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) |
92997e69 |
27 | #define scan_ident(a,b,c,d,e) S_scan_ident(aTHX_ a,b,c,d,e) |
78160085 |
28 | |
29 | STATIC void S_incline(pTHX_ char *s); |
30 | STATIC char* S_skipspace(pTHX_ char *s); |
31 | STATIC char * S_filter_gets(pTHX_ SV *sv, PerlIO *fp, STRLEN append); |
32 | STATIC char* S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims); |
33 | STATIC 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 */ |
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 |
60d774aa |
47 | #ifndef MEM_WRAP_CHECK_ |
48 | #define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t), |
49 | #endif |
14ab3472 |
50 | |
0ba8c7aa |
51 | #define SvPV_renew(sv,n) \ |
52 | STMT_START { SvLEN_set(sv, n); \ |
53 | SvPV_set((sv), (MEM_WRAP_CHECK_(n,char) \ |
54 | (char*)saferealloc((Malloc_t)SvPVX(sv), \ |
55 | (MEM_SIZE)((n))))); \ |
56 | } STMT_END |
e807ee50 |
57 | |
d6e7537a |
58 | #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x))) |
59 | |
c630715a |
60 | /* On MacOS, respect nonbreaking spaces */ |
61 | #ifdef MACOS_TRADITIONAL |
62 | #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t') |
63 | #else |
64 | #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t') |
65 | #endif |
e807ee50 |
66 | |
2688337b |
67 | /* |
68 | * Normally, during compile time, PL_curcop == &PL_compiling is true. However, |
69 | * Devel::Declare makes the interpreter call back to perl during compile time, |
70 | * which temporarily enters runtime. Then perl space calls various functions |
71 | * from this file, which are designed to work during compile time. They all |
72 | * happen to operate on PL_curcop, not PL_compiling. That doesn't make a |
73 | * difference in the core, but it does for Devel::Declare, which operates at |
74 | * runtime, but still wants to mangle the things that are about to be compiled. |
75 | * That's why we define our own PL_curcop and make it point to PL_compiling |
76 | * here. |
77 | */ |
78 | #define PL_curcop &PL_compiling |
79 | |
0ba8c7aa |
80 | #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline)) |
81 | |
e807ee50 |
82 | #define LEX_NORMAL 10 /* normal code (ie not within "...") */ |
83 | #define LEX_INTERPNORMAL 9 /* code within a string, eg "$foo[$x+1]" */ |
84 | #define LEX_INTERPCASEMOD 8 /* expecting a \U, \Q or \E etc */ |
85 | #define LEX_INTERPPUSH 7 /* starting a new sublex parse level */ |
86 | #define LEX_INTERPSTART 6 /* expecting the start of a $var */ |
87 | |
88 | /* at end of code, eg "$x" followed by: */ |
89 | #define LEX_INTERPEND 5 /* ... eg not one of [, { or -> */ |
90 | #define LEX_INTERPENDMAYBE 4 /* ... eg one of [, { or -> */ |
91 | |
92 | #define LEX_INTERPCONCAT 3 /* expecting anything, eg at start of |
93 | string or after \E, $foo, etc */ |
94 | #define LEX_INTERPCONST 2 /* NOT USED */ |
95 | #define LEX_FORMLINE 1 /* expecting a format line */ |
96 | #define LEX_KNOWNEXT 0 /* next token known; just return it */ |
97 | |
14ab3472 |
98 | /* and these two are my own madness (mst) */ |
71511651 |
99 | |
100 | #if PERL_REVISION == 5 && PERL_VERSION == 8 && PERL_SUBVERSION >= 8 |
101 | #define PERL_5_8_8_PLUS |
102 | #endif |
103 | |
e38f5a74 |
104 | #if PERL_REVISION == 5 && PERL_VERSION > 8 |
105 | #define PERL_5_9_PLUS |
106 | #endif |
107 | |
108 | #ifdef PERL_5_9_PLUS |
14ab3472 |
109 | /* 5.9+ moves a bunch of things to a PL_parser struct so we need to |
110 | declare the backcompat macros for things to still work (mst) */ |
111 | |
112 | /* XXX temporary backwards compatibility */ |
113 | #define PL_lex_brackets (PL_parser->lex_brackets) |
114 | #define PL_lex_brackstack (PL_parser->lex_brackstack) |
115 | #define PL_lex_casemods (PL_parser->lex_casemods) |
116 | #define PL_lex_casestack (PL_parser->lex_casestack) |
117 | #define PL_lex_defer (PL_parser->lex_defer) |
118 | #define PL_lex_dojoin (PL_parser->lex_dojoin) |
119 | #define PL_lex_expect (PL_parser->lex_expect) |
120 | #define PL_lex_formbrack (PL_parser->lex_formbrack) |
121 | #define PL_lex_inpat (PL_parser->lex_inpat) |
122 | #define PL_lex_inwhat (PL_parser->lex_inwhat) |
123 | #define PL_lex_op (PL_parser->lex_op) |
124 | #define PL_lex_repl (PL_parser->lex_repl) |
125 | #define PL_lex_starts (PL_parser->lex_starts) |
126 | #define PL_lex_stuff (PL_parser->lex_stuff) |
127 | #define PL_multi_start (PL_parser->multi_start) |
128 | #define PL_multi_open (PL_parser->multi_open) |
129 | #define PL_multi_close (PL_parser->multi_close) |
130 | #define PL_pending_ident (PL_parser->pending_ident) |
131 | #define PL_preambled (PL_parser->preambled) |
132 | #define PL_sublex_info (PL_parser->sublex_info) |
133 | #define PL_linestr (PL_parser->linestr) |
134 | #define PL_sublex_info (PL_parser->sublex_info) |
135 | #define PL_linestr (PL_parser->linestr) |
136 | #define PL_expect (PL_parser->expect) |
137 | #define PL_copline (PL_parser->copline) |
138 | #define PL_bufptr (PL_parser->bufptr) |
139 | #define PL_oldbufptr (PL_parser->oldbufptr) |
140 | #define PL_oldoldbufptr (PL_parser->oldoldbufptr) |
141 | #define PL_linestart (PL_parser->linestart) |
142 | #define PL_bufend (PL_parser->bufend) |
143 | #define PL_last_uni (PL_parser->last_uni) |
144 | #define PL_last_lop (PL_parser->last_lop) |
145 | #define PL_last_lop_op (PL_parser->last_lop_op) |
146 | #define PL_lex_state (PL_parser->lex_state) |
147 | #define PL_rsfp (PL_parser->rsfp) |
148 | #define PL_rsfp_filters (PL_parser->rsfp_filters) |
149 | #define PL_in_my (PL_parser->in_my) |
150 | #define PL_in_my_stash (PL_parser->in_my_stash) |
151 | #define PL_tokenbuf (PL_parser->tokenbuf) |
152 | #define PL_multi_end (PL_parser->multi_end) |
153 | #define PL_error_count (PL_parser->error_count) |
52eaf5fa |
154 | #define PL_nexttoke (PL_parser->nexttoke) |
155 | /* these are from the non-PERL_MAD path but I don't -think- I need |
14ab3472 |
156 | the PERL_MAD stuff since my code isn't really populating things (mst) */ |
52eaf5fa |
157 | # ifdef PERL_MAD |
158 | # define PL_curforce (PL_parser->curforce) |
159 | # define PL_lasttoke (PL_parser->lasttoke) |
160 | # else |
14ab3472 |
161 | # define PL_nexttype (PL_parser->nexttype) |
162 | # define PL_nextval (PL_parser->nextval) |
52eaf5fa |
163 | # endif |
164 | /* end of backcompat macros from 5.9 toke.c (mst) */ |
78160085 |
165 | #endif |
14ab3472 |
166 | |
167 | /* when ccflags include -DDEBUGGING we need this for earlier 5.8 perls */ |
168 | #ifndef SvPV_nolen_const |
169 | #define SvPV_nolen_const SvPV_nolen |
170 | #endif |
171 | |
71511651 |
172 | /* and now we're back to the toke.c stuff again (mst) */ |
173 | |
e807ee50 |
174 | static const char ident_too_long[] = |
175 | "Identifier too long"; |
176 | static const char c_without_g[] = |
177 | "Use of /c modifier is meaningless without /g"; |
178 | static const char c_in_subst[] = |
179 | "Use of /c modifier is meaningless in s///"; |
180 | |
181 | #ifdef USE_UTF8_SCRIPTS |
182 | # define UTF (!IN_BYTES) |
183 | #else |
184 | # define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || (PL_hints & HINT_UTF8)) |
185 | #endif |
186 | |
187 | /* Invoke the idxth filter function for the current rsfp. */ |
188 | /* maxlen 0 = read one text line */ |
189 | I32 |
190 | Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen) |
191 | { |
192 | filter_t funcp; |
193 | SV *datasv = NULL; |
194 | |
195 | if (!PL_rsfp_filters) |
196 | return -1; |
197 | if (idx > AvFILLp(PL_rsfp_filters)) { /* Any more filters? */ |
198 | /* Provide a default input filter to make life easy. */ |
199 | /* Note that we append to the line. This is handy. */ |
200 | DEBUG_P(PerlIO_printf(Perl_debug_log, |
201 | "filter_read %d: from rsfp\n", idx)); |
202 | if (maxlen) { |
203 | /* Want a block */ |
204 | int len ; |
205 | const int old_len = SvCUR(buf_sv); |
206 | |
207 | /* ensure buf_sv is large enough */ |
208 | SvGROW(buf_sv, (STRLEN)(old_len + maxlen)) ; |
209 | if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){ |
210 | if (PerlIO_error(PL_rsfp)) |
211 | return -1; /* error */ |
212 | else |
213 | return 0 ; /* end of file */ |
214 | } |
215 | SvCUR_set(buf_sv, old_len + len) ; |
216 | } else { |
217 | /* Want a line */ |
218 | if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) { |
219 | if (PerlIO_error(PL_rsfp)) |
220 | return -1; /* error */ |
221 | else |
222 | return 0 ; /* end of file */ |
223 | } |
224 | } |
225 | return SvCUR(buf_sv); |
226 | } |
227 | /* Skip this filter slot if filter has been deleted */ |
228 | if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef) { |
229 | DEBUG_P(PerlIO_printf(Perl_debug_log, |
230 | "filter_read %d: skipped (filter deleted)\n", |
231 | idx)); |
232 | return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */ |
233 | } |
234 | /* Get function pointer hidden within datasv */ |
235 | funcp = DPTR2FPTR(filter_t, IoANY(datasv)); |
236 | DEBUG_P(PerlIO_printf(Perl_debug_log, |
237 | "filter_read %d: via function %p (%s)\n", |
238 | idx, datasv, SvPV_nolen_const(datasv))); |
239 | /* Call function. The function is expected to */ |
240 | /* call "FILTER_READ(idx+1, buf_sv)" first. */ |
241 | /* Return: <0:error, =0:eof, >0:not eof */ |
242 | return (*funcp)(aTHX_ idx, buf_sv, maxlen); |
243 | } |
244 | |
245 | STATIC char * |
246 | S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append) |
247 | { |
248 | #ifdef PERL_CR_FILTER |
249 | if (!PL_rsfp_filters) { |
250 | filter_add(S_cr_textfilter,NULL); |
251 | } |
252 | #endif |
253 | if (PL_rsfp_filters) { |
254 | if (!append) |
255 | SvCUR_set(sv, 0); /* start with empty line */ |
256 | if (FILTER_READ(0, sv, 0) > 0) |
257 | return ( SvPVX(sv) ) ; |
258 | else |
259 | return Nullch ; |
260 | } |
261 | else |
262 | return (sv_gets(sv, fp, append)); |
263 | } |
264 | |
265 | /* |
266 | * S_skipspace |
267 | * Called to gobble the appropriate amount and type of whitespace. |
268 | * Skips comments as well. |
269 | */ |
270 | |
271 | STATIC char * |
272 | S_skipspace(pTHX_ register char *s) |
273 | { |
274 | if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) { |
275 | while (s < PL_bufend && SPACE_OR_TAB(*s)) |
276 | s++; |
277 | return s; |
278 | } |
279 | for (;;) { |
280 | STRLEN prevlen; |
281 | SSize_t oldprevlen, oldoldprevlen; |
282 | SSize_t oldloplen = 0, oldunilen = 0; |
283 | while (s < PL_bufend && isSPACE(*s)) { |
284 | if (*s++ == '\n' && PL_in_eval && !PL_rsfp) |
285 | incline(s); |
286 | } |
287 | |
288 | /* comment */ |
289 | if (s < PL_bufend && *s == '#') { |
290 | while (s < PL_bufend && *s != '\n') |
291 | s++; |
292 | if (s < PL_bufend) { |
293 | s++; |
294 | if (PL_in_eval && !PL_rsfp) { |
295 | incline(s); |
296 | continue; |
297 | } |
298 | } |
299 | } |
300 | |
301 | /* only continue to recharge the buffer if we're at the end |
302 | * of the buffer, we're not reading from a source filter, and |
303 | * we're in normal lexing mode |
304 | */ |
305 | if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat || |
306 | PL_lex_state == LEX_FORMLINE) |
307 | return s; |
308 | |
309 | /* try to recharge the buffer */ |
310 | if ((s = filter_gets(PL_linestr, PL_rsfp, |
311 | (prevlen = SvCUR(PL_linestr)))) == Nullch) |
312 | { |
313 | /* end of file. Add on the -p or -n magic */ |
314 | if (PL_minus_p) { |
315 | sv_setpv(PL_linestr, |
316 | ";}continue{print or die qq(-p destination: $!\\n);}"); |
317 | PL_minus_n = PL_minus_p = 0; |
318 | } |
319 | else if (PL_minus_n) { |
320 | sv_setpvn(PL_linestr, ";}", 2); |
321 | PL_minus_n = 0; |
322 | } |
323 | else |
324 | sv_setpvn(PL_linestr,";", 1); |
325 | |
326 | /* reset variables for next time we lex */ |
327 | PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart |
328 | = SvPVX(PL_linestr); |
329 | PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); |
330 | PL_last_lop = PL_last_uni = Nullch; |
331 | |
35fc8ada |
332 | /* In perl versions previous to p4-rawid: //depot/perl@32954 -P |
333 | * preprocessors were supported here. We don't support -P at all, even |
334 | * on perls that support it, and use the following chunk from blead |
335 | * perl. (rafl) |
336 | */ |
337 | |
338 | /* Close the filehandle. Could be from |
e807ee50 |
339 | * STDIN, or a regular file. If we were reading code from |
340 | * STDIN (because the commandline held no -e or filename) |
341 | * then we don't close it, we reset it so the code can |
342 | * read from STDIN too. |
343 | */ |
344 | |
35fc8ada |
345 | if ((PerlIO*)PL_rsfp == PerlIO_stdin()) |
e807ee50 |
346 | PerlIO_clearerr(PL_rsfp); |
347 | else |
348 | (void)PerlIO_close(PL_rsfp); |
349 | PL_rsfp = Nullfp; |
350 | return s; |
351 | } |
352 | |
353 | /* not at end of file, so we only read another line */ |
354 | /* make corresponding updates to old pointers, for yyerror() */ |
355 | oldprevlen = PL_oldbufptr - PL_bufend; |
356 | oldoldprevlen = PL_oldoldbufptr - PL_bufend; |
357 | if (PL_last_uni) |
358 | oldunilen = PL_last_uni - PL_bufend; |
359 | if (PL_last_lop) |
360 | oldloplen = PL_last_lop - PL_bufend; |
361 | PL_linestart = PL_bufptr = s + prevlen; |
362 | PL_bufend = s + SvCUR(PL_linestr); |
363 | s = PL_bufptr; |
364 | PL_oldbufptr = s + oldprevlen; |
365 | PL_oldoldbufptr = s + oldoldprevlen; |
366 | if (PL_last_uni) |
367 | PL_last_uni = s + oldunilen; |
368 | if (PL_last_lop) |
369 | PL_last_lop = s + oldloplen; |
370 | incline(s); |
371 | |
372 | /* debugger active and we're not compiling the debugger code, |
373 | * so store the line into the debugger's array of lines |
374 | */ |
375 | if (PERLDB_LINE && PL_curstash != PL_debstash) { |
376 | SV * const sv = NEWSV(85,0); |
377 | |
378 | sv_upgrade(sv, SVt_PVMG); |
379 | sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr); |
380 | (void)SvIOK_on(sv); |
381 | SvIV_set(sv, 0); |
382 | av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv); |
383 | } |
384 | } |
385 | } |
386 | |
387 | STATIC char * |
388 | S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp) |
389 | { |
390 | register char *d = dest; |
391 | register char * const e = d + destlen - 3; /* two-character token, ending NUL */ |
392 | for (;;) { |
393 | if (d >= e) |
394 | Perl_croak(aTHX_ ident_too_long); |
395 | if (isALNUM(*s)) /* UTF handled below */ |
396 | *d++ = *s++; |
397 | else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) { |
398 | *d++ = ':'; |
399 | *d++ = ':'; |
400 | s++; |
401 | } |
402 | else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') { |
403 | *d++ = *s++; |
404 | *d++ = *s++; |
405 | } |
406 | else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) { |
407 | char *t = s + UTF8SKIP(s); |
408 | while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t)) |
409 | t += UTF8SKIP(t); |
410 | if (d + (t - s) > e) |
411 | Perl_croak(aTHX_ ident_too_long); |
412 | Copy(s, d, t - s, char); |
413 | d += t - s; |
414 | s = t; |
415 | } |
416 | else { |
417 | *d = '\0'; |
418 | *slp = d - dest; |
419 | return s; |
420 | } |
421 | } |
422 | } |
423 | |
424 | /* |
425 | * S_incline |
426 | * This subroutine has nothing to do with tilting, whether at windmills |
427 | * or pinball tables. Its name is short for "increment line". It |
428 | * increments the current line number in CopLINE(PL_curcop) and checks |
429 | * to see whether the line starts with a comment of the form |
430 | * # line 500 "foo.pm" |
431 | * If so, it sets the current line number and file to the values in the comment. |
432 | */ |
433 | |
434 | STATIC void |
78160085 |
435 | S_incline(pTHX_ char *s) |
e807ee50 |
436 | { |
437 | char *t; |
78160085 |
438 | char *n; |
439 | char *e; |
e807ee50 |
440 | char ch; |
441 | |
442 | CopLINE_inc(PL_curcop); |
443 | if (*s++ != '#') |
444 | return; |
445 | while (SPACE_OR_TAB(*s)) s++; |
446 | if (strnEQ(s, "line", 4)) |
447 | s += 4; |
448 | else |
449 | return; |
450 | if (SPACE_OR_TAB(*s)) |
451 | s++; |
452 | else |
453 | return; |
454 | while (SPACE_OR_TAB(*s)) s++; |
455 | if (!isDIGIT(*s)) |
456 | return; |
457 | n = s; |
458 | while (isDIGIT(*s)) |
459 | s++; |
460 | while (SPACE_OR_TAB(*s)) |
461 | s++; |
462 | if (*s == '"' && (t = strchr(s+1, '"'))) { |
463 | s++; |
464 | e = t + 1; |
465 | } |
466 | else { |
78160085 |
467 | for (t = s; !isSPACE(*t); t++) ; |
e807ee50 |
468 | e = t; |
469 | } |
470 | while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f') |
471 | e++; |
472 | if (*e != '\n' && *e != '\0') |
473 | return; /* false alarm */ |
474 | |
475 | ch = *t; |
476 | *t = '\0'; |
477 | if (t - s > 0) { |
71511651 |
478 | /* this chunk was added to S_incline during 5.8.8. I don't know why but I don't |
479 | honestly care since I probably want to be bug-compatible anyway (mst) */ |
480 | |
481 | /* ... my kingdom for a perl parser in perl ... (mst) */ |
482 | |
483 | #ifdef PERL_5_8_8_PLUS |
e807ee50 |
484 | #ifndef USE_ITHREADS |
485 | const char *cf = CopFILE(PL_curcop); |
486 | if (cf && strlen(cf) > 7 && strnEQ(cf, "(eval ", 6)) { |
487 | /* must copy *{"::_<(eval N)[oldfilename:L]"} |
488 | * to *{"::_<newfilename"} */ |
489 | char smallbuf[256], smallbuf2[256]; |
490 | char *tmpbuf, *tmpbuf2; |
491 | GV **gvp, *gv2; |
492 | STRLEN tmplen = strlen(cf); |
493 | STRLEN tmplen2 = strlen(s); |
494 | if (tmplen + 3 < sizeof smallbuf) |
495 | tmpbuf = smallbuf; |
496 | else |
497 | Newx(tmpbuf, tmplen + 3, char); |
498 | if (tmplen2 + 3 < sizeof smallbuf2) |
499 | tmpbuf2 = smallbuf2; |
500 | else |
501 | Newx(tmpbuf2, tmplen2 + 3, char); |
502 | tmpbuf[0] = tmpbuf2[0] = '_'; |
503 | tmpbuf[1] = tmpbuf2[1] = '<'; |
504 | memcpy(tmpbuf + 2, cf, ++tmplen); |
505 | memcpy(tmpbuf2 + 2, s, ++tmplen2); |
506 | ++tmplen; ++tmplen2; |
507 | gvp = (GV**)hv_fetch(PL_defstash, tmpbuf, tmplen, FALSE); |
508 | if (gvp) { |
509 | gv2 = *(GV**)hv_fetch(PL_defstash, tmpbuf2, tmplen2, TRUE); |
510 | if (!isGV(gv2)) |
511 | gv_init(gv2, PL_defstash, tmpbuf2, tmplen2, FALSE); |
512 | /* adjust ${"::_<newfilename"} to store the new file name */ |
513 | GvSV(gv2) = newSVpvn(tmpbuf2 + 2, tmplen2 - 2); |
514 | GvHV(gv2) = (HV*)SvREFCNT_inc(GvHV(*gvp)); |
515 | GvAV(gv2) = (AV*)SvREFCNT_inc(GvAV(*gvp)); |
516 | } |
517 | if (tmpbuf != smallbuf) Safefree(tmpbuf); |
518 | if (tmpbuf2 != smallbuf2) Safefree(tmpbuf2); |
519 | } |
520 | #endif |
71511651 |
521 | #endif |
522 | /* second endif closes out the "are we 5.8.(8+)" conditional */ |
e807ee50 |
523 | CopFILE_free(PL_curcop); |
524 | CopFILE_set(PL_curcop, s); |
525 | } |
526 | *t = ch; |
527 | CopLINE_set(PL_curcop, atoi(n)-1); |
528 | } |
0ba8c7aa |
529 | |
530 | /* scan_str |
531 | takes: start position in buffer |
532 | keep_quoted preserve \ on the embedded delimiter(s) |
533 | keep_delims preserve the delimiters around the string |
534 | returns: position to continue reading from buffer |
535 | side-effects: multi_start, multi_close, lex_repl or lex_stuff, and |
536 | updates the read buffer. |
537 | |
538 | This subroutine pulls a string out of the input. It is called for: |
539 | q single quotes q(literal text) |
540 | ' single quotes 'literal text' |
541 | qq double quotes qq(interpolate $here please) |
542 | " double quotes "interpolate $here please" |
543 | qx backticks qx(/bin/ls -l) |
544 | ` backticks `/bin/ls -l` |
545 | qw quote words @EXPORT_OK = qw( func() $spam ) |
546 | m// regexp match m/this/ |
547 | s/// regexp substitute s/this/that/ |
548 | tr/// string transliterate tr/this/that/ |
549 | y/// string transliterate y/this/that/ |
550 | ($*@) sub prototypes sub foo ($) |
551 | (stuff) sub attr parameters sub foo : attr(stuff) |
552 | <> readline or globs <FOO>, <>, <$fh>, or <*.c> |
553 | |
554 | In most of these cases (all but <>, patterns and transliterate) |
555 | yylex() calls scan_str(). m// makes yylex() call scan_pat() which |
556 | calls scan_str(). s/// makes yylex() call scan_subst() which calls |
557 | scan_str(). tr/// and y/// make yylex() call scan_trans() which |
558 | calls scan_str(). |
559 | |
560 | It skips whitespace before the string starts, and treats the first |
561 | character as the delimiter. If the delimiter is one of ([{< then |
562 | the corresponding "close" character )]}> is used as the closing |
563 | delimiter. It allows quoting of delimiters, and if the string has |
564 | balanced delimiters ([{<>}]) it allows nesting. |
565 | |
566 | On success, the SV with the resulting string is put into lex_stuff or, |
567 | if that is already non-NULL, into lex_repl. The second case occurs only |
568 | when parsing the RHS of the special constructs s/// and tr/// (y///). |
569 | For convenience, the terminating delimiter character is stuffed into |
570 | SvIVX of the SV. |
571 | */ |
572 | |
573 | STATIC char * |
574 | S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims) |
575 | { |
576 | SV *sv; /* scalar value: string */ |
577 | char *tmps; /* temp string, used for delimiter matching */ |
578 | register char *s = start; /* current position in the buffer */ |
579 | register char term; /* terminating character */ |
580 | register char *to; /* current position in the sv's data */ |
581 | I32 brackets = 1; /* bracket nesting level */ |
582 | bool has_utf8 = FALSE; /* is there any utf8 content? */ |
583 | I32 termcode; /* terminating char. code */ |
71511651 |
584 | /* 5.8.7+ uses UTF8_MAXBYTES but also its utf8.h defs _MAXLEN to it so |
585 | I'm reasonably hopeful this won't destroy anything (mst) */ |
586 | U8 termstr[UTF8_MAXLEN]; /* terminating string */ |
0ba8c7aa |
587 | STRLEN termlen; /* length of terminating string */ |
588 | char *last = NULL; /* last position for nesting bracket */ |
589 | |
590 | /* skip space before the delimiter */ |
591 | if (isSPACE(*s)) |
592 | s = skipspace(s); |
593 | |
594 | /* mark where we are, in case we need to report errors */ |
595 | CLINE; |
596 | |
597 | /* after skipping whitespace, the next character is the terminator */ |
598 | term = *s; |
599 | if (!UTF) { |
600 | termcode = termstr[0] = term; |
601 | termlen = 1; |
602 | } |
603 | else { |
604 | termcode = utf8_to_uvchr((U8*)s, &termlen); |
605 | Copy(s, termstr, termlen, U8); |
606 | if (!UTF8_IS_INVARIANT(term)) |
607 | has_utf8 = TRUE; |
608 | } |
609 | |
610 | /* mark where we are */ |
611 | PL_multi_start = CopLINE(PL_curcop); |
612 | PL_multi_open = term; |
613 | |
614 | /* find corresponding closing delimiter */ |
615 | if (term && (tmps = strchr("([{< )]}> )]}>",term))) |
616 | termcode = termstr[0] = term = tmps[5]; |
617 | |
618 | PL_multi_close = term; |
619 | |
620 | /* create a new SV to hold the contents. 87 is leak category, I'm |
621 | assuming. 79 is the SV's initial length. What a random number. */ |
622 | sv = NEWSV(87,79); |
623 | sv_upgrade(sv, SVt_PVIV); |
624 | SvIV_set(sv, termcode); |
625 | (void)SvPOK_only(sv); /* validate pointer */ |
626 | |
627 | /* move past delimiter and try to read a complete string */ |
628 | if (keep_delims) |
629 | sv_catpvn(sv, s, termlen); |
630 | s += termlen; |
631 | for (;;) { |
632 | if (PL_encoding && !UTF) { |
633 | bool cont = TRUE; |
634 | |
635 | while (cont) { |
636 | int offset = s - SvPVX_const(PL_linestr); |
637 | const bool found = sv_cat_decode(sv, PL_encoding, PL_linestr, |
638 | &offset, (char*)termstr, termlen); |
639 | const char *ns = SvPVX_const(PL_linestr) + offset; |
640 | char *svlast = SvEND(sv) - 1; |
641 | |
642 | for (; s < ns; s++) { |
643 | if (*s == '\n' && !PL_rsfp) |
644 | CopLINE_inc(PL_curcop); |
645 | } |
646 | if (!found) |
647 | goto read_more_line; |
648 | else { |
649 | /* handle quoted delimiters */ |
650 | if (SvCUR(sv) > 1 && *(svlast-1) == '\\') { |
651 | const char *t; |
652 | for (t = svlast-2; t >= SvPVX_const(sv) && *t == '\\';) |
653 | t--; |
654 | if ((svlast-1 - t) % 2) { |
655 | if (!keep_quoted) { |
656 | *(svlast-1) = term; |
657 | *svlast = '\0'; |
658 | SvCUR_set(sv, SvCUR(sv) - 1); |
659 | } |
660 | continue; |
661 | } |
662 | } |
663 | if (PL_multi_open == PL_multi_close) { |
664 | cont = FALSE; |
665 | } |
666 | else { |
667 | const char *t; |
668 | char *w; |
669 | if (!last) |
670 | last = SvPVX(sv); |
671 | for (t = w = last; t < svlast; w++, t++) { |
672 | /* At here, all closes are "was quoted" one, |
673 | so we don't check PL_multi_close. */ |
674 | if (*t == '\\') { |
675 | if (!keep_quoted && *(t+1) == PL_multi_open) |
676 | t++; |
677 | else |
678 | *w++ = *t++; |
679 | } |
680 | else if (*t == PL_multi_open) |
681 | brackets++; |
682 | |
683 | *w = *t; |
684 | } |
685 | if (w < t) { |
686 | *w++ = term; |
687 | *w = '\0'; |
688 | SvCUR_set(sv, w - SvPVX_const(sv)); |
689 | } |
690 | last = w; |
691 | if (--brackets <= 0) |
692 | cont = FALSE; |
693 | } |
694 | } |
695 | } |
696 | if (!keep_delims) { |
697 | SvCUR_set(sv, SvCUR(sv) - 1); |
698 | *SvEND(sv) = '\0'; |
699 | } |
700 | break; |
701 | } |
702 | |
703 | /* extend sv if need be */ |
704 | SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1); |
705 | /* set 'to' to the next character in the sv's string */ |
706 | to = SvPVX(sv)+SvCUR(sv); |
707 | |
708 | /* if open delimiter is the close delimiter read unbridle */ |
709 | if (PL_multi_open == PL_multi_close) { |
710 | for (; s < PL_bufend; s++,to++) { |
711 | /* embedded newlines increment the current line number */ |
712 | if (*s == '\n' && !PL_rsfp) |
713 | CopLINE_inc(PL_curcop); |
714 | /* handle quoted delimiters */ |
715 | if (*s == '\\' && s+1 < PL_bufend && term != '\\') { |
716 | if (!keep_quoted && s[1] == term) |
717 | s++; |
718 | /* any other quotes are simply copied straight through */ |
719 | else |
720 | *to++ = *s++; |
721 | } |
722 | /* terminate when run out of buffer (the for() condition), or |
723 | have found the terminator */ |
724 | else if (*s == term) { |
725 | if (termlen == 1) |
726 | break; |
727 | if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen)) |
728 | break; |
729 | } |
730 | else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF) |
731 | has_utf8 = TRUE; |
732 | *to = *s; |
733 | } |
734 | } |
735 | |
736 | /* if the terminator isn't the same as the start character (e.g., |
737 | matched brackets), we have to allow more in the quoting, and |
738 | be prepared for nested brackets. |
739 | */ |
740 | else { |
741 | /* read until we run out of string, or we find the terminator */ |
742 | for (; s < PL_bufend; s++,to++) { |
743 | /* embedded newlines increment the line count */ |
744 | if (*s == '\n' && !PL_rsfp) |
745 | CopLINE_inc(PL_curcop); |
746 | /* backslashes can escape the open or closing characters */ |
747 | if (*s == '\\' && s+1 < PL_bufend) { |
748 | if (!keep_quoted && |
749 | ((s[1] == PL_multi_open) || (s[1] == PL_multi_close))) |
750 | s++; |
751 | else |
752 | *to++ = *s++; |
753 | } |
754 | /* allow nested opens and closes */ |
755 | else if (*s == PL_multi_close && --brackets <= 0) |
756 | break; |
757 | else if (*s == PL_multi_open) |
758 | brackets++; |
759 | else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF) |
760 | has_utf8 = TRUE; |
761 | *to = *s; |
762 | } |
763 | } |
764 | /* terminate the copied string and update the sv's end-of-string */ |
765 | *to = '\0'; |
766 | SvCUR_set(sv, to - SvPVX_const(sv)); |
767 | |
768 | /* |
769 | * this next chunk reads more into the buffer if we're not done yet |
770 | */ |
771 | |
772 | if (s < PL_bufend) |
773 | break; /* handle case where we are done yet :-) */ |
774 | |
775 | #ifndef PERL_STRICT_CR |
776 | if (to - SvPVX_const(sv) >= 2) { |
777 | if ((to[-2] == '\r' && to[-1] == '\n') || |
778 | (to[-2] == '\n' && to[-1] == '\r')) |
779 | { |
780 | to[-2] = '\n'; |
781 | to--; |
782 | SvCUR_set(sv, to - SvPVX_const(sv)); |
783 | } |
784 | else if (to[-1] == '\r') |
785 | to[-1] = '\n'; |
786 | } |
787 | else if (to - SvPVX_const(sv) == 1 && to[-1] == '\r') |
788 | to[-1] = '\n'; |
789 | #endif |
790 | |
791 | read_more_line: |
792 | /* if we're out of file, or a read fails, bail and reset the current |
793 | line marker so we can report where the unterminated string began |
794 | */ |
795 | if (!PL_rsfp || |
796 | !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) { |
797 | sv_free(sv); |
798 | CopLINE_set(PL_curcop, (line_t)PL_multi_start); |
799 | return Nullch; |
800 | } |
801 | /* we read a line, so increment our line counter */ |
802 | CopLINE_inc(PL_curcop); |
803 | |
804 | /* update debugger info */ |
805 | if (PERLDB_LINE && PL_curstash != PL_debstash) { |
806 | SV *sv = NEWSV(88,0); |
807 | |
808 | sv_upgrade(sv, SVt_PVMG); |
809 | sv_setsv(sv,PL_linestr); |
810 | (void)SvIOK_on(sv); |
811 | SvIV_set(sv, 0); |
812 | av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv); |
813 | } |
814 | |
815 | /* having changed the buffer, we must update PL_bufend */ |
816 | PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); |
817 | PL_last_lop = PL_last_uni = Nullch; |
818 | } |
819 | |
820 | /* at this point, we have successfully read the delimited string */ |
821 | |
822 | if (!PL_encoding || UTF) { |
823 | if (keep_delims) |
824 | sv_catpvn(sv, s, termlen); |
825 | s += termlen; |
826 | } |
827 | if (has_utf8 || PL_encoding) |
828 | SvUTF8_on(sv); |
829 | |
830 | PL_multi_end = CopLINE(PL_curcop); |
831 | |
832 | /* if we allocated too much space, give some back */ |
833 | if (SvCUR(sv) + 5 < SvLEN(sv)) { |
834 | SvLEN_set(sv, SvCUR(sv) + 1); |
71511651 |
835 | /* 5.8.8 uses SvPV_renew, no prior version actually has the damn thing (mst) */ |
836 | #ifdef PERL_5_8_8_PLUS |
0ba8c7aa |
837 | SvPV_renew(sv, SvLEN(sv)); |
71511651 |
838 | #else |
839 | Renew(SvPVX(sv), SvLEN(sv), char); |
840 | #endif |
0ba8c7aa |
841 | } |
842 | |
843 | /* decide whether this is the first or second quoted string we've read |
844 | for this op |
845 | */ |
846 | |
847 | if (PL_lex_stuff) |
848 | PL_lex_repl = sv; |
849 | else |
850 | PL_lex_stuff = sv; |
851 | return s; |
852 | } |
853 | |
854 | /* |
855 | * S_force_next |
856 | * When the lexer realizes it knows the next token (for instance, |
857 | * it is reordering tokens for the parser) then it can call S_force_next |
858 | * to know what token to return the next time the lexer is called. Caller |
859 | * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer |
860 | * handles the token correctly. |
861 | */ |
862 | |
863 | STATIC void |
864 | S_force_next(pTHX_ I32 type) |
865 | { |
52eaf5fa |
866 | #ifdef PERL_MAD |
6d6f4cad |
867 | dVAR; |
52eaf5fa |
868 | if (PL_curforce < 0) |
869 | start_force(PL_lasttoke); |
870 | PL_nexttoke[PL_curforce].next_type = type; |
871 | if (PL_lex_state != LEX_KNOWNEXT) |
872 | PL_lex_defer = PL_lex_state; |
873 | PL_lex_state = LEX_KNOWNEXT; |
874 | PL_lex_expect = PL_expect; |
875 | PL_curforce = -1; |
876 | #else |
0ba8c7aa |
877 | PL_nexttype[PL_nexttoke] = type; |
878 | PL_nexttoke++; |
879 | if (PL_lex_state != LEX_KNOWNEXT) { |
880 | PL_lex_defer = PL_lex_state; |
881 | PL_lex_expect = PL_expect; |
882 | PL_lex_state = LEX_KNOWNEXT; |
883 | } |
52eaf5fa |
884 | #endif |
0ba8c7aa |
885 | } |
92997e69 |
886 | |
887 | #define XFAKEBRACK 128 |
888 | |
889 | STATIC char * |
890 | S_scan_ident(pTHX_ register char *s, register const char *send, char *dest, STRLEN destlen, I32 ck_uni) |
891 | { |
892 | register char *d; |
893 | register char *e; |
894 | char *bracket = Nullch; |
895 | char funny = *s++; |
896 | |
897 | if (isSPACE(*s)) |
898 | s = skipspace(s); |
899 | d = dest; |
900 | e = d + destlen - 3; /* two-character token, ending NUL */ |
901 | if (isDIGIT(*s)) { |
902 | while (isDIGIT(*s)) { |
903 | if (d >= e) |
904 | Perl_croak(aTHX_ ident_too_long); |
905 | *d++ = *s++; |
906 | } |
907 | } |
908 | else { |
909 | for (;;) { |
910 | if (d >= e) |
911 | Perl_croak(aTHX_ ident_too_long); |
912 | if (isALNUM(*s)) /* UTF handled below */ |
913 | *d++ = *s++; |
914 | else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) { |
915 | *d++ = ':'; |
916 | *d++ = ':'; |
917 | s++; |
918 | } |
919 | else if (*s == ':' && s[1] == ':') { |
920 | *d++ = *s++; |
921 | *d++ = *s++; |
922 | } |
923 | else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) { |
924 | char *t = s + UTF8SKIP(s); |
925 | while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t)) |
926 | t += UTF8SKIP(t); |
927 | if (d + (t - s) > e) |
928 | Perl_croak(aTHX_ ident_too_long); |
929 | Copy(s, d, t - s, char); |
930 | d += t - s; |
931 | s = t; |
932 | } |
933 | else |
934 | break; |
935 | } |
936 | } |
937 | *d = '\0'; |
938 | d = dest; |
939 | if (*d) { |
940 | if (PL_lex_state != LEX_NORMAL) |
941 | PL_lex_state = LEX_INTERPENDMAYBE; |
942 | return s; |
943 | } |
944 | if (*s == '$' && s[1] && |
945 | (isALNUM_lazy_if(s+1,UTF) || s[1] == '$' || s[1] == '{' || strnEQ(s+1,"::",2)) ) |
946 | { |
947 | return s; |
948 | } |
949 | if (*s == '{') { |
950 | bracket = s; |
951 | s++; |
952 | } |
d6e7537a |
953 | /* we always call this with ck_uni == 0 (rafl) */ |
954 | /* |
92997e69 |
955 | else if (ck_uni) |
956 | check_uni(); |
d6e7537a |
957 | */ |
92997e69 |
958 | if (s < send) |
959 | *d = *s++; |
960 | d[1] = '\0'; |
961 | if (*d == '^' && *s && isCONTROLVAR(*s)) { |
962 | *d = toCTRL(*s); |
963 | s++; |
964 | } |
965 | if (bracket) { |
966 | if (isSPACE(s[-1])) { |
967 | while (s < send) { |
968 | const char ch = *s++; |
969 | if (!SPACE_OR_TAB(ch)) { |
970 | *d = ch; |
971 | break; |
972 | } |
973 | } |
974 | } |
975 | if (isIDFIRST_lazy_if(d,UTF)) { |
976 | d++; |
977 | if (UTF) { |
978 | e = s; |
979 | while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') { |
980 | e += UTF8SKIP(e); |
981 | while (e < send && UTF8_IS_CONTINUED(*e) && is_utf8_mark((U8*)e)) |
982 | e += UTF8SKIP(e); |
983 | } |
984 | Copy(s, d, e - s, char); |
985 | d += e - s; |
986 | s = e; |
987 | } |
988 | else { |
989 | while ((isALNUM(*s) || *s == ':') && d < e) |
990 | *d++ = *s++; |
991 | if (d >= e) |
992 | Perl_croak(aTHX_ ident_too_long); |
993 | } |
994 | *d = '\0'; |
995 | while (s < send && SPACE_OR_TAB(*s)) s++; |
996 | if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) { |
d6e7537a |
997 | /* we don't want perl to guess what is meant. the keyword |
998 | * parser decides that later. (rafl) |
999 | */ |
1000 | /* |
92997e69 |
1001 | if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) { |
1002 | const char *brack = *s == '[' ? "[...]" : "{...}"; |
1003 | Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS), |
1004 | "Ambiguous use of %c{%s%s} resolved to %c%s%s", |
1005 | funny, dest, brack, funny, dest, brack); |
1006 | } |
d6e7537a |
1007 | */ |
92997e69 |
1008 | bracket++; |
1009 | PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK); |
1010 | return s; |
1011 | } |
1012 | } |
1013 | /* Handle extended ${^Foo} variables |
1014 | * 1999-02-27 mjd-perl-patch@plover.com */ |
1015 | else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */ |
1016 | && isALNUM(*s)) |
1017 | { |
1018 | d++; |
1019 | while (isALNUM(*s) && d < e) { |
1020 | *d++ = *s++; |
1021 | } |
1022 | if (d >= e) |
1023 | Perl_croak(aTHX_ ident_too_long); |
1024 | *d = '\0'; |
1025 | } |
1026 | if (*s == '}') { |
1027 | s++; |
1028 | if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets) { |
1029 | PL_lex_state = LEX_INTERPEND; |
1030 | PL_expect = XREF; |
1031 | } |
1032 | if (funny == '#') |
1033 | funny = '@'; |
d6e7537a |
1034 | /* we don't want perl to guess what is meant. the keyword |
1035 | * parser decides that later. (rafl) |
1036 | */ |
1037 | /* |
92997e69 |
1038 | if (PL_lex_state == LEX_NORMAL) { |
1039 | if (ckWARN(WARN_AMBIGUOUS) && |
1040 | (keyword(dest, d - dest) || get_cv(dest, FALSE))) |
1041 | { |
1042 | Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS), |
1043 | "Ambiguous use of %c{%s} resolved to %c%s", |
1044 | funny, dest, funny, dest); |
1045 | } |
1046 | } |
d6e7537a |
1047 | */ |
92997e69 |
1048 | } |
1049 | else { |
1050 | s = bracket; /* let the parser handle it */ |
1051 | *dest = '\0'; |
1052 | } |
1053 | } |
d6e7537a |
1054 | /* don't intuit. we really just want the string. (rafl) */ |
1055 | /* |
92997e69 |
1056 | else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s)) |
1057 | PL_lex_state = LEX_INTERPEND; |
d6e7537a |
1058 | */ |
92997e69 |
1059 | return s; |
1060 | } |