Commit | Line | Data |
---|---|---|
a0d0e21e | 1 | /* toke.c |
a687059c | 2 | * |
bc89e66f | 3 | * Copyright (c) 1991-2001, Larry Wall |
a687059c | 4 | * |
d48672a2 LW |
5 | * You may distribute under the terms of either the GNU General Public |
6 | * License or the Artistic License, as specified in the README file. | |
378cc40b | 7 | * |
a0d0e21e LW |
8 | */ |
9 | ||
10 | /* | |
11 | * "It all comes from here, the stench and the peril." --Frodo | |
378cc40b LW |
12 | */ |
13 | ||
9cbb5ea2 GS |
14 | /* |
15 | * This file is the lexer for Perl. It's closely linked to the | |
4e553d73 | 16 | * parser, perly.y. |
ffb4593c NT |
17 | * |
18 | * The main routine is yylex(), which returns the next token. | |
19 | */ | |
20 | ||
378cc40b | 21 | #include "EXTERN.h" |
864dbfa3 | 22 | #define PERL_IN_TOKE_C |
378cc40b | 23 | #include "perl.h" |
378cc40b | 24 | |
d3b6f988 GS |
25 | #define yychar PL_yychar |
26 | #define yylval PL_yylval | |
27 | ||
fc36a67e | 28 | static char ident_too_long[] = "Identifier too long"; |
8903cb82 | 29 | |
51371543 | 30 | static void restore_rsfp(pTHXo_ void *f); |
6e3aabd6 GS |
31 | #ifndef PERL_NO_UTF16_FILTER |
32 | static I32 utf16_textfilter(pTHXo_ int idx, SV *sv, int maxlen); | |
33 | static I32 utf16rev_textfilter(pTHXo_ int idx, SV *sv, int maxlen); | |
34 | #endif | |
51371543 | 35 | |
9059aa12 LW |
36 | #define XFAKEBRACK 128 |
37 | #define XENUMMASK 127 | |
38 | ||
7e2040f0 | 39 | /*#define UTF (SvUTF8(PL_linestr) && !(PL_hints & HINT_BYTE))*/ |
a0ed51b3 LW |
40 | #define UTF (PL_hints & HINT_UTF8) |
41 | ||
4e553d73 | 42 | /* In variables name $^X, these are the legal values for X. |
2b92dfce GS |
43 | * 1999-02-27 mjd-perl-patch@plover.com */ |
44 | #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x))) | |
45 | ||
bf4acbe4 GS |
46 | /* On MacOS, respect nonbreaking spaces */ |
47 | #ifdef MACOS_TRADITIONAL | |
48 | #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t') | |
49 | #else | |
50 | #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t') | |
51 | #endif | |
52 | ||
ffb4593c NT |
53 | /* LEX_* are values for PL_lex_state, the state of the lexer. |
54 | * They are arranged oddly so that the guard on the switch statement | |
79072805 LW |
55 | * can get by with a single comparison (if the compiler is smart enough). |
56 | */ | |
57 | ||
fb73857a | 58 | /* #define LEX_NOTPARSING 11 is done in perl.h. */ |
59 | ||
55497cff | 60 | #define LEX_NORMAL 10 |
61 | #define LEX_INTERPNORMAL 9 | |
62 | #define LEX_INTERPCASEMOD 8 | |
63 | #define LEX_INTERPPUSH 7 | |
64 | #define LEX_INTERPSTART 6 | |
65 | #define LEX_INTERPEND 5 | |
66 | #define LEX_INTERPENDMAYBE 4 | |
67 | #define LEX_INTERPCONCAT 3 | |
68 | #define LEX_INTERPCONST 2 | |
69 | #define LEX_FORMLINE 1 | |
70 | #define LEX_KNOWNEXT 0 | |
79072805 | 71 | |
79072805 LW |
72 | #ifdef ff_next |
73 | #undef ff_next | |
d48672a2 LW |
74 | #endif |
75 | ||
a1a0e61e | 76 | #ifdef USE_PURE_BISON |
dba4d153 JH |
77 | # ifndef YYMAXLEVEL |
78 | # define YYMAXLEVEL 100 | |
79 | # endif | |
20141f0e IRC |
80 | YYSTYPE* yylval_pointer[YYMAXLEVEL]; |
81 | int* yychar_pointer[YYMAXLEVEL]; | |
6f202aea | 82 | int yyactlevel = -1; |
22c35a8c GS |
83 | # undef yylval |
84 | # undef yychar | |
20141f0e IRC |
85 | # define yylval (*yylval_pointer[yyactlevel]) |
86 | # define yychar (*yychar_pointer[yyactlevel]) | |
87 | # define PERL_YYLEX_PARAM yylval_pointer[yyactlevel],yychar_pointer[yyactlevel] | |
4e553d73 | 88 | # undef yylex |
dba4d153 | 89 | # define yylex() Perl_yylex_r(aTHX_ yylval_pointer[yyactlevel],yychar_pointer[yyactlevel]) |
a1a0e61e TD |
90 | #endif |
91 | ||
79072805 | 92 | #include "keywords.h" |
fe14fcc3 | 93 | |
ffb4593c NT |
94 | /* CLINE is a macro that ensures PL_copline has a sane value */ |
95 | ||
ae986130 LW |
96 | #ifdef CLINE |
97 | #undef CLINE | |
98 | #endif | |
57843af0 | 99 | #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline)) |
3280af22 | 100 | |
ffb4593c NT |
101 | /* |
102 | * Convenience functions to return different tokens and prime the | |
9cbb5ea2 | 103 | * lexer for the next token. They all take an argument. |
ffb4593c NT |
104 | * |
105 | * TOKEN : generic token (used for '(', DOLSHARP, etc) | |
106 | * OPERATOR : generic operator | |
107 | * AOPERATOR : assignment operator | |
108 | * PREBLOCK : beginning the block after an if, while, foreach, ... | |
109 | * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref) | |
110 | * PREREF : *EXPR where EXPR is not a simple identifier | |
111 | * TERM : expression term | |
112 | * LOOPX : loop exiting command (goto, last, dump, etc) | |
113 | * FTST : file test operator | |
114 | * FUN0 : zero-argument function | |
2d2e263d | 115 | * FUN1 : not used, except for not, which isn't a UNIOP |
ffb4593c NT |
116 | * BOop : bitwise or or xor |
117 | * BAop : bitwise and | |
118 | * SHop : shift operator | |
119 | * PWop : power operator | |
9cbb5ea2 | 120 | * PMop : pattern-matching operator |
ffb4593c NT |
121 | * Aop : addition-level operator |
122 | * Mop : multiplication-level operator | |
123 | * Eop : equality-testing operator | |
e5edeb50 | 124 | * Rop : relational operator <= != gt |
ffb4593c NT |
125 | * |
126 | * Also see LOP and lop() below. | |
127 | */ | |
128 | ||
998054bd SC |
129 | #ifdef DEBUGGING /* Serve -DT. */ |
130 | # define REPORT(x,retval) tokereport(x,s,(int)retval) | |
131 | # define REPORT2(x,retval) tokereport(x,s, yylval.ival) | |
132 | #else | |
133 | # define REPORT(x,retval) 1 | |
134 | # define REPORT2(x,retval) 1 | |
135 | #endif | |
136 | ||
137 | #define TOKEN(retval) return (REPORT2("token",retval), PL_bufptr = s,(int)retval) | |
138 | #define OPERATOR(retval) return (REPORT2("operator",retval), PL_expect = XTERM, PL_bufptr = s,(int)retval) | |
139 | #define AOPERATOR(retval) return ao((REPORT2("aop",retval), PL_expect = XTERM, PL_bufptr = s,(int)retval)) | |
140 | #define PREBLOCK(retval) return (REPORT2("preblock",retval), PL_expect = XBLOCK,PL_bufptr = s,(int)retval) | |
141 | #define PRETERMBLOCK(retval) return (REPORT2("pretermblock",retval), PL_expect = XTERMBLOCK,PL_bufptr = s,(int)retval) | |
142 | #define PREREF(retval) return (REPORT2("preref",retval), PL_expect = XREF,PL_bufptr = s,(int)retval) | |
143 | #define TERM(retval) return (CLINE, REPORT2("term",retval), PL_expect = XOPERATOR, PL_bufptr = s,(int)retval) | |
144 | #define LOOPX(f) return(yylval.ival=f, REPORT("loopx",f), PL_expect = XTERM,PL_bufptr = s,(int)LOOPEX) | |
145 | #define FTST(f) return(yylval.ival=f, REPORT("ftst",f), PL_expect = XTERM,PL_bufptr = s,(int)UNIOP) | |
146 | #define FUN0(f) return(yylval.ival = f, REPORT("fun0",f), PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC0) | |
147 | #define FUN1(f) return(yylval.ival = f, REPORT("fun1",f), PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC1) | |
148 | #define BOop(f) return ao((yylval.ival=f, REPORT("bitorop",f), PL_expect = XTERM,PL_bufptr = s,(int)BITOROP)) | |
149 | #define BAop(f) return ao((yylval.ival=f, REPORT("bitandop",f), PL_expect = XTERM,PL_bufptr = s,(int)BITANDOP)) | |
150 | #define SHop(f) return ao((yylval.ival=f, REPORT("shiftop",f), PL_expect = XTERM,PL_bufptr = s,(int)SHIFTOP)) | |
151 | #define PWop(f) return ao((yylval.ival=f, REPORT("powop",f), PL_expect = XTERM,PL_bufptr = s,(int)POWOP)) | |
152 | #define PMop(f) return(yylval.ival=f, REPORT("matchop",f), PL_expect = XTERM,PL_bufptr = s,(int)MATCHOP) | |
153 | #define Aop(f) return ao((yylval.ival=f, REPORT("add",f), PL_expect = XTERM,PL_bufptr = s,(int)ADDOP)) | |
154 | #define Mop(f) return ao((yylval.ival=f, REPORT("mul",f), PL_expect = XTERM,PL_bufptr = s,(int)MULOP)) | |
155 | #define Eop(f) return(yylval.ival=f, REPORT("eq",f), PL_expect = XTERM,PL_bufptr = s,(int)EQOP) | |
156 | #define Rop(f) return(yylval.ival=f, REPORT("rel",f), PL_expect = XTERM,PL_bufptr = s,(int)RELOP) | |
2f3197b3 | 157 | |
a687059c LW |
158 | /* This bit of chicanery makes a unary function followed by |
159 | * a parenthesis into a function with one argument, highest precedence. | |
160 | */ | |
2f3197b3 | 161 | #define UNI(f) return(yylval.ival = f, \ |
998054bd | 162 | REPORT("uni",f), \ |
3280af22 NIS |
163 | PL_expect = XTERM, \ |
164 | PL_bufptr = s, \ | |
165 | PL_last_uni = PL_oldbufptr, \ | |
166 | PL_last_lop_op = f, \ | |
a687059c LW |
167 | (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) ) |
168 | ||
79072805 | 169 | #define UNIBRACK(f) return(yylval.ival = f, \ |
998054bd | 170 | REPORT("uni",f), \ |
3280af22 NIS |
171 | PL_bufptr = s, \ |
172 | PL_last_uni = PL_oldbufptr, \ | |
79072805 LW |
173 | (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) ) |
174 | ||
9f68db38 | 175 | /* grandfather return to old style */ |
3280af22 | 176 | #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP) |
79072805 | 177 | |
2d00ba3b | 178 | STATIC void |
61b2116b | 179 | S_tokereport(pTHX_ char *thing, char* s, I32 rv) |
998054bd SC |
180 | { |
181 | SV *report; | |
182 | DEBUG_T({ | |
183 | report = newSVpv(thing, 0); | |
61b2116b | 184 | Perl_sv_catpvf(aTHX_ report, ":line %i:%i:", CopLINE(PL_curcop), rv); |
998054bd SC |
185 | |
186 | if (s - PL_bufptr > 0) | |
187 | sv_catpvn(report, PL_bufptr, s - PL_bufptr); | |
188 | else { | |
189 | if (PL_oldbufptr && *PL_oldbufptr) | |
190 | sv_catpv(report, PL_tokenbuf); | |
191 | } | |
192 | PerlIO_printf(Perl_debug_log, "### %s\n", SvPV_nolen(report)); | |
193 | }) | |
194 | } | |
195 | ||
ffb4593c NT |
196 | /* |
197 | * S_ao | |
198 | * | |
199 | * This subroutine detects &&= and ||= and turns an ANDAND or OROR | |
200 | * into an OP_ANDASSIGN or OP_ORASSIGN | |
201 | */ | |
202 | ||
76e3520e | 203 | STATIC int |
cea2e8a9 | 204 | S_ao(pTHX_ int toketype) |
a0d0e21e | 205 | { |
3280af22 NIS |
206 | if (*PL_bufptr == '=') { |
207 | PL_bufptr++; | |
a0d0e21e LW |
208 | if (toketype == ANDAND) |
209 | yylval.ival = OP_ANDASSIGN; | |
210 | else if (toketype == OROR) | |
211 | yylval.ival = OP_ORASSIGN; | |
212 | toketype = ASSIGNOP; | |
213 | } | |
214 | return toketype; | |
215 | } | |
216 | ||
ffb4593c NT |
217 | /* |
218 | * S_no_op | |
219 | * When Perl expects an operator and finds something else, no_op | |
220 | * prints the warning. It always prints "<something> found where | |
221 | * operator expected. It prints "Missing semicolon on previous line?" | |
222 | * if the surprise occurs at the start of the line. "do you need to | |
223 | * predeclare ..." is printed out for code like "sub bar; foo bar $x" | |
224 | * where the compiler doesn't know if foo is a method call or a function. | |
225 | * It prints "Missing operator before end of line" if there's nothing | |
226 | * after the missing operator, or "... before <...>" if there is something | |
227 | * after the missing operator. | |
228 | */ | |
229 | ||
76e3520e | 230 | STATIC void |
cea2e8a9 | 231 | S_no_op(pTHX_ char *what, char *s) |
463ee0b2 | 232 | { |
3280af22 NIS |
233 | char *oldbp = PL_bufptr; |
234 | bool is_first = (PL_oldbufptr == PL_linestart); | |
68dc0745 | 235 | |
1189a94a GS |
236 | if (!s) |
237 | s = oldbp; | |
07c798fb | 238 | else |
1189a94a | 239 | PL_bufptr = s; |
cea2e8a9 | 240 | yywarn(Perl_form(aTHX_ "%s found where operator expected", what)); |
748a9306 | 241 | if (is_first) |
cea2e8a9 | 242 | Perl_warn(aTHX_ "\t(Missing semicolon on previous line?)\n"); |
7e2040f0 | 243 | else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) { |
748a9306 | 244 | char *t; |
7e2040f0 | 245 | for (t = PL_oldoldbufptr; *t && (isALNUM_lazy_if(t,UTF) || *t == ':'); t++) ; |
3280af22 | 246 | if (t < PL_bufptr && isSPACE(*t)) |
cea2e8a9 | 247 | Perl_warn(aTHX_ "\t(Do you need to predeclare %.*s?)\n", |
3280af22 | 248 | t - PL_oldoldbufptr, PL_oldoldbufptr); |
748a9306 | 249 | } |
07c798fb HS |
250 | else { |
251 | assert(s >= oldbp); | |
cea2e8a9 | 252 | Perl_warn(aTHX_ "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp); |
07c798fb | 253 | } |
3280af22 | 254 | PL_bufptr = oldbp; |
8990e307 LW |
255 | } |
256 | ||
ffb4593c NT |
257 | /* |
258 | * S_missingterm | |
259 | * Complain about missing quote/regexp/heredoc terminator. | |
260 | * If it's called with (char *)NULL then it cauterizes the line buffer. | |
261 | * If we're in a delimited string and the delimiter is a control | |
262 | * character, it's reformatted into a two-char sequence like ^C. | |
263 | * This is fatal. | |
264 | */ | |
265 | ||
76e3520e | 266 | STATIC void |
cea2e8a9 | 267 | S_missingterm(pTHX_ char *s) |
8990e307 LW |
268 | { |
269 | char tmpbuf[3]; | |
270 | char q; | |
271 | if (s) { | |
272 | char *nl = strrchr(s,'\n'); | |
d2719217 | 273 | if (nl) |
8990e307 LW |
274 | *nl = '\0'; |
275 | } | |
9d116dd7 JH |
276 | else if ( |
277 | #ifdef EBCDIC | |
278 | iscntrl(PL_multi_close) | |
279 | #else | |
280 | PL_multi_close < 32 || PL_multi_close == 127 | |
281 | #endif | |
282 | ) { | |
8990e307 | 283 | *tmpbuf = '^'; |
3280af22 | 284 | tmpbuf[1] = toCTRL(PL_multi_close); |
8990e307 LW |
285 | s = "\\n"; |
286 | tmpbuf[2] = '\0'; | |
287 | s = tmpbuf; | |
288 | } | |
289 | else { | |
3280af22 | 290 | *tmpbuf = PL_multi_close; |
8990e307 LW |
291 | tmpbuf[1] = '\0'; |
292 | s = tmpbuf; | |
293 | } | |
294 | q = strchr(s,'"') ? '\'' : '"'; | |
cea2e8a9 | 295 | Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q); |
463ee0b2 | 296 | } |
79072805 | 297 | |
ffb4593c NT |
298 | /* |
299 | * Perl_deprecate | |
ffb4593c NT |
300 | */ |
301 | ||
79072805 | 302 | void |
864dbfa3 | 303 | Perl_deprecate(pTHX_ char *s) |
a0d0e21e | 304 | { |
599cee73 | 305 | if (ckWARN(WARN_DEPRECATED)) |
cea2e8a9 | 306 | Perl_warner(aTHX_ WARN_DEPRECATED, "Use of %s is deprecated", s); |
a0d0e21e LW |
307 | } |
308 | ||
ffb4593c NT |
309 | /* |
310 | * depcom | |
9cbb5ea2 | 311 | * Deprecate a comma-less variable list. |
ffb4593c NT |
312 | */ |
313 | ||
76e3520e | 314 | STATIC void |
cea2e8a9 | 315 | S_depcom(pTHX) |
a0d0e21e LW |
316 | { |
317 | deprecate("comma-less variable list"); | |
318 | } | |
319 | ||
ffb4593c | 320 | /* |
9cbb5ea2 GS |
321 | * experimental text filters for win32 carriage-returns, utf16-to-utf8 and |
322 | * utf16-to-utf8-reversed. | |
ffb4593c NT |
323 | */ |
324 | ||
c39cd008 GS |
325 | #ifdef PERL_CR_FILTER |
326 | static void | |
327 | strip_return(SV *sv) | |
328 | { | |
329 | register char *s = SvPVX(sv); | |
330 | register char *e = s + SvCUR(sv); | |
331 | /* outer loop optimized to do nothing if there are no CR-LFs */ | |
332 | while (s < e) { | |
333 | if (*s++ == '\r' && *s == '\n') { | |
334 | /* hit a CR-LF, need to copy the rest */ | |
335 | register char *d = s - 1; | |
336 | *d++ = *s++; | |
337 | while (s < e) { | |
338 | if (*s == '\r' && s[1] == '\n') | |
339 | s++; | |
340 | *d++ = *s++; | |
341 | } | |
342 | SvCUR(sv) -= s - d; | |
343 | return; | |
344 | } | |
345 | } | |
346 | } | |
a868473f | 347 | |
76e3520e | 348 | STATIC I32 |
c39cd008 | 349 | S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen) |
a868473f | 350 | { |
c39cd008 GS |
351 | I32 count = FILTER_READ(idx+1, sv, maxlen); |
352 | if (count > 0 && !maxlen) | |
353 | strip_return(sv); | |
354 | return count; | |
a868473f NIS |
355 | } |
356 | #endif | |
357 | ||
ffb4593c NT |
358 | /* |
359 | * Perl_lex_start | |
9cbb5ea2 GS |
360 | * Initialize variables. Uses the Perl save_stack to save its state (for |
361 | * recursive calls to the parser). | |
ffb4593c NT |
362 | */ |
363 | ||
a0d0e21e | 364 | void |
864dbfa3 | 365 | Perl_lex_start(pTHX_ SV *line) |
79072805 | 366 | { |
8990e307 LW |
367 | char *s; |
368 | STRLEN len; | |
369 | ||
3280af22 NIS |
370 | SAVEI32(PL_lex_dojoin); |
371 | SAVEI32(PL_lex_brackets); | |
3280af22 NIS |
372 | SAVEI32(PL_lex_casemods); |
373 | SAVEI32(PL_lex_starts); | |
374 | SAVEI32(PL_lex_state); | |
7766f137 | 375 | SAVEVPTR(PL_lex_inpat); |
3280af22 | 376 | SAVEI32(PL_lex_inwhat); |
18b09519 GS |
377 | if (PL_lex_state == LEX_KNOWNEXT) { |
378 | I32 toke = PL_nexttoke; | |
379 | while (--toke >= 0) { | |
380 | SAVEI32(PL_nexttype[toke]); | |
381 | SAVEVPTR(PL_nextval[toke]); | |
382 | } | |
383 | SAVEI32(PL_nexttoke); | |
18b09519 | 384 | } |
57843af0 | 385 | SAVECOPLINE(PL_curcop); |
3280af22 NIS |
386 | SAVEPPTR(PL_bufptr); |
387 | SAVEPPTR(PL_bufend); | |
388 | SAVEPPTR(PL_oldbufptr); | |
389 | SAVEPPTR(PL_oldoldbufptr); | |
390 | SAVEPPTR(PL_linestart); | |
391 | SAVESPTR(PL_linestr); | |
392 | SAVEPPTR(PL_lex_brackstack); | |
393 | SAVEPPTR(PL_lex_casestack); | |
c76ac1ee | 394 | SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp); |
3280af22 NIS |
395 | SAVESPTR(PL_lex_stuff); |
396 | SAVEI32(PL_lex_defer); | |
09bef843 | 397 | SAVEI32(PL_sublex_info.sub_inwhat); |
3280af22 | 398 | SAVESPTR(PL_lex_repl); |
bebdddfc GS |
399 | SAVEINT(PL_expect); |
400 | SAVEINT(PL_lex_expect); | |
3280af22 NIS |
401 | |
402 | PL_lex_state = LEX_NORMAL; | |
403 | PL_lex_defer = 0; | |
404 | PL_expect = XSTATE; | |
405 | PL_lex_brackets = 0; | |
3280af22 NIS |
406 | New(899, PL_lex_brackstack, 120, char); |
407 | New(899, PL_lex_casestack, 12, char); | |
408 | SAVEFREEPV(PL_lex_brackstack); | |
409 | SAVEFREEPV(PL_lex_casestack); | |
410 | PL_lex_casemods = 0; | |
411 | *PL_lex_casestack = '\0'; | |
412 | PL_lex_dojoin = 0; | |
413 | PL_lex_starts = 0; | |
414 | PL_lex_stuff = Nullsv; | |
415 | PL_lex_repl = Nullsv; | |
416 | PL_lex_inpat = 0; | |
76be56bc | 417 | PL_nexttoke = 0; |
3280af22 | 418 | PL_lex_inwhat = 0; |
09bef843 | 419 | PL_sublex_info.sub_inwhat = 0; |
3280af22 NIS |
420 | PL_linestr = line; |
421 | if (SvREADONLY(PL_linestr)) | |
422 | PL_linestr = sv_2mortal(newSVsv(PL_linestr)); | |
423 | s = SvPV(PL_linestr, len); | |
8990e307 | 424 | if (len && s[len-1] != ';') { |
3280af22 NIS |
425 | if (!(SvFLAGS(PL_linestr) & SVs_TEMP)) |
426 | PL_linestr = sv_2mortal(newSVsv(PL_linestr)); | |
427 | sv_catpvn(PL_linestr, "\n;", 2); | |
8990e307 | 428 | } |
3280af22 NIS |
429 | SvTEMP_off(PL_linestr); |
430 | PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr); | |
431 | PL_bufend = PL_bufptr + SvCUR(PL_linestr); | |
432 | SvREFCNT_dec(PL_rs); | |
79cb57f6 | 433 | PL_rs = newSVpvn("\n", 1); |
3280af22 | 434 | PL_rsfp = 0; |
79072805 | 435 | } |
a687059c | 436 | |
ffb4593c NT |
437 | /* |
438 | * Perl_lex_end | |
9cbb5ea2 GS |
439 | * Finalizer for lexing operations. Must be called when the parser is |
440 | * done with the lexer. | |
ffb4593c NT |
441 | */ |
442 | ||
463ee0b2 | 443 | void |
864dbfa3 | 444 | Perl_lex_end(pTHX) |
463ee0b2 | 445 | { |
3280af22 | 446 | PL_doextract = FALSE; |
463ee0b2 LW |
447 | } |
448 | ||
ffb4593c NT |
449 | /* |
450 | * S_incline | |
451 | * This subroutine has nothing to do with tilting, whether at windmills | |
452 | * or pinball tables. Its name is short for "increment line". It | |
57843af0 | 453 | * increments the current line number in CopLINE(PL_curcop) and checks |
ffb4593c | 454 | * to see whether the line starts with a comment of the form |
9cbb5ea2 GS |
455 | * # line 500 "foo.pm" |
456 | * If so, it sets the current line number and file to the values in the comment. | |
ffb4593c NT |
457 | */ |
458 | ||
76e3520e | 459 | STATIC void |
cea2e8a9 | 460 | S_incline(pTHX_ char *s) |
463ee0b2 LW |
461 | { |
462 | char *t; | |
463 | char *n; | |
73659bf1 | 464 | char *e; |
463ee0b2 | 465 | char ch; |
463ee0b2 | 466 | |
57843af0 | 467 | CopLINE_inc(PL_curcop); |
463ee0b2 LW |
468 | if (*s++ != '#') |
469 | return; | |
bf4acbe4 | 470 | while (SPACE_OR_TAB(*s)) s++; |
73659bf1 GS |
471 | if (strnEQ(s, "line", 4)) |
472 | s += 4; | |
473 | else | |
474 | return; | |
084592ab | 475 | if (SPACE_OR_TAB(*s)) |
73659bf1 | 476 | s++; |
4e553d73 | 477 | else |
73659bf1 | 478 | return; |
bf4acbe4 | 479 | while (SPACE_OR_TAB(*s)) s++; |
463ee0b2 LW |
480 | if (!isDIGIT(*s)) |
481 | return; | |
482 | n = s; | |
483 | while (isDIGIT(*s)) | |
484 | s++; | |
bf4acbe4 | 485 | while (SPACE_OR_TAB(*s)) |
463ee0b2 | 486 | s++; |
73659bf1 | 487 | if (*s == '"' && (t = strchr(s+1, '"'))) { |
463ee0b2 | 488 | s++; |
73659bf1 GS |
489 | e = t + 1; |
490 | } | |
463ee0b2 | 491 | else { |
463ee0b2 | 492 | for (t = s; !isSPACE(*t); t++) ; |
73659bf1 | 493 | e = t; |
463ee0b2 | 494 | } |
bf4acbe4 | 495 | while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f') |
73659bf1 GS |
496 | e++; |
497 | if (*e != '\n' && *e != '\0') | |
498 | return; /* false alarm */ | |
499 | ||
463ee0b2 LW |
500 | ch = *t; |
501 | *t = '\0'; | |
f4dd75d9 GS |
502 | if (t - s > 0) { |
503 | #ifdef USE_ITHREADS | |
504 | Safefree(CopFILE(PL_curcop)); | |
505 | #else | |
506 | SvREFCNT_dec(CopFILEGV(PL_curcop)); | |
507 | #endif | |
57843af0 | 508 | CopFILE_set(PL_curcop, s); |
f4dd75d9 | 509 | } |
463ee0b2 | 510 | *t = ch; |
57843af0 | 511 | CopLINE_set(PL_curcop, atoi(n)-1); |
463ee0b2 LW |
512 | } |
513 | ||
ffb4593c NT |
514 | /* |
515 | * S_skipspace | |
516 | * Called to gobble the appropriate amount and type of whitespace. | |
517 | * Skips comments as well. | |
518 | */ | |
519 | ||
76e3520e | 520 | STATIC char * |
cea2e8a9 | 521 | S_skipspace(pTHX_ register char *s) |
a687059c | 522 | { |
3280af22 | 523 | if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) { |
bf4acbe4 | 524 | while (s < PL_bufend && SPACE_OR_TAB(*s)) |
463ee0b2 LW |
525 | s++; |
526 | return s; | |
527 | } | |
528 | for (;;) { | |
fd049845 | 529 | STRLEN prevlen; |
09bef843 SB |
530 | SSize_t oldprevlen, oldoldprevlen; |
531 | SSize_t oldloplen, oldunilen; | |
60e6418e GS |
532 | while (s < PL_bufend && isSPACE(*s)) { |
533 | if (*s++ == '\n' && PL_in_eval && !PL_rsfp) | |
534 | incline(s); | |
535 | } | |
ffb4593c NT |
536 | |
537 | /* comment */ | |
3280af22 NIS |
538 | if (s < PL_bufend && *s == '#') { |
539 | while (s < PL_bufend && *s != '\n') | |
463ee0b2 | 540 | s++; |
60e6418e | 541 | if (s < PL_bufend) { |
463ee0b2 | 542 | s++; |
60e6418e GS |
543 | if (PL_in_eval && !PL_rsfp) { |
544 | incline(s); | |
545 | continue; | |
546 | } | |
547 | } | |
463ee0b2 | 548 | } |
ffb4593c NT |
549 | |
550 | /* only continue to recharge the buffer if we're at the end | |
551 | * of the buffer, we're not reading from a source filter, and | |
552 | * we're in normal lexing mode | |
553 | */ | |
09bef843 SB |
554 | if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat || |
555 | PL_lex_state == LEX_FORMLINE) | |
463ee0b2 | 556 | return s; |
ffb4593c NT |
557 | |
558 | /* try to recharge the buffer */ | |
9cbb5ea2 GS |
559 | if ((s = filter_gets(PL_linestr, PL_rsfp, |
560 | (prevlen = SvCUR(PL_linestr)))) == Nullch) | |
561 | { | |
562 | /* end of file. Add on the -p or -n magic */ | |
3280af22 NIS |
563 | if (PL_minus_n || PL_minus_p) { |
564 | sv_setpv(PL_linestr,PL_minus_p ? | |
08e9d68e DD |
565 | ";}continue{print or die qq(-p destination: $!\\n)" : |
566 | ""); | |
3280af22 NIS |
567 | sv_catpv(PL_linestr,";}"); |
568 | PL_minus_n = PL_minus_p = 0; | |
a0d0e21e LW |
569 | } |
570 | else | |
3280af22 | 571 | sv_setpv(PL_linestr,";"); |
ffb4593c NT |
572 | |
573 | /* reset variables for next time we lex */ | |
9cbb5ea2 GS |
574 | PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart |
575 | = SvPVX(PL_linestr); | |
3280af22 | 576 | PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); |
ffb4593c NT |
577 | |
578 | /* Close the filehandle. Could be from -P preprocessor, | |
579 | * STDIN, or a regular file. If we were reading code from | |
580 | * STDIN (because the commandline held no -e or filename) | |
581 | * then we don't close it, we reset it so the code can | |
582 | * read from STDIN too. | |
583 | */ | |
584 | ||
3280af22 NIS |
585 | if (PL_preprocess && !PL_in_eval) |
586 | (void)PerlProc_pclose(PL_rsfp); | |
587 | else if ((PerlIO*)PL_rsfp == PerlIO_stdin()) | |
588 | PerlIO_clearerr(PL_rsfp); | |
8990e307 | 589 | else |
3280af22 NIS |
590 | (void)PerlIO_close(PL_rsfp); |
591 | PL_rsfp = Nullfp; | |
463ee0b2 LW |
592 | return s; |
593 | } | |
ffb4593c NT |
594 | |
595 | /* not at end of file, so we only read another line */ | |
09bef843 SB |
596 | /* make corresponding updates to old pointers, for yyerror() */ |
597 | oldprevlen = PL_oldbufptr - PL_bufend; | |
598 | oldoldprevlen = PL_oldoldbufptr - PL_bufend; | |
599 | if (PL_last_uni) | |
600 | oldunilen = PL_last_uni - PL_bufend; | |
601 | if (PL_last_lop) | |
602 | oldloplen = PL_last_lop - PL_bufend; | |
3280af22 NIS |
603 | PL_linestart = PL_bufptr = s + prevlen; |
604 | PL_bufend = s + SvCUR(PL_linestr); | |
605 | s = PL_bufptr; | |
09bef843 SB |
606 | PL_oldbufptr = s + oldprevlen; |
607 | PL_oldoldbufptr = s + oldoldprevlen; | |
608 | if (PL_last_uni) | |
609 | PL_last_uni = s + oldunilen; | |
610 | if (PL_last_lop) | |
611 | PL_last_lop = s + oldloplen; | |
a0d0e21e | 612 | incline(s); |
ffb4593c NT |
613 | |
614 | /* debugger active and we're not compiling the debugger code, | |
615 | * so store the line into the debugger's array of lines | |
616 | */ | |
3280af22 | 617 | if (PERLDB_LINE && PL_curstash != PL_debstash) { |
8990e307 LW |
618 | SV *sv = NEWSV(85,0); |
619 | ||
620 | sv_upgrade(sv, SVt_PVMG); | |
3280af22 | 621 | sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr); |
57843af0 | 622 | av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv); |
8990e307 | 623 | } |
463ee0b2 | 624 | } |
a687059c | 625 | } |
378cc40b | 626 | |
ffb4593c NT |
627 | /* |
628 | * S_check_uni | |
629 | * Check the unary operators to ensure there's no ambiguity in how they're | |
630 | * used. An ambiguous piece of code would be: | |
631 | * rand + 5 | |
632 | * This doesn't mean rand() + 5. Because rand() is a unary operator, | |
633 | * the +5 is its argument. | |
634 | */ | |
635 | ||
76e3520e | 636 | STATIC void |
cea2e8a9 | 637 | S_check_uni(pTHX) |
ba106d47 | 638 | { |
2f3197b3 | 639 | char *s; |
a0d0e21e | 640 | char *t; |
2f3197b3 | 641 | |
3280af22 | 642 | if (PL_oldoldbufptr != PL_last_uni) |
2f3197b3 | 643 | return; |
3280af22 NIS |
644 | while (isSPACE(*PL_last_uni)) |
645 | PL_last_uni++; | |
7e2040f0 | 646 | for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ; |
3280af22 | 647 | if ((t = strchr(s, '(')) && t < PL_bufptr) |
a0d0e21e | 648 | return; |
0453d815 | 649 | if (ckWARN_d(WARN_AMBIGUOUS)){ |
f248d071 | 650 | char ch = *s; |
0453d815 | 651 | *s = '\0'; |
4e553d73 NIS |
652 | Perl_warner(aTHX_ WARN_AMBIGUOUS, |
653 | "Warning: Use of \"%s\" without parens is ambiguous", | |
0453d815 PM |
654 | PL_last_uni); |
655 | *s = ch; | |
656 | } | |
2f3197b3 LW |
657 | } |
658 | ||
ffb4593c NT |
659 | /* workaround to replace the UNI() macro with a function. Only the |
660 | * hints/uts.sh file mentions this. Other comments elsewhere in the | |
661 | * source indicate Microport Unix might need it too. | |
662 | */ | |
663 | ||
ffed7fef LW |
664 | #ifdef CRIPPLED_CC |
665 | ||
666 | #undef UNI | |
ffed7fef | 667 | #define UNI(f) return uni(f,s) |
ffed7fef | 668 | |
76e3520e | 669 | STATIC int |
cea2e8a9 | 670 | S_uni(pTHX_ I32 f, char *s) |
ffed7fef LW |
671 | { |
672 | yylval.ival = f; | |
3280af22 NIS |
673 | PL_expect = XTERM; |
674 | PL_bufptr = s; | |
8f872242 NIS |
675 | PL_last_uni = PL_oldbufptr; |
676 | PL_last_lop_op = f; | |
ffed7fef LW |
677 | if (*s == '(') |
678 | return FUNC1; | |
679 | s = skipspace(s); | |
680 | if (*s == '(') | |
681 | return FUNC1; | |
682 | else | |
683 | return UNIOP; | |
684 | } | |
685 | ||
a0d0e21e LW |
686 | #endif /* CRIPPLED_CC */ |
687 | ||
ffb4593c NT |
688 | /* |
689 | * LOP : macro to build a list operator. Its behaviour has been replaced | |
690 | * with a subroutine, S_lop() for which LOP is just another name. | |
691 | */ | |
692 | ||
a0d0e21e LW |
693 | #define LOP(f,x) return lop(f,x,s) |
694 | ||
ffb4593c NT |
695 | /* |
696 | * S_lop | |
697 | * Build a list operator (or something that might be one). The rules: | |
698 | * - if we have a next token, then it's a list operator [why?] | |
699 | * - if the next thing is an opening paren, then it's a function | |
700 | * - else it's a list operator | |
701 | */ | |
702 | ||
76e3520e | 703 | STATIC I32 |
a0be28da | 704 | S_lop(pTHX_ I32 f, int x, char *s) |
ffed7fef | 705 | { |
79072805 | 706 | yylval.ival = f; |
35c8bce7 | 707 | CLINE; |
998054bd | 708 | REPORT("lop", f); |
3280af22 NIS |
709 | PL_expect = x; |
710 | PL_bufptr = s; | |
711 | PL_last_lop = PL_oldbufptr; | |
712 | PL_last_lop_op = f; | |
713 | if (PL_nexttoke) | |
a0d0e21e | 714 | return LSTOP; |
79072805 LW |
715 | if (*s == '(') |
716 | return FUNC; | |
717 | s = skipspace(s); | |
718 | if (*s == '(') | |
719 | return FUNC; | |
720 | else | |
721 | return LSTOP; | |
722 | } | |
723 | ||
ffb4593c NT |
724 | /* |
725 | * S_force_next | |
9cbb5ea2 | 726 | * When the lexer realizes it knows the next token (for instance, |
ffb4593c | 727 | * it is reordering tokens for the parser) then it can call S_force_next |
9cbb5ea2 GS |
728 | * to know what token to return the next time the lexer is called. Caller |
729 | * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer | |
730 | * handles the token correctly. | |
ffb4593c NT |
731 | */ |
732 | ||
4e553d73 | 733 | STATIC void |
cea2e8a9 | 734 | S_force_next(pTHX_ I32 type) |
79072805 | 735 | { |
3280af22 NIS |
736 | PL_nexttype[PL_nexttoke] = type; |
737 | PL_nexttoke++; | |
738 | if (PL_lex_state != LEX_KNOWNEXT) { | |
739 | PL_lex_defer = PL_lex_state; | |
740 | PL_lex_expect = PL_expect; | |
741 | PL_lex_state = LEX_KNOWNEXT; | |
79072805 LW |
742 | } |
743 | } | |
744 | ||
ffb4593c NT |
745 | /* |
746 | * S_force_word | |
747 | * When the lexer knows the next thing is a word (for instance, it has | |
748 | * just seen -> and it knows that the next char is a word char, then | |
749 | * it calls S_force_word to stick the next word into the PL_next lookahead. | |
750 | * | |
751 | * Arguments: | |
b1b65b59 | 752 | * char *start : buffer position (must be within PL_linestr) |
ffb4593c NT |
753 | * int token : PL_next will be this type of bare word (e.g., METHOD,WORD) |
754 | * int check_keyword : if true, Perl checks to make sure the word isn't | |
755 | * a keyword (do this if the word is a label, e.g. goto FOO) | |
756 | * int allow_pack : if true, : characters will also be allowed (require, | |
757 | * use, etc. do this) | |
9cbb5ea2 | 758 | * int allow_initial_tick : used by the "sub" lexer only. |
ffb4593c NT |
759 | */ |
760 | ||
76e3520e | 761 | STATIC char * |
cea2e8a9 | 762 | S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick) |
79072805 | 763 | { |
463ee0b2 LW |
764 | register char *s; |
765 | STRLEN len; | |
4e553d73 | 766 | |
463ee0b2 LW |
767 | start = skipspace(start); |
768 | s = start; | |
7e2040f0 | 769 | if (isIDFIRST_lazy_if(s,UTF) || |
a0d0e21e | 770 | (allow_pack && *s == ':') || |
15f0808c | 771 | (allow_initial_tick && *s == '\'') ) |
a0d0e21e | 772 | { |
3280af22 NIS |
773 | s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len); |
774 | if (check_keyword && keyword(PL_tokenbuf, len)) | |
463ee0b2 LW |
775 | return start; |
776 | if (token == METHOD) { | |
777 | s = skipspace(s); | |
778 | if (*s == '(') | |
3280af22 | 779 | PL_expect = XTERM; |
463ee0b2 | 780 | else { |
3280af22 | 781 | PL_expect = XOPERATOR; |
463ee0b2 | 782 | } |
79072805 | 783 | } |
3280af22 NIS |
784 | PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0)); |
785 | PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE; | |
79072805 LW |
786 | force_next(token); |
787 | } | |
788 | return s; | |
789 | } | |
790 | ||
ffb4593c NT |
791 | /* |
792 | * S_force_ident | |
9cbb5ea2 | 793 | * Called when the lexer wants $foo *foo &foo etc, but the program |
ffb4593c NT |
794 | * text only contains the "foo" portion. The first argument is a pointer |
795 | * to the "foo", and the second argument is the type symbol to prefix. | |
796 | * Forces the next token to be a "WORD". | |
9cbb5ea2 | 797 | * Creates the symbol if it didn't already exist (via gv_fetchpv()). |
ffb4593c NT |
798 | */ |
799 | ||
76e3520e | 800 | STATIC void |
cea2e8a9 | 801 | S_force_ident(pTHX_ register char *s, int kind) |
79072805 LW |
802 | { |
803 | if (s && *s) { | |
11343788 | 804 | OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0)); |
3280af22 | 805 | PL_nextval[PL_nexttoke].opval = o; |
79072805 | 806 | force_next(WORD); |
748a9306 | 807 | if (kind) { |
11343788 | 808 | o->op_private = OPpCONST_ENTERED; |
55497cff | 809 | /* XXX see note in pp_entereval() for why we forgo typo |
810 | warnings if the symbol must be introduced in an eval. | |
811 | GSAR 96-10-12 */ | |
3280af22 | 812 | gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE, |
a0d0e21e LW |
813 | kind == '$' ? SVt_PV : |
814 | kind == '@' ? SVt_PVAV : | |
815 | kind == '%' ? SVt_PVHV : | |
816 | SVt_PVGV | |
817 | ); | |
748a9306 | 818 | } |
79072805 LW |
819 | } |
820 | } | |
821 | ||
1571675a GS |
822 | NV |
823 | Perl_str_to_version(pTHX_ SV *sv) | |
824 | { | |
825 | NV retval = 0.0; | |
826 | NV nshift = 1.0; | |
827 | STRLEN len; | |
828 | char *start = SvPVx(sv,len); | |
3aa33fe5 | 829 | bool utf = SvUTF8(sv) ? TRUE : FALSE; |
1571675a GS |
830 | char *end = start + len; |
831 | while (start < end) { | |
ba210ebe | 832 | STRLEN skip; |
1571675a GS |
833 | UV n; |
834 | if (utf) | |
dcad2880 | 835 | n = utf8_to_uv((U8*)start, len, &skip, 0); |
1571675a GS |
836 | else { |
837 | n = *(U8*)start; | |
838 | skip = 1; | |
839 | } | |
840 | retval += ((NV)n)/nshift; | |
841 | start += skip; | |
842 | nshift *= 1000; | |
843 | } | |
844 | return retval; | |
845 | } | |
846 | ||
4e553d73 | 847 | /* |
ffb4593c NT |
848 | * S_force_version |
849 | * Forces the next token to be a version number. | |
850 | */ | |
851 | ||
76e3520e | 852 | STATIC char * |
cea2e8a9 | 853 | S_force_version(pTHX_ char *s) |
89bfa8cd | 854 | { |
855 | OP *version = Nullop; | |
44dcb63b | 856 | char *d; |
89bfa8cd | 857 | |
858 | s = skipspace(s); | |
859 | ||
44dcb63b | 860 | d = s; |
dd629d5b | 861 | if (*d == 'v') |
44dcb63b | 862 | d++; |
44dcb63b | 863 | if (isDIGIT(*d)) { |
a7cb1f99 | 864 | for (; isDIGIT(*d) || *d == '_' || *d == '.'; d++); |
9f3d182e | 865 | if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) { |
dd629d5b | 866 | SV *ver; |
b73d6f50 | 867 | s = scan_num(s, &yylval); |
89bfa8cd | 868 | version = yylval.opval; |
dd629d5b GS |
869 | ver = cSVOPx(version)->op_sv; |
870 | if (SvPOK(ver) && !SvNIOK(ver)) { | |
155aba94 | 871 | (void)SvUPGRADE(ver, SVt_PVNV); |
1571675a GS |
872 | SvNVX(ver) = str_to_version(ver); |
873 | SvNOK_on(ver); /* hint that it is a version */ | |
44dcb63b | 874 | } |
89bfa8cd | 875 | } |
876 | } | |
877 | ||
878 | /* NOTE: The parser sees the package name and the VERSION swapped */ | |
3280af22 | 879 | PL_nextval[PL_nexttoke].opval = version; |
4e553d73 | 880 | force_next(WORD); |
89bfa8cd | 881 | |
882 | return (s); | |
883 | } | |
884 | ||
ffb4593c NT |
885 | /* |
886 | * S_tokeq | |
887 | * Tokenize a quoted string passed in as an SV. It finds the next | |
888 | * chunk, up to end of string or a backslash. It may make a new | |
889 | * SV containing that chunk (if HINT_NEW_STRING is on). It also | |
890 | * turns \\ into \. | |
891 | */ | |
892 | ||
76e3520e | 893 | STATIC SV * |
cea2e8a9 | 894 | S_tokeq(pTHX_ SV *sv) |
79072805 LW |
895 | { |
896 | register char *s; | |
897 | register char *send; | |
898 | register char *d; | |
b3ac6de7 IZ |
899 | STRLEN len = 0; |
900 | SV *pv = sv; | |
79072805 LW |
901 | |
902 | if (!SvLEN(sv)) | |
b3ac6de7 | 903 | goto finish; |
79072805 | 904 | |
a0d0e21e | 905 | s = SvPV_force(sv, len); |
21a311ee | 906 | if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1) |
b3ac6de7 | 907 | goto finish; |
463ee0b2 | 908 | send = s + len; |
79072805 LW |
909 | while (s < send && *s != '\\') |
910 | s++; | |
911 | if (s == send) | |
b3ac6de7 | 912 | goto finish; |
79072805 | 913 | d = s; |
3280af22 | 914 | if ( PL_hints & HINT_NEW_STRING ) |
79cb57f6 | 915 | pv = sv_2mortal(newSVpvn(SvPVX(pv), len)); |
79072805 LW |
916 | while (s < send) { |
917 | if (*s == '\\') { | |
a0d0e21e | 918 | if (s + 1 < send && (s[1] == '\\')) |
79072805 LW |
919 | s++; /* all that, just for this */ |
920 | } | |
921 | *d++ = *s++; | |
922 | } | |
923 | *d = '\0'; | |
463ee0b2 | 924 | SvCUR_set(sv, d - SvPVX(sv)); |
b3ac6de7 | 925 | finish: |
3280af22 | 926 | if ( PL_hints & HINT_NEW_STRING ) |
b3ac6de7 | 927 | return new_constant(NULL, 0, "q", sv, pv, "q"); |
79072805 LW |
928 | return sv; |
929 | } | |
930 | ||
ffb4593c NT |
931 | /* |
932 | * Now come three functions related to double-quote context, | |
933 | * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when | |
934 | * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They | |
935 | * interact with PL_lex_state, and create fake ( ... ) argument lists | |
936 | * to handle functions and concatenation. | |
937 | * They assume that whoever calls them will be setting up a fake | |
938 | * join call, because each subthing puts a ',' after it. This lets | |
939 | * "lower \luPpEr" | |
940 | * become | |
941 | * join($, , 'lower ', lcfirst( 'uPpEr', ) ,) | |
942 | * | |
943 | * (I'm not sure whether the spurious commas at the end of lcfirst's | |
944 | * arguments and join's arguments are created or not). | |
945 | */ | |
946 | ||
947 | /* | |
948 | * S_sublex_start | |
949 | * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST). | |
950 | * | |
951 | * Pattern matching will set PL_lex_op to the pattern-matching op to | |
952 | * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise). | |
953 | * | |
954 | * OP_CONST and OP_READLINE are easy--just make the new op and return. | |
955 | * | |
956 | * Everything else becomes a FUNC. | |
957 | * | |
958 | * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we | |
959 | * had an OP_CONST or OP_READLINE). This just sets us up for a | |
960 | * call to S_sublex_push(). | |
961 | */ | |
962 | ||
76e3520e | 963 | STATIC I32 |
cea2e8a9 | 964 | S_sublex_start(pTHX) |
79072805 LW |
965 | { |
966 | register I32 op_type = yylval.ival; | |
79072805 LW |
967 | |
968 | if (op_type == OP_NULL) { | |
3280af22 NIS |
969 | yylval.opval = PL_lex_op; |
970 | PL_lex_op = Nullop; | |
79072805 LW |
971 | return THING; |
972 | } | |
973 | if (op_type == OP_CONST || op_type == OP_READLINE) { | |
3280af22 | 974 | SV *sv = tokeq(PL_lex_stuff); |
b3ac6de7 IZ |
975 | |
976 | if (SvTYPE(sv) == SVt_PVIV) { | |
977 | /* Overloaded constants, nothing fancy: Convert to SVt_PV: */ | |
978 | STRLEN len; | |
979 | char *p; | |
980 | SV *nsv; | |
981 | ||
982 | p = SvPV(sv, len); | |
79cb57f6 | 983 | nsv = newSVpvn(p, len); |
01ec43d0 GS |
984 | if (SvUTF8(sv)) |
985 | SvUTF8_on(nsv); | |
b3ac6de7 IZ |
986 | SvREFCNT_dec(sv); |
987 | sv = nsv; | |
4e553d73 | 988 | } |
b3ac6de7 | 989 | yylval.opval = (OP*)newSVOP(op_type, 0, sv); |
3280af22 | 990 | PL_lex_stuff = Nullsv; |
79072805 LW |
991 | return THING; |
992 | } | |
993 | ||
3280af22 NIS |
994 | PL_sublex_info.super_state = PL_lex_state; |
995 | PL_sublex_info.sub_inwhat = op_type; | |
996 | PL_sublex_info.sub_op = PL_lex_op; | |
997 | PL_lex_state = LEX_INTERPPUSH; | |
55497cff | 998 | |
3280af22 NIS |
999 | PL_expect = XTERM; |
1000 | if (PL_lex_op) { | |
1001 | yylval.opval = PL_lex_op; | |
1002 | PL_lex_op = Nullop; | |
55497cff | 1003 | return PMFUNC; |
1004 | } | |
1005 | else | |
1006 | return FUNC; | |
1007 | } | |
1008 | ||
ffb4593c NT |
1009 | /* |
1010 | * S_sublex_push | |
1011 | * Create a new scope to save the lexing state. The scope will be | |
1012 | * ended in S_sublex_done. Returns a '(', starting the function arguments | |
1013 | * to the uc, lc, etc. found before. | |
1014 | * Sets PL_lex_state to LEX_INTERPCONCAT. | |
1015 | */ | |
1016 | ||
76e3520e | 1017 | STATIC I32 |
cea2e8a9 | 1018 | S_sublex_push(pTHX) |
55497cff | 1019 | { |
f46d017c | 1020 | ENTER; |
55497cff | 1021 | |
3280af22 NIS |
1022 | PL_lex_state = PL_sublex_info.super_state; |
1023 | SAVEI32(PL_lex_dojoin); | |
1024 | SAVEI32(PL_lex_brackets); | |
3280af22 NIS |
1025 | SAVEI32(PL_lex_casemods); |
1026 | SAVEI32(PL_lex_starts); | |
1027 | SAVEI32(PL_lex_state); | |
7766f137 | 1028 | SAVEVPTR(PL_lex_inpat); |
3280af22 | 1029 | SAVEI32(PL_lex_inwhat); |
57843af0 | 1030 | SAVECOPLINE(PL_curcop); |
3280af22 NIS |
1031 | SAVEPPTR(PL_bufptr); |
1032 | SAVEPPTR(PL_oldbufptr); | |
1033 | SAVEPPTR(PL_oldoldbufptr); | |
1034 | SAVEPPTR(PL_linestart); | |
1035 | SAVESPTR(PL_linestr); | |
1036 | SAVEPPTR(PL_lex_brackstack); | |
1037 | SAVEPPTR(PL_lex_casestack); | |
1038 | ||
1039 | PL_linestr = PL_lex_stuff; | |
1040 | PL_lex_stuff = Nullsv; | |
1041 | ||
9cbb5ea2 GS |
1042 | PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart |
1043 | = SvPVX(PL_linestr); | |
3280af22 NIS |
1044 | PL_bufend += SvCUR(PL_linestr); |
1045 | SAVEFREESV(PL_linestr); | |
1046 | ||
1047 | PL_lex_dojoin = FALSE; | |
1048 | PL_lex_brackets = 0; | |
3280af22 NIS |
1049 | New(899, PL_lex_brackstack, 120, char); |
1050 | New(899, PL_lex_casestack, 12, char); | |
1051 | SAVEFREEPV(PL_lex_brackstack); | |
1052 | SAVEFREEPV(PL_lex_casestack); | |
1053 | PL_lex_casemods = 0; | |
1054 | *PL_lex_casestack = '\0'; | |
1055 | PL_lex_starts = 0; | |
1056 | PL_lex_state = LEX_INTERPCONCAT; | |
57843af0 | 1057 | CopLINE_set(PL_curcop, PL_multi_start); |
3280af22 NIS |
1058 | |
1059 | PL_lex_inwhat = PL_sublex_info.sub_inwhat; | |
1060 | if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST) | |
1061 | PL_lex_inpat = PL_sublex_info.sub_op; | |
79072805 | 1062 | else |
3280af22 | 1063 | PL_lex_inpat = Nullop; |
79072805 | 1064 | |
55497cff | 1065 | return '('; |
79072805 LW |
1066 | } |
1067 | ||
ffb4593c NT |
1068 | /* |
1069 | * S_sublex_done | |
1070 | * Restores lexer state after a S_sublex_push. | |
1071 | */ | |
1072 | ||
76e3520e | 1073 | STATIC I32 |
cea2e8a9 | 1074 | S_sublex_done(pTHX) |
79072805 | 1075 | { |
3280af22 | 1076 | if (!PL_lex_starts++) { |
9aa983d2 JH |
1077 | SV *sv = newSVpvn("",0); |
1078 | if (SvUTF8(PL_linestr)) | |
1079 | SvUTF8_on(sv); | |
3280af22 | 1080 | PL_expect = XOPERATOR; |
9aa983d2 | 1081 | yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv); |
79072805 LW |
1082 | return THING; |
1083 | } | |
1084 | ||
3280af22 NIS |
1085 | if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */ |
1086 | PL_lex_state = LEX_INTERPCASEMOD; | |
cea2e8a9 | 1087 | return yylex(); |
79072805 LW |
1088 | } |
1089 | ||
ffb4593c | 1090 | /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */ |
3280af22 NIS |
1091 | if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) { |
1092 | PL_linestr = PL_lex_repl; | |
1093 | PL_lex_inpat = 0; | |
1094 | PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr); | |
1095 | PL_bufend += SvCUR(PL_linestr); | |
1096 | SAVEFREESV(PL_linestr); | |
1097 | PL_lex_dojoin = FALSE; | |
1098 | PL_lex_brackets = 0; | |
3280af22 NIS |
1099 | PL_lex_casemods = 0; |
1100 | *PL_lex_casestack = '\0'; | |
1101 | PL_lex_starts = 0; | |
25da4f38 | 1102 | if (SvEVALED(PL_lex_repl)) { |
3280af22 NIS |
1103 | PL_lex_state = LEX_INTERPNORMAL; |
1104 | PL_lex_starts++; | |
e9fa98b2 HS |
1105 | /* we don't clear PL_lex_repl here, so that we can check later |
1106 | whether this is an evalled subst; that means we rely on the | |
1107 | logic to ensure sublex_done() is called again only via the | |
1108 | branch (in yylex()) that clears PL_lex_repl, else we'll loop */ | |
79072805 | 1109 | } |
e9fa98b2 | 1110 | else { |
3280af22 | 1111 | PL_lex_state = LEX_INTERPCONCAT; |
e9fa98b2 HS |
1112 | PL_lex_repl = Nullsv; |
1113 | } | |
79072805 | 1114 | return ','; |
ffed7fef LW |
1115 | } |
1116 | else { | |
f46d017c | 1117 | LEAVE; |
3280af22 NIS |
1118 | PL_bufend = SvPVX(PL_linestr); |
1119 | PL_bufend += SvCUR(PL_linestr); | |
1120 | PL_expect = XOPERATOR; | |
09bef843 | 1121 | PL_sublex_info.sub_inwhat = 0; |
79072805 | 1122 | return ')'; |
ffed7fef LW |
1123 | } |
1124 | } | |
1125 | ||
02aa26ce NT |
1126 | /* |
1127 | scan_const | |
1128 | ||
1129 | Extracts a pattern, double-quoted string, or transliteration. This | |
1130 | is terrifying code. | |
1131 | ||
3280af22 NIS |
1132 | It looks at lex_inwhat and PL_lex_inpat to find out whether it's |
1133 | processing a pattern (PL_lex_inpat is true), a transliteration | |
02aa26ce NT |
1134 | (lex_inwhat & OP_TRANS is true), or a double-quoted string. |
1135 | ||
9b599b2a GS |
1136 | Returns a pointer to the character scanned up to. Iff this is |
1137 | advanced from the start pointer supplied (ie if anything was | |
1138 | successfully parsed), will leave an OP for the substring scanned | |
1139 | in yylval. Caller must intuit reason for not parsing further | |
1140 | by looking at the next characters herself. | |
1141 | ||
02aa26ce NT |
1142 | In patterns: |
1143 | backslashes: | |
1144 | double-quoted style: \r and \n | |
1145 | regexp special ones: \D \s | |
1146 | constants: \x3 | |
1147 | backrefs: \1 (deprecated in substitution replacements) | |
1148 | case and quoting: \U \Q \E | |
1149 | stops on @ and $, but not for $ as tail anchor | |
1150 | ||
1151 | In transliterations: | |
1152 | characters are VERY literal, except for - not at the start or end | |
1153 | of the string, which indicates a range. scan_const expands the | |
1154 | range to the full set of intermediate characters. | |
1155 | ||
1156 | In double-quoted strings: | |
1157 | backslashes: | |
1158 | double-quoted style: \r and \n | |
1159 | constants: \x3 | |
1160 | backrefs: \1 (deprecated) | |
1161 | case and quoting: \U \Q \E | |
1162 | stops on @ and $ | |
1163 | ||
1164 | scan_const does *not* construct ops to handle interpolated strings. | |
1165 | It stops processing as soon as it finds an embedded $ or @ variable | |
1166 | and leaves it to the caller to work out what's going on. | |
1167 | ||
1168 | @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @:foo. | |
1169 | ||
1170 | $ in pattern could be $foo or could be tail anchor. Assumption: | |
1171 | it's a tail anchor if $ is the last thing in the string, or if it's | |
1172 | followed by one of ")| \n\t" | |
1173 | ||
1174 | \1 (backreferences) are turned into $1 | |
1175 | ||
1176 | The structure of the code is | |
1177 | while (there's a character to process) { | |
1178 | handle transliteration ranges | |
1179 | skip regexp comments | |
1180 | skip # initiated comments in //x patterns | |
1181 | check for embedded @foo | |
1182 | check for embedded scalars | |
1183 | if (backslash) { | |
1184 | leave intact backslashes from leave (below) | |
1185 | deprecate \1 in strings and sub replacements | |
1186 | handle string-changing backslashes \l \U \Q \E, etc. | |
1187 | switch (what was escaped) { | |
1188 | handle - in a transliteration (becomes a literal -) | |
1189 | handle \132 octal characters | |
1190 | handle 0x15 hex characters | |
1191 | handle \cV (control V) | |
1192 | handle printf backslashes (\f, \r, \n, etc) | |
1193 | } (end switch) | |
1194 | } (end if backslash) | |
1195 | } (end while character to read) | |
4e553d73 | 1196 | |
02aa26ce NT |
1197 | */ |
1198 | ||
76e3520e | 1199 | STATIC char * |
cea2e8a9 | 1200 | S_scan_const(pTHX_ char *start) |
79072805 | 1201 | { |
3280af22 | 1202 | register char *send = PL_bufend; /* end of the constant */ |
02aa26ce NT |
1203 | SV *sv = NEWSV(93, send - start); /* sv for the constant */ |
1204 | register char *s = start; /* start of the constant */ | |
1205 | register char *d = SvPVX(sv); /* destination for copies */ | |
1206 | bool dorange = FALSE; /* are we in a translit range? */ | |
c2e66d9e | 1207 | bool didrange = FALSE; /* did we just finish a range? */ |
9aa983d2 JH |
1208 | bool has_utf8 = (PL_linestr && SvUTF8(PL_linestr)); |
1209 | /* the constant is UTF8 */ | |
012bcf8d GS |
1210 | UV uv; |
1211 | ||
ac2262e3 | 1212 | I32 utf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) |
a0ed51b3 LW |
1213 | ? (PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF)) |
1214 | : UTF; | |
89491803 | 1215 | I32 this_utf8 = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) |
9cbb5ea2 GS |
1216 | ? (PL_sublex_info.sub_op->op_private & (PL_lex_repl ? |
1217 | OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF)) | |
a0ed51b3 | 1218 | : UTF; |
dff6d3cd | 1219 | const char *leaveit = /* set of acceptably-backslashed characters */ |
3280af22 | 1220 | PL_lex_inpat |
4a2d328f | 1221 | ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#" |
9b599b2a | 1222 | : ""; |
79072805 LW |
1223 | |
1224 | while (s < send || dorange) { | |
02aa26ce | 1225 | /* get transliterations out of the way (they're most literal) */ |
3280af22 | 1226 | if (PL_lex_inwhat == OP_TRANS) { |
02aa26ce | 1227 | /* expand a range A-Z to the full set of characters. AIE! */ |
79072805 | 1228 | if (dorange) { |
02aa26ce | 1229 | I32 i; /* current expanded character */ |
8ada0baa | 1230 | I32 min; /* first character in range */ |
02aa26ce NT |
1231 | I32 max; /* last character in range */ |
1232 | ||
1233 | i = d - SvPVX(sv); /* remember current offset */ | |
9cbb5ea2 GS |
1234 | SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */ |
1235 | d = SvPVX(sv) + i; /* refresh d after realloc */ | |
02aa26ce NT |
1236 | d -= 2; /* eat the first char and the - */ |
1237 | ||
8ada0baa JH |
1238 | min = (U8)*d; /* first char in range */ |
1239 | max = (U8)d[1]; /* last char in range */ | |
1240 | ||
c2e66d9e | 1241 | if (min > max) { |
01ec43d0 GS |
1242 | Perl_croak(aTHX_ |
1243 | "Invalid [] range \"%c-%c\" in transliteration operator", | |
d2560b70 | 1244 | (char)min, (char)max); |
c2e66d9e GS |
1245 | } |
1246 | ||
8ada0baa JH |
1247 | #ifndef ASCIIish |
1248 | if ((isLOWER(min) && isLOWER(max)) || | |
1249 | (isUPPER(min) && isUPPER(max))) { | |
1250 | if (isLOWER(min)) { | |
1251 | for (i = min; i <= max; i++) | |
1252 | if (isLOWER(i)) | |
1253 | *d++ = i; | |
1254 | } else { | |
1255 | for (i = min; i <= max; i++) | |
1256 | if (isUPPER(i)) | |
1257 | *d++ = i; | |
1258 | } | |
1259 | } | |
1260 | else | |
1261 | #endif | |
1262 | for (i = min; i <= max; i++) | |
1263 | *d++ = i; | |
02aa26ce NT |
1264 | |
1265 | /* mark the range as done, and continue */ | |
79072805 | 1266 | dorange = FALSE; |
01ec43d0 | 1267 | didrange = TRUE; |
79072805 | 1268 | continue; |
4e553d73 | 1269 | } |
02aa26ce NT |
1270 | |
1271 | /* range begins (ignore - as first or last char) */ | |
79072805 | 1272 | else if (*s == '-' && s+1 < send && s != start) { |
4e553d73 | 1273 | if (didrange) { |
1fafa243 | 1274 | Perl_croak(aTHX_ "Ambiguous range in transliteration operator"); |
01ec43d0 | 1275 | } |
a0ed51b3 | 1276 | if (utf) { |
a176fa2a | 1277 | *d++ = (char)0xff; /* use illegal utf8 byte--see pmtrans */ |
a0ed51b3 LW |
1278 | s++; |
1279 | continue; | |
1280 | } | |
79072805 LW |
1281 | dorange = TRUE; |
1282 | s++; | |
01ec43d0 GS |
1283 | } |
1284 | else { | |
1285 | didrange = FALSE; | |
1286 | } | |
79072805 | 1287 | } |
02aa26ce NT |
1288 | |
1289 | /* if we get here, we're not doing a transliteration */ | |
1290 | ||
0f5d15d6 IZ |
1291 | /* skip for regexp comments /(?#comment)/ and code /(?{code})/, |
1292 | except for the last char, which will be done separately. */ | |
3280af22 | 1293 | else if (*s == '(' && PL_lex_inpat && s[1] == '?') { |
cc6b7395 IZ |
1294 | if (s[2] == '#') { |
1295 | while (s < send && *s != ')') | |
1296 | *d++ = *s++; | |
155aba94 GS |
1297 | } |
1298 | else if (s[2] == '{' /* This should match regcomp.c */ | |
1299 | || ((s[2] == 'p' || s[2] == '?') && s[3] == '{')) | |
1300 | { | |
cc6b7395 | 1301 | I32 count = 1; |
0f5d15d6 | 1302 | char *regparse = s + (s[2] == '{' ? 3 : 4); |
cc6b7395 IZ |
1303 | char c; |
1304 | ||
d9f97599 GS |
1305 | while (count && (c = *regparse)) { |
1306 | if (c == '\\' && regparse[1]) | |
1307 | regparse++; | |
4e553d73 | 1308 | else if (c == '{') |
cc6b7395 | 1309 | count++; |
4e553d73 | 1310 | else if (c == '}') |
cc6b7395 | 1311 | count--; |
d9f97599 | 1312 | regparse++; |
cc6b7395 | 1313 | } |
5bdf89e7 IZ |
1314 | if (*regparse != ')') { |
1315 | regparse--; /* Leave one char for continuation. */ | |
cc6b7395 | 1316 | yyerror("Sequence (?{...}) not terminated or not {}-balanced"); |
5bdf89e7 | 1317 | } |
0f5d15d6 | 1318 | while (s < regparse) |
cc6b7395 IZ |
1319 | *d++ = *s++; |
1320 | } | |
748a9306 | 1321 | } |
02aa26ce NT |
1322 | |
1323 | /* likewise skip #-initiated comments in //x patterns */ | |
3280af22 NIS |
1324 | else if (*s == '#' && PL_lex_inpat && |
1325 | ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) { | |
748a9306 LW |
1326 | while (s+1 < send && *s != '\n') |
1327 | *d++ = *s++; | |
1328 | } | |
02aa26ce | 1329 | |
5d1d4326 JH |
1330 | /* check for embedded arrays |
1331 | (@foo, @:foo, @'foo, @{foo}, @$foo, @+, @-) | |
1332 | */ | |
7e2040f0 | 1333 | else if (*s == '@' && s[1] |
5d1d4326 | 1334 | && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$+-", s[1]))) |
79072805 | 1335 | break; |
02aa26ce NT |
1336 | |
1337 | /* check for embedded scalars. only stop if we're sure it's a | |
1338 | variable. | |
1339 | */ | |
79072805 | 1340 | else if (*s == '$') { |
3280af22 | 1341 | if (!PL_lex_inpat) /* not a regexp, so $ must be var */ |
79072805 | 1342 | break; |
c277df42 | 1343 | if (s + 1 < send && !strchr("()| \n\t", s[1])) |
79072805 LW |
1344 | break; /* in regexp, $ might be tail anchor */ |
1345 | } | |
02aa26ce NT |
1346 | |
1347 | /* backslashes */ | |
79072805 LW |
1348 | if (*s == '\\' && s+1 < send) { |
1349 | s++; | |
02aa26ce NT |
1350 | |
1351 | /* some backslashes we leave behind */ | |
c9f97d15 | 1352 | if (*leaveit && *s && strchr(leaveit, *s)) { |
79072805 LW |
1353 | *d++ = '\\'; |
1354 | *d++ = *s++; | |
1355 | continue; | |
1356 | } | |
02aa26ce NT |
1357 | |
1358 | /* deprecate \1 in strings and substitution replacements */ | |
3280af22 | 1359 | if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat && |
a0d0e21e | 1360 | isDIGIT(*s) && *s != '0' && !isDIGIT(s[1])) |
79072805 | 1361 | { |
599cee73 | 1362 | if (ckWARN(WARN_SYNTAX)) |
cea2e8a9 | 1363 | Perl_warner(aTHX_ WARN_SYNTAX, "\\%c better written as $%c", *s, *s); |
79072805 LW |
1364 | *--s = '$'; |
1365 | break; | |
1366 | } | |
02aa26ce NT |
1367 | |
1368 | /* string-change backslash escapes */ | |
3280af22 | 1369 | if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) { |
79072805 LW |
1370 | --s; |
1371 | break; | |
1372 | } | |
02aa26ce NT |
1373 | |
1374 | /* if we get here, it's either a quoted -, or a digit */ | |
79072805 | 1375 | switch (*s) { |
02aa26ce NT |
1376 | |
1377 | /* quoted - in transliterations */ | |
79072805 | 1378 | case '-': |
3280af22 | 1379 | if (PL_lex_inwhat == OP_TRANS) { |
79072805 LW |
1380 | *d++ = *s++; |
1381 | continue; | |
1382 | } | |
1383 | /* FALL THROUGH */ | |
1384 | default: | |
11b8faa4 | 1385 | { |
7e84c16c | 1386 | if (ckWARN(WARN_MISC) && isALNUM(*s)) |
4e553d73 | 1387 | Perl_warner(aTHX_ WARN_MISC, |
11b8faa4 JH |
1388 | "Unrecognized escape \\%c passed through", |
1389 | *s); | |
1390 | /* default action is to copy the quoted character */ | |
f9a63242 | 1391 | goto default_action; |
11b8faa4 | 1392 | } |
02aa26ce NT |
1393 | |
1394 | /* \132 indicates an octal constant */ | |
79072805 LW |
1395 | case '0': case '1': case '2': case '3': |
1396 | case '4': case '5': case '6': case '7': | |
ba210ebe JH |
1397 | { |
1398 | STRLEN len = 0; /* disallow underscores */ | |
1399 | uv = (UV)scan_oct(s, 3, &len); | |
1400 | s += len; | |
1401 | } | |
012bcf8d | 1402 | goto NUM_ESCAPE_INSERT; |
02aa26ce NT |
1403 | |
1404 | /* \x24 indicates a hex constant */ | |
79072805 | 1405 | case 'x': |
a0ed51b3 LW |
1406 | ++s; |
1407 | if (*s == '{') { | |
1408 | char* e = strchr(s, '}'); | |
adaeee49 | 1409 | if (!e) { |
a0ed51b3 | 1410 | yyerror("Missing right brace on \\x{}"); |
adaeee49 GA |
1411 | e = s; |
1412 | } | |
89491803 | 1413 | else { |
ba210ebe JH |
1414 | STRLEN len = 1; /* allow underscores */ |
1415 | uv = (UV)scan_hex(s + 1, e - s - 1, &len); | |
1416 | } | |
1417 | s = e + 1; | |
a0ed51b3 LW |
1418 | } |
1419 | else { | |
ba210ebe JH |
1420 | { |
1421 | STRLEN len = 0; /* disallow underscores */ | |
1422 | uv = (UV)scan_hex(s, 2, &len); | |
1423 | s += len; | |
1424 | } | |
012bcf8d GS |
1425 | } |
1426 | ||
1427 | NUM_ESCAPE_INSERT: | |
1428 | /* Insert oct or hex escaped character. | |
301d3d20 JH |
1429 | * There will always enough room in sv since such |
1430 | * escapes will be longer than any UT-F8 sequence | |
1431 | * they can end up as. */ | |
ba7cea30 JH |
1432 | |
1433 | /* This spot is wrong for EBCDIC. Characters like | |
1434 | * the lowercase letters and digits are >127 in EBCDIC, | |
1435 | * so here they would need to be mapped to the Unicode | |
1436 | * repertoire. --jhi */ | |
1437 | ||
7948272d | 1438 | if (uv > 127) { |
9aa983d2 | 1439 | if (!has_utf8 && uv > 255) { |
301d3d20 JH |
1440 | /* Might need to recode whatever we have |
1441 | * accumulated so far if it contains any | |
1442 | * hibit chars. | |
1443 | * | |
1444 | * (Can't we keep track of that and avoid | |
1445 | * this rescan? --jhi) | |
012bcf8d GS |
1446 | */ |
1447 | int hicount = 0; | |
1448 | char *c; | |
301d3d20 | 1449 | |
012bcf8d | 1450 | for (c = SvPVX(sv); c < d; c++) { |
fd400ab9 | 1451 | if (UTF8_IS_CONTINUED(*c)) |
012bcf8d GS |
1452 | hicount++; |
1453 | } | |
1454 | if (hicount) { | |
1455 | char *old_pvx = SvPVX(sv); | |
1456 | char *src, *dst; | |
301d3d20 JH |
1457 | |
1458 | d = SvGROW(sv, | |
1459 | SvCUR(sv) + hicount + 1) + | |
1460 | (d - old_pvx); | |
012bcf8d GS |
1461 | |
1462 | src = d - 1; | |
1463 | d += hicount; | |
1464 | dst = d - 1; | |
1465 | ||
1466 | while (src < dst) { | |
fd400ab9 | 1467 | if (UTF8_IS_CONTINUED(*src)) { |
9b877dbb IH |
1468 | *dst-- = UTF8_EIGHT_BIT_LO(*src); |
1469 | *dst-- = UTF8_EIGHT_BIT_HI(*src--); | |
012bcf8d GS |
1470 | } |
1471 | else { | |
1472 | *dst-- = *src--; | |
1473 | } | |
1474 | } | |
1475 | } | |
1476 | } | |
1477 | ||
9aa983d2 | 1478 | if (has_utf8 || uv > 255) { |
012bcf8d | 1479 | d = (char*)uv_to_utf8((U8*)d, uv); |
4e553d73 | 1480 | has_utf8 = TRUE; |
f9a63242 JH |
1481 | if (PL_lex_inwhat == OP_TRANS && |
1482 | PL_sublex_info.sub_op) { | |
1483 | PL_sublex_info.sub_op->op_private |= | |
1484 | (PL_lex_repl ? OPpTRANS_FROM_UTF | |
1485 | : OPpTRANS_TO_UTF); | |
1486 | utf = TRUE; | |
1487 | } | |
012bcf8d | 1488 | } |
a0ed51b3 | 1489 | else { |
012bcf8d | 1490 | *d++ = (char)uv; |
a0ed51b3 | 1491 | } |
012bcf8d GS |
1492 | } |
1493 | else { | |
1494 | *d++ = (char)uv; | |
a0ed51b3 | 1495 | } |
79072805 | 1496 | continue; |
02aa26ce | 1497 | |
4a2d328f IZ |
1498 | /* \N{latin small letter a} is a named character */ |
1499 | case 'N': | |
423cee85 JH |
1500 | ++s; |
1501 | if (*s == '{') { | |
1502 | char* e = strchr(s, '}'); | |
155aba94 | 1503 | SV *res; |
423cee85 JH |
1504 | STRLEN len; |
1505 | char *str; | |
4e553d73 | 1506 | |
423cee85 | 1507 | if (!e) { |
5777a3f7 | 1508 | yyerror("Missing right brace on \\N{}"); |
423cee85 JH |
1509 | e = s - 1; |
1510 | goto cont_scan; | |
1511 | } | |
1512 | res = newSVpvn(s + 1, e - s - 1); | |
4e553d73 | 1513 | res = new_constant( Nullch, 0, "charnames", |
5777a3f7 | 1514 | res, Nullsv, "\\N{...}" ); |
f9a63242 JH |
1515 | if (has_utf8) |
1516 | sv_utf8_upgrade(res); | |
423cee85 | 1517 | str = SvPV(res,len); |
89491803 | 1518 | if (!has_utf8 && SvUTF8(res)) { |
f08d6ad9 GS |
1519 | char *ostart = SvPVX(sv); |
1520 | SvCUR_set(sv, d - ostart); | |
1521 | SvPOK_on(sv); | |
e4f3eed8 | 1522 | *d = '\0'; |
f08d6ad9 | 1523 | sv_utf8_upgrade(sv); |
d2f449dd SB |
1524 | /* this just broke our allocation above... */ |
1525 | SvGROW(sv, send - start); | |
f08d6ad9 | 1526 | d = SvPVX(sv) + SvCUR(sv); |
89491803 | 1527 | has_utf8 = TRUE; |
f08d6ad9 | 1528 | } |
423cee85 JH |
1529 | if (len > e - s + 4) { |
1530 | char *odest = SvPVX(sv); | |
1531 | ||
1532 | SvGROW(sv, (SvCUR(sv) + len - (e - s + 4))); | |
1533 | d = SvPVX(sv) + (d - odest); | |
1534 | } | |
1535 | Copy(str, d, len, char); | |
1536 | d += len; | |
1537 | SvREFCNT_dec(res); | |
1538 | cont_scan: | |
1539 | s = e + 1; | |
1540 | } | |
1541 | else | |
5777a3f7 | 1542 | yyerror("Missing braces on \\N{}"); |
423cee85 JH |
1543 | continue; |
1544 | ||
02aa26ce | 1545 | /* \c is a control character */ |
79072805 LW |
1546 | case 'c': |
1547 | s++; | |
9d116dd7 JH |
1548 | #ifdef EBCDIC |
1549 | *d = *s++; | |
1550 | if (isLOWER(*d)) | |
1551 | *d = toUPPER(*d); | |
4e553d73 | 1552 | *d = toCTRL(*d); |
774a9426 | 1553 | d++; |
9d116dd7 | 1554 | #else |
ba210ebe JH |
1555 | { |
1556 | U8 c = *s++; | |
1557 | *d++ = toCTRL(c); | |
1558 | } | |
9d116dd7 | 1559 | #endif |
79072805 | 1560 | continue; |
02aa26ce NT |
1561 | |
1562 | /* printf-style backslashes, formfeeds, newlines, etc */ | |
79072805 LW |
1563 | case 'b': |
1564 | *d++ = '\b'; | |
1565 | break; | |
1566 | case 'n': | |
1567 | *d++ = '\n'; | |
1568 | break; | |
1569 | case 'r': | |
1570 | *d++ = '\r'; | |
1571 | break; | |
1572 | case 'f': | |
1573 | *d++ = '\f'; | |
1574 | break; | |
1575 | case 't': | |
1576 | *d++ = '\t'; | |
1577 | break; | |
34a3fe2a PP |
1578 | #ifdef EBCDIC |
1579 | case 'e': | |
1580 | *d++ = '\047'; /* CP 1047 */ | |
1581 | break; | |
1582 | case 'a': | |
1583 | *d++ = '\057'; /* CP 1047 */ | |
1584 | break; | |
1585 | #else | |
79072805 LW |
1586 | case 'e': |
1587 | *d++ = '\033'; | |
1588 | break; | |
1589 | case 'a': | |
1590 | *d++ = '\007'; | |
1591 | break; | |
34a3fe2a | 1592 | #endif |
02aa26ce NT |
1593 | } /* end switch */ |
1594 | ||
79072805 LW |
1595 | s++; |
1596 | continue; | |
02aa26ce NT |
1597 | } /* end if (backslash) */ |
1598 | ||
f9a63242 | 1599 | default_action: |
fd400ab9 | 1600 | if (UTF8_IS_CONTINUED(*s) && (this_utf8 || has_utf8)) { |
a5a960be IRC |
1601 | STRLEN len = (STRLEN) -1; |
1602 | UV uv; | |
1603 | if (this_utf8) { | |
a0dbb045 | 1604 | uv = utf8_to_uv((U8*)s, send - s, &len, 0); |
a5a960be IRC |
1605 | } |
1606 | if (len == (STRLEN)-1) { | |
1607 | /* Illegal UTF8 (a high-bit byte), make it valid. */ | |
1608 | char *old_pvx = SvPVX(sv); | |
1609 | /* need space for one extra char (NOTE: SvCUR() not set here) */ | |
1610 | d = SvGROW(sv, SvLEN(sv) + 1) + (d - old_pvx); | |
1611 | d = (char*)uv_to_utf8((U8*)d, (U8)*s++); | |
1612 | } | |
1613 | else { | |
1614 | while (len--) | |
1615 | *d++ = *s++; | |
1616 | } | |
1617 | has_utf8 = TRUE; | |
f9a63242 JH |
1618 | if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) { |
1619 | PL_sublex_info.sub_op->op_private |= | |
1620 | (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF); | |
1621 | utf = TRUE; | |
1622 | } | |
a5a960be IRC |
1623 | continue; |
1624 | } | |
1625 | ||
f9a63242 | 1626 | *d++ = *s++; |
02aa26ce NT |
1627 | } /* while loop to process each character */ |
1628 | ||
1629 | /* terminate the string and set up the sv */ | |
79072805 | 1630 | *d = '\0'; |
463ee0b2 | 1631 | SvCUR_set(sv, d - SvPVX(sv)); |
79072805 | 1632 | SvPOK_on(sv); |
89491803 | 1633 | if (has_utf8) |
7e2040f0 | 1634 | SvUTF8_on(sv); |
79072805 | 1635 | |
02aa26ce | 1636 | /* shrink the sv if we allocated more than we used */ |
79072805 LW |
1637 | if (SvCUR(sv) + 5 < SvLEN(sv)) { |
1638 | SvLEN_set(sv, SvCUR(sv) + 1); | |
463ee0b2 | 1639 | Renew(SvPVX(sv), SvLEN(sv), char); |
79072805 | 1640 | } |
02aa26ce | 1641 | |
9b599b2a | 1642 | /* return the substring (via yylval) only if we parsed anything */ |
3280af22 NIS |
1643 | if (s > PL_bufptr) { |
1644 | if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) ) | |
4e553d73 | 1645 | sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"), |
b3ac6de7 | 1646 | sv, Nullsv, |
4e553d73 | 1647 | ( PL_lex_inwhat == OP_TRANS |
b3ac6de7 | 1648 | ? "tr" |
3280af22 | 1649 | : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat) |
b3ac6de7 IZ |
1650 | ? "s" |
1651 | : "qq"))); | |
79072805 | 1652 | yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv); |
b3ac6de7 | 1653 | } else |
8990e307 | 1654 | SvREFCNT_dec(sv); |
79072805 LW |
1655 | return s; |
1656 | } | |
1657 | ||
ffb4593c NT |
1658 | /* S_intuit_more |
1659 | * Returns TRUE if there's more to the expression (e.g., a subscript), | |
1660 | * FALSE otherwise. | |
ffb4593c NT |
1661 | * |
1662 | * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/ | |
1663 | * | |
1664 | * ->[ and ->{ return TRUE | |
1665 | * { and [ outside a pattern are always subscripts, so return TRUE | |
1666 | * if we're outside a pattern and it's not { or [, then return FALSE | |
1667 | * if we're in a pattern and the first char is a { | |
1668 | * {4,5} (any digits around the comma) returns FALSE | |
1669 | * if we're in a pattern and the first char is a [ | |
1670 | * [] returns FALSE | |
1671 | * [SOMETHING] has a funky algorithm to decide whether it's a | |
1672 | * character class or not. It has to deal with things like | |
1673 | * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/ | |
1674 | * anything else returns TRUE | |
1675 | */ | |
1676 | ||
9cbb5ea2 GS |
1677 | /* This is the one truly awful dwimmer necessary to conflate C and sed. */ |
1678 | ||
76e3520e | 1679 | STATIC int |
cea2e8a9 | 1680 | S_intuit_more(pTHX_ register char *s) |
79072805 | 1681 | { |
3280af22 | 1682 | if (PL_lex_brackets) |
79072805 LW |
1683 | return TRUE; |
1684 | if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{')) | |
1685 | return TRUE; | |
1686 | if (*s != '{' && *s != '[') | |
1687 | return FALSE; | |
3280af22 | 1688 | if (!PL_lex_inpat) |
79072805 LW |
1689 | return TRUE; |
1690 | ||
1691 | /* In a pattern, so maybe we have {n,m}. */ | |
1692 | if (*s == '{') { | |
1693 | s++; | |
1694 | if (!isDIGIT(*s)) | |
1695 | return TRUE; | |
1696 | while (isDIGIT(*s)) | |
1697 | s++; | |
1698 | if (*s == ',') | |
1699 | s++; | |
1700 | while (isDIGIT(*s)) | |
1701 | s++; | |
1702 | if (*s == '}') | |
1703 | return FALSE; | |
1704 | return TRUE; | |
1705 | ||
1706 | } | |
1707 | ||
1708 | /* On the other hand, maybe we have a character class */ | |
1709 | ||
1710 | s++; | |
1711 | if (*s == ']' || *s == '^') | |
1712 | return FALSE; | |
1713 | else { | |
ffb4593c | 1714 | /* this is terrifying, and it works */ |
79072805 LW |
1715 | int weight = 2; /* let's weigh the evidence */ |
1716 | char seen[256]; | |
f27ffc4a | 1717 | unsigned char un_char = 255, last_un_char; |
93a17b20 | 1718 | char *send = strchr(s,']'); |
3280af22 | 1719 | char tmpbuf[sizeof PL_tokenbuf * 4]; |
79072805 LW |
1720 | |
1721 | if (!send) /* has to be an expression */ | |
1722 | return TRUE; | |
1723 | ||
1724 | Zero(seen,256,char); | |
1725 | if (*s == '$') | |
1726 | weight -= 3; | |
1727 | else if (isDIGIT(*s)) { | |
1728 | if (s[1] != ']') { | |
1729 | if (isDIGIT(s[1]) && s[2] == ']') | |
1730 | weight -= 10; | |
1731 | } | |
1732 | else | |
1733 | weight -= 100; | |
1734 | } | |
1735 | for (; s < send; s++) { | |
1736 | last_un_char = un_char; | |
1737 | un_char = (unsigned char)*s; | |
1738 | switch (*s) { | |
1739 | case '@': | |
1740 | case '&': | |
1741 | case '$': | |
1742 | weight -= seen[un_char] * 10; | |
7e2040f0 | 1743 | if (isALNUM_lazy_if(s+1,UTF)) { |
8903cb82 | 1744 | scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE); |
a0d0e21e | 1745 | if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV)) |
79072805 LW |
1746 | weight -= 100; |
1747 | else | |
1748 | weight -= 10; | |
1749 | } | |
1750 | else if (*s == '$' && s[1] && | |
93a17b20 LW |
1751 | strchr("[#!%*<>()-=",s[1])) { |
1752 | if (/*{*/ strchr("])} =",s[2])) | |
79072805 LW |
1753 | weight -= 10; |
1754 | else | |
1755 | weight -= 1; | |
1756 | } | |
1757 | break; | |
1758 | case '\\': | |
1759 | un_char = 254; | |
1760 | if (s[1]) { | |
93a17b20 | 1761 | if (strchr("wds]",s[1])) |
79072805 LW |
1762 | weight += 100; |
1763 | else if (seen['\''] || seen['"']) | |
1764 | weight += 1; | |
93a17b20 | 1765 | else if (strchr("rnftbxcav",s[1])) |
79072805 LW |
1766 | weight += 40; |
1767 | else if (isDIGIT(s[1])) { | |
1768 | weight += 40; | |
1769 | while (s[1] && isDIGIT(s[1])) | |
1770 | s++; | |
1771 | } | |
1772 | } | |
1773 | else | |
1774 | weight += 100; | |
1775 | break; | |
1776 | case '-': | |
1777 | if (s[1] == '\\') | |
1778 | weight += 50; | |
93a17b20 | 1779 | if (strchr("aA01! ",last_un_char)) |
79072805 | 1780 | weight += 30; |
93a17b20 | 1781 | if (strchr("zZ79~",s[1])) |
79072805 | 1782 | weight += 30; |
f27ffc4a GS |
1783 | if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$')) |
1784 | weight -= 5; /* cope with negative subscript */ | |
79072805 LW |
1785 | break; |
1786 | default: | |
93a17b20 | 1787 | if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) && |
79072805 LW |
1788 | isALPHA(*s) && s[1] && isALPHA(s[1])) { |
1789 | char *d = tmpbuf; | |
1790 | while (isALPHA(*s)) | |
1791 | *d++ = *s++; | |
1792 | *d = '\0'; | |
1793 | if (keyword(tmpbuf, d - tmpbuf)) | |
1794 | weight -= 150; | |
1795 | } | |
1796 | if (un_char == last_un_char + 1) | |
1797 | weight += 5; | |
1798 | weight -= seen[un_char]; | |
1799 | break; | |
1800 | } | |
1801 | seen[un_char]++; | |
1802 | } | |
1803 | if (weight >= 0) /* probably a character class */ | |
1804 | return FALSE; | |
1805 | } | |
1806 | ||
1807 | return TRUE; | |
1808 | } | |
ffed7fef | 1809 | |
ffb4593c NT |
1810 | /* |
1811 | * S_intuit_method | |
1812 | * | |
1813 | * Does all the checking to disambiguate | |
1814 | * foo bar | |
1815 | * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise | |
1816 | * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args). | |
1817 | * | |
1818 | * First argument is the stuff after the first token, e.g. "bar". | |
1819 | * | |
1820 | * Not a method if bar is a filehandle. | |
1821 | * Not a method if foo is a subroutine prototyped to take a filehandle. | |
1822 | * Not a method if it's really "Foo $bar" | |
1823 | * Method if it's "foo $bar" | |
1824 | * Not a method if it's really "print foo $bar" | |
1825 | * Method if it's really "foo package::" (interpreted as package->foo) | |
1826 | * Not a method if bar is known to be a subroutne ("sub bar; foo bar") | |
3cb0bbe5 | 1827 | * Not a method if bar is a filehandle or package, but is quoted with |
ffb4593c NT |
1828 | * => |
1829 | */ | |
1830 | ||
76e3520e | 1831 | STATIC int |
cea2e8a9 | 1832 | S_intuit_method(pTHX_ char *start, GV *gv) |
a0d0e21e LW |
1833 | { |
1834 | char *s = start + (*start == '$'); | |
3280af22 | 1835 | char tmpbuf[sizeof PL_tokenbuf]; |
a0d0e21e LW |
1836 | STRLEN len; |
1837 | GV* indirgv; | |
1838 | ||
1839 | if (gv) { | |
b6c543e3 | 1840 | CV *cv; |
a0d0e21e LW |
1841 | if (GvIO(gv)) |
1842 | return 0; | |
b6c543e3 IZ |
1843 | if ((cv = GvCVu(gv))) { |
1844 | char *proto = SvPVX(cv); | |
1845 | if (proto) { | |
1846 | if (*proto == ';') | |
1847 | proto++; | |
1848 | if (*proto == '*') | |
1849 | return 0; | |
1850 | } | |
1851 | } else | |
a0d0e21e LW |
1852 | gv = 0; |
1853 | } | |
8903cb82 | 1854 | s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len); |
ffb4593c NT |
1855 | /* start is the beginning of the possible filehandle/object, |
1856 | * and s is the end of it | |
1857 | * tmpbuf is a copy of it | |
1858 | */ | |
1859 | ||
a0d0e21e | 1860 | if (*start == '$') { |
3280af22 | 1861 | if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf)) |
a0d0e21e LW |
1862 | return 0; |
1863 | s = skipspace(s); | |
3280af22 NIS |
1864 | PL_bufptr = start; |
1865 | PL_expect = XREF; | |
a0d0e21e LW |
1866 | return *s == '(' ? FUNCMETH : METHOD; |
1867 | } | |
1868 | if (!keyword(tmpbuf, len)) { | |
c3e0f903 GS |
1869 | if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') { |
1870 | len -= 2; | |
1871 | tmpbuf[len] = '\0'; | |
1872 | goto bare_package; | |
1873 | } | |
1874 | indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV); | |
8ebc5c01 | 1875 | if (indirgv && GvCVu(indirgv)) |
a0d0e21e LW |
1876 | return 0; |
1877 | /* filehandle or package name makes it a method */ | |
89bfa8cd | 1878 | if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) { |
a0d0e21e | 1879 | s = skipspace(s); |
3280af22 | 1880 | if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>') |
55497cff | 1881 | return 0; /* no assumptions -- "=>" quotes bearword */ |
c3e0f903 | 1882 | bare_package: |
3280af22 | 1883 | PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, |
79cb57f6 | 1884 | newSVpvn(tmpbuf,len)); |
3280af22 NIS |
1885 | PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE; |
1886 | PL_expect = XTERM; | |
a0d0e21e | 1887 | force_next(WORD); |
3280af22 | 1888 | PL_bufptr = s; |
a0d0e21e LW |
1889 | return *s == '(' ? FUNCMETH : METHOD; |
1890 | } | |
1891 | } | |
1892 | return 0; | |
1893 | } | |
1894 | ||
ffb4593c NT |
1895 | /* |
1896 | * S_incl_perldb | |
1897 | * Return a string of Perl code to load the debugger. If PERL5DB | |
1898 | * is set, it will return the contents of that, otherwise a | |
1899 | * compile-time require of perl5db.pl. | |
1900 | */ | |
1901 | ||
76e3520e | 1902 | STATIC char* |
cea2e8a9 | 1903 | S_incl_perldb(pTHX) |
a0d0e21e | 1904 | { |
3280af22 | 1905 | if (PL_perldb) { |
76e3520e | 1906 | char *pdb = PerlEnv_getenv("PERL5DB"); |
a0d0e21e LW |
1907 | |
1908 | if (pdb) | |
1909 | return pdb; | |
61bb5906 | 1910 | SETERRNO(0,SS$_NORMAL); |
a0d0e21e LW |
1911 | return "BEGIN { require 'perl5db.pl' }"; |
1912 | } | |
1913 | return ""; | |
1914 | } | |
1915 | ||
1916 | ||
16d20bd9 | 1917 | /* Encoded script support. filter_add() effectively inserts a |
4e553d73 | 1918 | * 'pre-processing' function into the current source input stream. |
16d20bd9 AD |
1919 | * Note that the filter function only applies to the current source file |
1920 | * (e.g., it will not affect files 'require'd or 'use'd by this one). | |
1921 | * | |
1922 | * The datasv parameter (which may be NULL) can be used to pass | |
1923 | * private data to this instance of the filter. The filter function | |
1924 | * can recover the SV using the FILTER_DATA macro and use it to | |
1925 | * store private buffers and state information. | |
1926 | * | |
1927 | * The supplied datasv parameter is upgraded to a PVIO type | |
4755096e | 1928 | * and the IoDIRP/IoANY field is used to store the function pointer, |
e0c19803 | 1929 | * and IOf_FAKE_DIRP is enabled on datasv to mark this as such. |
16d20bd9 AD |
1930 | * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for |
1931 | * private use must be set using malloc'd pointers. | |
1932 | */ | |
16d20bd9 AD |
1933 | |
1934 | SV * | |
864dbfa3 | 1935 | Perl_filter_add(pTHX_ filter_t funcp, SV *datasv) |
16d20bd9 | 1936 | { |
f4c556ac GS |
1937 | if (!funcp) |
1938 | return Nullsv; | |
1939 | ||
3280af22 NIS |
1940 | if (!PL_rsfp_filters) |
1941 | PL_rsfp_filters = newAV(); | |
16d20bd9 | 1942 | if (!datasv) |
8c52afec | 1943 | datasv = NEWSV(255,0); |
16d20bd9 | 1944 | if (!SvUPGRADE(datasv, SVt_PVIO)) |
cea2e8a9 | 1945 | Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO"); |
4755096e | 1946 | IoANY(datasv) = (void *)funcp; /* stash funcp into spare field */ |
e0c19803 | 1947 | IoFLAGS(datasv) |= IOf_FAKE_DIRP; |
f4c556ac GS |
1948 | DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n", |
1949 | funcp, SvPV_nolen(datasv))); | |
3280af22 NIS |
1950 | av_unshift(PL_rsfp_filters, 1); |
1951 | av_store(PL_rsfp_filters, 0, datasv) ; | |
16d20bd9 AD |
1952 | return(datasv); |
1953 | } | |
4e553d73 | 1954 | |
16d20bd9 AD |
1955 | |
1956 | /* Delete most recently added instance of this filter function. */ | |
a0d0e21e | 1957 | void |
864dbfa3 | 1958 | Perl_filter_del(pTHX_ filter_t funcp) |
16d20bd9 | 1959 | { |
e0c19803 | 1960 | SV *datasv; |
f4c556ac | 1961 | DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp)); |
3280af22 | 1962 | if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0) |
16d20bd9 AD |
1963 | return; |
1964 | /* if filter is on top of stack (usual case) just pop it off */ | |
e0c19803 | 1965 | datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters)); |
4755096e | 1966 | if (IoANY(datasv) == (void *)funcp) { |
e0c19803 | 1967 | IoFLAGS(datasv) &= ~IOf_FAKE_DIRP; |
4755096e | 1968 | IoANY(datasv) = (void *)NULL; |
3280af22 | 1969 | sv_free(av_pop(PL_rsfp_filters)); |
e50aee73 | 1970 | |
16d20bd9 AD |
1971 | return; |
1972 | } | |
1973 | /* we need to search for the correct entry and clear it */ | |
cea2e8a9 | 1974 | Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)"); |
16d20bd9 AD |
1975 | } |
1976 | ||
1977 | ||
1978 | /* Invoke the n'th filter function for the current rsfp. */ | |
1979 | I32 | |
864dbfa3 | 1980 | Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen) |
4e553d73 NIS |
1981 | |
1982 | ||
8ac85365 | 1983 | /* 0 = read one text line */ |
a0d0e21e | 1984 | { |
16d20bd9 AD |
1985 | filter_t funcp; |
1986 | SV *datasv = NULL; | |
e50aee73 | 1987 | |
3280af22 | 1988 | if (!PL_rsfp_filters) |
16d20bd9 | 1989 | return -1; |
3280af22 | 1990 | if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */ |
16d20bd9 AD |
1991 | /* Provide a default input filter to make life easy. */ |
1992 | /* Note that we append to the line. This is handy. */ | |
f4c556ac GS |
1993 | DEBUG_P(PerlIO_printf(Perl_debug_log, |
1994 | "filter_read %d: from rsfp\n", idx)); | |
4e553d73 | 1995 | if (maxlen) { |
16d20bd9 AD |
1996 | /* Want a block */ |
1997 | int len ; | |
1998 | int old_len = SvCUR(buf_sv) ; | |
1999 | ||
2000 | /* ensure buf_sv is large enough */ | |
2001 | SvGROW(buf_sv, old_len + maxlen) ; | |
3280af22 NIS |
2002 | if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){ |
2003 | if (PerlIO_error(PL_rsfp)) | |
37120919 AD |
2004 | return -1; /* error */ |
2005 | else | |
2006 | return 0 ; /* end of file */ | |
2007 | } | |
16d20bd9 AD |
2008 | SvCUR_set(buf_sv, old_len + len) ; |
2009 | } else { | |
2010 | /* Want a line */ | |
3280af22 NIS |
2011 | if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) { |
2012 | if (PerlIO_error(PL_rsfp)) | |
37120919 AD |
2013 | return -1; /* error */ |
2014 | else | |
2015 | return 0 ; /* end of file */ | |
2016 | } | |
16d20bd9 AD |
2017 | } |
2018 | return SvCUR(buf_sv); | |
2019 | } | |
2020 | /* Skip this filter slot if filter has been deleted */ | |
3280af22 | 2021 | if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){ |
f4c556ac GS |
2022 | DEBUG_P(PerlIO_printf(Perl_debug_log, |
2023 | "filter_read %d: skipped (filter deleted)\n", | |
2024 | idx)); | |
16d20bd9 AD |
2025 | return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */ |
2026 | } | |
2027 | /* Get function pointer hidden within datasv */ | |
4755096e | 2028 | funcp = (filter_t)IoANY(datasv); |
f4c556ac GS |
2029 | DEBUG_P(PerlIO_printf(Perl_debug_log, |
2030 | "filter_read %d: via function %p (%s)\n", | |
2031 | idx, funcp, SvPV_nolen(datasv))); | |
16d20bd9 AD |
2032 | /* Call function. The function is expected to */ |
2033 | /* call "FILTER_READ(idx+1, buf_sv)" first. */ | |
37120919 | 2034 | /* Return: <0:error, =0:eof, >0:not eof */ |
0cb96387 | 2035 | return (*funcp)(aTHXo_ idx, buf_sv, maxlen); |
16d20bd9 AD |
2036 | } |
2037 | ||
76e3520e | 2038 | STATIC char * |
cea2e8a9 | 2039 | S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append) |
16d20bd9 | 2040 | { |
c39cd008 | 2041 | #ifdef PERL_CR_FILTER |
3280af22 | 2042 | if (!PL_rsfp_filters) { |
c39cd008 | 2043 | filter_add(S_cr_textfilter,NULL); |
a868473f NIS |
2044 | } |
2045 | #endif | |
3280af22 | 2046 | if (PL_rsfp_filters) { |
16d20bd9 | 2047 | |
55497cff | 2048 | if (!append) |
2049 | SvCUR_set(sv, 0); /* start with empty line */ | |
16d20bd9 AD |
2050 | if (FILTER_READ(0, sv, 0) > 0) |
2051 | return ( SvPVX(sv) ) ; | |
2052 | else | |
2053 | return Nullch ; | |
2054 | } | |
9d116dd7 | 2055 | else |
fd049845 | 2056 | return (sv_gets(sv, fp, append)); |
a0d0e21e LW |
2057 | } |
2058 | ||
01ec43d0 GS |
2059 | STATIC HV * |
2060 | S_find_in_my_stash(pTHX_ char *pkgname, I32 len) | |
def3634b GS |
2061 | { |
2062 | GV *gv; | |
2063 | ||
01ec43d0 | 2064 | if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__")) |
def3634b GS |
2065 | return PL_curstash; |
2066 | ||
2067 | if (len > 2 && | |
2068 | (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') && | |
01ec43d0 GS |
2069 | (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV))) |
2070 | { | |
2071 | return GvHV(gv); /* Foo:: */ | |
def3634b GS |
2072 | } |
2073 | ||
2074 | /* use constant CLASS => 'MyClass' */ | |
2075 | if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) { | |
2076 | SV *sv; | |
2077 | if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) { | |
2078 | pkgname = SvPV_nolen(sv); | |
2079 | } | |
2080 | } | |
2081 | ||
2082 | return gv_stashpv(pkgname, FALSE); | |
2083 | } | |
a0d0e21e | 2084 | |
748a9306 LW |
2085 | #ifdef DEBUGGING |
2086 | static char* exp_name[] = | |
09bef843 SB |
2087 | { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK", |
2088 | "ATTRTERM", "TERMBLOCK" | |
2089 | }; | |
748a9306 | 2090 | #endif |
463ee0b2 | 2091 | |
02aa26ce NT |
2092 | /* |
2093 | yylex | |
2094 | ||
2095 | Works out what to call the token just pulled out of the input | |
2096 | stream. The yacc parser takes care of taking the ops we return and | |
2097 | stitching them into a tree. | |
2098 | ||
2099 | Returns: | |
2100 | PRIVATEREF | |
2101 | ||
2102 | Structure: | |
2103 | if read an identifier | |
2104 | if we're in a my declaration | |
2105 | croak if they tried to say my($foo::bar) | |
2106 | build the ops for a my() declaration | |
2107 | if it's an access to a my() variable | |
2108 | are we in a sort block? | |
2109 | croak if my($a); $a <=> $b | |
2110 | build ops for access to a my() variable | |
2111 | if in a dq string, and they've said @foo and we can't find @foo | |
2112 | croak | |
2113 | build ops for a bareword | |
2114 | if we already built the token before, use it. | |
2115 | */ | |
2116 | ||
dba4d153 | 2117 | #ifdef USE_PURE_BISON |
864dbfa3 | 2118 | int |
dba4d153 | 2119 | Perl_yylex_r(pTHX_ YYSTYPE *lvalp, int *lcharp) |
378cc40b | 2120 | { |
20141f0e IRC |
2121 | int r; |
2122 | ||
6f202aea | 2123 | yyactlevel++; |
20141f0e IRC |
2124 | yylval_pointer[yyactlevel] = lvalp; |
2125 | yychar_pointer[yyactlevel] = lcharp; | |
b73d6f50 IRC |
2126 | if (yyactlevel >= YYMAXLEVEL) |
2127 | Perl_croak(aTHX_ "panic: YYMAXLEVEL"); | |
20141f0e | 2128 | |
dba4d153 | 2129 | r = Perl_yylex(aTHX); |
20141f0e | 2130 | |
d8ae6756 IRC |
2131 | if (yyactlevel > 0) |
2132 | yyactlevel--; | |
20141f0e IRC |
2133 | |
2134 | return r; | |
2135 | } | |
dba4d153 | 2136 | #endif |
20141f0e | 2137 | |
dba4d153 JH |
2138 | #ifdef __SC__ |
2139 | #pragma segment Perl_yylex | |
2140 | #endif | |
dba4d153 | 2141 | int |
dba4d153 | 2142 | Perl_yylex(pTHX) |
20141f0e | 2143 | { |
79072805 | 2144 | register char *s; |
378cc40b | 2145 | register char *d; |
79072805 | 2146 | register I32 tmp; |
463ee0b2 | 2147 | STRLEN len; |
161b471a NIS |
2148 | GV *gv = Nullgv; |
2149 | GV **gvp = 0; | |
aa7440fb | 2150 | bool bof = FALSE; |
a687059c | 2151 | |
02aa26ce | 2152 | /* check if there's an identifier for us to look at */ |
3280af22 | 2153 | if (PL_pending_ident) { |
02aa26ce | 2154 | /* pit holds the identifier we read and pending_ident is reset */ |
3280af22 NIS |
2155 | char pit = PL_pending_ident; |
2156 | PL_pending_ident = 0; | |
bbce6d69 | 2157 | |
607df283 SC |
2158 | DEBUG_T({ PerlIO_printf(Perl_debug_log, |
2159 | "### Tokener saw identifier '%s'\n", PL_tokenbuf); }) | |
2160 | ||
02aa26ce NT |
2161 | /* if we're in a my(), we can't allow dynamics here. |
2162 | $foo'bar has already been turned into $foo::bar, so | |
2163 | just check for colons. | |
2164 | ||
2165 | if it's a legal name, the OP is a PADANY. | |
2166 | */ | |
3280af22 | 2167 | if (PL_in_my) { |
77ca0c92 | 2168 | if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */ |
1ec3e8de GS |
2169 | if (strchr(PL_tokenbuf,':')) |
2170 | yyerror(Perl_form(aTHX_ "No package name allowed for " | |
2171 | "variable %s in \"our\"", | |
2172 | PL_tokenbuf)); | |
77ca0c92 LW |
2173 | tmp = pad_allocmy(PL_tokenbuf); |
2174 | } | |
2175 | else { | |
2176 | if (strchr(PL_tokenbuf,':')) | |
2177 | yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf)); | |
02aa26ce | 2178 | |
77ca0c92 LW |
2179 | yylval.opval = newOP(OP_PADANY, 0); |
2180 | yylval.opval->op_targ = pad_allocmy(PL_tokenbuf); | |
2181 | return PRIVATEREF; | |
2182 | } | |
bbce6d69 | 2183 | } |
2184 | ||
4e553d73 | 2185 | /* |
02aa26ce NT |
2186 | build the ops for accesses to a my() variable. |
2187 | ||
2188 | Deny my($a) or my($b) in a sort block, *if* $a or $b is | |
2189 | then used in a comparison. This catches most, but not | |
2190 | all cases. For instance, it catches | |
2191 | sort { my($a); $a <=> $b } | |
2192 | but not | |
2193 | sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; } | |
2194 | (although why you'd do that is anyone's guess). | |
2195 | */ | |
2196 | ||
3280af22 | 2197 | if (!strchr(PL_tokenbuf,':')) { |
a863c7d1 | 2198 | #ifdef USE_THREADS |
54b9620d | 2199 | /* Check for single character per-thread SVs */ |
3280af22 NIS |
2200 | if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0' |
2201 | && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */ | |
2202 | && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD) | |
554b3eca | 2203 | { |
2faa37cc | 2204 | yylval.opval = newOP(OP_THREADSV, 0); |
a863c7d1 MB |
2205 | yylval.opval->op_targ = tmp; |
2206 | return PRIVATEREF; | |
2207 | } | |
2208 | #endif /* USE_THREADS */ | |
3280af22 | 2209 | if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) { |
f472eb5c | 2210 | SV *namesv = AvARRAY(PL_comppad_name)[tmp]; |
77ca0c92 | 2211 | /* might be an "our" variable" */ |
f472eb5c | 2212 | if (SvFLAGS(namesv) & SVpad_OUR) { |
77ca0c92 | 2213 | /* build ops for a bareword */ |
f472eb5c GS |
2214 | SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0); |
2215 | sv_catpvn(sym, "::", 2); | |
2216 | sv_catpv(sym, PL_tokenbuf+1); | |
2217 | yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym); | |
77ca0c92 | 2218 | yylval.opval->op_private = OPpCONST_ENTERED; |
f472eb5c | 2219 | gv_fetchpv(SvPVX(sym), |
77ca0c92 | 2220 | (PL_in_eval |
f472eb5c GS |
2221 | ? (GV_ADDMULTI | GV_ADDINEVAL) |
2222 | : TRUE | |
77ca0c92 LW |
2223 | ), |
2224 | ((PL_tokenbuf[0] == '$') ? SVt_PV | |
2225 | : (PL_tokenbuf[0] == '@') ? SVt_PVAV | |
2226 | : SVt_PVHV)); | |
2227 | return WORD; | |
2228 | } | |
2229 | ||
02aa26ce | 2230 | /* if it's a sort block and they're naming $a or $b */ |
3280af22 NIS |
2231 | if (PL_last_lop_op == OP_SORT && |
2232 | PL_tokenbuf[0] == '$' && | |
2233 | (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b') | |
2234 | && !PL_tokenbuf[2]) | |
bbce6d69 | 2235 | { |
3280af22 NIS |
2236 | for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart; |
2237 | d < PL_bufend && *d != '\n'; | |
a863c7d1 MB |
2238 | d++) |
2239 | { | |
2240 | if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) { | |
cea2e8a9 | 2241 | Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison", |
3280af22 | 2242 | PL_tokenbuf); |
a863c7d1 | 2243 | } |
bbce6d69 | 2244 | } |
2245 | } | |
bbce6d69 | 2246 | |
a863c7d1 MB |
2247 | yylval.opval = newOP(OP_PADANY, 0); |
2248 | yylval.opval->op_targ = tmp; | |
2249 | return PRIVATEREF; | |
2250 | } | |
bbce6d69 | 2251 | } |
2252 | ||
02aa26ce NT |
2253 | /* |
2254 | Whine if they've said @foo in a doublequoted string, | |
2255 | and @foo isn't a variable we can find in the symbol | |
2256 | table. | |
2257 | */ | |
3280af22 NIS |
2258 | if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) { |
2259 | GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV); | |
8593bda5 GS |
2260 | if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv))) |
2261 | && ckWARN(WARN_AMBIGUOUS)) | |
2262 | { | |
2263 | /* Downgraded from fatal to warning 20000522 mjd */ | |
2264 | Perl_warner(aTHX_ WARN_AMBIGUOUS, | |
2265 | "Possible unintended interpolation of %s in string", | |
2266 | PL_tokenbuf); | |
2267 | } | |
bbce6d69 | 2268 | } |
2269 | ||
02aa26ce | 2270 | /* build ops for a bareword */ |
3280af22 | 2271 | yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0)); |
bbce6d69 | 2272 | yylval.opval->op_private = OPpCONST_ENTERED; |
3280af22 NIS |
2273 | gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE, |
2274 | ((PL_tokenbuf[0] == '$') ? SVt_PV | |
2275 | : (PL_tokenbuf[0] == '@') ? SVt_PVAV | |
bbce6d69 | 2276 | : SVt_PVHV)); |
2277 | return WORD; | |
2278 | } | |
2279 | ||
02aa26ce NT |
2280 | /* no identifier pending identification */ |
2281 | ||
3280af22 | 2282 | switch (PL_lex_state) { |
79072805 LW |
2283 | #ifdef COMMENTARY |
2284 | case LEX_NORMAL: /* Some compilers will produce faster */ | |
2285 | case LEX_INTERPNORMAL: /* code if we comment these out. */ | |
2286 | break; | |
2287 | #endif | |
2288 | ||
09bef843 | 2289 | /* when we've already built the next token, just pull it out of the queue */ |
79072805 | 2290 | case LEX_KNOWNEXT: |
3280af22 NIS |
2291 | PL_nexttoke--; |
2292 | yylval = PL_nextval[PL_nexttoke]; | |
2293 | if (!PL_nexttoke) { | |
2294 | PL_lex_state = PL_lex_defer; | |
2295 | PL_expect = PL_lex_expect; | |
2296 | PL_lex_defer = LEX_NORMAL; | |
463ee0b2 | 2297 | } |
607df283 | 2298 | DEBUG_T({ PerlIO_printf(Perl_debug_log, |
4659c93f RB |
2299 | "### Next token after '%s' was known, type %"IVdf"\n", PL_bufptr, |
2300 | (IV)PL_nexttype[PL_nexttoke]); }) | |
607df283 | 2301 | |
3280af22 | 2302 | return(PL_nexttype[PL_nexttoke]); |
79072805 | 2303 | |
02aa26ce | 2304 | /* interpolated case modifiers like \L \U, including \Q and \E. |
3280af22 | 2305 | when we get here, PL_bufptr is at the \ |
02aa26ce | 2306 | */ |
79072805 LW |
2307 | case LEX_INTERPCASEMOD: |
2308 | #ifdef DEBUGGING | |
3280af22 | 2309 | if (PL_bufptr != PL_bufend && *PL_bufptr != '\\') |
cea2e8a9 | 2310 | Perl_croak(aTHX_ "panic: INTERPCASEMOD"); |
79072805 | 2311 | #endif |
02aa26ce | 2312 | /* handle \E or end of string */ |
3280af22 | 2313 | if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') { |
a0d0e21e | 2314 | char oldmod; |
02aa26ce NT |
2315 | |
2316 | /* if at a \E */ | |
3280af22 NIS |
2317 | if (PL_lex_casemods) { |
2318 | oldmod = PL_lex_casestack[--PL_lex_casemods]; | |
2319 | PL_lex_casestack[PL_lex_casemods] = '\0'; | |
02aa26ce | 2320 | |
3280af22 NIS |
2321 | if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) { |
2322 | PL_bufptr += 2; | |
2323 | PL_lex_state = LEX_INTERPCONCAT; | |
a0d0e21e | 2324 | } |
79072805 LW |
2325 | return ')'; |
2326 | } | |
3280af22 NIS |
2327 | if (PL_bufptr != PL_bufend) |
2328 | PL_bufptr += 2; | |
2329 | PL_lex_state = LEX_INTERPCONCAT; | |
cea2e8a9 | 2330 | return yylex(); |
79072805 LW |
2331 | } |
2332 | else { | |
607df283 SC |
2333 | DEBUG_T({ PerlIO_printf(Perl_debug_log, |
2334 | "### Saw case modifier at '%s'\n", PL_bufptr); }) | |
3280af22 | 2335 | s = PL_bufptr + 1; |
79072805 LW |
2336 | if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3)) |
2337 | tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */ | |
a0d0e21e | 2338 | if (strchr("LU", *s) && |
3280af22 | 2339 | (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U'))) |
a0d0e21e | 2340 | { |
3280af22 | 2341 | PL_lex_casestack[--PL_lex_casemods] = '\0'; |
a0d0e21e LW |
2342 | return ')'; |
2343 | } | |
3280af22 NIS |
2344 | if (PL_lex_casemods > 10) { |
2345 | char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char); | |
2346 | if (newlb != PL_lex_casestack) { | |
a0d0e21e | 2347 | SAVEFREEPV(newlb); |
3280af22 | 2348 | PL_lex_casestack = newlb; |
a0d0e21e LW |
2349 | } |
2350 | } | |
3280af22 NIS |
2351 | PL_lex_casestack[PL_lex_casemods++] = *s; |
2352 | PL_lex_casestack[PL_lex_casemods] = '\0'; | |
2353 | PL_lex_state = LEX_INTERPCONCAT; | |
2354 | PL_nextval[PL_nexttoke].ival = 0; | |
79072805 LW |
2355 | force_next('('); |
2356 | if (*s == 'l') | |
3280af22 | 2357 | PL_nextval[PL_nexttoke].ival = OP_LCFIRST; |
79072805 | 2358 | else if (*s == 'u') |
3280af22 | 2359 | PL_nextval[PL_nexttoke].ival = OP_UCFIRST; |
79072805 | 2360 | else if (*s == 'L') |
3280af22 | 2361 | PL_nextval[PL_nexttoke].ival = OP_LC; |
79072805 | 2362 | else if (*s == 'U') |
3280af22 | 2363 | PL_nextval[PL_nexttoke].ival = OP_UC; |
a0d0e21e | 2364 | else if (*s == 'Q') |
3280af22 | 2365 | PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA; |
79072805 | 2366 | else |
cea2e8a9 | 2367 | Perl_croak(aTHX_ "panic: yylex"); |
3280af22 | 2368 | PL_bufptr = s + 1; |
79072805 | 2369 | force_next(FUNC); |
3280af22 NIS |
2370 | if (PL_lex_starts) { |
2371 | s = PL_bufptr; | |
2372 | PL_lex_starts = 0; | |
79072805 LW |
2373 | Aop(OP_CONCAT); |
2374 | } | |
2375 | else | |
cea2e8a9 | 2376 | return yylex(); |
79072805 LW |
2377 | } |
2378 | ||
55497cff | 2379 | case LEX_INTERPPUSH: |
2380 | return sublex_push(); | |
2381 | ||
79072805 | 2382 | case LEX_INTERPSTART: |
3280af22 | 2383 | if (PL_bufptr == PL_bufend) |
79072805 | 2384 | return sublex_done(); |
607df283 SC |
2385 | DEBUG_T({ PerlIO_printf(Perl_debug_log, |
2386 | "### Interpolated variable at '%s'\n", PL_bufptr); }) | |
3280af22 NIS |
2387 | PL_expect = XTERM; |
2388 | PL_lex_dojoin = (*PL_bufptr == '@'); | |
2389 | PL_lex_state = LEX_INTERPNORMAL; | |
2390 | if (PL_lex_dojoin) { | |
2391 | PL_nextval[PL_nexttoke].ival = 0; | |
79072805 | 2392 | force_next(','); |
554b3eca | 2393 | #ifdef USE_THREADS |
533c011a NIS |
2394 | PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0); |
2395 | PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\""); | |
554b3eca MB |
2396 | force_next(PRIVATEREF); |
2397 | #else | |
a0d0e21e | 2398 | force_ident("\"", '$'); |
554b3eca | 2399 | #endif /* USE_THREADS */ |
3280af22 | 2400 | PL_nextval[PL_nexttoke].ival = 0; |
79072805 | 2401 | force_next('$'); |
3280af22 | 2402 | PL_nextval[PL_nexttoke].ival = 0; |
79072805 | 2403 | force_next('('); |
3280af22 | 2404 | PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */ |
79072805 LW |
2405 | force_next(FUNC); |
2406 | } | |
3280af22 NIS |
2407 | if (PL_lex_starts++) { |
2408 | s = PL_bufptr; | |
79072805 LW |
2409 | Aop(OP_CONCAT); |
2410 | } | |
cea2e8a9 | 2411 | return yylex(); |
79072805 LW |
2412 | |
2413 | case LEX_INTERPENDMAYBE: | |
3280af22 NIS |
2414 | if (intuit_more(PL_bufptr)) { |
2415 | PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */ | |
79072805 LW |
2416 | break; |
2417 | } | |
2418 | /* FALL THROUGH */ | |
2419 | ||
2420 | case LEX_INTERPEND: | |
3280af22 NIS |
2421 | if (PL_lex_dojoin) { |
2422 | PL_lex_dojoin = FALSE; | |
2423 | PL_lex_state = LEX_INTERPCONCAT; | |
79072805 LW |
2424 | return ')'; |
2425 | } | |
43a16006 | 2426 | if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl |
25da4f38 | 2427 | && SvEVALED(PL_lex_repl)) |
43a16006 | 2428 | { |
e9fa98b2 | 2429 | if (PL_bufptr != PL_bufend) |
cea2e8a9 | 2430 | Perl_croak(aTHX_ "Bad evalled substitution pattern"); |
e9fa98b2 HS |
2431 | PL_lex_repl = Nullsv; |
2432 | } | |
79072805 LW |
2433 | /* FALLTHROUGH */ |
2434 | case LEX_INTERPCONCAT: | |
2435 | #ifdef DEBUGGING | |
3280af22 | 2436 | if (PL_lex_brackets) |
cea2e8a9 | 2437 | Perl_croak(aTHX_ "panic: INTERPCONCAT"); |
79072805 | 2438 | #endif |
3280af22 | 2439 | if (PL_bufptr == PL_bufend) |
79072805 LW |
2440 | return sublex_done(); |
2441 | ||
3280af22 NIS |
2442 | if (SvIVX(PL_linestr) == '\'') { |
2443 | SV *sv = newSVsv(PL_linestr); | |
2444 | if (!PL_lex_inpat) | |
76e3520e | 2445 | sv = tokeq(sv); |
3280af22 | 2446 | else if ( PL_hints & HINT_NEW_RE ) |
b3ac6de7 | 2447 | sv = new_constant(NULL, 0, "qr", sv, sv, "q"); |
79072805 | 2448 | yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv); |
3280af22 | 2449 | s = PL_bufend; |
79072805 LW |
2450 | } |
2451 | else { | |
3280af22 | 2452 | s = scan_const(PL_bufptr); |
79072805 | 2453 | if (*s == '\\') |
3280af22 | 2454 | PL_lex_state = LEX_INTERPCASEMOD; |
79072805 | 2455 | else |
3280af22 | 2456 | PL_lex_state = LEX_INTERPSTART; |
79072805 LW |
2457 | } |
2458 | ||
3280af22 NIS |
2459 | if (s != PL_bufptr) { |
2460 | PL_nextval[PL_nexttoke] = yylval; | |
2461 | PL_expect = XTERM; | |
79072805 | 2462 | force_next(THING); |
3280af22 | 2463 | if (PL_lex_starts++) |
79072805 LW |
2464 | Aop(OP_CONCAT); |
2465 | else { | |
3280af22 | 2466 | PL_bufptr = s; |
cea2e8a9 | 2467 | return yylex(); |
79072805 LW |
2468 | } |
2469 | } | |
2470 | ||
cea2e8a9 | 2471 | return yylex(); |
a0d0e21e | 2472 | case LEX_FORMLINE: |
3280af22 NIS |
2473 | PL_lex_state = LEX_NORMAL; |
2474 | s = scan_formline(PL_bufptr); | |
2475 | if (!PL_lex_formbrack) | |
a0d0e21e LW |
2476 | goto rightbracket; |
2477 | OPERATOR(';'); | |
79072805 LW |
2478 | } |
2479 | ||
3280af22 NIS |
2480 | s = PL_bufptr; |
2481 | PL_oldoldbufptr = PL_oldbufptr; | |
2482 | PL_oldbufptr = s; | |
607df283 | 2483 | DEBUG_T( { |
bf49b057 GS |
2484 | PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n", |
2485 | exp_name[PL_expect], s); | |
79072805 | 2486 | } ) |
463ee0b2 LW |
2487 | |
2488 | retry: | |
378cc40b LW |
2489 | switch (*s) { |
2490 | default: | |
7e2040f0 | 2491 | if (isIDFIRST_lazy_if(s,UTF)) |
834a4ddd | 2492 | goto keylookup; |
cea2e8a9 | 2493 | Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255); |
e929a76b LW |
2494 | case 4: |
2495 | case 26: | |
2496 | goto fake_eof; /* emulate EOF on ^D or ^Z */ | |
378cc40b | 2497 | case 0: |
3280af22 NIS |
2498 | if (!PL_rsfp) { |
2499 | PL_last_uni = 0; | |
2500 | PL_last_lop = 0; | |
2501 | if (PL_lex_brackets) | |
d98d5fff | 2502 | yyerror("Missing right curly or square bracket"); |
4e553d73 | 2503 | DEBUG_T( { PerlIO_printf(Perl_debug_log, |
607df283 SC |
2504 | "### Tokener got EOF\n"); |
2505 | } ) | |
79072805 | 2506 | TOKEN(0); |
463ee0b2 | 2507 | } |
3280af22 | 2508 | if (s++ < PL_bufend) |
a687059c | 2509 | goto retry; /* ignore stray nulls */ |
3280af22 NIS |
2510 | PL_last_uni = 0; |
2511 | PL_last_lop = 0; | |
2512 | if (!PL_in_eval && !PL_preambled) { | |
2513 | PL_preambled = TRUE; | |
2514 | sv_setpv(PL_linestr,incl_perldb()); | |
2515 | if (SvCUR(PL_linestr)) | |
2516 | sv_catpv(PL_linestr,";"); | |
2517 | if (PL_preambleav){ | |
2518 | while(AvFILLp(PL_preambleav) >= 0) { | |
2519 | SV *tmpsv = av_shift(PL_preambleav); | |
2520 | sv_catsv(PL_linestr, tmpsv); | |
2521 | sv_catpv(PL_linestr, ";"); | |
91b7def8 | 2522 | sv_free(tmpsv); |
2523 | } | |
3280af22 NIS |
2524 | sv_free((SV*)PL_preambleav); |
2525 | PL_preambleav = NULL; | |
91b7def8 | 2526 | } |
3280af22 NIS |
2527 | if (PL_minus_n || PL_minus_p) { |
2528 | sv_catpv(PL_linestr, "LINE: while (<>) {"); | |
2529 | if (PL_minus_l) | |
2530 | sv_catpv(PL_linestr,"chomp;"); | |
2531 | if (PL_minus_a) { | |
8fd239a7 CS |
2532 | GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV); |
2533 | if (gv) | |
2534 | GvIMPORTED_AV_on(gv); | |
3280af22 NIS |
2535 | if (PL_minus_F) { |
2536 | if (strchr("/'\"", *PL_splitstr) | |
2537 | && strchr(PL_splitstr + 1, *PL_splitstr)) | |
cea2e8a9 | 2538 | Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr); |
54310121 | 2539 | else { |
2540 | char delim; | |
2541 | s = "'~#\200\1'"; /* surely one char is unused...*/ | |
3280af22 | 2542 | while (s[1] && strchr(PL_splitstr, *s)) s++; |
54310121 | 2543 | delim = *s; |
cea2e8a9 | 2544 | Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c", |
46fc3d4c | 2545 | "q" + (delim == '\''), delim); |
3280af22 | 2546 | for (s = PL_splitstr; *s; s++) { |
54310121 | 2547 | if (*s == '\\') |
3280af22 NIS |
2548 | sv_catpvn(PL_linestr, "\\", 1); |
2549 | sv_catpvn(PL_linestr, s, 1); | |
54310121 | 2550 | } |
cea2e8a9 | 2551 | Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim); |
54310121 | 2552 | } |
2304df62 AD |
2553 | } |
2554 | else | |
3280af22 | 2555 | sv_catpv(PL_linestr,"@F=split(' ');"); |
2304df62 | 2556 | } |
79072805 | 2557 | } |
3280af22 NIS |
2558 | sv_catpv(PL_linestr, "\n"); |
2559 | PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr); | |
2560 | PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); | |
2561 | if (PERLDB_LINE && PL_curstash != PL_debstash) { | |
a0d0e21e LW |
2562 | SV *sv = NEWSV(85,0); |
2563 | ||
2564 | sv_upgrade(sv, SVt_PVMG); | |
3280af22 | 2565 | sv_setsv(sv,PL_linestr); |
57843af0 | 2566 | av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv); |
a0d0e21e | 2567 | } |
79072805 | 2568 | goto retry; |
a687059c | 2569 | } |
e929a76b | 2570 | do { |
aa7440fb | 2571 | bof = PL_rsfp ? TRUE : FALSE; |
7e28d3af JH |
2572 | if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) { |
2573 | fake_eof: | |
2574 | if (PL_rsfp) { | |
2575 | if (PL_preprocess && !PL_in_eval) | |
2576 | (void)PerlProc_pclose(PL_rsfp); | |
2577 | else if ((PerlIO *)PL_rsfp == PerlIO_stdin()) | |
2578 | PerlIO_clearerr(PL_rsfp); | |
2579 | else | |
2580 | (void)PerlIO_close(PL_rsfp); | |
2581 | PL_rsfp = Nullfp; | |
2582 | PL_doextract = FALSE; | |
2583 | } | |
2584 | if (!PL_in_eval && (PL_minus_n || PL_minus_p)) { | |
2585 | sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : ""); | |
2586 | sv_catpv(PL_linestr,";}"); | |
2587 | PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr); | |
2588 | PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); | |
2589 | PL_minus_n = PL_minus_p = 0; | |
2590 | goto retry; | |
2591 | } | |
2592 | PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr); | |
2593 | sv_setpv(PL_linestr,""); | |
2594 | TOKEN(';'); /* not infinite loop because rsfp is NULL now */ | |
2595 | } | |
2596 | /* if it looks like the start of a BOM, check if it in fact is */ | |
2597 | else if (bof && (!*s || *(U8*)s == 0xEF || *(U8*)s >= 0xFE)) { | |
226017aa | 2598 | #ifdef PERLIO_IS_STDIO |
e3f494f1 JH |
2599 | # ifdef __GNU_LIBRARY__ |
2600 | # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */ | |
226017aa DD |
2601 | # define FTELL_FOR_PIPE_IS_BROKEN |
2602 | # endif | |
e3f494f1 JH |
2603 | # else |
2604 | # ifdef __GLIBC__ | |
2605 | # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */ | |
2606 | # define FTELL_FOR_PIPE_IS_BROKEN | |
2607 | # endif | |
2608 | # endif | |
226017aa DD |
2609 | # endif |
2610 | #endif | |
2611 | #ifdef FTELL_FOR_PIPE_IS_BROKEN | |
2612 | /* This loses the possibility to detect the bof | |
2613 | * situation on perl -P when the libc5 is being used. | |
2614 | * Workaround? Maybe attach some extra state to PL_rsfp? | |
2615 | */ | |
2616 | if (!PL_preprocess) | |
7e28d3af | 2617 | bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr); |
226017aa | 2618 | #else |
7e28d3af | 2619 | bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr); |
226017aa | 2620 | #endif |
7e28d3af | 2621 | if (bof) { |
3280af22 | 2622 | PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); |
7e28d3af | 2623 | s = swallow_bom((U8*)s); |
e929a76b | 2624 | } |
378cc40b | 2625 | } |
3280af22 | 2626 | if (PL_doextract) { |
a0d0e21e | 2627 | if (*s == '#' && s[1] == '!' && instr(s,"perl")) |
3280af22 | 2628 | PL_doextract = FALSE; |
a0d0e21e LW |
2629 | |
2630 | /* Incest with pod. */ | |
2631 | if (*s == '=' && strnEQ(s, "=cut", 4)) { | |
3280af22 NIS |
2632 | sv_setpv(PL_linestr, ""); |
2633 | PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr); | |
2634 | PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); | |
2635 | PL_doextract = FALSE; | |
a0d0e21e | 2636 | } |
4e553d73 | 2637 | } |
463ee0b2 | 2638 | incline(s); |
3280af22 NIS |
2639 | } while (PL_doextract); |
2640 | PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s; | |
2641 | if (PERLDB_LINE && PL_curstash != PL_debstash) { | |
79072805 | 2642 | SV *sv = NEWSV(85,0); |
a687059c | 2643 | |
93a17b20 | 2644 | sv_upgrade(sv, SVt_PVMG); |
3280af22 | 2645 | sv_setsv(sv,PL_linestr); |
57843af0 | 2646 | av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv); |
a687059c | 2647 | } |
3280af22 | 2648 | PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); |
57843af0 | 2649 | if (CopLINE(PL_curcop) == 1) { |
3280af22 | 2650 | while (s < PL_bufend && isSPACE(*s)) |
79072805 | 2651 | s++; |
a0d0e21e | 2652 | if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */ |
79072805 | 2653 | s++; |
44a8e56a | 2654 | d = Nullch; |
3280af22 | 2655 | if (!PL_in_eval) { |
44a8e56a | 2656 | if (*s == '#' && *(s+1) == '!') |
2657 | d = s + 2; | |
2658 | #ifdef ALTERNATE_SHEBANG | |
2659 | else { | |
2660 | static char as[] = ALTERNATE_SHEBANG; | |
2661 | if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1)) | |
2662 | d = s + (sizeof(as) - 1); | |
2663 | } | |
2664 | #endif /* ALTERNATE_SHEBANG */ | |
2665 | } | |
2666 | if (d) { | |
b8378b72 | 2667 | char *ipath; |
774d564b | 2668 | char *ipathend; |
b8378b72 | 2669 | |
774d564b | 2670 | while (isSPACE(*d)) |
b8378b72 CS |
2671 | d++; |
2672 | ipath = d; | |
774d564b | 2673 | while (*d && !isSPACE(*d)) |
2674 | d++; | |
2675 | ipathend = d; | |
2676 | ||
2677 | #ifdef ARG_ZERO_IS_SCRIPT | |
2678 | if (ipathend > ipath) { | |
2679 | /* | |
2680 | * HP-UX (at least) sets argv[0] to the script name, | |
2681 | * which makes $^X incorrect. And Digital UNIX and Linux, | |
2682 | * at least, set argv[0] to the basename of the Perl | |
2683 | * interpreter. So, having found "#!", we'll set it right. | |
2684 | */ | |
2685 | SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV)); | |
2686 | assert(SvPOK(x) || SvGMAGICAL(x)); | |
cc49e20b | 2687 | if (sv_eq(x, CopFILESV(PL_curcop))) { |
774d564b | 2688 | sv_setpvn(x, ipath, ipathend - ipath); |
9607fc9c | 2689 | SvSETMAGIC(x); |
2690 | } | |
774d564b | 2691 | TAINT_NOT; /* $^X is always tainted, but that's OK */ |
8ebc5c01 | 2692 | } |
774d564b | 2693 | #endif /* ARG_ZERO_IS_SCRIPT */ |
b8378b72 CS |
2694 | |
2695 | /* | |
2696 | * Look for options. | |
2697 | */ | |
748a9306 | 2698 | d = instr(s,"perl -"); |
84e30d1a | 2699 | if (!d) { |
748a9306 | 2700 | d = instr(s,"perl"); |
84e30d1a GS |
2701 | #if defined(DOSISH) |
2702 | /* avoid getting into infinite loops when shebang | |
2703 | * line contains "Perl" rather than "perl" */ | |
2704 | if (!d) { | |
2705 | for (d = ipathend-4; d >= ipath; --d) { | |
2706 | if ((*d == 'p' || *d == 'P') | |
2707 | && !ibcmp(d, "perl", 4)) | |
2708 | { | |
2709 | break; | |
2710 | } | |
2711 | } | |
2712 | if (d < ipath) | |
2713 | d = Nullch; | |
2714 | } | |
2715 | #endif | |
2716 | } | |
44a8e56a | 2717 | #ifdef ALTERNATE_SHEBANG |
2718 | /* | |
2719 | * If the ALTERNATE_SHEBANG on this system starts with a | |
2720 | * character that can be part of a Perl expression, then if | |
2721 | * we see it but not "perl", we're probably looking at the | |
2722 | * start of Perl code, not a request to hand off to some | |
2723 | * other interpreter. Similarly, if "perl" is there, but | |
2724 | * not in the first 'word' of the line, we assume the line | |
2725 | * contains the start of the Perl program. | |
44a8e56a | 2726 | */ |
2727 | if (d && *s != '#') { | |
774d564b | 2728 | char *c = ipath; |
44a8e56a | 2729 | while (*c && !strchr("; \t\r\n\f\v#", *c)) |
2730 | c++; | |
2731 | if (c < d) | |
2732 | d = Nullch; /* "perl" not in first word; ignore */ | |
2733 | else | |
2734 | *s = '#'; /* Don't try to parse shebang line */ | |
2735 | } | |
774d564b | 2736 | #endif /* ALTERNATE_SHEBANG */ |
bf4acbe4 | 2737 | #ifndef MACOS_TRADITIONAL |
748a9306 | 2738 | if (!d && |
44a8e56a | 2739 | *s == '#' && |
774d564b | 2740 | ipathend > ipath && |
3280af22 | 2741 | !PL_minus_c && |
748a9306 | 2742 | !instr(s,"indir") && |
3280af22 | 2743 | instr(PL_origargv[0],"perl")) |
748a9306 | 2744 | { |
9f68db38 | 2745 | char **newargv; |
9f68db38 | 2746 | |
774d564b | 2747 | *ipathend = '\0'; |
2748 | s = ipathend + 1; | |
3280af22 | 2749 | while (s < PL_bufend && isSPACE(*s)) |
9f68db38 | 2750 | s++; |
3280af22 NIS |
2751 | if (s < PL_bufend) { |
2752 | Newz(899,newargv,PL_origargc+3,char*); | |
9f68db38 | 2753 | newargv[1] = s; |
3280af22 | 2754 | while (s < PL_bufend && !isSPACE(*s)) |
9f68db38 LW |
2755 | s++; |
2756 | *s = '\0'; | |
3280af22 | 2757 | Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*); |
9f68db38 LW |
2758 | } |
2759 | else | |
3280af22 | 2760 | newargv = PL_origargv; |
774d564b | 2761 | newargv[0] = ipath; |
b4748376 | 2762 | PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv)); |
cea2e8a9 | 2763 | Perl_croak(aTHX_ "Can't exec %s", ipath); |
9f68db38 | 2764 | } |
bf4acbe4 | 2765 | #endif |
748a9306 | 2766 | if (d) { |
3280af22 NIS |
2767 | U32 oldpdb = PL_perldb; |
2768 | bool oldn = PL_minus_n; | |
2769 | bool oldp = PL_minus_p; | |
748a9306 LW |
2770 | |
2771 | while (*d && !isSPACE(*d)) d++; | |
bf4acbe4 | 2772 | while (SPACE_OR_TAB(*d)) d++; |
748a9306 LW |
2773 | |
2774 | if (*d++ == '-') { | |
8cc95fdb | 2775 | do { |
2776 | if (*d == 'M' || *d == 'm') { | |
2777 | char *m = d; | |
2778 | while (*d && !isSPACE(*d)) d++; | |
cea2e8a9 | 2779 | Perl_croak(aTHX_ "Too late for \"-%.*s\" option", |
8cc95fdb | 2780 | (int)(d - m), m); |
2781 | } | |
2782 | d = moreswitches(d); | |
2783 | } while (d); | |
155aba94 GS |
2784 | if ((PERLDB_LINE && !oldpdb) || |
2785 | ((PL_minus_n || PL_minus_p) && !(oldn || oldp))) | |
b084f20b | 2786 | /* if we have already added "LINE: while (<>) {", |
2787 | we must not do it again */ | |
748a9306 | 2788 | { |
3280af22 NIS |
2789 | sv_setpv(PL_linestr, ""); |
2790 | PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr); | |
2791 | PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); | |
2792 | PL_preambled = FALSE; | |
84902520 | 2793 | if (PERLDB_LINE) |
3280af22 | 2794 | (void)gv_fetchfile(PL_origfilename); |
748a9306 LW |
2795 | goto retry; |
2796 | } | |
a0d0e21e | 2797 | } |
79072805 | 2798 | } |
9f68db38 | 2799 | } |
79072805 | 2800 | } |
3280af22 NIS |
2801 | if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) { |
2802 | PL_bufptr = s; | |
2803 | PL_lex_state = LEX_FORMLINE; | |
cea2e8a9 | 2804 | return yylex(); |
ae986130 | 2805 | } |
378cc40b | 2806 | goto retry; |
4fdae800 | 2807 | case '\r': |
6a27c188 | 2808 | #ifdef PERL_STRICT_CR |
cea2e8a9 | 2809 | Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r'); |
4e553d73 | 2810 | Perl_croak(aTHX_ |
cc507455 | 2811 | "\t(Maybe you didn't strip carriage returns after a network transfer?)\n"); |
a868473f | 2812 | #endif |
4fdae800 | 2813 | case ' ': case '\t': case '\f': case 013: |
bf4acbe4 GS |
2814 | #ifdef MACOS_TRADITIONAL |
2815 | case '\312': | |
2816 | #endif | |
378cc40b LW |
2817 | s++; |
2818 | goto retry; | |
378cc40b | 2819 | case '#': |
e929a76b | 2820 | case '\n': |
3280af22 | 2821 | if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) { |
df0deb90 GS |
2822 | if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) { |
2823 | /* handle eval qq[#line 1 "foo"\n ...] */ | |
2824 | CopLINE_dec(PL_curcop); | |
2825 | incline(s); | |
2826 | } | |
3280af22 | 2827 | d = PL_bufend; |
a687059c | 2828 | while (s < d && *s != '\n') |
378cc40b | 2829 | s++; |
0f85fab0 | 2830 | if (s < d) |
378cc40b | 2831 | s++; |
463ee0b2 | 2832 | incline(s); |
3280af22 NIS |
2833 | if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) { |
2834 | PL_bufptr = s; | |
2835 | PL_lex_state = LEX_FORMLINE; | |
cea2e8a9 | 2836 | return yylex(); |
a687059c | 2837 | } |
378cc40b | 2838 | } |
a687059c | 2839 | else { |
378cc40b | 2840 | *s = '\0'; |
3280af22 | 2841 | PL_bufend = s; |
a687059c | 2842 | } |
378cc40b LW |
2843 | goto retry; |
2844 | case '-': | |
79072805 | 2845 | if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) { |
e5edeb50 JH |
2846 | I32 ftst = 0; |
2847 | ||
378cc40b | 2848 | s++; |
3280af22 | 2849 | PL_bufptr = s; |
748a9306 LW |
2850 | tmp = *s++; |
2851 | ||
bf4acbe4 | 2852 | while (s < PL_bufend && SPACE_OR_TAB(*s)) |
748a9306 LW |
2853 | s++; |
2854 | ||
2855 | if (strnEQ(s,"=>",2)) { | |
3280af22 | 2856 | s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE); |
4e553d73 | 2857 | DEBUG_T( { PerlIO_printf(Perl_debug_log, |
607df283 SC |
2858 | "### Saw unary minus before =>, forcing word '%s'\n", s); |
2859 | } ) | |
748a9306 LW |
2860 | OPERATOR('-'); /* unary minus */ |
2861 | } | |
3280af22 | 2862 | PL_last_uni = PL_oldbufptr; |
748a9306 | 2863 | switch (tmp) { |
e5edeb50 JH |
2864 | case 'r': ftst = OP_FTEREAD; break; |
2865 | case 'w': ftst = OP_FTEWRITE; break; | |
2866 | case 'x': ftst = OP_FTEEXEC; break; | |
2867 | case 'o': ftst = OP_FTEOWNED; break; | |
2868 | case 'R': ftst = OP_FTRREAD; break; | |
2869 | case 'W': ftst = OP_FTRWRITE; break; | |
2870 | case 'X': ftst = OP_FTREXEC; break; | |
2871 | case 'O': ftst = OP_FTROWNED; break; | |
2872 | case 'e': ftst = OP_FTIS; break; | |
2873 | case 'z': ftst = OP_FTZERO; break; | |
2874 | case 's': ftst = OP_FTSIZE; break; | |
2875 | case 'f': ftst = OP_FTFILE; break; | |
2876 | case 'd': ftst = OP_FTDIR; break; | |
2877 | case 'l': ftst = OP_FTLINK; break; | |
2878 | case 'p': ftst = OP_FTPIPE; break; | |
2879 | case 'S': ftst = OP_FTSOCK; break; | |
2880 | case 'u': ftst = OP_FTSUID; break; | |
2881 | case 'g': ftst = OP_FTSGID; break; | |
2882 | case 'k': ftst = OP_FTSVTX; break; | |
2883 | case 'b': ftst = OP_FTBLK; break; | |
2884 | case 'c': ftst = OP_FTCHR; break; | |
2885 | case 't': ftst = OP_FTTTY; break; | |
2886 | case 'T': ftst = OP_FTTEXT; break; | |
2887 | case 'B': ftst = OP_FTBINARY; break; | |
2888 | case 'M': case 'A': case 'C': | |
2889 | gv_fetchpv("\024",TRUE, SVt_PV); | |
2890 | switch (tmp) { | |
2891 | case 'M': ftst = OP_FTMTIME; break; | |
2892 | case 'A': ftst = OP_FTATIME; break; | |
2893 | case 'C': ftst = OP_FTCTIME; break; | |
2894 | default: break; | |
2895 | } | |
2896 | break; | |
378cc40b | 2897 | default: |
378cc40b LW |
2898 | break; |
2899 | } | |
e5edeb50 JH |
2900 | if (ftst) { |
2901 | PL_last_lop_op = ftst; | |
4e553d73 | 2902 | DEBUG_T( { PerlIO_printf(Perl_debug_log, |
0844c848 | 2903 | "### Saw file test %c\n", (int)ftst); |
e5edeb50 | 2904 | } ) |
e5edeb50 JH |
2905 | FTST(ftst); |
2906 | } | |
2907 | else { | |
2908 | /* Assume it was a minus followed by a one-letter named | |
2909 | * subroutine call (or a -bareword), then. */ | |
95c31fe3 | 2910 | DEBUG_T( { PerlIO_printf(Perl_debug_log, |
0844c848 RB |
2911 | "### %c looked like a file test but was not\n", |
2912 | (int)ftst); | |
95c31fe3 | 2913 | } ) |
e5edeb50 JH |
2914 | s -= 2; |
2915 | } | |
378cc40b | 2916 | } |
a687059c LW |
2917 | tmp = *s++; |
2918 | if (*s == tmp) { | |
2919 | s++; | |
3280af22 | 2920 | if (PL_expect == XOPERATOR) |
79072805 LW |
2921 | TERM(POSTDEC); |
2922 | else | |
2923 | OPERATOR(PREDEC); | |
2924 | } | |
2925 | else if (*s == '>') { | |
2926 | s++; | |
2927 | s = skipspace(s); | |
7e2040f0 | 2928 | if (isIDFIRST_lazy_if(s,UTF)) { |
a0d0e21e | 2929 | s = force_word(s,METHOD,FALSE,TRUE,FALSE); |
463ee0b2 | 2930 | TOKEN(ARROW); |
79072805 | 2931 | } |
748a9306 LW |
2932 | else if (*s == '$') |
2933 | OPERATOR(ARROW); | |
463ee0b2 | 2934 | else |
748a9306 | 2935 | TERM(ARROW); |
a687059c | 2936 | } |
3280af22 | 2937 | if (PL_expect == XOPERATOR) |
79072805 LW |
2938 | Aop(OP_SUBTRACT); |
2939 | else { | |
3280af22 | 2940 | if (isSPACE(*s) || !isSPACE(*PL_bufptr)) |
2f3197b3 | 2941 | check_uni(); |
79072805 | 2942 | OPERATOR('-'); /* unary minus */ |
2f3197b3 | 2943 | } |
79072805 | 2944 | |
378cc40b | 2945 | case '+': |
a687059c LW |
2946 | tmp = *s++; |
2947 | if (*s == tmp) { | |
378cc40b | 2948 | s++; |
3280af22 | 2949 | if (PL_expect == XOPERATOR) |
79072805 LW |
2950 | TERM(POSTINC); |
2951 | else | |
2952 | OPERATOR(PREINC); | |
378cc40b | 2953 | } |
3280af22 | 2954 | if (PL_expect == XOPERATOR) |
79072805 LW |
2955 | Aop(OP_ADD); |
2956 | else { | |
3280af22 | 2957 | if (isSPACE(*s) || !isSPACE(*PL_bufptr)) |
2f3197b3 | 2958 | check_uni(); |
a687059c | 2959 | OPERATOR('+'); |
2f3197b3 | 2960 | } |
a687059c | 2961 | |
378cc40b | 2962 | case '*': |
3280af22 NIS |
2963 | if (PL_expect != XOPERATOR) { |
2964 | s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE); | |
2965 | PL_expect = XOPERATOR; | |
2966 | force_ident(PL_tokenbuf, '*'); | |
2967 | if (!*PL_tokenbuf) | |
a0d0e21e | 2968 | PREREF('*'); |
79072805 | 2969 | TERM('*'); |
a687059c | 2970 | } |
79072805 LW |
2971 | s++; |
2972 | if (*s == '*') { | |
a687059c | 2973 | s++; |
79072805 | 2974 | PWop(OP_POW); |
a687059c | 2975 | } |
79072805 LW |
2976 | Mop(OP_MULTIPLY); |
2977 | ||
378cc40b | 2978 | case '%': |
3280af22 | 2979 | if (PL_expect == XOPERATOR) { |
bbce6d69 | 2980 | ++s; |
2981 | Mop(OP_MODULO); | |
a687059c | 2982 | } |
3280af22 NIS |
2983 | PL_tokenbuf[0] = '%'; |
2984 | s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE); | |
2985 | if (!PL_tokenbuf[1]) { | |
2986 | if (s == PL_bufend) | |
bbce6d69 | 2987 | yyerror("Final % should be \\% or %name"); |
2988 | PREREF('%'); | |
a687059c | 2989 | } |
3280af22 | 2990 | PL_pending_ident = '%'; |
bbce6d69 | 2991 | TERM('%'); |
a687059c | 2992 | |
378cc40b | 2993 | case '^': |
79072805 | 2994 | s++; |
a0d0e21e | 2995 | BOop(OP_BIT_XOR); |
79072805 | 2996 | case '[': |
3280af22 | 2997 | PL_lex_brackets++; |
79072805 | 2998 | /* FALL THROUGH */ |
378cc40b | 2999 | case '~': |
378cc40b | 3000 | case ',': |
378cc40b LW |
3001 | tmp = *s++; |
3002 | OPERATOR(tmp); | |
a0d0e21e LW |
3003 | case ':': |
3004 | if (s[1] == ':') { | |
3005 | len = 0; | |
3006 | goto just_a_word; | |
3007 | } | |
3008 | s++; | |
09bef843 SB |
3009 | switch (PL_expect) { |
3010 | OP *attrs; | |
3011 | case XOPERATOR: | |
3012 | if (!PL_in_my || PL_lex_state != LEX_NORMAL) | |
3013 | break; | |
3014 | PL_bufptr = s; /* update in case we back off */ | |
3015 | goto grabattrs; | |
3016 | case XATTRBLOCK: | |
3017 | PL_expect = XBLOCK; | |
3018 | goto grabattrs; | |
3019 | case XATTRTERM: | |
3020 | PL_expect = XTERMBLOCK; | |
3021 | grabattrs: | |
3022 | s = skipspace(s); | |
3023 | attrs = Nullop; | |
7e2040f0 | 3024 | while (isIDFIRST_lazy_if(s,UTF)) { |
09bef843 | 3025 | d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len); |
f9829d6b GS |
3026 | if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) { |
3027 | if (tmp < 0) tmp = -tmp; | |
3028 | switch (tmp) { | |
3029 | case KEY_or: | |
3030 | case KEY_and: | |
3031 | case KEY_for: | |
3032 | case KEY_unless: | |
3033 | case KEY_if: | |
3034 | case KEY_while: | |
3035 | case KEY_until: | |
3036 | goto got_attrs; | |
3037 | default: | |
3038 | break; | |
3039 | } | |
3040 | } | |
09bef843 SB |
3041 | if (*d == '(') { |
3042 | d = scan_str(d,TRUE,TRUE); | |
3043 | if (!d) { | |
3044 | if (PL_lex_stuff) { | |
3045 | SvREFCNT_dec(PL_lex_stuff); | |
3046 | PL_lex_stuff = Nullsv; | |
3047 | } | |
3048 | /* MUST advance bufptr here to avoid bogus | |
3049 | "at end of line" context messages from yyerror(). | |
3050 | */ | |
3051 | PL_bufptr = s + len; | |
3052 | yyerror("Unterminated attribute parameter in attribute list"); | |
3053 | if (attrs) | |
3054 | op_free(attrs); | |
3055 | return 0; /* EOF indicator */ | |
3056 | } | |
3057 | } | |
3058 | if (PL_lex_stuff) { | |
3059 | SV *sv = newSVpvn(s, len); | |
3060 | sv_catsv(sv, PL_lex_stuff); | |
3061 | attrs = append_elem(OP_LIST, attrs, | |
3062 | newSVOP(OP_CONST, 0, sv)); | |
3063 | SvREFCNT_dec(PL_lex_stuff); | |
3064 | PL_lex_stuff = Nullsv; | |
3065 | } | |
3066 | else { | |
78f9721b SM |
3067 | if (!PL_in_my && len == 6 && strnEQ(s, "lvalue", len)) |
3068 | CvLVALUE_on(PL_compcv); | |
3069 | else if (!PL_in_my && len == 6 && strnEQ(s, "locked", len)) | |
3070 | CvLOCKED_on(PL_compcv); | |
3071 | else if (!PL_in_my && len == 6 && strnEQ(s, "method", len)) | |
3072 | CvMETHOD_on(PL_compcv); | |
87ecf892 DM |
3073 | #ifdef USE_ITHREADS |
3074 | else if (PL_in_my == KEY_our && len == 6 && strnEQ(s, "shared", len)) | |
3075 | GvSHARED_on(cGVOPx_gv(yylval.opval)); | |
3076 | #endif | |
78f9721b SM |
3077 | /* After we've set the flags, it could be argued that |
3078 | we don't need to do the attributes.pm-based setting | |
3079 | process, and shouldn't bother appending recognized | |
3080 | flags. To experiment with that, uncomment the | |
3081 | following "else": */ | |
0256094b | 3082 | else |
78f9721b SM |
3083 | attrs = append_elem(OP_LIST, attrs, |
3084 | newSVOP(OP_CONST, 0, | |
3085 | newSVpvn(s, len))); | |
09bef843 SB |
3086 | } |
3087 | s = skipspace(d); | |
0120eecf | 3088 | if (*s == ':' && s[1] != ':') |
09bef843 | 3089 | s = skipspace(s+1); |
0120eecf GS |
3090 | else if (s == d) |
3091 | break; /* require real whitespace or :'s */ | |
09bef843 | 3092 | } |
f9829d6b GS |
3093 | tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */ |
3094 | if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) { | |
09bef843 SB |
3095 | char q = ((*s == '\'') ? '"' : '\''); |
3096 | /* If here for an expression, and parsed no attrs, back off. */ | |
3097 | if (tmp == '=' && !attrs) { | |
3098 | s = PL_bufptr; | |
3099 | break; | |
3100 | } | |
3101 | /* MUST advance bufptr here to avoid bogus "at end of line" | |
3102 | context messages from yyerror(). | |
3103 | */ | |
3104 | PL_bufptr = s; | |
3105 | if (!*s) | |
3106 | yyerror("Unterminated attribute list"); | |
3107 | else | |
3108 | yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list", | |
3109 | q, *s, q)); | |
3110 | if (attrs) | |
3111 | op_free(attrs); | |
3112 | OPERATOR(':'); | |
3113 | } | |
f9829d6b | 3114 | got_attrs: |
09bef843 SB |
3115 | if (attrs) { |
3116 | PL_nextval[PL_nexttoke].opval = attrs; | |
3117 | force_next(THING); | |
3118 | } | |
3119 | TOKEN(COLONATTR); | |
3120 | } | |
a0d0e21e | 3121 | OPERATOR(':'); |
8990e307 LW |
3122 | case '(': |
3123 | s++; | |
3280af22 NIS |
3124 | if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr) |
3125 | PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */ | |
a0d0e21e | 3126 | else |
3280af22 | 3127 | PL_expect = XTERM; |
a0d0e21e | 3128 | TOKEN('('); |
378cc40b | 3129 | case ';': |
f4dd75d9 | 3130 | CLINE; |
378cc40b LW |
3131 | tmp = *s++; |
3132 | OPERATOR(tmp); | |
3133 | case ')': | |
378cc40b | 3134 | tmp = *s++; |
16d20bd9 AD |
3135 | s = skipspace(s); |
3136 | if (*s == '{') | |
3137 | PREBLOCK(tmp); | |
378cc40b | 3138 | TERM(tmp); |
79072805 LW |
3139 | case ']': |
3140 | s++; | |
3280af22 | 3141 | if (PL_lex_brackets <= 0) |
d98d5fff | 3142 | yyerror("Unmatched right square bracket"); |
463ee0b2 | 3143 | else |
3280af22 NIS |
3144 | --PL_lex_brackets; |
3145 | if (PL_lex_state == LEX_INTERPNORMAL) { | |
3146 | if (PL_lex_brackets == 0) { | |
a0d0e21e | 3147 | if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>')) |
3280af22 | 3148 | PL_lex_state = LEX_INTERPEND; |
79072805 LW |
3149 | } |
3150 | } | |
4633a7c4 | 3151 | TERM(']'); |
79072805 LW |
3152 | case '{': |
3153 | leftbracket: | |
79072805 | 3154 | s++; |
3280af22 NIS |
3155 | if (PL_lex_brackets > 100) { |
3156 | char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char); | |
3157 | if (newlb != PL_lex_brackstack) { | |
8990e307 | 3158 | SAVEFREEPV(newlb); |
3280af22 | 3159 | PL_lex_brackstack = newlb; |
8990e307 LW |
3160 | } |
3161 | } | |
3280af22 | 3162 | switch (PL_expect) { |
a0d0e21e | 3163 | case XTERM: |
3280af22 | 3164 | if (PL_lex_formbrack) { |
a0d0e21e LW |
3165 | s--; |
3166 | PRETERMBLOCK(DO); | |
3167 | } | |
3280af22 NIS |
3168 | if (PL_oldoldbufptr == PL_last_lop) |
3169 | PL_lex_brackstack[PL_lex_brackets++] = XTERM; | |
a0d0e21e | 3170 | else |
3280af22 | 3171 | PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR; |
79072805 | 3172 | OPERATOR(HASHBRACK); |
a0d0e21e | 3173 | case XOPERATOR: |
bf4acbe4 | 3174 | while (s < PL_bufend && SPACE_OR_TAB(*s)) |
748a9306 | 3175 | s++; |
44a8e56a | 3176 | d = s; |
3280af22 NIS |
3177 | PL_tokenbuf[0] = '\0'; |
3178 | if (d < PL_bufend && *d == '-') { | |
3179 | PL_tokenbuf[0] = '-'; | |
44a8e56a | 3180 | d++; |
bf4acbe4 | 3181 | while (d < PL_bufend && SPACE_OR_TAB(*d)) |
44a8e56a | 3182 | d++; |
3183 | } | |
7e2040f0 | 3184 | if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) { |
3280af22 | 3185 | d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, |
8903cb82 | 3186 | FALSE, &len); |
bf4acbe4 | 3187 | while (d < PL_bufend && SPACE_OR_TAB(*d)) |
748a9306 LW |
3188 | d++; |
3189 | if (*d == '}') { | |
3280af22 | 3190 | char minus = (PL_tokenbuf[0] == '-'); |
44a8e56a | 3191 | s = force_word(s + minus, WORD, FALSE, TRUE, FALSE); |
7948272d NIS |
3192 | if (UTF && !IN_BYTE && is_utf8_string((U8*)PL_tokenbuf, 0) && |
3193 | PL_nextval[PL_nexttoke-1].opval) | |
3194 | SvUTF8_on(((SVOP*)PL_nextval[PL_nexttoke-1].opval)->op_sv); | |
44a8e56a | 3195 | if (minus) |
3196 | force_next('-'); | |
748a9306 LW |
3197 | } |
3198 | } | |
3199 | /* FALL THROUGH */ | |
09bef843 | 3200 | case XATTRBLOCK: |
748a9306 | 3201 | case XBLOCK: |
3280af22 NIS |
3202 | PL_lex_brackstack[PL_lex_brackets++] = XSTATE; |
3203 | PL_expect = XSTATE; | |
a0d0e21e | 3204 | break; |
09bef843 | 3205 | case XATTRTERM: |
a0d0e21e | 3206 | case XTERMBLOCK: |
3280af22 NIS |
3207 | PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR; |
3208 | PL_expect = XSTATE; | |
a0d0e21e LW |
3209 | break; |
3210 | default: { | |
3211 | char *t; | |
3280af22 NIS |
3212 | if (PL_oldoldbufptr == PL_last_lop) |
3213 | PL_lex_brackstack[PL_lex_brackets++] = XTERM; | |
a0d0e21e | 3214 | else |
3280af22 | 3215 | PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR; |
a0d0e21e | 3216 | s = skipspace(s); |
09ecc4b6 | 3217 | if (*s == '}') |
a0d0e21e | 3218 | OPERATOR(HASHBRACK); |
b8a4b1be GS |
3219 | /* This hack serves to disambiguate a pair of curlies |
3220 | * as being a block or an anon hash. Normally, expectation | |
3221 | * determines that, but in cases where we're not in a | |
3222 | * position to expect anything in particular (like inside | |
3223 | * eval"") we have to resolve the ambiguity. This code | |
3224 | * covers the case where the first term in the curlies is a | |
3225 | * quoted string. Most other cases need to be explicitly | |
3226 | * disambiguated by prepending a `+' before the opening | |
3227 | * curly in order to force resolution as an anon hash. | |
3228 | * | |
3229 | * XXX should probably propagate the outer expectation | |
3230 | * into eval"" to rely less on this hack, but that could | |
3231 | * potentially break current behavior of eval"". | |
3232 | * GSAR 97-07-21 | |
3233 | */ | |
3234 | t = s; | |
3235 | if (*s == '\'' || *s == '"' || *s == '`') { | |
3236 | /* common case: get past first string, handling escapes */ | |
3280af22 | 3237 | for (t++; t < PL_bufend && *t != *s;) |
b8a4b1be GS |
3238 | if (*t++ == '\\' && (*t == '\\' || *t == *s)) |
3239 | t++; | |
3240 | t++; | |
a0d0e21e | 3241 | } |
b8a4b1be | 3242 | else if (*s == 'q') { |
3280af22 | 3243 | if (++t < PL_bufend |
b8a4b1be | 3244 | && (!isALNUM(*t) |
3280af22 | 3245 | || ((*t == 'q' || *t == 'x') && ++t < PL_bufend |
0505442f GS |
3246 | && !isALNUM(*t)))) |
3247 | { | |
b8a4b1be GS |
3248 | char *tmps; |
3249 | char open, close, term; | |
3250 | I32 brackets = 1; | |
3251 | ||
3280af22 | 3252 | while (t < PL_bufend && isSPACE(*t)) |
b8a4b1be GS |
3253 | t++; |
3254 | term = *t; | |
3255 | open = term; | |
3256 | if (term && (tmps = strchr("([{< )]}> )]}>",term))) | |
3257 | term = tmps[5]; | |
3258 | close = term; | |
3259 | if (open == close) | |
3280af22 NIS |
3260 | for (t++; t < PL_bufend; t++) { |
3261 | if (*t == '\\' && t+1 < PL_bufend && open != '\\') | |
b8a4b1be | 3262 | t++; |
6d07e5e9 | 3263 | else if (*t == open) |
b8a4b1be GS |
3264 | break; |
3265 | } | |
3266 | else | |
3280af22 NIS |
3267 | for (t++; t < PL_bufend; t++) { |
3268 | if (*t == '\\' && t+1 < PL_bufend) | |
b8a4b1be | 3269 | t++; |
6d07e5e9 | 3270 | else if (*t == close && --brackets <= 0) |
b8a4b1be GS |
3271 | break; |
3272 | else if (*t == open) | |
3273 | brackets++; | |
3274 | } | |
3275 | } | |
3276 | t++; | |
a0d0e21e | 3277 | } |
7e2040f0 | 3278 | else if (isALNUM_lazy_if(t,UTF)) { |
0505442f | 3279 | t += UTF8SKIP(t); |
7e2040f0 | 3280 | while (t < PL_bufend && isALNUM_lazy_if(t,UTF)) |
0505442f | 3281 | t += UTF8SKIP(t); |
a0d0e21e | 3282 | } |
3280af22 | 3283 | while (t < PL_bufend && isSPACE(*t)) |
a0d0e21e | 3284 | t++; |
b8a4b1be GS |
3285 | /* if comma follows first term, call it an anon hash */ |
3286 | /* XXX it could be a comma expression with loop modifiers */ | |
3280af22 | 3287 | if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s))) |
b8a4b1be | 3288 | || (*t == '=' && t[1] == '>'))) |
a0d0e21e | 3289 | OPERATOR(HASHBRACK); |
3280af22 | 3290 | if (PL_expect == XREF) |
4e4e412b | 3291 | PL_expect = XTERM; |
a0d0e21e | 3292 | else { |
3280af22 NIS |
3293 | PL_lex_brackstack[PL_lex_brackets-1] = XSTATE; |
3294 | PL_expect = XSTATE; | |
a0d0e21e | 3295 | } |
8990e307 | 3296 | } |
a0d0e21e | 3297 | break; |
463ee0b2 | 3298 | } |
57843af0 | 3299 | yylval.ival = CopLINE(PL_curcop); |
79072805 | 3300 | if (isSPACE(*s) || *s == '#') |
3280af22 | 3301 | PL_copline = NOLINE; /* invalidate current command line number */ |
79072805 | 3302 | TOKEN('{'); |
378cc40b | 3303 | case '}': |
79072805 LW |
3304 | rightbracket: |
3305 | s++; | |
3280af22 | 3306 | if (PL_lex_brackets <= 0) |
d98d5fff | 3307 | yyerror("Unmatched right curly bracket"); |
463ee0b2 | 3308 | else |
3280af22 | 3309 | PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets]; |
c2e66d9e | 3310 | if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL) |
3280af22 NIS |
3311 | PL_lex_formbrack = 0; |
3312 | if (PL_lex_state == LEX_INTERPNORMAL) { | |
3313 | if (PL_lex_brackets == 0) { | |
9059aa12 LW |
3314 | if (PL_expect & XFAKEBRACK) { |
3315 | PL_expect &= XENUMMASK; | |
3280af22 NIS |
3316 | PL_lex_state = LEX_INTERPEND; |
3317 | PL_bufptr = s; | |
cea2e8a9 | 3318 | return yylex(); /* ignore fake brackets */ |
79072805 | 3319 | } |
fa83b5b6 | 3320 | if (*s == '-' && s[1] == '>') |
3280af22 | 3321 | PL_lex_state = LEX_INTERPENDMAYBE; |
fa83b5b6 | 3322 | else if (*s != '[' && *s != '{') |
3280af22 | 3323 | PL_lex_state = LEX_INTERPEND; |
79072805 LW |
3324 | } |
3325 | } | |
9059aa12 LW |
3326 | if (PL_expect & XFAKEBRACK) { |
3327 | PL_expect &= XENUMMASK; | |
3280af22 | 3328 | PL_bufptr = s; |
cea2e8a9 | 3329 | return yylex(); /* ignore fake brackets */ |
748a9306 | 3330 | } |
79072805 LW |
3331 | force_next('}'); |
3332 | TOKEN(';'); | |
378cc40b LW |
3333 | case '&': |
3334 | s++; | |
3335 | tmp = *s++; | |
3336 | if (tmp == '&') | |
a0d0e21e | 3337 | AOPERATOR(ANDAND); |
378cc40b | 3338 | s--; |
3280af22 | 3339 | if (PL_expect == XOPERATOR) { |
7e2040f0 GS |
3340 | if (ckWARN(WARN_SEMICOLON) |
3341 | && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart) | |
3342 | { | |
57843af0 | 3343 | CopLINE_dec(PL_curcop); |
cea2e8a9 | 3344 | Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi); |
57843af0 | 3345 | CopLINE_inc(PL_curcop); |
463ee0b2 | 3346 | } |
79072805 | 3347 | BAop(OP_BIT_AND); |
463ee0b2 | 3348 | } |
79072805 | 3349 | |
3280af22 NIS |
3350 | s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE); |
3351 | if (*PL_tokenbuf) { | |
3352 | PL_expect = XOPERATOR; | |
3353 | force_ident(PL_tokenbuf, '&'); | |
463ee0b2 | 3354 | } |
79072805 LW |
3355 | else |
3356 | PREREF('&'); | |
c07a80fd | 3357 | yylval.ival = (OPpENTERSUB_AMPER<<8); |
79072805 LW |
3358 | TERM('&'); |
3359 | ||
378cc40b LW |
3360 | case '|': |
3361 | s++; | |
3362 | tmp = *s++; | |
3363 | if (tmp == '|') | |
a0d0e21e | 3364 | AOPERATOR(OROR); |
378cc40b | 3365 | s--; |
79072805 | 3366 | BOop(OP_BIT_OR); |
378cc40b LW |
3367 | case '=': |
3368 | s++; | |
3369 | tmp = *s++; | |
3370 | if (tmp == '=') | |
79072805 LW |
3371 | Eop(OP_EQ); |
3372 | if (tmp == '>') | |
3373 | OPERATOR(','); | |
378cc40b | 3374 | if (tmp == '~') |
79072805 | 3375 | PMop(OP_MATCH); |
599cee73 | 3376 | if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp)) |
cea2e8a9 | 3377 | Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp); |
378cc40b | 3378 | s--; |
3280af22 NIS |
3379 | if (PL_expect == XSTATE && isALPHA(tmp) && |
3380 | (s == PL_linestart+1 || s[-2] == '\n') ) | |
748a9306 | 3381 | { |
3280af22 NIS |
3382 | if (PL_in_eval && !PL_rsfp) { |
3383 | d = PL_bufend; | |
a5f75d66 AD |
3384 | while (s < d) { |
3385 | if (*s++ == '\n') { | |
3386 | incline(s); | |
3387 | if (strnEQ(s,"=cut",4)) { | |
3388 | s = strchr(s,'\n'); | |
3389 | if (s) | |
3390 | s++; | |
3391 | else | |
3392 | s = d; | |
3393 | incline(s); | |
3394 | goto retry; | |
3395 | } | |
3396 | } | |
3397 | } | |
3398 | goto retry; | |
3399 | } | |
3280af22 NIS |
3400 | s = PL_bufend; |
3401 | PL_doextract = TRUE; | |
a0d0e21e LW |
3402 | goto retry; |
3403 | } | |
3280af22 | 3404 | if (PL_lex_brackets < PL_lex_formbrack) { |
a0d0e21e | 3405 | char *t; |
51882d45 | 3406 | #ifdef PERL_STRICT_CR |
bf4acbe4 | 3407 | for (t = s; SPACE_OR_TAB(*t); t++) ; |
51882d45 | 3408 | #else |
bf4acbe4 | 3409 | for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ; |
51882d45 | 3410 | #endif |
a0d0e21e LW |
3411 | if (*t == '\n' || *t == '#') { |
3412 | s--; | |
3280af22 | 3413 | PL_expect = XBLOCK; |
a0d0e21e LW |
3414 | goto leftbracket; |
3415 | } | |
79072805 | 3416 | } |
a0d0e21e LW |
3417 | yylval.ival = 0; |
3418 | OPERATOR(ASSIGNOP); | |
378cc40b LW |
3419 | case '!': |
3420 | s++; | |
3421 | tmp = *s++; | |
3422 | if (tmp == '=') | |
79072805 | 3423 | Eop(OP_NE); |
378cc40b | 3424 | if (tmp == '~') |
79072805 | 3425 | PMop(OP_NOT); |
378cc40b LW |
3426 | s--; |
3427 | OPERATOR('!'); | |
3428 | case '<': | |
3280af22 | 3429 | if (PL_expect != XOPERATOR) { |
93a17b20 | 3430 | if (s[1] != '<' && !strchr(s,'>')) |
2f3197b3 | 3431 | check_uni(); |
79072805 LW |
3432 | if (s[1] == '<') |
3433 | s = scan_heredoc(s); | |
3434 | else | |
3435 | s = scan_inputsymbol(s); | |
3436 | TERM(sublex_start()); | |
378cc40b LW |
3437 | } |
3438 | s++; | |
3439 | tmp = *s++; | |
3440 | if (tmp == '<') | |
79072805 | 3441 | SHop(OP_LEFT_SHIFT); |
395c3793 LW |
3442 | if (tmp == '=') { |
3443 | tmp = *s++; | |
3444 | if (tmp == '>') | |
79072805 | 3445 | Eop(OP_NCMP); |
395c3793 | 3446 | s--; |
79072805 | 3447 | Rop(OP_LE); |
395c3793 | 3448 | } |
378cc40b | 3449 | s--; |
79072805 | 3450 | Rop(OP_LT); |
378cc40b LW |
3451 | case '>': |
3452 | s++; | |
3453 | tmp = *s++; | |
3454 | if (tmp == '>') | |
79072805 | 3455 | SHop(OP_RIGHT_SHIFT); |
378cc40b | 3456 | if (tmp == '=') |
79072805 | 3457 | Rop(OP_GE); |
378cc40b | 3458 | s--; |
79072805 | 3459 | Rop(OP_GT); |
378cc40b LW |
3460 | |
3461 | case '$': | |
bbce6d69 | 3462 | CLINE; |
3463 | ||
3280af22 NIS |
3464 | if (PL_expect == XOPERATOR) { |
3465 | if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) { | |
3466 | PL_expect = XTERM; | |
a0d0e21e | 3467 | depcom(); |
bbce6d69 | 3468 | return ','; /* grandfather non-comma-format format */ |
a0d0e21e | 3469 | } |
8990e307 | 3470 | } |
a0d0e21e | 3471 | |
7e2040f0 | 3472 | if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) { |
3280af22 | 3473 | PL_tokenbuf[0] = '@'; |
376b8730 SM |
3474 | s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1, |
3475 | sizeof PL_tokenbuf - 1, FALSE); | |
3476 | if (PL_expect == XOPERATOR) | |
3477 | no_op("Array length", s); | |
3280af22 | 3478 | if (!PL_tokenbuf[1]) |
a0d0e21e | 3479 | PREREF(DOLSHARP); |
3280af22 NIS |
3480 | PL_expect = XOPERATOR; |
3481 | PL_pending_ident = '#'; | |
463ee0b2 | 3482 | TOKEN(DOLSHARP); |
79072805 | 3483 | } |
bbce6d69 | 3484 | |
3280af22 | 3485 | PL_tokenbuf[0] = '$'; |
376b8730 SM |
3486 | s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, |
3487 | sizeof PL_tokenbuf - 1, FALSE); | |
3488 | if (PL_expect == XOPERATOR) | |
3489 | no_op("Scalar", s); | |
3280af22 NIS |
3490 | if (!PL_tokenbuf[1]) { |
3491 | if (s == PL_bufend) | |
bbce6d69 | 3492 | yyerror("Final $ should be \\$ or $name"); |
3493 | PREREF('$'); | |
8990e307 | 3494 | } |
a0d0e21e | 3495 | |
bbce6d69 | 3496 | /* This kludge not intended to be bulletproof. */ |
3280af22 | 3497 | if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) { |
bbce6d69 | 3498 | yylval.opval = newSVOP(OP_CONST, 0, |
b448e4fe | 3499 | newSViv(PL_compiling.cop_arybase)); |
bbce6d69 | 3500 | yylval.opval->op_private = OPpCONST_ARYBASE; |
3501 | TERM(THING); | |
3502 | } | |
3503 | ||
ff68c719 | 3504 | d = s; |
69d2bceb | 3505 | tmp = (I32)*s; |
3280af22 | 3506 | if (PL_lex_state == LEX_NORMAL) |
ff68c719 | 3507 | s = skipspace(s); |
3508 | ||
3280af22 | 3509 | if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) { |
bbce6d69 | 3510 | char *t; |
3511 | if (*s == '[') { | |
3280af22 | 3512 | PL_tokenbuf[0] = '@'; |
599cee73 | 3513 | if (ckWARN(WARN_SYNTAX)) { |
bbce6d69 | 3514 | for(t = s + 1; |
7e2040f0 | 3515 | isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$'; |
bbce6d69 | 3516 | t++) ; |
a0d0e21e | 3517 | if (*t++ == ',') { |
3280af22 NIS |
3518 | PL_bufptr = skipspace(PL_bufptr); |
3519 | while (t < PL_bufend && *t != ']') | |
bbce6d69 | 3520 | t++; |
cea2e8a9 | 3521 | Perl_warner(aTHX_ WARN_SYNTAX, |
599cee73 PM |
3522 | "Multidimensional syntax %.*s not supported", |
3523 | (t - PL_bufptr) + 1, PL_bufptr); | |
a0d0e21e LW |
3524 | } |
3525 | } | |
bbce6d69 | 3526 | } |
3527 | else if (*s == '{') { | |
3280af22 | 3528 | PL_tokenbuf[0] = '%'; |
599cee73 | 3529 | if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") && |
bbce6d69 | 3530 | (t = strchr(s, '}')) && (t = strchr(t, '='))) |
3531 | { | |
3280af22 | 3532 | char tmpbuf[sizeof PL_tokenbuf]; |
a0d0e21e LW |
3533 | STRLEN len; |
3534 | for (t++; isSPACE(*t); t++) ; | |
7e2040f0 | 3535 | if (isIDFIRST_lazy_if(t,UTF)) { |
8903cb82 | 3536 | t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len); |
59a6d928 | 3537 | for (; isSPACE(*t); t++) ; |
864dbfa3 | 3538 | if (*t == ';' && get_cv(tmpbuf, FALSE)) |
cea2e8a9 | 3539 | Perl_warner(aTHX_ WARN_SYNTAX, |
599cee73 | 3540 | "You need to quote \"%s\"", tmpbuf); |
748a9306 | 3541 | } |
93a17b20 LW |
3542 | } |
3543 | } | |
2f3197b3 | 3544 | } |
bbce6d69 | 3545 | |
3280af22 | 3546 | PL_expect = XOPERATOR; |
69d2bceb | 3547 | if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) { |
3280af22 NIS |
3548 | bool islop = (PL_last_lop == PL_oldoldbufptr); |
3549 | if (!islop || PL_last_lop_op == OP_GREPSTART) | |
3550 | PL_expect = XOPERATOR; | |
bbce6d69 | 3551 | else if (strchr("$@\"'`q", *s)) |
3280af22 | 3552 | PL_expect = XTERM; /* e.g. print $fh "foo" */ |
7e2040f0 | 3553 | else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF)) |
3280af22 | 3554 | PL_expect = XTERM; /* e.g. print $fh &sub */ |
7e2040f0 | 3555 | else if (isIDFIRST_lazy_if(s,UTF)) { |
3280af22 | 3556 | char tmpbuf[sizeof PL_tokenbuf]; |
8903cb82 | 3557 | scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len); |
155aba94 | 3558 | if ((tmp = keyword(tmpbuf, len))) { |
84902520 TB |
3559 | /* binary operators exclude handle interpretations */ |
3560 | switch (tmp) { | |
3561 | case -KEY_x: | |
3562 | case -KEY_eq: | |
3563 | case -KEY_ne: | |
3564 | case -KEY_gt: | |
3565 | case -KEY_lt: | |
3566 | case -KEY_ge: | |
3567 | case -KEY_le: | |
3568 | case -KEY_cmp: | |
3569 | break; | |
3570 | default: | |
3280af22 | 3571 | PL_expect = XTERM; /* e.g. print $fh length() */ |
84902520 TB |
3572 | break; |
3573 | } | |
3574 | } | |
68dc0745 | 3575 | else { |
3576 | GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV); | |
3577 | if (gv && GvCVu(gv)) | |
3280af22 | 3578 | PL_expect = XTERM; /* e.g. print $fh subr() */ |
93a17b20 | 3579 | } |
93a17b20 | 3580 | } |
bbce6d69 | 3581 | else if (isDIGIT(*s)) |
3280af22 | 3582 | PL_expect = XTERM; /* e.g. print $fh 3 */ |
bbce6d69 | 3583 | else if (*s == '.' && isDIGIT(s[1])) |
3280af22 | 3584 | PL_expect = XTERM; /* e.g. print $fh .3 */ |
e0587a03 | 3585 | else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=') |
3280af22 | 3586 | PL_expect = XTERM; /* e.g. print $fh -1 */ |
e0587a03 | 3587 | else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=') |
3280af22 | 3588 | PL_expect = XTERM; /* print $fh <<"EOF" */ |
bbce6d69 | 3589 | } |
3280af22 | 3590 | PL_pending_ident = '$'; |
79072805 | 3591 | TOKEN('$'); |
378cc40b LW |
3592 | |
3593 | case '@': | |
3280af22 | 3594 | if (PL_expect == XOPERATOR) |
bbce6d69 | 3595 | no_op("Array", s); |
3280af22 NIS |
3596 | PL_tokenbuf[0] = '@'; |
3597 | s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE); | |
3598 | if (!PL_tokenbuf[1]) { | |
3599 | if (s == PL_bufend) | |
bbce6d69 | 3600 | yyerror("Final @ should be \\@ or @name"); |
3601 | PREREF('@'); | |
3602 | } | |
3280af22 | 3603 | if (PL_lex_state == LEX_NORMAL) |
ff68c719 | 3604 | s = skipspace(s); |
3280af22 | 3605 | if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) { |
bbce6d69 | 3606 | if (*s == '{') |
3280af22 | 3607 | PL_tokenbuf[0] = '%'; |
a0d0e21e LW |
3608 | |
3609 | /* Warn about @ where they meant $. */ | |
599cee73 | 3610 | if (ckWARN(WARN_SYNTAX)) { |
a0d0e21e LW |
3611 | if (*s == '[' || *s == '{') { |
3612 | char *t = s + 1; | |
7e2040f0 | 3613 | while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t))) |
a0d0e21e LW |
3614 | t++; |
3615 | if (*t == '}' || *t == ']') { | |
3616 | t++; | |
3280af22 | 3617 | PL_bufptr = skipspace(PL_bufptr); |
cea2e8a9 | 3618 | Perl_warner(aTHX_ WARN_SYNTAX, |
599cee73 | 3619 | "Scalar value %.*s better written as $%.*s", |
3280af22 | 3620 | t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1); |
a0d0e21e | 3621 | } |
93a17b20 LW |
3622 | } |
3623 | } | |
463ee0b2 | 3624 | } |
3280af22 | 3625 | PL_pending_ident = '@'; |
79072805 | 3626 | TERM('@'); |
378cc40b LW |
3627 | |
3628 | case '/': /* may either be division or pattern */ | |
3629 | case '?': /* may either be conditional or pattern */ | |
3280af22 | 3630 | if (PL_expect != XOPERATOR) { |
c277df42 | 3631 | /* Disable warning on "study /blah/" */ |
4e553d73 NIS |
3632 | if (PL_oldoldbufptr == PL_last_uni |
3633 | && (*PL_last_uni != 's' || s - PL_last_uni < 5 | |
7e2040f0 GS |
3634 | || memNE(PL_last_uni, "study", 5) |
3635 | || isALNUM_lazy_if(PL_last_uni+5,UTF))) | |
c277df42 | 3636 | check_uni(); |
8782bef2 | 3637 | s = scan_pat(s,OP_MATCH); |
79072805 | 3638 | TERM(sublex_start()); |
378cc40b LW |
3639 | } |
3640 | tmp = *s++; | |
a687059c | 3641 | if (tmp == '/') |
79072805 | 3642 | Mop(OP_DIVIDE); |
378cc40b LW |
3643 | OPERATOR(tmp); |
3644 | ||
3645 | case '.': | |
51882d45 GS |
3646 | if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack |
3647 | #ifdef PERL_STRICT_CR | |
3648 | && s[1] == '\n' | |
3649 | #else | |
3650 | && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n')) | |
3651 | #endif | |
3652 | && (s == PL_linestart || s[-1] == '\n') ) | |
3653 | { | |
3280af22 NIS |
3654 | PL_lex_formbrack = 0; |
3655 | PL_expect = XSTATE; | |
79072805 LW |
3656 | goto rightbracket; |
3657 | } | |
3280af22 | 3658 | if (PL_expect == XOPERATOR || !isDIGIT(s[1])) { |
378cc40b | 3659 | tmp = *s++; |
a687059c LW |
3660 | if (*s == tmp) { |
3661 | s++; | |
2f3197b3 LW |
3662 | if (*s == tmp) { |
3663 | s++; | |
79072805 | 3664 | yylval.ival = OPf_SPECIAL; |
2f3197b3 LW |
3665 | } |
3666 | else | |
79072805 | 3667 | yylval.ival = 0; |
378cc40b | 3668 | OPERATOR(DOTDOT); |
a687059c | 3669 | } |
3280af22 | 3670 | if (PL_expect != XOPERATOR) |
2f3197b3 | 3671 | check_uni(); |
79072805 | 3672 | Aop(OP_CONCAT); |
378cc40b LW |
3673 | } |
3674 | /* FALL THROUGH */ | |
3675 | case '0': case '1': case '2': case '3': case '4': | |
3676 | case '5': case '6': case '7': case '8': case '9': | |
b73d6f50 | 3677 | s = scan_num(s, &yylval); |
4e553d73 | 3678 | DEBUG_T( { PerlIO_printf(Perl_debug_log, |
607df283 SC |
3679 | "### Saw number in '%s'\n", s); |
3680 | } ) | |
3280af22 | 3681 | if (PL_expect == XOPERATOR) |
8990e307 | 3682 | no_op("Number",s); |
79072805 LW |
3683 | TERM(THING); |
3684 | ||
3685 | case '\'': | |
09bef843 | 3686 | s = scan_str(s,FALSE,FALSE); |
4e553d73 | 3687 | DEBUG_T( { PerlIO_printf(Perl_debug_log, |
607df283 SC |
3688 | "### Saw string in '%s'\n", s); |
3689 | } ) | |
3280af22 NIS |
3690 | if (PL_expect == XOPERATOR) { |
3691 | if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) { | |
3692 | PL_expect = XTERM; | |
a0d0e21e LW |
3693 | depcom(); |
3694 | return ','; /* grandfather non-comma-format format */ | |
3695 | } | |
463ee0b2 | 3696 | else |
8990e307 | 3697 | no_op("String",s); |
463ee0b2 | 3698 | } |
79072805 | 3699 | if (!s) |
85e6fe83 | 3700 | missingterm((char*)0); |
79072805 LW |
3701 | yylval.ival = OP_CONST; |
3702 | TERM(sublex_start()); | |
3703 | ||
3704 | case '"': | |
09bef843 | 3705 | s = scan_str(s,FALSE,FALSE); |
4e553d73 | 3706 | DEBUG_T( { PerlIO_printf(Perl_debug_log, |
607df283 SC |
3707 | "### Saw string in '%s'\n", s); |
3708 | } ) | |
3280af22 NIS |
3709 | if (PL_expect == XOPERATOR) { |
3710 | if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) { | |
3711 | PL_expect = XTERM; | |
a0d0e21e LW |
3712 | depcom(); |
3713 | return ','; /* grandfather non-comma-format format */ | |
3714 | } | |
463ee0b2 | 3715 | else |
8990e307 | 3716 | no_op("String",s); |
463ee0b2 | 3717 | } |
79072805 | 3718 | if (!s) |
85e6fe83 | 3719 | missingterm((char*)0); |
4633a7c4 | 3720 | yylval.ival = OP_CONST; |
3280af22 | 3721 | for (d = SvPV(PL_lex_stuff, len); len; len--, d++) { |
fd400ab9 | 3722 | if (*d == '$' || *d == '@' || *d == '\\' || UTF8_IS_CONTINUED(*d)) { |
4633a7c4 LW |
3723 | yylval.ival = OP_STRINGIFY; |
3724 | break; | |
3725 | } | |
3726 | } | |
79072805 LW |
3727 | TERM(sublex_start()); |
3728 | ||
3729 | case '`': | |
09bef843 | 3730 | s = scan_str(s,FALSE,FALSE); |
4e553d73 | 3731 | DEBUG_T( { PerlIO_printf(Perl_debug_log, |
607df283 SC |
3732 | "### Saw backtick string in '%s'\n", s); |
3733 | } ) | |
3280af22 | 3734 | if (PL_expect == XOPERATOR) |
8990e307 | 3735 | no_op("Backticks",s); |
79072805 | 3736 | if (!s) |
85e6fe83 | 3737 | missingterm((char*)0); |
79072805 LW |
3738 | yylval.ival = OP_BACKTICK; |
3739 | set_csh(); | |
3740 | TERM(sublex_start()); | |
3741 | ||
3742 | case '\\': | |
3743 | s++; | |
599cee73 | 3744 | if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s)) |
cea2e8a9 | 3745 | Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression", |
599cee73 | 3746 | *s, *s); |
3280af22 | 3747 | if (PL_expect == XOPERATOR) |
8990e307 | 3748 | no_op("Backslash",s); |
79072805 LW |
3749 | OPERATOR(REFGEN); |
3750 | ||
a7cb1f99 | 3751 | case 'v': |
e526c9e6 | 3752 | if (isDIGIT(s[1]) && PL_expect != XOPERATOR) { |
a7cb1f99 GS |
3753 | char *start = s; |
3754 | start++; | |
3755 | start++; | |
dd629d5b | 3756 | while (isDIGIT(*start) || *start == '_') |
a7cb1f99 GS |
3757 | start++; |
3758 | if (*start == '.' && isDIGIT(start[1])) { | |
b73d6f50 | 3759 | s = scan_num(s, &yylval); |
a7cb1f99 GS |
3760 | TERM(THING); |
3761 | } | |
e526c9e6 GS |
3762 | /* avoid v123abc() or $h{v1}, allow C<print v10;> */ |
3763 | else if (!isALPHA(*start) && (PL_expect == XTERM || PL_expect == XREF)) { | |
3764 | char c = *start; | |
3765 | GV *gv; | |
3766 | *start = '\0'; | |
3767 | gv = gv_fetchpv(s, FALSE, SVt_PVCV); | |
3768 | *start = c; | |
3769 | if (!gv) { | |
b73d6f50 | 3770 | s = scan_num(s, &yylval); |
e526c9e6 GS |
3771 | TERM(THING); |
3772 | } | |
3773 | } | |
a7cb1f99 GS |
3774 | } |
3775 | goto keylookup; | |
79072805 | 3776 | case 'x': |
3280af22 | 3777 | if (isDIGIT(s[1]) && PL_expect == XOPERATOR) { |
79072805 LW |
3778 | s++; |
3779 | Mop(OP_REPEAT); | |
2f3197b3 | 3780 | } |
79072805 LW |
3781 | goto keylookup; |
3782 | ||
378cc40b | 3783 | case '_': |
79072805 LW |
3784 | case 'a': case 'A': |
3785 | case 'b': case 'B': | |
3786 | case 'c': case 'C': | |
3787 | case 'd': case 'D': | |
3788 | case 'e': case 'E': | |
3789 | case 'f': case 'F': | |
3790 | case 'g': case 'G': | |
3791 | case 'h': case 'H': | |
3792 | case 'i': case 'I': | |
3793 | case 'j': case 'J': | |
3794 | case 'k': case 'K': | |
3795 | case 'l': case 'L': | |
3796 | case 'm': case 'M': | |
3797 | case 'n': case 'N': | |
3798 | case 'o': case 'O': | |
3799 | case 'p': case 'P': | |
3800 | case 'q': case 'Q': | |
3801 | case 'r': case 'R': | |
3802 | case 's': case 'S': | |
3803 | case 't': case 'T': | |
3804 | case 'u': case 'U': | |
a7cb1f99 | 3805 | case 'V': |
79072805 LW |
3806 | case 'w': case 'W': |
3807 | case 'X': | |
3808 | case 'y': case 'Y': | |
3809 | case 'z': case 'Z': | |
3810 | ||
49dc05e3 | 3811 | keylookup: { |
161b471a NIS |
3812 | gv = Nullgv; |
3813 | gvp = 0; | |
49dc05e3 | 3814 | |
3280af22 NIS |
3815 | PL_bufptr = s; |
3816 | s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len); | |
8ebc5c01 | 3817 | |
3818 | /* Some keywords can be followed by any delimiter, including ':' */ | |
155aba94 GS |
3819 | tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) || |
3820 | (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') || | |
3821 | (PL_tokenbuf[0] == 'q' && | |
3822 | strchr("qwxr", PL_tokenbuf[1]))))); | |
8ebc5c01 | 3823 | |
3824 | /* x::* is just a word, unless x is "CORE" */ | |
3280af22 | 3825 | if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE")) |
4633a7c4 LW |
3826 | goto just_a_word; |
3827 | ||
3643fb5f | 3828 | d = s; |
3280af22 | 3829 | while (d < PL_bufend && isSPACE(*d)) |
3643fb5f CS |
3830 | d++; /* no comments skipped here, or s### is misparsed */ |
3831 | ||
3832 | /* Is this a label? */ | |
3280af22 NIS |
3833 | if (!tmp && PL_expect == XSTATE |
3834 | && d < PL_bufend && *d == ':' && *(d + 1) != ':') { | |
8ebc5c01 | 3835 | s = d + 1; |
3280af22 | 3836 | yylval.pval = savepv(PL_tokenbuf); |
8ebc5c01 | 3837 | CLINE; |
3838 | TOKEN(LABEL); | |
3643fb5f CS |
3839 | } |
3840 | ||
3841 | /* Check for keywords */ | |
3280af22 | 3842 | tmp = keyword(PL_tokenbuf, len); |
748a9306 LW |
3843 | |
3844 | /* Is this a word before a => operator? */ | |
1c3923b3 | 3845 | if (*d == '=' && d[1] == '>') { |
748a9306 | 3846 | CLINE; |
3280af22 | 3847 | yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0)); |
748a9306 | 3848 | yylval.opval->op_private = OPpCONST_BARE; |
7948272d NIS |
3849 | if (UTF && !IN_BYTE && is_utf8_string((U8*)PL_tokenbuf, len)) |
3850 | SvUTF8_on(((SVOP*)yylval.opval)->op_sv); | |
748a9306 LW |
3851 | TERM(WORD); |
3852 | } | |
3853 | ||
a0d0e21e | 3854 | if (tmp < 0) { /* second-class keyword? */ |
56f7f34b CS |
3855 | GV *ogv = Nullgv; /* override (winner) */ |
3856 | GV *hgv = Nullgv; /* hidden (loser) */ | |
3280af22 | 3857 | if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) { |
56f7f34b | 3858 | CV *cv; |
3280af22 | 3859 | if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) && |
56f7f34b CS |
3860 | (cv = GvCVu(gv))) |
3861 | { | |
3862 | if (GvIMPORTED_CV(gv)) | |
3863 | ogv = gv; | |
3864 | else if (! CvMETHOD(cv)) | |
3865 | hgv = gv; | |
3866 | } | |
3867 | if (!ogv && | |
3280af22 NIS |
3868 | (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) && |
3869 | (gv = *gvp) != (GV*)&PL_sv_undef && | |
56f7f34b CS |
3870 | GvCVu(gv) && GvIMPORTED_CV(gv)) |
3871 | { | |
3872 | ogv = gv; | |
3873 | } | |
3874 | } | |
3875 | if (ogv) { | |
3876 | tmp = 0; /* overridden by import or by GLOBAL */ | |
6e7b2336 GS |
3877 | } |
3878 | else if (gv && !gvp | |
3879 | && -tmp==KEY_lock /* XXX generalizable kludge */ | |
d0456cad | 3880 | && GvCVu(gv) |
3280af22 | 3881 | && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE)) |
6e7b2336 GS |
3882 | { |
3883 | tmp = 0; /* any sub overrides "weak" keyword */ | |
a0d0e21e | 3884 | } |
56f7f34b CS |
3885 | else { /* no override */ |
3886 | tmp = -tmp; | |
3887 | gv = Nullgv; | |
3888 | gvp = 0; | |
4944e2f7 GS |
3889 | if (ckWARN(WARN_AMBIGUOUS) && hgv |
3890 | && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */ | |
cea2e8a9 | 3891 | Perl_warner(aTHX_ WARN_AMBIGUOUS, |
599cee73 | 3892 | "Ambiguous call resolved as CORE::%s(), %s", |
2f3ca594 | 3893 | GvENAME(hgv), "qualify as such or use &"); |
49dc05e3 | 3894 | } |
a0d0e21e LW |
3895 | } |
3896 | ||
3897 | reserved_word: | |
3898 | switch (tmp) { | |
79072805 LW |
3899 | |
3900 | default: /* not a keyword */ | |
93a17b20 | 3901 | just_a_word: { |
96e4d5b1 | 3902 | SV *sv; |
3280af22 | 3903 | char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]); |
8990e307 LW |
3904 | |
3905 | /* Get the rest if it looks like a package qualifier */ | |
3906 | ||
155aba94 | 3907 | if (*s == '\'' || (*s == ':' && s[1] == ':')) { |
c3e0f903 | 3908 | STRLEN morelen; |
3280af22 | 3909 | s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len, |
c3e0f903 GS |
3910 | TRUE, &morelen); |
3911 | if (!morelen) | |
cea2e8a9 | 3912 | Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf, |
ec2ab091 | 3913 | *s == '\'' ? "'" : "::"); |
c3e0f903 | 3914 | len += morelen; |
a0d0e21e | 3915 | } |
8990e307 | 3916 | |
3280af22 NIS |
3917 | if (PL_expect == XOPERATOR) { |
3918 | if (PL_bufptr == PL_linestart) { | |
57843af0 | 3919 | CopLINE_dec(PL_curcop); |
cea2e8a9 | 3920 | Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi); |
57843af0 | 3921 | CopLINE_inc(PL_curcop); |
463ee0b2 LW |
3922 | } |
3923 | else | |
54310121 | 3924 | no_op("Bareword",s); |
463ee0b2 | 3925 | } |
8990e307 | 3926 | |
c3e0f903 GS |
3927 | /* Look for a subroutine with this name in current package, |
3928 | unless name is "Foo::", in which case Foo is a bearword | |
3929 | (and a package name). */ | |
3930 | ||
3931 | if (len > 2 && | |
3280af22 | 3932 | PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':') |
c3e0f903 | 3933 | { |
e476b1b5 | 3934 | if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV)) |
4e553d73 | 3935 | Perl_warner(aTHX_ WARN_BAREWORD, |
599cee73 | 3936 | "Bareword \"%s\" refers to nonexistent package", |
3280af22 | 3937 | PL_tokenbuf); |
c3e0f903 | 3938 | len -= 2; |
3280af22 | 3939 | PL_tokenbuf[len] = '\0'; |
c3e0f903 GS |
3940 | gv = Nullgv; |
3941 | gvp = 0; | |
3942 | } | |
3943 | else { | |
3944 | len = 0; | |
3945 | if (!gv) | |
3280af22 | 3946 | gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV); |
c3e0f903 GS |
3947 | } |
3948 | ||
3949 | /* if we saw a global override before, get the right name */ | |
8990e307 | 3950 | |
49dc05e3 | 3951 | if (gvp) { |
79cb57f6 | 3952 | sv = newSVpvn("CORE::GLOBAL::",14); |
3280af22 | 3953 | sv_catpv(sv,PL_tokenbuf); |
49dc05e3 GS |
3954 | } |
3955 | else | |
3280af22 | 3956 | sv = newSVpv(PL_tokenbuf,0); |
8990e307 | 3957 | |
a0d0e21e LW |
3958 | /* Presume this is going to be a bareword of some sort. */ |
3959 | ||
3960 | CLINE; | |
49dc05e3 | 3961 | yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv); |
a0d0e21e LW |
3962 | yylval.opval->op_private = OPpCONST_BARE; |
3963 | ||
c3e0f903 GS |
3964 | /* And if "Foo::", then that's what it certainly is. */ |
3965 | ||
3966 | if (len) | |
3967 | goto safe_bareword; | |
3968 | ||
8990e307 LW |
3969 | /* See if it's the indirect object for a list operator. */ |
3970 | ||
3280af22 NIS |
3971 | if (PL_oldoldbufptr && |
3972 | PL_oldoldbufptr < PL_bufptr && | |
65cec589 GS |
3973 | (PL_oldoldbufptr == PL_last_lop |
3974 | || PL_oldoldbufptr == PL_last_uni) && | |
a0d0e21e | 3975 | /* NO SKIPSPACE BEFORE HERE! */ |
a9ef352a GS |
3976 | (PL_expect == XREF || |
3977 | ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF)) | |
a0d0e21e | 3978 | { |
748a9306 LW |
3979 | bool immediate_paren = *s == '('; |
3980 | ||
a0d0e21e LW |
3981 | /* (Now we can afford to cross potential line boundary.) */ |
3982 | s = skipspace(s); | |
3983 | ||
3984 | /* Two barewords in a row may indicate method call. */ | |
3985 | ||
7e2040f0 | 3986 | if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv))) |
a0d0e21e LW |
3987 | return tmp; |
3988 | ||
3989 | /* If not a declared subroutine, it's an indirect object. */ | |
3990 | /* (But it's an indir obj regardless for sort.) */ | |
3991 | ||
7948272d | 3992 | if ( !immediate_paren && (PL_last_lop_op == OP_SORT || |
f0670693 | 3993 | ((!gv || !GvCVu(gv)) && |
a9ef352a | 3994 | (PL_last_lop_op != OP_MAPSTART && |
f0670693 | 3995 | PL_last_lop_op != OP_GREPSTART)))) |
a9ef352a | 3996 | { |
3280af22 | 3997 | PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR; |
748a9306 | 3998 | goto bareword; |
93a17b20 LW |
3999 | } |
4000 | } | |
8990e307 | 4001 | |
8990e307 | 4002 | |
3280af22 | 4003 | PL_expect = XOPERATOR; |
8990e307 | 4004 | s = skipspace(s); |
1c3923b3 GS |
4005 | |
4006 | /* Is this a word before a => operator? */ | |
4007 | if (*s == '=' && s[1] == '>') { | |
4008 | CLINE; | |
4009 | sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf); | |
7948272d NIS |
4010 | if (UTF && !IN_BYTE && is_utf8_string((U8*)PL_tokenbuf, len)) |
4011 | SvUTF8_on(((SVOP*)yylval.opval)->op_sv); | |
1c3923b3 GS |
4012 | TERM(WORD); |
4013 | } | |
4014 | ||
4015 | /* If followed by a paren, it's certainly a subroutine. */ | |
93a17b20 | 4016 | if (*s == '(') { |
79072805 | 4017 | CLINE; |
96e4d5b1 | 4018 | if (gv && GvCVu(gv)) { |
bf4acbe4 | 4019 | for (d = s + 1; SPACE_OR_TAB(*d); d++) ; |
7a52d87a | 4020 | if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) { |
96e4d5b1 | 4021 | s = d + 1; |
4022 | goto its_constant; | |
4023 | } | |
4024 | } | |
3280af22 NIS |
4025 | PL_nextval[PL_nexttoke].opval = yylval.opval; |
4026 | PL_expect = XOPERATOR; | |
93a17b20 | 4027 | force_next(WORD); |
c07a80fd | 4028 | yylval.ival = 0; |
463ee0b2 | 4029 | TOKEN('&'); |
79072805 | 4030 | } |
93a17b20 | 4031 | |
a0d0e21e | 4032 | /* If followed by var or block, call it a method (unless sub) */ |
8990e307 | 4033 | |
8ebc5c01 | 4034 | if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) { |
3280af22 NIS |
4035 | PL_last_lop = PL_oldbufptr; |
4036 | PL_last_lop_op = OP_METHOD; | |
93a17b20 | 4037 | PREBLOCK(METHOD); |
463ee0b2 LW |
4038 | } |
4039 | ||
8990e307 LW |
4040 | /* If followed by a bareword, see if it looks like indir obj. */ |
4041 | ||
7e2040f0 | 4042 | if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp = intuit_method(s,gv))) |
a0d0e21e | 4043 | return tmp; |
93a17b20 | 4044 | |
8990e307 LW |
4045 | /* Not a method, so call it a subroutine (if defined) */ |
4046 | ||
8ebc5c01 | 4047 | if (gv && GvCVu(gv)) { |
46fc3d4c | 4048 | CV* cv; |
0453d815 PM |
4049 | if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS)) |
4050 | Perl_warner(aTHX_ WARN_AMBIGUOUS, | |
4051 | "Ambiguous use of -%s resolved as -&%s()", | |
3280af22 | 4052 | PL_tokenbuf, PL_tokenbuf); |
89bfa8cd | 4053 | /* Check for a constant sub */ |
46fc3d4c | 4054 | cv = GvCV(gv); |
96e4d5b1 | 4055 | if ((sv = cv_const_sv(cv))) { |
4056 | its_constant: | |
4057 | SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv); | |
4058 | ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv); | |
4059 | yylval.opval->op_private = 0; | |
4060 | TOKEN(WORD); | |
89bfa8cd | 4061 | } |
4062 | ||
a5f75d66 AD |
4063 | /* Resolve to GV now. */ |
4064 | op_free(yylval.opval); | |
4065 | yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv)); | |
9675f7ac | 4066 | yylval.opval->op_private |= OPpENTERSUB_NOPAREN; |
7a52d87a | 4067 | PL_last_lop = PL_oldbufptr; |
bf848113 | 4068 | PL_last_lop_op = OP_ENTERSUB; |
4633a7c4 LW |
4069 | /* Is there a prototype? */ |
4070 | if (SvPOK(cv)) { | |
4071 | STRLEN len; | |
7a52d87a | 4072 | char *proto = SvPV((SV*)cv, len); |
4633a7c4 LW |
4073 | if (!len) |
4074 | TERM(FUNC0SUB); | |
7a52d87a | 4075 | if (strEQ(proto, "$")) |
4633a7c4 | 4076 | OPERATOR(UNIOPSUB); |
7a52d87a | 4077 | if (*proto == '&' && *s == '{') { |
3280af22 | 4078 | sv_setpv(PL_subname,"__ANON__"); |
4633a7c4 LW |
4079 | PREBLOCK(LSTOPSUB); |
4080 | } | |
a9ef352a | 4081 | } |
3280af22 NIS |
4082 | PL_nextval[PL_nexttoke].opval = yylval.opval; |
4083 | PL_expect = XTERM; | |
8990e307 LW |
4084 | force_next(WORD); |
4085 | TOKEN(NOAMP); | |
4086 | } | |
748a9306 | 4087 | |
8990e307 LW |
4088 | /* Call it a bare word */ |
4089 | ||
5603f27d GS |
4090 | if (PL_hints & HINT_STRICT_SUBS) |
4091 | yylval.opval->op_private |= OPpCONST_STRICT; | |
4092 | else { | |
4093 | bareword: | |
4094 | if (ckWARN(WARN_RESERVED)) { | |
4095 | if (lastchar != '-') { | |
4096 | for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ; | |
4097 | if (!*d) | |
cea2e8a9 | 4098 | Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved, |
5603f27d GS |
4099 | PL_tokenbuf); |
4100 | } | |
748a9306 LW |
4101 | } |
4102 | } | |
c3e0f903 GS |
4103 | |
4104 | safe_bareword: | |
f248d071 | 4105 | if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) { |
0453d815 PM |
4106 | Perl_warner(aTHX_ WARN_AMBIGUOUS, |
4107 | "Operator or semicolon missing before %c%s", | |
3280af22 | 4108 | lastchar, PL_tokenbuf); |
0453d815 PM |
4109 | Perl_warner(aTHX_ WARN_AMBIGUOUS, |
4110 | "Ambiguous use of %c resolved as operator %c", | |
748a9306 LW |
4111 | lastchar, lastchar); |
4112 | } | |
93a17b20 | 4113 | TOKEN(WORD); |
79072805 | 4114 | } |
79072805 | 4115 | |
68dc0745 | 4116 | case KEY___FILE__: |
46fc3d4c | 4117 | yylval.opval = (OP*)newSVOP(OP_CONST, 0, |
ed094faf | 4118 | newSVpv(CopFILE(PL_curcop),0)); |
46fc3d4c | 4119 | TERM(THING); |
4120 | ||
79072805 | 4121 | case KEY___LINE__: |
cf2093f6 | 4122 | yylval.opval = (OP*)newSVOP(OP_CONST, 0, |
57843af0 | 4123 | Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop))); |
79072805 | 4124 | TERM(THING); |
68dc0745 | 4125 | |
4126 | case KEY___PACKAGE__: | |
4127 | yylval.opval = (OP*)newSVOP(OP_CONST, 0, | |
3280af22 NIS |
4128 | (PL_curstash |
4129 | ? newSVsv(PL_curstname) | |
4130 | : &PL_sv_undef)); | |
79072805 | 4131 | TERM(THING); |
79072805 | 4132 | |
e50aee73 | 4133 | case KEY___DATA__: |
79072805 LW |
4134 | case KEY___END__: { |
4135 | GV *gv; | |
79072805 LW |
4136 | |
4137 | /*SUPPRESS 560*/ | |
3280af22 | 4138 | if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) { |
e50aee73 | 4139 | char *pname = "main"; |
3280af22 NIS |
4140 | if (PL_tokenbuf[2] == 'D') |
4141 | pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash); | |
cea2e8a9 | 4142 | gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO); |
a5f75d66 | 4143 | GvMULTI_on(gv); |
79072805 | 4144 | if (!GvIO(gv)) |
a0d0e21e | 4145 | GvIOp(gv) = newIO(); |
3280af22 | 4146 | IoIFP(GvIOp(gv)) = PL_rsfp; |
a0d0e21e LW |
4147 | #if defined(HAS_FCNTL) && defined(F_SETFD) |
4148 | { | |
3280af22 | 4149 | int fd = PerlIO_fileno(PL_rsfp); |
a0d0e21e LW |
4150 | fcntl(fd,F_SETFD,fd >= 3); |
4151 | } | |
79072805 | 4152 | #endif |
fd049845 | 4153 | /* Mark this internal pseudo-handle as clean */ |
4154 | IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT; | |
3280af22 | 4155 | if (PL_preprocess) |
50952442 | 4156 | IoTYPE(GvIOp(gv)) = IoTYPE_PIPE; |
3280af22 | 4157 | else if ((PerlIO*)PL_rsfp == PerlIO_stdin()) |
50952442 | 4158 | IoTYPE(GvIOp(gv)) = IoTYPE_STD; |
79072805 | 4159 | else |
50952442 | 4160 | IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY; |
c39cd008 GS |
4161 | #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS) |
4162 | /* if the script was opened in binmode, we need to revert | |
53129d29 | 4163 | * it to text mode for compatibility; but only iff it has CRs |
c39cd008 | 4164 | * XXX this is a questionable hack at best. */ |
53129d29 GS |
4165 | if (PL_bufend-PL_bufptr > 2 |
4166 | && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r') | |
c39cd008 GS |
4167 | { |
4168 | Off_t loc = 0; | |
50952442 | 4169 | if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) { |
c39cd008 GS |
4170 | loc = PerlIO_tell(PL_rsfp); |
4171 | (void)PerlIO_seek(PL_rsfp, 0L, 0); | |
4172 | } | |
4173 | if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) { | |
c39cd008 GS |
4174 | if (loc > 0) |
4175 | PerlIO_seek(PL_rsfp, loc, 0); | |
4176 | } | |
4177 | } | |
4178 | #endif | |
7948272d NIS |
4179 | #ifdef PERLIO_LAYERS |
4180 | if (UTF && !IN_BYTE) | |
4181 | PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8"); | |
4182 | #endif | |
3280af22 | 4183 | PL_rsfp = Nullfp; |
79072805 LW |
4184 | } |
4185 | goto fake_eof; | |
e929a76b | 4186 | } |
de3bb511 | 4187 | |
8990e307 | 4188 | case KEY_AUTOLOAD: |
ed6116ce | 4189 | case KEY_DESTROY: |
79072805 | 4190 | case KEY_BEGIN: |
7d30b5c4 | 4191 | case KEY_CHECK: |
7d07dbc2 | 4192 | case KEY_INIT: |
7d30b5c4 | 4193 | case KEY_END: |
3280af22 NIS |
4194 | if (PL_expect == XSTATE) { |
4195 | s = PL_bufptr; | |
93a17b20 | 4196 | goto really_sub; |
79072805 LW |
4197 | } |
4198 | goto just_a_word; | |
4199 | ||
a0d0e21e LW |
4200 | case KEY_CORE: |
4201 | if (*s == ':' && s[1] == ':') { | |
4202 | s += 2; | |
748a9306 | 4203 | d = s; |
3280af22 | 4204 | s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len); |
6798c92b GS |
4205 | if (!(tmp = keyword(PL_tokenbuf, len))) |
4206 | Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf); | |
a0d0e21e LW |
4207 | if (tmp < 0) |
4208 | tmp = -tmp; | |
4209 | goto reserved_word; | |
4210 | } | |
4211 | goto just_a_word; | |
4212 | ||
463ee0b2 LW |
4213 | case KEY_abs: |
4214 | UNI(OP_ABS); | |
4215 | ||
79072805 LW |
4216 | case KEY_alarm: |
4217 | UNI(OP_ALARM); | |
4218 | ||
4219 | case KEY_accept: | |
a0d0e21e | 4220 | LOP(OP_ACCEPT,XTERM); |
79072805 | 4221 | |
463ee0b2 LW |
4222 | case KEY_and: |
4223 | OPERATOR(ANDOP); | |
4224 | ||
79072805 | 4225 | case KEY_atan2: |
a0d0e21e | 4226 | LOP(OP_ATAN2,XTERM); |
85e6fe83 | 4227 | |
79072805 | 4228 | case KEY_bind: |
a0d0e21e | 4229 | LOP(OP_BIND,XTERM); |
79072805 LW |
4230 | |
4231 | case KEY_binmode: | |
1c1fc3ea | 4232 | LOP(OP_BINMODE,XTERM); |
79072805 LW |
4233 | |
4234 | case KEY_bless: | |
a0d0e21e | 4235 | LOP(OP_BLESS,XTERM); |
79072805 LW |
4236 | |
4237 | case KEY_chop: | |
4238 | UNI(OP_CHOP); | |
4239 | ||
4240 | case KEY_continue: | |
4241 | PREBLOCK(CONTINUE); | |
4242 | ||
4243 | case KEY_chdir: | |
85e6fe83 | 4244 | (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */ |
79072805 LW |
4245 | UNI(OP_CHDIR); |
4246 | ||
4247 | case KEY_close: | |
4248 | UNI(OP_CLOSE); | |
4249 | ||
4250 | case KEY_closedir: | |
4251 | UNI(OP_CLOSEDIR); | |
4252 | ||
4253 | case KEY_cmp: | |
4254 | Eop(OP_SCMP); | |
4255 | ||
4256 | case KEY_caller: | |
4257 | UNI(OP_CALLER); | |
4258 | ||
4259 | case KEY_crypt: | |
4260 | #ifdef FCRYPT | |
f4c556ac GS |
4261 | if (!PL_cryptseen) { |
4262 | PL_cryptseen = TRUE; | |
de3bb511 | 4263 | init_des(); |
f4c556ac | 4264 | } |
a687059c | 4265 | #endif |
a0d0e21e | 4266 | LOP(OP_CRYPT,XTERM); |
79072805 LW |
4267 | |
4268 | case KEY_chmod: | |
e476b1b5 | 4269 | if (ckWARN(WARN_CHMOD)) { |
3280af22 | 4270 | for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ; |
748a9306 | 4271 | if (*d != '0' && isDIGIT(*d)) |
e476b1b5 | 4272 | Perl_warner(aTHX_ WARN_CHMOD, |
5a211162 | 4273 | "chmod() mode argument is missing initial 0"); |
748a9306 | 4274 | } |
a0d0e21e | 4275 | LOP(OP_CHMOD,XTERM); |
79072805 LW |
4276 | |
4277 | case KEY_chown: | |
a0d0e21e | 4278 | LOP(OP_CHOWN,XTERM); |
79072805 LW |
4279 | |
4280 | case KEY_connect: | |
a0d0e21e | 4281 | LOP(OP_CONNECT,XTERM); |
79072805 | 4282 | |
463ee0b2 LW |
4283 | case KEY_chr: |
4284 | UNI(OP_CHR); | |
4285 | ||
79072805 LW |
4286 | case KEY_cos: |
4287 | UNI(OP_COS); | |
4288 | ||
4289 | case KEY_chroot: | |
4290 | UNI(OP_CHROOT); | |
4291 | ||
4292 | case KEY_do: | |
4293 | s = skipspace(s); | |
4294 | if (*s == '{') | |
a0d0e21e | 4295 | PRETERMBLOCK(DO); |
79072805 | 4296 | if (*s != '\'') |
a0d0e21e | 4297 | s = force_word(s,WORD,FALSE,TRUE,FALSE); |
378cc40b | 4298 | OPERATOR(DO); |
79072805 LW |
4299 | |
4300 | case KEY_die: | |
3280af22 | 4301 | PL_hints |= HINT_BLOCK_SCOPE; |
a0d0e21e | 4302 | LOP(OP_DIE,XTERM); |
79072805 LW |
4303 | |
4304 | case KEY_defined: | |
4305 | UNI(OP_DEFINED); | |
4306 | ||
4307 | case KEY_delete: | |
a0d0e21e | 4308 | UNI(OP_DELETE); |
79072805 LW |
4309 | |
4310 | case KEY_dbmopen: | |
a0d0e21e LW |
4311 | gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV); |
4312 | LOP(OP_DBMOPEN,XTERM); | |
79072805 LW |
4313 | |
4314 | case KEY_dbmclose: | |
4315 | UNI(OP_DBMCLOSE); | |
4316 | ||
4317 | case KEY_dump: | |
a0d0e21e | 4318 | s = force_word(s,WORD,TRUE,FALSE,FALSE); |
79072805 LW |
4319 | LOOPX(OP_DUMP); |
4320 | ||
4321 | case KEY_else: | |
4322 | PREBLOCK(ELSE); | |
4323 | ||
4324 | case KEY_elsif: | |
57843af0 | 4325 | yylval.ival = CopLINE(PL_curcop); |
79072805 LW |
4326 | OPERATOR(ELSIF); |
4327 | ||
4328 | case KEY_eq: | |
4329 | Eop(OP_SEQ); | |
4330 | ||
a0d0e21e LW |
4331 | case KEY_exists: |
4332 | UNI(OP_EXISTS); | |
4e553d73 | 4333 | |
79072805 LW |
4334 | case KEY_exit: |
4335 | UNI(OP_EXIT); | |
4336 | ||
4337 | case KEY_eval: | |
79072805 | 4338 | s = skipspace(s); |
3280af22 | 4339 | PL_expect = (*s == '{') ? XTERMBLOCK : XTERM; |
463ee0b2 | 4340 | UNIBRACK(OP_ENTEREVAL); |
79072805 LW |
4341 | |
4342 | case KEY_eof: | |
4343 | UNI(OP_EOF); | |
4344 | ||
4345 | case KEY_exp: | |
4346 | UNI(OP_EXP); | |
4347 | ||
4348 | case KEY_each: | |
4349 | UNI(OP_EACH); | |
4350 | ||
4351 | case KEY_exec: | |
4352 | set_csh(); | |
a0d0e21e | 4353 | LOP(OP_EXEC,XREF); |
79072805 LW |
4354 | |
4355 | case KEY_endhostent: | |
4356 | FUN0(OP_EHOSTENT); | |
4357 | ||
4358 | case KEY_endnetent: | |
4359 | FUN0(OP_ENETENT); | |
4360 | ||
4361 | case KEY_endservent: | |
4362 | FUN0(OP_ESERVENT); | |
4363 | ||
4364 | case KEY_endprotoent: | |
4365 | FUN0(OP_EPROTOENT); | |
4366 | ||
4367 | case KEY_endpwent: | |
4368 | FUN0(OP_EPWENT); | |
4369 | ||
4370 | case KEY_endgrent: | |
4371 | FUN0(OP_EGRENT); | |
4372 | ||
4373 | case KEY_for: | |
4374 | case KEY_foreach: | |
57843af0 | 4375 | yylval.ival = CopLINE(PL_curcop); |
55497cff | 4376 | s = skipspace(s); |
7e2040f0 | 4377 | if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) { |
55497cff | 4378 | char *p = s; |
3280af22 | 4379 | if ((PL_bufend - p) >= 3 && |
55497cff | 4380 | strnEQ(p, "my", 2) && isSPACE(*(p + 2))) |
4381 | p += 2; | |
77ca0c92 LW |
4382 | else if ((PL_bufend - p) >= 4 && |
4383 | strnEQ(p, "our", 3) && isSPACE(*(p + 3))) | |
4384 | p += 3; | |
55497cff | 4385 | p = skipspace(p); |
7e2040f0 | 4386 | if (isIDFIRST_lazy_if(p,UTF)) { |
77ca0c92 LW |
4387 | p = scan_ident(p, PL_bufend, |
4388 | PL_tokenbuf, sizeof PL_tokenbuf, TRUE); | |
4389 | p = skipspace(p); | |
4390 | } | |
4391 | if (*p != '$') | |
cea2e8a9 | 4392 | Perl_croak(aTHX_ "Missing $ on loop variable"); |
55497cff | 4393 | } |
79072805 LW |
4394 | OPERATOR(FOR); |
4395 | ||
4396 | case KEY_formline: | |
a0d0e21e | 4397 | LOP(OP_FORMLINE,XTERM); |
79072805 LW |
4398 | |
4399 | case KEY_fork: | |
4400 | FUN0(OP_FORK); | |
4401 | ||
4402 | case KEY_fcntl: | |
a0d0e21e | 4403 | LOP(OP_FCNTL,XTERM); |
79072805 LW |
4404 | |
4405 | case KEY_fileno: | |
4406 | UNI(OP_FILENO); | |
4407 | ||
4408 | case KEY_flock: | |
a0d0e21e | 4409 | LOP(OP_FLOCK,XTERM); |
79072805 LW |
4410 | |
4411 | case KEY_gt: | |
4412 | Rop(OP_SGT); | |
4413 | ||
4414 | case KEY_ge: | |
4415 | Rop(OP_SGE); | |
4416 | ||
4417 | case KEY_grep: | |
2c38e13d | 4418 | LOP(OP_GREPSTART, XREF); |
79072805 LW |
4419 | |
4420 | case KEY_goto: | |
a0d0e21e | 4421 | s = force_word(s,WORD,TRUE,FALSE,FALSE); |
79072805 LW |
4422 | LOOPX(OP_GOTO); |
4423 | ||
4424 | case KEY_gmtime: | |
4425 | UNI(OP_GMTIME); | |
4426 | ||
4427 | case KEY_getc: | |
4428 | UNI(OP_GETC); | |
4429 | ||
4430 | case KEY_getppid: | |
4431 | FUN0(OP_GETPPID); | |
4432 | ||
4433 | case KEY_getpgrp: | |
4434 | UNI(OP_GETPGRP); | |
4435 | ||
4436 | case KEY_getpriority: | |
a0d0e21e | 4437 | LOP(OP_GETPRIORITY,XTERM); |
79072805 LW |
4438 | |
4439 | case KEY_getprotobyname: | |
4440 | UNI(OP_GPBYNAME); | |
4441 | ||
4442 | case KEY_getprotobynumber: | |
a0d0e21e | 4443 | LOP(OP_GPBYNUMBER,XTERM); |
79072805 LW |
4444 | |
4445 | case KEY_getprotoent: | |
4446 | FUN0(OP_GPROTOENT); | |
4447 | ||
4448 | case KEY_getpwent: | |
4449 | FUN0(OP_GPWENT); | |
4450 | ||
4451 | case KEY_getpwnam: | |
ff68c719 | 4452 | UNI(OP_GPWNAM); |
79072805 LW |
4453 | |
4454 | case KEY_getpwuid: | |
ff68c719 | 4455 | UNI(OP_GPWUID); |
79072805 LW |
4456 | |
4457 | case KEY_getpeername: | |
4458 | UNI(OP_GETPEERNAME); | |
4459 | ||
4460 | case KEY_gethostbyname: | |
4461 | UNI(OP_GHBYNAME); | |
4462 | ||
4463 | case KEY_gethostbyaddr: | |
a0d0e21e | 4464 | LOP(OP_GHBYADDR,XTERM); |
79072805 LW |
4465 | |
4466 | case KEY_gethostent: | |
4467 | FUN0(OP_GHOSTENT); | |
4468 | ||
4469 | case KEY_getnetbyname: | |
4470 | UNI(OP_GNBYNAME); | |
4471 | ||
4472 | case KEY_getnetbyaddr: | |
a0d0e21e | 4473 | LOP(OP_GNBYADDR,XTERM); |
79072805 LW |
4474 | |
4475 | case KEY_getnetent: | |
4476 | FUN0(OP_GNETENT); | |
4477 | ||
4478 | case KEY_getservbyname: | |
a0d0e21e | 4479 | LOP(OP_GSBYNAME,XTERM); |
79072805 LW |
4480 | |
4481 | case KEY_getservbyport: | |
a0d0e21e | 4482 | LOP(OP_GSBYPORT,XTERM); |
79072805 LW |
4483 | |
4484 | case KEY_getservent: | |
4485 | FUN0(OP_GSERVENT); | |
4486 | ||
4487 | case KEY_getsockname: | |
4488 | UNI(OP_GETSOCKNAME); | |
4489 | ||
4490 | case KEY_getsockopt: | |
a0d0e21e | 4491 | LOP(OP_GSOCKOPT,XTERM); |
79072805 LW |
4492 | |
4493 | case KEY_getgrent: | |
4494 | FUN0(OP_GGRENT); | |
4495 | ||
4496 | case KEY_getgrnam: | |
ff68c719 | 4497 | UNI(OP_GGRNAM); |
79072805 LW |
4498 | |
4499 | case KEY_getgrgid: | |
ff68c719 | 4500 | UNI(OP_GGRGID); |
79072805 LW |
4501 | |
4502 | case KEY_getlogin: | |
4503 | FUN0(OP_GETLOGIN); | |
4504 | ||
93a17b20 | 4505 | case KEY_glob: |
a0d0e21e LW |
4506 | set_csh(); |
4507 | LOP(OP_GLOB,XTERM); | |
93a17b20 | 4508 | |
79072805 LW |
4509 | case KEY_hex: |
4510 | UNI(OP_HEX); | |
4511 | ||
4512 | case KEY_if: | |
57843af0 | 4513 | yylval.ival = CopLINE(PL_curcop); |
79072805 LW |
4514 | OPERATOR(IF); |
4515 | ||
4516 | case KEY_index: | |
a0d0e21e | 4517 | LOP(OP_INDEX,XTERM); |
79072805 LW |
4518 | |
4519 | case KEY_int: | |
4520 | UNI(OP_INT); | |
4521 | ||
4522 | case KEY_ioctl: | |
a0d0e21e | 4523 | LOP(OP_IOCTL,XTERM); |
79072805 LW |
4524 | |
4525 | case KEY_join: | |
a0d0e21e | 4526 | LOP(OP_JOIN,XTERM); |
79072805 LW |
4527 | |
4528 | case KEY_keys: | |
4529 | UNI(OP_KEYS); | |
4530 | ||
4531 | case KEY_kill: | |
a0d0e21e | 4532 | LOP(OP_KILL,XTERM); |
79072805 LW |
4533 | |
4534 | case KEY_last: | |
a0d0e21e | 4535 | s = force_word(s,WORD,TRUE,FALSE,FALSE); |
79072805 | 4536 | LOOPX(OP_LAST); |
4e553d73 | 4537 | |
79072805 LW |
4538 | case KEY_lc: |
4539 | UNI(OP_LC); | |
4540 | ||
4541 | case KEY_lcfirst: | |
4542 | UNI(OP_LCFIRST); | |
4543 | ||
4544 | case KEY_local: | |
09bef843 | 4545 | yylval.ival = 0; |
79072805 LW |
4546 | OPERATOR(LOCAL); |
4547 | ||
4548 | case KEY_length: | |
4549 | UNI(OP_LENGTH); | |
4550 | ||
4551 | case KEY_lt: | |
4552 | Rop(OP_SLT); | |
4553 | ||
4554 | case KEY_le: | |
4555 | Rop(OP_SLE); | |
4556 | ||
4557 | case KEY_localtime: | |
4558 | UNI(OP_LOCALTIME); | |
4559 | ||
4560 | case KEY_log: | |
4561 | UNI(OP_LOG); | |
4562 | ||
4563 | case KEY_link: | |
a0d0e21e | 4564 | LOP(OP_LINK,XTERM); |
79072805 LW |
4565 | |
4566 | case KEY_listen: | |
a0d0e21e | 4567 | LOP(OP_LISTEN,XTERM); |
79072805 | 4568 | |
c0329465 MB |
4569 | case KEY_lock: |
4570 | UNI(OP_LOCK); | |
4571 | ||
79072805 LW |
4572 | case KEY_lstat: |
4573 | UNI(OP_LSTAT); | |
4574 | ||
4575 | case KEY_m: | |
8782bef2 | 4576 | s = scan_pat(s,OP_MATCH); |
79072805 LW |
4577 | TERM(sublex_start()); |
4578 | ||
a0d0e21e | 4579 | case KEY_map: |
2c38e13d | 4580 | LOP(OP_MAPSTART, XREF); |
4e4e412b | 4581 | |
79072805 | 4582 | case KEY_mkdir: |
a0d0e21e | 4583 | LOP(OP_MKDIR,XTERM); |
79072805 LW |
4584 | |
4585 | case KEY_msgctl: | |
a0d0e21e | 4586 | LOP(OP_MSGCTL,XTERM); |
79072805 LW |
4587 | |
4588 | case KEY_msgget: | |
a0d0e21e | 4589 | LOP(OP_MSGGET,XTERM); |
79072805 LW |
4590 | |
4591 | case KEY_msgrcv: | |
a0d0e21e | 4592 | LOP(OP_MSGRCV,XTERM); |
79072805 LW |
4593 | |
4594 | case KEY_msgsnd: | |
a0d0e21e | 4595 | LOP(OP_MSGSND,XTERM); |
79072805 | 4596 | |
77ca0c92 | 4597 | case KEY_our: |
93a17b20 | 4598 | case KEY_my: |
77ca0c92 | 4599 | PL_in_my = tmp; |
c750a3ec | 4600 | s = skipspace(s); |
7e2040f0 | 4601 | if (isIDFIRST_lazy_if(s,UTF)) { |
3280af22 | 4602 | s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len); |
09bef843 SB |
4603 | if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3)) |
4604 | goto really_sub; | |
def3634b | 4605 | PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len); |
3280af22 | 4606 | if (!PL_in_my_stash) { |
c750a3ec | 4607 | char tmpbuf[1024]; |
3280af22 NIS |
4608 | PL_bufptr = s; |
4609 | sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf); | |
c750a3ec MB |
4610 | yyerror(tmpbuf); |
4611 | } | |
4612 | } | |
09bef843 | 4613 | yylval.ival = 1; |
55497cff | 4614 | OPERATOR(MY); |
93a17b20 | 4615 | |
79072805 | 4616 | case KEY_next: |
a0d0e21e | 4617 | s = force_word(s,WORD,TRUE,FALSE,FALSE); |
79072805 LW |
4618 | LOOPX(OP_NEXT); |
4619 | ||
4620 | case KEY_ne: | |
4621 | Eop(OP_SNE); | |
4622 | ||
a0d0e21e | 4623 | case KEY_no: |
3280af22 | 4624 | if (PL_expect != XSTATE) |
a0d0e21e LW |
4625 | yyerror("\"no\" not allowed in expression"); |
4626 | s = force_word(s,WORD,FALSE,TRUE,FALSE); | |
89bfa8cd | 4627 | s = force_version(s); |
a0d0e21e LW |
4628 | yylval.ival = 0; |
4629 | OPERATOR(USE); | |
4630 | ||
4631 | case KEY_not: | |
2d2e263d LW |
4632 | if (*s == '(' || (s = skipspace(s), *s == '(')) |
4633 | FUN1(OP_NOT); | |
4634 | else | |
4635 | OPERATOR(NOTOP); | |
a0d0e21e | 4636 | |
79072805 | 4637 | case KEY_open: |
93a17b20 | 4638 | s = skipspace(s); |
7e2040f0 | 4639 | if (isIDFIRST_lazy_if(s,UTF)) { |
93a17b20 | 4640 | char *t; |
7e2040f0 | 4641 | for (d = s; isALNUM_lazy_if(d,UTF); d++) ; |
93a17b20 | 4642 | t = skipspace(d); |
e476b1b5 GS |
4643 | if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE)) |
4644 | Perl_warner(aTHX_ WARN_PRECEDENCE, | |
0453d815 PM |
4645 | "Precedence problem: open %.*s should be open(%.*s)", |
4646 | d-s,s, d-s,s); | |
93a17b20 | 4647 | } |
a0d0e21e | 4648 | LOP(OP_OPEN,XTERM); |
79072805 | 4649 | |
463ee0b2 | 4650 | case KEY_or: |
a0d0e21e | 4651 | yylval.ival = OP_OR; |
463ee0b2 LW |
4652 | OPERATOR(OROP); |
4653 | ||
79072805 LW |
4654 | case KEY_ord: |
4655 | UNI(OP_ORD); | |
4656 | ||
4657 | case KEY_oct: | |
4658 | UNI(OP_OCT); | |
4659 | ||
4660 | case KEY_opendir: | |
a0d0e21e | 4661 | LOP(OP_OPEN_DIR,XTERM); |
79072805 LW |
4662 | |
4663 | case KEY_print: | |
3280af22 | 4664 | checkcomma(s,PL_tokenbuf,"filehandle"); |
a0d0e21e | 4665 | LOP(OP_PRINT,XREF); |
79072805 LW |
4666 | |
4667 | case KEY_printf: | |
3280af22 | 4668 | checkcomma(s,PL_tokenbuf,"filehandle"); |
a0d0e21e | 4669 | LOP(OP_PRTF,XREF); |
79072805 | 4670 | |
c07a80fd | 4671 | case KEY_prototype: |
4672 | UNI(OP_PROTOTYPE); | |
4673 | ||
79072805 | 4674 | case KEY_push: |
a0d0e21e | 4675 | LOP(OP_PUSH,XTERM); |
79072805 LW |
4676 | |
4677 | case KEY_pop: | |
4678 | UNI(OP_POP); | |
4679 | ||
a0d0e21e LW |
4680 | case KEY_pos: |
4681 | UNI(OP_POS); | |
4e553d73 | 4682 | |
79072805 | 4683 | case KEY_pack: |
a0d0e21e | 4684 | LOP(OP_PACK,XTERM); |
79072805 LW |
4685 | |
4686 | case KEY_package: | |
a0d0e21e | 4687 | s = force_word(s,WORD,FALSE,TRUE,FALSE); |
79072805 LW |
4688 | OPERATOR(PACKAGE); |
4689 | ||
4690 | case KEY_pipe: | |
a0d0e21e | 4691 | LOP(OP_PIPE_OP,XTERM); |
79072805 LW |
4692 | |
4693 | case KEY_q: | |
09bef843 | 4694 | s = scan_str(s,FALSE,FALSE); |
79072805 | 4695 | if (!s) |
85e6fe83 | 4696 | missingterm((char*)0); |
79072805 LW |
4697 | yylval.ival = OP_CONST; |
4698 | TERM(sublex_start()); | |
4699 | ||
a0d0e21e LW |
4700 | case KEY_quotemeta: |
4701 | UNI(OP_QUOTEMETA); | |
4702 | ||
8990e307 | 4703 | case KEY_qw: |
09bef843 | 4704 | s = scan_str(s,FALSE,FALSE); |
8990e307 | 4705 | if (!s) |
85e6fe83 | 4706 | missingterm((char*)0); |
8127e0e3 GS |
4707 | force_next(')'); |
4708 | if (SvCUR(PL_lex_stuff)) { | |
4709 | OP *words = Nullop; | |
4710 | int warned = 0; | |
3280af22 | 4711 | d = SvPV_force(PL_lex_stuff, len); |
8127e0e3 | 4712 | while (len) { |
7948272d | 4713 | SV *sv; |
8127e0e3 GS |
4714 | for (; isSPACE(*d) && len; --len, ++d) ; |
4715 | if (len) { | |
4716 | char *b = d; | |
e476b1b5 | 4717 | if (!warned && ckWARN(WARN_QW)) { |
8127e0e3 GS |
4718 | for (; !isSPACE(*d) && len; --len, ++d) { |
4719 | if (*d == ',') { | |
e476b1b5 | 4720 | Perl_warner(aTHX_ WARN_QW, |
8127e0e3 GS |
4721 | "Possible attempt to separate words with commas"); |
4722 | ++warned; | |
4723 | } | |
4724 | else if (*d == '#') { | |
e476b1b5 | 4725 | Perl_warner(aTHX_ WARN_QW, |
8127e0e3 GS |
4726 | "Possible attempt to put comments in qw() list"); |
4727 | ++warned; | |
4728 | } | |
4729 | } | |
4730 | } | |
4731 | else { | |
4732 | for (; !isSPACE(*d) && len; --len, ++d) ; | |
4733 | } | |
7948272d NIS |
4734 | sv = newSVpvn(b, d-b); |
4735 | if (DO_UTF8(PL_lex_stuff)) | |
4736 | SvUTF8_on(sv); | |
8127e0e3 | 4737 | words = append_elem(OP_LIST, words, |
7948272d | 4738 | newSVOP(OP_CONST, 0, tokeq(sv))); |
55497cff | 4739 | } |
4740 | } | |
8127e0e3 GS |
4741 | if (words) { |
4742 | PL_nextval[PL_nexttoke].opval = words; | |
4743 | force_next(THING); | |
4744 | } | |
55497cff | 4745 | } |
8127e0e3 GS |
4746 | if (PL_lex_stuff) |
4747 | SvREFCNT_dec(PL_lex_stuff); | |
3280af22 | 4748 | PL_lex_stuff = Nullsv; |
3280af22 | 4749 | PL_expect = XTERM; |
8127e0e3 | 4750 | TOKEN('('); |
8990e307 | 4751 | |
79072805 | 4752 | case KEY_qq: |
9aa983d2 | 4753 | case KEY_qu: |
09bef843 | 4754 | s = scan_str(s,FALSE,FALSE); |
f9a63242 JH |
4755 | if (tmp == KEY_qu && |
4756 | is_utf8_string((U8*)SvPVX(PL_lex_stuff), SvCUR(PL_lex_stuff))) | |
9aa983d2 | 4757 | SvUTF8_on(PL_lex_stuff); |
79072805 | 4758 | if (!s) |
85e6fe83 | 4759 | missingterm((char*)0); |
a0d0e21e | 4760 | yylval.ival = OP_STRINGIFY; |
3280af22 NIS |
4761 | if (SvIVX(PL_lex_stuff) == '\'') |
4762 | SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */ | |
79072805 LW |
4763 | TERM(sublex_start()); |
4764 | ||
8782bef2 GB |
4765 | case KEY_qr: |
4766 | s = scan_pat(s,OP_QR); | |
4767 | TERM(sublex_start()); | |
4768 | ||
79072805 | 4769 | case KEY_qx: |
09bef843 | 4770 | s = scan_str(s,FALSE,FALSE); |
79072805 | 4771 | if (!s) |
85e6fe83 | 4772 | missingterm((char*)0); |
79072805 LW |
4773 | yylval.ival = OP_BACKTICK; |
4774 | set_csh(); | |
4775 | TERM(sublex_start()); | |
4776 | ||
4777 | case KEY_return: | |
4778 | OLDLOP(OP_RETURN); | |
4779 | ||
4780 | case KEY_require: | |
a7cb1f99 GS |
4781 | s = skipspace(s); |
4782 | if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) { | |
4783 | s = force_version(s); | |
4784 | } | |
4785 | else { | |
4786 | *PL_tokenbuf = '\0'; | |
4787 | s = force_word(s,WORD,TRUE,TRUE,FALSE); | |
7e2040f0 | 4788 | if (isIDFIRST_lazy_if(PL_tokenbuf,UTF)) |
a7cb1f99 GS |
4789 | gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE); |
4790 | else if (*s == '<') | |
4791 | yyerror("<> should be quotes"); | |
4792 | } | |
463ee0b2 | 4793 | UNI(OP_REQUIRE); |
79072805 LW |
4794 | |
4795 | case KEY_reset: | |
4796 | UNI(OP_RESET); | |
4797 | ||
4798 | case KEY_redo: | |
a0d0e21e | 4799 | s = force_word(s,WORD,TRUE,FALSE,FALSE); |
79072805 LW |
4800 | LOOPX(OP_REDO); |
4801 | ||
4802 | case KEY_rename: | |
a0d0e21e | 4803 | LOP(OP_RENAME,XTERM); |
79072805 LW |
4804 | |
4805 | case KEY_rand: | |
4806 | UNI(OP_RAND); | |
4807 | ||
4808 | case KEY_rmdir: | |
4809 | UNI(OP_RMDIR); | |
4810 | ||
4811 | case KEY_rindex: | |
a0d0e21e | 4812 | LOP(OP_RINDEX,XTERM); |
79072805 LW |
4813 | |
4814 | case KEY_read: | |
a0d0e21e | 4815 | LOP(OP_READ,XTERM); |
79072805 LW |
4816 | |
4817 | case KEY_readdir: | |
4818 | UNI(OP_READDIR); | |
4819 | ||
93a17b20 LW |
4820 | case KEY_readline: |
4821 | set_csh(); | |
4822 | UNI(OP_READLINE); | |
4823 | ||
4824 | case KEY_readpipe: | |
4825 | set_csh(); | |
4826 | UNI(OP_BACKTICK); | |
4827 | ||
79072805 LW |
4828 | case KEY_rewinddir: |
4829 | UNI(OP_REWINDDIR); | |
4830 | ||
4831 | case KEY_recv: | |
a0d0e21e | 4832 | LOP(OP_RECV,XTERM); |
79072805 LW |
4833 | |
4834 | case KEY_reverse: | |
a0d0e21e | 4835 | LOP(OP_REVERSE,XTERM); |
79072805 LW |
4836 | |
4837 | case KEY_readlink: | |
4838 | UNI(OP_READLINK); | |
4839 | ||
4840 | case KEY_ref: | |
4841 | UNI(OP_REF); | |
4842 | ||
4843 | case KEY_s: | |
4844 | s = scan_subst(s); | |
4845 | if (yylval.opval) | |
4846 | TERM(sublex_start()); | |
4847 | else | |
4848 | TOKEN(1); /* force error */ | |
4849 | ||
a0d0e21e LW |
4850 | case KEY_chomp: |
4851 | UNI(OP_CHOMP); | |
4e553d73 | 4852 | |
79072805 LW |
4853 | case KEY_scalar: |
4854 | UNI(OP_SCALAR); | |
4855 | ||
4856 | case KEY_select: | |
a0d0e21e | 4857 | LOP(OP_SELECT,XTERM); |
79072805 LW |
4858 | |
4859 | case KEY_seek: | |
a0d0e21e | 4860 | LOP(OP_SEEK,XTERM); |
79072805 LW |
4861 | |
4862 | case KEY_semctl: | |
a0d0e21e | 4863 | LOP(OP_SEMCTL,XTERM); |
79072805 LW |
4864 | |
4865 | case KEY_semget: | |
a0d0e21e | 4866 | LOP(OP_SEMGET,XTERM); |
79072805 LW |
4867 | |
4868 | case KEY_semop: | |
a0d0e21e | 4869 | LOP(OP_SEMOP,XTERM); |
79072805 LW |
4870 | |
4871 | case KEY_send: | |
a0d0e21e | 4872 | LOP(OP_SEND,XTERM); |
79072805 LW |
4873 | |
4874 | case KEY_setpgrp: | |
a0d0e21e | 4875 | LOP(OP_SETPGRP,XTERM); |
79072805 LW |
4876 | |
4877 | case KEY_setpriority: | |
a0d0e21e | 4878 | LOP(OP_SETPRIORITY,XTERM); |
79072805 LW |
4879 | |
4880 | case KEY_sethostent: | |
ff68c719 | 4881 | UNI(OP_SHOSTENT); |
79072805 LW |
4882 | |
4883 | case KEY_setnetent: | |
ff68c719 | 4884 | UNI(OP_SNETENT); |
79072805 LW |
4885 | |
4886 | case KEY_setservent: | |
ff68c719 | 4887 | UNI(OP_SSERVENT); |
79072805 LW |
4888 | |
4889 | case KEY_setprotoent: | |
ff68c719 | 4890 | UNI(OP_SPROTOENT); |
79072805 LW |
4891 | |
4892 | case KEY_setpwent: | |
4893 | FUN0(OP_SPWENT); | |
4894 | ||
4895 | case KEY_setgrent: | |
4896 | FUN0(OP_SGRENT); | |
4897 | ||
4898 | case KEY_seekdir: | |
a0d0e21e | 4899 | LOP(OP_SEEKDIR,XTERM); |
79072805 LW |
4900 | |
4901 | case KEY_setsockopt: | |
a0d0e21e | 4902 | LOP(OP_SSOCKOPT,XTERM); |
79072805 LW |
4903 | |
4904 | case KEY_shift: | |
4905 | UNI(OP_SHIFT); | |
4906 | ||
4907 | case KEY_shmctl: | |
a0d0e21e | 4908 | LOP(OP_SHMCTL,XTERM); |
79072805 LW |
4909 | |
4910 | case KEY_shmget: | |
a0d0e21e | 4911 | LOP(OP_SHMGET,XTERM); |
79072805 LW |
4912 | |
4913 | case KEY_shmread: | |
a0d0e21e | 4914 | LOP(OP_SHMREAD,XTERM); |
79072805 LW |
4915 | |
4916 | case KEY_shmwrite: | |
a0d0e21e | 4917 | LOP(OP_SHMWRITE,XTERM); |
79072805 LW |
4918 | |
4919 | case KEY_shutdown: | |
a0d0e21e | 4920 | LOP(OP_SHUTDOWN,XTERM); |
79072805 LW |
4921 | |
4922 | case KEY_sin: | |
4923 | UNI(OP_SIN); | |
4924 | ||
4925 | case KEY_sleep: | |
4926 | UNI(OP_SLEEP); | |
4927 | ||
4928 | case KEY_socket: | |
a0d0e21e | 4929 | LOP(OP_SOCKET,XTERM); |
79072805 LW |
4930 | |
4931 | case KEY_socketpair: | |
a0d0e21e | 4932 | LOP(OP_SOCKPAIR,XTERM); |
79072805 LW |
4933 | |
4934 | case KEY_sort: | |
3280af22 | 4935 | checkcomma(s,PL_tokenbuf,"subroutine name"); |
79072805 LW |
4936 | s = skipspace(s); |
4937 | if (*s == ';' || *s == ')') /* probably a close */ | |
cea2e8a9 | 4938 | Perl_croak(aTHX_ "sort is now a reserved word"); |
3280af22 | 4939 | PL_expect = XTERM; |
15f0808c | 4940 | s = force_word(s,WORD,TRUE,TRUE,FALSE); |
a0d0e21e | 4941 | LOP(OP_SORT,XREF); |
79072805 LW |
4942 | |
4943 | case KEY_split: | |
a0d0e21e | 4944 | LOP(OP_SPLIT,XTERM); |
79072805 LW |
4945 | |
4946 | case KEY_sprintf: | |
a0d0e21e | 4947 | LOP(OP_SPRINTF,XTERM); |
79072805 LW |
4948 | |
4949 | case KEY_splice: | |
a0d0e21e | 4950 | LOP(OP_SPLICE,XTERM); |
79072805 LW |
4951 | |
4952 | case KEY_sqrt: | |
4953 | UNI(OP_SQRT); | |
4954 | ||
4955 | case KEY_srand: | |
4956 | UNI(OP_SRAND); | |
4957 | ||
4958 | case KEY_stat: | |
4959 | UNI(OP_STAT); | |
4960 | ||
4961 | case KEY_study: | |
79072805 LW |
4962 | UNI(OP_STUDY); |
4963 | ||
4964 | case KEY_substr: | |
a0d0e21e | 4965 | LOP(OP_SUBSTR,XTERM); |
79072805 LW |
4966 | |
4967 | case KEY_format: | |
4968 | case KEY_sub: | |
93a17b20 | 4969 | really_sub: |
09bef843 | 4970 | { |
3280af22 | 4971 | char tmpbuf[sizeof PL_tokenbuf]; |
b1b65b59 | 4972 | SSize_t tboffset; |
09bef843 SB |
4973 | expectation attrful; |
4974 | bool have_name, have_proto; | |
4975 | int key = tmp; | |
4976 | ||
4977 | s = skipspace(s); | |
4978 | ||
7e2040f0 | 4979 | if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' || |
09bef843 SB |
4980 | (*s == ':' && s[1] == ':')) |
4981 | { | |
4982 | PL_expect = XBLOCK; | |
4983 | attrful = XATTRBLOCK; | |
b1b65b59 JH |
4984 | /* remember buffer pos'n for later force_word */ |
4985 | tboffset = s - PL_oldbufptr; | |
09bef843 SB |
4986 | d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len); |
4987 | if (strchr(tmpbuf, ':')) | |
4988 | sv_setpv(PL_subname, tmpbuf); | |
4989 | else { | |
4990 | sv_setsv(PL_subname,PL_curstname); | |
4991 | sv_catpvn(PL_subname,"::",2); | |
4992 | sv_catpvn(PL_subname,tmpbuf,len); | |
4993 | } | |
4994 | s = skipspace(d); | |
4995 | have_name = TRUE; | |
4996 | } | |
463ee0b2 | 4997 | else { |
09bef843 SB |
4998 | if (key == KEY_my) |
4999 | Perl_croak(aTHX_ "Missing name in \"my sub\""); | |
5000 | PL_expect = XTERMBLOCK; | |
5001 | attrful = XATTRTERM; | |
5002 | sv_setpv(PL_subname,"?"); | |
5003 | have_name = FALSE; | |
463ee0b2 | 5004 | } |
4633a7c4 | 5005 | |
09bef843 SB |
5006 | if (key == KEY_format) { |
5007 | if (*s == '=') | |
5008 | PL_lex_formbrack = PL_lex_brackets + 1; | |
5009 | if (have_name) | |
b1b65b59 JH |
5010 | (void) force_word(PL_oldbufptr + tboffset, WORD, |
5011 | FALSE, TRUE, TRUE); | |
09bef843 SB |
5012 | OPERATOR(FORMAT); |
5013 | } | |
79072805 | 5014 | |
09bef843 SB |
5015 | /* Look for a prototype */ |
5016 | if (*s == '(') { | |
5017 | char *p; | |
5018 | ||
5019 | s = scan_str(s,FALSE,FALSE); | |
5020 | if (!s) { | |
5021 | if (PL_lex_stuff) | |
5022 | SvREFCNT_dec(PL_lex_stuff); | |
5023 | PL_lex_stuff = Nullsv; | |
5024 | Perl_croak(aTHX_ "Prototype not terminated"); | |
5025 | } | |
5026 | /* strip spaces */ | |
5027 | d = SvPVX(PL_lex_stuff); | |
5028 | tmp = 0; | |
5029 | for (p = d; *p; ++p) { | |
5030 | if (!isSPACE(*p)) | |
5031 | d[tmp++] = *p; | |
5032 | } | |
5033 | d[tmp] = '\0'; | |
5034 | SvCUR(PL_lex_stuff) = tmp; | |
5035 | have_proto = TRUE; | |
68dc0745 | 5036 | |
09bef843 | 5037 | s = skipspace(s); |
4633a7c4 | 5038 | } |
09bef843 SB |
5039 | else |
5040 | have_proto = FALSE; | |
5041 | ||
5042 | if (*s == ':' && s[1] != ':') | |
5043 | PL_expect = attrful; | |
5044 | ||
5045 | if (have_proto) { | |
b1b65b59 JH |
5046 | PL_nextval[PL_nexttoke].opval = |
5047 | (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff); | |
09bef843 SB |
5048 | PL_lex_stuff = Nullsv; |
5049 | force_next(THING); | |
68dc0745 | 5050 | } |
09bef843 SB |
5051 | if (!have_name) { |
5052 | sv_setpv(PL_subname,"__ANON__"); | |
5053 | TOKEN(ANONSUB); | |
4633a7c4 | 5054 | } |
b1b65b59 JH |
5055 | (void) force_word(PL_oldbufptr + tboffset, WORD, |
5056 | FALSE, TRUE, TRUE); | |
09bef843 SB |
5057 | if (key == KEY_my) |
5058 | TOKEN(MYSUB); | |
5059 | TOKEN(SUB); | |
4633a7c4 | 5060 | } |
79072805 LW |
5061 | |
5062 | case KEY_system: | |
5063 | set_csh(); | |
a0d0e21e | 5064 | LOP(OP_SYSTEM,XREF); |
79072805 LW |
5065 | |
5066 | case KEY_symlink: | |
a0d0e21e | 5067 | LOP(OP_SYMLINK,XTERM); |
79072805 LW |
5068 | |
5069 | case KEY_syscall: | |
a0d0e21e | 5070 | LOP(OP_SYSCALL,XTERM); |
79072805 | 5071 | |
c07a80fd | 5072 | case KEY_sysopen: |
5073 | LOP(OP_SYSOPEN,XTERM); | |
5074 | ||
137443ea | 5075 | case KEY_sysseek: |
5076 | LOP(OP_SYSSEEK,XTERM); | |
5077 | ||
79072805 | 5078 | case KEY_sysread: |
a0d0e21e | 5079 | LOP(OP_SYSREAD,XTERM); |
79072805 LW |
5080 | |
5081 | case KEY_syswrite: | |
a0d0e21e | 5082 | LOP(OP_SYSWRITE,XTERM); |
79072805 LW |
5083 | |
5084 | case KEY_tr: | |
5085 | s = scan_trans(s); | |
5086 | TERM(sublex_start()); | |
5087 | ||
5088 | case KEY_tell: | |
5089 | UNI(OP_TELL); | |
5090 | ||
5091 | case KEY_telldir: | |
5092 | UNI(OP_TELLDIR); | |
5093 | ||
463ee0b2 | 5094 | case KEY_tie: |
a0d0e21e | 5095 | LOP(OP_TIE,XTERM); |
463ee0b2 | 5096 | |
c07a80fd | 5097 | case KEY_tied: |
5098 | UNI(OP_TIED); | |
5099 | ||
79072805 LW |
5100 | case KEY_time: |
5101 | FUN0(OP_TIME); | |
5102 | ||
5103 | case KEY_times: | |
5104 | FUN0(OP_TMS); | |
5105 | ||
5106 | case KEY_truncate: | |
a0d0e21e | 5107 | LOP(OP_TRUNCATE,XTERM); |
79072805 LW |
5108 | |
5109 | case KEY_uc: | |
5110 | UNI(OP_UC); | |
5111 | ||
5112 | case KEY_ucfirst: | |
5113 | UNI(OP_UCFIRST); | |
5114 | ||
463ee0b2 LW |
5115 | case KEY_untie: |
5116 | UNI(OP_UNTIE); | |
5117 | ||
79072805 | 5118 | case KEY_until: |
57843af0 | 5119 | yylval.ival = CopLINE(PL_curcop); |
79072805 LW |
5120 | OPERATOR(UNTIL); |
5121 | ||
5122 | case KEY_unless: | |
57843af0 | 5123 | yylval.ival = CopLINE(PL_curcop); |
79072805 LW |
5124 | OPERATOR(UNLESS); |
5125 | ||
5126 | case KEY_unlink: | |
a0d0e21e | 5127 | LOP(OP_UNLINK,XTERM); |
79072805 LW |
5128 | |
5129 | case KEY_undef: | |
5130 | UNI(OP_UNDEF); | |
5131 | ||
5132 | case KEY_unpack: | |
a0d0e21e | 5133 | LOP(OP_UNPACK,XTERM); |
79072805 LW |
5134 | |
5135 | case KEY_utime: | |
a0d0e21e | 5136 | LOP(OP_UTIME,XTERM); |
79072805 LW |
5137 | |
5138 | case KEY_umask: | |
e476b1b5 | 5139 | if (ckWARN(WARN_UMASK)) { |
3280af22 | 5140 | for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ; |
4e553d73 | 5141 | if (*d != '0' && isDIGIT(*d)) |
e476b1b5 | 5142 | Perl_warner(aTHX_ WARN_UMASK, |
4438c4b7 | 5143 | "umask: argument is missing initial 0"); |
748a9306 | 5144 | } |
79072805 LW |
5145 | UNI(OP_UMASK); |
5146 | ||
5147 | case KEY_unshift: | |
a0d0e21e LW |
5148 | LOP(OP_UNSHIFT,XTERM); |
5149 | ||
5150 | case KEY_use: | |
3280af22 | 5151 | if (PL_expect != XSTATE) |
a0d0e21e | 5152 | yyerror("\"use\" not allowed in expression"); |
89bfa8cd | 5153 | s = skipspace(s); |
a7cb1f99 | 5154 | if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) { |
89bfa8cd | 5155 | s = force_version(s); |
a7cb1f99 | 5156 | if (*s == ';' || (s = skipspace(s), *s == ';')) { |
3280af22 | 5157 | PL_nextval[PL_nexttoke].opval = Nullop; |
89bfa8cd | 5158 | force_next(WORD); |
5159 | } | |
5160 | } | |
5161 | else { | |
5162 | s = force_word(s,WORD,FALSE,TRUE,FALSE); | |
5163 | s = force_version(s); | |
5164 | } | |
a0d0e21e LW |
5165 | yylval.ival = 1; |
5166 | OPERATOR(USE); | |
79072805 LW |
5167 | |
5168 | case KEY_values: | |
5169 | UNI(OP_VALUES); | |
5170 | ||
5171 | case KEY_vec: | |
a0d0e21e | 5172 | LOP(OP_VEC,XTERM); |
79072805 LW |
5173 | |
5174 | case KEY_while: | |
57843af0 | 5175 | yylval.ival = CopLINE(PL_curcop); |
79072805 LW |
5176 | OPERATOR(WHILE); |
5177 | ||
5178 | case KEY_warn: | |
3280af22 | 5179 | PL_hints |= HINT_BLOCK_SCOPE; |
a0d0e21e | 5180 | LOP(OP_WARN,XTERM); |
79072805 LW |
5181 | |
5182 | case KEY_wait: | |
5183 | FUN0(OP_WAIT); | |
5184 | ||
5185 | case KEY_waitpid: | |
a0d0e21e | 5186 | LOP(OP_WAITPID,XTERM); |
79072805 LW |
5187 | |
5188 | case KEY_wantarray: | |
5189 | FUN0(OP_WANTARRAY); | |
5190 | ||
5191 | case KEY_write: | |
9d116dd7 JH |
5192 | #ifdef EBCDIC |
5193 | { | |
5194 | static char ctl_l[2]; | |
5195 | ||
4e553d73 | 5196 | if (ctl_l[0] == '\0') |
9d116dd7 JH |
5197 | ctl_l[0] = toCTRL('L'); |
5198 | gv_fetchpv(ctl_l,TRUE, SVt_PV); | |
5199 | } | |
5200 | #else | |
5201 | gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */ | |
5202 | #endif | |
79072805 LW |
5203 | UNI(OP_ENTERWRITE); |
5204 | ||
5205 | case KEY_x: | |
3280af22 | 5206 | if (PL_expect == XOPERATOR) |
79072805 LW |
5207 | Mop(OP_REPEAT); |
5208 | check_uni(); | |
5209 | goto just_a_word; | |
5210 | ||
a0d0e21e LW |
5211 | case KEY_xor: |
5212 | yylval.ival = OP_XOR; | |
5213 | OPERATOR(OROP); | |
5214 | ||
79072805 LW |
5215 | case KEY_y: |
5216 | s = scan_trans(s); | |
5217 | TERM(sublex_start()); | |
5218 | } | |
49dc05e3 | 5219 | }} |
79072805 | 5220 | } |
bf4acbe4 GS |
5221 | #ifdef __SC__ |
5222 | #pragma segment Main | |
5223 | #endif | |
79072805 LW |
5224 | |
5225 | I32 | |
864dbfa3 | 5226 | Perl_keyword(pTHX_ register char *d, I32 len) |
79072805 LW |
5227 | { |
5228 | switch (*d) { | |
5229 | case '_': | |
5230 | if (d[1] == '_') { | |
a0d0e21e | 5231 | if (strEQ(d,"__FILE__")) return -KEY___FILE__; |
68dc0745 | 5232 | if (strEQ(d,"__LINE__")) return -KEY___LINE__; |
5233 | if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__; | |
e50aee73 | 5234 | if (strEQ(d,"__DATA__")) return KEY___DATA__; |
79072805 LW |
5235 | if (strEQ(d,"__END__")) return KEY___END__; |
5236 | } | |
5237 | break; | |
8990e307 LW |
5238 | case 'A': |
5239 | if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD; | |
5240 | break; | |
79072805 | 5241 | case 'a': |
463ee0b2 LW |
5242 | switch (len) { |
5243 | case 3: | |
a0d0e21e LW |
5244 | if (strEQ(d,"and")) return -KEY_and; |
5245 | if (strEQ(d,"abs")) return -KEY_abs; | |
85e6fe83 | 5246 | break; |
463ee0b2 | 5247 | case 5: |
a0d0e21e LW |
5248 | if (strEQ(d,"alarm")) return -KEY_alarm; |
5249 | if (strEQ(d,"atan2")) return -KEY_atan2; | |
463ee0b2 LW |
5250 | break; |
5251 | case 6: | |
a0d0e21e | 5252 | if (strEQ(d,"accept")) return -KEY_accept; |
463ee0b2 LW |
5253 | break; |
5254 | } | |
79072805 LW |
5255 | break; |
5256 | case 'B': | |
5257 | if (strEQ(d,"BEGIN")) return KEY_BEGIN; | |
93a17b20 | 5258 | break; |
79072805 | 5259 | case 'b': |
a0d0e21e LW |
5260 | if (strEQ(d,"bless")) return -KEY_bless; |
5261 | if (strEQ(d,"bind")) return -KEY_bind; | |
5262 | if (strEQ(d,"binmode")) return -KEY_binmode; | |
5263 | break; | |
5264 | case 'C': | |
5265 | if (strEQ(d,"CORE")) return -KEY_CORE; | |
7d30b5c4 | 5266 | if (strEQ(d,"CHECK")) return KEY_CHECK; |
79072805 LW |
5267 | break; |
5268 | case 'c': | |
5269 | switch (len) { | |
5270 | case 3: | |
a0d0e21e LW |
5271 | if (strEQ(d,"cmp")) return -KEY_cmp; |
5272 | if (strEQ(d,"chr")) return -KEY_chr; | |
5273 | if (strEQ(d,"cos")) return -KEY_cos; | |
79072805 LW |
5274 | break; |
5275 | case 4: | |
79e5458b | 5276 | if (strEQ(d,"chop")) return -KEY_chop; |
79072805 LW |
5277 | break; |
5278 | case 5: | |
a0d0e21e LW |
5279 | if (strEQ(d,"close")) return -KEY_close; |
5280 | if (strEQ(d,"chdir")) return -KEY_chdir; | |
79e5458b | 5281 | if (strEQ(d,"chomp")) return -KEY_chomp; |
a0d0e21e LW |
5282 | if (strEQ(d,"chmod")) return -KEY_chmod; |
5283 | if (strEQ(d,"chown")) return -KEY_chown; | |
5284 | if (strEQ(d,"crypt")) return -KEY_crypt; | |
79072805 LW |
5285 | break; |
5286 | case 6: | |
a0d0e21e LW |
5287 | if (strEQ(d,"chroot")) return -KEY_chroot; |
5288 | if (strEQ(d,"caller")) return -KEY_caller; | |
79072805 LW |
5289 | break; |
5290 | case 7: | |
a0d0e21e | 5291 | if (strEQ(d,"connect")) return -KEY_connect; |
79072805 LW |
5292 | break; |
5293 | case 8: | |
a0d0e21e LW |
5294 | if (strEQ(d,"closedir")) return -KEY_closedir; |
5295 | if (strEQ(d,"continue")) return -KEY_continue; | |
79072805 LW |
5296 | break; |
5297 | } | |
5298 | break; | |
ed6116ce LW |
5299 | case 'D': |
5300 | if (strEQ(d,"DESTROY")) return KEY_DESTROY; | |
5301 | break; | |
79072805 LW |
5302 | case 'd': |
5303 | switch (len) { | |
5304 | case 2: | |
5305 | if (strEQ(d,"do")) return KEY_do; | |
5306 | break; | |
5307 | case 3: | |
a0d0e21e | 5308 | if (strEQ(d,"die")) return -KEY_die; |
79072805 LW |
5309 | break; |
5310 | case 4: | |
a0d0e21e | 5311 | if (strEQ(d,"dump")) return -KEY_dump; |
79072805 LW |
5312 | break; |
5313 | case 6: | |
5314 | if (strEQ(d,"delete")) return KEY_delete; | |
5315 | break; | |
5316 | case 7: | |
5317 | if (strEQ(d,"defined")) return KEY_defined; | |
a0d0e21e | 5318 | if (strEQ(d,"dbmopen")) return -KEY_dbmopen; |
79072805 LW |
5319 | break; |
5320 | case 8: | |
a0d0e21e | 5321 | if (strEQ(d,"dbmclose")) return -KEY_dbmclose; |
79072805 LW |
5322 | break; |
5323 | } | |
5324 | break; | |
5325 | case 'E': | |
79072805 LW |
5326 | if (strEQ(d,"END")) return KEY_END; |
5327 | break; | |
5328 | case 'e': | |
5329 | switch (len) { | |
5330 | case 2: | |
a0d0e21e | 5331 | if (strEQ(d,"eq")) return -KEY_eq; |
79072805 LW |
5332 | break; |
5333 | case 3: | |
a0d0e21e LW |
5334 | if (strEQ(d,"eof")) return -KEY_eof; |
5335 | if (strEQ(d,"exp")) return -KEY_exp; | |
79072805 LW |
5336 | break; |
5337 | case 4: | |
5338 | if (strEQ(d,"else")) return KEY_else; | |
a0d0e21e | 5339 | if (strEQ(d,"exit")) return -KEY_exit; |
79072805 | 5340 | if (strEQ(d,"eval")) return KEY_eval; |
a0d0e21e | 5341 | if (strEQ(d,"exec")) return -KEY_exec; |
3a6a8333 | 5342 | if (strEQ(d,"each")) return -KEY_each; |
79072805 LW |
5343 | break; |
5344 | case 5: | |
5345 | if (strEQ(d,"elsif")) return KEY_elsif; | |
5346 | break; | |
a0d0e21e LW |
5347 | case 6: |
5348 | if (strEQ(d,"exists")) return KEY_exists; | |
cea2e8a9 | 5349 | if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif"); |
a0d0e21e | 5350 | break; |
79072805 | 5351 | case 8: |
a0d0e21e LW |
5352 | if (strEQ(d,"endgrent")) return -KEY_endgrent; |
5353 | if (strEQ(d,"endpwent")) return -KEY_endpwent; | |
79072805 LW |
5354 | break; |
5355 | case 9: | |
a0d0e21e | 5356 | if (strEQ(d,"endnetent")) return -KEY_endnetent; |
79072805 LW |
5357 | break; |
5358 | case 10: | |
a0d0e21e LW |
5359 | if (strEQ(d,"endhostent")) return -KEY_endhostent; |
5360 | if (strEQ(d,"endservent")) return -KEY_endservent; | |
79072805 LW |
5361 | break; |
5362 | case 11: | |
a0d0e21e | 5363 | if (strEQ(d,"endprotoent")) return -KEY_endprotoent; |
79072805 | 5364 | break; |
a687059c | 5365 | } |
a687059c | 5366 | break; |
79072805 LW |
5367 | case 'f': |
5368 | switch (len) { | |
5369 | case 3: | |
5370 | if (strEQ(d,"for")) return KEY_for; | |
5371 | break; | |
5372 | case 4: | |
a0d0e21e | 5373 | if (strEQ(d,"fork")) return -KEY_fork; |
79072805 LW |
5374 | break; |
5375 | case 5: | |
a0d0e21e LW |
5376 | if (strEQ(d,"fcntl")) return -KEY_fcntl; |
5377 | if (strEQ(d,"flock")) return -KEY_flock; | |
79072805 LW |
5378 | break; |
5379 | case 6: | |
5380 | if (strEQ(d,"format")) return KEY_format; | |
a0d0e21e | 5381 | if (strEQ(d,"fileno")) return -KEY_fileno; |
79072805 LW |
5382 | break; |
5383 | case 7: | |
5384 | if (strEQ(d,"foreach")) return KEY_foreach; | |
5385 | break; | |
5386 | case 8: | |
a0d0e21e | 5387 | if (strEQ(d,"formline")) return -KEY_formline; |
79072805 | 5388 | break; |
378cc40b | 5389 | } |
a687059c | 5390 | break; |
79072805 | 5391 | case 'g': |
a687059c LW |
5392 | if (strnEQ(d,"get",3)) { |
5393 | d += 3; | |
5394 | if (*d == 'p') { | |
79072805 LW |
5395 | switch (len) { |
5396 | case 7: | |
a0d0e21e LW |
5397 | if (strEQ(d,"ppid")) return -KEY_getppid; |
5398 | if (strEQ(d,"pgrp")) return -KEY_getpgrp; | |
79072805 LW |
5399 | break; |
5400 | case 8: | |
a0d0e21e LW |
5401 | if (strEQ(d,"pwent")) return -KEY_getpwent; |
5402 | if (strEQ(d,"pwnam")) return -KEY_getpwnam; | |
5403 | if (strEQ(d,"pwuid")) return -KEY_getpwuid; | |
79072805 LW |
5404 | break; |
5405 | case 11: | |
a0d0e21e LW |
5406 | if (strEQ(d,"peername")) return -KEY_getpeername; |
5407 | if (strEQ(d,"protoent")) return -KEY_getprotoent; | |
5408 | if (strEQ(d,"priority")) return -KEY_getpriority; | |
79072805 LW |
5409 | break; |
5410 | case 14: | |
a0d0e21e | 5411 | if (strEQ(d,"protobyname")) return -KEY_getprotobyname; |
79072805 LW |
5412 | break; |
5413 | case 16: | |
a0d0e21e | 5414 | if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber; |
79072805 LW |
5415 | break; |
5416 | } | |
a687059c LW |
5417 | } |
5418 | else if (*d == 'h') { | |
a0d0e21e LW |
5419 | if (strEQ(d,"hostbyname")) return -KEY_gethostbyname; |
5420 | if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr; | |
5421 | if (strEQ(d,"hostent")) return -KEY_gethostent; | |
a687059c LW |
5422 | } |
5423 | else if (*d == 'n') { | |
a0d0e21e LW |
5424 | if (strEQ(d,"netbyname")) return -KEY_getnetbyname; |
5425 | if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr; | |
5426 | if (strEQ(d,"netent")) return -KEY_getnetent; | |
a687059c LW |
5427 | } |
5428 | else if (*d == 's') { | |
a0d0e21e LW |
5429 | if (strEQ(d,"servbyname")) return -KEY_getservbyname; |
5430 | if (strEQ(d,"servbyport")) return -KEY_getservbyport; | |
5431 | if (strEQ(d,"servent")) return -KEY_getservent; | |
5432 | if (strEQ(d,"sockname")) return -KEY_getsockname; | |
5433 | if (strEQ(d,"sockopt")) return -KEY_getsockopt; | |
a687059c LW |
5434 | } |
5435 | else if (*d == 'g') { | |
a0d0e21e LW |
5436 | if (strEQ(d,"grent")) return -KEY_getgrent; |
5437 | if (strEQ(d,"grnam")) return -KEY_getgrnam; | |
5438 | if (strEQ(d,"grgid")) return -KEY_getgrgid; | |
a687059c LW |
5439 | } |
5440 | else if (*d == 'l') { | |
a0d0e21e | 5441 | if (strEQ(d,"login")) return -KEY_getlogin; |
a687059c | 5442 | } |
a0d0e21e | 5443 | else if (strEQ(d,"c")) return -KEY_getc; |
79072805 | 5444 | break; |
a687059c | 5445 | } |
79072805 LW |
5446 | switch (len) { |
5447 | case 2: | |
a0d0e21e LW |
5448 | if (strEQ(d,"gt")) return -KEY_gt; |
5449 | if (strEQ(d,"ge")) return -KEY_ge; | |
79072805 LW |
5450 | break; |
5451 | case 4: | |
5452 | if (strEQ(d,"grep")) return KEY_grep; | |
5453 | if (strEQ(d,"goto")) return KEY_goto; | |
fb73857a | 5454 | if (strEQ(d,"glob")) return KEY_glob; |
79072805 LW |
5455 | break; |
5456 | case 6: | |
a0d0e21e | 5457 | if (strEQ(d,"gmtime")) return -KEY_gmtime; |
79072805 | 5458 | break; |
378cc40b | 5459 | } |
a687059c | 5460 | break; |
79072805 | 5461 | case 'h': |
a0d0e21e | 5462 | if (strEQ(d,"hex")) return -KEY_hex; |
a687059c | 5463 | break; |
7d07dbc2 MB |
5464 | case 'I': |
5465 | if (strEQ(d,"INIT")) return KEY_INIT; | |
5466 | break; | |
79072805 LW |
5467 | case 'i': |
5468 | switch (len) { | |
5469 | case 2: | |
5470 | if (strEQ(d,"if")) return KEY_if; | |
5471 | break; | |
5472 | case 3: | |
a0d0e21e | 5473 | if (strEQ(d,"int")) return -KEY_int; |
79072805 LW |
5474 | break; |
5475 | case 5: | |
a0d0e21e LW |
5476 | if (strEQ(d,"index")) return -KEY_index; |
5477 | if (strEQ(d,"ioctl")) return -KEY_ioctl; | |
79072805 LW |
5478 | break; |
5479 | } | |
a687059c | 5480 | break; |
79072805 | 5481 | case 'j': |
a0d0e21e | 5482 | if (strEQ(d,"join")) return -KEY_join; |
a687059c | 5483 | break; |
79072805 LW |
5484 | case 'k': |
5485 | if (len == 4) { | |
3a6a8333 | 5486 | if (strEQ(d,"keys")) return -KEY_keys; |
a0d0e21e | 5487 | if (strEQ(d,"kill")) return -KEY_kill; |
663a0e37 | 5488 | } |
79072805 | 5489 | break; |
79072805 LW |
5490 | case 'l': |
5491 | switch (len) { | |
5492 | case 2: | |
a0d0e21e LW |
5493 | if (strEQ(d,"lt")) return -KEY_lt; |
5494 | if (strEQ(d,"le")) return -KEY_le; | |
5495 | if (strEQ(d,"lc")) return -KEY_lc; | |
79072805 LW |
5496 | break; |
5497 | case 3: | |
a0d0e21e | 5498 | if (strEQ(d,"log")) return -KEY_log; |
79072805 LW |
5499 | break; |
5500 | case 4: | |
5501 | if (strEQ(d,"last")) return KEY_last; | |
a0d0e21e | 5502 | if (strEQ(d,"link")) return -KEY_link; |
c0329465 | 5503 | if (strEQ(d,"lock")) return -KEY_lock; |
395c3793 | 5504 | break; |
79072805 LW |
5505 | case 5: |
5506 | if (strEQ(d,"local")) return KEY_local; | |
a0d0e21e | 5507 | if (strEQ(d,"lstat")) return -KEY_lstat; |
79072805 LW |
5508 | break; |
5509 | case 6: | |
a0d0e21e LW |
5510 | if (strEQ(d,"length")) return -KEY_length; |
5511 | if (strEQ(d,"listen")) return -KEY_listen; | |
79072805 LW |
5512 | break; |
5513 | case 7: | |
a0d0e21e | 5514 | if (strEQ(d,"lcfirst")) return -KEY_lcfirst; |
79072805 LW |
5515 | break; |
5516 | case 9: | |
a0d0e21e | 5517 | if (strEQ(d,"localtime")) return -KEY_localtime; |
395c3793 LW |
5518 | break; |
5519 | } | |
a687059c | 5520 | break; |
79072805 LW |
5521 | case 'm': |
5522 | switch (len) { | |
5523 | case 1: return KEY_m; | |
93a17b20 LW |
5524 | case 2: |
5525 | if (strEQ(d,"my")) return KEY_my; | |
5526 | break; | |
a0d0e21e LW |
5527 | case 3: |
5528 | if (strEQ(d,"map")) return KEY_map; | |
5529 | break; | |
79072805 | 5530 | case 5: |
a0d0e21e | 5531 | if (strEQ(d,"mkdir")) return -KEY_mkdir; |
79072805 LW |
5532 | break; |
5533 | case 6: | |
a0d0e21e LW |
5534 | if (strEQ(d,"msgctl")) return -KEY_msgctl; |
5535 | if (strEQ(d,"msgget")) return -KEY_msgget; | |
5536 | if (strEQ(d,"msgrcv")) return -KEY_msgrcv; | |
5537 | if (strEQ(d,"msgsnd")) return -KEY_msgsnd; | |
79072805 LW |
5538 | break; |
5539 | } | |
a687059c | 5540 | break; |
79072805 LW |
5541 | case 'n': |
5542 | if (strEQ(d,"next")) return KEY_next; | |
a0d0e21e LW |
5543 | if (strEQ(d,"ne")) return -KEY_ne; |
5544 | if (strEQ(d,"not")) return -KEY_not; | |
5545 | if (strEQ(d,"no")) return KEY_no; | |
a687059c | 5546 | break; |
79072805 LW |
5547 | case 'o': |
5548 | switch (len) { | |
463ee0b2 | 5549 | case 2: |
a0d0e21e | 5550 | if (strEQ(d,"or")) return -KEY_or; |
463ee0b2 | 5551 | break; |
79072805 | 5552 | case 3: |
a0d0e21e LW |
5553 | if (strEQ(d,"ord")) return -KEY_ord; |
5554 | if (strEQ(d,"oct")) return -KEY_oct; | |
77ca0c92 | 5555 | if (strEQ(d,"our")) return KEY_our; |
79072805 LW |
5556 | break; |
5557 | case 4: | |
a0d0e21e | 5558 | if (strEQ(d,"open")) return -KEY_open; |
79072805 LW |
5559 | break; |
5560 | case 7: | |
a0d0e21e | 5561 | if (strEQ(d,"opendir")) return -KEY_opendir; |
79072805 | 5562 | break; |
fe14fcc3 | 5563 | } |
a687059c | 5564 | break; |
79072805 LW |
5565 | case 'p': |
5566 | switch (len) { | |
5567 | case 3: | |
4e553d73 | 5568 | if (strEQ(d,"pop")) return -KEY_pop; |
a0d0e21e | 5569 | if (strEQ(d,"pos")) return KEY_pos; |
79072805 LW |
5570 | break; |
5571 | case 4: | |
3a6a8333 | 5572 | if (strEQ(d,"push")) return -KEY_push; |
a0d0e21e LW |
5573 | if (strEQ(d,"pack")) return -KEY_pack; |
5574 | if (strEQ(d,"pipe")) return -KEY_pipe; | |
79072805 LW |
5575 | break; |
5576 | case 5: | |
5577 | if (strEQ(d,"print")) return KEY_print; | |
5578 | break; | |
5579 | case 6: | |
5580 | if (strEQ(d,"printf")) return KEY_printf; | |
5581 | break; | |
5582 | case 7: | |
5583 | if (strEQ(d,"package")) return KEY_package; | |
5584 | break; | |
c07a80fd | 5585 | case 9: |
5586 | if (strEQ(d,"prototype")) return KEY_prototype; | |
663a0e37 | 5587 | } |
79072805 LW |
5588 | break; |
5589 | case 'q': | |
5590 | if (len <= 2) { | |
5591 | if (strEQ(d,"q")) return KEY_q; | |
8782bef2 | 5592 | if (strEQ(d,"qr")) return KEY_qr; |
79072805 | 5593 | if (strEQ(d,"qq")) return KEY_qq; |
9aa983d2 | 5594 | if (strEQ(d,"qu")) return KEY_qu; |
8990e307 | 5595 | if (strEQ(d,"qw")) return KEY_qw; |
79072805 | 5596 | if (strEQ(d,"qx")) return KEY_qx; |
663a0e37 | 5597 | } |
a0d0e21e | 5598 | else if (strEQ(d,"quotemeta")) return -KEY_quotemeta; |
79072805 LW |
5599 | break; |
5600 | case 'r': | |
5601 | switch (len) { | |
5602 | case 3: | |
a0d0e21e | 5603 | if (strEQ(d,"ref")) return -KEY_ref; |
79072805 LW |
5604 | break; |
5605 | case 4: | |
a0d0e21e LW |
5606 | if (strEQ(d,"read")) return -KEY_read; |
5607 | if (strEQ(d,"rand")) return -KEY_rand; | |
5608 | if (strEQ(d,"recv")) return -KEY_recv; | |
79072805 LW |
5609 | if (strEQ(d,"redo")) return KEY_redo; |
5610 | break; | |
5611 | case 5: | |
a0d0e21e LW |
5612 | if (strEQ(d,"rmdir")) return -KEY_rmdir; |
5613 | if (strEQ(d,"reset")) return -KEY_reset; | |
79072805 LW |
5614 | break; |
5615 | case 6: | |
5616 | if (strEQ(d,"return")) return KEY_return; | |
a0d0e21e LW |
5617 | if (strEQ(d,"rename")) return -KEY_rename; |
5618 | if (strEQ(d,"rindex")) return -KEY_rindex; | |
79072805 LW |
5619 | break; |
5620 | case 7: | |
a0d0e21e LW |
5621 | if (strEQ(d,"require")) return -KEY_require; |
5622 | if (strEQ(d,"reverse")) return -KEY_reverse; | |
5623 | if (strEQ(d,"readdir")) return -KEY_readdir; | |
79072805 LW |
5624 | break; |
5625 | case 8: | |
a0d0e21e LW |
5626 | if (strEQ(d,"readlink")) return -KEY_readlink; |
5627 | if (strEQ(d,"readline")) return -KEY_readline; | |
5628 | if (strEQ(d,"readpipe")) return -KEY_readpipe; | |
79072805 LW |
5629 | break; |
5630 | case 9: | |
a0d0e21e | 5631 | if (strEQ(d,"rewinddir")) return -KEY_rewinddir; |
79072805 | 5632 | break; |
a687059c | 5633 | } |
79072805 LW |
5634 | break; |
5635 | case 's': | |
a687059c | 5636 | switch (d[1]) { |
79072805 | 5637 | case 0: return KEY_s; |
a687059c | 5638 | case 'c': |
79072805 | 5639 | if (strEQ(d,"scalar")) return KEY_scalar; |
a687059c LW |
5640 | break; |
5641 | case 'e': | |
79072805 LW |
5642 | switch (len) { |
5643 | case 4: | |
a0d0e21e LW |
5644 | if (strEQ(d,"seek")) return -KEY_seek; |
5645 | if (strEQ(d,"send")) return -KEY_send; | |
79072805 LW |
5646 | break; |
5647 | case 5: | |
a0d0e21e | 5648 | if (strEQ(d,"semop")) return -KEY_semop; |
79072805 LW |
5649 | break; |
5650 | case 6: | |
a0d0e21e LW |
5651 | if (strEQ(d,"select")) return -KEY_select; |
5652 | if (strEQ(d,"semctl")) return -KEY_semctl; | |
5653 | if (strEQ(d,"semget")) return -KEY_semget; | |
79072805 LW |
5654 | break; |
5655 | case 7: | |
a0d0e21e LW |
5656 | if (strEQ(d,"setpgrp")) return -KEY_setpgrp; |
5657 | if (strEQ(d,"seekdir")) return -KEY_seekdir; | |
79072805 LW |
5658 | break; |
5659 | case 8: | |
a0d0e21e LW |
5660 | if (strEQ(d,"setpwent")) return -KEY_setpwent; |
5661 | if (strEQ(d,"setgrent")) return -KEY_setgrent; | |
79072805 LW |
5662 | break; |
5663 | case 9: | |
a0d0e21e | 5664 | if (strEQ(d,"setnetent")) return -KEY_setnetent; |
79072805 LW |
5665 | break; |
5666 | case 10: | |
a0d0e21e LW |
5667 | if (strEQ(d,"setsockopt")) return -KEY_setsockopt; |
5668 | if (strEQ(d,"sethostent")) return -KEY_sethostent; | |
5669 | if (strEQ(d,"setservent")) return -KEY_setservent; | |
79072805 LW |
5670 | break; |
5671 | case 11: | |
a0d0e21e LW |
5672 | if (strEQ(d,"setpriority")) return -KEY_setpriority; |
5673 | if (strEQ(d,"setprotoent")) return -KEY_setprotoent; | |
79072805 LW |
5674 | break; |
5675 | } | |
a687059c LW |
5676 | break; |
5677 | case 'h': | |
79072805 LW |
5678 | switch (len) { |
5679 | case 5: | |
3a6a8333 | 5680 | if (strEQ(d,"shift")) return -KEY_shift; |
79072805 LW |
5681 | break; |
5682 | case 6: | |
a0d0e21e LW |
5683 | if (strEQ(d,"shmctl")) return -KEY_shmctl; |
5684 | if (strEQ(d,"shmget")) return -KEY_shmget; | |
79072805 LW |
5685 | break; |
5686 | case 7: | |
a0d0e21e | 5687 | if (strEQ(d,"shmread")) return -KEY_shmread; |
79072805 LW |
5688 | break; |
5689 | case 8: | |
a0d0e21e LW |
5690 | if (strEQ(d,"shmwrite")) return -KEY_shmwrite; |
5691 | if (strEQ(d,"shutdown")) return -KEY_shutdown; | |
79072805 LW |
5692 | break; |
5693 | } | |
a687059c LW |
5694 | break; |
5695 | case 'i': | |
a0d0e21e | 5696 | if (strEQ(d,"sin")) return -KEY_sin; |
a687059c LW |
5697 | break; |
5698 | case 'l': | |
a0d0e21e | 5699 | if (strEQ(d,"sleep")) return -KEY_sleep; |
a687059c LW |
5700 | break; |
5701 | case 'o': | |
79072805 | 5702 | if (strEQ(d,"sort")) return KEY_sort; |
a0d0e21e LW |
5703 | if (strEQ(d,"socket")) return -KEY_socket; |
5704 | if (strEQ(d,"socketpair")) return -KEY_socketpair; | |
a687059c LW |
5705 | break; |
5706 | case 'p': | |
79072805 | 5707 | if (strEQ(d,"split")) return KEY_split; |
a0d0e21e | 5708 | if (strEQ(d,"sprintf")) return -KEY_sprintf; |
3a6a8333 | 5709 | if (strEQ(d,"splice")) return -KEY_splice; |
a687059c LW |
5710 | break; |
5711 | case 'q': | |
a0d0e21e | 5712 | if (strEQ(d,"sqrt")) return -KEY_sqrt; |
a687059c LW |
5713 | break; |
5714 | case 'r': | |
a0d0e21e | 5715 | if (strEQ(d,"srand")) return -KEY_srand; |
a687059c LW |
5716 | break; |
5717 | case 't': | |
a0d0e21e | 5718 | if (strEQ(d,"stat")) return -KEY_stat; |
79072805 | 5719 | if (strEQ(d,"study")) return KEY_study; |
a687059c LW |
5720 | break; |
5721 | case 'u': | |
a0d0e21e | 5722 | if (strEQ(d,"substr")) return -KEY_substr; |
79072805 | 5723 | if (strEQ(d,"sub")) return KEY_sub; |
a687059c LW |
5724 | break; |
5725 | case 'y': | |
79072805 LW |
5726 | switch (len) { |
5727 | case 6: | |
a0d0e21e | 5728 | if (strEQ(d,"system")) return -KEY_system; |
79072805 LW |
5729 | break; |
5730 | case 7: | |
a0d0e21e LW |
5731 | if (strEQ(d,"symlink")) return -KEY_symlink; |
5732 | if (strEQ(d,"syscall")) return -KEY_syscall; | |
137443ea | 5733 | if (strEQ(d,"sysopen")) return -KEY_sysopen; |
5734 | if (strEQ(d,"sysread")) return -KEY_sysread; | |
5735 | if (strEQ(d,"sysseek")) return -KEY_sysseek; | |
79072805 LW |
5736 | break; |
5737 | case 8: | |
a0d0e21e | 5738 | if (strEQ(d,"syswrite")) return -KEY_syswrite; |
79072805 | 5739 | break; |
a687059c | 5740 | } |
a687059c LW |
5741 | break; |
5742 | } | |
5743 | break; | |
79072805 LW |
5744 | case 't': |
5745 | switch (len) { | |
5746 | case 2: | |
5747 | if (strEQ(d,"tr")) return KEY_tr; | |
5748 | break; | |
463ee0b2 LW |
5749 | case 3: |
5750 | if (strEQ(d,"tie")) return KEY_tie; | |
5751 | break; | |
79072805 | 5752 | case 4: |
a0d0e21e | 5753 | if (strEQ(d,"tell")) return -KEY_tell; |
c07a80fd | 5754 | if (strEQ(d,"tied")) return KEY_tied; |
a0d0e21e | 5755 | if (strEQ(d,"time")) return -KEY_time; |
79072805 LW |
5756 | break; |
5757 | case 5: | |
a0d0e21e | 5758 | if (strEQ(d,"times")) return -KEY_times; |
79072805 LW |
5759 | break; |
5760 | case 7: | |
a0d0e21e | 5761 | if (strEQ(d,"telldir")) return -KEY_telldir; |
79072805 LW |
5762 | break; |
5763 | case 8: | |
a0d0e21e | 5764 | if (strEQ(d,"truncate")) return -KEY_truncate; |
79072805 | 5765 | break; |
378cc40b | 5766 | } |
a687059c | 5767 | break; |
79072805 LW |
5768 | case 'u': |
5769 | switch (len) { | |
5770 | case 2: | |
a0d0e21e LW |
5771 | if (strEQ(d,"uc")) return -KEY_uc; |
5772 | break; | |
5773 | case 3: | |
5774 | if (strEQ(d,"use")) return KEY_use; | |
79072805 LW |
5775 | break; |
5776 | case 5: | |
5777 | if (strEQ(d,"undef")) return KEY_undef; | |
5778 | if (strEQ(d,"until")) return KEY_until; | |
463ee0b2 | 5779 | if (strEQ(d,"untie")) return KEY_untie; |
a0d0e21e LW |
5780 | if (strEQ(d,"utime")) return -KEY_utime; |
5781 | if (strEQ(d,"umask")) return -KEY_umask; | |
79072805 LW |
5782 | break; |
5783 | case 6: | |
5784 | if (strEQ(d,"unless")) return KEY_unless; | |
a0d0e21e LW |
5785 | if (strEQ(d,"unpack")) return -KEY_unpack; |
5786 | if (strEQ(d,"unlink")) return -KEY_unlink; | |
79072805 LW |
5787 | break; |
5788 | case 7: | |
3a6a8333 | 5789 | if (strEQ(d,"unshift")) return -KEY_unshift; |
a0d0e21e | 5790 | if (strEQ(d,"ucfirst")) return -KEY_ucfirst; |
79072805 | 5791 | break; |
a687059c LW |
5792 | } |
5793 | break; | |
79072805 | 5794 | case 'v': |
a0d0e21e LW |
5795 | if (strEQ(d,"values")) return -KEY_values; |
5796 | if (strEQ(d,"vec")) return -KEY_vec; | |
a687059c | 5797 | break; |
79072805 LW |
5798 | case 'w': |
5799 | switch (len) { | |
5800 | case 4: | |
a0d0e21e LW |
5801 | if (strEQ(d,"warn")) return -KEY_warn; |
5802 | if (strEQ(d,"wait")) return -KEY_wait; | |
79072805 LW |
5803 | break; |
5804 | case 5: | |
5805 | if (strEQ(d,"while")) return KEY_while; | |
a0d0e21e | 5806 | if (strEQ(d,"write")) return -KEY_write; |
79072805 LW |
5807 | break; |
5808 | case 7: | |
a0d0e21e | 5809 | if (strEQ(d,"waitpid")) return -KEY_waitpid; |
79072805 LW |
5810 | break; |
5811 | case 9: | |
a0d0e21e | 5812 | if (strEQ(d,"wantarray")) return -KEY_wantarray; |
79072805 | 5813 | break; |
2f3197b3 | 5814 | } |
a687059c | 5815 | break; |
79072805 | 5816 | case 'x': |
a0d0e21e LW |
5817 | if (len == 1) return -KEY_x; |
5818 | if (strEQ(d,"xor")) return -KEY_xor; | |
a687059c | 5819 | break; |
79072805 LW |
5820 | case 'y': |
5821 | if (len == 1) return KEY_y; | |
5822 | break; | |
5823 | case 'z': | |
a687059c LW |
5824 | break; |
5825 | } | |
79072805 | 5826 | return 0; |
a687059c LW |
5827 | } |
5828 | ||
76e3520e | 5829 | STATIC void |
cea2e8a9 | 5830 | S_checkcomma(pTHX_ register char *s, char *name, char *what) |
a687059c | 5831 | { |
2f3197b3 LW |
5832 | char *w; |
5833 | ||
d008e5eb | 5834 | if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */ |
d008e5eb GS |
5835 | if (ckWARN(WARN_SYNTAX)) { |
5836 | int level = 1; | |
5837 | for (w = s+2; *w && level; w++) { | |
5838 | if (*w == '(') | |
5839 | ++level; | |
5840 | else if (*w == ')') | |
5841 | --level; | |
5842 | } | |
5843 | if (*w) | |
5844 | for (; *w && isSPACE(*w); w++) ; | |
5845 | if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */ | |
65cec589 GS |
5846 | Perl_warner(aTHX_ WARN_SYNTAX, |
5847 | "%s (...) interpreted as function",name); | |
d008e5eb | 5848 | } |
2f3197b3 | 5849 | } |
3280af22 | 5850 | while (s < PL_bufend && isSPACE(*s)) |
2f3197b3 | 5851 | s++; |
a687059c LW |
5852 | if (*s == '(') |
5853 | s++; | |
3280af22 | 5854 | while (s < PL_bufend && isSPACE(*s)) |
a687059c | 5855 | s++; |
7e2040f0 | 5856 | if (isIDFIRST_lazy_if(s,UTF)) { |
2f3197b3 | 5857 | w = s++; |
7e2040f0 | 5858 | while (isALNUM_lazy_if(s,UTF)) |
a687059c | 5859 | s++; |
3280af22 | 5860 | while (s < PL_bufend && isSPACE(*s)) |
a687059c | 5861 | s++; |
e929a76b | 5862 | if (*s == ',') { |
463ee0b2 | 5863 | int kw; |
e929a76b | 5864 | *s = '\0'; |
864dbfa3 | 5865 | kw = keyword(w, s - w) || get_cv(w, FALSE) != 0; |
e929a76b | 5866 | *s = ','; |
463ee0b2 | 5867 | if (kw) |
e929a76b | 5868 | return; |
cea2e8a9 | 5869 | Perl_croak(aTHX_ "No comma allowed after %s", what); |
463ee0b2 LW |
5870 | } |
5871 | } | |
5872 | } | |
5873 | ||
423cee85 JH |
5874 | /* Either returns sv, or mortalizes sv and returns a new SV*. |
5875 | Best used as sv=new_constant(..., sv, ...). | |
5876 | If s, pv are NULL, calls subroutine with one argument, | |
5877 | and type is used with error messages only. */ | |
5878 | ||
b3ac6de7 | 5879 | STATIC SV * |
dff6d3cd | 5880 | S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv, |
9b0e499b | 5881 | const char *type) |
b3ac6de7 | 5882 | { |
b3ac6de7 | 5883 | dSP; |
3280af22 | 5884 | HV *table = GvHV(PL_hintgv); /* ^H */ |
b3ac6de7 | 5885 | SV *res; |
b3ac6de7 IZ |
5886 | SV **cvp; |
5887 | SV *cv, *typesv; | |
f0af216f | 5888 | const char *why1, *why2, *why3; |
4e553d73 | 5889 | |
f0af216f | 5890 | if (!table || !(PL_hints & HINT_LOCALIZE_HH)) { |
423cee85 JH |
5891 | SV *msg; |
5892 | ||
f0af216f | 5893 | why2 = strEQ(key,"charnames") |
41ab332f | 5894 | ? "(possibly a missing \"use charnames ...\")" |
f0af216f | 5895 | : ""; |
4e553d73 | 5896 | msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s", |
41ab332f JH |
5897 | (type ? type: "undef"), why2); |
5898 | ||
5899 | /* This is convoluted and evil ("goto considered harmful") | |
5900 | * but I do not understand the intricacies of all the different | |
5901 | * failure modes of %^H in here. The goal here is to make | |
5902 | * the most probable error message user-friendly. --jhi */ | |
5903 | ||
5904 | goto msgdone; | |
5905 | ||
423cee85 | 5906 | report: |
4e553d73 | 5907 | msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s", |
f0af216f | 5908 | (type ? type: "undef"), why1, why2, why3); |
41ab332f | 5909 | msgdone: |
423cee85 JH |
5910 | yyerror(SvPVX(msg)); |
5911 | SvREFCNT_dec(msg); | |
5912 | return sv; | |
5913 | } | |
b3ac6de7 IZ |
5914 | cvp = hv_fetch(table, key, strlen(key), FALSE); |
5915 | if (!cvp || !SvOK(*cvp)) { | |
423cee85 JH |
5916 | why1 = "$^H{"; |
5917 | why2 = key; | |
f0af216f | 5918 | why3 = "} is not defined"; |
423cee85 | 5919 | goto report; |
b3ac6de7 IZ |
5920 | } |
5921 | sv_2mortal(sv); /* Parent created it permanently */ | |
5922 | cv = *cvp; | |
423cee85 JH |
5923 | if (!pv && s) |
5924 | pv = sv_2mortal(newSVpvn(s, len)); | |
5925 | if (type && pv) | |
5926 | typesv = sv_2mortal(newSVpv(type, 0)); | |
b3ac6de7 | 5927 | else |
423cee85 | 5928 | typesv = &PL_sv_undef; |
4e553d73 | 5929 | |
e788e7d3 | 5930 | PUSHSTACKi(PERLSI_OVERLOAD); |
423cee85 JH |
5931 | ENTER ; |
5932 | SAVETMPS; | |
4e553d73 | 5933 | |
423cee85 | 5934 | PUSHMARK(SP) ; |
a5845cb7 | 5935 | EXTEND(sp, 3); |
423cee85 JH |
5936 | if (pv) |
5937 | PUSHs(pv); | |
b3ac6de7 | 5938 | PUSHs(sv); |
423cee85 JH |
5939 | if (pv) |
5940 | PUSHs(typesv); | |
b3ac6de7 | 5941 | PUTBACK; |
423cee85 | 5942 | call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL)); |
4e553d73 | 5943 | |
423cee85 | 5944 | SPAGAIN ; |
4e553d73 | 5945 | |
423cee85 | 5946 | /* Check the eval first */ |
9b0e499b | 5947 | if (!PL_in_eval && SvTRUE(ERRSV)) { |
423cee85 JH |
5948 | STRLEN n_a; |
5949 | sv_catpv(ERRSV, "Propagated"); | |
5950 | yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */ | |
e1f15930 | 5951 | (void)POPs; |
423cee85 JH |
5952 | res = SvREFCNT_inc(sv); |
5953 | } | |
5954 | else { | |
5955 | res = POPs; | |
e1f15930 | 5956 | (void)SvREFCNT_inc(res); |
423cee85 | 5957 | } |
4e553d73 | 5958 | |
423cee85 JH |
5959 | PUTBACK ; |
5960 | FREETMPS ; | |
5961 | LEAVE ; | |
b3ac6de7 | 5962 | POPSTACK; |
4e553d73 | 5963 | |
b3ac6de7 | 5964 | if (!SvOK(res)) { |
423cee85 JH |
5965 | why1 = "Call to &{$^H{"; |
5966 | why2 = key; | |
f0af216f | 5967 | why3 = "}} did not return a defined value"; |
423cee85 JH |
5968 | sv = res; |
5969 | goto report; | |
9b0e499b | 5970 | } |
423cee85 | 5971 | |
9b0e499b | 5972 | return res; |
b3ac6de7 | 5973 | } |
4e553d73 | 5974 | |
76e3520e | 5975 | STATIC char * |
cea2e8a9 | 5976 | S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp) |
463ee0b2 LW |
5977 | { |
5978 | register char *d = dest; | |
8903cb82 | 5979 | register char *e = d + destlen - 3; /* two-character token, ending NUL */ |
463ee0b2 | 5980 | for (;;) { |
8903cb82 | 5981 | if (d >= e) |
cea2e8a9 | 5982 | Perl_croak(aTHX_ ident_too_long); |
834a4ddd | 5983 | if (isALNUM(*s)) /* UTF handled below */ |
463ee0b2 | 5984 | *d++ = *s++; |
7e2040f0 | 5985 | else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) { |
463ee0b2 LW |
5986 | *d++ = ':'; |
5987 | *d++ = ':'; | |
5988 | s++; | |
5989 | } | |
c3e0f903 | 5990 | else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') { |
463ee0b2 LW |
5991 | *d++ = *s++; |
5992 | *d++ = *s++; | |
5993 | } | |
fd400ab9 | 5994 | else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) { |
a0ed51b3 | 5995 | char *t = s + UTF8SKIP(s); |
fd400ab9 | 5996 | while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t)) |
a0ed51b3 LW |
5997 | t += UTF8SKIP(t); |
5998 | if (d + (t - s) > e) | |
cea2e8a9 | 5999 | Perl_croak(aTHX_ ident_too_long); |
a0ed51b3 LW |
6000 | Copy(s, d, t - s, char); |
6001 | d += t - s; | |
6002 | s = t; | |
6003 | } | |
463ee0b2 LW |
6004 | else { |
6005 | *d = '\0'; | |
6006 | *slp = d - dest; | |
6007 | return s; | |
e929a76b | 6008 | } |
378cc40b LW |
6009 | } |
6010 | } | |
6011 | ||
76e3520e | 6012 | STATIC char * |
cea2e8a9 | 6013 | S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni) |
378cc40b LW |
6014 | { |
6015 | register char *d; | |
8903cb82 | 6016 | register char *e; |
79072805 | 6017 | char *bracket = 0; |
748a9306 | 6018 | char funny = *s++; |
378cc40b | 6019 | |
a0d0e21e LW |
6020 | if (isSPACE(*s)) |
6021 | s = skipspace(s); | |
378cc40b | 6022 | d = dest; |
8903cb82 | 6023 | e = d + destlen - 3; /* two-character token, ending NUL */ |
de3bb511 | 6024 | if (isDIGIT(*s)) { |
8903cb82 | 6025 | while (isDIGIT(*s)) { |
6026 | if (d >= e) | |
cea2e8a9 | 6027 | Perl_croak(aTHX_ ident_too_long); |
378cc40b | 6028 | *d++ = *s++; |
8903cb82 | 6029 | } |
378cc40b LW |
6030 | } |
6031 | else { | |
463ee0b2 | 6032 | for (;;) { |
8903cb82 | 6033 | if (d >= e) |
cea2e8a9 | 6034 | Perl_croak(aTHX_ ident_too_long); |
834a4ddd | 6035 | if (isALNUM(*s)) /* UTF handled below */ |
463ee0b2 | 6036 | *d++ = *s++; |
7e2040f0 | 6037 | else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) { |
463ee0b2 LW |
6038 | *d++ = ':'; |
6039 | *d++ = ':'; | |
6040 | s++; | |
6041 | } | |
a0d0e21e | 6042 | else if (*s == ':' && s[1] == ':') { |
463ee0b2 LW |
6043 | *d++ = *s++; |
6044 | *d++ = *s++; | |
6045 | } | |
fd400ab9 | 6046 | else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) { |
a0ed51b3 | 6047 | char *t = s + UTF8SKIP(s); |
fd400ab9 | 6048 | while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t)) |
a0ed51b3 LW |
6049 | t += UTF8SKIP(t); |
6050 | if (d + (t - s) > e) | |
cea2e8a9 | 6051 | Perl_croak(aTHX_ ident_too_long); |
a0ed51b3 LW |
6052 | Copy(s, d, t - s, char); |
6053 | d += t - s; | |
6054 | s = t; | |
6055 | } | |
463ee0b2 LW |
6056 | else |
6057 | break; | |
6058 | } | |
378cc40b LW |
6059 | } |
6060 | *d = '\0'; | |
6061 | d = dest; | |
79072805 | 6062 | if (*d) { |
3280af22 NIS |
6063 | if (PL_lex_state != LEX_NORMAL) |
6064 | PL_lex_state = LEX_INTERPENDMAYBE; | |
79072805 | 6065 | return s; |
378cc40b | 6066 | } |
748a9306 | 6067 | if (*s == '$' && s[1] && |
7e2040f0 | 6068 | (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) ) |
5cd24f17 | 6069 | { |
4810e5ec | 6070 | return s; |
5cd24f17 | 6071 | } |
79072805 LW |
6072 | if (*s == '{') { |
6073 | bracket = s; | |
6074 | s++; | |
6075 | } | |
6076 | else if (ck_uni) | |
6077 | check_uni(); | |
93a17b20 | 6078 | if (s < send) |
79072805 LW |
6079 | *d = *s++; |
6080 | d[1] = '\0'; | |
2b92dfce | 6081 | if (*d == '^' && *s && isCONTROLVAR(*s)) { |
bbce6d69 | 6082 | *d = toCTRL(*s); |
6083 | s++; | |
de3bb511 | 6084 | } |
79072805 | 6085 | if (bracket) { |
748a9306 | 6086 | if (isSPACE(s[-1])) { |
fa83b5b6 | 6087 | while (s < send) { |
6088 | char ch = *s++; | |
bf4acbe4 | 6089 | if (!SPACE_OR_TAB(ch)) { |
fa83b5b6 | 6090 | *d = ch; |
6091 | break; | |
6092 | } | |
6093 | } | |
748a9306 | 6094 | } |
7e2040f0 | 6095 | if (isIDFIRST_lazy_if(d,UTF)) { |
79072805 | 6096 | d++; |
a0ed51b3 LW |
6097 | if (UTF) { |
6098 | e = s; | |
155aba94 | 6099 | while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') { |
a0ed51b3 | 6100 | e += UTF8SKIP(e); |
fd400ab9 | 6101 | while (e < send && UTF8_IS_CONTINUED(*e) && is_utf8_mark((U8*)e)) |
a0ed51b3 LW |
6102 | e += UTF8SKIP(e); |
6103 | } | |
6104 | Copy(s, d, e - s, char); | |
6105 | d += e - s; | |
6106 | s = e; | |
6107 | } | |
6108 | else { | |
2b92dfce | 6109 | while ((isALNUM(*s) || *s == ':') && d < e) |
a0ed51b3 | 6110 | *d++ = *s++; |
2b92dfce | 6111 | if (d >= e) |
cea2e8a9 | 6112 | Perl_croak(aTHX_ ident_too_long); |
a0ed51b3 | 6113 | } |
79072805 | 6114 | *d = '\0'; |
bf4acbe4 | 6115 | while (s < send && SPACE_OR_TAB(*s)) s++; |
ff68c719 | 6116 | if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) { |
599cee73 | 6117 | if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) { |
dff6d3cd | 6118 | const char *brack = *s == '[' ? "[...]" : "{...}"; |
cea2e8a9 | 6119 | Perl_warner(aTHX_ WARN_AMBIGUOUS, |
599cee73 | 6120 | "Ambiguous use of %c{%s%s} resolved to %c%s%s", |
748a9306 LW |
6121 | funny, dest, brack, funny, dest, brack); |
6122 | } | |
79072805 | 6123 | bracket++; |
a0be28da | 6124 | PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK); |
79072805 LW |
6125 | return s; |
6126 | } | |
4e553d73 NIS |
6127 | } |
6128 | /* Handle extended ${^Foo} variables | |
2b92dfce GS |
6129 | * 1999-02-27 mjd-perl-patch@plover.com */ |
6130 | else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */ | |
6131 | && isALNUM(*s)) | |
6132 | { | |
6133 | d++; | |
6134 | while (isALNUM(*s) && d < e) { | |
6135 | *d++ = *s++; | |
6136 | } | |
6137 | if (d >= e) | |
cea2e8a9 | 6138 | Perl_croak(aTHX_ ident_too_long); |
2b92dfce | 6139 | *d = '\0'; |
79072805 LW |
6140 | } |
6141 | if (*s == '}') { | |
6142 | s++; | |
3280af22 NIS |
6143 | if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets) |
6144 | PL_lex_state = LEX_INTERPEND; | |
748a9306 LW |
6145 | if (funny == '#') |
6146 | funny = '@'; | |
d008e5eb | 6147 | if (PL_lex_state == LEX_NORMAL) { |
d008e5eb | 6148 | if (ckWARN(WARN_AMBIGUOUS) && |
864dbfa3 | 6149 | (keyword(dest, d - dest) || get_cv(dest, FALSE))) |
d008e5eb | 6150 | { |
cea2e8a9 | 6151 | Perl_warner(aTHX_ WARN_AMBIGUOUS, |
d008e5eb GS |
6152 | "Ambiguous use of %c{%s} resolved to %c%s", |
6153 | funny, dest, funny, dest); | |
6154 | } | |
6155 | } | |
79072805 LW |
6156 | } |
6157 | else { | |
6158 | s = bracket; /* let the parser handle it */ | |
93a17b20 | 6159 | *dest = '\0'; |
79072805 LW |
6160 | } |
6161 | } | |
3280af22 NIS |
6162 | else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s)) |
6163 | PL_lex_state = LEX_INTERPEND; | |
378cc40b LW |
6164 | return s; |
6165 | } | |
6166 | ||
cea2e8a9 GS |
6167 | void |
6168 | Perl_pmflag(pTHX_ U16 *pmfl, int ch) | |
a0d0e21e | 6169 | { |
bbce6d69 | 6170 | if (ch == 'i') |
a0d0e21e | 6171 | *pmfl |= PMf_FOLD; |
a0d0e21e LW |
6172 | else if (ch == 'g') |
6173 | *pmfl |= PMf_GLOBAL; | |
c90c0ff4 | 6174 | else if (ch == 'c') |
6175 | *pmfl |= PMf_CONTINUE; | |
a0d0e21e LW |
6176 | else if (ch == 'o') |
6177 | *pmfl |= PMf_KEEP; | |
6178 | else if (ch == 'm') | |
6179 | *pmfl |= PMf_MULTILINE; | |
6180 | else if (ch == 's') | |
6181 | *pmfl |= PMf_SINGLELINE; | |
6182 | else if (ch == 'x') | |
6183 | *pmfl |= PMf_EXTENDED; | |
6184 | } | |
378cc40b | 6185 | |
76e3520e | 6186 | STATIC char * |
cea2e8a9 | 6187 | S_scan_pat(pTHX_ char *start, I32 type) |
378cc40b | 6188 | { |
79072805 LW |
6189 | PMOP *pm; |
6190 | char *s; | |
378cc40b | 6191 | |
09bef843 | 6192 | s = scan_str(start,FALSE,FALSE); |
79072805 | 6193 | if (!s) { |
3280af22 NIS |
6194 | if (PL_lex_stuff) |
6195 | SvREFCNT_dec(PL_lex_stuff); | |
6196 | PL_lex_stuff = Nullsv; | |
cea2e8a9 | 6197 | Perl_croak(aTHX_ "Search pattern not terminated"); |
378cc40b | 6198 | } |
bbce6d69 | 6199 | |
8782bef2 | 6200 | pm = (PMOP*)newPMOP(type, 0); |
3280af22 | 6201 | if (PL_multi_open == '?') |
79072805 | 6202 | pm->op_pmflags |= PMf_ONCE; |
8782bef2 GB |
6203 | if(type == OP_QR) { |
6204 | while (*s && strchr("iomsx", *s)) | |
6205 | pmflag(&pm->op_pmflags,*s++); | |
6206 | } | |
6207 | else { | |
6208 | while (*s && strchr("iogcmsx", *s)) | |
6209 | pmflag(&pm->op_pmflags,*s++); | |
6210 | } | |
4633a7c4 | 6211 | pm->op_pmpermflags = pm->op_pmflags; |
bbce6d69 | 6212 | |
3280af22 | 6213 | PL_lex_op = (OP*)pm; |
79072805 | 6214 | yylval.ival = OP_MATCH; |
378cc40b LW |
6215 | return s; |
6216 | } | |
6217 | ||
76e3520e | 6218 | STATIC char * |
cea2e8a9 | 6219 | S_scan_subst(pTHX_ char *start) |
79072805 | 6220 | { |
a0d0e21e | 6221 | register char *s; |
79072805 | 6222 | register PMOP *pm; |
4fdae800 | 6223 | I32 first_start; |
79072805 LW |
6224 | I32 es = 0; |
6225 | ||
79072805 LW |
6226 | yylval.ival = OP_NULL; |
6227 | ||
09bef843 | 6228 | s = scan_str(start,FALSE,FALSE); |
79072805 LW |
6229 | |
6230 | if (!s) { | |
3280af22 NIS |
6231 | if (PL_lex_stuff) |
6232 | SvREFCNT_dec(PL_lex_stuff); | |
6233 | PL_lex_stuff = Nullsv; | |
cea2e8a9 | 6234 | Perl_croak(aTHX_ "Substitution pattern not terminated"); |
a687059c | 6235 | } |
79072805 | 6236 | |
3280af22 | 6237 | if (s[-1] == PL_multi_open) |
79072805 LW |
6238 | s--; |
6239 | ||
3280af22 | 6240 | first_start = PL_multi_start; |
09bef843 | 6241 | s = scan_str(s,FALSE,FALSE); |
79072805 | 6242 | if (!s) { |
3280af22 NIS |
6243 | if (PL_lex_stuff) |
6244 | SvREFCNT_dec(PL_lex_stuff); | |
6245 | PL_lex_stuff = Nullsv; | |
6246 | if (PL_lex_repl) | |
6247 | SvREFCNT_dec(PL_lex_repl); | |
6248 | PL_lex_repl = Nullsv; | |
cea2e8a9 | 6249 | Perl_croak(aTHX_ "Substitution replacement not terminated"); |
a687059c | 6250 | } |
3280af22 | 6251 | PL_multi_start = first_start; /* so whole substitution is taken together */ |
2f3197b3 | 6252 | |
79072805 | 6253 | pm = (PMOP*)newPMOP(OP_SUBST, 0); |
48c036b1 | 6254 | while (*s) { |
a687059c LW |
6255 | if (*s == 'e') { |
6256 | s++; | |
2f3197b3 | 6257 | es++; |
a687059c | 6258 | } |
b3eb6a9b | 6259 | else if (strchr("iogcmsx", *s)) |
a0d0e21e | 6260 | pmflag(&pm->op_pmflags,*s++); |
48c036b1 GS |
6261 | else |
6262 | break; | |
378cc40b | 6263 | } |
79072805 LW |
6264 | |
6265 | if (es) { | |
6266 | SV *repl; | |
0244c3a4 GS |
6267 | PL_sublex_info.super_bufptr = s; |
6268 | PL_sublex_info.super_bufend = PL_bufend; | |
6269 | PL_multi_end = 0; | |
79072805 | 6270 | pm->op_pmflags |= PMf_EVAL; |
79cb57f6 | 6271 | repl = newSVpvn("",0); |
463ee0b2 | 6272 | while (es-- > 0) |
a0d0e21e | 6273 | sv_catpv(repl, es ? "eval " : "do "); |
79072805 | 6274 | sv_catpvn(repl, "{ ", 2); |
3280af22 | 6275 | sv_catsv(repl, PL_lex_repl); |
79072805 | 6276 | sv_catpvn(repl, " };", 2); |
25da4f38 | 6277 | SvEVALED_on(repl); |
3280af22 NIS |
6278 | SvREFCNT_dec(PL_lex_repl); |
6279 | PL_lex_repl = repl; | |
378cc40b | 6280 | } |
79072805 | 6281 | |
4633a7c4 | 6282 | pm->op_pmpermflags = pm->op_pmflags; |
3280af22 | 6283 | PL_lex_op = (OP*)pm; |
79072805 | 6284 | yylval.ival = OP_SUBST; |
378cc40b LW |
6285 | return s; |
6286 | } | |
6287 | ||
76e3520e | 6288 | STATIC char * |
cea2e8a9 | 6289 | S_scan_trans(pTHX_ char *start) |
378cc40b | 6290 | { |
a0d0e21e | 6291 | register char* s; |
11343788 | 6292 | OP *o; |
79072805 LW |
6293 | short *tbl; |
6294 | I32 squash; | |
a0ed51b3 | 6295 | I32 del; |
79072805 | 6296 | I32 complement; |
a0ed51b3 LW |
6297 | I32 utf8; |
6298 | I32 count = 0; | |
79072805 LW |
6299 | |
6300 | yylval.ival = OP_NULL; | |
6301 | ||
09bef843 | 6302 | s = scan_str(start,FALSE,FALSE); |
79072805 | 6303 | if (!s) { |
3280af22 NIS |
6304 | if (PL_lex_stuff) |
6305 | SvREFCNT_dec(PL_lex_stuff); | |
6306 | PL_lex_stuff = Nullsv; | |
cea2e8a9 | 6307 | Perl_croak(aTHX_ "Transliteration pattern not terminated"); |
a687059c | 6308 | } |
3280af22 | 6309 | if (s[-1] == PL_multi_open) |
2f3197b3 LW |
6310 | s--; |
6311 | ||
09bef843 | 6312 | s = scan_str(s,FALSE,FALSE); |
79072805 | 6313 | if (!s) { |
3280af22 NIS |
6314 | if (PL_lex_stuff) |
6315 | SvREFCNT_dec(PL_lex_stuff); | |
6316 | PL_lex_stuff = Nullsv; | |
6317 | if (PL_lex_repl) | |
6318 | SvREFCNT_dec(PL_lex_repl); | |
6319 | PL_lex_repl = Nullsv; | |
cea2e8a9 | 6320 | Perl_croak(aTHX_ "Transliteration replacement not terminated"); |
a687059c | 6321 | } |
79072805 | 6322 | |
01ec43d0 GS |
6323 | New(803,tbl,256,short); |
6324 | o = newPVOP(OP_TRANS, 0, (char*)tbl); | |
2f3197b3 | 6325 | |
a0ed51b3 | 6326 | complement = del = squash = 0; |
6940069f | 6327 | while (strchr("cds", *s)) { |
395c3793 | 6328 | if (*s == 'c') |
79072805 | 6329 | complement = OPpTRANS_COMPLEMENT; |
395c3793 | 6330 | else if (*s == 'd') |
a0ed51b3 LW |
6331 | del = OPpTRANS_DELETE; |
6332 | else if (*s == 's') | |
79072805 | 6333 | squash = OPpTRANS_SQUASH; |
395c3793 LW |
6334 | s++; |
6335 | } | |
7948272d NIS |
6336 | o->op_private = del|squash|complement| |
6337 | (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)| | |
6338 | (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0); | |
79072805 | 6339 | |
3280af22 | 6340 | PL_lex_op = o; |
79072805 LW |
6341 | yylval.ival = OP_TRANS; |
6342 | return s; | |
6343 | } | |
6344 | ||
76e3520e | 6345 | STATIC char * |
cea2e8a9 | 6346 | S_scan_heredoc(pTHX_ register char *s) |
79072805 LW |
6347 | { |
6348 | SV *herewas; | |
6349 | I32 op_type = OP_SCALAR; | |
6350 | I32 len; | |
6351 | SV *tmpstr; | |
6352 | char term; | |
6353 | register char *d; | |
fc36a67e | 6354 | register char *e; |
4633a7c4 | 6355 | char *peek; |
3280af22 | 6356 | int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR)); |
79072805 LW |
6357 | |
6358 | s += 2; | |
3280af22 NIS |
6359 | d = PL_tokenbuf; |
6360 | e = PL_tokenbuf + sizeof PL_tokenbuf - 1; | |
fd2d0953 | 6361 | if (!outer) |
79072805 | 6362 | *d++ = '\n'; |
bf4acbe4 | 6363 | for (peek = s; SPACE_OR_TAB(*peek); peek++) ; |
4633a7c4 LW |
6364 | if (*peek && strchr("`'\"",*peek)) { |
6365 | s = peek; | |
79072805 | 6366 | term = *s++; |
3280af22 | 6367 | s = delimcpy(d, e, s, PL_bufend, term, &len); |
fc36a67e | 6368 | d += len; |
3280af22 | 6369 | if (s < PL_bufend) |
79072805 | 6370 | s++; |
79072805 LW |
6371 | } |
6372 | else { | |
6373 | if (*s == '\\') | |
6374 | s++, term = '\''; | |
6375 | else | |
6376 | term = '"'; | |
7e2040f0 | 6377 | if (!isALNUM_lazy_if(s,UTF)) |
4633a7c4 | 6378 | deprecate("bare << to mean <<\"\""); |
7e2040f0 | 6379 | for (; isALNUM_lazy_if(s,UTF); s++) { |
fc36a67e | 6380 | if (d < e) |
6381 | *d++ = *s; | |
6382 | } | |
6383 | } | |
3280af22 | 6384 | if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1) |
cea2e8a9 | 6385 | Perl_croak(aTHX_ "Delimiter for here document is too long"); |
79072805 LW |
6386 | *d++ = '\n'; |
6387 | *d = '\0'; | |
3280af22 | 6388 | len = d - PL_tokenbuf; |
6a27c188 | 6389 | #ifndef PERL_STRICT_CR |
f63a84b2 LW |
6390 | d = strchr(s, '\r'); |
6391 | if (d) { | |
6392 | char *olds = s; | |
6393 | s = d; | |
3280af22 | 6394 | while (s < PL_bufend) { |
f63a84b2 LW |
6395 | if (*s == '\r') { |
6396 | *d++ = '\n'; | |
6397 | if (*++s == '\n') | |
6398 | s++; | |
6399 | } | |
6400 | else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */ | |
6401 | *d++ = *s++; | |
6402 | s++; | |
6403 | } | |
6404 | else | |
6405 | *d++ = *s++; | |
6406 | } | |
6407 | *d = '\0'; | |
3280af22 NIS |
6408 | PL_bufend = d; |
6409 | SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr)); | |
f63a84b2 LW |
6410 | s = olds; |
6411 | } | |
6412 | #endif | |
79072805 | 6413 | d = "\n"; |
3280af22 | 6414 | if (outer || !(d=ninstr(s,PL_bufend,d,d+1))) |
79cb57f6 | 6415 | herewas = newSVpvn(s,PL_bufend-s); |
79072805 | 6416 | else |
79cb57f6 | 6417 | s--, herewas = newSVpvn(s,d-s); |
79072805 | 6418 | s += SvCUR(herewas); |
748a9306 | 6419 | |
8d6dde3e | 6420 | tmpstr = NEWSV(87,79); |
748a9306 LW |
6421 | sv_upgrade(tmpstr, SVt_PVIV); |
6422 | if (term == '\'') { | |
79072805 | 6423 | op_type = OP_CONST; |
748a9306 LW |
6424 | SvIVX(tmpstr) = -1; |
6425 | } | |
6426 | else if (term == '`') { | |
79072805 | 6427 | op_type = OP_BACKTICK; |
748a9306 LW |
6428 | SvIVX(tmpstr) = '\\'; |
6429 | } | |
79072805 LW |
6430 | |
6431 | CLINE; | |
57843af0 | 6432 | PL_multi_start = CopLINE(PL_curcop); |
3280af22 NIS |
6433 | PL_multi_open = PL_multi_close = '<'; |
6434 | term = *PL_tokenbuf; | |
0244c3a4 GS |
6435 | if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) { |
6436 | char *bufptr = PL_sublex_info.super_bufptr; | |
6437 | char *bufend = PL_sublex_info.super_bufend; | |
6438 | char *olds = s - SvCUR(herewas); | |
6439 | s = strchr(bufptr, '\n'); | |
6440 | if (!s) | |
6441 | s = bufend; | |
6442 | d = s; | |
6443 | while (s < bufend && | |
6444 | (*s != term || memNE(s,PL_tokenbuf,len)) ) { | |
6445 | if (*s++ == '\n') | |
57843af0 | 6446 | CopLINE_inc(PL_curcop); |
0244c3a4 GS |
6447 | } |
6448 | if (s >= bufend) { | |
57843af0 | 6449 | CopLINE_set(PL_curcop, PL_multi_start); |
0244c3a4 GS |
6450 | missingterm(PL_tokenbuf); |
6451 | } | |
6452 | sv_setpvn(herewas,bufptr,d-bufptr+1); | |
6453 | sv_setpvn(tmpstr,d+1,s-d); | |
6454 | s += len - 1; | |
6455 | sv_catpvn(herewas,s,bufend-s); | |
6456 | (void)strcpy(bufptr,SvPVX(herewas)); | |
6457 | ||
6458 | s = olds; | |
6459 | goto retval; | |
6460 | } | |
6461 | else if (!outer) { | |
79072805 | 6462 | d = s; |
3280af22 NIS |
6463 | while (s < PL_bufend && |
6464 | (*s != term || memNE(s,PL_tokenbuf,len)) ) { | |
79072805 | 6465 | if (*s++ == '\n') |
57843af0 | 6466 | CopLINE_inc(PL_curcop); |
79072805 | 6467 | } |
3280af22 | 6468 | if (s >= PL_bufend) { |
57843af0 | 6469 | CopLINE_set(PL_curcop, PL_multi_start); |
3280af22 | 6470 | missingterm(PL_tokenbuf); |
79072805 LW |
6471 | } |
6472 | sv_setpvn(tmpstr,d+1,s-d); | |
6473 | s += len - 1; | |
57843af0 | 6474 | CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */ |
49d8d3a1 | 6475 | |
3280af22 NIS |
6476 | sv_catpvn(herewas,s,PL_bufend-s); |
6477 | sv_setsv(PL_linestr,herewas); | |
6478 | PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr); | |
6479 | PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); | |
79072805 LW |
6480 | } |
6481 | else | |
6482 | sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */ | |
3280af22 | 6483 | while (s >= PL_bufend) { /* multiple line string? */ |
fd2d0953 | 6484 | if (!outer || |
3280af22 | 6485 | !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) { |
57843af0 | 6486 | CopLINE_set(PL_curcop, PL_multi_start); |
3280af22 | 6487 | missingterm(PL_tokenbuf); |
79072805 | 6488 | } |
57843af0 | 6489 | CopLINE_inc(PL_curcop); |
3280af22 | 6490 | PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); |
6a27c188 | 6491 | #ifndef PERL_STRICT_CR |
3280af22 | 6492 | if (PL_bufend - PL_linestart >= 2) { |
a1529941 NIS |
6493 | if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') || |
6494 | (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r')) | |
c6f14548 | 6495 | { |
3280af22 NIS |
6496 | PL_bufend[-2] = '\n'; |
6497 | PL_bufend--; | |
6498 | SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr)); | |
f63a84b2 | 6499 | } |
3280af22 NIS |
6500 | else if (PL_bufend[-1] == '\r') |
6501 | PL_bufend[-1] = '\n'; | |
f63a84b2 | 6502 | } |
3280af22 NIS |
6503 | else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r') |
6504 | PL_bufend[-1] = '\n'; | |
f63a84b2 | 6505 | #endif |
3280af22 | 6506 | if (PERLDB_LINE && PL_curstash != PL_debstash) { |
79072805 LW |
6507 | SV *sv = NEWSV(88,0); |
6508 | ||
93a17b20 | 6509 | sv_upgrade(sv, SVt_PVMG); |
3280af22 | 6510 | sv_setsv(sv,PL_linestr); |
57843af0 | 6511 | av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv); |
79072805 | 6512 | } |
3280af22 NIS |
6513 | if (*s == term && memEQ(s,PL_tokenbuf,len)) { |
6514 | s = PL_bufend - 1; | |
79072805 | 6515 | *s = ' '; |
3280af22 NIS |
6516 | sv_catsv(PL_linestr,herewas); |
6517 | PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); | |
79072805 LW |
6518 | } |
6519 | else { | |
3280af22 NIS |
6520 | s = PL_bufend; |
6521 | sv_catsv(tmpstr,PL_linestr); | |
395c3793 LW |
6522 | } |
6523 | } | |
79072805 | 6524 | s++; |
0244c3a4 | 6525 | retval: |
57843af0 | 6526 | PL_multi_end = CopLINE(PL_curcop); |
79072805 LW |
6527 | if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) { |
6528 | SvLEN_set(tmpstr, SvCUR(tmpstr) + 1); | |
463ee0b2 | 6529 | Renew(SvPVX(tmpstr), SvLEN(tmpstr), char); |
79072805 | 6530 | } |
8990e307 | 6531 | SvREFCNT_dec(herewas); |
7948272d NIS |
6532 | if (UTF && !IN_BYTE && is_utf8_string((U8*)SvPVX(tmpstr), SvCUR(tmpstr))) |
6533 | SvUTF8_on(tmpstr); | |
3280af22 | 6534 | PL_lex_stuff = tmpstr; |
79072805 LW |
6535 | yylval.ival = op_type; |
6536 | return s; | |
6537 | } | |
6538 | ||
02aa26ce NT |
6539 | /* scan_inputsymbol |
6540 | takes: current position in input buffer | |
6541 | returns: new position in input buffer | |
6542 | side-effects: yylval and lex_op are set. | |
6543 | ||
6544 | This code handles: | |
6545 | ||
6546 | <> read from ARGV | |
6547 | <FH> read from filehandle | |
6548 | <pkg::FH> read from package qualified filehandle | |
6549 | <pkg'FH> read from package qualified filehandle | |
6550 | <$fh> read from filehandle in $fh | |
6551 | <*.h> filename glob | |
6552 | ||
6553 | */ | |
6554 | ||
76e3520e | 6555 | STATIC char * |
cea2e8a9 | 6556 | S_scan_inputsymbol(pTHX_ char *start) |
79072805 | 6557 | { |
02aa26ce | 6558 | register char *s = start; /* current position in buffer */ |
79072805 | 6559 | register char *d; |
fc36a67e | 6560 | register char *e; |
1b420867 | 6561 | char *end; |
79072805 LW |
6562 | I32 len; |
6563 | ||
3280af22 NIS |
6564 | d = PL_tokenbuf; /* start of temp holding space */ |
6565 | e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */ | |
1b420867 GS |
6566 | end = strchr(s, '\n'); |
6567 | if (!end) | |
6568 | end = PL_bufend; | |
6569 | s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */ | |
02aa26ce NT |
6570 | |
6571 | /* die if we didn't have space for the contents of the <>, | |
1b420867 | 6572 | or if it didn't end, or if we see a newline |
02aa26ce NT |
6573 | */ |
6574 | ||
3280af22 | 6575 | if (len >= sizeof PL_tokenbuf) |
cea2e8a9 | 6576 | Perl_croak(aTHX_ "Excessively long <> operator"); |
1b420867 | 6577 | if (s >= end) |
cea2e8a9 | 6578 | Perl_croak(aTHX_ "Unterminated <> operator"); |
02aa26ce | 6579 | |
fc36a67e | 6580 | s++; |
02aa26ce NT |
6581 | |
6582 | /* check for <$fh> | |
6583 | Remember, only scalar variables are interpreted as filehandles by | |
6584 | this code. Anything more complex (e.g., <$fh{$num}>) will be | |
6585 | treated as a glob() call. | |
6586 | This code makes use of the fact that except for the $ at the front, | |
6587 | a scalar variable and a filehandle look the same. | |
6588 | */ | |
4633a7c4 | 6589 | if (*d == '$' && d[1]) d++; |
02aa26ce NT |
6590 | |
6591 | /* allow <Pkg'VALUE> or <Pkg::VALUE> */ | |
7e2040f0 | 6592 | while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':')) |
79072805 | 6593 | d++; |
02aa26ce NT |
6594 | |
6595 | /* If we've tried to read what we allow filehandles to look like, and | |
6596 | there's still text left, then it must be a glob() and not a getline. | |
6597 | Use scan_str to pull out the stuff between the <> and treat it | |
6598 | as nothing more than a string. | |
6599 | */ | |
6600 | ||
3280af22 | 6601 | if (d - PL_tokenbuf != len) { |
79072805 LW |
6602 | yylval.ival = OP_GLOB; |
6603 | set_csh(); | |
09bef843 | 6604 | s = scan_str(start,FALSE,FALSE); |
79072805 | 6605 | if (!s) |
cea2e8a9 | 6606 | Perl_croak(aTHX_ "Glob not terminated"); |
79072805 LW |
6607 | return s; |
6608 | } | |
395c3793 | 6609 | else { |
02aa26ce | 6610 | /* we're in a filehandle read situation */ |
3280af22 | 6611 | d = PL_tokenbuf; |
02aa26ce NT |
6612 | |
6613 | /* turn <> into <ARGV> */ | |
79072805 LW |
6614 | if (!len) |
6615 | (void)strcpy(d,"ARGV"); | |
02aa26ce NT |
6616 | |
6617 | /* if <$fh>, create the ops to turn the variable into a | |
6618 | filehandle | |
6619 | */ | |
79072805 | 6620 | if (*d == '$') { |
a0d0e21e | 6621 | I32 tmp; |
02aa26ce NT |
6622 | |
6623 | /* try to find it in the pad for this block, otherwise find | |
6624 | add symbol table ops | |
6625 | */ | |
11343788 MB |
6626 | if ((tmp = pad_findmy(d)) != NOT_IN_PAD) { |
6627 | OP *o = newOP(OP_PADSV, 0); | |
6628 | o->op_targ = tmp; | |
f5284f61 | 6629 | PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o); |
a0d0e21e LW |
6630 | } |
6631 | else { | |
6632 | GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV); | |
3280af22 | 6633 | PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, |
a0d0e21e | 6634 | newUNOP(OP_RV2SV, 0, |
f5284f61 | 6635 | newGVOP(OP_GV, 0, gv))); |
a0d0e21e | 6636 | } |
f5284f61 IZ |
6637 | PL_lex_op->op_flags |= OPf_SPECIAL; |
6638 | /* we created the ops in PL_lex_op, so make yylval.ival a null op */ | |
79072805 LW |
6639 | yylval.ival = OP_NULL; |
6640 | } | |
02aa26ce NT |
6641 | |
6642 | /* If it's none of the above, it must be a literal filehandle | |
6643 | (<Foo::BAR> or <FOO>) so build a simple readline OP */ | |
79072805 | 6644 | else { |
85e6fe83 | 6645 | GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO); |
3280af22 | 6646 | PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv)); |
79072805 LW |
6647 | yylval.ival = OP_NULL; |
6648 | } | |
6649 | } | |
02aa26ce | 6650 | |
79072805 LW |
6651 | return s; |
6652 | } | |
6653 | ||
02aa26ce NT |
6654 | |
6655 | /* scan_str | |
6656 | takes: start position in buffer | |
09bef843 SB |
6657 | keep_quoted preserve \ on the embedded delimiter(s) |
6658 | keep_delims preserve the delimiters around the string | |
02aa26ce NT |
6659 | returns: position to continue reading from buffer |
6660 | side-effects: multi_start, multi_close, lex_repl or lex_stuff, and | |
6661 | updates the read buffer. | |
6662 | ||
6663 | This subroutine pulls a string out of the input. It is called for: | |
6664 | q single quotes q(literal text) | |
6665 | ' single quotes 'literal text' | |
6666 | qq double quotes qq(interpolate $here please) | |
6667 | " double quotes "interpolate $here please" | |
6668 | qx backticks qx(/bin/ls -l) | |
6669 | ` backticks `/bin/ls -l` | |
6670 | qw quote words @EXPORT_OK = qw( func() $spam ) | |
6671 | m// regexp match m/this/ | |
6672 | s/// regexp substitute s/this/that/ | |
6673 | tr/// string transliterate tr/this/that/ | |
6674 | y/// string transliterate y/this/that/ | |
6675 | ($*@) sub prototypes sub foo ($) | |
09bef843 | 6676 | (stuff) sub attr parameters sub foo : attr(stuff) |
02aa26ce NT |
6677 | <> readline or globs <FOO>, <>, <$fh>, or <*.c> |
6678 | ||
6679 | In most of these cases (all but <>, patterns and transliterate) | |
6680 | yylex() calls scan_str(). m// makes yylex() call scan_pat() which | |
6681 | calls scan_str(). s/// makes yylex() call scan_subst() which calls | |
6682 | scan_str(). tr/// and y/// make yylex() call scan_trans() which | |
6683 | calls scan_str(). | |
4e553d73 | 6684 | |
02aa26ce NT |
6685 | It skips whitespace before the string starts, and treats the first |
6686 | character as the delimiter. If the delimiter is one of ([{< then | |
6687 | the corresponding "close" character )]}> is used as the closing | |
6688 | delimiter. It allows quoting of delimiters, and if the string has | |
6689 | balanced delimiters ([{<>}]) it allows nesting. | |
6690 | ||
6691 | The lexer always reads these strings into lex_stuff, except in the | |
6692 | case of the operators which take *two* arguments (s/// and tr///) | |
6693 | when it checks to see if lex_stuff is full (presumably with the 1st | |
6694 | arg to s or tr) and if so puts the string into lex_repl. | |
6695 | ||
6696 | */ | |
6697 | ||
76e3520e | 6698 | STATIC char * |
09bef843 | 6699 | S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims) |
79072805 | 6700 | { |
02aa26ce NT |
6701 | SV *sv; /* scalar value: string */ |
6702 | char *tmps; /* temp string, used for delimiter matching */ | |
6703 | register char *s = start; /* current position in the buffer */ | |
6704 | register char term; /* terminating character */ | |
6705 | register char *to; /* current position in the sv's data */ | |
6706 | I32 brackets = 1; /* bracket nesting level */ | |
89491803 | 6707 | bool has_utf8 = FALSE; /* is there any utf8 content? */ |
02aa26ce NT |
6708 | |
6709 | /* skip space before the delimiter */ | |
fb73857a | 6710 | if (isSPACE(*s)) |
6711 | s = skipspace(s); | |
02aa26ce NT |
6712 | |
6713 | /* mark where we are, in case we need to report errors */ | |
79072805 | 6714 | CLINE; |
02aa26ce NT |
6715 | |
6716 | /* after skipping whitespace, the next character is the terminator */ | |
a0d0e21e | 6717 | term = *s; |
fd400ab9 | 6718 | if (UTF8_IS_CONTINUED(term) && UTF) |
89491803 | 6719 | has_utf8 = TRUE; |
b1c7b182 | 6720 | |
02aa26ce | 6721 | /* mark where we are */ |
57843af0 | 6722 | PL_multi_start = CopLINE(PL_curcop); |
3280af22 | 6723 | PL_multi_open = term; |
02aa26ce NT |
6724 | |
6725 | /* find corresponding closing delimiter */ | |
93a17b20 | 6726 | if (term && (tmps = strchr("([{< )]}> )]}>",term))) |
79072805 | 6727 | term = tmps[5]; |
3280af22 | 6728 | PL_multi_close = term; |
79072805 | 6729 | |
02aa26ce | 6730 | /* create a new SV to hold the contents. 87 is leak category, I'm |
8d6dde3e IZ |
6731 | assuming. 79 is the SV's initial length. What a random number. */ |
6732 | sv = NEWSV(87,79); | |
ed6116ce LW |
6733 | sv_upgrade(sv, SVt_PVIV); |
6734 | SvIVX(sv) = term; | |
a0d0e21e | 6735 | (void)SvPOK_only(sv); /* validate pointer */ |
02aa26ce NT |
6736 | |
6737 | /* move past delimiter and try to read a complete string */ | |
09bef843 SB |
6738 | if (keep_delims) |
6739 | sv_catpvn(sv, s, 1); | |
93a17b20 LW |
6740 | s++; |
6741 | for (;;) { | |
02aa26ce | 6742 | /* extend sv if need be */ |
3280af22 | 6743 | SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1); |
02aa26ce | 6744 | /* set 'to' to the next character in the sv's string */ |
463ee0b2 | 6745 | to = SvPVX(sv)+SvCUR(sv); |
09bef843 | 6746 | |
02aa26ce | 6747 | /* if open delimiter is the close delimiter read unbridle */ |
3280af22 NIS |
6748 | if (PL_multi_open == PL_multi_close) { |
6749 | for (; s < PL_bufend; s++,to++) { | |
02aa26ce | 6750 | /* embedded newlines increment the current line number */ |
3280af22 | 6751 | if (*s == '\n' && !PL_rsfp) |
57843af0 | 6752 | CopLINE_inc(PL_curcop); |
02aa26ce | 6753 | /* handle quoted delimiters */ |
3280af22 | 6754 | if (*s == '\\' && s+1 < PL_bufend && term != '\\') { |
09bef843 | 6755 | if (!keep_quoted && s[1] == term) |
a0d0e21e | 6756 | s++; |
02aa26ce | 6757 | /* any other quotes are simply copied straight through */ |
a0d0e21e LW |
6758 | else |
6759 | *to++ = *s++; | |
6760 | } | |
02aa26ce NT |
6761 | /* terminate when run out of buffer (the for() condition), or |
6762 | have found the terminator */ | |
93a17b20 LW |
6763 | else if (*s == term) |
6764 | break; | |
fd400ab9 | 6765 | else if (!has_utf8 && UTF8_IS_CONTINUED(*s) && UTF) |
89491803 | 6766 | has_utf8 = TRUE; |
93a17b20 LW |
6767 | *to = *s; |
6768 | } | |
6769 | } | |
02aa26ce NT |
6770 | |
6771 | /* if the terminator isn't the same as the start character (e.g., | |
6772 | matched brackets), we have to allow more in the quoting, and | |
6773 | be prepared for nested brackets. | |
6774 | */ | |
93a17b20 | 6775 | else { |
02aa26ce | 6776 | /* read until we run out of string, or we find the terminator */ |
3280af22 | 6777 | for (; s < PL_bufend; s++,to++) { |
02aa26ce | 6778 | /* embedded newlines increment the line count */ |
3280af22 | 6779 | if (*s == '\n' && !PL_rsfp) |
57843af0 | 6780 | CopLINE_inc(PL_curcop); |
02aa26ce | 6781 | /* backslashes can escape the open or closing characters */ |
3280af22 | 6782 | if (*s == '\\' && s+1 < PL_bufend) { |
09bef843 SB |
6783 | if (!keep_quoted && |
6784 | ((s[1] == PL_multi_open) || (s[1] == PL_multi_close))) | |
a0d0e21e LW |
6785 | s++; |
6786 | else | |
6787 | *to++ = *s++; | |
6788 | } | |
02aa26ce | 6789 | /* allow nested opens and closes */ |
3280af22 | 6790 | else if (*s == PL_multi_close && --brackets <= 0) |
93a17b20 | 6791 | break; |
3280af22 | 6792 | else if (*s == PL_multi_open) |
93a17b20 | 6793 | brackets++; |
fd400ab9 | 6794 | else if (!has_utf8 && UTF8_IS_CONTINUED(*s) && UTF) |
89491803 | 6795 | has_utf8 = TRUE; |
93a17b20 LW |
6796 | *to = *s; |
6797 | } | |
6798 | } | |
02aa26ce | 6799 | /* terminate the copied string and update the sv's end-of-string */ |
93a17b20 | 6800 | *to = '\0'; |
463ee0b2 | 6801 | SvCUR_set(sv, to - SvPVX(sv)); |
93a17b20 | 6802 | |
02aa26ce NT |
6803 | /* |
6804 | * this next chunk reads more into the buffer if we're not done yet | |
6805 | */ | |
6806 | ||
b1c7b182 GS |
6807 | if (s < PL_bufend) |
6808 | break; /* handle case where we are done yet :-) */ | |
79072805 | 6809 | |
6a27c188 | 6810 | #ifndef PERL_STRICT_CR |
f63a84b2 | 6811 | if (to - SvPVX(sv) >= 2) { |
c6f14548 GS |
6812 | if ((to[-2] == '\r' && to[-1] == '\n') || |
6813 | (to[-2] == '\n' && to[-1] == '\r')) | |
6814 | { | |
f63a84b2 LW |
6815 | to[-2] = '\n'; |
6816 | to--; | |
6817 | SvCUR_set(sv, to - SvPVX(sv)); | |
6818 | } | |
6819 | else if (to[-1] == '\r') | |
6820 | to[-1] = '\n'; | |
6821 | } | |
6822 | else if (to - SvPVX(sv) == 1 && to[-1] == '\r') | |
6823 | to[-1] = '\n'; | |
6824 | #endif | |
6825 | ||
02aa26ce NT |
6826 | /* if we're out of file, or a read fails, bail and reset the current |
6827 | line marker so we can report where the unterminated string began | |
6828 | */ | |
3280af22 NIS |
6829 | if (!PL_rsfp || |
6830 | !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) { | |
c07a80fd | 6831 | sv_free(sv); |
57843af0 | 6832 | CopLINE_set(PL_curcop, PL_multi_start); |
79072805 LW |
6833 | return Nullch; |
6834 | } | |
02aa26ce | 6835 | /* we read a line, so increment our line counter */ |
57843af0 | 6836 | CopLINE_inc(PL_curcop); |
a0ed51b3 | 6837 | |
02aa26ce | 6838 | /* update debugger info */ |
3280af22 | 6839 | if (PERLDB_LINE && PL_curstash != PL_debstash) { |
79072805 LW |
6840 | SV *sv = NEWSV(88,0); |
6841 | ||
93a17b20 | 6842 | sv_upgrade(sv, SVt_PVMG); |
3280af22 | 6843 | sv_setsv(sv,PL_linestr); |
57843af0 | 6844 | av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv); |
395c3793 | 6845 | } |
a0ed51b3 | 6846 | |
3280af22 NIS |
6847 | /* having changed the buffer, we must update PL_bufend */ |
6848 | PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); | |
378cc40b | 6849 | } |
4e553d73 | 6850 | |
02aa26ce NT |
6851 | /* at this point, we have successfully read the delimited string */ |
6852 | ||
09bef843 SB |
6853 | if (keep_delims) |
6854 | sv_catpvn(sv, s, 1); | |
89491803 | 6855 | if (has_utf8) |
b1c7b182 | 6856 | SvUTF8_on(sv); |
57843af0 | 6857 | PL_multi_end = CopLINE(PL_curcop); |
79072805 | 6858 | s++; |
02aa26ce NT |
6859 | |
6860 | /* if we allocated too much space, give some back */ | |
93a17b20 LW |
6861 | if (SvCUR(sv) + 5 < SvLEN(sv)) { |
6862 | SvLEN_set(sv, SvCUR(sv) + 1); | |
463ee0b2 | 6863 | Renew(SvPVX(sv), SvLEN(sv), char); |
79072805 | 6864 | } |
02aa26ce NT |
6865 | |
6866 | /* decide whether this is the first or second quoted string we've read | |
6867 | for this op | |
6868 | */ | |
4e553d73 | 6869 | |
3280af22 NIS |
6870 | if (PL_lex_stuff) |
6871 | PL_lex_repl = sv; | |
79072805 | 6872 | else |
3280af22 | 6873 | PL_lex_stuff = sv; |
378cc40b LW |
6874 | return s; |
6875 | } | |
6876 | ||
02aa26ce NT |
6877 | /* |
6878 | scan_num | |
6879 | takes: pointer to position in buffer | |
6880 | returns: pointer to new position in buffer | |
6881 | side-effects: builds ops for the constant in yylval.op | |
6882 | ||
6883 | Read a number in any of the formats that Perl accepts: | |
6884 | ||
4f19785b | 6885 | 0(x[0-7A-F]+)|([0-7]+)|(b[01]) |
02aa26ce NT |
6886 | [\d_]+(\.[\d_]*)?[Ee](\d+) |
6887 | ||
6888 | Underbars (_) are allowed in decimal numbers. If -w is on, | |
6889 | underbars before a decimal point must be at three digit intervals. | |
6890 | ||
3280af22 | 6891 | Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the |
02aa26ce NT |
6892 | thing it reads. |
6893 | ||
6894 | If it reads a number without a decimal point or an exponent, it will | |
6895 | try converting the number to an integer and see if it can do so | |
6896 | without loss of precision. | |
6897 | */ | |
4e553d73 | 6898 | |
378cc40b | 6899 | char * |
b73d6f50 | 6900 | Perl_scan_num(pTHX_ char *start, YYSTYPE* lvalp) |
378cc40b | 6901 | { |
02aa26ce NT |
6902 | register char *s = start; /* current position in buffer */ |
6903 | register char *d; /* destination in temp buffer */ | |
6904 | register char *e; /* end of temp buffer */ | |
86554af2 | 6905 | NV nv; /* number read, as a double */ |
a7cb1f99 | 6906 | SV *sv = Nullsv; /* place to put the converted number */ |
b8403495 | 6907 | bool floatit; /* boolean: int or float? */ |
02aa26ce | 6908 | char *lastub = 0; /* position of last underbar */ |
fc36a67e | 6909 | static char number_too_long[] = "Number too long"; |
378cc40b | 6910 | |
02aa26ce NT |
6911 | /* We use the first character to decide what type of number this is */ |
6912 | ||
378cc40b | 6913 | switch (*s) { |
79072805 | 6914 | default: |
cea2e8a9 | 6915 | Perl_croak(aTHX_ "panic: scan_num"); |
4e553d73 | 6916 | |
02aa26ce | 6917 | /* if it starts with a 0, it could be an octal number, a decimal in |
a7cb1f99 | 6918 | 0.13 disguise, or a hexadecimal number, or a binary number. */ |
378cc40b LW |
6919 | case '0': |
6920 | { | |
02aa26ce NT |
6921 | /* variables: |
6922 | u holds the "number so far" | |
4f19785b WSI |
6923 | shift the power of 2 of the base |
6924 | (hex == 4, octal == 3, binary == 1) | |
02aa26ce NT |
6925 | overflowed was the number more than we can hold? |
6926 | ||
6927 | Shift is used when we add a digit. It also serves as an "are | |
4f19785b WSI |
6928 | we in octal/hex/binary?" indicator to disallow hex characters |
6929 | when in octal mode. | |
02aa26ce | 6930 | */ |
9e24b6e2 JH |
6931 | NV n = 0.0; |
6932 | UV u = 0; | |
79072805 | 6933 | I32 shift; |
9e24b6e2 JH |
6934 | bool overflowed = FALSE; |
6935 | static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 }; | |
6936 | static char* bases[5] = { "", "binary", "", "octal", | |
6937 | "hexadecimal" }; | |
6938 | static char* Bases[5] = { "", "Binary", "", "Octal", | |
6939 | "Hexadecimal" }; | |
6940 | static char *maxima[5] = { "", | |
6941 | "0b11111111111111111111111111111111", | |
6942 | "", | |
893fe2c2 | 6943 | "037777777777", |
9e24b6e2 JH |
6944 | "0xffffffff" }; |
6945 | char *base, *Base, *max; | |
378cc40b | 6946 | |
02aa26ce | 6947 | /* check for hex */ |
378cc40b LW |
6948 | if (s[1] == 'x') { |
6949 | shift = 4; | |
6950 | s += 2; | |
4f19785b WSI |
6951 | } else if (s[1] == 'b') { |
6952 | shift = 1; | |
6953 | s += 2; | |
378cc40b | 6954 | } |
02aa26ce | 6955 | /* check for a decimal in disguise */ |
b78218b7 | 6956 | else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E') |
378cc40b | 6957 | goto decimal; |
02aa26ce | 6958 | /* so it must be octal */ |
378cc40b LW |
6959 | else |
6960 | shift = 3; | |
9e24b6e2 JH |
6961 | |
6962 | base = bases[shift]; | |
6963 | Base = Bases[shift]; | |
6964 | max = maxima[shift]; | |
02aa26ce | 6965 | |
4f19785b | 6966 | /* read the rest of the number */ |
378cc40b | 6967 | for (;;) { |
9e24b6e2 | 6968 | /* x is used in the overflow test, |
893fe2c2 | 6969 | b is the digit we're adding on. */ |
9e24b6e2 | 6970 | UV x, b; |
55497cff | 6971 | |
378cc40b | 6972 | switch (*s) { |
02aa26ce NT |
6973 | |
6974 | /* if we don't mention it, we're done */ | |
378cc40b LW |
6975 | default: |
6976 | goto out; | |
02aa26ce NT |
6977 | |
6978 | /* _ are ignored */ | |
de3bb511 LW |
6979 | case '_': |
6980 | s++; | |
6981 | break; | |
02aa26ce NT |
6982 | |
6983 | /* 8 and 9 are not octal */ | |
378cc40b | 6984 | case '8': case '9': |
4f19785b | 6985 | if (shift == 3) |
cea2e8a9 | 6986 | yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s)); |
378cc40b | 6987 | /* FALL THROUGH */ |
02aa26ce NT |
6988 | |
6989 | /* octal digits */ | |
4f19785b | 6990 | case '2': case '3': case '4': |
378cc40b | 6991 | case '5': case '6': case '7': |
4f19785b | 6992 | if (shift == 1) |
cea2e8a9 | 6993 | yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s)); |
4f19785b WSI |
6994 | /* FALL THROUGH */ |
6995 | ||
6996 | case '0': case '1': | |
02aa26ce | 6997 | b = *s++ & 15; /* ASCII digit -> value of digit */ |
55497cff | 6998 | goto digit; |
02aa26ce NT |
6999 | |
7000 | /* hex digits */ | |
378cc40b LW |
7001 | case 'a': case 'b': case 'c': case 'd': case 'e': case 'f': |
7002 | case 'A': case 'B': case 'C': case 'D': case 'E': case 'F': | |
02aa26ce | 7003 | /* make sure they said 0x */ |
378cc40b LW |
7004 | if (shift != 4) |
7005 | goto out; | |
55497cff | 7006 | b = (*s++ & 7) + 9; |
02aa26ce NT |
7007 | |
7008 | /* Prepare to put the digit we have onto the end | |
7009 | of the number so far. We check for overflows. | |
7010 | */ | |
7011 | ||
55497cff | 7012 | digit: |
9e24b6e2 JH |
7013 | if (!overflowed) { |
7014 | x = u << shift; /* make room for the digit */ | |
7015 | ||
7016 | if ((x >> shift) != u | |
7017 | && !(PL_hints & HINT_NEW_BINARY)) { | |
9e24b6e2 JH |
7018 | overflowed = TRUE; |
7019 | n = (NV) u; | |
767a6a26 PM |
7020 | if (ckWARN_d(WARN_OVERFLOW)) |
7021 | Perl_warner(aTHX_ WARN_OVERFLOW, | |
9e24b6e2 JH |
7022 | "Integer overflow in %s number", |
7023 | base); | |
7024 | } else | |
7025 | u = x | b; /* add the digit to the end */ | |
7026 | } | |
7027 | if (overflowed) { | |
7028 | n *= nvshift[shift]; | |
7029 | /* If an NV has not enough bits in its | |
7030 | * mantissa to represent an UV this summing of | |
7031 | * small low-order numbers is a waste of time | |
7032 | * (because the NV cannot preserve the | |
7033 | * low-order bits anyway): we could just | |
7034 | * remember when did we overflow and in the | |
7035 | * end just multiply n by the right | |
7036 | * amount. */ | |
7037 | n += (NV) b; | |
55497cff | 7038 | } |
378cc40b LW |
7039 | break; |
7040 | } | |
7041 | } | |
02aa26ce NT |
7042 | |
7043 | /* if we get here, we had success: make a scalar value from | |
7044 | the number. | |
7045 | */ | |
378cc40b | 7046 | out: |
79072805 | 7047 | sv = NEWSV(92,0); |
9e24b6e2 | 7048 | if (overflowed) { |
767a6a26 PM |
7049 | if (ckWARN(WARN_PORTABLE) && n > 4294967295.0) |
7050 | Perl_warner(aTHX_ WARN_PORTABLE, | |
9e24b6e2 JH |
7051 | "%s number > %s non-portable", |
7052 | Base, max); | |
7053 | sv_setnv(sv, n); | |
7054 | } | |
7055 | else { | |
15041a67 | 7056 | #if UVSIZE > 4 |
767a6a26 PM |
7057 | if (ckWARN(WARN_PORTABLE) && u > 0xffffffff) |
7058 | Perl_warner(aTHX_ WARN_PORTABLE, | |
9e24b6e2 JH |
7059 | "%s number > %s non-portable", |
7060 | Base, max); | |
2cc4c2dc | 7061 | #endif |
9e24b6e2 JH |
7062 | sv_setuv(sv, u); |
7063 | } | |
2cc4c2dc | 7064 | if (PL_hints & HINT_NEW_BINARY) |
b3ac6de7 | 7065 | sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL); |
378cc40b LW |
7066 | } |
7067 | break; | |
02aa26ce NT |
7068 | |
7069 | /* | |
7070 | handle decimal numbers. | |
7071 | we're also sent here when we read a 0 as the first digit | |
7072 | */ | |
378cc40b LW |
7073 | case '1': case '2': case '3': case '4': case '5': |
7074 | case '6': case '7': case '8': case '9': case '.': | |
7075 | decimal: | |
3280af22 NIS |
7076 | d = PL_tokenbuf; |
7077 | e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */ | |
79072805 | 7078 | floatit = FALSE; |
02aa26ce NT |
7079 | |
7080 | /* read next group of digits and _ and copy into d */ | |
de3bb511 | 7081 | while (isDIGIT(*s) || *s == '_') { |
4e553d73 | 7082 | /* skip underscores, checking for misplaced ones |
02aa26ce NT |
7083 | if -w is on |
7084 | */ | |
93a17b20 | 7085 | if (*s == '_') { |
599cee73 | 7086 | if (ckWARN(WARN_SYNTAX) && lastub && s - lastub != 3) |
cea2e8a9 | 7087 | Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number"); |
93a17b20 LW |
7088 | lastub = ++s; |
7089 | } | |
fc36a67e | 7090 | else { |
02aa26ce | 7091 | /* check for end of fixed-length buffer */ |
fc36a67e | 7092 | if (d >= e) |
cea2e8a9 | 7093 | Perl_croak(aTHX_ number_too_long); |
02aa26ce | 7094 | /* if we're ok, copy the character */ |
378cc40b | 7095 | *d++ = *s++; |
fc36a67e | 7096 | } |
378cc40b | 7097 | } |
02aa26ce NT |
7098 | |
7099 | /* final misplaced underbar check */ | |
d008e5eb | 7100 | if (lastub && s - lastub != 3) { |
d008e5eb | 7101 | if (ckWARN(WARN_SYNTAX)) |
cea2e8a9 | 7102 | Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number"); |
d008e5eb | 7103 | } |
02aa26ce NT |
7104 | |
7105 | /* read a decimal portion if there is one. avoid | |
7106 | 3..5 being interpreted as the number 3. followed | |
7107 | by .5 | |
7108 | */ | |
2f3197b3 | 7109 | if (*s == '.' && s[1] != '.') { |
79072805 | 7110 | floatit = TRUE; |
378cc40b | 7111 | *d++ = *s++; |
02aa26ce NT |
7112 | |
7113 | /* copy, ignoring underbars, until we run out of | |
7114 | digits. Note: no misplaced underbar checks! | |
7115 | */ | |
fc36a67e | 7116 | for (; isDIGIT(*s) || *s == '_'; s++) { |
02aa26ce | 7117 | /* fixed length buffer check */ |
fc36a67e | 7118 | if (d >= e) |
cea2e8a9 | 7119 | Perl_croak(aTHX_ number_too_long); |
fc36a67e | 7120 | if (*s != '_') |
7121 | *d++ = *s; | |
378cc40b | 7122 | } |
dd629d5b GS |
7123 | if (*s == '.' && isDIGIT(s[1])) { |
7124 | /* oops, it's really a v-string, but without the "v" */ | |
7125 | s = start - 1; | |
7126 | goto vstring; | |
7127 | } | |
378cc40b | 7128 | } |
02aa26ce NT |
7129 | |
7130 | /* read exponent part, if present */ | |
93a17b20 | 7131 | if (*s && strchr("eE",*s) && strchr("+-0123456789",s[1])) { |
79072805 LW |
7132 | floatit = TRUE; |
7133 | s++; | |
02aa26ce NT |
7134 | |
7135 | /* regardless of whether user said 3E5 or 3e5, use lower 'e' */ | |
79072805 | 7136 | *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */ |
02aa26ce NT |
7137 | |
7138 | /* allow positive or negative exponent */ | |
378cc40b LW |
7139 | if (*s == '+' || *s == '-') |
7140 | *d++ = *s++; | |
02aa26ce NT |
7141 | |
7142 | /* read digits of exponent (no underbars :-) */ | |
fc36a67e | 7143 | while (isDIGIT(*s)) { |
7144 | if (d >= e) | |
cea2e8a9 | 7145 | Perl_croak(aTHX_ number_too_long); |
378cc40b | 7146 | *d++ = *s++; |
fc36a67e | 7147 | } |
378cc40b | 7148 | } |
02aa26ce NT |
7149 | |
7150 | /* terminate the string */ | |
378cc40b | 7151 | *d = '\0'; |
02aa26ce NT |
7152 | |
7153 | /* make an sv from the string */ | |
79072805 | 7154 | sv = NEWSV(92,0); |
097ee67d | 7155 | |
86554af2 | 7156 | #if defined(Strtol) && defined(Strtoul) |
0b7fceb9 MU |
7157 | |
7158 | /* | |
0b7fceb9 MU |
7159 | strtol/strtoll sets errno to ERANGE if the number is too big |
7160 | for an integer. We try to do an integer conversion first | |
7161 | if no characters indicating "float" have been found. | |
7162 | */ | |
7163 | ||
7164 | if (!floatit) { | |
0b7fceb9 MU |
7165 | IV iv; |
7166 | UV uv; | |
7167 | errno = 0; | |
c239479b | 7168 | if (*PL_tokenbuf == '-') |
96989be3 | 7169 | iv = Strtol(PL_tokenbuf, (char**)NULL, 10); |
c239479b | 7170 | else |
96989be3 | 7171 | uv = Strtoul(PL_tokenbuf, (char**)NULL, 10); |
55eb892c | 7172 | if (errno) |
86554af2 | 7173 | floatit = TRUE; /* Probably just too large. */ |
0b7fceb9 MU |
7174 | else if (*PL_tokenbuf == '-') |
7175 | sv_setiv(sv, iv); | |
86554af2 JH |
7176 | else if (uv <= IV_MAX) |
7177 | sv_setiv(sv, uv); /* Prefer IVs over UVs. */ | |
0b7fceb9 | 7178 | else |
c239479b | 7179 | sv_setuv(sv, uv); |
0b7fceb9 MU |
7180 | } |
7181 | if (floatit) { | |
86554af2 JH |
7182 | nv = Atof(PL_tokenbuf); |
7183 | sv_setnv(sv, nv); | |
7184 | } | |
7185 | #else | |
7186 | /* | |
7187 | No working strtou?ll?. | |
7188 | ||
7189 | Unfortunately atol() doesn't do range checks (returning | |
7190 | LONG_MIN/LONG_MAX, and setting errno to ERANGE on overflows) | |
7191 | everywhere [1], so we cannot use use atol() (or atoll()). | |
7192 | If we could, they would be used, as Atol(), very much like | |
7193 | Strtol() and Strtoul() are used above. | |
7194 | ||
7195 | [1] XXX Configure test needed to check for atol() | |
d6c14000 JH |
7196 | (and atoll()) overflow behaviour XXX |
7197 | ||
7198 | --jhi | |
86554af2 JH |
7199 | |
7200 | We need to do this the hard way. */ | |
7201 | ||
7202 | nv = Atof(PL_tokenbuf); | |
7203 | ||
7204 | /* See if we can make do with an integer value without loss of | |
7205 | precision. We use U_V to cast to a UV, because some | |
7206 | compilers have issues. Then we try casting it back and see | |
7207 | if it was the same [1]. We only do this if we know we | |
7208 | specifically read an integer. If floatit is true, then we | |
4e553d73 | 7209 | don't need to do the conversion at all. |
86554af2 JH |
7210 | |
7211 | [1] Note that this is lossy if our NVs cannot preserve our | |
d6c14000 JH |
7212 | UVs. There are metaconfig defines NV_PRESERVES_UV (a boolean) |
7213 | and NV_PRESERVES_UV_BITS (a number), but in general we really | |
7214 | do hope all such potentially lossy platforms have strtou?ll? | |
7215 | to do a lossless IV/UV conversion. | |
7216 | ||
7217 | Maybe could do some tricks with DBL_DIG, LDBL_DIG and | |
7218 | DBL_MANT_DIG and LDBL_MANT_DIG (these are already available | |
7219 | as NV_DIG and NV_MANT_DIG)? | |
4e553d73 | 7220 | |
d6c14000 | 7221 | --jhi |
86554af2 JH |
7222 | */ |
7223 | { | |
7224 | UV uv = U_V(nv); | |
7225 | if (!floatit && (NV)uv == nv) { | |
7226 | if (uv <= IV_MAX) | |
7227 | sv_setiv(sv, uv); /* Prefer IVs over UVs. */ | |
7228 | else | |
7229 | sv_setuv(sv, uv); | |
7230 | } | |
7231 | else | |
7232 | sv_setnv(sv, nv); | |
96989be3 | 7233 | } |
0b7fceb9 | 7234 | #endif |
b8403495 JH |
7235 | if ( floatit ? (PL_hints & HINT_NEW_FLOAT) : |
7236 | (PL_hints & HINT_NEW_INTEGER) ) | |
4e553d73 | 7237 | sv = new_constant(PL_tokenbuf, d - PL_tokenbuf, |
b8403495 JH |
7238 | (floatit ? "float" : "integer"), |
7239 | sv, Nullsv, NULL); | |
378cc40b | 7240 | break; |
0b7fceb9 | 7241 | |
e312add1 | 7242 | /* if it starts with a v, it could be a v-string */ |
a7cb1f99 | 7243 | case 'v': |
dd629d5b | 7244 | vstring: |
a7cb1f99 | 7245 | { |
a7cb1f99 GS |
7246 | char *pos = s; |
7247 | pos++; | |
dd629d5b | 7248 | while (isDIGIT(*pos) || *pos == '_') |
a7cb1f99 | 7249 | pos++; |
e526c9e6 | 7250 | if (!isALPHA(*pos)) { |
9aa983d2 | 7251 | UV rev, revmax = 0; |
ad391ad9 | 7252 | U8 tmpbuf[UTF8_MAXLEN+1]; |
a7cb1f99 | 7253 | U8 *tmpend; |
a7cb1f99 GS |
7254 | s++; /* get past 'v' */ |
7255 | ||
7256 | sv = NEWSV(92,5); | |
a7cb1f99 GS |
7257 | sv_setpvn(sv, "", 0); |
7258 | ||
e526c9e6 | 7259 | for (;;) { |
3cb0bbe5 GS |
7260 | if (*s == '0' && isDIGIT(s[1])) |
7261 | yyerror("Octal number in vector unsupported"); | |
dd629d5b GS |
7262 | rev = 0; |
7263 | { | |
7264 | /* this is atoi() that tolerates underscores */ | |
7265 | char *end = pos; | |
7266 | UV mult = 1; | |
7267 | while (--end >= s) { | |
7268 | UV orev; | |
7269 | if (*end == '_') | |
7270 | continue; | |
7271 | orev = rev; | |
7272 | rev += (*end - '0') * mult; | |
7273 | mult *= 10; | |
7274 | if (orev > rev && ckWARN_d(WARN_OVERFLOW)) | |
7275 | Perl_warner(aTHX_ WARN_OVERFLOW, | |
7276 | "Integer overflow in decimal number"); | |
7277 | } | |
7278 | } | |
e526c9e6 | 7279 | tmpend = uv_to_utf8(tmpbuf, rev); |
9aa983d2 JH |
7280 | if (rev > revmax) |
7281 | revmax = rev; | |
e526c9e6 GS |
7282 | sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf); |
7283 | if (*pos == '.' && isDIGIT(pos[1])) | |
7284 | s = ++pos; | |
3818b22b | 7285 | else { |
e526c9e6 GS |
7286 | s = pos; |
7287 | break; | |
3818b22b | 7288 | } |
dd629d5b | 7289 | while (isDIGIT(*pos) || *pos == '_') |
e526c9e6 GS |
7290 | pos++; |
7291 | } | |
a7cb1f99 GS |
7292 | |
7293 | SvPOK_on(sv); | |
a7cb1f99 | 7294 | SvREADONLY_on(sv); |
9aa983d2 | 7295 | if (revmax > 127) { |
3818b22b | 7296 | SvUTF8_on(sv); |
9aa983d2 | 7297 | if (revmax < 256) |
7948272d | 7298 | sv_utf8_downgrade(sv, TRUE); |
560a288e | 7299 | } |
a7cb1f99 GS |
7300 | } |
7301 | } | |
7302 | break; | |
79072805 | 7303 | } |
a687059c | 7304 | |
02aa26ce NT |
7305 | /* make the op for the constant and return */ |
7306 | ||
a7cb1f99 | 7307 | if (sv) |
b73d6f50 | 7308 | lvalp->opval = newSVOP(OP_CONST, 0, sv); |
a7cb1f99 | 7309 | else |
b73d6f50 | 7310 | lvalp->opval = Nullop; |
a687059c | 7311 | |
378cc40b LW |
7312 | return s; |
7313 | } | |
7314 | ||
76e3520e | 7315 | STATIC char * |
cea2e8a9 | 7316 | S_scan_formline(pTHX_ register char *s) |
378cc40b | 7317 | { |
79072805 | 7318 | register char *eol; |
378cc40b | 7319 | register char *t; |
79cb57f6 | 7320 | SV *stuff = newSVpvn("",0); |
79072805 | 7321 | bool needargs = FALSE; |
378cc40b | 7322 | |
79072805 | 7323 | while (!needargs) { |
c2e66d9e | 7324 | if (*s == '.' || *s == /*{*/'}') { |
79072805 | 7325 | /*SUPPRESS 530*/ |
51882d45 | 7326 | #ifdef PERL_STRICT_CR |
bf4acbe4 | 7327 | for (t = s+1;SPACE_OR_TAB(*t); t++) ; |
51882d45 | 7328 | #else |
bf4acbe4 | 7329 | for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ; |
51882d45 | 7330 | #endif |
6a65c6a0 | 7331 | if (*t == '\n' || t == PL_bufend) |
79072805 LW |
7332 | break; |
7333 | } | |
3280af22 | 7334 | if (PL_in_eval && !PL_rsfp) { |
93a17b20 | 7335 | eol = strchr(s,'\n'); |
0f85fab0 | 7336 | if (!eol++) |
3280af22 | 7337 | eol = PL_bufend; |
0f85fab0 LW |
7338 | } |
7339 | else | |
3280af22 | 7340 | eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); |
79072805 | 7341 | if (*s != '#') { |
a0d0e21e LW |
7342 | for (t = s; t < eol; t++) { |
7343 | if (*t == '~' && t[1] == '~' && SvCUR(stuff)) { | |
7344 | needargs = FALSE; | |
7345 | goto enough; /* ~~ must be first line in formline */ | |
378cc40b | 7346 | } |
a0d0e21e LW |
7347 | if (*t == '@' || *t == '^') |
7348 | needargs = TRUE; | |
378cc40b | 7349 | } |
a0d0e21e | 7350 | sv_catpvn(stuff, s, eol-s); |
2dc4c65b GS |
7351 | #ifndef PERL_STRICT_CR |
7352 | if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') { | |
7353 | char *end = SvPVX(stuff) + SvCUR(stuff); | |
7354 | end[-2] = '\n'; | |
7355 | end[-1] = '\0'; | |
7356 | SvCUR(stuff)--; | |
7357 | } | |
7358 | #endif | |
79072805 LW |
7359 | } |
7360 | s = eol; | |
3280af22 NIS |
7361 | if (PL_rsfp) { |
7362 | s = filter_gets(PL_linestr, PL_rsfp, 0); | |
7363 | PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr); | |
7364 | PL_bufend = PL_bufptr + SvCUR(PL_linestr); | |
79072805 | 7365 | if (!s) { |
3280af22 | 7366 | s = PL_bufptr; |
79072805 | 7367 | yyerror("Format not terminated"); |
378cc40b LW |
7368 | break; |
7369 | } | |
378cc40b | 7370 | } |
463ee0b2 | 7371 | incline(s); |
79072805 | 7372 | } |
a0d0e21e LW |
7373 | enough: |
7374 | if (SvCUR(stuff)) { | |
3280af22 | 7375 | PL_expect = XTERM; |
79072805 | 7376 | if (needargs) { |
3280af22 NIS |
7377 | PL_lex_state = LEX_NORMAL; |
7378 | PL_nextval[PL_nexttoke].ival = 0; | |
79072805 LW |
7379 | force_next(','); |
7380 | } | |
a0d0e21e | 7381 | else |
3280af22 NIS |
7382 | PL_lex_state = LEX_FORMLINE; |
7383 | PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff); | |
79072805 | 7384 | force_next(THING); |
3280af22 | 7385 | PL_nextval[PL_nexttoke].ival = OP_FORMLINE; |
79072805 | 7386 | force_next(LSTOP); |
378cc40b | 7387 | } |
79072805 | 7388 | else { |
8990e307 | 7389 | SvREFCNT_dec(stuff); |
3280af22 NIS |
7390 | PL_lex_formbrack = 0; |
7391 | PL_bufptr = s; | |
79072805 LW |
7392 | } |
7393 | return s; | |
378cc40b | 7394 | } |
a687059c | 7395 | |
76e3520e | 7396 | STATIC void |
cea2e8a9 | 7397 | S_set_csh(pTHX) |
a687059c | 7398 | { |
ae986130 | 7399 | #ifdef CSH |
3280af22 NIS |
7400 | if (!PL_cshlen) |
7401 | PL_cshlen = strlen(PL_cshname); | |
ae986130 | 7402 | #endif |
a687059c | 7403 | } |
463ee0b2 | 7404 | |
ba6d6ac9 | 7405 | I32 |
864dbfa3 | 7406 | Perl_start_subparse(pTHX_ I32 is_format, U32 flags) |
8990e307 | 7407 | { |
3280af22 NIS |
7408 | I32 oldsavestack_ix = PL_savestack_ix; |
7409 | CV* outsidecv = PL_compcv; | |
748a9306 | 7410 | AV* comppadlist; |
8990e307 | 7411 | |
3280af22 NIS |
7412 | if (PL_compcv) { |
7413 | assert(SvTYPE(PL_compcv) == SVt_PVCV); | |
e9a444f0 | 7414 | } |
7766f137 | 7415 | SAVEI32(PL_subline); |
3280af22 NIS |
7416 | save_item(PL_subname); |
7417 | SAVEI32(PL_padix); | |
354992b1 | 7418 | SAVECOMPPAD(); |
3280af22 NIS |
7419 | SAVESPTR(PL_comppad_name); |
7420 | SAVESPTR(PL_compcv); | |
7421 | SAVEI32(PL_comppad_name_fill); | |
7422 | SAVEI32(PL_min_intro_pending); | |
7423 | SAVEI32(PL_max_intro_pending); | |
7424 | SAVEI32(PL_pad_reset_pending); | |
7425 | ||
7426 | PL_compcv = (CV*)NEWSV(1104,0); | |
7427 | sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV); | |
7428 | CvFLAGS(PL_compcv) |= flags; | |
7429 | ||
7430 | PL_comppad = newAV(); | |
7431 | av_push(PL_comppad, Nullsv); | |
7432 | PL_curpad = AvARRAY(PL_comppad); | |
7433 | PL_comppad_name = newAV(); | |
7434 | PL_comppad_name_fill = 0; | |
7435 | PL_min_intro_pending = 0; | |
7436 | PL_padix = 0; | |
57843af0 | 7437 | PL_subline = CopLINE(PL_curcop); |
6d4ff0d2 | 7438 | #ifdef USE_THREADS |
79cb57f6 | 7439 | av_store(PL_comppad_name, 0, newSVpvn("@_", 2)); |
533c011a NIS |
7440 | PL_curpad[0] = (SV*)newAV(); |
7441 | SvPADMY_on(PL_curpad[0]); /* XXX Needed? */ | |
6d4ff0d2 | 7442 | #endif /* USE_THREADS */ |
748a9306 LW |
7443 | |
7444 | comppadlist = newAV(); | |
7445 | AvREAL_off(comppadlist); | |
3280af22 NIS |
7446 | av_store(comppadlist, 0, (SV*)PL_comppad_name); |
7447 | av_store(comppadlist, 1, (SV*)PL_comppad); | |
748a9306 | 7448 | |
3280af22 NIS |
7449 | CvPADLIST(PL_compcv) = comppadlist; |
7450 | CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv); | |
11343788 | 7451 | #ifdef USE_THREADS |
533c011a NIS |
7452 | CvOWNER(PL_compcv) = 0; |
7453 | New(666, CvMUTEXP(PL_compcv), 1, perl_mutex); | |
7454 | MUTEX_INIT(CvMUTEXP(PL_compcv)); | |
11343788 | 7455 | #endif /* USE_THREADS */ |
748a9306 | 7456 | |
8990e307 LW |
7457 | return oldsavestack_ix; |
7458 | } | |
7459 | ||
084592ab CN |
7460 | #ifdef __SC__ |
7461 | #pragma segment Perl_yylex | |
7462 | #endif | |
8990e307 | 7463 | int |
864dbfa3 | 7464 | Perl_yywarn(pTHX_ char *s) |
8990e307 | 7465 | { |
faef0170 | 7466 | PL_in_eval |= EVAL_WARNONLY; |
748a9306 | 7467 | yyerror(s); |
faef0170 | 7468 | PL_in_eval &= ~EVAL_WARNONLY; |
748a9306 | 7469 | return 0; |
8990e307 LW |
7470 | } |
7471 | ||
7472 | int | |
864dbfa3 | 7473 | Perl_yyerror(pTHX_ char *s) |
463ee0b2 | 7474 | { |
68dc0745 | 7475 | char *where = NULL; |
7476 | char *context = NULL; | |
7477 | int contlen = -1; | |
46fc3d4c | 7478 | SV *msg; |
463ee0b2 | 7479 | |
3280af22 | 7480 | if (!yychar || (yychar == ';' && !PL_rsfp)) |
54310121 | 7481 | where = "at EOF"; |
3280af22 NIS |
7482 | else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 && |
7483 | PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) { | |
7484 | while (isSPACE(*PL_oldoldbufptr)) | |
7485 | PL_oldoldbufptr++; | |
7486 | context = PL_oldoldbufptr; | |
7487 | contlen = PL_bufptr - PL_oldoldbufptr; | |
463ee0b2 | 7488 | } |
3280af22 NIS |
7489 | else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 && |
7490 | PL_oldbufptr != PL_bufptr) { | |
7491 | while (isSPACE(*PL_oldbufptr)) | |
7492 | PL_oldbufptr++; | |
7493 | context = PL_oldbufptr; | |
7494 | contlen = PL_bufptr - PL_oldbufptr; | |
463ee0b2 LW |
7495 | } |
7496 | else if (yychar > 255) | |
68dc0745 | 7497 | where = "next token ???"; |
cdfb297e GS |
7498 | #ifdef USE_PURE_BISON |
7499 | /* GNU Bison sets the value -2 */ | |
7500 | else if (yychar == -2) { | |
7501 | #else | |
463ee0b2 | 7502 | else if ((yychar & 127) == 127) { |
cdfb297e | 7503 | #endif |
3280af22 NIS |
7504 | if (PL_lex_state == LEX_NORMAL || |
7505 | (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL)) | |
68dc0745 | 7506 | where = "at end of line"; |
3280af22 | 7507 | else if (PL_lex_inpat) |
68dc0745 | 7508 | where = "within pattern"; |
463ee0b2 | 7509 | else |
68dc0745 | 7510 | where = "within string"; |
463ee0b2 | 7511 | } |
46fc3d4c | 7512 | else { |
79cb57f6 | 7513 | SV *where_sv = sv_2mortal(newSVpvn("next char ", 10)); |
46fc3d4c | 7514 | if (yychar < 32) |
cea2e8a9 | 7515 | Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar)); |
46fc3d4c | 7516 | else if (isPRINT_LC(yychar)) |
cea2e8a9 | 7517 | Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar); |
463ee0b2 | 7518 | else |
cea2e8a9 | 7519 | Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255); |
46fc3d4c | 7520 | where = SvPVX(where_sv); |
463ee0b2 | 7521 | } |
46fc3d4c | 7522 | msg = sv_2mortal(newSVpv(s, 0)); |
ed094faf GS |
7523 | Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ", |
7524 | CopFILE(PL_curcop), (IV)CopLINE(PL_curcop)); | |
68dc0745 | 7525 | if (context) |
cea2e8a9 | 7526 | Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context); |
463ee0b2 | 7527 | else |
cea2e8a9 | 7528 | Perl_sv_catpvf(aTHX_ msg, "%s\n", where); |
57843af0 | 7529 | if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) { |
cf2093f6 | 7530 | Perl_sv_catpvf(aTHX_ msg, |
57def98f | 7531 | " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n", |
cf2093f6 | 7532 | (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start); |
3280af22 | 7533 | PL_multi_end = 0; |
a0d0e21e | 7534 | } |
faef0170 | 7535 | if (PL_in_eval & EVAL_WARNONLY) |
894356b3 | 7536 | Perl_warn(aTHX_ "%"SVf, msg); |
463ee0b2 | 7537 | else |
5a844595 | 7538 | qerror(msg); |
c7d6bfb2 GS |
7539 | if (PL_error_count >= 10) { |
7540 | if (PL_in_eval && SvCUR(ERRSV)) | |
d2560b70 | 7541 | Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n", |
c7d6bfb2 GS |
7542 | ERRSV, CopFILE(PL_curcop)); |
7543 | else | |
7544 | Perl_croak(aTHX_ "%s has too many errors.\n", | |
7545 | CopFILE(PL_curcop)); | |
7546 | } | |
3280af22 NIS |
7547 | PL_in_my = 0; |
7548 | PL_in_my_stash = Nullhv; | |
463ee0b2 LW |
7549 | return 0; |
7550 | } | |
084592ab CN |
7551 | #ifdef __SC__ |
7552 | #pragma segment Main | |
7553 | #endif | |
4e35701f | 7554 | |
b250498f | 7555 | STATIC char* |
3ae08724 | 7556 | S_swallow_bom(pTHX_ U8 *s) |
01ec43d0 | 7557 | { |
b250498f GS |
7558 | STRLEN slen; |
7559 | slen = SvCUR(PL_linestr); | |
7560 | switch (*s) { | |
4e553d73 NIS |
7561 | case 0xFF: |
7562 | if (s[1] == 0xFE) { | |
01ec43d0 | 7563 | /* UTF-16 little-endian */ |
3ae08724 | 7564 | if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */ |
01ec43d0 GS |
7565 | Perl_croak(aTHX_ "Unsupported script encoding"); |
7566 | #ifndef PERL_NO_UTF16_FILTER | |
dea0fc0b | 7567 | DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-LE script encoding\n")); |
3ae08724 | 7568 | s += 2; |
dea0fc0b JH |
7569 | if (PL_bufend > (char*)s) { |
7570 | U8 *news; | |
7571 | I32 newlen; | |
7572 | ||
7573 | filter_add(utf16rev_textfilter, NULL); | |
7574 | New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8); | |
f72f5f89 JH |
7575 | PL_bufend = (char*)utf16_to_utf8_reversed(s, news, |
7576 | PL_bufend - (char*)s - 1, | |
dea0fc0b JH |
7577 | &newlen); |
7578 | Copy(news, s, newlen, U8); | |
7579 | SvCUR_set(PL_linestr, newlen); | |
7580 | PL_bufend = SvPVX(PL_linestr) + newlen; | |
7581 | news[newlen++] = '\0'; | |
7582 | Safefree(news); | |
7583 | } | |
b250498f | 7584 | #else |
01ec43d0 | 7585 | Perl_croak(aTHX_ "Unsupported script encoding"); |
b250498f | 7586 | #endif |
01ec43d0 GS |
7587 | } |
7588 | break; | |
78ae23f5 | 7589 | case 0xFE: |
3ae08724 | 7590 | if (s[1] == 0xFF) { /* UTF-16 big-endian */ |
01ec43d0 | 7591 | #ifndef PERL_NO_UTF16_FILTER |
dea0fc0b JH |
7592 | DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding\n")); |
7593 | s += 2; | |
7594 | if (PL_bufend > (char *)s) { | |
7595 | U8 *news; | |
7596 | I32 newlen; | |
7597 | ||
7598 | filter_add(utf16_textfilter, NULL); | |
7599 | New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8); | |
7600 | PL_bufend = (char*)utf16_to_utf8(s, news, | |
7601 | PL_bufend - (char*)s, | |
7602 | &newlen); | |
7603 | Copy(news, s, newlen, U8); | |
7604 | SvCUR_set(PL_linestr, newlen); | |
7605 | PL_bufend = SvPVX(PL_linestr) + newlen; | |
7606 | news[newlen++] = '\0'; | |
7607 | Safefree(news); | |
7608 | } | |
b250498f | 7609 | #else |
01ec43d0 | 7610 | Perl_croak(aTHX_ "Unsupported script encoding"); |
b250498f | 7611 | #endif |
01ec43d0 GS |
7612 | } |
7613 | break; | |
3ae08724 GS |
7614 | case 0xEF: |
7615 | if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) { | |
dea0fc0b | 7616 | DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-8 script encoding\n")); |
01ec43d0 GS |
7617 | s += 3; /* UTF-8 */ |
7618 | } | |
7619 | break; | |
7620 | case 0: | |
7621 | if (slen > 3 && s[1] == 0 && /* UTF-32 big-endian */ | |
3ae08724 | 7622 | s[2] == 0xFE && s[3] == 0xFF) |
01ec43d0 GS |
7623 | { |
7624 | Perl_croak(aTHX_ "Unsupported script encoding"); | |
7625 | } | |
7626 | } | |
b8f84bb2 | 7627 | return (char*)s; |
b250498f | 7628 | } |
4755096e GS |
7629 | |
7630 | #ifdef PERL_OBJECT | |
7631 | #include "XSUB.h" | |
7632 | #endif | |
7633 | ||
7634 | /* | |
7635 | * restore_rsfp | |
7636 | * Restore a source filter. | |
7637 | */ | |
7638 | ||
7639 | static void | |
7640 | restore_rsfp(pTHXo_ void *f) | |
7641 | { | |
7642 | PerlIO *fp = (PerlIO*)f; | |
7643 | ||
7644 | if (PL_rsfp == PerlIO_stdin()) | |
7645 | PerlIO_clearerr(PL_rsfp); | |
7646 | else if (PL_rsfp && (PL_rsfp != fp)) | |
7647 | PerlIO_close(PL_rsfp); | |
7648 | PL_rsfp = fp; | |
7649 | } | |
6e3aabd6 GS |
7650 | |
7651 | #ifndef PERL_NO_UTF16_FILTER | |
7652 | static I32 | |
7653 | utf16_textfilter(pTHXo_ int idx, SV *sv, int maxlen) | |
7654 | { | |
7655 | I32 count = FILTER_READ(idx+1, sv, maxlen); | |
7656 | if (count) { | |
7657 | U8* tmps; | |
7658 | U8* tend; | |
dea0fc0b | 7659 | I32 newlen; |
6e3aabd6 | 7660 | New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8); |
f72f5f89 JH |
7661 | if (!*SvPV_nolen(sv)) |
7662 | /* Game over, but don't feed an odd-length string to utf16_to_utf8 */ | |
7663 | return count; | |
4e553d73 | 7664 | |
dea0fc0b | 7665 | tend = utf16_to_utf8((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen); |
6e3aabd6 GS |
7666 | sv_usepvn(sv, (char*)tmps, tend - tmps); |
7667 | } | |
7668 | return count; | |
7669 | } | |
7670 | ||
7671 | static I32 | |
7672 | utf16rev_textfilter(pTHXo_ int idx, SV *sv, int maxlen) | |
7673 | { | |
7674 | I32 count = FILTER_READ(idx+1, sv, maxlen); | |
7675 | if (count) { | |
7676 | U8* tmps; | |
7677 | U8* tend; | |
dea0fc0b | 7678 | I32 newlen; |
f72f5f89 JH |
7679 | if (!*SvPV_nolen(sv)) |
7680 | /* Game over, but don't feed an odd-length string to utf16_to_utf8 */ | |
7681 | return count; | |
7682 | ||
6e3aabd6 | 7683 | New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8); |
dea0fc0b | 7684 | tend = utf16_to_utf8_reversed((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen); |
6e3aabd6 GS |
7685 | sv_usepvn(sv, (char*)tmps, tend - tmps); |
7686 | } | |
7687 | return count; | |
7688 | } | |
7689 | #endif |