This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Put utf8_downgrade in CRIPPLED_CC versions of SvPVbyte* before I forget.
[perl5.git] / toke.c
CommitLineData
a0d0e21e 1/* toke.c
a687059c 2 *
bc89e66f 3 * Copyright (c) 1991-2001, Larry Wall
a687059c 4 *
d48672a2
LW
5 * You may distribute under the terms of either the GNU General Public
6 * License or the Artistic License, as specified in the README file.
378cc40b 7 *
a0d0e21e
LW
8 */
9
10/*
11 * "It all comes from here, the stench and the peril." --Frodo
378cc40b
LW
12 */
13
9cbb5ea2
GS
14/*
15 * This file is the lexer for Perl. It's closely linked to the
4e553d73 16 * parser, perly.y.
ffb4593c
NT
17 *
18 * The main routine is yylex(), which returns the next token.
19 */
20
378cc40b 21#include "EXTERN.h"
864dbfa3 22#define PERL_IN_TOKE_C
378cc40b 23#include "perl.h"
378cc40b 24
d3b6f988
GS
25#define yychar PL_yychar
26#define yylval PL_yylval
27
fc36a67e 28static char ident_too_long[] = "Identifier too long";
8903cb82 29
51371543 30static void restore_rsfp(pTHXo_ void *f);
6e3aabd6
GS
31#ifndef PERL_NO_UTF16_FILTER
32static I32 utf16_textfilter(pTHXo_ int idx, SV *sv, int maxlen);
33static I32 utf16rev_textfilter(pTHXo_ int idx, SV *sv, int maxlen);
34#endif
51371543 35
9059aa12
LW
36#define XFAKEBRACK 128
37#define XENUMMASK 127
38
7e2040f0 39/*#define UTF (SvUTF8(PL_linestr) && !(PL_hints & HINT_BYTE))*/
a0ed51b3
LW
40#define UTF (PL_hints & HINT_UTF8)
41
4e553d73 42/* In variables name $^X, these are the legal values for X.
2b92dfce
GS
43 * 1999-02-27 mjd-perl-patch@plover.com */
44#define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
45
bf4acbe4
GS
46/* On MacOS, respect nonbreaking spaces */
47#ifdef MACOS_TRADITIONAL
48#define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
49#else
50#define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
51#endif
52
ffb4593c
NT
53/* LEX_* are values for PL_lex_state, the state of the lexer.
54 * They are arranged oddly so that the guard on the switch statement
79072805
LW
55 * can get by with a single comparison (if the compiler is smart enough).
56 */
57
fb73857a
PP
58/* #define LEX_NOTPARSING 11 is done in perl.h. */
59
55497cff
PP
60#define LEX_NORMAL 10
61#define LEX_INTERPNORMAL 9
62#define LEX_INTERPCASEMOD 8
63#define LEX_INTERPPUSH 7
64#define LEX_INTERPSTART 6
65#define LEX_INTERPEND 5
66#define LEX_INTERPENDMAYBE 4
67#define LEX_INTERPCONCAT 3
68#define LEX_INTERPCONST 2
69#define LEX_FORMLINE 1
70#define LEX_KNOWNEXT 0
79072805 71
79072805
LW
72#ifdef ff_next
73#undef ff_next
d48672a2
LW
74#endif
75
a1a0e61e 76#ifdef USE_PURE_BISON
dba4d153
JH
77# ifndef YYMAXLEVEL
78# define YYMAXLEVEL 100
79# endif
20141f0e
RI
80YYSTYPE* yylval_pointer[YYMAXLEVEL];
81int* yychar_pointer[YYMAXLEVEL];
6f202aea 82int yyactlevel = -1;
22c35a8c
GS
83# undef yylval
84# undef yychar
20141f0e
RI
85# define yylval (*yylval_pointer[yyactlevel])
86# define yychar (*yychar_pointer[yyactlevel])
87# define PERL_YYLEX_PARAM yylval_pointer[yyactlevel],yychar_pointer[yyactlevel]
4e553d73 88# undef yylex
dba4d153 89# define yylex() Perl_yylex_r(aTHX_ yylval_pointer[yyactlevel],yychar_pointer[yyactlevel])
a1a0e61e
TD
90#endif
91
79072805 92#include "keywords.h"
fe14fcc3 93
ffb4593c
NT
94/* CLINE is a macro that ensures PL_copline has a sane value */
95
ae986130
LW
96#ifdef CLINE
97#undef CLINE
98#endif
57843af0 99#define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
3280af22 100
ffb4593c
NT
101/*
102 * Convenience functions to return different tokens and prime the
9cbb5ea2 103 * lexer for the next token. They all take an argument.
ffb4593c
NT
104 *
105 * TOKEN : generic token (used for '(', DOLSHARP, etc)
106 * OPERATOR : generic operator
107 * AOPERATOR : assignment operator
108 * PREBLOCK : beginning the block after an if, while, foreach, ...
109 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
110 * PREREF : *EXPR where EXPR is not a simple identifier
111 * TERM : expression term
112 * LOOPX : loop exiting command (goto, last, dump, etc)
113 * FTST : file test operator
114 * FUN0 : zero-argument function
2d2e263d 115 * FUN1 : not used, except for not, which isn't a UNIOP
ffb4593c
NT
116 * BOop : bitwise or or xor
117 * BAop : bitwise and
118 * SHop : shift operator
119 * PWop : power operator
9cbb5ea2 120 * PMop : pattern-matching operator
ffb4593c
NT
121 * Aop : addition-level operator
122 * Mop : multiplication-level operator
123 * Eop : equality-testing operator
e5edeb50 124 * Rop : relational operator <= != gt
ffb4593c
NT
125 *
126 * Also see LOP and lop() below.
127 */
128
075953c3
JH
129/* Note that REPORT() and REPORT2() will be expressions that supply
130 * their own trailing comma, not suitable for statements as such. */
998054bd 131#ifdef DEBUGGING /* Serve -DT. */
075953c3
JH
132# define REPORT(x,retval) tokereport(x,s,(int)retval),
133# define REPORT2(x,retval) tokereport(x,s, yylval.ival),
998054bd 134#else
075953c3
JH
135# define REPORT(x,retval)
136# define REPORT2(x,retval)
998054bd
SC
137#endif
138
075953c3
JH
139#define TOKEN(retval) return (REPORT2("token",retval) PL_bufptr = s,(int)retval)
140#define OPERATOR(retval) return (REPORT2("operator",retval) PL_expect = XTERM, PL_bufptr = s,(int)retval)
141#define AOPERATOR(retval) return ao((REPORT2("aop",retval) PL_expect = XTERM, PL_bufptr = s,(int)retval))
142#define PREBLOCK(retval) return (REPORT2("preblock",retval) PL_expect = XBLOCK,PL_bufptr = s,(int)retval)
143#define PRETERMBLOCK(retval) return (REPORT2("pretermblock",retval) PL_expect = XTERMBLOCK,PL_bufptr = s,(int)retval)
144#define PREREF(retval) return (REPORT2("preref",retval) PL_expect = XREF,PL_bufptr = s,(int)retval)
145#define TERM(retval) return (CLINE, REPORT2("term",retval) PL_expect = XOPERATOR, PL_bufptr = s,(int)retval)
146#define LOOPX(f) return(yylval.ival=f, REPORT("loopx",f) PL_expect = XTERM,PL_bufptr = s,(int)LOOPEX)
147#define FTST(f) return(yylval.ival=f, REPORT("ftst",f) PL_expect = XTERM,PL_bufptr = s,(int)UNIOP)
148#define FUN0(f) return(yylval.ival = f, REPORT("fun0",f) PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC0)
149#define FUN1(f) return(yylval.ival = f, REPORT("fun1",f) PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC1)
150#define BOop(f) return ao((yylval.ival=f, REPORT("bitorop",f) PL_expect = XTERM,PL_bufptr = s,(int)BITOROP))
151#define BAop(f) return ao((yylval.ival=f, REPORT("bitandop",f) PL_expect = XTERM,PL_bufptr = s,(int)BITANDOP))
152#define SHop(f) return ao((yylval.ival=f, REPORT("shiftop",f) PL_expect = XTERM,PL_bufptr = s,(int)SHIFTOP))
153#define PWop(f) return ao((yylval.ival=f, REPORT("powop",f) PL_expect = XTERM,PL_bufptr = s,(int)POWOP))
154#define PMop(f) return(yylval.ival=f, REPORT("matchop",f) PL_expect = XTERM,PL_bufptr = s,(int)MATCHOP)
155#define Aop(f) return ao((yylval.ival=f, REPORT("add",f) PL_expect = XTERM,PL_bufptr = s,(int)ADDOP))
156#define Mop(f) return ao((yylval.ival=f, REPORT("mul",f) PL_expect = XTERM,PL_bufptr = s,(int)MULOP))
157#define Eop(f) return(yylval.ival=f, REPORT("eq",f) PL_expect = XTERM,PL_bufptr = s,(int)EQOP)
158#define Rop(f) return(yylval.ival=f, REPORT("rel",f) PL_expect = XTERM,PL_bufptr = s,(int)RELOP)
2f3197b3 159
a687059c
LW
160/* This bit of chicanery makes a unary function followed by
161 * a parenthesis into a function with one argument, highest precedence.
162 */
2f3197b3 163#define UNI(f) return(yylval.ival = f, \
075953c3 164 REPORT("uni",f) \
3280af22
NIS
165 PL_expect = XTERM, \
166 PL_bufptr = s, \
167 PL_last_uni = PL_oldbufptr, \
168 PL_last_lop_op = f, \
a687059c
LW
169 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
170
79072805 171#define UNIBRACK(f) return(yylval.ival = f, \
075953c3 172 REPORT("uni",f) \
3280af22
NIS
173 PL_bufptr = s, \
174 PL_last_uni = PL_oldbufptr, \
79072805
LW
175 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
176
9f68db38 177/* grandfather return to old style */
3280af22 178#define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
79072805 179
2d00ba3b 180STATIC void
61b2116b 181S_tokereport(pTHX_ char *thing, char* s, I32 rv)
9041c2e3 182{
998054bd
SC
183 SV *report;
184 DEBUG_T({
185 report = newSVpv(thing, 0);
61b2116b 186 Perl_sv_catpvf(aTHX_ report, ":line %i:%i:", CopLINE(PL_curcop), rv);
998054bd
SC
187
188 if (s - PL_bufptr > 0)
189 sv_catpvn(report, PL_bufptr, s - PL_bufptr);
190 else {
191 if (PL_oldbufptr && *PL_oldbufptr)
192 sv_catpv(report, PL_tokenbuf);
193 }
194 PerlIO_printf(Perl_debug_log, "### %s\n", SvPV_nolen(report));
195 })
196}
197
ffb4593c
NT
198/*
199 * S_ao
200 *
201 * This subroutine detects &&= and ||= and turns an ANDAND or OROR
202 * into an OP_ANDASSIGN or OP_ORASSIGN
203 */
204
76e3520e 205STATIC int
cea2e8a9 206S_ao(pTHX_ int toketype)
a0d0e21e 207{
3280af22
NIS
208 if (*PL_bufptr == '=') {
209 PL_bufptr++;
a0d0e21e
LW
210 if (toketype == ANDAND)
211 yylval.ival = OP_ANDASSIGN;
212 else if (toketype == OROR)
213 yylval.ival = OP_ORASSIGN;
214 toketype = ASSIGNOP;
215 }
216 return toketype;
217}
218
ffb4593c
NT
219/*
220 * S_no_op
221 * When Perl expects an operator and finds something else, no_op
222 * prints the warning. It always prints "<something> found where
223 * operator expected. It prints "Missing semicolon on previous line?"
224 * if the surprise occurs at the start of the line. "do you need to
225 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
226 * where the compiler doesn't know if foo is a method call or a function.
227 * It prints "Missing operator before end of line" if there's nothing
228 * after the missing operator, or "... before <...>" if there is something
229 * after the missing operator.
230 */
231
76e3520e 232STATIC void
cea2e8a9 233S_no_op(pTHX_ char *what, char *s)
463ee0b2 234{
3280af22
NIS
235 char *oldbp = PL_bufptr;
236 bool is_first = (PL_oldbufptr == PL_linestart);
68dc0745 237
1189a94a
GS
238 if (!s)
239 s = oldbp;
07c798fb 240 else
1189a94a 241 PL_bufptr = s;
cea2e8a9 242 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
748a9306 243 if (is_first)
cea2e8a9 244 Perl_warn(aTHX_ "\t(Missing semicolon on previous line?)\n");
7e2040f0 245 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
748a9306 246 char *t;
7e2040f0 247 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy_if(t,UTF) || *t == ':'); t++) ;
3280af22 248 if (t < PL_bufptr && isSPACE(*t))
cea2e8a9 249 Perl_warn(aTHX_ "\t(Do you need to predeclare %.*s?)\n",
3280af22 250 t - PL_oldoldbufptr, PL_oldoldbufptr);
748a9306 251 }
07c798fb
HS
252 else {
253 assert(s >= oldbp);
cea2e8a9 254 Perl_warn(aTHX_ "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp);
07c798fb 255 }
3280af22 256 PL_bufptr = oldbp;
8990e307
LW
257}
258
ffb4593c
NT
259/*
260 * S_missingterm
261 * Complain about missing quote/regexp/heredoc terminator.
262 * If it's called with (char *)NULL then it cauterizes the line buffer.
263 * If we're in a delimited string and the delimiter is a control
264 * character, it's reformatted into a two-char sequence like ^C.
265 * This is fatal.
266 */
267
76e3520e 268STATIC void
cea2e8a9 269S_missingterm(pTHX_ char *s)
8990e307
LW
270{
271 char tmpbuf[3];
272 char q;
273 if (s) {
274 char *nl = strrchr(s,'\n');
d2719217 275 if (nl)
8990e307
LW
276 *nl = '\0';
277 }
9d116dd7
JH
278 else if (
279#ifdef EBCDIC
280 iscntrl(PL_multi_close)
281#else
282 PL_multi_close < 32 || PL_multi_close == 127
283#endif
284 ) {
8990e307 285 *tmpbuf = '^';
3280af22 286 tmpbuf[1] = toCTRL(PL_multi_close);
8990e307
LW
287 s = "\\n";
288 tmpbuf[2] = '\0';
289 s = tmpbuf;
290 }
291 else {
3280af22 292 *tmpbuf = PL_multi_close;
8990e307
LW
293 tmpbuf[1] = '\0';
294 s = tmpbuf;
295 }
296 q = strchr(s,'"') ? '\'' : '"';
cea2e8a9 297 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
463ee0b2 298}
79072805 299
ffb4593c
NT
300/*
301 * Perl_deprecate
ffb4593c
NT
302 */
303
79072805 304void
864dbfa3 305Perl_deprecate(pTHX_ char *s)
a0d0e21e 306{
599cee73 307 if (ckWARN(WARN_DEPRECATED))
cea2e8a9 308 Perl_warner(aTHX_ WARN_DEPRECATED, "Use of %s is deprecated", s);
a0d0e21e
LW
309}
310
ffb4593c
NT
311/*
312 * depcom
9cbb5ea2 313 * Deprecate a comma-less variable list.
ffb4593c
NT
314 */
315
76e3520e 316STATIC void
cea2e8a9 317S_depcom(pTHX)
a0d0e21e
LW
318{
319 deprecate("comma-less variable list");
320}
321
ffb4593c 322/*
9cbb5ea2
GS
323 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
324 * utf16-to-utf8-reversed.
ffb4593c
NT
325 */
326
c39cd008
GS
327#ifdef PERL_CR_FILTER
328static void
329strip_return(SV *sv)
330{
331 register char *s = SvPVX(sv);
332 register char *e = s + SvCUR(sv);
333 /* outer loop optimized to do nothing if there are no CR-LFs */
334 while (s < e) {
335 if (*s++ == '\r' && *s == '\n') {
336 /* hit a CR-LF, need to copy the rest */
337 register char *d = s - 1;
338 *d++ = *s++;
339 while (s < e) {
340 if (*s == '\r' && s[1] == '\n')
341 s++;
342 *d++ = *s++;
343 }
344 SvCUR(sv) -= s - d;
345 return;
346 }
347 }
348}
a868473f 349
76e3520e 350STATIC I32
c39cd008 351S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
a868473f 352{
c39cd008
GS
353 I32 count = FILTER_READ(idx+1, sv, maxlen);
354 if (count > 0 && !maxlen)
355 strip_return(sv);
356 return count;
a868473f
NIS
357}
358#endif
359
ffb4593c
NT
360/*
361 * Perl_lex_start
9cbb5ea2
GS
362 * Initialize variables. Uses the Perl save_stack to save its state (for
363 * recursive calls to the parser).
ffb4593c
NT
364 */
365
a0d0e21e 366void
864dbfa3 367Perl_lex_start(pTHX_ SV *line)
79072805 368{
8990e307
LW
369 char *s;
370 STRLEN len;
371
3280af22
NIS
372 SAVEI32(PL_lex_dojoin);
373 SAVEI32(PL_lex_brackets);
3280af22
NIS
374 SAVEI32(PL_lex_casemods);
375 SAVEI32(PL_lex_starts);
376 SAVEI32(PL_lex_state);
7766f137 377 SAVEVPTR(PL_lex_inpat);
3280af22 378 SAVEI32(PL_lex_inwhat);
18b09519
GS
379 if (PL_lex_state == LEX_KNOWNEXT) {
380 I32 toke = PL_nexttoke;
381 while (--toke >= 0) {
382 SAVEI32(PL_nexttype[toke]);
383 SAVEVPTR(PL_nextval[toke]);
384 }
385 SAVEI32(PL_nexttoke);
18b09519 386 }
57843af0 387 SAVECOPLINE(PL_curcop);
3280af22
NIS
388 SAVEPPTR(PL_bufptr);
389 SAVEPPTR(PL_bufend);
390 SAVEPPTR(PL_oldbufptr);
391 SAVEPPTR(PL_oldoldbufptr);
207e3d1a
JH
392 SAVEPPTR(PL_last_lop);
393 SAVEPPTR(PL_last_uni);
3280af22
NIS
394 SAVEPPTR(PL_linestart);
395 SAVESPTR(PL_linestr);
396 SAVEPPTR(PL_lex_brackstack);
397 SAVEPPTR(PL_lex_casestack);
c76ac1ee 398 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
3280af22
NIS
399 SAVESPTR(PL_lex_stuff);
400 SAVEI32(PL_lex_defer);
09bef843 401 SAVEI32(PL_sublex_info.sub_inwhat);
3280af22 402 SAVESPTR(PL_lex_repl);
bebdddfc
GS
403 SAVEINT(PL_expect);
404 SAVEINT(PL_lex_expect);
3280af22
NIS
405
406 PL_lex_state = LEX_NORMAL;
407 PL_lex_defer = 0;
408 PL_expect = XSTATE;
409 PL_lex_brackets = 0;
3280af22
NIS
410 New(899, PL_lex_brackstack, 120, char);
411 New(899, PL_lex_casestack, 12, char);
412 SAVEFREEPV(PL_lex_brackstack);
413 SAVEFREEPV(PL_lex_casestack);
414 PL_lex_casemods = 0;
415 *PL_lex_casestack = '\0';
416 PL_lex_dojoin = 0;
417 PL_lex_starts = 0;
418 PL_lex_stuff = Nullsv;
419 PL_lex_repl = Nullsv;
420 PL_lex_inpat = 0;
76be56bc 421 PL_nexttoke = 0;
3280af22 422 PL_lex_inwhat = 0;
09bef843 423 PL_sublex_info.sub_inwhat = 0;
3280af22
NIS
424 PL_linestr = line;
425 if (SvREADONLY(PL_linestr))
426 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
427 s = SvPV(PL_linestr, len);
8990e307 428 if (len && s[len-1] != ';') {
3280af22
NIS
429 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
430 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
431 sv_catpvn(PL_linestr, "\n;", 2);
8990e307 432 }
3280af22
NIS
433 SvTEMP_off(PL_linestr);
434 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
435 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
207e3d1a 436 PL_last_lop = PL_last_uni = Nullch;
3280af22 437 SvREFCNT_dec(PL_rs);
79cb57f6 438 PL_rs = newSVpvn("\n", 1);
3280af22 439 PL_rsfp = 0;
79072805 440}
a687059c 441
ffb4593c
NT
442/*
443 * Perl_lex_end
9cbb5ea2
GS
444 * Finalizer for lexing operations. Must be called when the parser is
445 * done with the lexer.
ffb4593c
NT
446 */
447
463ee0b2 448void
864dbfa3 449Perl_lex_end(pTHX)
463ee0b2 450{
3280af22 451 PL_doextract = FALSE;
463ee0b2
LW
452}
453
ffb4593c
NT
454/*
455 * S_incline
456 * This subroutine has nothing to do with tilting, whether at windmills
457 * or pinball tables. Its name is short for "increment line". It
57843af0 458 * increments the current line number in CopLINE(PL_curcop) and checks
ffb4593c 459 * to see whether the line starts with a comment of the form
9cbb5ea2
GS
460 * # line 500 "foo.pm"
461 * If so, it sets the current line number and file to the values in the comment.
ffb4593c
NT
462 */
463
76e3520e 464STATIC void
cea2e8a9 465S_incline(pTHX_ char *s)
463ee0b2
LW
466{
467 char *t;
468 char *n;
73659bf1 469 char *e;
463ee0b2 470 char ch;
463ee0b2 471
57843af0 472 CopLINE_inc(PL_curcop);
463ee0b2
LW
473 if (*s++ != '#')
474 return;
bf4acbe4 475 while (SPACE_OR_TAB(*s)) s++;
73659bf1
GS
476 if (strnEQ(s, "line", 4))
477 s += 4;
478 else
479 return;
084592ab 480 if (SPACE_OR_TAB(*s))
73659bf1 481 s++;
4e553d73 482 else
73659bf1 483 return;
bf4acbe4 484 while (SPACE_OR_TAB(*s)) s++;
463ee0b2
LW
485 if (!isDIGIT(*s))
486 return;
487 n = s;
488 while (isDIGIT(*s))
489 s++;
bf4acbe4 490 while (SPACE_OR_TAB(*s))
463ee0b2 491 s++;
73659bf1 492 if (*s == '"' && (t = strchr(s+1, '"'))) {
463ee0b2 493 s++;
73659bf1
GS
494 e = t + 1;
495 }
463ee0b2 496 else {
463ee0b2 497 for (t = s; !isSPACE(*t); t++) ;
73659bf1 498 e = t;
463ee0b2 499 }
bf4acbe4 500 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
73659bf1
GS
501 e++;
502 if (*e != '\n' && *e != '\0')
503 return; /* false alarm */
504
463ee0b2
LW
505 ch = *t;
506 *t = '\0';
f4dd75d9
GS
507 if (t - s > 0) {
508#ifdef USE_ITHREADS
509 Safefree(CopFILE(PL_curcop));
510#else
511 SvREFCNT_dec(CopFILEGV(PL_curcop));
512#endif
57843af0 513 CopFILE_set(PL_curcop, s);
f4dd75d9 514 }
463ee0b2 515 *t = ch;
57843af0 516 CopLINE_set(PL_curcop, atoi(n)-1);
463ee0b2
LW
517}
518
ffb4593c
NT
519/*
520 * S_skipspace
521 * Called to gobble the appropriate amount and type of whitespace.
522 * Skips comments as well.
523 */
524
76e3520e 525STATIC char *
cea2e8a9 526S_skipspace(pTHX_ register char *s)
a687059c 527{
3280af22 528 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
bf4acbe4 529 while (s < PL_bufend && SPACE_OR_TAB(*s))
463ee0b2
LW
530 s++;
531 return s;
532 }
533 for (;;) {
fd049845 534 STRLEN prevlen;
09bef843
SB
535 SSize_t oldprevlen, oldoldprevlen;
536 SSize_t oldloplen, oldunilen;
60e6418e
GS
537 while (s < PL_bufend && isSPACE(*s)) {
538 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
539 incline(s);
540 }
ffb4593c
NT
541
542 /* comment */
3280af22
NIS
543 if (s < PL_bufend && *s == '#') {
544 while (s < PL_bufend && *s != '\n')
463ee0b2 545 s++;
60e6418e 546 if (s < PL_bufend) {
463ee0b2 547 s++;
60e6418e
GS
548 if (PL_in_eval && !PL_rsfp) {
549 incline(s);
550 continue;
551 }
552 }
463ee0b2 553 }
ffb4593c
NT
554
555 /* only continue to recharge the buffer if we're at the end
556 * of the buffer, we're not reading from a source filter, and
557 * we're in normal lexing mode
558 */
09bef843
SB
559 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
560 PL_lex_state == LEX_FORMLINE)
463ee0b2 561 return s;
ffb4593c
NT
562
563 /* try to recharge the buffer */
9cbb5ea2
GS
564 if ((s = filter_gets(PL_linestr, PL_rsfp,
565 (prevlen = SvCUR(PL_linestr)))) == Nullch)
566 {
567 /* end of file. Add on the -p or -n magic */
3280af22
NIS
568 if (PL_minus_n || PL_minus_p) {
569 sv_setpv(PL_linestr,PL_minus_p ?
08e9d68e
DD
570 ";}continue{print or die qq(-p destination: $!\\n)" :
571 "");
3280af22
NIS
572 sv_catpv(PL_linestr,";}");
573 PL_minus_n = PL_minus_p = 0;
a0d0e21e
LW
574 }
575 else
3280af22 576 sv_setpv(PL_linestr,";");
ffb4593c
NT
577
578 /* reset variables for next time we lex */
9cbb5ea2
GS
579 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
580 = SvPVX(PL_linestr);
3280af22 581 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
207e3d1a 582 PL_last_lop = PL_last_uni = Nullch;
ffb4593c
NT
583
584 /* Close the filehandle. Could be from -P preprocessor,
585 * STDIN, or a regular file. If we were reading code from
586 * STDIN (because the commandline held no -e or filename)
587 * then we don't close it, we reset it so the code can
588 * read from STDIN too.
589 */
590
3280af22
NIS
591 if (PL_preprocess && !PL_in_eval)
592 (void)PerlProc_pclose(PL_rsfp);
593 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
594 PerlIO_clearerr(PL_rsfp);
8990e307 595 else
3280af22
NIS
596 (void)PerlIO_close(PL_rsfp);
597 PL_rsfp = Nullfp;
463ee0b2
LW
598 return s;
599 }
ffb4593c
NT
600
601 /* not at end of file, so we only read another line */
09bef843
SB
602 /* make corresponding updates to old pointers, for yyerror() */
603 oldprevlen = PL_oldbufptr - PL_bufend;
604 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
605 if (PL_last_uni)
606 oldunilen = PL_last_uni - PL_bufend;
607 if (PL_last_lop)
608 oldloplen = PL_last_lop - PL_bufend;
3280af22
NIS
609 PL_linestart = PL_bufptr = s + prevlen;
610 PL_bufend = s + SvCUR(PL_linestr);
611 s = PL_bufptr;
09bef843
SB
612 PL_oldbufptr = s + oldprevlen;
613 PL_oldoldbufptr = s + oldoldprevlen;
614 if (PL_last_uni)
615 PL_last_uni = s + oldunilen;
616 if (PL_last_lop)
617 PL_last_lop = s + oldloplen;
a0d0e21e 618 incline(s);
ffb4593c
NT
619
620 /* debugger active and we're not compiling the debugger code,
621 * so store the line into the debugger's array of lines
622 */
3280af22 623 if (PERLDB_LINE && PL_curstash != PL_debstash) {
8990e307
LW
624 SV *sv = NEWSV(85,0);
625
626 sv_upgrade(sv, SVt_PVMG);
3280af22 627 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
57843af0 628 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
8990e307 629 }
463ee0b2 630 }
a687059c 631}
378cc40b 632
ffb4593c
NT
633/*
634 * S_check_uni
635 * Check the unary operators to ensure there's no ambiguity in how they're
636 * used. An ambiguous piece of code would be:
637 * rand + 5
638 * This doesn't mean rand() + 5. Because rand() is a unary operator,
639 * the +5 is its argument.
640 */
641
76e3520e 642STATIC void
cea2e8a9 643S_check_uni(pTHX)
ba106d47 644{
2f3197b3 645 char *s;
a0d0e21e 646 char *t;
2f3197b3 647
3280af22 648 if (PL_oldoldbufptr != PL_last_uni)
2f3197b3 649 return;
3280af22
NIS
650 while (isSPACE(*PL_last_uni))
651 PL_last_uni++;
7e2040f0 652 for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ;
3280af22 653 if ((t = strchr(s, '(')) && t < PL_bufptr)
a0d0e21e 654 return;
0453d815 655 if (ckWARN_d(WARN_AMBIGUOUS)){
f248d071 656 char ch = *s;
0453d815 657 *s = '\0';
4e553d73
NIS
658 Perl_warner(aTHX_ WARN_AMBIGUOUS,
659 "Warning: Use of \"%s\" without parens is ambiguous",
0453d815
PM
660 PL_last_uni);
661 *s = ch;
662 }
2f3197b3
LW
663}
664
ffb4593c
NT
665/* workaround to replace the UNI() macro with a function. Only the
666 * hints/uts.sh file mentions this. Other comments elsewhere in the
667 * source indicate Microport Unix might need it too.
668 */
669
ffed7fef
LW
670#ifdef CRIPPLED_CC
671
672#undef UNI
ffed7fef 673#define UNI(f) return uni(f,s)
ffed7fef 674
76e3520e 675STATIC int
cea2e8a9 676S_uni(pTHX_ I32 f, char *s)
ffed7fef
LW
677{
678 yylval.ival = f;
3280af22
NIS
679 PL_expect = XTERM;
680 PL_bufptr = s;
8f872242
NIS
681 PL_last_uni = PL_oldbufptr;
682 PL_last_lop_op = f;
ffed7fef
LW
683 if (*s == '(')
684 return FUNC1;
685 s = skipspace(s);
686 if (*s == '(')
687 return FUNC1;
688 else
689 return UNIOP;
690}
691
a0d0e21e
LW
692#endif /* CRIPPLED_CC */
693
ffb4593c
NT
694/*
695 * LOP : macro to build a list operator. Its behaviour has been replaced
696 * with a subroutine, S_lop() for which LOP is just another name.
697 */
698
a0d0e21e
LW
699#define LOP(f,x) return lop(f,x,s)
700
ffb4593c
NT
701/*
702 * S_lop
703 * Build a list operator (or something that might be one). The rules:
704 * - if we have a next token, then it's a list operator [why?]
705 * - if the next thing is an opening paren, then it's a function
706 * - else it's a list operator
707 */
708
76e3520e 709STATIC I32
a0be28da 710S_lop(pTHX_ I32 f, int x, char *s)
ffed7fef 711{
79072805 712 yylval.ival = f;
35c8bce7 713 CLINE;
075953c3 714 REPORT("lop", f)
3280af22
NIS
715 PL_expect = x;
716 PL_bufptr = s;
717 PL_last_lop = PL_oldbufptr;
718 PL_last_lop_op = f;
719 if (PL_nexttoke)
a0d0e21e 720 return LSTOP;
79072805
LW
721 if (*s == '(')
722 return FUNC;
723 s = skipspace(s);
724 if (*s == '(')
725 return FUNC;
726 else
727 return LSTOP;
728}
729
ffb4593c
NT
730/*
731 * S_force_next
9cbb5ea2 732 * When the lexer realizes it knows the next token (for instance,
ffb4593c 733 * it is reordering tokens for the parser) then it can call S_force_next
9cbb5ea2
GS
734 * to know what token to return the next time the lexer is called. Caller
735 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
736 * handles the token correctly.
ffb4593c
NT
737 */
738
4e553d73 739STATIC void
cea2e8a9 740S_force_next(pTHX_ I32 type)
79072805 741{
3280af22
NIS
742 PL_nexttype[PL_nexttoke] = type;
743 PL_nexttoke++;
744 if (PL_lex_state != LEX_KNOWNEXT) {
745 PL_lex_defer = PL_lex_state;
746 PL_lex_expect = PL_expect;
747 PL_lex_state = LEX_KNOWNEXT;
79072805
LW
748 }
749}
750
ffb4593c
NT
751/*
752 * S_force_word
753 * When the lexer knows the next thing is a word (for instance, it has
754 * just seen -> and it knows that the next char is a word char, then
755 * it calls S_force_word to stick the next word into the PL_next lookahead.
756 *
757 * Arguments:
b1b65b59 758 * char *start : buffer position (must be within PL_linestr)
ffb4593c
NT
759 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
760 * int check_keyword : if true, Perl checks to make sure the word isn't
761 * a keyword (do this if the word is a label, e.g. goto FOO)
762 * int allow_pack : if true, : characters will also be allowed (require,
763 * use, etc. do this)
9cbb5ea2 764 * int allow_initial_tick : used by the "sub" lexer only.
ffb4593c
NT
765 */
766
76e3520e 767STATIC char *
cea2e8a9 768S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
79072805 769{
463ee0b2
LW
770 register char *s;
771 STRLEN len;
4e553d73 772
463ee0b2
LW
773 start = skipspace(start);
774 s = start;
7e2040f0 775 if (isIDFIRST_lazy_if(s,UTF) ||
a0d0e21e 776 (allow_pack && *s == ':') ||
15f0808c 777 (allow_initial_tick && *s == '\'') )
a0d0e21e 778 {
3280af22
NIS
779 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
780 if (check_keyword && keyword(PL_tokenbuf, len))
463ee0b2
LW
781 return start;
782 if (token == METHOD) {
783 s = skipspace(s);
784 if (*s == '(')
3280af22 785 PL_expect = XTERM;
463ee0b2 786 else {
3280af22 787 PL_expect = XOPERATOR;
463ee0b2 788 }
79072805 789 }
3280af22
NIS
790 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0));
791 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
79072805
LW
792 force_next(token);
793 }
794 return s;
795}
796
ffb4593c
NT
797/*
798 * S_force_ident
9cbb5ea2 799 * Called when the lexer wants $foo *foo &foo etc, but the program
ffb4593c
NT
800 * text only contains the "foo" portion. The first argument is a pointer
801 * to the "foo", and the second argument is the type symbol to prefix.
802 * Forces the next token to be a "WORD".
9cbb5ea2 803 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
ffb4593c
NT
804 */
805
76e3520e 806STATIC void
cea2e8a9 807S_force_ident(pTHX_ register char *s, int kind)
79072805
LW
808{
809 if (s && *s) {
11343788 810 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
3280af22 811 PL_nextval[PL_nexttoke].opval = o;
79072805 812 force_next(WORD);
748a9306 813 if (kind) {
11343788 814 o->op_private = OPpCONST_ENTERED;
55497cff
PP
815 /* XXX see note in pp_entereval() for why we forgo typo
816 warnings if the symbol must be introduced in an eval.
817 GSAR 96-10-12 */
3280af22 818 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
a0d0e21e
LW
819 kind == '$' ? SVt_PV :
820 kind == '@' ? SVt_PVAV :
821 kind == '%' ? SVt_PVHV :
822 SVt_PVGV
823 );
748a9306 824 }
79072805
LW
825 }
826}
827
1571675a
GS
828NV
829Perl_str_to_version(pTHX_ SV *sv)
830{
831 NV retval = 0.0;
832 NV nshift = 1.0;
833 STRLEN len;
834 char *start = SvPVx(sv,len);
3aa33fe5 835 bool utf = SvUTF8(sv) ? TRUE : FALSE;
1571675a
GS
836 char *end = start + len;
837 while (start < end) {
ba210ebe 838 STRLEN skip;
1571675a
GS
839 UV n;
840 if (utf)
9041c2e3 841 n = utf8n_to_uvchr((U8*)start, len, &skip, 0);
1571675a
GS
842 else {
843 n = *(U8*)start;
844 skip = 1;
845 }
846 retval += ((NV)n)/nshift;
847 start += skip;
848 nshift *= 1000;
849 }
850 return retval;
851}
852
4e553d73 853/*
ffb4593c
NT
854 * S_force_version
855 * Forces the next token to be a version number.
856 */
857
76e3520e 858STATIC char *
cea2e8a9 859S_force_version(pTHX_ char *s)
89bfa8cd
PP
860{
861 OP *version = Nullop;
44dcb63b 862 char *d;
89bfa8cd
PP
863
864 s = skipspace(s);
865
44dcb63b 866 d = s;
dd629d5b 867 if (*d == 'v')
44dcb63b 868 d++;
44dcb63b 869 if (isDIGIT(*d)) {
a7cb1f99 870 for (; isDIGIT(*d) || *d == '_' || *d == '.'; d++);
9f3d182e 871 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
dd629d5b 872 SV *ver;
b73d6f50 873 s = scan_num(s, &yylval);
89bfa8cd 874 version = yylval.opval;
dd629d5b
GS
875 ver = cSVOPx(version)->op_sv;
876 if (SvPOK(ver) && !SvNIOK(ver)) {
155aba94 877 (void)SvUPGRADE(ver, SVt_PVNV);
1571675a
GS
878 SvNVX(ver) = str_to_version(ver);
879 SvNOK_on(ver); /* hint that it is a version */
44dcb63b 880 }
89bfa8cd
PP
881 }
882 }
883
884 /* NOTE: The parser sees the package name and the VERSION swapped */
3280af22 885 PL_nextval[PL_nexttoke].opval = version;
4e553d73 886 force_next(WORD);
89bfa8cd
PP
887
888 return (s);
889}
890
ffb4593c
NT
891/*
892 * S_tokeq
893 * Tokenize a quoted string passed in as an SV. It finds the next
894 * chunk, up to end of string or a backslash. It may make a new
895 * SV containing that chunk (if HINT_NEW_STRING is on). It also
896 * turns \\ into \.
897 */
898
76e3520e 899STATIC SV *
cea2e8a9 900S_tokeq(pTHX_ SV *sv)
79072805
LW
901{
902 register char *s;
903 register char *send;
904 register char *d;
b3ac6de7
IZ
905 STRLEN len = 0;
906 SV *pv = sv;
79072805
LW
907
908 if (!SvLEN(sv))
b3ac6de7 909 goto finish;
79072805 910
a0d0e21e 911 s = SvPV_force(sv, len);
21a311ee 912 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
b3ac6de7 913 goto finish;
463ee0b2 914 send = s + len;
79072805
LW
915 while (s < send && *s != '\\')
916 s++;
917 if (s == send)
b3ac6de7 918 goto finish;
79072805 919 d = s;
3280af22 920 if ( PL_hints & HINT_NEW_STRING )
79cb57f6 921 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
79072805
LW
922 while (s < send) {
923 if (*s == '\\') {
a0d0e21e 924 if (s + 1 < send && (s[1] == '\\'))
79072805
LW
925 s++; /* all that, just for this */
926 }
927 *d++ = *s++;
928 }
929 *d = '\0';
463ee0b2 930 SvCUR_set(sv, d - SvPVX(sv));
b3ac6de7 931 finish:
3280af22 932 if ( PL_hints & HINT_NEW_STRING )
b3ac6de7 933 return new_constant(NULL, 0, "q", sv, pv, "q");
79072805
LW
934 return sv;
935}
936
ffb4593c
NT
937/*
938 * Now come three functions related to double-quote context,
939 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
940 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
941 * interact with PL_lex_state, and create fake ( ... ) argument lists
942 * to handle functions and concatenation.
943 * They assume that whoever calls them will be setting up a fake
944 * join call, because each subthing puts a ',' after it. This lets
945 * "lower \luPpEr"
946 * become
947 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
948 *
949 * (I'm not sure whether the spurious commas at the end of lcfirst's
950 * arguments and join's arguments are created or not).
951 */
952
953/*
954 * S_sublex_start
955 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
956 *
957 * Pattern matching will set PL_lex_op to the pattern-matching op to
958 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
959 *
960 * OP_CONST and OP_READLINE are easy--just make the new op and return.
961 *
962 * Everything else becomes a FUNC.
963 *
964 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
965 * had an OP_CONST or OP_READLINE). This just sets us up for a
966 * call to S_sublex_push().
967 */
968
76e3520e 969STATIC I32
cea2e8a9 970S_sublex_start(pTHX)
79072805
LW
971{
972 register I32 op_type = yylval.ival;
79072805
LW
973
974 if (op_type == OP_NULL) {
3280af22
NIS
975 yylval.opval = PL_lex_op;
976 PL_lex_op = Nullop;
79072805
LW
977 return THING;
978 }
979 if (op_type == OP_CONST || op_type == OP_READLINE) {
3280af22 980 SV *sv = tokeq(PL_lex_stuff);
b3ac6de7
IZ
981
982 if (SvTYPE(sv) == SVt_PVIV) {
983 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
984 STRLEN len;
985 char *p;
986 SV *nsv;
987
988 p = SvPV(sv, len);
79cb57f6 989 nsv = newSVpvn(p, len);
01ec43d0
GS
990 if (SvUTF8(sv))
991 SvUTF8_on(nsv);
b3ac6de7
IZ
992 SvREFCNT_dec(sv);
993 sv = nsv;
4e553d73 994 }
b3ac6de7 995 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
3280af22 996 PL_lex_stuff = Nullsv;
79072805
LW
997 return THING;
998 }
999
3280af22
NIS
1000 PL_sublex_info.super_state = PL_lex_state;
1001 PL_sublex_info.sub_inwhat = op_type;
1002 PL_sublex_info.sub_op = PL_lex_op;
1003 PL_lex_state = LEX_INTERPPUSH;
55497cff 1004
3280af22
NIS
1005 PL_expect = XTERM;
1006 if (PL_lex_op) {
1007 yylval.opval = PL_lex_op;
1008 PL_lex_op = Nullop;
55497cff
PP
1009 return PMFUNC;
1010 }
1011 else
1012 return FUNC;
1013}
1014
ffb4593c
NT
1015/*
1016 * S_sublex_push
1017 * Create a new scope to save the lexing state. The scope will be
1018 * ended in S_sublex_done. Returns a '(', starting the function arguments
1019 * to the uc, lc, etc. found before.
1020 * Sets PL_lex_state to LEX_INTERPCONCAT.
1021 */
1022
76e3520e 1023STATIC I32
cea2e8a9 1024S_sublex_push(pTHX)
55497cff 1025{
f46d017c 1026 ENTER;
55497cff 1027
3280af22
NIS
1028 PL_lex_state = PL_sublex_info.super_state;
1029 SAVEI32(PL_lex_dojoin);
1030 SAVEI32(PL_lex_brackets);
3280af22
NIS
1031 SAVEI32(PL_lex_casemods);
1032 SAVEI32(PL_lex_starts);
1033 SAVEI32(PL_lex_state);
7766f137 1034 SAVEVPTR(PL_lex_inpat);
3280af22 1035 SAVEI32(PL_lex_inwhat);
57843af0 1036 SAVECOPLINE(PL_curcop);
3280af22
NIS
1037 SAVEPPTR(PL_bufptr);
1038 SAVEPPTR(PL_oldbufptr);
1039 SAVEPPTR(PL_oldoldbufptr);
207e3d1a
JH
1040 SAVEPPTR(PL_last_lop);
1041 SAVEPPTR(PL_last_uni);
3280af22
NIS
1042 SAVEPPTR(PL_linestart);
1043 SAVESPTR(PL_linestr);
1044 SAVEPPTR(PL_lex_brackstack);
1045 SAVEPPTR(PL_lex_casestack);
1046
1047 PL_linestr = PL_lex_stuff;
1048 PL_lex_stuff = Nullsv;
1049
9cbb5ea2
GS
1050 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1051 = SvPVX(PL_linestr);
3280af22 1052 PL_bufend += SvCUR(PL_linestr);
207e3d1a 1053 PL_last_lop = PL_last_uni = Nullch;
3280af22
NIS
1054 SAVEFREESV(PL_linestr);
1055
1056 PL_lex_dojoin = FALSE;
1057 PL_lex_brackets = 0;
3280af22
NIS
1058 New(899, PL_lex_brackstack, 120, char);
1059 New(899, PL_lex_casestack, 12, char);
1060 SAVEFREEPV(PL_lex_brackstack);
1061 SAVEFREEPV(PL_lex_casestack);
1062 PL_lex_casemods = 0;
1063 *PL_lex_casestack = '\0';
1064 PL_lex_starts = 0;
1065 PL_lex_state = LEX_INTERPCONCAT;
57843af0 1066 CopLINE_set(PL_curcop, PL_multi_start);
3280af22
NIS
1067
1068 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1069 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1070 PL_lex_inpat = PL_sublex_info.sub_op;
79072805 1071 else
3280af22 1072 PL_lex_inpat = Nullop;
79072805 1073
55497cff 1074 return '(';
79072805
LW
1075}
1076
ffb4593c
NT
1077/*
1078 * S_sublex_done
1079 * Restores lexer state after a S_sublex_push.
1080 */
1081
76e3520e 1082STATIC I32
cea2e8a9 1083S_sublex_done(pTHX)
79072805 1084{
3280af22 1085 if (!PL_lex_starts++) {
9aa983d2
JH
1086 SV *sv = newSVpvn("",0);
1087 if (SvUTF8(PL_linestr))
1088 SvUTF8_on(sv);
3280af22 1089 PL_expect = XOPERATOR;
9aa983d2 1090 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
79072805
LW
1091 return THING;
1092 }
1093
3280af22
NIS
1094 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1095 PL_lex_state = LEX_INTERPCASEMOD;
cea2e8a9 1096 return yylex();
79072805
LW
1097 }
1098
ffb4593c 1099 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
3280af22
NIS
1100 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1101 PL_linestr = PL_lex_repl;
1102 PL_lex_inpat = 0;
1103 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1104 PL_bufend += SvCUR(PL_linestr);
207e3d1a 1105 PL_last_lop = PL_last_uni = Nullch;
3280af22
NIS
1106 SAVEFREESV(PL_linestr);
1107 PL_lex_dojoin = FALSE;
1108 PL_lex_brackets = 0;
3280af22
NIS
1109 PL_lex_casemods = 0;
1110 *PL_lex_casestack = '\0';
1111 PL_lex_starts = 0;
25da4f38 1112 if (SvEVALED(PL_lex_repl)) {
3280af22
NIS
1113 PL_lex_state = LEX_INTERPNORMAL;
1114 PL_lex_starts++;
e9fa98b2
HS
1115 /* we don't clear PL_lex_repl here, so that we can check later
1116 whether this is an evalled subst; that means we rely on the
1117 logic to ensure sublex_done() is called again only via the
1118 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
79072805 1119 }
e9fa98b2 1120 else {
3280af22 1121 PL_lex_state = LEX_INTERPCONCAT;
e9fa98b2
HS
1122 PL_lex_repl = Nullsv;
1123 }
79072805 1124 return ',';
ffed7fef
LW
1125 }
1126 else {
f46d017c 1127 LEAVE;
3280af22
NIS
1128 PL_bufend = SvPVX(PL_linestr);
1129 PL_bufend += SvCUR(PL_linestr);
1130 PL_expect = XOPERATOR;
09bef843 1131 PL_sublex_info.sub_inwhat = 0;
79072805 1132 return ')';
ffed7fef
LW
1133 }
1134}
1135
02aa26ce
NT
1136/*
1137 scan_const
1138
1139 Extracts a pattern, double-quoted string, or transliteration. This
1140 is terrifying code.
1141
3280af22
NIS
1142 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1143 processing a pattern (PL_lex_inpat is true), a transliteration
02aa26ce
NT
1144 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1145
9b599b2a
GS
1146 Returns a pointer to the character scanned up to. Iff this is
1147 advanced from the start pointer supplied (ie if anything was
1148 successfully parsed), will leave an OP for the substring scanned
1149 in yylval. Caller must intuit reason for not parsing further
1150 by looking at the next characters herself.
1151
02aa26ce
NT
1152 In patterns:
1153 backslashes:
1154 double-quoted style: \r and \n
1155 regexp special ones: \D \s
1156 constants: \x3
1157 backrefs: \1 (deprecated in substitution replacements)
1158 case and quoting: \U \Q \E
1159 stops on @ and $, but not for $ as tail anchor
1160
1161 In transliterations:
1162 characters are VERY literal, except for - not at the start or end
1163 of the string, which indicates a range. scan_const expands the
1164 range to the full set of intermediate characters.
1165
1166 In double-quoted strings:
1167 backslashes:
1168 double-quoted style: \r and \n
1169 constants: \x3
1170 backrefs: \1 (deprecated)
1171 case and quoting: \U \Q \E
1172 stops on @ and $
1173
1174 scan_const does *not* construct ops to handle interpolated strings.
1175 It stops processing as soon as it finds an embedded $ or @ variable
1176 and leaves it to the caller to work out what's going on.
1177
1178 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @:foo.
1179
1180 $ in pattern could be $foo or could be tail anchor. Assumption:
1181 it's a tail anchor if $ is the last thing in the string, or if it's
1182 followed by one of ")| \n\t"
1183
1184 \1 (backreferences) are turned into $1
1185
1186 The structure of the code is
1187 while (there's a character to process) {
1188 handle transliteration ranges
1189 skip regexp comments
1190 skip # initiated comments in //x patterns
1191 check for embedded @foo
1192 check for embedded scalars
1193 if (backslash) {
1194 leave intact backslashes from leave (below)
1195 deprecate \1 in strings and sub replacements
1196 handle string-changing backslashes \l \U \Q \E, etc.
1197 switch (what was escaped) {
1198 handle - in a transliteration (becomes a literal -)
1199 handle \132 octal characters
1200 handle 0x15 hex characters
1201 handle \cV (control V)
1202 handle printf backslashes (\f, \r, \n, etc)
1203 } (end switch)
1204 } (end if backslash)
1205 } (end while character to read)
4e553d73 1206
02aa26ce
NT
1207*/
1208
76e3520e 1209STATIC char *
cea2e8a9 1210S_scan_const(pTHX_ char *start)
79072805 1211{
3280af22 1212 register char *send = PL_bufend; /* end of the constant */
02aa26ce
NT
1213 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1214 register char *s = start; /* start of the constant */
1215 register char *d = SvPVX(sv); /* destination for copies */
1216 bool dorange = FALSE; /* are we in a translit range? */
c2e66d9e 1217 bool didrange = FALSE; /* did we just finish a range? */
9aa983d2
JH
1218 bool has_utf8 = (PL_linestr && SvUTF8(PL_linestr));
1219 /* the constant is UTF8 */
012bcf8d
GS
1220 UV uv;
1221
ac2262e3 1222 I32 utf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
a0ed51b3
LW
1223 ? (PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
1224 : UTF;
89491803 1225 I32 this_utf8 = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
9cbb5ea2
GS
1226 ? (PL_sublex_info.sub_op->op_private & (PL_lex_repl ?
1227 OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF))
a0ed51b3 1228 : UTF;
dff6d3cd 1229 const char *leaveit = /* set of acceptably-backslashed characters */
3280af22 1230 PL_lex_inpat
4a2d328f 1231 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
9b599b2a 1232 : "";
79072805
LW
1233
1234 while (s < send || dorange) {
02aa26ce 1235 /* get transliterations out of the way (they're most literal) */
3280af22 1236 if (PL_lex_inwhat == OP_TRANS) {
02aa26ce 1237 /* expand a range A-Z to the full set of characters. AIE! */
79072805 1238 if (dorange) {
1ba5c669
JH
1239 I32 i; /* current expanded character */
1240 I32 min; /* first character in range */
1241 I32 max; /* last character in range */
02aa26ce 1242
8973db79
JH
1243 if (utf) {
1244 char *c = (char*)utf8_hop((U8*)d, -1);
1245 char *e = d++;
1246 while (e-- > c)
1247 *(e + 1) = *e;
8b107d6c 1248 *c = (char)0xff;
8973db79
JH
1249 /* mark the range as done, and continue */
1250 dorange = FALSE;
1251 didrange = TRUE;
1252 continue;
1253 }
02aa26ce 1254 i = d - SvPVX(sv); /* remember current offset */
9cbb5ea2
GS
1255 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1256 d = SvPVX(sv) + i; /* refresh d after realloc */
02aa26ce
NT
1257 d -= 2; /* eat the first char and the - */
1258
8ada0baa
JH
1259 min = (U8)*d; /* first char in range */
1260 max = (U8)d[1]; /* last char in range */
1261
c2e66d9e 1262 if (min > max) {
01ec43d0 1263 Perl_croak(aTHX_
1ba5c669
JH
1264 "Invalid [] range \"%c-%c\" in transliteration operator",
1265 (char)min, (char)max);
c2e66d9e
GS
1266 }
1267
1ba5c669 1268#ifndef ASCIIish
8ada0baa
JH
1269 if ((isLOWER(min) && isLOWER(max)) ||
1270 (isUPPER(min) && isUPPER(max))) {
1271 if (isLOWER(min)) {
1272 for (i = min; i <= max; i++)
1273 if (isLOWER(i))
1274 *d++ = i;
1275 } else {
1276 for (i = min; i <= max; i++)
1277 if (isUPPER(i))
1278 *d++ = i;
1279 }
1280 }
1281 else
1282#endif
1283 for (i = min; i <= max; i++)
1284 *d++ = i;
02aa26ce
NT
1285
1286 /* mark the range as done, and continue */
79072805 1287 dorange = FALSE;
01ec43d0 1288 didrange = TRUE;
79072805 1289 continue;
4e553d73 1290 }
02aa26ce
NT
1291
1292 /* range begins (ignore - as first or last char) */
79072805 1293 else if (*s == '-' && s+1 < send && s != start) {
4e553d73 1294 if (didrange) {
1fafa243 1295 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
01ec43d0 1296 }
a0ed51b3 1297 if (utf) {
a176fa2a 1298 *d++ = (char)0xff; /* use illegal utf8 byte--see pmtrans */
a0ed51b3
LW
1299 s++;
1300 continue;
1301 }
79072805
LW
1302 dorange = TRUE;
1303 s++;
01ec43d0
GS
1304 }
1305 else {
1306 didrange = FALSE;
1307 }
79072805 1308 }
02aa26ce
NT
1309
1310 /* if we get here, we're not doing a transliteration */
1311
0f5d15d6
IZ
1312 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1313 except for the last char, which will be done separately. */
3280af22 1314 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
cc6b7395
IZ
1315 if (s[2] == '#') {
1316 while (s < send && *s != ')')
1317 *d++ = *s++;
155aba94
GS
1318 }
1319 else if (s[2] == '{' /* This should match regcomp.c */
1320 || ((s[2] == 'p' || s[2] == '?') && s[3] == '{'))
1321 {
cc6b7395 1322 I32 count = 1;
0f5d15d6 1323 char *regparse = s + (s[2] == '{' ? 3 : 4);
cc6b7395
IZ
1324 char c;
1325
d9f97599
GS
1326 while (count && (c = *regparse)) {
1327 if (c == '\\' && regparse[1])
1328 regparse++;
4e553d73 1329 else if (c == '{')
cc6b7395 1330 count++;
4e553d73 1331 else if (c == '}')
cc6b7395 1332 count--;
d9f97599 1333 regparse++;
cc6b7395 1334 }
5bdf89e7
IZ
1335 if (*regparse != ')') {
1336 regparse--; /* Leave one char for continuation. */
cc6b7395 1337 yyerror("Sequence (?{...}) not terminated or not {}-balanced");
5bdf89e7 1338 }
0f5d15d6 1339 while (s < regparse)
cc6b7395
IZ
1340 *d++ = *s++;
1341 }
748a9306 1342 }
02aa26ce
NT
1343
1344 /* likewise skip #-initiated comments in //x patterns */
3280af22
NIS
1345 else if (*s == '#' && PL_lex_inpat &&
1346 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
748a9306
LW
1347 while (s+1 < send && *s != '\n')
1348 *d++ = *s++;
1349 }
02aa26ce 1350
5d1d4326
JH
1351 /* check for embedded arrays
1352 (@foo, @:foo, @'foo, @{foo}, @$foo, @+, @-)
1353 */
7e2040f0 1354 else if (*s == '@' && s[1]
5d1d4326 1355 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$+-", s[1])))
79072805 1356 break;
02aa26ce
NT
1357
1358 /* check for embedded scalars. only stop if we're sure it's a
1359 variable.
1360 */
79072805 1361 else if (*s == '$') {
3280af22 1362 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
79072805 1363 break;
c277df42 1364 if (s + 1 < send && !strchr("()| \n\t", s[1]))
79072805
LW
1365 break; /* in regexp, $ might be tail anchor */
1366 }
02aa26ce
NT
1367
1368 /* backslashes */
79072805
LW
1369 if (*s == '\\' && s+1 < send) {
1370 s++;
02aa26ce
NT
1371
1372 /* some backslashes we leave behind */
c9f97d15 1373 if (*leaveit && *s && strchr(leaveit, *s)) {
79072805
LW
1374 *d++ = '\\';
1375 *d++ = *s++;
1376 continue;
1377 }
02aa26ce
NT
1378
1379 /* deprecate \1 in strings and substitution replacements */
3280af22 1380 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
a0d0e21e 1381 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
79072805 1382 {
599cee73 1383 if (ckWARN(WARN_SYNTAX))
cea2e8a9 1384 Perl_warner(aTHX_ WARN_SYNTAX, "\\%c better written as $%c", *s, *s);
79072805
LW
1385 *--s = '$';
1386 break;
1387 }
02aa26ce
NT
1388
1389 /* string-change backslash escapes */
3280af22 1390 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
79072805
LW
1391 --s;
1392 break;
1393 }
02aa26ce
NT
1394
1395 /* if we get here, it's either a quoted -, or a digit */
79072805 1396 switch (*s) {
02aa26ce
NT
1397
1398 /* quoted - in transliterations */
79072805 1399 case '-':
3280af22 1400 if (PL_lex_inwhat == OP_TRANS) {
79072805
LW
1401 *d++ = *s++;
1402 continue;
1403 }
1404 /* FALL THROUGH */
1405 default:
11b8faa4 1406 {
7e84c16c 1407 if (ckWARN(WARN_MISC) && isALNUM(*s))
4e553d73 1408 Perl_warner(aTHX_ WARN_MISC,
11b8faa4
JH
1409 "Unrecognized escape \\%c passed through",
1410 *s);
1411 /* default action is to copy the quoted character */
f9a63242 1412 goto default_action;
11b8faa4 1413 }
02aa26ce
NT
1414
1415 /* \132 indicates an octal constant */
79072805
LW
1416 case '0': case '1': case '2': case '3':
1417 case '4': case '5': case '6': case '7':
ba210ebe
JH
1418 {
1419 STRLEN len = 0; /* disallow underscores */
1420 uv = (UV)scan_oct(s, 3, &len);
1421 s += len;
1422 }
012bcf8d 1423 goto NUM_ESCAPE_INSERT;
02aa26ce
NT
1424
1425 /* \x24 indicates a hex constant */
79072805 1426 case 'x':
a0ed51b3
LW
1427 ++s;
1428 if (*s == '{') {
1429 char* e = strchr(s, '}');
adaeee49 1430 if (!e) {
a0ed51b3 1431 yyerror("Missing right brace on \\x{}");
adaeee49
GA
1432 e = s;
1433 }
89491803 1434 else {
ba210ebe
JH
1435 STRLEN len = 1; /* allow underscores */
1436 uv = (UV)scan_hex(s + 1, e - s - 1, &len);
1437 }
1438 s = e + 1;
a0ed51b3
LW
1439 }
1440 else {
ba210ebe
JH
1441 {
1442 STRLEN len = 0; /* disallow underscores */
1443 uv = (UV)scan_hex(s, 2, &len);
1444 s += len;
1445 }
012bcf8d
GS
1446 }
1447
1448 NUM_ESCAPE_INSERT:
1449 /* Insert oct or hex escaped character.
301d3d20
JH
1450 * There will always enough room in sv since such
1451 * escapes will be longer than any UT-F8 sequence
1452 * they can end up as. */
ba7cea30
JH
1453
1454 /* This spot is wrong for EBCDIC. Characters like
1455 * the lowercase letters and digits are >127 in EBCDIC,
1456 * so here they would need to be mapped to the Unicode
1457 * repertoire. --jhi */
1458
7948272d 1459 if (uv > 127) {
9aa983d2 1460 if (!has_utf8 && uv > 255) {
301d3d20
JH
1461 /* Might need to recode whatever we have
1462 * accumulated so far if it contains any
1463 * hibit chars.
1464 *
1465 * (Can't we keep track of that and avoid
1466 * this rescan? --jhi)
012bcf8d
GS
1467 */
1468 int hicount = 0;
1469 char *c;
301d3d20 1470
012bcf8d 1471 for (c = SvPVX(sv); c < d; c++) {
fd400ab9 1472 if (UTF8_IS_CONTINUED(*c))
012bcf8d
GS
1473 hicount++;
1474 }
1475 if (hicount) {
1476 char *old_pvx = SvPVX(sv);
1477 char *src, *dst;
9041c2e3 1478
301d3d20 1479 d = SvGROW(sv,
8973db79 1480 SvLEN(sv) + hicount + 1) +
301d3d20 1481 (d - old_pvx);
012bcf8d
GS
1482
1483 src = d - 1;
1484 d += hicount;
1485 dst = d - 1;
1486
1487 while (src < dst) {
fd400ab9 1488 if (UTF8_IS_CONTINUED(*src)) {
9b877dbb
IH
1489 *dst-- = UTF8_EIGHT_BIT_LO(*src);
1490 *dst-- = UTF8_EIGHT_BIT_HI(*src--);
012bcf8d
GS
1491 }
1492 else {
1493 *dst-- = *src--;
1494 }
1495 }
1496 }
1497 }
1498
9aa983d2 1499 if (has_utf8 || uv > 255) {
9041c2e3 1500 d = (char*)uvchr_to_utf8((U8*)d, uv);
4e553d73 1501 has_utf8 = TRUE;
f9a63242
JH
1502 if (PL_lex_inwhat == OP_TRANS &&
1503 PL_sublex_info.sub_op) {
1504 PL_sublex_info.sub_op->op_private |=
1505 (PL_lex_repl ? OPpTRANS_FROM_UTF
1506 : OPpTRANS_TO_UTF);
1507 utf = TRUE;
1508 }
012bcf8d 1509 }
a0ed51b3 1510 else {
012bcf8d 1511 *d++ = (char)uv;
a0ed51b3 1512 }
012bcf8d
GS
1513 }
1514 else {
1515 *d++ = (char)uv;
a0ed51b3 1516 }
79072805 1517 continue;
02aa26ce 1518
4a2d328f
IZ
1519 /* \N{latin small letter a} is a named character */
1520 case 'N':
55eda711 1521 ++s;
423cee85
JH
1522 if (*s == '{') {
1523 char* e = strchr(s, '}');
155aba94 1524 SV *res;
423cee85
JH
1525 STRLEN len;
1526 char *str;
4e553d73 1527
423cee85 1528 if (!e) {
5777a3f7 1529 yyerror("Missing right brace on \\N{}");
423cee85
JH
1530 e = s - 1;
1531 goto cont_scan;
1532 }
55eda711
JH
1533 res = newSVpvn(s + 1, e - s - 1);
1534 res = new_constant( Nullch, 0, "charnames",
1535 res, Nullsv, "\\N{...}" );
f9a63242
JH
1536 if (has_utf8)
1537 sv_utf8_upgrade(res);
423cee85 1538 str = SvPV(res,len);
89491803 1539 if (!has_utf8 && SvUTF8(res)) {
f08d6ad9
GS
1540 char *ostart = SvPVX(sv);
1541 SvCUR_set(sv, d - ostart);
1542 SvPOK_on(sv);
e4f3eed8 1543 *d = '\0';
f08d6ad9 1544 sv_utf8_upgrade(sv);
d2f449dd
SB
1545 /* this just broke our allocation above... */
1546 SvGROW(sv, send - start);
f08d6ad9 1547 d = SvPVX(sv) + SvCUR(sv);
89491803 1548 has_utf8 = TRUE;
f08d6ad9 1549 }
423cee85
JH
1550 if (len > e - s + 4) {
1551 char *odest = SvPVX(sv);
1552
8973db79 1553 SvGROW(sv, (SvLEN(sv) + len - (e - s + 4)));
423cee85
JH
1554 d = SvPVX(sv) + (d - odest);
1555 }
1556 Copy(str, d, len, char);
1557 d += len;
1558 SvREFCNT_dec(res);
1559 cont_scan:
1560 s = e + 1;
1561 }
1562 else
5777a3f7 1563 yyerror("Missing braces on \\N{}");
423cee85
JH
1564 continue;
1565
02aa26ce 1566 /* \c is a control character */
79072805
LW
1567 case 'c':
1568 s++;
9d116dd7
JH
1569#ifdef EBCDIC
1570 *d = *s++;
1571 if (isLOWER(*d))
1572 *d = toUPPER(*d);
4e553d73 1573 *d = toCTRL(*d);
774a9426 1574 d++;
9d116dd7 1575#else
ba210ebe
JH
1576 {
1577 U8 c = *s++;
1578 *d++ = toCTRL(c);
1579 }
9d116dd7 1580#endif
79072805 1581 continue;
02aa26ce
NT
1582
1583 /* printf-style backslashes, formfeeds, newlines, etc */
79072805
LW
1584 case 'b':
1585 *d++ = '\b';
1586 break;
1587 case 'n':
1588 *d++ = '\n';
1589 break;
1590 case 'r':
1591 *d++ = '\r';
1592 break;
1593 case 'f':
1594 *d++ = '\f';
1595 break;
1596 case 't':
1597 *d++ = '\t';
1598 break;
34a3fe2a
PP
1599#ifdef EBCDIC
1600 case 'e':
1601 *d++ = '\047'; /* CP 1047 */
1602 break;
1603 case 'a':
1604 *d++ = '\057'; /* CP 1047 */
1605 break;
1606#else
79072805
LW
1607 case 'e':
1608 *d++ = '\033';
1609 break;
1610 case 'a':
1611 *d++ = '\007';
1612 break;
34a3fe2a 1613#endif
02aa26ce
NT
1614 } /* end switch */
1615
79072805
LW
1616 s++;
1617 continue;
02aa26ce
NT
1618 } /* end if (backslash) */
1619
f9a63242 1620 default_action:
fd400ab9 1621 if (UTF8_IS_CONTINUED(*s) && (this_utf8 || has_utf8)) {
a5a960be
RI
1622 STRLEN len = (STRLEN) -1;
1623 UV uv;
1624 if (this_utf8) {
9041c2e3 1625 uv = utf8n_to_uvchr((U8*)s, send - s, &len, 0);
a5a960be
RI
1626 }
1627 if (len == (STRLEN)-1) {
1628 /* Illegal UTF8 (a high-bit byte), make it valid. */
1629 char *old_pvx = SvPVX(sv);
1630 /* need space for one extra char (NOTE: SvCUR() not set here) */
1631 d = SvGROW(sv, SvLEN(sv) + 1) + (d - old_pvx);
9041c2e3 1632 d = (char*)uvchr_to_utf8((U8*)d, (U8)*s++);
a5a960be
RI
1633 }
1634 else {
1635 while (len--)
1636 *d++ = *s++;
1637 }
1638 has_utf8 = TRUE;
f9a63242
JH
1639 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1640 PL_sublex_info.sub_op->op_private |=
1641 (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1642 utf = TRUE;
1643 }
a5a960be
RI
1644 continue;
1645 }
1646
f9a63242 1647 *d++ = *s++;
02aa26ce
NT
1648 } /* while loop to process each character */
1649
1650 /* terminate the string and set up the sv */
79072805 1651 *d = '\0';
463ee0b2 1652 SvCUR_set(sv, d - SvPVX(sv));
79072805 1653 SvPOK_on(sv);
89491803 1654 if (has_utf8)
7e2040f0 1655 SvUTF8_on(sv);
79072805 1656
02aa26ce 1657 /* shrink the sv if we allocated more than we used */
79072805
LW
1658 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1659 SvLEN_set(sv, SvCUR(sv) + 1);
463ee0b2 1660 Renew(SvPVX(sv), SvLEN(sv), char);
79072805 1661 }
02aa26ce 1662
9b599b2a 1663 /* return the substring (via yylval) only if we parsed anything */
3280af22
NIS
1664 if (s > PL_bufptr) {
1665 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
4e553d73 1666 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
b3ac6de7 1667 sv, Nullsv,
4e553d73 1668 ( PL_lex_inwhat == OP_TRANS
b3ac6de7 1669 ? "tr"
3280af22 1670 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
b3ac6de7
IZ
1671 ? "s"
1672 : "qq")));
79072805 1673 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
b3ac6de7 1674 } else
8990e307 1675 SvREFCNT_dec(sv);
79072805
LW
1676 return s;
1677}
1678
ffb4593c
NT
1679/* S_intuit_more
1680 * Returns TRUE if there's more to the expression (e.g., a subscript),
1681 * FALSE otherwise.
ffb4593c
NT
1682 *
1683 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1684 *
1685 * ->[ and ->{ return TRUE
1686 * { and [ outside a pattern are always subscripts, so return TRUE
1687 * if we're outside a pattern and it's not { or [, then return FALSE
1688 * if we're in a pattern and the first char is a {
1689 * {4,5} (any digits around the comma) returns FALSE
1690 * if we're in a pattern and the first char is a [
1691 * [] returns FALSE
1692 * [SOMETHING] has a funky algorithm to decide whether it's a
1693 * character class or not. It has to deal with things like
1694 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1695 * anything else returns TRUE
1696 */
1697
9cbb5ea2
GS
1698/* This is the one truly awful dwimmer necessary to conflate C and sed. */
1699
76e3520e 1700STATIC int
cea2e8a9 1701S_intuit_more(pTHX_ register char *s)
79072805 1702{
3280af22 1703 if (PL_lex_brackets)
79072805
LW
1704 return TRUE;
1705 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1706 return TRUE;
1707 if (*s != '{' && *s != '[')
1708 return FALSE;
3280af22 1709 if (!PL_lex_inpat)
79072805
LW
1710 return TRUE;
1711
1712 /* In a pattern, so maybe we have {n,m}. */
1713 if (*s == '{') {
1714 s++;
1715 if (!isDIGIT(*s))
1716 return TRUE;
1717 while (isDIGIT(*s))
1718 s++;
1719 if (*s == ',')
1720 s++;
1721 while (isDIGIT(*s))
1722 s++;
1723 if (*s == '}')
1724 return FALSE;
1725 return TRUE;
1726
1727 }
1728
1729 /* On the other hand, maybe we have a character class */
1730
1731 s++;
1732 if (*s == ']' || *s == '^')
1733 return FALSE;
1734 else {
ffb4593c 1735 /* this is terrifying, and it works */
79072805
LW
1736 int weight = 2; /* let's weigh the evidence */
1737 char seen[256];
f27ffc4a 1738 unsigned char un_char = 255, last_un_char;
93a17b20 1739 char *send = strchr(s,']');
3280af22 1740 char tmpbuf[sizeof PL_tokenbuf * 4];
79072805
LW
1741
1742 if (!send) /* has to be an expression */
1743 return TRUE;
1744
1745 Zero(seen,256,char);
1746 if (*s == '$')
1747 weight -= 3;
1748 else if (isDIGIT(*s)) {
1749 if (s[1] != ']') {
1750 if (isDIGIT(s[1]) && s[2] == ']')
1751 weight -= 10;
1752 }
1753 else
1754 weight -= 100;
1755 }
1756 for (; s < send; s++) {
1757 last_un_char = un_char;
1758 un_char = (unsigned char)*s;
1759 switch (*s) {
1760 case '@':
1761 case '&':
1762 case '$':
1763 weight -= seen[un_char] * 10;
7e2040f0 1764 if (isALNUM_lazy_if(s+1,UTF)) {
8903cb82 1765 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
a0d0e21e 1766 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
79072805
LW
1767 weight -= 100;
1768 else
1769 weight -= 10;
1770 }
1771 else if (*s == '$' && s[1] &&
93a17b20
LW
1772 strchr("[#!%*<>()-=",s[1])) {
1773 if (/*{*/ strchr("])} =",s[2]))
79072805
LW
1774 weight -= 10;
1775 else
1776 weight -= 1;
1777 }
1778 break;
1779 case '\\':
1780 un_char = 254;
1781 if (s[1]) {
93a17b20 1782 if (strchr("wds]",s[1]))
79072805
LW
1783 weight += 100;
1784 else if (seen['\''] || seen['"'])
1785 weight += 1;
93a17b20 1786 else if (strchr("rnftbxcav",s[1]))
79072805
LW
1787 weight += 40;
1788 else if (isDIGIT(s[1])) {
1789 weight += 40;
1790 while (s[1] && isDIGIT(s[1]))
1791 s++;
1792 }
1793 }
1794 else
1795 weight += 100;
1796 break;
1797 case '-':
1798 if (s[1] == '\\')
1799 weight += 50;
93a17b20 1800 if (strchr("aA01! ",last_un_char))
79072805 1801 weight += 30;
93a17b20 1802 if (strchr("zZ79~",s[1]))
79072805 1803 weight += 30;
f27ffc4a
GS
1804 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1805 weight -= 5; /* cope with negative subscript */
79072805
LW
1806 break;
1807 default:
93a17b20 1808 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
79072805
LW
1809 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1810 char *d = tmpbuf;
1811 while (isALPHA(*s))
1812 *d++ = *s++;
1813 *d = '\0';
1814 if (keyword(tmpbuf, d - tmpbuf))
1815 weight -= 150;
1816 }
1817 if (un_char == last_un_char + 1)
1818 weight += 5;
1819 weight -= seen[un_char];
1820 break;
1821 }
1822 seen[un_char]++;
1823 }
1824 if (weight >= 0) /* probably a character class */
1825 return FALSE;
1826 }
1827
1828 return TRUE;
1829}
ffed7fef 1830
ffb4593c
NT
1831/*
1832 * S_intuit_method
1833 *
1834 * Does all the checking to disambiguate
1835 * foo bar
1836 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1837 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1838 *
1839 * First argument is the stuff after the first token, e.g. "bar".
1840 *
1841 * Not a method if bar is a filehandle.
1842 * Not a method if foo is a subroutine prototyped to take a filehandle.
1843 * Not a method if it's really "Foo $bar"
1844 * Method if it's "foo $bar"
1845 * Not a method if it's really "print foo $bar"
1846 * Method if it's really "foo package::" (interpreted as package->foo)
1847 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
3cb0bbe5 1848 * Not a method if bar is a filehandle or package, but is quoted with
ffb4593c
NT
1849 * =>
1850 */
1851
76e3520e 1852STATIC int
cea2e8a9 1853S_intuit_method(pTHX_ char *start, GV *gv)
a0d0e21e
LW
1854{
1855 char *s = start + (*start == '$');
3280af22 1856 char tmpbuf[sizeof PL_tokenbuf];
a0d0e21e
LW
1857 STRLEN len;
1858 GV* indirgv;
1859
1860 if (gv) {
b6c543e3 1861 CV *cv;
a0d0e21e
LW
1862 if (GvIO(gv))
1863 return 0;
b6c543e3
IZ
1864 if ((cv = GvCVu(gv))) {
1865 char *proto = SvPVX(cv);
1866 if (proto) {
1867 if (*proto == ';')
1868 proto++;
1869 if (*proto == '*')
1870 return 0;
1871 }
1872 } else
a0d0e21e
LW
1873 gv = 0;
1874 }
8903cb82 1875 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
ffb4593c
NT
1876 /* start is the beginning of the possible filehandle/object,
1877 * and s is the end of it
1878 * tmpbuf is a copy of it
1879 */
1880
a0d0e21e 1881 if (*start == '$') {
3280af22 1882 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
a0d0e21e
LW
1883 return 0;
1884 s = skipspace(s);
3280af22
NIS
1885 PL_bufptr = start;
1886 PL_expect = XREF;
a0d0e21e
LW
1887 return *s == '(' ? FUNCMETH : METHOD;
1888 }
1889 if (!keyword(tmpbuf, len)) {
c3e0f903
GS
1890 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1891 len -= 2;
1892 tmpbuf[len] = '\0';
1893 goto bare_package;
1894 }
1895 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
8ebc5c01 1896 if (indirgv && GvCVu(indirgv))
a0d0e21e
LW
1897 return 0;
1898 /* filehandle or package name makes it a method */
89bfa8cd 1899 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
a0d0e21e 1900 s = skipspace(s);
3280af22 1901 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
55497cff 1902 return 0; /* no assumptions -- "=>" quotes bearword */
c3e0f903 1903 bare_package:
3280af22 1904 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
79cb57f6 1905 newSVpvn(tmpbuf,len));
3280af22
NIS
1906 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1907 PL_expect = XTERM;
a0d0e21e 1908 force_next(WORD);
3280af22 1909 PL_bufptr = s;
a0d0e21e
LW
1910 return *s == '(' ? FUNCMETH : METHOD;
1911 }
1912 }
1913 return 0;
1914}
1915
ffb4593c
NT
1916/*
1917 * S_incl_perldb
1918 * Return a string of Perl code to load the debugger. If PERL5DB
1919 * is set, it will return the contents of that, otherwise a
1920 * compile-time require of perl5db.pl.
1921 */
1922
76e3520e 1923STATIC char*
cea2e8a9 1924S_incl_perldb(pTHX)
a0d0e21e 1925{
3280af22 1926 if (PL_perldb) {
76e3520e 1927 char *pdb = PerlEnv_getenv("PERL5DB");
a0d0e21e
LW
1928
1929 if (pdb)
1930 return pdb;
61bb5906 1931 SETERRNO(0,SS$_NORMAL);
a0d0e21e
LW
1932 return "BEGIN { require 'perl5db.pl' }";
1933 }
1934 return "";
1935}
1936
1937
16d20bd9 1938/* Encoded script support. filter_add() effectively inserts a
4e553d73 1939 * 'pre-processing' function into the current source input stream.
16d20bd9
AD
1940 * Note that the filter function only applies to the current source file
1941 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1942 *
1943 * The datasv parameter (which may be NULL) can be used to pass
1944 * private data to this instance of the filter. The filter function
1945 * can recover the SV using the FILTER_DATA macro and use it to
1946 * store private buffers and state information.
1947 *
1948 * The supplied datasv parameter is upgraded to a PVIO type
4755096e 1949 * and the IoDIRP/IoANY field is used to store the function pointer,
e0c19803 1950 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
16d20bd9
AD
1951 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1952 * private use must be set using malloc'd pointers.
1953 */
16d20bd9
AD
1954
1955SV *
864dbfa3 1956Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
16d20bd9 1957{
f4c556ac
GS
1958 if (!funcp)
1959 return Nullsv;
1960
3280af22
NIS
1961 if (!PL_rsfp_filters)
1962 PL_rsfp_filters = newAV();
16d20bd9 1963 if (!datasv)
8c52afec 1964 datasv = NEWSV(255,0);
16d20bd9 1965 if (!SvUPGRADE(datasv, SVt_PVIO))
cea2e8a9 1966 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
4755096e 1967 IoANY(datasv) = (void *)funcp; /* stash funcp into spare field */
e0c19803 1968 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
f4c556ac
GS
1969 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1970 funcp, SvPV_nolen(datasv)));
3280af22
NIS
1971 av_unshift(PL_rsfp_filters, 1);
1972 av_store(PL_rsfp_filters, 0, datasv) ;
16d20bd9
AD
1973 return(datasv);
1974}
4e553d73 1975
16d20bd9
AD
1976
1977/* Delete most recently added instance of this filter function. */
a0d0e21e 1978void
864dbfa3 1979Perl_filter_del(pTHX_ filter_t funcp)
16d20bd9 1980{
e0c19803 1981 SV *datasv;
f4c556ac 1982 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp));
3280af22 1983 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
16d20bd9
AD
1984 return;
1985 /* if filter is on top of stack (usual case) just pop it off */
e0c19803 1986 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
4755096e 1987 if (IoANY(datasv) == (void *)funcp) {
e0c19803 1988 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
4755096e 1989 IoANY(datasv) = (void *)NULL;
3280af22 1990 sv_free(av_pop(PL_rsfp_filters));
e50aee73 1991
16d20bd9
AD
1992 return;
1993 }
1994 /* we need to search for the correct entry and clear it */
cea2e8a9 1995 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
16d20bd9
AD
1996}
1997
1998
1999/* Invoke the n'th filter function for the current rsfp. */
2000I32
864dbfa3 2001Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
4e553d73
NIS
2002
2003
8ac85365 2004 /* 0 = read one text line */
a0d0e21e 2005{
16d20bd9
AD
2006 filter_t funcp;
2007 SV *datasv = NULL;
e50aee73 2008
3280af22 2009 if (!PL_rsfp_filters)
16d20bd9 2010 return -1;
3280af22 2011 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
16d20bd9
AD
2012 /* Provide a default input filter to make life easy. */
2013 /* Note that we append to the line. This is handy. */
f4c556ac
GS
2014 DEBUG_P(PerlIO_printf(Perl_debug_log,
2015 "filter_read %d: from rsfp\n", idx));
4e553d73 2016 if (maxlen) {
16d20bd9
AD
2017 /* Want a block */
2018 int len ;
2019 int old_len = SvCUR(buf_sv) ;
2020
2021 /* ensure buf_sv is large enough */
2022 SvGROW(buf_sv, old_len + maxlen) ;
3280af22
NIS
2023 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
2024 if (PerlIO_error(PL_rsfp))
37120919
AD
2025 return -1; /* error */
2026 else
2027 return 0 ; /* end of file */
2028 }
16d20bd9
AD
2029 SvCUR_set(buf_sv, old_len + len) ;
2030 } else {
2031 /* Want a line */
3280af22
NIS
2032 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
2033 if (PerlIO_error(PL_rsfp))
37120919
AD
2034 return -1; /* error */
2035 else
2036 return 0 ; /* end of file */
2037 }
16d20bd9
AD
2038 }
2039 return SvCUR(buf_sv);
2040 }
2041 /* Skip this filter slot if filter has been deleted */
3280af22 2042 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
f4c556ac
GS
2043 DEBUG_P(PerlIO_printf(Perl_debug_log,
2044 "filter_read %d: skipped (filter deleted)\n",
2045 idx));
16d20bd9
AD
2046 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
2047 }
2048 /* Get function pointer hidden within datasv */
4755096e 2049 funcp = (filter_t)IoANY(datasv);
f4c556ac
GS
2050 DEBUG_P(PerlIO_printf(Perl_debug_log,
2051 "filter_read %d: via function %p (%s)\n",
2052 idx, funcp, SvPV_nolen(datasv)));
16d20bd9
AD
2053 /* Call function. The function is expected to */
2054 /* call "FILTER_READ(idx+1, buf_sv)" first. */
37120919 2055 /* Return: <0:error, =0:eof, >0:not eof */
0cb96387 2056 return (*funcp)(aTHXo_ idx, buf_sv, maxlen);
16d20bd9
AD
2057}
2058
76e3520e 2059STATIC char *
cea2e8a9 2060S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
16d20bd9 2061{
c39cd008 2062#ifdef PERL_CR_FILTER
3280af22 2063 if (!PL_rsfp_filters) {
c39cd008 2064 filter_add(S_cr_textfilter,NULL);
a868473f
NIS
2065 }
2066#endif
3280af22 2067 if (PL_rsfp_filters) {
16d20bd9 2068
55497cff
PP
2069 if (!append)
2070 SvCUR_set(sv, 0); /* start with empty line */
16d20bd9
AD
2071 if (FILTER_READ(0, sv, 0) > 0)
2072 return ( SvPVX(sv) ) ;
2073 else
2074 return Nullch ;
2075 }
9d116dd7 2076 else
fd049845 2077 return (sv_gets(sv, fp, append));
a0d0e21e
LW
2078}
2079
01ec43d0
GS
2080STATIC HV *
2081S_find_in_my_stash(pTHX_ char *pkgname, I32 len)
def3634b
GS
2082{
2083 GV *gv;
2084
01ec43d0 2085 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
def3634b
GS
2086 return PL_curstash;
2087
2088 if (len > 2 &&
2089 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
01ec43d0
GS
2090 (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV)))
2091 {
2092 return GvHV(gv); /* Foo:: */
def3634b
GS
2093 }
2094
2095 /* use constant CLASS => 'MyClass' */
2096 if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) {
2097 SV *sv;
2098 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2099 pkgname = SvPV_nolen(sv);
2100 }
2101 }
2102
2103 return gv_stashpv(pkgname, FALSE);
2104}
a0d0e21e 2105
748a9306
LW
2106#ifdef DEBUGGING
2107 static char* exp_name[] =
09bef843
SB
2108 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2109 "ATTRTERM", "TERMBLOCK"
2110 };
748a9306 2111#endif
463ee0b2 2112
02aa26ce
NT
2113/*
2114 yylex
2115
2116 Works out what to call the token just pulled out of the input
2117 stream. The yacc parser takes care of taking the ops we return and
2118 stitching them into a tree.
2119
2120 Returns:
2121 PRIVATEREF
2122
2123 Structure:
2124 if read an identifier
2125 if we're in a my declaration
2126 croak if they tried to say my($foo::bar)
2127 build the ops for a my() declaration
2128 if it's an access to a my() variable
2129 are we in a sort block?
2130 croak if my($a); $a <=> $b
2131 build ops for access to a my() variable
2132 if in a dq string, and they've said @foo and we can't find @foo
2133 croak
2134 build ops for a bareword
2135 if we already built the token before, use it.
2136*/
2137
dba4d153 2138#ifdef USE_PURE_BISON
864dbfa3 2139int
dba4d153 2140Perl_yylex_r(pTHX_ YYSTYPE *lvalp, int *lcharp)
378cc40b 2141{
20141f0e
RI
2142 int r;
2143
6f202aea 2144 yyactlevel++;
20141f0e
RI
2145 yylval_pointer[yyactlevel] = lvalp;
2146 yychar_pointer[yyactlevel] = lcharp;
b73d6f50
RI
2147 if (yyactlevel >= YYMAXLEVEL)
2148 Perl_croak(aTHX_ "panic: YYMAXLEVEL");
20141f0e 2149
dba4d153 2150 r = Perl_yylex(aTHX);
20141f0e 2151
d8ae6756
RI
2152 if (yyactlevel > 0)
2153 yyactlevel--;
20141f0e
RI
2154
2155 return r;
2156}
dba4d153 2157#endif
20141f0e 2158
dba4d153
JH
2159#ifdef __SC__
2160#pragma segment Perl_yylex
2161#endif
dba4d153 2162int
dba4d153 2163Perl_yylex(pTHX)
20141f0e 2164{
79072805 2165 register char *s;
378cc40b 2166 register char *d;
79072805 2167 register I32 tmp;
463ee0b2 2168 STRLEN len;
161b471a
NIS
2169 GV *gv = Nullgv;
2170 GV **gvp = 0;
aa7440fb 2171 bool bof = FALSE;
a687059c 2172
02aa26ce 2173 /* check if there's an identifier for us to look at */
3280af22 2174 if (PL_pending_ident) {
02aa26ce 2175 /* pit holds the identifier we read and pending_ident is reset */
3280af22
NIS
2176 char pit = PL_pending_ident;
2177 PL_pending_ident = 0;
bbce6d69 2178
607df283
SC
2179 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2180 "### Tokener saw identifier '%s'\n", PL_tokenbuf); })
2181
02aa26ce
NT
2182 /* if we're in a my(), we can't allow dynamics here.
2183 $foo'bar has already been turned into $foo::bar, so
2184 just check for colons.
2185
2186 if it's a legal name, the OP is a PADANY.
2187 */
3280af22 2188 if (PL_in_my) {
77ca0c92 2189 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
1ec3e8de
GS
2190 if (strchr(PL_tokenbuf,':'))
2191 yyerror(Perl_form(aTHX_ "No package name allowed for "
2192 "variable %s in \"our\"",
2193 PL_tokenbuf));
77ca0c92
LW
2194 tmp = pad_allocmy(PL_tokenbuf);
2195 }
2196 else {
2197 if (strchr(PL_tokenbuf,':'))
2198 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
02aa26ce 2199
77ca0c92
LW
2200 yylval.opval = newOP(OP_PADANY, 0);
2201 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
2202 return PRIVATEREF;
2203 }
bbce6d69
PP
2204 }
2205
4e553d73 2206 /*
02aa26ce
NT
2207 build the ops for accesses to a my() variable.
2208
2209 Deny my($a) or my($b) in a sort block, *if* $a or $b is
2210 then used in a comparison. This catches most, but not
2211 all cases. For instance, it catches
2212 sort { my($a); $a <=> $b }
2213 but not
2214 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
2215 (although why you'd do that is anyone's guess).
2216 */
2217
3280af22 2218 if (!strchr(PL_tokenbuf,':')) {
a863c7d1 2219#ifdef USE_THREADS
54b9620d 2220 /* Check for single character per-thread SVs */
3280af22
NIS
2221 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
2222 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
2223 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
554b3eca 2224 {
2faa37cc 2225 yylval.opval = newOP(OP_THREADSV, 0);
a863c7d1
MB
2226 yylval.opval->op_targ = tmp;
2227 return PRIVATEREF;
2228 }
2229#endif /* USE_THREADS */
3280af22 2230 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
f472eb5c 2231 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
77ca0c92 2232 /* might be an "our" variable" */
f472eb5c 2233 if (SvFLAGS(namesv) & SVpad_OUR) {
77ca0c92 2234 /* build ops for a bareword */
f472eb5c
GS
2235 SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0);
2236 sv_catpvn(sym, "::", 2);
2237 sv_catpv(sym, PL_tokenbuf+1);
2238 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
77ca0c92 2239 yylval.opval->op_private = OPpCONST_ENTERED;
f472eb5c 2240 gv_fetchpv(SvPVX(sym),
77ca0c92 2241 (PL_in_eval
f472eb5c
GS
2242 ? (GV_ADDMULTI | GV_ADDINEVAL)
2243 : TRUE
77ca0c92
LW
2244 ),
2245 ((PL_tokenbuf[0] == '$') ? SVt_PV
2246 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2247 : SVt_PVHV));
2248 return WORD;
2249 }
2250
02aa26ce 2251 /* if it's a sort block and they're naming $a or $b */
3280af22
NIS
2252 if (PL_last_lop_op == OP_SORT &&
2253 PL_tokenbuf[0] == '$' &&
2254 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
2255 && !PL_tokenbuf[2])
bbce6d69 2256 {
3280af22
NIS
2257 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
2258 d < PL_bufend && *d != '\n';
a863c7d1
MB
2259 d++)
2260 {
2261 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
cea2e8a9 2262 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
3280af22 2263 PL_tokenbuf);
a863c7d1 2264 }
bbce6d69
PP
2265 }
2266 }
bbce6d69 2267
a863c7d1
MB
2268 yylval.opval = newOP(OP_PADANY, 0);
2269 yylval.opval->op_targ = tmp;
2270 return PRIVATEREF;
2271 }
bbce6d69
PP
2272 }
2273
02aa26ce
NT
2274 /*
2275 Whine if they've said @foo in a doublequoted string,
2276 and @foo isn't a variable we can find in the symbol
2277 table.
2278 */
3280af22
NIS
2279 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
2280 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
8593bda5
GS
2281 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
2282 && ckWARN(WARN_AMBIGUOUS))
2283 {
2284 /* Downgraded from fatal to warning 20000522 mjd */
2285 Perl_warner(aTHX_ WARN_AMBIGUOUS,
2286 "Possible unintended interpolation of %s in string",
2287 PL_tokenbuf);
2288 }
bbce6d69
PP
2289 }
2290
02aa26ce 2291 /* build ops for a bareword */
3280af22 2292 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
bbce6d69 2293 yylval.opval->op_private = OPpCONST_ENTERED;
3280af22
NIS
2294 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
2295 ((PL_tokenbuf[0] == '$') ? SVt_PV
2296 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
bbce6d69
PP
2297 : SVt_PVHV));
2298 return WORD;
2299 }
2300
02aa26ce
NT
2301 /* no identifier pending identification */
2302
3280af22 2303 switch (PL_lex_state) {
79072805
LW
2304#ifdef COMMENTARY
2305 case LEX_NORMAL: /* Some compilers will produce faster */
2306 case LEX_INTERPNORMAL: /* code if we comment these out. */
2307 break;
2308#endif
2309
09bef843 2310 /* when we've already built the next token, just pull it out of the queue */
79072805 2311 case LEX_KNOWNEXT:
3280af22
NIS
2312 PL_nexttoke--;
2313 yylval = PL_nextval[PL_nexttoke];
2314 if (!PL_nexttoke) {
2315 PL_lex_state = PL_lex_defer;
2316 PL_expect = PL_lex_expect;
2317 PL_lex_defer = LEX_NORMAL;
463ee0b2 2318 }
607df283 2319 DEBUG_T({ PerlIO_printf(Perl_debug_log,
4659c93f
RB
2320 "### Next token after '%s' was known, type %"IVdf"\n", PL_bufptr,
2321 (IV)PL_nexttype[PL_nexttoke]); })
607df283 2322
3280af22 2323 return(PL_nexttype[PL_nexttoke]);
79072805 2324
02aa26ce 2325 /* interpolated case modifiers like \L \U, including \Q and \E.
3280af22 2326 when we get here, PL_bufptr is at the \
02aa26ce 2327 */
79072805
LW
2328 case LEX_INTERPCASEMOD:
2329#ifdef DEBUGGING
3280af22 2330 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
cea2e8a9 2331 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
79072805 2332#endif
02aa26ce 2333 /* handle \E or end of string */
3280af22 2334 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
a0d0e21e 2335 char oldmod;
02aa26ce
NT
2336
2337 /* if at a \E */
3280af22
NIS
2338 if (PL_lex_casemods) {
2339 oldmod = PL_lex_casestack[--PL_lex_casemods];
2340 PL_lex_casestack[PL_lex_casemods] = '\0';
02aa26ce 2341
3280af22
NIS
2342 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2343 PL_bufptr += 2;
2344 PL_lex_state = LEX_INTERPCONCAT;
a0d0e21e 2345 }
79072805
LW
2346 return ')';
2347 }
3280af22
NIS
2348 if (PL_bufptr != PL_bufend)
2349 PL_bufptr += 2;
2350 PL_lex_state = LEX_INTERPCONCAT;
cea2e8a9 2351 return yylex();
79072805
LW
2352 }
2353 else {
607df283
SC
2354 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2355 "### Saw case modifier at '%s'\n", PL_bufptr); })
3280af22 2356 s = PL_bufptr + 1;
79072805
LW
2357 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2358 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
a0d0e21e 2359 if (strchr("LU", *s) &&
3280af22 2360 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
a0d0e21e 2361 {
3280af22 2362 PL_lex_casestack[--PL_lex_casemods] = '\0';
a0d0e21e
LW
2363 return ')';
2364 }
3280af22
NIS
2365 if (PL_lex_casemods > 10) {
2366 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2367 if (newlb != PL_lex_casestack) {
a0d0e21e 2368 SAVEFREEPV(newlb);
3280af22 2369 PL_lex_casestack = newlb;
a0d0e21e
LW
2370 }
2371 }
3280af22
NIS
2372 PL_lex_casestack[PL_lex_casemods++] = *s;
2373 PL_lex_casestack[PL_lex_casemods] = '\0';
2374 PL_lex_state = LEX_INTERPCONCAT;
2375 PL_nextval[PL_nexttoke].ival = 0;
79072805
LW
2376 force_next('(');
2377 if (*s == 'l')
3280af22 2378 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
79072805 2379 else if (*s == 'u')
3280af22 2380 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
79072805 2381 else if (*s == 'L')
3280af22 2382 PL_nextval[PL_nexttoke].ival = OP_LC;
79072805 2383 else if (*s == 'U')
3280af22 2384 PL_nextval[PL_nexttoke].ival = OP_UC;
a0d0e21e 2385 else if (*s == 'Q')
3280af22 2386 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
79072805 2387 else
cea2e8a9 2388 Perl_croak(aTHX_ "panic: yylex");
3280af22 2389 PL_bufptr = s + 1;
79072805 2390 force_next(FUNC);
3280af22
NIS
2391 if (PL_lex_starts) {
2392 s = PL_bufptr;
2393 PL_lex_starts = 0;
79072805
LW
2394 Aop(OP_CONCAT);
2395 }
2396 else
cea2e8a9 2397 return yylex();
79072805
LW
2398 }
2399
55497cff
PP
2400 case LEX_INTERPPUSH:
2401 return sublex_push();
2402
79072805 2403 case LEX_INTERPSTART:
3280af22 2404 if (PL_bufptr == PL_bufend)
79072805 2405 return sublex_done();
607df283
SC
2406 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2407 "### Interpolated variable at '%s'\n", PL_bufptr); })
3280af22
NIS
2408 PL_expect = XTERM;
2409 PL_lex_dojoin = (*PL_bufptr == '@');
2410 PL_lex_state = LEX_INTERPNORMAL;
2411 if (PL_lex_dojoin) {
2412 PL_nextval[PL_nexttoke].ival = 0;
79072805 2413 force_next(',');
554b3eca 2414#ifdef USE_THREADS
533c011a
NIS
2415 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2416 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
554b3eca
MB
2417 force_next(PRIVATEREF);
2418#else
a0d0e21e 2419 force_ident("\"", '$');
554b3eca 2420#endif /* USE_THREADS */
3280af22 2421 PL_nextval[PL_nexttoke].ival = 0;
79072805 2422 force_next('$');
3280af22 2423 PL_nextval[PL_nexttoke].ival = 0;
79072805 2424 force_next('(');
3280af22 2425 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
79072805
LW
2426 force_next(FUNC);
2427 }
3280af22
NIS
2428 if (PL_lex_starts++) {
2429 s = PL_bufptr;
79072805
LW
2430 Aop(OP_CONCAT);
2431 }
cea2e8a9 2432 return yylex();
79072805
LW
2433
2434 case LEX_INTERPENDMAYBE:
3280af22
NIS
2435 if (intuit_more(PL_bufptr)) {
2436 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
79072805
LW
2437 break;
2438 }
2439 /* FALL THROUGH */
2440
2441 case LEX_INTERPEND:
3280af22
NIS
2442 if (PL_lex_dojoin) {
2443 PL_lex_dojoin = FALSE;
2444 PL_lex_state = LEX_INTERPCONCAT;
79072805
LW
2445 return ')';
2446 }
43a16006 2447 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
25da4f38 2448 && SvEVALED(PL_lex_repl))
43a16006 2449 {
e9fa98b2 2450 if (PL_bufptr != PL_bufend)
cea2e8a9 2451 Perl_croak(aTHX_ "Bad evalled substitution pattern");
e9fa98b2
HS
2452 PL_lex_repl = Nullsv;
2453 }
79072805
LW
2454 /* FALLTHROUGH */
2455 case LEX_INTERPCONCAT:
2456#ifdef DEBUGGING
3280af22 2457 if (PL_lex_brackets)
cea2e8a9 2458 Perl_croak(aTHX_ "panic: INTERPCONCAT");
79072805 2459#endif
3280af22 2460 if (PL_bufptr == PL_bufend)
79072805
LW
2461 return sublex_done();
2462
3280af22
NIS
2463 if (SvIVX(PL_linestr) == '\'') {
2464 SV *sv = newSVsv(PL_linestr);
2465 if (!PL_lex_inpat)
76e3520e 2466 sv = tokeq(sv);
3280af22 2467 else if ( PL_hints & HINT_NEW_RE )
b3ac6de7 2468 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
79072805 2469 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3280af22 2470 s = PL_bufend;
79072805
LW
2471 }
2472 else {
3280af22 2473 s = scan_const(PL_bufptr);
79072805 2474 if (*s == '\\')
3280af22 2475 PL_lex_state = LEX_INTERPCASEMOD;
79072805 2476 else
3280af22 2477 PL_lex_state = LEX_INTERPSTART;
79072805
LW
2478 }
2479
3280af22
NIS
2480 if (s != PL_bufptr) {
2481 PL_nextval[PL_nexttoke] = yylval;
2482 PL_expect = XTERM;
79072805 2483 force_next(THING);
3280af22 2484 if (PL_lex_starts++)
79072805
LW
2485 Aop(OP_CONCAT);
2486 else {
3280af22 2487 PL_bufptr = s;
cea2e8a9 2488 return yylex();
79072805
LW
2489 }
2490 }
2491
cea2e8a9 2492 return yylex();
a0d0e21e 2493 case LEX_FORMLINE:
3280af22
NIS
2494 PL_lex_state = LEX_NORMAL;
2495 s = scan_formline(PL_bufptr);
2496 if (!PL_lex_formbrack)
a0d0e21e
LW
2497 goto rightbracket;
2498 OPERATOR(';');
79072805
LW
2499 }
2500
3280af22
NIS
2501 s = PL_bufptr;
2502 PL_oldoldbufptr = PL_oldbufptr;
2503 PL_oldbufptr = s;
607df283 2504 DEBUG_T( {
bf49b057
GS
2505 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2506 exp_name[PL_expect], s);
79072805 2507 } )
463ee0b2
LW
2508
2509 retry:
378cc40b
LW
2510 switch (*s) {
2511 default:
7e2040f0 2512 if (isIDFIRST_lazy_if(s,UTF))
834a4ddd 2513 goto keylookup;
cea2e8a9 2514 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
e929a76b
LW
2515 case 4:
2516 case 26:
2517 goto fake_eof; /* emulate EOF on ^D or ^Z */
378cc40b 2518 case 0:
3280af22
NIS
2519 if (!PL_rsfp) {
2520 PL_last_uni = 0;
2521 PL_last_lop = 0;
2522 if (PL_lex_brackets)
d98d5fff 2523 yyerror("Missing right curly or square bracket");
4e553d73 2524 DEBUG_T( { PerlIO_printf(Perl_debug_log,
607df283
SC
2525 "### Tokener got EOF\n");
2526 } )
79072805 2527 TOKEN(0);
463ee0b2 2528 }
3280af22 2529 if (s++ < PL_bufend)
a687059c 2530 goto retry; /* ignore stray nulls */
3280af22
NIS
2531 PL_last_uni = 0;
2532 PL_last_lop = 0;
2533 if (!PL_in_eval && !PL_preambled) {
2534 PL_preambled = TRUE;
2535 sv_setpv(PL_linestr,incl_perldb());
2536 if (SvCUR(PL_linestr))
2537 sv_catpv(PL_linestr,";");
2538 if (PL_preambleav){
2539 while(AvFILLp(PL_preambleav) >= 0) {
2540 SV *tmpsv = av_shift(PL_preambleav);
2541 sv_catsv(PL_linestr, tmpsv);
2542 sv_catpv(PL_linestr, ";");
91b7def8
PP
2543 sv_free(tmpsv);
2544 }
3280af22
NIS
2545 sv_free((SV*)PL_preambleav);
2546 PL_preambleav = NULL;
91b7def8 2547 }
3280af22
NIS
2548 if (PL_minus_n || PL_minus_p) {
2549 sv_catpv(PL_linestr, "LINE: while (<>) {");
2550 if (PL_minus_l)
2551 sv_catpv(PL_linestr,"chomp;");
2552 if (PL_minus_a) {
8fd239a7
CS
2553 GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
2554 if (gv)
2555 GvIMPORTED_AV_on(gv);
3280af22
NIS
2556 if (PL_minus_F) {
2557 if (strchr("/'\"", *PL_splitstr)
2558 && strchr(PL_splitstr + 1, *PL_splitstr))
cea2e8a9 2559 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
54310121
PP
2560 else {
2561 char delim;
2562 s = "'~#\200\1'"; /* surely one char is unused...*/
3280af22 2563 while (s[1] && strchr(PL_splitstr, *s)) s++;
54310121 2564 delim = *s;
cea2e8a9 2565 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c",
46fc3d4c 2566 "q" + (delim == '\''), delim);
3280af22 2567 for (s = PL_splitstr; *s; s++) {
54310121 2568 if (*s == '\\')
3280af22
NIS
2569 sv_catpvn(PL_linestr, "\\", 1);
2570 sv_catpvn(PL_linestr, s, 1);
54310121 2571 }
cea2e8a9 2572 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
54310121 2573 }
2304df62
AD
2574 }
2575 else
3280af22 2576 sv_catpv(PL_linestr,"@F=split(' ');");
2304df62 2577 }
79072805 2578 }
3280af22
NIS
2579 sv_catpv(PL_linestr, "\n");
2580 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2581 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
207e3d1a 2582 PL_last_lop = PL_last_uni = Nullch;
3280af22 2583 if (PERLDB_LINE && PL_curstash != PL_debstash) {
a0d0e21e
LW
2584 SV *sv = NEWSV(85,0);
2585
2586 sv_upgrade(sv, SVt_PVMG);
3280af22 2587 sv_setsv(sv,PL_linestr);
57843af0 2588 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
a0d0e21e 2589 }
79072805 2590 goto retry;
a687059c 2591 }
e929a76b 2592 do {
aa7440fb 2593 bof = PL_rsfp ? TRUE : FALSE;
7e28d3af
JH
2594 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2595 fake_eof:
2596 if (PL_rsfp) {
2597 if (PL_preprocess && !PL_in_eval)
2598 (void)PerlProc_pclose(PL_rsfp);
2599 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2600 PerlIO_clearerr(PL_rsfp);
2601 else
2602 (void)PerlIO_close(PL_rsfp);
2603 PL_rsfp = Nullfp;
2604 PL_doextract = FALSE;
2605 }
2606 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2607 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2608 sv_catpv(PL_linestr,";}");
2609 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2610 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
207e3d1a 2611 PL_last_lop = PL_last_uni = Nullch;
7e28d3af
JH
2612 PL_minus_n = PL_minus_p = 0;
2613 goto retry;
2614 }
2615 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
207e3d1a 2616 PL_last_lop = PL_last_uni = Nullch;
7e28d3af
JH
2617 sv_setpv(PL_linestr,"");
2618 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2619 }
2620 /* if it looks like the start of a BOM, check if it in fact is */
2621 else if (bof && (!*s || *(U8*)s == 0xEF || *(U8*)s >= 0xFE)) {
226017aa 2622#ifdef PERLIO_IS_STDIO
e3f494f1
JH
2623# ifdef __GNU_LIBRARY__
2624# if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
226017aa
DD
2625# define FTELL_FOR_PIPE_IS_BROKEN
2626# endif
e3f494f1
JH
2627# else
2628# ifdef __GLIBC__
2629# if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
2630# define FTELL_FOR_PIPE_IS_BROKEN
2631# endif
2632# endif
226017aa
DD
2633# endif
2634#endif
2635#ifdef FTELL_FOR_PIPE_IS_BROKEN
2636 /* This loses the possibility to detect the bof
2637 * situation on perl -P when the libc5 is being used.
2638 * Workaround? Maybe attach some extra state to PL_rsfp?
2639 */
2640 if (!PL_preprocess)
7e28d3af 2641 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
226017aa 2642#else
7e28d3af 2643 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
226017aa 2644#endif
7e28d3af 2645 if (bof) {
3280af22 2646 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7e28d3af 2647 s = swallow_bom((U8*)s);
e929a76b 2648 }
378cc40b 2649 }
3280af22 2650 if (PL_doextract) {
a0d0e21e 2651 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
3280af22 2652 PL_doextract = FALSE;
a0d0e21e
LW
2653
2654 /* Incest with pod. */
2655 if (*s == '=' && strnEQ(s, "=cut", 4)) {
3280af22
NIS
2656 sv_setpv(PL_linestr, "");
2657 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2658 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
207e3d1a 2659 PL_last_lop = PL_last_uni = Nullch;
3280af22 2660 PL_doextract = FALSE;
a0d0e21e 2661 }
4e553d73 2662 }
463ee0b2 2663 incline(s);
3280af22
NIS
2664 } while (PL_doextract);
2665 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2666 if (PERLDB_LINE && PL_curstash != PL_debstash) {
79072805 2667 SV *sv = NEWSV(85,0);
a687059c 2668
93a17b20 2669 sv_upgrade(sv, SVt_PVMG);
3280af22 2670 sv_setsv(sv,PL_linestr);
57843af0 2671 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
a687059c 2672 }
3280af22 2673 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
207e3d1a 2674 PL_last_lop = PL_last_uni = Nullch;
57843af0 2675 if (CopLINE(PL_curcop) == 1) {
3280af22 2676 while (s < PL_bufend && isSPACE(*s))
79072805 2677 s++;
a0d0e21e 2678 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
79072805 2679 s++;
44a8e56a 2680 d = Nullch;
3280af22 2681 if (!PL_in_eval) {
44a8e56a
PP
2682 if (*s == '#' && *(s+1) == '!')
2683 d = s + 2;
2684#ifdef ALTERNATE_SHEBANG
2685 else {
2686 static char as[] = ALTERNATE_SHEBANG;
2687 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2688 d = s + (sizeof(as) - 1);
2689 }
2690#endif /* ALTERNATE_SHEBANG */
2691 }
2692 if (d) {
b8378b72 2693 char *ipath;
774d564b 2694 char *ipathend;
b8378b72 2695
774d564b 2696 while (isSPACE(*d))
b8378b72
CS
2697 d++;
2698 ipath = d;
774d564b
PP
2699 while (*d && !isSPACE(*d))
2700 d++;
2701 ipathend = d;
2702
2703#ifdef ARG_ZERO_IS_SCRIPT
2704 if (ipathend > ipath) {
2705 /*
2706 * HP-UX (at least) sets argv[0] to the script name,
2707 * which makes $^X incorrect. And Digital UNIX and Linux,
2708 * at least, set argv[0] to the basename of the Perl
2709 * interpreter. So, having found "#!", we'll set it right.
2710 */
2711 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2712 assert(SvPOK(x) || SvGMAGICAL(x));
cc49e20b 2713 if (sv_eq(x, CopFILESV(PL_curcop))) {
774d564b 2714 sv_setpvn(x, ipath, ipathend - ipath);
9607fc9c
PP
2715 SvSETMAGIC(x);
2716 }
774d564b 2717 TAINT_NOT; /* $^X is always tainted, but that's OK */
8ebc5c01 2718 }
774d564b 2719#endif /* ARG_ZERO_IS_SCRIPT */
b8378b72
CS
2720
2721 /*
2722 * Look for options.
2723 */
748a9306 2724 d = instr(s,"perl -");
84e30d1a 2725 if (!d) {
748a9306 2726 d = instr(s,"perl");
84e30d1a
GS
2727#if defined(DOSISH)
2728 /* avoid getting into infinite loops when shebang
2729 * line contains "Perl" rather than "perl" */
2730 if (!d) {
2731 for (d = ipathend-4; d >= ipath; --d) {
2732 if ((*d == 'p' || *d == 'P')
2733 && !ibcmp(d, "perl", 4))
2734 {
2735 break;
2736 }
2737 }
2738 if (d < ipath)
2739 d = Nullch;
2740 }
2741#endif
2742 }
44a8e56a
PP
2743#ifdef ALTERNATE_SHEBANG
2744 /*
2745 * If the ALTERNATE_SHEBANG on this system starts with a
2746 * character that can be part of a Perl expression, then if
2747 * we see it but not "perl", we're probably looking at the
2748 * start of Perl code, not a request to hand off to some
2749 * other interpreter. Similarly, if "perl" is there, but
2750 * not in the first 'word' of the line, we assume the line
2751 * contains the start of the Perl program.
44a8e56a
PP
2752 */
2753 if (d && *s != '#') {
774d564b 2754 char *c = ipath;
44a8e56a
PP
2755 while (*c && !strchr("; \t\r\n\f\v#", *c))
2756 c++;
2757 if (c < d)
2758 d = Nullch; /* "perl" not in first word; ignore */
2759 else
2760 *s = '#'; /* Don't try to parse shebang line */
2761 }
774d564b 2762#endif /* ALTERNATE_SHEBANG */
bf4acbe4 2763#ifndef MACOS_TRADITIONAL
748a9306 2764 if (!d &&
44a8e56a 2765 *s == '#' &&
774d564b 2766 ipathend > ipath &&
3280af22 2767 !PL_minus_c &&
748a9306 2768 !instr(s,"indir") &&
3280af22 2769 instr(PL_origargv[0],"perl"))
748a9306 2770 {
9f68db38 2771 char **newargv;
9f68db38 2772
774d564b
PP
2773 *ipathend = '\0';
2774 s = ipathend + 1;
3280af22 2775 while (s < PL_bufend && isSPACE(*s))
9f68db38 2776 s++;
3280af22
NIS
2777 if (s < PL_bufend) {
2778 Newz(899,newargv,PL_origargc+3,char*);
9f68db38 2779 newargv[1] = s;
3280af22 2780 while (s < PL_bufend && !isSPACE(*s))
9f68db38
LW
2781 s++;
2782 *s = '\0';
3280af22 2783 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
9f68db38
LW
2784 }
2785 else
3280af22 2786 newargv = PL_origargv;
774d564b 2787 newargv[0] = ipath;
b4748376 2788 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
cea2e8a9 2789 Perl_croak(aTHX_ "Can't exec %s", ipath);
9f68db38 2790 }
bf4acbe4 2791#endif
748a9306 2792 if (d) {
3280af22
NIS
2793 U32 oldpdb = PL_perldb;
2794 bool oldn = PL_minus_n;
2795 bool oldp = PL_minus_p;
748a9306
LW
2796
2797 while (*d && !isSPACE(*d)) d++;
bf4acbe4 2798 while (SPACE_OR_TAB(*d)) d++;
748a9306
LW
2799
2800 if (*d++ == '-') {
8cc95fdb
PP
2801 do {
2802 if (*d == 'M' || *d == 'm') {
2803 char *m = d;
2804 while (*d && !isSPACE(*d)) d++;
cea2e8a9 2805 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
8cc95fdb
PP
2806 (int)(d - m), m);
2807 }
2808 d = moreswitches(d);
2809 } while (d);
155aba94
GS
2810 if ((PERLDB_LINE && !oldpdb) ||
2811 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
b084f20b
PP
2812 /* if we have already added "LINE: while (<>) {",
2813 we must not do it again */
748a9306 2814 {
3280af22
NIS
2815 sv_setpv(PL_linestr, "");
2816 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2817 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
207e3d1a 2818 PL_last_lop = PL_last_uni = Nullch;
3280af22 2819 PL_preambled = FALSE;
84902520 2820 if (PERLDB_LINE)
3280af22 2821 (void)gv_fetchfile(PL_origfilename);
748a9306
LW
2822 goto retry;
2823 }
a0d0e21e 2824 }
79072805 2825 }
9f68db38 2826 }
79072805 2827 }
3280af22
NIS
2828 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2829 PL_bufptr = s;
2830 PL_lex_state = LEX_FORMLINE;
cea2e8a9 2831 return yylex();
ae986130 2832 }
378cc40b 2833 goto retry;
4fdae800 2834 case '\r':
6a27c188 2835#ifdef PERL_STRICT_CR
cea2e8a9 2836 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
4e553d73 2837 Perl_croak(aTHX_
cc507455 2838 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
a868473f 2839#endif
4fdae800 2840 case ' ': case '\t': case '\f': case 013:
bf4acbe4
GS
2841#ifdef MACOS_TRADITIONAL
2842 case '\312':
2843#endif
378cc40b
LW
2844 s++;
2845 goto retry;
378cc40b 2846 case '#':
e929a76b 2847 case '\n':
3280af22 2848 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
df0deb90
GS
2849 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2850 /* handle eval qq[#line 1 "foo"\n ...] */
2851 CopLINE_dec(PL_curcop);
2852 incline(s);
2853 }
3280af22 2854 d = PL_bufend;
a687059c 2855 while (s < d && *s != '\n')
378cc40b 2856 s++;
0f85fab0 2857 if (s < d)
378cc40b 2858 s++;
463ee0b2 2859 incline(s);
3280af22
NIS
2860 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2861 PL_bufptr = s;
2862 PL_lex_state = LEX_FORMLINE;
cea2e8a9 2863 return yylex();
a687059c 2864 }
378cc40b 2865 }
a687059c 2866 else {
378cc40b 2867 *s = '\0';
3280af22 2868 PL_bufend = s;
a687059c 2869 }
378cc40b
LW
2870 goto retry;
2871 case '-':
79072805 2872 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
e5edeb50
JH
2873 I32 ftst = 0;
2874
378cc40b 2875 s++;
3280af22 2876 PL_bufptr = s;
748a9306
LW
2877 tmp = *s++;
2878
bf4acbe4 2879 while (s < PL_bufend && SPACE_OR_TAB(*s))
748a9306
LW
2880 s++;
2881
2882 if (strnEQ(s,"=>",2)) {
3280af22 2883 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
4e553d73 2884 DEBUG_T( { PerlIO_printf(Perl_debug_log,
607df283
SC
2885 "### Saw unary minus before =>, forcing word '%s'\n", s);
2886 } )
748a9306
LW
2887 OPERATOR('-'); /* unary minus */
2888 }
3280af22 2889 PL_last_uni = PL_oldbufptr;
748a9306 2890 switch (tmp) {
e5edeb50
JH
2891 case 'r': ftst = OP_FTEREAD; break;
2892 case 'w': ftst = OP_FTEWRITE; break;
2893 case 'x': ftst = OP_FTEEXEC; break;
2894 case 'o': ftst = OP_FTEOWNED; break;
2895 case 'R': ftst = OP_FTRREAD; break;
2896 case 'W': ftst = OP_FTRWRITE; break;
2897 case 'X': ftst = OP_FTREXEC; break;
2898 case 'O': ftst = OP_FTROWNED; break;
2899 case 'e': ftst = OP_FTIS; break;
2900 case 'z': ftst = OP_FTZERO; break;
2901 case 's': ftst = OP_FTSIZE; break;
2902 case 'f': ftst = OP_FTFILE; break;
2903 case 'd': ftst = OP_FTDIR; break;
2904 case 'l': ftst = OP_FTLINK; break;
2905 case 'p': ftst = OP_FTPIPE; break;
2906 case 'S': ftst = OP_FTSOCK; break;
2907 case 'u': ftst = OP_FTSUID; break;
2908 case 'g': ftst = OP_FTSGID; break;
2909 case 'k': ftst = OP_FTSVTX; break;
2910 case 'b': ftst = OP_FTBLK; break;
2911 case 'c': ftst = OP_FTCHR; break;
2912 case 't': ftst = OP_FTTTY; break;
2913 case 'T': ftst = OP_FTTEXT; break;
2914 case 'B': ftst = OP_FTBINARY; break;
2915 case 'M': case 'A': case 'C':
2916 gv_fetchpv("\024",TRUE, SVt_PV);
2917 switch (tmp) {
2918 case 'M': ftst = OP_FTMTIME; break;
2919 case 'A': ftst = OP_FTATIME; break;
2920 case 'C': ftst = OP_FTCTIME; break;
2921 default: break;
2922 }
2923 break;
378cc40b 2924 default:
378cc40b
LW
2925 break;
2926 }
e5edeb50
JH
2927 if (ftst) {
2928 PL_last_lop_op = ftst;
4e553d73 2929 DEBUG_T( { PerlIO_printf(Perl_debug_log,
0844c848 2930 "### Saw file test %c\n", (int)ftst);
e5edeb50 2931 } )
e5edeb50
JH
2932 FTST(ftst);
2933 }
2934 else {
2935 /* Assume it was a minus followed by a one-letter named
2936 * subroutine call (or a -bareword), then. */
95c31fe3 2937 DEBUG_T( { PerlIO_printf(Perl_debug_log,
0844c848
RB
2938 "### %c looked like a file test but was not\n",
2939 (int)ftst);
95c31fe3 2940 } )
e5edeb50
JH
2941 s -= 2;
2942 }
378cc40b 2943 }
a687059c
LW
2944 tmp = *s++;
2945 if (*s == tmp) {
2946 s++;
3280af22 2947 if (PL_expect == XOPERATOR)
79072805
LW
2948 TERM(POSTDEC);
2949 else
2950 OPERATOR(PREDEC);
2951 }
2952 else if (*s == '>') {
2953 s++;
2954 s = skipspace(s);
7e2040f0 2955 if (isIDFIRST_lazy_if(s,UTF)) {
a0d0e21e 2956 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
463ee0b2 2957 TOKEN(ARROW);
79072805 2958 }
748a9306
LW
2959 else if (*s == '$')
2960 OPERATOR(ARROW);
463ee0b2 2961 else
748a9306 2962 TERM(ARROW);
a687059c 2963 }
3280af22 2964 if (PL_expect == XOPERATOR)
79072805
LW
2965 Aop(OP_SUBTRACT);
2966 else {
3280af22 2967 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2f3197b3 2968 check_uni();
79072805 2969 OPERATOR('-'); /* unary minus */
2f3197b3 2970 }
79072805 2971
378cc40b 2972 case '+':
a687059c
LW
2973 tmp = *s++;
2974 if (*s == tmp) {
378cc40b 2975 s++;
3280af22 2976 if (PL_expect == XOPERATOR)
79072805
LW
2977 TERM(POSTINC);
2978 else
2979 OPERATOR(PREINC);
378cc40b 2980 }
3280af22 2981 if (PL_expect == XOPERATOR)
79072805
LW
2982 Aop(OP_ADD);
2983 else {
3280af22 2984 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2f3197b3 2985 check_uni();
a687059c 2986 OPERATOR('+');
2f3197b3 2987 }
a687059c 2988
378cc40b 2989 case '*':
3280af22
NIS
2990 if (PL_expect != XOPERATOR) {
2991 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2992 PL_expect = XOPERATOR;
2993 force_ident(PL_tokenbuf, '*');
2994 if (!*PL_tokenbuf)
a0d0e21e 2995 PREREF('*');
79072805 2996 TERM('*');
a687059c 2997 }
79072805
LW
2998 s++;
2999 if (*s == '*') {
a687059c 3000 s++;
79072805 3001 PWop(OP_POW);
a687059c 3002 }
79072805
LW
3003 Mop(OP_MULTIPLY);
3004
378cc40b 3005 case '%':
3280af22 3006 if (PL_expect == XOPERATOR) {
bbce6d69
PP
3007 ++s;
3008 Mop(OP_MODULO);
a687059c 3009 }
3280af22
NIS
3010 PL_tokenbuf[0] = '%';
3011 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
3012 if (!PL_tokenbuf[1]) {
3013 if (s == PL_bufend)
bbce6d69
PP
3014 yyerror("Final % should be \\% or %name");
3015 PREREF('%');
a687059c 3016 }
3280af22 3017 PL_pending_ident = '%';
bbce6d69 3018 TERM('%');
a687059c 3019
378cc40b 3020 case '^':
79072805 3021 s++;
a0d0e21e 3022 BOop(OP_BIT_XOR);
79072805 3023 case '[':
3280af22 3024 PL_lex_brackets++;
79072805 3025 /* FALL THROUGH */
378cc40b 3026 case '~':
378cc40b 3027 case ',':
378cc40b
LW
3028 tmp = *s++;
3029 OPERATOR(tmp);
a0d0e21e
LW
3030 case ':':
3031 if (s[1] == ':') {
3032 len = 0;
3033 goto just_a_word;
3034 }
3035 s++;
09bef843
SB
3036 switch (PL_expect) {
3037 OP *attrs;
3038 case XOPERATOR:
3039 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
3040 break;
3041 PL_bufptr = s; /* update in case we back off */
3042 goto grabattrs;
3043 case XATTRBLOCK:
3044 PL_expect = XBLOCK;
3045 goto grabattrs;
3046 case XATTRTERM:
3047 PL_expect = XTERMBLOCK;
3048 grabattrs:
3049 s = skipspace(s);
3050 attrs = Nullop;
7e2040f0 3051 while (isIDFIRST_lazy_if(s,UTF)) {
09bef843 3052 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
f9829d6b
GS
3053 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
3054 if (tmp < 0) tmp = -tmp;
3055 switch (tmp) {
3056 case KEY_or:
3057 case KEY_and:
3058 case KEY_for:
3059 case KEY_unless:
3060 case KEY_if:
3061 case KEY_while:
3062 case KEY_until:
3063 goto got_attrs;
3064 default:
3065 break;
3066 }
3067 }
09bef843
SB
3068 if (*d == '(') {
3069 d = scan_str(d,TRUE,TRUE);
3070 if (!d) {
09bef843
SB
3071 /* MUST advance bufptr here to avoid bogus
3072 "at end of line" context messages from yyerror().
3073 */
3074 PL_bufptr = s + len;
3075 yyerror("Unterminated attribute parameter in attribute list");
3076 if (attrs)
3077 op_free(attrs);
3078 return 0; /* EOF indicator */
3079 }
3080 }
3081 if (PL_lex_stuff) {
3082 SV *sv = newSVpvn(s, len);
3083 sv_catsv(sv, PL_lex_stuff);
3084 attrs = append_elem(OP_LIST, attrs,
3085 newSVOP(OP_CONST, 0, sv));
3086 SvREFCNT_dec(PL_lex_stuff);
3087 PL_lex_stuff = Nullsv;
3088 }
3089 else {
78f9721b
SM
3090 if (!PL_in_my && len == 6 && strnEQ(s, "lvalue", len))
3091 CvLVALUE_on(PL_compcv);
3092 else if (!PL_in_my && len == 6 && strnEQ(s, "locked", len))
3093 CvLOCKED_on(PL_compcv);
3094 else if (!PL_in_my && len == 6 && strnEQ(s, "method", len))
3095 CvMETHOD_on(PL_compcv);
87ecf892
DM
3096#ifdef USE_ITHREADS
3097 else if (PL_in_my == KEY_our && len == 6 && strnEQ(s, "shared", len))
3098 GvSHARED_on(cGVOPx_gv(yylval.opval));
3099#endif
78f9721b
SM
3100 /* After we've set the flags, it could be argued that
3101 we don't need to do the attributes.pm-based setting
3102 process, and shouldn't bother appending recognized
3103 flags. To experiment with that, uncomment the
3104 following "else": */
0256094b 3105 else
78f9721b
SM
3106 attrs = append_elem(OP_LIST, attrs,
3107 newSVOP(OP_CONST, 0,
3108 newSVpvn(s, len)));
09bef843
SB
3109 }
3110 s = skipspace(d);
0120eecf 3111 if (*s == ':' && s[1] != ':')
09bef843 3112 s = skipspace(s+1);
0120eecf
GS
3113 else if (s == d)
3114 break; /* require real whitespace or :'s */
09bef843 3115 }
f9829d6b
GS
3116 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
3117 if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) {
09bef843
SB
3118 char q = ((*s == '\'') ? '"' : '\'');
3119 /* If here for an expression, and parsed no attrs, back off. */
3120 if (tmp == '=' && !attrs) {
3121 s = PL_bufptr;
3122 break;
3123 }
3124 /* MUST advance bufptr here to avoid bogus "at end of line"
3125 context messages from yyerror().
3126 */
3127 PL_bufptr = s;
3128 if (!*s)
3129 yyerror("Unterminated attribute list");
3130 else
3131 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
3132 q, *s, q));
3133 if (attrs)
3134 op_free(attrs);
3135 OPERATOR(':');
3136 }
f9829d6b 3137 got_attrs:
09bef843
SB
3138 if (attrs) {
3139 PL_nextval[PL_nexttoke].opval = attrs;
3140 force_next(THING);
3141 }
3142 TOKEN(COLONATTR);
3143 }
a0d0e21e 3144 OPERATOR(':');
8990e307
LW
3145 case '(':
3146 s++;
3280af22
NIS
3147 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
3148 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
a0d0e21e 3149 else
3280af22 3150 PL_expect = XTERM;
a0d0e21e 3151 TOKEN('(');
378cc40b 3152 case ';':
f4dd75d9 3153 CLINE;
378cc40b
LW
3154 tmp = *s++;
3155 OPERATOR(tmp);
3156 case ')':
378cc40b 3157 tmp = *s++;
16d20bd9
AD
3158 s = skipspace(s);
3159 if (*s == '{')
3160 PREBLOCK(tmp);
378cc40b 3161 TERM(tmp);
79072805
LW
3162 case ']':
3163 s++;
3280af22 3164 if (PL_lex_brackets <= 0)
d98d5fff 3165 yyerror("Unmatched right square bracket");
463ee0b2 3166 else
3280af22
NIS
3167 --PL_lex_brackets;
3168 if (PL_lex_state == LEX_INTERPNORMAL) {
3169 if (PL_lex_brackets == 0) {
a0d0e21e 3170 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
3280af22 3171 PL_lex_state = LEX_INTERPEND;
79072805
LW
3172 }
3173 }
4633a7c4 3174 TERM(']');
79072805
LW
3175 case '{':
3176 leftbracket:
79072805 3177 s++;
3280af22
NIS
3178 if (PL_lex_brackets > 100) {
3179 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
3180 if (newlb != PL_lex_brackstack) {
8990e307 3181 SAVEFREEPV(newlb);
3280af22 3182 PL_lex_brackstack = newlb;
8990e307
LW
3183 }
3184 }
3280af22 3185 switch (PL_expect) {
a0d0e21e 3186 case XTERM:
3280af22 3187 if (PL_lex_formbrack) {
a0d0e21e
LW
3188 s--;
3189 PRETERMBLOCK(DO);
3190 }
3280af22
NIS
3191 if (PL_oldoldbufptr == PL_last_lop)
3192 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
a0d0e21e 3193 else
3280af22 3194 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
79072805 3195 OPERATOR(HASHBRACK);
a0d0e21e 3196 case XOPERATOR:
bf4acbe4 3197 while (s < PL_bufend && SPACE_OR_TAB(*s))
748a9306 3198 s++;
44a8e56a 3199 d = s;
3280af22
NIS
3200 PL_tokenbuf[0] = '\0';
3201 if (d < PL_bufend && *d == '-') {
3202 PL_tokenbuf[0] = '-';
44a8e56a 3203 d++;
bf4acbe4 3204 while (d < PL_bufend && SPACE_OR_TAB(*d))
44a8e56a
PP
3205 d++;
3206 }
7e2040f0 3207 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3280af22 3208 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
8903cb82 3209 FALSE, &len);
bf4acbe4 3210 while (d < PL_bufend && SPACE_OR_TAB(*d))
748a9306
LW
3211 d++;
3212 if (*d == '}') {
3280af22 3213 char minus = (PL_tokenbuf[0] == '-');
44a8e56a
PP
3214 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3215 if (minus)
3216 force_next('-');
748a9306
LW
3217 }
3218 }
3219 /* FALL THROUGH */
09bef843 3220 case XATTRBLOCK:
748a9306 3221 case XBLOCK:
3280af22
NIS
3222 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3223 PL_expect = XSTATE;
a0d0e21e 3224 break;
09bef843 3225 case XATTRTERM:
a0d0e21e 3226 case XTERMBLOCK:
3280af22
NIS
3227 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3228 PL_expect = XSTATE;
a0d0e21e
LW
3229 break;
3230 default: {
3231 char *t;
3280af22
NIS
3232 if (PL_oldoldbufptr == PL_last_lop)
3233 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
a0d0e21e 3234 else
3280af22 3235 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
a0d0e21e 3236 s = skipspace(s);
09ecc4b6 3237 if (*s == '}')
a0d0e21e 3238 OPERATOR(HASHBRACK);
b8a4b1be
GS
3239 /* This hack serves to disambiguate a pair of curlies
3240 * as being a block or an anon hash. Normally, expectation
3241 * determines that, but in cases where we're not in a
3242 * position to expect anything in particular (like inside
3243 * eval"") we have to resolve the ambiguity. This code
3244 * covers the case where the first term in the curlies is a
3245 * quoted string. Most other cases need to be explicitly
3246 * disambiguated by prepending a `+' before the opening
3247 * curly in order to force resolution as an anon hash.
3248 *
3249 * XXX should probably propagate the outer expectation
3250 * into eval"" to rely less on this hack, but that could
3251 * potentially break current behavior of eval"".
3252 * GSAR 97-07-21
3253 */
3254 t = s;
3255 if (*s == '\'' || *s == '"' || *s == '`') {
3256 /* common case: get past first string, handling escapes */
3280af22 3257 for (t++; t < PL_bufend && *t != *s;)
b8a4b1be
GS
3258 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3259 t++;
3260 t++;
a0d0e21e 3261 }
b8a4b1be 3262 else if (*s == 'q') {
3280af22 3263 if (++t < PL_bufend
b8a4b1be 3264 && (!isALNUM(*t)
3280af22 3265 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
0505442f
GS
3266 && !isALNUM(*t))))
3267 {
b8a4b1be
GS
3268 char *tmps;
3269 char open, close, term;
3270 I32 brackets = 1;
3271
3280af22 3272 while (t < PL_bufend && isSPACE(*t))
b8a4b1be
GS
3273 t++;
3274 term = *t;
3275 open = term;
3276 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3277 term = tmps[5];
3278 close = term;
3279 if (open == close)
3280af22
NIS
3280 for (t++; t < PL_bufend; t++) {
3281 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
b8a4b1be 3282 t++;
6d07e5e9 3283 else if (*t == open)
b8a4b1be
GS
3284 break;
3285 }
3286 else
3280af22
NIS
3287 for (t++; t < PL_bufend; t++) {
3288 if (*t == '\\' && t+1 < PL_bufend)
b8a4b1be 3289 t++;
6d07e5e9 3290 else if (*t == close && --brackets <= 0)
b8a4b1be
GS
3291 break;
3292 else if (*t == open)
3293 brackets++;
3294 }
3295 }
3296 t++;
a0d0e21e 3297 }
7e2040f0 3298 else if (isALNUM_lazy_if(t,UTF)) {
0505442f 3299 t += UTF8SKIP(t);
7e2040f0 3300 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
0505442f 3301 t += UTF8SKIP(t);
a0d0e21e 3302 }
3280af22 3303 while (t < PL_bufend && isSPACE(*t))
a0d0e21e 3304 t++;
b8a4b1be
GS
3305 /* if comma follows first term, call it an anon hash */
3306 /* XXX it could be a comma expression with loop modifiers */
3280af22 3307 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
b8a4b1be 3308 || (*t == '=' && t[1] == '>')))
a0d0e21e 3309 OPERATOR(HASHBRACK);
3280af22 3310 if (PL_expect == XREF)
4e4e412b 3311 PL_expect = XTERM;
a0d0e21e 3312 else {
3280af22
NIS
3313 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3314 PL_expect = XSTATE;
a0d0e21e 3315 }
8990e307 3316 }
a0d0e21e 3317 break;
463ee0b2 3318 }
57843af0 3319 yylval.ival = CopLINE(PL_curcop);
79072805 3320 if (isSPACE(*s) || *s == '#')
3280af22 3321 PL_copline = NOLINE; /* invalidate current command line number */
79072805 3322 TOKEN('{');
378cc40b 3323 case '}':
79072805
LW
3324 rightbracket:
3325 s++;
3280af22 3326 if (PL_lex_brackets <= 0)
d98d5fff 3327 yyerror("Unmatched right curly bracket");
463ee0b2 3328 else
3280af22 3329 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
c2e66d9e 3330 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3280af22
NIS
3331 PL_lex_formbrack = 0;
3332 if (PL_lex_state == LEX_INTERPNORMAL) {
3333 if (PL_lex_brackets == 0) {
9059aa12
LW
3334 if (PL_expect & XFAKEBRACK) {
3335 PL_expect &= XENUMMASK;
3280af22
NIS
3336 PL_lex_state = LEX_INTERPEND;
3337 PL_bufptr = s;
cea2e8a9 3338 return yylex(); /* ignore fake brackets */
79072805 3339 }
fa83b5b6 3340 if (*s == '-' && s[1] == '>')
3280af22 3341 PL_lex_state = LEX_INTERPENDMAYBE;
fa83b5b6 3342 else if (*s != '[' && *s != '{')
3280af22 3343 PL_lex_state = LEX_INTERPEND;
79072805
LW
3344 }
3345 }
9059aa12
LW
3346 if (PL_expect & XFAKEBRACK) {
3347 PL_expect &= XENUMMASK;
3280af22 3348 PL_bufptr = s;
cea2e8a9 3349 return yylex(); /* ignore fake brackets */
748a9306 3350 }
79072805
LW
3351 force_next('}');
3352 TOKEN(';');
378cc40b
LW
3353 case '&':
3354 s++;
3355 tmp = *s++;
3356 if (tmp == '&')
a0d0e21e 3357 AOPERATOR(ANDAND);
378cc40b 3358 s--;
3280af22 3359 if (PL_expect == XOPERATOR) {
7e2040f0
GS
3360 if (ckWARN(WARN_SEMICOLON)
3361 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3362 {
57843af0 3363 CopLINE_dec(PL_curcop);
cea2e8a9 3364 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
57843af0 3365 CopLINE_inc(PL_curcop);
463ee0b2 3366 }
79072805 3367 BAop(OP_BIT_AND);
463ee0b2 3368 }
79072805 3369
3280af22
NIS
3370 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3371 if (*PL_tokenbuf) {
3372 PL_expect = XOPERATOR;
3373 force_ident(PL_tokenbuf, '&');
463ee0b2 3374 }
79072805
LW
3375 else
3376 PREREF('&');
c07a80fd 3377 yylval.ival = (OPpENTERSUB_AMPER<<8);
79072805
LW
3378 TERM('&');
3379
378cc40b
LW
3380 case '|':
3381 s++;
3382 tmp = *s++;
3383 if (tmp == '|')
a0d0e21e 3384 AOPERATOR(OROR);
378cc40b 3385 s--;
79072805 3386 BOop(OP_BIT_OR);
378cc40b
LW
3387 case '=':
3388 s++;
3389 tmp = *s++;
3390 if (tmp == '=')
79072805
LW
3391 Eop(OP_EQ);
3392 if (tmp == '>')
3393 OPERATOR(',');
378cc40b 3394 if (tmp == '~')
79072805 3395 PMop(OP_MATCH);
599cee73 3396 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
cea2e8a9 3397 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
378cc40b 3398 s--;
3280af22
NIS
3399 if (PL_expect == XSTATE && isALPHA(tmp) &&
3400 (s == PL_linestart+1 || s[-2] == '\n') )
748a9306 3401 {
3280af22
NIS
3402 if (PL_in_eval && !PL_rsfp) {
3403 d = PL_bufend;
a5f75d66
AD
3404 while (s < d) {
3405 if (*s++ == '\n') {
3406 incline(s);
3407 if (strnEQ(s,"=cut",4)) {
3408 s = strchr(s,'\n');
3409 if (s)
3410 s++;
3411 else
3412 s = d;
3413 incline(s);
3414 goto retry;
3415 }
3416 }
3417 }
3418 goto retry;
3419 }
3280af22
NIS
3420 s = PL_bufend;
3421 PL_doextract = TRUE;
a0d0e21e
LW
3422 goto retry;
3423 }
3280af22 3424 if (PL_lex_brackets < PL_lex_formbrack) {
a0d0e21e 3425 char *t;
51882d45 3426#ifdef PERL_STRICT_CR
bf4acbe4 3427 for (t = s; SPACE_OR_TAB(*t); t++) ;
51882d45 3428#else
bf4acbe4 3429 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
51882d45 3430#endif
a0d0e21e
LW
3431 if (*t == '\n' || *t == '#') {
3432 s--;
3280af22 3433 PL_expect = XBLOCK;
a0d0e21e
LW
3434 goto leftbracket;
3435 }
79072805 3436 }
a0d0e21e
LW
3437 yylval.ival = 0;
3438 OPERATOR(ASSIGNOP);
378cc40b
LW
3439 case '!':
3440 s++;
3441 tmp = *s++;
3442 if (tmp == '=')
79072805 3443 Eop(OP_NE);
378cc40b 3444 if (tmp == '~')
79072805 3445 PMop(OP_NOT);
378cc40b
LW
3446 s--;
3447 OPERATOR('!');
3448 case '<':
3280af22 3449 if (PL_expect != XOPERATOR) {
93a17b20 3450 if (s[1] != '<' && !strchr(s,'>'))
2f3197b3 3451 check_uni();
79072805
LW
3452 if (s[1] == '<')
3453 s = scan_heredoc(s);
3454 else
3455 s = scan_inputsymbol(s);
3456 TERM(sublex_start());
378cc40b
LW
3457 }
3458 s++;
3459 tmp = *s++;
3460 if (tmp == '<')
79072805 3461 SHop(OP_LEFT_SHIFT);
395c3793
LW
3462 if (tmp == '=') {
3463 tmp = *s++;
3464 if (tmp == '>')
79072805 3465 Eop(OP_NCMP);
395c3793 3466 s--;
79072805 3467 Rop(OP_LE);
395c3793 3468 }
378cc40b 3469 s--;
79072805 3470 Rop(OP_LT);
378cc40b
LW
3471 case '>':
3472 s++;
3473 tmp = *s++;
3474 if (tmp == '>')
79072805 3475 SHop(OP_RIGHT_SHIFT);
378cc40b 3476 if (tmp == '=')
79072805 3477 Rop(OP_GE);
378cc40b 3478 s--;
79072805 3479 Rop(OP_GT);
378cc40b
LW
3480
3481 case '$':
bbce6d69
PP
3482 CLINE;
3483
3280af22
NIS
3484 if (PL_expect == XOPERATOR) {
3485 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3486 PL_expect = XTERM;
a0d0e21e 3487 depcom();
bbce6d69 3488 return ','; /* grandfather non-comma-format format */
a0d0e21e 3489 }
8990e307 3490 }
a0d0e21e 3491
7e2040f0 3492 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3280af22 3493 PL_tokenbuf[0] = '@';
376b8730
SM
3494 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3495 sizeof PL_tokenbuf - 1, FALSE);
3496 if (PL_expect == XOPERATOR)
3497 no_op("Array length", s);
3280af22 3498 if (!PL_tokenbuf[1])
a0d0e21e 3499 PREREF(DOLSHARP);
3280af22
NIS
3500 PL_expect = XOPERATOR;
3501 PL_pending_ident = '#';
463ee0b2 3502 TOKEN(DOLSHARP);
79072805 3503 }
bbce6d69 3504
3280af22 3505 PL_tokenbuf[0] = '$';
376b8730
SM
3506 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3507 sizeof PL_tokenbuf - 1, FALSE);
3508 if (PL_expect == XOPERATOR)
3509 no_op("Scalar", s);
3280af22
NIS
3510 if (!PL_tokenbuf[1]) {
3511 if (s == PL_bufend)
bbce6d69
PP
3512 yyerror("Final $ should be \\$ or $name");
3513 PREREF('$');
8990e307 3514 }
a0d0e21e 3515
bbce6d69 3516 /* This kludge not intended to be bulletproof. */
3280af22 3517 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
bbce6d69 3518 yylval.opval = newSVOP(OP_CONST, 0,
b448e4fe 3519 newSViv(PL_compiling.cop_arybase));
bbce6d69
PP
3520 yylval.opval->op_private = OPpCONST_ARYBASE;
3521 TERM(THING);
3522 }
3523
ff68c719 3524 d = s;
69d2bceb 3525 tmp = (I32)*s;
3280af22 3526 if (PL_lex_state == LEX_NORMAL)
ff68c719
PP
3527 s = skipspace(s);
3528
3280af22 3529 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
bbce6d69
PP
3530 char *t;
3531 if (*s == '[') {
3280af22 3532 PL_tokenbuf[0] = '@';
599cee73 3533 if (ckWARN(WARN_SYNTAX)) {
bbce6d69 3534 for(t = s + 1;
7e2040f0 3535 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
bbce6d69 3536 t++) ;
a0d0e21e 3537 if (*t++ == ',') {
3280af22
NIS
3538 PL_bufptr = skipspace(PL_bufptr);
3539 while (t < PL_bufend && *t != ']')
bbce6d69 3540 t++;
cea2e8a9 3541 Perl_warner(aTHX_ WARN_SYNTAX,
599cee73
PM
3542 "Multidimensional syntax %.*s not supported",
3543 (t - PL_bufptr) + 1, PL_bufptr);
a0d0e21e
LW
3544 }
3545 }
bbce6d69
PP
3546 }
3547 else if (*s == '{') {
3280af22 3548 PL_tokenbuf[0] = '%';
599cee73 3549 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
bbce6d69
PP
3550 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3551 {
3280af22 3552 char tmpbuf[sizeof PL_tokenbuf];
a0d0e21e
LW
3553 STRLEN len;
3554 for (t++; isSPACE(*t); t++) ;
7e2040f0 3555 if (isIDFIRST_lazy_if(t,UTF)) {
8903cb82<