3 * Copyright (c) 1991-2002, Larry Wall
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.
11 * "It all comes from here, the stench and the peril." --Frodo
15 * This file is the lexer for Perl. It's closely linked to the
18 * The main routine is yylex(), which returns the next token.
22 #define PERL_IN_TOKE_C
25 #define yychar PL_yychar
26 #define yylval PL_yylval
28 static char ident_too_long[] = "Identifier too long";
29 static char c_without_g[] = "Use of /c modifier is meaningless without /g";
30 static char c_in_subst[] = "Use of /c modifier is meaningless in s///";
32 static void restore_rsfp(pTHX_ void *f);
33 #ifndef PERL_NO_UTF16_FILTER
34 static I32 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen);
35 static I32 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen);
38 #define XFAKEBRACK 128
41 #ifdef USE_UTF8_SCRIPTS
42 # define UTF (!IN_BYTES)
44 # ifdef EBCDIC /* For now 'use utf8' does not affect tokenizer on EBCDIC */
45 # define UTF (PL_linestr && DO_UTF8(PL_linestr))
47 # define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || (PL_hints & HINT_UTF8))
51 /* In variables named $^X, these are the legal values for X.
52 * 1999-02-27 mjd-perl-patch@plover.com */
53 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
55 /* On MacOS, respect nonbreaking spaces */
56 #ifdef MACOS_TRADITIONAL
57 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
59 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
62 /* LEX_* are values for PL_lex_state, the state of the lexer.
63 * They are arranged oddly so that the guard on the switch statement
64 * can get by with a single comparison (if the compiler is smart enough).
67 /* #define LEX_NOTPARSING 11 is done in perl.h. */
70 #define LEX_INTERPNORMAL 9
71 #define LEX_INTERPCASEMOD 8
72 #define LEX_INTERPPUSH 7
73 #define LEX_INTERPSTART 6
74 #define LEX_INTERPEND 5
75 #define LEX_INTERPENDMAYBE 4
76 #define LEX_INTERPCONCAT 3
77 #define LEX_INTERPCONST 2
78 #define LEX_FORMLINE 1
79 #define LEX_KNOWNEXT 0
87 # define YYMAXLEVEL 100
89 YYSTYPE* yylval_pointer[YYMAXLEVEL];
90 int* yychar_pointer[YYMAXLEVEL];
94 # define yylval (*yylval_pointer[yyactlevel])
95 # define yychar (*yychar_pointer[yyactlevel])
96 # define PERL_YYLEX_PARAM yylval_pointer[yyactlevel],yychar_pointer[yyactlevel]
98 # define yylex() Perl_yylex_r(aTHX_ yylval_pointer[yyactlevel],yychar_pointer[yyactlevel])
101 #include "keywords.h"
103 /* CLINE is a macro that ensures PL_copline has a sane value */
108 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
111 * Convenience functions to return different tokens and prime the
112 * lexer for the next token. They all take an argument.
114 * TOKEN : generic token (used for '(', DOLSHARP, etc)
115 * OPERATOR : generic operator
116 * AOPERATOR : assignment operator
117 * PREBLOCK : beginning the block after an if, while, foreach, ...
118 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
119 * PREREF : *EXPR where EXPR is not a simple identifier
120 * TERM : expression term
121 * LOOPX : loop exiting command (goto, last, dump, etc)
122 * FTST : file test operator
123 * FUN0 : zero-argument function
124 * FUN1 : not used, except for not, which isn't a UNIOP
125 * BOop : bitwise or or xor
127 * SHop : shift operator
128 * PWop : power operator
129 * PMop : pattern-matching operator
130 * Aop : addition-level operator
131 * Mop : multiplication-level operator
132 * Eop : equality-testing operator
133 * Rop : relational operator <= != gt
135 * Also see LOP and lop() below.
138 /* Note that REPORT() and REPORT2() will be expressions that supply
139 * their own trailing comma, not suitable for statements as such. */
140 #ifdef DEBUGGING /* Serve -DT. */
141 # define REPORT(x,retval) tokereport(x,s,(int)retval),
142 # define REPORT2(x,retval) tokereport(x,s, yylval.ival),
144 # define REPORT(x,retval)
145 # define REPORT2(x,retval)
148 #define TOKEN(retval) return (REPORT2("token",retval) PL_bufptr = s,(int)retval)
149 #define OPERATOR(retval) return (REPORT2("operator",retval) PL_expect = XTERM, PL_bufptr = s,(int)retval)
150 #define AOPERATOR(retval) return ao((REPORT2("aop",retval) PL_expect = XTERM, PL_bufptr = s,(int)retval))
151 #define PREBLOCK(retval) return (REPORT2("preblock",retval) PL_expect = XBLOCK,PL_bufptr = s,(int)retval)
152 #define PRETERMBLOCK(retval) return (REPORT2("pretermblock",retval) PL_expect = XTERMBLOCK,PL_bufptr = s,(int)retval)
153 #define PREREF(retval) return (REPORT2("preref",retval) PL_expect = XREF,PL_bufptr = s,(int)retval)
154 #define TERM(retval) return (CLINE, REPORT2("term",retval) PL_expect = XOPERATOR, PL_bufptr = s,(int)retval)
155 #define LOOPX(f) return(yylval.ival=f, REPORT("loopx",f) PL_expect = XTERM,PL_bufptr = s,(int)LOOPEX)
156 #define FTST(f) return(yylval.ival=f, REPORT("ftst",f) PL_expect = XTERM,PL_bufptr = s,(int)UNIOP)
157 #define FUN0(f) return(yylval.ival = f, REPORT("fun0",f) PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC0)
158 #define FUN1(f) return(yylval.ival = f, REPORT("fun1",f) PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC1)
159 #define BOop(f) return ao((yylval.ival=f, REPORT("bitorop",f) PL_expect = XTERM,PL_bufptr = s,(int)BITOROP))
160 #define BAop(f) return ao((yylval.ival=f, REPORT("bitandop",f) PL_expect = XTERM,PL_bufptr = s,(int)BITANDOP))
161 #define SHop(f) return ao((yylval.ival=f, REPORT("shiftop",f) PL_expect = XTERM,PL_bufptr = s,(int)SHIFTOP))
162 #define PWop(f) return ao((yylval.ival=f, REPORT("powop",f) PL_expect = XTERM,PL_bufptr = s,(int)POWOP))
163 #define PMop(f) return(yylval.ival=f, REPORT("matchop",f) PL_expect = XTERM,PL_bufptr = s,(int)MATCHOP)
164 #define Aop(f) return ao((yylval.ival=f, REPORT("add",f) PL_expect = XTERM,PL_bufptr = s,(int)ADDOP))
165 #define Mop(f) return ao((yylval.ival=f, REPORT("mul",f) PL_expect = XTERM,PL_bufptr = s,(int)MULOP))
166 #define Eop(f) return(yylval.ival=f, REPORT("eq",f) PL_expect = XTERM,PL_bufptr = s,(int)EQOP)
167 #define Rop(f) return(yylval.ival=f, REPORT("rel",f) PL_expect = XTERM,PL_bufptr = s,(int)RELOP)
169 /* This bit of chicanery makes a unary function followed by
170 * a parenthesis into a function with one argument, highest precedence.
172 #define UNI(f) return(yylval.ival = f, \
176 PL_last_uni = PL_oldbufptr, \
177 PL_last_lop_op = f, \
178 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
180 #define UNIBRACK(f) return(yylval.ival = f, \
183 PL_last_uni = PL_oldbufptr, \
184 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
186 /* grandfather return to old style */
187 #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
192 S_tokereport(pTHX_ char *thing, char* s, I32 rv)
195 SV* report = newSVpv(thing, 0);
196 Perl_sv_catpvf(aTHX_ report, ":line %d:%"IVdf":", CopLINE(PL_curcop),
199 if (s - PL_bufptr > 0)
200 sv_catpvn(report, PL_bufptr, s - PL_bufptr);
202 if (PL_oldbufptr && *PL_oldbufptr)
203 sv_catpv(report, PL_tokenbuf);
205 PerlIO_printf(Perl_debug_log, "### %s\n", SvPV_nolen(report));
214 * This subroutine detects &&= and ||= and turns an ANDAND or OROR
215 * into an OP_ANDASSIGN or OP_ORASSIGN
219 S_ao(pTHX_ int toketype)
221 if (*PL_bufptr == '=') {
223 if (toketype == ANDAND)
224 yylval.ival = OP_ANDASSIGN;
225 else if (toketype == OROR)
226 yylval.ival = OP_ORASSIGN;
234 * When Perl expects an operator and finds something else, no_op
235 * prints the warning. It always prints "<something> found where
236 * operator expected. It prints "Missing semicolon on previous line?"
237 * if the surprise occurs at the start of the line. "do you need to
238 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
239 * where the compiler doesn't know if foo is a method call or a function.
240 * It prints "Missing operator before end of line" if there's nothing
241 * after the missing operator, or "... before <...>" if there is something
242 * after the missing operator.
246 S_no_op(pTHX_ char *what, char *s)
248 char *oldbp = PL_bufptr;
249 bool is_first = (PL_oldbufptr == PL_linestart);
255 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
257 Perl_warn(aTHX_ "\t(Missing semicolon on previous line?)\n");
258 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
260 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy_if(t,UTF) || *t == ':'); t++) ;
261 if (t < PL_bufptr && isSPACE(*t))
262 Perl_warn(aTHX_ "\t(Do you need to predeclare %.*s?)\n",
263 t - PL_oldoldbufptr, PL_oldoldbufptr);
267 Perl_warn(aTHX_ "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp);
274 * Complain about missing quote/regexp/heredoc terminator.
275 * If it's called with (char *)NULL then it cauterizes the line buffer.
276 * If we're in a delimited string and the delimiter is a control
277 * character, it's reformatted into a two-char sequence like ^C.
282 S_missingterm(pTHX_ char *s)
287 char *nl = strrchr(s,'\n');
293 iscntrl(PL_multi_close)
295 PL_multi_close < 32 || PL_multi_close == 127
299 tmpbuf[1] = toCTRL(PL_multi_close);
305 *tmpbuf = (char)PL_multi_close;
309 q = strchr(s,'"') ? '\'' : '"';
310 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
318 Perl_deprecate(pTHX_ char *s)
320 if (ckWARN(WARN_DEPRECATED))
321 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED), "Use of %s is deprecated", s);
325 Perl_deprecate_old(pTHX_ char *s)
327 /* This function should NOT be called for any new deprecated warnings */
328 /* Use Perl_deprecate instead */
330 /* It is here to maintain backward compatibility with the pre-5.8 */
331 /* warnings category hierarchy. The "deprecated" category used to */
332 /* live under the "syntax" category. It is now a top-level category */
333 /* in its own right. */
335 if (ckWARN2(WARN_DEPRECATED, WARN_SYNTAX))
336 Perl_warner(aTHX_ packWARN2(WARN_DEPRECATED, WARN_SYNTAX),
337 "Use of %s is deprecated", s);
342 * Deprecate a comma-less variable list.
348 deprecate_old("comma-less variable list");
352 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
353 * utf16-to-utf8-reversed.
356 #ifdef PERL_CR_FILTER
360 register char *s = SvPVX(sv);
361 register char *e = s + SvCUR(sv);
362 /* outer loop optimized to do nothing if there are no CR-LFs */
364 if (*s++ == '\r' && *s == '\n') {
365 /* hit a CR-LF, need to copy the rest */
366 register char *d = s - 1;
369 if (*s == '\r' && s[1] == '\n')
380 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
382 I32 count = FILTER_READ(idx+1, sv, maxlen);
383 if (count > 0 && !maxlen)
391 * Initialize variables. Uses the Perl save_stack to save its state (for
392 * recursive calls to the parser).
396 Perl_lex_start(pTHX_ SV *line)
401 SAVEI32(PL_lex_dojoin);
402 SAVEI32(PL_lex_brackets);
403 SAVEI32(PL_lex_casemods);
404 SAVEI32(PL_lex_starts);
405 SAVEI32(PL_lex_state);
406 SAVEVPTR(PL_lex_inpat);
407 SAVEI32(PL_lex_inwhat);
408 if (PL_lex_state == LEX_KNOWNEXT) {
409 I32 toke = PL_nexttoke;
410 while (--toke >= 0) {
411 SAVEI32(PL_nexttype[toke]);
412 SAVEVPTR(PL_nextval[toke]);
414 SAVEI32(PL_nexttoke);
416 SAVECOPLINE(PL_curcop);
419 SAVEPPTR(PL_oldbufptr);
420 SAVEPPTR(PL_oldoldbufptr);
421 SAVEPPTR(PL_last_lop);
422 SAVEPPTR(PL_last_uni);
423 SAVEPPTR(PL_linestart);
424 SAVESPTR(PL_linestr);
425 SAVEPPTR(PL_lex_brackstack);
426 SAVEPPTR(PL_lex_casestack);
427 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
428 SAVESPTR(PL_lex_stuff);
429 SAVEI32(PL_lex_defer);
430 SAVEI32(PL_sublex_info.sub_inwhat);
431 SAVESPTR(PL_lex_repl);
433 SAVEINT(PL_lex_expect);
435 PL_lex_state = LEX_NORMAL;
439 New(899, PL_lex_brackstack, 120, char);
440 New(899, PL_lex_casestack, 12, char);
441 SAVEFREEPV(PL_lex_brackstack);
442 SAVEFREEPV(PL_lex_casestack);
444 *PL_lex_casestack = '\0';
447 PL_lex_stuff = Nullsv;
448 PL_lex_repl = Nullsv;
452 PL_sublex_info.sub_inwhat = 0;
454 if (SvREADONLY(PL_linestr))
455 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
456 s = SvPV(PL_linestr, len);
457 if (len && s[len-1] != ';') {
458 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
459 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
460 sv_catpvn(PL_linestr, "\n;", 2);
462 SvTEMP_off(PL_linestr);
463 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
464 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
465 PL_last_lop = PL_last_uni = Nullch;
471 * Finalizer for lexing operations. Must be called when the parser is
472 * done with the lexer.
478 PL_doextract = FALSE;
483 * This subroutine has nothing to do with tilting, whether at windmills
484 * or pinball tables. Its name is short for "increment line". It
485 * increments the current line number in CopLINE(PL_curcop) and checks
486 * to see whether the line starts with a comment of the form
487 * # line 500 "foo.pm"
488 * If so, it sets the current line number and file to the values in the comment.
492 S_incline(pTHX_ char *s)
499 CopLINE_inc(PL_curcop);
502 while (SPACE_OR_TAB(*s)) s++;
503 if (strnEQ(s, "line", 4))
507 if (SPACE_OR_TAB(*s))
511 while (SPACE_OR_TAB(*s)) s++;
517 while (SPACE_OR_TAB(*s))
519 if (*s == '"' && (t = strchr(s+1, '"'))) {
524 for (t = s; !isSPACE(*t); t++) ;
527 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
529 if (*e != '\n' && *e != '\0')
530 return; /* false alarm */
535 CopFILE_free(PL_curcop);
536 CopFILE_set(PL_curcop, s);
539 CopLINE_set(PL_curcop, atoi(n)-1);
544 * Called to gobble the appropriate amount and type of whitespace.
545 * Skips comments as well.
549 S_skipspace(pTHX_ register char *s)
551 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
552 while (s < PL_bufend && SPACE_OR_TAB(*s))
558 SSize_t oldprevlen, oldoldprevlen;
559 SSize_t oldloplen = 0, oldunilen = 0;
560 while (s < PL_bufend && isSPACE(*s)) {
561 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
566 if (s < PL_bufend && *s == '#') {
567 while (s < PL_bufend && *s != '\n')
571 if (PL_in_eval && !PL_rsfp) {
578 /* only continue to recharge the buffer if we're at the end
579 * of the buffer, we're not reading from a source filter, and
580 * we're in normal lexing mode
582 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
583 PL_lex_state == LEX_FORMLINE)
586 /* try to recharge the buffer */
587 if ((s = filter_gets(PL_linestr, PL_rsfp,
588 (prevlen = SvCUR(PL_linestr)))) == Nullch)
590 /* end of file. Add on the -p or -n magic */
591 if (PL_minus_n || PL_minus_p) {
592 sv_setpv(PL_linestr,PL_minus_p ?
593 ";}continue{print or die qq(-p destination: $!\\n)" :
595 sv_catpv(PL_linestr,";}");
596 PL_minus_n = PL_minus_p = 0;
599 sv_setpv(PL_linestr,";");
601 /* reset variables for next time we lex */
602 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
604 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
605 PL_last_lop = PL_last_uni = Nullch;
607 /* Close the filehandle. Could be from -P preprocessor,
608 * STDIN, or a regular file. If we were reading code from
609 * STDIN (because the commandline held no -e or filename)
610 * then we don't close it, we reset it so the code can
611 * read from STDIN too.
614 if (PL_preprocess && !PL_in_eval)
615 (void)PerlProc_pclose(PL_rsfp);
616 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
617 PerlIO_clearerr(PL_rsfp);
619 (void)PerlIO_close(PL_rsfp);
624 /* not at end of file, so we only read another line */
625 /* make corresponding updates to old pointers, for yyerror() */
626 oldprevlen = PL_oldbufptr - PL_bufend;
627 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
629 oldunilen = PL_last_uni - PL_bufend;
631 oldloplen = PL_last_lop - PL_bufend;
632 PL_linestart = PL_bufptr = s + prevlen;
633 PL_bufend = s + SvCUR(PL_linestr);
635 PL_oldbufptr = s + oldprevlen;
636 PL_oldoldbufptr = s + oldoldprevlen;
638 PL_last_uni = s + oldunilen;
640 PL_last_lop = s + oldloplen;
643 /* debugger active and we're not compiling the debugger code,
644 * so store the line into the debugger's array of lines
646 if (PERLDB_LINE && PL_curstash != PL_debstash) {
647 SV *sv = NEWSV(85,0);
649 sv_upgrade(sv, SVt_PVMG);
650 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
653 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
660 * Check the unary operators to ensure there's no ambiguity in how they're
661 * used. An ambiguous piece of code would be:
663 * This doesn't mean rand() + 5. Because rand() is a unary operator,
664 * the +5 is its argument.
673 if (PL_oldoldbufptr != PL_last_uni)
675 while (isSPACE(*PL_last_uni))
677 for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ;
678 if ((t = strchr(s, '(')) && t < PL_bufptr)
680 if (ckWARN_d(WARN_AMBIGUOUS)){
683 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
684 "Warning: Use of \"%s\" without parens is ambiguous",
691 * LOP : macro to build a list operator. Its behaviour has been replaced
692 * with a subroutine, S_lop() for which LOP is just another name.
695 #define LOP(f,x) return lop(f,x,s)
699 * Build a list operator (or something that might be one). The rules:
700 * - if we have a next token, then it's a list operator [why?]
701 * - if the next thing is an opening paren, then it's a function
702 * - else it's a list operator
706 S_lop(pTHX_ I32 f, int x, char *s)
713 PL_last_lop = PL_oldbufptr;
714 PL_last_lop_op = (OPCODE)f;
728 * When the lexer realizes it knows the next token (for instance,
729 * it is reordering tokens for the parser) then it can call S_force_next
730 * to know what token to return the next time the lexer is called. Caller
731 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
732 * handles the token correctly.
736 S_force_next(pTHX_ I32 type)
738 PL_nexttype[PL_nexttoke] = type;
740 if (PL_lex_state != LEX_KNOWNEXT) {
741 PL_lex_defer = PL_lex_state;
742 PL_lex_expect = PL_expect;
743 PL_lex_state = LEX_KNOWNEXT;
749 * When the lexer knows the next thing is a word (for instance, it has
750 * just seen -> and it knows that the next char is a word char, then
751 * it calls S_force_word to stick the next word into the PL_next lookahead.
754 * char *start : buffer position (must be within PL_linestr)
755 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
756 * int check_keyword : if true, Perl checks to make sure the word isn't
757 * a keyword (do this if the word is a label, e.g. goto FOO)
758 * int allow_pack : if true, : characters will also be allowed (require,
760 * int allow_initial_tick : used by the "sub" lexer only.
764 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
769 start = skipspace(start);
771 if (isIDFIRST_lazy_if(s,UTF) ||
772 (allow_pack && *s == ':') ||
773 (allow_initial_tick && *s == '\'') )
775 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
776 if (check_keyword && keyword(PL_tokenbuf, len))
778 if (token == METHOD) {
783 PL_expect = XOPERATOR;
786 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0));
787 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
795 * Called when the lexer wants $foo *foo &foo etc, but the program
796 * text only contains the "foo" portion. The first argument is a pointer
797 * to the "foo", and the second argument is the type symbol to prefix.
798 * Forces the next token to be a "WORD".
799 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
803 S_force_ident(pTHX_ register char *s, int kind)
806 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
807 PL_nextval[PL_nexttoke].opval = o;
810 o->op_private = OPpCONST_ENTERED;
811 /* XXX see note in pp_entereval() for why we forgo typo
812 warnings if the symbol must be introduced in an eval.
814 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
815 kind == '$' ? SVt_PV :
816 kind == '@' ? SVt_PVAV :
817 kind == '%' ? SVt_PVHV :
825 Perl_str_to_version(pTHX_ SV *sv)
830 char *start = SvPVx(sv,len);
831 bool utf = SvUTF8(sv) ? TRUE : FALSE;
832 char *end = start + len;
833 while (start < end) {
837 n = utf8n_to_uvchr((U8*)start, len, &skip, 0);
842 retval += ((NV)n)/nshift;
851 * Forces the next token to be a version number.
852 * If the next token appears to be an invalid version number, (e.g. "v2b"),
853 * and if "guessing" is TRUE, then no new token is created (and the caller
854 * must use an alternative parsing method).
858 S_force_version(pTHX_ char *s, int guessing)
860 OP *version = Nullop;
869 while (isDIGIT(*d) || *d == '_' || *d == '.')
871 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
873 s = scan_num(s, &yylval);
874 version = yylval.opval;
875 ver = cSVOPx(version)->op_sv;
876 if (SvPOK(ver) && !SvNIOK(ver)) {
877 (void)SvUPGRADE(ver, SVt_PVNV);
878 SvNVX(ver) = str_to_version(ver);
879 SvNOK_on(ver); /* hint that it is a version */
886 /* NOTE: The parser sees the package name and the VERSION swapped */
887 PL_nextval[PL_nexttoke].opval = version;
895 * Tokenize a quoted string passed in as an SV. It finds the next
896 * chunk, up to end of string or a backslash. It may make a new
897 * SV containing that chunk (if HINT_NEW_STRING is on). It also
902 S_tokeq(pTHX_ SV *sv)
913 s = SvPV_force(sv, len);
914 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
917 while (s < send && *s != '\\')
922 if ( PL_hints & HINT_NEW_STRING ) {
923 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
929 if (s + 1 < send && (s[1] == '\\'))
930 s++; /* all that, just for this */
935 SvCUR_set(sv, d - SvPVX(sv));
937 if ( PL_hints & HINT_NEW_STRING )
938 return new_constant(NULL, 0, "q", sv, pv, "q");
943 * Now come three functions related to double-quote context,
944 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
945 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
946 * interact with PL_lex_state, and create fake ( ... ) argument lists
947 * to handle functions and concatenation.
948 * They assume that whoever calls them will be setting up a fake
949 * join call, because each subthing puts a ',' after it. This lets
952 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
954 * (I'm not sure whether the spurious commas at the end of lcfirst's
955 * arguments and join's arguments are created or not).
960 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
962 * Pattern matching will set PL_lex_op to the pattern-matching op to
963 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
965 * OP_CONST and OP_READLINE are easy--just make the new op and return.
967 * Everything else becomes a FUNC.
969 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
970 * had an OP_CONST or OP_READLINE). This just sets us up for a
971 * call to S_sublex_push().
977 register I32 op_type = yylval.ival;
979 if (op_type == OP_NULL) {
980 yylval.opval = PL_lex_op;
984 if (op_type == OP_CONST || op_type == OP_READLINE) {
985 SV *sv = tokeq(PL_lex_stuff);
987 if (SvTYPE(sv) == SVt_PVIV) {
988 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
994 nsv = newSVpvn(p, len);
1000 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
1001 PL_lex_stuff = Nullsv;
1005 PL_sublex_info.super_state = PL_lex_state;
1006 PL_sublex_info.sub_inwhat = op_type;
1007 PL_sublex_info.sub_op = PL_lex_op;
1008 PL_lex_state = LEX_INTERPPUSH;
1012 yylval.opval = PL_lex_op;
1022 * Create a new scope to save the lexing state. The scope will be
1023 * ended in S_sublex_done. Returns a '(', starting the function arguments
1024 * to the uc, lc, etc. found before.
1025 * Sets PL_lex_state to LEX_INTERPCONCAT.
1033 PL_lex_state = PL_sublex_info.super_state;
1034 SAVEI32(PL_lex_dojoin);
1035 SAVEI32(PL_lex_brackets);
1036 SAVEI32(PL_lex_casemods);
1037 SAVEI32(PL_lex_starts);
1038 SAVEI32(PL_lex_state);
1039 SAVEVPTR(PL_lex_inpat);
1040 SAVEI32(PL_lex_inwhat);
1041 SAVECOPLINE(PL_curcop);
1042 SAVEPPTR(PL_bufptr);
1043 SAVEPPTR(PL_bufend);
1044 SAVEPPTR(PL_oldbufptr);
1045 SAVEPPTR(PL_oldoldbufptr);
1046 SAVEPPTR(PL_last_lop);
1047 SAVEPPTR(PL_last_uni);
1048 SAVEPPTR(PL_linestart);
1049 SAVESPTR(PL_linestr);
1050 SAVEPPTR(PL_lex_brackstack);
1051 SAVEPPTR(PL_lex_casestack);
1053 PL_linestr = PL_lex_stuff;
1054 PL_lex_stuff = Nullsv;
1056 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1057 = SvPVX(PL_linestr);
1058 PL_bufend += SvCUR(PL_linestr);
1059 PL_last_lop = PL_last_uni = Nullch;
1060 SAVEFREESV(PL_linestr);
1062 PL_lex_dojoin = FALSE;
1063 PL_lex_brackets = 0;
1064 New(899, PL_lex_brackstack, 120, char);
1065 New(899, PL_lex_casestack, 12, char);
1066 SAVEFREEPV(PL_lex_brackstack);
1067 SAVEFREEPV(PL_lex_casestack);
1068 PL_lex_casemods = 0;
1069 *PL_lex_casestack = '\0';
1071 PL_lex_state = LEX_INTERPCONCAT;
1072 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
1074 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1075 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1076 PL_lex_inpat = PL_sublex_info.sub_op;
1078 PL_lex_inpat = Nullop;
1085 * Restores lexer state after a S_sublex_push.
1091 if (!PL_lex_starts++) {
1092 SV *sv = newSVpvn("",0);
1093 if (SvUTF8(PL_linestr))
1095 PL_expect = XOPERATOR;
1096 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1100 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1101 PL_lex_state = LEX_INTERPCASEMOD;
1105 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1106 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1107 PL_linestr = PL_lex_repl;
1109 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1110 PL_bufend += SvCUR(PL_linestr);
1111 PL_last_lop = PL_last_uni = Nullch;
1112 SAVEFREESV(PL_linestr);
1113 PL_lex_dojoin = FALSE;
1114 PL_lex_brackets = 0;
1115 PL_lex_casemods = 0;
1116 *PL_lex_casestack = '\0';
1118 if (SvEVALED(PL_lex_repl)) {
1119 PL_lex_state = LEX_INTERPNORMAL;
1121 /* we don't clear PL_lex_repl here, so that we can check later
1122 whether this is an evalled subst; that means we rely on the
1123 logic to ensure sublex_done() is called again only via the
1124 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1127 PL_lex_state = LEX_INTERPCONCAT;
1128 PL_lex_repl = Nullsv;
1134 PL_bufend = SvPVX(PL_linestr);
1135 PL_bufend += SvCUR(PL_linestr);
1136 PL_expect = XOPERATOR;
1137 PL_sublex_info.sub_inwhat = 0;
1145 Extracts a pattern, double-quoted string, or transliteration. This
1148 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1149 processing a pattern (PL_lex_inpat is true), a transliteration
1150 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1152 Returns a pointer to the character scanned up to. Iff this is
1153 advanced from the start pointer supplied (ie if anything was
1154 successfully parsed), will leave an OP for the substring scanned
1155 in yylval. Caller must intuit reason for not parsing further
1156 by looking at the next characters herself.
1160 double-quoted style: \r and \n
1161 regexp special ones: \D \s
1163 backrefs: \1 (deprecated in substitution replacements)
1164 case and quoting: \U \Q \E
1165 stops on @ and $, but not for $ as tail anchor
1167 In transliterations:
1168 characters are VERY literal, except for - not at the start or end
1169 of the string, which indicates a range. scan_const expands the
1170 range to the full set of intermediate characters.
1172 In double-quoted strings:
1174 double-quoted style: \r and \n
1176 backrefs: \1 (deprecated)
1177 case and quoting: \U \Q \E
1180 scan_const does *not* construct ops to handle interpolated strings.
1181 It stops processing as soon as it finds an embedded $ or @ variable
1182 and leaves it to the caller to work out what's going on.
1184 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @::foo.
1186 $ in pattern could be $foo or could be tail anchor. Assumption:
1187 it's a tail anchor if $ is the last thing in the string, or if it's
1188 followed by one of ")| \n\t"
1190 \1 (backreferences) are turned into $1
1192 The structure of the code is
1193 while (there's a character to process) {
1194 handle transliteration ranges
1195 skip regexp comments
1196 skip # initiated comments in //x patterns
1197 check for embedded @foo
1198 check for embedded scalars
1200 leave intact backslashes from leave (below)
1201 deprecate \1 in strings and sub replacements
1202 handle string-changing backslashes \l \U \Q \E, etc.
1203 switch (what was escaped) {
1204 handle - in a transliteration (becomes a literal -)
1205 handle \132 octal characters
1206 handle 0x15 hex characters
1207 handle \cV (control V)
1208 handle printf backslashes (\f, \r, \n, etc)
1210 } (end if backslash)
1211 } (end while character to read)
1216 S_scan_const(pTHX_ char *start)
1218 register char *send = PL_bufend; /* end of the constant */
1219 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1220 register char *s = start; /* start of the constant */
1221 register char *d = SvPVX(sv); /* destination for copies */
1222 bool dorange = FALSE; /* are we in a translit range? */
1223 bool didrange = FALSE; /* did we just finish a range? */
1224 I32 has_utf8 = FALSE; /* Output constant is UTF8 */
1225 I32 this_utf8 = UTF; /* The source string is assumed to be UTF8 */
1228 const char *leaveit = /* set of acceptably-backslashed characters */
1230 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
1233 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1234 /* If we are doing a trans and we know we want UTF8 set expectation */
1235 has_utf8 = PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF);
1236 this_utf8 = PL_sublex_info.sub_op->op_private & (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1240 while (s < send || dorange) {
1241 /* get transliterations out of the way (they're most literal) */
1242 if (PL_lex_inwhat == OP_TRANS) {
1243 /* expand a range A-Z to the full set of characters. AIE! */
1245 I32 i; /* current expanded character */
1246 I32 min; /* first character in range */
1247 I32 max; /* last character in range */
1250 char *c = (char*)utf8_hop((U8*)d, -1);
1254 *c = (char)UTF_TO_NATIVE(0xff);
1255 /* mark the range as done, and continue */
1261 i = d - SvPVX(sv); /* remember current offset */
1262 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1263 d = SvPVX(sv) + i; /* refresh d after realloc */
1264 d -= 2; /* eat the first char and the - */
1266 min = (U8)*d; /* first char in range */
1267 max = (U8)d[1]; /* last char in range */
1271 "Invalid [] range \"%c-%c\" in transliteration operator",
1272 (char)min, (char)max);
1276 if ((isLOWER(min) && isLOWER(max)) ||
1277 (isUPPER(min) && isUPPER(max))) {
1279 for (i = min; i <= max; i++)
1281 *d++ = NATIVE_TO_NEED(has_utf8,i);
1283 for (i = min; i <= max; i++)
1285 *d++ = NATIVE_TO_NEED(has_utf8,i);
1290 for (i = min; i <= max; i++)
1293 /* mark the range as done, and continue */
1299 /* range begins (ignore - as first or last char) */
1300 else if (*s == '-' && s+1 < send && s != start) {
1302 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
1305 *d++ = (char)UTF_TO_NATIVE(0xff); /* use illegal utf8 byte--see pmtrans */
1317 /* if we get here, we're not doing a transliteration */
1319 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1320 except for the last char, which will be done separately. */
1321 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1323 while (s < send && *s != ')')
1324 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1326 else if (s[2] == '{' /* This should match regcomp.c */
1327 || ((s[2] == 'p' || s[2] == '?') && s[3] == '{'))
1330 char *regparse = s + (s[2] == '{' ? 3 : 4);
1333 while (count && (c = *regparse)) {
1334 if (c == '\\' && regparse[1])
1342 if (*regparse != ')') {
1343 regparse--; /* Leave one char for continuation. */
1344 yyerror("Sequence (?{...}) not terminated or not {}-balanced");
1346 while (s < regparse)
1347 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1351 /* likewise skip #-initiated comments in //x patterns */
1352 else if (*s == '#' && PL_lex_inpat &&
1353 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1354 while (s+1 < send && *s != '\n')
1355 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1358 /* check for embedded arrays
1359 (@foo, @::foo, @'foo, @{foo}, @$foo, @+, @-)
1361 else if (*s == '@' && s[1]
1362 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$+-", s[1])))
1365 /* check for embedded scalars. only stop if we're sure it's a
1368 else if (*s == '$') {
1369 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1371 if (s + 1 < send && !strchr("()| \r\n\t", s[1]))
1372 break; /* in regexp, $ might be tail anchor */
1375 /* End of else if chain - OP_TRANS rejoin rest */
1378 if (*s == '\\' && s+1 < send) {
1381 /* some backslashes we leave behind */
1382 if (*leaveit && *s && strchr(leaveit, *s)) {
1383 *d++ = NATIVE_TO_NEED(has_utf8,'\\');
1384 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1388 /* deprecate \1 in strings and substitution replacements */
1389 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1390 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1392 if (ckWARN(WARN_SYNTAX))
1393 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "\\%c better written as $%c", *s, *s);
1398 /* string-change backslash escapes */
1399 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1404 /* if we get here, it's either a quoted -, or a digit */
1407 /* quoted - in transliterations */
1409 if (PL_lex_inwhat == OP_TRANS) {
1416 if (ckWARN(WARN_MISC) &&
1419 Perl_warner(aTHX_ packWARN(WARN_MISC),
1420 "Unrecognized escape \\%c passed through",
1422 /* default action is to copy the quoted character */
1423 goto default_action;
1426 /* \132 indicates an octal constant */
1427 case '0': case '1': case '2': case '3':
1428 case '4': case '5': case '6': case '7':
1432 uv = grok_oct(s, &len, &flags, NULL);
1435 goto NUM_ESCAPE_INSERT;
1437 /* \x24 indicates a hex constant */
1441 char* e = strchr(s, '}');
1442 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
1443 PERL_SCAN_DISALLOW_PREFIX;
1448 yyerror("Missing right brace on \\x{}");
1452 uv = grok_hex(s, &len, &flags, NULL);
1458 I32 flags = PERL_SCAN_DISALLOW_PREFIX;
1459 uv = grok_hex(s, &len, &flags, NULL);
1465 /* Insert oct or hex escaped character.
1466 * There will always enough room in sv since such
1467 * escapes will be longer than any UTF-8 sequence
1468 * they can end up as. */
1470 /* We need to map to chars to ASCII before doing the tests
1473 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(uv))) {
1474 if (!has_utf8 && uv > 255) {
1475 /* Might need to recode whatever we have
1476 * accumulated so far if it contains any
1479 * (Can't we keep track of that and avoid
1480 * this rescan? --jhi)
1484 for (c = (U8 *) SvPVX(sv); c < (U8 *)d; c++) {
1485 if (!NATIVE_IS_INVARIANT(*c)) {
1490 STRLEN offset = d - SvPVX(sv);
1492 d = SvGROW(sv, SvLEN(sv) + hicount + 1) + offset;
1496 while (src >= (U8 *)SvPVX(sv)) {
1497 if (!NATIVE_IS_INVARIANT(*src)) {
1498 U8 ch = NATIVE_TO_ASCII(*src);
1499 *dst-- = (U8)UTF8_EIGHT_BIT_LO(ch);
1500 *dst-- = (U8)UTF8_EIGHT_BIT_HI(ch);
1510 if (has_utf8 || uv > 255) {
1511 d = (char*)uvchr_to_utf8((U8*)d, uv);
1513 if (PL_lex_inwhat == OP_TRANS &&
1514 PL_sublex_info.sub_op) {
1515 PL_sublex_info.sub_op->op_private |=
1516 (PL_lex_repl ? OPpTRANS_FROM_UTF
1529 /* \N{LATIN SMALL LETTER A} is a named character */
1533 char* e = strchr(s, '}');
1539 yyerror("Missing right brace on \\N{}");
1543 res = newSVpvn(s + 1, e - s - 1);
1544 res = new_constant( Nullch, 0, "charnames",
1545 res, Nullsv, "\\N{...}" );
1547 sv_utf8_upgrade(res);
1548 str = SvPV(res,len);
1549 #ifdef EBCDIC_NEVER_MIND
1550 /* charnames uses pack U and that has been
1551 * recently changed to do the below uni->native
1552 * mapping, so this would be redundant (and wrong,
1553 * the code point would be doubly converted).
1554 * But leave this in just in case the pack U change
1555 * gets revoked, but the semantics is still
1556 * desireable for charnames. --jhi */
1558 UV uv = utf8_to_uvchr((U8*)str, 0);
1561 U8 tmpbuf[UTF8_MAXLEN+1], *d;
1563 d = uvchr_to_utf8(tmpbuf, UNI_TO_NATIVE(uv));
1564 sv_setpvn(res, (char *)tmpbuf, d - tmpbuf);
1565 str = SvPV(res, len);
1569 if (!has_utf8 && SvUTF8(res)) {
1570 char *ostart = SvPVX(sv);
1571 SvCUR_set(sv, d - ostart);
1574 sv_utf8_upgrade(sv);
1575 /* this just broke our allocation above... */
1576 SvGROW(sv, (STRLEN)(send - start));
1577 d = SvPVX(sv) + SvCUR(sv);
1580 if (len > (STRLEN)(e - s + 4)) { /* I _guess_ 4 is \N{} --jhi */
1581 char *odest = SvPVX(sv);
1583 SvGROW(sv, (SvLEN(sv) + len - (e - s + 4)));
1584 d = SvPVX(sv) + (d - odest);
1586 Copy(str, d, len, char);
1593 yyerror("Missing braces on \\N{}");
1596 /* \c is a control character */
1605 *d++ = NATIVE_TO_NEED(has_utf8,toCTRL(c));
1609 /* printf-style backslashes, formfeeds, newlines, etc */
1611 *d++ = NATIVE_TO_NEED(has_utf8,'\b');
1614 *d++ = NATIVE_TO_NEED(has_utf8,'\n');
1617 *d++ = NATIVE_TO_NEED(has_utf8,'\r');
1620 *d++ = NATIVE_TO_NEED(has_utf8,'\f');
1623 *d++ = NATIVE_TO_NEED(has_utf8,'\t');
1626 *d++ = ASCII_TO_NEED(has_utf8,'\033');
1629 *d++ = ASCII_TO_NEED(has_utf8,'\007');
1635 } /* end if (backslash) */
1638 /* If we started with encoded form, or already know we want it
1639 and then encode the next character */
1640 if ((has_utf8 || this_utf8) && !NATIVE_IS_INVARIANT((U8)(*s))) {
1642 UV uv = (this_utf8) ? utf8n_to_uvchr((U8*)s, send - s, &len, 0) : (UV) ((U8) *s);
1643 STRLEN need = UNISKIP(NATIVE_TO_UNI(uv));
1646 /* encoded value larger than old, need extra space (NOTE: SvCUR() not set here) */
1647 STRLEN off = d - SvPVX(sv);
1648 d = SvGROW(sv, SvLEN(sv) + (need-len)) + off;
1650 d = (char*)uvchr_to_utf8((U8*)d, uv);
1654 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1656 } /* while loop to process each character */
1658 /* terminate the string and set up the sv */
1660 SvCUR_set(sv, d - SvPVX(sv));
1661 if (SvCUR(sv) >= SvLEN(sv))
1662 Perl_croak(aTHX_ "panic: constant overflowed allocated space");
1665 if (PL_encoding && !has_utf8) {
1666 sv_recode_to_utf8(sv, PL_encoding);
1671 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1672 PL_sublex_info.sub_op->op_private |=
1673 (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1677 /* shrink the sv if we allocated more than we used */
1678 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1679 SvLEN_set(sv, SvCUR(sv) + 1);
1680 Renew(SvPVX(sv), SvLEN(sv), char);
1683 /* return the substring (via yylval) only if we parsed anything */
1684 if (s > PL_bufptr) {
1685 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1686 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1688 ( PL_lex_inwhat == OP_TRANS
1690 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1693 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1700 * Returns TRUE if there's more to the expression (e.g., a subscript),
1703 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1705 * ->[ and ->{ return TRUE
1706 * { and [ outside a pattern are always subscripts, so return TRUE
1707 * if we're outside a pattern and it's not { or [, then return FALSE
1708 * if we're in a pattern and the first char is a {
1709 * {4,5} (any digits around the comma) returns FALSE
1710 * if we're in a pattern and the first char is a [
1712 * [SOMETHING] has a funky algorithm to decide whether it's a
1713 * character class or not. It has to deal with things like
1714 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1715 * anything else returns TRUE
1718 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1721 S_intuit_more(pTHX_ register char *s)
1723 if (PL_lex_brackets)
1725 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1727 if (*s != '{' && *s != '[')
1732 /* In a pattern, so maybe we have {n,m}. */
1749 /* On the other hand, maybe we have a character class */
1752 if (*s == ']' || *s == '^')
1755 /* this is terrifying, and it works */
1756 int weight = 2; /* let's weigh the evidence */
1758 unsigned char un_char = 255, last_un_char;
1759 char *send = strchr(s,']');
1760 char tmpbuf[sizeof PL_tokenbuf * 4];
1762 if (!send) /* has to be an expression */
1765 Zero(seen,256,char);
1768 else if (isDIGIT(*s)) {
1770 if (isDIGIT(s[1]) && s[2] == ']')
1776 for (; s < send; s++) {
1777 last_un_char = un_char;
1778 un_char = (unsigned char)*s;
1783 weight -= seen[un_char] * 10;
1784 if (isALNUM_lazy_if(s+1,UTF)) {
1785 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1786 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1791 else if (*s == '$' && s[1] &&
1792 strchr("[#!%*<>()-=",s[1])) {
1793 if (/*{*/ strchr("])} =",s[2]))
1802 if (strchr("wds]",s[1]))
1804 else if (seen['\''] || seen['"'])
1806 else if (strchr("rnftbxcav",s[1]))
1808 else if (isDIGIT(s[1])) {
1810 while (s[1] && isDIGIT(s[1]))
1820 if (strchr("aA01! ",last_un_char))
1822 if (strchr("zZ79~",s[1]))
1824 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1825 weight -= 5; /* cope with negative subscript */
1828 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1829 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1834 if (keyword(tmpbuf, d - tmpbuf))
1837 if (un_char == last_un_char + 1)
1839 weight -= seen[un_char];
1844 if (weight >= 0) /* probably a character class */
1854 * Does all the checking to disambiguate
1856 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1857 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1859 * First argument is the stuff after the first token, e.g. "bar".
1861 * Not a method if bar is a filehandle.
1862 * Not a method if foo is a subroutine prototyped to take a filehandle.
1863 * Not a method if it's really "Foo $bar"
1864 * Method if it's "foo $bar"
1865 * Not a method if it's really "print foo $bar"
1866 * Method if it's really "foo package::" (interpreted as package->foo)
1867 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1868 * Not a method if bar is a filehandle or package, but is quoted with
1873 S_intuit_method(pTHX_ char *start, GV *gv)
1875 char *s = start + (*start == '$');
1876 char tmpbuf[sizeof PL_tokenbuf];
1884 if ((cv = GvCVu(gv))) {
1885 char *proto = SvPVX(cv);
1895 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1896 /* start is the beginning of the possible filehandle/object,
1897 * and s is the end of it
1898 * tmpbuf is a copy of it
1901 if (*start == '$') {
1902 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1907 return *s == '(' ? FUNCMETH : METHOD;
1909 if (!keyword(tmpbuf, len)) {
1910 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1915 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1916 if (indirgv && GvCVu(indirgv))
1918 /* filehandle or package name makes it a method */
1919 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1921 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1922 return 0; /* no assumptions -- "=>" quotes bearword */
1924 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1925 newSVpvn(tmpbuf,len));
1926 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1930 return *s == '(' ? FUNCMETH : METHOD;
1938 * Return a string of Perl code to load the debugger. If PERL5DB
1939 * is set, it will return the contents of that, otherwise a
1940 * compile-time require of perl5db.pl.
1947 char *pdb = PerlEnv_getenv("PERL5DB");
1951 SETERRNO(0,SS$_NORMAL);
1952 return "BEGIN { require 'perl5db.pl' }";
1958 /* Encoded script support. filter_add() effectively inserts a
1959 * 'pre-processing' function into the current source input stream.
1960 * Note that the filter function only applies to the current source file
1961 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1963 * The datasv parameter (which may be NULL) can be used to pass
1964 * private data to this instance of the filter. The filter function
1965 * can recover the SV using the FILTER_DATA macro and use it to
1966 * store private buffers and state information.
1968 * The supplied datasv parameter is upgraded to a PVIO type
1969 * and the IoDIRP/IoANY field is used to store the function pointer,
1970 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1971 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1972 * private use must be set using malloc'd pointers.
1976 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1981 if (!PL_rsfp_filters)
1982 PL_rsfp_filters = newAV();
1984 datasv = NEWSV(255,0);
1985 if (!SvUPGRADE(datasv, SVt_PVIO))
1986 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1987 IoANY(datasv) = (void *)funcp; /* stash funcp into spare field */
1988 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
1989 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1990 (void*)funcp, SvPV_nolen(datasv)));
1991 av_unshift(PL_rsfp_filters, 1);
1992 av_store(PL_rsfp_filters, 0, datasv) ;
1997 /* Delete most recently added instance of this filter function. */
1999 Perl_filter_del(pTHX_ filter_t funcp)
2002 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", (void*)funcp));
2003 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
2005 /* if filter is on top of stack (usual case) just pop it off */
2006 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
2007 if (IoANY(datasv) == (void *)funcp) {
2008 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
2009 IoANY(datasv) = (void *)NULL;
2010 sv_free(av_pop(PL_rsfp_filters));
2014 /* we need to search for the correct entry and clear it */
2015 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
2019 /* Invoke the n'th filter function for the current rsfp. */
2021 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
2024 /* 0 = read one text line */
2029 if (!PL_rsfp_filters)
2031 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
2032 /* Provide a default input filter to make life easy. */
2033 /* Note that we append to the line. This is handy. */
2034 DEBUG_P(PerlIO_printf(Perl_debug_log,
2035 "filter_read %d: from rsfp\n", idx));
2039 int old_len = SvCUR(buf_sv) ;
2041 /* ensure buf_sv is large enough */
2042 SvGROW(buf_sv, (STRLEN)(old_len + maxlen)) ;
2043 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
2044 if (PerlIO_error(PL_rsfp))
2045 return -1; /* error */
2047 return 0 ; /* end of file */
2049 SvCUR_set(buf_sv, old_len + len) ;
2052 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
2053 if (PerlIO_error(PL_rsfp))
2054 return -1; /* error */
2056 return 0 ; /* end of file */
2059 return SvCUR(buf_sv);
2061 /* Skip this filter slot if filter has been deleted */
2062 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
2063 DEBUG_P(PerlIO_printf(Perl_debug_log,
2064 "filter_read %d: skipped (filter deleted)\n",
2066 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
2068 /* Get function pointer hidden within datasv */
2069 funcp = (filter_t)IoANY(datasv);
2070 DEBUG_P(PerlIO_printf(Perl_debug_log,
2071 "filter_read %d: via function %p (%s)\n",
2072 idx, (void*)funcp, SvPV_nolen(datasv)));
2073 /* Call function. The function is expected to */
2074 /* call "FILTER_READ(idx+1, buf_sv)" first. */
2075 /* Return: <0:error, =0:eof, >0:not eof */
2076 return (*funcp)(aTHX_ idx, buf_sv, maxlen);
2080 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
2082 #ifdef PERL_CR_FILTER
2083 if (!PL_rsfp_filters) {
2084 filter_add(S_cr_textfilter,NULL);
2087 if (PL_rsfp_filters) {
2090 SvCUR_set(sv, 0); /* start with empty line */
2091 if (FILTER_READ(0, sv, 0) > 0)
2092 return ( SvPVX(sv) ) ;
2097 return (sv_gets(sv, fp, append));
2101 S_find_in_my_stash(pTHX_ char *pkgname, I32 len)
2105 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2109 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2110 (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV)))
2112 return GvHV(gv); /* Foo:: */
2115 /* use constant CLASS => 'MyClass' */
2116 if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) {
2118 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2119 pkgname = SvPV_nolen(sv);
2123 return gv_stashpv(pkgname, FALSE);
2127 static char* exp_name[] =
2128 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2129 "ATTRTERM", "TERMBLOCK"
2136 Works out what to call the token just pulled out of the input
2137 stream. The yacc parser takes care of taking the ops we return and
2138 stitching them into a tree.
2144 if read an identifier
2145 if we're in a my declaration
2146 croak if they tried to say my($foo::bar)
2147 build the ops for a my() declaration
2148 if it's an access to a my() variable
2149 are we in a sort block?
2150 croak if my($a); $a <=> $b
2151 build ops for access to a my() variable
2152 if in a dq string, and they've said @foo and we can't find @foo
2154 build ops for a bareword
2155 if we already built the token before, use it.
2158 #ifdef USE_PURE_BISON
2160 Perl_yylex_r(pTHX_ YYSTYPE *lvalp, int *lcharp)
2165 yylval_pointer[yyactlevel] = lvalp;
2166 yychar_pointer[yyactlevel] = lcharp;
2167 if (yyactlevel >= YYMAXLEVEL)
2168 Perl_croak(aTHX_ "panic: YYMAXLEVEL");
2170 r = Perl_yylex(aTHX);
2180 #pragma segment Perl_yylex
2193 /* check if there's an identifier for us to look at */
2194 if (PL_pending_ident)
2195 return S_pending_ident(aTHX);
2197 /* no identifier pending identification */
2199 switch (PL_lex_state) {
2201 case LEX_NORMAL: /* Some compilers will produce faster */
2202 case LEX_INTERPNORMAL: /* code if we comment these out. */
2206 /* when we've already built the next token, just pull it out of the queue */
2209 yylval = PL_nextval[PL_nexttoke];
2211 PL_lex_state = PL_lex_defer;
2212 PL_expect = PL_lex_expect;
2213 PL_lex_defer = LEX_NORMAL;
2215 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2216 "### Next token after '%s' was known, type %"IVdf"\n", PL_bufptr,
2217 (IV)PL_nexttype[PL_nexttoke]); });
2219 return(PL_nexttype[PL_nexttoke]);
2221 /* interpolated case modifiers like \L \U, including \Q and \E.
2222 when we get here, PL_bufptr is at the \
2224 case LEX_INTERPCASEMOD:
2226 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2227 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2229 /* handle \E or end of string */
2230 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2234 if (PL_lex_casemods) {
2235 oldmod = PL_lex_casestack[--PL_lex_casemods];
2236 PL_lex_casestack[PL_lex_casemods] = '\0';
2238 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2240 PL_lex_state = LEX_INTERPCONCAT;
2244 if (PL_bufptr != PL_bufend)
2246 PL_lex_state = LEX_INTERPCONCAT;
2250 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2251 "### Saw case modifier at '%s'\n", PL_bufptr); });
2253 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2254 tmp = *s, *s = s[2], s[2] = (char)tmp; /* misordered... */
2255 if (strchr("LU", *s) &&
2256 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2258 PL_lex_casestack[--PL_lex_casemods] = '\0';
2261 if (PL_lex_casemods > 10) {
2262 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2263 if (newlb != PL_lex_casestack) {
2265 PL_lex_casestack = newlb;
2268 PL_lex_casestack[PL_lex_casemods++] = *s;
2269 PL_lex_casestack[PL_lex_casemods] = '\0';
2270 PL_lex_state = LEX_INTERPCONCAT;
2271 PL_nextval[PL_nexttoke].ival = 0;
2274 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2276 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2278 PL_nextval[PL_nexttoke].ival = OP_LC;
2280 PL_nextval[PL_nexttoke].ival = OP_UC;
2282 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2284 Perl_croak(aTHX_ "panic: yylex");
2287 if (PL_lex_starts) {
2296 case LEX_INTERPPUSH:
2297 return sublex_push();
2299 case LEX_INTERPSTART:
2300 if (PL_bufptr == PL_bufend)
2301 return sublex_done();
2302 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2303 "### Interpolated variable at '%s'\n", PL_bufptr); });
2305 PL_lex_dojoin = (*PL_bufptr == '@');
2306 PL_lex_state = LEX_INTERPNORMAL;
2307 if (PL_lex_dojoin) {
2308 PL_nextval[PL_nexttoke].ival = 0;
2310 #ifdef USE_5005THREADS
2311 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2312 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2313 force_next(PRIVATEREF);
2315 force_ident("\"", '$');
2316 #endif /* USE_5005THREADS */
2317 PL_nextval[PL_nexttoke].ival = 0;
2319 PL_nextval[PL_nexttoke].ival = 0;
2321 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2324 if (PL_lex_starts++) {
2330 case LEX_INTERPENDMAYBE:
2331 if (intuit_more(PL_bufptr)) {
2332 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2338 if (PL_lex_dojoin) {
2339 PL_lex_dojoin = FALSE;
2340 PL_lex_state = LEX_INTERPCONCAT;
2343 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2344 && SvEVALED(PL_lex_repl))
2346 if (PL_bufptr != PL_bufend)
2347 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2348 PL_lex_repl = Nullsv;
2351 case LEX_INTERPCONCAT:
2353 if (PL_lex_brackets)
2354 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2356 if (PL_bufptr == PL_bufend)
2357 return sublex_done();
2359 if (SvIVX(PL_linestr) == '\'') {
2360 SV *sv = newSVsv(PL_linestr);
2363 else if ( PL_hints & HINT_NEW_RE )
2364 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2365 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2369 s = scan_const(PL_bufptr);
2371 PL_lex_state = LEX_INTERPCASEMOD;
2373 PL_lex_state = LEX_INTERPSTART;
2376 if (s != PL_bufptr) {
2377 PL_nextval[PL_nexttoke] = yylval;
2380 if (PL_lex_starts++)
2390 PL_lex_state = LEX_NORMAL;
2391 s = scan_formline(PL_bufptr);
2392 if (!PL_lex_formbrack)
2398 PL_oldoldbufptr = PL_oldbufptr;
2401 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2402 exp_name[PL_expect], s);
2408 if (isIDFIRST_lazy_if(s,UTF))
2410 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2413 goto fake_eof; /* emulate EOF on ^D or ^Z */
2418 if (PL_lex_brackets)
2419 yyerror("Missing right curly or square bracket");
2420 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2421 "### Tokener got EOF\n");
2425 if (s++ < PL_bufend)
2426 goto retry; /* ignore stray nulls */
2429 if (!PL_in_eval && !PL_preambled) {
2430 PL_preambled = TRUE;
2431 sv_setpv(PL_linestr,incl_perldb());
2432 if (SvCUR(PL_linestr))
2433 sv_catpv(PL_linestr,";");
2435 while(AvFILLp(PL_preambleav) >= 0) {
2436 SV *tmpsv = av_shift(PL_preambleav);
2437 sv_catsv(PL_linestr, tmpsv);
2438 sv_catpv(PL_linestr, ";");
2441 sv_free((SV*)PL_preambleav);
2442 PL_preambleav = NULL;
2444 if (PL_minus_n || PL_minus_p) {
2445 sv_catpv(PL_linestr, "LINE: while (<>) {");
2447 sv_catpv(PL_linestr,"chomp;");
2450 if (strchr("/'\"", *PL_splitstr)
2451 && strchr(PL_splitstr + 1, *PL_splitstr))
2452 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s);", PL_splitstr);
2455 s = "'~#\200\1'"; /* surely one char is unused...*/
2456 while (s[1] && strchr(PL_splitstr, *s)) s++;
2458 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s%c",
2459 "q" + (delim == '\''), delim);
2460 for (s = PL_splitstr; *s; s++) {
2462 sv_catpvn(PL_linestr, "\\", 1);
2463 sv_catpvn(PL_linestr, s, 1);
2465 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2469 sv_catpv(PL_linestr,"our @F=split(' ');");
2472 sv_catpv(PL_linestr, "\n");
2473 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2474 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2475 PL_last_lop = PL_last_uni = Nullch;
2476 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2477 SV *sv = NEWSV(85,0);
2479 sv_upgrade(sv, SVt_PVMG);
2480 sv_setsv(sv,PL_linestr);
2483 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2488 bof = PL_rsfp ? TRUE : FALSE;
2489 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2492 if (PL_preprocess && !PL_in_eval)
2493 (void)PerlProc_pclose(PL_rsfp);
2494 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2495 PerlIO_clearerr(PL_rsfp);
2497 (void)PerlIO_close(PL_rsfp);
2499 PL_doextract = FALSE;
2501 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2502 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2503 sv_catpv(PL_linestr,";}");
2504 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2505 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2506 PL_last_lop = PL_last_uni = Nullch;
2507 PL_minus_n = PL_minus_p = 0;
2510 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2511 PL_last_lop = PL_last_uni = Nullch;
2512 sv_setpv(PL_linestr,"");
2513 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2515 /* if it looks like the start of a BOM, check if it in fact is */
2516 else if (bof && (!*s || *(U8*)s == 0xEF || *(U8*)s >= 0xFE)) {
2517 #ifdef PERLIO_IS_STDIO
2518 # ifdef __GNU_LIBRARY__
2519 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
2520 # define FTELL_FOR_PIPE_IS_BROKEN
2524 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
2525 # define FTELL_FOR_PIPE_IS_BROKEN
2530 #ifdef FTELL_FOR_PIPE_IS_BROKEN
2531 /* This loses the possibility to detect the bof
2532 * situation on perl -P when the libc5 is being used.
2533 * Workaround? Maybe attach some extra state to PL_rsfp?
2536 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
2538 bof = PerlIO_tell(PL_rsfp) == (Off_t)SvCUR(PL_linestr);
2541 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2542 s = swallow_bom((U8*)s);
2546 /* Incest with pod. */
2547 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2548 sv_setpv(PL_linestr, "");
2549 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2550 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2551 PL_last_lop = PL_last_uni = Nullch;
2552 PL_doextract = FALSE;
2556 } while (PL_doextract);
2557 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2558 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2559 SV *sv = NEWSV(85,0);
2561 sv_upgrade(sv, SVt_PVMG);
2562 sv_setsv(sv,PL_linestr);
2565 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2567 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2568 PL_last_lop = PL_last_uni = Nullch;
2569 if (CopLINE(PL_curcop) == 1) {
2570 while (s < PL_bufend && isSPACE(*s))
2572 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2576 if (*s == '#' && *(s+1) == '!')
2578 #ifdef ALTERNATE_SHEBANG
2580 static char as[] = ALTERNATE_SHEBANG;
2581 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2582 d = s + (sizeof(as) - 1);
2584 #endif /* ALTERNATE_SHEBANG */
2593 while (*d && !isSPACE(*d))
2597 #ifdef ARG_ZERO_IS_SCRIPT
2598 if (ipathend > ipath) {
2600 * HP-UX (at least) sets argv[0] to the script name,
2601 * which makes $^X incorrect. And Digital UNIX and Linux,
2602 * at least, set argv[0] to the basename of the Perl
2603 * interpreter. So, having found "#!", we'll set it right.
2605 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV)); /* $^X */
2606 assert(SvPOK(x) || SvGMAGICAL(x));
2607 if (sv_eq(x, CopFILESV(PL_curcop))) {
2608 sv_setpvn(x, ipath, ipathend - ipath);
2611 TAINT_NOT; /* $^X is always tainted, but that's OK */
2613 #endif /* ARG_ZERO_IS_SCRIPT */
2618 d = instr(s,"perl -");
2620 d = instr(s,"perl");
2622 /* avoid getting into infinite loops when shebang
2623 * line contains "Perl" rather than "perl" */
2625 for (d = ipathend-4; d >= ipath; --d) {
2626 if ((*d == 'p' || *d == 'P')
2627 && !ibcmp(d, "perl", 4))
2637 #ifdef ALTERNATE_SHEBANG
2639 * If the ALTERNATE_SHEBANG on this system starts with a
2640 * character that can be part of a Perl expression, then if
2641 * we see it but not "perl", we're probably looking at the
2642 * start of Perl code, not a request to hand off to some
2643 * other interpreter. Similarly, if "perl" is there, but
2644 * not in the first 'word' of the line, we assume the line
2645 * contains the start of the Perl program.
2647 if (d && *s != '#') {
2649 while (*c && !strchr("; \t\r\n\f\v#", *c))
2652 d = Nullch; /* "perl" not in first word; ignore */
2654 *s = '#'; /* Don't try to parse shebang line */
2656 #endif /* ALTERNATE_SHEBANG */
2657 #ifndef MACOS_TRADITIONAL
2662 !instr(s,"indir") &&
2663 instr(PL_origargv[0],"perl"))
2669 while (s < PL_bufend && isSPACE(*s))
2671 if (s < PL_bufend) {
2672 Newz(899,newargv,PL_origargc+3,char*);
2674 while (s < PL_bufend && !isSPACE(*s))
2677 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2680 newargv = PL_origargv;
2682 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
2683 Perl_croak(aTHX_ "Can't exec %s", ipath);
2687 U32 oldpdb = PL_perldb;
2688 bool oldn = PL_minus_n;
2689 bool oldp = PL_minus_p;
2691 while (*d && !isSPACE(*d)) d++;
2692 while (SPACE_OR_TAB(*d)) d++;
2695 bool switches_done = PL_doswitches;
2697 if (*d == 'M' || *d == 'm') {
2699 while (*d && !isSPACE(*d)) d++;
2700 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2703 d = moreswitches(d);
2705 if ((PERLDB_LINE && !oldpdb) ||
2706 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2707 /* if we have already added "LINE: while (<>) {",
2708 we must not do it again */
2710 sv_setpv(PL_linestr, "");
2711 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2712 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2713 PL_last_lop = PL_last_uni = Nullch;
2714 PL_preambled = FALSE;
2716 (void)gv_fetchfile(PL_origfilename);
2719 if (PL_doswitches && !switches_done) {
2720 int argc = PL_origargc;
2721 char **argv = PL_origargv;
2724 } while (argc && argv[0][0] == '-' && argv[0][1]);
2725 init_argv_symbols(argc,argv);
2731 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2733 PL_lex_state = LEX_FORMLINE;
2738 #ifdef PERL_STRICT_CR
2739 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2741 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
2743 case ' ': case '\t': case '\f': case 013:
2744 #ifdef MACOS_TRADITIONAL
2751 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2752 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2753 /* handle eval qq[#line 1 "foo"\n ...] */
2754 CopLINE_dec(PL_curcop);
2758 while (s < d && *s != '\n')
2762 else if (s > d) /* Found by Ilya: feed random input to Perl. */
2763 Perl_croak(aTHX_ "panic: input overflow");
2765 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2767 PL_lex_state = LEX_FORMLINE;
2777 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2784 while (s < PL_bufend && SPACE_OR_TAB(*s))
2787 if (strnEQ(s,"=>",2)) {
2788 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2789 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2790 "### Saw unary minus before =>, forcing word '%s'\n", s);
2792 OPERATOR('-'); /* unary minus */
2794 PL_last_uni = PL_oldbufptr;
2796 case 'r': ftst = OP_FTEREAD; break;
2797 case 'w': ftst = OP_FTEWRITE; break;
2798 case 'x': ftst = OP_FTEEXEC; break;
2799 case 'o': ftst = OP_FTEOWNED; break;
2800 case 'R': ftst = OP_FTRREAD; break;
2801 case 'W': ftst = OP_FTRWRITE; break;
2802 case 'X': ftst = OP_FTREXEC; break;
2803 case 'O': ftst = OP_FTROWNED; break;
2804 case 'e': ftst = OP_FTIS; break;
2805 case 'z': ftst = OP_FTZERO; break;
2806 case 's': ftst = OP_FTSIZE; break;
2807 case 'f': ftst = OP_FTFILE; break;
2808 case 'd': ftst = OP_FTDIR; break;
2809 case 'l': ftst = OP_FTLINK; break;
2810 case 'p': ftst = OP_FTPIPE; break;
2811 case 'S': ftst = OP_FTSOCK; break;
2812 case 'u': ftst = OP_FTSUID; break;
2813 case 'g': ftst = OP_FTSGID; break;
2814 case 'k': ftst = OP_FTSVTX; break;
2815 case 'b': ftst = OP_FTBLK; break;
2816 case 'c': ftst = OP_FTCHR; break;
2817 case 't': ftst = OP_FTTTY; break;
2818 case 'T': ftst = OP_FTTEXT; break;
2819 case 'B': ftst = OP_FTBINARY; break;
2820 case 'M': case 'A': case 'C':
2821 gv_fetchpv("\024",TRUE, SVt_PV);
2823 case 'M': ftst = OP_FTMTIME; break;
2824 case 'A': ftst = OP_FTATIME; break;
2825 case 'C': ftst = OP_FTCTIME; break;
2833 PL_last_lop_op = (OPCODE)ftst;
2834 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2835 "### Saw file test %c\n", (int)ftst);
2840 /* Assume it was a minus followed by a one-letter named
2841 * subroutine call (or a -bareword), then. */
2842 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2843 "### %c looked like a file test but was not\n",
2852 if (PL_expect == XOPERATOR)
2857 else if (*s == '>') {
2860 if (isIDFIRST_lazy_if(s,UTF)) {
2861 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2869 if (PL_expect == XOPERATOR)
2872 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2874 OPERATOR('-'); /* unary minus */
2881 if (PL_expect == XOPERATOR)
2886 if (PL_expect == XOPERATOR)
2889 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2895 if (PL_expect != XOPERATOR) {
2896 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2897 PL_expect = XOPERATOR;
2898 force_ident(PL_tokenbuf, '*');
2911 if (PL_expect == XOPERATOR) {
2915 PL_tokenbuf[0] = '%';
2916 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2917 if (!PL_tokenbuf[1]) {
2919 yyerror("Final % should be \\% or %name");
2922 PL_pending_ident = '%';
2941 switch (PL_expect) {
2944 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2946 PL_bufptr = s; /* update in case we back off */
2952 PL_expect = XTERMBLOCK;
2956 while (isIDFIRST_lazy_if(s,UTF)) {
2957 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2958 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2959 if (tmp < 0) tmp = -tmp;
2974 d = scan_str(d,TRUE,TRUE);
2976 /* MUST advance bufptr here to avoid bogus
2977 "at end of line" context messages from yyerror().
2979 PL_bufptr = s + len;
2980 yyerror("Unterminated attribute parameter in attribute list");
2983 return 0; /* EOF indicator */
2987 SV *sv = newSVpvn(s, len);
2988 sv_catsv(sv, PL_lex_stuff);
2989 attrs = append_elem(OP_LIST, attrs,
2990 newSVOP(OP_CONST, 0, sv));
2991 SvREFCNT_dec(PL_lex_stuff);
2992 PL_lex_stuff = Nullsv;
2995 /* NOTE: any CV attrs applied here need to be part of
2996 the CVf_BUILTIN_ATTRS define in cv.h! */
2997 if (!PL_in_my && len == 6 && strnEQ(s, "lvalue", len))
2998 CvLVALUE_on(PL_compcv);
2999 else if (!PL_in_my && len == 6 && strnEQ(s, "locked", len))
3000 CvLOCKED_on(PL_compcv);
3001 else if (!PL_in_my && len == 6 && strnEQ(s, "method", len))
3002 CvMETHOD_on(PL_compcv);
3004 else if (PL_in_my == KEY_our && len == 6 &&
3005 strnEQ(s, "unique", len))
3006 GvUNIQUE_on(cGVOPx_gv(yylval.opval));
3008 /* After we've set the flags, it could be argued that
3009 we don't need to do the attributes.pm-based setting
3010 process, and shouldn't bother appending recognized
3011 flags. To experiment with that, uncomment the
3012 following "else". (Note that's already been
3013 uncommented. That keeps the above-applied built-in
3014 attributes from being intercepted (and possibly
3015 rejected) by a package's attribute routines, but is
3016 justified by the performance win for the common case
3017 of applying only built-in attributes.) */
3019 attrs = append_elem(OP_LIST, attrs,
3020 newSVOP(OP_CONST, 0,
3024 if (*s == ':' && s[1] != ':')
3027 break; /* require real whitespace or :'s */
3029 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
3030 if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) {
3031 char q = ((*s == '\'') ? '"' : '\'');
3032 /* If here for an expression, and parsed no attrs, back off. */
3033 if (tmp == '=' && !attrs) {
3037 /* MUST advance bufptr here to avoid bogus "at end of line"
3038 context messages from yyerror().
3042 yyerror("Unterminated attribute list");
3044 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
3052 PL_nextval[PL_nexttoke].opval = attrs;
3060 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
3061 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
3077 if (PL_lex_brackets <= 0)
3078 yyerror("Unmatched right square bracket");
3081 if (PL_lex_state == LEX_INTERPNORMAL) {
3082 if (PL_lex_brackets == 0) {
3083 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
3084 PL_lex_state = LEX_INTERPEND;
3091 if (PL_lex_brackets > 100) {
3092 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
3093 if (newlb != PL_lex_brackstack) {
3095 PL_lex_brackstack = newlb;
3098 switch (PL_expect) {
3100 if (PL_lex_formbrack) {
3104 if (PL_oldoldbufptr == PL_last_lop)
3105 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3107 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3108 OPERATOR(HASHBRACK);
3110 while (s < PL_bufend && SPACE_OR_TAB(*s))
3113 PL_tokenbuf[0] = '\0';
3114 if (d < PL_bufend && *d == '-') {
3115 PL_tokenbuf[0] = '-';
3117 while (d < PL_bufend && SPACE_OR_TAB(*d))
3120 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3121 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3123 while (d < PL_bufend && SPACE_OR_TAB(*d))
3126 char minus = (PL_tokenbuf[0] == '-');
3127 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3135 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3140 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3145 if (PL_oldoldbufptr == PL_last_lop)
3146 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3148 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3151 if (PL_expect == XREF && PL_lex_state == LEX_INTERPNORMAL) {
3153 /* This hack is to get the ${} in the message. */
3155 yyerror("syntax error");
3158 OPERATOR(HASHBRACK);
3160 /* This hack serves to disambiguate a pair of curlies
3161 * as being a block or an anon hash. Normally, expectation
3162 * determines that, but in cases where we're not in a
3163 * position to expect anything in particular (like inside
3164 * eval"") we have to resolve the ambiguity. This code
3165 * covers the case where the first term in the curlies is a
3166 * quoted string. Most other cases need to be explicitly
3167 * disambiguated by prepending a `+' before the opening
3168 * curly in order to force resolution as an anon hash.
3170 * XXX should probably propagate the outer expectation
3171 * into eval"" to rely less on this hack, but that could
3172 * potentially break current behavior of eval"".
3176 if (*s == '\'' || *s == '"' || *s == '`') {
3177 /* common case: get past first string, handling escapes */
3178 for (t++; t < PL_bufend && *t != *s;)
3179 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3183 else if (*s == 'q') {
3186 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3190 char open, close, term;
3193 while (t < PL_bufend && isSPACE(*t))
3197 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3201 for (t++; t < PL_bufend; t++) {
3202 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3204 else if (*t == open)
3208 for (t++; t < PL_bufend; t++) {
3209 if (*t == '\\' && t+1 < PL_bufend)
3211 else if (*t == close && --brackets <= 0)
3213 else if (*t == open)
3219 else if (isALNUM_lazy_if(t,UTF)) {
3221 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3224 while (t < PL_bufend && isSPACE(*t))
3226 /* if comma follows first term, call it an anon hash */
3227 /* XXX it could be a comma expression with loop modifiers */
3228 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3229 || (*t == '=' && t[1] == '>')))
3230 OPERATOR(HASHBRACK);
3231 if (PL_expect == XREF)
3234 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3240 yylval.ival = CopLINE(PL_curcop);
3241 if (isSPACE(*s) || *s == '#')
3242 PL_copline = NOLINE; /* invalidate current command line number */
3247 if (PL_lex_brackets <= 0)
3248 yyerror("Unmatched right curly bracket");
3250 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3251 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3252 PL_lex_formbrack = 0;
3253 if (PL_lex_state == LEX_INTERPNORMAL) {
3254 if (PL_lex_brackets == 0) {
3255 if (PL_expect & XFAKEBRACK) {
3256 PL_expect &= XENUMMASK;
3257 PL_lex_state = LEX_INTERPEND;
3259 return yylex(); /* ignore fake brackets */
3261 if (*s == '-' && s[1] == '>')
3262 PL_lex_state = LEX_INTERPENDMAYBE;
3263 else if (*s != '[' && *s != '{')
3264 PL_lex_state = LEX_INTERPEND;
3267 if (PL_expect & XFAKEBRACK) {
3268 PL_expect &= XENUMMASK;
3270 return yylex(); /* ignore fake brackets */
3280 if (PL_expect == XOPERATOR) {
3281 if (ckWARN(WARN_SEMICOLON)
3282 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3284 CopLINE_dec(PL_curcop);
3285 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
3286 CopLINE_inc(PL_curcop);
3291 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3293 PL_expect = XOPERATOR;
3294 force_ident(PL_tokenbuf, '&');
3298 yylval.ival = (OPpENTERSUB_AMPER<<8);
3317 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3318 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Reversed %c= operator",(int)tmp);
3320 if (PL_expect == XSTATE && isALPHA(tmp) &&
3321 (s == PL_linestart+1 || s[-2] == '\n') )
3323 if (PL_in_eval && !PL_rsfp) {
3328 if (strnEQ(s,"=cut",4)) {
3342 PL_doextract = TRUE;
3345 if (PL_lex_brackets < PL_lex_formbrack) {
3347 #ifdef PERL_STRICT_CR
3348 for (t = s; SPACE_OR_TAB(*t); t++) ;
3350 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3352 if (*t == '\n' || *t == '#') {
3370 if (PL_expect != XOPERATOR) {
3371 if (s[1] != '<' && !strchr(s,'>'))
3374 s = scan_heredoc(s);
3376 s = scan_inputsymbol(s);
3377 TERM(sublex_start());
3382 SHop(OP_LEFT_SHIFT);
3396 SHop(OP_RIGHT_SHIFT);
3405 if (PL_expect == XOPERATOR) {
3406 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3409 return ','; /* grandfather non-comma-format format */
3413 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3414 PL_tokenbuf[0] = '@';
3415 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3416 sizeof PL_tokenbuf - 1, FALSE);
3417 if (PL_expect == XOPERATOR)
3418 no_op("Array length", s);
3419 if (!PL_tokenbuf[1])
3421 PL_expect = XOPERATOR;
3422 PL_pending_ident = '#';
3426 PL_tokenbuf[0] = '$';
3427 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3428 sizeof PL_tokenbuf - 1, FALSE);
3429 if (PL_expect == XOPERATOR)
3431 if (!PL_tokenbuf[1]) {
3433 yyerror("Final $ should be \\$ or $name");
3437 /* This kludge not intended to be bulletproof. */
3438 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3439 yylval.opval = newSVOP(OP_CONST, 0,
3440 newSViv(PL_compiling.cop_arybase));
3441 yylval.opval->op_private = OPpCONST_ARYBASE;
3447 if (PL_lex_state == LEX_NORMAL)
3450 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3453 PL_tokenbuf[0] = '@';
3454 if (ckWARN(WARN_SYNTAX)) {
3456 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3459 PL_bufptr = skipspace(PL_bufptr);
3460 while (t < PL_bufend && *t != ']')
3462 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3463 "Multidimensional syntax %.*s not supported",
3464 (t - PL_bufptr) + 1, PL_bufptr);
3468 else if (*s == '{') {
3469 PL_tokenbuf[0] = '%';
3470 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3471 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3473 char tmpbuf[sizeof PL_tokenbuf];
3475 for (t++; isSPACE(*t); t++) ;
3476 if (isIDFIRST_lazy_if(t,UTF)) {
3477 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3478 for (; isSPACE(*t); t++) ;
3479 if (*t == ';' && get_cv(tmpbuf, FALSE))
3480 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3481 "You need to quote \"%s\"", tmpbuf);
3487 PL_expect = XOPERATOR;
3488 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3489 bool islop = (PL_last_lop == PL_oldoldbufptr);
3490 if (!islop || PL_last_lop_op == OP_GREPSTART)
3491 PL_expect = XOPERATOR;
3492 else if (strchr("$@\"'`q", *s))
3493 PL_expect = XTERM; /* e.g. print $fh "foo" */
3494 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3495 PL_expect = XTERM; /* e.g. print $fh &sub */
3496 else if (isIDFIRST_lazy_if(s,UTF)) {
3497 char tmpbuf[sizeof PL_tokenbuf];
3498 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3499 if ((tmp = keyword(tmpbuf, len))) {
3500 /* binary operators exclude handle interpretations */
3512 PL_expect = XTERM; /* e.g. print $fh length() */
3517 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3518 if (gv && GvCVu(gv))
3519 PL_expect = XTERM; /* e.g. print $fh subr() */
3522 else if (isDIGIT(*s))
3523 PL_expect = XTERM; /* e.g. print $fh 3 */
3524 else if (*s == '.' && isDIGIT(s[1]))
3525 PL_expect = XTERM; /* e.g. print $fh .3 */
3526 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3527 PL_expect = XTERM; /* e.g. print $fh -1 */
3528 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3529 PL_expect = XTERM; /* print $fh <<"EOF" */
3531 PL_pending_ident = '$';
3535 if (PL_expect == XOPERATOR)
3537 PL_tokenbuf[0] = '@';
3538 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3539 if (!PL_tokenbuf[1]) {
3541 yyerror("Final @ should be \\@ or @name");
3544 if (PL_lex_state == LEX_NORMAL)
3546 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3548 PL_tokenbuf[0] = '%';
3550 /* Warn about @ where they meant $. */
3551 if (ckWARN(WARN_SYNTAX)) {
3552 if (*s == '[' || *s == '{') {
3554 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3556 if (*t == '}' || *t == ']') {
3558 PL_bufptr = skipspace(PL_bufptr);
3559 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3560 "Scalar value %.*s better written as $%.*s",
3561 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3566 PL_pending_ident = '@';
3569 case '/': /* may either be division or pattern */
3570 case '?': /* may either be conditional or pattern */
3571 if (PL_expect != XOPERATOR) {
3572 /* Disable warning on "study /blah/" */
3573 if (PL_oldoldbufptr == PL_last_uni
3574 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3575 || memNE(PL_last_uni, "study", 5)
3576 || isALNUM_lazy_if(PL_last_uni+5,UTF)))
3578 s = scan_pat(s,OP_MATCH);
3579 TERM(sublex_start());
3587 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3588 #ifdef PERL_STRICT_CR
3591 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3593 && (s == PL_linestart || s[-1] == '\n') )
3595 PL_lex_formbrack = 0;
3599 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3605 yylval.ival = OPf_SPECIAL;
3611 if (PL_expect != XOPERATOR)
3616 case '0': case '1': case '2': case '3': case '4':
3617 case '5': case '6': case '7': case '8': case '9':
3618 s = scan_num(s, &yylval);
3619 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3620 "### Saw number in '%s'\n", s);
3622 if (PL_expect == XOPERATOR)
3627 s = scan_str(s,FALSE,FALSE);
3628 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3629 "### Saw string before '%s'\n", s);
3631 if (PL_expect == XOPERATOR) {
3632 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3635 return ','; /* grandfather non-comma-format format */
3641 missingterm((char*)0);
3642 yylval.ival = OP_CONST;
3643 TERM(sublex_start());
3646 s = scan_str(s,FALSE,FALSE);
3647 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3648 "### Saw string before '%s'\n", s);
3650 if (PL_expect == XOPERATOR) {
3651 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3654 return ','; /* grandfather non-comma-format format */
3660 missingterm((char*)0);
3661 yylval.ival = OP_CONST;
3662 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3663 if (*d == '$' || *d == '@' || *d == '\\' || !UTF8_IS_INVARIANT((U8)*d)) {
3664 yylval.ival = OP_STRINGIFY;
3668 TERM(sublex_start());
3671 s = scan_str(s,FALSE,FALSE);
3672 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3673 "### Saw backtick string before '%s'\n", s);
3675 if (PL_expect == XOPERATOR)
3676 no_op("Backticks",s);
3678 missingterm((char*)0);
3679 yylval.ival = OP_BACKTICK;
3681 TERM(sublex_start());
3685 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3686 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),"Can't use \\%c to mean $%c in expression",
3688 if (PL_expect == XOPERATOR)
3689 no_op("Backslash",s);
3693 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3697 while (isDIGIT(*start) || *start == '_')
3699 if (*start == '.' && isDIGIT(start[1])) {
3700 s = scan_num(s, &yylval);
3703 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3704 else if (!isALPHA(*start) && (PL_expect == XTERM || PL_expect == XREF || PL_expect == XSTATE)) {
3708 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3711 s = scan_num(s, &yylval);
3718 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3757 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3759 /* Some keywords can be followed by any delimiter, including ':' */
3760 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
3761 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3762 (PL_tokenbuf[0] == 'q' &&
3763 strchr("qwxr", PL_tokenbuf[1])))));
3765 /* x::* is just a word, unless x is "CORE" */
3766 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3770 while (d < PL_bufend && isSPACE(*d))
3771 d++; /* no comments skipped here, or s### is misparsed */
3773 /* Is this a label? */
3774 if (!tmp && PL_expect == XSTATE
3775 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3777 yylval.pval = savepv(PL_tokenbuf);
3782 /* Check for keywords */
3783 tmp = keyword(PL_tokenbuf, len);
3785 /* Is this a word before a => operator? */
3786 if (*d == '=' && d[1] == '>') {
3788 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3789 yylval.opval->op_private = OPpCONST_BARE;
3790 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
3791 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
3795 if (tmp < 0) { /* second-class keyword? */
3796 GV *ogv = Nullgv; /* override (winner) */
3797 GV *hgv = Nullgv; /* hidden (loser) */
3798 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3800 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3803 if (GvIMPORTED_CV(gv))
3805 else if (! CvMETHOD(cv))
3809 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3810 (gv = *gvp) != (GV*)&PL_sv_undef &&
3811 GvCVu(gv) && GvIMPORTED_CV(gv))
3817 tmp = 0; /* overridden by import or by GLOBAL */
3820 && -tmp==KEY_lock /* XXX generalizable kludge */
3822 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3824 tmp = 0; /* any sub overrides "weak" keyword */
3826 else { /* no override */
3828 if (tmp == KEY_dump && ckWARN(WARN_MISC)) {
3829 Perl_warner(aTHX_ packWARN(WARN_MISC),
3830 "dump() better written as CORE::dump()");
3834 if (ckWARN(WARN_AMBIGUOUS) && hgv
3835 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3836 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
3837 "Ambiguous call resolved as CORE::%s(), %s",
3838 GvENAME(hgv), "qualify as such or use &");
3845 default: /* not a keyword */
3849 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3851 /* Get the rest if it looks like a package qualifier */
3853 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
3855 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3858 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3859 *s == '\'' ? "'" : "::");
3864 if (PL_expect == XOPERATOR) {
3865 if (PL_bufptr == PL_linestart) {
3866 CopLINE_dec(PL_curcop);
3867 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
3868 CopLINE_inc(PL_curcop);
3871 no_op("Bareword",s);
3874 /* Look for a subroutine with this name in current package,
3875 unless name is "Foo::", in which case Foo is a bearword
3876 (and a package name). */
3879 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3881 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3882 Perl_warner(aTHX_ packWARN(WARN_BAREWORD),
3883 "Bareword \"%s\" refers to nonexistent package",
3886 PL_tokenbuf[len] = '\0';
3893 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3896 /* if we saw a global override before, get the right name */
3899 sv = newSVpvn("CORE::GLOBAL::",14);
3900 sv_catpv(sv,PL_tokenbuf);
3903 sv = newSVpv(PL_tokenbuf,0);
3905 /* Presume this is going to be a bareword of some sort. */
3908 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3909 yylval.opval->op_private = OPpCONST_BARE;
3911 /* And if "Foo::", then that's what it certainly is. */
3916 /* See if it's the indirect object for a list operator. */
3918 if (PL_oldoldbufptr &&
3919 PL_oldoldbufptr < PL_bufptr &&
3920 (PL_oldoldbufptr == PL_last_lop
3921 || PL_oldoldbufptr == PL_last_uni) &&
3922 /* NO SKIPSPACE BEFORE HERE! */
3923 (PL_expect == XREF ||
3924 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3926 bool immediate_paren = *s == '(';
3928 /* (Now we can afford to cross potential line boundary.) */
3931 /* Two barewords in a row may indicate method call. */
3933 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
3936 /* If not a declared subroutine, it's an indirect object. */
3937 /* (But it's an indir obj regardless for sort.) */
3939 if ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
3940 ((!gv || !GvCVu(gv)) &&
3941 (PL_last_lop_op != OP_MAPSTART &&
3942 PL_last_lop_op != OP_GREPSTART))))
3944 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3949 PL_expect = XOPERATOR;
3952 /* Is this a word before a => operator? */
3953 if (*s == '=' && s[1] == '>' && !pkgname) {
3955 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
3956 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
3957 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
3961 /* If followed by a paren, it's certainly a subroutine. */
3964 if (gv && GvCVu(gv)) {
3965 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
3966 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3971 PL_nextval[PL_nexttoke].opval = yylval.opval;
3972 PL_expect = XOPERATOR;
3978 /* If followed by var or block, call it a method (unless sub) */
3980 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3981 PL_last_lop = PL_oldbufptr;
3982 PL_last_lop_op = OP_METHOD;
3986 /* If followed by a bareword, see if it looks like indir obj. */
3988 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp = intuit_method(s,gv)))
3991 /* Not a method, so call it a subroutine (if defined) */
3993 if (gv && GvCVu(gv)) {
3995 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
3996 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
3997 "Ambiguous use of -%s resolved as -&%s()",
3998 PL_tokenbuf, PL_tokenbuf);
3999 /* Check for a constant sub */
4001 if ((sv = cv_const_sv(cv))) {
4003 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
4004 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
4005 yylval.opval->op_private = 0;
4009 /* Resolve to GV now. */
4010 op_free(yylval.opval);
4011 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
4012 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
4013 PL_last_lop = PL_oldbufptr;
4014 PL_last_lop_op = OP_ENTERSUB;
4015 /* Is there a prototype? */
4018 char *proto = SvPV((SV*)cv, len);
4021 if (strEQ(proto, "$"))
4023 if (*proto == '&' && *s == '{') {
4024 sv_setpv(PL_subname, PL_curstash ?
4025 "__ANON__" : "__ANON__::__ANON__");
4029 PL_nextval[PL_nexttoke].opval = yylval.opval;
4035 /* Call it a bare word */
4037 if (PL_hints & HINT_STRICT_SUBS)
4038 yylval.opval->op_private |= OPpCONST_STRICT;
4041 if (ckWARN(WARN_RESERVED)) {
4042 if (lastchar != '-') {
4043 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
4044 if (!*d && strNE(PL_tokenbuf,"main"))
4045 Perl_warner(aTHX_ packWARN(WARN_RESERVED), PL_warn_reserved,
4052 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
4053 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4054 "Operator or semicolon missing before %c%s",
4055 lastchar, PL_tokenbuf);
4056 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4057 "Ambiguous use of %c resolved as operator %c",
4058 lastchar, lastchar);
4064 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4065 newSVpv(CopFILE(PL_curcop),0));
4069 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4070 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
4073 case KEY___PACKAGE__:
4074 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4076 ? newSVsv(PL_curstname)
4085 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
4086 char *pname = "main";
4087 if (PL_tokenbuf[2] == 'D')
4088 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
4089 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
4092 GvIOp(gv) = newIO();
4093 IoIFP(GvIOp(gv)) = PL_rsfp;
4094 #if defined(HAS_FCNTL) && defined(F_SETFD)
4096 int fd = PerlIO_fileno(PL_rsfp);
4097 fcntl(fd,F_SETFD,fd >= 3);
4100 /* Mark this internal pseudo-handle as clean */
4101 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
4103 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
4104 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
4105 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
4107 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
4108 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
4109 /* if the script was opened in binmode, we need to revert
4110 * it to text mode for compatibility; but only iff it has CRs
4111 * XXX this is a questionable hack at best. */
4112 if (PL_bufend-PL_bufptr > 2
4113 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
4116 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
4117 loc = PerlIO_tell(PL_rsfp);
4118 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4121 if (PerlLIO_setmode(PL_rsfp, O_TEXT) != -1) {
4123 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4124 #endif /* NETWARE */
4125 #ifdef PERLIO_IS_STDIO /* really? */
4126 # if defined(__BORLANDC__)
4127 /* XXX see note in do_binmode() */
4128 ((FILE*)PL_rsfp)->flags &= ~_F_BIN;
4132 PerlIO_seek(PL_rsfp, loc, 0);
4136 #ifdef PERLIO_LAYERS
4137 if (UTF && !IN_BYTES)
4138 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
4151 if (PL_expect == XSTATE) {
4158 if (*s == ':' && s[1] == ':') {
4161 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4162 if (!(tmp = keyword(PL_tokenbuf, len)))
4163 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4177 LOP(OP_ACCEPT,XTERM);
4183 LOP(OP_ATAN2,XTERM);
4189 LOP(OP_BINMODE,XTERM);
4192 LOP(OP_BLESS,XTERM);
4201 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4218 if (!PL_cryptseen) {
4219 PL_cryptseen = TRUE;
4223 LOP(OP_CRYPT,XTERM);
4226 LOP(OP_CHMOD,XTERM);
4229 LOP(OP_CHOWN,XTERM);
4232 LOP(OP_CONNECT,XTERM);
4248 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4252 PL_hints |= HINT_BLOCK_SCOPE;
4262 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4263 LOP(OP_DBMOPEN,XTERM);
4269 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4276 yylval.ival = CopLINE(PL_curcop);
4290 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4291 UNIBRACK(OP_ENTEREVAL);
4306 case KEY_endhostent:
4312 case KEY_endservent:
4315 case KEY_endprotoent:
4326 yylval.ival = CopLINE(PL_curcop);
4328 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4330 if ((PL_bufend - p) >= 3 &&
4331 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4333 else if ((PL_bufend - p) >= 4 &&
4334 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4337 if (isIDFIRST_lazy_if(p,UTF)) {
4338 p = scan_ident(p, PL_bufend,
4339 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4343 Perl_croak(aTHX_ "Missing $ on loop variable");
4348 LOP(OP_FORMLINE,XTERM);
4354 LOP(OP_FCNTL,XTERM);
4360 LOP(OP_FLOCK,XTERM);
4369 LOP(OP_GREPSTART, XREF);
4372 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4387 case KEY_getpriority:
4388 LOP(OP_GETPRIORITY,XTERM);
4390 case KEY_getprotobyname:
4393 case KEY_getprotobynumber:
4394 LOP(OP_GPBYNUMBER,XTERM);
4396 case KEY_getprotoent:
4408 case KEY_getpeername:
4409 UNI(OP_GETPEERNAME);
4411 case KEY_gethostbyname:
4414 case KEY_gethostbyaddr:
4415 LOP(OP_GHBYADDR,XTERM);
4417 case KEY_gethostent:
4420 case KEY_getnetbyname:
4423 case KEY_getnetbyaddr:
4424 LOP(OP_GNBYADDR,XTERM);
4429 case KEY_getservbyname:
4430 LOP(OP_GSBYNAME,XTERM);
4432 case KEY_getservbyport:
4433 LOP(OP_GSBYPORT,XTERM);
4435 case KEY_getservent:
4438 case KEY_getsockname:
4439 UNI(OP_GETSOCKNAME);
4441 case KEY_getsockopt:
4442 LOP(OP_GSOCKOPT,XTERM);
4464 yylval.ival = CopLINE(PL_curcop);
4468 LOP(OP_INDEX,XTERM);
4474 LOP(OP_IOCTL,XTERM);
4486 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4518 LOP(OP_LISTEN,XTERM);
4527 s = scan_pat(s,OP_MATCH);
4528 TERM(sublex_start());
4531 LOP(OP_MAPSTART, XREF);
4534 LOP(OP_MKDIR,XTERM);
4537 LOP(OP_MSGCTL,XTERM);
4540 LOP(OP_MSGGET,XTERM);
4543 LOP(OP_MSGRCV,XTERM);
4546 LOP(OP_MSGSND,XTERM);
4552 if (isIDFIRST_lazy_if(s,UTF)) {
4553 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4554 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4556 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
4557 if (!PL_in_my_stash) {
4560 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4568 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4575 if (PL_expect != XSTATE)
4576 yyerror("\"no\" not allowed in expression");
4577 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4578 s = force_version(s, FALSE);
4583 if (*s == '(' || (s = skipspace(s), *s == '('))
4590 if (isIDFIRST_lazy_if(s,UTF)) {
4592 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4594 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE))
4595 Perl_warner(aTHX_ packWARN(WARN_PRECEDENCE),
4596 "Precedence problem: open %.*s should be open(%.*s)",
4602 yylval.ival = OP_OR;
4612 LOP(OP_OPEN_DIR,XTERM);
4615 checkcomma(s,PL_tokenbuf,"filehandle");
4619 checkcomma(s,PL_tokenbuf,"filehandle");
4638 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4642 LOP(OP_PIPE_OP,XTERM);
4645 s = scan_str(s,FALSE,FALSE);
4647 missingterm((char*)0);
4648 yylval.ival = OP_CONST;
4649 TERM(sublex_start());
4655 s = scan_str(s,FALSE,FALSE);
4657 missingterm((char*)0);
4659 if (SvCUR(PL_lex_stuff)) {
4662 d = SvPV_force(PL_lex_stuff, len);
4665 for (; isSPACE(*d) && len; --len, ++d) ;
4668 if (!warned && ckWARN(WARN_QW)) {
4669 for (; !isSPACE(*d) && len; --len, ++d) {
4671 Perl_warner(aTHX_ packWARN(WARN_QW),
4672 "Possible attempt to separate words with commas");
4675 else if (*d == '#') {
4676 Perl_warner(aTHX_ packWARN(WARN_QW),
4677 "Possible attempt to put comments in qw() list");
4683 for (; !isSPACE(*d) && len; --len, ++d) ;
4685 sv = newSVpvn(b, d-b);
4686 if (DO_UTF8(PL_lex_stuff))
4688 words = append_elem(OP_LIST, words,
4689 newSVOP(OP_CONST, 0, tokeq(sv)));
4693 PL_nextval[PL_nexttoke].opval = words;
4698 SvREFCNT_dec(PL_lex_stuff);
4699 PL_lex_stuff = Nullsv;
4705 s = scan_str(s,FALSE,FALSE);
4707 missingterm((char*)0);
4708 yylval.ival = OP_STRINGIFY;
4709 if (SvIVX(PL_lex_stuff) == '\'')
4710 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4711 TERM(sublex_start());
4714 s = scan_pat(s,OP_QR);
4715 TERM(sublex_start());
4718 s = scan_str(s,FALSE,FALSE);
4720 missingterm((char*)0);
4721 yylval.ival = OP_BACKTICK;
4723 TERM(sublex_start());
4731 s = force_version(s, FALSE);
4733 else if (*s != 'v' || !isDIGIT(s[1])
4734 || (s = force_version(s, TRUE), *s == 'v'))
4736 *PL_tokenbuf = '\0';
4737 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4738 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4739 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4741 yyerror("<> should be quotes");
4749 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4753 LOP(OP_RENAME,XTERM);
4762 LOP(OP_RINDEX,XTERM);
4785 LOP(OP_REVERSE,XTERM);
4796 TERM(sublex_start());
4798 TOKEN(1); /* force error */
4807 LOP(OP_SELECT,XTERM);
4813 LOP(OP_SEMCTL,XTERM);
4816 LOP(OP_SEMGET,XTERM);
4819 LOP(OP_SEMOP,XTERM);
4825 LOP(OP_SETPGRP,XTERM);
4827 case KEY_setpriority:
4828 LOP(OP_SETPRIORITY,XTERM);
4830 case KEY_sethostent:
4836 case KEY_setservent:
4839 case KEY_setprotoent:
4849 LOP(OP_SEEKDIR,XTERM);
4851 case KEY_setsockopt:
4852 LOP(OP_SSOCKOPT,XTERM);
4858 LOP(OP_SHMCTL,XTERM);
4861 LOP(OP_SHMGET,XTERM);
4864 LOP(OP_SHMREAD,XTERM);
4867 LOP(OP_SHMWRITE,XTERM);
4870 LOP(OP_SHUTDOWN,XTERM);
4879 LOP(OP_SOCKET,XTERM);
4881 case KEY_socketpair:
4882 LOP(OP_SOCKPAIR,XTERM);
4885 checkcomma(s,PL_tokenbuf,"subroutine name");
4887 if (*s == ';' || *s == ')') /* probably a close */
4888 Perl_croak(aTHX_ "sort is now a reserved word");
4890 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4894 LOP(OP_SPLIT,XTERM);
4897 LOP(OP_SPRINTF,XTERM);
4900 LOP(OP_SPLICE,XTERM);
4915 LOP(OP_SUBSTR,XTERM);
4921 char tmpbuf[sizeof PL_tokenbuf];
4922 SSize_t tboffset = 0;
4923 expectation attrful;
4924 bool have_name, have_proto, bad_proto;
4929 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
4930 (*s == ':' && s[1] == ':'))
4933 attrful = XATTRBLOCK;
4934 /* remember buffer pos'n for later force_word */
4935 tboffset = s - PL_oldbufptr;
4936 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4937 if (strchr(tmpbuf, ':'))
4938 sv_setpv(PL_subname, tmpbuf);
4940 sv_setsv(PL_subname,PL_curstname);
4941 sv_catpvn(PL_subname,"::",2);
4942 sv_catpvn(PL_subname,tmpbuf,len);
4949 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4950 PL_expect = XTERMBLOCK;
4951 attrful = XATTRTERM;
4952 sv_setpv(PL_subname,"?");
4956 if (key == KEY_format) {
4958 PL_lex_formbrack = PL_lex_brackets + 1;
4960 (void) force_word(PL_oldbufptr + tboffset, WORD,
4965 /* Look for a prototype */
4969 s = scan_str(s,FALSE,FALSE);
4971 Perl_croak(aTHX_ "Prototype not terminated");
4972 /* strip spaces and check for bad characters */
4973 d = SvPVX(PL_lex_stuff);
4976 for (p = d; *p; ++p) {
4979 if (!strchr("$@%*;[]&\\", *p))
4984 if (bad_proto && ckWARN(WARN_SYNTAX))
4985 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4986 "Illegal character in prototype for %s : %s",
4987 SvPVX(PL_subname), d);
4988 SvCUR(PL_lex_stuff) = tmp;
4996 if (*s == ':' && s[1] != ':')
4997 PL_expect = attrful;
5000 PL_nextval[PL_nexttoke].opval =
5001 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
5002 PL_lex_stuff = Nullsv;
5006 sv_setpv(PL_subname,
5007 PL_curstash ? "__ANON__" : "__ANON__::__ANON__");
5010 (void) force_word(PL_oldbufptr + tboffset, WORD,
5019 LOP(OP_SYSTEM,XREF);
5022 LOP(OP_SYMLINK,XTERM);
5025 LOP(OP_SYSCALL,XTERM);
5028 LOP(OP_SYSOPEN,XTERM);
5031 LOP(OP_SYSSEEK,XTERM);
5034 LOP(OP_SYSREAD,XTERM);
5037 LOP(OP_SYSWRITE,XTERM);
5041 TERM(sublex_start());
5062 LOP(OP_TRUNCATE,XTERM);
5074 yylval.ival = CopLINE(PL_curcop);
5078 yylval.ival = CopLINE(PL_curcop);
5082 LOP(OP_UNLINK,XTERM);
5088 LOP(OP_UNPACK,XTERM);
5091 LOP(OP_UTIME,XTERM);
5097 LOP(OP_UNSHIFT,XTERM);
5100 if (PL_expect != XSTATE)
5101 yyerror("\"use\" not allowed in expression");
5103 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
5104 s = force_version(s, TRUE);
5105 if (*s == ';' || (s = skipspace(s), *s == ';')) {
5106 PL_nextval[PL_nexttoke].opval = Nullop;
5109 else if (*s == 'v') {
5110 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5111 s = force_version(s, FALSE);
5115 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5116 s = force_version(s, FALSE);
5128 yylval.ival = CopLINE(PL_curcop);
5132 PL_hints |= HINT_BLOCK_SCOPE;
5139 LOP(OP_WAITPID,XTERM);
5148 ctl_l[0] = toCTRL('L');
5150 gv_fetchpv(ctl_l,TRUE, SVt_PV);
5153 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
5158 if (PL_expect == XOPERATOR)
5164 yylval.ival = OP_XOR;
5169 TERM(sublex_start());
5174 #pragma segment Main
5178 S_pending_ident(pTHX)
5182 /* pit holds the identifier we read and pending_ident is reset */
5183 char pit = PL_pending_ident;
5184 PL_pending_ident = 0;
5186 DEBUG_T({ PerlIO_printf(Perl_debug_log,
5187 "### Tokener saw identifier '%s'\n", PL_tokenbuf); });
5189 /* if we're in a my(), we can't allow dynamics here.
5190 $foo'bar has already been turned into $foo::bar, so
5191 just check for colons.
5193 if it's a legal name, the OP is a PADANY.
5196 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
5197 if (strchr(PL_tokenbuf,':'))
5198 yyerror(Perl_form(aTHX_ "No package name allowed for "
5199 "variable %s in \"our\"",
5201 tmp = pad_allocmy(PL_tokenbuf);
5204 if (strchr(PL_tokenbuf,':'))
5205 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
5207 yylval.opval = newOP(OP_PADANY, 0);
5208 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
5214 build the ops for accesses to a my() variable.
5216 Deny my($a) or my($b) in a sort block, *if* $a or $b is
5217 then used in a comparison. This catches most, but not
5218 all cases. For instance, it catches
5219 sort { my($a); $a <=> $b }
5221 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
5222 (although why you'd do that is anyone's guess).
5225 if (!strchr(PL_tokenbuf,':')) {
5226 #ifdef USE_5005THREADS
5227 /* Check for single character per-thread SVs */
5228 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
5229 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
5230 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
5232 yylval.opval = newOP(OP_THREADSV, 0);
5233 yylval.opval->op_targ = tmp;
5236 #endif /* USE_5005THREADS */
5237 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
5238 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
5239 /* might be an "our" variable" */
5240 if (SvFLAGS(namesv) & SVpad_OUR) {
5241 /* build ops for a bareword */
5242 SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0);
5243 sv_catpvn(sym, "::", 2);
5244 sv_catpv(sym, PL_tokenbuf+1);
5245 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
5246 yylval.opval->op_private = OPpCONST_ENTERED;
5247 gv_fetchpv(SvPVX(sym),
5249 ? (GV_ADDMULTI | GV_ADDINEVAL)
5252 ((PL_tokenbuf[0] == '$') ? SVt_PV
5253 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5258 /* if it's a sort block and they're naming $a or $b */
5259 if (PL_last_lop_op == OP_SORT &&
5260 PL_tokenbuf[0] == '$' &&
5261 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
5264 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
5265 d < PL_bufend && *d != '\n';
5268 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
5269 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
5275 yylval.opval = newOP(OP_PADANY, 0);
5276 yylval.opval->op_targ = tmp;
5282 Whine if they've said @foo in a doublequoted string,
5283 and @foo isn't a variable we can find in the symbol
5286 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
5287 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
5288 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
5289 && ckWARN(WARN_AMBIGUOUS))
5291 /* Downgraded from fatal to warning 20000522 mjd */
5292 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5293 "Possible unintended interpolation of %s in string",
5298 /* build ops for a bareword */
5299 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
5300 yylval.opval->op_private = OPpCONST_ENTERED;
5301 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
5302 ((PL_tokenbuf[0] == '$') ? SVt_PV
5303 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5309 Perl_keyword(pTHX_ register char *d, I32 len)
5314 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
5315 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
5316 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
5317 if (strEQ(d,"__DATA__")) return KEY___DATA__;
5318 if (strEQ(d,"__END__")) return KEY___END__;
5322 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
5327 if (strEQ(d,"and")) return -KEY_and;
5328 if (strEQ(d,"abs")) return -KEY_abs;
5331 if (strEQ(d,"alarm")) return -KEY_alarm;
5332 if (strEQ(d,"atan2")) return -KEY_atan2;
5335 if (strEQ(d,"accept")) return -KEY_accept;
5340 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5343 if (strEQ(d,"bless")) return -KEY_bless;
5344 if (strEQ(d,"bind")) return -KEY_bind;
5345 if (strEQ(d,"binmode")) return -KEY_binmode;
5348 if (strEQ(d,"CORE")) return -KEY_CORE;
5349 if (strEQ(d,"CHECK")) return KEY_CHECK;
5354 if (strEQ(d,"cmp")) return -KEY_cmp;
5355 if (strEQ(d,"chr")) return -KEY_chr;
5356 if (strEQ(d,"cos")) return -KEY_cos;
5359 if (strEQ(d,"chop")) return -KEY_chop;
5362 if (strEQ(d,"close")) return -KEY_close;
5363 if (strEQ(d,"chdir")) return -KEY_chdir;
5364 if (strEQ(d,"chomp")) return -KEY_chomp;
5365 if (strEQ(d,"chmod")) return -KEY_chmod;
5366 if (strEQ(d,"chown")) return -KEY_chown;
5367 if (strEQ(d,"crypt")) return -KEY_crypt;
5370 if (strEQ(d,"chroot")) return -KEY_chroot;
5371 if (strEQ(d,"caller")) return -KEY_caller;
5374 if (strEQ(d,"connect")) return -KEY_connect;
5377 if (strEQ(d,"closedir")) return -KEY_closedir;
5378 if (strEQ(d,"continue")) return -KEY_continue;
5383 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5388 if (strEQ(d,"do")) return KEY_do;
5391 if (strEQ(d,"die")) return -KEY_die;
5394 if (strEQ(d,"dump")) return -KEY_dump;
5397 if (strEQ(d,"delete")) return KEY_delete;
5400 if (strEQ(d,"defined")) return KEY_defined;
5401 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5404 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5409 if (strEQ(d,"END")) return KEY_END;
5414 if (strEQ(d,"eq")) return -KEY_eq;
5417 if (strEQ(d,"eof")) return -KEY_eof;
5418 if (strEQ(d,"exp")) return -KEY_exp;
5421 if (strEQ(d,"else")) return KEY_else;
5422 if (strEQ(d,"exit")) return -KEY_exit;
5423 if (strEQ(d,"eval")) return KEY_eval;
5424 if (strEQ(d,"exec")) return -KEY_exec;
5425 if (strEQ(d,"each")) return -KEY_each;
5428 if (strEQ(d,"elsif")) return KEY_elsif;
5431 if (strEQ(d,"exists")) return KEY_exists;
5432 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5435 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5436 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5439 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5442 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5443 if (strEQ(d,"endservent")) return -KEY_endservent;
5446 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5453 if (strEQ(d,"for")) return KEY_for;
5456 if (strEQ(d,"fork")) return -KEY_fork;
5459 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5460 if (strEQ(d,"flock")) return -KEY_flock;
5463 if (strEQ(d,"format")) return KEY_format;
5464 if (strEQ(d,"fileno")) return -KEY_fileno;
5467 if (strEQ(d,"foreach")) return KEY_foreach;
5470 if (strEQ(d,"formline")) return -KEY_formline;
5475 if (strnEQ(d,"get",3)) {
5480 if (strEQ(d,"ppid")) return -KEY_getppid;
5481 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5484 if (strEQ(d,"pwent")) return -KEY_getpwent;
5485 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5486 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5489 if (strEQ(d,"peername")) return -KEY_getpeername;
5490 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5491 if (strEQ(d,"priority")) return -KEY_getpriority;
5494 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5497 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5501 else if (*d == 'h') {
5502 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5503 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5504 if (strEQ(d,"hostent")) return -KEY_gethostent;
5506 else if (*d == 'n') {
5507 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5508 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5509 if (strEQ(d,"netent")) return -KEY_getnetent;
5511 else if (*d == 's') {
5512 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5513 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5514 if (strEQ(d,"servent")) return -KEY_getservent;
5515 if (strEQ(d,"sockname")) return -KEY_getsockname;
5516 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5518 else if (*d == 'g') {
5519 if (strEQ(d,"grent")) return -KEY_getgrent;
5520 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5521 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5523 else if (*d == 'l') {
5524 if (strEQ(d,"login")) return -KEY_getlogin;
5526 else if (strEQ(d,"c")) return -KEY_getc;
5531 if (strEQ(d,"gt")) return -KEY_gt;
5532 if (strEQ(d,"ge")) return -KEY_ge;
5535 if (strEQ(d,"grep")) return KEY_grep;
5536 if (strEQ(d,"goto")) return KEY_goto;
5537 if (strEQ(d,"glob")) return KEY_glob;
5540 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5545 if (strEQ(d,"hex")) return -KEY_hex;
5548 if (strEQ(d,"INIT")) return KEY_INIT;
5553 if (strEQ(d,"if")) return KEY_if;
5556 if (strEQ(d,"int")) return -KEY_int;
5559 if (strEQ(d,"index")) return -KEY_index;
5560 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5565 if (strEQ(d,"join")) return -KEY_join;
5569 if (strEQ(d,"keys")) return -KEY_keys;
5570 if (strEQ(d,"kill")) return -KEY_kill;
5576 if (strEQ(d,"lt")) return -KEY_lt;
5577 if (strEQ(d,"le")) return -KEY_le;
5578 if (strEQ(d,"lc")) return -KEY_lc;
5581 if (strEQ(d,"log")) return -KEY_log;
5584 if (strEQ(d,"last")) return KEY_last;
5585 if (strEQ(d,"link")) return -KEY_link;
5586 if (strEQ(d,"lock")) return -KEY_lock;
5589 if (strEQ(d,"local")) return KEY_local;
5590 if (strEQ(d,"lstat")) return -KEY_lstat;
5593 if (strEQ(d,"length")) return -KEY_length;
5594 if (strEQ(d,"listen")) return -KEY_listen;
5597 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5600 if (strEQ(d,"localtime")) return -KEY_localtime;
5606 case 1: return KEY_m;
5608 if (strEQ(d,"my")) return KEY_my;
5611 if (strEQ(d,"map")) return KEY_map;
5614 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5617 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5618 if (strEQ(d,"msgget")) return -KEY_msgget;
5619 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5620 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5625 if (strEQ(d,"next")) return KEY_next;
5626 if (strEQ(d,"ne")) return -KEY_ne;
5627 if (strEQ(d,"not")) return -KEY_not;
5628 if (strEQ(d,"no")) return KEY_no;
5633 if (strEQ(d,"or")) return -KEY_or;
5636 if (strEQ(d,"ord")) return -KEY_ord;
5637 if (strEQ(d,"oct")) return -KEY_oct;
5638 if (strEQ(d,"our")) return KEY_our;
5641 if (strEQ(d,"open")) return -KEY_open;
5644 if (strEQ(d,"opendir")) return -KEY_opendir;
5651 if (strEQ(d,"pop")) return -KEY_pop;
5652 if (strEQ(d,"pos")) return KEY_pos;
5655 if (strEQ(d,"push")) return -KEY_push;
5656 if (strEQ(d,"pack")) return -KEY_pack;
5657 if (strEQ(d,"pipe")) return -KEY_pipe;
5660 if (strEQ(d,"print")) return KEY_print;
5663 if (strEQ(d,"printf")) return KEY_printf;
5666 if (strEQ(d,"package")) return KEY_package;
5669 if (strEQ(d,"prototype")) return KEY_prototype;
5674 if (strEQ(d,"q")) return KEY_q;
5675 if (strEQ(d,"qr")) return KEY_qr;
5676 if (strEQ(d,"qq")) return KEY_qq;
5677 if (strEQ(d,"qw")) return KEY_qw;
5678 if (strEQ(d,"qx")) return KEY_qx;
5680 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5685 if (strEQ(d,"ref")) return -KEY_ref;
5688 if (strEQ(d,"read")) return -KEY_read;
5689 if (strEQ(d,"rand")) return -KEY_rand;
5690 if (strEQ(d,"recv")) return -KEY_recv;
5691 if (strEQ(d,"redo")) return KEY_redo;
5694 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5695 if (strEQ(d,"reset")) return -KEY_reset;
5698 if (strEQ(d,"return")) return KEY_return;
5699 if (strEQ(d,"rename")) return -KEY_rename;
5700 if (strEQ(d,"rindex")) return -KEY_rindex;
5703 if (strEQ(d,"require")) return KEY_require;
5704 if (strEQ(d,"reverse")) return -KEY_reverse;
5705 if (strEQ(d,"readdir")) return -KEY_readdir;
5708 if (strEQ(d,"readlink")) return -KEY_readlink;
5709 if (strEQ(d,"readline")) return -KEY_readline;
5710 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5713 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5719 case 0: return KEY_s;
5721 if (strEQ(d,"scalar")) return KEY_scalar;
5726 if (strEQ(d,"seek")) return -KEY_seek;
5727 if (strEQ(d,"send")) return -KEY_send;
5730 if (strEQ(d,"semop")) return -KEY_semop;
5733 if (strEQ(d,"select")) return -KEY_select;
5734 if (strEQ(d,"semctl")) return -KEY_semctl;
5735 if (strEQ(d,"semget")) return -KEY_semget;
5738 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5739 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5742 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5743 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5746 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5749 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5750 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5751 if (strEQ(d,"setservent")) return -KEY_setservent;
5754 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5755 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5762 if (strEQ(d,"shift")) return -KEY_shift;
5765 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5766 if (strEQ(d,"shmget")) return -KEY_shmget;
5769 if (strEQ(d,"shmread")) return -KEY_shmread;
5772 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5773 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5778 if (strEQ(d,"sin")) return -KEY_sin;
5781 if (strEQ(d,"sleep")) return -KEY_sleep;
5784 if (strEQ(d,"sort")) return KEY_sort;
5785 if (strEQ(d,"socket")) return -KEY_socket;
5786 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5789 if (strEQ(d,"split")) return KEY_split;
5790 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5791 if (strEQ(d,"splice")) return -KEY_splice;
5794 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5797 if (strEQ(d,"srand")) return -KEY_srand;
5800 if (strEQ(d,"stat")) return -KEY_stat;
5801 if (strEQ(d,"study")) return KEY_study;
5804 if (strEQ(d,"substr")) return -KEY_substr;
5805 if (strEQ(d,"sub")) return KEY_sub;
5810 if (strEQ(d,"system")) return -KEY_system;
5813 if (strEQ(d,"symlink")) return -KEY_symlink;
5814 if (strEQ(d,"syscall")) return -KEY_syscall;
5815 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5816 if (strEQ(d,"sysread")) return -KEY_sysread;
5817 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5820 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5829 if (strEQ(d,"tr")) return KEY_tr;
5832 if (strEQ(d,"tie")) return KEY_tie;
5835 if (strEQ(d,"tell")) return -KEY_tell;
5836 if (strEQ(d,"tied")) return KEY_tied;
5837 if (strEQ(d,"time")) return -KEY_time;
5840 if (strEQ(d,"times")) return -KEY_times;
5843 if (strEQ(d,"telldir")) return -KEY_telldir;
5846 if (strEQ(d,"truncate")) return -KEY_truncate;
5853 if (strEQ(d,"uc")) return -KEY_uc;
5856 if (strEQ(d,"use")) return KEY_use;
5859 if (strEQ(d,"undef")) return KEY_undef;
5860 if (strEQ(d,"until")) return KEY_until;
5861 if (strEQ(d,"untie")) return KEY_untie;
5862 if (strEQ(d,"utime")) return -KEY_utime;
5863 if (strEQ(d,"umask")) return -KEY_umask;
5866 if (strEQ(d,"unless")) return KEY_unless;
5867 if (strEQ(d,"unpack")) return -KEY_unpack;
5868 if (strEQ(d,"unlink")) return -KEY_unlink;
5871 if (strEQ(d,"unshift")) return -KEY_unshift;
5872 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5877 if (strEQ(d,"values")) return -KEY_values;
5878 if (strEQ(d,"vec")) return -KEY_vec;
5883 if (strEQ(d,"warn")) return -KEY_warn;
5884 if (strEQ(d,"wait")) return -KEY_wait;
5887 if (strEQ(d,"while")) return KEY_while;
5888 if (strEQ(d,"write")) return -KEY_write;
5891 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5894 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5899 if (len == 1) return -KEY_x;
5900 if (strEQ(d,"xor")) return -KEY_xor;
5903 if (len == 1) return KEY_y;
5912 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5916 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5917 if (ckWARN(WARN_SYNTAX)) {
5919 for (w = s+2; *w && level; w++) {
5926 for (; *w && isSPACE(*w); w++) ;
5927 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5928 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5929 "%s (...) interpreted as function",name);
5932 while (s < PL_bufend && isSPACE(*s))
5936 while (s < PL_bufend && isSPACE(*s))
5938 if (isIDFIRST_lazy_if(s,UTF)) {
5940 while (isALNUM_lazy_if(s,UTF))
5942 while (s < PL_bufend && isSPACE(*s))
5947 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5951 Perl_croak(aTHX_ "No comma allowed after %s", what);
5956 /* Either returns sv, or mortalizes sv and returns a new SV*.
5957 Best used as sv=new_constant(..., sv, ...).
5958 If s, pv are NULL, calls subroutine with one argument,
5959 and type is used with error messages only. */
5962 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5966 HV *table = GvHV(PL_hintgv); /* ^H */
5970 const char *why1, *why2, *why3;
5972 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
5975 why2 = strEQ(key,"charnames")
5976 ? "(possibly a missing \"use charnames ...\")"
5978 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
5979 (type ? type: "undef"), why2);
5981 /* This is convoluted and evil ("goto considered harmful")
5982 * but I do not understand the intricacies of all the different
5983 * failure modes of %^H in here. The goal here is to make
5984 * the most probable error message user-friendly. --jhi */
5989 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
5990 (type ? type: "undef"), why1, why2, why3);
5992 yyerror(SvPVX(msg));
5996 cvp = hv_fetch(table, key, strlen(key), FALSE);
5997 if (!cvp || !SvOK(*cvp)) {
6000 why3 = "} is not defined";
6003 sv_2mortal(sv); /* Parent created it permanently */
6006 pv = sv_2mortal(newSVpvn(s, len));
6008 typesv = sv_2mortal(newSVpv(type, 0));
6010 typesv = &PL_sv_undef;
6012 PUSHSTACKi(PERLSI_OVERLOAD);
6024 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
6028 /* Check the eval first */
6029 if (!PL_in_eval && SvTRUE(ERRSV)) {
6031 sv_catpv(ERRSV, "Propagated");
6032 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
6034 res = SvREFCNT_inc(sv);
6038 (void)SvREFCNT_inc(res);
6047 why1 = "Call to &{$^H{";
6049 why3 = "}} did not return a defined value";
6058 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
6060 register char *d = dest;
6061 register char *e = d + destlen - 3; /* two-character token, ending NUL */
6064 Perl_croak(aTHX_ ident_too_long);
6065 if (isALNUM(*s)) /* UTF handled below */
6067 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
6072 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
6076 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
6077 char *t = s + UTF8SKIP(s);
6078 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
6080 if (d + (t - s) > e)
6081 Perl_croak(aTHX_ ident_too_long);
6082 Copy(s, d, t - s, char);
6095 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
6105 e = d + destlen - 3; /* two-character token, ending NUL */
6107 while (isDIGIT(*s)) {
6109 Perl_croak(aTHX_ ident_too_long);
6116 Perl_croak(aTHX_ ident_too_long);
6117 if (isALNUM(*s)) /* UTF handled below */
6119 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
6124 else if (*s == ':' && s[1] == ':') {
6128 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
6129 char *t = s + UTF8SKIP(s);
6130 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
6132 if (d + (t - s) > e)
6133 Perl_croak(aTHX_ ident_too_long);
6134 Copy(s, d, t - s, char);
6145 if (PL_lex_state != LEX_NORMAL)
6146 PL_lex_state = LEX_INTERPENDMAYBE;
6149 if (*s == '$' && s[1] &&
6150 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
6163 if (*d == '^' && *s && isCONTROLVAR(*s)) {
6168 if (isSPACE(s[-1])) {
6171 if (!SPACE_OR_TAB(ch)) {
6177 if (isIDFIRST_lazy_if(d,UTF)) {
6181 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
6183 while (e < send && UTF8_IS_CONTINUED(*e) && is_utf8_mark((U8*)e))
6186 Copy(s, d, e - s, char);
6191 while ((isALNUM(*s) || *s == ':') && d < e)
6194 Perl_croak(aTHX_ ident_too_long);
6197 while (s < send && SPACE_OR_TAB(*s)) s++;
6198 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
6199 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
6200 const char *brack = *s == '[' ? "[...]" : "{...}";
6201 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
6202 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
6203 funny, dest, brack, funny, dest, brack);
6206 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
6210 /* Handle extended ${^Foo} variables
6211 * 1999-02-27 mjd-perl-patch@plover.com */
6212 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
6216 while (isALNUM(*s) && d < e) {
6220 Perl_croak(aTHX_ ident_too_long);
6225 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
6226 PL_lex_state = LEX_INTERPEND;
6229 if (PL_lex_state == LEX_NORMAL) {
6230 if (ckWARN(WARN_AMBIGUOUS) &&
6231 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
6233 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
6234 "Ambiguous use of %c{%s} resolved to %c%s",
6235 funny, dest, funny, dest);
6240 s = bracket; /* let the parser handle it */
6244 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
6245 PL_lex_state = LEX_INTERPEND;
6250 Perl_pmflag(pTHX_ U32* pmfl, int ch)
6255 *pmfl |= PMf_GLOBAL;
6257 *pmfl |= PMf_CONTINUE;
6261 *pmfl |= PMf_MULTILINE;
6263 *pmfl |= PMf_SINGLELINE;
6265 *pmfl |= PMf_EXTENDED;
6269 S_scan_pat(pTHX_ char *start, I32 type)
6274 s = scan_str(start,FALSE,FALSE);
6276 Perl_croak(aTHX_ "Search pattern not terminated");
6278 pm = (PMOP*)newPMOP(type, 0);
6279 if (PL_multi_open == '?')
6280 pm->op_pmflags |= PMf_ONCE;
6282 while (*s && strchr("iomsx", *s))
6283 pmflag(&pm->op_pmflags,*s++);
6286 while (*s && strchr("iogcmsx", *s))
6287 pmflag(&pm->op_pmflags,*s++);
6289 /* issue a warning if /c is specified,but /g is not */
6290 if (ckWARN(WARN_REGEXP) &&
6291 (pm->op_pmflags & PMf_CONTINUE) && !(pm->op_pmflags & PMf_GLOBAL))
6293 Perl_warner(aTHX_ packWARN(WARN_REGEXP), c_without_g);
6296 pm->op_pmpermflags = pm->op_pmflags;
6298 PL_lex_op = (OP*)pm;
6299 yylval.ival = OP_MATCH;
6304 S_scan_subst(pTHX_ char *start)
6311 yylval.ival = OP_NULL;
6313 s = scan_str(start,FALSE,FALSE);
6316 Perl_croak(aTHX_ "Substitution pattern not terminated");
6318 if (s[-1] == PL_multi_open)
6321 first_start = PL_multi_start;
6322 s = scan_str(s,FALSE,FALSE);
6325 SvREFCNT_dec(PL_lex_stuff);
6326 PL_lex_stuff = Nullsv;
6328 Perl_croak(aTHX_ "Substitution replacement not terminated");
6330 PL_multi_start = first_start; /* so whole substitution is taken together */
6332 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6338 else if (strchr("iogcmsx", *s))
6339 pmflag(&pm->op_pmflags,*s++);
6344 /* /c is not meaningful with s/// */
6345 if (ckWARN(WARN_REGEXP) && (pm->op_pmflags & PMf_CONTINUE))
6347 Perl_warner(aTHX_ packWARN(WARN_REGEXP), c_in_subst);
6352 PL_sublex_info.super_bufptr = s;
6353 PL_sublex_info.super_bufend = PL_bufend;
6355 pm->op_pmflags |= PMf_EVAL;
6356 repl = newSVpvn("",0);
6358 sv_catpv(repl, es ? "eval " : "do ");
6359 sv_catpvn(repl, "{ ", 2);
6360 sv_catsv(repl, PL_lex_repl);
6361 sv_catpvn(repl, " };", 2);
6363 SvREFCNT_dec(PL_lex_repl);
6367 pm->op_pmpermflags = pm->op_pmflags;
6368 PL_lex_op = (OP*)pm;
6369 yylval.ival = OP_SUBST;
6374 S_scan_trans(pTHX_ char *start)
6383 yylval.ival = OP_NULL;
6385 s = scan_str(start,FALSE,FALSE);
6387 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6388 if (s[-1] == PL_multi_open)
6391 s = scan_str(s,FALSE,FALSE);
6394 SvREFCNT_dec(PL_lex_stuff);
6395 PL_lex_stuff = Nullsv;
6397 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6400 complement = del = squash = 0;
6401 while (strchr("cds", *s)) {
6403 complement = OPpTRANS_COMPLEMENT;
6405 del = OPpTRANS_DELETE;
6407 squash = OPpTRANS_SQUASH;
6411 New(803, tbl, complement&&!del?258:256, short);
6412 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6413 o->op_private = del|squash|complement|
6414 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
6415 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
6418 yylval.ival = OP_TRANS;
6423 S_scan_heredoc(pTHX_ register char *s)
6426 I32 op_type = OP_SCALAR;
6433 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6437 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6440 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
6441 if (*peek && strchr("`'\"",*peek)) {
6444 s = delimcpy(d, e, s, PL_bufend, term, &len);
6454 if (!isALNUM_lazy_if(s,UTF))
6455 deprecate_old("bare << to mean <<\"\"");
6456 for (; isALNUM_lazy_if(s,UTF); s++) {
6461 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6462 Perl_croak(aTHX_ "Delimiter for here document is too long");
6465 len = d - PL_tokenbuf;
6466 #ifndef PERL_STRICT_CR
6467 d = strchr(s, '\r');
6471 while (s < PL_bufend) {
6477 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6486 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6491 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6492 herewas = newSVpvn(s,PL_bufend-s);
6494 s--, herewas = newSVpvn(s,d-s);
6495 s += SvCUR(herewas);
6497 tmpstr = NEWSV(87,79);
6498 sv_upgrade(tmpstr, SVt_PVIV);
6503 else if (term == '`') {
6504 op_type = OP_BACKTICK;
6505 SvIVX(tmpstr) = '\\';
6509 PL_multi_start = CopLINE(PL_curcop);
6510 PL_multi_open = PL_multi_close = '<';
6511 term = *PL_tokenbuf;
6512 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6513 char *bufptr = PL_sublex_info.super_bufptr;
6514 char *bufend = PL_sublex_info.super_bufend;
6515 char *olds = s - SvCUR(herewas);
6516 s = strchr(bufptr, '\n');
6520 while (s < bufend &&
6521 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6523 CopLINE_inc(PL_curcop);
6526 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
6527 missingterm(PL_tokenbuf);
6529 sv_setpvn(herewas,bufptr,d-bufptr+1);
6530 sv_setpvn(tmpstr,d+1,s-d);
6532 sv_catpvn(herewas,s,bufend-s);
6533 (void)strcpy(bufptr,SvPVX(herewas));
6540 while (s < PL_bufend &&
6541 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6543 CopLINE_inc(PL_curcop);
6545 if (s >= PL_bufend) {
6546 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
6547 missingterm(PL_tokenbuf);
6549 sv_setpvn(tmpstr,d+1,s-d);
6551 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6553 sv_catpvn(herewas,s,PL_bufend-s);
6554 sv_setsv(PL_linestr,herewas);
6555 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6556 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6557 PL_last_lop = PL_last_uni = Nullch;
6560 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6561 while (s >= PL_bufend) { /* multiple line string? */
6563 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6564 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
6565 missingterm(PL_tokenbuf);
6567 CopLINE_inc(PL_curcop);
6568 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6569 PL_last_lop = PL_last_uni = Nullch;
6570 #ifndef PERL_STRICT_CR
6571 if (PL_bufend - PL_linestart >= 2) {
6572 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6573 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6575 PL_bufend[-2] = '\n';
6577 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6579 else if (PL_bufend[-1] == '\r')
6580 PL_bufend[-1] = '\n';
6582 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6583 PL_bufend[-1] = '\n';
6585 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6586 SV *sv = NEWSV(88,0);
6588 sv_upgrade(sv, SVt_PVMG);
6589 sv_setsv(sv,PL_linestr);
6592 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6594 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6597 sv_catsv(PL_linestr,herewas);
6598 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6602 sv_catsv(tmpstr,PL_linestr);
6607 PL_multi_end = CopLINE(PL_curcop);
6608 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6609 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6610 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6612 SvREFCNT_dec(herewas);
6613 if (UTF && !IN_BYTES && is_utf8_string((U8*)SvPVX(tmpstr), SvCUR(tmpstr)))
6615 PL_lex_stuff = tmpstr;
6616 yylval.ival = op_type;
6621 takes: current position in input buffer
6622 returns: new position in input buffer
6623 side-effects: yylval and lex_op are set.
6628 <FH> read from filehandle
6629 <pkg::FH> read from package qualified filehandle
6630 <pkg'FH> read from package qualified filehandle
6631 <$fh> read from filehandle in $fh
6637 S_scan_inputsymbol(pTHX_ char *start)
6639 register char *s = start; /* current position in buffer */
6645 d = PL_tokenbuf; /* start of temp holding space */
6646 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6647 end = strchr(s, '\n');
6650 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6652 /* die if we didn't have space for the contents of the <>,
6653 or if it didn't end, or if we see a newline
6656 if (len >= sizeof PL_tokenbuf)
6657 Perl_croak(aTHX_ "Excessively long <> operator");
6659 Perl_croak(aTHX_ "Unterminated <> operator");
6664 Remember, only scalar variables are interpreted as filehandles by
6665 this code. Anything more complex (e.g., <$fh{$num}>) will be
6666 treated as a glob() call.
6667 This code makes use of the fact that except for the $ at the front,
6668 a scalar variable and a filehandle look the same.
6670 if (*d == '$' && d[1]) d++;
6672 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6673 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6676 /* If we've tried to read what we allow filehandles to look like, and
6677 there's still text left, then it must be a glob() and not a getline.
6678 Use scan_str to pull out the stuff between the <> and treat it
6679 as nothing more than a string.
6682 if (d - PL_tokenbuf != len) {
6683 yylval.ival = OP_GLOB;
6685 s = scan_str(start,FALSE,FALSE);
6687 Perl_croak(aTHX_ "Glob not terminated");
6691 bool readline_overriden = FALSE;
6692 GV *gv_readline = Nullgv;
6694 /* we're in a filehandle read situation */
6697 /* turn <> into <ARGV> */
6699 (void)strcpy(d,"ARGV");
6701 /* Check whether readline() is overriden */
6702 if (((gv_readline = gv_fetchpv("readline", FALSE, SVt_PVCV))
6703 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline))
6705 ((gvp = (GV**)hv_fetch(PL_globalstash, "readline", 8, FALSE))
6706 && (gv_readline = *gvp) != (GV*)&PL_sv_undef
6707 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline)))
6708 readline_overriden = TRUE;
6710 /* if <$fh>, create the ops to turn the variable into a
6716 /* try to find it in the pad for this block, otherwise find
6717 add symbol table ops
6719 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6720 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
6721 if (SvFLAGS(namesv) & SVpad_OUR) {
6722 SV *sym = sv_2mortal(newSVpv(HvNAME(GvSTASH(namesv)),0));
6723 sv_catpvn(sym, "::", 2);
6729 OP *o = newOP(OP_PADSV, 0);
6731 PL_lex_op = readline_overriden
6732 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
6733 append_elem(OP_LIST, o,
6734 newCVREF(0, newGVOP(OP_GV,0,gv_readline))))
6735 : (OP*)newUNOP(OP_READLINE, 0, o);
6744 ? (GV_ADDMULTI | GV_ADDINEVAL)
6747 PL_lex_op = readline_overriden
6748 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
6749 append_elem(OP_LIST,
6750 newUNOP(OP_RV2SV, 0, newGVOP(OP_GV, 0, gv)),
6751 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
6752 : (OP*)newUNOP(OP_READLINE, 0,
6753 newUNOP(OP_RV2SV, 0,
6754 newGVOP(OP_GV, 0, gv)));
6756 if (!readline_overriden)
6757 PL_lex_op->op_flags |= OPf_SPECIAL;
6758 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6759 yylval.ival = OP_NULL;
6762 /* If it's none of the above, it must be a literal filehandle
6763 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6765 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6766 PL_lex_op = readline_overriden
6767 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
6768 append_elem(OP_LIST,
6769 newGVOP(OP_GV, 0, gv),
6770 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
6771 : (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6772 yylval.ival = OP_NULL;
6781 takes: start position in buffer
6782 keep_quoted preserve \ on the embedded delimiter(s)
6783 keep_delims preserve the delimiters around the string
6784 returns: position to continue reading from buffer
6785 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6786 updates the read buffer.
6788 This subroutine pulls a string out of the input. It is called for:
6789 q single quotes q(literal text)
6790 ' single quotes 'literal text'
6791 qq double quotes qq(interpolate $here please)
6792 " double quotes "interpolate $here please"
6793 qx backticks qx(/bin/ls -l)
6794 ` backticks `/bin/ls -l`
6795 qw quote words @EXPORT_OK = qw( func() $spam )
6796 m// regexp match m/this/
6797 s/// regexp substitute s/this/that/
6798 tr/// string transliterate tr/this/that/
6799 y/// string transliterate y/this/that/
6800 ($*@) sub prototypes sub foo ($)
6801 (stuff) sub attr parameters sub foo : attr(stuff)
6802 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6804 In most of these cases (all but <>, patterns and transliterate)
6805 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6806 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6807 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6810 It skips whitespace before the string starts, and treats the first
6811 character as the delimiter. If the delimiter is one of ([{< then
6812 the corresponding "close" character )]}> is used as the closing
6813 delimiter. It allows quoting of delimiters, and if the string has
6814 balanced delimiters ([{<>}]) it allows nesting.
6816 On success, the SV with the resulting string is put into lex_stuff or,
6817 if that is already non-NULL, into lex_repl. The second case occurs only
6818 when parsing the RHS of the special constructs s/// and tr/// (y///).
6819 For convenience, the terminating delimiter character is stuffed into
6824 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6826 SV *sv; /* scalar value: string */
6827 char *tmps; /* temp string, used for delimiter matching */
6828 register char *s = start; /* current position in the buffer */
6829 register char term; /* terminating character */
6830 register char *to; /* current position in the sv's data */
6831 I32 brackets = 1; /* bracket nesting level */
6832 bool has_utf8 = FALSE; /* is there any utf8 content? */
6834 /* skip space before the delimiter */
6838 /* mark where we are, in case we need to report errors */
6841 /* after skipping whitespace, the next character is the terminator */
6843 if (!UTF8_IS_INVARIANT((U8)term) && UTF)
6846 /* mark where we are */
6847 PL_multi_start = CopLINE(PL_curcop);
6848 PL_multi_open = term;
6850 /* find corresponding closing delimiter */
6851 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6853 PL_multi_close = term;
6855 /* create a new SV to hold the contents. 87 is leak category, I'm
6856 assuming. 79 is the SV's initial length. What a random number. */
6858 sv_upgrade(sv, SVt_PVIV);
6860 (void)SvPOK_only(sv); /* validate pointer */
6862 /* move past delimiter and try to read a complete string */
6864 sv_catpvn(sv, s, 1);
6867 /* extend sv if need be */
6868 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6869 /* set 'to' to the next character in the sv's string */
6870 to = SvPVX(sv)+SvCUR(sv);
6872 /* if open delimiter is the close delimiter read unbridle */
6873 if (PL_multi_open == PL_multi_close) {
6874 for (; s < PL_bufend; s++,to++) {
6875 /* embedded newlines increment the current line number */
6876 if (*s == '\n' && !PL_rsfp)
6877 CopLINE_inc(PL_curcop);
6878 /* handle quoted delimiters */
6879 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6880 if (!keep_quoted && s[1] == term)
6882 /* any other quotes are simply copied straight through */
6886 /* terminate when run out of buffer (the for() condition), or
6887 have found the terminator */
6888 else if (*s == term)
6890 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
6896 /* if the terminator isn't the same as the start character (e.g.,
6897 matched brackets), we have to allow more in the quoting, and
6898 be prepared for nested brackets.
6901 /* read until we run out of string, or we find the terminator */
6902 for (; s < PL_bufend; s++,to++) {
6903 /* embedded newlines increment the line count */
6904 if (*s == '\n' && !PL_rsfp)
6905 CopLINE_inc(PL_curcop);
6906 /* backslashes can escape the open or closing characters */
6907 if (*s == '\\' && s+1 < PL_bufend) {
6909 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6914 /* allow nested opens and closes */
6915 else if (*s == PL_multi_close && --brackets <= 0)
6917 else if (*s == PL_multi_open)
6919 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
6924 /* terminate the copied string and update the sv's end-of-string */
6926 SvCUR_set(sv, to - SvPVX(sv));
6929 * this next chunk reads more into the buffer if we're not done yet
6933 break; /* handle case where we are done yet :-) */
6935 #ifndef PERL_STRICT_CR
6936 if (to - SvPVX(sv) >= 2) {
6937 if ((to[-2] == '\r' && to[-1] == '\n') ||
6938 (to[-2] == '\n' && to[-1] == '\r'))
6942 SvCUR_set(sv, to - SvPVX(sv));
6944 else if (to[-1] == '\r')
6947 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6951 /* if we're out of file, or a read fails, bail and reset the current
6952 line marker so we can report where the unterminated string began
6955 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6957 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
6960 /* we read a line, so increment our line counter */
6961 CopLINE_inc(PL_curcop);
6963 /* update debugger info */
6964 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6965 SV *sv = NEWSV(88,0);
6967 sv_upgrade(sv, SVt_PVMG);
6968 sv_setsv(sv,PL_linestr);
6971 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6974 /* having changed the buffer, we must update PL_bufend */
6975 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6976 PL_last_lop = PL_last_uni = Nullch;
6979 /* at this point, we have successfully read the delimited string */
6982 sv_catpvn(sv, s, 1);
6985 PL_multi_end = CopLINE(PL_curcop);
6988 /* if we allocated too much space, give some back */
6989 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6990 SvLEN_set(sv, SvCUR(sv) + 1);
6991 Renew(SvPVX(sv), SvLEN(sv), char);
6994 /* decide whether this is the first or second quoted string we've read
7007 takes: pointer to position in buffer
7008 returns: pointer to new position in buffer
7009 side-effects: builds ops for the constant in yylval.op
7011 Read a number in any of the formats that Perl accepts:
7013 \d(_?\d)*(\.(\d(_?\d)*)?)?[Ee][\+\-]?(\d(_?\d)*) 12 12.34 12.
7014 \.\d(_?\d)*[Ee][\+\-]?(\d(_?\d)*) .34
7017 0x[0-9A-Fa-f](_?[0-9A-Fa-f])*
7019 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
7022 If it reads a number without a decimal point or an exponent, it will
7023 try converting the number to an integer and see if it can do so
7024 without loss of precision.
7028 Perl_scan_num(pTHX_ char *start, YYSTYPE* lvalp)
7030 register char *s = start; /* current position in buffer */
7031 register char *d; /* destination in temp buffer */
7032 register char *e; /* end of temp buffer */
7033 NV nv; /* number read, as a double */
7034 SV *sv = Nullsv; /* place to put the converted number */
7035 bool floatit; /* boolean: int or float? */
7036 char *lastub = 0; /* position of last underbar */
7037 static char number_too_long[] = "Number too long";
7039 /* We use the first character to decide what type of number this is */
7043 Perl_croak(aTHX_ "panic: scan_num");
7045 /* if it starts with a 0, it could be an octal number, a decimal in
7046 0.13 disguise, or a hexadecimal number, or a binary number. */
7050 u holds the "number so far"
7051 shift the power of 2 of the base
7052 (hex == 4, octal == 3, binary == 1)
7053 overflowed was the number more than we can hold?
7055 Shift is used when we add a digit. It also serves as an "are
7056 we in octal/hex/binary?" indicator to disallow hex characters
7062 bool overflowed = FALSE;
7063 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
7064 static char* bases[5] = { "", "binary", "", "octal",
7066 static char* Bases[5] = { "", "Binary", "", "Octal",
7068 static char *maxima[5] = { "",
7069 "0b11111111111111111111111111111111",
7073 char *base, *Base, *max;
7079 } else if (s[1] == 'b') {
7083 /* check for a decimal in disguise */
7084 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
7086 /* so it must be octal */
7093 if (ckWARN(WARN_SYNTAX))
7094 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7095 "Misplaced _ in number");
7099 base = bases[shift];
7100 Base = Bases[shift];
7101 max = maxima[shift];
7103 /* read the rest of the number */
7105 /* x is used in the overflow test,
7106 b is the digit we're adding on. */
7111 /* if we don't mention it, we're done */
7115 /* _ are ignored -- but warned about if consecutive */
7117 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7118 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7119 "Misplaced _ in number");
7123 /* 8 and 9 are not octal */
7126 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
7130 case '2': case '3': case '4':
7131 case '5': case '6': case '7':
7133 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
7137 b = *s++ & 15; /* ASCII digit -> value of digit */
7141 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
7142 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
7143 /* make sure they said 0x */
7148 /* Prepare to put the digit we have onto the end
7149 of the number so far. We check for overflows.
7154 x = u << shift; /* make room for the digit */
7156 if ((x >> shift) != u
7157 && !(PL_hints & HINT_NEW_BINARY)) {
7160 if (ckWARN_d(WARN_OVERFLOW))
7161 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
7162 "Integer overflow in %s number",
7165 u = x | b; /* add the digit to the end */
7168 n *= nvshift[shift];
7169 /* If an NV has not enough bits in its
7170 * mantissa to represent an UV this summing of
7171 * small low-order numbers is a waste of time
7172 * (because the NV cannot preserve the
7173 * low-order bits anyway): we could just
7174 * remember when did we overflow and in the
7175 * end just multiply n by the right
7183 /* if we get here, we had success: make a scalar value from
7188 /* final misplaced underbar check */
7190 if (ckWARN(WARN_SYNTAX))
7191 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
7196 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
7197 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
7198 "%s number > %s non-portable",
7204 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
7205 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
7206 "%s number > %s non-portable",
7211 if (PL_hints & HINT_NEW_BINARY)
7212 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
7217 handle decimal numbers.
7218 we're also sent here when we read a 0 as the first digit
7220 case '1': case '2': case '3': case '4': case '5':
7221 case '6': case '7': case '8': case '9': case '.':
7224 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
7227 /* read next group of digits and _ and copy into d */
7228 while (isDIGIT(*s) || *s == '_') {
7229 /* skip underscores, checking for misplaced ones
7233 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7234 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7235 "Misplaced _ in number");
7239 /* check for end of fixed-length buffer */
7241 Perl_croak(aTHX_ number_too_long);
7242 /* if we're ok, copy the character */
7247 /* final misplaced underbar check */
7248 if (lastub && s == lastub + 1) {
7249 if (ckWARN(WARN_SYNTAX))
7250 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
7253 /* read a decimal portion if there is one. avoid
7254 3..5 being interpreted as the number 3. followed
7257 if (*s == '.' && s[1] != '.') {
7262 if (ckWARN(WARN_SYNTAX))
7263 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7264 "Misplaced _ in number");
7268 /* copy, ignoring underbars, until we run out of digits.
7270 for (; isDIGIT(*s) || *s == '_'; s++) {
7271 /* fixed length buffer check */
7273 Perl_croak(aTHX_ number_too_long);
7275 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7276 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7277 "Misplaced _ in number");
7283 /* fractional part ending in underbar? */
7285 if (ckWARN(WARN_SYNTAX))
7286 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7287 "Misplaced _ in number");
7289 if (*s == '.' && isDIGIT(s[1])) {
7290 /* oops, it's really a v-string, but without the "v" */
7296 /* read exponent part, if present */
7297 if (*s && strchr("eE",*s) && strchr("+-0123456789_", s[1])) {
7301 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
7302 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
7304 /* stray preinitial _ */
7306 if (ckWARN(WARN_SYNTAX))
7307 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7308 "Misplaced _ in number");
7312 /* allow positive or negative exponent */
7313 if (*s == '+' || *s == '-')
7316 /* stray initial _ */
7318 if (ckWARN(WARN_SYNTAX))
7319 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7320 "Misplaced _ in number");
7324 /* read digits of exponent */
7325 while (isDIGIT(*s) || *s == '_') {
7328 Perl_croak(aTHX_ number_too_long);
7332 if (ckWARN(WARN_SYNTAX) &&
7333 ((lastub && s == lastub + 1) ||
7334 (!isDIGIT(s[1]) && s[1] != '_')))
7335 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7336 "Misplaced _ in number");
7343 /* make an sv from the string */
7347 We try to do an integer conversion first if no characters
7348 indicating "float" have been found.
7353 int flags = grok_number (PL_tokenbuf, d - PL_tokenbuf, &uv);
7355 if (flags == IS_NUMBER_IN_UV) {
7357 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7360 } else if (flags == (IS_NUMBER_IN_UV | IS_NUMBER_NEG)) {
7361 if (uv <= (UV) IV_MIN)
7362 sv_setiv(sv, -(IV)uv);
7369 /* terminate the string */
7371 nv = Atof(PL_tokenbuf);
7375 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7376 (PL_hints & HINT_NEW_INTEGER) )
7377 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7378 (floatit ? "float" : "integer"),
7382 /* if it starts with a v, it could be a v-string */
7385 sv = NEWSV(92,5); /* preallocate storage space */
7386 s = new_vstring(s,sv);
7390 /* make the op for the constant and return */
7393 lvalp->opval = newSVOP(OP_CONST, 0, sv);
7395 lvalp->opval = Nullop;
7401 S_scan_formline(pTHX_ register char *s)
7405 SV *stuff = newSVpvn("",0);
7406 bool needargs = FALSE;
7409 if (*s == '.' || *s == /*{*/'}') {
7411 #ifdef PERL_STRICT_CR
7412 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
7414 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
7416 if (*t == '\n' || t == PL_bufend)
7419 if (PL_in_eval && !PL_rsfp) {
7420 eol = strchr(s,'\n');
7425 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7427 for (t = s; t < eol; t++) {
7428 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7430 goto enough; /* ~~ must be first line in formline */
7432 if (*t == '@' || *t == '^')
7436 sv_catpvn(stuff, s, eol-s);
7437 #ifndef PERL_STRICT_CR
7438 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7439 char *end = SvPVX(stuff) + SvCUR(stuff);
7451 s = filter_gets(PL_linestr, PL_rsfp, 0);
7452 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7453 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7454 PL_last_lop = PL_last_uni = Nullch;
7457 yyerror("Format not terminated");
7467 PL_lex_state = LEX_NORMAL;
7468 PL_nextval[PL_nexttoke].ival = 0;
7472 PL_lex_state = LEX_FORMLINE;
7473 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7475 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7479 SvREFCNT_dec(stuff);
7480 PL_lex_formbrack = 0;
7491 PL_cshlen = strlen(PL_cshname);
7496 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7498 I32 oldsavestack_ix = PL_savestack_ix;
7499 CV* outsidecv = PL_compcv;
7503 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7505 SAVEI32(PL_subline);
7506 save_item(PL_subname);
7509 SAVESPTR(PL_comppad_name);
7510 SAVESPTR(PL_compcv);
7511 SAVEI32(PL_comppad_name_fill);
7512 SAVEI32(PL_min_intro_pending);
7513 SAVEI32(PL_max_intro_pending);
7514 SAVEI32(PL_pad_reset_pending);
7516 PL_compcv = (CV*)NEWSV(1104,0);
7517 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7518 CvFLAGS(PL_compcv) |= flags;
7520 PL_comppad = newAV();
7521 av_push(PL_comppad, Nullsv);
7522 PL_curpad = AvARRAY(PL_comppad);
7523 PL_comppad_name = newAV();
7524 PL_comppad_name_fill = 0;
7525 PL_min_intro_pending = 0;
7527 PL_subline = CopLINE(PL_curcop);
7528 #ifdef USE_5005THREADS
7529 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
7530 PL_curpad[0] = (SV*)newAV();
7531 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
7532 #endif /* USE_5005THREADS */
7534 comppadlist = newAV();
7535 AvREAL_off(comppadlist);
7536 av_store(comppadlist, 0, (SV*)PL_comppad_name);
7537 av_store(comppadlist, 1, (SV*)PL_comppad);
7539 CvPADLIST(PL_compcv) = comppadlist;
7540 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7541 #ifdef USE_5005THREADS
7542 CvOWNER(PL_compcv) = 0;
7543 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7544 MUTEX_INIT(CvMUTEXP(PL_compcv));
7545 #endif /* USE_5005THREADS */
7547 return oldsavestack_ix;
7551 #pragma segment Perl_yylex
7554 Perl_yywarn(pTHX_ char *s)
7556 PL_in_eval |= EVAL_WARNONLY;
7558 PL_in_eval &= ~EVAL_WARNONLY;
7563 Perl_yyerror(pTHX_ char *s)
7566 char *context = NULL;
7570 if (!yychar || (yychar == ';' && !PL_rsfp))
7572 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7573 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7574 while (isSPACE(*PL_oldoldbufptr))
7576 context = PL_oldoldbufptr;
7577 contlen = PL_bufptr - PL_oldoldbufptr;
7579 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7580 PL_oldbufptr != PL_bufptr) {
7581 while (isSPACE(*PL_oldbufptr))
7583 context = PL_oldbufptr;
7584 contlen = PL_bufptr - PL_oldbufptr;
7586 else if (yychar > 255)
7587 where = "next token ???";
7588 #ifdef USE_PURE_BISON
7589 /* GNU Bison sets the value -2 */
7590 else if (yychar == -2) {
7592 else if ((yychar & 127) == 127) {
7594 if (PL_lex_state == LEX_NORMAL ||
7595 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7596 where = "at end of line";
7597 else if (PL_lex_inpat)
7598 where = "within pattern";
7600 where = "within string";
7603 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7605 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7606 else if (isPRINT_LC(yychar))
7607 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7609 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7610 where = SvPVX(where_sv);
7612 msg = sv_2mortal(newSVpv(s, 0));
7613 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7614 OutCopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7616 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7618 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7619 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7620 Perl_sv_catpvf(aTHX_ msg,
7621 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7622 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7625 if (PL_in_eval & EVAL_WARNONLY)
7626 Perl_warn(aTHX_ "%"SVf, msg);
7629 if (PL_error_count >= 10) {
7630 if (PL_in_eval && SvCUR(ERRSV))
7631 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
7632 ERRSV, OutCopFILE(PL_curcop));
7634 Perl_croak(aTHX_ "%s has too many errors.\n",
7635 OutCopFILE(PL_curcop));
7638 PL_in_my_stash = Nullhv;
7642 #pragma segment Main
7646 S_swallow_bom(pTHX_ U8 *s)
7649 slen = SvCUR(PL_linestr);
7653 /* UTF-16 little-endian */
7654 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
7655 Perl_croak(aTHX_ "Unsupported script encoding");
7656 #ifndef PERL_NO_UTF16_FILTER
7657 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-LE script encoding\n"));
7659 if (PL_bufend > (char*)s) {
7663 filter_add(utf16rev_textfilter, NULL);
7664 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7665 PL_bufend = (char*)utf16_to_utf8_reversed(s, news,
7666 PL_bufend - (char*)s - 1,
7668 Copy(news, s, newlen, U8);
7669 SvCUR_set(PL_linestr, newlen);
7670 PL_bufend = SvPVX(PL_linestr) + newlen;
7671 news[newlen++] = '\0';
7675 Perl_croak(aTHX_ "Unsupported script encoding");
7680 if (s[1] == 0xFF) { /* UTF-16 big-endian */
7681 #ifndef PERL_NO_UTF16_FILTER
7682 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding\n"));
7684 if (PL_bufend > (char *)s) {
7688 filter_add(utf16_textfilter, NULL);
7689 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7690 PL_bufend = (char*)utf16_to_utf8(s, news,
7691 PL_bufend - (char*)s,
7693 Copy(news, s, newlen, U8);
7694 SvCUR_set(PL_linestr, newlen);
7695 PL_bufend = SvPVX(PL_linestr) + newlen;
7696 news[newlen++] = '\0';
7700 Perl_croak(aTHX_ "Unsupported script encoding");
7705 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
7706 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-8 script encoding\n"));
7711 if (slen > 3 && s[1] == 0 && /* UTF-32 big-endian */
7712 s[2] == 0xFE && s[3] == 0xFF)
7714 Perl_croak(aTHX_ "Unsupported script encoding");
7722 * Restore a source filter.
7726 restore_rsfp(pTHX_ void *f)
7728 PerlIO *fp = (PerlIO*)f;
7730 if (PL_rsfp == PerlIO_stdin())
7731 PerlIO_clearerr(PL_rsfp);
7732 else if (PL_rsfp && (PL_rsfp != fp))
7733 PerlIO_close(PL_rsfp);
7737 #ifndef PERL_NO_UTF16_FILTER
7739 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
7741 I32 count = FILTER_READ(idx+1, sv, maxlen);
7746 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7747 if (!*SvPV_nolen(sv))
7748 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7751 tend = utf16_to_utf8((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7752 sv_usepvn(sv, (char*)tmps, tend - tmps);
7758 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
7760 I32 count = FILTER_READ(idx+1, sv, maxlen);
7765 if (!*SvPV_nolen(sv))
7766 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7769 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7770 tend = utf16_to_utf8_reversed((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7771 sv_usepvn(sv, (char*)tmps, tend - tmps);