1 /* stolen_chunk_of_toke.c - from perl 5.8.8 toke.c
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
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.
12 * "It all comes from here, the stench and the peril." --Frodo
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
20 /* the following #defines are stolen from assorted headers, not toke.c (mst) */
22 #define skipspace(a) S_skipspace(aTHX_ a, 0)
23 #define peekspace(a) S_skipspace(aTHX_ a, 1)
24 #define skipspace_force(a) S_skipspace(aTHX_ a, 2)
25 #define incline(a) S_incline(aTHX_ a)
26 #define filter_gets(a,b,c) S_filter_gets(aTHX_ a,b,c)
27 #define scan_str(a,b,c) S_scan_str(aTHX_ a,b,c)
28 #define scan_word(a,b,c,d,e) S_scan_word(aTHX_ a,b,c,d,e)
29 #define scan_ident(a,b,c,d,e) S_scan_ident(aTHX_ a,b,c,d,e)
31 STATIC void S_incline(pTHX_ char *s);
32 STATIC char* S_skipspace(pTHX_ char *s, int incline);
33 STATIC char * S_filter_gets(pTHX_ SV *sv, PerlIO *fp, STRLEN append);
34 STATIC char* S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims);
35 STATIC char* S_scan_word(pTHX_ char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp);
37 #define DPTR2FPTR(t,p) ((t)PTR2nat(p)) /* data pointer to function pointer */
38 #define FPTR2DPTR(t,p) ((t)PTR2nat(p)) /* function pointer to data pointer */
39 #define PTR2nat(p) (PTRV)(p) /* pointer to integer of PTRSIZE */
41 /* conditionalise these two because as of 5.9.5 we already get them from
44 #define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)safemalloc((MEM_SIZE)((n)*sizeof(t)))))
47 #define SvPVX_const(sv) ((const char*) (0 + SvPVX(sv)))
49 #ifndef MEM_WRAP_CHECK_
50 #define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
53 #define SvPV_renew(sv,n) \
54 STMT_START { SvLEN_set(sv, n); \
55 SvPV_set((sv), (MEM_WRAP_CHECK_(n,char) \
56 (char*)saferealloc((Malloc_t)SvPVX(sv), \
60 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
62 /* On MacOS, respect nonbreaking spaces */
63 #ifdef MACOS_TRADITIONAL
64 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
66 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
70 * Normally, during compile time, PL_curcop == &PL_compiling is true. However,
71 * Devel::Declare makes the interpreter call back to perl during compile time,
72 * which temporarily enters runtime. Then perl space calls various functions
73 * from this file, which are designed to work during compile time. They all
74 * happen to operate on PL_curcop, not PL_compiling. That doesn't make a
75 * difference in the core, but it does for Devel::Declare, which operates at
76 * runtime, but still wants to mangle the things that are about to be compiled.
77 * That's why we define our own PL_curcop and make it point to PL_compiling
81 #define PL_curcop (&PL_compiling)
83 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
85 #define LEX_NORMAL 10 /* normal code (ie not within "...") */
86 #define LEX_INTERPNORMAL 9 /* code within a string, eg "$foo[$x+1]" */
87 #define LEX_INTERPCASEMOD 8 /* expecting a \U, \Q or \E etc */
88 #define LEX_INTERPPUSH 7 /* starting a new sublex parse level */
89 #define LEX_INTERPSTART 6 /* expecting the start of a $var */
91 /* at end of code, eg "$x" followed by: */
92 #define LEX_INTERPEND 5 /* ... eg not one of [, { or -> */
93 #define LEX_INTERPENDMAYBE 4 /* ... eg one of [, { or -> */
95 #define LEX_INTERPCONCAT 3 /* expecting anything, eg at start of
96 string or after \E, $foo, etc */
97 #define LEX_INTERPCONST 2 /* NOT USED */
98 #define LEX_FORMLINE 1 /* expecting a format line */
99 #define LEX_KNOWNEXT 0 /* next token known; just return it */
101 /* and these two are my own madness (mst) */
103 #if PERL_REVISION == 5 && PERL_VERSION == 8 && PERL_SUBVERSION >= 8
104 #define PERL_5_8_8_PLUS
107 #if PERL_REVISION == 5 && PERL_VERSION > 8
108 #define PERL_5_9_PLUS
112 /* 5.9+ moves a bunch of things to a PL_parser struct so we need to
113 declare the backcompat macros for things to still work (mst) */
115 /* XXX temporary backwards compatibility */
116 #define PL_lex_brackets (PL_parser->lex_brackets)
117 #define PL_lex_brackstack (PL_parser->lex_brackstack)
118 #define PL_lex_casemods (PL_parser->lex_casemods)
119 #define PL_lex_casestack (PL_parser->lex_casestack)
120 #define PL_lex_defer (PL_parser->lex_defer)
121 #define PL_lex_dojoin (PL_parser->lex_dojoin)
122 #define PL_lex_expect (PL_parser->lex_expect)
123 #define PL_lex_formbrack (PL_parser->lex_formbrack)
124 #define PL_lex_inpat (PL_parser->lex_inpat)
125 #define PL_lex_inwhat (PL_parser->lex_inwhat)
126 #define PL_lex_op (PL_parser->lex_op)
127 #define PL_lex_repl (PL_parser->lex_repl)
128 #define PL_lex_starts (PL_parser->lex_starts)
129 #define PL_lex_stuff (PL_parser->lex_stuff)
130 #define PL_multi_start (PL_parser->multi_start)
131 #define PL_multi_open (PL_parser->multi_open)
132 #define PL_multi_close (PL_parser->multi_close)
133 #define PL_pending_ident (PL_parser->pending_ident)
134 #define PL_preambled (PL_parser->preambled)
135 #define PL_sublex_info (PL_parser->sublex_info)
136 #define PL_linestr (PL_parser->linestr)
137 #define PL_sublex_info (PL_parser->sublex_info)
138 #define PL_linestr (PL_parser->linestr)
139 #define PL_expect (PL_parser->expect)
140 #define PL_copline (PL_parser->copline)
141 #define PL_bufptr (PL_parser->bufptr)
142 #define PL_oldbufptr (PL_parser->oldbufptr)
143 #define PL_oldoldbufptr (PL_parser->oldoldbufptr)
144 #define PL_linestart (PL_parser->linestart)
145 #define PL_bufend (PL_parser->bufend)
146 #define PL_last_uni (PL_parser->last_uni)
147 #define PL_last_lop (PL_parser->last_lop)
148 #define PL_last_lop_op (PL_parser->last_lop_op)
149 #define PL_lex_state (PL_parser->lex_state)
150 #define PL_rsfp (PL_parser->rsfp)
151 #define PL_rsfp_filters (PL_parser->rsfp_filters)
152 #define PL_in_my (PL_parser->in_my)
153 #define PL_in_my_stash (PL_parser->in_my_stash)
154 #define PL_tokenbuf (PL_parser->tokenbuf)
155 #define PL_multi_end (PL_parser->multi_end)
156 #define PL_error_count (PL_parser->error_count)
157 #define PL_nexttoke (PL_parser->nexttoke)
158 /* these are from the non-PERL_MAD path but I don't -think- I need
159 the PERL_MAD stuff since my code isn't really populating things (mst) */
161 # define PL_curforce (PL_parser->curforce)
162 # define PL_lasttoke (PL_parser->lasttoke)
164 # define PL_nexttype (PL_parser->nexttype)
165 # define PL_nextval (PL_parser->nextval)
167 /* end of backcompat macros from 5.9 toke.c (mst) */
170 /* when ccflags include -DDEBUGGING we need this for earlier 5.8 perls */
171 #ifndef SvPV_nolen_const
172 #define SvPV_nolen_const SvPV_nolen
175 /* and now we're back to the toke.c stuff again (mst) */
177 static const char ident_too_long[] =
178 "Identifier too long";
179 static const char c_without_g[] =
180 "Use of /c modifier is meaningless without /g";
181 static const char c_in_subst[] =
182 "Use of /c modifier is meaningless in s///";
184 #ifdef USE_UTF8_SCRIPTS
185 # define UTF (!IN_BYTES)
187 # define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || (PL_hints & HINT_UTF8))
190 /* Invoke the idxth filter function for the current rsfp. */
191 /* maxlen 0 = read one text line */
193 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
198 if (!PL_rsfp_filters)
200 if (idx > AvFILLp(PL_rsfp_filters)) { /* Any more filters? */
201 /* Provide a default input filter to make life easy. */
202 /* Note that we append to the line. This is handy. */
203 DEBUG_P(PerlIO_printf(Perl_debug_log,
204 "filter_read %d: from rsfp\n", idx));
208 const int old_len = SvCUR(buf_sv);
210 /* ensure buf_sv is large enough */
211 SvGROW(buf_sv, (STRLEN)(old_len + maxlen)) ;
212 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
213 if (PerlIO_error(PL_rsfp))
214 return -1; /* error */
216 return 0 ; /* end of file */
218 SvCUR_set(buf_sv, old_len + len) ;
221 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
222 if (PerlIO_error(PL_rsfp))
223 return -1; /* error */
225 return 0 ; /* end of file */
228 return SvCUR(buf_sv);
230 /* Skip this filter slot if filter has been deleted */
231 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef) {
232 DEBUG_P(PerlIO_printf(Perl_debug_log,
233 "filter_read %d: skipped (filter deleted)\n",
235 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
237 /* Get function pointer hidden within datasv */
238 funcp = DPTR2FPTR(filter_t, IoANY(datasv));
239 DEBUG_P(PerlIO_printf(Perl_debug_log,
240 "filter_read %d: via function %p (%s)\n",
241 idx, datasv, SvPV_nolen_const(datasv)));
242 /* Call function. The function is expected to */
243 /* call "FILTER_READ(idx+1, buf_sv)" first. */
244 /* Return: <0:error, =0:eof, >0:not eof */
245 return (*funcp)(aTHX_ idx, buf_sv, maxlen);
249 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
251 #ifdef PERL_CR_FILTER
252 if (!PL_rsfp_filters) {
253 filter_add(S_cr_textfilter,NULL);
256 if (PL_rsfp_filters) {
258 SvCUR_set(sv, 0); /* start with empty line */
259 if (FILTER_READ(0, sv, 0) > 0)
260 return ( SvPVX(sv) ) ;
265 return (sv_gets(sv, fp, append));
270 * Called to gobble the appropriate amount and type of whitespace.
271 * Skips comments as well.
275 S_skipspace(pTHX_ register char *s, int incline)
277 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
278 while (s < PL_bufend && SPACE_OR_TAB(*s))
284 SSize_t oldprevlen, oldoldprevlen;
285 SSize_t oldloplen = 0, oldunilen = 0;
286 while (s < PL_bufend && isSPACE(*s)) {
287 if (*s++ == '\n' && ((incline == 2) || PL_in_eval && !PL_rsfp && !incline))
292 if (s < PL_bufend && *s == '#') {
293 while (s < PL_bufend && *s != '\n')
297 if (PL_in_eval && !PL_rsfp && !incline) {
304 /* also skip leading whitespace on the beginning of a line before deciding
305 * whether or not to recharge the linestr. --rafl
307 while (s < PL_bufend && isSPACE(*s)) {
308 if (*s++ == '\n' && PL_in_eval && !PL_rsfp && !incline)
312 /* only continue to recharge the buffer if we're at the end
313 * of the buffer, we're not reading from a source filter, and
314 * we're in normal lexing mode
316 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
317 PL_lex_state == LEX_FORMLINE)
320 /* try to recharge the buffer */
321 if ((s = filter_gets(PL_linestr, PL_rsfp,
322 (prevlen = SvCUR(PL_linestr)))) == Nullch)
324 /* end of file. Add on the -p or -n magic */
327 ";}continue{print or die qq(-p destination: $!\\n);}");
328 PL_minus_n = PL_minus_p = 0;
330 else if (PL_minus_n) {
331 sv_setpvn(PL_linestr, ";}", 2);
335 sv_setpvn(PL_linestr,";", 1);
337 /* reset variables for next time we lex */
338 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
340 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
341 PL_last_lop = PL_last_uni = Nullch;
343 /* In perl versions previous to p4-rawid: //depot/perl@32954 -P
344 * preprocessors were supported here. We don't support -P at all, even
345 * on perls that support it, and use the following chunk from blead
349 /* Close the filehandle. Could be from
350 * STDIN, or a regular file. If we were reading code from
351 * STDIN (because the commandline held no -e or filename)
352 * then we don't close it, we reset it so the code can
353 * read from STDIN too.
356 if ((PerlIO*)PL_rsfp == PerlIO_stdin())
357 PerlIO_clearerr(PL_rsfp);
359 (void)PerlIO_close(PL_rsfp);
364 /* not at end of file, so we only read another line */
365 /* make corresponding updates to old pointers, for yyerror() */
366 oldprevlen = PL_oldbufptr - PL_bufend;
367 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
369 oldunilen = PL_last_uni - PL_bufend;
371 oldloplen = PL_last_lop - PL_bufend;
372 PL_linestart = PL_bufptr = s + prevlen;
373 PL_bufend = s + SvCUR(PL_linestr);
375 PL_oldbufptr = s + oldprevlen;
376 PL_oldoldbufptr = s + oldoldprevlen;
378 PL_last_uni = s + oldunilen;
380 PL_last_lop = s + oldloplen;
384 /* debugger active and we're not compiling the debugger code,
385 * so store the line into the debugger's array of lines
387 if (PERLDB_LINE && PL_curstash != PL_debstash) {
388 SV * const sv = NEWSV(85,0);
390 sv_upgrade(sv, SVt_PVMG);
391 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
394 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
400 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
402 register char *d = dest;
403 register char * const e = d + destlen - 3; /* two-character token, ending NUL */
406 Perl_croak(aTHX_ ident_too_long);
407 if (isALNUM(*s)) /* UTF handled below */
409 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
414 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
418 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
419 char *t = s + UTF8SKIP(s);
420 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
423 Perl_croak(aTHX_ ident_too_long);
424 Copy(s, d, t - s, char);
438 * This subroutine has nothing to do with tilting, whether at windmills
439 * or pinball tables. Its name is short for "increment line". It
440 * increments the current line number in CopLINE(PL_curcop) and checks
441 * to see whether the line starts with a comment of the form
442 * # line 500 "foo.pm"
443 * If so, it sets the current line number and file to the values in the comment.
447 S_incline(pTHX_ char *s)
454 CopLINE_inc(PL_curcop);
457 while (SPACE_OR_TAB(*s)) s++;
458 if (strnEQ(s, "line", 4))
462 if (SPACE_OR_TAB(*s))
466 while (SPACE_OR_TAB(*s)) s++;
472 while (SPACE_OR_TAB(*s))
474 if (*s == '"' && (t = strchr(s+1, '"'))) {
479 for (t = s; !isSPACE(*t); t++) ;
482 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
484 if (*e != '\n' && *e != '\0')
485 return; /* false alarm */
490 /* this chunk was added to S_incline during 5.8.8. I don't know why but I don't
491 honestly care since I probably want to be bug-compatible anyway (mst) */
493 /* ... my kingdom for a perl parser in perl ... (mst) */
495 #ifdef PERL_5_8_8_PLUS
497 const char *cf = CopFILE(PL_curcop);
498 if (cf && strlen(cf) > 7 && strnEQ(cf, "(eval ", 6)) {
499 /* must copy *{"::_<(eval N)[oldfilename:L]"}
500 * to *{"::_<newfilename"} */
501 char smallbuf[256], smallbuf2[256];
502 char *tmpbuf, *tmpbuf2;
504 STRLEN tmplen = strlen(cf);
505 STRLEN tmplen2 = strlen(s);
506 if (tmplen + 3 < sizeof smallbuf)
509 Newx(tmpbuf, tmplen + 3, char);
510 if (tmplen2 + 3 < sizeof smallbuf2)
513 Newx(tmpbuf2, tmplen2 + 3, char);
514 tmpbuf[0] = tmpbuf2[0] = '_';
515 tmpbuf[1] = tmpbuf2[1] = '<';
516 memcpy(tmpbuf + 2, cf, ++tmplen);
517 memcpy(tmpbuf2 + 2, s, ++tmplen2);
519 gvp = (GV**)hv_fetch(PL_defstash, tmpbuf, tmplen, FALSE);
521 gv2 = *(GV**)hv_fetch(PL_defstash, tmpbuf2, tmplen2, TRUE);
523 gv_init(gv2, PL_defstash, tmpbuf2, tmplen2, FALSE);
524 /* adjust ${"::_<newfilename"} to store the new file name */
525 GvSV(gv2) = newSVpvn(tmpbuf2 + 2, tmplen2 - 2);
526 GvHV(gv2) = (HV*)SvREFCNT_inc(GvHV(*gvp));
527 GvAV(gv2) = (AV*)SvREFCNT_inc(GvAV(*gvp));
529 if (tmpbuf != smallbuf) Safefree(tmpbuf);
530 if (tmpbuf2 != smallbuf2) Safefree(tmpbuf2);
534 /* second endif closes out the "are we 5.8.(8+)" conditional */
535 CopFILE_free(PL_curcop);
536 CopFILE_set(PL_curcop, s);
539 CopLINE_set(PL_curcop, atoi(n)-1);
543 takes: start position in buffer
544 keep_quoted preserve \ on the embedded delimiter(s)
545 keep_delims preserve the delimiters around the string
546 returns: position to continue reading from buffer
547 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
548 updates the read buffer.
550 This subroutine pulls a string out of the input. It is called for:
551 q single quotes q(literal text)
552 ' single quotes 'literal text'
553 qq double quotes qq(interpolate $here please)
554 " double quotes "interpolate $here please"
555 qx backticks qx(/bin/ls -l)
556 ` backticks `/bin/ls -l`
557 qw quote words @EXPORT_OK = qw( func() $spam )
558 m// regexp match m/this/
559 s/// regexp substitute s/this/that/
560 tr/// string transliterate tr/this/that/
561 y/// string transliterate y/this/that/
562 ($*@) sub prototypes sub foo ($)
563 (stuff) sub attr parameters sub foo : attr(stuff)
564 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
566 In most of these cases (all but <>, patterns and transliterate)
567 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
568 calls scan_str(). s/// makes yylex() call scan_subst() which calls
569 scan_str(). tr/// and y/// make yylex() call scan_trans() which
572 It skips whitespace before the string starts, and treats the first
573 character as the delimiter. If the delimiter is one of ([{< then
574 the corresponding "close" character )]}> is used as the closing
575 delimiter. It allows quoting of delimiters, and if the string has
576 balanced delimiters ([{<>}]) it allows nesting.
578 On success, the SV with the resulting string is put into lex_stuff or,
579 if that is already non-NULL, into lex_repl. The second case occurs only
580 when parsing the RHS of the special constructs s/// and tr/// (y///).
581 For convenience, the terminating delimiter character is stuffed into
586 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
588 SV *sv; /* scalar value: string */
589 char *tmps; /* temp string, used for delimiter matching */
590 register char *s = start; /* current position in the buffer */
591 register char term; /* terminating character */
592 register char *to; /* current position in the sv's data */
593 I32 brackets = 1; /* bracket nesting level */
594 bool has_utf8 = FALSE; /* is there any utf8 content? */
595 I32 termcode; /* terminating char. code */
596 /* 5.8.7+ uses UTF8_MAXBYTES but also its utf8.h defs _MAXLEN to it so
597 I'm reasonably hopeful this won't destroy anything (mst) */
598 U8 termstr[UTF8_MAXLEN]; /* terminating string */
599 STRLEN termlen; /* length of terminating string */
600 char *last = NULL; /* last position for nesting bracket */
602 /* skip space before the delimiter */
606 /* mark where we are, in case we need to report errors */
609 /* after skipping whitespace, the next character is the terminator */
612 termcode = termstr[0] = term;
616 termcode = utf8_to_uvchr((U8*)s, &termlen);
617 Copy(s, termstr, termlen, U8);
618 if (!UTF8_IS_INVARIANT(term))
622 /* mark where we are */
623 PL_multi_start = CopLINE(PL_curcop);
624 PL_multi_open = term;
626 /* find corresponding closing delimiter */
627 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
628 termcode = termstr[0] = term = tmps[5];
630 PL_multi_close = term;
632 /* create a new SV to hold the contents. 87 is leak category, I'm
633 assuming. 79 is the SV's initial length. What a random number. */
635 sv_upgrade(sv, SVt_PVIV);
636 SvIV_set(sv, termcode);
637 (void)SvPOK_only(sv); /* validate pointer */
639 /* move past delimiter and try to read a complete string */
641 sv_catpvn(sv, s, termlen);
644 if (PL_encoding && !UTF) {
648 int offset = s - SvPVX_const(PL_linestr);
649 const bool found = sv_cat_decode(sv, PL_encoding, PL_linestr,
650 &offset, (char*)termstr, termlen);
651 const char *ns = SvPVX_const(PL_linestr) + offset;
652 char *svlast = SvEND(sv) - 1;
654 for (; s < ns; s++) {
655 if (*s == '\n' && !PL_rsfp)
656 CopLINE_inc(PL_curcop);
661 /* handle quoted delimiters */
662 if (SvCUR(sv) > 1 && *(svlast-1) == '\\') {
664 for (t = svlast-2; t >= SvPVX_const(sv) && *t == '\\';)
666 if ((svlast-1 - t) % 2) {
670 SvCUR_set(sv, SvCUR(sv) - 1);
675 if (PL_multi_open == PL_multi_close) {
683 for (t = w = last; t < svlast; w++, t++) {
684 /* At here, all closes are "was quoted" one,
685 so we don't check PL_multi_close. */
687 if (!keep_quoted && *(t+1) == PL_multi_open)
692 else if (*t == PL_multi_open)
700 SvCUR_set(sv, w - SvPVX_const(sv));
709 SvCUR_set(sv, SvCUR(sv) - 1);
715 /* extend sv if need be */
716 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
717 /* set 'to' to the next character in the sv's string */
718 to = SvPVX(sv)+SvCUR(sv);
720 /* if open delimiter is the close delimiter read unbridle */
721 if (PL_multi_open == PL_multi_close) {
722 for (; s < PL_bufend; s++,to++) {
723 /* embedded newlines increment the current line number */
724 if (*s == '\n' && !PL_rsfp)
725 CopLINE_inc(PL_curcop);
726 /* handle quoted delimiters */
727 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
728 if (!keep_quoted && s[1] == term)
730 /* any other quotes are simply copied straight through */
734 /* terminate when run out of buffer (the for() condition), or
735 have found the terminator */
736 else if (*s == term) {
739 if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen))
742 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
748 /* if the terminator isn't the same as the start character (e.g.,
749 matched brackets), we have to allow more in the quoting, and
750 be prepared for nested brackets.
753 /* read until we run out of string, or we find the terminator */
754 for (; s < PL_bufend; s++,to++) {
755 /* embedded newlines increment the line count */
756 if (*s == '\n' && !PL_rsfp)
757 CopLINE_inc(PL_curcop);
758 /* backslashes can escape the open or closing characters */
759 if (*s == '\\' && s+1 < PL_bufend) {
761 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
766 /* allow nested opens and closes */
767 else if (*s == PL_multi_close && --brackets <= 0)
769 else if (*s == PL_multi_open)
771 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
776 /* terminate the copied string and update the sv's end-of-string */
778 SvCUR_set(sv, to - SvPVX_const(sv));
781 * this next chunk reads more into the buffer if we're not done yet
785 break; /* handle case where we are done yet :-) */
787 #ifndef PERL_STRICT_CR
788 if (to - SvPVX_const(sv) >= 2) {
789 if ((to[-2] == '\r' && to[-1] == '\n') ||
790 (to[-2] == '\n' && to[-1] == '\r'))
794 SvCUR_set(sv, to - SvPVX_const(sv));
796 else if (to[-1] == '\r')
799 else if (to - SvPVX_const(sv) == 1 && to[-1] == '\r')
804 /* if we're out of file, or a read fails, bail and reset the current
805 line marker so we can report where the unterminated string began
808 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
810 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
813 /* we read a line, so increment our line counter */
814 CopLINE_inc(PL_curcop);
816 /* update debugger info */
817 if (PERLDB_LINE && PL_curstash != PL_debstash) {
818 SV *sv = NEWSV(88,0);
820 sv_upgrade(sv, SVt_PVMG);
821 sv_setsv(sv,PL_linestr);
824 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
827 /* having changed the buffer, we must update PL_bufend */
828 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
829 PL_last_lop = PL_last_uni = Nullch;
832 /* at this point, we have successfully read the delimited string */
834 if (!PL_encoding || UTF) {
836 sv_catpvn(sv, s, termlen);
839 if (has_utf8 || PL_encoding)
842 PL_multi_end = CopLINE(PL_curcop);
844 /* if we allocated too much space, give some back */
845 if (SvCUR(sv) + 5 < SvLEN(sv)) {
846 SvLEN_set(sv, SvCUR(sv) + 1);
847 /* 5.8.8 uses SvPV_renew, no prior version actually has the damn thing (mst) */
848 #ifdef PERL_5_8_8_PLUS
849 SvPV_renew(sv, SvLEN(sv));
851 Renew(SvPVX(sv), SvLEN(sv), char);
855 /* decide whether this is the first or second quoted string we've read
868 * When the lexer realizes it knows the next token (for instance,
869 * it is reordering tokens for the parser) then it can call S_force_next
870 * to know what token to return the next time the lexer is called. Caller
871 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
872 * handles the token correctly.
876 S_force_next(pTHX_ I32 type)
881 start_force(PL_lasttoke);
882 PL_nexttoke[PL_curforce].next_type = type;
883 if (PL_lex_state != LEX_KNOWNEXT)
884 PL_lex_defer = PL_lex_state;
885 PL_lex_state = LEX_KNOWNEXT;
886 PL_lex_expect = PL_expect;
889 PL_nexttype[PL_nexttoke] = type;
891 if (PL_lex_state != LEX_KNOWNEXT) {
892 PL_lex_defer = PL_lex_state;
893 PL_lex_expect = PL_expect;
894 PL_lex_state = LEX_KNOWNEXT;
899 #define XFAKEBRACK 128
902 S_scan_ident(pTHX_ register char *s, register const char *send, char *dest, STRLEN destlen, I32 ck_uni)
906 char *bracket = Nullch;
912 e = d + destlen - 3; /* two-character token, ending NUL */
914 while (isDIGIT(*s)) {
916 Perl_croak(aTHX_ ident_too_long);
923 Perl_croak(aTHX_ ident_too_long);
924 if (isALNUM(*s)) /* UTF handled below */
926 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
931 else if (*s == ':' && s[1] == ':') {
935 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
936 char *t = s + UTF8SKIP(s);
937 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
940 Perl_croak(aTHX_ ident_too_long);
941 Copy(s, d, t - s, char);
952 if (PL_lex_state != LEX_NORMAL)
953 PL_lex_state = LEX_INTERPENDMAYBE;
956 if (*s == '$' && s[1] &&
957 (isALNUM_lazy_if(s+1,UTF) || s[1] == '$' || s[1] == '{' || strnEQ(s+1,"::",2)) )
965 /* we always call this with ck_uni == 0 (rafl) */
973 if (*d == '^' && *s && isCONTROLVAR(*s)) {
978 if (isSPACE(s[-1])) {
980 const char ch = *s++;
981 if (!SPACE_OR_TAB(ch)) {
987 if (isIDFIRST_lazy_if(d,UTF)) {
991 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
993 while (e < send && UTF8_IS_CONTINUED(*e) && is_utf8_mark((U8*)e))
996 Copy(s, d, e - s, char);
1001 while ((isALNUM(*s) || *s == ':') && d < e)
1004 Perl_croak(aTHX_ ident_too_long);
1007 while (s < send && SPACE_OR_TAB(*s)) s++;
1008 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
1009 /* we don't want perl to guess what is meant. the keyword
1010 * parser decides that later. (rafl)
1013 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
1014 const char *brack = *s == '[' ? "[...]" : "{...}";
1015 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
1016 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
1017 funny, dest, brack, funny, dest, brack);
1021 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
1025 /* Handle extended ${^Foo} variables
1026 * 1999-02-27 mjd-perl-patch@plover.com */
1027 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
1031 while (isALNUM(*s) && d < e) {
1035 Perl_croak(aTHX_ ident_too_long);
1040 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets) {
1041 PL_lex_state = LEX_INTERPEND;
1046 /* we don't want perl to guess what is meant. the keyword
1047 * parser decides that later. (rafl)
1050 if (PL_lex_state == LEX_NORMAL) {
1051 if (ckWARN(WARN_AMBIGUOUS) &&
1052 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
1054 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
1055 "Ambiguous use of %c{%s} resolved to %c%s",
1056 funny, dest, funny, dest);
1062 s = bracket; /* let the parser handle it */
1066 /* don't intuit. we really just want the string. (rafl) */
1068 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
1069 PL_lex_state = LEX_INTERPEND;