Commit | Line | Data |
---|---|---|
a0d0e21e | 1 | /* toke.c |
a687059c | 2 | * |
3818b22b | 3 | * Copyright (c) 1991-2000, 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 | |
ffb4593c NT |
16 | * parser, perly.y. |
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); |
51371543 | 31 | |
9059aa12 LW |
32 | #define XFAKEBRACK 128 |
33 | #define XENUMMASK 127 | |
34 | ||
7e2040f0 | 35 | /*#define UTF (SvUTF8(PL_linestr) && !(PL_hints & HINT_BYTE))*/ |
a0ed51b3 LW |
36 | #define UTF (PL_hints & HINT_UTF8) |
37 | ||
2b92dfce GS |
38 | /* In variables name $^X, these are the legal values for X. |
39 | * 1999-02-27 mjd-perl-patch@plover.com */ | |
40 | #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x))) | |
41 | ||
ffb4593c NT |
42 | /* LEX_* are values for PL_lex_state, the state of the lexer. |
43 | * They are arranged oddly so that the guard on the switch statement | |
79072805 LW |
44 | * can get by with a single comparison (if the compiler is smart enough). |
45 | */ | |
46 | ||
fb73857a | 47 | /* #define LEX_NOTPARSING 11 is done in perl.h. */ |
48 | ||
55497cff | 49 | #define LEX_NORMAL 10 |
50 | #define LEX_INTERPNORMAL 9 | |
51 | #define LEX_INTERPCASEMOD 8 | |
52 | #define LEX_INTERPPUSH 7 | |
53 | #define LEX_INTERPSTART 6 | |
54 | #define LEX_INTERPEND 5 | |
55 | #define LEX_INTERPENDMAYBE 4 | |
56 | #define LEX_INTERPCONCAT 3 | |
57 | #define LEX_INTERPCONST 2 | |
58 | #define LEX_FORMLINE 1 | |
59 | #define LEX_KNOWNEXT 0 | |
79072805 | 60 | |
a790bc05 | 61 | /* XXX If this causes problems, set i_unistd=undef in the hint file. */ |
62 | #ifdef I_UNISTD | |
63 | # include <unistd.h> /* Needed for execv() */ | |
64 | #endif | |
65 | ||
66 | ||
79072805 LW |
67 | #ifdef ff_next |
68 | #undef ff_next | |
d48672a2 LW |
69 | #endif |
70 | ||
a1a0e61e TD |
71 | #ifdef USE_PURE_BISON |
72 | YYSTYPE* yylval_pointer = NULL; | |
73 | int* yychar_pointer = NULL; | |
22c35a8c GS |
74 | # undef yylval |
75 | # undef yychar | |
e4bfbdd4 JH |
76 | # define yylval (*yylval_pointer) |
77 | # define yychar (*yychar_pointer) | |
78 | # define PERL_YYLEX_PARAM yylval_pointer,yychar_pointer | |
cea2e8a9 GS |
79 | # undef yylex |
80 | # define yylex() Perl_yylex(aTHX_ yylval_pointer, yychar_pointer) | |
a1a0e61e TD |
81 | #endif |
82 | ||
79072805 | 83 | #include "keywords.h" |
fe14fcc3 | 84 | |
ffb4593c NT |
85 | /* CLINE is a macro that ensures PL_copline has a sane value */ |
86 | ||
ae986130 LW |
87 | #ifdef CLINE |
88 | #undef CLINE | |
89 | #endif | |
57843af0 | 90 | #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline)) |
3280af22 | 91 | |
ffb4593c NT |
92 | /* |
93 | * Convenience functions to return different tokens and prime the | |
9cbb5ea2 | 94 | * lexer for the next token. They all take an argument. |
ffb4593c NT |
95 | * |
96 | * TOKEN : generic token (used for '(', DOLSHARP, etc) | |
97 | * OPERATOR : generic operator | |
98 | * AOPERATOR : assignment operator | |
99 | * PREBLOCK : beginning the block after an if, while, foreach, ... | |
100 | * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref) | |
101 | * PREREF : *EXPR where EXPR is not a simple identifier | |
102 | * TERM : expression term | |
103 | * LOOPX : loop exiting command (goto, last, dump, etc) | |
104 | * FTST : file test operator | |
105 | * FUN0 : zero-argument function | |
2d2e263d | 106 | * FUN1 : not used, except for not, which isn't a UNIOP |
ffb4593c NT |
107 | * BOop : bitwise or or xor |
108 | * BAop : bitwise and | |
109 | * SHop : shift operator | |
110 | * PWop : power operator | |
9cbb5ea2 | 111 | * PMop : pattern-matching operator |
ffb4593c NT |
112 | * Aop : addition-level operator |
113 | * Mop : multiplication-level operator | |
114 | * Eop : equality-testing operator | |
115 | * Rop : relational operator <= != gt | |
116 | * | |
117 | * Also see LOP and lop() below. | |
118 | */ | |
119 | ||
3280af22 NIS |
120 | #define TOKEN(retval) return (PL_bufptr = s,(int)retval) |
121 | #define OPERATOR(retval) return (PL_expect = XTERM,PL_bufptr = s,(int)retval) | |
122 | #define AOPERATOR(retval) return ao((PL_expect = XTERM,PL_bufptr = s,(int)retval)) | |
123 | #define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s,(int)retval) | |
124 | #define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s,(int)retval) | |
125 | #define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s,(int)retval) | |
126 | #define TERM(retval) return (CLINE, PL_expect = XOPERATOR,PL_bufptr = s,(int)retval) | |
127 | #define LOOPX(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LOOPEX) | |
128 | #define FTST(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)UNIOP) | |
129 | #define FUN0(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC0) | |
130 | #define FUN1(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC1) | |
131 | #define BOop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITOROP)) | |
132 | #define BAop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITANDOP)) | |
133 | #define SHop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)SHIFTOP)) | |
134 | #define PWop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)POWOP)) | |
135 | #define PMop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MATCHOP) | |
136 | #define Aop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)ADDOP)) | |
137 | #define Mop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MULOP)) | |
138 | #define Eop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)EQOP) | |
139 | #define Rop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)RELOP) | |
2f3197b3 | 140 | |
a687059c LW |
141 | /* This bit of chicanery makes a unary function followed by |
142 | * a parenthesis into a function with one argument, highest precedence. | |
143 | */ | |
2f3197b3 | 144 | #define UNI(f) return(yylval.ival = f, \ |
3280af22 NIS |
145 | PL_expect = XTERM, \ |
146 | PL_bufptr = s, \ | |
147 | PL_last_uni = PL_oldbufptr, \ | |
148 | PL_last_lop_op = f, \ | |
a687059c LW |
149 | (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) ) |
150 | ||
79072805 | 151 | #define UNIBRACK(f) return(yylval.ival = f, \ |
3280af22 NIS |
152 | PL_bufptr = s, \ |
153 | PL_last_uni = PL_oldbufptr, \ | |
79072805 LW |
154 | (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) ) |
155 | ||
9f68db38 | 156 | /* grandfather return to old style */ |
3280af22 | 157 | #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP) |
79072805 | 158 | |
ffb4593c NT |
159 | /* |
160 | * S_ao | |
161 | * | |
162 | * This subroutine detects &&= and ||= and turns an ANDAND or OROR | |
163 | * into an OP_ANDASSIGN or OP_ORASSIGN | |
164 | */ | |
165 | ||
76e3520e | 166 | STATIC int |
cea2e8a9 | 167 | S_ao(pTHX_ int toketype) |
a0d0e21e | 168 | { |
3280af22 NIS |
169 | if (*PL_bufptr == '=') { |
170 | PL_bufptr++; | |
a0d0e21e LW |
171 | if (toketype == ANDAND) |
172 | yylval.ival = OP_ANDASSIGN; | |
173 | else if (toketype == OROR) | |
174 | yylval.ival = OP_ORASSIGN; | |
175 | toketype = ASSIGNOP; | |
176 | } | |
177 | return toketype; | |
178 | } | |
179 | ||
ffb4593c NT |
180 | /* |
181 | * S_no_op | |
182 | * When Perl expects an operator and finds something else, no_op | |
183 | * prints the warning. It always prints "<something> found where | |
184 | * operator expected. It prints "Missing semicolon on previous line?" | |
185 | * if the surprise occurs at the start of the line. "do you need to | |
186 | * predeclare ..." is printed out for code like "sub bar; foo bar $x" | |
187 | * where the compiler doesn't know if foo is a method call or a function. | |
188 | * It prints "Missing operator before end of line" if there's nothing | |
189 | * after the missing operator, or "... before <...>" if there is something | |
190 | * after the missing operator. | |
191 | */ | |
192 | ||
76e3520e | 193 | STATIC void |
cea2e8a9 | 194 | S_no_op(pTHX_ char *what, char *s) |
463ee0b2 | 195 | { |
3280af22 NIS |
196 | char *oldbp = PL_bufptr; |
197 | bool is_first = (PL_oldbufptr == PL_linestart); | |
68dc0745 | 198 | |
1189a94a GS |
199 | if (!s) |
200 | s = oldbp; | |
201 | else { | |
202 | assert(s >= oldbp); | |
203 | PL_bufptr = s; | |
204 | } | |
cea2e8a9 | 205 | yywarn(Perl_form(aTHX_ "%s found where operator expected", what)); |
748a9306 | 206 | if (is_first) |
cea2e8a9 | 207 | Perl_warn(aTHX_ "\t(Missing semicolon on previous line?)\n"); |
7e2040f0 | 208 | else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) { |
748a9306 | 209 | char *t; |
7e2040f0 | 210 | for (t = PL_oldoldbufptr; *t && (isALNUM_lazy_if(t,UTF) || *t == ':'); t++) ; |
3280af22 | 211 | if (t < PL_bufptr && isSPACE(*t)) |
cea2e8a9 | 212 | Perl_warn(aTHX_ "\t(Do you need to predeclare %.*s?)\n", |
3280af22 | 213 | t - PL_oldoldbufptr, PL_oldoldbufptr); |
748a9306 LW |
214 | } |
215 | else | |
cea2e8a9 | 216 | Perl_warn(aTHX_ "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp); |
3280af22 | 217 | PL_bufptr = oldbp; |
8990e307 LW |
218 | } |
219 | ||
ffb4593c NT |
220 | /* |
221 | * S_missingterm | |
222 | * Complain about missing quote/regexp/heredoc terminator. | |
223 | * If it's called with (char *)NULL then it cauterizes the line buffer. | |
224 | * If we're in a delimited string and the delimiter is a control | |
225 | * character, it's reformatted into a two-char sequence like ^C. | |
226 | * This is fatal. | |
227 | */ | |
228 | ||
76e3520e | 229 | STATIC void |
cea2e8a9 | 230 | S_missingterm(pTHX_ char *s) |
8990e307 LW |
231 | { |
232 | char tmpbuf[3]; | |
233 | char q; | |
234 | if (s) { | |
235 | char *nl = strrchr(s,'\n'); | |
d2719217 | 236 | if (nl) |
8990e307 LW |
237 | *nl = '\0'; |
238 | } | |
9d116dd7 JH |
239 | else if ( |
240 | #ifdef EBCDIC | |
241 | iscntrl(PL_multi_close) | |
242 | #else | |
243 | PL_multi_close < 32 || PL_multi_close == 127 | |
244 | #endif | |
245 | ) { | |
8990e307 | 246 | *tmpbuf = '^'; |
3280af22 | 247 | tmpbuf[1] = toCTRL(PL_multi_close); |
8990e307 LW |
248 | s = "\\n"; |
249 | tmpbuf[2] = '\0'; | |
250 | s = tmpbuf; | |
251 | } | |
252 | else { | |
3280af22 | 253 | *tmpbuf = PL_multi_close; |
8990e307 LW |
254 | tmpbuf[1] = '\0'; |
255 | s = tmpbuf; | |
256 | } | |
257 | q = strchr(s,'"') ? '\'' : '"'; | |
cea2e8a9 | 258 | Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q); |
463ee0b2 | 259 | } |
79072805 | 260 | |
ffb4593c NT |
261 | /* |
262 | * Perl_deprecate | |
ffb4593c NT |
263 | */ |
264 | ||
79072805 | 265 | void |
864dbfa3 | 266 | Perl_deprecate(pTHX_ char *s) |
a0d0e21e | 267 | { |
d008e5eb | 268 | dTHR; |
599cee73 | 269 | if (ckWARN(WARN_DEPRECATED)) |
cea2e8a9 | 270 | Perl_warner(aTHX_ WARN_DEPRECATED, "Use of %s is deprecated", s); |
a0d0e21e LW |
271 | } |
272 | ||
ffb4593c NT |
273 | /* |
274 | * depcom | |
9cbb5ea2 | 275 | * Deprecate a comma-less variable list. |
ffb4593c NT |
276 | */ |
277 | ||
76e3520e | 278 | STATIC void |
cea2e8a9 | 279 | S_depcom(pTHX) |
a0d0e21e LW |
280 | { |
281 | deprecate("comma-less variable list"); | |
282 | } | |
283 | ||
ffb4593c | 284 | /* |
9cbb5ea2 GS |
285 | * experimental text filters for win32 carriage-returns, utf16-to-utf8 and |
286 | * utf16-to-utf8-reversed. | |
ffb4593c NT |
287 | */ |
288 | ||
c39cd008 GS |
289 | #ifdef PERL_CR_FILTER |
290 | static void | |
291 | strip_return(SV *sv) | |
292 | { | |
293 | register char *s = SvPVX(sv); | |
294 | register char *e = s + SvCUR(sv); | |
295 | /* outer loop optimized to do nothing if there are no CR-LFs */ | |
296 | while (s < e) { | |
297 | if (*s++ == '\r' && *s == '\n') { | |
298 | /* hit a CR-LF, need to copy the rest */ | |
299 | register char *d = s - 1; | |
300 | *d++ = *s++; | |
301 | while (s < e) { | |
302 | if (*s == '\r' && s[1] == '\n') | |
303 | s++; | |
304 | *d++ = *s++; | |
305 | } | |
306 | SvCUR(sv) -= s - d; | |
307 | return; | |
308 | } | |
309 | } | |
310 | } | |
a868473f | 311 | |
76e3520e | 312 | STATIC I32 |
c39cd008 | 313 | S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen) |
a868473f | 314 | { |
c39cd008 GS |
315 | I32 count = FILTER_READ(idx+1, sv, maxlen); |
316 | if (count > 0 && !maxlen) | |
317 | strip_return(sv); | |
318 | return count; | |
a868473f NIS |
319 | } |
320 | #endif | |
321 | ||
155aba94 | 322 | #if 0 |
a0ed51b3 | 323 | STATIC I32 |
cea2e8a9 | 324 | S_utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen) |
a0ed51b3 LW |
325 | { |
326 | I32 count = FILTER_READ(idx+1, sv, maxlen); | |
327 | if (count) { | |
dfe13c55 GS |
328 | U8* tmps; |
329 | U8* tend; | |
330 | New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8); | |
a0ed51b3 | 331 | tend = utf16_to_utf8((U16*)SvPVX(sv), tmps, SvCUR(sv)); |
dfe13c55 | 332 | sv_usepvn(sv, (char*)tmps, tend - tmps); |
a0ed51b3 LW |
333 | } |
334 | return count; | |
335 | } | |
336 | ||
337 | STATIC I32 | |
cea2e8a9 | 338 | S_utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen) |
a0ed51b3 LW |
339 | { |
340 | I32 count = FILTER_READ(idx+1, sv, maxlen); | |
341 | if (count) { | |
dfe13c55 GS |
342 | U8* tmps; |
343 | U8* tend; | |
344 | New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8); | |
a0ed51b3 | 345 | tend = utf16_to_utf8_reversed((U16*)SvPVX(sv), tmps, SvCUR(sv)); |
dfe13c55 | 346 | sv_usepvn(sv, (char*)tmps, tend - tmps); |
a0ed51b3 LW |
347 | } |
348 | return count; | |
349 | } | |
155aba94 | 350 | #endif |
a868473f | 351 | |
ffb4593c NT |
352 | /* |
353 | * Perl_lex_start | |
9cbb5ea2 GS |
354 | * Initialize variables. Uses the Perl save_stack to save its state (for |
355 | * recursive calls to the parser). | |
ffb4593c NT |
356 | */ |
357 | ||
a0d0e21e | 358 | void |
864dbfa3 | 359 | Perl_lex_start(pTHX_ SV *line) |
79072805 | 360 | { |
0f15f207 | 361 | dTHR; |
8990e307 LW |
362 | char *s; |
363 | STRLEN len; | |
364 | ||
3280af22 NIS |
365 | SAVEI32(PL_lex_dojoin); |
366 | SAVEI32(PL_lex_brackets); | |
3280af22 NIS |
367 | SAVEI32(PL_lex_casemods); |
368 | SAVEI32(PL_lex_starts); | |
369 | SAVEI32(PL_lex_state); | |
7766f137 | 370 | SAVEVPTR(PL_lex_inpat); |
3280af22 | 371 | SAVEI32(PL_lex_inwhat); |
18b09519 GS |
372 | if (PL_lex_state == LEX_KNOWNEXT) { |
373 | I32 toke = PL_nexttoke; | |
374 | while (--toke >= 0) { | |
375 | SAVEI32(PL_nexttype[toke]); | |
376 | SAVEVPTR(PL_nextval[toke]); | |
377 | } | |
378 | SAVEI32(PL_nexttoke); | |
379 | PL_nexttoke = 0; | |
380 | } | |
57843af0 | 381 | SAVECOPLINE(PL_curcop); |
3280af22 NIS |
382 | SAVEPPTR(PL_bufptr); |
383 | SAVEPPTR(PL_bufend); | |
384 | SAVEPPTR(PL_oldbufptr); | |
385 | SAVEPPTR(PL_oldoldbufptr); | |
386 | SAVEPPTR(PL_linestart); | |
387 | SAVESPTR(PL_linestr); | |
388 | SAVEPPTR(PL_lex_brackstack); | |
389 | SAVEPPTR(PL_lex_casestack); | |
c76ac1ee | 390 | SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp); |
3280af22 NIS |
391 | SAVESPTR(PL_lex_stuff); |
392 | SAVEI32(PL_lex_defer); | |
09bef843 | 393 | SAVEI32(PL_sublex_info.sub_inwhat); |
3280af22 | 394 | SAVESPTR(PL_lex_repl); |
bebdddfc GS |
395 | SAVEINT(PL_expect); |
396 | SAVEINT(PL_lex_expect); | |
3280af22 NIS |
397 | |
398 | PL_lex_state = LEX_NORMAL; | |
399 | PL_lex_defer = 0; | |
400 | PL_expect = XSTATE; | |
401 | PL_lex_brackets = 0; | |
3280af22 NIS |
402 | New(899, PL_lex_brackstack, 120, char); |
403 | New(899, PL_lex_casestack, 12, char); | |
404 | SAVEFREEPV(PL_lex_brackstack); | |
405 | SAVEFREEPV(PL_lex_casestack); | |
406 | PL_lex_casemods = 0; | |
407 | *PL_lex_casestack = '\0'; | |
408 | PL_lex_dojoin = 0; | |
409 | PL_lex_starts = 0; | |
410 | PL_lex_stuff = Nullsv; | |
411 | PL_lex_repl = Nullsv; | |
412 | PL_lex_inpat = 0; | |
413 | PL_lex_inwhat = 0; | |
09bef843 | 414 | PL_sublex_info.sub_inwhat = 0; |
3280af22 NIS |
415 | PL_linestr = line; |
416 | if (SvREADONLY(PL_linestr)) | |
417 | PL_linestr = sv_2mortal(newSVsv(PL_linestr)); | |
418 | s = SvPV(PL_linestr, len); | |
8990e307 | 419 | if (len && s[len-1] != ';') { |
3280af22 NIS |
420 | if (!(SvFLAGS(PL_linestr) & SVs_TEMP)) |
421 | PL_linestr = sv_2mortal(newSVsv(PL_linestr)); | |
422 | sv_catpvn(PL_linestr, "\n;", 2); | |
8990e307 | 423 | } |
3280af22 NIS |
424 | SvTEMP_off(PL_linestr); |
425 | PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr); | |
426 | PL_bufend = PL_bufptr + SvCUR(PL_linestr); | |
427 | SvREFCNT_dec(PL_rs); | |
79cb57f6 | 428 | PL_rs = newSVpvn("\n", 1); |
3280af22 | 429 | PL_rsfp = 0; |
79072805 | 430 | } |
a687059c | 431 | |
ffb4593c NT |
432 | /* |
433 | * Perl_lex_end | |
9cbb5ea2 GS |
434 | * Finalizer for lexing operations. Must be called when the parser is |
435 | * done with the lexer. | |
ffb4593c NT |
436 | */ |
437 | ||
463ee0b2 | 438 | void |
864dbfa3 | 439 | Perl_lex_end(pTHX) |
463ee0b2 | 440 | { |
3280af22 | 441 | PL_doextract = FALSE; |
463ee0b2 LW |
442 | } |
443 | ||
ffb4593c NT |
444 | /* |
445 | * S_incline | |
446 | * This subroutine has nothing to do with tilting, whether at windmills | |
447 | * or pinball tables. Its name is short for "increment line". It | |
57843af0 | 448 | * increments the current line number in CopLINE(PL_curcop) and checks |
ffb4593c | 449 | * to see whether the line starts with a comment of the form |
9cbb5ea2 GS |
450 | * # line 500 "foo.pm" |
451 | * If so, it sets the current line number and file to the values in the comment. | |
ffb4593c NT |
452 | */ |
453 | ||
76e3520e | 454 | STATIC void |
cea2e8a9 | 455 | S_incline(pTHX_ char *s) |
463ee0b2 | 456 | { |
0f15f207 | 457 | dTHR; |
463ee0b2 LW |
458 | char *t; |
459 | char *n; | |
73659bf1 | 460 | char *e; |
463ee0b2 | 461 | char ch; |
463ee0b2 | 462 | |
57843af0 | 463 | CopLINE_inc(PL_curcop); |
463ee0b2 LW |
464 | if (*s++ != '#') |
465 | return; | |
466 | while (*s == ' ' || *s == '\t') s++; | |
73659bf1 GS |
467 | if (strnEQ(s, "line", 4)) |
468 | s += 4; | |
469 | else | |
470 | return; | |
471 | if (*s == ' ' || *s == '\t') | |
472 | s++; | |
473 | else | |
474 | return; | |
475 | while (*s == ' ' || *s == '\t') s++; | |
463ee0b2 LW |
476 | if (!isDIGIT(*s)) |
477 | return; | |
478 | n = s; | |
479 | while (isDIGIT(*s)) | |
480 | s++; | |
481 | while (*s == ' ' || *s == '\t') | |
482 | s++; | |
73659bf1 | 483 | if (*s == '"' && (t = strchr(s+1, '"'))) { |
463ee0b2 | 484 | s++; |
73659bf1 GS |
485 | e = t + 1; |
486 | } | |
463ee0b2 | 487 | else { |
463ee0b2 | 488 | for (t = s; !isSPACE(*t); t++) ; |
73659bf1 | 489 | e = t; |
463ee0b2 | 490 | } |
73659bf1 GS |
491 | while (*e == ' ' || *e == '\t' || *e == '\r' || *e == '\f') |
492 | e++; | |
493 | if (*e != '\n' && *e != '\0') | |
494 | return; /* false alarm */ | |
495 | ||
463ee0b2 LW |
496 | ch = *t; |
497 | *t = '\0'; | |
498 | if (t - s > 0) | |
57843af0 | 499 | CopFILE_set(PL_curcop, s); |
463ee0b2 | 500 | *t = ch; |
57843af0 | 501 | CopLINE_set(PL_curcop, atoi(n)-1); |
463ee0b2 LW |
502 | } |
503 | ||
ffb4593c NT |
504 | /* |
505 | * S_skipspace | |
506 | * Called to gobble the appropriate amount and type of whitespace. | |
507 | * Skips comments as well. | |
508 | */ | |
509 | ||
76e3520e | 510 | STATIC char * |
cea2e8a9 | 511 | S_skipspace(pTHX_ register char *s) |
a687059c | 512 | { |
11343788 | 513 | dTHR; |
3280af22 NIS |
514 | if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) { |
515 | while (s < PL_bufend && (*s == ' ' || *s == '\t')) | |
463ee0b2 LW |
516 | s++; |
517 | return s; | |
518 | } | |
519 | for (;;) { | |
fd049845 | 520 | STRLEN prevlen; |
09bef843 SB |
521 | SSize_t oldprevlen, oldoldprevlen; |
522 | SSize_t oldloplen, oldunilen; | |
60e6418e GS |
523 | while (s < PL_bufend && isSPACE(*s)) { |
524 | if (*s++ == '\n' && PL_in_eval && !PL_rsfp) | |
525 | incline(s); | |
526 | } | |
ffb4593c NT |
527 | |
528 | /* comment */ | |
3280af22 NIS |
529 | if (s < PL_bufend && *s == '#') { |
530 | while (s < PL_bufend && *s != '\n') | |
463ee0b2 | 531 | s++; |
60e6418e | 532 | if (s < PL_bufend) { |
463ee0b2 | 533 | s++; |
60e6418e GS |
534 | if (PL_in_eval && !PL_rsfp) { |
535 | incline(s); | |
536 | continue; | |
537 | } | |
538 | } | |
463ee0b2 | 539 | } |
ffb4593c NT |
540 | |
541 | /* only continue to recharge the buffer if we're at the end | |
542 | * of the buffer, we're not reading from a source filter, and | |
543 | * we're in normal lexing mode | |
544 | */ | |
09bef843 SB |
545 | if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat || |
546 | PL_lex_state == LEX_FORMLINE) | |
463ee0b2 | 547 | return s; |
ffb4593c NT |
548 | |
549 | /* try to recharge the buffer */ | |
9cbb5ea2 GS |
550 | if ((s = filter_gets(PL_linestr, PL_rsfp, |
551 | (prevlen = SvCUR(PL_linestr)))) == Nullch) | |
552 | { | |
553 | /* end of file. Add on the -p or -n magic */ | |
3280af22 NIS |
554 | if (PL_minus_n || PL_minus_p) { |
555 | sv_setpv(PL_linestr,PL_minus_p ? | |
08e9d68e DD |
556 | ";}continue{print or die qq(-p destination: $!\\n)" : |
557 | ""); | |
3280af22 NIS |
558 | sv_catpv(PL_linestr,";}"); |
559 | PL_minus_n = PL_minus_p = 0; | |
a0d0e21e LW |
560 | } |
561 | else | |
3280af22 | 562 | sv_setpv(PL_linestr,";"); |
ffb4593c NT |
563 | |
564 | /* reset variables for next time we lex */ | |
9cbb5ea2 GS |
565 | PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart |
566 | = SvPVX(PL_linestr); | |
3280af22 | 567 | PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); |
ffb4593c NT |
568 | |
569 | /* Close the filehandle. Could be from -P preprocessor, | |
570 | * STDIN, or a regular file. If we were reading code from | |
571 | * STDIN (because the commandline held no -e or filename) | |
572 | * then we don't close it, we reset it so the code can | |
573 | * read from STDIN too. | |
574 | */ | |
575 | ||
3280af22 NIS |
576 | if (PL_preprocess && !PL_in_eval) |
577 | (void)PerlProc_pclose(PL_rsfp); | |
578 | else if ((PerlIO*)PL_rsfp == PerlIO_stdin()) | |
579 | PerlIO_clearerr(PL_rsfp); | |
8990e307 | 580 | else |
3280af22 NIS |
581 | (void)PerlIO_close(PL_rsfp); |
582 | PL_rsfp = Nullfp; | |
463ee0b2 LW |
583 | return s; |
584 | } | |
ffb4593c NT |
585 | |
586 | /* not at end of file, so we only read another line */ | |
09bef843 SB |
587 | /* make corresponding updates to old pointers, for yyerror() */ |
588 | oldprevlen = PL_oldbufptr - PL_bufend; | |
589 | oldoldprevlen = PL_oldoldbufptr - PL_bufend; | |
590 | if (PL_last_uni) | |
591 | oldunilen = PL_last_uni - PL_bufend; | |
592 | if (PL_last_lop) | |
593 | oldloplen = PL_last_lop - PL_bufend; | |
3280af22 NIS |
594 | PL_linestart = PL_bufptr = s + prevlen; |
595 | PL_bufend = s + SvCUR(PL_linestr); | |
596 | s = PL_bufptr; | |
09bef843 SB |
597 | PL_oldbufptr = s + oldprevlen; |
598 | PL_oldoldbufptr = s + oldoldprevlen; | |
599 | if (PL_last_uni) | |
600 | PL_last_uni = s + oldunilen; | |
601 | if (PL_last_lop) | |
602 | PL_last_lop = s + oldloplen; | |
a0d0e21e | 603 | incline(s); |
ffb4593c NT |
604 | |
605 | /* debugger active and we're not compiling the debugger code, | |
606 | * so store the line into the debugger's array of lines | |
607 | */ | |
3280af22 | 608 | if (PERLDB_LINE && PL_curstash != PL_debstash) { |
8990e307 LW |
609 | SV *sv = NEWSV(85,0); |
610 | ||
611 | sv_upgrade(sv, SVt_PVMG); | |
3280af22 | 612 | sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr); |
57843af0 | 613 | av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv); |
8990e307 | 614 | } |
463ee0b2 | 615 | } |
a687059c | 616 | } |
378cc40b | 617 | |
ffb4593c NT |
618 | /* |
619 | * S_check_uni | |
620 | * Check the unary operators to ensure there's no ambiguity in how they're | |
621 | * used. An ambiguous piece of code would be: | |
622 | * rand + 5 | |
623 | * This doesn't mean rand() + 5. Because rand() is a unary operator, | |
624 | * the +5 is its argument. | |
625 | */ | |
626 | ||
76e3520e | 627 | STATIC void |
cea2e8a9 | 628 | S_check_uni(pTHX) |
ba106d47 | 629 | { |
2f3197b3 | 630 | char *s; |
a0d0e21e | 631 | char *t; |
0453d815 | 632 | dTHR; |
2f3197b3 | 633 | |
3280af22 | 634 | if (PL_oldoldbufptr != PL_last_uni) |
2f3197b3 | 635 | return; |
3280af22 NIS |
636 | while (isSPACE(*PL_last_uni)) |
637 | PL_last_uni++; | |
7e2040f0 | 638 | for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ; |
3280af22 | 639 | if ((t = strchr(s, '(')) && t < PL_bufptr) |
a0d0e21e | 640 | return; |
0453d815 | 641 | if (ckWARN_d(WARN_AMBIGUOUS)){ |
f248d071 | 642 | char ch = *s; |
0453d815 PM |
643 | *s = '\0'; |
644 | Perl_warner(aTHX_ WARN_AMBIGUOUS, | |
645 | "Warning: Use of \"%s\" without parens is ambiguous", | |
646 | PL_last_uni); | |
647 | *s = ch; | |
648 | } | |
2f3197b3 LW |
649 | } |
650 | ||
ffb4593c NT |
651 | /* workaround to replace the UNI() macro with a function. Only the |
652 | * hints/uts.sh file mentions this. Other comments elsewhere in the | |
653 | * source indicate Microport Unix might need it too. | |
654 | */ | |
655 | ||
ffed7fef LW |
656 | #ifdef CRIPPLED_CC |
657 | ||
658 | #undef UNI | |
ffed7fef | 659 | #define UNI(f) return uni(f,s) |
ffed7fef | 660 | |
76e3520e | 661 | STATIC int |
cea2e8a9 | 662 | S_uni(pTHX_ I32 f, char *s) |
ffed7fef LW |
663 | { |
664 | yylval.ival = f; | |
3280af22 NIS |
665 | PL_expect = XTERM; |
666 | PL_bufptr = s; | |
8f872242 NIS |
667 | PL_last_uni = PL_oldbufptr; |
668 | PL_last_lop_op = f; | |
ffed7fef LW |
669 | if (*s == '(') |
670 | return FUNC1; | |
671 | s = skipspace(s); | |
672 | if (*s == '(') | |
673 | return FUNC1; | |
674 | else | |
675 | return UNIOP; | |
676 | } | |
677 | ||
a0d0e21e LW |
678 | #endif /* CRIPPLED_CC */ |
679 | ||
ffb4593c NT |
680 | /* |
681 | * LOP : macro to build a list operator. Its behaviour has been replaced | |
682 | * with a subroutine, S_lop() for which LOP is just another name. | |
683 | */ | |
684 | ||
a0d0e21e LW |
685 | #define LOP(f,x) return lop(f,x,s) |
686 | ||
ffb4593c NT |
687 | /* |
688 | * S_lop | |
689 | * Build a list operator (or something that might be one). The rules: | |
690 | * - if we have a next token, then it's a list operator [why?] | |
691 | * - if the next thing is an opening paren, then it's a function | |
692 | * - else it's a list operator | |
693 | */ | |
694 | ||
76e3520e | 695 | STATIC I32 |
a0be28da | 696 | S_lop(pTHX_ I32 f, int x, char *s) |
ffed7fef | 697 | { |
0f15f207 | 698 | dTHR; |
79072805 | 699 | yylval.ival = f; |
35c8bce7 | 700 | CLINE; |
3280af22 NIS |
701 | PL_expect = x; |
702 | PL_bufptr = s; | |
703 | PL_last_lop = PL_oldbufptr; | |
704 | PL_last_lop_op = f; | |
705 | if (PL_nexttoke) | |
a0d0e21e | 706 | return LSTOP; |
79072805 LW |
707 | if (*s == '(') |
708 | return FUNC; | |
709 | s = skipspace(s); | |
710 | if (*s == '(') | |
711 | return FUNC; | |
712 | else | |
713 | return LSTOP; | |
714 | } | |
715 | ||
ffb4593c NT |
716 | /* |
717 | * S_force_next | |
9cbb5ea2 | 718 | * When the lexer realizes it knows the next token (for instance, |
ffb4593c | 719 | * it is reordering tokens for the parser) then it can call S_force_next |
9cbb5ea2 GS |
720 | * to know what token to return the next time the lexer is called. Caller |
721 | * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer | |
722 | * handles the token correctly. | |
ffb4593c NT |
723 | */ |
724 | ||
76e3520e | 725 | STATIC void |
cea2e8a9 | 726 | S_force_next(pTHX_ I32 type) |
79072805 | 727 | { |
3280af22 NIS |
728 | PL_nexttype[PL_nexttoke] = type; |
729 | PL_nexttoke++; | |
730 | if (PL_lex_state != LEX_KNOWNEXT) { | |
731 | PL_lex_defer = PL_lex_state; | |
732 | PL_lex_expect = PL_expect; | |
733 | PL_lex_state = LEX_KNOWNEXT; | |
79072805 LW |
734 | } |
735 | } | |
736 | ||
ffb4593c NT |
737 | /* |
738 | * S_force_word | |
739 | * When the lexer knows the next thing is a word (for instance, it has | |
740 | * just seen -> and it knows that the next char is a word char, then | |
741 | * it calls S_force_word to stick the next word into the PL_next lookahead. | |
742 | * | |
743 | * Arguments: | |
b1b65b59 | 744 | * char *start : buffer position (must be within PL_linestr) |
ffb4593c NT |
745 | * int token : PL_next will be this type of bare word (e.g., METHOD,WORD) |
746 | * int check_keyword : if true, Perl checks to make sure the word isn't | |
747 | * a keyword (do this if the word is a label, e.g. goto FOO) | |
748 | * int allow_pack : if true, : characters will also be allowed (require, | |
749 | * use, etc. do this) | |
9cbb5ea2 | 750 | * int allow_initial_tick : used by the "sub" lexer only. |
ffb4593c NT |
751 | */ |
752 | ||
76e3520e | 753 | STATIC char * |
cea2e8a9 | 754 | S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick) |
79072805 | 755 | { |
463ee0b2 LW |
756 | register char *s; |
757 | STRLEN len; | |
758 | ||
759 | start = skipspace(start); | |
760 | s = start; | |
7e2040f0 | 761 | if (isIDFIRST_lazy_if(s,UTF) || |
a0d0e21e | 762 | (allow_pack && *s == ':') || |
15f0808c | 763 | (allow_initial_tick && *s == '\'') ) |
a0d0e21e | 764 | { |
3280af22 NIS |
765 | s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len); |
766 | if (check_keyword && keyword(PL_tokenbuf, len)) | |
463ee0b2 LW |
767 | return start; |
768 | if (token == METHOD) { | |
769 | s = skipspace(s); | |
770 | if (*s == '(') | |
3280af22 | 771 | PL_expect = XTERM; |
463ee0b2 | 772 | else { |
3280af22 | 773 | PL_expect = XOPERATOR; |
463ee0b2 | 774 | } |
79072805 | 775 | } |
3280af22 NIS |
776 | PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0)); |
777 | PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE; | |
79072805 LW |
778 | force_next(token); |
779 | } | |
780 | return s; | |
781 | } | |
782 | ||
ffb4593c NT |
783 | /* |
784 | * S_force_ident | |
9cbb5ea2 | 785 | * Called when the lexer wants $foo *foo &foo etc, but the program |
ffb4593c NT |
786 | * text only contains the "foo" portion. The first argument is a pointer |
787 | * to the "foo", and the second argument is the type symbol to prefix. | |
788 | * Forces the next token to be a "WORD". | |
9cbb5ea2 | 789 | * Creates the symbol if it didn't already exist (via gv_fetchpv()). |
ffb4593c NT |
790 | */ |
791 | ||
76e3520e | 792 | STATIC void |
cea2e8a9 | 793 | S_force_ident(pTHX_ register char *s, int kind) |
79072805 LW |
794 | { |
795 | if (s && *s) { | |
11343788 | 796 | OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0)); |
3280af22 | 797 | PL_nextval[PL_nexttoke].opval = o; |
79072805 | 798 | force_next(WORD); |
748a9306 | 799 | if (kind) { |
e858de61 | 800 | dTHR; /* just for in_eval */ |
11343788 | 801 | o->op_private = OPpCONST_ENTERED; |
55497cff | 802 | /* XXX see note in pp_entereval() for why we forgo typo |
803 | warnings if the symbol must be introduced in an eval. | |
804 | GSAR 96-10-12 */ | |
3280af22 | 805 | gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE, |
a0d0e21e LW |
806 | kind == '$' ? SVt_PV : |
807 | kind == '@' ? SVt_PVAV : | |
808 | kind == '%' ? SVt_PVHV : | |
809 | SVt_PVGV | |
810 | ); | |
748a9306 | 811 | } |
79072805 LW |
812 | } |
813 | } | |
814 | ||
1571675a GS |
815 | NV |
816 | Perl_str_to_version(pTHX_ SV *sv) | |
817 | { | |
818 | NV retval = 0.0; | |
819 | NV nshift = 1.0; | |
820 | STRLEN len; | |
821 | char *start = SvPVx(sv,len); | |
3aa33fe5 | 822 | bool utf = SvUTF8(sv) ? TRUE : FALSE; |
1571675a GS |
823 | char *end = start + len; |
824 | while (start < end) { | |
825 | I32 skip; | |
826 | UV n; | |
827 | if (utf) | |
828 | n = utf8_to_uv((U8*)start, &skip); | |
829 | else { | |
830 | n = *(U8*)start; | |
831 | skip = 1; | |
832 | } | |
833 | retval += ((NV)n)/nshift; | |
834 | start += skip; | |
835 | nshift *= 1000; | |
836 | } | |
837 | return retval; | |
838 | } | |
839 | ||
ffb4593c NT |
840 | /* |
841 | * S_force_version | |
842 | * Forces the next token to be a version number. | |
843 | */ | |
844 | ||
76e3520e | 845 | STATIC char * |
cea2e8a9 | 846 | S_force_version(pTHX_ char *s) |
89bfa8cd | 847 | { |
848 | OP *version = Nullop; | |
44dcb63b | 849 | char *d; |
89bfa8cd | 850 | |
851 | s = skipspace(s); | |
852 | ||
44dcb63b | 853 | d = s; |
dd629d5b | 854 | if (*d == 'v') |
44dcb63b | 855 | d++; |
44dcb63b | 856 | if (isDIGIT(*d)) { |
a7cb1f99 | 857 | for (; isDIGIT(*d) || *d == '_' || *d == '.'; d++); |
9f3d182e | 858 | if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) { |
dd629d5b | 859 | SV *ver; |
89bfa8cd | 860 | s = scan_num(s); |
89bfa8cd | 861 | version = yylval.opval; |
dd629d5b GS |
862 | ver = cSVOPx(version)->op_sv; |
863 | if (SvPOK(ver) && !SvNIOK(ver)) { | |
155aba94 | 864 | (void)SvUPGRADE(ver, SVt_PVNV); |
1571675a GS |
865 | SvNVX(ver) = str_to_version(ver); |
866 | SvNOK_on(ver); /* hint that it is a version */ | |
44dcb63b | 867 | } |
89bfa8cd | 868 | } |
869 | } | |
870 | ||
871 | /* NOTE: The parser sees the package name and the VERSION swapped */ | |
3280af22 | 872 | PL_nextval[PL_nexttoke].opval = version; |
89bfa8cd | 873 | force_next(WORD); |
874 | ||
875 | return (s); | |
876 | } | |
877 | ||
ffb4593c NT |
878 | /* |
879 | * S_tokeq | |
880 | * Tokenize a quoted string passed in as an SV. It finds the next | |
881 | * chunk, up to end of string or a backslash. It may make a new | |
882 | * SV containing that chunk (if HINT_NEW_STRING is on). It also | |
883 | * turns \\ into \. | |
884 | */ | |
885 | ||
76e3520e | 886 | STATIC SV * |
cea2e8a9 | 887 | S_tokeq(pTHX_ SV *sv) |
79072805 LW |
888 | { |
889 | register char *s; | |
890 | register char *send; | |
891 | register char *d; | |
b3ac6de7 IZ |
892 | STRLEN len = 0; |
893 | SV *pv = sv; | |
79072805 LW |
894 | |
895 | if (!SvLEN(sv)) | |
b3ac6de7 | 896 | goto finish; |
79072805 | 897 | |
a0d0e21e | 898 | s = SvPV_force(sv, len); |
21a311ee | 899 | if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1) |
b3ac6de7 | 900 | goto finish; |
463ee0b2 | 901 | send = s + len; |
79072805 LW |
902 | while (s < send && *s != '\\') |
903 | s++; | |
904 | if (s == send) | |
b3ac6de7 | 905 | goto finish; |
79072805 | 906 | d = s; |
3280af22 | 907 | if ( PL_hints & HINT_NEW_STRING ) |
79cb57f6 | 908 | pv = sv_2mortal(newSVpvn(SvPVX(pv), len)); |
79072805 LW |
909 | while (s < send) { |
910 | if (*s == '\\') { | |
a0d0e21e | 911 | if (s + 1 < send && (s[1] == '\\')) |
79072805 LW |
912 | s++; /* all that, just for this */ |
913 | } | |
914 | *d++ = *s++; | |
915 | } | |
916 | *d = '\0'; | |
463ee0b2 | 917 | SvCUR_set(sv, d - SvPVX(sv)); |
b3ac6de7 | 918 | finish: |
3280af22 | 919 | if ( PL_hints & HINT_NEW_STRING ) |
b3ac6de7 | 920 | return new_constant(NULL, 0, "q", sv, pv, "q"); |
79072805 LW |
921 | return sv; |
922 | } | |
923 | ||
ffb4593c NT |
924 | /* |
925 | * Now come three functions related to double-quote context, | |
926 | * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when | |
927 | * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They | |
928 | * interact with PL_lex_state, and create fake ( ... ) argument lists | |
929 | * to handle functions and concatenation. | |
930 | * They assume that whoever calls them will be setting up a fake | |
931 | * join call, because each subthing puts a ',' after it. This lets | |
932 | * "lower \luPpEr" | |
933 | * become | |
934 | * join($, , 'lower ', lcfirst( 'uPpEr', ) ,) | |
935 | * | |
936 | * (I'm not sure whether the spurious commas at the end of lcfirst's | |
937 | * arguments and join's arguments are created or not). | |
938 | */ | |
939 | ||
940 | /* | |
941 | * S_sublex_start | |
942 | * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST). | |
943 | * | |
944 | * Pattern matching will set PL_lex_op to the pattern-matching op to | |
945 | * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise). | |
946 | * | |
947 | * OP_CONST and OP_READLINE are easy--just make the new op and return. | |
948 | * | |
949 | * Everything else becomes a FUNC. | |
950 | * | |
951 | * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we | |
952 | * had an OP_CONST or OP_READLINE). This just sets us up for a | |
953 | * call to S_sublex_push(). | |
954 | */ | |
955 | ||
76e3520e | 956 | STATIC I32 |
cea2e8a9 | 957 | S_sublex_start(pTHX) |
79072805 LW |
958 | { |
959 | register I32 op_type = yylval.ival; | |
79072805 LW |
960 | |
961 | if (op_type == OP_NULL) { | |
3280af22 NIS |
962 | yylval.opval = PL_lex_op; |
963 | PL_lex_op = Nullop; | |
79072805 LW |
964 | return THING; |
965 | } | |
966 | if (op_type == OP_CONST || op_type == OP_READLINE) { | |
3280af22 | 967 | SV *sv = tokeq(PL_lex_stuff); |
b3ac6de7 IZ |
968 | |
969 | if (SvTYPE(sv) == SVt_PVIV) { | |
970 | /* Overloaded constants, nothing fancy: Convert to SVt_PV: */ | |
971 | STRLEN len; | |
972 | char *p; | |
973 | SV *nsv; | |
974 | ||
975 | p = SvPV(sv, len); | |
79cb57f6 | 976 | nsv = newSVpvn(p, len); |
b3ac6de7 IZ |
977 | SvREFCNT_dec(sv); |
978 | sv = nsv; | |
979 | } | |
980 | yylval.opval = (OP*)newSVOP(op_type, 0, sv); | |
3280af22 | 981 | PL_lex_stuff = Nullsv; |
79072805 LW |
982 | return THING; |
983 | } | |
984 | ||
3280af22 NIS |
985 | PL_sublex_info.super_state = PL_lex_state; |
986 | PL_sublex_info.sub_inwhat = op_type; | |
987 | PL_sublex_info.sub_op = PL_lex_op; | |
988 | PL_lex_state = LEX_INTERPPUSH; | |
55497cff | 989 | |
3280af22 NIS |
990 | PL_expect = XTERM; |
991 | if (PL_lex_op) { | |
992 | yylval.opval = PL_lex_op; | |
993 | PL_lex_op = Nullop; | |
55497cff | 994 | return PMFUNC; |
995 | } | |
996 | else | |
997 | return FUNC; | |
998 | } | |
999 | ||
ffb4593c NT |
1000 | /* |
1001 | * S_sublex_push | |
1002 | * Create a new scope to save the lexing state. The scope will be | |
1003 | * ended in S_sublex_done. Returns a '(', starting the function arguments | |
1004 | * to the uc, lc, etc. found before. | |
1005 | * Sets PL_lex_state to LEX_INTERPCONCAT. | |
1006 | */ | |
1007 | ||
76e3520e | 1008 | STATIC I32 |
cea2e8a9 | 1009 | S_sublex_push(pTHX) |
55497cff | 1010 | { |
0f15f207 | 1011 | dTHR; |
f46d017c | 1012 | ENTER; |
55497cff | 1013 | |
3280af22 NIS |
1014 | PL_lex_state = PL_sublex_info.super_state; |
1015 | SAVEI32(PL_lex_dojoin); | |
1016 | SAVEI32(PL_lex_brackets); | |
3280af22 NIS |
1017 | SAVEI32(PL_lex_casemods); |
1018 | SAVEI32(PL_lex_starts); | |
1019 | SAVEI32(PL_lex_state); | |
7766f137 | 1020 | SAVEVPTR(PL_lex_inpat); |
3280af22 | 1021 | SAVEI32(PL_lex_inwhat); |
57843af0 | 1022 | SAVECOPLINE(PL_curcop); |
3280af22 NIS |
1023 | SAVEPPTR(PL_bufptr); |
1024 | SAVEPPTR(PL_oldbufptr); | |
1025 | SAVEPPTR(PL_oldoldbufptr); | |
1026 | SAVEPPTR(PL_linestart); | |
1027 | SAVESPTR(PL_linestr); | |
1028 | SAVEPPTR(PL_lex_brackstack); | |
1029 | SAVEPPTR(PL_lex_casestack); | |
1030 | ||
1031 | PL_linestr = PL_lex_stuff; | |
1032 | PL_lex_stuff = Nullsv; | |
1033 | ||
9cbb5ea2 GS |
1034 | PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart |
1035 | = SvPVX(PL_linestr); | |
3280af22 NIS |
1036 | PL_bufend += SvCUR(PL_linestr); |
1037 | SAVEFREESV(PL_linestr); | |
1038 | ||
1039 | PL_lex_dojoin = FALSE; | |
1040 | PL_lex_brackets = 0; | |
3280af22 NIS |
1041 | New(899, PL_lex_brackstack, 120, char); |
1042 | New(899, PL_lex_casestack, 12, char); | |
1043 | SAVEFREEPV(PL_lex_brackstack); | |
1044 | SAVEFREEPV(PL_lex_casestack); | |
1045 | PL_lex_casemods = 0; | |
1046 | *PL_lex_casestack = '\0'; | |
1047 | PL_lex_starts = 0; | |
1048 | PL_lex_state = LEX_INTERPCONCAT; | |
57843af0 | 1049 | CopLINE_set(PL_curcop, PL_multi_start); |
3280af22 NIS |
1050 | |
1051 | PL_lex_inwhat = PL_sublex_info.sub_inwhat; | |
1052 | if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST) | |
1053 | PL_lex_inpat = PL_sublex_info.sub_op; | |
79072805 | 1054 | else |
3280af22 | 1055 | PL_lex_inpat = Nullop; |
79072805 | 1056 | |
55497cff | 1057 | return '('; |
79072805 LW |
1058 | } |
1059 | ||
ffb4593c NT |
1060 | /* |
1061 | * S_sublex_done | |
1062 | * Restores lexer state after a S_sublex_push. | |
1063 | */ | |
1064 | ||
76e3520e | 1065 | STATIC I32 |
cea2e8a9 | 1066 | S_sublex_done(pTHX) |
79072805 | 1067 | { |
3280af22 NIS |
1068 | if (!PL_lex_starts++) { |
1069 | PL_expect = XOPERATOR; | |
79cb57f6 | 1070 | yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpvn("",0)); |
79072805 LW |
1071 | return THING; |
1072 | } | |
1073 | ||
3280af22 NIS |
1074 | if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */ |
1075 | PL_lex_state = LEX_INTERPCASEMOD; | |
cea2e8a9 | 1076 | return yylex(); |
79072805 LW |
1077 | } |
1078 | ||
ffb4593c | 1079 | /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */ |
3280af22 NIS |
1080 | if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) { |
1081 | PL_linestr = PL_lex_repl; | |
1082 | PL_lex_inpat = 0; | |
1083 | PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr); | |
1084 | PL_bufend += SvCUR(PL_linestr); | |
1085 | SAVEFREESV(PL_linestr); | |
1086 | PL_lex_dojoin = FALSE; | |
1087 | PL_lex_brackets = 0; | |
3280af22 NIS |
1088 | PL_lex_casemods = 0; |
1089 | *PL_lex_casestack = '\0'; | |
1090 | PL_lex_starts = 0; | |
25da4f38 | 1091 | if (SvEVALED(PL_lex_repl)) { |
3280af22 NIS |
1092 | PL_lex_state = LEX_INTERPNORMAL; |
1093 | PL_lex_starts++; | |
e9fa98b2 HS |
1094 | /* we don't clear PL_lex_repl here, so that we can check later |
1095 | whether this is an evalled subst; that means we rely on the | |
1096 | logic to ensure sublex_done() is called again only via the | |
1097 | branch (in yylex()) that clears PL_lex_repl, else we'll loop */ | |
79072805 | 1098 | } |
e9fa98b2 | 1099 | else { |
3280af22 | 1100 | PL_lex_state = LEX_INTERPCONCAT; |
e9fa98b2 HS |
1101 | PL_lex_repl = Nullsv; |
1102 | } | |
79072805 | 1103 | return ','; |
ffed7fef LW |
1104 | } |
1105 | else { | |
f46d017c | 1106 | LEAVE; |
3280af22 NIS |
1107 | PL_bufend = SvPVX(PL_linestr); |
1108 | PL_bufend += SvCUR(PL_linestr); | |
1109 | PL_expect = XOPERATOR; | |
09bef843 | 1110 | PL_sublex_info.sub_inwhat = 0; |
79072805 | 1111 | return ')'; |
ffed7fef LW |
1112 | } |
1113 | } | |
1114 | ||
02aa26ce NT |
1115 | /* |
1116 | scan_const | |
1117 | ||
1118 | Extracts a pattern, double-quoted string, or transliteration. This | |
1119 | is terrifying code. | |
1120 | ||
3280af22 NIS |
1121 | It looks at lex_inwhat and PL_lex_inpat to find out whether it's |
1122 | processing a pattern (PL_lex_inpat is true), a transliteration | |
02aa26ce NT |
1123 | (lex_inwhat & OP_TRANS is true), or a double-quoted string. |
1124 | ||
9b599b2a GS |
1125 | Returns a pointer to the character scanned up to. Iff this is |
1126 | advanced from the start pointer supplied (ie if anything was | |
1127 | successfully parsed), will leave an OP for the substring scanned | |
1128 | in yylval. Caller must intuit reason for not parsing further | |
1129 | by looking at the next characters herself. | |
1130 | ||
02aa26ce NT |
1131 | In patterns: |
1132 | backslashes: | |
1133 | double-quoted style: \r and \n | |
1134 | regexp special ones: \D \s | |
1135 | constants: \x3 | |
1136 | backrefs: \1 (deprecated in substitution replacements) | |
1137 | case and quoting: \U \Q \E | |
1138 | stops on @ and $, but not for $ as tail anchor | |
1139 | ||
1140 | In transliterations: | |
1141 | characters are VERY literal, except for - not at the start or end | |
1142 | of the string, which indicates a range. scan_const expands the | |
1143 | range to the full set of intermediate characters. | |
1144 | ||
1145 | In double-quoted strings: | |
1146 | backslashes: | |
1147 | double-quoted style: \r and \n | |
1148 | constants: \x3 | |
1149 | backrefs: \1 (deprecated) | |
1150 | case and quoting: \U \Q \E | |
1151 | stops on @ and $ | |
1152 | ||
1153 | scan_const does *not* construct ops to handle interpolated strings. | |
1154 | It stops processing as soon as it finds an embedded $ or @ variable | |
1155 | and leaves it to the caller to work out what's going on. | |
1156 | ||
1157 | @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @:foo. | |
1158 | ||
1159 | $ in pattern could be $foo or could be tail anchor. Assumption: | |
1160 | it's a tail anchor if $ is the last thing in the string, or if it's | |
1161 | followed by one of ")| \n\t" | |
1162 | ||
1163 | \1 (backreferences) are turned into $1 | |
1164 | ||
1165 | The structure of the code is | |
1166 | while (there's a character to process) { | |
1167 | handle transliteration ranges | |
1168 | skip regexp comments | |
1169 | skip # initiated comments in //x patterns | |
1170 | check for embedded @foo | |
1171 | check for embedded scalars | |
1172 | if (backslash) { | |
1173 | leave intact backslashes from leave (below) | |
1174 | deprecate \1 in strings and sub replacements | |
1175 | handle string-changing backslashes \l \U \Q \E, etc. | |
1176 | switch (what was escaped) { | |
1177 | handle - in a transliteration (becomes a literal -) | |
1178 | handle \132 octal characters | |
1179 | handle 0x15 hex characters | |
1180 | handle \cV (control V) | |
1181 | handle printf backslashes (\f, \r, \n, etc) | |
1182 | } (end switch) | |
1183 | } (end if backslash) | |
1184 | } (end while character to read) | |
1185 | ||
1186 | */ | |
1187 | ||
76e3520e | 1188 | STATIC char * |
cea2e8a9 | 1189 | S_scan_const(pTHX_ char *start) |
79072805 | 1190 | { |
3280af22 | 1191 | register char *send = PL_bufend; /* end of the constant */ |
02aa26ce NT |
1192 | SV *sv = NEWSV(93, send - start); /* sv for the constant */ |
1193 | register char *s = start; /* start of the constant */ | |
1194 | register char *d = SvPVX(sv); /* destination for copies */ | |
1195 | bool dorange = FALSE; /* are we in a translit range? */ | |
7e2040f0 | 1196 | bool has_utf = FALSE; /* embedded \x{} */ |
02aa26ce | 1197 | I32 len; /* ? */ |
012bcf8d GS |
1198 | UV uv; |
1199 | ||
ac2262e3 | 1200 | I32 utf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) |
a0ed51b3 LW |
1201 | ? (PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF)) |
1202 | : UTF; | |
ac2262e3 | 1203 | I32 thisutf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) |
9cbb5ea2 GS |
1204 | ? (PL_sublex_info.sub_op->op_private & (PL_lex_repl ? |
1205 | OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF)) | |
a0ed51b3 | 1206 | : UTF; |
dff6d3cd | 1207 | const char *leaveit = /* set of acceptably-backslashed characters */ |
3280af22 | 1208 | PL_lex_inpat |
4a2d328f | 1209 | ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#" |
9b599b2a | 1210 | : ""; |
79072805 LW |
1211 | |
1212 | while (s < send || dorange) { | |
02aa26ce | 1213 | /* get transliterations out of the way (they're most literal) */ |
3280af22 | 1214 | if (PL_lex_inwhat == OP_TRANS) { |
02aa26ce | 1215 | /* expand a range A-Z to the full set of characters. AIE! */ |
79072805 | 1216 | if (dorange) { |
02aa26ce | 1217 | I32 i; /* current expanded character */ |
8ada0baa | 1218 | I32 min; /* first character in range */ |
02aa26ce NT |
1219 | I32 max; /* last character in range */ |
1220 | ||
1221 | i = d - SvPVX(sv); /* remember current offset */ | |
9cbb5ea2 GS |
1222 | SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */ |
1223 | d = SvPVX(sv) + i; /* refresh d after realloc */ | |
02aa26ce NT |
1224 | d -= 2; /* eat the first char and the - */ |
1225 | ||
8ada0baa JH |
1226 | min = (U8)*d; /* first char in range */ |
1227 | max = (U8)d[1]; /* last char in range */ | |
1228 | ||
1229 | #ifndef ASCIIish | |
1230 | if ((isLOWER(min) && isLOWER(max)) || | |
1231 | (isUPPER(min) && isUPPER(max))) { | |
1232 | if (isLOWER(min)) { | |
1233 | for (i = min; i <= max; i++) | |
1234 | if (isLOWER(i)) | |
1235 | *d++ = i; | |
1236 | } else { | |
1237 | for (i = min; i <= max; i++) | |
1238 | if (isUPPER(i)) | |
1239 | *d++ = i; | |
1240 | } | |
1241 | } | |
1242 | else | |
1243 | #endif | |
1244 | for (i = min; i <= max; i++) | |
1245 | *d++ = i; | |
02aa26ce NT |
1246 | |
1247 | /* mark the range as done, and continue */ | |
79072805 LW |
1248 | dorange = FALSE; |
1249 | continue; | |
1250 | } | |
02aa26ce NT |
1251 | |
1252 | /* range begins (ignore - as first or last char) */ | |
79072805 | 1253 | else if (*s == '-' && s+1 < send && s != start) { |
a0ed51b3 | 1254 | if (utf) { |
a176fa2a | 1255 | *d++ = (char)0xff; /* use illegal utf8 byte--see pmtrans */ |
a0ed51b3 LW |
1256 | s++; |
1257 | continue; | |
1258 | } | |
79072805 LW |
1259 | dorange = TRUE; |
1260 | s++; | |
1261 | } | |
1262 | } | |
02aa26ce NT |
1263 | |
1264 | /* if we get here, we're not doing a transliteration */ | |
1265 | ||
0f5d15d6 IZ |
1266 | /* skip for regexp comments /(?#comment)/ and code /(?{code})/, |
1267 | except for the last char, which will be done separately. */ | |
3280af22 | 1268 | else if (*s == '(' && PL_lex_inpat && s[1] == '?') { |
cc6b7395 IZ |
1269 | if (s[2] == '#') { |
1270 | while (s < send && *s != ')') | |
1271 | *d++ = *s++; | |
155aba94 GS |
1272 | } |
1273 | else if (s[2] == '{' /* This should match regcomp.c */ | |
1274 | || ((s[2] == 'p' || s[2] == '?') && s[3] == '{')) | |
1275 | { | |
cc6b7395 | 1276 | I32 count = 1; |
0f5d15d6 | 1277 | char *regparse = s + (s[2] == '{' ? 3 : 4); |
cc6b7395 IZ |
1278 | char c; |
1279 | ||
d9f97599 GS |
1280 | while (count && (c = *regparse)) { |
1281 | if (c == '\\' && regparse[1]) | |
1282 | regparse++; | |
cc6b7395 IZ |
1283 | else if (c == '{') |
1284 | count++; | |
1285 | else if (c == '}') | |
1286 | count--; | |
d9f97599 | 1287 | regparse++; |
cc6b7395 | 1288 | } |
5bdf89e7 IZ |
1289 | if (*regparse != ')') { |
1290 | regparse--; /* Leave one char for continuation. */ | |
cc6b7395 | 1291 | yyerror("Sequence (?{...}) not terminated or not {}-balanced"); |
5bdf89e7 | 1292 | } |
0f5d15d6 | 1293 | while (s < regparse) |
cc6b7395 IZ |
1294 | *d++ = *s++; |
1295 | } | |
748a9306 | 1296 | } |
02aa26ce NT |
1297 | |
1298 | /* likewise skip #-initiated comments in //x patterns */ | |
3280af22 NIS |
1299 | else if (*s == '#' && PL_lex_inpat && |
1300 | ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) { | |
748a9306 LW |
1301 | while (s+1 < send && *s != '\n') |
1302 | *d++ = *s++; | |
1303 | } | |
02aa26ce NT |
1304 | |
1305 | /* check for embedded arrays (@foo, @:foo, @'foo, @{foo}, @$foo) */ | |
7e2040f0 GS |
1306 | else if (*s == '@' && s[1] |
1307 | && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$", s[1]))) | |
79072805 | 1308 | break; |
02aa26ce NT |
1309 | |
1310 | /* check for embedded scalars. only stop if we're sure it's a | |
1311 | variable. | |
1312 | */ | |
79072805 | 1313 | else if (*s == '$') { |
3280af22 | 1314 | if (!PL_lex_inpat) /* not a regexp, so $ must be var */ |
79072805 | 1315 | break; |
c277df42 | 1316 | if (s + 1 < send && !strchr("()| \n\t", s[1])) |
79072805 LW |
1317 | break; /* in regexp, $ might be tail anchor */ |
1318 | } | |
02aa26ce | 1319 | |
a0ed51b3 LW |
1320 | /* (now in tr/// code again) */ |
1321 | ||
d008e5eb | 1322 | if (*s & 0x80 && thisutf) { |
012bcf8d GS |
1323 | (void)utf8_to_uv((U8*)s, &len); |
1324 | if (len == 1) { | |
1325 | /* illegal UTF8, make it valid */ | |
012bcf8d | 1326 | char *old_pvx = SvPVX(sv); |
bee7c419 GS |
1327 | /* need space for one extra char (NOTE: SvCUR() not set here) */ |
1328 | d = SvGROW(sv, SvLEN(sv) + 1) + (d - old_pvx); | |
012bcf8d GS |
1329 | d = (char*)uv_to_utf8((U8*)d, (U8)*s++); |
1330 | } | |
1331 | else { | |
1332 | while (len--) | |
1333 | *d++ = *s++; | |
1334 | } | |
1335 | has_utf = TRUE; | |
1336 | continue; | |
a0ed51b3 LW |
1337 | } |
1338 | ||
02aa26ce | 1339 | /* backslashes */ |
79072805 LW |
1340 | if (*s == '\\' && s+1 < send) { |
1341 | s++; | |
02aa26ce NT |
1342 | |
1343 | /* some backslashes we leave behind */ | |
c9f97d15 | 1344 | if (*leaveit && *s && strchr(leaveit, *s)) { |
79072805 LW |
1345 | *d++ = '\\'; |
1346 | *d++ = *s++; | |
1347 | continue; | |
1348 | } | |
02aa26ce NT |
1349 | |
1350 | /* deprecate \1 in strings and substitution replacements */ | |
3280af22 | 1351 | if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat && |
a0d0e21e | 1352 | isDIGIT(*s) && *s != '0' && !isDIGIT(s[1])) |
79072805 | 1353 | { |
d008e5eb | 1354 | dTHR; /* only for ckWARN */ |
599cee73 | 1355 | if (ckWARN(WARN_SYNTAX)) |
cea2e8a9 | 1356 | Perl_warner(aTHX_ WARN_SYNTAX, "\\%c better written as $%c", *s, *s); |
79072805 LW |
1357 | *--s = '$'; |
1358 | break; | |
1359 | } | |
02aa26ce NT |
1360 | |
1361 | /* string-change backslash escapes */ | |
3280af22 | 1362 | if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) { |
79072805 LW |
1363 | --s; |
1364 | break; | |
1365 | } | |
02aa26ce NT |
1366 | |
1367 | /* if we get here, it's either a quoted -, or a digit */ | |
79072805 | 1368 | switch (*s) { |
02aa26ce NT |
1369 | |
1370 | /* quoted - in transliterations */ | |
79072805 | 1371 | case '-': |
3280af22 | 1372 | if (PL_lex_inwhat == OP_TRANS) { |
79072805 LW |
1373 | *d++ = *s++; |
1374 | continue; | |
1375 | } | |
1376 | /* FALL THROUGH */ | |
1377 | default: | |
11b8faa4 JH |
1378 | { |
1379 | dTHR; | |
7e84c16c | 1380 | if (ckWARN(WARN_MISC) && isALNUM(*s)) |
e476b1b5 | 1381 | Perl_warner(aTHX_ WARN_MISC, |
11b8faa4 JH |
1382 | "Unrecognized escape \\%c passed through", |
1383 | *s); | |
1384 | /* default action is to copy the quoted character */ | |
1385 | *d++ = *s++; | |
1386 | continue; | |
1387 | } | |
02aa26ce NT |
1388 | |
1389 | /* \132 indicates an octal constant */ | |
79072805 LW |
1390 | case '0': case '1': case '2': case '3': |
1391 | case '4': case '5': case '6': case '7': | |
b21ed0a9 | 1392 | len = 0; /* disallow underscores */ |
012bcf8d | 1393 | uv = (UV)scan_oct(s, 3, &len); |
79072805 | 1394 | s += len; |
012bcf8d | 1395 | goto NUM_ESCAPE_INSERT; |
02aa26ce NT |
1396 | |
1397 | /* \x24 indicates a hex constant */ | |
79072805 | 1398 | case 'x': |
a0ed51b3 LW |
1399 | ++s; |
1400 | if (*s == '{') { | |
1401 | char* e = strchr(s, '}'); | |
adaeee49 | 1402 | if (!e) { |
a0ed51b3 | 1403 | yyerror("Missing right brace on \\x{}"); |
adaeee49 GA |
1404 | e = s; |
1405 | } | |
b21ed0a9 | 1406 | len = 1; /* allow underscores */ |
012bcf8d GS |
1407 | uv = (UV)scan_hex(s + 1, e - s - 1, &len); |
1408 | s = e + 1; | |
a0ed51b3 LW |
1409 | } |
1410 | else { | |
b21ed0a9 | 1411 | len = 0; /* disallow underscores */ |
012bcf8d GS |
1412 | uv = (UV)scan_hex(s, 2, &len); |
1413 | s += len; | |
1414 | } | |
1415 | ||
1416 | NUM_ESCAPE_INSERT: | |
1417 | /* Insert oct or hex escaped character. | |
1418 | * There will always enough room in sv since such escapes will | |
1419 | * be longer than any utf8 sequence they can end up as | |
1420 | */ | |
1421 | if (uv > 127) { | |
1422 | if (!thisutf && !has_utf && uv > 255) { | |
1423 | /* might need to recode whatever we have accumulated so far | |
1424 | * if it contains any hibit chars | |
1425 | */ | |
1426 | int hicount = 0; | |
1427 | char *c; | |
1428 | for (c = SvPVX(sv); c < d; c++) { | |
1429 | if (*c & 0x80) | |
1430 | hicount++; | |
1431 | } | |
1432 | if (hicount) { | |
1433 | char *old_pvx = SvPVX(sv); | |
1434 | char *src, *dst; | |
1435 | d = SvGROW(sv, SvCUR(sv) + hicount + 1) + (d - old_pvx); | |
1436 | ||
1437 | src = d - 1; | |
1438 | d += hicount; | |
1439 | dst = d - 1; | |
1440 | ||
1441 | while (src < dst) { | |
1442 | if (*src & 0x80) { | |
1443 | dst--; | |
1444 | uv_to_utf8((U8*)dst, (U8)*src--); | |
1445 | dst--; | |
1446 | } | |
1447 | else { | |
1448 | *dst-- = *src--; | |
1449 | } | |
1450 | } | |
1451 | } | |
1452 | } | |
1453 | ||
1454 | if (thisutf || uv > 255) { | |
1455 | d = (char*)uv_to_utf8((U8*)d, uv); | |
7e2040f0 | 1456 | has_utf = TRUE; |
012bcf8d | 1457 | } |
a0ed51b3 | 1458 | else { |
012bcf8d | 1459 | *d++ = (char)uv; |
a0ed51b3 | 1460 | } |
012bcf8d GS |
1461 | } |
1462 | else { | |
1463 | *d++ = (char)uv; | |
a0ed51b3 | 1464 | } |
79072805 | 1465 | continue; |
02aa26ce | 1466 | |
4a2d328f IZ |
1467 | /* \N{latin small letter a} is a named character */ |
1468 | case 'N': | |
423cee85 JH |
1469 | ++s; |
1470 | if (*s == '{') { | |
1471 | char* e = strchr(s, '}'); | |
155aba94 | 1472 | SV *res; |
423cee85 JH |
1473 | STRLEN len; |
1474 | char *str; | |
423cee85 JH |
1475 | |
1476 | if (!e) { | |
5777a3f7 | 1477 | yyerror("Missing right brace on \\N{}"); |
423cee85 JH |
1478 | e = s - 1; |
1479 | goto cont_scan; | |
1480 | } | |
1481 | res = newSVpvn(s + 1, e - s - 1); | |
1482 | res = new_constant( Nullch, 0, "charnames", | |
5777a3f7 | 1483 | res, Nullsv, "\\N{...}" ); |
423cee85 | 1484 | str = SvPV(res,len); |
f08d6ad9 GS |
1485 | if (!has_utf && SvUTF8(res)) { |
1486 | char *ostart = SvPVX(sv); | |
1487 | SvCUR_set(sv, d - ostart); | |
1488 | SvPOK_on(sv); | |
1489 | sv_utf8_upgrade(sv); | |
1490 | d = SvPVX(sv) + SvCUR(sv); | |
e1992b6d | 1491 | has_utf = TRUE; |
f08d6ad9 | 1492 | } |
423cee85 JH |
1493 | if (len > e - s + 4) { |
1494 | char *odest = SvPVX(sv); | |
1495 | ||
1496 | SvGROW(sv, (SvCUR(sv) + len - (e - s + 4))); | |
1497 | d = SvPVX(sv) + (d - odest); | |
1498 | } | |
1499 | Copy(str, d, len, char); | |
1500 | d += len; | |
1501 | SvREFCNT_dec(res); | |
1502 | cont_scan: | |
1503 | s = e + 1; | |
1504 | } | |
1505 | else | |
5777a3f7 | 1506 | yyerror("Missing braces on \\N{}"); |
423cee85 JH |
1507 | continue; |
1508 | ||
02aa26ce | 1509 | /* \c is a control character */ |
79072805 LW |
1510 | case 'c': |
1511 | s++; | |
9d116dd7 JH |
1512 | #ifdef EBCDIC |
1513 | *d = *s++; | |
1514 | if (isLOWER(*d)) | |
1515 | *d = toUPPER(*d); | |
774a9426 GS |
1516 | *d = toCTRL(*d); |
1517 | d++; | |
9d116dd7 | 1518 | #else |
bbce6d69 | 1519 | len = *s++; |
1520 | *d++ = toCTRL(len); | |
9d116dd7 | 1521 | #endif |
79072805 | 1522 | continue; |
02aa26ce NT |
1523 | |
1524 | /* printf-style backslashes, formfeeds, newlines, etc */ | |
79072805 LW |
1525 | case 'b': |
1526 | *d++ = '\b'; | |
1527 | break; | |
1528 | case 'n': | |
1529 | *d++ = '\n'; | |
1530 | break; | |
1531 | case 'r': | |
1532 | *d++ = '\r'; | |
1533 | break; | |
1534 | case 'f': | |
1535 | *d++ = '\f'; | |
1536 | break; | |
1537 | case 't': | |
1538 | *d++ = '\t'; | |
1539 | break; | |
34a3fe2a PP |
1540 | #ifdef EBCDIC |
1541 | case 'e': | |
1542 | *d++ = '\047'; /* CP 1047 */ | |
1543 | break; | |
1544 | case 'a': | |
1545 | *d++ = '\057'; /* CP 1047 */ | |
1546 | break; | |
1547 | #else | |
79072805 LW |
1548 | case 'e': |
1549 | *d++ = '\033'; | |
1550 | break; | |
1551 | case 'a': | |
1552 | *d++ = '\007'; | |
1553 | break; | |
34a3fe2a | 1554 | #endif |
02aa26ce NT |
1555 | } /* end switch */ |
1556 | ||
79072805 LW |
1557 | s++; |
1558 | continue; | |
02aa26ce NT |
1559 | } /* end if (backslash) */ |
1560 | ||
79072805 | 1561 | *d++ = *s++; |
02aa26ce NT |
1562 | } /* while loop to process each character */ |
1563 | ||
1564 | /* terminate the string and set up the sv */ | |
79072805 | 1565 | *d = '\0'; |
463ee0b2 | 1566 | SvCUR_set(sv, d - SvPVX(sv)); |
79072805 | 1567 | SvPOK_on(sv); |
7e2040f0 GS |
1568 | if (has_utf) |
1569 | SvUTF8_on(sv); | |
79072805 | 1570 | |
02aa26ce | 1571 | /* shrink the sv if we allocated more than we used */ |
79072805 LW |
1572 | if (SvCUR(sv) + 5 < SvLEN(sv)) { |
1573 | SvLEN_set(sv, SvCUR(sv) + 1); | |
463ee0b2 | 1574 | Renew(SvPVX(sv), SvLEN(sv), char); |
79072805 | 1575 | } |
02aa26ce | 1576 | |
9b599b2a | 1577 | /* return the substring (via yylval) only if we parsed anything */ |
3280af22 NIS |
1578 | if (s > PL_bufptr) { |
1579 | if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) ) | |
1580 | sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"), | |
b3ac6de7 | 1581 | sv, Nullsv, |
3280af22 | 1582 | ( PL_lex_inwhat == OP_TRANS |
b3ac6de7 | 1583 | ? "tr" |
3280af22 | 1584 | : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat) |
b3ac6de7 IZ |
1585 | ? "s" |
1586 | : "qq"))); | |
79072805 | 1587 | yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv); |
b3ac6de7 | 1588 | } else |
8990e307 | 1589 | SvREFCNT_dec(sv); |
79072805 LW |
1590 | return s; |
1591 | } | |
1592 | ||
ffb4593c NT |
1593 | /* S_intuit_more |
1594 | * Returns TRUE if there's more to the expression (e.g., a subscript), | |
1595 | * FALSE otherwise. | |
ffb4593c NT |
1596 | * |
1597 | * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/ | |
1598 | * | |
1599 | * ->[ and ->{ return TRUE | |
1600 | * { and [ outside a pattern are always subscripts, so return TRUE | |
1601 | * if we're outside a pattern and it's not { or [, then return FALSE | |
1602 | * if we're in a pattern and the first char is a { | |
1603 | * {4,5} (any digits around the comma) returns FALSE | |
1604 | * if we're in a pattern and the first char is a [ | |
1605 | * [] returns FALSE | |
1606 | * [SOMETHING] has a funky algorithm to decide whether it's a | |
1607 | * character class or not. It has to deal with things like | |
1608 | * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/ | |
1609 | * anything else returns TRUE | |
1610 | */ | |
1611 | ||
9cbb5ea2 GS |
1612 | /* This is the one truly awful dwimmer necessary to conflate C and sed. */ |
1613 | ||
76e3520e | 1614 | STATIC int |
cea2e8a9 | 1615 | S_intuit_more(pTHX_ register char *s) |
79072805 | 1616 | { |
3280af22 | 1617 | if (PL_lex_brackets) |
79072805 LW |
1618 | return TRUE; |
1619 | if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{')) | |
1620 | return TRUE; | |
1621 | if (*s != '{' && *s != '[') | |
1622 | return FALSE; | |
3280af22 | 1623 | if (!PL_lex_inpat) |
79072805 LW |
1624 | return TRUE; |
1625 | ||
1626 | /* In a pattern, so maybe we have {n,m}. */ | |
1627 | if (*s == '{') { | |
1628 | s++; | |
1629 | if (!isDIGIT(*s)) | |
1630 | return TRUE; | |
1631 | while (isDIGIT(*s)) | |
1632 | s++; | |
1633 | if (*s == ',') | |
1634 | s++; | |
1635 | while (isDIGIT(*s)) | |
1636 | s++; | |
1637 | if (*s == '}') | |
1638 | return FALSE; | |
1639 | return TRUE; | |
1640 | ||
1641 | } | |
1642 | ||
1643 | /* On the other hand, maybe we have a character class */ | |
1644 | ||
1645 | s++; | |
1646 | if (*s == ']' || *s == '^') | |
1647 | return FALSE; | |
1648 | else { | |
ffb4593c | 1649 | /* this is terrifying, and it works */ |
79072805 LW |
1650 | int weight = 2; /* let's weigh the evidence */ |
1651 | char seen[256]; | |
f27ffc4a | 1652 | unsigned char un_char = 255, last_un_char; |
93a17b20 | 1653 | char *send = strchr(s,']'); |
3280af22 | 1654 | char tmpbuf[sizeof PL_tokenbuf * 4]; |
79072805 LW |
1655 | |
1656 | if (!send) /* has to be an expression */ | |
1657 | return TRUE; | |
1658 | ||
1659 | Zero(seen,256,char); | |
1660 | if (*s == '$') | |
1661 | weight -= 3; | |
1662 | else if (isDIGIT(*s)) { | |
1663 | if (s[1] != ']') { | |
1664 | if (isDIGIT(s[1]) && s[2] == ']') | |
1665 | weight -= 10; | |
1666 | } | |
1667 | else | |
1668 | weight -= 100; | |
1669 | } | |
1670 | for (; s < send; s++) { | |
1671 | last_un_char = un_char; | |
1672 | un_char = (unsigned char)*s; | |
1673 | switch (*s) { | |
1674 | case '@': | |
1675 | case '&': | |
1676 | case '$': | |
1677 | weight -= seen[un_char] * 10; | |
7e2040f0 | 1678 | if (isALNUM_lazy_if(s+1,UTF)) { |
8903cb82 | 1679 | scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE); |
a0d0e21e | 1680 | if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV)) |
79072805 LW |
1681 | weight -= 100; |
1682 | else | |
1683 | weight -= 10; | |
1684 | } | |
1685 | else if (*s == '$' && s[1] && | |
93a17b20 LW |
1686 | strchr("[#!%*<>()-=",s[1])) { |
1687 | if (/*{*/ strchr("])} =",s[2])) | |
79072805 LW |
1688 | weight -= 10; |
1689 | else | |
1690 | weight -= 1; | |
1691 | } | |
1692 | break; | |
1693 | case '\\': | |
1694 | un_char = 254; | |
1695 | if (s[1]) { | |
93a17b20 | 1696 | if (strchr("wds]",s[1])) |
79072805 LW |
1697 | weight += 100; |
1698 | else if (seen['\''] || seen['"']) | |
1699 | weight += 1; | |
93a17b20 | 1700 | else if (strchr("rnftbxcav",s[1])) |
79072805 LW |
1701 | weight += 40; |
1702 | else if (isDIGIT(s[1])) { | |
1703 | weight += 40; | |
1704 | while (s[1] && isDIGIT(s[1])) | |
1705 | s++; | |
1706 | } | |
1707 | } | |
1708 | else | |
1709 | weight += 100; | |
1710 | break; | |
1711 | case '-': | |
1712 | if (s[1] == '\\') | |
1713 | weight += 50; | |
93a17b20 | 1714 | if (strchr("aA01! ",last_un_char)) |
79072805 | 1715 | weight += 30; |
93a17b20 | 1716 | if (strchr("zZ79~",s[1])) |
79072805 | 1717 | weight += 30; |
f27ffc4a GS |
1718 | if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$')) |
1719 | weight -= 5; /* cope with negative subscript */ | |
79072805 LW |
1720 | break; |
1721 | default: | |
93a17b20 | 1722 | if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) && |
79072805 LW |
1723 | isALPHA(*s) && s[1] && isALPHA(s[1])) { |
1724 | char *d = tmpbuf; | |
1725 | while (isALPHA(*s)) | |
1726 | *d++ = *s++; | |
1727 | *d = '\0'; | |
1728 | if (keyword(tmpbuf, d - tmpbuf)) | |
1729 | weight -= 150; | |
1730 | } | |
1731 | if (un_char == last_un_char + 1) | |
1732 | weight += 5; | |
1733 | weight -= seen[un_char]; | |
1734 | break; | |
1735 | } | |
1736 | seen[un_char]++; | |
1737 | } | |
1738 | if (weight >= 0) /* probably a character class */ | |
1739 | return FALSE; | |
1740 | } | |
1741 | ||
1742 | return TRUE; | |
1743 | } | |
ffed7fef | 1744 | |
ffb4593c NT |
1745 | /* |
1746 | * S_intuit_method | |
1747 | * | |
1748 | * Does all the checking to disambiguate | |
1749 | * foo bar | |
1750 | * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise | |
1751 | * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args). | |
1752 | * | |
1753 | * First argument is the stuff after the first token, e.g. "bar". | |
1754 | * | |
1755 | * Not a method if bar is a filehandle. | |
1756 | * Not a method if foo is a subroutine prototyped to take a filehandle. | |
1757 | * Not a method if it's really "Foo $bar" | |
1758 | * Method if it's "foo $bar" | |
1759 | * Not a method if it's really "print foo $bar" | |
1760 | * Method if it's really "foo package::" (interpreted as package->foo) | |
1761 | * Not a method if bar is known to be a subroutne ("sub bar; foo bar") | |
3cb0bbe5 | 1762 | * Not a method if bar is a filehandle or package, but is quoted with |
ffb4593c NT |
1763 | * => |
1764 | */ | |
1765 | ||
76e3520e | 1766 | STATIC int |
cea2e8a9 | 1767 | S_intuit_method(pTHX_ char *start, GV *gv) |
a0d0e21e LW |
1768 | { |
1769 | char *s = start + (*start == '$'); | |
3280af22 | 1770 | char tmpbuf[sizeof PL_tokenbuf]; |
a0d0e21e LW |
1771 | STRLEN len; |
1772 | GV* indirgv; | |
1773 | ||
1774 | if (gv) { | |
b6c543e3 | 1775 | CV *cv; |
a0d0e21e LW |
1776 | if (GvIO(gv)) |
1777 | return 0; | |
b6c543e3 IZ |
1778 | if ((cv = GvCVu(gv))) { |
1779 | char *proto = SvPVX(cv); | |
1780 | if (proto) { | |
1781 | if (*proto == ';') | |
1782 | proto++; | |
1783 | if (*proto == '*') | |
1784 | return 0; | |
1785 | } | |
1786 | } else | |
a0d0e21e LW |
1787 | gv = 0; |
1788 | } | |
8903cb82 | 1789 | s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len); |
ffb4593c NT |
1790 | /* start is the beginning of the possible filehandle/object, |
1791 | * and s is the end of it | |
1792 | * tmpbuf is a copy of it | |
1793 | */ | |
1794 | ||
a0d0e21e | 1795 | if (*start == '$') { |
3280af22 | 1796 | if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf)) |
a0d0e21e LW |
1797 | return 0; |
1798 | s = skipspace(s); | |
3280af22 NIS |
1799 | PL_bufptr = start; |
1800 | PL_expect = XREF; | |
a0d0e21e LW |
1801 | return *s == '(' ? FUNCMETH : METHOD; |
1802 | } | |
1803 | if (!keyword(tmpbuf, len)) { | |
c3e0f903 GS |
1804 | if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') { |
1805 | len -= 2; | |
1806 | tmpbuf[len] = '\0'; | |
1807 | goto bare_package; | |
1808 | } | |
1809 | indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV); | |
8ebc5c01 | 1810 | if (indirgv && GvCVu(indirgv)) |
a0d0e21e LW |
1811 | return 0; |
1812 | /* filehandle or package name makes it a method */ | |
89bfa8cd | 1813 | if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) { |
a0d0e21e | 1814 | s = skipspace(s); |
3280af22 | 1815 | if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>') |
55497cff | 1816 | return 0; /* no assumptions -- "=>" quotes bearword */ |
c3e0f903 | 1817 | bare_package: |
3280af22 | 1818 | PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, |
79cb57f6 | 1819 | newSVpvn(tmpbuf,len)); |
3280af22 NIS |
1820 | PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE; |
1821 | PL_expect = XTERM; | |
a0d0e21e | 1822 | force_next(WORD); |
3280af22 | 1823 | PL_bufptr = s; |
a0d0e21e LW |
1824 | return *s == '(' ? FUNCMETH : METHOD; |
1825 | } | |
1826 | } | |
1827 | return 0; | |
1828 | } | |
1829 | ||
ffb4593c NT |
1830 | /* |
1831 | * S_incl_perldb | |
1832 | * Return a string of Perl code to load the debugger. If PERL5DB | |
1833 | * is set, it will return the contents of that, otherwise a | |
1834 | * compile-time require of perl5db.pl. | |
1835 | */ | |
1836 | ||
76e3520e | 1837 | STATIC char* |
cea2e8a9 | 1838 | S_incl_perldb(pTHX) |
a0d0e21e | 1839 | { |
3280af22 | 1840 | if (PL_perldb) { |
76e3520e | 1841 | char *pdb = PerlEnv_getenv("PERL5DB"); |
a0d0e21e LW |
1842 | |
1843 | if (pdb) | |
1844 | return pdb; | |
61bb5906 | 1845 | SETERRNO(0,SS$_NORMAL); |
a0d0e21e LW |
1846 | return "BEGIN { require 'perl5db.pl' }"; |
1847 | } | |
1848 | return ""; | |
1849 | } | |
1850 | ||
1851 | ||
16d20bd9 AD |
1852 | /* Encoded script support. filter_add() effectively inserts a |
1853 | * 'pre-processing' function into the current source input stream. | |
1854 | * Note that the filter function only applies to the current source file | |
1855 | * (e.g., it will not affect files 'require'd or 'use'd by this one). | |
1856 | * | |
1857 | * The datasv parameter (which may be NULL) can be used to pass | |
1858 | * private data to this instance of the filter. The filter function | |
1859 | * can recover the SV using the FILTER_DATA macro and use it to | |
1860 | * store private buffers and state information. | |
1861 | * | |
1862 | * The supplied datasv parameter is upgraded to a PVIO type | |
e0c19803 GS |
1863 | * and the IoDIRP field is used to store the function pointer, |
1864 | * and IOf_FAKE_DIRP is enabled on datasv to mark this as such. | |
16d20bd9 AD |
1865 | * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for |
1866 | * private use must be set using malloc'd pointers. | |
1867 | */ | |
16d20bd9 AD |
1868 | |
1869 | SV * | |
864dbfa3 | 1870 | Perl_filter_add(pTHX_ filter_t funcp, SV *datasv) |
16d20bd9 | 1871 | { |
f4c556ac GS |
1872 | if (!funcp) |
1873 | return Nullsv; | |
1874 | ||
3280af22 NIS |
1875 | if (!PL_rsfp_filters) |
1876 | PL_rsfp_filters = newAV(); | |
16d20bd9 | 1877 | if (!datasv) |
8c52afec | 1878 | datasv = NEWSV(255,0); |
16d20bd9 | 1879 | if (!SvUPGRADE(datasv, SVt_PVIO)) |
cea2e8a9 | 1880 | Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO"); |
16d20bd9 | 1881 | IoDIRP(datasv) = (DIR*)funcp; /* stash funcp into spare field */ |
e0c19803 | 1882 | IoFLAGS(datasv) |= IOf_FAKE_DIRP; |
f4c556ac GS |
1883 | DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n", |
1884 | funcp, SvPV_nolen(datasv))); | |
3280af22 NIS |
1885 | av_unshift(PL_rsfp_filters, 1); |
1886 | av_store(PL_rsfp_filters, 0, datasv) ; | |
16d20bd9 AD |
1887 | return(datasv); |
1888 | } | |
1889 | ||
1890 | ||
1891 | /* Delete most recently added instance of this filter function. */ | |
a0d0e21e | 1892 | void |
864dbfa3 | 1893 | Perl_filter_del(pTHX_ filter_t funcp) |
16d20bd9 | 1894 | { |
e0c19803 | 1895 | SV *datasv; |
f4c556ac | 1896 | DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp)); |
3280af22 | 1897 | if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0) |
16d20bd9 AD |
1898 | return; |
1899 | /* if filter is on top of stack (usual case) just pop it off */ | |
e0c19803 GS |
1900 | datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters)); |
1901 | if (IoDIRP(datasv) == (DIR*)funcp) { | |
1902 | IoFLAGS(datasv) &= ~IOf_FAKE_DIRP; | |
1903 | IoDIRP(datasv) = (DIR*)NULL; | |
3280af22 | 1904 | sv_free(av_pop(PL_rsfp_filters)); |
e50aee73 | 1905 | |
16d20bd9 AD |
1906 | return; |
1907 | } | |
1908 | /* we need to search for the correct entry and clear it */ | |
cea2e8a9 | 1909 | Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)"); |
16d20bd9 AD |
1910 | } |
1911 | ||
1912 | ||
1913 | /* Invoke the n'th filter function for the current rsfp. */ | |
1914 | I32 | |
864dbfa3 | 1915 | Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen) |
8ac85365 NIS |
1916 | |
1917 | ||
1918 | /* 0 = read one text line */ | |
a0d0e21e | 1919 | { |
16d20bd9 AD |
1920 | filter_t funcp; |
1921 | SV *datasv = NULL; | |
e50aee73 | 1922 | |
3280af22 | 1923 | if (!PL_rsfp_filters) |
16d20bd9 | 1924 | return -1; |
3280af22 | 1925 | if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */ |
16d20bd9 AD |
1926 | /* Provide a default input filter to make life easy. */ |
1927 | /* Note that we append to the line. This is handy. */ | |
f4c556ac GS |
1928 | DEBUG_P(PerlIO_printf(Perl_debug_log, |
1929 | "filter_read %d: from rsfp\n", idx)); | |
16d20bd9 AD |
1930 | if (maxlen) { |
1931 | /* Want a block */ | |
1932 | int len ; | |
1933 | int old_len = SvCUR(buf_sv) ; | |
1934 | ||
1935 | /* ensure buf_sv is large enough */ | |
1936 | SvGROW(buf_sv, old_len + maxlen) ; | |
3280af22 NIS |
1937 | if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){ |
1938 | if (PerlIO_error(PL_rsfp)) | |
37120919 AD |
1939 | return -1; /* error */ |
1940 | else | |
1941 | return 0 ; /* end of file */ | |
1942 | } | |
16d20bd9 AD |
1943 | SvCUR_set(buf_sv, old_len + len) ; |
1944 | } else { | |
1945 | /* Want a line */ | |
3280af22 NIS |
1946 | if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) { |
1947 | if (PerlIO_error(PL_rsfp)) | |
37120919 AD |
1948 | return -1; /* error */ |
1949 | else | |
1950 | return 0 ; /* end of file */ | |
1951 | } | |
16d20bd9 AD |
1952 | } |
1953 | return SvCUR(buf_sv); | |
1954 | } | |
1955 | /* Skip this filter slot if filter has been deleted */ | |
3280af22 | 1956 | if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){ |
f4c556ac GS |
1957 | DEBUG_P(PerlIO_printf(Perl_debug_log, |
1958 | "filter_read %d: skipped (filter deleted)\n", | |
1959 | idx)); | |
16d20bd9 AD |
1960 | return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */ |
1961 | } | |
1962 | /* Get function pointer hidden within datasv */ | |
1963 | funcp = (filter_t)IoDIRP(datasv); | |
f4c556ac GS |
1964 | DEBUG_P(PerlIO_printf(Perl_debug_log, |
1965 | "filter_read %d: via function %p (%s)\n", | |
1966 | idx, funcp, SvPV_nolen(datasv))); | |
16d20bd9 AD |
1967 | /* Call function. The function is expected to */ |
1968 | /* call "FILTER_READ(idx+1, buf_sv)" first. */ | |
37120919 | 1969 | /* Return: <0:error, =0:eof, >0:not eof */ |
0cb96387 | 1970 | return (*funcp)(aTHXo_ idx, buf_sv, maxlen); |
16d20bd9 AD |
1971 | } |
1972 | ||
76e3520e | 1973 | STATIC char * |
cea2e8a9 | 1974 | S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append) |
16d20bd9 | 1975 | { |
c39cd008 | 1976 | #ifdef PERL_CR_FILTER |
3280af22 | 1977 | if (!PL_rsfp_filters) { |
c39cd008 | 1978 | filter_add(S_cr_textfilter,NULL); |
a868473f NIS |
1979 | } |
1980 | #endif | |
3280af22 | 1981 | if (PL_rsfp_filters) { |
16d20bd9 | 1982 | |
55497cff | 1983 | if (!append) |
1984 | SvCUR_set(sv, 0); /* start with empty line */ | |
16d20bd9 AD |
1985 | if (FILTER_READ(0, sv, 0) > 0) |
1986 | return ( SvPVX(sv) ) ; | |
1987 | else | |
1988 | return Nullch ; | |
1989 | } | |
9d116dd7 | 1990 | else |
fd049845 | 1991 | return (sv_gets(sv, fp, append)); |
a0d0e21e LW |
1992 | } |
1993 | ||
1994 | ||
748a9306 LW |
1995 | #ifdef DEBUGGING |
1996 | static char* exp_name[] = | |
09bef843 SB |
1997 | { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK", |
1998 | "ATTRTERM", "TERMBLOCK" | |
1999 | }; | |
748a9306 | 2000 | #endif |
463ee0b2 | 2001 | |
02aa26ce NT |
2002 | /* |
2003 | yylex | |
2004 | ||
2005 | Works out what to call the token just pulled out of the input | |
2006 | stream. The yacc parser takes care of taking the ops we return and | |
2007 | stitching them into a tree. | |
2008 | ||
2009 | Returns: | |
2010 | PRIVATEREF | |
2011 | ||
2012 | Structure: | |
2013 | if read an identifier | |
2014 | if we're in a my declaration | |
2015 | croak if they tried to say my($foo::bar) | |
2016 | build the ops for a my() declaration | |
2017 | if it's an access to a my() variable | |
2018 | are we in a sort block? | |
2019 | croak if my($a); $a <=> $b | |
2020 | build ops for access to a my() variable | |
2021 | if in a dq string, and they've said @foo and we can't find @foo | |
2022 | croak | |
2023 | build ops for a bareword | |
2024 | if we already built the token before, use it. | |
2025 | */ | |
2026 | ||
864dbfa3 GS |
2027 | int |
2028 | #ifdef USE_PURE_BISON | |
cea2e8a9 | 2029 | Perl_yylex(pTHX_ YYSTYPE *lvalp, int *lcharp) |
864dbfa3 | 2030 | #else |
cea2e8a9 | 2031 | Perl_yylex(pTHX) |
864dbfa3 | 2032 | #endif |
378cc40b | 2033 | { |
11343788 | 2034 | dTHR; |
79072805 | 2035 | register char *s; |
378cc40b | 2036 | register char *d; |
79072805 | 2037 | register I32 tmp; |
463ee0b2 | 2038 | STRLEN len; |
161b471a NIS |
2039 | GV *gv = Nullgv; |
2040 | GV **gvp = 0; | |
a687059c | 2041 | |
a1a0e61e TD |
2042 | #ifdef USE_PURE_BISON |
2043 | yylval_pointer = lvalp; | |
2044 | yychar_pointer = lcharp; | |
2045 | #endif | |
2046 | ||
02aa26ce | 2047 | /* check if there's an identifier for us to look at */ |
3280af22 | 2048 | if (PL_pending_ident) { |
02aa26ce | 2049 | /* pit holds the identifier we read and pending_ident is reset */ |
3280af22 NIS |
2050 | char pit = PL_pending_ident; |
2051 | PL_pending_ident = 0; | |
bbce6d69 | 2052 | |
02aa26ce NT |
2053 | /* if we're in a my(), we can't allow dynamics here. |
2054 | $foo'bar has already been turned into $foo::bar, so | |
2055 | just check for colons. | |
2056 | ||
2057 | if it's a legal name, the OP is a PADANY. | |
2058 | */ | |
3280af22 | 2059 | if (PL_in_my) { |
77ca0c92 | 2060 | if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */ |
1ec3e8de GS |
2061 | if (strchr(PL_tokenbuf,':')) |
2062 | yyerror(Perl_form(aTHX_ "No package name allowed for " | |
2063 | "variable %s in \"our\"", | |
2064 | PL_tokenbuf)); | |
77ca0c92 LW |
2065 | tmp = pad_allocmy(PL_tokenbuf); |
2066 | } | |
2067 | else { | |
2068 | if (strchr(PL_tokenbuf,':')) | |
2069 | yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf)); | |
02aa26ce | 2070 | |
77ca0c92 LW |
2071 | yylval.opval = newOP(OP_PADANY, 0); |
2072 | yylval.opval->op_targ = pad_allocmy(PL_tokenbuf); | |
2073 | return PRIVATEREF; | |
2074 | } | |
bbce6d69 | 2075 | } |
2076 | ||
02aa26ce NT |
2077 | /* |
2078 | build the ops for accesses to a my() variable. | |
2079 | ||
2080 | Deny my($a) or my($b) in a sort block, *if* $a or $b is | |
2081 | then used in a comparison. This catches most, but not | |
2082 | all cases. For instance, it catches | |
2083 | sort { my($a); $a <=> $b } | |
2084 | but not | |
2085 | sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; } | |
2086 | (although why you'd do that is anyone's guess). | |
2087 | */ | |
2088 | ||
3280af22 | 2089 | if (!strchr(PL_tokenbuf,':')) { |
a863c7d1 | 2090 | #ifdef USE_THREADS |
54b9620d | 2091 | /* Check for single character per-thread SVs */ |
3280af22 NIS |
2092 | if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0' |
2093 | && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */ | |
2094 | && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD) | |
554b3eca | 2095 | { |
2faa37cc | 2096 | yylval.opval = newOP(OP_THREADSV, 0); |
a863c7d1 MB |
2097 | yylval.opval->op_targ = tmp; |
2098 | return PRIVATEREF; | |
2099 | } | |
2100 | #endif /* USE_THREADS */ | |
3280af22 | 2101 | if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) { |
f472eb5c | 2102 | SV *namesv = AvARRAY(PL_comppad_name)[tmp]; |
77ca0c92 | 2103 | /* might be an "our" variable" */ |
f472eb5c | 2104 | if (SvFLAGS(namesv) & SVpad_OUR) { |
77ca0c92 | 2105 | /* build ops for a bareword */ |
f472eb5c GS |
2106 | SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0); |
2107 | sv_catpvn(sym, "::", 2); | |
2108 | sv_catpv(sym, PL_tokenbuf+1); | |
2109 | yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym); | |
77ca0c92 | 2110 | yylval.opval->op_private = OPpCONST_ENTERED; |
f472eb5c | 2111 | gv_fetchpv(SvPVX(sym), |
77ca0c92 | 2112 | (PL_in_eval |
f472eb5c GS |
2113 | ? (GV_ADDMULTI | GV_ADDINEVAL) |
2114 | : TRUE | |
77ca0c92 LW |
2115 | ), |
2116 | ((PL_tokenbuf[0] == '$') ? SVt_PV | |
2117 | : (PL_tokenbuf[0] == '@') ? SVt_PVAV | |
2118 | : SVt_PVHV)); | |
2119 | return WORD; | |
2120 | } | |
2121 | ||
02aa26ce | 2122 | /* if it's a sort block and they're naming $a or $b */ |
3280af22 NIS |
2123 | if (PL_last_lop_op == OP_SORT && |
2124 | PL_tokenbuf[0] == '$' && | |
2125 | (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b') | |
2126 | && !PL_tokenbuf[2]) | |
bbce6d69 | 2127 | { |
3280af22 NIS |
2128 | for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart; |
2129 | d < PL_bufend && *d != '\n'; | |
a863c7d1 MB |
2130 | d++) |
2131 | { | |
2132 | if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) { | |
cea2e8a9 | 2133 | Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison", |
3280af22 | 2134 | PL_tokenbuf); |
a863c7d1 | 2135 | } |
bbce6d69 | 2136 | } |
2137 | } | |
bbce6d69 | 2138 | |
a863c7d1 MB |
2139 | yylval.opval = newOP(OP_PADANY, 0); |
2140 | yylval.opval->op_targ = tmp; | |
2141 | return PRIVATEREF; | |
2142 | } | |
bbce6d69 | 2143 | } |
2144 | ||
02aa26ce NT |
2145 | /* |
2146 | Whine if they've said @foo in a doublequoted string, | |
2147 | and @foo isn't a variable we can find in the symbol | |
2148 | table. | |
2149 | */ | |
3280af22 NIS |
2150 | if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) { |
2151 | GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV); | |
8593bda5 GS |
2152 | if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv))) |
2153 | && ckWARN(WARN_AMBIGUOUS)) | |
2154 | { | |
2155 | /* Downgraded from fatal to warning 20000522 mjd */ | |
2156 | Perl_warner(aTHX_ WARN_AMBIGUOUS, | |
2157 | "Possible unintended interpolation of %s in string", | |
2158 | PL_tokenbuf); | |
2159 | } | |
bbce6d69 | 2160 | } |
2161 | ||
02aa26ce | 2162 | /* build ops for a bareword */ |
3280af22 | 2163 | yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0)); |
bbce6d69 | 2164 | yylval.opval->op_private = OPpCONST_ENTERED; |
3280af22 NIS |
2165 | gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE, |
2166 | ((PL_tokenbuf[0] == '$') ? SVt_PV | |
2167 | : (PL_tokenbuf[0] == '@') ? SVt_PVAV | |
bbce6d69 | 2168 | : SVt_PVHV)); |
2169 | return WORD; | |
2170 | } | |
2171 | ||
02aa26ce NT |
2172 | /* no identifier pending identification */ |
2173 | ||
3280af22 | 2174 | switch (PL_lex_state) { |
79072805 LW |
2175 | #ifdef COMMENTARY |
2176 | case LEX_NORMAL: /* Some compilers will produce faster */ | |
2177 | case LEX_INTERPNORMAL: /* code if we comment these out. */ | |
2178 | break; | |
2179 | #endif | |
2180 | ||
09bef843 | 2181 | /* when we've already built the next token, just pull it out of the queue */ |
79072805 | 2182 | case LEX_KNOWNEXT: |
3280af22 NIS |
2183 | PL_nexttoke--; |
2184 | yylval = PL_nextval[PL_nexttoke]; | |
2185 | if (!PL_nexttoke) { | |
2186 | PL_lex_state = PL_lex_defer; | |
2187 | PL_expect = PL_lex_expect; | |
2188 | PL_lex_defer = LEX_NORMAL; | |
463ee0b2 | 2189 | } |
3280af22 | 2190 | return(PL_nexttype[PL_nexttoke]); |
79072805 | 2191 | |
02aa26ce | 2192 | /* interpolated case modifiers like \L \U, including \Q and \E. |
3280af22 | 2193 | when we get here, PL_bufptr is at the \ |
02aa26ce | 2194 | */ |
79072805 LW |
2195 | case LEX_INTERPCASEMOD: |
2196 | #ifdef DEBUGGING | |
3280af22 | 2197 | if (PL_bufptr != PL_bufend && *PL_bufptr != '\\') |
cea2e8a9 | 2198 | Perl_croak(aTHX_ "panic: INTERPCASEMOD"); |
79072805 | 2199 | #endif |
02aa26ce | 2200 | /* handle \E or end of string */ |
3280af22 | 2201 | if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') { |
a0d0e21e | 2202 | char oldmod; |
02aa26ce NT |
2203 | |
2204 | /* if at a \E */ | |
3280af22 NIS |
2205 | if (PL_lex_casemods) { |
2206 | oldmod = PL_lex_casestack[--PL_lex_casemods]; | |
2207 | PL_lex_casestack[PL_lex_casemods] = '\0'; | |
02aa26ce | 2208 | |
3280af22 NIS |
2209 | if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) { |
2210 | PL_bufptr += 2; | |
2211 | PL_lex_state = LEX_INTERPCONCAT; | |
a0d0e21e | 2212 | } |
79072805 LW |
2213 | return ')'; |
2214 | } | |
3280af22 NIS |
2215 | if (PL_bufptr != PL_bufend) |
2216 | PL_bufptr += 2; | |
2217 | PL_lex_state = LEX_INTERPCONCAT; | |
cea2e8a9 | 2218 | return yylex(); |
79072805 LW |
2219 | } |
2220 | else { | |
3280af22 | 2221 | s = PL_bufptr + 1; |
79072805 LW |
2222 | if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3)) |
2223 | tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */ | |
a0d0e21e | 2224 | if (strchr("LU", *s) && |
3280af22 | 2225 | (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U'))) |
a0d0e21e | 2226 | { |
3280af22 | 2227 | PL_lex_casestack[--PL_lex_casemods] = '\0'; |
a0d0e21e LW |
2228 | return ')'; |
2229 | } | |
3280af22 NIS |
2230 | if (PL_lex_casemods > 10) { |
2231 | char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char); | |
2232 | if (newlb != PL_lex_casestack) { | |
a0d0e21e | 2233 | SAVEFREEPV(newlb); |
3280af22 | 2234 | PL_lex_casestack = newlb; |
a0d0e21e LW |
2235 | } |
2236 | } | |
3280af22 NIS |
2237 | PL_lex_casestack[PL_lex_casemods++] = *s; |
2238 | PL_lex_casestack[PL_lex_casemods] = '\0'; | |
2239 | PL_lex_state = LEX_INTERPCONCAT; | |
2240 | PL_nextval[PL_nexttoke].ival = 0; | |
79072805 LW |
2241 | force_next('('); |
2242 | if (*s == 'l') | |
3280af22 | 2243 | PL_nextval[PL_nexttoke].ival = OP_LCFIRST; |
79072805 | 2244 | else if (*s == 'u') |
3280af22 | 2245 | PL_nextval[PL_nexttoke].ival = OP_UCFIRST; |
79072805 | 2246 | else if (*s == 'L') |
3280af22 | 2247 | PL_nextval[PL_nexttoke].ival = OP_LC; |
79072805 | 2248 | else if (*s == 'U') |
3280af22 | 2249 | PL_nextval[PL_nexttoke].ival = OP_UC; |
a0d0e21e | 2250 | else if (*s == 'Q') |
3280af22 | 2251 | PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA; |
79072805 | 2252 | else |
cea2e8a9 | 2253 | Perl_croak(aTHX_ "panic: yylex"); |
3280af22 | 2254 | PL_bufptr = s + 1; |
79072805 | 2255 | force_next(FUNC); |
3280af22 NIS |
2256 | if (PL_lex_starts) { |
2257 | s = PL_bufptr; | |
2258 | PL_lex_starts = 0; | |
79072805 LW |
2259 | Aop(OP_CONCAT); |
2260 | } | |
2261 | else | |
cea2e8a9 | 2262 | return yylex(); |
79072805 LW |
2263 | } |
2264 | ||
55497cff | 2265 | case LEX_INTERPPUSH: |
2266 | return sublex_push(); | |
2267 | ||
79072805 | 2268 | case LEX_INTERPSTART: |
3280af22 | 2269 | if (PL_bufptr == PL_bufend) |
79072805 | 2270 | return sublex_done(); |
3280af22 NIS |
2271 | PL_expect = XTERM; |
2272 | PL_lex_dojoin = (*PL_bufptr == '@'); | |
2273 | PL_lex_state = LEX_INTERPNORMAL; | |
2274 | if (PL_lex_dojoin) { | |
2275 | PL_nextval[PL_nexttoke].ival = 0; | |
79072805 | 2276 | force_next(','); |
554b3eca | 2277 | #ifdef USE_THREADS |
533c011a NIS |
2278 | PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0); |
2279 | PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\""); | |
554b3eca MB |
2280 | force_next(PRIVATEREF); |
2281 | #else | |
a0d0e21e | 2282 | force_ident("\"", '$'); |
554b3eca | 2283 | #endif /* USE_THREADS */ |
3280af22 | 2284 | PL_nextval[PL_nexttoke].ival = 0; |
79072805 | 2285 | force_next('$'); |
3280af22 | 2286 | PL_nextval[PL_nexttoke].ival = 0; |
79072805 | 2287 | force_next('('); |
3280af22 | 2288 | PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */ |
79072805 LW |
2289 | force_next(FUNC); |
2290 | } | |
3280af22 NIS |
2291 | if (PL_lex_starts++) { |
2292 | s = PL_bufptr; | |
79072805 LW |
2293 | Aop(OP_CONCAT); |
2294 | } | |
cea2e8a9 | 2295 | return yylex(); |
79072805 LW |
2296 | |
2297 | case LEX_INTERPENDMAYBE: | |
3280af22 NIS |
2298 | if (intuit_more(PL_bufptr)) { |
2299 | PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */ | |
79072805 LW |
2300 | break; |
2301 | } | |
2302 | /* FALL THROUGH */ | |
2303 | ||
2304 | case LEX_INTERPEND: | |
3280af22 NIS |
2305 | if (PL_lex_dojoin) { |
2306 | PL_lex_dojoin = FALSE; | |
2307 | PL_lex_state = LEX_INTERPCONCAT; | |
79072805 LW |
2308 | return ')'; |
2309 | } | |
43a16006 | 2310 | if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl |
25da4f38 | 2311 | && SvEVALED(PL_lex_repl)) |
43a16006 | 2312 | { |
e9fa98b2 | 2313 | if (PL_bufptr != PL_bufend) |
cea2e8a9 | 2314 | Perl_croak(aTHX_ "Bad evalled substitution pattern"); |
e9fa98b2 HS |
2315 | PL_lex_repl = Nullsv; |
2316 | } | |
79072805 LW |
2317 | /* FALLTHROUGH */ |
2318 | case LEX_INTERPCONCAT: | |
2319 | #ifdef DEBUGGING | |
3280af22 | 2320 | if (PL_lex_brackets) |
cea2e8a9 | 2321 | Perl_croak(aTHX_ "panic: INTERPCONCAT"); |
79072805 | 2322 | #endif |
3280af22 | 2323 | if (PL_bufptr == PL_bufend) |
79072805 LW |
2324 | return sublex_done(); |
2325 | ||
3280af22 NIS |
2326 | if (SvIVX(PL_linestr) == '\'') { |
2327 | SV *sv = newSVsv(PL_linestr); | |
2328 | if (!PL_lex_inpat) | |
76e3520e | 2329 | sv = tokeq(sv); |
3280af22 | 2330 | else if ( PL_hints & HINT_NEW_RE ) |
b3ac6de7 | 2331 | sv = new_constant(NULL, 0, "qr", sv, sv, "q"); |
79072805 | 2332 | yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv); |
3280af22 | 2333 | s = PL_bufend; |
79072805 LW |
2334 | } |
2335 | else { | |
3280af22 | 2336 | s = scan_const(PL_bufptr); |
79072805 | 2337 | if (*s == '\\') |
3280af22 | 2338 | PL_lex_state = LEX_INTERPCASEMOD; |
79072805 | 2339 | else |
3280af22 | 2340 | PL_lex_state = LEX_INTERPSTART; |
79072805 LW |
2341 | } |
2342 | ||
3280af22 NIS |
2343 | if (s != PL_bufptr) { |
2344 | PL_nextval[PL_nexttoke] = yylval; | |
2345 | PL_expect = XTERM; | |
79072805 | 2346 | force_next(THING); |
3280af22 | 2347 | if (PL_lex_starts++) |
79072805 LW |
2348 | Aop(OP_CONCAT); |
2349 | else { | |
3280af22 | 2350 | PL_bufptr = s; |
cea2e8a9 | 2351 | return yylex(); |
79072805 LW |
2352 | } |
2353 | } | |
2354 | ||
cea2e8a9 | 2355 | return yylex(); |
a0d0e21e | 2356 | case LEX_FORMLINE: |
3280af22 NIS |
2357 | PL_lex_state = LEX_NORMAL; |
2358 | s = scan_formline(PL_bufptr); | |
2359 | if (!PL_lex_formbrack) | |
a0d0e21e LW |
2360 | goto rightbracket; |
2361 | OPERATOR(';'); | |
79072805 LW |
2362 | } |
2363 | ||
3280af22 NIS |
2364 | s = PL_bufptr; |
2365 | PL_oldoldbufptr = PL_oldbufptr; | |
2366 | PL_oldbufptr = s; | |
79072805 | 2367 | DEBUG_p( { |
bf49b057 GS |
2368 | PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n", |
2369 | exp_name[PL_expect], s); | |
79072805 | 2370 | } ) |
463ee0b2 LW |
2371 | |
2372 | retry: | |
378cc40b LW |
2373 | switch (*s) { |
2374 | default: | |
7e2040f0 | 2375 | if (isIDFIRST_lazy_if(s,UTF)) |
834a4ddd | 2376 | goto keylookup; |
cea2e8a9 | 2377 | Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255); |
e929a76b LW |
2378 | case 4: |
2379 | case 26: | |
2380 | goto fake_eof; /* emulate EOF on ^D or ^Z */ | |
378cc40b | 2381 | case 0: |
3280af22 NIS |
2382 | if (!PL_rsfp) { |
2383 | PL_last_uni = 0; | |
2384 | PL_last_lop = 0; | |
2385 | if (PL_lex_brackets) | |
d98d5fff | 2386 | yyerror("Missing right curly or square bracket"); |
79072805 | 2387 | TOKEN(0); |
463ee0b2 | 2388 | } |
3280af22 | 2389 | if (s++ < PL_bufend) |
a687059c | 2390 | goto retry; /* ignore stray nulls */ |
3280af22 NIS |
2391 | PL_last_uni = 0; |
2392 | PL_last_lop = 0; | |
2393 | if (!PL_in_eval && !PL_preambled) { | |
2394 | PL_preambled = TRUE; | |
2395 | sv_setpv(PL_linestr,incl_perldb()); | |
2396 | if (SvCUR(PL_linestr)) | |
2397 | sv_catpv(PL_linestr,";"); | |
2398 | if (PL_preambleav){ | |
2399 | while(AvFILLp(PL_preambleav) >= 0) { | |
2400 | SV *tmpsv = av_shift(PL_preambleav); | |
2401 | sv_catsv(PL_linestr, tmpsv); | |
2402 | sv_catpv(PL_linestr, ";"); | |
91b7def8 | 2403 | sv_free(tmpsv); |
2404 | } | |
3280af22 NIS |
2405 | sv_free((SV*)PL_preambleav); |
2406 | PL_preambleav = NULL; | |
91b7def8 | 2407 | } |
3280af22 NIS |
2408 | if (PL_minus_n || PL_minus_p) { |
2409 | sv_catpv(PL_linestr, "LINE: while (<>) {"); | |
2410 | if (PL_minus_l) | |
2411 | sv_catpv(PL_linestr,"chomp;"); | |
2412 | if (PL_minus_a) { | |
8fd239a7 CS |
2413 | GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV); |
2414 | if (gv) | |
2415 | GvIMPORTED_AV_on(gv); | |
3280af22 NIS |
2416 | if (PL_minus_F) { |
2417 | if (strchr("/'\"", *PL_splitstr) | |
2418 | && strchr(PL_splitstr + 1, *PL_splitstr)) | |
cea2e8a9 | 2419 | Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr); |
54310121 | 2420 | else { |
2421 | char delim; | |
2422 | s = "'~#\200\1'"; /* surely one char is unused...*/ | |
3280af22 | 2423 | while (s[1] && strchr(PL_splitstr, *s)) s++; |
54310121 | 2424 | delim = *s; |
cea2e8a9 | 2425 | Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c", |
46fc3d4c | 2426 | "q" + (delim == '\''), delim); |
3280af22 | 2427 | for (s = PL_splitstr; *s; s++) { |
54310121 | 2428 | if (*s == '\\') |
3280af22 NIS |
2429 | sv_catpvn(PL_linestr, "\\", 1); |
2430 | sv_catpvn(PL_linestr, s, 1); | |
54310121 | 2431 | } |
cea2e8a9 | 2432 | Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim); |
54310121 | 2433 | } |
2304df62 AD |
2434 | } |
2435 | else | |
3280af22 | 2436 | sv_catpv(PL_linestr,"@F=split(' ');"); |
2304df62 | 2437 | } |
79072805 | 2438 | } |
3280af22 NIS |
2439 | sv_catpv(PL_linestr, "\n"); |
2440 | PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr); | |
2441 | PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); | |
2442 | if (PERLDB_LINE && PL_curstash != PL_debstash) { | |
a0d0e21e LW |
2443 | SV *sv = NEWSV(85,0); |
2444 | ||
2445 | sv_upgrade(sv, SVt_PVMG); | |
3280af22 | 2446 | sv_setsv(sv,PL_linestr); |
57843af0 | 2447 | av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv); |
a0d0e21e | 2448 | } |
79072805 | 2449 | goto retry; |
a687059c | 2450 | } |
e929a76b | 2451 | do { |
3280af22 | 2452 | if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) { |
e929a76b | 2453 | fake_eof: |
3280af22 NIS |
2454 | if (PL_rsfp) { |
2455 | if (PL_preprocess && !PL_in_eval) | |
2456 | (void)PerlProc_pclose(PL_rsfp); | |
2457 | else if ((PerlIO *)PL_rsfp == PerlIO_stdin()) | |
2458 | PerlIO_clearerr(PL_rsfp); | |
395c3793 | 2459 | else |
3280af22 NIS |
2460 | (void)PerlIO_close(PL_rsfp); |
2461 | PL_rsfp = Nullfp; | |
4a9ae47a | 2462 | PL_doextract = FALSE; |
395c3793 | 2463 | } |
3280af22 NIS |
2464 | if (!PL_in_eval && (PL_minus_n || PL_minus_p)) { |
2465 | sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : ""); | |
2466 | sv_catpv(PL_linestr,";}"); | |
2467 | PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr); | |
2468 | PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); | |
2469 | PL_minus_n = PL_minus_p = 0; | |
e929a76b LW |
2470 | goto retry; |
2471 | } | |
3280af22 NIS |
2472 | PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr); |
2473 | sv_setpv(PL_linestr,""); | |
79072805 | 2474 | TOKEN(';'); /* not infinite loop because rsfp is NULL now */ |
378cc40b | 2475 | } |
3280af22 | 2476 | if (PL_doextract) { |
a0d0e21e | 2477 | if (*s == '#' && s[1] == '!' && instr(s,"perl")) |
3280af22 | 2478 | PL_doextract = FALSE; |
a0d0e21e LW |
2479 | |
2480 | /* Incest with pod. */ | |
2481 | if (*s == '=' && strnEQ(s, "=cut", 4)) { | |
3280af22 NIS |
2482 | sv_setpv(PL_linestr, ""); |
2483 | PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr); | |
2484 | PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); | |
2485 | PL_doextract = FALSE; | |
a0d0e21e LW |
2486 | } |
2487 | } | |
463ee0b2 | 2488 | incline(s); |
3280af22 NIS |
2489 | } while (PL_doextract); |
2490 | PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s; | |
2491 | if (PERLDB_LINE && PL_curstash != PL_debstash) { | |
79072805 | 2492 | SV *sv = NEWSV(85,0); |
a687059c | 2493 | |
93a17b20 | 2494 | sv_upgrade(sv, SVt_PVMG); |
3280af22 | 2495 | sv_setsv(sv,PL_linestr); |
57843af0 | 2496 | av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv); |
a687059c | 2497 | } |
3280af22 | 2498 | PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); |
57843af0 | 2499 | if (CopLINE(PL_curcop) == 1) { |
3280af22 | 2500 | while (s < PL_bufend && isSPACE(*s)) |
79072805 | 2501 | s++; |
a0d0e21e | 2502 | if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */ |
79072805 | 2503 | s++; |
44a8e56a | 2504 | d = Nullch; |
3280af22 | 2505 | if (!PL_in_eval) { |
44a8e56a | 2506 | if (*s == '#' && *(s+1) == '!') |
2507 | d = s + 2; | |
2508 | #ifdef ALTERNATE_SHEBANG | |
2509 | else { | |
2510 | static char as[] = ALTERNATE_SHEBANG; | |
2511 | if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1)) | |
2512 | d = s + (sizeof(as) - 1); | |
2513 | } | |
2514 | #endif /* ALTERNATE_SHEBANG */ | |
2515 | } | |
2516 | if (d) { | |
b8378b72 | 2517 | char *ipath; |
774d564b | 2518 | char *ipathend; |
b8378b72 | 2519 | |
774d564b | 2520 | while (isSPACE(*d)) |
b8378b72 CS |
2521 | d++; |
2522 | ipath = d; | |
774d564b | 2523 | while (*d && !isSPACE(*d)) |
2524 | d++; | |
2525 | ipathend = d; | |
2526 | ||
2527 | #ifdef ARG_ZERO_IS_SCRIPT | |
2528 | if (ipathend > ipath) { | |
2529 | /* | |
2530 | * HP-UX (at least) sets argv[0] to the script name, | |
2531 | * which makes $^X incorrect. And Digital UNIX and Linux, | |
2532 | * at least, set argv[0] to the basename of the Perl | |
2533 | * interpreter. So, having found "#!", we'll set it right. | |
2534 | */ | |
2535 | SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV)); | |
2536 | assert(SvPOK(x) || SvGMAGICAL(x)); | |
cc49e20b | 2537 | if (sv_eq(x, CopFILESV(PL_curcop))) { |
774d564b | 2538 | sv_setpvn(x, ipath, ipathend - ipath); |
9607fc9c | 2539 | SvSETMAGIC(x); |
2540 | } | |
774d564b | 2541 | TAINT_NOT; /* $^X is always tainted, but that's OK */ |
8ebc5c01 | 2542 | } |
774d564b | 2543 | #endif /* ARG_ZERO_IS_SCRIPT */ |
b8378b72 CS |
2544 | |
2545 | /* | |
2546 | * Look for options. | |
2547 | */ | |
748a9306 | 2548 | d = instr(s,"perl -"); |
84e30d1a | 2549 | if (!d) { |
748a9306 | 2550 | d = instr(s,"perl"); |
84e30d1a GS |
2551 | #if defined(DOSISH) |
2552 | /* avoid getting into infinite loops when shebang | |
2553 | * line contains "Perl" rather than "perl" */ | |
2554 | if (!d) { | |
2555 | for (d = ipathend-4; d >= ipath; --d) { | |
2556 | if ((*d == 'p' || *d == 'P') | |
2557 | && !ibcmp(d, "perl", 4)) | |
2558 | { | |
2559 | break; | |
2560 | } | |
2561 | } | |
2562 | if (d < ipath) | |
2563 | d = Nullch; | |
2564 | } | |
2565 | #endif | |
2566 | } | |
44a8e56a | 2567 | #ifdef ALTERNATE_SHEBANG |
2568 | /* | |
2569 | * If the ALTERNATE_SHEBANG on this system starts with a | |
2570 | * character that can be part of a Perl expression, then if | |
2571 | * we see it but not "perl", we're probably looking at the | |
2572 | * start of Perl code, not a request to hand off to some | |
2573 | * other interpreter. Similarly, if "perl" is there, but | |
2574 | * not in the first 'word' of the line, we assume the line | |
2575 | * contains the start of the Perl program. | |
44a8e56a | 2576 | */ |
2577 | if (d && *s != '#') { | |
774d564b | 2578 | char *c = ipath; |
44a8e56a | 2579 | while (*c && !strchr("; \t\r\n\f\v#", *c)) |
2580 | c++; | |
2581 | if (c < d) | |
2582 | d = Nullch; /* "perl" not in first word; ignore */ | |
2583 | else | |
2584 | *s = '#'; /* Don't try to parse shebang line */ | |
2585 | } | |
774d564b | 2586 | #endif /* ALTERNATE_SHEBANG */ |
748a9306 | 2587 | if (!d && |
44a8e56a | 2588 | *s == '#' && |
774d564b | 2589 | ipathend > ipath && |
3280af22 | 2590 | !PL_minus_c && |
748a9306 | 2591 | !instr(s,"indir") && |
3280af22 | 2592 | instr(PL_origargv[0],"perl")) |
748a9306 | 2593 | { |
9f68db38 | 2594 | char **newargv; |
9f68db38 | 2595 | |
774d564b | 2596 | *ipathend = '\0'; |
2597 | s = ipathend + 1; | |
3280af22 | 2598 | while (s < PL_bufend && isSPACE(*s)) |
9f68db38 | 2599 | s++; |
3280af22 NIS |
2600 | if (s < PL_bufend) { |
2601 | Newz(899,newargv,PL_origargc+3,char*); | |
9f68db38 | 2602 | newargv[1] = s; |
3280af22 | 2603 | while (s < PL_bufend && !isSPACE(*s)) |
9f68db38 LW |
2604 | s++; |
2605 | *s = '\0'; | |
3280af22 | 2606 | Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*); |
9f68db38 LW |
2607 | } |
2608 | else | |
3280af22 | 2609 | newargv = PL_origargv; |
774d564b | 2610 | newargv[0] = ipath; |
80252599 | 2611 | PerlProc_execv(ipath, newargv); |
cea2e8a9 | 2612 | Perl_croak(aTHX_ "Can't exec %s", ipath); |
9f68db38 | 2613 | } |
748a9306 | 2614 | if (d) { |
3280af22 NIS |
2615 | U32 oldpdb = PL_perldb; |
2616 | bool oldn = PL_minus_n; | |
2617 | bool oldp = PL_minus_p; | |
748a9306 LW |
2618 | |
2619 | while (*d && !isSPACE(*d)) d++; | |
89bfa8cd | 2620 | while (*d == ' ' || *d == '\t') d++; |
748a9306 LW |
2621 | |
2622 | if (*d++ == '-') { | |
8cc95fdb | 2623 | do { |
2624 | if (*d == 'M' || *d == 'm') { | |
2625 | char *m = d; | |
2626 | while (*d && !isSPACE(*d)) d++; | |
cea2e8a9 | 2627 | Perl_croak(aTHX_ "Too late for \"-%.*s\" option", |
8cc95fdb | 2628 | (int)(d - m), m); |
2629 | } | |
2630 | d = moreswitches(d); | |
2631 | } while (d); | |
155aba94 GS |
2632 | if ((PERLDB_LINE && !oldpdb) || |
2633 | ((PL_minus_n || PL_minus_p) && !(oldn || oldp))) | |
b084f20b | 2634 | /* if we have already added "LINE: while (<>) {", |
2635 | we must not do it again */ | |
748a9306 | 2636 | { |
3280af22 NIS |
2637 | sv_setpv(PL_linestr, ""); |
2638 | PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr); | |
2639 | PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); | |
2640 | PL_preambled = FALSE; | |
84902520 | 2641 | if (PERLDB_LINE) |
3280af22 | 2642 | (void)gv_fetchfile(PL_origfilename); |
748a9306 LW |
2643 | goto retry; |
2644 | } | |
a0d0e21e | 2645 | } |
79072805 | 2646 | } |
9f68db38 | 2647 | } |
79072805 | 2648 | } |
3280af22 NIS |
2649 | if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) { |
2650 | PL_bufptr = s; | |
2651 | PL_lex_state = LEX_FORMLINE; | |
cea2e8a9 | 2652 | return yylex(); |
ae986130 | 2653 | } |
378cc40b | 2654 | goto retry; |
4fdae800 | 2655 | case '\r': |
6a27c188 | 2656 | #ifdef PERL_STRICT_CR |
cea2e8a9 GS |
2657 | Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r'); |
2658 | Perl_croak(aTHX_ | |
cc507455 | 2659 | "\t(Maybe you didn't strip carriage returns after a network transfer?)\n"); |
a868473f | 2660 | #endif |
4fdae800 | 2661 | case ' ': case '\t': case '\f': case 013: |
378cc40b LW |
2662 | s++; |
2663 | goto retry; | |
378cc40b | 2664 | case '#': |
e929a76b | 2665 | case '\n': |
3280af22 | 2666 | if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) { |
df0deb90 GS |
2667 | if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) { |
2668 | /* handle eval qq[#line 1 "foo"\n ...] */ | |
2669 | CopLINE_dec(PL_curcop); | |
2670 | incline(s); | |
2671 | } | |
3280af22 | 2672 | d = PL_bufend; |
a687059c | 2673 | while (s < d && *s != '\n') |
378cc40b | 2674 | s++; |
0f85fab0 | 2675 | if (s < d) |
378cc40b | 2676 | s++; |
463ee0b2 | 2677 | incline(s); |
3280af22 NIS |
2678 | if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) { |
2679 | PL_bufptr = s; | |
2680 | PL_lex_state = LEX_FORMLINE; | |
cea2e8a9 | 2681 | return yylex(); |
a687059c | 2682 | } |
378cc40b | 2683 | } |
a687059c | 2684 | else { |
378cc40b | 2685 | *s = '\0'; |
3280af22 | 2686 | PL_bufend = s; |
a687059c | 2687 | } |
378cc40b LW |
2688 | goto retry; |
2689 | case '-': | |
79072805 | 2690 | if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) { |
378cc40b | 2691 | s++; |
3280af22 | 2692 | PL_bufptr = s; |
748a9306 LW |
2693 | tmp = *s++; |
2694 | ||
3280af22 | 2695 | while (s < PL_bufend && (*s == ' ' || *s == '\t')) |
748a9306 LW |
2696 | s++; |
2697 | ||
2698 | if (strnEQ(s,"=>",2)) { | |
3280af22 | 2699 | s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE); |
748a9306 LW |
2700 | OPERATOR('-'); /* unary minus */ |
2701 | } | |
3280af22 NIS |
2702 | PL_last_uni = PL_oldbufptr; |
2703 | PL_last_lop_op = OP_FTEREAD; /* good enough */ | |
748a9306 | 2704 | switch (tmp) { |
79072805 LW |
2705 | case 'r': FTST(OP_FTEREAD); |
2706 | case 'w': FTST(OP_FTEWRITE); | |
2707 | case 'x': FTST(OP_FTEEXEC); | |
2708 | case 'o': FTST(OP_FTEOWNED); | |
2709 | case 'R': FTST(OP_FTRREAD); | |
2710 | case 'W': FTST(OP_FTRWRITE); | |
2711 | case 'X': FTST(OP_FTREXEC); | |
2712 | case 'O': FTST(OP_FTROWNED); | |
2713 | case 'e': FTST(OP_FTIS); | |
2714 | case 'z': FTST(OP_FTZERO); | |
2715 | case 's': FTST(OP_FTSIZE); | |
2716 | case 'f': FTST(OP_FTFILE); | |
2717 | case 'd': FTST(OP_FTDIR); | |
2718 | case 'l': FTST(OP_FTLINK); | |
2719 | case 'p': FTST(OP_FTPIPE); | |
2720 | case 'S': FTST(OP_FTSOCK); | |
2721 | case 'u': FTST(OP_FTSUID); | |
2722 | case 'g': FTST(OP_FTSGID); | |
2723 | case 'k': FTST(OP_FTSVTX); | |
2724 | case 'b': FTST(OP_FTBLK); | |
2725 | case 'c': FTST(OP_FTCHR); | |
2726 | case 't': FTST(OP_FTTTY); | |
2727 | case 'T': FTST(OP_FTTEXT); | |
2728 | case 'B': FTST(OP_FTBINARY); | |
85e6fe83 LW |
2729 | case 'M': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTMTIME); |
2730 | case 'A': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTATIME); | |
2731 | case 'C': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTCTIME); | |
378cc40b | 2732 | default: |
cea2e8a9 | 2733 | Perl_croak(aTHX_ "Unrecognized file test: -%c", (int)tmp); |
378cc40b LW |
2734 | break; |
2735 | } | |
2736 | } | |
a687059c LW |
2737 | tmp = *s++; |
2738 | if (*s == tmp) { | |
2739 | s++; | |
3280af22 | 2740 | if (PL_expect == XOPERATOR) |
79072805 LW |
2741 | TERM(POSTDEC); |
2742 | else | |
2743 | OPERATOR(PREDEC); | |
2744 | } | |
2745 | else if (*s == '>') { | |
2746 | s++; | |
2747 | s = skipspace(s); | |
7e2040f0 | 2748 | if (isIDFIRST_lazy_if(s,UTF)) { |
a0d0e21e | 2749 | s = force_word(s,METHOD,FALSE,TRUE,FALSE); |
463ee0b2 | 2750 | TOKEN(ARROW); |
79072805 | 2751 | } |
748a9306 LW |
2752 | else if (*s == '$') |
2753 | OPERATOR(ARROW); | |
463ee0b2 | 2754 | else |
748a9306 | 2755 | TERM(ARROW); |
a687059c | 2756 | } |
3280af22 | 2757 | if (PL_expect == XOPERATOR) |
79072805 LW |
2758 | Aop(OP_SUBTRACT); |
2759 | else { | |
3280af22 | 2760 | if (isSPACE(*s) || !isSPACE(*PL_bufptr)) |
2f3197b3 | 2761 | check_uni(); |
79072805 | 2762 | OPERATOR('-'); /* unary minus */ |
2f3197b3 | 2763 | } |
79072805 | 2764 | |
378cc40b | 2765 | case '+': |
a687059c LW |
2766 | tmp = *s++; |
2767 | if (*s == tmp) { | |
378cc40b | 2768 | s++; |
3280af22 | 2769 | if (PL_expect == XOPERATOR) |
79072805 LW |
2770 | TERM(POSTINC); |
2771 | else | |
2772 | OPERATOR(PREINC); | |
378cc40b | 2773 | } |
3280af22 | 2774 | if (PL_expect == XOPERATOR) |
79072805 LW |
2775 | Aop(OP_ADD); |
2776 | else { | |
3280af22 | 2777 | if (isSPACE(*s) || !isSPACE(*PL_bufptr)) |
2f3197b3 | 2778 | check_uni(); |
a687059c | 2779 | OPERATOR('+'); |
2f3197b3 | 2780 | } |
a687059c | 2781 | |
378cc40b | 2782 | case '*': |
3280af22 NIS |
2783 | if (PL_expect != XOPERATOR) { |
2784 | s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE); | |
2785 | PL_expect = XOPERATOR; | |
2786 | force_ident(PL_tokenbuf, '*'); | |
2787 | if (!*PL_tokenbuf) | |
a0d0e21e | 2788 | PREREF('*'); |
79072805 | 2789 | TERM('*'); |
a687059c | 2790 | } |
79072805 LW |
2791 | s++; |
2792 | if (*s == '*') { | |
a687059c | 2793 | s++; |
79072805 | 2794 | PWop(OP_POW); |
a687059c | 2795 | } |
79072805 LW |
2796 | Mop(OP_MULTIPLY); |
2797 | ||
378cc40b | 2798 | case '%': |
3280af22 | 2799 | if (PL_expect == XOPERATOR) { |
bbce6d69 | 2800 | ++s; |
2801 | Mop(OP_MODULO); | |
a687059c | 2802 | } |
3280af22 NIS |
2803 | PL_tokenbuf[0] = '%'; |
2804 | s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE); | |
2805 | if (!PL_tokenbuf[1]) { | |
2806 | if (s == PL_bufend) | |
bbce6d69 | 2807 | yyerror("Final % should be \\% or %name"); |
2808 | PREREF('%'); | |
a687059c | 2809 | } |
3280af22 | 2810 | PL_pending_ident = '%'; |
bbce6d69 | 2811 | TERM('%'); |
a687059c | 2812 | |
378cc40b | 2813 | case '^': |
79072805 | 2814 | s++; |
a0d0e21e | 2815 | BOop(OP_BIT_XOR); |
79072805 | 2816 | case '[': |
3280af22 | 2817 | PL_lex_brackets++; |
79072805 | 2818 | /* FALL THROUGH */ |
378cc40b | 2819 | case '~': |
378cc40b | 2820 | case ',': |
378cc40b LW |
2821 | tmp = *s++; |
2822 | OPERATOR(tmp); | |
a0d0e21e LW |
2823 | case ':': |
2824 | if (s[1] == ':') { | |
2825 | len = 0; | |
2826 | goto just_a_word; | |
2827 | } | |
2828 | s++; | |
09bef843 SB |
2829 | switch (PL_expect) { |
2830 | OP *attrs; | |
2831 | case XOPERATOR: | |
2832 | if (!PL_in_my || PL_lex_state != LEX_NORMAL) | |
2833 | break; | |
2834 | PL_bufptr = s; /* update in case we back off */ | |
2835 | goto grabattrs; | |
2836 | case XATTRBLOCK: | |
2837 | PL_expect = XBLOCK; | |
2838 | goto grabattrs; | |
2839 | case XATTRTERM: | |
2840 | PL_expect = XTERMBLOCK; | |
2841 | grabattrs: | |
2842 | s = skipspace(s); | |
2843 | attrs = Nullop; | |
7e2040f0 | 2844 | while (isIDFIRST_lazy_if(s,UTF)) { |
09bef843 | 2845 | d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len); |
f9829d6b GS |
2846 | if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) { |
2847 | if (tmp < 0) tmp = -tmp; | |
2848 | switch (tmp) { | |
2849 | case KEY_or: | |
2850 | case KEY_and: | |
2851 | case KEY_for: | |
2852 | case KEY_unless: | |
2853 | case KEY_if: | |
2854 | case KEY_while: | |
2855 | case KEY_until: | |
2856 | goto got_attrs; | |
2857 | default: | |
2858 | break; | |
2859 | } | |
2860 | } | |
09bef843 SB |
2861 | if (*d == '(') { |
2862 | d = scan_str(d,TRUE,TRUE); | |
2863 | if (!d) { | |
2864 | if (PL_lex_stuff) { | |
2865 | SvREFCNT_dec(PL_lex_stuff); | |
2866 | PL_lex_stuff = Nullsv; | |
2867 | } | |
2868 | /* MUST advance bufptr here to avoid bogus | |
2869 | "at end of line" context messages from yyerror(). | |
2870 | */ | |
2871 | PL_bufptr = s + len; | |
2872 | yyerror("Unterminated attribute parameter in attribute list"); | |
2873 | if (attrs) | |
2874 | op_free(attrs); | |
2875 | return 0; /* EOF indicator */ | |
2876 | } | |
2877 | } | |
2878 | if (PL_lex_stuff) { | |
2879 | SV *sv = newSVpvn(s, len); | |
2880 | sv_catsv(sv, PL_lex_stuff); | |
2881 | attrs = append_elem(OP_LIST, attrs, | |
2882 | newSVOP(OP_CONST, 0, sv)); | |
2883 | SvREFCNT_dec(PL_lex_stuff); | |
2884 | PL_lex_stuff = Nullsv; | |
2885 | } | |
2886 | else { | |
2887 | attrs = append_elem(OP_LIST, attrs, | |
2888 | newSVOP(OP_CONST, 0, | |
2889 | newSVpvn(s, len))); | |
2890 | } | |
2891 | s = skipspace(d); | |
0120eecf | 2892 | if (*s == ':' && s[1] != ':') |
09bef843 | 2893 | s = skipspace(s+1); |
0120eecf GS |
2894 | else if (s == d) |
2895 | break; /* require real whitespace or :'s */ | |
09bef843 | 2896 | } |
f9829d6b GS |
2897 | tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */ |
2898 | if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) { | |
09bef843 SB |
2899 | char q = ((*s == '\'') ? '"' : '\''); |
2900 | /* If here for an expression, and parsed no attrs, back off. */ | |
2901 | if (tmp == '=' && !attrs) { | |
2902 | s = PL_bufptr; | |
2903 | break; | |
2904 | } | |
2905 | /* MUST advance bufptr here to avoid bogus "at end of line" | |
2906 | context messages from yyerror(). | |
2907 | */ | |
2908 | PL_bufptr = s; | |
2909 | if (!*s) | |
2910 | yyerror("Unterminated attribute list"); | |
2911 | else | |
2912 | yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list", | |
2913 | q, *s, q)); | |
2914 | if (attrs) | |
2915 | op_free(attrs); | |
2916 | OPERATOR(':'); | |
2917 | } | |
f9829d6b | 2918 | got_attrs: |
09bef843 SB |
2919 | if (attrs) { |
2920 | PL_nextval[PL_nexttoke].opval = attrs; | |
2921 | force_next(THING); | |
2922 | } | |
2923 | TOKEN(COLONATTR); | |
2924 | } | |
a0d0e21e | 2925 | OPERATOR(':'); |
8990e307 LW |
2926 | case '(': |
2927 | s++; | |
3280af22 NIS |
2928 | if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr) |
2929 | PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */ | |
a0d0e21e | 2930 | else |
3280af22 | 2931 | PL_expect = XTERM; |
a0d0e21e | 2932 | TOKEN('('); |
378cc40b | 2933 | case ';': |
57843af0 GS |
2934 | if (CopLINE(PL_curcop) < PL_copline) |
2935 | PL_copline = CopLINE(PL_curcop); | |
378cc40b LW |
2936 | tmp = *s++; |
2937 | OPERATOR(tmp); | |
2938 | case ')': | |
378cc40b | 2939 | tmp = *s++; |
16d20bd9 AD |
2940 | s = skipspace(s); |
2941 | if (*s == '{') | |
2942 | PREBLOCK(tmp); | |
378cc40b | 2943 | TERM(tmp); |
79072805 LW |
2944 | case ']': |
2945 | s++; | |
3280af22 | 2946 | if (PL_lex_brackets <= 0) |
d98d5fff | 2947 | yyerror("Unmatched right square bracket"); |
463ee0b2 | 2948 | else |
3280af22 NIS |
2949 | --PL_lex_brackets; |
2950 | if (PL_lex_state == LEX_INTERPNORMAL) { | |
2951 | if (PL_lex_brackets == 0) { | |
a0d0e21e | 2952 | if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>')) |
3280af22 | 2953 | PL_lex_state = LEX_INTERPEND; |
79072805 LW |
2954 | } |
2955 | } | |
4633a7c4 | 2956 | TERM(']'); |
79072805 LW |
2957 | case '{': |
2958 | leftbracket: | |
79072805 | 2959 | s++; |
3280af22 NIS |
2960 | if (PL_lex_brackets > 100) { |
2961 | char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char); | |
2962 | if (newlb != PL_lex_brackstack) { | |
8990e307 | 2963 | SAVEFREEPV(newlb); |
3280af22 | 2964 | PL_lex_brackstack = newlb; |
8990e307 LW |
2965 | } |
2966 | } | |
3280af22 | 2967 | switch (PL_expect) { |
a0d0e21e | 2968 | case XTERM: |
3280af22 | 2969 | if (PL_lex_formbrack) { |
a0d0e21e LW |
2970 | s--; |
2971 | PRETERMBLOCK(DO); | |
2972 | } | |
3280af22 NIS |
2973 | if (PL_oldoldbufptr == PL_last_lop) |
2974 | PL_lex_brackstack[PL_lex_brackets++] = XTERM; | |
a0d0e21e | 2975 | else |
3280af22 | 2976 | PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR; |
79072805 | 2977 | OPERATOR(HASHBRACK); |
a0d0e21e | 2978 | case XOPERATOR: |
3280af22 | 2979 | while (s < PL_bufend && (*s == ' ' || *s == '\t')) |
748a9306 | 2980 | s++; |
44a8e56a | 2981 | d = s; |
3280af22 NIS |
2982 | PL_tokenbuf[0] = '\0'; |
2983 | if (d < PL_bufend && *d == '-') { | |
2984 | PL_tokenbuf[0] = '-'; | |
44a8e56a | 2985 | d++; |
3280af22 | 2986 | while (d < PL_bufend && (*d == ' ' || *d == '\t')) |
44a8e56a | 2987 | d++; |
2988 | } | |
7e2040f0 | 2989 | if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) { |
3280af22 | 2990 | d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, |
8903cb82 | 2991 | FALSE, &len); |
3280af22 | 2992 | while (d < PL_bufend && (*d == ' ' || *d == '\t')) |
748a9306 LW |
2993 | d++; |
2994 | if (*d == '}') { | |
3280af22 | 2995 | char minus = (PL_tokenbuf[0] == '-'); |
44a8e56a | 2996 | s = force_word(s + minus, WORD, FALSE, TRUE, FALSE); |
2997 | if (minus) | |
2998 | force_next('-'); | |
748a9306 LW |
2999 | } |
3000 | } | |
3001 | /* FALL THROUGH */ | |
09bef843 | 3002 | case XATTRBLOCK: |
748a9306 | 3003 | case XBLOCK: |
3280af22 NIS |
3004 | PL_lex_brackstack[PL_lex_brackets++] = XSTATE; |
3005 | PL_expect = XSTATE; | |
a0d0e21e | 3006 | break; |
09bef843 | 3007 | case XATTRTERM: |
a0d0e21e | 3008 | case XTERMBLOCK: |
3280af22 NIS |
3009 | PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR; |
3010 | PL_expect = XSTATE; | |
a0d0e21e LW |
3011 | break; |
3012 | default: { | |
3013 | char *t; | |
3280af22 NIS |
3014 | if (PL_oldoldbufptr == PL_last_lop) |
3015 | PL_lex_brackstack[PL_lex_brackets++] = XTERM; | |
a0d0e21e | 3016 | else |
3280af22 | 3017 | PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR; |
a0d0e21e | 3018 | s = skipspace(s); |
09ecc4b6 | 3019 | if (*s == '}') |
a0d0e21e | 3020 | OPERATOR(HASHBRACK); |
b8a4b1be GS |
3021 | /* This hack serves to disambiguate a pair of curlies |
3022 | * as being a block or an anon hash. Normally, expectation | |
3023 | * determines that, but in cases where we're not in a | |
3024 | * position to expect anything in particular (like inside | |
3025 | * eval"") we have to resolve the ambiguity. This code | |
3026 | * covers the case where the first term in the curlies is a | |
3027 | * quoted string. Most other cases need to be explicitly | |
3028 | * disambiguated by prepending a `+' before the opening | |
3029 | * curly in order to force resolution as an anon hash. | |
3030 | * | |
3031 | * XXX should probably propagate the outer expectation | |
3032 | * into eval"" to rely less on this hack, but that could | |
3033 | * potentially break current behavior of eval"". | |
3034 | * GSAR 97-07-21 | |
3035 | */ | |
3036 | t = s; | |
3037 | if (*s == '\'' || *s == '"' || *s == '`') { | |
3038 | /* common case: get past first string, handling escapes */ | |
3280af22 | 3039 | for (t++; t < PL_bufend && *t != *s;) |
b8a4b1be GS |
3040 | if (*t++ == '\\' && (*t == '\\' || *t == *s)) |
3041 | t++; | |
3042 | t++; | |
a0d0e21e | 3043 | } |
b8a4b1be | 3044 | else if (*s == 'q') { |
3280af22 | 3045 | if (++t < PL_bufend |
b8a4b1be | 3046 | && (!isALNUM(*t) |
3280af22 | 3047 | || ((*t == 'q' || *t == 'x') && ++t < PL_bufend |
0505442f GS |
3048 | && !isALNUM(*t)))) |
3049 | { | |
b8a4b1be GS |
3050 | char *tmps; |
3051 | char open, close, term; | |
3052 | I32 brackets = 1; | |
3053 | ||
3280af22 | 3054 | while (t < PL_bufend && isSPACE(*t)) |
b8a4b1be GS |
3055 | t++; |
3056 | term = *t; | |
3057 | open = term; | |
3058 | if (term && (tmps = strchr("([{< )]}> )]}>",term))) | |
3059 | term = tmps[5]; | |
3060 | close = term; | |
3061 | if (open == close) | |
3280af22 NIS |
3062 | for (t++; t < PL_bufend; t++) { |
3063 | if (*t == '\\' && t+1 < PL_bufend && open != '\\') | |
b8a4b1be | 3064 | t++; |
6d07e5e9 | 3065 | else if (*t == open) |
b8a4b1be GS |
3066 | break; |
3067 | } | |
3068 | else | |
3280af22 NIS |
3069 | for (t++; t < PL_bufend; t++) { |
3070 | if (*t == '\\' && t+1 < PL_bufend) | |
b8a4b1be | 3071 | t++; |
6d07e5e9 | 3072 | else if (*t == close && --brackets <= 0) |
b8a4b1be GS |
3073 | break; |
3074 | else if (*t == open) | |
3075 | brackets++; | |
3076 | } | |
3077 | } | |
3078 | t++; | |
a0d0e21e | 3079 | } |
7e2040f0 | 3080 | else if (isALNUM_lazy_if(t,UTF)) { |
0505442f | 3081 | t += UTF8SKIP(t); |
7e2040f0 | 3082 | while (t < PL_bufend && isALNUM_lazy_if(t,UTF)) |
0505442f | 3083 | t += UTF8SKIP(t); |
a0d0e21e | 3084 | } |
3280af22 | 3085 | while (t < PL_bufend && isSPACE(*t)) |
a0d0e21e | 3086 | t++; |
b8a4b1be GS |
3087 | /* if comma follows first term, call it an anon hash */ |
3088 | /* XXX it could be a comma expression with loop modifiers */ | |
3280af22 | 3089 | if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s))) |
b8a4b1be | 3090 | || (*t == '=' && t[1] == '>'))) |
a0d0e21e | 3091 | OPERATOR(HASHBRACK); |
3280af22 | 3092 | if (PL_expect == XREF) |
4e4e412b | 3093 | PL_expect = XTERM; |
a0d0e21e | 3094 | else { |
3280af22 NIS |
3095 | PL_lex_brackstack[PL_lex_brackets-1] = XSTATE; |
3096 | PL_expect = XSTATE; | |
a0d0e21e | 3097 | } |
8990e307 | 3098 | } |
a0d0e21e | 3099 | break; |
463ee0b2 | 3100 | } |
57843af0 | 3101 | yylval.ival = CopLINE(PL_curcop); |
79072805 | 3102 | if (isSPACE(*s) || *s == '#') |
3280af22 | 3103 | PL_copline = NOLINE; /* invalidate current command line number */ |
79072805 | 3104 | TOKEN('{'); |
378cc40b | 3105 | case '}': |
79072805 LW |
3106 | rightbracket: |
3107 | s++; | |
3280af22 | 3108 | if (PL_lex_brackets <= 0) |
d98d5fff | 3109 | yyerror("Unmatched right curly bracket"); |
463ee0b2 | 3110 | else |
3280af22 NIS |
3111 | PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets]; |
3112 | if (PL_lex_brackets < PL_lex_formbrack) | |
3113 | PL_lex_formbrack = 0; | |
3114 | if (PL_lex_state == LEX_INTERPNORMAL) { | |
3115 | if (PL_lex_brackets == 0) { | |
9059aa12 LW |
3116 | if (PL_expect & XFAKEBRACK) { |
3117 | PL_expect &= XENUMMASK; | |
3280af22 NIS |
3118 | PL_lex_state = LEX_INTERPEND; |
3119 | PL_bufptr = s; | |
cea2e8a9 | 3120 | return yylex(); /* ignore fake brackets */ |
79072805 | 3121 | } |
fa83b5b6 | 3122 | if (*s == '-' && s[1] == '>') |
3280af22 | 3123 | PL_lex_state = LEX_INTERPENDMAYBE; |
fa83b5b6 | 3124 | else if (*s != '[' && *s != '{') |
3280af22 | 3125 | PL_lex_state = LEX_INTERPEND; |
79072805 LW |
3126 | } |
3127 | } | |
9059aa12 LW |
3128 | if (PL_expect & XFAKEBRACK) { |
3129 | PL_expect &= XENUMMASK; | |
3280af22 | 3130 | PL_bufptr = s; |
cea2e8a9 | 3131 | return yylex(); /* ignore fake brackets */ |
748a9306 | 3132 | } |
79072805 LW |
3133 | force_next('}'); |
3134 | TOKEN(';'); | |
378cc40b LW |
3135 | case '&': |
3136 | s++; | |
3137 | tmp = *s++; | |
3138 | if (tmp == '&') | |
a0d0e21e | 3139 | AOPERATOR(ANDAND); |
378cc40b | 3140 | s--; |
3280af22 | 3141 | if (PL_expect == XOPERATOR) { |
7e2040f0 GS |
3142 | if (ckWARN(WARN_SEMICOLON) |
3143 | && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart) | |
3144 | { | |
57843af0 | 3145 | CopLINE_dec(PL_curcop); |
cea2e8a9 | 3146 | Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi); |
57843af0 | 3147 | CopLINE_inc(PL_curcop); |
463ee0b2 | 3148 | } |
79072805 | 3149 | BAop(OP_BIT_AND); |
463ee0b2 | 3150 | } |
79072805 | 3151 | |
3280af22 NIS |
3152 | s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE); |
3153 | if (*PL_tokenbuf) { | |
3154 | PL_expect = XOPERATOR; | |
3155 | force_ident(PL_tokenbuf, '&'); | |
463ee0b2 | 3156 | } |
79072805 LW |
3157 | else |
3158 | PREREF('&'); | |
c07a80fd | 3159 | yylval.ival = (OPpENTERSUB_AMPER<<8); |
79072805 LW |
3160 | TERM('&'); |
3161 | ||
378cc40b LW |
3162 | case '|': |
3163 | s++; | |
3164 | tmp = *s++; | |
3165 | if (tmp == '|') | |
a0d0e21e | 3166 | AOPERATOR(OROR); |
378cc40b | 3167 | s--; |
79072805 | 3168 | BOop(OP_BIT_OR); |
378cc40b LW |
3169 | case '=': |
3170 | s++; | |
3171 | tmp = *s++; | |
3172 | if (tmp == '=') | |
79072805 LW |
3173 | Eop(OP_EQ); |
3174 | if (tmp == '>') | |
3175 | OPERATOR(','); | |
378cc40b | 3176 | if (tmp == '~') |
79072805 | 3177 | PMop(OP_MATCH); |
599cee73 | 3178 | if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp)) |
cea2e8a9 | 3179 | Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp); |
378cc40b | 3180 | s--; |
3280af22 NIS |
3181 | if (PL_expect == XSTATE && isALPHA(tmp) && |
3182 | (s == PL_linestart+1 || s[-2] == '\n') ) | |
748a9306 | 3183 | { |
3280af22 NIS |
3184 | if (PL_in_eval && !PL_rsfp) { |
3185 | d = PL_bufend; | |
a5f75d66 AD |
3186 | while (s < d) { |
3187 | if (*s++ == '\n') { | |
3188 | incline(s); | |
3189 | if (strnEQ(s,"=cut",4)) { | |
3190 | s = strchr(s,'\n'); | |
3191 | if (s) | |
3192 | s++; | |
3193 | else | |
3194 | s = d; | |
3195 | incline(s); | |
3196 | goto retry; | |
3197 | } | |
3198 | } | |
3199 | } | |
3200 | goto retry; | |
3201 | } | |
3280af22 NIS |
3202 | s = PL_bufend; |
3203 | PL_doextract = TRUE; | |
a0d0e21e LW |
3204 | goto retry; |
3205 | } | |
3280af22 | 3206 | if (PL_lex_brackets < PL_lex_formbrack) { |
a0d0e21e | 3207 | char *t; |
51882d45 | 3208 | #ifdef PERL_STRICT_CR |
a0d0e21e | 3209 | for (t = s; *t == ' ' || *t == '\t'; t++) ; |
51882d45 GS |
3210 | #else |
3211 | for (t = s; *t == ' ' || *t == '\t' || *t == '\r'; t++) ; | |
3212 | #endif | |
a0d0e21e LW |
3213 | if (*t == '\n' || *t == '#') { |
3214 | s--; | |
3280af22 | 3215 | PL_expect = XBLOCK; |
a0d0e21e LW |
3216 | goto leftbracket; |
3217 | } | |
79072805 | 3218 | } |
a0d0e21e LW |
3219 | yylval.ival = 0; |
3220 | OPERATOR(ASSIGNOP); | |
378cc40b LW |
3221 | case '!': |
3222 | s++; | |
3223 | tmp = *s++; | |
3224 | if (tmp == '=') | |
79072805 | 3225 | Eop(OP_NE); |
378cc40b | 3226 | if (tmp == '~') |
79072805 | 3227 | PMop(OP_NOT); |
378cc40b LW |
3228 | s--; |
3229 | OPERATOR('!'); | |
3230 | case '<': | |
3280af22 | 3231 | if (PL_expect != XOPERATOR) { |
93a17b20 | 3232 | if (s[1] != '<' && !strchr(s,'>')) |
2f3197b3 | 3233 | check_uni(); |
79072805 LW |
3234 | if (s[1] == '<') |
3235 | s = scan_heredoc(s); | |
3236 | else | |
3237 | s = scan_inputsymbol(s); | |
3238 | TERM(sublex_start()); | |
378cc40b LW |
3239 | } |
3240 | s++; | |
3241 | tmp = *s++; | |
3242 | if (tmp == '<') | |
79072805 | 3243 | SHop(OP_LEFT_SHIFT); |
395c3793 LW |
3244 | if (tmp == '=') { |
3245 | tmp = *s++; | |
3246 | if (tmp == '>') | |
79072805 | 3247 | Eop(OP_NCMP); |
395c3793 | 3248 | s--; |
79072805 | 3249 | Rop(OP_LE); |
395c3793 | 3250 | } |
378cc40b | 3251 | s--; |
79072805 | 3252 | Rop(OP_LT); |
378cc40b LW |
3253 | case '>': |
3254 | s++; | |
3255 | tmp = *s++; | |
3256 | if (tmp == '>') | |
79072805 | 3257 | SHop(OP_RIGHT_SHIFT); |
378cc40b | 3258 | if (tmp == '=') |
79072805 | 3259 | Rop(OP_GE); |
378cc40b | 3260 | s--; |
79072805 | 3261 | Rop(OP_GT); |
378cc40b LW |
3262 | |
3263 | case '$': | |
bbce6d69 | 3264 | CLINE; |
3265 | ||
3280af22 NIS |
3266 | if (PL_expect == XOPERATOR) { |
3267 | if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) { | |
3268 | PL_expect = XTERM; | |
a0d0e21e | 3269 | depcom(); |
bbce6d69 | 3270 | return ','; /* grandfather non-comma-format format */ |
a0d0e21e | 3271 | } |
8990e307 | 3272 | } |
a0d0e21e | 3273 | |
7e2040f0 | 3274 | if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) { |
3280af22 | 3275 | PL_tokenbuf[0] = '@'; |
376b8730 SM |
3276 | s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1, |
3277 | sizeof PL_tokenbuf - 1, FALSE); | |
3278 | if (PL_expect == XOPERATOR) | |
3279 | no_op("Array length", s); | |
3280af22 | 3280 | if (!PL_tokenbuf[1]) |
a0d0e21e | 3281 | PREREF(DOLSHARP); |
3280af22 NIS |
3282 | PL_expect = XOPERATOR; |
3283 | PL_pending_ident = '#'; | |
463ee0b2 | 3284 | TOKEN(DOLSHARP); |
79072805 | 3285 | } |
bbce6d69 | 3286 | |
3280af22 | 3287 | PL_tokenbuf[0] = '$'; |
376b8730 SM |
3288 | s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, |
3289 | sizeof PL_tokenbuf - 1, FALSE); | |
3290 | if (PL_expect == XOPERATOR) | |
3291 | no_op("Scalar", s); | |
3280af22 NIS |
3292 | if (!PL_tokenbuf[1]) { |
3293 | if (s == PL_bufend) | |
bbce6d69 | 3294 | yyerror("Final $ should be \\$ or $name"); |
3295 | PREREF('$'); | |
8990e307 | 3296 | } |
a0d0e21e | 3297 | |
bbce6d69 | 3298 | /* This kludge not intended to be bulletproof. */ |
3280af22 | 3299 | if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) { |
bbce6d69 | 3300 | yylval.opval = newSVOP(OP_CONST, 0, |
b448e4fe | 3301 | newSViv(PL_compiling.cop_arybase)); |
bbce6d69 | 3302 | yylval.opval->op_private = OPpCONST_ARYBASE; |
3303 | TERM(THING); | |
3304 | } | |
3305 | ||
ff68c719 | 3306 | d = s; |
69d2bceb | 3307 | tmp = (I32)*s; |
3280af22 | 3308 | if (PL_lex_state == LEX_NORMAL) |
ff68c719 | 3309 | s = skipspace(s); |
3310 | ||
3280af22 | 3311 | if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) { |
bbce6d69 | 3312 | char *t; |
3313 | if (*s == '[') { | |
3280af22 | 3314 | PL_tokenbuf[0] = '@'; |
599cee73 | 3315 | if (ckWARN(WARN_SYNTAX)) { |
bbce6d69 | 3316 | for(t = s + 1; |
7e2040f0 | 3317 | isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$'; |
bbce6d69 | 3318 | t++) ; |
a0d0e21e | 3319 | if (*t++ == ',') { |
3280af22 NIS |
3320 | PL_bufptr = skipspace(PL_bufptr); |
3321 | while (t < PL_bufend && *t != ']') | |
bbce6d69 | 3322 | t++; |
cea2e8a9 | 3323 | Perl_warner(aTHX_ WARN_SYNTAX, |
599cee73 PM |
3324 | "Multidimensional syntax %.*s not supported", |
3325 | (t - PL_bufptr) + 1, PL_bufptr); | |
a0d0e21e LW |
3326 | } |
3327 | } | |
bbce6d69 | 3328 | } |
3329 | else if (*s == '{') { | |
3280af22 | 3330 | PL_tokenbuf[0] = '%'; |
599cee73 | 3331 | if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") && |
bbce6d69 | 3332 | (t = strchr(s, '}')) && (t = strchr(t, '='))) |
3333 | { | |
3280af22 | 3334 | char tmpbuf[sizeof PL_tokenbuf]; |
a0d0e21e LW |
3335 | STRLEN len; |
3336 | for (t++; isSPACE(*t); t++) ; | |
7e2040f0 | 3337 | if (isIDFIRST_lazy_if(t,UTF)) { |
8903cb82 | 3338 | t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len); |
59a6d928 | 3339 | for (; isSPACE(*t); t++) ; |
864dbfa3 | 3340 | if (*t == ';' && get_cv(tmpbuf, FALSE)) |
cea2e8a9 | 3341 | Perl_warner(aTHX_ WARN_SYNTAX, |
599cee73 | 3342 | "You need to quote \"%s\"", tmpbuf); |
748a9306 | 3343 | } |
93a17b20 LW |
3344 | } |
3345 | } | |
2f3197b3 | 3346 | } |
bbce6d69 | 3347 | |
3280af22 | 3348 | PL_expect = XOPERATOR; |
69d2bceb | 3349 | if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) { |
3280af22 NIS |
3350 | bool islop = (PL_last_lop == PL_oldoldbufptr); |
3351 | if (!islop || PL_last_lop_op == OP_GREPSTART) | |
3352 | PL_expect = XOPERATOR; | |
bbce6d69 | 3353 | else if (strchr("$@\"'`q", *s)) |
3280af22 | 3354 | PL_expect = XTERM; /* e.g. print $fh "foo" */ |
7e2040f0 | 3355 | else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF)) |
3280af22 | 3356 | PL_expect = XTERM; /* e.g. print $fh &sub */ |
7e2040f0 | 3357 | else if (isIDFIRST_lazy_if(s,UTF)) { |
3280af22 | 3358 | char tmpbuf[sizeof PL_tokenbuf]; |
8903cb82 | 3359 | scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len); |
155aba94 | 3360 | if ((tmp = keyword(tmpbuf, len))) { |
84902520 TB |
3361 | /* binary operators exclude handle interpretations */ |
3362 | switch (tmp) { | |
3363 | case -KEY_x: | |
3364 | case -KEY_eq: | |
3365 | case -KEY_ne: | |
3366 | case -KEY_gt: | |
3367 | case -KEY_lt: | |
3368 | case -KEY_ge: | |
3369 | case -KEY_le: | |
3370 | case -KEY_cmp: | |
3371 | break; | |
3372 | default: | |
3280af22 | 3373 | PL_expect = XTERM; /* e.g. print $fh length() */ |
84902520 TB |
3374 | break; |
3375 | } | |
3376 | } | |
68dc0745 | 3377 | else { |
3378 | GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV); | |
3379 | if (gv && GvCVu(gv)) | |
3280af22 | 3380 | PL_expect = XTERM; /* e.g. print $fh subr() */ |
93a17b20 | 3381 | } |
93a17b20 | 3382 | } |
bbce6d69 | 3383 | else if (isDIGIT(*s)) |
3280af22 | 3384 | PL_expect = XTERM; /* e.g. print $fh 3 */ |
bbce6d69 | 3385 | else if (*s == '.' && isDIGIT(s[1])) |
3280af22 | 3386 | PL_expect = XTERM; /* e.g. print $fh .3 */ |
e0587a03 | 3387 | else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=') |
3280af22 | 3388 | PL_expect = XTERM; /* e.g. print $fh -1 */ |
e0587a03 | 3389 | else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=') |
3280af22 | 3390 | PL_expect = XTERM; /* print $fh <<"EOF" */ |
bbce6d69 | 3391 | } |
3280af22 | 3392 | PL_pending_ident = '$'; |
79072805 | 3393 | TOKEN('$'); |
378cc40b LW |
3394 | |
3395 | case '@': | |
3280af22 | 3396 | if (PL_expect == XOPERATOR) |
bbce6d69 | 3397 | no_op("Array", s); |
3280af22 NIS |
3398 | PL_tokenbuf[0] = '@'; |
3399 | s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE); | |
3400 | if (!PL_tokenbuf[1]) { | |
3401 | if (s == PL_bufend) | |
bbce6d69 | 3402 | yyerror("Final @ should be \\@ or @name"); |
3403 | PREREF('@'); | |
3404 | } | |
3280af22 | 3405 | if (PL_lex_state == LEX_NORMAL) |
ff68c719 | 3406 | s = skipspace(s); |
3280af22 | 3407 | if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) { |
bbce6d69 | 3408 | if (*s == '{') |
3280af22 | 3409 | PL_tokenbuf[0] = '%'; |
a0d0e21e LW |
3410 | |
3411 | /* Warn about @ where they meant $. */ | |
599cee73 | 3412 | if (ckWARN(WARN_SYNTAX)) { |
a0d0e21e LW |
3413 | if (*s == '[' || *s == '{') { |
3414 | char *t = s + 1; | |
7e2040f0 | 3415 | while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t))) |
a0d0e21e LW |
3416 | t++; |
3417 | if (*t == '}' || *t == ']') { | |
3418 | t++; | |
3280af22 | 3419 | PL_bufptr = skipspace(PL_bufptr); |
cea2e8a9 | 3420 | Perl_warner(aTHX_ WARN_SYNTAX, |
599cee73 | 3421 | "Scalar value %.*s better written as $%.*s", |
3280af22 | 3422 | t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1); |
a0d0e21e | 3423 | } |
93a17b20 LW |
3424 | } |
3425 | } | |
463ee0b2 | 3426 | } |
3280af22 | 3427 | PL_pending_ident = '@'; |
79072805 | 3428 | TERM('@'); |
378cc40b LW |
3429 | |
3430 | case '/': /* may either be division or pattern */ | |
3431 | case '?': /* may either be conditional or pattern */ | |
3280af22 | 3432 | if (PL_expect != XOPERATOR) { |
c277df42 | 3433 | /* Disable warning on "study /blah/" */ |
3280af22 NIS |
3434 | if (PL_oldoldbufptr == PL_last_uni |
3435 | && (*PL_last_uni != 's' || s - PL_last_uni < 5 | |
7e2040f0 GS |
3436 | || memNE(PL_last_uni, "study", 5) |
3437 | || isALNUM_lazy_if(PL_last_uni+5,UTF))) | |
c277df42 | 3438 | check_uni(); |
8782bef2 | 3439 | s = scan_pat(s,OP_MATCH); |
79072805 | 3440 | TERM(sublex_start()); |
378cc40b LW |
3441 | } |
3442 | tmp = *s++; | |
a687059c | 3443 | if (tmp == '/') |
79072805 | 3444 | Mop(OP_DIVIDE); |
378cc40b LW |
3445 | OPERATOR(tmp); |
3446 | ||
3447 | case '.': | |
51882d45 GS |
3448 | if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack |
3449 | #ifdef PERL_STRICT_CR | |
3450 | && s[1] == '\n' | |
3451 | #else | |
3452 | && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n')) | |
3453 | #endif | |
3454 | && (s == PL_linestart || s[-1] == '\n') ) | |
3455 | { | |
3280af22 NIS |
3456 | PL_lex_formbrack = 0; |
3457 | PL_expect = XSTATE; | |
79072805 LW |
3458 | goto rightbracket; |
3459 | } | |
3280af22 | 3460 | if (PL_expect == XOPERATOR || !isDIGIT(s[1])) { |
378cc40b | 3461 | tmp = *s++; |
a687059c LW |
3462 | if (*s == tmp) { |
3463 | s++; | |
2f3197b3 LW |
3464 | if (*s == tmp) { |
3465 | s++; | |
79072805 | 3466 | yylval.ival = OPf_SPECIAL; |
2f3197b3 LW |
3467 | } |
3468 | else | |
79072805 | 3469 | yylval.ival = 0; |
378cc40b | 3470 | OPERATOR(DOTDOT); |
a687059c | 3471 | } |
3280af22 | 3472 | if (PL_expect != XOPERATOR) |
2f3197b3 | 3473 | check_uni(); |
79072805 | 3474 | Aop(OP_CONCAT); |
378cc40b LW |
3475 | } |
3476 | /* FALL THROUGH */ | |
3477 | case '0': case '1': case '2': case '3': case '4': | |
3478 | case '5': case '6': case '7': case '8': case '9': | |
79072805 | 3479 | s = scan_num(s); |
3280af22 | 3480 | if (PL_expect == XOPERATOR) |
8990e307 | 3481 | no_op("Number",s); |
79072805 LW |
3482 | TERM(THING); |
3483 | ||
3484 | case '\'': | |
09bef843 | 3485 | s = scan_str(s,FALSE,FALSE); |
3280af22 NIS |
3486 | if (PL_expect == XOPERATOR) { |
3487 | if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) { | |
3488 | PL_expect = XTERM; | |
a0d0e21e LW |
3489 | depcom(); |
3490 | return ','; /* grandfather non-comma-format format */ | |
3491 | } | |
463ee0b2 | 3492 | else |
8990e307 | 3493 | no_op("String",s); |
463ee0b2 | 3494 | } |
79072805 | 3495 | if (!s) |
85e6fe83 | 3496 | missingterm((char*)0); |
79072805 LW |
3497 | yylval.ival = OP_CONST; |
3498 | TERM(sublex_start()); | |
3499 | ||
3500 | case '"': | |
09bef843 | 3501 | s = scan_str(s,FALSE,FALSE); |
3280af22 NIS |
3502 | if (PL_expect == XOPERATOR) { |
3503 | if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) { | |
3504 | PL_expect = XTERM; | |
a0d0e21e LW |
3505 | depcom(); |
3506 | return ','; /* grandfather non-comma-format format */ | |
3507 | } | |
463ee0b2 | 3508 | else |
8990e307 | 3509 | no_op("String",s); |
463ee0b2 | 3510 | } |
79072805 | 3511 | if (!s) |
85e6fe83 | 3512 | missingterm((char*)0); |
4633a7c4 | 3513 | yylval.ival = OP_CONST; |
3280af22 | 3514 | for (d = SvPV(PL_lex_stuff, len); len; len--, d++) { |
a0ed51b3 | 3515 | if (*d == '$' || *d == '@' || *d == '\\' || *d & 0x80) { |
4633a7c4 LW |
3516 | yylval.ival = OP_STRINGIFY; |
3517 | break; | |
3518 | } | |
3519 | } | |
79072805 LW |
3520 | TERM(sublex_start()); |
3521 | ||
3522 | case '`': | |
09bef843 | 3523 | s = scan_str(s,FALSE,FALSE); |
3280af22 | 3524 | if (PL_expect == XOPERATOR) |
8990e307 | 3525 | no_op("Backticks",s); |
79072805 | 3526 | if (!s) |
85e6fe83 | 3527 | missingterm((char*)0); |
79072805 LW |
3528 | yylval.ival = OP_BACKTICK; |
3529 | set_csh(); | |
3530 | TERM(sublex_start()); | |
3531 | ||
3532 | case '\\': | |
3533 | s++; | |
599cee73 | 3534 | if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s)) |
cea2e8a9 | 3535 | Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression", |
599cee73 | 3536 | *s, *s); |
3280af22 | 3537 | if (PL_expect == XOPERATOR) |
8990e307 | 3538 | no_op("Backslash",s); |
79072805 LW |
3539 | OPERATOR(REFGEN); |
3540 | ||
a7cb1f99 | 3541 | case 'v': |
e526c9e6 | 3542 | if (isDIGIT(s[1]) && PL_expect != XOPERATOR) { |
a7cb1f99 GS |
3543 | char *start = s; |
3544 | start++; | |
3545 | start++; | |
dd629d5b | 3546 | while (isDIGIT(*start) || *start == '_') |
a7cb1f99 GS |
3547 | start++; |
3548 | if (*start == '.' && isDIGIT(start[1])) { | |
3549 | s = scan_num(s); | |
3550 | TERM(THING); | |
3551 | } | |
e526c9e6 GS |
3552 | /* avoid v123abc() or $h{v1}, allow C<print v10;> */ |
3553 | else if (!isALPHA(*start) && (PL_expect == XTERM || PL_expect == XREF)) { | |
3554 | char c = *start; | |
3555 | GV *gv; | |
3556 | *start = '\0'; | |
3557 | gv = gv_fetchpv(s, FALSE, SVt_PVCV); | |
3558 | *start = c; | |
3559 | if (!gv) { | |
3560 | s = scan_num(s); | |
3561 | TERM(THING); | |
3562 | } | |
3563 | } | |
a7cb1f99 GS |
3564 | } |
3565 | goto keylookup; | |
79072805 | 3566 | case 'x': |
3280af22 | 3567 | if (isDIGIT(s[1]) && PL_expect == XOPERATOR) { |
79072805 LW |
3568 | s++; |
3569 | Mop(OP_REPEAT); | |
2f3197b3 | 3570 | } |
79072805 LW |
3571 | goto keylookup; |
3572 | ||
378cc40b | 3573 | case '_': |
79072805 LW |
3574 | case 'a': case 'A': |
3575 | case 'b': case 'B': | |
3576 | case 'c': case 'C': | |
3577 | case 'd': case 'D': | |
3578 | case 'e': case 'E': | |
3579 | case 'f': case 'F': | |
3580 | case 'g': case 'G': | |
3581 | case 'h': case 'H': | |
3582 | case 'i': case 'I': | |
3583 | case 'j': case 'J': | |
3584 | case 'k': case 'K': | |
3585 | case 'l': case 'L': | |
3586 | case 'm': case 'M': | |
3587 | case 'n': case 'N': | |
3588 | case 'o': case 'O': | |
3589 | case 'p': case 'P': | |
3590 | case 'q': case 'Q': | |
3591 | case 'r': case 'R': | |
3592 | case 's': case 'S': | |
3593 | case 't': case 'T': | |
3594 | case 'u': case 'U': | |
a7cb1f99 | 3595 | case 'V': |
79072805 LW |
3596 | case 'w': case 'W': |
3597 | case 'X': | |
3598 | case 'y': case 'Y': | |
3599 | case 'z': case 'Z': | |
3600 | ||
49dc05e3 | 3601 | keylookup: { |