Add strip_indent to Ctx::Simple.
[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
c630715a 56/* On MacOS, respect nonbreaking spaces */
57#ifdef MACOS_TRADITIONAL
58#define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
59#else
60#define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
61#endif
e807ee50 62
0ba8c7aa 63#define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
64
e807ee50 65#define LEX_NORMAL 10 /* normal code (ie not within "...") */
66#define LEX_INTERPNORMAL 9 /* code within a string, eg "$foo[$x+1]" */
67#define LEX_INTERPCASEMOD 8 /* expecting a \U, \Q or \E etc */
68#define LEX_INTERPPUSH 7 /* starting a new sublex parse level */
69#define LEX_INTERPSTART 6 /* expecting the start of a $var */
70
71 /* at end of code, eg "$x" followed by: */
72#define LEX_INTERPEND 5 /* ... eg not one of [, { or -> */
73#define LEX_INTERPENDMAYBE 4 /* ... eg one of [, { or -> */
74
75#define LEX_INTERPCONCAT 3 /* expecting anything, eg at start of
76 string or after \E, $foo, etc */
77#define LEX_INTERPCONST 2 /* NOT USED */
78#define LEX_FORMLINE 1 /* expecting a format line */
79#define LEX_KNOWNEXT 0 /* next token known; just return it */
80
14ab3472 81/* and these two are my own madness (mst) */
71511651 82
83#if PERL_REVISION == 5 && PERL_VERSION == 8 && PERL_SUBVERSION >= 8
84#define PERL_5_8_8_PLUS
85#endif
86
e38f5a74 87#if PERL_REVISION == 5 && PERL_VERSION > 8
88#define PERL_5_9_PLUS
89#endif
90
91#ifdef PERL_5_9_PLUS
14ab3472 92/* 5.9+ moves a bunch of things to a PL_parser struct so we need to
93 declare the backcompat macros for things to still work (mst) */
94
95/* XXX temporary backwards compatibility */
96#define PL_lex_brackets (PL_parser->lex_brackets)
97#define PL_lex_brackstack (PL_parser->lex_brackstack)
98#define PL_lex_casemods (PL_parser->lex_casemods)
99#define PL_lex_casestack (PL_parser->lex_casestack)
100#define PL_lex_defer (PL_parser->lex_defer)
101#define PL_lex_dojoin (PL_parser->lex_dojoin)
102#define PL_lex_expect (PL_parser->lex_expect)
103#define PL_lex_formbrack (PL_parser->lex_formbrack)
104#define PL_lex_inpat (PL_parser->lex_inpat)
105#define PL_lex_inwhat (PL_parser->lex_inwhat)
106#define PL_lex_op (PL_parser->lex_op)
107#define PL_lex_repl (PL_parser->lex_repl)
108#define PL_lex_starts (PL_parser->lex_starts)
109#define PL_lex_stuff (PL_parser->lex_stuff)
110#define PL_multi_start (PL_parser->multi_start)
111#define PL_multi_open (PL_parser->multi_open)
112#define PL_multi_close (PL_parser->multi_close)
113#define PL_pending_ident (PL_parser->pending_ident)
114#define PL_preambled (PL_parser->preambled)
115#define PL_sublex_info (PL_parser->sublex_info)
116#define PL_linestr (PL_parser->linestr)
117#define PL_sublex_info (PL_parser->sublex_info)
118#define PL_linestr (PL_parser->linestr)
119#define PL_expect (PL_parser->expect)
120#define PL_copline (PL_parser->copline)
121#define PL_bufptr (PL_parser->bufptr)
122#define PL_oldbufptr (PL_parser->oldbufptr)
123#define PL_oldoldbufptr (PL_parser->oldoldbufptr)
124#define PL_linestart (PL_parser->linestart)
125#define PL_bufend (PL_parser->bufend)
126#define PL_last_uni (PL_parser->last_uni)
127#define PL_last_lop (PL_parser->last_lop)
128#define PL_last_lop_op (PL_parser->last_lop_op)
129#define PL_lex_state (PL_parser->lex_state)
130#define PL_rsfp (PL_parser->rsfp)
131#define PL_rsfp_filters (PL_parser->rsfp_filters)
132#define PL_in_my (PL_parser->in_my)
133#define PL_in_my_stash (PL_parser->in_my_stash)
134#define PL_tokenbuf (PL_parser->tokenbuf)
135#define PL_multi_end (PL_parser->multi_end)
136#define PL_error_count (PL_parser->error_count)
137/* these three are from the non-PERL_MAD path but I don't -think- I need
138 the PERL_MAD stuff since my code isn't really populating things (mst) */
139# define PL_nexttoke (PL_parser->nexttoke)
140# define PL_nexttype (PL_parser->nexttype)
141# define PL_nextval (PL_parser->nextval)
142/* end of backcompat macros form 5.9 toke.c (mst) */
78160085 143#endif
14ab3472 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
35fc8ada 310 /* In perl versions previous to p4-rawid: //depot/perl@32954 -P
311 * preprocessors were supported here. We don't support -P at all, even
312 * on perls that support it, and use the following chunk from blead
313 * perl. (rafl)
314 */
315
316 /* Close the filehandle. Could be from
e807ee50 317 * STDIN, or a regular file. If we were reading code from
318 * STDIN (because the commandline held no -e or filename)
319 * then we don't close it, we reset it so the code can
320 * read from STDIN too.
321 */
322
35fc8ada 323 if ((PerlIO*)PL_rsfp == PerlIO_stdin())
e807ee50 324 PerlIO_clearerr(PL_rsfp);
325 else
326 (void)PerlIO_close(PL_rsfp);
327 PL_rsfp = Nullfp;
328 return s;
329 }
330
331 /* not at end of file, so we only read another line */
332 /* make corresponding updates to old pointers, for yyerror() */
333 oldprevlen = PL_oldbufptr - PL_bufend;
334 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
335 if (PL_last_uni)
336 oldunilen = PL_last_uni - PL_bufend;
337 if (PL_last_lop)
338 oldloplen = PL_last_lop - PL_bufend;
339 PL_linestart = PL_bufptr = s + prevlen;
340 PL_bufend = s + SvCUR(PL_linestr);
341 s = PL_bufptr;
342 PL_oldbufptr = s + oldprevlen;
343 PL_oldoldbufptr = s + oldoldprevlen;
344 if (PL_last_uni)
345 PL_last_uni = s + oldunilen;
346 if (PL_last_lop)
347 PL_last_lop = s + oldloplen;
348 incline(s);
349
350 /* debugger active and we're not compiling the debugger code,
351 * so store the line into the debugger's array of lines
352 */
353 if (PERLDB_LINE && PL_curstash != PL_debstash) {
354 SV * const sv = NEWSV(85,0);
355
356 sv_upgrade(sv, SVt_PVMG);
357 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
358 (void)SvIOK_on(sv);
359 SvIV_set(sv, 0);
360 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
361 }
362 }
363}
364
365STATIC char *
366S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
367{
368 register char *d = dest;
369 register char * const e = d + destlen - 3; /* two-character token, ending NUL */
370 for (;;) {
371 if (d >= e)
372 Perl_croak(aTHX_ ident_too_long);
373 if (isALNUM(*s)) /* UTF handled below */
374 *d++ = *s++;
375 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
376 *d++ = ':';
377 *d++ = ':';
378 s++;
379 }
380 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
381 *d++ = *s++;
382 *d++ = *s++;
383 }
384 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
385 char *t = s + UTF8SKIP(s);
386 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
387 t += UTF8SKIP(t);
388 if (d + (t - s) > e)
389 Perl_croak(aTHX_ ident_too_long);
390 Copy(s, d, t - s, char);
391 d += t - s;
392 s = t;
393 }
394 else {
395 *d = '\0';
396 *slp = d - dest;
397 return s;
398 }
399 }
400}
401
402/*
403 * S_incline
404 * This subroutine has nothing to do with tilting, whether at windmills
405 * or pinball tables. Its name is short for "increment line". It
406 * increments the current line number in CopLINE(PL_curcop) and checks
407 * to see whether the line starts with a comment of the form
408 * # line 500 "foo.pm"
409 * If so, it sets the current line number and file to the values in the comment.
410 */
411
412STATIC void
78160085 413S_incline(pTHX_ char *s)
e807ee50 414{
415 char *t;
78160085 416 char *n;
417 char *e;
e807ee50 418 char ch;
419
420 CopLINE_inc(PL_curcop);
421 if (*s++ != '#')
422 return;
423 while (SPACE_OR_TAB(*s)) s++;
424 if (strnEQ(s, "line", 4))
425 s += 4;
426 else
427 return;
428 if (SPACE_OR_TAB(*s))
429 s++;
430 else
431 return;
432 while (SPACE_OR_TAB(*s)) s++;
433 if (!isDIGIT(*s))
434 return;
435 n = s;
436 while (isDIGIT(*s))
437 s++;
438 while (SPACE_OR_TAB(*s))
439 s++;
440 if (*s == '"' && (t = strchr(s+1, '"'))) {
441 s++;
442 e = t + 1;
443 }
444 else {
78160085 445 for (t = s; !isSPACE(*t); t++) ;
e807ee50 446 e = t;
447 }
448 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
449 e++;
450 if (*e != '\n' && *e != '\0')
451 return; /* false alarm */
452
453 ch = *t;
454 *t = '\0';
455 if (t - s > 0) {
71511651 456/* this chunk was added to S_incline during 5.8.8. I don't know why but I don't
457 honestly care since I probably want to be bug-compatible anyway (mst) */
458
459/* ... my kingdom for a perl parser in perl ... (mst) */
460
461#ifdef PERL_5_8_8_PLUS
e807ee50 462#ifndef USE_ITHREADS
463 const char *cf = CopFILE(PL_curcop);
464 if (cf && strlen(cf) > 7 && strnEQ(cf, "(eval ", 6)) {
465 /* must copy *{"::_<(eval N)[oldfilename:L]"}
466 * to *{"::_<newfilename"} */
467 char smallbuf[256], smallbuf2[256];
468 char *tmpbuf, *tmpbuf2;
469 GV **gvp, *gv2;
470 STRLEN tmplen = strlen(cf);
471 STRLEN tmplen2 = strlen(s);
472 if (tmplen + 3 < sizeof smallbuf)
473 tmpbuf = smallbuf;
474 else
475 Newx(tmpbuf, tmplen + 3, char);
476 if (tmplen2 + 3 < sizeof smallbuf2)
477 tmpbuf2 = smallbuf2;
478 else
479 Newx(tmpbuf2, tmplen2 + 3, char);
480 tmpbuf[0] = tmpbuf2[0] = '_';
481 tmpbuf[1] = tmpbuf2[1] = '<';
482 memcpy(tmpbuf + 2, cf, ++tmplen);
483 memcpy(tmpbuf2 + 2, s, ++tmplen2);
484 ++tmplen; ++tmplen2;
485 gvp = (GV**)hv_fetch(PL_defstash, tmpbuf, tmplen, FALSE);
486 if (gvp) {
487 gv2 = *(GV**)hv_fetch(PL_defstash, tmpbuf2, tmplen2, TRUE);
488 if (!isGV(gv2))
489 gv_init(gv2, PL_defstash, tmpbuf2, tmplen2, FALSE);
490 /* adjust ${"::_<newfilename"} to store the new file name */
491 GvSV(gv2) = newSVpvn(tmpbuf2 + 2, tmplen2 - 2);
492 GvHV(gv2) = (HV*)SvREFCNT_inc(GvHV(*gvp));
493 GvAV(gv2) = (AV*)SvREFCNT_inc(GvAV(*gvp));
494 }
495 if (tmpbuf != smallbuf) Safefree(tmpbuf);
496 if (tmpbuf2 != smallbuf2) Safefree(tmpbuf2);
497 }
498#endif
71511651 499#endif
500/* second endif closes out the "are we 5.8.(8+)" conditional */
e807ee50 501 CopFILE_free(PL_curcop);
502 CopFILE_set(PL_curcop, s);
503 }
504 *t = ch;
505 CopLINE_set(PL_curcop, atoi(n)-1);
506}
0ba8c7aa 507
508/* scan_str
509 takes: start position in buffer
510 keep_quoted preserve \ on the embedded delimiter(s)
511 keep_delims preserve the delimiters around the string
512 returns: position to continue reading from buffer
513 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
514 updates the read buffer.
515
516 This subroutine pulls a string out of the input. It is called for:
517 q single quotes q(literal text)
518 ' single quotes 'literal text'
519 qq double quotes qq(interpolate $here please)
520 " double quotes "interpolate $here please"
521 qx backticks qx(/bin/ls -l)
522 ` backticks `/bin/ls -l`
523 qw quote words @EXPORT_OK = qw( func() $spam )
524 m// regexp match m/this/
525 s/// regexp substitute s/this/that/
526 tr/// string transliterate tr/this/that/
527 y/// string transliterate y/this/that/
528 ($*@) sub prototypes sub foo ($)
529 (stuff) sub attr parameters sub foo : attr(stuff)
530 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
531
532 In most of these cases (all but <>, patterns and transliterate)
533 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
534 calls scan_str(). s/// makes yylex() call scan_subst() which calls
535 scan_str(). tr/// and y/// make yylex() call scan_trans() which
536 calls scan_str().
537
538 It skips whitespace before the string starts, and treats the first
539 character as the delimiter. If the delimiter is one of ([{< then
540 the corresponding "close" character )]}> is used as the closing
541 delimiter. It allows quoting of delimiters, and if the string has
542 balanced delimiters ([{<>}]) it allows nesting.
543
544 On success, the SV with the resulting string is put into lex_stuff or,
545 if that is already non-NULL, into lex_repl. The second case occurs only
546 when parsing the RHS of the special constructs s/// and tr/// (y///).
547 For convenience, the terminating delimiter character is stuffed into
548 SvIVX of the SV.
549*/
550
551STATIC char *
552S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
553{
554 SV *sv; /* scalar value: string */
555 char *tmps; /* temp string, used for delimiter matching */
556 register char *s = start; /* current position in the buffer */
557 register char term; /* terminating character */
558 register char *to; /* current position in the sv's data */
559 I32 brackets = 1; /* bracket nesting level */
560 bool has_utf8 = FALSE; /* is there any utf8 content? */
561 I32 termcode; /* terminating char. code */
71511651 562 /* 5.8.7+ uses UTF8_MAXBYTES but also its utf8.h defs _MAXLEN to it so
563 I'm reasonably hopeful this won't destroy anything (mst) */
564 U8 termstr[UTF8_MAXLEN]; /* terminating string */
0ba8c7aa 565 STRLEN termlen; /* length of terminating string */
566 char *last = NULL; /* last position for nesting bracket */
567
568 /* skip space before the delimiter */
569 if (isSPACE(*s))
570 s = skipspace(s);
571
572 /* mark where we are, in case we need to report errors */
573 CLINE;
574
575 /* after skipping whitespace, the next character is the terminator */
576 term = *s;
577 if (!UTF) {
578 termcode = termstr[0] = term;
579 termlen = 1;
580 }
581 else {
582 termcode = utf8_to_uvchr((U8*)s, &termlen);
583 Copy(s, termstr, termlen, U8);
584 if (!UTF8_IS_INVARIANT(term))
585 has_utf8 = TRUE;
586 }
587
588 /* mark where we are */
589 PL_multi_start = CopLINE(PL_curcop);
590 PL_multi_open = term;
591
592 /* find corresponding closing delimiter */
593 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
594 termcode = termstr[0] = term = tmps[5];
595
596 PL_multi_close = term;
597
598 /* create a new SV to hold the contents. 87 is leak category, I'm
599 assuming. 79 is the SV's initial length. What a random number. */
600 sv = NEWSV(87,79);
601 sv_upgrade(sv, SVt_PVIV);
602 SvIV_set(sv, termcode);
603 (void)SvPOK_only(sv); /* validate pointer */
604
605 /* move past delimiter and try to read a complete string */
606 if (keep_delims)
607 sv_catpvn(sv, s, termlen);
608 s += termlen;
609 for (;;) {
610 if (PL_encoding && !UTF) {
611 bool cont = TRUE;
612
613 while (cont) {
614 int offset = s - SvPVX_const(PL_linestr);
615 const bool found = sv_cat_decode(sv, PL_encoding, PL_linestr,
616 &offset, (char*)termstr, termlen);
617 const char *ns = SvPVX_const(PL_linestr) + offset;
618 char *svlast = SvEND(sv) - 1;
619
620 for (; s < ns; s++) {
621 if (*s == '\n' && !PL_rsfp)
622 CopLINE_inc(PL_curcop);
623 }
624 if (!found)
625 goto read_more_line;
626 else {
627 /* handle quoted delimiters */
628 if (SvCUR(sv) > 1 && *(svlast-1) == '\\') {
629 const char *t;
630 for (t = svlast-2; t >= SvPVX_const(sv) && *t == '\\';)
631 t--;
632 if ((svlast-1 - t) % 2) {
633 if (!keep_quoted) {
634 *(svlast-1) = term;
635 *svlast = '\0';
636 SvCUR_set(sv, SvCUR(sv) - 1);
637 }
638 continue;
639 }
640 }
641 if (PL_multi_open == PL_multi_close) {
642 cont = FALSE;
643 }
644 else {
645 const char *t;
646 char *w;
647 if (!last)
648 last = SvPVX(sv);
649 for (t = w = last; t < svlast; w++, t++) {
650 /* At here, all closes are "was quoted" one,
651 so we don't check PL_multi_close. */
652 if (*t == '\\') {
653 if (!keep_quoted && *(t+1) == PL_multi_open)
654 t++;
655 else
656 *w++ = *t++;
657 }
658 else if (*t == PL_multi_open)
659 brackets++;
660
661 *w = *t;
662 }
663 if (w < t) {
664 *w++ = term;
665 *w = '\0';
666 SvCUR_set(sv, w - SvPVX_const(sv));
667 }
668 last = w;
669 if (--brackets <= 0)
670 cont = FALSE;
671 }
672 }
673 }
674 if (!keep_delims) {
675 SvCUR_set(sv, SvCUR(sv) - 1);
676 *SvEND(sv) = '\0';
677 }
678 break;
679 }
680
681 /* extend sv if need be */
682 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
683 /* set 'to' to the next character in the sv's string */
684 to = SvPVX(sv)+SvCUR(sv);
685
686 /* if open delimiter is the close delimiter read unbridle */
687 if (PL_multi_open == PL_multi_close) {
688 for (; s < PL_bufend; s++,to++) {
689 /* embedded newlines increment the current line number */
690 if (*s == '\n' && !PL_rsfp)
691 CopLINE_inc(PL_curcop);
692 /* handle quoted delimiters */
693 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
694 if (!keep_quoted && s[1] == term)
695 s++;
696 /* any other quotes are simply copied straight through */
697 else
698 *to++ = *s++;
699 }
700 /* terminate when run out of buffer (the for() condition), or
701 have found the terminator */
702 else if (*s == term) {
703 if (termlen == 1)
704 break;
705 if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen))
706 break;
707 }
708 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
709 has_utf8 = TRUE;
710 *to = *s;
711 }
712 }
713
714 /* if the terminator isn't the same as the start character (e.g.,
715 matched brackets), we have to allow more in the quoting, and
716 be prepared for nested brackets.
717 */
718 else {
719 /* read until we run out of string, or we find the terminator */
720 for (; s < PL_bufend; s++,to++) {
721 /* embedded newlines increment the line count */
722 if (*s == '\n' && !PL_rsfp)
723 CopLINE_inc(PL_curcop);
724 /* backslashes can escape the open or closing characters */
725 if (*s == '\\' && s+1 < PL_bufend) {
726 if (!keep_quoted &&
727 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
728 s++;
729 else
730 *to++ = *s++;
731 }
732 /* allow nested opens and closes */
733 else if (*s == PL_multi_close && --brackets <= 0)
734 break;
735 else if (*s == PL_multi_open)
736 brackets++;
737 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
738 has_utf8 = TRUE;
739 *to = *s;
740 }
741 }
742 /* terminate the copied string and update the sv's end-of-string */
743 *to = '\0';
744 SvCUR_set(sv, to - SvPVX_const(sv));
745
746 /*
747 * this next chunk reads more into the buffer if we're not done yet
748 */
749
750 if (s < PL_bufend)
751 break; /* handle case where we are done yet :-) */
752
753#ifndef PERL_STRICT_CR
754 if (to - SvPVX_const(sv) >= 2) {
755 if ((to[-2] == '\r' && to[-1] == '\n') ||
756 (to[-2] == '\n' && to[-1] == '\r'))
757 {
758 to[-2] = '\n';
759 to--;
760 SvCUR_set(sv, to - SvPVX_const(sv));
761 }
762 else if (to[-1] == '\r')
763 to[-1] = '\n';
764 }
765 else if (to - SvPVX_const(sv) == 1 && to[-1] == '\r')
766 to[-1] = '\n';
767#endif
768
769 read_more_line:
770 /* if we're out of file, or a read fails, bail and reset the current
771 line marker so we can report where the unterminated string began
772 */
773 if (!PL_rsfp ||
774 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
775 sv_free(sv);
776 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
777 return Nullch;
778 }
779 /* we read a line, so increment our line counter */
780 CopLINE_inc(PL_curcop);
781
782 /* update debugger info */
783 if (PERLDB_LINE && PL_curstash != PL_debstash) {
784 SV *sv = NEWSV(88,0);
785
786 sv_upgrade(sv, SVt_PVMG);
787 sv_setsv(sv,PL_linestr);
788 (void)SvIOK_on(sv);
789 SvIV_set(sv, 0);
790 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
791 }
792
793 /* having changed the buffer, we must update PL_bufend */
794 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
795 PL_last_lop = PL_last_uni = Nullch;
796 }
797
798 /* at this point, we have successfully read the delimited string */
799
800 if (!PL_encoding || UTF) {
801 if (keep_delims)
802 sv_catpvn(sv, s, termlen);
803 s += termlen;
804 }
805 if (has_utf8 || PL_encoding)
806 SvUTF8_on(sv);
807
808 PL_multi_end = CopLINE(PL_curcop);
809
810 /* if we allocated too much space, give some back */
811 if (SvCUR(sv) + 5 < SvLEN(sv)) {
812 SvLEN_set(sv, SvCUR(sv) + 1);
71511651 813/* 5.8.8 uses SvPV_renew, no prior version actually has the damn thing (mst) */
814#ifdef PERL_5_8_8_PLUS
0ba8c7aa 815 SvPV_renew(sv, SvLEN(sv));
71511651 816#else
817 Renew(SvPVX(sv), SvLEN(sv), char);
818#endif
0ba8c7aa 819 }
820
821 /* decide whether this is the first or second quoted string we've read
822 for this op
823 */
824
825 if (PL_lex_stuff)
826 PL_lex_repl = sv;
827 else
828 PL_lex_stuff = sv;
829 return s;
830}
831
832/*
833 * S_force_next
834 * When the lexer realizes it knows the next token (for instance,
835 * it is reordering tokens for the parser) then it can call S_force_next
836 * to know what token to return the next time the lexer is called. Caller
837 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
838 * handles the token correctly.
839 */
840
841STATIC void
842S_force_next(pTHX_ I32 type)
843{
844 PL_nexttype[PL_nexttoke] = type;
845 PL_nexttoke++;
846 if (PL_lex_state != LEX_KNOWNEXT) {
847 PL_lex_defer = PL_lex_state;
848 PL_lex_expect = PL_expect;
849 PL_lex_state = LEX_KNOWNEXT;
850 }
851}
92997e69 852
853#define XFAKEBRACK 128
854
855STATIC char *
856S_scan_ident(pTHX_ register char *s, register const char *send, char *dest, STRLEN destlen, I32 ck_uni)
857{
858 register char *d;
859 register char *e;
860 char *bracket = Nullch;
861 char funny = *s++;
862
863 if (isSPACE(*s))
864 s = skipspace(s);
865 d = dest;
866 e = d + destlen - 3; /* two-character token, ending NUL */
867 if (isDIGIT(*s)) {
868 while (isDIGIT(*s)) {
869 if (d >= e)
870 Perl_croak(aTHX_ ident_too_long);
871 *d++ = *s++;
872 }
873 }
874 else {
875 for (;;) {
876 if (d >= e)
877 Perl_croak(aTHX_ ident_too_long);
878 if (isALNUM(*s)) /* UTF handled below */
879 *d++ = *s++;
880 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
881 *d++ = ':';
882 *d++ = ':';
883 s++;
884 }
885 else if (*s == ':' && s[1] == ':') {
886 *d++ = *s++;
887 *d++ = *s++;
888 }
889 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
890 char *t = s + UTF8SKIP(s);
891 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
892 t += UTF8SKIP(t);
893 if (d + (t - s) > e)
894 Perl_croak(aTHX_ ident_too_long);
895 Copy(s, d, t - s, char);
896 d += t - s;
897 s = t;
898 }
899 else
900 break;
901 }
902 }
903 *d = '\0';
904 d = dest;
905 if (*d) {
906 if (PL_lex_state != LEX_NORMAL)
907 PL_lex_state = LEX_INTERPENDMAYBE;
908 return s;
909 }
910 if (*s == '$' && s[1] &&
911 (isALNUM_lazy_if(s+1,UTF) || s[1] == '$' || s[1] == '{' || strnEQ(s+1,"::",2)) )
912 {
913 return s;
914 }
915 if (*s == '{') {
916 bracket = s;
917 s++;
918 }
919 else if (ck_uni)
920 check_uni();
921 if (s < send)
922 *d = *s++;
923 d[1] = '\0';
924 if (*d == '^' && *s && isCONTROLVAR(*s)) {
925 *d = toCTRL(*s);
926 s++;
927 }
928 if (bracket) {
929 if (isSPACE(s[-1])) {
930 while (s < send) {
931 const char ch = *s++;
932 if (!SPACE_OR_TAB(ch)) {
933 *d = ch;
934 break;
935 }
936 }
937 }
938 if (isIDFIRST_lazy_if(d,UTF)) {
939 d++;
940 if (UTF) {
941 e = s;
942 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
943 e += UTF8SKIP(e);
944 while (e < send && UTF8_IS_CONTINUED(*e) && is_utf8_mark((U8*)e))
945 e += UTF8SKIP(e);
946 }
947 Copy(s, d, e - s, char);
948 d += e - s;
949 s = e;
950 }
951 else {
952 while ((isALNUM(*s) || *s == ':') && d < e)
953 *d++ = *s++;
954 if (d >= e)
955 Perl_croak(aTHX_ ident_too_long);
956 }
957 *d = '\0';
958 while (s < send && SPACE_OR_TAB(*s)) s++;
959 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
960 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
961 const char *brack = *s == '[' ? "[...]" : "{...}";
962 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
963 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
964 funny, dest, brack, funny, dest, brack);
965 }
966 bracket++;
967 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
968 return s;
969 }
970 }
971 /* Handle extended ${^Foo} variables
972 * 1999-02-27 mjd-perl-patch@plover.com */
973 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
974 && isALNUM(*s))
975 {
976 d++;
977 while (isALNUM(*s) && d < e) {
978 *d++ = *s++;
979 }
980 if (d >= e)
981 Perl_croak(aTHX_ ident_too_long);
982 *d = '\0';
983 }
984 if (*s == '}') {
985 s++;
986 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets) {
987 PL_lex_state = LEX_INTERPEND;
988 PL_expect = XREF;
989 }
990 if (funny == '#')
991 funny = '@';
992 if (PL_lex_state == LEX_NORMAL) {
993 if (ckWARN(WARN_AMBIGUOUS) &&
994 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
995 {
996 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
997 "Ambiguous use of %c{%s} resolved to %c%s",
998 funny, dest, funny, dest);
999 }
1000 }
1001 }
1002 else {
1003 s = bracket; /* let the parser handle it */
1004 *dest = '\0';
1005 }
1006 }
1007 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
1008 PL_lex_state = LEX_INTERPEND;
1009 return s;
1010}