This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
[PATCH] Re: chomp/chop prototype changed?
[perl5.git] / toke.c
CommitLineData
a0d0e21e 1/* toke.c
a687059c 2 *
bc89e66f 3 * Copyright (c) 1991-2001, Larry Wall
a687059c 4 *
d48672a2
LW
5 * You may distribute under the terms of either the GNU General Public
6 * License or the Artistic License, as specified in the README file.
378cc40b 7 *
a0d0e21e
LW
8 */
9
10/*
11 * "It all comes from here, the stench and the peril." --Frodo
378cc40b
LW
12 */
13
9cbb5ea2
GS
14/*
15 * This file is the lexer for Perl. It's closely linked to the
4e553d73 16 * parser, perly.y.
ffb4593c
NT
17 *
18 * The main routine is yylex(), which returns the next token.
19 */
20
378cc40b 21#include "EXTERN.h"
864dbfa3 22#define PERL_IN_TOKE_C
378cc40b 23#include "perl.h"
378cc40b 24
d3b6f988
GS
25#define yychar PL_yychar
26#define yylval PL_yylval
27
fc36a67e 28static char ident_too_long[] = "Identifier too long";
8903cb82 29
acfe0abc 30static void restore_rsfp(pTHX_ void *f);
6e3aabd6 31#ifndef PERL_NO_UTF16_FILTER
acfe0abc
GS
32static I32 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen);
33static I32 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen);
6e3aabd6 34#endif
51371543 35
9059aa12
LW
36#define XFAKEBRACK 128
37#define XENUMMASK 127
38
39e02b42
JH
39#ifdef USE_UTF8_SCRIPTS
40# define UTF (!IN_BYTES)
2b9d42f0 41#else
39e02b42
JH
42# ifdef EBCDIC /* For now 'use utf8' does not affect tokenizer on EBCDIC */
43# define UTF (PL_linestr && DO_UTF8(PL_linestr))
44# else
45# define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || (PL_hints & HINT_UTF8))
46# endif
2b9d42f0 47#endif
a0ed51b3 48
61f0cdd9 49/* In variables named $^X, these are the legal values for X.
2b92dfce
GS
50 * 1999-02-27 mjd-perl-patch@plover.com */
51#define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
52
bf4acbe4
GS
53/* On MacOS, respect nonbreaking spaces */
54#ifdef MACOS_TRADITIONAL
55#define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
56#else
57#define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
58#endif
59
ffb4593c
NT
60/* LEX_* are values for PL_lex_state, the state of the lexer.
61 * They are arranged oddly so that the guard on the switch statement
79072805
LW
62 * can get by with a single comparison (if the compiler is smart enough).
63 */
64
fb73857a
PP
65/* #define LEX_NOTPARSING 11 is done in perl.h. */
66
55497cff
PP
67#define LEX_NORMAL 10
68#define LEX_INTERPNORMAL 9
69#define LEX_INTERPCASEMOD 8
70#define LEX_INTERPPUSH 7
71#define LEX_INTERPSTART 6
72#define LEX_INTERPEND 5
73#define LEX_INTERPENDMAYBE 4
74#define LEX_INTERPCONCAT 3
75#define LEX_INTERPCONST 2
76#define LEX_FORMLINE 1
77#define LEX_KNOWNEXT 0
79072805 78
79072805
LW
79#ifdef ff_next
80#undef ff_next
d48672a2
LW
81#endif
82
a1a0e61e 83#ifdef USE_PURE_BISON
dba4d153
JH
84# ifndef YYMAXLEVEL
85# define YYMAXLEVEL 100
86# endif
20141f0e
RI
87YYSTYPE* yylval_pointer[YYMAXLEVEL];
88int* yychar_pointer[YYMAXLEVEL];
6f202aea 89int yyactlevel = -1;
22c35a8c
GS
90# undef yylval
91# undef yychar
20141f0e
RI
92# define yylval (*yylval_pointer[yyactlevel])
93# define yychar (*yychar_pointer[yyactlevel])
94# define PERL_YYLEX_PARAM yylval_pointer[yyactlevel],yychar_pointer[yyactlevel]
4e553d73 95# undef yylex
dba4d153 96# define yylex() Perl_yylex_r(aTHX_ yylval_pointer[yyactlevel],yychar_pointer[yyactlevel])
a1a0e61e
TD
97#endif
98
79072805 99#include "keywords.h"
fe14fcc3 100
ffb4593c
NT
101/* CLINE is a macro that ensures PL_copline has a sane value */
102
ae986130
LW
103#ifdef CLINE
104#undef CLINE
105#endif
57843af0 106#define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
3280af22 107
ffb4593c
NT
108/*
109 * Convenience functions to return different tokens and prime the
9cbb5ea2 110 * lexer for the next token. They all take an argument.
ffb4593c
NT
111 *
112 * TOKEN : generic token (used for '(', DOLSHARP, etc)
113 * OPERATOR : generic operator
114 * AOPERATOR : assignment operator
115 * PREBLOCK : beginning the block after an if, while, foreach, ...
116 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
117 * PREREF : *EXPR where EXPR is not a simple identifier
118 * TERM : expression term
119 * LOOPX : loop exiting command (goto, last, dump, etc)
120 * FTST : file test operator
121 * FUN0 : zero-argument function
2d2e263d 122 * FUN1 : not used, except for not, which isn't a UNIOP
ffb4593c
NT
123 * BOop : bitwise or or xor
124 * BAop : bitwise and
125 * SHop : shift operator
126 * PWop : power operator
9cbb5ea2 127 * PMop : pattern-matching operator
ffb4593c
NT
128 * Aop : addition-level operator
129 * Mop : multiplication-level operator
130 * Eop : equality-testing operator
e5edeb50 131 * Rop : relational operator <= != gt
ffb4593c
NT
132 *
133 * Also see LOP and lop() below.
134 */
135
075953c3
JH
136/* Note that REPORT() and REPORT2() will be expressions that supply
137 * their own trailing comma, not suitable for statements as such. */
998054bd 138#ifdef DEBUGGING /* Serve -DT. */
075953c3
JH
139# define REPORT(x,retval) tokereport(x,s,(int)retval),
140# define REPORT2(x,retval) tokereport(x,s, yylval.ival),
998054bd 141#else
075953c3
JH
142# define REPORT(x,retval)
143# define REPORT2(x,retval)
998054bd
SC
144#endif
145
075953c3
JH
146#define TOKEN(retval) return (REPORT2("token",retval) PL_bufptr = s,(int)retval)
147#define OPERATOR(retval) return (REPORT2("operator",retval) PL_expect = XTERM, PL_bufptr = s,(int)retval)
148#define AOPERATOR(retval) return ao((REPORT2("aop",retval) PL_expect = XTERM, PL_bufptr = s,(int)retval))
149#define PREBLOCK(retval) return (REPORT2("preblock",retval) PL_expect = XBLOCK,PL_bufptr = s,(int)retval)
150#define PRETERMBLOCK(retval) return (REPORT2("pretermblock",retval) PL_expect = XTERMBLOCK,PL_bufptr = s,(int)retval)
151#define PREREF(retval) return (REPORT2("preref",retval) PL_expect = XREF,PL_bufptr = s,(int)retval)
152#define TERM(retval) return (CLINE, REPORT2("term",retval) PL_expect = XOPERATOR, PL_bufptr = s,(int)retval)
153#define LOOPX(f) return(yylval.ival=f, REPORT("loopx",f) PL_expect = XTERM,PL_bufptr = s,(int)LOOPEX)
154#define FTST(f) return(yylval.ival=f, REPORT("ftst",f) PL_expect = XTERM,PL_bufptr = s,(int)UNIOP)
155#define FUN0(f) return(yylval.ival = f, REPORT("fun0",f) PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC0)
156#define FUN1(f) return(yylval.ival = f, REPORT("fun1",f) PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC1)
157#define BOop(f) return ao((yylval.ival=f, REPORT("bitorop",f) PL_expect = XTERM,PL_bufptr = s,(int)BITOROP))
158#define BAop(f) return ao((yylval.ival=f, REPORT("bitandop",f) PL_expect = XTERM,PL_bufptr = s,(int)BITANDOP))
159#define SHop(f) return ao((yylval.ival=f, REPORT("shiftop",f) PL_expect = XTERM,PL_bufptr = s,(int)SHIFTOP))
160#define PWop(f) return ao((yylval.ival=f, REPORT("powop",f) PL_expect = XTERM,PL_bufptr = s,(int)POWOP))
161#define PMop(f) return(yylval.ival=f, REPORT("matchop",f) PL_expect = XTERM,PL_bufptr = s,(int)MATCHOP)
162#define Aop(f) return ao((yylval.ival=f, REPORT("add",f) PL_expect = XTERM,PL_bufptr = s,(int)ADDOP))
163#define Mop(f) return ao((yylval.ival=f, REPORT("mul",f) PL_expect = XTERM,PL_bufptr = s,(int)MULOP))
164#define Eop(f) return(yylval.ival=f, REPORT("eq",f) PL_expect = XTERM,PL_bufptr = s,(int)EQOP)
165#define Rop(f) return(yylval.ival=f, REPORT("rel",f) PL_expect = XTERM,PL_bufptr = s,(int)RELOP)
2f3197b3 166
a687059c
LW
167/* This bit of chicanery makes a unary function followed by
168 * a parenthesis into a function with one argument, highest precedence.
169 */
2f3197b3 170#define UNI(f) return(yylval.ival = f, \
075953c3 171 REPORT("uni",f) \
3280af22
NIS
172 PL_expect = XTERM, \
173 PL_bufptr = s, \
174 PL_last_uni = PL_oldbufptr, \
175 PL_last_lop_op = f, \
a687059c
LW
176 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
177
79072805 178#define UNIBRACK(f) return(yylval.ival = f, \
075953c3 179 REPORT("uni",f) \
3280af22
NIS
180 PL_bufptr = s, \
181 PL_last_uni = PL_oldbufptr, \
79072805
LW
182 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
183
9f68db38 184/* grandfather return to old style */
3280af22 185#define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
79072805 186
8fa7f367
JH
187#ifdef DEBUGGING
188
2d00ba3b 189STATIC void
61b2116b 190S_tokereport(pTHX_ char *thing, char* s, I32 rv)
9041c2e3 191{
998054bd 192 DEBUG_T({
9c5ffd7c 193 SV* report = newSVpv(thing, 0);
29b291f7
RB
194 Perl_sv_catpvf(aTHX_ report, ":line %d:%"IVdf":", CopLINE(PL_curcop),
195 (IV)rv);
998054bd
SC
196
197 if (s - PL_bufptr > 0)
198 sv_catpvn(report, PL_bufptr, s - PL_bufptr);
199 else {
200 if (PL_oldbufptr && *PL_oldbufptr)
201 sv_catpv(report, PL_tokenbuf);
202 }
203 PerlIO_printf(Perl_debug_log, "### %s\n", SvPV_nolen(report));
5f80b19c 204 });
998054bd
SC
205}
206
8fa7f367
JH
207#endif
208
ffb4593c
NT
209/*
210 * S_ao
211 *
212 * This subroutine detects &&= and ||= and turns an ANDAND or OROR
213 * into an OP_ANDASSIGN or OP_ORASSIGN
214 */
215
76e3520e 216STATIC int
cea2e8a9 217S_ao(pTHX_ int toketype)
a0d0e21e 218{
3280af22
NIS
219 if (*PL_bufptr == '=') {
220 PL_bufptr++;
a0d0e21e
LW
221 if (toketype == ANDAND)
222 yylval.ival = OP_ANDASSIGN;
223 else if (toketype == OROR)
224 yylval.ival = OP_ORASSIGN;
225 toketype = ASSIGNOP;
226 }
227 return toketype;
228}
229
ffb4593c
NT
230/*
231 * S_no_op
232 * When Perl expects an operator and finds something else, no_op
233 * prints the warning. It always prints "<something> found where
234 * operator expected. It prints "Missing semicolon on previous line?"
235 * if the surprise occurs at the start of the line. "do you need to
236 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
237 * where the compiler doesn't know if foo is a method call or a function.
238 * It prints "Missing operator before end of line" if there's nothing
239 * after the missing operator, or "... before <...>" if there is something
240 * after the missing operator.
241 */
242
76e3520e 243STATIC void
cea2e8a9 244S_no_op(pTHX_ char *what, char *s)
463ee0b2 245{
3280af22
NIS
246 char *oldbp = PL_bufptr;
247 bool is_first = (PL_oldbufptr == PL_linestart);
68dc0745 248
1189a94a
GS
249 if (!s)
250 s = oldbp;
07c798fb 251 else
1189a94a 252 PL_bufptr = s;
cea2e8a9 253 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
748a9306 254 if (is_first)
cea2e8a9 255 Perl_warn(aTHX_ "\t(Missing semicolon on previous line?)\n");
7e2040f0 256 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
748a9306 257 char *t;
7e2040f0 258 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy_if(t,UTF) || *t == ':'); t++) ;
3280af22 259 if (t < PL_bufptr && isSPACE(*t))
cea2e8a9 260 Perl_warn(aTHX_ "\t(Do you need to predeclare %.*s?)\n",
3280af22 261 t - PL_oldoldbufptr, PL_oldoldbufptr);
748a9306 262 }
07c798fb
HS
263 else {
264 assert(s >= oldbp);
cea2e8a9 265 Perl_warn(aTHX_ "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp);
07c798fb 266 }
3280af22 267 PL_bufptr = oldbp;
8990e307
LW
268}
269
ffb4593c
NT
270/*
271 * S_missingterm
272 * Complain about missing quote/regexp/heredoc terminator.
273 * If it's called with (char *)NULL then it cauterizes the line buffer.
274 * If we're in a delimited string and the delimiter is a control
275 * character, it's reformatted into a two-char sequence like ^C.
276 * This is fatal.
277 */
278
76e3520e 279STATIC void
cea2e8a9 280S_missingterm(pTHX_ char *s)
8990e307
LW
281{
282 char tmpbuf[3];
283 char q;
284 if (s) {
285 char *nl = strrchr(s,'\n');
d2719217 286 if (nl)
8990e307
LW
287 *nl = '\0';
288 }
9d116dd7
JH
289 else if (
290#ifdef EBCDIC
291 iscntrl(PL_multi_close)
292#else
293 PL_multi_close < 32 || PL_multi_close == 127
294#endif
295 ) {
8990e307 296 *tmpbuf = '^';
3280af22 297 tmpbuf[1] = toCTRL(PL_multi_close);
8990e307
LW
298 s = "\\n";
299 tmpbuf[2] = '\0';
300 s = tmpbuf;
301 }
302 else {
3280af22 303 *tmpbuf = PL_multi_close;
8990e307
LW
304 tmpbuf[1] = '\0';
305 s = tmpbuf;
306 }
307 q = strchr(s,'"') ? '\'' : '"';
cea2e8a9 308 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
463ee0b2 309}
79072805 310
ffb4593c
NT
311/*
312 * Perl_deprecate
ffb4593c
NT
313 */
314
79072805 315void
864dbfa3 316Perl_deprecate(pTHX_ char *s)
a0d0e21e 317{
599cee73 318 if (ckWARN(WARN_DEPRECATED))
cea2e8a9 319 Perl_warner(aTHX_ WARN_DEPRECATED, "Use of %s is deprecated", s);
a0d0e21e
LW
320}
321
ffb4593c
NT
322/*
323 * depcom
9cbb5ea2 324 * Deprecate a comma-less variable list.
ffb4593c
NT
325 */
326
76e3520e 327STATIC void
cea2e8a9 328S_depcom(pTHX)
a0d0e21e
LW
329{
330 deprecate("comma-less variable list");
331}
332
ffb4593c 333/*
9cbb5ea2
GS
334 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
335 * utf16-to-utf8-reversed.
ffb4593c
NT
336 */
337
c39cd008
GS
338#ifdef PERL_CR_FILTER
339static void
340strip_return(SV *sv)
341{
342 register char *s = SvPVX(sv);
343 register char *e = s + SvCUR(sv);
344 /* outer loop optimized to do nothing if there are no CR-LFs */
345 while (s < e) {
346 if (*s++ == '\r' && *s == '\n') {
347 /* hit a CR-LF, need to copy the rest */
348 register char *d = s - 1;
349 *d++ = *s++;
350 while (s < e) {
351 if (*s == '\r' && s[1] == '\n')
352 s++;
353 *d++ = *s++;
354 }
355 SvCUR(sv) -= s - d;
356 return;
357 }
358 }
359}
a868473f 360
76e3520e 361STATIC I32
c39cd008 362S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
a868473f 363{
c39cd008
GS
364 I32 count = FILTER_READ(idx+1, sv, maxlen);
365 if (count > 0 && !maxlen)
366 strip_return(sv);
367 return count;
a868473f
NIS
368}
369#endif
370
ffb4593c
NT
371/*
372 * Perl_lex_start
9cbb5ea2
GS
373 * Initialize variables. Uses the Perl save_stack to save its state (for
374 * recursive calls to the parser).
ffb4593c
NT
375 */
376
a0d0e21e 377void
864dbfa3 378Perl_lex_start(pTHX_ SV *line)
79072805 379{
8990e307
LW
380 char *s;
381 STRLEN len;
382
3280af22
NIS
383 SAVEI32(PL_lex_dojoin);
384 SAVEI32(PL_lex_brackets);
3280af22
NIS
385 SAVEI32(PL_lex_casemods);
386 SAVEI32(PL_lex_starts);
387 SAVEI32(PL_lex_state);
7766f137 388 SAVEVPTR(PL_lex_inpat);
3280af22 389 SAVEI32(PL_lex_inwhat);
18b09519
GS
390 if (PL_lex_state == LEX_KNOWNEXT) {
391 I32 toke = PL_nexttoke;
392 while (--toke >= 0) {
393 SAVEI32(PL_nexttype[toke]);
394 SAVEVPTR(PL_nextval[toke]);
395 }
396 SAVEI32(PL_nexttoke);
18b09519 397 }
57843af0 398 SAVECOPLINE(PL_curcop);
3280af22
NIS
399 SAVEPPTR(PL_bufptr);
400 SAVEPPTR(PL_bufend);
401 SAVEPPTR(PL_oldbufptr);
402 SAVEPPTR(PL_oldoldbufptr);
207e3d1a
JH
403 SAVEPPTR(PL_last_lop);
404 SAVEPPTR(PL_last_uni);
3280af22
NIS
405 SAVEPPTR(PL_linestart);
406 SAVESPTR(PL_linestr);
407 SAVEPPTR(PL_lex_brackstack);
408 SAVEPPTR(PL_lex_casestack);
c76ac1ee 409 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
3280af22
NIS
410 SAVESPTR(PL_lex_stuff);
411 SAVEI32(PL_lex_defer);
09bef843 412 SAVEI32(PL_sublex_info.sub_inwhat);
3280af22 413 SAVESPTR(PL_lex_repl);
bebdddfc
GS
414 SAVEINT(PL_expect);
415 SAVEINT(PL_lex_expect);
3280af22
NIS
416
417 PL_lex_state = LEX_NORMAL;
418 PL_lex_defer = 0;
419 PL_expect = XSTATE;
420 PL_lex_brackets = 0;
3280af22
NIS
421 New(899, PL_lex_brackstack, 120, char);
422 New(899, PL_lex_casestack, 12, char);
423 SAVEFREEPV(PL_lex_brackstack);
424 SAVEFREEPV(PL_lex_casestack);
425 PL_lex_casemods = 0;
426 *PL_lex_casestack = '\0';
427 PL_lex_dojoin = 0;
428 PL_lex_starts = 0;
429 PL_lex_stuff = Nullsv;
430 PL_lex_repl = Nullsv;
431 PL_lex_inpat = 0;
76be56bc 432 PL_nexttoke = 0;
3280af22 433 PL_lex_inwhat = 0;
09bef843 434 PL_sublex_info.sub_inwhat = 0;
3280af22
NIS
435 PL_linestr = line;
436 if (SvREADONLY(PL_linestr))
437 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
438 s = SvPV(PL_linestr, len);
8990e307 439 if (len && s[len-1] != ';') {
3280af22
NIS
440 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
441 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
442 sv_catpvn(PL_linestr, "\n;", 2);
8990e307 443 }
3280af22
NIS
444 SvTEMP_off(PL_linestr);
445 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
446 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
207e3d1a 447 PL_last_lop = PL_last_uni = Nullch;
3280af22 448 PL_rsfp = 0;
79072805 449}
a687059c 450
ffb4593c
NT
451/*
452 * Perl_lex_end
9cbb5ea2
GS
453 * Finalizer for lexing operations. Must be called when the parser is
454 * done with the lexer.
ffb4593c
NT
455 */
456
463ee0b2 457void
864dbfa3 458Perl_lex_end(pTHX)
463ee0b2 459{
3280af22 460 PL_doextract = FALSE;
463ee0b2
LW
461}
462
ffb4593c
NT
463/*
464 * S_incline
465 * This subroutine has nothing to do with tilting, whether at windmills
466 * or pinball tables. Its name is short for "increment line". It
57843af0 467 * increments the current line number in CopLINE(PL_curcop) and checks
ffb4593c 468 * to see whether the line starts with a comment of the form
9cbb5ea2
GS
469 * # line 500 "foo.pm"
470 * If so, it sets the current line number and file to the values in the comment.
ffb4593c
NT
471 */
472
76e3520e 473STATIC void
cea2e8a9 474S_incline(pTHX_ char *s)
463ee0b2
LW
475{
476 char *t;
477 char *n;
73659bf1 478 char *e;
463ee0b2 479 char ch;
463ee0b2 480
57843af0 481 CopLINE_inc(PL_curcop);
463ee0b2
LW
482 if (*s++ != '#')
483 return;
bf4acbe4 484 while (SPACE_OR_TAB(*s)) s++;
73659bf1
GS
485 if (strnEQ(s, "line", 4))
486 s += 4;
487 else
488 return;
084592ab 489 if (SPACE_OR_TAB(*s))
73659bf1 490 s++;
4e553d73 491 else
73659bf1 492 return;
bf4acbe4 493 while (SPACE_OR_TAB(*s)) s++;
463ee0b2
LW
494 if (!isDIGIT(*s))
495 return;
496 n = s;
497 while (isDIGIT(*s))
498 s++;
bf4acbe4 499 while (SPACE_OR_TAB(*s))
463ee0b2 500 s++;
73659bf1 501 if (*s == '"' && (t = strchr(s+1, '"'))) {
463ee0b2 502 s++;
73659bf1
GS
503 e = t + 1;
504 }
463ee0b2 505 else {
463ee0b2 506 for (t = s; !isSPACE(*t); t++) ;
73659bf1 507 e = t;
463ee0b2 508 }
bf4acbe4 509 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
73659bf1
GS
510 e++;
511 if (*e != '\n' && *e != '\0')
512 return; /* false alarm */
513
463ee0b2
LW
514 ch = *t;
515 *t = '\0';
f4dd75d9
GS
516 if (t - s > 0) {
517#ifdef USE_ITHREADS
518 Safefree(CopFILE(PL_curcop));
519#else
520 SvREFCNT_dec(CopFILEGV(PL_curcop));
521#endif
57843af0 522 CopFILE_set(PL_curcop, s);
f4dd75d9 523 }
463ee0b2 524 *t = ch;
57843af0 525 CopLINE_set(PL_curcop, atoi(n)-1);
463ee0b2
LW
526}
527
ffb4593c
NT
528/*
529 * S_skipspace
530 * Called to gobble the appropriate amount and type of whitespace.
531 * Skips comments as well.
532 */
533
76e3520e 534STATIC char *
cea2e8a9 535S_skipspace(pTHX_ register char *s)
a687059c 536{
3280af22 537 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
bf4acbe4 538 while (s < PL_bufend && SPACE_OR_TAB(*s))
463ee0b2
LW
539 s++;
540 return s;
541 }
542 for (;;) {
fd049845 543 STRLEN prevlen;
09bef843 544 SSize_t oldprevlen, oldoldprevlen;
9c5ffd7c 545 SSize_t oldloplen = 0, oldunilen = 0;
60e6418e
GS
546 while (s < PL_bufend && isSPACE(*s)) {
547 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
548 incline(s);
549 }
ffb4593c
NT
550
551 /* comment */
3280af22
NIS
552 if (s < PL_bufend && *s == '#') {
553 while (s < PL_bufend && *s != '\n')
463ee0b2 554 s++;
60e6418e 555 if (s < PL_bufend) {
463ee0b2 556 s++;
60e6418e
GS
557 if (PL_in_eval && !PL_rsfp) {
558 incline(s);
559 continue;
560 }
561 }
463ee0b2 562 }
ffb4593c
NT
563
564 /* only continue to recharge the buffer if we're at the end
565 * of the buffer, we're not reading from a source filter, and
566 * we're in normal lexing mode
567 */
09bef843
SB
568 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
569 PL_lex_state == LEX_FORMLINE)
463ee0b2 570 return s;
ffb4593c
NT
571
572 /* try to recharge the buffer */
9cbb5ea2
GS
573 if ((s = filter_gets(PL_linestr, PL_rsfp,
574 (prevlen = SvCUR(PL_linestr)))) == Nullch)
575 {
576 /* end of file. Add on the -p or -n magic */
3280af22
NIS
577 if (PL_minus_n || PL_minus_p) {
578 sv_setpv(PL_linestr,PL_minus_p ?
08e9d68e
DD
579 ";}continue{print or die qq(-p destination: $!\\n)" :
580 "");
3280af22
NIS
581 sv_catpv(PL_linestr,";}");
582 PL_minus_n = PL_minus_p = 0;
a0d0e21e
LW
583 }
584 else
3280af22 585 sv_setpv(PL_linestr,";");
ffb4593c
NT
586
587 /* reset variables for next time we lex */
9cbb5ea2
GS
588 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
589 = SvPVX(PL_linestr);
3280af22 590 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
207e3d1a 591 PL_last_lop = PL_last_uni = Nullch;
ffb4593c
NT
592
593 /* Close the filehandle. Could be from -P preprocessor,
594 * STDIN, or a regular file. If we were reading code from
595 * STDIN (because the commandline held no -e or filename)
596 * then we don't close it, we reset it so the code can
597 * read from STDIN too.
598 */
599
3280af22
NIS
600 if (PL_preprocess && !PL_in_eval)
601 (void)PerlProc_pclose(PL_rsfp);
602 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
603 PerlIO_clearerr(PL_rsfp);
8990e307 604 else
3280af22
NIS
605 (void)PerlIO_close(PL_rsfp);
606 PL_rsfp = Nullfp;
463ee0b2
LW
607 return s;
608 }
ffb4593c
NT
609
610 /* not at end of file, so we only read another line */
09bef843
SB
611 /* make corresponding updates to old pointers, for yyerror() */
612 oldprevlen = PL_oldbufptr - PL_bufend;
613 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
614 if (PL_last_uni)
615 oldunilen = PL_last_uni - PL_bufend;
616 if (PL_last_lop)
617 oldloplen = PL_last_lop - PL_bufend;
3280af22
NIS
618 PL_linestart = PL_bufptr = s + prevlen;
619 PL_bufend = s + SvCUR(PL_linestr);
620 s = PL_bufptr;
09bef843
SB
621 PL_oldbufptr = s + oldprevlen;
622 PL_oldoldbufptr = s + oldoldprevlen;
623 if (PL_last_uni)
624 PL_last_uni = s + oldunilen;
625 if (PL_last_lop)
626 PL_last_lop = s + oldloplen;
a0d0e21e 627 incline(s);
ffb4593c
NT
628
629 /* debugger active and we're not compiling the debugger code,
630 * so store the line into the debugger's array of lines
631 */
3280af22 632 if (PERLDB_LINE && PL_curstash != PL_debstash) {
8990e307
LW
633 SV *sv = NEWSV(85,0);
634
635 sv_upgrade(sv, SVt_PVMG);
3280af22 636 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
0ac0412a
MJD
637 (void)SvIOK_on(sv);
638 SvIVX(sv) = 0;
57843af0 639 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
8990e307 640 }
463ee0b2 641 }
a687059c 642}
378cc40b 643
ffb4593c
NT
644/*
645 * S_check_uni
646 * Check the unary operators to ensure there's no ambiguity in how they're
647 * used. An ambiguous piece of code would be:
648 * rand + 5
649 * This doesn't mean rand() + 5. Because rand() is a unary operator,
650 * the +5 is its argument.
651 */
652
76e3520e 653STATIC void
cea2e8a9 654S_check_uni(pTHX)
ba106d47 655{
2f3197b3 656 char *s;
a0d0e21e 657 char *t;
2f3197b3 658
3280af22 659 if (PL_oldoldbufptr != PL_last_uni)
2f3197b3 660 return;
3280af22
NIS
661 while (isSPACE(*PL_last_uni))
662 PL_last_uni++;
7e2040f0 663 for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ;
3280af22 664 if ((t = strchr(s, '(')) && t < PL_bufptr)
a0d0e21e 665 return;
0453d815 666 if (ckWARN_d(WARN_AMBIGUOUS)){
f248d071 667 char ch = *s;
0453d815 668 *s = '\0';
4e553d73
NIS
669 Perl_warner(aTHX_ WARN_AMBIGUOUS,
670 "Warning: Use of \"%s\" without parens is ambiguous",
0453d815
PM
671 PL_last_uni);
672 *s = ch;
673 }
2f3197b3
LW
674}
675
ffb4593c
NT
676/* workaround to replace the UNI() macro with a function. Only the
677 * hints/uts.sh file mentions this. Other comments elsewhere in the
678 * source indicate Microport Unix might need it too.
679 */
680
ffed7fef
LW
681#ifdef CRIPPLED_CC
682
683#undef UNI
ffed7fef 684#define UNI(f) return uni(f,s)
ffed7fef 685
76e3520e 686STATIC int
cea2e8a9 687S_uni(pTHX_ I32 f, char *s)
ffed7fef
LW
688{
689 yylval.ival = f;
3280af22
NIS
690 PL_expect = XTERM;
691 PL_bufptr = s;
8f872242
NIS
692 PL_last_uni = PL_oldbufptr;
693 PL_last_lop_op = f;
ffed7fef
LW
694 if (*s == '(')
695 return FUNC1;
696 s = skipspace(s);
697 if (*s == '(')
698 return FUNC1;
699 else
700 return UNIOP;
701}
702
a0d0e21e
LW
703#endif /* CRIPPLED_CC */
704
ffb4593c
NT
705/*
706 * LOP : macro to build a list operator. Its behaviour has been replaced
707 * with a subroutine, S_lop() for which LOP is just another name.
708 */
709
a0d0e21e
LW
710#define LOP(f,x) return lop(f,x,s)
711
ffb4593c
NT
712/*
713 * S_lop
714 * Build a list operator (or something that might be one). The rules:
715 * - if we have a next token, then it's a list operator [why?]
716 * - if the next thing is an opening paren, then it's a function
717 * - else it's a list operator
718 */
719
76e3520e 720STATIC I32
a0be28da 721S_lop(pTHX_ I32 f, int x, char *s)
ffed7fef 722{
79072805 723 yylval.ival = f;
35c8bce7 724 CLINE;
075953c3 725 REPORT("lop", f)
3280af22
NIS
726 PL_expect = x;
727 PL_bufptr = s;
728 PL_last_lop = PL_oldbufptr;
729 PL_last_lop_op = f;
730 if (PL_nexttoke)
a0d0e21e 731 return LSTOP;
79072805
LW
732 if (*s == '(')
733 return FUNC;
734 s = skipspace(s);
735 if (*s == '(')
736 return FUNC;
737 else
738 return LSTOP;
739}
740
ffb4593c
NT
741/*
742 * S_force_next
9cbb5ea2 743 * When the lexer realizes it knows the next token (for instance,
ffb4593c 744 * it is reordering tokens for the parser) then it can call S_force_next
9cbb5ea2
GS
745 * to know what token to return the next time the lexer is called. Caller
746 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
747 * handles the token correctly.
ffb4593c
NT
748 */
749
4e553d73 750STATIC void
cea2e8a9 751S_force_next(pTHX_ I32 type)
79072805 752{
3280af22
NIS
753 PL_nexttype[PL_nexttoke] = type;
754 PL_nexttoke++;
755 if (PL_lex_state != LEX_KNOWNEXT) {
756 PL_lex_defer = PL_lex_state;
757 PL_lex_expect = PL_expect;
758 PL_lex_state = LEX_KNOWNEXT;
79072805
LW
759 }
760}
761
ffb4593c
NT
762/*
763 * S_force_word
764 * When the lexer knows the next thing is a word (for instance, it has
765 * just seen -> and it knows that the next char is a word char, then
766 * it calls S_force_word to stick the next word into the PL_next lookahead.
767 *
768 * Arguments:
b1b65b59 769 * char *start : buffer position (must be within PL_linestr)
ffb4593c
NT
770 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
771 * int check_keyword : if true, Perl checks to make sure the word isn't
772 * a keyword (do this if the word is a label, e.g. goto FOO)
773 * int allow_pack : if true, : characters will also be allowed (require,
774 * use, etc. do this)
9cbb5ea2 775 * int allow_initial_tick : used by the "sub" lexer only.
ffb4593c
NT
776 */
777
76e3520e 778STATIC char *
cea2e8a9 779S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
79072805 780{
463ee0b2
LW
781 register char *s;
782 STRLEN len;
4e553d73 783
463ee0b2
LW
784 start = skipspace(start);
785 s = start;
7e2040f0 786 if (isIDFIRST_lazy_if(s,UTF) ||
a0d0e21e 787 (allow_pack && *s == ':') ||
15f0808c 788 (allow_initial_tick && *s == '\'') )
a0d0e21e 789 {
3280af22
NIS
790 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
791 if (check_keyword && keyword(PL_tokenbuf, len))
463ee0b2
LW
792 return start;
793 if (token == METHOD) {
794 s = skipspace(s);
795 if (*s == '(')
3280af22 796 PL_expect = XTERM;
463ee0b2 797 else {
3280af22 798 PL_expect = XOPERATOR;
463ee0b2 799 }
79072805 800 }
3280af22
NIS
801 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0));
802 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
79072805
LW
803 force_next(token);
804 }
805 return s;
806}
807
ffb4593c
NT
808/*
809 * S_force_ident
9cbb5ea2 810 * Called when the lexer wants $foo *foo &foo etc, but the program
ffb4593c
NT
811 * text only contains the "foo" portion. The first argument is a pointer
812 * to the "foo", and the second argument is the type symbol to prefix.
813 * Forces the next token to be a "WORD".
9cbb5ea2 814 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
ffb4593c
NT
815 */
816
76e3520e 817STATIC void
cea2e8a9 818S_force_ident(pTHX_ register char *s, int kind)
79072805
LW
819{
820 if (s && *s) {
11343788 821 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
3280af22 822 PL_nextval[PL_nexttoke].opval = o;
79072805 823 force_next(WORD);
748a9306 824 if (kind) {
11343788 825 o->op_private = OPpCONST_ENTERED;
55497cff
PP
826 /* XXX see note in pp_entereval() for why we forgo typo
827 warnings if the symbol must be introduced in an eval.
828 GSAR 96-10-12 */
3280af22 829 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
a0d0e21e
LW
830 kind == '$' ? SVt_PV :
831 kind == '@' ? SVt_PVAV :
832 kind == '%' ? SVt_PVHV :
833 SVt_PVGV
834 );
748a9306 835 }
79072805
LW
836 }
837}
838
1571675a
GS
839NV
840Perl_str_to_version(pTHX_ SV *sv)
841{
842 NV retval = 0.0;
843 NV nshift = 1.0;
844 STRLEN len;
845 char *start = SvPVx(sv,len);
3aa33fe5 846 bool utf = SvUTF8(sv) ? TRUE : FALSE;
1571675a
GS
847 char *end = start + len;
848 while (start < end) {
ba210ebe 849 STRLEN skip;
1571675a
GS
850 UV n;
851 if (utf)
9041c2e3 852 n = utf8n_to_uvchr((U8*)start, len, &skip, 0);
1571675a
GS
853 else {
854 n = *(U8*)start;
855 skip = 1;
856 }
857 retval += ((NV)n)/nshift;
858 start += skip;
859 nshift *= 1000;
860 }
861 return retval;
862}
863
4e553d73 864/*
ffb4593c
NT
865 * S_force_version
866 * Forces the next token to be a version number.
e759cc13
RGS
867 * If the next token appears to be an invalid version number, (e.g. "v2b"),
868 * and if "guessing" is TRUE, then no new token is created (and the caller
869 * must use an alternative parsing method).
ffb4593c
NT
870 */
871
76e3520e 872STATIC char *
e759cc13 873S_force_version(pTHX_ char *s, int guessing)
89bfa8cd
PP
874{
875 OP *version = Nullop;
44dcb63b 876 char *d;
89bfa8cd
PP
877
878 s = skipspace(s);
879
44dcb63b 880 d = s;
dd629d5b 881 if (*d == 'v')
44dcb63b 882 d++;
44dcb63b 883 if (isDIGIT(*d)) {
e759cc13
RGS
884 while (isDIGIT(*d) || *d == '_' || *d == '.')
885 d++;
9f3d182e 886 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
dd629d5b 887 SV *ver;
b73d6f50 888 s = scan_num(s, &yylval);
89bfa8cd 889 version = yylval.opval;
dd629d5b
GS
890 ver = cSVOPx(version)->op_sv;
891 if (SvPOK(ver) && !SvNIOK(ver)) {
155aba94 892 (void)SvUPGRADE(ver, SVt_PVNV);
1571675a
GS
893 SvNVX(ver) = str_to_version(ver);
894 SvNOK_on(ver); /* hint that it is a version */
44dcb63b 895 }
89bfa8cd 896 }
e759cc13
RGS
897 else if (guessing)
898 return s;
89bfa8cd
PP
899 }
900
901 /* NOTE: The parser sees the package name and the VERSION swapped */
3280af22 902 PL_nextval[PL_nexttoke].opval = version;
4e553d73 903 force_next(WORD);
89bfa8cd 904
e759cc13 905 return s;
89bfa8cd
PP
906}
907
ffb4593c
NT
908/*
909 * S_tokeq
910 * Tokenize a quoted string passed in as an SV. It finds the next
911 * chunk, up to end of string or a backslash. It may make a new
912 * SV containing that chunk (if HINT_NEW_STRING is on). It also
913 * turns \\ into \.
914 */
915
76e3520e 916STATIC SV *
cea2e8a9 917S_tokeq(pTHX_ SV *sv)
79072805
LW
918{
919 register char *s;
920 register char *send;
921 register char *d;
b3ac6de7
IZ
922 STRLEN len = 0;
923 SV *pv = sv;
79072805
LW
924
925 if (!SvLEN(sv))
b3ac6de7 926 goto finish;
79072805 927
a0d0e21e 928 s = SvPV_force(sv, len);
21a311ee 929 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
b3ac6de7 930 goto finish;
463ee0b2 931 send = s + len;
79072805
LW
932 while (s < send && *s != '\\')
933 s++;
934 if (s == send)
b3ac6de7 935 goto finish;
79072805 936 d = s;
be4731d2 937 if ( PL_hints & HINT_NEW_STRING ) {
79cb57f6 938 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
be4731d2
NIS
939 if (SvUTF8(sv))
940 SvUTF8_on(pv);
941 }
79072805
LW
942 while (s < send) {
943 if (*s == '\\') {
a0d0e21e 944 if (s + 1 < send && (s[1] == '\\'))
79072805
LW
945 s++; /* all that, just for this */
946 }
947 *d++ = *s++;
948 }
949 *d = '\0';
463ee0b2 950 SvCUR_set(sv, d - SvPVX(sv));
b3ac6de7 951 finish:
3280af22 952 if ( PL_hints & HINT_NEW_STRING )
b3ac6de7 953 return new_constant(NULL, 0, "q", sv, pv, "q");
79072805
LW
954 return sv;
955}
956
ffb4593c
NT
957/*
958 * Now come three functions related to double-quote context,
959 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
960 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
961 * interact with PL_lex_state, and create fake ( ... ) argument lists
962 * to handle functions and concatenation.
963 * They assume that whoever calls them will be setting up a fake
964 * join call, because each subthing puts a ',' after it. This lets
965 * "lower \luPpEr"
966 * become
967 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
968 *
969 * (I'm not sure whether the spurious commas at the end of lcfirst's
970 * arguments and join's arguments are created or not).
971 */
972
973/*
974 * S_sublex_start
975 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
976 *
977 * Pattern matching will set PL_lex_op to the pattern-matching op to
978 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
979 *
980 * OP_CONST and OP_READLINE are easy--just make the new op and return.
981 *
982 * Everything else becomes a FUNC.
983 *
984 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
985 * had an OP_CONST or OP_READLINE). This just sets us up for a
986 * call to S_sublex_push().
987 */
988
76e3520e 989STATIC I32
cea2e8a9 990S_sublex_start(pTHX)
79072805
LW
991{
992 register I32 op_type = yylval.ival;
79072805
LW
993
994 if (op_type == OP_NULL) {
3280af22
NIS
995 yylval.opval = PL_lex_op;
996 PL_lex_op = Nullop;
79072805
LW
997 return THING;
998 }
999 if (op_type == OP_CONST || op_type == OP_READLINE) {
3280af22 1000 SV *sv = tokeq(PL_lex_stuff);
b3ac6de7
IZ
1001
1002 if (SvTYPE(sv) == SVt_PVIV) {
1003 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
1004 STRLEN len;
1005 char *p;
1006 SV *nsv;
1007
1008 p = SvPV(sv, len);
79cb57f6 1009 nsv = newSVpvn(p, len);
01ec43d0
GS
1010 if (SvUTF8(sv))
1011 SvUTF8_on(nsv);
b3ac6de7
IZ
1012 SvREFCNT_dec(sv);
1013 sv = nsv;
4e553d73 1014 }
b3ac6de7 1015 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
3280af22 1016 PL_lex_stuff = Nullsv;
79072805
LW
1017 return THING;
1018 }
1019
3280af22
NIS
1020 PL_sublex_info.super_state = PL_lex_state;
1021 PL_sublex_info.sub_inwhat = op_type;
1022 PL_sublex_info.sub_op = PL_lex_op;
1023 PL_lex_state = LEX_INTERPPUSH;
55497cff 1024
3280af22
NIS
1025 PL_expect = XTERM;
1026 if (PL_lex_op) {
1027 yylval.opval = PL_lex_op;
1028 PL_lex_op = Nullop;
55497cff
PP
1029 return PMFUNC;
1030 }
1031 else
1032 return FUNC;
1033}
1034
ffb4593c
NT
1035/*
1036 * S_sublex_push
1037 * Create a new scope to save the lexing state. The scope will be
1038 * ended in S_sublex_done. Returns a '(', starting the function arguments
1039 * to the uc, lc, etc. found before.
1040 * Sets PL_lex_state to LEX_INTERPCONCAT.
1041 */
1042
76e3520e 1043STATIC I32
cea2e8a9 1044S_sublex_push(pTHX)
55497cff 1045{
f46d017c 1046 ENTER;
55497cff 1047
3280af22
NIS
1048 PL_lex_state = PL_sublex_info.super_state;
1049 SAVEI32(PL_lex_dojoin);
1050 SAVEI32(PL_lex_brackets);
3280af22
NIS
1051 SAVEI32(PL_lex_casemods);
1052 SAVEI32(PL_lex_starts);
1053 SAVEI32(PL_lex_state);
7766f137 1054 SAVEVPTR(PL_lex_inpat);
3280af22 1055 SAVEI32(PL_lex_inwhat);
57843af0 1056 SAVECOPLINE(PL_curcop);
3280af22 1057 SAVEPPTR(PL_bufptr);
8452ff4b 1058 SAVEPPTR(PL_bufend);
3280af22
NIS
1059 SAVEPPTR(PL_oldbufptr);
1060 SAVEPPTR(PL_oldoldbufptr);
207e3d1a
JH
1061 SAVEPPTR(PL_last_lop);
1062 SAVEPPTR(PL_last_uni);
3280af22
NIS
1063 SAVEPPTR(PL_linestart);
1064 SAVESPTR(PL_linestr);
1065 SAVEPPTR(PL_lex_brackstack);
1066 SAVEPPTR(PL_lex_casestack);
1067
1068 PL_linestr = PL_lex_stuff;
1069 PL_lex_stuff = Nullsv;
1070
9cbb5ea2
GS
1071 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1072 = SvPVX(PL_linestr);
3280af22 1073 PL_bufend += SvCUR(PL_linestr);
207e3d1a 1074 PL_last_lop = PL_last_uni = Nullch;
3280af22
NIS
1075 SAVEFREESV(PL_linestr);
1076
1077 PL_lex_dojoin = FALSE;
1078 PL_lex_brackets = 0;
3280af22
NIS
1079 New(899, PL_lex_brackstack, 120, char);
1080 New(899, PL_lex_casestack, 12, char);
1081 SAVEFREEPV(PL_lex_brackstack);
1082 SAVEFREEPV(PL_lex_casestack);
1083 PL_lex_casemods = 0;
1084 *PL_lex_casestack = '\0';
1085 PL_lex_starts = 0;
1086 PL_lex_state = LEX_INTERPCONCAT;
57843af0 1087 CopLINE_set(PL_curcop, PL_multi_start);
3280af22
NIS
1088
1089 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1090 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1091 PL_lex_inpat = PL_sublex_info.sub_op;
79072805 1092 else
3280af22 1093 PL_lex_inpat = Nullop;
79072805 1094
55497cff 1095 return '(';
79072805
LW
1096}
1097
ffb4593c
NT
1098/*
1099 * S_sublex_done
1100 * Restores lexer state after a S_sublex_push.
1101 */
1102
76e3520e 1103STATIC I32
cea2e8a9 1104S_sublex_done(pTHX)
79072805 1105{
3280af22 1106 if (!PL_lex_starts++) {
9aa983d2
JH
1107 SV *sv = newSVpvn("",0);
1108 if (SvUTF8(PL_linestr))
1109 SvUTF8_on(sv);
3280af22 1110 PL_expect = XOPERATOR;
9aa983d2 1111 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
79072805
LW
1112 return THING;
1113 }
1114
3280af22
NIS
1115 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1116 PL_lex_state = LEX_INTERPCASEMOD;
cea2e8a9 1117 return yylex();
79072805
LW
1118 }
1119
ffb4593c 1120 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
3280af22
NIS
1121 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1122 PL_linestr = PL_lex_repl;
1123 PL_lex_inpat = 0;
1124 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1125 PL_bufend += SvCUR(PL_linestr);
207e3d1a 1126 PL_last_lop = PL_last_uni = Nullch;
3280af22
NIS
1127 SAVEFREESV(PL_linestr);
1128 PL_lex_dojoin = FALSE;
1129 PL_lex_brackets = 0;
3280af22
NIS
1130 PL_lex_casemods = 0;
1131 *PL_lex_casestack = '\0';
1132 PL_lex_starts = 0;
25da4f38 1133 if (SvEVALED(PL_lex_repl)) {
3280af22
NIS
1134 PL_lex_state = LEX_INTERPNORMAL;
1135 PL_lex_starts++;
e9fa98b2
HS
1136 /* we don't clear PL_lex_repl here, so that we can check later
1137 whether this is an evalled subst; that means we rely on the
1138 logic to ensure sublex_done() is called again only via the
1139 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
79072805 1140 }
e9fa98b2 1141 else {
3280af22 1142 PL_lex_state = LEX_INTERPCONCAT;
e9fa98b2
HS
1143 PL_lex_repl = Nullsv;
1144 }
79072805 1145 return ',';
ffed7fef
LW
1146 }
1147 else {
f46d017c 1148 LEAVE;
3280af22
NIS
1149 PL_bufend = SvPVX(PL_linestr);
1150 PL_bufend += SvCUR(PL_linestr);
1151 PL_expect = XOPERATOR;
09bef843 1152 PL_sublex_info.sub_inwhat = 0;
79072805 1153 return ')';
ffed7fef
LW
1154 }
1155}
1156
02aa26ce
NT
1157/*
1158 scan_const
1159
1160 Extracts a pattern, double-quoted string, or transliteration. This
1161 is terrifying code.
1162
3280af22
NIS
1163 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1164 processing a pattern (PL_lex_inpat is true), a transliteration
02aa26ce
NT
1165 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1166
9b599b2a
GS
1167 Returns a pointer to the character scanned up to. Iff this is
1168 advanced from the start pointer supplied (ie if anything was
1169 successfully parsed), will leave an OP for the substring scanned
1170 in yylval. Caller must intuit reason for not parsing further
1171 by looking at the next characters herself.
1172
02aa26ce
NT
1173 In patterns:
1174 backslashes:
1175 double-quoted style: \r and \n
1176 regexp special ones: \D \s
1177 constants: \x3
1178 backrefs: \1 (deprecated in substitution replacements)
1179 case and quoting: \U \Q \E
1180 stops on @ and $, but not for $ as tail anchor
1181
1182 In transliterations:
1183 characters are VERY literal, except for - not at the start or end
1184 of the string, which indicates a range. scan_const expands the
1185 range to the full set of intermediate characters.
1186
1187 In double-quoted strings:
1188 backslashes:
1189 double-quoted style: \r and \n
1190 constants: \x3
1191 backrefs: \1 (deprecated)
1192 case and quoting: \U \Q \E
1193 stops on @ and $
1194
1195 scan_const does *not* construct ops to handle interpolated strings.
1196 It stops processing as soon as it finds an embedded $ or @ variable
1197 and leaves it to the caller to work out what's going on.
1198
1199 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @:foo.
1200
1201 $ in pattern could be $foo or could be tail anchor. Assumption:
1202 it's a tail anchor if $ is the last thing in the string, or if it's
1203 followed by one of ")| \n\t"
1204
1205 \1 (backreferences) are turned into $1
1206
1207 The structure of the code is
1208 while (there's a character to process) {
1209 handle transliteration ranges
1210 skip regexp comments
1211 skip # initiated comments in //x patterns
1212 check for embedded @foo
1213 check for embedded scalars
1214 if (backslash) {
1215 leave intact backslashes from leave (below)
1216 deprecate \1 in strings and sub replacements
1217 handle string-changing backslashes \l \U \Q \E, etc.
1218 switch (what was escaped) {
1219 handle - in a transliteration (becomes a literal -)
1220 handle \132 octal characters
1221 handle 0x15 hex characters
1222 handle \cV (control V)
1223 handle printf backslashes (\f, \r, \n, etc)
1224 } (end switch)
1225 } (end if backslash)
1226 } (end while character to read)
4e553d73 1227
02aa26ce
NT
1228*/
1229
76e3520e 1230STATIC char *
cea2e8a9 1231S_scan_const(pTHX_ char *start)
79072805 1232{
3280af22 1233 register char *send = PL_bufend; /* end of the constant */
02aa26ce
NT
1234 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1235 register char *s = start; /* start of the constant */
1236 register char *d = SvPVX(sv); /* destination for copies */
1237 bool dorange = FALSE; /* are we in a translit range? */
c2e66d9e 1238 bool didrange = FALSE; /* did we just finish a range? */
2b9d42f0
NIS
1239 I32 has_utf8 = FALSE; /* Output constant is UTF8 */
1240 I32 this_utf8 = UTF; /* The source string is assumed to be UTF8 */
012bcf8d
GS
1241 UV uv;
1242
dff6d3cd 1243 const char *leaveit = /* set of acceptably-backslashed characters */
3280af22 1244 PL_lex_inpat
4a2d328f 1245 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
9b599b2a 1246 : "";
79072805 1247
2b9d42f0
NIS
1248 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1249 /* If we are doing a trans and we know we want UTF8 set expectation */
1250 has_utf8 = PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF);
1251 this_utf8 = PL_sublex_info.sub_op->op_private & (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1252 }
1253
1254
79072805 1255 while (s < send || dorange) {
02aa26ce 1256 /* get transliterations out of the way (they're most literal) */
3280af22 1257 if (PL_lex_inwhat == OP_TRANS) {
02aa26ce 1258 /* expand a range A-Z to the full set of characters. AIE! */
79072805 1259 if (dorange) {
1ba5c669
JH
1260 I32 i; /* current expanded character */
1261 I32 min; /* first character in range */
1262 I32 max; /* last character in range */
02aa26ce 1263
2b9d42f0 1264 if (has_utf8) {
8973db79
JH
1265 char *c = (char*)utf8_hop((U8*)d, -1);
1266 char *e = d++;
1267 while (e-- > c)
1268 *(e + 1) = *e;
25716404 1269 *c = (char)UTF_TO_NATIVE(0xff);
8973db79
JH
1270 /* mark the range as done, and continue */
1271 dorange = FALSE;
1272 didrange = TRUE;
1273 continue;
1274 }
2b9d42f0 1275
02aa26ce 1276 i = d - SvPVX(sv); /* remember current offset */
9cbb5ea2
GS
1277 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1278 d = SvPVX(sv) + i; /* refresh d after realloc */
02aa26ce
NT
1279 d -= 2; /* eat the first char and the - */
1280
8ada0baa
JH
1281 min = (U8)*d; /* first char in range */
1282 max = (U8)d[1]; /* last char in range */
1283
c2e66d9e 1284 if (min > max) {
01ec43d0 1285 Perl_croak(aTHX_
1ba5c669
JH
1286 "Invalid [] range \"%c-%c\" in transliteration operator",
1287 (char)min, (char)max);
c2e66d9e
GS
1288 }
1289
c7f1f016 1290#ifdef EBCDIC
8ada0baa
JH
1291 if ((isLOWER(min) && isLOWER(max)) ||
1292 (isUPPER(min) && isUPPER(max))) {
1293 if (isLOWER(min)) {
1294 for (i = min; i <= max; i++)
1295 if (isLOWER(i))
db42d148 1296 *d++ = NATIVE_TO_NEED(has_utf8,i);
8ada0baa
JH
1297 } else {
1298 for (i = min; i <= max; i++)
1299 if (isUPPER(i))
db42d148 1300 *d++ = NATIVE_TO_NEED(has_utf8,i);
8ada0baa
JH
1301 }
1302 }
1303 else
1304#endif
1305 for (i = min; i <= max; i++)
1306 *d++ = i;
02aa26ce
NT
1307
1308 /* mark the range as done, and continue */
79072805 1309 dorange = FALSE;
01ec43d0 1310 didrange = TRUE;
79072805 1311 continue;
4e553d73 1312 }
02aa26ce
NT
1313
1314 /* range begins (ignore - as first or last char) */
79072805 1315 else if (*s == '-' && s+1 < send && s != start) {
4e553d73 1316 if (didrange) {
1fafa243 1317 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
01ec43d0 1318 }
2b9d42f0 1319 if (has_utf8) {
25716404 1320 *d++ = (char)UTF_TO_NATIVE(0xff); /* use illegal utf8 byte--see pmtrans */
a0ed51b3
LW
1321 s++;
1322 continue;
1323 }
79072805
LW
1324 dorange = TRUE;
1325 s++;
01ec43d0
GS
1326 }
1327 else {
1328 didrange = FALSE;
1329 }
79072805 1330 }
02aa26ce
NT
1331
1332 /* if we get here, we're not doing a transliteration */
1333
0f5d15d6
IZ
1334 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1335 except for the last char, which will be done separately. */
3280af22 1336 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
cc6b7395
IZ
1337 if (s[2] == '#') {
1338 while (s < send && *s != ')')
db42d148 1339 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
155aba94
GS
1340 }
1341 else if (s[2] == '{' /* This should match regcomp.c */
1342 || ((s[2] == 'p' || s[2] == '?') && s[3] == '{'))
1343 {
cc6b7395 1344 I32 count = 1;
0f5d15d6 1345 char *regparse = s + (s[2] == '{' ? 3 : 4);
cc6b7395
IZ
1346 char c;
1347
d9f97599
GS
1348 while (count && (c = *regparse)) {
1349 if (c == '\\' && regparse[1])
1350 regparse++;
4e553d73 1351 else if (c == '{')
cc6b7395 1352 count++;
4e553d73 1353 else if (c == '}')
cc6b7395 1354 count--;
d9f97599 1355 regparse++;
cc6b7395 1356 }
5bdf89e7
IZ
1357 if (*regparse != ')') {
1358 regparse--; /* Leave one char for continuation. */
cc6b7395 1359 yyerror("Sequence (?{...}) not terminated or not {}-balanced");
5bdf89e7 1360 }
0f5d15d6 1361 while (s < regparse)
db42d148 1362 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
cc6b7395 1363 }
748a9306 1364 }
02aa26ce
NT
1365
1366 /* likewise skip #-initiated comments in //x patterns */
3280af22
NIS
1367 else if (*s == '#' && PL_lex_inpat &&
1368 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
748a9306 1369 while (s+1 < send && *s != '\n')
db42d148 1370 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
748a9306 1371 }
02aa26ce 1372
5d1d4326
JH
1373 /* check for embedded arrays
1374 (@foo, @:foo, @'foo, @{foo}, @$foo, @+, @-)
1375 */
7e2040f0 1376 else if (*s == '@' && s[1]
5d1d4326 1377 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$+-", s[1])))
79072805 1378 break;
02aa26ce
NT
1379
1380 /* check for embedded scalars. only stop if we're sure it's a
1381 variable.
1382 */
79072805 1383 else if (*s == '$') {
3280af22 1384 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
79072805 1385 break;
6002328a 1386 if (s + 1 < send && !strchr("()| \r\n\t", s[1]))
79072805
LW
1387 break; /* in regexp, $ might be tail anchor */
1388 }
02aa26ce 1389
2b9d42f0
NIS
1390 /* End of else if chain - OP_TRANS rejoin rest */
1391
02aa26ce 1392 /* backslashes */
79072805
LW
1393 if (*s == '\\' && s+1 < send) {
1394 s++;
02aa26ce
NT
1395
1396 /* some backslashes we leave behind */
c9f97d15 1397 if (*leaveit && *s && strchr(leaveit, *s)) {
db42d148
NIS
1398 *d++ = NATIVE_TO_NEED(has_utf8,'\\');
1399 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
79072805
LW
1400 continue;
1401 }
02aa26ce
NT
1402
1403 /* deprecate \1 in strings and substitution replacements */
3280af22 1404 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
a0d0e21e 1405 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
79072805 1406 {
599cee73 1407 if (ckWARN(WARN_SYNTAX))
cea2e8a9 1408 Perl_warner(aTHX_ WARN_SYNTAX, "\\%c better written as $%c", *s, *s);
79072805
LW
1409 *--s = '$';
1410 break;
1411 }
02aa26ce
NT
1412
1413 /* string-change backslash escapes */
3280af22 1414 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
79072805
LW
1415 --s;
1416 break;
1417 }
02aa26ce
NT
1418
1419 /* if we get here, it's either a quoted -, or a digit */
79072805 1420 switch (*s) {
02aa26ce
NT
1421
1422 /* quoted - in transliterations */
79072805 1423 case '-':
3280af22 1424 if (PL_lex_inwhat == OP_TRANS) {
79072805
LW
1425 *d++ = *s++;
1426 continue;
1427 }
1428 /* FALL THROUGH */
1429 default:
11b8faa4 1430 {
7e84c16c 1431 if (ckWARN(WARN_MISC) && isALNUM(*s))
4e553d73 1432 Perl_warner(aTHX_ WARN_MISC,
11b8faa4
JH
1433 "Unrecognized escape \\%c passed through",
1434 *s);
1435 /* default action is to copy the quoted character */
f9a63242 1436 goto default_action;
11b8faa4 1437 }
02aa26ce
NT
1438
1439 /* \132 indicates an octal constant */
79072805
LW
1440 case '0': case '1': case '2': case '3':
1441 case '4': case '5': case '6': case '7':
ba210ebe 1442 {
53305cf1
NC
1443 I32 flags = 0;
1444 STRLEN len = 3;
1445 uv = grok_oct(s, &len, &flags, NULL);
ba210ebe
JH
1446 s += len;
1447 }
012bcf8d 1448 goto NUM_ESCAPE_INSERT;
02aa26ce
NT
1449
1450 /* \x24 indicates a hex constant */
79072805 1451 case 'x':
a0ed51b3
LW
1452 ++s;
1453 if (*s == '{') {
1454 char* e = strchr(s, '}');
a4c04bdc
NC
1455 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
1456 PERL_SCAN_DISALLOW_PREFIX;
53305cf1 1457 STRLEN len;
355860ce 1458
53305cf1 1459 ++s;
adaeee49 1460 if (!e) {
a0ed51b3 1461 yyerror("Missing right brace on \\x{}");
355860ce 1462 continue;
ba210ebe 1463 }
53305cf1
NC
1464 len = e - s;
1465 uv = grok_hex(s, &len, &flags, NULL);
ba210ebe 1466 s = e + 1;
a0ed51b3
LW
1467 }
1468 else {
ba210ebe 1469 {
53305cf1 1470 STRLEN len = 2;
a4c04bdc 1471 I32 flags = PERL_SCAN_DISALLOW_PREFIX;
53305cf1 1472 uv = grok_hex(s, &len, &flags, NULL);
ba210ebe
JH
1473 s += len;
1474 }
012bcf8d
GS
1475 }
1476
1477 NUM_ESCAPE_INSERT:
1478 /* Insert oct or hex escaped character.
301d3d20 1479 * There will always enough room in sv since such
db42d148 1480 * escapes will be longer than any UTF-8 sequence
301d3d20 1481 * they can end up as. */
ba7cea30 1482
c7f1f016
NIS
1483 /* We need to map to chars to ASCII before doing the tests
1484 to cover EBCDIC
1485 */
c4d5f83a 1486 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(uv))) {
9aa983d2 1487 if (!has_utf8 && uv > 255) {
301d3d20
JH
1488 /* Might need to recode whatever we have
1489 * accumulated so far if it contains any
1490 * hibit chars.
1491 *
1492 * (Can't we keep track of that and avoid
1493 * this rescan? --jhi)
012bcf8d 1494 */
c7f1f016 1495 int hicount = 0;
63cd0674
NIS
1496 U8 *c;
1497 for (c = (U8 *) SvPVX(sv); c < (U8 *)d; c++) {
c4d5f83a 1498 if (!NATIVE_IS_INVARIANT(*c)) {
012bcf8d 1499 hicount++;
db42d148 1500 }
012bcf8d 1501 }
63cd0674 1502 if (hicount) {
db42d148
NIS
1503 STRLEN offset = d - SvPVX(sv);
1504 U8 *src, *dst;
1505 d = SvGROW(sv, SvLEN(sv) + hicount + 1) + offset;
1506 src = (U8 *)d - 1;
1507 dst = src+hicount;
1508 d += hicount;
1509 while (src >= (U8 *)SvPVX(sv)) {
c4d5f83a 1510 if (!NATIVE_IS_INVARIANT(*src)) {
63cd0674 1511 U8 ch = NATIVE_TO_ASCII(*src);
db42d148
NIS
1512 *dst-- = UTF8_EIGHT_BIT_LO(ch);
1513 *dst-- = UTF8_EIGHT_BIT_HI(ch);
012bcf8d
GS
1514 }
1515 else {
63cd0674 1516 *dst-- = *src;
012bcf8d 1517 }
c7f1f016 1518 src--;
012bcf8d
GS
1519 }
1520 }
1521 }
1522
9aa983d2 1523 if (has_utf8 || uv > 255) {
9041c2e3 1524 d = (char*)uvchr_to_utf8((U8*)d, uv);
4e553d73 1525 has_utf8 = TRUE;
f9a63242
JH
1526 if (PL_lex_inwhat == OP_TRANS &&
1527 PL_sublex_info.sub_op) {
1528 PL_sublex_info.sub_op->op_private |=
1529 (PL_lex_repl ? OPpTRANS_FROM_UTF
1530 : OPpTRANS_TO_UTF);
f9a63242 1531 }
012bcf8d 1532 }
a0ed51b3 1533 else {
012bcf8d 1534 *d++ = (char)uv;
a0ed51b3 1535 }
012bcf8d
GS
1536 }
1537 else {
c4d5f83a 1538 *d++ = (char) uv;
a0ed51b3 1539 }
79072805 1540 continue;
02aa26ce 1541
b239daa5 1542 /* \N{LATIN SMALL LETTER A} is a named character */
4a2d328f 1543 case 'N':
55eda711 1544 ++s;
423cee85
JH
1545 if (*s == '{') {
1546 char* e = strchr(s, '}');
155aba94 1547 SV *res;
423cee85
JH
1548 STRLEN len;
1549 char *str;
4e553d73 1550
423cee85 1551 if (!e) {
5777a3f7 1552 yyerror("Missing right brace on \\N{}");
423cee85
JH
1553 e = s - 1;
1554 goto cont_scan;
1555 }
55eda711
JH
1556 res = newSVpvn(s + 1, e - s - 1);
1557 res = new_constant( Nullch, 0, "charnames",
1558 res, Nullsv, "\\N{...}" );
f9a63242
JH
1559 if (has_utf8)
1560 sv_utf8_upgrade(res);
423cee85 1561 str = SvPV(res,len);
89491803 1562 if (!has_utf8 && SvUTF8(res)) {
f08d6ad9
GS
1563 char *ostart = SvPVX(sv);
1564 SvCUR_set(sv, d - ostart);
1565 SvPOK_on(sv);
e4f3eed8 1566 *d = '\0';
f08d6ad9 1567 sv_utf8_upgrade(sv);
d2f449dd
SB
1568 /* this just broke our allocation above... */
1569 SvGROW(sv, send - start);
f08d6ad9 1570 d = SvPVX(sv) + SvCUR(sv);
89491803 1571 has_utf8 = TRUE;
f08d6ad9 1572 }
b239daa5 1573 if (len > e - s + 4) { /* I _guess_ 4 is \N{} --jhi */
423cee85
JH
1574 char *odest = SvPVX(sv);
1575
8973db79 1576 SvGROW(sv, (SvLEN(sv) + len - (e - s + 4)));
423cee85
JH
1577 d = SvPVX(sv) + (d - odest);
1578 }
1579 Copy(str, d, len, char);
1580 d += len;
1581 SvREFCNT_dec(res);
1582 cont_scan:
1583 s = e + 1;
1584 }
1585 else
5777a3f7 1586 yyerror("Missing braces on \\N{}");
423cee85
JH
1587 continue;
1588
02aa26ce 1589 /* \c is a control character */
79072805
LW
1590 case 'c':
1591 s++;
ba210ebe
JH
1592 {
1593 U8 c = *s++;
c7f1f016
NIS
1594#ifdef EBCDIC
1595 if (isLOWER(c))
1596 c = toUPPER(c);
1597#endif
db42d148 1598 *d++ = NATIVE_TO_NEED(has_utf8,toCTRL(c));
ba210ebe 1599 }
79072805 1600 continue;
02aa26ce
NT
1601
1602 /* printf-style backslashes, formfeeds, newlines, etc */
79072805 1603 case 'b':
db42d148 1604 *d++ = NATIVE_TO_NEED(has_utf8,'\b');
79072805
LW
1605 break;
1606 case 'n':
db42d148 1607 *d++ = NATIVE_TO_NEED(has_utf8,'\n');
79072805
LW
1608 break;
1609 case 'r':
db42d148 1610 *d++ = NATIVE_TO_NEED(has_utf8,'\r');
79072805
LW
1611 break;
1612 case 'f':
db42d148 1613 *d++ = NATIVE_TO_NEED(has_utf8,'\f');
79072805
LW
1614 break;
1615 case 't':
db42d148 1616 *d++ = NATIVE_TO_NEED(has_utf8,'\t');
79072805 1617 break;
34a3fe2a 1618 case 'e':
db42d148 1619 *d++ = ASCII_TO_NEED(has_utf8,'\033');
34a3fe2a
PP
1620 break;
1621 case 'a':
db42d148 1622 *d++ = ASCII_TO_NEED(has_utf8,'\007');
79072805 1623 break;
02aa26ce
NT
1624 } /* end switch */
1625
79072805
LW
1626 s++;
1627 continue;
02aa26ce
NT
1628 } /* end if (backslash) */
1629
f9a63242 1630 default_action:
2b9d42f0
NIS
1631 /* If we started with encoded form, or already know we want it
1632 and then encode the next character */
1633 if ((has_utf8 || this_utf8) && !NATIVE_IS_INVARIANT((U8)(*s))) {
1634 STRLEN len = 1;
1635 UV uv = (this_utf8) ? utf8n_to_uvchr((U8*)s, send - s, &len, 0) : (UV) ((U8) *s);
1636 STRLEN need = UNISKIP(NATIVE_TO_UNI(uv));
1637 s += len;
1638 if (need > len) {
1639 /* encoded value larger than old, need extra space (NOTE: SvCUR() not set here) */
1640 STRLEN off = d - SvPVX(sv);
1641 d = SvGROW(sv, SvLEN(sv) + (need-len)) + off;
1642 }
1643 d = (char*)uvchr_to_utf8((U8*)d, uv);
1644 has_utf8 = TRUE;
1645 }
1646 else {
1647 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1648 }
02aa26ce
NT
1649 } /* while loop to process each character */
1650
1651 /* terminate the string and set up the sv */
79072805 1652 *d = '\0';
463ee0b2 1653 SvCUR_set(sv, d - SvPVX(sv));
2b9d42f0 1654 if (SvCUR(sv) >= SvLEN(sv))
585602fa 1655 Perl_croak(aTHX_ "panic: constant overflowed allocated space");
2b9d42f0 1656
79072805 1657 SvPOK_on(sv);
9f4817db 1658 if (PL_encoding && !has_utf8) {
5d170f3a 1659 Perl_sv_recode_to_utf8(aTHX_ sv, PL_encoding);
9f4817db
JH
1660 has_utf8 = TRUE;
1661 }
2b9d42f0 1662 if (has_utf8) {
7e2040f0 1663 SvUTF8_on(sv);
2b9d42f0
NIS
1664 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1665 PL_sublex_info.sub_op->op_private |=
1666 (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1667 }
1668 }
79072805 1669
02aa26ce 1670 /* shrink the sv if we allocated more than we used */
79072805
LW
1671 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1672 SvLEN_set(sv, SvCUR(sv) + 1);
463ee0b2 1673 Renew(SvPVX(sv), SvLEN(sv), char);
79072805 1674 }
02aa26ce 1675
9b599b2a 1676 /* return the substring (via yylval) only if we parsed anything */
3280af22
NIS
1677 if (s > PL_bufptr) {
1678 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
4e553d73 1679 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
b3ac6de7 1680 sv, Nullsv,
4e553d73 1681 ( PL_lex_inwhat == OP_TRANS
b3ac6de7 1682 ? "tr"
3280af22 1683 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
b3ac6de7
IZ
1684 ? "s"
1685 : "qq")));
79072805 1686 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
b3ac6de7 1687 } else
8990e307 1688 SvREFCNT_dec(sv);
79072805
LW
1689 return s;
1690}
1691
ffb4593c
NT
1692/* S_intuit_more
1693 * Returns TRUE if there's more to the expression (e.g., a subscript),
1694 * FALSE otherwise.
ffb4593c
NT
1695 *
1696 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1697 *
1698 * ->[ and ->{ return TRUE
1699 * { and [ outside a pattern are always subscripts, so return TRUE
1700 * if we're outside a pattern and it's not { or [, then return FALSE
1701 * if we're in a pattern and the first char is a {
1702 * {4,5} (any digits around the comma) returns FALSE
1703 * if we're in a pattern and the first char is a [
1704 * [] returns FALSE
1705 * [SOMETHING] has a funky algorithm to decide whether it's a
1706 * character class or not. It has to deal with things like
1707 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1708 * anything else returns TRUE
1709 */
1710
9cbb5ea2
GS
1711/* This is the one truly awful dwimmer necessary to conflate C and sed. */
1712
76e3520e 1713STATIC int
cea2e8a9 1714S_intuit_more(pTHX_ register char *s)
79072805 1715{
3280af22 1716 if (PL_lex_brackets)
79072805
LW
1717 return TRUE;
1718 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1719 return TRUE;
1720 if (*s != '{' && *s != '[')
1721 return FALSE;
3280af22 1722 if (!PL_lex_inpat)
79072805
LW
1723 return TRUE;
1724
1725 /* In a pattern, so maybe we have {n,m}. */
1726 if (*s == '{') {
1727 s++;
1728 if (!isDIGIT(*s))
1729 return TRUE;
1730 while (isDIGIT(*s))
1731 s++;
1732 if (*s == ',')
1733 s++;
1734 while (isDIGIT(*s))
1735 s++;
1736 if (*s == '}')
1737 return FALSE;
1738 return TRUE;
1739
1740 }
1741
1742 /* On the other hand, maybe we have a character class */
1743
1744 s++;
1745 if (*s == ']' || *s == '^')
1746 return FALSE;
1747 else {
ffb4593c 1748 /* this is terrifying, and it works */
79072805
LW
1749 int weight = 2; /* let's weigh the evidence */
1750 char seen[256];
f27ffc4a 1751 unsigned char un_char = 255, last_un_char;
93a17b20 1752 char *send = strchr(s,']');
3280af22 1753 char tmpbuf[sizeof PL_tokenbuf * 4];
79072805
LW
1754
1755 if (!send) /* has to be an expression */
1756 return TRUE;
1757
1758 Zero(seen,256,char);
1759 if (*s == '$')
1760 weight -= 3;
1761 else if (isDIGIT(*s)) {
1762 if (s[1] != ']') {
1763 if (isDIGIT(s[1]) && s[2] == ']')
1764 weight -= 10;
1765 }
1766 else
1767 weight -= 100;
1768 }
1769 for (; s < send; s++) {
1770 last_un_char = un_char;
1771 un_char = (unsigned char)*s;
1772 switch (*s) {
1773 case '@':
1774 case '&':
1775 case '$':
1776 weight -= seen[un_char] * 10;
7e2040f0 1777 if (isALNUM_lazy_if(s+1,UTF)) {
8903cb82 1778 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
a0d0e21e 1779 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
79072805
LW
1780 weight -= 100;
1781 else
1782 weight -= 10;
1783 }
1784 else if (*s == '$' && s[1] &&
93a17b20
LW
1785 strchr("[#!%*<>()-=",s[1])) {
1786 if (/*{*/ strchr("])} =",s[2]))
79072805
LW
1787 weight -= 10;
1788 else
1789 weight -= 1;
1790 }
1791 break;
1792 case '\\':
1793 un_char = 254;
1794 if (s[1]) {
93a17b20 1795 if (strchr("wds]",s[1]))
79072805
LW
1796 weight += 100;
1797 else if (seen['\''] || seen['"'])
1798 weight += 1;
93a17b20 1799 else if (strchr("rnftbxcav",s[1]))
79072805
LW
1800 weight += 40;
1801 else if (isDIGIT(s[1])) {
1802 weight += 40;
1803 while (s[1] && isDIGIT(s[1]))
1804 s++;
1805 }
1806 }
1807 else
1808 weight += 100;
1809 break;
1810 case '-':
1811 if (s[1] == '\\')
1812 weight += 50;
93a17b20 1813 if (strchr("aA01! ",last_un_char))
79072805 1814 weight += 30;
93a17b20 1815 if (strchr("zZ79~",s[1]))
79072805 1816 weight += 30;
f27ffc4a
GS
1817 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1818 weight -= 5; /* cope with negative subscript */
79072805
LW
1819 break;
1820 default:
93a17b20 1821 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
79072805
LW
1822 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1823 char *d = tmpbuf;
1824 while (isALPHA(*s))
1825 *d++ = *s++;
1826 *d = '\0';
1827 if (keyword(tmpbuf, d - tmpbuf))
1828 weight -= 150;
1829 }
1830 if (un_char == last_un_char + 1)
1831 weight += 5;
1832 weight -= seen[un_char];
1833 break;
1834 }
1835 seen[un_char]++;
1836 }
1837 if (weight >= 0) /* probably a character class */
1838 return FALSE;
1839 }
1840
1841 return TRUE;
1842}
ffed7fef 1843
ffb4593c
NT
1844/*
1845 * S_intuit_method
1846 *
1847 * Does all the checking to disambiguate
1848 * foo bar
1849 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1850 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1851 *
1852 * First argument is the stuff after the first token, e.g. "bar".
1853 *
1854 * Not a method if bar is a filehandle.
1855 * Not a method if foo is a subroutine prototyped to take a filehandle.
1856 * Not a method if it's really "Foo $bar"
1857 * Method if it's "foo $bar"
1858 * Not a method if it's really "print foo $bar"
1859 * Method if it's really "foo package::" (interpreted as package->foo)
1860 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
3cb0bbe5 1861 * Not a method if bar is a filehandle or package, but is quoted with
ffb4593c
NT
1862 * =>
1863 */
1864
76e3520e 1865STATIC int
cea2e8a9 1866S_intuit_method(pTHX_ char *start, GV *gv)
a0d0e21e
LW
1867{
1868 char *s = start + (*start == '$');
3280af22 1869 char tmpbuf[sizeof PL_tokenbuf];
a0d0e21e
LW
1870 STRLEN len;
1871 GV* indirgv;
1872
1873 if (gv) {
b6c543e3 1874 CV *cv;
a0d0e21e
LW
1875 if (GvIO(gv))
1876 return 0;
b6c543e3
IZ
1877 if ((cv = GvCVu(gv))) {
1878 char *proto = SvPVX(cv);
1879 if (proto) {
1880 if (*proto == ';')
1881 proto++;
1882 if (*proto == '*')
1883 return 0;
1884 }
1885 } else
a0d0e21e
LW
1886 gv = 0;
1887 }
8903cb82 1888 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
ffb4593c
NT
1889 /* start is the beginning of the possible filehandle/object,
1890 * and s is the end of it
1891 * tmpbuf is a copy of it
1892 */
1893
a0d0e21e 1894 if (*start == '$') {
3280af22 1895 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
a0d0e21e
LW
1896 return 0;
1897 s = skipspace(s);
3280af22
NIS
1898 PL_bufptr = start;
1899 PL_expect = XREF;
a0d0e21e
LW
1900 return *s == '(' ? FUNCMETH : METHOD;
1901 }
1902 if (!keyword(tmpbuf, len)) {
c3e0f903
GS
1903 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1904 len -= 2;
1905 tmpbuf[len] = '\0';
1906 goto bare_package;
1907 }
1908 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
8ebc5c01 1909 if (indirgv && GvCVu(indirgv))
a0d0e21e
LW
1910 return 0;
1911 /* filehandle or package name makes it a method */
89bfa8cd 1912 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
a0d0e21e 1913 s = skipspace(s);
3280af22 1914 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
55497cff 1915 return 0; /* no assumptions -- "=>" quotes bearword */
c3e0f903 1916 bare_package:
3280af22 1917 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
79cb57f6 1918 newSVpvn(tmpbuf,len));
3280af22
NIS
1919 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1920 PL_expect = XTERM;
a0d0e21e 1921 force_next(WORD);
3280af22 1922 PL_bufptr = s;
a0d0e21e
LW
1923 return *s == '(' ? FUNCMETH : METHOD;
1924 }
1925 }
1926 return 0;
1927}
1928
ffb4593c
NT
1929/*
1930 * S_incl_perldb
1931 * Return a string of Perl code to load the debugger. If PERL5DB
1932 * is set, it will return the contents of that, otherwise a
1933 * compile-time require of perl5db.pl.
1934 */
1935
76e3520e 1936STATIC char*
cea2e8a9 1937S_incl_perldb(pTHX)
a0d0e21e 1938{
3280af22 1939 if (PL_perldb) {
76e3520e 1940 char *pdb = PerlEnv_getenv("PERL5DB");
a0d0e21e
LW
1941
1942 if (pdb)
1943 return pdb;
91487cfc 1944 SETERRNO(0,SS$_NORMAL);
a0d0e21e
LW
1945 return "BEGIN { require 'perl5db.pl' }";
1946 }
1947 return "";
1948}
1949
1950
16d20bd9 1951/* Encoded script support. filter_add() effectively inserts a
4e553d73 1952 * 'pre-processing' function into the current source input stream.
16d20bd9
AD
1953 * Note that the filter function only applies to the current source file
1954 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1955 *
1956 * The datasv parameter (which may be NULL) can be used to pass
1957 * private data to this instance of the filter. The filter function
1958 * can recover the SV using the FILTER_DATA macro and use it to
1959 * store private buffers and state information.
1960 *
1961 * The supplied datasv parameter is upgraded to a PVIO type
4755096e 1962 * and the IoDIRP/IoANY field is used to store the function pointer,
e0c19803 1963 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
16d20bd9
AD
1964 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1965 * private use must be set using malloc'd pointers.
1966 */
16d20bd9
AD
1967
1968SV *
864dbfa3 1969Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
16d20bd9 1970{
f4c556ac
GS
1971 if (!funcp)
1972 return Nullsv;
1973
3280af22
NIS
1974 if (!PL_rsfp_filters)
1975 PL_rsfp_filters = newAV();
16d20bd9 1976 if (!datasv)
8c52afec 1977 datasv = NEWSV(255,0);
16d20bd9 1978 if (!SvUPGRADE(datasv, SVt_PVIO))
cea2e8a9 1979 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
4755096e 1980 IoANY(datasv) = (void *)funcp; /* stash funcp into spare field */
e0c19803 1981 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
f4c556ac 1982 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
fe5a182c 1983 (void*)funcp, SvPV_nolen(datasv)));
3280af22
NIS
1984 av_unshift(PL_rsfp_filters, 1);
1985 av_store(PL_rsfp_filters, 0, datasv) ;
16d20bd9
AD
1986 return(datasv);
1987}
4e553d73 1988
16d20bd9
AD
1989
1990/* Delete most recently added instance of this filter function. */
a0d0e21e 1991void
864dbfa3 1992Perl_filter_del(pTHX_ filter_t funcp)
16d20bd9 1993{
e0c19803 1994 SV *datasv;
fe5a182c 1995 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", (void*)funcp));
3280af22 1996 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
16d20bd9
AD
1997 return;
1998 /* if filter is on top of stack (usual case) just pop it off */
e0c19803 1999 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
4755096e 2000 if (IoANY(datasv) == (void *)funcp) {
e0c19803 2001 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
4755096e 2002 IoANY(datasv) = (void *)NULL;
3280af22 2003 sv_free(av_pop(PL_rsfp_filters));
e50aee73 2004
16d20bd9
AD
2005 return;
2006 }
2007 /* we need to search for the correct entry and clear it */
cea2e8a9 2008 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
16d20bd9
AD
2009}
2010
2011
2012/* Invoke the n'th filter function for the current rsfp. */
2013I32
864dbfa3 2014Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
4e553d73
NIS
2015
2016
8ac85365 2017 /* 0 = read one text line */
a0d0e21e 2018{
16d20bd9
AD
2019 filter_t funcp;
2020 SV *datasv = NULL;
e50aee73 2021
3280af22 2022 if (!PL_rsfp_filters)
16d20bd9 2023 return -1;
3280af22 2024 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
16d20bd9
AD
2025 /* Provide a default input filter to make life easy. */
2026 /* Note that we append to the line. This is handy. */
f4c556ac
GS
2027 DEBUG_P(PerlIO_printf(Perl_debug_log,
2028 "filter_read %d: from rsfp\n", idx));
4e553d73 2029 if (maxlen) {
16d20bd9
AD
2030 /* Want a block */
2031 int len ;
2032 int old_len = SvCUR(buf_sv) ;
2033
2034 /* ensure buf_sv is large enough */
2035 SvGROW(buf_sv, old_len + maxlen) ;
3280af22
NIS
2036 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
2037 if (PerlIO_error(PL_rsfp))
37120919
AD
2038 return -1; /* error */
2039 else
2040 return 0 ; /* end of file */
2041 }
16d20bd9
AD
2042 SvCUR_set(buf_sv, old_len + len) ;
2043 } else {
2044 /* Want a line */
3280af22
NIS
2045 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
2046 if (PerlIO_error(PL_rsfp))
37120919
AD
2047 return -1; /* error */
2048 else
2049 return 0 ; /* end of file */
2050 }
16d20bd9
AD
2051 }
2052 return SvCUR(buf_sv);
2053 }
2054 /* Skip this filter slot if filter has been deleted */
3280af22 2055 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
f4c556ac
GS
2056 DEBUG_P(PerlIO_printf(Perl_debug_log,
2057 "filter_read %d: skipped (filter deleted)\n",
2058 idx));
16d20bd9
AD
2059 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
2060 }
2061 /* Get function pointer hidden within datasv */
4755096e 2062 funcp = (filter_t)IoANY(datasv);
f4c556ac
GS
2063 DEBUG_P(PerlIO_printf(Perl_debug_log,
2064 "filter_read %d: via function %p (%s)\n",
fe5a182c 2065 idx, (void*)funcp, SvPV_nolen(datasv)));
16d20bd9
AD
2066 /* Call function. The function is expected to */
2067 /* call "FILTER_READ(idx+1, buf_sv)" first. */
37120919 2068 /* Return: <0:error, =0:eof, >0:not eof */
acfe0abc 2069 return (*funcp)(aTHX_ idx, buf_sv, maxlen);
16d20bd9
AD
2070}
2071
76e3520e 2072STATIC char *
cea2e8a9 2073S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
16d20bd9 2074{
c39cd008 2075#ifdef PERL_CR_FILTER
3280af22 2076 if (!PL_rsfp_filters) {
c39cd008 2077 filter_add(S_cr_textfilter,NULL);
a868473f
NIS
2078 }
2079#endif
3280af22 2080 if (PL_rsfp_filters) {
16d20bd9 2081
55497cff
PP
2082 if (!append)
2083 SvCUR_set(sv, 0); /* start with empty line */
16d20bd9
AD
2084 if (FILTER_READ(0, sv, 0) > 0)
2085 return ( SvPVX(sv) ) ;
2086 else
2087 return Nullch ;
2088 }
9d116dd7 2089 else
fd049845 2090 return (sv_gets(sv, fp, append));
a0d0e21e
LW
2091}
2092
01ec43d0
GS
2093STATIC HV *
2094S_find_in_my_stash(pTHX_ char *pkgname, I32 len)
def3634b
GS
2095{
2096 GV *gv;
2097
01ec43d0 2098 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
def3634b
GS
2099 return PL_curstash;
2100
2101 if (len > 2 &&
2102 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
01ec43d0
GS
2103 (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV)))
2104 {
2105 return GvHV(gv); /* Foo:: */
def3634b
GS
2106 }
2107
2108 /* use constant CLASS => 'MyClass' */
2109 if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) {
2110 SV *sv;
2111 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2112 pkgname = SvPV_nolen(sv);
2113 }
2114 }
2115
2116 return gv_stashpv(pkgname, FALSE);
2117}
a0d0e21e 2118
748a9306
LW
2119#ifdef DEBUGGING
2120 static char* exp_name[] =
09bef843
SB
2121 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2122 "ATTRTERM", "TERMBLOCK"
2123 };
748a9306 2124#endif
463ee0b2 2125
02aa26ce
NT
2126/*
2127 yylex
2128
2129 Works out what to call the token just pulled out of the input
2130 stream. The yacc parser takes care of taking the ops we return and
2131 stitching them into a tree.
2132
2133 Returns:
2134 PRIVATEREF
2135
2136 Structure:
2137 if read an identifier
2138 if we're in a my declaration
2139 croak if they tried to say my($foo::bar)
2140 build the ops for a my() declaration
2141 if it's an access to a my() variable
2142 are we in a sort block?
2143 croak if my($a); $a <=> $b
2144 build ops for access to a my() variable
2145 if in a dq string, and they've said @foo and we can't find @foo
2146 croak
2147 build ops for a bareword
2148 if we already built the token before, use it.
2149*/
2150
dba4d153 2151#ifdef USE_PURE_BISON
864dbfa3 2152int
dba4d153 2153Perl_yylex_r(pTHX_ YYSTYPE *lvalp, int *lcharp)
378cc40b 2154{
20141f0e
RI
2155 int r;
2156
6f202aea 2157 yyactlevel++;
20141f0e
RI
2158 yylval_pointer[yyactlevel] = lvalp;
2159 yychar_pointer[yyactlevel] = lcharp;
b73d6f50
RI
2160 if (yyactlevel >= YYMAXLEVEL)
2161 Perl_croak(aTHX_ "panic: YYMAXLEVEL");
20141f0e 2162
dba4d153 2163 r = Perl_yylex(aTHX);
20141f0e 2164
d8ae6756
RI
2165 if (yyactlevel > 0)
2166 yyactlevel--;
20141f0e
RI
2167
2168 return r;
2169}
dba4d153 2170#endif
20141f0e 2171
dba4d153
JH
2172#ifdef __SC__
2173#pragma segment Perl_yylex
2174#endif
dba4d153 2175int
dba4d153 2176Perl_yylex(pTHX)
20141f0e 2177{
79072805 2178 register char *s;
378cc40b 2179 register char *d;
79072805 2180 register I32 tmp;
463ee0b2 2181 STRLEN len;
161b471a
NIS
2182 GV *gv = Nullgv;
2183 GV **gvp = 0;
aa7440fb 2184 bool bof = FALSE;
a687059c 2185
02aa26ce 2186 /* check if there's an identifier for us to look at */
8eceec63 2187 if (PL_pending_ident)
e930465f 2188 return S_pending_ident(aTHX);
bbce6d69 2189
02aa26ce
NT
2190 /* no identifier pending identification */
2191
3280af22 2192 switch (PL_lex_state) {
79072805
LW
2193#ifdef COMMENTARY
2194 case LEX_NORMAL: /* Some compilers will produce faster */
2195 case LEX_INTERPNORMAL: /* code if we comment these out. */
2196 break;
2197#endif
2198
09bef843 2199 /* when we've already built the next token, just pull it out of the queue */
79072805 2200 case LEX_KNOWNEXT:
3280af22
NIS
2201 PL_nexttoke--;
2202 yylval = PL_nextval[PL_nexttoke];
2203 if (!PL_nexttoke) {
2204 PL_lex_state = PL_lex_defer;
2205 PL_expect = PL_lex_expect;
2206 PL_lex_defer = LEX_NORMAL;
463ee0b2 2207 }
607df283 2208 DEBUG_T({ PerlIO_printf(Perl_debug_log,
4659c93f 2209 "### Next token after '%s' was known, type %"IVdf"\n", PL_bufptr,
5f80b19c 2210 (IV)PL_nexttype[PL_nexttoke]); });
607df283 2211
3280af22 2212 return(PL_nexttype[PL_nexttoke]);
79072805 2213
02aa26ce 2214 /* interpolated case modifiers like \L \U, including \Q and \E.
3280af22 2215 when we get here, PL_bufptr is at the \
02aa26ce 2216 */
79072805
LW
2217 case LEX_INTERPCASEMOD:
2218#ifdef DEBUGGING
3280af22 2219 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
cea2e8a9 2220 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
79072805 2221#endif
02aa26ce 2222 /* handle \E or end of string */
3280af22 2223 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
a0d0e21e 2224 char oldmod;
02aa26ce
NT
2225
2226 /* if at a \E */
3280af22
NIS
2227 if (PL_lex_casemods) {
2228 oldmod = PL_lex_casestack[--PL_lex_casemods];
2229 PL_lex_casestack[PL_lex_casemods] = '\0';
02aa26ce 2230
3280af22
NIS
2231 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2232 PL_bufptr += 2;
2233 PL_lex_state = LEX_INTERPCONCAT;
a0d0e21e 2234 }
79072805
LW
2235 return ')';
2236 }
3280af22
NIS
2237 if (PL_bufptr != PL_bufend)
2238 PL_bufptr += 2;
2239 PL_lex_state = LEX_INTERPCONCAT;
cea2e8a9 2240 return yylex();
79072805
LW
2241 }
2242 else {
607df283 2243 DEBUG_T({ PerlIO_printf(Perl_debug_log,
5f80b19c 2244 "### Saw case modifier at '%s'\n", PL_bufptr); });
3280af22 2245 s = PL_bufptr + 1;
79072805
LW
2246 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2247 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
a0d0e21e 2248 if (strchr("LU", *s) &&
3280af22 2249 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
a0d0e21e 2250 {
3280af22 2251 PL_lex_casestack[--PL_lex_casemods] = '\0';
a0d0e21e
LW
2252 return ')';
2253 }
3280af22
NIS
2254 if (PL_lex_casemods > 10) {
2255 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2256 if (newlb != PL_lex_casestack) {
a0d0e21e 2257 SAVEFREEPV(newlb);
3280af22 2258 PL_lex_casestack = newlb;
a0d0e21e
LW
2259 }
2260 }
3280af22
NIS
2261 PL_lex_casestack[PL_lex_casemods++] = *s;
2262 PL_lex_casestack[PL_lex_casemods] = '\0';
2263 PL_lex_state = LEX_INTERPCONCAT;
2264 PL_nextval[PL_nexttoke].ival = 0;
79072805
LW
2265 force_next('(');
2266 if (*s == 'l')
3280af22 2267 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
79072805 2268 else if (*s == 'u')
3280af22 2269 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
79072805 2270 else if (*s == 'L')
3280af22 2271 PL_nextval[PL_nexttoke].ival = OP_LC;
79072805 2272 else if (*s == 'U')
3280af22 2273 PL_nextval[PL_nexttoke].ival = OP_UC;
a0d0e21e 2274 else if (*s == 'Q')
3280af22 2275 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
79072805 2276 else
cea2e8a9 2277 Perl_croak(aTHX_ "panic: yylex");
3280af22 2278 PL_bufptr = s + 1;
79072805 2279 force_next(FUNC);
3280af22
NIS
2280 if (PL_lex_starts) {
2281 s = PL_bufptr;
2282 PL_lex_starts = 0;
79072805
LW
2283 Aop(OP_CONCAT);
2284 }
2285 else
cea2e8a9 2286 return yylex();
79072805
LW
2287 }
2288
55497cff
PP
2289 case LEX_INTERPPUSH:
2290 return sublex_push();
2291
79072805 2292 case LEX_INTERPSTART:
3280af22 2293 if (PL_bufptr == PL_bufend)
79072805 2294 return sublex_done();
607df283 2295 DEBUG_T({ PerlIO_printf(Perl_debug_log,
5f80b19c 2296 "### Interpolated variable at '%s'\n", PL_bufptr); });
3280af22
NIS
2297 PL_expect = XTERM;
2298 PL_lex_dojoin = (*PL_bufptr == '@');
2299 PL_lex_state = LEX_INTERPNORMAL;
2300 if (PL_lex_dojoin) {
2301 PL_nextval[PL_nexttoke].ival = 0;
79072805 2302 force_next(',');
4d1ff10f 2303#ifdef USE_5005THREADS
533c011a
NIS
2304 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2305 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
554b3eca
MB
2306 force_next(PRIVATEREF);
2307#else
a0d0e21e 2308 force_ident("\"", '$');
4d1ff10f 2309#endif /* USE_5005THREADS */
3280af22 2310 PL_nextval[PL_nexttoke].ival = 0;
79072805 2311 force_next('$');
3280af22 2312 PL_nextval[PL_nexttoke].ival = 0;
79072805 2313 force_next('(');
3280af22 2314 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
79072805
LW
2315 force_next(FUNC);
2316 }
3280af22
NIS
2317 if (PL_lex_starts++) {
2318 s = PL_bufptr;
79072805
LW
2319 Aop(OP_CONCAT);
2320 }
cea2e8a9 2321 return yylex();
79072805
LW
2322
2323 case LEX_INTERPENDMAYBE:
3280af22
NIS
2324 if (intuit_more(PL_bufptr)) {
2325 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
79072805
LW
2326 break;
2327 }
2328 /* FALL THROUGH */
2329
2330 case LEX_INTERPEND:
3280af22
NIS
2331 if (PL_lex_dojoin) {
2332 PL_lex_dojoin = FALSE;
2333 PL_lex_state = LEX_INTERPCONCAT;
79072805
LW
2334 return ')';
2335 }
43a16006 2336 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
25da4f38 2337 && SvEVALED(PL_lex_repl))
43a16006 2338 {
e9fa98b2 2339 if (PL_bufptr != PL_bufend)
cea2e8a9 2340 Perl_croak(aTHX_ "Bad evalled substitution pattern");
e9fa98b2
HS
2341 PL_lex_repl = Nullsv;
2342 }
79072805
LW
2343 /* FALLTHROUGH */
2344 case LEX_INTERPCONCAT:
2345#ifdef DEBUGGING
3280af22 2346 if (PL_lex_brackets)
cea2e8a9 2347 Perl_croak(aTHX_ "panic: INTERPCONCAT");
79072805 2348#endif
3280af22 2349 if (PL_bufptr == PL_bufend)
79072805
LW
2350 return sublex_done();
2351
3280af22
NIS
2352 if (SvIVX(PL_linestr) == '\'') {
2353 SV *sv = newSVsv(PL_linestr);
2354 if (!PL_lex_inpat)
76e3520e 2355 sv = tokeq(sv);
3280af22 2356 else if ( PL_hints & HINT_NEW_RE )
b3ac6de7 2357 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
79072805 2358 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3280af22 2359 s = PL_bufend;
79072805
LW
2360 }
2361 else {
3280af22 2362 s = scan_const(PL_bufptr);
79072805 2363 if (*s == '\\')
3280af22 2364 PL_lex_state = LEX_INTERPCASEMOD;
79072805 2365 else
3280af22 2366 PL_lex_state = LEX_INTERPSTART;
79072805
LW
2367 }
2368
3280af22
NIS
2369 if (s != PL_bufptr) {
2370 PL_nextval[PL_nexttoke] = yylval;
2371 PL_expect = XTERM;
79072805 2372 force_next(THING);
3280af22 2373 if (PL_lex_starts++)
79072805
LW
2374 Aop(OP_CONCAT);
2375 else {
3280af22 2376 PL_bufptr = s;
cea2e8a9 2377 return yylex();
79072805
LW
2378 }
2379 }
2380
cea2e8a9 2381 return yylex();
a0d0e21e 2382 case LEX_FORMLINE:
3280af22
NIS
2383 PL_lex_state = LEX_NORMAL;
2384 s = scan_formline(PL_bufptr);
2385 if (!PL_lex_formbrack)
a0d0e21e
LW
2386 goto rightbracket;
2387 OPERATOR(';');
79072805
LW
2388 }
2389
3280af22
NIS
2390 s = PL_bufptr;
2391 PL_oldoldbufptr = PL_oldbufptr;
2392 PL_oldbufptr = s;
607df283 2393 DEBUG_T( {
bf49b057
GS
2394 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2395 exp_name[PL_expect], s);
5f80b19c 2396 } );
463ee0b2
LW
2397
2398 retry:
378cc40b
LW
2399 switch (*s) {
2400 default:
7e2040f0 2401 if (isIDFIRST_lazy_if(s,UTF))
834a4ddd 2402 goto keylookup;
cea2e8a9 2403 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
e929a76b
LW
2404 case 4:
2405 case 26:
2406 goto fake_eof; /* emulate EOF on ^D or ^Z */
378cc40b 2407 case 0:
3280af22
NIS
2408 if (!PL_rsfp) {
2409 PL_last_uni = 0;
2410 PL_last_lop = 0;
2411 if (PL_lex_brackets)
d98d5fff 2412 yyerror("Missing right curly or square bracket");
4e553d73 2413 DEBUG_T( { PerlIO_printf(Perl_debug_log,
607df283 2414 "### Tokener got EOF\n");
5f80b19c 2415 } );
79072805 2416 TOKEN(0);
463ee0b2 2417 }
3280af22 2418 if (s++ < PL_bufend)
a687059c 2419 goto retry; /* ignore stray nulls */
3280af22
NIS
2420 PL_last_uni = 0;
2421 PL_last_lop = 0;
2422 if (!PL_in_eval && !PL_preambled) {
2423 PL_preambled = TRUE;
2424 sv_setpv(PL_linestr,incl_perldb());
2425 if (SvCUR(PL_linestr))
2426 sv_catpv(PL_linestr,";");
2427 if (PL_preambleav){
2428 while(AvFILLp(PL_preambleav) >= 0) {
2429 SV *tmpsv = av_shift(PL_preambleav);
2430 sv_catsv(PL_linestr, tmpsv);
2431 sv_catpv(PL_linestr, ";");
91b7def8
PP
2432 sv_free(tmpsv);
2433 }
3280af22
NIS
2434 sv_free((SV*)PL_preambleav);
2435 PL_preambleav = NULL;
91b7def8 2436 }
3280af22
NIS
2437 if (PL_minus_n || PL_minus_p) {
2438 sv_catpv(PL_linestr, "LINE: while (<>) {");
2439 if (PL_minus_l)
2440 sv_catpv(PL_linestr,"chomp;");
2441 if (PL_minus_a) {
3280af22
NIS
2442 if (PL_minus_F) {
2443 if (strchr("/'\"", *PL_splitstr)
2444 && strchr(PL_splitstr + 1, *PL_splitstr))
3db68c4c 2445 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s);", PL_splitstr);
54310121
PP
2446 else {
2447 char delim;
2448 s = "'~#\200\1'"; /* surely one char is unused...*/
3280af22 2449 while (s[1] && strchr(PL_splitstr, *s)) s++;
54310121 2450 delim = *s;
75c72d73 2451 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s%c",
46fc3d4c 2452 "q" + (delim == '\''), delim);
3280af22 2453 for (s = PL_splitstr; *s; s++) {
54310121 2454 if (*s == '\\')
3280af22
NIS
2455 sv_catpvn(PL_linestr, "\\", 1);
2456 sv_catpvn(PL_linestr, s, 1);
54310121 2457 }
cea2e8a9 2458 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
54310121 2459 }
2304df62
AD
2460 }
2461 else
75c72d73 2462 sv_catpv(PL_linestr,"our @F=split(' ');");
2304df62 2463 }
79072805 2464 }
3280af22
NIS
2465 sv_catpv(PL_linestr, "\n");
2466 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2467 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
207e3d1a 2468 PL_last_lop = PL_last_uni = Nullch;
3280af22 2469 if (PERLDB_LINE && PL_curstash != PL_debstash) {
a0d0e21e
LW
2470 SV *sv = NEWSV(85,0);
2471
2472 sv_upgrade(sv, SVt_PVMG);
3280af22 2473 sv_setsv(sv,PL_linestr);
0ac0412a
MJD
2474 (void)SvIOK_on(sv);
2475 SvIVX(sv) = 0;
57843af0 2476 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
a0d0e21e 2477 }
79072805 2478 goto retry;
a687059c 2479 }
e929a76b 2480 do {
aa7440fb 2481 bof = PL_rsfp ? TRUE : FALSE;
7e28d3af
JH
2482 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2483 fake_eof:
2484 if (PL_rsfp) {
2485 if (PL_preprocess && !PL_in_eval)
2486 (void)PerlProc_pclose(PL_rsfp);
2487 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2488 PerlIO_clearerr(PL_rsfp);
2489 else
2490 (void)PerlIO_close(PL_rsfp);
2491 PL_rsfp = Nullfp;
2492 PL_doextract = FALSE;
2493 }
2494 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2495 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2496 sv_catpv(PL_linestr,";}");
2497 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2498 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
207e3d1a 2499 PL_last_lop = PL_last_uni = Nullch;
7e28d3af
JH
2500 PL_minus_n = PL_minus_p = 0;
2501 goto retry;
2502 }
2503 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
207e3d1a 2504 PL_last_lop = PL_last_uni = Nullch;
7e28d3af
JH
2505 sv_setpv(PL_linestr,"");
2506 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2507 }
2508 /* if it looks like the start of a BOM, check if it in fact is */
2509 else if (bof && (!*s || *(U8*)s == 0xEF || *(U8*)s >= 0xFE)) {
226017aa 2510#ifdef PERLIO_IS_STDIO
e3f494f1
JH
2511# ifdef __GNU_LIBRARY__
2512# if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
226017aa
DD
2513# define FTELL_FOR_PIPE_IS_BROKEN
2514# endif
e3f494f1
JH
2515# else
2516# ifdef __GLIBC__
2517# if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
2518# define FTELL_FOR_PIPE_IS_BROKEN
2519# endif
2520# endif
226017aa
DD
2521# endif
2522#endif
2523#ifdef FTELL_FOR_PIPE_IS_BROKEN
2524 /* This loses the possibility to detect the bof
2525 * situation on perl -P when the libc5 is being used.
2526 * Workaround? Maybe attach some extra state to PL_rsfp?
2527 */
2528 if (!PL_preprocess)
7e28d3af 2529 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
226017aa 2530#else
7e28d3af 2531 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
226017aa 2532#endif
7e28d3af 2533 if (bof) {
3280af22 2534 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7e28d3af 2535 s = swallow_bom((U8*)s);
e929a76b 2536 }
378cc40b 2537 }
3280af22 2538 if (PL_doextract) {
a0d0e21e 2539 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
3280af22 2540 PL_doextract = FALSE;
a0d0e21e
LW
2541
2542 /* Incest with pod. */
2543 if (*s == '=' && strnEQ(s, "=cut", 4)) {
3280af22
NIS
2544 sv_setpv(PL_linestr, "");
2545 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2546 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
207e3d1a 2547 PL_last_lop = PL_last_uni = Nullch;
3280af22 2548 PL_doextract = FALSE;
a0d0e21e 2549 }
4e553d73 2550 }
463ee0b2 2551 incline(s);
3280af22
NIS
2552 } while (PL_doextract);
2553 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2554 if (PERLDB_LINE && PL_curstash != PL_debstash) {
79072805 2555 SV *sv = NEWSV(85,0);
a687059c 2556
93a17b20 2557 sv_upgrade(sv, SVt_PVMG);
3280af22 2558 sv_setsv(sv,PL_linestr);
0ac0412a
MJD
2559 (void)SvIOK_on(sv);
2560 SvIVX(sv) = 0;
57843af0 2561 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
a687059c 2562 }
3280af22 2563 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
207e3d1a 2564 PL_last_lop = PL_last_uni = Nullch;
57843af0 2565 if (CopLINE(PL_curcop) == 1) {
3280af22 2566 while (s < PL_bufend && isSPACE(*s))
79072805 2567 s++;
a0d0e21e 2568 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
79072805 2569 s++;
44a8e56a 2570 d = Nullch;
3280af22 2571 if (!PL_in_eval) {
44a8e56a
PP
2572 if (*s == '#' && *(s+1) == '!')
2573 d = s + 2;
2574#ifdef ALTERNATE_SHEBANG
2575 else {
2576 static char as[] = ALTERNATE_SHEBANG;
2577 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2578 d = s + (sizeof(as) - 1);
2579 }
2580#endif /* ALTERNATE_SHEBANG */
2581 }
2582 if (d) {
b8378b72 2583 char *ipath;
774d564b 2584 char *ipathend;
b8378b72 2585
774d564b 2586 while (isSPACE(*d))
b8378b72
CS
2587 d++;
2588 ipath = d;
774d564b
PP
2589 while (*d && !isSPACE(*d))
2590 d++;
2591 ipathend = d;
2592
2593#ifdef ARG_ZERO_IS_SCRIPT
2594 if (ipathend > ipath) {
2595 /*
2596 * HP-UX (at least) sets argv[0] to the script name,
2597 * which makes $^X incorrect. And Digital UNIX and Linux,
2598 * at least, set argv[0] to the basename of the Perl
2599 * interpreter. So, having found "#!", we'll set it right.
2600 */
ee2f7564 2601 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV)); /* $^X */
774d564b 2602 assert(SvPOK(x) || SvGMAGICAL(x));
cc49e20b 2603 if (sv_eq(x, CopFILESV(PL_curcop))) {
774d564b 2604 sv_setpvn(x, ipath, ipathend - ipath);
9607fc9c
PP
2605 SvSETMAGIC(x);
2606 }
774d564b 2607 TAINT_NOT; /* $^X is always tainted, but that's OK */
8ebc5c01 2608 }
774d564b 2609#endif /* ARG_ZERO_IS_SCRIPT */
b8378b72
CS
2610
2611 /*
2612 * Look for options.
2613 */
748a9306 2614 d = instr(s,"perl -");
84e30d1a 2615 if (!d) {
748a9306 2616 d = instr(s,"perl");
84e30d1a
GS
2617#if defined(DOSISH)
2618 /* avoid getting into infinite loops when shebang
2619 * line contains "Perl" rather than "perl" */
2620 if (!d) {
2621 for (d = ipathend-4; d >= ipath; --d) {
2622 if ((*d == 'p' || *d == 'P')
2623 && !ibcmp(d, "perl", 4))
2624 {
2625 break;
2626 }
2627 }
2628 if (d < ipath)
2629 d = Nullch;
2630 }
2631#endif
2632 }
44a8e56a
PP
2633#ifdef ALTERNATE_SHEBANG
2634 /*
2635 * If the ALTERNATE_SHEBANG on this system starts with a
2636 * character that can be part of a Perl expression, then if
2637 * we see it but not "perl", we're probably looking at the
2638 * start of Perl code, not a request to hand off to some
2639 * other interpreter. Similarly, if "perl" is there, but
2640 * not in the first 'word' of the line, we assume the line
2641 * contains the start of the Perl program.
44a8e56a
PP
2642 */
2643 if (d && *s != '#') {
774d564b 2644 char *c = ipath;
44a8e56a
PP
2645 while (*c && !strchr("; \t\r\n\f\v#", *c))
2646 c++;
2647 if (c < d)
2648 d = Nullch; /* "perl" not in first word; ignore */
2649 else
2650 *s = '#'; /* Don't try to parse shebang line */
2651 }
774d564b 2652#endif /* ALTERNATE_SHEBANG */
bf4acbe4 2653#ifndef MACOS_TRADITIONAL
748a9306 2654 if (!d &&
44a8e56a 2655 *s == '#' &&
774d564b 2656 ipathend > ipath &&
3280af22 2657 !PL_minus_c &&
748a9306 2658 !instr(s,"indir") &&
3280af22 2659 instr(PL_origargv[0],"perl"))
748a9306 2660 {
9f68db38 2661 char **newargv;
9f68db38 2662
774d564b
PP
2663 *ipathend = '\0';
2664 s = ipathend + 1;
3280af22 2665 while (s < PL_bufend && isSPACE(*s))
9f68db38 2666 s++;
3280af22
NIS
2667 if (s < PL_bufend) {
2668 Newz(899,newargv,PL_origargc+3,char*);
9f68db38 2669 newargv[1] = s;
3280af22 2670 while (s < PL_bufend && !isSPACE(*s))
9f68db38
LW
2671 s++;
2672 *s = '\0';
3280af22 2673 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
9f68db38
LW
2674 }
2675 else
3280af22 2676 newargv = PL_origargv;
774d564b 2677 newargv[0] = ipath;
b4748376 2678 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
cea2e8a9 2679 Perl_croak(aTHX_ "Can't exec %s", ipath);
9f68db38 2680 }
bf4acbe4 2681#endif
748a9306 2682 if (d) {
3280af22
NIS
2683 U32 oldpdb = PL_perldb;
2684 bool oldn = PL_minus_n;
2685 bool oldp = PL_minus_p;
748a9306
LW
2686
2687 while (*d && !isSPACE(*d)) d++;
bf4acbe4 2688 while (SPACE_OR_TAB(*d)) d++;
748a9306
LW
2689
2690 if (*d++ == '-') {
a11ec5a9 2691 bool switches_done = PL_doswitches;
8cc95fdb
PP
2692 do {
2693 if (*d == 'M' || *d == 'm') {
2694 char *m = d;
2695 while (*d && !isSPACE(*d)) d++;
cea2e8a9 2696 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
8cc95fdb
PP
2697 (int)(d - m), m);
2698 }
2699 d = moreswitches(d);
2700 } while (d);
155aba94
GS
2701 if ((PERLDB_LINE && !oldpdb) ||
2702 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
b084f20b
PP
2703 /* if we have already added "LINE: while (<>) {",
2704 we must not do it again */
748a9306 2705 {
3280af22
NIS
2706 sv_setpv(PL_linestr, "");
2707 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2708 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
207e3d1a 2709 PL_last_lop = PL_last_uni = Nullch;
3280af22 2710 PL_preambled = FALSE;
84902520 2711 if (PERLDB_LINE)
3280af22 2712 (void)gv_fetchfile(PL_origfilename);
748a9306
LW
2713 goto retry;
2714 }
a11ec5a9
RGS
2715 if (PL_doswitches && !switches_done) {
2716 int argc = PL_origargc;
2717 char **argv = PL_origargv;
2718 do {
2719 argc--,argv++;
2720 } while (argc && argv[0][0] == '-' && argv[0][1]);
2721 init_argv_symbols(argc,argv);
2722 }
a0d0e21e 2723 }
79072805 2724 }
9f68db38 2725 }
79072805 2726 }
3280af22
NIS
2727 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2728 PL_bufptr = s;
2729 PL_lex_state = LEX_FORMLINE;
cea2e8a9 2730 return yylex();
ae986130 2731 }
378cc40b 2732 goto retry;
4fdae800 2733 case '\r':
6a27c188 2734#ifdef PERL_STRICT_CR
cea2e8a9 2735 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
4e553d73 2736 Perl_croak(aTHX_
cc507455 2737 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
a868473f 2738#endif
4fdae800 2739 case ' ': case '\t': case '\f': case 013:
bf4acbe4
GS
2740#ifdef MACOS_TRADITIONAL
2741 case '\312':
2742#endif
378cc40b
LW
2743 s++;
2744 goto retry;
378cc40b 2745 case '#':
e929a76b 2746 case '\n':
3280af22 2747 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
df0deb90
GS
2748 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2749 /* handle eval qq[#line 1 "foo"\n ...] */
2750 CopLINE_dec(PL_curcop);
2751 incline(s);
2752 }
3280af22 2753 d = PL_bufend;
a687059c 2754 while (s < d && *s != '\n')
378cc40b 2755 s++;
0f85fab0 2756 if (s < d)
378cc40b 2757 s++;
78c267c1 2758 else if (s > d) /* Found by Ilya: feed random input to Perl. */
a8406387 2759 Perl_croak(aTHX_ "panic: input overflow");
463ee0b2 2760 incline(s);
3280af22
NIS
2761 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2762 PL_bufptr = s;
2763 PL_lex_state = LEX_FORMLINE;
cea2e8a9 2764 return yylex();
a687059c 2765 }
378cc40b 2766 }
a687059c 2767 else {
378cc40b 2768 *s = '\0';
3280af22 2769 PL_bufend = s;
a687059c 2770 }
378cc40b
LW
2771 goto retry;
2772 case '-':
79072805 2773 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
e5edeb50
JH
2774 I32 ftst = 0;
2775
378cc40b 2776 s++;
3280af22 2777 PL_bufptr = s;
748a9306
LW
2778 tmp = *s++;
2779
bf4acbe4 2780 while (s < PL_bufend && SPACE_OR_TAB(*s))
748a9306
LW
2781 s++;
2782
2783 if (strnEQ(s,"=>",2)) {
3280af22 2784 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
4e553d73 2785 DEBUG_T( { PerlIO_printf(Perl_debug_log,
607df283 2786 "### Saw unary minus before =>, forcing word '%s'\n", s);
5f80b19c 2787 } );
748a9306
LW
2788 OPERATOR('-'); /* unary minus */
2789 }
3280af22 2790 PL_last_uni = PL_oldbufptr;
748a9306 2791 switch (tmp) {
e5edeb50
JH
2792 case 'r': ftst = OP_FTEREAD; break;
2793 case 'w': ftst = OP_FTEWRITE; break;
2794 case 'x': ftst = OP_FTEEXEC; break;
2795 case 'o': ftst = OP_FTEOWNED; break;
2796 case 'R': ftst = OP_FTRREAD; break;
2797 case 'W': ftst = OP_FTRWRITE; break;
2798 case 'X': ftst = OP_FTREXEC; break;
2799 case 'O': ftst = OP_FTROWNED; break;
2800 case 'e': ftst = OP_FTIS; break;
2801 case 'z': ftst = OP_FTZERO; break;
2802 case 's': ftst = OP_FTSIZE; break;
2803 case 'f': ftst = OP_FTFILE; break;
2804 case 'd': ftst = OP_FTDIR; break;
2805 case 'l': ftst = OP_FTLINK; break;
2806 case 'p': ftst = OP_FTPIPE; break;
2807 case 'S': ftst = OP_FTSOCK; break;
2808 case 'u': ftst = OP_FTSUID; break;
2809 case 'g': ftst = OP_FTSGID; break;
2810 case 'k': ftst = OP_FTSVTX; break;
2811 case 'b': ftst = OP_FTBLK; break;
2812 case 'c': ftst = OP_FTCHR; break;
2813 case 't': ftst = OP_FTTTY; break;
2814 case 'T': ftst = OP_FTTEXT; break;
2815 case 'B': ftst = OP_FTBINARY; break;
2816 case 'M': case 'A': case 'C':
2817 gv_fetchpv("\024",TRUE, SVt_PV);
2818 switch (tmp) {
2819 case 'M': ftst = OP_FTMTIME; break;
2820 case 'A': ftst = OP_FTATIME; break;
2821 case 'C': ftst = OP_FTCTIME; break;
2822 default: break;
2823 }
2824 break;
378cc40b 2825 default:
378cc40b
LW
2826 break;
2827 }
e5edeb50
JH
2828 if (ftst) {
2829 PL_last_lop_op = ftst;
4e553d73 2830 DEBUG_T( { PerlIO_printf(Perl_debug_log,
0844c848 2831 "### Saw file test %c\n", (int)ftst);
5f80b19c 2832 } );
e5edeb50
JH
2833 FTST(ftst);
2834 }
2835 else {
2836 /* Assume it was a minus followed by a one-letter named
2837 * subroutine call (or a -bareword), then. */
95c31fe3 2838 DEBUG_T( { PerlIO_printf(Perl_debug_log,
0844c848
RB
2839 "### %c looked like a file test but was not\n",
2840 (int)ftst);
5f80b19c 2841 } );
e5edeb50
JH
2842 s -= 2;
2843 }
378cc40b 2844 }
a687059c
LW
2845 tmp = *s++;
2846 if (*s == tmp) {
2847 s++;
3280af22 2848 if (PL_expect == XOPERATOR)
79072805
LW
2849 TERM(POSTDEC);
2850 else
2851 OPERATOR(PREDEC);
2852 }
2853 else if (*s == '>') {
2854 s++;
2855 s = skipspace(s);
7e2040f0 2856 if (isIDFIRST_lazy_if(s,UTF)) {
a0d0e21e 2857 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
463ee0b2 2858 TOKEN(ARROW);
79072805 2859 }
748a9306
LW
2860 else if (*s == '$')
2861 OPERATOR(ARROW);
463ee0b2 2862 else
748a9306 2863 TERM(ARROW);
a687059c 2864 }
3280af22 2865 if (PL_expect == XOPERATOR)
79072805
LW
2866 Aop(OP_SUBTRACT);
2867 else {
3280af22 2868 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2f3197b3 2869 check_uni();
79072805 2870 OPERATOR('-'); /* unary minus */
2f3197b3 2871 }
79072805 2872
378cc40b 2873 case '+':
a687059c
LW
2874 tmp = *s++;
2875 if (*s == tmp) {
378cc40b 2876 s++;
3280af22 2877 if (PL_expect == XOPERATOR)
79072805
LW
2878 TERM(POSTINC);
2879 else
2880 OPERATOR(PREINC);
378cc40b 2881 }
3280af22 2882 if (PL_expect == XOPERATOR)
79072805
LW
2883 Aop(OP_ADD);
2884 else {
3280af22 2885 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2f3197b3 2886 check_uni();
a687059c 2887 OPERATOR('+');
2f3197b3 2888 }
a687059c 2889
378cc40b 2890 case '*':
3280af22
NIS
2891 if (PL_expect != XOPERATOR) {
2892 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2893 PL_expect = XOPERATOR;
2894 force_ident(PL_tokenbuf, '*');
2895 if (!*PL_tokenbuf)
a0d0e21e 2896 PREREF('*');
79072805 2897 TERM('*');
a687059c 2898 }
79072805
LW
2899 s++;
2900 if (*s == '*') {
a687059c 2901 s++;
79072805 2902 PWop(OP_POW);
a687059c 2903 }
79072805
LW
2904 Mop(OP_MULTIPLY);
2905
378cc40b 2906 case '%':
3280af22 2907 if (PL_expect == XOPERATOR) {
bbce6d69
PP
2908 ++s;
2909 Mop(OP_MODULO);
a687059c 2910 }
3280af22
NIS
2911 PL_tokenbuf[0] = '%';
2912 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2913 if (!PL_tokenbuf[1]) {
2914 if (s == PL_bufend)
bbce6d69
PP
2915 yyerror("Final % should be \\% or %name");
2916 PREREF('%');
a687059c 2917 }
3280af22 2918 PL_pending_ident = '%';
bbce6d69 2919 TERM('%');
a687059c 2920
378cc40b 2921 case '^':
79072805 2922 s++;
a0d0e21e 2923 BOop(OP_BIT_XOR);
79072805 2924 case '[':
3280af22 2925 PL_lex_brackets++;
79072805 2926 /* FALL THROUGH */
378cc40b 2927 case '~':
378cc40b 2928 case ',':
378cc40b
LW
2929 tmp = *s++;
2930 OPERATOR(tmp);
a0d0e21e
LW
2931 case ':':
2932 if (s[1] == ':') {
2933 len = 0;
2934 goto just_a_word;
2935 }
2936 s++;
09bef843
SB
2937 switch (PL_expect) {
2938 OP *attrs;
2939 case XOPERATOR:
2940 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2941 break;
2942 PL_bufptr = s; /* update in case we back off */
2943 goto grabattrs;
2944 case XATTRBLOCK:
2945 PL_expect = XBLOCK;
2946 goto grabattrs;
2947 case XATTRTERM:
2948 PL_expect = XTERMBLOCK;
2949 grabattrs:
2950 s = skipspace(s);
2951 attrs = Nullop;
7e2040f0 2952 while (isIDFIRST_lazy_if(s,UTF)) {
09bef843 2953 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
f9829d6b
GS
2954 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2955 if (tmp < 0) tmp = -tmp;
2956 switch (tmp) {
2957 case KEY_or:
2958 case KEY_and:
2959 case KEY_for:
2960 case KEY_unless:
2961 case KEY_if:
2962 case KEY_while:
2963 case KEY_until:
2964 goto got_attrs;
2965 default:
2966 break;
2967 }
2968 }
09bef843
SB
2969 if (*d == '(') {
2970 d = scan_str(d,TRUE,TRUE);
2971 if (!d) {
09bef843
SB
2972 /* MUST advance bufptr here to avoid bogus
2973 "at end of line" context messages from yyerror().
2974 */
2975 PL_bufptr = s + len;
2976 yyerror("Unterminated attribute parameter in attribute list");
2977 if (attrs)
2978 op_free(attrs);
2979 return 0; /* EOF indicator */
2980 }
2981 }
2982 if (PL_lex_stuff) {
2983 SV *sv = newSVpvn(s, len);
2984 sv_catsv(sv, PL_lex_stuff);
2985 attrs = append_elem(OP_LIST, attrs,
2986 newSVOP(OP_CONST, 0, sv));
2987 SvREFCNT_dec(PL_lex_stuff);
2988 PL_lex_stuff = Nullsv;
2989 }
2990 else {
78f9721b
SM
2991 if (!PL_in_my && len == 6 && strnEQ(s, "lvalue", len))
2992 CvLVALUE_on(PL_compcv);
2993 else if (!PL_in_my && len == 6 && strnEQ(s, "locked", len))
2994 CvLOCKED_on(PL_compcv);
2995 else if (!PL_in_my && len == 6 && strnEQ(s, "method", len))
2996 CvMETHOD_on(PL_compcv);
87ecf892 2997#ifdef USE_ITHREADS
c8a3bf85 2998 else if (PL_in_my == KEY_our && len == 6 && strnEQ(s, "unique", len))
7fb37951 2999 GvUNIQUE_on(cGVOPx_gv(yylval.opval));
87ecf892 3000#endif
78f9721b
SM
3001 /* After we've set the flags, it could be argued that
3002 we don't need to do the attributes.pm-based setting
3003 process, and shouldn't bother appending recognized
3004 flags. To experiment with that, uncomment the
3005 following "else": */
0256094b 3006 else
78f9721b
SM
3007 attrs = append_elem(OP_LIST, attrs,
3008 newSVOP(OP_CONST, 0,
3009 newSVpvn(s, len)));
09bef843
SB
3010 }
3011 s = skipspace(d);
0120eecf 3012 if (*s == ':' && s[1] != ':')
09bef843 3013 s = skipspace(s+1);
0120eecf
GS
3014 else if (s == d)
3015 break; /* require real whitespace or :'s */
09bef843 3016 }
f9829d6b
GS
3017 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
3018 if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) {
09bef843
SB
3019 char q = ((*s == '\'') ? '"' : '\'');
3020 /* If here for an expression, and parsed no attrs, back off. */
3021 if (tmp == '=' && !attrs) {
3022 s = PL_bufptr;
3023 break;
3024 }
3025 /* MUST advance bufptr here to avoid bogus "at end of line"
3026 context messages from yyerror().
3027 */
3028 PL_bufptr = s;
3029 if (!*s)
3030 yyerror("Unterminated attribute list");
3031 else
3032 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
3033 q, *s, q));
3034 if (attrs)
3035 op_free(attrs);
3036 OPERATOR(':');
3037 }
f9829d6b 3038 got_attrs:
09bef843
SB
3039 if (attrs) {
3040 PL_nextval[PL_nexttoke].opval = attrs;
3041 force_next(THING);
3042 }
3043 TOKEN(COLONATTR);
3044 }
a0d0e21e 3045 OPERATOR(':');
8990e307
LW
3046 case '(':
3047 s++;
3280af22
NIS
3048 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
3049 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
a0d0e21e 3050 else
3280af22 3051 PL_expect = XTERM;
a0d0e21e 3052 TOKEN('(');
378cc40b 3053 case ';':
f4dd75d9 3054 CLINE;
378cc40b
LW
3055 tmp = *s++;
3056 OPERATOR(tmp);
3057 case ')':
378cc40b 3058 tmp = *s++;
16d20bd9
AD
3059 s = skipspace(s);
3060 if (*s == '{')
3061 PREBLOCK(tmp);
378cc40b 3062 TERM(tmp);
79072805
LW
3063 case ']':
3064 s++;
3280af22 3065 if (PL_lex_brackets <= 0)
d98d5fff 3066 yyerror("Unmatched right square bracket");
463ee0b2 3067 else
3280af22
NIS
3068 --PL_lex_brackets;
3069 if (PL_lex_state == LEX_INTERPNORMAL) {
3070 if (PL_lex_brackets == 0) {
a0d0e21e 3071 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
3280af22 3072 PL_lex_state = LEX_INTERPEND;
79072805
LW
3073 }
3074 }
4633a7c4 3075 TERM(']');
79072805
LW
3076 case '{':
3077 leftbracket:
79072805 3078 s++;
3280af22
NIS
3079 if (PL_lex_brackets > 100) {
3080 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
3081 if (newlb != PL_lex_brackstack) {
8990e307 3082 SAVEFREEPV(newlb);
3280af22 3083 PL_lex_brackstack = newlb;
8990e307
LW
3084 }
3085 }
3280af22 3086 switch (PL_expect) {
a0d0e21e 3087 case XTERM:
3280af22 3088 if (PL_lex_formbrack) {
a0d0e21e
LW
3089 s--;
3090 PRETERMBLOCK(DO);
3091 }
3280af22
NIS
3092 if (PL_oldoldbufptr == PL_last_lop)
3093 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
a0d0e21e 3094 else
3280af22 3095 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
79072805 3096 OPERATOR(HASHBRACK);
a0d0e21e 3097 case XOPERATOR:
bf4acbe4 3098 while (s < PL_bufend && SPACE_OR_TAB(*s))
748a9306 3099 s++;
44a8e56a 3100 d = s;
3280af22
NIS
3101 PL_tokenbuf[0] = '\0';
3102 if (d < PL_bufend && *d == '-') {
3103 PL_tokenbuf[0] = '-';
44a8e56a 3104 d++;
bf4acbe4 3105 while (d < PL_bufend && SPACE_OR_TAB(*d))
44a8e56a
PP
3106 d++;
3107 }
7e2040f0 3108 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3280af22 3109 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
8903cb82 3110 FALSE, &len);
bf4acbe4 3111 while (d < PL_bufend && SPACE_OR_TAB(*d))
748a9306
LW
3112 d++;
3113 if (*d == '}') {
3280af22 3114 char minus = (PL_tokenbuf[0] == '-');
44a8e56a
PP
3115 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3116 if (minus)
3117 force_next('-');
748a9306
LW
3118 }
3119 }
3120 /* FALL THROUGH */
09bef843 3121 case XATTRBLOCK:
748a9306 3122 case XBLOCK:
3280af22
NIS
3123 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3124 PL_expect = XSTATE;
a0d0e21e 3125 break;
09bef843 3126 case XATTRTERM:
a0d0e21e 3127 case XTERMBLOCK:
3280af22
NIS
3128 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3129 PL_expect = XSTATE;
a0d0e21e
LW
3130 break;
3131 default: {
3132 char *t;
3280af22
NIS
3133 if (PL_oldoldbufptr == PL_last_lop)
3134 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
a0d0e21e 3135 else
3280af22 3136 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
a0d0e21e 3137 s = skipspace(s);
8452ff4b
SB
3138 if (*s == '}') {
3139 if (PL_expect == XREF && PL_lex_state == LEX_INTERPNORMAL) {
3140 PL_expect = XTERM;
3141 /* This hack is to get the ${} in the message. */
3142 PL_bufptr = s+1;
3143 yyerror("syntax error");
3144 break;
3145 }
a0d0e21e 3146 OPERATOR(HASHBRACK);
8452ff4b 3147 }
b8a4b1be
GS
3148 /* This hack serves to disambiguate a pair of curlies
3149 * as being a block or an anon hash. Normally, expectation
3150 * determines that, but in cases where we're not in a
3151 * position to expect anything in particular (like inside
3152 * eval"") we have to resolve the ambiguity. This code
3153 * covers the case where the first term in the curlies is a
3154 * quoted string. Most other cases need to be explicitly
3155 * disambiguated by prepending a `+' before the opening
3156 * curly in order to force resolution as an anon hash.
3157 *
3158 * XXX should probably propagate the outer expectation
3159 * into eval"" to rely less on this hack, but that could
3160 * potentially break current behavior of eval"".
3161 * GSAR 97-07-21
3162 */
3163 t = s;
3164 if (*s == '\'' || *s == '"' || *s == '`') {
3165 /* common case: get past first string, handling escapes */
3280af22 3166 for (t++; t < PL_bufend && *t != *s;)
b8a4b1be
GS
3167 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3168 t++;
3169 t++;
a0d0e21e 3170 }
b8a4b1be 3171 else if (*s == 'q') {
3280af22 3172 if (++t < PL_bufend
b8a4b1be 3173 && (!isALNUM(*t)
3280af22 3174 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
0505442f
GS
3175 && !isALNUM(*t))))
3176 {
b8a4b1be
GS
3177 char *tmps;
3178 char open, close, term;
3179 I32 brackets = 1;
3180
3280af22 3181 while (t < PL_bufend && isSPACE(*t))
b8a4b1be
GS
3182 t++;
3183 term = *t;
3184 open = term;
3185 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3186 term = tmps[5];
3187 close = term;
3188 if (open == close)
3280af22
NIS
3189 for (t++; t < PL_bufend; t++) {
3190 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
b8a4b1be 3191 t++;
6d07e5e9 3192 else if (*t == open)
b8a4b1be
GS
3193 break;
3194 }
3195 else
3280af22
NIS
3196 for (t++; t < PL_bufend; t++) {
3197 if (*t == '\\' && t+1 < PL_bufend)
b8a4b1be 3198 t++;
6d07e5e9 3199 else if (*t == close && --brackets <= 0)
b8a4b1be
GS
3200 break;
3201 else if (*t == open)
3202 brackets++;
3203 }
3204 }
3205 t++;
a0d0e21e 3206 }
7e2040f0 3207 else if (isALNUM_lazy_if(t,UTF)) {
0505442f 3208 t += UTF8SKIP(t);
7e2040f0 3209 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
0505442f 3210 t += UTF8SKIP(t);
a0d0e21e 3211 }
3280af22 3212 while (t < PL_bufend && isSPACE(*t))
a0d0e21e 3213 t++;
b8a4b1be
GS
3214 /* if comma follows first term, call it an anon hash */
3215 /* XXX it could be a comma expression with loop modifiers */
3280af22 3216 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
b8a4b1be 3217 || (*t == '=' && t[1] == '>')))
a0d0e21e 3218 OPERATOR(HASHBRACK);
3280af22 3219 if (PL_expect == XREF)
4e4e412b 3220 PL_expect = XTERM;
a0d0e21e 3221 else {
3280af22
NIS
3222 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3223 PL_expect = XSTATE;
a0d0e21e 3224 }
8990e307 3225 }
a0d0e21e 3226 break;
463ee0b2 3227 }
57843af0 3228 yylval.ival = CopLINE(PL_curcop);
79072805 3229 if (isSPACE(*s) || *s == '#')
3280af22 3230 PL_copline = NOLINE; /* invalidate current command line number */
79072805 3231 TOKEN('{');
378cc40b 3232 case '}':
79072805
LW
3233 rightbracket:
3234 s++;
3280af22 3235 if (PL_lex_brackets <= 0)
d98d5fff 3236 yyerror("Unmatched right curly bracket");
463ee0b2 3237 else
3280af22 3238 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
c2e66d9e 3239 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3280af22
NIS
3240 PL_lex_formbrack = 0;
3241 if (PL_lex_state == LEX_INTERPNORMAL) {
3242 if (PL_lex_brackets == 0) {
9059aa12
LW
3243 if (PL_expect & XFAKEBRACK) {
3244 PL_expect &= XENUMMASK;
3280af22
NIS
3245 PL_lex_state = LEX_INTERPEND;
3246 PL_bufptr = s;
cea2e8a9 3247 return yylex(); /* ignore fake brackets */
79072805 3248 }
fa83b5b6 3249 if (*s == '-' && s[1] == '>')
3280af22 3250 PL_lex_state = LEX_INTERPENDMAYBE;
fa83b5b6 3251 else if (*s != '[' && *s != '{')
3280af22 3252 PL_lex_state = LEX_INTERPEND;
79072805
LW
3253 }
3254 }
9059aa12
LW
3255 if (PL_expect & XFAKEBRACK) {
3256 PL_expect &= XENUMMASK;
3280af22 3257 PL_bufptr = s;
cea2e8a9 3258 return yylex(); /* ignore fake brackets */
748a9306 3259 }
79072805
LW
3260 force_next('}');
3261 TOKEN(';');
378cc40b
LW
3262 case '&':
3263 s++;
3264 tmp = *s++;
3265 if (tmp == '&')
a0d0e21e 3266 AOPERATOR(ANDAND);
378cc40b 3267 s--;
3280af22 3268 if (PL_expect == XOPERATOR) {
7e2040f0
GS
3269 if (ckWARN(WARN_SEMICOLON)
3270 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3271 {
57843af0 3272 CopLINE_dec(PL_curcop);
cea2e8a9 3273 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
57843af0 3274 CopLINE_inc(PL_curcop);
463ee0b2 3275 }
79072805 3276 BAop(OP_BIT_AND);
463ee0b2 3277 }
79072805 3278
3280af22
NIS
3279 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3280 if (*PL_tokenbuf) {
3281 PL_expect = XOPERATOR;
3282 force_ident(PL_tokenbuf, '&');
463ee0b2 3283 }
79072805
LW
3284 else
3285 PREREF('&');
c07a80fd 3286 yylval.ival = (OPpENTERSUB_AMPER<<8);
79072805
LW
3287 TERM('&');
3288
378cc40b
LW
3289 case '|':
3290 s++;
3291 tmp = *s++;
3292 if (tmp == '|')
a0d0e21e 3293 AOPERATOR(OROR);
378cc40b 3294 s--;
79072805 3295 BOop(OP_BIT_OR);
378cc40b
LW
3296 case '=':
3297 s++;
3298 tmp = *s++;
3299 if (tmp == '=')
79072805
LW
3300 Eop(OP_EQ);
3301 if (tmp == '>')
3302 OPERATOR(',');
378cc40b 3303 if (tmp == '~')
79072805 3304 PMop(OP_MATCH);
599cee73 3305 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
cea2e8a9 3306 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
378cc40b 3307 s--;
3280af22
NIS
3308 if (PL_expect == XSTATE && isALPHA(tmp) &&
3309 (s == PL_linestart+1 || s[-2] == '\n') )
748a9306 3310 {
3280af22
NIS
3311 if (PL_in_eval && !PL_rsfp) {
3312 d = PL_bufend;
a5f75d66
AD
3313 while (s < d) {
3314 if (*s++ == '\n') {
3315 incline(s);
3316 if (strnEQ(s,"=cut",4)) {
3317 s = strchr(s,'\n');
3318 if (s)
3319 s++;
3320 else
3321 s = d;
3322 incline(s);
3323 goto retry;
3324 }
3325 }
3326 }
3327 goto retry;
3328 }
3280af22
NIS
3329 s = PL_bufend;
3330 PL_doextract = TRUE;
a0d0e21e
LW
3331 goto retry;
3332 }
3280af22 3333 if (PL_lex_brackets < PL_lex_formbrack) {
a0d0e21e 3334 char *t;
51882d45 3335#ifdef PERL_STRICT_CR
bf4acbe4 3336 for (t = s; SPACE_OR_TAB(*t); t++) ;
51882d45 3337#else
bf4acbe4 3338 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
51882d45 3339#endif
a0d0e21e
LW
3340 if (*t == '\n' || *t == '#') {
3341 s--;
3280af22 3342 PL_expect = XBLOCK;
a0d0e21e
LW
3343 goto leftbracket;
3344 }
79072805 3345 }
a0d0e21e
LW
3346 yylval.ival = 0;
3347 OPERATOR(ASSIGNOP);
378cc40b
LW
3348 case '!':
3349 s++;
3350 tmp = *s++;
3351 if (tmp == '=')
79072805 3352 Eop(OP_NE);
378cc40b 3353 if (tmp == '~')
79072805 3354 PMop(OP_NOT);
378cc40b
LW
3355 s--;
3356 OPERATOR('!');
3357 case '<':
3280af22 3358 if (PL_expect != XOPERATOR) {
93a17b20 3359 if (s[1] != '<' && !strchr(s,'>'))
2f3197b3 3360 check_uni();
79072805
LW
3361 if (s[1] == '<')
3362 s = scan_heredoc(s);
3363 else
3364 s = scan_inputsymbol(s);
3365 TERM(sublex_start());
378cc40b
LW
3366 }
3367 s++;
3368 tmp = *s++;
3369 if (tmp == '<')
79072805 3370 SHop(OP_LEFT_SHIFT);
395c3793
LW
3371 if (tmp == '=') {
3372 tmp = *s++;
3373 if (tmp == '>')
79072805 3374 Eop(OP_NCMP);
395c3793 3375 s--;
79072805 3376 Rop(OP_LE);
395c3793 3377 }
378cc40b 3378 s--;
79072805 3379 Rop(OP_LT);
378cc40b
LW
3380 case '>':
3381 s++;
3382 tmp = *s++;
3383 if (tmp == '>')
79072805 3384 SHop(OP_RIGHT_SHIFT);
378cc40b 3385 if (tmp == '=')
79072805 3386 Rop(OP_GE);
378cc40b 3387 s--;
79072805 3388 Rop(OP_GT);
378cc40b
LW
3389
3390 case '$':
bbce6d69
PP
3391 CLINE;
3392
3280af22
NIS
3393 if (PL_expect == XOPERATOR) {
3394 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3395 PL_expect = XTERM;
a0d0e21e 3396 depcom();
bbce6d69 3397 return ','; /* grandfather non-comma-format format */
a0d0e21e 3398 }
8990e307 3399 }
a0d0e21e 3400
7e2040f0 3401 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3280af22 3402 PL_tokenbuf[0] = '@';
376b8730
SM
3403 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3404 sizeof PL_tokenbuf - 1, FALSE);
3405 if (PL_expect == XOPERATOR)
3406 no_op("Array length", s);
3280af22 3407 if (!PL_tokenbuf[1])
a0d0e21e 3408 PREREF(DOLSHARP);
3280af22
NIS
3409 PL_expect = XOPERATOR;
3410 PL_pending_ident = '#';
463ee0b2 3411 TOKEN(DOLSHARP);
79072805 3412 }
bbce6d69 3413
3280af22 3414 PL_tokenbuf[0] = '$';
376b8730
SM
3415 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3416 sizeof PL_tokenbuf - 1, FALSE);
3417 if (PL_expect == XOPERATOR)
3418 no_op("Scalar", s);
3280af22
NIS
3419 if (!PL_tokenbuf[1]) {
3420 if (s == PL_bufend)
bbce6d69
PP
3421 yyerror("Final $ should be \\$ or $name");
3422 PREREF('$');
8990e307 3423 }
a0d0e21e 3424
bbce6d69 3425 /* This kludge not intended to be bulletproof. */
3280af22 3426 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
bbce6d69 3427 yylval.opval = newSVOP(OP_CONST, 0,
b448e4fe 3428 newSViv(PL_compiling.cop_arybase));
bbce6d69
PP
3429 yylval.opval->op_private = OPpCONST_ARYBASE;
3430 TERM(THING);
3431 }
3432
ff68c719 3433 d = s;
69d2bceb 3434 tmp = (I32)*s;
3280af22 3435 if (PL_lex_state == LEX_NORMAL)
ff68c719
PP
3436 s = skipspace(s);
3437
3280af22 3438 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
bbce6d69
PP
3439 char *t;
3440 if (*s == '[') {
3280af22 3441 PL_tokenbuf[0] = '@';
599cee73 3442 if (ckWARN(WARN_SYNTAX)) {
bbce6d69 3443 for(t = s + 1;
7e2040f0 3444 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
bbce6d69 3445 t++) ;
a0d0e21e 3446 if (*t++ == ',') {
3280af22
NIS
3447 PL_bufptr = skipspace(PL_bufptr);
3448 while (t < PL_bufend && *t != ']')
bbce6d69 3449 t++;
cea2e8a9 3450 Perl_warner(aTHX_ WARN_SYNTAX,
599cee73
PM
3451 "Multidimensional syntax %.*s not supported",
3452 (t - PL_bufptr) + 1, PL_bufptr);
a0d0e21e
LW
3453 }
3454 }
bbce6d69
PP
3455 }
3456 else if (*s == '{') {
3280af22 3457 PL_tokenbuf[0] = '%';
599cee73 3458 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
bbce6d69
PP
3459 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3460 {
3280af22 3461 char tmpbuf[sizeof PL_tokenbuf];
a0d0e21e
LW
3462 STRLEN len;
3463 for (t++; isSPACE(*t); t++) ;
7e2040f0 3464 if (isIDFIRST_lazy_if(t,UTF)) {
8903cb82 3465 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
59a6d928 3466 for (; isSPACE(*t); t++) ;
864dbfa3 3467 if (*t == ';' && get_cv(tmpbuf, FALSE))
cea2e8a9 3468 Perl_warner(aTHX_ WARN_SYNTAX,
599cee73 3469 "You need to quote \"%s\"", tmpbuf);
748a9306 3470 }
93a17b20
LW
3471 }
3472 }
2f3197b3 3473 }
bbce6d69 3474
3280af22 3475 PL_expect = XOPERATOR;
69d2bceb 3476 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3280af22
NIS
3477 bool islop = (PL_last_lop == PL_oldoldbufptr);
3478 if (!islop || PL_last_lop_op == OP_GREPSTART)
3479 PL_expect = XOPERATOR;
bbce6d69 3480 else if (strchr("$@\"'`q", *s))
3280af22 3481 PL_expect = XTERM; /* e.g. print $fh "foo" */
7e2040f0 3482 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3280af22 3483 PL_expect = XTERM; /* e.g. print $fh &sub */
7e2040f0 3484 else if (isIDFIRST_lazy_if(s,UTF)) {
3280af22 3485 char tmpbuf[sizeof PL_tokenbuf];
8903cb82 3486 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
155aba94 3487 if ((tmp = keyword(tmpbuf, len))) {
84902520
TB
3488 /* binary operators exclude handle interpretations */
3489 switch (tmp) {
3490 case -KEY_x:
3491 case -KEY_eq:
3492 case -KEY_ne:
3493 case -KEY_gt:
3494 case -KEY_lt:
3495 case -KEY_ge:
3496 case -KEY_le:
3497 case -KEY_cmp:
3498 break;
3499 default:
3280af22 3500 PL_expect = XTERM; /* e.g. print $fh length() */
84902520
TB
3501 break;
3502 }
3503 }
68dc0745
PP
3504 else {
3505 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3506 if (gv && GvCVu(gv))
3280af22 3507 PL_expect = XTERM; /* e.g. print $fh subr() */
93a17b20 3508 }
93a17b20 3509 }
bbce6d69 3510 else if (isDIGIT(*s))
3280af22 3511 PL_expect = XTERM; /* e.g. print $fh 3 */