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 $foo::bar, so
1973 just check for colons.
1975 if it's a legal name, the OP is a PADANY.
1978 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
1979 if (strchr(PL_tokenbuf,':'))
1980 yyerror(Perl_form(aTHX_ "No package name allowed for "
1981 "variable %s in \"our\"",
1983 tmp = pad_allocmy(PL_tokenbuf);
1986 if (strchr(PL_tokenbuf,':'))
1987 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
1989 yylval.opval = newOP(OP_PADANY, 0);
1990 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
1996 build the ops for accesses to a my() variable.
1998 Deny my($a) or my($b) in a sort block, *if* $a or $b is
1999 then used in a comparison. This catches most, but not
2000 all cases. For instance, it catches
2001 sort { my($a); $a <=> $b }
2003 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
2004 (although why you'd do that is anyone's guess).
2007 if (!strchr(PL_tokenbuf,':')) {
2009 /* Check for single character per-thread SVs */
2010 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
2011 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
2012 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
2014 yylval.opval = newOP(OP_THREADSV, 0);
2015 yylval.opval->op_targ = tmp;
2018 #endif /* USE_THREADS */
2019 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
2020 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
2021 /* might be an "our" variable" */
2022 if (SvFLAGS(namesv) & SVpad_OUR) {
2023 /* build ops for a bareword */
2024 SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0);
2025 sv_catpvn(sym, "::", 2);
2026 sv_catpv(sym, PL_tokenbuf+1);
2027 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
2028 yylval.opval->op_private = OPpCONST_ENTERED;
2029 gv_fetchpv(SvPVX(sym),
2031 ? (GV_ADDMULTI | GV_ADDINEVAL)
2034 ((PL_tokenbuf[0] == '$') ? SVt_PV
2035 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2040 /* if it's a sort block and they're naming $a or $b */
2041 if (PL_last_lop_op == OP_SORT &&
2042 PL_tokenbuf[0] == '$' &&
2043 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
2046 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
2047 d < PL_bufend && *d != '\n';
2050 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
2051 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
2057 yylval.opval = newOP(OP_PADANY, 0);
2058 yylval.opval->op_targ = tmp;
2064 Whine if they've said @foo in a doublequoted string,
2065 and @foo isn't a variable we can find in the symbol
2068 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
2069 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
2070 if (!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
2071 yyerror(Perl_form(aTHX_ "In string, %s now must be written as \\%s",
2072 PL_tokenbuf, PL_tokenbuf));
2075 /* build ops for a bareword */
2076 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2077 yylval.opval->op_private = OPpCONST_ENTERED;
2078 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
2079 ((PL_tokenbuf[0] == '$') ? SVt_PV
2080 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2085 /* no identifier pending identification */
2087 switch (PL_lex_state) {
2089 case LEX_NORMAL: /* Some compilers will produce faster */
2090 case LEX_INTERPNORMAL: /* code if we comment these out. */
2094 /* when we've already built the next token, just pull it out of the queue */
2097 yylval = PL_nextval[PL_nexttoke];
2099 PL_lex_state = PL_lex_defer;
2100 PL_expect = PL_lex_expect;
2101 PL_lex_defer = LEX_NORMAL;
2103 return(PL_nexttype[PL_nexttoke]);
2105 /* interpolated case modifiers like \L \U, including \Q and \E.
2106 when we get here, PL_bufptr is at the \
2108 case LEX_INTERPCASEMOD:
2110 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2111 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2113 /* handle \E or end of string */
2114 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2118 if (PL_lex_casemods) {
2119 oldmod = PL_lex_casestack[--PL_lex_casemods];
2120 PL_lex_casestack[PL_lex_casemods] = '\0';
2122 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2124 PL_lex_state = LEX_INTERPCONCAT;
2128 if (PL_bufptr != PL_bufend)
2130 PL_lex_state = LEX_INTERPCONCAT;
2135 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2136 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2137 if (strchr("LU", *s) &&
2138 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2140 PL_lex_casestack[--PL_lex_casemods] = '\0';
2143 if (PL_lex_casemods > 10) {
2144 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2145 if (newlb != PL_lex_casestack) {
2147 PL_lex_casestack = newlb;
2150 PL_lex_casestack[PL_lex_casemods++] = *s;
2151 PL_lex_casestack[PL_lex_casemods] = '\0';
2152 PL_lex_state = LEX_INTERPCONCAT;
2153 PL_nextval[PL_nexttoke].ival = 0;
2156 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2158 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2160 PL_nextval[PL_nexttoke].ival = OP_LC;
2162 PL_nextval[PL_nexttoke].ival = OP_UC;
2164 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2166 Perl_croak(aTHX_ "panic: yylex");
2169 if (PL_lex_starts) {
2178 case LEX_INTERPPUSH:
2179 return sublex_push();
2181 case LEX_INTERPSTART:
2182 if (PL_bufptr == PL_bufend)
2183 return sublex_done();
2185 PL_lex_dojoin = (*PL_bufptr == '@');
2186 PL_lex_state = LEX_INTERPNORMAL;
2187 if (PL_lex_dojoin) {
2188 PL_nextval[PL_nexttoke].ival = 0;
2191 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2192 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2193 force_next(PRIVATEREF);
2195 force_ident("\"", '$');
2196 #endif /* USE_THREADS */
2197 PL_nextval[PL_nexttoke].ival = 0;
2199 PL_nextval[PL_nexttoke].ival = 0;
2201 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2204 if (PL_lex_starts++) {
2210 case LEX_INTERPENDMAYBE:
2211 if (intuit_more(PL_bufptr)) {
2212 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2218 if (PL_lex_dojoin) {
2219 PL_lex_dojoin = FALSE;
2220 PL_lex_state = LEX_INTERPCONCAT;
2223 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2224 && SvEVALED(PL_lex_repl))
2226 if (PL_bufptr != PL_bufend)
2227 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2228 PL_lex_repl = Nullsv;
2231 case LEX_INTERPCONCAT:
2233 if (PL_lex_brackets)
2234 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2236 if (PL_bufptr == PL_bufend)
2237 return sublex_done();
2239 if (SvIVX(PL_linestr) == '\'') {
2240 SV *sv = newSVsv(PL_linestr);
2243 else if ( PL_hints & HINT_NEW_RE )
2244 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2245 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2249 s = scan_const(PL_bufptr);
2251 PL_lex_state = LEX_INTERPCASEMOD;
2253 PL_lex_state = LEX_INTERPSTART;
2256 if (s != PL_bufptr) {
2257 PL_nextval[PL_nexttoke] = yylval;
2260 if (PL_lex_starts++)
2270 PL_lex_state = LEX_NORMAL;
2271 s = scan_formline(PL_bufptr);
2272 if (!PL_lex_formbrack)
2278 PL_oldoldbufptr = PL_oldbufptr;
2281 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2282 exp_name[PL_expect], s);
2288 if (isIDFIRST_lazy(s))
2290 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2293 goto fake_eof; /* emulate EOF on ^D or ^Z */
2298 if (PL_lex_brackets)
2299 yyerror("Missing right curly or square bracket");
2302 if (s++ < PL_bufend)
2303 goto retry; /* ignore stray nulls */
2306 if (!PL_in_eval && !PL_preambled) {
2307 PL_preambled = TRUE;
2308 sv_setpv(PL_linestr,incl_perldb());
2309 if (SvCUR(PL_linestr))
2310 sv_catpv(PL_linestr,";");
2312 while(AvFILLp(PL_preambleav) >= 0) {
2313 SV *tmpsv = av_shift(PL_preambleav);
2314 sv_catsv(PL_linestr, tmpsv);
2315 sv_catpv(PL_linestr, ";");
2318 sv_free((SV*)PL_preambleav);
2319 PL_preambleav = NULL;
2321 if (PL_minus_n || PL_minus_p) {
2322 sv_catpv(PL_linestr, "LINE: while (<>) {");
2324 sv_catpv(PL_linestr,"chomp;");
2326 GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
2328 GvIMPORTED_AV_on(gv);
2330 if (strchr("/'\"", *PL_splitstr)
2331 && strchr(PL_splitstr + 1, *PL_splitstr))
2332 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2335 s = "'~#\200\1'"; /* surely one char is unused...*/
2336 while (s[1] && strchr(PL_splitstr, *s)) s++;
2338 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c",
2339 "q" + (delim == '\''), delim);
2340 for (s = PL_splitstr; *s; s++) {
2342 sv_catpvn(PL_linestr, "\\", 1);
2343 sv_catpvn(PL_linestr, s, 1);
2345 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2349 sv_catpv(PL_linestr,"@F=split(' ');");
2352 sv_catpv(PL_linestr, "\n");
2353 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2354 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2355 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2356 SV *sv = NEWSV(85,0);
2358 sv_upgrade(sv, SVt_PVMG);
2359 sv_setsv(sv,PL_linestr);
2360 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2365 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2368 if (PL_preprocess && !PL_in_eval)
2369 (void)PerlProc_pclose(PL_rsfp);
2370 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2371 PerlIO_clearerr(PL_rsfp);
2373 (void)PerlIO_close(PL_rsfp);
2375 PL_doextract = FALSE;
2377 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2378 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2379 sv_catpv(PL_linestr,";}");
2380 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2381 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2382 PL_minus_n = PL_minus_p = 0;
2385 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2386 sv_setpv(PL_linestr,"");
2387 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2390 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2391 PL_doextract = FALSE;
2393 /* Incest with pod. */
2394 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2395 sv_setpv(PL_linestr, "");
2396 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2397 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2398 PL_doextract = FALSE;
2402 } while (PL_doextract);
2403 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2404 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2405 SV *sv = NEWSV(85,0);
2407 sv_upgrade(sv, SVt_PVMG);
2408 sv_setsv(sv,PL_linestr);
2409 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2411 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2412 if (CopLINE(PL_curcop) == 1) {
2413 while (s < PL_bufend && isSPACE(*s))
2415 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2419 if (*s == '#' && *(s+1) == '!')
2421 #ifdef ALTERNATE_SHEBANG
2423 static char as[] = ALTERNATE_SHEBANG;
2424 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2425 d = s + (sizeof(as) - 1);
2427 #endif /* ALTERNATE_SHEBANG */
2436 while (*d && !isSPACE(*d))
2440 #ifdef ARG_ZERO_IS_SCRIPT
2441 if (ipathend > ipath) {
2443 * HP-UX (at least) sets argv[0] to the script name,
2444 * which makes $^X incorrect. And Digital UNIX and Linux,
2445 * at least, set argv[0] to the basename of the Perl
2446 * interpreter. So, having found "#!", we'll set it right.
2448 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2449 assert(SvPOK(x) || SvGMAGICAL(x));
2450 if (sv_eq(x, CopFILESV(PL_curcop))) {
2451 sv_setpvn(x, ipath, ipathend - ipath);
2454 TAINT_NOT; /* $^X is always tainted, but that's OK */
2456 #endif /* ARG_ZERO_IS_SCRIPT */
2461 d = instr(s,"perl -");
2463 d = instr(s,"perl");
2465 /* avoid getting into infinite loops when shebang
2466 * line contains "Perl" rather than "perl" */
2468 for (d = ipathend-4; d >= ipath; --d) {
2469 if ((*d == 'p' || *d == 'P')
2470 && !ibcmp(d, "perl", 4))
2480 #ifdef ALTERNATE_SHEBANG
2482 * If the ALTERNATE_SHEBANG on this system starts with a
2483 * character that can be part of a Perl expression, then if
2484 * we see it but not "perl", we're probably looking at the
2485 * start of Perl code, not a request to hand off to some
2486 * other interpreter. Similarly, if "perl" is there, but
2487 * not in the first 'word' of the line, we assume the line
2488 * contains the start of the Perl program.
2490 if (d && *s != '#') {
2492 while (*c && !strchr("; \t\r\n\f\v#", *c))
2495 d = Nullch; /* "perl" not in first word; ignore */
2497 *s = '#'; /* Don't try to parse shebang line */
2499 #endif /* ALTERNATE_SHEBANG */
2504 !instr(s,"indir") &&
2505 instr(PL_origargv[0],"perl"))
2511 while (s < PL_bufend && isSPACE(*s))
2513 if (s < PL_bufend) {
2514 Newz(899,newargv,PL_origargc+3,char*);
2516 while (s < PL_bufend && !isSPACE(*s))
2519 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2522 newargv = PL_origargv;
2524 PerlProc_execv(ipath, newargv);
2525 Perl_croak(aTHX_ "Can't exec %s", ipath);
2528 U32 oldpdb = PL_perldb;
2529 bool oldn = PL_minus_n;
2530 bool oldp = PL_minus_p;
2532 while (*d && !isSPACE(*d)) d++;
2533 while (*d == ' ' || *d == '\t') d++;
2537 if (*d == 'M' || *d == 'm') {
2539 while (*d && !isSPACE(*d)) d++;
2540 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2543 d = moreswitches(d);
2545 if (PERLDB_LINE && !oldpdb ||
2546 ( PL_minus_n || PL_minus_p ) && !(oldn || oldp) )
2547 /* if we have already added "LINE: while (<>) {",
2548 we must not do it again */
2550 sv_setpv(PL_linestr, "");
2551 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2552 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2553 PL_preambled = FALSE;
2555 (void)gv_fetchfile(PL_origfilename);
2562 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2564 PL_lex_state = LEX_FORMLINE;
2569 #ifdef PERL_STRICT_CR
2570 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2572 "(Maybe you didn't strip carriage returns after a network transfer?)\n");
2574 case ' ': case '\t': case '\f': case 013:
2579 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2581 while (s < d && *s != '\n')
2586 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2588 PL_lex_state = LEX_FORMLINE;
2598 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2603 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
2606 if (strnEQ(s,"=>",2)) {
2607 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2608 OPERATOR('-'); /* unary minus */
2610 PL_last_uni = PL_oldbufptr;
2611 PL_last_lop_op = OP_FTEREAD; /* good enough */
2613 case 'r': FTST(OP_FTEREAD);
2614 case 'w': FTST(OP_FTEWRITE);
2615 case 'x': FTST(OP_FTEEXEC);
2616 case 'o': FTST(OP_FTEOWNED);
2617 case 'R': FTST(OP_FTRREAD);
2618 case 'W': FTST(OP_FTRWRITE);
2619 case 'X': FTST(OP_FTREXEC);
2620 case 'O': FTST(OP_FTROWNED);
2621 case 'e': FTST(OP_FTIS);
2622 case 'z': FTST(OP_FTZERO);
2623 case 's': FTST(OP_FTSIZE);
2624 case 'f': FTST(OP_FTFILE);
2625 case 'd': FTST(OP_FTDIR);
2626 case 'l': FTST(OP_FTLINK);
2627 case 'p': FTST(OP_FTPIPE);
2628 case 'S': FTST(OP_FTSOCK);
2629 case 'u': FTST(OP_FTSUID);
2630 case 'g': FTST(OP_FTSGID);
2631 case 'k': FTST(OP_FTSVTX);
2632 case 'b': FTST(OP_FTBLK);
2633 case 'c': FTST(OP_FTCHR);
2634 case 't': FTST(OP_FTTTY);
2635 case 'T': FTST(OP_FTTEXT);
2636 case 'B': FTST(OP_FTBINARY);
2637 case 'M': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTMTIME);
2638 case 'A': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTATIME);
2639 case 'C': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTCTIME);
2641 Perl_croak(aTHX_ "Unrecognized file test: -%c", (int)tmp);
2648 if (PL_expect == XOPERATOR)
2653 else if (*s == '>') {
2656 if (isIDFIRST_lazy(s)) {
2657 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2665 if (PL_expect == XOPERATOR)
2668 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2670 OPERATOR('-'); /* unary minus */
2677 if (PL_expect == XOPERATOR)
2682 if (PL_expect == XOPERATOR)
2685 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2691 if (PL_expect != XOPERATOR) {
2692 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2693 PL_expect = XOPERATOR;
2694 force_ident(PL_tokenbuf, '*');
2707 if (PL_expect == XOPERATOR) {
2711 PL_tokenbuf[0] = '%';
2712 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2713 if (!PL_tokenbuf[1]) {
2715 yyerror("Final % should be \\% or %name");
2718 PL_pending_ident = '%';
2737 switch (PL_expect) {
2740 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2742 PL_bufptr = s; /* update in case we back off */
2748 PL_expect = XTERMBLOCK;
2752 while (isIDFIRST_lazy(s)) {
2753 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2754 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2755 if (tmp < 0) tmp = -tmp;
2770 d = scan_str(d,TRUE,TRUE);
2773 SvREFCNT_dec(PL_lex_stuff);
2774 PL_lex_stuff = Nullsv;
2776 /* MUST advance bufptr here to avoid bogus
2777 "at end of line" context messages from yyerror().
2779 PL_bufptr = s + len;
2780 yyerror("Unterminated attribute parameter in attribute list");
2783 return 0; /* EOF indicator */
2787 SV *sv = newSVpvn(s, len);
2788 sv_catsv(sv, PL_lex_stuff);
2789 attrs = append_elem(OP_LIST, attrs,
2790 newSVOP(OP_CONST, 0, sv));
2791 SvREFCNT_dec(PL_lex_stuff);
2792 PL_lex_stuff = Nullsv;
2795 attrs = append_elem(OP_LIST, attrs,
2796 newSVOP(OP_CONST, 0,
2803 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
2804 if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) {
2805 char q = ((*s == '\'') ? '"' : '\'');
2806 /* If here for an expression, and parsed no attrs, back off. */
2807 if (tmp == '=' && !attrs) {
2811 /* MUST advance bufptr here to avoid bogus "at end of line"
2812 context messages from yyerror().
2816 yyerror("Unterminated attribute list");
2818 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
2826 PL_nextval[PL_nexttoke].opval = attrs;
2834 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
2835 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
2840 if (CopLINE(PL_curcop) < PL_copline)
2841 PL_copline = CopLINE(PL_curcop);
2852 if (PL_lex_brackets <= 0)
2853 yyerror("Unmatched right square bracket");
2856 if (PL_lex_state == LEX_INTERPNORMAL) {
2857 if (PL_lex_brackets == 0) {
2858 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
2859 PL_lex_state = LEX_INTERPEND;
2866 if (PL_lex_brackets > 100) {
2867 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
2868 if (newlb != PL_lex_brackstack) {
2870 PL_lex_brackstack = newlb;
2873 switch (PL_expect) {
2875 if (PL_lex_formbrack) {
2879 if (PL_oldoldbufptr == PL_last_lop)
2880 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
2882 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2883 OPERATOR(HASHBRACK);
2885 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
2888 PL_tokenbuf[0] = '\0';
2889 if (d < PL_bufend && *d == '-') {
2890 PL_tokenbuf[0] = '-';
2892 while (d < PL_bufend && (*d == ' ' || *d == '\t'))
2895 if (d < PL_bufend && isIDFIRST_lazy(d)) {
2896 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
2898 while (d < PL_bufend && (*d == ' ' || *d == '\t'))
2901 char minus = (PL_tokenbuf[0] == '-');
2902 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
2910 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
2915 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2920 if (PL_oldoldbufptr == PL_last_lop)
2921 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
2923 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2926 OPERATOR(HASHBRACK);
2927 /* This hack serves to disambiguate a pair of curlies
2928 * as being a block or an anon hash. Normally, expectation
2929 * determines that, but in cases where we're not in a
2930 * position to expect anything in particular (like inside
2931 * eval"") we have to resolve the ambiguity. This code
2932 * covers the case where the first term in the curlies is a
2933 * quoted string. Most other cases need to be explicitly
2934 * disambiguated by prepending a `+' before the opening
2935 * curly in order to force resolution as an anon hash.
2937 * XXX should probably propagate the outer expectation
2938 * into eval"" to rely less on this hack, but that could
2939 * potentially break current behavior of eval"".
2943 if (*s == '\'' || *s == '"' || *s == '`') {
2944 /* common case: get past first string, handling escapes */
2945 for (t++; t < PL_bufend && *t != *s;)
2946 if (*t++ == '\\' && (*t == '\\' || *t == *s))
2950 else if (*s == 'q') {
2953 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
2957 char open, close, term;
2960 while (t < PL_bufend && isSPACE(*t))
2964 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
2968 for (t++; t < PL_bufend; t++) {
2969 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
2971 else if (*t == open)
2975 for (t++; t < PL_bufend; t++) {
2976 if (*t == '\\' && t+1 < PL_bufend)
2978 else if (*t == close && --brackets <= 0)
2980 else if (*t == open)
2986 else if (isALNUM_lazy(t)) {
2988 while (t < PL_bufend && isALNUM_lazy(t))
2991 while (t < PL_bufend && isSPACE(*t))
2993 /* if comma follows first term, call it an anon hash */
2994 /* XXX it could be a comma expression with loop modifiers */
2995 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
2996 || (*t == '=' && t[1] == '>')))
2997 OPERATOR(HASHBRACK);
2998 if (PL_expect == XREF)
3001 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3007 yylval.ival = CopLINE(PL_curcop);
3008 if (isSPACE(*s) || *s == '#')
3009 PL_copline = NOLINE; /* invalidate current command line number */
3014 if (PL_lex_brackets <= 0)
3015 yyerror("Unmatched right curly bracket");
3017 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3018 if (PL_lex_brackets < PL_lex_formbrack)
3019 PL_lex_formbrack = 0;
3020 if (PL_lex_state == LEX_INTERPNORMAL) {
3021 if (PL_lex_brackets == 0) {
3022 if (PL_expect & XFAKEBRACK) {
3023 PL_expect &= XENUMMASK;
3024 PL_lex_state = LEX_INTERPEND;
3026 return yylex(); /* ignore fake brackets */
3028 if (*s == '-' && s[1] == '>')
3029 PL_lex_state = LEX_INTERPENDMAYBE;
3030 else if (*s != '[' && *s != '{')
3031 PL_lex_state = LEX_INTERPEND;
3034 if (PL_expect & XFAKEBRACK) {
3035 PL_expect &= XENUMMASK;
3037 return yylex(); /* ignore fake brackets */
3047 if (PL_expect == XOPERATOR) {
3048 if (ckWARN(WARN_SEMICOLON) && isIDFIRST_lazy(s) && PL_bufptr == PL_linestart) {
3049 CopLINE_dec(PL_curcop);
3050 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3051 CopLINE_inc(PL_curcop);
3056 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3058 PL_expect = XOPERATOR;
3059 force_ident(PL_tokenbuf, '&');
3063 yylval.ival = (OPpENTERSUB_AMPER<<8);
3082 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3083 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3085 if (PL_expect == XSTATE && isALPHA(tmp) &&
3086 (s == PL_linestart+1 || s[-2] == '\n') )
3088 if (PL_in_eval && !PL_rsfp) {
3093 if (strnEQ(s,"=cut",4)) {
3107 PL_doextract = TRUE;
3110 if (PL_lex_brackets < PL_lex_formbrack) {
3112 #ifdef PERL_STRICT_CR
3113 for (t = s; *t == ' ' || *t == '\t'; t++) ;
3115 for (t = s; *t == ' ' || *t == '\t' || *t == '\r'; t++) ;
3117 if (*t == '\n' || *t == '#') {
3135 if (PL_expect != XOPERATOR) {
3136 if (s[1] != '<' && !strchr(s,'>'))
3139 s = scan_heredoc(s);
3141 s = scan_inputsymbol(s);
3142 TERM(sublex_start());
3147 SHop(OP_LEFT_SHIFT);
3161 SHop(OP_RIGHT_SHIFT);
3170 if (PL_expect == XOPERATOR) {
3171 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3174 return ','; /* grandfather non-comma-format format */
3178 if (s[1] == '#' && (isIDFIRST_lazy(s+2) || strchr("{$:+-", s[2]))) {
3179 PL_tokenbuf[0] = '@';
3180 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3181 sizeof PL_tokenbuf - 1, FALSE);
3182 if (PL_expect == XOPERATOR)
3183 no_op("Array length", s);
3184 if (!PL_tokenbuf[1])
3186 PL_expect = XOPERATOR;
3187 PL_pending_ident = '#';
3191 PL_tokenbuf[0] = '$';
3192 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3193 sizeof PL_tokenbuf - 1, FALSE);
3194 if (PL_expect == XOPERATOR)
3196 if (!PL_tokenbuf[1]) {
3198 yyerror("Final $ should be \\$ or $name");
3202 /* This kludge not intended to be bulletproof. */
3203 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3204 yylval.opval = newSVOP(OP_CONST, 0,
3205 newSViv((IV)PL_compiling.cop_arybase));
3206 yylval.opval->op_private = OPpCONST_ARYBASE;
3212 if (PL_lex_state == LEX_NORMAL)
3215 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3218 PL_tokenbuf[0] = '@';
3219 if (ckWARN(WARN_SYNTAX)) {
3221 isSPACE(*t) || isALNUM_lazy(t) || *t == '$';
3224 PL_bufptr = skipspace(PL_bufptr);
3225 while (t < PL_bufend && *t != ']')
3227 Perl_warner(aTHX_ WARN_SYNTAX,
3228 "Multidimensional syntax %.*s not supported",
3229 (t - PL_bufptr) + 1, PL_bufptr);
3233 else if (*s == '{') {
3234 PL_tokenbuf[0] = '%';
3235 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3236 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3238 char tmpbuf[sizeof PL_tokenbuf];
3240 for (t++; isSPACE(*t); t++) ;
3241 if (isIDFIRST_lazy(t)) {
3242 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3243 for (; isSPACE(*t); t++) ;
3244 if (*t == ';' && get_cv(tmpbuf, FALSE))
3245 Perl_warner(aTHX_ WARN_SYNTAX,
3246 "You need to quote \"%s\"", tmpbuf);
3252 PL_expect = XOPERATOR;
3253 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3254 bool islop = (PL_last_lop == PL_oldoldbufptr);
3255 if (!islop || PL_last_lop_op == OP_GREPSTART)
3256 PL_expect = XOPERATOR;
3257 else if (strchr("$@\"'`q", *s))
3258 PL_expect = XTERM; /* e.g. print $fh "foo" */
3259 else if (strchr("&*<%", *s) && isIDFIRST_lazy(s+1))
3260 PL_expect = XTERM; /* e.g. print $fh &sub */
3261 else if (isIDFIRST_lazy(s)) {
3262 char tmpbuf[sizeof PL_tokenbuf];
3263 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3264 if (tmp = keyword(tmpbuf, len)) {
3265 /* binary operators exclude handle interpretations */
3277 PL_expect = XTERM; /* e.g. print $fh length() */
3282 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3283 if (gv && GvCVu(gv))
3284 PL_expect = XTERM; /* e.g. print $fh subr() */
3287 else if (isDIGIT(*s))
3288 PL_expect = XTERM; /* e.g. print $fh 3 */
3289 else if (*s == '.' && isDIGIT(s[1]))
3290 PL_expect = XTERM; /* e.g. print $fh .3 */
3291 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3292 PL_expect = XTERM; /* e.g. print $fh -1 */
3293 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3294 PL_expect = XTERM; /* print $fh <<"EOF" */
3296 PL_pending_ident = '$';
3300 if (PL_expect == XOPERATOR)
3302 PL_tokenbuf[0] = '@';
3303 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3304 if (!PL_tokenbuf[1]) {
3306 yyerror("Final @ should be \\@ or @name");
3309 if (PL_lex_state == LEX_NORMAL)
3311 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3313 PL_tokenbuf[0] = '%';
3315 /* Warn about @ where they meant $. */
3316 if (ckWARN(WARN_SYNTAX)) {
3317 if (*s == '[' || *s == '{') {
3319 while (*t && (isALNUM_lazy(t) || strchr(" \t$#+-'\"", *t)))
3321 if (*t == '}' || *t == ']') {
3323 PL_bufptr = skipspace(PL_bufptr);
3324 Perl_warner(aTHX_ WARN_SYNTAX,
3325 "Scalar value %.*s better written as $%.*s",
3326 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3331 PL_pending_ident = '@';
3334 case '/': /* may either be division or pattern */
3335 case '?': /* may either be conditional or pattern */
3336 if (PL_expect != XOPERATOR) {
3337 /* Disable warning on "study /blah/" */
3338 if (PL_oldoldbufptr == PL_last_uni
3339 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3340 || memNE(PL_last_uni, "study", 5) || isALNUM_lazy(PL_last_uni+5)))
3342 s = scan_pat(s,OP_MATCH);
3343 TERM(sublex_start());
3351 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3352 #ifdef PERL_STRICT_CR
3355 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3357 && (s == PL_linestart || s[-1] == '\n') )
3359 PL_lex_formbrack = 0;
3363 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3369 yylval.ival = OPf_SPECIAL;
3375 if (PL_expect != XOPERATOR)
3380 case '0': case '1': case '2': case '3': case '4':
3381 case '5': case '6': case '7': case '8': case '9':
3383 if (PL_expect == XOPERATOR)
3388 s = scan_str(s,FALSE,FALSE);
3389 if (PL_expect == XOPERATOR) {
3390 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3393 return ','; /* grandfather non-comma-format format */
3399 missingterm((char*)0);
3400 yylval.ival = OP_CONST;
3401 TERM(sublex_start());
3404 s = scan_str(s,FALSE,FALSE);
3405 if (PL_expect == XOPERATOR) {
3406 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3409 return ','; /* grandfather non-comma-format format */
3415 missingterm((char*)0);
3416 yylval.ival = OP_CONST;
3417 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3418 if (*d == '$' || *d == '@' || *d == '\\' || *d & 0x80) {
3419 yylval.ival = OP_STRINGIFY;
3423 TERM(sublex_start());
3426 s = scan_str(s,FALSE,FALSE);
3427 if (PL_expect == XOPERATOR)
3428 no_op("Backticks",s);
3430 missingterm((char*)0);
3431 yylval.ival = OP_BACKTICK;
3433 TERM(sublex_start());
3437 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3438 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3440 if (PL_expect == XOPERATOR)
3441 no_op("Backslash",s);
3445 if (isDIGIT(s[1]) && PL_expect == XTERM) {
3449 while (isDIGIT(*start))
3451 if (*start == '.' && isDIGIT(start[1])) {
3458 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3498 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3500 /* Some keywords can be followed by any delimiter, including ':' */
3501 tmp = (len == 1 && strchr("msyq", PL_tokenbuf[0]) ||
3502 len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3503 (PL_tokenbuf[0] == 'q' &&
3504 strchr("qwxr", PL_tokenbuf[1]))));
3506 /* x::* is just a word, unless x is "CORE" */
3507 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3511 while (d < PL_bufend && isSPACE(*d))
3512 d++; /* no comments skipped here, or s### is misparsed */
3514 /* Is this a label? */
3515 if (!tmp && PL_expect == XSTATE
3516 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3518 yylval.pval = savepv(PL_tokenbuf);
3523 /* Check for keywords */
3524 tmp = keyword(PL_tokenbuf, len);
3526 /* Is this a word before a => operator? */
3527 if (strnEQ(d,"=>",2)) {
3529 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3530 yylval.opval->op_private = OPpCONST_BARE;
3534 if (tmp < 0) { /* second-class keyword? */
3535 GV *ogv = Nullgv; /* override (winner) */
3536 GV *hgv = Nullgv; /* hidden (loser) */
3537 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3539 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3542 if (GvIMPORTED_CV(gv))
3544 else if (! CvMETHOD(cv))
3548 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3549 (gv = *gvp) != (GV*)&PL_sv_undef &&
3550 GvCVu(gv) && GvIMPORTED_CV(gv))
3556 tmp = 0; /* overridden by import or by GLOBAL */
3559 && -tmp==KEY_lock /* XXX generalizable kludge */
3561 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3563 tmp = 0; /* any sub overrides "weak" keyword */
3565 else { /* no override */
3569 if (ckWARN(WARN_AMBIGUOUS) && hgv
3570 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3571 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3572 "Ambiguous call resolved as CORE::%s(), %s",
3573 GvENAME(hgv), "qualify as such or use &");
3580 default: /* not a keyword */
3583 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3585 /* Get the rest if it looks like a package qualifier */
3587 if (*s == '\'' || *s == ':' && s[1] == ':') {
3589 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3592 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3593 *s == '\'' ? "'" : "::");
3597 if (PL_expect == XOPERATOR) {
3598 if (PL_bufptr == PL_linestart) {
3599 CopLINE_dec(PL_curcop);
3600 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3601 CopLINE_inc(PL_curcop);
3604 no_op("Bareword",s);
3607 /* Look for a subroutine with this name in current package,
3608 unless name is "Foo::", in which case Foo is a bearword
3609 (and a package name). */
3612 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3614 if (ckWARN(WARN_UNSAFE) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3615 Perl_warner(aTHX_ WARN_UNSAFE,
3616 "Bareword \"%s\" refers to nonexistent package",
3619 PL_tokenbuf[len] = '\0';
3626 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3629 /* if we saw a global override before, get the right name */
3632 sv = newSVpvn("CORE::GLOBAL::",14);
3633 sv_catpv(sv,PL_tokenbuf);
3636 sv = newSVpv(PL_tokenbuf,0);
3638 /* Presume this is going to be a bareword of some sort. */
3641 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3642 yylval.opval->op_private = OPpCONST_BARE;
3644 /* And if "Foo::", then that's what it certainly is. */
3649 /* See if it's the indirect object for a list operator. */
3651 if (PL_oldoldbufptr &&
3652 PL_oldoldbufptr < PL_bufptr &&
3653 (PL_oldoldbufptr == PL_last_lop
3654 || PL_oldoldbufptr == PL_last_uni) &&
3655 /* NO SKIPSPACE BEFORE HERE! */
3656 (PL_expect == XREF ||
3657 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3659 bool immediate_paren = *s == '(';
3661 /* (Now we can afford to cross potential line boundary.) */
3664 /* Two barewords in a row may indicate method call. */
3666 if ((isIDFIRST_lazy(s) || *s == '$') && (tmp=intuit_method(s,gv)))
3669 /* If not a declared subroutine, it's an indirect object. */
3670 /* (But it's an indir obj regardless for sort.) */
3672 if ((PL_last_lop_op == OP_SORT ||
3673 (!immediate_paren && (!gv || !GvCVu(gv)))) &&
3674 (PL_last_lop_op != OP_MAPSTART &&
3675 PL_last_lop_op != OP_GREPSTART))
3677 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3682 /* If followed by a paren, it's certainly a subroutine. */
3684 PL_expect = XOPERATOR;
3688 if (gv && GvCVu(gv)) {
3689 for (d = s + 1; *d == ' ' || *d == '\t'; d++) ;
3690 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3695 PL_nextval[PL_nexttoke].opval = yylval.opval;
3696 PL_expect = XOPERATOR;
3702 /* If followed by var or block, call it a method (unless sub) */
3704 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3705 PL_last_lop = PL_oldbufptr;
3706 PL_last_lop_op = OP_METHOD;
3710 /* If followed by a bareword, see if it looks like indir obj. */
3712 if ((isIDFIRST_lazy(s) || *s == '$') && (tmp = intuit_method(s,gv)))
3715 /* Not a method, so call it a subroutine (if defined) */
3717 if (gv && GvCVu(gv)) {
3719 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
3720 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3721 "Ambiguous use of -%s resolved as -&%s()",
3722 PL_tokenbuf, PL_tokenbuf);
3723 /* Check for a constant sub */
3725 if ((sv = cv_const_sv(cv))) {
3727 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
3728 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
3729 yylval.opval->op_private = 0;
3733 /* Resolve to GV now. */
3734 op_free(yylval.opval);
3735 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
3736 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
3737 PL_last_lop = PL_oldbufptr;
3738 PL_last_lop_op = OP_ENTERSUB;
3739 /* Is there a prototype? */
3742 char *proto = SvPV((SV*)cv, len);
3745 if (strEQ(proto, "$"))
3747 if (*proto == '&' && *s == '{') {
3748 sv_setpv(PL_subname,"__ANON__");
3752 PL_nextval[PL_nexttoke].opval = yylval.opval;
3758 /* Call it a bare word */
3760 if (PL_hints & HINT_STRICT_SUBS)
3761 yylval.opval->op_private |= OPpCONST_STRICT;
3764 if (ckWARN(WARN_RESERVED)) {
3765 if (lastchar != '-') {
3766 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
3768 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
3775 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
3776 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3777 "Operator or semicolon missing before %c%s",
3778 lastchar, PL_tokenbuf);
3779 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3780 "Ambiguous use of %c resolved as operator %c",
3781 lastchar, lastchar);
3787 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3788 newSVpv(CopFILE(PL_curcop),0));
3792 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3793 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
3796 case KEY___PACKAGE__:
3797 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3799 ? newSVsv(PL_curstname)
3808 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
3809 char *pname = "main";
3810 if (PL_tokenbuf[2] == 'D')
3811 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
3812 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
3815 GvIOp(gv) = newIO();
3816 IoIFP(GvIOp(gv)) = PL_rsfp;
3817 #if defined(HAS_FCNTL) && defined(F_SETFD)
3819 int fd = PerlIO_fileno(PL_rsfp);
3820 fcntl(fd,F_SETFD,fd >= 3);
3823 /* Mark this internal pseudo-handle as clean */
3824 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
3826 IoTYPE(GvIOp(gv)) = '|';
3827 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
3828 IoTYPE(GvIOp(gv)) = '-';
3830 IoTYPE(GvIOp(gv)) = '<';
3831 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
3832 /* if the script was opened in binmode, we need to revert
3833 * it to text mode for compatibility; but only iff it has CRs
3834 * XXX this is a questionable hack at best. */
3835 if (PL_bufend-PL_bufptr > 2
3836 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
3839 if (IoTYPE(GvIOp(gv)) == '<') {
3840 loc = PerlIO_tell(PL_rsfp);
3841 (void)PerlIO_seek(PL_rsfp, 0L, 0);
3843 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
3844 #if defined(__BORLANDC__)
3845 /* XXX see note in do_binmode() */
3846 ((FILE*)PL_rsfp)->flags |= _F_BIN;
3849 PerlIO_seek(PL_rsfp, loc, 0);
3864 if (PL_expect == XSTATE) {
3871 if (*s == ':' && s[1] == ':') {
3874 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3875 tmp = keyword(PL_tokenbuf, len);
3889 LOP(OP_ACCEPT,XTERM);
3895 LOP(OP_ATAN2,XTERM);
3904 LOP(OP_BLESS,XTERM);
3913 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
3930 if (!PL_cryptseen) {
3931 PL_cryptseen = TRUE;
3935 LOP(OP_CRYPT,XTERM);
3938 if (ckWARN(WARN_OCTAL)) {
3939 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
3940 if (*d != '0' && isDIGIT(*d))
3941 Perl_warner(aTHX_ WARN_OCTAL,
3942 "chmod: mode argument is missing initial 0");
3944 LOP(OP_CHMOD,XTERM);
3947 LOP(OP_CHOWN,XTERM);
3950 LOP(OP_CONNECT,XTERM);
3966 s = force_word(s,WORD,FALSE,TRUE,FALSE);
3970 PL_hints |= HINT_BLOCK_SCOPE;
3980 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
3981 LOP(OP_DBMOPEN,XTERM);
3987 s = force_word(s,WORD,TRUE,FALSE,FALSE);
3994 yylval.ival = CopLINE(PL_curcop);
4008 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4009 UNIBRACK(OP_ENTEREVAL);
4024 case KEY_endhostent:
4030 case KEY_endservent:
4033 case KEY_endprotoent:
4044 yylval.ival = CopLINE(PL_curcop);
4046 if (PL_expect == XSTATE && isIDFIRST_lazy(s)) {
4048 if ((PL_bufend - p) >= 3 &&
4049 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4051 else if ((PL_bufend - p) >= 4 &&
4052 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4055 if (isIDFIRST_lazy(p)) {
4056 p = scan_ident(p, PL_bufend,
4057 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4061 Perl_croak(aTHX_ "Missing $ on loop variable");
4066 LOP(OP_FORMLINE,XTERM);
4072 LOP(OP_FCNTL,XTERM);
4078 LOP(OP_FLOCK,XTERM);
4087 LOP(OP_GREPSTART, *s == '(' ? XTERM : XREF);
4090 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4105 case KEY_getpriority:
4106 LOP(OP_GETPRIORITY,XTERM);
4108 case KEY_getprotobyname:
4111 case KEY_getprotobynumber:
4112 LOP(OP_GPBYNUMBER,XTERM);
4114 case KEY_getprotoent:
4126 case KEY_getpeername:
4127 UNI(OP_GETPEERNAME);
4129 case KEY_gethostbyname:
4132 case KEY_gethostbyaddr:
4133 LOP(OP_GHBYADDR,XTERM);
4135 case KEY_gethostent:
4138 case KEY_getnetbyname:
4141 case KEY_getnetbyaddr:
4142 LOP(OP_GNBYADDR,XTERM);
4147 case KEY_getservbyname:
4148 LOP(OP_GSBYNAME,XTERM);
4150 case KEY_getservbyport:
4151 LOP(OP_GSBYPORT,XTERM);
4153 case KEY_getservent:
4156 case KEY_getsockname:
4157 UNI(OP_GETSOCKNAME);
4159 case KEY_getsockopt:
4160 LOP(OP_GSOCKOPT,XTERM);
4182 yylval.ival = CopLINE(PL_curcop);
4186 LOP(OP_INDEX,XTERM);
4192 LOP(OP_IOCTL,XTERM);
4204 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4236 LOP(OP_LISTEN,XTERM);
4245 s = scan_pat(s,OP_MATCH);
4246 TERM(sublex_start());
4249 LOP(OP_MAPSTART, *s == '(' ? XTERM : XREF);
4252 LOP(OP_MKDIR,XTERM);
4255 LOP(OP_MSGCTL,XTERM);
4258 LOP(OP_MSGGET,XTERM);
4261 LOP(OP_MSGRCV,XTERM);
4264 LOP(OP_MSGSND,XTERM);
4270 if (isIDFIRST_lazy(s)) {
4271 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4272 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4274 PL_in_my_stash = gv_stashpv(PL_tokenbuf, FALSE);
4275 if (!PL_in_my_stash) {
4278 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4286 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4293 if (PL_expect != XSTATE)
4294 yyerror("\"no\" not allowed in expression");
4295 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4296 s = force_version(s);
4301 if (*s == '(' || (s = skipspace(s), *s == '('))
4308 if (isIDFIRST_lazy(s)) {
4310 for (d = s; isALNUM_lazy(d); d++) ;
4312 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_AMBIGUOUS))
4313 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4314 "Precedence problem: open %.*s should be open(%.*s)",
4320 yylval.ival = OP_OR;
4330 LOP(OP_OPEN_DIR,XTERM);
4333 checkcomma(s,PL_tokenbuf,"filehandle");
4337 checkcomma(s,PL_tokenbuf,"filehandle");
4356 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4360 LOP(OP_PIPE_OP,XTERM);
4363 s = scan_str(s,FALSE,FALSE);
4365 missingterm((char*)0);
4366 yylval.ival = OP_CONST;
4367 TERM(sublex_start());
4373 s = scan_str(s,FALSE,FALSE);
4375 missingterm((char*)0);
4377 if (SvCUR(PL_lex_stuff)) {
4380 d = SvPV_force(PL_lex_stuff, len);
4382 for (; isSPACE(*d) && len; --len, ++d) ;
4385 if (!warned && ckWARN(WARN_SYNTAX)) {
4386 for (; !isSPACE(*d) && len; --len, ++d) {
4388 Perl_warner(aTHX_ WARN_SYNTAX,
4389 "Possible attempt to separate words with commas");
4392 else if (*d == '#') {
4393 Perl_warner(aTHX_ WARN_SYNTAX,
4394 "Possible attempt to put comments in qw() list");
4400 for (; !isSPACE(*d) && len; --len, ++d) ;
4402 words = append_elem(OP_LIST, words,
4403 newSVOP(OP_CONST, 0, newSVpvn(b, d-b)));
4407 PL_nextval[PL_nexttoke].opval = words;
4412 SvREFCNT_dec(PL_lex_stuff);
4413 PL_lex_stuff = Nullsv;
4418 s = scan_str(s,FALSE,FALSE);
4420 missingterm((char*)0);
4421 yylval.ival = OP_STRINGIFY;
4422 if (SvIVX(PL_lex_stuff) == '\'')
4423 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4424 TERM(sublex_start());
4427 s = scan_pat(s,OP_QR);
4428 TERM(sublex_start());
4431 s = scan_str(s,FALSE,FALSE);
4433 missingterm((char*)0);
4434 yylval.ival = OP_BACKTICK;
4436 TERM(sublex_start());
4443 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4444 s = force_version(s);
4447 *PL_tokenbuf = '\0';
4448 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4449 if (isIDFIRST_lazy(PL_tokenbuf))
4450 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4452 yyerror("<> should be quotes");
4460 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4464 LOP(OP_RENAME,XTERM);
4473 LOP(OP_RINDEX,XTERM);
4496 LOP(OP_REVERSE,XTERM);
4507 TERM(sublex_start());
4509 TOKEN(1); /* force error */
4518 LOP(OP_SELECT,XTERM);
4524 LOP(OP_SEMCTL,XTERM);
4527 LOP(OP_SEMGET,XTERM);
4530 LOP(OP_SEMOP,XTERM);
4536 LOP(OP_SETPGRP,XTERM);
4538 case KEY_setpriority:
4539 LOP(OP_SETPRIORITY,XTERM);
4541 case KEY_sethostent:
4547 case KEY_setservent:
4550 case KEY_setprotoent:
4560 LOP(OP_SEEKDIR,XTERM);
4562 case KEY_setsockopt:
4563 LOP(OP_SSOCKOPT,XTERM);
4569 LOP(OP_SHMCTL,XTERM);
4572 LOP(OP_SHMGET,XTERM);
4575 LOP(OP_SHMREAD,XTERM);
4578 LOP(OP_SHMWRITE,XTERM);
4581 LOP(OP_SHUTDOWN,XTERM);
4590 LOP(OP_SOCKET,XTERM);
4592 case KEY_socketpair:
4593 LOP(OP_SOCKPAIR,XTERM);
4596 checkcomma(s,PL_tokenbuf,"subroutine name");
4598 if (*s == ';' || *s == ')') /* probably a close */
4599 Perl_croak(aTHX_ "sort is now a reserved word");
4601 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4605 LOP(OP_SPLIT,XTERM);
4608 LOP(OP_SPRINTF,XTERM);
4611 LOP(OP_SPLICE,XTERM);
4626 LOP(OP_SUBSTR,XTERM);
4632 char tmpbuf[sizeof PL_tokenbuf];
4634 expectation attrful;
4635 bool have_name, have_proto;
4640 if (isIDFIRST_lazy(s) || *s == '\'' ||
4641 (*s == ':' && s[1] == ':'))
4644 attrful = XATTRBLOCK;
4645 /* remember buffer pos'n for later force_word */
4646 tboffset = s - PL_oldbufptr;
4647 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4648 if (strchr(tmpbuf, ':'))
4649 sv_setpv(PL_subname, tmpbuf);
4651 sv_setsv(PL_subname,PL_curstname);
4652 sv_catpvn(PL_subname,"::",2);
4653 sv_catpvn(PL_subname,tmpbuf,len);
4660 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4661 PL_expect = XTERMBLOCK;
4662 attrful = XATTRTERM;
4663 sv_setpv(PL_subname,"?");
4667 if (key == KEY_format) {
4669 PL_lex_formbrack = PL_lex_brackets + 1;
4671 (void) force_word(PL_oldbufptr + tboffset, WORD,
4676 /* Look for a prototype */
4680 s = scan_str(s,FALSE,FALSE);
4683 SvREFCNT_dec(PL_lex_stuff);
4684 PL_lex_stuff = Nullsv;
4685 Perl_croak(aTHX_ "Prototype not terminated");
4688 d = SvPVX(PL_lex_stuff);
4690 for (p = d; *p; ++p) {
4695 SvCUR(PL_lex_stuff) = tmp;
4703 if (*s == ':' && s[1] != ':')
4704 PL_expect = attrful;
4707 PL_nextval[PL_nexttoke].opval =
4708 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
4709 PL_lex_stuff = Nullsv;
4713 sv_setpv(PL_subname,"__ANON__");
4716 (void) force_word(PL_oldbufptr + tboffset, WORD,
4725 LOP(OP_SYSTEM,XREF);
4728 LOP(OP_SYMLINK,XTERM);
4731 LOP(OP_SYSCALL,XTERM);
4734 LOP(OP_SYSOPEN,XTERM);
4737 LOP(OP_SYSSEEK,XTERM);
4740 LOP(OP_SYSREAD,XTERM);
4743 LOP(OP_SYSWRITE,XTERM);
4747 TERM(sublex_start());
4768 LOP(OP_TRUNCATE,XTERM);
4780 yylval.ival = CopLINE(PL_curcop);
4784 yylval.ival = CopLINE(PL_curcop);
4788 LOP(OP_UNLINK,XTERM);
4794 LOP(OP_UNPACK,XTERM);
4797 LOP(OP_UTIME,XTERM);
4800 if (ckWARN(WARN_OCTAL)) {
4801 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4802 if (*d != '0' && isDIGIT(*d))
4803 Perl_warner(aTHX_ WARN_OCTAL,
4804 "umask: argument is missing initial 0");
4809 LOP(OP_UNSHIFT,XTERM);
4812 if (PL_expect != XSTATE)
4813 yyerror("\"use\" not allowed in expression");
4815 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4816 s = force_version(s);
4817 if (*s == ';' || (s = skipspace(s), *s == ';')) {
4818 PL_nextval[PL_nexttoke].opval = Nullop;
4823 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4824 s = force_version(s);
4836 yylval.ival = CopLINE(PL_curcop);
4840 PL_hints |= HINT_BLOCK_SCOPE;
4847 LOP(OP_WAITPID,XTERM);
4855 static char ctl_l[2];
4857 if (ctl_l[0] == '\0')
4858 ctl_l[0] = toCTRL('L');
4859 gv_fetchpv(ctl_l,TRUE, SVt_PV);
4862 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
4867 if (PL_expect == XOPERATOR)
4873 yylval.ival = OP_XOR;
4878 TERM(sublex_start());
4884 Perl_keyword(pTHX_ register char *d, I32 len)
4889 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
4890 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
4891 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
4892 if (strEQ(d,"__DATA__")) return KEY___DATA__;
4893 if (strEQ(d,"__END__")) return KEY___END__;
4897 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
4902 if (strEQ(d,"and")) return -KEY_and;
4903 if (strEQ(d,"abs")) return -KEY_abs;
4906 if (strEQ(d,"alarm")) return -KEY_alarm;
4907 if (strEQ(d,"atan2")) return -KEY_atan2;
4910 if (strEQ(d,"accept")) return -KEY_accept;
4915 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
4918 if (strEQ(d,"bless")) return -KEY_bless;
4919 if (strEQ(d,"bind")) return -KEY_bind;
4920 if (strEQ(d,"binmode")) return -KEY_binmode;
4923 if (strEQ(d,"CORE")) return -KEY_CORE;
4928 if (strEQ(d,"cmp")) return -KEY_cmp;
4929 if (strEQ(d,"chr")) return -KEY_chr;
4930 if (strEQ(d,"cos")) return -KEY_cos;
4933 if (strEQ(d,"chop")) return KEY_chop;
4936 if (strEQ(d,"close")) return -KEY_close;
4937 if (strEQ(d,"chdir")) return -KEY_chdir;
4938 if (strEQ(d,"chomp")) return KEY_chomp;
4939 if (strEQ(d,"chmod")) return -KEY_chmod;
4940 if (strEQ(d,"chown")) return -KEY_chown;
4941 if (strEQ(d,"crypt")) return -KEY_crypt;
4944 if (strEQ(d,"chroot")) return -KEY_chroot;
4945 if (strEQ(d,"caller")) return -KEY_caller;
4948 if (strEQ(d,"connect")) return -KEY_connect;
4951 if (strEQ(d,"closedir")) return -KEY_closedir;
4952 if (strEQ(d,"continue")) return -KEY_continue;
4957 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
4962 if (strEQ(d,"do")) return KEY_do;
4965 if (strEQ(d,"die")) return -KEY_die;
4968 if (strEQ(d,"dump")) return -KEY_dump;
4971 if (strEQ(d,"delete")) return KEY_delete;
4974 if (strEQ(d,"defined")) return KEY_defined;
4975 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
4978 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
4983 if (strEQ(d,"EQ")) { deprecate(d); return -KEY_eq;}
4984 if (strEQ(d,"END")) return KEY_END;
4989 if (strEQ(d,"eq")) return -KEY_eq;
4992 if (strEQ(d,"eof")) return -KEY_eof;
4993 if (strEQ(d,"exp")) return -KEY_exp;
4996 if (strEQ(d,"else")) return KEY_else;
4997 if (strEQ(d,"exit")) return -KEY_exit;
4998 if (strEQ(d,"eval")) return KEY_eval;
4999 if (strEQ(d,"exec")) return -KEY_exec;
5000 if (strEQ(d,"each")) return KEY_each;
5003 if (strEQ(d,"elsif")) return KEY_elsif;
5006 if (strEQ(d,"exists")) return KEY_exists;
5007 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5010 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5011 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5014 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5017 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5018 if (strEQ(d,"endservent")) return -KEY_endservent;
5021 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5028 if (strEQ(d,"for")) return KEY_for;
5031 if (strEQ(d,"fork")) return -KEY_fork;
5034 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5035 if (strEQ(d,"flock")) return -KEY_flock;
5038 if (strEQ(d,"format")) return KEY_format;
5039 if (strEQ(d,"fileno")) return -KEY_fileno;
5042 if (strEQ(d,"foreach")) return KEY_foreach;
5045 if (strEQ(d,"formline")) return -KEY_formline;
5051 if (strEQ(d,"GT")) { deprecate(d); return -KEY_gt;}
5052 if (strEQ(d,"GE")) { deprecate(d); return -KEY_ge;}
5056 if (strnEQ(d,"get",3)) {
5061 if (strEQ(d,"ppid")) return -KEY_getppid;
5062 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5065 if (strEQ(d,"pwent")) return -KEY_getpwent;
5066 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5067 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5070 if (strEQ(d,"peername")) return -KEY_getpeername;
5071 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5072 if (strEQ(d,"priority")) return -KEY_getpriority;
5075 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5078 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5082 else if (*d == 'h') {
5083 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5084 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5085 if (strEQ(d,"hostent")) return -KEY_gethostent;
5087 else if (*d == 'n') {
5088 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5089 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5090 if (strEQ(d,"netent")) return -KEY_getnetent;
5092 else if (*d == 's') {
5093 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5094 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5095 if (strEQ(d,"servent")) return -KEY_getservent;
5096 if (strEQ(d,"sockname")) return -KEY_getsockname;
5097 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5099 else if (*d == 'g') {
5100 if (strEQ(d,"grent")) return -KEY_getgrent;
5101 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5102 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5104 else if (*d == 'l') {
5105 if (strEQ(d,"login")) return -KEY_getlogin;
5107 else if (strEQ(d,"c")) return -KEY_getc;
5112 if (strEQ(d,"gt")) return -KEY_gt;
5113 if (strEQ(d,"ge")) return -KEY_ge;
5116 if (strEQ(d,"grep")) return KEY_grep;
5117 if (strEQ(d,"goto")) return KEY_goto;
5118 if (strEQ(d,"glob")) return KEY_glob;
5121 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5126 if (strEQ(d,"hex")) return -KEY_hex;
5129 if (strEQ(d,"INIT")) return KEY_INIT;
5134 if (strEQ(d,"if")) return KEY_if;
5137 if (strEQ(d,"int")) return -KEY_int;
5140 if (strEQ(d,"index")) return -KEY_index;
5141 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5146 if (strEQ(d,"join")) return -KEY_join;
5150 if (strEQ(d,"keys")) return KEY_keys;
5151 if (strEQ(d,"kill")) return -KEY_kill;
5156 if (strEQ(d,"LT")) { deprecate(d); return -KEY_lt;}
5157 if (strEQ(d,"LE")) { deprecate(d); return -KEY_le;}
5163 if (strEQ(d,"lt")) return -KEY_lt;
5164 if (strEQ(d,"le")) return -KEY_le;
5165 if (strEQ(d,"lc")) return -KEY_lc;
5168 if (strEQ(d,"log")) return -KEY_log;
5171 if (strEQ(d,"last")) return KEY_last;
5172 if (strEQ(d,"link")) return -KEY_link;
5173 if (strEQ(d,"lock")) return -KEY_lock;
5176 if (strEQ(d,"local")) return KEY_local;
5177 if (strEQ(d,"lstat")) return -KEY_lstat;
5180 if (strEQ(d,"length")) return -KEY_length;
5181 if (strEQ(d,"listen")) return -KEY_listen;
5184 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5187 if (strEQ(d,"localtime")) return -KEY_localtime;
5193 case 1: return KEY_m;
5195 if (strEQ(d,"my")) return KEY_my;
5198 if (strEQ(d,"map")) return KEY_map;
5201 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5204 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5205 if (strEQ(d,"msgget")) return -KEY_msgget;
5206 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5207 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5212 if (strEQ(d,"NE")) { deprecate(d); return -KEY_ne;}
5215 if (strEQ(d,"next")) return KEY_next;
5216 if (strEQ(d,"ne")) return -KEY_ne;
5217 if (strEQ(d,"not")) return -KEY_not;
5218 if (strEQ(d,"no")) return KEY_no;
5223 if (strEQ(d,"or")) return -KEY_or;
5226 if (strEQ(d,"ord")) return -KEY_ord;
5227 if (strEQ(d,"oct")) return -KEY_oct;
5228 if (strEQ(d,"our")) return KEY_our;
5231 if (strEQ(d,"open")) return -KEY_open;
5234 if (strEQ(d,"opendir")) return -KEY_opendir;
5241 if (strEQ(d,"pop")) return KEY_pop;
5242 if (strEQ(d,"pos")) return KEY_pos;
5245 if (strEQ(d,"push")) return KEY_push;
5246 if (strEQ(d,"pack")) return -KEY_pack;
5247 if (strEQ(d,"pipe")) return -KEY_pipe;
5250 if (strEQ(d,"print")) return KEY_print;
5253 if (strEQ(d,"printf")) return KEY_printf;
5256 if (strEQ(d,"package")) return KEY_package;
5259 if (strEQ(d,"prototype")) return KEY_prototype;
5264 if (strEQ(d,"q")) return KEY_q;
5265 if (strEQ(d,"qr")) return KEY_qr;
5266 if (strEQ(d,"qq")) return KEY_qq;
5267 if (strEQ(d,"qw")) return KEY_qw;
5268 if (strEQ(d,"qx")) return KEY_qx;
5270 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5275 if (strEQ(d,"ref")) return -KEY_ref;
5278 if (strEQ(d,"read")) return -KEY_read;
5279 if (strEQ(d,"rand")) return -KEY_rand;
5280 if (strEQ(d,"recv")) return -KEY_recv;
5281 if (strEQ(d,"redo")) return KEY_redo;
5284 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5285 if (strEQ(d,"reset")) return -KEY_reset;
5288 if (strEQ(d,"return")) return KEY_return;
5289 if (strEQ(d,"rename")) return -KEY_rename;
5290 if (strEQ(d,"rindex")) return -KEY_rindex;
5293 if (strEQ(d,"require")) return -KEY_require;
5294 if (strEQ(d,"reverse")) return -KEY_reverse;
5295 if (strEQ(d,"readdir")) return -KEY_readdir;
5298 if (strEQ(d,"readlink")) return -KEY_readlink;
5299 if (strEQ(d,"readline")) return -KEY_readline;
5300 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5303 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5308 if (strEQ(d,"STOP")) return KEY_STOP;
5312 case 0: return KEY_s;
5314 if (strEQ(d,"scalar")) return KEY_scalar;
5319 if (strEQ(d,"seek")) return -KEY_seek;
5320 if (strEQ(d,"send")) return -KEY_send;
5323 if (strEQ(d,"semop")) return -KEY_semop;
5326 if (strEQ(d,"select")) return -KEY_select;
5327 if (strEQ(d,"semctl")) return -KEY_semctl;
5328 if (strEQ(d,"semget")) return -KEY_semget;
5331 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5332 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5335 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5336 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5339 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5342 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5343 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5344 if (strEQ(d,"setservent")) return -KEY_setservent;
5347 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5348 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5355 if (strEQ(d,"shift")) return KEY_shift;
5358 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5359 if (strEQ(d,"shmget")) return -KEY_shmget;
5362 if (strEQ(d,"shmread")) return -KEY_shmread;
5365 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5366 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5371 if (strEQ(d,"sin")) return -KEY_sin;
5374 if (strEQ(d,"sleep")) return -KEY_sleep;
5377 if (strEQ(d,"sort")) return KEY_sort;
5378 if (strEQ(d,"socket")) return -KEY_socket;
5379 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5382 if (strEQ(d,"split")) return KEY_split;
5383 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5384 if (strEQ(d,"splice")) return KEY_splice;
5387 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5390 if (strEQ(d,"srand")) return -KEY_srand;
5393 if (strEQ(d,"stat")) return -KEY_stat;
5394 if (strEQ(d,"study")) return KEY_study;
5397 if (strEQ(d,"substr")) return -KEY_substr;
5398 if (strEQ(d,"sub")) return KEY_sub;
5403 if (strEQ(d,"system")) return -KEY_system;
5406 if (strEQ(d,"symlink")) return -KEY_symlink;
5407 if (strEQ(d,"syscall")) return -KEY_syscall;
5408 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5409 if (strEQ(d,"sysread")) return -KEY_sysread;
5410 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5413 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5422 if (strEQ(d,"tr")) return KEY_tr;
5425 if (strEQ(d,"tie")) return KEY_tie;
5428 if (strEQ(d,"tell")) return -KEY_tell;
5429 if (strEQ(d,"tied")) return KEY_tied;
5430 if (strEQ(d,"time")) return -KEY_time;
5433 if (strEQ(d,"times")) return -KEY_times;
5436 if (strEQ(d,"telldir")) return -KEY_telldir;
5439 if (strEQ(d,"truncate")) return -KEY_truncate;
5446 if (strEQ(d,"uc")) return -KEY_uc;
5449 if (strEQ(d,"use")) return KEY_use;
5452 if (strEQ(d,"undef")) return KEY_undef;
5453 if (strEQ(d,"until")) return KEY_until;
5454 if (strEQ(d,"untie")) return KEY_untie;
5455 if (strEQ(d,"utime")) return -KEY_utime;
5456 if (strEQ(d,"umask")) return -KEY_umask;
5459 if (strEQ(d,"unless")) return KEY_unless;
5460 if (strEQ(d,"unpack")) return -KEY_unpack;
5461 if (strEQ(d,"unlink")) return -KEY_unlink;
5464 if (strEQ(d,"unshift")) return KEY_unshift;
5465 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5470 if (strEQ(d,"values")) return -KEY_values;
5471 if (strEQ(d,"vec")) return -KEY_vec;
5476 if (strEQ(d,"warn")) return -KEY_warn;
5477 if (strEQ(d,"wait")) return -KEY_wait;
5480 if (strEQ(d,"while")) return KEY_while;
5481 if (strEQ(d,"write")) return -KEY_write;
5484 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5487 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5492 if (len == 1) return -KEY_x;
5493 if (strEQ(d,"xor")) return -KEY_xor;
5496 if (len == 1) return KEY_y;
5505 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5509 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5510 dTHR; /* only for ckWARN */
5511 if (ckWARN(WARN_SYNTAX)) {
5513 for (w = s+2; *w && level; w++) {
5520 for (; *w && isSPACE(*w); w++) ;
5521 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5522 Perl_warner(aTHX_ WARN_SYNTAX,
5523 "%s (...) interpreted as function",name);
5526 while (s < PL_bufend && isSPACE(*s))
5530 while (s < PL_bufend && isSPACE(*s))
5532 if (isIDFIRST_lazy(s)) {
5534 while (isALNUM_lazy(s))
5536 while (s < PL_bufend && isSPACE(*s))
5541 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5545 Perl_croak(aTHX_ "No comma allowed after %s", what);
5550 /* Either returns sv, or mortalizes sv and returns a new SV*.
5551 Best used as sv=new_constant(..., sv, ...).
5552 If s, pv are NULL, calls subroutine with one argument,
5553 and type is used with error messages only. */
5556 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5560 HV *table = GvHV(PL_hintgv); /* ^H */
5564 const char *why, *why1, *why2;
5566 if (!(PL_hints & HINT_LOCALIZE_HH)) {
5569 why = "%^H is not localized";
5573 msg = Perl_newSVpvf(aTHX_ "constant(%s): %s%s%s",
5574 (type ? type: "undef"), why1, why2, why);
5575 yyerror(SvPVX(msg));
5580 why = "%^H is not defined";
5583 cvp = hv_fetch(table, key, strlen(key), FALSE);
5584 if (!cvp || !SvOK(*cvp)) {
5585 why = "} is not defined";
5590 sv_2mortal(sv); /* Parent created it permanently */
5593 pv = sv_2mortal(newSVpvn(s, len));
5595 typesv = sv_2mortal(newSVpv(type, 0));
5597 typesv = &PL_sv_undef;
5599 PUSHSTACKi(PERLSI_OVERLOAD);
5612 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5616 /* Check the eval first */
5617 if (!PL_in_eval && SvTRUE(ERRSV))
5620 sv_catpv(ERRSV, "Propagated");
5621 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5623 res = SvREFCNT_inc(sv);
5627 (void)SvREFCNT_inc(res);
5636 why = "}} did not return a defined value";
5637 why1 = "Call to &{$^H{";
5647 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
5649 register char *d = dest;
5650 register char *e = d + destlen - 3; /* two-character token, ending NUL */
5653 Perl_croak(aTHX_ ident_too_long);
5654 if (isALNUM(*s)) /* UTF handled below */
5656 else if (*s == '\'' && allow_package && isIDFIRST_lazy(s+1)) {
5661 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
5665 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5666 char *t = s + UTF8SKIP(s);
5667 while (*t & 0x80 && is_utf8_mark((U8*)t))
5669 if (d + (t - s) > e)
5670 Perl_croak(aTHX_ ident_too_long);
5671 Copy(s, d, t - s, char);
5684 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
5694 e = d + destlen - 3; /* two-character token, ending NUL */
5696 while (isDIGIT(*s)) {
5698 Perl_croak(aTHX_ ident_too_long);
5705 Perl_croak(aTHX_ ident_too_long);
5706 if (isALNUM(*s)) /* UTF handled below */
5708 else if (*s == '\'' && isIDFIRST_lazy(s+1)) {
5713 else if (*s == ':' && s[1] == ':') {
5717 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5718 char *t = s + UTF8SKIP(s);
5719 while (*t & 0x80 && is_utf8_mark((U8*)t))
5721 if (d + (t - s) > e)
5722 Perl_croak(aTHX_ ident_too_long);
5723 Copy(s, d, t - s, char);
5734 if (PL_lex_state != LEX_NORMAL)
5735 PL_lex_state = LEX_INTERPENDMAYBE;
5738 if (*s == '$' && s[1] &&
5739 (isALNUM_lazy(s+1) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
5752 if (*d == '^' && *s && isCONTROLVAR(*s)) {
5757 if (isSPACE(s[-1])) {
5760 if (ch != ' ' && ch != '\t') {
5766 if (isIDFIRST_lazy(d)) {
5770 while (e < send && isALNUM_lazy(e) || *e == ':') {
5772 while (e < send && *e & 0x80 && is_utf8_mark((U8*)e))
5775 Copy(s, d, e - s, char);
5780 while ((isALNUM(*s) || *s == ':') && d < e)
5783 Perl_croak(aTHX_ ident_too_long);
5786 while (s < send && (*s == ' ' || *s == '\t')) s++;
5787 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
5788 dTHR; /* only for ckWARN */
5789 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
5790 const char *brack = *s == '[' ? "[...]" : "{...}";
5791 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5792 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
5793 funny, dest, brack, funny, dest, brack);
5796 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
5800 /* Handle extended ${^Foo} variables
5801 * 1999-02-27 mjd-perl-patch@plover.com */
5802 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
5806 while (isALNUM(*s) && d < e) {
5810 Perl_croak(aTHX_ ident_too_long);
5815 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
5816 PL_lex_state = LEX_INTERPEND;
5819 if (PL_lex_state == LEX_NORMAL) {
5820 dTHR; /* only for ckWARN */
5821 if (ckWARN(WARN_AMBIGUOUS) &&
5822 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
5824 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5825 "Ambiguous use of %c{%s} resolved to %c%s",
5826 funny, dest, funny, dest);
5831 s = bracket; /* let the parser handle it */
5835 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
5836 PL_lex_state = LEX_INTERPEND;
5841 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
5846 *pmfl |= PMf_GLOBAL;
5848 *pmfl |= PMf_CONTINUE;
5852 *pmfl |= PMf_MULTILINE;
5854 *pmfl |= PMf_SINGLELINE;
5856 *pmfl |= PMf_EXTENDED;
5860 S_scan_pat(pTHX_ char *start, I32 type)
5865 s = scan_str(start,FALSE,FALSE);
5868 SvREFCNT_dec(PL_lex_stuff);
5869 PL_lex_stuff = Nullsv;
5870 Perl_croak(aTHX_ "Search pattern not terminated");
5873 pm = (PMOP*)newPMOP(type, 0);
5874 if (PL_multi_open == '?')
5875 pm->op_pmflags |= PMf_ONCE;
5877 while (*s && strchr("iomsx", *s))
5878 pmflag(&pm->op_pmflags,*s++);
5881 while (*s && strchr("iogcmsx", *s))
5882 pmflag(&pm->op_pmflags,*s++);
5884 pm->op_pmpermflags = pm->op_pmflags;
5886 PL_lex_op = (OP*)pm;
5887 yylval.ival = OP_MATCH;
5892 S_scan_subst(pTHX_ char *start)
5899 yylval.ival = OP_NULL;
5901 s = scan_str(start,FALSE,FALSE);
5905 SvREFCNT_dec(PL_lex_stuff);
5906 PL_lex_stuff = Nullsv;
5907 Perl_croak(aTHX_ "Substitution pattern not terminated");
5910 if (s[-1] == PL_multi_open)
5913 first_start = PL_multi_start;
5914 s = scan_str(s,FALSE,FALSE);
5917 SvREFCNT_dec(PL_lex_stuff);
5918 PL_lex_stuff = Nullsv;
5920 SvREFCNT_dec(PL_lex_repl);
5921 PL_lex_repl = Nullsv;
5922 Perl_croak(aTHX_ "Substitution replacement not terminated");
5924 PL_multi_start = first_start; /* so whole substitution is taken together */
5926 pm = (PMOP*)newPMOP(OP_SUBST, 0);
5932 else if (strchr("iogcmsx", *s))
5933 pmflag(&pm->op_pmflags,*s++);
5940 PL_sublex_info.super_bufptr = s;
5941 PL_sublex_info.super_bufend = PL_bufend;
5943 pm->op_pmflags |= PMf_EVAL;
5944 repl = newSVpvn("",0);
5946 sv_catpv(repl, es ? "eval " : "do ");
5947 sv_catpvn(repl, "{ ", 2);
5948 sv_catsv(repl, PL_lex_repl);
5949 sv_catpvn(repl, " };", 2);
5951 SvREFCNT_dec(PL_lex_repl);
5955 pm->op_pmpermflags = pm->op_pmflags;
5956 PL_lex_op = (OP*)pm;
5957 yylval.ival = OP_SUBST;
5962 S_scan_trans(pTHX_ char *start)
5973 yylval.ival = OP_NULL;
5975 s = scan_str(start,FALSE,FALSE);
5978 SvREFCNT_dec(PL_lex_stuff);
5979 PL_lex_stuff = Nullsv;
5980 Perl_croak(aTHX_ "Transliteration pattern not terminated");
5982 if (s[-1] == PL_multi_open)
5985 s = scan_str(s,FALSE,FALSE);
5988 SvREFCNT_dec(PL_lex_stuff);
5989 PL_lex_stuff = Nullsv;
5991 SvREFCNT_dec(PL_lex_repl);
5992 PL_lex_repl = Nullsv;
5993 Perl_croak(aTHX_ "Transliteration replacement not terminated");
5997 o = newSVOP(OP_TRANS, 0, 0);
5998 utf8 = OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF;
6001 New(803,tbl,256,short);
6002 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6006 complement = del = squash = 0;
6007 while (strchr("cdsCU", *s)) {
6009 complement = OPpTRANS_COMPLEMENT;
6011 del = OPpTRANS_DELETE;
6013 squash = OPpTRANS_SQUASH;
6018 utf8 &= ~OPpTRANS_FROM_UTF;
6020 utf8 |= OPpTRANS_FROM_UTF;
6024 utf8 &= ~OPpTRANS_TO_UTF;
6026 utf8 |= OPpTRANS_TO_UTF;
6029 Perl_croak(aTHX_ "Too many /C and /U options");
6034 o->op_private = del|squash|complement|utf8;
6037 yylval.ival = OP_TRANS;
6042 S_scan_heredoc(pTHX_ register char *s)
6046 I32 op_type = OP_SCALAR;
6053 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6057 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6060 for (peek = s; *peek == ' ' || *peek == '\t'; peek++) ;
6061 if (*peek && strchr("`'\"",*peek)) {
6064 s = delimcpy(d, e, s, PL_bufend, term, &len);
6074 if (!isALNUM_lazy(s))
6075 deprecate("bare << to mean <<\"\"");
6076 for (; isALNUM_lazy(s); s++) {
6081 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6082 Perl_croak(aTHX_ "Delimiter for here document is too long");
6085 len = d - PL_tokenbuf;
6086 #ifndef PERL_STRICT_CR
6087 d = strchr(s, '\r');
6091 while (s < PL_bufend) {
6097 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6106 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6111 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6112 herewas = newSVpvn(s,PL_bufend-s);
6114 s--, herewas = newSVpvn(s,d-s);
6115 s += SvCUR(herewas);
6117 tmpstr = NEWSV(87,79);
6118 sv_upgrade(tmpstr, SVt_PVIV);
6123 else if (term == '`') {
6124 op_type = OP_BACKTICK;
6125 SvIVX(tmpstr) = '\\';
6129 PL_multi_start = CopLINE(PL_curcop);
6130 PL_multi_open = PL_multi_close = '<';
6131 term = *PL_tokenbuf;
6132 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6133 char *bufptr = PL_sublex_info.super_bufptr;
6134 char *bufend = PL_sublex_info.super_bufend;
6135 char *olds = s - SvCUR(herewas);
6136 s = strchr(bufptr, '\n');
6140 while (s < bufend &&
6141 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6143 CopLINE_inc(PL_curcop);
6146 CopLINE_set(PL_curcop, PL_multi_start);
6147 missingterm(PL_tokenbuf);
6149 sv_setpvn(herewas,bufptr,d-bufptr+1);
6150 sv_setpvn(tmpstr,d+1,s-d);
6152 sv_catpvn(herewas,s,bufend-s);
6153 (void)strcpy(bufptr,SvPVX(herewas));
6160 while (s < PL_bufend &&
6161 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6163 CopLINE_inc(PL_curcop);
6165 if (s >= PL_bufend) {
6166 CopLINE_set(PL_curcop, PL_multi_start);
6167 missingterm(PL_tokenbuf);
6169 sv_setpvn(tmpstr,d+1,s-d);
6171 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6173 sv_catpvn(herewas,s,PL_bufend-s);
6174 sv_setsv(PL_linestr,herewas);
6175 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6176 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6179 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6180 while (s >= PL_bufend) { /* multiple line string? */
6182 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6183 CopLINE_set(PL_curcop, PL_multi_start);
6184 missingterm(PL_tokenbuf);
6186 CopLINE_inc(PL_curcop);
6187 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6188 #ifndef PERL_STRICT_CR
6189 if (PL_bufend - PL_linestart >= 2) {
6190 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6191 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6193 PL_bufend[-2] = '\n';
6195 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6197 else if (PL_bufend[-1] == '\r')
6198 PL_bufend[-1] = '\n';
6200 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6201 PL_bufend[-1] = '\n';
6203 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6204 SV *sv = NEWSV(88,0);
6206 sv_upgrade(sv, SVt_PVMG);
6207 sv_setsv(sv,PL_linestr);
6208 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6210 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6213 sv_catsv(PL_linestr,herewas);
6214 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6218 sv_catsv(tmpstr,PL_linestr);
6223 PL_multi_end = CopLINE(PL_curcop);
6224 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6225 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6226 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6228 SvREFCNT_dec(herewas);
6229 PL_lex_stuff = tmpstr;
6230 yylval.ival = op_type;
6235 takes: current position in input buffer
6236 returns: new position in input buffer
6237 side-effects: yylval and lex_op are set.
6242 <FH> read from filehandle
6243 <pkg::FH> read from package qualified filehandle
6244 <pkg'FH> read from package qualified filehandle
6245 <$fh> read from filehandle in $fh
6251 S_scan_inputsymbol(pTHX_ char *start)
6253 register char *s = start; /* current position in buffer */
6259 d = PL_tokenbuf; /* start of temp holding space */
6260 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6261 end = strchr(s, '\n');
6264 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6266 /* die if we didn't have space for the contents of the <>,
6267 or if it didn't end, or if we see a newline
6270 if (len >= sizeof PL_tokenbuf)
6271 Perl_croak(aTHX_ "Excessively long <> operator");
6273 Perl_croak(aTHX_ "Unterminated <> operator");
6278 Remember, only scalar variables are interpreted as filehandles by
6279 this code. Anything more complex (e.g., <$fh{$num}>) will be
6280 treated as a glob() call.
6281 This code makes use of the fact that except for the $ at the front,
6282 a scalar variable and a filehandle look the same.
6284 if (*d == '$' && d[1]) d++;
6286 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6287 while (*d && (isALNUM_lazy(d) || *d == '\'' || *d == ':'))
6290 /* If we've tried to read what we allow filehandles to look like, and
6291 there's still text left, then it must be a glob() and not a getline.
6292 Use scan_str to pull out the stuff between the <> and treat it
6293 as nothing more than a string.
6296 if (d - PL_tokenbuf != len) {
6297 yylval.ival = OP_GLOB;
6299 s = scan_str(start,FALSE,FALSE);
6301 Perl_croak(aTHX_ "Glob not terminated");
6305 /* we're in a filehandle read situation */
6308 /* turn <> into <ARGV> */
6310 (void)strcpy(d,"ARGV");
6312 /* if <$fh>, create the ops to turn the variable into a
6318 /* try to find it in the pad for this block, otherwise find
6319 add symbol table ops
6321 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6322 OP *o = newOP(OP_PADSV, 0);
6324 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o);
6327 GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV);
6328 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0,
6329 newUNOP(OP_RV2SV, 0,
6330 newGVOP(OP_GV, 0, gv)));
6332 PL_lex_op->op_flags |= OPf_SPECIAL;
6333 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6334 yylval.ival = OP_NULL;
6337 /* If it's none of the above, it must be a literal filehandle
6338 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6340 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6341 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6342 yylval.ival = OP_NULL;
6351 takes: start position in buffer
6352 keep_quoted preserve \ on the embedded delimiter(s)
6353 keep_delims preserve the delimiters around the string
6354 returns: position to continue reading from buffer
6355 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6356 updates the read buffer.
6358 This subroutine pulls a string out of the input. It is called for:
6359 q single quotes q(literal text)
6360 ' single quotes 'literal text'
6361 qq double quotes qq(interpolate $here please)
6362 " double quotes "interpolate $here please"
6363 qx backticks qx(/bin/ls -l)
6364 ` backticks `/bin/ls -l`
6365 qw quote words @EXPORT_OK = qw( func() $spam )
6366 m// regexp match m/this/
6367 s/// regexp substitute s/this/that/
6368 tr/// string transliterate tr/this/that/
6369 y/// string transliterate y/this/that/
6370 ($*@) sub prototypes sub foo ($)
6371 (stuff) sub attr parameters sub foo : attr(stuff)
6372 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6374 In most of these cases (all but <>, patterns and transliterate)
6375 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6376 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6377 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6380 It skips whitespace before the string starts, and treats the first
6381 character as the delimiter. If the delimiter is one of ([{< then
6382 the corresponding "close" character )]}> is used as the closing
6383 delimiter. It allows quoting of delimiters, and if the string has
6384 balanced delimiters ([{<>}]) it allows nesting.
6386 The lexer always reads these strings into lex_stuff, except in the
6387 case of the operators which take *two* arguments (s/// and tr///)
6388 when it checks to see if lex_stuff is full (presumably with the 1st
6389 arg to s or tr) and if so puts the string into lex_repl.
6394 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6397 SV *sv; /* scalar value: string */
6398 char *tmps; /* temp string, used for delimiter matching */
6399 register char *s = start; /* current position in the buffer */
6400 register char term; /* terminating character */
6401 register char *to; /* current position in the sv's data */
6402 I32 brackets = 1; /* bracket nesting level */
6404 /* skip space before the delimiter */
6408 /* mark where we are, in case we need to report errors */
6411 /* after skipping whitespace, the next character is the terminator */
6413 /* mark where we are */
6414 PL_multi_start = CopLINE(PL_curcop);
6415 PL_multi_open = term;
6417 /* find corresponding closing delimiter */
6418 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6420 PL_multi_close = term;
6422 /* create a new SV to hold the contents. 87 is leak category, I'm
6423 assuming. 79 is the SV's initial length. What a random number. */
6425 sv_upgrade(sv, SVt_PVIV);
6427 (void)SvPOK_only(sv); /* validate pointer */
6429 /* move past delimiter and try to read a complete string */
6431 sv_catpvn(sv, s, 1);
6434 /* extend sv if need be */
6435 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6436 /* set 'to' to the next character in the sv's string */
6437 to = SvPVX(sv)+SvCUR(sv);
6439 /* if open delimiter is the close delimiter read unbridle */
6440 if (PL_multi_open == PL_multi_close) {
6441 for (; s < PL_bufend; s++,to++) {
6442 /* embedded newlines increment the current line number */
6443 if (*s == '\n' && !PL_rsfp)
6444 CopLINE_inc(PL_curcop);
6445 /* handle quoted delimiters */
6446 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6447 if (!keep_quoted && s[1] == term)
6449 /* any other quotes are simply copied straight through */
6453 /* terminate when run out of buffer (the for() condition), or
6454 have found the terminator */
6455 else if (*s == term)
6461 /* if the terminator isn't the same as the start character (e.g.,
6462 matched brackets), we have to allow more in the quoting, and
6463 be prepared for nested brackets.
6466 /* read until we run out of string, or we find the terminator */
6467 for (; s < PL_bufend; s++,to++) {
6468 /* embedded newlines increment the line count */
6469 if (*s == '\n' && !PL_rsfp)
6470 CopLINE_inc(PL_curcop);
6471 /* backslashes can escape the open or closing characters */
6472 if (*s == '\\' && s+1 < PL_bufend) {
6474 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6479 /* allow nested opens and closes */
6480 else if (*s == PL_multi_close && --brackets <= 0)
6482 else if (*s == PL_multi_open)
6487 /* terminate the copied string and update the sv's end-of-string */
6489 SvCUR_set(sv, to - SvPVX(sv));
6492 * this next chunk reads more into the buffer if we're not done yet
6495 if (s < PL_bufend) break; /* handle case where we are done yet :-) */
6497 #ifndef PERL_STRICT_CR
6498 if (to - SvPVX(sv) >= 2) {
6499 if ((to[-2] == '\r' && to[-1] == '\n') ||
6500 (to[-2] == '\n' && to[-1] == '\r'))
6504 SvCUR_set(sv, to - SvPVX(sv));
6506 else if (to[-1] == '\r')
6509 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6513 /* if we're out of file, or a read fails, bail and reset the current
6514 line marker so we can report where the unterminated string began
6517 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6519 CopLINE_set(PL_curcop, PL_multi_start);
6522 /* we read a line, so increment our line counter */
6523 CopLINE_inc(PL_curcop);
6525 /* update debugger info */
6526 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6527 SV *sv = NEWSV(88,0);
6529 sv_upgrade(sv, SVt_PVMG);
6530 sv_setsv(sv,PL_linestr);
6531 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6534 /* having changed the buffer, we must update PL_bufend */
6535 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6538 /* at this point, we have successfully read the delimited string */
6541 sv_catpvn(sv, s, 1);
6542 PL_multi_end = CopLINE(PL_curcop);
6545 /* if we allocated too much space, give some back */
6546 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6547 SvLEN_set(sv, SvCUR(sv) + 1);
6548 Renew(SvPVX(sv), SvLEN(sv), char);
6551 /* decide whether this is the first or second quoted string we've read
6564 takes: pointer to position in buffer
6565 returns: pointer to new position in buffer
6566 side-effects: builds ops for the constant in yylval.op
6568 Read a number in any of the formats that Perl accepts:
6570 0(x[0-7A-F]+)|([0-7]+)|(b[01])
6571 [\d_]+(\.[\d_]*)?[Ee](\d+)
6573 Underbars (_) are allowed in decimal numbers. If -w is on,
6574 underbars before a decimal point must be at three digit intervals.
6576 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6579 If it reads a number without a decimal point or an exponent, it will
6580 try converting the number to an integer and see if it can do so
6581 without loss of precision.
6585 Perl_scan_num(pTHX_ char *start)
6587 register char *s = start; /* current position in buffer */
6588 register char *d; /* destination in temp buffer */
6589 register char *e; /* end of temp buffer */
6590 IV tryiv; /* used to see if it can be an IV */
6591 NV value; /* number read, as a double */
6592 SV *sv = Nullsv; /* place to put the converted number */
6593 bool floatit; /* boolean: int or float? */
6594 char *lastub = 0; /* position of last underbar */
6595 static char number_too_long[] = "Number too long";
6597 /* We use the first character to decide what type of number this is */
6601 Perl_croak(aTHX_ "panic: scan_num");
6603 /* if it starts with a 0, it could be an octal number, a decimal in
6604 0.13 disguise, or a hexadecimal number, or a binary number. */
6608 u holds the "number so far"
6609 shift the power of 2 of the base
6610 (hex == 4, octal == 3, binary == 1)
6611 overflowed was the number more than we can hold?
6613 Shift is used when we add a digit. It also serves as an "are
6614 we in octal/hex/binary?" indicator to disallow hex characters
6621 bool overflowed = FALSE;
6622 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6623 static char* bases[5] = { "", "binary", "", "octal",
6625 static char* Bases[5] = { "", "Binary", "", "Octal",
6627 static char *maxima[5] = { "",
6628 "0b11111111111111111111111111111111",
6632 char *base, *Base, *max;
6638 } else if (s[1] == 'b') {
6642 /* check for a decimal in disguise */
6643 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
6645 /* so it must be octal */
6649 base = bases[shift];
6650 Base = Bases[shift];
6651 max = maxima[shift];
6653 /* read the rest of the number */
6655 /* x is used in the overflow test,
6656 b is the digit we're adding on. */
6661 /* if we don't mention it, we're done */
6670 /* 8 and 9 are not octal */
6673 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
6677 case '2': case '3': case '4':
6678 case '5': case '6': case '7':
6680 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
6684 b = *s++ & 15; /* ASCII digit -> value of digit */
6688 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
6689 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
6690 /* make sure they said 0x */
6695 /* Prepare to put the digit we have onto the end
6696 of the number so far. We check for overflows.
6701 x = u << shift; /* make room for the digit */
6703 if ((x >> shift) != u
6704 && !(PL_hints & HINT_NEW_BINARY)) {
6708 if (ckWARN_d(WARN_OVERFLOW))
6709 Perl_warner(aTHX_ WARN_OVERFLOW,
6710 "Integer overflow in %s number",
6713 u = x | b; /* add the digit to the end */
6716 n *= nvshift[shift];
6717 /* If an NV has not enough bits in its
6718 * mantissa to represent an UV this summing of
6719 * small low-order numbers is a waste of time
6720 * (because the NV cannot preserve the
6721 * low-order bits anyway): we could just
6722 * remember when did we overflow and in the
6723 * end just multiply n by the right
6731 /* if we get here, we had success: make a scalar value from
6738 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
6739 Perl_warner(aTHX_ WARN_PORTABLE,
6740 "%s number > %s non-portable",
6747 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
6748 Perl_warner(aTHX_ WARN_PORTABLE,
6749 "%s number > %s non-portable",
6754 if (PL_hints & HINT_NEW_BINARY)
6755 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
6760 handle decimal numbers.
6761 we're also sent here when we read a 0 as the first digit
6763 case '1': case '2': case '3': case '4': case '5':
6764 case '6': case '7': case '8': case '9': case '.':
6767 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
6770 /* read next group of digits and _ and copy into d */
6771 while (isDIGIT(*s) || *s == '_') {
6772 /* skip underscores, checking for misplaced ones
6776 dTHR; /* only for ckWARN */
6777 if (ckWARN(WARN_SYNTAX) && lastub && s - lastub != 3)
6778 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6782 /* check for end of fixed-length buffer */
6784 Perl_croak(aTHX_ number_too_long);
6785 /* if we're ok, copy the character */
6790 /* final misplaced underbar check */
6791 if (lastub && s - lastub != 3) {
6793 if (ckWARN(WARN_SYNTAX))
6794 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6797 /* read a decimal portion if there is one. avoid
6798 3..5 being interpreted as the number 3. followed
6801 if (*s == '.' && s[1] != '.') {
6805 /* copy, ignoring underbars, until we run out of
6806 digits. Note: no misplaced underbar checks!
6808 for (; isDIGIT(*s) || *s == '_'; s++) {
6809 /* fixed length buffer check */
6811 Perl_croak(aTHX_ number_too_long);
6817 /* read exponent part, if present */
6818 if (*s && strchr("eE",*s) && strchr("+-0123456789",s[1])) {
6822 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
6823 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
6825 /* allow positive or negative exponent */
6826 if (*s == '+' || *s == '-')
6829 /* read digits of exponent (no underbars :-) */
6830 while (isDIGIT(*s)) {
6832 Perl_croak(aTHX_ number_too_long);
6837 /* terminate the string */
6840 /* make an sv from the string */
6843 value = Atof(PL_tokenbuf);
6846 See if we can make do with an integer value without loss of
6847 precision. We use I_V to cast to an int, because some
6848 compilers have issues. Then we try casting it back and see
6849 if it was the same. We only do this if we know we
6850 specifically read an integer.
6852 Note: if floatit is true, then we don't need to do the
6856 if (!floatit && (NV)tryiv == value)
6857 sv_setiv(sv, tryiv);
6859 sv_setnv(sv, value);
6860 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
6861 (PL_hints & HINT_NEW_INTEGER) )
6862 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
6863 (floatit ? "float" : "integer"),
6866 /* if it starts with a v, it could be a version number */
6871 while (isDIGIT(*pos))
6873 if (*pos == '.' && isDIGIT(pos[1])) {
6878 s++; /* get past 'v' */
6881 SvUPGRADE(sv, SVt_PVNV);
6882 sv_setpvn(sv, "", 0);
6887 while (isDIGIT(*pos))
6890 tmpend = uv_to_utf8(tmpbuf, rev);
6892 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
6894 SvNVX(sv) += (NV)rev/nshift;
6896 } while (*pos == '.' && isDIGIT(pos[1]));
6900 tmpend = uv_to_utf8(tmpbuf, rev);
6902 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
6904 SvNVX(sv) += (NV)rev/nshift;
6915 /* make the op for the constant and return */
6918 yylval.opval = newSVOP(OP_CONST, 0, sv);
6920 yylval.opval = Nullop;
6926 S_scan_formline(pTHX_ register char *s)
6931 SV *stuff = newSVpvn("",0);
6932 bool needargs = FALSE;
6935 if (*s == '.' || *s == '}') {
6937 #ifdef PERL_STRICT_CR
6938 for (t = s+1;*t == ' ' || *t == '\t'; t++) ;
6940 for (t = s+1;*t == ' ' || *t == '\t' || *t == '\r'; t++) ;
6942 if (*t == '\n' || t == PL_bufend)
6945 if (PL_in_eval && !PL_rsfp) {
6946 eol = strchr(s,'\n');
6951 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6953 for (t = s; t < eol; t++) {
6954 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
6956 goto enough; /* ~~ must be first line in formline */
6958 if (*t == '@' || *t == '^')
6961 sv_catpvn(stuff, s, eol-s);
6962 #ifndef PERL_STRICT_CR
6963 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
6964 char *end = SvPVX(stuff) + SvCUR(stuff);
6973 s = filter_gets(PL_linestr, PL_rsfp, 0);
6974 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
6975 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
6978 yyerror("Format not terminated");
6988 PL_lex_state = LEX_NORMAL;
6989 PL_nextval[PL_nexttoke].ival = 0;
6993 PL_lex_state = LEX_FORMLINE;
6994 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
6996 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7000 SvREFCNT_dec(stuff);
7001 PL_lex_formbrack = 0;
7012 PL_cshlen = strlen(PL_cshname);
7017 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7020 I32 oldsavestack_ix = PL_savestack_ix;
7021 CV* outsidecv = PL_compcv;
7025 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7027 SAVEI32(PL_subline);
7028 save_item(PL_subname);
7030 SAVEVPTR(PL_curpad);
7031 SAVESPTR(PL_comppad);
7032 SAVESPTR(PL_comppad_name);
7033 SAVESPTR(PL_compcv);
7034 SAVEI32(PL_comppad_name_fill);
7035 SAVEI32(PL_min_intro_pending);
7036 SAVEI32(PL_max_intro_pending);
7037 SAVEI32(PL_pad_reset_pending);
7039 PL_compcv = (CV*)NEWSV(1104,0);
7040 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7041 CvFLAGS(PL_compcv) |= flags;
7043 PL_comppad = newAV();
7044 av_push(PL_comppad, Nullsv);
7045 PL_curpad = AvARRAY(PL_comppad);
7046 PL_comppad_name = newAV();
7047 PL_comppad_name_fill = 0;
7048 PL_min_intro_pending = 0;
7050 PL_subline = CopLINE(PL_curcop);
7052 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
7053 PL_curpad[0] = (SV*)newAV();
7054 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
7055 #endif /* USE_THREADS */
7057 comppadlist = newAV();
7058 AvREAL_off(comppadlist);
7059 av_store(comppadlist, 0, (SV*)PL_comppad_name);
7060 av_store(comppadlist, 1, (SV*)PL_comppad);
7062 CvPADLIST(PL_compcv) = comppadlist;
7063 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7065 CvOWNER(PL_compcv) = 0;
7066 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7067 MUTEX_INIT(CvMUTEXP(PL_compcv));
7068 #endif /* USE_THREADS */
7070 return oldsavestack_ix;
7074 Perl_yywarn(pTHX_ char *s)
7077 PL_in_eval |= EVAL_WARNONLY;
7079 PL_in_eval &= ~EVAL_WARNONLY;
7084 Perl_yyerror(pTHX_ char *s)
7088 char *context = NULL;
7092 if (!yychar || (yychar == ';' && !PL_rsfp))
7094 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7095 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7096 while (isSPACE(*PL_oldoldbufptr))
7098 context = PL_oldoldbufptr;
7099 contlen = PL_bufptr - PL_oldoldbufptr;
7101 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7102 PL_oldbufptr != PL_bufptr) {
7103 while (isSPACE(*PL_oldbufptr))
7105 context = PL_oldbufptr;
7106 contlen = PL_bufptr - PL_oldbufptr;
7108 else if (yychar > 255)
7109 where = "next token ???";
7110 else if ((yychar & 127) == 127) {
7111 if (PL_lex_state == LEX_NORMAL ||
7112 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7113 where = "at end of line";
7114 else if (PL_lex_inpat)
7115 where = "within pattern";
7117 where = "within string";
7120 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7122 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7123 else if (isPRINT_LC(yychar))
7124 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7126 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7127 where = SvPVX(where_sv);
7129 msg = sv_2mortal(newSVpv(s, 0));
7130 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7131 CopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7133 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7135 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7136 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7137 Perl_sv_catpvf(aTHX_ msg,
7138 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7139 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7142 if (PL_in_eval & EVAL_WARNONLY)
7143 Perl_warn(aTHX_ "%"SVf, msg);
7146 if (PL_error_count >= 10)
7147 Perl_croak(aTHX_ "%s has too many errors.\n", CopFILE(PL_curcop));
7149 PL_in_my_stash = Nullhv;
7160 * Restore a source filter.
7164 restore_rsfp(pTHXo_ void *f)
7166 PerlIO *fp = (PerlIO*)f;
7168 if (PL_rsfp == PerlIO_stdin())
7169 PerlIO_clearerr(PL_rsfp);
7170 else if (PL_rsfp && (PL_rsfp != fp))
7171 PerlIO_close(PL_rsfp);