Fail hard if strip_names_and_args fails.
[p5sagit/Devel-Declare.git] / stolen_chunk_of_toke.c
CommitLineData
e807ee50 1/* stolen_chunk_of_toke.c - from perl 5.8.8 toke.c
2 *
3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
4 * 2000, 2001, 2002, 2003, 2004, 2005, 2006, by Larry Wall and others
5 *
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
8 *
9 */
10
11/*
12 * "It all comes from here, the stench and the peril." --Frodo
13 */
14
15/*
16 * this is all blatantly stolen. I sincerely hopes it doesn't fuck anything
17 * up but if it does blame me (Matt S Trout), not the poor original authors
18 */
19
71511651 20/* the following #defines are stolen from assorted headers, not toke.c (mst) */
e807ee50 21
78160085 22#define skipspace(a) S_skipspace(aTHX_ a)
23#define incline(a) S_incline(aTHX_ a)
24#define filter_gets(a,b,c) S_filter_gets(aTHX_ a,b,c)
25#define scan_str(a,b,c) S_scan_str(aTHX_ a,b,c)
26#define scan_word(a,b,c,d,e) S_scan_word(aTHX_ a,b,c,d,e)
92997e69 27#define scan_ident(a,b,c,d,e) S_scan_ident(aTHX_ a,b,c,d,e)
78160085 28
29STATIC void S_incline(pTHX_ char *s);
30STATIC char* S_skipspace(pTHX_ char *s);
31STATIC char * S_filter_gets(pTHX_ SV *sv, PerlIO *fp, STRLEN append);
32STATIC char* S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims);
33STATIC char* S_scan_word(pTHX_ char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp);
34
e807ee50 35#define DPTR2FPTR(t,p) ((t)PTR2nat(p)) /* data pointer to function pointer */
36#define FPTR2DPTR(t,p) ((t)PTR2nat(p)) /* function pointer to data pointer */
c630715a 37#define PTR2nat(p) (PTRV)(p) /* pointer to integer of PTRSIZE */
e807ee50 38#define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
14ab3472 39
40/* conditionalise these two because as of 5.9.5 we already get them from
41 the headers (mst) */
42#ifndef Newx
e807ee50 43#define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)safemalloc((MEM_SIZE)((n)*sizeof(t)))))
14ab3472 44#endif
45#ifndef SvPVX_const
0ba8c7aa 46#define SvPVX_const(sv) ((const char*) (0 + SvPVX(sv)))
14ab3472 47#endif
48
0ba8c7aa 49#define SvPV_renew(sv,n) \
50 STMT_START { SvLEN_set(sv, n); \
51 SvPV_set((sv), (MEM_WRAP_CHECK_(n,char) \
52 (char*)saferealloc((Malloc_t)SvPVX(sv), \
53 (MEM_SIZE)((n))))); \
54 } STMT_END
e807ee50 55
d6e7537a 56#define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
57
c630715a 58/* On MacOS, respect nonbreaking spaces */
59#ifdef MACOS_TRADITIONAL
60#define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
61#else
62#define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
63#endif
e807ee50 64
0ba8c7aa 65#define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
66
e807ee50 67#define LEX_NORMAL 10 /* normal code (ie not within "...") */
68#define LEX_INTERPNORMAL 9 /* code within a string, eg "$foo[$x+1]" */
69#define LEX_INTERPCASEMOD 8 /* expecting a \U, \Q or \E etc */
70#define LEX_INTERPPUSH 7 /* starting a new sublex parse level */
71#define LEX_INTERPSTART 6 /* expecting the start of a $var */
72
73 /* at end of code, eg "$x" followed by: */
74#define LEX_INTERPEND 5 /* ... eg not one of [, { or -> */
75#define LEX_INTERPENDMAYBE 4 /* ... eg one of [, { or -> */
76
77#define LEX_INTERPCONCAT 3 /* expecting anything, eg at start of
78 string or after \E, $foo, etc */
79#define LEX_INTERPCONST 2 /* NOT USED */
80#define LEX_FORMLINE 1 /* expecting a format line */
81#define LEX_KNOWNEXT 0 /* next token known; just return it */
82
14ab3472 83/* and these two are my own madness (mst) */
71511651 84
85#if PERL_REVISION == 5 && PERL_VERSION == 8 && PERL_SUBVERSION >= 8
86#define PERL_5_8_8_PLUS
87#endif
88
e38f5a74 89#if PERL_REVISION == 5 && PERL_VERSION > 8
90#define PERL_5_9_PLUS
91#endif
92
93#ifdef PERL_5_9_PLUS
14ab3472 94/* 5.9+ moves a bunch of things to a PL_parser struct so we need to
95 declare the backcompat macros for things to still work (mst) */
96
97/* XXX temporary backwards compatibility */
98#define PL_lex_brackets (PL_parser->lex_brackets)
99#define PL_lex_brackstack (PL_parser->lex_brackstack)
100#define PL_lex_casemods (PL_parser->lex_casemods)
101#define PL_lex_casestack (PL_parser->lex_casestack)
102#define PL_lex_defer (PL_parser->lex_defer)
103#define PL_lex_dojoin (PL_parser->lex_dojoin)
104#define PL_lex_expect (PL_parser->lex_expect)
105#define PL_lex_formbrack (PL_parser->lex_formbrack)
106#define PL_lex_inpat (PL_parser->lex_inpat)
107#define PL_lex_inwhat (PL_parser->lex_inwhat)
108#define PL_lex_op (PL_parser->lex_op)
109#define PL_lex_repl (PL_parser->lex_repl)
110#define PL_lex_starts (PL_parser->lex_starts)
111#define PL_lex_stuff (PL_parser->lex_stuff)
112#define PL_multi_start (PL_parser->multi_start)
113#define PL_multi_open (PL_parser->multi_open)
114#define PL_multi_close (PL_parser->multi_close)
115#define PL_pending_ident (PL_parser->pending_ident)
116#define PL_preambled (PL_parser->preambled)
117#define PL_sublex_info (PL_parser->sublex_info)
118#define PL_linestr (PL_parser->linestr)
119#define PL_sublex_info (PL_parser->sublex_info)
120#define PL_linestr (PL_parser->linestr)
121#define PL_expect (PL_parser->expect)
122#define PL_copline (PL_parser->copline)
123#define PL_bufptr (PL_parser->bufptr)
124#define PL_oldbufptr (PL_parser->oldbufptr)
125#define PL_oldoldbufptr (PL_parser->oldoldbufptr)
126#define PL_linestart (PL_parser->linestart)
127#define PL_bufend (PL_parser->bufend)
128#define PL_last_uni (PL_parser->last_uni)
129#define PL_last_lop (PL_parser->last_lop)
130#define PL_last_lop_op (PL_parser->last_lop_op)
131#define PL_lex_state (PL_parser->lex_state)
132#define PL_rsfp (PL_parser->rsfp)
133#define PL_rsfp_filters (PL_parser->rsfp_filters)
134#define PL_in_my (PL_parser->in_my)
135#define PL_in_my_stash (PL_parser->in_my_stash)
136#define PL_tokenbuf (PL_parser->tokenbuf)
137#define PL_multi_end (PL_parser->multi_end)
138#define PL_error_count (PL_parser->error_count)
139/* these three are from the non-PERL_MAD path but I don't -think- I need
140 the PERL_MAD stuff since my code isn't really populating things (mst) */
141# define PL_nexttoke (PL_parser->nexttoke)
142# define PL_nexttype (PL_parser->nexttype)
143# define PL_nextval (PL_parser->nextval)
144/* end of backcompat macros form 5.9 toke.c (mst) */
78160085 145#endif
14ab3472 146
147/* when ccflags include -DDEBUGGING we need this for earlier 5.8 perls */
148#ifndef SvPV_nolen_const
149#define SvPV_nolen_const SvPV_nolen
150#endif
151
71511651 152/* and now we're back to the toke.c stuff again (mst) */
153
e807ee50 154static const char ident_too_long[] =
155 "Identifier too long";
156static const char c_without_g[] =
157 "Use of /c modifier is meaningless without /g";
158static const char c_in_subst[] =
159 "Use of /c modifier is meaningless in s///";
160
161#ifdef USE_UTF8_SCRIPTS
162# define UTF (!IN_BYTES)
163#else
164# define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || (PL_hints & HINT_UTF8))
165#endif
166
167/* Invoke the idxth filter function for the current rsfp. */
168/* maxlen 0 = read one text line */
169I32
170Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
171{
172 filter_t funcp;
173 SV *datasv = NULL;
174
175 if (!PL_rsfp_filters)
176 return -1;
177 if (idx > AvFILLp(PL_rsfp_filters)) { /* Any more filters? */
178 /* Provide a default input filter to make life easy. */
179 /* Note that we append to the line. This is handy. */
180 DEBUG_P(PerlIO_printf(Perl_debug_log,
181 "filter_read %d: from rsfp\n", idx));
182 if (maxlen) {
183 /* Want a block */
184 int len ;
185 const int old_len = SvCUR(buf_sv);
186
187 /* ensure buf_sv is large enough */
188 SvGROW(buf_sv, (STRLEN)(old_len + maxlen)) ;
189 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
190 if (PerlIO_error(PL_rsfp))
191 return -1; /* error */
192 else
193 return 0 ; /* end of file */
194 }
195 SvCUR_set(buf_sv, old_len + len) ;
196 } else {
197 /* Want a line */
198 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
199 if (PerlIO_error(PL_rsfp))
200 return -1; /* error */
201 else
202 return 0 ; /* end of file */
203 }
204 }
205 return SvCUR(buf_sv);
206 }
207 /* Skip this filter slot if filter has been deleted */
208 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef) {
209 DEBUG_P(PerlIO_printf(Perl_debug_log,
210 "filter_read %d: skipped (filter deleted)\n",
211 idx));
212 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
213 }
214 /* Get function pointer hidden within datasv */
215 funcp = DPTR2FPTR(filter_t, IoANY(datasv));
216 DEBUG_P(PerlIO_printf(Perl_debug_log,
217 "filter_read %d: via function %p (%s)\n",
218 idx, datasv, SvPV_nolen_const(datasv)));
219 /* Call function. The function is expected to */
220 /* call "FILTER_READ(idx+1, buf_sv)" first. */
221 /* Return: <0:error, =0:eof, >0:not eof */
222 return (*funcp)(aTHX_ idx, buf_sv, maxlen);
223}
224
225STATIC char *
226S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
227{
228#ifdef PERL_CR_FILTER
229 if (!PL_rsfp_filters) {
230 filter_add(S_cr_textfilter,NULL);
231 }
232#endif
233 if (PL_rsfp_filters) {
234 if (!append)
235 SvCUR_set(sv, 0); /* start with empty line */
236 if (FILTER_READ(0, sv, 0) > 0)
237 return ( SvPVX(sv) ) ;
238 else
239 return Nullch ;
240 }
241 else
242 return (sv_gets(sv, fp, append));
243}
244
245/*
246 * S_skipspace
247 * Called to gobble the appropriate amount and type of whitespace.
248 * Skips comments as well.
249 */
250
251STATIC char *
252S_skipspace(pTHX_ register char *s)
253{
254 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
255 while (s < PL_bufend && SPACE_OR_TAB(*s))
256 s++;
257 return s;
258 }
259 for (;;) {
260 STRLEN prevlen;
261 SSize_t oldprevlen, oldoldprevlen;
262 SSize_t oldloplen = 0, oldunilen = 0;
263 while (s < PL_bufend && isSPACE(*s)) {
264 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
265 incline(s);
266 }
267
268 /* comment */
269 if (s < PL_bufend && *s == '#') {
270 while (s < PL_bufend && *s != '\n')
271 s++;
272 if (s < PL_bufend) {
273 s++;
274 if (PL_in_eval && !PL_rsfp) {
275 incline(s);
276 continue;
277 }
278 }
279 }
280
281 /* only continue to recharge the buffer if we're at the end
282 * of the buffer, we're not reading from a source filter, and
283 * we're in normal lexing mode
284 */
285 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
286 PL_lex_state == LEX_FORMLINE)
287 return s;
288
289 /* try to recharge the buffer */
290 if ((s = filter_gets(PL_linestr, PL_rsfp,
291 (prevlen = SvCUR(PL_linestr)))) == Nullch)
292 {
293 /* end of file. Add on the -p or -n magic */
294 if (PL_minus_p) {
295 sv_setpv(PL_linestr,
296 ";}continue{print or die qq(-p destination: $!\\n);}");
297 PL_minus_n = PL_minus_p = 0;
298 }
299 else if (PL_minus_n) {
300 sv_setpvn(PL_linestr, ";}", 2);
301 PL_minus_n = 0;
302 }
303 else
304 sv_setpvn(PL_linestr,";", 1);
305
306 /* reset variables for next time we lex */
307 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
308 = SvPVX(PL_linestr);
309 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
310 PL_last_lop = PL_last_uni = Nullch;
311
35fc8ada 312 /* In perl versions previous to p4-rawid: //depot/perl@32954 -P
313 * preprocessors were supported here. We don't support -P at all, even
314 * on perls that support it, and use the following chunk from blead
315 * perl. (rafl)
316 */
317
318 /* Close the filehandle. Could be from
e807ee50 319 * STDIN, or a regular file. If we were reading code from
320 * STDIN (because the commandline held no -e or filename)
321 * then we don't close it, we reset it so the code can
322 * read from STDIN too.
323 */
324
35fc8ada 325 if ((PerlIO*)PL_rsfp == PerlIO_stdin())
e807ee50 326 PerlIO_clearerr(PL_rsfp);
327 else
328 (void)PerlIO_close(PL_rsfp);
329 PL_rsfp = Nullfp;
330 return s;
331 }
332
333 /* not at end of file, so we only read another line */
334 /* make corresponding updates to old pointers, for yyerror() */
335 oldprevlen = PL_oldbufptr - PL_bufend;
336 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
337 if (PL_last_uni)
338 oldunilen = PL_last_uni - PL_bufend;
339 if (PL_last_lop)
340 oldloplen = PL_last_lop - PL_bufend;
341 PL_linestart = PL_bufptr = s + prevlen;
342 PL_bufend = s + SvCUR(PL_linestr);
343 s = PL_bufptr;
344 PL_oldbufptr = s + oldprevlen;
345 PL_oldoldbufptr = s + oldoldprevlen;
346 if (PL_last_uni)
347 PL_last_uni = s + oldunilen;
348 if (PL_last_lop)
349 PL_last_lop = s + oldloplen;
350 incline(s);
351
352 /* debugger active and we're not compiling the debugger code,
353 * so store the line into the debugger's array of lines
354 */
355 if (PERLDB_LINE && PL_curstash != PL_debstash) {
356 SV * const sv = NEWSV(85,0);
357
358 sv_upgrade(sv, SVt_PVMG);
359 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
360 (void)SvIOK_on(sv);
361 SvIV_set(sv, 0);
362 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
363 }
364 }
365}
366
367STATIC char *
368S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
369{
370 register char *d = dest;
371 register char * const e = d + destlen - 3; /* two-character token, ending NUL */
372 for (;;) {
373 if (d >= e)
374 Perl_croak(aTHX_ ident_too_long);
375 if (isALNUM(*s)) /* UTF handled below */
376 *d++ = *s++;
377 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
378 *d++ = ':';
379 *d++ = ':';
380 s++;
381 }
382 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
383 *d++ = *s++;
384 *d++ = *s++;
385 }
386 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
387 char *t = s + UTF8SKIP(s);
388 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
389 t += UTF8SKIP(t);
390 if (d + (t - s) > e)
391 Perl_croak(aTHX_ ident_too_long);
392 Copy(s, d, t - s, char);
393 d += t - s;
394 s = t;
395 }
396 else {
397 *d = '\0';
398 *slp = d - dest;
399 return s;
400 }
401 }
402}
403
404/*
405 * S_incline
406 * This subroutine has nothing to do with tilting, whether at windmills
407 * or pinball tables. Its name is short for "increment line". It
408 * increments the current line number in CopLINE(PL_curcop) and checks
409 * to see whether the line starts with a comment of the form
410 * # line 500 "foo.pm"
411 * If so, it sets the current line number and file to the values in the comment.
412 */
413
414STATIC void
78160085 415S_incline(pTHX_ char *s)
e807ee50 416{
417 char *t;
78160085 418 char *n;
419 char *e;
e807ee50 420 char ch;
421
422 CopLINE_inc(PL_curcop);
423 if (*s++ != '#')
424 return;
425 while (SPACE_OR_TAB(*s)) s++;
426 if (strnEQ(s, "line", 4))
427 s += 4;
428 else
429 return;
430 if (SPACE_OR_TAB(*s))
431 s++;
432 else
433 return;
434 while (SPACE_OR_TAB(*s)) s++;
435 if (!isDIGIT(*s))
436 return;
437 n = s;
438 while (isDIGIT(*s))
439 s++;
440 while (SPACE_OR_TAB(*s))
441 s++;
442 if (*s == '"' && (t = strchr(s+1, '"'))) {
443 s++;
444 e = t + 1;
445 }
446 else {
78160085 447 for (t = s; !isSPACE(*t); t++) ;
e807ee50 448 e = t;
449 }
450 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
451 e++;
452 if (*e != '\n' && *e != '\0')
453 return; /* false alarm */
454
455 ch = *t;
456 *t = '\0';
457 if (t - s > 0) {
71511651 458/* this chunk was added to S_incline during 5.8.8. I don't know why but I don't
459 honestly care since I probably want to be bug-compatible anyway (mst) */
460
461/* ... my kingdom for a perl parser in perl ... (mst) */
462
463#ifdef PERL_5_8_8_PLUS
e807ee50 464#ifndef USE_ITHREADS
465 const char *cf = CopFILE(PL_curcop);
466 if (cf && strlen(cf) > 7 && strnEQ(cf, "(eval ", 6)) {
467 /* must copy *{"::_<(eval N)[oldfilename:L]"}
468 * to *{"::_<newfilename"} */
469 char smallbuf[256], smallbuf2[256];
470 char *tmpbuf, *tmpbuf2;
471 GV **gvp, *gv2;
472 STRLEN tmplen = strlen(cf);
473 STRLEN tmplen2 = strlen(s);
474 if (tmplen + 3 < sizeof smallbuf)
475 tmpbuf = smallbuf;
476 else
477 Newx(tmpbuf, tmplen + 3, char);
478 if (tmplen2 + 3 < sizeof smallbuf2)
479 tmpbuf2 = smallbuf2;
480 else
481 Newx(tmpbuf2, tmplen2 + 3, char);
482 tmpbuf[0] = tmpbuf2[0] = '_';
483 tmpbuf[1] = tmpbuf2[1] = '<';
484 memcpy(tmpbuf + 2, cf, ++tmplen);
485 memcpy(tmpbuf2 + 2, s, ++tmplen2);
486 ++tmplen; ++tmplen2;
487 gvp = (GV**)hv_fetch(PL_defstash, tmpbuf, tmplen, FALSE);
488 if (gvp) {
489 gv2 = *(GV**)hv_fetch(PL_defstash, tmpbuf2, tmplen2, TRUE);
490 if (!isGV(gv2))
491 gv_init(gv2, PL_defstash, tmpbuf2, tmplen2, FALSE);
492 /* adjust ${"::_<newfilename"} to store the new file name */
493 GvSV(gv2) = newSVpvn(tmpbuf2 + 2, tmplen2 - 2);
494 GvHV(gv2) = (HV*)SvREFCNT_inc(GvHV(*gvp));
495 GvAV(gv2) = (AV*)SvREFCNT_inc(GvAV(*gvp));
496 }
497 if (tmpbuf != smallbuf) Safefree(tmpbuf);
498 if (tmpbuf2 != smallbuf2) Safefree(tmpbuf2);
499 }
500#endif
71511651 501#endif
502/* second endif closes out the "are we 5.8.(8+)" conditional */
e807ee50 503 CopFILE_free(PL_curcop);
504 CopFILE_set(PL_curcop, s);
505 }
506 *t = ch;
507 CopLINE_set(PL_curcop, atoi(n)-1);
508}
0ba8c7aa 509
510/* scan_str
511 takes: start position in buffer
512 keep_quoted preserve \ on the embedded delimiter(s)
513 keep_delims preserve the delimiters around the string
514 returns: position to continue reading from buffer
515 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
516 updates the read buffer.
517
518 This subroutine pulls a string out of the input. It is called for:
519 q single quotes q(literal text)
520 ' single quotes 'literal text'
521 qq double quotes qq(interpolate $here please)
522 " double quotes "interpolate $here please"
523 qx backticks qx(/bin/ls -l)
524 ` backticks `/bin/ls -l`
525 qw quote words @EXPORT_OK = qw( func() $spam )
526 m// regexp match m/this/
527 s/// regexp substitute s/this/that/
528 tr/// string transliterate tr/this/that/
529 y/// string transliterate y/this/that/
530 ($*@) sub prototypes sub foo ($)
531 (stuff) sub attr parameters sub foo : attr(stuff)
532 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
533
534 In most of these cases (all but <>, patterns and transliterate)
535 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
536 calls scan_str(). s/// makes yylex() call scan_subst() which calls
537 scan_str(). tr/// and y/// make yylex() call scan_trans() which
538 calls scan_str().
539
540 It skips whitespace before the string starts, and treats the first
541 character as the delimiter. If the delimiter is one of ([{< then
542 the corresponding "close" character )]}> is used as the closing
543 delimiter. It allows quoting of delimiters, and if the string has
544 balanced delimiters ([{<>}]) it allows nesting.
545
546 On success, the SV with the resulting string is put into lex_stuff or,
547 if that is already non-NULL, into lex_repl. The second case occurs only
548 when parsing the RHS of the special constructs s/// and tr/// (y///).
549 For convenience, the terminating delimiter character is stuffed into
550 SvIVX of the SV.
551*/
552
553STATIC char *
554S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
555{
556 SV *sv; /* scalar value: string */
557 char *tmps; /* temp string, used for delimiter matching */
558 register char *s = start; /* current position in the buffer */
559 register char term; /* terminating character */
560 register char *to; /* current position in the sv's data */
561 I32 brackets = 1; /* bracket nesting level */
562 bool has_utf8 = FALSE; /* is there any utf8 content? */
563 I32 termcode; /* terminating char. code */
71511651 564 /* 5.8.7+ uses UTF8_MAXBYTES but also its utf8.h defs _MAXLEN to it so
565 I'm reasonably hopeful this won't destroy anything (mst) */
566 U8 termstr[UTF8_MAXLEN]; /* terminating string */
0ba8c7aa 567 STRLEN termlen; /* length of terminating string */
568 char *last = NULL; /* last position for nesting bracket */
569
570 /* skip space before the delimiter */
571 if (isSPACE(*s))
572 s = skipspace(s);
573
574 /* mark where we are, in case we need to report errors */
575 CLINE;
576
577 /* after skipping whitespace, the next character is the terminator */
578 term = *s;
579 if (!UTF) {
580 termcode = termstr[0] = term;
581 termlen = 1;
582 }
583 else {
584 termcode = utf8_to_uvchr((U8*)s, &termlen);
585 Copy(s, termstr, termlen, U8);
586 if (!UTF8_IS_INVARIANT(term))
587 has_utf8 = TRUE;
588 }
589
590 /* mark where we are */
591 PL_multi_start = CopLINE(PL_curcop);
592 PL_multi_open = term;
593
594 /* find corresponding closing delimiter */
595 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
596 termcode = termstr[0] = term = tmps[5];
597
598 PL_multi_close = term;
599
600 /* create a new SV to hold the contents. 87 is leak category, I'm
601 assuming. 79 is the SV's initial length. What a random number. */
602 sv = NEWSV(87,79);
603 sv_upgrade(sv, SVt_PVIV);
604 SvIV_set(sv, termcode);
605 (void)SvPOK_only(sv); /* validate pointer */
606
607 /* move past delimiter and try to read a complete string */
608 if (keep_delims)
609 sv_catpvn(sv, s, termlen);
610 s += termlen;
611 for (;;) {
612 if (PL_encoding && !UTF) {
613 bool cont = TRUE;
614
615 while (cont) {
616 int offset = s - SvPVX_const(PL_linestr);
617 const bool found = sv_cat_decode(sv, PL_encoding, PL_linestr,
618 &offset, (char*)termstr, termlen);
619 const char *ns = SvPVX_const(PL_linestr) + offset;
620 char *svlast = SvEND(sv) - 1;
621
622 for (; s < ns; s++) {
623 if (*s == '\n' && !PL_rsfp)
624 CopLINE_inc(PL_curcop);
625 }
626 if (!found)
627 goto read_more_line;
628 else {
629 /* handle quoted delimiters */
630 if (SvCUR(sv) > 1 && *(svlast-1) == '\\') {
631 const char *t;
632 for (t = svlast-2; t >= SvPVX_const(sv) && *t == '\\';)
633 t--;
634 if ((svlast-1 - t) % 2) {
635 if (!keep_quoted) {
636 *(svlast-1) = term;
637 *svlast = '\0';
638 SvCUR_set(sv, SvCUR(sv) - 1);
639 }
640 continue;
641 }
642 }
643 if (PL_multi_open == PL_multi_close) {
644 cont = FALSE;
645 }
646 else {
647 const char *t;
648 char *w;
649 if (!last)
650 last = SvPVX(sv);
651 for (t = w = last; t < svlast; w++, t++) {
652 /* At here, all closes are "was quoted" one,
653 so we don't check PL_multi_close. */
654 if (*t == '\\') {
655 if (!keep_quoted && *(t+1) == PL_multi_open)
656 t++;
657 else
658 *w++ = *t++;
659 }
660 else if (*t == PL_multi_open)
661 brackets++;
662
663 *w = *t;
664 }
665 if (w < t) {
666 *w++ = term;
667 *w = '\0';
668 SvCUR_set(sv, w - SvPVX_const(sv));
669 }
670 last = w;
671 if (--brackets <= 0)
672 cont = FALSE;
673 }
674 }
675 }
676 if (!keep_delims) {
677 SvCUR_set(sv, SvCUR(sv) - 1);
678 *SvEND(sv) = '\0';
679 }
680 break;
681 }
682
683 /* extend sv if need be */
684 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
685 /* set 'to' to the next character in the sv's string */
686 to = SvPVX(sv)+SvCUR(sv);
687
688 /* if open delimiter is the close delimiter read unbridle */
689 if (PL_multi_open == PL_multi_close) {
690 for (; s < PL_bufend; s++,to++) {
691 /* embedded newlines increment the current line number */
692 if (*s == '\n' && !PL_rsfp)
693 CopLINE_inc(PL_curcop);
694 /* handle quoted delimiters */
695 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
696 if (!keep_quoted && s[1] == term)
697 s++;
698 /* any other quotes are simply copied straight through */
699 else
700 *to++ = *s++;
701 }
702 /* terminate when run out of buffer (the for() condition), or
703 have found the terminator */
704 else if (*s == term) {
705 if (termlen == 1)
706 break;
707 if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen))
708 break;
709 }
710 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
711 has_utf8 = TRUE;
712 *to = *s;
713 }
714 }
715
716 /* if the terminator isn't the same as the start character (e.g.,
717 matched brackets), we have to allow more in the quoting, and
718 be prepared for nested brackets.
719 */
720 else {
721 /* read until we run out of string, or we find the terminator */
722 for (; s < PL_bufend; s++,to++) {
723 /* embedded newlines increment the line count */
724 if (*s == '\n' && !PL_rsfp)
725 CopLINE_inc(PL_curcop);
726 /* backslashes can escape the open or closing characters */
727 if (*s == '\\' && s+1 < PL_bufend) {
728 if (!keep_quoted &&
729 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
730 s++;
731 else
732 *to++ = *s++;
733 }
734 /* allow nested opens and closes */
735 else if (*s == PL_multi_close && --brackets <= 0)
736 break;
737 else if (*s == PL_multi_open)
738 brackets++;
739 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
740 has_utf8 = TRUE;
741 *to = *s;
742 }
743 }
744 /* terminate the copied string and update the sv's end-of-string */
745 *to = '\0';
746 SvCUR_set(sv, to - SvPVX_const(sv));
747
748 /*
749 * this next chunk reads more into the buffer if we're not done yet
750 */
751
752 if (s < PL_bufend)
753 break; /* handle case where we are done yet :-) */
754
755#ifndef PERL_STRICT_CR
756 if (to - SvPVX_const(sv) >= 2) {
757 if ((to[-2] == '\r' && to[-1] == '\n') ||
758 (to[-2] == '\n' && to[-1] == '\r'))
759 {
760 to[-2] = '\n';
761 to--;
762 SvCUR_set(sv, to - SvPVX_const(sv));
763 }
764 else if (to[-1] == '\r')
765 to[-1] = '\n';
766 }
767 else if (to - SvPVX_const(sv) == 1 && to[-1] == '\r')
768 to[-1] = '\n';
769#endif
770
771 read_more_line:
772 /* if we're out of file, or a read fails, bail and reset the current
773 line marker so we can report where the unterminated string began
774 */
775 if (!PL_rsfp ||
776 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
777 sv_free(sv);
778 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
779 return Nullch;
780 }
781 /* we read a line, so increment our line counter */
782 CopLINE_inc(PL_curcop);
783
784 /* update debugger info */
785 if (PERLDB_LINE && PL_curstash != PL_debstash) {
786 SV *sv = NEWSV(88,0);
787
788 sv_upgrade(sv, SVt_PVMG);
789 sv_setsv(sv,PL_linestr);
790 (void)SvIOK_on(sv);
791 SvIV_set(sv, 0);
792 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
793 }
794
795 /* having changed the buffer, we must update PL_bufend */
796 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
797 PL_last_lop = PL_last_uni = Nullch;
798 }
799
800 /* at this point, we have successfully read the delimited string */
801
802 if (!PL_encoding || UTF) {
803 if (keep_delims)
804 sv_catpvn(sv, s, termlen);
805 s += termlen;
806 }
807 if (has_utf8 || PL_encoding)
808 SvUTF8_on(sv);
809
810 PL_multi_end = CopLINE(PL_curcop);
811
812 /* if we allocated too much space, give some back */
813 if (SvCUR(sv) + 5 < SvLEN(sv)) {
814 SvLEN_set(sv, SvCUR(sv) + 1);
71511651 815/* 5.8.8 uses SvPV_renew, no prior version actually has the damn thing (mst) */
816#ifdef PERL_5_8_8_PLUS
0ba8c7aa 817 SvPV_renew(sv, SvLEN(sv));
71511651 818#else
819 Renew(SvPVX(sv), SvLEN(sv), char);
820#endif
0ba8c7aa 821 }
822
823 /* decide whether this is the first or second quoted string we've read
824 for this op
825 */
826
827 if (PL_lex_stuff)
828 PL_lex_repl = sv;
829 else
830 PL_lex_stuff = sv;
831 return s;
832}
833
834/*
835 * S_force_next
836 * When the lexer realizes it knows the next token (for instance,
837 * it is reordering tokens for the parser) then it can call S_force_next
838 * to know what token to return the next time the lexer is called. Caller
839 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
840 * handles the token correctly.
841 */
842
843STATIC void
844S_force_next(pTHX_ I32 type)
845{
846 PL_nexttype[PL_nexttoke] = type;
847 PL_nexttoke++;
848 if (PL_lex_state != LEX_KNOWNEXT) {
849 PL_lex_defer = PL_lex_state;
850 PL_lex_expect = PL_expect;
851 PL_lex_state = LEX_KNOWNEXT;
852 }
853}
92997e69 854
855#define XFAKEBRACK 128
856
857STATIC char *
858S_scan_ident(pTHX_ register char *s, register const char *send, char *dest, STRLEN destlen, I32 ck_uni)
859{
860 register char *d;
861 register char *e;
862 char *bracket = Nullch;
863 char funny = *s++;
864
865 if (isSPACE(*s))
866 s = skipspace(s);
867 d = dest;
868 e = d + destlen - 3; /* two-character token, ending NUL */
869 if (isDIGIT(*s)) {
870 while (isDIGIT(*s)) {
871 if (d >= e)
872 Perl_croak(aTHX_ ident_too_long);
873 *d++ = *s++;
874 }
875 }
876 else {
877 for (;;) {
878 if (d >= e)
879 Perl_croak(aTHX_ ident_too_long);
880 if (isALNUM(*s)) /* UTF handled below */
881 *d++ = *s++;
882 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
883 *d++ = ':';
884 *d++ = ':';
885 s++;
886 }
887 else if (*s == ':' && s[1] == ':') {
888 *d++ = *s++;
889 *d++ = *s++;
890 }
891 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
892 char *t = s + UTF8SKIP(s);
893 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
894 t += UTF8SKIP(t);
895 if (d + (t - s) > e)
896 Perl_croak(aTHX_ ident_too_long);
897 Copy(s, d, t - s, char);
898 d += t - s;
899 s = t;
900 }
901 else
902 break;
903 }
904 }
905 *d = '\0';
906 d = dest;
907 if (*d) {
908 if (PL_lex_state != LEX_NORMAL)
909 PL_lex_state = LEX_INTERPENDMAYBE;
910 return s;
911 }
912 if (*s == '$' && s[1] &&
913 (isALNUM_lazy_if(s+1,UTF) || s[1] == '$' || s[1] == '{' || strnEQ(s+1,"::",2)) )
914 {
915 return s;
916 }
917 if (*s == '{') {
918 bracket = s;
919 s++;
920 }
d6e7537a 921 /* we always call this with ck_uni == 0 (rafl) */
922 /*
92997e69 923 else if (ck_uni)
924 check_uni();
d6e7537a 925 */
92997e69 926 if (s < send)
927 *d = *s++;
928 d[1] = '\0';
929 if (*d == '^' && *s && isCONTROLVAR(*s)) {
930 *d = toCTRL(*s);
931 s++;
932 }
933 if (bracket) {
934 if (isSPACE(s[-1])) {
935 while (s < send) {
936 const char ch = *s++;
937 if (!SPACE_OR_TAB(ch)) {
938 *d = ch;
939 break;
940 }
941 }
942 }
943 if (isIDFIRST_lazy_if(d,UTF)) {
944 d++;
945 if (UTF) {
946 e = s;
947 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
948 e += UTF8SKIP(e);
949 while (e < send && UTF8_IS_CONTINUED(*e) && is_utf8_mark((U8*)e))
950 e += UTF8SKIP(e);
951 }
952 Copy(s, d, e - s, char);
953 d += e - s;
954 s = e;
955 }
956 else {
957 while ((isALNUM(*s) || *s == ':') && d < e)
958 *d++ = *s++;
959 if (d >= e)
960 Perl_croak(aTHX_ ident_too_long);
961 }
962 *d = '\0';
963 while (s < send && SPACE_OR_TAB(*s)) s++;
964 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
d6e7537a 965 /* we don't want perl to guess what is meant. the keyword
966 * parser decides that later. (rafl)
967 */
968 /*
92997e69 969 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
970 const char *brack = *s == '[' ? "[...]" : "{...}";
971 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
972 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
973 funny, dest, brack, funny, dest, brack);
974 }
d6e7537a 975 */
92997e69 976 bracket++;
977 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
978 return s;
979 }
980 }
981 /* Handle extended ${^Foo} variables
982 * 1999-02-27 mjd-perl-patch@plover.com */
983 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
984 && isALNUM(*s))
985 {
986 d++;
987 while (isALNUM(*s) && d < e) {
988 *d++ = *s++;
989 }
990 if (d >= e)
991 Perl_croak(aTHX_ ident_too_long);
992 *d = '\0';
993 }
994 if (*s == '}') {
995 s++;
996 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets) {
997 PL_lex_state = LEX_INTERPEND;
998 PL_expect = XREF;
999 }
1000 if (funny == '#')
1001 funny = '@';
d6e7537a 1002 /* we don't want perl to guess what is meant. the keyword
1003 * parser decides that later. (rafl)
1004 */
1005 /*
92997e69 1006 if (PL_lex_state == LEX_NORMAL) {
1007 if (ckWARN(WARN_AMBIGUOUS) &&
1008 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
1009 {
1010 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
1011 "Ambiguous use of %c{%s} resolved to %c%s",
1012 funny, dest, funny, dest);
1013 }
1014 }
d6e7537a 1015 */
92997e69 1016 }
1017 else {
1018 s = bracket; /* let the parser handle it */
1019 *dest = '\0';
1020 }
1021 }
d6e7537a 1022 /* don't intuit. we really just want the string. (rafl) */
1023 /*
92997e69 1024 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
1025 PL_lex_state = LEX_INTERPEND;
d6e7537a 1026 */
92997e69 1027 return s;
1028}