Commit | Line | Data |
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 | |
28 | STATIC void S_incline(pTHX_ char *s); |
29 | STATIC char* S_skipspace(pTHX_ char *s); |
30 | STATIC char * S_filter_gets(pTHX_ SV *sv, PerlIO *fp, STRLEN append); |
31 | STATIC char* S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims); |
32 | STATIC 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 | |
e38f5a74 |
86 | #if PERL_REVISION == 5 && PERL_VERSION > 8 |
87 | #define PERL_5_9_PLUS |
88 | #endif |
89 | |
90 | #ifdef PERL_5_9_PLUS |
14ab3472 |
91 | /* 5.9+ moves a bunch of things to a PL_parser struct so we need to |
92 | declare the backcompat macros for things to still work (mst) */ |
93 | |
94 | /* XXX temporary backwards compatibility */ |
95 | #define PL_lex_brackets (PL_parser->lex_brackets) |
96 | #define PL_lex_brackstack (PL_parser->lex_brackstack) |
97 | #define PL_lex_casemods (PL_parser->lex_casemods) |
98 | #define PL_lex_casestack (PL_parser->lex_casestack) |
99 | #define PL_lex_defer (PL_parser->lex_defer) |
100 | #define PL_lex_dojoin (PL_parser->lex_dojoin) |
101 | #define PL_lex_expect (PL_parser->lex_expect) |
102 | #define PL_lex_formbrack (PL_parser->lex_formbrack) |
103 | #define PL_lex_inpat (PL_parser->lex_inpat) |
104 | #define PL_lex_inwhat (PL_parser->lex_inwhat) |
105 | #define PL_lex_op (PL_parser->lex_op) |
106 | #define PL_lex_repl (PL_parser->lex_repl) |
107 | #define PL_lex_starts (PL_parser->lex_starts) |
108 | #define PL_lex_stuff (PL_parser->lex_stuff) |
109 | #define PL_multi_start (PL_parser->multi_start) |
110 | #define PL_multi_open (PL_parser->multi_open) |
111 | #define PL_multi_close (PL_parser->multi_close) |
112 | #define PL_pending_ident (PL_parser->pending_ident) |
113 | #define PL_preambled (PL_parser->preambled) |
114 | #define PL_sublex_info (PL_parser->sublex_info) |
115 | #define PL_linestr (PL_parser->linestr) |
116 | #define PL_sublex_info (PL_parser->sublex_info) |
117 | #define PL_linestr (PL_parser->linestr) |
118 | #define PL_expect (PL_parser->expect) |
119 | #define PL_copline (PL_parser->copline) |
120 | #define PL_bufptr (PL_parser->bufptr) |
121 | #define PL_oldbufptr (PL_parser->oldbufptr) |
122 | #define PL_oldoldbufptr (PL_parser->oldoldbufptr) |
123 | #define PL_linestart (PL_parser->linestart) |
124 | #define PL_bufend (PL_parser->bufend) |
125 | #define PL_last_uni (PL_parser->last_uni) |
126 | #define PL_last_lop (PL_parser->last_lop) |
127 | #define PL_last_lop_op (PL_parser->last_lop_op) |
128 | #define PL_lex_state (PL_parser->lex_state) |
129 | #define PL_rsfp (PL_parser->rsfp) |
130 | #define PL_rsfp_filters (PL_parser->rsfp_filters) |
131 | #define PL_in_my (PL_parser->in_my) |
132 | #define PL_in_my_stash (PL_parser->in_my_stash) |
133 | #define PL_tokenbuf (PL_parser->tokenbuf) |
134 | #define PL_multi_end (PL_parser->multi_end) |
135 | #define PL_error_count (PL_parser->error_count) |
136 | /* these three are from the non-PERL_MAD path but I don't -think- I need |
137 | the PERL_MAD stuff since my code isn't really populating things (mst) */ |
138 | # define PL_nexttoke (PL_parser->nexttoke) |
139 | # define PL_nexttype (PL_parser->nexttype) |
140 | # define PL_nextval (PL_parser->nextval) |
141 | /* end of backcompat macros form 5.9 toke.c (mst) */ |
78160085 |
142 | #endif |
14ab3472 |
143 | |
144 | /* when ccflags include -DDEBUGGING we need this for earlier 5.8 perls */ |
145 | #ifndef SvPV_nolen_const |
146 | #define SvPV_nolen_const SvPV_nolen |
147 | #endif |
148 | |
71511651 |
149 | /* and now we're back to the toke.c stuff again (mst) */ |
150 | |
e807ee50 |
151 | static const char ident_too_long[] = |
152 | "Identifier too long"; |
153 | static const char c_without_g[] = |
154 | "Use of /c modifier is meaningless without /g"; |
155 | static const char c_in_subst[] = |
156 | "Use of /c modifier is meaningless in s///"; |
157 | |
158 | #ifdef USE_UTF8_SCRIPTS |
159 | # define UTF (!IN_BYTES) |
160 | #else |
161 | # define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || (PL_hints & HINT_UTF8)) |
162 | #endif |
163 | |
164 | /* Invoke the idxth filter function for the current rsfp. */ |
165 | /* maxlen 0 = read one text line */ |
166 | I32 |
167 | Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen) |
168 | { |
169 | filter_t funcp; |
170 | SV *datasv = NULL; |
171 | |
172 | if (!PL_rsfp_filters) |
173 | return -1; |
174 | if (idx > AvFILLp(PL_rsfp_filters)) { /* Any more filters? */ |
175 | /* Provide a default input filter to make life easy. */ |
176 | /* Note that we append to the line. This is handy. */ |
177 | DEBUG_P(PerlIO_printf(Perl_debug_log, |
178 | "filter_read %d: from rsfp\n", idx)); |
179 | if (maxlen) { |
180 | /* Want a block */ |
181 | int len ; |
182 | const int old_len = SvCUR(buf_sv); |
183 | |
184 | /* ensure buf_sv is large enough */ |
185 | SvGROW(buf_sv, (STRLEN)(old_len + maxlen)) ; |
186 | if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){ |
187 | if (PerlIO_error(PL_rsfp)) |
188 | return -1; /* error */ |
189 | else |
190 | return 0 ; /* end of file */ |
191 | } |
192 | SvCUR_set(buf_sv, old_len + len) ; |
193 | } else { |
194 | /* Want a line */ |
195 | if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) { |
196 | if (PerlIO_error(PL_rsfp)) |
197 | return -1; /* error */ |
198 | else |
199 | return 0 ; /* end of file */ |
200 | } |
201 | } |
202 | return SvCUR(buf_sv); |
203 | } |
204 | /* Skip this filter slot if filter has been deleted */ |
205 | if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef) { |
206 | DEBUG_P(PerlIO_printf(Perl_debug_log, |
207 | "filter_read %d: skipped (filter deleted)\n", |
208 | idx)); |
209 | return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */ |
210 | } |
211 | /* Get function pointer hidden within datasv */ |
212 | funcp = DPTR2FPTR(filter_t, IoANY(datasv)); |
213 | DEBUG_P(PerlIO_printf(Perl_debug_log, |
214 | "filter_read %d: via function %p (%s)\n", |
215 | idx, datasv, SvPV_nolen_const(datasv))); |
216 | /* Call function. The function is expected to */ |
217 | /* call "FILTER_READ(idx+1, buf_sv)" first. */ |
218 | /* Return: <0:error, =0:eof, >0:not eof */ |
219 | return (*funcp)(aTHX_ idx, buf_sv, maxlen); |
220 | } |
221 | |
222 | STATIC char * |
223 | S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append) |
224 | { |
225 | #ifdef PERL_CR_FILTER |
226 | if (!PL_rsfp_filters) { |
227 | filter_add(S_cr_textfilter,NULL); |
228 | } |
229 | #endif |
230 | if (PL_rsfp_filters) { |
231 | if (!append) |
232 | SvCUR_set(sv, 0); /* start with empty line */ |
233 | if (FILTER_READ(0, sv, 0) > 0) |
234 | return ( SvPVX(sv) ) ; |
235 | else |
236 | return Nullch ; |
237 | } |
238 | else |
239 | return (sv_gets(sv, fp, append)); |
240 | } |
241 | |
242 | /* |
243 | * S_skipspace |
244 | * Called to gobble the appropriate amount and type of whitespace. |
245 | * Skips comments as well. |
246 | */ |
247 | |
248 | STATIC char * |
249 | S_skipspace(pTHX_ register char *s) |
250 | { |
251 | if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) { |
252 | while (s < PL_bufend && SPACE_OR_TAB(*s)) |
253 | s++; |
254 | return s; |
255 | } |
256 | for (;;) { |
257 | STRLEN prevlen; |
258 | SSize_t oldprevlen, oldoldprevlen; |
259 | SSize_t oldloplen = 0, oldunilen = 0; |
260 | while (s < PL_bufend && isSPACE(*s)) { |
261 | if (*s++ == '\n' && PL_in_eval && !PL_rsfp) |
262 | incline(s); |
263 | } |
264 | |
265 | /* comment */ |
266 | if (s < PL_bufend && *s == '#') { |
267 | while (s < PL_bufend && *s != '\n') |
268 | s++; |
269 | if (s < PL_bufend) { |
270 | s++; |
271 | if (PL_in_eval && !PL_rsfp) { |
272 | incline(s); |
273 | continue; |
274 | } |
275 | } |
276 | } |
277 | |
278 | /* only continue to recharge the buffer if we're at the end |
279 | * of the buffer, we're not reading from a source filter, and |
280 | * we're in normal lexing mode |
281 | */ |
282 | if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat || |
283 | PL_lex_state == LEX_FORMLINE) |
284 | return s; |
285 | |
286 | /* try to recharge the buffer */ |
287 | if ((s = filter_gets(PL_linestr, PL_rsfp, |
288 | (prevlen = SvCUR(PL_linestr)))) == Nullch) |
289 | { |
290 | /* end of file. Add on the -p or -n magic */ |
291 | if (PL_minus_p) { |
292 | sv_setpv(PL_linestr, |
293 | ";}continue{print or die qq(-p destination: $!\\n);}"); |
294 | PL_minus_n = PL_minus_p = 0; |
295 | } |
296 | else if (PL_minus_n) { |
297 | sv_setpvn(PL_linestr, ";}", 2); |
298 | PL_minus_n = 0; |
299 | } |
300 | else |
301 | sv_setpvn(PL_linestr,";", 1); |
302 | |
303 | /* reset variables for next time we lex */ |
304 | PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart |
305 | = SvPVX(PL_linestr); |
306 | PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); |
307 | PL_last_lop = PL_last_uni = Nullch; |
308 | |
35fc8ada |
309 | /* In perl versions previous to p4-rawid: //depot/perl@32954 -P |
310 | * preprocessors were supported here. We don't support -P at all, even |
311 | * on perls that support it, and use the following chunk from blead |
312 | * perl. (rafl) |
313 | */ |
314 | |
315 | /* Close the filehandle. Could be from |
e807ee50 |
316 | * STDIN, or a regular file. If we were reading code from |
317 | * STDIN (because the commandline held no -e or filename) |
318 | * then we don't close it, we reset it so the code can |
319 | * read from STDIN too. |
320 | */ |
321 | |
35fc8ada |
322 | if ((PerlIO*)PL_rsfp == PerlIO_stdin()) |
e807ee50 |
323 | PerlIO_clearerr(PL_rsfp); |
324 | else |
325 | (void)PerlIO_close(PL_rsfp); |
326 | PL_rsfp = Nullfp; |
327 | return s; |
328 | } |
329 | |
330 | /* not at end of file, so we only read another line */ |
331 | /* make corresponding updates to old pointers, for yyerror() */ |
332 | oldprevlen = PL_oldbufptr - PL_bufend; |
333 | oldoldprevlen = PL_oldoldbufptr - PL_bufend; |
334 | if (PL_last_uni) |
335 | oldunilen = PL_last_uni - PL_bufend; |
336 | if (PL_last_lop) |
337 | oldloplen = PL_last_lop - PL_bufend; |
338 | PL_linestart = PL_bufptr = s + prevlen; |
339 | PL_bufend = s + SvCUR(PL_linestr); |
340 | s = PL_bufptr; |
341 | PL_oldbufptr = s + oldprevlen; |
342 | PL_oldoldbufptr = s + oldoldprevlen; |
343 | if (PL_last_uni) |
344 | PL_last_uni = s + oldunilen; |
345 | if (PL_last_lop) |
346 | PL_last_lop = s + oldloplen; |
347 | incline(s); |
348 | |
349 | /* debugger active and we're not compiling the debugger code, |
350 | * so store the line into the debugger's array of lines |
351 | */ |
352 | if (PERLDB_LINE && PL_curstash != PL_debstash) { |
353 | SV * const sv = NEWSV(85,0); |
354 | |
355 | sv_upgrade(sv, SVt_PVMG); |
356 | sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr); |
357 | (void)SvIOK_on(sv); |
358 | SvIV_set(sv, 0); |
359 | av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv); |
360 | } |
361 | } |
362 | } |
363 | |
364 | STATIC char * |
365 | S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp) |
366 | { |
367 | register char *d = dest; |
368 | register char * const e = d + destlen - 3; /* two-character token, ending NUL */ |
369 | for (;;) { |
370 | if (d >= e) |
371 | Perl_croak(aTHX_ ident_too_long); |
372 | if (isALNUM(*s)) /* UTF handled below */ |
373 | *d++ = *s++; |
374 | else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) { |
375 | *d++ = ':'; |
376 | *d++ = ':'; |
377 | s++; |
378 | } |
379 | else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') { |
380 | *d++ = *s++; |
381 | *d++ = *s++; |
382 | } |
383 | else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) { |
384 | char *t = s + UTF8SKIP(s); |
385 | while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t)) |
386 | t += UTF8SKIP(t); |
387 | if (d + (t - s) > e) |
388 | Perl_croak(aTHX_ ident_too_long); |
389 | Copy(s, d, t - s, char); |
390 | d += t - s; |
391 | s = t; |
392 | } |
393 | else { |
394 | *d = '\0'; |
395 | *slp = d - dest; |
396 | return s; |
397 | } |
398 | } |
399 | } |
400 | |
401 | /* |
402 | * S_incline |
403 | * This subroutine has nothing to do with tilting, whether at windmills |
404 | * or pinball tables. Its name is short for "increment line". It |
405 | * increments the current line number in CopLINE(PL_curcop) and checks |
406 | * to see whether the line starts with a comment of the form |
407 | * # line 500 "foo.pm" |
408 | * If so, it sets the current line number and file to the values in the comment. |
409 | */ |
410 | |
411 | STATIC void |
78160085 |
412 | S_incline(pTHX_ char *s) |
e807ee50 |
413 | { |
414 | char *t; |
78160085 |
415 | char *n; |
416 | char *e; |
e807ee50 |
417 | char ch; |
418 | |
419 | CopLINE_inc(PL_curcop); |
420 | if (*s++ != '#') |
421 | return; |
422 | while (SPACE_OR_TAB(*s)) s++; |
423 | if (strnEQ(s, "line", 4)) |
424 | s += 4; |
425 | else |
426 | return; |
427 | if (SPACE_OR_TAB(*s)) |
428 | s++; |
429 | else |
430 | return; |
431 | while (SPACE_OR_TAB(*s)) s++; |
432 | if (!isDIGIT(*s)) |
433 | return; |
434 | n = s; |
435 | while (isDIGIT(*s)) |
436 | s++; |
437 | while (SPACE_OR_TAB(*s)) |
438 | s++; |
439 | if (*s == '"' && (t = strchr(s+1, '"'))) { |
440 | s++; |
441 | e = t + 1; |
442 | } |
443 | else { |
78160085 |
444 | for (t = s; !isSPACE(*t); t++) ; |
e807ee50 |
445 | e = t; |
446 | } |
447 | while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f') |
448 | e++; |
449 | if (*e != '\n' && *e != '\0') |
450 | return; /* false alarm */ |
451 | |
452 | ch = *t; |
453 | *t = '\0'; |
454 | if (t - s > 0) { |
71511651 |
455 | /* this chunk was added to S_incline during 5.8.8. I don't know why but I don't |
456 | honestly care since I probably want to be bug-compatible anyway (mst) */ |
457 | |
458 | /* ... my kingdom for a perl parser in perl ... (mst) */ |
459 | |
460 | #ifdef PERL_5_8_8_PLUS |
e807ee50 |
461 | #ifndef USE_ITHREADS |
462 | const char *cf = CopFILE(PL_curcop); |
463 | if (cf && strlen(cf) > 7 && strnEQ(cf, "(eval ", 6)) { |
464 | /* must copy *{"::_<(eval N)[oldfilename:L]"} |
465 | * to *{"::_<newfilename"} */ |
466 | char smallbuf[256], smallbuf2[256]; |
467 | char *tmpbuf, *tmpbuf2; |
468 | GV **gvp, *gv2; |
469 | STRLEN tmplen = strlen(cf); |
470 | STRLEN tmplen2 = strlen(s); |
471 | if (tmplen + 3 < sizeof smallbuf) |
472 | tmpbuf = smallbuf; |
473 | else |
474 | Newx(tmpbuf, tmplen + 3, char); |
475 | if (tmplen2 + 3 < sizeof smallbuf2) |
476 | tmpbuf2 = smallbuf2; |
477 | else |
478 | Newx(tmpbuf2, tmplen2 + 3, char); |
479 | tmpbuf[0] = tmpbuf2[0] = '_'; |
480 | tmpbuf[1] = tmpbuf2[1] = '<'; |
481 | memcpy(tmpbuf + 2, cf, ++tmplen); |
482 | memcpy(tmpbuf2 + 2, s, ++tmplen2); |
483 | ++tmplen; ++tmplen2; |
484 | gvp = (GV**)hv_fetch(PL_defstash, tmpbuf, tmplen, FALSE); |
485 | if (gvp) { |
486 | gv2 = *(GV**)hv_fetch(PL_defstash, tmpbuf2, tmplen2, TRUE); |
487 | if (!isGV(gv2)) |
488 | gv_init(gv2, PL_defstash, tmpbuf2, tmplen2, FALSE); |
489 | /* adjust ${"::_<newfilename"} to store the new file name */ |
490 | GvSV(gv2) = newSVpvn(tmpbuf2 + 2, tmplen2 - 2); |
491 | GvHV(gv2) = (HV*)SvREFCNT_inc(GvHV(*gvp)); |
492 | GvAV(gv2) = (AV*)SvREFCNT_inc(GvAV(*gvp)); |
493 | } |
494 | if (tmpbuf != smallbuf) Safefree(tmpbuf); |
495 | if (tmpbuf2 != smallbuf2) Safefree(tmpbuf2); |
496 | } |
497 | #endif |
71511651 |
498 | #endif |
499 | /* second endif closes out the "are we 5.8.(8+)" conditional */ |
e807ee50 |
500 | CopFILE_free(PL_curcop); |
501 | CopFILE_set(PL_curcop, s); |
502 | } |
503 | *t = ch; |
504 | CopLINE_set(PL_curcop, atoi(n)-1); |
505 | } |
0ba8c7aa |
506 | |
507 | /* scan_str |
508 | takes: start position in buffer |
509 | keep_quoted preserve \ on the embedded delimiter(s) |
510 | keep_delims preserve the delimiters around the string |
511 | returns: position to continue reading from buffer |
512 | side-effects: multi_start, multi_close, lex_repl or lex_stuff, and |
513 | updates the read buffer. |
514 | |
515 | This subroutine pulls a string out of the input. It is called for: |
516 | q single quotes q(literal text) |
517 | ' single quotes 'literal text' |
518 | qq double quotes qq(interpolate $here please) |
519 | " double quotes "interpolate $here please" |
520 | qx backticks qx(/bin/ls -l) |
521 | ` backticks `/bin/ls -l` |
522 | qw quote words @EXPORT_OK = qw( func() $spam ) |
523 | m// regexp match m/this/ |
524 | s/// regexp substitute s/this/that/ |
525 | tr/// string transliterate tr/this/that/ |
526 | y/// string transliterate y/this/that/ |
527 | ($*@) sub prototypes sub foo ($) |
528 | (stuff) sub attr parameters sub foo : attr(stuff) |
529 | <> readline or globs <FOO>, <>, <$fh>, or <*.c> |
530 | |
531 | In most of these cases (all but <>, patterns and transliterate) |
532 | yylex() calls scan_str(). m// makes yylex() call scan_pat() which |
533 | calls scan_str(). s/// makes yylex() call scan_subst() which calls |
534 | scan_str(). tr/// and y/// make yylex() call scan_trans() which |
535 | calls scan_str(). |
536 | |
537 | It skips whitespace before the string starts, and treats the first |
538 | character as the delimiter. If the delimiter is one of ([{< then |
539 | the corresponding "close" character )]}> is used as the closing |
540 | delimiter. It allows quoting of delimiters, and if the string has |
541 | balanced delimiters ([{<>}]) it allows nesting. |
542 | |
543 | On success, the SV with the resulting string is put into lex_stuff or, |
544 | if that is already non-NULL, into lex_repl. The second case occurs only |
545 | when parsing the RHS of the special constructs s/// and tr/// (y///). |
546 | For convenience, the terminating delimiter character is stuffed into |
547 | SvIVX of the SV. |
548 | */ |
549 | |
550 | STATIC char * |
551 | S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims) |
552 | { |
553 | SV *sv; /* scalar value: string */ |
554 | char *tmps; /* temp string, used for delimiter matching */ |
555 | register char *s = start; /* current position in the buffer */ |
556 | register char term; /* terminating character */ |
557 | register char *to; /* current position in the sv's data */ |
558 | I32 brackets = 1; /* bracket nesting level */ |
559 | bool has_utf8 = FALSE; /* is there any utf8 content? */ |
560 | I32 termcode; /* terminating char. code */ |
71511651 |
561 | /* 5.8.7+ uses UTF8_MAXBYTES but also its utf8.h defs _MAXLEN to it so |
562 | I'm reasonably hopeful this won't destroy anything (mst) */ |
563 | U8 termstr[UTF8_MAXLEN]; /* terminating string */ |
0ba8c7aa |
564 | STRLEN termlen; /* length of terminating string */ |
565 | char *last = NULL; /* last position for nesting bracket */ |
566 | |
567 | /* skip space before the delimiter */ |
568 | if (isSPACE(*s)) |
569 | s = skipspace(s); |
570 | |
571 | /* mark where we are, in case we need to report errors */ |
572 | CLINE; |
573 | |
574 | /* after skipping whitespace, the next character is the terminator */ |
575 | term = *s; |
576 | if (!UTF) { |
577 | termcode = termstr[0] = term; |
578 | termlen = 1; |
579 | } |
580 | else { |
581 | termcode = utf8_to_uvchr((U8*)s, &termlen); |
582 | Copy(s, termstr, termlen, U8); |
583 | if (!UTF8_IS_INVARIANT(term)) |
584 | has_utf8 = TRUE; |
585 | } |
586 | |
587 | /* mark where we are */ |
588 | PL_multi_start = CopLINE(PL_curcop); |
589 | PL_multi_open = term; |
590 | |
591 | /* find corresponding closing delimiter */ |
592 | if (term && (tmps = strchr("([{< )]}> )]}>",term))) |
593 | termcode = termstr[0] = term = tmps[5]; |
594 | |
595 | PL_multi_close = term; |
596 | |
597 | /* create a new SV to hold the contents. 87 is leak category, I'm |
598 | assuming. 79 is the SV's initial length. What a random number. */ |
599 | sv = NEWSV(87,79); |
600 | sv_upgrade(sv, SVt_PVIV); |
601 | SvIV_set(sv, termcode); |
602 | (void)SvPOK_only(sv); /* validate pointer */ |
603 | |
604 | /* move past delimiter and try to read a complete string */ |
605 | if (keep_delims) |
606 | sv_catpvn(sv, s, termlen); |
607 | s += termlen; |
608 | for (;;) { |
609 | if (PL_encoding && !UTF) { |
610 | bool cont = TRUE; |
611 | |
612 | while (cont) { |
613 | int offset = s - SvPVX_const(PL_linestr); |
614 | const bool found = sv_cat_decode(sv, PL_encoding, PL_linestr, |
615 | &offset, (char*)termstr, termlen); |
616 | const char *ns = SvPVX_const(PL_linestr) + offset; |
617 | char *svlast = SvEND(sv) - 1; |
618 | |
619 | for (; s < ns; s++) { |
620 | if (*s == '\n' && !PL_rsfp) |
621 | CopLINE_inc(PL_curcop); |
622 | } |
623 | if (!found) |
624 | goto read_more_line; |
625 | else { |
626 | /* handle quoted delimiters */ |
627 | if (SvCUR(sv) > 1 && *(svlast-1) == '\\') { |
628 | const char *t; |
629 | for (t = svlast-2; t >= SvPVX_const(sv) && *t == '\\';) |
630 | t--; |
631 | if ((svlast-1 - t) % 2) { |
632 | if (!keep_quoted) { |
633 | *(svlast-1) = term; |
634 | *svlast = '\0'; |
635 | SvCUR_set(sv, SvCUR(sv) - 1); |
636 | } |
637 | continue; |
638 | } |
639 | } |
640 | if (PL_multi_open == PL_multi_close) { |
641 | cont = FALSE; |
642 | } |
643 | else { |
644 | const char *t; |
645 | char *w; |
646 | if (!last) |
647 | last = SvPVX(sv); |
648 | for (t = w = last; t < svlast; w++, t++) { |
649 | /* At here, all closes are "was quoted" one, |
650 | so we don't check PL_multi_close. */ |
651 | if (*t == '\\') { |
652 | if (!keep_quoted && *(t+1) == PL_multi_open) |
653 | t++; |
654 | else |
655 | *w++ = *t++; |
656 | } |
657 | else if (*t == PL_multi_open) |
658 | brackets++; |
659 | |
660 | *w = *t; |
661 | } |
662 | if (w < t) { |
663 | *w++ = term; |
664 | *w = '\0'; |
665 | SvCUR_set(sv, w - SvPVX_const(sv)); |
666 | } |
667 | last = w; |
668 | if (--brackets <= 0) |
669 | cont = FALSE; |
670 | } |
671 | } |
672 | } |
673 | if (!keep_delims) { |
674 | SvCUR_set(sv, SvCUR(sv) - 1); |
675 | *SvEND(sv) = '\0'; |
676 | } |
677 | break; |
678 | } |
679 | |
680 | /* extend sv if need be */ |
681 | SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1); |
682 | /* set 'to' to the next character in the sv's string */ |
683 | to = SvPVX(sv)+SvCUR(sv); |
684 | |
685 | /* if open delimiter is the close delimiter read unbridle */ |
686 | if (PL_multi_open == PL_multi_close) { |
687 | for (; s < PL_bufend; s++,to++) { |
688 | /* embedded newlines increment the current line number */ |
689 | if (*s == '\n' && !PL_rsfp) |
690 | CopLINE_inc(PL_curcop); |
691 | /* handle quoted delimiters */ |
692 | if (*s == '\\' && s+1 < PL_bufend && term != '\\') { |
693 | if (!keep_quoted && s[1] == term) |
694 | s++; |
695 | /* any other quotes are simply copied straight through */ |
696 | else |
697 | *to++ = *s++; |
698 | } |
699 | /* terminate when run out of buffer (the for() condition), or |
700 | have found the terminator */ |
701 | else if (*s == term) { |
702 | if (termlen == 1) |
703 | break; |
704 | if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen)) |
705 | break; |
706 | } |
707 | else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF) |
708 | has_utf8 = TRUE; |
709 | *to = *s; |
710 | } |
711 | } |
712 | |
713 | /* if the terminator isn't the same as the start character (e.g., |
714 | matched brackets), we have to allow more in the quoting, and |
715 | be prepared for nested brackets. |
716 | */ |
717 | else { |
718 | /* read until we run out of string, or we find the terminator */ |
719 | for (; s < PL_bufend; s++,to++) { |
720 | /* embedded newlines increment the line count */ |
721 | if (*s == '\n' && !PL_rsfp) |
722 | CopLINE_inc(PL_curcop); |
723 | /* backslashes can escape the open or closing characters */ |
724 | if (*s == '\\' && s+1 < PL_bufend) { |
725 | if (!keep_quoted && |
726 | ((s[1] == PL_multi_open) || (s[1] == PL_multi_close))) |
727 | s++; |
728 | else |
729 | *to++ = *s++; |
730 | } |
731 | /* allow nested opens and closes */ |
732 | else if (*s == PL_multi_close && --brackets <= 0) |
733 | break; |
734 | else if (*s == PL_multi_open) |
735 | brackets++; |
736 | else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF) |
737 | has_utf8 = TRUE; |
738 | *to = *s; |
739 | } |
740 | } |
741 | /* terminate the copied string and update the sv's end-of-string */ |
742 | *to = '\0'; |
743 | SvCUR_set(sv, to - SvPVX_const(sv)); |
744 | |
745 | /* |
746 | * this next chunk reads more into the buffer if we're not done yet |
747 | */ |
748 | |
749 | if (s < PL_bufend) |
750 | break; /* handle case where we are done yet :-) */ |
751 | |
752 | #ifndef PERL_STRICT_CR |
753 | if (to - SvPVX_const(sv) >= 2) { |
754 | if ((to[-2] == '\r' && to[-1] == '\n') || |
755 | (to[-2] == '\n' && to[-1] == '\r')) |
756 | { |
757 | to[-2] = '\n'; |
758 | to--; |
759 | SvCUR_set(sv, to - SvPVX_const(sv)); |
760 | } |
761 | else if (to[-1] == '\r') |
762 | to[-1] = '\n'; |
763 | } |
764 | else if (to - SvPVX_const(sv) == 1 && to[-1] == '\r') |
765 | to[-1] = '\n'; |
766 | #endif |
767 | |
768 | read_more_line: |
769 | /* if we're out of file, or a read fails, bail and reset the current |
770 | line marker so we can report where the unterminated string began |
771 | */ |
772 | if (!PL_rsfp || |
773 | !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) { |
774 | sv_free(sv); |
775 | CopLINE_set(PL_curcop, (line_t)PL_multi_start); |
776 | return Nullch; |
777 | } |
778 | /* we read a line, so increment our line counter */ |
779 | CopLINE_inc(PL_curcop); |
780 | |
781 | /* update debugger info */ |
782 | if (PERLDB_LINE && PL_curstash != PL_debstash) { |
783 | SV *sv = NEWSV(88,0); |
784 | |
785 | sv_upgrade(sv, SVt_PVMG); |
786 | sv_setsv(sv,PL_linestr); |
787 | (void)SvIOK_on(sv); |
788 | SvIV_set(sv, 0); |
789 | av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv); |
790 | } |
791 | |
792 | /* having changed the buffer, we must update PL_bufend */ |
793 | PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); |
794 | PL_last_lop = PL_last_uni = Nullch; |
795 | } |
796 | |
797 | /* at this point, we have successfully read the delimited string */ |
798 | |
799 | if (!PL_encoding || UTF) { |
800 | if (keep_delims) |
801 | sv_catpvn(sv, s, termlen); |
802 | s += termlen; |
803 | } |
804 | if (has_utf8 || PL_encoding) |
805 | SvUTF8_on(sv); |
806 | |
807 | PL_multi_end = CopLINE(PL_curcop); |
808 | |
809 | /* if we allocated too much space, give some back */ |
810 | if (SvCUR(sv) + 5 < SvLEN(sv)) { |
811 | SvLEN_set(sv, SvCUR(sv) + 1); |
71511651 |
812 | /* 5.8.8 uses SvPV_renew, no prior version actually has the damn thing (mst) */ |
813 | #ifdef PERL_5_8_8_PLUS |
0ba8c7aa |
814 | SvPV_renew(sv, SvLEN(sv)); |
71511651 |
815 | #else |
816 | Renew(SvPVX(sv), SvLEN(sv), char); |
817 | #endif |
0ba8c7aa |
818 | } |
819 | |
820 | /* decide whether this is the first or second quoted string we've read |
821 | for this op |
822 | */ |
823 | |
824 | if (PL_lex_stuff) |
825 | PL_lex_repl = sv; |
826 | else |
827 | PL_lex_stuff = sv; |
828 | return s; |
829 | } |
830 | |
831 | /* |
832 | * S_force_next |
833 | * When the lexer realizes it knows the next token (for instance, |
834 | * it is reordering tokens for the parser) then it can call S_force_next |
835 | * to know what token to return the next time the lexer is called. Caller |
836 | * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer |
837 | * handles the token correctly. |
838 | */ |
839 | |
840 | STATIC void |
841 | S_force_next(pTHX_ I32 type) |
842 | { |
843 | PL_nexttype[PL_nexttoke] = type; |
844 | PL_nexttoke++; |
845 | if (PL_lex_state != LEX_KNOWNEXT) { |
846 | PL_lex_defer = PL_lex_state; |
847 | PL_lex_expect = PL_expect; |
848 | PL_lex_state = LEX_KNOWNEXT; |
849 | } |
850 | } |