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);
32 #define XFAKEBRACK 128
35 #define UTF (PL_hints & HINT_UTF8)
37 * Note: we try to be careful never to call the isXXX_utf8() functions
38 * unless we're pretty sure we've seen the beginning of a UTF-8 character
39 * (that is, the two high bits are set). Otherwise we risk loading in the
40 * heavy-duty SWASHINIT and SWASHGET routines unnecessarily.
42 #define isIDFIRST_lazy(p) ((!UTF || (*((U8*)p) < 0xc0)) \
44 : isIDFIRST_utf8((U8*)p))
45 #define isALNUM_lazy(p) ((!UTF || (*((U8*)p) < 0xc0)) \
47 : isALNUM_utf8((U8*)p))
49 /* In variables name $^X, these are the legal values for X.
50 * 1999-02-27 mjd-perl-patch@plover.com */
51 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
53 /* LEX_* are values for PL_lex_state, the state of the lexer.
54 * They are arranged oddly so that the guard on the switch statement
55 * can get by with a single comparison (if the compiler is smart enough).
58 /* #define LEX_NOTPARSING 11 is done in perl.h. */
61 #define LEX_INTERPNORMAL 9
62 #define LEX_INTERPCASEMOD 8
63 #define LEX_INTERPPUSH 7
64 #define LEX_INTERPSTART 6
65 #define LEX_INTERPEND 5
66 #define LEX_INTERPENDMAYBE 4
67 #define LEX_INTERPCONCAT 3
68 #define LEX_INTERPCONST 2
69 #define LEX_FORMLINE 1
70 #define LEX_KNOWNEXT 0
79 /* XXX If this causes problems, set i_unistd=undef in the hint file. */
81 # include <unistd.h> /* Needed for execv() */
90 YYSTYPE* yylval_pointer = NULL;
91 int* yychar_pointer = NULL;
94 # define yylval (*yylval_pointer)
95 # define yychar (*yychar_pointer)
96 # define PERL_YYLEX_PARAM yylval_pointer,yychar_pointer
98 # define yylex() Perl_yylex(aTHX_ yylval_pointer, yychar_pointer)
101 #include "keywords.h"
103 /* CLINE is a macro that ensures PL_copline has a sane value */
108 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
111 * Convenience functions to return different tokens and prime the
112 * lexer for the next token. They all take an argument.
114 * TOKEN : generic token (used for '(', DOLSHARP, etc)
115 * OPERATOR : generic operator
116 * AOPERATOR : assignment operator
117 * PREBLOCK : beginning the block after an if, while, foreach, ...
118 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
119 * PREREF : *EXPR where EXPR is not a simple identifier
120 * TERM : expression term
121 * LOOPX : loop exiting command (goto, last, dump, etc)
122 * FTST : file test operator
123 * FUN0 : zero-argument function
124 * FUN1 : not used, except for not, which isn't a UNIOP
125 * BOop : bitwise or or xor
127 * SHop : shift operator
128 * PWop : power operator
129 * PMop : pattern-matching operator
130 * Aop : addition-level operator
131 * Mop : multiplication-level operator
132 * Eop : equality-testing operator
133 * Rop : relational operator <= != gt
135 * Also see LOP and lop() below.
138 #define TOKEN(retval) return (PL_bufptr = s,(int)retval)
139 #define OPERATOR(retval) return (PL_expect = XTERM,PL_bufptr = s,(int)retval)
140 #define AOPERATOR(retval) return ao((PL_expect = XTERM,PL_bufptr = s,(int)retval))
141 #define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s,(int)retval)
142 #define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s,(int)retval)
143 #define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s,(int)retval)
144 #define TERM(retval) return (CLINE, PL_expect = XOPERATOR,PL_bufptr = s,(int)retval)
145 #define LOOPX(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LOOPEX)
146 #define FTST(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)UNIOP)
147 #define FUN0(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC0)
148 #define FUN1(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC1)
149 #define BOop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITOROP))
150 #define BAop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITANDOP))
151 #define SHop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)SHIFTOP))
152 #define PWop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)POWOP))
153 #define PMop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MATCHOP)
154 #define Aop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)ADDOP))
155 #define Mop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MULOP))
156 #define Eop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)EQOP)
157 #define Rop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)RELOP)
159 /* This bit of chicanery makes a unary function followed by
160 * a parenthesis into a function with one argument, highest precedence.
162 #define UNI(f) return(yylval.ival = f, \
165 PL_last_uni = PL_oldbufptr, \
166 PL_last_lop_op = f, \
167 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
169 #define UNIBRACK(f) return(yylval.ival = f, \
171 PL_last_uni = PL_oldbufptr, \
172 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
174 /* grandfather return to old style */
175 #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
180 * This subroutine detects &&= and ||= and turns an ANDAND or OROR
181 * into an OP_ANDASSIGN or OP_ORASSIGN
185 S_ao(pTHX_ int toketype)
187 if (*PL_bufptr == '=') {
189 if (toketype == ANDAND)
190 yylval.ival = OP_ANDASSIGN;
191 else if (toketype == OROR)
192 yylval.ival = OP_ORASSIGN;
200 * When Perl expects an operator and finds something else, no_op
201 * prints the warning. It always prints "<something> found where
202 * operator expected. It prints "Missing semicolon on previous line?"
203 * if the surprise occurs at the start of the line. "do you need to
204 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
205 * where the compiler doesn't know if foo is a method call or a function.
206 * It prints "Missing operator before end of line" if there's nothing
207 * after the missing operator, or "... before <...>" if there is something
208 * after the missing operator.
212 S_no_op(pTHX_ char *what, char *s)
214 char *oldbp = PL_bufptr;
215 bool is_first = (PL_oldbufptr == PL_linestart);
223 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
225 Perl_warn(aTHX_ "\t(Missing semicolon on previous line?)\n");
226 else if (PL_oldoldbufptr && isIDFIRST_lazy(PL_oldoldbufptr)) {
228 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy(t) || *t == ':'); t++) ;
229 if (t < PL_bufptr && isSPACE(*t))
230 Perl_warn(aTHX_ "\t(Do you need to predeclare %.*s?)\n",
231 t - PL_oldoldbufptr, PL_oldoldbufptr);
234 Perl_warn(aTHX_ "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp);
240 * Complain about missing quote/regexp/heredoc terminator.
241 * If it's called with (char *)NULL then it cauterizes the line buffer.
242 * If we're in a delimited string and the delimiter is a control
243 * character, it's reformatted into a two-char sequence like ^C.
248 S_missingterm(pTHX_ char *s)
253 char *nl = strrchr(s,'\n');
259 iscntrl(PL_multi_close)
261 PL_multi_close < 32 || PL_multi_close == 127
265 tmpbuf[1] = toCTRL(PL_multi_close);
271 *tmpbuf = PL_multi_close;
275 q = strchr(s,'"') ? '\'' : '"';
276 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
284 Perl_deprecate(pTHX_ char *s)
287 if (ckWARN(WARN_DEPRECATED))
288 Perl_warner(aTHX_ WARN_DEPRECATED, "Use of %s is deprecated", s);
293 * Deprecate a comma-less variable list.
299 deprecate("comma-less variable list");
303 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
304 * utf16-to-utf8-reversed.
307 #ifdef PERL_CR_FILTER
311 register char *s = SvPVX(sv);
312 register char *e = s + SvCUR(sv);
313 /* outer loop optimized to do nothing if there are no CR-LFs */
315 if (*s++ == '\r' && *s == '\n') {
316 /* hit a CR-LF, need to copy the rest */
317 register char *d = s - 1;
320 if (*s == '\r' && s[1] == '\n')
331 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
333 I32 count = FILTER_READ(idx+1, sv, maxlen);
334 if (count > 0 && !maxlen)
341 S_utf16_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((U16*)SvPVX(sv), tmps, SvCUR(sv));
349 sv_usepvn(sv, (char*)tmps, tend - tmps);
356 S_utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
358 I32 count = FILTER_READ(idx+1, sv, maxlen);
362 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
363 tend = utf16_to_utf8_reversed((U16*)SvPVX(sv), tmps, SvCUR(sv));
364 sv_usepvn(sv, (char*)tmps, tend - tmps);
372 * Initialize variables. Uses the Perl save_stack to save its state (for
373 * recursive calls to the parser).
377 Perl_lex_start(pTHX_ SV *line)
383 SAVEI32(PL_lex_dojoin);
384 SAVEI32(PL_lex_brackets);
385 SAVEI32(PL_lex_casemods);
386 SAVEI32(PL_lex_starts);
387 SAVEI32(PL_lex_state);
388 SAVEVPTR(PL_lex_inpat);
389 SAVEI32(PL_lex_inwhat);
390 SAVECOPLINE(PL_curcop);
393 SAVEPPTR(PL_oldbufptr);
394 SAVEPPTR(PL_oldoldbufptr);
395 SAVEPPTR(PL_linestart);
396 SAVESPTR(PL_linestr);
397 SAVEPPTR(PL_lex_brackstack);
398 SAVEPPTR(PL_lex_casestack);
399 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
400 SAVESPTR(PL_lex_stuff);
401 SAVEI32(PL_lex_defer);
402 SAVEI32(PL_sublex_info.sub_inwhat);
403 SAVESPTR(PL_lex_repl);
405 SAVEINT(PL_lex_expect);
407 PL_lex_state = LEX_NORMAL;
411 New(899, PL_lex_brackstack, 120, char);
412 New(899, PL_lex_casestack, 12, char);
413 SAVEFREEPV(PL_lex_brackstack);
414 SAVEFREEPV(PL_lex_casestack);
416 *PL_lex_casestack = '\0';
419 PL_lex_stuff = Nullsv;
420 PL_lex_repl = Nullsv;
423 PL_sublex_info.sub_inwhat = 0;
425 if (SvREADONLY(PL_linestr))
426 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
427 s = SvPV(PL_linestr, len);
428 if (len && s[len-1] != ';') {
429 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
430 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
431 sv_catpvn(PL_linestr, "\n;", 2);
433 SvTEMP_off(PL_linestr);
434 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
435 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
437 PL_rs = newSVpvn("\n", 1);
443 * Finalizer for lexing operations. Must be called when the parser is
444 * done with the lexer.
450 PL_doextract = FALSE;
455 * This subroutine has nothing to do with tilting, whether at windmills
456 * or pinball tables. Its name is short for "increment line". It
457 * increments the current line number in CopLINE(PL_curcop) and checks
458 * to see whether the line starts with a comment of the form
459 * # line 500 "foo.pm"
460 * If so, it sets the current line number and file to the values in the comment.
464 S_incline(pTHX_ char *s)
472 CopLINE_inc(PL_curcop);
475 while (*s == ' ' || *s == '\t') s++;
476 if (strnEQ(s, "line ", 5)) {
485 while (*s == ' ' || *s == '\t')
487 if (*s == '"' && (t = strchr(s+1, '"')))
491 return; /* false alarm */
492 for (t = s; !isSPACE(*t); t++) ;
497 CopFILE_set(PL_curcop, s);
499 CopLINE_set(PL_curcop, atoi(n)-1);
504 * Called to gobble the appropriate amount and type of whitespace.
505 * Skips comments as well.
509 S_skipspace(pTHX_ register char *s)
512 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
513 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
519 SSize_t oldprevlen, oldoldprevlen;
520 SSize_t oldloplen, oldunilen;
521 while (s < PL_bufend && isSPACE(*s)) {
522 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
527 if (s < PL_bufend && *s == '#') {
528 while (s < PL_bufend && *s != '\n')
532 if (PL_in_eval && !PL_rsfp) {
539 /* only continue to recharge the buffer if we're at the end
540 * of the buffer, we're not reading from a source filter, and
541 * we're in normal lexing mode
543 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
544 PL_lex_state == LEX_FORMLINE)
547 /* try to recharge the buffer */
548 if ((s = filter_gets(PL_linestr, PL_rsfp,
549 (prevlen = SvCUR(PL_linestr)))) == Nullch)
551 /* end of file. Add on the -p or -n magic */
552 if (PL_minus_n || PL_minus_p) {
553 sv_setpv(PL_linestr,PL_minus_p ?
554 ";}continue{print or die qq(-p destination: $!\\n)" :
556 sv_catpv(PL_linestr,";}");
557 PL_minus_n = PL_minus_p = 0;
560 sv_setpv(PL_linestr,";");
562 /* reset variables for next time we lex */
563 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
565 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
567 /* Close the filehandle. Could be from -P preprocessor,
568 * STDIN, or a regular file. If we were reading code from
569 * STDIN (because the commandline held no -e or filename)
570 * then we don't close it, we reset it so the code can
571 * read from STDIN too.
574 if (PL_preprocess && !PL_in_eval)
575 (void)PerlProc_pclose(PL_rsfp);
576 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
577 PerlIO_clearerr(PL_rsfp);
579 (void)PerlIO_close(PL_rsfp);
584 /* not at end of file, so we only read another line */
585 /* make corresponding updates to old pointers, for yyerror() */
586 oldprevlen = PL_oldbufptr - PL_bufend;
587 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
589 oldunilen = PL_last_uni - PL_bufend;
591 oldloplen = PL_last_lop - PL_bufend;
592 PL_linestart = PL_bufptr = s + prevlen;
593 PL_bufend = s + SvCUR(PL_linestr);
595 PL_oldbufptr = s + oldprevlen;
596 PL_oldoldbufptr = s + oldoldprevlen;
598 PL_last_uni = s + oldunilen;
600 PL_last_lop = s + oldloplen;
603 /* debugger active and we're not compiling the debugger code,
604 * so store the line into the debugger's array of lines
606 if (PERLDB_LINE && PL_curstash != PL_debstash) {
607 SV *sv = NEWSV(85,0);
609 sv_upgrade(sv, SVt_PVMG);
610 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
611 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
618 * Check the unary operators to ensure there's no ambiguity in how they're
619 * used. An ambiguous piece of code would be:
621 * This doesn't mean rand() + 5. Because rand() is a unary operator,
622 * the +5 is its argument.
632 if (PL_oldoldbufptr != PL_last_uni)
634 while (isSPACE(*PL_last_uni))
636 for (s = PL_last_uni; isALNUM_lazy(s) || *s == '-'; s++) ;
637 if ((t = strchr(s, '(')) && t < PL_bufptr)
639 if (ckWARN_d(WARN_AMBIGUOUS)){
642 Perl_warner(aTHX_ WARN_AMBIGUOUS,
643 "Warning: Use of \"%s\" without parens is ambiguous",
649 /* workaround to replace the UNI() macro with a function. Only the
650 * hints/uts.sh file mentions this. Other comments elsewhere in the
651 * source indicate Microport Unix might need it too.
657 #define UNI(f) return uni(f,s)
660 S_uni(pTHX_ I32 f, char *s)
665 PL_last_uni = PL_oldbufptr;
676 #endif /* CRIPPLED_CC */
679 * LOP : macro to build a list operator. Its behaviour has been replaced
680 * with a subroutine, S_lop() for which LOP is just another name.
683 #define LOP(f,x) return lop(f,x,s)
687 * Build a list operator (or something that might be one). The rules:
688 * - if we have a next token, then it's a list operator [why?]
689 * - if the next thing is an opening paren, then it's a function
690 * - else it's a list operator
694 S_lop(pTHX_ I32 f, int x, char *s)
701 PL_last_lop = PL_oldbufptr;
716 * When the lexer realizes it knows the next token (for instance,
717 * it is reordering tokens for the parser) then it can call S_force_next
718 * to know what token to return the next time the lexer is called. Caller
719 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
720 * handles the token correctly.
724 S_force_next(pTHX_ I32 type)
726 PL_nexttype[PL_nexttoke] = type;
728 if (PL_lex_state != LEX_KNOWNEXT) {
729 PL_lex_defer = PL_lex_state;
730 PL_lex_expect = PL_expect;
731 PL_lex_state = LEX_KNOWNEXT;
737 * When the lexer knows the next thing is a word (for instance, it has
738 * just seen -> and it knows that the next char is a word char, then
739 * it calls S_force_word to stick the next word into the PL_next lookahead.
742 * char *start : buffer position (must be within PL_linestr)
743 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
744 * int check_keyword : if true, Perl checks to make sure the word isn't
745 * a keyword (do this if the word is a label, e.g. goto FOO)
746 * int allow_pack : if true, : characters will also be allowed (require,
748 * int allow_initial_tick : used by the "sub" lexer only.
752 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
757 start = skipspace(start);
759 if (isIDFIRST_lazy(s) ||
760 (allow_pack && *s == ':') ||
761 (allow_initial_tick && *s == '\'') )
763 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
764 if (check_keyword && keyword(PL_tokenbuf, len))
766 if (token == METHOD) {
771 PL_expect = XOPERATOR;
774 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0));
775 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
783 * Called when the lexer wants $foo *foo &foo etc, but the program
784 * text only contains the "foo" portion. The first argument is a pointer
785 * to the "foo", and the second argument is the type symbol to prefix.
786 * Forces the next token to be a "WORD".
787 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
791 S_force_ident(pTHX_ register char *s, int kind)
794 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
795 PL_nextval[PL_nexttoke].opval = o;
798 dTHR; /* just for in_eval */
799 o->op_private = OPpCONST_ENTERED;
800 /* XXX see note in pp_entereval() for why we forgo typo
801 warnings if the symbol must be introduced in an eval.
803 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
804 kind == '$' ? SVt_PV :
805 kind == '@' ? SVt_PVAV :
806 kind == '%' ? SVt_PVHV :
815 * Forces the next token to be a version number.
819 S_force_version(pTHX_ char *s)
821 OP *version = Nullop;
825 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
829 for (; isDIGIT(*d) || *d == '_' || *d == '.'; d++);
830 if ((*d == ';' || isSPACE(*d)) && *(skipspace(d)) != ',') {
832 /* real VERSION number -- GBARR */
833 version = yylval.opval;
837 /* NOTE: The parser sees the package name and the VERSION swapped */
838 PL_nextval[PL_nexttoke].opval = version;
846 * Tokenize a quoted string passed in as an SV. It finds the next
847 * chunk, up to end of string or a backslash. It may make a new
848 * SV containing that chunk (if HINT_NEW_STRING is on). It also
853 S_tokeq(pTHX_ SV *sv)
864 s = SvPV_force(sv, len);
868 while (s < send && *s != '\\')
873 if ( PL_hints & HINT_NEW_STRING )
874 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
877 if (s + 1 < send && (s[1] == '\\'))
878 s++; /* all that, just for this */
883 SvCUR_set(sv, d - SvPVX(sv));
885 if ( PL_hints & HINT_NEW_STRING )
886 return new_constant(NULL, 0, "q", sv, pv, "q");
891 * Now come three functions related to double-quote context,
892 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
893 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
894 * interact with PL_lex_state, and create fake ( ... ) argument lists
895 * to handle functions and concatenation.
896 * They assume that whoever calls them will be setting up a fake
897 * join call, because each subthing puts a ',' after it. This lets
900 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
902 * (I'm not sure whether the spurious commas at the end of lcfirst's
903 * arguments and join's arguments are created or not).
908 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
910 * Pattern matching will set PL_lex_op to the pattern-matching op to
911 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
913 * OP_CONST and OP_READLINE are easy--just make the new op and return.
915 * Everything else becomes a FUNC.
917 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
918 * had an OP_CONST or OP_READLINE). This just sets us up for a
919 * call to S_sublex_push().
925 register I32 op_type = yylval.ival;
927 if (op_type == OP_NULL) {
928 yylval.opval = PL_lex_op;
932 if (op_type == OP_CONST || op_type == OP_READLINE) {
933 SV *sv = tokeq(PL_lex_stuff);
935 if (SvTYPE(sv) == SVt_PVIV) {
936 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
942 nsv = newSVpvn(p, len);
946 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
947 PL_lex_stuff = Nullsv;
951 PL_sublex_info.super_state = PL_lex_state;
952 PL_sublex_info.sub_inwhat = op_type;
953 PL_sublex_info.sub_op = PL_lex_op;
954 PL_lex_state = LEX_INTERPPUSH;
958 yylval.opval = PL_lex_op;
968 * Create a new scope to save the lexing state. The scope will be
969 * ended in S_sublex_done. Returns a '(', starting the function arguments
970 * to the uc, lc, etc. found before.
971 * Sets PL_lex_state to LEX_INTERPCONCAT.
980 PL_lex_state = PL_sublex_info.super_state;
981 SAVEI32(PL_lex_dojoin);
982 SAVEI32(PL_lex_brackets);
983 SAVEI32(PL_lex_casemods);
984 SAVEI32(PL_lex_starts);
985 SAVEI32(PL_lex_state);
986 SAVEVPTR(PL_lex_inpat);
987 SAVEI32(PL_lex_inwhat);
988 SAVECOPLINE(PL_curcop);
990 SAVEPPTR(PL_oldbufptr);
991 SAVEPPTR(PL_oldoldbufptr);
992 SAVEPPTR(PL_linestart);
993 SAVESPTR(PL_linestr);
994 SAVEPPTR(PL_lex_brackstack);
995 SAVEPPTR(PL_lex_casestack);
997 PL_linestr = PL_lex_stuff;
998 PL_lex_stuff = Nullsv;
1000 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1001 = SvPVX(PL_linestr);
1002 PL_bufend += SvCUR(PL_linestr);
1003 SAVEFREESV(PL_linestr);
1005 PL_lex_dojoin = FALSE;
1006 PL_lex_brackets = 0;
1007 New(899, PL_lex_brackstack, 120, char);
1008 New(899, PL_lex_casestack, 12, char);
1009 SAVEFREEPV(PL_lex_brackstack);
1010 SAVEFREEPV(PL_lex_casestack);
1011 PL_lex_casemods = 0;
1012 *PL_lex_casestack = '\0';
1014 PL_lex_state = LEX_INTERPCONCAT;
1015 CopLINE_set(PL_curcop, PL_multi_start);
1017 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1018 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1019 PL_lex_inpat = PL_sublex_info.sub_op;
1021 PL_lex_inpat = Nullop;
1028 * Restores lexer state after a S_sublex_push.
1034 if (!PL_lex_starts++) {
1035 PL_expect = XOPERATOR;
1036 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpvn("",0));
1040 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1041 PL_lex_state = LEX_INTERPCASEMOD;
1045 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1046 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1047 PL_linestr = PL_lex_repl;
1049 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1050 PL_bufend += SvCUR(PL_linestr);
1051 SAVEFREESV(PL_linestr);
1052 PL_lex_dojoin = FALSE;
1053 PL_lex_brackets = 0;
1054 PL_lex_casemods = 0;
1055 *PL_lex_casestack = '\0';
1057 if (SvEVALED(PL_lex_repl)) {
1058 PL_lex_state = LEX_INTERPNORMAL;
1060 /* we don't clear PL_lex_repl here, so that we can check later
1061 whether this is an evalled subst; that means we rely on the
1062 logic to ensure sublex_done() is called again only via the
1063 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1066 PL_lex_state = LEX_INTERPCONCAT;
1067 PL_lex_repl = Nullsv;
1073 PL_bufend = SvPVX(PL_linestr);
1074 PL_bufend += SvCUR(PL_linestr);
1075 PL_expect = XOPERATOR;
1076 PL_sublex_info.sub_inwhat = 0;
1084 Extracts a pattern, double-quoted string, or transliteration. This
1087 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1088 processing a pattern (PL_lex_inpat is true), a transliteration
1089 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1091 Returns a pointer to the character scanned up to. Iff this is
1092 advanced from the start pointer supplied (ie if anything was
1093 successfully parsed), will leave an OP for the substring scanned
1094 in yylval. Caller must intuit reason for not parsing further
1095 by looking at the next characters herself.
1099 double-quoted style: \r and \n
1100 regexp special ones: \D \s
1102 backrefs: \1 (deprecated in substitution replacements)
1103 case and quoting: \U \Q \E
1104 stops on @ and $, but not for $ as tail anchor
1106 In transliterations:
1107 characters are VERY literal, except for - not at the start or end
1108 of the string, which indicates a range. scan_const expands the
1109 range to the full set of intermediate characters.
1111 In double-quoted strings:
1113 double-quoted style: \r and \n
1115 backrefs: \1 (deprecated)
1116 case and quoting: \U \Q \E
1119 scan_const does *not* construct ops to handle interpolated strings.
1120 It stops processing as soon as it finds an embedded $ or @ variable
1121 and leaves it to the caller to work out what's going on.
1123 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @:foo.
1125 $ in pattern could be $foo or could be tail anchor. Assumption:
1126 it's a tail anchor if $ is the last thing in the string, or if it's
1127 followed by one of ")| \n\t"
1129 \1 (backreferences) are turned into $1
1131 The structure of the code is
1132 while (there's a character to process) {
1133 handle transliteration ranges
1134 skip regexp comments
1135 skip # initiated comments in //x patterns
1136 check for embedded @foo
1137 check for embedded scalars
1139 leave intact backslashes from leave (below)
1140 deprecate \1 in strings and sub replacements
1141 handle string-changing backslashes \l \U \Q \E, etc.
1142 switch (what was escaped) {
1143 handle - in a transliteration (becomes a literal -)
1144 handle \132 octal characters
1145 handle 0x15 hex characters
1146 handle \cV (control V)
1147 handle printf backslashes (\f, \r, \n, etc)
1149 } (end if backslash)
1150 } (end while character to read)
1155 S_scan_const(pTHX_ char *start)
1157 register char *send = PL_bufend; /* end of the constant */
1158 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1159 register char *s = start; /* start of the constant */
1160 register char *d = SvPVX(sv); /* destination for copies */
1161 bool dorange = FALSE; /* are we in a translit range? */
1163 I32 utf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1164 ? (PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
1166 I32 thisutf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1167 ? (PL_sublex_info.sub_op->op_private & (PL_lex_repl ?
1168 OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF))
1170 const char *leaveit = /* set of acceptably-backslashed characters */
1172 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
1175 while (s < send || dorange) {
1176 /* get transliterations out of the way (they're most literal) */
1177 if (PL_lex_inwhat == OP_TRANS) {
1178 /* expand a range A-Z to the full set of characters. AIE! */
1180 I32 i; /* current expanded character */
1181 I32 min; /* first character in range */
1182 I32 max; /* last character in range */
1184 i = d - SvPVX(sv); /* remember current offset */
1185 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1186 d = SvPVX(sv) + i; /* refresh d after realloc */
1187 d -= 2; /* eat the first char and the - */
1189 min = (U8)*d; /* first char in range */
1190 max = (U8)d[1]; /* last char in range */
1193 if ((isLOWER(min) && isLOWER(max)) ||
1194 (isUPPER(min) && isUPPER(max))) {
1196 for (i = min; i <= max; i++)
1200 for (i = min; i <= max; i++)
1207 for (i = min; i <= max; i++)
1210 /* mark the range as done, and continue */
1215 /* range begins (ignore - as first or last char) */
1216 else if (*s == '-' && s+1 < send && s != start) {
1218 *d++ = (char)0xff; /* use illegal utf8 byte--see pmtrans */
1227 /* if we get here, we're not doing a transliteration */
1229 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1230 except for the last char, which will be done separately. */
1231 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1233 while (s < send && *s != ')')
1235 } else if (s[2] == '{'
1236 || s[2] == 'p' && s[3] == '{') { /* This should march regcomp.c */
1238 char *regparse = s + (s[2] == '{' ? 3 : 4);
1241 while (count && (c = *regparse)) {
1242 if (c == '\\' && regparse[1])
1250 if (*regparse != ')') {
1251 regparse--; /* Leave one char for continuation. */
1252 yyerror("Sequence (?{...}) not terminated or not {}-balanced");
1254 while (s < regparse)
1259 /* likewise skip #-initiated comments in //x patterns */
1260 else if (*s == '#' && PL_lex_inpat &&
1261 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1262 while (s+1 < send && *s != '\n')
1266 /* check for embedded arrays (@foo, @:foo, @'foo, @{foo}, @$foo) */
1267 else if (*s == '@' && s[1] && (isALNUM_lazy(s+1) || strchr(":'{$", s[1])))
1270 /* check for embedded scalars. only stop if we're sure it's a
1273 else if (*s == '$') {
1274 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1276 if (s + 1 < send && !strchr("()| \n\t", s[1]))
1277 break; /* in regexp, $ might be tail anchor */
1280 /* (now in tr/// code again) */
1282 if (*s & 0x80 && thisutf) {
1283 dTHR; /* only for ckWARN */
1284 if (ckWARN(WARN_UTF8)) {
1285 (void)utf8_to_uv((U8*)s, &len); /* could cvt latin-1 to utf8 here... */
1295 if (*s == '\\' && s+1 < send) {
1298 /* some backslashes we leave behind */
1299 if (*leaveit && *s && strchr(leaveit, *s)) {
1305 /* deprecate \1 in strings and substitution replacements */
1306 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1307 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1309 dTHR; /* only for ckWARN */
1310 if (ckWARN(WARN_SYNTAX))
1311 Perl_warner(aTHX_ WARN_SYNTAX, "\\%c better written as $%c", *s, *s);
1316 /* string-change backslash escapes */
1317 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1322 /* if we get here, it's either a quoted -, or a digit */
1325 /* quoted - in transliterations */
1327 if (PL_lex_inwhat == OP_TRANS) {
1335 if (ckWARN(WARN_UNSAFE) && isALPHA(*s))
1336 Perl_warner(aTHX_ WARN_UNSAFE,
1337 "Unrecognized escape \\%c passed through",
1339 /* default action is to copy the quoted character */
1344 /* \132 indicates an octal constant */
1345 case '0': case '1': case '2': case '3':
1346 case '4': case '5': case '6': case '7':
1347 *d++ = (char)scan_oct(s, 3, &len);
1351 /* \x24 indicates a hex constant */
1355 char* e = strchr(s, '}');
1358 yyerror("Missing right brace on \\x{}");
1363 if (ckWARN(WARN_UTF8))
1364 Perl_warner(aTHX_ WARN_UTF8,
1365 "Use of \\x{} without utf8 declaration");
1367 /* note: utf always shorter than hex */
1368 d = (char*)uv_to_utf8((U8*)d,
1369 (UV)scan_hex(s + 1, e - s - 1, &len));
1373 UV uv = (UV)scan_hex(s, 2, &len);
1374 if (utf && PL_lex_inwhat == OP_TRANS &&
1375 utf != (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
1377 d = (char*)uv_to_utf8((U8*)d, uv); /* doing a CU or UC */
1380 if (uv >= 127 && UTF) {
1382 if (ckWARN(WARN_UTF8))
1383 Perl_warner(aTHX_ WARN_UTF8,
1384 "\\x%.*s will produce malformed UTF-8 character; use \\x{%.*s} for that",
1385 (int)len,s,(int)len,s);
1393 /* \N{latin small letter a} is a named character */
1397 char* e = strchr(s, '}');
1406 yyerror("Missing right brace on \\N{}");
1410 res = newSVpvn(s + 1, e - s - 1);
1411 res = new_constant( Nullch, 0, "charnames",
1412 res, Nullsv, "\\N{...}" );
1413 str = SvPV(res,len);
1414 if (len > e - s + 4) {
1415 char *odest = SvPVX(sv);
1417 SvGROW(sv, (SvCUR(sv) + len - (e - s + 4)));
1418 d = SvPVX(sv) + (d - odest);
1420 Copy(str, d, len, char);
1427 yyerror("Missing braces on \\N{}");
1430 /* \c is a control character */
1444 /* printf-style backslashes, formfeeds, newlines, etc */
1462 *d++ = '\047'; /* CP 1047 */
1465 *d++ = '\057'; /* CP 1047 */
1479 } /* end if (backslash) */
1482 } /* while loop to process each character */
1484 /* terminate the string and set up the sv */
1486 SvCUR_set(sv, d - SvPVX(sv));
1489 /* shrink the sv if we allocated more than we used */
1490 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1491 SvLEN_set(sv, SvCUR(sv) + 1);
1492 Renew(SvPVX(sv), SvLEN(sv), char);
1495 /* return the substring (via yylval) only if we parsed anything */
1496 if (s > PL_bufptr) {
1497 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1498 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1500 ( PL_lex_inwhat == OP_TRANS
1502 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1505 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1512 * Returns TRUE if there's more to the expression (e.g., a subscript),
1515 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1517 * ->[ and ->{ return TRUE
1518 * { and [ outside a pattern are always subscripts, so return TRUE
1519 * if we're outside a pattern and it's not { or [, then return FALSE
1520 * if we're in a pattern and the first char is a {
1521 * {4,5} (any digits around the comma) returns FALSE
1522 * if we're in a pattern and the first char is a [
1524 * [SOMETHING] has a funky algorithm to decide whether it's a
1525 * character class or not. It has to deal with things like
1526 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1527 * anything else returns TRUE
1530 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1533 S_intuit_more(pTHX_ register char *s)
1535 if (PL_lex_brackets)
1537 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1539 if (*s != '{' && *s != '[')
1544 /* In a pattern, so maybe we have {n,m}. */
1561 /* On the other hand, maybe we have a character class */
1564 if (*s == ']' || *s == '^')
1567 /* this is terrifying, and it works */
1568 int weight = 2; /* let's weigh the evidence */
1570 unsigned char un_char = 255, last_un_char;
1571 char *send = strchr(s,']');
1572 char tmpbuf[sizeof PL_tokenbuf * 4];
1574 if (!send) /* has to be an expression */
1577 Zero(seen,256,char);
1580 else if (isDIGIT(*s)) {
1582 if (isDIGIT(s[1]) && s[2] == ']')
1588 for (; s < send; s++) {
1589 last_un_char = un_char;
1590 un_char = (unsigned char)*s;
1595 weight -= seen[un_char] * 10;
1596 if (isALNUM_lazy(s+1)) {
1597 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1598 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1603 else if (*s == '$' && s[1] &&
1604 strchr("[#!%*<>()-=",s[1])) {
1605 if (/*{*/ strchr("])} =",s[2]))
1614 if (strchr("wds]",s[1]))
1616 else if (seen['\''] || seen['"'])
1618 else if (strchr("rnftbxcav",s[1]))
1620 else if (isDIGIT(s[1])) {
1622 while (s[1] && isDIGIT(s[1]))
1632 if (strchr("aA01! ",last_un_char))
1634 if (strchr("zZ79~",s[1]))
1636 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1637 weight -= 5; /* cope with negative subscript */
1640 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1641 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1646 if (keyword(tmpbuf, d - tmpbuf))
1649 if (un_char == last_un_char + 1)
1651 weight -= seen[un_char];
1656 if (weight >= 0) /* probably a character class */
1666 * Does all the checking to disambiguate
1668 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1669 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1671 * First argument is the stuff after the first token, e.g. "bar".
1673 * Not a method if bar is a filehandle.
1674 * Not a method if foo is a subroutine prototyped to take a filehandle.
1675 * Not a method if it's really "Foo $bar"
1676 * Method if it's "foo $bar"
1677 * Not a method if it's really "print foo $bar"
1678 * Method if it's really "foo package::" (interpreted as package->foo)
1679 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1680 * Not a method if bar is a filehandle or package, but is quotd with
1685 S_intuit_method(pTHX_ char *start, GV *gv)
1687 char *s = start + (*start == '$');
1688 char tmpbuf[sizeof PL_tokenbuf];
1696 if ((cv = GvCVu(gv))) {
1697 char *proto = SvPVX(cv);
1707 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1708 /* start is the beginning of the possible filehandle/object,
1709 * and s is the end of it
1710 * tmpbuf is a copy of it
1713 if (*start == '$') {
1714 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1719 return *s == '(' ? FUNCMETH : METHOD;
1721 if (!keyword(tmpbuf, len)) {
1722 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1727 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1728 if (indirgv && GvCVu(indirgv))
1730 /* filehandle or package name makes it a method */
1731 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1733 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1734 return 0; /* no assumptions -- "=>" quotes bearword */
1736 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1737 newSVpvn(tmpbuf,len));
1738 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1742 return *s == '(' ? FUNCMETH : METHOD;
1750 * Return a string of Perl code to load the debugger. If PERL5DB
1751 * is set, it will return the contents of that, otherwise a
1752 * compile-time require of perl5db.pl.
1759 char *pdb = PerlEnv_getenv("PERL5DB");
1763 SETERRNO(0,SS$_NORMAL);
1764 return "BEGIN { require 'perl5db.pl' }";
1770 /* Encoded script support. filter_add() effectively inserts a
1771 * 'pre-processing' function into the current source input stream.
1772 * Note that the filter function only applies to the current source file
1773 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1775 * The datasv parameter (which may be NULL) can be used to pass
1776 * private data to this instance of the filter. The filter function
1777 * can recover the SV using the FILTER_DATA macro and use it to
1778 * store private buffers and state information.
1780 * The supplied datasv parameter is upgraded to a PVIO type
1781 * and the IoDIRP field is used to store the function pointer,
1782 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1783 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1784 * private use must be set using malloc'd pointers.
1788 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1793 if (!PL_rsfp_filters)
1794 PL_rsfp_filters = newAV();
1796 datasv = NEWSV(255,0);
1797 if (!SvUPGRADE(datasv, SVt_PVIO))
1798 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1799 IoDIRP(datasv) = (DIR*)funcp; /* stash funcp into spare field */
1800 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
1801 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1802 funcp, SvPV_nolen(datasv)));
1803 av_unshift(PL_rsfp_filters, 1);
1804 av_store(PL_rsfp_filters, 0, datasv) ;
1809 /* Delete most recently added instance of this filter function. */
1811 Perl_filter_del(pTHX_ filter_t funcp)
1814 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp));
1815 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
1817 /* if filter is on top of stack (usual case) just pop it off */
1818 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
1819 if (IoDIRP(datasv) == (DIR*)funcp) {
1820 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
1821 IoDIRP(datasv) = (DIR*)NULL;
1822 sv_free(av_pop(PL_rsfp_filters));
1826 /* we need to search for the correct entry and clear it */
1827 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
1831 /* Invoke the n'th filter function for the current rsfp. */
1833 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
1836 /* 0 = read one text line */
1841 if (!PL_rsfp_filters)
1843 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
1844 /* Provide a default input filter to make life easy. */
1845 /* Note that we append to the line. This is handy. */
1846 DEBUG_P(PerlIO_printf(Perl_debug_log,
1847 "filter_read %d: from rsfp\n", idx));
1851 int old_len = SvCUR(buf_sv) ;
1853 /* ensure buf_sv is large enough */
1854 SvGROW(buf_sv, old_len + maxlen) ;
1855 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
1856 if (PerlIO_error(PL_rsfp))
1857 return -1; /* error */
1859 return 0 ; /* end of file */
1861 SvCUR_set(buf_sv, old_len + len) ;
1864 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
1865 if (PerlIO_error(PL_rsfp))
1866 return -1; /* error */
1868 return 0 ; /* end of file */
1871 return SvCUR(buf_sv);
1873 /* Skip this filter slot if filter has been deleted */
1874 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
1875 DEBUG_P(PerlIO_printf(Perl_debug_log,
1876 "filter_read %d: skipped (filter deleted)\n",
1878 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
1880 /* Get function pointer hidden within datasv */
1881 funcp = (filter_t)IoDIRP(datasv);
1882 DEBUG_P(PerlIO_printf(Perl_debug_log,
1883 "filter_read %d: via function %p (%s)\n",
1884 idx, funcp, SvPV_nolen(datasv)));
1885 /* Call function. The function is expected to */
1886 /* call "FILTER_READ(idx+1, buf_sv)" first. */
1887 /* Return: <0:error, =0:eof, >0:not eof */
1888 return (*funcp)(aTHXo_ idx, buf_sv, maxlen);
1892 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
1894 #ifdef PERL_CR_FILTER
1895 if (!PL_rsfp_filters) {
1896 filter_add(S_cr_textfilter,NULL);
1899 if (PL_rsfp_filters) {
1902 SvCUR_set(sv, 0); /* start with empty line */
1903 if (FILTER_READ(0, sv, 0) > 0)
1904 return ( SvPVX(sv) ) ;
1909 return (sv_gets(sv, fp, append));
1914 static char* exp_name[] =
1915 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
1916 "ATTRTERM", "TERMBLOCK"
1923 Works out what to call the token just pulled out of the input
1924 stream. The yacc parser takes care of taking the ops we return and
1925 stitching them into a tree.
1931 if read an identifier
1932 if we're in a my declaration
1933 croak if they tried to say my($foo::bar)
1934 build the ops for a my() declaration
1935 if it's an access to a my() variable
1936 are we in a sort block?
1937 croak if my($a); $a <=> $b
1938 build ops for access to a my() variable
1939 if in a dq string, and they've said @foo and we can't find @foo
1941 build ops for a bareword
1942 if we already built the token before, use it.
1946 #ifdef USE_PURE_BISON
1947 Perl_yylex(pTHX_ YYSTYPE *lvalp, int *lcharp)
1960 #ifdef USE_PURE_BISON
1961 yylval_pointer = lvalp;
1962 yychar_pointer = lcharp;
1965 /* check if there's an identifier for us to look at */
1966 if (PL_pending_ident) {
1967 /* pit holds the identifier we read and pending_ident is reset */
1968 char pit = PL_pending_ident;
1969 PL_pending_ident = 0;
1971 /* if we're in a my(), we can't allow dynamics here.
1972 $foo'bar has already been turned into $foo::bar, so
1973 just check for colons.
1975 if it's a legal name, the OP is a PADANY.
1978 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
1979 tmp = pad_allocmy(PL_tokenbuf);
1982 if (strchr(PL_tokenbuf,':'))
1983 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
1985 yylval.opval = newOP(OP_PADANY, 0);
1986 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
1992 build the ops for accesses to a my() variable.
1994 Deny my($a) or my($b) in a sort block, *if* $a or $b is
1995 then used in a comparison. This catches most, but not
1996 all cases. For instance, it catches
1997 sort { my($a); $a <=> $b }
1999 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
2000 (although why you'd do that is anyone's guess).
2003 if (!strchr(PL_tokenbuf,':')) {
2005 /* Check for single character per-thread SVs */
2006 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
2007 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
2008 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
2010 yylval.opval = newOP(OP_THREADSV, 0);
2011 yylval.opval->op_targ = tmp;
2014 #endif /* USE_THREADS */
2015 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
2016 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
2017 /* might be an "our" variable" */
2018 if (SvFLAGS(namesv) & SVpad_OUR) {
2019 /* build ops for a bareword */
2020 SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0);
2021 sv_catpvn(sym, "::", 2);
2022 sv_catpv(sym, PL_tokenbuf+1);
2023 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
2024 yylval.opval->op_private = OPpCONST_ENTERED;
2025 gv_fetchpv(SvPVX(sym),
2027 ? (GV_ADDMULTI | GV_ADDINEVAL)
2030 ((PL_tokenbuf[0] == '$') ? SVt_PV
2031 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2036 /* if it's a sort block and they're naming $a or $b */
2037 if (PL_last_lop_op == OP_SORT &&
2038 PL_tokenbuf[0] == '$' &&
2039 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
2042 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
2043 d < PL_bufend && *d != '\n';
2046 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
2047 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
2053 yylval.opval = newOP(OP_PADANY, 0);
2054 yylval.opval->op_targ = tmp;
2060 Whine if they've said @foo in a doublequoted string,
2061 and @foo isn't a variable we can find in the symbol
2064 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
2065 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
2066 if (!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
2067 yyerror(Perl_form(aTHX_ "In string, %s now must be written as \\%s",
2068 PL_tokenbuf, PL_tokenbuf));
2071 /* build ops for a bareword */
2072 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2073 yylval.opval->op_private = OPpCONST_ENTERED;
2074 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
2075 ((PL_tokenbuf[0] == '$') ? SVt_PV
2076 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2081 /* no identifier pending identification */
2083 switch (PL_lex_state) {
2085 case LEX_NORMAL: /* Some compilers will produce faster */
2086 case LEX_INTERPNORMAL: /* code if we comment these out. */
2090 /* when we've already built the next token, just pull it out of the queue */
2093 yylval = PL_nextval[PL_nexttoke];
2095 PL_lex_state = PL_lex_defer;
2096 PL_expect = PL_lex_expect;
2097 PL_lex_defer = LEX_NORMAL;
2099 return(PL_nexttype[PL_nexttoke]);
2101 /* interpolated case modifiers like \L \U, including \Q and \E.
2102 when we get here, PL_bufptr is at the \
2104 case LEX_INTERPCASEMOD:
2106 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2107 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2109 /* handle \E or end of string */
2110 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2114 if (PL_lex_casemods) {
2115 oldmod = PL_lex_casestack[--PL_lex_casemods];
2116 PL_lex_casestack[PL_lex_casemods] = '\0';
2118 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2120 PL_lex_state = LEX_INTERPCONCAT;
2124 if (PL_bufptr != PL_bufend)
2126 PL_lex_state = LEX_INTERPCONCAT;
2131 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2132 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2133 if (strchr("LU", *s) &&
2134 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2136 PL_lex_casestack[--PL_lex_casemods] = '\0';
2139 if (PL_lex_casemods > 10) {
2140 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2141 if (newlb != PL_lex_casestack) {
2143 PL_lex_casestack = newlb;
2146 PL_lex_casestack[PL_lex_casemods++] = *s;
2147 PL_lex_casestack[PL_lex_casemods] = '\0';
2148 PL_lex_state = LEX_INTERPCONCAT;
2149 PL_nextval[PL_nexttoke].ival = 0;
2152 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2154 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2156 PL_nextval[PL_nexttoke].ival = OP_LC;
2158 PL_nextval[PL_nexttoke].ival = OP_UC;
2160 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2162 Perl_croak(aTHX_ "panic: yylex");
2165 if (PL_lex_starts) {
2174 case LEX_INTERPPUSH:
2175 return sublex_push();
2177 case LEX_INTERPSTART:
2178 if (PL_bufptr == PL_bufend)
2179 return sublex_done();
2181 PL_lex_dojoin = (*PL_bufptr == '@');
2182 PL_lex_state = LEX_INTERPNORMAL;
2183 if (PL_lex_dojoin) {
2184 PL_nextval[PL_nexttoke].ival = 0;
2187 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2188 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2189 force_next(PRIVATEREF);
2191 force_ident("\"", '$');
2192 #endif /* USE_THREADS */
2193 PL_nextval[PL_nexttoke].ival = 0;
2195 PL_nextval[PL_nexttoke].ival = 0;
2197 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2200 if (PL_lex_starts++) {
2206 case LEX_INTERPENDMAYBE:
2207 if (intuit_more(PL_bufptr)) {
2208 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2214 if (PL_lex_dojoin) {
2215 PL_lex_dojoin = FALSE;
2216 PL_lex_state = LEX_INTERPCONCAT;
2219 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2220 && SvEVALED(PL_lex_repl))
2222 if (PL_bufptr != PL_bufend)
2223 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2224 PL_lex_repl = Nullsv;
2227 case LEX_INTERPCONCAT:
2229 if (PL_lex_brackets)
2230 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2232 if (PL_bufptr == PL_bufend)
2233 return sublex_done();
2235 if (SvIVX(PL_linestr) == '\'') {
2236 SV *sv = newSVsv(PL_linestr);
2239 else if ( PL_hints & HINT_NEW_RE )
2240 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2241 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2245 s = scan_const(PL_bufptr);
2247 PL_lex_state = LEX_INTERPCASEMOD;
2249 PL_lex_state = LEX_INTERPSTART;
2252 if (s != PL_bufptr) {
2253 PL_nextval[PL_nexttoke] = yylval;
2256 if (PL_lex_starts++)
2266 PL_lex_state = LEX_NORMAL;
2267 s = scan_formline(PL_bufptr);
2268 if (!PL_lex_formbrack)
2274 PL_oldoldbufptr = PL_oldbufptr;
2277 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2278 exp_name[PL_expect], s);
2284 if (isIDFIRST_lazy(s))
2286 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2289 goto fake_eof; /* emulate EOF on ^D or ^Z */
2294 if (PL_lex_brackets)
2295 yyerror("Missing right curly or square bracket");
2298 if (s++ < PL_bufend)
2299 goto retry; /* ignore stray nulls */
2302 if (!PL_in_eval && !PL_preambled) {
2303 PL_preambled = TRUE;
2304 sv_setpv(PL_linestr,incl_perldb());
2305 if (SvCUR(PL_linestr))
2306 sv_catpv(PL_linestr,";");
2308 while(AvFILLp(PL_preambleav) >= 0) {
2309 SV *tmpsv = av_shift(PL_preambleav);
2310 sv_catsv(PL_linestr, tmpsv);
2311 sv_catpv(PL_linestr, ";");
2314 sv_free((SV*)PL_preambleav);
2315 PL_preambleav = NULL;
2317 if (PL_minus_n || PL_minus_p) {
2318 sv_catpv(PL_linestr, "LINE: while (<>) {");
2320 sv_catpv(PL_linestr,"chomp;");
2322 GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
2324 GvIMPORTED_AV_on(gv);
2326 if (strchr("/'\"", *PL_splitstr)
2327 && strchr(PL_splitstr + 1, *PL_splitstr))
2328 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2331 s = "'~#\200\1'"; /* surely one char is unused...*/
2332 while (s[1] && strchr(PL_splitstr, *s)) s++;
2334 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c",
2335 "q" + (delim == '\''), delim);
2336 for (s = PL_splitstr; *s; s++) {
2338 sv_catpvn(PL_linestr, "\\", 1);
2339 sv_catpvn(PL_linestr, s, 1);
2341 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2345 sv_catpv(PL_linestr,"@F=split(' ');");
2348 sv_catpv(PL_linestr, "\n");
2349 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2350 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2351 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2352 SV *sv = NEWSV(85,0);
2354 sv_upgrade(sv, SVt_PVMG);
2355 sv_setsv(sv,PL_linestr);
2356 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2361 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2364 if (PL_preprocess && !PL_in_eval)
2365 (void)PerlProc_pclose(PL_rsfp);
2366 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2367 PerlIO_clearerr(PL_rsfp);
2369 (void)PerlIO_close(PL_rsfp);
2371 PL_doextract = FALSE;
2373 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2374 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2375 sv_catpv(PL_linestr,";}");
2376 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2377 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2378 PL_minus_n = PL_minus_p = 0;
2381 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2382 sv_setpv(PL_linestr,"");
2383 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2386 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2387 PL_doextract = FALSE;
2389 /* Incest with pod. */
2390 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2391 sv_setpv(PL_linestr, "");
2392 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2393 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2394 PL_doextract = FALSE;
2398 } while (PL_doextract);
2399 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2400 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2401 SV *sv = NEWSV(85,0);
2403 sv_upgrade(sv, SVt_PVMG);
2404 sv_setsv(sv,PL_linestr);
2405 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2407 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2408 if (CopLINE(PL_curcop) == 1) {
2409 while (s < PL_bufend && isSPACE(*s))
2411 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2415 if (*s == '#' && *(s+1) == '!')
2417 #ifdef ALTERNATE_SHEBANG
2419 static char as[] = ALTERNATE_SHEBANG;
2420 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2421 d = s + (sizeof(as) - 1);
2423 #endif /* ALTERNATE_SHEBANG */
2432 while (*d && !isSPACE(*d))
2436 #ifdef ARG_ZERO_IS_SCRIPT
2437 if (ipathend > ipath) {
2439 * HP-UX (at least) sets argv[0] to the script name,
2440 * which makes $^X incorrect. And Digital UNIX and Linux,
2441 * at least, set argv[0] to the basename of the Perl
2442 * interpreter. So, having found "#!", we'll set it right.
2444 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2445 assert(SvPOK(x) || SvGMAGICAL(x));
2446 if (sv_eq(x, CopFILESV(PL_curcop))) {
2447 sv_setpvn(x, ipath, ipathend - ipath);
2450 TAINT_NOT; /* $^X is always tainted, but that's OK */
2452 #endif /* ARG_ZERO_IS_SCRIPT */
2457 d = instr(s,"perl -");
2459 d = instr(s,"perl");
2461 /* avoid getting into infinite loops when shebang
2462 * line contains "Perl" rather than "perl" */
2464 for (d = ipathend-4; d >= ipath; --d) {
2465 if ((*d == 'p' || *d == 'P')
2466 && !ibcmp(d, "perl", 4))
2476 #ifdef ALTERNATE_SHEBANG
2478 * If the ALTERNATE_SHEBANG on this system starts with a
2479 * character that can be part of a Perl expression, then if
2480 * we see it but not "perl", we're probably looking at the
2481 * start of Perl code, not a request to hand off to some
2482 * other interpreter. Similarly, if "perl" is there, but
2483 * not in the first 'word' of the line, we assume the line
2484 * contains the start of the Perl program.
2486 if (d && *s != '#') {
2488 while (*c && !strchr("; \t\r\n\f\v#", *c))
2491 d = Nullch; /* "perl" not in first word; ignore */
2493 *s = '#'; /* Don't try to parse shebang line */
2495 #endif /* ALTERNATE_SHEBANG */
2500 !instr(s,"indir") &&
2501 instr(PL_origargv[0],"perl"))
2507 while (s < PL_bufend && isSPACE(*s))
2509 if (s < PL_bufend) {
2510 Newz(899,newargv,PL_origargc+3,char*);
2512 while (s < PL_bufend && !isSPACE(*s))
2515 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2518 newargv = PL_origargv;
2520 PerlProc_execv(ipath, newargv);
2521 Perl_croak(aTHX_ "Can't exec %s", ipath);
2524 U32 oldpdb = PL_perldb;
2525 bool oldn = PL_minus_n;
2526 bool oldp = PL_minus_p;
2528 while (*d && !isSPACE(*d)) d++;
2529 while (*d == ' ' || *d == '\t') d++;
2533 if (*d == 'M' || *d == 'm') {
2535 while (*d && !isSPACE(*d)) d++;
2536 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2539 d = moreswitches(d);
2541 if (PERLDB_LINE && !oldpdb ||
2542 ( PL_minus_n || PL_minus_p ) && !(oldn || oldp) )
2543 /* if we have already added "LINE: while (<>) {",
2544 we must not do it again */
2546 sv_setpv(PL_linestr, "");
2547 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2548 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2549 PL_preambled = FALSE;
2551 (void)gv_fetchfile(PL_origfilename);
2558 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2560 PL_lex_state = LEX_FORMLINE;
2565 #ifdef PERL_STRICT_CR
2566 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2568 "(Maybe you didn't strip carriage returns after a network transfer?)\n");
2570 case ' ': case '\t': case '\f': case 013:
2575 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2577 while (s < d && *s != '\n')
2582 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2584 PL_lex_state = LEX_FORMLINE;
2594 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2599 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
2602 if (strnEQ(s,"=>",2)) {
2603 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2604 OPERATOR('-'); /* unary minus */
2606 PL_last_uni = PL_oldbufptr;
2607 PL_last_lop_op = OP_FTEREAD; /* good enough */
2609 case 'r': FTST(OP_FTEREAD);
2610 case 'w': FTST(OP_FTEWRITE);
2611 case 'x': FTST(OP_FTEEXEC);
2612 case 'o': FTST(OP_FTEOWNED);
2613 case 'R': FTST(OP_FTRREAD);
2614 case 'W': FTST(OP_FTRWRITE);
2615 case 'X': FTST(OP_FTREXEC);
2616 case 'O': FTST(OP_FTROWNED);
2617 case 'e': FTST(OP_FTIS);
2618 case 'z': FTST(OP_FTZERO);
2619 case 's': FTST(OP_FTSIZE);
2620 case 'f': FTST(OP_FTFILE);
2621 case 'd': FTST(OP_FTDIR);
2622 case 'l': FTST(OP_FTLINK);
2623 case 'p': FTST(OP_FTPIPE);
2624 case 'S': FTST(OP_FTSOCK);
2625 case 'u': FTST(OP_FTSUID);
2626 case 'g': FTST(OP_FTSGID);
2627 case 'k': FTST(OP_FTSVTX);
2628 case 'b': FTST(OP_FTBLK);
2629 case 'c': FTST(OP_FTCHR);
2630 case 't': FTST(OP_FTTTY);
2631 case 'T': FTST(OP_FTTEXT);
2632 case 'B': FTST(OP_FTBINARY);
2633 case 'M': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTMTIME);
2634 case 'A': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTATIME);
2635 case 'C': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTCTIME);
2637 Perl_croak(aTHX_ "Unrecognized file test: -%c", (int)tmp);
2644 if (PL_expect == XOPERATOR)
2649 else if (*s == '>') {
2652 if (isIDFIRST_lazy(s)) {
2653 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2661 if (PL_expect == XOPERATOR)
2664 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2666 OPERATOR('-'); /* unary minus */
2673 if (PL_expect == XOPERATOR)
2678 if (PL_expect == XOPERATOR)
2681 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2687 if (PL_expect != XOPERATOR) {
2688 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2689 PL_expect = XOPERATOR;
2690 force_ident(PL_tokenbuf, '*');
2703 if (PL_expect == XOPERATOR) {
2707 PL_tokenbuf[0] = '%';
2708 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2709 if (!PL_tokenbuf[1]) {
2711 yyerror("Final % should be \\% or %name");
2714 PL_pending_ident = '%';
2733 switch (PL_expect) {
2736 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2738 PL_bufptr = s; /* update in case we back off */
2744 PL_expect = XTERMBLOCK;
2748 while (isIDFIRST_lazy(s)) {
2749 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2751 d = scan_str(d,TRUE,TRUE);
2754 SvREFCNT_dec(PL_lex_stuff);
2755 PL_lex_stuff = Nullsv;
2757 /* MUST advance bufptr here to avoid bogus
2758 "at end of line" context messages from yyerror().
2760 PL_bufptr = s + len;
2761 yyerror("Unterminated attribute parameter in attribute list");
2764 return 0; /* EOF indicator */
2768 SV *sv = newSVpvn(s, len);
2769 sv_catsv(sv, PL_lex_stuff);
2770 attrs = append_elem(OP_LIST, attrs,
2771 newSVOP(OP_CONST, 0, sv));
2772 SvREFCNT_dec(PL_lex_stuff);
2773 PL_lex_stuff = Nullsv;
2776 attrs = append_elem(OP_LIST, attrs,
2777 newSVOP(OP_CONST, 0,
2784 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}' for vi */
2785 if (*s != ';' && *s != tmp) {
2786 char q = ((*s == '\'') ? '"' : '\'');
2787 /* If here for an expression, and parsed no attrs, back off. */
2788 if (tmp == '=' && !attrs) {
2792 /* MUST advance bufptr here to avoid bogus "at end of line"
2793 context messages from yyerror().
2797 yyerror("Unterminated attribute list");
2799 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
2806 PL_nextval[PL_nexttoke].opval = attrs;
2814 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
2815 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
2820 if (CopLINE(PL_curcop) < PL_copline)
2821 PL_copline = CopLINE(PL_curcop);
2832 if (PL_lex_brackets <= 0)
2833 yyerror("Unmatched right square bracket");
2836 if (PL_lex_state == LEX_INTERPNORMAL) {
2837 if (PL_lex_brackets == 0) {
2838 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
2839 PL_lex_state = LEX_INTERPEND;
2846 if (PL_lex_brackets > 100) {
2847 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
2848 if (newlb != PL_lex_brackstack) {
2850 PL_lex_brackstack = newlb;
2853 switch (PL_expect) {
2855 if (PL_lex_formbrack) {
2859 if (PL_oldoldbufptr == PL_last_lop)
2860 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
2862 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2863 OPERATOR(HASHBRACK);
2865 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
2868 PL_tokenbuf[0] = '\0';
2869 if (d < PL_bufend && *d == '-') {
2870 PL_tokenbuf[0] = '-';
2872 while (d < PL_bufend && (*d == ' ' || *d == '\t'))
2875 if (d < PL_bufend && isIDFIRST_lazy(d)) {
2876 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
2878 while (d < PL_bufend && (*d == ' ' || *d == '\t'))
2881 char minus = (PL_tokenbuf[0] == '-');
2882 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
2890 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
2895 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2900 if (PL_oldoldbufptr == PL_last_lop)
2901 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
2903 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2906 OPERATOR(HASHBRACK);
2907 /* This hack serves to disambiguate a pair of curlies
2908 * as being a block or an anon hash. Normally, expectation
2909 * determines that, but in cases where we're not in a
2910 * position to expect anything in particular (like inside
2911 * eval"") we have to resolve the ambiguity. This code
2912 * covers the case where the first term in the curlies is a
2913 * quoted string. Most other cases need to be explicitly
2914 * disambiguated by prepending a `+' before the opening
2915 * curly in order to force resolution as an anon hash.
2917 * XXX should probably propagate the outer expectation
2918 * into eval"" to rely less on this hack, but that could
2919 * potentially break current behavior of eval"".
2923 if (*s == '\'' || *s == '"' || *s == '`') {
2924 /* common case: get past first string, handling escapes */
2925 for (t++; t < PL_bufend && *t != *s;)
2926 if (*t++ == '\\' && (*t == '\\' || *t == *s))
2930 else if (*s == 'q') {
2933 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
2937 char open, close, term;
2940 while (t < PL_bufend && isSPACE(*t))
2944 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
2948 for (t++; t < PL_bufend; t++) {
2949 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
2951 else if (*t == open)
2955 for (t++; t < PL_bufend; t++) {
2956 if (*t == '\\' && t+1 < PL_bufend)
2958 else if (*t == close && --brackets <= 0)
2960 else if (*t == open)
2966 else if (isALNUM_lazy(t)) {
2968 while (t < PL_bufend && isALNUM_lazy(t))
2971 while (t < PL_bufend && isSPACE(*t))
2973 /* if comma follows first term, call it an anon hash */
2974 /* XXX it could be a comma expression with loop modifiers */
2975 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
2976 || (*t == '=' && t[1] == '>')))
2977 OPERATOR(HASHBRACK);
2978 if (PL_expect == XREF)
2981 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
2987 yylval.ival = CopLINE(PL_curcop);
2988 if (isSPACE(*s) || *s == '#')
2989 PL_copline = NOLINE; /* invalidate current command line number */
2994 if (PL_lex_brackets <= 0)
2995 yyerror("Unmatched right curly bracket");
2997 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
2998 if (PL_lex_brackets < PL_lex_formbrack)
2999 PL_lex_formbrack = 0;
3000 if (PL_lex_state == LEX_INTERPNORMAL) {
3001 if (PL_lex_brackets == 0) {
3002 if (PL_expect & XFAKEBRACK) {
3003 PL_expect &= XENUMMASK;
3004 PL_lex_state = LEX_INTERPEND;
3006 return yylex(); /* ignore fake brackets */
3008 if (*s == '-' && s[1] == '>')
3009 PL_lex_state = LEX_INTERPENDMAYBE;
3010 else if (*s != '[' && *s != '{')
3011 PL_lex_state = LEX_INTERPEND;
3014 if (PL_expect & XFAKEBRACK) {
3015 PL_expect &= XENUMMASK;
3017 return yylex(); /* ignore fake brackets */
3027 if (PL_expect == XOPERATOR) {
3028 if (ckWARN(WARN_SEMICOLON) && isIDFIRST_lazy(s) && PL_bufptr == PL_linestart) {
3029 CopLINE_dec(PL_curcop);
3030 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3031 CopLINE_inc(PL_curcop);
3036 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3038 PL_expect = XOPERATOR;
3039 force_ident(PL_tokenbuf, '&');
3043 yylval.ival = (OPpENTERSUB_AMPER<<8);
3062 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3063 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3065 if (PL_expect == XSTATE && isALPHA(tmp) &&
3066 (s == PL_linestart+1 || s[-2] == '\n') )
3068 if (PL_in_eval && !PL_rsfp) {
3073 if (strnEQ(s,"=cut",4)) {
3087 PL_doextract = TRUE;
3090 if (PL_lex_brackets < PL_lex_formbrack) {
3092 #ifdef PERL_STRICT_CR
3093 for (t = s; *t == ' ' || *t == '\t'; t++) ;
3095 for (t = s; *t == ' ' || *t == '\t' || *t == '\r'; t++) ;
3097 if (*t == '\n' || *t == '#') {
3115 if (PL_expect != XOPERATOR) {
3116 if (s[1] != '<' && !strchr(s,'>'))
3119 s = scan_heredoc(s);
3121 s = scan_inputsymbol(s);
3122 TERM(sublex_start());
3127 SHop(OP_LEFT_SHIFT);
3141 SHop(OP_RIGHT_SHIFT);
3150 if (PL_expect == XOPERATOR) {
3151 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3154 return ','; /* grandfather non-comma-format format */
3158 if (s[1] == '#' && (isIDFIRST_lazy(s+2) || strchr("{$:+-", s[2]))) {
3159 PL_tokenbuf[0] = '@';
3160 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3161 sizeof PL_tokenbuf - 1, FALSE);
3162 if (PL_expect == XOPERATOR)
3163 no_op("Array length", s);
3164 if (!PL_tokenbuf[1])
3166 PL_expect = XOPERATOR;
3167 PL_pending_ident = '#';
3171 PL_tokenbuf[0] = '$';
3172 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3173 sizeof PL_tokenbuf - 1, FALSE);
3174 if (PL_expect == XOPERATOR)
3176 if (!PL_tokenbuf[1]) {
3178 yyerror("Final $ should be \\$ or $name");
3182 /* This kludge not intended to be bulletproof. */
3183 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3184 yylval.opval = newSVOP(OP_CONST, 0,
3185 newSViv((IV)PL_compiling.cop_arybase));
3186 yylval.opval->op_private = OPpCONST_ARYBASE;
3192 if (PL_lex_state == LEX_NORMAL)
3195 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3198 PL_tokenbuf[0] = '@';
3199 if (ckWARN(WARN_SYNTAX)) {
3201 isSPACE(*t) || isALNUM_lazy(t) || *t == '$';
3204 PL_bufptr = skipspace(PL_bufptr);
3205 while (t < PL_bufend && *t != ']')
3207 Perl_warner(aTHX_ WARN_SYNTAX,
3208 "Multidimensional syntax %.*s not supported",
3209 (t - PL_bufptr) + 1, PL_bufptr);
3213 else if (*s == '{') {
3214 PL_tokenbuf[0] = '%';
3215 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3216 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3218 char tmpbuf[sizeof PL_tokenbuf];
3220 for (t++; isSPACE(*t); t++) ;
3221 if (isIDFIRST_lazy(t)) {
3222 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3223 for (; isSPACE(*t); t++) ;
3224 if (*t == ';' && get_cv(tmpbuf, FALSE))
3225 Perl_warner(aTHX_ WARN_SYNTAX,
3226 "You need to quote \"%s\"", tmpbuf);
3232 PL_expect = XOPERATOR;
3233 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3234 bool islop = (PL_last_lop == PL_oldoldbufptr);
3235 if (!islop || PL_last_lop_op == OP_GREPSTART)
3236 PL_expect = XOPERATOR;
3237 else if (strchr("$@\"'`q", *s))
3238 PL_expect = XTERM; /* e.g. print $fh "foo" */
3239 else if (strchr("&*<%", *s) && isIDFIRST_lazy(s+1))
3240 PL_expect = XTERM; /* e.g. print $fh &sub */
3241 else if (isIDFIRST_lazy(s)) {
3242 char tmpbuf[sizeof PL_tokenbuf];
3243 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3244 if (tmp = keyword(tmpbuf, len)) {
3245 /* binary operators exclude handle interpretations */
3257 PL_expect = XTERM; /* e.g. print $fh length() */
3262 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3263 if (gv && GvCVu(gv))
3264 PL_expect = XTERM; /* e.g. print $fh subr() */
3267 else if (isDIGIT(*s))
3268 PL_expect = XTERM; /* e.g. print $fh 3 */
3269 else if (*s == '.' && isDIGIT(s[1]))
3270 PL_expect = XTERM; /* e.g. print $fh .3 */
3271 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3272 PL_expect = XTERM; /* e.g. print $fh -1 */
3273 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3274 PL_expect = XTERM; /* print $fh <<"EOF" */
3276 PL_pending_ident = '$';
3280 if (PL_expect == XOPERATOR)
3282 PL_tokenbuf[0] = '@';
3283 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3284 if (!PL_tokenbuf[1]) {
3286 yyerror("Final @ should be \\@ or @name");
3289 if (PL_lex_state == LEX_NORMAL)
3291 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3293 PL_tokenbuf[0] = '%';
3295 /* Warn about @ where they meant $. */
3296 if (ckWARN(WARN_SYNTAX)) {
3297 if (*s == '[' || *s == '{') {
3299 while (*t && (isALNUM_lazy(t) || strchr(" \t$#+-'\"", *t)))
3301 if (*t == '}' || *t == ']') {
3303 PL_bufptr = skipspace(PL_bufptr);
3304 Perl_warner(aTHX_ WARN_SYNTAX,
3305 "Scalar value %.*s better written as $%.*s",
3306 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3311 PL_pending_ident = '@';
3314 case '/': /* may either be division or pattern */
3315 case '?': /* may either be conditional or pattern */
3316 if (PL_expect != XOPERATOR) {
3317 /* Disable warning on "study /blah/" */
3318 if (PL_oldoldbufptr == PL_last_uni
3319 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3320 || memNE(PL_last_uni, "study", 5) || isALNUM_lazy(PL_last_uni+5)))
3322 s = scan_pat(s,OP_MATCH);
3323 TERM(sublex_start());
3331 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3332 #ifdef PERL_STRICT_CR
3335 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3337 && (s == PL_linestart || s[-1] == '\n') )
3339 PL_lex_formbrack = 0;
3343 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3349 yylval.ival = OPf_SPECIAL;
3355 if (PL_expect != XOPERATOR)
3360 case '0': case '1': case '2': case '3': case '4':
3361 case '5': case '6': case '7': case '8': case '9':
3363 if (PL_expect == XOPERATOR)
3368 s = scan_str(s,FALSE,FALSE);
3369 if (PL_expect == XOPERATOR) {
3370 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3373 return ','; /* grandfather non-comma-format format */
3379 missingterm((char*)0);
3380 yylval.ival = OP_CONST;
3381 TERM(sublex_start());
3384 s = scan_str(s,FALSE,FALSE);
3385 if (PL_expect == XOPERATOR) {
3386 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3389 return ','; /* grandfather non-comma-format format */
3395 missingterm((char*)0);
3396 yylval.ival = OP_CONST;
3397 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3398 if (*d == '$' || *d == '@' || *d == '\\' || *d & 0x80) {
3399 yylval.ival = OP_STRINGIFY;
3403 TERM(sublex_start());
3406 s = scan_str(s,FALSE,FALSE);
3407 if (PL_expect == XOPERATOR)
3408 no_op("Backticks",s);
3410 missingterm((char*)0);
3411 yylval.ival = OP_BACKTICK;
3413 TERM(sublex_start());
3417 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3418 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3420 if (PL_expect == XOPERATOR)
3421 no_op("Backslash",s);
3425 if (isDIGIT(s[1]) && PL_expect == XTERM) {
3429 while (isDIGIT(*start))
3431 if (*start == '.' && isDIGIT(start[1])) {
3438 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3478 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3480 /* Some keywords can be followed by any delimiter, including ':' */
3481 tmp = (len == 1 && strchr("msyq", PL_tokenbuf[0]) ||
3482 len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3483 (PL_tokenbuf[0] == 'q' &&
3484 strchr("qwxr", PL_tokenbuf[1]))));
3486 /* x::* is just a word, unless x is "CORE" */
3487 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3491 while (d < PL_bufend && isSPACE(*d))
3492 d++; /* no comments skipped here, or s### is misparsed */
3494 /* Is this a label? */
3495 if (!tmp && PL_expect == XSTATE
3496 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3498 yylval.pval = savepv(PL_tokenbuf);
3503 /* Check for keywords */
3504 tmp = keyword(PL_tokenbuf, len);
3506 /* Is this a word before a => operator? */
3507 if (strnEQ(d,"=>",2)) {
3509 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3510 yylval.opval->op_private = OPpCONST_BARE;
3514 if (tmp < 0) { /* second-class keyword? */
3515 GV *ogv = Nullgv; /* override (winner) */
3516 GV *hgv = Nullgv; /* hidden (loser) */
3517 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3519 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3522 if (GvIMPORTED_CV(gv))
3524 else if (! CvMETHOD(cv))
3528 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3529 (gv = *gvp) != (GV*)&PL_sv_undef &&
3530 GvCVu(gv) && GvIMPORTED_CV(gv))
3536 tmp = 0; /* overridden by import or by GLOBAL */
3539 && -tmp==KEY_lock /* XXX generalizable kludge */
3541 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3543 tmp = 0; /* any sub overrides "weak" keyword */
3545 else { /* no override */
3549 if (ckWARN(WARN_AMBIGUOUS) && hgv
3550 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3551 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3552 "Ambiguous call resolved as CORE::%s(), %s",
3553 GvENAME(hgv), "qualify as such or use &");
3560 default: /* not a keyword */
3563 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3565 /* Get the rest if it looks like a package qualifier */
3567 if (*s == '\'' || *s == ':' && s[1] == ':') {
3569 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3572 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3573 *s == '\'' ? "'" : "::");
3577 if (PL_expect == XOPERATOR) {
3578 if (PL_bufptr == PL_linestart) {
3579 CopLINE_dec(PL_curcop);
3580 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3581 CopLINE_inc(PL_curcop);
3584 no_op("Bareword",s);
3587 /* Look for a subroutine with this name in current package,
3588 unless name is "Foo::", in which case Foo is a bearword
3589 (and a package name). */
3592 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3594 if (ckWARN(WARN_UNSAFE) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3595 Perl_warner(aTHX_ WARN_UNSAFE,
3596 "Bareword \"%s\" refers to nonexistent package",
3599 PL_tokenbuf[len] = '\0';
3606 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3609 /* if we saw a global override before, get the right name */
3612 sv = newSVpvn("CORE::GLOBAL::",14);
3613 sv_catpv(sv,PL_tokenbuf);
3616 sv = newSVpv(PL_tokenbuf,0);
3618 /* Presume this is going to be a bareword of some sort. */
3621 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3622 yylval.opval->op_private = OPpCONST_BARE;
3624 /* And if "Foo::", then that's what it certainly is. */
3629 /* See if it's the indirect object for a list operator. */
3631 if (PL_oldoldbufptr &&
3632 PL_oldoldbufptr < PL_bufptr &&
3633 (PL_oldoldbufptr == PL_last_lop
3634 || PL_oldoldbufptr == PL_last_uni) &&
3635 /* NO SKIPSPACE BEFORE HERE! */
3636 (PL_expect == XREF ||
3637 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3639 bool immediate_paren = *s == '(';
3641 /* (Now we can afford to cross potential line boundary.) */
3644 /* Two barewords in a row may indicate method call. */
3646 if ((isIDFIRST_lazy(s) || *s == '$') && (tmp=intuit_method(s,gv)))
3649 /* If not a declared subroutine, it's an indirect object. */
3650 /* (But it's an indir obj regardless for sort.) */
3652 if ((PL_last_lop_op == OP_SORT ||
3653 (!immediate_paren && (!gv || !GvCVu(gv)))) &&
3654 (PL_last_lop_op != OP_MAPSTART &&
3655 PL_last_lop_op != OP_GREPSTART))
3657 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3662 /* If followed by a paren, it's certainly a subroutine. */
3664 PL_expect = XOPERATOR;
3668 if (gv && GvCVu(gv)) {
3669 for (d = s + 1; *d == ' ' || *d == '\t'; d++) ;
3670 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3675 PL_nextval[PL_nexttoke].opval = yylval.opval;
3676 PL_expect = XOPERATOR;
3682 /* If followed by var or block, call it a method (unless sub) */
3684 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3685 PL_last_lop = PL_oldbufptr;
3686 PL_last_lop_op = OP_METHOD;
3690 /* If followed by a bareword, see if it looks like indir obj. */
3692 if ((isIDFIRST_lazy(s) || *s == '$') && (tmp = intuit_method(s,gv)))
3695 /* Not a method, so call it a subroutine (if defined) */
3697 if (gv && GvCVu(gv)) {
3699 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
3700 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3701 "Ambiguous use of -%s resolved as -&%s()",
3702 PL_tokenbuf, PL_tokenbuf);
3703 /* Check for a constant sub */
3705 if ((sv = cv_const_sv(cv))) {
3707 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
3708 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
3709 yylval.opval->op_private = 0;
3713 /* Resolve to GV now. */
3714 op_free(yylval.opval);
3715 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
3716 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
3717 PL_last_lop = PL_oldbufptr;
3718 PL_last_lop_op = OP_ENTERSUB;
3719 /* Is there a prototype? */
3722 char *proto = SvPV((SV*)cv, len);
3725 if (strEQ(proto, "$"))
3727 if (*proto == '&' && *s == '{') {
3728 sv_setpv(PL_subname,"__ANON__");
3732 PL_nextval[PL_nexttoke].opval = yylval.opval;
3738 /* Call it a bare word */
3740 if (PL_hints & HINT_STRICT_SUBS)
3741 yylval.opval->op_private |= OPpCONST_STRICT;
3744 if (ckWARN(WARN_RESERVED)) {
3745 if (lastchar != '-') {
3746 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
3748 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
3755 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
3756 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3757 "Operator or semicolon missing before %c%s",
3758 lastchar, PL_tokenbuf);
3759 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3760 "Ambiguous use of %c resolved as operator %c",
3761 lastchar, lastchar);
3767 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3768 newSVpv(CopFILE(PL_curcop),0));
3772 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3773 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
3776 case KEY___PACKAGE__:
3777 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3779 ? newSVsv(PL_curstname)
3788 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
3789 char *pname = "main";
3790 if (PL_tokenbuf[2] == 'D')
3791 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
3792 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
3795 GvIOp(gv) = newIO();
3796 IoIFP(GvIOp(gv)) = PL_rsfp;
3797 #if defined(HAS_FCNTL) && defined(F_SETFD)
3799 int fd = PerlIO_fileno(PL_rsfp);
3800 fcntl(fd,F_SETFD,fd >= 3);
3803 /* Mark this internal pseudo-handle as clean */
3804 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
3806 IoTYPE(GvIOp(gv)) = '|';
3807 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
3808 IoTYPE(GvIOp(gv)) = '-';
3810 IoTYPE(GvIOp(gv)) = '<';
3811 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
3812 /* if the script was opened in binmode, we need to revert
3813 * it to text mode for compatibility; but only iff it has CRs
3814 * XXX this is a questionable hack at best. */
3815 if (PL_bufend-PL_bufptr > 2
3816 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
3819 if (IoTYPE(GvIOp(gv)) == '<') {
3820 loc = PerlIO_tell(PL_rsfp);
3821 (void)PerlIO_seek(PL_rsfp, 0L, 0);
3823 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
3824 #if defined(__BORLANDC__)
3825 /* XXX see note in do_binmode() */
3826 ((FILE*)PL_rsfp)->flags |= _F_BIN;
3829 PerlIO_seek(PL_rsfp, loc, 0);
3844 if (PL_expect == XSTATE) {
3851 if (*s == ':' && s[1] == ':') {
3854 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3855 tmp = keyword(PL_tokenbuf, len);
3869 LOP(OP_ACCEPT,XTERM);
3875 LOP(OP_ATAN2,XTERM);
3884 LOP(OP_BLESS,XTERM);
3893 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
3910 if (!PL_cryptseen) {
3911 PL_cryptseen = TRUE;
3915 LOP(OP_CRYPT,XTERM);
3918 if (ckWARN(WARN_OCTAL)) {
3919 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
3920 if (*d != '0' && isDIGIT(*d))
3921 Perl_warner(aTHX_ WARN_OCTAL,
3922 "chmod: mode argument is missing initial 0");
3924 LOP(OP_CHMOD,XTERM);
3927 LOP(OP_CHOWN,XTERM);
3930 LOP(OP_CONNECT,XTERM);
3946 s = force_word(s,WORD,FALSE,TRUE,FALSE);
3950 PL_hints |= HINT_BLOCK_SCOPE;
3960 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
3961 LOP(OP_DBMOPEN,XTERM);
3967 s = force_word(s,WORD,TRUE,FALSE,FALSE);
3974 yylval.ival = CopLINE(PL_curcop);
3988 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
3989 UNIBRACK(OP_ENTEREVAL);
4004 case KEY_endhostent:
4010 case KEY_endservent:
4013 case KEY_endprotoent:
4024 yylval.ival = CopLINE(PL_curcop);
4026 if (PL_expect == XSTATE && isIDFIRST_lazy(s)) {
4028 if ((PL_bufend - p) >= 3 &&
4029 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4031 else if ((PL_bufend - p) >= 4 &&
4032 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4035 if (isIDFIRST_lazy(p)) {
4036 p = scan_ident(p, PL_bufend,
4037 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4041 Perl_croak(aTHX_ "Missing $ on loop variable");
4046 LOP(OP_FORMLINE,XTERM);
4052 LOP(OP_FCNTL,XTERM);
4058 LOP(OP_FLOCK,XTERM);
4067 LOP(OP_GREPSTART, *s == '(' ? XTERM : XREF);
4070 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4085 case KEY_getpriority:
4086 LOP(OP_GETPRIORITY,XTERM);
4088 case KEY_getprotobyname:
4091 case KEY_getprotobynumber:
4092 LOP(OP_GPBYNUMBER,XTERM);
4094 case KEY_getprotoent:
4106 case KEY_getpeername:
4107 UNI(OP_GETPEERNAME);
4109 case KEY_gethostbyname:
4112 case KEY_gethostbyaddr:
4113 LOP(OP_GHBYADDR,XTERM);
4115 case KEY_gethostent:
4118 case KEY_getnetbyname:
4121 case KEY_getnetbyaddr:
4122 LOP(OP_GNBYADDR,XTERM);
4127 case KEY_getservbyname:
4128 LOP(OP_GSBYNAME,XTERM);
4130 case KEY_getservbyport:
4131 LOP(OP_GSBYPORT,XTERM);
4133 case KEY_getservent:
4136 case KEY_getsockname:
4137 UNI(OP_GETSOCKNAME);
4139 case KEY_getsockopt:
4140 LOP(OP_GSOCKOPT,XTERM);
4162 yylval.ival = CopLINE(PL_curcop);
4166 LOP(OP_INDEX,XTERM);
4172 LOP(OP_IOCTL,XTERM);
4184 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4216 LOP(OP_LISTEN,XTERM);
4225 s = scan_pat(s,OP_MATCH);
4226 TERM(sublex_start());
4229 LOP(OP_MAPSTART, *s == '(' ? XTERM : XREF);
4232 LOP(OP_MKDIR,XTERM);
4235 LOP(OP_MSGCTL,XTERM);
4238 LOP(OP_MSGGET,XTERM);
4241 LOP(OP_MSGRCV,XTERM);
4244 LOP(OP_MSGSND,XTERM);
4250 if (isIDFIRST_lazy(s)) {
4251 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4252 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4254 PL_in_my_stash = gv_stashpv(PL_tokenbuf, FALSE);
4255 if (!PL_in_my_stash) {
4258 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4266 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4273 if (PL_expect != XSTATE)
4274 yyerror("\"no\" not allowed in expression");
4275 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4276 s = force_version(s);
4281 if (*s == '(' || (s = skipspace(s), *s == '('))
4288 if (isIDFIRST_lazy(s)) {
4290 for (d = s; isALNUM_lazy(d); d++) ;
4292 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_AMBIGUOUS))
4293 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4294 "Precedence problem: open %.*s should be open(%.*s)",
4300 yylval.ival = OP_OR;
4310 LOP(OP_OPEN_DIR,XTERM);
4313 checkcomma(s,PL_tokenbuf,"filehandle");
4317 checkcomma(s,PL_tokenbuf,"filehandle");
4336 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4340 LOP(OP_PIPE_OP,XTERM);
4343 s = scan_str(s,FALSE,FALSE);
4345 missingterm((char*)0);
4346 yylval.ival = OP_CONST;
4347 TERM(sublex_start());
4353 s = scan_str(s,FALSE,FALSE);
4355 missingterm((char*)0);
4357 if (SvCUR(PL_lex_stuff)) {
4360 d = SvPV_force(PL_lex_stuff, len);
4362 for (; isSPACE(*d) && len; --len, ++d) ;
4365 if (!warned && ckWARN(WARN_SYNTAX)) {
4366 for (; !isSPACE(*d) && len; --len, ++d) {
4368 Perl_warner(aTHX_ WARN_SYNTAX,
4369 "Possible attempt to separate words with commas");
4372 else if (*d == '#') {
4373 Perl_warner(aTHX_ WARN_SYNTAX,
4374 "Possible attempt to put comments in qw() list");
4380 for (; !isSPACE(*d) && len; --len, ++d) ;
4382 words = append_elem(OP_LIST, words,
4383 newSVOP(OP_CONST, 0, newSVpvn(b, d-b)));
4387 PL_nextval[PL_nexttoke].opval = words;
4392 SvREFCNT_dec(PL_lex_stuff);
4393 PL_lex_stuff = Nullsv;
4398 s = scan_str(s,FALSE,FALSE);
4400 missingterm((char*)0);
4401 yylval.ival = OP_STRINGIFY;
4402 if (SvIVX(PL_lex_stuff) == '\'')
4403 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4404 TERM(sublex_start());
4407 s = scan_pat(s,OP_QR);
4408 TERM(sublex_start());
4411 s = scan_str(s,FALSE,FALSE);
4413 missingterm((char*)0);
4414 yylval.ival = OP_BACKTICK;
4416 TERM(sublex_start());
4423 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4424 s = force_version(s);
4427 *PL_tokenbuf = '\0';
4428 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4429 if (isIDFIRST_lazy(PL_tokenbuf))
4430 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4432 yyerror("<> should be quotes");
4440 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4444 LOP(OP_RENAME,XTERM);
4453 LOP(OP_RINDEX,XTERM);
4476 LOP(OP_REVERSE,XTERM);
4487 TERM(sublex_start());
4489 TOKEN(1); /* force error */
4498 LOP(OP_SELECT,XTERM);
4504 LOP(OP_SEMCTL,XTERM);
4507 LOP(OP_SEMGET,XTERM);
4510 LOP(OP_SEMOP,XTERM);
4516 LOP(OP_SETPGRP,XTERM);
4518 case KEY_setpriority:
4519 LOP(OP_SETPRIORITY,XTERM);
4521 case KEY_sethostent:
4527 case KEY_setservent:
4530 case KEY_setprotoent:
4540 LOP(OP_SEEKDIR,XTERM);
4542 case KEY_setsockopt:
4543 LOP(OP_SSOCKOPT,XTERM);
4549 LOP(OP_SHMCTL,XTERM);
4552 LOP(OP_SHMGET,XTERM);
4555 LOP(OP_SHMREAD,XTERM);
4558 LOP(OP_SHMWRITE,XTERM);
4561 LOP(OP_SHUTDOWN,XTERM);
4570 LOP(OP_SOCKET,XTERM);
4572 case KEY_socketpair:
4573 LOP(OP_SOCKPAIR,XTERM);
4576 checkcomma(s,PL_tokenbuf,"subroutine name");
4578 if (*s == ';' || *s == ')') /* probably a close */
4579 Perl_croak(aTHX_ "sort is now a reserved word");
4581 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4585 LOP(OP_SPLIT,XTERM);
4588 LOP(OP_SPRINTF,XTERM);
4591 LOP(OP_SPLICE,XTERM);
4606 LOP(OP_SUBSTR,XTERM);
4612 char tmpbuf[sizeof PL_tokenbuf];
4614 expectation attrful;
4615 bool have_name, have_proto;
4620 if (isIDFIRST_lazy(s) || *s == '\'' ||
4621 (*s == ':' && s[1] == ':'))
4624 attrful = XATTRBLOCK;
4625 /* remember buffer pos'n for later force_word */
4626 tboffset = s - PL_oldbufptr;
4627 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4628 if (strchr(tmpbuf, ':'))
4629 sv_setpv(PL_subname, tmpbuf);
4631 sv_setsv(PL_subname,PL_curstname);
4632 sv_catpvn(PL_subname,"::",2);
4633 sv_catpvn(PL_subname,tmpbuf,len);
4640 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4641 PL_expect = XTERMBLOCK;
4642 attrful = XATTRTERM;
4643 sv_setpv(PL_subname,"?");
4647 if (key == KEY_format) {
4649 PL_lex_formbrack = PL_lex_brackets + 1;
4651 (void) force_word(PL_oldbufptr + tboffset, WORD,
4656 /* Look for a prototype */
4660 s = scan_str(s,FALSE,FALSE);
4663 SvREFCNT_dec(PL_lex_stuff);
4664 PL_lex_stuff = Nullsv;
4665 Perl_croak(aTHX_ "Prototype not terminated");
4668 d = SvPVX(PL_lex_stuff);
4670 for (p = d; *p; ++p) {
4675 SvCUR(PL_lex_stuff) = tmp;
4683 if (*s == ':' && s[1] != ':')
4684 PL_expect = attrful;
4687 PL_nextval[PL_nexttoke].opval =
4688 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
4689 PL_lex_stuff = Nullsv;
4693 sv_setpv(PL_subname,"__ANON__");
4696 (void) force_word(PL_oldbufptr + tboffset, WORD,
4705 LOP(OP_SYSTEM,XREF);
4708 LOP(OP_SYMLINK,XTERM);
4711 LOP(OP_SYSCALL,XTERM);
4714 LOP(OP_SYSOPEN,XTERM);
4717 LOP(OP_SYSSEEK,XTERM);
4720 LOP(OP_SYSREAD,XTERM);
4723 LOP(OP_SYSWRITE,XTERM);
4727 TERM(sublex_start());
4748 LOP(OP_TRUNCATE,XTERM);
4760 yylval.ival = CopLINE(PL_curcop);
4764 yylval.ival = CopLINE(PL_curcop);
4768 LOP(OP_UNLINK,XTERM);
4774 LOP(OP_UNPACK,XTERM);
4777 LOP(OP_UTIME,XTERM);
4780 if (ckWARN(WARN_OCTAL)) {
4781 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4782 if (*d != '0' && isDIGIT(*d))
4783 Perl_warner(aTHX_ WARN_OCTAL,
4784 "umask: argument is missing initial 0");
4789 LOP(OP_UNSHIFT,XTERM);
4792 if (PL_expect != XSTATE)
4793 yyerror("\"use\" not allowed in expression");
4795 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4796 s = force_version(s);
4797 if (*s == ';' || (s = skipspace(s), *s == ';')) {
4798 PL_nextval[PL_nexttoke].opval = Nullop;
4803 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4804 s = force_version(s);
4816 yylval.ival = CopLINE(PL_curcop);
4820 PL_hints |= HINT_BLOCK_SCOPE;
4827 LOP(OP_WAITPID,XTERM);
4835 static char ctl_l[2];
4837 if (ctl_l[0] == '\0')
4838 ctl_l[0] = toCTRL('L');
4839 gv_fetchpv(ctl_l,TRUE, SVt_PV);
4842 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
4847 if (PL_expect == XOPERATOR)
4853 yylval.ival = OP_XOR;
4858 TERM(sublex_start());
4864 Perl_keyword(pTHX_ register char *d, I32 len)
4869 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
4870 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
4871 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
4872 if (strEQ(d,"__DATA__")) return KEY___DATA__;
4873 if (strEQ(d,"__END__")) return KEY___END__;
4877 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
4882 if (strEQ(d,"and")) return -KEY_and;
4883 if (strEQ(d,"abs")) return -KEY_abs;
4886 if (strEQ(d,"alarm")) return -KEY_alarm;
4887 if (strEQ(d,"atan2")) return -KEY_atan2;
4890 if (strEQ(d,"accept")) return -KEY_accept;
4895 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
4898 if (strEQ(d,"bless")) return -KEY_bless;
4899 if (strEQ(d,"bind")) return -KEY_bind;
4900 if (strEQ(d,"binmode")) return -KEY_binmode;
4903 if (strEQ(d,"CORE")) return -KEY_CORE;
4908 if (strEQ(d,"cmp")) return -KEY_cmp;
4909 if (strEQ(d,"chr")) return -KEY_chr;
4910 if (strEQ(d,"cos")) return -KEY_cos;
4913 if (strEQ(d,"chop")) return KEY_chop;
4916 if (strEQ(d,"close")) return -KEY_close;
4917 if (strEQ(d,"chdir")) return -KEY_chdir;
4918 if (strEQ(d,"chomp")) return KEY_chomp;
4919 if (strEQ(d,"chmod")) return -KEY_chmod;
4920 if (strEQ(d,"chown")) return -KEY_chown;
4921 if (strEQ(d,"crypt")) return -KEY_crypt;
4924 if (strEQ(d,"chroot")) return -KEY_chroot;
4925 if (strEQ(d,"caller")) return -KEY_caller;
4928 if (strEQ(d,"connect")) return -KEY_connect;
4931 if (strEQ(d,"closedir")) return -KEY_closedir;
4932 if (strEQ(d,"continue")) return -KEY_continue;
4937 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
4942 if (strEQ(d,"do")) return KEY_do;
4945 if (strEQ(d,"die")) return -KEY_die;
4948 if (strEQ(d,"dump")) return -KEY_dump;
4951 if (strEQ(d,"delete")) return KEY_delete;
4954 if (strEQ(d,"defined")) return KEY_defined;
4955 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
4958 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
4963 if (strEQ(d,"EQ")) { deprecate(d); return -KEY_eq;}
4964 if (strEQ(d,"END")) return KEY_END;
4969 if (strEQ(d,"eq")) return -KEY_eq;
4972 if (strEQ(d,"eof")) return -KEY_eof;
4973 if (strEQ(d,"exp")) return -KEY_exp;
4976 if (strEQ(d,"else")) return KEY_else;
4977 if (strEQ(d,"exit")) return -KEY_exit;
4978 if (strEQ(d,"eval")) return KEY_eval;
4979 if (strEQ(d,"exec")) return -KEY_exec;
4980 if (strEQ(d,"each")) return KEY_each;
4983 if (strEQ(d,"elsif")) return KEY_elsif;
4986 if (strEQ(d,"exists")) return KEY_exists;
4987 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
4990 if (strEQ(d,"endgrent")) return -KEY_endgrent;
4991 if (strEQ(d,"endpwent")) return -KEY_endpwent;
4994 if (strEQ(d,"endnetent")) return -KEY_endnetent;
4997 if (strEQ(d,"endhostent")) return -KEY_endhostent;
4998 if (strEQ(d,"endservent")) return -KEY_endservent;
5001 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5008 if (strEQ(d,"for")) return KEY_for;
5011 if (strEQ(d,"fork")) return -KEY_fork;
5014 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5015 if (strEQ(d,"flock")) return -KEY_flock;
5018 if (strEQ(d,"format")) return KEY_format;
5019 if (strEQ(d,"fileno")) return -KEY_fileno;
5022 if (strEQ(d,"foreach")) return KEY_foreach;
5025 if (strEQ(d,"formline")) return -KEY_formline;
5031 if (strEQ(d,"GT")) { deprecate(d); return -KEY_gt;}
5032 if (strEQ(d,"GE")) { deprecate(d); return -KEY_ge;}
5036 if (strnEQ(d,"get",3)) {
5041 if (strEQ(d,"ppid")) return -KEY_getppid;
5042 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5045 if (strEQ(d,"pwent")) return -KEY_getpwent;
5046 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5047 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5050 if (strEQ(d,"peername")) return -KEY_getpeername;
5051 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5052 if (strEQ(d,"priority")) return -KEY_getpriority;
5055 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5058 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5062 else if (*d == 'h') {
5063 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5064 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5065 if (strEQ(d,"hostent")) return -KEY_gethostent;
5067 else if (*d == 'n') {
5068 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5069 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5070 if (strEQ(d,"netent")) return -KEY_getnetent;
5072 else if (*d == 's') {
5073 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5074 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5075 if (strEQ(d,"servent")) return -KEY_getservent;
5076 if (strEQ(d,"sockname")) return -KEY_getsockname;
5077 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5079 else if (*d == 'g') {
5080 if (strEQ(d,"grent")) return -KEY_getgrent;
5081 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5082 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5084 else if (*d == 'l') {
5085 if (strEQ(d,"login")) return -KEY_getlogin;
5087 else if (strEQ(d,"c")) return -KEY_getc;
5092 if (strEQ(d,"gt")) return -KEY_gt;
5093 if (strEQ(d,"ge")) return -KEY_ge;
5096 if (strEQ(d,"grep")) return KEY_grep;
5097 if (strEQ(d,"goto")) return KEY_goto;
5098 if (strEQ(d,"glob")) return KEY_glob;
5101 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5106 if (strEQ(d,"hex")) return -KEY_hex;
5109 if (strEQ(d,"INIT")) return KEY_INIT;
5114 if (strEQ(d,"if")) return KEY_if;
5117 if (strEQ(d,"int")) return -KEY_int;
5120 if (strEQ(d,"index")) return -KEY_index;
5121 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5126 if (strEQ(d,"join")) return -KEY_join;
5130 if (strEQ(d,"keys")) return KEY_keys;
5131 if (strEQ(d,"kill")) return -KEY_kill;
5136 if (strEQ(d,"LT")) { deprecate(d); return -KEY_lt;}
5137 if (strEQ(d,"LE")) { deprecate(d); return -KEY_le;}
5143 if (strEQ(d,"lt")) return -KEY_lt;
5144 if (strEQ(d,"le")) return -KEY_le;
5145 if (strEQ(d,"lc")) return -KEY_lc;
5148 if (strEQ(d,"log")) return -KEY_log;
5151 if (strEQ(d,"last")) return KEY_last;
5152 if (strEQ(d,"link")) return -KEY_link;
5153 if (strEQ(d,"lock")) return -KEY_lock;
5156 if (strEQ(d,"local")) return KEY_local;
5157 if (strEQ(d,"lstat")) return -KEY_lstat;
5160 if (strEQ(d,"length")) return -KEY_length;
5161 if (strEQ(d,"listen")) return -KEY_listen;
5164 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5167 if (strEQ(d,"localtime")) return -KEY_localtime;
5173 case 1: return KEY_m;
5175 if (strEQ(d,"my")) return KEY_my;
5178 if (strEQ(d,"map")) return KEY_map;
5181 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5184 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5185 if (strEQ(d,"msgget")) return -KEY_msgget;
5186 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5187 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5192 if (strEQ(d,"NE")) { deprecate(d); return -KEY_ne;}
5195 if (strEQ(d,"next")) return KEY_next;
5196 if (strEQ(d,"ne")) return -KEY_ne;
5197 if (strEQ(d,"not")) return -KEY_not;
5198 if (strEQ(d,"no")) return KEY_no;
5203 if (strEQ(d,"or")) return -KEY_or;
5206 if (strEQ(d,"ord")) return -KEY_ord;
5207 if (strEQ(d,"oct")) return -KEY_oct;
5208 if (strEQ(d,"our")) return KEY_our;
5211 if (strEQ(d,"open")) return -KEY_open;
5214 if (strEQ(d,"opendir")) return -KEY_opendir;
5221 if (strEQ(d,"pop")) return KEY_pop;
5222 if (strEQ(d,"pos")) return KEY_pos;
5225 if (strEQ(d,"push")) return KEY_push;
5226 if (strEQ(d,"pack")) return -KEY_pack;
5227 if (strEQ(d,"pipe")) return -KEY_pipe;
5230 if (strEQ(d,"print")) return KEY_print;
5233 if (strEQ(d,"printf")) return KEY_printf;
5236 if (strEQ(d,"package")) return KEY_package;
5239 if (strEQ(d,"prototype")) return KEY_prototype;
5244 if (strEQ(d,"q")) return KEY_q;
5245 if (strEQ(d,"qr")) return KEY_qr;
5246 if (strEQ(d,"qq")) return KEY_qq;
5247 if (strEQ(d,"qw")) return KEY_qw;
5248 if (strEQ(d,"qx")) return KEY_qx;
5250 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5255 if (strEQ(d,"ref")) return -KEY_ref;
5258 if (strEQ(d,"read")) return -KEY_read;
5259 if (strEQ(d,"rand")) return -KEY_rand;
5260 if (strEQ(d,"recv")) return -KEY_recv;
5261 if (strEQ(d,"redo")) return KEY_redo;
5264 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5265 if (strEQ(d,"reset")) return -KEY_reset;
5268 if (strEQ(d,"return")) return KEY_return;
5269 if (strEQ(d,"rename")) return -KEY_rename;
5270 if (strEQ(d,"rindex")) return -KEY_rindex;
5273 if (strEQ(d,"require")) return -KEY_require;
5274 if (strEQ(d,"reverse")) return -KEY_reverse;
5275 if (strEQ(d,"readdir")) return -KEY_readdir;
5278 if (strEQ(d,"readlink")) return -KEY_readlink;
5279 if (strEQ(d,"readline")) return -KEY_readline;
5280 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5283 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5288 if (strEQ(d,"STOP")) return KEY_STOP;
5292 case 0: return KEY_s;
5294 if (strEQ(d,"scalar")) return KEY_scalar;
5299 if (strEQ(d,"seek")) return -KEY_seek;
5300 if (strEQ(d,"send")) return -KEY_send;
5303 if (strEQ(d,"semop")) return -KEY_semop;
5306 if (strEQ(d,"select")) return -KEY_select;
5307 if (strEQ(d,"semctl")) return -KEY_semctl;
5308 if (strEQ(d,"semget")) return -KEY_semget;
5311 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5312 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5315 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5316 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5319 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5322 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5323 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5324 if (strEQ(d,"setservent")) return -KEY_setservent;
5327 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5328 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5335 if (strEQ(d,"shift")) return KEY_shift;
5338 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5339 if (strEQ(d,"shmget")) return -KEY_shmget;
5342 if (strEQ(d,"shmread")) return -KEY_shmread;
5345 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5346 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5351 if (strEQ(d,"sin")) return -KEY_sin;
5354 if (strEQ(d,"sleep")) return -KEY_sleep;
5357 if (strEQ(d,"sort")) return KEY_sort;
5358 if (strEQ(d,"socket")) return -KEY_socket;
5359 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5362 if (strEQ(d,"split")) return KEY_split;
5363 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5364 if (strEQ(d,"splice")) return KEY_splice;
5367 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5370 if (strEQ(d,"srand")) return -KEY_srand;
5373 if (strEQ(d,"stat")) return -KEY_stat;
5374 if (strEQ(d,"study")) return KEY_study;
5377 if (strEQ(d,"substr")) return -KEY_substr;
5378 if (strEQ(d,"sub")) return KEY_sub;
5383 if (strEQ(d,"system")) return -KEY_system;
5386 if (strEQ(d,"symlink")) return -KEY_symlink;
5387 if (strEQ(d,"syscall")) return -KEY_syscall;
5388 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5389 if (strEQ(d,"sysread")) return -KEY_sysread;
5390 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5393 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5402 if (strEQ(d,"tr")) return KEY_tr;
5405 if (strEQ(d,"tie")) return KEY_tie;
5408 if (strEQ(d,"tell")) return -KEY_tell;
5409 if (strEQ(d,"tied")) return KEY_tied;
5410 if (strEQ(d,"time")) return -KEY_time;
5413 if (strEQ(d,"times")) return -KEY_times;
5416 if (strEQ(d,"telldir")) return -KEY_telldir;
5419 if (strEQ(d,"truncate")) return -KEY_truncate;
5426 if (strEQ(d,"uc")) return -KEY_uc;
5429 if (strEQ(d,"use")) return KEY_use;
5432 if (strEQ(d,"undef")) return KEY_undef;
5433 if (strEQ(d,"until")) return KEY_until;
5434 if (strEQ(d,"untie")) return KEY_untie;
5435 if (strEQ(d,"utime")) return -KEY_utime;
5436 if (strEQ(d,"umask")) return -KEY_umask;
5439 if (strEQ(d,"unless")) return KEY_unless;
5440 if (strEQ(d,"unpack")) return -KEY_unpack;
5441 if (strEQ(d,"unlink")) return -KEY_unlink;
5444 if (strEQ(d,"unshift")) return KEY_unshift;
5445 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5450 if (strEQ(d,"values")) return -KEY_values;
5451 if (strEQ(d,"vec")) return -KEY_vec;
5456 if (strEQ(d,"warn")) return -KEY_warn;
5457 if (strEQ(d,"wait")) return -KEY_wait;
5460 if (strEQ(d,"while")) return KEY_while;
5461 if (strEQ(d,"write")) return -KEY_write;
5464 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5467 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5472 if (len == 1) return -KEY_x;
5473 if (strEQ(d,"xor")) return -KEY_xor;
5476 if (len == 1) return KEY_y;
5485 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5489 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5490 dTHR; /* only for ckWARN */
5491 if (ckWARN(WARN_SYNTAX)) {
5493 for (w = s+2; *w && level; w++) {
5500 for (; *w && isSPACE(*w); w++) ;
5501 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5502 Perl_warner(aTHX_ WARN_SYNTAX,
5503 "%s (...) interpreted as function",name);
5506 while (s < PL_bufend && isSPACE(*s))
5510 while (s < PL_bufend && isSPACE(*s))
5512 if (isIDFIRST_lazy(s)) {
5514 while (isALNUM_lazy(s))
5516 while (s < PL_bufend && isSPACE(*s))
5521 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5525 Perl_croak(aTHX_ "No comma allowed after %s", what);
5530 /* Either returns sv, or mortalizes sv and returns a new SV*.
5531 Best used as sv=new_constant(..., sv, ...).
5532 If s, pv are NULL, calls subroutine with one argument,
5533 and type is used with error messages only. */
5536 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5540 HV *table = GvHV(PL_hintgv); /* ^H */
5544 const char *why, *why1, *why2;
5546 if (!(PL_hints & HINT_LOCALIZE_HH)) {
5549 why = "%^H is not localized";
5553 msg = Perl_newSVpvf(aTHX_ "constant(%s): %s%s%s",
5554 (type ? type: "undef"), why1, why2, why);
5555 yyerror(SvPVX(msg));
5560 why = "%^H is not defined";
5563 cvp = hv_fetch(table, key, strlen(key), FALSE);
5564 if (!cvp || !SvOK(*cvp)) {
5565 why = "} is not defined";
5570 sv_2mortal(sv); /* Parent created it permanently */
5573 pv = sv_2mortal(newSVpvn(s, len));
5575 typesv = sv_2mortal(newSVpv(type, 0));
5577 typesv = &PL_sv_undef;
5579 PUSHSTACKi(PERLSI_OVERLOAD);
5592 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5596 /* Check the eval first */
5597 if (!PL_in_eval && SvTRUE(ERRSV))
5600 sv_catpv(ERRSV, "Propagated");
5601 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5603 res = SvREFCNT_inc(sv);
5607 (void)SvREFCNT_inc(res);
5616 why = "}} did not return a defined value";
5617 why1 = "Call to &{$^H{";
5627 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
5629 register char *d = dest;
5630 register char *e = d + destlen - 3; /* two-character token, ending NUL */
5633 Perl_croak(aTHX_ ident_too_long);
5634 if (isALNUM(*s)) /* UTF handled below */
5636 else if (*s == '\'' && allow_package && isIDFIRST_lazy(s+1)) {
5641 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
5645 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5646 char *t = s + UTF8SKIP(s);
5647 while (*t & 0x80 && is_utf8_mark((U8*)t))
5649 if (d + (t - s) > e)
5650 Perl_croak(aTHX_ ident_too_long);
5651 Copy(s, d, t - s, char);
5664 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
5674 e = d + destlen - 3; /* two-character token, ending NUL */
5676 while (isDIGIT(*s)) {
5678 Perl_croak(aTHX_ ident_too_long);
5685 Perl_croak(aTHX_ ident_too_long);
5686 if (isALNUM(*s)) /* UTF handled below */
5688 else if (*s == '\'' && isIDFIRST_lazy(s+1)) {
5693 else if (*s == ':' && s[1] == ':') {
5697 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5698 char *t = s + UTF8SKIP(s);
5699 while (*t & 0x80 && is_utf8_mark((U8*)t))
5701 if (d + (t - s) > e)
5702 Perl_croak(aTHX_ ident_too_long);
5703 Copy(s, d, t - s, char);
5714 if (PL_lex_state != LEX_NORMAL)
5715 PL_lex_state = LEX_INTERPENDMAYBE;
5718 if (*s == '$' && s[1] &&
5719 (isALNUM_lazy(s+1) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
5732 if (*d == '^' && *s && isCONTROLVAR(*s)) {
5737 if (isSPACE(s[-1])) {
5740 if (ch != ' ' && ch != '\t') {
5746 if (isIDFIRST_lazy(d)) {
5750 while (e < send && isALNUM_lazy(e) || *e == ':') {
5752 while (e < send && *e & 0x80 && is_utf8_mark((U8*)e))
5755 Copy(s, d, e - s, char);
5760 while ((isALNUM(*s) || *s == ':') && d < e)
5763 Perl_croak(aTHX_ ident_too_long);
5766 while (s < send && (*s == ' ' || *s == '\t')) s++;
5767 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
5768 dTHR; /* only for ckWARN */
5769 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
5770 const char *brack = *s == '[' ? "[...]" : "{...}";
5771 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5772 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
5773 funny, dest, brack, funny, dest, brack);
5776 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
5780 /* Handle extended ${^Foo} variables
5781 * 1999-02-27 mjd-perl-patch@plover.com */
5782 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
5786 while (isALNUM(*s) && d < e) {
5790 Perl_croak(aTHX_ ident_too_long);
5795 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
5796 PL_lex_state = LEX_INTERPEND;
5799 if (PL_lex_state == LEX_NORMAL) {
5800 dTHR; /* only for ckWARN */
5801 if (ckWARN(WARN_AMBIGUOUS) &&
5802 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
5804 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5805 "Ambiguous use of %c{%s} resolved to %c%s",
5806 funny, dest, funny, dest);
5811 s = bracket; /* let the parser handle it */
5815 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
5816 PL_lex_state = LEX_INTERPEND;
5821 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
5826 *pmfl |= PMf_GLOBAL;
5828 *pmfl |= PMf_CONTINUE;
5832 *pmfl |= PMf_MULTILINE;
5834 *pmfl |= PMf_SINGLELINE;
5836 *pmfl |= PMf_EXTENDED;
5840 S_scan_pat(pTHX_ char *start, I32 type)
5845 s = scan_str(start,FALSE,FALSE);
5848 SvREFCNT_dec(PL_lex_stuff);
5849 PL_lex_stuff = Nullsv;
5850 Perl_croak(aTHX_ "Search pattern not terminated");
5853 pm = (PMOP*)newPMOP(type, 0);
5854 if (PL_multi_open == '?')
5855 pm->op_pmflags |= PMf_ONCE;
5857 while (*s && strchr("iomsx", *s))
5858 pmflag(&pm->op_pmflags,*s++);
5861 while (*s && strchr("iogcmsx", *s))
5862 pmflag(&pm->op_pmflags,*s++);
5864 pm->op_pmpermflags = pm->op_pmflags;
5866 PL_lex_op = (OP*)pm;
5867 yylval.ival = OP_MATCH;
5872 S_scan_subst(pTHX_ char *start)
5879 yylval.ival = OP_NULL;
5881 s = scan_str(start,FALSE,FALSE);
5885 SvREFCNT_dec(PL_lex_stuff);
5886 PL_lex_stuff = Nullsv;
5887 Perl_croak(aTHX_ "Substitution pattern not terminated");
5890 if (s[-1] == PL_multi_open)
5893 first_start = PL_multi_start;
5894 s = scan_str(s,FALSE,FALSE);
5897 SvREFCNT_dec(PL_lex_stuff);
5898 PL_lex_stuff = Nullsv;
5900 SvREFCNT_dec(PL_lex_repl);
5901 PL_lex_repl = Nullsv;
5902 Perl_croak(aTHX_ "Substitution replacement not terminated");
5904 PL_multi_start = first_start; /* so whole substitution is taken together */
5906 pm = (PMOP*)newPMOP(OP_SUBST, 0);
5912 else if (strchr("iogcmsx", *s))
5913 pmflag(&pm->op_pmflags,*s++);
5920 PL_sublex_info.super_bufptr = s;
5921 PL_sublex_info.super_bufend = PL_bufend;
5923 pm->op_pmflags |= PMf_EVAL;
5924 repl = newSVpvn("",0);
5926 sv_catpv(repl, es ? "eval " : "do ");
5927 sv_catpvn(repl, "{ ", 2);
5928 sv_catsv(repl, PL_lex_repl);
5929 sv_catpvn(repl, " };", 2);
5931 SvREFCNT_dec(PL_lex_repl);
5935 pm->op_pmpermflags = pm->op_pmflags;
5936 PL_lex_op = (OP*)pm;
5937 yylval.ival = OP_SUBST;
5942 S_scan_trans(pTHX_ char *start)
5953 yylval.ival = OP_NULL;
5955 s = scan_str(start,FALSE,FALSE);
5958 SvREFCNT_dec(PL_lex_stuff);
5959 PL_lex_stuff = Nullsv;
5960 Perl_croak(aTHX_ "Transliteration pattern not terminated");
5962 if (s[-1] == PL_multi_open)
5965 s = scan_str(s,FALSE,FALSE);
5968 SvREFCNT_dec(PL_lex_stuff);
5969 PL_lex_stuff = Nullsv;
5971 SvREFCNT_dec(PL_lex_repl);
5972 PL_lex_repl = Nullsv;
5973 Perl_croak(aTHX_ "Transliteration replacement not terminated");
5977 o = newSVOP(OP_TRANS, 0, 0);
5978 utf8 = OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF;
5981 New(803,tbl,256,short);
5982 o = newPVOP(OP_TRANS, 0, (char*)tbl);
5986 complement = del = squash = 0;
5987 while (strchr("cdsCU", *s)) {
5989 complement = OPpTRANS_COMPLEMENT;
5991 del = OPpTRANS_DELETE;
5993 squash = OPpTRANS_SQUASH;
5998 utf8 &= ~OPpTRANS_FROM_UTF;
6000 utf8 |= OPpTRANS_FROM_UTF;
6004 utf8 &= ~OPpTRANS_TO_UTF;
6006 utf8 |= OPpTRANS_TO_UTF;
6009 Perl_croak(aTHX_ "Too many /C and /U options");
6014 o->op_private = del|squash|complement|utf8;
6017 yylval.ival = OP_TRANS;
6022 S_scan_heredoc(pTHX_ register char *s)
6026 I32 op_type = OP_SCALAR;
6033 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6037 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6040 for (peek = s; *peek == ' ' || *peek == '\t'; peek++) ;
6041 if (*peek && strchr("`'\"",*peek)) {
6044 s = delimcpy(d, e, s, PL_bufend, term, &len);
6054 if (!isALNUM_lazy(s))
6055 deprecate("bare << to mean <<\"\"");
6056 for (; isALNUM_lazy(s); s++) {
6061 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6062 Perl_croak(aTHX_ "Delimiter for here document is too long");
6065 len = d - PL_tokenbuf;
6066 #ifndef PERL_STRICT_CR
6067 d = strchr(s, '\r');
6071 while (s < PL_bufend) {
6077 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6086 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6091 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6092 herewas = newSVpvn(s,PL_bufend-s);
6094 s--, herewas = newSVpvn(s,d-s);
6095 s += SvCUR(herewas);
6097 tmpstr = NEWSV(87,79);
6098 sv_upgrade(tmpstr, SVt_PVIV);
6103 else if (term == '`') {
6104 op_type = OP_BACKTICK;
6105 SvIVX(tmpstr) = '\\';
6109 PL_multi_start = CopLINE(PL_curcop);
6110 PL_multi_open = PL_multi_close = '<';
6111 term = *PL_tokenbuf;
6112 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6113 char *bufptr = PL_sublex_info.super_bufptr;
6114 char *bufend = PL_sublex_info.super_bufend;
6115 char *olds = s - SvCUR(herewas);
6116 s = strchr(bufptr, '\n');
6120 while (s < bufend &&
6121 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6123 CopLINE_inc(PL_curcop);
6126 CopLINE_set(PL_curcop, PL_multi_start);
6127 missingterm(PL_tokenbuf);
6129 sv_setpvn(herewas,bufptr,d-bufptr+1);
6130 sv_setpvn(tmpstr,d+1,s-d);
6132 sv_catpvn(herewas,s,bufend-s);
6133 (void)strcpy(bufptr,SvPVX(herewas));
6140 while (s < PL_bufend &&
6141 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6143 CopLINE_inc(PL_curcop);
6145 if (s >= PL_bufend) {
6146 CopLINE_set(PL_curcop, PL_multi_start);
6147 missingterm(PL_tokenbuf);
6149 sv_setpvn(tmpstr,d+1,s-d);
6151 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6153 sv_catpvn(herewas,s,PL_bufend-s);
6154 sv_setsv(PL_linestr,herewas);
6155 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6156 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6159 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6160 while (s >= PL_bufend) { /* multiple line string? */
6162 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6163 CopLINE_set(PL_curcop, PL_multi_start);
6164 missingterm(PL_tokenbuf);
6166 CopLINE_inc(PL_curcop);
6167 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6168 #ifndef PERL_STRICT_CR
6169 if (PL_bufend - PL_linestart >= 2) {
6170 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6171 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6173 PL_bufend[-2] = '\n';
6175 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6177 else if (PL_bufend[-1] == '\r')
6178 PL_bufend[-1] = '\n';
6180 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6181 PL_bufend[-1] = '\n';
6183 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6184 SV *sv = NEWSV(88,0);
6186 sv_upgrade(sv, SVt_PVMG);
6187 sv_setsv(sv,PL_linestr);
6188 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6190 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6193 sv_catsv(PL_linestr,herewas);
6194 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6198 sv_catsv(tmpstr,PL_linestr);
6203 PL_multi_end = CopLINE(PL_curcop);
6204 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6205 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6206 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6208 SvREFCNT_dec(herewas);
6209 PL_lex_stuff = tmpstr;
6210 yylval.ival = op_type;
6215 takes: current position in input buffer
6216 returns: new position in input buffer
6217 side-effects: yylval and lex_op are set.
6222 <FH> read from filehandle
6223 <pkg::FH> read from package qualified filehandle
6224 <pkg'FH> read from package qualified filehandle
6225 <$fh> read from filehandle in $fh
6231 S_scan_inputsymbol(pTHX_ char *start)
6233 register char *s = start; /* current position in buffer */
6239 d = PL_tokenbuf; /* start of temp holding space */
6240 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6241 end = strchr(s, '\n');
6244 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6246 /* die if we didn't have space for the contents of the <>,
6247 or if it didn't end, or if we see a newline
6250 if (len >= sizeof PL_tokenbuf)
6251 Perl_croak(aTHX_ "Excessively long <> operator");
6253 Perl_croak(aTHX_ "Unterminated <> operator");
6258 Remember, only scalar variables are interpreted as filehandles by
6259 this code. Anything more complex (e.g., <$fh{$num}>) will be
6260 treated as a glob() call.
6261 This code makes use of the fact that except for the $ at the front,
6262 a scalar variable and a filehandle look the same.
6264 if (*d == '$' && d[1]) d++;
6266 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6267 while (*d && (isALNUM_lazy(d) || *d == '\'' || *d == ':'))
6270 /* If we've tried to read what we allow filehandles to look like, and
6271 there's still text left, then it must be a glob() and not a getline.
6272 Use scan_str to pull out the stuff between the <> and treat it
6273 as nothing more than a string.
6276 if (d - PL_tokenbuf != len) {
6277 yylval.ival = OP_GLOB;
6279 s = scan_str(start,FALSE,FALSE);
6281 Perl_croak(aTHX_ "Glob not terminated");
6285 /* we're in a filehandle read situation */
6288 /* turn <> into <ARGV> */
6290 (void)strcpy(d,"ARGV");
6292 /* if <$fh>, create the ops to turn the variable into a
6298 /* try to find it in the pad for this block, otherwise find
6299 add symbol table ops
6301 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6302 OP *o = newOP(OP_PADSV, 0);
6304 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o);
6307 GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV);
6308 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0,
6309 newUNOP(OP_RV2SV, 0,
6310 newGVOP(OP_GV, 0, gv)));
6312 PL_lex_op->op_flags |= OPf_SPECIAL;
6313 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6314 yylval.ival = OP_NULL;
6317 /* If it's none of the above, it must be a literal filehandle
6318 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6320 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6321 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6322 yylval.ival = OP_NULL;
6331 takes: start position in buffer
6332 keep_quoted preserve \ on the embedded delimiter(s)
6333 keep_delims preserve the delimiters around the string
6334 returns: position to continue reading from buffer
6335 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6336 updates the read buffer.
6338 This subroutine pulls a string out of the input. It is called for:
6339 q single quotes q(literal text)
6340 ' single quotes 'literal text'
6341 qq double quotes qq(interpolate $here please)
6342 " double quotes "interpolate $here please"
6343 qx backticks qx(/bin/ls -l)
6344 ` backticks `/bin/ls -l`
6345 qw quote words @EXPORT_OK = qw( func() $spam )
6346 m// regexp match m/this/
6347 s/// regexp substitute s/this/that/
6348 tr/// string transliterate tr/this/that/
6349 y/// string transliterate y/this/that/
6350 ($*@) sub prototypes sub foo ($)
6351 (stuff) sub attr parameters sub foo : attr(stuff)
6352 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6354 In most of these cases (all but <>, patterns and transliterate)
6355 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6356 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6357 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6360 It skips whitespace before the string starts, and treats the first
6361 character as the delimiter. If the delimiter is one of ([{< then
6362 the corresponding "close" character )]}> is used as the closing
6363 delimiter. It allows quoting of delimiters, and if the string has
6364 balanced delimiters ([{<>}]) it allows nesting.
6366 The lexer always reads these strings into lex_stuff, except in the
6367 case of the operators which take *two* arguments (s/// and tr///)
6368 when it checks to see if lex_stuff is full (presumably with the 1st
6369 arg to s or tr) and if so puts the string into lex_repl.
6374 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6377 SV *sv; /* scalar value: string */
6378 char *tmps; /* temp string, used for delimiter matching */
6379 register char *s = start; /* current position in the buffer */
6380 register char term; /* terminating character */
6381 register char *to; /* current position in the sv's data */
6382 I32 brackets = 1; /* bracket nesting level */
6384 /* skip space before the delimiter */
6388 /* mark where we are, in case we need to report errors */
6391 /* after skipping whitespace, the next character is the terminator */
6393 /* mark where we are */
6394 PL_multi_start = CopLINE(PL_curcop);
6395 PL_multi_open = term;
6397 /* find corresponding closing delimiter */
6398 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6400 PL_multi_close = term;
6402 /* create a new SV to hold the contents. 87 is leak category, I'm
6403 assuming. 79 is the SV's initial length. What a random number. */
6405 sv_upgrade(sv, SVt_PVIV);
6407 (void)SvPOK_only(sv); /* validate pointer */
6409 /* move past delimiter and try to read a complete string */
6411 sv_catpvn(sv, s, 1);
6414 /* extend sv if need be */
6415 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6416 /* set 'to' to the next character in the sv's string */
6417 to = SvPVX(sv)+SvCUR(sv);
6419 /* if open delimiter is the close delimiter read unbridle */
6420 if (PL_multi_open == PL_multi_close) {
6421 for (; s < PL_bufend; s++,to++) {
6422 /* embedded newlines increment the current line number */
6423 if (*s == '\n' && !PL_rsfp)
6424 CopLINE_inc(PL_curcop);
6425 /* handle quoted delimiters */
6426 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6427 if (!keep_quoted && s[1] == term)
6429 /* any other quotes are simply copied straight through */
6433 /* terminate when run out of buffer (the for() condition), or
6434 have found the terminator */
6435 else if (*s == term)
6441 /* if the terminator isn't the same as the start character (e.g.,
6442 matched brackets), we have to allow more in the quoting, and
6443 be prepared for nested brackets.
6446 /* read until we run out of string, or we find the terminator */
6447 for (; s < PL_bufend; s++,to++) {
6448 /* embedded newlines increment the line count */
6449 if (*s == '\n' && !PL_rsfp)
6450 CopLINE_inc(PL_curcop);
6451 /* backslashes can escape the open or closing characters */
6452 if (*s == '\\' && s+1 < PL_bufend) {
6454 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6459 /* allow nested opens and closes */
6460 else if (*s == PL_multi_close && --brackets <= 0)
6462 else if (*s == PL_multi_open)
6467 /* terminate the copied string and update the sv's end-of-string */
6469 SvCUR_set(sv, to - SvPVX(sv));
6472 * this next chunk reads more into the buffer if we're not done yet
6475 if (s < PL_bufend) break; /* handle case where we are done yet :-) */
6477 #ifndef PERL_STRICT_CR
6478 if (to - SvPVX(sv) >= 2) {
6479 if ((to[-2] == '\r' && to[-1] == '\n') ||
6480 (to[-2] == '\n' && to[-1] == '\r'))
6484 SvCUR_set(sv, to - SvPVX(sv));
6486 else if (to[-1] == '\r')
6489 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6493 /* if we're out of file, or a read fails, bail and reset the current
6494 line marker so we can report where the unterminated string began
6497 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6499 CopLINE_set(PL_curcop, PL_multi_start);
6502 /* we read a line, so increment our line counter */
6503 CopLINE_inc(PL_curcop);
6505 /* update debugger info */
6506 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6507 SV *sv = NEWSV(88,0);
6509 sv_upgrade(sv, SVt_PVMG);
6510 sv_setsv(sv,PL_linestr);
6511 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6514 /* having changed the buffer, we must update PL_bufend */
6515 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6518 /* at this point, we have successfully read the delimited string */
6521 sv_catpvn(sv, s, 1);
6522 PL_multi_end = CopLINE(PL_curcop);
6525 /* if we allocated too much space, give some back */
6526 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6527 SvLEN_set(sv, SvCUR(sv) + 1);
6528 Renew(SvPVX(sv), SvLEN(sv), char);
6531 /* decide whether this is the first or second quoted string we've read
6544 takes: pointer to position in buffer
6545 returns: pointer to new position in buffer
6546 side-effects: builds ops for the constant in yylval.op
6548 Read a number in any of the formats that Perl accepts:
6550 0(x[0-7A-F]+)|([0-7]+)|(b[01])
6551 [\d_]+(\.[\d_]*)?[Ee](\d+)
6553 Underbars (_) are allowed in decimal numbers. If -w is on,
6554 underbars before a decimal point must be at three digit intervals.
6556 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6559 If it reads a number without a decimal point or an exponent, it will
6560 try converting the number to an integer and see if it can do so
6561 without loss of precision.
6565 Perl_scan_num(pTHX_ char *start)
6567 register char *s = start; /* current position in buffer */
6568 register char *d; /* destination in temp buffer */
6569 register char *e; /* end of temp buffer */
6570 IV tryiv; /* used to see if it can be an IV */
6571 NV value; /* number read, as a double */
6572 SV *sv = Nullsv; /* place to put the converted number */
6573 bool floatit; /* boolean: int or float? */
6574 char *lastub = 0; /* position of last underbar */
6575 static char number_too_long[] = "Number too long";
6577 /* We use the first character to decide what type of number this is */
6581 Perl_croak(aTHX_ "panic: scan_num");
6583 /* if it starts with a 0, it could be an octal number, a decimal in
6584 0.13 disguise, or a hexadecimal number, or a binary number. */
6588 u holds the "number so far"
6589 shift the power of 2 of the base
6590 (hex == 4, octal == 3, binary == 1)
6591 overflowed was the number more than we can hold?
6593 Shift is used when we add a digit. It also serves as an "are
6594 we in octal/hex/binary?" indicator to disallow hex characters
6601 bool overflowed = FALSE;
6602 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6603 static char* bases[5] = { "", "binary", "", "octal",
6605 static char* Bases[5] = { "", "Binary", "", "Octal",
6607 static char *maxima[5] = { "",
6608 "0b11111111111111111111111111111111",
6612 char *base, *Base, *max;
6618 } else if (s[1] == 'b') {
6622 /* check for a decimal in disguise */
6623 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
6625 /* so it must be octal */
6629 base = bases[shift];
6630 Base = Bases[shift];
6631 max = maxima[shift];
6633 /* read the rest of the number */
6635 /* x is used in the overflow test,
6636 b is the digit we're adding on. */
6641 /* if we don't mention it, we're done */
6650 /* 8 and 9 are not octal */
6653 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
6657 case '2': case '3': case '4':
6658 case '5': case '6': case '7':
6660 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
6664 b = *s++ & 15; /* ASCII digit -> value of digit */
6668 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
6669 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
6670 /* make sure they said 0x */
6675 /* Prepare to put the digit we have onto the end
6676 of the number so far. We check for overflows.
6681 x = u << shift; /* make room for the digit */
6683 if ((x >> shift) != u
6684 && !(PL_hints & HINT_NEW_BINARY)) {
6688 if (ckWARN_d(WARN_OVERFLOW))
6689 Perl_warner(aTHX_ WARN_OVERFLOW,
6690 "Integer overflow in %s number",
6693 u = x | b; /* add the digit to the end */
6696 n *= nvshift[shift];
6697 /* If an NV has not enough bits in its
6698 * mantissa to represent an UV this summing of
6699 * small low-order numbers is a waste of time
6700 * (because the NV cannot preserve the
6701 * low-order bits anyway): we could just
6702 * remember when did we overflow and in the
6703 * end just multiply n by the right
6711 /* if we get here, we had success: make a scalar value from
6718 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
6719 Perl_warner(aTHX_ WARN_PORTABLE,
6720 "%s number > %s non-portable",
6727 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
6728 Perl_warner(aTHX_ WARN_PORTABLE,
6729 "%s number > %s non-portable",
6734 if (PL_hints & HINT_NEW_BINARY)
6735 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
6740 handle decimal numbers.
6741 we're also sent here when we read a 0 as the first digit
6743 case '1': case '2': case '3': case '4': case '5':
6744 case '6': case '7': case '8': case '9': case '.':
6747 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
6750 /* read next group of digits and _ and copy into d */
6751 while (isDIGIT(*s) || *s == '_') {
6752 /* skip underscores, checking for misplaced ones
6756 dTHR; /* only for ckWARN */
6757 if (ckWARN(WARN_SYNTAX) && lastub && s - lastub != 3)
6758 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6762 /* check for end of fixed-length buffer */
6764 Perl_croak(aTHX_ number_too_long);
6765 /* if we're ok, copy the character */
6770 /* final misplaced underbar check */
6771 if (lastub && s - lastub != 3) {
6773 if (ckWARN(WARN_SYNTAX))
6774 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6777 /* read a decimal portion if there is one. avoid
6778 3..5 being interpreted as the number 3. followed
6781 if (*s == '.' && s[1] != '.') {
6785 /* copy, ignoring underbars, until we run out of
6786 digits. Note: no misplaced underbar checks!
6788 for (; isDIGIT(*s) || *s == '_'; s++) {
6789 /* fixed length buffer check */
6791 Perl_croak(aTHX_ number_too_long);
6797 /* read exponent part, if present */
6798 if (*s && strchr("eE",*s) && strchr("+-0123456789",s[1])) {
6802 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
6803 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
6805 /* allow positive or negative exponent */
6806 if (*s == '+' || *s == '-')
6809 /* read digits of exponent (no underbars :-) */
6810 while (isDIGIT(*s)) {
6812 Perl_croak(aTHX_ number_too_long);
6817 /* terminate the string */
6820 /* make an sv from the string */
6823 value = Atof(PL_tokenbuf);
6826 See if we can make do with an integer value without loss of
6827 precision. We use I_V to cast to an int, because some
6828 compilers have issues. Then we try casting it back and see
6829 if it was the same. We only do this if we know we
6830 specifically read an integer.
6832 Note: if floatit is true, then we don't need to do the
6836 if (!floatit && (NV)tryiv == value)
6837 sv_setiv(sv, tryiv);
6839 sv_setnv(sv, value);
6840 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
6841 (PL_hints & HINT_NEW_INTEGER) )
6842 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
6843 (floatit ? "float" : "integer"),
6846 /* if it starts with a v, it could be a version number */
6851 while (isDIGIT(*pos))
6853 if (*pos == '.' && isDIGIT(pos[1])) {
6858 s++; /* get past 'v' */
6861 SvUPGRADE(sv, SVt_PVNV);
6862 sv_setpvn(sv, "", 0);
6867 while (isDIGIT(*pos))
6870 tmpend = uv_to_utf8(tmpbuf, rev);
6872 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
6874 SvNVX(sv) += (NV)rev/nshift;
6876 } while (*pos == '.' && isDIGIT(pos[1]));
6880 tmpend = uv_to_utf8(tmpbuf, rev);
6882 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
6884 SvNVX(sv) += (NV)rev/nshift;
6895 /* make the op for the constant and return */
6898 yylval.opval = newSVOP(OP_CONST, 0, sv);
6900 yylval.opval = Nullop;
6906 S_scan_formline(pTHX_ register char *s)
6911 SV *stuff = newSVpvn("",0);
6912 bool needargs = FALSE;
6915 if (*s == '.' || *s == '}') {
6917 #ifdef PERL_STRICT_CR
6918 for (t = s+1;*t == ' ' || *t == '\t'; t++) ;
6920 for (t = s+1;*t == ' ' || *t == '\t' || *t == '\r'; t++) ;
6922 if (*t == '\n' || t == PL_bufend)
6925 if (PL_in_eval && !PL_rsfp) {
6926 eol = strchr(s,'\n');
6931 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6933 for (t = s; t < eol; t++) {
6934 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
6936 goto enough; /* ~~ must be first line in formline */
6938 if (*t == '@' || *t == '^')
6941 sv_catpvn(stuff, s, eol-s);
6942 #ifndef PERL_STRICT_CR
6943 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
6944 char *end = SvPVX(stuff) + SvCUR(stuff);
6953 s = filter_gets(PL_linestr, PL_rsfp, 0);
6954 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
6955 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
6958 yyerror("Format not terminated");
6968 PL_lex_state = LEX_NORMAL;
6969 PL_nextval[PL_nexttoke].ival = 0;
6973 PL_lex_state = LEX_FORMLINE;
6974 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
6976 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
6980 SvREFCNT_dec(stuff);
6981 PL_lex_formbrack = 0;
6992 PL_cshlen = strlen(PL_cshname);
6997 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7000 I32 oldsavestack_ix = PL_savestack_ix;
7001 CV* outsidecv = PL_compcv;
7005 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7007 SAVEI32(PL_subline);
7008 save_item(PL_subname);
7010 SAVEVPTR(PL_curpad);
7011 SAVESPTR(PL_comppad);
7012 SAVESPTR(PL_comppad_name);
7013 SAVESPTR(PL_compcv);
7014 SAVEI32(PL_comppad_name_fill);
7015 SAVEI32(PL_min_intro_pending);
7016 SAVEI32(PL_max_intro_pending);
7017 SAVEI32(PL_pad_reset_pending);
7019 PL_compcv = (CV*)NEWSV(1104,0);
7020 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7021 CvFLAGS(PL_compcv) |= flags;
7023 PL_comppad = newAV();
7024 av_push(PL_comppad, Nullsv);
7025 PL_curpad = AvARRAY(PL_comppad);
7026 PL_comppad_name = newAV();
7027 PL_comppad_name_fill = 0;
7028 PL_min_intro_pending = 0;
7030 PL_subline = CopLINE(PL_curcop);
7032 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
7033 PL_curpad[0] = (SV*)newAV();
7034 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
7035 #endif /* USE_THREADS */
7037 comppadlist = newAV();
7038 AvREAL_off(comppadlist);
7039 av_store(comppadlist, 0, (SV*)PL_comppad_name);
7040 av_store(comppadlist, 1, (SV*)PL_comppad);
7042 CvPADLIST(PL_compcv) = comppadlist;
7043 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7045 CvOWNER(PL_compcv) = 0;
7046 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7047 MUTEX_INIT(CvMUTEXP(PL_compcv));
7048 #endif /* USE_THREADS */
7050 return oldsavestack_ix;
7054 Perl_yywarn(pTHX_ char *s)
7057 PL_in_eval |= EVAL_WARNONLY;
7059 PL_in_eval &= ~EVAL_WARNONLY;
7064 Perl_yyerror(pTHX_ char *s)
7068 char *context = NULL;
7072 if (!yychar || (yychar == ';' && !PL_rsfp))
7074 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7075 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7076 while (isSPACE(*PL_oldoldbufptr))
7078 context = PL_oldoldbufptr;
7079 contlen = PL_bufptr - PL_oldoldbufptr;
7081 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7082 PL_oldbufptr != PL_bufptr) {
7083 while (isSPACE(*PL_oldbufptr))
7085 context = PL_oldbufptr;
7086 contlen = PL_bufptr - PL_oldbufptr;
7088 else if (yychar > 255)
7089 where = "next token ???";
7090 else if ((yychar & 127) == 127) {
7091 if (PL_lex_state == LEX_NORMAL ||
7092 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7093 where = "at end of line";
7094 else if (PL_lex_inpat)
7095 where = "within pattern";
7097 where = "within string";
7100 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7102 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7103 else if (isPRINT_LC(yychar))
7104 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7106 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7107 where = SvPVX(where_sv);
7109 msg = sv_2mortal(newSVpv(s, 0));
7110 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7111 CopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7113 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7115 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7116 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7117 Perl_sv_catpvf(aTHX_ msg,
7118 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7119 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7122 if (PL_in_eval & EVAL_WARNONLY)
7123 Perl_warn(aTHX_ "%"SVf, msg);
7126 if (PL_error_count >= 10)
7127 Perl_croak(aTHX_ "%s has too many errors.\n", CopFILE(PL_curcop));
7129 PL_in_my_stash = Nullhv;
7140 * Restore a source filter.
7144 restore_rsfp(pTHXo_ void *f)
7146 PerlIO *fp = (PerlIO*)f;
7148 if (PL_rsfp == PerlIO_stdin())
7149 PerlIO_clearerr(PL_rsfp);
7150 else if (PL_rsfp && (PL_rsfp != fp))
7151 PerlIO_close(PL_rsfp);