3 * Copyright (c) 1991-2000, Larry Wall
5 * You may distribute under the terms of either the GNU General Public
6 * License or the Artistic License, as specified in the README file.
11 * "It all comes from here, the stench and the peril." --Frodo
15 * This file is the lexer for Perl. It's closely linked to the
18 * The main routine is yylex(), which returns the next token.
22 #define PERL_IN_TOKE_C
25 #define yychar PL_yychar
26 #define yylval PL_yylval
28 static char ident_too_long[] = "Identifier too long";
30 static void restore_rsfp(pTHXo_ void *f);
31 #ifndef PERL_NO_UTF16_FILTER
32 static I32 utf16_textfilter(pTHXo_ int idx, SV *sv, int maxlen);
33 static I32 utf16rev_textfilter(pTHXo_ int idx, SV *sv, int maxlen);
36 #define XFAKEBRACK 128
39 /*#define UTF (SvUTF8(PL_linestr) && !(PL_hints & HINT_BYTE))*/
40 #define UTF (PL_hints & HINT_UTF8)
42 /* In variables name $^X, these are the legal values for X.
43 * 1999-02-27 mjd-perl-patch@plover.com */
44 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
46 /* On MacOS, respect nonbreaking spaces */
47 #ifdef MACOS_TRADITIONAL
48 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
50 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
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
78 # define YYMAXLEVEL 100
80 YYSTYPE* yylval_pointer[YYMAXLEVEL];
81 int* yychar_pointer[YYMAXLEVEL];
85 # define yylval (*yylval_pointer[yyactlevel])
86 # define yychar (*yychar_pointer[yyactlevel])
87 # define PERL_YYLEX_PARAM yylval_pointer[yyactlevel],yychar_pointer[yyactlevel]
89 # define yylex() Perl_yylex_r(aTHX_ yylval_pointer[yyactlevel],yychar_pointer[yyactlevel])
94 /* CLINE is a macro that ensures PL_copline has a sane value */
99 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
102 * Convenience functions to return different tokens and prime the
103 * lexer for the next token. They all take an argument.
105 * TOKEN : generic token (used for '(', DOLSHARP, etc)
106 * OPERATOR : generic operator
107 * AOPERATOR : assignment operator
108 * PREBLOCK : beginning the block after an if, while, foreach, ...
109 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
110 * PREREF : *EXPR where EXPR is not a simple identifier
111 * TERM : expression term
112 * LOOPX : loop exiting command (goto, last, dump, etc)
113 * FTST : file test operator
114 * FUN0 : zero-argument function
115 * FUN1 : not used, except for not, which isn't a UNIOP
116 * BOop : bitwise or or xor
118 * SHop : shift operator
119 * PWop : power operator
120 * PMop : pattern-matching operator
121 * Aop : addition-level operator
122 * Mop : multiplication-level operator
123 * Eop : equality-testing operator
124 * Rop : relational operator <= != gt
126 * Also see LOP and lop() below.
129 #define TOKEN(retval) return (PL_bufptr = s,(int)retval)
130 #define OPERATOR(retval) return (PL_expect = XTERM,PL_bufptr = s,(int)retval)
131 #define AOPERATOR(retval) return ao((PL_expect = XTERM,PL_bufptr = s,(int)retval))
132 #define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s,(int)retval)
133 #define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s,(int)retval)
134 #define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s,(int)retval)
135 #define TERM(retval) return (CLINE, PL_expect = XOPERATOR,PL_bufptr = s,(int)retval)
136 #define LOOPX(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LOOPEX)
137 #define FTST(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)UNIOP)
138 #define FUN0(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC0)
139 #define FUN1(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC1)
140 #define BOop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITOROP))
141 #define BAop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITANDOP))
142 #define SHop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)SHIFTOP))
143 #define PWop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)POWOP))
144 #define PMop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MATCHOP)
145 #define Aop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)ADDOP))
146 #define Mop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MULOP))
147 #define Eop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)EQOP)
148 #define Rop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)RELOP)
150 /* This bit of chicanery makes a unary function followed by
151 * a parenthesis into a function with one argument, highest precedence.
153 #define UNI(f) return(yylval.ival = f, \
156 PL_last_uni = PL_oldbufptr, \
157 PL_last_lop_op = f, \
158 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
160 #define UNIBRACK(f) return(yylval.ival = f, \
162 PL_last_uni = PL_oldbufptr, \
163 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
165 /* grandfather return to old style */
166 #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
171 * This subroutine detects &&= and ||= and turns an ANDAND or OROR
172 * into an OP_ANDASSIGN or OP_ORASSIGN
176 S_ao(pTHX_ int toketype)
178 if (*PL_bufptr == '=') {
180 if (toketype == ANDAND)
181 yylval.ival = OP_ANDASSIGN;
182 else if (toketype == OROR)
183 yylval.ival = OP_ORASSIGN;
191 * When Perl expects an operator and finds something else, no_op
192 * prints the warning. It always prints "<something> found where
193 * operator expected. It prints "Missing semicolon on previous line?"
194 * if the surprise occurs at the start of the line. "do you need to
195 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
196 * where the compiler doesn't know if foo is a method call or a function.
197 * It prints "Missing operator before end of line" if there's nothing
198 * after the missing operator, or "... before <...>" if there is something
199 * after the missing operator.
203 S_no_op(pTHX_ char *what, char *s)
205 char *oldbp = PL_bufptr;
206 bool is_first = (PL_oldbufptr == PL_linestart);
212 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
214 Perl_warn(aTHX_ "\t(Missing semicolon on previous line?)\n");
215 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
217 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy_if(t,UTF) || *t == ':'); t++) ;
218 if (t < PL_bufptr && isSPACE(*t))
219 Perl_warn(aTHX_ "\t(Do you need to predeclare %.*s?)\n",
220 t - PL_oldoldbufptr, PL_oldoldbufptr);
224 Perl_warn(aTHX_ "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp);
231 * Complain about missing quote/regexp/heredoc terminator.
232 * If it's called with (char *)NULL then it cauterizes the line buffer.
233 * If we're in a delimited string and the delimiter is a control
234 * character, it's reformatted into a two-char sequence like ^C.
239 S_missingterm(pTHX_ char *s)
244 char *nl = strrchr(s,'\n');
250 iscntrl(PL_multi_close)
252 PL_multi_close < 32 || PL_multi_close == 127
256 tmpbuf[1] = toCTRL(PL_multi_close);
262 *tmpbuf = PL_multi_close;
266 q = strchr(s,'"') ? '\'' : '"';
267 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
275 Perl_deprecate(pTHX_ char *s)
277 if (ckWARN(WARN_DEPRECATED))
278 Perl_warner(aTHX_ WARN_DEPRECATED, "Use of %s is deprecated", s);
283 * Deprecate a comma-less variable list.
289 deprecate("comma-less variable list");
293 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
294 * utf16-to-utf8-reversed.
297 #ifdef PERL_CR_FILTER
301 register char *s = SvPVX(sv);
302 register char *e = s + SvCUR(sv);
303 /* outer loop optimized to do nothing if there are no CR-LFs */
305 if (*s++ == '\r' && *s == '\n') {
306 /* hit a CR-LF, need to copy the rest */
307 register char *d = s - 1;
310 if (*s == '\r' && s[1] == '\n')
321 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
323 I32 count = FILTER_READ(idx+1, sv, maxlen);
324 if (count > 0 && !maxlen)
332 * Initialize variables. Uses the Perl save_stack to save its state (for
333 * recursive calls to the parser).
337 Perl_lex_start(pTHX_ SV *line)
342 SAVEI32(PL_lex_dojoin);
343 SAVEI32(PL_lex_brackets);
344 SAVEI32(PL_lex_casemods);
345 SAVEI32(PL_lex_starts);
346 SAVEI32(PL_lex_state);
347 SAVEVPTR(PL_lex_inpat);
348 SAVEI32(PL_lex_inwhat);
349 if (PL_lex_state == LEX_KNOWNEXT) {
350 I32 toke = PL_nexttoke;
351 while (--toke >= 0) {
352 SAVEI32(PL_nexttype[toke]);
353 SAVEVPTR(PL_nextval[toke]);
355 SAVEI32(PL_nexttoke);
357 SAVECOPLINE(PL_curcop);
360 SAVEPPTR(PL_oldbufptr);
361 SAVEPPTR(PL_oldoldbufptr);
362 SAVEPPTR(PL_linestart);
363 SAVESPTR(PL_linestr);
364 SAVEPPTR(PL_lex_brackstack);
365 SAVEPPTR(PL_lex_casestack);
366 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
367 SAVESPTR(PL_lex_stuff);
368 SAVEI32(PL_lex_defer);
369 SAVEI32(PL_sublex_info.sub_inwhat);
370 SAVESPTR(PL_lex_repl);
372 SAVEINT(PL_lex_expect);
374 PL_lex_state = LEX_NORMAL;
378 New(899, PL_lex_brackstack, 120, char);
379 New(899, PL_lex_casestack, 12, char);
380 SAVEFREEPV(PL_lex_brackstack);
381 SAVEFREEPV(PL_lex_casestack);
383 *PL_lex_casestack = '\0';
386 PL_lex_stuff = Nullsv;
387 PL_lex_repl = Nullsv;
391 PL_sublex_info.sub_inwhat = 0;
393 if (SvREADONLY(PL_linestr))
394 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
395 s = SvPV(PL_linestr, len);
396 if (len && s[len-1] != ';') {
397 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
398 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
399 sv_catpvn(PL_linestr, "\n;", 2);
401 SvTEMP_off(PL_linestr);
402 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
403 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
405 PL_rs = newSVpvn("\n", 1);
411 * Finalizer for lexing operations. Must be called when the parser is
412 * done with the lexer.
418 PL_doextract = FALSE;
423 * This subroutine has nothing to do with tilting, whether at windmills
424 * or pinball tables. Its name is short for "increment line". It
425 * increments the current line number in CopLINE(PL_curcop) and checks
426 * to see whether the line starts with a comment of the form
427 * # line 500 "foo.pm"
428 * If so, it sets the current line number and file to the values in the comment.
432 S_incline(pTHX_ char *s)
439 CopLINE_inc(PL_curcop);
442 while (SPACE_OR_TAB(*s)) s++;
443 if (strnEQ(s, "line", 4))
447 if (*s == ' ' || *s == '\t')
451 while (SPACE_OR_TAB(*s)) s++;
457 while (SPACE_OR_TAB(*s))
459 if (*s == '"' && (t = strchr(s+1, '"'))) {
464 for (t = s; !isSPACE(*t); t++) ;
467 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
469 if (*e != '\n' && *e != '\0')
470 return; /* false alarm */
476 Safefree(CopFILE(PL_curcop));
478 SvREFCNT_dec(CopFILEGV(PL_curcop));
480 CopFILE_set(PL_curcop, s);
483 CopLINE_set(PL_curcop, atoi(n)-1);
488 * Called to gobble the appropriate amount and type of whitespace.
489 * Skips comments as well.
493 S_skipspace(pTHX_ register char *s)
495 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
496 while (s < PL_bufend && SPACE_OR_TAB(*s))
502 SSize_t oldprevlen, oldoldprevlen;
503 SSize_t oldloplen, oldunilen;
504 while (s < PL_bufend && isSPACE(*s)) {
505 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
510 if (s < PL_bufend && *s == '#') {
511 while (s < PL_bufend && *s != '\n')
515 if (PL_in_eval && !PL_rsfp) {
522 /* only continue to recharge the buffer if we're at the end
523 * of the buffer, we're not reading from a source filter, and
524 * we're in normal lexing mode
526 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
527 PL_lex_state == LEX_FORMLINE)
530 /* try to recharge the buffer */
531 if ((s = filter_gets(PL_linestr, PL_rsfp,
532 (prevlen = SvCUR(PL_linestr)))) == Nullch)
534 /* end of file. Add on the -p or -n magic */
535 if (PL_minus_n || PL_minus_p) {
536 sv_setpv(PL_linestr,PL_minus_p ?
537 ";}continue{print or die qq(-p destination: $!\\n)" :
539 sv_catpv(PL_linestr,";}");
540 PL_minus_n = PL_minus_p = 0;
543 sv_setpv(PL_linestr,";");
545 /* reset variables for next time we lex */
546 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
548 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
550 /* Close the filehandle. Could be from -P preprocessor,
551 * STDIN, or a regular file. If we were reading code from
552 * STDIN (because the commandline held no -e or filename)
553 * then we don't close it, we reset it so the code can
554 * read from STDIN too.
557 if (PL_preprocess && !PL_in_eval)
558 (void)PerlProc_pclose(PL_rsfp);
559 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
560 PerlIO_clearerr(PL_rsfp);
562 (void)PerlIO_close(PL_rsfp);
567 /* not at end of file, so we only read another line */
568 /* make corresponding updates to old pointers, for yyerror() */
569 oldprevlen = PL_oldbufptr - PL_bufend;
570 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
572 oldunilen = PL_last_uni - PL_bufend;
574 oldloplen = PL_last_lop - PL_bufend;
575 PL_linestart = PL_bufptr = s + prevlen;
576 PL_bufend = s + SvCUR(PL_linestr);
578 PL_oldbufptr = s + oldprevlen;
579 PL_oldoldbufptr = s + oldoldprevlen;
581 PL_last_uni = s + oldunilen;
583 PL_last_lop = s + oldloplen;
586 /* debugger active and we're not compiling the debugger code,
587 * so store the line into the debugger's array of lines
589 if (PERLDB_LINE && PL_curstash != PL_debstash) {
590 SV *sv = NEWSV(85,0);
592 sv_upgrade(sv, SVt_PVMG);
593 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
594 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
601 * Check the unary operators to ensure there's no ambiguity in how they're
602 * used. An ambiguous piece of code would be:
604 * This doesn't mean rand() + 5. Because rand() is a unary operator,
605 * the +5 is its argument.
614 if (PL_oldoldbufptr != PL_last_uni)
616 while (isSPACE(*PL_last_uni))
618 for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ;
619 if ((t = strchr(s, '(')) && t < PL_bufptr)
621 if (ckWARN_d(WARN_AMBIGUOUS)){
624 Perl_warner(aTHX_ WARN_AMBIGUOUS,
625 "Warning: Use of \"%s\" without parens is ambiguous",
631 /* workaround to replace the UNI() macro with a function. Only the
632 * hints/uts.sh file mentions this. Other comments elsewhere in the
633 * source indicate Microport Unix might need it too.
639 #define UNI(f) return uni(f,s)
642 S_uni(pTHX_ I32 f, char *s)
647 PL_last_uni = PL_oldbufptr;
658 #endif /* CRIPPLED_CC */
661 * LOP : macro to build a list operator. Its behaviour has been replaced
662 * with a subroutine, S_lop() for which LOP is just another name.
665 #define LOP(f,x) return lop(f,x,s)
669 * Build a list operator (or something that might be one). The rules:
670 * - if we have a next token, then it's a list operator [why?]
671 * - if the next thing is an opening paren, then it's a function
672 * - else it's a list operator
676 S_lop(pTHX_ I32 f, int x, char *s)
682 PL_last_lop = PL_oldbufptr;
697 * When the lexer realizes it knows the next token (for instance,
698 * it is reordering tokens for the parser) then it can call S_force_next
699 * to know what token to return the next time the lexer is called. Caller
700 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
701 * handles the token correctly.
705 S_force_next(pTHX_ I32 type)
707 PL_nexttype[PL_nexttoke] = type;
709 if (PL_lex_state != LEX_KNOWNEXT) {
710 PL_lex_defer = PL_lex_state;
711 PL_lex_expect = PL_expect;
712 PL_lex_state = LEX_KNOWNEXT;
718 * When the lexer knows the next thing is a word (for instance, it has
719 * just seen -> and it knows that the next char is a word char, then
720 * it calls S_force_word to stick the next word into the PL_next lookahead.
723 * char *start : buffer position (must be within PL_linestr)
724 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
725 * int check_keyword : if true, Perl checks to make sure the word isn't
726 * a keyword (do this if the word is a label, e.g. goto FOO)
727 * int allow_pack : if true, : characters will also be allowed (require,
729 * int allow_initial_tick : used by the "sub" lexer only.
733 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
738 start = skipspace(start);
740 if (isIDFIRST_lazy_if(s,UTF) ||
741 (allow_pack && *s == ':') ||
742 (allow_initial_tick && *s == '\'') )
744 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
745 if (check_keyword && keyword(PL_tokenbuf, len))
747 if (token == METHOD) {
752 PL_expect = XOPERATOR;
755 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0));
756 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
764 * Called when the lexer wants $foo *foo &foo etc, but the program
765 * text only contains the "foo" portion. The first argument is a pointer
766 * to the "foo", and the second argument is the type symbol to prefix.
767 * Forces the next token to be a "WORD".
768 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
772 S_force_ident(pTHX_ register char *s, int kind)
775 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
776 PL_nextval[PL_nexttoke].opval = o;
779 o->op_private = OPpCONST_ENTERED;
780 /* XXX see note in pp_entereval() for why we forgo typo
781 warnings if the symbol must be introduced in an eval.
783 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
784 kind == '$' ? SVt_PV :
785 kind == '@' ? SVt_PVAV :
786 kind == '%' ? SVt_PVHV :
794 Perl_str_to_version(pTHX_ SV *sv)
799 char *start = SvPVx(sv,len);
800 bool utf = SvUTF8(sv) ? TRUE : FALSE;
801 char *end = start + len;
802 while (start < end) {
806 n = utf8_to_uv((U8*)start, len, &skip, 0);
811 retval += ((NV)n)/nshift;
820 * Forces the next token to be a version number.
824 S_force_version(pTHX_ char *s)
826 OP *version = Nullop;
835 for (; isDIGIT(*d) || *d == '_' || *d == '.'; d++);
836 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
838 s = scan_num(s, &yylval);
839 version = yylval.opval;
840 ver = cSVOPx(version)->op_sv;
841 if (SvPOK(ver) && !SvNIOK(ver)) {
842 (void)SvUPGRADE(ver, SVt_PVNV);
843 SvNVX(ver) = str_to_version(ver);
844 SvNOK_on(ver); /* hint that it is a version */
849 /* NOTE: The parser sees the package name and the VERSION swapped */
850 PL_nextval[PL_nexttoke].opval = version;
858 * Tokenize a quoted string passed in as an SV. It finds the next
859 * chunk, up to end of string or a backslash. It may make a new
860 * SV containing that chunk (if HINT_NEW_STRING is on). It also
865 S_tokeq(pTHX_ SV *sv)
876 s = SvPV_force(sv, len);
877 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
880 while (s < send && *s != '\\')
885 if ( PL_hints & HINT_NEW_STRING )
886 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
889 if (s + 1 < send && (s[1] == '\\'))
890 s++; /* all that, just for this */
895 SvCUR_set(sv, d - SvPVX(sv));
897 if ( PL_hints & HINT_NEW_STRING )
898 return new_constant(NULL, 0, "q", sv, pv, "q");
903 * Now come three functions related to double-quote context,
904 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
905 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
906 * interact with PL_lex_state, and create fake ( ... ) argument lists
907 * to handle functions and concatenation.
908 * They assume that whoever calls them will be setting up a fake
909 * join call, because each subthing puts a ',' after it. This lets
912 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
914 * (I'm not sure whether the spurious commas at the end of lcfirst's
915 * arguments and join's arguments are created or not).
920 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
922 * Pattern matching will set PL_lex_op to the pattern-matching op to
923 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
925 * OP_CONST and OP_READLINE are easy--just make the new op and return.
927 * Everything else becomes a FUNC.
929 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
930 * had an OP_CONST or OP_READLINE). This just sets us up for a
931 * call to S_sublex_push().
937 register I32 op_type = yylval.ival;
939 if (op_type == OP_NULL) {
940 yylval.opval = PL_lex_op;
944 if (op_type == OP_CONST || op_type == OP_READLINE) {
945 SV *sv = tokeq(PL_lex_stuff);
947 if (SvTYPE(sv) == SVt_PVIV) {
948 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
954 nsv = newSVpvn(p, len);
960 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
961 PL_lex_stuff = Nullsv;
965 PL_sublex_info.super_state = PL_lex_state;
966 PL_sublex_info.sub_inwhat = op_type;
967 PL_sublex_info.sub_op = PL_lex_op;
968 PL_lex_state = LEX_INTERPPUSH;
972 yylval.opval = PL_lex_op;
982 * Create a new scope to save the lexing state. The scope will be
983 * ended in S_sublex_done. Returns a '(', starting the function arguments
984 * to the uc, lc, etc. found before.
985 * Sets PL_lex_state to LEX_INTERPCONCAT.
993 PL_lex_state = PL_sublex_info.super_state;
994 SAVEI32(PL_lex_dojoin);
995 SAVEI32(PL_lex_brackets);
996 SAVEI32(PL_lex_casemods);
997 SAVEI32(PL_lex_starts);
998 SAVEI32(PL_lex_state);
999 SAVEVPTR(PL_lex_inpat);
1000 SAVEI32(PL_lex_inwhat);
1001 SAVECOPLINE(PL_curcop);
1002 SAVEPPTR(PL_bufptr);
1003 SAVEPPTR(PL_oldbufptr);
1004 SAVEPPTR(PL_oldoldbufptr);
1005 SAVEPPTR(PL_linestart);
1006 SAVESPTR(PL_linestr);
1007 SAVEPPTR(PL_lex_brackstack);
1008 SAVEPPTR(PL_lex_casestack);
1010 PL_linestr = PL_lex_stuff;
1011 PL_lex_stuff = Nullsv;
1013 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1014 = SvPVX(PL_linestr);
1015 PL_bufend += SvCUR(PL_linestr);
1016 SAVEFREESV(PL_linestr);
1018 PL_lex_dojoin = FALSE;
1019 PL_lex_brackets = 0;
1020 New(899, PL_lex_brackstack, 120, char);
1021 New(899, PL_lex_casestack, 12, char);
1022 SAVEFREEPV(PL_lex_brackstack);
1023 SAVEFREEPV(PL_lex_casestack);
1024 PL_lex_casemods = 0;
1025 *PL_lex_casestack = '\0';
1027 PL_lex_state = LEX_INTERPCONCAT;
1028 CopLINE_set(PL_curcop, PL_multi_start);
1030 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1031 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1032 PL_lex_inpat = PL_sublex_info.sub_op;
1034 PL_lex_inpat = Nullop;
1041 * Restores lexer state after a S_sublex_push.
1047 if (!PL_lex_starts++) {
1048 PL_expect = XOPERATOR;
1049 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpvn("",0));
1053 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1054 PL_lex_state = LEX_INTERPCASEMOD;
1058 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1059 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1060 PL_linestr = PL_lex_repl;
1062 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1063 PL_bufend += SvCUR(PL_linestr);
1064 SAVEFREESV(PL_linestr);
1065 PL_lex_dojoin = FALSE;
1066 PL_lex_brackets = 0;
1067 PL_lex_casemods = 0;
1068 *PL_lex_casestack = '\0';
1070 if (SvEVALED(PL_lex_repl)) {
1071 PL_lex_state = LEX_INTERPNORMAL;
1073 /* we don't clear PL_lex_repl here, so that we can check later
1074 whether this is an evalled subst; that means we rely on the
1075 logic to ensure sublex_done() is called again only via the
1076 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1079 PL_lex_state = LEX_INTERPCONCAT;
1080 PL_lex_repl = Nullsv;
1086 PL_bufend = SvPVX(PL_linestr);
1087 PL_bufend += SvCUR(PL_linestr);
1088 PL_expect = XOPERATOR;
1089 PL_sublex_info.sub_inwhat = 0;
1097 Extracts a pattern, double-quoted string, or transliteration. This
1100 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1101 processing a pattern (PL_lex_inpat is true), a transliteration
1102 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1104 Returns a pointer to the character scanned up to. Iff this is
1105 advanced from the start pointer supplied (ie if anything was
1106 successfully parsed), will leave an OP for the substring scanned
1107 in yylval. Caller must intuit reason for not parsing further
1108 by looking at the next characters herself.
1112 double-quoted style: \r and \n
1113 regexp special ones: \D \s
1115 backrefs: \1 (deprecated in substitution replacements)
1116 case and quoting: \U \Q \E
1117 stops on @ and $, but not for $ as tail anchor
1119 In transliterations:
1120 characters are VERY literal, except for - not at the start or end
1121 of the string, which indicates a range. scan_const expands the
1122 range to the full set of intermediate characters.
1124 In double-quoted strings:
1126 double-quoted style: \r and \n
1128 backrefs: \1 (deprecated)
1129 case and quoting: \U \Q \E
1132 scan_const does *not* construct ops to handle interpolated strings.
1133 It stops processing as soon as it finds an embedded $ or @ variable
1134 and leaves it to the caller to work out what's going on.
1136 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @:foo.
1138 $ in pattern could be $foo or could be tail anchor. Assumption:
1139 it's a tail anchor if $ is the last thing in the string, or if it's
1140 followed by one of ")| \n\t"
1142 \1 (backreferences) are turned into $1
1144 The structure of the code is
1145 while (there's a character to process) {
1146 handle transliteration ranges
1147 skip regexp comments
1148 skip # initiated comments in //x patterns
1149 check for embedded @foo
1150 check for embedded scalars
1152 leave intact backslashes from leave (below)
1153 deprecate \1 in strings and sub replacements
1154 handle string-changing backslashes \l \U \Q \E, etc.
1155 switch (what was escaped) {
1156 handle - in a transliteration (becomes a literal -)
1157 handle \132 octal characters
1158 handle 0x15 hex characters
1159 handle \cV (control V)
1160 handle printf backslashes (\f, \r, \n, etc)
1162 } (end if backslash)
1163 } (end while character to read)
1168 S_scan_const(pTHX_ char *start)
1170 register char *send = PL_bufend; /* end of the constant */
1171 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1172 register char *s = start; /* start of the constant */
1173 register char *d = SvPVX(sv); /* destination for copies */
1174 bool dorange = FALSE; /* are we in a translit range? */
1175 bool didrange = FALSE; /* did we just finish a range? */
1176 bool has_utf8 = FALSE; /* embedded \x{} */
1179 I32 utf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1180 ? (PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
1182 I32 this_utf8 = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1183 ? (PL_sublex_info.sub_op->op_private & (PL_lex_repl ?
1184 OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF))
1186 const char *leaveit = /* set of acceptably-backslashed characters */
1188 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
1191 while (s < send || dorange) {
1192 /* get transliterations out of the way (they're most literal) */
1193 if (PL_lex_inwhat == OP_TRANS) {
1194 /* expand a range A-Z to the full set of characters. AIE! */
1196 I32 i; /* current expanded character */
1197 I32 min; /* first character in range */
1198 I32 max; /* last character in range */
1200 i = d - SvPVX(sv); /* remember current offset */
1201 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1202 d = SvPVX(sv) + i; /* refresh d after realloc */
1203 d -= 2; /* eat the first char and the - */
1205 min = (U8)*d; /* first char in range */
1206 max = (U8)d[1]; /* last char in range */
1210 "Invalid [] range \"%c-%c\" in transliteration operator",
1211 (char)min, (char)max);
1215 if ((isLOWER(min) && isLOWER(max)) ||
1216 (isUPPER(min) && isUPPER(max))) {
1218 for (i = min; i <= max; i++)
1222 for (i = min; i <= max; i++)
1229 for (i = min; i <= max; i++)
1232 /* mark the range as done, and continue */
1238 /* range begins (ignore - as first or last char) */
1239 else if (*s == '-' && s+1 < send && s != start) {
1241 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
1244 *d++ = (char)0xff; /* use illegal utf8 byte--see pmtrans */
1256 /* if we get here, we're not doing a transliteration */
1258 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1259 except for the last char, which will be done separately. */
1260 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1262 while (s < send && *s != ')')
1265 else if (s[2] == '{' /* This should match regcomp.c */
1266 || ((s[2] == 'p' || s[2] == '?') && s[3] == '{'))
1269 char *regparse = s + (s[2] == '{' ? 3 : 4);
1272 while (count && (c = *regparse)) {
1273 if (c == '\\' && regparse[1])
1281 if (*regparse != ')') {
1282 regparse--; /* Leave one char for continuation. */
1283 yyerror("Sequence (?{...}) not terminated or not {}-balanced");
1285 while (s < regparse)
1290 /* likewise skip #-initiated comments in //x patterns */
1291 else if (*s == '#' && PL_lex_inpat &&
1292 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1293 while (s+1 < send && *s != '\n')
1297 /* check for embedded arrays
1298 (@foo, @:foo, @'foo, @{foo}, @$foo, @+, @-)
1300 else if (*s == '@' && s[1]
1301 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$+-", s[1])))
1304 /* check for embedded scalars. only stop if we're sure it's a
1307 else if (*s == '$') {
1308 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1310 if (s + 1 < send && !strchr("()| \n\t", s[1]))
1311 break; /* in regexp, $ might be tail anchor */
1314 /* (now in tr/// code again) */
1316 if (*s & 0x80 && (this_utf8 || has_utf8)) {
1317 STRLEN len = (STRLEN) -1;
1320 uv = utf8_to_uv((U8*)s, send - s, &len, UTF8_CHECK_ONLY);
1322 if (len == (STRLEN)-1) {
1323 /* Illegal UTF8 (a high-bit byte), make it valid. */
1324 char *old_pvx = SvPVX(sv);
1325 /* need space for one extra char (NOTE: SvCUR() not set here) */
1326 d = SvGROW(sv, SvLEN(sv) + 1) + (d - old_pvx);
1327 d = (char*)uv_to_utf8((U8*)d, (U8)*s++);
1338 if (*s == '\\' && s+1 < send) {
1341 /* some backslashes we leave behind */
1342 if (*leaveit && *s && strchr(leaveit, *s)) {
1348 /* deprecate \1 in strings and substitution replacements */
1349 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1350 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1352 if (ckWARN(WARN_SYNTAX))
1353 Perl_warner(aTHX_ WARN_SYNTAX, "\\%c better written as $%c", *s, *s);
1358 /* string-change backslash escapes */
1359 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1364 /* if we get here, it's either a quoted -, or a digit */
1367 /* quoted - in transliterations */
1369 if (PL_lex_inwhat == OP_TRANS) {
1376 if (ckWARN(WARN_MISC) && isALNUM(*s))
1377 Perl_warner(aTHX_ WARN_MISC,
1378 "Unrecognized escape \\%c passed through",
1380 /* default action is to copy the quoted character */
1385 /* \132 indicates an octal constant */
1386 case '0': case '1': case '2': case '3':
1387 case '4': case '5': case '6': case '7':
1389 STRLEN len = 0; /* disallow underscores */
1390 uv = (UV)scan_oct(s, 3, &len);
1393 goto NUM_ESCAPE_INSERT;
1395 /* \x24 indicates a hex constant */
1399 char* e = strchr(s, '}');
1401 yyerror("Missing right brace on \\x{}");
1405 STRLEN len = 1; /* allow underscores */
1406 uv = (UV)scan_hex(s + 1, e - s - 1, &len);
1413 STRLEN len = 0; /* disallow underscores */
1414 uv = (UV)scan_hex(s, 2, &len);
1420 /* Insert oct or hex escaped character.
1421 * There will always enough room in sv since such escapes will
1422 * be longer than any utf8 sequence they can end up as
1424 if (uv > 127 || has_utf8) {
1425 if (!this_utf8 && !has_utf8 && uv > 255) {
1426 /* might need to recode whatever we have accumulated so far
1427 * if it contains any hibit chars
1431 for (c = SvPVX(sv); c < d; c++) {
1436 char *old_pvx = SvPVX(sv);
1438 d = SvGROW(sv, SvCUR(sv) + hicount + 1) + (d - old_pvx);
1447 uv_to_utf8((U8*)dst, (U8)*src--);
1457 if (has_utf8 || uv > 255) {
1458 d = (char*)uv_to_utf8((U8*)d, uv);
1470 /* \N{latin small letter a} is a named character */
1474 char* e = strchr(s, '}');
1480 yyerror("Missing right brace on \\N{}");
1484 res = newSVpvn(s + 1, e - s - 1);
1485 res = new_constant( Nullch, 0, "charnames",
1486 res, Nullsv, "\\N{...}" );
1487 str = SvPV(res,len);
1488 if (!has_utf8 && SvUTF8(res)) {
1489 char *ostart = SvPVX(sv);
1490 SvCUR_set(sv, d - ostart);
1493 sv_utf8_upgrade(sv);
1494 /* this just broke our allocation above... */
1495 SvGROW(sv, send - start);
1496 d = SvPVX(sv) + SvCUR(sv);
1499 if (len > e - s + 4) {
1500 char *odest = SvPVX(sv);
1502 SvGROW(sv, (SvCUR(sv) + len - (e - s + 4)));
1503 d = SvPVX(sv) + (d - odest);
1505 Copy(str, d, len, char);
1512 yyerror("Missing braces on \\N{}");
1515 /* \c is a control character */
1532 /* printf-style backslashes, formfeeds, newlines, etc */
1550 *d++ = '\047'; /* CP 1047 */
1553 *d++ = '\057'; /* CP 1047 */
1567 } /* end if (backslash) */
1570 } /* while loop to process each character */
1572 /* terminate the string and set up the sv */
1574 SvCUR_set(sv, d - SvPVX(sv));
1579 /* shrink the sv if we allocated more than we used */
1580 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1581 SvLEN_set(sv, SvCUR(sv) + 1);
1582 Renew(SvPVX(sv), SvLEN(sv), char);
1585 /* return the substring (via yylval) only if we parsed anything */
1586 if (s > PL_bufptr) {
1587 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1588 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1590 ( PL_lex_inwhat == OP_TRANS
1592 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1595 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1602 * Returns TRUE if there's more to the expression (e.g., a subscript),
1605 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1607 * ->[ and ->{ return TRUE
1608 * { and [ outside a pattern are always subscripts, so return TRUE
1609 * if we're outside a pattern and it's not { or [, then return FALSE
1610 * if we're in a pattern and the first char is a {
1611 * {4,5} (any digits around the comma) returns FALSE
1612 * if we're in a pattern and the first char is a [
1614 * [SOMETHING] has a funky algorithm to decide whether it's a
1615 * character class or not. It has to deal with things like
1616 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1617 * anything else returns TRUE
1620 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1623 S_intuit_more(pTHX_ register char *s)
1625 if (PL_lex_brackets)
1627 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1629 if (*s != '{' && *s != '[')
1634 /* In a pattern, so maybe we have {n,m}. */
1651 /* On the other hand, maybe we have a character class */
1654 if (*s == ']' || *s == '^')
1657 /* this is terrifying, and it works */
1658 int weight = 2; /* let's weigh the evidence */
1660 unsigned char un_char = 255, last_un_char;
1661 char *send = strchr(s,']');
1662 char tmpbuf[sizeof PL_tokenbuf * 4];
1664 if (!send) /* has to be an expression */
1667 Zero(seen,256,char);
1670 else if (isDIGIT(*s)) {
1672 if (isDIGIT(s[1]) && s[2] == ']')
1678 for (; s < send; s++) {
1679 last_un_char = un_char;
1680 un_char = (unsigned char)*s;
1685 weight -= seen[un_char] * 10;
1686 if (isALNUM_lazy_if(s+1,UTF)) {
1687 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1688 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1693 else if (*s == '$' && s[1] &&
1694 strchr("[#!%*<>()-=",s[1])) {
1695 if (/*{*/ strchr("])} =",s[2]))
1704 if (strchr("wds]",s[1]))
1706 else if (seen['\''] || seen['"'])
1708 else if (strchr("rnftbxcav",s[1]))
1710 else if (isDIGIT(s[1])) {
1712 while (s[1] && isDIGIT(s[1]))
1722 if (strchr("aA01! ",last_un_char))
1724 if (strchr("zZ79~",s[1]))
1726 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1727 weight -= 5; /* cope with negative subscript */
1730 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1731 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1736 if (keyword(tmpbuf, d - tmpbuf))
1739 if (un_char == last_un_char + 1)
1741 weight -= seen[un_char];
1746 if (weight >= 0) /* probably a character class */
1756 * Does all the checking to disambiguate
1758 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1759 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1761 * First argument is the stuff after the first token, e.g. "bar".
1763 * Not a method if bar is a filehandle.
1764 * Not a method if foo is a subroutine prototyped to take a filehandle.
1765 * Not a method if it's really "Foo $bar"
1766 * Method if it's "foo $bar"
1767 * Not a method if it's really "print foo $bar"
1768 * Method if it's really "foo package::" (interpreted as package->foo)
1769 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1770 * Not a method if bar is a filehandle or package, but is quoted with
1775 S_intuit_method(pTHX_ char *start, GV *gv)
1777 char *s = start + (*start == '$');
1778 char tmpbuf[sizeof PL_tokenbuf];
1786 if ((cv = GvCVu(gv))) {
1787 char *proto = SvPVX(cv);
1797 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1798 /* start is the beginning of the possible filehandle/object,
1799 * and s is the end of it
1800 * tmpbuf is a copy of it
1803 if (*start == '$') {
1804 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1809 return *s == '(' ? FUNCMETH : METHOD;
1811 if (!keyword(tmpbuf, len)) {
1812 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1817 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1818 if (indirgv && GvCVu(indirgv))
1820 /* filehandle or package name makes it a method */
1821 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1823 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1824 return 0; /* no assumptions -- "=>" quotes bearword */
1826 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1827 newSVpvn(tmpbuf,len));
1828 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1832 return *s == '(' ? FUNCMETH : METHOD;
1840 * Return a string of Perl code to load the debugger. If PERL5DB
1841 * is set, it will return the contents of that, otherwise a
1842 * compile-time require of perl5db.pl.
1849 char *pdb = PerlEnv_getenv("PERL5DB");
1853 SETERRNO(0,SS$_NORMAL);
1854 return "BEGIN { require 'perl5db.pl' }";
1860 /* Encoded script support. filter_add() effectively inserts a
1861 * 'pre-processing' function into the current source input stream.
1862 * Note that the filter function only applies to the current source file
1863 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1865 * The datasv parameter (which may be NULL) can be used to pass
1866 * private data to this instance of the filter. The filter function
1867 * can recover the SV using the FILTER_DATA macro and use it to
1868 * store private buffers and state information.
1870 * The supplied datasv parameter is upgraded to a PVIO type
1871 * and the IoDIRP/IoANY field is used to store the function pointer,
1872 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1873 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1874 * private use must be set using malloc'd pointers.
1878 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1883 if (!PL_rsfp_filters)
1884 PL_rsfp_filters = newAV();
1886 datasv = NEWSV(255,0);
1887 if (!SvUPGRADE(datasv, SVt_PVIO))
1888 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1889 IoANY(datasv) = (void *)funcp; /* stash funcp into spare field */
1890 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
1891 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1892 funcp, SvPV_nolen(datasv)));
1893 av_unshift(PL_rsfp_filters, 1);
1894 av_store(PL_rsfp_filters, 0, datasv) ;
1899 /* Delete most recently added instance of this filter function. */
1901 Perl_filter_del(pTHX_ filter_t funcp)
1904 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp));
1905 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
1907 /* if filter is on top of stack (usual case) just pop it off */
1908 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
1909 if (IoANY(datasv) == (void *)funcp) {
1910 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
1911 IoANY(datasv) = (void *)NULL;
1912 sv_free(av_pop(PL_rsfp_filters));
1916 /* we need to search for the correct entry and clear it */
1917 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
1921 /* Invoke the n'th filter function for the current rsfp. */
1923 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
1926 /* 0 = read one text line */
1931 if (!PL_rsfp_filters)
1933 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
1934 /* Provide a default input filter to make life easy. */
1935 /* Note that we append to the line. This is handy. */
1936 DEBUG_P(PerlIO_printf(Perl_debug_log,
1937 "filter_read %d: from rsfp\n", idx));
1941 int old_len = SvCUR(buf_sv) ;
1943 /* ensure buf_sv is large enough */
1944 SvGROW(buf_sv, old_len + maxlen) ;
1945 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
1946 if (PerlIO_error(PL_rsfp))
1947 return -1; /* error */
1949 return 0 ; /* end of file */
1951 SvCUR_set(buf_sv, old_len + len) ;
1954 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
1955 if (PerlIO_error(PL_rsfp))
1956 return -1; /* error */
1958 return 0 ; /* end of file */
1961 return SvCUR(buf_sv);
1963 /* Skip this filter slot if filter has been deleted */
1964 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
1965 DEBUG_P(PerlIO_printf(Perl_debug_log,
1966 "filter_read %d: skipped (filter deleted)\n",
1968 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
1970 /* Get function pointer hidden within datasv */
1971 funcp = (filter_t)IoANY(datasv);
1972 DEBUG_P(PerlIO_printf(Perl_debug_log,
1973 "filter_read %d: via function %p (%s)\n",
1974 idx, funcp, SvPV_nolen(datasv)));
1975 /* Call function. The function is expected to */
1976 /* call "FILTER_READ(idx+1, buf_sv)" first. */
1977 /* Return: <0:error, =0:eof, >0:not eof */
1978 return (*funcp)(aTHXo_ idx, buf_sv, maxlen);
1982 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
1984 #ifdef PERL_CR_FILTER
1985 if (!PL_rsfp_filters) {
1986 filter_add(S_cr_textfilter,NULL);
1989 if (PL_rsfp_filters) {
1992 SvCUR_set(sv, 0); /* start with empty line */
1993 if (FILTER_READ(0, sv, 0) > 0)
1994 return ( SvPVX(sv) ) ;
1999 return (sv_gets(sv, fp, append));
2003 S_find_in_my_stash(pTHX_ char *pkgname, I32 len)
2007 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2011 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2012 (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV)))
2014 return GvHV(gv); /* Foo:: */
2017 /* use constant CLASS => 'MyClass' */
2018 if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) {
2020 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2021 pkgname = SvPV_nolen(sv);
2025 return gv_stashpv(pkgname, FALSE);
2029 static char* exp_name[] =
2030 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2031 "ATTRTERM", "TERMBLOCK"
2038 Works out what to call the token just pulled out of the input
2039 stream. The yacc parser takes care of taking the ops we return and
2040 stitching them into a tree.
2046 if read an identifier
2047 if we're in a my declaration
2048 croak if they tried to say my($foo::bar)
2049 build the ops for a my() declaration
2050 if it's an access to a my() variable
2051 are we in a sort block?
2052 croak if my($a); $a <=> $b
2053 build ops for access to a my() variable
2054 if in a dq string, and they've said @foo and we can't find @foo
2056 build ops for a bareword
2057 if we already built the token before, use it.
2060 #ifdef USE_PURE_BISON
2062 #pragma segment Perl_yylex_r
2065 Perl_yylex_r(pTHX_ YYSTYPE *lvalp, int *lcharp)
2069 yylval_pointer[yyactlevel] = lvalp;
2070 yychar_pointer[yyactlevel] = lcharp;
2072 if (yyactlevel >= YYMAXLEVEL)
2073 Perl_croak(aTHX_ "panic: YYMAXLEVEL");
2075 r = Perl_yylex(aTHX);
2084 #pragma segment Perl_yylex
2088 #ifdef USE_PURE_BISON
2089 Perl_yylex(pTHX_ YYSTYPE *lvalp, int *lcharp)
2101 /* check if there's an identifier for us to look at */
2102 if (PL_pending_ident) {
2103 /* pit holds the identifier we read and pending_ident is reset */
2104 char pit = PL_pending_ident;
2105 PL_pending_ident = 0;
2107 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2108 "### Tokener saw identifier '%s'\n", PL_tokenbuf); })
2110 /* if we're in a my(), we can't allow dynamics here.
2111 $foo'bar has already been turned into $foo::bar, so
2112 just check for colons.
2114 if it's a legal name, the OP is a PADANY.
2117 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
2118 if (strchr(PL_tokenbuf,':'))
2119 yyerror(Perl_form(aTHX_ "No package name allowed for "
2120 "variable %s in \"our\"",
2122 tmp = pad_allocmy(PL_tokenbuf);
2125 if (strchr(PL_tokenbuf,':'))
2126 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
2128 yylval.opval = newOP(OP_PADANY, 0);
2129 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
2135 build the ops for accesses to a my() variable.
2137 Deny my($a) or my($b) in a sort block, *if* $a or $b is
2138 then used in a comparison. This catches most, but not
2139 all cases. For instance, it catches
2140 sort { my($a); $a <=> $b }
2142 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
2143 (although why you'd do that is anyone's guess).
2146 if (!strchr(PL_tokenbuf,':')) {
2148 /* Check for single character per-thread SVs */
2149 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
2150 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
2151 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
2153 yylval.opval = newOP(OP_THREADSV, 0);
2154 yylval.opval->op_targ = tmp;
2157 #endif /* USE_THREADS */
2158 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
2159 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
2160 /* might be an "our" variable" */
2161 if (SvFLAGS(namesv) & SVpad_OUR) {
2162 /* build ops for a bareword */
2163 SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0);
2164 sv_catpvn(sym, "::", 2);
2165 sv_catpv(sym, PL_tokenbuf+1);
2166 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
2167 yylval.opval->op_private = OPpCONST_ENTERED;
2168 gv_fetchpv(SvPVX(sym),
2170 ? (GV_ADDMULTI | GV_ADDINEVAL)
2173 ((PL_tokenbuf[0] == '$') ? SVt_PV
2174 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2179 /* if it's a sort block and they're naming $a or $b */
2180 if (PL_last_lop_op == OP_SORT &&
2181 PL_tokenbuf[0] == '$' &&
2182 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
2185 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
2186 d < PL_bufend && *d != '\n';
2189 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
2190 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
2196 yylval.opval = newOP(OP_PADANY, 0);
2197 yylval.opval->op_targ = tmp;
2203 Whine if they've said @foo in a doublequoted string,
2204 and @foo isn't a variable we can find in the symbol
2207 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
2208 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
2209 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
2210 && ckWARN(WARN_AMBIGUOUS))
2212 /* Downgraded from fatal to warning 20000522 mjd */
2213 Perl_warner(aTHX_ WARN_AMBIGUOUS,
2214 "Possible unintended interpolation of %s in string",
2219 /* build ops for a bareword */
2220 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2221 yylval.opval->op_private = OPpCONST_ENTERED;
2222 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
2223 ((PL_tokenbuf[0] == '$') ? SVt_PV
2224 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2229 /* no identifier pending identification */
2231 switch (PL_lex_state) {
2233 case LEX_NORMAL: /* Some compilers will produce faster */
2234 case LEX_INTERPNORMAL: /* code if we comment these out. */
2238 /* when we've already built the next token, just pull it out of the queue */
2241 yylval = PL_nextval[PL_nexttoke];
2243 PL_lex_state = PL_lex_defer;
2244 PL_expect = PL_lex_expect;
2245 PL_lex_defer = LEX_NORMAL;
2247 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2248 "### Next token after '%s' was known, type %"IVdf"\n", PL_bufptr,
2249 (IV)PL_nexttype[PL_nexttoke]); })
2251 return(PL_nexttype[PL_nexttoke]);
2253 /* interpolated case modifiers like \L \U, including \Q and \E.
2254 when we get here, PL_bufptr is at the \
2256 case LEX_INTERPCASEMOD:
2258 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2259 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2261 /* handle \E or end of string */
2262 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2266 if (PL_lex_casemods) {
2267 oldmod = PL_lex_casestack[--PL_lex_casemods];
2268 PL_lex_casestack[PL_lex_casemods] = '\0';
2270 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2272 PL_lex_state = LEX_INTERPCONCAT;
2276 if (PL_bufptr != PL_bufend)
2278 PL_lex_state = LEX_INTERPCONCAT;
2282 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2283 "### Saw case modifier at '%s'\n", PL_bufptr); })
2285 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2286 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2287 if (strchr("LU", *s) &&
2288 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2290 PL_lex_casestack[--PL_lex_casemods] = '\0';
2293 if (PL_lex_casemods > 10) {
2294 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2295 if (newlb != PL_lex_casestack) {
2297 PL_lex_casestack = newlb;
2300 PL_lex_casestack[PL_lex_casemods++] = *s;
2301 PL_lex_casestack[PL_lex_casemods] = '\0';
2302 PL_lex_state = LEX_INTERPCONCAT;
2303 PL_nextval[PL_nexttoke].ival = 0;
2306 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2308 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2310 PL_nextval[PL_nexttoke].ival = OP_LC;
2312 PL_nextval[PL_nexttoke].ival = OP_UC;
2314 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2316 Perl_croak(aTHX_ "panic: yylex");
2319 if (PL_lex_starts) {
2328 case LEX_INTERPPUSH:
2329 return sublex_push();
2331 case LEX_INTERPSTART:
2332 if (PL_bufptr == PL_bufend)
2333 return sublex_done();
2334 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2335 "### Interpolated variable at '%s'\n", PL_bufptr); })
2337 PL_lex_dojoin = (*PL_bufptr == '@');
2338 PL_lex_state = LEX_INTERPNORMAL;
2339 if (PL_lex_dojoin) {
2340 PL_nextval[PL_nexttoke].ival = 0;
2343 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2344 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2345 force_next(PRIVATEREF);
2347 force_ident("\"", '$');
2348 #endif /* USE_THREADS */
2349 PL_nextval[PL_nexttoke].ival = 0;
2351 PL_nextval[PL_nexttoke].ival = 0;
2353 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2356 if (PL_lex_starts++) {
2362 case LEX_INTERPENDMAYBE:
2363 if (intuit_more(PL_bufptr)) {
2364 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2370 if (PL_lex_dojoin) {
2371 PL_lex_dojoin = FALSE;
2372 PL_lex_state = LEX_INTERPCONCAT;
2375 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2376 && SvEVALED(PL_lex_repl))
2378 if (PL_bufptr != PL_bufend)
2379 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2380 PL_lex_repl = Nullsv;
2383 case LEX_INTERPCONCAT:
2385 if (PL_lex_brackets)
2386 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2388 if (PL_bufptr == PL_bufend)
2389 return sublex_done();
2391 if (SvIVX(PL_linestr) == '\'') {
2392 SV *sv = newSVsv(PL_linestr);
2395 else if ( PL_hints & HINT_NEW_RE )
2396 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2397 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2401 s = scan_const(PL_bufptr);
2403 PL_lex_state = LEX_INTERPCASEMOD;
2405 PL_lex_state = LEX_INTERPSTART;
2408 if (s != PL_bufptr) {
2409 PL_nextval[PL_nexttoke] = yylval;
2412 if (PL_lex_starts++)
2422 PL_lex_state = LEX_NORMAL;
2423 s = scan_formline(PL_bufptr);
2424 if (!PL_lex_formbrack)
2430 PL_oldoldbufptr = PL_oldbufptr;
2433 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2434 exp_name[PL_expect], s);
2440 if (isIDFIRST_lazy_if(s,UTF))
2442 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2445 goto fake_eof; /* emulate EOF on ^D or ^Z */
2450 if (PL_lex_brackets)
2451 yyerror("Missing right curly or square bracket");
2452 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2453 "### Tokener got EOF\n");
2457 if (s++ < PL_bufend)
2458 goto retry; /* ignore stray nulls */
2461 if (!PL_in_eval && !PL_preambled) {
2462 PL_preambled = TRUE;
2463 sv_setpv(PL_linestr,incl_perldb());
2464 if (SvCUR(PL_linestr))
2465 sv_catpv(PL_linestr,";");
2467 while(AvFILLp(PL_preambleav) >= 0) {
2468 SV *tmpsv = av_shift(PL_preambleav);
2469 sv_catsv(PL_linestr, tmpsv);
2470 sv_catpv(PL_linestr, ";");
2473 sv_free((SV*)PL_preambleav);
2474 PL_preambleav = NULL;
2476 if (PL_minus_n || PL_minus_p) {
2477 sv_catpv(PL_linestr, "LINE: while (<>) {");
2479 sv_catpv(PL_linestr,"chomp;");
2481 GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
2483 GvIMPORTED_AV_on(gv);
2485 if (strchr("/'\"", *PL_splitstr)
2486 && strchr(PL_splitstr + 1, *PL_splitstr))
2487 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2490 s = "'~#\200\1'"; /* surely one char is unused...*/
2491 while (s[1] && strchr(PL_splitstr, *s)) s++;
2493 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c",
2494 "q" + (delim == '\''), delim);
2495 for (s = PL_splitstr; *s; s++) {
2497 sv_catpvn(PL_linestr, "\\", 1);
2498 sv_catpvn(PL_linestr, s, 1);
2500 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2504 sv_catpv(PL_linestr,"@F=split(' ');");
2507 sv_catpv(PL_linestr, "\n");
2508 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2509 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2510 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2511 SV *sv = NEWSV(85,0);
2513 sv_upgrade(sv, SVt_PVMG);
2514 sv_setsv(sv,PL_linestr);
2515 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2520 bool bof = PL_rsfp ? TRUE : FALSE;
2522 #ifdef PERLIO_IS_STDIO
2523 # ifdef __GNU_LIBRARY__
2524 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
2525 # define FTELL_FOR_PIPE_IS_BROKEN
2529 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
2530 # define FTELL_FOR_PIPE_IS_BROKEN
2535 #ifdef FTELL_FOR_PIPE_IS_BROKEN
2536 /* This loses the possibility to detect the bof
2537 * situation on perl -P when the libc5 is being used.
2538 * Workaround? Maybe attach some extra state to PL_rsfp?
2541 bof = PerlIO_tell(PL_rsfp) == 0;
2543 bof = PerlIO_tell(PL_rsfp) == 0;
2546 s = filter_gets(PL_linestr, PL_rsfp, 0);
2550 if (PL_preprocess && !PL_in_eval)
2551 (void)PerlProc_pclose(PL_rsfp);
2552 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2553 PerlIO_clearerr(PL_rsfp);
2555 (void)PerlIO_close(PL_rsfp);
2557 PL_doextract = FALSE;
2559 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2560 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2561 sv_catpv(PL_linestr,";}");
2562 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2563 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2564 PL_minus_n = PL_minus_p = 0;
2567 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2568 sv_setpv(PL_linestr,"");
2569 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2571 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2572 s = swallow_bom((U8*)s);
2575 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2576 PL_doextract = FALSE;
2578 /* Incest with pod. */
2579 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2580 sv_setpv(PL_linestr, "");
2581 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2582 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2583 PL_doextract = FALSE;
2587 } while (PL_doextract);
2588 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2589 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2590 SV *sv = NEWSV(85,0);
2592 sv_upgrade(sv, SVt_PVMG);
2593 sv_setsv(sv,PL_linestr);
2594 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2596 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2597 if (CopLINE(PL_curcop) == 1) {
2598 while (s < PL_bufend && isSPACE(*s))
2600 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2604 if (*s == '#' && *(s+1) == '!')
2606 #ifdef ALTERNATE_SHEBANG
2608 static char as[] = ALTERNATE_SHEBANG;
2609 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2610 d = s + (sizeof(as) - 1);
2612 #endif /* ALTERNATE_SHEBANG */
2621 while (*d && !isSPACE(*d))
2625 #ifdef ARG_ZERO_IS_SCRIPT
2626 if (ipathend > ipath) {
2628 * HP-UX (at least) sets argv[0] to the script name,
2629 * which makes $^X incorrect. And Digital UNIX and Linux,
2630 * at least, set argv[0] to the basename of the Perl
2631 * interpreter. So, having found "#!", we'll set it right.
2633 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2634 assert(SvPOK(x) || SvGMAGICAL(x));
2635 if (sv_eq(x, CopFILESV(PL_curcop))) {
2636 sv_setpvn(x, ipath, ipathend - ipath);
2639 TAINT_NOT; /* $^X is always tainted, but that's OK */
2641 #endif /* ARG_ZERO_IS_SCRIPT */
2646 d = instr(s,"perl -");
2648 d = instr(s,"perl");
2650 /* avoid getting into infinite loops when shebang
2651 * line contains "Perl" rather than "perl" */
2653 for (d = ipathend-4; d >= ipath; --d) {
2654 if ((*d == 'p' || *d == 'P')
2655 && !ibcmp(d, "perl", 4))
2665 #ifdef ALTERNATE_SHEBANG
2667 * If the ALTERNATE_SHEBANG on this system starts with a
2668 * character that can be part of a Perl expression, then if
2669 * we see it but not "perl", we're probably looking at the
2670 * start of Perl code, not a request to hand off to some
2671 * other interpreter. Similarly, if "perl" is there, but
2672 * not in the first 'word' of the line, we assume the line
2673 * contains the start of the Perl program.
2675 if (d && *s != '#') {
2677 while (*c && !strchr("; \t\r\n\f\v#", *c))
2680 d = Nullch; /* "perl" not in first word; ignore */
2682 *s = '#'; /* Don't try to parse shebang line */
2684 #endif /* ALTERNATE_SHEBANG */
2685 #ifndef MACOS_TRADITIONAL
2690 !instr(s,"indir") &&
2691 instr(PL_origargv[0],"perl"))
2697 while (s < PL_bufend && isSPACE(*s))
2699 if (s < PL_bufend) {
2700 Newz(899,newargv,PL_origargc+3,char*);
2702 while (s < PL_bufend && !isSPACE(*s))
2705 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2708 newargv = PL_origargv;
2710 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
2711 Perl_croak(aTHX_ "Can't exec %s", ipath);
2715 U32 oldpdb = PL_perldb;
2716 bool oldn = PL_minus_n;
2717 bool oldp = PL_minus_p;
2719 while (*d && !isSPACE(*d)) d++;
2720 while (SPACE_OR_TAB(*d)) d++;
2724 if (*d == 'M' || *d == 'm') {
2726 while (*d && !isSPACE(*d)) d++;
2727 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2730 d = moreswitches(d);
2732 if ((PERLDB_LINE && !oldpdb) ||
2733 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2734 /* if we have already added "LINE: while (<>) {",
2735 we must not do it again */
2737 sv_setpv(PL_linestr, "");
2738 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2739 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2740 PL_preambled = FALSE;
2742 (void)gv_fetchfile(PL_origfilename);
2749 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2751 PL_lex_state = LEX_FORMLINE;
2756 #ifdef PERL_STRICT_CR
2757 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2759 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
2761 case ' ': case '\t': case '\f': case 013:
2762 #ifdef MACOS_TRADITIONAL
2769 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2770 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2771 /* handle eval qq[#line 1 "foo"\n ...] */
2772 CopLINE_dec(PL_curcop);
2776 while (s < d && *s != '\n')
2781 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2783 PL_lex_state = LEX_FORMLINE;
2793 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2800 while (s < PL_bufend && SPACE_OR_TAB(*s))
2803 if (strnEQ(s,"=>",2)) {
2804 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2805 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2806 "### Saw unary minus before =>, forcing word '%s'\n", s);
2808 OPERATOR('-'); /* unary minus */
2810 PL_last_uni = PL_oldbufptr;
2812 case 'r': ftst = OP_FTEREAD; break;
2813 case 'w': ftst = OP_FTEWRITE; break;
2814 case 'x': ftst = OP_FTEEXEC; break;
2815 case 'o': ftst = OP_FTEOWNED; break;
2816 case 'R': ftst = OP_FTRREAD; break;
2817 case 'W': ftst = OP_FTRWRITE; break;
2818 case 'X': ftst = OP_FTREXEC; break;
2819 case 'O': ftst = OP_FTROWNED; break;
2820 case 'e': ftst = OP_FTIS; break;
2821 case 'z': ftst = OP_FTZERO; break;
2822 case 's': ftst = OP_FTSIZE; break;
2823 case 'f': ftst = OP_FTFILE; break;
2824 case 'd': ftst = OP_FTDIR; break;
2825 case 'l': ftst = OP_FTLINK; break;
2826 case 'p': ftst = OP_FTPIPE; break;
2827 case 'S': ftst = OP_FTSOCK; break;
2828 case 'u': ftst = OP_FTSUID; break;
2829 case 'g': ftst = OP_FTSGID; break;
2830 case 'k': ftst = OP_FTSVTX; break;
2831 case 'b': ftst = OP_FTBLK; break;
2832 case 'c': ftst = OP_FTCHR; break;
2833 case 't': ftst = OP_FTTTY; break;
2834 case 'T': ftst = OP_FTTEXT; break;
2835 case 'B': ftst = OP_FTBINARY; break;
2836 case 'M': case 'A': case 'C':
2837 gv_fetchpv("\024",TRUE, SVt_PV);
2839 case 'M': ftst = OP_FTMTIME; break;
2840 case 'A': ftst = OP_FTATIME; break;
2841 case 'C': ftst = OP_FTCTIME; break;
2849 PL_last_lop_op = ftst;
2850 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2851 "### Saw file test %c\n", ftst);
2853 if (*s == '(' && ckWARN(WARN_AMBIGUOUS))
2854 Perl_warner(aTHX_ WARN_AMBIGUOUS,
2855 "Ambiguous -%c() resolved as a file test",
2860 /* Assume it was a minus followed by a one-letter named
2861 * subroutine call (or a -bareword), then. */
2868 if (PL_expect == XOPERATOR)
2873 else if (*s == '>') {
2876 if (isIDFIRST_lazy_if(s,UTF)) {
2877 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2885 if (PL_expect == XOPERATOR)
2888 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2890 OPERATOR('-'); /* unary minus */
2897 if (PL_expect == XOPERATOR)
2902 if (PL_expect == XOPERATOR)
2905 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2911 if (PL_expect != XOPERATOR) {
2912 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2913 PL_expect = XOPERATOR;
2914 force_ident(PL_tokenbuf, '*');
2927 if (PL_expect == XOPERATOR) {
2931 PL_tokenbuf[0] = '%';
2932 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2933 if (!PL_tokenbuf[1]) {
2935 yyerror("Final % should be \\% or %name");
2938 PL_pending_ident = '%';
2957 switch (PL_expect) {
2960 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2962 PL_bufptr = s; /* update in case we back off */
2968 PL_expect = XTERMBLOCK;
2972 while (isIDFIRST_lazy_if(s,UTF)) {
2973 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2974 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2975 if (tmp < 0) tmp = -tmp;
2990 d = scan_str(d,TRUE,TRUE);
2993 SvREFCNT_dec(PL_lex_stuff);
2994 PL_lex_stuff = Nullsv;
2996 /* MUST advance bufptr here to avoid bogus
2997 "at end of line" context messages from yyerror().
2999 PL_bufptr = s + len;
3000 yyerror("Unterminated attribute parameter in attribute list");
3003 return 0; /* EOF indicator */
3007 SV *sv = newSVpvn(s, len);
3008 sv_catsv(sv, PL_lex_stuff);
3009 attrs = append_elem(OP_LIST, attrs,
3010 newSVOP(OP_CONST, 0, sv));
3011 SvREFCNT_dec(PL_lex_stuff);
3012 PL_lex_stuff = Nullsv;
3015 attrs = append_elem(OP_LIST, attrs,
3016 newSVOP(OP_CONST, 0,
3020 if (*s == ':' && s[1] != ':')
3023 break; /* require real whitespace or :'s */
3025 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
3026 if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) {
3027 char q = ((*s == '\'') ? '"' : '\'');
3028 /* If here for an expression, and parsed no attrs, back off. */
3029 if (tmp == '=' && !attrs) {
3033 /* MUST advance bufptr here to avoid bogus "at end of line"
3034 context messages from yyerror().
3038 yyerror("Unterminated attribute list");
3040 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
3048 PL_nextval[PL_nexttoke].opval = attrs;
3056 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
3057 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
3073 if (PL_lex_brackets <= 0)
3074 yyerror("Unmatched right square bracket");
3077 if (PL_lex_state == LEX_INTERPNORMAL) {
3078 if (PL_lex_brackets == 0) {
3079 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
3080 PL_lex_state = LEX_INTERPEND;
3087 if (PL_lex_brackets > 100) {
3088 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
3089 if (newlb != PL_lex_brackstack) {
3091 PL_lex_brackstack = newlb;
3094 switch (PL_expect) {
3096 if (PL_lex_formbrack) {
3100 if (PL_oldoldbufptr == PL_last_lop)
3101 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3103 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3104 OPERATOR(HASHBRACK);
3106 while (s < PL_bufend && SPACE_OR_TAB(*s))
3109 PL_tokenbuf[0] = '\0';
3110 if (d < PL_bufend && *d == '-') {
3111 PL_tokenbuf[0] = '-';
3113 while (d < PL_bufend && SPACE_OR_TAB(*d))
3116 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3117 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3119 while (d < PL_bufend && SPACE_OR_TAB(*d))
3122 char minus = (PL_tokenbuf[0] == '-');
3123 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3131 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3136 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3141 if (PL_oldoldbufptr == PL_last_lop)
3142 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3144 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3147 OPERATOR(HASHBRACK);
3148 /* This hack serves to disambiguate a pair of curlies
3149 * as being a block or an anon hash. Normally, expectation
3150 * determines that, but in cases where we're not in a
3151 * position to expect anything in particular (like inside
3152 * eval"") we have to resolve the ambiguity. This code
3153 * covers the case where the first term in the curlies is a
3154 * quoted string. Most other cases need to be explicitly
3155 * disambiguated by prepending a `+' before the opening
3156 * curly in order to force resolution as an anon hash.
3158 * XXX should probably propagate the outer expectation
3159 * into eval"" to rely less on this hack, but that could
3160 * potentially break current behavior of eval"".
3164 if (*s == '\'' || *s == '"' || *s == '`') {
3165 /* common case: get past first string, handling escapes */
3166 for (t++; t < PL_bufend && *t != *s;)
3167 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3171 else if (*s == 'q') {
3174 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3178 char open, close, term;
3181 while (t < PL_bufend && isSPACE(*t))
3185 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3189 for (t++; t < PL_bufend; t++) {
3190 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3192 else if (*t == open)
3196 for (t++; t < PL_bufend; t++) {
3197 if (*t == '\\' && t+1 < PL_bufend)
3199 else if (*t == close && --brackets <= 0)
3201 else if (*t == open)
3207 else if (isALNUM_lazy_if(t,UTF)) {
3209 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3212 while (t < PL_bufend && isSPACE(*t))
3214 /* if comma follows first term, call it an anon hash */
3215 /* XXX it could be a comma expression with loop modifiers */
3216 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3217 || (*t == '=' && t[1] == '>')))
3218 OPERATOR(HASHBRACK);
3219 if (PL_expect == XREF)
3222 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3228 yylval.ival = CopLINE(PL_curcop);
3229 if (isSPACE(*s) || *s == '#')
3230 PL_copline = NOLINE; /* invalidate current command line number */
3235 if (PL_lex_brackets <= 0)
3236 yyerror("Unmatched right curly bracket");
3238 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3239 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3240 PL_lex_formbrack = 0;
3241 if (PL_lex_state == LEX_INTERPNORMAL) {
3242 if (PL_lex_brackets == 0) {
3243 if (PL_expect & XFAKEBRACK) {
3244 PL_expect &= XENUMMASK;
3245 PL_lex_state = LEX_INTERPEND;
3247 return yylex(); /* ignore fake brackets */
3249 if (*s == '-' && s[1] == '>')
3250 PL_lex_state = LEX_INTERPENDMAYBE;
3251 else if (*s != '[' && *s != '{')
3252 PL_lex_state = LEX_INTERPEND;
3255 if (PL_expect & XFAKEBRACK) {
3256 PL_expect &= XENUMMASK;
3258 return yylex(); /* ignore fake brackets */
3268 if (PL_expect == XOPERATOR) {
3269 if (ckWARN(WARN_SEMICOLON)
3270 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3272 CopLINE_dec(PL_curcop);
3273 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3274 CopLINE_inc(PL_curcop);
3279 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3281 PL_expect = XOPERATOR;
3282 force_ident(PL_tokenbuf, '&');
3286 yylval.ival = (OPpENTERSUB_AMPER<<8);
3305 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3306 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3308 if (PL_expect == XSTATE && isALPHA(tmp) &&
3309 (s == PL_linestart+1 || s[-2] == '\n') )
3311 if (PL_in_eval && !PL_rsfp) {
3316 if (strnEQ(s,"=cut",4)) {
3330 PL_doextract = TRUE;
3333 if (PL_lex_brackets < PL_lex_formbrack) {
3335 #ifdef PERL_STRICT_CR
3336 for (t = s; SPACE_OR_TAB(*t); t++) ;
3338 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3340 if (*t == '\n' || *t == '#') {
3358 if (PL_expect != XOPERATOR) {
3359 if (s[1] != '<' && !strchr(s,'>'))
3362 s = scan_heredoc(s);
3364 s = scan_inputsymbol(s);
3365 TERM(sublex_start());
3370 SHop(OP_LEFT_SHIFT);
3384 SHop(OP_RIGHT_SHIFT);
3393 if (PL_expect == XOPERATOR) {
3394 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3397 return ','; /* grandfather non-comma-format format */
3401 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3402 PL_tokenbuf[0] = '@';
3403 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3404 sizeof PL_tokenbuf - 1, FALSE);
3405 if (PL_expect == XOPERATOR)
3406 no_op("Array length", s);
3407 if (!PL_tokenbuf[1])
3409 PL_expect = XOPERATOR;
3410 PL_pending_ident = '#';
3414 PL_tokenbuf[0] = '$';
3415 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3416 sizeof PL_tokenbuf - 1, FALSE);
3417 if (PL_expect == XOPERATOR)
3419 if (!PL_tokenbuf[1]) {
3421 yyerror("Final $ should be \\$ or $name");
3425 /* This kludge not intended to be bulletproof. */
3426 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3427 yylval.opval = newSVOP(OP_CONST, 0,
3428 newSViv(PL_compiling.cop_arybase));
3429 yylval.opval->op_private = OPpCONST_ARYBASE;
3435 if (PL_lex_state == LEX_NORMAL)
3438 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3441 PL_tokenbuf[0] = '@';
3442 if (ckWARN(WARN_SYNTAX)) {
3444 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3447 PL_bufptr = skipspace(PL_bufptr);
3448 while (t < PL_bufend && *t != ']')
3450 Perl_warner(aTHX_ WARN_SYNTAX,
3451 "Multidimensional syntax %.*s not supported",
3452 (t - PL_bufptr) + 1, PL_bufptr);
3456 else if (*s == '{') {
3457 PL_tokenbuf[0] = '%';
3458 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3459 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3461 char tmpbuf[sizeof PL_tokenbuf];
3463 for (t++; isSPACE(*t); t++) ;
3464 if (isIDFIRST_lazy_if(t,UTF)) {
3465 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3466 for (; isSPACE(*t); t++) ;
3467 if (*t == ';' && get_cv(tmpbuf, FALSE))
3468 Perl_warner(aTHX_ WARN_SYNTAX,
3469 "You need to quote \"%s\"", tmpbuf);
3475 PL_expect = XOPERATOR;
3476 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3477 bool islop = (PL_last_lop == PL_oldoldbufptr);
3478 if (!islop || PL_last_lop_op == OP_GREPSTART)
3479 PL_expect = XOPERATOR;
3480 else if (strchr("$@\"'`q", *s))
3481 PL_expect = XTERM; /* e.g. print $fh "foo" */
3482 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3483 PL_expect = XTERM; /* e.g. print $fh &sub */
3484 else if (isIDFIRST_lazy_if(s,UTF)) {
3485 char tmpbuf[sizeof PL_tokenbuf];
3486 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3487 if ((tmp = keyword(tmpbuf, len))) {
3488 /* binary operators exclude handle interpretations */
3500 PL_expect = XTERM; /* e.g. print $fh length() */
3505 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3506 if (gv && GvCVu(gv))
3507 PL_expect = XTERM; /* e.g. print $fh subr() */
3510 else if (isDIGIT(*s))
3511 PL_expect = XTERM; /* e.g. print $fh 3 */
3512 else if (*s == '.' && isDIGIT(s[1]))
3513 PL_expect = XTERM; /* e.g. print $fh .3 */
3514 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3515 PL_expect = XTERM; /* e.g. print $fh -1 */
3516 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3517 PL_expect = XTERM; /* print $fh <<"EOF" */
3519 PL_pending_ident = '$';
3523 if (PL_expect == XOPERATOR)
3525 PL_tokenbuf[0] = '@';
3526 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3527 if (!PL_tokenbuf[1]) {
3529 yyerror("Final @ should be \\@ or @name");
3532 if (PL_lex_state == LEX_NORMAL)
3534 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3536 PL_tokenbuf[0] = '%';
3538 /* Warn about @ where they meant $. */
3539 if (ckWARN(WARN_SYNTAX)) {
3540 if (*s == '[' || *s == '{') {
3542 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3544 if (*t == '}' || *t == ']') {
3546 PL_bufptr = skipspace(PL_bufptr);
3547 Perl_warner(aTHX_ WARN_SYNTAX,
3548 "Scalar value %.*s better written as $%.*s",
3549 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3554 PL_pending_ident = '@';
3557 case '/': /* may either be division or pattern */
3558 case '?': /* may either be conditional or pattern */
3559 if (PL_expect != XOPERATOR) {
3560 /* Disable warning on "study /blah/" */
3561 if (PL_oldoldbufptr == PL_last_uni
3562 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3563 || memNE(PL_last_uni, "study", 5)
3564 || isALNUM_lazy_if(PL_last_uni+5,UTF)))
3566 s = scan_pat(s,OP_MATCH);
3567 TERM(sublex_start());
3575 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3576 #ifdef PERL_STRICT_CR
3579 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3581 && (s == PL_linestart || s[-1] == '\n') )
3583 PL_lex_formbrack = 0;
3587 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3593 yylval.ival = OPf_SPECIAL;
3599 if (PL_expect != XOPERATOR)
3604 case '0': case '1': case '2': case '3': case '4':
3605 case '5': case '6': case '7': case '8': case '9':
3606 s = scan_num(s, &yylval);
3607 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3608 "### Saw number in '%s'\n", s);
3610 if (PL_expect == XOPERATOR)
3615 s = scan_str(s,FALSE,FALSE);
3616 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3617 "### Saw string in '%s'\n", s);
3619 if (PL_expect == XOPERATOR) {
3620 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3623 return ','; /* grandfather non-comma-format format */
3629 missingterm((char*)0);
3630 yylval.ival = OP_CONST;
3631 TERM(sublex_start());
3634 s = scan_str(s,FALSE,FALSE);
3635 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3636 "### Saw string in '%s'\n", s);
3638 if (PL_expect == XOPERATOR) {
3639 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3642 return ','; /* grandfather non-comma-format format */
3648 missingterm((char*)0);
3649 yylval.ival = OP_CONST;
3650 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3651 if (*d == '$' || *d == '@' || *d == '\\' || *d & 0x80) {
3652 yylval.ival = OP_STRINGIFY;
3656 TERM(sublex_start());
3659 s = scan_str(s,FALSE,FALSE);
3660 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3661 "### Saw backtick string in '%s'\n", s);
3663 if (PL_expect == XOPERATOR)
3664 no_op("Backticks",s);
3666 missingterm((char*)0);
3667 yylval.ival = OP_BACKTICK;
3669 TERM(sublex_start());
3673 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3674 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3676 if (PL_expect == XOPERATOR)
3677 no_op("Backslash",s);
3681 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3685 while (isDIGIT(*start) || *start == '_')
3687 if (*start == '.' && isDIGIT(start[1])) {
3688 s = scan_num(s, &yylval);
3691 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3692 else if (!isALPHA(*start) && (PL_expect == XTERM || PL_expect == XREF)) {
3696 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3699 s = scan_num(s, &yylval);
3706 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3745 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3747 /* Some keywords can be followed by any delimiter, including ':' */
3748 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
3749 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3750 (PL_tokenbuf[0] == 'q' &&
3751 strchr("qwxr", PL_tokenbuf[1])))));
3753 /* x::* is just a word, unless x is "CORE" */
3754 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3758 while (d < PL_bufend && isSPACE(*d))
3759 d++; /* no comments skipped here, or s### is misparsed */
3761 /* Is this a label? */
3762 if (!tmp && PL_expect == XSTATE
3763 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3765 yylval.pval = savepv(PL_tokenbuf);
3770 /* Check for keywords */
3771 tmp = keyword(PL_tokenbuf, len);
3773 /* Is this a word before a => operator? */
3774 if (*d == '=' && d[1] == '>') {
3776 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3777 yylval.opval->op_private = OPpCONST_BARE;
3781 if (tmp < 0) { /* second-class keyword? */
3782 GV *ogv = Nullgv; /* override (winner) */
3783 GV *hgv = Nullgv; /* hidden (loser) */
3784 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3786 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3789 if (GvIMPORTED_CV(gv))
3791 else if (! CvMETHOD(cv))
3795 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3796 (gv = *gvp) != (GV*)&PL_sv_undef &&
3797 GvCVu(gv) && GvIMPORTED_CV(gv))
3803 tmp = 0; /* overridden by import or by GLOBAL */
3806 && -tmp==KEY_lock /* XXX generalizable kludge */
3808 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3810 tmp = 0; /* any sub overrides "weak" keyword */
3812 else { /* no override */
3816 if (ckWARN(WARN_AMBIGUOUS) && hgv
3817 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3818 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3819 "Ambiguous call resolved as CORE::%s(), %s",
3820 GvENAME(hgv), "qualify as such or use &");
3827 default: /* not a keyword */
3830 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3832 /* Get the rest if it looks like a package qualifier */
3834 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
3836 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3839 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3840 *s == '\'' ? "'" : "::");
3844 if (PL_expect == XOPERATOR) {
3845 if (PL_bufptr == PL_linestart) {
3846 CopLINE_dec(PL_curcop);
3847 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3848 CopLINE_inc(PL_curcop);
3851 no_op("Bareword",s);
3854 /* Look for a subroutine with this name in current package,
3855 unless name is "Foo::", in which case Foo is a bearword
3856 (and a package name). */
3859 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3861 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3862 Perl_warner(aTHX_ WARN_BAREWORD,
3863 "Bareword \"%s\" refers to nonexistent package",
3866 PL_tokenbuf[len] = '\0';
3873 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3876 /* if we saw a global override before, get the right name */
3879 sv = newSVpvn("CORE::GLOBAL::",14);
3880 sv_catpv(sv,PL_tokenbuf);
3883 sv = newSVpv(PL_tokenbuf,0);
3885 /* Presume this is going to be a bareword of some sort. */
3888 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3889 yylval.opval->op_private = OPpCONST_BARE;
3891 /* And if "Foo::", then that's what it certainly is. */
3896 /* See if it's the indirect object for a list operator. */
3898 if (PL_oldoldbufptr &&
3899 PL_oldoldbufptr < PL_bufptr &&
3900 (PL_oldoldbufptr == PL_last_lop
3901 || PL_oldoldbufptr == PL_last_uni) &&
3902 /* NO SKIPSPACE BEFORE HERE! */
3903 (PL_expect == XREF ||
3904 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3906 bool immediate_paren = *s == '(';
3908 /* (Now we can afford to cross potential line boundary.) */
3911 /* Two barewords in a row may indicate method call. */
3913 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
3916 /* If not a declared subroutine, it's an indirect object. */
3917 /* (But it's an indir obj regardless for sort.) */
3919 if ((PL_last_lop_op == OP_SORT ||
3920 (!immediate_paren && (!gv || !GvCVu(gv)))) &&
3921 (PL_last_lop_op != OP_MAPSTART &&
3922 PL_last_lop_op != OP_GREPSTART))
3924 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3930 PL_expect = XOPERATOR;
3933 /* Is this a word before a => operator? */
3934 if (*s == '=' && s[1] == '>') {
3936 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
3940 /* If followed by a paren, it's certainly a subroutine. */
3943 if (gv && GvCVu(gv)) {
3944 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
3945 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3950 PL_nextval[PL_nexttoke].opval = yylval.opval;
3951 PL_expect = XOPERATOR;
3957 /* If followed by var or block, call it a method (unless sub) */
3959 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3960 PL_last_lop = PL_oldbufptr;
3961 PL_last_lop_op = OP_METHOD;
3965 /* If followed by a bareword, see if it looks like indir obj. */
3967 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp = intuit_method(s,gv)))
3970 /* Not a method, so call it a subroutine (if defined) */
3972 if (gv && GvCVu(gv)) {
3974 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
3975 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3976 "Ambiguous use of -%s resolved as -&%s()",
3977 PL_tokenbuf, PL_tokenbuf);
3978 /* Check for a constant sub */
3980 if ((sv = cv_const_sv(cv))) {
3982 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
3983 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
3984 yylval.opval->op_private = 0;
3988 /* Resolve to GV now. */
3989 op_free(yylval.opval);
3990 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
3991 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
3992 PL_last_lop = PL_oldbufptr;
3993 PL_last_lop_op = OP_ENTERSUB;
3994 /* Is there a prototype? */
3997 char *proto = SvPV((SV*)cv, len);
4000 if (strEQ(proto, "$"))
4002 if (*proto == '&' && *s == '{') {
4003 sv_setpv(PL_subname,"__ANON__");
4007 PL_nextval[PL_nexttoke].opval = yylval.opval;
4013 /* Call it a bare word */
4015 if (PL_hints & HINT_STRICT_SUBS)
4016 yylval.opval->op_private |= OPpCONST_STRICT;
4019 if (ckWARN(WARN_RESERVED)) {
4020 if (lastchar != '-') {
4021 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
4023 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
4030 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
4031 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4032 "Operator or semicolon missing before %c%s",
4033 lastchar, PL_tokenbuf);
4034 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4035 "Ambiguous use of %c resolved as operator %c",
4036 lastchar, lastchar);
4042 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4043 newSVpv(CopFILE(PL_curcop),0));
4047 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4048 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
4051 case KEY___PACKAGE__:
4052 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4054 ? newSVsv(PL_curstname)
4063 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
4064 char *pname = "main";
4065 if (PL_tokenbuf[2] == 'D')
4066 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
4067 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
4070 GvIOp(gv) = newIO();
4071 IoIFP(GvIOp(gv)) = PL_rsfp;
4072 #if defined(HAS_FCNTL) && defined(F_SETFD)
4074 int fd = PerlIO_fileno(PL_rsfp);
4075 fcntl(fd,F_SETFD,fd >= 3);
4078 /* Mark this internal pseudo-handle as clean */
4079 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
4081 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
4082 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
4083 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
4085 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
4086 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
4087 /* if the script was opened in binmode, we need to revert
4088 * it to text mode for compatibility; but only iff it has CRs
4089 * XXX this is a questionable hack at best. */
4090 if (PL_bufend-PL_bufptr > 2
4091 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
4094 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
4095 loc = PerlIO_tell(PL_rsfp);
4096 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4098 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4099 #if defined(__BORLANDC__)
4100 /* XXX see note in do_binmode() */
4101 ((FILE*)PL_rsfp)->flags |= _F_BIN;
4104 PerlIO_seek(PL_rsfp, loc, 0);
4119 if (PL_expect == XSTATE) {
4126 if (*s == ':' && s[1] == ':') {
4129 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4130 if (!(tmp = keyword(PL_tokenbuf, len)))
4131 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4145 LOP(OP_ACCEPT,XTERM);
4151 LOP(OP_ATAN2,XTERM);
4157 LOP(OP_BINMODE,XTERM);
4160 LOP(OP_BLESS,XTERM);
4169 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4186 if (!PL_cryptseen) {
4187 PL_cryptseen = TRUE;
4191 LOP(OP_CRYPT,XTERM);
4194 if (ckWARN(WARN_CHMOD)) {
4195 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4196 if (*d != '0' && isDIGIT(*d))
4197 Perl_warner(aTHX_ WARN_CHMOD,
4198 "chmod() mode argument is missing initial 0");
4200 LOP(OP_CHMOD,XTERM);
4203 LOP(OP_CHOWN,XTERM);
4206 LOP(OP_CONNECT,XTERM);
4222 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4226 PL_hints |= HINT_BLOCK_SCOPE;
4236 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4237 LOP(OP_DBMOPEN,XTERM);
4243 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4250 yylval.ival = CopLINE(PL_curcop);
4264 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4265 UNIBRACK(OP_ENTEREVAL);
4280 case KEY_endhostent:
4286 case KEY_endservent:
4289 case KEY_endprotoent:
4300 yylval.ival = CopLINE(PL_curcop);
4302 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4304 if ((PL_bufend - p) >= 3 &&
4305 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4307 else if ((PL_bufend - p) >= 4 &&
4308 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4311 if (isIDFIRST_lazy_if(p,UTF)) {
4312 p = scan_ident(p, PL_bufend,
4313 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4317 Perl_croak(aTHX_ "Missing $ on loop variable");
4322 LOP(OP_FORMLINE,XTERM);
4328 LOP(OP_FCNTL,XTERM);
4334 LOP(OP_FLOCK,XTERM);
4343 LOP(OP_GREPSTART, XREF);
4346 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4361 case KEY_getpriority:
4362 LOP(OP_GETPRIORITY,XTERM);
4364 case KEY_getprotobyname:
4367 case KEY_getprotobynumber:
4368 LOP(OP_GPBYNUMBER,XTERM);
4370 case KEY_getprotoent:
4382 case KEY_getpeername:
4383 UNI(OP_GETPEERNAME);
4385 case KEY_gethostbyname:
4388 case KEY_gethostbyaddr:
4389 LOP(OP_GHBYADDR,XTERM);
4391 case KEY_gethostent:
4394 case KEY_getnetbyname:
4397 case KEY_getnetbyaddr:
4398 LOP(OP_GNBYADDR,XTERM);
4403 case KEY_getservbyname:
4404 LOP(OP_GSBYNAME,XTERM);
4406 case KEY_getservbyport:
4407 LOP(OP_GSBYPORT,XTERM);
4409 case KEY_getservent:
4412 case KEY_getsockname:
4413 UNI(OP_GETSOCKNAME);
4415 case KEY_getsockopt:
4416 LOP(OP_GSOCKOPT,XTERM);
4438 yylval.ival = CopLINE(PL_curcop);
4442 LOP(OP_INDEX,XTERM);
4448 LOP(OP_IOCTL,XTERM);
4460 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4492 LOP(OP_LISTEN,XTERM);
4501 s = scan_pat(s,OP_MATCH);
4502 TERM(sublex_start());
4505 LOP(OP_MAPSTART, XREF);
4508 LOP(OP_MKDIR,XTERM);
4511 LOP(OP_MSGCTL,XTERM);
4514 LOP(OP_MSGGET,XTERM);
4517 LOP(OP_MSGRCV,XTERM);
4520 LOP(OP_MSGSND,XTERM);
4526 if (isIDFIRST_lazy_if(s,UTF)) {
4527 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4528 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4530 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
4531 if (!PL_in_my_stash) {
4534 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4542 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4549 if (PL_expect != XSTATE)
4550 yyerror("\"no\" not allowed in expression");
4551 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4552 s = force_version(s);
4557 if (*s == '(' || (s = skipspace(s), *s == '('))
4564 if (isIDFIRST_lazy_if(s,UTF)) {
4566 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4568 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE))
4569 Perl_warner(aTHX_ WARN_PRECEDENCE,
4570 "Precedence problem: open %.*s should be open(%.*s)",
4576 yylval.ival = OP_OR;
4586 LOP(OP_OPEN_DIR,XTERM);
4589 checkcomma(s,PL_tokenbuf,"filehandle");
4593 checkcomma(s,PL_tokenbuf,"filehandle");
4612 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4616 LOP(OP_PIPE_OP,XTERM);
4619 s = scan_str(s,FALSE,FALSE);
4621 missingterm((char*)0);
4622 yylval.ival = OP_CONST;
4623 TERM(sublex_start());
4629 s = scan_str(s,FALSE,FALSE);
4631 missingterm((char*)0);
4633 if (SvCUR(PL_lex_stuff)) {
4636 d = SvPV_force(PL_lex_stuff, len);
4638 for (; isSPACE(*d) && len; --len, ++d) ;
4641 if (!warned && ckWARN(WARN_QW)) {
4642 for (; !isSPACE(*d) && len; --len, ++d) {
4644 Perl_warner(aTHX_ WARN_QW,
4645 "Possible attempt to separate words with commas");
4648 else if (*d == '#') {
4649 Perl_warner(aTHX_ WARN_QW,
4650 "Possible attempt to put comments in qw() list");
4656 for (; !isSPACE(*d) && len; --len, ++d) ;
4658 words = append_elem(OP_LIST, words,
4659 newSVOP(OP_CONST, 0, tokeq(newSVpvn(b, d-b))));
4663 PL_nextval[PL_nexttoke].opval = words;
4668 SvREFCNT_dec(PL_lex_stuff);
4669 PL_lex_stuff = Nullsv;
4674 s = scan_str(s,FALSE,FALSE);
4676 missingterm((char*)0);
4677 yylval.ival = OP_STRINGIFY;
4678 if (SvIVX(PL_lex_stuff) == '\'')
4679 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4680 TERM(sublex_start());
4683 s = scan_pat(s,OP_QR);
4684 TERM(sublex_start());
4687 s = scan_str(s,FALSE,FALSE);
4689 missingterm((char*)0);
4690 yylval.ival = OP_BACKTICK;
4692 TERM(sublex_start());
4699 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4700 s = force_version(s);
4703 *PL_tokenbuf = '\0';
4704 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4705 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4706 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4708 yyerror("<> should be quotes");
4716 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4720 LOP(OP_RENAME,XTERM);
4729 LOP(OP_RINDEX,XTERM);
4752 LOP(OP_REVERSE,XTERM);
4763 TERM(sublex_start());
4765 TOKEN(1); /* force error */
4774 LOP(OP_SELECT,XTERM);
4780 LOP(OP_SEMCTL,XTERM);
4783 LOP(OP_SEMGET,XTERM);
4786 LOP(OP_SEMOP,XTERM);
4792 LOP(OP_SETPGRP,XTERM);
4794 case KEY_setpriority:
4795 LOP(OP_SETPRIORITY,XTERM);
4797 case KEY_sethostent:
4803 case KEY_setservent:
4806 case KEY_setprotoent:
4816 LOP(OP_SEEKDIR,XTERM);
4818 case KEY_setsockopt:
4819 LOP(OP_SSOCKOPT,XTERM);
4825 LOP(OP_SHMCTL,XTERM);
4828 LOP(OP_SHMGET,XTERM);
4831 LOP(OP_SHMREAD,XTERM);
4834 LOP(OP_SHMWRITE,XTERM);
4837 LOP(OP_SHUTDOWN,XTERM);
4846 LOP(OP_SOCKET,XTERM);
4848 case KEY_socketpair:
4849 LOP(OP_SOCKPAIR,XTERM);
4852 checkcomma(s,PL_tokenbuf,"subroutine name");
4854 if (*s == ';' || *s == ')') /* probably a close */
4855 Perl_croak(aTHX_ "sort is now a reserved word");
4857 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4861 LOP(OP_SPLIT,XTERM);
4864 LOP(OP_SPRINTF,XTERM);
4867 LOP(OP_SPLICE,XTERM);
4882 LOP(OP_SUBSTR,XTERM);
4888 char tmpbuf[sizeof PL_tokenbuf];
4890 expectation attrful;
4891 bool have_name, have_proto;
4896 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
4897 (*s == ':' && s[1] == ':'))
4900 attrful = XATTRBLOCK;
4901 /* remember buffer pos'n for later force_word */
4902 tboffset = s - PL_oldbufptr;
4903 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4904 if (strchr(tmpbuf, ':'))
4905 sv_setpv(PL_subname, tmpbuf);
4907 sv_setsv(PL_subname,PL_curstname);
4908 sv_catpvn(PL_subname,"::",2);
4909 sv_catpvn(PL_subname,tmpbuf,len);
4916 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4917 PL_expect = XTERMBLOCK;
4918 attrful = XATTRTERM;
4919 sv_setpv(PL_subname,"?");
4923 if (key == KEY_format) {
4925 PL_lex_formbrack = PL_lex_brackets + 1;
4927 (void) force_word(PL_oldbufptr + tboffset, WORD,
4932 /* Look for a prototype */
4936 s = scan_str(s,FALSE,FALSE);
4939 SvREFCNT_dec(PL_lex_stuff);
4940 PL_lex_stuff = Nullsv;
4941 Perl_croak(aTHX_ "Prototype not terminated");
4944 d = SvPVX(PL_lex_stuff);
4946 for (p = d; *p; ++p) {
4951 SvCUR(PL_lex_stuff) = tmp;
4959 if (*s == ':' && s[1] != ':')
4960 PL_expect = attrful;
4963 PL_nextval[PL_nexttoke].opval =
4964 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
4965 PL_lex_stuff = Nullsv;
4969 sv_setpv(PL_subname,"__ANON__");
4972 (void) force_word(PL_oldbufptr + tboffset, WORD,
4981 LOP(OP_SYSTEM,XREF);
4984 LOP(OP_SYMLINK,XTERM);
4987 LOP(OP_SYSCALL,XTERM);
4990 LOP(OP_SYSOPEN,XTERM);
4993 LOP(OP_SYSSEEK,XTERM);
4996 LOP(OP_SYSREAD,XTERM);
4999 LOP(OP_SYSWRITE,XTERM);
5003 TERM(sublex_start());
5024 LOP(OP_TRUNCATE,XTERM);
5036 yylval.ival = CopLINE(PL_curcop);
5040 yylval.ival = CopLINE(PL_curcop);
5044 LOP(OP_UNLINK,XTERM);
5050 LOP(OP_UNPACK,XTERM);
5053 LOP(OP_UTIME,XTERM);
5056 if (ckWARN(WARN_UMASK)) {
5057 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
5058 if (*d != '0' && isDIGIT(*d))
5059 Perl_warner(aTHX_ WARN_UMASK,
5060 "umask: argument is missing initial 0");
5065 LOP(OP_UNSHIFT,XTERM);
5068 if (PL_expect != XSTATE)
5069 yyerror("\"use\" not allowed in expression");
5071 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
5072 s = force_version(s);
5073 if (*s == ';' || (s = skipspace(s), *s == ';')) {
5074 PL_nextval[PL_nexttoke].opval = Nullop;
5079 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5080 s = force_version(s);
5092 yylval.ival = CopLINE(PL_curcop);
5096 PL_hints |= HINT_BLOCK_SCOPE;
5103 LOP(OP_WAITPID,XTERM);
5111 static char ctl_l[2];
5113 if (ctl_l[0] == '\0')
5114 ctl_l[0] = toCTRL('L');
5115 gv_fetchpv(ctl_l,TRUE, SVt_PV);
5118 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
5123 if (PL_expect == XOPERATOR)
5129 yylval.ival = OP_XOR;
5134 TERM(sublex_start());
5139 #pragma segment Main
5143 Perl_keyword(pTHX_ register char *d, I32 len)
5148 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
5149 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
5150 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
5151 if (strEQ(d,"__DATA__")) return KEY___DATA__;
5152 if (strEQ(d,"__END__")) return KEY___END__;
5156 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
5161 if (strEQ(d,"and")) return -KEY_and;
5162 if (strEQ(d,"abs")) return -KEY_abs;
5165 if (strEQ(d,"alarm")) return -KEY_alarm;
5166 if (strEQ(d,"atan2")) return -KEY_atan2;
5169 if (strEQ(d,"accept")) return -KEY_accept;
5174 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5177 if (strEQ(d,"bless")) return -KEY_bless;
5178 if (strEQ(d,"bind")) return -KEY_bind;
5179 if (strEQ(d,"binmode")) return -KEY_binmode;
5182 if (strEQ(d,"CORE")) return -KEY_CORE;
5183 if (strEQ(d,"CHECK")) return KEY_CHECK;
5188 if (strEQ(d,"cmp")) return -KEY_cmp;
5189 if (strEQ(d,"chr")) return -KEY_chr;
5190 if (strEQ(d,"cos")) return -KEY_cos;
5193 if (strEQ(d,"chop")) return -KEY_chop;
5196 if (strEQ(d,"close")) return -KEY_close;
5197 if (strEQ(d,"chdir")) return -KEY_chdir;
5198 if (strEQ(d,"chomp")) return -KEY_chomp;
5199 if (strEQ(d,"chmod")) return -KEY_chmod;
5200 if (strEQ(d,"chown")) return -KEY_chown;
5201 if (strEQ(d,"crypt")) return -KEY_crypt;
5204 if (strEQ(d,"chroot")) return -KEY_chroot;
5205 if (strEQ(d,"caller")) return -KEY_caller;
5208 if (strEQ(d,"connect")) return -KEY_connect;
5211 if (strEQ(d,"closedir")) return -KEY_closedir;
5212 if (strEQ(d,"continue")) return -KEY_continue;
5217 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5222 if (strEQ(d,"do")) return KEY_do;
5225 if (strEQ(d,"die")) return -KEY_die;
5228 if (strEQ(d,"dump")) return -KEY_dump;
5231 if (strEQ(d,"delete")) return KEY_delete;
5234 if (strEQ(d,"defined")) return KEY_defined;
5235 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5238 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5243 if (strEQ(d,"END")) return KEY_END;
5248 if (strEQ(d,"eq")) return -KEY_eq;
5251 if (strEQ(d,"eof")) return -KEY_eof;
5252 if (strEQ(d,"exp")) return -KEY_exp;
5255 if (strEQ(d,"else")) return KEY_else;
5256 if (strEQ(d,"exit")) return -KEY_exit;
5257 if (strEQ(d,"eval")) return KEY_eval;
5258 if (strEQ(d,"exec")) return -KEY_exec;
5259 if (strEQ(d,"each")) return -KEY_each;
5262 if (strEQ(d,"elsif")) return KEY_elsif;
5265 if (strEQ(d,"exists")) return KEY_exists;
5266 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5269 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5270 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5273 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5276 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5277 if (strEQ(d,"endservent")) return -KEY_endservent;
5280 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5287 if (strEQ(d,"for")) return KEY_for;
5290 if (strEQ(d,"fork")) return -KEY_fork;
5293 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5294 if (strEQ(d,"flock")) return -KEY_flock;
5297 if (strEQ(d,"format")) return KEY_format;
5298 if (strEQ(d,"fileno")) return -KEY_fileno;
5301 if (strEQ(d,"foreach")) return KEY_foreach;
5304 if (strEQ(d,"formline")) return -KEY_formline;
5309 if (strnEQ(d,"get",3)) {
5314 if (strEQ(d,"ppid")) return -KEY_getppid;
5315 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5318 if (strEQ(d,"pwent")) return -KEY_getpwent;
5319 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5320 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5323 if (strEQ(d,"peername")) return -KEY_getpeername;
5324 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5325 if (strEQ(d,"priority")) return -KEY_getpriority;
5328 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5331 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5335 else if (*d == 'h') {
5336 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5337 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5338 if (strEQ(d,"hostent")) return -KEY_gethostent;
5340 else if (*d == 'n') {
5341 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5342 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5343 if (strEQ(d,"netent")) return -KEY_getnetent;
5345 else if (*d == 's') {
5346 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5347 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5348 if (strEQ(d,"servent")) return -KEY_getservent;
5349 if (strEQ(d,"sockname")) return -KEY_getsockname;
5350 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5352 else if (*d == 'g') {
5353 if (strEQ(d,"grent")) return -KEY_getgrent;
5354 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5355 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5357 else if (*d == 'l') {
5358 if (strEQ(d,"login")) return -KEY_getlogin;
5360 else if (strEQ(d,"c")) return -KEY_getc;
5365 if (strEQ(d,"gt")) return -KEY_gt;
5366 if (strEQ(d,"ge")) return -KEY_ge;
5369 if (strEQ(d,"grep")) return KEY_grep;
5370 if (strEQ(d,"goto")) return KEY_goto;
5371 if (strEQ(d,"glob")) return KEY_glob;
5374 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5379 if (strEQ(d,"hex")) return -KEY_hex;
5382 if (strEQ(d,"INIT")) return KEY_INIT;
5387 if (strEQ(d,"if")) return KEY_if;
5390 if (strEQ(d,"int")) return -KEY_int;
5393 if (strEQ(d,"index")) return -KEY_index;
5394 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5399 if (strEQ(d,"join")) return -KEY_join;
5403 if (strEQ(d,"keys")) return -KEY_keys;
5404 if (strEQ(d,"kill")) return -KEY_kill;
5410 if (strEQ(d,"lt")) return -KEY_lt;
5411 if (strEQ(d,"le")) return -KEY_le;
5412 if (strEQ(d,"lc")) return -KEY_lc;
5415 if (strEQ(d,"log")) return -KEY_log;
5418 if (strEQ(d,"last")) return KEY_last;
5419 if (strEQ(d,"link")) return -KEY_link;
5420 if (strEQ(d,"lock")) return -KEY_lock;
5423 if (strEQ(d,"local")) return KEY_local;
5424 if (strEQ(d,"lstat")) return -KEY_lstat;
5427 if (strEQ(d,"length")) return -KEY_length;
5428 if (strEQ(d,"listen")) return -KEY_listen;
5431 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5434 if (strEQ(d,"localtime")) return -KEY_localtime;
5440 case 1: return KEY_m;
5442 if (strEQ(d,"my")) return KEY_my;
5445 if (strEQ(d,"map")) return KEY_map;
5448 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5451 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5452 if (strEQ(d,"msgget")) return -KEY_msgget;
5453 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5454 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5459 if (strEQ(d,"next")) return KEY_next;
5460 if (strEQ(d,"ne")) return -KEY_ne;
5461 if (strEQ(d,"not")) return -KEY_not;
5462 if (strEQ(d,"no")) return KEY_no;
5467 if (strEQ(d,"or")) return -KEY_or;
5470 if (strEQ(d,"ord")) return -KEY_ord;
5471 if (strEQ(d,"oct")) return -KEY_oct;
5472 if (strEQ(d,"our")) return KEY_our;
5475 if (strEQ(d,"open")) return -KEY_open;
5478 if (strEQ(d,"opendir")) return -KEY_opendir;
5485 if (strEQ(d,"pop")) return -KEY_pop;
5486 if (strEQ(d,"pos")) return KEY_pos;
5489 if (strEQ(d,"push")) return -KEY_push;
5490 if (strEQ(d,"pack")) return -KEY_pack;
5491 if (strEQ(d,"pipe")) return -KEY_pipe;
5494 if (strEQ(d,"print")) return KEY_print;
5497 if (strEQ(d,"printf")) return KEY_printf;
5500 if (strEQ(d,"package")) return KEY_package;
5503 if (strEQ(d,"prototype")) return KEY_prototype;
5508 if (strEQ(d,"q")) return KEY_q;
5509 if (strEQ(d,"qr")) return KEY_qr;
5510 if (strEQ(d,"qq")) return KEY_qq;
5511 if (strEQ(d,"qw")) return KEY_qw;
5512 if (strEQ(d,"qx")) return KEY_qx;
5514 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5519 if (strEQ(d,"ref")) return -KEY_ref;
5522 if (strEQ(d,"read")) return -KEY_read;
5523 if (strEQ(d,"rand")) return -KEY_rand;
5524 if (strEQ(d,"recv")) return -KEY_recv;
5525 if (strEQ(d,"redo")) return KEY_redo;
5528 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5529 if (strEQ(d,"reset")) return -KEY_reset;
5532 if (strEQ(d,"return")) return KEY_return;
5533 if (strEQ(d,"rename")) return -KEY_rename;
5534 if (strEQ(d,"rindex")) return -KEY_rindex;
5537 if (strEQ(d,"require")) return -KEY_require;
5538 if (strEQ(d,"reverse")) return -KEY_reverse;
5539 if (strEQ(d,"readdir")) return -KEY_readdir;
5542 if (strEQ(d,"readlink")) return -KEY_readlink;
5543 if (strEQ(d,"readline")) return -KEY_readline;
5544 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5547 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5553 case 0: return KEY_s;
5555 if (strEQ(d,"scalar")) return KEY_scalar;
5560 if (strEQ(d,"seek")) return -KEY_seek;
5561 if (strEQ(d,"send")) return -KEY_send;
5564 if (strEQ(d,"semop")) return -KEY_semop;
5567 if (strEQ(d,"select")) return -KEY_select;
5568 if (strEQ(d,"semctl")) return -KEY_semctl;
5569 if (strEQ(d,"semget")) return -KEY_semget;
5572 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5573 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5576 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5577 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5580 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5583 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5584 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5585 if (strEQ(d,"setservent")) return -KEY_setservent;
5588 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5589 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5596 if (strEQ(d,"shift")) return -KEY_shift;
5599 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5600 if (strEQ(d,"shmget")) return -KEY_shmget;
5603 if (strEQ(d,"shmread")) return -KEY_shmread;
5606 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5607 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5612 if (strEQ(d,"sin")) return -KEY_sin;
5615 if (strEQ(d,"sleep")) return -KEY_sleep;
5618 if (strEQ(d,"sort")) return KEY_sort;
5619 if (strEQ(d,"socket")) return -KEY_socket;
5620 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5623 if (strEQ(d,"split")) return KEY_split;
5624 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5625 if (strEQ(d,"splice")) return -KEY_splice;
5628 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5631 if (strEQ(d,"srand")) return -KEY_srand;
5634 if (strEQ(d,"stat")) return -KEY_stat;
5635 if (strEQ(d,"study")) return KEY_study;
5638 if (strEQ(d,"substr")) return -KEY_substr;
5639 if (strEQ(d,"sub")) return KEY_sub;
5644 if (strEQ(d,"system")) return -KEY_system;
5647 if (strEQ(d,"symlink")) return -KEY_symlink;
5648 if (strEQ(d,"syscall")) return -KEY_syscall;
5649 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5650 if (strEQ(d,"sysread")) return -KEY_sysread;
5651 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5654 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5663 if (strEQ(d,"tr")) return KEY_tr;
5666 if (strEQ(d,"tie")) return KEY_tie;
5669 if (strEQ(d,"tell")) return -KEY_tell;
5670 if (strEQ(d,"tied")) return KEY_tied;
5671 if (strEQ(d,"time")) return -KEY_time;
5674 if (strEQ(d,"times")) return -KEY_times;
5677 if (strEQ(d,"telldir")) return -KEY_telldir;
5680 if (strEQ(d,"truncate")) return -KEY_truncate;
5687 if (strEQ(d,"uc")) return -KEY_uc;
5690 if (strEQ(d,"use")) return KEY_use;
5693 if (strEQ(d,"undef")) return KEY_undef;
5694 if (strEQ(d,"until")) return KEY_until;
5695 if (strEQ(d,"untie")) return KEY_untie;
5696 if (strEQ(d,"utime")) return -KEY_utime;
5697 if (strEQ(d,"umask")) return -KEY_umask;
5700 if (strEQ(d,"unless")) return KEY_unless;
5701 if (strEQ(d,"unpack")) return -KEY_unpack;
5702 if (strEQ(d,"unlink")) return -KEY_unlink;
5705 if (strEQ(d,"unshift")) return -KEY_unshift;
5706 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5711 if (strEQ(d,"values")) return -KEY_values;
5712 if (strEQ(d,"vec")) return -KEY_vec;
5717 if (strEQ(d,"warn")) return -KEY_warn;
5718 if (strEQ(d,"wait")) return -KEY_wait;
5721 if (strEQ(d,"while")) return KEY_while;
5722 if (strEQ(d,"write")) return -KEY_write;
5725 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5728 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5733 if (len == 1) return -KEY_x;
5734 if (strEQ(d,"xor")) return -KEY_xor;
5737 if (len == 1) return KEY_y;
5746 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5750 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5751 if (ckWARN(WARN_SYNTAX)) {
5753 for (w = s+2; *w && level; w++) {
5760 for (; *w && isSPACE(*w); w++) ;
5761 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5762 Perl_warner(aTHX_ WARN_SYNTAX,
5763 "%s (...) interpreted as function",name);
5766 while (s < PL_bufend && isSPACE(*s))
5770 while (s < PL_bufend && isSPACE(*s))
5772 if (isIDFIRST_lazy_if(s,UTF)) {
5774 while (isALNUM_lazy_if(s,UTF))
5776 while (s < PL_bufend && isSPACE(*s))
5781 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5785 Perl_croak(aTHX_ "No comma allowed after %s", what);
5790 /* Either returns sv, or mortalizes sv and returns a new SV*.
5791 Best used as sv=new_constant(..., sv, ...).
5792 If s, pv are NULL, calls subroutine with one argument,
5793 and type is used with error messages only. */
5796 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5800 HV *table = GvHV(PL_hintgv); /* ^H */
5804 const char *why1, *why2, *why3;
5806 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
5809 why2 = strEQ(key,"charnames")
5810 ? "(possibly a missing \"use charnames ...\")"
5812 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
5813 (type ? type: "undef"), why2);
5815 /* This is convoluted and evil ("goto considered harmful")
5816 * but I do not understand the intricacies of all the different
5817 * failure modes of %^H in here. The goal here is to make
5818 * the most probable error message user-friendly. --jhi */
5823 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
5824 (type ? type: "undef"), why1, why2, why3);
5826 yyerror(SvPVX(msg));
5830 cvp = hv_fetch(table, key, strlen(key), FALSE);
5831 if (!cvp || !SvOK(*cvp)) {
5834 why3 = "} is not defined";
5837 sv_2mortal(sv); /* Parent created it permanently */
5840 pv = sv_2mortal(newSVpvn(s, len));
5842 typesv = sv_2mortal(newSVpv(type, 0));
5844 typesv = &PL_sv_undef;
5846 PUSHSTACKi(PERLSI_OVERLOAD);
5858 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5862 /* Check the eval first */
5863 if (!PL_in_eval && SvTRUE(ERRSV)) {
5865 sv_catpv(ERRSV, "Propagated");
5866 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5868 res = SvREFCNT_inc(sv);
5872 (void)SvREFCNT_inc(res);
5881 why1 = "Call to &{$^H{";
5883 why3 = "}} did not return a defined value";
5892 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
5894 register char *d = dest;
5895 register char *e = d + destlen - 3; /* two-character token, ending NUL */
5898 Perl_croak(aTHX_ ident_too_long);
5899 if (isALNUM(*s)) /* UTF handled below */
5901 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
5906 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
5910 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5911 char *t = s + UTF8SKIP(s);
5912 while (*t & 0x80 && is_utf8_mark((U8*)t))
5914 if (d + (t - s) > e)
5915 Perl_croak(aTHX_ ident_too_long);
5916 Copy(s, d, t - s, char);
5929 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
5939 e = d + destlen - 3; /* two-character token, ending NUL */
5941 while (isDIGIT(*s)) {
5943 Perl_croak(aTHX_ ident_too_long);
5950 Perl_croak(aTHX_ ident_too_long);
5951 if (isALNUM(*s)) /* UTF handled below */
5953 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
5958 else if (*s == ':' && s[1] == ':') {
5962 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5963 char *t = s + UTF8SKIP(s);
5964 while (*t & 0x80 && is_utf8_mark((U8*)t))
5966 if (d + (t - s) > e)
5967 Perl_croak(aTHX_ ident_too_long);
5968 Copy(s, d, t - s, char);
5979 if (PL_lex_state != LEX_NORMAL)
5980 PL_lex_state = LEX_INTERPENDMAYBE;
5983 if (*s == '$' && s[1] &&
5984 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
5997 if (*d == '^' && *s && isCONTROLVAR(*s)) {
6002 if (isSPACE(s[-1])) {
6005 if (!SPACE_OR_TAB(ch)) {
6011 if (isIDFIRST_lazy_if(d,UTF)) {
6015 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
6017 while (e < send && *e & 0x80 && is_utf8_mark((U8*)e))
6020 Copy(s, d, e - s, char);
6025 while ((isALNUM(*s) || *s == ':') && d < e)
6028 Perl_croak(aTHX_ ident_too_long);
6031 while (s < send && SPACE_OR_TAB(*s)) s++;
6032 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
6033 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
6034 const char *brack = *s == '[' ? "[...]" : "{...}";
6035 Perl_warner(aTHX_ WARN_AMBIGUOUS,
6036 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
6037 funny, dest, brack, funny, dest, brack);
6040 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
6044 /* Handle extended ${^Foo} variables
6045 * 1999-02-27 mjd-perl-patch@plover.com */
6046 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
6050 while (isALNUM(*s) && d < e) {
6054 Perl_croak(aTHX_ ident_too_long);
6059 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
6060 PL_lex_state = LEX_INTERPEND;
6063 if (PL_lex_state == LEX_NORMAL) {
6064 if (ckWARN(WARN_AMBIGUOUS) &&
6065 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
6067 Perl_warner(aTHX_ WARN_AMBIGUOUS,
6068 "Ambiguous use of %c{%s} resolved to %c%s",
6069 funny, dest, funny, dest);
6074 s = bracket; /* let the parser handle it */
6078 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
6079 PL_lex_state = LEX_INTERPEND;
6084 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
6089 *pmfl |= PMf_GLOBAL;
6091 *pmfl |= PMf_CONTINUE;
6095 *pmfl |= PMf_MULTILINE;
6097 *pmfl |= PMf_SINGLELINE;
6099 *pmfl |= PMf_EXTENDED;
6103 S_scan_pat(pTHX_ char *start, I32 type)
6108 s = scan_str(start,FALSE,FALSE);
6111 SvREFCNT_dec(PL_lex_stuff);
6112 PL_lex_stuff = Nullsv;
6113 Perl_croak(aTHX_ "Search pattern not terminated");
6116 pm = (PMOP*)newPMOP(type, 0);
6117 if (PL_multi_open == '?')
6118 pm->op_pmflags |= PMf_ONCE;
6120 while (*s && strchr("iomsx", *s))
6121 pmflag(&pm->op_pmflags,*s++);
6124 while (*s && strchr("iogcmsx", *s))
6125 pmflag(&pm->op_pmflags,*s++);
6127 pm->op_pmpermflags = pm->op_pmflags;
6129 PL_lex_op = (OP*)pm;
6130 yylval.ival = OP_MATCH;
6135 S_scan_subst(pTHX_ char *start)
6142 yylval.ival = OP_NULL;
6144 s = scan_str(start,FALSE,FALSE);
6148 SvREFCNT_dec(PL_lex_stuff);
6149 PL_lex_stuff = Nullsv;
6150 Perl_croak(aTHX_ "Substitution pattern not terminated");
6153 if (s[-1] == PL_multi_open)
6156 first_start = PL_multi_start;
6157 s = scan_str(s,FALSE,FALSE);
6160 SvREFCNT_dec(PL_lex_stuff);
6161 PL_lex_stuff = Nullsv;
6163 SvREFCNT_dec(PL_lex_repl);
6164 PL_lex_repl = Nullsv;
6165 Perl_croak(aTHX_ "Substitution replacement not terminated");
6167 PL_multi_start = first_start; /* so whole substitution is taken together */
6169 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6175 else if (strchr("iogcmsx", *s))
6176 pmflag(&pm->op_pmflags,*s++);
6183 PL_sublex_info.super_bufptr = s;
6184 PL_sublex_info.super_bufend = PL_bufend;
6186 pm->op_pmflags |= PMf_EVAL;
6187 repl = newSVpvn("",0);
6189 sv_catpv(repl, es ? "eval " : "do ");
6190 sv_catpvn(repl, "{ ", 2);
6191 sv_catsv(repl, PL_lex_repl);
6192 sv_catpvn(repl, " };", 2);
6194 SvREFCNT_dec(PL_lex_repl);
6198 pm->op_pmpermflags = pm->op_pmflags;
6199 PL_lex_op = (OP*)pm;
6200 yylval.ival = OP_SUBST;
6205 S_scan_trans(pTHX_ char *start)
6216 yylval.ival = OP_NULL;
6218 s = scan_str(start,FALSE,FALSE);
6221 SvREFCNT_dec(PL_lex_stuff);
6222 PL_lex_stuff = Nullsv;
6223 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6225 if (s[-1] == PL_multi_open)
6228 s = scan_str(s,FALSE,FALSE);
6231 SvREFCNT_dec(PL_lex_stuff);
6232 PL_lex_stuff = Nullsv;
6234 SvREFCNT_dec(PL_lex_repl);
6235 PL_lex_repl = Nullsv;
6236 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6239 New(803,tbl,256,short);
6240 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6242 complement = del = squash = 0;
6243 while (strchr("cds", *s)) {
6245 complement = OPpTRANS_COMPLEMENT;
6247 del = OPpTRANS_DELETE;
6249 squash = OPpTRANS_SQUASH;
6252 o->op_private = del|squash|complement;
6255 yylval.ival = OP_TRANS;
6260 S_scan_heredoc(pTHX_ register char *s)
6263 I32 op_type = OP_SCALAR;
6270 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6274 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6277 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
6278 if (*peek && strchr("`'\"",*peek)) {
6281 s = delimcpy(d, e, s, PL_bufend, term, &len);
6291 if (!isALNUM_lazy_if(s,UTF))
6292 deprecate("bare << to mean <<\"\"");
6293 for (; isALNUM_lazy_if(s,UTF); s++) {
6298 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6299 Perl_croak(aTHX_ "Delimiter for here document is too long");
6302 len = d - PL_tokenbuf;
6303 #ifndef PERL_STRICT_CR
6304 d = strchr(s, '\r');
6308 while (s < PL_bufend) {
6314 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6323 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6328 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6329 herewas = newSVpvn(s,PL_bufend-s);
6331 s--, herewas = newSVpvn(s,d-s);
6332 s += SvCUR(herewas);
6334 tmpstr = NEWSV(87,79);
6335 sv_upgrade(tmpstr, SVt_PVIV);
6340 else if (term == '`') {
6341 op_type = OP_BACKTICK;
6342 SvIVX(tmpstr) = '\\';
6346 PL_multi_start = CopLINE(PL_curcop);
6347 PL_multi_open = PL_multi_close = '<';
6348 term = *PL_tokenbuf;
6349 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6350 char *bufptr = PL_sublex_info.super_bufptr;
6351 char *bufend = PL_sublex_info.super_bufend;
6352 char *olds = s - SvCUR(herewas);
6353 s = strchr(bufptr, '\n');
6357 while (s < bufend &&
6358 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6360 CopLINE_inc(PL_curcop);
6363 CopLINE_set(PL_curcop, PL_multi_start);
6364 missingterm(PL_tokenbuf);
6366 sv_setpvn(herewas,bufptr,d-bufptr+1);
6367 sv_setpvn(tmpstr,d+1,s-d);
6369 sv_catpvn(herewas,s,bufend-s);
6370 (void)strcpy(bufptr,SvPVX(herewas));
6377 while (s < PL_bufend &&
6378 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6380 CopLINE_inc(PL_curcop);
6382 if (s >= PL_bufend) {
6383 CopLINE_set(PL_curcop, PL_multi_start);
6384 missingterm(PL_tokenbuf);
6386 sv_setpvn(tmpstr,d+1,s-d);
6388 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6390 sv_catpvn(herewas,s,PL_bufend-s);
6391 sv_setsv(PL_linestr,herewas);
6392 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6393 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6396 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6397 while (s >= PL_bufend) { /* multiple line string? */
6399 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6400 CopLINE_set(PL_curcop, PL_multi_start);
6401 missingterm(PL_tokenbuf);
6403 CopLINE_inc(PL_curcop);
6404 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6405 #ifndef PERL_STRICT_CR
6406 if (PL_bufend - PL_linestart >= 2) {
6407 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6408 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6410 PL_bufend[-2] = '\n';
6412 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6414 else if (PL_bufend[-1] == '\r')
6415 PL_bufend[-1] = '\n';
6417 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6418 PL_bufend[-1] = '\n';
6420 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6421 SV *sv = NEWSV(88,0);
6423 sv_upgrade(sv, SVt_PVMG);
6424 sv_setsv(sv,PL_linestr);
6425 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6427 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6430 sv_catsv(PL_linestr,herewas);
6431 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6435 sv_catsv(tmpstr,PL_linestr);
6440 PL_multi_end = CopLINE(PL_curcop);
6441 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6442 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6443 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6445 SvREFCNT_dec(herewas);
6446 PL_lex_stuff = tmpstr;
6447 yylval.ival = op_type;
6452 takes: current position in input buffer
6453 returns: new position in input buffer
6454 side-effects: yylval and lex_op are set.
6459 <FH> read from filehandle
6460 <pkg::FH> read from package qualified filehandle
6461 <pkg'FH> read from package qualified filehandle
6462 <$fh> read from filehandle in $fh
6468 S_scan_inputsymbol(pTHX_ char *start)
6470 register char *s = start; /* current position in buffer */
6476 d = PL_tokenbuf; /* start of temp holding space */
6477 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6478 end = strchr(s, '\n');
6481 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6483 /* die if we didn't have space for the contents of the <>,
6484 or if it didn't end, or if we see a newline
6487 if (len >= sizeof PL_tokenbuf)
6488 Perl_croak(aTHX_ "Excessively long <> operator");
6490 Perl_croak(aTHX_ "Unterminated <> operator");
6495 Remember, only scalar variables are interpreted as filehandles by
6496 this code. Anything more complex (e.g., <$fh{$num}>) will be
6497 treated as a glob() call.
6498 This code makes use of the fact that except for the $ at the front,
6499 a scalar variable and a filehandle look the same.
6501 if (*d == '$' && d[1]) d++;
6503 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6504 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6507 /* If we've tried to read what we allow filehandles to look like, and
6508 there's still text left, then it must be a glob() and not a getline.
6509 Use scan_str to pull out the stuff between the <> and treat it
6510 as nothing more than a string.
6513 if (d - PL_tokenbuf != len) {
6514 yylval.ival = OP_GLOB;
6516 s = scan_str(start,FALSE,FALSE);
6518 Perl_croak(aTHX_ "Glob not terminated");
6522 /* we're in a filehandle read situation */
6525 /* turn <> into <ARGV> */
6527 (void)strcpy(d,"ARGV");
6529 /* if <$fh>, create the ops to turn the variable into a
6535 /* try to find it in the pad for this block, otherwise find
6536 add symbol table ops
6538 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6539 OP *o = newOP(OP_PADSV, 0);
6541 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o);
6544 GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV);
6545 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0,
6546 newUNOP(OP_RV2SV, 0,
6547 newGVOP(OP_GV, 0, gv)));
6549 PL_lex_op->op_flags |= OPf_SPECIAL;
6550 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6551 yylval.ival = OP_NULL;
6554 /* If it's none of the above, it must be a literal filehandle
6555 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6557 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6558 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6559 yylval.ival = OP_NULL;
6568 takes: start position in buffer
6569 keep_quoted preserve \ on the embedded delimiter(s)
6570 keep_delims preserve the delimiters around the string
6571 returns: position to continue reading from buffer
6572 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6573 updates the read buffer.
6575 This subroutine pulls a string out of the input. It is called for:
6576 q single quotes q(literal text)
6577 ' single quotes 'literal text'
6578 qq double quotes qq(interpolate $here please)
6579 " double quotes "interpolate $here please"
6580 qx backticks qx(/bin/ls -l)
6581 ` backticks `/bin/ls -l`
6582 qw quote words @EXPORT_OK = qw( func() $spam )
6583 m// regexp match m/this/
6584 s/// regexp substitute s/this/that/
6585 tr/// string transliterate tr/this/that/
6586 y/// string transliterate y/this/that/
6587 ($*@) sub prototypes sub foo ($)
6588 (stuff) sub attr parameters sub foo : attr(stuff)
6589 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6591 In most of these cases (all but <>, patterns and transliterate)
6592 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6593 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6594 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6597 It skips whitespace before the string starts, and treats the first
6598 character as the delimiter. If the delimiter is one of ([{< then
6599 the corresponding "close" character )]}> is used as the closing
6600 delimiter. It allows quoting of delimiters, and if the string has
6601 balanced delimiters ([{<>}]) it allows nesting.
6603 The lexer always reads these strings into lex_stuff, except in the
6604 case of the operators which take *two* arguments (s/// and tr///)
6605 when it checks to see if lex_stuff is full (presumably with the 1st
6606 arg to s or tr) and if so puts the string into lex_repl.
6611 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6613 SV *sv; /* scalar value: string */
6614 char *tmps; /* temp string, used for delimiter matching */
6615 register char *s = start; /* current position in the buffer */
6616 register char term; /* terminating character */
6617 register char *to; /* current position in the sv's data */
6618 I32 brackets = 1; /* bracket nesting level */
6619 bool has_utf8 = FALSE; /* is there any utf8 content? */
6621 /* skip space before the delimiter */
6625 /* mark where we are, in case we need to report errors */
6628 /* after skipping whitespace, the next character is the terminator */
6630 if ((term & 0x80) && UTF)
6633 /* mark where we are */
6634 PL_multi_start = CopLINE(PL_curcop);
6635 PL_multi_open = term;
6637 /* find corresponding closing delimiter */
6638 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6640 PL_multi_close = term;
6642 /* create a new SV to hold the contents. 87 is leak category, I'm
6643 assuming. 79 is the SV's initial length. What a random number. */
6645 sv_upgrade(sv, SVt_PVIV);
6647 (void)SvPOK_only(sv); /* validate pointer */
6649 /* move past delimiter and try to read a complete string */
6651 sv_catpvn(sv, s, 1);
6654 /* extend sv if need be */
6655 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6656 /* set 'to' to the next character in the sv's string */
6657 to = SvPVX(sv)+SvCUR(sv);
6659 /* if open delimiter is the close delimiter read unbridle */
6660 if (PL_multi_open == PL_multi_close) {
6661 for (; s < PL_bufend; s++,to++) {
6662 /* embedded newlines increment the current line number */
6663 if (*s == '\n' && !PL_rsfp)
6664 CopLINE_inc(PL_curcop);
6665 /* handle quoted delimiters */
6666 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6667 if (!keep_quoted && s[1] == term)
6669 /* any other quotes are simply copied straight through */
6673 /* terminate when run out of buffer (the for() condition), or
6674 have found the terminator */
6675 else if (*s == term)
6677 else if (!has_utf8 && (*s & 0x80) && UTF)
6683 /* if the terminator isn't the same as the start character (e.g.,
6684 matched brackets), we have to allow more in the quoting, and
6685 be prepared for nested brackets.
6688 /* read until we run out of string, or we find the terminator */
6689 for (; s < PL_bufend; s++,to++) {
6690 /* embedded newlines increment the line count */
6691 if (*s == '\n' && !PL_rsfp)
6692 CopLINE_inc(PL_curcop);
6693 /* backslashes can escape the open or closing characters */
6694 if (*s == '\\' && s+1 < PL_bufend) {
6696 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6701 /* allow nested opens and closes */
6702 else if (*s == PL_multi_close && --brackets <= 0)
6704 else if (*s == PL_multi_open)
6706 else if (!has_utf8 && (*s & 0x80) && UTF)
6711 /* terminate the copied string and update the sv's end-of-string */
6713 SvCUR_set(sv, to - SvPVX(sv));
6716 * this next chunk reads more into the buffer if we're not done yet
6720 break; /* handle case where we are done yet :-) */
6722 #ifndef PERL_STRICT_CR
6723 if (to - SvPVX(sv) >= 2) {
6724 if ((to[-2] == '\r' && to[-1] == '\n') ||
6725 (to[-2] == '\n' && to[-1] == '\r'))
6729 SvCUR_set(sv, to - SvPVX(sv));
6731 else if (to[-1] == '\r')
6734 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6738 /* if we're out of file, or a read fails, bail and reset the current
6739 line marker so we can report where the unterminated string began
6742 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6744 CopLINE_set(PL_curcop, PL_multi_start);
6747 /* we read a line, so increment our line counter */
6748 CopLINE_inc(PL_curcop);
6750 /* update debugger info */
6751 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6752 SV *sv = NEWSV(88,0);
6754 sv_upgrade(sv, SVt_PVMG);
6755 sv_setsv(sv,PL_linestr);
6756 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6759 /* having changed the buffer, we must update PL_bufend */
6760 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6763 /* at this point, we have successfully read the delimited string */
6766 sv_catpvn(sv, s, 1);
6769 PL_multi_end = CopLINE(PL_curcop);
6772 /* if we allocated too much space, give some back */
6773 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6774 SvLEN_set(sv, SvCUR(sv) + 1);
6775 Renew(SvPVX(sv), SvLEN(sv), char);
6778 /* decide whether this is the first or second quoted string we've read
6791 takes: pointer to position in buffer
6792 returns: pointer to new position in buffer
6793 side-effects: builds ops for the constant in yylval.op
6795 Read a number in any of the formats that Perl accepts:
6797 0(x[0-7A-F]+)|([0-7]+)|(b[01])
6798 [\d_]+(\.[\d_]*)?[Ee](\d+)
6800 Underbars (_) are allowed in decimal numbers. If -w is on,
6801 underbars before a decimal point must be at three digit intervals.
6803 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6806 If it reads a number without a decimal point or an exponent, it will
6807 try converting the number to an integer and see if it can do so
6808 without loss of precision.
6812 Perl_scan_num(pTHX_ char *start, YYSTYPE* lvalp)
6814 register char *s = start; /* current position in buffer */
6815 register char *d; /* destination in temp buffer */
6816 register char *e; /* end of temp buffer */
6817 NV nv; /* number read, as a double */
6818 SV *sv = Nullsv; /* place to put the converted number */
6819 bool floatit; /* boolean: int or float? */
6820 char *lastub = 0; /* position of last underbar */
6821 static char number_too_long[] = "Number too long";
6823 /* We use the first character to decide what type of number this is */
6827 Perl_croak(aTHX_ "panic: scan_num");
6829 /* if it starts with a 0, it could be an octal number, a decimal in
6830 0.13 disguise, or a hexadecimal number, or a binary number. */
6834 u holds the "number so far"
6835 shift the power of 2 of the base
6836 (hex == 4, octal == 3, binary == 1)
6837 overflowed was the number more than we can hold?
6839 Shift is used when we add a digit. It also serves as an "are
6840 we in octal/hex/binary?" indicator to disallow hex characters
6846 bool overflowed = FALSE;
6847 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6848 static char* bases[5] = { "", "binary", "", "octal",
6850 static char* Bases[5] = { "", "Binary", "", "Octal",
6852 static char *maxima[5] = { "",
6853 "0b11111111111111111111111111111111",
6857 char *base, *Base, *max;
6863 } else if (s[1] == 'b') {
6867 /* check for a decimal in disguise */
6868 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
6870 /* so it must be octal */
6874 base = bases[shift];
6875 Base = Bases[shift];
6876 max = maxima[shift];
6878 /* read the rest of the number */
6880 /* x is used in the overflow test,
6881 b is the digit we're adding on. */
6886 /* if we don't mention it, we're done */
6895 /* 8 and 9 are not octal */
6898 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
6902 case '2': case '3': case '4':
6903 case '5': case '6': case '7':
6905 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
6909 b = *s++ & 15; /* ASCII digit -> value of digit */
6913 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
6914 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
6915 /* make sure they said 0x */
6920 /* Prepare to put the digit we have onto the end
6921 of the number so far. We check for overflows.
6926 x = u << shift; /* make room for the digit */
6928 if ((x >> shift) != u
6929 && !(PL_hints & HINT_NEW_BINARY)) {
6932 if (ckWARN_d(WARN_OVERFLOW))
6933 Perl_warner(aTHX_ WARN_OVERFLOW,
6934 "Integer overflow in %s number",
6937 u = x | b; /* add the digit to the end */
6940 n *= nvshift[shift];
6941 /* If an NV has not enough bits in its
6942 * mantissa to represent an UV this summing of
6943 * small low-order numbers is a waste of time
6944 * (because the NV cannot preserve the
6945 * low-order bits anyway): we could just
6946 * remember when did we overflow and in the
6947 * end just multiply n by the right
6955 /* if we get here, we had success: make a scalar value from
6961 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
6962 Perl_warner(aTHX_ WARN_PORTABLE,
6963 "%s number > %s non-portable",
6969 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
6970 Perl_warner(aTHX_ WARN_PORTABLE,
6971 "%s number > %s non-portable",
6976 if (PL_hints & HINT_NEW_BINARY)
6977 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
6982 handle decimal numbers.
6983 we're also sent here when we read a 0 as the first digit
6985 case '1': case '2': case '3': case '4': case '5':
6986 case '6': case '7': case '8': case '9': case '.':
6989 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
6992 /* read next group of digits and _ and copy into d */
6993 while (isDIGIT(*s) || *s == '_') {
6994 /* skip underscores, checking for misplaced ones
6998 if (ckWARN(WARN_SYNTAX) && lastub && s - lastub != 3)
6999 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
7003 /* check for end of fixed-length buffer */
7005 Perl_croak(aTHX_ number_too_long);
7006 /* if we're ok, copy the character */
7011 /* final misplaced underbar check */
7012 if (lastub && s - lastub != 3) {
7013 if (ckWARN(WARN_SYNTAX))
7014 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
7017 /* read a decimal portion if there is one. avoid
7018 3..5 being interpreted as the number 3. followed
7021 if (*s == '.' && s[1] != '.') {
7025 /* copy, ignoring underbars, until we run out of
7026 digits. Note: no misplaced underbar checks!
7028 for (; isDIGIT(*s) || *s == '_'; s++) {
7029 /* fixed length buffer check */
7031 Perl_croak(aTHX_ number_too_long);
7035 if (*s == '.' && isDIGIT(s[1])) {
7036 /* oops, it's really a v-string, but without the "v" */
7042 /* read exponent part, if present */
7043 if (*s && strchr("eE",*s) && strchr("+-0123456789",s[1])) {
7047 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
7048 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
7050 /* allow positive or negative exponent */
7051 if (*s == '+' || *s == '-')
7054 /* read digits of exponent (no underbars :-) */
7055 while (isDIGIT(*s)) {
7057 Perl_croak(aTHX_ number_too_long);
7062 /* terminate the string */
7065 /* make an sv from the string */
7068 #if defined(Strtol) && defined(Strtoul)
7071 strtol/strtoll sets errno to ERANGE if the number is too big
7072 for an integer. We try to do an integer conversion first
7073 if no characters indicating "float" have been found.
7080 if (*PL_tokenbuf == '-')
7081 iv = Strtol(PL_tokenbuf, (char**)NULL, 10);
7083 uv = Strtoul(PL_tokenbuf, (char**)NULL, 10);
7085 floatit = TRUE; /* Probably just too large. */
7086 else if (*PL_tokenbuf == '-')
7088 else if (uv <= IV_MAX)
7089 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7094 nv = Atof(PL_tokenbuf);
7099 No working strtou?ll?.
7101 Unfortunately atol() doesn't do range checks (returning
7102 LONG_MIN/LONG_MAX, and setting errno to ERANGE on overflows)
7103 everywhere [1], so we cannot use use atol() (or atoll()).
7104 If we could, they would be used, as Atol(), very much like
7105 Strtol() and Strtoul() are used above.
7107 [1] XXX Configure test needed to check for atol()
7108 (and atoll()) overflow behaviour XXX
7112 We need to do this the hard way. */
7114 nv = Atof(PL_tokenbuf);
7116 /* See if we can make do with an integer value without loss of
7117 precision. We use U_V to cast to a UV, because some
7118 compilers have issues. Then we try casting it back and see
7119 if it was the same [1]. We only do this if we know we
7120 specifically read an integer. If floatit is true, then we
7121 don't need to do the conversion at all.
7123 [1] Note that this is lossy if our NVs cannot preserve our
7124 UVs. There are metaconfig defines NV_PRESERVES_UV (a boolean)
7125 and NV_PRESERVES_UV_BITS (a number), but in general we really
7126 do hope all such potentially lossy platforms have strtou?ll?
7127 to do a lossless IV/UV conversion.
7129 Maybe could do some tricks with DBL_DIG, LDBL_DIG and
7130 DBL_MANT_DIG and LDBL_MANT_DIG (these are already available
7131 as NV_DIG and NV_MANT_DIG)?
7137 if (!floatit && (NV)uv == nv) {
7139 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7147 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7148 (PL_hints & HINT_NEW_INTEGER) )
7149 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7150 (floatit ? "float" : "integer"),
7154 /* if it starts with a v, it could be a v-string */
7160 while (isDIGIT(*pos) || *pos == '_')
7162 if (!isALPHA(*pos)) {
7164 U8 tmpbuf[UTF8_MAXLEN+1];
7167 s++; /* get past 'v' */
7170 sv_setpvn(sv, "", 0);
7173 if (*s == '0' && isDIGIT(s[1]))
7174 yyerror("Octal number in vector unsupported");
7177 /* this is atoi() that tolerates underscores */
7180 while (--end >= s) {
7185 rev += (*end - '0') * mult;
7187 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
7188 Perl_warner(aTHX_ WARN_OVERFLOW,
7189 "Integer overflow in decimal number");
7192 tmpend = uv_to_utf8(tmpbuf, rev);
7193 utf8 = utf8 || rev > 127;
7194 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
7195 if (*pos == '.' && isDIGIT(pos[1]))
7201 while (isDIGIT(*pos) || *pos == '_')
7209 sv_utf8_downgrade(sv, TRUE);
7216 /* make the op for the constant and return */
7219 lvalp->opval = newSVOP(OP_CONST, 0, sv);
7221 lvalp->opval = Nullop;
7227 S_scan_formline(pTHX_ register char *s)
7231 SV *stuff = newSVpvn("",0);
7232 bool needargs = FALSE;
7235 if (*s == '.' || *s == /*{*/'}') {
7237 #ifdef PERL_STRICT_CR
7238 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
7240 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
7242 if (*t == '\n' || t == PL_bufend)
7245 if (PL_in_eval && !PL_rsfp) {
7246 eol = strchr(s,'\n');
7251 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7253 for (t = s; t < eol; t++) {
7254 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7256 goto enough; /* ~~ must be first line in formline */
7258 if (*t == '@' || *t == '^')
7261 sv_catpvn(stuff, s, eol-s);
7262 #ifndef PERL_STRICT_CR
7263 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7264 char *end = SvPVX(stuff) + SvCUR(stuff);
7273 s = filter_gets(PL_linestr, PL_rsfp, 0);
7274 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7275 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7278 yyerror("Format not terminated");
7288 PL_lex_state = LEX_NORMAL;
7289 PL_nextval[PL_nexttoke].ival = 0;
7293 PL_lex_state = LEX_FORMLINE;
7294 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7296 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7300 SvREFCNT_dec(stuff);
7301 PL_lex_formbrack = 0;
7312 PL_cshlen = strlen(PL_cshname);
7317 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7319 I32 oldsavestack_ix = PL_savestack_ix;
7320 CV* outsidecv = PL_compcv;
7324 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7326 SAVEI32(PL_subline);
7327 save_item(PL_subname);
7330 SAVESPTR(PL_comppad_name);
7331 SAVESPTR(PL_compcv);
7332 SAVEI32(PL_comppad_name_fill);
7333 SAVEI32(PL_min_intro_pending);
7334 SAVEI32(PL_max_intro_pending);
7335 SAVEI32(PL_pad_reset_pending);
7337 PL_compcv = (CV*)NEWSV(1104,0);
7338 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7339 CvFLAGS(PL_compcv) |= flags;
7341 PL_comppad = newAV();
7342 av_push(PL_comppad, Nullsv);
7343 PL_curpad = AvARRAY(PL_comppad);
7344 PL_comppad_name = newAV();
7345 PL_comppad_name_fill = 0;
7346 PL_min_intro_pending = 0;
7348 PL_subline = CopLINE(PL_curcop);
7350 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
7351 PL_curpad[0] = (SV*)newAV();
7352 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
7353 #endif /* USE_THREADS */
7355 comppadlist = newAV();
7356 AvREAL_off(comppadlist);
7357 av_store(comppadlist, 0, (SV*)PL_comppad_name);
7358 av_store(comppadlist, 1, (SV*)PL_comppad);
7360 CvPADLIST(PL_compcv) = comppadlist;
7361 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7363 CvOWNER(PL_compcv) = 0;
7364 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7365 MUTEX_INIT(CvMUTEXP(PL_compcv));
7366 #endif /* USE_THREADS */
7368 return oldsavestack_ix;
7372 Perl_yywarn(pTHX_ char *s)
7374 PL_in_eval |= EVAL_WARNONLY;
7376 PL_in_eval &= ~EVAL_WARNONLY;
7381 Perl_yyerror(pTHX_ char *s)
7384 char *context = NULL;
7388 if (!yychar || (yychar == ';' && !PL_rsfp))
7390 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7391 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7392 while (isSPACE(*PL_oldoldbufptr))
7394 context = PL_oldoldbufptr;
7395 contlen = PL_bufptr - PL_oldoldbufptr;
7397 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7398 PL_oldbufptr != PL_bufptr) {
7399 while (isSPACE(*PL_oldbufptr))
7401 context = PL_oldbufptr;
7402 contlen = PL_bufptr - PL_oldbufptr;
7404 else if (yychar > 255)
7405 where = "next token ???";
7406 #ifdef USE_PURE_BISON
7407 /* GNU Bison sets the value -2 */
7408 else if (yychar == -2) {
7410 else if ((yychar & 127) == 127) {
7412 if (PL_lex_state == LEX_NORMAL ||
7413 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7414 where = "at end of line";
7415 else if (PL_lex_inpat)
7416 where = "within pattern";
7418 where = "within string";
7421 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7423 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7424 else if (isPRINT_LC(yychar))
7425 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7427 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7428 where = SvPVX(where_sv);
7430 msg = sv_2mortal(newSVpv(s, 0));
7431 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7432 CopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7434 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7436 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7437 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7438 Perl_sv_catpvf(aTHX_ msg,
7439 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7440 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7443 if (PL_in_eval & EVAL_WARNONLY)
7444 Perl_warn(aTHX_ "%"SVf, msg);
7447 if (PL_error_count >= 10) {
7448 if (PL_in_eval && SvCUR(ERRSV))
7449 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
7450 ERRSV, CopFILE(PL_curcop));
7452 Perl_croak(aTHX_ "%s has too many errors.\n",
7453 CopFILE(PL_curcop));
7456 PL_in_my_stash = Nullhv;
7461 S_swallow_bom(pTHX_ U8 *s)
7464 slen = SvCUR(PL_linestr);
7468 /* UTF-16 little-endian */
7469 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
7470 Perl_croak(aTHX_ "Unsupported script encoding");
7471 #ifndef PERL_NO_UTF16_FILTER
7472 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-LE script encoding\n"));
7474 if (PL_bufend > (char*)s) {
7478 filter_add(utf16rev_textfilter, NULL);
7479 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7480 PL_bufend = (char*)utf16_to_utf8_reversed(s, news,
7481 PL_bufend - (char*)s - 1,
7483 Copy(news, s, newlen, U8);
7484 SvCUR_set(PL_linestr, newlen);
7485 PL_bufend = SvPVX(PL_linestr) + newlen;
7486 news[newlen++] = '\0';
7490 Perl_croak(aTHX_ "Unsupported script encoding");
7495 if (s[1] == 0xFF) { /* UTF-16 big-endian */
7496 #ifndef PERL_NO_UTF16_FILTER
7497 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding\n"));
7499 if (PL_bufend > (char *)s) {
7503 filter_add(utf16_textfilter, NULL);
7504 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7505 PL_bufend = (char*)utf16_to_utf8(s, news,
7506 PL_bufend - (char*)s,
7508 Copy(news, s, newlen, U8);
7509 SvCUR_set(PL_linestr, newlen);
7510 PL_bufend = SvPVX(PL_linestr) + newlen;
7511 news[newlen++] = '\0';
7515 Perl_croak(aTHX_ "Unsupported script encoding");
7520 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
7521 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-8 script encoding\n"));
7526 if (slen > 3 && s[1] == 0 && /* UTF-32 big-endian */
7527 s[2] == 0xFE && s[3] == 0xFF)
7529 Perl_croak(aTHX_ "Unsupported script encoding");
7541 * Restore a source filter.
7545 restore_rsfp(pTHXo_ void *f)
7547 PerlIO *fp = (PerlIO*)f;
7549 if (PL_rsfp == PerlIO_stdin())
7550 PerlIO_clearerr(PL_rsfp);
7551 else if (PL_rsfp && (PL_rsfp != fp))
7552 PerlIO_close(PL_rsfp);
7556 #ifndef PERL_NO_UTF16_FILTER
7558 utf16_textfilter(pTHXo_ int idx, SV *sv, int maxlen)
7560 I32 count = FILTER_READ(idx+1, sv, maxlen);
7565 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7566 if (!*SvPV_nolen(sv))
7567 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7570 tend = utf16_to_utf8((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7571 sv_usepvn(sv, (char*)tmps, tend - tmps);
7577 utf16rev_textfilter(pTHXo_ int idx, SV *sv, int maxlen)
7579 I32 count = FILTER_READ(idx+1, sv, maxlen);
7584 if (!*SvPV_nolen(sv))
7585 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7588 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7589 tend = utf16_to_utf8_reversed((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7590 sv_usepvn(sv, (char*)tmps, tend - tmps);