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 DPTR2FPTR(t,p) ((t)PTR2nat(p)) /* data pointer to function pointer */
23 #define FPTR2DPTR(t,p) ((t)PTR2nat(p)) /* function pointer to data pointer */
24 #define PTR2nat(p) (PTRV)(p) /* pointer to integer of PTRSIZE */
25 #define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
26 #define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)safemalloc((MEM_SIZE)((n)*sizeof(t)))))
27 #define SvPVX_const(sv) ((const char*) (0 + SvPVX(sv)))
28 #define SvPV_renew(sv,n) \
29 STMT_START { SvLEN_set(sv, n); \
30 SvPV_set((sv), (MEM_WRAP_CHECK_(n,char) \
31 (char*)saferealloc((Malloc_t)SvPVX(sv), \
35 /* On MacOS, respect nonbreaking spaces */
36 #ifdef MACOS_TRADITIONAL
37 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
39 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
42 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
44 #define LEX_NORMAL 10 /* normal code (ie not within "...") */
45 #define LEX_INTERPNORMAL 9 /* code within a string, eg "$foo[$x+1]" */
46 #define LEX_INTERPCASEMOD 8 /* expecting a \U, \Q or \E etc */
47 #define LEX_INTERPPUSH 7 /* starting a new sublex parse level */
48 #define LEX_INTERPSTART 6 /* expecting the start of a $var */
50 /* at end of code, eg "$x" followed by: */
51 #define LEX_INTERPEND 5 /* ... eg not one of [, { or -> */
52 #define LEX_INTERPENDMAYBE 4 /* ... eg one of [, { or -> */
54 #define LEX_INTERPCONCAT 3 /* expecting anything, eg at start of
55 string or after \E, $foo, etc */
56 #define LEX_INTERPCONST 2 /* NOT USED */
57 #define LEX_FORMLINE 1 /* expecting a format line */
58 #define LEX_KNOWNEXT 0 /* next token known; just return it */
60 /* and this one is my own madness (mst) */
62 #if PERL_REVISION == 5 && PERL_VERSION == 8 && PERL_SUBVERSION >= 8
63 #define PERL_5_8_8_PLUS
66 /* and now we're back to the toke.c stuff again (mst) */
68 static const char ident_too_long[] =
69 "Identifier too long";
70 static const char c_without_g[] =
71 "Use of /c modifier is meaningless without /g";
72 static const char c_in_subst[] =
73 "Use of /c modifier is meaningless in s///";
75 #ifdef USE_UTF8_SCRIPTS
76 # define UTF (!IN_BYTES)
78 # define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || (PL_hints & HINT_UTF8))
81 /* Invoke the idxth filter function for the current rsfp. */
82 /* maxlen 0 = read one text line */
84 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
91 if (idx > AvFILLp(PL_rsfp_filters)) { /* Any more filters? */
92 /* Provide a default input filter to make life easy. */
93 /* Note that we append to the line. This is handy. */
94 DEBUG_P(PerlIO_printf(Perl_debug_log,
95 "filter_read %d: from rsfp\n", idx));
99 const int old_len = SvCUR(buf_sv);
101 /* ensure buf_sv is large enough */
102 SvGROW(buf_sv, (STRLEN)(old_len + maxlen)) ;
103 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
104 if (PerlIO_error(PL_rsfp))
105 return -1; /* error */
107 return 0 ; /* end of file */
109 SvCUR_set(buf_sv, old_len + len) ;
112 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
113 if (PerlIO_error(PL_rsfp))
114 return -1; /* error */
116 return 0 ; /* end of file */
119 return SvCUR(buf_sv);
121 /* Skip this filter slot if filter has been deleted */
122 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef) {
123 DEBUG_P(PerlIO_printf(Perl_debug_log,
124 "filter_read %d: skipped (filter deleted)\n",
126 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
128 /* Get function pointer hidden within datasv */
129 funcp = DPTR2FPTR(filter_t, IoANY(datasv));
130 DEBUG_P(PerlIO_printf(Perl_debug_log,
131 "filter_read %d: via function %p (%s)\n",
132 idx, datasv, SvPV_nolen_const(datasv)));
133 /* Call function. The function is expected to */
134 /* call "FILTER_READ(idx+1, buf_sv)" first. */
135 /* Return: <0:error, =0:eof, >0:not eof */
136 return (*funcp)(aTHX_ idx, buf_sv, maxlen);
140 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
142 #ifdef PERL_CR_FILTER
143 if (!PL_rsfp_filters) {
144 filter_add(S_cr_textfilter,NULL);
147 if (PL_rsfp_filters) {
149 SvCUR_set(sv, 0); /* start with empty line */
150 if (FILTER_READ(0, sv, 0) > 0)
151 return ( SvPVX(sv) ) ;
156 return (sv_gets(sv, fp, append));
161 * Called to gobble the appropriate amount and type of whitespace.
162 * Skips comments as well.
166 S_skipspace(pTHX_ register char *s)
168 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
169 while (s < PL_bufend && SPACE_OR_TAB(*s))
175 SSize_t oldprevlen, oldoldprevlen;
176 SSize_t oldloplen = 0, oldunilen = 0;
177 while (s < PL_bufend && isSPACE(*s)) {
178 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
183 if (s < PL_bufend && *s == '#') {
184 while (s < PL_bufend && *s != '\n')
188 if (PL_in_eval && !PL_rsfp) {
195 /* only continue to recharge the buffer if we're at the end
196 * of the buffer, we're not reading from a source filter, and
197 * we're in normal lexing mode
199 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
200 PL_lex_state == LEX_FORMLINE)
203 /* try to recharge the buffer */
204 if ((s = filter_gets(PL_linestr, PL_rsfp,
205 (prevlen = SvCUR(PL_linestr)))) == Nullch)
207 /* end of file. Add on the -p or -n magic */
210 ";}continue{print or die qq(-p destination: $!\\n);}");
211 PL_minus_n = PL_minus_p = 0;
213 else if (PL_minus_n) {
214 sv_setpvn(PL_linestr, ";}", 2);
218 sv_setpvn(PL_linestr,";", 1);
220 /* reset variables for next time we lex */
221 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
223 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
224 PL_last_lop = PL_last_uni = Nullch;
226 /* Close the filehandle. Could be from -P preprocessor,
227 * STDIN, or a regular file. If we were reading code from
228 * STDIN (because the commandline held no -e or filename)
229 * then we don't close it, we reset it so the code can
230 * read from STDIN too.
233 if (PL_preprocess && !PL_in_eval)
234 (void)PerlProc_pclose(PL_rsfp);
235 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
236 PerlIO_clearerr(PL_rsfp);
238 (void)PerlIO_close(PL_rsfp);
243 /* not at end of file, so we only read another line */
244 /* make corresponding updates to old pointers, for yyerror() */
245 oldprevlen = PL_oldbufptr - PL_bufend;
246 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
248 oldunilen = PL_last_uni - PL_bufend;
250 oldloplen = PL_last_lop - PL_bufend;
251 PL_linestart = PL_bufptr = s + prevlen;
252 PL_bufend = s + SvCUR(PL_linestr);
254 PL_oldbufptr = s + oldprevlen;
255 PL_oldoldbufptr = s + oldoldprevlen;
257 PL_last_uni = s + oldunilen;
259 PL_last_lop = s + oldloplen;
262 /* debugger active and we're not compiling the debugger code,
263 * so store the line into the debugger's array of lines
265 if (PERLDB_LINE && PL_curstash != PL_debstash) {
266 SV * const sv = NEWSV(85,0);
268 sv_upgrade(sv, SVt_PVMG);
269 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
272 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
278 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
280 register char *d = dest;
281 register char * const e = d + destlen - 3; /* two-character token, ending NUL */
284 Perl_croak(aTHX_ ident_too_long);
285 if (isALNUM(*s)) /* UTF handled below */
287 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
292 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
296 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
297 char *t = s + UTF8SKIP(s);
298 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
301 Perl_croak(aTHX_ ident_too_long);
302 Copy(s, d, t - s, char);
316 * This subroutine has nothing to do with tilting, whether at windmills
317 * or pinball tables. Its name is short for "increment line". It
318 * increments the current line number in CopLINE(PL_curcop) and checks
319 * to see whether the line starts with a comment of the form
320 * # line 500 "foo.pm"
321 * If so, it sets the current line number and file to the values in the comment.
325 S_incline(pTHX_ char *s)
332 CopLINE_inc(PL_curcop);
335 while (SPACE_OR_TAB(*s)) s++;
336 if (strnEQ(s, "line", 4))
340 if (SPACE_OR_TAB(*s))
344 while (SPACE_OR_TAB(*s)) s++;
350 while (SPACE_OR_TAB(*s))
352 if (*s == '"' && (t = strchr(s+1, '"'))) {
357 for (t = s; !isSPACE(*t); t++) ;
360 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
362 if (*e != '\n' && *e != '\0')
363 return; /* false alarm */
368 /* this chunk was added to S_incline during 5.8.8. I don't know why but I don't
369 honestly care since I probably want to be bug-compatible anyway (mst) */
371 /* ... my kingdom for a perl parser in perl ... (mst) */
373 #ifdef PERL_5_8_8_PLUS
375 const char *cf = CopFILE(PL_curcop);
376 if (cf && strlen(cf) > 7 && strnEQ(cf, "(eval ", 6)) {
377 /* must copy *{"::_<(eval N)[oldfilename:L]"}
378 * to *{"::_<newfilename"} */
379 char smallbuf[256], smallbuf2[256];
380 char *tmpbuf, *tmpbuf2;
382 STRLEN tmplen = strlen(cf);
383 STRLEN tmplen2 = strlen(s);
384 if (tmplen + 3 < sizeof smallbuf)
387 Newx(tmpbuf, tmplen + 3, char);
388 if (tmplen2 + 3 < sizeof smallbuf2)
391 Newx(tmpbuf2, tmplen2 + 3, char);
392 tmpbuf[0] = tmpbuf2[0] = '_';
393 tmpbuf[1] = tmpbuf2[1] = '<';
394 memcpy(tmpbuf + 2, cf, ++tmplen);
395 memcpy(tmpbuf2 + 2, s, ++tmplen2);
397 gvp = (GV**)hv_fetch(PL_defstash, tmpbuf, tmplen, FALSE);
399 gv2 = *(GV**)hv_fetch(PL_defstash, tmpbuf2, tmplen2, TRUE);
401 gv_init(gv2, PL_defstash, tmpbuf2, tmplen2, FALSE);
402 /* adjust ${"::_<newfilename"} to store the new file name */
403 GvSV(gv2) = newSVpvn(tmpbuf2 + 2, tmplen2 - 2);
404 GvHV(gv2) = (HV*)SvREFCNT_inc(GvHV(*gvp));
405 GvAV(gv2) = (AV*)SvREFCNT_inc(GvAV(*gvp));
407 if (tmpbuf != smallbuf) Safefree(tmpbuf);
408 if (tmpbuf2 != smallbuf2) Safefree(tmpbuf2);
412 /* second endif closes out the "are we 5.8.(8+)" conditional */
413 CopFILE_free(PL_curcop);
414 CopFILE_set(PL_curcop, s);
417 CopLINE_set(PL_curcop, atoi(n)-1);
421 takes: start position in buffer
422 keep_quoted preserve \ on the embedded delimiter(s)
423 keep_delims preserve the delimiters around the string
424 returns: position to continue reading from buffer
425 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
426 updates the read buffer.
428 This subroutine pulls a string out of the input. It is called for:
429 q single quotes q(literal text)
430 ' single quotes 'literal text'
431 qq double quotes qq(interpolate $here please)
432 " double quotes "interpolate $here please"
433 qx backticks qx(/bin/ls -l)
434 ` backticks `/bin/ls -l`
435 qw quote words @EXPORT_OK = qw( func() $spam )
436 m// regexp match m/this/
437 s/// regexp substitute s/this/that/
438 tr/// string transliterate tr/this/that/
439 y/// string transliterate y/this/that/
440 ($*@) sub prototypes sub foo ($)
441 (stuff) sub attr parameters sub foo : attr(stuff)
442 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
444 In most of these cases (all but <>, patterns and transliterate)
445 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
446 calls scan_str(). s/// makes yylex() call scan_subst() which calls
447 scan_str(). tr/// and y/// make yylex() call scan_trans() which
450 It skips whitespace before the string starts, and treats the first
451 character as the delimiter. If the delimiter is one of ([{< then
452 the corresponding "close" character )]}> is used as the closing
453 delimiter. It allows quoting of delimiters, and if the string has
454 balanced delimiters ([{<>}]) it allows nesting.
456 On success, the SV with the resulting string is put into lex_stuff or,
457 if that is already non-NULL, into lex_repl. The second case occurs only
458 when parsing the RHS of the special constructs s/// and tr/// (y///).
459 For convenience, the terminating delimiter character is stuffed into
464 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
466 SV *sv; /* scalar value: string */
467 char *tmps; /* temp string, used for delimiter matching */
468 register char *s = start; /* current position in the buffer */
469 register char term; /* terminating character */
470 register char *to; /* current position in the sv's data */
471 I32 brackets = 1; /* bracket nesting level */
472 bool has_utf8 = FALSE; /* is there any utf8 content? */
473 I32 termcode; /* terminating char. code */
474 /* 5.8.7+ uses UTF8_MAXBYTES but also its utf8.h defs _MAXLEN to it so
475 I'm reasonably hopeful this won't destroy anything (mst) */
476 U8 termstr[UTF8_MAXLEN]; /* terminating string */
477 STRLEN termlen; /* length of terminating string */
478 char *last = NULL; /* last position for nesting bracket */
480 /* skip space before the delimiter */
484 /* mark where we are, in case we need to report errors */
487 /* after skipping whitespace, the next character is the terminator */
490 termcode = termstr[0] = term;
494 termcode = utf8_to_uvchr((U8*)s, &termlen);
495 Copy(s, termstr, termlen, U8);
496 if (!UTF8_IS_INVARIANT(term))
500 /* mark where we are */
501 PL_multi_start = CopLINE(PL_curcop);
502 PL_multi_open = term;
504 /* find corresponding closing delimiter */
505 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
506 termcode = termstr[0] = term = tmps[5];
508 PL_multi_close = term;
510 /* create a new SV to hold the contents. 87 is leak category, I'm
511 assuming. 79 is the SV's initial length. What a random number. */
513 sv_upgrade(sv, SVt_PVIV);
514 SvIV_set(sv, termcode);
515 (void)SvPOK_only(sv); /* validate pointer */
517 /* move past delimiter and try to read a complete string */
519 sv_catpvn(sv, s, termlen);
522 if (PL_encoding && !UTF) {
526 int offset = s - SvPVX_const(PL_linestr);
527 const bool found = sv_cat_decode(sv, PL_encoding, PL_linestr,
528 &offset, (char*)termstr, termlen);
529 const char *ns = SvPVX_const(PL_linestr) + offset;
530 char *svlast = SvEND(sv) - 1;
532 for (; s < ns; s++) {
533 if (*s == '\n' && !PL_rsfp)
534 CopLINE_inc(PL_curcop);
539 /* handle quoted delimiters */
540 if (SvCUR(sv) > 1 && *(svlast-1) == '\\') {
542 for (t = svlast-2; t >= SvPVX_const(sv) && *t == '\\';)
544 if ((svlast-1 - t) % 2) {
548 SvCUR_set(sv, SvCUR(sv) - 1);
553 if (PL_multi_open == PL_multi_close) {
561 for (t = w = last; t < svlast; w++, t++) {
562 /* At here, all closes are "was quoted" one,
563 so we don't check PL_multi_close. */
565 if (!keep_quoted && *(t+1) == PL_multi_open)
570 else if (*t == PL_multi_open)
578 SvCUR_set(sv, w - SvPVX_const(sv));
587 SvCUR_set(sv, SvCUR(sv) - 1);
593 /* extend sv if need be */
594 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
595 /* set 'to' to the next character in the sv's string */
596 to = SvPVX(sv)+SvCUR(sv);
598 /* if open delimiter is the close delimiter read unbridle */
599 if (PL_multi_open == PL_multi_close) {
600 for (; s < PL_bufend; s++,to++) {
601 /* embedded newlines increment the current line number */
602 if (*s == '\n' && !PL_rsfp)
603 CopLINE_inc(PL_curcop);
604 /* handle quoted delimiters */
605 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
606 if (!keep_quoted && s[1] == term)
608 /* any other quotes are simply copied straight through */
612 /* terminate when run out of buffer (the for() condition), or
613 have found the terminator */
614 else if (*s == term) {
617 if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen))
620 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
626 /* if the terminator isn't the same as the start character (e.g.,
627 matched brackets), we have to allow more in the quoting, and
628 be prepared for nested brackets.
631 /* read until we run out of string, or we find the terminator */
632 for (; s < PL_bufend; s++,to++) {
633 /* embedded newlines increment the line count */
634 if (*s == '\n' && !PL_rsfp)
635 CopLINE_inc(PL_curcop);
636 /* backslashes can escape the open or closing characters */
637 if (*s == '\\' && s+1 < PL_bufend) {
639 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
644 /* allow nested opens and closes */
645 else if (*s == PL_multi_close && --brackets <= 0)
647 else if (*s == PL_multi_open)
649 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
654 /* terminate the copied string and update the sv's end-of-string */
656 SvCUR_set(sv, to - SvPVX_const(sv));
659 * this next chunk reads more into the buffer if we're not done yet
663 break; /* handle case where we are done yet :-) */
665 #ifndef PERL_STRICT_CR
666 if (to - SvPVX_const(sv) >= 2) {
667 if ((to[-2] == '\r' && to[-1] == '\n') ||
668 (to[-2] == '\n' && to[-1] == '\r'))
672 SvCUR_set(sv, to - SvPVX_const(sv));
674 else if (to[-1] == '\r')
677 else if (to - SvPVX_const(sv) == 1 && to[-1] == '\r')
682 /* if we're out of file, or a read fails, bail and reset the current
683 line marker so we can report where the unterminated string began
686 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
688 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
691 /* we read a line, so increment our line counter */
692 CopLINE_inc(PL_curcop);
694 /* update debugger info */
695 if (PERLDB_LINE && PL_curstash != PL_debstash) {
696 SV *sv = NEWSV(88,0);
698 sv_upgrade(sv, SVt_PVMG);
699 sv_setsv(sv,PL_linestr);
702 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
705 /* having changed the buffer, we must update PL_bufend */
706 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
707 PL_last_lop = PL_last_uni = Nullch;
710 /* at this point, we have successfully read the delimited string */
712 if (!PL_encoding || UTF) {
714 sv_catpvn(sv, s, termlen);
717 if (has_utf8 || PL_encoding)
720 PL_multi_end = CopLINE(PL_curcop);
722 /* if we allocated too much space, give some back */
723 if (SvCUR(sv) + 5 < SvLEN(sv)) {
724 SvLEN_set(sv, SvCUR(sv) + 1);
725 /* 5.8.8 uses SvPV_renew, no prior version actually has the damn thing (mst) */
726 #ifdef PERL_5_8_8_PLUS
727 SvPV_renew(sv, SvLEN(sv));
729 Renew(SvPVX(sv), SvLEN(sv), char);
733 /* decide whether this is the first or second quoted string we've read
746 * When the lexer realizes it knows the next token (for instance,
747 * it is reordering tokens for the parser) then it can call S_force_next
748 * to know what token to return the next time the lexer is called. Caller
749 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
750 * handles the token correctly.
754 S_force_next(pTHX_ I32 type)
756 PL_nexttype[PL_nexttoke] = type;
758 if (PL_lex_state != LEX_KNOWNEXT) {
759 PL_lex_defer = PL_lex_state;
760 PL_lex_expect = PL_expect;
761 PL_lex_state = LEX_KNOWNEXT;