projects / p5sagit/Devel-Declare.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | history | HEAD
Update for latest B::Hooks::OP::Check API.
[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/* we also need this because we define PERL_CORE so handy.h doesn't provide
140 it for us (mst) */
552a9b31 141#ifndef NEWSV
14ab3472 142#define NEWSV(x,len) newSV(len)
143#endif
144
145/* when ccflags include -DDEBUGGING we need this for earlier 5.8 perls */
146#ifndef SvPV_nolen_const
147#define SvPV_nolen_const SvPV_nolen
148#endif
149
71511651 150/* and now we're back to the toke.c stuff again (mst) */
151
e807ee50 152static const char ident_too_long[] =
153 "Identifier too long";
154static const char c_without_g[] =
155 "Use of /c modifier is meaningless without /g";
156static const char c_in_subst[] =
157 "Use of /c modifier is meaningless in s///";
158
159#ifdef USE_UTF8_SCRIPTS
160# define UTF (!IN_BYTES)
161#else
162# define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || (PL_hints & HINT_UTF8))
163#endif
164
165/* Invoke the idxth filter function for the current rsfp. */
166/* maxlen 0 = read one text line */
167I32
168Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
169{
170 filter_t funcp;
171 SV *datasv = NULL;
172
173 if (!PL_rsfp_filters)
174 return -1;
175 if (idx > AvFILLp(PL_rsfp_filters)) { /* Any more filters? */
176 /* Provide a default input filter to make life easy. */
177 /* Note that we append to the line. This is handy. */
178 DEBUG_P(PerlIO_printf(Perl_debug_log,
179 "filter_read %d: from rsfp\n", idx));
180 if (maxlen) {
181 /* Want a block */
182 int len ;
183 const int old_len = SvCUR(buf_sv);
184
185 /* ensure buf_sv is large enough */
186 SvGROW(buf_sv, (STRLEN)(old_len + maxlen)) ;
187 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
188 if (PerlIO_error(PL_rsfp))
189 return -1; /* error */
190 else
191 return 0 ; /* end of file */
192 }
193 SvCUR_set(buf_sv, old_len + len) ;
194 } else {
195 /* Want a line */
196 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
197 if (PerlIO_error(PL_rsfp))
198 return -1; /* error */
199 else
200 return 0 ; /* end of file */
201 }
202 }
203 return SvCUR(buf_sv);
204 }
205 /* Skip this filter slot if filter has been deleted */
206 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef) {
207 DEBUG_P(PerlIO_printf(Perl_debug_log,
208 "filter_read %d: skipped (filter deleted)\n",
209 idx));
210 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
211 }
212 /* Get function pointer hidden within datasv */
213 funcp = DPTR2FPTR(filter_t, IoANY(datasv));
214 DEBUG_P(PerlIO_printf(Perl_debug_log,
215 "filter_read %d: via function %p (%s)\n",
216 idx, datasv, SvPV_nolen_const(datasv)));
217 /* Call function. The function is expected to */
218 /* call "FILTER_READ(idx+1, buf_sv)" first. */
219 /* Return: <0:error, =0:eof, >0:not eof */
220 return (*funcp)(aTHX_ idx, buf_sv, maxlen);
221}
222
223STATIC char *
224S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
225{
226#ifdef PERL_CR_FILTER
227 if (!PL_rsfp_filters) {
228 filter_add(S_cr_textfilter,NULL);
229 }
230#endif
231 if (PL_rsfp_filters) {
232 if (!append)
233 SvCUR_set(sv, 0); /* start with empty line */
234 if (FILTER_READ(0, sv, 0) > 0)
235 return ( SvPVX(sv) ) ;
236 else
237 return Nullch ;
238 }
239 else
240 return (sv_gets(sv, fp, append));
241}
242
243/*
244 * S_skipspace
245 * Called to gobble the appropriate amount and type of whitespace.
246 * Skips comments as well.
247 */
248
249STATIC char *
250S_skipspace(pTHX_ register char *s)
251{
252 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
253 while (s < PL_bufend && SPACE_OR_TAB(*s))
254 s++;
255 return s;
256 }
257 for (;;) {
258 STRLEN prevlen;
259 SSize_t oldprevlen, oldoldprevlen;
260 SSize_t oldloplen = 0, oldunilen = 0;
261 while (s < PL_bufend && isSPACE(*s)) {
262 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
263 incline(s);
264 }
265
266 /* comment */
267 if (s < PL_bufend && *s == '#') {
268 while (s < PL_bufend && *s != '\n')
269 s++;
270 if (s < PL_bufend) {
271 s++;
272 if (PL_in_eval && !PL_rsfp) {
273 incline(s);
274 continue;
275 }
276 }
277 }
278
279 /* only continue to recharge the buffer if we're at the end
280 * of the buffer, we're not reading from a source filter, and
281 * we're in normal lexing mode
282 */
283 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
284 PL_lex_state == LEX_FORMLINE)
285 return s;
286
287 /* try to recharge the buffer */
288 if ((s = filter_gets(PL_linestr, PL_rsfp,
289 (prevlen = SvCUR(PL_linestr)))) == Nullch)
290 {
291 /* end of file. Add on the -p or -n magic */
292 if (PL_minus_p) {
293 sv_setpv(PL_linestr,
294 ";}continue{print or die qq(-p destination: $!\\n);}");
295 PL_minus_n = PL_minus_p = 0;
296 }
297 else if (PL_minus_n) {
298 sv_setpvn(PL_linestr, ";}", 2);
299 PL_minus_n = 0;
300 }
301 else
302 sv_setpvn(PL_linestr,";", 1);
303
304 /* reset variables for next time we lex */
305 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
306 = SvPVX(PL_linestr);
307 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
308 PL_last_lop = PL_last_uni = Nullch;
309
310 /* Close the filehandle. Could be from -P preprocessor,
311 * STDIN, or a regular file. If we were reading code from
312 * STDIN (because the commandline held no -e or filename)
313 * then we don't close it, we reset it so the code can
314 * read from STDIN too.
315 */
316
317 if (PL_preprocess && !PL_in_eval)
318 (void)PerlProc_pclose(PL_rsfp);
319 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
320 PerlIO_clearerr(PL_rsfp);
321 else
322 (void)PerlIO_close(PL_rsfp);
323 PL_rsfp = Nullfp;
324 return s;
325 }
326
327 /* not at end of file, so we only read another line */
328 /* make corresponding updates to old pointers, for yyerror() */
329 oldprevlen = PL_oldbufptr - PL_bufend;
330 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
331 if (PL_last_uni)
332 oldunilen = PL_last_uni - PL_bufend;
333 if (PL_last_lop)
334 oldloplen = PL_last_lop - PL_bufend;
335 PL_linestart = PL_bufptr = s + prevlen;
336 PL_bufend = s + SvCUR(PL_linestr);
337 s = PL_bufptr;
338 PL_oldbufptr = s + oldprevlen;
339 PL_oldoldbufptr = s + oldoldprevlen;
340 if (PL_last_uni)
341 PL_last_uni = s + oldunilen;
342 if (PL_last_lop)
343 PL_last_lop = s + oldloplen;
344 incline(s);
345
346 /* debugger active and we're not compiling the debugger code,
347 * so store the line into the debugger's array of lines
348 */
349 if (PERLDB_LINE && PL_curstash != PL_debstash) {
350 SV * const sv = NEWSV(85,0);
351
352 sv_upgrade(sv, SVt_PVMG);
353 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
354 (void)SvIOK_on(sv);
355 SvIV_set(sv, 0);
356 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
357 }
358 }
359}
360
361STATIC char *
362S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
363{
364 register char *d = dest;
365 register char * const e = d + destlen - 3; /* two-character token, ending NUL */
366 for (;;) {
367 if (d >= e)
368 Perl_croak(aTHX_ ident_too_long);
369 if (isALNUM(*s)) /* UTF handled below */
370 *d++ = *s++;
371 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
372 *d++ = ':';
373 *d++ = ':';
374 s++;
375 }
376 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
377 *d++ = *s++;
378 *d++ = *s++;
379 }
380 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
381 char *t = s + UTF8SKIP(s);
382 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
383 t += UTF8SKIP(t);
384 if (d + (t - s) > e)
385 Perl_croak(aTHX_ ident_too_long);
386 Copy(s, d, t - s, char);
387 d += t - s;
388 s = t;
389 }
390 else {
391 *d = '\0';
392 *slp = d - dest;
393 return s;
394 }
395 }
396}
397
398/*
399 * S_incline
400 * This subroutine has nothing to do with tilting, whether at windmills
401 * or pinball tables. Its name is short for "increment line". It
402 * increments the current line number in CopLINE(PL_curcop) and checks
403 * to see whether the line starts with a comment of the form
404 * # line 500 "foo.pm"
405 * If so, it sets the current line number and file to the values in the comment.
406 */
407
408STATIC void
78160085 409S_incline(pTHX_ char *s)
e807ee50 410{
411 char *t;
78160085 412 char *n;
413 char *e;
e807ee50 414 char ch;
415
416 CopLINE_inc(PL_curcop);
417 if (*s++ != '#')
418 return;
419 while (SPACE_OR_TAB(*s)) s++;
420 if (strnEQ(s, "line", 4))
421 s += 4;
422 else
423 return;
424 if (SPACE_OR_TAB(*s))
425 s++;
426 else
427 return;
428 while (SPACE_OR_TAB(*s)) s++;
429 if (!isDIGIT(*s))
430 return;
431 n = s;
432 while (isDIGIT(*s))
433 s++;
434 while (SPACE_OR_TAB(*s))
435 s++;
436 if (*s == '"' && (t = strchr(s+1, '"'))) {
437 s++;
438 e = t + 1;
439 }
440 else {
78160085 441 for (t = s; !isSPACE(*t); t++) ;
e807ee50 442 e = t;
443 }
444 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
445 e++;
446 if (*e != '\n' && *e != '\0')
447 return; /* false alarm */
448
449 ch = *t;
450 *t = '\0';
451 if (t - s > 0) {
71511651 452/* this chunk was added to S_incline during 5.8.8. I don't know why but I don't
453 honestly care since I probably want to be bug-compatible anyway (mst) */
454
455/* ... my kingdom for a perl parser in perl ... (mst) */
456
457#ifdef PERL_5_8_8_PLUS
e807ee50 458#ifndef USE_ITHREADS
459 const char *cf = CopFILE(PL_curcop);
460 if (cf && strlen(cf) > 7 && strnEQ(cf, "(eval ", 6)) {
461 /* must copy *{"::_<(eval N)[oldfilename:L]"}
462 * to *{"::_<newfilename"} */
463 char smallbuf[256], smallbuf2[256];
464 char *tmpbuf, *tmpbuf2;
465 GV **gvp, *gv2;
466 STRLEN tmplen = strlen(cf);
467 STRLEN tmplen2 = strlen(s);
468 if (tmplen + 3 < sizeof smallbuf)
469 tmpbuf = smallbuf;
470 else
471 Newx(tmpbuf, tmplen + 3, char);
472 if (tmplen2 + 3 < sizeof smallbuf2)
473 tmpbuf2 = smallbuf2;
474 else
475 Newx(tmpbuf2, tmplen2 + 3, char);
476 tmpbuf[0] = tmpbuf2[0] = '_';
477 tmpbuf[1] = tmpbuf2[1] = '<';
478 memcpy(tmpbuf + 2, cf, ++tmplen);
479 memcpy(tmpbuf2 + 2, s, ++tmplen2);
480 ++tmplen; ++tmplen2;
481 gvp = (GV**)hv_fetch(PL_defstash, tmpbuf, tmplen, FALSE);
482 if (gvp) {
483 gv2 = *(GV**)hv_fetch(PL_defstash, tmpbuf2, tmplen2, TRUE);
484 if (!isGV(gv2))
485 gv_init(gv2, PL_defstash, tmpbuf2, tmplen2, FALSE);
486 /* adjust ${"::_<newfilename"} to store the new file name */
487 GvSV(gv2) = newSVpvn(tmpbuf2 + 2, tmplen2 - 2);
488 GvHV(gv2) = (HV*)SvREFCNT_inc(GvHV(*gvp));
489 GvAV(gv2) = (AV*)SvREFCNT_inc(GvAV(*gvp));
490 }
491 if (tmpbuf != smallbuf) Safefree(tmpbuf);
492 if (tmpbuf2 != smallbuf2) Safefree(tmpbuf2);
493 }
494#endif
71511651 495#endif
496/* second endif closes out the "are we 5.8.(8+)" conditional */
e807ee50 497 CopFILE_free(PL_curcop);
498 CopFILE_set(PL_curcop, s);
499 }
500 *t = ch;
501 CopLINE_set(PL_curcop, atoi(n)-1);
502}
0ba8c7aa 503
504/* scan_str
505 takes: start position in buffer
506 keep_quoted preserve \ on the embedded delimiter(s)
507 keep_delims preserve the delimiters around the string
508 returns: position to continue reading from buffer
509 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
510 updates the read buffer.
511
512 This subroutine pulls a string out of the input. It is called for:
513 q single quotes q(literal text)
514 ' single quotes 'literal text'
515 qq double quotes qq(interpolate $here please)
516 " double quotes "interpolate $here please"
517 qx backticks qx(/bin/ls -l)
518 ` backticks `/bin/ls -l`
519 qw quote words @EXPORT_OK = qw( func() $spam )
520 m// regexp match m/this/
521 s/// regexp substitute s/this/that/
522 tr/// string transliterate tr/this/that/
523 y/// string transliterate y/this/that/
524 ($*@) sub prototypes sub foo ($)
525 (stuff) sub attr parameters sub foo : attr(stuff)
526 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
527
528 In most of these cases (all but <>, patterns and transliterate)
529 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
530 calls scan_str(). s/// makes yylex() call scan_subst() which calls
531 scan_str(). tr/// and y/// make yylex() call scan_trans() which
532 calls scan_str().
533
534 It skips whitespace before the string starts, and treats the first
535 character as the delimiter. If the delimiter is one of ([{< then
536 the corresponding "close" character )]}> is used as the closing
537 delimiter. It allows quoting of delimiters, and if the string has
538 balanced delimiters ([{<>}]) it allows nesting.
539
540 On success, the SV with the resulting string is put into lex_stuff or,
541 if that is already non-NULL, into lex_repl. The second case occurs only
542 when parsing the RHS of the special constructs s/// and tr/// (y///).
543 For convenience, the terminating delimiter character is stuffed into
544 SvIVX of the SV.
545*/
546
547STATIC char *
548S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
549{
550 SV *sv; /* scalar value: string */
551 char *tmps; /* temp string, used for delimiter matching */
552 register char *s = start; /* current position in the buffer */
553 register char term; /* terminating character */
554 register char *to; /* current position in the sv's data */
555 I32 brackets = 1; /* bracket nesting level */
556 bool has_utf8 = FALSE; /* is there any utf8 content? */
557 I32 termcode; /* terminating char. code */
71511651 558 /* 5.8.7+ uses UTF8_MAXBYTES but also its utf8.h defs _MAXLEN to it so
559 I'm reasonably hopeful this won't destroy anything (mst) */
560 U8 termstr[UTF8_MAXLEN]; /* terminating string */
0ba8c7aa 561 STRLEN termlen; /* length of terminating string */
562 char *last = NULL; /* last position for nesting bracket */
563
564 /* skip space before the delimiter */
565 if (isSPACE(*s))
566 s = skipspace(s);
567
568 /* mark where we are, in case we need to report errors */
569 CLINE;
570
571 /* after skipping whitespace, the next character is the terminator */
572 term = *s;
573 if (!UTF) {
574 termcode = termstr[0] = term;
575 termlen = 1;
576 }
577 else {
578 termcode = utf8_to_uvchr((U8*)s, &termlen);
579 Copy(s, termstr, termlen, U8);
580 if (!UTF8_IS_INVARIANT(term))
581 has_utf8 = TRUE;
582 }
583
584 /* mark where we are */
585 PL_multi_start = CopLINE(PL_curcop);
586 PL_multi_open = term;
587
588 /* find corresponding closing delimiter */
589 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
590 termcode = termstr[0] = term = tmps[5];
591
592 PL_multi_close = term;
593
594 /* create a new SV to hold the contents. 87 is leak category, I'm
595 assuming. 79 is the SV's initial length. What a random number. */
596 sv = NEWSV(87,79);
597 sv_upgrade(sv, SVt_PVIV);
598 SvIV_set(sv, termcode);
599 (void)SvPOK_only(sv); /* validate pointer */
600
601 /* move past delimiter and try to read a complete string */
602 if (keep_delims)
603 sv_catpvn(sv, s, termlen);
604 s += termlen;
605 for (;;) {
606 if (PL_encoding && !UTF) {
607 bool cont = TRUE;
608
609 while (cont) {
610 int offset = s - SvPVX_const(PL_linestr);
611 const bool found = sv_cat_decode(sv, PL_encoding, PL_linestr,
612 &offset, (char*)termstr, termlen);
613 const char *ns = SvPVX_const(PL_linestr) + offset;
614 char *svlast = SvEND(sv) - 1;
615
616 for (; s < ns; s++) {
617 if (*s == '\n' && !PL_rsfp)
618 CopLINE_inc(PL_curcop);
619 }
620 if (!found)
621 goto read_more_line;
622 else {
623 /* handle quoted delimiters */
624 if (SvCUR(sv) > 1 && *(svlast-1) == '\\') {
625 const char *t;
626 for (t = svlast-2; t >= SvPVX_const(sv) && *t == '\\';)
627 t--;
628 if ((svlast-1 - t) % 2) {
629 if (!keep_quoted) {
630 *(svlast-1) = term;
631 *svlast = '\0';
632 SvCUR_set(sv, SvCUR(sv) - 1);
633 }
634 continue;
635 }
636 }
637 if (PL_multi_open == PL_multi_close) {
638 cont = FALSE;
639 }
640 else {
641 const char *t;
642 char *w;
643 if (!last)
644 last = SvPVX(sv);
645 for (t = w = last; t < svlast; w++, t++) {
646 /* At here, all closes are "was quoted" one,
647 so we don't check PL_multi_close. */
648 if (*t == '\\') {
649 if (!keep_quoted && *(t+1) == PL_multi_open)
650 t++;
651 else
652 *w++ = *t++;
653 }
654 else if (*t == PL_multi_open)
655 brackets++;
656
657 *w = *t;
658 }
659 if (w < t) {
660 *w++ = term;
661 *w = '\0';
662 SvCUR_set(sv, w - SvPVX_const(sv));
663 }
664 last = w;
665 if (--brackets <= 0)
666 cont = FALSE;
667 }
668 }
669 }
670 if (!keep_delims) {
671 SvCUR_set(sv, SvCUR(sv) - 1);
672 *SvEND(sv) = '\0';
673 }
674 break;
675 }
676
677 /* extend sv if need be */
678 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
679 /* set 'to' to the next character in the sv's string */
680 to = SvPVX(sv)+SvCUR(sv);
681
682 /* if open delimiter is the close delimiter read unbridle */
683 if (PL_multi_open == PL_multi_close) {
684 for (; s < PL_bufend; s++,to++) {
685 /* embedded newlines increment the current line number */
686 if (*s == '\n' && !PL_rsfp)
687 CopLINE_inc(PL_curcop);
688 /* handle quoted delimiters */
689 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
690 if (!keep_quoted && s[1] == term)
691 s++;
692 /* any other quotes are simply copied straight through */
693 else
694 *to++ = *s++;
695 }
696 /* terminate when run out of buffer (the for() condition), or
697 have found the terminator */
698 else if (*s == term) {
699 if (termlen == 1)
700 break;
701 if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen))
702 break;
703 }
704 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
705 has_utf8 = TRUE;
706 *to = *s;
707 }
708 }
709
710 /* if the terminator isn't the same as the start character (e.g.,
711 matched brackets), we have to allow more in the quoting, and
712 be prepared for nested brackets.
713 */
714 else {
715 /* read until we run out of string, or we find the terminator */
716 for (; s < PL_bufend; s++,to++) {
717 /* embedded newlines increment the line count */
718 if (*s == '\n' && !PL_rsfp)
719 CopLINE_inc(PL_curcop);
720 /* backslashes can escape the open or closing characters */
721 if (*s == '\\' && s+1 < PL_bufend) {
722 if (!keep_quoted &&
723 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
724 s++;
725 else
726 *to++ = *s++;
727 }
728 /* allow nested opens and closes */
729 else if (*s == PL_multi_close && --brackets <= 0)
730 break;
731 else if (*s == PL_multi_open)
732 brackets++;
733 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
734 has_utf8 = TRUE;
735 *to = *s;
736 }
737 }
738 /* terminate the copied string and update the sv's end-of-string */
739 *to = '\0';
740 SvCUR_set(sv, to - SvPVX_const(sv));
741
742 /*
743 * this next chunk reads more into the buffer if we're not done yet
744 */
745
746 if (s < PL_bufend)
747 break; /* handle case where we are done yet :-) */
748
749#ifndef PERL_STRICT_CR
750 if (to - SvPVX_const(sv) >= 2) {
751 if ((to[-2] == '\r' && to[-1] == '\n') ||
752 (to[-2] == '\n' && to[-1] == '\r'))
753 {
754 to[-2] = '\n';
755 to--;
756 SvCUR_set(sv, to - SvPVX_const(sv));
757 }
758 else if (to[-1] == '\r')
759 to[-1] = '\n';
760 }
761 else if (to - SvPVX_const(sv) == 1 && to[-1] == '\r')
762 to[-1] = '\n';
763#endif
764
765 read_more_line:
766 /* if we're out of file, or a read fails, bail and reset the current
767 line marker so we can report where the unterminated string began
768 */
769 if (!PL_rsfp ||
770 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
771 sv_free(sv);
772 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
773 return Nullch;
774 }
775 /* we read a line, so increment our line counter */
776 CopLINE_inc(PL_curcop);
777
778 /* update debugger info */
779 if (PERLDB_LINE && PL_curstash != PL_debstash) {
780 SV *sv = NEWSV(88,0);
781
782 sv_upgrade(sv, SVt_PVMG);
783 sv_setsv(sv,PL_linestr);
784 (void)SvIOK_on(sv);
785 SvIV_set(sv, 0);
786 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
787 }
788
789 /* having changed the buffer, we must update PL_bufend */
790 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
791 PL_last_lop = PL_last_uni = Nullch;
792 }
793
794 /* at this point, we have successfully read the delimited string */
795
796 if (!PL_encoding || UTF) {
797 if (keep_delims)
798 sv_catpvn(sv, s, termlen);
799 s += termlen;
800 }
801 if (has_utf8 || PL_encoding)
802 SvUTF8_on(sv);
803
804 PL_multi_end = CopLINE(PL_curcop);
805
806 /* if we allocated too much space, give some back */
807 if (SvCUR(sv) + 5 < SvLEN(sv)) {
808 SvLEN_set(sv, SvCUR(sv) + 1);
71511651 809/* 5.8.8 uses SvPV_renew, no prior version actually has the damn thing (mst) */
810#ifdef PERL_5_8_8_PLUS
0ba8c7aa 811 SvPV_renew(sv, SvLEN(sv));
71511651 812#else
813 Renew(SvPVX(sv), SvLEN(sv), char);
814#endif
0ba8c7aa 815 }
816
817 /* decide whether this is the first or second quoted string we've read
818 for this op
819 */
820
821 if (PL_lex_stuff)
822 PL_lex_repl = sv;
823 else
824 PL_lex_stuff = sv;
825 return s;
826}
827
828/*
829 * S_force_next
830 * When the lexer realizes it knows the next token (for instance,
831 * it is reordering tokens for the parser) then it can call S_force_next
832 * to know what token to return the next time the lexer is called. Caller
833 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
834 * handles the token correctly.
835 */
836
837STATIC void
838S_force_next(pTHX_ I32 type)
839{
840 PL_nexttype[PL_nexttoke] = type;
841 PL_nexttoke++;
842 if (PL_lex_state != LEX_KNOWNEXT) {
843 PL_lex_defer = PL_lex_state;
844 PL_lex_expect = PL_expect;
845 PL_lex_state = LEX_KNOWNEXT;
846 }
847}
Unnamed repository; edit this file to name it for gitweb.