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 /* On MacOS, respect nonbreaking spaces */
53 #ifdef MACOS_TRADITIONAL
54 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
56 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
59 /* LEX_* are values for PL_lex_state, the state of the lexer.
60 * They are arranged oddly so that the guard on the switch statement
61 * can get by with a single comparison (if the compiler is smart enough).
64 /* #define LEX_NOTPARSING 11 is done in perl.h. */
67 #define LEX_INTERPNORMAL 9
68 #define LEX_INTERPCASEMOD 8
69 #define LEX_INTERPPUSH 7
70 #define LEX_INTERPSTART 6
71 #define LEX_INTERPEND 5
72 #define LEX_INTERPENDMAYBE 4
73 #define LEX_INTERPCONCAT 3
74 #define LEX_INTERPCONST 2
75 #define LEX_FORMLINE 1
76 #define LEX_KNOWNEXT 0
85 /* XXX If this causes problems, set i_unistd=undef in the hint file. */
87 # include <unistd.h> /* Needed for execv() */
96 YYSTYPE* yylval_pointer = NULL;
97 int* yychar_pointer = NULL;
100 # define yylval (*yylval_pointer)
101 # define yychar (*yychar_pointer)
102 # define PERL_YYLEX_PARAM yylval_pointer,yychar_pointer
104 # define yylex() Perl_yylex(aTHX_ yylval_pointer, yychar_pointer)
107 #include "keywords.h"
109 /* CLINE is a macro that ensures PL_copline has a sane value */
114 #define CLINE (PL_copline = (PL_curcop->cop_line < PL_copline ? PL_curcop->cop_line : PL_copline))
117 * Convenience functions to return different tokens and prime the
118 * lexer for the next token. They all take an argument.
120 * TOKEN : generic token (used for '(', DOLSHARP, etc)
121 * OPERATOR : generic operator
122 * AOPERATOR : assignment operator
123 * PREBLOCK : beginning the block after an if, while, foreach, ...
124 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
125 * PREREF : *EXPR where EXPR is not a simple identifier
126 * TERM : expression term
127 * LOOPX : loop exiting command (goto, last, dump, etc)
128 * FTST : file test operator
129 * FUN0 : zero-argument function
130 * FUN1 : not used, except for not, which isn't a UNIOP
131 * BOop : bitwise or or xor
133 * SHop : shift operator
134 * PWop : power operator
135 * PMop : pattern-matching operator
136 * Aop : addition-level operator
137 * Mop : multiplication-level operator
138 * Eop : equality-testing operator
139 * Rop : relational operator <= != gt
141 * Also see LOP and lop() below.
144 #define TOKEN(retval) return (PL_bufptr = s,(int)retval)
145 #define OPERATOR(retval) return (PL_expect = XTERM,PL_bufptr = s,(int)retval)
146 #define AOPERATOR(retval) return ao((PL_expect = XTERM,PL_bufptr = s,(int)retval))
147 #define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s,(int)retval)
148 #define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s,(int)retval)
149 #define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s,(int)retval)
150 #define TERM(retval) return (CLINE, PL_expect = XOPERATOR,PL_bufptr = s,(int)retval)
151 #define LOOPX(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LOOPEX)
152 #define FTST(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)UNIOP)
153 #define FUN0(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC0)
154 #define FUN1(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC1)
155 #define BOop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITOROP))
156 #define BAop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITANDOP))
157 #define SHop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)SHIFTOP))
158 #define PWop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)POWOP))
159 #define PMop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MATCHOP)
160 #define Aop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)ADDOP))
161 #define Mop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MULOP))
162 #define Eop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)EQOP)
163 #define Rop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)RELOP)
165 /* This bit of chicanery makes a unary function followed by
166 * a parenthesis into a function with one argument, highest precedence.
168 #define UNI(f) return(yylval.ival = f, \
171 PL_last_uni = PL_oldbufptr, \
172 PL_last_lop_op = f, \
173 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
175 #define UNIBRACK(f) return(yylval.ival = f, \
177 PL_last_uni = PL_oldbufptr, \
178 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
180 /* grandfather return to old style */
181 #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
186 * This subroutine detects &&= and ||= and turns an ANDAND or OROR
187 * into an OP_ANDASSIGN or OP_ORASSIGN
191 S_ao(pTHX_ int toketype)
193 if (*PL_bufptr == '=') {
195 if (toketype == ANDAND)
196 yylval.ival = OP_ANDASSIGN;
197 else if (toketype == OROR)
198 yylval.ival = OP_ORASSIGN;
206 * When Perl expects an operator and finds something else, no_op
207 * prints the warning. It always prints "<something> found where
208 * operator expected. It prints "Missing semicolon on previous line?"
209 * if the surprise occurs at the start of the line. "do you need to
210 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
211 * where the compiler doesn't know if foo is a method call or a function.
212 * It prints "Missing operator before end of line" if there's nothing
213 * after the missing operator, or "... before <...>" if there is something
214 * after the missing operator.
218 S_no_op(pTHX_ char *what, char *s)
220 char *oldbp = PL_bufptr;
221 bool is_first = (PL_oldbufptr == PL_linestart);
229 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
231 Perl_warn(aTHX_ "\t(Missing semicolon on previous line?)\n");
232 else if (PL_oldoldbufptr && isIDFIRST_lazy(PL_oldoldbufptr)) {
234 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy(t) || *t == ':'); t++) ;
235 if (t < PL_bufptr && isSPACE(*t))
236 Perl_warn(aTHX_ "\t(Do you need to predeclare %.*s?)\n",
237 t - PL_oldoldbufptr, PL_oldoldbufptr);
240 Perl_warn(aTHX_ "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp);
246 * Complain about missing quote/regexp/heredoc terminator.
247 * If it's called with (char *)NULL then it cauterizes the line buffer.
248 * If we're in a delimited string and the delimiter is a control
249 * character, it's reformatted into a two-char sequence like ^C.
254 S_missingterm(pTHX_ char *s)
259 char *nl = strrchr(s,'\n');
265 iscntrl(PL_multi_close)
267 PL_multi_close < 32 || PL_multi_close == 127
271 tmpbuf[1] = toCTRL(PL_multi_close);
277 *tmpbuf = PL_multi_close;
281 q = strchr(s,'"') ? '\'' : '"';
282 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
290 Perl_deprecate(pTHX_ char *s)
293 if (ckWARN(WARN_DEPRECATED))
294 Perl_warner(aTHX_ WARN_DEPRECATED, "Use of %s is deprecated", s);
299 * Deprecate a comma-less variable list.
305 deprecate("comma-less variable list");
309 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
310 * utf16-to-utf8-reversed.
316 S_win32_textfilter(pTHX_ int idx, SV *sv, int maxlen)
318 I32 count = FILTER_READ(idx+1, sv, maxlen);
319 if (count > 0 && !maxlen)
320 win32_strip_return(sv);
326 S_utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
328 I32 count = FILTER_READ(idx+1, sv, maxlen);
332 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
333 tend = utf16_to_utf8((U16*)SvPVX(sv), tmps, SvCUR(sv));
334 sv_usepvn(sv, (char*)tmps, tend - tmps);
341 S_utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
343 I32 count = FILTER_READ(idx+1, sv, maxlen);
347 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
348 tend = utf16_to_utf8_reversed((U16*)SvPVX(sv), tmps, SvCUR(sv));
349 sv_usepvn(sv, (char*)tmps, tend - tmps);
357 * Initialize variables. Uses the Perl save_stack to save its state (for
358 * recursive calls to the parser).
362 Perl_lex_start(pTHX_ SV *line)
368 SAVEI32(PL_lex_dojoin);
369 SAVEI32(PL_lex_brackets);
370 SAVEI32(PL_lex_fakebrack);
371 SAVEI32(PL_lex_casemods);
372 SAVEI32(PL_lex_starts);
373 SAVEI32(PL_lex_state);
374 SAVESPTR(PL_lex_inpat);
375 SAVEI32(PL_lex_inwhat);
376 SAVEI16(PL_curcop->cop_line);
379 SAVEPPTR(PL_oldbufptr);
380 SAVEPPTR(PL_oldoldbufptr);
381 SAVEPPTR(PL_linestart);
382 SAVESPTR(PL_linestr);
383 SAVEPPTR(PL_lex_brackstack);
384 SAVEPPTR(PL_lex_casestack);
385 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
386 SAVESPTR(PL_lex_stuff);
387 SAVEI32(PL_lex_defer);
388 SAVEI32(PL_sublex_info.sub_inwhat);
389 SAVESPTR(PL_lex_repl);
390 SAVEDESTRUCTOR_X(restore_expect, PL_tokenbuf + PL_expect); /* encode as pointer */
391 SAVEDESTRUCTOR_X(restore_lex_expect, PL_tokenbuf + PL_expect);
393 PL_lex_state = LEX_NORMAL;
397 PL_lex_fakebrack = 0;
398 New(899, PL_lex_brackstack, 120, char);
399 New(899, PL_lex_casestack, 12, char);
400 SAVEFREEPV(PL_lex_brackstack);
401 SAVEFREEPV(PL_lex_casestack);
403 *PL_lex_casestack = '\0';
406 PL_lex_stuff = Nullsv;
407 PL_lex_repl = Nullsv;
410 PL_sublex_info.sub_inwhat = 0;
412 if (SvREADONLY(PL_linestr))
413 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
414 s = SvPV(PL_linestr, len);
415 if (len && s[len-1] != ';') {
416 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
417 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
418 sv_catpvn(PL_linestr, "\n;", 2);
420 SvTEMP_off(PL_linestr);
421 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
422 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
424 PL_rs = newSVpvn("\n", 1);
430 * Finalizer for lexing operations. Must be called when the parser is
431 * done with the lexer.
437 PL_doextract = FALSE;
442 * This subroutine has nothing to do with tilting, whether at windmills
443 * or pinball tables. Its name is short for "increment line". It
444 * increments the current line number in PL_curcop->cop_line and checks
445 * to see whether the line starts with a comment of the form
446 * # line 500 "foo.pm"
447 * If so, it sets the current line number and file to the values in the comment.
451 S_incline(pTHX_ char *s)
460 PL_curcop->cop_line++;
463 while (SPACE_OR_TAB(*s)) s++;
464 if (strnEQ(s, "line ", 5)) {
473 while (SPACE_OR_TAB(*s))
475 if (*s == '"' && (t = strchr(s+1, '"')))
479 return; /* false alarm */
480 for (t = s; !isSPACE(*t); t++) ;
485 CopFILEGV_set(PL_curcop, gv_fetchfile(s));
487 CopFILEGV_set(PL_curcop, gv_fetchfile(PL_origfilename));
489 PL_curcop->cop_line = atoi(n)-1;
494 * Called to gobble the appropriate amount and type of whitespace.
495 * Skips comments as well.
499 S_skipspace(pTHX_ register char *s)
502 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
503 while (s < PL_bufend && SPACE_OR_TAB(*s))
509 SSize_t oldprevlen, oldoldprevlen;
510 SSize_t oldloplen, oldunilen;
511 while (s < PL_bufend && isSPACE(*s)) {
512 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
517 if (s < PL_bufend && *s == '#') {
518 while (s < PL_bufend && *s != '\n')
522 if (PL_in_eval && !PL_rsfp) {
529 /* only continue to recharge the buffer if we're at the end
530 * of the buffer, we're not reading from a source filter, and
531 * we're in normal lexing mode
533 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
534 PL_lex_state == LEX_FORMLINE)
537 /* try to recharge the buffer */
538 if ((s = filter_gets(PL_linestr, PL_rsfp,
539 (prevlen = SvCUR(PL_linestr)))) == Nullch)
541 /* end of file. Add on the -p or -n magic */
542 if (PL_minus_n || PL_minus_p) {
543 sv_setpv(PL_linestr,PL_minus_p ?
544 ";}continue{print or die qq(-p destination: $!\\n)" :
546 sv_catpv(PL_linestr,";}");
547 PL_minus_n = PL_minus_p = 0;
550 sv_setpv(PL_linestr,";");
552 /* reset variables for next time we lex */
553 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
555 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
557 /* Close the filehandle. Could be from -P preprocessor,
558 * STDIN, or a regular file. If we were reading code from
559 * STDIN (because the commandline held no -e or filename)
560 * then we don't close it, we reset it so the code can
561 * read from STDIN too.
564 if (PL_preprocess && !PL_in_eval)
565 (void)PerlProc_pclose(PL_rsfp);
566 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
567 PerlIO_clearerr(PL_rsfp);
569 (void)PerlIO_close(PL_rsfp);
574 /* not at end of file, so we only read another line */
575 /* make corresponding updates to old pointers, for yyerror() */
576 oldprevlen = PL_oldbufptr - PL_bufend;
577 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
579 oldunilen = PL_last_uni - PL_bufend;
581 oldloplen = PL_last_lop - PL_bufend;
582 PL_linestart = PL_bufptr = s + prevlen;
583 PL_bufend = s + SvCUR(PL_linestr);
585 PL_oldbufptr = s + oldprevlen;
586 PL_oldoldbufptr = s + oldoldprevlen;
588 PL_last_uni = s + oldunilen;
590 PL_last_lop = s + oldloplen;
593 /* debugger active and we're not compiling the debugger code,
594 * so store the line into the debugger's array of lines
596 if (PERLDB_LINE && PL_curstash != PL_debstash) {
597 SV *sv = NEWSV(85,0);
599 sv_upgrade(sv, SVt_PVMG);
600 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
601 av_store(CopFILEAV(PL_curcop),(I32)PL_curcop->cop_line,sv);
608 * Check the unary operators to ensure there's no ambiguity in how they're
609 * used. An ambiguous piece of code would be:
611 * This doesn't mean rand() + 5. Because rand() is a unary operator,
612 * the +5 is its argument.
622 if (PL_oldoldbufptr != PL_last_uni)
624 while (isSPACE(*PL_last_uni))
626 for (s = PL_last_uni; isALNUM_lazy(s) || *s == '-'; s++) ;
627 if ((t = strchr(s, '(')) && t < PL_bufptr)
629 if (ckWARN_d(WARN_AMBIGUOUS)){
632 Perl_warner(aTHX_ WARN_AMBIGUOUS,
633 "Warning: Use of \"%s\" without parens is ambiguous",
639 /* workaround to replace the UNI() macro with a function. Only the
640 * hints/uts.sh file mentions this. Other comments elsewhere in the
641 * source indicate Microport Unix might need it too.
647 #define UNI(f) return uni(f,s)
650 S_uni(pTHX_ I32 f, char *s)
655 PL_last_uni = PL_oldbufptr;
666 #endif /* CRIPPLED_CC */
669 * LOP : macro to build a list operator. Its behaviour has been replaced
670 * with a subroutine, S_lop() for which LOP is just another name.
673 #define LOP(f,x) return lop(f,x,s)
677 * Build a list operator (or something that might be one). The rules:
678 * - if we have a next token, then it's a list operator [why?]
679 * - if the next thing is an opening paren, then it's a function
680 * - else it's a list operator
684 S_lop(pTHX_ I32 f, expectation x, char *s)
691 PL_last_lop = PL_oldbufptr;
706 * When the lexer realizes it knows the next token (for instance,
707 * it is reordering tokens for the parser) then it can call S_force_next
708 * to know what token to return the next time the lexer is called. Caller
709 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
710 * handles the token correctly.
714 S_force_next(pTHX_ I32 type)
716 PL_nexttype[PL_nexttoke] = type;
718 if (PL_lex_state != LEX_KNOWNEXT) {
719 PL_lex_defer = PL_lex_state;
720 PL_lex_expect = PL_expect;
721 PL_lex_state = LEX_KNOWNEXT;
727 * When the lexer knows the next thing is a word (for instance, it has
728 * just seen -> and it knows that the next char is a word char, then
729 * it calls S_force_word to stick the next word into the PL_next lookahead.
732 * char *start : buffer position (must be within PL_linestr)
733 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
734 * int check_keyword : if true, Perl checks to make sure the word isn't
735 * a keyword (do this if the word is a label, e.g. goto FOO)
736 * int allow_pack : if true, : characters will also be allowed (require,
738 * int allow_initial_tick : used by the "sub" lexer only.
742 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
747 start = skipspace(start);
749 if (isIDFIRST_lazy(s) ||
750 (allow_pack && *s == ':') ||
751 (allow_initial_tick && *s == '\'') )
753 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
754 if (check_keyword && keyword(PL_tokenbuf, len))
756 if (token == METHOD) {
761 PL_expect = XOPERATOR;
764 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0));
765 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
773 * Called when the lexer wants $foo *foo &foo etc, but the program
774 * text only contains the "foo" portion. The first argument is a pointer
775 * to the "foo", and the second argument is the type symbol to prefix.
776 * Forces the next token to be a "WORD".
777 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
781 S_force_ident(pTHX_ register char *s, int kind)
784 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
785 PL_nextval[PL_nexttoke].opval = o;
788 dTHR; /* just for in_eval */
789 o->op_private = OPpCONST_ENTERED;
790 /* XXX see note in pp_entereval() for why we forgo typo
791 warnings if the symbol must be introduced in an eval.
793 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
794 kind == '$' ? SVt_PV :
795 kind == '@' ? SVt_PVAV :
796 kind == '%' ? SVt_PVHV :
805 * Forces the next token to be a version number.
809 S_force_version(pTHX_ char *s)
811 OP *version = Nullop;
815 /* default VERSION number -- GBARR */
820 for( d=s, c = 1; isDIGIT(*d) || *d == '_' || (*d == '.' && c--); d++);
821 if((*d == ';' || isSPACE(*d)) && *(skipspace(d)) != ',') {
823 /* real VERSION number -- GBARR */
824 version = yylval.opval;
828 /* NOTE: The parser sees the package name and the VERSION swapped */
829 PL_nextval[PL_nexttoke].opval = version;
837 * Tokenize a quoted string passed in as an SV. It finds the next
838 * chunk, up to end of string or a backslash. It may make a new
839 * SV containing that chunk (if HINT_NEW_STRING is on). It also
844 S_tokeq(pTHX_ SV *sv)
855 s = SvPV_force(sv, len);
859 while (s < send && *s != '\\')
864 if ( PL_hints & HINT_NEW_STRING )
865 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
868 if (s + 1 < send && (s[1] == '\\'))
869 s++; /* all that, just for this */
874 SvCUR_set(sv, d - SvPVX(sv));
876 if ( PL_hints & HINT_NEW_STRING )
877 return new_constant(NULL, 0, "q", sv, pv, "q");
882 * Now come three functions related to double-quote context,
883 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
884 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
885 * interact with PL_lex_state, and create fake ( ... ) argument lists
886 * to handle functions and concatenation.
887 * They assume that whoever calls them will be setting up a fake
888 * join call, because each subthing puts a ',' after it. This lets
891 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
893 * (I'm not sure whether the spurious commas at the end of lcfirst's
894 * arguments and join's arguments are created or not).
899 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
901 * Pattern matching will set PL_lex_op to the pattern-matching op to
902 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
904 * OP_CONST and OP_READLINE are easy--just make the new op and return.
906 * Everything else becomes a FUNC.
908 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
909 * had an OP_CONST or OP_READLINE). This just sets us up for a
910 * call to S_sublex_push().
916 register I32 op_type = yylval.ival;
918 if (op_type == OP_NULL) {
919 yylval.opval = PL_lex_op;
923 if (op_type == OP_CONST || op_type == OP_READLINE) {
924 SV *sv = tokeq(PL_lex_stuff);
926 if (SvTYPE(sv) == SVt_PVIV) {
927 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
933 nsv = newSVpvn(p, len);
937 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
938 PL_lex_stuff = Nullsv;
942 PL_sublex_info.super_state = PL_lex_state;
943 PL_sublex_info.sub_inwhat = op_type;
944 PL_sublex_info.sub_op = PL_lex_op;
945 PL_lex_state = LEX_INTERPPUSH;
949 yylval.opval = PL_lex_op;
959 * Create a new scope to save the lexing state. The scope will be
960 * ended in S_sublex_done. Returns a '(', starting the function arguments
961 * to the uc, lc, etc. found before.
962 * Sets PL_lex_state to LEX_INTERPCONCAT.
971 PL_lex_state = PL_sublex_info.super_state;
972 SAVEI32(PL_lex_dojoin);
973 SAVEI32(PL_lex_brackets);
974 SAVEI32(PL_lex_fakebrack);
975 SAVEI32(PL_lex_casemods);
976 SAVEI32(PL_lex_starts);
977 SAVEI32(PL_lex_state);
978 SAVESPTR(PL_lex_inpat);
979 SAVEI32(PL_lex_inwhat);
980 SAVEI16(PL_curcop->cop_line);
982 SAVEPPTR(PL_oldbufptr);
983 SAVEPPTR(PL_oldoldbufptr);
984 SAVEPPTR(PL_linestart);
985 SAVESPTR(PL_linestr);
986 SAVEPPTR(PL_lex_brackstack);
987 SAVEPPTR(PL_lex_casestack);
989 PL_linestr = PL_lex_stuff;
990 PL_lex_stuff = Nullsv;
992 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
994 PL_bufend += SvCUR(PL_linestr);
995 SAVEFREESV(PL_linestr);
997 PL_lex_dojoin = FALSE;
999 PL_lex_fakebrack = 0;
1000 New(899, PL_lex_brackstack, 120, char);
1001 New(899, PL_lex_casestack, 12, char);
1002 SAVEFREEPV(PL_lex_brackstack);
1003 SAVEFREEPV(PL_lex_casestack);
1004 PL_lex_casemods = 0;
1005 *PL_lex_casestack = '\0';
1007 PL_lex_state = LEX_INTERPCONCAT;
1008 PL_curcop->cop_line = PL_multi_start;
1010 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1011 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1012 PL_lex_inpat = PL_sublex_info.sub_op;
1014 PL_lex_inpat = Nullop;
1021 * Restores lexer state after a S_sublex_push.
1027 if (!PL_lex_starts++) {
1028 PL_expect = XOPERATOR;
1029 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpvn("",0));
1033 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1034 PL_lex_state = LEX_INTERPCASEMOD;
1038 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1039 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1040 PL_linestr = PL_lex_repl;
1042 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1043 PL_bufend += SvCUR(PL_linestr);
1044 SAVEFREESV(PL_linestr);
1045 PL_lex_dojoin = FALSE;
1046 PL_lex_brackets = 0;
1047 PL_lex_fakebrack = 0;
1048 PL_lex_casemods = 0;
1049 *PL_lex_casestack = '\0';
1051 if (SvEVALED(PL_lex_repl)) {
1052 PL_lex_state = LEX_INTERPNORMAL;
1054 /* we don't clear PL_lex_repl here, so that we can check later
1055 whether this is an evalled subst; that means we rely on the
1056 logic to ensure sublex_done() is called again only via the
1057 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1060 PL_lex_state = LEX_INTERPCONCAT;
1061 PL_lex_repl = Nullsv;
1067 PL_bufend = SvPVX(PL_linestr);
1068 PL_bufend += SvCUR(PL_linestr);
1069 PL_expect = XOPERATOR;
1070 PL_sublex_info.sub_inwhat = 0;
1078 Extracts a pattern, double-quoted string, or transliteration. This
1081 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1082 processing a pattern (PL_lex_inpat is true), a transliteration
1083 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1085 Returns a pointer to the character scanned up to. Iff this is
1086 advanced from the start pointer supplied (ie if anything was
1087 successfully parsed), will leave an OP for the substring scanned
1088 in yylval. Caller must intuit reason for not parsing further
1089 by looking at the next characters herself.
1093 double-quoted style: \r and \n
1094 regexp special ones: \D \s
1096 backrefs: \1 (deprecated in substitution replacements)
1097 case and quoting: \U \Q \E
1098 stops on @ and $, but not for $ as tail anchor
1100 In transliterations:
1101 characters are VERY literal, except for - not at the start or end
1102 of the string, which indicates a range. scan_const expands the
1103 range to the full set of intermediate characters.
1105 In double-quoted strings:
1107 double-quoted style: \r and \n
1109 backrefs: \1 (deprecated)
1110 case and quoting: \U \Q \E
1113 scan_const does *not* construct ops to handle interpolated strings.
1114 It stops processing as soon as it finds an embedded $ or @ variable
1115 and leaves it to the caller to work out what's going on.
1117 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @:foo.
1119 $ in pattern could be $foo or could be tail anchor. Assumption:
1120 it's a tail anchor if $ is the last thing in the string, or if it's
1121 followed by one of ")| \n\t"
1123 \1 (backreferences) are turned into $1
1125 The structure of the code is
1126 while (there's a character to process) {
1127 handle transliteration ranges
1128 skip regexp comments
1129 skip # initiated comments in //x patterns
1130 check for embedded @foo
1131 check for embedded scalars
1133 leave intact backslashes from leave (below)
1134 deprecate \1 in strings and sub replacements
1135 handle string-changing backslashes \l \U \Q \E, etc.
1136 switch (what was escaped) {
1137 handle - in a transliteration (becomes a literal -)
1138 handle \132 octal characters
1139 handle 0x15 hex characters
1140 handle \cV (control V)
1141 handle printf backslashes (\f, \r, \n, etc)
1143 } (end if backslash)
1144 } (end while character to read)
1149 S_scan_const(pTHX_ char *start)
1151 register char *send = PL_bufend; /* end of the constant */
1152 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1153 register char *s = start; /* start of the constant */
1154 register char *d = SvPVX(sv); /* destination for copies */
1155 bool dorange = FALSE; /* are we in a translit range? */
1157 I32 utf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1158 ? (PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
1160 I32 thisutf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1161 ? (PL_sublex_info.sub_op->op_private & (PL_lex_repl ?
1162 OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF))
1164 const char *leaveit = /* set of acceptably-backslashed characters */
1166 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
1169 while (s < send || dorange) {
1170 /* get transliterations out of the way (they're most literal) */
1171 if (PL_lex_inwhat == OP_TRANS) {
1172 /* expand a range A-Z to the full set of characters. AIE! */
1174 I32 i; /* current expanded character */
1175 I32 min; /* first character in range */
1176 I32 max; /* last character in range */
1178 i = d - SvPVX(sv); /* remember current offset */
1179 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1180 d = SvPVX(sv) + i; /* refresh d after realloc */
1181 d -= 2; /* eat the first char and the - */
1183 min = (U8)*d; /* first char in range */
1184 max = (U8)d[1]; /* last char in range */
1187 if ((isLOWER(min) && isLOWER(max)) ||
1188 (isUPPER(min) && isUPPER(max))) {
1190 for (i = min; i <= max; i++)
1194 for (i = min; i <= max; i++)
1201 for (i = min; i <= max; i++)
1204 /* mark the range as done, and continue */
1209 /* range begins (ignore - as first or last char) */
1210 else if (*s == '-' && s+1 < send && s != start) {
1212 *d++ = (char)0xff; /* use illegal utf8 byte--see pmtrans */
1221 /* if we get here, we're not doing a transliteration */
1223 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1224 except for the last char, which will be done separately. */
1225 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1227 while (s < send && *s != ')')
1229 } else if (s[2] == '{'
1230 || s[2] == 'p' && s[3] == '{') { /* This should march regcomp.c */
1232 char *regparse = s + (s[2] == '{' ? 3 : 4);
1235 while (count && (c = *regparse)) {
1236 if (c == '\\' && regparse[1])
1244 if (*regparse != ')') {
1245 regparse--; /* Leave one char for continuation. */
1246 yyerror("Sequence (?{...}) not terminated or not {}-balanced");
1248 while (s < regparse)
1253 /* likewise skip #-initiated comments in //x patterns */
1254 else if (*s == '#' && PL_lex_inpat &&
1255 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1256 while (s+1 < send && *s != '\n')
1260 /* check for embedded arrays (@foo, @:foo, @'foo, @{foo}, @$foo) */
1261 else if (*s == '@' && s[1] && (isALNUM_lazy(s+1) || strchr(":'{$", s[1])))
1264 /* check for embedded scalars. only stop if we're sure it's a
1267 else if (*s == '$') {
1268 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1270 if (s + 1 < send && !strchr("()| \n\t", s[1]))
1271 break; /* in regexp, $ might be tail anchor */
1274 /* (now in tr/// code again) */
1276 if (*s & 0x80 && thisutf) {
1277 dTHR; /* only for ckWARN */
1278 if (ckWARN(WARN_UTF8)) {
1279 (void)utf8_to_uv((U8*)s, &len); /* could cvt latin-1 to utf8 here... */
1289 if (*s == '\\' && s+1 < send) {
1292 /* some backslashes we leave behind */
1293 if (*leaveit && *s && strchr(leaveit, *s)) {
1299 /* deprecate \1 in strings and substitution replacements */
1300 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1301 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1303 dTHR; /* only for ckWARN */
1304 if (ckWARN(WARN_SYNTAX))
1305 Perl_warner(aTHX_ WARN_SYNTAX, "\\%c better written as $%c", *s, *s);
1310 /* string-change backslash escapes */
1311 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1316 /* if we get here, it's either a quoted -, or a digit */
1319 /* quoted - in transliterations */
1321 if (PL_lex_inwhat == OP_TRANS) {
1329 if (ckWARN(WARN_UNSAFE) && isALPHA(*s))
1330 Perl_warner(aTHX_ WARN_UNSAFE,
1331 "Unrecognized escape \\%c passed through",
1333 /* default action is to copy the quoted character */
1338 /* \132 indicates an octal constant */
1339 case '0': case '1': case '2': case '3':
1340 case '4': case '5': case '6': case '7':
1341 *d++ = (char)scan_oct(s, 3, &len);
1345 /* \x24 indicates a hex constant */
1349 char* e = strchr(s, '}');
1352 yyerror("Missing right brace on \\x{}");
1357 if (ckWARN(WARN_UTF8))
1358 Perl_warner(aTHX_ WARN_UTF8,
1359 "Use of \\x{} without utf8 declaration");
1361 /* note: utf always shorter than hex */
1362 d = (char*)uv_to_utf8((U8*)d,
1363 (UV)scan_hex(s + 1, e - s - 1, &len));
1367 UV uv = (UV)scan_hex(s, 2, &len);
1368 if (utf && PL_lex_inwhat == OP_TRANS &&
1369 utf != (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
1371 d = (char*)uv_to_utf8((U8*)d, uv); /* doing a CU or UC */
1374 if (uv >= 127 && UTF) {
1376 if (ckWARN(WARN_UTF8))
1377 Perl_warner(aTHX_ WARN_UTF8,
1378 "\\x%.*s will produce malformed UTF-8 character; use \\x{%.*s} for that",
1387 /* \N{latin small letter a} is a named character */
1391 char* e = strchr(s, '}');
1400 yyerror("Missing right brace on \\N{}");
1404 res = newSVpvn(s + 1, e - s - 1);
1405 res = new_constant( Nullch, 0, "charnames",
1406 res, Nullsv, "\\N{...}" );
1407 str = SvPV(res,len);
1408 if (len > e - s + 4) {
1409 char *odest = SvPVX(sv);
1411 SvGROW(sv, (SvCUR(sv) + len - (e - s + 4)));
1412 d = SvPVX(sv) + (d - odest);
1414 Copy(str, d, len, char);
1421 yyerror("Missing braces on \\N{}");
1424 /* \c is a control character */
1438 /* printf-style backslashes, formfeeds, newlines, etc */
1456 *d++ = '\047'; /* CP 1047 */
1459 *d++ = '\057'; /* CP 1047 */
1473 } /* end if (backslash) */
1476 } /* while loop to process each character */
1478 /* terminate the string and set up the sv */
1480 SvCUR_set(sv, d - SvPVX(sv));
1483 /* shrink the sv if we allocated more than we used */
1484 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1485 SvLEN_set(sv, SvCUR(sv) + 1);
1486 Renew(SvPVX(sv), SvLEN(sv), char);
1489 /* return the substring (via yylval) only if we parsed anything */
1490 if (s > PL_bufptr) {
1491 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1492 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1494 ( PL_lex_inwhat == OP_TRANS
1496 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1499 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1506 * Returns TRUE if there's more to the expression (e.g., a subscript),
1509 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1511 * ->[ and ->{ return TRUE
1512 * { and [ outside a pattern are always subscripts, so return TRUE
1513 * if we're outside a pattern and it's not { or [, then return FALSE
1514 * if we're in a pattern and the first char is a {
1515 * {4,5} (any digits around the comma) returns FALSE
1516 * if we're in a pattern and the first char is a [
1518 * [SOMETHING] has a funky algorithm to decide whether it's a
1519 * character class or not. It has to deal with things like
1520 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1521 * anything else returns TRUE
1524 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1527 S_intuit_more(pTHX_ register char *s)
1529 if (PL_lex_brackets)
1531 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1533 if (*s != '{' && *s != '[')
1538 /* In a pattern, so maybe we have {n,m}. */
1555 /* On the other hand, maybe we have a character class */
1558 if (*s == ']' || *s == '^')
1561 /* this is terrifying, and it works */
1562 int weight = 2; /* let's weigh the evidence */
1564 unsigned char un_char = 255, last_un_char;
1565 char *send = strchr(s,']');
1566 char tmpbuf[sizeof PL_tokenbuf * 4];
1568 if (!send) /* has to be an expression */
1571 Zero(seen,256,char);
1574 else if (isDIGIT(*s)) {
1576 if (isDIGIT(s[1]) && s[2] == ']')
1582 for (; s < send; s++) {
1583 last_un_char = un_char;
1584 un_char = (unsigned char)*s;
1589 weight -= seen[un_char] * 10;
1590 if (isALNUM_lazy(s+1)) {
1591 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1592 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1597 else if (*s == '$' && s[1] &&
1598 strchr("[#!%*<>()-=",s[1])) {
1599 if (/*{*/ strchr("])} =",s[2]))
1608 if (strchr("wds]",s[1]))
1610 else if (seen['\''] || seen['"'])
1612 else if (strchr("rnftbxcav",s[1]))
1614 else if (isDIGIT(s[1])) {
1616 while (s[1] && isDIGIT(s[1]))
1626 if (strchr("aA01! ",last_un_char))
1628 if (strchr("zZ79~",s[1]))
1630 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1631 weight -= 5; /* cope with negative subscript */
1634 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1635 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1640 if (keyword(tmpbuf, d - tmpbuf))
1643 if (un_char == last_un_char + 1)
1645 weight -= seen[un_char];
1650 if (weight >= 0) /* probably a character class */
1660 * Does all the checking to disambiguate
1662 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1663 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1665 * First argument is the stuff after the first token, e.g. "bar".
1667 * Not a method if bar is a filehandle.
1668 * Not a method if foo is a subroutine prototyped to take a filehandle.
1669 * Not a method if it's really "Foo $bar"
1670 * Method if it's "foo $bar"
1671 * Not a method if it's really "print foo $bar"
1672 * Method if it's really "foo package::" (interpreted as package->foo)
1673 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1674 * Not a method if bar is a filehandle or package, but is quotd with
1679 S_intuit_method(pTHX_ char *start, GV *gv)
1681 char *s = start + (*start == '$');
1682 char tmpbuf[sizeof PL_tokenbuf];
1690 if ((cv = GvCVu(gv))) {
1691 char *proto = SvPVX(cv);
1701 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1702 /* start is the beginning of the possible filehandle/object,
1703 * and s is the end of it
1704 * tmpbuf is a copy of it
1707 if (*start == '$') {
1708 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1713 return *s == '(' ? FUNCMETH : METHOD;
1715 if (!keyword(tmpbuf, len)) {
1716 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1721 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1722 if (indirgv && GvCVu(indirgv))
1724 /* filehandle or package name makes it a method */
1725 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1727 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1728 return 0; /* no assumptions -- "=>" quotes bearword */
1730 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1731 newSVpvn(tmpbuf,len));
1732 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1736 return *s == '(' ? FUNCMETH : METHOD;
1744 * Return a string of Perl code to load the debugger. If PERL5DB
1745 * is set, it will return the contents of that, otherwise a
1746 * compile-time require of perl5db.pl.
1753 char *pdb = PerlEnv_getenv("PERL5DB");
1757 SETERRNO(0,SS$_NORMAL);
1758 return "BEGIN { require 'perl5db.pl' }";
1764 /* Encoded script support. filter_add() effectively inserts a
1765 * 'pre-processing' function into the current source input stream.
1766 * Note that the filter function only applies to the current source file
1767 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1769 * The datasv parameter (which may be NULL) can be used to pass
1770 * private data to this instance of the filter. The filter function
1771 * can recover the SV using the FILTER_DATA macro and use it to
1772 * store private buffers and state information.
1774 * The supplied datasv parameter is upgraded to a PVIO type
1775 * and the IoDIRP field is used to store the function pointer.
1776 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1777 * private use must be set using malloc'd pointers.
1781 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1786 if (!PL_rsfp_filters)
1787 PL_rsfp_filters = newAV();
1789 datasv = NEWSV(255,0);
1790 if (!SvUPGRADE(datasv, SVt_PVIO))
1791 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1792 IoDIRP(datasv) = (DIR*)funcp; /* stash funcp into spare field */
1793 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1794 funcp, SvPV_nolen(datasv)));
1795 av_unshift(PL_rsfp_filters, 1);
1796 av_store(PL_rsfp_filters, 0, datasv) ;
1801 /* Delete most recently added instance of this filter function. */
1803 Perl_filter_del(pTHX_ filter_t funcp)
1805 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp));
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. */
1835 DEBUG_P(PerlIO_printf(Perl_debug_log,
1836 "filter_read %d: from rsfp\n", idx));
1840 int old_len = SvCUR(buf_sv) ;
1842 /* ensure buf_sv is large enough */
1843 SvGROW(buf_sv, old_len + maxlen) ;
1844 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
1845 if (PerlIO_error(PL_rsfp))
1846 return -1; /* error */
1848 return 0 ; /* end of file */
1850 SvCUR_set(buf_sv, old_len + len) ;
1853 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
1854 if (PerlIO_error(PL_rsfp))
1855 return -1; /* error */
1857 return 0 ; /* end of file */
1860 return SvCUR(buf_sv);
1862 /* Skip this filter slot if filter has been deleted */
1863 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
1864 DEBUG_P(PerlIO_printf(Perl_debug_log,
1865 "filter_read %d: skipped (filter deleted)\n",
1867 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
1869 /* Get function pointer hidden within datasv */
1870 funcp = (filter_t)IoDIRP(datasv);
1871 DEBUG_P(PerlIO_printf(Perl_debug_log,
1872 "filter_read %d: via function %p (%s)\n",
1873 idx, funcp, SvPV_nolen(datasv)));
1874 /* Call function. The function is expected to */
1875 /* call "FILTER_READ(idx+1, buf_sv)" first. */
1876 /* Return: <0:error, =0:eof, >0:not eof */
1877 return (*funcp)(aTHXo_ idx, buf_sv, maxlen);
1881 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
1884 if (!PL_rsfp_filters) {
1885 filter_add(win32_textfilter,NULL);
1888 if (PL_rsfp_filters) {
1891 SvCUR_set(sv, 0); /* start with empty line */
1892 if (FILTER_READ(0, sv, 0) > 0)
1893 return ( SvPVX(sv) ) ;
1898 return (sv_gets(sv, fp, append));
1903 static char* exp_name[] =
1904 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
1905 "ATTRTERM", "TERMBLOCK"
1912 Works out what to call the token just pulled out of the input
1913 stream. The yacc parser takes care of taking the ops we return and
1914 stitching them into a tree.
1920 if read an identifier
1921 if we're in a my declaration
1922 croak if they tried to say my($foo::bar)
1923 build the ops for a my() declaration
1924 if it's an access to a my() variable
1925 are we in a sort block?
1926 croak if my($a); $a <=> $b
1927 build ops for access to a my() variable
1928 if in a dq string, and they've said @foo and we can't find @foo
1930 build ops for a bareword
1931 if we already built the token before, use it.
1935 #ifdef USE_PURE_BISON
1936 Perl_yylex(pTHX_ YYSTYPE *lvalp, int *lcharp)
1949 #ifdef USE_PURE_BISON
1950 yylval_pointer = lvalp;
1951 yychar_pointer = lcharp;
1954 /* check if there's an identifier for us to look at */
1955 if (PL_pending_ident) {
1956 /* pit holds the identifier we read and pending_ident is reset */
1957 char pit = PL_pending_ident;
1958 PL_pending_ident = 0;
1960 /* if we're in a my(), we can't allow dynamics here.
1961 $foo'bar has already been turned into $foo::bar, so
1962 just check for colons.
1964 if it's a legal name, the OP is a PADANY.
1967 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
1968 tmp = pad_allocmy(PL_tokenbuf);
1971 if (strchr(PL_tokenbuf,':'))
1972 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
1974 yylval.opval = newOP(OP_PADANY, 0);
1975 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
1981 build the ops for accesses to a my() variable.
1983 Deny my($a) or my($b) in a sort block, *if* $a or $b is
1984 then used in a comparison. This catches most, but not
1985 all cases. For instance, it catches
1986 sort { my($a); $a <=> $b }
1988 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
1989 (although why you'd do that is anyone's guess).
1992 if (!strchr(PL_tokenbuf,':')) {
1994 /* Check for single character per-thread SVs */
1995 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
1996 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
1997 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
1999 yylval.opval = newOP(OP_THREADSV, 0);
2000 yylval.opval->op_targ = tmp;
2003 #endif /* USE_THREADS */
2004 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
2005 /* might be an "our" variable" */
2006 if (SvFLAGS(AvARRAY(PL_comppad_name)[tmp]) & SVpad_OUR) {
2007 /* build ops for a bareword */
2008 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2009 yylval.opval->op_private = OPpCONST_ENTERED;
2010 gv_fetchpv(PL_tokenbuf+1,
2012 ? (GV_ADDMULTI | GV_ADDINEVAL | GV_ADDOUR)
2015 ((PL_tokenbuf[0] == '$') ? SVt_PV
2016 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2021 /* if it's a sort block and they're naming $a or $b */
2022 if (PL_last_lop_op == OP_SORT &&
2023 PL_tokenbuf[0] == '$' &&
2024 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
2027 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
2028 d < PL_bufend && *d != '\n';
2031 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
2032 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
2038 yylval.opval = newOP(OP_PADANY, 0);
2039 yylval.opval->op_targ = tmp;
2045 Whine if they've said @foo in a doublequoted string,
2046 and @foo isn't a variable we can find in the symbol
2049 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
2050 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
2051 if (!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
2052 yyerror(Perl_form(aTHX_ "In string, %s now must be written as \\%s",
2053 PL_tokenbuf, PL_tokenbuf));
2056 /* build ops for a bareword */
2057 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2058 yylval.opval->op_private = OPpCONST_ENTERED;
2059 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
2060 ((PL_tokenbuf[0] == '$') ? SVt_PV
2061 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2066 /* no identifier pending identification */
2068 switch (PL_lex_state) {
2070 case LEX_NORMAL: /* Some compilers will produce faster */
2071 case LEX_INTERPNORMAL: /* code if we comment these out. */
2075 /* when we've already built the next token, just pull it out of the queue */
2078 yylval = PL_nextval[PL_nexttoke];
2080 PL_lex_state = PL_lex_defer;
2081 PL_expect = PL_lex_expect;
2082 PL_lex_defer = LEX_NORMAL;
2084 return(PL_nexttype[PL_nexttoke]);
2086 /* interpolated case modifiers like \L \U, including \Q and \E.
2087 when we get here, PL_bufptr is at the \
2089 case LEX_INTERPCASEMOD:
2091 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2092 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2094 /* handle \E or end of string */
2095 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2099 if (PL_lex_casemods) {
2100 oldmod = PL_lex_casestack[--PL_lex_casemods];
2101 PL_lex_casestack[PL_lex_casemods] = '\0';
2103 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2105 PL_lex_state = LEX_INTERPCONCAT;
2109 if (PL_bufptr != PL_bufend)
2111 PL_lex_state = LEX_INTERPCONCAT;
2116 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2117 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2118 if (strchr("LU", *s) &&
2119 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2121 PL_lex_casestack[--PL_lex_casemods] = '\0';
2124 if (PL_lex_casemods > 10) {
2125 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2126 if (newlb != PL_lex_casestack) {
2128 PL_lex_casestack = newlb;
2131 PL_lex_casestack[PL_lex_casemods++] = *s;
2132 PL_lex_casestack[PL_lex_casemods] = '\0';
2133 PL_lex_state = LEX_INTERPCONCAT;
2134 PL_nextval[PL_nexttoke].ival = 0;
2137 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2139 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2141 PL_nextval[PL_nexttoke].ival = OP_LC;
2143 PL_nextval[PL_nexttoke].ival = OP_UC;
2145 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2147 Perl_croak(aTHX_ "panic: yylex");
2150 if (PL_lex_starts) {
2159 case LEX_INTERPPUSH:
2160 return sublex_push();
2162 case LEX_INTERPSTART:
2163 if (PL_bufptr == PL_bufend)
2164 return sublex_done();
2166 PL_lex_dojoin = (*PL_bufptr == '@');
2167 PL_lex_state = LEX_INTERPNORMAL;
2168 if (PL_lex_dojoin) {
2169 PL_nextval[PL_nexttoke].ival = 0;
2172 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2173 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2174 force_next(PRIVATEREF);
2176 force_ident("\"", '$');
2177 #endif /* USE_THREADS */
2178 PL_nextval[PL_nexttoke].ival = 0;
2180 PL_nextval[PL_nexttoke].ival = 0;
2182 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2185 if (PL_lex_starts++) {
2191 case LEX_INTERPENDMAYBE:
2192 if (intuit_more(PL_bufptr)) {
2193 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2199 if (PL_lex_dojoin) {
2200 PL_lex_dojoin = FALSE;
2201 PL_lex_state = LEX_INTERPCONCAT;
2204 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2205 && SvEVALED(PL_lex_repl))
2207 if (PL_bufptr != PL_bufend)
2208 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2209 PL_lex_repl = Nullsv;
2212 case LEX_INTERPCONCAT:
2214 if (PL_lex_brackets)
2215 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2217 if (PL_bufptr == PL_bufend)
2218 return sublex_done();
2220 if (SvIVX(PL_linestr) == '\'') {
2221 SV *sv = newSVsv(PL_linestr);
2224 else if ( PL_hints & HINT_NEW_RE )
2225 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2226 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2230 s = scan_const(PL_bufptr);
2232 PL_lex_state = LEX_INTERPCASEMOD;
2234 PL_lex_state = LEX_INTERPSTART;
2237 if (s != PL_bufptr) {
2238 PL_nextval[PL_nexttoke] = yylval;
2241 if (PL_lex_starts++)
2251 PL_lex_state = LEX_NORMAL;
2252 s = scan_formline(PL_bufptr);
2253 if (!PL_lex_formbrack)
2259 PL_oldoldbufptr = PL_oldbufptr;
2262 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2263 exp_name[PL_expect], s);
2269 if (isIDFIRST_lazy(s))
2271 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2274 goto fake_eof; /* emulate EOF on ^D or ^Z */
2279 if (PL_lex_brackets)
2280 yyerror("Missing right curly or square bracket");
2283 if (s++ < PL_bufend)
2284 goto retry; /* ignore stray nulls */
2287 if (!PL_in_eval && !PL_preambled) {
2288 PL_preambled = TRUE;
2289 sv_setpv(PL_linestr,incl_perldb());
2290 if (SvCUR(PL_linestr))
2291 sv_catpv(PL_linestr,";");
2293 while(AvFILLp(PL_preambleav) >= 0) {
2294 SV *tmpsv = av_shift(PL_preambleav);
2295 sv_catsv(PL_linestr, tmpsv);
2296 sv_catpv(PL_linestr, ";");
2299 sv_free((SV*)PL_preambleav);
2300 PL_preambleav = NULL;
2302 if (PL_minus_n || PL_minus_p) {
2303 sv_catpv(PL_linestr, "LINE: while (<>) {");
2305 sv_catpv(PL_linestr,"chomp;");
2307 GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
2309 GvIMPORTED_AV_on(gv);
2311 if (strchr("/'\"", *PL_splitstr)
2312 && strchr(PL_splitstr + 1, *PL_splitstr))
2313 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2316 s = "'~#\200\1'"; /* surely one char is unused...*/
2317 while (s[1] && strchr(PL_splitstr, *s)) s++;
2319 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c",
2320 "q" + (delim == '\''), delim);
2321 for (s = PL_splitstr; *s; s++) {
2323 sv_catpvn(PL_linestr, "\\", 1);
2324 sv_catpvn(PL_linestr, s, 1);
2326 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2330 sv_catpv(PL_linestr,"@F=split(' ');");
2333 sv_catpv(PL_linestr, "\n");
2334 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2335 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2336 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2337 SV *sv = NEWSV(85,0);
2339 sv_upgrade(sv, SVt_PVMG);
2340 sv_setsv(sv,PL_linestr);
2341 av_store(CopFILEAV(PL_curcop),(I32)PL_curcop->cop_line,sv);
2346 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2349 if (PL_preprocess && !PL_in_eval)
2350 (void)PerlProc_pclose(PL_rsfp);
2351 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2352 PerlIO_clearerr(PL_rsfp);
2354 (void)PerlIO_close(PL_rsfp);
2356 PL_doextract = FALSE;
2358 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2359 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2360 sv_catpv(PL_linestr,";}");
2361 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2362 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2363 PL_minus_n = PL_minus_p = 0;
2366 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2367 sv_setpv(PL_linestr,"");
2368 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2371 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2372 PL_doextract = FALSE;
2374 /* Incest with pod. */
2375 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2376 sv_setpv(PL_linestr, "");
2377 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2378 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2379 PL_doextract = FALSE;
2383 } while (PL_doextract);
2384 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2385 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2386 SV *sv = NEWSV(85,0);
2388 sv_upgrade(sv, SVt_PVMG);
2389 sv_setsv(sv,PL_linestr);
2390 av_store(CopFILEAV(PL_curcop),(I32)PL_curcop->cop_line,sv);
2392 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2393 if (PL_curcop->cop_line == 1) {
2394 while (s < PL_bufend && isSPACE(*s))
2396 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2400 if (*s == '#' && *(s+1) == '!')
2402 #ifdef ALTERNATE_SHEBANG
2404 static char as[] = ALTERNATE_SHEBANG;
2405 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2406 d = s + (sizeof(as) - 1);
2408 #endif /* ALTERNATE_SHEBANG */
2417 while (*d && !isSPACE(*d))
2421 #ifdef ARG_ZERO_IS_SCRIPT
2422 if (ipathend > ipath) {
2424 * HP-UX (at least) sets argv[0] to the script name,
2425 * which makes $^X incorrect. And Digital UNIX and Linux,
2426 * at least, set argv[0] to the basename of the Perl
2427 * interpreter. So, having found "#!", we'll set it right.
2429 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2430 assert(SvPOK(x) || SvGMAGICAL(x));
2431 if (sv_eq(x, CopFILESV(PL_curcop))) {
2432 sv_setpvn(x, ipath, ipathend - ipath);
2435 TAINT_NOT; /* $^X is always tainted, but that's OK */
2437 #endif /* ARG_ZERO_IS_SCRIPT */
2442 d = instr(s,"perl -");
2444 d = instr(s,"perl");
2446 /* avoid getting into infinite loops when shebang
2447 * line contains "Perl" rather than "perl" */
2449 for (d = ipathend-4; d >= ipath; --d) {
2450 if ((*d == 'p' || *d == 'P')
2451 && !ibcmp(d, "perl", 4))
2461 #ifdef ALTERNATE_SHEBANG
2463 * If the ALTERNATE_SHEBANG on this system starts with a
2464 * character that can be part of a Perl expression, then if
2465 * we see it but not "perl", we're probably looking at the
2466 * start of Perl code, not a request to hand off to some
2467 * other interpreter. Similarly, if "perl" is there, but
2468 * not in the first 'word' of the line, we assume the line
2469 * contains the start of the Perl program.
2471 if (d && *s != '#') {
2473 while (*c && !strchr("; \t\r\n\f\v#", *c))
2476 d = Nullch; /* "perl" not in first word; ignore */
2478 *s = '#'; /* Don't try to parse shebang line */
2480 #endif /* ALTERNATE_SHEBANG */
2481 #ifndef MACOS_TRADITIONAL
2486 !instr(s,"indir") &&
2487 instr(PL_origargv[0],"perl"))
2493 while (s < PL_bufend && isSPACE(*s))
2495 if (s < PL_bufend) {
2496 Newz(899,newargv,PL_origargc+3,char*);
2498 while (s < PL_bufend && !isSPACE(*s))
2501 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2504 newargv = PL_origargv;
2506 PerlProc_execv(ipath, newargv);
2507 Perl_croak(aTHX_ "Can't exec %s", ipath);
2511 U32 oldpdb = PL_perldb;
2512 bool oldn = PL_minus_n;
2513 bool oldp = PL_minus_p;
2515 while (*d && !isSPACE(*d)) d++;
2516 while (SPACE_OR_TAB(*d)) d++;
2520 if (*d == 'M' || *d == 'm') {
2522 while (*d && !isSPACE(*d)) d++;
2523 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2526 d = moreswitches(d);
2528 if (PERLDB_LINE && !oldpdb ||
2529 ( PL_minus_n || PL_minus_p ) && !(oldn || oldp) )
2530 /* if we have already added "LINE: while (<>) {",
2531 we must not do it again */
2533 sv_setpv(PL_linestr, "");
2534 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2535 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2536 PL_preambled = FALSE;
2538 (void)gv_fetchfile(PL_origfilename);
2545 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2547 PL_lex_state = LEX_FORMLINE;
2552 #ifdef PERL_STRICT_CR
2553 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2555 "(Maybe you didn't strip carriage returns after a network transfer?)\n");
2557 case ' ': case '\t': case '\f': case 013:
2558 #ifdef MACOS_TRADITIONAL
2559 case '\312': /* Them nonbreaking spaces again */
2565 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2567 while (s < d && *s != '\n')
2572 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2574 PL_lex_state = LEX_FORMLINE;
2584 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2589 while (s < PL_bufend && SPACE_OR_TAB(*s))
2592 if (strnEQ(s,"=>",2)) {
2593 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2594 OPERATOR('-'); /* unary minus */
2596 PL_last_uni = PL_oldbufptr;
2597 PL_last_lop_op = OP_FTEREAD; /* good enough */
2599 case 'r': FTST(OP_FTEREAD);
2600 case 'w': FTST(OP_FTEWRITE);
2601 case 'x': FTST(OP_FTEEXEC);
2602 case 'o': FTST(OP_FTEOWNED);
2603 case 'R': FTST(OP_FTRREAD);
2604 case 'W': FTST(OP_FTRWRITE);
2605 case 'X': FTST(OP_FTREXEC);
2606 case 'O': FTST(OP_FTROWNED);
2607 case 'e': FTST(OP_FTIS);
2608 case 'z': FTST(OP_FTZERO);
2609 case 's': FTST(OP_FTSIZE);
2610 case 'f': FTST(OP_FTFILE);
2611 case 'd': FTST(OP_FTDIR);
2612 case 'l': FTST(OP_FTLINK);
2613 case 'p': FTST(OP_FTPIPE);
2614 case 'S': FTST(OP_FTSOCK);
2615 case 'u': FTST(OP_FTSUID);
2616 case 'g': FTST(OP_FTSGID);
2617 case 'k': FTST(OP_FTSVTX);
2618 case 'b': FTST(OP_FTBLK);
2619 case 'c': FTST(OP_FTCHR);
2620 case 't': FTST(OP_FTTTY);
2621 case 'T': FTST(OP_FTTEXT);
2622 case 'B': FTST(OP_FTBINARY);
2623 case 'M': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTMTIME);
2624 case 'A': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTATIME);
2625 case 'C': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTCTIME);
2627 Perl_croak(aTHX_ "Unrecognized file test: -%c", (int)tmp);
2634 if (PL_expect == XOPERATOR)
2639 else if (*s == '>') {
2642 if (isIDFIRST_lazy(s)) {
2643 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2651 if (PL_expect == XOPERATOR)
2654 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2656 OPERATOR('-'); /* unary minus */
2663 if (PL_expect == XOPERATOR)
2668 if (PL_expect == XOPERATOR)
2671 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2677 if (PL_expect != XOPERATOR) {
2678 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2679 PL_expect = XOPERATOR;
2680 force_ident(PL_tokenbuf, '*');
2693 if (PL_expect == XOPERATOR) {
2697 PL_tokenbuf[0] = '%';
2698 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2699 if (!PL_tokenbuf[1]) {
2701 yyerror("Final % should be \\% or %name");
2704 PL_pending_ident = '%';
2723 switch (PL_expect) {
2726 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2728 PL_bufptr = s; /* update in case we back off */
2734 PL_expect = XTERMBLOCK;
2738 while (isIDFIRST_lazy(s)) {
2739 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2741 d = scan_str(d,TRUE,TRUE);
2744 SvREFCNT_dec(PL_lex_stuff);
2745 PL_lex_stuff = Nullsv;
2747 /* MUST advance bufptr here to avoid bogus
2748 "at end of line" context messages from yyerror().
2750 PL_bufptr = s + len;
2751 yyerror("Unterminated attribute parameter in attribute list");
2754 return 0; /* EOF indicator */
2758 SV *sv = newSVpvn(s, len);
2759 sv_catsv(sv, PL_lex_stuff);
2760 attrs = append_elem(OP_LIST, attrs,
2761 newSVOP(OP_CONST, 0, sv));
2762 SvREFCNT_dec(PL_lex_stuff);
2763 PL_lex_stuff = Nullsv;
2766 attrs = append_elem(OP_LIST, attrs,
2767 newSVOP(OP_CONST, 0,
2774 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}' for vi */
2775 if (*s != ';' && *s != tmp) {
2776 char q = ((*s == '\'') ? '"' : '\'');
2777 /* If here for an expression, and parsed no attrs, back off. */
2778 if (tmp == '=' && !attrs) {
2782 /* MUST advance bufptr here to avoid bogus "at end of line"
2783 context messages from yyerror().
2787 yyerror("Unterminated attribute list");
2789 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
2796 PL_nextval[PL_nexttoke].opval = attrs;
2804 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
2805 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
2810 if (PL_curcop->cop_line < PL_copline)
2811 PL_copline = PL_curcop->cop_line;
2822 if (PL_lex_brackets <= 0)
2823 yyerror("Unmatched right square bracket");
2826 if (PL_lex_state == LEX_INTERPNORMAL) {
2827 if (PL_lex_brackets == 0) {
2828 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
2829 PL_lex_state = LEX_INTERPEND;
2836 if (PL_lex_brackets > 100) {
2837 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
2838 if (newlb != PL_lex_brackstack) {
2840 PL_lex_brackstack = newlb;
2843 switch (PL_expect) {
2845 if (PL_lex_formbrack) {
2849 if (PL_oldoldbufptr == PL_last_lop)
2850 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
2852 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2853 OPERATOR(HASHBRACK);
2855 while (s < PL_bufend && SPACE_OR_TAB(*s))
2858 PL_tokenbuf[0] = '\0';
2859 if (d < PL_bufend && *d == '-') {
2860 PL_tokenbuf[0] = '-';
2862 while (d < PL_bufend && SPACE_OR_TAB(*d))
2865 if (d < PL_bufend && isIDFIRST_lazy(d)) {
2866 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
2868 while (d < PL_bufend && SPACE_OR_TAB(*d))
2871 char minus = (PL_tokenbuf[0] == '-');
2872 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
2880 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
2885 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2890 if (PL_oldoldbufptr == PL_last_lop)
2891 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
2893 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2896 OPERATOR(HASHBRACK);
2897 /* This hack serves to disambiguate a pair of curlies
2898 * as being a block or an anon hash. Normally, expectation
2899 * determines that, but in cases where we're not in a
2900 * position to expect anything in particular (like inside
2901 * eval"") we have to resolve the ambiguity. This code
2902 * covers the case where the first term in the curlies is a
2903 * quoted string. Most other cases need to be explicitly
2904 * disambiguated by prepending a `+' before the opening
2905 * curly in order to force resolution as an anon hash.
2907 * XXX should probably propagate the outer expectation
2908 * into eval"" to rely less on this hack, but that could
2909 * potentially break current behavior of eval"".
2913 if (*s == '\'' || *s == '"' || *s == '`') {
2914 /* common case: get past first string, handling escapes */
2915 for (t++; t < PL_bufend && *t != *s;)
2916 if (*t++ == '\\' && (*t == '\\' || *t == *s))
2920 else if (*s == 'q') {
2923 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
2924 && !isALNUM(*t)))) {
2926 char open, close, term;
2929 while (t < PL_bufend && isSPACE(*t))
2933 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
2937 for (t++; t < PL_bufend; t++) {
2938 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
2940 else if (*t == open)
2944 for (t++; t < PL_bufend; t++) {
2945 if (*t == '\\' && t+1 < PL_bufend)
2947 else if (*t == close && --brackets <= 0)
2949 else if (*t == open)
2955 else if (isIDFIRST_lazy(s)) {
2956 for (t++; t < PL_bufend && isALNUM_lazy(t); t++) ;
2958 while (t < PL_bufend && isSPACE(*t))
2960 /* if comma follows first term, call it an anon hash */
2961 /* XXX it could be a comma expression with loop modifiers */
2962 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
2963 || (*t == '=' && t[1] == '>')))
2964 OPERATOR(HASHBRACK);
2965 if (PL_expect == XREF)
2968 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
2974 yylval.ival = PL_curcop->cop_line;
2975 if (isSPACE(*s) || *s == '#')
2976 PL_copline = NOLINE; /* invalidate current command line number */
2981 if (PL_lex_brackets <= 0)
2982 yyerror("Unmatched right curly bracket");
2984 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
2985 if (PL_lex_brackets < PL_lex_formbrack)
2986 PL_lex_formbrack = 0;
2987 if (PL_lex_state == LEX_INTERPNORMAL) {
2988 if (PL_lex_brackets == 0) {
2989 if (PL_lex_fakebrack) {
2990 PL_lex_state = LEX_INTERPEND;
2992 return yylex(); /* ignore fake brackets */
2994 if (*s == '-' && s[1] == '>')
2995 PL_lex_state = LEX_INTERPENDMAYBE;
2996 else if (*s != '[' && *s != '{')
2997 PL_lex_state = LEX_INTERPEND;
3000 if (PL_lex_brackets < PL_lex_fakebrack) {
3002 PL_lex_fakebrack = 0;
3003 return yylex(); /* ignore fake brackets */
3013 if (PL_expect == XOPERATOR) {
3014 if (ckWARN(WARN_SEMICOLON) && isIDFIRST_lazy(s) && PL_bufptr == PL_linestart) {
3015 PL_curcop->cop_line--;
3016 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3017 PL_curcop->cop_line++;
3022 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3024 PL_expect = XOPERATOR;
3025 force_ident(PL_tokenbuf, '&');
3029 yylval.ival = (OPpENTERSUB_AMPER<<8);
3048 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3049 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3051 if (PL_expect == XSTATE && isALPHA(tmp) &&
3052 (s == PL_linestart+1 || s[-2] == '\n') )
3054 if (PL_in_eval && !PL_rsfp) {
3059 if (strnEQ(s,"=cut",4)) {
3073 PL_doextract = TRUE;
3076 if (PL_lex_brackets < PL_lex_formbrack) {
3078 #ifdef PERL_STRICT_CR
3079 for (t = s; SPACE_OR_TAB(*t); t++) ;
3081 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3083 if (*t == '\n' || *t == '#') {
3101 if (PL_expect != XOPERATOR) {
3102 if (s[1] != '<' && !strchr(s,'>'))
3105 s = scan_heredoc(s);
3107 s = scan_inputsymbol(s);
3108 TERM(sublex_start());
3113 SHop(OP_LEFT_SHIFT);
3127 SHop(OP_RIGHT_SHIFT);
3136 if (PL_expect == XOPERATOR) {
3137 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3140 return ','; /* grandfather non-comma-format format */
3144 if (s[1] == '#' && (isIDFIRST_lazy(s+2) || strchr("{$:+-", s[2]))) {
3145 PL_tokenbuf[0] = '@';
3146 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3147 sizeof PL_tokenbuf - 1, FALSE);
3148 if (PL_expect == XOPERATOR)
3149 no_op("Array length", s);
3150 if (!PL_tokenbuf[1])
3152 PL_expect = XOPERATOR;
3153 PL_pending_ident = '#';
3157 PL_tokenbuf[0] = '$';
3158 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3159 sizeof PL_tokenbuf - 1, FALSE);
3160 if (PL_expect == XOPERATOR)
3162 if (!PL_tokenbuf[1]) {
3164 yyerror("Final $ should be \\$ or $name");
3168 /* This kludge not intended to be bulletproof. */
3169 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3170 yylval.opval = newSVOP(OP_CONST, 0,
3171 newSViv((IV)PL_compiling.cop_arybase));
3172 yylval.opval->op_private = OPpCONST_ARYBASE;
3178 if (PL_lex_state == LEX_NORMAL)
3181 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3184 PL_tokenbuf[0] = '@';
3185 if (ckWARN(WARN_SYNTAX)) {
3187 isSPACE(*t) || isALNUM_lazy(t) || *t == '$';
3190 PL_bufptr = skipspace(PL_bufptr);
3191 while (t < PL_bufend && *t != ']')
3193 Perl_warner(aTHX_ WARN_SYNTAX,
3194 "Multidimensional syntax %.*s not supported",
3195 (t - PL_bufptr) + 1, PL_bufptr);
3199 else if (*s == '{') {
3200 PL_tokenbuf[0] = '%';
3201 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3202 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3204 char tmpbuf[sizeof PL_tokenbuf];
3206 for (t++; isSPACE(*t); t++) ;
3207 if (isIDFIRST_lazy(t)) {
3208 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3209 for (; isSPACE(*t); t++) ;
3210 if (*t == ';' && get_cv(tmpbuf, FALSE))
3211 Perl_warner(aTHX_ WARN_SYNTAX,
3212 "You need to quote \"%s\"", tmpbuf);
3218 PL_expect = XOPERATOR;
3219 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3220 bool islop = (PL_last_lop == PL_oldoldbufptr);
3221 if (!islop || PL_last_lop_op == OP_GREPSTART)
3222 PL_expect = XOPERATOR;
3223 else if (strchr("$@\"'`q", *s))
3224 PL_expect = XTERM; /* e.g. print $fh "foo" */
3225 else if (strchr("&*<%", *s) && isIDFIRST_lazy(s+1))
3226 PL_expect = XTERM; /* e.g. print $fh &sub */
3227 else if (isIDFIRST_lazy(s)) {
3228 char tmpbuf[sizeof PL_tokenbuf];
3229 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3230 if (tmp = keyword(tmpbuf, len)) {
3231 /* binary operators exclude handle interpretations */
3243 PL_expect = XTERM; /* e.g. print $fh length() */
3248 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3249 if (gv && GvCVu(gv))
3250 PL_expect = XTERM; /* e.g. print $fh subr() */
3253 else if (isDIGIT(*s))
3254 PL_expect = XTERM; /* e.g. print $fh 3 */
3255 else if (*s == '.' && isDIGIT(s[1]))
3256 PL_expect = XTERM; /* e.g. print $fh .3 */
3257 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3258 PL_expect = XTERM; /* e.g. print $fh -1 */
3259 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3260 PL_expect = XTERM; /* print $fh <<"EOF" */
3262 PL_pending_ident = '$';
3266 if (PL_expect == XOPERATOR)
3268 PL_tokenbuf[0] = '@';
3269 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3270 if (!PL_tokenbuf[1]) {
3272 yyerror("Final @ should be \\@ or @name");
3275 if (PL_lex_state == LEX_NORMAL)
3277 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3279 PL_tokenbuf[0] = '%';
3281 /* Warn about @ where they meant $. */
3282 if (ckWARN(WARN_SYNTAX)) {
3283 if (*s == '[' || *s == '{') {
3285 while (*t && (isALNUM_lazy(t) || strchr(" \t$#+-'\"", *t)))
3287 if (*t == '}' || *t == ']') {
3289 PL_bufptr = skipspace(PL_bufptr);
3290 Perl_warner(aTHX_ WARN_SYNTAX,
3291 "Scalar value %.*s better written as $%.*s",
3292 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3297 PL_pending_ident = '@';
3300 case '/': /* may either be division or pattern */
3301 case '?': /* may either be conditional or pattern */
3302 if (PL_expect != XOPERATOR) {
3303 /* Disable warning on "study /blah/" */
3304 if (PL_oldoldbufptr == PL_last_uni
3305 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3306 || memNE(PL_last_uni, "study", 5) || isALNUM_lazy(PL_last_uni+5)))
3308 s = scan_pat(s,OP_MATCH);
3309 TERM(sublex_start());
3317 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3318 #ifdef PERL_STRICT_CR
3321 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3323 && (s == PL_linestart || s[-1] == '\n') )
3325 PL_lex_formbrack = 0;
3329 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3335 yylval.ival = OPf_SPECIAL;
3341 if (PL_expect != XOPERATOR)
3346 case '0': case '1': case '2': case '3': case '4':
3347 case '5': case '6': case '7': case '8': case '9':
3349 if (PL_expect == XOPERATOR)
3354 s = scan_str(s,FALSE,FALSE);
3355 if (PL_expect == XOPERATOR) {
3356 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3359 return ','; /* grandfather non-comma-format format */
3365 missingterm((char*)0);
3366 yylval.ival = OP_CONST;
3367 TERM(sublex_start());
3370 s = scan_str(s,FALSE,FALSE);
3371 if (PL_expect == XOPERATOR) {
3372 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3375 return ','; /* grandfather non-comma-format format */
3381 missingterm((char*)0);
3382 yylval.ival = OP_CONST;
3383 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3384 if (*d == '$' || *d == '@' || *d == '\\' || *d & 0x80) {
3385 yylval.ival = OP_STRINGIFY;
3389 TERM(sublex_start());
3392 s = scan_str(s,FALSE,FALSE);
3393 if (PL_expect == XOPERATOR)
3394 no_op("Backticks",s);
3396 missingterm((char*)0);
3397 yylval.ival = OP_BACKTICK;
3399 TERM(sublex_start());
3403 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3404 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3406 if (PL_expect == XOPERATOR)
3407 no_op("Backslash",s);
3411 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3451 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3453 /* Some keywords can be followed by any delimiter, including ':' */
3454 tmp = (len == 1 && strchr("msyq", PL_tokenbuf[0]) ||
3455 len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3456 (PL_tokenbuf[0] == 'q' &&
3457 strchr("qwxr", PL_tokenbuf[1]))));
3459 /* x::* is just a word, unless x is "CORE" */
3460 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3464 while (d < PL_bufend && isSPACE(*d))
3465 d++; /* no comments skipped here, or s### is misparsed */
3467 /* Is this a label? */
3468 if (!tmp && PL_expect == XSTATE
3469 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3471 yylval.pval = savepv(PL_tokenbuf);
3476 /* Check for keywords */
3477 tmp = keyword(PL_tokenbuf, len);
3479 /* Is this a word before a => operator? */
3480 if (strnEQ(d,"=>",2)) {
3482 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3483 yylval.opval->op_private = OPpCONST_BARE;
3487 if (tmp < 0) { /* second-class keyword? */
3488 GV *ogv = Nullgv; /* override (winner) */
3489 GV *hgv = Nullgv; /* hidden (loser) */
3490 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3492 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3495 if (GvIMPORTED_CV(gv))
3497 else if (! CvMETHOD(cv))
3501 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3502 (gv = *gvp) != (GV*)&PL_sv_undef &&
3503 GvCVu(gv) && GvIMPORTED_CV(gv))
3509 tmp = 0; /* overridden by import or by GLOBAL */
3512 && -tmp==KEY_lock /* XXX generalizable kludge */
3513 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3515 tmp = 0; /* any sub overrides "weak" keyword */
3517 else { /* no override */
3521 if (ckWARN(WARN_AMBIGUOUS) && hgv
3522 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3523 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3524 "Ambiguous call resolved as CORE::%s(), %s",
3525 GvENAME(hgv), "qualify as such or use &");
3532 default: /* not a keyword */
3535 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3537 /* Get the rest if it looks like a package qualifier */
3539 if (*s == '\'' || *s == ':' && s[1] == ':') {
3541 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3544 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3545 *s == '\'' ? "'" : "::");
3549 if (PL_expect == XOPERATOR) {
3550 if (PL_bufptr == PL_linestart) {
3551 PL_curcop->cop_line--;
3552 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3553 PL_curcop->cop_line++;
3556 no_op("Bareword",s);
3559 /* Look for a subroutine with this name in current package,
3560 unless name is "Foo::", in which case Foo is a bearword
3561 (and a package name). */
3564 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3566 if (ckWARN(WARN_UNSAFE) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3567 Perl_warner(aTHX_ WARN_UNSAFE,
3568 "Bareword \"%s\" refers to nonexistent package",
3571 PL_tokenbuf[len] = '\0';
3578 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3581 /* if we saw a global override before, get the right name */
3584 sv = newSVpvn("CORE::GLOBAL::",14);
3585 sv_catpv(sv,PL_tokenbuf);
3588 sv = newSVpv(PL_tokenbuf,0);
3590 /* Presume this is going to be a bareword of some sort. */
3593 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3594 yylval.opval->op_private = OPpCONST_BARE;
3596 /* And if "Foo::", then that's what it certainly is. */
3601 /* See if it's the indirect object for a list operator. */
3603 if (PL_oldoldbufptr &&
3604 PL_oldoldbufptr < PL_bufptr &&
3605 (PL_oldoldbufptr == PL_last_lop
3606 || 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; SPACE_OR_TAB(*d); 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(CopFILESV(PL_curcop)));
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)) = '<';
3794 if (PL_expect == XSTATE) {
3801 if (*s == ':' && s[1] == ':') {
3804 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3805 tmp = keyword(PL_tokenbuf, len);
3819 LOP(OP_ACCEPT,XTERM);
3825 LOP(OP_ATAN2,XTERM);
3834 LOP(OP_BLESS,XTERM);
3843 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
3860 if (!PL_cryptseen) {
3861 PL_cryptseen = TRUE;
3865 LOP(OP_CRYPT,XTERM);
3868 if (ckWARN(WARN_OCTAL)) {
3869 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
3870 if (*d != '0' && isDIGIT(*d))
3871 Perl_warner(aTHX_ WARN_OCTAL,
3872 "chmod: mode argument is missing initial 0");
3874 LOP(OP_CHMOD,XTERM);
3877 LOP(OP_CHOWN,XTERM);
3880 LOP(OP_CONNECT,XTERM);
3896 s = force_word(s,WORD,FALSE,TRUE,FALSE);
3900 PL_hints |= HINT_BLOCK_SCOPE;
3910 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
3911 LOP(OP_DBMOPEN,XTERM);
3917 s = force_word(s,WORD,TRUE,FALSE,FALSE);
3924 yylval.ival = PL_curcop->cop_line;
3938 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
3939 UNIBRACK(OP_ENTEREVAL);
3954 case KEY_endhostent:
3960 case KEY_endservent:
3963 case KEY_endprotoent:
3974 yylval.ival = PL_curcop->cop_line;
3976 if (PL_expect == XSTATE && isIDFIRST_lazy(s)) {
3978 if ((PL_bufend - p) >= 3 &&
3979 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
3981 else if ((PL_bufend - p) >= 4 &&
3982 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
3985 if (isIDFIRST_lazy(p)) {
3986 p = scan_ident(p, PL_bufend,
3987 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3991 Perl_croak(aTHX_ "Missing $ on loop variable");
3996 LOP(OP_FORMLINE,XTERM);
4002 LOP(OP_FCNTL,XTERM);
4008 LOP(OP_FLOCK,XTERM);
4017 LOP(OP_GREPSTART, *s == '(' ? XTERM : XREF);
4020 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4035 case KEY_getpriority:
4036 LOP(OP_GETPRIORITY,XTERM);
4038 case KEY_getprotobyname:
4041 case KEY_getprotobynumber:
4042 LOP(OP_GPBYNUMBER,XTERM);
4044 case KEY_getprotoent:
4056 case KEY_getpeername:
4057 UNI(OP_GETPEERNAME);
4059 case KEY_gethostbyname:
4062 case KEY_gethostbyaddr:
4063 LOP(OP_GHBYADDR,XTERM);
4065 case KEY_gethostent:
4068 case KEY_getnetbyname:
4071 case KEY_getnetbyaddr:
4072 LOP(OP_GNBYADDR,XTERM);
4077 case KEY_getservbyname:
4078 LOP(OP_GSBYNAME,XTERM);
4080 case KEY_getservbyport:
4081 LOP(OP_GSBYPORT,XTERM);
4083 case KEY_getservent:
4086 case KEY_getsockname:
4087 UNI(OP_GETSOCKNAME);
4089 case KEY_getsockopt:
4090 LOP(OP_GSOCKOPT,XTERM);
4112 yylval.ival = PL_curcop->cop_line;
4116 LOP(OP_INDEX,XTERM);
4122 LOP(OP_IOCTL,XTERM);
4134 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4166 LOP(OP_LISTEN,XTERM);
4175 s = scan_pat(s,OP_MATCH);
4176 TERM(sublex_start());
4179 LOP(OP_MAPSTART, *s == '(' ? XTERM : XREF);
4182 LOP(OP_MKDIR,XTERM);
4185 LOP(OP_MSGCTL,XTERM);
4188 LOP(OP_MSGGET,XTERM);
4191 LOP(OP_MSGRCV,XTERM);
4194 LOP(OP_MSGSND,XTERM);
4200 if (isIDFIRST_lazy(s)) {
4201 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4202 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4204 PL_in_my_stash = gv_stashpv(PL_tokenbuf, FALSE);
4205 if (!PL_in_my_stash) {
4208 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4216 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4223 if (PL_expect != XSTATE)
4224 yyerror("\"no\" not allowed in expression");
4225 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4226 s = force_version(s);
4231 if (*s == '(' || (s = skipspace(s), *s == '('))
4238 if (isIDFIRST_lazy(s)) {
4240 for (d = s; isALNUM_lazy(d); d++) ;
4242 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_AMBIGUOUS))
4243 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4244 "Precedence problem: open %.*s should be open(%.*s)",
4250 yylval.ival = OP_OR;
4260 LOP(OP_OPEN_DIR,XTERM);
4263 checkcomma(s,PL_tokenbuf,"filehandle");
4267 checkcomma(s,PL_tokenbuf,"filehandle");
4286 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4290 LOP(OP_PIPE_OP,XTERM);
4293 s = scan_str(s,FALSE,FALSE);
4295 missingterm((char*)0);
4296 yylval.ival = OP_CONST;
4297 TERM(sublex_start());
4303 s = scan_str(s,FALSE,FALSE);
4305 missingterm((char*)0);
4307 if (SvCUR(PL_lex_stuff)) {
4310 d = SvPV_force(PL_lex_stuff, len);
4312 for (; isSPACE(*d) && len; --len, ++d) ;
4315 if (!warned && ckWARN(WARN_SYNTAX)) {
4316 for (; !isSPACE(*d) && len; --len, ++d) {
4318 Perl_warner(aTHX_ WARN_SYNTAX,
4319 "Possible attempt to separate words with commas");
4322 else if (*d == '#') {
4323 Perl_warner(aTHX_ WARN_SYNTAX,
4324 "Possible attempt to put comments in qw() list");
4330 for (; !isSPACE(*d) && len; --len, ++d) ;
4332 words = append_elem(OP_LIST, words,
4333 newSVOP(OP_CONST, 0, newSVpvn(b, d-b)));
4337 PL_nextval[PL_nexttoke].opval = words;
4342 SvREFCNT_dec(PL_lex_stuff);
4343 PL_lex_stuff = Nullsv;
4348 s = scan_str(s,FALSE,FALSE);
4350 missingterm((char*)0);
4351 yylval.ival = OP_STRINGIFY;
4352 if (SvIVX(PL_lex_stuff) == '\'')
4353 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4354 TERM(sublex_start());
4357 s = scan_pat(s,OP_QR);
4358 TERM(sublex_start());
4361 s = scan_str(s,FALSE,FALSE);
4363 missingterm((char*)0);
4364 yylval.ival = OP_BACKTICK;
4366 TERM(sublex_start());
4372 *PL_tokenbuf = '\0';
4373 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4374 if (isIDFIRST_lazy(PL_tokenbuf))
4375 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4377 yyerror("<> should be quotes");
4384 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4388 LOP(OP_RENAME,XTERM);
4397 LOP(OP_RINDEX,XTERM);
4420 LOP(OP_REVERSE,XTERM);
4431 TERM(sublex_start());
4433 TOKEN(1); /* force error */
4442 LOP(OP_SELECT,XTERM);
4448 LOP(OP_SEMCTL,XTERM);
4451 LOP(OP_SEMGET,XTERM);
4454 LOP(OP_SEMOP,XTERM);
4460 LOP(OP_SETPGRP,XTERM);
4462 case KEY_setpriority:
4463 LOP(OP_SETPRIORITY,XTERM);
4465 case KEY_sethostent:
4471 case KEY_setservent:
4474 case KEY_setprotoent:
4484 LOP(OP_SEEKDIR,XTERM);
4486 case KEY_setsockopt:
4487 LOP(OP_SSOCKOPT,XTERM);
4493 LOP(OP_SHMCTL,XTERM);
4496 LOP(OP_SHMGET,XTERM);
4499 LOP(OP_SHMREAD,XTERM);
4502 LOP(OP_SHMWRITE,XTERM);
4505 LOP(OP_SHUTDOWN,XTERM);
4514 LOP(OP_SOCKET,XTERM);
4516 case KEY_socketpair:
4517 LOP(OP_SOCKPAIR,XTERM);
4520 checkcomma(s,PL_tokenbuf,"subroutine name");
4522 if (*s == ';' || *s == ')') /* probably a close */
4523 Perl_croak(aTHX_ "sort is now a reserved word");
4525 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4529 LOP(OP_SPLIT,XTERM);
4532 LOP(OP_SPRINTF,XTERM);
4535 LOP(OP_SPLICE,XTERM);
4550 LOP(OP_SUBSTR,XTERM);
4556 char tmpbuf[sizeof PL_tokenbuf];
4558 expectation attrful;
4559 bool have_name, have_proto;
4564 if (isIDFIRST_lazy(s) || *s == '\'' ||
4565 (*s == ':' && s[1] == ':'))
4568 attrful = XATTRBLOCK;
4569 /* remember buffer pos'n for later force_word */
4570 tboffset = s - PL_oldbufptr;
4571 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4572 if (strchr(tmpbuf, ':'))
4573 sv_setpv(PL_subname, tmpbuf);
4575 sv_setsv(PL_subname,PL_curstname);
4576 sv_catpvn(PL_subname,"::",2);
4577 sv_catpvn(PL_subname,tmpbuf,len);
4584 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4585 PL_expect = XTERMBLOCK;
4586 attrful = XATTRTERM;
4587 sv_setpv(PL_subname,"?");
4591 if (key == KEY_format) {
4593 PL_lex_formbrack = PL_lex_brackets + 1;
4595 (void) force_word(PL_oldbufptr + tboffset, WORD,
4600 /* Look for a prototype */
4604 s = scan_str(s,FALSE,FALSE);
4607 SvREFCNT_dec(PL_lex_stuff);
4608 PL_lex_stuff = Nullsv;
4609 Perl_croak(aTHX_ "Prototype not terminated");
4612 d = SvPVX(PL_lex_stuff);
4614 for (p = d; *p; ++p) {
4619 SvCUR(PL_lex_stuff) = tmp;
4627 if (*s == ':' && s[1] != ':')
4628 PL_expect = attrful;
4631 PL_nextval[PL_nexttoke].opval =
4632 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
4633 PL_lex_stuff = Nullsv;
4637 sv_setpv(PL_subname,"__ANON__");
4640 (void) force_word(PL_oldbufptr + tboffset, WORD,
4649 LOP(OP_SYSTEM,XREF);
4652 LOP(OP_SYMLINK,XTERM);
4655 LOP(OP_SYSCALL,XTERM);
4658 LOP(OP_SYSOPEN,XTERM);
4661 LOP(OP_SYSSEEK,XTERM);
4664 LOP(OP_SYSREAD,XTERM);
4667 LOP(OP_SYSWRITE,XTERM);
4671 TERM(sublex_start());
4692 LOP(OP_TRUNCATE,XTERM);
4704 yylval.ival = PL_curcop->cop_line;
4708 yylval.ival = PL_curcop->cop_line;
4712 LOP(OP_UNLINK,XTERM);
4718 LOP(OP_UNPACK,XTERM);
4721 LOP(OP_UTIME,XTERM);
4724 if (ckWARN(WARN_OCTAL)) {
4725 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4726 if (*d != '0' && isDIGIT(*d))
4727 Perl_warner(aTHX_ WARN_OCTAL,
4728 "umask: argument is missing initial 0");
4733 LOP(OP_UNSHIFT,XTERM);
4736 if (PL_expect != XSTATE)
4737 yyerror("\"use\" not allowed in expression");
4740 s = force_version(s);
4741 if(*s == ';' || (s = skipspace(s), *s == ';')) {
4742 PL_nextval[PL_nexttoke].opval = Nullop;
4747 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4748 s = force_version(s);
4760 yylval.ival = PL_curcop->cop_line;
4764 PL_hints |= HINT_BLOCK_SCOPE;
4771 LOP(OP_WAITPID,XTERM);
4779 static char ctl_l[2];
4781 if (ctl_l[0] == '\0')
4782 ctl_l[0] = toCTRL('L');
4783 gv_fetchpv(ctl_l,TRUE, SVt_PV);
4786 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
4791 if (PL_expect == XOPERATOR)
4797 yylval.ival = OP_XOR;
4802 TERM(sublex_start());
4808 Perl_keyword(pTHX_ register char *d, I32 len)
4813 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
4814 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
4815 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
4816 if (strEQ(d,"__DATA__")) return KEY___DATA__;
4817 if (strEQ(d,"__END__")) return KEY___END__;
4821 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
4826 if (strEQ(d,"and")) return -KEY_and;
4827 if (strEQ(d,"abs")) return -KEY_abs;
4830 if (strEQ(d,"alarm")) return -KEY_alarm;
4831 if (strEQ(d,"atan2")) return -KEY_atan2;
4834 if (strEQ(d,"accept")) return -KEY_accept;
4839 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
4842 if (strEQ(d,"bless")) return -KEY_bless;
4843 if (strEQ(d,"bind")) return -KEY_bind;
4844 if (strEQ(d,"binmode")) return -KEY_binmode;
4847 if (strEQ(d,"CORE")) return -KEY_CORE;
4852 if (strEQ(d,"cmp")) return -KEY_cmp;
4853 if (strEQ(d,"chr")) return -KEY_chr;
4854 if (strEQ(d,"cos")) return -KEY_cos;
4857 if (strEQ(d,"chop")) return KEY_chop;
4860 if (strEQ(d,"close")) return -KEY_close;
4861 if (strEQ(d,"chdir")) return -KEY_chdir;
4862 if (strEQ(d,"chomp")) return KEY_chomp;
4863 if (strEQ(d,"chmod")) return -KEY_chmod;
4864 if (strEQ(d,"chown")) return -KEY_chown;
4865 if (strEQ(d,"crypt")) return -KEY_crypt;
4868 if (strEQ(d,"chroot")) return -KEY_chroot;
4869 if (strEQ(d,"caller")) return -KEY_caller;
4872 if (strEQ(d,"connect")) return -KEY_connect;
4875 if (strEQ(d,"closedir")) return -KEY_closedir;
4876 if (strEQ(d,"continue")) return -KEY_continue;
4881 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
4886 if (strEQ(d,"do")) return KEY_do;
4889 if (strEQ(d,"die")) return -KEY_die;
4892 if (strEQ(d,"dump")) return -KEY_dump;
4895 if (strEQ(d,"delete")) return KEY_delete;
4898 if (strEQ(d,"defined")) return KEY_defined;
4899 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
4902 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
4907 if (strEQ(d,"EQ")) { deprecate(d); return -KEY_eq;}
4908 if (strEQ(d,"END")) return KEY_END;
4913 if (strEQ(d,"eq")) return -KEY_eq;
4916 if (strEQ(d,"eof")) return -KEY_eof;
4917 if (strEQ(d,"exp")) return -KEY_exp;
4920 if (strEQ(d,"else")) return KEY_else;
4921 if (strEQ(d,"exit")) return -KEY_exit;
4922 if (strEQ(d,"eval")) return KEY_eval;
4923 if (strEQ(d,"exec")) return -KEY_exec;
4924 if (strEQ(d,"each")) return KEY_each;
4927 if (strEQ(d,"elsif")) return KEY_elsif;
4930 if (strEQ(d,"exists")) return KEY_exists;
4931 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
4934 if (strEQ(d,"endgrent")) return -KEY_endgrent;
4935 if (strEQ(d,"endpwent")) return -KEY_endpwent;
4938 if (strEQ(d,"endnetent")) return -KEY_endnetent;
4941 if (strEQ(d,"endhostent")) return -KEY_endhostent;
4942 if (strEQ(d,"endservent")) return -KEY_endservent;
4945 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
4952 if (strEQ(d,"for")) return KEY_for;
4955 if (strEQ(d,"fork")) return -KEY_fork;
4958 if (strEQ(d,"fcntl")) return -KEY_fcntl;
4959 if (strEQ(d,"flock")) return -KEY_flock;
4962 if (strEQ(d,"format")) return KEY_format;
4963 if (strEQ(d,"fileno")) return -KEY_fileno;
4966 if (strEQ(d,"foreach")) return KEY_foreach;
4969 if (strEQ(d,"formline")) return -KEY_formline;
4975 if (strEQ(d,"GT")) { deprecate(d); return -KEY_gt;}
4976 if (strEQ(d,"GE")) { deprecate(d); return -KEY_ge;}
4980 if (strnEQ(d,"get",3)) {
4985 if (strEQ(d,"ppid")) return -KEY_getppid;
4986 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
4989 if (strEQ(d,"pwent")) return -KEY_getpwent;
4990 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
4991 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
4994 if (strEQ(d,"peername")) return -KEY_getpeername;
4995 if (strEQ(d,"protoent")) return -KEY_getprotoent;
4996 if (strEQ(d,"priority")) return -KEY_getpriority;
4999 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5002 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5006 else if (*d == 'h') {
5007 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5008 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5009 if (strEQ(d,"hostent")) return -KEY_gethostent;
5011 else if (*d == 'n') {
5012 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5013 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5014 if (strEQ(d,"netent")) return -KEY_getnetent;
5016 else if (*d == 's') {
5017 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5018 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5019 if (strEQ(d,"servent")) return -KEY_getservent;
5020 if (strEQ(d,"sockname")) return -KEY_getsockname;
5021 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5023 else if (*d == 'g') {
5024 if (strEQ(d,"grent")) return -KEY_getgrent;
5025 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5026 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5028 else if (*d == 'l') {
5029 if (strEQ(d,"login")) return -KEY_getlogin;
5031 else if (strEQ(d,"c")) return -KEY_getc;
5036 if (strEQ(d,"gt")) return -KEY_gt;
5037 if (strEQ(d,"ge")) return -KEY_ge;
5040 if (strEQ(d,"grep")) return KEY_grep;
5041 if (strEQ(d,"goto")) return KEY_goto;
5042 if (strEQ(d,"glob")) return KEY_glob;
5045 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5050 if (strEQ(d,"hex")) return -KEY_hex;
5053 if (strEQ(d,"INIT")) return KEY_INIT;
5058 if (strEQ(d,"if")) return KEY_if;
5061 if (strEQ(d,"int")) return -KEY_int;
5064 if (strEQ(d,"index")) return -KEY_index;
5065 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5070 if (strEQ(d,"join")) return -KEY_join;
5074 if (strEQ(d,"keys")) return KEY_keys;
5075 if (strEQ(d,"kill")) return -KEY_kill;
5080 if (strEQ(d,"LT")) { deprecate(d); return -KEY_lt;}
5081 if (strEQ(d,"LE")) { deprecate(d); return -KEY_le;}
5087 if (strEQ(d,"lt")) return -KEY_lt;
5088 if (strEQ(d,"le")) return -KEY_le;
5089 if (strEQ(d,"lc")) return -KEY_lc;
5092 if (strEQ(d,"log")) return -KEY_log;
5095 if (strEQ(d,"last")) return KEY_last;
5096 if (strEQ(d,"link")) return -KEY_link;
5097 if (strEQ(d,"lock")) return -KEY_lock;
5100 if (strEQ(d,"local")) return KEY_local;
5101 if (strEQ(d,"lstat")) return -KEY_lstat;
5104 if (strEQ(d,"length")) return -KEY_length;
5105 if (strEQ(d,"listen")) return -KEY_listen;
5108 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5111 if (strEQ(d,"localtime")) return -KEY_localtime;
5117 case 1: return KEY_m;
5119 if (strEQ(d,"my")) return KEY_my;
5122 if (strEQ(d,"map")) return KEY_map;
5125 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5128 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5129 if (strEQ(d,"msgget")) return -KEY_msgget;
5130 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5131 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5136 if (strEQ(d,"NE")) { deprecate(d); return -KEY_ne;}
5139 if (strEQ(d,"next")) return KEY_next;
5140 if (strEQ(d,"ne")) return -KEY_ne;
5141 if (strEQ(d,"not")) return -KEY_not;
5142 if (strEQ(d,"no")) return KEY_no;
5147 if (strEQ(d,"or")) return -KEY_or;
5150 if (strEQ(d,"ord")) return -KEY_ord;
5151 if (strEQ(d,"oct")) return -KEY_oct;
5152 if (strEQ(d,"our")) return KEY_our;
5155 if (strEQ(d,"open")) return -KEY_open;
5158 if (strEQ(d,"opendir")) return -KEY_opendir;
5165 if (strEQ(d,"pop")) return KEY_pop;
5166 if (strEQ(d,"pos")) return KEY_pos;
5169 if (strEQ(d,"push")) return KEY_push;
5170 if (strEQ(d,"pack")) return -KEY_pack;
5171 if (strEQ(d,"pipe")) return -KEY_pipe;
5174 if (strEQ(d,"print")) return KEY_print;
5177 if (strEQ(d,"printf")) return KEY_printf;
5180 if (strEQ(d,"package")) return KEY_package;
5183 if (strEQ(d,"prototype")) return KEY_prototype;
5188 if (strEQ(d,"q")) return KEY_q;
5189 if (strEQ(d,"qr")) return KEY_qr;
5190 if (strEQ(d,"qq")) return KEY_qq;
5191 if (strEQ(d,"qw")) return KEY_qw;
5192 if (strEQ(d,"qx")) return KEY_qx;
5194 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5199 if (strEQ(d,"ref")) return -KEY_ref;
5202 if (strEQ(d,"read")) return -KEY_read;
5203 if (strEQ(d,"rand")) return -KEY_rand;
5204 if (strEQ(d,"recv")) return -KEY_recv;
5205 if (strEQ(d,"redo")) return KEY_redo;
5208 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5209 if (strEQ(d,"reset")) return -KEY_reset;
5212 if (strEQ(d,"return")) return KEY_return;
5213 if (strEQ(d,"rename")) return -KEY_rename;
5214 if (strEQ(d,"rindex")) return -KEY_rindex;
5217 if (strEQ(d,"require")) return -KEY_require;
5218 if (strEQ(d,"reverse")) return -KEY_reverse;
5219 if (strEQ(d,"readdir")) return -KEY_readdir;
5222 if (strEQ(d,"readlink")) return -KEY_readlink;
5223 if (strEQ(d,"readline")) return -KEY_readline;
5224 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5227 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5232 if (strEQ(d,"STOP")) return KEY_STOP;
5236 case 0: return KEY_s;
5238 if (strEQ(d,"scalar")) return KEY_scalar;
5243 if (strEQ(d,"seek")) return -KEY_seek;
5244 if (strEQ(d,"send")) return -KEY_send;
5247 if (strEQ(d,"semop")) return -KEY_semop;
5250 if (strEQ(d,"select")) return -KEY_select;
5251 if (strEQ(d,"semctl")) return -KEY_semctl;
5252 if (strEQ(d,"semget")) return -KEY_semget;
5255 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5256 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5259 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5260 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5263 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5266 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5267 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5268 if (strEQ(d,"setservent")) return -KEY_setservent;
5271 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5272 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5279 if (strEQ(d,"shift")) return KEY_shift;
5282 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5283 if (strEQ(d,"shmget")) return -KEY_shmget;
5286 if (strEQ(d,"shmread")) return -KEY_shmread;
5289 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5290 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5295 if (strEQ(d,"sin")) return -KEY_sin;
5298 if (strEQ(d,"sleep")) return -KEY_sleep;
5301 if (strEQ(d,"sort")) return KEY_sort;
5302 if (strEQ(d,"socket")) return -KEY_socket;
5303 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5306 if (strEQ(d,"split")) return KEY_split;
5307 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5308 if (strEQ(d,"splice")) return KEY_splice;
5311 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5314 if (strEQ(d,"srand")) return -KEY_srand;
5317 if (strEQ(d,"stat")) return -KEY_stat;
5318 if (strEQ(d,"study")) return KEY_study;
5321 if (strEQ(d,"substr")) return -KEY_substr;
5322 if (strEQ(d,"sub")) return KEY_sub;
5327 if (strEQ(d,"system")) return -KEY_system;
5330 if (strEQ(d,"symlink")) return -KEY_symlink;
5331 if (strEQ(d,"syscall")) return -KEY_syscall;
5332 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5333 if (strEQ(d,"sysread")) return -KEY_sysread;
5334 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5337 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5346 if (strEQ(d,"tr")) return KEY_tr;
5349 if (strEQ(d,"tie")) return KEY_tie;
5352 if (strEQ(d,"tell")) return -KEY_tell;
5353 if (strEQ(d,"tied")) return KEY_tied;
5354 if (strEQ(d,"time")) return -KEY_time;
5357 if (strEQ(d,"times")) return -KEY_times;
5360 if (strEQ(d,"telldir")) return -KEY_telldir;
5363 if (strEQ(d,"truncate")) return -KEY_truncate;
5370 if (strEQ(d,"uc")) return -KEY_uc;
5373 if (strEQ(d,"use")) return KEY_use;
5376 if (strEQ(d,"undef")) return KEY_undef;
5377 if (strEQ(d,"until")) return KEY_until;
5378 if (strEQ(d,"untie")) return KEY_untie;
5379 if (strEQ(d,"utime")) return -KEY_utime;
5380 if (strEQ(d,"umask")) return -KEY_umask;
5383 if (strEQ(d,"unless")) return KEY_unless;
5384 if (strEQ(d,"unpack")) return -KEY_unpack;
5385 if (strEQ(d,"unlink")) return -KEY_unlink;
5388 if (strEQ(d,"unshift")) return KEY_unshift;
5389 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5394 if (strEQ(d,"values")) return -KEY_values;
5395 if (strEQ(d,"vec")) return -KEY_vec;
5400 if (strEQ(d,"warn")) return -KEY_warn;
5401 if (strEQ(d,"wait")) return -KEY_wait;
5404 if (strEQ(d,"while")) return KEY_while;
5405 if (strEQ(d,"write")) return -KEY_write;
5408 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5411 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5416 if (len == 1) return -KEY_x;
5417 if (strEQ(d,"xor")) return -KEY_xor;
5420 if (len == 1) return KEY_y;
5429 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5433 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5434 dTHR; /* only for ckWARN */
5435 if (ckWARN(WARN_SYNTAX)) {
5437 for (w = s+2; *w && level; w++) {
5444 for (; *w && isSPACE(*w); w++) ;
5445 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5446 Perl_warner(aTHX_ WARN_SYNTAX,
5447 "%s (...) interpreted as function",name);
5450 while (s < PL_bufend && isSPACE(*s))
5454 while (s < PL_bufend && isSPACE(*s))
5456 if (isIDFIRST_lazy(s)) {
5458 while (isALNUM_lazy(s))
5460 while (s < PL_bufend && isSPACE(*s))
5465 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5469 Perl_croak(aTHX_ "No comma allowed after %s", what);
5474 /* Either returns sv, or mortalizes sv and returns a new SV*.
5475 Best used as sv=new_constant(..., sv, ...).
5476 If s, pv are NULL, calls subroutine with one argument,
5477 and type is used with error messages only. */
5480 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5484 HV *table = GvHV(PL_hintgv); /* ^H */
5488 const char *why, *why1, *why2;
5490 if (!(PL_hints & HINT_LOCALIZE_HH)) {
5493 why = "%^H is not localized";
5497 msg = Perl_newSVpvf(aTHX_ "constant(%s): %s%s%s",
5498 (type ? type: "undef"), why1, why2, why);
5499 yyerror(SvPVX(msg));
5504 why = "%^H is not defined";
5507 cvp = hv_fetch(table, key, strlen(key), FALSE);
5508 if (!cvp || !SvOK(*cvp)) {
5509 why = "} is not defined";
5514 sv_2mortal(sv); /* Parent created it permanently */
5517 pv = sv_2mortal(newSVpvn(s, len));
5519 typesv = sv_2mortal(newSVpv(type, 0));
5521 typesv = &PL_sv_undef;
5523 PUSHSTACKi(PERLSI_OVERLOAD);
5536 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5540 /* Check the eval first */
5541 if (!PL_in_eval && SvTRUE(ERRSV))
5544 sv_catpv(ERRSV, "Propagated");
5545 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5547 res = SvREFCNT_inc(sv);
5551 (void)SvREFCNT_inc(res);
5560 why = "}} did not return a defined value";
5561 why1 = "Call to &{$^H{";
5571 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
5573 register char *d = dest;
5574 register char *e = d + destlen - 3; /* two-character token, ending NUL */
5577 Perl_croak(aTHX_ ident_too_long);
5578 if (isALNUM(*s)) /* UTF handled below */
5580 else if (*s == '\'' && allow_package && isIDFIRST_lazy(s+1)) {
5585 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
5589 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5590 char *t = s + UTF8SKIP(s);
5591 while (*t & 0x80 && is_utf8_mark((U8*)t))
5593 if (d + (t - s) > e)
5594 Perl_croak(aTHX_ ident_too_long);
5595 Copy(s, d, t - s, char);
5608 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
5615 if (PL_lex_brackets == 0)
5616 PL_lex_fakebrack = 0;
5620 e = d + destlen - 3; /* two-character token, ending NUL */
5622 while (isDIGIT(*s)) {
5624 Perl_croak(aTHX_ ident_too_long);
5631 Perl_croak(aTHX_ ident_too_long);
5632 if (isALNUM(*s)) /* UTF handled below */
5634 else if (*s == '\'' && isIDFIRST_lazy(s+1)) {
5639 else if (*s == ':' && s[1] == ':') {
5643 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5644 char *t = s + UTF8SKIP(s);
5645 while (*t & 0x80 && is_utf8_mark((U8*)t))
5647 if (d + (t - s) > e)
5648 Perl_croak(aTHX_ ident_too_long);
5649 Copy(s, d, t - s, char);
5660 if (PL_lex_state != LEX_NORMAL)
5661 PL_lex_state = LEX_INTERPENDMAYBE;
5664 if (*s == '$' && s[1] &&
5665 (isALNUM_lazy(s+1) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
5678 if (*d == '^' && *s && isCONTROLVAR(*s)) {
5683 if (isSPACE(s[-1])) {
5686 if (!SPACE_OR_TAB(ch)) {
5692 if (isIDFIRST_lazy(d)) {
5696 while (e < send && isALNUM_lazy(e) || *e == ':') {
5698 while (e < send && *e & 0x80 && is_utf8_mark((U8*)e))
5701 Copy(s, d, e - s, char);
5706 while ((isALNUM(*s) || *s == ':') && d < e)
5709 Perl_croak(aTHX_ ident_too_long);
5712 while (s < send && SPACE_OR_TAB(*s)) s++;
5713 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
5714 dTHR; /* only for ckWARN */
5715 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
5716 const char *brack = *s == '[' ? "[...]" : "{...}";
5717 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5718 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
5719 funny, dest, brack, funny, dest, brack);
5721 PL_lex_fakebrack = PL_lex_brackets+1;
5723 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
5727 /* Handle extended ${^Foo} variables
5728 * 1999-02-27 mjd-perl-patch@plover.com */
5729 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
5733 while (isALNUM(*s) && d < e) {
5737 Perl_croak(aTHX_ ident_too_long);
5742 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
5743 PL_lex_state = LEX_INTERPEND;
5746 if (PL_lex_state == LEX_NORMAL) {
5747 dTHR; /* only for ckWARN */
5748 if (ckWARN(WARN_AMBIGUOUS) &&
5749 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
5751 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5752 "Ambiguous use of %c{%s} resolved to %c%s",
5753 funny, dest, funny, dest);
5758 s = bracket; /* let the parser handle it */
5762 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
5763 PL_lex_state = LEX_INTERPEND;
5768 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
5773 *pmfl |= PMf_GLOBAL;
5775 *pmfl |= PMf_CONTINUE;
5779 *pmfl |= PMf_MULTILINE;
5781 *pmfl |= PMf_SINGLELINE;
5783 *pmfl |= PMf_EXTENDED;
5787 S_scan_pat(pTHX_ char *start, I32 type)
5792 s = scan_str(start,FALSE,FALSE);
5795 SvREFCNT_dec(PL_lex_stuff);
5796 PL_lex_stuff = Nullsv;
5797 Perl_croak(aTHX_ "Search pattern not terminated");
5800 pm = (PMOP*)newPMOP(type, 0);
5801 if (PL_multi_open == '?')
5802 pm->op_pmflags |= PMf_ONCE;
5804 while (*s && strchr("iomsx", *s))
5805 pmflag(&pm->op_pmflags,*s++);
5808 while (*s && strchr("iogcmsx", *s))
5809 pmflag(&pm->op_pmflags,*s++);
5811 pm->op_pmpermflags = pm->op_pmflags;
5813 PL_lex_op = (OP*)pm;
5814 yylval.ival = OP_MATCH;
5819 S_scan_subst(pTHX_ char *start)
5826 yylval.ival = OP_NULL;
5828 s = scan_str(start,FALSE,FALSE);
5832 SvREFCNT_dec(PL_lex_stuff);
5833 PL_lex_stuff = Nullsv;
5834 Perl_croak(aTHX_ "Substitution pattern not terminated");
5837 if (s[-1] == PL_multi_open)
5840 first_start = PL_multi_start;
5841 s = scan_str(s,FALSE,FALSE);
5844 SvREFCNT_dec(PL_lex_stuff);
5845 PL_lex_stuff = Nullsv;
5847 SvREFCNT_dec(PL_lex_repl);
5848 PL_lex_repl = Nullsv;
5849 Perl_croak(aTHX_ "Substitution replacement not terminated");
5851 PL_multi_start = first_start; /* so whole substitution is taken together */
5853 pm = (PMOP*)newPMOP(OP_SUBST, 0);
5859 else if (strchr("iogcmsx", *s))
5860 pmflag(&pm->op_pmflags,*s++);
5867 PL_sublex_info.super_bufptr = s;
5868 PL_sublex_info.super_bufend = PL_bufend;
5870 pm->op_pmflags |= PMf_EVAL;
5871 repl = newSVpvn("",0);
5873 sv_catpv(repl, es ? "eval " : "do ");
5874 sv_catpvn(repl, "{ ", 2);
5875 sv_catsv(repl, PL_lex_repl);
5876 sv_catpvn(repl, " };", 2);
5878 SvREFCNT_dec(PL_lex_repl);
5882 pm->op_pmpermflags = pm->op_pmflags;
5883 PL_lex_op = (OP*)pm;
5884 yylval.ival = OP_SUBST;
5889 S_scan_trans(pTHX_ char *start)
5900 yylval.ival = OP_NULL;
5902 s = scan_str(start,FALSE,FALSE);
5905 SvREFCNT_dec(PL_lex_stuff);
5906 PL_lex_stuff = Nullsv;
5907 Perl_croak(aTHX_ "Transliteration pattern not terminated");
5909 if (s[-1] == PL_multi_open)
5912 s = scan_str(s,FALSE,FALSE);
5915 SvREFCNT_dec(PL_lex_stuff);
5916 PL_lex_stuff = Nullsv;
5918 SvREFCNT_dec(PL_lex_repl);
5919 PL_lex_repl = Nullsv;
5920 Perl_croak(aTHX_ "Transliteration replacement not terminated");
5924 o = newSVOP(OP_TRANS, 0, 0);
5925 utf8 = OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF;
5928 New(803,tbl,256,short);
5929 o = newPVOP(OP_TRANS, 0, (char*)tbl);
5933 complement = del = squash = 0;
5934 while (strchr("cdsCU", *s)) {
5936 complement = OPpTRANS_COMPLEMENT;
5938 del = OPpTRANS_DELETE;
5940 squash = OPpTRANS_SQUASH;
5945 utf8 &= ~OPpTRANS_FROM_UTF;
5947 utf8 |= OPpTRANS_FROM_UTF;
5951 utf8 &= ~OPpTRANS_TO_UTF;
5953 utf8 |= OPpTRANS_TO_UTF;
5956 Perl_croak(aTHX_ "Too many /C and /U options");
5961 o->op_private = del|squash|complement|utf8;
5964 yylval.ival = OP_TRANS;
5969 S_scan_heredoc(pTHX_ register char *s)
5973 I32 op_type = OP_SCALAR;
5980 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
5984 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
5987 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
5988 if (*peek && strchr("`'\"",*peek)) {
5991 s = delimcpy(d, e, s, PL_bufend, term, &len);
6001 if (!isALNUM_lazy(s))
6002 deprecate("bare << to mean <<\"\"");
6003 for (; isALNUM_lazy(s); s++) {
6008 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6009 Perl_croak(aTHX_ "Delimiter for here document is too long");
6012 len = d - PL_tokenbuf;
6013 #ifndef PERL_STRICT_CR
6014 d = strchr(s, '\r');
6018 while (s < PL_bufend) {
6024 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6033 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6038 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6039 herewas = newSVpvn(s,PL_bufend-s);
6041 s--, herewas = newSVpvn(s,d-s);
6042 s += SvCUR(herewas);
6044 tmpstr = NEWSV(87,79);
6045 sv_upgrade(tmpstr, SVt_PVIV);
6050 else if (term == '`') {
6051 op_type = OP_BACKTICK;
6052 SvIVX(tmpstr) = '\\';
6056 PL_multi_start = PL_curcop->cop_line;
6057 PL_multi_open = PL_multi_close = '<';
6058 term = *PL_tokenbuf;
6059 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6060 char *bufptr = PL_sublex_info.super_bufptr;
6061 char *bufend = PL_sublex_info.super_bufend;
6062 char *olds = s - SvCUR(herewas);
6063 s = strchr(bufptr, '\n');
6067 while (s < bufend &&
6068 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6070 PL_curcop->cop_line++;
6073 PL_curcop->cop_line = PL_multi_start;
6074 missingterm(PL_tokenbuf);
6076 sv_setpvn(herewas,bufptr,d-bufptr+1);
6077 sv_setpvn(tmpstr,d+1,s-d);
6079 sv_catpvn(herewas,s,bufend-s);
6080 (void)strcpy(bufptr,SvPVX(herewas));
6087 while (s < PL_bufend &&
6088 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6090 PL_curcop->cop_line++;
6092 if (s >= PL_bufend) {
6093 PL_curcop->cop_line = PL_multi_start;
6094 missingterm(PL_tokenbuf);
6096 sv_setpvn(tmpstr,d+1,s-d);
6098 PL_curcop->cop_line++; /* the preceding stmt passes a newline */
6100 sv_catpvn(herewas,s,PL_bufend-s);
6101 sv_setsv(PL_linestr,herewas);
6102 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6103 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6106 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6107 while (s >= PL_bufend) { /* multiple line string? */
6109 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6110 PL_curcop->cop_line = PL_multi_start;
6111 missingterm(PL_tokenbuf);
6113 PL_curcop->cop_line++;
6114 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6115 #ifndef PERL_STRICT_CR
6116 if (PL_bufend - PL_linestart >= 2) {
6117 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6118 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6120 PL_bufend[-2] = '\n';
6122 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6124 else if (PL_bufend[-1] == '\r')
6125 PL_bufend[-1] = '\n';
6127 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6128 PL_bufend[-1] = '\n';
6130 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6131 SV *sv = NEWSV(88,0);
6133 sv_upgrade(sv, SVt_PVMG);
6134 sv_setsv(sv,PL_linestr);
6135 av_store(CopFILEAV(PL_curcop), (I32)PL_curcop->cop_line,sv);
6137 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6140 sv_catsv(PL_linestr,herewas);
6141 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6145 sv_catsv(tmpstr,PL_linestr);
6150 PL_multi_end = PL_curcop->cop_line;
6151 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6152 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6153 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6155 SvREFCNT_dec(herewas);
6156 PL_lex_stuff = tmpstr;
6157 yylval.ival = op_type;
6162 takes: current position in input buffer
6163 returns: new position in input buffer
6164 side-effects: yylval and lex_op are set.
6169 <FH> read from filehandle
6170 <pkg::FH> read from package qualified filehandle
6171 <pkg'FH> read from package qualified filehandle
6172 <$fh> read from filehandle in $fh
6178 S_scan_inputsymbol(pTHX_ char *start)
6180 register char *s = start; /* current position in buffer */
6186 d = PL_tokenbuf; /* start of temp holding space */
6187 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6188 end = strchr(s, '\n');
6191 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6193 /* die if we didn't have space for the contents of the <>,
6194 or if it didn't end, or if we see a newline
6197 if (len >= sizeof PL_tokenbuf)
6198 Perl_croak(aTHX_ "Excessively long <> operator");
6200 Perl_croak(aTHX_ "Unterminated <> operator");
6205 Remember, only scalar variables are interpreted as filehandles by
6206 this code. Anything more complex (e.g., <$fh{$num}>) will be
6207 treated as a glob() call.
6208 This code makes use of the fact that except for the $ at the front,
6209 a scalar variable and a filehandle look the same.
6211 if (*d == '$' && d[1]) d++;
6213 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6214 while (*d && (isALNUM_lazy(d) || *d == '\'' || *d == ':'))
6217 /* If we've tried to read what we allow filehandles to look like, and
6218 there's still text left, then it must be a glob() and not a getline.
6219 Use scan_str to pull out the stuff between the <> and treat it
6220 as nothing more than a string.
6223 if (d - PL_tokenbuf != len) {
6224 yylval.ival = OP_GLOB;
6226 s = scan_str(start,FALSE,FALSE);
6228 Perl_croak(aTHX_ "Glob not terminated");
6232 /* we're in a filehandle read situation */
6235 /* turn <> into <ARGV> */
6237 (void)strcpy(d,"ARGV");
6239 /* if <$fh>, create the ops to turn the variable into a
6245 /* try to find it in the pad for this block, otherwise find
6246 add symbol table ops
6248 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6249 OP *o = newOP(OP_PADSV, 0);
6251 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o);
6254 GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV);
6255 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0,
6256 newUNOP(OP_RV2SV, 0,
6257 newGVOP(OP_GV, 0, gv)));
6259 PL_lex_op->op_flags |= OPf_SPECIAL;
6260 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6261 yylval.ival = OP_NULL;
6264 /* If it's none of the above, it must be a literal filehandle
6265 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6267 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6268 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6269 yylval.ival = OP_NULL;
6278 takes: start position in buffer
6279 keep_quoted preserve \ on the embedded delimiter(s)
6280 keep_delims preserve the delimiters around the string
6281 returns: position to continue reading from buffer
6282 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6283 updates the read buffer.
6285 This subroutine pulls a string out of the input. It is called for:
6286 q single quotes q(literal text)
6287 ' single quotes 'literal text'
6288 qq double quotes qq(interpolate $here please)
6289 " double quotes "interpolate $here please"
6290 qx backticks qx(/bin/ls -l)
6291 ` backticks `/bin/ls -l`
6292 qw quote words @EXPORT_OK = qw( func() $spam )
6293 m// regexp match m/this/
6294 s/// regexp substitute s/this/that/
6295 tr/// string transliterate tr/this/that/
6296 y/// string transliterate y/this/that/
6297 ($*@) sub prototypes sub foo ($)
6298 (stuff) sub attr parameters sub foo : attr(stuff)
6299 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6301 In most of these cases (all but <>, patterns and transliterate)
6302 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6303 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6304 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6307 It skips whitespace before the string starts, and treats the first
6308 character as the delimiter. If the delimiter is one of ([{< then
6309 the corresponding "close" character )]}> is used as the closing
6310 delimiter. It allows quoting of delimiters, and if the string has
6311 balanced delimiters ([{<>}]) it allows nesting.
6313 The lexer always reads these strings into lex_stuff, except in the
6314 case of the operators which take *two* arguments (s/// and tr///)
6315 when it checks to see if lex_stuff is full (presumably with the 1st
6316 arg to s or tr) and if so puts the string into lex_repl.
6321 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6324 SV *sv; /* scalar value: string */
6325 char *tmps; /* temp string, used for delimiter matching */
6326 register char *s = start; /* current position in the buffer */
6327 register char term; /* terminating character */
6328 register char *to; /* current position in the sv's data */
6329 I32 brackets = 1; /* bracket nesting level */
6331 /* skip space before the delimiter */
6335 /* mark where we are, in case we need to report errors */
6338 /* after skipping whitespace, the next character is the terminator */
6340 /* mark where we are */
6341 PL_multi_start = PL_curcop->cop_line;
6342 PL_multi_open = term;
6344 /* find corresponding closing delimiter */
6345 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6347 PL_multi_close = term;
6349 /* create a new SV to hold the contents. 87 is leak category, I'm
6350 assuming. 79 is the SV's initial length. What a random number. */
6352 sv_upgrade(sv, SVt_PVIV);
6354 (void)SvPOK_only(sv); /* validate pointer */
6356 /* move past delimiter and try to read a complete string */
6358 sv_catpvn(sv, s, 1);
6361 /* extend sv if need be */
6362 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6363 /* set 'to' to the next character in the sv's string */
6364 to = SvPVX(sv)+SvCUR(sv);
6366 /* if open delimiter is the close delimiter read unbridle */
6367 if (PL_multi_open == PL_multi_close) {
6368 for (; s < PL_bufend; s++,to++) {
6369 /* embedded newlines increment the current line number */
6370 if (*s == '\n' && !PL_rsfp)
6371 PL_curcop->cop_line++;
6372 /* handle quoted delimiters */
6373 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6374 if (!keep_quoted && s[1] == term)
6376 /* any other quotes are simply copied straight through */
6380 /* terminate when run out of buffer (the for() condition), or
6381 have found the terminator */
6382 else if (*s == term)
6388 /* if the terminator isn't the same as the start character (e.g.,
6389 matched brackets), we have to allow more in the quoting, and
6390 be prepared for nested brackets.
6393 /* read until we run out of string, or we find the terminator */
6394 for (; s < PL_bufend; s++,to++) {
6395 /* embedded newlines increment the line count */
6396 if (*s == '\n' && !PL_rsfp)
6397 PL_curcop->cop_line++;
6398 /* backslashes can escape the open or closing characters */
6399 if (*s == '\\' && s+1 < PL_bufend) {
6401 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6406 /* allow nested opens and closes */
6407 else if (*s == PL_multi_close && --brackets <= 0)
6409 else if (*s == PL_multi_open)
6414 /* terminate the copied string and update the sv's end-of-string */
6416 SvCUR_set(sv, to - SvPVX(sv));
6419 * this next chunk reads more into the buffer if we're not done yet
6422 if (s < PL_bufend) break; /* handle case where we are done yet :-) */
6424 #ifndef PERL_STRICT_CR
6425 if (to - SvPVX(sv) >= 2) {
6426 if ((to[-2] == '\r' && to[-1] == '\n') ||
6427 (to[-2] == '\n' && to[-1] == '\r'))
6431 SvCUR_set(sv, to - SvPVX(sv));
6433 else if (to[-1] == '\r')
6436 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6440 /* if we're out of file, or a read fails, bail and reset the current
6441 line marker so we can report where the unterminated string began
6444 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6446 PL_curcop->cop_line = PL_multi_start;
6449 /* we read a line, so increment our line counter */
6450 PL_curcop->cop_line++;
6452 /* update debugger info */
6453 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6454 SV *sv = NEWSV(88,0);
6456 sv_upgrade(sv, SVt_PVMG);
6457 sv_setsv(sv,PL_linestr);
6458 av_store(CopFILEAV(PL_curcop), (I32)PL_curcop->cop_line, sv);
6461 /* having changed the buffer, we must update PL_bufend */
6462 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6465 /* at this point, we have successfully read the delimited string */
6468 sv_catpvn(sv, s, 1);
6469 PL_multi_end = PL_curcop->cop_line;
6472 /* if we allocated too much space, give some back */
6473 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6474 SvLEN_set(sv, SvCUR(sv) + 1);
6475 Renew(SvPVX(sv), SvLEN(sv), char);
6478 /* decide whether this is the first or second quoted string we've read
6491 takes: pointer to position in buffer
6492 returns: pointer to new position in buffer
6493 side-effects: builds ops for the constant in yylval.op
6495 Read a number in any of the formats that Perl accepts:
6497 0(x[0-7A-F]+)|([0-7]+)|(b[01])
6498 [\d_]+(\.[\d_]*)?[Ee](\d+)
6500 Underbars (_) are allowed in decimal numbers. If -w is on,
6501 underbars before a decimal point must be at three digit intervals.
6503 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6506 If it reads a number without a decimal point or an exponent, it will
6507 try converting the number to an integer and see if it can do so
6508 without loss of precision.
6512 Perl_scan_num(pTHX_ char *start)
6514 register char *s = start; /* current position in buffer */
6515 register char *d; /* destination in temp buffer */
6516 register char *e; /* end of temp buffer */
6517 IV tryiv; /* used to see if it can be an IV */
6518 NV value; /* number read, as a double */
6519 SV *sv; /* place to put the converted number */
6520 bool floatit; /* boolean: int or float? */
6521 char *lastub = 0; /* position of last underbar */
6522 static char number_too_long[] = "Number too long";
6524 /* We use the first character to decide what type of number this is */
6528 Perl_croak(aTHX_ "panic: scan_num");
6530 /* if it starts with a 0, it could be an octal number, a decimal in
6531 0.13 disguise, or a hexadecimal number, or a binary number.
6536 u holds the "number so far"
6537 shift the power of 2 of the base
6538 (hex == 4, octal == 3, binary == 1)
6539 overflowed was the number more than we can hold?
6541 Shift is used when we add a digit. It also serves as an "are
6542 we in octal/hex/binary?" indicator to disallow hex characters
6549 bool overflowed = FALSE;
6550 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6551 static char* bases[5] = { "", "binary", "", "octal",
6553 static char* Bases[5] = { "", "Binary", "", "Octal",
6555 static char *maxima[5] = { "",
6556 "0b11111111111111111111111111111111",
6560 char *base, *Base, *max;
6566 } else if (s[1] == 'b') {
6570 /* check for a decimal in disguise */
6571 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
6573 /* so it must be octal */
6577 base = bases[shift];
6578 Base = Bases[shift];
6579 max = maxima[shift];
6581 /* read the rest of the number */
6583 /* x is used in the overflow test,
6584 b is the digit we're adding on. */
6589 /* if we don't mention it, we're done */
6598 /* 8 and 9 are not octal */
6601 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
6605 case '2': case '3': case '4':
6606 case '5': case '6': case '7':
6608 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
6612 b = *s++ & 15; /* ASCII digit -> value of digit */
6616 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
6617 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
6618 /* make sure they said 0x */
6623 /* Prepare to put the digit we have onto the end
6624 of the number so far. We check for overflows.
6629 x = u << shift; /* make room for the digit */
6631 if ((x >> shift) != u
6632 && !(PL_hints & HINT_NEW_BINARY)) {
6636 if (ckWARN_d(WARN_OVERFLOW))
6637 Perl_warner(aTHX_ WARN_OVERFLOW,
6638 "Integer overflow in %s number",
6641 u = x | b; /* add the digit to the end */
6644 n *= nvshift[shift];
6645 /* If an NV has not enough bits in its
6646 * mantissa to represent an UV this summing of
6647 * small low-order numbers is a waste of time
6648 * (because the NV cannot preserve the
6649 * low-order bits anyway): we could just
6650 * remember when did we overflow and in the
6651 * end just multiply n by the right
6659 /* if we get here, we had success: make a scalar value from
6666 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
6667 Perl_warner(aTHX_ WARN_PORTABLE,
6668 "%s number > %s non-portable",
6675 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
6676 Perl_warner(aTHX_ WARN_PORTABLE,
6677 "%s number > %s non-portable",
6682 if (PL_hints & HINT_NEW_BINARY)
6683 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
6688 handle decimal numbers.
6689 we're also sent here when we read a 0 as the first digit
6691 case '1': case '2': case '3': case '4': case '5':
6692 case '6': case '7': case '8': case '9': case '.':
6695 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
6698 /* read next group of digits and _ and copy into d */
6699 while (isDIGIT(*s) || *s == '_') {
6700 /* skip underscores, checking for misplaced ones
6704 dTHR; /* only for ckWARN */
6705 if (ckWARN(WARN_SYNTAX) && lastub && s - lastub != 3)
6706 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6710 /* check for end of fixed-length buffer */
6712 Perl_croak(aTHX_ number_too_long);
6713 /* if we're ok, copy the character */
6718 /* final misplaced underbar check */
6719 if (lastub && s - lastub != 3) {
6721 if (ckWARN(WARN_SYNTAX))
6722 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6725 /* read a decimal portion if there is one. avoid
6726 3..5 being interpreted as the number 3. followed
6729 if (*s == '.' && s[1] != '.') {
6733 /* copy, ignoring underbars, until we run out of
6734 digits. Note: no misplaced underbar checks!
6736 for (; isDIGIT(*s) || *s == '_'; s++) {
6737 /* fixed length buffer check */
6739 Perl_croak(aTHX_ number_too_long);
6745 /* read exponent part, if present */
6746 if (*s && strchr("eE",*s) && strchr("+-0123456789",s[1])) {
6750 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
6751 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
6753 /* allow positive or negative exponent */
6754 if (*s == '+' || *s == '-')
6757 /* read digits of exponent (no underbars :-) */
6758 while (isDIGIT(*s)) {
6760 Perl_croak(aTHX_ number_too_long);
6765 /* terminate the string */
6768 /* make an sv from the string */
6771 value = Atof(PL_tokenbuf);
6774 See if we can make do with an integer value without loss of
6775 precision. We use I_V to cast to an int, because some
6776 compilers have issues. Then we try casting it back and see
6777 if it was the same. We only do this if we know we
6778 specifically read an integer.
6780 Note: if floatit is true, then we don't need to do the
6784 if (!floatit && (NV)tryiv == value)
6785 sv_setiv(sv, tryiv);
6787 sv_setnv(sv, value);
6788 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
6789 (PL_hints & HINT_NEW_INTEGER) )
6790 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
6791 (floatit ? "float" : "integer"),
6796 /* make the op for the constant and return */
6798 yylval.opval = newSVOP(OP_CONST, 0, sv);
6804 S_scan_formline(pTHX_ register char *s)
6809 SV *stuff = newSVpvn("",0);
6810 bool needargs = FALSE;
6813 if (*s == '.' || *s == '}') {
6815 #ifdef PERL_STRICT_CR
6816 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
6818 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
6820 if (*t == '\n' || t == PL_bufend)
6823 if (PL_in_eval && !PL_rsfp) {
6824 eol = strchr(s,'\n');
6829 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6831 for (t = s; t < eol; t++) {
6832 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
6834 goto enough; /* ~~ must be first line in formline */
6836 if (*t == '@' || *t == '^')
6839 sv_catpvn(stuff, s, eol-s);
6843 s = filter_gets(PL_linestr, PL_rsfp, 0);
6844 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
6845 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
6848 yyerror("Format not terminated");
6858 PL_lex_state = LEX_NORMAL;
6859 PL_nextval[PL_nexttoke].ival = 0;
6863 PL_lex_state = LEX_FORMLINE;
6864 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
6866 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
6870 SvREFCNT_dec(stuff);
6871 PL_lex_formbrack = 0;
6882 PL_cshlen = strlen(PL_cshname);
6887 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
6890 I32 oldsavestack_ix = PL_savestack_ix;
6891 CV* outsidecv = PL_compcv;
6895 assert(SvTYPE(PL_compcv) == SVt_PVCV);
6897 save_I32(&PL_subline);
6898 save_item(PL_subname);
6900 SAVESPTR(PL_curpad);
6901 SAVESPTR(PL_comppad);
6902 SAVESPTR(PL_comppad_name);
6903 SAVESPTR(PL_compcv);
6904 SAVEI32(PL_comppad_name_fill);
6905 SAVEI32(PL_min_intro_pending);
6906 SAVEI32(PL_max_intro_pending);
6907 SAVEI32(PL_pad_reset_pending);
6909 PL_compcv = (CV*)NEWSV(1104,0);
6910 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
6911 CvFLAGS(PL_compcv) |= flags;
6913 PL_comppad = newAV();
6914 av_push(PL_comppad, Nullsv);
6915 PL_curpad = AvARRAY(PL_comppad);
6916 PL_comppad_name = newAV();
6917 PL_comppad_name_fill = 0;
6918 PL_min_intro_pending = 0;
6920 PL_subline = PL_curcop->cop_line;
6922 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
6923 PL_curpad[0] = (SV*)newAV();
6924 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
6925 #endif /* USE_THREADS */
6927 comppadlist = newAV();
6928 AvREAL_off(comppadlist);
6929 av_store(comppadlist, 0, (SV*)PL_comppad_name);
6930 av_store(comppadlist, 1, (SV*)PL_comppad);
6932 CvPADLIST(PL_compcv) = comppadlist;
6933 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
6935 CvOWNER(PL_compcv) = 0;
6936 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
6937 MUTEX_INIT(CvMUTEXP(PL_compcv));
6938 #endif /* USE_THREADS */
6940 return oldsavestack_ix;
6944 Perl_yywarn(pTHX_ char *s)
6947 PL_in_eval |= EVAL_WARNONLY;
6949 PL_in_eval &= ~EVAL_WARNONLY;
6954 Perl_yyerror(pTHX_ char *s)
6958 char *context = NULL;
6962 if (!yychar || (yychar == ';' && !PL_rsfp))
6964 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
6965 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
6966 while (isSPACE(*PL_oldoldbufptr))
6968 context = PL_oldoldbufptr;
6969 contlen = PL_bufptr - PL_oldoldbufptr;
6971 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
6972 PL_oldbufptr != PL_bufptr) {
6973 while (isSPACE(*PL_oldbufptr))
6975 context = PL_oldbufptr;
6976 contlen = PL_bufptr - PL_oldbufptr;
6978 else if (yychar > 255)
6979 where = "next token ???";
6980 else if ((yychar & 127) == 127) {
6981 if (PL_lex_state == LEX_NORMAL ||
6982 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
6983 where = "at end of line";
6984 else if (PL_lex_inpat)
6985 where = "within pattern";
6987 where = "within string";
6990 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
6992 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
6993 else if (isPRINT_LC(yychar))
6994 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
6996 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
6997 where = SvPVX(where_sv);
6999 msg = sv_2mortal(newSVpv(s, 0));
7000 Perl_sv_catpvf(aTHX_ msg, " at %_ line %"IVdf", ",
7001 CopFILESV(PL_curcop), (IV)PL_curcop->cop_line);
7003 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7005 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7006 if (PL_multi_start < PL_multi_end &&
7007 (U32)(PL_curcop->cop_line - PL_multi_end) <= 1) {
7008 Perl_sv_catpvf(aTHX_ msg,
7009 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7010 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7013 if (PL_in_eval & EVAL_WARNONLY)
7014 Perl_warn(aTHX_ "%_", msg);
7017 if (PL_error_count >= 10)
7018 Perl_croak(aTHX_ "%_ has too many errors.\n", CopFILESV(PL_curcop));
7020 PL_in_my_stash = Nullhv;
7031 * Restore a source filter.
7035 restore_rsfp(pTHXo_ void *f)
7037 PerlIO *fp = (PerlIO*)f;
7039 if (PL_rsfp == PerlIO_stdin())
7040 PerlIO_clearerr(PL_rsfp);
7041 else if (PL_rsfp && (PL_rsfp != fp))
7042 PerlIO_close(PL_rsfp);
7048 * Restores the state of PL_expect when the lexing that begun with a
7049 * start_lex() call has ended.
7053 restore_expect(pTHXo_ void *e)
7055 /* a safe way to store a small integer in a pointer */
7056 PL_expect = (expectation)((char *)e - PL_tokenbuf);
7060 * restore_lex_expect
7061 * Restores the state of PL_lex_expect when the lexing that begun with a
7062 * start_lex() call has ended.
7066 restore_lex_expect(pTHXo_ void *e)
7068 /* a safe way to store a small integer in a pointer */
7069 PL_lex_expect = (expectation)((char *)e - PL_tokenbuf);