3 * Copyright (c) 1991-1999, Larry Wall
5 * You may distribute under the terms of either the GNU General Public
6 * License or the Artistic License, as specified in the README file.
11 * "It all comes from here, the stench and the peril." --Frodo
15 * This file is the lexer for Perl. It's closely linked to the
18 * The main routine is yylex(), which returns the next token.
22 #define PERL_IN_TOKE_C
25 #define yychar PL_yychar
26 #define yylval PL_yylval
28 static char ident_too_long[] = "Identifier too long";
30 static void restore_rsfp(pTHXo_ void *f);
31 static void restore_expect(pTHXo_ void *e);
32 static void restore_lex_expect(pTHXo_ void *e);
34 #define UTF (PL_hints & HINT_UTF8)
36 * Note: we try to be careful never to call the isXXX_utf8() functions
37 * unless we're pretty sure we've seen the beginning of a UTF-8 character
38 * (that is, the two high bits are set). Otherwise we risk loading in the
39 * heavy-duty SWASHINIT and SWASHGET routines unnecessarily.
41 #define isIDFIRST_lazy(p) ((!UTF || (*((U8*)p) < 0xc0)) \
43 : isIDFIRST_utf8((U8*)p))
44 #define isALNUM_lazy(p) ((!UTF || (*((U8*)p) < 0xc0)) \
46 : isALNUM_utf8((U8*)p))
48 /* In variables name $^X, these are the legal values for X.
49 * 1999-02-27 mjd-perl-patch@plover.com */
50 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
52 /* LEX_* are values for PL_lex_state, the state of the lexer.
53 * They are arranged oddly so that the guard on the switch statement
54 * can get by with a single comparison (if the compiler is smart enough).
57 /* #define LEX_NOTPARSING 11 is done in perl.h. */
60 #define LEX_INTERPNORMAL 9
61 #define LEX_INTERPCASEMOD 8
62 #define LEX_INTERPPUSH 7
63 #define LEX_INTERPSTART 6
64 #define LEX_INTERPEND 5
65 #define LEX_INTERPENDMAYBE 4
66 #define LEX_INTERPCONCAT 3
67 #define LEX_INTERPCONST 2
68 #define LEX_FORMLINE 1
69 #define LEX_KNOWNEXT 0
78 /* XXX If this causes problems, set i_unistd=undef in the hint file. */
80 # include <unistd.h> /* Needed for execv() */
89 YYSTYPE* yylval_pointer = NULL;
90 int* yychar_pointer = NULL;
93 # define yylval (*yylval_pointer)
94 # define yychar (*yychar_pointer)
95 # define PERL_YYLEX_PARAM yylval_pointer,yychar_pointer
97 # define yylex() Perl_yylex(aTHX_ yylval_pointer, yychar_pointer)
100 #include "keywords.h"
102 /* CLINE is a macro that ensures PL_copline has a sane value */
107 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
110 * Convenience functions to return different tokens and prime the
111 * lexer for the next token. They all take an argument.
113 * TOKEN : generic token (used for '(', DOLSHARP, etc)
114 * OPERATOR : generic operator
115 * AOPERATOR : assignment operator
116 * PREBLOCK : beginning the block after an if, while, foreach, ...
117 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
118 * PREREF : *EXPR where EXPR is not a simple identifier
119 * TERM : expression term
120 * LOOPX : loop exiting command (goto, last, dump, etc)
121 * FTST : file test operator
122 * FUN0 : zero-argument function
123 * FUN1 : not used, except for not, which isn't a UNIOP
124 * BOop : bitwise or or xor
126 * SHop : shift operator
127 * PWop : power operator
128 * PMop : pattern-matching operator
129 * Aop : addition-level operator
130 * Mop : multiplication-level operator
131 * Eop : equality-testing operator
132 * Rop : relational operator <= != gt
134 * Also see LOP and lop() below.
137 #define TOKEN(retval) return (PL_bufptr = s,(int)retval)
138 #define OPERATOR(retval) return (PL_expect = XTERM,PL_bufptr = s,(int)retval)
139 #define AOPERATOR(retval) return ao((PL_expect = XTERM,PL_bufptr = s,(int)retval))
140 #define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s,(int)retval)
141 #define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s,(int)retval)
142 #define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s,(int)retval)
143 #define TERM(retval) return (CLINE, PL_expect = XOPERATOR,PL_bufptr = s,(int)retval)
144 #define LOOPX(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LOOPEX)
145 #define FTST(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)UNIOP)
146 #define FUN0(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC0)
147 #define FUN1(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC1)
148 #define BOop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITOROP))
149 #define BAop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITANDOP))
150 #define SHop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)SHIFTOP))
151 #define PWop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)POWOP))
152 #define PMop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MATCHOP)
153 #define Aop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)ADDOP))
154 #define Mop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MULOP))
155 #define Eop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)EQOP)
156 #define Rop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)RELOP)
158 /* This bit of chicanery makes a unary function followed by
159 * a parenthesis into a function with one argument, highest precedence.
161 #define UNI(f) return(yylval.ival = f, \
164 PL_last_uni = PL_oldbufptr, \
165 PL_last_lop_op = f, \
166 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
168 #define UNIBRACK(f) return(yylval.ival = f, \
170 PL_last_uni = PL_oldbufptr, \
171 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
173 /* grandfather return to old style */
174 #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
179 * This subroutine detects &&= and ||= and turns an ANDAND or OROR
180 * into an OP_ANDASSIGN or OP_ORASSIGN
184 S_ao(pTHX_ int toketype)
186 if (*PL_bufptr == '=') {
188 if (toketype == ANDAND)
189 yylval.ival = OP_ANDASSIGN;
190 else if (toketype == OROR)
191 yylval.ival = OP_ORASSIGN;
199 * When Perl expects an operator and finds something else, no_op
200 * prints the warning. It always prints "<something> found where
201 * operator expected. It prints "Missing semicolon on previous line?"
202 * if the surprise occurs at the start of the line. "do you need to
203 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
204 * where the compiler doesn't know if foo is a method call or a function.
205 * It prints "Missing operator before end of line" if there's nothing
206 * after the missing operator, or "... before <...>" if there is something
207 * after the missing operator.
211 S_no_op(pTHX_ char *what, char *s)
213 char *oldbp = PL_bufptr;
214 bool is_first = (PL_oldbufptr == PL_linestart);
222 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
224 Perl_warn(aTHX_ "\t(Missing semicolon on previous line?)\n");
225 else if (PL_oldoldbufptr && isIDFIRST_lazy(PL_oldoldbufptr)) {
227 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy(t) || *t == ':'); t++) ;
228 if (t < PL_bufptr && isSPACE(*t))
229 Perl_warn(aTHX_ "\t(Do you need to predeclare %.*s?)\n",
230 t - PL_oldoldbufptr, PL_oldoldbufptr);
233 Perl_warn(aTHX_ "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp);
239 * Complain about missing quote/regexp/heredoc terminator.
240 * If it's called with (char *)NULL then it cauterizes the line buffer.
241 * If we're in a delimited string and the delimiter is a control
242 * character, it's reformatted into a two-char sequence like ^C.
247 S_missingterm(pTHX_ char *s)
252 char *nl = strrchr(s,'\n');
258 iscntrl(PL_multi_close)
260 PL_multi_close < 32 || PL_multi_close == 127
264 tmpbuf[1] = toCTRL(PL_multi_close);
270 *tmpbuf = PL_multi_close;
274 q = strchr(s,'"') ? '\'' : '"';
275 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
283 Perl_deprecate(pTHX_ char *s)
286 if (ckWARN(WARN_DEPRECATED))
287 Perl_warner(aTHX_ WARN_DEPRECATED, "Use of %s is deprecated", s);
292 * Deprecate a comma-less variable list.
298 deprecate("comma-less variable list");
302 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
303 * utf16-to-utf8-reversed.
309 S_win32_textfilter(pTHX_ int idx, SV *sv, int maxlen)
311 I32 count = FILTER_READ(idx+1, sv, maxlen);
312 if (count > 0 && !maxlen)
313 win32_strip_return(sv);
319 S_utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
321 I32 count = FILTER_READ(idx+1, sv, maxlen);
325 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
326 tend = utf16_to_utf8((U16*)SvPVX(sv), tmps, SvCUR(sv));
327 sv_usepvn(sv, (char*)tmps, tend - tmps);
334 S_utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
336 I32 count = FILTER_READ(idx+1, sv, maxlen);
340 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
341 tend = utf16_to_utf8_reversed((U16*)SvPVX(sv), tmps, SvCUR(sv));
342 sv_usepvn(sv, (char*)tmps, tend - tmps);
350 * Initialize variables. Uses the Perl save_stack to save its state (for
351 * recursive calls to the parser).
355 Perl_lex_start(pTHX_ SV *line)
361 SAVEI32(PL_lex_dojoin);
362 SAVEI32(PL_lex_brackets);
363 SAVEI32(PL_lex_fakebrack);
364 SAVEI32(PL_lex_casemods);
365 SAVEI32(PL_lex_starts);
366 SAVEI32(PL_lex_state);
367 SAVESPTR(PL_lex_inpat);
368 SAVEI32(PL_lex_inwhat);
369 SAVECOPLINE(PL_curcop);
372 SAVEPPTR(PL_oldbufptr);
373 SAVEPPTR(PL_oldoldbufptr);
374 SAVEPPTR(PL_linestart);
375 SAVESPTR(PL_linestr);
376 SAVEPPTR(PL_lex_brackstack);
377 SAVEPPTR(PL_lex_casestack);
378 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
379 SAVESPTR(PL_lex_stuff);
380 SAVEI32(PL_lex_defer);
381 SAVEI32(PL_sublex_info.sub_inwhat);
382 SAVESPTR(PL_lex_repl);
383 SAVEDESTRUCTOR_X(restore_expect, PL_tokenbuf + PL_expect); /* encode as pointer */
384 SAVEDESTRUCTOR_X(restore_lex_expect, PL_tokenbuf + PL_expect);
386 PL_lex_state = LEX_NORMAL;
390 PL_lex_fakebrack = 0;
391 New(899, PL_lex_brackstack, 120, char);
392 New(899, PL_lex_casestack, 12, char);
393 SAVEFREEPV(PL_lex_brackstack);
394 SAVEFREEPV(PL_lex_casestack);
396 *PL_lex_casestack = '\0';
399 PL_lex_stuff = Nullsv;
400 PL_lex_repl = Nullsv;
403 PL_sublex_info.sub_inwhat = 0;
405 if (SvREADONLY(PL_linestr))
406 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
407 s = SvPV(PL_linestr, len);
408 if (len && s[len-1] != ';') {
409 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
410 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
411 sv_catpvn(PL_linestr, "\n;", 2);
413 SvTEMP_off(PL_linestr);
414 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
415 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
417 PL_rs = newSVpvn("\n", 1);
423 * Finalizer for lexing operations. Must be called when the parser is
424 * done with the lexer.
430 PL_doextract = FALSE;
435 * This subroutine has nothing to do with tilting, whether at windmills
436 * or pinball tables. Its name is short for "increment line". It
437 * increments the current line number in CopLINE(PL_curcop) and checks
438 * to see whether the line starts with a comment of the form
439 * # line 500 "foo.pm"
440 * If so, it sets the current line number and file to the values in the comment.
444 S_incline(pTHX_ char *s)
452 CopLINE_inc(PL_curcop);
455 while (*s == ' ' || *s == '\t') s++;
456 if (strnEQ(s, "line ", 5)) {
465 while (*s == ' ' || *s == '\t')
467 if (*s == '"' && (t = strchr(s+1, '"')))
471 return; /* false alarm */
472 for (t = s; !isSPACE(*t); t++) ;
477 CopFILE_set(PL_curcop, s);
479 CopFILE_set(PL_curcop, PL_origfilename);
481 CopLINE_set(PL_curcop, atoi(n)-1);
486 * Called to gobble the appropriate amount and type of whitespace.
487 * Skips comments as well.
491 S_skipspace(pTHX_ register char *s)
494 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
495 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
501 SSize_t oldprevlen, oldoldprevlen;
502 SSize_t oldloplen, oldunilen;
503 while (s < PL_bufend && isSPACE(*s)) {
504 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
509 if (s < PL_bufend && *s == '#') {
510 while (s < PL_bufend && *s != '\n')
514 if (PL_in_eval && !PL_rsfp) {
521 /* only continue to recharge the buffer if we're at the end
522 * of the buffer, we're not reading from a source filter, and
523 * we're in normal lexing mode
525 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
526 PL_lex_state == LEX_FORMLINE)
529 /* try to recharge the buffer */
530 if ((s = filter_gets(PL_linestr, PL_rsfp,
531 (prevlen = SvCUR(PL_linestr)))) == Nullch)
533 /* end of file. Add on the -p or -n magic */
534 if (PL_minus_n || PL_minus_p) {
535 sv_setpv(PL_linestr,PL_minus_p ?
536 ";}continue{print or die qq(-p destination: $!\\n)" :
538 sv_catpv(PL_linestr,";}");
539 PL_minus_n = PL_minus_p = 0;
542 sv_setpv(PL_linestr,";");
544 /* reset variables for next time we lex */
545 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
547 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
549 /* Close the filehandle. Could be from -P preprocessor,
550 * STDIN, or a regular file. If we were reading code from
551 * STDIN (because the commandline held no -e or filename)
552 * then we don't close it, we reset it so the code can
553 * read from STDIN too.
556 if (PL_preprocess && !PL_in_eval)
557 (void)PerlProc_pclose(PL_rsfp);
558 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
559 PerlIO_clearerr(PL_rsfp);
561 (void)PerlIO_close(PL_rsfp);
566 /* not at end of file, so we only read another line */
567 /* make corresponding updates to old pointers, for yyerror() */
568 oldprevlen = PL_oldbufptr - PL_bufend;
569 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
571 oldunilen = PL_last_uni - PL_bufend;
573 oldloplen = PL_last_lop - PL_bufend;
574 PL_linestart = PL_bufptr = s + prevlen;
575 PL_bufend = s + SvCUR(PL_linestr);
577 PL_oldbufptr = s + oldprevlen;
578 PL_oldoldbufptr = s + oldoldprevlen;
580 PL_last_uni = s + oldunilen;
582 PL_last_lop = s + oldloplen;
585 /* debugger active and we're not compiling the debugger code,
586 * so store the line into the debugger's array of lines
588 if (PERLDB_LINE && PL_curstash != PL_debstash) {
589 SV *sv = NEWSV(85,0);
591 sv_upgrade(sv, SVt_PVMG);
592 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
593 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
600 * Check the unary operators to ensure there's no ambiguity in how they're
601 * used. An ambiguous piece of code would be:
603 * This doesn't mean rand() + 5. Because rand() is a unary operator,
604 * the +5 is its argument.
614 if (PL_oldoldbufptr != PL_last_uni)
616 while (isSPACE(*PL_last_uni))
618 for (s = PL_last_uni; isALNUM_lazy(s) || *s == '-'; s++) ;
619 if ((t = strchr(s, '(')) && t < PL_bufptr)
621 if (ckWARN_d(WARN_AMBIGUOUS)){
624 Perl_warner(aTHX_ WARN_AMBIGUOUS,
625 "Warning: Use of \"%s\" without parens is ambiguous",
631 /* workaround to replace the UNI() macro with a function. Only the
632 * hints/uts.sh file mentions this. Other comments elsewhere in the
633 * source indicate Microport Unix might need it too.
639 #define UNI(f) return uni(f,s)
642 S_uni(pTHX_ I32 f, char *s)
647 PL_last_uni = PL_oldbufptr;
658 #endif /* CRIPPLED_CC */
661 * LOP : macro to build a list operator. Its behaviour has been replaced
662 * with a subroutine, S_lop() for which LOP is just another name.
665 #define LOP(f,x) return lop(f,x,s)
669 * Build a list operator (or something that might be one). The rules:
670 * - if we have a next token, then it's a list operator [why?]
671 * - if the next thing is an opening paren, then it's a function
672 * - else it's a list operator
676 S_lop(pTHX_ I32 f, expectation x, char *s)
683 PL_last_lop = PL_oldbufptr;
698 * When the lexer realizes it knows the next token (for instance,
699 * it is reordering tokens for the parser) then it can call S_force_next
700 * to know what token to return the next time the lexer is called. Caller
701 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
702 * handles the token correctly.
706 S_force_next(pTHX_ I32 type)
708 PL_nexttype[PL_nexttoke] = type;
710 if (PL_lex_state != LEX_KNOWNEXT) {
711 PL_lex_defer = PL_lex_state;
712 PL_lex_expect = PL_expect;
713 PL_lex_state = LEX_KNOWNEXT;
719 * When the lexer knows the next thing is a word (for instance, it has
720 * just seen -> and it knows that the next char is a word char, then
721 * it calls S_force_word to stick the next word into the PL_next lookahead.
724 * char *start : buffer position (must be within PL_linestr)
725 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
726 * int check_keyword : if true, Perl checks to make sure the word isn't
727 * a keyword (do this if the word is a label, e.g. goto FOO)
728 * int allow_pack : if true, : characters will also be allowed (require,
730 * int allow_initial_tick : used by the "sub" lexer only.
734 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
739 start = skipspace(start);
741 if (isIDFIRST_lazy(s) ||
742 (allow_pack && *s == ':') ||
743 (allow_initial_tick && *s == '\'') )
745 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
746 if (check_keyword && keyword(PL_tokenbuf, len))
748 if (token == METHOD) {
753 PL_expect = XOPERATOR;
756 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0));
757 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
765 * Called when the lexer wants $foo *foo &foo etc, but the program
766 * text only contains the "foo" portion. The first argument is a pointer
767 * to the "foo", and the second argument is the type symbol to prefix.
768 * Forces the next token to be a "WORD".
769 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
773 S_force_ident(pTHX_ register char *s, int kind)
776 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
777 PL_nextval[PL_nexttoke].opval = o;
780 dTHR; /* just for in_eval */
781 o->op_private = OPpCONST_ENTERED;
782 /* XXX see note in pp_entereval() for why we forgo typo
783 warnings if the symbol must be introduced in an eval.
785 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
786 kind == '$' ? SVt_PV :
787 kind == '@' ? SVt_PVAV :
788 kind == '%' ? SVt_PVHV :
797 * Forces the next token to be a version number.
801 S_force_version(pTHX_ char *s)
803 OP *version = Nullop;
807 /* default VERSION number -- GBARR */
812 for( d=s, c = 1; isDIGIT(*d) || *d == '_' || (*d == '.' && c--); d++);
813 if((*d == ';' || isSPACE(*d)) && *(skipspace(d)) != ',') {
815 /* real VERSION number -- GBARR */
816 version = yylval.opval;
820 /* NOTE: The parser sees the package name and the VERSION swapped */
821 PL_nextval[PL_nexttoke].opval = version;
829 * Tokenize a quoted string passed in as an SV. It finds the next
830 * chunk, up to end of string or a backslash. It may make a new
831 * SV containing that chunk (if HINT_NEW_STRING is on). It also
836 S_tokeq(pTHX_ SV *sv)
847 s = SvPV_force(sv, len);
851 while (s < send && *s != '\\')
856 if ( PL_hints & HINT_NEW_STRING )
857 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
860 if (s + 1 < send && (s[1] == '\\'))
861 s++; /* all that, just for this */
866 SvCUR_set(sv, d - SvPVX(sv));
868 if ( PL_hints & HINT_NEW_STRING )
869 return new_constant(NULL, 0, "q", sv, pv, "q");
874 * Now come three functions related to double-quote context,
875 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
876 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
877 * interact with PL_lex_state, and create fake ( ... ) argument lists
878 * to handle functions and concatenation.
879 * They assume that whoever calls them will be setting up a fake
880 * join call, because each subthing puts a ',' after it. This lets
883 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
885 * (I'm not sure whether the spurious commas at the end of lcfirst's
886 * arguments and join's arguments are created or not).
891 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
893 * Pattern matching will set PL_lex_op to the pattern-matching op to
894 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
896 * OP_CONST and OP_READLINE are easy--just make the new op and return.
898 * Everything else becomes a FUNC.
900 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
901 * had an OP_CONST or OP_READLINE). This just sets us up for a
902 * call to S_sublex_push().
908 register I32 op_type = yylval.ival;
910 if (op_type == OP_NULL) {
911 yylval.opval = PL_lex_op;
915 if (op_type == OP_CONST || op_type == OP_READLINE) {
916 SV *sv = tokeq(PL_lex_stuff);
918 if (SvTYPE(sv) == SVt_PVIV) {
919 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
925 nsv = newSVpvn(p, len);
929 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
930 PL_lex_stuff = Nullsv;
934 PL_sublex_info.super_state = PL_lex_state;
935 PL_sublex_info.sub_inwhat = op_type;
936 PL_sublex_info.sub_op = PL_lex_op;
937 PL_lex_state = LEX_INTERPPUSH;
941 yylval.opval = PL_lex_op;
951 * Create a new scope to save the lexing state. The scope will be
952 * ended in S_sublex_done. Returns a '(', starting the function arguments
953 * to the uc, lc, etc. found before.
954 * Sets PL_lex_state to LEX_INTERPCONCAT.
963 PL_lex_state = PL_sublex_info.super_state;
964 SAVEI32(PL_lex_dojoin);
965 SAVEI32(PL_lex_brackets);
966 SAVEI32(PL_lex_fakebrack);
967 SAVEI32(PL_lex_casemods);
968 SAVEI32(PL_lex_starts);
969 SAVEI32(PL_lex_state);
970 SAVESPTR(PL_lex_inpat);
971 SAVEI32(PL_lex_inwhat);
972 SAVECOPLINE(PL_curcop);
974 SAVEPPTR(PL_oldbufptr);
975 SAVEPPTR(PL_oldoldbufptr);
976 SAVEPPTR(PL_linestart);
977 SAVESPTR(PL_linestr);
978 SAVEPPTR(PL_lex_brackstack);
979 SAVEPPTR(PL_lex_casestack);
981 PL_linestr = PL_lex_stuff;
982 PL_lex_stuff = Nullsv;
984 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
986 PL_bufend += SvCUR(PL_linestr);
987 SAVEFREESV(PL_linestr);
989 PL_lex_dojoin = FALSE;
991 PL_lex_fakebrack = 0;
992 New(899, PL_lex_brackstack, 120, char);
993 New(899, PL_lex_casestack, 12, char);
994 SAVEFREEPV(PL_lex_brackstack);
995 SAVEFREEPV(PL_lex_casestack);
997 *PL_lex_casestack = '\0';
999 PL_lex_state = LEX_INTERPCONCAT;
1000 CopLINE_set(PL_curcop, PL_multi_start);
1002 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1003 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1004 PL_lex_inpat = PL_sublex_info.sub_op;
1006 PL_lex_inpat = Nullop;
1013 * Restores lexer state after a S_sublex_push.
1019 if (!PL_lex_starts++) {
1020 PL_expect = XOPERATOR;
1021 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpvn("",0));
1025 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1026 PL_lex_state = LEX_INTERPCASEMOD;
1030 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1031 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1032 PL_linestr = PL_lex_repl;
1034 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1035 PL_bufend += SvCUR(PL_linestr);
1036 SAVEFREESV(PL_linestr);
1037 PL_lex_dojoin = FALSE;
1038 PL_lex_brackets = 0;
1039 PL_lex_fakebrack = 0;
1040 PL_lex_casemods = 0;
1041 *PL_lex_casestack = '\0';
1043 if (SvEVALED(PL_lex_repl)) {
1044 PL_lex_state = LEX_INTERPNORMAL;
1046 /* we don't clear PL_lex_repl here, so that we can check later
1047 whether this is an evalled subst; that means we rely on the
1048 logic to ensure sublex_done() is called again only via the
1049 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1052 PL_lex_state = LEX_INTERPCONCAT;
1053 PL_lex_repl = Nullsv;
1059 PL_bufend = SvPVX(PL_linestr);
1060 PL_bufend += SvCUR(PL_linestr);
1061 PL_expect = XOPERATOR;
1062 PL_sublex_info.sub_inwhat = 0;
1070 Extracts a pattern, double-quoted string, or transliteration. This
1073 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1074 processing a pattern (PL_lex_inpat is true), a transliteration
1075 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1077 Returns a pointer to the character scanned up to. Iff this is
1078 advanced from the start pointer supplied (ie if anything was
1079 successfully parsed), will leave an OP for the substring scanned
1080 in yylval. Caller must intuit reason for not parsing further
1081 by looking at the next characters herself.
1085 double-quoted style: \r and \n
1086 regexp special ones: \D \s
1088 backrefs: \1 (deprecated in substitution replacements)
1089 case and quoting: \U \Q \E
1090 stops on @ and $, but not for $ as tail anchor
1092 In transliterations:
1093 characters are VERY literal, except for - not at the start or end
1094 of the string, which indicates a range. scan_const expands the
1095 range to the full set of intermediate characters.
1097 In double-quoted strings:
1099 double-quoted style: \r and \n
1101 backrefs: \1 (deprecated)
1102 case and quoting: \U \Q \E
1105 scan_const does *not* construct ops to handle interpolated strings.
1106 It stops processing as soon as it finds an embedded $ or @ variable
1107 and leaves it to the caller to work out what's going on.
1109 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @:foo.
1111 $ in pattern could be $foo or could be tail anchor. Assumption:
1112 it's a tail anchor if $ is the last thing in the string, or if it's
1113 followed by one of ")| \n\t"
1115 \1 (backreferences) are turned into $1
1117 The structure of the code is
1118 while (there's a character to process) {
1119 handle transliteration ranges
1120 skip regexp comments
1121 skip # initiated comments in //x patterns
1122 check for embedded @foo
1123 check for embedded scalars
1125 leave intact backslashes from leave (below)
1126 deprecate \1 in strings and sub replacements
1127 handle string-changing backslashes \l \U \Q \E, etc.
1128 switch (what was escaped) {
1129 handle - in a transliteration (becomes a literal -)
1130 handle \132 octal characters
1131 handle 0x15 hex characters
1132 handle \cV (control V)
1133 handle printf backslashes (\f, \r, \n, etc)
1135 } (end if backslash)
1136 } (end while character to read)
1141 S_scan_const(pTHX_ char *start)
1143 register char *send = PL_bufend; /* end of the constant */
1144 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1145 register char *s = start; /* start of the constant */
1146 register char *d = SvPVX(sv); /* destination for copies */
1147 bool dorange = FALSE; /* are we in a translit range? */
1149 I32 utf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1150 ? (PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
1152 I32 thisutf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1153 ? (PL_sublex_info.sub_op->op_private & (PL_lex_repl ?
1154 OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF))
1156 const char *leaveit = /* set of acceptably-backslashed characters */
1158 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
1161 while (s < send || dorange) {
1162 /* get transliterations out of the way (they're most literal) */
1163 if (PL_lex_inwhat == OP_TRANS) {
1164 /* expand a range A-Z to the full set of characters. AIE! */
1166 I32 i; /* current expanded character */
1167 I32 min; /* first character in range */
1168 I32 max; /* last character in range */
1170 i = d - SvPVX(sv); /* remember current offset */
1171 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1172 d = SvPVX(sv) + i; /* refresh d after realloc */
1173 d -= 2; /* eat the first char and the - */
1175 min = (U8)*d; /* first char in range */
1176 max = (U8)d[1]; /* last char in range */
1179 if ((isLOWER(min) && isLOWER(max)) ||
1180 (isUPPER(min) && isUPPER(max))) {
1182 for (i = min; i <= max; i++)
1186 for (i = min; i <= max; i++)
1193 for (i = min; i <= max; i++)
1196 /* mark the range as done, and continue */
1201 /* range begins (ignore - as first or last char) */
1202 else if (*s == '-' && s+1 < send && s != start) {
1204 *d++ = (char)0xff; /* use illegal utf8 byte--see pmtrans */
1213 /* if we get here, we're not doing a transliteration */
1215 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1216 except for the last char, which will be done separately. */
1217 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1219 while (s < send && *s != ')')
1221 } else if (s[2] == '{'
1222 || s[2] == 'p' && s[3] == '{') { /* This should march regcomp.c */
1224 char *regparse = s + (s[2] == '{' ? 3 : 4);
1227 while (count && (c = *regparse)) {
1228 if (c == '\\' && regparse[1])
1236 if (*regparse != ')') {
1237 regparse--; /* Leave one char for continuation. */
1238 yyerror("Sequence (?{...}) not terminated or not {}-balanced");
1240 while (s < regparse)
1245 /* likewise skip #-initiated comments in //x patterns */
1246 else if (*s == '#' && PL_lex_inpat &&
1247 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1248 while (s+1 < send && *s != '\n')
1252 /* check for embedded arrays (@foo, @:foo, @'foo, @{foo}, @$foo) */
1253 else if (*s == '@' && s[1] && (isALNUM_lazy(s+1) || strchr(":'{$", s[1])))
1256 /* check for embedded scalars. only stop if we're sure it's a
1259 else if (*s == '$') {
1260 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1262 if (s + 1 < send && !strchr("()| \n\t", s[1]))
1263 break; /* in regexp, $ might be tail anchor */
1266 /* (now in tr/// code again) */
1268 if (*s & 0x80 && thisutf) {
1269 dTHR; /* only for ckWARN */
1270 if (ckWARN(WARN_UTF8)) {
1271 (void)utf8_to_uv((U8*)s, &len); /* could cvt latin-1 to utf8 here... */
1281 if (*s == '\\' && s+1 < send) {
1284 /* some backslashes we leave behind */
1285 if (*leaveit && *s && strchr(leaveit, *s)) {
1291 /* deprecate \1 in strings and substitution replacements */
1292 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1293 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1295 dTHR; /* only for ckWARN */
1296 if (ckWARN(WARN_SYNTAX))
1297 Perl_warner(aTHX_ WARN_SYNTAX, "\\%c better written as $%c", *s, *s);
1302 /* string-change backslash escapes */
1303 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1308 /* if we get here, it's either a quoted -, or a digit */
1311 /* quoted - in transliterations */
1313 if (PL_lex_inwhat == OP_TRANS) {
1321 if (ckWARN(WARN_UNSAFE) && isALPHA(*s))
1322 Perl_warner(aTHX_ WARN_UNSAFE,
1323 "Unrecognized escape \\%c passed through",
1325 /* default action is to copy the quoted character */
1330 /* \132 indicates an octal constant */
1331 case '0': case '1': case '2': case '3':
1332 case '4': case '5': case '6': case '7':
1333 *d++ = (char)scan_oct(s, 3, &len);
1337 /* \x24 indicates a hex constant */
1341 char* e = strchr(s, '}');
1344 yyerror("Missing right brace on \\x{}");
1349 if (ckWARN(WARN_UTF8))
1350 Perl_warner(aTHX_ WARN_UTF8,
1351 "Use of \\x{} without utf8 declaration");
1353 /* note: utf always shorter than hex */
1354 d = (char*)uv_to_utf8((U8*)d,
1355 (UV)scan_hex(s + 1, e - s - 1, &len));
1359 UV uv = (UV)scan_hex(s, 2, &len);
1360 if (utf && PL_lex_inwhat == OP_TRANS &&
1361 utf != (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
1363 d = (char*)uv_to_utf8((U8*)d, uv); /* doing a CU or UC */
1366 if (uv >= 127 && UTF) {
1368 if (ckWARN(WARN_UTF8))
1369 Perl_warner(aTHX_ WARN_UTF8,
1370 "\\x%.*s will produce malformed UTF-8 character; use \\x{%.*s} for that",
1379 /* \N{latin small letter a} is a named character */
1383 char* e = strchr(s, '}');
1392 yyerror("Missing right brace on \\N{}");
1396 res = newSVpvn(s + 1, e - s - 1);
1397 res = new_constant( Nullch, 0, "charnames",
1398 res, Nullsv, "\\N{...}" );
1399 str = SvPV(res,len);
1400 if (len > e - s + 4) {
1401 char *odest = SvPVX(sv);
1403 SvGROW(sv, (SvCUR(sv) + len - (e - s + 4)));
1404 d = SvPVX(sv) + (d - odest);
1406 Copy(str, d, len, char);
1413 yyerror("Missing braces on \\N{}");
1416 /* \c is a control character */
1430 /* printf-style backslashes, formfeeds, newlines, etc */
1448 *d++ = '\047'; /* CP 1047 */
1451 *d++ = '\057'; /* CP 1047 */
1465 } /* end if (backslash) */
1468 } /* while loop to process each character */
1470 /* terminate the string and set up the sv */
1472 SvCUR_set(sv, d - SvPVX(sv));
1475 /* shrink the sv if we allocated more than we used */
1476 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1477 SvLEN_set(sv, SvCUR(sv) + 1);
1478 Renew(SvPVX(sv), SvLEN(sv), char);
1481 /* return the substring (via yylval) only if we parsed anything */
1482 if (s > PL_bufptr) {
1483 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1484 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1486 ( PL_lex_inwhat == OP_TRANS
1488 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1491 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1498 * Returns TRUE if there's more to the expression (e.g., a subscript),
1501 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1503 * ->[ and ->{ return TRUE
1504 * { and [ outside a pattern are always subscripts, so return TRUE
1505 * if we're outside a pattern and it's not { or [, then return FALSE
1506 * if we're in a pattern and the first char is a {
1507 * {4,5} (any digits around the comma) returns FALSE
1508 * if we're in a pattern and the first char is a [
1510 * [SOMETHING] has a funky algorithm to decide whether it's a
1511 * character class or not. It has to deal with things like
1512 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1513 * anything else returns TRUE
1516 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1519 S_intuit_more(pTHX_ register char *s)
1521 if (PL_lex_brackets)
1523 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1525 if (*s != '{' && *s != '[')
1530 /* In a pattern, so maybe we have {n,m}. */
1547 /* On the other hand, maybe we have a character class */
1550 if (*s == ']' || *s == '^')
1553 /* this is terrifying, and it works */
1554 int weight = 2; /* let's weigh the evidence */
1556 unsigned char un_char = 255, last_un_char;
1557 char *send = strchr(s,']');
1558 char tmpbuf[sizeof PL_tokenbuf * 4];
1560 if (!send) /* has to be an expression */
1563 Zero(seen,256,char);
1566 else if (isDIGIT(*s)) {
1568 if (isDIGIT(s[1]) && s[2] == ']')
1574 for (; s < send; s++) {
1575 last_un_char = un_char;
1576 un_char = (unsigned char)*s;
1581 weight -= seen[un_char] * 10;
1582 if (isALNUM_lazy(s+1)) {
1583 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1584 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1589 else if (*s == '$' && s[1] &&
1590 strchr("[#!%*<>()-=",s[1])) {
1591 if (/*{*/ strchr("])} =",s[2]))
1600 if (strchr("wds]",s[1]))
1602 else if (seen['\''] || seen['"'])
1604 else if (strchr("rnftbxcav",s[1]))
1606 else if (isDIGIT(s[1])) {
1608 while (s[1] && isDIGIT(s[1]))
1618 if (strchr("aA01! ",last_un_char))
1620 if (strchr("zZ79~",s[1]))
1622 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1623 weight -= 5; /* cope with negative subscript */
1626 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1627 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1632 if (keyword(tmpbuf, d - tmpbuf))
1635 if (un_char == last_un_char + 1)
1637 weight -= seen[un_char];
1642 if (weight >= 0) /* probably a character class */
1652 * Does all the checking to disambiguate
1654 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1655 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1657 * First argument is the stuff after the first token, e.g. "bar".
1659 * Not a method if bar is a filehandle.
1660 * Not a method if foo is a subroutine prototyped to take a filehandle.
1661 * Not a method if it's really "Foo $bar"
1662 * Method if it's "foo $bar"
1663 * Not a method if it's really "print foo $bar"
1664 * Method if it's really "foo package::" (interpreted as package->foo)
1665 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1666 * Not a method if bar is a filehandle or package, but is quotd with
1671 S_intuit_method(pTHX_ char *start, GV *gv)
1673 char *s = start + (*start == '$');
1674 char tmpbuf[sizeof PL_tokenbuf];
1682 if ((cv = GvCVu(gv))) {
1683 char *proto = SvPVX(cv);
1693 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1694 /* start is the beginning of the possible filehandle/object,
1695 * and s is the end of it
1696 * tmpbuf is a copy of it
1699 if (*start == '$') {
1700 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1705 return *s == '(' ? FUNCMETH : METHOD;
1707 if (!keyword(tmpbuf, len)) {
1708 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1713 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1714 if (indirgv && GvCVu(indirgv))
1716 /* filehandle or package name makes it a method */
1717 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1719 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1720 return 0; /* no assumptions -- "=>" quotes bearword */
1722 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1723 newSVpvn(tmpbuf,len));
1724 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1728 return *s == '(' ? FUNCMETH : METHOD;
1736 * Return a string of Perl code to load the debugger. If PERL5DB
1737 * is set, it will return the contents of that, otherwise a
1738 * compile-time require of perl5db.pl.
1745 char *pdb = PerlEnv_getenv("PERL5DB");
1749 SETERRNO(0,SS$_NORMAL);
1750 return "BEGIN { require 'perl5db.pl' }";
1756 /* Encoded script support. filter_add() effectively inserts a
1757 * 'pre-processing' function into the current source input stream.
1758 * Note that the filter function only applies to the current source file
1759 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1761 * The datasv parameter (which may be NULL) can be used to pass
1762 * private data to this instance of the filter. The filter function
1763 * can recover the SV using the FILTER_DATA macro and use it to
1764 * store private buffers and state information.
1766 * The supplied datasv parameter is upgraded to a PVIO type
1767 * and the IoDIRP field is used to store the function pointer,
1768 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1769 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1770 * private use must be set using malloc'd pointers.
1774 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1779 if (!PL_rsfp_filters)
1780 PL_rsfp_filters = newAV();
1782 datasv = NEWSV(255,0);
1783 if (!SvUPGRADE(datasv, SVt_PVIO))
1784 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1785 IoDIRP(datasv) = (DIR*)funcp; /* stash funcp into spare field */
1786 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
1787 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1788 funcp, SvPV_nolen(datasv)));
1789 av_unshift(PL_rsfp_filters, 1);
1790 av_store(PL_rsfp_filters, 0, datasv) ;
1795 /* Delete most recently added instance of this filter function. */
1797 Perl_filter_del(pTHX_ filter_t funcp)
1800 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp));
1801 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
1803 /* if filter is on top of stack (usual case) just pop it off */
1804 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
1805 if (IoDIRP(datasv) == (DIR*)funcp) {
1806 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
1807 IoDIRP(datasv) = (DIR*)NULL;
1808 sv_free(av_pop(PL_rsfp_filters));
1812 /* we need to search for the correct entry and clear it */
1813 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
1817 /* Invoke the n'th filter function for the current rsfp. */
1819 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
1822 /* 0 = read one text line */
1827 if (!PL_rsfp_filters)
1829 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
1830 /* Provide a default input filter to make life easy. */
1831 /* Note that we append to the line. This is handy. */
1832 DEBUG_P(PerlIO_printf(Perl_debug_log,
1833 "filter_read %d: from rsfp\n", idx));
1837 int old_len = SvCUR(buf_sv) ;
1839 /* ensure buf_sv is large enough */
1840 SvGROW(buf_sv, old_len + maxlen) ;
1841 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
1842 if (PerlIO_error(PL_rsfp))
1843 return -1; /* error */
1845 return 0 ; /* end of file */
1847 SvCUR_set(buf_sv, old_len + len) ;
1850 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
1851 if (PerlIO_error(PL_rsfp))
1852 return -1; /* error */
1854 return 0 ; /* end of file */
1857 return SvCUR(buf_sv);
1859 /* Skip this filter slot if filter has been deleted */
1860 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
1861 DEBUG_P(PerlIO_printf(Perl_debug_log,
1862 "filter_read %d: skipped (filter deleted)\n",
1864 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
1866 /* Get function pointer hidden within datasv */
1867 funcp = (filter_t)IoDIRP(datasv);
1868 DEBUG_P(PerlIO_printf(Perl_debug_log,
1869 "filter_read %d: via function %p (%s)\n",
1870 idx, funcp, SvPV_nolen(datasv)));
1871 /* Call function. The function is expected to */
1872 /* call "FILTER_READ(idx+1, buf_sv)" first. */
1873 /* Return: <0:error, =0:eof, >0:not eof */
1874 return (*funcp)(aTHXo_ idx, buf_sv, maxlen);
1878 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
1881 if (!PL_rsfp_filters) {
1882 filter_add(win32_textfilter,NULL);
1885 if (PL_rsfp_filters) {
1888 SvCUR_set(sv, 0); /* start with empty line */
1889 if (FILTER_READ(0, sv, 0) > 0)
1890 return ( SvPVX(sv) ) ;
1895 return (sv_gets(sv, fp, append));
1900 static char* exp_name[] =
1901 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
1902 "ATTRTERM", "TERMBLOCK"
1909 Works out what to call the token just pulled out of the input
1910 stream. The yacc parser takes care of taking the ops we return and
1911 stitching them into a tree.
1917 if read an identifier
1918 if we're in a my declaration
1919 croak if they tried to say my($foo::bar)
1920 build the ops for a my() declaration
1921 if it's an access to a my() variable
1922 are we in a sort block?
1923 croak if my($a); $a <=> $b
1924 build ops for access to a my() variable
1925 if in a dq string, and they've said @foo and we can't find @foo
1927 build ops for a bareword
1928 if we already built the token before, use it.
1932 #ifdef USE_PURE_BISON
1933 Perl_yylex(pTHX_ YYSTYPE *lvalp, int *lcharp)
1946 #ifdef USE_PURE_BISON
1947 yylval_pointer = lvalp;
1948 yychar_pointer = lcharp;
1951 /* check if there's an identifier for us to look at */
1952 if (PL_pending_ident) {
1953 /* pit holds the identifier we read and pending_ident is reset */
1954 char pit = PL_pending_ident;
1955 PL_pending_ident = 0;
1957 /* if we're in a my(), we can't allow dynamics here.
1958 $foo'bar has already been turned into $foo::bar, so
1959 just check for colons.
1961 if it's a legal name, the OP is a PADANY.
1964 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
1965 tmp = pad_allocmy(PL_tokenbuf);
1968 if (strchr(PL_tokenbuf,':'))
1969 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
1971 yylval.opval = newOP(OP_PADANY, 0);
1972 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
1978 build the ops for accesses to a my() variable.
1980 Deny my($a) or my($b) in a sort block, *if* $a or $b is
1981 then used in a comparison. This catches most, but not
1982 all cases. For instance, it catches
1983 sort { my($a); $a <=> $b }
1985 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
1986 (although why you'd do that is anyone's guess).
1989 if (!strchr(PL_tokenbuf,':')) {
1991 /* Check for single character per-thread SVs */
1992 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
1993 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
1994 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
1996 yylval.opval = newOP(OP_THREADSV, 0);
1997 yylval.opval->op_targ = tmp;
2000 #endif /* USE_THREADS */
2001 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
2002 /* might be an "our" variable" */
2003 if (SvFLAGS(AvARRAY(PL_comppad_name)[tmp]) & SVpad_OUR) {
2004 /* build ops for a bareword */
2005 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2006 yylval.opval->op_private = OPpCONST_ENTERED;
2007 gv_fetchpv(PL_tokenbuf+1,
2009 ? (GV_ADDMULTI | GV_ADDINEVAL | GV_ADDOUR)
2012 ((PL_tokenbuf[0] == '$') ? SVt_PV
2013 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2018 /* if it's a sort block and they're naming $a or $b */
2019 if (PL_last_lop_op == OP_SORT &&
2020 PL_tokenbuf[0] == '$' &&
2021 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
2024 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
2025 d < PL_bufend && *d != '\n';
2028 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
2029 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
2035 yylval.opval = newOP(OP_PADANY, 0);
2036 yylval.opval->op_targ = tmp;
2042 Whine if they've said @foo in a doublequoted string,
2043 and @foo isn't a variable we can find in the symbol
2046 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
2047 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
2048 if (!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
2049 yyerror(Perl_form(aTHX_ "In string, %s now must be written as \\%s",
2050 PL_tokenbuf, PL_tokenbuf));
2053 /* build ops for a bareword */
2054 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2055 yylval.opval->op_private = OPpCONST_ENTERED;
2056 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
2057 ((PL_tokenbuf[0] == '$') ? SVt_PV
2058 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2063 /* no identifier pending identification */
2065 switch (PL_lex_state) {
2067 case LEX_NORMAL: /* Some compilers will produce faster */
2068 case LEX_INTERPNORMAL: /* code if we comment these out. */
2072 /* when we've already built the next token, just pull it out of the queue */
2075 yylval = PL_nextval[PL_nexttoke];
2077 PL_lex_state = PL_lex_defer;
2078 PL_expect = PL_lex_expect;
2079 PL_lex_defer = LEX_NORMAL;
2081 return(PL_nexttype[PL_nexttoke]);
2083 /* interpolated case modifiers like \L \U, including \Q and \E.
2084 when we get here, PL_bufptr is at the \
2086 case LEX_INTERPCASEMOD:
2088 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2089 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2091 /* handle \E or end of string */
2092 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2096 if (PL_lex_casemods) {
2097 oldmod = PL_lex_casestack[--PL_lex_casemods];
2098 PL_lex_casestack[PL_lex_casemods] = '\0';
2100 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2102 PL_lex_state = LEX_INTERPCONCAT;
2106 if (PL_bufptr != PL_bufend)
2108 PL_lex_state = LEX_INTERPCONCAT;
2113 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2114 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2115 if (strchr("LU", *s) &&
2116 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2118 PL_lex_casestack[--PL_lex_casemods] = '\0';
2121 if (PL_lex_casemods > 10) {
2122 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2123 if (newlb != PL_lex_casestack) {
2125 PL_lex_casestack = newlb;
2128 PL_lex_casestack[PL_lex_casemods++] = *s;
2129 PL_lex_casestack[PL_lex_casemods] = '\0';
2130 PL_lex_state = LEX_INTERPCONCAT;
2131 PL_nextval[PL_nexttoke].ival = 0;
2134 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2136 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2138 PL_nextval[PL_nexttoke].ival = OP_LC;
2140 PL_nextval[PL_nexttoke].ival = OP_UC;
2142 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2144 Perl_croak(aTHX_ "panic: yylex");
2147 if (PL_lex_starts) {
2156 case LEX_INTERPPUSH:
2157 return sublex_push();
2159 case LEX_INTERPSTART:
2160 if (PL_bufptr == PL_bufend)
2161 return sublex_done();
2163 PL_lex_dojoin = (*PL_bufptr == '@');
2164 PL_lex_state = LEX_INTERPNORMAL;
2165 if (PL_lex_dojoin) {
2166 PL_nextval[PL_nexttoke].ival = 0;
2169 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2170 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2171 force_next(PRIVATEREF);
2173 force_ident("\"", '$');
2174 #endif /* USE_THREADS */
2175 PL_nextval[PL_nexttoke].ival = 0;
2177 PL_nextval[PL_nexttoke].ival = 0;
2179 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2182 if (PL_lex_starts++) {
2188 case LEX_INTERPENDMAYBE:
2189 if (intuit_more(PL_bufptr)) {
2190 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2196 if (PL_lex_dojoin) {
2197 PL_lex_dojoin = FALSE;
2198 PL_lex_state = LEX_INTERPCONCAT;
2201 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2202 && SvEVALED(PL_lex_repl))
2204 if (PL_bufptr != PL_bufend)
2205 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2206 PL_lex_repl = Nullsv;
2209 case LEX_INTERPCONCAT:
2211 if (PL_lex_brackets)
2212 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2214 if (PL_bufptr == PL_bufend)
2215 return sublex_done();
2217 if (SvIVX(PL_linestr) == '\'') {
2218 SV *sv = newSVsv(PL_linestr);
2221 else if ( PL_hints & HINT_NEW_RE )
2222 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2223 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2227 s = scan_const(PL_bufptr);
2229 PL_lex_state = LEX_INTERPCASEMOD;
2231 PL_lex_state = LEX_INTERPSTART;
2234 if (s != PL_bufptr) {
2235 PL_nextval[PL_nexttoke] = yylval;
2238 if (PL_lex_starts++)
2248 PL_lex_state = LEX_NORMAL;
2249 s = scan_formline(PL_bufptr);
2250 if (!PL_lex_formbrack)
2256 PL_oldoldbufptr = PL_oldbufptr;
2259 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2260 exp_name[PL_expect], s);
2266 if (isIDFIRST_lazy(s))
2268 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2271 goto fake_eof; /* emulate EOF on ^D or ^Z */
2276 if (PL_lex_brackets)
2277 yyerror("Missing right curly or square bracket");
2280 if (s++ < PL_bufend)
2281 goto retry; /* ignore stray nulls */
2284 if (!PL_in_eval && !PL_preambled) {
2285 PL_preambled = TRUE;
2286 sv_setpv(PL_linestr,incl_perldb());
2287 if (SvCUR(PL_linestr))
2288 sv_catpv(PL_linestr,";");
2290 while(AvFILLp(PL_preambleav) >= 0) {
2291 SV *tmpsv = av_shift(PL_preambleav);
2292 sv_catsv(PL_linestr, tmpsv);
2293 sv_catpv(PL_linestr, ";");
2296 sv_free((SV*)PL_preambleav);
2297 PL_preambleav = NULL;
2299 if (PL_minus_n || PL_minus_p) {
2300 sv_catpv(PL_linestr, "LINE: while (<>) {");
2302 sv_catpv(PL_linestr,"chomp;");
2304 GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
2306 GvIMPORTED_AV_on(gv);
2308 if (strchr("/'\"", *PL_splitstr)
2309 && strchr(PL_splitstr + 1, *PL_splitstr))
2310 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2313 s = "'~#\200\1'"; /* surely one char is unused...*/
2314 while (s[1] && strchr(PL_splitstr, *s)) s++;
2316 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c",
2317 "q" + (delim == '\''), delim);
2318 for (s = PL_splitstr; *s; s++) {
2320 sv_catpvn(PL_linestr, "\\", 1);
2321 sv_catpvn(PL_linestr, s, 1);
2323 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2327 sv_catpv(PL_linestr,"@F=split(' ');");
2330 sv_catpv(PL_linestr, "\n");
2331 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2332 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2333 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2334 SV *sv = NEWSV(85,0);
2336 sv_upgrade(sv, SVt_PVMG);
2337 sv_setsv(sv,PL_linestr);
2338 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2343 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2346 if (PL_preprocess && !PL_in_eval)
2347 (void)PerlProc_pclose(PL_rsfp);
2348 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2349 PerlIO_clearerr(PL_rsfp);
2351 (void)PerlIO_close(PL_rsfp);
2353 PL_doextract = FALSE;
2355 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2356 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2357 sv_catpv(PL_linestr,";}");
2358 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2359 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2360 PL_minus_n = PL_minus_p = 0;
2363 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2364 sv_setpv(PL_linestr,"");
2365 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2368 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2369 PL_doextract = FALSE;
2371 /* Incest with pod. */
2372 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2373 sv_setpv(PL_linestr, "");
2374 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2375 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2376 PL_doextract = FALSE;
2380 } while (PL_doextract);
2381 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2382 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2383 SV *sv = NEWSV(85,0);
2385 sv_upgrade(sv, SVt_PVMG);
2386 sv_setsv(sv,PL_linestr);
2387 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2389 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2390 if (CopLINE(PL_curcop) == 1) {
2391 while (s < PL_bufend && isSPACE(*s))
2393 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2397 if (*s == '#' && *(s+1) == '!')
2399 #ifdef ALTERNATE_SHEBANG
2401 static char as[] = ALTERNATE_SHEBANG;
2402 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2403 d = s + (sizeof(as) - 1);
2405 #endif /* ALTERNATE_SHEBANG */
2414 while (*d && !isSPACE(*d))
2418 #ifdef ARG_ZERO_IS_SCRIPT
2419 if (ipathend > ipath) {
2421 * HP-UX (at least) sets argv[0] to the script name,
2422 * which makes $^X incorrect. And Digital UNIX and Linux,
2423 * at least, set argv[0] to the basename of the Perl
2424 * interpreter. So, having found "#!", we'll set it right.
2426 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2427 assert(SvPOK(x) || SvGMAGICAL(x));
2428 if (sv_eq(x, CopFILESV(PL_curcop))) {
2429 sv_setpvn(x, ipath, ipathend - ipath);
2432 TAINT_NOT; /* $^X is always tainted, but that's OK */
2434 #endif /* ARG_ZERO_IS_SCRIPT */
2439 d = instr(s,"perl -");
2441 d = instr(s,"perl");
2443 /* avoid getting into infinite loops when shebang
2444 * line contains "Perl" rather than "perl" */
2446 for (d = ipathend-4; d >= ipath; --d) {
2447 if ((*d == 'p' || *d == 'P')
2448 && !ibcmp(d, "perl", 4))
2458 #ifdef ALTERNATE_SHEBANG
2460 * If the ALTERNATE_SHEBANG on this system starts with a
2461 * character that can be part of a Perl expression, then if
2462 * we see it but not "perl", we're probably looking at the
2463 * start of Perl code, not a request to hand off to some
2464 * other interpreter. Similarly, if "perl" is there, but
2465 * not in the first 'word' of the line, we assume the line
2466 * contains the start of the Perl program.
2468 if (d && *s != '#') {
2470 while (*c && !strchr("; \t\r\n\f\v#", *c))
2473 d = Nullch; /* "perl" not in first word; ignore */
2475 *s = '#'; /* Don't try to parse shebang line */
2477 #endif /* ALTERNATE_SHEBANG */
2482 !instr(s,"indir") &&
2483 instr(PL_origargv[0],"perl"))
2489 while (s < PL_bufend && isSPACE(*s))
2491 if (s < PL_bufend) {
2492 Newz(899,newargv,PL_origargc+3,char*);
2494 while (s < PL_bufend && !isSPACE(*s))
2497 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2500 newargv = PL_origargv;
2502 PerlProc_execv(ipath, newargv);
2503 Perl_croak(aTHX_ "Can't exec %s", ipath);
2506 U32 oldpdb = PL_perldb;
2507 bool oldn = PL_minus_n;
2508 bool oldp = PL_minus_p;
2510 while (*d && !isSPACE(*d)) d++;
2511 while (*d == ' ' || *d == '\t') d++;
2515 if (*d == 'M' || *d == 'm') {
2517 while (*d && !isSPACE(*d)) d++;
2518 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2521 d = moreswitches(d);
2523 if (PERLDB_LINE && !oldpdb ||
2524 ( PL_minus_n || PL_minus_p ) && !(oldn || oldp) )
2525 /* if we have already added "LINE: while (<>) {",
2526 we must not do it again */
2528 sv_setpv(PL_linestr, "");
2529 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2530 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2531 PL_preambled = FALSE;
2533 (void)gv_fetchfile(PL_origfilename);
2540 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2542 PL_lex_state = LEX_FORMLINE;
2547 #ifdef PERL_STRICT_CR
2548 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2550 "(Maybe you didn't strip carriage returns after a network transfer?)\n");
2552 case ' ': case '\t': case '\f': case 013:
2557 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2559 while (s < d && *s != '\n')
2564 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2566 PL_lex_state = LEX_FORMLINE;
2576 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2581 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
2584 if (strnEQ(s,"=>",2)) {
2585 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2586 OPERATOR('-'); /* unary minus */
2588 PL_last_uni = PL_oldbufptr;
2589 PL_last_lop_op = OP_FTEREAD; /* good enough */
2591 case 'r': FTST(OP_FTEREAD);
2592 case 'w': FTST(OP_FTEWRITE);
2593 case 'x': FTST(OP_FTEEXEC);
2594 case 'o': FTST(OP_FTEOWNED);
2595 case 'R': FTST(OP_FTRREAD);
2596 case 'W': FTST(OP_FTRWRITE);
2597 case 'X': FTST(OP_FTREXEC);
2598 case 'O': FTST(OP_FTROWNED);
2599 case 'e': FTST(OP_FTIS);
2600 case 'z': FTST(OP_FTZERO);
2601 case 's': FTST(OP_FTSIZE);
2602 case 'f': FTST(OP_FTFILE);
2603 case 'd': FTST(OP_FTDIR);
2604 case 'l': FTST(OP_FTLINK);
2605 case 'p': FTST(OP_FTPIPE);
2606 case 'S': FTST(OP_FTSOCK);
2607 case 'u': FTST(OP_FTSUID);
2608 case 'g': FTST(OP_FTSGID);
2609 case 'k': FTST(OP_FTSVTX);
2610 case 'b': FTST(OP_FTBLK);
2611 case 'c': FTST(OP_FTCHR);
2612 case 't': FTST(OP_FTTTY);
2613 case 'T': FTST(OP_FTTEXT);
2614 case 'B': FTST(OP_FTBINARY);
2615 case 'M': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTMTIME);
2616 case 'A': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTATIME);
2617 case 'C': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTCTIME);
2619 Perl_croak(aTHX_ "Unrecognized file test: -%c", (int)tmp);
2626 if (PL_expect == XOPERATOR)
2631 else if (*s == '>') {
2634 if (isIDFIRST_lazy(s)) {
2635 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2643 if (PL_expect == XOPERATOR)
2646 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2648 OPERATOR('-'); /* unary minus */
2655 if (PL_expect == XOPERATOR)
2660 if (PL_expect == XOPERATOR)
2663 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2669 if (PL_expect != XOPERATOR) {
2670 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2671 PL_expect = XOPERATOR;
2672 force_ident(PL_tokenbuf, '*');
2685 if (PL_expect == XOPERATOR) {
2689 PL_tokenbuf[0] = '%';
2690 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2691 if (!PL_tokenbuf[1]) {
2693 yyerror("Final % should be \\% or %name");
2696 PL_pending_ident = '%';
2715 switch (PL_expect) {
2718 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2720 PL_bufptr = s; /* update in case we back off */
2726 PL_expect = XTERMBLOCK;
2730 while (isIDFIRST_lazy(s)) {
2731 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2733 d = scan_str(d,TRUE,TRUE);
2736 SvREFCNT_dec(PL_lex_stuff);
2737 PL_lex_stuff = Nullsv;
2739 /* MUST advance bufptr here to avoid bogus
2740 "at end of line" context messages from yyerror().
2742 PL_bufptr = s + len;
2743 yyerror("Unterminated attribute parameter in attribute list");
2746 return 0; /* EOF indicator */
2750 SV *sv = newSVpvn(s, len);
2751 sv_catsv(sv, PL_lex_stuff);
2752 attrs = append_elem(OP_LIST, attrs,
2753 newSVOP(OP_CONST, 0, sv));
2754 SvREFCNT_dec(PL_lex_stuff);
2755 PL_lex_stuff = Nullsv;
2758 attrs = append_elem(OP_LIST, attrs,
2759 newSVOP(OP_CONST, 0,
2766 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}' for vi */
2767 if (*s != ';' && *s != tmp) {
2768 char q = ((*s == '\'') ? '"' : '\'');
2769 /* If here for an expression, and parsed no attrs, back off. */
2770 if (tmp == '=' && !attrs) {
2774 /* MUST advance bufptr here to avoid bogus "at end of line"
2775 context messages from yyerror().
2779 yyerror("Unterminated attribute list");
2781 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
2788 PL_nextval[PL_nexttoke].opval = attrs;
2796 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
2797 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
2802 if (CopLINE(PL_curcop) < PL_copline)
2803 PL_copline = CopLINE(PL_curcop);
2814 if (PL_lex_brackets <= 0)
2815 yyerror("Unmatched right square bracket");
2818 if (PL_lex_state == LEX_INTERPNORMAL) {
2819 if (PL_lex_brackets == 0) {
2820 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
2821 PL_lex_state = LEX_INTERPEND;
2828 if (PL_lex_brackets > 100) {
2829 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
2830 if (newlb != PL_lex_brackstack) {
2832 PL_lex_brackstack = newlb;
2835 switch (PL_expect) {
2837 if (PL_lex_formbrack) {
2841 if (PL_oldoldbufptr == PL_last_lop)
2842 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
2844 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2845 OPERATOR(HASHBRACK);
2847 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
2850 PL_tokenbuf[0] = '\0';
2851 if (d < PL_bufend && *d == '-') {
2852 PL_tokenbuf[0] = '-';
2854 while (d < PL_bufend && (*d == ' ' || *d == '\t'))
2857 if (d < PL_bufend && isIDFIRST_lazy(d)) {
2858 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
2860 while (d < PL_bufend && (*d == ' ' || *d == '\t'))
2863 char minus = (PL_tokenbuf[0] == '-');
2864 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
2872 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
2877 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2882 if (PL_oldoldbufptr == PL_last_lop)
2883 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
2885 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2888 OPERATOR(HASHBRACK);
2889 /* This hack serves to disambiguate a pair of curlies
2890 * as being a block or an anon hash. Normally, expectation
2891 * determines that, but in cases where we're not in a
2892 * position to expect anything in particular (like inside
2893 * eval"") we have to resolve the ambiguity. This code
2894 * covers the case where the first term in the curlies is a
2895 * quoted string. Most other cases need to be explicitly
2896 * disambiguated by prepending a `+' before the opening
2897 * curly in order to force resolution as an anon hash.
2899 * XXX should probably propagate the outer expectation
2900 * into eval"" to rely less on this hack, but that could
2901 * potentially break current behavior of eval"".
2905 if (*s == '\'' || *s == '"' || *s == '`') {
2906 /* common case: get past first string, handling escapes */
2907 for (t++; t < PL_bufend && *t != *s;)
2908 if (*t++ == '\\' && (*t == '\\' || *t == *s))
2912 else if (*s == 'q') {
2915 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
2916 && !isALNUM(*t)))) {
2918 char open, close, term;
2921 while (t < PL_bufend && isSPACE(*t))
2925 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
2929 for (t++; t < PL_bufend; t++) {
2930 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
2932 else if (*t == open)
2936 for (t++; t < PL_bufend; t++) {
2937 if (*t == '\\' && t+1 < PL_bufend)
2939 else if (*t == close && --brackets <= 0)
2941 else if (*t == open)
2947 else if (isIDFIRST_lazy(s)) {
2948 for (t++; t < PL_bufend && isALNUM_lazy(t); t++) ;
2950 while (t < PL_bufend && isSPACE(*t))
2952 /* if comma follows first term, call it an anon hash */
2953 /* XXX it could be a comma expression with loop modifiers */
2954 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
2955 || (*t == '=' && t[1] == '>')))
2956 OPERATOR(HASHBRACK);
2957 if (PL_expect == XREF)
2960 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
2966 yylval.ival = CopLINE(PL_curcop);
2967 if (isSPACE(*s) || *s == '#')
2968 PL_copline = NOLINE; /* invalidate current command line number */
2973 if (PL_lex_brackets <= 0)
2974 yyerror("Unmatched right curly bracket");
2976 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
2977 if (PL_lex_brackets < PL_lex_formbrack)
2978 PL_lex_formbrack = 0;
2979 if (PL_lex_state == LEX_INTERPNORMAL) {
2980 if (PL_lex_brackets == 0) {
2981 if (PL_lex_fakebrack) {
2982 PL_lex_state = LEX_INTERPEND;
2984 return yylex(); /* ignore fake brackets */
2986 if (*s == '-' && s[1] == '>')
2987 PL_lex_state = LEX_INTERPENDMAYBE;
2988 else if (*s != '[' && *s != '{')
2989 PL_lex_state = LEX_INTERPEND;
2992 if (PL_lex_brackets < PL_lex_fakebrack) {
2994 PL_lex_fakebrack = 0;
2995 return yylex(); /* ignore fake brackets */
3005 if (PL_expect == XOPERATOR) {
3006 if (ckWARN(WARN_SEMICOLON) && isIDFIRST_lazy(s) && PL_bufptr == PL_linestart) {
3007 CopLINE_dec(PL_curcop);
3008 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3009 CopLINE_inc(PL_curcop);
3014 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3016 PL_expect = XOPERATOR;
3017 force_ident(PL_tokenbuf, '&');
3021 yylval.ival = (OPpENTERSUB_AMPER<<8);
3040 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3041 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3043 if (PL_expect == XSTATE && isALPHA(tmp) &&
3044 (s == PL_linestart+1 || s[-2] == '\n') )
3046 if (PL_in_eval && !PL_rsfp) {
3051 if (strnEQ(s,"=cut",4)) {
3065 PL_doextract = TRUE;
3068 if (PL_lex_brackets < PL_lex_formbrack) {
3070 #ifdef PERL_STRICT_CR
3071 for (t = s; *t == ' ' || *t == '\t'; t++) ;
3073 for (t = s; *t == ' ' || *t == '\t' || *t == '\r'; t++) ;
3075 if (*t == '\n' || *t == '#') {
3093 if (PL_expect != XOPERATOR) {
3094 if (s[1] != '<' && !strchr(s,'>'))
3097 s = scan_heredoc(s);
3099 s = scan_inputsymbol(s);
3100 TERM(sublex_start());
3105 SHop(OP_LEFT_SHIFT);
3119 SHop(OP_RIGHT_SHIFT);
3128 if (PL_expect == XOPERATOR) {
3129 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3132 return ','; /* grandfather non-comma-format format */
3136 if (s[1] == '#' && (isIDFIRST_lazy(s+2) || strchr("{$:+-", s[2]))) {
3137 PL_tokenbuf[0] = '@';
3138 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3139 sizeof PL_tokenbuf - 1, FALSE);
3140 if (PL_expect == XOPERATOR)
3141 no_op("Array length", s);
3142 if (!PL_tokenbuf[1])
3144 PL_expect = XOPERATOR;
3145 PL_pending_ident = '#';
3149 PL_tokenbuf[0] = '$';
3150 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3151 sizeof PL_tokenbuf - 1, FALSE);
3152 if (PL_expect == XOPERATOR)
3154 if (!PL_tokenbuf[1]) {
3156 yyerror("Final $ should be \\$ or $name");
3160 /* This kludge not intended to be bulletproof. */
3161 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3162 yylval.opval = newSVOP(OP_CONST, 0,
3163 newSViv((IV)PL_compiling.cop_arybase));
3164 yylval.opval->op_private = OPpCONST_ARYBASE;
3170 if (PL_lex_state == LEX_NORMAL)
3173 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3176 PL_tokenbuf[0] = '@';
3177 if (ckWARN(WARN_SYNTAX)) {
3179 isSPACE(*t) || isALNUM_lazy(t) || *t == '$';
3182 PL_bufptr = skipspace(PL_bufptr);
3183 while (t < PL_bufend && *t != ']')
3185 Perl_warner(aTHX_ WARN_SYNTAX,
3186 "Multidimensional syntax %.*s not supported",
3187 (t - PL_bufptr) + 1, PL_bufptr);
3191 else if (*s == '{') {
3192 PL_tokenbuf[0] = '%';
3193 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3194 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3196 char tmpbuf[sizeof PL_tokenbuf];
3198 for (t++; isSPACE(*t); t++) ;
3199 if (isIDFIRST_lazy(t)) {
3200 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3201 for (; isSPACE(*t); t++) ;
3202 if (*t == ';' && get_cv(tmpbuf, FALSE))
3203 Perl_warner(aTHX_ WARN_SYNTAX,
3204 "You need to quote \"%s\"", tmpbuf);
3210 PL_expect = XOPERATOR;
3211 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3212 bool islop = (PL_last_lop == PL_oldoldbufptr);
3213 if (!islop || PL_last_lop_op == OP_GREPSTART)
3214 PL_expect = XOPERATOR;
3215 else if (strchr("$@\"'`q", *s))
3216 PL_expect = XTERM; /* e.g. print $fh "foo" */
3217 else if (strchr("&*<%", *s) && isIDFIRST_lazy(s+1))
3218 PL_expect = XTERM; /* e.g. print $fh &sub */
3219 else if (isIDFIRST_lazy(s)) {
3220 char tmpbuf[sizeof PL_tokenbuf];
3221 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3222 if (tmp = keyword(tmpbuf, len)) {
3223 /* binary operators exclude handle interpretations */
3235 PL_expect = XTERM; /* e.g. print $fh length() */
3240 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3241 if (gv && GvCVu(gv))
3242 PL_expect = XTERM; /* e.g. print $fh subr() */
3245 else if (isDIGIT(*s))
3246 PL_expect = XTERM; /* e.g. print $fh 3 */
3247 else if (*s == '.' && isDIGIT(s[1]))
3248 PL_expect = XTERM; /* e.g. print $fh .3 */
3249 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3250 PL_expect = XTERM; /* e.g. print $fh -1 */
3251 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3252 PL_expect = XTERM; /* print $fh <<"EOF" */
3254 PL_pending_ident = '$';
3258 if (PL_expect == XOPERATOR)
3260 PL_tokenbuf[0] = '@';
3261 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3262 if (!PL_tokenbuf[1]) {
3264 yyerror("Final @ should be \\@ or @name");
3267 if (PL_lex_state == LEX_NORMAL)
3269 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3271 PL_tokenbuf[0] = '%';
3273 /* Warn about @ where they meant $. */
3274 if (ckWARN(WARN_SYNTAX)) {
3275 if (*s == '[' || *s == '{') {
3277 while (*t && (isALNUM_lazy(t) || strchr(" \t$#+-'\"", *t)))
3279 if (*t == '}' || *t == ']') {
3281 PL_bufptr = skipspace(PL_bufptr);
3282 Perl_warner(aTHX_ WARN_SYNTAX,
3283 "Scalar value %.*s better written as $%.*s",
3284 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3289 PL_pending_ident = '@';
3292 case '/': /* may either be division or pattern */
3293 case '?': /* may either be conditional or pattern */
3294 if (PL_expect != XOPERATOR) {
3295 /* Disable warning on "study /blah/" */
3296 if (PL_oldoldbufptr == PL_last_uni
3297 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3298 || memNE(PL_last_uni, "study", 5) || isALNUM_lazy(PL_last_uni+5)))
3300 s = scan_pat(s,OP_MATCH);
3301 TERM(sublex_start());
3309 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3310 #ifdef PERL_STRICT_CR
3313 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3315 && (s == PL_linestart || s[-1] == '\n') )
3317 PL_lex_formbrack = 0;
3321 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3327 yylval.ival = OPf_SPECIAL;
3333 if (PL_expect != XOPERATOR)
3338 case '0': case '1': case '2': case '3': case '4':
3339 case '5': case '6': case '7': case '8': case '9':
3341 if (PL_expect == XOPERATOR)
3346 s = scan_str(s,FALSE,FALSE);
3347 if (PL_expect == XOPERATOR) {
3348 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3351 return ','; /* grandfather non-comma-format format */
3357 missingterm((char*)0);
3358 yylval.ival = OP_CONST;
3359 TERM(sublex_start());
3362 s = scan_str(s,FALSE,FALSE);
3363 if (PL_expect == XOPERATOR) {
3364 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3367 return ','; /* grandfather non-comma-format format */
3373 missingterm((char*)0);
3374 yylval.ival = OP_CONST;
3375 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3376 if (*d == '$' || *d == '@' || *d == '\\' || *d & 0x80) {
3377 yylval.ival = OP_STRINGIFY;
3381 TERM(sublex_start());
3384 s = scan_str(s,FALSE,FALSE);
3385 if (PL_expect == XOPERATOR)
3386 no_op("Backticks",s);
3388 missingterm((char*)0);
3389 yylval.ival = OP_BACKTICK;
3391 TERM(sublex_start());
3395 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3396 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3398 if (PL_expect == XOPERATOR)
3399 no_op("Backslash",s);
3403 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3443 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3445 /* Some keywords can be followed by any delimiter, including ':' */
3446 tmp = (len == 1 && strchr("msyq", PL_tokenbuf[0]) ||
3447 len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3448 (PL_tokenbuf[0] == 'q' &&
3449 strchr("qwxr", PL_tokenbuf[1]))));
3451 /* x::* is just a word, unless x is "CORE" */
3452 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3456 while (d < PL_bufend && isSPACE(*d))
3457 d++; /* no comments skipped here, or s### is misparsed */
3459 /* Is this a label? */
3460 if (!tmp && PL_expect == XSTATE
3461 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3463 yylval.pval = savepv(PL_tokenbuf);
3468 /* Check for keywords */
3469 tmp = keyword(PL_tokenbuf, len);
3471 /* Is this a word before a => operator? */
3472 if (strnEQ(d,"=>",2)) {
3474 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3475 yylval.opval->op_private = OPpCONST_BARE;
3479 if (tmp < 0) { /* second-class keyword? */
3480 GV *ogv = Nullgv; /* override (winner) */
3481 GV *hgv = Nullgv; /* hidden (loser) */
3482 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3484 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3487 if (GvIMPORTED_CV(gv))
3489 else if (! CvMETHOD(cv))
3493 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3494 (gv = *gvp) != (GV*)&PL_sv_undef &&
3495 GvCVu(gv) && GvIMPORTED_CV(gv))
3501 tmp = 0; /* overridden by import or by GLOBAL */
3504 && -tmp==KEY_lock /* XXX generalizable kludge */
3505 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3507 tmp = 0; /* any sub overrides "weak" keyword */
3509 else { /* no override */
3513 if (ckWARN(WARN_AMBIGUOUS) && hgv
3514 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3515 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3516 "Ambiguous call resolved as CORE::%s(), %s",
3517 GvENAME(hgv), "qualify as such or use &");
3524 default: /* not a keyword */
3527 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3529 /* Get the rest if it looks like a package qualifier */
3531 if (*s == '\'' || *s == ':' && s[1] == ':') {
3533 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3536 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3537 *s == '\'' ? "'" : "::");
3541 if (PL_expect == XOPERATOR) {
3542 if (PL_bufptr == PL_linestart) {
3543 CopLINE_dec(PL_curcop);
3544 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3545 CopLINE_inc(PL_curcop);
3548 no_op("Bareword",s);
3551 /* Look for a subroutine with this name in current package,
3552 unless name is "Foo::", in which case Foo is a bearword
3553 (and a package name). */
3556 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3558 if (ckWARN(WARN_UNSAFE) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3559 Perl_warner(aTHX_ WARN_UNSAFE,
3560 "Bareword \"%s\" refers to nonexistent package",
3563 PL_tokenbuf[len] = '\0';
3570 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3573 /* if we saw a global override before, get the right name */
3576 sv = newSVpvn("CORE::GLOBAL::",14);
3577 sv_catpv(sv,PL_tokenbuf);
3580 sv = newSVpv(PL_tokenbuf,0);
3582 /* Presume this is going to be a bareword of some sort. */
3585 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3586 yylval.opval->op_private = OPpCONST_BARE;
3588 /* And if "Foo::", then that's what it certainly is. */
3593 /* See if it's the indirect object for a list operator. */
3595 if (PL_oldoldbufptr &&
3596 PL_oldoldbufptr < PL_bufptr &&
3597 (PL_oldoldbufptr == PL_last_lop
3598 || PL_oldoldbufptr == PL_last_uni) &&
3599 /* NO SKIPSPACE BEFORE HERE! */
3600 (PL_expect == XREF ||
3601 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3603 bool immediate_paren = *s == '(';
3605 /* (Now we can afford to cross potential line boundary.) */
3608 /* Two barewords in a row may indicate method call. */
3610 if ((isIDFIRST_lazy(s) || *s == '$') && (tmp=intuit_method(s,gv)))
3613 /* If not a declared subroutine, it's an indirect object. */
3614 /* (But it's an indir obj regardless for sort.) */
3616 if ((PL_last_lop_op == OP_SORT ||
3617 (!immediate_paren && (!gv || !GvCVu(gv)))) &&
3618 (PL_last_lop_op != OP_MAPSTART &&
3619 PL_last_lop_op != OP_GREPSTART))
3621 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3626 /* If followed by a paren, it's certainly a subroutine. */
3628 PL_expect = XOPERATOR;
3632 if (gv && GvCVu(gv)) {
3633 for (d = s + 1; *d == ' ' || *d == '\t'; d++) ;
3634 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3639 PL_nextval[PL_nexttoke].opval = yylval.opval;
3640 PL_expect = XOPERATOR;
3646 /* If followed by var or block, call it a method (unless sub) */
3648 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3649 PL_last_lop = PL_oldbufptr;
3650 PL_last_lop_op = OP_METHOD;
3654 /* If followed by a bareword, see if it looks like indir obj. */
3656 if ((isIDFIRST_lazy(s) || *s == '$') && (tmp = intuit_method(s,gv)))
3659 /* Not a method, so call it a subroutine (if defined) */
3661 if (gv && GvCVu(gv)) {
3663 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
3664 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3665 "Ambiguous use of -%s resolved as -&%s()",
3666 PL_tokenbuf, PL_tokenbuf);
3667 /* Check for a constant sub */
3669 if ((sv = cv_const_sv(cv))) {
3671 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
3672 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
3673 yylval.opval->op_private = 0;
3677 /* Resolve to GV now. */
3678 op_free(yylval.opval);
3679 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
3680 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
3681 PL_last_lop = PL_oldbufptr;
3682 PL_last_lop_op = OP_ENTERSUB;
3683 /* Is there a prototype? */
3686 char *proto = SvPV((SV*)cv, len);
3689 if (strEQ(proto, "$"))
3691 if (*proto == '&' && *s == '{') {
3692 sv_setpv(PL_subname,"__ANON__");
3696 PL_nextval[PL_nexttoke].opval = yylval.opval;
3702 /* Call it a bare word */
3704 if (PL_hints & HINT_STRICT_SUBS)
3705 yylval.opval->op_private |= OPpCONST_STRICT;
3708 if (ckWARN(WARN_RESERVED)) {
3709 if (lastchar != '-') {
3710 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
3712 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
3719 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
3720 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3721 "Operator or semicolon missing before %c%s",
3722 lastchar, PL_tokenbuf);
3723 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3724 "Ambiguous use of %c resolved as operator %c",
3725 lastchar, lastchar);
3731 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3732 newSVsv(CopFILESV(PL_curcop)));
3736 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3737 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
3740 case KEY___PACKAGE__:
3741 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3743 ? newSVsv(PL_curstname)
3752 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
3753 char *pname = "main";
3754 if (PL_tokenbuf[2] == 'D')
3755 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
3756 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
3759 GvIOp(gv) = newIO();
3760 IoIFP(GvIOp(gv)) = PL_rsfp;
3761 #if defined(HAS_FCNTL) && defined(F_SETFD)
3763 int fd = PerlIO_fileno(PL_rsfp);
3764 fcntl(fd,F_SETFD,fd >= 3);
3767 /* Mark this internal pseudo-handle as clean */
3768 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
3770 IoTYPE(GvIOp(gv)) = '|';
3771 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
3772 IoTYPE(GvIOp(gv)) = '-';
3774 IoTYPE(GvIOp(gv)) = '<';
3786 if (PL_expect == XSTATE) {
3793 if (*s == ':' && s[1] == ':') {
3796 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3797 tmp = keyword(PL_tokenbuf, len);
3811 LOP(OP_ACCEPT,XTERM);
3817 LOP(OP_ATAN2,XTERM);
3826 LOP(OP_BLESS,XTERM);
3835 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
3852 if (!PL_cryptseen) {
3853 PL_cryptseen = TRUE;
3857 LOP(OP_CRYPT,XTERM);
3860 if (ckWARN(WARN_OCTAL)) {
3861 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
3862 if (*d != '0' && isDIGIT(*d))
3863 Perl_warner(aTHX_ WARN_OCTAL,
3864 "chmod: mode argument is missing initial 0");
3866 LOP(OP_CHMOD,XTERM);
3869 LOP(OP_CHOWN,XTERM);
3872 LOP(OP_CONNECT,XTERM);
3888 s = force_word(s,WORD,FALSE,TRUE,FALSE);
3892 PL_hints |= HINT_BLOCK_SCOPE;
3902 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
3903 LOP(OP_DBMOPEN,XTERM);
3909 s = force_word(s,WORD,TRUE,FALSE,FALSE);
3916 yylval.ival = CopLINE(PL_curcop);
3930 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
3931 UNIBRACK(OP_ENTEREVAL);
3946 case KEY_endhostent:
3952 case KEY_endservent:
3955 case KEY_endprotoent:
3966 yylval.ival = CopLINE(PL_curcop);
3968 if (PL_expect == XSTATE && isIDFIRST_lazy(s)) {
3970 if ((PL_bufend - p) >= 3 &&
3971 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
3973 else if ((PL_bufend - p) >= 4 &&
3974 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
3977 if (isIDFIRST_lazy(p)) {
3978 p = scan_ident(p, PL_bufend,
3979 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3983 Perl_croak(aTHX_ "Missing $ on loop variable");
3988 LOP(OP_FORMLINE,XTERM);
3994 LOP(OP_FCNTL,XTERM);
4000 LOP(OP_FLOCK,XTERM);
4009 LOP(OP_GREPSTART, *s == '(' ? XTERM : XREF);
4012 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4027 case KEY_getpriority:
4028 LOP(OP_GETPRIORITY,XTERM);
4030 case KEY_getprotobyname:
4033 case KEY_getprotobynumber:
4034 LOP(OP_GPBYNUMBER,XTERM);
4036 case KEY_getprotoent:
4048 case KEY_getpeername:
4049 UNI(OP_GETPEERNAME);
4051 case KEY_gethostbyname:
4054 case KEY_gethostbyaddr:
4055 LOP(OP_GHBYADDR,XTERM);
4057 case KEY_gethostent:
4060 case KEY_getnetbyname:
4063 case KEY_getnetbyaddr:
4064 LOP(OP_GNBYADDR,XTERM);
4069 case KEY_getservbyname:
4070 LOP(OP_GSBYNAME,XTERM);
4072 case KEY_getservbyport:
4073 LOP(OP_GSBYPORT,XTERM);
4075 case KEY_getservent:
4078 case KEY_getsockname:
4079 UNI(OP_GETSOCKNAME);
4081 case KEY_getsockopt:
4082 LOP(OP_GSOCKOPT,XTERM);
4104 yylval.ival = CopLINE(PL_curcop);
4108 LOP(OP_INDEX,XTERM);
4114 LOP(OP_IOCTL,XTERM);
4126 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4158 LOP(OP_LISTEN,XTERM);
4167 s = scan_pat(s,OP_MATCH);
4168 TERM(sublex_start());
4171 LOP(OP_MAPSTART, *s == '(' ? XTERM : XREF);
4174 LOP(OP_MKDIR,XTERM);
4177 LOP(OP_MSGCTL,XTERM);
4180 LOP(OP_MSGGET,XTERM);
4183 LOP(OP_MSGRCV,XTERM);
4186 LOP(OP_MSGSND,XTERM);
4192 if (isIDFIRST_lazy(s)) {
4193 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4194 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4196 PL_in_my_stash = gv_stashpv(PL_tokenbuf, FALSE);
4197 if (!PL_in_my_stash) {
4200 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4208 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4215 if (PL_expect != XSTATE)
4216 yyerror("\"no\" not allowed in expression");
4217 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4218 s = force_version(s);
4223 if (*s == '(' || (s = skipspace(s), *s == '('))
4230 if (isIDFIRST_lazy(s)) {
4232 for (d = s; isALNUM_lazy(d); d++) ;
4234 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_AMBIGUOUS))
4235 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4236 "Precedence problem: open %.*s should be open(%.*s)",
4242 yylval.ival = OP_OR;
4252 LOP(OP_OPEN_DIR,XTERM);
4255 checkcomma(s,PL_tokenbuf,"filehandle");
4259 checkcomma(s,PL_tokenbuf,"filehandle");
4278 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4282 LOP(OP_PIPE_OP,XTERM);
4285 s = scan_str(s,FALSE,FALSE);
4287 missingterm((char*)0);
4288 yylval.ival = OP_CONST;
4289 TERM(sublex_start());
4295 s = scan_str(s,FALSE,FALSE);
4297 missingterm((char*)0);
4299 if (SvCUR(PL_lex_stuff)) {
4302 d = SvPV_force(PL_lex_stuff, len);
4304 for (; isSPACE(*d) && len; --len, ++d) ;
4307 if (!warned && ckWARN(WARN_SYNTAX)) {
4308 for (; !isSPACE(*d) && len; --len, ++d) {
4310 Perl_warner(aTHX_ WARN_SYNTAX,
4311 "Possible attempt to separate words with commas");
4314 else if (*d == '#') {
4315 Perl_warner(aTHX_ WARN_SYNTAX,
4316 "Possible attempt to put comments in qw() list");
4322 for (; !isSPACE(*d) && len; --len, ++d) ;
4324 words = append_elem(OP_LIST, words,
4325 newSVOP(OP_CONST, 0, newSVpvn(b, d-b)));
4329 PL_nextval[PL_nexttoke].opval = words;
4334 SvREFCNT_dec(PL_lex_stuff);
4335 PL_lex_stuff = Nullsv;
4340 s = scan_str(s,FALSE,FALSE);
4342 missingterm((char*)0);
4343 yylval.ival = OP_STRINGIFY;
4344 if (SvIVX(PL_lex_stuff) == '\'')
4345 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4346 TERM(sublex_start());
4349 s = scan_pat(s,OP_QR);
4350 TERM(sublex_start());
4353 s = scan_str(s,FALSE,FALSE);
4355 missingterm((char*)0);
4356 yylval.ival = OP_BACKTICK;
4358 TERM(sublex_start());
4364 *PL_tokenbuf = '\0';
4365 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4366 if (isIDFIRST_lazy(PL_tokenbuf))
4367 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4369 yyerror("<> should be quotes");
4376 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4380 LOP(OP_RENAME,XTERM);
4389 LOP(OP_RINDEX,XTERM);
4412 LOP(OP_REVERSE,XTERM);
4423 TERM(sublex_start());