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
123 * FUN1 : not used, except for not, which isn't a UNIOP
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_X(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_X(restore_expect, PL_tokenbuf + PL_expect); /* encode as pointer */
384 SAVEDESTRUCTOR_X(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 CopFILEGV_set(PL_curcop, gv_fetchfile(s));
479 CopFILEGV_set(PL_curcop, 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(CopFILEAV(PL_curcop),(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 $1
1117 The structure of the code is
1118 while (there's a character to process) {
1119 handle transliteration ranges
1120 skip regexp comments
1121 skip # initiated comments in //x patterns
1122 check for embedded @foo
1123 check for embedded scalars
1125 leave intact backslashes from leave (below)
1126 deprecate \1 in strings and sub replacements
1127 handle string-changing backslashes \l \U \Q \E, etc.
1128 switch (what was escaped) {
1129 handle - in a transliteration (becomes a literal -)
1130 handle \132 octal characters
1131 handle 0x15 hex characters
1132 handle \cV (control V)
1133 handle printf backslashes (\f, \r, \n, etc)
1135 } (end if backslash)
1136 } (end while character to read)
1141 S_scan_const(pTHX_ char *start)
1143 register char *send = PL_bufend; /* end of the constant */
1144 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1145 register char *s = start; /* start of the constant */
1146 register char *d = SvPVX(sv); /* destination for copies */
1147 bool dorange = FALSE; /* are we in a translit range? */
1149 I32 utf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1150 ? (PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
1152 I32 thisutf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1153 ? (PL_sublex_info.sub_op->op_private & (PL_lex_repl ?
1154 OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF))
1156 const char *leaveit = /* set of acceptably-backslashed characters */
1158 ? "\\.^$@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++ = (char)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 (UV)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)
1778 if (!PL_rsfp_filters)
1779 PL_rsfp_filters = newAV();
1781 datasv = NEWSV(255,0);
1782 if (!SvUPGRADE(datasv, SVt_PVIO))
1783 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1784 IoDIRP(datasv) = (DIR*)funcp; /* stash funcp into spare field */
1785 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1786 funcp, SvPV_nolen(datasv)));
1787 av_unshift(PL_rsfp_filters, 1);
1788 av_store(PL_rsfp_filters, 0, datasv) ;
1793 /* Delete most recently added instance of this filter function. */
1795 Perl_filter_del(pTHX_ filter_t funcp)
1797 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp));
1798 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
1800 /* if filter is on top of stack (usual case) just pop it off */
1801 if (IoDIRP(FILTER_DATA(AvFILLp(PL_rsfp_filters))) == (DIR*)funcp){
1802 IoDIRP(FILTER_DATA(AvFILLp(PL_rsfp_filters))) = NULL;
1803 sv_free(av_pop(PL_rsfp_filters));
1807 /* we need to search for the correct entry and clear it */
1808 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
1812 /* Invoke the n'th filter function for the current rsfp. */
1814 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
1817 /* 0 = read one text line */
1822 if (!PL_rsfp_filters)
1824 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
1825 /* Provide a default input filter to make life easy. */
1826 /* Note that we append to the line. This is handy. */
1827 DEBUG_P(PerlIO_printf(Perl_debug_log,
1828 "filter_read %d: from rsfp\n", idx));
1832 int old_len = SvCUR(buf_sv) ;
1834 /* ensure buf_sv is large enough */
1835 SvGROW(buf_sv, old_len + maxlen) ;
1836 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
1837 if (PerlIO_error(PL_rsfp))
1838 return -1; /* error */
1840 return 0 ; /* end of file */
1842 SvCUR_set(buf_sv, old_len + len) ;
1845 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
1846 if (PerlIO_error(PL_rsfp))
1847 return -1; /* error */
1849 return 0 ; /* end of file */
1852 return SvCUR(buf_sv);
1854 /* Skip this filter slot if filter has been deleted */
1855 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
1856 DEBUG_P(PerlIO_printf(Perl_debug_log,
1857 "filter_read %d: skipped (filter deleted)\n",
1859 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
1861 /* Get function pointer hidden within datasv */
1862 funcp = (filter_t)IoDIRP(datasv);
1863 DEBUG_P(PerlIO_printf(Perl_debug_log,
1864 "filter_read %d: via function %p (%s)\n",
1865 idx, funcp, SvPV_nolen(datasv)));
1866 /* Call function. The function is expected to */
1867 /* call "FILTER_READ(idx+1, buf_sv)" first. */
1868 /* Return: <0:error, =0:eof, >0:not eof */
1869 return (*funcp)(aTHXo_ idx, buf_sv, maxlen);
1873 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
1876 if (!PL_rsfp_filters) {
1877 filter_add(win32_textfilter,NULL);
1880 if (PL_rsfp_filters) {
1883 SvCUR_set(sv, 0); /* start with empty line */
1884 if (FILTER_READ(0, sv, 0) > 0)
1885 return ( SvPVX(sv) ) ;
1890 return (sv_gets(sv, fp, append));
1895 static char* exp_name[] =
1896 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
1897 "ATTRTERM", "TERMBLOCK"
1904 Works out what to call the token just pulled out of the input
1905 stream. The yacc parser takes care of taking the ops we return and
1906 stitching them into a tree.
1912 if read an identifier
1913 if we're in a my declaration
1914 croak if they tried to say my($foo::bar)
1915 build the ops for a my() declaration
1916 if it's an access to a my() variable
1917 are we in a sort block?
1918 croak if my($a); $a <=> $b
1919 build ops for access to a my() variable
1920 if in a dq string, and they've said @foo and we can't find @foo
1922 build ops for a bareword
1923 if we already built the token before, use it.
1927 #ifdef USE_PURE_BISON
1928 Perl_yylex(pTHX_ YYSTYPE *lvalp, int *lcharp)
1941 #ifdef USE_PURE_BISON
1942 yylval_pointer = lvalp;
1943 yychar_pointer = lcharp;
1946 /* check if there's an identifier for us to look at */
1947 if (PL_pending_ident) {
1948 /* pit holds the identifier we read and pending_ident is reset */
1949 char pit = PL_pending_ident;
1950 PL_pending_ident = 0;
1952 /* if we're in a my(), we can't allow dynamics here.
1953 $foo'bar has already been turned into $foo::bar, so
1954 just check for colons.
1956 if it's a legal name, the OP is a PADANY.
1959 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
1960 tmp = pad_allocmy(PL_tokenbuf);
1963 if (strchr(PL_tokenbuf,':'))
1964 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
1966 yylval.opval = newOP(OP_PADANY, 0);
1967 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
1973 build the ops for accesses to a my() variable.
1975 Deny my($a) or my($b) in a sort block, *if* $a or $b is
1976 then used in a comparison. This catches most, but not
1977 all cases. For instance, it catches
1978 sort { my($a); $a <=> $b }
1980 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
1981 (although why you'd do that is anyone's guess).
1984 if (!strchr(PL_tokenbuf,':')) {
1986 /* Check for single character per-thread SVs */
1987 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
1988 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
1989 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
1991 yylval.opval = newOP(OP_THREADSV, 0);
1992 yylval.opval->op_targ = tmp;
1995 #endif /* USE_THREADS */
1996 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
1997 /* might be an "our" variable" */
1998 if (SvFLAGS(AvARRAY(PL_comppad_name)[tmp]) & SVpad_OUR) {
1999 /* build ops for a bareword */
2000 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2001 yylval.opval->op_private = OPpCONST_ENTERED;
2002 gv_fetchpv(PL_tokenbuf+1,
2004 ? (GV_ADDMULTI | GV_ADDINEVAL | GV_ADDOUR)
2007 ((PL_tokenbuf[0] == '$') ? SVt_PV
2008 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2013 /* if it's a sort block and they're naming $a or $b */
2014 if (PL_last_lop_op == OP_SORT &&
2015 PL_tokenbuf[0] == '$' &&
2016 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
2019 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
2020 d < PL_bufend && *d != '\n';
2023 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
2024 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
2030 yylval.opval = newOP(OP_PADANY, 0);
2031 yylval.opval->op_targ = tmp;
2037 Whine if they've said @foo in a doublequoted string,
2038 and @foo isn't a variable we can find in the symbol
2041 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
2042 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
2043 if (!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
2044 yyerror(Perl_form(aTHX_ "In string, %s now must be written as \\%s",
2045 PL_tokenbuf, PL_tokenbuf));
2048 /* build ops for a bareword */
2049 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2050 yylval.opval->op_private = OPpCONST_ENTERED;
2051 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
2052 ((PL_tokenbuf[0] == '$') ? SVt_PV
2053 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2058 /* no identifier pending identification */
2060 switch (PL_lex_state) {
2062 case LEX_NORMAL: /* Some compilers will produce faster */
2063 case LEX_INTERPNORMAL: /* code if we comment these out. */
2067 /* when we've already built the next token, just pull it out of the queue */
2070 yylval = PL_nextval[PL_nexttoke];
2072 PL_lex_state = PL_lex_defer;
2073 PL_expect = PL_lex_expect;
2074 PL_lex_defer = LEX_NORMAL;
2076 return(PL_nexttype[PL_nexttoke]);
2078 /* interpolated case modifiers like \L \U, including \Q and \E.
2079 when we get here, PL_bufptr is at the \
2081 case LEX_INTERPCASEMOD:
2083 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2084 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2086 /* handle \E or end of string */
2087 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2091 if (PL_lex_casemods) {
2092 oldmod = PL_lex_casestack[--PL_lex_casemods];
2093 PL_lex_casestack[PL_lex_casemods] = '\0';
2095 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2097 PL_lex_state = LEX_INTERPCONCAT;
2101 if (PL_bufptr != PL_bufend)
2103 PL_lex_state = LEX_INTERPCONCAT;
2108 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2109 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2110 if (strchr("LU", *s) &&
2111 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2113 PL_lex_casestack[--PL_lex_casemods] = '\0';
2116 if (PL_lex_casemods > 10) {
2117 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2118 if (newlb != PL_lex_casestack) {
2120 PL_lex_casestack = newlb;
2123 PL_lex_casestack[PL_lex_casemods++] = *s;
2124 PL_lex_casestack[PL_lex_casemods] = '\0';
2125 PL_lex_state = LEX_INTERPCONCAT;
2126 PL_nextval[PL_nexttoke].ival = 0;
2129 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2131 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2133 PL_nextval[PL_nexttoke].ival = OP_LC;
2135 PL_nextval[PL_nexttoke].ival = OP_UC;
2137 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2139 Perl_croak(aTHX_ "panic: yylex");
2142 if (PL_lex_starts) {
2151 case LEX_INTERPPUSH:
2152 return sublex_push();
2154 case LEX_INTERPSTART:
2155 if (PL_bufptr == PL_bufend)
2156 return sublex_done();
2158 PL_lex_dojoin = (*PL_bufptr == '@');
2159 PL_lex_state = LEX_INTERPNORMAL;
2160 if (PL_lex_dojoin) {
2161 PL_nextval[PL_nexttoke].ival = 0;
2164 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2165 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2166 force_next(PRIVATEREF);
2168 force_ident("\"", '$');
2169 #endif /* USE_THREADS */
2170 PL_nextval[PL_nexttoke].ival = 0;
2172 PL_nextval[PL_nexttoke].ival = 0;
2174 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2177 if (PL_lex_starts++) {
2183 case LEX_INTERPENDMAYBE:
2184 if (intuit_more(PL_bufptr)) {
2185 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2191 if (PL_lex_dojoin) {
2192 PL_lex_dojoin = FALSE;
2193 PL_lex_state = LEX_INTERPCONCAT;
2196 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2197 && SvEVALED(PL_lex_repl))
2199 if (PL_bufptr != PL_bufend)
2200 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2201 PL_lex_repl = Nullsv;
2204 case LEX_INTERPCONCAT:
2206 if (PL_lex_brackets)
2207 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2209 if (PL_bufptr == PL_bufend)
2210 return sublex_done();
2212 if (SvIVX(PL_linestr) == '\'') {
2213 SV *sv = newSVsv(PL_linestr);
2216 else if ( PL_hints & HINT_NEW_RE )
2217 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2218 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2222 s = scan_const(PL_bufptr);
2224 PL_lex_state = LEX_INTERPCASEMOD;
2226 PL_lex_state = LEX_INTERPSTART;
2229 if (s != PL_bufptr) {
2230 PL_nextval[PL_nexttoke] = yylval;
2233 if (PL_lex_starts++)
2243 PL_lex_state = LEX_NORMAL;
2244 s = scan_formline(PL_bufptr);
2245 if (!PL_lex_formbrack)
2251 PL_oldoldbufptr = PL_oldbufptr;
2254 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2255 exp_name[PL_expect], s);
2261 if (isIDFIRST_lazy(s))
2263 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2266 goto fake_eof; /* emulate EOF on ^D or ^Z */
2271 if (PL_lex_brackets)
2272 yyerror("Missing right curly or square bracket");
2275 if (s++ < PL_bufend)
2276 goto retry; /* ignore stray nulls */
2279 if (!PL_in_eval && !PL_preambled) {
2280 PL_preambled = TRUE;
2281 sv_setpv(PL_linestr,incl_perldb());
2282 if (SvCUR(PL_linestr))
2283 sv_catpv(PL_linestr,";");
2285 while(AvFILLp(PL_preambleav) >= 0) {
2286 SV *tmpsv = av_shift(PL_preambleav);
2287 sv_catsv(PL_linestr, tmpsv);
2288 sv_catpv(PL_linestr, ";");
2291 sv_free((SV*)PL_preambleav);
2292 PL_preambleav = NULL;
2294 if (PL_minus_n || PL_minus_p) {
2295 sv_catpv(PL_linestr, "LINE: while (<>) {");
2297 sv_catpv(PL_linestr,"chomp;");
2299 GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
2301 GvIMPORTED_AV_on(gv);
2303 if (strchr("/'\"", *PL_splitstr)
2304 && strchr(PL_splitstr + 1, *PL_splitstr))
2305 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2308 s = "'~#\200\1'"; /* surely one char is unused...*/
2309 while (s[1] && strchr(PL_splitstr, *s)) s++;
2311 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c",
2312 "q" + (delim == '\''), delim);
2313 for (s = PL_splitstr; *s; s++) {
2315 sv_catpvn(PL_linestr, "\\", 1);
2316 sv_catpvn(PL_linestr, s, 1);
2318 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2322 sv_catpv(PL_linestr,"@F=split(' ');");
2325 sv_catpv(PL_linestr, "\n");
2326 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2327 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2328 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2329 SV *sv = NEWSV(85,0);
2331 sv_upgrade(sv, SVt_PVMG);
2332 sv_setsv(sv,PL_linestr);
2333 av_store(CopFILEAV(PL_curcop),(I32)PL_curcop->cop_line,sv);
2338 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2341 if (PL_preprocess && !PL_in_eval)
2342 (void)PerlProc_pclose(PL_rsfp);
2343 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2344 PerlIO_clearerr(PL_rsfp);
2346 (void)PerlIO_close(PL_rsfp);
2348 PL_doextract = FALSE;
2350 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2351 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2352 sv_catpv(PL_linestr,";}");
2353 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2354 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2355 PL_minus_n = PL_minus_p = 0;
2358 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2359 sv_setpv(PL_linestr,"");
2360 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2363 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2364 PL_doextract = FALSE;
2366 /* Incest with pod. */
2367 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2368 sv_setpv(PL_linestr, "");
2369 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2370 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2371 PL_doextract = FALSE;
2375 } while (PL_doextract);
2376 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2377 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2378 SV *sv = NEWSV(85,0);
2380 sv_upgrade(sv, SVt_PVMG);
2381 sv_setsv(sv,PL_linestr);
2382 av_store(CopFILEAV(PL_curcop),(I32)PL_curcop->cop_line,sv);
2384 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2385 if (PL_curcop->cop_line == 1) {
2386 while (s < PL_bufend && isSPACE(*s))
2388 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2392 if (*s == '#' && *(s+1) == '!')
2394 #ifdef ALTERNATE_SHEBANG
2396 static char as[] = ALTERNATE_SHEBANG;
2397 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2398 d = s + (sizeof(as) - 1);
2400 #endif /* ALTERNATE_SHEBANG */
2409 while (*d && !isSPACE(*d))
2413 #ifdef ARG_ZERO_IS_SCRIPT
2414 if (ipathend > ipath) {
2416 * HP-UX (at least) sets argv[0] to the script name,
2417 * which makes $^X incorrect. And Digital UNIX and Linux,
2418 * at least, set argv[0] to the basename of the Perl
2419 * interpreter. So, having found "#!", we'll set it right.
2421 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2422 assert(SvPOK(x) || SvGMAGICAL(x));
2423 if (sv_eq(x, CopFILESV(PL_curcop))) {
2424 sv_setpvn(x, ipath, ipathend - ipath);
2427 TAINT_NOT; /* $^X is always tainted, but that's OK */
2429 #endif /* ARG_ZERO_IS_SCRIPT */
2434 d = instr(s,"perl -");
2436 d = instr(s,"perl");
2438 /* avoid getting into infinite loops when shebang
2439 * line contains "Perl" rather than "perl" */
2441 for (d = ipathend-4; d >= ipath; --d) {
2442 if ((*d == 'p' || *d == 'P')
2443 && !ibcmp(d, "perl", 4))
2453 #ifdef ALTERNATE_SHEBANG
2455 * If the ALTERNATE_SHEBANG on this system starts with a
2456 * character that can be part of a Perl expression, then if
2457 * we see it but not "perl", we're probably looking at the
2458 * start of Perl code, not a request to hand off to some
2459 * other interpreter. Similarly, if "perl" is there, but
2460 * not in the first 'word' of the line, we assume the line
2461 * contains the start of the Perl program.
2463 if (d && *s != '#') {
2465 while (*c && !strchr("; \t\r\n\f\v#", *c))
2468 d = Nullch; /* "perl" not in first word; ignore */
2470 *s = '#'; /* Don't try to parse shebang line */
2472 #endif /* ALTERNATE_SHEBANG */
2477 !instr(s,"indir") &&
2478 instr(PL_origargv[0],"perl"))
2484 while (s < PL_bufend && isSPACE(*s))
2486 if (s < PL_bufend) {
2487 Newz(899,newargv,PL_origargc+3,char*);
2489 while (s < PL_bufend && !isSPACE(*s))
2492 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2495 newargv = PL_origargv;
2497 PerlProc_execv(ipath, newargv);
2498 Perl_croak(aTHX_ "Can't exec %s", ipath);
2501 U32 oldpdb = PL_perldb;
2502 bool oldn = PL_minus_n;
2503 bool oldp = PL_minus_p;
2505 while (*d && !isSPACE(*d)) d++;
2506 while (*d == ' ' || *d == '\t') d++;
2510 if (*d == 'M' || *d == 'm') {
2512 while (*d && !isSPACE(*d)) d++;
2513 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2516 d = moreswitches(d);
2518 if (PERLDB_LINE && !oldpdb ||
2519 ( PL_minus_n || PL_minus_p ) && !(oldn || oldp) )
2520 /* if we have already added "LINE: while (<>) {",
2521 we must not do it again */
2523 sv_setpv(PL_linestr, "");
2524 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2525 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2526 PL_preambled = FALSE;
2528 (void)gv_fetchfile(PL_origfilename);
2535 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2537 PL_lex_state = LEX_FORMLINE;
2542 #ifdef PERL_STRICT_CR
2543 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2545 "(Maybe you didn't strip carriage returns after a network transfer?)\n");
2547 case ' ': case '\t': case '\f': case 013:
2552 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2554 while (s < d && *s != '\n')
2559 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2561 PL_lex_state = LEX_FORMLINE;
2571 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2576 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
2579 if (strnEQ(s,"=>",2)) {
2580 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2581 OPERATOR('-'); /* unary minus */
2583 PL_last_uni = PL_oldbufptr;
2584 PL_last_lop_op = OP_FTEREAD; /* good enough */
2586 case 'r': FTST(OP_FTEREAD);
2587 case 'w': FTST(OP_FTEWRITE);
2588 case 'x': FTST(OP_FTEEXEC);
2589 case 'o': FTST(OP_FTEOWNED);
2590 case 'R': FTST(OP_FTRREAD);
2591 case 'W': FTST(OP_FTRWRITE);
2592 case 'X': FTST(OP_FTREXEC);
2593 case 'O': FTST(OP_FTROWNED);
2594 case 'e': FTST(OP_FTIS);
2595 case 'z': FTST(OP_FTZERO);
2596 case 's': FTST(OP_FTSIZE);
2597 case 'f': FTST(OP_FTFILE);
2598 case 'd': FTST(OP_FTDIR);
2599 case 'l': FTST(OP_FTLINK);
2600 case 'p': FTST(OP_FTPIPE);
2601 case 'S': FTST(OP_FTSOCK);
2602 case 'u': FTST(OP_FTSUID);
2603 case 'g': FTST(OP_FTSGID);
2604 case 'k': FTST(OP_FTSVTX);
2605 case 'b': FTST(OP_FTBLK);
2606 case 'c': FTST(OP_FTCHR);
2607 case 't': FTST(OP_FTTTY);
2608 case 'T': FTST(OP_FTTEXT);
2609 case 'B': FTST(OP_FTBINARY);
2610 case 'M': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTMTIME);
2611 case 'A': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTATIME);
2612 case 'C': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTCTIME);
2614 Perl_croak(aTHX_ "Unrecognized file test: -%c", (int)tmp);
2621 if (PL_expect == XOPERATOR)
2626 else if (*s == '>') {
2629 if (isIDFIRST_lazy(s)) {
2630 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2638 if (PL_expect == XOPERATOR)
2641 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2643 OPERATOR('-'); /* unary minus */
2650 if (PL_expect == XOPERATOR)
2655 if (PL_expect == XOPERATOR)
2658 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2664 if (PL_expect != XOPERATOR) {
2665 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2666 PL_expect = XOPERATOR;
2667 force_ident(PL_tokenbuf, '*');
2680 if (PL_expect == XOPERATOR) {
2684 PL_tokenbuf[0] = '%';
2685 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2686 if (!PL_tokenbuf[1]) {
2688 yyerror("Final % should be \\% or %name");
2691 PL_pending_ident = '%';
2710 switch (PL_expect) {
2713 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2715 PL_bufptr = s; /* update in case we back off */
2721 PL_expect = XTERMBLOCK;
2725 while (isIDFIRST_lazy(s)) {
2726 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2728 d = scan_str(d,TRUE,TRUE);
2731 SvREFCNT_dec(PL_lex_stuff);
2732 PL_lex_stuff = Nullsv;
2734 /* MUST advance bufptr here to avoid bogus
2735 "at end of line" context messages from yyerror().
2737 PL_bufptr = s + len;
2738 yyerror("Unterminated attribute parameter in attribute list");
2741 return 0; /* EOF indicator */
2745 SV *sv = newSVpvn(s, len);
2746 sv_catsv(sv, PL_lex_stuff);
2747 attrs = append_elem(OP_LIST, attrs,
2748 newSVOP(OP_CONST, 0, sv));
2749 SvREFCNT_dec(PL_lex_stuff);
2750 PL_lex_stuff = Nullsv;
2753 attrs = append_elem(OP_LIST, attrs,
2754 newSVOP(OP_CONST, 0,
2761 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}' for vi */
2762 if (*s != ';' && *s != tmp) {
2763 char q = ((*s == '\'') ? '"' : '\'');
2764 /* If here for an expression, and parsed no attrs, back off. */
2765 if (tmp == '=' && !attrs) {
2769 /* MUST advance bufptr here to avoid bogus "at end of line"
2770 context messages from yyerror().
2774 yyerror("Unterminated attribute list");
2776 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
2783 PL_nextval[PL_nexttoke].opval = attrs;
2791 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
2792 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
2797 if (PL_curcop->cop_line < PL_copline)
2798 PL_copline = PL_curcop->cop_line;
2809 if (PL_lex_brackets <= 0)
2810 yyerror("Unmatched right square bracket");
2813 if (PL_lex_state == LEX_INTERPNORMAL) {
2814 if (PL_lex_brackets == 0) {
2815 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
2816 PL_lex_state = LEX_INTERPEND;
2823 if (PL_lex_brackets > 100) {
2824 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
2825 if (newlb != PL_lex_brackstack) {
2827 PL_lex_brackstack = newlb;
2830 switch (PL_expect) {
2832 if (PL_lex_formbrack) {
2836 if (PL_oldoldbufptr == PL_last_lop)
2837 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
2839 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2840 OPERATOR(HASHBRACK);
2842 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
2845 PL_tokenbuf[0] = '\0';
2846 if (d < PL_bufend && *d == '-') {
2847 PL_tokenbuf[0] = '-';
2849 while (d < PL_bufend && (*d == ' ' || *d == '\t'))
2852 if (d < PL_bufend && isIDFIRST_lazy(d)) {
2853 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
2855 while (d < PL_bufend && (*d == ' ' || *d == '\t'))
2858 char minus = (PL_tokenbuf[0] == '-');
2859 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
2867 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
2872 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2877 if (PL_oldoldbufptr == PL_last_lop)
2878 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
2880 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2883 OPERATOR(HASHBRACK);
2884 /* This hack serves to disambiguate a pair of curlies
2885 * as being a block or an anon hash. Normally, expectation
2886 * determines that, but in cases where we're not in a
2887 * position to expect anything in particular (like inside
2888 * eval"") we have to resolve the ambiguity. This code
2889 * covers the case where the first term in the curlies is a
2890 * quoted string. Most other cases need to be explicitly
2891 * disambiguated by prepending a `+' before the opening
2892 * curly in order to force resolution as an anon hash.
2894 * XXX should probably propagate the outer expectation
2895 * into eval"" to rely less on this hack, but that could
2896 * potentially break current behavior of eval"".
2900 if (*s == '\'' || *s == '"' || *s == '`') {
2901 /* common case: get past first string, handling escapes */
2902 for (t++; t < PL_bufend && *t != *s;)
2903 if (*t++ == '\\' && (*t == '\\' || *t == *s))
2907 else if (*s == 'q') {
2910 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
2911 && !isALNUM(*t)))) {
2913 char open, close, term;
2916 while (t < PL_bufend && isSPACE(*t))
2920 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
2924 for (t++; t < PL_bufend; t++) {
2925 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
2927 else if (*t == open)
2931 for (t++; t < PL_bufend; t++) {
2932 if (*t == '\\' && t+1 < PL_bufend)
2934 else if (*t == close && --brackets <= 0)
2936 else if (*t == open)
2942 else if (isIDFIRST_lazy(s)) {
2943 for (t++; t < PL_bufend && isALNUM_lazy(t); t++) ;
2945 while (t < PL_bufend && isSPACE(*t))
2947 /* if comma follows first term, call it an anon hash */
2948 /* XXX it could be a comma expression with loop modifiers */
2949 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
2950 || (*t == '=' && t[1] == '>')))
2951 OPERATOR(HASHBRACK);
2952 if (PL_expect == XREF)
2955 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
2961 yylval.ival = PL_curcop->cop_line;
2962 if (isSPACE(*s) || *s == '#')
2963 PL_copline = NOLINE; /* invalidate current command line number */
2968 if (PL_lex_brackets <= 0)
2969 yyerror("Unmatched right curly bracket");
2971 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
2972 if (PL_lex_brackets < PL_lex_formbrack)
2973 PL_lex_formbrack = 0;
2974 if (PL_lex_state == LEX_INTERPNORMAL) {
2975 if (PL_lex_brackets == 0) {
2976 if (PL_lex_fakebrack) {
2977 PL_lex_state = LEX_INTERPEND;
2979 return yylex(); /* ignore fake brackets */
2981 if (*s == '-' && s[1] == '>')
2982 PL_lex_state = LEX_INTERPENDMAYBE;
2983 else if (*s != '[' && *s != '{')
2984 PL_lex_state = LEX_INTERPEND;
2987 if (PL_lex_brackets < PL_lex_fakebrack) {
2989 PL_lex_fakebrack = 0;
2990 return yylex(); /* ignore fake brackets */
3000 if (PL_expect == XOPERATOR) {
3001 if (ckWARN(WARN_SEMICOLON) && isIDFIRST_lazy(s) && PL_bufptr == PL_linestart) {
3002 PL_curcop->cop_line--;
3003 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3004 PL_curcop->cop_line++;
3009 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3011 PL_expect = XOPERATOR;
3012 force_ident(PL_tokenbuf, '&');
3016 yylval.ival = (OPpENTERSUB_AMPER<<8);
3035 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3036 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3038 if (PL_expect == XSTATE && isALPHA(tmp) &&
3039 (s == PL_linestart+1 || s[-2] == '\n') )
3041 if (PL_in_eval && !PL_rsfp) {
3046 if (strnEQ(s,"=cut",4)) {
3060 PL_doextract = TRUE;
3063 if (PL_lex_brackets < PL_lex_formbrack) {
3065 #ifdef PERL_STRICT_CR
3066 for (t = s; *t == ' ' || *t == '\t'; t++) ;
3068 for (t = s; *t == ' ' || *t == '\t' || *t == '\r'; t++) ;
3070 if (*t == '\n' || *t == '#') {
3088 if (PL_expect != XOPERATOR) {
3089 if (s[1] != '<' && !strchr(s,'>'))
3092 s = scan_heredoc(s);
3094 s = scan_inputsymbol(s);
3095 TERM(sublex_start());
3100 SHop(OP_LEFT_SHIFT);
3114 SHop(OP_RIGHT_SHIFT);
3123 if (PL_expect == XOPERATOR) {
3124 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3127 return ','; /* grandfather non-comma-format format */
3131 if (s[1] == '#' && (isIDFIRST_lazy(s+2) || strchr("{$:+-", s[2]))) {
3132 PL_tokenbuf[0] = '@';
3133 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3134 sizeof PL_tokenbuf - 1, FALSE);
3135 if (PL_expect == XOPERATOR)
3136 no_op("Array length", s);
3137 if (!PL_tokenbuf[1])
3139 PL_expect = XOPERATOR;
3140 PL_pending_ident = '#';
3144 PL_tokenbuf[0] = '$';
3145 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3146 sizeof PL_tokenbuf - 1, FALSE);
3147 if (PL_expect == XOPERATOR)
3149 if (!PL_tokenbuf[1]) {
3151 yyerror("Final $ should be \\$ or $name");
3155 /* This kludge not intended to be bulletproof. */
3156 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3157 yylval.opval = newSVOP(OP_CONST, 0,
3158 newSViv((IV)PL_compiling.cop_arybase));
3159 yylval.opval->op_private = OPpCONST_ARYBASE;
3165 if (PL_lex_state == LEX_NORMAL)
3168 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3171 PL_tokenbuf[0] = '@';
3172 if (ckWARN(WARN_SYNTAX)) {
3174 isSPACE(*t) || isALNUM_lazy(t) || *t == '$';
3177 PL_bufptr = skipspace(PL_bufptr);
3178 while (t < PL_bufend && *t != ']')
3180 Perl_warner(aTHX_ WARN_SYNTAX,
3181 "Multidimensional syntax %.*s not supported",
3182 (t - PL_bufptr) + 1, PL_bufptr);
3186 else if (*s == '{') {
3187 PL_tokenbuf[0] = '%';
3188 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3189 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3191 char tmpbuf[sizeof PL_tokenbuf];
3193 for (t++; isSPACE(*t); t++) ;
3194 if (isIDFIRST_lazy(t)) {
3195 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3196 for (; isSPACE(*t); t++) ;
3197 if (*t == ';' && get_cv(tmpbuf, FALSE))
3198 Perl_warner(aTHX_ WARN_SYNTAX,
3199 "You need to quote \"%s\"", tmpbuf);
3205 PL_expect = XOPERATOR;
3206 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3207 bool islop = (PL_last_lop == PL_oldoldbufptr);
3208 if (!islop || PL_last_lop_op == OP_GREPSTART)
3209 PL_expect = XOPERATOR;
3210 else if (strchr("$@\"'`q", *s))
3211 PL_expect = XTERM; /* e.g. print $fh "foo" */
3212 else if (strchr("&*<%", *s) && isIDFIRST_lazy(s+1))
3213 PL_expect = XTERM; /* e.g. print $fh &sub */
3214 else if (isIDFIRST_lazy(s)) {
3215 char tmpbuf[sizeof PL_tokenbuf];
3216 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3217 if (tmp = keyword(tmpbuf, len)) {
3218 /* binary operators exclude handle interpretations */
3230 PL_expect = XTERM; /* e.g. print $fh length() */
3235 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3236 if (gv && GvCVu(gv))
3237 PL_expect = XTERM; /* e.g. print $fh subr() */
3240 else if (isDIGIT(*s))
3241 PL_expect = XTERM; /* e.g. print $fh 3 */
3242 else if (*s == '.' && isDIGIT(s[1]))
3243 PL_expect = XTERM; /* e.g. print $fh .3 */
3244 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3245 PL_expect = XTERM; /* e.g. print $fh -1 */
3246 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3247 PL_expect = XTERM; /* print $fh <<"EOF" */
3249 PL_pending_ident = '$';
3253 if (PL_expect == XOPERATOR)
3255 PL_tokenbuf[0] = '@';
3256 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3257 if (!PL_tokenbuf[1]) {
3259 yyerror("Final @ should be \\@ or @name");
3262 if (PL_lex_state == LEX_NORMAL)
3264 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3266 PL_tokenbuf[0] = '%';
3268 /* Warn about @ where they meant $. */
3269 if (ckWARN(WARN_SYNTAX)) {
3270 if (*s == '[' || *s == '{') {
3272 while (*t && (isALNUM_lazy(t) || strchr(" \t$#+-'\"", *t)))
3274 if (*t == '}' || *t == ']') {
3276 PL_bufptr = skipspace(PL_bufptr);
3277 Perl_warner(aTHX_ WARN_SYNTAX,
3278 "Scalar value %.*s better written as $%.*s",
3279 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3284 PL_pending_ident = '@';
3287 case '/': /* may either be division or pattern */
3288 case '?': /* may either be conditional or pattern */
3289 if (PL_expect != XOPERATOR) {
3290 /* Disable warning on "study /blah/" */
3291 if (PL_oldoldbufptr == PL_last_uni
3292 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3293 || memNE(PL_last_uni, "study", 5) || isALNUM_lazy(PL_last_uni+5)))
3295 s = scan_pat(s,OP_MATCH);
3296 TERM(sublex_start());
3304 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3305 #ifdef PERL_STRICT_CR
3308 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3310 && (s == PL_linestart || s[-1] == '\n') )
3312 PL_lex_formbrack = 0;
3316 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3322 yylval.ival = OPf_SPECIAL;
3328 if (PL_expect != XOPERATOR)
3333 case '0': case '1': case '2': case '3': case '4':
3334 case '5': case '6': case '7': case '8': case '9':
3336 if (PL_expect == XOPERATOR)
3341 s = scan_str(s,FALSE,FALSE);
3342 if (PL_expect == XOPERATOR) {
3343 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3346 return ','; /* grandfather non-comma-format format */
3352 missingterm((char*)0);
3353 yylval.ival = OP_CONST;
3354 TERM(sublex_start());
3357 s = scan_str(s,FALSE,FALSE);
3358 if (PL_expect == XOPERATOR) {
3359 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3362 return ','; /* grandfather non-comma-format format */
3368 missingterm((char*)0);
3369 yylval.ival = OP_CONST;
3370 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3371 if (*d == '$' || *d == '@' || *d == '\\' || *d & 0x80) {
3372 yylval.ival = OP_STRINGIFY;
3376 TERM(sublex_start());
3379 s = scan_str(s,FALSE,FALSE);
3380 if (PL_expect == XOPERATOR)
3381 no_op("Backticks",s);
3383 missingterm((char*)0);
3384 yylval.ival = OP_BACKTICK;
3386 TERM(sublex_start());
3390 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3391 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3393 if (PL_expect == XOPERATOR)
3394 no_op("Backslash",s);
3398 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3438 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3440 /* Some keywords can be followed by any delimiter, including ':' */
3441 tmp = (len == 1 && strchr("msyq", PL_tokenbuf[0]) ||
3442 len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3443 (PL_tokenbuf[0] == 'q' &&
3444 strchr("qwxr", PL_tokenbuf[1]))));
3446 /* x::* is just a word, unless x is "CORE" */
3447 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3451 while (d < PL_bufend && isSPACE(*d))
3452 d++; /* no comments skipped here, or s### is misparsed */
3454 /* Is this a label? */
3455 if (!tmp && PL_expect == XSTATE
3456 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3458 yylval.pval = savepv(PL_tokenbuf);
3463 /* Check for keywords */
3464 tmp = keyword(PL_tokenbuf, len);
3466 /* Is this a word before a => operator? */
3467 if (strnEQ(d,"=>",2)) {
3469 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3470 yylval.opval->op_private = OPpCONST_BARE;
3474 if (tmp < 0) { /* second-class keyword? */
3475 GV *ogv = Nullgv; /* override (winner) */
3476 GV *hgv = Nullgv; /* hidden (loser) */
3477 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3479 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3482 if (GvIMPORTED_CV(gv))
3484 else if (! CvMETHOD(cv))
3488 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3489 (gv = *gvp) != (GV*)&PL_sv_undef &&
3490 GvCVu(gv) && GvIMPORTED_CV(gv))
3496 tmp = 0; /* overridden by import or by GLOBAL */
3499 && -tmp==KEY_lock /* XXX generalizable kludge */
3500 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3502 tmp = 0; /* any sub overrides "weak" keyword */
3504 else { /* no override */
3508 if (ckWARN(WARN_AMBIGUOUS) && hgv
3509 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3510 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3511 "Ambiguous call resolved as CORE::%s(), %s",
3512 GvENAME(hgv), "qualify as such or use &");
3519 default: /* not a keyword */
3522 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3524 /* Get the rest if it looks like a package qualifier */
3526 if (*s == '\'' || *s == ':' && s[1] == ':') {
3528 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3531 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3532 *s == '\'' ? "'" : "::");
3536 if (PL_expect == XOPERATOR) {
3537 if (PL_bufptr == PL_linestart) {
3538 PL_curcop->cop_line--;
3539 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3540 PL_curcop->cop_line++;
3543 no_op("Bareword",s);
3546 /* Look for a subroutine with this name in current package,
3547 unless name is "Foo::", in which case Foo is a bearword
3548 (and a package name). */
3551 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3553 if (ckWARN(WARN_UNSAFE) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3554 Perl_warner(aTHX_ WARN_UNSAFE,
3555 "Bareword \"%s\" refers to nonexistent package",
3558 PL_tokenbuf[len] = '\0';
3565 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3568 /* if we saw a global override before, get the right name */
3571 sv = newSVpvn("CORE::GLOBAL::",14);
3572 sv_catpv(sv,PL_tokenbuf);
3575 sv = newSVpv(PL_tokenbuf,0);
3577 /* Presume this is going to be a bareword of some sort. */
3580 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3581 yylval.opval->op_private = OPpCONST_BARE;
3583 /* And if "Foo::", then that's what it certainly is. */
3588 /* See if it's the indirect object for a list operator. */
3590 if (PL_oldoldbufptr &&
3591 PL_oldoldbufptr < PL_bufptr &&
3592 (PL_oldoldbufptr == PL_last_lop
3593 || PL_oldoldbufptr == PL_last_uni) &&
3594 /* NO SKIPSPACE BEFORE HERE! */
3595 (PL_expect == XREF ||
3596 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3598 bool immediate_paren = *s == '(';
3600 /* (Now we can afford to cross potential line boundary.) */
3603 /* Two barewords in a row may indicate method call. */
3605 if ((isIDFIRST_lazy(s) || *s == '$') && (tmp=intuit_method(s,gv)))
3608 /* If not a declared subroutine, it's an indirect object. */
3609 /* (But it's an indir obj regardless for sort.) */
3611 if ((PL_last_lop_op == OP_SORT ||
3612 (!immediate_paren && (!gv || !GvCVu(gv)))) &&
3613 (PL_last_lop_op != OP_MAPSTART &&
3614 PL_last_lop_op != OP_GREPSTART))
3616 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3621 /* If followed by a paren, it's certainly a subroutine. */
3623 PL_expect = XOPERATOR;
3627 if (gv && GvCVu(gv)) {
3628 for (d = s + 1; *d == ' ' || *d == '\t'; d++) ;
3629 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3634 PL_nextval[PL_nexttoke].opval = yylval.opval;
3635 PL_expect = XOPERATOR;
3641 /* If followed by var or block, call it a method (unless sub) */
3643 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3644 PL_last_lop = PL_oldbufptr;
3645 PL_last_lop_op = OP_METHOD;
3649 /* If followed by a bareword, see if it looks like indir obj. */
3651 if ((isIDFIRST_lazy(s) || *s == '$') && (tmp = intuit_method(s,gv)))
3654 /* Not a method, so call it a subroutine (if defined) */
3656 if (gv && GvCVu(gv)) {
3658 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
3659 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3660 "Ambiguous use of -%s resolved as -&%s()",
3661 PL_tokenbuf, PL_tokenbuf);
3662 /* Check for a constant sub */
3664 if ((sv = cv_const_sv(cv))) {
3666 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
3667 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
3668 yylval.opval->op_private = 0;
3672 /* Resolve to GV now. */
3673 op_free(yylval.opval);
3674 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
3675 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
3676 PL_last_lop = PL_oldbufptr;
3677 PL_last_lop_op = OP_ENTERSUB;
3678 /* Is there a prototype? */
3681 char *proto = SvPV((SV*)cv, len);
3684 if (strEQ(proto, "$"))
3686 if (*proto == '&' && *s == '{') {
3687 sv_setpv(PL_subname,"__ANON__");
3691 PL_nextval[PL_nexttoke].opval = yylval.opval;
3697 /* Call it a bare word */
3699 if (PL_hints & HINT_STRICT_SUBS)
3700 yylval.opval->op_private |= OPpCONST_STRICT;
3703 if (ckWARN(WARN_RESERVED)) {
3704 if (lastchar != '-') {
3705 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
3707 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
3714 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
3715 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3716 "Operator or semicolon missing before %c%s",
3717 lastchar, PL_tokenbuf);
3718 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3719 "Ambiguous use of %c resolved as operator %c",
3720 lastchar, lastchar);
3726 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3727 newSVsv(CopFILESV(PL_curcop)));
3731 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3732 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)PL_curcop->cop_line));
3735 case KEY___PACKAGE__:
3736 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3738 ? newSVsv(PL_curstname)
3747 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
3748 char *pname = "main";
3749 if (PL_tokenbuf[2] == 'D')
3750 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
3751 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
3754 GvIOp(gv) = newIO();
3755 IoIFP(GvIOp(gv)) = PL_rsfp;
3756 #if defined(HAS_FCNTL) && defined(F_SETFD)
3758 int fd = PerlIO_fileno(PL_rsfp);
3759 fcntl(fd,F_SETFD,fd >= 3);
3762 /* Mark this internal pseudo-handle as clean */
3763 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
3765 IoTYPE(GvIOp(gv)) = '|';
3766 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
3767 IoTYPE(GvIOp(gv)) = '-';
3769 IoTYPE(GvIOp(gv)) = '<';
3781 if (PL_expect == XSTATE) {
3788 if (*s == ':' && s[1] == ':') {
3791 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3792 tmp = keyword(PL_tokenbuf, len);
3806 LOP(OP_ACCEPT,XTERM);
3812 LOP(OP_ATAN2,XTERM);
3821 LOP(OP_BLESS,XTERM);
3830 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
3847 if (!PL_cryptseen) {
3848 PL_cryptseen = TRUE;
3852 LOP(OP_CRYPT,XTERM);
3855 if (ckWARN(WARN_OCTAL)) {
3856 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
3857 if (*d != '0' && isDIGIT(*d))
3858 Perl_warner(aTHX_ WARN_OCTAL,
3859 "chmod: mode argument is missing initial 0");
3861 LOP(OP_CHMOD,XTERM);
3864 LOP(OP_CHOWN,XTERM);
3867 LOP(OP_CONNECT,XTERM);
3883 s = force_word(s,WORD,FALSE,TRUE,FALSE);
3887 PL_hints |= HINT_BLOCK_SCOPE;
3897 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
3898 LOP(OP_DBMOPEN,XTERM);
3904 s = force_word(s,WORD,TRUE,FALSE,FALSE);
3911 yylval.ival = PL_curcop->cop_line;
3925 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
3926 UNIBRACK(OP_ENTEREVAL);
3941 case KEY_endhostent:
3947 case KEY_endservent:
3950 case KEY_endprotoent:
3961 yylval.ival = PL_curcop->cop_line;
3963 if (PL_expect == XSTATE && isIDFIRST_lazy(s)) {
3965 if ((PL_bufend - p) >= 3 &&
3966 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
3968 else if ((PL_bufend - p) >= 4 &&
3969 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
3972 if (isIDFIRST_lazy(p)) {
3973 p = scan_ident(p, PL_bufend,
3974 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3978 Perl_croak(aTHX_ "Missing $ on loop variable");
3983 LOP(OP_FORMLINE,XTERM);
3989 LOP(OP_FCNTL,XTERM);
3995 LOP(OP_FLOCK,XTERM);
4004 LOP(OP_GREPSTART, *s == '(' ? XTERM : XREF);
4007 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4022 case KEY_getpriority:
4023 LOP(OP_GETPRIORITY,XTERM);
4025 case KEY_getprotobyname:
4028 case KEY_getprotobynumber:
4029 LOP(OP_GPBYNUMBER,XTERM);
4031 case KEY_getprotoent:
4043 case KEY_getpeername:
4044 UNI(OP_GETPEERNAME);
4046 case KEY_gethostbyname:
4049 case KEY_gethostbyaddr:
4050 LOP(OP_GHBYADDR,XTERM);
4052 case KEY_gethostent:
4055 case KEY_getnetbyname:
4058 case KEY_getnetbyaddr:
4059 LOP(OP_GNBYADDR,XTERM);
4064 case KEY_getservbyname:
4065 LOP(OP_GSBYNAME,XTERM);
4067 case KEY_getservbyport:
4068 LOP(OP_GSBYPORT,XTERM);
4070 case KEY_getservent:
4073 case KEY_getsockname:
4074 UNI(OP_GETSOCKNAME);
4076 case KEY_getsockopt:
4077 LOP(OP_GSOCKOPT,XTERM);
4099 yylval.ival = PL_curcop->cop_line;
4103 LOP(OP_INDEX,XTERM);
4109 LOP(OP_IOCTL,XTERM);
4121 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4153 LOP(OP_LISTEN,XTERM);
4162 s = scan_pat(s,OP_MATCH);
4163 TERM(sublex_start());
4166 LOP(OP_MAPSTART, *s == '(' ? XTERM : XREF);
4169 LOP(OP_MKDIR,XTERM);
4172 LOP(OP_MSGCTL,XTERM);
4175 LOP(OP_MSGGET,XTERM);
4178 LOP(OP_MSGRCV,XTERM);
4181 LOP(OP_MSGSND,XTERM);
4187 if (isIDFIRST_lazy(s)) {
4188 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4189 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4191 PL_in_my_stash = gv_stashpv(PL_tokenbuf, FALSE);
4192 if (!PL_in_my_stash) {
4195 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4203 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4210 if (PL_expect != XSTATE)
4211 yyerror("\"no\" not allowed in expression");
4212 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4213 s = force_version(s);
4218 if (*s == '(' || (s = skipspace(s), *s == '('))
4225 if (isIDFIRST_lazy(s)) {
4227 for (d = s; isALNUM_lazy(d); d++) ;
4229 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_AMBIGUOUS))
4230 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4231 "Precedence problem: open %.*s should be open(%.*s)",
4237 yylval.ival = OP_OR;
4247 LOP(OP_OPEN_DIR,XTERM);
4250 checkcomma(s,PL_tokenbuf,"filehandle");
4254 checkcomma(s,PL_tokenbuf,"filehandle");
4273 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4277 LOP(OP_PIPE_OP,XTERM);
4280 s = scan_str(s,FALSE,FALSE);
4282 missingterm((char*)0);
4283 yylval.ival = OP_CONST;
4284 TERM(sublex_start());
4290 s = scan_str(s,FALSE,FALSE);
4292 missingterm((char*)0);
4294 if (SvCUR(PL_lex_stuff)) {
4297 d = SvPV_force(PL_lex_stuff, len);
4299 for (; isSPACE(*d) && len; --len, ++d) ;
4302 if (!warned && ckWARN(WARN_SYNTAX)) {
4303 for (; !isSPACE(*d) && len; --len, ++d) {
4305 Perl_warner(aTHX_ WARN_SYNTAX,
4306 "Possible attempt to separate words with commas");
4309 else if (*d == '#') {
4310 Perl_warner(aTHX_ WARN_SYNTAX,
4311 "Possible attempt to put comments in qw() list");
4317 for (; !isSPACE(*d) && len; --len, ++d) ;
4319 words = append_elem(OP_LIST, words,
4320 newSVOP(OP_CONST, 0, newSVpvn(b, d-b)));
4324 PL_nextval[PL_nexttoke].opval = words;
4329 SvREFCNT_dec(PL_lex_stuff);
4330 PL_lex_stuff = Nullsv;
4335 s = scan_str(s,FALSE,FALSE);
4337 missingterm((char*)0);
4338 yylval.ival = OP_STRINGIFY;
4339 if (SvIVX(PL_lex_stuff) == '\'')
4340 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4341 TERM(sublex_start());
4344 s = scan_pat(s,OP_QR);
4345 TERM(sublex_start());
4348 s = scan_str(s,FALSE,FALSE);
4350 missingterm((char*)0);
4351 yylval.ival = OP_BACKTICK;
4353 TERM(sublex_start());
4359 *PL_tokenbuf = '\0';
4360 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4361 if (isIDFIRST_lazy(PL_tokenbuf))
4362 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4364 yyerror("<> should be quotes");
4371 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4375 LOP(OP_RENAME,XTERM);
4384 LOP(OP_RINDEX,XTERM);
4407 LOP(OP_REVERSE,XTERM);
4418 TERM(sublex_start());
4420 TOKEN(1); /* force error */
4429 LOP(OP_SELECT,XTERM);
4435 LOP(OP_SEMCTL,XTERM);
4438 LOP(OP_SEMGET,XTERM);
4441 LOP(OP_SEMOP,XTERM);
4447 LOP(OP_SETPGRP,XTERM);
4449 case KEY_setpriority:
4450 LOP(OP_SETPRIORITY,XTERM);
4452 case KEY_sethostent:
4458 case KEY_setservent:
4461 case KEY_setprotoent:
4471 LOP(OP_SEEKDIR,XTERM);
4473 case KEY_setsockopt:
4474 LOP(OP_SSOCKOPT,XTERM);
4480 LOP(OP_SHMCTL,XTERM);
4483 LOP(OP_SHMGET,XTERM);
4486 LOP(OP_SHMREAD,XTERM);
4489 LOP(OP_SHMWRITE,XTERM);
4492 LOP(OP_SHUTDOWN,XTERM);
4501 LOP(OP_SOCKET,XTERM);
4503 case KEY_socketpair:
4504 LOP(OP_SOCKPAIR,XTERM);
4507 checkcomma(s,PL_tokenbuf,"subroutine name");
4509 if (*s == ';' || *s == ')') /* probably a close */
4510 Perl_croak(aTHX_ "sort is now a reserved word");
4512 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4516 LOP(OP_SPLIT,XTERM);
4519 LOP(OP_SPRINTF,XTERM);
4522 LOP(OP_SPLICE,XTERM);
4537 LOP(OP_SUBSTR,XTERM);
4543 char tmpbuf[sizeof PL_tokenbuf];
4545 expectation attrful;
4546 bool have_name, have_proto;
4551 if (isIDFIRST_lazy(s) || *s == '\'' ||
4552 (*s == ':' && s[1] == ':'))
4555 attrful = XATTRBLOCK;
4556 /* remember buffer pos'n for later force_word */
4557 tboffset = s - PL_oldbufptr;
4558 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4559 if (strchr(tmpbuf, ':'))
4560 sv_setpv(PL_subname, tmpbuf);
4562 sv_setsv(PL_subname,PL_curstname);
4563 sv_catpvn(PL_subname,"::",2);
4564 sv_catpvn(PL_subname,tmpbuf,len);
4571 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4572 PL_expect = XTERMBLOCK;
4573 attrful = XATTRTERM;
4574 sv_setpv(PL_subname,"?");
4578 if (key == KEY_format) {
4580 PL_lex_formbrack = PL_lex_brackets + 1;
4582 (void) force_word(PL_oldbufptr + tboffset, WORD,
4587 /* Look for a prototype */
4591 s = scan_str(s,FALSE,FALSE);
4594 SvREFCNT_dec(PL_lex_stuff);
4595 PL_lex_stuff = Nullsv;
4596 Perl_croak(aTHX_ "Prototype not terminated");
4599 d = SvPVX(PL_lex_stuff);
4601 for (p = d; *p; ++p) {
4606 SvCUR(PL_lex_stuff) = tmp;
4614 if (*s == ':' && s[1] != ':')
4615 PL_expect = attrful;
4618 PL_nextval[PL_nexttoke].opval =
4619 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
4620 PL_lex_stuff = Nullsv;
4624 sv_setpv(PL_subname,"__ANON__");
4627 (void) force_word(PL_oldbufptr + tboffset, WORD,
4636 LOP(OP_SYSTEM,XREF);
4639 LOP(OP_SYMLINK,XTERM);
4642 LOP(OP_SYSCALL,XTERM);
4645 LOP(OP_SYSOPEN,XTERM);
4648 LOP(OP_SYSSEEK,XTERM);
4651 LOP(OP_SYSREAD,XTERM);
4654 LOP(OP_SYSWRITE,XTERM);
4658 TERM(sublex_start());
4679 LOP(OP_TRUNCATE,XTERM);
4691 yylval.ival = PL_curcop->cop_line;
4695 yylval.ival = PL_curcop->cop_line;
4699 LOP(OP_UNLINK,XTERM);
4705 LOP(OP_UNPACK,XTERM);
4708 LOP(OP_UTIME,XTERM);
4711 if (ckWARN(WARN_OCTAL)) {
4712 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4713 if (*d != '0' && isDIGIT(*d))
4714 Perl_warner(aTHX_ WARN_OCTAL,
4715 "umask: argument is missing initial 0");
4720 LOP(OP_UNSHIFT,XTERM);
4723 if (PL_expect != XSTATE)
4724 yyerror("\"use\" not allowed in expression");
4727 s = force_version(s);
4728 if(*s == ';' || (s = skipspace(s), *s == ';')) {
4729 PL_nextval[PL_nexttoke].opval = Nullop;
4734 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4735 s = force_version(s);
4747 yylval.ival = PL_curcop->cop_line;
4751 PL_hints |= HINT_BLOCK_SCOPE;
4758 LOP(OP_WAITPID,XTERM);
4766 static char ctl_l[2];
4768 if (ctl_l[0] == '\0')
4769 ctl_l[0] = toCTRL('L');
4770 gv_fetchpv(ctl_l,TRUE, SVt_PV);
4773 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
4778 if (PL_expect == XOPERATOR)
4784 yylval.ival = OP_XOR;
4789 TERM(sublex_start());
4795 Perl_keyword(pTHX_ register char *d, I32 len)
4800 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
4801 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
4802 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
4803 if (strEQ(d,"__DATA__")) return KEY___DATA__;
4804 if (strEQ(d,"__END__")) return KEY___END__;
4808 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
4813 if (strEQ(d,"and")) return -KEY_and;
4814 if (strEQ(d,"abs")) return -KEY_abs;
4817 if (strEQ(d,"alarm")) return -KEY_alarm;
4818 if (strEQ(d,"atan2")) return -KEY_atan2;
4821 if (strEQ(d,"accept")) return -KEY_accept;
4826 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
4829 if (strEQ(d,"bless")) return -KEY_bless;
4830 if (strEQ(d,"bind")) return -KEY_bind;
4831 if (strEQ(d,"binmode")) return -KEY_binmode;
4834 if (strEQ(d,"CORE")) return -KEY_CORE;
4839 if (strEQ(d,"cmp")) return -KEY_cmp;
4840 if (strEQ(d,"chr")) return -KEY_chr;
4841 if (strEQ(d,"cos")) return -KEY_cos;
4844 if (strEQ(d,"chop")) return KEY_chop;
4847 if (strEQ(d,"close")) return -KEY_close;
4848 if (strEQ(d,"chdir")) return -KEY_chdir;
4849 if (strEQ(d,"chomp")) return KEY_chomp;
4850 if (strEQ(d,"chmod")) return -KEY_chmod;
4851 if (strEQ(d,"chown")) return -KEY_chown;
4852 if (strEQ(d,"crypt")) return -KEY_crypt;
4855 if (strEQ(d,"chroot")) return -KEY_chroot;
4856 if (strEQ(d,"caller")) return -KEY_caller;
4859 if (strEQ(d,"connect")) return -KEY_connect;
4862 if (strEQ(d,"closedir")) return -KEY_closedir;
4863 if (strEQ(d,"continue")) return -KEY_continue;
4868 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
4873 if (strEQ(d,"do")) return KEY_do;
4876 if (strEQ(d,"die")) return -KEY_die;
4879 if (strEQ(d,"dump")) return -KEY_dump;
4882 if (strEQ(d,"delete")) return KEY_delete;
4885 if (strEQ(d,"defined")) return KEY_defined;
4886 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
4889 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
4894 if (strEQ(d,"EQ")) { deprecate(d); return -KEY_eq;}
4895 if (strEQ(d,"END")) return KEY_END;
4900 if (strEQ(d,"eq")) return -KEY_eq;
4903 if (strEQ(d,"eof")) return -KEY_eof;
4904 if (strEQ(d,"exp")) return -KEY_exp;
4907 if (strEQ(d,"else")) return KEY_else;
4908 if (strEQ(d,"exit")) return -KEY_exit;
4909 if (strEQ(d,"eval")) return KEY_eval;
4910 if (strEQ(d,"exec")) return -KEY_exec;
4911 if (strEQ(d,"each")) return KEY_each;
4914 if (strEQ(d,"elsif")) return KEY_elsif;
4917 if (strEQ(d,"exists")) return KEY_exists;
4918 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
4921 if (strEQ(d,"endgrent")) return -KEY_endgrent;
4922 if (strEQ(d,"endpwent")) return -KEY_endpwent;
4925 if (strEQ(d,"endnetent")) return -KEY_endnetent;
4928 if (strEQ(d,"endhostent")) return -KEY_endhostent;
4929 if (strEQ(d,"endservent")) return -KEY_endservent;
4932 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
4939 if (strEQ(d,"for")) return KEY_for;
4942 if (strEQ(d,"fork")) return -KEY_fork;
4945 if (strEQ(d,"fcntl")) return -KEY_fcntl;
4946 if (strEQ(d,"flock")) return -KEY_flock;
4949 if (strEQ(d,"format")) return KEY_format;
4950 if (strEQ(d,"fileno")) return -KEY_fileno;
4953 if (strEQ(d,"foreach")) return KEY_foreach;
4956 if (strEQ(d,"formline")) return -KEY_formline;
4962 if (strEQ(d,"GT")) { deprecate(d); return -KEY_gt;}
4963 if (strEQ(d,"GE")) { deprecate(d); return -KEY_ge;}
4967 if (strnEQ(d,"get",3)) {
4972 if (strEQ(d,"ppid")) return -KEY_getppid;
4973 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
4976 if (strEQ(d,"pwent")) return -KEY_getpwent;
4977 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
4978 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
4981 if (strEQ(d,"peername")) return -KEY_getpeername;
4982 if (strEQ(d,"protoent")) return -KEY_getprotoent;
4983 if (strEQ(d,"priority")) return -KEY_getpriority;
4986 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
4989 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
4993 else if (*d == 'h') {
4994 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
4995 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
4996 if (strEQ(d,"hostent")) return -KEY_gethostent;
4998 else if (*d == 'n') {
4999 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5000 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5001 if (strEQ(d,"netent")) return -KEY_getnetent;
5003 else if (*d == 's') {
5004 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5005 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5006 if (strEQ(d,"servent")) return -KEY_getservent;
5007 if (strEQ(d,"sockname")) return -KEY_getsockname;
5008 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5010 else if (*d == 'g') {
5011 if (strEQ(d,"grent")) return -KEY_getgrent;
5012 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5013 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5015 else if (*d == 'l') {
5016 if (strEQ(d,"login")) return -KEY_getlogin;
5018 else if (strEQ(d,"c")) return -KEY_getc;
5023 if (strEQ(d,"gt")) return -KEY_gt;
5024 if (strEQ(d,"ge")) return -KEY_ge;
5027 if (strEQ(d,"grep")) return KEY_grep;
5028 if (strEQ(d,"goto")) return KEY_goto;
5029 if (strEQ(d,"glob")) return KEY_glob;
5032 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5037 if (strEQ(d,"hex")) return -KEY_hex;
5040 if (strEQ(d,"INIT")) return KEY_INIT;
5045 if (strEQ(d,"if")) return KEY_if;
5048 if (strEQ(d,"int")) return -KEY_int;
5051 if (strEQ(d,"index")) return -KEY_index;
5052 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5057 if (strEQ(d,"join")) return -KEY_join;
5061 if (strEQ(d,"keys")) return KEY_keys;
5062 if (strEQ(d,"kill")) return -KEY_kill;
5067 if (strEQ(d,"LT")) { deprecate(d); return -KEY_lt;}
5068 if (strEQ(d,"LE")) { deprecate(d); return -KEY_le;}
5074 if (strEQ(d,"lt")) return -KEY_lt;
5075 if (strEQ(d,"le")) return -KEY_le;
5076 if (strEQ(d,"lc")) return -KEY_lc;
5079 if (strEQ(d,"log")) return -KEY_log;
5082 if (strEQ(d,"last")) return KEY_last;
5083 if (strEQ(d,"link")) return -KEY_link;
5084 if (strEQ(d,"lock")) return -KEY_lock;
5087 if (strEQ(d,"local")) return KEY_local;
5088 if (strEQ(d,"lstat")) return -KEY_lstat;
5091 if (strEQ(d,"length")) return -KEY_length;
5092 if (strEQ(d,"listen")) return -KEY_listen;
5095 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5098 if (strEQ(d,"localtime")) return -KEY_localtime;
5104 case 1: return KEY_m;
5106 if (strEQ(d,"my")) return KEY_my;
5109 if (strEQ(d,"map")) return KEY_map;
5112 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5115 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5116 if (strEQ(d,"msgget")) return -KEY_msgget;
5117 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5118 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5123 if (strEQ(d,"NE")) { deprecate(d); return -KEY_ne;}
5126 if (strEQ(d,"next")) return KEY_next;
5127 if (strEQ(d,"ne")) return -KEY_ne;
5128 if (strEQ(d,"not")) return -KEY_not;
5129 if (strEQ(d,"no")) return KEY_no;
5134 if (strEQ(d,"or")) return -KEY_or;
5137 if (strEQ(d,"ord")) return -KEY_ord;
5138 if (strEQ(d,"oct")) return -KEY_oct;
5139 if (strEQ(d,"our")) return KEY_our;
5142 if (strEQ(d,"open")) return -KEY_open;
5145 if (strEQ(d,"opendir")) return -KEY_opendir;
5152 if (strEQ(d,"pop")) return KEY_pop;
5153 if (strEQ(d,"pos")) return KEY_pos;
5156 if (strEQ(d,"push")) return KEY_push;
5157 if (strEQ(d,"pack")) return -KEY_pack;
5158 if (strEQ(d,"pipe")) return -KEY_pipe;
5161 if (strEQ(d,"print")) return KEY_print;
5164 if (strEQ(d,"printf")) return KEY_printf;
5167 if (strEQ(d,"package")) return KEY_package;
5170 if (strEQ(d,"prototype")) return KEY_prototype;
5175 if (strEQ(d,"q")) return KEY_q;
5176 if (strEQ(d,"qr")) return KEY_qr;
5177 if (strEQ(d,"qq")) return KEY_qq;
5178 if (strEQ(d,"qw")) return KEY_qw;
5179 if (strEQ(d,"qx")) return KEY_qx;
5181 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5186 if (strEQ(d,"ref")) return -KEY_ref;
5189 if (strEQ(d,"read")) return -KEY_read;
5190 if (strEQ(d,"rand")) return -KEY_rand;
5191 if (strEQ(d,"recv")) return -KEY_recv;
5192 if (strEQ(d,"redo")) return KEY_redo;
5195 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5196 if (strEQ(d,"reset")) return -KEY_reset;
5199 if (strEQ(d,"return")) return KEY_return;
5200 if (strEQ(d,"rename")) return -KEY_rename;
5201 if (strEQ(d,"rindex")) return -KEY_rindex;
5204 if (strEQ(d,"require")) return -KEY_require;
5205 if (strEQ(d,"reverse")) return -KEY_reverse;
5206 if (strEQ(d,"readdir")) return -KEY_readdir;
5209 if (strEQ(d,"readlink")) return -KEY_readlink;
5210 if (strEQ(d,"readline")) return -KEY_readline;
5211 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5214 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5219 if (strEQ(d,"STOP")) return KEY_STOP;
5223 case 0: return KEY_s;
5225 if (strEQ(d,"scalar")) return KEY_scalar;
5230 if (strEQ(d,"seek")) return -KEY_seek;
5231 if (strEQ(d,"send")) return -KEY_send;
5234 if (strEQ(d,"semop")) return -KEY_semop;
5237 if (strEQ(d,"select")) return -KEY_select;
5238 if (strEQ(d,"semctl")) return -KEY_semctl;
5239 if (strEQ(d,"semget")) return -KEY_semget;
5242 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5243 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5246 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5247 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5250 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5253 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5254 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5255 if (strEQ(d,"setservent")) return -KEY_setservent;
5258 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5259 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5266 if (strEQ(d,"shift")) return KEY_shift;
5269 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5270 if (strEQ(d,"shmget")) return -KEY_shmget;
5273 if (strEQ(d,"shmread")) return -KEY_shmread;
5276 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5277 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5282 if (strEQ(d,"sin")) return -KEY_sin;
5285 if (strEQ(d,"sleep")) return -KEY_sleep;
5288 if (strEQ(d,"sort")) return KEY_sort;
5289 if (strEQ(d,"socket")) return -KEY_socket;
5290 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5293 if (strEQ(d,"split")) return KEY_split;
5294 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5295 if (strEQ(d,"splice")) return KEY_splice;
5298 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5301 if (strEQ(d,"srand")) return -KEY_srand;
5304 if (strEQ(d,"stat")) return -KEY_stat;
5305 if (strEQ(d,"study")) return KEY_study;
5308 if (strEQ(d,"substr")) return -KEY_substr;
5309 if (strEQ(d,"sub")) return KEY_sub;
5314 if (strEQ(d,"system")) return -KEY_system;
5317 if (strEQ(d,"symlink")) return -KEY_symlink;
5318 if (strEQ(d,"syscall")) return -KEY_syscall;
5319 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5320 if (strEQ(d,"sysread")) return -KEY_sysread;
5321 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5324 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5333 if (strEQ(d,"tr")) return KEY_tr;
5336 if (strEQ(d,"tie")) return KEY_tie;
5339 if (strEQ(d,"tell")) return -KEY_tell;
5340 if (strEQ(d,"tied")) return KEY_tied;
5341 if (strEQ(d,"time")) return -KEY_time;
5344 if (strEQ(d,"times")) return -KEY_times;
5347 if (strEQ(d,"telldir")) return -KEY_telldir;
5350 if (strEQ(d,"truncate")) return -KEY_truncate;
5357 if (strEQ(d,"uc")) return -KEY_uc;
5360 if (strEQ(d,"use")) return KEY_use;
5363 if (strEQ(d,"undef")) return KEY_undef;
5364 if (strEQ(d,"until")) return KEY_until;
5365 if (strEQ(d,"untie")) return KEY_untie;
5366 if (strEQ(d,"utime")) return -KEY_utime;
5367 if (strEQ(d,"umask")) return -KEY_umask;
5370 if (strEQ(d,"unless")) return KEY_unless;
5371 if (strEQ(d,"unpack")) return -KEY_unpack;
5372 if (strEQ(d,"unlink")) return -KEY_unlink;
5375 if (strEQ(d,"unshift")) return KEY_unshift;
5376 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5381 if (strEQ(d,"values")) return -KEY_values;
5382 if (strEQ(d,"vec")) return -KEY_vec;
5387 if (strEQ(d,"warn")) return -KEY_warn;
5388 if (strEQ(d,"wait")) return -KEY_wait;
5391 if (strEQ(d,"while")) return KEY_while;
5392 if (strEQ(d,"write")) return -KEY_write;
5395 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5398 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5403 if (len == 1) return -KEY_x;
5404 if (strEQ(d,"xor")) return -KEY_xor;
5407 if (len == 1) return KEY_y;
5416 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5420 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5421 dTHR; /* only for ckWARN */
5422 if (ckWARN(WARN_SYNTAX)) {
5424 for (w = s+2; *w && level; w++) {
5431 for (; *w && isSPACE(*w); w++) ;
5432 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5433 Perl_warner(aTHX_ WARN_SYNTAX,
5434 "%s (...) interpreted as function",name);
5437 while (s < PL_bufend && isSPACE(*s))
5441 while (s < PL_bufend && isSPACE(*s))
5443 if (isIDFIRST_lazy(s)) {
5445 while (isALNUM_lazy(s))
5447 while (s < PL_bufend && isSPACE(*s))
5452 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5456 Perl_croak(aTHX_ "No comma allowed after %s", what);
5461 /* Either returns sv, or mortalizes sv and returns a new SV*.
5462 Best used as sv=new_constant(..., sv, ...).
5463 If s, pv are NULL, calls subroutine with one argument,
5464 and type is used with error messages only. */
5467 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5471 HV *table = GvHV(PL_hintgv); /* ^H */
5475 const char *why, *why1, *why2;
5477 if (!(PL_hints & HINT_LOCALIZE_HH)) {
5480 why = "%^H is not localized";
5484 msg = Perl_newSVpvf(aTHX_ "constant(%s): %s%s%s",
5485 (type ? type: "undef"), why1, why2, why);
5486 yyerror(SvPVX(msg));
5491 why = "%^H is not defined";
5494 cvp = hv_fetch(table, key, strlen(key), FALSE);
5495 if (!cvp || !SvOK(*cvp)) {
5496 why = "} is not defined";
5501 sv_2mortal(sv); /* Parent created it permanently */
5504 pv = sv_2mortal(newSVpvn(s, len));
5506 typesv = sv_2mortal(newSVpv(type, 0));
5508 typesv = &PL_sv_undef;
5510 PUSHSTACKi(PERLSI_OVERLOAD);
5523 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5527 /* Check the eval first */
5528 if (!PL_in_eval && SvTRUE(ERRSV))
5531 sv_catpv(ERRSV, "Propagated");
5532 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5534 res = SvREFCNT_inc(sv);
5538 (void)SvREFCNT_inc(res);
5547 why = "}} did not return a defined value";
5548 why1 = "Call to &{$^H{";
5558 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
5560 register char *d = dest;
5561 register char *e = d + destlen - 3; /* two-character token, ending NUL */
5564 Perl_croak(aTHX_ ident_too_long);
5565 if (isALNUM(*s)) /* UTF handled below */
5567 else if (*s == '\'' && allow_package && isIDFIRST_lazy(s+1)) {
5572 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
5576 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5577 char *t = s + UTF8SKIP(s);
5578 while (*t & 0x80 && is_utf8_mark((U8*)t))
5580 if (d + (t - s) > e)
5581 Perl_croak(aTHX_ ident_too_long);
5582 Copy(s, d, t - s, char);
5595 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
5602 if (PL_lex_brackets == 0)
5603 PL_lex_fakebrack = 0;
5607 e = d + destlen - 3; /* two-character token, ending NUL */
5609 while (isDIGIT(*s)) {
5611 Perl_croak(aTHX_ ident_too_long);
5618 Perl_croak(aTHX_ ident_too_long);
5619 if (isALNUM(*s)) /* UTF handled below */
5621 else if (*s == '\'' && isIDFIRST_lazy(s+1)) {
5626 else if (*s == ':' && s[1] == ':') {
5630 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5631 char *t = s + UTF8SKIP(s);
5632 while (*t & 0x80 && is_utf8_mark((U8*)t))
5634 if (d + (t - s) > e)
5635 Perl_croak(aTHX_ ident_too_long);
5636 Copy(s, d, t - s, char);
5647 if (PL_lex_state != LEX_NORMAL)
5648 PL_lex_state = LEX_INTERPENDMAYBE;
5651 if (*s == '$' && s[1] &&
5652 (isALNUM_lazy(s+1) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
5665 if (*d == '^' && *s && isCONTROLVAR(*s)) {
5670 if (isSPACE(s[-1])) {
5673 if (ch != ' ' && ch != '\t') {
5679 if (isIDFIRST_lazy(d)) {
5683 while (e < send && isALNUM_lazy(e) || *e == ':') {
5685 while (e < send && *e & 0x80 && is_utf8_mark((U8*)e))
5688 Copy(s, d, e - s, char);
5693 while ((isALNUM(*s) || *s == ':') && d < e)
5696 Perl_croak(aTHX_ ident_too_long);
5699 while (s < send && (*s == ' ' || *s == '\t')) s++;
5700 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
5701 dTHR; /* only for ckWARN */
5702 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
5703 const char *brack = *s == '[' ? "[...]" : "{...}";
5704 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5705 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
5706 funny, dest, brack, funny, dest, brack);
5708 PL_lex_fakebrack = PL_lex_brackets+1;
5710 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
5714 /* Handle extended ${^Foo} variables
5715 * 1999-02-27 mjd-perl-patch@plover.com */
5716 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
5720 while (isALNUM(*s) && d < e) {
5724 Perl_croak(aTHX_ ident_too_long);
5729 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
5730 PL_lex_state = LEX_INTERPEND;
5733 if (PL_lex_state == LEX_NORMAL) {
5734 dTHR; /* only for ckWARN */
5735 if (ckWARN(WARN_AMBIGUOUS) &&
5736 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
5738 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5739 "Ambiguous use of %c{%s} resolved to %c%s",
5740 funny, dest, funny, dest);
5745 s = bracket; /* let the parser handle it */
5749 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
5750 PL_lex_state = LEX_INTERPEND;
5755 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
5760 *pmfl |= PMf_GLOBAL;
5762 *pmfl |= PMf_CONTINUE;
5766 *pmfl |= PMf_MULTILINE;
5768 *pmfl |= PMf_SINGLELINE;
5770 *pmfl |= PMf_EXTENDED;
5774 S_scan_pat(pTHX_ char *start, I32 type)
5779 s = scan_str(start,FALSE,FALSE);
5782 SvREFCNT_dec(PL_lex_stuff);
5783 PL_lex_stuff = Nullsv;
5784 Perl_croak(aTHX_ "Search pattern not terminated");
5787 pm = (PMOP*)newPMOP(type, 0);
5788 if (PL_multi_open == '?')
5789 pm->op_pmflags |= PMf_ONCE;
5791 while (*s && strchr("iomsx", *s))
5792 pmflag(&pm->op_pmflags,*s++);
5795 while (*s && strchr("iogcmsx", *s))
5796 pmflag(&pm->op_pmflags,*s++);
5798 pm->op_pmpermflags = pm->op_pmflags;
5800 PL_lex_op = (OP*)pm;
5801 yylval.ival = OP_MATCH;
5806 S_scan_subst(pTHX_ char *start)
5813 yylval.ival = OP_NULL;
5815 s = scan_str(start,FALSE,FALSE);
5819 SvREFCNT_dec(PL_lex_stuff);
5820 PL_lex_stuff = Nullsv;
5821 Perl_croak(aTHX_ "Substitution pattern not terminated");
5824 if (s[-1] == PL_multi_open)
5827 first_start = PL_multi_start;
5828 s = scan_str(s,FALSE,FALSE);
5831 SvREFCNT_dec(PL_lex_stuff);
5832 PL_lex_stuff = Nullsv;
5834 SvREFCNT_dec(PL_lex_repl);
5835 PL_lex_repl = Nullsv;
5836 Perl_croak(aTHX_ "Substitution replacement not terminated");
5838 PL_multi_start = first_start; /* so whole substitution is taken together */
5840 pm = (PMOP*)newPMOP(OP_SUBST, 0);
5846 else if (strchr("iogcmsx", *s))
5847 pmflag(&pm->op_pmflags,*s++);
5854 PL_sublex_info.super_bufptr = s;
5855 PL_sublex_info.super_bufend = PL_bufend;
5857 pm->op_pmflags |= PMf_EVAL;
5858 repl = newSVpvn("",0);
5860 sv_catpv(repl, es ? "eval " : "do ");
5861 sv_catpvn(repl, "{ ", 2);
5862 sv_catsv(repl, PL_lex_repl);
5863 sv_catpvn(repl, " };", 2);
5865 SvREFCNT_dec(PL_lex_repl);
5869 pm->op_pmpermflags = pm->op_pmflags;
5870 PL_lex_op = (OP*)pm;
5871 yylval.ival = OP_SUBST;
5876 S_scan_trans(pTHX_ char *start)
5887 yylval.ival = OP_NULL;
5889 s = scan_str(start,FALSE,FALSE);
5892 SvREFCNT_dec(PL_lex_stuff);
5893 PL_lex_stuff = Nullsv;
5894 Perl_croak(aTHX_ "Transliteration pattern not terminated");
5896 if (s[-1] == PL_multi_open)
5899 s = scan_str(s,FALSE,FALSE);
5902 SvREFCNT_dec(PL_lex_stuff);
5903 PL_lex_stuff = Nullsv;
5905 SvREFCNT_dec(PL_lex_repl);
5906 PL_lex_repl = Nullsv;
5907 Perl_croak(aTHX_ "Transliteration replacement not terminated");
5911 o = newSVOP(OP_TRANS, 0, 0);
5912 utf8 = OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF;
5915 New(803,tbl,256,short);
5916 o = newPVOP(OP_TRANS, 0, (char*)tbl);
5920 complement = del = squash = 0;
5921 while (strchr("cdsCU", *s)) {
5923 complement = OPpTRANS_COMPLEMENT;
5925 del = OPpTRANS_DELETE;
5927 squash = OPpTRANS_SQUASH;
5932 utf8 &= ~OPpTRANS_FROM_UTF;
5934 utf8 |= OPpTRANS_FROM_UTF;
5938 utf8 &= ~OPpTRANS_TO_UTF;
5940 utf8 |= OPpTRANS_TO_UTF;
5943 Perl_croak(aTHX_ "Too many /C and /U options");
5948 o->op_private = del|squash|complement|utf8;
5951 yylval.ival = OP_TRANS;
5956 S_scan_heredoc(pTHX_ register char *s)
5960 I32 op_type = OP_SCALAR;
5967 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
5971 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
5974 for (peek = s; *peek == ' ' || *peek == '\t'; peek++) ;
5975 if (*peek && strchr("`'\"",*peek)) {
5978 s = delimcpy(d, e, s, PL_bufend, term, &len);
5988 if (!isALNUM_lazy(s))
5989 deprecate("bare << to mean <<\"\"");
5990 for (; isALNUM_lazy(s); s++) {
5995 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
5996 Perl_croak(aTHX_ "Delimiter for here document is too long");
5999 len = d - PL_tokenbuf;
6000 #ifndef PERL_STRICT_CR
6001 d = strchr(s, '\r');
6005 while (s < PL_bufend) {
6011 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6020 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6025 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6026 herewas = newSVpvn(s,PL_bufend-s);
6028 s--, herewas = newSVpvn(s,d-s);
6029 s += SvCUR(herewas);
6031 tmpstr = NEWSV(87,79);
6032 sv_upgrade(tmpstr, SVt_PVIV);
6037 else if (term == '`') {
6038 op_type = OP_BACKTICK;
6039 SvIVX(tmpstr) = '\\';
6043 PL_multi_start = PL_curcop->cop_line;
6044 PL_multi_open = PL_multi_close = '<';
6045 term = *PL_tokenbuf;
6046 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6047 char *bufptr = PL_sublex_info.super_bufptr;
6048 char *bufend = PL_sublex_info.super_bufend;
6049 char *olds = s - SvCUR(herewas);
6050 s = strchr(bufptr, '\n');
6054 while (s < bufend &&
6055 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6057 PL_curcop->cop_line++;
6060 PL_curcop->cop_line = PL_multi_start;
6061 missingterm(PL_tokenbuf);
6063 sv_setpvn(herewas,bufptr,d-bufptr+1);
6064 sv_setpvn(tmpstr,d+1,s-d);
6066 sv_catpvn(herewas,s,bufend-s);
6067 (void)strcpy(bufptr,SvPVX(herewas));
6074 while (s < PL_bufend &&
6075 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6077 PL_curcop->cop_line++;
6079 if (s >= PL_bufend) {
6080 PL_curcop->cop_line = PL_multi_start;
6081 missingterm(PL_tokenbuf);
6083 sv_setpvn(tmpstr,d+1,s-d);
6085 PL_curcop->cop_line++; /* the preceding stmt passes a newline */
6087 sv_catpvn(herewas,s,PL_bufend-s);
6088 sv_setsv(PL_linestr,herewas);
6089 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6090 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6093 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6094 while (s >= PL_bufend) { /* multiple line string? */
6096 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6097 PL_curcop->cop_line = PL_multi_start;
6098 missingterm(PL_tokenbuf);
6100 PL_curcop->cop_line++;
6101 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6102 #ifndef PERL_STRICT_CR
6103 if (PL_bufend - PL_linestart >= 2) {
6104 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6105 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6107 PL_bufend[-2] = '\n';
6109 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6111 else if (PL_bufend[-1] == '\r')
6112 PL_bufend[-1] = '\n';
6114 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6115 PL_bufend[-1] = '\n';
6117 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6118 SV *sv = NEWSV(88,0);
6120 sv_upgrade(sv, SVt_PVMG);
6121 sv_setsv(sv,PL_linestr);
6122 av_store(CopFILEAV(PL_curcop), (I32)PL_curcop->cop_line,sv);
6124 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6127 sv_catsv(PL_linestr,herewas);
6128 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6132 sv_catsv(tmpstr,PL_linestr);
6137 PL_multi_end = PL_curcop->cop_line;
6138 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6139 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6140 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6142 SvREFCNT_dec(herewas);
6143 PL_lex_stuff = tmpstr;
6144 yylval.ival = op_type;
6149 takes: current position in input buffer
6150 returns: new position in input buffer
6151 side-effects: yylval and lex_op are set.
6156 <FH> read from filehandle
6157 <pkg::FH> read from package qualified filehandle
6158 <pkg'FH> read from package qualified filehandle
6159 <$fh> read from filehandle in $fh
6165 S_scan_inputsymbol(pTHX_ char *start)
6167 register char *s = start; /* current position in buffer */
6173 d = PL_tokenbuf; /* start of temp holding space */
6174 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6175 end = strchr(s, '\n');
6178 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6180 /* die if we didn't have space for the contents of the <>,
6181 or if it didn't end, or if we see a newline
6184 if (len >= sizeof PL_tokenbuf)
6185 Perl_croak(aTHX_ "Excessively long <> operator");
6187 Perl_croak(aTHX_ "Unterminated <> operator");
6192 Remember, only scalar variables are interpreted as filehandles by
6193 this code. Anything more complex (e.g., <$fh{$num}>) will be
6194 treated as a glob() call.
6195 This code makes use of the fact that except for the $ at the front,
6196 a scalar variable and a filehandle look the same.
6198 if (*d == '$' && d[1]) d++;
6200 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6201 while (*d && (isALNUM_lazy(d) || *d == '\'' || *d == ':'))
6204 /* If we've tried to read what we allow filehandles to look like, and
6205 there's still text left, then it must be a glob() and not a getline.
6206 Use scan_str to pull out the stuff between the <> and treat it
6207 as nothing more than a string.
6210 if (d - PL_tokenbuf != len) {
6211 yylval.ival = OP_GLOB;
6213 s = scan_str(start,FALSE,FALSE);
6215 Perl_croak(aTHX_ "Glob not terminated");
6219 /* we're in a filehandle read situation */
6222 /* turn <> into <ARGV> */
6224 (void)strcpy(d,"ARGV");
6226 /* if <$fh>, create the ops to turn the variable into a
6232 /* try to find it in the pad for this block, otherwise find
6233 add symbol table ops
6235 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6236 OP *o = newOP(OP_PADSV, 0);
6238 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o);
6241 GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV);
6242 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0,
6243 newUNOP(OP_RV2SV, 0,
6244 newGVOP(OP_GV, 0, gv)));
6246 PL_lex_op->op_flags |= OPf_SPECIAL;
6247 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6248 yylval.ival = OP_NULL;
6251 /* If it's none of the above, it must be a literal filehandle
6252 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6254 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6255 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6256 yylval.ival = OP_NULL;
6265 takes: start position in buffer
6266 keep_quoted preserve \ on the embedded delimiter(s)
6267 keep_delims preserve the delimiters around the string
6268 returns: position to continue reading from buffer
6269 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6270 updates the read buffer.
6272 This subroutine pulls a string out of the input. It is called for:
6273 q single quotes q(literal text)
6274 ' single quotes 'literal text'
6275 qq double quotes qq(interpolate $here please)
6276 " double quotes "interpolate $here please"
6277 qx backticks qx(/bin/ls -l)
6278 ` backticks `/bin/ls -l`
6279 qw quote words @EXPORT_OK = qw( func() $spam )
6280 m// regexp match m/this/
6281 s/// regexp substitute s/this/that/
6282 tr/// string transliterate tr/this/that/
6283 y/// string transliterate y/this/that/
6284 ($*@) sub prototypes sub foo ($)
6285 (stuff) sub attr parameters sub foo : attr(stuff)
6286 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6288 In most of these cases (all but <>, patterns and transliterate)
6289 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6290 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6291 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6294 It skips whitespace before the string starts, and treats the first
6295 character as the delimiter. If the delimiter is one of ([{< then
6296 the corresponding "close" character )]}> is used as the closing
6297 delimiter. It allows quoting of delimiters, and if the string has
6298 balanced delimiters ([{<>}]) it allows nesting.
6300 The lexer always reads these strings into lex_stuff, except in the
6301 case of the operators which take *two* arguments (s/// and tr///)
6302 when it checks to see if lex_stuff is full (presumably with the 1st
6303 arg to s or tr) and if so puts the string into lex_repl.
6308 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6311 SV *sv; /* scalar value: string */
6312 char *tmps; /* temp string, used for delimiter matching */
6313 register char *s = start; /* current position in the buffer */
6314 register char term; /* terminating character */
6315 register char *to; /* current position in the sv's data */
6316 I32 brackets = 1; /* bracket nesting level */
6318 /* skip space before the delimiter */
6322 /* mark where we are, in case we need to report errors */
6325 /* after skipping whitespace, the next character is the terminator */
6327 /* mark where we are */
6328 PL_multi_start = PL_curcop->cop_line;
6329 PL_multi_open = term;
6331 /* find corresponding closing delimiter */
6332 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6334 PL_multi_close = term;
6336 /* create a new SV to hold the contents. 87 is leak category, I'm
6337 assuming. 79 is the SV's initial length. What a random number. */
6339 sv_upgrade(sv, SVt_PVIV);
6341 (void)SvPOK_only(sv); /* validate pointer */
6343 /* move past delimiter and try to read a complete string */
6345 sv_catpvn(sv, s, 1);
6348 /* extend sv if need be */
6349 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6350 /* set 'to' to the next character in the sv's string */
6351 to = SvPVX(sv)+SvCUR(sv);
6353 /* if open delimiter is the close delimiter read unbridle */
6354 if (PL_multi_open == PL_multi_close) {
6355 for (; s < PL_bufend; s++,to++) {
6356 /* embedded newlines increment the current line number */
6357 if (*s == '\n' && !PL_rsfp)
6358 PL_curcop->cop_line++;
6359 /* handle quoted delimiters */
6360 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6361 if (!keep_quoted && s[1] == term)
6363 /* any other quotes are simply copied straight through */
6367 /* terminate when run out of buffer (the for() condition), or
6368 have found the terminator */
6369 else if (*s == term)
6375 /* if the terminator isn't the same as the start character (e.g.,
6376 matched brackets), we have to allow more in the quoting, and
6377 be prepared for nested brackets.
6380 /* read until we run out of string, or we find the terminator */
6381 for (; s < PL_bufend; s++,to++) {
6382 /* embedded newlines increment the line count */
6383 if (*s == '\n' && !PL_rsfp)
6384 PL_curcop->cop_line++;
6385 /* backslashes can escape the open or closing characters */
6386 if (*s == '\\' && s+1 < PL_bufend) {
6388 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6393 /* allow nested opens and closes */
6394 else if (*s == PL_multi_close && --brackets <= 0)
6396 else if (*s == PL_multi_open)
6401 /* terminate the copied string and update the sv's end-of-string */
6403 SvCUR_set(sv, to - SvPVX(sv));
6406 * this next chunk reads more into the buffer if we're not done yet
6409 if (s < PL_bufend) break; /* handle case where we are done yet :-) */
6411 #ifndef PERL_STRICT_CR
6412 if (to - SvPVX(sv) >= 2) {
6413 if ((to[-2] == '\r' && to[-1] == '\n') ||
6414 (to[-2] == '\n' && to[-1] == '\r'))
6418 SvCUR_set(sv, to - SvPVX(sv));
6420 else if (to[-1] == '\r')
6423 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6427 /* if we're out of file, or a read fails, bail and reset the current
6428 line marker so we can report where the unterminated string began
6431 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6433 PL_curcop->cop_line = PL_multi_start;
6436 /* we read a line, so increment our line counter */
6437 PL_curcop->cop_line++;
6439 /* update debugger info */
6440 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6441 SV *sv = NEWSV(88,0);
6443 sv_upgrade(sv, SVt_PVMG);
6444 sv_setsv(sv,PL_linestr);
6445 av_store(CopFILEAV(PL_curcop), (I32)PL_curcop->cop_line, sv);
6448 /* having changed the buffer, we must update PL_bufend */
6449 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6452 /* at this point, we have successfully read the delimited string */
6455 sv_catpvn(sv, s, 1);
6456 PL_multi_end = PL_curcop->cop_line;
6459 /* if we allocated too much space, give some back */
6460 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6461 SvLEN_set(sv, SvCUR(sv) + 1);
6462 Renew(SvPVX(sv), SvLEN(sv), char);
6465 /* decide whether this is the first or second quoted string we've read
6478 takes: pointer to position in buffer
6479 returns: pointer to new position in buffer
6480 side-effects: builds ops for the constant in yylval.op
6482 Read a number in any of the formats that Perl accepts:
6484 0(x[0-7A-F]+)|([0-7]+)|(b[01])
6485 [\d_]+(\.[\d_]*)?[Ee](\d+)
6487 Underbars (_) are allowed in decimal numbers. If -w is on,
6488 underbars before a decimal point must be at three digit intervals.
6490 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6493 If it reads a number without a decimal point or an exponent, it will
6494 try converting the number to an integer and see if it can do so
6495 without loss of precision.
6499 Perl_scan_num(pTHX_ char *start)
6501 register char *s = start; /* current position in buffer */
6502 register char *d; /* destination in temp buffer */
6503 register char *e; /* end of temp buffer */
6504 IV tryiv; /* used to see if it can be an IV */
6505 NV value; /* number read, as a double */
6506 SV *sv; /* place to put the converted number */
6507 bool floatit; /* boolean: int or float? */
6508 char *lastub = 0; /* position of last underbar */
6509 static char number_too_long[] = "Number too long";
6511 /* We use the first character to decide what type of number this is */
6515 Perl_croak(aTHX_ "panic: scan_num");
6517 /* if it starts with a 0, it could be an octal number, a decimal in
6518 0.13 disguise, or a hexadecimal number, or a binary number.
6523 u holds the "number so far"
6524 shift the power of 2 of the base
6525 (hex == 4, octal == 3, binary == 1)
6526 overflowed was the number more than we can hold?
6528 Shift is used when we add a digit. It also serves as an "are
6529 we in octal/hex/binary?" indicator to disallow hex characters
6536 bool overflowed = FALSE;
6537 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6538 static char* bases[5] = { "", "binary", "", "octal",
6540 static char* Bases[5] = { "", "Binary", "", "Octal",
6542 static char *maxima[5] = { "",
6543 "0b11111111111111111111111111111111",
6547 char *base, *Base, *max;
6553 } else if (s[1] == 'b') {
6557 /* check for a decimal in disguise */
6558 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
6560 /* so it must be octal */
6564 base = bases[shift];
6565 Base = Bases[shift];
6566 max = maxima[shift];
6568 /* read the rest of the number */
6570 /* x is used in the overflow test,
6571 b is the digit we're adding on. */
6576 /* if we don't mention it, we're done */
6585 /* 8 and 9 are not octal */
6588 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
6592 case '2': case '3': case '4':
6593 case '5': case '6': case '7':
6595 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
6599 b = *s++ & 15; /* ASCII digit -> value of digit */
6603 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
6604 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
6605 /* make sure they said 0x */
6610 /* Prepare to put the digit we have onto the end
6611 of the number so far. We check for overflows.
6616 x = u << shift; /* make room for the digit */
6618 if ((x >> shift) != u
6619 && !(PL_hints & HINT_NEW_BINARY)) {
6623 if (ckWARN_d(WARN_OVERFLOW))
6624 Perl_warner(aTHX_ WARN_OVERFLOW,
6625 "Integer overflow in %s number",
6628 u = x | b; /* add the digit to the end */
6631 n *= nvshift[shift];
6632 /* If an NV has not enough bits in its
6633 * mantissa to represent an UV this summing of
6634 * small low-order numbers is a waste of time
6635 * (because the NV cannot preserve the
6636 * low-order bits anyway): we could just
6637 * remember when did we overflow and in the
6638 * end just multiply n by the right
6646 /* if we get here, we had success: make a scalar value from
6653 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
6654 Perl_warner(aTHX_ WARN_PORTABLE,
6655 "%s number > %s non-portable",
6662 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
6663 Perl_warner(aTHX_ WARN_PORTABLE,
6664 "%s number > %s non-portable",
6669 if (PL_hints & HINT_NEW_BINARY)
6670 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
6675 handle decimal numbers.
6676 we're also sent here when we read a 0 as the first digit
6678 case '1': case '2': case '3': case '4': case '5':
6679 case '6': case '7': case '8': case '9': case '.':
6682 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
6685 /* read next group of digits and _ and copy into d */
6686 while (isDIGIT(*s) || *s == '_') {
6687 /* skip underscores, checking for misplaced ones
6691 dTHR; /* only for ckWARN */
6692 if (ckWARN(WARN_SYNTAX) && lastub && s - lastub != 3)
6693 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6697 /* check for end of fixed-length buffer */
6699 Perl_croak(aTHX_ number_too_long);
6700 /* if we're ok, copy the character */
6705 /* final misplaced underbar check */
6706 if (lastub && s - lastub != 3) {
6708 if (ckWARN(WARN_SYNTAX))
6709 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6712 /* read a decimal portion if there is one. avoid
6713 3..5 being interpreted as the number 3. followed
6716 if (*s == '.' && s[1] != '.') {
6720 /* copy, ignoring underbars, until we run out of
6721 digits. Note: no misplaced underbar checks!
6723 for (; isDIGIT(*s) || *s == '_'; s++) {
6724 /* fixed length buffer check */
6726 Perl_croak(aTHX_ number_too_long);
6732 /* read exponent part, if present */
6733 if (*s && strchr("eE",*s) && strchr("+-0123456789",s[1])) {
6737 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
6738 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
6740 /* allow positive or negative exponent */
6741 if (*s == '+' || *s == '-')
6744 /* read digits of exponent (no underbars :-) */
6745 while (isDIGIT(*s)) {
6747 Perl_croak(aTHX_ number_too_long);
6752 /* terminate the string */
6755 /* make an sv from the string */
6758 value = Atof(PL_tokenbuf);
6761 See if we can make do with an integer value without loss of
6762 precision. We use I_V to cast to an int, because some
6763 compilers have issues. Then we try casting it back and see
6764 if it was the same. We only do this if we know we
6765 specifically read an integer.
6767 Note: if floatit is true, then we don't need to do the
6771 if (!floatit && (NV)tryiv == value)
6772 sv_setiv(sv, tryiv);
6774 sv_setnv(sv, value);
6775 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
6776 (PL_hints & HINT_NEW_INTEGER) )
6777 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
6778 (floatit ? "float" : "integer"),
6783 /* make the op for the constant and return */
6785 yylval.opval = newSVOP(OP_CONST, 0, sv);
6791 S_scan_formline(pTHX_ register char *s)
6796 SV *stuff = newSVpvn("",0);
6797 bool needargs = FALSE;
6800 if (*s == '.' || *s == '}') {
6802 #ifdef PERL_STRICT_CR
6803 for (t = s+1;*t == ' ' || *t == '\t'; t++) ;
6805 for (t = s+1;*t == ' ' || *t == '\t' || *t == '\r'; t++) ;
6807 if (*t == '\n' || t == PL_bufend)
6810 if (PL_in_eval && !PL_rsfp) {
6811 eol = strchr(s,'\n');
6816 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6818 for (t = s; t < eol; t++) {
6819 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
6821 goto enough; /* ~~ must be first line in formline */
6823 if (*t == '@' || *t == '^')
6826 sv_catpvn(stuff, s, eol-s);
6830 s = filter_gets(PL_linestr, PL_rsfp, 0);
6831 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
6832 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
6835 yyerror("Format not terminated");
6845 PL_lex_state = LEX_NORMAL;
6846 PL_nextval[PL_nexttoke].ival = 0;
6850 PL_lex_state = LEX_FORMLINE;
6851 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
6853 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
6857 SvREFCNT_dec(stuff);
6858 PL_lex_formbrack = 0;
6869 PL_cshlen = strlen(PL_cshname);
6874 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
6877 I32 oldsavestack_ix = PL_savestack_ix;
6878 CV* outsidecv = PL_compcv;
6882 assert(SvTYPE(PL_compcv) == SVt_PVCV);
6884 save_I32(&PL_subline);
6885 save_item(PL_subname);
6887 SAVESPTR(PL_curpad);
6888 SAVESPTR(PL_comppad);
6889 SAVESPTR(PL_comppad_name);
6890 SAVESPTR(PL_compcv);
6891 SAVEI32(PL_comppad_name_fill);
6892 SAVEI32(PL_min_intro_pending);
6893 SAVEI32(PL_max_intro_pending);
6894 SAVEI32(PL_pad_reset_pending);
6896 PL_compcv = (CV*)NEWSV(1104,0);
6897 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
6898 CvFLAGS(PL_compcv) |= flags;
6900 PL_comppad = newAV();
6901 av_push(PL_comppad, Nullsv);
6902 PL_curpad = AvARRAY(PL_comppad);
6903 PL_comppad_name = newAV();
6904 PL_comppad_name_fill = 0;
6905 PL_min_intro_pending = 0;
6907 PL_subline = PL_curcop->cop_line;
6909 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
6910 PL_curpad[0] = (SV*)newAV();
6911 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
6912 #endif /* USE_THREADS */
6914 comppadlist = newAV();
6915 AvREAL_off(comppadlist);
6916 av_store(comppadlist, 0, (SV*)PL_comppad_name);
6917 av_store(comppadlist, 1, (SV*)PL_comppad);
6919 CvPADLIST(PL_compcv) = comppadlist;
6920 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
6922 CvOWNER(PL_compcv) = 0;
6923 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
6924 MUTEX_INIT(CvMUTEXP(PL_compcv));
6925 #endif /* USE_THREADS */
6927 return oldsavestack_ix;
6931 Perl_yywarn(pTHX_ char *s)
6934 PL_in_eval |= EVAL_WARNONLY;
6936 PL_in_eval &= ~EVAL_WARNONLY;
6941 Perl_yyerror(pTHX_ char *s)
6945 char *context = NULL;
6949 if (!yychar || (yychar == ';' && !PL_rsfp))
6951 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
6952 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
6953 while (isSPACE(*PL_oldoldbufptr))
6955 context = PL_oldoldbufptr;
6956 contlen = PL_bufptr - PL_oldoldbufptr;
6958 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
6959 PL_oldbufptr != PL_bufptr) {
6960 while (isSPACE(*PL_oldbufptr))
6962 context = PL_oldbufptr;
6963 contlen = PL_bufptr - PL_oldbufptr;
6965 else if (yychar > 255)
6966 where = "next token ???";
6967 else if ((yychar & 127) == 127) {
6968 if (PL_lex_state == LEX_NORMAL ||
6969 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
6970 where = "at end of line";
6971 else if (PL_lex_inpat)
6972 where = "within pattern";
6974 where = "within string";
6977 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
6979 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
6980 else if (isPRINT_LC(yychar))
6981 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
6983 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
6984 where = SvPVX(where_sv);
6986 msg = sv_2mortal(newSVpv(s, 0));
6987 Perl_sv_catpvf(aTHX_ msg, " at %_ line %"IVdf", ",
6988 CopFILESV(PL_curcop), (IV)PL_curcop->cop_line);
6990 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
6992 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
6993 if (PL_multi_start < PL_multi_end && (U32)(PL_curcop->cop_line - PL_multi_end) <= 1) {
6994 Perl_sv_catpvf(aTHX_ msg,
6995 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
6996 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
6999 if (PL_in_eval & EVAL_WARNONLY)
7000 Perl_warn(aTHX_ "%_", msg);
7003 if (PL_error_count >= 10)
7004 Perl_croak(aTHX_ "%_ has too many errors.\n", CopFILESV(PL_curcop));
7006 PL_in_my_stash = Nullhv;
7017 * Restore a source filter.
7021 restore_rsfp(pTHXo_ void *f)
7023 PerlIO *fp = (PerlIO*)f;
7025 if (PL_rsfp == PerlIO_stdin())
7026 PerlIO_clearerr(PL_rsfp);
7027 else if (PL_rsfp && (PL_rsfp != fp))
7028 PerlIO_close(PL_rsfp);
7034 * Restores the state of PL_expect when the lexing that begun with a
7035 * start_lex() call has ended.
7039 restore_expect(pTHXo_ void *e)
7041 /* a safe way to store a small integer in a pointer */
7042 PL_expect = (expectation)((char *)e - PL_tokenbuf);
7046 * restore_lex_expect
7047 * Restores the state of PL_lex_expect when the lexing that begun with a
7048 * start_lex() call has ended.
7052 restore_lex_expect(pTHXo_ void *e)
7054 /* a safe way to store a small integer in a pointer */
7055 PL_lex_expect = (expectation)((char *)e - PL_tokenbuf);