This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Re: [ID 20010920.007] q and qq does not work after do
[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);
57843af0 637 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
8990e307 638 }
463ee0b2 639 }
a687059c 640}
378cc40b 641
ffb4593c
NT
642/*
643 * S_check_uni
644 * Check the unary operators to ensure there's no ambiguity in how they're
645 * used. An ambiguous piece of code would be:
646 * rand + 5
647 * This doesn't mean rand() + 5. Because rand() is a unary operator,
648 * the +5 is its argument.
649 */
650
76e3520e 651STATIC void
cea2e8a9 652S_check_uni(pTHX)
ba106d47 653{
2f3197b3 654 char *s;
a0d0e21e 655 char *t;
2f3197b3 656
3280af22 657 if (PL_oldoldbufptr != PL_last_uni)
2f3197b3 658 return;
3280af22
NIS
659 while (isSPACE(*PL_last_uni))
660 PL_last_uni++;
7e2040f0 661 for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ;
3280af22 662 if ((t = strchr(s, '(')) && t < PL_bufptr)
a0d0e21e 663 return;
0453d815 664 if (ckWARN_d(WARN_AMBIGUOUS)){
f248d071 665 char ch = *s;
0453d815 666 *s = '\0';
4e553d73
NIS
667 Perl_warner(aTHX_ WARN_AMBIGUOUS,
668 "Warning: Use of \"%s\" without parens is ambiguous",
0453d815
PM
669 PL_last_uni);
670 *s = ch;
671 }
2f3197b3
LW
672}
673
ffb4593c
NT
674/* workaround to replace the UNI() macro with a function. Only the
675 * hints/uts.sh file mentions this. Other comments elsewhere in the
676 * source indicate Microport Unix might need it too.
677 */
678
ffed7fef
LW
679#ifdef CRIPPLED_CC
680
681#undef UNI
ffed7fef 682#define UNI(f) return uni(f,s)
ffed7fef 683
76e3520e 684STATIC int
cea2e8a9 685S_uni(pTHX_ I32 f, char *s)
ffed7fef
LW
686{
687 yylval.ival = f;
3280af22
NIS
688 PL_expect = XTERM;
689 PL_bufptr = s;
8f872242
NIS
690 PL_last_uni = PL_oldbufptr;
691 PL_last_lop_op = f;
ffed7fef
LW
692 if (*s == '(')
693 return FUNC1;
694 s = skipspace(s);
695 if (*s == '(')
696 return FUNC1;
697 else
698 return UNIOP;
699}
700
a0d0e21e
LW
701#endif /* CRIPPLED_CC */
702
ffb4593c
NT
703/*
704 * LOP : macro to build a list operator. Its behaviour has been replaced
705 * with a subroutine, S_lop() for which LOP is just another name.
706 */
707
a0d0e21e
LW
708#define LOP(f,x) return lop(f,x,s)
709
ffb4593c
NT
710/*
711 * S_lop
712 * Build a list operator (or something that might be one). The rules:
713 * - if we have a next token, then it's a list operator [why?]
714 * - if the next thing is an opening paren, then it's a function
715 * - else it's a list operator
716 */
717
76e3520e 718STATIC I32
a0be28da 719S_lop(pTHX_ I32 f, int x, char *s)
ffed7fef 720{
79072805 721 yylval.ival = f;
35c8bce7 722 CLINE;
075953c3 723 REPORT("lop", f)
3280af22
NIS
724 PL_expect = x;
725 PL_bufptr = s;
726 PL_last_lop = PL_oldbufptr;
727 PL_last_lop_op = f;
728 if (PL_nexttoke)
a0d0e21e 729 return LSTOP;
79072805
LW
730 if (*s == '(')
731 return FUNC;
732 s = skipspace(s);
733 if (*s == '(')
734 return FUNC;
735 else
736 return LSTOP;
737}
738
ffb4593c
NT
739/*
740 * S_force_next
9cbb5ea2 741 * When the lexer realizes it knows the next token (for instance,
ffb4593c 742 * it is reordering tokens for the parser) then it can call S_force_next
9cbb5ea2
GS
743 * to know what token to return the next time the lexer is called. Caller
744 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
745 * handles the token correctly.
ffb4593c
NT
746 */
747
4e553d73 748STATIC void
cea2e8a9 749S_force_next(pTHX_ I32 type)
79072805 750{
3280af22
NIS
751 PL_nexttype[PL_nexttoke] = type;
752 PL_nexttoke++;
753 if (PL_lex_state != LEX_KNOWNEXT) {
754 PL_lex_defer = PL_lex_state;
755 PL_lex_expect = PL_expect;
756 PL_lex_state = LEX_KNOWNEXT;
79072805
LW
757 }
758}
759
ffb4593c
NT
760/*
761 * S_force_word
762 * When the lexer knows the next thing is a word (for instance, it has
763 * just seen -> and it knows that the next char is a word char, then
764 * it calls S_force_word to stick the next word into the PL_next lookahead.
765 *
766 * Arguments:
b1b65b59 767 * char *start : buffer position (must be within PL_linestr)
ffb4593c
NT
768 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
769 * int check_keyword : if true, Perl checks to make sure the word isn't
770 * a keyword (do this if the word is a label, e.g. goto FOO)
771 * int allow_pack : if true, : characters will also be allowed (require,
772 * use, etc. do this)
9cbb5ea2 773 * int allow_initial_tick : used by the "sub" lexer only.
ffb4593c
NT
774 */
775
76e3520e 776STATIC char *
cea2e8a9 777S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
79072805 778{
463ee0b2
LW
779 register char *s;
780 STRLEN len;
4e553d73 781
463ee0b2
LW
782 start = skipspace(start);
783 s = start;
7e2040f0 784 if (isIDFIRST_lazy_if(s,UTF) ||
a0d0e21e 785 (allow_pack && *s == ':') ||
15f0808c 786 (allow_initial_tick && *s == '\'') )
a0d0e21e 787 {
3280af22
NIS
788 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
789 if (check_keyword && keyword(PL_tokenbuf, len))
463ee0b2
LW
790 return start;
791 if (token == METHOD) {
792 s = skipspace(s);
793 if (*s == '(')
3280af22 794 PL_expect = XTERM;
463ee0b2 795 else {
3280af22 796 PL_expect = XOPERATOR;
463ee0b2 797 }
79072805 798 }
3280af22
NIS
799 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0));
800 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
79072805
LW
801 force_next(token);
802 }
803 return s;
804}
805
ffb4593c
NT
806/*
807 * S_force_ident
9cbb5ea2 808 * Called when the lexer wants $foo *foo &foo etc, but the program
ffb4593c
NT
809 * text only contains the "foo" portion. The first argument is a pointer
810 * to the "foo", and the second argument is the type symbol to prefix.
811 * Forces the next token to be a "WORD".
9cbb5ea2 812 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
ffb4593c
NT
813 */
814
76e3520e 815STATIC void
cea2e8a9 816S_force_ident(pTHX_ register char *s, int kind)
79072805
LW
817{
818 if (s && *s) {
11343788 819 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
3280af22 820 PL_nextval[PL_nexttoke].opval = o;
79072805 821 force_next(WORD);
748a9306 822 if (kind) {
11343788 823 o->op_private = OPpCONST_ENTERED;
55497cff
PP
824 /* XXX see note in pp_entereval() for why we forgo typo
825 warnings if the symbol must be introduced in an eval.
826 GSAR 96-10-12 */
3280af22 827 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
a0d0e21e
LW
828 kind == '$' ? SVt_PV :
829 kind == '@' ? SVt_PVAV :
830 kind == '%' ? SVt_PVHV :
831 SVt_PVGV
832 );
748a9306 833 }
79072805
LW
834 }
835}
836
1571675a
GS
837NV
838Perl_str_to_version(pTHX_ SV *sv)
839{
840 NV retval = 0.0;
841 NV nshift = 1.0;
842 STRLEN len;
843 char *start = SvPVx(sv,len);
3aa33fe5 844 bool utf = SvUTF8(sv) ? TRUE : FALSE;
1571675a
GS
845 char *end = start + len;
846 while (start < end) {
ba210ebe 847 STRLEN skip;
1571675a
GS
848 UV n;
849 if (utf)
9041c2e3 850 n = utf8n_to_uvchr((U8*)start, len, &skip, 0);
1571675a
GS
851 else {
852 n = *(U8*)start;
853 skip = 1;
854 }
855 retval += ((NV)n)/nshift;
856 start += skip;
857 nshift *= 1000;
858 }
859 return retval;
860}
861
4e553d73 862/*
ffb4593c
NT
863 * S_force_version
864 * Forces the next token to be a version number.
e759cc13
RGS
865 * If the next token appears to be an invalid version number, (e.g. "v2b"),
866 * and if "guessing" is TRUE, then no new token is created (and the caller
867 * must use an alternative parsing method).
ffb4593c
NT
868 */
869
76e3520e 870STATIC char *
e759cc13 871S_force_version(pTHX_ char *s, int guessing)
89bfa8cd
PP
872{
873 OP *version = Nullop;
44dcb63b 874 char *d;
89bfa8cd
PP
875
876 s = skipspace(s);
877
44dcb63b 878 d = s;
dd629d5b 879 if (*d == 'v')
44dcb63b 880 d++;
44dcb63b 881 if (isDIGIT(*d)) {
e759cc13
RGS
882 while (isDIGIT(*d) || *d == '_' || *d == '.')
883 d++;
9f3d182e 884 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
dd629d5b 885 SV *ver;
b73d6f50 886 s = scan_num(s, &yylval);
89bfa8cd 887 version = yylval.opval;
dd629d5b
GS
888 ver = cSVOPx(version)->op_sv;
889 if (SvPOK(ver) && !SvNIOK(ver)) {
155aba94 890 (void)SvUPGRADE(ver, SVt_PVNV);
1571675a
GS
891 SvNVX(ver) = str_to_version(ver);
892 SvNOK_on(ver); /* hint that it is a version */
44dcb63b 893 }
89bfa8cd 894 }
e759cc13
RGS
895 else if (guessing)
896 return s;
89bfa8cd
PP
897 }
898
899 /* NOTE: The parser sees the package name and the VERSION swapped */
3280af22 900 PL_nextval[PL_nexttoke].opval = version;
4e553d73 901 force_next(WORD);
89bfa8cd 902
e759cc13 903 return s;
89bfa8cd
PP
904}
905
ffb4593c
NT
906/*
907 * S_tokeq
908 * Tokenize a quoted string passed in as an SV. It finds the next
909 * chunk, up to end of string or a backslash. It may make a new
910 * SV containing that chunk (if HINT_NEW_STRING is on). It also
911 * turns \\ into \.
912 */
913
76e3520e 914STATIC SV *
cea2e8a9 915S_tokeq(pTHX_ SV *sv)
79072805
LW
916{
917 register char *s;
918 register char *send;
919 register char *d;
b3ac6de7
IZ
920 STRLEN len = 0;
921 SV *pv = sv;
79072805
LW
922
923 if (!SvLEN(sv))
b3ac6de7 924 goto finish;
79072805 925
a0d0e21e 926 s = SvPV_force(sv, len);
21a311ee 927 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
b3ac6de7 928 goto finish;
463ee0b2 929 send = s + len;
79072805
LW
930 while (s < send && *s != '\\')
931 s++;
932 if (s == send)
b3ac6de7 933 goto finish;
79072805 934 d = s;
be4731d2 935 if ( PL_hints & HINT_NEW_STRING ) {
79cb57f6 936 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
be4731d2
NIS
937 if (SvUTF8(sv))
938 SvUTF8_on(pv);
939 }
79072805
LW
940 while (s < send) {
941 if (*s == '\\') {
a0d0e21e 942 if (s + 1 < send && (s[1] == '\\'))
79072805
LW
943 s++; /* all that, just for this */
944 }
945 *d++ = *s++;
946 }
947 *d = '\0';
463ee0b2 948 SvCUR_set(sv, d - SvPVX(sv));
b3ac6de7 949 finish:
3280af22 950 if ( PL_hints & HINT_NEW_STRING )
b3ac6de7 951 return new_constant(NULL, 0, "q", sv, pv, "q");
79072805
LW
952 return sv;
953}
954
ffb4593c
NT
955/*
956 * Now come three functions related to double-quote context,
957 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
958 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
959 * interact with PL_lex_state, and create fake ( ... ) argument lists
960 * to handle functions and concatenation.
961 * They assume that whoever calls them will be setting up a fake
962 * join call, because each subthing puts a ',' after it. This lets
963 * "lower \luPpEr"
964 * become
965 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
966 *
967 * (I'm not sure whether the spurious commas at the end of lcfirst's
968 * arguments and join's arguments are created or not).
969 */
970
971/*
972 * S_sublex_start
973 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
974 *
975 * Pattern matching will set PL_lex_op to the pattern-matching op to
976 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
977 *
978 * OP_CONST and OP_READLINE are easy--just make the new op and return.
979 *
980 * Everything else becomes a FUNC.
981 *
982 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
983 * had an OP_CONST or OP_READLINE). This just sets us up for a
984 * call to S_sublex_push().
985 */
986
76e3520e 987STATIC I32
cea2e8a9 988S_sublex_start(pTHX)
79072805
LW
989{
990 register I32 op_type = yylval.ival;
79072805
LW
991
992 if (op_type == OP_NULL) {
3280af22
NIS
993 yylval.opval = PL_lex_op;
994 PL_lex_op = Nullop;
79072805
LW
995 return THING;
996 }
997 if (op_type == OP_CONST || op_type == OP_READLINE) {
3280af22 998 SV *sv = tokeq(PL_lex_stuff);
b3ac6de7
IZ
999
1000 if (SvTYPE(sv) == SVt_PVIV) {
1001 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
1002 STRLEN len;
1003 char *p;
1004 SV *nsv;
1005
1006 p = SvPV(sv, len);
79cb57f6 1007 nsv = newSVpvn(p, len);
01ec43d0
GS
1008 if (SvUTF8(sv))
1009 SvUTF8_on(nsv);
b3ac6de7
IZ
1010 SvREFCNT_dec(sv);
1011 sv = nsv;
4e553d73 1012 }
b3ac6de7 1013 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
3280af22 1014 PL_lex_stuff = Nullsv;
79072805
LW
1015 return THING;
1016 }
1017
3280af22
NIS
1018 PL_sublex_info.super_state = PL_lex_state;
1019 PL_sublex_info.sub_inwhat = op_type;
1020 PL_sublex_info.sub_op = PL_lex_op;
1021 PL_lex_state = LEX_INTERPPUSH;
55497cff 1022
3280af22
NIS
1023 PL_expect = XTERM;
1024 if (PL_lex_op) {
1025 yylval.opval = PL_lex_op;
1026 PL_lex_op = Nullop;
55497cff
PP
1027 return PMFUNC;
1028 }
1029 else
1030 return FUNC;
1031}
1032
ffb4593c
NT
1033/*
1034 * S_sublex_push
1035 * Create a new scope to save the lexing state. The scope will be
1036 * ended in S_sublex_done. Returns a '(', starting the function arguments
1037 * to the uc, lc, etc. found before.
1038 * Sets PL_lex_state to LEX_INTERPCONCAT.
1039 */
1040
76e3520e 1041STATIC I32
cea2e8a9 1042S_sublex_push(pTHX)
55497cff 1043{
f46d017c 1044 ENTER;
55497cff 1045
3280af22
NIS
1046 PL_lex_state = PL_sublex_info.super_state;
1047 SAVEI32(PL_lex_dojoin);
1048 SAVEI32(PL_lex_brackets);
3280af22
NIS
1049 SAVEI32(PL_lex_casemods);
1050 SAVEI32(PL_lex_starts);
1051 SAVEI32(PL_lex_state);
7766f137 1052 SAVEVPTR(PL_lex_inpat);
3280af22 1053 SAVEI32(PL_lex_inwhat);
57843af0 1054 SAVECOPLINE(PL_curcop);
3280af22 1055 SAVEPPTR(PL_bufptr);
8452ff4b 1056 SAVEPPTR(PL_bufend);
3280af22
NIS
1057 SAVEPPTR(PL_oldbufptr);
1058 SAVEPPTR(PL_oldoldbufptr);
207e3d1a
JH
1059 SAVEPPTR(PL_last_lop);
1060 SAVEPPTR(PL_last_uni);
3280af22
NIS
1061 SAVEPPTR(PL_linestart);
1062 SAVESPTR(PL_linestr);
1063 SAVEPPTR(PL_lex_brackstack);
1064 SAVEPPTR(PL_lex_casestack);
1065
1066 PL_linestr = PL_lex_stuff;
1067 PL_lex_stuff = Nullsv;
1068
9cbb5ea2
GS
1069 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1070 = SvPVX(PL_linestr);
3280af22 1071 PL_bufend += SvCUR(PL_linestr);
207e3d1a 1072 PL_last_lop = PL_last_uni = Nullch;
3280af22
NIS
1073 SAVEFREESV(PL_linestr);
1074
1075 PL_lex_dojoin = FALSE;
1076 PL_lex_brackets = 0;
3280af22
NIS
1077 New(899, PL_lex_brackstack, 120, char);
1078 New(899, PL_lex_casestack, 12, char);
1079 SAVEFREEPV(PL_lex_brackstack);
1080 SAVEFREEPV(PL_lex_casestack);
1081 PL_lex_casemods = 0;
1082 *PL_lex_casestack = '\0';
1083 PL_lex_starts = 0;
1084 PL_lex_state = LEX_INTERPCONCAT;
57843af0 1085 CopLINE_set(PL_curcop, PL_multi_start);
3280af22
NIS
1086
1087 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1088 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1089 PL_lex_inpat = PL_sublex_info.sub_op;
79072805 1090 else
3280af22 1091 PL_lex_inpat = Nullop;
79072805 1092
55497cff 1093 return '(';
79072805
LW
1094}
1095
ffb4593c
NT
1096/*
1097 * S_sublex_done
1098 * Restores lexer state after a S_sublex_push.
1099 */
1100
76e3520e 1101STATIC I32
cea2e8a9 1102S_sublex_done(pTHX)
79072805 1103{
3280af22 1104 if (!PL_lex_starts++) {
9aa983d2
JH
1105 SV *sv = newSVpvn("",0);
1106 if (SvUTF8(PL_linestr))
1107 SvUTF8_on(sv);
3280af22 1108 PL_expect = XOPERATOR;
9aa983d2 1109 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
79072805
LW
1110 return THING;
1111 }
1112
3280af22
NIS
1113 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1114 PL_lex_state = LEX_INTERPCASEMOD;
cea2e8a9 1115 return yylex();
79072805
LW
1116 }
1117
ffb4593c 1118 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
3280af22
NIS
1119 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1120 PL_linestr = PL_lex_repl;
1121 PL_lex_inpat = 0;
1122 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1123 PL_bufend += SvCUR(PL_linestr);
207e3d1a 1124 PL_last_lop = PL_last_uni = Nullch;
3280af22
NIS
1125 SAVEFREESV(PL_linestr);
1126 PL_lex_dojoin = FALSE;
1127 PL_lex_brackets = 0;
3280af22
NIS
1128 PL_lex_casemods = 0;
1129 *PL_lex_casestack = '\0';
1130 PL_lex_starts = 0;
25da4f38 1131 if (SvEVALED(PL_lex_repl)) {
3280af22
NIS
1132 PL_lex_state = LEX_INTERPNORMAL;
1133 PL_lex_starts++;
e9fa98b2
HS
1134 /* we don't clear PL_lex_repl here, so that we can check later
1135 whether this is an evalled subst; that means we rely on the
1136 logic to ensure sublex_done() is called again only via the
1137 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
79072805 1138 }
e9fa98b2 1139 else {
3280af22 1140 PL_lex_state = LEX_INTERPCONCAT;
e9fa98b2
HS
1141 PL_lex_repl = Nullsv;
1142 }
79072805 1143 return ',';
ffed7fef
LW
1144 }
1145 else {
f46d017c 1146 LEAVE;
3280af22
NIS
1147 PL_bufend = SvPVX(PL_linestr);
1148 PL_bufend += SvCUR(PL_linestr);
1149 PL_expect = XOPERATOR;
09bef843 1150 PL_sublex_info.sub_inwhat = 0;
79072805 1151 return ')';
ffed7fef
LW
1152 }
1153}
1154
02aa26ce
NT
1155/*
1156 scan_const
1157
1158 Extracts a pattern, double-quoted string, or transliteration. This
1159 is terrifying code.
1160
3280af22
NIS
1161 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1162 processing a pattern (PL_lex_inpat is true), a transliteration
02aa26ce
NT
1163 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1164
9b599b2a
GS
1165 Returns a pointer to the character scanned up to. Iff this is
1166 advanced from the start pointer supplied (ie if anything was
1167 successfully parsed), will leave an OP for the substring scanned
1168 in yylval. Caller must intuit reason for not parsing further
1169 by looking at the next characters herself.
1170
02aa26ce
NT
1171 In patterns:
1172 backslashes:
1173 double-quoted style: \r and \n
1174 regexp special ones: \D \s
1175 constants: \x3
1176 backrefs: \1 (deprecated in substitution replacements)
1177 case and quoting: \U \Q \E
1178 stops on @ and $, but not for $ as tail anchor
1179
1180 In transliterations:
1181 characters are VERY literal, except for - not at the start or end
1182 of the string, which indicates a range. scan_const expands the
1183 range to the full set of intermediate characters.
1184
1185 In double-quoted strings:
1186 backslashes:
1187 double-quoted style: \r and \n
1188 constants: \x3
1189 backrefs: \1 (deprecated)
1190 case and quoting: \U \Q \E
1191 stops on @ and $
1192
1193 scan_const does *not* construct ops to handle interpolated strings.
1194 It stops processing as soon as it finds an embedded $ or @ variable
1195 and leaves it to the caller to work out what's going on.
1196
1197 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @:foo.
1198
1199 $ in pattern could be $foo or could be tail anchor. Assumption:
1200 it's a tail anchor if $ is the last thing in the string, or if it's
1201 followed by one of ")| \n\t"
1202
1203 \1 (backreferences) are turned into $1
1204
1205 The structure of the code is
1206 while (there's a character to process) {
1207 handle transliteration ranges
1208 skip regexp comments
1209 skip # initiated comments in //x patterns
1210 check for embedded @foo
1211 check for embedded scalars
1212 if (backslash) {
1213 leave intact backslashes from leave (below)
1214 deprecate \1 in strings and sub replacements
1215 handle string-changing backslashes \l \U \Q \E, etc.
1216 switch (what was escaped) {
1217 handle - in a transliteration (becomes a literal -)
1218 handle \132 octal characters
1219 handle 0x15 hex characters
1220 handle \cV (control V)
1221 handle printf backslashes (\f, \r, \n, etc)
1222 } (end switch)
1223 } (end if backslash)
1224 } (end while character to read)
4e553d73 1225
02aa26ce
NT
1226*/
1227
76e3520e 1228STATIC char *
cea2e8a9 1229S_scan_const(pTHX_ char *start)
79072805 1230{
3280af22 1231 register char *send = PL_bufend; /* end of the constant */
02aa26ce
NT
1232 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1233 register char *s = start; /* start of the constant */
1234 register char *d = SvPVX(sv); /* destination for copies */
1235 bool dorange = FALSE; /* are we in a translit range? */
c2e66d9e 1236 bool didrange = FALSE; /* did we just finish a range? */
2b9d42f0
NIS
1237 I32 has_utf8 = FALSE; /* Output constant is UTF8 */
1238 I32 this_utf8 = UTF; /* The source string is assumed to be UTF8 */
012bcf8d
GS
1239 UV uv;
1240
dff6d3cd 1241 const char *leaveit = /* set of acceptably-backslashed characters */
3280af22 1242 PL_lex_inpat
4a2d328f 1243 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
9b599b2a 1244 : "";
79072805 1245
2b9d42f0
NIS
1246 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1247 /* If we are doing a trans and we know we want UTF8 set expectation */
1248 has_utf8 = PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF);
1249 this_utf8 = PL_sublex_info.sub_op->op_private & (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1250 }
1251
1252
79072805 1253 while (s < send || dorange) {
02aa26ce 1254 /* get transliterations out of the way (they're most literal) */
3280af22 1255 if (PL_lex_inwhat == OP_TRANS) {
02aa26ce 1256 /* expand a range A-Z to the full set of characters. AIE! */
79072805 1257 if (dorange) {
1ba5c669
JH
1258 I32 i; /* current expanded character */
1259 I32 min; /* first character in range */
1260 I32 max; /* last character in range */
02aa26ce 1261
2b9d42f0 1262 if (has_utf8) {
8973db79
JH
1263 char *c = (char*)utf8_hop((U8*)d, -1);
1264 char *e = d++;
1265 while (e-- > c)
1266 *(e + 1) = *e;
25716404 1267 *c = (char)UTF_TO_NATIVE(0xff);
8973db79
JH
1268 /* mark the range as done, and continue */
1269 dorange = FALSE;
1270 didrange = TRUE;
1271 continue;
1272 }
2b9d42f0 1273
02aa26ce 1274 i = d - SvPVX(sv); /* remember current offset */
9cbb5ea2
GS
1275 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1276 d = SvPVX(sv) + i; /* refresh d after realloc */
02aa26ce
NT
1277 d -= 2; /* eat the first char and the - */
1278
8ada0baa
JH
1279 min = (U8)*d; /* first char in range */
1280 max = (U8)d[1]; /* last char in range */
1281
c2e66d9e 1282 if (min > max) {
01ec43d0 1283 Perl_croak(aTHX_
1ba5c669
JH
1284 "Invalid [] range \"%c-%c\" in transliteration operator",
1285 (char)min, (char)max);
c2e66d9e
GS
1286 }
1287
c7f1f016 1288#ifdef EBCDIC
8ada0baa
JH
1289 if ((isLOWER(min) && isLOWER(max)) ||
1290 (isUPPER(min) && isUPPER(max))) {
1291 if (isLOWER(min)) {
1292 for (i = min; i <= max; i++)
1293 if (isLOWER(i))
db42d148 1294 *d++ = NATIVE_TO_NEED(has_utf8,i);
8ada0baa
JH
1295 } else {
1296 for (i = min; i <= max; i++)
1297 if (isUPPER(i))
db42d148 1298 *d++ = NATIVE_TO_NEED(has_utf8,i);
8ada0baa
JH
1299 }
1300 }
1301 else
1302#endif
1303 for (i = min; i <= max; i++)
1304 *d++ = i;
02aa26ce
NT
1305
1306 /* mark the range as done, and continue */
79072805 1307 dorange = FALSE;
01ec43d0 1308 didrange = TRUE;
79072805 1309 continue;
4e553d73 1310 }
02aa26ce
NT
1311
1312 /* range begins (ignore - as first or last char) */
79072805 1313 else if (*s == '-' && s+1 < send && s != start) {
4e553d73 1314 if (didrange) {
1fafa243 1315 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
01ec43d0 1316 }
2b9d42f0 1317 if (has_utf8) {
25716404 1318 *d++ = (char)UTF_TO_NATIVE(0xff); /* use illegal utf8 byte--see pmtrans */
a0ed51b3
LW
1319 s++;
1320 continue;
1321 }
79072805
LW
1322 dorange = TRUE;
1323 s++;
01ec43d0
GS
1324 }
1325 else {
1326 didrange = FALSE;
1327 }
79072805 1328 }
02aa26ce
NT
1329
1330 /* if we get here, we're not doing a transliteration */
1331
0f5d15d6
IZ
1332 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1333 except for the last char, which will be done separately. */
3280af22 1334 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
cc6b7395
IZ
1335 if (s[2] == '#') {
1336 while (s < send && *s != ')')
db42d148 1337 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
155aba94
GS
1338 }
1339 else if (s[2] == '{' /* This should match regcomp.c */
1340 || ((s[2] == 'p' || s[2] == '?') && s[3] == '{'))
1341 {
cc6b7395 1342 I32 count = 1;
0f5d15d6 1343 char *regparse = s + (s[2] == '{' ? 3 : 4);
cc6b7395
IZ
1344 char c;
1345
d9f97599
GS
1346 while (count && (c = *regparse)) {
1347 if (c == '\\' && regparse[1])
1348 regparse++;
4e553d73 1349 else if (c == '{')
cc6b7395 1350 count++;
4e553d73 1351 else if (c == '}')
cc6b7395 1352 count--;
d9f97599 1353 regparse++;
cc6b7395 1354 }
5bdf89e7
IZ
1355 if (*regparse != ')') {
1356 regparse--; /* Leave one char for continuation. */
cc6b7395 1357 yyerror("Sequence (?{...}) not terminated or not {}-balanced");
5bdf89e7 1358 }
0f5d15d6 1359 while (s < regparse)
db42d148 1360 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
cc6b7395 1361 }
748a9306 1362 }
02aa26ce
NT
1363
1364 /* likewise skip #-initiated comments in //x patterns */
3280af22
NIS
1365 else if (*s == '#' && PL_lex_inpat &&
1366 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
748a9306 1367 while (s+1 < send && *s != '\n')
db42d148 1368 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
748a9306 1369 }
02aa26ce 1370
5d1d4326
JH
1371 /* check for embedded arrays
1372 (@foo, @:foo, @'foo, @{foo}, @$foo, @+, @-)
1373 */
7e2040f0 1374 else if (*s == '@' && s[1]
5d1d4326 1375 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$+-", s[1])))
79072805 1376 break;
02aa26ce
NT
1377
1378 /* check for embedded scalars. only stop if we're sure it's a
1379 variable.
1380 */
79072805 1381 else if (*s == '$') {
3280af22 1382 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
79072805 1383 break;
6002328a 1384 if (s + 1 < send && !strchr("()| \r\n\t", s[1]))
79072805
LW
1385 break; /* in regexp, $ might be tail anchor */
1386 }
02aa26ce 1387
2b9d42f0
NIS
1388 /* End of else if chain - OP_TRANS rejoin rest */
1389
02aa26ce 1390 /* backslashes */
79072805
LW
1391 if (*s == '\\' && s+1 < send) {
1392 s++;
02aa26ce
NT
1393
1394 /* some backslashes we leave behind */
c9f97d15 1395 if (*leaveit && *s && strchr(leaveit, *s)) {
db42d148
NIS
1396 *d++ = NATIVE_TO_NEED(has_utf8,'\\');
1397 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
79072805
LW
1398 continue;
1399 }
02aa26ce
NT
1400
1401 /* deprecate \1 in strings and substitution replacements */
3280af22 1402 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
a0d0e21e 1403 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
79072805 1404 {
599cee73 1405 if (ckWARN(WARN_SYNTAX))
cea2e8a9 1406 Perl_warner(aTHX_ WARN_SYNTAX, "\\%c better written as $%c", *s, *s);
79072805
LW
1407 *--s = '$';
1408 break;
1409 }
02aa26ce
NT
1410
1411 /* string-change backslash escapes */
3280af22 1412 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
79072805
LW
1413 --s;
1414 break;
1415 }
02aa26ce
NT
1416
1417 /* if we get here, it's either a quoted -, or a digit */
79072805 1418 switch (*s) {
02aa26ce
NT
1419
1420 /* quoted - in transliterations */
79072805 1421 case '-':
3280af22 1422 if (PL_lex_inwhat == OP_TRANS) {
79072805
LW
1423 *d++ = *s++;
1424 continue;
1425 }
1426 /* FALL THROUGH */
1427 default:
11b8faa4 1428 {
7e84c16c 1429 if (ckWARN(WARN_MISC) && isALNUM(*s))
4e553d73 1430 Perl_warner(aTHX_ WARN_MISC,
11b8faa4
JH
1431 "Unrecognized escape \\%c passed through",
1432 *s);
1433 /* default action is to copy the quoted character */
f9a63242 1434 goto default_action;
11b8faa4 1435 }
02aa26ce
NT
1436
1437 /* \132 indicates an octal constant */
79072805
LW
1438 case '0': case '1': case '2': case '3':
1439 case '4': case '5': case '6': case '7':
ba210ebe 1440 {
53305cf1
NC
1441 I32 flags = 0;
1442 STRLEN len = 3;
1443 uv = grok_oct(s, &len, &flags, NULL);
ba210ebe
JH
1444 s += len;
1445 }
012bcf8d 1446 goto NUM_ESCAPE_INSERT;
02aa26ce
NT
1447
1448 /* \x24 indicates a hex constant */
79072805 1449 case 'x':
a0ed51b3
LW
1450 ++s;
1451 if (*s == '{') {
1452 char* e = strchr(s, '}');
a4c04bdc
NC
1453 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
1454 PERL_SCAN_DISALLOW_PREFIX;
53305cf1 1455 STRLEN len;
355860ce 1456
53305cf1 1457 ++s;
adaeee49 1458 if (!e) {
a0ed51b3 1459 yyerror("Missing right brace on \\x{}");
355860ce 1460 continue;
ba210ebe 1461 }
53305cf1
NC
1462 len = e - s;
1463 uv = grok_hex(s, &len, &flags, NULL);
ba210ebe 1464 s = e + 1;
a0ed51b3
LW
1465 }
1466 else {
ba210ebe 1467 {
53305cf1 1468 STRLEN len = 2;
a4c04bdc 1469 I32 flags = PERL_SCAN_DISALLOW_PREFIX;
53305cf1 1470 uv = grok_hex(s, &len, &flags, NULL);
ba210ebe
JH
1471 s += len;
1472 }
012bcf8d
GS
1473 }
1474
1475 NUM_ESCAPE_INSERT:
1476 /* Insert oct or hex escaped character.
301d3d20 1477 * There will always enough room in sv since such
db42d148 1478 * escapes will be longer than any UTF-8 sequence
301d3d20 1479 * they can end up as. */
ba7cea30 1480
c7f1f016
NIS
1481 /* We need to map to chars to ASCII before doing the tests
1482 to cover EBCDIC
1483 */
c4d5f83a 1484 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(uv))) {
9aa983d2 1485 if (!has_utf8 && uv > 255) {
301d3d20
JH
1486 /* Might need to recode whatever we have
1487 * accumulated so far if it contains any
1488 * hibit chars.
1489 *
1490 * (Can't we keep track of that and avoid
1491 * this rescan? --jhi)
012bcf8d 1492 */
c7f1f016 1493 int hicount = 0;
63cd0674
NIS
1494 U8 *c;
1495 for (c = (U8 *) SvPVX(sv); c < (U8 *)d; c++) {
c4d5f83a 1496 if (!NATIVE_IS_INVARIANT(*c)) {
012bcf8d 1497 hicount++;
db42d148 1498 }
012bcf8d 1499 }
63cd0674 1500 if (hicount) {
db42d148
NIS
1501 STRLEN offset = d - SvPVX(sv);
1502 U8 *src, *dst;
1503 d = SvGROW(sv, SvLEN(sv) + hicount + 1) + offset;
1504 src = (U8 *)d - 1;
1505 dst = src+hicount;
1506 d += hicount;
1507 while (src >= (U8 *)SvPVX(sv)) {
c4d5f83a 1508 if (!NATIVE_IS_INVARIANT(*src)) {
63cd0674 1509 U8 ch = NATIVE_TO_ASCII(*src);
db42d148
NIS
1510 *dst-- = UTF8_EIGHT_BIT_LO(ch);
1511 *dst-- = UTF8_EIGHT_BIT_HI(ch);
012bcf8d
GS
1512 }
1513 else {
63cd0674 1514 *dst-- = *src;
012bcf8d 1515 }
c7f1f016 1516 src--;
012bcf8d
GS
1517 }
1518 }
1519 }
1520
9aa983d2 1521 if (has_utf8 || uv > 255) {
9041c2e3 1522 d = (char*)uvchr_to_utf8((U8*)d, uv);
4e553d73 1523 has_utf8 = TRUE;
f9a63242
JH
1524 if (PL_lex_inwhat == OP_TRANS &&
1525 PL_sublex_info.sub_op) {
1526 PL_sublex_info.sub_op->op_private |=
1527 (PL_lex_repl ? OPpTRANS_FROM_UTF
1528 : OPpTRANS_TO_UTF);
f9a63242 1529 }
012bcf8d 1530 }
a0ed51b3 1531 else {
012bcf8d 1532 *d++ = (char)uv;
a0ed51b3 1533 }
012bcf8d
GS
1534 }
1535 else {
c4d5f83a 1536 *d++ = (char) uv;
a0ed51b3 1537 }
79072805 1538 continue;
02aa26ce 1539
4a2d328f
IZ
1540 /* \N{latin small letter a} is a named character */
1541 case 'N':
55eda711 1542 ++s;
423cee85
JH
1543 if (*s == '{') {
1544 char* e = strchr(s, '}');
155aba94 1545 SV *res;
423cee85
JH
1546 STRLEN len;
1547 char *str;
4e553d73 1548
423cee85 1549 if (!e) {
5777a3f7 1550 yyerror("Missing right brace on \\N{}");
423cee85
JH
1551 e = s - 1;
1552 goto cont_scan;
1553 }
55eda711
JH
1554 res = newSVpvn(s + 1, e - s - 1);
1555 res = new_constant( Nullch, 0, "charnames",
1556 res, Nullsv, "\\N{...}" );
f9a63242
JH
1557 if (has_utf8)
1558 sv_utf8_upgrade(res);
423cee85 1559 str = SvPV(res,len);
89491803 1560 if (!has_utf8 && SvUTF8(res)) {
f08d6ad9
GS
1561 char *ostart = SvPVX(sv);
1562 SvCUR_set(sv, d - ostart);
1563 SvPOK_on(sv);
e4f3eed8 1564 *d = '\0';
f08d6ad9 1565 sv_utf8_upgrade(sv);
d2f449dd
SB
1566 /* this just broke our allocation above... */
1567 SvGROW(sv, send - start);
f08d6ad9 1568 d = SvPVX(sv) + SvCUR(sv);
89491803 1569 has_utf8 = TRUE;
f08d6ad9 1570 }
423cee85
JH
1571 if (len > e - s + 4) {
1572 char *odest = SvPVX(sv);
1573
8973db79 1574 SvGROW(sv, (SvLEN(sv) + len - (e - s + 4)));
423cee85
JH
1575 d = SvPVX(sv) + (d - odest);
1576 }
1577 Copy(str, d, len, char);
1578 d += len;
1579 SvREFCNT_dec(res);
1580 cont_scan:
1581 s = e + 1;
1582 }
1583 else
5777a3f7 1584 yyerror("Missing braces on \\N{}");
423cee85
JH
1585 continue;
1586
02aa26ce 1587 /* \c is a control character */
79072805
LW
1588 case 'c':
1589 s++;
ba210ebe
JH
1590 {
1591 U8 c = *s++;
c7f1f016
NIS
1592#ifdef EBCDIC
1593 if (isLOWER(c))
1594 c = toUPPER(c);
1595#endif
db42d148 1596 *d++ = NATIVE_TO_NEED(has_utf8,toCTRL(c));
ba210ebe 1597 }
79072805 1598 continue;
02aa26ce
NT
1599
1600 /* printf-style backslashes, formfeeds, newlines, etc */
79072805 1601 case 'b':
db42d148 1602 *d++ = NATIVE_TO_NEED(has_utf8,'\b');
79072805
LW
1603 break;
1604 case 'n':
db42d148 1605 *d++ = NATIVE_TO_NEED(has_utf8,'\n');
79072805
LW
1606 break;
1607 case 'r':
db42d148 1608 *d++ = NATIVE_TO_NEED(has_utf8,'\r');
79072805
LW
1609 break;
1610 case 'f':
db42d148 1611 *d++ = NATIVE_TO_NEED(has_utf8,'\f');
79072805
LW
1612 break;
1613 case 't':
db42d148 1614 *d++ = NATIVE_TO_NEED(has_utf8,'\t');
79072805 1615 break;
34a3fe2a 1616 case 'e':
db42d148 1617 *d++ = ASCII_TO_NEED(has_utf8,'\033');
34a3fe2a
PP
1618 break;
1619 case 'a':
db42d148 1620 *d++ = ASCII_TO_NEED(has_utf8,'\007');
79072805 1621 break;
02aa26ce
NT
1622 } /* end switch */
1623
79072805
LW
1624 s++;
1625 continue;
02aa26ce
NT
1626 } /* end if (backslash) */
1627
f9a63242 1628 default_action:
2b9d42f0
NIS
1629 /* If we started with encoded form, or already know we want it
1630 and then encode the next character */
1631 if ((has_utf8 || this_utf8) && !NATIVE_IS_INVARIANT((U8)(*s))) {
1632 STRLEN len = 1;
1633 UV uv = (this_utf8) ? utf8n_to_uvchr((U8*)s, send - s, &len, 0) : (UV) ((U8) *s);
1634 STRLEN need = UNISKIP(NATIVE_TO_UNI(uv));
1635 s += len;
1636 if (need > len) {
1637 /* encoded value larger than old, need extra space (NOTE: SvCUR() not set here) */
1638 STRLEN off = d - SvPVX(sv);
1639 d = SvGROW(sv, SvLEN(sv) + (need-len)) + off;
1640 }
1641 d = (char*)uvchr_to_utf8((U8*)d, uv);
1642 has_utf8 = TRUE;
1643 }
1644 else {
1645 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1646 }
02aa26ce
NT
1647 } /* while loop to process each character */
1648
1649 /* terminate the string and set up the sv */
79072805 1650 *d = '\0';
463ee0b2 1651 SvCUR_set(sv, d - SvPVX(sv));
2b9d42f0 1652 if (SvCUR(sv) >= SvLEN(sv))
585602fa 1653 Perl_croak(aTHX_ "panic: constant overflowed allocated space");
2b9d42f0 1654
79072805 1655 SvPOK_on(sv);
2b9d42f0 1656 if (has_utf8) {
7e2040f0 1657 SvUTF8_on(sv);
2b9d42f0
NIS
1658 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1659 PL_sublex_info.sub_op->op_private |=
1660 (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1661 }
1662 }
79072805 1663
02aa26ce 1664 /* shrink the sv if we allocated more than we used */
79072805
LW
1665 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1666 SvLEN_set(sv, SvCUR(sv) + 1);
463ee0b2 1667 Renew(SvPVX(sv), SvLEN(sv), char);
79072805 1668 }
02aa26ce 1669
9b599b2a 1670 /* return the substring (via yylval) only if we parsed anything */
3280af22
NIS
1671 if (s > PL_bufptr) {
1672 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
4e553d73 1673 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
b3ac6de7 1674 sv, Nullsv,
4e553d73 1675 ( PL_lex_inwhat == OP_TRANS
b3ac6de7 1676 ? "tr"
3280af22 1677 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
b3ac6de7
IZ
1678 ? "s"
1679 : "qq")));
79072805 1680 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
b3ac6de7 1681 } else
8990e307 1682 SvREFCNT_dec(sv);
79072805
LW
1683 return s;
1684}
1685
ffb4593c
NT
1686/* S_intuit_more
1687 * Returns TRUE if there's more to the expression (e.g., a subscript),
1688 * FALSE otherwise.
ffb4593c
NT
1689 *
1690 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1691 *
1692 * ->[ and ->{ return TRUE
1693 * { and [ outside a pattern are always subscripts, so return TRUE
1694 * if we're outside a pattern and it's not { or [, then return FALSE
1695 * if we're in a pattern and the first char is a {
1696 * {4,5} (any digits around the comma) returns FALSE
1697 * if we're in a pattern and the first char is a [
1698 * [] returns FALSE
1699 * [SOMETHING] has a funky algorithm to decide whether it's a
1700 * character class or not. It has to deal with things like
1701 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1702 * anything else returns TRUE
1703 */
1704
9cbb5ea2
GS
1705/* This is the one truly awful dwimmer necessary to conflate C and sed. */
1706
76e3520e 1707STATIC int
cea2e8a9 1708S_intuit_more(pTHX_ register char *s)
79072805 1709{
3280af22 1710 if (PL_lex_brackets)
79072805
LW
1711 return TRUE;
1712 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1713 return TRUE;
1714 if (*s != '{' && *s != '[')
1715 return FALSE;
3280af22 1716 if (!PL_lex_inpat)
79072805
LW
1717 return TRUE;
1718
1719 /* In a pattern, so maybe we have {n,m}. */
1720 if (*s == '{') {
1721 s++;
1722 if (!isDIGIT(*s))
1723 return TRUE;
1724 while (isDIGIT(*s))
1725 s++;
1726 if (*s == ',')
1727 s++;
1728 while (isDIGIT(*s))
1729 s++;
1730 if (*s == '}')
1731 return FALSE;
1732 return TRUE;
1733
1734 }
1735
1736 /* On the other hand, maybe we have a character class */
1737
1738 s++;
1739 if (*s == ']' || *s == '^')
1740 return FALSE;
1741 else {
ffb4593c 1742 /* this is terrifying, and it works */
79072805
LW
1743 int weight = 2; /* let's weigh the evidence */
1744 char seen[256];
f27ffc4a 1745 unsigned char un_char = 255, last_un_char;
93a17b20 1746 char *send = strchr(s,']');
3280af22 1747 char tmpbuf[sizeof PL_tokenbuf * 4];
79072805
LW
1748
1749 if (!send) /* has to be an expression */
1750 return TRUE;
1751
1752 Zero(seen,256,char);
1753 if (*s == '$')
1754 weight -= 3;
1755 else if (isDIGIT(*s)) {
1756 if (s[1] != ']') {
1757 if (isDIGIT(s[1]) && s[2] == ']')
1758 weight -= 10;
1759 }
1760 else
1761 weight -= 100;
1762 }
1763 for (; s < send; s++) {
1764 last_un_char = un_char;
1765 un_char = (unsigned char)*s;
1766 switch (*s) {
1767 case '@':
1768 case '&':
1769 case '$':
1770 weight -= seen[un_char] * 10;
7e2040f0 1771 if (isALNUM_lazy_if(s+1,UTF)) {
8903cb82 1772 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
a0d0e21e 1773 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
79072805
LW
1774 weight -= 100;
1775 else
1776 weight -= 10;
1777 }
1778 else if (*s == '$' && s[1] &&
93a17b20
LW
1779 strchr("[#!%*<>()-=",s[1])) {
1780 if (/*{*/ strchr("])} =",s[2]))
79072805
LW
1781 weight -= 10;
1782 else
1783 weight -= 1;
1784 }
1785 break;
1786 case '\\':
1787 un_char = 254;
1788 if (s[1]) {
93a17b20 1789 if (strchr("wds]",s[1]))
79072805
LW
1790 weight += 100;
1791 else if (seen['\''] || seen['"'])
1792 weight += 1;
93a17b20 1793 else if (strchr("rnftbxcav",s[1]))
79072805
LW
1794 weight += 40;
1795 else if (isDIGIT(s[1])) {
1796 weight += 40;
1797 while (s[1] && isDIGIT(s[1]))
1798 s++;
1799 }
1800 }
1801 else
1802 weight += 100;
1803 break;
1804 case '-':
1805 if (s[1] == '\\')
1806 weight += 50;
93a17b20 1807 if (strchr("aA01! ",last_un_char))
79072805 1808 weight += 30;
93a17b20 1809 if (strchr("zZ79~",s[1]))
79072805 1810 weight += 30;
f27ffc4a
GS
1811 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1812 weight -= 5; /* cope with negative subscript */
79072805
LW
1813 break;
1814 default:
93a17b20 1815 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
79072805
LW
1816 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1817 char *d = tmpbuf;
1818 while (isALPHA(*s))
1819 *d++ = *s++;
1820 *d = '\0';
1821 if (keyword(tmpbuf, d - tmpbuf))
1822 weight -= 150;
1823 }
1824 if (un_char == last_un_char + 1)
1825 weight += 5;
1826 weight -= seen[un_char];
1827 break;
1828 }
1829 seen[un_char]++;
1830 }
1831 if (weight >= 0) /* probably a character class */
1832 return FALSE;
1833 }
1834
1835 return TRUE;
1836}
ffed7fef 1837
ffb4593c
NT
1838/*
1839 * S_intuit_method
1840 *
1841 * Does all the checking to disambiguate
1842 * foo bar
1843 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1844 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1845 *
1846 * First argument is the stuff after the first token, e.g. "bar".
1847 *
1848 * Not a method if bar is a filehandle.
1849 * Not a method if foo is a subroutine prototyped to take a filehandle.
1850 * Not a method if it's really "Foo $bar"
1851 * Method if it's "foo $bar"
1852 * Not a method if it's really "print foo $bar"
1853 * Method if it's really "foo package::" (interpreted as package->foo)
1854 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
3cb0bbe5 1855 * Not a method if bar is a filehandle or package, but is quoted with
ffb4593c
NT
1856 * =>
1857 */
1858
76e3520e 1859STATIC int
cea2e8a9 1860S_intuit_method(pTHX_ char *start, GV *gv)
a0d0e21e
LW
1861{
1862 char *s = start + (*start == '$');
3280af22 1863 char tmpbuf[sizeof PL_tokenbuf];
a0d0e21e
LW
1864 STRLEN len;
1865 GV* indirgv;
1866
1867 if (gv) {
b6c543e3 1868 CV *cv;
a0d0e21e
LW
1869 if (GvIO(gv))
1870 return 0;
b6c543e3
IZ
1871 if ((cv = GvCVu(gv))) {
1872 char *proto = SvPVX(cv);
1873 if (proto) {
1874 if (*proto == ';')
1875 proto++;
1876 if (*proto == '*')
1877 return 0;
1878 }
1879 } else
a0d0e21e
LW
1880 gv = 0;
1881 }
8903cb82 1882 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
ffb4593c
NT
1883 /* start is the beginning of the possible filehandle/object,
1884 * and s is the end of it
1885 * tmpbuf is a copy of it
1886 */
1887
a0d0e21e 1888 if (*start == '$') {
3280af22 1889 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
a0d0e21e
LW
1890 return 0;
1891 s = skipspace(s);
3280af22
NIS
1892 PL_bufptr = start;
1893 PL_expect = XREF;
a0d0e21e
LW
1894 return *s == '(' ? FUNCMETH : METHOD;
1895 }
1896 if (!keyword(tmpbuf, len)) {
c3e0f903
GS
1897 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1898 len -= 2;
1899 tmpbuf[len] = '\0';
1900 goto bare_package;
1901 }
1902 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
8ebc5c01 1903 if (indirgv && GvCVu(indirgv))
a0d0e21e
LW
1904 return 0;
1905 /* filehandle or package name makes it a method */
89bfa8cd 1906 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
a0d0e21e 1907 s = skipspace(s);
3280af22 1908 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
55497cff 1909 return 0; /* no assumptions -- "=>" quotes bearword */
c3e0f903 1910 bare_package:
3280af22 1911 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
79cb57f6 1912 newSVpvn(tmpbuf,len));
3280af22
NIS
1913 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1914 PL_expect = XTERM;
a0d0e21e 1915 force_next(WORD);
3280af22 1916 PL_bufptr = s;
a0d0e21e
LW
1917 return *s == '(' ? FUNCMETH : METHOD;
1918 }
1919 }
1920 return 0;
1921}
1922
ffb4593c
NT
1923/*
1924 * S_incl_perldb
1925 * Return a string of Perl code to load the debugger. If PERL5DB
1926 * is set, it will return the contents of that, otherwise a
1927 * compile-time require of perl5db.pl.
1928 */
1929
76e3520e 1930STATIC char*
cea2e8a9 1931S_incl_perldb(pTHX)
a0d0e21e 1932{
3280af22 1933 if (PL_perldb) {
76e3520e 1934 char *pdb = PerlEnv_getenv("PERL5DB");
a0d0e21e
LW
1935
1936 if (pdb)
1937 return pdb;
61bb5906 1938 SETERRNO(0,SS$_NORMAL);
a0d0e21e
LW
1939 return "BEGIN { require 'perl5db.pl' }";
1940 }
1941 return "";
1942}
1943
1944
16d20bd9 1945/* Encoded script support. filter_add() effectively inserts a
4e553d73 1946 * 'pre-processing' function into the current source input stream.
16d20bd9
AD
1947 * Note that the filter function only applies to the current source file
1948 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1949 *
1950 * The datasv parameter (which may be NULL) can be used to pass
1951 * private data to this instance of the filter. The filter function
1952 * can recover the SV using the FILTER_DATA macro and use it to
1953 * store private buffers and state information.
1954 *
1955 * The supplied datasv parameter is upgraded to a PVIO type
4755096e 1956 * and the IoDIRP/IoANY field is used to store the function pointer,
e0c19803 1957 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
16d20bd9
AD
1958 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1959 * private use must be set using malloc'd pointers.
1960 */
16d20bd9
AD
1961
1962SV *
864dbfa3 1963Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
16d20bd9 1964{
f4c556ac
GS
1965 if (!funcp)
1966 return Nullsv;
1967
3280af22
NIS
1968 if (!PL_rsfp_filters)
1969 PL_rsfp_filters = newAV();
16d20bd9 1970 if (!datasv)
8c52afec 1971 datasv = NEWSV(255,0);
16d20bd9 1972 if (!SvUPGRADE(datasv, SVt_PVIO))
cea2e8a9 1973 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
4755096e 1974 IoANY(datasv) = (void *)funcp; /* stash funcp into spare field */
e0c19803 1975 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
f4c556ac
GS
1976 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1977 funcp, SvPV_nolen(datasv)));
3280af22
NIS
1978 av_unshift(PL_rsfp_filters, 1);
1979 av_store(PL_rsfp_filters, 0, datasv) ;
16d20bd9
AD
1980 return(datasv);
1981}
4e553d73 1982
16d20bd9
AD
1983
1984/* Delete most recently added instance of this filter function. */
a0d0e21e 1985void
864dbfa3 1986Perl_filter_del(pTHX_ filter_t funcp)
16d20bd9 1987{
e0c19803 1988 SV *datasv;
f4c556ac 1989 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp));
3280af22 1990 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
16d20bd9
AD
1991 return;
1992 /* if filter is on top of stack (usual case) just pop it off */
e0c19803 1993 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
4755096e 1994 if (IoANY(datasv) == (void *)funcp) {
e0c19803 1995 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
4755096e 1996 IoANY(datasv) = (void *)NULL;
3280af22 1997 sv_free(av_pop(PL_rsfp_filters));
e50aee73 1998
16d20bd9
AD
1999 return;
2000 }
2001 /* we need to search for the correct entry and clear it */
cea2e8a9 2002 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
16d20bd9
AD
2003}
2004
2005
2006/* Invoke the n'th filter function for the current rsfp. */
2007I32
864dbfa3 2008Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
4e553d73
NIS
2009
2010
8ac85365 2011 /* 0 = read one text line */
a0d0e21e 2012{
16d20bd9
AD
2013 filter_t funcp;
2014 SV *datasv = NULL;
e50aee73 2015
3280af22 2016 if (!PL_rsfp_filters)
16d20bd9 2017 return -1;
3280af22 2018 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
16d20bd9
AD
2019 /* Provide a default input filter to make life easy. */
2020 /* Note that we append to the line. This is handy. */
f4c556ac
GS
2021 DEBUG_P(PerlIO_printf(Perl_debug_log,
2022 "filter_read %d: from rsfp\n", idx));
4e553d73 2023 if (maxlen) {
16d20bd9
AD
2024 /* Want a block */
2025 int len ;
2026 int old_len = SvCUR(buf_sv) ;
2027
2028 /* ensure buf_sv is large enough */
2029 SvGROW(buf_sv, old_len + maxlen) ;
3280af22
NIS
2030 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
2031 if (PerlIO_error(PL_rsfp))
37120919
AD
2032 return -1; /* error */
2033 else
2034 return 0 ; /* end of file */
2035 }
16d20bd9
AD
2036 SvCUR_set(buf_sv, old_len + len) ;
2037 } else {
2038 /* Want a line */
3280af22
NIS
2039 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
2040 if (PerlIO_error(PL_rsfp))
37120919
AD
2041 return -1; /* error */
2042 else
2043 return 0 ; /* end of file */
2044 }
16d20bd9
AD
2045 }
2046 return SvCUR(buf_sv);
2047 }
2048 /* Skip this filter slot if filter has been deleted */
3280af22 2049 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
f4c556ac
GS
2050 DEBUG_P(PerlIO_printf(Perl_debug_log,
2051 "filter_read %d: skipped (filter deleted)\n",
2052 idx));
16d20bd9
AD
2053 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
2054 }
2055 /* Get function pointer hidden within datasv */
4755096e 2056 funcp = (filter_t)IoANY(datasv);
f4c556ac
GS
2057 DEBUG_P(PerlIO_printf(Perl_debug_log,
2058 "filter_read %d: via function %p (%s)\n",
2059 idx, funcp, SvPV_nolen(datasv)));
16d20bd9
AD
2060 /* Call function. The function is expected to */
2061 /* call "FILTER_READ(idx+1, buf_sv)" first. */
37120919 2062 /* Return: <0:error, =0:eof, >0:not eof */
acfe0abc 2063 return (*funcp)(aTHX_ idx, buf_sv, maxlen);
16d20bd9
AD
2064}
2065
76e3520e 2066STATIC char *
cea2e8a9 2067S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
16d20bd9 2068{
c39cd008 2069#ifdef PERL_CR_FILTER
3280af22 2070 if (!PL_rsfp_filters) {
c39cd008 2071 filter_add(S_cr_textfilter,NULL);
a868473f
NIS
2072 }
2073#endif
3280af22 2074 if (PL_rsfp_filters) {
16d20bd9 2075
55497cff
PP
2076 if (!append)
2077 SvCUR_set(sv, 0); /* start with empty line */
16d20bd9
AD
2078 if (FILTER_READ(0, sv, 0) > 0)
2079 return ( SvPVX(sv) ) ;
2080 else
2081 return Nullch ;
2082 }
9d116dd7 2083 else
fd049845 2084 return (sv_gets(sv, fp, append));
a0d0e21e
LW
2085}
2086
01ec43d0
GS
2087STATIC HV *
2088S_find_in_my_stash(pTHX_ char *pkgname, I32 len)
def3634b
GS
2089{
2090 GV *gv;
2091
01ec43d0 2092 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
def3634b
GS
2093 return PL_curstash;
2094
2095 if (len > 2 &&
2096 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
01ec43d0
GS
2097 (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV)))
2098 {
2099 return GvHV(gv); /* Foo:: */
def3634b
GS
2100 }
2101
2102 /* use constant CLASS => 'MyClass' */
2103 if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) {
2104 SV *sv;
2105 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2106 pkgname = SvPV_nolen(sv);
2107 }
2108 }
2109
2110 return gv_stashpv(pkgname, FALSE);
2111}
a0d0e21e 2112
748a9306
LW
2113#ifdef DEBUGGING
2114 static char* exp_name[] =
09bef843
SB
2115 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2116 "ATTRTERM", "TERMBLOCK"
2117 };
748a9306 2118#endif
463ee0b2 2119
02aa26ce
NT
2120/*
2121 yylex
2122
2123 Works out what to call the token just pulled out of the input
2124 stream. The yacc parser takes care of taking the ops we return and
2125 stitching them into a tree.
2126
2127 Returns:
2128 PRIVATEREF
2129
2130 Structure:
2131 if read an identifier
2132 if we're in a my declaration
2133 croak if they tried to say my($foo::bar)
2134 build the ops for a my() declaration
2135 if it's an access to a my() variable
2136 are we in a sort block?
2137 croak if my($a); $a <=> $b
2138 build ops for access to a my() variable
2139 if in a dq string, and they've said @foo and we can't find @foo
2140 croak
2141 build ops for a bareword
2142 if we already built the token before, use it.
2143*/
2144
dba4d153 2145#ifdef USE_PURE_BISON
864dbfa3 2146int
dba4d153 2147Perl_yylex_r(pTHX_ YYSTYPE *lvalp, int *lcharp)
378cc40b 2148{
20141f0e
RI
2149 int r;
2150
6f202aea 2151 yyactlevel++;
20141f0e
RI
2152 yylval_pointer[yyactlevel] = lvalp;
2153 yychar_pointer[yyactlevel] = lcharp;
b73d6f50
RI
2154 if (yyactlevel >= YYMAXLEVEL)
2155 Perl_croak(aTHX_ "panic: YYMAXLEVEL");
20141f0e 2156
dba4d153 2157 r = Perl_yylex(aTHX);
20141f0e 2158
d8ae6756
RI
2159 if (yyactlevel > 0)
2160 yyactlevel--;
20141f0e
RI
2161
2162 return r;
2163}
dba4d153 2164#endif
20141f0e 2165
dba4d153
JH
2166#ifdef __SC__
2167#pragma segment Perl_yylex
2168#endif
dba4d153 2169int
dba4d153 2170Perl_yylex(pTHX)
20141f0e 2171{
79072805 2172 register char *s;
378cc40b 2173 register char *d;
79072805 2174 register I32 tmp;
463ee0b2 2175 STRLEN len;
161b471a
NIS
2176 GV *gv = Nullgv;
2177 GV **gvp = 0;
aa7440fb 2178 bool bof = FALSE;
a687059c 2179
02aa26ce 2180 /* check if there's an identifier for us to look at */
8eceec63 2181 if (PL_pending_ident)
e930465f 2182 return S_pending_ident(aTHX);
bbce6d69 2183
02aa26ce
NT
2184 /* no identifier pending identification */
2185
3280af22 2186 switch (PL_lex_state) {
79072805
LW
2187#ifdef COMMENTARY
2188 case LEX_NORMAL: /* Some compilers will produce faster */
2189 case LEX_INTERPNORMAL: /* code if we comment these out. */
2190 break;
2191#endif
2192
09bef843 2193 /* when we've already built the next token, just pull it out of the queue */
79072805 2194 case LEX_KNOWNEXT:
3280af22
NIS
2195 PL_nexttoke--;
2196 yylval = PL_nextval[PL_nexttoke];
2197 if (!PL_nexttoke) {
2198 PL_lex_state = PL_lex_defer;
2199 PL_expect = PL_lex_expect;
2200 PL_lex_defer = LEX_NORMAL;
463ee0b2 2201 }
607df283 2202 DEBUG_T({ PerlIO_printf(Perl_debug_log,
4659c93f 2203 "### Next token after '%s' was known, type %"IVdf"\n", PL_bufptr,
5f80b19c 2204 (IV)PL_nexttype[PL_nexttoke]); });
607df283 2205
3280af22 2206 return(PL_nexttype[PL_nexttoke]);
79072805 2207
02aa26ce 2208 /* interpolated case modifiers like \L \U, including \Q and \E.
3280af22 2209 when we get here, PL_bufptr is at the \
02aa26ce 2210 */
79072805
LW
2211 case LEX_INTERPCASEMOD:
2212#ifdef DEBUGGING
3280af22 2213 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
cea2e8a9 2214 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
79072805 2215#endif
02aa26ce 2216 /* handle \E or end of string */
3280af22 2217 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
a0d0e21e 2218 char oldmod;
02aa26ce
NT
2219
2220 /* if at a \E */
3280af22
NIS
2221 if (PL_lex_casemods) {
2222 oldmod = PL_lex_casestack[--PL_lex_casemods];
2223 PL_lex_casestack[PL_lex_casemods] = '\0';
02aa26ce 2224
3280af22
NIS
2225 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2226 PL_bufptr += 2;
2227 PL_lex_state = LEX_INTERPCONCAT;
a0d0e21e 2228 }
79072805
LW
2229 return ')';
2230 }
3280af22
NIS
2231 if (PL_bufptr != PL_bufend)
2232 PL_bufptr += 2;
2233 PL_lex_state = LEX_INTERPCONCAT;
cea2e8a9 2234 return yylex();
79072805
LW
2235 }
2236 else {
607df283 2237 DEBUG_T({ PerlIO_printf(Perl_debug_log,
5f80b19c 2238 "### Saw case modifier at '%s'\n", PL_bufptr); });
3280af22 2239 s = PL_bufptr + 1;
79072805
LW
2240 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2241 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
a0d0e21e 2242 if (strchr("LU", *s) &&
3280af22 2243 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
a0d0e21e 2244 {
3280af22 2245 PL_lex_casestack[--PL_lex_casemods] = '\0';
a0d0e21e
LW
2246 return ')';
2247 }
3280af22
NIS
2248 if (PL_lex_casemods > 10) {
2249 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2250 if (newlb != PL_lex_casestack) {
a0d0e21e 2251 SAVEFREEPV(newlb);
3280af22 2252 PL_lex_casestack = newlb;
a0d0e21e
LW
2253 }
2254 }
3280af22
NIS
2255 PL_lex_casestack[PL_lex_casemods++] = *s;
2256 PL_lex_casestack[PL_lex_casemods] = '\0';
2257 PL_lex_state = LEX_INTERPCONCAT;
2258 PL_nextval[PL_nexttoke].ival = 0;
79072805
LW
2259 force_next('(');
2260 if (*s == 'l')
3280af22 2261 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
79072805 2262 else if (*s == 'u')
3280af22 2263 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
79072805 2264 else if (*s == 'L')
3280af22 2265 PL_nextval[PL_nexttoke].ival = OP_LC;
79072805 2266 else if (*s == 'U')
3280af22 2267 PL_nextval[PL_nexttoke].ival = OP_UC;
a0d0e21e 2268 else if (*s == 'Q')
3280af22 2269 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
79072805 2270 else
cea2e8a9 2271 Perl_croak(aTHX_ "panic: yylex");
3280af22 2272 PL_bufptr = s + 1;
79072805 2273 force_next(FUNC);
3280af22
NIS
2274 if (PL_lex_starts) {
2275 s = PL_bufptr;
2276 PL_lex_starts = 0;
79072805
LW
2277 Aop(OP_CONCAT);
2278 }
2279 else
cea2e8a9 2280 return yylex();
79072805
LW
2281 }
2282
55497cff
PP
2283 case LEX_INTERPPUSH:
2284 return sublex_push();
2285
79072805 2286 case LEX_INTERPSTART:
3280af22 2287 if (PL_bufptr == PL_bufend)
79072805 2288 return sublex_done();
607df283 2289 DEBUG_T({ PerlIO_printf(Perl_debug_log,
5f80b19c 2290 "### Interpolated variable at '%s'\n", PL_bufptr); });
3280af22
NIS
2291 PL_expect = XTERM;
2292 PL_lex_dojoin = (*PL_bufptr == '@');
2293 PL_lex_state = LEX_INTERPNORMAL;
2294 if (PL_lex_dojoin) {
2295 PL_nextval[PL_nexttoke].ival = 0;
79072805 2296 force_next(',');
4d1ff10f 2297#ifdef USE_5005THREADS
533c011a
NIS
2298 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2299 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
554b3eca
MB
2300 force_next(PRIVATEREF);
2301#else
a0d0e21e 2302 force_ident("\"", '$');
4d1ff10f 2303#endif /* USE_5005THREADS */
3280af22 2304 PL_nextval[PL_nexttoke].ival = 0;
79072805 2305 force_next('$');
3280af22 2306 PL_nextval[PL_nexttoke].ival = 0;
79072805 2307 force_next('(');
3280af22 2308 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
79072805
LW
2309 force_next(FUNC);
2310 }
3280af22
NIS
2311 if (PL_lex_starts++) {
2312 s = PL_bufptr;
79072805
LW
2313 Aop(OP_CONCAT);
2314 }
cea2e8a9 2315 return yylex();
79072805
LW
2316
2317 case LEX_INTERPENDMAYBE:
3280af22
NIS
2318 if (intuit_more(PL_bufptr)) {
2319 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
79072805
LW
2320 break;
2321 }
2322 /* FALL THROUGH */
2323
2324 case LEX_INTERPEND:
3280af22
NIS
2325 if (PL_lex_dojoin) {
2326 PL_lex_dojoin = FALSE;
2327 PL_lex_state = LEX_INTERPCONCAT;
79072805
LW
2328 return ')';
2329 }
43a16006 2330 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
25da4f38 2331 && SvEVALED(PL_lex_repl))
43a16006 2332 {
e9fa98b2 2333 if (PL_bufptr != PL_bufend)
cea2e8a9 2334 Perl_croak(aTHX_ "Bad evalled substitution pattern");
e9fa98b2
HS
2335 PL_lex_repl = Nullsv;
2336 }
79072805
LW
2337 /* FALLTHROUGH */
2338 case LEX_INTERPCONCAT:
2339#ifdef DEBUGGING
3280af22 2340 if (PL_lex_brackets)
cea2e8a9 2341 Perl_croak(aTHX_ "panic: INTERPCONCAT");
79072805 2342#endif
3280af22 2343 if (PL_bufptr == PL_bufend)
79072805
LW
2344 return sublex_done();
2345
3280af22
NIS
2346 if (SvIVX(PL_linestr) == '\'') {
2347 SV *sv = newSVsv(PL_linestr);
2348 if (!PL_lex_inpat)
76e3520e 2349 sv = tokeq(sv);
3280af22 2350 else if ( PL_hints & HINT_NEW_RE )
b3ac6de7 2351 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
79072805 2352 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3280af22 2353 s = PL_bufend;
79072805
LW
2354 }
2355 else {
3280af22 2356 s = scan_const(PL_bufptr);
79072805 2357 if (*s == '\\')
3280af22 2358 PL_lex_state = LEX_INTERPCASEMOD;
79072805 2359 else
3280af22 2360 PL_lex_state = LEX_INTERPSTART;
79072805
LW
2361 }
2362
3280af22
NIS
2363 if (s != PL_bufptr) {
2364 PL_nextval[PL_nexttoke] = yylval;
2365 PL_expect = XTERM;
79072805 2366 force_next(THING);
3280af22 2367 if (PL_lex_starts++)
79072805
LW
2368 Aop(OP_CONCAT);
2369 else {
3280af22 2370 PL_bufptr = s;
cea2e8a9 2371 return yylex();
79072805
LW
2372 }
2373 }
2374
cea2e8a9 2375 return yylex();
a0d0e21e 2376 case LEX_FORMLINE:
3280af22
NIS
2377 PL_lex_state = LEX_NORMAL;
2378 s = scan_formline(PL_bufptr);
2379 if (!PL_lex_formbrack)
a0d0e21e
LW
2380 goto rightbracket;
2381 OPERATOR(';');
79072805
LW
2382 }
2383
3280af22
NIS
2384 s = PL_bufptr;
2385 PL_oldoldbufptr = PL_oldbufptr;
2386 PL_oldbufptr = s;
607df283 2387 DEBUG_T( {
bf49b057
GS
2388 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2389 exp_name[PL_expect], s);
5f80b19c 2390 } );
463ee0b2
LW
2391
2392 retry:
378cc40b
LW
2393 switch (*s) {
2394 default:
7e2040f0 2395 if (isIDFIRST_lazy_if(s,UTF))
834a4ddd 2396 goto keylookup;
cea2e8a9 2397 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
e929a76b
LW
2398 case 4:
2399 case 26:
2400 goto fake_eof; /* emulate EOF on ^D or ^Z */
378cc40b 2401 case 0:
3280af22
NIS
2402 if (!PL_rsfp) {
2403 PL_last_uni = 0;
2404 PL_last_lop = 0;
2405 if (PL_lex_brackets)
d98d5fff 2406 yyerror("Missing right curly or square bracket");
4e553d73 2407 DEBUG_T( { PerlIO_printf(Perl_debug_log,
607df283 2408 "### Tokener got EOF\n");
5f80b19c 2409 } );
79072805 2410 TOKEN(0);
463ee0b2 2411 }
3280af22 2412 if (s++ < PL_bufend)
a687059c 2413 goto retry; /* ignore stray nulls */
3280af22
NIS
2414 PL_last_uni = 0;
2415 PL_last_lop = 0;
2416 if (!PL_in_eval && !PL_preambled) {
2417 PL_preambled = TRUE;
2418 sv_setpv(PL_linestr,incl_perldb());
2419 if (SvCUR(PL_linestr))
2420 sv_catpv(PL_linestr,";");
2421 if (PL_preambleav){
2422 while(AvFILLp(PL_preambleav) >= 0) {
2423 SV *tmpsv = av_shift(PL_preambleav);
2424 sv_catsv(PL_linestr, tmpsv);
2425 sv_catpv(PL_linestr, ";");
91b7def8
PP
2426 sv_free(tmpsv);
2427 }
3280af22
NIS
2428 sv_free((SV*)PL_preambleav);
2429 PL_preambleav = NULL;
91b7def8 2430 }
3280af22
NIS
2431 if (PL_minus_n || PL_minus_p) {
2432 sv_catpv(PL_linestr, "LINE: while (<>) {");
2433 if (PL_minus_l)
2434 sv_catpv(PL_linestr,"chomp;");
2435 if (PL_minus_a) {
3280af22
NIS
2436 if (PL_minus_F) {
2437 if (strchr("/'\"", *PL_splitstr)
2438 && strchr(PL_splitstr + 1, *PL_splitstr))
cea2e8a9 2439 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
54310121
PP
2440 else {
2441 char delim;
2442 s = "'~#\200\1'"; /* surely one char is unused...*/
3280af22 2443 while (s[1] && strchr(PL_splitstr, *s)) s++;
54310121 2444 delim = *s;
75c72d73 2445 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s%c",
46fc3d4c 2446 "q" + (delim == '\''), delim);
3280af22 2447 for (s = PL_splitstr; *s; s++) {
54310121 2448 if (*s == '\\')
3280af22
NIS
2449 sv_catpvn(PL_linestr, "\\", 1);
2450 sv_catpvn(PL_linestr, s, 1);
54310121 2451 }
cea2e8a9 2452 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
54310121 2453 }
2304df62
AD
2454 }
2455 else
75c72d73 2456 sv_catpv(PL_linestr,"our @F=split(' ');");
2304df62 2457 }
79072805 2458 }
3280af22
NIS
2459 sv_catpv(PL_linestr, "\n");
2460 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2461 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
207e3d1a 2462 PL_last_lop = PL_last_uni = Nullch;
3280af22 2463 if (PERLDB_LINE && PL_curstash != PL_debstash) {
a0d0e21e
LW
2464 SV *sv = NEWSV(85,0);
2465
2466 sv_upgrade(sv, SVt_PVMG);
3280af22 2467 sv_setsv(sv,PL_linestr);
57843af0 2468 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
a0d0e21e 2469 }
79072805 2470 goto retry;
a687059c 2471 }
e929a76b 2472 do {
aa7440fb 2473 bof = PL_rsfp ? TRUE : FALSE;
7e28d3af
JH
2474 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2475 fake_eof:
2476 if (PL_rsfp) {
2477 if (PL_preprocess && !PL_in_eval)
2478 (void)PerlProc_pclose(PL_rsfp);
2479 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2480 PerlIO_clearerr(PL_rsfp);
2481 else
2482 (void)PerlIO_close(PL_rsfp);
2483 PL_rsfp = Nullfp;
2484 PL_doextract = FALSE;
2485 }
2486 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2487 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2488 sv_catpv(PL_linestr,";}");
2489 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2490 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
207e3d1a 2491 PL_last_lop = PL_last_uni = Nullch;
7e28d3af
JH
2492 PL_minus_n = PL_minus_p = 0;
2493 goto retry;
2494 }
2495 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
207e3d1a 2496 PL_last_lop = PL_last_uni = Nullch;
7e28d3af
JH
2497 sv_setpv(PL_linestr,"");
2498 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2499 }
2500 /* if it looks like the start of a BOM, check if it in fact is */
2501 else if (bof && (!*s || *(U8*)s == 0xEF || *(U8*)s >= 0xFE)) {
226017aa 2502#ifdef PERLIO_IS_STDIO
e3f494f1
JH
2503# ifdef __GNU_LIBRARY__
2504# if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
226017aa
DD
2505# define FTELL_FOR_PIPE_IS_BROKEN
2506# endif
e3f494f1
JH
2507# else
2508# ifdef __GLIBC__
2509# if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
2510# define FTELL_FOR_PIPE_IS_BROKEN
2511# endif
2512# endif
226017aa
DD
2513# endif
2514#endif
2515#ifdef FTELL_FOR_PIPE_IS_BROKEN
2516 /* This loses the possibility to detect the bof
2517 * situation on perl -P when the libc5 is being used.
2518 * Workaround? Maybe attach some extra state to PL_rsfp?
2519 */
2520 if (!PL_preprocess)
7e28d3af 2521 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
226017aa 2522#else
7e28d3af 2523 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
226017aa 2524#endif
7e28d3af 2525 if (bof) {
3280af22 2526 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7e28d3af 2527 s = swallow_bom((U8*)s);
e929a76b 2528 }
378cc40b 2529 }
3280af22 2530 if (PL_doextract) {
a0d0e21e 2531 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
3280af22 2532 PL_doextract = FALSE;
a0d0e21e
LW
2533
2534 /* Incest with pod. */
2535 if (*s == '=' && strnEQ(s, "=cut", 4)) {
3280af22
NIS
2536 sv_setpv(PL_linestr, "");
2537 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2538 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
207e3d1a 2539 PL_last_lop = PL_last_uni = Nullch;
3280af22 2540 PL_doextract = FALSE;
a0d0e21e 2541 }
4e553d73 2542 }
463ee0b2 2543 incline(s);
3280af22
NIS
2544 } while (PL_doextract);
2545 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2546 if (PERLDB_LINE && PL_curstash != PL_debstash) {
79072805 2547 SV *sv = NEWSV(85,0);
a687059c 2548
93a17b20 2549 sv_upgrade(sv, SVt_PVMG);
3280af22 2550 sv_setsv(sv,PL_linestr);
57843af0 2551 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
a687059c 2552 }
3280af22 2553 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
207e3d1a 2554 PL_last_lop = PL_last_uni = Nullch;
57843af0 2555 if (CopLINE(PL_curcop) == 1) {
3280af22 2556 while (s < PL_bufend && isSPACE(*s))
79072805 2557 s++;
a0d0e21e 2558 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
79072805 2559 s++;
44a8e56a 2560 d = Nullch;
3280af22 2561 if (!PL_in_eval) {
44a8e56a
PP
2562 if (*s == '#' && *(s+1) == '!')
2563 d = s + 2;
2564#ifdef ALTERNATE_SHEBANG
2565 else {
2566 static char as[] = ALTERNATE_SHEBANG;
2567 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2568 d = s + (sizeof(as) - 1);
2569 }
2570#endif /* ALTERNATE_SHEBANG */
2571 }
2572 if (d) {
b8378b72 2573 char *ipath;
774d564b 2574 char *ipathend;
b8378b72 2575
774d564b 2576 while (isSPACE(*d))
b8378b72
CS
2577 d++;
2578 ipath = d;
774d564b
PP
2579 while (*d && !isSPACE(*d))
2580 d++;
2581 ipathend = d;
2582
2583#ifdef ARG_ZERO_IS_SCRIPT
2584 if (ipathend > ipath) {
2585 /*
2586 * HP-UX (at least) sets argv[0] to the script name,
2587 * which makes $^X incorrect. And Digital UNIX and Linux,
2588 * at least, set argv[0] to the basename of the Perl
2589 * interpreter. So, having found "#!", we'll set it right.
2590 */
2591 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2592 assert(SvPOK(x) || SvGMAGICAL(x));
cc49e20b 2593 if (sv_eq(x, CopFILESV(PL_curcop))) {
774d564b 2594 sv_setpvn(x, ipath, ipathend - ipath);
9607fc9c
PP
2595 SvSETMAGIC(x);
2596 }
774d564b 2597 TAINT_NOT; /* $^X is always tainted, but that's OK */
8ebc5c01 2598 }
774d564b 2599#endif /* ARG_ZERO_IS_SCRIPT */
b8378b72
CS
2600
2601 /*
2602 * Look for options.
2603 */
748a9306 2604 d = instr(s,"perl -");
84e30d1a 2605 if (!d) {
748a9306 2606 d = instr(s,"perl");
84e30d1a
GS
2607#if defined(DOSISH)
2608 /* avoid getting into infinite loops when shebang
2609 * line contains "Perl" rather than "perl" */
2610 if (!d) {
2611 for (d = ipathend-4; d >= ipath; --d) {
2612 if ((*d == 'p' || *d == 'P')
2613 && !ibcmp(d, "perl", 4))
2614 {
2615 break;
2616 }
2617 }
2618 if (d < ipath)
2619 d = Nullch;
2620 }
2621#endif
2622 }
44a8e56a
PP
2623#ifdef ALTERNATE_SHEBANG
2624 /*
2625 * If the ALTERNATE_SHEBANG on this system starts with a
2626 * character that can be part of a Perl expression, then if
2627 * we see it but not "perl", we're probably looking at the
2628 * start of Perl code, not a request to hand off to some
2629 * other interpreter. Similarly, if "perl" is there, but
2630 * not in the first 'word' of the line, we assume the line
2631 * contains the start of the Perl program.
44a8e56a
PP
2632 */
2633 if (d && *s != '#') {
774d564b 2634 char *c = ipath;
44a8e56a
PP
2635 while (*c && !strchr("; \t\r\n\f\v#", *c))
2636 c++;
2637 if (c < d)
2638 d = Nullch; /* "perl" not in first word; ignore */
2639 else
2640 *s = '#'; /* Don't try to parse shebang line */
2641 }
774d564b 2642#endif /* ALTERNATE_SHEBANG */
bf4acbe4 2643#ifndef MACOS_TRADITIONAL
748a9306 2644 if (!d &&
44a8e56a 2645 *s == '#' &&
774d564b 2646 ipathend > ipath &&
3280af22 2647 !PL_minus_c &&
748a9306 2648 !instr(s,"indir") &&
3280af22 2649 instr(PL_origargv[0],"perl"))
748a9306 2650 {
9f68db38 2651 char **newargv;
9f68db38 2652
774d564b
PP
2653 *ipathend = '\0';
2654 s = ipathend + 1;
3280af22 2655 while (s < PL_bufend && isSPACE(*s))
9f68db38 2656 s++;
3280af22
NIS
2657 if (s < PL_bufend) {
2658 Newz(899,newargv,PL_origargc+3,char*);
9f68db38 2659 newargv[1] = s;
3280af22 2660 while (s < PL_bufend && !isSPACE(*s))
9f68db38
LW
2661 s++;
2662 *s = '\0';
3280af22 2663 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
9f68db38
LW
2664 }
2665 else
3280af22 2666 newargv = PL_origargv;
774d564b 2667 newargv[0] = ipath;
b4748376 2668 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
cea2e8a9 2669 Perl_croak(aTHX_ "Can't exec %s", ipath);
9f68db38 2670 }
bf4acbe4 2671#endif
748a9306 2672 if (d) {
3280af22
NIS
2673 U32 oldpdb = PL_perldb;
2674 bool oldn = PL_minus_n;
2675 bool oldp = PL_minus_p;
748a9306
LW
2676
2677 while (*d && !isSPACE(*d)) d++;
bf4acbe4 2678 while (SPACE_OR_TAB(*d)) d++;
748a9306
LW
2679
2680 if (*d++ == '-') {
8cc95fdb
PP
2681 do {
2682 if (*d == 'M' || *d == 'm') {
2683 char *m = d;
2684 while (*d && !isSPACE(*d)) d++;
cea2e8a9 2685 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
8cc95fdb
PP
2686 (int)(d - m), m);
2687 }
2688 d = moreswitches(d);
2689 } while (d);
155aba94
GS
2690 if ((PERLDB_LINE && !oldpdb) ||
2691 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
b084f20b
PP
2692 /* if we have already added "LINE: while (<>) {",
2693 we must not do it again */
748a9306 2694 {
3280af22
NIS
2695 sv_setpv(PL_linestr, "");
2696 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2697 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
207e3d1a 2698 PL_last_lop = PL_last_uni = Nullch;
3280af22 2699 PL_preambled = FALSE;
84902520 2700 if (PERLDB_LINE)
3280af22 2701 (void)gv_fetchfile(PL_origfilename);
748a9306
LW
2702 goto retry;
2703 }
a0d0e21e 2704 }
79072805 2705 }
9f68db38 2706 }
79072805 2707 }
3280af22
NIS
2708 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2709 PL_bufptr = s;
2710 PL_lex_state = LEX_FORMLINE;
cea2e8a9 2711 return yylex();
ae986130 2712 }
378cc40b 2713 goto retry;
4fdae800 2714 case '\r':
6a27c188 2715#ifdef PERL_STRICT_CR
cea2e8a9 2716 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
4e553d73 2717 Perl_croak(aTHX_
cc507455 2718 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
a868473f 2719#endif
4fdae800 2720 case ' ': case '\t': case '\f': case 013:
bf4acbe4
GS
2721#ifdef MACOS_TRADITIONAL
2722 case '\312':
2723#endif
378cc40b
LW
2724 s++;
2725 goto retry;
378cc40b 2726 case '#':
e929a76b 2727 case '\n':
3280af22 2728 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
df0deb90
GS
2729 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2730 /* handle eval qq[#line 1 "foo"\n ...] */
2731 CopLINE_dec(PL_curcop);
2732 incline(s);
2733 }
3280af22 2734 d = PL_bufend;
a687059c 2735 while (s < d && *s != '\n')
378cc40b 2736 s++;
0f85fab0 2737 if (s < d)
378cc40b 2738 s++;
78c267c1 2739 else if (s > d) /* Found by Ilya: feed random input to Perl. */
a8406387 2740 Perl_croak(aTHX_ "panic: input overflow");
463ee0b2 2741 incline(s);
3280af22
NIS
2742 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2743 PL_bufptr = s;
2744 PL_lex_state = LEX_FORMLINE;
cea2e8a9 2745 return yylex();
a687059c 2746 }
378cc40b 2747 }
a687059c 2748 else {
378cc40b 2749 *s = '\0';
3280af22 2750 PL_bufend = s;
a687059c 2751 }
378cc40b
LW
2752 goto retry;
2753 case '-':
79072805 2754 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
e5edeb50
JH
2755 I32 ftst = 0;
2756
378cc40b 2757 s++;
3280af22 2758 PL_bufptr = s;
748a9306
LW
2759 tmp = *s++;
2760
bf4acbe4 2761 while (s < PL_bufend && SPACE_OR_TAB(*s))
748a9306
LW
2762 s++;
2763
2764 if (strnEQ(s,"=>",2)) {
3280af22 2765 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
4e553d73 2766 DEBUG_T( { PerlIO_printf(Perl_debug_log,
607df283 2767 "### Saw unary minus before =>, forcing word '%s'\n", s);
5f80b19c 2768 } );
748a9306
LW
2769 OPERATOR('-'); /* unary minus */
2770 }
3280af22 2771 PL_last_uni = PL_oldbufptr;
748a9306 2772 switch (tmp) {
e5edeb50
JH
2773 case 'r': ftst = OP_FTEREAD; break;
2774 case 'w': ftst = OP_FTEWRITE; break;
2775 case 'x': ftst = OP_FTEEXEC; break;
2776 case 'o': ftst = OP_FTEOWNED; break;
2777 case 'R': ftst = OP_FTRREAD; break;
2778 case 'W': ftst = OP_FTRWRITE; break;
2779 case 'X': ftst = OP_FTREXEC; break;
2780 case 'O': ftst = OP_FTROWNED; break;
2781 case 'e': ftst = OP_FTIS; break;
2782 case 'z': ftst = OP_FTZERO; break;
2783 case 's': ftst = OP_FTSIZE; break;
2784 case 'f': ftst = OP_FTFILE; break;
2785 case 'd': ftst = OP_FTDIR; break;
2786 case 'l': ftst = OP_FTLINK; break;
2787 case 'p': ftst = OP_FTPIPE; break;
2788 case 'S': ftst = OP_FTSOCK; break;
2789 case 'u': ftst = OP_FTSUID; break;
2790 case 'g': ftst = OP_FTSGID; break;
2791 case 'k': ftst = OP_FTSVTX; break;
2792 case 'b': ftst = OP_FTBLK; break;
2793 case 'c': ftst = OP_FTCHR; break;
2794 case 't': ftst = OP_FTTTY; break;
2795 case 'T': ftst = OP_FTTEXT; break;
2796 case 'B': ftst = OP_FTBINARY; break;
2797 case 'M': case 'A': case 'C':
2798 gv_fetchpv("\024",TRUE, SVt_PV);
2799 switch (tmp) {
2800 case 'M': ftst = OP_FTMTIME; break;
2801 case 'A': ftst = OP_FTATIME; break;
2802 case 'C': ftst = OP_FTCTIME; break;
2803 default: break;
2804 }
2805 break;
378cc40b 2806 default:
378cc40b
LW
2807 break;
2808 }
e5edeb50
JH
2809 if (ftst) {
2810 PL_last_lop_op = ftst;
4e553d73 2811 DEBUG_T( { PerlIO_printf(Perl_debug_log,
0844c848 2812 "### Saw file test %c\n", (int)ftst);
5f80b19c 2813 } );
e5edeb50
JH
2814 FTST(ftst);
2815 }
2816 else {
2817 /* Assume it was a minus followed by a one-letter named
2818 * subroutine call (or a -bareword), then. */
95c31fe3 2819 DEBUG_T( { PerlIO_printf(Perl_debug_log,
0844c848
RB
2820 "### %c looked like a file test but was not\n",
2821 (int)ftst);
5f80b19c 2822 } );
e5edeb50
JH
2823 s -= 2;
2824 }
378cc40b 2825 }
a687059c
LW
2826 tmp = *s++;
2827 if (*s == tmp) {
2828 s++;
3280af22 2829 if (PL_expect == XOPERATOR)
79072805
LW
2830 TERM(POSTDEC);
2831 else
2832 OPERATOR(PREDEC);
2833 }
2834 else if (*s == '>') {
2835 s++;
2836 s = skipspace(s);
7e2040f0 2837 if (isIDFIRST_lazy_if(s,UTF)) {
a0d0e21e 2838 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
463ee0b2 2839 TOKEN(ARROW);
79072805 2840 }
748a9306
LW
2841 else if (*s == '$')
2842 OPERATOR(ARROW);
463ee0b2 2843 else
748a9306 2844 TERM(ARROW);
a687059c 2845 }
3280af22 2846 if (PL_expect == XOPERATOR)
79072805
LW
2847 Aop(OP_SUBTRACT);
2848 else {
3280af22 2849 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2f3197b3 2850 check_uni();
79072805 2851 OPERATOR('-'); /* unary minus */
2f3197b3 2852 }
79072805 2853
378cc40b 2854 case '+':
a687059c
LW
2855 tmp = *s++;
2856 if (*s == tmp) {
378cc40b 2857 s++;
3280af22 2858 if (PL_expect == XOPERATOR)
79072805
LW
2859 TERM(POSTINC);
2860 else
2861 OPERATOR(PREINC);
378cc40b 2862 }
3280af22 2863 if (PL_expect == XOPERATOR)
79072805
LW
2864 Aop(OP_ADD);
2865 else {
3280af22 2866 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2f3197b3 2867 check_uni();
a687059c 2868 OPERATOR('+');
2f3197b3 2869 }
a687059c 2870
378cc40b 2871 case '*':
3280af22
NIS
2872 if (PL_expect != XOPERATOR) {
2873 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2874 PL_expect = XOPERATOR;
2875 force_ident(PL_tokenbuf, '*');
2876 if (!*PL_tokenbuf)
a0d0e21e 2877 PREREF('*');
79072805 2878 TERM('*');
a687059c 2879 }
79072805
LW
2880 s++;
2881 if (*s == '*') {
a687059c 2882 s++;
79072805 2883 PWop(OP_POW);
a687059c 2884 }
79072805
LW
2885 Mop(OP_MULTIPLY);
2886
378cc40b 2887 case '%':
3280af22 2888 if (PL_expect == XOPERATOR) {
bbce6d69
PP
2889 ++s;
2890 Mop(OP_MODULO);
a687059c 2891 }
3280af22
NIS
2892 PL_tokenbuf[0] = '%';
2893 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2894 if (!PL_tokenbuf[1]) {
2895 if (s == PL_bufend)
bbce6d69
PP
2896 yyerror("Final % should be \\% or %name");
2897 PREREF('%');
a687059c 2898 }
3280af22 2899 PL_pending_ident = '%';
bbce6d69 2900 TERM('%');
a687059c 2901
378cc40b 2902 case '^':
79072805 2903 s++;
a0d0e21e 2904 BOop(OP_BIT_XOR);
79072805 2905 case '[':
3280af22 2906 PL_lex_brackets++;
79072805 2907 /* FALL THROUGH */
378cc40b 2908 case '~':
378cc40b 2909 case ',':
378cc40b
LW
2910 tmp = *s++;
2911 OPERATOR(tmp);
a0d0e21e
LW
2912 case ':':
2913 if (s[1] == ':') {
2914 len = 0;
2915 goto just_a_word;
2916 }
2917 s++;
09bef843
SB
2918 switch (PL_expect) {
2919 OP *attrs;
2920 case XOPERATOR:
2921 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2922 break;
2923 PL_bufptr = s; /* update in case we back off */
2924 goto grabattrs;
2925 case XATTRBLOCK:
2926 PL_expect = XBLOCK;
2927 goto grabattrs;
2928 case XATTRTERM:
2929 PL_expect = XTERMBLOCK;
2930 grabattrs:
2931 s = skipspace(s);
2932 attrs = Nullop;
7e2040f0 2933 while (isIDFIRST_lazy_if(s,UTF)) {
09bef843 2934 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
f9829d6b
GS
2935 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2936 if (tmp < 0) tmp = -tmp;
2937 switch (tmp) {
2938 case KEY_or:
2939 case KEY_and:
2940 case KEY_for:
2941 case KEY_unless:
2942 case KEY_if:
2943 case KEY_while:
2944 case KEY_until:
2945 goto got_attrs;
2946 default:
2947 break;
2948 }
2949 }
09bef843
SB
2950 if (*d == '(') {
2951 d = scan_str(d,TRUE,TRUE);
2952 if (!d) {
09bef843
SB
2953 /* MUST advance bufptr here to avoid bogus
2954 "at end of line" context messages from yyerror().
2955 */
2956 PL_bufptr = s + len;
2957 yyerror("Unterminated attribute parameter in attribute list");
2958 if (attrs)
2959 op_free(attrs);
2960 return 0; /* EOF indicator */
2961 }
2962 }
2963 if (PL_lex_stuff) {
2964 SV *sv = newSVpvn(s, len);
2965 sv_catsv(sv, PL_lex_stuff);
2966 attrs = append_elem(OP_LIST, attrs,
2967 newSVOP(OP_CONST, 0, sv));
2968 SvREFCNT_dec(PL_lex_stuff);
2969 PL_lex_stuff = Nullsv;
2970 }
2971 else {
78f9721b
SM
2972 if (!PL_in_my && len == 6 && strnEQ(s, "lvalue", len))
2973 CvLVALUE_on(PL_compcv);
2974 else if (!PL_in_my && len == 6 && strnEQ(s, "locked", len))
2975 CvLOCKED_on(PL_compcv);
2976 else if (!PL_in_my && len == 6 && strnEQ(s, "method", len))
2977 CvMETHOD_on(PL_compcv);
87ecf892 2978#ifdef USE_ITHREADS
c8a3bf85 2979 else if (PL_in_my == KEY_our && len == 6 && strnEQ(s, "unique", len))
7fb37951 2980 GvUNIQUE_on(cGVOPx_gv(yylval.opval));
87ecf892 2981#endif
78f9721b
SM
2982 /* After we've set the flags, it could be argued that
2983 we don't need to do the attributes.pm-based setting
2984 process, and shouldn't bother appending recognized
2985 flags. To experiment with that, uncomment the
2986 following "else": */
0256094b 2987 else
78f9721b
SM
2988 attrs = append_elem(OP_LIST, attrs,
2989 newSVOP(OP_CONST, 0,
2990 newSVpvn(s, len)));
09bef843
SB
2991 }
2992 s = skipspace(d);
0120eecf 2993 if (*s == ':' && s[1] != ':')
09bef843 2994 s = skipspace(s+1);
0120eecf
GS
2995 else if (s == d)
2996 break; /* require real whitespace or :'s */
09bef843 2997 }
f9829d6b
GS
2998 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
2999 if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) {
09bef843
SB
3000 char q = ((*s == '\'') ? '"' : '\'');
3001 /* If here for an expression, and parsed no attrs, back off. */
3002 if (tmp == '=' && !attrs) {
3003 s = PL_bufptr;
3004 break;
3005 }
3006 /* MUST advance bufptr here to avoid bogus "at end of line"
3007 context messages from yyerror().
3008 */
3009 PL_bufptr = s;
3010 if (!*s)
3011 yyerror("Unterminated attribute list");
3012 else
3013 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
3014 q, *s, q));
3015 if (attrs)
3016 op_free(attrs);
3017 OPERATOR(':');
3018 }
f9829d6b 3019 got_attrs:
09bef843
SB
3020 if (attrs) {
3021 PL_nextval[PL_nexttoke].opval = attrs;
3022 force_next(THING);
3023 }
3024 TOKEN(COLONATTR);
3025 }
a0d0e21e 3026 OPERATOR(':');
8990e307
LW
3027 case '(':
3028 s++;
3280af22
NIS
3029 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
3030 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
a0d0e21e 3031 else
3280af22 3032 PL_expect = XTERM;
a0d0e21e 3033 TOKEN('(');
378cc40b 3034 case ';':
f4dd75d9 3035 CLINE;
378cc40b
LW
3036 tmp = *s++;
3037 OPERATOR(tmp);
3038 case ')':
378cc40b 3039 tmp = *s++;
16d20bd9
AD
3040 s = skipspace(s);
3041 if (*s == '{')
3042 PREBLOCK(tmp);
378cc40b 3043 TERM(tmp);
79072805
LW
3044 case ']':
3045 s++;
3280af22 3046 if (PL_lex_brackets <= 0)
d98d5fff 3047 yyerror("Unmatched right square bracket");
463ee0b2 3048 else
3280af22
NIS
3049 --PL_lex_brackets;
3050 if (PL_lex_state == LEX_INTERPNORMAL) {
3051 if (PL_lex_brackets == 0) {
a0d0e21e 3052 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
3280af22 3053 PL_lex_state = LEX_INTERPEND;
79072805
LW
3054 }
3055 }
4633a7c4 3056 TERM(']');
79072805
LW
3057 case '{':
3058 leftbracket:
79072805 3059 s++;
3280af22
NIS
3060 if (PL_lex_brackets > 100) {
3061 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
3062 if (newlb != PL_lex_brackstack) {
8990e307 3063 SAVEFREEPV(newlb);
3280af22 3064 PL_lex_brackstack = newlb;
8990e307
LW
3065 }
3066 }
3280af22 3067 switch (PL_expect) {
a0d0e21e 3068 case XTERM:
3280af22 3069 if (PL_lex_formbrack) {
a0d0e21e
LW
3070 s--;
3071 PRETERMBLOCK(DO);
3072 }
3280af22
NIS
3073 if (PL_oldoldbufptr == PL_last_lop)
3074 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
a0d0e21e 3075 else
3280af22 3076 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
79072805 3077 OPERATOR(HASHBRACK);
a0d0e21e 3078 case XOPERATOR:
bf4acbe4 3079 while (s < PL_bufend && SPACE_OR_TAB(*s))
748a9306 3080 s++;
44a8e56a 3081 d = s;
3280af22
NIS
3082 PL_tokenbuf[0] = '\0';
3083 if (d < PL_bufend && *d == '-') {
3084 PL_tokenbuf[0] = '-';
44a8e56a 3085 d++;
bf4acbe4 3086 while (d < PL_bufend && SPACE_OR_TAB(*d))
44a8e56a
PP
3087 d++;
3088 }
7e2040f0 3089 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3280af22 3090 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
8903cb82 3091 FALSE, &len);
bf4acbe4 3092 while (d < PL_bufend && SPACE_OR_TAB(*d))
748a9306
LW
3093 d++;
3094 if (*d == '}') {
3280af22 3095 char minus = (PL_tokenbuf[0] == '-');
44a8e56a
PP
3096 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3097 if (minus)
3098 force_next('-');
748a9306
LW
3099 }
3100 }
3101 /* FALL THROUGH */
09bef843 3102 case XATTRBLOCK:
748a9306 3103 case XBLOCK:
3280af22
NIS
3104 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3105 PL_expect = XSTATE;
a0d0e21e 3106 break;
09bef843 3107 case XATTRTERM:
a0d0e21e 3108 case XTERMBLOCK:
3280af22
NIS
3109 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3110 PL_expect = XSTATE;
a0d0e21e
LW
3111 break;
3112 default: {
3113 char *t;
3280af22
NIS
3114 if (PL_oldoldbufptr == PL_last_lop)
3115 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
a0d0e21e 3116 else
3280af22 3117 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
a0d0e21e 3118 s = skipspace(s);
8452ff4b
SB
3119 if (*s == '}') {
3120 if (PL_expect == XREF && PL_lex_state == LEX_INTERPNORMAL) {
3121 PL_expect = XTERM;
3122 /* This hack is to get the ${} in the message. */
3123 PL_bufptr = s+1;
3124 yyerror("syntax error");
3125 break;
3126 }
a0d0e21e 3127 OPERATOR(HASHBRACK);
8452ff4b 3128 }
b8a4b1be
GS
3129 /* This hack serves to disambiguate a pair of curlies
3130 * as being a block or an anon hash. Normally, expectation
3131 * determines that, but in cases where we're not in a
3132 * position to expect anything in particular (like inside
3133 * eval"") we have to resolve the ambiguity. This code
3134 * covers the case where the first term in the curlies is a
3135 * quoted string. Most other cases need to be explicitly
3136 * disambiguated by prepending a `+' before the opening
3137 * curly in order to force resolution as an anon hash.
3138 *
3139 * XXX should probably propagate the outer expectation
3140 * into eval"" to rely less on this hack, but that could
3141 * potentially break current behavior of eval"".
3142 * GSAR 97-07-21
3143 */
3144 t = s;
3145 if (*s == '\'' || *s == '"' || *s == '`') {
3146 /* common case: get past first string, handling escapes */
3280af22 3147 for (t++; t < PL_bufend && *t != *s;)
b8a4b1be
GS
3148 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3149 t++;
3150 t++;
a0d0e21e 3151 }
b8a4b1be 3152 else if (*s == 'q') {
3280af22 3153 if (++t < PL_bufend
b8a4b1be 3154 && (!isALNUM(*t)
3280af22 3155 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
0505442f
GS
3156 && !isALNUM(*t))))
3157 {
b8a4b1be
GS
3158 char *tmps;
3159 char open, close, term;
3160 I32 brackets = 1;
3161
3280af22 3162 while (t < PL_bufend && isSPACE(*t))
b8a4b1be
GS
3163 t++;
3164 term = *t;
3165 open = term;
3166 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3167 term = tmps[5];
3168 close = term;
3169 if (open == close)
3280af22
NIS
3170 for (t++; t < PL_bufend; t++) {
3171 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
b8a4b1be 3172 t++;
6d07e5e9 3173 else if (*t == open)
b8a4b1be
GS
3174 break;
3175 }
3176 else
3280af22
NIS
3177 for (t++; t < PL_bufend; t++) {
3178 if (*t == '\\' && t+1 < PL_bufend)
b8a4b1be 3179 t++;
6d07e5e9 3180 else if (*t == close && --brackets <= 0)
b8a4b1be
GS
3181 break;
3182 else if (*t == open)
3183 brackets++;
3184 }
3185 }
3186 t++;
a0d0e21e 3187 }
7e2040f0 3188 else if (isALNUM_lazy_if(t,UTF)) {
0505442f 3189 t += UTF8SKIP(t);
7e2040f0 3190 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
0505442f 3191 t += UTF8SKIP(t);
a0d0e21e 3192 }
3280af22 3193 while (t < PL_bufend && isSPACE(*t))
a0d0e21e 3194 t++;
b8a4b1be
GS
3195 /* if comma follows first term, call it an anon hash */
3196 /* XXX it could be a comma expression with loop modifiers */
3280af22 3197 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
b8a4b1be 3198 || (*t == '=' && t[1] == '>')))
a0d0e21e 3199 OPERATOR(HASHBRACK);
3280af22 3200 if (PL_expect == XREF)
4e4e412b 3201 PL_expect = XTERM;
a0d0e21e 3202 else {
3280af22
NIS
3203 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3204 PL_expect = XSTATE;
a0d0e21e 3205 }
8990e307 3206 }
a0d0e21e 3207 break;
463ee0b2 3208 }
57843af0 3209 yylval.ival = CopLINE(PL_curcop);
79072805 3210 if (isSPACE(*s) || *s == '#')
3280af22 3211 PL_copline = NOLINE; /* invalidate current command line number */
79072805 3212 TOKEN('{');
378cc40b 3213 case '}':
79072805
LW
3214 rightbracket:
3215 s++;
3280af22 3216 if (PL_lex_brackets <= 0)
d98d5fff 3217 yyerror("Unmatched right curly bracket");
463ee0b2 3218 else
3280af22 3219 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
c2e66d9e 3220 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3280af22
NIS
3221 PL_lex_formbrack = 0;
3222 if (PL_lex_state == LEX_INTERPNORMAL) {
3223 if (PL_lex_brackets == 0) {
9059aa12
LW
3224 if (PL_expect & XFAKEBRACK) {
3225 PL_expect &= XENUMMASK;
3280af22
NIS
3226 PL_lex_state = LEX_INTERPEND;
3227 PL_bufptr = s;
cea2e8a9 3228 return yylex(); /* ignore fake brackets */
79072805 3229 }
fa83b5b6 3230 if (*s == '-' && s[1] == '>')
3280af22 3231 PL_lex_state = LEX_INTERPENDMAYBE;
fa83b5b6 3232 else if (*s != '[' && *s != '{')
3280af22 3233 PL_lex_state = LEX_INTERPEND;
79072805
LW
3234 }
3235 }
9059aa12
LW
3236 if (PL_expect & XFAKEBRACK) {
3237 PL_expect &= XENUMMASK;
3280af22 3238 PL_bufptr = s;
cea2e8a9 3239 return yylex(); /* ignore fake brackets */
748a9306 3240 }
79072805
LW
3241 force_next('}');
3242 TOKEN(';');
378cc40b
LW
3243 case '&':
3244 s++;
3245 tmp = *s++;
3246 if (tmp == '&')
a0d0e21e 3247 AOPERATOR(ANDAND);
378cc40b 3248 s--;
3280af22 3249 if (PL_expect == XOPERATOR) {
7e2040f0
GS
3250 if (ckWARN(WARN_SEMICOLON)
3251 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3252 {
57843af0 3253 CopLINE_dec(PL_curcop);
cea2e8a9 3254 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
57843af0 3255 CopLINE_inc(PL_curcop);
463ee0b2 3256 }
79072805 3257 BAop(OP_BIT_AND);
463ee0b2 3258 }
79072805 3259
3280af22
NIS
3260 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3261 if (*PL_tokenbuf) {
3262 PL_expect = XOPERATOR;
3263 force_ident(PL_tokenbuf, '&');
463ee0b2 3264 }
79072805
LW
3265 else
3266 PREREF('&');
c07a80fd 3267 yylval.ival = (OPpENTERSUB_AMPER<<8);
79072805
LW
3268 TERM('&');
3269
378cc40b
LW
3270 case '|':
3271 s++;
3272 tmp = *s++;
3273 if (tmp == '|')
a0d0e21e 3274 AOPERATOR(OROR);
378cc40b 3275 s--;
79072805 3276 BOop(OP_BIT_OR);
378cc40b
LW
3277 case '=':
3278 s++;
3279 tmp = *s++;
3280 if (tmp == '=')
79072805
LW
3281 Eop(OP_EQ);
3282 if (tmp == '>')
3283 OPERATOR(',');
378cc40b 3284 if (tmp == '~')
79072805 3285 PMop(OP_MATCH);
599cee73 3286 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
cea2e8a9 3287 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
378cc40b 3288 s--;
3280af22
NIS
3289 if (PL_expect == XSTATE && isALPHA(tmp) &&
3290 (s == PL_linestart+1 || s[-2] == '\n') )
748a9306 3291 {
3280af22
NIS
3292 if (PL_in_eval && !PL_rsfp) {
3293 d = PL_bufend;
a5f75d66
AD
3294 while (s < d) {
3295 if (*s++ == '\n') {
3296 incline(s);
3297 if (strnEQ(s,"=cut",4)) {
3298 s = strchr(s,'\n');
3299 if (s)
3300 s++;
3301 else
3302 s = d;
3303 incline(s);
3304 goto retry;
3305 }
3306 }
3307 }
3308 goto retry;
3309 }
3280af22
NIS
3310 s = PL_bufend;
3311 PL_doextract = TRUE;
a0d0e21e
LW
3312 goto retry;
3313 }
3280af22 3314 if (PL_lex_brackets < PL_lex_formbrack) {
a0d0e21e 3315 char *t;
51882d45 3316#ifdef PERL_STRICT_CR
bf4acbe4 3317 for (t = s; SPACE_OR_TAB(*t); t++) ;
51882d45 3318#else
bf4acbe4 3319 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
51882d45 3320#endif
a0d0e21e
LW
3321 if (*t == '\n' || *t == '#') {
3322 s--;
3280af22 3323 PL_expect = XBLOCK;
a0d0e21e
LW
3324 goto leftbracket;
3325 }
79072805 3326 }
a0d0e21e
LW
3327 yylval.ival = 0;
3328 OPERATOR(ASSIGNOP);
378cc40b
LW
3329 case '!':
3330 s++;
3331 tmp = *s++;
3332 if (tmp == '=')
79072805 3333 Eop(OP_NE);
378cc40b 3334 if (tmp == '~')
79072805 3335 PMop(OP_NOT);
378cc40b
LW
3336 s--;
3337 OPERATOR('!');
3338 case '<':
3280af22 3339 if (PL_expect != XOPERATOR) {
93a17b20 3340 if (s[1] != '<' && !strchr(s,'>'))
2f3197b3 3341 check_uni();
79072805
LW
3342 if (s[1] == '<')
3343 s = scan_heredoc(s);
3344 else
3345 s = scan_inputsymbol(s);
3346 TERM(sublex_start());
378cc40b
LW
3347 }
3348 s++;
3349 tmp = *s++;
3350 if (tmp == '<')
79072805 3351 SHop(OP_LEFT_SHIFT);
395c3793
LW
3352 if (tmp == '=') {
3353 tmp = *s++;
3354 if (tmp == '>')
79072805 3355 Eop(OP_NCMP);
395c3793 3356 s--;
79072805 3357 Rop(OP_LE);
395c3793 3358 }
378cc40b 3359 s--;
79072805 3360 Rop(OP_LT);
378cc40b
LW
3361 case '>':
3362 s++;
3363 tmp = *s++;
3364 if (tmp == '>')
79072805 3365 SHop(OP_RIGHT_SHIFT);
378cc40b 3366 if (tmp == '=')
79072805 3367 Rop(OP_GE);
378cc40b 3368 s--;
79072805 3369 Rop(OP_GT);
378cc40b
LW
3370
3371 case '$':
bbce6d69
PP
3372 CLINE;
3373
3280af22
NIS
3374 if (PL_expect == XOPERATOR) {
3375 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3376 PL_expect = XTERM;
a0d0e21e 3377 depcom();
bbce6d69 3378 return ','; /* grandfather non-comma-format format */
a0d0e21e 3379 }
8990e307 3380 }
a0d0e21e 3381
7e2040f0 3382 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3280af22 3383 PL_tokenbuf[0] = '@';
376b8730
SM
3384 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3385 sizeof PL_tokenbuf - 1, FALSE);
3386 if (PL_expect == XOPERATOR)
3387 no_op("Array length", s);
3280af22 3388 if (!PL_tokenbuf[1])
a0d0e21e 3389 PREREF(DOLSHARP);
3280af22
NIS
3390 PL_expect = XOPERATOR;
3391 PL_pending_ident = '#';
463ee0b2 3392 TOKEN(DOLSHARP);
79072805 3393 }
bbce6d69 3394
3280af22 3395 PL_tokenbuf[0] = '$';
376b8730
SM
3396 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3397 sizeof PL_tokenbuf - 1, FALSE);
3398 if (PL_expect == XOPERATOR)
3399 no_op("Scalar", s);
3280af22
NIS
3400 if (!PL_tokenbuf[1]) {
3401 if (s == PL_bufend)
bbce6d69
PP
3402 yyerror("Final $ should be \\$ or $name");
3403 PREREF('$');
8990e307 3404 }
a0d0e21e 3405
bbce6d69 3406 /* This kludge not intended to be bulletproof. */
3280af22 3407 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
bbce6d69 3408 yylval.opval = newSVOP(OP_CONST, 0,
b448e4fe 3409 newSViv(PL_compiling.cop_arybase));
bbce6d69
PP
3410 yylval.opval->op_private = OPpCONST_ARYBASE;
3411 TERM(THING);
3412 }
3413
ff68c719 3414 d = s;
69d2bceb 3415 tmp = (I32)*s;
3280af22 3416 if (PL_lex_state == LEX_NORMAL)
ff68c719
PP
3417 s = skipspace(s);
3418
3280af22 3419 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
bbce6d69
PP
3420 char *t;
3421 if (*s == '[') {
3280af22 3422 PL_tokenbuf[0] = '@';
599cee73 3423 if (ckWARN(WARN_SYNTAX)) {
bbce6d69 3424 for(t = s + 1;
7e2040f0 3425 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
bbce6d69 3426 t++) ;
a0d0e21e 3427 if (*t++ == ',') {
3280af22
NIS
3428 PL_bufptr = skipspace(PL_bufptr);
3429 while (t < PL_bufend && *t != ']')
bbce6d69 3430 t++;
cea2e8a9 3431 Perl_warner(aTHX_ WARN_SYNTAX,
599cee73
PM
3432 "Multidimensional syntax %.*s not supported",
3433 (t - PL_bufptr) + 1, PL_bufptr);
a0d0e21e
LW
3434 }
3435 }
bbce6d69
PP
3436 }
3437 else if (*s == '{') {
3280af22 3438 PL_tokenbuf[0] = '%';
599cee73 3439 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
bbce6d69
PP
3440 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3441 {
3280af22 3442 char tmpbuf[sizeof PL_tokenbuf];
a0d0e21e
LW
3443 STRLEN len;
3444 for (t++; isSPACE(*t); t++) ;
7e2040f0 3445 if (isIDFIRST_lazy_if(t,UTF)) {
8903cb82 3446 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
59a6d928 3447 for (; isSPACE(*t); t++) ;
864dbfa3 3448 if (*t == ';' && get_cv(tmpbuf, FALSE))
cea2e8a9 3449 Perl_warner(aTHX_ WARN_SYNTAX,
599cee73 3450 "You need to quote \"%s\"", tmpbuf);
748a9306 3451 }
93a17b20
LW
3452 }
3453 }
2f3197b3 3454 }
bbce6d69 3455
3280af22 3456 PL_expect = XOPERATOR;
69d2bceb 3457 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3280af22
NIS
3458 bool islop = (PL_last_lop == PL_oldoldbufptr);
3459 if (!islop || PL_last_lop_op == OP_GREPSTART)
3460 PL_expect = XOPERATOR;
bbce6d69 3461 else if (strchr("$@\"'`q", *s))
3280af22 3462 PL_expect = XTERM; /* e.g. print $fh "foo" */
7e2040f0 3463 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3280af22 3464 PL_expect = XTERM; /* e.g. print $fh &sub */
7e2040f0 3465 else if (isIDFIRST_lazy_if(s,UTF)) {
3280af22 3466 char tmpbuf[sizeof PL_tokenbuf];
8903cb82 3467 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
155aba94 3468 if ((tmp = keyword(tmpbuf, len))) {
84902520
TB
3469 /* binary operators exclude handle interpretations */
3470 switch (tmp) {
3471 case -KEY_x:
3472 case -KEY_eq:
3473 case -KEY_ne:
3474 case -KEY_gt:
3475 case -KEY_lt:
3476 case -KEY_ge:
3477 case -KEY_le:
3478 case -KEY_cmp:
3479 break;
3480 default:
3280af22 3481 PL_expect = XTERM; /* e.g. print $fh length() */
84902520
TB
3482 break;
3483 }
3484 }
68dc0745
PP
3485 else {
3486 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3487 if (gv && GvCVu(gv))
3280af22 3488 PL_expect = XTERM; /* e.g. print $fh subr() */
93a17b20 3489 }
93a17b20 3490 }
bbce6d69 3491 else if (isDIGIT(*s))
3280af22 3492 PL_expect = XTERM; /* e.g. print $fh 3 */
bbce6d69 3493 else if (*s == '.' && isDIGIT(s[1]))
3280af22 3494 PL_expect = XTERM; /* e.g. print $fh .3 */
e0587a03 3495 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3280af22 3496 PL_expect = XTERM; /* e.g. print $fh -1 */
e0587a03 3497 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3280af22 3498 PL_expect = XTERM; /* print $fh <<"EOF" */
bbce6d69 3499 }
3280af22 3500 PL_pending_ident = '$';
79072805 3501 TOKEN('$');
378cc40b
LW
3502
3503 case '@':
3280af22 3504 if (PL_expect == XOPERATOR)
bbce6d69 3505 no_op("Array", s);
3280af22
NIS
3506 PL_tokenbuf[0] = '@';
3507 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3508 if (!PL_tokenbuf[1]) {
3509 if (s == PL_bufend)
bbce6d69
PP
3510 yyerror("Final @ should be \\@ or @name");
3511 PREREF('@');</