projects / p5sagit/Devel-Declare.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | history | HEAD
Don't define NEWSV. handy.h does that for us.
[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)
27
28STATIC void S_incline(pTHX_ char *s);
29STATIC char* S_skipspace(pTHX_ char *s);
30STATIC char * S_filter_gets(pTHX_ SV *sv, PerlIO *fp, STRLEN append);
31STATIC char* S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims);
32STATIC char* S_scan_word(pTHX_ char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp);
33
e807ee50 34#define DPTR2FPTR(t,p) ((t)PTR2nat(p)) /* data pointer to function pointer */
35#define FPTR2DPTR(t,p) ((t)PTR2nat(p)) /* function pointer to data pointer */
c630715a 36#define PTR2nat(p) (PTRV)(p) /* pointer to integer of PTRSIZE */
e807ee50 37#define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
14ab3472 38
39/* conditionalise these two because as of 5.9.5 we already get them from
40 the headers (mst) */
41#ifndef Newx
e807ee50 42#define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)safemalloc((MEM_SIZE)((n)*sizeof(t)))))
14ab3472 43#endif
44#ifndef SvPVX_const
0ba8c7aa 45#define SvPVX_const(sv) ((const char*) (0 + SvPVX(sv)))
14ab3472 46#endif
47
0ba8c7aa 48#define SvPV_renew(sv,n) \
49 STMT_START { SvLEN_set(sv, n); \
50 SvPV_set((sv), (MEM_WRAP_CHECK_(n,char) \
51 (char*)saferealloc((Malloc_t)SvPVX(sv), \
52 (MEM_SIZE)((n))))); \
53 } STMT_END
e807ee50 54
c630715a 55/* On MacOS, respect nonbreaking spaces */
56#ifdef MACOS_TRADITIONAL
57#define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
58#else
59#define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
60#endif
e807ee50 61
0ba8c7aa 62#define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
63
e807ee50 64#define LEX_NORMAL 10 /* normal code (ie not within "...") */
65#define LEX_INTERPNORMAL 9 /* code within a string, eg "$foo[$x+1]" */
66#define LEX_INTERPCASEMOD 8 /* expecting a \U, \Q or \E etc */
67#define LEX_INTERPPUSH 7 /* starting a new sublex parse level */
68#define LEX_INTERPSTART 6 /* expecting the start of a $var */
69
70 /* at end of code, eg "$x" followed by: */
71#define LEX_INTERPEND 5 /* ... eg not one of [, { or -> */
72#define LEX_INTERPENDMAYBE 4 /* ... eg one of [, { or -> */
73
74#define LEX_INTERPCONCAT 3 /* expecting anything, eg at start of
75 string or after \E, $foo, etc */
76#define LEX_INTERPCONST 2 /* NOT USED */
77#define LEX_FORMLINE 1 /* expecting a format line */
78#define LEX_KNOWNEXT 0 /* next token known; just return it */
79
14ab3472 80/* and these two are my own madness (mst) */
71511651 81
82#if PERL_REVISION == 5 && PERL_VERSION == 8 && PERL_SUBVERSION >= 8
83#define PERL_5_8_8_PLUS
84#endif
85
5e9a4dee 86#ifdef PL_parser
14ab3472 87/* 5.9+ moves a bunch of things to a PL_parser struct so we need to
88 declare the backcompat macros for things to still work (mst) */
89
90/* XXX temporary backwards compatibility */
91#define PL_lex_brackets (PL_parser->lex_brackets)
92#define PL_lex_brackstack (PL_parser->lex_brackstack)
93#define PL_lex_casemods (PL_parser->lex_casemods)
94#define PL_lex_casestack (PL_parser->lex_casestack)
95#define PL_lex_defer (PL_parser->lex_defer)
96#define PL_lex_dojoin (PL_parser->lex_dojoin)
97#define PL_lex_expect (PL_parser->lex_expect)
98#define PL_lex_formbrack (PL_parser->lex_formbrack)
99#define PL_lex_inpat (PL_parser->lex_inpat)
100#define PL_lex_inwhat (PL_parser->lex_inwhat)
101#define PL_lex_op (PL_parser->lex_op)
102#define PL_lex_repl (PL_parser->lex_repl)
103#define PL_lex_starts (PL_parser->lex_starts)
104#define PL_lex_stuff (PL_parser->lex_stuff)
105#define PL_multi_start (PL_parser->multi_start)
106#define PL_multi_open (PL_parser->multi_open)
107#define PL_multi_close (PL_parser->multi_close)
108#define PL_pending_ident (PL_parser->pending_ident)
109#define PL_preambled (PL_parser->preambled)
110#define PL_sublex_info (PL_parser->sublex_info)
111#define PL_linestr (PL_parser->linestr)
112#define PL_sublex_info (PL_parser->sublex_info)
113#define PL_linestr (PL_parser->linestr)
114#define PL_expect (PL_parser->expect)
115#define PL_copline (PL_parser->copline)
116#define PL_bufptr (PL_parser->bufptr)
117#define PL_oldbufptr (PL_parser->oldbufptr)
118#define PL_oldoldbufptr (PL_parser->oldoldbufptr)
119#define PL_linestart (PL_parser->linestart)
120#define PL_bufend (PL_parser->bufend)
121#define PL_last_uni (PL_parser->last_uni)
122#define PL_last_lop (PL_parser->last_lop)
123#define PL_last_lop_op (PL_parser->last_lop_op)
124#define PL_lex_state (PL_parser->lex_state)
125#define PL_rsfp (PL_parser->rsfp)
126#define PL_rsfp_filters (PL_parser->rsfp_filters)
127#define PL_in_my (PL_parser->in_my)
128#define PL_in_my_stash (PL_parser->in_my_stash)
129#define PL_tokenbuf (PL_parser->tokenbuf)
130#define PL_multi_end (PL_parser->multi_end)
131#define PL_error_count (PL_parser->error_count)
132/* these three are from the non-PERL_MAD path but I don't -think- I need
133 the PERL_MAD stuff since my code isn't really populating things (mst) */
134# define PL_nexttoke (PL_parser->nexttoke)
135# define PL_nexttype (PL_parser->nexttype)
136# define PL_nextval (PL_parser->nextval)
137/* end of backcompat macros form 5.9 toke.c (mst) */
78160085 138#endif
14ab3472 139
140/* when ccflags include -DDEBUGGING we need this for earlier 5.8 perls */
141#ifndef SvPV_nolen_const
142#define SvPV_nolen_const SvPV_nolen
143#endif
144
71511651 145/* and now we're back to the toke.c stuff again (mst) */
146
e807ee50 147static const char ident_too_long[] =
148 "Identifier too long";
149static const char c_without_g[] =
150 "Use of /c modifier is meaningless without /g";
151static const char c_in_subst[] =
152 "Use of /c modifier is meaningless in s///";
153
154#ifdef USE_UTF8_SCRIPTS
155# define UTF (!IN_BYTES)
156#else
157# define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || (PL_hints & HINT_UTF8))
158#endif
159
160/* Invoke the idxth filter function for the current rsfp. */
161/* maxlen 0 = read one text line */
162I32
163Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
164{
165 filter_t funcp;
166 SV *datasv = NULL;
167
168 if (!PL_rsfp_filters)
169 return -1;
170 if (idx > AvFILLp(PL_rsfp_filters)) { /* Any more filters? */
171 /* Provide a default input filter to make life easy. */
172 /* Note that we append to the line. This is handy. */
173 DEBUG_P(PerlIO_printf(Perl_debug_log,
174 "filter_read %d: from rsfp\n", idx));
175 if (maxlen) {
176 /* Want a block */
177 int len ;
178 const int old_len = SvCUR(buf_sv);
179
180 /* ensure buf_sv is large enough */
181 SvGROW(buf_sv, (STRLEN)(old_len + maxlen)) ;
182 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
183 if (PerlIO_error(PL_rsfp))
184 return -1; /* error */
185 else
186 return 0 ; /* end of file */
187 }
188 SvCUR_set(buf_sv, old_len + len) ;
189 } else {
190 /* Want a line */
191 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
192 if (PerlIO_error(PL_rsfp))
193 return -1; /* error */
194 else
195 return 0 ; /* end of file */
196 }
197 }
198 return SvCUR(buf_sv);
199 }
200 /* Skip this filter slot if filter has been deleted */
201 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef) {
202 DEBUG_P(PerlIO_printf(Perl_debug_log,
203 "filter_read %d: skipped (filter deleted)\n",
204 idx));
205 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
206 }
207 /* Get function pointer hidden within datasv */
208 funcp = DPTR2FPTR(filter_t, IoANY(datasv));
209 DEBUG_P(PerlIO_printf(Perl_debug_log,
210 "filter_read %d: via function %p (%s)\n",
211 idx, datasv, SvPV_nolen_const(datasv)));
212 /* Call function. The function is expected to */
213 /* call "FILTER_READ(idx+1, buf_sv)" first. */
214 /* Return: <0:error, =0:eof, >0:not eof */
215 return (*funcp)(aTHX_ idx, buf_sv, maxlen);
216}
217
218STATIC char *
219S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
220{
221#ifdef PERL_CR_FILTER
222 if (!PL_rsfp_filters) {
223 filter_add(S_cr_textfilter,NULL);
224 }
225#endif
226 if (PL_rsfp_filters) {
227 if (!append)
228 SvCUR_set(sv, 0); /* start with empty line */
229 if (FILTER_READ(0, sv, 0) > 0)
230 return ( SvPVX(sv) ) ;
231 else
232 return Nullch ;
233 }
234 else
235 return (sv_gets(sv, fp, append));
236}
237
238/*
239 * S_skipspace
240 * Called to gobble the appropriate amount and type of whitespace.
241 * Skips comments as well.
242 */
243
244STATIC char *
245S_skipspace(pTHX_ register char *s)
246{
247 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
248 while (s < PL_bufend && SPACE_OR_TAB(*s))
249 s++;
250 return s;
251 }
252 for (;;) {
253 STRLEN prevlen;
254 SSize_t oldprevlen, oldoldprevlen;
255 SSize_t oldloplen = 0, oldunilen = 0;
256 while (s < PL_bufend && isSPACE(*s)) {
257 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
258 incline(s);
259 }
260
261 /* comment */
262 if (s < PL_bufend && *s == '#') {
263 while (s < PL_bufend && *s != '\n')
264 s++;
265 if (s < PL_bufend) {
266 s++;
267 if (PL_in_eval && !PL_rsfp) {
268 incline(s);
269 continue;
270 }
271 }
272 }
273
274 /* only continue to recharge the buffer if we're at the end
275 * of the buffer, we're not reading from a source filter, and
276 * we're in normal lexing mode
277 */
278 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
279 PL_lex_state == LEX_FORMLINE)
280 return s;
281
282 /* try to recharge the buffer */
283 if ((s = filter_gets(PL_linestr, PL_rsfp,
284 (prevlen = SvCUR(PL_linestr)))) == Nullch)
285 {
286 /* end of file. Add on the -p or -n magic */
287 if (PL_minus_p) {
288 sv_setpv(PL_linestr,
289 ";}continue{print or die qq(-p destination: $!\\n);}");
290 PL_minus_n = PL_minus_p = 0;
291 }
292 else if (PL_minus_n) {
293 sv_setpvn(PL_linestr, ";}", 2);
294 PL_minus_n = 0;
295 }
296 else
297 sv_setpvn(PL_linestr,";", 1);
298
299 /* reset variables for next time we lex */
300 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
301 = SvPVX(PL_linestr);
302 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
303 PL_last_lop = PL_last_uni = Nullch;
304
305 /* Close the filehandle. Could be from -P preprocessor,
306 * STDIN, or a regular file. If we were reading code from
307 * STDIN (because the commandline held no -e or filename)
308 * then we don't close it, we reset it so the code can
309 * read from STDIN too.
310 */
311
312 if (PL_preprocess && !PL_in_eval)
313 (void)PerlProc_pclose(PL_rsfp);
314 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
315 PerlIO_clearerr(PL_rsfp);
316 else
317 (void)PerlIO_close(PL_rsfp);
318 PL_rsfp = Nullfp;
319 return s;
320 }
321
322 /* not at end of file, so we only read another line */
323 /* make corresponding updates to old pointers, for yyerror() */
324 oldprevlen = PL_oldbufptr - PL_bufend;
325 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
326 if (PL_last_uni)
327 oldunilen = PL_last_uni - PL_bufend;
328 if (PL_last_lop)
329 oldloplen = PL_last_lop - PL_bufend;
330 PL_linestart = PL_bufptr = s + prevlen;
331 PL_bufend = s + SvCUR(PL_linestr);
332 s = PL_bufptr;
333 PL_oldbufptr = s + oldprevlen;
334 PL_oldoldbufptr = s + oldoldprevlen;
335 if (PL_last_uni)
336 PL_last_uni = s + oldunilen;
337 if (PL_last_lop)
338 PL_last_lop = s + oldloplen;
339 incline(s);
340
341 /* debugger active and we're not compiling the debugger code,
342 * so store the line into the debugger's array of lines
343 */
344 if (PERLDB_LINE && PL_curstash != PL_debstash) {
345 SV * const sv = NEWSV(85,0);
346
347 sv_upgrade(sv, SVt_PVMG);
348 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
349 (void)SvIOK_on(sv);
350 SvIV_set(sv, 0);
351 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
352 }
353 }
354}
355
356STATIC char *
357S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
358{
359 register char *d = dest;
360 register char * const e = d + destlen - 3; /* two-character token, ending NUL */
361 for (;;) {
362 if (d >= e)
363 Perl_croak(aTHX_ ident_too_long);
364 if (isALNUM(*s)) /* UTF handled below */
365 *d++ = *s++;
366 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
367 *d++ = ':';
368 *d++ = ':';
369 s++;
370 }
371 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
372 *d++ = *s++;
373 *d++ = *s++;
374 }
375 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
376 char *t = s + UTF8SKIP(s);
377 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
378 t += UTF8SKIP(t);
379 if (d + (t - s) > e)
380 Perl_croak(aTHX_ ident_too_long);
381 Copy(s, d, t - s, char);
382 d += t - s;
383 s = t;
384 }
385 else {
386 *d = '\0';
387 *slp = d - dest;
388 return s;
389 }
390 }
391}
392
393/*
394 * S_incline
395 * This subroutine has nothing to do with tilting, whether at windmills
396 * or pinball tables. Its name is short for "increment line". It
397 * increments the current line number in CopLINE(PL_curcop) and checks
398 * to see whether the line starts with a comment of the form
399 * # line 500 "foo.pm"
400 * If so, it sets the current line number and file to the values in the comment.
401 */
402
403STATIC void
78160085 404S_incline(pTHX_ char *s)
e807ee50 405{
406 char *t;
78160085 407 char *n;
408 char *e;
e807ee50 409 char ch;
410
411 CopLINE_inc(PL_curcop);
412 if (*s++ != '#')
413 return;
414 while (SPACE_OR_TAB(*s)) s++;
415 if (strnEQ(s, "line", 4))
416 s += 4;
417 else
418 return;
419 if (SPACE_OR_TAB(*s))
420 s++;
421 else
422 return;
423 while (SPACE_OR_TAB(*s)) s++;
424 if (!isDIGIT(*s))
425 return;
426 n = s;
427 while (isDIGIT(*s))
428 s++;
429 while (SPACE_OR_TAB(*s))
430 s++;
431 if (*s == '"' && (t = strchr(s+1, '"'))) {
432 s++;
433 e = t + 1;
434 }
435 else {
78160085 436 for (t = s; !isSPACE(*t); t++) ;
e807ee50 437 e = t;
438 }
439 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
440 e++;
441 if (*e != '\n' && *e != '\0')
442 return; /* false alarm */
443
444 ch = *t;
445 *t = '\0';
446 if (t - s > 0) {
71511651 447/* this chunk was added to S_incline during 5.8.8. I don't know why but I don't
448 honestly care since I probably want to be bug-compatible anyway (mst) */
449
450/* ... my kingdom for a perl parser in perl ... (mst) */
451
452#ifdef PERL_5_8_8_PLUS
e807ee50 453#ifndef USE_ITHREADS
454 const char *cf = CopFILE(PL_curcop);
455 if (cf && strlen(cf) > 7 && strnEQ(cf, "(eval ", 6)) {
456 /* must copy *{"::_<(eval N)[oldfilename:L]"}
457 * to *{"::_<newfilename"} */
458 char smallbuf[256], smallbuf2[256];
459 char *tmpbuf, *tmpbuf2;
460 GV **gvp, *gv2;
461 STRLEN tmplen = strlen(cf);
462 STRLEN tmplen2 = strlen(s);
463 if (tmplen + 3 < sizeof smallbuf)
464 tmpbuf = smallbuf;
465 else
466 Newx(tmpbuf, tmplen + 3, char);
467 if (tmplen2 + 3 < sizeof smallbuf2)
468 tmpbuf2 = smallbuf2;
469 else
470 Newx(tmpbuf2, tmplen2 + 3, char);
471 tmpbuf[0] = tmpbuf2[0] = '_';
472 tmpbuf[1] = tmpbuf2[1] = '<';
473 memcpy(tmpbuf + 2, cf, ++tmplen);
474 memcpy(tmpbuf2 + 2, s, ++tmplen2);
475 ++tmplen; ++tmplen2;
476 gvp = (GV**)hv_fetch(PL_defstash, tmpbuf, tmplen, FALSE);
477 if (gvp) {
478 gv2 = *(GV**)hv_fetch(PL_defstash, tmpbuf2, tmplen2, TRUE);
479 if (!isGV(gv2))
480 gv_init(gv2, PL_defstash, tmpbuf2, tmplen2, FALSE);
481 /* adjust ${"::_<newfilename"} to store the new file name */
482 GvSV(gv2) = newSVpvn(tmpbuf2 + 2, tmplen2 - 2);
483 GvHV(gv2) = (HV*)SvREFCNT_inc(GvHV(*gvp));
484 GvAV(gv2) = (AV*)SvREFCNT_inc(GvAV(*gvp));
485 }
486 if (tmpbuf != smallbuf) Safefree(tmpbuf);
487 if (tmpbuf2 != smallbuf2) Safefree(tmpbuf2);
488 }
489#endif
71511651 490#endif
491/* second endif closes out the "are we 5.8.(8+)" conditional */
e807ee50 492 CopFILE_free(PL_curcop);
493 CopFILE_set(PL_curcop, s);
494 }
495 *t = ch;
496 CopLINE_set(PL_curcop, atoi(n)-1);
497}
0ba8c7aa 498
499/* scan_str
500 takes: start position in buffer
501 keep_quoted preserve \ on the embedded delimiter(s)
502 keep_delims preserve the delimiters around the string
503 returns: position to continue reading from buffer
504 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
505 updates the read buffer.
506
507 This subroutine pulls a string out of the input. It is called for:
508 q single quotes q(literal text)
509 ' single quotes 'literal text'
510 qq double quotes qq(interpolate $here please)
511 " double quotes "interpolate $here please"
512 qx backticks qx(/bin/ls -l)
513 ` backticks `/bin/ls -l`
514 qw quote words @EXPORT_OK = qw( func() $spam )
515 m// regexp match m/this/
516 s/// regexp substitute s/this/that/
517 tr/// string transliterate tr/this/that/
518 y/// string transliterate y/this/that/
519 ($*@) sub prototypes sub foo ($)
520 (stuff) sub attr parameters sub foo : attr(stuff)
521 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
522
523 In most of these cases (all but <>, patterns and transliterate)
524 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
525 calls scan_str(). s/// makes yylex() call scan_subst() which calls
526 scan_str(). tr/// and y/// make yylex() call scan_trans() which
527 calls scan_str().
528
529 It skips whitespace before the string starts, and treats the first
530 character as the delimiter. If the delimiter is one of ([{< then
531 the corresponding "close" character )]}> is used as the closing
532 delimiter. It allows quoting of delimiters, and if the string has
533 balanced delimiters ([{<>}]) it allows nesting.
534
535 On success, the SV with the resulting string is put into lex_stuff or,
536 if that is already non-NULL, into lex_repl. The second case occurs only
537 when parsing the RHS of the special constructs s/// and tr/// (y///).
538 For convenience, the terminating delimiter character is stuffed into
539 SvIVX of the SV.
540*/
541
542STATIC char *
543S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
544{
545 SV *sv; /* scalar value: string */
546 char *tmps; /* temp string, used for delimiter matching */
547 register char *s = start; /* current position in the buffer */
548 register char term; /* terminating character */
549 register char *to; /* current position in the sv's data */
550 I32 brackets = 1; /* bracket nesting level */
551 bool has_utf8 = FALSE; /* is there any utf8 content? */
552 I32 termcode; /* terminating char. code */
71511651 553 /* 5.8.7+ uses UTF8_MAXBYTES but also its utf8.h defs _MAXLEN to it so
554 I'm reasonably hopeful this won't destroy anything (mst) */
555 U8 termstr[UTF8_MAXLEN]; /* terminating string */
0ba8c7aa 556 STRLEN termlen; /* length of terminating string */
557 char *last = NULL; /* last position for nesting bracket */
558
559 /* skip space before the delimiter */
560 if (isSPACE(*s))
561 s = skipspace(s);
562
563 /* mark where we are, in case we need to report errors */
564 CLINE;
565
566 /* after skipping whitespace, the next character is the terminator */
567 term = *s;
568 if (!UTF) {
569 termcode = termstr[0] = term;
570 termlen = 1;
571 }
572 else {
573 termcode = utf8_to_uvchr((U8*)s, &termlen);
574 Copy(s, termstr, termlen, U8);
575 if (!UTF8_IS_INVARIANT(term))
576 has_utf8 = TRUE;
577 }
578
579 /* mark where we are */
580 PL_multi_start = CopLINE(PL_curcop);
581 PL_multi_open = term;
582
583 /* find corresponding closing delimiter */
584 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
585 termcode = termstr[0] = term = tmps[5];
586
587 PL_multi_close = term;
588
589 /* create a new SV to hold the contents. 87 is leak category, I'm
590 assuming. 79 is the SV's initial length. What a random number. */
591 sv = NEWSV(87,79);
592 sv_upgrade(sv, SVt_PVIV);
593 SvIV_set(sv, termcode);
594 (void)SvPOK_only(sv); /* validate pointer */
595
596 /* move past delimiter and try to read a complete string */
597 if (keep_delims)
598 sv_catpvn(sv, s, termlen);
599 s += termlen;
600 for (;;) {
601 if (PL_encoding && !UTF) {
602 bool cont = TRUE;
603
604 while (cont) {
605 int offset = s - SvPVX_const(PL_linestr);
606 const bool found = sv_cat_decode(sv, PL_encoding, PL_linestr,
607 &offset, (char*)termstr, termlen);
608 const char *ns = SvPVX_const(PL_linestr) + offset;
609 char *svlast = SvEND(sv) - 1;
610
611 for (; s < ns; s++) {
612 if (*s == '\n' && !PL_rsfp)
613 CopLINE_inc(PL_curcop);
614 }
615 if (!found)
616 goto read_more_line;
617 else {
618 /* handle quoted delimiters */
619 if (SvCUR(sv) > 1 && *(svlast-1) == '\\') {
620 const char *t;
621 for (t = svlast-2; t >= SvPVX_const(sv) && *t == '\\';)
622 t--;
623 if ((svlast-1 - t) % 2) {
624 if (!keep_quoted) {
625 *(svlast-1) = term;
626 *svlast = '\0';
627 SvCUR_set(sv, SvCUR(sv) - 1);
628 }
629 continue;
630 }
631 }
632 if (PL_multi_open == PL_multi_close) {
633 cont = FALSE;
634 }
635 else {
636 const char *t;
637 char *w;
638 if (!last)
639 last = SvPVX(sv);
640 for (t = w = last; t < svlast; w++, t++) {
641 /* At here, all closes are "was quoted" one,
642 so we don't check PL_multi_close. */
643 if (*t == '\\') {
644 if (!keep_quoted && *(t+1) == PL_multi_open)
645 t++;
646 else
647 *w++ = *t++;
648 }
649 else if (*t == PL_multi_open)
650 brackets++;
651
652 *w = *t;
653 }
654 if (w < t) {
655 *w++ = term;
656 *w = '\0';
657 SvCUR_set(sv, w - SvPVX_const(sv));
658 }
659 last = w;
660 if (--brackets <= 0)
661 cont = FALSE;
662 }
663 }
664 }
665 if (!keep_delims) {
666 SvCUR_set(sv, SvCUR(sv) - 1);
667 *SvEND(sv) = '\0';
668 }
669 break;
670 }
671
672 /* extend sv if need be */
673 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
674 /* set 'to' to the next character in the sv's string */
675 to = SvPVX(sv)+SvCUR(sv);
676
677 /* if open delimiter is the close delimiter read unbridle */
678 if (PL_multi_open == PL_multi_close) {
679 for (; s < PL_bufend; s++,to++) {
680 /* embedded newlines increment the current line number */
681 if (*s == '\n' && !PL_rsfp)
682 CopLINE_inc(PL_curcop);
683 /* handle quoted delimiters */
684 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
685 if (!keep_quoted && s[1] == term)
686 s++;
687 /* any other quotes are simply copied straight through */
688 else
689 *to++ = *s++;
690 }
691 /* terminate when run out of buffer (the for() condition), or
692 have found the terminator */
693 else if (*s == term) {
694 if (termlen == 1)
695 break;
696 if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen))
697 break;
698 }
699 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
700 has_utf8 = TRUE;
701 *to = *s;
702 }
703 }
704
705 /* if the terminator isn't the same as the start character (e.g.,
706 matched brackets), we have to allow more in the quoting, and
707 be prepared for nested brackets.
708 */
709 else {
710 /* read until we run out of string, or we find the terminator */
711 for (; s < PL_bufend; s++,to++) {
712 /* embedded newlines increment the line count */
713 if (*s == '\n' && !PL_rsfp)
714 CopLINE_inc(PL_curcop);
715 /* backslashes can escape the open or closing characters */
716 if (*s == '\\' && s+1 < PL_bufend) {
717 if (!keep_quoted &&
718 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
719 s++;
720 else
721 *to++ = *s++;
722 }
723 /* allow nested opens and closes */
724 else if (*s == PL_multi_close && --brackets <= 0)
725 break;
726 else if (*s == PL_multi_open)
727 brackets++;
728 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
729 has_utf8 = TRUE;
730 *to = *s;
731 }
732 }
733 /* terminate the copied string and update the sv's end-of-string */
734 *to = '\0';
735 SvCUR_set(sv, to - SvPVX_const(sv));
736
737 /*
738 * this next chunk reads more into the buffer if we're not done yet
739 */
740
741 if (s < PL_bufend)
742 break; /* handle case where we are done yet :-) */
743
744#ifndef PERL_STRICT_CR
745 if (to - SvPVX_const(sv) >= 2) {
746 if ((to[-2] == '\r' && to[-1] == '\n') ||
747 (to[-2] == '\n' && to[-1] == '\r'))
748 {
749 to[-2] = '\n';
750 to--;
751 SvCUR_set(sv, to - SvPVX_const(sv));
752 }
753 else if (to[-1] == '\r')
754 to[-1] = '\n';
755 }
756 else if (to - SvPVX_const(sv) == 1 && to[-1] == '\r')
757 to[-1] = '\n';
758#endif
759
760 read_more_line:
761 /* if we're out of file, or a read fails, bail and reset the current
762 line marker so we can report where the unterminated string began
763 */
764 if (!PL_rsfp ||
765 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
766 sv_free(sv);
767 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
768 return Nullch;
769 }
770 /* we read a line, so increment our line counter */
771 CopLINE_inc(PL_curcop);
772
773 /* update debugger info */
774 if (PERLDB_LINE && PL_curstash != PL_debstash) {
775 SV *sv = NEWSV(88,0);
776
777 sv_upgrade(sv, SVt_PVMG);
778 sv_setsv(sv,PL_linestr);
779 (void)SvIOK_on(sv);
780 SvIV_set(sv, 0);
781 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
782 }
783
784 /* having changed the buffer, we must update PL_bufend */
785 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
786 PL_last_lop = PL_last_uni = Nullch;
787 }
788
789 /* at this point, we have successfully read the delimited string */
790
791 if (!PL_encoding || UTF) {
792 if (keep_delims)
793 sv_catpvn(sv, s, termlen);
794 s += termlen;
795 }
796 if (has_utf8 || PL_encoding)
797 SvUTF8_on(sv);
798
799 PL_multi_end = CopLINE(PL_curcop);
800
801 /* if we allocated too much space, give some back */
802 if (SvCUR(sv) + 5 < SvLEN(sv)) {
803 SvLEN_set(sv, SvCUR(sv) + 1);
71511651 804/* 5.8.8 uses SvPV_renew, no prior version actually has the damn thing (mst) */
805#ifdef PERL_5_8_8_PLUS
0ba8c7aa 806 SvPV_renew(sv, SvLEN(sv));
71511651 807#else
808 Renew(SvPVX(sv), SvLEN(sv), char);
809#endif
0ba8c7aa 810 }
811
812 /* decide whether this is the first or second quoted string we've read
813 for this op
814 */
815
816 if (PL_lex_stuff)
817 PL_lex_repl = sv;
818 else
819 PL_lex_stuff = sv;
820 return s;
821}
822
823/*
824 * S_force_next
825 * When the lexer realizes it knows the next token (for instance,
826 * it is reordering tokens for the parser) then it can call S_force_next
827 * to know what token to return the next time the lexer is called. Caller
828 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
829 * handles the token correctly.
830 */
831
832STATIC void
833S_force_next(pTHX_ I32 type)
834{
835 PL_nexttype[PL_nexttoke] = type;
836 PL_nexttoke++;
837 if (PL_lex_state != LEX_KNOWNEXT) {
838 PL_lex_defer = PL_lex_state;
839 PL_lex_expect = PL_expect;
840 PL_lex_state = LEX_KNOWNEXT;
841 }
842}
Unnamed repository; edit this file to name it for gitweb.