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 | |
20 | /* the following #defines are stolen from assorted headers, not toke.c */ |
21 | |
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 MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t), |
25 | #define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)safemalloc((MEM_SIZE)((n)*sizeof(t))))) |
26 | |
27 | |
28 | #define LEX_NORMAL 10 /* normal code (ie not within "...") */ |
29 | #define LEX_INTERPNORMAL 9 /* code within a string, eg "$foo[$x+1]" */ |
30 | #define LEX_INTERPCASEMOD 8 /* expecting a \U, \Q or \E etc */ |
31 | #define LEX_INTERPPUSH 7 /* starting a new sublex parse level */ |
32 | #define LEX_INTERPSTART 6 /* expecting the start of a $var */ |
33 | |
34 | /* at end of code, eg "$x" followed by: */ |
35 | #define LEX_INTERPEND 5 /* ... eg not one of [, { or -> */ |
36 | #define LEX_INTERPENDMAYBE 4 /* ... eg one of [, { or -> */ |
37 | |
38 | #define LEX_INTERPCONCAT 3 /* expecting anything, eg at start of |
39 | string or after \E, $foo, etc */ |
40 | #define LEX_INTERPCONST 2 /* NOT USED */ |
41 | #define LEX_FORMLINE 1 /* expecting a format line */ |
42 | #define LEX_KNOWNEXT 0 /* next token known; just return it */ |
43 | |
44 | static const char ident_too_long[] = |
45 | "Identifier too long"; |
46 | static const char c_without_g[] = |
47 | "Use of /c modifier is meaningless without /g"; |
48 | static const char c_in_subst[] = |
49 | "Use of /c modifier is meaningless in s///"; |
50 | |
51 | #ifdef USE_UTF8_SCRIPTS |
52 | # define UTF (!IN_BYTES) |
53 | #else |
54 | # define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || (PL_hints & HINT_UTF8)) |
55 | #endif |
56 | |
57 | /* Invoke the idxth filter function for the current rsfp. */ |
58 | /* maxlen 0 = read one text line */ |
59 | I32 |
60 | Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen) |
61 | { |
62 | filter_t funcp; |
63 | SV *datasv = NULL; |
64 | |
65 | if (!PL_rsfp_filters) |
66 | return -1; |
67 | if (idx > AvFILLp(PL_rsfp_filters)) { /* Any more filters? */ |
68 | /* Provide a default input filter to make life easy. */ |
69 | /* Note that we append to the line. This is handy. */ |
70 | DEBUG_P(PerlIO_printf(Perl_debug_log, |
71 | "filter_read %d: from rsfp\n", idx)); |
72 | if (maxlen) { |
73 | /* Want a block */ |
74 | int len ; |
75 | const int old_len = SvCUR(buf_sv); |
76 | |
77 | /* ensure buf_sv is large enough */ |
78 | SvGROW(buf_sv, (STRLEN)(old_len + maxlen)) ; |
79 | if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){ |
80 | if (PerlIO_error(PL_rsfp)) |
81 | return -1; /* error */ |
82 | else |
83 | return 0 ; /* end of file */ |
84 | } |
85 | SvCUR_set(buf_sv, old_len + len) ; |
86 | } else { |
87 | /* Want a line */ |
88 | if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) { |
89 | if (PerlIO_error(PL_rsfp)) |
90 | return -1; /* error */ |
91 | else |
92 | return 0 ; /* end of file */ |
93 | } |
94 | } |
95 | return SvCUR(buf_sv); |
96 | } |
97 | /* Skip this filter slot if filter has been deleted */ |
98 | if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef) { |
99 | DEBUG_P(PerlIO_printf(Perl_debug_log, |
100 | "filter_read %d: skipped (filter deleted)\n", |
101 | idx)); |
102 | return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */ |
103 | } |
104 | /* Get function pointer hidden within datasv */ |
105 | funcp = DPTR2FPTR(filter_t, IoANY(datasv)); |
106 | DEBUG_P(PerlIO_printf(Perl_debug_log, |
107 | "filter_read %d: via function %p (%s)\n", |
108 | idx, datasv, SvPV_nolen_const(datasv))); |
109 | /* Call function. The function is expected to */ |
110 | /* call "FILTER_READ(idx+1, buf_sv)" first. */ |
111 | /* Return: <0:error, =0:eof, >0:not eof */ |
112 | return (*funcp)(aTHX_ idx, buf_sv, maxlen); |
113 | } |
114 | |
115 | STATIC char * |
116 | S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append) |
117 | { |
118 | #ifdef PERL_CR_FILTER |
119 | if (!PL_rsfp_filters) { |
120 | filter_add(S_cr_textfilter,NULL); |
121 | } |
122 | #endif |
123 | if (PL_rsfp_filters) { |
124 | if (!append) |
125 | SvCUR_set(sv, 0); /* start with empty line */ |
126 | if (FILTER_READ(0, sv, 0) > 0) |
127 | return ( SvPVX(sv) ) ; |
128 | else |
129 | return Nullch ; |
130 | } |
131 | else |
132 | return (sv_gets(sv, fp, append)); |
133 | } |
134 | |
135 | /* |
136 | * S_skipspace |
137 | * Called to gobble the appropriate amount and type of whitespace. |
138 | * Skips comments as well. |
139 | */ |
140 | |
141 | STATIC char * |
142 | S_skipspace(pTHX_ register char *s) |
143 | { |
144 | if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) { |
145 | while (s < PL_bufend && SPACE_OR_TAB(*s)) |
146 | s++; |
147 | return s; |
148 | } |
149 | for (;;) { |
150 | STRLEN prevlen; |
151 | SSize_t oldprevlen, oldoldprevlen; |
152 | SSize_t oldloplen = 0, oldunilen = 0; |
153 | while (s < PL_bufend && isSPACE(*s)) { |
154 | if (*s++ == '\n' && PL_in_eval && !PL_rsfp) |
155 | incline(s); |
156 | } |
157 | |
158 | /* comment */ |
159 | if (s < PL_bufend && *s == '#') { |
160 | while (s < PL_bufend && *s != '\n') |
161 | s++; |
162 | if (s < PL_bufend) { |
163 | s++; |
164 | if (PL_in_eval && !PL_rsfp) { |
165 | incline(s); |
166 | continue; |
167 | } |
168 | } |
169 | } |
170 | |
171 | /* only continue to recharge the buffer if we're at the end |
172 | * of the buffer, we're not reading from a source filter, and |
173 | * we're in normal lexing mode |
174 | */ |
175 | if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat || |
176 | PL_lex_state == LEX_FORMLINE) |
177 | return s; |
178 | |
179 | /* try to recharge the buffer */ |
180 | if ((s = filter_gets(PL_linestr, PL_rsfp, |
181 | (prevlen = SvCUR(PL_linestr)))) == Nullch) |
182 | { |
183 | /* end of file. Add on the -p or -n magic */ |
184 | if (PL_minus_p) { |
185 | sv_setpv(PL_linestr, |
186 | ";}continue{print or die qq(-p destination: $!\\n);}"); |
187 | PL_minus_n = PL_minus_p = 0; |
188 | } |
189 | else if (PL_minus_n) { |
190 | sv_setpvn(PL_linestr, ";}", 2); |
191 | PL_minus_n = 0; |
192 | } |
193 | else |
194 | sv_setpvn(PL_linestr,";", 1); |
195 | |
196 | /* reset variables for next time we lex */ |
197 | PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart |
198 | = SvPVX(PL_linestr); |
199 | PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); |
200 | PL_last_lop = PL_last_uni = Nullch; |
201 | |
202 | /* Close the filehandle. Could be from -P preprocessor, |
203 | * STDIN, or a regular file. If we were reading code from |
204 | * STDIN (because the commandline held no -e or filename) |
205 | * then we don't close it, we reset it so the code can |
206 | * read from STDIN too. |
207 | */ |
208 | |
209 | if (PL_preprocess && !PL_in_eval) |
210 | (void)PerlProc_pclose(PL_rsfp); |
211 | else if ((PerlIO*)PL_rsfp == PerlIO_stdin()) |
212 | PerlIO_clearerr(PL_rsfp); |
213 | else |
214 | (void)PerlIO_close(PL_rsfp); |
215 | PL_rsfp = Nullfp; |
216 | return s; |
217 | } |
218 | |
219 | /* not at end of file, so we only read another line */ |
220 | /* make corresponding updates to old pointers, for yyerror() */ |
221 | oldprevlen = PL_oldbufptr - PL_bufend; |
222 | oldoldprevlen = PL_oldoldbufptr - PL_bufend; |
223 | if (PL_last_uni) |
224 | oldunilen = PL_last_uni - PL_bufend; |
225 | if (PL_last_lop) |
226 | oldloplen = PL_last_lop - PL_bufend; |
227 | PL_linestart = PL_bufptr = s + prevlen; |
228 | PL_bufend = s + SvCUR(PL_linestr); |
229 | s = PL_bufptr; |
230 | PL_oldbufptr = s + oldprevlen; |
231 | PL_oldoldbufptr = s + oldoldprevlen; |
232 | if (PL_last_uni) |
233 | PL_last_uni = s + oldunilen; |
234 | if (PL_last_lop) |
235 | PL_last_lop = s + oldloplen; |
236 | incline(s); |
237 | |
238 | /* debugger active and we're not compiling the debugger code, |
239 | * so store the line into the debugger's array of lines |
240 | */ |
241 | if (PERLDB_LINE && PL_curstash != PL_debstash) { |
242 | SV * const sv = NEWSV(85,0); |
243 | |
244 | sv_upgrade(sv, SVt_PVMG); |
245 | sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr); |
246 | (void)SvIOK_on(sv); |
247 | SvIV_set(sv, 0); |
248 | av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv); |
249 | } |
250 | } |
251 | } |
252 | |
253 | STATIC char * |
254 | S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp) |
255 | { |
256 | register char *d = dest; |
257 | register char * const e = d + destlen - 3; /* two-character token, ending NUL */ |
258 | for (;;) { |
259 | if (d >= e) |
260 | Perl_croak(aTHX_ ident_too_long); |
261 | if (isALNUM(*s)) /* UTF handled below */ |
262 | *d++ = *s++; |
263 | else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) { |
264 | *d++ = ':'; |
265 | *d++ = ':'; |
266 | s++; |
267 | } |
268 | else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') { |
269 | *d++ = *s++; |
270 | *d++ = *s++; |
271 | } |
272 | else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) { |
273 | char *t = s + UTF8SKIP(s); |
274 | while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t)) |
275 | t += UTF8SKIP(t); |
276 | if (d + (t - s) > e) |
277 | Perl_croak(aTHX_ ident_too_long); |
278 | Copy(s, d, t - s, char); |
279 | d += t - s; |
280 | s = t; |
281 | } |
282 | else { |
283 | *d = '\0'; |
284 | *slp = d - dest; |
285 | return s; |
286 | } |
287 | } |
288 | } |
289 | |
290 | /* |
291 | * S_incline |
292 | * This subroutine has nothing to do with tilting, whether at windmills |
293 | * or pinball tables. Its name is short for "increment line". It |
294 | * increments the current line number in CopLINE(PL_curcop) and checks |
295 | * to see whether the line starts with a comment of the form |
296 | * # line 500 "foo.pm" |
297 | * If so, it sets the current line number and file to the values in the comment. |
298 | */ |
299 | |
300 | STATIC void |
301 | S_incline(pTHX_ char *s) |
302 | { |
303 | char *t; |
304 | char *n; |
305 | char *e; |
306 | char ch; |
307 | |
308 | CopLINE_inc(PL_curcop); |
309 | if (*s++ != '#') |
310 | return; |
311 | while (SPACE_OR_TAB(*s)) s++; |
312 | if (strnEQ(s, "line", 4)) |
313 | s += 4; |
314 | else |
315 | return; |
316 | if (SPACE_OR_TAB(*s)) |
317 | s++; |
318 | else |
319 | return; |
320 | while (SPACE_OR_TAB(*s)) s++; |
321 | if (!isDIGIT(*s)) |
322 | return; |
323 | n = s; |
324 | while (isDIGIT(*s)) |
325 | s++; |
326 | while (SPACE_OR_TAB(*s)) |
327 | s++; |
328 | if (*s == '"' && (t = strchr(s+1, '"'))) { |
329 | s++; |
330 | e = t + 1; |
331 | } |
332 | else { |
333 | for (t = s; !isSPACE(*t); t++) ; |
334 | e = t; |
335 | } |
336 | while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f') |
337 | e++; |
338 | if (*e != '\n' && *e != '\0') |
339 | return; /* false alarm */ |
340 | |
341 | ch = *t; |
342 | *t = '\0'; |
343 | if (t - s > 0) { |
344 | #ifndef USE_ITHREADS |
345 | const char *cf = CopFILE(PL_curcop); |
346 | if (cf && strlen(cf) > 7 && strnEQ(cf, "(eval ", 6)) { |
347 | /* must copy *{"::_<(eval N)[oldfilename:L]"} |
348 | * to *{"::_<newfilename"} */ |
349 | char smallbuf[256], smallbuf2[256]; |
350 | char *tmpbuf, *tmpbuf2; |
351 | GV **gvp, *gv2; |
352 | STRLEN tmplen = strlen(cf); |
353 | STRLEN tmplen2 = strlen(s); |
354 | if (tmplen + 3 < sizeof smallbuf) |
355 | tmpbuf = smallbuf; |
356 | else |
357 | Newx(tmpbuf, tmplen + 3, char); |
358 | if (tmplen2 + 3 < sizeof smallbuf2) |
359 | tmpbuf2 = smallbuf2; |
360 | else |
361 | Newx(tmpbuf2, tmplen2 + 3, char); |
362 | tmpbuf[0] = tmpbuf2[0] = '_'; |
363 | tmpbuf[1] = tmpbuf2[1] = '<'; |
364 | memcpy(tmpbuf + 2, cf, ++tmplen); |
365 | memcpy(tmpbuf2 + 2, s, ++tmplen2); |
366 | ++tmplen; ++tmplen2; |
367 | gvp = (GV**)hv_fetch(PL_defstash, tmpbuf, tmplen, FALSE); |
368 | if (gvp) { |
369 | gv2 = *(GV**)hv_fetch(PL_defstash, tmpbuf2, tmplen2, TRUE); |
370 | if (!isGV(gv2)) |
371 | gv_init(gv2, PL_defstash, tmpbuf2, tmplen2, FALSE); |
372 | /* adjust ${"::_<newfilename"} to store the new file name */ |
373 | GvSV(gv2) = newSVpvn(tmpbuf2 + 2, tmplen2 - 2); |
374 | GvHV(gv2) = (HV*)SvREFCNT_inc(GvHV(*gvp)); |
375 | GvAV(gv2) = (AV*)SvREFCNT_inc(GvAV(*gvp)); |
376 | } |
377 | if (tmpbuf != smallbuf) Safefree(tmpbuf); |
378 | if (tmpbuf2 != smallbuf2) Safefree(tmpbuf2); |
379 | } |
380 | #endif |
381 | CopFILE_free(PL_curcop); |
382 | CopFILE_set(PL_curcop, s); |
383 | } |
384 | *t = ch; |
385 | CopLINE_set(PL_curcop, atoi(n)-1); |
386 | } |