3 * Copyright (c) 1991-1999, Larry Wall
5 * You may distribute under the terms of either the GNU General Public
6 * License or the Artistic License, as specified in the README file.
11 * "It all comes from here, the stench and the peril." --Frodo
15 * This file is the lexer for Perl. It's closely linked to the
18 * The main routine is yylex(), which returns the next token.
22 #define PERL_IN_TOKE_C
25 #define yychar PL_yychar
26 #define yylval PL_yylval
28 static char ident_too_long[] = "Identifier too long";
30 static void restore_rsfp(pTHXo_ void *f);
31 static void restore_expect(pTHXo_ void *e);
32 static void restore_lex_expect(pTHXo_ void *e);
34 #define UTF (PL_hints & HINT_UTF8)
36 * Note: we try to be careful never to call the isXXX_utf8() functions
37 * unless we're pretty sure we've seen the beginning of a UTF-8 character
38 * (that is, the two high bits are set). Otherwise we risk loading in the
39 * heavy-duty SWASHINIT and SWASHGET routines unnecessarily.
41 #define isIDFIRST_lazy(p) ((!UTF || (*((U8*)p) < 0xc0)) \
43 : isIDFIRST_utf8((U8*)p))
44 #define isALNUM_lazy(p) ((!UTF || (*((U8*)p) < 0xc0)) \
46 : isALNUM_utf8((U8*)p))
48 /* In variables name $^X, these are the legal values for X.
49 * 1999-02-27 mjd-perl-patch@plover.com */
50 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
52 /* LEX_* are values for PL_lex_state, the state of the lexer.
53 * They are arranged oddly so that the guard on the switch statement
54 * can get by with a single comparison (if the compiler is smart enough).
57 /* #define LEX_NOTPARSING 11 is done in perl.h. */
60 #define LEX_INTERPNORMAL 9
61 #define LEX_INTERPCASEMOD 8
62 #define LEX_INTERPPUSH 7
63 #define LEX_INTERPSTART 6
64 #define LEX_INTERPEND 5
65 #define LEX_INTERPENDMAYBE 4
66 #define LEX_INTERPCONCAT 3
67 #define LEX_INTERPCONST 2
68 #define LEX_FORMLINE 1
69 #define LEX_KNOWNEXT 0
78 /* XXX If this causes problems, set i_unistd=undef in the hint file. */
80 # include <unistd.h> /* Needed for execv() */
89 YYSTYPE* yylval_pointer = NULL;
90 int* yychar_pointer = NULL;
93 # define yylval (*yylval_pointer)
94 # define yychar (*yychar_pointer)
95 # define PERL_YYLEX_PARAM yylval_pointer,yychar_pointer
97 # define yylex() Perl_yylex(aTHX_ yylval_pointer, yychar_pointer)
100 #include "keywords.h"
102 /* CLINE is a macro that ensures PL_copline has a sane value */
107 #define CLINE (PL_copline = (PL_curcop->cop_line < PL_copline ? PL_curcop->cop_line : PL_copline))
110 * Convenience functions to return different tokens and prime the
111 * lexer for the next token. They all take an argument.
113 * TOKEN : generic token (used for '(', DOLSHARP, etc)
114 * OPERATOR : generic operator
115 * AOPERATOR : assignment operator
116 * PREBLOCK : beginning the block after an if, while, foreach, ...
117 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
118 * PREREF : *EXPR where EXPR is not a simple identifier
119 * TERM : expression term
120 * LOOPX : loop exiting command (goto, last, dump, etc)
121 * FTST : file test operator
122 * FUN0 : zero-argument function
124 * BOop : bitwise or or xor
126 * SHop : shift operator
127 * PWop : power operator
128 * PMop : pattern-matching operator
129 * Aop : addition-level operator
130 * Mop : multiplication-level operator
131 * Eop : equality-testing operator
132 * Rop : relational operator <= != gt
134 * Also see LOP and lop() below.
137 #define TOKEN(retval) return (PL_bufptr = s,(int)retval)
138 #define OPERATOR(retval) return (PL_expect = XTERM,PL_bufptr = s,(int)retval)
139 #define AOPERATOR(retval) return ao((PL_expect = XTERM,PL_bufptr = s,(int)retval))
140 #define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s,(int)retval)
141 #define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s,(int)retval)
142 #define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s,(int)retval)
143 #define TERM(retval) return (CLINE, PL_expect = XOPERATOR,PL_bufptr = s,(int)retval)
144 #define LOOPX(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LOOPEX)
145 #define FTST(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)UNIOP)
146 #define FUN0(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC0)
147 #define FUN1(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC1)
148 #define BOop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITOROP))
149 #define BAop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITANDOP))
150 #define SHop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)SHIFTOP))
151 #define PWop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)POWOP))
152 #define PMop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MATCHOP)
153 #define Aop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)ADDOP))
154 #define Mop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MULOP))
155 #define Eop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)EQOP)
156 #define Rop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)RELOP)
158 /* This bit of chicanery makes a unary function followed by
159 * a parenthesis into a function with one argument, highest precedence.
161 #define UNI(f) return(yylval.ival = f, \
164 PL_last_uni = PL_oldbufptr, \
165 PL_last_lop_op = f, \
166 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
168 #define UNIBRACK(f) return(yylval.ival = f, \
170 PL_last_uni = PL_oldbufptr, \
171 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
173 /* grandfather return to old style */
174 #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
179 * This subroutine detects &&= and ||= and turns an ANDAND or OROR
180 * into an OP_ANDASSIGN or OP_ORASSIGN
184 S_ao(pTHX_ int toketype)
186 if (*PL_bufptr == '=') {
188 if (toketype == ANDAND)
189 yylval.ival = OP_ANDASSIGN;
190 else if (toketype == OROR)
191 yylval.ival = OP_ORASSIGN;
199 * When Perl expects an operator and finds something else, no_op
200 * prints the warning. It always prints "<something> found where
201 * operator expected. It prints "Missing semicolon on previous line?"
202 * if the surprise occurs at the start of the line. "do you need to
203 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
204 * where the compiler doesn't know if foo is a method call or a function.
205 * It prints "Missing operator before end of line" if there's nothing
206 * after the missing operator, or "... before <...>" if there is something
207 * after the missing operator.
211 S_no_op(pTHX_ char *what, char *s)
213 char *oldbp = PL_bufptr;
214 bool is_first = (PL_oldbufptr == PL_linestart);
222 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
224 Perl_warn(aTHX_ "\t(Missing semicolon on previous line?)\n");
225 else if (PL_oldoldbufptr && isIDFIRST_lazy(PL_oldoldbufptr)) {
227 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy(t) || *t == ':'); t++) ;
228 if (t < PL_bufptr && isSPACE(*t))
229 Perl_warn(aTHX_ "\t(Do you need to predeclare %.*s?)\n",
230 t - PL_oldoldbufptr, PL_oldoldbufptr);
233 Perl_warn(aTHX_ "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp);
239 * Complain about missing quote/regexp/heredoc terminator.
240 * If it's called with (char *)NULL then it cauterizes the line buffer.
241 * If we're in a delimited string and the delimiter is a control
242 * character, it's reformatted into a two-char sequence like ^C.
247 S_missingterm(pTHX_ char *s)
252 char *nl = strrchr(s,'\n');
258 iscntrl(PL_multi_close)
260 PL_multi_close < 32 || PL_multi_close == 127
264 tmpbuf[1] = toCTRL(PL_multi_close);
270 *tmpbuf = PL_multi_close;
274 q = strchr(s,'"') ? '\'' : '"';
275 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
283 Perl_deprecate(pTHX_ char *s)
286 if (ckWARN(WARN_DEPRECATED))
287 Perl_warner(aTHX_ WARN_DEPRECATED, "Use of %s is deprecated", s);
292 * Deprecate a comma-less variable list.
298 deprecate("comma-less variable list");
302 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
303 * utf16-to-utf8-reversed.
309 S_win32_textfilter(pTHX_ int idx, SV *sv, int maxlen)
311 I32 count = FILTER_READ(idx+1, sv, maxlen);
312 if (count > 0 && !maxlen)
313 win32_strip_return(sv);
319 S_utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
321 I32 count = FILTER_READ(idx+1, sv, maxlen);
325 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
326 tend = utf16_to_utf8((U16*)SvPVX(sv), tmps, SvCUR(sv));
327 sv_usepvn(sv, (char*)tmps, tend - tmps);
334 S_utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
336 I32 count = FILTER_READ(idx+1, sv, maxlen);
340 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
341 tend = utf16_to_utf8_reversed((U16*)SvPVX(sv), tmps, SvCUR(sv));
342 sv_usepvn(sv, (char*)tmps, tend - tmps);
350 * Initialize variables. Uses the Perl save_stack to save its state (for
351 * recursive calls to the parser).
355 Perl_lex_start(pTHX_ SV *line)
361 SAVEI32(PL_lex_dojoin);
362 SAVEI32(PL_lex_brackets);
363 SAVEI32(PL_lex_fakebrack);
364 SAVEI32(PL_lex_casemods);
365 SAVEI32(PL_lex_starts);
366 SAVEI32(PL_lex_state);
367 SAVESPTR(PL_lex_inpat);
368 SAVEI32(PL_lex_inwhat);
369 SAVEI16(PL_curcop->cop_line);
372 SAVEPPTR(PL_oldbufptr);
373 SAVEPPTR(PL_oldoldbufptr);
374 SAVEPPTR(PL_linestart);
375 SAVESPTR(PL_linestr);
376 SAVEPPTR(PL_lex_brackstack);
377 SAVEPPTR(PL_lex_casestack);
378 SAVEDESTRUCTOR(restore_rsfp, PL_rsfp);
379 SAVESPTR(PL_lex_stuff);
380 SAVEI32(PL_lex_defer);
381 SAVEI32(PL_sublex_info.sub_inwhat);
382 SAVESPTR(PL_lex_repl);
383 SAVEDESTRUCTOR(restore_expect, PL_tokenbuf + PL_expect); /* encode as pointer */
384 SAVEDESTRUCTOR(restore_lex_expect, PL_tokenbuf + PL_expect);
386 PL_lex_state = LEX_NORMAL;
390 PL_lex_fakebrack = 0;
391 New(899, PL_lex_brackstack, 120, char);
392 New(899, PL_lex_casestack, 12, char);
393 SAVEFREEPV(PL_lex_brackstack);
394 SAVEFREEPV(PL_lex_casestack);
396 *PL_lex_casestack = '\0';
399 PL_lex_stuff = Nullsv;
400 PL_lex_repl = Nullsv;
403 PL_sublex_info.sub_inwhat = 0;
405 if (SvREADONLY(PL_linestr))
406 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
407 s = SvPV(PL_linestr, len);
408 if (len && s[len-1] != ';') {
409 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
410 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
411 sv_catpvn(PL_linestr, "\n;", 2);
413 SvTEMP_off(PL_linestr);
414 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
415 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
417 PL_rs = newSVpvn("\n", 1);
423 * Finalizer for lexing operations. Must be called when the parser is
424 * done with the lexer.
430 PL_doextract = FALSE;
435 * This subroutine has nothing to do with tilting, whether at windmills
436 * or pinball tables. Its name is short for "increment line". It
437 * increments the current line number in PL_curcop->cop_line and checks
438 * to see whether the line starts with a comment of the form
439 * # line 500 "foo.pm"
440 * If so, it sets the current line number and file to the values in the comment.
444 S_incline(pTHX_ char *s)
452 PL_curcop->cop_line++;
455 while (*s == ' ' || *s == '\t') s++;
456 if (strnEQ(s, "line ", 5)) {
465 while (*s == ' ' || *s == '\t')
467 if (*s == '"' && (t = strchr(s+1, '"')))
471 return; /* false alarm */
472 for (t = s; !isSPACE(*t); t++) ;
477 PL_curcop->cop_filegv = gv_fetchfile(s);
479 PL_curcop->cop_filegv = gv_fetchfile(PL_origfilename);
481 PL_curcop->cop_line = atoi(n)-1;
486 * Called to gobble the appropriate amount and type of whitespace.
487 * Skips comments as well.
491 S_skipspace(pTHX_ register char *s)
494 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
495 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
501 SSize_t oldprevlen, oldoldprevlen;
502 SSize_t oldloplen, oldunilen;
503 while (s < PL_bufend && isSPACE(*s)) {
504 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
509 if (s < PL_bufend && *s == '#') {
510 while (s < PL_bufend && *s != '\n')
514 if (PL_in_eval && !PL_rsfp) {
521 /* only continue to recharge the buffer if we're at the end
522 * of the buffer, we're not reading from a source filter, and
523 * we're in normal lexing mode
525 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
526 PL_lex_state == LEX_FORMLINE)
529 /* try to recharge the buffer */
530 if ((s = filter_gets(PL_linestr, PL_rsfp,
531 (prevlen = SvCUR(PL_linestr)))) == Nullch)
533 /* end of file. Add on the -p or -n magic */
534 if (PL_minus_n || PL_minus_p) {
535 sv_setpv(PL_linestr,PL_minus_p ?
536 ";}continue{print or die qq(-p destination: $!\\n)" :
538 sv_catpv(PL_linestr,";}");
539 PL_minus_n = PL_minus_p = 0;
542 sv_setpv(PL_linestr,";");
544 /* reset variables for next time we lex */
545 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
547 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
549 /* Close the filehandle. Could be from -P preprocessor,
550 * STDIN, or a regular file. If we were reading code from
551 * STDIN (because the commandline held no -e or filename)
552 * then we don't close it, we reset it so the code can
553 * read from STDIN too.
556 if (PL_preprocess && !PL_in_eval)
557 (void)PerlProc_pclose(PL_rsfp);
558 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
559 PerlIO_clearerr(PL_rsfp);
561 (void)PerlIO_close(PL_rsfp);
566 /* not at end of file, so we only read another line */
567 /* make corresponding updates to old pointers, for yyerror() */
568 oldprevlen = PL_oldbufptr - PL_bufend;
569 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
571 oldunilen = PL_last_uni - PL_bufend;
573 oldloplen = PL_last_lop - PL_bufend;
574 PL_linestart = PL_bufptr = s + prevlen;
575 PL_bufend = s + SvCUR(PL_linestr);
577 PL_oldbufptr = s + oldprevlen;
578 PL_oldoldbufptr = s + oldoldprevlen;
580 PL_last_uni = s + oldunilen;
582 PL_last_lop = s + oldloplen;
585 /* debugger active and we're not compiling the debugger code,
586 * so store the line into the debugger's array of lines
588 if (PERLDB_LINE && PL_curstash != PL_debstash) {
589 SV *sv = NEWSV(85,0);
591 sv_upgrade(sv, SVt_PVMG);
592 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
593 av_store(GvAV(PL_curcop->cop_filegv),(I32)PL_curcop->cop_line,sv);
600 * Check the unary operators to ensure there's no ambiguity in how they're
601 * used. An ambiguous piece of code would be:
603 * This doesn't mean rand() + 5. Because rand() is a unary operator,
604 * the +5 is its argument.
614 if (PL_oldoldbufptr != PL_last_uni)
616 while (isSPACE(*PL_last_uni))
618 for (s = PL_last_uni; isALNUM_lazy(s) || *s == '-'; s++) ;
619 if ((t = strchr(s, '(')) && t < PL_bufptr)
621 if (ckWARN_d(WARN_AMBIGUOUS)){
624 Perl_warner(aTHX_ WARN_AMBIGUOUS,
625 "Warning: Use of \"%s\" without parens is ambiguous",
631 /* workaround to replace the UNI() macro with a function. Only the
632 * hints/uts.sh file mentions this. Other comments elsewhere in the
633 * source indicate Microport Unix might need it too.
639 #define UNI(f) return uni(f,s)
642 S_uni(pTHX_ I32 f, char *s)
647 PL_last_uni = PL_oldbufptr;
658 #endif /* CRIPPLED_CC */
661 * LOP : macro to build a list operator. Its behaviour has been replaced
662 * with a subroutine, S_lop() for which LOP is just another name.
665 #define LOP(f,x) return lop(f,x,s)
669 * Build a list operator (or something that might be one). The rules:
670 * - if we have a next token, then it's a list operator [why?]
671 * - if the next thing is an opening paren, then it's a function
672 * - else it's a list operator
676 S_lop(pTHX_ I32 f, expectation x, char *s)
683 PL_last_lop = PL_oldbufptr;
698 * When the lexer realizes it knows the next token (for instance,
699 * it is reordering tokens for the parser) then it can call S_force_next
700 * to know what token to return the next time the lexer is called. Caller
701 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
702 * handles the token correctly.
706 S_force_next(pTHX_ I32 type)
708 PL_nexttype[PL_nexttoke] = type;
710 if (PL_lex_state != LEX_KNOWNEXT) {
711 PL_lex_defer = PL_lex_state;
712 PL_lex_expect = PL_expect;
713 PL_lex_state = LEX_KNOWNEXT;
719 * When the lexer knows the next thing is a word (for instance, it has
720 * just seen -> and it knows that the next char is a word char, then
721 * it calls S_force_word to stick the next word into the PL_next lookahead.
724 * char *start : buffer position (must be within PL_linestr)
725 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
726 * int check_keyword : if true, Perl checks to make sure the word isn't
727 * a keyword (do this if the word is a label, e.g. goto FOO)
728 * int allow_pack : if true, : characters will also be allowed (require,
730 * int allow_initial_tick : used by the "sub" lexer only.
734 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
739 start = skipspace(start);
741 if (isIDFIRST_lazy(s) ||
742 (allow_pack && *s == ':') ||
743 (allow_initial_tick && *s == '\'') )
745 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
746 if (check_keyword && keyword(PL_tokenbuf, len))
748 if (token == METHOD) {
753 PL_expect = XOPERATOR;
756 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0));
757 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
765 * Called when the lexer wants $foo *foo &foo etc, but the program
766 * text only contains the "foo" portion. The first argument is a pointer
767 * to the "foo", and the second argument is the type symbol to prefix.
768 * Forces the next token to be a "WORD".
769 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
773 S_force_ident(pTHX_ register char *s, int kind)
776 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
777 PL_nextval[PL_nexttoke].opval = o;
780 dTHR; /* just for in_eval */
781 o->op_private = OPpCONST_ENTERED;
782 /* XXX see note in pp_entereval() for why we forgo typo
783 warnings if the symbol must be introduced in an eval.
785 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
786 kind == '$' ? SVt_PV :
787 kind == '@' ? SVt_PVAV :
788 kind == '%' ? SVt_PVHV :
797 * Forces the next token to be a version number.
801 S_force_version(pTHX_ char *s)
803 OP *version = Nullop;
807 /* default VERSION number -- GBARR */
812 for( d=s, c = 1; isDIGIT(*d) || *d == '_' || (*d == '.' && c--); d++);
813 if((*d == ';' || isSPACE(*d)) && *(skipspace(d)) != ',') {
815 /* real VERSION number -- GBARR */
816 version = yylval.opval;
820 /* NOTE: The parser sees the package name and the VERSION swapped */
821 PL_nextval[PL_nexttoke].opval = version;
829 * Tokenize a quoted string passed in as an SV. It finds the next
830 * chunk, up to end of string or a backslash. It may make a new
831 * SV containing that chunk (if HINT_NEW_STRING is on). It also
836 S_tokeq(pTHX_ SV *sv)
847 s = SvPV_force(sv, len);
851 while (s < send && *s != '\\')
856 if ( PL_hints & HINT_NEW_STRING )
857 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
860 if (s + 1 < send && (s[1] == '\\'))
861 s++; /* all that, just for this */
866 SvCUR_set(sv, d - SvPVX(sv));
868 if ( PL_hints & HINT_NEW_STRING )
869 return new_constant(NULL, 0, "q", sv, pv, "q");
874 * Now come three functions related to double-quote context,
875 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
876 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
877 * interact with PL_lex_state, and create fake ( ... ) argument lists
878 * to handle functions and concatenation.
879 * They assume that whoever calls them will be setting up a fake
880 * join call, because each subthing puts a ',' after it. This lets
883 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
885 * (I'm not sure whether the spurious commas at the end of lcfirst's
886 * arguments and join's arguments are created or not).
891 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
893 * Pattern matching will set PL_lex_op to the pattern-matching op to
894 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
896 * OP_CONST and OP_READLINE are easy--just make the new op and return.
898 * Everything else becomes a FUNC.
900 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
901 * had an OP_CONST or OP_READLINE). This just sets us up for a
902 * call to S_sublex_push().
908 register I32 op_type = yylval.ival;
910 if (op_type == OP_NULL) {
911 yylval.opval = PL_lex_op;
915 if (op_type == OP_CONST || op_type == OP_READLINE) {
916 SV *sv = tokeq(PL_lex_stuff);
918 if (SvTYPE(sv) == SVt_PVIV) {
919 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
925 nsv = newSVpvn(p, len);
929 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
930 PL_lex_stuff = Nullsv;
934 PL_sublex_info.super_state = PL_lex_state;
935 PL_sublex_info.sub_inwhat = op_type;
936 PL_sublex_info.sub_op = PL_lex_op;
937 PL_lex_state = LEX_INTERPPUSH;
941 yylval.opval = PL_lex_op;
951 * Create a new scope to save the lexing state. The scope will be
952 * ended in S_sublex_done. Returns a '(', starting the function arguments
953 * to the uc, lc, etc. found before.
954 * Sets PL_lex_state to LEX_INTERPCONCAT.
963 PL_lex_state = PL_sublex_info.super_state;
964 SAVEI32(PL_lex_dojoin);
965 SAVEI32(PL_lex_brackets);
966 SAVEI32(PL_lex_fakebrack);
967 SAVEI32(PL_lex_casemods);
968 SAVEI32(PL_lex_starts);
969 SAVEI32(PL_lex_state);
970 SAVESPTR(PL_lex_inpat);
971 SAVEI32(PL_lex_inwhat);
972 SAVEI16(PL_curcop->cop_line);
974 SAVEPPTR(PL_oldbufptr);
975 SAVEPPTR(PL_oldoldbufptr);
976 SAVEPPTR(PL_linestart);
977 SAVESPTR(PL_linestr);
978 SAVEPPTR(PL_lex_brackstack);
979 SAVEPPTR(PL_lex_casestack);
981 PL_linestr = PL_lex_stuff;
982 PL_lex_stuff = Nullsv;
984 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
986 PL_bufend += SvCUR(PL_linestr);
987 SAVEFREESV(PL_linestr);
989 PL_lex_dojoin = FALSE;
991 PL_lex_fakebrack = 0;
992 New(899, PL_lex_brackstack, 120, char);
993 New(899, PL_lex_casestack, 12, char);
994 SAVEFREEPV(PL_lex_brackstack);
995 SAVEFREEPV(PL_lex_casestack);
997 *PL_lex_casestack = '\0';
999 PL_lex_state = LEX_INTERPCONCAT;
1000 PL_curcop->cop_line = PL_multi_start;
1002 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1003 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1004 PL_lex_inpat = PL_sublex_info.sub_op;
1006 PL_lex_inpat = Nullop;
1013 * Restores lexer state after a S_sublex_push.
1019 if (!PL_lex_starts++) {
1020 PL_expect = XOPERATOR;
1021 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpvn("",0));
1025 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1026 PL_lex_state = LEX_INTERPCASEMOD;
1030 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1031 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1032 PL_linestr = PL_lex_repl;
1034 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1035 PL_bufend += SvCUR(PL_linestr);
1036 SAVEFREESV(PL_linestr);
1037 PL_lex_dojoin = FALSE;
1038 PL_lex_brackets = 0;
1039 PL_lex_fakebrack = 0;
1040 PL_lex_casemods = 0;
1041 *PL_lex_casestack = '\0';
1043 if (SvEVALED(PL_lex_repl)) {
1044 PL_lex_state = LEX_INTERPNORMAL;
1046 /* we don't clear PL_lex_repl here, so that we can check later
1047 whether this is an evalled subst; that means we rely on the
1048 logic to ensure sublex_done() is called again only via the
1049 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1052 PL_lex_state = LEX_INTERPCONCAT;
1053 PL_lex_repl = Nullsv;
1059 PL_bufend = SvPVX(PL_linestr);
1060 PL_bufend += SvCUR(PL_linestr);
1061 PL_expect = XOPERATOR;
1062 PL_sublex_info.sub_inwhat = 0;
1070 Extracts a pattern, double-quoted string, or transliteration. This
1073 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1074 processing a pattern (PL_lex_inpat is true), a transliteration
1075 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1077 Returns a pointer to the character scanned up to. Iff this is
1078 advanced from the start pointer supplied (ie if anything was
1079 successfully parsed), will leave an OP for the substring scanned
1080 in yylval. Caller must intuit reason for not parsing further
1081 by looking at the next characters herself.
1085 double-quoted style: \r and \n
1086 regexp special ones: \D \s
1088 backrefs: \1 (deprecated in substitution replacements)
1089 case and quoting: \U \Q \E
1090 stops on @ and $, but not for $ as tail anchor
1092 In transliterations:
1093 characters are VERY literal, except for - not at the start or end
1094 of the string, which indicates a range. scan_const expands the
1095 range to the full set of intermediate characters.
1097 In double-quoted strings:
1099 double-quoted style: \r and \n
1101 backrefs: \1 (deprecated)
1102 case and quoting: \U \Q \E
1105 scan_const does *not* construct ops to handle interpolated strings.
1106 It stops processing as soon as it finds an embedded $ or @ variable
1107 and leaves it to the caller to work out what's going on.
1109 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @:foo.
1111 $ in pattern could be $foo or could be tail anchor. Assumption:
1112 it's a tail anchor if $ is the last thing in the string, or if it's
1113 followed by one of ")| \n\t"
1115 \1 (backreferences) are turned into $1
1117 The structure of the code is
1118 while (there's a character to process) {
1119 handle transliteration ranges
1120 skip regexp comments
1121 skip # initiated comments in //x patterns
1122 check for embedded @foo
1123 check for embedded scalars
1125 leave intact backslashes from leave (below)
1126 deprecate \1 in strings and sub replacements
1127 handle string-changing backslashes \l \U \Q \E, etc.
1128 switch (what was escaped) {
1129 handle - in a transliteration (becomes a literal -)
1130 handle \132 octal characters
1131 handle 0x15 hex characters
1132 handle \cV (control V)
1133 handle printf backslashes (\f, \r, \n, etc)
1135 } (end if backslash)
1136 } (end while character to read)
1141 S_scan_const(pTHX_ char *start)
1143 register char *send = PL_bufend; /* end of the constant */
1144 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1145 register char *s = start; /* start of the constant */
1146 register char *d = SvPVX(sv); /* destination for copies */
1147 bool dorange = FALSE; /* are we in a translit range? */
1149 I32 utf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1150 ? (PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
1152 I32 thisutf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1153 ? (PL_sublex_info.sub_op->op_private & (PL_lex_repl ?
1154 OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF))
1156 char *leaveit = /* set of acceptably-backslashed characters */
1158 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
1161 while (s < send || dorange) {
1162 /* get transliterations out of the way (they're most literal) */
1163 if (PL_lex_inwhat == OP_TRANS) {
1164 /* expand a range A-Z to the full set of characters. AIE! */
1166 I32 i; /* current expanded character */
1167 I32 min; /* first character in range */
1168 I32 max; /* last character in range */
1170 i = d - SvPVX(sv); /* remember current offset */
1171 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1172 d = SvPVX(sv) + i; /* refresh d after realloc */
1173 d -= 2; /* eat the first char and the - */
1175 min = (U8)*d; /* first char in range */
1176 max = (U8)d[1]; /* last char in range */
1179 if ((isLOWER(min) && isLOWER(max)) ||
1180 (isUPPER(min) && isUPPER(max))) {
1182 for (i = min; i <= max; i++)
1186 for (i = min; i <= max; i++)
1193 for (i = min; i <= max; i++)
1196 /* mark the range as done, and continue */
1201 /* range begins (ignore - as first or last char) */
1202 else if (*s == '-' && s+1 < send && s != start) {
1204 *d++ = (char)0xff; /* use illegal utf8 byte--see pmtrans */
1213 /* if we get here, we're not doing a transliteration */
1215 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1216 except for the last char, which will be done separately. */
1217 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1219 while (s < send && *s != ')')
1221 } else if (s[2] == '{'
1222 || s[2] == 'p' && s[3] == '{') { /* This should march regcomp.c */
1224 char *regparse = s + (s[2] == '{' ? 3 : 4);
1227 while (count && (c = *regparse)) {
1228 if (c == '\\' && regparse[1])
1236 if (*regparse != ')') {
1237 regparse--; /* Leave one char for continuation. */
1238 yyerror("Sequence (?{...}) not terminated or not {}-balanced");
1240 while (s < regparse)
1245 /* likewise skip #-initiated comments in //x patterns */
1246 else if (*s == '#' && PL_lex_inpat &&
1247 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1248 while (s+1 < send && *s != '\n')
1252 /* check for embedded arrays (@foo, @:foo, @'foo, @{foo}, @$foo) */
1253 else if (*s == '@' && s[1] && (isALNUM_lazy(s+1) || strchr(":'{$", s[1])))
1256 /* check for embedded scalars. only stop if we're sure it's a
1259 else if (*s == '$') {
1260 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1262 if (s + 1 < send && !strchr("()| \n\t", s[1]))
1263 break; /* in regexp, $ might be tail anchor */
1266 /* (now in tr/// code again) */
1268 if (*s & 0x80 && thisutf) {
1269 dTHR; /* only for ckWARN */
1270 if (ckWARN(WARN_UTF8)) {
1271 (void)utf8_to_uv((U8*)s, &len); /* could cvt latin-1 to utf8 here... */
1281 if (*s == '\\' && s+1 < send) {
1284 /* some backslashes we leave behind */
1285 if (*leaveit && *s && strchr(leaveit, *s)) {
1291 /* deprecate \1 in strings and substitution replacements */
1292 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1293 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1295 dTHR; /* only for ckWARN */
1296 if (ckWARN(WARN_SYNTAX))
1297 Perl_warner(aTHX_ WARN_SYNTAX, "\\%c better written as $%c", *s, *s);
1302 /* string-change backslash escapes */
1303 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1308 /* if we get here, it's either a quoted -, or a digit */
1311 /* quoted - in transliterations */
1313 if (PL_lex_inwhat == OP_TRANS) {
1321 if (ckWARN(WARN_UNSAFE) && isALPHA(*s))
1322 Perl_warner(aTHX_ WARN_UNSAFE,
1323 "Unrecognized escape \\%c passed through",
1325 /* default action is to copy the quoted character */
1330 /* \132 indicates an octal constant */
1331 case '0': case '1': case '2': case '3':
1332 case '4': case '5': case '6': case '7':
1333 *d++ = scan_oct(s, 3, &len);
1337 /* \x24 indicates a hex constant */
1341 char* e = strchr(s, '}');
1344 yyerror("Missing right brace on \\x{}");
1349 if (ckWARN(WARN_UTF8))
1350 Perl_warner(aTHX_ WARN_UTF8,
1351 "Use of \\x{} without utf8 declaration");
1353 /* note: utf always shorter than hex */
1354 d = (char*)uv_to_utf8((U8*)d,
1355 scan_hex(s + 1, e - s - 1, &len));
1359 UV uv = (UV)scan_hex(s, 2, &len);
1360 if (utf && PL_lex_inwhat == OP_TRANS &&
1361 utf != (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
1363 d = (char*)uv_to_utf8((U8*)d, uv); /* doing a CU or UC */
1366 if (uv >= 127 && UTF) {
1368 if (ckWARN(WARN_UTF8))
1369 Perl_warner(aTHX_ WARN_UTF8,
1370 "\\x%.*s will produce malformed UTF-8 character; use \\x{%.*s} for that",
1379 /* \N{latin small letter a} is a named character */
1383 char* e = strchr(s, '}');
1392 yyerror("Missing right brace on \\N{}");
1396 res = newSVpvn(s + 1, e - s - 1);
1397 res = new_constant( Nullch, 0, "charnames",
1398 res, Nullsv, "\\N{...}" );
1399 str = SvPV(res,len);
1400 if (len > e - s + 4) {
1401 char *odest = SvPVX(sv);
1403 SvGROW(sv, (SvCUR(sv) + len - (e - s + 4)));
1404 d = SvPVX(sv) + (d - odest);
1406 Copy(str, d, len, char);
1413 yyerror("Missing braces on \\N{}");
1416 /* \c is a control character */
1430 /* printf-style backslashes, formfeeds, newlines, etc */
1448 *d++ = '\047'; /* CP 1047 */
1451 *d++ = '\057'; /* CP 1047 */
1465 } /* end if (backslash) */
1468 } /* while loop to process each character */
1470 /* terminate the string and set up the sv */
1472 SvCUR_set(sv, d - SvPVX(sv));
1475 /* shrink the sv if we allocated more than we used */
1476 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1477 SvLEN_set(sv, SvCUR(sv) + 1);
1478 Renew(SvPVX(sv), SvLEN(sv), char);
1481 /* return the substring (via yylval) only if we parsed anything */
1482 if (s > PL_bufptr) {
1483 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1484 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1486 ( PL_lex_inwhat == OP_TRANS
1488 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1491 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1498 * Returns TRUE if there's more to the expression (e.g., a subscript),
1501 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1503 * ->[ and ->{ return TRUE
1504 * { and [ outside a pattern are always subscripts, so return TRUE
1505 * if we're outside a pattern and it's not { or [, then return FALSE
1506 * if we're in a pattern and the first char is a {
1507 * {4,5} (any digits around the comma) returns FALSE
1508 * if we're in a pattern and the first char is a [
1510 * [SOMETHING] has a funky algorithm to decide whether it's a
1511 * character class or not. It has to deal with things like
1512 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1513 * anything else returns TRUE
1516 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1519 S_intuit_more(pTHX_ register char *s)
1521 if (PL_lex_brackets)
1523 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1525 if (*s != '{' && *s != '[')
1530 /* In a pattern, so maybe we have {n,m}. */
1547 /* On the other hand, maybe we have a character class */
1550 if (*s == ']' || *s == '^')
1553 /* this is terrifying, and it works */
1554 int weight = 2; /* let's weigh the evidence */
1556 unsigned char un_char = 255, last_un_char;
1557 char *send = strchr(s,']');
1558 char tmpbuf[sizeof PL_tokenbuf * 4];
1560 if (!send) /* has to be an expression */
1563 Zero(seen,256,char);
1566 else if (isDIGIT(*s)) {
1568 if (isDIGIT(s[1]) && s[2] == ']')
1574 for (; s < send; s++) {
1575 last_un_char = un_char;
1576 un_char = (unsigned char)*s;
1581 weight -= seen[un_char] * 10;
1582 if (isALNUM_lazy(s+1)) {
1583 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1584 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1589 else if (*s == '$' && s[1] &&
1590 strchr("[#!%*<>()-=",s[1])) {
1591 if (/*{*/ strchr("])} =",s[2]))
1600 if (strchr("wds]",s[1]))
1602 else if (seen['\''] || seen['"'])
1604 else if (strchr("rnftbxcav",s[1]))
1606 else if (isDIGIT(s[1])) {
1608 while (s[1] && isDIGIT(s[1]))
1618 if (strchr("aA01! ",last_un_char))
1620 if (strchr("zZ79~",s[1]))
1622 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1623 weight -= 5; /* cope with negative subscript */
1626 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1627 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1632 if (keyword(tmpbuf, d - tmpbuf))
1635 if (un_char == last_un_char + 1)
1637 weight -= seen[un_char];
1642 if (weight >= 0) /* probably a character class */
1652 * Does all the checking to disambiguate
1654 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1655 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1657 * First argument is the stuff after the first token, e.g. "bar".
1659 * Not a method if bar is a filehandle.
1660 * Not a method if foo is a subroutine prototyped to take a filehandle.
1661 * Not a method if it's really "Foo $bar"
1662 * Method if it's "foo $bar"
1663 * Not a method if it's really "print foo $bar"
1664 * Method if it's really "foo package::" (interpreted as package->foo)
1665 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1666 * Not a method if bar is a filehandle or package, but is quotd with
1671 S_intuit_method(pTHX_ char *start, GV *gv)
1673 char *s = start + (*start == '$');
1674 char tmpbuf[sizeof PL_tokenbuf];
1682 if ((cv = GvCVu(gv))) {
1683 char *proto = SvPVX(cv);
1693 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1694 /* start is the beginning of the possible filehandle/object,
1695 * and s is the end of it
1696 * tmpbuf is a copy of it
1699 if (*start == '$') {
1700 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1705 return *s == '(' ? FUNCMETH : METHOD;
1707 if (!keyword(tmpbuf, len)) {
1708 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1713 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1714 if (indirgv && GvCVu(indirgv))
1716 /* filehandle or package name makes it a method */
1717 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1719 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1720 return 0; /* no assumptions -- "=>" quotes bearword */
1722 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1723 newSVpvn(tmpbuf,len));
1724 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1728 return *s == '(' ? FUNCMETH : METHOD;
1736 * Return a string of Perl code to load the debugger. If PERL5DB
1737 * is set, it will return the contents of that, otherwise a
1738 * compile-time require of perl5db.pl.
1745 char *pdb = PerlEnv_getenv("PERL5DB");
1749 SETERRNO(0,SS$_NORMAL);
1750 return "BEGIN { require 'perl5db.pl' }";
1756 /* Encoded script support. filter_add() effectively inserts a
1757 * 'pre-processing' function into the current source input stream.
1758 * Note that the filter function only applies to the current source file
1759 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1761 * The datasv parameter (which may be NULL) can be used to pass
1762 * private data to this instance of the filter. The filter function
1763 * can recover the SV using the FILTER_DATA macro and use it to
1764 * store private buffers and state information.
1766 * The supplied datasv parameter is upgraded to a PVIO type
1767 * and the IoDIRP field is used to store the function pointer.
1768 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1769 * private use must be set using malloc'd pointers.
1773 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1775 if (!funcp){ /* temporary handy debugging hack to be deleted */
1776 PL_filter_debug = atoi((char*)datasv);
1779 if (!PL_rsfp_filters)
1780 PL_rsfp_filters = newAV();
1782 datasv = NEWSV(255,0);
1783 if (!SvUPGRADE(datasv, SVt_PVIO))
1784 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1785 IoDIRP(datasv) = (DIR*)funcp; /* stash funcp into spare field */
1787 if (PL_filter_debug) {
1789 Perl_warn(aTHX_ "filter_add func %p (%s)", funcp, SvPV(datasv, n_a));
1791 #endif /* DEBUGGING */
1792 av_unshift(PL_rsfp_filters, 1);
1793 av_store(PL_rsfp_filters, 0, datasv) ;
1798 /* Delete most recently added instance of this filter function. */
1800 Perl_filter_del(pTHX_ filter_t funcp)
1803 if (PL_filter_debug)
1804 Perl_warn(aTHX_ "filter_del func %p", funcp);
1805 #endif /* DEBUGGING */
1806 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
1808 /* if filter is on top of stack (usual case) just pop it off */
1809 if (IoDIRP(FILTER_DATA(AvFILLp(PL_rsfp_filters))) == (DIR*)funcp){
1810 IoDIRP(FILTER_DATA(AvFILLp(PL_rsfp_filters))) = NULL;
1811 sv_free(av_pop(PL_rsfp_filters));
1815 /* we need to search for the correct entry and clear it */
1816 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
1820 /* Invoke the n'th filter function for the current rsfp. */
1822 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
1825 /* 0 = read one text line */
1830 if (!PL_rsfp_filters)
1832 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
1833 /* Provide a default input filter to make life easy. */
1834 /* Note that we append to the line. This is handy. */
1836 if (PL_filter_debug)
1837 Perl_warn(aTHX_ "filter_read %d: from rsfp\n", idx);
1838 #endif /* DEBUGGING */
1842 int old_len = SvCUR(buf_sv) ;
1844 /* ensure buf_sv is large enough */
1845 SvGROW(buf_sv, old_len + maxlen) ;
1846 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
1847 if (PerlIO_error(PL_rsfp))
1848 return -1; /* error */
1850 return 0 ; /* end of file */
1852 SvCUR_set(buf_sv, old_len + len) ;
1855 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
1856 if (PerlIO_error(PL_rsfp))
1857 return -1; /* error */
1859 return 0 ; /* end of file */
1862 return SvCUR(buf_sv);
1864 /* Skip this filter slot if filter has been deleted */
1865 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
1867 if (PL_filter_debug)
1868 Perl_warn(aTHX_ "filter_read %d: skipped (filter deleted)\n", idx);
1869 #endif /* DEBUGGING */
1870 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
1872 /* Get function pointer hidden within datasv */
1873 funcp = (filter_t)IoDIRP(datasv);
1875 if (PL_filter_debug) {
1877 Perl_warn(aTHX_ "filter_read %d: via function %p (%s)\n",
1878 idx, funcp, SvPV(datasv,n_a));
1880 #endif /* DEBUGGING */
1881 /* Call function. The function is expected to */
1882 /* call "FILTER_READ(idx+1, buf_sv)" first. */
1883 /* Return: <0:error, =0:eof, >0:not eof */
1884 return (*funcp)(aTHXo_ idx, buf_sv, maxlen);
1888 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
1891 if (!PL_rsfp_filters) {
1892 filter_add(win32_textfilter,NULL);
1895 if (PL_rsfp_filters) {
1898 SvCUR_set(sv, 0); /* start with empty line */
1899 if (FILTER_READ(0, sv, 0) > 0)
1900 return ( SvPVX(sv) ) ;
1905 return (sv_gets(sv, fp, append));
1910 static char* exp_name[] =
1911 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
1912 "ATTRTERM", "TERMBLOCK"
1919 Works out what to call the token just pulled out of the input
1920 stream. The yacc parser takes care of taking the ops we return and
1921 stitching them into a tree.
1927 if read an identifier
1928 if we're in a my declaration
1929 croak if they tried to say my($foo::bar)
1930 build the ops for a my() declaration
1931 if it's an access to a my() variable
1932 are we in a sort block?
1933 croak if my($a); $a <=> $b
1934 build ops for access to a my() variable
1935 if in a dq string, and they've said @foo and we can't find @foo
1937 build ops for a bareword
1938 if we already built the token before, use it.
1942 #ifdef USE_PURE_BISON
1943 Perl_yylex(pTHX_ YYSTYPE *lvalp, int *lcharp)
1956 #ifdef USE_PURE_BISON
1957 yylval_pointer = lvalp;
1958 yychar_pointer = lcharp;
1961 /* check if there's an identifier for us to look at */
1962 if (PL_pending_ident) {
1963 /* pit holds the identifier we read and pending_ident is reset */
1964 char pit = PL_pending_ident;
1965 PL_pending_ident = 0;
1967 /* if we're in a my(), we can't allow dynamics here.
1968 $foo'bar has already been turned into $foo::bar, so
1969 just check for colons.
1971 if it's a legal name, the OP is a PADANY.
1974 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
1975 tmp = pad_allocmy(PL_tokenbuf);
1978 if (strchr(PL_tokenbuf,':'))
1979 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
1981 yylval.opval = newOP(OP_PADANY, 0);
1982 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
1988 build the ops for accesses to a my() variable.
1990 Deny my($a) or my($b) in a sort block, *if* $a or $b is
1991 then used in a comparison. This catches most, but not
1992 all cases. For instance, it catches
1993 sort { my($a); $a <=> $b }
1995 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
1996 (although why you'd do that is anyone's guess).
1999 if (!strchr(PL_tokenbuf,':')) {
2001 /* Check for single character per-thread SVs */
2002 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
2003 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
2004 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
2006 yylval.opval = newOP(OP_THREADSV, 0);
2007 yylval.opval->op_targ = tmp;
2010 #endif /* USE_THREADS */
2011 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
2012 /* might be an "our" variable" */
2013 if (SvFLAGS(AvARRAY(PL_comppad_name)[tmp]) & SVpad_OUR) {
2014 /* build ops for a bareword */
2015 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2016 yylval.opval->op_private = OPpCONST_ENTERED;
2017 gv_fetchpv(PL_tokenbuf+1,
2019 ? (GV_ADDMULTI | GV_ADDINEVAL | GV_ADDOUR)
2022 ((PL_tokenbuf[0] == '$') ? SVt_PV
2023 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2028 /* if it's a sort block and they're naming $a or $b */
2029 if (PL_last_lop_op == OP_SORT &&
2030 PL_tokenbuf[0] == '$' &&
2031 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
2034 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
2035 d < PL_bufend && *d != '\n';
2038 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
2039 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
2045 yylval.opval = newOP(OP_PADANY, 0);
2046 yylval.opval->op_targ = tmp;
2052 Whine if they've said @foo in a doublequoted string,
2053 and @foo isn't a variable we can find in the symbol
2056 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
2057 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
2058 if (!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
2059 yyerror(Perl_form(aTHX_ "In string, %s now must be written as \\%s",
2060 PL_tokenbuf, PL_tokenbuf));
2063 /* build ops for a bareword */
2064 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2065 yylval.opval->op_private = OPpCONST_ENTERED;
2066 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
2067 ((PL_tokenbuf[0] == '$') ? SVt_PV
2068 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2073 /* no identifier pending identification */
2075 switch (PL_lex_state) {
2077 case LEX_NORMAL: /* Some compilers will produce faster */
2078 case LEX_INTERPNORMAL: /* code if we comment these out. */
2082 /* when we've already built the next token, just pull it out of the queue */
2085 yylval = PL_nextval[PL_nexttoke];
2087 PL_lex_state = PL_lex_defer;
2088 PL_expect = PL_lex_expect;
2089 PL_lex_defer = LEX_NORMAL;
2091 return(PL_nexttype[PL_nexttoke]);
2093 /* interpolated case modifiers like \L \U, including \Q and \E.
2094 when we get here, PL_bufptr is at the \
2096 case LEX_INTERPCASEMOD:
2098 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2099 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2101 /* handle \E or end of string */
2102 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2106 if (PL_lex_casemods) {
2107 oldmod = PL_lex_casestack[--PL_lex_casemods];
2108 PL_lex_casestack[PL_lex_casemods] = '\0';
2110 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2112 PL_lex_state = LEX_INTERPCONCAT;
2116 if (PL_bufptr != PL_bufend)
2118 PL_lex_state = LEX_INTERPCONCAT;
2123 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2124 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2125 if (strchr("LU", *s) &&
2126 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2128 PL_lex_casestack[--PL_lex_casemods] = '\0';
2131 if (PL_lex_casemods > 10) {
2132 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2133 if (newlb != PL_lex_casestack) {
2135 PL_lex_casestack = newlb;
2138 PL_lex_casestack[PL_lex_casemods++] = *s;
2139 PL_lex_casestack[PL_lex_casemods] = '\0';
2140 PL_lex_state = LEX_INTERPCONCAT;
2141 PL_nextval[PL_nexttoke].ival = 0;
2144 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2146 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2148 PL_nextval[PL_nexttoke].ival = OP_LC;
2150 PL_nextval[PL_nexttoke].ival = OP_UC;
2152 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2154 Perl_croak(aTHX_ "panic: yylex");
2157 if (PL_lex_starts) {
2166 case LEX_INTERPPUSH:
2167 return sublex_push();
2169 case LEX_INTERPSTART:
2170 if (PL_bufptr == PL_bufend)
2171 return sublex_done();
2173 PL_lex_dojoin = (*PL_bufptr == '@');
2174 PL_lex_state = LEX_INTERPNORMAL;
2175 if (PL_lex_dojoin) {
2176 PL_nextval[PL_nexttoke].ival = 0;
2179 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2180 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2181 force_next(PRIVATEREF);
2183 force_ident("\"", '$');
2184 #endif /* USE_THREADS */
2185 PL_nextval[PL_nexttoke].ival = 0;
2187 PL_nextval[PL_nexttoke].ival = 0;
2189 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2192 if (PL_lex_starts++) {
2198 case LEX_INTERPENDMAYBE:
2199 if (intuit_more(PL_bufptr)) {
2200 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2206 if (PL_lex_dojoin) {
2207 PL_lex_dojoin = FALSE;
2208 PL_lex_state = LEX_INTERPCONCAT;
2211 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2212 && SvEVALED(PL_lex_repl))
2214 if (PL_bufptr != PL_bufend)
2215 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2216 PL_lex_repl = Nullsv;
2219 case LEX_INTERPCONCAT:
2221 if (PL_lex_brackets)
2222 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2224 if (PL_bufptr == PL_bufend)
2225 return sublex_done();
2227 if (SvIVX(PL_linestr) == '\'') {
2228 SV *sv = newSVsv(PL_linestr);
2231 else if ( PL_hints & HINT_NEW_RE )
2232 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2233 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2237 s = scan_const(PL_bufptr);
2239 PL_lex_state = LEX_INTERPCASEMOD;
2241 PL_lex_state = LEX_INTERPSTART;
2244 if (s != PL_bufptr) {
2245 PL_nextval[PL_nexttoke] = yylval;
2248 if (PL_lex_starts++)
2258 PL_lex_state = LEX_NORMAL;
2259 s = scan_formline(PL_bufptr);
2260 if (!PL_lex_formbrack)
2266 PL_oldoldbufptr = PL_oldbufptr;
2269 PerlIO_printf(PerlIO_stderr(), "### Tokener expecting %s at %s\n", exp_name[PL_expect], s);
2275 if (isIDFIRST_lazy(s))
2277 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2280 goto fake_eof; /* emulate EOF on ^D or ^Z */
2285 if (PL_lex_brackets)
2286 yyerror("Missing right curly or square bracket");
2289 if (s++ < PL_bufend)
2290 goto retry; /* ignore stray nulls */
2293 if (!PL_in_eval && !PL_preambled) {
2294 PL_preambled = TRUE;
2295 sv_setpv(PL_linestr,incl_perldb());
2296 if (SvCUR(PL_linestr))
2297 sv_catpv(PL_linestr,";");
2299 while(AvFILLp(PL_preambleav) >= 0) {
2300 SV *tmpsv = av_shift(PL_preambleav);
2301 sv_catsv(PL_linestr, tmpsv);
2302 sv_catpv(PL_linestr, ";");
2305 sv_free((SV*)PL_preambleav);
2306 PL_preambleav = NULL;
2308 if (PL_minus_n || PL_minus_p) {
2309 sv_catpv(PL_linestr, "LINE: while (<>) {");
2311 sv_catpv(PL_linestr,"chomp;");
2313 GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
2315 GvIMPORTED_AV_on(gv);
2317 if (strchr("/'\"", *PL_splitstr)
2318 && strchr(PL_splitstr + 1, *PL_splitstr))
2319 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2322 s = "'~#\200\1'"; /* surely one char is unused...*/
2323 while (s[1] && strchr(PL_splitstr, *s)) s++;
2325 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c",
2326 "q" + (delim == '\''), delim);
2327 for (s = PL_splitstr; *s; s++) {
2329 sv_catpvn(PL_linestr, "\\", 1);
2330 sv_catpvn(PL_linestr, s, 1);
2332 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2336 sv_catpv(PL_linestr,"@F=split(' ');");
2339 sv_catpv(PL_linestr, "\n");
2340 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2341 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2342 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2343 SV *sv = NEWSV(85,0);
2345 sv_upgrade(sv, SVt_PVMG);
2346 sv_setsv(sv,PL_linestr);
2347 av_store(GvAV(PL_curcop->cop_filegv),(I32)PL_curcop->cop_line,sv);
2352 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2355 if (PL_preprocess && !PL_in_eval)
2356 (void)PerlProc_pclose(PL_rsfp);
2357 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2358 PerlIO_clearerr(PL_rsfp);
2360 (void)PerlIO_close(PL_rsfp);
2362 PL_doextract = FALSE;
2364 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2365 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2366 sv_catpv(PL_linestr,";}");
2367 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2368 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2369 PL_minus_n = PL_minus_p = 0;
2372 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2373 sv_setpv(PL_linestr,"");
2374 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2377 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2378 PL_doextract = FALSE;
2380 /* Incest with pod. */
2381 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2382 sv_setpv(PL_linestr, "");
2383 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2384 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2385 PL_doextract = FALSE;
2389 } while (PL_doextract);
2390 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2391 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2392 SV *sv = NEWSV(85,0);
2394 sv_upgrade(sv, SVt_PVMG);
2395 sv_setsv(sv,PL_linestr);
2396 av_store(GvAV(PL_curcop->cop_filegv),(I32)PL_curcop->cop_line,sv);
2398 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2399 if (PL_curcop->cop_line == 1) {
2400 while (s < PL_bufend && isSPACE(*s))
2402 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2406 if (*s == '#' && *(s+1) == '!')
2408 #ifdef ALTERNATE_SHEBANG
2410 static char as[] = ALTERNATE_SHEBANG;
2411 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2412 d = s + (sizeof(as) - 1);
2414 #endif /* ALTERNATE_SHEBANG */
2423 while (*d && !isSPACE(*d))
2427 #ifdef ARG_ZERO_IS_SCRIPT
2428 if (ipathend > ipath) {
2430 * HP-UX (at least) sets argv[0] to the script name,
2431 * which makes $^X incorrect. And Digital UNIX and Linux,
2432 * at least, set argv[0] to the basename of the Perl
2433 * interpreter. So, having found "#!", we'll set it right.
2435 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2436 assert(SvPOK(x) || SvGMAGICAL(x));
2437 if (sv_eq(x, GvSV(PL_curcop->cop_filegv))) {
2438 sv_setpvn(x, ipath, ipathend - ipath);
2441 TAINT_NOT; /* $^X is always tainted, but that's OK */
2443 #endif /* ARG_ZERO_IS_SCRIPT */
2448 d = instr(s,"perl -");
2450 d = instr(s,"perl");
2452 /* avoid getting into infinite loops when shebang
2453 * line contains "Perl" rather than "perl" */
2455 for (d = ipathend-4; d >= ipath; --d) {
2456 if ((*d == 'p' || *d == 'P')
2457 && !ibcmp(d, "perl", 4))
2467 #ifdef ALTERNATE_SHEBANG
2469 * If the ALTERNATE_SHEBANG on this system starts with a
2470 * character that can be part of a Perl expression, then if
2471 * we see it but not "perl", we're probably looking at the
2472 * start of Perl code, not a request to hand off to some
2473 * other interpreter. Similarly, if "perl" is there, but
2474 * not in the first 'word' of the line, we assume the line
2475 * contains the start of the Perl program.
2477 if (d && *s != '#') {
2479 while (*c && !strchr("; \t\r\n\f\v#", *c))
2482 d = Nullch; /* "perl" not in first word; ignore */
2484 *s = '#'; /* Don't try to parse shebang line */
2486 #endif /* ALTERNATE_SHEBANG */
2491 !instr(s,"indir") &&
2492 instr(PL_origargv[0],"perl"))
2498 while (s < PL_bufend && isSPACE(*s))
2500 if (s < PL_bufend) {
2501 Newz(899,newargv,PL_origargc+3,char*);
2503 while (s < PL_bufend && !isSPACE(*s))
2506 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2509 newargv = PL_origargv;
2511 PerlProc_execv(ipath, newargv);
2512 Perl_croak(aTHX_ "Can't exec %s", ipath);
2515 U32 oldpdb = PL_perldb;
2516 bool oldn = PL_minus_n;
2517 bool oldp = PL_minus_p;
2519 while (*d && !isSPACE(*d)) d++;
2520 while (*d == ' ' || *d == '\t') d++;
2524 if (*d == 'M' || *d == 'm') {
2526 while (*d && !isSPACE(*d)) d++;
2527 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2530 d = moreswitches(d);
2532 if (PERLDB_LINE && !oldpdb ||
2533 ( PL_minus_n || PL_minus_p ) && !(oldn || oldp) )
2534 /* if we have already added "LINE: while (<>) {",
2535 we must not do it again */
2537 sv_setpv(PL_linestr, "");
2538 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2539 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2540 PL_preambled = FALSE;
2542 (void)gv_fetchfile(PL_origfilename);
2549 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2551 PL_lex_state = LEX_FORMLINE;
2556 #ifdef PERL_STRICT_CR
2557 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2559 "(Maybe you didn't strip carriage returns after a network transfer?)\n");
2561 case ' ': case '\t': case '\f': case 013:
2566 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2568 while (s < d && *s != '\n')
2573 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2575 PL_lex_state = LEX_FORMLINE;
2585 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2590 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
2593 if (strnEQ(s,"=>",2)) {
2594 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2595 OPERATOR('-'); /* unary minus */
2597 PL_last_uni = PL_oldbufptr;
2598 PL_last_lop_op = OP_FTEREAD; /* good enough */
2600 case 'r': FTST(OP_FTEREAD);
2601 case 'w': FTST(OP_FTEWRITE);
2602 case 'x': FTST(OP_FTEEXEC);
2603 case 'o': FTST(OP_FTEOWNED);
2604 case 'R': FTST(OP_FTRREAD);
2605 case 'W': FTST(OP_FTRWRITE);
2606 case 'X': FTST(OP_FTREXEC);
2607 case 'O': FTST(OP_FTROWNED);
2608 case 'e': FTST(OP_FTIS);
2609 case 'z': FTST(OP_FTZERO);
2610 case 's': FTST(OP_FTSIZE);
2611 case 'f': FTST(OP_FTFILE);
2612 case 'd': FTST(OP_FTDIR);
2613 case 'l': FTST(OP_FTLINK);
2614 case 'p': FTST(OP_FTPIPE);
2615 case 'S': FTST(OP_FTSOCK);
2616 case 'u': FTST(OP_FTSUID);
2617 case 'g': FTST(OP_FTSGID);
2618 case 'k': FTST(OP_FTSVTX);
2619 case 'b': FTST(OP_FTBLK);
2620 case 'c': FTST(OP_FTCHR);
2621 case 't': FTST(OP_FTTTY);
2622 case 'T': FTST(OP_FTTEXT);
2623 case 'B': FTST(OP_FTBINARY);
2624 case 'M': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTMTIME);
2625 case 'A': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTATIME);
2626 case 'C': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTCTIME);
2628 Perl_croak(aTHX_ "Unrecognized file test: -%c", (int)tmp);
2635 if (PL_expect == XOPERATOR)
2640 else if (*s == '>') {
2643 if (isIDFIRST_lazy(s)) {
2644 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2652 if (PL_expect == XOPERATOR)
2655 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2657 OPERATOR('-'); /* unary minus */
2664 if (PL_expect == XOPERATOR)
2669 if (PL_expect == XOPERATOR)
2672 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2678 if (PL_expect != XOPERATOR) {
2679 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2680 PL_expect = XOPERATOR;
2681 force_ident(PL_tokenbuf, '*');
2694 if (PL_expect == XOPERATOR) {
2698 PL_tokenbuf[0] = '%';
2699 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2700 if (!PL_tokenbuf[1]) {
2702 yyerror("Final % should be \\% or %name");
2705 PL_pending_ident = '%';
2724 switch (PL_expect) {
2727 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2729 PL_bufptr = s; /* update in case we back off */
2735 PL_expect = XTERMBLOCK;
2739 while (isIDFIRST_lazy(s)) {
2740 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2742 d = scan_str(d,TRUE,TRUE);
2745 SvREFCNT_dec(PL_lex_stuff);
2746 PL_lex_stuff = Nullsv;
2748 /* MUST advance bufptr here to avoid bogus
2749 "at end of line" context messages from yyerror().
2751 PL_bufptr = s + len;
2752 yyerror("Unterminated attribute parameter in attribute list");
2755 return 0; /* EOF indicator */
2759 SV *sv = newSVpvn(s, len);
2760 sv_catsv(sv, PL_lex_stuff);
2761 attrs = append_elem(OP_LIST, attrs,
2762 newSVOP(OP_CONST, 0, sv));
2763 SvREFCNT_dec(PL_lex_stuff);
2764 PL_lex_stuff = Nullsv;
2767 attrs = append_elem(OP_LIST, attrs,
2768 newSVOP(OP_CONST, 0,
2775 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}' for vi */
2776 if (*s != ';' && *s != tmp) {
2777 char q = ((*s == '\'') ? '"' : '\'');
2778 /* If here for an expression, and parsed no attrs, back off. */
2779 if (tmp == '=' && !attrs) {
2783 /* MUST advance bufptr here to avoid bogus "at end of line"
2784 context messages from yyerror().
2788 yyerror("Unterminated attribute list");
2790 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
2797 PL_nextval[PL_nexttoke].opval = attrs;
2805 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
2806 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
2811 if (PL_curcop->cop_line < PL_copline)
2812 PL_copline = PL_curcop->cop_line;
2823 if (PL_lex_brackets <= 0)
2824 yyerror("Unmatched right square bracket");
2827 if (PL_lex_state == LEX_INTERPNORMAL) {
2828 if (PL_lex_brackets == 0) {
2829 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
2830 PL_lex_state = LEX_INTERPEND;
2837 if (PL_lex_brackets > 100) {
2838 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
2839 if (newlb != PL_lex_brackstack) {
2841 PL_lex_brackstack = newlb;
2844 switch (PL_expect) {
2846 if (PL_lex_formbrack) {
2850 if (PL_oldoldbufptr == PL_last_lop)
2851 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
2853 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2854 OPERATOR(HASHBRACK);
2856 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
2859 PL_tokenbuf[0] = '\0';
2860 if (d < PL_bufend && *d == '-') {
2861 PL_tokenbuf[0] = '-';
2863 while (d < PL_bufend && (*d == ' ' || *d == '\t'))
2866 if (d < PL_bufend && isIDFIRST_lazy(d)) {
2867 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
2869 while (d < PL_bufend && (*d == ' ' || *d == '\t'))
2872 char minus = (PL_tokenbuf[0] == '-');
2873 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
2881 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
2886 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2891 if (PL_oldoldbufptr == PL_last_lop)
2892 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
2894 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2897 OPERATOR(HASHBRACK);
2898 /* This hack serves to disambiguate a pair of curlies
2899 * as being a block or an anon hash. Normally, expectation
2900 * determines that, but in cases where we're not in a
2901 * position to expect anything in particular (like inside
2902 * eval"") we have to resolve the ambiguity. This code
2903 * covers the case where the first term in the curlies is a
2904 * quoted string. Most other cases need to be explicitly
2905 * disambiguated by prepending a `+' before the opening
2906 * curly in order to force resolution as an anon hash.
2908 * XXX should probably propagate the outer expectation
2909 * into eval"" to rely less on this hack, but that could
2910 * potentially break current behavior of eval"".
2914 if (*s == '\'' || *s == '"' || *s == '`') {
2915 /* common case: get past first string, handling escapes */
2916 for (t++; t < PL_bufend && *t != *s;)
2917 if (*t++ == '\\' && (*t == '\\' || *t == *s))
2921 else if (*s == 'q') {
2924 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
2925 && !isALNUM(*t)))) {
2927 char open, close, term;
2930 while (t < PL_bufend && isSPACE(*t))
2934 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
2938 for (t++; t < PL_bufend; t++) {
2939 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
2941 else if (*t == open)
2945 for (t++; t < PL_bufend; t++) {
2946 if (*t == '\\' && t+1 < PL_bufend)
2948 else if (*t == close && --brackets <= 0)
2950 else if (*t == open)
2956 else if (isIDFIRST_lazy(s)) {
2957 for (t++; t < PL_bufend && isALNUM_lazy(t); t++) ;
2959 while (t < PL_bufend && isSPACE(*t))
2961 /* if comma follows first term, call it an anon hash */
2962 /* XXX it could be a comma expression with loop modifiers */
2963 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
2964 || (*t == '=' && t[1] == '>')))
2965 OPERATOR(HASHBRACK);
2966 if (PL_expect == XREF)
2969 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
2975 yylval.ival = PL_curcop->cop_line;
2976 if (isSPACE(*s) || *s == '#')
2977 PL_copline = NOLINE; /* invalidate current command line number */
2982 if (PL_lex_brackets <= 0)
2983 yyerror("Unmatched right curly bracket");
2985 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
2986 if (PL_lex_brackets < PL_lex_formbrack)
2987 PL_lex_formbrack = 0;
2988 if (PL_lex_state == LEX_INTERPNORMAL) {
2989 if (PL_lex_brackets == 0) {
2990 if (PL_lex_fakebrack) {
2991 PL_lex_state = LEX_INTERPEND;
2993 return yylex(); /* ignore fake brackets */
2995 if (*s == '-' && s[1] == '>')
2996 PL_lex_state = LEX_INTERPENDMAYBE;
2997 else if (*s != '[' && *s != '{')
2998 PL_lex_state = LEX_INTERPEND;
3001 if (PL_lex_brackets < PL_lex_fakebrack) {
3003 PL_lex_fakebrack = 0;
3004 return yylex(); /* ignore fake brackets */
3014 if (PL_expect == XOPERATOR) {
3015 if (ckWARN(WARN_SEMICOLON) && isIDFIRST_lazy(s) && PL_bufptr == PL_linestart) {
3016 PL_curcop->cop_line--;
3017 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3018 PL_curcop->cop_line++;
3023 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3025 PL_expect = XOPERATOR;
3026 force_ident(PL_tokenbuf, '&');
3030 yylval.ival = (OPpENTERSUB_AMPER<<8);
3049 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3050 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3052 if (PL_expect == XSTATE && isALPHA(tmp) &&
3053 (s == PL_linestart+1 || s[-2] == '\n') )
3055 if (PL_in_eval && !PL_rsfp) {
3060 if (strnEQ(s,"=cut",4)) {
3074 PL_doextract = TRUE;
3077 if (PL_lex_brackets < PL_lex_formbrack) {
3079 #ifdef PERL_STRICT_CR
3080 for (t = s; *t == ' ' || *t == '\t'; t++) ;
3082 for (t = s; *t == ' ' || *t == '\t' || *t == '\r'; t++) ;
3084 if (*t == '\n' || *t == '#') {
3102 if (PL_expect != XOPERATOR) {
3103 if (s[1] != '<' && !strchr(s,'>'))
3106 s = scan_heredoc(s);
3108 s = scan_inputsymbol(s);
3109 TERM(sublex_start());
3114 SHop(OP_LEFT_SHIFT);
3128 SHop(OP_RIGHT_SHIFT);
3137 if (PL_expect == XOPERATOR) {
3138 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3141 return ','; /* grandfather non-comma-format format */
3145 if (s[1] == '#' && (isIDFIRST_lazy(s+2) || strchr("{$:+-", s[2]))) {
3146 PL_tokenbuf[0] = '@';
3147 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3148 sizeof PL_tokenbuf - 1, FALSE);
3149 if (PL_expect == XOPERATOR)
3150 no_op("Array length", s);
3151 if (!PL_tokenbuf[1])
3153 PL_expect = XOPERATOR;
3154 PL_pending_ident = '#';
3158 PL_tokenbuf[0] = '$';
3159 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3160 sizeof PL_tokenbuf - 1, FALSE);
3161 if (PL_expect == XOPERATOR)
3163 if (!PL_tokenbuf[1]) {
3165 yyerror("Final $ should be \\$ or $name");
3169 /* This kludge not intended to be bulletproof. */
3170 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3171 yylval.opval = newSVOP(OP_CONST, 0,
3172 newSViv((IV)PL_compiling.cop_arybase));
3173 yylval.opval->op_private = OPpCONST_ARYBASE;
3179 if (PL_lex_state == LEX_NORMAL)
3182 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3185 PL_tokenbuf[0] = '@';
3186 if (ckWARN(WARN_SYNTAX)) {
3188 isSPACE(*t) || isALNUM_lazy(t) || *t == '$';
3191 PL_bufptr = skipspace(PL_bufptr);
3192 while (t < PL_bufend && *t != ']')
3194 Perl_warner(aTHX_ WARN_SYNTAX,
3195 "Multidimensional syntax %.*s not supported",
3196 (t - PL_bufptr) + 1, PL_bufptr);
3200 else if (*s == '{') {
3201 PL_tokenbuf[0] = '%';
3202 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3203 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3205 char tmpbuf[sizeof PL_tokenbuf];
3207 for (t++; isSPACE(*t); t++) ;
3208 if (isIDFIRST_lazy(t)) {
3209 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3210 for (; isSPACE(*t); t++) ;
3211 if (*t == ';' && get_cv(tmpbuf, FALSE))
3212 Perl_warner(aTHX_ WARN_SYNTAX,
3213 "You need to quote \"%s\"", tmpbuf);
3219 PL_expect = XOPERATOR;
3220 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3221 bool islop = (PL_last_lop == PL_oldoldbufptr);
3222 if (!islop || PL_last_lop_op == OP_GREPSTART)
3223 PL_expect = XOPERATOR;
3224 else if (strchr("$@\"'`q", *s))
3225 PL_expect = XTERM; /* e.g. print $fh "foo" */
3226 else if (strchr("&*<%", *s) && isIDFIRST_lazy(s+1))
3227 PL_expect = XTERM; /* e.g. print $fh &sub */
3228 else if (isIDFIRST_lazy(s)) {
3229 char tmpbuf[sizeof PL_tokenbuf];
3230 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3231 if (tmp = keyword(tmpbuf, len)) {
3232 /* binary operators exclude handle interpretations */
3244 PL_expect = XTERM; /* e.g. print $fh length() */
3249 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3250 if (gv && GvCVu(gv))
3251 PL_expect = XTERM; /* e.g. print $fh subr() */
3254 else if (isDIGIT(*s))
3255 PL_expect = XTERM; /* e.g. print $fh 3 */
3256 else if (*s == '.' && isDIGIT(s[1]))
3257 PL_expect = XTERM; /* e.g. print $fh .3 */
3258 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3259 PL_expect = XTERM; /* e.g. print $fh -1 */
3260 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3261 PL_expect = XTERM; /* print $fh <<"EOF" */
3263 PL_pending_ident = '$';
3267 if (PL_expect == XOPERATOR)
3269 PL_tokenbuf[0] = '@';
3270 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3271 if (!PL_tokenbuf[1]) {
3273 yyerror("Final @ should be \\@ or @name");
3276 if (PL_lex_state == LEX_NORMAL)
3278 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3280 PL_tokenbuf[0] = '%';
3282 /* Warn about @ where they meant $. */
3283 if (ckWARN(WARN_SYNTAX)) {
3284 if (*s == '[' || *s == '{') {
3286 while (*t && (isALNUM_lazy(t) || strchr(" \t$#+-'\"", *t)))
3288 if (*t == '}' || *t == ']') {
3290 PL_bufptr = skipspace(PL_bufptr);
3291 Perl_warner(aTHX_ WARN_SYNTAX,
3292 "Scalar value %.*s better written as $%.*s",
3293 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3298 PL_pending_ident = '@';
3301 case '/': /* may either be division or pattern */
3302 case '?': /* may either be conditional or pattern */
3303 if (PL_expect != XOPERATOR) {
3304 /* Disable warning on "study /blah/" */
3305 if (PL_oldoldbufptr == PL_last_uni
3306 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3307 || memNE(PL_last_uni, "study", 5) || isALNUM_lazy(PL_last_uni+5)))
3309 s = scan_pat(s,OP_MATCH);
3310 TERM(sublex_start());
3318 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3319 #ifdef PERL_STRICT_CR
3322 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3324 && (s == PL_linestart || s[-1] == '\n') )
3326 PL_lex_formbrack = 0;
3330 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3336 yylval.ival = OPf_SPECIAL;
3342 if (PL_expect != XOPERATOR)
3347 case '0': case '1': case '2': case '3': case '4':
3348 case '5': case '6': case '7': case '8': case '9':
3350 if (PL_expect == XOPERATOR)
3355 s = scan_str(s,FALSE,FALSE);
3356 if (PL_expect == XOPERATOR) {
3357 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3360 return ','; /* grandfather non-comma-format format */
3366 missingterm((char*)0);
3367 yylval.ival = OP_CONST;
3368 TERM(sublex_start());
3371 s = scan_str(s,FALSE,FALSE);
3372 if (PL_expect == XOPERATOR) {
3373 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3376 return ','; /* grandfather non-comma-format format */
3382 missingterm((char*)0);
3383 yylval.ival = OP_CONST;
3384 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3385 if (*d == '$' || *d == '@' || *d == '\\' || *d & 0x80) {
3386 yylval.ival = OP_STRINGIFY;
3390 TERM(sublex_start());
3393 s = scan_str(s,FALSE,FALSE);
3394 if (PL_expect == XOPERATOR)
3395 no_op("Backticks",s);
3397 missingterm((char*)0);
3398 yylval.ival = OP_BACKTICK;
3400 TERM(sublex_start());
3404 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3405 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3407 if (PL_expect == XOPERATOR)
3408 no_op("Backslash",s);
3412 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3452 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3454 /* Some keywords can be followed by any delimiter, including ':' */
3455 tmp = (len == 1 && strchr("msyq", PL_tokenbuf[0]) ||
3456 len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3457 (PL_tokenbuf[0] == 'q' &&
3458 strchr("qwxr", PL_tokenbuf[1]))));
3460 /* x::* is just a word, unless x is "CORE" */
3461 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3465 while (d < PL_bufend && isSPACE(*d))
3466 d++; /* no comments skipped here, or s### is misparsed */
3468 /* Is this a label? */
3469 if (!tmp && PL_expect == XSTATE
3470 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3472 yylval.pval = savepv(PL_tokenbuf);
3477 /* Check for keywords */
3478 tmp = keyword(PL_tokenbuf, len);
3480 /* Is this a word before a => operator? */
3481 if (strnEQ(d,"=>",2)) {
3483 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3484 yylval.opval->op_private = OPpCONST_BARE;
3488 if (tmp < 0) { /* second-class keyword? */
3489 GV *ogv = Nullgv; /* override (winner) */
3490 GV *hgv = Nullgv; /* hidden (loser) */
3491 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3493 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3496 if (GvIMPORTED_CV(gv))
3498 else if (! CvMETHOD(cv))
3502 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3503 (gv = *gvp) != (GV*)&PL_sv_undef &&
3504 GvCVu(gv) && GvIMPORTED_CV(gv))
3510 tmp = 0; /* overridden by import or by GLOBAL */
3513 && -tmp==KEY_lock /* XXX generalizable kludge */
3514 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3516 tmp = 0; /* any sub overrides "weak" keyword */
3518 else { /* no override */
3522 if (ckWARN(WARN_AMBIGUOUS) && hgv
3523 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3524 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3525 "Ambiguous call resolved as CORE::%s(), %s",
3526 GvENAME(hgv), "qualify as such or use &");
3533 default: /* not a keyword */
3536 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3538 /* Get the rest if it looks like a package qualifier */
3540 if (*s == '\'' || *s == ':' && s[1] == ':') {
3542 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3545 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3546 *s == '\'' ? "'" : "::");
3550 if (PL_expect == XOPERATOR) {
3551 if (PL_bufptr == PL_linestart) {
3552 PL_curcop->cop_line--;
3553 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3554 PL_curcop->cop_line++;
3557 no_op("Bareword",s);
3560 /* Look for a subroutine with this name in current package,
3561 unless name is "Foo::", in which case Foo is a bearword
3562 (and a package name). */
3565 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3567 if (ckWARN(WARN_UNSAFE) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3568 Perl_warner(aTHX_ WARN_UNSAFE,
3569 "Bareword \"%s\" refers to nonexistent package",
3572 PL_tokenbuf[len] = '\0';
3579 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3582 /* if we saw a global override before, get the right name */
3585 sv = newSVpvn("CORE::GLOBAL::",14);
3586 sv_catpv(sv,PL_tokenbuf);
3589 sv = newSVpv(PL_tokenbuf,0);
3591 /* Presume this is going to be a bareword of some sort. */
3594 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3595 yylval.opval->op_private = OPpCONST_BARE;
3597 /* And if "Foo::", then that's what it certainly is. */
3602 /* See if it's the indirect object for a list operator. */
3604 if (PL_oldoldbufptr &&
3605 PL_oldoldbufptr < PL_bufptr &&
3606 (PL_oldoldbufptr == PL_last_lop || PL_oldoldbufptr == PL_last_uni) &&
3607 /* NO SKIPSPACE BEFORE HERE! */
3608 (PL_expect == XREF ||
3609 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3611 bool immediate_paren = *s == '(';
3613 /* (Now we can afford to cross potential line boundary.) */
3616 /* Two barewords in a row may indicate method call. */
3618 if ((isIDFIRST_lazy(s) || *s == '$') && (tmp=intuit_method(s,gv)))
3621 /* If not a declared subroutine, it's an indirect object. */
3622 /* (But it's an indir obj regardless for sort.) */
3624 if ((PL_last_lop_op == OP_SORT ||
3625 (!immediate_paren && (!gv || !GvCVu(gv)))) &&
3626 (PL_last_lop_op != OP_MAPSTART &&
3627 PL_last_lop_op != OP_GREPSTART))
3629 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3634 /* If followed by a paren, it's certainly a subroutine. */
3636 PL_expect = XOPERATOR;
3640 if (gv && GvCVu(gv)) {
3641 for (d = s + 1; *d == ' ' || *d == '\t'; d++) ;
3642 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3647 PL_nextval[PL_nexttoke].opval = yylval.opval;
3648 PL_expect = XOPERATOR;
3654 /* If followed by var or block, call it a method (unless sub) */
3656 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3657 PL_last_lop = PL_oldbufptr;
3658 PL_last_lop_op = OP_METHOD;
3662 /* If followed by a bareword, see if it looks like indir obj. */
3664 if ((isIDFIRST_lazy(s) || *s == '$') && (tmp = intuit_method(s,gv)))
3667 /* Not a method, so call it a subroutine (if defined) */
3669 if (gv && GvCVu(gv)) {
3671 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
3672 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3673 "Ambiguous use of -%s resolved as -&%s()",
3674 PL_tokenbuf, PL_tokenbuf);
3675 /* Check for a constant sub */
3677 if ((sv = cv_const_sv(cv))) {
3679 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
3680 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
3681 yylval.opval->op_private = 0;
3685 /* Resolve to GV now. */
3686 op_free(yylval.opval);
3687 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
3688 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
3689 PL_last_lop = PL_oldbufptr;
3690 PL_last_lop_op = OP_ENTERSUB;
3691 /* Is there a prototype? */
3694 char *proto = SvPV((SV*)cv, len);
3697 if (strEQ(proto, "$"))
3699 if (*proto == '&' && *s == '{') {
3700 sv_setpv(PL_subname,"__ANON__");
3704 PL_nextval[PL_nexttoke].opval = yylval.opval;
3710 /* Call it a bare word */
3712 if (PL_hints & HINT_STRICT_SUBS)
3713 yylval.opval->op_private |= OPpCONST_STRICT;
3716 if (ckWARN(WARN_RESERVED)) {
3717 if (lastchar != '-') {
3718 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
3720 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
3727 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
3728 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3729 "Operator or semicolon missing before %c%s",
3730 lastchar, PL_tokenbuf);
3731 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3732 "Ambiguous use of %c resolved as operator %c",
3733 lastchar, lastchar);
3739 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3740 newSVsv(GvSV(PL_curcop->cop_filegv)));
3744 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3745 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)PL_curcop->cop_line));
3748 case KEY___PACKAGE__:
3749 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3751 ? newSVsv(PL_curstname)
3760 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
3761 char *pname = "main";
3762 if (PL_tokenbuf[2] == 'D')
3763 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
3764 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
3767 GvIOp(gv) = newIO();
3768 IoIFP(GvIOp(gv)) = PL_rsfp;
3769 #if defined(HAS_FCNTL) && defined(F_SETFD)
3771 int fd = PerlIO_fileno(PL_rsfp);
3772 fcntl(fd,F_SETFD,fd >= 3);
3775 /* Mark this internal pseudo-handle as clean */
3776 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
3778 IoTYPE(GvIOp(gv)) = '|';
3779 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
3780 IoTYPE(GvIOp(gv)) = '-';
3782 IoTYPE(GvIOp(gv)) = '<';
3793 if (PL_expect == XSTATE) {
3800 if (*s == ':' && s[1] == ':') {
3803 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3804 tmp = keyword(PL_tokenbuf, len);
3818 LOP(OP_ACCEPT,XTERM);
3824 LOP(OP_ATAN2,XTERM);
3833 LOP(OP_BLESS,XTERM);
3842 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
3859 if (!PL_cryptseen++)
3862 LOP(OP_CRYPT,XTERM);
3865 if (ckWARN(WARN_OCTAL)) {
3866 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
3867 if (*d != '0' && isDIGIT(*d))
3868 Perl_warner(aTHX_ WARN_OCTAL,
3869 "chmod: mode argument is missing initial 0");
3871 LOP(OP_CHMOD,XTERM);
3874 LOP(OP_CHOWN,XTERM);
3877 LOP(OP_CONNECT,XTERM);
3893 s = force_word(s,WORD,FALSE,TRUE,FALSE);
3897 PL_hints |= HINT_BLOCK_SCOPE;
3907 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
3908 LOP(OP_DBMOPEN,XTERM);
3914 s = force_word(s,WORD,TRUE,FALSE,FALSE);
3921 yylval.ival = PL_curcop->cop_line;
3935 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
3936 UNIBRACK(OP_ENTEREVAL);
3951 case KEY_endhostent:
3957 case KEY_endservent:
3960 case KEY_endprotoent:
3971 yylval.ival = PL_curcop->cop_line;
3973 if (PL_expect == XSTATE && isIDFIRST_lazy(s)) {
3975 if ((PL_bufend - p) >= 3 &&
3976 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
3978 else if ((PL_bufend - p) >= 4 &&
3979 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
3982 if (isIDFIRST_lazy(p)) {
3983 p = scan_ident(p, PL_bufend,
3984 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3988 Perl_croak(aTHX_ "Missing $ on loop variable");
3993 LOP(OP_FORMLINE,XTERM);
3999 LOP(OP_FCNTL,XTERM);
4005 LOP(OP_FLOCK,XTERM);
4014 LOP(OP_GREPSTART, *s == '(' ? XTERM : XREF);
4017 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4032 case KEY_getpriority:
4033 LOP(OP_GETPRIORITY,XTERM);
4035 case KEY_getprotobyname:
4038 case KEY_getprotobynumber:
4039 LOP(OP_GPBYNUMBER,XTERM);
4041 case KEY_getprotoent:
4053 case KEY_getpeername:
4054 UNI(OP_GETPEERNAME);
4056 case KEY_gethostbyname:
4059 case KEY_gethostbyaddr:
4060 LOP(OP_GHBYADDR,XTERM);
4062 case KEY_gethostent:
4065 case KEY_getnetbyname:
4068 case KEY_getnetbyaddr:
4069 LOP(OP_GNBYADDR,XTERM);
4074 case KEY_getservbyname:
4075 LOP(OP_GSBYNAME,XTERM);
4077 case KEY_getservbyport:
4078 LOP(OP_GSBYPORT,XTERM);
4080 case KEY_getservent:
4083 case KEY_getsockname:
4084 UNI(OP_GETSOCKNAME);
4086 case KEY_getsockopt:
4087 LOP(OP_GSOCKOPT,XTERM);
4109 yylval.ival = PL_curcop->cop_line;
4113 LOP(OP_INDEX,XTERM);
4119 LOP(OP_IOCTL,XTERM);
4131 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4163 LOP(OP_LISTEN,XTERM);
4172 s = scan_pat(s,OP_MATCH);
4173 TERM(sublex_start());
4176 LOP(OP_MAPSTART, *s == '(' ? XTERM : XREF);
4179 LOP(OP_MKDIR,XTERM);
4182 LOP(OP_MSGCTL,XTERM);
4185 LOP(OP_MSGGET,XTERM);
4188 LOP(OP_MSGRCV,XTERM);
4191 LOP(OP_MSGSND,XTERM);
4197 if (isIDFIRST_lazy(s)) {
4198 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4199 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4201 PL_in_my_stash = gv_stashpv(PL_tokenbuf, FALSE);
4202 if (!PL_in_my_stash) {
4205 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4213 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4220 if (PL_expect != XSTATE)
4221 yyerror("\"no\" not allowed in expression");
4222 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4223 s = force_version(s);
4232 if (isIDFIRST_lazy(s)) {
4234 for (d = s; isALNUM_lazy(d); d++) ;
4236 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_AMBIGUOUS))
4237 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4238 "Precedence problem: open %.*s should be open(%.*s)",
4244 yylval.ival = OP_OR;
4254 LOP(OP_OPEN_DIR,XTERM);
4257 checkcomma(s,PL_tokenbuf,"filehandle");
4261 checkcomma(s,PL_tokenbuf,"filehandle");
4280 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4284 LOP(OP_PIPE_OP,XTERM);
4287 s = scan_str(s,FALSE,FALSE);
4289 missingterm((char*)0);
4290 yylval.ival = OP_CONST;
4291 TERM(sublex_start());
4297 s = scan_str(s,FALSE,FALSE);
4299 missingterm((char*)0);
4301 if (SvCUR(PL_lex_stuff)) {
4304 d = SvPV_force(PL_lex_stuff, len);
4306 for (; isSPACE(*d) && len; --len, ++d) ;
4309 if (!warned && ckWARN(WARN_SYNTAX)) {
4310 for (; !isSPACE(*d) && len; --len, ++d) {
4312 Perl_warner(aTHX_ WARN_SYNTAX,
4313 "Possible attempt to separate words with commas");
4316 else if (*d == '#') {
4317 Perl_warner(aTHX_ WARN_SYNTAX,
4318 "Possible attempt to put comments in qw() list");
4324 for (; !isSPACE(*d) && len; --len, ++d) ;
4326 words = append_elem(OP_LIST, words,
4327 newSVOP(OP_CONST, 0, newSVpvn(b, d-b)));
4331 PL_nextval[PL_nexttoke].opval = words;
4336 SvREFCNT_dec(PL_lex_stuff);
4337 PL_lex_stuff = Nullsv;
4342 s = scan_str(s,FALSE,FALSE);
4344 missingterm((char*)0);
4345 yylval.ival = OP_STRINGIFY;
4346 if (SvIVX(PL_lex_stuff) == '\'')
4347 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4348 TERM(sublex_start());
4351 s = scan_pat(s,OP_QR);
4352 TERM(sublex_start());
4355 s = scan_str(s,FALSE,FALSE);
4357 missingterm((char*)0);
4358 yylval.ival = OP_BACKTICK;
4360 TERM(sublex_start());
4366 *PL_tokenbuf = '\0';
4367 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4368 if (isIDFIRST_lazy(PL_tokenbuf))
4369 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4371 yyerror("<> should be quotes");
4378 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4382 LOP(OP_RENAME,XTERM);
4391 LOP(OP_RINDEX,XTERM);
4414 LOP(OP_REVERSE,XTERM);
4425 TERM(sublex_start());
4427 TOKEN(1); /* force error */
4436 LOP(OP_SELECT,XTERM);
4442 LOP(OP_SEMCTL,XTERM);
4445 LOP(OP_SEMGET,XTERM);
4448 LOP(OP_SEMOP,XTERM);
4454 LOP(OP_SETPGRP,XTERM);
4456 case KEY_setpriority:
4457 LOP(OP_SETPRIORITY,XTERM);
4459 case KEY_sethostent:
4465 case KEY_setservent:
4468 case KEY_setprotoent:
4478 LOP(OP_SEEKDIR,XTERM);
4480 case KEY_setsockopt:
4481 LOP(OP_SSOCKOPT,XTERM);
4487 LOP(OP_SHMCTL,XTERM);
4490 LOP(OP_SHMGET,XTERM);
4493 LOP(OP_SHMREAD,XTERM);
4496 LOP(OP_SHMWRITE,XTERM);
4499 LOP(OP_SHUTDOWN,XTERM);
4508 LOP(OP_SOCKET,XTERM);
4510 case KEY_socketpair:
4511 LOP(OP_SOCKPAIR,XTERM);
4514 checkcomma(s,PL_tokenbuf,"subroutine name");
4516 if (*s == ';' || *s == ')') /* probably a close */
4517 Perl_croak(aTHX_ "sort is now a reserved word");
4519 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4523 LOP(OP_SPLIT,XTERM);
4526 LOP(OP_SPRINTF,XTERM);
4529 LOP(OP_SPLICE,XTERM);
4545 LOP(OP_SUBSTR,XTERM);
4551 char tmpbuf[sizeof PL_tokenbuf];
4553 expectation attrful;
4554 bool have_name, have_proto;
4559 if (isIDFIRST_lazy(s) || *s == '\'' ||
4560 (*s == ':' && s[1] == ':'))
4563 attrful = XATTRBLOCK;
4564 /* remember buffer pos'n for later force_word */
4565 tboffset = s - PL_oldbufptr;
4566 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4567 if (strchr(tmpbuf, ':'))
4568 sv_setpv(PL_subname, tmpbuf);
4570 sv_setsv(PL_subname,PL_curstname);
4571 sv_catpvn(PL_subname,"::",2);
4572 sv_catpvn(PL_subname,tmpbuf,len);
4579 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4580 PL_expect = XTERMBLOCK;
4581 attrful = XATTRTERM;
4582 sv_setpv(PL_subname,"?");
4586 if (key == KEY_format) {
4588 PL_lex_formbrack = PL_lex_brackets + 1;
4590 (void) force_word(PL_oldbufptr + tboffset, WORD,
4595 /* Look for a prototype */
4599 s = scan_str(s,FALSE,FALSE);
4602 SvREFCNT_dec(PL_lex_stuff);
4603 PL_lex_stuff = Nullsv;
4604 Perl_croak(aTHX_ "Prototype not terminated");
4607 d = SvPVX(PL_lex_stuff);
4609 for (p = d; *p; ++p) {
4614 SvCUR(PL_lex_stuff) = tmp;
4622 if (*s == ':' && s[1] != ':')
4623 PL_expect = attrful;
4626 PL_nextval[PL_nexttoke].opval =
4627 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
4628 PL_lex_stuff = Nullsv;
4632 sv_setpv(PL_subname,"__ANON__");
4635 (void) force_word(PL_oldbufptr + tboffset, WORD,
4644 LOP(OP_SYSTEM,XREF);
4647 LOP(OP_SYMLINK,XTERM);
4650 LOP(OP_SYSCALL,XTERM);
4653 LOP(OP_SYSOPEN,XTERM);
4656 LOP(OP_SYSSEEK,XTERM);
4659 LOP(OP_SYSREAD,XTERM);
4662 LOP(OP_SYSWRITE,XTERM);
4666 TERM(sublex_start());
4687 LOP(OP_TRUNCATE,XTERM);
4699 yylval.ival = PL_curcop->cop_line;
4703 yylval.ival = PL_curcop->cop_line;
4707 LOP(OP_UNLINK,XTERM);
4713 LOP(OP_UNPACK,XTERM);
4716 LOP(OP_UTIME,XTERM);
4719 if (ckWARN(WARN_OCTAL)) {
4720 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4721 if (*d != '0' && isDIGIT(*d))
4722 Perl_warner(aTHX_ WARN_OCTAL,
4723 "umask: argument is missing initial 0");
4728 LOP(OP_UNSHIFT,XTERM);
4731 if (PL_expect != XSTATE)
4732 yyerror("\"use\" not allowed in expression");
4735 s = force_version(s);
4736 if(*s == ';' || (s = skipspace(s), *s == ';')) {
4737 PL_nextval[PL_nexttoke].opval = Nullop;
4742 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4743 s = force_version(s);
4756 yylval.ival = PL_curcop->cop_line;
4760 PL_hints |= HINT_BLOCK_SCOPE;
4767 LOP(OP_WAITPID,XTERM);
4775 static char ctl_l[2];
4777 if (ctl_l[0] == '\0')
4778 ctl_l[0] = toCTRL('L');
4779 gv_fetchpv(ctl_l,TRUE, SVt_PV);
4782 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
4787 if (PL_expect == XOPERATOR)
4793 yylval.ival = OP_XOR;
4798 TERM(sublex_start());
4804 Perl_keyword(pTHX_ register char *d, I32 len)
4809 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
4810 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
4811 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
4812 if (strEQ(d,"__DATA__")) return KEY___DATA__;
4813 if (strEQ(d,"__END__")) return KEY___END__;
4817 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
4822 if (strEQ(d,"and")) return -KEY_and;
4823 if (strEQ(d,"abs")) return -KEY_abs;
4826 if (strEQ(d,"alarm")) return -KEY_alarm;
4827 if (strEQ(d,"atan2")) return -KEY_atan2;
4830 if (strEQ(d,"accept")) return -KEY_accept;
4835 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
4838 if (strEQ(d,"bless")) return -KEY_bless;
4839 if (strEQ(d,"bind")) return -KEY_bind;
4840 if (strEQ(d,"binmode")) return -KEY_binmode;
4843 if (strEQ(d,"CORE")) return -KEY_CORE;
4848 if (strEQ(d,"cmp")) return -KEY_cmp;
4849 if (strEQ(d,"chr")) return -KEY_chr;
4850 if (strEQ(d,"cos")) return -KEY_cos;
4853 if (strEQ(d,"chop")) return KEY_chop;
4856 if (strEQ(d,"close")) return -KEY_close;
4857 if (strEQ(d,"chdir")) return -KEY_chdir;
4858 if (strEQ(d,"chomp")) return KEY_chomp;
4859 if (strEQ(d,"chmod")) return -KEY_chmod;
4860 if (strEQ(d,"chown")) return -KEY_chown;
4861 if (strEQ(d,"crypt")) return -KEY_crypt;
4864 if (strEQ(d,"chroot")) return -KEY_chroot;
4865 if (strEQ(d,"caller")) return -KEY_caller;
4868 if (strEQ(d,"connect")) return -KEY_connect;
4871 if (strEQ(d,"closedir")) return -KEY_closedir;
4872 if (strEQ(d,"continue")) return -KEY_continue;
4877 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
4882 if (strEQ(d,"do")) return KEY_do;
4885 if (strEQ(d,"die")) return -KEY_die;
4888 if (strEQ(d,"dump")) return -KEY_dump;
4891 if (strEQ(d,"delete")) return KEY_delete;
4894 if (strEQ(d,"defined")) return KEY_defined;
4895 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
4898 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
4903 if (strEQ(d,"EQ")) { deprecate(d); return -KEY_eq;}
4904 if (strEQ(d,"END")) return KEY_END;
4909 if (strEQ(d,"eq")) return -KEY_eq;
4912 if (strEQ(d,"eof")) return -KEY_eof;
4913 if (strEQ(d,"exp")) return -KEY_exp;
4916 if (strEQ(d,"else")) return KEY_else;
4917 if (strEQ(d,"exit")) return -KEY_exit;
4918 if (strEQ(d,"eval")) return KEY_eval;
4919 if (strEQ(d,"exec")) return -KEY_exec;
4920 if (strEQ(d,"each")) return KEY_each;
4923 if (strEQ(d,"elsif")) return KEY_elsif;
4926 if (strEQ(d,"exists")) return KEY_exists;
4927 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
4930 if (strEQ(d,"endgrent")) return -KEY_endgrent;
4931 if (strEQ(d,"endpwent")) return -KEY_endpwent;
4934 if (strEQ(d,"endnetent")) return -KEY_endnetent;
4937 if (strEQ(d,"endhostent")) return -KEY_endhostent;
4938 if (strEQ(d,"endservent")) return -KEY_endservent;
4941 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
4948 if (strEQ(d,"for")) return KEY_for;
4951 if (strEQ(d,"fork")) return -KEY_fork;
4954 if (strEQ(d,"fcntl")) return -KEY_fcntl;
4955 if (strEQ(d,"flock")) return -KEY_flock;
4958 if (strEQ(d,"format")) return KEY_format;
4959 if (strEQ(d,"fileno")) return -KEY_fileno;
4962 if (strEQ(d,"foreach")) return KEY_foreach;
4965 if (strEQ(d,"formline")) return -KEY_formline;
4971 if (strEQ(d,"GT")) { deprecate(d); return -KEY_gt;}
4972 if (strEQ(d,"GE")) { deprecate(d); return -KEY_ge;}
4976 if (strnEQ(d,"get",3)) {
4981 if (strEQ(d,"ppid")) return -KEY_getppid;
4982 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
4985 if (strEQ(d,"pwent")) return -KEY_getpwent;
4986 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
4987 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
4990 if (strEQ(d,"peername")) return -KEY_getpeername;
4991 if (strEQ(d,"protoent")) return -KEY_getprotoent;
4992 if (strEQ(d,"priority")) return -KEY_getpriority;
4995 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
4998 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5002 else if (*d == 'h') {
5003 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5004 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5005 if (strEQ(d,"hostent")) return -KEY_gethostent;
5007 else if (*d == 'n') {
5008 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5009 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5010 if (strEQ(d,"netent")) return -KEY_getnetent;
5012 else if (*d == 's') {
5013 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5014 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5015 if (strEQ(d,"servent")) return -KEY_getservent;
5016 if (strEQ(d,"sockname")) return -KEY_getsockname;
5017 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5019 else if (*d == 'g') {
5020 if (strEQ(d,"grent")) return -KEY_getgrent;
5021 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5022 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5024 else if (*d == 'l') {
5025 if (strEQ(d,"login")) return -KEY_getlogin;
5027 else if (strEQ(d,"c")) return -KEY_getc;
5032 if (strEQ(d,"gt")) return -KEY_gt;
5033 if (strEQ(d,"ge")) return -KEY_ge;
5036 if (strEQ(d,"grep")) return KEY_grep;
5037 if (strEQ(d,"goto")) return KEY_goto;
5038 if (strEQ(d,"glob")) return KEY_glob;
5041 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5046 if (strEQ(d,"hex")) return -KEY_hex;
5049 if (strEQ(d,"INIT")) return KEY_INIT;
5054 if (strEQ(d,"if")) return KEY_if;
5057 if (strEQ(d,"int")) return -KEY_int;
5060 if (strEQ(d,"index")) return -KEY_index;
5061 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5066 if (strEQ(d,"join")) return -KEY_join;
5070 if (strEQ(d,"keys")) return KEY_keys;
5071 if (strEQ(d,"kill")) return -KEY_kill;
5076 if (strEQ(d,"LT")) { deprecate(d); return -KEY_lt;}
5077 if (strEQ(d,"LE")) { deprecate(d); return -KEY_le;}
5083 if (strEQ(d,"lt")) return -KEY_lt;
5084 if (strEQ(d,"le")) return -KEY_le;
5085 if (strEQ(d,"lc")) return -KEY_lc;
5088 if (strEQ(d,"log")) return -KEY_log;
5091 if (strEQ(d,"last")) return KEY_last;
5092 if (strEQ(d,"link")) return -KEY_link;
5093 if (strEQ(d,"lock")) return -KEY_lock;
5096 if (strEQ(d,"local")) return KEY_local;
5097 if (strEQ(d,"lstat")) return -KEY_lstat;
5100 if (strEQ(d,"length")) return -KEY_length;
5101 if (strEQ(d,"listen")) return -KEY_listen;
5104 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5107 if (strEQ(d,"localtime")) return -KEY_localtime;
5113 case 1: return KEY_m;
5115 if (strEQ(d,"my")) return KEY_my;
5118 if (strEQ(d,"map")) return KEY_map;
5121 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5124 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5125 if (strEQ(d,"msgget")) return -KEY_msgget;
5126 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5127 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5132 if (strEQ(d,"NE")) { deprecate(d); return -KEY_ne;}
5135 if (strEQ(d,"next")) return KEY_next;
5136 if (strEQ(d,"ne")) return -KEY_ne;
5137 if (strEQ(d,"not")) return -KEY_not;
5138 if (strEQ(d,"no")) return KEY_no;
5143 if (strEQ(d,"or")) return -KEY_or;
5146 if (strEQ(d,"ord")) return -KEY_ord;
5147 if (strEQ(d,"oct")) return -KEY_oct;
5148 if (strEQ(d,"our")) return KEY_our;
5151 if (strEQ(d,"open")) return -KEY_open;
5154 if (strEQ(d,"opendir")) return -KEY_opendir;
5161 if (strEQ(d,"pop")) return KEY_pop;
5162 if (strEQ(d,"pos")) return KEY_pos;
5165 if (strEQ(d,"push")) return KEY_push;
5166 if (strEQ(d,"pack")) return -KEY_pack;
5167 if (strEQ(d,"pipe")) return -KEY_pipe;
5170 if (strEQ(d,"print")) return KEY_print;
5173 if (strEQ(d,"printf")) return KEY_printf;
5176 if (strEQ(d,"package")) return KEY_package;
5179 if (strEQ(d,"prototype")) return KEY_prototype;
5184 if (strEQ(d,"q")) return KEY_q;
5185 if (strEQ(d,"qr")) return KEY_qr;
5186 if (strEQ(d,"qq")) return KEY_qq;
5187 if (strEQ(d,"qw")) return KEY_qw;
5188 if (strEQ(d,"qx")) return KEY_qx;
5190 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5195 if (strEQ(d,"ref")) return -KEY_ref;
5198 if (strEQ(d,"read")) return -KEY_read;
5199 if (strEQ(d,"rand")) return -KEY_rand;
5200 if (strEQ(d,"recv")) return -KEY_recv;
5201 if (strEQ(d,"redo")) return KEY_redo;
5204 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5205 if (strEQ(d,"reset")) return -KEY_reset;
5208 if (strEQ(d,"return")) return KEY_return;
5209 if (strEQ(d,"rename")) return -KEY_rename;
5210 if (strEQ(d,"rindex")) return -KEY_rindex;
5213 if (strEQ(d,"require")) return -KEY_require;
5214 if (strEQ(d,"reverse")) return -KEY_reverse;
5215 if (strEQ(d,"readdir")) return -KEY_readdir;
5218 if (strEQ(d,"readlink")) return -KEY_readlink;
5219 if (strEQ(d,"readline")) return -KEY_readline;
5220 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5223 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5229 case 0: return KEY_s;
5231 if (strEQ(d,"scalar")) return KEY_scalar;
5236 if (strEQ(d,"seek")) return -KEY_seek;
5237 if (strEQ(d,"send")) return -KEY_send;
5240 if (strEQ(d,"semop")) return -KEY_semop;
5243 if (strEQ(d,"select")) return -KEY_select;
5244 if (strEQ(d,"semctl")) return -KEY_semctl;
5245 if (strEQ(d,"semget")) return -KEY_semget;
5248 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5249 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5252 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5253 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5256 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5259 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5260 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5261 if (strEQ(d,"setservent")) return -KEY_setservent;
5264 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5265 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5272 if (strEQ(d,"shift")) return KEY_shift;
5275 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5276 if (strEQ(d,"shmget")) return -KEY_shmget;
5279 if (strEQ(d,"shmread")) return -KEY_shmread;
5282 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5283 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5288 if (strEQ(d,"sin")) return -KEY_sin;
5291 if (strEQ(d,"sleep")) return -KEY_sleep;
5294 if (strEQ(d,"sort")) return KEY_sort;
5295 if (strEQ(d,"socket")) return -KEY_socket;
5296 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5299 if (strEQ(d,"split")) return KEY_split;
5300 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5301 if (strEQ(d,"splice")) return KEY_splice;
5304 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5307 if (strEQ(d,"srand")) return -KEY_srand;
5310 if (strEQ(d,"stat")) return -KEY_stat;
5311 if (strEQ(d,"study")) return KEY_study;
5314 if (strEQ(d,"substr")) return -KEY_substr;
5315 if (strEQ(d,"sub")) return KEY_sub;
5320 if (strEQ(d,"system")) return -KEY_system;
5323 if (strEQ(d,"symlink")) return -KEY_symlink;
5324 if (strEQ(d,"syscall")) return -KEY_syscall;
5325 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5326 if (strEQ(d,"sysread")) return -KEY_sysread;
5327 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5330 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5339 if (strEQ(d,"tr")) return KEY_tr;
5342 if (strEQ(d,"tie")) return KEY_tie;
5345 if (strEQ(d,"tell")) return -KEY_tell;
5346 if (strEQ(d,"tied")) return KEY_tied;
5347 if (strEQ(d,"time")) return -KEY_time;
5350 if (strEQ(d,"times")) return -KEY_times;
5353 if (strEQ(d,"telldir")) return -KEY_telldir;
5356 if (strEQ(d,"truncate")) return -KEY_truncate;
5363 if (strEQ(d,"uc")) return -KEY_uc;
5366 if (strEQ(d,"use")) return KEY_use;
5369 if (strEQ(d,"undef")) return KEY_undef;
5370 if (strEQ(d,"until")) return KEY_until;
5371 if (strEQ(d,"untie")) return KEY_untie;
5372 if (strEQ(d,"utime")) return -KEY_utime;
5373 if (strEQ(d,"umask")) return -KEY_umask;
5376 if (strEQ(d,"unless")) return KEY_unless;
5377 if (strEQ(d,"unpack")) return -KEY_unpack;
5378 if (strEQ(d,"unlink")) return -KEY_unlink;
5381 if (strEQ(d,"unshift")) return KEY_unshift;
5382 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5387 if (strEQ(d,"values")) return -KEY_values;
5388 if (strEQ(d,"vec")) return -KEY_vec;
5393 if (strEQ(d,"warn")) return -KEY_warn;
5394 if (strEQ(d,"wait")) return -KEY_wait;
5397 if (strEQ(d,"while")) return KEY_while;
5398 if (strEQ(d,"write")) return -KEY_write;
5401 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5404 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5409 if (len == 1) return -KEY_x;
5410 if (strEQ(d,"xor")) return -KEY_xor;
5413 if (len == 1) return KEY_y;
5422 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5426 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5427 dTHR; /* only for ckWARN */
5428 if (ckWARN(WARN_SYNTAX)) {
5430 for (w = s+2; *w && level; w++) {
5437 for (; *w && isSPACE(*w); w++) ;
5438 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5439 Perl_warner(aTHX_ WARN_SYNTAX, "%s (...) interpreted as function",name);
5442 while (s < PL_bufend && isSPACE(*s))
5446 while (s < PL_bufend && isSPACE(*s))
5448 if (isIDFIRST_lazy(s)) {
5450 while (isALNUM_lazy(s))
5452 while (s < PL_bufend && isSPACE(*s))
5457 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5461 Perl_croak(aTHX_ "No comma allowed after %s", what);
5466 /* Either returns sv, or mortalizes sv and returns a new SV*.
5467 Best used as sv=new_constant(..., sv, ...).
5468 If s, pv are NULL, calls subroutine with one argument,
5469 and type is used with error messages only. */
5472 S_new_constant(pTHX_ char *s, STRLEN len, char *key, SV *sv, SV *pv, char *type)
5475 HV *table = GvHV(PL_hintgv); /* ^H */
5479 char *why, *why1, *why2;
5481 if (!(PL_hints & HINT_LOCALIZE_HH)) {
5484 why = "%^H is not localized";
5488 msg = Perl_newSVpvf(aTHX_ "constant(%s): %s%s%s",
5489 (type ? type: "undef"), why1, why2, why);
5490 yyerror(SvPVX(msg));
5495 why = "%^H is not defined";
5498 cvp = hv_fetch(table, key, strlen(key), FALSE);
5499 if (!cvp || !SvOK(*cvp)) {
5500 why = "} is not defined";
5505 sv_2mortal(sv); /* Parent created it permanently */
5508 pv = sv_2mortal(newSVpvn(s, len));
5510 typesv = sv_2mortal(newSVpv(type, 0));
5512 typesv = &PL_sv_undef;
5514 PUSHSTACKi(PERLSI_OVERLOAD);
5527 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5531 /* Check the eval first */
5532 if (!PL_in_eval && SvTRUE(ERRSV))
5535 sv_catpv(ERRSV, "Propagated");
5536 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5538 res = SvREFCNT_inc(sv);
5551 why = "}} did not return a defined value";
5552 why1 = "Call to &{$^H{";
5562 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
5564 register char *d = dest;
5565 register char *e = d + destlen - 3; /* two-character token, ending NUL */
5568 Perl_croak(aTHX_ ident_too_long);
5569 if (isALNUM(*s)) /* UTF handled below */
5571 else if (*s == '\'' && allow_package && isIDFIRST_lazy(s+1)) {
5576 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
5580 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5581 char *t = s + UTF8SKIP(s);
5582 while (*t & 0x80 && is_utf8_mark((U8*)t))
5584 if (d + (t - s) > e)
5585 Perl_croak(aTHX_ ident_too_long);
5586 Copy(s, d, t - s, char);
5599 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
5606 if (PL_lex_brackets == 0)
5607 PL_lex_fakebrack = 0;
5611 e = d + destlen - 3; /* two-character token, ending NUL */
5613 while (isDIGIT(*s)) {
5615 Perl_croak(aTHX_ ident_too_long);
5622 Perl_croak(aTHX_ ident_too_long);
5623 if (isALNUM(*s)) /* UTF handled below */
5625 else if (*s == '\'' && isIDFIRST_lazy(s+1)) {
5630 else if (*s == ':' && s[1] == ':') {
5634 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5635 char *t = s + UTF8SKIP(s);
5636 while (*t & 0x80 && is_utf8_mark((U8*)t))
5638 if (d + (t - s) > e)
5639 Perl_croak(aTHX_ ident_too_long);
5640 Copy(s, d, t - s, char);
5651 if (PL_lex_state != LEX_NORMAL)
5652 PL_lex_state = LEX_INTERPENDMAYBE;
5655 if (*s == '$' && s[1] &&
5656 (isALNUM_lazy(s+1) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
5669 if (*d == '^' && *s && isCONTROLVAR(*s)) {
5674 if (isSPACE(s[-1])) {
5677 if (ch != ' ' && ch != '\t') {
5683 if (isIDFIRST_lazy(d)) {
5687 while (e < send && isALNUM_lazy(e) || *e == ':') {
5689 while (e < send && *e & 0x80 && is_utf8_mark((U8*)e))
5692 Copy(s, d, e - s, char);
5697 while ((isALNUM(*s) || *s == ':') && d < e)
5700 Perl_croak(aTHX_ ident_too_long);
5703 while (s < send && (*s == ' ' || *s == '\t')) s++;
5704 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
5705 dTHR; /* only for ckWARN */
5706 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
5707 char *brack = *s == '[' ? "[...]" : "{...}";
5708 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5709 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
5710 funny, dest, brack, funny, dest, brack);
5712 PL_lex_fakebrack = PL_lex_brackets+1;
5714 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
5718 /* Handle extended ${^Foo} variables
5719 * 1999-02-27 mjd-perl-patch@plover.com */
5720 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
5724 while (isALNUM(*s) && d < e) {
5728 Perl_croak(aTHX_ ident_too_long);
5733 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
5734 PL_lex_state = LEX_INTERPEND;
5737 if (PL_lex_state == LEX_NORMAL) {
5738 dTHR; /* only for ckWARN */
5739 if (ckWARN(WARN_AMBIGUOUS) &&
5740 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
5742 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5743 "Ambiguous use of %c{%s} resolved to %c%s",
5744 funny, dest, funny, dest);
5749 s = bracket; /* let the parser handle it */
5753 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
5754 PL_lex_state = LEX_INTERPEND;
5759 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
5764 *pmfl |= PMf_GLOBAL;
5766 *pmfl |= PMf_CONTINUE;
5770 *pmfl |= PMf_MULTILINE;
5772 *pmfl |= PMf_SINGLELINE;
5774 *pmfl |= PMf_EXTENDED;
5778 S_scan_pat(pTHX_ char *start, I32 type)
5783 s = scan_str(start,FALSE,FALSE);
5786 SvREFCNT_dec(PL_lex_stuff);
5787 PL_lex_stuff = Nullsv;
5788 Perl_croak(aTHX_ "Search pattern not terminated");
5791 pm = (PMOP*)newPMOP(type, 0);
5792 if (PL_multi_open == '?')
5793 pm->op_pmflags |= PMf_ONCE;
5795 while (*s && strchr("iomsx", *s))
5796 pmflag(&pm->op_pmflags,*s++);
5799 while (*s && strchr("iogcmsx", *s))
5800 pmflag(&pm->op_pmflags,*s++);
5802 pm->op_pmpermflags = pm->op_pmflags;
5804 PL_lex_op = (OP*)pm;
5805 yylval.ival = OP_MATCH;
5810 S_scan_subst(pTHX_ char *start)
5817 yylval.ival = OP_NULL;
5819 s = scan_str(start,FALSE,FALSE);
5823 SvREFCNT_dec(PL_lex_stuff);
5824 PL_lex_stuff = Nullsv;
5825 Perl_croak(aTHX_ "Substitution pattern not terminated");
5828 if (s[-1] == PL_multi_open)
5831 first_start = PL_multi_start;
5832 s = scan_str(s,FALSE,FALSE);
5835 SvREFCNT_dec(PL_lex_stuff);
5836 PL_lex_stuff = Nullsv;
5838 SvREFCNT_dec(PL_lex_repl);
5839 PL_lex_repl = Nullsv;
5840 Perl_croak(aTHX_ "Substitution replacement not terminated");
5842 PL_multi_start = first_start; /* so whole substitution is taken together */
5844 pm = (PMOP*)newPMOP(OP_SUBST, 0);
5850 else if (strchr("iogcmsx", *s))
5851 pmflag(&pm->op_pmflags,*s++);
5858 PL_sublex_info.super_bufptr = s;
5859 PL_sublex_info.super_bufend = PL_bufend;
5861 pm->op_pmflags |= PMf_EVAL;
5862 repl = newSVpvn("",0);
5864 sv_catpv(repl, es ? "eval " : "do ");
5865 sv_catpvn(repl, "{ ", 2);
5866 sv_catsv(repl, PL_lex_repl);
5867 sv_catpvn(repl, " };", 2);
5869 SvREFCNT_dec(PL_lex_repl);
5873 pm->op_pmpermflags = pm->op_pmflags;
5874 PL_lex_op = (OP*)pm;
5875 yylval.ival = OP_SUBST;
5880 S_scan_trans(pTHX_ char *start)
5891 yylval.ival = OP_NULL;
5893 s = scan_str(start,FALSE,FALSE);
5896 SvREFCNT_dec(PL_lex_stuff);
5897 PL_lex_stuff = Nullsv;
5898 Perl_croak(aTHX_ "Transliteration pattern not terminated");
5900 if (s[-1] == PL_multi_open)
5903 s = scan_str(s,FALSE,FALSE);
5906 SvREFCNT_dec(PL_lex_stuff);
5907 PL_lex_stuff = Nullsv;
5909 SvREFCNT_dec(PL_lex_repl);
5910 PL_lex_repl = Nullsv;
5911 Perl_croak(aTHX_ "Transliteration replacement not terminated");
5915 o = newSVOP(OP_TRANS, 0, 0);
5916 utf8 = OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF;
5919 New(803,tbl,256,short);
5920 o = newPVOP(OP_TRANS, 0, (char*)tbl);
5924 complement = del = squash = 0;
5925 while (strchr("cdsCU", *s)) {
5927 complement = OPpTRANS_COMPLEMENT;
5929 del = OPpTRANS_DELETE;
5931 squash = OPpTRANS_SQUASH;
5936 utf8 &= ~OPpTRANS_FROM_UTF;
5938 utf8 |= OPpTRANS_FROM_UTF;
5942 utf8 &= ~OPpTRANS_TO_UTF;
5944 utf8 |= OPpTRANS_TO_UTF;
5947 Perl_croak(aTHX_ "Too many /C and /U options");
5952 o->op_private = del|squash|complement|utf8;
5955 yylval.ival = OP_TRANS;
5960 S_scan_heredoc(pTHX_ register char *s)
5964 I32 op_type = OP_SCALAR;
5971 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
5975 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
5978 for (peek = s; *peek == ' ' || *peek == '\t'; peek++) ;
5979 if (*peek && strchr("`'\"",*peek)) {
5982 s = delimcpy(d, e, s, PL_bufend, term, &len);
5992 if (!isALNUM_lazy(s))
5993 deprecate("bare << to mean <<\"\"");
5994 for (; isALNUM_lazy(s); s++) {
5999 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6000 Perl_croak(aTHX_ "Delimiter for here document is too long");
6003 len = d - PL_tokenbuf;
6004 #ifndef PERL_STRICT_CR
6005 d = strchr(s, '\r');
6009 while (s < PL_bufend) {
6015 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6024 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6029 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6030 herewas = newSVpvn(s,PL_bufend-s);
6032 s--, herewas = newSVpvn(s,d-s);
6033 s += SvCUR(herewas);
6035 tmpstr = NEWSV(87,79);
6036 sv_upgrade(tmpstr, SVt_PVIV);
6041 else if (term == '`') {
6042 op_type = OP_BACKTICK;
6043 SvIVX(tmpstr) = '\\';
6047 PL_multi_start = PL_curcop->cop_line;
6048 PL_multi_open = PL_multi_close = '<';
6049 term = *PL_tokenbuf;
6050 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6051 char *bufptr = PL_sublex_info.super_bufptr;
6052 char *bufend = PL_sublex_info.super_bufend;
6053 char *olds = s - SvCUR(herewas);
6054 s = strchr(bufptr, '\n');
6058 while (s < bufend &&
6059 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6061 PL_curcop->cop_line++;
6064 PL_curcop->cop_line = PL_multi_start;
6065 missingterm(PL_tokenbuf);
6067 sv_setpvn(herewas,bufptr,d-bufptr+1);
6068 sv_setpvn(tmpstr,d+1,s-d);
6070 sv_catpvn(herewas,s,bufend-s);
6071 (void)strcpy(bufptr,SvPVX(herewas));
6078 while (s < PL_bufend &&
6079 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6081 PL_curcop->cop_line++;
6083 if (s >= PL_bufend) {
6084 PL_curcop->cop_line = PL_multi_start;
6085 missingterm(PL_tokenbuf);
6087 sv_setpvn(tmpstr,d+1,s-d);
6089 PL_curcop->cop_line++; /* the preceding stmt passes a newline */
6091 sv_catpvn(herewas,s,PL_bufend-s);
6092 sv_setsv(PL_linestr,herewas);
6093 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6094 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6097 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6098 while (s >= PL_bufend) { /* multiple line string? */
6100 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6101 PL_curcop->cop_line = PL_multi_start;
6102 missingterm(PL_tokenbuf);
6104 PL_curcop->cop_line++;
6105 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6106 #ifndef PERL_STRICT_CR
6107 if (PL_bufend - PL_linestart >= 2) {
6108 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6109 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6111 PL_bufend[-2] = '\n';
6113 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6115 else if (PL_bufend[-1] == '\r')
6116 PL_bufend[-1] = '\n';
6118 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6119 PL_bufend[-1] = '\n';
6121 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6122 SV *sv = NEWSV(88,0);
6124 sv_upgrade(sv, SVt_PVMG);
6125 sv_setsv(sv,PL_linestr);
6126 av_store(GvAV(PL_curcop->cop_filegv),
6127 (I32)PL_curcop->cop_line,sv);
6129 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6132 sv_catsv(PL_linestr,herewas);
6133 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6137 sv_catsv(tmpstr,PL_linestr);
6142 PL_multi_end = PL_curcop->cop_line;
6143 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6144 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6145 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6147 SvREFCNT_dec(herewas);
6148 PL_lex_stuff = tmpstr;
6149 yylval.ival = op_type;
6154 takes: current position in input buffer
6155 returns: new position in input buffer
6156 side-effects: yylval and lex_op are set.
6161 <FH> read from filehandle
6162 <pkg::FH> read from package qualified filehandle
6163 <pkg'FH> read from package qualified filehandle
6164 <$fh> read from filehandle in $fh
6170 S_scan_inputsymbol(pTHX_ char *start)
6172 register char *s = start; /* current position in buffer */
6178 d = PL_tokenbuf; /* start of temp holding space */
6179 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6180 end = strchr(s, '\n');
6183 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6185 /* die if we didn't have space for the contents of the <>,
6186 or if it didn't end, or if we see a newline
6189 if (len >= sizeof PL_tokenbuf)
6190 Perl_croak(aTHX_ "Excessively long <> operator");
6192 Perl_croak(aTHX_ "Unterminated <> operator");
6197 Remember, only scalar variables are interpreted as filehandles by
6198 this code. Anything more complex (e.g., <$fh{$num}>) will be
6199 treated as a glob() call.
6200 This code makes use of the fact that except for the $ at the front,
6201 a scalar variable and a filehandle look the same.
6203 if (*d == '$' && d[1]) d++;
6205 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6206 while (*d && (isALNUM_lazy(d) || *d == '\'' || *d == ':'))
6209 /* If we've tried to read what we allow filehandles to look like, and
6210 there's still text left, then it must be a glob() and not a getline.
6211 Use scan_str to pull out the stuff between the <> and treat it
6212 as nothing more than a string.
6215 if (d - PL_tokenbuf != len) {
6216 yylval.ival = OP_GLOB;
6218 s = scan_str(start,FALSE,FALSE);
6220 Perl_croak(aTHX_ "Glob not terminated");
6224 /* we're in a filehandle read situation */
6227 /* turn <> into <ARGV> */
6229 (void)strcpy(d,"ARGV");
6231 /* if <$fh>, create the ops to turn the variable into a
6237 /* try to find it in the pad for this block, otherwise find
6238 add symbol table ops
6240 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6241 OP *o = newOP(OP_PADSV, 0);
6243 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o);
6246 GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV);
6247 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0,
6248 newUNOP(OP_RV2SV, 0,
6249 newGVOP(OP_GV, 0, gv)));
6251 PL_lex_op->op_flags |= OPf_SPECIAL;
6252 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6253 yylval.ival = OP_NULL;
6256 /* If it's none of the above, it must be a literal filehandle
6257 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6259 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6260 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6261 yylval.ival = OP_NULL;
6270 takes: start position in buffer
6271 keep_quoted preserve \ on the embedded delimiter(s)
6272 keep_delims preserve the delimiters around the string
6273 returns: position to continue reading from buffer
6274 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6275 updates the read buffer.
6277 This subroutine pulls a string out of the input. It is called for:
6278 q single quotes q(literal text)
6279 ' single quotes 'literal text'
6280 qq double quotes qq(interpolate $here please)
6281 " double quotes "interpolate $here please"
6282 qx backticks qx(/bin/ls -l)
6283 ` backticks `/bin/ls -l`
6284 qw quote words @EXPORT_OK = qw( func() $spam )
6285 m// regexp match m/this/
6286 s/// regexp substitute s/this/that/
6287 tr/// string transliterate tr/this/that/
6288 y/// string transliterate y/this/that/
6289 ($*@) sub prototypes sub foo ($)
6290 (stuff) sub attr parameters sub foo : attr(stuff)
6291 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6293 In most of these cases (all but <>, patterns and transliterate)
6294 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6295 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6296 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6299 It skips whitespace before the string starts, and treats the first
6300 character as the delimiter. If the delimiter is one of ([{< then
6301 the corresponding "close" character )]}> is used as the closing
6302 delimiter. It allows quoting of delimiters, and if the string has
6303 balanced delimiters ([{<>}]) it allows nesting.
6305 The lexer always reads these strings into lex_stuff, except in the
6306 case of the operators which take *two* arguments (s/// and tr///)
6307 when it checks to see if lex_stuff is full (presumably with the 1st
6308 arg to s or tr) and if so puts the string into lex_repl.
6313 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6316 SV *sv; /* scalar value: string */
6317 char *tmps; /* temp string, used for delimiter matching */
6318 register char *s = start; /* current position in the buffer */
6319 register char term; /* terminating character */
6320 register char *to; /* current position in the sv's data */
6321 I32 brackets = 1; /* bracket nesting level */
6323 /* skip space before the delimiter */
6327 /* mark where we are, in case we need to report errors */
6330 /* after skipping whitespace, the next character is the terminator */
6332 /* mark where we are */
6333 PL_multi_start = PL_curcop->cop_line;
6334 PL_multi_open = term;
6336 /* find corresponding closing delimiter */
6337 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6339 PL_multi_close = term;
6341 /* create a new SV to hold the contents. 87 is leak category, I'm
6342 assuming. 79 is the SV's initial length. What a random number. */
6344 sv_upgrade(sv, SVt_PVIV);
6346 (void)SvPOK_only(sv); /* validate pointer */
6348 /* move past delimiter and try to read a complete string */
6350 sv_catpvn(sv, s, 1);
6353 /* extend sv if need be */
6354 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6355 /* set 'to' to the next character in the sv's string */
6356 to = SvPVX(sv)+SvCUR(sv);
6358 /* if open delimiter is the close delimiter read unbridle */
6359 if (PL_multi_open == PL_multi_close) {
6360 for (; s < PL_bufend; s++,to++) {
6361 /* embedded newlines increment the current line number */
6362 if (*s == '\n' && !PL_rsfp)
6363 PL_curcop->cop_line++;
6364 /* handle quoted delimiters */
6365 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6366 if (!keep_quoted && s[1] == term)
6368 /* any other quotes are simply copied straight through */
6372 /* terminate when run out of buffer (the for() condition), or
6373 have found the terminator */
6374 else if (*s == term)
6380 /* if the terminator isn't the same as the start character (e.g.,
6381 matched brackets), we have to allow more in the quoting, and
6382 be prepared for nested brackets.
6385 /* read until we run out of string, or we find the terminator */
6386 for (; s < PL_bufend; s++,to++) {
6387 /* embedded newlines increment the line count */
6388 if (*s == '\n' && !PL_rsfp)
6389 PL_curcop->cop_line++;
6390 /* backslashes can escape the open or closing characters */
6391 if (*s == '\\' && s+1 < PL_bufend) {
6393 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6398 /* allow nested opens and closes */
6399 else if (*s == PL_multi_close && --brackets <= 0)
6401 else if (*s == PL_multi_open)
6406 /* terminate the copied string and update the sv's end-of-string */
6408 SvCUR_set(sv, to - SvPVX(sv));
6411 * this next chunk reads more into the buffer if we're not done yet
6414 if (s < PL_bufend) break; /* handle case where we are done yet :-) */
6416 #ifndef PERL_STRICT_CR
6417 if (to - SvPVX(sv) >= 2) {
6418 if ((to[-2] == '\r' && to[-1] == '\n') ||
6419 (to[-2] == '\n' && to[-1] == '\r'))
6423 SvCUR_set(sv, to - SvPVX(sv));
6425 else if (to[-1] == '\r')
6428 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6432 /* if we're out of file, or a read fails, bail and reset the current
6433 line marker so we can report where the unterminated string began
6436 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6438 PL_curcop->cop_line = PL_multi_start;
6441 /* we read a line, so increment our line counter */
6442 PL_curcop->cop_line++;
6444 /* update debugger info */
6445 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6446 SV *sv = NEWSV(88,0);
6448 sv_upgrade(sv, SVt_PVMG);
6449 sv_setsv(sv,PL_linestr);
6450 av_store(GvAV(PL_curcop->cop_filegv),
6451 (I32)PL_curcop->cop_line, sv);
6454 /* having changed the buffer, we must update PL_bufend */
6455 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6458 /* at this point, we have successfully read the delimited string */
6461 sv_catpvn(sv, s, 1);
6462 PL_multi_end = PL_curcop->cop_line;
6465 /* if we allocated too much space, give some back */
6466 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6467 SvLEN_set(sv, SvCUR(sv) + 1);
6468 Renew(SvPVX(sv), SvLEN(sv), char);
6471 /* decide whether this is the first or second quoted string we've read
6484 takes: pointer to position in buffer
6485 returns: pointer to new position in buffer
6486 side-effects: builds ops for the constant in yylval.op
6488 Read a number in any of the formats that Perl accepts:
6490 0(x[0-7A-F]+)|([0-7]+)|(b[01])
6491 [\d_]+(\.[\d_]*)?[Ee](\d+)
6493 Underbars (_) are allowed in decimal numbers. If -w is on,
6494 underbars before a decimal point must be at three digit intervals.
6496 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6499 If it reads a number without a decimal point or an exponent, it will
6500 try converting the number to an integer and see if it can do so
6501 without loss of precision.
6505 Perl_scan_num(pTHX_ char *start)
6507 register char *s = start; /* current position in buffer */
6508 register char *d; /* destination in temp buffer */
6509 register char *e; /* end of temp buffer */
6510 IV tryiv; /* used to see if it can be an IV */
6511 NV value; /* number read, as a double */
6512 SV *sv; /* place to put the converted number */
6513 bool floatit; /* boolean: int or float? */
6514 char *lastub = 0; /* position of last underbar */
6515 static char number_too_long[] = "Number too long";
6517 /* We use the first character to decide what type of number this is */
6521 Perl_croak(aTHX_ "panic: scan_num");
6523 /* if it starts with a 0, it could be an octal number, a decimal in
6524 0.13 disguise, or a hexadecimal number, or a binary number.
6529 u holds the "number so far"
6530 shift the power of 2 of the base
6531 (hex == 4, octal == 3, binary == 1)
6532 overflowed was the number more than we can hold?
6534 Shift is used when we add a digit. It also serves as an "are
6535 we in octal/hex/binary?" indicator to disallow hex characters
6542 bool overflowed = FALSE;
6543 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6544 static char* bases[5] = { "", "binary", "", "octal",
6546 static char* Bases[5] = { "", "Binary", "", "Octal",
6548 static char *maxima[5] = { "",
6549 "0b11111111111111111111111111111111",
6553 char *base, *Base, *max;
6559 } else if (s[1] == 'b') {
6563 /* check for a decimal in disguise */
6564 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
6566 /* so it must be octal */
6570 base = bases[shift];
6571 Base = Bases[shift];
6572 max = maxima[shift];
6574 /* read the rest of the number */
6576 /* x is used in the overflow test,
6577 b is the digit we're adding on. */
6582 /* if we don't mention it, we're done */
6591 /* 8 and 9 are not octal */
6594 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
6598 case '2': case '3': case '4':
6599 case '5': case '6': case '7':
6601 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
6605 b = *s++ & 15; /* ASCII digit -> value of digit */
6609 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
6610 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
6611 /* make sure they said 0x */
6616 /* Prepare to put the digit we have onto the end
6617 of the number so far. We check for overflows.
6622 x = u << shift; /* make room for the digit */
6624 if ((x >> shift) != u
6625 && !(PL_hints & HINT_NEW_BINARY)) {
6629 if (ckWARN_d(WARN_OVERFLOW))
6630 Perl_warner(aTHX_ WARN_OVERFLOW,
6631 "Integer overflow in %s number",
6634 u = x | b; /* add the digit to the end */
6637 n *= nvshift[shift];
6638 /* If an NV has not enough bits in its
6639 * mantissa to represent an UV this summing of
6640 * small low-order numbers is a waste of time
6641 * (because the NV cannot preserve the
6642 * low-order bits anyway): we could just
6643 * remember when did we overflow and in the
6644 * end just multiply n by the right
6652 /* if we get here, we had success: make a scalar value from
6659 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
6660 Perl_warner(aTHX_ WARN_PORTABLE,
6661 "%s number > %s non-portable",
6668 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
6669 Perl_warner(aTHX_ WARN_PORTABLE,
6670 "%s number > %s non-portable",
6675 if (PL_hints & HINT_NEW_BINARY)
6676 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
6681 handle decimal numbers.
6682 we're also sent here when we read a 0 as the first digit
6684 case '1': case '2': case '3': case '4': case '5':
6685 case '6': case '7': case '8': case '9': case '.':
6688 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
6691 /* read next group of digits and _ and copy into d */
6692 while (isDIGIT(*s) || *s == '_') {
6693 /* skip underscores, checking for misplaced ones
6697 dTHR; /* only for ckWARN */
6698 if (ckWARN(WARN_SYNTAX) && lastub && s - lastub != 3)
6699 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6703 /* check for end of fixed-length buffer */
6705 Perl_croak(aTHX_ number_too_long);
6706 /* if we're ok, copy the character */
6711 /* final misplaced underbar check */
6712 if (lastub && s - lastub != 3) {
6714 if (ckWARN(WARN_SYNTAX))
6715 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6718 /* read a decimal portion if there is one. avoid
6719 3..5 being interpreted as the number 3. followed
6722 if (*s == '.' && s[1] != '.') {
6726 /* copy, ignoring underbars, until we run out of
6727 digits. Note: no misplaced underbar checks!
6729 for (; isDIGIT(*s) || *s == '_'; s++) {
6730 /* fixed length buffer check */
6732 Perl_croak(aTHX_ number_too_long);
6738 /* read exponent part, if present */
6739 if (*s && strchr("eE",*s) && strchr("+-0123456789",s[1])) {
6743 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
6744 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
6746 /* allow positive or negative exponent */
6747 if (*s == '+' || *s == '-')
6750 /* read digits of exponent (no underbars :-) */
6751 while (isDIGIT(*s)) {
6753 Perl_croak(aTHX_ number_too_long);
6758 /* terminate the string */
6761 /* make an sv from the string */
6764 value = Atof(PL_tokenbuf);
6767 See if we can make do with an integer value without loss of
6768 precision. We use I_V to cast to an int, because some
6769 compilers have issues. Then we try casting it back and see
6770 if it was the same. We only do this if we know we
6771 specifically read an integer.
6773 Note: if floatit is true, then we don't need to do the
6777 if (!floatit && (NV)tryiv == value)
6778 sv_setiv(sv, tryiv);
6780 sv_setnv(sv, value);
6781 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
6782 (PL_hints & HINT_NEW_INTEGER) )
6783 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
6784 (floatit ? "float" : "integer"),
6789 /* make the op for the constant and return */
6791 yylval.opval = newSVOP(OP_CONST, 0, sv);
6797 S_scan_formline(pTHX_ register char *s)
6802 SV *stuff = newSVpvn("",0);
6803 bool needargs = FALSE;
6806 if (*s == '.' || *s == '}') {
6808 #ifdef PERL_STRICT_CR
6809 for (t = s+1;*t == ' ' || *t == '\t'; t++) ;
6811 for (t = s+1;*t == ' ' || *t == '\t' || *t == '\r'; t++) ;
6813 if (*t == '\n' || t == PL_bufend)
6816 if (PL_in_eval && !PL_rsfp) {
6817 eol = strchr(s,'\n');
6822 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6824 for (t = s; t < eol; t++) {
6825 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
6827 goto enough; /* ~~ must be first line in formline */
6829 if (*t == '@' || *t == '^')
6832 sv_catpvn(stuff, s, eol-s);
6836 s = filter_gets(PL_linestr, PL_rsfp, 0);
6837 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
6838 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
6841 yyerror("Format not terminated");
6851 PL_lex_state = LEX_NORMAL;
6852 PL_nextval[PL_nexttoke].ival = 0;
6856 PL_lex_state = LEX_FORMLINE;
6857 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
6859 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
6863 SvREFCNT_dec(stuff);
6864 PL_lex_formbrack = 0;
6875 PL_cshlen = strlen(PL_cshname);
6880 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
6883 I32 oldsavestack_ix = PL_savestack_ix;
6884 CV* outsidecv = PL_compcv;
6888 assert(SvTYPE(PL_compcv) == SVt_PVCV);
6890 save_I32(&PL_subline);
6891 save_item(PL_subname);
6893 SAVESPTR(PL_curpad);
6894 SAVESPTR(PL_comppad);
6895 SAVESPTR(PL_comppad_name);
6896 SAVESPTR(PL_compcv);
6897 SAVEI32(PL_comppad_name_fill);
6898 SAVEI32(PL_min_intro_pending);
6899 SAVEI32(PL_max_intro_pending);
6900 SAVEI32(PL_pad_reset_pending);
6902 PL_compcv = (CV*)NEWSV(1104,0);
6903 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
6904 CvFLAGS(PL_compcv) |= flags;
6906 PL_comppad = newAV();
6907 av_push(PL_comppad, Nullsv);
6908 PL_curpad = AvARRAY(PL_comppad);
6909 PL_comppad_name = newAV();
6910 PL_comppad_name_fill = 0;
6911 PL_min_intro_pending = 0;
6913 PL_subline = PL_curcop->cop_line;
6915 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
6916 PL_curpad[0] = (SV*)newAV();
6917 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
6918 #endif /* USE_THREADS */
6920 comppadlist = newAV();
6921 AvREAL_off(comppadlist);
6922 av_store(comppadlist, 0, (SV*)PL_comppad_name);
6923 av_store(comppadlist, 1, (SV*)PL_comppad);
6925 CvPADLIST(PL_compcv) = comppadlist;
6926 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
6928 CvOWNER(PL_compcv) = 0;
6929 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
6930 MUTEX_INIT(CvMUTEXP(PL_compcv));
6931 #endif /* USE_THREADS */
6933 return oldsavestack_ix;
6937 Perl_yywarn(pTHX_ char *s)
6940 PL_in_eval |= EVAL_WARNONLY;
6942 PL_in_eval &= ~EVAL_WARNONLY;
6947 Perl_yyerror(pTHX_ char *s)
6951 char *context = NULL;
6955 if (!yychar || (yychar == ';' && !PL_rsfp))
6957 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
6958 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
6959 while (isSPACE(*PL_oldoldbufptr))
6961 context = PL_oldoldbufptr;
6962 contlen = PL_bufptr - PL_oldoldbufptr;
6964 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
6965 PL_oldbufptr != PL_bufptr) {
6966 while (isSPACE(*PL_oldbufptr))
6968 context = PL_oldbufptr;
6969 contlen = PL_bufptr - PL_oldbufptr;
6971 else if (yychar > 255)
6972 where = "next token ???";
6973 else if ((yychar & 127) == 127) {
6974 if (PL_lex_state == LEX_NORMAL ||
6975 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
6976 where = "at end of line";
6977 else if (PL_lex_inpat)
6978 where = "within pattern";
6980 where = "within string";
6983 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
6985 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
6986 else if (isPRINT_LC(yychar))
6987 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
6989 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
6990 where = SvPVX(where_sv);
6992 msg = sv_2mortal(newSVpv(s, 0));
6993 Perl_sv_catpvf(aTHX_ msg, " at %_ line %"IVdf", ",
6994 GvSV(PL_curcop->cop_filegv), (IV)PL_curcop->cop_line);
6996 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
6998 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
6999 if (PL_multi_start < PL_multi_end && (U32)(PL_curcop->cop_line - PL_multi_end) <= 1) {
7000 Perl_sv_catpvf(aTHX_ msg,
7001 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7002 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7005 if (PL_in_eval & EVAL_WARNONLY)
7006 Perl_warn(aTHX_ "%_", msg);
7009 if (PL_error_count >= 10)
7010 Perl_croak(aTHX_ "%_ has too many errors.\n", GvSV(PL_curcop->cop_filegv));
7012 PL_in_my_stash = Nullhv;
7024 * Restore a source filter.
7028 restore_rsfp(pTHXo_ void *f)
7030 PerlIO *fp = (PerlIO*)f;
7032 if (PL_rsfp == PerlIO_stdin())
7033 PerlIO_clearerr(PL_rsfp);
7034 else if (PL_rsfp && (PL_rsfp != fp))
7035 PerlIO_close(PL_rsfp);
7041 * Restores the state of PL_expect when the lexing that begun with a
7042 * start_lex() call has ended.
7046 restore_expect(pTHXo_ void *e)
7048 /* a safe way to store a small integer in a pointer */
7049 PL_expect = (expectation)((char *)e - PL_tokenbuf);
7053 * restore_lex_expect
7054 * Restores the state of PL_lex_expect when the lexing that begun with a
7055 * start_lex() call has ended.
7059 restore_lex_expect(pTHXo_ void *e)
7061 /* a safe way to store a small integer in a pointer */
7062 PL_lex_expect = (expectation)((char *)e - PL_tokenbuf);