3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
4 * 2000, 2001, 2002, 2003, 2004, by Larry Wall and others
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
12 * "It all comes from here, the stench and the peril." --Frodo
16 * This file is the lexer for Perl. It's closely linked to the
19 * The main routine is yylex(), which returns the next token.
23 #define PERL_IN_TOKE_C
26 #define yychar (*PL_yycharp)
27 #define yylval (*PL_yylvalp)
29 static char ident_too_long[] = "Identifier too long";
30 static char c_without_g[] = "Use of /c modifier is meaningless without /g";
31 static char c_in_subst[] = "Use of /c modifier is meaningless in s///";
33 static void restore_rsfp(pTHX_ void *f);
34 #ifndef PERL_NO_UTF16_FILTER
35 static I32 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen);
36 static I32 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen);
39 #define XFAKEBRACK 128
42 #ifdef USE_UTF8_SCRIPTS
43 # define UTF (!IN_BYTES)
45 # define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || (PL_hints & HINT_UTF8))
48 /* In variables named $^X, these are the legal values for X.
49 * 1999-02-27 mjd-perl-patch@plover.com */
50 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
52 /* On MacOS, respect nonbreaking spaces */
53 #ifdef MACOS_TRADITIONAL
54 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
56 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
59 /* LEX_* are values for PL_lex_state, the state of the lexer.
60 * They are arranged oddly so that the guard on the switch statement
61 * can get by with a single comparison (if the compiler is smart enough).
64 /* #define LEX_NOTPARSING 11 is done in perl.h. */
67 #define LEX_INTERPNORMAL 9
68 #define LEX_INTERPCASEMOD 8
69 #define LEX_INTERPPUSH 7
70 #define LEX_INTERPSTART 6
71 #define LEX_INTERPEND 5
72 #define LEX_INTERPENDMAYBE 4
73 #define LEX_INTERPCONCAT 3
74 #define LEX_INTERPCONST 2
75 #define LEX_FORMLINE 1
76 #define LEX_KNOWNEXT 0
79 static char* lex_state_names[] = {
100 /* CLINE is a macro that ensures PL_copline has a sane value */
105 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
108 * Convenience functions to return different tokens and prime the
109 * lexer for the next token. They all take an argument.
111 * TOKEN : generic token (used for '(', DOLSHARP, etc)
112 * OPERATOR : generic operator
113 * AOPERATOR : assignment operator
114 * PREBLOCK : beginning the block after an if, while, foreach, ...
115 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
116 * PREREF : *EXPR where EXPR is not a simple identifier
117 * TERM : expression term
118 * LOOPX : loop exiting command (goto, last, dump, etc)
119 * FTST : file test operator
120 * FUN0 : zero-argument function
121 * FUN1 : not used, except for not, which isn't a UNIOP
122 * BOop : bitwise or or xor
124 * SHop : shift operator
125 * PWop : power operator
126 * PMop : pattern-matching operator
127 * Aop : addition-level operator
128 * Mop : multiplication-level operator
129 * Eop : equality-testing operator
130 * Rop : relational operator <= != gt
132 * Also see LOP and lop() below.
135 #ifdef DEBUGGING /* Serve -DT. */
136 # define REPORT(retval) tokereport(s,(int)retval)
138 # define REPORT(retval) (retval)
141 #define TOKEN(retval) return ( PL_bufptr = s, REPORT(retval))
142 #define OPERATOR(retval) return (PL_expect = XTERM, PL_bufptr = s, REPORT(retval))
143 #define AOPERATOR(retval) return ao((PL_expect = XTERM, PL_bufptr = s, REPORT(retval)))
144 #define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s, REPORT(retval))
145 #define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s, REPORT(retval))
146 #define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s, REPORT(retval))
147 #define TERM(retval) return (CLINE, PL_expect = XOPERATOR, PL_bufptr = s, REPORT(retval))
148 #define LOOPX(f) return (yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)LOOPEX))
149 #define FTST(f) return (yylval.ival=f, PL_expect=XTERMORDORDOR, PL_bufptr=s, REPORT((int)UNIOP))
150 #define FUN0(f) return (yylval.ival=f, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC0))
151 #define FUN1(f) return (yylval.ival=f, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC1))
152 #define BOop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)BITOROP)))
153 #define BAop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)BITANDOP)))
154 #define SHop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)SHIFTOP)))
155 #define PWop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)POWOP)))
156 #define PMop(f) return(yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)MATCHOP))
157 #define Aop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)ADDOP)))
158 #define Mop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)MULOP)))
159 #define Eop(f) return (yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)EQOP))
160 #define Rop(f) return (yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)RELOP))
162 /* This bit of chicanery makes a unary function followed by
163 * a parenthesis into a function with one argument, highest precedence.
164 * The UNIDOR macro is for unary functions that can be followed by the //
165 * operator (such as C<shift // 0>).
167 #define UNI2(f,x) return ( \
171 PL_last_uni = PL_oldbufptr, \
172 PL_last_lop_op = f, \
174 (*s == '(' || (s = skipspace(s), *s == '(') \
175 ? (int)FUNC1 : (int)UNIOP)))
176 #define UNI(f) UNI2(f,XTERM)
177 #define UNIDOR(f) UNI2(f,XTERMORDORDOR)
179 #define UNIBRACK(f) return ( \
182 PL_last_uni = PL_oldbufptr, \
184 (*s == '(' || (s = skipspace(s), *s == '(') \
185 ? (int)FUNC1 : (int)UNIOP)))
187 /* grandfather return to old style */
188 #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
192 /* how to interpret the yylval associated with the token */
196 TOKENTYPE_OPNUM, /* yylval.ival contains an opcode number */
202 static struct debug_tokens { int token, type; char *name;} debug_tokens[] =
204 { ADDOP, TOKENTYPE_OPNUM, "ADDOP" },
205 { ANDAND, TOKENTYPE_NONE, "ANDAND" },
206 { ANDOP, TOKENTYPE_NONE, "ANDOP" },
207 { ANONSUB, TOKENTYPE_IVAL, "ANONSUB" },
208 { ARROW, TOKENTYPE_NONE, "ARROW" },
209 { ASSIGNOP, TOKENTYPE_OPNUM, "ASSIGNOP" },
210 { BITANDOP, TOKENTYPE_OPNUM, "BITANDOP" },
211 { BITOROP, TOKENTYPE_OPNUM, "BITOROP" },
212 { COLONATTR, TOKENTYPE_NONE, "COLONATTR" },
213 { CONTINUE, TOKENTYPE_NONE, "CONTINUE" },
214 { DO, TOKENTYPE_NONE, "DO" },
215 { DOLSHARP, TOKENTYPE_NONE, "DOLSHARP" },
216 { DORDOR, TOKENTYPE_NONE, "DORDOR" },
217 { DOROP, TOKENTYPE_OPNUM, "DOROP" },
218 { DOTDOT, TOKENTYPE_IVAL, "DOTDOT" },
219 { ELSE, TOKENTYPE_NONE, "ELSE" },
220 { ELSIF, TOKENTYPE_IVAL, "ELSIF" },
221 { EQOP, TOKENTYPE_OPNUM, "EQOP" },
222 { FOR, TOKENTYPE_IVAL, "FOR" },
223 { FORMAT, TOKENTYPE_NONE, "FORMAT" },
224 { FUNC, TOKENTYPE_OPNUM, "FUNC" },
225 { FUNC0, TOKENTYPE_OPNUM, "FUNC0" },
226 { FUNC0SUB, TOKENTYPE_OPVAL, "FUNC0SUB" },
227 { FUNC1, TOKENTYPE_OPNUM, "FUNC1" },
228 { FUNCMETH, TOKENTYPE_OPVAL, "FUNCMETH" },
229 { HASHBRACK, TOKENTYPE_NONE, "HASHBRACK" },
230 { IF, TOKENTYPE_IVAL, "IF" },
231 { LABEL, TOKENTYPE_PVAL, "LABEL" },
232 { LOCAL, TOKENTYPE_IVAL, "LOCAL" },
233 { LOOPEX, TOKENTYPE_OPNUM, "LOOPEX" },
234 { LSTOP, TOKENTYPE_OPNUM, "LSTOP" },
235 { LSTOPSUB, TOKENTYPE_OPVAL, "LSTOPSUB" },
236 { MATCHOP, TOKENTYPE_OPNUM, "MATCHOP" },
237 { METHOD, TOKENTYPE_OPVAL, "METHOD" },
238 { MULOP, TOKENTYPE_OPNUM, "MULOP" },
239 { MY, TOKENTYPE_IVAL, "MY" },
240 { MYSUB, TOKENTYPE_NONE, "MYSUB" },
241 { NOAMP, TOKENTYPE_NONE, "NOAMP" },
242 { NOTOP, TOKENTYPE_NONE, "NOTOP" },
243 { OROP, TOKENTYPE_IVAL, "OROP" },
244 { OROR, TOKENTYPE_NONE, "OROR" },
245 { PACKAGE, TOKENTYPE_NONE, "PACKAGE" },
246 { PMFUNC, TOKENTYPE_OPVAL, "PMFUNC" },
247 { POSTDEC, TOKENTYPE_NONE, "POSTDEC" },
248 { POSTINC, TOKENTYPE_NONE, "POSTINC" },
249 { POWOP, TOKENTYPE_OPNUM, "POWOP" },
250 { PREDEC, TOKENTYPE_NONE, "PREDEC" },
251 { PREINC, TOKENTYPE_NONE, "PREINC" },
252 { PRIVATEREF, TOKENTYPE_OPVAL, "PRIVATEREF" },
253 { REFGEN, TOKENTYPE_NONE, "REFGEN" },
254 { RELOP, TOKENTYPE_OPNUM, "RELOP" },
255 { SHIFTOP, TOKENTYPE_OPNUM, "SHIFTOP" },
256 { SUB, TOKENTYPE_NONE, "SUB" },
257 { THING, TOKENTYPE_OPVAL, "THING" },
258 { UMINUS, TOKENTYPE_NONE, "UMINUS" },
259 { UNIOP, TOKENTYPE_OPNUM, "UNIOP" },
260 { UNIOPSUB, TOKENTYPE_OPVAL, "UNIOPSUB" },
261 { UNLESS, TOKENTYPE_IVAL, "UNLESS" },
262 { UNTIL, TOKENTYPE_IVAL, "UNTIL" },
263 { USE, TOKENTYPE_IVAL, "USE" },
264 { WHILE, TOKENTYPE_IVAL, "WHILE" },
265 { WORD, TOKENTYPE_OPVAL, "WORD" },
266 { 0, TOKENTYPE_NONE, 0 }
269 /* dump the returned token in rv, plus any optional arg in yylval */
272 S_tokereport(pTHX_ char* s, I32 rv)
276 enum token_type type = TOKENTYPE_NONE;
277 struct debug_tokens *p;
278 SV* report = NEWSV(0, 60);
280 Perl_sv_catpvf(aTHX_ report, "<== ");
282 for (p = debug_tokens; p->token; p++) {
283 if (p->token == (int)rv) {
290 Perl_sv_catpvf(aTHX_ report, "%s", name);
291 else if ((char)rv > ' ' && (char)rv < '~')
292 Perl_sv_catpvf(aTHX_ report, "'%c'", (char)rv);
294 Perl_sv_catpvf(aTHX_ report, "EOF");
296 Perl_sv_catpvf(aTHX_ report, "?? %"IVdf, (IV)rv);
299 case TOKENTYPE_GVVAL: /* doesn't appear to be used */
302 Perl_sv_catpvf(aTHX_ report, "(ival=%"IVdf")", yylval.ival);
304 case TOKENTYPE_OPNUM:
305 Perl_sv_catpvf(aTHX_ report, "(ival=op_%s)",
306 PL_op_name[yylval.ival]);
309 Perl_sv_catpvf(aTHX_ report, "(pval=\"%s\")", yylval.pval);
311 case TOKENTYPE_OPVAL:
313 Perl_sv_catpvf(aTHX_ report, "(opval=op_%s)",
314 PL_op_name[yylval.opval->op_type]);
316 Perl_sv_catpv(aTHX_ report, "(opval=null)");
319 Perl_sv_catpvf(aTHX_ report, " at line %d [", CopLINE(PL_curcop));
320 if (s - PL_bufptr > 0)
321 sv_catpvn(report, PL_bufptr, s - PL_bufptr);
323 if (PL_oldbufptr && *PL_oldbufptr)
324 sv_catpv(report, PL_tokenbuf);
326 PerlIO_printf(Perl_debug_log, "### %s]\n", SvPV_nolen(report));
336 * This subroutine detects &&=, ||=, and //= and turns an ANDAND, OROR or DORDOR
337 * into an OP_ANDASSIGN, OP_ORASSIGN, or OP_DORASSIGN
341 S_ao(pTHX_ int toketype)
343 if (*PL_bufptr == '=') {
345 if (toketype == ANDAND)
346 yylval.ival = OP_ANDASSIGN;
347 else if (toketype == OROR)
348 yylval.ival = OP_ORASSIGN;
349 else if (toketype == DORDOR)
350 yylval.ival = OP_DORASSIGN;
358 * When Perl expects an operator and finds something else, no_op
359 * prints the warning. It always prints "<something> found where
360 * operator expected. It prints "Missing semicolon on previous line?"
361 * if the surprise occurs at the start of the line. "do you need to
362 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
363 * where the compiler doesn't know if foo is a method call or a function.
364 * It prints "Missing operator before end of line" if there's nothing
365 * after the missing operator, or "... before <...>" if there is something
366 * after the missing operator.
370 S_no_op(pTHX_ char *what, char *s)
372 char *oldbp = PL_bufptr;
373 bool is_first = (PL_oldbufptr == PL_linestart);
379 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
380 if (ckWARN_d(WARN_SYNTAX)) {
382 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
383 "\t(Missing semicolon on previous line?)\n");
384 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
386 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy_if(t,UTF) || *t == ':'); t++) ;
387 if (t < PL_bufptr && isSPACE(*t))
388 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
389 "\t(Do you need to predeclare %.*s?)\n",
390 t - PL_oldoldbufptr, PL_oldoldbufptr);
394 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
395 "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp);
403 * Complain about missing quote/regexp/heredoc terminator.
404 * If it's called with (char *)NULL then it cauterizes the line buffer.
405 * If we're in a delimited string and the delimiter is a control
406 * character, it's reformatted into a two-char sequence like ^C.
411 S_missingterm(pTHX_ char *s)
416 char *nl = strrchr(s,'\n');
422 iscntrl(PL_multi_close)
424 PL_multi_close < 32 || PL_multi_close == 127
428 tmpbuf[1] = toCTRL(PL_multi_close);
434 *tmpbuf = (char)PL_multi_close;
438 q = strchr(s,'"') ? '\'' : '"';
439 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
447 Perl_deprecate(pTHX_ char *s)
449 if (ckWARN(WARN_DEPRECATED))
450 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED), "Use of %s is deprecated", s);
454 Perl_deprecate_old(pTHX_ char *s)
456 /* This function should NOT be called for any new deprecated warnings */
457 /* Use Perl_deprecate instead */
459 /* It is here to maintain backward compatibility with the pre-5.8 */
460 /* warnings category hierarchy. The "deprecated" category used to */
461 /* live under the "syntax" category. It is now a top-level category */
462 /* in its own right. */
464 if (ckWARN2(WARN_DEPRECATED, WARN_SYNTAX))
465 Perl_warner(aTHX_ packWARN2(WARN_DEPRECATED, WARN_SYNTAX),
466 "Use of %s is deprecated", s);
471 * Deprecate a comma-less variable list.
477 deprecate_old("comma-less variable list");
481 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
482 * utf16-to-utf8-reversed.
485 #ifdef PERL_CR_FILTER
489 register char *s = SvPVX(sv);
490 register char *e = s + SvCUR(sv);
491 /* outer loop optimized to do nothing if there are no CR-LFs */
493 if (*s++ == '\r' && *s == '\n') {
494 /* hit a CR-LF, need to copy the rest */
495 register char *d = s - 1;
498 if (*s == '\r' && s[1] == '\n')
509 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
511 I32 count = FILTER_READ(idx+1, sv, maxlen);
512 if (count > 0 && !maxlen)
520 * Initialize variables. Uses the Perl save_stack to save its state (for
521 * recursive calls to the parser).
525 Perl_lex_start(pTHX_ SV *line)
530 SAVEI32(PL_lex_dojoin);
531 SAVEI32(PL_lex_brackets);
532 SAVEI32(PL_lex_casemods);
533 SAVEI32(PL_lex_starts);
534 SAVEI32(PL_lex_state);
535 SAVEVPTR(PL_lex_inpat);
536 SAVEI32(PL_lex_inwhat);
537 if (PL_lex_state == LEX_KNOWNEXT) {
538 I32 toke = PL_nexttoke;
539 while (--toke >= 0) {
540 SAVEI32(PL_nexttype[toke]);
541 SAVEVPTR(PL_nextval[toke]);
543 SAVEI32(PL_nexttoke);
545 SAVECOPLINE(PL_curcop);
548 SAVEPPTR(PL_oldbufptr);
549 SAVEPPTR(PL_oldoldbufptr);
550 SAVEPPTR(PL_last_lop);
551 SAVEPPTR(PL_last_uni);
552 SAVEPPTR(PL_linestart);
553 SAVESPTR(PL_linestr);
554 SAVEGENERICPV(PL_lex_brackstack);
555 SAVEGENERICPV(PL_lex_casestack);
556 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
557 SAVESPTR(PL_lex_stuff);
558 SAVEI32(PL_lex_defer);
559 SAVEI32(PL_sublex_info.sub_inwhat);
560 SAVESPTR(PL_lex_repl);
562 SAVEINT(PL_lex_expect);
564 PL_lex_state = LEX_NORMAL;
568 New(899, PL_lex_brackstack, 120, char);
569 New(899, PL_lex_casestack, 12, char);
571 *PL_lex_casestack = '\0';
574 PL_lex_stuff = Nullsv;
575 PL_lex_repl = Nullsv;
579 PL_sublex_info.sub_inwhat = 0;
581 if (SvREADONLY(PL_linestr))
582 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
583 s = SvPV(PL_linestr, len);
584 if (!len || s[len-1] != ';') {
585 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
586 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
587 sv_catpvn(PL_linestr, "\n;", 2);
589 SvTEMP_off(PL_linestr);
590 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
591 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
592 PL_last_lop = PL_last_uni = Nullch;
598 * Finalizer for lexing operations. Must be called when the parser is
599 * done with the lexer.
605 PL_doextract = FALSE;
610 * This subroutine has nothing to do with tilting, whether at windmills
611 * or pinball tables. Its name is short for "increment line". It
612 * increments the current line number in CopLINE(PL_curcop) and checks
613 * to see whether the line starts with a comment of the form
614 * # line 500 "foo.pm"
615 * If so, it sets the current line number and file to the values in the comment.
619 S_incline(pTHX_ char *s)
626 CopLINE_inc(PL_curcop);
629 while (SPACE_OR_TAB(*s)) s++;
630 if (strnEQ(s, "line", 4))
634 if (SPACE_OR_TAB(*s))
638 while (SPACE_OR_TAB(*s)) s++;
644 while (SPACE_OR_TAB(*s))
646 if (*s == '"' && (t = strchr(s+1, '"'))) {
651 for (t = s; !isSPACE(*t); t++) ;
654 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
656 if (*e != '\n' && *e != '\0')
657 return; /* false alarm */
662 CopFILE_free(PL_curcop);
663 CopFILE_set(PL_curcop, s);
666 CopLINE_set(PL_curcop, atoi(n)-1);
671 * Called to gobble the appropriate amount and type of whitespace.
672 * Skips comments as well.
676 S_skipspace(pTHX_ register char *s)
678 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
679 while (s < PL_bufend && SPACE_OR_TAB(*s))
685 SSize_t oldprevlen, oldoldprevlen;
686 SSize_t oldloplen = 0, oldunilen = 0;
687 while (s < PL_bufend && isSPACE(*s)) {
688 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
693 if (s < PL_bufend && *s == '#') {
694 while (s < PL_bufend && *s != '\n')
698 if (PL_in_eval && !PL_rsfp) {
705 /* only continue to recharge the buffer if we're at the end
706 * of the buffer, we're not reading from a source filter, and
707 * we're in normal lexing mode
709 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
710 PL_lex_state == LEX_FORMLINE)
713 /* try to recharge the buffer */
714 if ((s = filter_gets(PL_linestr, PL_rsfp,
715 (prevlen = SvCUR(PL_linestr)))) == Nullch)
717 /* end of file. Add on the -p or -n magic */
718 if (PL_minus_n || PL_minus_p) {
719 sv_setpv(PL_linestr,PL_minus_p ?
720 ";}continue{print or die qq(-p destination: $!\\n)" :
722 sv_catpv(PL_linestr,";}");
723 PL_minus_n = PL_minus_p = 0;
726 sv_setpv(PL_linestr,";");
728 /* reset variables for next time we lex */
729 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
731 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
732 PL_last_lop = PL_last_uni = Nullch;
734 /* Close the filehandle. Could be from -P preprocessor,
735 * STDIN, or a regular file. If we were reading code from
736 * STDIN (because the commandline held no -e or filename)
737 * then we don't close it, we reset it so the code can
738 * read from STDIN too.
741 if (PL_preprocess && !PL_in_eval)
742 (void)PerlProc_pclose(PL_rsfp);
743 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
744 PerlIO_clearerr(PL_rsfp);
746 (void)PerlIO_close(PL_rsfp);
751 /* not at end of file, so we only read another line */
752 /* make corresponding updates to old pointers, for yyerror() */
753 oldprevlen = PL_oldbufptr - PL_bufend;
754 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
756 oldunilen = PL_last_uni - PL_bufend;
758 oldloplen = PL_last_lop - PL_bufend;
759 PL_linestart = PL_bufptr = s + prevlen;
760 PL_bufend = s + SvCUR(PL_linestr);
762 PL_oldbufptr = s + oldprevlen;
763 PL_oldoldbufptr = s + oldoldprevlen;
765 PL_last_uni = s + oldunilen;
767 PL_last_lop = s + oldloplen;
770 /* debugger active and we're not compiling the debugger code,
771 * so store the line into the debugger's array of lines
773 if (PERLDB_LINE && PL_curstash != PL_debstash) {
774 SV *sv = NEWSV(85,0);
776 sv_upgrade(sv, SVt_PVMG);
777 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
780 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
787 * Check the unary operators to ensure there's no ambiguity in how they're
788 * used. An ambiguous piece of code would be:
790 * This doesn't mean rand() + 5. Because rand() is a unary operator,
791 * the +5 is its argument.
800 if (PL_oldoldbufptr != PL_last_uni)
802 while (isSPACE(*PL_last_uni))
804 for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ;
805 if ((t = strchr(s, '(')) && t < PL_bufptr)
807 if (ckWARN_d(WARN_AMBIGUOUS)){
810 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
811 "Warning: Use of \"%s\" without parentheses is ambiguous",
818 * LOP : macro to build a list operator. Its behaviour has been replaced
819 * with a subroutine, S_lop() for which LOP is just another name.
822 #define LOP(f,x) return lop(f,x,s)
826 * Build a list operator (or something that might be one). The rules:
827 * - if we have a next token, then it's a list operator [why?]
828 * - if the next thing is an opening paren, then it's a function
829 * - else it's a list operator
833 S_lop(pTHX_ I32 f, int x, char *s)
839 PL_last_lop = PL_oldbufptr;
840 PL_last_lop_op = (OPCODE)f;
842 return REPORT(LSTOP);
849 return REPORT(LSTOP);
854 * When the lexer realizes it knows the next token (for instance,
855 * it is reordering tokens for the parser) then it can call S_force_next
856 * to know what token to return the next time the lexer is called. Caller
857 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
858 * handles the token correctly.
862 S_force_next(pTHX_ I32 type)
864 PL_nexttype[PL_nexttoke] = type;
866 if (PL_lex_state != LEX_KNOWNEXT) {
867 PL_lex_defer = PL_lex_state;
868 PL_lex_expect = PL_expect;
869 PL_lex_state = LEX_KNOWNEXT;
875 * When the lexer knows the next thing is a word (for instance, it has
876 * just seen -> and it knows that the next char is a word char, then
877 * it calls S_force_word to stick the next word into the PL_next lookahead.
880 * char *start : buffer position (must be within PL_linestr)
881 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
882 * int check_keyword : if true, Perl checks to make sure the word isn't
883 * a keyword (do this if the word is a label, e.g. goto FOO)
884 * int allow_pack : if true, : characters will also be allowed (require,
886 * int allow_initial_tick : used by the "sub" lexer only.
890 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
895 start = skipspace(start);
897 if (isIDFIRST_lazy_if(s,UTF) ||
898 (allow_pack && *s == ':') ||
899 (allow_initial_tick && *s == '\'') )
901 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
902 if (check_keyword && keyword(PL_tokenbuf, len))
904 if (token == METHOD) {
909 PL_expect = XOPERATOR;
912 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0));
913 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
914 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
915 SvUTF8_on(((SVOP*)PL_nextval[PL_nexttoke].opval)->op_sv);
923 * Called when the lexer wants $foo *foo &foo etc, but the program
924 * text only contains the "foo" portion. The first argument is a pointer
925 * to the "foo", and the second argument is the type symbol to prefix.
926 * Forces the next token to be a "WORD".
927 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
931 S_force_ident(pTHX_ register char *s, int kind)
934 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
935 PL_nextval[PL_nexttoke].opval = o;
938 o->op_private = OPpCONST_ENTERED;
939 /* XXX see note in pp_entereval() for why we forgo typo
940 warnings if the symbol must be introduced in an eval.
942 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
943 kind == '$' ? SVt_PV :
944 kind == '@' ? SVt_PVAV :
945 kind == '%' ? SVt_PVHV :
953 Perl_str_to_version(pTHX_ SV *sv)
958 char *start = SvPVx(sv,len);
959 bool utf = SvUTF8(sv) ? TRUE : FALSE;
960 char *end = start + len;
961 while (start < end) {
965 n = utf8n_to_uvchr((U8*)start, len, &skip, 0);
970 retval += ((NV)n)/nshift;
979 * Forces the next token to be a version number.
980 * If the next token appears to be an invalid version number, (e.g. "v2b"),
981 * and if "guessing" is TRUE, then no new token is created (and the caller
982 * must use an alternative parsing method).
986 S_force_version(pTHX_ char *s, int guessing)
988 OP *version = Nullop;
997 while (isDIGIT(*d) || *d == '_' || *d == '.')
999 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
1001 s = scan_num(s, &yylval);
1002 version = yylval.opval;
1003 ver = cSVOPx(version)->op_sv;
1004 if (SvPOK(ver) && !SvNIOK(ver)) {
1005 (void)SvUPGRADE(ver, SVt_PVNV);
1006 SvNVX(ver) = str_to_version(ver);
1007 SvNOK_on(ver); /* hint that it is a version */
1014 /* NOTE: The parser sees the package name and the VERSION swapped */
1015 PL_nextval[PL_nexttoke].opval = version;
1023 * Tokenize a quoted string passed in as an SV. It finds the next
1024 * chunk, up to end of string or a backslash. It may make a new
1025 * SV containing that chunk (if HINT_NEW_STRING is on). It also
1030 S_tokeq(pTHX_ SV *sv)
1033 register char *send;
1041 s = SvPV_force(sv, len);
1042 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
1045 while (s < send && *s != '\\')
1050 if ( PL_hints & HINT_NEW_STRING ) {
1051 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
1057 if (s + 1 < send && (s[1] == '\\'))
1058 s++; /* all that, just for this */
1063 SvCUR_set(sv, d - SvPVX(sv));
1065 if ( PL_hints & HINT_NEW_STRING )
1066 return new_constant(NULL, 0, "q", sv, pv, "q");
1071 * Now come three functions related to double-quote context,
1072 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
1073 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
1074 * interact with PL_lex_state, and create fake ( ... ) argument lists
1075 * to handle functions and concatenation.
1076 * They assume that whoever calls them will be setting up a fake
1077 * join call, because each subthing puts a ',' after it. This lets
1080 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
1082 * (I'm not sure whether the spurious commas at the end of lcfirst's
1083 * arguments and join's arguments are created or not).
1088 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
1090 * Pattern matching will set PL_lex_op to the pattern-matching op to
1091 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
1093 * OP_CONST and OP_READLINE are easy--just make the new op and return.
1095 * Everything else becomes a FUNC.
1097 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
1098 * had an OP_CONST or OP_READLINE). This just sets us up for a
1099 * call to S_sublex_push().
1103 S_sublex_start(pTHX)
1105 register I32 op_type = yylval.ival;
1107 if (op_type == OP_NULL) {
1108 yylval.opval = PL_lex_op;
1112 if (op_type == OP_CONST || op_type == OP_READLINE) {
1113 SV *sv = tokeq(PL_lex_stuff);
1115 if (SvTYPE(sv) == SVt_PVIV) {
1116 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
1122 nsv = newSVpvn(p, len);
1128 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
1129 PL_lex_stuff = Nullsv;
1130 /* Allow <FH> // "foo" */
1131 if (op_type == OP_READLINE)
1132 PL_expect = XTERMORDORDOR;
1136 PL_sublex_info.super_state = PL_lex_state;
1137 PL_sublex_info.sub_inwhat = op_type;
1138 PL_sublex_info.sub_op = PL_lex_op;
1139 PL_lex_state = LEX_INTERPPUSH;
1143 yylval.opval = PL_lex_op;
1153 * Create a new scope to save the lexing state. The scope will be
1154 * ended in S_sublex_done. Returns a '(', starting the function arguments
1155 * to the uc, lc, etc. found before.
1156 * Sets PL_lex_state to LEX_INTERPCONCAT.
1164 PL_lex_state = PL_sublex_info.super_state;
1165 SAVEI32(PL_lex_dojoin);
1166 SAVEI32(PL_lex_brackets);
1167 SAVEI32(PL_lex_casemods);
1168 SAVEI32(PL_lex_starts);
1169 SAVEI32(PL_lex_state);
1170 SAVEVPTR(PL_lex_inpat);
1171 SAVEI32(PL_lex_inwhat);
1172 SAVECOPLINE(PL_curcop);
1173 SAVEPPTR(PL_bufptr);
1174 SAVEPPTR(PL_bufend);
1175 SAVEPPTR(PL_oldbufptr);
1176 SAVEPPTR(PL_oldoldbufptr);
1177 SAVEPPTR(PL_last_lop);
1178 SAVEPPTR(PL_last_uni);
1179 SAVEPPTR(PL_linestart);
1180 SAVESPTR(PL_linestr);
1181 SAVEGENERICPV(PL_lex_brackstack);
1182 SAVEGENERICPV(PL_lex_casestack);
1184 PL_linestr = PL_lex_stuff;
1185 PL_lex_stuff = Nullsv;
1187 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1188 = SvPVX(PL_linestr);
1189 PL_bufend += SvCUR(PL_linestr);
1190 PL_last_lop = PL_last_uni = Nullch;
1191 SAVEFREESV(PL_linestr);
1193 PL_lex_dojoin = FALSE;
1194 PL_lex_brackets = 0;
1195 New(899, PL_lex_brackstack, 120, char);
1196 New(899, PL_lex_casestack, 12, char);
1197 PL_lex_casemods = 0;
1198 *PL_lex_casestack = '\0';
1200 PL_lex_state = LEX_INTERPCONCAT;
1201 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
1203 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1204 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1205 PL_lex_inpat = PL_sublex_info.sub_op;
1207 PL_lex_inpat = Nullop;
1214 * Restores lexer state after a S_sublex_push.
1220 if (!PL_lex_starts++) {
1221 SV *sv = newSVpvn("",0);
1222 if (SvUTF8(PL_linestr))
1224 PL_expect = XOPERATOR;
1225 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1229 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1230 PL_lex_state = LEX_INTERPCASEMOD;
1234 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1235 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1236 PL_linestr = PL_lex_repl;
1238 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1239 PL_bufend += SvCUR(PL_linestr);
1240 PL_last_lop = PL_last_uni = Nullch;
1241 SAVEFREESV(PL_linestr);
1242 PL_lex_dojoin = FALSE;
1243 PL_lex_brackets = 0;
1244 PL_lex_casemods = 0;
1245 *PL_lex_casestack = '\0';
1247 if (SvEVALED(PL_lex_repl)) {
1248 PL_lex_state = LEX_INTERPNORMAL;
1250 /* we don't clear PL_lex_repl here, so that we can check later
1251 whether this is an evalled subst; that means we rely on the
1252 logic to ensure sublex_done() is called again only via the
1253 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1256 PL_lex_state = LEX_INTERPCONCAT;
1257 PL_lex_repl = Nullsv;
1263 PL_bufend = SvPVX(PL_linestr);
1264 PL_bufend += SvCUR(PL_linestr);
1265 PL_expect = XOPERATOR;
1266 PL_sublex_info.sub_inwhat = 0;
1274 Extracts a pattern, double-quoted string, or transliteration. This
1277 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1278 processing a pattern (PL_lex_inpat is true), a transliteration
1279 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1281 Returns a pointer to the character scanned up to. Iff this is
1282 advanced from the start pointer supplied (ie if anything was
1283 successfully parsed), will leave an OP for the substring scanned
1284 in yylval. Caller must intuit reason for not parsing further
1285 by looking at the next characters herself.
1289 double-quoted style: \r and \n
1290 regexp special ones: \D \s
1292 backrefs: \1 (deprecated in substitution replacements)
1293 case and quoting: \U \Q \E
1294 stops on @ and $, but not for $ as tail anchor
1296 In transliterations:
1297 characters are VERY literal, except for - not at the start or end
1298 of the string, which indicates a range. scan_const expands the
1299 range to the full set of intermediate characters.
1301 In double-quoted strings:
1303 double-quoted style: \r and \n
1305 backrefs: \1 (deprecated)
1306 case and quoting: \U \Q \E
1309 scan_const does *not* construct ops to handle interpolated strings.
1310 It stops processing as soon as it finds an embedded $ or @ variable
1311 and leaves it to the caller to work out what's going on.
1313 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @::foo.
1315 $ in pattern could be $foo or could be tail anchor. Assumption:
1316 it's a tail anchor if $ is the last thing in the string, or if it's
1317 followed by one of ")| \n\t"
1319 \1 (backreferences) are turned into $1
1321 The structure of the code is
1322 while (there's a character to process) {
1323 handle transliteration ranges
1324 skip regexp comments
1325 skip # initiated comments in //x patterns
1326 check for embedded @foo
1327 check for embedded scalars
1329 leave intact backslashes from leave (below)
1330 deprecate \1 in strings and sub replacements
1331 handle string-changing backslashes \l \U \Q \E, etc.
1332 switch (what was escaped) {
1333 handle - in a transliteration (becomes a literal -)
1334 handle \132 octal characters
1335 handle 0x15 hex characters
1336 handle \cV (control V)
1337 handle printf backslashes (\f, \r, \n, etc)
1339 } (end if backslash)
1340 } (end while character to read)
1345 S_scan_const(pTHX_ char *start)
1347 register char *send = PL_bufend; /* end of the constant */
1348 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1349 register char *s = start; /* start of the constant */
1350 register char *d = SvPVX(sv); /* destination for copies */
1351 bool dorange = FALSE; /* are we in a translit range? */
1352 bool didrange = FALSE; /* did we just finish a range? */
1353 I32 has_utf8 = FALSE; /* Output constant is UTF8 */
1354 I32 this_utf8 = UTF; /* The source string is assumed to be UTF8 */
1357 const char *leaveit = /* set of acceptably-backslashed characters */
1359 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxz0123456789[{]} \t\n\r\f\v#"
1362 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1363 /* If we are doing a trans and we know we want UTF8 set expectation */
1364 has_utf8 = PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF);
1365 this_utf8 = PL_sublex_info.sub_op->op_private & (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1369 while (s < send || dorange) {
1370 /* get transliterations out of the way (they're most literal) */
1371 if (PL_lex_inwhat == OP_TRANS) {
1372 /* expand a range A-Z to the full set of characters. AIE! */
1374 I32 i; /* current expanded character */
1375 I32 min; /* first character in range */
1376 I32 max; /* last character in range */
1379 char *c = (char*)utf8_hop((U8*)d, -1);
1383 *c = (char)UTF_TO_NATIVE(0xff);
1384 /* mark the range as done, and continue */
1390 i = d - SvPVX(sv); /* remember current offset */
1391 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1392 d = SvPVX(sv) + i; /* refresh d after realloc */
1393 d -= 2; /* eat the first char and the - */
1395 min = (U8)*d; /* first char in range */
1396 max = (U8)d[1]; /* last char in range */
1400 "Invalid range \"%c-%c\" in transliteration operator",
1401 (char)min, (char)max);
1405 if ((isLOWER(min) && isLOWER(max)) ||
1406 (isUPPER(min) && isUPPER(max))) {
1408 for (i = min; i <= max; i++)
1410 *d++ = NATIVE_TO_NEED(has_utf8,i);
1412 for (i = min; i <= max; i++)
1414 *d++ = NATIVE_TO_NEED(has_utf8,i);
1419 for (i = min; i <= max; i++)
1422 /* mark the range as done, and continue */
1428 /* range begins (ignore - as first or last char) */
1429 else if (*s == '-' && s+1 < send && s != start) {
1431 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
1434 *d++ = (char)UTF_TO_NATIVE(0xff); /* use illegal utf8 byte--see pmtrans */
1446 /* if we get here, we're not doing a transliteration */
1448 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1449 except for the last char, which will be done separately. */
1450 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1452 while (s+1 < send && *s != ')')
1453 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1455 else if (s[2] == '{' /* This should match regcomp.c */
1456 || ((s[2] == 'p' || s[2] == '?') && s[3] == '{'))
1459 char *regparse = s + (s[2] == '{' ? 3 : 4);
1462 while (count && (c = *regparse)) {
1463 if (c == '\\' && regparse[1])
1471 if (*regparse != ')')
1472 regparse--; /* Leave one char for continuation. */
1473 while (s < regparse)
1474 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1478 /* likewise skip #-initiated comments in //x patterns */
1479 else if (*s == '#' && PL_lex_inpat &&
1480 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1481 while (s+1 < send && *s != '\n')
1482 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1485 /* check for embedded arrays
1486 (@foo, @::foo, @'foo, @{foo}, @$foo, @+, @-)
1488 else if (*s == '@' && s[1]
1489 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$+-", s[1])))
1492 /* check for embedded scalars. only stop if we're sure it's a
1495 else if (*s == '$') {
1496 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1498 if (s + 1 < send && !strchr("()| \r\n\t", s[1]))
1499 break; /* in regexp, $ might be tail anchor */
1502 /* End of else if chain - OP_TRANS rejoin rest */
1505 if (*s == '\\' && s+1 < send) {
1508 /* some backslashes we leave behind */
1509 if (*leaveit && *s && strchr(leaveit, *s)) {
1510 *d++ = NATIVE_TO_NEED(has_utf8,'\\');
1511 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1515 /* deprecate \1 in strings and substitution replacements */
1516 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1517 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1519 if (ckWARN(WARN_SYNTAX))
1520 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "\\%c better written as $%c", *s, *s);
1525 /* string-change backslash escapes */
1526 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1531 /* if we get here, it's either a quoted -, or a digit */
1534 /* quoted - in transliterations */
1536 if (PL_lex_inwhat == OP_TRANS) {
1543 if (ckWARN(WARN_MISC) &&
1546 Perl_warner(aTHX_ packWARN(WARN_MISC),
1547 "Unrecognized escape \\%c passed through",
1549 /* default action is to copy the quoted character */
1550 goto default_action;
1553 /* \132 indicates an octal constant */
1554 case '0': case '1': case '2': case '3':
1555 case '4': case '5': case '6': case '7':
1559 uv = grok_oct(s, &len, &flags, NULL);
1562 goto NUM_ESCAPE_INSERT;
1564 /* \x24 indicates a hex constant */
1568 char* e = strchr(s, '}');
1569 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
1570 PERL_SCAN_DISALLOW_PREFIX;
1575 yyerror("Missing right brace on \\x{}");
1579 uv = grok_hex(s, &len, &flags, NULL);
1585 I32 flags = PERL_SCAN_DISALLOW_PREFIX;
1586 uv = grok_hex(s, &len, &flags, NULL);
1592 /* Insert oct or hex escaped character.
1593 * There will always enough room in sv since such
1594 * escapes will be longer than any UTF-8 sequence
1595 * they can end up as. */
1597 /* We need to map to chars to ASCII before doing the tests
1600 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(uv))) {
1601 if (!has_utf8 && uv > 255) {
1602 /* Might need to recode whatever we have
1603 * accumulated so far if it contains any
1606 * (Can't we keep track of that and avoid
1607 * this rescan? --jhi)
1611 for (c = (U8 *) SvPVX(sv); c < (U8 *)d; c++) {
1612 if (!NATIVE_IS_INVARIANT(*c)) {
1617 STRLEN offset = d - SvPVX(sv);
1619 d = SvGROW(sv, SvLEN(sv) + hicount + 1) + offset;
1623 while (src >= (U8 *)SvPVX(sv)) {
1624 if (!NATIVE_IS_INVARIANT(*src)) {
1625 U8 ch = NATIVE_TO_ASCII(*src);
1626 *dst-- = (U8)UTF8_EIGHT_BIT_LO(ch);
1627 *dst-- = (U8)UTF8_EIGHT_BIT_HI(ch);
1637 if (has_utf8 || uv > 255) {
1638 d = (char*)uvchr_to_utf8((U8*)d, uv);
1640 if (PL_lex_inwhat == OP_TRANS &&
1641 PL_sublex_info.sub_op) {
1642 PL_sublex_info.sub_op->op_private |=
1643 (PL_lex_repl ? OPpTRANS_FROM_UTF
1656 /* \N{LATIN SMALL LETTER A} is a named character */
1660 char* e = strchr(s, '}');
1666 yyerror("Missing right brace on \\N{}");
1670 if (e > s + 2 && s[1] == 'U' && s[2] == '+') {
1672 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
1673 PERL_SCAN_DISALLOW_PREFIX;
1676 uv = grok_hex(s, &len, &flags, NULL);
1678 goto NUM_ESCAPE_INSERT;
1680 res = newSVpvn(s + 1, e - s - 1);
1681 res = new_constant( Nullch, 0, "charnames",
1682 res, Nullsv, "\\N{...}" );
1684 sv_utf8_upgrade(res);
1685 str = SvPV(res,len);
1686 #ifdef EBCDIC_NEVER_MIND
1687 /* charnames uses pack U and that has been
1688 * recently changed to do the below uni->native
1689 * mapping, so this would be redundant (and wrong,
1690 * the code point would be doubly converted).
1691 * But leave this in just in case the pack U change
1692 * gets revoked, but the semantics is still
1693 * desireable for charnames. --jhi */
1695 UV uv = utf8_to_uvchr((U8*)str, 0);
1698 U8 tmpbuf[UTF8_MAXLEN+1], *d;
1700 d = uvchr_to_utf8(tmpbuf, UNI_TO_NATIVE(uv));
1701 sv_setpvn(res, (char *)tmpbuf, d - tmpbuf);
1702 str = SvPV(res, len);
1706 if (!has_utf8 && SvUTF8(res)) {
1707 char *ostart = SvPVX(sv);
1708 SvCUR_set(sv, d - ostart);
1711 sv_utf8_upgrade(sv);
1712 /* this just broke our allocation above... */
1713 SvGROW(sv, (STRLEN)(send - start));
1714 d = SvPVX(sv) + SvCUR(sv);
1717 if (len > (STRLEN)(e - s + 4)) { /* I _guess_ 4 is \N{} --jhi */
1718 char *odest = SvPVX(sv);
1720 SvGROW(sv, (SvLEN(sv) + len - (e - s + 4)));
1721 d = SvPVX(sv) + (d - odest);
1723 Copy(str, d, len, char);
1730 yyerror("Missing braces on \\N{}");
1733 /* \c is a control character */
1742 *d++ = NATIVE_TO_NEED(has_utf8,toCTRL(c));
1745 yyerror("Missing control char name in \\c");
1749 /* printf-style backslashes, formfeeds, newlines, etc */
1751 *d++ = NATIVE_TO_NEED(has_utf8,'\b');
1754 *d++ = NATIVE_TO_NEED(has_utf8,'\n');
1757 *d++ = NATIVE_TO_NEED(has_utf8,'\r');
1760 *d++ = NATIVE_TO_NEED(has_utf8,'\f');
1763 *d++ = NATIVE_TO_NEED(has_utf8,'\t');
1766 *d++ = ASCII_TO_NEED(has_utf8,'\033');
1769 *d++ = ASCII_TO_NEED(has_utf8,'\007');
1775 } /* end if (backslash) */
1778 /* If we started with encoded form, or already know we want it
1779 and then encode the next character */
1780 if ((has_utf8 || this_utf8) && !NATIVE_IS_INVARIANT((U8)(*s))) {
1782 UV uv = (this_utf8) ? utf8n_to_uvchr((U8*)s, send - s, &len, 0) : (UV) ((U8) *s);
1783 STRLEN need = UNISKIP(NATIVE_TO_UNI(uv));
1786 /* encoded value larger than old, need extra space (NOTE: SvCUR() not set here) */
1787 STRLEN off = d - SvPVX(sv);
1788 d = SvGROW(sv, SvLEN(sv) + (need-len)) + off;
1790 d = (char*)uvchr_to_utf8((U8*)d, uv);
1794 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1796 } /* while loop to process each character */
1798 /* terminate the string and set up the sv */
1800 SvCUR_set(sv, d - SvPVX(sv));
1801 if (SvCUR(sv) >= SvLEN(sv))
1802 Perl_croak(aTHX_ "panic: constant overflowed allocated space");
1805 if (PL_encoding && !has_utf8) {
1806 sv_recode_to_utf8(sv, PL_encoding);
1812 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1813 PL_sublex_info.sub_op->op_private |=
1814 (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1818 /* shrink the sv if we allocated more than we used */
1819 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1820 SvLEN_set(sv, SvCUR(sv) + 1);
1821 Renew(SvPVX(sv), SvLEN(sv), char);
1824 /* return the substring (via yylval) only if we parsed anything */
1825 if (s > PL_bufptr) {
1826 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1827 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1829 ( PL_lex_inwhat == OP_TRANS
1831 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1834 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1841 * Returns TRUE if there's more to the expression (e.g., a subscript),
1844 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1846 * ->[ and ->{ return TRUE
1847 * { and [ outside a pattern are always subscripts, so return TRUE
1848 * if we're outside a pattern and it's not { or [, then return FALSE
1849 * if we're in a pattern and the first char is a {
1850 * {4,5} (any digits around the comma) returns FALSE
1851 * if we're in a pattern and the first char is a [
1853 * [SOMETHING] has a funky algorithm to decide whether it's a
1854 * character class or not. It has to deal with things like
1855 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1856 * anything else returns TRUE
1859 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1862 S_intuit_more(pTHX_ register char *s)
1864 if (PL_lex_brackets)
1866 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1868 if (*s != '{' && *s != '[')
1873 /* In a pattern, so maybe we have {n,m}. */
1890 /* On the other hand, maybe we have a character class */
1893 if (*s == ']' || *s == '^')
1896 /* this is terrifying, and it works */
1897 int weight = 2; /* let's weigh the evidence */
1899 unsigned char un_char = 255, last_un_char;
1900 char *send = strchr(s,']');
1901 char tmpbuf[sizeof PL_tokenbuf * 4];
1903 if (!send) /* has to be an expression */
1906 Zero(seen,256,char);
1909 else if (isDIGIT(*s)) {
1911 if (isDIGIT(s[1]) && s[2] == ']')
1917 for (; s < send; s++) {
1918 last_un_char = un_char;
1919 un_char = (unsigned char)*s;
1924 weight -= seen[un_char] * 10;
1925 if (isALNUM_lazy_if(s+1,UTF)) {
1926 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1927 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1932 else if (*s == '$' && s[1] &&
1933 strchr("[#!%*<>()-=",s[1])) {
1934 if (/*{*/ strchr("])} =",s[2]))
1943 if (strchr("wds]",s[1]))
1945 else if (seen['\''] || seen['"'])
1947 else if (strchr("rnftbxcav",s[1]))
1949 else if (isDIGIT(s[1])) {
1951 while (s[1] && isDIGIT(s[1]))
1961 if (strchr("aA01! ",last_un_char))
1963 if (strchr("zZ79~",s[1]))
1965 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1966 weight -= 5; /* cope with negative subscript */
1969 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1970 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1975 if (keyword(tmpbuf, d - tmpbuf))
1978 if (un_char == last_un_char + 1)
1980 weight -= seen[un_char];
1985 if (weight >= 0) /* probably a character class */
1995 * Does all the checking to disambiguate
1997 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1998 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
2000 * First argument is the stuff after the first token, e.g. "bar".
2002 * Not a method if bar is a filehandle.
2003 * Not a method if foo is a subroutine prototyped to take a filehandle.
2004 * Not a method if it's really "Foo $bar"
2005 * Method if it's "foo $bar"
2006 * Not a method if it's really "print foo $bar"
2007 * Method if it's really "foo package::" (interpreted as package->foo)
2008 * Not a method if bar is known to be a subroutine ("sub bar; foo bar")
2009 * Not a method if bar is a filehandle or package, but is quoted with
2014 S_intuit_method(pTHX_ char *start, GV *gv)
2016 char *s = start + (*start == '$');
2017 char tmpbuf[sizeof PL_tokenbuf];
2025 if ((cv = GvCVu(gv))) {
2026 char *proto = SvPVX(cv);
2036 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
2037 /* start is the beginning of the possible filehandle/object,
2038 * and s is the end of it
2039 * tmpbuf is a copy of it
2042 if (*start == '$') {
2043 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
2048 return *s == '(' ? FUNCMETH : METHOD;
2050 if (!keyword(tmpbuf, len)) {
2051 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
2056 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
2057 if (indirgv && GvCVu(indirgv))
2059 /* filehandle or package name makes it a method */
2060 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
2062 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
2063 return 0; /* no assumptions -- "=>" quotes bearword */
2065 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
2066 newSVpvn(tmpbuf,len));
2067 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
2071 return *s == '(' ? FUNCMETH : METHOD;
2079 * Return a string of Perl code to load the debugger. If PERL5DB
2080 * is set, it will return the contents of that, otherwise a
2081 * compile-time require of perl5db.pl.
2088 char *pdb = PerlEnv_getenv("PERL5DB");
2092 SETERRNO(0,SS_NORMAL);
2093 return "BEGIN { require 'perl5db.pl' }";
2099 /* Encoded script support. filter_add() effectively inserts a
2100 * 'pre-processing' function into the current source input stream.
2101 * Note that the filter function only applies to the current source file
2102 * (e.g., it will not affect files 'require'd or 'use'd by this one).
2104 * The datasv parameter (which may be NULL) can be used to pass
2105 * private data to this instance of the filter. The filter function
2106 * can recover the SV using the FILTER_DATA macro and use it to
2107 * store private buffers and state information.
2109 * The supplied datasv parameter is upgraded to a PVIO type
2110 * and the IoDIRP/IoANY field is used to store the function pointer,
2111 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
2112 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
2113 * private use must be set using malloc'd pointers.
2117 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
2122 if (!PL_rsfp_filters)
2123 PL_rsfp_filters = newAV();
2125 datasv = NEWSV(255,0);
2126 if (!SvUPGRADE(datasv, SVt_PVIO))
2127 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
2128 IoANY(datasv) = (void *)funcp; /* stash funcp into spare field */
2129 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
2130 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
2131 (void*)funcp, SvPV_nolen(datasv)));
2132 av_unshift(PL_rsfp_filters, 1);
2133 av_store(PL_rsfp_filters, 0, datasv) ;
2138 /* Delete most recently added instance of this filter function. */
2140 Perl_filter_del(pTHX_ filter_t funcp)
2143 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", (void*)funcp));
2144 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
2146 /* if filter is on top of stack (usual case) just pop it off */
2147 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
2148 if (IoANY(datasv) == (void *)funcp) {
2149 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
2150 IoANY(datasv) = (void *)NULL;
2151 sv_free(av_pop(PL_rsfp_filters));
2155 /* we need to search for the correct entry and clear it */
2156 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
2160 /* Invoke the n'th filter function for the current rsfp. */
2162 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
2165 /* 0 = read one text line */
2170 if (!PL_rsfp_filters)
2172 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
2173 /* Provide a default input filter to make life easy. */
2174 /* Note that we append to the line. This is handy. */
2175 DEBUG_P(PerlIO_printf(Perl_debug_log,
2176 "filter_read %d: from rsfp\n", idx));
2180 int old_len = SvCUR(buf_sv) ;
2182 /* ensure buf_sv is large enough */
2183 SvGROW(buf_sv, (STRLEN)(old_len + maxlen)) ;
2184 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
2185 if (PerlIO_error(PL_rsfp))
2186 return -1; /* error */
2188 return 0 ; /* end of file */
2190 SvCUR_set(buf_sv, old_len + len) ;
2193 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
2194 if (PerlIO_error(PL_rsfp))
2195 return -1; /* error */
2197 return 0 ; /* end of file */
2200 return SvCUR(buf_sv);
2202 /* Skip this filter slot if filter has been deleted */
2203 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
2204 DEBUG_P(PerlIO_printf(Perl_debug_log,
2205 "filter_read %d: skipped (filter deleted)\n",
2207 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
2209 /* Get function pointer hidden within datasv */
2210 funcp = (filter_t)IoANY(datasv);
2211 DEBUG_P(PerlIO_printf(Perl_debug_log,
2212 "filter_read %d: via function %p (%s)\n",
2213 idx, (void*)funcp, SvPV_nolen(datasv)));
2214 /* Call function. The function is expected to */
2215 /* call "FILTER_READ(idx+1, buf_sv)" first. */
2216 /* Return: <0:error, =0:eof, >0:not eof */
2217 return (*funcp)(aTHX_ idx, buf_sv, maxlen);
2221 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
2223 #ifdef PERL_CR_FILTER
2224 if (!PL_rsfp_filters) {
2225 filter_add(S_cr_textfilter,NULL);
2228 if (PL_rsfp_filters) {
2231 SvCUR_set(sv, 0); /* start with empty line */
2232 if (FILTER_READ(0, sv, 0) > 0)
2233 return ( SvPVX(sv) ) ;
2238 return (sv_gets(sv, fp, append));
2242 S_find_in_my_stash(pTHX_ char *pkgname, I32 len)
2246 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2250 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2251 (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV)))
2253 return GvHV(gv); /* Foo:: */
2256 /* use constant CLASS => 'MyClass' */
2257 if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) {
2259 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2260 pkgname = SvPV_nolen(sv);
2264 return gv_stashpv(pkgname, FALSE);
2268 static char* exp_name[] =
2269 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2270 "ATTRTERM", "TERMBLOCK", "TERMORDORDOR"
2277 Works out what to call the token just pulled out of the input
2278 stream. The yacc parser takes care of taking the ops we return and
2279 stitching them into a tree.
2285 if read an identifier
2286 if we're in a my declaration
2287 croak if they tried to say my($foo::bar)
2288 build the ops for a my() declaration
2289 if it's an access to a my() variable
2290 are we in a sort block?
2291 croak if my($a); $a <=> $b
2292 build ops for access to a my() variable
2293 if in a dq string, and they've said @foo and we can't find @foo
2295 build ops for a bareword
2296 if we already built the token before, use it.
2301 #pragma segment Perl_yylex
2313 I32 orig_keyword = 0;
2316 PerlIO_printf(Perl_debug_log, "### LEX_%s\n",
2317 lex_state_names[PL_lex_state]);
2319 /* check if there's an identifier for us to look at */
2320 if (PL_pending_ident)
2321 return REPORT(S_pending_ident(aTHX));
2323 /* no identifier pending identification */
2325 switch (PL_lex_state) {
2327 case LEX_NORMAL: /* Some compilers will produce faster */
2328 case LEX_INTERPNORMAL: /* code if we comment these out. */
2332 /* when we've already built the next token, just pull it out of the queue */
2335 yylval = PL_nextval[PL_nexttoke];
2337 PL_lex_state = PL_lex_defer;
2338 PL_expect = PL_lex_expect;
2339 PL_lex_defer = LEX_NORMAL;
2341 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2342 "### Next token after '%s' was known, type %"IVdf"\n", PL_bufptr,
2343 (IV)PL_nexttype[PL_nexttoke]); });
2345 return REPORT(PL_nexttype[PL_nexttoke]);
2347 /* interpolated case modifiers like \L \U, including \Q and \E.
2348 when we get here, PL_bufptr is at the \
2350 case LEX_INTERPCASEMOD:
2352 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2353 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2355 /* handle \E or end of string */
2356 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2360 if (PL_lex_casemods) {
2361 oldmod = PL_lex_casestack[--PL_lex_casemods];
2362 PL_lex_casestack[PL_lex_casemods] = '\0';
2364 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2366 PL_lex_state = LEX_INTERPCONCAT;
2370 if (PL_bufptr != PL_bufend)
2372 PL_lex_state = LEX_INTERPCONCAT;
2376 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2377 "### Saw case modifier at '%s'\n", PL_bufptr); });
2379 if (s[1] == '\\' && s[2] == 'E') {
2381 PL_lex_state = LEX_INTERPCONCAT;
2385 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2386 tmp = *s, *s = s[2], s[2] = (char)tmp; /* misordered... */
2387 if (strchr("LU", *s) &&
2388 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U'))) {
2389 PL_lex_casestack[--PL_lex_casemods] = '\0';
2392 if (PL_lex_casemods > 10)
2393 Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2394 PL_lex_casestack[PL_lex_casemods++] = *s;
2395 PL_lex_casestack[PL_lex_casemods] = '\0';
2396 PL_lex_state = LEX_INTERPCONCAT;
2397 PL_nextval[PL_nexttoke].ival = 0;
2400 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2402 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2404 PL_nextval[PL_nexttoke].ival = OP_LC;
2406 PL_nextval[PL_nexttoke].ival = OP_UC;
2408 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2410 Perl_croak(aTHX_ "panic: yylex");
2414 if (PL_lex_starts) {
2423 case LEX_INTERPPUSH:
2424 return REPORT(sublex_push());
2426 case LEX_INTERPSTART:
2427 if (PL_bufptr == PL_bufend)
2428 return REPORT(sublex_done());
2429 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2430 "### Interpolated variable at '%s'\n", PL_bufptr); });
2432 PL_lex_dojoin = (*PL_bufptr == '@');
2433 PL_lex_state = LEX_INTERPNORMAL;
2434 if (PL_lex_dojoin) {
2435 PL_nextval[PL_nexttoke].ival = 0;
2437 force_ident("\"", '$');
2438 PL_nextval[PL_nexttoke].ival = 0;
2440 PL_nextval[PL_nexttoke].ival = 0;
2442 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2445 if (PL_lex_starts++) {
2451 case LEX_INTERPENDMAYBE:
2452 if (intuit_more(PL_bufptr)) {
2453 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2459 if (PL_lex_dojoin) {
2460 PL_lex_dojoin = FALSE;
2461 PL_lex_state = LEX_INTERPCONCAT;
2464 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2465 && SvEVALED(PL_lex_repl))
2467 if (PL_bufptr != PL_bufend)
2468 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2469 PL_lex_repl = Nullsv;
2472 case LEX_INTERPCONCAT:
2474 if (PL_lex_brackets)
2475 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2477 if (PL_bufptr == PL_bufend)
2478 return REPORT(sublex_done());
2480 if (SvIVX(PL_linestr) == '\'') {
2481 SV *sv = newSVsv(PL_linestr);
2484 else if ( PL_hints & HINT_NEW_RE )
2485 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2486 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2490 s = scan_const(PL_bufptr);
2492 PL_lex_state = LEX_INTERPCASEMOD;
2494 PL_lex_state = LEX_INTERPSTART;
2497 if (s != PL_bufptr) {
2498 PL_nextval[PL_nexttoke] = yylval;
2501 if (PL_lex_starts++)
2511 PL_lex_state = LEX_NORMAL;
2512 s = scan_formline(PL_bufptr);
2513 if (!PL_lex_formbrack)
2519 PL_oldoldbufptr = PL_oldbufptr;
2522 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at [%s]\n",
2523 exp_name[PL_expect], s);
2529 if (isIDFIRST_lazy_if(s,UTF))
2531 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2534 goto fake_eof; /* emulate EOF on ^D or ^Z */
2539 if (PL_lex_brackets) {
2540 if (PL_lex_formbrack)
2541 yyerror("Format not terminated");
2543 yyerror("Missing right curly or square bracket");
2545 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2546 "### Tokener got EOF\n");
2550 if (s++ < PL_bufend)
2551 goto retry; /* ignore stray nulls */
2554 if (!PL_in_eval && !PL_preambled) {
2555 PL_preambled = TRUE;
2556 sv_setpv(PL_linestr,incl_perldb());
2557 if (SvCUR(PL_linestr))
2558 sv_catpv(PL_linestr,";");
2560 while(AvFILLp(PL_preambleav) >= 0) {
2561 SV *tmpsv = av_shift(PL_preambleav);
2562 sv_catsv(PL_linestr, tmpsv);
2563 sv_catpv(PL_linestr, ";");
2566 sv_free((SV*)PL_preambleav);
2567 PL_preambleav = NULL;
2569 if (PL_minus_n || PL_minus_p) {
2570 sv_catpv(PL_linestr, "LINE: while (<>) {");
2572 sv_catpv(PL_linestr,"chomp;");
2575 if (strchr("/'\"", *PL_splitstr)
2576 && strchr(PL_splitstr + 1, *PL_splitstr))
2577 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s);", PL_splitstr);
2580 s = "'~#\200\1'"; /* surely one char is unused...*/
2581 while (s[1] && strchr(PL_splitstr, *s)) s++;
2583 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s%c",
2584 "q" + (delim == '\''), delim);
2585 for (s = PL_splitstr; *s; s++) {
2587 sv_catpvn(PL_linestr, "\\", 1);
2588 sv_catpvn(PL_linestr, s, 1);
2590 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2594 sv_catpv(PL_linestr,"our @F=split(' ');");
2597 sv_catpv(PL_linestr, "\n");
2598 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2599 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2600 PL_last_lop = PL_last_uni = Nullch;
2601 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2602 SV *sv = NEWSV(85,0);
2604 sv_upgrade(sv, SVt_PVMG);
2605 sv_setsv(sv,PL_linestr);
2608 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2613 bof = PL_rsfp ? TRUE : FALSE;
2614 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2617 if (PL_preprocess && !PL_in_eval)
2618 (void)PerlProc_pclose(PL_rsfp);
2619 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2620 PerlIO_clearerr(PL_rsfp);
2622 (void)PerlIO_close(PL_rsfp);
2624 PL_doextract = FALSE;
2626 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2627 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2628 sv_catpv(PL_linestr,";}");
2629 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2630 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2631 PL_last_lop = PL_last_uni = Nullch;
2632 PL_minus_n = PL_minus_p = 0;
2635 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2636 PL_last_lop = PL_last_uni = Nullch;
2637 sv_setpv(PL_linestr,"");
2638 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2640 /* If it looks like the start of a BOM or raw UTF-16,
2641 * check if it in fact is. */
2647 #ifdef PERLIO_IS_STDIO
2648 # ifdef __GNU_LIBRARY__
2649 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
2650 # define FTELL_FOR_PIPE_IS_BROKEN
2654 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
2655 # define FTELL_FOR_PIPE_IS_BROKEN
2660 #ifdef FTELL_FOR_PIPE_IS_BROKEN
2661 /* This loses the possibility to detect the bof
2662 * situation on perl -P when the libc5 is being used.
2663 * Workaround? Maybe attach some extra state to PL_rsfp?
2666 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
2668 bof = PerlIO_tell(PL_rsfp) == (Off_t)SvCUR(PL_linestr);
2671 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2672 s = swallow_bom((U8*)s);
2676 /* Incest with pod. */
2677 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2678 sv_setpv(PL_linestr, "");
2679 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2680 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2681 PL_last_lop = PL_last_uni = Nullch;
2682 PL_doextract = FALSE;
2686 } while (PL_doextract);
2687 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2688 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2689 SV *sv = NEWSV(85,0);
2691 sv_upgrade(sv, SVt_PVMG);
2692 sv_setsv(sv,PL_linestr);
2695 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2697 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2698 PL_last_lop = PL_last_uni = Nullch;
2699 if (CopLINE(PL_curcop) == 1) {
2700 while (s < PL_bufend && isSPACE(*s))
2702 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2706 if (*s == '#' && *(s+1) == '!')
2708 #ifdef ALTERNATE_SHEBANG
2710 static char as[] = ALTERNATE_SHEBANG;
2711 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2712 d = s + (sizeof(as) - 1);
2714 #endif /* ALTERNATE_SHEBANG */
2723 while (*d && !isSPACE(*d))
2727 #ifdef ARG_ZERO_IS_SCRIPT
2728 if (ipathend > ipath) {
2730 * HP-UX (at least) sets argv[0] to the script name,
2731 * which makes $^X incorrect. And Digital UNIX and Linux,
2732 * at least, set argv[0] to the basename of the Perl
2733 * interpreter. So, having found "#!", we'll set it right.
2735 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV)); /* $^X */
2736 assert(SvPOK(x) || SvGMAGICAL(x));
2737 if (sv_eq(x, CopFILESV(PL_curcop))) {
2738 sv_setpvn(x, ipath, ipathend - ipath);
2744 char *bstart = SvPV(CopFILESV(PL_curcop),blen);
2745 char *lstart = SvPV(x,llen);
2747 bstart += blen - llen;
2748 if (strnEQ(bstart, lstart, llen) && bstart[-1] == '/') {
2749 sv_setpvn(x, ipath, ipathend - ipath);
2754 TAINT_NOT; /* $^X is always tainted, but that's OK */
2756 #endif /* ARG_ZERO_IS_SCRIPT */
2761 d = instr(s,"perl -");
2763 d = instr(s,"perl");
2765 /* avoid getting into infinite loops when shebang
2766 * line contains "Perl" rather than "perl" */
2768 for (d = ipathend-4; d >= ipath; --d) {
2769 if ((*d == 'p' || *d == 'P')
2770 && !ibcmp(d, "perl", 4))
2780 #ifdef ALTERNATE_SHEBANG
2782 * If the ALTERNATE_SHEBANG on this system starts with a
2783 * character that can be part of a Perl expression, then if
2784 * we see it but not "perl", we're probably looking at the
2785 * start of Perl code, not a request to hand off to some
2786 * other interpreter. Similarly, if "perl" is there, but
2787 * not in the first 'word' of the line, we assume the line
2788 * contains the start of the Perl program.
2790 if (d && *s != '#') {
2792 while (*c && !strchr("; \t\r\n\f\v#", *c))
2795 d = Nullch; /* "perl" not in first word; ignore */
2797 *s = '#'; /* Don't try to parse shebang line */
2799 #endif /* ALTERNATE_SHEBANG */
2800 #ifndef MACOS_TRADITIONAL
2805 !instr(s,"indir") &&
2806 instr(PL_origargv[0],"perl"))
2812 while (s < PL_bufend && isSPACE(*s))
2814 if (s < PL_bufend) {
2815 Newz(899,newargv,PL_origargc+3,char*);
2817 while (s < PL_bufend && !isSPACE(*s))
2820 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2823 newargv = PL_origargv;
2826 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
2828 Perl_croak(aTHX_ "Can't exec %s", ipath);
2832 U32 oldpdb = PL_perldb;
2833 bool oldn = PL_minus_n;
2834 bool oldp = PL_minus_p;
2836 while (*d && !isSPACE(*d)) d++;
2837 while (SPACE_OR_TAB(*d)) d++;
2840 bool switches_done = PL_doswitches;
2842 if (*d == 'M' || *d == 'm') {
2844 while (*d && !isSPACE(*d)) d++;
2845 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2848 d = moreswitches(d);
2850 if (PL_doswitches && !switches_done) {
2851 int argc = PL_origargc;
2852 char **argv = PL_origargv;
2855 } while (argc && argv[0][0] == '-' && argv[0][1]);
2856 init_argv_symbols(argc,argv);
2858 if ((PERLDB_LINE && !oldpdb) ||
2859 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2860 /* if we have already added "LINE: while (<>) {",
2861 we must not do it again */
2863 sv_setpv(PL_linestr, "");
2864 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2865 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2866 PL_last_lop = PL_last_uni = Nullch;
2867 PL_preambled = FALSE;
2869 (void)gv_fetchfile(PL_origfilename);
2872 if (PL_doswitches && !switches_done) {
2873 int argc = PL_origargc;
2874 char **argv = PL_origargv;
2877 } while (argc && argv[0][0] == '-' && argv[0][1]);
2878 init_argv_symbols(argc,argv);
2884 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2886 PL_lex_state = LEX_FORMLINE;
2891 #ifdef PERL_STRICT_CR
2892 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2894 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
2896 case ' ': case '\t': case '\f': case 013:
2897 #ifdef MACOS_TRADITIONAL
2904 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2905 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2906 /* handle eval qq[#line 1 "foo"\n ...] */
2907 CopLINE_dec(PL_curcop);
2911 while (s < d && *s != '\n')
2915 else if (s > d) /* Found by Ilya: feed random input to Perl. */
2916 Perl_croak(aTHX_ "panic: input overflow");
2918 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2920 PL_lex_state = LEX_FORMLINE;
2930 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2937 while (s < PL_bufend && SPACE_OR_TAB(*s))
2940 if (strnEQ(s,"=>",2)) {
2941 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2942 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2943 "### Saw unary minus before =>, forcing word '%s'\n", s);
2945 OPERATOR('-'); /* unary minus */
2947 PL_last_uni = PL_oldbufptr;
2949 case 'r': ftst = OP_FTEREAD; break;
2950 case 'w': ftst = OP_FTEWRITE; break;
2951 case 'x': ftst = OP_FTEEXEC; break;
2952 case 'o': ftst = OP_FTEOWNED; break;
2953 case 'R': ftst = OP_FTRREAD; break;
2954 case 'W': ftst = OP_FTRWRITE; break;
2955 case 'X': ftst = OP_FTREXEC; break;
2956 case 'O': ftst = OP_FTROWNED; break;
2957 case 'e': ftst = OP_FTIS; break;
2958 case 'z': ftst = OP_FTZERO; break;
2959 case 's': ftst = OP_FTSIZE; break;
2960 case 'f': ftst = OP_FTFILE; break;
2961 case 'd': ftst = OP_FTDIR; break;
2962 case 'l': ftst = OP_FTLINK; break;
2963 case 'p': ftst = OP_FTPIPE; break;
2964 case 'S': ftst = OP_FTSOCK; break;
2965 case 'u': ftst = OP_FTSUID; break;
2966 case 'g': ftst = OP_FTSGID; break;
2967 case 'k': ftst = OP_FTSVTX; break;
2968 case 'b': ftst = OP_FTBLK; break;
2969 case 'c': ftst = OP_FTCHR; break;
2970 case 't': ftst = OP_FTTTY; break;
2971 case 'T': ftst = OP_FTTEXT; break;
2972 case 'B': ftst = OP_FTBINARY; break;
2973 case 'M': case 'A': case 'C':
2974 gv_fetchpv("\024",TRUE, SVt_PV);
2976 case 'M': ftst = OP_FTMTIME; break;
2977 case 'A': ftst = OP_FTATIME; break;
2978 case 'C': ftst = OP_FTCTIME; break;
2986 PL_last_lop_op = (OPCODE)ftst;
2987 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2988 "### Saw file test %c\n", (int)ftst);
2993 /* Assume it was a minus followed by a one-letter named
2994 * subroutine call (or a -bareword), then. */
2995 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2996 "### '-%c' looked like a file test but was not\n",
3005 if (PL_expect == XOPERATOR)
3010 else if (*s == '>') {
3013 if (isIDFIRST_lazy_if(s,UTF)) {
3014 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
3022 if (PL_expect == XOPERATOR)
3025 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
3027 OPERATOR('-'); /* unary minus */
3034 if (PL_expect == XOPERATOR)
3039 if (PL_expect == XOPERATOR)
3042 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
3048 if (PL_expect != XOPERATOR) {
3049 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3050 PL_expect = XOPERATOR;
3051 force_ident(PL_tokenbuf, '*');
3064 if (PL_expect == XOPERATOR) {
3068 PL_tokenbuf[0] = '%';
3069 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
3070 if (!PL_tokenbuf[1]) {
3073 PL_pending_ident = '%';
3092 switch (PL_expect) {
3095 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
3097 PL_bufptr = s; /* update in case we back off */
3103 PL_expect = XTERMBLOCK;
3107 while (isIDFIRST_lazy_if(s,UTF)) {
3108 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3109 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
3110 if (tmp < 0) tmp = -tmp;
3126 d = scan_str(d,TRUE,TRUE);
3128 /* MUST advance bufptr here to avoid bogus
3129 "at end of line" context messages from yyerror().
3131 PL_bufptr = s + len;
3132 yyerror("Unterminated attribute parameter in attribute list");
3135 return REPORT(0); /* EOF indicator */
3139 SV *sv = newSVpvn(s, len);
3140 sv_catsv(sv, PL_lex_stuff);
3141 attrs = append_elem(OP_LIST, attrs,
3142 newSVOP(OP_CONST, 0, sv));
3143 SvREFCNT_dec(PL_lex_stuff);
3144 PL_lex_stuff = Nullsv;
3147 if (len == 6 && strnEQ(s, "unique", len)) {
3148 if (PL_in_my == KEY_our)
3150 GvUNIQUE_on(cGVOPx_gv(yylval.opval));
3152 ; /* skip to avoid loading attributes.pm */
3155 Perl_croak(aTHX_ "The 'unique' attribute may only be applied to 'our' variables");
3158 /* NOTE: any CV attrs applied here need to be part of
3159 the CVf_BUILTIN_ATTRS define in cv.h! */
3160 else if (!PL_in_my && len == 6 && strnEQ(s, "lvalue", len))
3161 CvLVALUE_on(PL_compcv);
3162 else if (!PL_in_my && len == 6 && strnEQ(s, "locked", len))
3163 CvLOCKED_on(PL_compcv);
3164 else if (!PL_in_my && len == 6 && strnEQ(s, "method", len))
3165 CvMETHOD_on(PL_compcv);
3166 else if (!PL_in_my && len == 9 && strnEQ(s, "assertion", len))
3167 CvASSERTION_on(PL_compcv);
3168 /* After we've set the flags, it could be argued that
3169 we don't need to do the attributes.pm-based setting
3170 process, and shouldn't bother appending recognized
3171 flags. To experiment with that, uncomment the
3172 following "else". (Note that's already been
3173 uncommented. That keeps the above-applied built-in
3174 attributes from being intercepted (and possibly
3175 rejected) by a package's attribute routines, but is
3176 justified by the performance win for the common case
3177 of applying only built-in attributes.) */
3179 attrs = append_elem(OP_LIST, attrs,
3180 newSVOP(OP_CONST, 0,
3184 if (*s == ':' && s[1] != ':')
3187 break; /* require real whitespace or :'s */
3189 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
3190 if (*s != ';' && *s != '}' && *s != tmp && (tmp != '=' || *s != ')')) {
3191 char q = ((*s == '\'') ? '"' : '\'');
3192 /* If here for an expression, and parsed no attrs, back off. */
3193 if (tmp == '=' && !attrs) {
3197 /* MUST advance bufptr here to avoid bogus "at end of line"
3198 context messages from yyerror().
3202 yyerror("Unterminated attribute list");
3204 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
3212 PL_nextval[PL_nexttoke].opval = attrs;
3220 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
3221 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
3238 if (PL_lex_brackets <= 0)
3239 yyerror("Unmatched right square bracket");
3242 if (PL_lex_state == LEX_INTERPNORMAL) {
3243 if (PL_lex_brackets == 0) {
3244 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
3245 PL_lex_state = LEX_INTERPEND;
3252 if (PL_lex_brackets > 100) {
3253 Renew(PL_lex_brackstack, PL_lex_brackets + 10, char);
3255 switch (PL_expect) {
3257 if (PL_lex_formbrack) {
3261 if (PL_oldoldbufptr == PL_last_lop)
3262 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3264 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3265 OPERATOR(HASHBRACK);
3267 while (s < PL_bufend && SPACE_OR_TAB(*s))
3270 PL_tokenbuf[0] = '\0';
3271 if (d < PL_bufend && *d == '-') {
3272 PL_tokenbuf[0] = '-';
3274 while (d < PL_bufend && SPACE_OR_TAB(*d))
3277 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3278 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3280 while (d < PL_bufend && SPACE_OR_TAB(*d))
3283 char minus = (PL_tokenbuf[0] == '-');
3284 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3292 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3297 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3302 if (PL_oldoldbufptr == PL_last_lop)
3303 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3305 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3308 if (PL_expect == XREF && PL_lex_state == LEX_INTERPNORMAL) {
3310 /* This hack is to get the ${} in the message. */
3312 yyerror("syntax error");
3315 OPERATOR(HASHBRACK);
3317 /* This hack serves to disambiguate a pair of curlies
3318 * as being a block or an anon hash. Normally, expectation
3319 * determines that, but in cases where we're not in a
3320 * position to expect anything in particular (like inside
3321 * eval"") we have to resolve the ambiguity. This code
3322 * covers the case where the first term in the curlies is a
3323 * quoted string. Most other cases need to be explicitly
3324 * disambiguated by prepending a `+' before the opening
3325 * curly in order to force resolution as an anon hash.
3327 * XXX should probably propagate the outer expectation
3328 * into eval"" to rely less on this hack, but that could
3329 * potentially break current behavior of eval"".
3333 if (*s == '\'' || *s == '"' || *s == '`') {
3334 /* common case: get past first string, handling escapes */
3335 for (t++; t < PL_bufend && *t != *s;)
3336 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3340 else if (*s == 'q') {
3343 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3346 /* skip q//-like construct */
3348 char open, close, term;
3351 while (t < PL_bufend && isSPACE(*t))
3353 /* check for q => */
3354 if (t+1 < PL_bufend && t[0] == '=' && t[1] == '>') {
3355 OPERATOR(HASHBRACK);
3359 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3363 for (t++; t < PL_bufend; t++) {
3364 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3366 else if (*t == open)
3370 for (t++; t < PL_bufend; t++) {
3371 if (*t == '\\' && t+1 < PL_bufend)
3373 else if (*t == close && --brackets <= 0)
3375 else if (*t == open)
3382 /* skip plain q word */
3383 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3386 else if (isALNUM_lazy_if(t,UTF)) {
3388 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3391 while (t < PL_bufend && isSPACE(*t))
3393 /* if comma follows first term, call it an anon hash */
3394 /* XXX it could be a comma expression with loop modifiers */
3395 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3396 || (*t == '=' && t[1] == '>')))
3397 OPERATOR(HASHBRACK);
3398 if (PL_expect == XREF)
3401 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3407 yylval.ival = CopLINE(PL_curcop);
3408 if (isSPACE(*s) || *s == '#')
3409 PL_copline = NOLINE; /* invalidate current command line number */
3414 if (PL_lex_brackets <= 0)
3415 yyerror("Unmatched right curly bracket");
3417 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3418 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3419 PL_lex_formbrack = 0;
3420 if (PL_lex_state == LEX_INTERPNORMAL) {
3421 if (PL_lex_brackets == 0) {
3422 if (PL_expect & XFAKEBRACK) {
3423 PL_expect &= XENUMMASK;
3424 PL_lex_state = LEX_INTERPEND;
3426 return yylex(); /* ignore fake brackets */
3428 if (*s == '-' && s[1] == '>')
3429 PL_lex_state = LEX_INTERPENDMAYBE;
3430 else if (*s != '[' && *s != '{')
3431 PL_lex_state = LEX_INTERPEND;
3434 if (PL_expect & XFAKEBRACK) {
3435 PL_expect &= XENUMMASK;
3437 return yylex(); /* ignore fake brackets */
3447 if (PL_expect == XOPERATOR) {
3448 if (ckWARN(WARN_SEMICOLON)
3449 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3451 CopLINE_dec(PL_curcop);
3452 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
3453 CopLINE_inc(PL_curcop);
3458 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3460 PL_expect = XOPERATOR;
3461 force_ident(PL_tokenbuf, '&');
3465 yylval.ival = (OPpENTERSUB_AMPER<<8);
3484 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3485 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Reversed %c= operator",(int)tmp);
3487 if (PL_expect == XSTATE && isALPHA(tmp) &&
3488 (s == PL_linestart+1 || s[-2] == '\n') )
3490 if (PL_in_eval && !PL_rsfp) {
3495 if (strnEQ(s,"=cut",4)) {
3509 PL_doextract = TRUE;
3512 if (PL_lex_brackets < PL_lex_formbrack) {
3514 #ifdef PERL_STRICT_CR
3515 for (t = s; SPACE_OR_TAB(*t); t++) ;
3517 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3519 if (*t == '\n' || *t == '#') {
3531 /* was this !=~ where !~ was meant?
3532 * warn on m:!=~\s+([/?]|[msy]\W|tr\W): */
3534 if (*s == '~' && ckWARN(WARN_SYNTAX)) {
3537 while (t < PL_bufend && isSPACE(*t))
3540 if (*t == '/' || *t == '?' ||
3541 ((*t == 'm' || *t == 's' || *t == 'y') && !isALNUM(t[1])) ||
3542 (*t == 't' && t[1] == 'r' && !isALNUM(t[2])))
3543 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3544 "!=~ should be !~");
3553 if (PL_expect != XOPERATOR) {
3554 if (s[1] != '<' && !strchr(s,'>'))
3557 s = scan_heredoc(s);
3559 s = scan_inputsymbol(s);
3560 TERM(sublex_start());
3565 SHop(OP_LEFT_SHIFT);
3579 SHop(OP_RIGHT_SHIFT);
3588 if (PL_expect == XOPERATOR) {
3589 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3592 return REPORT(','); /* grandfather non-comma-format format */
3596 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3597 PL_tokenbuf[0] = '@';
3598 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3599 sizeof PL_tokenbuf - 1, FALSE);
3600 if (PL_expect == XOPERATOR)
3601 no_op("Array length", s);
3602 if (!PL_tokenbuf[1])
3604 PL_expect = XOPERATOR;
3605 PL_pending_ident = '#';
3609 PL_tokenbuf[0] = '$';
3610 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3611 sizeof PL_tokenbuf - 1, FALSE);
3612 if (PL_expect == XOPERATOR)
3614 if (!PL_tokenbuf[1]) {
3616 yyerror("Final $ should be \\$ or $name");
3620 /* This kludge not intended to be bulletproof. */
3621 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3622 yylval.opval = newSVOP(OP_CONST, 0,
3623 newSViv(PL_compiling.cop_arybase));
3624 yylval.opval->op_private = OPpCONST_ARYBASE;
3630 if (PL_lex_state == LEX_NORMAL)
3633 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3636 PL_tokenbuf[0] = '@';
3637 if (ckWARN(WARN_SYNTAX)) {
3639 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3642 PL_bufptr = skipspace(PL_bufptr);
3643 while (t < PL_bufend && *t != ']')
3645 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3646 "Multidimensional syntax %.*s not supported",
3647 (t - PL_bufptr) + 1, PL_bufptr);
3651 else if (*s == '{') {
3652 PL_tokenbuf[0] = '%';
3653 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3654 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3656 char tmpbuf[sizeof PL_tokenbuf];
3658 for (t++; isSPACE(*t); t++) ;
3659 if (isIDFIRST_lazy_if(t,UTF)) {
3660 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3661 for (; isSPACE(*t); t++) ;
3662 if (*t == ';' && get_cv(tmpbuf, FALSE))
3663 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3664 "You need to quote \"%s\"", tmpbuf);
3670 PL_expect = XOPERATOR;
3671 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3672 bool islop = (PL_last_lop == PL_oldoldbufptr);
3673 if (!islop || PL_last_lop_op == OP_GREPSTART)
3674 PL_expect = XOPERATOR;
3675 else if (strchr("$@\"'`q", *s))
3676 PL_expect = XTERM; /* e.g. print $fh "foo" */
3677 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3678 PL_expect = XTERM; /* e.g. print $fh &sub */
3679 else if (isIDFIRST_lazy_if(s,UTF)) {
3680 char tmpbuf[sizeof PL_tokenbuf];
3681 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3682 if ((tmp = keyword(tmpbuf, len))) {
3683 /* binary operators exclude handle interpretations */
3695 PL_expect = XTERM; /* e.g. print $fh length() */
3700 PL_expect = XTERM; /* e.g. print $fh subr() */
3703 else if (isDIGIT(*s))
3704 PL_expect = XTERM; /* e.g. print $fh 3 */
3705 else if (*s == '.' && isDIGIT(s[1]))
3706 PL_expect = XTERM; /* e.g. print $fh .3 */
3707 else if (strchr("?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3708 PL_expect = XTERM; /* e.g. print $fh -1 */
3709 else if (*s == '/' && !isSPACE(s[1]) && s[1] != '=' && s[1] != '/')
3710 PL_expect = XTERM; /* e.g. print $fh /.../
3711 XXX except DORDOR operator */
3712 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3713 PL_expect = XTERM; /* print $fh <<"EOF" */
3715 PL_pending_ident = '$';
3719 if (PL_expect == XOPERATOR)
3721 PL_tokenbuf[0] = '@';
3722 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3723 if (!PL_tokenbuf[1]) {
3726 if (PL_lex_state == LEX_NORMAL)
3728 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3730 PL_tokenbuf[0] = '%';
3732 /* Warn about @ where they meant $. */
3733 if (ckWARN(WARN_SYNTAX)) {
3734 if (*s == '[' || *s == '{') {
3736 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3738 if (*t == '}' || *t == ']') {
3740 PL_bufptr = skipspace(PL_bufptr);
3741 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3742 "Scalar value %.*s better written as $%.*s",
3743 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3748 PL_pending_ident = '@';
3751 case '/': /* may be division, defined-or, or pattern */
3752 if (PL_expect == XTERMORDORDOR && s[1] == '/') {
3756 case '?': /* may either be conditional or pattern */
3757 if(PL_expect == XOPERATOR) {
3765 /* A // operator. */
3775 /* Disable warning on "study /blah/" */
3776 if (PL_oldoldbufptr == PL_last_uni
3777 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3778 || memNE(PL_last_uni, "study", 5)
3779 || isALNUM_lazy_if(PL_last_uni+5,UTF)
3782 s = scan_pat(s,OP_MATCH);
3783 TERM(sublex_start());
3787 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3788 #ifdef PERL_STRICT_CR
3791 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3793 && (s == PL_linestart || s[-1] == '\n') )
3795 PL_lex_formbrack = 0;
3799 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3805 yylval.ival = OPf_SPECIAL;
3811 if (PL_expect != XOPERATOR)
3816 case '0': case '1': case '2': case '3': case '4':
3817 case '5': case '6': case '7': case '8': case '9':
3818 s = scan_num(s, &yylval);
3819 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3820 "### Saw number in '%s'\n", s);
3822 if (PL_expect == XOPERATOR)
3827 s = scan_str(s,FALSE,FALSE);
3828 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3829 "### Saw string before '%s'\n", s);
3831 if (PL_expect == XOPERATOR) {
3832 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3835 return REPORT(','); /* grandfather non-comma-format format */
3841 missingterm((char*)0);
3842 yylval.ival = OP_CONST;
3843 TERM(sublex_start());
3846 s = scan_str(s,FALSE,FALSE);
3847 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3848 "### Saw string before '%s'\n", s);
3850 if (PL_expect == XOPERATOR) {
3851 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3854 return REPORT(','); /* grandfather non-comma-format format */
3860 missingterm((char*)0);
3861 yylval.ival = OP_CONST;
3862 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3863 if (*d == '$' || *d == '@' || *d == '\\' || !UTF8_IS_INVARIANT((U8)*d)) {
3864 yylval.ival = OP_STRINGIFY;
3868 TERM(sublex_start());
3871 s = scan_str(s,FALSE,FALSE);
3872 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3873 "### Saw backtick string before '%s'\n", s);
3875 if (PL_expect == XOPERATOR)
3876 no_op("Backticks",s);
3878 missingterm((char*)0);
3879 yylval.ival = OP_BACKTICK;
3881 TERM(sublex_start());
3885 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3886 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),"Can't use \\%c to mean $%c in expression",
3888 if (PL_expect == XOPERATOR)
3889 no_op("Backslash",s);
3893 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3897 while (isDIGIT(*start) || *start == '_')
3899 if (*start == '.' && isDIGIT(start[1])) {
3900 s = scan_num(s, &yylval);
3903 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3904 else if (!isALPHA(*start) && (PL_expect == XTERM
3905 || PL_expect == XREF || PL_expect == XSTATE
3906 || PL_expect == XTERMORDORDOR)) {
3910 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3913 s = scan_num(s, &yylval);
3920 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3960 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3962 /* Some keywords can be followed by any delimiter, including ':' */
3963 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
3964 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3965 (PL_tokenbuf[0] == 'q' &&
3966 strchr("qwxr", PL_tokenbuf[1])))));
3968 /* x::* is just a word, unless x is "CORE" */
3969 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3973 while (d < PL_bufend && isSPACE(*d))
3974 d++; /* no comments skipped here, or s### is misparsed */
3976 /* Is this a label? */
3977 if (!tmp && PL_expect == XSTATE
3978 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3980 yylval.pval = savepv(PL_tokenbuf);
3985 /* Check for keywords */
3986 tmp = keyword(PL_tokenbuf, len);
3988 /* Is this a word before a => operator? */
3989 if (*d == '=' && d[1] == '>') {
3991 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3992 yylval.opval->op_private = OPpCONST_BARE;
3993 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
3994 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
3998 if (tmp < 0) { /* second-class keyword? */
3999 GV *ogv = Nullgv; /* override (winner) */
4000 GV *hgv = Nullgv; /* hidden (loser) */
4001 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
4003 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
4006 if (GvIMPORTED_CV(gv))
4008 else if (! CvMETHOD(cv))
4012 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
4013 (gv = *gvp) != (GV*)&PL_sv_undef &&
4014 GvCVu(gv) && GvIMPORTED_CV(gv))
4021 tmp = 0; /* overridden by import or by GLOBAL */
4024 && -tmp==KEY_lock /* XXX generalizable kludge */
4026 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
4028 tmp = 0; /* any sub overrides "weak" keyword */
4030 else { /* no override */
4032 if (tmp == KEY_dump && ckWARN(WARN_MISC)) {
4033 Perl_warner(aTHX_ packWARN(WARN_MISC),
4034 "dump() better written as CORE::dump()");
4038 if (ckWARN(WARN_AMBIGUOUS) && hgv
4039 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
4040 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4041 "Ambiguous call resolved as CORE::%s(), %s",
4042 GvENAME(hgv), "qualify as such or use &");
4049 default: /* not a keyword */
4053 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
4055 /* Get the rest if it looks like a package qualifier */
4057 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
4059 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
4062 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
4063 *s == '\'' ? "'" : "::");
4068 if (PL_expect == XOPERATOR) {
4069 if (PL_bufptr == PL_linestart) {
4070 CopLINE_dec(PL_curcop);
4071 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
4072 CopLINE_inc(PL_curcop);
4075 no_op("Bareword",s);
4078 /* Look for a subroutine with this name in current package,
4079 unless name is "Foo::", in which case Foo is a bearword
4080 (and a package name). */
4083 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
4085 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
4086 Perl_warner(aTHX_ packWARN(WARN_BAREWORD),
4087 "Bareword \"%s\" refers to nonexistent package",
4090 PL_tokenbuf[len] = '\0';
4097 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
4100 /* if we saw a global override before, get the right name */
4103 sv = newSVpvn("CORE::GLOBAL::",14);
4104 sv_catpv(sv,PL_tokenbuf);
4107 sv = newSVpv(PL_tokenbuf,0);
4109 /* Presume this is going to be a bareword of some sort. */
4112 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
4113 yylval.opval->op_private = OPpCONST_BARE;
4114 /* UTF-8 package name? */
4115 if (UTF && !IN_BYTES &&
4116 is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
4119 /* And if "Foo::", then that's what it certainly is. */
4124 /* See if it's the indirect object for a list operator. */
4126 if (PL_oldoldbufptr &&
4127 PL_oldoldbufptr < PL_bufptr &&
4128 (PL_oldoldbufptr == PL_last_lop
4129 || PL_oldoldbufptr == PL_last_uni) &&
4130 /* NO SKIPSPACE BEFORE HERE! */
4131 (PL_expect == XREF ||
4132 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
4134 bool immediate_paren = *s == '(';
4136 /* (Now we can afford to cross potential line boundary.) */
4139 /* Two barewords in a row may indicate method call. */
4141 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
4144 /* If not a declared subroutine, it's an indirect object. */
4145 /* (But it's an indir obj regardless for sort.) */
4147 if ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
4148 ((!gv || !GvCVu(gv)) &&
4149 (PL_last_lop_op != OP_MAPSTART &&
4150 PL_last_lop_op != OP_GREPSTART))))
4152 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
4157 PL_expect = XOPERATOR;
4160 /* Is this a word before a => operator? */
4161 if (*s == '=' && s[1] == '>' && !pkgname) {
4163 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
4164 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
4165 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
4169 /* If followed by a paren, it's certainly a subroutine. */
4172 if (gv && GvCVu(gv)) {
4173 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
4174 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
4179 PL_nextval[PL_nexttoke].opval = yylval.opval;
4180 PL_expect = XOPERATOR;
4186 /* If followed by var or block, call it a method (unless sub) */
4188 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
4189 PL_last_lop = PL_oldbufptr;
4190 PL_last_lop_op = OP_METHOD;
4194 /* If followed by a bareword, see if it looks like indir obj. */
4197 && (isIDFIRST_lazy_if(s,UTF) || *s == '$')
4198 && (tmp = intuit_method(s,gv)))
4201 /* Not a method, so call it a subroutine (if defined) */
4203 if (gv && GvCVu(gv)) {
4205 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
4206 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4207 "Ambiguous use of -%s resolved as -&%s()",
4208 PL_tokenbuf, PL_tokenbuf);
4209 /* Check for a constant sub */
4211 if ((sv = cv_const_sv(cv))) {
4213 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
4214 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
4215 yylval.opval->op_private = 0;
4219 /* Resolve to GV now. */
4220 op_free(yylval.opval);
4221 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
4222 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
4223 PL_last_lop = PL_oldbufptr;
4224 PL_last_lop_op = OP_ENTERSUB;
4225 /* Is there a prototype? */
4228 char *proto = SvPV((SV*)cv, len);
4231 if (strEQ(proto, "$"))
4233 while (*proto == ';')
4235 if (*proto == '&' && *s == '{') {
4236 sv_setpv(PL_subname, PL_curstash ?
4237 "__ANON__" : "__ANON__::__ANON__");
4241 PL_nextval[PL_nexttoke].opval = yylval.opval;
4247 /* Call it a bare word */
4249 if (PL_hints & HINT_STRICT_SUBS)
4250 yylval.opval->op_private |= OPpCONST_STRICT;
4253 if (ckWARN(WARN_RESERVED)) {
4254 if (lastchar != '-') {
4255 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
4256 if (!*d && !gv_stashpv(PL_tokenbuf,FALSE))
4257 Perl_warner(aTHX_ packWARN(WARN_RESERVED), PL_warn_reserved,
4264 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
4265 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4266 "Operator or semicolon missing before %c%s",
4267 lastchar, PL_tokenbuf);
4268 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4269 "Ambiguous use of %c resolved as operator %c",
4270 lastchar, lastchar);
4276 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4277 newSVpv(CopFILE(PL_curcop),0));
4281 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4282 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
4285 case KEY___PACKAGE__:
4286 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4288 ? newSVsv(PL_curstname)
4297 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
4298 char *pname = "main";
4299 if (PL_tokenbuf[2] == 'D')
4300 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
4301 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
4304 GvIOp(gv) = newIO();
4305 IoIFP(GvIOp(gv)) = PL_rsfp;
4306 #if defined(HAS_FCNTL) && defined(F_SETFD)
4308 int fd = PerlIO_fileno(PL_rsfp);
4309 fcntl(fd,F_SETFD,fd >= 3);
4312 /* Mark this internal pseudo-handle as clean */
4313 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
4315 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
4316 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
4317 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
4319 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
4320 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
4321 /* if the script was opened in binmode, we need to revert
4322 * it to text mode for compatibility; but only iff it has CRs
4323 * XXX this is a questionable hack at best. */
4324 if (PL_bufend-PL_bufptr > 2
4325 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
4328 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
4329 loc = PerlIO_tell(PL_rsfp);
4330 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4333 if (PerlLIO_setmode(PL_rsfp, O_TEXT) != -1) {
4335 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4336 #endif /* NETWARE */
4337 #ifdef PERLIO_IS_STDIO /* really? */
4338 # if defined(__BORLANDC__)
4339 /* XXX see note in do_binmode() */
4340 ((FILE*)PL_rsfp)->flags &= ~_F_BIN;
4344 PerlIO_seek(PL_rsfp, loc, 0);
4348 #ifdef PERLIO_LAYERS
4351 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
4352 else if (PL_encoding) {
4359 XPUSHs(PL_encoding);
4361 call_method("name", G_SCALAR);
4365 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL,
4366 Perl_form(aTHX_ ":encoding(%"SVf")",
4384 if (PL_expect == XSTATE) {
4391 if (*s == ':' && s[1] == ':') {
4394 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4395 if (!(tmp = keyword(PL_tokenbuf, len)))
4396 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4410 LOP(OP_ACCEPT,XTERM);
4416 LOP(OP_ATAN2,XTERM);
4422 LOP(OP_BINMODE,XTERM);
4425 LOP(OP_BLESS,XTERM);
4434 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4451 if (!PL_cryptseen) {
4452 PL_cryptseen = TRUE;
4456 LOP(OP_CRYPT,XTERM);
4459 LOP(OP_CHMOD,XTERM);
4462 LOP(OP_CHOWN,XTERM);
4465 LOP(OP_CONNECT,XTERM);
4481 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4485 PL_hints |= HINT_BLOCK_SCOPE;
4495 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4496 LOP(OP_DBMOPEN,XTERM);
4502 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4509 yylval.ival = CopLINE(PL_curcop);
4523 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4524 UNIBRACK(OP_ENTEREVAL);
4542 case KEY_endhostent:
4548 case KEY_endservent:
4551 case KEY_endprotoent:
4562 yylval.ival = CopLINE(PL_curcop);
4564 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4566 if ((PL_bufend - p) >= 3 &&
4567 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4569 else if ((PL_bufend - p) >= 4 &&
4570 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4573 if (isIDFIRST_lazy_if(p,UTF)) {
4574 p = scan_ident(p, PL_bufend,
4575 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4579 Perl_croak(aTHX_ "Missing $ on loop variable");
4584 LOP(OP_FORMLINE,XTERM);
4590 LOP(OP_FCNTL,XTERM);
4596 LOP(OP_FLOCK,XTERM);
4605 LOP(OP_GREPSTART, XREF);
4608 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4623 case KEY_getpriority:
4624 LOP(OP_GETPRIORITY,XTERM);
4626 case KEY_getprotobyname:
4629 case KEY_getprotobynumber:
4630 LOP(OP_GPBYNUMBER,XTERM);
4632 case KEY_getprotoent:
4644 case KEY_getpeername:
4645 UNI(OP_GETPEERNAME);
4647 case KEY_gethostbyname:
4650 case KEY_gethostbyaddr:
4651 LOP(OP_GHBYADDR,XTERM);
4653 case KEY_gethostent:
4656 case KEY_getnetbyname:
4659 case KEY_getnetbyaddr:
4660 LOP(OP_GNBYADDR,XTERM);
4665 case KEY_getservbyname:
4666 LOP(OP_GSBYNAME,XTERM);
4668 case KEY_getservbyport:
4669 LOP(OP_GSBYPORT,XTERM);
4671 case KEY_getservent:
4674 case KEY_getsockname:
4675 UNI(OP_GETSOCKNAME);
4677 case KEY_getsockopt:
4678 LOP(OP_GSOCKOPT,XTERM);
4700 yylval.ival = CopLINE(PL_curcop);
4704 LOP(OP_INDEX,XTERM);
4710 LOP(OP_IOCTL,XTERM);
4722 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4754 LOP(OP_LISTEN,XTERM);
4763 s = scan_pat(s,OP_MATCH);
4764 TERM(sublex_start());
4767 LOP(OP_MAPSTART, XREF);
4770 LOP(OP_MKDIR,XTERM);
4773 LOP(OP_MSGCTL,XTERM);
4776 LOP(OP_MSGGET,XTERM);
4779 LOP(OP_MSGRCV,XTERM);
4782 LOP(OP_MSGSND,XTERM);
4788 if (isIDFIRST_lazy_if(s,UTF)) {
4789 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4790 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4792 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
4793 if (!PL_in_my_stash) {
4796 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4804 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4811 if (PL_expect != XSTATE)
4812 yyerror("\"no\" not allowed in expression");
4813 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4814 s = force_version(s, FALSE);
4819 if (*s == '(' || (s = skipspace(s), *s == '('))
4826 if (isIDFIRST_lazy_if(s,UTF)) {
4828 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4829 for (t=d; *t && isSPACE(*t); t++) ;
4830 if ( *t && strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE)
4832 && !(t[0] == '=' && t[1] == '>')
4834 Perl_warner(aTHX_ packWARN(WARN_PRECEDENCE),
4835 "Precedence problem: open %.*s should be open(%.*s)",
4836 d - s, s, d - s, s);
4842 yylval.ival = OP_OR;
4852 LOP(OP_OPEN_DIR,XTERM);
4855 checkcomma(s,PL_tokenbuf,"filehandle");
4859 checkcomma(s,PL_tokenbuf,"filehandle");
4878 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4882 LOP(OP_PIPE_OP,XTERM);
4885 s = scan_str(s,FALSE,FALSE);
4887 missingterm((char*)0);
4888 yylval.ival = OP_CONST;
4889 TERM(sublex_start());
4895 s = scan_str(s,FALSE,FALSE);
4897 missingterm((char*)0);
4899 if (SvCUR(PL_lex_stuff)) {
4902 d = SvPV_force(PL_lex_stuff, len);
4905 for (; isSPACE(*d) && len; --len, ++d) ;
4908 if (!warned && ckWARN(WARN_QW)) {
4909 for (; !isSPACE(*d) && len; --len, ++d) {
4911 Perl_warner(aTHX_ packWARN(WARN_QW),
4912 "Possible attempt to separate words with commas");
4915 else if (*d == '#') {
4916 Perl_warner(aTHX_ packWARN(WARN_QW),
4917 "Possible attempt to put comments in qw() list");
4923 for (; !isSPACE(*d) && len; --len, ++d) ;
4925 sv = newSVpvn(b, d-b);
4926 if (DO_UTF8(PL_lex_stuff))
4928 words = append_elem(OP_LIST, words,
4929 newSVOP(OP_CONST, 0, tokeq(sv)));
4933 PL_nextval[PL_nexttoke].opval = words;
4938 SvREFCNT_dec(PL_lex_stuff);
4939 PL_lex_stuff = Nullsv;
4945 s = scan_str(s,FALSE,FALSE);
4947 missingterm((char*)0);
4948 yylval.ival = OP_STRINGIFY;
4949 if (SvIVX(PL_lex_stuff) == '\'')
4950 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4951 TERM(sublex_start());
4954 s = scan_pat(s,OP_QR);
4955 TERM(sublex_start());
4958 s = scan_str(s,FALSE,FALSE);
4960 missingterm((char*)0);
4961 yylval.ival = OP_BACKTICK;
4963 TERM(sublex_start());
4971 s = force_version(s, FALSE);
4973 else if (*s != 'v' || !isDIGIT(s[1])
4974 || (s = force_version(s, TRUE), *s == 'v'))
4976 *PL_tokenbuf = '\0';
4977 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4978 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4979 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4981 yyerror("<> should be quotes");
4989 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4993 LOP(OP_RENAME,XTERM);
5002 LOP(OP_RINDEX,XTERM);
5012 UNIDOR(OP_READLINE);
5025 LOP(OP_REVERSE,XTERM);
5028 UNIDOR(OP_READLINK);
5036 TERM(sublex_start());
5038 TOKEN(1); /* force error */
5047 LOP(OP_SELECT,XTERM);
5053 LOP(OP_SEMCTL,XTERM);
5056 LOP(OP_SEMGET,XTERM);
5059 LOP(OP_SEMOP,XTERM);
5065 LOP(OP_SETPGRP,XTERM);
5067 case KEY_setpriority:
5068 LOP(OP_SETPRIORITY,XTERM);
5070 case KEY_sethostent:
5076 case KEY_setservent:
5079 case KEY_setprotoent:
5089 LOP(OP_SEEKDIR,XTERM);
5091 case KEY_setsockopt:
5092 LOP(OP_SSOCKOPT,XTERM);
5098 LOP(OP_SHMCTL,XTERM);
5101 LOP(OP_SHMGET,XTERM);
5104 LOP(OP_SHMREAD,XTERM);
5107 LOP(OP_SHMWRITE,XTERM);
5110 LOP(OP_SHUTDOWN,XTERM);
5119 LOP(OP_SOCKET,XTERM);
5121 case KEY_socketpair:
5122 LOP(OP_SOCKPAIR,XTERM);
5125 checkcomma(s,PL_tokenbuf,"subroutine name");
5127 if (*s == ';' || *s == ')') /* probably a close */
5128 Perl_croak(aTHX_ "sort is now a reserved word");
5130 s = force_word(s,WORD,TRUE,TRUE,FALSE);
5134 LOP(OP_SPLIT,XTERM);
5137 LOP(OP_SPRINTF,XTERM);
5140 LOP(OP_SPLICE,XTERM);
5155 LOP(OP_SUBSTR,XTERM);
5161 char tmpbuf[sizeof PL_tokenbuf];
5162 SSize_t tboffset = 0;
5163 expectation attrful;
5164 bool have_name, have_proto, bad_proto;
5169 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
5170 (*s == ':' && s[1] == ':'))
5173 attrful = XATTRBLOCK;
5174 /* remember buffer pos'n for later force_word */
5175 tboffset = s - PL_oldbufptr;
5176 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
5177 if (strchr(tmpbuf, ':'))
5178 sv_setpv(PL_subname, tmpbuf);
5180 sv_setsv(PL_subname,PL_curstname);
5181 sv_catpvn(PL_subname,"::",2);
5182 sv_catpvn(PL_subname,tmpbuf,len);
5189 Perl_croak(aTHX_ "Missing name in \"my sub\"");
5190 PL_expect = XTERMBLOCK;
5191 attrful = XATTRTERM;
5192 sv_setpv(PL_subname,"?");
5196 if (key == KEY_format) {
5198 PL_lex_formbrack = PL_lex_brackets + 1;
5200 (void) force_word(PL_oldbufptr + tboffset, WORD,
5205 /* Look for a prototype */
5209 s = scan_str(s,FALSE,FALSE);
5211 Perl_croak(aTHX_ "Prototype not terminated");
5212 /* strip spaces and check for bad characters */
5213 d = SvPVX(PL_lex_stuff);
5216 for (p = d; *p; ++p) {
5219 if (!strchr("$@%*;[]&\\", *p))
5224 if (bad_proto && ckWARN(WARN_SYNTAX))
5225 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5226 "Illegal character in prototype for %"SVf" : %s",
5228 SvCUR(PL_lex_stuff) = tmp;
5236 if (*s == ':' && s[1] != ':')
5237 PL_expect = attrful;
5238 else if (*s != '{' && key == KEY_sub) {
5240 Perl_croak(aTHX_ "Illegal declaration of anonymous subroutine");
5242 Perl_croak(aTHX_ "Illegal declaration of subroutine %"SVf, PL_subname);
5246 PL_nextval[PL_nexttoke].opval =
5247 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
5248 PL_lex_stuff = Nullsv;
5252 sv_setpv(PL_subname,
5253 PL_curstash ? "__ANON__" : "__ANON__::__ANON__");
5256 (void) force_word(PL_oldbufptr + tboffset, WORD,
5265 LOP(OP_SYSTEM,XREF);
5268 LOP(OP_SYMLINK,XTERM);
5271 LOP(OP_SYSCALL,XTERM);
5274 LOP(OP_SYSOPEN,XTERM);
5277 LOP(OP_SYSSEEK,XTERM);
5280 LOP(OP_SYSREAD,XTERM);
5283 LOP(OP_SYSWRITE,XTERM);
5287 TERM(sublex_start());
5308 LOP(OP_TRUNCATE,XTERM);
5320 yylval.ival = CopLINE(PL_curcop);
5324 yylval.ival = CopLINE(PL_curcop);
5328 LOP(OP_UNLINK,XTERM);
5334 LOP(OP_UNPACK,XTERM);
5337 LOP(OP_UTIME,XTERM);
5343 LOP(OP_UNSHIFT,XTERM);
5346 if (PL_expect != XSTATE)
5347 yyerror("\"use\" not allowed in expression");
5349 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
5350 s = force_version(s, TRUE);
5351 if (*s == ';' || (s = skipspace(s), *s == ';')) {
5352 PL_nextval[PL_nexttoke].opval = Nullop;
5355 else if (*s == 'v') {
5356 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5357 s = force_version(s, FALSE);
5361 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5362 s = force_version(s, FALSE);
5374 yylval.ival = CopLINE(PL_curcop);
5378 PL_hints |= HINT_BLOCK_SCOPE;
5385 LOP(OP_WAITPID,XTERM);
5394 ctl_l[0] = toCTRL('L');
5396 gv_fetchpv(ctl_l,TRUE, SVt_PV);
5399 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
5404 if (PL_expect == XOPERATOR)
5410 yylval.ival = OP_XOR;
5415 TERM(sublex_start());
5420 #pragma segment Main
5424 S_pending_ident(pTHX)
5427 register I32 tmp = 0;
5428 /* pit holds the identifier we read and pending_ident is reset */
5429 char pit = PL_pending_ident;
5430 PL_pending_ident = 0;
5432 DEBUG_T({ PerlIO_printf(Perl_debug_log,
5433 "### Tokener saw identifier '%s'\n", PL_tokenbuf); });
5435 /* if we're in a my(), we can't allow dynamics here.
5436 $foo'bar has already been turned into $foo::bar, so
5437 just check for colons.
5439 if it's a legal name, the OP is a PADANY.
5442 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
5443 if (strchr(PL_tokenbuf,':'))
5444 yyerror(Perl_form(aTHX_ "No package name allowed for "
5445 "variable %s in \"our\"",
5447 tmp = allocmy(PL_tokenbuf);
5450 if (strchr(PL_tokenbuf,':'))
5451 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
5453 yylval.opval = newOP(OP_PADANY, 0);
5454 yylval.opval->op_targ = allocmy(PL_tokenbuf);
5460 build the ops for accesses to a my() variable.
5462 Deny my($a) or my($b) in a sort block, *if* $a or $b is
5463 then used in a comparison. This catches most, but not
5464 all cases. For instance, it catches
5465 sort { my($a); $a <=> $b }
5467 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
5468 (although why you'd do that is anyone's guess).
5471 if (!strchr(PL_tokenbuf,':')) {
5473 tmp = pad_findmy(PL_tokenbuf);
5474 if (tmp != NOT_IN_PAD) {
5475 /* might be an "our" variable" */
5476 if (PAD_COMPNAME_FLAGS(tmp) & SVpad_OUR) {
5477 /* build ops for a bareword */
5478 SV *sym = newSVpv(HvNAME(PAD_COMPNAME_OURSTASH(tmp)), 0);
5479 sv_catpvn(sym, "::", 2);
5480 sv_catpv(sym, PL_tokenbuf+1);
5481 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
5482 yylval.opval->op_private = OPpCONST_ENTERED;
5483 gv_fetchpv(SvPVX(sym),
5485 ? (GV_ADDMULTI | GV_ADDINEVAL)
5488 ((PL_tokenbuf[0] == '$') ? SVt_PV
5489 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5494 /* if it's a sort block and they're naming $a or $b */
5495 if (PL_last_lop_op == OP_SORT &&
5496 PL_tokenbuf[0] == '$' &&
5497 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
5500 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
5501 d < PL_bufend && *d != '\n';
5504 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
5505 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
5511 yylval.opval = newOP(OP_PADANY, 0);
5512 yylval.opval->op_targ = tmp;
5518 Whine if they've said @foo in a doublequoted string,
5519 and @foo isn't a variable we can find in the symbol
5522 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
5523 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
5524 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
5525 && ckWARN(WARN_AMBIGUOUS))
5527 /* Downgraded from fatal to warning 20000522 mjd */
5528 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5529 "Possible unintended interpolation of %s in string",
5534 /* build ops for a bareword */
5535 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
5536 yylval.opval->op_private = OPpCONST_ENTERED;
5537 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
5538 ((PL_tokenbuf[0] == '$') ? SVt_PV
5539 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5545 Perl_keyword(pTHX_ register char *d, I32 len)
5550 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
5551 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
5552 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
5553 if (strEQ(d,"__DATA__")) return KEY___DATA__;
5554 if (strEQ(d,"__END__")) return KEY___END__;
5558 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
5563 if (strEQ(d,"and")) return -KEY_and;
5564 if (strEQ(d,"abs")) return -KEY_abs;
5567 if (strEQ(d,"alarm")) return -KEY_alarm;
5568 if (strEQ(d,"atan2")) return -KEY_atan2;
5571 if (strEQ(d,"accept")) return -KEY_accept;
5576 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5579 if (strEQ(d,"bless")) return -KEY_bless;
5580 if (strEQ(d,"bind")) return -KEY_bind;
5581 if (strEQ(d,"binmode")) return -KEY_binmode;
5584 if (strEQ(d,"CORE")) return -KEY_CORE;
5585 if (strEQ(d,"CHECK")) return KEY_CHECK;
5590 if (strEQ(d,"cmp")) return -KEY_cmp;
5591 if (strEQ(d,"chr")) return -KEY_chr;
5592 if (strEQ(d,"cos")) return -KEY_cos;
5595 if (strEQ(d,"chop")) return -KEY_chop;
5598 if (strEQ(d,"close")) return -KEY_close;
5599 if (strEQ(d,"chdir")) return -KEY_chdir;
5600 if (strEQ(d,"chomp")) return -KEY_chomp;
5601 if (strEQ(d,"chmod")) return -KEY_chmod;
5602 if (strEQ(d,"chown")) return -KEY_chown;
5603 if (strEQ(d,"crypt")) return -KEY_crypt;
5606 if (strEQ(d,"chroot")) return -KEY_chroot;
5607 if (strEQ(d,"caller")) return -KEY_caller;
5610 if (strEQ(d,"connect")) return -KEY_connect;
5613 if (strEQ(d,"closedir")) return -KEY_closedir;
5614 if (strEQ(d,"continue")) return -KEY_continue;
5619 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5624 if (strEQ(d,"do")) return KEY_do;
5627 if (strEQ(d,"die")) return -KEY_die;
5630 if (strEQ(d,"dump")) return -KEY_dump;
5633 if (strEQ(d,"delete")) return KEY_delete;
5636 if (strEQ(d,"defined")) return KEY_defined;
5637 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5640 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5645 if (strEQ(d,"END")) return KEY_END;
5650 if (strEQ(d,"eq")) return -KEY_eq;
5653 if (strEQ(d,"eof")) return -KEY_eof;
5654 if (strEQ(d,"err")) return -KEY_err;
5655 if (strEQ(d,"exp")) return -KEY_exp;
5658 if (strEQ(d,"else")) return KEY_else;
5659 if (strEQ(d,"exit")) return -KEY_exit;
5660 if (strEQ(d,"eval")) return KEY_eval;
5661 if (strEQ(d,"exec")) return -KEY_exec;
5662 if (strEQ(d,"each")) return -KEY_each;
5665 if (strEQ(d,"elsif")) return KEY_elsif;
5668 if (strEQ(d,"exists")) return KEY_exists;
5669 if (strEQ(d,"elseif") && ckWARN_d(WARN_SYNTAX))
5670 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5671 "elseif should be elsif");
5674 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5675 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5678 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5681 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5682 if (strEQ(d,"endservent")) return -KEY_endservent;
5685 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5692 if (strEQ(d,"for")) return KEY_for;
5695 if (strEQ(d,"fork")) return -KEY_fork;
5698 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5699 if (strEQ(d,"flock")) return -KEY_flock;
5702 if (strEQ(d,"format")) return KEY_format;
5703 if (strEQ(d,"fileno")) return -KEY_fileno;
5706 if (strEQ(d,"foreach")) return KEY_foreach;
5709 if (strEQ(d,"formline")) return -KEY_formline;
5714 if (strnEQ(d,"get",3)) {
5719 if (strEQ(d,"ppid")) return -KEY_getppid;
5720 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5723 if (strEQ(d,"pwent")) return -KEY_getpwent;
5724 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5725 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5728 if (strEQ(d,"peername")) return -KEY_getpeername;
5729 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5730 if (strEQ(d,"priority")) return -KEY_getpriority;
5733 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5736 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5740 else if (*d == 'h') {
5741 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5742 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5743 if (strEQ(d,"hostent")) return -KEY_gethostent;
5745 else if (*d == 'n') {
5746 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5747 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5748 if (strEQ(d,"netent")) return -KEY_getnetent;
5750 else if (*d == 's') {
5751 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5752 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5753 if (strEQ(d,"servent")) return -KEY_getservent;
5754 if (strEQ(d,"sockname")) return -KEY_getsockname;
5755 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5757 else if (*d == 'g') {
5758 if (strEQ(d,"grent")) return -KEY_getgrent;
5759 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5760 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5762 else if (*d == 'l') {
5763 if (strEQ(d,"login")) return -KEY_getlogin;
5765 else if (strEQ(d,"c")) return -KEY_getc;
5770 if (strEQ(d,"gt")) return -KEY_gt;
5771 if (strEQ(d,"ge")) return -KEY_ge;
5774 if (strEQ(d,"grep")) return KEY_grep;
5775 if (strEQ(d,"goto")) return KEY_goto;
5776 if (strEQ(d,"glob")) return KEY_glob;
5779 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5784 if (strEQ(d,"hex")) return -KEY_hex;
5787 if (strEQ(d,"INIT")) return KEY_INIT;
5792 if (strEQ(d,"if")) return KEY_if;
5795 if (strEQ(d,"int")) return -KEY_int;
5798 if (strEQ(d,"index")) return -KEY_index;
5799 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5804 if (strEQ(d,"join")) return -KEY_join;
5808 if (strEQ(d,"keys")) return -KEY_keys;
5809 if (strEQ(d,"kill")) return -KEY_kill;
5815 if (strEQ(d,"lt")) return -KEY_lt;
5816 if (strEQ(d,"le")) return -KEY_le;
5817 if (strEQ(d,"lc")) return -KEY_lc;
5820 if (strEQ(d,"log")) return -KEY_log;
5823 if (strEQ(d,"last")) return KEY_last;
5824 if (strEQ(d,"link")) return -KEY_link;
5825 if (strEQ(d,"lock")) return -KEY_lock;
5828 if (strEQ(d,"local")) return KEY_local;
5829 if (strEQ(d,"lstat")) return -KEY_lstat;
5832 if (strEQ(d,"length")) return -KEY_length;
5833 if (strEQ(d,"listen")) return -KEY_listen;
5836 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5839 if (strEQ(d,"localtime")) return -KEY_localtime;
5845 case 1: return KEY_m;
5847 if (strEQ(d,"my")) return KEY_my;
5850 if (strEQ(d,"map")) return KEY_map;
5853 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5856 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5857 if (strEQ(d,"msgget")) return -KEY_msgget;
5858 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5859 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5864 if (strEQ(d,"next")) return KEY_next;
5865 if (strEQ(d,"ne")) return -KEY_ne;
5866 if (strEQ(d,"not")) return -KEY_not;
5867 if (strEQ(d,"no")) return KEY_no;
5872 if (strEQ(d,"or")) return -KEY_or;
5875 if (strEQ(d,"ord")) return -KEY_ord;
5876 if (strEQ(d,"oct")) return -KEY_oct;
5877 if (strEQ(d,"our")) return KEY_our;
5880 if (strEQ(d,"open")) return -KEY_open;
5883 if (strEQ(d,"opendir")) return -KEY_opendir;
5890 if (strEQ(d,"pop")) return -KEY_pop;
5891 if (strEQ(d,"pos")) return KEY_pos;
5894 if (strEQ(d,"push")) return -KEY_push;
5895 if (strEQ(d,"pack")) return -KEY_pack;
5896 if (strEQ(d,"pipe")) return -KEY_pipe;
5899 if (strEQ(d,"print")) return KEY_print;
5902 if (strEQ(d,"printf")) return KEY_printf;
5905 if (strEQ(d,"package")) return KEY_package;
5908 if (strEQ(d,"prototype")) return KEY_prototype;
5913 if (strEQ(d,"q")) return KEY_q;
5914 if (strEQ(d,"qr")) return KEY_qr;
5915 if (strEQ(d,"qq")) return KEY_qq;
5916 if (strEQ(d,"qw")) return KEY_qw;
5917 if (strEQ(d,"qx")) return KEY_qx;
5919 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5924 if (strEQ(d,"ref")) return -KEY_ref;
5927 if (strEQ(d,"read")) return -KEY_read;
5928 if (strEQ(d,"rand")) return -KEY_rand;
5929 if (strEQ(d,"recv")) return -KEY_recv;
5930 if (strEQ(d,"redo")) return KEY_redo;
5933 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5934 if (strEQ(d,"reset")) return -KEY_reset;
5937 if (strEQ(d,"return")) return KEY_return;
5938 if (strEQ(d,"rename")) return -KEY_rename;
5939 if (strEQ(d,"rindex")) return -KEY_rindex;
5942 if (strEQ(d,"require")) return KEY_require;
5943 if (strEQ(d,"reverse")) return -KEY_reverse;
5944 if (strEQ(d,"readdir")) return -KEY_readdir;
5947 if (strEQ(d,"readlink")) return -KEY_readlink;
5948 if (strEQ(d,"readline")) return -KEY_readline;
5949 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5952 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5958 case 0: return KEY_s;
5960 if (strEQ(d,"scalar")) return KEY_scalar;
5965 if (strEQ(d,"seek")) return -KEY_seek;
5966 if (strEQ(d,"send")) return -KEY_send;
5969 if (strEQ(d,"semop")) return -KEY_semop;
5972 if (strEQ(d,"select")) return -KEY_select;
5973 if (strEQ(d,"semctl")) return -KEY_semctl;
5974 if (strEQ(d,"semget")) return -KEY_semget;
5977 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5978 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5981 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5982 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5985 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5988 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5989 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5990 if (strEQ(d,"setservent")) return -KEY_setservent;
5993 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5994 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
6001 if (strEQ(d,"shift")) return -KEY_shift;
6004 if (strEQ(d,"shmctl")) return -KEY_shmctl;
6005 if (strEQ(d,"shmget")) return -KEY_shmget;
6008 if (strEQ(d,"shmread")) return -KEY_shmread;
6011 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
6012 if (strEQ(d,"shutdown")) return -KEY_shutdown;
6017 if (strEQ(d,"sin")) return -KEY_sin;
6020 if (strEQ(d,"sleep")) return -KEY_sleep;
6023 if (strEQ(d,"sort")) return KEY_sort;
6024 if (strEQ(d,"socket")) return -KEY_socket;
6025 if (strEQ(d,"socketpair")) return -KEY_socketpair;
6028 if (strEQ(d,"split")) return KEY_split;
6029 if (strEQ(d,"sprintf")) return -KEY_sprintf;
6030 if (strEQ(d,"splice")) return -KEY_splice;
6033 if (strEQ(d,"sqrt")) return -KEY_sqrt;
6036 if (strEQ(d,"srand")) return -KEY_srand;
6039 if (strEQ(d,"stat")) return -KEY_stat;
6040 if (strEQ(d,"study")) return KEY_study;
6043 if (strEQ(d,"substr")) return -KEY_substr;
6044 if (strEQ(d,"sub")) return KEY_sub;
6049 if (strEQ(d,"system")) return -KEY_system;
6052 if (strEQ(d,"symlink")) return -KEY_symlink;
6053 if (strEQ(d,"syscall")) return -KEY_syscall;
6054 if (strEQ(d,"sysopen")) return -KEY_sysopen;
6055 if (strEQ(d,"sysread")) return -KEY_sysread;
6056 if (strEQ(d,"sysseek")) return -KEY_sysseek;
6059 if (strEQ(d,"syswrite")) return -KEY_syswrite;
6068 if (strEQ(d,"tr")) return KEY_tr;
6071 if (strEQ(d,"tie")) return KEY_tie;
6074 if (strEQ(d,"tell")) return -KEY_tell;
6075 if (strEQ(d,"tied")) return KEY_tied;
6076 if (strEQ(d,"time")) return -KEY_time;
6079 if (strEQ(d,"times")) return -KEY_times;
6082 if (strEQ(d,"telldir")) return -KEY_telldir;
6085 if (strEQ(d,"truncate")) return -KEY_truncate;
6092 if (strEQ(d,"uc")) return -KEY_uc;
6095 if (strEQ(d,"use")) return KEY_use;
6098 if (strEQ(d,"undef")) return KEY_undef;
6099 if (strEQ(d,"until")) return KEY_until;
6100 if (strEQ(d,"untie")) return KEY_untie;
6101 if (strEQ(d,"utime")) return -KEY_utime;
6102 if (strEQ(d,"umask")) return -KEY_umask;
6105 if (strEQ(d,"unless")) return KEY_unless;
6106 if (strEQ(d,"unpack")) return -KEY_unpack;
6107 if (strEQ(d,"unlink")) return -KEY_unlink;
6110 if (strEQ(d,"unshift")) return -KEY_unshift;
6111 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
6116 if (strEQ(d,"values")) return -KEY_values;
6117 if (strEQ(d,"vec")) return -KEY_vec;
6122 if (strEQ(d,"warn")) return -KEY_warn;
6123 if (strEQ(d,"wait")) return -KEY_wait;
6126 if (strEQ(d,"while")) return KEY_while;
6127 if (strEQ(d,"write")) return -KEY_write;
6130 if (strEQ(d,"waitpid")) return -KEY_waitpid;
6133 if (strEQ(d,"wantarray")) return -KEY_wantarray;
6138 if (len == 1) return -KEY_x;
6139 if (strEQ(d,"xor")) return -KEY_xor;
6142 if (len == 1) return KEY_y;
6151 S_checkcomma(pTHX_ register char *s, char *name, char *what)
6155 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
6156 if (ckWARN(WARN_SYNTAX)) {
6158 for (w = s+2; *w && level; w++) {
6165 for (; *w && isSPACE(*w); w++) ;
6166 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
6167 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
6168 "%s (...) interpreted as function",name);
6171 while (s < PL_bufend && isSPACE(*s))
6175 while (s < PL_bufend && isSPACE(*s))
6177 if (isIDFIRST_lazy_if(s,UTF)) {
6179 while (isALNUM_lazy_if(s,UTF))
6181 while (s < PL_bufend && isSPACE(*s))
6186 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
6190 Perl_croak(aTHX_ "No comma allowed after %s", what);
6195 /* Either returns sv, or mortalizes sv and returns a new SV*.
6196 Best used as sv=new_constant(..., sv, ...).
6197 If s, pv are NULL, calls subroutine with one argument,
6198 and type is used with error messages only. */
6201 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
6205 HV *table = GvHV(PL_hintgv); /* ^H */
6209 const char *why1, *why2, *why3;
6211 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
6214 why2 = strEQ(key,"charnames")
6215 ? "(possibly a missing \"use charnames ...\")"
6217 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
6218 (type ? type: "undef"), why2);
6220 /* This is convoluted and evil ("goto considered harmful")
6221 * but I do not understand the intricacies of all the different
6222 * failure modes of %^H in here. The goal here is to make
6223 * the most probable error message user-friendly. --jhi */
6228 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
6229 (type ? type: "undef"), why1, why2, why3);
6231 yyerror(SvPVX(msg));
6235 cvp = hv_fetch(table, key, strlen(key), FALSE);
6236 if (!cvp || !SvOK(*cvp)) {
6239 why3 = "} is not defined";
6242 sv_2mortal(sv); /* Parent created it permanently */
6245 pv = sv_2mortal(newSVpvn(s, len));
6247 typesv = sv_2mortal(newSVpv(type, 0));
6249 typesv = &PL_sv_undef;
6251 PUSHSTACKi(PERLSI_OVERLOAD);
6263 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
6267 /* Check the eval first */
6268 if (!PL_in_eval && SvTRUE(ERRSV)) {
6270 sv_catpv(ERRSV, "Propagated");
6271 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
6273 res = SvREFCNT_inc(sv);
6277 (void)SvREFCNT_inc(res);
6286 why1 = "Call to &{$^H{";
6288 why3 = "}} did not return a defined value";
6297 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
6299 register char *d = dest;
6300 register char *e = d + destlen - 3; /* two-character token, ending NUL */
6303 Perl_croak(aTHX_ ident_too_long);
6304 if (isALNUM(*s)) /* UTF handled below */
6306 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
6311 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
6315 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
6316 char *t = s + UTF8SKIP(s);
6317 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
6319 if (d + (t - s) > e)
6320 Perl_croak(aTHX_ ident_too_long);
6321 Copy(s, d, t - s, char);
6334 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
6344 e = d + destlen - 3; /* two-character token, ending NUL */
6346 while (isDIGIT(*s)) {
6348 Perl_croak(aTHX_ ident_too_long);
6355 Perl_croak(aTHX_ ident_too_long);
6356 if (isALNUM(*s)) /* UTF handled below */
6358 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
6363 else if (*s == ':' && s[1] == ':') {
6367 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
6368 char *t = s + UTF8SKIP(s);
6369 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
6371 if (d + (t - s) > e)
6372 Perl_croak(aTHX_ ident_too_long);
6373 Copy(s, d, t - s, char);
6384 if (PL_lex_state != LEX_NORMAL)
6385 PL_lex_state = LEX_INTERPENDMAYBE;
6388 if (*s == '$' && s[1] &&
6389 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
6402 if (*d == '^' && *s && isCONTROLVAR(*s)) {
6407 if (isSPACE(s[-1])) {
6410 if (!SPACE_OR_TAB(ch)) {
6416 if (isIDFIRST_lazy_if(d,UTF)) {
6420 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
6422 while (e < send && UTF8_IS_CONTINUED(*e) && is_utf8_mark((U8*)e))
6425 Copy(s, d, e - s, char);
6430 while ((isALNUM(*s) || *s == ':') && d < e)
6433 Perl_croak(aTHX_ ident_too_long);
6436 while (s < send && SPACE_OR_TAB(*s)) s++;
6437 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
6438 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
6439 const char *brack = *s == '[' ? "[...]" : "{...}";
6440 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
6441 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
6442 funny, dest, brack, funny, dest, brack);
6445 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
6449 /* Handle extended ${^Foo} variables
6450 * 1999-02-27 mjd-perl-patch@plover.com */
6451 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
6455 while (isALNUM(*s) && d < e) {
6459 Perl_croak(aTHX_ ident_too_long);
6464 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets) {
6465 PL_lex_state = LEX_INTERPEND;
6470 if (PL_lex_state == LEX_NORMAL) {
6471 if (ckWARN(WARN_AMBIGUOUS) &&
6472 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
6474 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
6475 "Ambiguous use of %c{%s} resolved to %c%s",
6476 funny, dest, funny, dest);
6481 s = bracket; /* let the parser handle it */
6485 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
6486 PL_lex_state = LEX_INTERPEND;
6491 Perl_pmflag(pTHX_ U32* pmfl, int ch)
6496 *pmfl |= PMf_GLOBAL;
6498 *pmfl |= PMf_CONTINUE;
6502 *pmfl |= PMf_MULTILINE;
6504 *pmfl |= PMf_SINGLELINE;
6506 *pmfl |= PMf_EXTENDED;
6510 S_scan_pat(pTHX_ char *start, I32 type)
6515 s = scan_str(start,FALSE,FALSE);
6517 Perl_croak(aTHX_ "Search pattern not terminated");
6519 pm = (PMOP*)newPMOP(type, 0);
6520 if (PL_multi_open == '?')
6521 pm->op_pmflags |= PMf_ONCE;
6523 while (*s && strchr("iomsx", *s))
6524 pmflag(&pm->op_pmflags,*s++);
6527 while (*s && strchr("iogcmsx", *s))
6528 pmflag(&pm->op_pmflags,*s++);
6530 /* issue a warning if /c is specified,but /g is not */
6531 if (ckWARN(WARN_REGEXP) &&
6532 (pm->op_pmflags & PMf_CONTINUE) && !(pm->op_pmflags & PMf_GLOBAL))
6534 Perl_warner(aTHX_ packWARN(WARN_REGEXP), c_without_g);
6537 pm->op_pmpermflags = pm->op_pmflags;
6539 PL_lex_op = (OP*)pm;
6540 yylval.ival = OP_MATCH;
6545 S_scan_subst(pTHX_ char *start)
6552 yylval.ival = OP_NULL;
6554 s = scan_str(start,FALSE,FALSE);
6557 Perl_croak(aTHX_ "Substitution pattern not terminated");
6559 if (s[-1] == PL_multi_open)
6562 first_start = PL_multi_start;
6563 s = scan_str(s,FALSE,FALSE);
6566 SvREFCNT_dec(PL_lex_stuff);
6567 PL_lex_stuff = Nullsv;
6569 Perl_croak(aTHX_ "Substitution replacement not terminated");
6571 PL_multi_start = first_start; /* so whole substitution is taken together */
6573 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6579 else if (strchr("iogcmsx", *s))
6580 pmflag(&pm->op_pmflags,*s++);
6585 /* /c is not meaningful with s/// */
6586 if (ckWARN(WARN_REGEXP) && (pm->op_pmflags & PMf_CONTINUE))
6588 Perl_warner(aTHX_ packWARN(WARN_REGEXP), c_in_subst);
6593 PL_sublex_info.super_bufptr = s;
6594 PL_sublex_info.super_bufend = PL_bufend;
6596 pm->op_pmflags |= PMf_EVAL;
6597 repl = newSVpvn("",0);
6599 sv_catpv(repl, es ? "eval " : "do ");
6600 sv_catpvn(repl, "{ ", 2);
6601 sv_catsv(repl, PL_lex_repl);
6602 sv_catpvn(repl, " };", 2);
6604 SvREFCNT_dec(PL_lex_repl);
6608 pm->op_pmpermflags = pm->op_pmflags;
6609 PL_lex_op = (OP*)pm;
6610 yylval.ival = OP_SUBST;
6615 S_scan_trans(pTHX_ char *start)
6624 yylval.ival = OP_NULL;
6626 s = scan_str(start,FALSE,FALSE);
6628 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6629 if (s[-1] == PL_multi_open)
6632 s = scan_str(s,FALSE,FALSE);
6635 SvREFCNT_dec(PL_lex_stuff);
6636 PL_lex_stuff = Nullsv;
6638 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6641 complement = del = squash = 0;
6642 while (strchr("cds", *s)) {
6644 complement = OPpTRANS_COMPLEMENT;
6646 del = OPpTRANS_DELETE;
6648 squash = OPpTRANS_SQUASH;
6652 New(803, tbl, complement&&!del?258:256, short);
6653 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6654 o->op_private &= ~OPpTRANS_ALL;
6655 o->op_private |= del|squash|complement|
6656 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
6657 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
6660 yylval.ival = OP_TRANS;
6665 S_scan_heredoc(pTHX_ register char *s)
6668 I32 op_type = OP_SCALAR;
6675 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6679 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6682 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
6683 if (*peek && strchr("`'\"",*peek)) {
6686 s = delimcpy(d, e, s, PL_bufend, term, &len);
6696 if (!isALNUM_lazy_if(s,UTF))
6697 deprecate_old("bare << to mean <<\"\"");
6698 for (; isALNUM_lazy_if(s,UTF); s++) {
6703 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6704 Perl_croak(aTHX_ "Delimiter for here document is too long");
6707 len = d - PL_tokenbuf;
6708 #ifndef PERL_STRICT_CR
6709 d = strchr(s, '\r');
6713 while (s < PL_bufend) {
6719 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6728 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6733 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6734 herewas = newSVpvn(s,PL_bufend-s);
6736 s--, herewas = newSVpvn(s,d-s);
6737 s += SvCUR(herewas);
6739 tmpstr = NEWSV(87,79);
6740 sv_upgrade(tmpstr, SVt_PVIV);
6745 else if (term == '`') {
6746 op_type = OP_BACKTICK;
6747 SvIVX(tmpstr) = '\\';
6751 PL_multi_start = CopLINE(PL_curcop);
6752 PL_multi_open = PL_multi_close = '<';
6753 term = *PL_tokenbuf;
6754 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6755 char *bufptr = PL_sublex_info.super_bufptr;
6756 char *bufend = PL_sublex_info.super_bufend;
6757 char *olds = s - SvCUR(herewas);
6758 s = strchr(bufptr, '\n');
6762 while (s < bufend &&
6763 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6765 CopLINE_inc(PL_curcop);
6768 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
6769 missingterm(PL_tokenbuf);
6771 sv_setpvn(herewas,bufptr,d-bufptr+1);
6772 sv_setpvn(tmpstr,d+1,s-d);
6774 sv_catpvn(herewas,s,bufend-s);
6775 Copy(SvPVX(herewas),bufptr,SvCUR(herewas) + 1,char);
6782 while (s < PL_bufend &&
6783 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6785 CopLINE_inc(PL_curcop);
6787 if (s >= PL_bufend) {
6788 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
6789 missingterm(PL_tokenbuf);
6791 sv_setpvn(tmpstr,d+1,s-d);
6793 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6795 sv_catpvn(herewas,s,PL_bufend-s);
6796 sv_setsv(PL_linestr,herewas);
6797 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6798 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6799 PL_last_lop = PL_last_uni = Nullch;
6802 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6803 while (s >= PL_bufend) { /* multiple line string? */
6805 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6806 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
6807 missingterm(PL_tokenbuf);
6809 CopLINE_inc(PL_curcop);
6810 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6811 PL_last_lop = PL_last_uni = Nullch;
6812 #ifndef PERL_STRICT_CR
6813 if (PL_bufend - PL_linestart >= 2) {
6814 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6815 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6817 PL_bufend[-2] = '\n';
6819 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6821 else if (PL_bufend[-1] == '\r')
6822 PL_bufend[-1] = '\n';
6824 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6825 PL_bufend[-1] = '\n';
6827 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6828 SV *sv = NEWSV(88,0);
6830 sv_upgrade(sv, SVt_PVMG);
6831 sv_setsv(sv,PL_linestr);
6834 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6836 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6839 sv_catsv(PL_linestr,herewas);
6840 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6844 sv_catsv(tmpstr,PL_linestr);
6849 PL_multi_end = CopLINE(PL_curcop);
6850 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6851 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6852 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6854 SvREFCNT_dec(herewas);
6856 if (UTF && is_utf8_string((U8*)SvPVX(tmpstr), SvCUR(tmpstr)))
6858 else if (PL_encoding)
6859 sv_recode_to_utf8(tmpstr, PL_encoding);
6861 PL_lex_stuff = tmpstr;
6862 yylval.ival = op_type;
6867 takes: current position in input buffer
6868 returns: new position in input buffer
6869 side-effects: yylval and lex_op are set.
6874 <FH> read from filehandle
6875 <pkg::FH> read from package qualified filehandle
6876 <pkg'FH> read from package qualified filehandle
6877 <$fh> read from filehandle in $fh
6883 S_scan_inputsymbol(pTHX_ char *start)
6885 register char *s = start; /* current position in buffer */
6891 d = PL_tokenbuf; /* start of temp holding space */
6892 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6893 end = strchr(s, '\n');
6896 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6898 /* die if we didn't have space for the contents of the <>,
6899 or if it didn't end, or if we see a newline
6902 if (len >= sizeof PL_tokenbuf)
6903 Perl_croak(aTHX_ "Excessively long <> operator");
6905 Perl_croak(aTHX_ "Unterminated <> operator");
6910 Remember, only scalar variables are interpreted as filehandles by
6911 this code. Anything more complex (e.g., <$fh{$num}>) will be
6912 treated as a glob() call.
6913 This code makes use of the fact that except for the $ at the front,
6914 a scalar variable and a filehandle look the same.
6916 if (*d == '$' && d[1]) d++;
6918 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6919 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6922 /* If we've tried to read what we allow filehandles to look like, and
6923 there's still text left, then it must be a glob() and not a getline.
6924 Use scan_str to pull out the stuff between the <> and treat it
6925 as nothing more than a string.
6928 if (d - PL_tokenbuf != len) {
6929 yylval.ival = OP_GLOB;
6931 s = scan_str(start,FALSE,FALSE);
6933 Perl_croak(aTHX_ "Glob not terminated");
6937 bool readline_overriden = FALSE;
6938 GV *gv_readline = Nullgv;
6940 /* we're in a filehandle read situation */
6943 /* turn <> into <ARGV> */
6945 Copy("ARGV",d,5,char);
6947 /* Check whether readline() is overriden */
6948 if (((gv_readline = gv_fetchpv("readline", FALSE, SVt_PVCV))
6949 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline))
6951 ((gvp = (GV**)hv_fetch(PL_globalstash, "readline", 8, FALSE))
6952 && (gv_readline = *gvp) != (GV*)&PL_sv_undef
6953 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline)))
6954 readline_overriden = TRUE;
6956 /* if <$fh>, create the ops to turn the variable into a
6962 /* try to find it in the pad for this block, otherwise find
6963 add symbol table ops
6965 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6966 if (PAD_COMPNAME_FLAGS(tmp) & SVpad_OUR) {
6967 SV *sym = sv_2mortal(
6968 newSVpv(HvNAME(PAD_COMPNAME_OURSTASH(tmp)),0));
6969 sv_catpvn(sym, "::", 2);
6975 OP *o = newOP(OP_PADSV, 0);
6977 PL_lex_op = readline_overriden
6978 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
6979 append_elem(OP_LIST, o,
6980 newCVREF(0, newGVOP(OP_GV,0,gv_readline))))
6981 : (OP*)newUNOP(OP_READLINE, 0, o);
6990 ? (GV_ADDMULTI | GV_ADDINEVAL)
6993 PL_lex_op = readline_overriden
6994 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
6995 append_elem(OP_LIST,
6996 newUNOP(OP_RV2SV, 0, newGVOP(OP_GV, 0, gv)),
6997 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
6998 : (OP*)newUNOP(OP_READLINE, 0,
6999 newUNOP(OP_RV2SV, 0,
7000 newGVOP(OP_GV, 0, gv)));
7002 if (!readline_overriden)
7003 PL_lex_op->op_flags |= OPf_SPECIAL;
7004 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
7005 yylval.ival = OP_NULL;
7008 /* If it's none of the above, it must be a literal filehandle
7009 (<Foo::BAR> or <FOO>) so build a simple readline OP */
7011 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
7012 PL_lex_op = readline_overriden
7013 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
7014 append_elem(OP_LIST,
7015 newGVOP(OP_GV, 0, gv),
7016 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
7017 : (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
7018 yylval.ival = OP_NULL;
7027 takes: start position in buffer
7028 keep_quoted preserve \ on the embedded delimiter(s)
7029 keep_delims preserve the delimiters around the string
7030 returns: position to continue reading from buffer
7031 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
7032 updates the read buffer.
7034 This subroutine pulls a string out of the input. It is called for:
7035 q single quotes q(literal text)
7036 ' single quotes 'literal text'
7037 qq double quotes qq(interpolate $here please)
7038 " double quotes "interpolate $here please"
7039 qx backticks qx(/bin/ls -l)
7040 ` backticks `/bin/ls -l`
7041 qw quote words @EXPORT_OK = qw( func() $spam )
7042 m// regexp match m/this/
7043 s/// regexp substitute s/this/that/
7044 tr/// string transliterate tr/this/that/
7045 y/// string transliterate y/this/that/
7046 ($*@) sub prototypes sub foo ($)
7047 (stuff) sub attr parameters sub foo : attr(stuff)
7048 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
7050 In most of these cases (all but <>, patterns and transliterate)
7051 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
7052 calls scan_str(). s/// makes yylex() call scan_subst() which calls
7053 scan_str(). tr/// and y/// make yylex() call scan_trans() which
7056 It skips whitespace before the string starts, and treats the first
7057 character as the delimiter. If the delimiter is one of ([{< then
7058 the corresponding "close" character )]}> is used as the closing
7059 delimiter. It allows quoting of delimiters, and if the string has
7060 balanced delimiters ([{<>}]) it allows nesting.
7062 On success, the SV with the resulting string is put into lex_stuff or,
7063 if that is already non-NULL, into lex_repl. The second case occurs only
7064 when parsing the RHS of the special constructs s/// and tr/// (y///).
7065 For convenience, the terminating delimiter character is stuffed into
7070 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
7072 SV *sv; /* scalar value: string */
7073 char *tmps; /* temp string, used for delimiter matching */
7074 register char *s = start; /* current position in the buffer */
7075 register char term; /* terminating character */
7076 register char *to; /* current position in the sv's data */
7077 I32 brackets = 1; /* bracket nesting level */
7078 bool has_utf8 = FALSE; /* is there any utf8 content? */
7079 I32 termcode; /* terminating char. code */
7080 U8 termstr[UTF8_MAXLEN]; /* terminating string */
7081 STRLEN termlen; /* length of terminating string */
7082 char *last = NULL; /* last position for nesting bracket */
7084 /* skip space before the delimiter */
7088 /* mark where we are, in case we need to report errors */
7091 /* after skipping whitespace, the next character is the terminator */
7094 termcode = termstr[0] = term;
7098 termcode = utf8_to_uvchr((U8*)s, &termlen);
7099 Copy(s, termstr, termlen, U8);
7100 if (!UTF8_IS_INVARIANT(term))
7104 /* mark where we are */
7105 PL_multi_start = CopLINE(PL_curcop);
7106 PL_multi_open = term;
7108 /* find corresponding closing delimiter */
7109 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
7110 termcode = termstr[0] = term = tmps[5];
7112 PL_multi_close = term;
7114 /* create a new SV to hold the contents. 87 is leak category, I'm
7115 assuming. 79 is the SV's initial length. What a random number. */
7117 sv_upgrade(sv, SVt_PVIV);
7118 SvIVX(sv) = termcode;
7119 (void)SvPOK_only(sv); /* validate pointer */
7121 /* move past delimiter and try to read a complete string */
7123 sv_catpvn(sv, s, termlen);
7126 if (PL_encoding && !UTF) {
7130 int offset = s - SvPVX(PL_linestr);
7131 bool found = sv_cat_decode(sv, PL_encoding, PL_linestr,
7132 &offset, (char*)termstr, termlen);
7133 char *ns = SvPVX(PL_linestr) + offset;
7134 char *svlast = SvEND(sv) - 1;
7136 for (; s < ns; s++) {
7137 if (*s == '\n' && !PL_rsfp)
7138 CopLINE_inc(PL_curcop);
7141 goto read_more_line;
7143 /* handle quoted delimiters */
7144 if (SvCUR(sv) > 1 && *(svlast-1) == '\\') {
7146 for (t = svlast-2; t >= SvPVX(sv) && *t == '\\';)
7148 if ((svlast-1 - t) % 2) {
7152 SvCUR_set(sv, SvCUR(sv) - 1);
7157 if (PL_multi_open == PL_multi_close) {
7164 for (w = t = last; t < svlast; w++, t++) {
7165 /* At here, all closes are "was quoted" one,
7166 so we don't check PL_multi_close. */
7168 if (!keep_quoted && *(t+1) == PL_multi_open)
7173 else if (*t == PL_multi_open)
7181 SvCUR_set(sv, w - SvPVX(sv));
7184 if (--brackets <= 0)
7190 SvCUR_set(sv, SvCUR(sv) - 1);
7196 /* extend sv if need be */
7197 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
7198 /* set 'to' to the next character in the sv's string */
7199 to = SvPVX(sv)+SvCUR(sv);
7201 /* if open delimiter is the close delimiter read unbridle */
7202 if (PL_multi_open == PL_multi_close) {
7203 for (; s < PL_bufend; s++,to++) {
7204 /* embedded newlines increment the current line number */
7205 if (*s == '\n' && !PL_rsfp)
7206 CopLINE_inc(PL_curcop);
7207 /* handle quoted delimiters */
7208 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
7209 if (!keep_quoted && s[1] == term)
7211 /* any other quotes are simply copied straight through */
7215 /* terminate when run out of buffer (the for() condition), or
7216 have found the terminator */
7217 else if (*s == term) {
7220 if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen))
7223 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
7229 /* if the terminator isn't the same as the start character (e.g.,
7230 matched brackets), we have to allow more in the quoting, and
7231 be prepared for nested brackets.
7234 /* read until we run out of string, or we find the terminator */
7235 for (; s < PL_bufend; s++,to++) {
7236 /* embedded newlines increment the line count */
7237 if (*s == '\n' && !PL_rsfp)
7238 CopLINE_inc(PL_curcop);
7239 /* backslashes can escape the open or closing characters */
7240 if (*s == '\\' && s+1 < PL_bufend) {
7242 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
7247 /* allow nested opens and closes */
7248 else if (*s == PL_multi_close && --brackets <= 0)
7250 else if (*s == PL_multi_open)
7252 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
7257 /* terminate the copied string and update the sv's end-of-string */
7259 SvCUR_set(sv, to - SvPVX(sv));
7262 * this next chunk reads more into the buffer if we're not done yet
7266 break; /* handle case where we are done yet :-) */
7268 #ifndef PERL_STRICT_CR
7269 if (to - SvPVX(sv) >= 2) {
7270 if ((to[-2] == '\r' && to[-1] == '\n') ||
7271 (to[-2] == '\n' && to[-1] == '\r'))
7275 SvCUR_set(sv, to - SvPVX(sv));
7277 else if (to[-1] == '\r')
7280 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
7285 /* if we're out of file, or a read fails, bail and reset the current
7286 line marker so we can report where the unterminated string began
7289 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
7291 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
7294 /* we read a line, so increment our line counter */
7295 CopLINE_inc(PL_curcop);
7297 /* update debugger info */
7298 if (PERLDB_LINE && PL_curstash != PL_debstash) {
7299 SV *sv = NEWSV(88,0);
7301 sv_upgrade(sv, SVt_PVMG);
7302 sv_setsv(sv,PL_linestr);
7305 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
7308 /* having changed the buffer, we must update PL_bufend */
7309 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7310 PL_last_lop = PL_last_uni = Nullch;
7313 /* at this point, we have successfully read the delimited string */
7315 if (!PL_encoding || UTF) {
7317 sv_catpvn(sv, s, termlen);
7320 if (has_utf8 || PL_encoding)
7323 PL_multi_end = CopLINE(PL_curcop);
7325 /* if we allocated too much space, give some back */
7326 if (SvCUR(sv) + 5 < SvLEN(sv)) {
7327 SvLEN_set(sv, SvCUR(sv) + 1);
7328 Renew(SvPVX(sv), SvLEN(sv), char);
7331 /* decide whether this is the first or second quoted string we've read
7344 takes: pointer to position in buffer
7345 returns: pointer to new position in buffer
7346 side-effects: builds ops for the constant in yylval.op
7348 Read a number in any of the formats that Perl accepts:
7350 \d(_?\d)*(\.(\d(_?\d)*)?)?[Ee][\+\-]?(\d(_?\d)*) 12 12.34 12.
7351 \.\d(_?\d)*[Ee][\+\-]?(\d(_?\d)*) .34
7354 0x[0-9A-Fa-f](_?[0-9A-Fa-f])*
7356 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
7359 If it reads a number without a decimal point or an exponent, it will
7360 try converting the number to an integer and see if it can do so
7361 without loss of precision.
7365 Perl_scan_num(pTHX_ char *start, YYSTYPE* lvalp)
7367 register char *s = start; /* current position in buffer */
7368 register char *d; /* destination in temp buffer */
7369 register char *e; /* end of temp buffer */
7370 NV nv; /* number read, as a double */
7371 SV *sv = Nullsv; /* place to put the converted number */
7372 bool floatit; /* boolean: int or float? */
7373 char *lastub = 0; /* position of last underbar */
7374 static char number_too_long[] = "Number too long";
7376 /* We use the first character to decide what type of number this is */
7380 Perl_croak(aTHX_ "panic: scan_num");
7382 /* if it starts with a 0, it could be an octal number, a decimal in
7383 0.13 disguise, or a hexadecimal number, or a binary number. */
7387 u holds the "number so far"
7388 shift the power of 2 of the base
7389 (hex == 4, octal == 3, binary == 1)
7390 overflowed was the number more than we can hold?
7392 Shift is used when we add a digit. It also serves as an "are
7393 we in octal/hex/binary?" indicator to disallow hex characters
7399 bool overflowed = FALSE;
7400 bool just_zero = TRUE; /* just plain 0 or binary number? */
7401 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
7402 static char* bases[5] = { "", "binary", "", "octal",
7404 static char* Bases[5] = { "", "Binary", "", "Octal",
7406 static char *maxima[5] = { "",
7407 "0b11111111111111111111111111111111",
7411 char *base, *Base, *max;
7418 } else if (s[1] == 'b') {
7423 /* check for a decimal in disguise */
7424 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
7426 /* so it must be octal */
7433 if (ckWARN(WARN_SYNTAX))
7434 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7435 "Misplaced _ in number");
7439 base = bases[shift];
7440 Base = Bases[shift];
7441 max = maxima[shift];
7443 /* read the rest of the number */
7445 /* x is used in the overflow test,
7446 b is the digit we're adding on. */
7451 /* if we don't mention it, we're done */
7455 /* _ are ignored -- but warned about if consecutive */
7457 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7458 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7459 "Misplaced _ in number");
7463 /* 8 and 9 are not octal */
7466 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
7470 case '2': case '3': case '4':
7471 case '5': case '6': case '7':
7473 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
7477 b = *s++ & 15; /* ASCII digit -> value of digit */
7481 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
7482 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
7483 /* make sure they said 0x */
7488 /* Prepare to put the digit we have onto the end
7489 of the number so far. We check for overflows.
7495 x = u << shift; /* make room for the digit */
7497 if ((x >> shift) != u
7498 && !(PL_hints & HINT_NEW_BINARY)) {
7501 if (ckWARN_d(WARN_OVERFLOW))
7502 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
7503 "Integer overflow in %s number",
7506 u = x | b; /* add the digit to the end */
7509 n *= nvshift[shift];
7510 /* If an NV has not enough bits in its
7511 * mantissa to represent an UV this summing of
7512 * small low-order numbers is a waste of time
7513 * (because the NV cannot preserve the
7514 * low-order bits anyway): we could just
7515 * remember when did we overflow and in the
7516 * end just multiply n by the right
7524 /* if we get here, we had success: make a scalar value from
7529 /* final misplaced underbar check */
7531 if (ckWARN(WARN_SYNTAX))
7532 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
7537 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
7538 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
7539 "%s number > %s non-portable",
7545 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
7546 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
7547 "%s number > %s non-portable",
7552 if (just_zero && (PL_hints & HINT_NEW_INTEGER))
7553 sv = new_constant(start, s - start, "integer",
7555 else if (PL_hints & HINT_NEW_BINARY)
7556 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
7561 handle decimal numbers.
7562 we're also sent here when we read a 0 as the first digit
7564 case '1': case '2': case '3': case '4': case '5':
7565 case '6': case '7': case '8': case '9': case '.':
7568 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
7571 /* read next group of digits and _ and copy into d */
7572 while (isDIGIT(*s) || *s == '_') {
7573 /* skip underscores, checking for misplaced ones
7577 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7578 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7579 "Misplaced _ in number");
7583 /* check for end of fixed-length buffer */
7585 Perl_croak(aTHX_ number_too_long);
7586 /* if we're ok, copy the character */
7591 /* final misplaced underbar check */
7592 if (lastub && s == lastub + 1) {
7593 if (ckWARN(WARN_SYNTAX))
7594 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
7597 /* read a decimal portion if there is one. avoid
7598 3..5 being interpreted as the number 3. followed
7601 if (*s == '.' && s[1] != '.') {
7606 if (ckWARN(WARN_SYNTAX))
7607 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7608 "Misplaced _ in number");
7612 /* copy, ignoring underbars, until we run out of digits.
7614 for (; isDIGIT(*s) || *s == '_'; s++) {
7615 /* fixed length buffer check */
7617 Perl_croak(aTHX_ number_too_long);
7619 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7620 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7621 "Misplaced _ in number");
7627 /* fractional part ending in underbar? */
7629 if (ckWARN(WARN_SYNTAX))
7630 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7631 "Misplaced _ in number");
7633 if (*s == '.' && isDIGIT(s[1])) {
7634 /* oops, it's really a v-string, but without the "v" */
7640 /* read exponent part, if present */
7641 if (*s && strchr("eE",*s) && strchr("+-0123456789_", s[1])) {
7645 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
7646 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
7648 /* stray preinitial _ */
7650 if (ckWARN(WARN_SYNTAX))
7651 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7652 "Misplaced _ in number");
7656 /* allow positive or negative exponent */
7657 if (*s == '+' || *s == '-')
7660 /* stray initial _ */
7662 if (ckWARN(WARN_SYNTAX))
7663 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7664 "Misplaced _ in number");
7668 /* read digits of exponent */
7669 while (isDIGIT(*s) || *s == '_') {
7672 Perl_croak(aTHX_ number_too_long);
7676 if (ckWARN(WARN_SYNTAX) &&
7677 ((lastub && s == lastub + 1) ||
7678 (!isDIGIT(s[1]) && s[1] != '_')))
7679 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7680 "Misplaced _ in number");
7687 /* make an sv from the string */
7691 We try to do an integer conversion first if no characters
7692 indicating "float" have been found.
7697 int flags = grok_number (PL_tokenbuf, d - PL_tokenbuf, &uv);
7699 if (flags == IS_NUMBER_IN_UV) {
7701 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7704 } else if (flags == (IS_NUMBER_IN_UV | IS_NUMBER_NEG)) {
7705 if (uv <= (UV) IV_MIN)
7706 sv_setiv(sv, -(IV)uv);
7713 /* terminate the string */
7715 nv = Atof(PL_tokenbuf);
7719 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7720 (PL_hints & HINT_NEW_INTEGER) )
7721 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7722 (floatit ? "float" : "integer"),
7726 /* if it starts with a v, it could be a v-string */
7729 sv = NEWSV(92,5); /* preallocate storage space */
7730 s = scan_vstring(s,sv);
7734 /* make the op for the constant and return */
7737 lvalp->opval = newSVOP(OP_CONST, 0, sv);
7739 lvalp->opval = Nullop;
7745 S_scan_formline(pTHX_ register char *s)
7749 SV *stuff = newSVpvn("",0);
7750 bool needargs = FALSE;
7756 #ifdef PERL_STRICT_CR
7757 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
7759 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
7761 if (*t == '\n' || t == PL_bufend) {
7766 if (PL_in_eval && !PL_rsfp) {
7767 eol = memchr(s,'\n',PL_bufend-s);
7772 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7774 for (t = s; t < eol; t++) {
7775 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7777 goto enough; /* ~~ must be first line in formline */
7779 if (*t == '@' || *t == '^')
7783 sv_catpvn(stuff, s, eol-s);
7784 #ifndef PERL_STRICT_CR
7785 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7786 char *end = SvPVX(stuff) + SvCUR(stuff);
7798 s = filter_gets(PL_linestr, PL_rsfp, 0);
7799 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7800 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7801 PL_last_lop = PL_last_uni = Nullch;
7813 PL_lex_state = LEX_NORMAL;
7814 PL_nextval[PL_nexttoke].ival = 0;
7818 PL_lex_state = LEX_FORMLINE;
7820 if (UTF && is_utf8_string((U8*)SvPVX(stuff), SvCUR(stuff)))
7822 else if (PL_encoding)
7823 sv_recode_to_utf8(stuff, PL_encoding);
7825 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7827 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7831 SvREFCNT_dec(stuff);
7833 PL_lex_formbrack = 0;
7844 PL_cshlen = strlen(PL_cshname);
7849 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7851 I32 oldsavestack_ix = PL_savestack_ix;
7852 CV* outsidecv = PL_compcv;
7855 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7857 SAVEI32(PL_subline);
7858 save_item(PL_subname);
7859 SAVESPTR(PL_compcv);
7861 PL_compcv = (CV*)NEWSV(1104,0);
7862 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7863 CvFLAGS(PL_compcv) |= flags;
7865 PL_subline = CopLINE(PL_curcop);
7866 CvPADLIST(PL_compcv) = pad_new(padnew_SAVE|padnew_SAVESUB);
7867 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7868 CvOUTSIDE_SEQ(PL_compcv) = PL_cop_seqmax;
7870 return oldsavestack_ix;
7874 #pragma segment Perl_yylex
7877 Perl_yywarn(pTHX_ char *s)
7879 PL_in_eval |= EVAL_WARNONLY;
7881 PL_in_eval &= ~EVAL_WARNONLY;
7886 Perl_yyerror(pTHX_ char *s)
7889 char *context = NULL;
7893 if (!yychar || (yychar == ';' && !PL_rsfp))
7895 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7896 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7899 The code below is removed for NetWare because it abends/crashes on NetWare
7900 when the script has error such as not having the closing quotes like:
7902 Checking of white spaces is anyway done in NetWare code.
7905 while (isSPACE(*PL_oldoldbufptr))
7908 context = PL_oldoldbufptr;
7909 contlen = PL_bufptr - PL_oldoldbufptr;
7911 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7912 PL_oldbufptr != PL_bufptr) {
7915 The code below is removed for NetWare because it abends/crashes on NetWare
7916 when the script has error such as not having the closing quotes like:
7918 Checking of white spaces is anyway done in NetWare code.
7921 while (isSPACE(*PL_oldbufptr))
7924 context = PL_oldbufptr;
7925 contlen = PL_bufptr - PL_oldbufptr;
7927 else if (yychar > 255)
7928 where = "next token ???";
7929 else if (yychar == -2) { /* YYEMPTY */
7930 if (PL_lex_state == LEX_NORMAL ||
7931 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7932 where = "at end of line";
7933 else if (PL_lex_inpat)
7934 where = "within pattern";
7936 where = "within string";
7939 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7941 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7942 else if (isPRINT_LC(yychar))
7943 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7945 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7946 where = SvPVX(where_sv);
7948 msg = sv_2mortal(newSVpv(s, 0));
7949 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7950 OutCopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7952 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7954 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7955 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7956 Perl_sv_catpvf(aTHX_ msg,
7957 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7958 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7961 if (PL_in_eval & EVAL_WARNONLY && ckWARN_d(WARN_SYNTAX))
7962 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "%"SVf, msg);
7965 if (PL_error_count >= 10) {
7966 if (PL_in_eval && SvCUR(ERRSV))
7967 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
7968 ERRSV, OutCopFILE(PL_curcop));
7970 Perl_croak(aTHX_ "%s has too many errors.\n",
7971 OutCopFILE(PL_curcop));
7974 PL_in_my_stash = Nullhv;
7978 #pragma segment Main
7982 S_swallow_bom(pTHX_ U8 *s)
7985 slen = SvCUR(PL_linestr);
7989 /* UTF-16 little-endian? (or UTF32-LE?) */
7990 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
7991 Perl_croak(aTHX_ "Unsupported script encoding UTF32-LE");
7992 #ifndef PERL_NO_UTF16_FILTER
7993 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF16-LE script encoding (BOM)\n");
7996 if (PL_bufend > (char*)s) {
8000 filter_add(utf16rev_textfilter, NULL);
8001 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
8003 (char*)utf16_to_utf8_reversed(s, news,
8004 PL_bufend - (char*)s - 1,
8006 sv_setpvn(PL_linestr, (const char*)news, newlen);
8008 SvUTF8_on(PL_linestr);
8009 s = (U8*)SvPVX(PL_linestr);
8010 PL_bufend = SvPVX(PL_linestr) + newlen;
8013 Perl_croak(aTHX_ "Unsupported script encoding UTF16-LE");
8018 if (s[1] == 0xFF) { /* UTF-16 big-endian? */
8019 #ifndef PERL_NO_UTF16_FILTER
8020 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (BOM)\n");
8023 if (PL_bufend > (char *)s) {
8027 filter_add(utf16_textfilter, NULL);
8028 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
8030 (char*)utf16_to_utf8(s, news,
8031 PL_bufend - (char*)s,
8033 sv_setpvn(PL_linestr, (const char*)news, newlen);
8035 SvUTF8_on(PL_linestr);
8036 s = (U8*)SvPVX(PL_linestr);
8037 PL_bufend = SvPVX(PL_linestr) + newlen;
8040 Perl_croak(aTHX_ "Unsupported script encoding UTF16-BE");
8045 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
8046 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-8 script encoding (BOM)\n");
8053 if (s[2] == 0xFE && s[3] == 0xFF) {
8054 /* UTF-32 big-endian */
8055 Perl_croak(aTHX_ "Unsupported script encoding UTF32-BE");
8058 else if (s[2] == 0 && s[3] != 0) {
8061 * are a good indicator of UTF-16BE. */
8062 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (no BOM)\n");
8067 if (slen > 3 && s[1] == 0 && s[2] != 0 && s[3] == 0) {
8070 * are a good indicator of UTF-16LE. */
8071 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16LE script encoding (no BOM)\n");
8080 * Restore a source filter.
8084 restore_rsfp(pTHX_ void *f)
8086 PerlIO *fp = (PerlIO*)f;
8088 if (PL_rsfp == PerlIO_stdin())
8089 PerlIO_clearerr(PL_rsfp);
8090 else if (PL_rsfp && (PL_rsfp != fp))
8091 PerlIO_close(PL_rsfp);
8095 #ifndef PERL_NO_UTF16_FILTER
8097 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
8099 I32 count = FILTER_READ(idx+1, sv, maxlen);
8104 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
8105 if (!*SvPV_nolen(sv))
8106 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
8109 tend = utf16_to_utf8((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
8110 sv_usepvn(sv, (char*)tmps, tend - tmps);
8116 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
8118 I32 count = FILTER_READ(idx+1, sv, maxlen);
8123 if (!*SvPV_nolen(sv))
8124 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
8127 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
8128 tend = utf16_to_utf8_reversed((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
8129 sv_usepvn(sv, (char*)tmps, tend - tmps);
8136 Returns a pointer to the next character after the parsed
8137 vstring, as well as updating the passed in sv.
8139 Function must be called like
8142 s = scan_vstring(s,sv);
8144 The sv should already be large enough to store the vstring
8145 passed in, for performance reasons.
8150 Perl_scan_vstring(pTHX_ char *s, SV *sv)
8154 if (*pos == 'v') pos++; /* get past 'v' */
8155 while (pos < PL_bufend && (isDIGIT(*pos) || *pos == '_'))
8158 /* this may not be a v-string if followed by => */
8160 while (next < PL_bufend && isSPACE(*next))
8162 if ((PL_bufend - next) >= 2 && *next == '=' && next[1] == '>' ) {
8163 /* return string not v-string */
8164 sv_setpvn(sv,(char *)s,pos-s);
8169 if (!isALPHA(*pos)) {
8171 U8 tmpbuf[UTF8_MAXLEN+1];
8174 if (*s == 'v') s++; /* get past 'v' */
8176 sv_setpvn(sv, "", 0);
8181 /* this is atoi() that tolerates underscores */
8184 while (--end >= s) {
8189 rev += (*end - '0') * mult;
8191 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
8192 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
8193 "Integer overflow in decimal number");
8197 if (rev > 0x7FFFFFFF)
8198 Perl_croak(aTHX_ "In EBCDIC the v-string components cannot exceed 2147483647");
8200 /* Append native character for the rev point */
8201 tmpend = uvchr_to_utf8(tmpbuf, rev);
8202 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
8203 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(rev)))
8205 if (pos + 1 < PL_bufend && *pos == '.' && isDIGIT(pos[1]))
8211 while (pos < PL_bufend && (isDIGIT(*pos) || *pos == '_'))
8215 sv_magic(sv,NULL,PERL_MAGIC_vstring,(const char*)start, pos-start);