debug flag, extra test
[p5sagit/Devel-Declare.git] / stolen_chunk_of_toke.c
CommitLineData
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 */
c630715a 24#define PTR2nat(p) (PTRV)(p) /* pointer to integer of PTRSIZE */
e807ee50 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
c630715a 28/* On MacOS, respect nonbreaking spaces */
29#ifdef MACOS_TRADITIONAL
30#define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
31#else
32#define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
33#endif
e807ee50 34
35#define LEX_NORMAL 10 /* normal code (ie not within "...") */
36#define LEX_INTERPNORMAL 9 /* code within a string, eg "$foo[$x+1]" */
37#define LEX_INTERPCASEMOD 8 /* expecting a \U, \Q or \E etc */
38#define LEX_INTERPPUSH 7 /* starting a new sublex parse level */
39#define LEX_INTERPSTART 6 /* expecting the start of a $var */
40
41 /* at end of code, eg "$x" followed by: */
42#define LEX_INTERPEND 5 /* ... eg not one of [, { or -> */
43#define LEX_INTERPENDMAYBE 4 /* ... eg one of [, { or -> */
44
45#define LEX_INTERPCONCAT 3 /* expecting anything, eg at start of
46 string or after \E, $foo, etc */
47#define LEX_INTERPCONST 2 /* NOT USED */
48#define LEX_FORMLINE 1 /* expecting a format line */
49#define LEX_KNOWNEXT 0 /* next token known; just return it */
50
51static const char ident_too_long[] =
52 "Identifier too long";
53static const char c_without_g[] =
54 "Use of /c modifier is meaningless without /g";
55static const char c_in_subst[] =
56 "Use of /c modifier is meaningless in s///";
57
58#ifdef USE_UTF8_SCRIPTS
59# define UTF (!IN_BYTES)
60#else
61# define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || (PL_hints & HINT_UTF8))
62#endif
63
64/* Invoke the idxth filter function for the current rsfp. */
65/* maxlen 0 = read one text line */
66I32
67Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
68{
69 filter_t funcp;
70 SV *datasv = NULL;
71
72 if (!PL_rsfp_filters)
73 return -1;
74 if (idx > AvFILLp(PL_rsfp_filters)) { /* Any more filters? */
75 /* Provide a default input filter to make life easy. */
76 /* Note that we append to the line. This is handy. */
77 DEBUG_P(PerlIO_printf(Perl_debug_log,
78 "filter_read %d: from rsfp\n", idx));
79 if (maxlen) {
80 /* Want a block */
81 int len ;
82 const int old_len = SvCUR(buf_sv);
83
84 /* ensure buf_sv is large enough */
85 SvGROW(buf_sv, (STRLEN)(old_len + maxlen)) ;
86 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
87 if (PerlIO_error(PL_rsfp))
88 return -1; /* error */
89 else
90 return 0 ; /* end of file */
91 }
92 SvCUR_set(buf_sv, old_len + len) ;
93 } else {
94 /* Want a line */
95 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
96 if (PerlIO_error(PL_rsfp))
97 return -1; /* error */
98 else
99 return 0 ; /* end of file */
100 }
101 }
102 return SvCUR(buf_sv);
103 }
104 /* Skip this filter slot if filter has been deleted */
105 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef) {
106 DEBUG_P(PerlIO_printf(Perl_debug_log,
107 "filter_read %d: skipped (filter deleted)\n",
108 idx));
109 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
110 }
111 /* Get function pointer hidden within datasv */
112 funcp = DPTR2FPTR(filter_t, IoANY(datasv));
113 DEBUG_P(PerlIO_printf(Perl_debug_log,
114 "filter_read %d: via function %p (%s)\n",
115 idx, datasv, SvPV_nolen_const(datasv)));
116 /* Call function. The function is expected to */
117 /* call "FILTER_READ(idx+1, buf_sv)" first. */
118 /* Return: <0:error, =0:eof, >0:not eof */
119 return (*funcp)(aTHX_ idx, buf_sv, maxlen);
120}
121
122STATIC char *
123S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
124{
125#ifdef PERL_CR_FILTER
126 if (!PL_rsfp_filters) {
127 filter_add(S_cr_textfilter,NULL);
128 }
129#endif
130 if (PL_rsfp_filters) {
131 if (!append)
132 SvCUR_set(sv, 0); /* start with empty line */
133 if (FILTER_READ(0, sv, 0) > 0)
134 return ( SvPVX(sv) ) ;
135 else
136 return Nullch ;
137 }
138 else
139 return (sv_gets(sv, fp, append));
140}
141
142/*
143 * S_skipspace
144 * Called to gobble the appropriate amount and type of whitespace.
145 * Skips comments as well.
146 */
147
148STATIC char *
149S_skipspace(pTHX_ register char *s)
150{
151 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
152 while (s < PL_bufend && SPACE_OR_TAB(*s))
153 s++;
154 return s;
155 }
156 for (;;) {
157 STRLEN prevlen;
158 SSize_t oldprevlen, oldoldprevlen;
159 SSize_t oldloplen = 0, oldunilen = 0;
160 while (s < PL_bufend && isSPACE(*s)) {
161 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
162 incline(s);
163 }
164
165 /* comment */
166 if (s < PL_bufend && *s == '#') {
167 while (s < PL_bufend && *s != '\n')
168 s++;
169 if (s < PL_bufend) {
170 s++;
171 if (PL_in_eval && !PL_rsfp) {
172 incline(s);
173 continue;
174 }
175 }
176 }
177
178 /* only continue to recharge the buffer if we're at the end
179 * of the buffer, we're not reading from a source filter, and
180 * we're in normal lexing mode
181 */
182 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
183 PL_lex_state == LEX_FORMLINE)
184 return s;
185
186 /* try to recharge the buffer */
187 if ((s = filter_gets(PL_linestr, PL_rsfp,
188 (prevlen = SvCUR(PL_linestr)))) == Nullch)
189 {
190 /* end of file. Add on the -p or -n magic */
191 if (PL_minus_p) {
192 sv_setpv(PL_linestr,
193 ";}continue{print or die qq(-p destination: $!\\n);}");
194 PL_minus_n = PL_minus_p = 0;
195 }
196 else if (PL_minus_n) {
197 sv_setpvn(PL_linestr, ";}", 2);
198 PL_minus_n = 0;
199 }
200 else
201 sv_setpvn(PL_linestr,";", 1);
202
203 /* reset variables for next time we lex */
204 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
205 = SvPVX(PL_linestr);
206 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
207 PL_last_lop = PL_last_uni = Nullch;
208
209 /* Close the filehandle. Could be from -P preprocessor,
210 * STDIN, or a regular file. If we were reading code from
211 * STDIN (because the commandline held no -e or filename)
212 * then we don't close it, we reset it so the code can
213 * read from STDIN too.
214 */
215
216 if (PL_preprocess && !PL_in_eval)
217 (void)PerlProc_pclose(PL_rsfp);
218 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
219 PerlIO_clearerr(PL_rsfp);
220 else
221 (void)PerlIO_close(PL_rsfp);
222 PL_rsfp = Nullfp;
223 return s;
224 }
225
226 /* not at end of file, so we only read another line */
227 /* make corresponding updates to old pointers, for yyerror() */
228 oldprevlen = PL_oldbufptr - PL_bufend;
229 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
230 if (PL_last_uni)
231 oldunilen = PL_last_uni - PL_bufend;
232 if (PL_last_lop)
233 oldloplen = PL_last_lop - PL_bufend;
234 PL_linestart = PL_bufptr = s + prevlen;
235 PL_bufend = s + SvCUR(PL_linestr);
236 s = PL_bufptr;
237 PL_oldbufptr = s + oldprevlen;
238 PL_oldoldbufptr = s + oldoldprevlen;
239 if (PL_last_uni)
240 PL_last_uni = s + oldunilen;
241 if (PL_last_lop)
242 PL_last_lop = s + oldloplen;
243 incline(s);
244
245 /* debugger active and we're not compiling the debugger code,
246 * so store the line into the debugger's array of lines
247 */
248 if (PERLDB_LINE && PL_curstash != PL_debstash) {
249 SV * const sv = NEWSV(85,0);
250
251 sv_upgrade(sv, SVt_PVMG);
252 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
253 (void)SvIOK_on(sv);
254 SvIV_set(sv, 0);
255 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
256 }
257 }
258}
259
260STATIC char *
261S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
262{
263 register char *d = dest;
264 register char * const e = d + destlen - 3; /* two-character token, ending NUL */
265 for (;;) {
266 if (d >= e)
267 Perl_croak(aTHX_ ident_too_long);
268 if (isALNUM(*s)) /* UTF handled below */
269 *d++ = *s++;
270 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
271 *d++ = ':';
272 *d++ = ':';
273 s++;
274 }
275 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
276 *d++ = *s++;
277 *d++ = *s++;
278 }
279 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
280 char *t = s + UTF8SKIP(s);
281 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
282 t += UTF8SKIP(t);
283 if (d + (t - s) > e)
284 Perl_croak(aTHX_ ident_too_long);
285 Copy(s, d, t - s, char);
286 d += t - s;
287 s = t;
288 }
289 else {
290 *d = '\0';
291 *slp = d - dest;
292 return s;
293 }
294 }
295}
296
297/*
298 * S_incline
299 * This subroutine has nothing to do with tilting, whether at windmills
300 * or pinball tables. Its name is short for "increment line". It
301 * increments the current line number in CopLINE(PL_curcop) and checks
302 * to see whether the line starts with a comment of the form
303 * # line 500 "foo.pm"
304 * If so, it sets the current line number and file to the values in the comment.
305 */
306
307STATIC void
308S_incline(pTHX_ char *s)
309{
310 char *t;
311 char *n;
312 char *e;
313 char ch;
314
315 CopLINE_inc(PL_curcop);
316 if (*s++ != '#')
317 return;
318 while (SPACE_OR_TAB(*s)) s++;
319 if (strnEQ(s, "line", 4))
320 s += 4;
321 else
322 return;
323 if (SPACE_OR_TAB(*s))
324 s++;
325 else
326 return;
327 while (SPACE_OR_TAB(*s)) s++;
328 if (!isDIGIT(*s))
329 return;
330 n = s;
331 while (isDIGIT(*s))
332 s++;
333 while (SPACE_OR_TAB(*s))
334 s++;
335 if (*s == '"' && (t = strchr(s+1, '"'))) {
336 s++;
337 e = t + 1;
338 }
339 else {
340 for (t = s; !isSPACE(*t); t++) ;
341 e = t;
342 }
343 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
344 e++;
345 if (*e != '\n' && *e != '\0')
346 return; /* false alarm */
347
348 ch = *t;
349 *t = '\0';
350 if (t - s > 0) {
351#ifndef USE_ITHREADS
352 const char *cf = CopFILE(PL_curcop);
353 if (cf && strlen(cf) > 7 && strnEQ(cf, "(eval ", 6)) {
354 /* must copy *{"::_<(eval N)[oldfilename:L]"}
355 * to *{"::_<newfilename"} */
356 char smallbuf[256], smallbuf2[256];
357 char *tmpbuf, *tmpbuf2;
358 GV **gvp, *gv2;
359 STRLEN tmplen = strlen(cf);
360 STRLEN tmplen2 = strlen(s);
361 if (tmplen + 3 < sizeof smallbuf)
362 tmpbuf = smallbuf;
363 else
364 Newx(tmpbuf, tmplen + 3, char);
365 if (tmplen2 + 3 < sizeof smallbuf2)
366 tmpbuf2 = smallbuf2;
367 else
368 Newx(tmpbuf2, tmplen2 + 3, char);
369 tmpbuf[0] = tmpbuf2[0] = '_';
370 tmpbuf[1] = tmpbuf2[1] = '<';
371 memcpy(tmpbuf + 2, cf, ++tmplen);
372 memcpy(tmpbuf2 + 2, s, ++tmplen2);
373 ++tmplen; ++tmplen2;
374 gvp = (GV**)hv_fetch(PL_defstash, tmpbuf, tmplen, FALSE);
375 if (gvp) {
376 gv2 = *(GV**)hv_fetch(PL_defstash, tmpbuf2, tmplen2, TRUE);
377 if (!isGV(gv2))
378 gv_init(gv2, PL_defstash, tmpbuf2, tmplen2, FALSE);
379 /* adjust ${"::_<newfilename"} to store the new file name */
380 GvSV(gv2) = newSVpvn(tmpbuf2 + 2, tmplen2 - 2);
381 GvHV(gv2) = (HV*)SvREFCNT_inc(GvHV(*gvp));
382 GvAV(gv2) = (AV*)SvREFCNT_inc(GvAV(*gvp));
383 }
384 if (tmpbuf != smallbuf) Safefree(tmpbuf);
385 if (tmpbuf2 != smallbuf2) Safefree(tmpbuf2);
386 }
387#endif
388 CopFILE_free(PL_curcop);
389 CopFILE_set(PL_curcop, s);
390 }
391 *t = ch;
392 CopLINE_set(PL_curcop, atoi(n)-1);
393}