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 CopFILE_set(PL_curcop, PL_origfilename);
501 CopLINE_set(PL_curcop, atoi(n)-1);
506 * Called to gobble the appropriate amount and type of whitespace.
507 * Skips comments as well.
511 S_skipspace(pTHX_ register char *s)
514 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
515 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
521 SSize_t oldprevlen, oldoldprevlen;
522 SSize_t oldloplen, oldunilen;
523 while (s < PL_bufend && isSPACE(*s)) {
524 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
529 if (s < PL_bufend && *s == '#') {
530 while (s < PL_bufend && *s != '\n')
534 if (PL_in_eval && !PL_rsfp) {
541 /* only continue to recharge the buffer if we're at the end
542 * of the buffer, we're not reading from a source filter, and
543 * we're in normal lexing mode
545 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
546 PL_lex_state == LEX_FORMLINE)
549 /* try to recharge the buffer */
550 if ((s = filter_gets(PL_linestr, PL_rsfp,
551 (prevlen = SvCUR(PL_linestr)))) == Nullch)
553 /* end of file. Add on the -p or -n magic */
554 if (PL_minus_n || PL_minus_p) {
555 sv_setpv(PL_linestr,PL_minus_p ?
556 ";}continue{print or die qq(-p destination: $!\\n)" :
558 sv_catpv(PL_linestr,";}");
559 PL_minus_n = PL_minus_p = 0;
562 sv_setpv(PL_linestr,";");
564 /* reset variables for next time we lex */
565 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
567 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
569 /* Close the filehandle. Could be from -P preprocessor,
570 * STDIN, or a regular file. If we were reading code from
571 * STDIN (because the commandline held no -e or filename)
572 * then we don't close it, we reset it so the code can
573 * read from STDIN too.
576 if (PL_preprocess && !PL_in_eval)
577 (void)PerlProc_pclose(PL_rsfp);
578 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
579 PerlIO_clearerr(PL_rsfp);
581 (void)PerlIO_close(PL_rsfp);
586 /* not at end of file, so we only read another line */
587 /* make corresponding updates to old pointers, for yyerror() */
588 oldprevlen = PL_oldbufptr - PL_bufend;
589 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
591 oldunilen = PL_last_uni - PL_bufend;
593 oldloplen = PL_last_lop - PL_bufend;
594 PL_linestart = PL_bufptr = s + prevlen;
595 PL_bufend = s + SvCUR(PL_linestr);
597 PL_oldbufptr = s + oldprevlen;
598 PL_oldoldbufptr = s + oldoldprevlen;
600 PL_last_uni = s + oldunilen;
602 PL_last_lop = s + oldloplen;
605 /* debugger active and we're not compiling the debugger code,
606 * so store the line into the debugger's array of lines
608 if (PERLDB_LINE && PL_curstash != PL_debstash) {
609 SV *sv = NEWSV(85,0);
611 sv_upgrade(sv, SVt_PVMG);
612 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
613 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
620 * Check the unary operators to ensure there's no ambiguity in how they're
621 * used. An ambiguous piece of code would be:
623 * This doesn't mean rand() + 5. Because rand() is a unary operator,
624 * the +5 is its argument.
634 if (PL_oldoldbufptr != PL_last_uni)
636 while (isSPACE(*PL_last_uni))
638 for (s = PL_last_uni; isALNUM_lazy(s) || *s == '-'; s++) ;
639 if ((t = strchr(s, '(')) && t < PL_bufptr)
641 if (ckWARN_d(WARN_AMBIGUOUS)){
644 Perl_warner(aTHX_ WARN_AMBIGUOUS,
645 "Warning: Use of \"%s\" without parens is ambiguous",
651 /* workaround to replace the UNI() macro with a function. Only the
652 * hints/uts.sh file mentions this. Other comments elsewhere in the
653 * source indicate Microport Unix might need it too.
659 #define UNI(f) return uni(f,s)
662 S_uni(pTHX_ I32 f, char *s)
667 PL_last_uni = PL_oldbufptr;
678 #endif /* CRIPPLED_CC */
681 * LOP : macro to build a list operator. Its behaviour has been replaced
682 * with a subroutine, S_lop() for which LOP is just another name.
685 #define LOP(f,x) return lop(f,x,s)
689 * Build a list operator (or something that might be one). The rules:
690 * - if we have a next token, then it's a list operator [why?]
691 * - if the next thing is an opening paren, then it's a function
692 * - else it's a list operator
696 S_lop(pTHX_ I32 f, int x, char *s)
703 PL_last_lop = PL_oldbufptr;
718 * When the lexer realizes it knows the next token (for instance,
719 * it is reordering tokens for the parser) then it can call S_force_next
720 * to know what token to return the next time the lexer is called. Caller
721 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
722 * handles the token correctly.
726 S_force_next(pTHX_ I32 type)
728 PL_nexttype[PL_nexttoke] = type;
730 if (PL_lex_state != LEX_KNOWNEXT) {
731 PL_lex_defer = PL_lex_state;
732 PL_lex_expect = PL_expect;
733 PL_lex_state = LEX_KNOWNEXT;
739 * When the lexer knows the next thing is a word (for instance, it has
740 * just seen -> and it knows that the next char is a word char, then
741 * it calls S_force_word to stick the next word into the PL_next lookahead.
744 * char *start : buffer position (must be within PL_linestr)
745 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
746 * int check_keyword : if true, Perl checks to make sure the word isn't
747 * a keyword (do this if the word is a label, e.g. goto FOO)
748 * int allow_pack : if true, : characters will also be allowed (require,
750 * int allow_initial_tick : used by the "sub" lexer only.
754 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
759 start = skipspace(start);
761 if (isIDFIRST_lazy(s) ||
762 (allow_pack && *s == ':') ||
763 (allow_initial_tick && *s == '\'') )
765 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
766 if (check_keyword && keyword(PL_tokenbuf, len))
768 if (token == METHOD) {
773 PL_expect = XOPERATOR;
776 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0));
777 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
785 * Called when the lexer wants $foo *foo &foo etc, but the program
786 * text only contains the "foo" portion. The first argument is a pointer
787 * to the "foo", and the second argument is the type symbol to prefix.
788 * Forces the next token to be a "WORD".
789 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
793 S_force_ident(pTHX_ register char *s, int kind)
796 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
797 PL_nextval[PL_nexttoke].opval = o;
800 dTHR; /* just for in_eval */
801 o->op_private = OPpCONST_ENTERED;
802 /* XXX see note in pp_entereval() for why we forgo typo
803 warnings if the symbol must be introduced in an eval.
805 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
806 kind == '$' ? SVt_PV :
807 kind == '@' ? SVt_PVAV :
808 kind == '%' ? SVt_PVHV :
817 * Forces the next token to be a version number.
821 S_force_version(pTHX_ char *s)
823 OP *version = Nullop;
827 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
831 for (; isDIGIT(*d) || *d == '_' || *d == '.'; d++);
832 if ((*d == ';' || isSPACE(*d)) && *(skipspace(d)) != ',') {
834 /* real VERSION number -- GBARR */
835 version = yylval.opval;
839 /* NOTE: The parser sees the package name and the VERSION swapped */
840 PL_nextval[PL_nexttoke].opval = version;
848 * Tokenize a quoted string passed in as an SV. It finds the next
849 * chunk, up to end of string or a backslash. It may make a new
850 * SV containing that chunk (if HINT_NEW_STRING is on). It also
855 S_tokeq(pTHX_ SV *sv)
866 s = SvPV_force(sv, len);
870 while (s < send && *s != '\\')
875 if ( PL_hints & HINT_NEW_STRING )
876 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
879 if (s + 1 < send && (s[1] == '\\'))
880 s++; /* all that, just for this */
885 SvCUR_set(sv, d - SvPVX(sv));
887 if ( PL_hints & HINT_NEW_STRING )
888 return new_constant(NULL, 0, "q", sv, pv, "q");
893 * Now come three functions related to double-quote context,
894 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
895 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
896 * interact with PL_lex_state, and create fake ( ... ) argument lists
897 * to handle functions and concatenation.
898 * They assume that whoever calls them will be setting up a fake
899 * join call, because each subthing puts a ',' after it. This lets
902 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
904 * (I'm not sure whether the spurious commas at the end of lcfirst's
905 * arguments and join's arguments are created or not).
910 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
912 * Pattern matching will set PL_lex_op to the pattern-matching op to
913 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
915 * OP_CONST and OP_READLINE are easy--just make the new op and return.
917 * Everything else becomes a FUNC.
919 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
920 * had an OP_CONST or OP_READLINE). This just sets us up for a
921 * call to S_sublex_push().
927 register I32 op_type = yylval.ival;
929 if (op_type == OP_NULL) {
930 yylval.opval = PL_lex_op;
934 if (op_type == OP_CONST || op_type == OP_READLINE) {
935 SV *sv = tokeq(PL_lex_stuff);
937 if (SvTYPE(sv) == SVt_PVIV) {
938 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
944 nsv = newSVpvn(p, len);
948 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
949 PL_lex_stuff = Nullsv;
953 PL_sublex_info.super_state = PL_lex_state;
954 PL_sublex_info.sub_inwhat = op_type;
955 PL_sublex_info.sub_op = PL_lex_op;
956 PL_lex_state = LEX_INTERPPUSH;
960 yylval.opval = PL_lex_op;
970 * Create a new scope to save the lexing state. The scope will be
971 * ended in S_sublex_done. Returns a '(', starting the function arguments
972 * to the uc, lc, etc. found before.
973 * Sets PL_lex_state to LEX_INTERPCONCAT.
982 PL_lex_state = PL_sublex_info.super_state;
983 SAVEI32(PL_lex_dojoin);
984 SAVEI32(PL_lex_brackets);
985 SAVEI32(PL_lex_casemods);
986 SAVEI32(PL_lex_starts);
987 SAVEI32(PL_lex_state);
988 SAVEVPTR(PL_lex_inpat);
989 SAVEI32(PL_lex_inwhat);
990 SAVECOPLINE(PL_curcop);
992 SAVEPPTR(PL_oldbufptr);
993 SAVEPPTR(PL_oldoldbufptr);
994 SAVEPPTR(PL_linestart);
995 SAVESPTR(PL_linestr);
996 SAVEPPTR(PL_lex_brackstack);
997 SAVEPPTR(PL_lex_casestack);
999 PL_linestr = PL_lex_stuff;
1000 PL_lex_stuff = Nullsv;
1002 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1003 = SvPVX(PL_linestr);
1004 PL_bufend += SvCUR(PL_linestr);
1005 SAVEFREESV(PL_linestr);
1007 PL_lex_dojoin = FALSE;
1008 PL_lex_brackets = 0;
1009 New(899, PL_lex_brackstack, 120, char);
1010 New(899, PL_lex_casestack, 12, char);
1011 SAVEFREEPV(PL_lex_brackstack);
1012 SAVEFREEPV(PL_lex_casestack);
1013 PL_lex_casemods = 0;
1014 *PL_lex_casestack = '\0';
1016 PL_lex_state = LEX_INTERPCONCAT;
1017 CopLINE_set(PL_curcop, PL_multi_start);
1019 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1020 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1021 PL_lex_inpat = PL_sublex_info.sub_op;
1023 PL_lex_inpat = Nullop;
1030 * Restores lexer state after a S_sublex_push.
1036 if (!PL_lex_starts++) {
1037 PL_expect = XOPERATOR;
1038 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpvn("",0));
1042 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1043 PL_lex_state = LEX_INTERPCASEMOD;
1047 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1048 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1049 PL_linestr = PL_lex_repl;
1051 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1052 PL_bufend += SvCUR(PL_linestr);
1053 SAVEFREESV(PL_linestr);
1054 PL_lex_dojoin = FALSE;
1055 PL_lex_brackets = 0;
1056 PL_lex_casemods = 0;
1057 *PL_lex_casestack = '\0';
1059 if (SvEVALED(PL_lex_repl)) {
1060 PL_lex_state = LEX_INTERPNORMAL;
1062 /* we don't clear PL_lex_repl here, so that we can check later
1063 whether this is an evalled subst; that means we rely on the
1064 logic to ensure sublex_done() is called again only via the
1065 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1068 PL_lex_state = LEX_INTERPCONCAT;
1069 PL_lex_repl = Nullsv;
1075 PL_bufend = SvPVX(PL_linestr);
1076 PL_bufend += SvCUR(PL_linestr);
1077 PL_expect = XOPERATOR;
1078 PL_sublex_info.sub_inwhat = 0;
1086 Extracts a pattern, double-quoted string, or transliteration. This
1089 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1090 processing a pattern (PL_lex_inpat is true), a transliteration
1091 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1093 Returns a pointer to the character scanned up to. Iff this is
1094 advanced from the start pointer supplied (ie if anything was
1095 successfully parsed), will leave an OP for the substring scanned
1096 in yylval. Caller must intuit reason for not parsing further
1097 by looking at the next characters herself.
1101 double-quoted style: \r and \n
1102 regexp special ones: \D \s
1104 backrefs: \1 (deprecated in substitution replacements)
1105 case and quoting: \U \Q \E
1106 stops on @ and $, but not for $ as tail anchor
1108 In transliterations:
1109 characters are VERY literal, except for - not at the start or end
1110 of the string, which indicates a range. scan_const expands the
1111 range to the full set of intermediate characters.
1113 In double-quoted strings:
1115 double-quoted style: \r and \n
1117 backrefs: \1 (deprecated)
1118 case and quoting: \U \Q \E
1121 scan_const does *not* construct ops to handle interpolated strings.
1122 It stops processing as soon as it finds an embedded $ or @ variable
1123 and leaves it to the caller to work out what's going on.
1125 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @:foo.
1127 $ in pattern could be $foo or could be tail anchor. Assumption:
1128 it's a tail anchor if $ is the last thing in the string, or if it's
1129 followed by one of ")| \n\t"
1131 \1 (backreferences) are turned into $1
1133 The structure of the code is
1134 while (there's a character to process) {
1135 handle transliteration ranges
1136 skip regexp comments
1137 skip # initiated comments in //x patterns
1138 check for embedded @foo
1139 check for embedded scalars
1141 leave intact backslashes from leave (below)
1142 deprecate \1 in strings and sub replacements
1143 handle string-changing backslashes \l \U \Q \E, etc.
1144 switch (what was escaped) {
1145 handle - in a transliteration (becomes a literal -)
1146 handle \132 octal characters
1147 handle 0x15 hex characters
1148 handle \cV (control V)
1149 handle printf backslashes (\f, \r, \n, etc)
1151 } (end if backslash)
1152 } (end while character to read)
1157 S_scan_const(pTHX_ char *start)
1159 register char *send = PL_bufend; /* end of the constant */
1160 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1161 register char *s = start; /* start of the constant */
1162 register char *d = SvPVX(sv); /* destination for copies */
1163 bool dorange = FALSE; /* are we in a translit range? */
1165 I32 utf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1166 ? (PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
1168 I32 thisutf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1169 ? (PL_sublex_info.sub_op->op_private & (PL_lex_repl ?
1170 OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF))
1172 const char *leaveit = /* set of acceptably-backslashed characters */
1174 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
1177 while (s < send || dorange) {
1178 /* get transliterations out of the way (they're most literal) */
1179 if (PL_lex_inwhat == OP_TRANS) {
1180 /* expand a range A-Z to the full set of characters. AIE! */
1182 I32 i; /* current expanded character */
1183 I32 min; /* first character in range */
1184 I32 max; /* last character in range */
1186 i = d - SvPVX(sv); /* remember current offset */
1187 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1188 d = SvPVX(sv) + i; /* refresh d after realloc */
1189 d -= 2; /* eat the first char and the - */
1191 min = (U8)*d; /* first char in range */
1192 max = (U8)d[1]; /* last char in range */
1195 if ((isLOWER(min) && isLOWER(max)) ||
1196 (isUPPER(min) && isUPPER(max))) {
1198 for (i = min; i <= max; i++)
1202 for (i = min; i <= max; i++)
1209 for (i = min; i <= max; i++)
1212 /* mark the range as done, and continue */
1217 /* range begins (ignore - as first or last char) */
1218 else if (*s == '-' && s+1 < send && s != start) {
1220 *d++ = (char)0xff; /* use illegal utf8 byte--see pmtrans */
1229 /* if we get here, we're not doing a transliteration */
1231 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1232 except for the last char, which will be done separately. */
1233 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1235 while (s < send && *s != ')')
1237 } else if (s[2] == '{'
1238 || s[2] == 'p' && s[3] == '{') { /* This should march regcomp.c */
1240 char *regparse = s + (s[2] == '{' ? 3 : 4);
1243 while (count && (c = *regparse)) {
1244 if (c == '\\' && regparse[1])
1252 if (*regparse != ')') {
1253 regparse--; /* Leave one char for continuation. */
1254 yyerror("Sequence (?{...}) not terminated or not {}-balanced");
1256 while (s < regparse)
1261 /* likewise skip #-initiated comments in //x patterns */
1262 else if (*s == '#' && PL_lex_inpat &&
1263 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1264 while (s+1 < send && *s != '\n')
1268 /* check for embedded arrays (@foo, @:foo, @'foo, @{foo}, @$foo) */
1269 else if (*s == '@' && s[1] && (isALNUM_lazy(s+1) || strchr(":'{$", s[1])))
1272 /* check for embedded scalars. only stop if we're sure it's a
1275 else if (*s == '$') {
1276 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1278 if (s + 1 < send && !strchr("()| \n\t", s[1]))
1279 break; /* in regexp, $ might be tail anchor */
1282 /* (now in tr/// code again) */
1284 if (*s & 0x80 && thisutf) {
1285 dTHR; /* only for ckWARN */
1286 if (ckWARN(WARN_UTF8)) {
1287 (void)utf8_to_uv((U8*)s, &len); /* could cvt latin-1 to utf8 here... */
1297 if (*s == '\\' && s+1 < send) {
1300 /* some backslashes we leave behind */
1301 if (*leaveit && *s && strchr(leaveit, *s)) {
1307 /* deprecate \1 in strings and substitution replacements */
1308 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1309 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1311 dTHR; /* only for ckWARN */
1312 if (ckWARN(WARN_SYNTAX))
1313 Perl_warner(aTHX_ WARN_SYNTAX, "\\%c better written as $%c", *s, *s);
1318 /* string-change backslash escapes */
1319 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1324 /* if we get here, it's either a quoted -, or a digit */
1327 /* quoted - in transliterations */
1329 if (PL_lex_inwhat == OP_TRANS) {
1337 if (ckWARN(WARN_UNSAFE) && isALPHA(*s))
1338 Perl_warner(aTHX_ WARN_UNSAFE,
1339 "Unrecognized escape \\%c passed through",
1341 /* default action is to copy the quoted character */
1346 /* \132 indicates an octal constant */
1347 case '0': case '1': case '2': case '3':
1348 case '4': case '5': case '6': case '7':
1349 *d++ = (char)scan_oct(s, 3, &len);
1353 /* \x24 indicates a hex constant */
1357 char* e = strchr(s, '}');
1360 yyerror("Missing right brace on \\x{}");
1365 if (ckWARN(WARN_UTF8))
1366 Perl_warner(aTHX_ WARN_UTF8,
1367 "Use of \\x{} without utf8 declaration");
1369 /* note: utf always shorter than hex */
1370 d = (char*)uv_to_utf8((U8*)d,
1371 (UV)scan_hex(s + 1, e - s - 1, &len));
1375 UV uv = (UV)scan_hex(s, 2, &len);
1376 if (utf && PL_lex_inwhat == OP_TRANS &&
1377 utf != (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
1379 d = (char*)uv_to_utf8((U8*)d, uv); /* doing a CU or UC */
1382 if (uv >= 127 && UTF) {
1384 if (ckWARN(WARN_UTF8))
1385 Perl_warner(aTHX_ WARN_UTF8,
1386 "\\x%.*s will produce malformed UTF-8 character; use \\x{%.*s} for that",
1395 /* \N{latin small letter a} is a named character */
1399 char* e = strchr(s, '}');
1408 yyerror("Missing right brace on \\N{}");
1412 res = newSVpvn(s + 1, e - s - 1);
1413 res = new_constant( Nullch, 0, "charnames",
1414 res, Nullsv, "\\N{...}" );
1415 str = SvPV(res,len);
1416 if (len > e - s + 4) {
1417 char *odest = SvPVX(sv);
1419 SvGROW(sv, (SvCUR(sv) + len - (e - s + 4)));
1420 d = SvPVX(sv) + (d - odest);
1422 Copy(str, d, len, char);
1429 yyerror("Missing braces on \\N{}");
1432 /* \c is a control character */
1446 /* printf-style backslashes, formfeeds, newlines, etc */
1464 *d++ = '\047'; /* CP 1047 */
1467 *d++ = '\057'; /* CP 1047 */
1481 } /* end if (backslash) */
1484 } /* while loop to process each character */
1486 /* terminate the string and set up the sv */
1488 SvCUR_set(sv, d - SvPVX(sv));
1491 /* shrink the sv if we allocated more than we used */
1492 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1493 SvLEN_set(sv, SvCUR(sv) + 1);
1494 Renew(SvPVX(sv), SvLEN(sv), char);
1497 /* return the substring (via yylval) only if we parsed anything */
1498 if (s > PL_bufptr) {
1499 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1500 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1502 ( PL_lex_inwhat == OP_TRANS
1504 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1507 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1514 * Returns TRUE if there's more to the expression (e.g., a subscript),
1517 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1519 * ->[ and ->{ return TRUE
1520 * { and [ outside a pattern are always subscripts, so return TRUE
1521 * if we're outside a pattern and it's not { or [, then return FALSE
1522 * if we're in a pattern and the first char is a {
1523 * {4,5} (any digits around the comma) returns FALSE
1524 * if we're in a pattern and the first char is a [
1526 * [SOMETHING] has a funky algorithm to decide whether it's a
1527 * character class or not. It has to deal with things like
1528 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1529 * anything else returns TRUE
1532 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1535 S_intuit_more(pTHX_ register char *s)
1537 if (PL_lex_brackets)
1539 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1541 if (*s != '{' && *s != '[')
1546 /* In a pattern, so maybe we have {n,m}. */
1563 /* On the other hand, maybe we have a character class */
1566 if (*s == ']' || *s == '^')
1569 /* this is terrifying, and it works */
1570 int weight = 2; /* let's weigh the evidence */
1572 unsigned char un_char = 255, last_un_char;
1573 char *send = strchr(s,']');
1574 char tmpbuf[sizeof PL_tokenbuf * 4];
1576 if (!send) /* has to be an expression */
1579 Zero(seen,256,char);
1582 else if (isDIGIT(*s)) {
1584 if (isDIGIT(s[1]) && s[2] == ']')
1590 for (; s < send; s++) {
1591 last_un_char = un_char;
1592 un_char = (unsigned char)*s;
1597 weight -= seen[un_char] * 10;
1598 if (isALNUM_lazy(s+1)) {
1599 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1600 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1605 else if (*s == '$' && s[1] &&
1606 strchr("[#!%*<>()-=",s[1])) {
1607 if (/*{*/ strchr("])} =",s[2]))
1616 if (strchr("wds]",s[1]))
1618 else if (seen['\''] || seen['"'])
1620 else if (strchr("rnftbxcav",s[1]))
1622 else if (isDIGIT(s[1])) {
1624 while (s[1] && isDIGIT(s[1]))
1634 if (strchr("aA01! ",last_un_char))
1636 if (strchr("zZ79~",s[1]))
1638 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1639 weight -= 5; /* cope with negative subscript */
1642 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1643 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1648 if (keyword(tmpbuf, d - tmpbuf))
1651 if (un_char == last_un_char + 1)
1653 weight -= seen[un_char];
1658 if (weight >= 0) /* probably a character class */
1668 * Does all the checking to disambiguate
1670 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1671 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1673 * First argument is the stuff after the first token, e.g. "bar".
1675 * Not a method if bar is a filehandle.
1676 * Not a method if foo is a subroutine prototyped to take a filehandle.
1677 * Not a method if it's really "Foo $bar"
1678 * Method if it's "foo $bar"
1679 * Not a method if it's really "print foo $bar"
1680 * Method if it's really "foo package::" (interpreted as package->foo)
1681 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1682 * Not a method if bar is a filehandle or package, but is quotd with
1687 S_intuit_method(pTHX_ char *start, GV *gv)
1689 char *s = start + (*start == '$');
1690 char tmpbuf[sizeof PL_tokenbuf];
1698 if ((cv = GvCVu(gv))) {
1699 char *proto = SvPVX(cv);
1709 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1710 /* start is the beginning of the possible filehandle/object,
1711 * and s is the end of it
1712 * tmpbuf is a copy of it
1715 if (*start == '$') {
1716 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1721 return *s == '(' ? FUNCMETH : METHOD;
1723 if (!keyword(tmpbuf, len)) {
1724 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1729 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1730 if (indirgv && GvCVu(indirgv))
1732 /* filehandle or package name makes it a method */
1733 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1735 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1736 return 0; /* no assumptions -- "=>" quotes bearword */
1738 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1739 newSVpvn(tmpbuf,len));
1740 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1744 return *s == '(' ? FUNCMETH : METHOD;
1752 * Return a string of Perl code to load the debugger. If PERL5DB
1753 * is set, it will return the contents of that, otherwise a
1754 * compile-time require of perl5db.pl.
1761 char *pdb = PerlEnv_getenv("PERL5DB");
1765 SETERRNO(0,SS$_NORMAL);
1766 return "BEGIN { require 'perl5db.pl' }";
1772 /* Encoded script support. filter_add() effectively inserts a
1773 * 'pre-processing' function into the current source input stream.
1774 * Note that the filter function only applies to the current source file
1775 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1777 * The datasv parameter (which may be NULL) can be used to pass
1778 * private data to this instance of the filter. The filter function
1779 * can recover the SV using the FILTER_DATA macro and use it to
1780 * store private buffers and state information.
1782 * The supplied datasv parameter is upgraded to a PVIO type
1783 * and the IoDIRP field is used to store the function pointer,
1784 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1785 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1786 * private use must be set using malloc'd pointers.
1790 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1795 if (!PL_rsfp_filters)
1796 PL_rsfp_filters = newAV();
1798 datasv = NEWSV(255,0);
1799 if (!SvUPGRADE(datasv, SVt_PVIO))
1800 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1801 IoDIRP(datasv) = (DIR*)funcp; /* stash funcp into spare field */
1802 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
1803 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1804 funcp, SvPV_nolen(datasv)));
1805 av_unshift(PL_rsfp_filters, 1);
1806 av_store(PL_rsfp_filters, 0, datasv) ;
1811 /* Delete most recently added instance of this filter function. */
1813 Perl_filter_del(pTHX_ filter_t funcp)
1816 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp));
1817 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
1819 /* if filter is on top of stack (usual case) just pop it off */
1820 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
1821 if (IoDIRP(datasv) == (DIR*)funcp) {
1822 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
1823 IoDIRP(datasv) = (DIR*)NULL;
1824 sv_free(av_pop(PL_rsfp_filters));
1828 /* we need to search for the correct entry and clear it */
1829 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
1833 /* Invoke the n'th filter function for the current rsfp. */
1835 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
1838 /* 0 = read one text line */
1843 if (!PL_rsfp_filters)
1845 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
1846 /* Provide a default input filter to make life easy. */
1847 /* Note that we append to the line. This is handy. */
1848 DEBUG_P(PerlIO_printf(Perl_debug_log,
1849 "filter_read %d: from rsfp\n", idx));
1853 int old_len = SvCUR(buf_sv) ;
1855 /* ensure buf_sv is large enough */
1856 SvGROW(buf_sv, old_len + maxlen) ;
1857 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
1858 if (PerlIO_error(PL_rsfp))
1859 return -1; /* error */
1861 return 0 ; /* end of file */
1863 SvCUR_set(buf_sv, old_len + len) ;
1866 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
1867 if (PerlIO_error(PL_rsfp))
1868 return -1; /* error */
1870 return 0 ; /* end of file */
1873 return SvCUR(buf_sv);
1875 /* Skip this filter slot if filter has been deleted */
1876 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
1877 DEBUG_P(PerlIO_printf(Perl_debug_log,
1878 "filter_read %d: skipped (filter deleted)\n",
1880 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
1882 /* Get function pointer hidden within datasv */
1883 funcp = (filter_t)IoDIRP(datasv);
1884 DEBUG_P(PerlIO_printf(Perl_debug_log,
1885 "filter_read %d: via function %p (%s)\n",
1886 idx, funcp, SvPV_nolen(datasv)));
1887 /* Call function. The function is expected to */
1888 /* call "FILTER_READ(idx+1, buf_sv)" first. */
1889 /* Return: <0:error, =0:eof, >0:not eof */
1890 return (*funcp)(aTHXo_ idx, buf_sv, maxlen);
1894 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
1896 #ifdef PERL_CR_FILTER
1897 if (!PL_rsfp_filters) {
1898 filter_add(S_cr_textfilter,NULL);
1901 if (PL_rsfp_filters) {
1904 SvCUR_set(sv, 0); /* start with empty line */
1905 if (FILTER_READ(0, sv, 0) > 0)
1906 return ( SvPVX(sv) ) ;
1911 return (sv_gets(sv, fp, append));
1916 static char* exp_name[] =
1917 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
1918 "ATTRTERM", "TERMBLOCK"
1925 Works out what to call the token just pulled out of the input
1926 stream. The yacc parser takes care of taking the ops we return and
1927 stitching them into a tree.
1933 if read an identifier
1934 if we're in a my declaration
1935 croak if they tried to say my($foo::bar)
1936 build the ops for a my() declaration
1937 if it's an access to a my() variable
1938 are we in a sort block?
1939 croak if my($a); $a <=> $b
1940 build ops for access to a my() variable
1941 if in a dq string, and they've said @foo and we can't find @foo
1943 build ops for a bareword
1944 if we already built the token before, use it.
1948 #ifdef USE_PURE_BISON
1949 Perl_yylex(pTHX_ YYSTYPE *lvalp, int *lcharp)
1962 #ifdef USE_PURE_BISON
1963 yylval_pointer = lvalp;
1964 yychar_pointer = lcharp;
1967 /* check if there's an identifier for us to look at */
1968 if (PL_pending_ident) {
1969 /* pit holds the identifier we read and pending_ident is reset */
1970 char pit = PL_pending_ident;
1971 PL_pending_ident = 0;
1973 /* if we're in a my(), we can't allow dynamics here.
1974 $foo'bar has already been turned into $foo::bar, so
1975 just check for colons.
1977 if it's a legal name, the OP is a PADANY.
1980 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
1981 tmp = pad_allocmy(PL_tokenbuf);
1984 if (strchr(PL_tokenbuf,':'))
1985 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
1987 yylval.opval = newOP(OP_PADANY, 0);
1988 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
1994 build the ops for accesses to a my() variable.
1996 Deny my($a) or my($b) in a sort block, *if* $a or $b is
1997 then used in a comparison. This catches most, but not
1998 all cases. For instance, it catches
1999 sort { my($a); $a <=> $b }
2001 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
2002 (although why you'd do that is anyone's guess).
2005 if (!strchr(PL_tokenbuf,':')) {
2007 /* Check for single character per-thread SVs */
2008 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
2009 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
2010 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
2012 yylval.opval = newOP(OP_THREADSV, 0);
2013 yylval.opval->op_targ = tmp;
2016 #endif /* USE_THREADS */
2017 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
2018 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
2019 /* might be an "our" variable" */
2020 if (SvFLAGS(namesv) & SVpad_OUR) {
2021 /* build ops for a bareword */
2022 SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0);
2023 sv_catpvn(sym, "::", 2);
2024 sv_catpv(sym, PL_tokenbuf+1);
2025 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
2026 yylval.opval->op_private = OPpCONST_ENTERED;
2027 gv_fetchpv(SvPVX(sym),
2029 ? (GV_ADDMULTI | GV_ADDINEVAL)
2032 ((PL_tokenbuf[0] == '$') ? SVt_PV
2033 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2038 /* if it's a sort block and they're naming $a or $b */
2039 if (PL_last_lop_op == OP_SORT &&
2040 PL_tokenbuf[0] == '$' &&
2041 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
2044 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
2045 d < PL_bufend && *d != '\n';
2048 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
2049 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
2055 yylval.opval = newOP(OP_PADANY, 0);
2056 yylval.opval->op_targ = tmp;
2062 Whine if they've said @foo in a doublequoted string,
2063 and @foo isn't a variable we can find in the symbol
2066 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
2067 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
2068 if (!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
2069 yyerror(Perl_form(aTHX_ "In string, %s now must be written as \\%s",
2070 PL_tokenbuf, PL_tokenbuf));
2073 /* build ops for a bareword */
2074 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2075 yylval.opval->op_private = OPpCONST_ENTERED;
2076 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
2077 ((PL_tokenbuf[0] == '$') ? SVt_PV
2078 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2083 /* no identifier pending identification */
2085 switch (PL_lex_state) {
2087 case LEX_NORMAL: /* Some compilers will produce faster */
2088 case LEX_INTERPNORMAL: /* code if we comment these out. */
2092 /* when we've already built the next token, just pull it out of the queue */
2095 yylval = PL_nextval[PL_nexttoke];
2097 PL_lex_state = PL_lex_defer;
2098 PL_expect = PL_lex_expect;
2099 PL_lex_defer = LEX_NORMAL;
2101 return(PL_nexttype[PL_nexttoke]);
2103 /* interpolated case modifiers like \L \U, including \Q and \E.
2104 when we get here, PL_bufptr is at the \
2106 case LEX_INTERPCASEMOD:
2108 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2109 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2111 /* handle \E or end of string */
2112 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2116 if (PL_lex_casemods) {
2117 oldmod = PL_lex_casestack[--PL_lex_casemods];
2118 PL_lex_casestack[PL_lex_casemods] = '\0';
2120 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2122 PL_lex_state = LEX_INTERPCONCAT;
2126 if (PL_bufptr != PL_bufend)
2128 PL_lex_state = LEX_INTERPCONCAT;
2133 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2134 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2135 if (strchr("LU", *s) &&
2136 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2138 PL_lex_casestack[--PL_lex_casemods] = '\0';
2141 if (PL_lex_casemods > 10) {
2142 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2143 if (newlb != PL_lex_casestack) {
2145 PL_lex_casestack = newlb;
2148 PL_lex_casestack[PL_lex_casemods++] = *s;
2149 PL_lex_casestack[PL_lex_casemods] = '\0';
2150 PL_lex_state = LEX_INTERPCONCAT;
2151 PL_nextval[PL_nexttoke].ival = 0;
2154 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2156 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2158 PL_nextval[PL_nexttoke].ival = OP_LC;
2160 PL_nextval[PL_nexttoke].ival = OP_UC;
2162 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2164 Perl_croak(aTHX_ "panic: yylex");
2167 if (PL_lex_starts) {
2176 case LEX_INTERPPUSH:
2177 return sublex_push();
2179 case LEX_INTERPSTART:
2180 if (PL_bufptr == PL_bufend)
2181 return sublex_done();
2183 PL_lex_dojoin = (*PL_bufptr == '@');
2184 PL_lex_state = LEX_INTERPNORMAL;
2185 if (PL_lex_dojoin) {
2186 PL_nextval[PL_nexttoke].ival = 0;
2189 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2190 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2191 force_next(PRIVATEREF);
2193 force_ident("\"", '$');
2194 #endif /* USE_THREADS */
2195 PL_nextval[PL_nexttoke].ival = 0;
2197 PL_nextval[PL_nexttoke].ival = 0;
2199 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2202 if (PL_lex_starts++) {
2208 case LEX_INTERPENDMAYBE:
2209 if (intuit_more(PL_bufptr)) {
2210 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2216 if (PL_lex_dojoin) {
2217 PL_lex_dojoin = FALSE;
2218 PL_lex_state = LEX_INTERPCONCAT;
2221 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2222 && SvEVALED(PL_lex_repl))
2224 if (PL_bufptr != PL_bufend)
2225 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2226 PL_lex_repl = Nullsv;
2229 case LEX_INTERPCONCAT:
2231 if (PL_lex_brackets)
2232 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2234 if (PL_bufptr == PL_bufend)
2235 return sublex_done();
2237 if (SvIVX(PL_linestr) == '\'') {
2238 SV *sv = newSVsv(PL_linestr);
2241 else if ( PL_hints & HINT_NEW_RE )
2242 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2243 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2247 s = scan_const(PL_bufptr);
2249 PL_lex_state = LEX_INTERPCASEMOD;
2251 PL_lex_state = LEX_INTERPSTART;
2254 if (s != PL_bufptr) {
2255 PL_nextval[PL_nexttoke] = yylval;
2258 if (PL_lex_starts++)
2268 PL_lex_state = LEX_NORMAL;
2269 s = scan_formline(PL_bufptr);
2270 if (!PL_lex_formbrack)
2276 PL_oldoldbufptr = PL_oldbufptr;
2279 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2280 exp_name[PL_expect], s);
2286 if (isIDFIRST_lazy(s))
2288 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2291 goto fake_eof; /* emulate EOF on ^D or ^Z */
2296 if (PL_lex_brackets)
2297 yyerror("Missing right curly or square bracket");
2300 if (s++ < PL_bufend)
2301 goto retry; /* ignore stray nulls */
2304 if (!PL_in_eval && !PL_preambled) {
2305 PL_preambled = TRUE;
2306 sv_setpv(PL_linestr,incl_perldb());
2307 if (SvCUR(PL_linestr))
2308 sv_catpv(PL_linestr,";");
2310 while(AvFILLp(PL_preambleav) >= 0) {
2311 SV *tmpsv = av_shift(PL_preambleav);
2312 sv_catsv(PL_linestr, tmpsv);
2313 sv_catpv(PL_linestr, ";");
2316 sv_free((SV*)PL_preambleav);
2317 PL_preambleav = NULL;
2319 if (PL_minus_n || PL_minus_p) {
2320 sv_catpv(PL_linestr, "LINE: while (<>) {");
2322 sv_catpv(PL_linestr,"chomp;");
2324 GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
2326 GvIMPORTED_AV_on(gv);
2328 if (strchr("/'\"", *PL_splitstr)
2329 && strchr(PL_splitstr + 1, *PL_splitstr))
2330 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2333 s = "'~#\200\1'"; /* surely one char is unused...*/
2334 while (s[1] && strchr(PL_splitstr, *s)) s++;
2336 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c",
2337 "q" + (delim == '\''), delim);
2338 for (s = PL_splitstr; *s; s++) {
2340 sv_catpvn(PL_linestr, "\\", 1);
2341 sv_catpvn(PL_linestr, s, 1);
2343 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2347 sv_catpv(PL_linestr,"@F=split(' ');");
2350 sv_catpv(PL_linestr, "\n");
2351 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2352 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2353 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2354 SV *sv = NEWSV(85,0);
2356 sv_upgrade(sv, SVt_PVMG);
2357 sv_setsv(sv,PL_linestr);
2358 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2363 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2366 if (PL_preprocess && !PL_in_eval)
2367 (void)PerlProc_pclose(PL_rsfp);
2368 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2369 PerlIO_clearerr(PL_rsfp);
2371 (void)PerlIO_close(PL_rsfp);
2373 PL_doextract = FALSE;
2375 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2376 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2377 sv_catpv(PL_linestr,";}");
2378 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2379 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2380 PL_minus_n = PL_minus_p = 0;
2383 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2384 sv_setpv(PL_linestr,"");
2385 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2388 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2389 PL_doextract = FALSE;
2391 /* Incest with pod. */
2392 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2393 sv_setpv(PL_linestr, "");
2394 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2395 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2396 PL_doextract = FALSE;
2400 } while (PL_doextract);
2401 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2402 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2403 SV *sv = NEWSV(85,0);
2405 sv_upgrade(sv, SVt_PVMG);
2406 sv_setsv(sv,PL_linestr);
2407 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2409 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2410 if (CopLINE(PL_curcop) == 1) {
2411 while (s < PL_bufend && isSPACE(*s))
2413 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2417 if (*s == '#' && *(s+1) == '!')
2419 #ifdef ALTERNATE_SHEBANG
2421 static char as[] = ALTERNATE_SHEBANG;
2422 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2423 d = s + (sizeof(as) - 1);
2425 #endif /* ALTERNATE_SHEBANG */
2434 while (*d && !isSPACE(*d))
2438 #ifdef ARG_ZERO_IS_SCRIPT
2439 if (ipathend > ipath) {
2441 * HP-UX (at least) sets argv[0] to the script name,
2442 * which makes $^X incorrect. And Digital UNIX and Linux,
2443 * at least, set argv[0] to the basename of the Perl
2444 * interpreter. So, having found "#!", we'll set it right.
2446 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2447 assert(SvPOK(x) || SvGMAGICAL(x));
2448 if (sv_eq(x, CopFILESV(PL_curcop))) {
2449 sv_setpvn(x, ipath, ipathend - ipath);
2452 TAINT_NOT; /* $^X is always tainted, but that's OK */
2454 #endif /* ARG_ZERO_IS_SCRIPT */
2459 d = instr(s,"perl -");
2461 d = instr(s,"perl");
2463 /* avoid getting into infinite loops when shebang
2464 * line contains "Perl" rather than "perl" */
2466 for (d = ipathend-4; d >= ipath; --d) {
2467 if ((*d == 'p' || *d == 'P')
2468 && !ibcmp(d, "perl", 4))
2478 #ifdef ALTERNATE_SHEBANG
2480 * If the ALTERNATE_SHEBANG on this system starts with a
2481 * character that can be part of a Perl expression, then if
2482 * we see it but not "perl", we're probably looking at the
2483 * start of Perl code, not a request to hand off to some
2484 * other interpreter. Similarly, if "perl" is there, but
2485 * not in the first 'word' of the line, we assume the line
2486 * contains the start of the Perl program.
2488 if (d && *s != '#') {
2490 while (*c && !strchr("; \t\r\n\f\v#", *c))
2493 d = Nullch; /* "perl" not in first word; ignore */
2495 *s = '#'; /* Don't try to parse shebang line */
2497 #endif /* ALTERNATE_SHEBANG */
2502 !instr(s,"indir") &&
2503 instr(PL_origargv[0],"perl"))
2509 while (s < PL_bufend && isSPACE(*s))
2511 if (s < PL_bufend) {
2512 Newz(899,newargv,PL_origargc+3,char*);
2514 while (s < PL_bufend && !isSPACE(*s))
2517 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2520 newargv = PL_origargv;
2522 PerlProc_execv(ipath, newargv);
2523 Perl_croak(aTHX_ "Can't exec %s", ipath);
2526 U32 oldpdb = PL_perldb;
2527 bool oldn = PL_minus_n;
2528 bool oldp = PL_minus_p;
2530 while (*d && !isSPACE(*d)) d++;
2531 while (*d == ' ' || *d == '\t') d++;
2535 if (*d == 'M' || *d == 'm') {
2537 while (*d && !isSPACE(*d)) d++;
2538 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2541 d = moreswitches(d);
2543 if (PERLDB_LINE && !oldpdb ||
2544 ( PL_minus_n || PL_minus_p ) && !(oldn || oldp) )
2545 /* if we have already added "LINE: while (<>) {",
2546 we must not do it again */
2548 sv_setpv(PL_linestr, "");
2549 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2550 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2551 PL_preambled = FALSE;
2553 (void)gv_fetchfile(PL_origfilename);
2560 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2562 PL_lex_state = LEX_FORMLINE;
2567 #ifdef PERL_STRICT_CR
2568 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2570 "(Maybe you didn't strip carriage returns after a network transfer?)\n");
2572 case ' ': case '\t': case '\f': case 013:
2577 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2579 while (s < d && *s != '\n')
2584 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2586 PL_lex_state = LEX_FORMLINE;
2596 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2601 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
2604 if (strnEQ(s,"=>",2)) {
2605 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2606 OPERATOR('-'); /* unary minus */
2608 PL_last_uni = PL_oldbufptr;
2609 PL_last_lop_op = OP_FTEREAD; /* good enough */
2611 case 'r': FTST(OP_FTEREAD);
2612 case 'w': FTST(OP_FTEWRITE);
2613 case 'x': FTST(OP_FTEEXEC);
2614 case 'o': FTST(OP_FTEOWNED);
2615 case 'R': FTST(OP_FTRREAD);
2616 case 'W': FTST(OP_FTRWRITE);
2617 case 'X': FTST(OP_FTREXEC);
2618 case 'O': FTST(OP_FTROWNED);
2619 case 'e': FTST(OP_FTIS);
2620 case 'z': FTST(OP_FTZERO);
2621 case 's': FTST(OP_FTSIZE);
2622 case 'f': FTST(OP_FTFILE);
2623 case 'd': FTST(OP_FTDIR);
2624 case 'l': FTST(OP_FTLINK);
2625 case 'p': FTST(OP_FTPIPE);
2626 case 'S': FTST(OP_FTSOCK);
2627 case 'u': FTST(OP_FTSUID);
2628 case 'g': FTST(OP_FTSGID);
2629 case 'k': FTST(OP_FTSVTX);
2630 case 'b': FTST(OP_FTBLK);
2631 case 'c': FTST(OP_FTCHR);
2632 case 't': FTST(OP_FTTTY);
2633 case 'T': FTST(OP_FTTEXT);
2634 case 'B': FTST(OP_FTBINARY);
2635 case 'M': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTMTIME);
2636 case 'A': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTATIME);
2637 case 'C': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTCTIME);
2639 Perl_croak(aTHX_ "Unrecognized file test: -%c", (int)tmp);
2646 if (PL_expect == XOPERATOR)
2651 else if (*s == '>') {
2654 if (isIDFIRST_lazy(s)) {
2655 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2663 if (PL_expect == XOPERATOR)
2666 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2668 OPERATOR('-'); /* unary minus */
2675 if (PL_expect == XOPERATOR)
2680 if (PL_expect == XOPERATOR)
2683 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2689 if (PL_expect != XOPERATOR) {
2690 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2691 PL_expect = XOPERATOR;
2692 force_ident(PL_tokenbuf, '*');
2705 if (PL_expect == XOPERATOR) {
2709 PL_tokenbuf[0] = '%';
2710 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2711 if (!PL_tokenbuf[1]) {
2713 yyerror("Final % should be \\% or %name");
2716 PL_pending_ident = '%';
2735 switch (PL_expect) {
2738 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2740 PL_bufptr = s; /* update in case we back off */
2746 PL_expect = XTERMBLOCK;
2750 while (isIDFIRST_lazy(s)) {
2751 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2753 d = scan_str(d,TRUE,TRUE);
2756 SvREFCNT_dec(PL_lex_stuff);
2757 PL_lex_stuff = Nullsv;
2759 /* MUST advance bufptr here to avoid bogus
2760 "at end of line" context messages from yyerror().
2762 PL_bufptr = s + len;
2763 yyerror("Unterminated attribute parameter in attribute list");
2766 return 0; /* EOF indicator */
2770 SV *sv = newSVpvn(s, len);
2771 sv_catsv(sv, PL_lex_stuff);
2772 attrs = append_elem(OP_LIST, attrs,
2773 newSVOP(OP_CONST, 0, sv));
2774 SvREFCNT_dec(PL_lex_stuff);
2775 PL_lex_stuff = Nullsv;
2778 attrs = append_elem(OP_LIST, attrs,
2779 newSVOP(OP_CONST, 0,
2786 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}' for vi */
2787 if (*s != ';' && *s != tmp) {
2788 char q = ((*s == '\'') ? '"' : '\'');
2789 /* If here for an expression, and parsed no attrs, back off. */
2790 if (tmp == '=' && !attrs) {
2794 /* MUST advance bufptr here to avoid bogus "at end of line"
2795 context messages from yyerror().
2799 yyerror("Unterminated attribute list");
2801 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
2808 PL_nextval[PL_nexttoke].opval = attrs;
2816 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
2817 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
2822 if (CopLINE(PL_curcop) < PL_copline)
2823 PL_copline = CopLINE(PL_curcop);
2834 if (PL_lex_brackets <= 0)
2835 yyerror("Unmatched right square bracket");
2838 if (PL_lex_state == LEX_INTERPNORMAL) {
2839 if (PL_lex_brackets == 0) {
2840 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
2841 PL_lex_state = LEX_INTERPEND;
2848 if (PL_lex_brackets > 100) {
2849 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
2850 if (newlb != PL_lex_brackstack) {
2852 PL_lex_brackstack = newlb;
2855 switch (PL_expect) {
2857 if (PL_lex_formbrack) {
2861 if (PL_oldoldbufptr == PL_last_lop)
2862 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
2864 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2865 OPERATOR(HASHBRACK);
2867 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
2870 PL_tokenbuf[0] = '\0';
2871 if (d < PL_bufend && *d == '-') {
2872 PL_tokenbuf[0] = '-';
2874 while (d < PL_bufend && (*d == ' ' || *d == '\t'))
2877 if (d < PL_bufend && isIDFIRST_lazy(d)) {
2878 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
2880 while (d < PL_bufend && (*d == ' ' || *d == '\t'))
2883 char minus = (PL_tokenbuf[0] == '-');
2884 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
2892 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
2897 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2902 if (PL_oldoldbufptr == PL_last_lop)
2903 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
2905 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
2908 OPERATOR(HASHBRACK);
2909 /* This hack serves to disambiguate a pair of curlies
2910 * as being a block or an anon hash. Normally, expectation
2911 * determines that, but in cases where we're not in a
2912 * position to expect anything in particular (like inside
2913 * eval"") we have to resolve the ambiguity. This code
2914 * covers the case where the first term in the curlies is a
2915 * quoted string. Most other cases need to be explicitly
2916 * disambiguated by prepending a `+' before the opening
2917 * curly in order to force resolution as an anon hash.
2919 * XXX should probably propagate the outer expectation
2920 * into eval"" to rely less on this hack, but that could
2921 * potentially break current behavior of eval"".
2925 if (*s == '\'' || *s == '"' || *s == '`') {
2926 /* common case: get past first string, handling escapes */
2927 for (t++; t < PL_bufend && *t != *s;)
2928 if (*t++ == '\\' && (*t == '\\' || *t == *s))
2932 else if (*s == 'q') {
2935 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
2939 char open, close, term;
2942 while (t < PL_bufend && isSPACE(*t))
2946 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
2950 for (t++; t < PL_bufend; t++) {
2951 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
2953 else if (*t == open)
2957 for (t++; t < PL_bufend; t++) {
2958 if (*t == '\\' && t+1 < PL_bufend)
2960 else if (*t == close && --brackets <= 0)
2962 else if (*t == open)
2968 else if (isALNUM_lazy(t)) {
2970 while (t < PL_bufend && isALNUM_lazy(t))
2973 while (t < PL_bufend && isSPACE(*t))
2975 /* if comma follows first term, call it an anon hash */
2976 /* XXX it could be a comma expression with loop modifiers */
2977 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
2978 || (*t == '=' && t[1] == '>')))
2979 OPERATOR(HASHBRACK);
2980 if (PL_expect == XREF)
2983 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
2989 yylval.ival = CopLINE(PL_curcop);
2990 if (isSPACE(*s) || *s == '#')
2991 PL_copline = NOLINE; /* invalidate current command line number */
2996 if (PL_lex_brackets <= 0)
2997 yyerror("Unmatched right curly bracket");
2999 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3000 if (PL_lex_brackets < PL_lex_formbrack)
3001 PL_lex_formbrack = 0;
3002 if (PL_lex_state == LEX_INTERPNORMAL) {
3003 if (PL_lex_brackets == 0) {
3004 if (PL_expect & XFAKEBRACK) {
3005 PL_expect &= XENUMMASK;
3006 PL_lex_state = LEX_INTERPEND;
3008 return yylex(); /* ignore fake brackets */
3010 if (*s == '-' && s[1] == '>')
3011 PL_lex_state = LEX_INTERPENDMAYBE;
3012 else if (*s != '[' && *s != '{')
3013 PL_lex_state = LEX_INTERPEND;
3016 if (PL_expect & XFAKEBRACK) {
3017 PL_expect &= XENUMMASK;
3019 return yylex(); /* ignore fake brackets */
3029 if (PL_expect == XOPERATOR) {
3030 if (ckWARN(WARN_SEMICOLON) && isIDFIRST_lazy(s) && PL_bufptr == PL_linestart) {
3031 CopLINE_dec(PL_curcop);
3032 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3033 CopLINE_inc(PL_curcop);
3038 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3040 PL_expect = XOPERATOR;
3041 force_ident(PL_tokenbuf, '&');
3045 yylval.ival = (OPpENTERSUB_AMPER<<8);
3064 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3065 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3067 if (PL_expect == XSTATE && isALPHA(tmp) &&
3068 (s == PL_linestart+1 || s[-2] == '\n') )
3070 if (PL_in_eval && !PL_rsfp) {
3075 if (strnEQ(s,"=cut",4)) {
3089 PL_doextract = TRUE;
3092 if (PL_lex_brackets < PL_lex_formbrack) {
3094 #ifdef PERL_STRICT_CR
3095 for (t = s; *t == ' ' || *t == '\t'; t++) ;
3097 for (t = s; *t == ' ' || *t == '\t' || *t == '\r'; t++) ;
3099 if (*t == '\n' || *t == '#') {
3117 if (PL_expect != XOPERATOR) {
3118 if (s[1] != '<' && !strchr(s,'>'))
3121 s = scan_heredoc(s);
3123 s = scan_inputsymbol(s);
3124 TERM(sublex_start());
3129 SHop(OP_LEFT_SHIFT);
3143 SHop(OP_RIGHT_SHIFT);
3152 if (PL_expect == XOPERATOR) {
3153 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3156 return ','; /* grandfather non-comma-format format */
3160 if (s[1] == '#' && (isIDFIRST_lazy(s+2) || strchr("{$:+-", s[2]))) {
3161 PL_tokenbuf[0] = '@';
3162 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3163 sizeof PL_tokenbuf - 1, FALSE);
3164 if (PL_expect == XOPERATOR)
3165 no_op("Array length", s);
3166 if (!PL_tokenbuf[1])
3168 PL_expect = XOPERATOR;
3169 PL_pending_ident = '#';
3173 PL_tokenbuf[0] = '$';
3174 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3175 sizeof PL_tokenbuf - 1, FALSE);
3176 if (PL_expect == XOPERATOR)
3178 if (!PL_tokenbuf[1]) {
3180 yyerror("Final $ should be \\$ or $name");
3184 /* This kludge not intended to be bulletproof. */
3185 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3186 yylval.opval = newSVOP(OP_CONST, 0,
3187 newSViv((IV)PL_compiling.cop_arybase));
3188 yylval.opval->op_private = OPpCONST_ARYBASE;
3194 if (PL_lex_state == LEX_NORMAL)
3197 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3200 PL_tokenbuf[0] = '@';
3201 if (ckWARN(WARN_SYNTAX)) {
3203 isSPACE(*t) || isALNUM_lazy(t) || *t == '$';
3206 PL_bufptr = skipspace(PL_bufptr);
3207 while (t < PL_bufend && *t != ']')
3209 Perl_warner(aTHX_ WARN_SYNTAX,
3210 "Multidimensional syntax %.*s not supported",
3211 (t - PL_bufptr) + 1, PL_bufptr);
3215 else if (*s == '{') {
3216 PL_tokenbuf[0] = '%';
3217 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3218 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3220 char tmpbuf[sizeof PL_tokenbuf];
3222 for (t++; isSPACE(*t); t++) ;
3223 if (isIDFIRST_lazy(t)) {
3224 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3225 for (; isSPACE(*t); t++) ;
3226 if (*t == ';' && get_cv(tmpbuf, FALSE))
3227 Perl_warner(aTHX_ WARN_SYNTAX,
3228 "You need to quote \"%s\"", tmpbuf);
3234 PL_expect = XOPERATOR;
3235 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3236 bool islop = (PL_last_lop == PL_oldoldbufptr);
3237 if (!islop || PL_last_lop_op == OP_GREPSTART)
3238 PL_expect = XOPERATOR;
3239 else if (strchr("$@\"'`q", *s))
3240 PL_expect = XTERM; /* e.g. print $fh "foo" */
3241 else if (strchr("&*<%", *s) && isIDFIRST_lazy(s+1))
3242 PL_expect = XTERM; /* e.g. print $fh &sub */
3243 else if (isIDFIRST_lazy(s)) {
3244 char tmpbuf[sizeof PL_tokenbuf];
3245 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3246 if (tmp = keyword(tmpbuf, len)) {
3247 /* binary operators exclude handle interpretations */
3259 PL_expect = XTERM; /* e.g. print $fh length() */
3264 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3265 if (gv && GvCVu(gv))
3266 PL_expect = XTERM; /* e.g. print $fh subr() */
3269 else if (isDIGIT(*s))
3270 PL_expect = XTERM; /* e.g. print $fh 3 */
3271 else if (*s == '.' && isDIGIT(s[1]))
3272 PL_expect = XTERM; /* e.g. print $fh .3 */
3273 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3274 PL_expect = XTERM; /* e.g. print $fh -1 */
3275 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3276 PL_expect = XTERM; /* print $fh <<"EOF" */
3278 PL_pending_ident = '$';
3282 if (PL_expect == XOPERATOR)
3284 PL_tokenbuf[0] = '@';
3285 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3286 if (!PL_tokenbuf[1]) {
3288 yyerror("Final @ should be \\@ or @name");
3291 if (PL_lex_state == LEX_NORMAL)
3293 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3295 PL_tokenbuf[0] = '%';
3297 /* Warn about @ where they meant $. */
3298 if (ckWARN(WARN_SYNTAX)) {
3299 if (*s == '[' || *s == '{') {
3301 while (*t && (isALNUM_lazy(t) || strchr(" \t$#+-'\"", *t)))
3303 if (*t == '}' || *t == ']') {
3305 PL_bufptr = skipspace(PL_bufptr);
3306 Perl_warner(aTHX_ WARN_SYNTAX,
3307 "Scalar value %.*s better written as $%.*s",
3308 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3313 PL_pending_ident = '@';
3316 case '/': /* may either be division or pattern */
3317 case '?': /* may either be conditional or pattern */
3318 if (PL_expect != XOPERATOR) {
3319 /* Disable warning on "study /blah/" */
3320 if (PL_oldoldbufptr == PL_last_uni
3321 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3322 || memNE(PL_last_uni, "study", 5) || isALNUM_lazy(PL_last_uni+5)))
3324 s = scan_pat(s,OP_MATCH);
3325 TERM(sublex_start());
3333 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3334 #ifdef PERL_STRICT_CR
3337 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3339 && (s == PL_linestart || s[-1] == '\n') )
3341 PL_lex_formbrack = 0;
3345 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3351 yylval.ival = OPf_SPECIAL;
3357 if (PL_expect != XOPERATOR)
3362 case '0': case '1': case '2': case '3': case '4':
3363 case '5': case '6': case '7': case '8': case '9':
3365 if (PL_expect == XOPERATOR)
3370 s = scan_str(s,FALSE,FALSE);
3371 if (PL_expect == XOPERATOR) {
3372 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3375 return ','; /* grandfather non-comma-format format */
3381 missingterm((char*)0);
3382 yylval.ival = OP_CONST;
3383 TERM(sublex_start());
3386 s = scan_str(s,FALSE,FALSE);
3387 if (PL_expect == XOPERATOR) {
3388 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3391 return ','; /* grandfather non-comma-format format */
3397 missingterm((char*)0);
3398 yylval.ival = OP_CONST;
3399 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3400 if (*d == '$' || *d == '@' || *d == '\\' || *d & 0x80) {
3401 yylval.ival = OP_STRINGIFY;
3405 TERM(sublex_start());
3408 s = scan_str(s,FALSE,FALSE);
3409 if (PL_expect == XOPERATOR)
3410 no_op("Backticks",s);
3412 missingterm((char*)0);
3413 yylval.ival = OP_BACKTICK;
3415 TERM(sublex_start());
3419 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3420 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3422 if (PL_expect == XOPERATOR)
3423 no_op("Backslash",s);
3427 if (isDIGIT(s[1]) && PL_expect == XTERM) {
3431 while (isDIGIT(*start))
3433 if (*start == '.' && isDIGIT(start[1])) {
3440 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3480 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3482 /* Some keywords can be followed by any delimiter, including ':' */
3483 tmp = (len == 1 && strchr("msyq", PL_tokenbuf[0]) ||
3484 len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3485 (PL_tokenbuf[0] == 'q' &&
3486 strchr("qwxr", PL_tokenbuf[1]))));
3488 /* x::* is just a word, unless x is "CORE" */
3489 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3493 while (d < PL_bufend && isSPACE(*d))
3494 d++; /* no comments skipped here, or s### is misparsed */
3496 /* Is this a label? */
3497 if (!tmp && PL_expect == XSTATE
3498 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3500 yylval.pval = savepv(PL_tokenbuf);
3505 /* Check for keywords */
3506 tmp = keyword(PL_tokenbuf, len);
3508 /* Is this a word before a => operator? */
3509 if (strnEQ(d,"=>",2)) {
3511 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3512 yylval.opval->op_private = OPpCONST_BARE;
3516 if (tmp < 0) { /* second-class keyword? */
3517 GV *ogv = Nullgv; /* override (winner) */
3518 GV *hgv = Nullgv; /* hidden (loser) */
3519 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3521 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3524 if (GvIMPORTED_CV(gv))
3526 else if (! CvMETHOD(cv))
3530 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3531 (gv = *gvp) != (GV*)&PL_sv_undef &&
3532 GvCVu(gv) && GvIMPORTED_CV(gv))
3538 tmp = 0; /* overridden by import or by GLOBAL */
3541 && -tmp==KEY_lock /* XXX generalizable kludge */
3543 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3545 tmp = 0; /* any sub overrides "weak" keyword */
3547 else { /* no override */
3551 if (ckWARN(WARN_AMBIGUOUS) && hgv
3552 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3553 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3554 "Ambiguous call resolved as CORE::%s(), %s",
3555 GvENAME(hgv), "qualify as such or use &");
3562 default: /* not a keyword */
3565 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3567 /* Get the rest if it looks like a package qualifier */
3569 if (*s == '\'' || *s == ':' && s[1] == ':') {
3571 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3574 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3575 *s == '\'' ? "'" : "::");
3579 if (PL_expect == XOPERATOR) {
3580 if (PL_bufptr == PL_linestart) {
3581 CopLINE_dec(PL_curcop);
3582 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3583 CopLINE_inc(PL_curcop);
3586 no_op("Bareword",s);
3589 /* Look for a subroutine with this name in current package,
3590 unless name is "Foo::", in which case Foo is a bearword
3591 (and a package name). */
3594 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3596 if (ckWARN(WARN_UNSAFE) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3597 Perl_warner(aTHX_ WARN_UNSAFE,
3598 "Bareword \"%s\" refers to nonexistent package",
3601 PL_tokenbuf[len] = '\0';
3608 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3611 /* if we saw a global override before, get the right name */
3614 sv = newSVpvn("CORE::GLOBAL::",14);
3615 sv_catpv(sv,PL_tokenbuf);
3618 sv = newSVpv(PL_tokenbuf,0);
3620 /* Presume this is going to be a bareword of some sort. */
3623 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3624 yylval.opval->op_private = OPpCONST_BARE;
3626 /* And if "Foo::", then that's what it certainly is. */
3631 /* See if it's the indirect object for a list operator. */
3633 if (PL_oldoldbufptr &&
3634 PL_oldoldbufptr < PL_bufptr &&
3635 (PL_oldoldbufptr == PL_last_lop
3636 || PL_oldoldbufptr == PL_last_uni) &&
3637 /* NO SKIPSPACE BEFORE HERE! */
3638 (PL_expect == XREF ||
3639 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3641 bool immediate_paren = *s == '(';
3643 /* (Now we can afford to cross potential line boundary.) */
3646 /* Two barewords in a row may indicate method call. */
3648 if ((isIDFIRST_lazy(s) || *s == '$') && (tmp=intuit_method(s,gv)))
3651 /* If not a declared subroutine, it's an indirect object. */
3652 /* (But it's an indir obj regardless for sort.) */
3654 if ((PL_last_lop_op == OP_SORT ||
3655 (!immediate_paren && (!gv || !GvCVu(gv)))) &&
3656 (PL_last_lop_op != OP_MAPSTART &&
3657 PL_last_lop_op != OP_GREPSTART))
3659 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3664 /* If followed by a paren, it's certainly a subroutine. */
3666 PL_expect = XOPERATOR;
3670 if (gv && GvCVu(gv)) {
3671 for (d = s + 1; *d == ' ' || *d == '\t'; d++) ;
3672 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3677 PL_nextval[PL_nexttoke].opval = yylval.opval;
3678 PL_expect = XOPERATOR;
3684 /* If followed by var or block, call it a method (unless sub) */
3686 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3687 PL_last_lop = PL_oldbufptr;
3688 PL_last_lop_op = OP_METHOD;
3692 /* If followed by a bareword, see if it looks like indir obj. */
3694 if ((isIDFIRST_lazy(s) || *s == '$') && (tmp = intuit_method(s,gv)))
3697 /* Not a method, so call it a subroutine (if defined) */
3699 if (gv && GvCVu(gv)) {
3701 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
3702 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3703 "Ambiguous use of -%s resolved as -&%s()",
3704 PL_tokenbuf, PL_tokenbuf);
3705 /* Check for a constant sub */
3707 if ((sv = cv_const_sv(cv))) {
3709 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
3710 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
3711 yylval.opval->op_private = 0;
3715 /* Resolve to GV now. */
3716 op_free(yylval.opval);
3717 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
3718 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
3719 PL_last_lop = PL_oldbufptr;
3720 PL_last_lop_op = OP_ENTERSUB;
3721 /* Is there a prototype? */
3724 char *proto = SvPV((SV*)cv, len);
3727 if (strEQ(proto, "$"))
3729 if (*proto == '&' && *s == '{') {
3730 sv_setpv(PL_subname,"__ANON__");
3734 PL_nextval[PL_nexttoke].opval = yylval.opval;
3740 /* Call it a bare word */
3742 if (PL_hints & HINT_STRICT_SUBS)
3743 yylval.opval->op_private |= OPpCONST_STRICT;
3746 if (ckWARN(WARN_RESERVED)) {
3747 if (lastchar != '-') {
3748 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
3750 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
3757 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
3758 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3759 "Operator or semicolon missing before %c%s",
3760 lastchar, PL_tokenbuf);
3761 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3762 "Ambiguous use of %c resolved as operator %c",
3763 lastchar, lastchar);
3769 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3770 newSVpv(CopFILE(PL_curcop),0));
3774 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3775 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
3778 case KEY___PACKAGE__:
3779 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
3781 ? newSVsv(PL_curstname)
3790 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
3791 char *pname = "main";
3792 if (PL_tokenbuf[2] == 'D')
3793 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
3794 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
3797 GvIOp(gv) = newIO();
3798 IoIFP(GvIOp(gv)) = PL_rsfp;
3799 #if defined(HAS_FCNTL) && defined(F_SETFD)
3801 int fd = PerlIO_fileno(PL_rsfp);
3802 fcntl(fd,F_SETFD,fd >= 3);
3805 /* Mark this internal pseudo-handle as clean */
3806 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
3808 IoTYPE(GvIOp(gv)) = '|';
3809 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
3810 IoTYPE(GvIOp(gv)) = '-';
3812 IoTYPE(GvIOp(gv)) = '<';
3813 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
3814 /* if the script was opened in binmode, we need to revert
3815 * it to text mode for compatibility; but only iff it has CRs
3816 * XXX this is a questionable hack at best. */
3817 if (PL_bufend-PL_bufptr > 2
3818 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
3821 if (IoTYPE(GvIOp(gv)) == '<') {
3822 loc = PerlIO_tell(PL_rsfp);
3823 (void)PerlIO_seek(PL_rsfp, 0L, 0);
3825 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
3826 #if defined(__BORLANDC__)
3827 /* XXX see note in do_binmode() */
3828 ((FILE*)PL_rsfp)->flags |= _F_BIN;
3831 PerlIO_seek(PL_rsfp, loc, 0);
3846 if (PL_expect == XSTATE) {
3853 if (*s == ':' && s[1] == ':') {
3856 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3857 tmp = keyword(PL_tokenbuf, len);
3871 LOP(OP_ACCEPT,XTERM);
3877 LOP(OP_ATAN2,XTERM);
3886 LOP(OP_BLESS,XTERM);
3895 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
3912 if (!PL_cryptseen) {
3913 PL_cryptseen = TRUE;
3917 LOP(OP_CRYPT,XTERM);
3920 if (ckWARN(WARN_OCTAL)) {
3921 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
3922 if (*d != '0' && isDIGIT(*d))
3923 Perl_warner(aTHX_ WARN_OCTAL,
3924 "chmod: mode argument is missing initial 0");
3926 LOP(OP_CHMOD,XTERM);
3929 LOP(OP_CHOWN,XTERM);
3932 LOP(OP_CONNECT,XTERM);
3948 s = force_word(s,WORD,FALSE,TRUE,FALSE);
3952 PL_hints |= HINT_BLOCK_SCOPE;
3962 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
3963 LOP(OP_DBMOPEN,XTERM);
3969 s = force_word(s,WORD,TRUE,FALSE,FALSE);
3976 yylval.ival = CopLINE(PL_curcop);
3990 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
3991 UNIBRACK(OP_ENTEREVAL);
4006 case KEY_endhostent:
4012 case KEY_endservent:
4015 case KEY_endprotoent:
4026 yylval.ival = CopLINE(PL_curcop);
4028 if (PL_expect == XSTATE && isIDFIRST_lazy(s)) {
4030 if ((PL_bufend - p) >= 3 &&
4031 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4033 else if ((PL_bufend - p) >= 4 &&
4034 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4037 if (isIDFIRST_lazy(p)) {
4038 p = scan_ident(p, PL_bufend,
4039 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4043 Perl_croak(aTHX_ "Missing $ on loop variable");
4048 LOP(OP_FORMLINE,XTERM);
4054 LOP(OP_FCNTL,XTERM);
4060 LOP(OP_FLOCK,XTERM);
4069 LOP(OP_GREPSTART, *s == '(' ? XTERM : XREF);
4072 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4087 case KEY_getpriority:
4088 LOP(OP_GETPRIORITY,XTERM);
4090 case KEY_getprotobyname:
4093 case KEY_getprotobynumber:
4094 LOP(OP_GPBYNUMBER,XTERM);
4096 case KEY_getprotoent:
4108 case KEY_getpeername:
4109 UNI(OP_GETPEERNAME);
4111 case KEY_gethostbyname:
4114 case KEY_gethostbyaddr:
4115 LOP(OP_GHBYADDR,XTERM);
4117 case KEY_gethostent:
4120 case KEY_getnetbyname:
4123 case KEY_getnetbyaddr:
4124 LOP(OP_GNBYADDR,XTERM);
4129 case KEY_getservbyname:
4130 LOP(OP_GSBYNAME,XTERM);
4132 case KEY_getservbyport:
4133 LOP(OP_GSBYPORT,XTERM);
4135 case KEY_getservent:
4138 case KEY_getsockname:
4139 UNI(OP_GETSOCKNAME);
4141 case KEY_getsockopt:
4142 LOP(OP_GSOCKOPT,XTERM);
4164 yylval.ival = CopLINE(PL_curcop);
4168 LOP(OP_INDEX,XTERM);
4174 LOP(OP_IOCTL,XTERM);
4186 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4218 LOP(OP_LISTEN,XTERM);
4227 s = scan_pat(s,OP_MATCH);
4228 TERM(sublex_start());
4231 LOP(OP_MAPSTART, *s == '(' ? XTERM : XREF);
4234 LOP(OP_MKDIR,XTERM);
4237 LOP(OP_MSGCTL,XTERM);
4240 LOP(OP_MSGGET,XTERM);
4243 LOP(OP_MSGRCV,XTERM);
4246 LOP(OP_MSGSND,XTERM);
4252 if (isIDFIRST_lazy(s)) {
4253 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4254 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4256 PL_in_my_stash = gv_stashpv(PL_tokenbuf, FALSE);
4257 if (!PL_in_my_stash) {
4260 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4268 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4275 if (PL_expect != XSTATE)
4276 yyerror("\"no\" not allowed in expression");
4277 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4278 s = force_version(s);
4283 if (*s == '(' || (s = skipspace(s), *s == '('))
4290 if (isIDFIRST_lazy(s)) {
4292 for (d = s; isALNUM_lazy(d); d++) ;
4294 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_AMBIGUOUS))
4295 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4296 "Precedence problem: open %.*s should be open(%.*s)",
4302 yylval.ival = OP_OR;
4312 LOP(OP_OPEN_DIR,XTERM);
4315 checkcomma(s,PL_tokenbuf,"filehandle");
4319 checkcomma(s,PL_tokenbuf,"filehandle");
4338 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4342 LOP(OP_PIPE_OP,XTERM);
4345 s = scan_str(s,FALSE,FALSE);
4347 missingterm((char*)0);
4348 yylval.ival = OP_CONST;
4349 TERM(sublex_start());
4355 s = scan_str(s,FALSE,FALSE);
4357 missingterm((char*)0);
4359 if (SvCUR(PL_lex_stuff)) {
4362 d = SvPV_force(PL_lex_stuff, len);
4364 for (; isSPACE(*d) && len; --len, ++d) ;
4367 if (!warned && ckWARN(WARN_SYNTAX)) {
4368 for (; !isSPACE(*d) && len; --len, ++d) {
4370 Perl_warner(aTHX_ WARN_SYNTAX,
4371 "Possible attempt to separate words with commas");
4374 else if (*d == '#') {
4375 Perl_warner(aTHX_ WARN_SYNTAX,
4376 "Possible attempt to put comments in qw() list");
4382 for (; !isSPACE(*d) && len; --len, ++d) ;
4384 words = append_elem(OP_LIST, words,
4385 newSVOP(OP_CONST, 0, newSVpvn(b, d-b)));
4389 PL_nextval[PL_nexttoke].opval = words;
4394 SvREFCNT_dec(PL_lex_stuff);
4395 PL_lex_stuff = Nullsv;
4400 s = scan_str(s,FALSE,FALSE);
4402 missingterm((char*)0);
4403 yylval.ival = OP_STRINGIFY;
4404 if (SvIVX(PL_lex_stuff) == '\'')
4405 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4406 TERM(sublex_start());
4409 s = scan_pat(s,OP_QR);
4410 TERM(sublex_start());
4413 s = scan_str(s,FALSE,FALSE);
4415 missingterm((char*)0);
4416 yylval.ival = OP_BACKTICK;
4418 TERM(sublex_start());
4425 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4426 s = force_version(s);
4429 *PL_tokenbuf = '\0';
4430 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4431 if (isIDFIRST_lazy(PL_tokenbuf))
4432 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4434 yyerror("<> should be quotes");
4442 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4446 LOP(OP_RENAME,XTERM);
4455 LOP(OP_RINDEX,XTERM);
4478 LOP(OP_REVERSE,XTERM);
4489 TERM(sublex_start());
4491 TOKEN(1); /* force error */
4500 LOP(OP_SELECT,XTERM);
4506 LOP(OP_SEMCTL,XTERM);
4509 LOP(OP_SEMGET,XTERM);
4512 LOP(OP_SEMOP,XTERM);
4518 LOP(OP_SETPGRP,XTERM);
4520 case KEY_setpriority:
4521 LOP(OP_SETPRIORITY,XTERM);
4523 case KEY_sethostent:
4529 case KEY_setservent:
4532 case KEY_setprotoent:
4542 LOP(OP_SEEKDIR,XTERM);
4544 case KEY_setsockopt:
4545 LOP(OP_SSOCKOPT,XTERM);
4551 LOP(OP_SHMCTL,XTERM);
4554 LOP(OP_SHMGET,XTERM);
4557 LOP(OP_SHMREAD,XTERM);
4560 LOP(OP_SHMWRITE,XTERM);
4563 LOP(OP_SHUTDOWN,XTERM);
4572 LOP(OP_SOCKET,XTERM);
4574 case KEY_socketpair:
4575 LOP(OP_SOCKPAIR,XTERM);
4578 checkcomma(s,PL_tokenbuf,"subroutine name");
4580 if (*s == ';' || *s == ')') /* probably a close */
4581 Perl_croak(aTHX_ "sort is now a reserved word");
4583 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4587 LOP(OP_SPLIT,XTERM);
4590 LOP(OP_SPRINTF,XTERM);
4593 LOP(OP_SPLICE,XTERM);
4608 LOP(OP_SUBSTR,XTERM);
4614 char tmpbuf[sizeof PL_tokenbuf];
4616 expectation attrful;
4617 bool have_name, have_proto;
4622 if (isIDFIRST_lazy(s) || *s == '\'' ||
4623 (*s == ':' && s[1] == ':'))
4626 attrful = XATTRBLOCK;
4627 /* remember buffer pos'n for later force_word */
4628 tboffset = s - PL_oldbufptr;
4629 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4630 if (strchr(tmpbuf, ':'))
4631 sv_setpv(PL_subname, tmpbuf);
4633 sv_setsv(PL_subname,PL_curstname);
4634 sv_catpvn(PL_subname,"::",2);
4635 sv_catpvn(PL_subname,tmpbuf,len);
4642 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4643 PL_expect = XTERMBLOCK;
4644 attrful = XATTRTERM;
4645 sv_setpv(PL_subname,"?");
4649 if (key == KEY_format) {
4651 PL_lex_formbrack = PL_lex_brackets + 1;
4653 (void) force_word(PL_oldbufptr + tboffset, WORD,
4658 /* Look for a prototype */
4662 s = scan_str(s,FALSE,FALSE);
4665 SvREFCNT_dec(PL_lex_stuff);
4666 PL_lex_stuff = Nullsv;
4667 Perl_croak(aTHX_ "Prototype not terminated");
4670 d = SvPVX(PL_lex_stuff);
4672 for (p = d; *p; ++p) {
4677 SvCUR(PL_lex_stuff) = tmp;
4685 if (*s == ':' && s[1] != ':')
4686 PL_expect = attrful;
4689 PL_nextval[PL_nexttoke].opval =
4690 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
4691 PL_lex_stuff = Nullsv;
4695 sv_setpv(PL_subname,"__ANON__");
4698 (void) force_word(PL_oldbufptr + tboffset, WORD,
4707 LOP(OP_SYSTEM,XREF);
4710 LOP(OP_SYMLINK,XTERM);
4713 LOP(OP_SYSCALL,XTERM);
4716 LOP(OP_SYSOPEN,XTERM);
4719 LOP(OP_SYSSEEK,XTERM);
4722 LOP(OP_SYSREAD,XTERM);
4725 LOP(OP_SYSWRITE,XTERM);
4729 TERM(sublex_start());
4750 LOP(OP_TRUNCATE,XTERM);
4762 yylval.ival = CopLINE(PL_curcop);
4766 yylval.ival = CopLINE(PL_curcop);
4770 LOP(OP_UNLINK,XTERM);
4776 LOP(OP_UNPACK,XTERM);
4779 LOP(OP_UTIME,XTERM);
4782 if (ckWARN(WARN_OCTAL)) {
4783 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4784 if (*d != '0' && isDIGIT(*d))
4785 Perl_warner(aTHX_ WARN_OCTAL,
4786 "umask: argument is missing initial 0");
4791 LOP(OP_UNSHIFT,XTERM);
4794 if (PL_expect != XSTATE)
4795 yyerror("\"use\" not allowed in expression");
4797 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4798 s = force_version(s);
4799 if (*s == ';' || (s = skipspace(s), *s == ';')) {
4800 PL_nextval[PL_nexttoke].opval = Nullop;
4805 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4806 s = force_version(s);
4818 yylval.ival = CopLINE(PL_curcop);
4822 PL_hints |= HINT_BLOCK_SCOPE;
4829 LOP(OP_WAITPID,XTERM);
4837 static char ctl_l[2];
4839 if (ctl_l[0] == '\0')
4840 ctl_l[0] = toCTRL('L');
4841 gv_fetchpv(ctl_l,TRUE, SVt_PV);
4844 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
4849 if (PL_expect == XOPERATOR)
4855 yylval.ival = OP_XOR;
4860 TERM(sublex_start());
4866 Perl_keyword(pTHX_ register char *d, I32 len)
4871 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
4872 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
4873 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
4874 if (strEQ(d,"__DATA__")) return KEY___DATA__;
4875 if (strEQ(d,"__END__")) return KEY___END__;
4879 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
4884 if (strEQ(d,"and")) return -KEY_and;
4885 if (strEQ(d,"abs")) return -KEY_abs;
4888 if (strEQ(d,"alarm")) return -KEY_alarm;
4889 if (strEQ(d,"atan2")) return -KEY_atan2;
4892 if (strEQ(d,"accept")) return -KEY_accept;
4897 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
4900 if (strEQ(d,"bless")) return -KEY_bless;
4901 if (strEQ(d,"bind")) return -KEY_bind;
4902 if (strEQ(d,"binmode")) return -KEY_binmode;
4905 if (strEQ(d,"CORE")) return -KEY_CORE;
4910 if (strEQ(d,"cmp")) return -KEY_cmp;
4911 if (strEQ(d,"chr")) return -KEY_chr;
4912 if (strEQ(d,"cos")) return -KEY_cos;
4915 if (strEQ(d,"chop")) return KEY_chop;
4918 if (strEQ(d,"close")) return -KEY_close;
4919 if (strEQ(d,"chdir")) return -KEY_chdir;
4920 if (strEQ(d,"chomp")) return KEY_chomp;
4921 if (strEQ(d,"chmod")) return -KEY_chmod;
4922 if (strEQ(d,"chown")) return -KEY_chown;
4923 if (strEQ(d,"crypt")) return -KEY_crypt;
4926 if (strEQ(d,"chroot")) return -KEY_chroot;
4927 if (strEQ(d,"caller")) return -KEY_caller;
4930 if (strEQ(d,"connect")) return -KEY_connect;
4933 if (strEQ(d,"closedir")) return -KEY_closedir;
4934 if (strEQ(d,"continue")) return -KEY_continue;
4939 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
4944 if (strEQ(d,"do")) return KEY_do;
4947 if (strEQ(d,"die")) return -KEY_die;
4950 if (strEQ(d,"dump")) return -KEY_dump;
4953 if (strEQ(d,"delete")) return KEY_delete;
4956 if (strEQ(d,"defined")) return KEY_defined;
4957 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
4960 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
4965 if (strEQ(d,"EQ")) { deprecate(d); return -KEY_eq;}
4966 if (strEQ(d,"END")) return KEY_END;
4971 if (strEQ(d,"eq")) return -KEY_eq;
4974 if (strEQ(d,"eof")) return -KEY_eof;
4975 if (strEQ(d,"exp")) return -KEY_exp;
4978 if (strEQ(d,"else")) return KEY_else;
4979 if (strEQ(d,"exit")) return -KEY_exit;
4980 if (strEQ(d,"eval")) return KEY_eval;
4981 if (strEQ(d,"exec")) return -KEY_exec;
4982 if (strEQ(d,"each")) return KEY_each;
4985 if (strEQ(d,"elsif")) return KEY_elsif;
4988 if (strEQ(d,"exists")) return KEY_exists;
4989 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
4992 if (strEQ(d,"endgrent")) return -KEY_endgrent;
4993 if (strEQ(d,"endpwent")) return -KEY_endpwent;
4996 if (strEQ(d,"endnetent")) return -KEY_endnetent;
4999 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5000 if (strEQ(d,"endservent")) return -KEY_endservent;
5003 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5010 if (strEQ(d,"for")) return KEY_for;
5013 if (strEQ(d,"fork")) return -KEY_fork;
5016 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5017 if (strEQ(d,"flock")) return -KEY_flock;
5020 if (strEQ(d,"format")) return KEY_format;
5021 if (strEQ(d,"fileno")) return -KEY_fileno;
5024 if (strEQ(d,"foreach")) return KEY_foreach;
5027 if (strEQ(d,"formline")) return -KEY_formline;
5033 if (strEQ(d,"GT")) { deprecate(d); return -KEY_gt;}
5034 if (strEQ(d,"GE")) { deprecate(d); return -KEY_ge;}
5038 if (strnEQ(d,"get",3)) {
5043 if (strEQ(d,"ppid")) return -KEY_getppid;
5044 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5047 if (strEQ(d,"pwent")) return -KEY_getpwent;
5048 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5049 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5052 if (strEQ(d,"peername")) return -KEY_getpeername;
5053 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5054 if (strEQ(d,"priority")) return -KEY_getpriority;
5057 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5060 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5064 else if (*d == 'h') {
5065 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5066 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5067 if (strEQ(d,"hostent")) return -KEY_gethostent;
5069 else if (*d == 'n') {
5070 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5071 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5072 if (strEQ(d,"netent")) return -KEY_getnetent;
5074 else if (*d == 's') {
5075 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5076 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5077 if (strEQ(d,"servent")) return -KEY_getservent;
5078 if (strEQ(d,"sockname")) return -KEY_getsockname;
5079 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5081 else if (*d == 'g') {
5082 if (strEQ(d,"grent")) return -KEY_getgrent;
5083 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5084 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5086 else if (*d == 'l') {
5087 if (strEQ(d,"login")) return -KEY_getlogin;
5089 else if (strEQ(d,"c")) return -KEY_getc;
5094 if (strEQ(d,"gt")) return -KEY_gt;
5095 if (strEQ(d,"ge")) return -KEY_ge;
5098 if (strEQ(d,"grep")) return KEY_grep;
5099 if (strEQ(d,"goto")) return KEY_goto;
5100 if (strEQ(d,"glob")) return KEY_glob;
5103 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5108 if (strEQ(d,"hex")) return -KEY_hex;
5111 if (strEQ(d,"INIT")) return KEY_INIT;
5116 if (strEQ(d,"if")) return KEY_if;
5119 if (strEQ(d,"int")) return -KEY_int;
5122 if (strEQ(d,"index")) return -KEY_index;
5123 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5128 if (strEQ(d,"join")) return -KEY_join;
5132 if (strEQ(d,"keys")) return KEY_keys;
5133 if (strEQ(d,"kill")) return -KEY_kill;
5138 if (strEQ(d,"LT")) { deprecate(d); return -KEY_lt;}
5139 if (strEQ(d,"LE")) { deprecate(d); return -KEY_le;}
5145 if (strEQ(d,"lt")) return -KEY_lt;
5146 if (strEQ(d,"le")) return -KEY_le;
5147 if (strEQ(d,"lc")) return -KEY_lc;
5150 if (strEQ(d,"log")) return -KEY_log;
5153 if (strEQ(d,"last")) return KEY_last;
5154 if (strEQ(d,"link")) return -KEY_link;
5155 if (strEQ(d,"lock")) return -KEY_lock;
5158 if (strEQ(d,"local")) return KEY_local;
5159 if (strEQ(d,"lstat")) return -KEY_lstat;
5162 if (strEQ(d,"length")) return -KEY_length;
5163 if (strEQ(d,"listen")) return -KEY_listen;
5166 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5169 if (strEQ(d,"localtime")) return -KEY_localtime;
5175 case 1: return KEY_m;
5177 if (strEQ(d,"my")) return KEY_my;
5180 if (strEQ(d,"map")) return KEY_map;
5183 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5186 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5187 if (strEQ(d,"msgget")) return -KEY_msgget;
5188 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5189 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5194 if (strEQ(d,"NE")) { deprecate(d); return -KEY_ne;}
5197 if (strEQ(d,"next")) return KEY_next;
5198 if (strEQ(d,"ne")) return -KEY_ne;
5199 if (strEQ(d,"not")) return -KEY_not;
5200 if (strEQ(d,"no")) return KEY_no;
5205 if (strEQ(d,"or")) return -KEY_or;
5208 if (strEQ(d,"ord")) return -KEY_ord;
5209 if (strEQ(d,"oct")) return -KEY_oct;
5210 if (strEQ(d,"our")) return KEY_our;
5213 if (strEQ(d,"open")) return -KEY_open;
5216 if (strEQ(d,"opendir")) return -KEY_opendir;
5223 if (strEQ(d,"pop")) return KEY_pop;
5224 if (strEQ(d,"pos")) return KEY_pos;
5227 if (strEQ(d,"push")) return KEY_push;
5228 if (strEQ(d,"pack")) return -KEY_pack;
5229 if (strEQ(d,"pipe")) return -KEY_pipe;
5232 if (strEQ(d,"print")) return KEY_print;
5235 if (strEQ(d,"printf")) return KEY_printf;
5238 if (strEQ(d,"package")) return KEY_package;
5241 if (strEQ(d,"prototype")) return KEY_prototype;
5246 if (strEQ(d,"q")) return KEY_q;
5247 if (strEQ(d,"qr")) return KEY_qr;
5248 if (strEQ(d,"qq")) return KEY_qq;
5249 if (strEQ(d,"qw")) return KEY_qw;
5250 if (strEQ(d,"qx")) return KEY_qx;
5252 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5257 if (strEQ(d,"ref")) return -KEY_ref;
5260 if (strEQ(d,"read")) return -KEY_read;
5261 if (strEQ(d,"rand")) return -KEY_rand;
5262 if (strEQ(d,"recv")) return -KEY_recv;
5263 if (strEQ(d,"redo")) return KEY_redo;
5266 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5267 if (strEQ(d,"reset")) return -KEY_reset;
5270 if (strEQ(d,"return")) return KEY_return;
5271 if (strEQ(d,"rename")) return -KEY_rename;
5272 if (strEQ(d,"rindex")) return -KEY_rindex;
5275 if (strEQ(d,"require")) return -KEY_require;
5276 if (strEQ(d,"reverse")) return -KEY_reverse;
5277 if (strEQ(d,"readdir")) return -KEY_readdir;
5280 if (strEQ(d,"readlink")) return -KEY_readlink;
5281 if (strEQ(d,"readline")) return -KEY_readline;
5282 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5285 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5290 if (strEQ(d,"STOP")) return KEY_STOP;
5294 case 0: return KEY_s;
5296 if (strEQ(d,"scalar")) return KEY_scalar;
5301 if (strEQ(d,"seek")) return -KEY_seek;
5302 if (strEQ(d,"send")) return -KEY_send;
5305 if (strEQ(d,"semop")) return -KEY_semop;
5308 if (strEQ(d,"select")) return -KEY_select;
5309 if (strEQ(d,"semctl")) return -KEY_semctl;
5310 if (strEQ(d,"semget")) return -KEY_semget;
5313 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5314 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5317 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5318 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5321 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5324 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5325 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5326 if (strEQ(d,"setservent")) return -KEY_setservent;
5329 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5330 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5337 if (strEQ(d,"shift")) return KEY_shift;
5340 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5341 if (strEQ(d,"shmget")) return -KEY_shmget;
5344 if (strEQ(d,"shmread")) return -KEY_shmread;
5347 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5348 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5353 if (strEQ(d,"sin")) return -KEY_sin;
5356 if (strEQ(d,"sleep")) return -KEY_sleep;
5359 if (strEQ(d,"sort")) return KEY_sort;
5360 if (strEQ(d,"socket")) return -KEY_socket;
5361 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5364 if (strEQ(d,"split")) return KEY_split;
5365 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5366 if (strEQ(d,"splice")) return KEY_splice;
5369 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5372 if (strEQ(d,"srand")) return -KEY_srand;
5375 if (strEQ(d,"stat")) return -KEY_stat;
5376 if (strEQ(d,"study")) return KEY_study;
5379 if (strEQ(d,"substr")) return -KEY_substr;
5380 if (strEQ(d,"sub")) return KEY_sub;
5385 if (strEQ(d,"system")) return -KEY_system;
5388 if (strEQ(d,"symlink")) return -KEY_symlink;
5389 if (strEQ(d,"syscall")) return -KEY_syscall;
5390 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5391 if (strEQ(d,"sysread")) return -KEY_sysread;
5392 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5395 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5404 if (strEQ(d,"tr")) return KEY_tr;
5407 if (strEQ(d,"tie")) return KEY_tie;
5410 if (strEQ(d,"tell")) return -KEY_tell;
5411 if (strEQ(d,"tied")) return KEY_tied;
5412 if (strEQ(d,"time")) return -KEY_time;
5415 if (strEQ(d,"times")) return -KEY_times;
5418 if (strEQ(d,"telldir")) return -KEY_telldir;
5421 if (strEQ(d,"truncate")) return -KEY_truncate;
5428 if (strEQ(d,"uc")) return -KEY_uc;
5431 if (strEQ(d,"use")) return KEY_use;
5434 if (strEQ(d,"undef")) return KEY_undef;
5435 if (strEQ(d,"until")) return KEY_until;
5436 if (strEQ(d,"untie")) return KEY_untie;
5437 if (strEQ(d,"utime")) return -KEY_utime;
5438 if (strEQ(d,"umask")) return -KEY_umask;
5441 if (strEQ(d,"unless")) return KEY_unless;
5442 if (strEQ(d,"unpack")) return -KEY_unpack;
5443 if (strEQ(d,"unlink")) return -KEY_unlink;
5446 if (strEQ(d,"unshift")) return KEY_unshift;
5447 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5452 if (strEQ(d,"values")) return -KEY_values;
5453 if (strEQ(d,"vec")) return -KEY_vec;
5458 if (strEQ(d,"warn")) return -KEY_warn;
5459 if (strEQ(d,"wait")) return -KEY_wait;
5462 if (strEQ(d,"while")) return KEY_while;
5463 if (strEQ(d,"write")) return -KEY_write;
5466 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5469 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5474 if (len == 1) return -KEY_x;
5475 if (strEQ(d,"xor")) return -KEY_xor;
5478 if (len == 1) return KEY_y;
5487 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5491 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5492 dTHR; /* only for ckWARN */
5493 if (ckWARN(WARN_SYNTAX)) {
5495 for (w = s+2; *w && level; w++) {
5502 for (; *w && isSPACE(*w); w++) ;
5503 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5504 Perl_warner(aTHX_ WARN_SYNTAX,
5505 "%s (...) interpreted as function",name);
5508 while (s < PL_bufend && isSPACE(*s))
5512 while (s < PL_bufend && isSPACE(*s))
5514 if (isIDFIRST_lazy(s)) {
5516 while (isALNUM_lazy(s))
5518 while (s < PL_bufend && isSPACE(*s))
5523 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5527 Perl_croak(aTHX_ "No comma allowed after %s", what);
5532 /* Either returns sv, or mortalizes sv and returns a new SV*.
5533 Best used as sv=new_constant(..., sv, ...).
5534 If s, pv are NULL, calls subroutine with one argument,
5535 and type is used with error messages only. */
5538 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5542 HV *table = GvHV(PL_hintgv); /* ^H */
5546 const char *why, *why1, *why2;
5548 if (!(PL_hints & HINT_LOCALIZE_HH)) {
5551 why = "%^H is not localized";
5555 msg = Perl_newSVpvf(aTHX_ "constant(%s): %s%s%s",
5556 (type ? type: "undef"), why1, why2, why);
5557 yyerror(SvPVX(msg));
5562 why = "%^H is not defined";
5565 cvp = hv_fetch(table, key, strlen(key), FALSE);
5566 if (!cvp || !SvOK(*cvp)) {
5567 why = "} is not defined";
5572 sv_2mortal(sv); /* Parent created it permanently */
5575 pv = sv_2mortal(newSVpvn(s, len));
5577 typesv = sv_2mortal(newSVpv(type, 0));
5579 typesv = &PL_sv_undef;
5581 PUSHSTACKi(PERLSI_OVERLOAD);
5594 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5598 /* Check the eval first */
5599 if (!PL_in_eval && SvTRUE(ERRSV))
5602 sv_catpv(ERRSV, "Propagated");
5603 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5605 res = SvREFCNT_inc(sv);
5609 (void)SvREFCNT_inc(res);
5618 why = "}} did not return a defined value";
5619 why1 = "Call to &{$^H{";
5629 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
5631 register char *d = dest;
5632 register char *e = d + destlen - 3; /* two-character token, ending NUL */
5635 Perl_croak(aTHX_ ident_too_long);
5636 if (isALNUM(*s)) /* UTF handled below */
5638 else if (*s == '\'' && allow_package && isIDFIRST_lazy(s+1)) {
5643 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
5647 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5648 char *t = s + UTF8SKIP(s);
5649 while (*t & 0x80 && is_utf8_mark((U8*)t))
5651 if (d + (t - s) > e)
5652 Perl_croak(aTHX_ ident_too_long);
5653 Copy(s, d, t - s, char);
5666 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
5676 e = d + destlen - 3; /* two-character token, ending NUL */
5678 while (isDIGIT(*s)) {
5680 Perl_croak(aTHX_ ident_too_long);
5687 Perl_croak(aTHX_ ident_too_long);
5688 if (isALNUM(*s)) /* UTF handled below */
5690 else if (*s == '\'' && isIDFIRST_lazy(s+1)) {
5695 else if (*s == ':' && s[1] == ':') {
5699 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5700 char *t = s + UTF8SKIP(s);
5701 while (*t & 0x80 && is_utf8_mark((U8*)t))
5703 if (d + (t - s) > e)
5704 Perl_croak(aTHX_ ident_too_long);
5705 Copy(s, d, t - s, char);
5716 if (PL_lex_state != LEX_NORMAL)
5717 PL_lex_state = LEX_INTERPENDMAYBE;
5720 if (*s == '$' && s[1] &&
5721 (isALNUM_lazy(s+1) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
5734 if (*d == '^' && *s && isCONTROLVAR(*s)) {
5739 if (isSPACE(s[-1])) {
5742 if (ch != ' ' && ch != '\t') {
5748 if (isIDFIRST_lazy(d)) {
5752 while (e < send && isALNUM_lazy(e) || *e == ':') {
5754 while (e < send && *e & 0x80 && is_utf8_mark((U8*)e))
5757 Copy(s, d, e - s, char);
5762 while ((isALNUM(*s) || *s == ':') && d < e)
5765 Perl_croak(aTHX_ ident_too_long);
5768 while (s < send && (*s == ' ' || *s == '\t')) s++;
5769 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
5770 dTHR; /* only for ckWARN */
5771 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
5772 const char *brack = *s == '[' ? "[...]" : "{...}";
5773 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5774 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
5775 funny, dest, brack, funny, dest, brack);
5778 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
5782 /* Handle extended ${^Foo} variables
5783 * 1999-02-27 mjd-perl-patch@plover.com */
5784 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
5788 while (isALNUM(*s) && d < e) {
5792 Perl_croak(aTHX_ ident_too_long);
5797 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
5798 PL_lex_state = LEX_INTERPEND;
5801 if (PL_lex_state == LEX_NORMAL) {
5802 dTHR; /* only for ckWARN */
5803 if (ckWARN(WARN_AMBIGUOUS) &&
5804 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
5806 Perl_warner(aTHX_ WARN_AMBIGUOUS,
5807 "Ambiguous use of %c{%s} resolved to %c%s",
5808 funny, dest, funny, dest);
5813 s = bracket; /* let the parser handle it */
5817 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
5818 PL_lex_state = LEX_INTERPEND;
5823 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
5828 *pmfl |= PMf_GLOBAL;
5830 *pmfl |= PMf_CONTINUE;
5834 *pmfl |= PMf_MULTILINE;
5836 *pmfl |= PMf_SINGLELINE;
5838 *pmfl |= PMf_EXTENDED;
5842 S_scan_pat(pTHX_ char *start, I32 type)
5847 s = scan_str(start,FALSE,FALSE);
5850 SvREFCNT_dec(PL_lex_stuff);
5851 PL_lex_stuff = Nullsv;
5852 Perl_croak(aTHX_ "Search pattern not terminated");
5855 pm = (PMOP*)newPMOP(type, 0);
5856 if (PL_multi_open == '?')
5857 pm->op_pmflags |= PMf_ONCE;
5859 while (*s && strchr("iomsx", *s))
5860 pmflag(&pm->op_pmflags,*s++);
5863 while (*s && strchr("iogcmsx", *s))
5864 pmflag(&pm->op_pmflags,*s++);
5866 pm->op_pmpermflags = pm->op_pmflags;
5868 PL_lex_op = (OP*)pm;
5869 yylval.ival = OP_MATCH;
5874 S_scan_subst(pTHX_ char *start)
5881 yylval.ival = OP_NULL;
5883 s = scan_str(start,FALSE,FALSE);
5887 SvREFCNT_dec(PL_lex_stuff);
5888 PL_lex_stuff = Nullsv;
5889 Perl_croak(aTHX_ "Substitution pattern not terminated");
5892 if (s[-1] == PL_multi_open)
5895 first_start = PL_multi_start;
5896 s = scan_str(s,FALSE,FALSE);
5899 SvREFCNT_dec(PL_lex_stuff);
5900 PL_lex_stuff = Nullsv;
5902 SvREFCNT_dec(PL_lex_repl);
5903 PL_lex_repl = Nullsv;
5904 Perl_croak(aTHX_ "Substitution replacement not terminated");
5906 PL_multi_start = first_start; /* so whole substitution is taken together */
5908 pm = (PMOP*)newPMOP(OP_SUBST, 0);
5914 else if (strchr("iogcmsx", *s))
5915 pmflag(&pm->op_pmflags,*s++);
5922 PL_sublex_info.super_bufptr = s;
5923 PL_sublex_info.super_bufend = PL_bufend;
5925 pm->op_pmflags |= PMf_EVAL;
5926 repl = newSVpvn("",0);
5928 sv_catpv(repl, es ? "eval " : "do ");
5929 sv_catpvn(repl, "{ ", 2);
5930 sv_catsv(repl, PL_lex_repl);
5931 sv_catpvn(repl, " };", 2);
5933 SvREFCNT_dec(PL_lex_repl);
5937 pm->op_pmpermflags = pm->op_pmflags;
5938 PL_lex_op = (OP*)pm;
5939 yylval.ival = OP_SUBST;
5944 S_scan_trans(pTHX_ char *start)
5955 yylval.ival = OP_NULL;
5957 s = scan_str(start,FALSE,FALSE);
5960 SvREFCNT_dec(PL_lex_stuff);
5961 PL_lex_stuff = Nullsv;
5962 Perl_croak(aTHX_ "Transliteration pattern not terminated");
5964 if (s[-1] == PL_multi_open)
5967 s = scan_str(s,FALSE,FALSE);
5970 SvREFCNT_dec(PL_lex_stuff);
5971 PL_lex_stuff = Nullsv;
5973 SvREFCNT_dec(PL_lex_repl);
5974 PL_lex_repl = Nullsv;
5975 Perl_croak(aTHX_ "Transliteration replacement not terminated");
5979 o = newSVOP(OP_TRANS, 0, 0);
5980 utf8 = OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF;
5983 New(803,tbl,256,short);
5984 o = newPVOP(OP_TRANS, 0, (char*)tbl);
5988 complement = del = squash = 0;
5989 while (strchr("cdsCU", *s)) {
5991 complement = OPpTRANS_COMPLEMENT;
5993 del = OPpTRANS_DELETE;
5995 squash = OPpTRANS_SQUASH;
6000 utf8 &= ~OPpTRANS_FROM_UTF;
6002 utf8 |= OPpTRANS_FROM_UTF;
6006 utf8 &= ~OPpTRANS_TO_UTF;
6008 utf8 |= OPpTRANS_TO_UTF;
6011 Perl_croak(aTHX_ "Too many /C and /U options");
6016 o->op_private = del|squash|complement|utf8;
6019 yylval.ival = OP_TRANS;
6024 S_scan_heredoc(pTHX_ register char *s)
6028 I32 op_type = OP_SCALAR;
6035 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6039 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6042 for (peek = s; *peek == ' ' || *peek == '\t'; peek++) ;
6043 if (*peek && strchr("`'\"",*peek)) {
6046 s = delimcpy(d, e, s, PL_bufend, term, &len);
6056 if (!isALNUM_lazy(s))
6057 deprecate("bare << to mean <<\"\"");
6058 for (; isALNUM_lazy(s); s++) {
6063 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6064 Perl_croak(aTHX_ "Delimiter for here document is too long");
6067 len = d - PL_tokenbuf;
6068 #ifndef PERL_STRICT_CR
6069 d = strchr(s, '\r');
6073 while (s < PL_bufend) {
6079 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6088 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6093 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6094 herewas = newSVpvn(s,PL_bufend-s);
6096 s--, herewas = newSVpvn(s,d-s);
6097 s += SvCUR(herewas);
6099 tmpstr = NEWSV(87,79);
6100 sv_upgrade(tmpstr, SVt_PVIV);
6105 else if (term == '`') {
6106 op_type = OP_BACKTICK;
6107 SvIVX(tmpstr) = '\\';
6111 PL_multi_start = CopLINE(PL_curcop);
6112 PL_multi_open = PL_multi_close = '<';
6113 term = *PL_tokenbuf;
6114 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6115 char *bufptr = PL_sublex_info.super_bufptr;
6116 char *bufend = PL_sublex_info.super_bufend;
6117 char *olds = s - SvCUR(herewas);
6118 s = strchr(bufptr, '\n');
6122 while (s < bufend &&
6123 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6125 CopLINE_inc(PL_curcop);
6128 CopLINE_set(PL_curcop, PL_multi_start);
6129 missingterm(PL_tokenbuf);
6131 sv_setpvn(herewas,bufptr,d-bufptr+1);
6132 sv_setpvn(tmpstr,d+1,s-d);
6134 sv_catpvn(herewas,s,bufend-s);
6135 (void)strcpy(bufptr,SvPVX(herewas));
6142 while (s < PL_bufend &&
6143 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6145 CopLINE_inc(PL_curcop);
6147 if (s >= PL_bufend) {
6148 CopLINE_set(PL_curcop, PL_multi_start);
6149 missingterm(PL_tokenbuf);
6151 sv_setpvn(tmpstr,d+1,s-d);
6153 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6155 sv_catpvn(herewas,s,PL_bufend-s);
6156 sv_setsv(PL_linestr,herewas);
6157 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6158 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6161 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6162 while (s >= PL_bufend) { /* multiple line string? */
6164 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6165 CopLINE_set(PL_curcop, PL_multi_start);
6166 missingterm(PL_tokenbuf);
6168 CopLINE_inc(PL_curcop);
6169 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6170 #ifndef PERL_STRICT_CR
6171 if (PL_bufend - PL_linestart >= 2) {
6172 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6173 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6175 PL_bufend[-2] = '\n';
6177 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6179 else if (PL_bufend[-1] == '\r')
6180 PL_bufend[-1] = '\n';
6182 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6183 PL_bufend[-1] = '\n';
6185 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6186 SV *sv = NEWSV(88,0);
6188 sv_upgrade(sv, SVt_PVMG);
6189 sv_setsv(sv,PL_linestr);
6190 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6192 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6195 sv_catsv(PL_linestr,herewas);
6196 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6200 sv_catsv(tmpstr,PL_linestr);
6205 PL_multi_end = CopLINE(PL_curcop);
6206 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6207 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6208 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6210 SvREFCNT_dec(herewas);
6211 PL_lex_stuff = tmpstr;
6212 yylval.ival = op_type;
6217 takes: current position in input buffer
6218 returns: new position in input buffer
6219 side-effects: yylval and lex_op are set.
6224 <FH> read from filehandle
6225 <pkg::FH> read from package qualified filehandle
6226 <pkg'FH> read from package qualified filehandle
6227 <$fh> read from filehandle in $fh
6233 S_scan_inputsymbol(pTHX_ char *start)
6235 register char *s = start; /* current position in buffer */
6241 d = PL_tokenbuf; /* start of temp holding space */
6242 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6243 end = strchr(s, '\n');
6246 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6248 /* die if we didn't have space for the contents of the <>,
6249 or if it didn't end, or if we see a newline
6252 if (len >= sizeof PL_tokenbuf)
6253 Perl_croak(aTHX_ "Excessively long <> operator");
6255 Perl_croak(aTHX_ "Unterminated <> operator");
6260 Remember, only scalar variables are interpreted as filehandles by
6261 this code. Anything more complex (e.g., <$fh{$num}>) will be
6262 treated as a glob() call.
6263 This code makes use of the fact that except for the $ at the front,
6264 a scalar variable and a filehandle look the same.
6266 if (*d == '$' && d[1]) d++;
6268 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6269 while (*d && (isALNUM_lazy(d) || *d == '\'' || *d == ':'))
6272 /* If we've tried to read what we allow filehandles to look like, and
6273 there's still text left, then it must be a glob() and not a getline.
6274 Use scan_str to pull out the stuff between the <> and treat it
6275 as nothing more than a string.
6278 if (d - PL_tokenbuf != len) {
6279 yylval.ival = OP_GLOB;
6281 s = scan_str(start,FALSE,FALSE);
6283 Perl_croak(aTHX_ "Glob not terminated");
6287 /* we're in a filehandle read situation */
6290 /* turn <> into <ARGV> */
6292 (void)strcpy(d,"ARGV");
6294 /* if <$fh>, create the ops to turn the variable into a
6300 /* try to find it in the pad for this block, otherwise find
6301 add symbol table ops
6303 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6304 OP *o = newOP(OP_PADSV, 0);
6306 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o);
6309 GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV);
6310 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0,
6311 newUNOP(OP_RV2SV, 0,
6312 newGVOP(OP_GV, 0, gv)));
6314 PL_lex_op->op_flags |= OPf_SPECIAL;
6315 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6316 yylval.ival = OP_NULL;
6319 /* If it's none of the above, it must be a literal filehandle
6320 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6322 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6323 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6324 yylval.ival = OP_NULL;
6333 takes: start position in buffer
6334 keep_quoted preserve \ on the embedded delimiter(s)
6335 keep_delims preserve the delimiters around the string
6336 returns: position to continue reading from buffer
6337 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6338 updates the read buffer.
6340 This subroutine pulls a string out of the input. It is called for:
6341 q single quotes q(literal text)
6342 ' single quotes 'literal text'
6343 qq double quotes qq(interpolate $here please)
6344 " double quotes "interpolate $here please"
6345 qx backticks qx(/bin/ls -l)
6346 ` backticks `/bin/ls -l`
6347 qw quote words @EXPORT_OK = qw( func() $spam )
6348 m// regexp match m/this/
6349 s/// regexp substitute s/this/that/
6350 tr/// string transliterate tr/this/that/
6351 y/// string transliterate y/this/that/
6352 ($*@) sub prototypes sub foo ($)
6353 (stuff) sub attr parameters sub foo : attr(stuff)
6354 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6356 In most of these cases (all but <>, patterns and transliterate)
6357 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6358 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6359 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6362 It skips whitespace before the string starts, and treats the first
6363 character as the delimiter. If the delimiter is one of ([{< then
6364 the corresponding "close" character )]}> is used as the closing
6365 delimiter. It allows quoting of delimiters, and if the string has
6366 balanced delimiters ([{<>}]) it allows nesting.
6368 The lexer always reads these strings into lex_stuff, except in the
6369 case of the operators which take *two* arguments (s/// and tr///)
6370 when it checks to see if lex_stuff is full (presumably with the 1st
6371 arg to s or tr) and if so puts the string into lex_repl.
6376 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6379 SV *sv; /* scalar value: string */
6380 char *tmps; /* temp string, used for delimiter matching */
6381 register char *s = start; /* current position in the buffer */
6382 register char term; /* terminating character */
6383 register char *to; /* current position in the sv's data */
6384 I32 brackets = 1; /* bracket nesting level */
6386 /* skip space before the delimiter */
6390 /* mark where we are, in case we need to report errors */
6393 /* after skipping whitespace, the next character is the terminator */
6395 /* mark where we are */
6396 PL_multi_start = CopLINE(PL_curcop);
6397 PL_multi_open = term;
6399 /* find corresponding closing delimiter */
6400 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6402 PL_multi_close = term;
6404 /* create a new SV to hold the contents. 87 is leak category, I'm
6405 assuming. 79 is the SV's initial length. What a random number. */
6407 sv_upgrade(sv, SVt_PVIV);
6409 (void)SvPOK_only(sv); /* validate pointer */
6411 /* move past delimiter and try to read a complete string */
6413 sv_catpvn(sv, s, 1);
6416 /* extend sv if need be */
6417 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6418 /* set 'to' to the next character in the sv's string */
6419 to = SvPVX(sv)+SvCUR(sv);
6421 /* if open delimiter is the close delimiter read unbridle */
6422 if (PL_multi_open == PL_multi_close) {
6423 for (; s < PL_bufend; s++,to++) {
6424 /* embedded newlines increment the current line number */
6425 if (*s == '\n' && !PL_rsfp)
6426 CopLINE_inc(PL_curcop);
6427 /* handle quoted delimiters */
6428 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6429 if (!keep_quoted && s[1] == term)
6431 /* any other quotes are simply copied straight through */
6435 /* terminate when run out of buffer (the for() condition), or
6436 have found the terminator */
6437 else if (*s == term)
6443 /* if the terminator isn't the same as the start character (e.g.,
6444 matched brackets), we have to allow more in the quoting, and
6445 be prepared for nested brackets.
6448 /* read until we run out of string, or we find the terminator */
6449 for (; s < PL_bufend; s++,to++) {
6450 /* embedded newlines increment the line count */
6451 if (*s == '\n' && !PL_rsfp)
6452 CopLINE_inc(PL_curcop);
6453 /* backslashes can escape the open or closing characters */
6454 if (*s == '\\' && s+1 < PL_bufend) {
6456 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6461 /* allow nested opens and closes */
6462 else if (*s == PL_multi_close && --brackets <= 0)
6464 else if (*s == PL_multi_open)
6469 /* terminate the copied string and update the sv's end-of-string */
6471 SvCUR_set(sv, to - SvPVX(sv));
6474 * this next chunk reads more into the buffer if we're not done yet
6477 if (s < PL_bufend) break; /* handle case where we are done yet :-) */
6479 #ifndef PERL_STRICT_CR
6480 if (to - SvPVX(sv) >= 2) {
6481 if ((to[-2] == '\r' && to[-1] == '\n') ||
6482 (to[-2] == '\n' && to[-1] == '\r'))
6486 SvCUR_set(sv, to - SvPVX(sv));
6488 else if (to[-1] == '\r')
6491 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6495 /* if we're out of file, or a read fails, bail and reset the current
6496 line marker so we can report where the unterminated string began
6499 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6501 CopLINE_set(PL_curcop, PL_multi_start);
6504 /* we read a line, so increment our line counter */
6505 CopLINE_inc(PL_curcop);
6507 /* update debugger info */
6508 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6509 SV *sv = NEWSV(88,0);
6511 sv_upgrade(sv, SVt_PVMG);
6512 sv_setsv(sv,PL_linestr);
6513 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6516 /* having changed the buffer, we must update PL_bufend */
6517 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6520 /* at this point, we have successfully read the delimited string */
6523 sv_catpvn(sv, s, 1);
6524 PL_multi_end = CopLINE(PL_curcop);
6527 /* if we allocated too much space, give some back */
6528 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6529 SvLEN_set(sv, SvCUR(sv) + 1);
6530 Renew(SvPVX(sv), SvLEN(sv), char);
6533 /* decide whether this is the first or second quoted string we've read
6546 takes: pointer to position in buffer
6547 returns: pointer to new position in buffer
6548 side-effects: builds ops for the constant in yylval.op
6550 Read a number in any of the formats that Perl accepts:
6552 0(x[0-7A-F]+)|([0-7]+)|(b[01])
6553 [\d_]+(\.[\d_]*)?[Ee](\d+)
6555 Underbars (_) are allowed in decimal numbers. If -w is on,
6556 underbars before a decimal point must be at three digit intervals.
6558 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6561 If it reads a number without a decimal point or an exponent, it will
6562 try converting the number to an integer and see if it can do so
6563 without loss of precision.
6567 Perl_scan_num(pTHX_ char *start)
6569 register char *s = start; /* current position in buffer */
6570 register char *d; /* destination in temp buffer */
6571 register char *e; /* end of temp buffer */
6572 IV tryiv; /* used to see if it can be an IV */
6573 NV value; /* number read, as a double */
6574 SV *sv = Nullsv; /* place to put the converted number */
6575 bool floatit; /* boolean: int or float? */
6576 char *lastub = 0; /* position of last underbar */
6577 static char number_too_long[] = "Number too long";
6579 /* We use the first character to decide what type of number this is */
6583 Perl_croak(aTHX_ "panic: scan_num");
6585 /* if it starts with a 0, it could be an octal number, a decimal in
6586 0.13 disguise, or a hexadecimal number, or a binary number. */
6590 u holds the "number so far"
6591 shift the power of 2 of the base
6592 (hex == 4, octal == 3, binary == 1)
6593 overflowed was the number more than we can hold?
6595 Shift is used when we add a digit. It also serves as an "are
6596 we in octal/hex/binary?" indicator to disallow hex characters
6603 bool overflowed = FALSE;
6604 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6605 static char* bases[5] = { "", "binary", "", "octal",
6607 static char* Bases[5] = { "", "Binary", "", "Octal",
6609 static char *maxima[5] = { "",
6610 "0b11111111111111111111111111111111",
6614 char *base, *Base, *max;
6620 } else if (s[1] == 'b') {
6624 /* check for a decimal in disguise */
6625 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
6627 /* so it must be octal */
6631 base = bases[shift];
6632 Base = Bases[shift];
6633 max = maxima[shift];
6635 /* read the rest of the number */
6637 /* x is used in the overflow test,
6638 b is the digit we're adding on. */
6643 /* if we don't mention it, we're done */
6652 /* 8 and 9 are not octal */
6655 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
6659 case '2': case '3': case '4':
6660 case '5': case '6': case '7':
6662 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
6666 b = *s++ & 15; /* ASCII digit -> value of digit */
6670 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
6671 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
6672 /* make sure they said 0x */
6677 /* Prepare to put the digit we have onto the end
6678 of the number so far. We check for overflows.
6683 x = u << shift; /* make room for the digit */
6685 if ((x >> shift) != u
6686 && !(PL_hints & HINT_NEW_BINARY)) {
6690 if (ckWARN_d(WARN_OVERFLOW))
6691 Perl_warner(aTHX_ WARN_OVERFLOW,
6692 "Integer overflow in %s number",
6695 u = x | b; /* add the digit to the end */
6698 n *= nvshift[shift];
6699 /* If an NV has not enough bits in its
6700 * mantissa to represent an UV this summing of
6701 * small low-order numbers is a waste of time
6702 * (because the NV cannot preserve the
6703 * low-order bits anyway): we could just
6704 * remember when did we overflow and in the
6705 * end just multiply n by the right
6713 /* if we get here, we had success: make a scalar value from
6720 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
6721 Perl_warner(aTHX_ WARN_PORTABLE,
6722 "%s number > %s non-portable",
6729 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
6730 Perl_warner(aTHX_ WARN_PORTABLE,
6731 "%s number > %s non-portable",
6736 if (PL_hints & HINT_NEW_BINARY)
6737 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
6742 handle decimal numbers.
6743 we're also sent here when we read a 0 as the first digit
6745 case '1': case '2': case '3': case '4': case '5':
6746 case '6': case '7': case '8': case '9': case '.':
6749 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
6752 /* read next group of digits and _ and copy into d */
6753 while (isDIGIT(*s) || *s == '_') {
6754 /* skip underscores, checking for misplaced ones
6758 dTHR; /* only for ckWARN */
6759 if (ckWARN(WARN_SYNTAX) && lastub && s - lastub != 3)
6760 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6764 /* check for end of fixed-length buffer */
6766 Perl_croak(aTHX_ number_too_long);
6767 /* if we're ok, copy the character */
6772 /* final misplaced underbar check */
6773 if (lastub && s - lastub != 3) {
6775 if (ckWARN(WARN_SYNTAX))
6776 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
6779 /* read a decimal portion if there is one. avoid
6780 3..5 being interpreted as the number 3. followed
6783 if (*s == '.' && s[1] != '.') {
6787 /* copy, ignoring underbars, until we run out of
6788 digits. Note: no misplaced underbar checks!
6790 for (; isDIGIT(*s) || *s == '_'; s++) {
6791 /* fixed length buffer check */
6793 Perl_croak(aTHX_ number_too_long);
6799 /* read exponent part, if present */
6800 if (*s && strchr("eE",*s) && strchr("+-0123456789",s[1])) {
6804 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
6805 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
6807 /* allow positive or negative exponent */
6808 if (*s == '+' || *s == '-')
6811 /* read digits of exponent (no underbars :-) */
6812 while (isDIGIT(*s)) {
6814 Perl_croak(aTHX_ number_too_long);
6819 /* terminate the string */
6822 /* make an sv from the string */
6825 value = Atof(PL_tokenbuf);
6828 See if we can make do with an integer value without loss of
6829 precision. We use I_V to cast to an int, because some
6830 compilers have issues. Then we try casting it back and see
6831 if it was the same. We only do this if we know we
6832 specifically read an integer.
6834 Note: if floatit is true, then we don't need to do the
6838 if (!floatit && (NV)tryiv == value)
6839 sv_setiv(sv, tryiv);
6841 sv_setnv(sv, value);
6842 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
6843 (PL_hints & HINT_NEW_INTEGER) )
6844 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
6845 (floatit ? "float" : "integer"),
6848 /* if it starts with a v, it could be a version number */
6853 while (isDIGIT(*pos))
6855 if (*pos == '.' && isDIGIT(pos[1])) {
6860 s++; /* get past 'v' */
6863 SvUPGRADE(sv, SVt_PVNV);
6864 sv_setpvn(sv, "", 0);
6869 while (isDIGIT(*pos))
6872 tmpend = uv_to_utf8(tmpbuf, rev);
6874 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
6876 SvNVX(sv) += (NV)rev/nshift;
6878 } while (*pos == '.' && isDIGIT(pos[1]));
6882 tmpend = uv_to_utf8(tmpbuf, rev);
6884 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
6886 SvNVX(sv) += (NV)rev/nshift;
6897 /* make the op for the constant and return */
6900 yylval.opval = newSVOP(OP_CONST, 0, sv);
6902 yylval.opval = Nullop;
6908 S_scan_formline(pTHX_ register char *s)
6913 SV *stuff = newSVpvn("",0);
6914 bool needargs = FALSE;
6917 if (*s == '.' || *s == '}') {
6919 #ifdef PERL_STRICT_CR
6920 for (t = s+1;*t == ' ' || *t == '\t'; t++) ;
6922 for (t = s+1;*t == ' ' || *t == '\t' || *t == '\r'; t++) ;
6924 if (*t == '\n' || t == PL_bufend)
6927 if (PL_in_eval && !PL_rsfp) {
6928 eol = strchr(s,'\n');
6933 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6935 for (t = s; t < eol; t++) {
6936 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
6938 goto enough; /* ~~ must be first line in formline */
6940 if (*t == '@' || *t == '^')
6943 sv_catpvn(stuff, s, eol-s);
6944 #ifndef PERL_STRICT_CR
6945 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
6946 char *end = SvPVX(stuff) + SvCUR(stuff);
6955 s = filter_gets(PL_linestr, PL_rsfp, 0);
6956 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
6957 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
6960 yyerror("Format not terminated");
6970 PL_lex_state = LEX_NORMAL;
6971 PL_nextval[PL_nexttoke].ival = 0;
6975 PL_lex_state = LEX_FORMLINE;
6976 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
6978 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
6982 SvREFCNT_dec(stuff);
6983 PL_lex_formbrack = 0;
6994 PL_cshlen = strlen(PL_cshname);
6999 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7002 I32 oldsavestack_ix = PL_savestack_ix;
7003 CV* outsidecv = PL_compcv;
7007 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7009 SAVEI32(PL_subline);
7010 save_item(PL_subname);
7012 SAVEVPTR(PL_curpad);
7013 SAVESPTR(PL_comppad);
7014 SAVESPTR(PL_comppad_name);
7015 SAVESPTR(PL_compcv);
7016 SAVEI32(PL_comppad_name_fill);
7017 SAVEI32(PL_min_intro_pending);
7018 SAVEI32(PL_max_intro_pending);
7019 SAVEI32(PL_pad_reset_pending);
7021 PL_compcv = (CV*)NEWSV(1104,0);
7022 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7023 CvFLAGS(PL_compcv) |= flags;
7025 PL_comppad = newAV();
7026 av_push(PL_comppad, Nullsv);
7027 PL_curpad = AvARRAY(PL_comppad);
7028 PL_comppad_name = newAV();
7029 PL_comppad_name_fill = 0;
7030 PL_min_intro_pending = 0;
7032 PL_subline = CopLINE(PL_curcop);
7034 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
7035 PL_curpad[0] = (SV*)newAV();
7036 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
7037 #endif /* USE_THREADS */
7039 comppadlist = newAV();
7040 AvREAL_off(comppadlist);
7041 av_store(comppadlist, 0, (SV*)PL_comppad_name);
7042 av_store(comppadlist, 1, (SV*)PL_comppad);
7044 CvPADLIST(PL_compcv) = comppadlist;
7045 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7047 CvOWNER(PL_compcv) = 0;
7048 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7049 MUTEX_INIT(CvMUTEXP(PL_compcv));
7050 #endif /* USE_THREADS */
7052 return oldsavestack_ix;
7056 Perl_yywarn(pTHX_ char *s)
7059 PL_in_eval |= EVAL_WARNONLY;
7061 PL_in_eval &= ~EVAL_WARNONLY;
7066 Perl_yyerror(pTHX_ char *s)
7070 char *context = NULL;
7074 if (!yychar || (yychar == ';' && !PL_rsfp))
7076 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7077 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7078 while (isSPACE(*PL_oldoldbufptr))
7080 context = PL_oldoldbufptr;
7081 contlen = PL_bufptr - PL_oldoldbufptr;
7083 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7084 PL_oldbufptr != PL_bufptr) {
7085 while (isSPACE(*PL_oldbufptr))
7087 context = PL_oldbufptr;
7088 contlen = PL_bufptr - PL_oldbufptr;
7090 else if (yychar > 255)
7091 where = "next token ???";
7092 else if ((yychar & 127) == 127) {
7093 if (PL_lex_state == LEX_NORMAL ||
7094 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7095 where = "at end of line";
7096 else if (PL_lex_inpat)
7097 where = "within pattern";
7099 where = "within string";
7102 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7104 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7105 else if (isPRINT_LC(yychar))
7106 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7108 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7109 where = SvPVX(where_sv);
7111 msg = sv_2mortal(newSVpv(s, 0));
7112 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7113 CopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7115 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7117 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7118 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7119 Perl_sv_catpvf(aTHX_ msg,
7120 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7121 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7124 if (PL_in_eval & EVAL_WARNONLY)
7125 Perl_warn(aTHX_ "%_", msg);
7128 if (PL_error_count >= 10)
7129 Perl_croak(aTHX_ "%s has too many errors.\n", CopFILE(PL_curcop));
7131 PL_in_my_stash = Nullhv;
7142 * Restore a source filter.
7146 restore_rsfp(pTHXo_ void *f)
7148 PerlIO *fp = (PerlIO*)f;
7150 if (PL_rsfp == PerlIO_stdin())
7151 PerlIO_clearerr(PL_rsfp);
7152 else if (PL_rsfp && (PL_rsfp != fp))
7153 PerlIO_close(PL_rsfp);