3 * Copyright (c) 1991-1999, 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";
30 static void restore_rsfp(pTHXo_ void *f);
31 static void restore_expect(pTHXo_ void *e);
32 static void restore_lex_expect(pTHXo_ void *e);
34 #define UTF (PL_hints & HINT_UTF8)
36 * Note: we try to be careful never to call the isXXX_utf8() functions
37 * unless we're pretty sure we've seen the beginning of a UTF-8 character
38 * (that is, the two high bits are set). Otherwise we risk loading in the
39 * heavy-duty SWASHINIT and SWASHGET routines unnecessarily.
41 #define isIDFIRST_lazy(p) ((!UTF || (*((U8*)p) < 0xc0)) \
43 : isIDFIRST_utf8((U8*)p))
44 #define isALNUM_lazy(p) ((!UTF || (*((U8*)p) < 0xc0)) \
46 : isALNUM_utf8((U8*)p))
48 /* In variables name $^X, these are the legal values for X.
49 * 1999-02-27 mjd-perl-patch@plover.com */
50 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
52 /* LEX_* are values for PL_lex_state, the state of the lexer.
53 * They are arranged oddly so that the guard on the switch statement
54 * can get by with a single comparison (if the compiler is smart enough).
57 /* #define LEX_NOTPARSING 11 is done in perl.h. */
60 #define LEX_INTERPNORMAL 9
61 #define LEX_INTERPCASEMOD 8
62 #define LEX_INTERPPUSH 7
63 #define LEX_INTERPSTART 6
64 #define LEX_INTERPEND 5
65 #define LEX_INTERPENDMAYBE 4
66 #define LEX_INTERPCONCAT 3
67 #define LEX_INTERPCONST 2
68 #define LEX_FORMLINE 1
69 #define LEX_KNOWNEXT 0
78 /* XXX If this causes problems, set i_unistd=undef in the hint file. */
80 # include <unistd.h> /* Needed for execv() */
89 YYSTYPE* yylval_pointer = NULL;
90 int* yychar_pointer = NULL;
93 # define yylval (*yylval_pointer)
94 # define yychar (*yychar_pointer)
95 # define PERL_YYLEX_PARAM yylval_pointer,yychar_pointer
97 # define yylex() Perl_yylex(aTHX_ yylval_pointer, yychar_pointer)
100 #include "keywords.h"
102 /* CLINE is a macro that ensures PL_copline has a sane value */
107 #define CLINE (PL_copline = (PL_curcop->cop_line < PL_copline ? PL_curcop->cop_line : PL_copline))
110 * Convenience functions to return different tokens and prime the
111 * lexer for the next token. They all take an argument.
113 * TOKEN : generic token (used for '(', DOLSHARP, etc)
114 * OPERATOR : generic operator
115 * AOPERATOR : assignment operator
116 * PREBLOCK : beginning the block after an if, while, foreach, ...
117 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
118 * PREREF : *EXPR where EXPR is not a simple identifier
119 * TERM : expression term
120 * LOOPX : loop exiting command (goto, last, dump, etc)
121 * FTST : file test operator
122 * FUN0 : zero-argument function
124 * BOop : bitwise or or xor
126 * SHop : shift operator
127 * PWop : power operator
128 * PMop : pattern-matching operator
129 * Aop : addition-level operator
130 * Mop : multiplication-level operator
131 * Eop : equality-testing operator
132 * Rop : relational operator <= != gt
134 * Also see LOP and lop() below.
137 #define TOKEN(retval) return (PL_bufptr = s,(int)retval)
138 #define OPERATOR(retval) return (PL_expect = XTERM,PL_bufptr = s,(int)retval)
139 #define AOPERATOR(retval) return ao((PL_expect = XTERM,PL_bufptr = s,(int)retval))
140 #define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s,(int)retval)
141 #define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s,(int)retval)
142 #define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s,(int)retval)
143 #define TERM(retval) return (CLINE, PL_expect = XOPERATOR,PL_bufptr = s,(int)retval)
144 #define LOOPX(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LOOPEX)
145 #define FTST(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)UNIOP)
146 #define FUN0(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC0)
147 #define FUN1(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC1)
148 #define BOop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITOROP))
149 #define BAop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITANDOP))
150 #define SHop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)SHIFTOP))
151 #define PWop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)POWOP))
152 #define PMop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MATCHOP)
153 #define Aop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)ADDOP))
154 #define Mop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MULOP))
155 #define Eop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)EQOP)
156 #define Rop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)RELOP)
158 /* This bit of chicanery makes a unary function followed by
159 * a parenthesis into a function with one argument, highest precedence.
161 #define UNI(f) return(yylval.ival = f, \
164 PL_last_uni = PL_oldbufptr, \
165 PL_last_lop_op = f, \
166 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
168 #define UNIBRACK(f) return(yylval.ival = f, \
170 PL_last_uni = PL_oldbufptr, \
171 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
173 /* grandfather return to old style */
174 #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
179 * This subroutine detects &&= and ||= and turns an ANDAND or OROR
180 * into an OP_ANDASSIGN or OP_ORASSIGN
184 S_ao(pTHX_ int toketype)
186 if (*PL_bufptr == '=') {
188 if (toketype == ANDAND)
189 yylval.ival = OP_ANDASSIGN;
190 else if (toketype == OROR)
191 yylval.ival = OP_ORASSIGN;
199 * When Perl expects an operator and finds something else, no_op
200 * prints the warning. It always prints "<something> found where
201 * operator expected. It prints "Missing semicolon on previous line?"
202 * if the surprise occurs at the start of the line. "do you need to
203 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
204 * where the compiler doesn't know if foo is a method call or a function.
205 * It prints "Missing operator before end of line" if there's nothing
206 * after the missing operator, or "... before <...>" if there is something
207 * after the missing operator.
211 S_no_op(pTHX_ char *what, char *s)
213 char *oldbp = PL_bufptr;
214 bool is_first = (PL_oldbufptr == PL_linestart);
222 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
224 Perl_warn(aTHX_ "\t(Missing semicolon on previous line?)\n");
225 else if (PL_oldoldbufptr && isIDFIRST_lazy(PL_oldoldbufptr)) {
227 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy(t) || *t == ':'); t++) ;
228 if (t < PL_bufptr && isSPACE(*t))
229 Perl_warn(aTHX_ "\t(Do you need to predeclare %.*s?)\n",
230 t - PL_oldoldbufptr, PL_oldoldbufptr);
233 Perl_warn(aTHX_ "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp);
239 * Complain about missing quote/regexp/heredoc terminator.
240 * If it's called with (char *)NULL then it cauterizes the line buffer.
241 * If we're in a delimited string and the delimiter is a control
242 * character, it's reformatted into a two-char sequence like ^C.
247 S_missingterm(pTHX_ char *s)
252 char *nl = strrchr(s,'\n');
258 iscntrl(PL_multi_close)
260 PL_multi_close < 32 || PL_multi_close == 127
264 tmpbuf[1] = toCTRL(PL_multi_close);
270 *tmpbuf = PL_multi_close;
274 q = strchr(s,'"') ? '\'' : '"';
275 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
283 Perl_deprecate(pTHX_ char *s)
286 if (ckWARN(WARN_DEPRECATED))
287 Perl_warner(aTHX_ WARN_DEPRECATED, "Use of %s is deprecated", s);
292 * Deprecate a comma-less variable list.
298 deprecate("comma-less variable list");
302 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
303 * utf16-to-utf8-reversed.
309 S_win32_textfilter(pTHX_ int idx, SV *sv, int maxlen)
311 I32 count = FILTER_READ(idx+1, sv, maxlen);
312 if (count > 0 && !maxlen)
313 win32_strip_return(sv);
319 S_utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
321 I32 count = FILTER_READ(idx+1, sv, maxlen);
325 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
326 tend = utf16_to_utf8((U16*)SvPVX(sv), tmps, SvCUR(sv));
327 sv_usepvn(sv, (char*)tmps, tend - tmps);
334 S_utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
336 I32 count = FILTER_READ(idx+1, sv, maxlen);
340 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
341 tend = utf16_to_utf8_reversed((U16*)SvPVX(sv), tmps, SvCUR(sv));
342 sv_usepvn(sv, (char*)tmps, tend - tmps);
350 * Initialize variables. Uses the Perl save_stack to save its state (for
351 * recursive calls to the parser).
355 Perl_lex_start(pTHX_ SV *line)
361 SAVEI32(PL_lex_dojoin);
362 SAVEI32(PL_lex_brackets);
363 SAVEI32(PL_lex_fakebrack);
364 SAVEI32(PL_lex_casemods);
365 SAVEI32(PL_lex_starts);
366 SAVEI32(PL_lex_state);
367 SAVESPTR(PL_lex_inpat);
368 SAVEI32(PL_lex_inwhat);
369 SAVEI16(PL_curcop->cop_line);
372 SAVEPPTR(PL_oldbufptr);
373 SAVEPPTR(PL_oldoldbufptr);
374 SAVEPPTR(PL_linestart);
375 SAVESPTR(PL_linestr);
376 SAVEPPTR(PL_lex_brackstack);
377 SAVEPPTR(PL_lex_casestack);
378 SAVEDESTRUCTOR(restore_rsfp, PL_rsfp);
379 SAVESPTR(PL_lex_stuff);
380 SAVEI32(PL_lex_defer);
381 SAVEI32(PL_sublex_info.sub_inwhat);
382 SAVESPTR(PL_lex_repl);
383 SAVEDESTRUCTOR(restore_expect, PL_tokenbuf + PL_expect); /* encode as pointer */
384 SAVEDESTRUCTOR(restore_lex_expect, PL_tokenbuf + PL_expect);
386 PL_lex_state = LEX_NORMAL;
390 PL_lex_fakebrack = 0;
391 New(899, PL_lex_brackstack, 120, char);
392 New(899, PL_lex_casestack, 12, char);
393 SAVEFREEPV(PL_lex_brackstack);
394 SAVEFREEPV(PL_lex_casestack);
396 *PL_lex_casestack = '\0';
399 PL_lex_stuff = Nullsv;
400 PL_lex_repl = Nullsv;
403 PL_sublex_info.sub_inwhat = 0;
405 if (SvREADONLY(PL_linestr))
406 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
407 s = SvPV(PL_linestr, len);
408 if (len && s[len-1] != ';') {
409 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
410 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
411 sv_catpvn(PL_linestr, "\n;", 2);
413 SvTEMP_off(PL_linestr);
414 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
415 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
417 PL_rs = newSVpvn("\n", 1);
423 * Finalizer for lexing operations. Must be called when the parser is
424 * done with the lexer.
430 PL_doextract = FALSE;
435 * This subroutine has nothing to do with tilting, whether at windmills
436 * or pinball tables. Its name is short for "increment line". It
437 * increments the current line number in PL_curcop->cop_line and checks
438 * to see whether the line starts with a comment of the form
439 * # line 500 "foo.pm"
440 * If so, it sets the current line number and file to the values in the comment.
444 S_incline(pTHX_ char *s)
452 PL_curcop->cop_line++;
455 while (*s == ' ' || *s == '\t') s++;
456 if (strnEQ(s, "line ", 5)) {
465 while (*s == ' ' || *s == '\t')
467 if (*s == '"' && (t = strchr(s+1, '"')))
471 return; /* false alarm */
472 for (t = s; !isSPACE(*t); t++) ;
477 PL_curcop->cop_filegv = gv_fetchfile(s);
479 PL_curcop->cop_filegv = gv_fetchfile(PL_origfilename);
481 PL_curcop->cop_line = atoi(n)-1;
486 * Called to gobble the appropriate amount and type of whitespace.
487 * Skips comments as well.
491 S_skipspace(pTHX_ register char *s)
494 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
495 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
501 SSize_t oldprevlen, oldoldprevlen;
502 SSize_t oldloplen, oldunilen;
503 while (s < PL_bufend && isSPACE(*s)) {
504 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
509 if (s < PL_bufend && *s == '#') {
510 while (s < PL_bufend && *s != '\n')
514 if (PL_in_eval && !PL_rsfp) {
521 /* only continue to recharge the buffer if we're at the end
522 * of the buffer, we're not reading from a source filter, and
523 * we're in normal lexing mode
525 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
526 PL_lex_state == LEX_FORMLINE)
529 /* try to recharge the buffer */
530 if ((s = filter_gets(PL_linestr, PL_rsfp,
531 (prevlen = SvCUR(PL_linestr)))) == Nullch)
533 /* end of file. Add on the -p or -n magic */
534 if (PL_minus_n || PL_minus_p) {
535 sv_setpv(PL_linestr,PL_minus_p ?
536 ";}continue{print or die qq(-p destination: $!\\n)" :
538 sv_catpv(PL_linestr,";}");
539 PL_minus_n = PL_minus_p = 0;
542 sv_setpv(PL_linestr,";");
544 /* reset variables for next time we lex */
545 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
547 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
549 /* Close the filehandle. Could be from -P preprocessor,
550 * STDIN, or a regular file. If we were reading code from
551 * STDIN (because the commandline held no -e or filename)
552 * then we don't close it, we reset it so the code can
553 * read from STDIN too.
556 if (PL_preprocess && !PL_in_eval)
557 (void)PerlProc_pclose(PL_rsfp);
558 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
559 PerlIO_clearerr(PL_rsfp);
561 (void)PerlIO_close(PL_rsfp);
566 /* not at end of file, so we only read another line */
567 /* make corresponding updates to old pointers, for yyerror() */
568 oldprevlen = PL_oldbufptr - PL_bufend;
569 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
571 oldunilen = PL_last_uni - PL_bufend;
573 oldloplen = PL_last_lop - PL_bufend;
574 PL_linestart = PL_bufptr = s + prevlen;
575 PL_bufend = s + SvCUR(PL_linestr);
577 PL_oldbufptr = s + oldprevlen;
578 PL_oldoldbufptr = s + oldoldprevlen;
580 PL_last_uni = s + oldunilen;
582 PL_last_lop = s + oldloplen;
585 /* debugger active and we're not compiling the debugger code,
586 * so store the line into the debugger's array of lines
588 if (PERLDB_LINE && PL_curstash != PL_debstash) {
589 SV *sv = NEWSV(85,0);
591 sv_upgrade(sv, SVt_PVMG);
592 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
593 av_store(GvAV(PL_curcop->cop_filegv),(I32)PL_curcop->cop_line,sv);
600 * Check the unary operators to ensure there's no ambiguity in how they're
601 * used. An ambiguous piece of code would be:
603 * This doesn't mean rand() + 5. Because rand() is a unary operator,
604 * the +5 is its argument.
614 if (PL_oldoldbufptr != PL_last_uni)
616 while (isSPACE(*PL_last_uni))
618 for (s = PL_last_uni; isALNUM_lazy(s) || *s == '-'; s++) ;
619 if ((t = strchr(s, '(')) && t < PL_bufptr)
621 if (ckWARN_d(WARN_AMBIGUOUS)){
624 Perl_warner(aTHX_ WARN_AMBIGUOUS,
625 "Warning: Use of \"%s\" without parens is ambiguous",
631 /* workaround to replace the UNI() macro with a function. Only the
632 * hints/uts.sh file mentions this. Other comments elsewhere in the
633 * source indicate Microport Unix might need it too.
639 #define UNI(f) return uni(f,s)
642 S_uni(pTHX_ I32 f, char *s)
647 PL_last_uni = PL_oldbufptr;
658 #endif /* CRIPPLED_CC */
661 * LOP : macro to build a list operator. Its behaviour has been replaced
662 * with a subroutine, S_lop() for which LOP is just another name.
665 #define LOP(f,x) return lop(f,x,s)
669 * Build a list operator (or something that might be one). The rules:
670 * - if we have a next token, then it's a list operator [why?]
671 * - if the next thing is an opening paren, then it's a function
672 * - else it's a list operator
676 S_lop(pTHX_ I32 f, expectation x, char *s)
683 PL_last_lop = PL_oldbufptr;
698 * When the lexer realizes it knows the next token (for instance,
699 * it is reordering tokens for the parser) then it can call S_force_next
700 * to know what token to return the next time the lexer is called. Caller
701 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
702 * handles the token correctly.
706 S_force_next(pTHX_ I32 type)
708 PL_nexttype[PL_nexttoke] = type;
710 if (PL_lex_state != LEX_KNOWNEXT) {
711 PL_lex_defer = PL_lex_state;
712 PL_lex_expect = PL_expect;
713 PL_lex_state = LEX_KNOWNEXT;
719 * When the lexer knows the next thing is a word (for instance, it has
720 * just seen -> and it knows that the next char is a word char, then
721 * it calls S_force_word to stick the next word into the PL_next lookahead.
724 * char *start : buffer position (must be within PL_linestr)
725 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
726 * int check_keyword : if true, Perl checks to make sure the word isn't
727 * a keyword (do this if the word is a label, e.g. goto FOO)
728 * int allow_pack : if true, : characters will also be allowed (require,
730 * int allow_initial_tick : used by the "sub" lexer only.
734 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
739 start = skipspace(start);
741 if (isIDFIRST_lazy(s) ||
742 (allow_pack && *s == ':') ||
743 (allow_initial_tick && *s == '\'') )
745 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
746 if (check_keyword && keyword(PL_tokenbuf, len))
748 if (token == METHOD) {
753 PL_expect = XOPERATOR;
756 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0));
757 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
765 * Called when the lexer wants $foo *foo &foo etc, but the program
766 * text only contains the "foo" portion. The first argument is a pointer
767 * to the "foo", and the second argument is the type symbol to prefix.
768 * Forces the next token to be a "WORD".
769 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
773 S_force_ident(pTHX_ register char *s, int kind)
776 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
777 PL_nextval[PL_nexttoke].opval = o;
780 dTHR; /* just for in_eval */
781 o->op_private = OPpCONST_ENTERED;
782 /* XXX see note in pp_entereval() for why we forgo typo
783 warnings if the symbol must be introduced in an eval.
785 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
786 kind == '$' ? SVt_PV :
787 kind == '@' ? SVt_PVAV :
788 kind == '%' ? SVt_PVHV :
797 * Forces the next token to be a version number.
801 S_force_version(pTHX_ char *s)
803 OP *version = Nullop;
807 /* default VERSION number -- GBARR */
812 for( d=s, c = 1; isDIGIT(*d) || *d == '_' || (*d == '.' && c--); d++);
813 if((*d == ';' || isSPACE(*d)) && *(skipspace(d)) != ',') {
815 /* real VERSION number -- GBARR */
816 version = yylval.opval;
820 /* NOTE: The parser sees the package name and the VERSION swapped */
821 PL_nextval[PL_nexttoke].opval = version;
829 * Tokenize a quoted string passed in as an SV. It finds the next
830 * chunk, up to end of string or a backslash. It may make a new
831 * SV containing that chunk (if HINT_NEW_STRING is on). It also
836 S_tokeq(pTHX_ SV *sv)
847 s = SvPV_force(sv, len);
851 while (s < send && *s != '\\')
856 if ( PL_hints & HINT_NEW_STRING )
857 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
860 if (s + 1 < send && (s[1] == '\\'))
861 s++; /* all that, just for this */
866 SvCUR_set(sv, d - SvPVX(sv));
868 if ( PL_hints & HINT_NEW_STRING )
869 return new_constant(NULL, 0, "q", sv, pv, "q");
874 * Now come three functions related to double-quote context,
875 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
876 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
877 * interact with PL_lex_state, and create fake ( ... ) argument lists
878 * to handle functions and concatenation.
879 * They assume that whoever calls them will be setting up a fake
880 * join call, because each subthing puts a ',' after it. This lets
883 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
885 * (I'm not sure whether the spurious commas at the end of lcfirst's
886 * arguments and join's arguments are created or not).
891 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
893 * Pattern matching will set PL_lex_op to the pattern-matching op to
894 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
896 * OP_CONST and OP_READLINE are easy--just make the new op and return.
898 * Everything else becomes a FUNC.
900 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
901 * had an OP_CONST or OP_READLINE). This just sets us up for a
902 * call to S_sublex_push().
908 register I32 op_type = yylval.ival;
910 if (op_type == OP_NULL) {
911 yylval.opval = PL_lex_op;
915 if (op_type == OP_CONST || op_type == OP_READLINE) {
916 SV *sv = tokeq(PL_lex_stuff);
918 if (SvTYPE(sv) == SVt_PVIV) {
919 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
925 nsv = newSVpvn(p, len);
929 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
930 PL_lex_stuff = Nullsv;
934 PL_sublex_info.super_state = PL_lex_state;
935 PL_sublex_info.sub_inwhat = op_type;
936 PL_sublex_info.sub_op = PL_lex_op;
937 PL_lex_state = LEX_INTERPPUSH;
941 yylval.opval = PL_lex_op;
951 * Create a new scope to save the lexing state. The scope will be
952 * ended in S_sublex_done. Returns a '(', starting the function arguments
953 * to the uc, lc, etc. found before.
954 * Sets PL_lex_state to LEX_INTERPCONCAT.
963 PL_lex_state = PL_sublex_info.super_state;
964 SAVEI32(PL_lex_dojoin);
965 SAVEI32(PL_lex_brackets);
966 SAVEI32(PL_lex_fakebrack);
967 SAVEI32(PL_lex_casemods);
968 SAVEI32(PL_lex_starts);
969 SAVEI32(PL_lex_state);
970 SAVESPTR(PL_lex_inpat);
971 SAVEI32(PL_lex_inwhat);
972 SAVEI16(PL_curcop->cop_line);
974 SAVEPPTR(PL_oldbufptr);
975 SAVEPPTR(PL_oldoldbufptr);
976 SAVEPPTR(PL_linestart);
977 SAVESPTR(PL_linestr);
978 SAVEPPTR(PL_lex_brackstack);
979 SAVEPPTR(PL_lex_casestack);
981 PL_linestr = PL_lex_stuff;
982 PL_lex_stuff = Nullsv;
984 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
986 PL_bufend += SvCUR(PL_linestr);
987 SAVEFREESV(PL_linestr);
989 PL_lex_dojoin = FALSE;
991 PL_lex_fakebrack = 0;
992 New(899, PL_lex_brackstack, 120, char);
993 New(899, PL_lex_casestack, 12, char);
994 SAVEFREEPV(PL_lex_brackstack);
995 SAVEFREEPV(PL_lex_casestack);
997 *PL_lex_casestack = '\0';
999 PL_lex_state = LEX_INTERPCONCAT;
1000 PL_curcop->cop_line = PL_multi_start;
1002 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1003 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1004 PL_lex_inpat = PL_sublex_info.sub_op;
1006 PL_lex_inpat = Nullop;
1013 * Restores lexer state after a S_sublex_push.
1019 if (!PL_lex_starts++) {
1020 PL_expect = XOPERATOR;
1021 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpvn("",0));
1025 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1026 PL_lex_state = LEX_INTERPCASEMOD;
1030 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1031 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1032 PL_linestr = PL_lex_repl;
1034 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1035 PL_bufend += SvCUR(PL_linestr);
1036 SAVEFREESV(PL_linestr);
1037 PL_lex_dojoin = FALSE;
1038 PL_lex_brackets = 0;
1039 PL_lex_fakebrack = 0;
1040 PL_lex_casemods = 0;
1041 *PL_lex_casestack = '\0';
1043 if (SvEVALED(PL_lex_repl)) {
1044 PL_lex_state = LEX_INTERPNORMAL;
1046 /* we don't clear PL_lex_repl here, so that we can check later
1047 whether this is an evalled subst; that means we rely on the
1048 logic to ensure sublex_done() is called again only via the
1049 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1052 PL_lex_state = LEX_INTERPCONCAT;
1053 PL_lex_repl = Nullsv;
1059 PL_bufend = SvPVX(PL_linestr);
1060 PL_bufend += SvCUR(PL_linestr);
1061 PL_expect = XOPERATOR;
1062 PL_sublex_info.sub_inwhat = 0;
1070 Extracts a pattern, double-quoted string, or transliteration. This
1073 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1074 processing a pattern (PL_lex_inpat is true), a transliteration
1075 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1077 Returns a pointer to the character scanned up to. Iff this is
1078 advanced from the start pointer supplied (ie if anything was
1079 successfully parsed), will leave an OP for the substring scanned
1080 in yylval. Caller must intuit reason for not parsing further
1081 by looking at the next characters herself.
1085 double-quoted style: \r and \n
1086 regexp special ones: \D \s
1088 backrefs: \1 (deprecated in substitution replacements)
1089 case and quoting: \U \Q \E
1090 stops on @ and $, but not for $ as tail anchor
1092 In transliterations:
1093 characters are VERY literal, except for - not at the start or end
1094 of the string, which indicates a range. scan_const expands the
1095 range to the full set of intermediate characters.
1097 In double-quoted strings:
1099 double-quoted style: \r and \n
1101 backrefs: \1 (deprecated)
1102 case and quoting: \U \Q \E
1105 scan_const does *not* construct ops to handle interpolated strings.
1106 It stops processing as soon as it finds an embedded $ or @ variable
1107 and leaves it to the caller to work out what's going on.
1109 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @:foo.
1111 $ in pattern could be $foo or could be tail anchor. Assumption:
1112 it's a tail anchor if $ is the last thing in the string, or if it's
1113 followed by one of ")| \n\t"
1115 \1 (backreferences) are turned into $1
1117 The structure of the code is
1118 while (there's a character to process) {
1119 handle transliteration ranges
1120 skip regexp comments
1121 skip # initiated comments in //x patterns
1122 check for embedded @foo
1123 check for embedded scalars
1125 leave intact backslashes from leave (below)
1126 deprecate \1 in strings and sub replacements
1127 handle string-changing backslashes \l \U \Q \E, etc.
1128 switch (what was escaped) {
1129 handle - in a transliteration (becomes a literal -)
1130 handle \132 octal characters
1131 handle 0x15 hex characters
1132 handle \cV (control V)
1133 handle printf backslashes (\f, \r, \n, etc)
1135 } (end if backslash)
1136 } (end while character to read)
1141 S_scan_const(pTHX_ char *start)
1143 register char *send = PL_bufend; /* end of the constant */
1144 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1145 register char *s = start; /* start of the constant */
1146 register char *d = SvPVX(sv); /* destination for copies */
1147 bool dorange = FALSE; /* are we in a translit range? */
1149 I32 utf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1150 ? (PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
1152 I32 thisutf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1153 ? (PL_sublex_info.sub_op->op_private & (PL_lex_repl ?
1154 OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF))
1156 char *leaveit = /* set of acceptably-backslashed characters */
1158 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
1161 while (s < send || dorange) {
1162 /* get transliterations out of the way (they're most literal) */
1163 if (PL_lex_inwhat == OP_TRANS) {
1164 /* expand a range A-Z to the full set of characters. AIE! */
1166 I32 i; /* current expanded character */
1167 I32 min; /* first character in range */
1168 I32 max; /* last character in range */
1170 i = d - SvPVX(sv); /* remember current offset */
1171 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1172 d = SvPVX(sv) + i; /* refresh d after realloc */
1173 d -= 2; /* eat the first char and the - */
1175 min = (U8)*d; /* first char in range */
1176 max = (U8)d[1]; /* last char in range */
1179 if ((isLOWER(min) && isLOWER(max)) ||
1180 (isUPPER(min) && isUPPER(max))) {
1182 for (i = min; i <= max; i++)
1186 for (i = min; i <= max; i++)
1193 for (i = min; i <= max; i++)
1196 /* mark the range as done, and continue */
1201 /* range begins (ignore - as first or last char) */
1202 else if (*s == '-' && s+1 < send && s != start) {
1204 *d++ = (char)0xff; /* use illegal utf8 byte--see pmtrans */
1213 /* if we get here, we're not doing a transliteration */
1215 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1216 except for the last char, which will be done separately. */
1217 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1219 while (s < send && *s != ')')
1221 } else if (s[2] == '{'
1222 || s[2] == 'p' && s[3] == '{') { /* This should march regcomp.c */
1224 char *regparse = s + (s[2] == '{' ? 3 : 4);
1227 while (count && (c = *regparse)) {
1228 if (c == '\\' && regparse[1])
1236 if (*regparse != ')') {
1237 regparse--; /* Leave one char for continuation. */
1238 yyerror("Sequence (?{...}) not terminated or not {}-balanced");
1240 while (s < regparse)
1245 /* likewise skip #-initiated comments in //x patterns */
1246 else if (*s == '#' && PL_lex_inpat &&
1247 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1248 while (s+1 < send && *s != '\n')
1252 /* check for embedded arrays (@foo, @:foo, @'foo, @{foo}, @$foo) */
1253 else if (*s == '@' && s[1] && (isALNUM_lazy(s+1) || strchr(":'{$", s[1])))
1256 /* check for embedded scalars. only stop if we're sure it's a
1259 else if (*s == '$') {
1260 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1262 if (s + 1 < send && !strchr("()| \n\t", s[1]))
1263 break; /* in regexp, $ might be tail anchor */
1266 /* (now in tr/// code again) */
1268 if (*s & 0x80 && thisutf) {
1269 dTHR; /* only for ckWARN */
1270 if (ckWARN(WARN_UTF8)) {
1271 (void)utf8_to_uv((U8*)s, &len); /* could cvt latin-1 to utf8 here... */
1281 if (*s == '\\' && s+1 < send) {
1284 /* some backslashes we leave behind */
1285 if (*leaveit && *s && strchr(leaveit, *s)) {
1291 /* deprecate \1 in strings and substitution replacements */
1292 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1293 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1295 dTHR; /* only for ckWARN */
1296 if (ckWARN(WARN_SYNTAX))
1297 Perl_warner(aTHX_ WARN_SYNTAX, "\\%c better written as $%c", *s, *s);
1302 /* string-change backslash escapes */
1303 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1308 /* if we get here, it's either a quoted -, or a digit */
1311 /* quoted - in transliterations */
1313 if (PL_lex_inwhat == OP_TRANS) {
1321 if (ckWARN(WARN_UNSAFE) && isALPHA(*s))
1322 Perl_warner(aTHX_ WARN_UNSAFE,
1323 "Unrecognized escape \\%c passed through",
1325 /* default action is to copy the quoted character */
1330 /* \132 indicates an octal constant */
1331 case '0': case '1': case '2': case '3':
1332 case '4': case '5': case '6': case '7':
1333 *d++ = scan_oct(s, 3, &len);
1337 /* \x24 indicates a hex constant */
1341 char* e = strchr(s, '}');
1344 yyerror("Missing right brace on \\x{}");
1349 if (ckWARN(WARN_UTF8))
1350 Perl_warner(aTHX_ WARN_UTF8,
1351 "Use of \\x{} without utf8 declaration");
1353 /* note: utf always shorter than hex */
1354 d = (char*)uv_to_utf8((U8*)d,
1355 scan_hex(s + 1, e - s - 1, &len));
1359 UV uv = (UV)scan_hex(s, 2, &len);
1360 if (utf && PL_lex_inwhat == OP_TRANS &&
1361 utf != (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
1363 d = (char*)uv_to_utf8((U8*)d, uv); /* doing a CU or UC */
1366 if (uv >= 127 && UTF) {
1368 if (ckWARN(WARN_UTF8))
1369 Perl_warner(aTHX_ WARN_UTF8,
1370 "\\x%.*s will produce malformed UTF-8 character; use \\x{%.*s} for that",
1379 /* \N{latin small letter a} is a named character */
1383 char* e = strchr(s, '}');
1392 yyerror("Missing right brace on \\N{}");
1396 res = newSVpvn(s + 1, e - s - 1);
1397 res = new_constant( Nullch, 0, "charnames",
1398 res, Nullsv, "\\N{...}" );
1399 str = SvPV(res,len);
1400 if (len > e - s + 4) {
1401 char *odest = SvPVX(sv);
1403 SvGROW(sv, (SvCUR(sv) + len - (e - s + 4)));
1404 d = SvPVX(sv) + (d - odest);
1406 Copy(str, d, len, char);
1413 yyerror("Missing braces on \\N{}");
1416 /* \c is a control character */
1430 /* printf-style backslashes, formfeeds, newlines, etc */
1448 *d++ = '\047'; /* CP 1047 */
1451 *d++ = '\057'; /* CP 1047 */
1465 } /* end if (backslash) */
1468 } /* while loop to process each character */
1470 /* terminate the string and set up the sv */
1472 SvCUR_set(sv, d - SvPVX(sv));
1475 /* shrink the sv if we allocated more than we used */
1476 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1477 SvLEN_set(sv, SvCUR(sv) + 1);
1478 Renew(SvPVX(sv), SvLEN(sv), char);
1481 /* return the substring (via yylval) only if we parsed anything */
1482 if (s > PL_bufptr) {
1483 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1484 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1486 ( PL_lex_inwhat == OP_TRANS
1488 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1491 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1498 * Returns TRUE if there's more to the expression (e.g., a subscript),
1501 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1503 * ->[ and ->{ return TRUE
1504 * { and [ outside a pattern are always subscripts, so return TRUE
1505 * if we're outside a pattern and it's not { or [, then return FALSE
1506 * if we're in a pattern and the first char is a {
1507 * {4,5} (any digits around the comma) returns FALSE
1508 * if we're in a pattern and the first char is a [
1510 * [SOMETHING] has a funky algorithm to decide whether it's a
1511 * character class or not. It has to deal with things like
1512 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1513 * anything else returns TRUE
1516 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1519 S_intuit_more(pTHX_ register char *s)
1521 if (PL_lex_brackets)
1523 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1525 if (*s != '{' && *s != '[')
1530 /* In a pattern, so maybe we have {n,m}. */
1547 /* On the other hand, maybe we have a character class */
1550 if (*s == ']' || *s == '^')
1553 /* this is terrifying, and it works */
1554 int weight = 2; /* let's weigh the evidence */
1556 unsigned char un_char = 255, last_un_char;
1557 char *send = strchr(s,']');
1558 char tmpbuf[sizeof PL_tokenbuf * 4];
1560 if (!send) /* has to be an expression */
1563 Zero(seen,256,char);
1566 else if (isDIGIT(*s)) {
1568 if (isDIGIT(s[1]) && s[2] == ']')
1574 for (; s < send; s++) {
1575 last_un_char = un_char;
1576 un_char = (unsigned char)*s;
1581 weight -= seen[un_char] * 10;
1582 if (isALNUM_lazy(s+1)) {
1583 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1584 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1589 else if (*s == '$' && s[1] &&
1590 strchr("[#!%*<>()-=",s[1])) {
1591 if (/*{*/ strchr("])} =",s[2]))
1600 if (strchr("wds]",s[1]))
1602 else if (seen['\''] || seen['"'])
1604 else if (strchr("rnftbxcav",s[1]))
1606 else if (isDIGIT(s[1])) {
1608 while (s[1] && isDIGIT(s[1]))
1618 if (strchr("aA01! ",last_un_char))
1620 if (strchr("zZ79~",s[1]))
1622 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1623 weight -= 5; /* cope with negative subscript */
1626 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1627 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1632 if (keyword(tmpbuf, d - tmpbuf))
1635 if (un_char == last_un_char + 1)
1637 weight -= seen[un_char];
1642 if (weight >= 0) /* probably a character class */
1652 * Does all the checking to disambiguate
1654 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1655 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1657 * First argument is the stuff after the first token, e.g. "bar".
1659 * Not a method if bar is a filehandle.
1660 * Not a method if foo is a subroutine prototyped to take a filehandle.
1661 * Not a method if it's really "Foo $bar"
1662 * Method if it's "foo $bar"
1663 * Not a method if it's really "print foo $bar"
1664 * Method if it's really "foo package::" (interpreted as package->foo)
1665 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1666 * Not a method if bar is a filehandle or package, but is quotd with
1671 S_intuit_method(pTHX_ char *start, GV *gv)
1673 char *s = start + (*start == '$');
1674 char tmpbuf[sizeof PL_tokenbuf];
1682 if ((cv = GvCVu(gv))) {
1683 char *proto = SvPVX(cv);
1693 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1694 /* start is the beginning of the possible filehandle/object,
1695 * and s is the end of it
1696 * tmpbuf is a copy of it
1699 if (*start == '$') {
1700 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1705 return *s == '(' ? FUNCMETH : METHOD;
1707 if (!keyword(tmpbuf, len)) {
1708 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1713 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1714 if (indirgv && GvCVu(indirgv))
1716 /* filehandle or package name makes it a method */
1717 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1719 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1720 return 0; /* no assumptions -- "=>" quotes bearword */
1722 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1723 newSVpvn(tmpbuf,len));
1724 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1728 return *s == '(' ? FUNCMETH : METHOD;
1736 * Return a string of Perl code to load the debugger. If PERL5DB
1737 * is set, it will return the contents of that, otherwise a
1738 * compile-time require of perl5db.pl.
1745 char *pdb = PerlEnv_getenv("PERL5DB");
1749 SETERRNO(0,SS$_NORMAL);
1750 return "BEGIN { require 'perl5db.pl' }";
1756 /* Encoded script support. filter_add() effectively inserts a
1757 * 'pre-processing' function into the current source input stream.
1758 * Note that the filter function only applies to the current source file
1759 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1761 * The datasv parameter (which may be NULL) can be used to pass
1762 * private data to this instance of the filter. The filter function
1763 * can recover the SV using the FILTER_DATA macro and use it to
1764 * store private buffers and state information.
1766 * The supplied datasv parameter is upgraded to a PVIO type
1767 * and the IoDIRP field is used to store the function pointer.
1768 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1769 * private use must be set using malloc'd pointers.
1773 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1775 if (!funcp){ /* temporary handy debugging hack to be deleted */
1776 PL_filter_debug = atoi((char*)datasv);
1779 if (!PL_rsfp_filters)
1780 PL_rsfp_filters = newAV();
1782 datasv = NEWSV(255,0);
1783 if (!SvUPGRADE(datasv, SVt_PVIO))
1784 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1785 IoDIRP(datasv) = (DIR*)funcp; /* stash funcp into spare field */
1787 if (PL_filter_debug) {
1789 Perl_warn(aTHX_ "filter_add func %p (%s)", funcp, SvPV(datasv, n_a));
1791 #endif /* DEBUGGING */
1792 av_unshift(PL_rsfp_filters, 1);
1793 av_store(PL_rsfp_filters, 0, datasv) ;
1798 /* Delete most recently added instance of this filter function. */
1800 Perl_filter_del(pTHX_ filter_t funcp)
1803 if (PL_filter_debug)
1804 Perl_warn(aTHX_ "filter_del func %p", funcp);
1805 #endif /* DEBUGGING */
1806 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
1808 /* if filter is on top of stack (usual case) just pop it off */
1809 if (IoDIRP(FILTER_DATA(AvFILLp(PL_rsfp_filters))) == (DIR*)funcp){
1810 IoDIRP(FILTER_DATA(AvFILLp(PL_rsfp_filters))) = NULL;
1811 sv_free(av_pop(PL_rsfp_filters));
1815 /* we need to search for the correct entry and clear it */
1816 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
1820 /* Invoke the n'th filter function for the current rsfp. */
1822 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
1825 /* 0 = read one text line */
1830 if (!PL_rsfp_filters)
1832 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
1833 /* Provide a default input filter to make life easy. */
1834 /* Note that we append to the line. This is handy. */
1836 if (PL_filter_debug)
1837 Perl_warn(aTHX_ "filter_read %d: from rsfp\n", idx);
1838 #endif /* DEBUGGING */
1842 int old_len = SvCUR(buf_sv) ;
1844 /* ensure buf_sv is large enough */
1845 SvGROW(buf_sv, old_len + maxlen) ;
1846 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
1847 if (PerlIO_error(PL_rsfp))
1848 return -1; /* error */
1850 return 0 ; /* end of file */
1852 SvCUR_set(buf_sv, old_len + len) ;
1855 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
1856 if (PerlIO_error(PL_rsfp))
1857 return -1; /* error */
1859 return 0 ; /* end of file */
1862 return SvCUR(buf_sv);
1864 /* Skip this filter slot if filter has been deleted */
1865 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
1867 if (PL_filter_debug)
1868 Perl_warn(aTHX_ "filter_read %d: skipped (filter deleted)\n", idx);
1869 #endif /* DEBUGGING */
1870 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
1872 /* Get function pointer hidden within datasv */
1873 funcp = (filter_t)IoDIRP(datasv);
1875 if (PL_filter_debug) {
1877 Perl_warn(aTHX_ "filter_read %d: via function %p (%s)\n",
1878 idx, funcp, SvPV(datasv,n_a));
1880 #endif /* DEBUGGING */
1881 /* Call function. The function is expected to */
1882 /* call "FILTER_READ(idx+1, buf_sv)" first. */
1883 /* Return: <0:error, =0:eof, >0:not eof */
1884 return (*funcp)(aTHXo_ idx, buf_sv, maxlen);
1888 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
1891 if (!PL_rsfp_filters) {
1892 filter_add(win32_textfilter,NULL);
1895 if (PL_rsfp_filters) {
1898 SvCUR_set(sv, 0); /* start with empty line */
1899 if (FILTER_READ(0, sv, 0) > 0)
1900 return ( SvPVX(sv) ) ;
1905 return (sv_gets(sv, fp, append));
1910 static char* exp_name[] =
1911 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
1912 "ATTRTERM", "TERMBLOCK"
1919 Works out what to call the token just pulled out of the input
1920 stream. The yacc parser takes care of taking the ops we return and
1921 stitching them into a tree.
1927 if read an identifier
1928 if we're in a my declaration
1929 croak if they tried to say my($foo::bar)
1930 build the ops for a my() declaration
1931 if it's an access to a my() variable
1932 are we in a sort block?
1933 croak if my($a); $a <=> $b
1934 build ops for access to a my() variable
1935 if in a dq string, and they've said @foo and we can't find @foo
1937 build ops for a bareword
1938 if we already built the token before, use it.
1942 #ifdef USE_PURE_BISON
1943 Perl_yylex(pTHX_ YYSTYPE *lvalp, int *lcharp)
1956 #ifdef USE_PURE_BISON
1957 yylval_pointer = lvalp;
1958 yychar_pointer = lcharp;
1961 /* check if there's an identifier for us to look at */
1962 if (PL_pending_ident) {
1963 /* pit holds the identifier we read and pending_ident is reset */
1964 char pit = PL_pending_ident;
1965 PL_pending_ident = 0;
1967 /* if we're in a my(), we can't allow dynamics here.
1968 $foo'bar has already been turned into $foo::bar, so
1969 just check for colons.
1971 if it's a legal name, the OP is a PADANY.
1974 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
1975 tmp = pad_allocmy(PL_tokenbuf);
1978 if (strchr(PL_tokenbuf,':'))
1979 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
1981 yylval.opval = newOP(OP_PADANY, 0);
1982 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
1988 build the ops for accesses to a my() variable.
1990 Deny my($a) or my($b) in a sort block, *if* $a or $b is
1991 then used in a comparison. This catches most, but not
1992 all cases. For instance, it catches
1993 sort { my($a); $a <=> $b }
1995 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
1996 (although why you'd do that is anyone's guess).
1999 if (!strchr(PL_tokenbuf,':')) {
2001 /* Check for single character per-thread SVs */
2002 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
2003 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
2004 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
2006 yylval.opval = newOP(OP_THREADSV, 0);
2007 yylval.opval->op_targ = tmp;
2010 #endif /* USE_THREADS */
2011 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
2012 /* might be an "our" variable" */
2013 if (SvFLAGS(AvARRAY(PL_comppad_name)[tmp]) & SVpad_OUR) {
2014 /* build ops for a bareword */
2015 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2016 yylval.opval->op_private = OPpCONST_ENTERED;
2017 gv_fetchpv(PL_tokenbuf+1,
2019 ? (GV_ADDMULTI | GV_ADDINEVAL | GV_ADDOUR)
2022 ((PL_tokenbuf[0] == '$') ? SVt_PV
2023 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2028 /* if it's a sort block and they're naming $a or $b */
2029 if (PL_last_lop_op == OP_SORT &&
2030 PL_tokenbuf[0] == '$' &&
2031 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
2034 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
2035 d < PL_bufend && *d != '\n';
2038 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
2039 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
2045 yylval.opval = newOP(OP_PADANY, 0);
2046 yylval.opval->op_targ = tmp;
2052 Whine if they've said @foo in a doublequoted string,
2053 and @foo isn't a variable we can find in the symbol
2056 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
2057 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
2058 if (!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
2059 yyerror(Perl_form(aTHX_ "In string, %s now must be written as \\%s",
2060 PL_tokenbuf, PL_tokenbuf));
2063 /* build ops for a bareword */
2064 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2065 yylval.opval->op_private = OPpCONST_ENTERED;
2066 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
2067 ((PL_tokenbuf[0] == '$') ? SVt_PV
2068 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2073 /* no identifier pending identification */
2075 switch (PL_lex_state) {
2077 case LEX_NORMAL: /* Some compilers will produce faster */
2078 case LEX_INTERPNORMAL: /* code if we comment these out. */
2082 /* when we've already built the next token, just pull it out of the queue */
2085 yylval = PL_nextval[PL_nexttoke];
2087 PL_lex_state = PL_lex_defer;
2088 PL_expect = PL_lex_expect;
2089 PL_lex_defer = LEX_NORMAL;
2091 return(PL_nexttype[PL_nexttoke]);
2093 /* interpolated case modifiers like \L \U, including \Q and \E.
2094 when we get here, PL_bufptr is at the \
2096 case LEX_INTERPCASEMOD:
2098 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2099 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2101 /* handle \E or end of string */
2102 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2106 if (PL_lex_casemods) {
2107 oldmod = PL_lex_casestack[--PL_lex_casemods];
2108 PL_lex_casestack[PL_lex_casemods] = '\0';
2110 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2112 PL_lex_state = LEX_INTERPCONCAT;
2116 if (PL_bufptr != PL_bufend)
2118 PL_lex_state = LEX_INTERPCONCAT;
2123 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2124 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2125 if (strchr("LU", *s) &&
2126 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2128 PL_lex_casestack[--PL_lex_casemods] = '\0';
2131 if (PL_lex_casemods > 10) {
2132 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2133 if (newlb != PL_lex_casestack) {
2135 PL_lex_casestack = newlb;
2138 PL_lex_casestack[PL_lex_casemods++] = *s;
2139 PL_lex_casestack[PL_lex_casemods] = '\0';
2140 PL_lex_state = LEX_INTERPCONCAT;
2141 PL_nextval[PL_nexttoke].ival = 0;
2144 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2146 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2148 PL_nextval[PL_nexttoke].ival = OP_LC;
2150 PL_nextval[PL_nexttoke].ival = OP_UC;
2152 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2154 Perl_croak(aTHX_ "panic: yylex");
2157 if (PL_lex_starts) {
2166 case LEX_INTERPPUSH:
2167 return sublex_push();
2169 case LEX_INTERPSTART:
2170 if (PL_bufptr == PL_bufend)
2171 return sublex_done();
2173 PL_lex_dojoin = (*PL_bufptr == '@');
2174 PL_lex_state = LEX_INTERPNORMAL;
2175 if (PL_lex_dojoin) {
2176 PL_nextval[PL_nexttoke].ival = 0;
2179 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2180 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2181 force_next(PRIVATEREF);
2183 force_ident("\"", '$');
2184 #endif /* USE_THREADS */
2185 PL_nextval[PL_nexttoke].ival = 0;
2187 PL_nextval[PL_nexttoke].ival = 0;
2189 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2192 if (PL_lex_starts++) {
2198 case LEX_INTERPENDMAYBE:
2199 if (intuit_more(PL_bufptr)) {
2200 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2206 if (PL_lex_dojoin) {
2207 PL_lex_dojoin = FALSE;
2208 PL_lex_state = LEX_INTERPCONCAT;
2211 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2212 && SvEVALED(PL_lex_repl))
2214 if (PL_bufptr != PL_bufend)
2215 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2216 PL_lex_repl = Nullsv;
2219 case LEX_INTERPCONCAT:
2221 if (PL_lex_brackets)
2222 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2224 if (PL_bufptr == PL_bufend)
2225 return sublex_done();
2227 if (SvIVX(PL_linestr) == '\'') {
2228 SV *sv = newSVsv(PL_linestr);
2231 else if ( PL_hints & HINT_NEW_RE )
2232 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2233 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2237 s = scan_const(PL_bufptr);
2239 PL_lex_state = LEX_INTERPCASEMOD;
2241 PL_lex_state = LEX_INTERPSTART;
2244 if (s != PL_bufptr) {
2245 PL_nextval[PL_nexttoke] = yylval;
2248 if (PL_lex_starts++)
2258 PL_lex_state = LEX_NORMAL;
2259 s = scan_formline(PL_bufptr);
2260 if (!PL_lex_formbrack)
2266 PL_oldoldbufptr = PL_oldbufptr;
2269 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2270 exp_name[PL_expect], s);
2276 if (isIDFIRST_lazy(s))
2278 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2281 goto fake_eof; /* emulate EOF on ^D or ^Z */
2286 if (PL_lex_brackets)
2287 yyerror("Missing right curly or square bracket");
2290 if (s++ < PL_bufend)
2291 goto retry; /* ignore stray nulls */
2294 if (!PL_in_eval && !PL_preambled) {
2295 PL_preambled = TRUE;
2296 sv_setpv(PL_linestr,incl_perldb());
2297 if (SvCUR(PL_linestr))
2298 sv_catpv(PL_linestr,";");
2300 while(AvFILLp(PL_preambleav) >= 0) {
2301 SV *tmpsv = av_shift(PL_preambleav);
2302 sv_catsv(PL_linestr, tmpsv);
2303 sv_catpv(PL_linestr, ";");
2306 sv_free((SV*)PL_preambleav);
2307 PL_preambleav = NULL;
2309 if (PL_minus_n || PL_minus_p) {
2310 sv_catpv(PL_linestr, "LINE: while (<>) {");
2312 sv_catpv(PL_linestr,"chomp;");
2314 GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
2316 GvIMPORTED_AV_on(gv);
2318 if (strchr("/'\"", *PL_splitstr)
2319 && strchr(PL_splitstr + 1, *PL_splitstr))
2320 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2323 s = "'~#\200\1'"; /* surely one char is unused...*/
2324 while (s[1] && strchr(PL_splitstr, *s)) s++;
2326 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c",
2327 "q" + (delim == '\''), delim);
2328 for (s = PL_splitstr; *s; s++) {
2330 sv_catpvn(PL_linestr, "\\", 1);
2331 sv_catpvn(PL_linestr, s, 1);
2333 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2337 sv_catpv(PL_linestr,"@F=split(' ');");
2340 sv_catpv(PL_linestr, "\n");
2341 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2342 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2343 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2344 SV *sv = NEWSV(85,0);
2346 sv_upgrade(sv, SVt_PVMG);
2347 sv_setsv(sv,PL_linestr);
2348 av_store(GvAV(PL_curcop->cop_filegv),(I32)PL_curcop->cop_line,sv);
2353 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2356 if (PL_preprocess && !PL_in_eval)
2357 (void)PerlProc_pclose(PL_rsfp);
2358 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2359 PerlIO_clearerr(PL_rsfp);
2361 (void)PerlIO_close(PL_rsfp);
2363 PL_doextract = FALSE;
2365 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2366 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2367 sv_catpv(PL_linestr,";}");
2368 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2369 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2370 PL_minus_n = PL_minus_p = 0;
2373 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2374 sv_setpv(PL_linestr,"");
2375 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2378 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2379 PL_doextract = FALSE;
2381 /* Incest with pod. */
2382 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2383 sv_setpv(PL_linestr, "");
2384 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2385 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2386 PL_doextract = FALSE;
2390 } while (PL_doextract);
2391 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2392 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2393 SV *sv = NEWSV(85,0);
2395 sv_upgrade(sv, SVt_PVMG);
2396 sv_setsv(sv,PL_linestr);
2397 av_store(GvAV(PL_curcop->cop_filegv),(I32)PL_curcop->cop_line,sv);
2399 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2400 if (PL_curcop->cop_line == 1) {
2401 while (s < PL_bufend && isSPACE(*s))
2403 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2407 if (*s == '#' && *(s+1) == '!')
2409 #ifdef ALTERNATE_SHEBANG
2411 static char as[] = ALTERNATE_SHEBANG;
2412 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2413 d = s + (sizeof(as) - 1);
2415 #endif /* ALTERNATE_SHEBANG */
2424 while (*d && !isSPACE(*d))
2428 #ifdef ARG_ZERO_IS_SCRIPT
2429 if (ipathend > ipath) {
2431 * HP-UX (at least) sets argv[0] to the script name,
2432 * which makes $^X incorrect. And Digital UNIX and Linux,
2433 * at least, set argv[0] to the basename of the Perl
2434 * interpreter. So, having found "#!", we'll set it right.
2436 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2437 assert(SvPOK(x) || SvGMAGICAL(x));
2438 if (sv_eq(x, GvSV(PL_curcop->cop_filegv))) {
2439 sv_setpvn(x, ipath, ipathend - ipath);
2442 TAINT_NOT; /* $^X is always tainted, but that's OK */
2444 #endif /* ARG_ZERO_IS_SCRIPT */
2449 d = instr(s,"perl -");
2451 d = instr(s,"perl");
2453 /* avoid getting into infinite loops when shebang
2454 * line contains "Perl" rather than "perl" */
2456 for (d = ipathend-4; d >= ipath; --d) {
2457 if ((*d == 'p' || *d == 'P')
2458 && !ibcmp(d, "perl", 4))
2468 #ifdef ALTERNATE_SHEBANG
2470 * If the ALTERNATE_SHEBANG on this system starts with a
2471 * character that can be part of a Perl expression, then if
2472 * we see it but not "perl", we're probably looking at the
2473 * start of Perl code, not a request to hand off to some
2474 * other interpreter. Similarly, if "perl" is there, but
2475 * not in the first 'word' of the line, we assume the line
2476 * contains the start of the Perl program.
2478 if (d && *s != '#') {
2480 while (*c && !strchr("; \t\r\n\f\v#", *c))
2483 d = Nullch; /* "perl" not in first word; ignore */
2485 *s = '#'; /* Don't try to parse shebang line */
2487 #endif /* ALTERNATE_SHEBANG */
2492 !instr(s,"indir") &&
2493 instr(PL_origargv[0],"perl"))
2499 while (s < PL_bufend && isSPACE(*s))
2501 if (s < PL_bufend) {
2502 Newz(899,newargv,PL_origargc+3,char*);
2504 while (s < PL_bufend && !isSPACE(*s))
2507 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2510 newargv = PL_origargv;
2512 PerlProc_execv(ipath, newargv);
2513 Perl_croak(aTHX_ "Can't exec %s", ipath);
2516 U32 oldpdb = PL_perldb;
2517 bool oldn = PL_minus_n;
2518 bool oldp = PL_minus_p;
2520 while (*d && !isSPACE(*d)) d++;
2521 while (*d == ' ' || *d == '\t') d++;
2525 if (*d == 'M' || *d == 'm') {
2527 while (*d && !isSPACE(*d)) d++;
2528 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2531 d = moreswitches(d);
2533 if (PERLDB_LINE && !oldpdb ||
2534 ( PL_minus_n || PL_minus_p ) && !(oldn || oldp) )
2535 /* if we have already added "LINE: while (<>) {",
2536 we must not do it again */
2538 sv_setpv(PL_linestr, "");
2539 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2540 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2541 PL_preambled = FALSE;
2543 (void)gv_fetchfile(PL_origfilename);
2550 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2552 PL_lex_state = LEX_FORMLINE;
2557 #ifdef PERL_STRICT_CR
2558 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2560 "(Maybe you didn't strip carriage returns after a network transfer?)\n");
2562 case ' ': case '\t': case '\f': case 013:
2567 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2569 while (s < d && *s != '\n')
2574 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2576 PL_lex_state = LEX_FORMLINE;
2586 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2591 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
2594 if (strnEQ(s,"=>",2)) {
2595 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2596 OPERATOR('-'); /* unary minus */
2598 PL_last_uni = PL_oldbufptr;
2599 PL_last_lop_op = OP_FTEREAD; /* good enough */
2601 case 'r': FTST(OP_FTEREAD);
2602 case 'w': FTST(OP_FTEWRITE);
2603 case 'x': FTST(OP_FTEEXEC);
2604 case 'o': FTST(OP_FTEOWNED);
2605 case 'R': FTST(OP_FTRREAD);
2606 case 'W': FTST(OP_FTRWRITE);
2607 case 'X': FTST(OP_FTREXEC);
2608 case 'O': FTST(OP_FTROWNED);
2609 case 'e': FTST(OP_FTIS);
2610 case 'z': FTST(OP_FTZERO);
2611 case 's': FTST(OP_FTSIZE);
2612 case 'f': FTST(OP_FTFILE);
2613 case 'd': FTST(OP_FTDIR);
2614 case 'l': FTST(OP_FTLINK);
2615 case 'p': FTST(OP_FTPIPE);
2616 case 'S': FTST(OP_FTSOCK);
2617 case 'u': FTST(OP_FTSUID);
2618 case 'g': FTST(OP_FTSGID);
2619 case 'k': FTST(OP_FTSVTX);
2620 case 'b': FTST(OP_FTBLK);
2621 case 'c': FTST(OP_FTCHR);
2622 case 't': FTST(OP_FTTTY);
2623 case 'T': FTST(OP_FTTEXT);
2624 case 'B': FTST(OP_FTBINARY);
2625 case 'M': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTMTIME);
2626 case 'A': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTATIME);
2627 case 'C': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTCTIME);
2629 Perl_croak(aTHX_ "Unrecognized file test: -%c", (int)tmp);
2636 if (PL_expect == XOPERATOR)
2641 else if (*s == '>') {
2644 if (isIDFIRST_lazy(s)) {
2645 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2653 if (PL_expect == XOPERATOR)
2656 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2658 OPERATOR('-'); /* unary minus */
2665 if (PL_expect == XOPERATOR)
2670 if (PL_expect == XOPERATOR)
2673 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2679 if (PL_expect != XOPERATOR) {
2680 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2681 PL_expect = XOPERATOR;
2682 force_ident(PL_tokenbuf, '*');
2695 if (PL_expect == XOPERATOR) {
2699 PL_tokenbuf[0] = '%';
2700 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2701 if (!PL_tokenbuf[1]) {
2703 yyerror("Final % should be \\% or %name");
2706 PL_pending_ident = '%';
2725 switch (PL_expect) {
2728 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2730 PL_bufptr = s; /* update in case we back off */
2736 PL_expect = XTERMBLOCK;
2740 while (isIDFIRST_lazy(s)) {
2741 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2743 d = scan_str(d,TRUE,TRUE);
2746 SvREFCNT_dec(PL_lex_stuff);
2747 PL_lex_stuff = Nullsv;
2749 /* MUST advance bufptr here to avoid bogus
2750 "at end of line" context messages from yyerror().
2752 PL_bufptr = s + len;
2753 yyerror("Unterminated attribute parameter in attribute list");
2756 return 0; /* EOF indicator */
2760 SV *sv = newSVpvn(s, len);
2761 sv_catsv(sv, PL_lex_stuff);
2762 attrs = append_elem(OP_LIST, attrs,
2763 newSVOP(OP_CONST, 0, sv));
2764 SvREFCNT_dec(PL_lex_stuff);
2765 PL_lex_stuff = Nullsv;
2768 attrs = append_elem(OP_LIST, attrs,
2769 newSVOP(OP_CONST, 0,
2776 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}' for vi */
2777 if (*s != ';' && *s != tmp) {
2778 char q = ((*s == '\'') ? '"' : '\'');
2779 /* If here for an expression, and parsed no attrs, back off. */
2780 if (tmp == '=' && !attrs) {
2784 /* MUST advance bufptr here to avoid bogus "at end of line"
2785 context messages from yyerror().
2789 yyerror("Unterminated attribute list");
2791 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
2798 PL_nextval[PL_nexttoke].opval = attrs;
2806 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
2807 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
2812 if (PL_curcop->cop_line < PL_copline)
2813 PL_copline = PL_curcop->cop_line;
2824 if (PL_lex_brackets <= 0)
2825 yyerror("Unmatched right square bracket");
2828 if (PL_lex_state == LEX_INTERPNORMAL) {
2829 if (PL_lex_brackets == 0) {
2830 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
2831 PL_lex_state = LEX_INTERPEND;
2838 if (PL_lex_brackets > 100) {
2839 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
2840 if (newlb != PL_lex_brackstack) {
2842 PL_lex_brackstack = newlb;
2845 switch (PL_expect) {
2847 if (PL_lex_formbrack) {
2851 if (PL_oldoldbufptr == PL_last_lop)
2852 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
2854 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2855 OPERATOR(HASHBRACK);
2857 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
2860 PL_tokenbuf[0] = '\0';
2861 if (d < PL_bufend && *d == '-') {
2862 PL_tokenbuf[0] = '-';
2864 while (d < PL_bufend && (*d == ' ' || *d == '\t'))
2867 if (d < PL_bufend && isIDFIRST_lazy(d)) {
2868 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
2870 while (d < PL_bufend && (*d == ' ' || *d == '\t'))
2873 char minus = (PL_tokenbuf[0] == '-');
2874 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
2882 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
2887 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2892 if (PL_oldoldbufptr == PL_last_lop)
2893 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
2895 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2898 OPERATOR(HASHBRACK);
2899 /* This hack serves to disambiguate a pair of curlies
2900 * as being a block or an anon hash. Normally, expectation
2901 * determines that, but in cases where we're not in a
2902 * position to expect anything in particular (like inside
2903 * eval"") we have to resolve the ambiguity. This code
2904 * covers the case where the first term in the curlies is a
2905 * quoted string. Most other cases need to be explicitly
2906 * disambiguated by prepending a `+' before the opening
2907 * curly in order to force resolution as an anon hash.
2909 * XXX should probably propagate the outer expectation
2910 * into eval"" to rely less on this hack, but that could
2911 * potentially break current behavior of eval"".
2915 if (*s == '\'' || *s == '"' || *s == '`') {
2916 /* common case: get past first string, handling escapes */
2917 for (t++; t < PL_bufend && *t != *s;)
2918 if (*t++ == '\\' && (*t == '\\' || *t == *s))
2922 else if (*s == 'q') {
2925 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
2926 && !isALNUM(*t)))) {
2928 char open, close, term;
2931 while (t < PL_bufend && isSPACE(*t))
2935 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
2939 for (t++; t < PL_bufend; t++) {
2940 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
2942 else if (*t == open)
2946 for (t++; t < PL_bufend; t++) {
2947 if (*t == '\\' && t+1 < PL_bufend)
2949 else if (*t == close && --brackets <= 0)
2951 else if (*t == open)
2957 else if (isIDFIRST_lazy(s)) {
2958 for (t++; t < PL_bufend && isALNUM_lazy(t); t++) ;
2960 while (t < PL_bufend && isSPACE(*t))
2962 /* if comma follows first term, call it an anon hash */
2963 /* XXX it could be a comma expression with loop modifiers */
2964 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
2965 || (*t == '=' && t[1] == '>')))
2966 OPERATOR(HASHBRACK);
2967 if (PL_expect == XREF)
2970 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
2976 yylval.ival = PL_curcop->cop_line;
2977 if (isSPACE(*s) || *s == '#')
2978 PL_copline = NOLINE; /* invalidate current command line number */
2983 if (PL_lex_brackets <= 0)
2984 yyerror("Unmatched right curly bracket");
2986 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
2987 if (PL_lex_brackets < PL_lex_formbrack)
2988 PL_lex_formbrack = 0;
2989 if (PL_lex_state == LEX_INTERPNORMAL) {
2990 if (PL_lex_brackets == 0) {
2991 if (PL_lex_fakebrack) {
2992 PL_lex_state = LEX_INTERPEND;
2994 return yylex(); /* ignore fake brackets */
2996 if (*s == '-' && s[1] == '>')
2997 PL_lex_state = LEX_INTERPENDMAYBE;
2998 else if (*s != '[' && *s != '{')
2999 PL_lex_state = LEX_INTERPEND;
3002 if (PL_lex_brackets < PL_lex_fakebrack) {
3004 PL_lex_fakebrack = 0;
3005 return yylex(); /* ignore fake brackets */
3015 if (PL_expect == XOPERATOR) {
3016 if (ckWARN(WARN_SEMICOLON) && isIDFIRST_lazy(s) && PL_bufptr == PL_linestart) {
3017 PL_curcop->cop_line--;
3018 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3019 PL_curcop->cop_line++;
3024 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3026 PL_expect = XOPERATOR;
3027 force_ident(PL_tokenbuf, '&');
3031 yylval.ival = (OPpENTERSUB_AMPER<<8);
3050 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3051 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3053 if (PL_expect == XSTATE && isALPHA(tmp) &&
3054 (s == PL_linestart+1 || s[-2] == '\n') )
3056 if (PL_in_eval && !PL_rsfp) {
3061 if (strnEQ(s,"=cut",4)) {
3075 PL_doextract = TRUE;
3078 if (PL_lex_brackets < PL_lex_formbrack) {
3080 #ifdef PERL_STRICT_CR
3081 for (t = s; *t == ' ' || *t == '\t'; t++) ;
3083 for (t = s; *t == ' ' || *t == '\t' || *t == '\r'; t++) ;
3085 if (*t == '\n' || *t == '#') {
3103 if (PL_expect != XOPERATOR) {
3104 if (s[1] != '<' && !strchr(s,'>'))
3107 s = scan_heredoc(s);
3109 s = scan_inputsymbol(s);
3110 TERM(sublex_start());
3115 SHop(OP_LEFT_SHIFT);
3129 SHop(OP_RIGHT_SHIFT);
3138 if (PL_expect == XOPERATOR) {
3139 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3142 return ','; /* grandfather non-comma-format format */
3146 if (s[1] == '#' && (isIDFIRST_lazy(s+2) || strchr("{$:+-", s[2]))) {
3147 PL_tokenbuf[0] = '@';
3148 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3149 sizeof PL_tokenbuf - 1, FALSE);
3150 if (PL_expect == XOPERATOR)
3151 no_op("Array length", s);
3152 if (!PL_tokenbuf[1])
3154 PL_expect = XOPERATOR;
3155 PL_pending_ident = '#';
3159 PL_tokenbuf[0] = '$';
3160 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3161 sizeof PL_tokenbuf - 1, FALSE);
3162 if (PL_expect == XOPERATOR)
3164 if (!PL_tokenbuf[1]) {
3166 yyerror("Final $ should be \\$ or $name");
3170 /* This kludge not intended to be bulletproof. */
3171 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3172 yylval.opval = newSVOP(OP_CONST, 0,
3173 newSViv((IV)PL_compiling.cop_arybase));
3174 yylval.opval->op_private = OPpCONST_ARYBASE;
3180 if (PL_lex_state == LEX_NORMAL)
3183 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3186 PL_tokenbuf[0] = '@';
3187 if (ckWARN(WARN_SYNTAX)) {
3189 isSPACE(*t) || isALNUM_lazy(t) || *t == '$';
3192 PL_bufptr = skipspace(PL_bufptr);
3193 while (t < PL_bufend && *t != ']')
3195 Perl_warner(aTHX_ WARN_SYNTAX,
3196 "Multidimensional syntax %.*s not supported",
3197 (t - PL_bufptr) + 1, PL_bufptr);
3201 else if (*s == '{') {
3202 PL_tokenbuf[0] = '%';
3203 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3204 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3206 char tmpbuf[sizeof PL_tokenbuf];
3208 for (t++; isSPACE(*t); t++) ;
3209 if (isIDFIRST_lazy(t)) {
3210 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3211 for (; isSPACE(*t); t++) ;
3212 if (*t == ';' && get_cv(tmpbuf, FALSE))
3213 Perl_warner(aTHX_ WARN_SYNTAX,
3214 "You need to quote \"%s\"", tmpbuf);
3220 PL_expect = XOPERATOR;
3221 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3222 bool islop = (PL_last_lop == PL_oldoldbufptr);
3223 if (!islop || PL_last_lop_op == OP_GREPSTART)
3224 PL_expect = XOPERATOR;
3225 else if (strchr("$@\"'`q", *s))
3226 PL_expect = XTERM; /* e.g. print $fh "foo" */
3227 else if (strchr("&*<%", *s) && isIDFIRST_lazy(s+1))
3228 PL_expect = XTERM; /* e.g. print $fh &sub */
3229 else if (isIDFIRST_lazy(s)) {
3230 char tmpbuf[sizeof PL_tokenbuf];
3231 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3232 if (tmp = keyword(tmpbuf, len)) {
3233 /* binary operators exclude handle interpretations */
3245 PL_expect = XTERM; /* e.g. print $fh length() */
3250 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3251 if (gv && GvCVu(gv))
3252 PL_expect = XTERM; /* e.g. print $fh subr() */
3255 else if (isDIGIT(*s))
3256 PL_expect = XTERM; /* e.g. print $fh 3 */
3257 else if (*s == '.' && isDIGIT(s[1]))
3258 PL_expect = XTERM; /* e.g. print $fh .3 */
3259 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3260 PL_expect = XTERM; /* e.g. print $fh -1 */
3261 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3262 PL_expect = XTERM; /* print $fh <<"EOF" */
3264 PL_pending_ident = '$';
3268 if (PL_expect == XOPERATOR)
3270 PL_tokenbuf[0] = '@';
3271 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3272 if (!PL_tokenbuf[1]) {
3274 yyerror("Final @ should be \\@ or @name");
3277 if (PL_lex_state == LEX_NORMAL)
3279 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3281 PL_tokenbuf[0] = '%';
3283 /* Warn about @ where they meant $. */
3284 if (ckWARN(WARN_SYNTAX)) {
3285 if (*s == '[' || *s == '{') {
3287 while (*t && (isALNUM_lazy(t) || strchr(" \t$#+-'\"", *t)))
3289 if (*t == '}' || *t == ']') {
3291 PL_bufptr = skipspace(PL_bufptr);
3292 Perl_warner(aTHX_ WARN_SYNTAX,
3293 "Scalar value %.*s better written as $%.*s",
3294 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3299 PL_pending_ident = '@';
3302 case '/': /* may either be division or pattern */
3303 case '?': /* may either be conditional or pattern */
3304 if (PL_expect != XOPERATOR) {
3305 /* Disable warning on "study /blah/" */
3306 if (PL_oldoldbufptr == PL_last_uni
3307 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3308 || memNE(PL_last_uni, "study", 5) || isALNUM_lazy(PL_last_uni+5)))
3310 s = scan_pat(s,OP_MATCH);
3311 TERM(sublex_start());
3319 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3320 #ifdef PERL_STRICT_CR
3323 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3325 && (s == PL_linestart || s[-1] == '\n') )
3327 PL_lex_formbrack = 0;
3331 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3337 yylval.ival = OPf_SPECIAL;
3343 if (PL_expect != XOPERATOR)
3348 case '0': case '1': case '2': case '3': case '4':
3349 case '5': case '6': case '7': case '8': case '9':
3351 if (PL_expect == XOPERATOR)
3356 s = scan_str(s,FALSE,FALSE);
3357 if (PL_expect == XOPERATOR) {
3358 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3361 return ','; /* grandfather non-comma-format format */
3367 missingterm((char*)0);
3368 yylval.ival = OP_CONST;
3369 TERM(sublex_start());
3372 s = scan_str(s,FALSE,FALSE);
3373 if (PL_expect == XOPERATOR) {
3374 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3377 return ','; /* grandfather non-comma-format format */
3383 missingterm((char*)0);
3384 yylval.ival = OP_CONST;
3385 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3386 if (*d == '$' || *d == '@' || *d == '\\' || *d & 0x80) {
3387 yylval.ival = OP_STRINGIFY;
3391 TERM(sublex_start());
3394 s = scan_str(s,FALSE,FALSE);
3395 if (PL_expect == XOPERATOR)
3396 no_op("Backticks",s);
3398 missingterm((char*)0);
3399 yylval.ival = OP_BACKTICK;
3401 TERM(sublex_start());
3405 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3406 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3408 if (PL_expect == XOPERATOR)
3409 no_op("Backslash",s);
3413 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3453 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3455 /* Some keywords can be followed by any delimiter, including ':' */
3456 tmp = (len == 1 && strchr("msyq", PL_tokenbuf[0]) ||
3457 len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3458 (PL_tokenbuf[0] == 'q' &&
3459 strchr("qwxr", PL_tokenbuf[1]))));
3461 /* x::* is just a word, unless x is "CORE" */
3462 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3466 while (d < PL_bufend && isSPACE(*d))
3467 d++; /* no comments skipped here, or s### is misparsed */
3469 /* Is this a label? */
3470 if (!tmp && PL_expect == XSTATE
3471 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3473 yylval.pval = savepv(PL_tokenbuf);
3478 /* Check for keywords */
3479 tmp = keyword(PL_tokenbuf, len);
3481 /* Is this a word before a => operator? */
3482 if (strnEQ(d,"=>",2)) {
3484 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3485 yylval.opval->op_private = OPpCONST_BARE;
3489 if (tmp < 0) { /* second-class keyword? */
3490 GV *ogv = Nullgv; /* override (winner) */
3491 GV *hgv = Nullgv; /* hidden (loser) */
3492 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3494 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3497 if (GvIMPORTED_CV(gv))
3499 else if (! CvMETHOD(cv))
3503 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3504 (gv = *gvp) != (GV*)&PL_sv_undef &&
3505 GvCVu(gv) && GvIMPORTED_CV(gv))
3511 tmp = 0; /* overridden by import or by GLOBAL */
3514 && -tmp==KEY_lock /* XXX generalizable kludge */
3515 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3517 tmp = 0; /* any sub overrides "weak" keyword */
3519 else { /* no override */
3523 if (ckWARN(WARN_AMBIGUOUS) && hgv
3524 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3525 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3526 "Ambiguous call resolved as CORE::%s(), %s",
3527 GvENAME(hgv), "qualify as such or use &");
3534 default: /* not a keyword */
3537 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3539 /* Get the rest if it looks like a package qualifier */
3541 if (*s == '\'' || *s == ':' && s[1] == ':') {
3543 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3546 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3547 *s == '\'' ? "'" : "::");
3551 if (PL_expect == XOPERATOR) {
3552 if (PL_bufptr == PL_linestart) {
3553 PL_curcop->cop_line--;
3554 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3555 PL_curcop->cop_line++;
3558 no_op("Bareword",s);
3561 /* Look for a subroutine with this name in current package,
3562 unless name is "Foo::", in which case Foo is a bearword
3563 (and a package name). */
3566 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3568 if (ckWARN(WARN_UNSAFE) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3569 Perl_warner(aTHX_ WARN_UNSAFE,
3570 "Bareword \"%s\" refers to nonexistent package",
3573 PL_tokenbuf[len] = '\0';
3580 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3583 /* if we saw a global override before, get the right name */
3586 sv = newSVpvn("CORE::GLOBAL::",14);
3587 sv_catpv(sv,PL_tokenbuf);
3590 sv = newSVpv(PL_tokenbuf,0);
3592 /* Presume this is going to be a bareword of some sort. */
3595 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3596 yylval.opval->op_private = OPpCONST_BARE;
3598 /* And if "Foo::", then that's what it certainly is. */
3603 /* See if it's the indirect object for a list operator. */
3605 if (PL_oldoldbufptr &&
3606 PL_oldoldbufptr < PL_bufptr &&
3607 (PL_oldoldbufptr == PL_last_lop
3608 || PL_oldoldbufptr == PL_last_uni) &&
3609 /* NO SKIPSPACE BEFORE HERE! */
3610 (PL_expect == XREF ||
3611 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3613 bool immediate_paren = *s == '(';
3615 /* (Now we can afford to cross potential line boundary.) */
3618 /* Two barewords in a row may indicate method call. */
3620 if ((isIDFIRST_lazy(s) || *s == '$') && (tmp=intuit_method(s,gv)))
3623 /* If not a declared subroutine, it's an indirect object. */
3624 /* (But it's an indir obj regardless for sort.) */
3626 if ((PL_last_lop_op == OP_SORT ||
3627 (!immediate_paren && (!gv || !GvCVu(gv)))) &&
3628 (PL_last_lop_op != OP_MAPSTART &&
3629 PL_last_lop_op != OP_GREPSTART))
3631 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3636 /* If followed by a paren, it's certainly a subroutine. */
3638 PL_expect = XOPERATOR;
3642 if (gv && GvCVu(gv)) {
3643 for (d = s + 1; *d == ' ' || *d == '\t'; d++) ;
3644 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3649 PL_nextval[PL_nexttoke].opval = yylval.opval;
3650 PL_expect = XOPERATOR;
3656 /* If followed by var or block, call it a method (unless sub) */
3658 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3659 PL_last_lop = PL_oldbufptr;
3660 PL_last_lop_op = OP_METHOD;
3664 /* If followed by a bareword, see if it looks like indir obj. */
3666 if ((isIDFIRST_lazy(s) || *s == '$') && (tmp = intuit_method(s,gv)))
3669 /* Not a method, so call it a subroutine (if defined) */
3671 if (gv && GvCVu(gv)) {
3673 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
3674 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3675 "Ambiguous use of -%s resolved as -&%s()",
3676 PL_tokenbuf, PL_tokenbuf);
3677 /* Check for a constant sub */
3679 if ((sv = cv_const_sv(cv))) {
3681 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
3682 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
3683 yylval.opval->op_private = 0;
3687 /* Resolve to GV now. */
3688 op_free(yylval.opval);
3689 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
3690 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
3691 PL_last_lop = PL_oldbufptr;
3692 PL_last_lop_op = OP_ENTERSUB;
3693 /* Is there a prototype? */
3696 char *proto = SvPV((SV*)cv, len);
3699 if (strEQ(proto, "$"))
3701 if (*proto == '&' && *s == '{') {
3702 sv_setpv(PL_subname,"__ANON__");
3706 PL_nextval[PL_nexttoke].opval = yylval.opval;
3712 /* Call it a bare word */
3714 if (PL_hints & HINT_STRICT_SUBS)
3715 yylval.opval->op_private |= OPpCONST_STRICT;
3718 if (ckWARN(WARN_RESERVED)) {
3719 if (lastchar != '-') {
3720 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
3722 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
3729 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
3730 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3731 "Operator or semicolon missing before %c%s",
3732 lastchar, PL_tokenbuf);
3733 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3734 "Ambiguous use of %c resolved as operator %c",
3735 lastchar, lastchar);
3741 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3742 newSVsv(GvSV(PL_curcop->cop_filegv)));
3746 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3747 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)PL_curcop->cop_line));
3750 case KEY___PACKAGE__:
3751 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3753 ? newSVsv(PL_curstname)
3762 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
3763 char *pname = "main";
3764 if (PL_tokenbuf[2] == 'D')
3765 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
3766 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
3769 GvIOp(gv) = newIO();
3770 IoIFP(GvIOp(gv)) = PL_rsfp;
3771 #if defined(HAS_FCNTL) && defined(F_SETFD)
3773 int fd = PerlIO_fileno(PL_rsfp);
3774 fcntl(fd,F_SETFD,fd >= 3);
3777 /* Mark this internal pseudo-handle as clean */
3778 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
3780 IoTYPE(GvIOp(gv)) = '|';
3781 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
3782 IoTYPE(GvIOp(gv)) = '-';
3784 IoTYPE(GvIOp(gv)) = '<';
3795 if (PL_expect == XSTATE) {
3802 if (*s == ':' && s[1] == ':') {
3805 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3806 tmp = keyword(PL_tokenbuf, len);
3820 LOP(OP_ACCEPT,XTERM);
3826 LOP(OP_ATAN2,XTERM);
3835 LOP(OP_BLESS,XTERM);
3844 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
3861 if (!PL_cryptseen++)
3864 LOP(OP_CRYPT,XTERM);
3867 if (ckWARN(WARN_OCTAL)) {
3868 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
3869 if (*d != '0' && isDIGIT(*d))
3870 Perl_warner(aTHX_ WARN_OCTAL,
3871 "chmod: mode argument is missing initial 0");
3873 LOP(OP_CHMOD,XTERM);
3876 LOP(OP_CHOWN,XTERM);
3879 LOP(OP_CONNECT,XTERM);
3895 s = force_word(s,WORD,FALSE,TRUE,FALSE);
3899 PL_hints |= HINT_BLOCK_SCOPE;
3909 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
3910 LOP(OP_DBMOPEN,XTERM);
3916 s = force_word(s,WORD,TRUE,FALSE,FALSE);
3923 yylval.ival = PL_curcop->cop_line;
3937 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
3938 UNIBRACK(OP_ENTEREVAL);
3953 case KEY_endhostent:
3959 case KEY_endservent:
3962 case KEY_endprotoent:
3973 yylval.ival = PL_curcop->cop_line;
3975 if (PL_expect == XSTATE && isIDFIRST_lazy(s)) {
3977 if ((PL_bufend - p) >= 3 &&
3978 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
3980 else if ((PL_bufend - p) >= 4 &&
3981 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
3984 if (isIDFIRST_lazy(p)) {
3985 p = scan_ident(p, PL_bufend,
3986 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3990 Perl_croak(aTHX_ "Missing $ on loop variable");
3995 LOP(OP_FORMLINE,XTERM);
4001 LOP(OP_FCNTL,XTERM);
4007 LOP(OP_FLOCK,XTERM);
4016 LOP(OP_GREPSTART, *s == '(' ? XTERM : XREF);
4019 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4034 case KEY_getpriority:
4035 LOP(OP_GETPRIORITY,XTERM);
4037 case KEY_getprotobyname:
4040 case KEY_getprotobynumber:
4041 LOP(OP_GPBYNUMBER,XTERM);
4043 case KEY_getprotoent:
4055 case KEY_getpeername:
4056 UNI(OP_GETPEERNAME);
4058 case KEY_gethostbyname:
4061 case KEY_gethostbyaddr:
4062 LOP(OP_GHBYADDR,XTERM);
4064 case KEY_gethostent:
4067 case KEY_getnetbyname:
4070 case KEY_getnetbyaddr:
4071 LOP(OP_GNBYADDR,XTERM);
4076 case KEY_getservbyname:
4077 LOP(OP_GSBYNAME,XTERM);
4079 case KEY_getservbyport:
4080 LOP(OP_GSBYPORT,XTERM);
4082 case KEY_getservent:
4085 case KEY_getsockname:
4086 UNI(OP_GETSOCKNAME);
4088 case KEY_getsockopt:
4089 LOP(OP_GSOCKOPT,XTERM);
4111 yylval.ival = PL_curcop->cop_line;
4115 LOP(OP_INDEX,XTERM);
4121 LOP(OP_IOCTL,XTERM);
4133 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4165 LOP(OP_LISTEN,XTERM);
4174 s = scan_pat(s,OP_MATCH);
4175 TERM(sublex_start());
4178 LOP(OP_MAPSTART, *s == '(' ? XTERM : XREF);
4181 LOP(OP_MKDIR,XTERM);
4184 LOP(OP_MSGCTL,XTERM);
4187 LOP(OP_MSGGET,XTERM);
4190 LOP(OP_MSGRCV,XTERM);
4193 LOP(OP_MSGSND,XTERM);
4199 if (isIDFIRST_lazy(s)) {
4200 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4201 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4203 PL_in_my_stash = gv_stashpv(PL_tokenbuf, FALSE);
4204 if (!PL_in_my_stash) {
4207 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4215 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4222 if (PL_expect != XSTATE)
4223 yyerror("\"no\" not allowed in expression");
4224 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4225 s = force_version(s);
4234 if (isIDFIRST_lazy(s)) {
4236 for (d = s; isALNUM_lazy(d); d++) ;
4238 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_AMBIGUOUS))
4239 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4240 "Precedence problem: open %.*s should be open(%.*s)",
4246 yylval.ival = OP_OR;
4256 LOP(OP_OPEN_DIR,XTERM);
4259 checkcomma(s,PL_tokenbuf,"filehandle");
4263 checkcomma(s,PL_tokenbuf,"filehandle");
4282 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4286 LOP(OP_PIPE_OP,XTERM);
4289 s = scan_str(s,FALSE,FALSE);
4291 missingterm((char*)0);
4292 yylval.ival = OP_CONST;
4293 TERM(sublex_start());
4299 s = scan_str(s,FALSE,FALSE);
4301 missingterm((char*)0);
4303 if (SvCUR(PL_lex_stuff)) {
4306 d = SvPV_force(PL_lex_stuff, len);
4308 for (; isSPACE(*d) && len; --len, ++d) ;
4311 if (!warned && ckWARN(WARN_SYNTAX)) {
4312 for (; !isSPACE(*d) && len; --len, ++d) {
4314 Perl_warner(aTHX_ WARN_SYNTAX,
4315 "Possible attempt to separate words with commas");
4318 else if (*d == '#') {
4319 Perl_warner(aTHX_ WARN_SYNTAX,
4320 "Possible attempt to put comments in qw() list");
4326 for (; !isSPACE(*d) && len; --len, ++d) ;
4328 words = append_elem(OP_LIST, words,
4329 newSVOP(OP_CONST, 0, newSVpvn(b, d-b)));
4333 PL_nextval[PL_nexttoke].opval = words;
4338 SvREFCNT_dec(PL_lex_stuff);
4339 PL_lex_stuff = Nullsv;
4344 s = scan_str(s,FALSE,FALSE);
4346 missingterm((char*)0);
4347 yylval.ival = OP_STRINGIFY;
4348 if (SvIVX(PL_lex_stuff) == '\'')
4349 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4350 TERM(sublex_start());
4353 s = scan_pat(s,OP_QR);
4354 TERM(sublex_start());
4357 s = scan_str(s,FALSE,FALSE);
4359 missingterm((char*)0);
4360 yylval.ival = OP_BACKTICK;
4362 TERM(sublex_start());
4368 *PL_tokenbuf = '\0';
4369 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4370 if (isIDFIRST_lazy(PL_tokenbuf))
4371 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4373 yyerror("<> should be quotes");
4380 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4384 LOP(OP_RENAME,XTERM);
4393 LOP(OP_RINDEX,XTERM);
4416 LOP(OP_REVERSE,XTERM);
4427 TERM(sublex_start());
4429 TOKEN(1); /* force error */
4438 LOP(OP_SELECT,XTERM);
4444 LOP(OP_SEMCTL,XTERM);
4447 LOP(OP_SEMGET,XTERM);
4450 LOP(OP_SEMOP,XTERM);
4456 LOP(OP_SETPGRP,XTERM);
4458 case KEY_setpriority:
4459 LOP(OP_SETPRIORITY,XTERM);
4461 case KEY_sethostent:
4467 case KEY_setservent:
4470 case KEY_setprotoent:
4480 LOP(OP_SEEKDIR,XTERM);
4482 case KEY_setsockopt:
4483 LOP(OP_SSOCKOPT,XTERM);
4489 LOP(OP_SHMCTL,XTERM);
4492 LOP(OP_SHMGET,XTERM);
4495 LOP(OP_SHMREAD,XTERM);
4498 LOP(OP_SHMWRITE,XTERM);
4501 LOP(OP_SHUTDOWN,XTERM);
4510 LOP(OP_SOCKET,XTERM);
4512 case KEY_socketpair:
4513 LOP(OP_SOCKPAIR,XTERM);
4516 checkcomma(s,PL_tokenbuf,"subroutine name");
4518 if (*s == ';' || *s == ')') /* probably a close */
4519 Perl_croak(aTHX_ "sort is now a reserved word");
4521 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4525 LOP(OP_SPLIT,XTERM);
4528 LOP(OP_SPRINTF,XTERM);
4531 LOP(OP_SPLICE,XTERM);
4547 LOP(OP_SUBSTR,XTERM);
4553 char tmpbuf[sizeof PL_tokenbuf];
4555 expectation attrful;
4556 bool have_name, have_proto;
4561 if (isIDFIRST_lazy(s) || *s == '\'' ||
4562 (*s == ':' && s[1] == ':'))
4565 attrful = XATTRBLOCK;
4566 /* remember buffer pos'n for later force_word */
4567 tboffset = s - PL_oldbufptr;
4568 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4569 if (strchr(tmpbuf, ':'))
4570 sv_setpv(PL_subname, tmpbuf);
4572 sv_setsv(PL_subname,PL_curstname);
4573 sv_catpvn(PL_subname,"::",2);
4574 sv_catpvn(PL_subname,tmpbuf,len);
4581 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4582 PL_expect = XTERMBLOCK;
4583 attrful = XATTRTERM;
4584 sv_setpv(PL_subname,"?");
4588 if (key == KEY_format) {
4590 PL_lex_formbrack = PL_lex_brackets + 1;
4592 (void) force_word(PL_oldbufptr + tboffset, WORD,
4597 /* Look for a prototype */
4601 s = scan_str(s,FALSE,FALSE);
4604 SvREFCNT_dec(PL_lex_stuff);
4605 PL_lex_stuff = Nullsv;
4606 Perl_croak(aTHX_ "Prototype not terminated");
4609 d = SvPVX(PL_lex_stuff);
4611 for (p = d; *p; ++p) {
4616 SvCUR(PL_lex_stuff) = tmp;
4624 if (*s == ':' && s[1] != ':')
4625 PL_expect = attrful;
4628 PL_nextval[PL_nexttoke].opval =
4629 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
4630 PL_lex_stuff = Nullsv;
4634 sv_setpv(PL_subname,"__ANON__");
4637 (void) force_word(PL_oldbufptr + tboffset, WORD,
4646 LOP(OP_SYSTEM,XREF);
4649 LOP(OP_SYMLINK,XTERM);
4652 LOP(OP_SYSCALL,XTERM);
4655 LOP(OP_SYSOPEN,XTERM);
4658 LOP(OP_SYSSEEK,XTERM);
4661 LOP(OP_SYSREAD,XTERM);
4664 LOP(OP_SYSWRITE,XTERM);
4668 TERM(sublex_start());
4689 LOP(OP_TRUNCATE,XTERM);
4701 yylval.ival = PL_curcop->cop_line;
4705 yylval.ival = PL_curcop->cop_line;
4709 LOP(OP_UNLINK,XTERM);
4715 LOP(OP_UNPACK,XTERM);
4718 LOP(OP_UTIME,XTERM);
4721 if (ckWARN(WARN_OCTAL)) {
4722 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4723 if (*d != '0' && isDIGIT(*d))
4724 Perl_warner(aTHX_ WARN_OCTAL,
4725 "umask: argument is missing initial 0");
4730 LOP(OP_UNSHIFT,XTERM);
4733 if (PL_expect != XSTATE)
4734 yyerror("\"use\" not allowed in expression");
4737 s = force_version(s);
4738 if(*s == ';' || (s = skipspace(s), *s == ';')) {
4739 PL_nextval[PL_nexttoke].opval = Nullop;
4744 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4745 s = force_version(s);
4758 yylval.ival = PL_curcop->cop_line;
4762 PL_hints |= HINT_BLOCK_SCOPE;
4769 LOP(OP_WAITPID,XTERM);
4777 static char ctl_l[2];
4779 if (ctl_l[0] == '\0')
4780 ctl_l[0] = toCTRL('L');
4781 gv_fetchpv(ctl_l,TRUE, SVt_PV);
4784 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
4789 if (PL_expect == XOPERATOR)
4795 yylval.ival = OP_XOR;
4800 TERM(sublex_start());
4806 Perl_keyword(pTHX_ register char *d, I32 len)
4811 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
4812 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
4813 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
4814 if (strEQ(d,"__DATA__")) return KEY___DATA__;
4815 if (strEQ(d,"__END__")) return KEY___END__;
4819 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
4824 if (strEQ(d,"and")) return -KEY_and;
4825 if (strEQ(d,"abs")) return -KEY_abs;
4828 if (strEQ(d,"alarm")) return -KEY_alarm;
4829 if (strEQ(d,"atan2")) return -KEY_atan2;
4832 if (strEQ(d,"accept")) return -KEY_accept;
4837 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
4840 if (strEQ(d,"bless")) return -KEY_bless;
4841 if (strEQ(d,"bind")) return -KEY_bind;
4842 if (strEQ(d,"binmode")) return -KEY_binmode;
4845 if (strEQ(d,"CORE")) return -KEY_CORE;
4850 if (strEQ(d,"cmp")) return -KEY_cmp;
4851 if (strEQ(d,"chr")) return -KEY_chr;
4852 if (strEQ(d,"cos")) return -KEY_cos;
4855 if (strEQ(d,"chop")) return KEY_chop;
4858 if (strEQ(d,"close")) return -KEY_close;
4859 if (strEQ(d,"chdir")) return -KEY_chdir;
4860 if (strEQ(d,"chomp")) return KEY_chomp;
4861 if (strEQ(d,"chmod")) return -KEY_chmod;
4862 if (strEQ(d,"chown")) return -KEY_chown;
4863 if (strEQ(d,"crypt")) return -KEY_crypt;
4866 if (strEQ(d,"chroot")) return -KEY_chroot;
4867 if (strEQ(d,"caller")) return -KEY_caller;
4870 if (strEQ(d,"connect")) return -KEY_connect;
4873 if (strEQ(d,"closedir")) return -KEY_closedir;
4874 if (strEQ(d,"continue")) return -KEY_continue;
4879 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
4884 if (strEQ(d,"do")) return KEY_do;
4887 if (strEQ(d,"die")) return -KEY_die;
4890 if (strEQ(d,"dump")) return -KEY_dump;
4893 if (strEQ(d,"delete")) return KEY_delete;
4896 if (strEQ(d,"defined")) return KEY_defined;
4897 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
4900 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
4905 if (strEQ(d,"EQ")) { deprecate(d); return -KEY_eq;}
4906 if (strEQ(d,"END")) return KEY_END;
4911 if (strEQ(d,"eq")) return -KEY_eq;
4914 if (strEQ(d,"eof")) return -KEY_eof;
4915 if (strEQ(d,"exp")) return -KEY_exp;
4918 if (strEQ(d,"else")) return KEY_else;
4919 if (strEQ(d,"exit")) return -KEY_exit;
4920 if (strEQ(d,"eval")) return KEY_eval;
4921 if (strEQ(d,"exec")) return -KEY_exec;
4922 if (strEQ(d,"each")) return KEY_each;
4925 if (strEQ(d,"elsif")) return KEY_elsif;
4928 if (strEQ(d,"exists")) return KEY_exists;
4929 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
4932 if (strEQ(d,"endgrent")) return -KEY_endgrent;
4933 if (strEQ(d,"endpwent")) return -KEY_endpwent;
4936 if (strEQ(d,"endnetent")) return -KEY_endnetent;
4939 if (strEQ(d,"endhostent")) return -KEY_endhostent;
4940 if (strEQ(d,"endservent")) return -KEY_endservent;
4943 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
4950 if (strEQ(d,"for")) return KEY_for;
4953 if (strEQ(d,"fork")) return -KEY_fork;
4956 if (strEQ(d,"fcntl")) return -KEY_fcntl;
4957 if (strEQ(d,"flock")) return -KEY_flock;
4960 if (strEQ(d,"format")) return KEY_format;
4961 if (strEQ(d,"fileno")) return -KEY_fileno;
4964 if (strEQ(d,"foreach")) return KEY_foreach;
4967 if (strEQ(d,"formline")) return -KEY_formline;
4973 if (strEQ(d,"GT")) { deprecate(d); return -KEY_gt;}
4974 if (strEQ(d,"GE")) { deprecate(d); return -KEY_ge;}
4978 if (strnEQ(d,"get",3)) {
4983 if (strEQ(d,"ppid")) return -KEY_getppid;
4984 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
4987 if (strEQ(d,"pwent")) return -KEY_getpwent;
4988 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
4989 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
4992 if (strEQ(d,"peername")) return -KEY_getpeername;
4993 if (strEQ(d,"protoent")) return -KEY_getprotoent;
4994 if (strEQ(d,"priority")) return -KEY_getpriority;
4997 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5000 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5004 else if (*d == 'h') {
5005 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5006 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5007 if (strEQ(d,"hostent")) return -KEY_gethostent;
5009 else if (*d == 'n') {
5010 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5011 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5012 if (strEQ(d,"netent")) return -KEY_getnetent;
5014 else if (*d == 's') {
5015 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5016 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5017 if (strEQ(d,"servent")) return -KEY_getservent;
5018 if (strEQ(d,"sockname")) return -KEY_getsockname;
5019 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5021 else if (*d == 'g') {
5022 if (strEQ(d,"grent")) return -KEY_getgrent;
5023 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5024 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5026 else if (*d == 'l') {
5027 if (strEQ(d,"login")) return -KEY_getlogin;
5029 else if (strEQ(d,"c")) return -KEY_getc;
5034 if (strEQ(d,"gt")) return -KEY_gt;
5035 if (strEQ(d,"ge")) return -KEY_ge;
5038 if (strEQ(d,"grep")) return KEY_grep;
5039 if (strEQ(d,"goto")) return KEY_goto;
5040 if (strEQ(d,"glob")) return KEY_glob;
5043 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5048 if (strEQ(d,"hex")) return -KEY_hex;
5051 if (strEQ(d,"INIT")) return KEY_INIT;
5056 if (strEQ(d,"if")) return KEY_if;
5059 if (strEQ(d,"int")) return -KEY_int;
5062 if (strEQ(d,"index")) return -KEY_index;
5063 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5068 if (strEQ(d,"join")) return -KEY_join;
5072 if (strEQ(d,"keys")) return KEY_keys;
5073 if (strEQ(d,"kill")) return -KEY_kill;
5078 if (strEQ(d,"LT")) { deprecate(d); return -KEY_lt;}
5079 if (strEQ(d,"LE")) { deprecate(d); return -KEY_le;}
5085 if (strEQ(d,"lt")) return -KEY_lt;
5086 if (strEQ(d,"le")) return -KEY_le;
5087 if (strEQ(d,"lc")) return -KEY_lc;
5090 if (strEQ(d,"log")) return -KEY_log;
5093 if (strEQ(d,"last")) return KEY_last;
5094 if (strEQ(d,"link")) return -KEY_link;
5095 if (strEQ(d,"lock")) return -KEY_lock;
5098 if (strEQ(d,"local")) return KEY_local;
5099 if (strEQ(d,"lstat")) return -KEY_lstat;
5102 if (strEQ(d,"length")) return -KEY_length;
5103 if (strEQ(d,"listen")) return -KEY_listen;
5106 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5109 if (strEQ(d,"localtime")) return -KEY_localtime;
5115 case 1: return KEY_m;
5117 if (strEQ(d,"my")) return KEY_my;
5120 if (strEQ(d,"map")) return KEY_map;
5123 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5126 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5127 if (strEQ(d,"msgget")) return -KEY_msgget;
5128 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5129 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5134 if (strEQ(d,"NE")) { deprecate(d); return -KEY_ne;}
5137 if (strEQ(d,"next")) return KEY_next;
5138 if (strEQ(d,"ne")) return -KEY_ne;
5139 if (strEQ(d,"not")) return -KEY_not;
5140 if (strEQ(d,"no")) return KEY_no;
5145 if (strEQ(d,"or")) return -KEY_or;
5148 if (strEQ(d,"ord")) return -KEY_ord;
5149 if (strEQ(d,"oct")) return -KEY_oct;
5150 if (strEQ(d,"our")) return KEY_our;
5153 if (strEQ(d,"open")) return -KEY_open;
5156 if (strEQ(d,"opendir")) return -KEY_opendir;
5163 if (strEQ(d,"pop")) return KEY_pop;
5164 if (strEQ(d,"pos")) return KEY_pos;
5167 if (strEQ(d,"push")) return KEY_push;
5168 if (strEQ(d,"pack")) return -KEY_pack;
5169 if (strEQ(d,"pipe")) return -KEY_pipe;
5172 if (strEQ(d,"print")) return KEY_print;
5175 if (strEQ(d,"printf")) return KEY_printf;
5178 if (strEQ(d,"package")) return KEY_package;
5181 if (strEQ(d,"prototype")) return KEY_prototype;
5186 if (strEQ(d,"q")) return KEY_q;
5187 if (strEQ(d,"qr")) return KEY_qr;
5188 if (strEQ(d,"qq")) return KEY_qq;
5189 if (strEQ(d,"qw")) return KEY_qw;
5190 if (strEQ(d,"qx")) return KEY_qx;
5192 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5197 if (strEQ(d,"ref")) return -KEY_ref;
5200 if (strEQ(d,"read")) return -KEY_read;
5201 if (strEQ(d,"rand")) return -KEY_rand;
5202 if (strEQ(d,"recv")) return -KEY_recv;
5203 if (strEQ(d,"redo")) return KEY_redo;
5206 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5207 if (strEQ(d,"reset")) return -KEY_reset;
5210 if (strEQ(d,"return")) return KEY_return;
5211 if (strEQ(d,"rename")) return -KEY_rename;
5212 if (strEQ(d,"rindex")) return -KEY_rindex;
5215 if (strEQ(d,"require")) return -KEY_require;
5216 if (strEQ(d,"reverse")) return -KEY_reverse;
5217 if (strEQ(d,"readdir")) return -KEY_readdir;
5220 if (strEQ(d,"readlink")) return -KEY_readlink;
5221 if (strEQ(d,"readline")) return -KEY_readline;
5222 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5225 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5231 case 0: return KEY_s;
5233 if (strEQ(d,"scalar")) return KEY_scalar;
5238 if (strEQ(d,"seek")) return -KEY_seek;
5239 if (strEQ(d,"send")) return -KEY_send;
5242 if (strEQ(d,"semop")) return -KEY_semop;
5245 if (strEQ(d,"select")) return -KEY_select;
5246 if (strEQ(d,"semctl")) return -KEY_semctl;
5247 if (strEQ(d,"semget")) return -KEY_semget;
5250 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5251 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5254 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5255 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5258 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5261 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5262 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5263 if (strEQ(d,"setservent")) return -KEY_setservent;
5266 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5267 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5274 if (strEQ(d,"shift")) return KEY_shift;
5277 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5278 if (strEQ(d,"shmget")) return -KEY_shmget;
5281 if (strEQ(d,"shmread")) return -KEY_shmread;
5284 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5285 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5290 if (strEQ(d,"sin")) return -KEY_sin;
5293 if (strEQ(d,"sleep")) return -KEY_sleep;
5296 if (strEQ(d,"sort")) return KEY_sort;
5297 if (strEQ(d,"socket")) return -KEY_socket;
5298 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5301 if (strEQ(d,"split")) return KEY_split;
5302 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5303 if (strEQ(d,"splice")) return KEY_splice;
5306 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5309 if (strEQ(d,"srand")) return -KEY_srand;
5312 if (strEQ(d,"stat")) return -KEY_stat;
5313 if (strEQ(d,"study")) return KEY_study;
5316 if (strEQ(d,"substr")) return -KEY_substr;
5317 if (strEQ(d,"sub")) return KEY_sub;
5322 if (strEQ(d,"system")) return -KEY_system;
5325 if (strEQ(d,"symlink")) return -KEY_symlink;
5326 if (strEQ(d,"syscall")) return -KEY_syscall;
5327 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5328 if (strEQ(d,"sysread")) return -KEY_sysread;
5329 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5332 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5341 if (strEQ(d,"tr")) return KEY_tr;
5344 if (strEQ(d,"tie")) return KEY_tie;
5347 if (strEQ(d,"tell")) return -KEY_tell;
5348 if (strEQ(d,"tied")) return KEY_tied;
5349 if (strEQ(d,"time")) return -KEY_time;
5352 if (strEQ(d,"times")) return -KEY_times;
5355 if (strEQ(d,"telldir")) return -KEY_telldir;
5358 if (strEQ(d,"truncate")) return -KEY_truncate;
5365 if (strEQ(d,"uc")) return -KEY_uc;
5368 if (strEQ(d,"use")) return KEY_use;
5371 if (strEQ(d,"undef")) return KEY_undef;
5372 if (strEQ(d,"until")) return KEY_until;
5373 if (strEQ(d,"untie")) return KEY_untie;
5374 if (strEQ(d,"utime")) return -KEY_utime;
5375 if (strEQ(d,"umask")) return -KEY_umask;
5378 if (strEQ(d,"unless")) return KEY_unless;
5379 if (strEQ(d,"unpack")) return -KEY_unpack;
5380 if (strEQ(d,"unlink")) return -KEY_unlink;
5383 if (strEQ(d,"unshift")) return KEY_unshift;
5384 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5389 if (strEQ(d,"values")) return -KEY_values;
5390 if (strEQ(d,"vec")) return -KEY_vec;
5395 if (strEQ(d,"warn")) return -KEY_warn;
5396 if (strEQ(d,"wait")) return -KEY_wait;
5399 if (strEQ(d,"while")) return KEY_while;
5400 if (strEQ(d,"write")) return -KEY_write;
5403 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5406 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5411 if (len == 1) return -KEY_x;
5412 if (strEQ(d,"xor")) return -KEY_xor;
5415 if (len == 1) return KEY_y;
5424 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5428 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5429 dTHR; /* only for ckWARN */
5430 if (ckWARN(WARN_SYNTAX)) {
5432 for (w = s+2; *w && level; w++) {
5439 for (; *w && isSPACE(*w); w++) ;
5440 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5441 Perl_warner(aTHX_ WARN_SYNTAX,
5442 "%s (...) interpreted as function",name);
5445 while (s < PL_bufend && isSPACE(*s))
5449 while (s < PL_bufend && isSPACE(*s))
5451 if (isIDFIRST_lazy(s)) {
5453 while (isALNUM_lazy(s))
5455 while (s < PL_bufend && isSPACE(*s))
5460 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5464 Perl_croak(aTHX_ "No comma allowed after %s", what);
5469 /* Either returns sv, or mortalizes sv and returns a new SV*.
5470 Best used as sv=new_constant(..., sv, ...).
5471 If s, pv are NULL, calls subroutine with one argument,
5472 and type is used with error messages only. */
5475 S_new_constant(pTHX_ char *s, STRLEN len, char *key, SV *sv, SV *pv, char *type)
5478 HV *table = GvHV(PL_hintgv); /* ^H */
5482 char *why, *why1, *why2;
5484 if (!(PL_hints & HINT_LOCALIZE_HH)) {
5487 why = "%^H is not localized";
5491 msg = Perl_newSVpvf(aTHX_ "constant(%s): %s%s%s",
5492 (type ? type: "undef"), why1, why2, why);
5493 yyerror(SvPVX(msg));
5498 why = "%^H is not defined";
5501 cvp = hv_fetch(table, key, strlen(key), FALSE);
5502 if (!cvp || !SvOK(*cvp)) {
5503 why = "} is not defined";
5508 sv_2mortal(sv); /* Parent created it permanently */
5511 pv = sv_2mortal(newSVpvn(s, len));
5513 typesv = sv_2mortal(newSVpv(type, 0));
5515 typesv = &PL_sv_undef;
5517 PUSHSTACKi(PERLSI_OVERLOAD);
5530 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5534 /* Check the eval first */
5535 if (!PL_in_eval && SvTRUE(ERRSV))
5538 sv_catpv(ERRSV, "Propagated");
5539 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5541 res = SvREFCNT_inc(sv);
5554 why = "}} did not return a defined value";
5555 why1 = "Call to &{$^H{";
5565 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
5567 register char *d = dest;
5568 register char *e = d + destlen - 3; /* two-character token, ending NUL */
5571 Perl_croak(aTHX_ ident_too_long);
5572 if (isALNUM(*s)) /* UTF handled below */
5574 else if (*s == '\'' && allow_package && isIDFIRST_lazy(s+1)) {
5579 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
5583 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5584 char *t = s + UTF8SKIP(s);
5585 while (*t & 0x80 && is_utf8_mark((U8*)t))
5587 if (d + (t - s) > e)
5588 Perl_croak(aTHX_ ident_too_long);
5589 Copy(s, d, t - s, char);
5602 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
5609 if (PL_lex_brackets == 0)
5610 PL_lex_fakebrack = 0;
5614 e = d + destlen - 3; /* two-character token, ending NUL */
5616 while (isDIGIT(*s)) {
5618 Perl_croak(aTHX_ ident_too_long);
5625 Perl_croak(aTHX_ ident_too_long);
5626 if (isALNUM(*s)) /* UTF handled below */
5628 else if (*s == '\'' && isIDFIRST_lazy(s+1)) {
5633 else if (*s == ':' && s[1] == ':') {
5637 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5638 char *t = s + UTF8SKIP(s);
5639 while (*t & 0x80 && is_utf8_mark((U8*)t))
5641 if (d + (t - s) > e)
5642 Perl_croak(aTHX_ ident_too_long);
5643 Copy(s, d, t - s, char);
5654 if (PL_lex_state != LEX_NORMAL)
5655 PL_lex_state = LEX_INTERPENDMAYBE;
5658 if (*s == '$' && s[1] &&
5659 (isALNUM_lazy(s+1) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
5672 if (*d == '^' && *s && isCONTROLVAR(*s)) {
5677 if (isSPACE(s[-1])) {
5680 if (ch != ' ' && ch != '\t') {
5686 if (isIDFIRST_lazy(d)) {
5690 while (e < send && isALNUM_lazy(e) || *e == ':') {
5692 while (e < send && *e & 0x80 && is_utf8_mark((U8*)e))
5695 Copy(s, d, e - s, char);
5700 while ((isALNUM(*s) || *s == ':') && d < e)
5703 Perl_croak(aTHX_ ident_too_long);
5706 while (s < send && (*s == ' ' || *s == '\t')) s++;
5707 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
5708 dTHR; /* only for ckWARN */
5709 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
5710 char *brack = *s == '[' ? "[...]" : "{...}";
5711 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5712 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
5713 funny, dest, brack, funny, dest, brack);
5715 PL_lex_fakebrack = PL_lex_brackets+1;
5717 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
5721 /* Handle extended ${^Foo} variables
5722 * 1999-02-27 mjd-perl-patch@plover.com */
5723 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
5727 while (isALNUM(*s) && d < e) {
5731 Perl_croak(aTHX_ ident_too_long);
5736 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
5737 PL_lex_state = LEX_INTERPEND;
5740 if (PL_lex_state == LEX_NORMAL) {
5741 dTHR; /* only for ckWARN */
5742 if (ckWARN(WARN_AMBIGUOUS) &&
5743 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
5745 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5746 "Ambiguous use of %c{%s} resolved to %c%s",
5747 funny, dest, funny, dest);
5752 s = bracket; /* let the parser handle it */
5756 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
5757 PL_lex_state = LEX_INTERPEND;
5762 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
5767 *pmfl |= PMf_GLOBAL;
5769 *pmfl |= PMf_CONTINUE;
5773 *pmfl |= PMf_MULTILINE;
5775 *pmfl |= PMf_SINGLELINE;
5777 *pmfl |= PMf_EXTENDED;
5781 S_scan_pat(pTHX_ char *start, I32 type)
5786 s = scan_str(start,FALSE,FALSE);
5789 SvREFCNT_dec(PL_lex_stuff);
5790 PL_lex_stuff = Nullsv;
5791 Perl_croak(aTHX_ "Search pattern not terminated");
5794 pm = (PMOP*)newPMOP(type, 0);
5795 if (PL_multi_open == '?')
5796 pm->op_pmflags |= PMf_ONCE;
5798 while (*s && strchr("iomsx", *s))
5799 pmflag(&pm->op_pmflags,*s++);
5802 while (*s && strchr("iogcmsx", *s))
5803 pmflag(&pm->op_pmflags,*s++);
5805 pm->op_pmpermflags = pm->op_pmflags;
5807 PL_lex_op = (OP*)pm;
5808 yylval.ival = OP_MATCH;
5813 S_scan_subst(pTHX_ char *start)
5820 yylval.ival = OP_NULL;
5822 s = scan_str(start,FALSE,FALSE);
5826 SvREFCNT_dec(PL_lex_stuff);
5827 PL_lex_stuff = Nullsv;
5828 Perl_croak(aTHX_ "Substitution pattern not terminated");
5831 if (s[-1] == PL_multi_open)
5834 first_start = PL_multi_start;
5835 s = scan_str(s,FALSE,FALSE);
5838 SvREFCNT_dec(PL_lex_stuff);
5839 PL_lex_stuff = Nullsv;
5841 SvREFCNT_dec(PL_lex_repl);
5842 PL_lex_repl = Nullsv;
5843 Perl_croak(aTHX_ "Substitution replacement not terminated");
5845 PL_multi_start = first_start; /* so whole substitution is taken together */
5847 pm = (PMOP*)newPMOP(OP_SUBST, 0);
5853 else if (strchr("iogcmsx", *s))
5854 pmflag(&pm->op_pmflags,*s++);
5861 PL_sublex_info.super_bufptr = s;
5862 PL_sublex_info.super_bufend = PL_bufend;
5864 pm->op_pmflags |= PMf_EVAL;
5865 repl = newSVpvn("",0);
5867 sv_catpv(repl, es ? "eval " : "do ");
5868 sv_catpvn(repl, "{ ", 2);
5869 sv_catsv(repl, PL_lex_repl);
5870 sv_catpvn(repl, " };", 2);
5872 SvREFCNT_dec(PL_lex_repl);
5876 pm->op_pmpermflags = pm->op_pmflags;
5877 PL_lex_op = (OP*)pm;
5878 yylval.ival = OP_SUBST;
5883 S_scan_trans(pTHX_ char *start)
5894 yylval.ival = OP_NULL;
5896 s = scan_str(start,FALSE,FALSE);
5899 SvREFCNT_dec(PL_lex_stuff);
5900 PL_lex_stuff = Nullsv;
5901 Perl_croak(aTHX_ "Transliteration pattern not terminated");
5903 if (s[-1] == PL_multi_open)
5906 s = scan_str(s,FALSE,FALSE);
5909 SvREFCNT_dec(PL_lex_stuff);
5910 PL_lex_stuff = Nullsv;
5912 SvREFCNT_dec(PL_lex_repl);
5913 PL_lex_repl = Nullsv;
5914 Perl_croak(aTHX_ "Transliteration replacement not terminated");
5918 o = newSVOP(OP_TRANS, 0, 0);
5919 utf8 = OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF;
5922 New(803,tbl,256,short);
5923 o = newPVOP(OP_TRANS, 0, (char*)tbl);
5927 complement = del = squash = 0;
5928 while (strchr("cdsCU", *s)) {
5930 complement = OPpTRANS_COMPLEMENT;
5932 del = OPpTRANS_DELETE;
5934 squash = OPpTRANS_SQUASH;
5939 utf8 &= ~OPpTRANS_FROM_UTF;
5941 utf8 |= OPpTRANS_FROM_UTF;
5945 utf8 &= ~OPpTRANS_TO_UTF;
5947 utf8 |= OPpTRANS_TO_UTF;
5950 Perl_croak(aTHX_ "Too many /C and /U options");
5955 o->op_private = del|squash|complement|utf8;
5958 yylval.ival = OP_TRANS;
5963 S_scan_heredoc(pTHX_ register char *s)
5967 I32 op_type = OP_SCALAR;
5974 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
5978 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
5981 for (peek = s; *peek == ' ' || *peek == '\t'; peek++) ;
5982 if (*peek && strchr("`'\"",*peek)) {
5985 s = delimcpy(d, e, s, PL_bufend, term, &len);
5995 if (!isALNUM_lazy(s))
5996 deprecate("bare << to mean <<\"\"");
5997 for (; isALNUM_lazy(s); s++) {
6002 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6003 Perl_croak(aTHX_ "Delimiter for here document is too long");
6006 len = d - PL_tokenbuf;
6007 #ifndef PERL_STRICT_CR
6008 d = strchr(s, '\r');
6012 while (s < PL_bufend) {
6018 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6027 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6032 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6033 herewas = newSVpvn(s,PL_bufend-s);
6035 s--, herewas = newSVpvn(s,d-s);
6036 s += SvCUR(herewas);
6038 tmpstr = NEWSV(87,79);
6039 sv_upgrade(tmpstr, SVt_PVIV);
6044 else if (term == '`') {
6045 op_type = OP_BACKTICK;
6046 SvIVX(tmpstr) = '\\';
6050 PL_multi_start = PL_curcop->cop_line;
6051 PL_multi_open = PL_multi_close = '<';
6052 term = *PL_tokenbuf;
6053 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6054 char *bufptr = PL_sublex_info.super_bufptr;
6055 char *bufend = PL_sublex_info.super_bufend;
6056 char *olds = s - SvCUR(herewas);
6057 s = strchr(bufptr, '\n');
6061 while (s < bufend &&
6062 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6064 PL_curcop->cop_line++;
6067 PL_curcop->cop_line = PL_multi_start;
6068 missingterm(PL_tokenbuf);
6070 sv_setpvn(herewas,bufptr,d-bufptr+1);
6071 sv_setpvn(tmpstr,d+1,s-d);
6073 sv_catpvn(herewas,s,bufend-s);
6074 (void)strcpy(bufptr,SvPVX(herewas));
6081 while (s < PL_bufend &&
6082 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6084 PL_curcop->cop_line++;
6086 if (s >= PL_bufend) {
6087 PL_curcop->cop_line = PL_multi_start;
6088 missingterm(PL_tokenbuf);
6090 sv_setpvn(tmpstr,d+1,s-d);
6092 PL_curcop->cop_line++; /* the preceding stmt passes a newline */
6094 sv_catpvn(herewas,s,PL_bufend-s);
6095 sv_setsv(PL_linestr,herewas);
6096 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6097 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6100 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6101 while (s >= PL_bufend) { /* multiple line string? */
6103 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6104 PL_curcop->cop_line = PL_multi_start;
6105 missingterm(PL_tokenbuf);
6107 PL_curcop->cop_line++;
6108 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6109 #ifndef PERL_STRICT_CR
6110 if (PL_bufend - PL_linestart >= 2) {
6111 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6112 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6114 PL_bufend[-2] = '\n';
6116 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6118 else if (PL_bufend[-1] == '\r')
6119 PL_bufend[-1] = '\n';
6121 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6122 PL_bufend[-1] = '\n';
6124 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6125 SV *sv = NEWSV(88,0);
6127 sv_upgrade(sv, SVt_PVMG);
6128 sv_setsv(sv,PL_linestr);
6129 av_store(GvAV(PL_curcop->cop_filegv),
6130 (I32)PL_curcop->cop_line,sv);
6132 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6135 sv_catsv(PL_linestr,herewas);
6136 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6140 sv_catsv(tmpstr,PL_linestr);
6145 PL_multi_end = PL_curcop->cop_line;
6146 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6147 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6148 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6150 SvREFCNT_dec(herewas);
6151 PL_lex_stuff = tmpstr;
6152 yylval.ival = op_type;
6157 takes: current position in input buffer
6158 returns: new position in input buffer
6159 side-effects: yylval and lex_op are set.
6164 <FH> read from filehandle
6165 <pkg::FH> read from package qualified filehandle
6166 <pkg'FH> read from package qualified filehandle
6167 <$fh> read from filehandle in $fh
6173 S_scan_inputsymbol(pTHX_ char *start)
6175 register char *s = start; /* current position in buffer */
6181 d = PL_tokenbuf; /* start of temp holding space */
6182 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6183 end = strchr(s, '\n');
6186 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6188 /* die if we didn't have space for the contents of the <>,
6189 or if it didn't end, or if we see a newline
6192 if (len >= sizeof PL_tokenbuf)
6193 Perl_croak(aTHX_ "Excessively long <> operator");
6195 Perl_croak(aTHX_ "Unterminated <> operator");
6200 Remember, only scalar variables are interpreted as filehandles by
6201 this code. Anything more complex (e.g., <$fh{$num}>) will be
6202 treated as a glob() call.
6203 This code makes use of the fact that except for the $ at the front,
6204 a scalar variable and a filehandle look the same.
6206 if (*d == '$' && d[1]) d++;
6208 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6209 while (*d && (isALNUM_lazy(d) || *d == '\'' || *d == ':'))
6212 /* If we've tried to read what we allow filehandles to look like, and
6213 there's still text left, then it must be a glob() and not a getline.
6214 Use scan_str to pull out the stuff between the <> and treat it
6215 as nothing more than a string.
6218 if (d - PL_tokenbuf != len) {
6219 yylval.ival = OP_GLOB;
6221 s = scan_str(start,FALSE,FALSE);
6223 Perl_croak(aTHX_ "Glob not terminated");
6227 /* we're in a filehandle read situation */
6230 /* turn <> into <ARGV> */
6232 (void)strcpy(d,"ARGV");
6234 /* if <$fh>, create the ops to turn the variable into a
6240 /* try to find it in the pad for this block, otherwise find
6241 add symbol table ops
6243 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6244 OP *o = newOP(OP_PADSV, 0);
6246 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o);
6249 GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV);
6250 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0,
6251 newUNOP(OP_RV2SV, 0,
6252 newGVOP(OP_GV, 0, gv)));
6254 PL_lex_op->op_flags |= OPf_SPECIAL;
6255 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6256 yylval.ival = OP_NULL;
6259 /* If it's none of the above, it must be a literal filehandle
6260 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6262 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6263 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6264 yylval.ival = OP_NULL;
6273 takes: start position in buffer
6274 keep_quoted preserve \ on the embedded delimiter(s)
6275 keep_delims preserve the delimiters around the string
6276 returns: position to continue reading from buffer
6277 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6278 updates the read buffer.
6280 This subroutine pulls a string out of the input. It is called for:
6281 q single quotes q(literal text)
6282 ' single quotes 'literal text'
6283 qq double quotes qq(interpolate $here please)
6284 " double quotes "interpolate $here please"
6285 qx backticks qx(/bin/ls -l)
6286 ` backticks `/bin/ls -l`
6287 qw quote words @EXPORT_OK = qw( func() $spam )
6288 m// regexp match m/this/
6289 s/// regexp substitute s/this/that/
6290 tr/// string transliterate tr/this/that/
6291 y/// string transliterate y/this/that/
6292 ($*@) sub prototypes sub foo ($)
6293 (stuff) sub attr parameters sub foo : attr(stuff)
6294 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6296 In most of these cases (all but <>, patterns and transliterate)
6297 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6298 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6299 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6302 It skips whitespace before the string starts, and treats the first
6303 character as the delimiter. If the delimiter is one of ([{< then
6304 the corresponding "close" character )]}> is used as the closing
6305 delimiter. It allows quoting of delimiters, and if the string has
6306 balanced delimiters ([{<>}]) it allows nesting.
6308 The lexer always reads these strings into lex_stuff, except in the
6309 case of the operators which take *two* arguments (s/// and tr///)
6310 when it checks to see if lex_stuff is full (presumably with the 1st
6311 arg to s or tr) and if so puts the string into lex_repl.
6316 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6319 SV *sv; /* scalar value: string */
6320 char *tmps; /* temp string, used for delimiter matching */
6321 register char *s = start; /* current position in the buffer */
6322 register char term; /* terminating character */
6323 register char *to; /* current position in the sv's data */
6324 I32 brackets = 1; /* bracket nesting level */
6326 /* skip space before the delimiter */
6330 /* mark where we are, in case we need to report errors */
6333 /* after skipping whitespace, the next character is the terminator */
6335 /* mark where we are */
6336 PL_multi_start = PL_curcop->cop_line;
6337 PL_multi_open = term;
6339 /* find corresponding closing delimiter */
6340 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6342 PL_multi_close = term;
6344 /* create a new SV to hold the contents. 87 is leak category, I'm
6345 assuming. 79 is the SV's initial length. What a random number. */
6347 sv_upgrade(sv, SVt_PVIV);
6349 (void)SvPOK_only(sv); /* validate pointer */
6351 /* move past delimiter and try to read a complete string */
6353 sv_catpvn(sv, s, 1);
6356 /* extend sv if need be */
6357 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6358 /* set 'to' to the next character in the sv's string */
6359 to = SvPVX(sv)+SvCUR(sv);
6361 /* if open delimiter is the close delimiter read unbridle */
6362 if (PL_multi_open == PL_multi_close) {
6363 for (; s < PL_bufend; s++,to++) {
6364 /* embedded newlines increment the current line number */
6365 if (*s == '\n' && !PL_rsfp)
6366 PL_curcop->cop_line++;
6367 /* handle quoted delimiters */
6368 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6369 if (!keep_quoted && s[1] == term)
6371 /* any other quotes are simply copied straight through */
6375 /* terminate when run out of buffer (the for() condition), or
6376 have found the terminator */
6377 else if (*s == term)
6383 /* if the terminator isn't the same as the start character (e.g.,
6384 matched brackets), we have to allow more in the quoting, and
6385 be prepared for nested brackets.
6388 /* read until we run out of string, or we find the terminator */
6389 for (; s < PL_bufend; s++,to++) {
6390 /* embedded newlines increment the line count */
6391 if (*s == '\n' && !PL_rsfp)
6392 PL_curcop->cop_line++;
6393 /* backslashes can escape the open or closing characters */
6394 if (*s == '\\' && s+1 < PL_bufend) {
6396 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6401 /* allow nested opens and closes */
6402 else if (*s == PL_multi_close && --brackets <= 0)
6404 else if (*s == PL_multi_open)
6409 /* terminate the copied string and update the sv's end-of-string */
6411 SvCUR_set(sv, to - SvPVX(sv));
6414 * this next chunk reads more into the buffer if we're not done yet
6417 if (s < PL_bufend) break; /* handle case where we are done yet :-) */
6419 #ifndef PERL_STRICT_CR
6420 if (to - SvPVX(sv) >= 2) {
6421 if ((to[-2] == '\r' && to[-1] == '\n') ||
6422 (to[-2] == '\n' && to[-1] == '\r'))
6426 SvCUR_set(sv, to - SvPVX(sv));
6428 else if (to[-1] == '\r')
6431 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6435 /* if we're out of file, or a read fails, bail and reset the current
6436 line marker so we can report where the unterminated string began
6439 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6441 PL_curcop->cop_line = PL_multi_start;
6444 /* we read a line, so increment our line counter */
6445 PL_curcop->cop_line++;
6447 /* update debugger info */
6448 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6449 SV *sv = NEWSV(88,0);
6451 sv_upgrade(sv, SVt_PVMG);
6452 sv_setsv(sv,PL_linestr);
6453 av_store(GvAV(PL_curcop->cop_filegv),
6454 (I32)PL_curcop->cop_line, sv);
6457 /* having changed the buffer, we must update PL_bufend */
6458 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6461 /* at this point, we have successfully read the delimited string */
6464 sv_catpvn(sv, s, 1);
6465 PL_multi_end = PL_curcop->cop_line;
6468 /* if we allocated too much space, give some back */
6469 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6470 SvLEN_set(sv, SvCUR(sv) + 1);
6471 Renew(SvPVX(sv), SvLEN(sv), char);
6474 /* decide whether this is the first or second quoted string we've read
6487 takes: pointer to position in buffer
6488 returns: pointer to new position in buffer
6489 side-effects: builds ops for the constant in yylval.op
6491 Read a number in any of the formats that Perl accepts:
6493 0(x[0-7A-F]+)|([0-7]+)|(b[01])
6494 [\d_]+(\.[\d_]*)?[Ee](\d+)
6496 Underbars (_) are allowed in decimal numbers. If -w is on,
6497 underbars before a decimal point must be at three digit intervals.
6499 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6502 If it reads a number without a decimal point or an exponent, it will
6503 try converting the number to an integer and see if it can do so
6504 without loss of precision.
6508 Perl_scan_num(pTHX_ char *start)
6510 register char *s = start; /* current position in buffer */
6511 register char *d; /* destination in temp buffer */
6512 register char *e; /* end of temp buffer */
6513 IV tryiv; /* used to see if it can be an IV */
6514 NV value; /* number read, as a double */
6515 SV *sv; /* place to put the converted number */
6516 bool floatit; /* boolean: int or float? */
6517 char *lastub = 0; /* position of last underbar */
6518 static char number_too_long[] = "Number too long";
6520 /* We use the first character to decide what type of number this is */
6524 Perl_croak(aTHX_ "panic: scan_num");
6526 /* if it starts with a 0, it could be an octal number, a decimal in
6527 0.13 disguise, or a hexadecimal number, or a binary number.
6532 u holds the "number so far"
6533 shift the power of 2 of the base
6534 (hex == 4, octal == 3, binary == 1)
6535 overflowed was the number more than we can hold?
6537 Shift is used when we add a digit. It also serves as an "are
6538 we in octal/hex/binary?" indicator to disallow hex characters
6545 bool overflowed = FALSE;
6546 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6547 static char* bases[5] = { "", "binary", "", "octal",
6549 static char* Bases[5] = { "", "Binary", "", "Octal",
6551 static char *maxima[5] = { "",
6552 "0b11111111111111111111111111111111",
6556 char *base, *Base, *max;
6562 } else if (s[1] == 'b') {
6566 /* check for a decimal in disguise */
6567 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
6569 /* so it must be octal */
6573 base = bases[shift];
6574 Base = Bases[shift];
6575 max = maxima[shift];
6577 /* read the rest of the number */
6579 /* x is used in the overflow test,
6580 b is the digit we're adding on. */
6585 /* if we don't mention it, we're done */
6594 /* 8 and 9 are not octal */
6597 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
6601 case '2': case '3': case '4':
6602 case '5': case '6': case '7':
6604 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
6608 b = *s++ & 15; /* ASCII digit -> value of digit */
6612 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
6613 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
6614 /* make sure they said 0x */
6619 /* Prepare to put the digit we have onto the end
6620 of the number so far. We check for overflows.
6625 x = u << shift; /* make room for the digit */
6627 if ((x >> shift) != u
6628 && !(PL_hints & HINT_NEW_BINARY)) {
6632 if (ckWARN_d(WARN_OVERFLOW))
6633 Perl_warner(aTHX_ WARN_OVERFLOW,
6634 "Integer overflow in %s number",
6637 u = x | b; /* add the digit to the end */
6640 n *= nvshift[shift];
6641 /* If an NV has not enough bits in its
6642 * mantissa to represent an UV this summing of
6643 * small low-order numbers is a waste of time
6644 * (because the NV cannot preserve the
6645 * low-order bits anyway): we could just
6646 * remember when did we overflow and in the
6647 * end just multiply n by the right
6655 /* if we get here, we had success: make a scalar value from
6662 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
6663 Perl_warner(aTHX_ WARN_PORTABLE,
6664 "%s number > %s non-portable",
6671 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
6672 Perl_warner(aTHX_ WARN_PORTABLE,
6673 "%s number > %s non-portable",
6678 if (PL_hints & HINT_NEW_BINARY)
6679 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
6684 handle decimal numbers.
6685 we're also sent here when we read a 0 as the first digit
6687 case '1': case '2': case '3': case '4': case '5':
6688 case '6': case '7': case '8': case '9': case '.':
6691 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
6694 /* read next group of digits and _ and copy into d */
6695 while (isDIGIT(*s) || *s == '_') {
6696 /* skip underscores, checking for misplaced ones
6700 dTHR; /* only for ckWARN */
6701 if (ckWARN(WARN_SYNTAX) && lastub && s - lastub != 3)
6702 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6706 /* check for end of fixed-length buffer */
6708 Perl_croak(aTHX_ number_too_long);
6709 /* if we're ok, copy the character */
6714 /* final misplaced underbar check */
6715 if (lastub && s - lastub != 3) {
6717 if (ckWARN(WARN_SYNTAX))
6718 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6721 /* read a decimal portion if there is one. avoid
6722 3..5 being interpreted as the number 3. followed
6725 if (*s == '.' && s[1] != '.') {
6729 /* copy, ignoring underbars, until we run out of
6730 digits. Note: no misplaced underbar checks!
6732 for (; isDIGIT(*s) || *s == '_'; s++) {
6733 /* fixed length buffer check */
6735 Perl_croak(aTHX_ number_too_long);
6741 /* read exponent part, if present */
6742 if (*s && strchr("eE",*s) && strchr("+-0123456789",s[1])) {
6746 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
6747 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
6749 /* allow positive or negative exponent */
6750 if (*s == '+' || *s == '-')
6753 /* read digits of exponent (no underbars :-) */
6754 while (isDIGIT(*s)) {
6756 Perl_croak(aTHX_ number_too_long);
6761 /* terminate the string */
6764 /* make an sv from the string */
6767 value = Atof(PL_tokenbuf);
6770 See if we can make do with an integer value without loss of
6771 precision. We use I_V to cast to an int, because some
6772 compilers have issues. Then we try casting it back and see
6773 if it was the same. We only do this if we know we
6774 specifically read an integer.
6776 Note: if floatit is true, then we don't need to do the
6780 if (!floatit && (NV)tryiv == value)
6781 sv_setiv(sv, tryiv);
6783 sv_setnv(sv, value);
6784 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
6785 (PL_hints & HINT_NEW_INTEGER) )
6786 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
6787 (floatit ? "float" : "integer"),
6792 /* make the op for the constant and return */
6794 yylval.opval = newSVOP(OP_CONST, 0, sv);
6800 S_scan_formline(pTHX_ register char *s)
6805 SV *stuff = newSVpvn("",0);
6806 bool needargs = FALSE;
6809 if (*s == '.' || *s == '}') {
6811 #ifdef PERL_STRICT_CR
6812 for (t = s+1;*t == ' ' || *t == '\t'; t++) ;
6814 for (t = s+1;*t == ' ' || *t == '\t' || *t == '\r'; t++) ;
6816 if (*t == '\n' || t == PL_bufend)
6819 if (PL_in_eval && !PL_rsfp) {
6820 eol = strchr(s,'\n');
6825 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6827 for (t = s; t < eol; t++) {
6828 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
6830 goto enough; /* ~~ must be first line in formline */
6832 if (*t == '@' || *t == '^')
6835 sv_catpvn(stuff, s, eol-s);
6839 s = filter_gets(PL_linestr, PL_rsfp, 0);
6840 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
6841 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
6844 yyerror("Format not terminated");
6854 PL_lex_state = LEX_NORMAL;
6855 PL_nextval[PL_nexttoke].ival = 0;
6859 PL_lex_state = LEX_FORMLINE;
6860 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
6862 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
6866 SvREFCNT_dec(stuff);
6867 PL_lex_formbrack = 0;
6878 PL_cshlen = strlen(PL_cshname);
6883 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
6886 I32 oldsavestack_ix = PL_savestack_ix;
6887 CV* outsidecv = PL_compcv;
6891 assert(SvTYPE(PL_compcv) == SVt_PVCV);
6893 save_I32(&PL_subline);
6894 save_item(PL_subname);
6896 SAVESPTR(PL_curpad);
6897 SAVESPTR(PL_comppad);
6898 SAVESPTR(PL_comppad_name);
6899 SAVESPTR(PL_compcv);
6900 SAVEI32(PL_comppad_name_fill);
6901 SAVEI32(PL_min_intro_pending);
6902 SAVEI32(PL_max_intro_pending);
6903 SAVEI32(PL_pad_reset_pending);
6905 PL_compcv = (CV*)NEWSV(1104,0);
6906 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
6907 CvFLAGS(PL_compcv) |= flags;
6909 PL_comppad = newAV();
6910 av_push(PL_comppad, Nullsv);
6911 PL_curpad = AvARRAY(PL_comppad);
6912 PL_comppad_name = newAV();
6913 PL_comppad_name_fill = 0;
6914 PL_min_intro_pending = 0;
6916 PL_subline = PL_curcop->cop_line;
6918 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
6919 PL_curpad[0] = (SV*)newAV();
6920 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
6921 #endif /* USE_THREADS */
6923 comppadlist = newAV();
6924 AvREAL_off(comppadlist);
6925 av_store(comppadlist, 0, (SV*)PL_comppad_name);
6926 av_store(comppadlist, 1, (SV*)PL_comppad);
6928 CvPADLIST(PL_compcv) = comppadlist;
6929 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
6931 CvOWNER(PL_compcv) = 0;
6932 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
6933 MUTEX_INIT(CvMUTEXP(PL_compcv));
6934 #endif /* USE_THREADS */
6936 return oldsavestack_ix;
6940 Perl_yywarn(pTHX_ char *s)
6943 PL_in_eval |= EVAL_WARNONLY;
6945 PL_in_eval &= ~EVAL_WARNONLY;
6950 Perl_yyerror(pTHX_ char *s)
6954 char *context = NULL;
6958 if (!yychar || (yychar == ';' && !PL_rsfp))
6960 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
6961 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
6962 while (isSPACE(*PL_oldoldbufptr))
6964 context = PL_oldoldbufptr;
6965 contlen = PL_bufptr - PL_oldoldbufptr;
6967 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
6968 PL_oldbufptr != PL_bufptr) {
6969 while (isSPACE(*PL_oldbufptr))
6971 context = PL_oldbufptr;
6972 contlen = PL_bufptr - PL_oldbufptr;
6974 else if (yychar > 255)
6975 where = "next token ???";
6976 else if ((yychar & 127) == 127) {
6977 if (PL_lex_state == LEX_NORMAL ||
6978 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
6979 where = "at end of line";
6980 else if (PL_lex_inpat)
6981 where = "within pattern";
6983 where = "within string";
6986 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
6988 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
6989 else if (isPRINT_LC(yychar))
6990 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
6992 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
6993 where = SvPVX(where_sv);
6995 msg = sv_2mortal(newSVpv(s, 0));
6996 Perl_sv_catpvf(aTHX_ msg, " at %_ line %"IVdf", ",
6997 GvSV(PL_curcop->cop_filegv), (IV)PL_curcop->cop_line);
6999 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7001 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7002 if (PL_multi_start < PL_multi_end && (U32)(PL_curcop->cop_line - PL_multi_end) <= 1) {
7003 Perl_sv_catpvf(aTHX_ msg,
7004 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7005 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7008 if (PL_in_eval & EVAL_WARNONLY)
7009 Perl_warn(aTHX_ "%_", msg);
7012 if (PL_error_count >= 10)
7013 Perl_croak(aTHX_ "%_ has too many errors.\n", GvSV(PL_curcop->cop_filegv));
7015 PL_in_my_stash = Nullhv;
7027 * Restore a source filter.
7031 restore_rsfp(pTHXo_ void *f)
7033 PerlIO *fp = (PerlIO*)f;
7035 if (PL_rsfp == PerlIO_stdin())
7036 PerlIO_clearerr(PL_rsfp);
7037 else if (PL_rsfp && (PL_rsfp != fp))
7038 PerlIO_close(PL_rsfp);
7044 * Restores the state of PL_expect when the lexing that begun with a
7045 * start_lex() call has ended.
7049 restore_expect(pTHXo_ void *e)
7051 /* a safe way to store a small integer in a pointer */
7052 PL_expect = (expectation)((char *)e - PL_tokenbuf);
7056 * restore_lex_expect
7057 * Restores the state of PL_lex_expect when the lexing that begun with a
7058 * start_lex() call has ended.
7062 restore_lex_expect(pTHXo_ void *e)
7064 /* a safe way to store a small integer in a pointer */
7065 PL_lex_expect = (expectation)((char *)e - PL_tokenbuf);