3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
4 * 2000, 2001, 2002, 2003, 2004, 2005, 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 const char ident_too_long[] =
30 "Identifier too long";
31 static const char c_without_g[] =
32 "Use of /c modifier is meaningless without /g";
33 static const char c_in_subst[] =
34 "Use of /c modifier is meaningless in s///";
36 static void restore_rsfp(pTHX_ void *f);
37 #ifndef PERL_NO_UTF16_FILTER
38 static I32 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen);
39 static I32 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen);
42 #define XFAKEBRACK 128
45 #ifdef USE_UTF8_SCRIPTS
46 # define UTF (!IN_BYTES)
48 # define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || (PL_hints & HINT_UTF8))
51 /* In variables named $^X, these are the legal values for X.
52 * 1999-02-27 mjd-perl-patch@plover.com */
53 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
55 /* On MacOS, respect nonbreaking spaces */
56 #ifdef MACOS_TRADITIONAL
57 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
59 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
62 /* LEX_* are values for PL_lex_state, the state of the lexer.
63 * They are arranged oddly so that the guard on the switch statement
64 * can get by with a single comparison (if the compiler is smart enough).
67 /* #define LEX_NOTPARSING 11 is done in perl.h. */
70 #define LEX_INTERPNORMAL 9
71 #define LEX_INTERPCASEMOD 8
72 #define LEX_INTERPPUSH 7
73 #define LEX_INTERPSTART 6
74 #define LEX_INTERPEND 5
75 #define LEX_INTERPENDMAYBE 4
76 #define LEX_INTERPCONCAT 3
77 #define LEX_INTERPCONST 2
78 #define LEX_FORMLINE 1
79 #define LEX_KNOWNEXT 0
82 static const char* const lex_state_names[] = {
101 #include "keywords.h"
103 /* CLINE is a macro that ensures PL_copline has a sane value */
108 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
111 * Convenience functions to return different tokens and prime the
112 * lexer for the next token. They all take an argument.
114 * TOKEN : generic token (used for '(', DOLSHARP, etc)
115 * OPERATOR : generic operator
116 * AOPERATOR : assignment operator
117 * PREBLOCK : beginning the block after an if, while, foreach, ...
118 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
119 * PREREF : *EXPR where EXPR is not a simple identifier
120 * TERM : expression term
121 * LOOPX : loop exiting command (goto, last, dump, etc)
122 * FTST : file test operator
123 * FUN0 : zero-argument function
124 * FUN1 : not used, except for not, which isn't a UNIOP
125 * BOop : bitwise or or xor
127 * SHop : shift operator
128 * PWop : power operator
129 * PMop : pattern-matching operator
130 * Aop : addition-level operator
131 * Mop : multiplication-level operator
132 * Eop : equality-testing operator
133 * Rop : relational operator <= != gt
135 * Also see LOP and lop() below.
138 #ifdef DEBUGGING /* Serve -DT. */
139 # define REPORT(retval) tokereport(s,(int)retval)
141 # define REPORT(retval) (retval)
144 #define TOKEN(retval) return ( PL_bufptr = s, REPORT(retval))
145 #define OPERATOR(retval) return (PL_expect = XTERM, PL_bufptr = s, REPORT(retval))
146 #define AOPERATOR(retval) return ao((PL_expect = XTERM, PL_bufptr = s, REPORT(retval)))
147 #define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s, REPORT(retval))
148 #define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s, REPORT(retval))
149 #define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s, REPORT(retval))
150 #define TERM(retval) return (CLINE, PL_expect = XOPERATOR, PL_bufptr = s, REPORT(retval))
151 #define LOOPX(f) return (yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)LOOPEX))
152 #define FTST(f) return (yylval.ival=f, PL_expect=XTERMORDORDOR, PL_bufptr=s, REPORT((int)UNIOP))
153 #define FUN0(f) return (yylval.ival=f, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC0))
154 #define FUN1(f) return (yylval.ival=f, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC1))
155 #define BOop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)BITOROP)))
156 #define BAop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)BITANDOP)))
157 #define SHop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)SHIFTOP)))
158 #define PWop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)POWOP)))
159 #define PMop(f) return(yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)MATCHOP))
160 #define Aop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)ADDOP)))
161 #define Mop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)MULOP)))
162 #define Eop(f) return (yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)EQOP))
163 #define Rop(f) return (yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)RELOP))
165 /* This bit of chicanery makes a unary function followed by
166 * a parenthesis into a function with one argument, highest precedence.
167 * The UNIDOR macro is for unary functions that can be followed by the //
168 * operator (such as C<shift // 0>).
170 #define UNI2(f,x) { \
174 PL_last_uni = PL_oldbufptr; \
175 PL_last_lop_op = f; \
177 return REPORT( (int)FUNC1 ); \
179 return REPORT( *s=='(' ? (int)FUNC1 : (int)UNIOP ); \
181 #define UNI(f) UNI2(f,XTERM)
182 #define UNIDOR(f) UNI2(f,XTERMORDORDOR)
184 #define UNIBRACK(f) { \
187 PL_last_uni = PL_oldbufptr; \
189 return REPORT( (int)FUNC1 ); \
191 return REPORT( (*s == '(') ? (int)FUNC1 : (int)UNIOP ); \
194 /* grandfather return to old style */
195 #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
199 /* how to interpret the yylval associated with the token */
203 TOKENTYPE_OPNUM, /* yylval.ival contains an opcode number */
209 static struct debug_tokens { const int token, type; const char *name; }
210 const debug_tokens[] =
212 { ADDOP, TOKENTYPE_OPNUM, "ADDOP" },
213 { ANDAND, TOKENTYPE_NONE, "ANDAND" },
214 { ANDOP, TOKENTYPE_NONE, "ANDOP" },
215 { ANONSUB, TOKENTYPE_IVAL, "ANONSUB" },
216 { ARROW, TOKENTYPE_NONE, "ARROW" },
217 { ASSIGNOP, TOKENTYPE_OPNUM, "ASSIGNOP" },
218 { BITANDOP, TOKENTYPE_OPNUM, "BITANDOP" },
219 { BITOROP, TOKENTYPE_OPNUM, "BITOROP" },
220 { COLONATTR, TOKENTYPE_NONE, "COLONATTR" },
221 { CONTINUE, TOKENTYPE_NONE, "CONTINUE" },
222 { DO, TOKENTYPE_NONE, "DO" },
223 { DOLSHARP, TOKENTYPE_NONE, "DOLSHARP" },
224 { DORDOR, TOKENTYPE_NONE, "DORDOR" },
225 { DOROP, TOKENTYPE_OPNUM, "DOROP" },
226 { DOTDOT, TOKENTYPE_IVAL, "DOTDOT" },
227 { ELSE, TOKENTYPE_NONE, "ELSE" },
228 { ELSIF, TOKENTYPE_IVAL, "ELSIF" },
229 { EQOP, TOKENTYPE_OPNUM, "EQOP" },
230 { FOR, TOKENTYPE_IVAL, "FOR" },
231 { FORMAT, TOKENTYPE_NONE, "FORMAT" },
232 { FUNC, TOKENTYPE_OPNUM, "FUNC" },
233 { FUNC0, TOKENTYPE_OPNUM, "FUNC0" },
234 { FUNC0SUB, TOKENTYPE_OPVAL, "FUNC0SUB" },
235 { FUNC1, TOKENTYPE_OPNUM, "FUNC1" },
236 { FUNCMETH, TOKENTYPE_OPVAL, "FUNCMETH" },
237 { HASHBRACK, TOKENTYPE_NONE, "HASHBRACK" },
238 { IF, TOKENTYPE_IVAL, "IF" },
239 { LABEL, TOKENTYPE_PVAL, "LABEL" },
240 { LOCAL, TOKENTYPE_IVAL, "LOCAL" },
241 { LOOPEX, TOKENTYPE_OPNUM, "LOOPEX" },
242 { LSTOP, TOKENTYPE_OPNUM, "LSTOP" },
243 { LSTOPSUB, TOKENTYPE_OPVAL, "LSTOPSUB" },
244 { MATCHOP, TOKENTYPE_OPNUM, "MATCHOP" },
245 { METHOD, TOKENTYPE_OPVAL, "METHOD" },
246 { MULOP, TOKENTYPE_OPNUM, "MULOP" },
247 { MY, TOKENTYPE_IVAL, "MY" },
248 { MYSUB, TOKENTYPE_NONE, "MYSUB" },
249 { NOAMP, TOKENTYPE_NONE, "NOAMP" },
250 { NOTOP, TOKENTYPE_NONE, "NOTOP" },
251 { OROP, TOKENTYPE_IVAL, "OROP" },
252 { OROR, TOKENTYPE_NONE, "OROR" },
253 { PACKAGE, TOKENTYPE_NONE, "PACKAGE" },
254 { PMFUNC, TOKENTYPE_OPVAL, "PMFUNC" },
255 { POSTDEC, TOKENTYPE_NONE, "POSTDEC" },
256 { POSTINC, TOKENTYPE_NONE, "POSTINC" },
257 { POWOP, TOKENTYPE_OPNUM, "POWOP" },
258 { PREDEC, TOKENTYPE_NONE, "PREDEC" },
259 { PREINC, TOKENTYPE_NONE, "PREINC" },
260 { PRIVATEREF, TOKENTYPE_OPVAL, "PRIVATEREF" },
261 { REFGEN, TOKENTYPE_NONE, "REFGEN" },
262 { RELOP, TOKENTYPE_OPNUM, "RELOP" },
263 { SHIFTOP, TOKENTYPE_OPNUM, "SHIFTOP" },
264 { SUB, TOKENTYPE_NONE, "SUB" },
265 { THING, TOKENTYPE_OPVAL, "THING" },
266 { UMINUS, TOKENTYPE_NONE, "UMINUS" },
267 { UNIOP, TOKENTYPE_OPNUM, "UNIOP" },
268 { UNIOPSUB, TOKENTYPE_OPVAL, "UNIOPSUB" },
269 { UNLESS, TOKENTYPE_IVAL, "UNLESS" },
270 { UNTIL, TOKENTYPE_IVAL, "UNTIL" },
271 { USE, TOKENTYPE_IVAL, "USE" },
272 { WHILE, TOKENTYPE_IVAL, "WHILE" },
273 { WORD, TOKENTYPE_OPVAL, "WORD" },
274 { 0, TOKENTYPE_NONE, 0 }
277 /* dump the returned token in rv, plus any optional arg in yylval */
280 S_tokereport(pTHX_ const char* s, I32 rv)
283 const char *name = Nullch;
284 enum token_type type = TOKENTYPE_NONE;
285 const struct debug_tokens *p;
286 SV* const report = newSVpvn("<== ", 4);
288 for (p = debug_tokens; p->token; p++) {
289 if (p->token == (int)rv) {
296 Perl_sv_catpv(aTHX_ report, name);
297 else if ((char)rv > ' ' && (char)rv < '~')
298 Perl_sv_catpvf(aTHX_ report, "'%c'", (char)rv);
300 Perl_sv_catpv(aTHX_ report, "EOF");
302 Perl_sv_catpvf(aTHX_ report, "?? %"IVdf, (IV)rv);
305 case TOKENTYPE_GVVAL: /* doesn't appear to be used */
308 Perl_sv_catpvf(aTHX_ report, "(ival=%"IVdf")", (IV)yylval.ival);
310 case TOKENTYPE_OPNUM:
311 Perl_sv_catpvf(aTHX_ report, "(ival=op_%s)",
312 PL_op_name[yylval.ival]);
315 Perl_sv_catpvf(aTHX_ report, "(pval=\"%s\")", yylval.pval);
317 case TOKENTYPE_OPVAL:
319 Perl_sv_catpvf(aTHX_ report, "(opval=op_%s)",
320 PL_op_name[yylval.opval->op_type]);
322 Perl_sv_catpv(aTHX_ report, "(opval=null)");
325 Perl_sv_catpvf(aTHX_ report, " at line %"IVdf" [", (IV)CopLINE(PL_curcop));
326 if (s - PL_bufptr > 0)
327 sv_catpvn(report, PL_bufptr, s - PL_bufptr);
329 if (PL_oldbufptr && *PL_oldbufptr)
330 sv_catpv(report, PL_tokenbuf);
332 PerlIO_printf(Perl_debug_log, "### %s]\n", SvPV_nolen_const(report));
342 * This subroutine detects &&=, ||=, and //= and turns an ANDAND, OROR or DORDOR
343 * into an OP_ANDASSIGN, OP_ORASSIGN, or OP_DORASSIGN
347 S_ao(pTHX_ int toketype)
349 if (*PL_bufptr == '=') {
351 if (toketype == ANDAND)
352 yylval.ival = OP_ANDASSIGN;
353 else if (toketype == OROR)
354 yylval.ival = OP_ORASSIGN;
355 else if (toketype == DORDOR)
356 yylval.ival = OP_DORASSIGN;
364 * When Perl expects an operator and finds something else, no_op
365 * prints the warning. It always prints "<something> found where
366 * operator expected. It prints "Missing semicolon on previous line?"
367 * if the surprise occurs at the start of the line. "do you need to
368 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
369 * where the compiler doesn't know if foo is a method call or a function.
370 * It prints "Missing operator before end of line" if there's nothing
371 * after the missing operator, or "... before <...>" if there is something
372 * after the missing operator.
376 S_no_op(pTHX_ const char *what, char *s)
378 char * const oldbp = PL_bufptr;
379 const bool is_first = (PL_oldbufptr == PL_linestart);
385 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
386 if (ckWARN_d(WARN_SYNTAX)) {
388 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
389 "\t(Missing semicolon on previous line?)\n");
390 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
392 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy_if(t,UTF) || *t == ':'); t++) ;
393 if (t < PL_bufptr && isSPACE(*t))
394 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
395 "\t(Do you need to predeclare %.*s?)\n",
396 t - PL_oldoldbufptr, PL_oldoldbufptr);
400 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
401 "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp);
409 * Complain about missing quote/regexp/heredoc terminator.
410 * If it's called with (char *)NULL then it cauterizes the line buffer.
411 * If we're in a delimited string and the delimiter is a control
412 * character, it's reformatted into a two-char sequence like ^C.
417 S_missingterm(pTHX_ char *s)
422 char * const nl = strrchr(s,'\n');
428 iscntrl(PL_multi_close)
430 PL_multi_close < 32 || PL_multi_close == 127
434 tmpbuf[1] = toCTRL(PL_multi_close);
439 *tmpbuf = (char)PL_multi_close;
443 q = strchr(s,'"') ? '\'' : '"';
444 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
452 Perl_deprecate(pTHX_ const char *s)
454 if (ckWARN(WARN_DEPRECATED))
455 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED), "Use of %s is deprecated", s);
459 Perl_deprecate_old(pTHX_ const char *s)
461 /* This function should NOT be called for any new deprecated warnings */
462 /* Use Perl_deprecate instead */
464 /* It is here to maintain backward compatibility with the pre-5.8 */
465 /* warnings category hierarchy. The "deprecated" category used to */
466 /* live under the "syntax" category. It is now a top-level category */
467 /* in its own right. */
469 if (ckWARN2(WARN_DEPRECATED, WARN_SYNTAX))
470 Perl_warner(aTHX_ packWARN2(WARN_DEPRECATED, WARN_SYNTAX),
471 "Use of %s is deprecated", s);
476 * Deprecate a comma-less variable list.
482 deprecate_old("comma-less variable list");
486 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
487 * utf16-to-utf8-reversed.
490 #ifdef PERL_CR_FILTER
494 register const char *s = SvPVX_const(sv);
495 register const char * const e = s + SvCUR(sv);
496 /* outer loop optimized to do nothing if there are no CR-LFs */
498 if (*s++ == '\r' && *s == '\n') {
499 /* hit a CR-LF, need to copy the rest */
500 register char *d = s - 1;
503 if (*s == '\r' && s[1] == '\n')
514 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
516 const I32 count = FILTER_READ(idx+1, sv, maxlen);
517 if (count > 0 && !maxlen)
525 * Initialize variables. Uses the Perl save_stack to save its state (for
526 * recursive calls to the parser).
530 Perl_lex_start(pTHX_ SV *line)
535 SAVEI32(PL_lex_dojoin);
536 SAVEI32(PL_lex_brackets);
537 SAVEI32(PL_lex_casemods);
538 SAVEI32(PL_lex_starts);
539 SAVEI32(PL_lex_state);
540 SAVEVPTR(PL_lex_inpat);
541 SAVEI32(PL_lex_inwhat);
542 if (PL_lex_state == LEX_KNOWNEXT) {
543 I32 toke = PL_nexttoke;
544 while (--toke >= 0) {
545 SAVEI32(PL_nexttype[toke]);
546 SAVEVPTR(PL_nextval[toke]);
548 SAVEI32(PL_nexttoke);
550 SAVECOPLINE(PL_curcop);
553 SAVEPPTR(PL_oldbufptr);
554 SAVEPPTR(PL_oldoldbufptr);
555 SAVEPPTR(PL_last_lop);
556 SAVEPPTR(PL_last_uni);
557 SAVEPPTR(PL_linestart);
558 SAVESPTR(PL_linestr);
559 SAVEGENERICPV(PL_lex_brackstack);
560 SAVEGENERICPV(PL_lex_casestack);
561 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
562 SAVESPTR(PL_lex_stuff);
563 SAVEI32(PL_lex_defer);
564 SAVEI32(PL_sublex_info.sub_inwhat);
565 SAVESPTR(PL_lex_repl);
567 SAVEINT(PL_lex_expect);
569 PL_lex_state = LEX_NORMAL;
573 Newx(PL_lex_brackstack, 120, char);
574 Newx(PL_lex_casestack, 12, char);
576 *PL_lex_casestack = '\0';
579 PL_lex_stuff = Nullsv;
580 PL_lex_repl = Nullsv;
584 PL_sublex_info.sub_inwhat = 0;
586 if (SvREADONLY(PL_linestr))
587 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
588 s = SvPV_const(PL_linestr, len);
589 if (!len || s[len-1] != ';') {
590 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
591 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
592 sv_catpvn(PL_linestr, "\n;", 2);
594 SvTEMP_off(PL_linestr);
595 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
596 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
597 PL_last_lop = PL_last_uni = Nullch;
603 * Finalizer for lexing operations. Must be called when the parser is
604 * done with the lexer.
610 PL_doextract = FALSE;
615 * This subroutine has nothing to do with tilting, whether at windmills
616 * or pinball tables. Its name is short for "increment line". It
617 * increments the current line number in CopLINE(PL_curcop) and checks
618 * to see whether the line starts with a comment of the form
619 * # line 500 "foo.pm"
620 * If so, it sets the current line number and file to the values in the comment.
624 S_incline(pTHX_ char *s)
631 CopLINE_inc(PL_curcop);
634 while (SPACE_OR_TAB(*s)) s++;
635 if (strnEQ(s, "line", 4))
639 if (SPACE_OR_TAB(*s))
643 while (SPACE_OR_TAB(*s)) s++;
649 while (SPACE_OR_TAB(*s))
651 if (*s == '"' && (t = strchr(s+1, '"'))) {
656 for (t = s; !isSPACE(*t); t++) ;
659 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
661 if (*e != '\n' && *e != '\0')
662 return; /* false alarm */
667 CopFILE_free(PL_curcop);
668 CopFILE_set(PL_curcop, s);
671 CopLINE_set(PL_curcop, atoi(n)-1);
676 * Called to gobble the appropriate amount and type of whitespace.
677 * Skips comments as well.
681 S_skipspace(pTHX_ register char *s)
683 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
684 while (s < PL_bufend && SPACE_OR_TAB(*s))
690 SSize_t oldprevlen, oldoldprevlen;
691 SSize_t oldloplen = 0, oldunilen = 0;
692 while (s < PL_bufend && isSPACE(*s)) {
693 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
698 if (s < PL_bufend && *s == '#') {
699 while (s < PL_bufend && *s != '\n')
703 if (PL_in_eval && !PL_rsfp) {
710 /* only continue to recharge the buffer if we're at the end
711 * of the buffer, we're not reading from a source filter, and
712 * we're in normal lexing mode
714 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
715 PL_lex_state == LEX_FORMLINE)
718 /* try to recharge the buffer */
719 if ((s = filter_gets(PL_linestr, PL_rsfp,
720 (prevlen = SvCUR(PL_linestr)))) == Nullch)
722 /* end of file. Add on the -p or -n magic */
725 ";}continue{print or die qq(-p destination: $!\\n);}");
726 PL_minus_n = PL_minus_p = 0;
728 else if (PL_minus_n) {
729 sv_setpvn(PL_linestr, ";}", 2);
733 sv_setpvn(PL_linestr,";", 1);
735 /* reset variables for next time we lex */
736 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
738 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
739 PL_last_lop = PL_last_uni = Nullch;
741 /* Close the filehandle. Could be from -P preprocessor,
742 * STDIN, or a regular file. If we were reading code from
743 * STDIN (because the commandline held no -e or filename)
744 * then we don't close it, we reset it so the code can
745 * read from STDIN too.
748 if (PL_preprocess && !PL_in_eval)
749 (void)PerlProc_pclose(PL_rsfp);
750 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
751 PerlIO_clearerr(PL_rsfp);
753 (void)PerlIO_close(PL_rsfp);
758 /* not at end of file, so we only read another line */
759 /* make corresponding updates to old pointers, for yyerror() */
760 oldprevlen = PL_oldbufptr - PL_bufend;
761 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
763 oldunilen = PL_last_uni - PL_bufend;
765 oldloplen = PL_last_lop - PL_bufend;
766 PL_linestart = PL_bufptr = s + prevlen;
767 PL_bufend = s + SvCUR(PL_linestr);
769 PL_oldbufptr = s + oldprevlen;
770 PL_oldoldbufptr = s + oldoldprevlen;
772 PL_last_uni = s + oldunilen;
774 PL_last_lop = s + oldloplen;
777 /* debugger active and we're not compiling the debugger code,
778 * so store the line into the debugger's array of lines
780 if (PERLDB_LINE && PL_curstash != PL_debstash) {
781 SV * const sv = NEWSV(85,0);
783 sv_upgrade(sv, SVt_PVMG);
784 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
787 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
794 * Check the unary operators to ensure there's no ambiguity in how they're
795 * used. An ambiguous piece of code would be:
797 * This doesn't mean rand() + 5. Because rand() is a unary operator,
798 * the +5 is its argument.
807 if (PL_oldoldbufptr != PL_last_uni)
809 while (isSPACE(*PL_last_uni))
811 for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ;
812 if ((t = strchr(s, '(')) && t < PL_bufptr)
814 if (ckWARN_d(WARN_AMBIGUOUS)){
817 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
818 "Warning: Use of \"%s\" without parentheses is ambiguous",
825 * LOP : macro to build a list operator. Its behaviour has been replaced
826 * with a subroutine, S_lop() for which LOP is just another name.
829 #define LOP(f,x) return lop(f,x,s)
833 * Build a list operator (or something that might be one). The rules:
834 * - if we have a next token, then it's a list operator [why?]
835 * - if the next thing is an opening paren, then it's a function
836 * - else it's a list operator
840 S_lop(pTHX_ I32 f, int x, char *s)
846 PL_last_lop = PL_oldbufptr;
847 PL_last_lop_op = (OPCODE)f;
849 return REPORT(LSTOP);
856 return REPORT(LSTOP);
861 * When the lexer realizes it knows the next token (for instance,
862 * it is reordering tokens for the parser) then it can call S_force_next
863 * to know what token to return the next time the lexer is called. Caller
864 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
865 * handles the token correctly.
869 S_force_next(pTHX_ I32 type)
871 PL_nexttype[PL_nexttoke] = type;
873 if (PL_lex_state != LEX_KNOWNEXT) {
874 PL_lex_defer = PL_lex_state;
875 PL_lex_expect = PL_expect;
876 PL_lex_state = LEX_KNOWNEXT;
881 S_newSV_maybe_utf8(pTHX_ const char *start, STRLEN len)
883 SV * const sv = newSVpvn(start,len);
884 if (UTF && !IN_BYTES && is_utf8_string((const U8*)start, len))
891 * When the lexer knows the next thing is a word (for instance, it has
892 * just seen -> and it knows that the next char is a word char, then
893 * it calls S_force_word to stick the next word into the PL_next lookahead.
896 * char *start : buffer position (must be within PL_linestr)
897 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
898 * int check_keyword : if true, Perl checks to make sure the word isn't
899 * a keyword (do this if the word is a label, e.g. goto FOO)
900 * int allow_pack : if true, : characters will also be allowed (require,
902 * int allow_initial_tick : used by the "sub" lexer only.
906 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
911 start = skipspace(start);
913 if (isIDFIRST_lazy_if(s,UTF) ||
914 (allow_pack && *s == ':') ||
915 (allow_initial_tick && *s == '\'') )
917 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
918 if (check_keyword && keyword(PL_tokenbuf, len))
920 if (token == METHOD) {
925 PL_expect = XOPERATOR;
928 PL_nextval[PL_nexttoke].opval
929 = (OP*)newSVOP(OP_CONST,0,
930 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
931 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
939 * Called when the lexer wants $foo *foo &foo etc, but the program
940 * text only contains the "foo" portion. The first argument is a pointer
941 * to the "foo", and the second argument is the type symbol to prefix.
942 * Forces the next token to be a "WORD".
943 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
947 S_force_ident(pTHX_ register const char *s, int kind)
950 OP* const o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
951 PL_nextval[PL_nexttoke].opval = o;
954 o->op_private = OPpCONST_ENTERED;
955 /* XXX see note in pp_entereval() for why we forgo typo
956 warnings if the symbol must be introduced in an eval.
958 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
959 kind == '$' ? SVt_PV :
960 kind == '@' ? SVt_PVAV :
961 kind == '%' ? SVt_PVHV :
969 Perl_str_to_version(pTHX_ SV *sv)
974 const char *start = SvPV_const(sv,len);
975 const char * const end = start + len;
976 const bool utf = SvUTF8(sv) ? TRUE : FALSE;
977 while (start < end) {
981 n = utf8n_to_uvchr((U8*)start, len, &skip, 0);
986 retval += ((NV)n)/nshift;
995 * Forces the next token to be a version number.
996 * If the next token appears to be an invalid version number, (e.g. "v2b"),
997 * and if "guessing" is TRUE, then no new token is created (and the caller
998 * must use an alternative parsing method).
1002 S_force_version(pTHX_ char *s, int guessing)
1004 OP *version = Nullop;
1013 while (isDIGIT(*d) || *d == '_' || *d == '.')
1015 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
1017 s = scan_num(s, &yylval);
1018 version = yylval.opval;
1019 ver = cSVOPx(version)->op_sv;
1020 if (SvPOK(ver) && !SvNIOK(ver)) {
1021 SvUPGRADE(ver, SVt_PVNV);
1022 SvNV_set(ver, str_to_version(ver));
1023 SvNOK_on(ver); /* hint that it is a version */
1030 /* NOTE: The parser sees the package name and the VERSION swapped */
1031 PL_nextval[PL_nexttoke].opval = version;
1039 * Tokenize a quoted string passed in as an SV. It finds the next
1040 * chunk, up to end of string or a backslash. It may make a new
1041 * SV containing that chunk (if HINT_NEW_STRING is on). It also
1046 S_tokeq(pTHX_ SV *sv)
1049 register char *send;
1057 s = SvPV_force(sv, len);
1058 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
1061 while (s < send && *s != '\\')
1066 if ( PL_hints & HINT_NEW_STRING ) {
1067 pv = sv_2mortal(newSVpvn(SvPVX_const(pv), len));
1073 if (s + 1 < send && (s[1] == '\\'))
1074 s++; /* all that, just for this */
1079 SvCUR_set(sv, d - SvPVX_const(sv));
1081 if ( PL_hints & HINT_NEW_STRING )
1082 return new_constant(NULL, 0, "q", sv, pv, "q");
1087 * Now come three functions related to double-quote context,
1088 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
1089 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
1090 * interact with PL_lex_state, and create fake ( ... ) argument lists
1091 * to handle functions and concatenation.
1092 * They assume that whoever calls them will be setting up a fake
1093 * join call, because each subthing puts a ',' after it. This lets
1096 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
1098 * (I'm not sure whether the spurious commas at the end of lcfirst's
1099 * arguments and join's arguments are created or not).
1104 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
1106 * Pattern matching will set PL_lex_op to the pattern-matching op to
1107 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
1109 * OP_CONST and OP_READLINE are easy--just make the new op and return.
1111 * Everything else becomes a FUNC.
1113 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
1114 * had an OP_CONST or OP_READLINE). This just sets us up for a
1115 * call to S_sublex_push().
1119 S_sublex_start(pTHX)
1121 const register I32 op_type = yylval.ival;
1123 if (op_type == OP_NULL) {
1124 yylval.opval = PL_lex_op;
1128 if (op_type == OP_CONST || op_type == OP_READLINE) {
1129 SV *sv = tokeq(PL_lex_stuff);
1131 if (SvTYPE(sv) == SVt_PVIV) {
1132 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
1134 const char *p = SvPV_const(sv, len);
1135 SV * const nsv = newSVpvn(p, len);
1141 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
1142 PL_lex_stuff = Nullsv;
1143 /* Allow <FH> // "foo" */
1144 if (op_type == OP_READLINE)
1145 PL_expect = XTERMORDORDOR;
1149 PL_sublex_info.super_state = PL_lex_state;
1150 PL_sublex_info.sub_inwhat = op_type;
1151 PL_sublex_info.sub_op = PL_lex_op;
1152 PL_lex_state = LEX_INTERPPUSH;
1156 yylval.opval = PL_lex_op;
1166 * Create a new scope to save the lexing state. The scope will be
1167 * ended in S_sublex_done. Returns a '(', starting the function arguments
1168 * to the uc, lc, etc. found before.
1169 * Sets PL_lex_state to LEX_INTERPCONCAT.
1178 PL_lex_state = PL_sublex_info.super_state;
1179 SAVEI32(PL_lex_dojoin);
1180 SAVEI32(PL_lex_brackets);
1181 SAVEI32(PL_lex_casemods);
1182 SAVEI32(PL_lex_starts);
1183 SAVEI32(PL_lex_state);
1184 SAVEVPTR(PL_lex_inpat);
1185 SAVEI32(PL_lex_inwhat);
1186 SAVECOPLINE(PL_curcop);
1187 SAVEPPTR(PL_bufptr);
1188 SAVEPPTR(PL_bufend);
1189 SAVEPPTR(PL_oldbufptr);
1190 SAVEPPTR(PL_oldoldbufptr);
1191 SAVEPPTR(PL_last_lop);
1192 SAVEPPTR(PL_last_uni);
1193 SAVEPPTR(PL_linestart);
1194 SAVESPTR(PL_linestr);
1195 SAVEGENERICPV(PL_lex_brackstack);
1196 SAVEGENERICPV(PL_lex_casestack);
1198 PL_linestr = PL_lex_stuff;
1199 PL_lex_stuff = Nullsv;
1201 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1202 = SvPVX(PL_linestr);
1203 PL_bufend += SvCUR(PL_linestr);
1204 PL_last_lop = PL_last_uni = Nullch;
1205 SAVEFREESV(PL_linestr);
1207 PL_lex_dojoin = FALSE;
1208 PL_lex_brackets = 0;
1209 Newx(PL_lex_brackstack, 120, char);
1210 Newx(PL_lex_casestack, 12, char);
1211 PL_lex_casemods = 0;
1212 *PL_lex_casestack = '\0';
1214 PL_lex_state = LEX_INTERPCONCAT;
1215 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
1217 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1218 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1219 PL_lex_inpat = PL_sublex_info.sub_op;
1221 PL_lex_inpat = Nullop;
1228 * Restores lexer state after a S_sublex_push.
1235 if (!PL_lex_starts++) {
1236 SV * const sv = newSVpvn("",0);
1237 if (SvUTF8(PL_linestr))
1239 PL_expect = XOPERATOR;
1240 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1244 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1245 PL_lex_state = LEX_INTERPCASEMOD;
1249 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1250 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1251 PL_linestr = PL_lex_repl;
1253 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1254 PL_bufend += SvCUR(PL_linestr);
1255 PL_last_lop = PL_last_uni = Nullch;
1256 SAVEFREESV(PL_linestr);
1257 PL_lex_dojoin = FALSE;
1258 PL_lex_brackets = 0;
1259 PL_lex_casemods = 0;
1260 *PL_lex_casestack = '\0';
1262 if (SvEVALED(PL_lex_repl)) {
1263 PL_lex_state = LEX_INTERPNORMAL;
1265 /* we don't clear PL_lex_repl here, so that we can check later
1266 whether this is an evalled subst; that means we rely on the
1267 logic to ensure sublex_done() is called again only via the
1268 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1271 PL_lex_state = LEX_INTERPCONCAT;
1272 PL_lex_repl = Nullsv;
1278 PL_bufend = SvPVX(PL_linestr);
1279 PL_bufend += SvCUR(PL_linestr);
1280 PL_expect = XOPERATOR;
1281 PL_sublex_info.sub_inwhat = 0;
1289 Extracts a pattern, double-quoted string, or transliteration. This
1292 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1293 processing a pattern (PL_lex_inpat is true), a transliteration
1294 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1296 Returns a pointer to the character scanned up to. Iff this is
1297 advanced from the start pointer supplied (ie if anything was
1298 successfully parsed), will leave an OP for the substring scanned
1299 in yylval. Caller must intuit reason for not parsing further
1300 by looking at the next characters herself.
1304 double-quoted style: \r and \n
1305 regexp special ones: \D \s
1307 backrefs: \1 (deprecated in substitution replacements)
1308 case and quoting: \U \Q \E
1309 stops on @ and $, but not for $ as tail anchor
1311 In transliterations:
1312 characters are VERY literal, except for - not at the start or end
1313 of the string, which indicates a range. scan_const expands the
1314 range to the full set of intermediate characters.
1316 In double-quoted strings:
1318 double-quoted style: \r and \n
1320 backrefs: \1 (deprecated)
1321 case and quoting: \U \Q \E
1324 scan_const does *not* construct ops to handle interpolated strings.
1325 It stops processing as soon as it finds an embedded $ or @ variable
1326 and leaves it to the caller to work out what's going on.
1328 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @::foo.
1330 $ in pattern could be $foo or could be tail anchor. Assumption:
1331 it's a tail anchor if $ is the last thing in the string, or if it's
1332 followed by one of ")| \n\t"
1334 \1 (backreferences) are turned into $1
1336 The structure of the code is
1337 while (there's a character to process) {
1338 handle transliteration ranges
1339 skip regexp comments
1340 skip # initiated comments in //x patterns
1341 check for embedded @foo
1342 check for embedded scalars
1344 leave intact backslashes from leave (below)
1345 deprecate \1 in strings and sub replacements
1346 handle string-changing backslashes \l \U \Q \E, etc.
1347 switch (what was escaped) {
1348 handle - in a transliteration (becomes a literal -)
1349 handle \132 octal characters
1350 handle 0x15 hex characters
1351 handle \cV (control V)
1352 handle printf backslashes (\f, \r, \n, etc)
1354 } (end if backslash)
1355 } (end while character to read)
1360 S_scan_const(pTHX_ char *start)
1362 register char *send = PL_bufend; /* end of the constant */
1363 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1364 register char *s = start; /* start of the constant */
1365 register char *d = SvPVX(sv); /* destination for copies */
1366 bool dorange = FALSE; /* are we in a translit range? */
1367 bool didrange = FALSE; /* did we just finish a range? */
1368 I32 has_utf8 = FALSE; /* Output constant is UTF8 */
1369 I32 this_utf8 = UTF; /* The source string is assumed to be UTF8 */
1372 UV literal_endpoint = 0;
1375 const char *leaveit = /* set of acceptably-backslashed characters */
1377 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxz0123456789[{]} \t\n\r\f\v#"
1380 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1381 /* If we are doing a trans and we know we want UTF8 set expectation */
1382 has_utf8 = PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF);
1383 this_utf8 = PL_sublex_info.sub_op->op_private & (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1387 while (s < send || dorange) {
1388 /* get transliterations out of the way (they're most literal) */
1389 if (PL_lex_inwhat == OP_TRANS) {
1390 /* expand a range A-Z to the full set of characters. AIE! */
1392 I32 i; /* current expanded character */
1393 I32 min; /* first character in range */
1394 I32 max; /* last character in range */
1397 char * const c = (char*)utf8_hop((U8*)d, -1);
1401 *c = (char)UTF_TO_NATIVE(0xff);
1402 /* mark the range as done, and continue */
1408 i = d - SvPVX_const(sv); /* remember current offset */
1409 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1410 d = SvPVX(sv) + i; /* refresh d after realloc */
1411 d -= 2; /* eat the first char and the - */
1413 min = (U8)*d; /* first char in range */
1414 max = (U8)d[1]; /* last char in range */
1418 "Invalid range \"%c-%c\" in transliteration operator",
1419 (char)min, (char)max);
1423 if (literal_endpoint == 2 &&
1424 ((isLOWER(min) && isLOWER(max)) ||
1425 (isUPPER(min) && isUPPER(max)))) {
1427 for (i = min; i <= max; i++)
1429 *d++ = NATIVE_TO_NEED(has_utf8,i);
1431 for (i = min; i <= max; i++)
1433 *d++ = NATIVE_TO_NEED(has_utf8,i);
1438 for (i = min; i <= max; i++)
1441 /* mark the range as done, and continue */
1445 literal_endpoint = 0;
1450 /* range begins (ignore - as first or last char) */
1451 else if (*s == '-' && s+1 < send && s != start) {
1453 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
1456 *d++ = (char)UTF_TO_NATIVE(0xff); /* use illegal utf8 byte--see pmtrans */
1466 literal_endpoint = 0;
1471 /* if we get here, we're not doing a transliteration */
1473 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1474 except for the last char, which will be done separately. */
1475 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1477 while (s+1 < send && *s != ')')
1478 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1480 else if (s[2] == '{' /* This should match regcomp.c */
1481 || ((s[2] == 'p' || s[2] == '?') && s[3] == '{'))
1484 char *regparse = s + (s[2] == '{' ? 3 : 4);
1487 while (count && (c = *regparse)) {
1488 if (c == '\\' && regparse[1])
1496 if (*regparse != ')')
1497 regparse--; /* Leave one char for continuation. */
1498 while (s < regparse)
1499 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1503 /* likewise skip #-initiated comments in //x patterns */
1504 else if (*s == '#' && PL_lex_inpat &&
1505 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1506 while (s+1 < send && *s != '\n')
1507 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1510 /* check for embedded arrays
1511 (@foo, @::foo, @'foo, @{foo}, @$foo, @+, @-)
1513 else if (*s == '@' && s[1]
1514 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$+-", s[1])))
1517 /* check for embedded scalars. only stop if we're sure it's a
1520 else if (*s == '$') {
1521 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1523 if (s + 1 < send && !strchr("()| \r\n\t", s[1]))
1524 break; /* in regexp, $ might be tail anchor */
1527 /* End of else if chain - OP_TRANS rejoin rest */
1530 if (*s == '\\' && s+1 < send) {
1533 /* some backslashes we leave behind */
1534 if (*leaveit && *s && strchr(leaveit, *s)) {
1535 *d++ = NATIVE_TO_NEED(has_utf8,'\\');
1536 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1540 /* deprecate \1 in strings and substitution replacements */
1541 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1542 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1544 if (ckWARN(WARN_SYNTAX))
1545 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "\\%c better written as $%c", *s, *s);
1550 /* string-change backslash escapes */
1551 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1556 /* if we get here, it's either a quoted -, or a digit */
1559 /* quoted - in transliterations */
1561 if (PL_lex_inwhat == OP_TRANS) {
1571 Perl_warner(aTHX_ packWARN(WARN_MISC),
1572 "Unrecognized escape \\%c passed through",
1574 /* default action is to copy the quoted character */
1575 goto default_action;
1578 /* \132 indicates an octal constant */
1579 case '0': case '1': case '2': case '3':
1580 case '4': case '5': case '6': case '7':
1584 uv = grok_oct(s, &len, &flags, NULL);
1587 goto NUM_ESCAPE_INSERT;
1589 /* \x24 indicates a hex constant */
1593 char* const e = strchr(s, '}');
1594 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
1595 PERL_SCAN_DISALLOW_PREFIX;
1600 yyerror("Missing right brace on \\x{}");
1604 uv = grok_hex(s, &len, &flags, NULL);
1610 I32 flags = PERL_SCAN_DISALLOW_PREFIX;
1611 uv = grok_hex(s, &len, &flags, NULL);
1617 /* Insert oct or hex escaped character.
1618 * There will always enough room in sv since such
1619 * escapes will be longer than any UTF-8 sequence
1620 * they can end up as. */
1622 /* We need to map to chars to ASCII before doing the tests
1625 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(uv))) {
1626 if (!has_utf8 && uv > 255) {
1627 /* Might need to recode whatever we have
1628 * accumulated so far if it contains any
1631 * (Can't we keep track of that and avoid
1632 * this rescan? --jhi)
1636 for (c = (U8 *) SvPVX(sv); c < (U8 *)d; c++) {
1637 if (!NATIVE_IS_INVARIANT(*c)) {
1642 const STRLEN offset = d - SvPVX_const(sv);
1644 d = SvGROW(sv, SvLEN(sv) + hicount + 1) + offset;
1648 while (src >= (const U8 *)SvPVX_const(sv)) {
1649 if (!NATIVE_IS_INVARIANT(*src)) {
1650 const U8 ch = NATIVE_TO_ASCII(*src);
1651 *dst-- = (U8)UTF8_EIGHT_BIT_LO(ch);
1652 *dst-- = (U8)UTF8_EIGHT_BIT_HI(ch);
1662 if (has_utf8 || uv > 255) {
1663 d = (char*)uvchr_to_utf8((U8*)d, uv);
1665 if (PL_lex_inwhat == OP_TRANS &&
1666 PL_sublex_info.sub_op) {
1667 PL_sublex_info.sub_op->op_private |=
1668 (PL_lex_repl ? OPpTRANS_FROM_UTF
1681 /* \N{LATIN SMALL LETTER A} is a named character */
1685 char* e = strchr(s, '}');
1691 yyerror("Missing right brace on \\N{}");
1695 if (e > s + 2 && s[1] == 'U' && s[2] == '+') {
1697 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
1698 PERL_SCAN_DISALLOW_PREFIX;
1701 uv = grok_hex(s, &len, &flags, NULL);
1703 goto NUM_ESCAPE_INSERT;
1705 res = newSVpvn(s + 1, e - s - 1);
1706 res = new_constant( Nullch, 0, "charnames",
1707 res, Nullsv, "\\N{...}" );
1709 sv_utf8_upgrade(res);
1710 str = SvPV_const(res,len);
1711 #ifdef EBCDIC_NEVER_MIND
1712 /* charnames uses pack U and that has been
1713 * recently changed to do the below uni->native
1714 * mapping, so this would be redundant (and wrong,
1715 * the code point would be doubly converted).
1716 * But leave this in just in case the pack U change
1717 * gets revoked, but the semantics is still
1718 * desireable for charnames. --jhi */
1720 UV uv = utf8_to_uvchr((const U8*)str, 0);
1723 U8 tmpbuf[UTF8_MAXBYTES+1], *d;
1725 d = uvchr_to_utf8(tmpbuf, UNI_TO_NATIVE(uv));
1726 sv_setpvn(res, (char *)tmpbuf, d - tmpbuf);
1727 str = SvPV_const(res, len);
1731 if (!has_utf8 && SvUTF8(res)) {
1732 const char * const ostart = SvPVX_const(sv);
1733 SvCUR_set(sv, d - ostart);
1736 sv_utf8_upgrade(sv);
1737 /* this just broke our allocation above... */
1738 SvGROW(sv, (STRLEN)(send - start));
1739 d = SvPVX(sv) + SvCUR(sv);
1742 if (len > (STRLEN)(e - s + 4)) { /* I _guess_ 4 is \N{} --jhi */
1743 const char * const odest = SvPVX_const(sv);
1745 SvGROW(sv, (SvLEN(sv) + len - (e - s + 4)));
1746 d = SvPVX(sv) + (d - odest);
1748 Copy(str, d, len, char);
1755 yyerror("Missing braces on \\N{}");
1758 /* \c is a control character */
1767 *d++ = NATIVE_TO_NEED(has_utf8,toCTRL(c));
1770 yyerror("Missing control char name in \\c");
1774 /* printf-style backslashes, formfeeds, newlines, etc */
1776 *d++ = NATIVE_TO_NEED(has_utf8,'\b');
1779 *d++ = NATIVE_TO_NEED(has_utf8,'\n');
1782 *d++ = NATIVE_TO_NEED(has_utf8,'\r');
1785 *d++ = NATIVE_TO_NEED(has_utf8,'\f');
1788 *d++ = NATIVE_TO_NEED(has_utf8,'\t');
1791 *d++ = ASCII_TO_NEED(has_utf8,'\033');
1794 *d++ = ASCII_TO_NEED(has_utf8,'\007');
1800 } /* end if (backslash) */
1807 /* If we started with encoded form, or already know we want it
1808 and then encode the next character */
1809 if ((has_utf8 || this_utf8) && !NATIVE_IS_INVARIANT((U8)(*s))) {
1811 const UV uv = (this_utf8) ? utf8n_to_uvchr((U8*)s, send - s, &len, 0) : (UV) ((U8) *s);
1812 const STRLEN need = UNISKIP(NATIVE_TO_UNI(uv));
1815 /* encoded value larger than old, need extra space (NOTE: SvCUR() not set here) */
1816 const STRLEN off = d - SvPVX_const(sv);
1817 d = SvGROW(sv, SvLEN(sv) + (need-len)) + off;
1819 d = (char*)uvchr_to_utf8((U8*)d, uv);
1823 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1825 } /* while loop to process each character */
1827 /* terminate the string and set up the sv */
1829 SvCUR_set(sv, d - SvPVX_const(sv));
1830 if (SvCUR(sv) >= SvLEN(sv))
1831 Perl_croak(aTHX_ "panic: constant overflowed allocated space");
1834 if (PL_encoding && !has_utf8) {
1835 sv_recode_to_utf8(sv, PL_encoding);
1841 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1842 PL_sublex_info.sub_op->op_private |=
1843 (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1847 /* shrink the sv if we allocated more than we used */
1848 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1849 SvPV_shrink_to_cur(sv);
1852 /* return the substring (via yylval) only if we parsed anything */
1853 if (s > PL_bufptr) {
1854 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1855 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1857 ( PL_lex_inwhat == OP_TRANS
1859 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1862 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1869 * Returns TRUE if there's more to the expression (e.g., a subscript),
1872 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1874 * ->[ and ->{ return TRUE
1875 * { and [ outside a pattern are always subscripts, so return TRUE
1876 * if we're outside a pattern and it's not { or [, then return FALSE
1877 * if we're in a pattern and the first char is a {
1878 * {4,5} (any digits around the comma) returns FALSE
1879 * if we're in a pattern and the first char is a [
1881 * [SOMETHING] has a funky algorithm to decide whether it's a
1882 * character class or not. It has to deal with things like
1883 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1884 * anything else returns TRUE
1887 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1890 S_intuit_more(pTHX_ register char *s)
1892 if (PL_lex_brackets)
1894 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1896 if (*s != '{' && *s != '[')
1901 /* In a pattern, so maybe we have {n,m}. */
1918 /* On the other hand, maybe we have a character class */
1921 if (*s == ']' || *s == '^')
1924 /* this is terrifying, and it works */
1925 int weight = 2; /* let's weigh the evidence */
1927 unsigned char un_char = 255, last_un_char;
1928 const char * const send = strchr(s,']');
1929 char tmpbuf[sizeof PL_tokenbuf * 4];
1931 if (!send) /* has to be an expression */
1934 Zero(seen,256,char);
1937 else if (isDIGIT(*s)) {
1939 if (isDIGIT(s[1]) && s[2] == ']')
1945 for (; s < send; s++) {
1946 last_un_char = un_char;
1947 un_char = (unsigned char)*s;
1952 weight -= seen[un_char] * 10;
1953 if (isALNUM_lazy_if(s+1,UTF)) {
1954 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1955 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1960 else if (*s == '$' && s[1] &&
1961 strchr("[#!%*<>()-=",s[1])) {
1962 if (/*{*/ strchr("])} =",s[2]))
1971 if (strchr("wds]",s[1]))
1973 else if (seen['\''] || seen['"'])
1975 else if (strchr("rnftbxcav",s[1]))
1977 else if (isDIGIT(s[1])) {
1979 while (s[1] && isDIGIT(s[1]))
1989 if (strchr("aA01! ",last_un_char))
1991 if (strchr("zZ79~",s[1]))
1993 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1994 weight -= 5; /* cope with negative subscript */
1997 if (!isALNUM(last_un_char)
1998 && !(last_un_char == '$' || last_un_char == '@'
1999 || last_un_char == '&')
2000 && isALPHA(*s) && s[1] && isALPHA(s[1])) {
2005 if (keyword(tmpbuf, d - tmpbuf))
2008 if (un_char == last_un_char + 1)
2010 weight -= seen[un_char];
2015 if (weight >= 0) /* probably a character class */
2025 * Does all the checking to disambiguate
2027 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
2028 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
2030 * First argument is the stuff after the first token, e.g. "bar".
2032 * Not a method if bar is a filehandle.
2033 * Not a method if foo is a subroutine prototyped to take a filehandle.
2034 * Not a method if it's really "Foo $bar"
2035 * Method if it's "foo $bar"
2036 * Not a method if it's really "print foo $bar"
2037 * Method if it's really "foo package::" (interpreted as package->foo)
2038 * Not a method if bar is known to be a subroutine ("sub bar; foo bar")
2039 * Not a method if bar is a filehandle or package, but is quoted with
2044 S_intuit_method(pTHX_ char *start, GV *gv)
2046 char *s = start + (*start == '$');
2047 char tmpbuf[sizeof PL_tokenbuf];
2055 if ((cv = GvCVu(gv))) {
2056 const char *proto = SvPVX_const(cv);
2066 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
2067 /* start is the beginning of the possible filehandle/object,
2068 * and s is the end of it
2069 * tmpbuf is a copy of it
2072 if (*start == '$') {
2073 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
2078 return *s == '(' ? FUNCMETH : METHOD;
2080 if (!keyword(tmpbuf, len)) {
2081 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
2086 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
2087 if (indirgv && GvCVu(indirgv))
2089 /* filehandle or package name makes it a method */
2090 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
2092 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
2093 return 0; /* no assumptions -- "=>" quotes bearword */
2095 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
2096 newSVpvn(tmpbuf,len));
2097 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
2101 return *s == '(' ? FUNCMETH : METHOD;
2109 * Return a string of Perl code to load the debugger. If PERL5DB
2110 * is set, it will return the contents of that, otherwise a
2111 * compile-time require of perl5db.pl.
2118 const char * const pdb = PerlEnv_getenv("PERL5DB");
2122 SETERRNO(0,SS_NORMAL);
2123 return "BEGIN { require 'perl5db.pl' }";
2129 /* Encoded script support. filter_add() effectively inserts a
2130 * 'pre-processing' function into the current source input stream.
2131 * Note that the filter function only applies to the current source file
2132 * (e.g., it will not affect files 'require'd or 'use'd by this one).
2134 * The datasv parameter (which may be NULL) can be used to pass
2135 * private data to this instance of the filter. The filter function
2136 * can recover the SV using the FILTER_DATA macro and use it to
2137 * store private buffers and state information.
2139 * The supplied datasv parameter is upgraded to a PVIO type
2140 * and the IoDIRP/IoANY field is used to store the function pointer,
2141 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
2142 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
2143 * private use must be set using malloc'd pointers.
2147 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
2152 if (!PL_rsfp_filters)
2153 PL_rsfp_filters = newAV();
2155 datasv = NEWSV(255,0);
2156 SvUPGRADE(datasv, SVt_PVIO);
2157 IoANY(datasv) = FPTR2DPTR(void *, funcp); /* stash funcp into spare field */
2158 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
2159 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
2160 IoANY(datasv), SvPV_nolen(datasv)));
2161 av_unshift(PL_rsfp_filters, 1);
2162 av_store(PL_rsfp_filters, 0, datasv) ;
2167 /* Delete most recently added instance of this filter function. */
2169 Perl_filter_del(pTHX_ filter_t funcp)
2174 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", FPTR2DPTR(XPVIO *, funcp)));
2176 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
2178 /* if filter is on top of stack (usual case) just pop it off */
2179 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
2180 if (IoANY(datasv) == FPTR2DPTR(void *, funcp)) {
2181 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
2182 IoANY(datasv) = (void *)NULL;
2183 sv_free(av_pop(PL_rsfp_filters));
2187 /* we need to search for the correct entry and clear it */
2188 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
2192 /* Invoke the idxth filter function for the current rsfp. */
2193 /* maxlen 0 = read one text line */
2195 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
2200 if (!PL_rsfp_filters)
2202 if (idx > AvFILLp(PL_rsfp_filters)) { /* Any more filters? */
2203 /* Provide a default input filter to make life easy. */
2204 /* Note that we append to the line. This is handy. */
2205 DEBUG_P(PerlIO_printf(Perl_debug_log,
2206 "filter_read %d: from rsfp\n", idx));
2210 const int old_len = SvCUR(buf_sv);
2212 /* ensure buf_sv is large enough */
2213 SvGROW(buf_sv, (STRLEN)(old_len + maxlen)) ;
2214 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
2215 if (PerlIO_error(PL_rsfp))
2216 return -1; /* error */
2218 return 0 ; /* end of file */
2220 SvCUR_set(buf_sv, old_len + len) ;
2223 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
2224 if (PerlIO_error(PL_rsfp))
2225 return -1; /* error */
2227 return 0 ; /* end of file */
2230 return SvCUR(buf_sv);
2232 /* Skip this filter slot if filter has been deleted */
2233 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef) {
2234 DEBUG_P(PerlIO_printf(Perl_debug_log,
2235 "filter_read %d: skipped (filter deleted)\n",
2237 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
2239 /* Get function pointer hidden within datasv */
2240 funcp = DPTR2FPTR(filter_t, IoANY(datasv));
2241 DEBUG_P(PerlIO_printf(Perl_debug_log,
2242 "filter_read %d: via function %p (%s)\n",
2243 idx, datasv, SvPV_nolen_const(datasv)));
2244 /* Call function. The function is expected to */
2245 /* call "FILTER_READ(idx+1, buf_sv)" first. */
2246 /* Return: <0:error, =0:eof, >0:not eof */
2247 return (*funcp)(aTHX_ idx, buf_sv, maxlen);
2251 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
2253 #ifdef PERL_CR_FILTER
2254 if (!PL_rsfp_filters) {
2255 filter_add(S_cr_textfilter,NULL);
2258 if (PL_rsfp_filters) {
2260 SvCUR_set(sv, 0); /* start with empty line */
2261 if (FILTER_READ(0, sv, 0) > 0)
2262 return ( SvPVX(sv) ) ;
2267 return (sv_gets(sv, fp, append));
2271 S_find_in_my_stash(pTHX_ const char *pkgname, I32 len)
2275 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2279 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2280 (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV)))
2282 return GvHV(gv); /* Foo:: */
2285 /* use constant CLASS => 'MyClass' */
2286 if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) {
2288 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2289 pkgname = SvPV_nolen_const(sv);
2293 return gv_stashpv(pkgname, FALSE);
2297 S_tokenize_use(int is_use, char *s) {
2298 if (PL_expect != XSTATE)
2299 yyerror(Perl_form(aTHX_ "\"%s\" not allowed in expression",
2300 is_use ? "use" : "no"));
2302 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
2303 s = force_version(s, TRUE);
2304 if (*s == ';' || (s = skipspace(s), *s == ';')) {
2305 PL_nextval[PL_nexttoke].opval = Nullop;
2308 else if (*s == 'v') {
2309 s = force_word(s,WORD,FALSE,TRUE,FALSE);
2310 s = force_version(s, FALSE);
2314 s = force_word(s,WORD,FALSE,TRUE,FALSE);
2315 s = force_version(s, FALSE);
2317 yylval.ival = is_use;
2321 static const char* const exp_name[] =
2322 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2323 "ATTRTERM", "TERMBLOCK", "TERMORDORDOR"
2330 Works out what to call the token just pulled out of the input
2331 stream. The yacc parser takes care of taking the ops we return and
2332 stitching them into a tree.
2338 if read an identifier
2339 if we're in a my declaration
2340 croak if they tried to say my($foo::bar)
2341 build the ops for a my() declaration
2342 if it's an access to a my() variable
2343 are we in a sort block?
2344 croak if my($a); $a <=> $b
2345 build ops for access to a my() variable
2346 if in a dq string, and they've said @foo and we can't find @foo
2348 build ops for a bareword
2349 if we already built the token before, use it.
2354 #pragma segment Perl_yylex
2359 register char *s = PL_bufptr;
2366 I32 orig_keyword = 0;
2369 PerlIO_printf(Perl_debug_log, "### LEX_%s\n",
2370 lex_state_names[PL_lex_state]);
2372 /* check if there's an identifier for us to look at */
2373 if (PL_pending_ident)
2374 return REPORT(S_pending_ident(aTHX));
2376 /* no identifier pending identification */
2378 switch (PL_lex_state) {
2380 case LEX_NORMAL: /* Some compilers will produce faster */
2381 case LEX_INTERPNORMAL: /* code if we comment these out. */
2385 /* when we've already built the next token, just pull it out of the queue */
2388 yylval = PL_nextval[PL_nexttoke];
2390 PL_lex_state = PL_lex_defer;
2391 PL_expect = PL_lex_expect;
2392 PL_lex_defer = LEX_NORMAL;
2394 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2395 "### Next token after '%s' was known, type %"IVdf"\n", PL_bufptr,
2396 (IV)PL_nexttype[PL_nexttoke]); });
2398 return REPORT(PL_nexttype[PL_nexttoke]);
2400 /* interpolated case modifiers like \L \U, including \Q and \E.
2401 when we get here, PL_bufptr is at the \
2403 case LEX_INTERPCASEMOD:
2405 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2406 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2408 /* handle \E or end of string */
2409 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2411 if (PL_lex_casemods) {
2412 const char oldmod = PL_lex_casestack[--PL_lex_casemods];
2413 PL_lex_casestack[PL_lex_casemods] = '\0';
2415 if (PL_bufptr != PL_bufend
2416 && (oldmod == 'L' || oldmod == 'U' || oldmod == 'Q')) {
2418 PL_lex_state = LEX_INTERPCONCAT;
2422 if (PL_bufptr != PL_bufend)
2424 PL_lex_state = LEX_INTERPCONCAT;
2428 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2429 "### Saw case modifier at '%s'\n", PL_bufptr); });
2431 if (s[1] == '\\' && s[2] == 'E') {
2433 PL_lex_state = LEX_INTERPCONCAT;
2437 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2438 tmp = *s, *s = s[2], s[2] = (char)tmp; /* misordered... */
2439 if ((*s == 'L' || *s == 'U') &&
2440 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U'))) {
2441 PL_lex_casestack[--PL_lex_casemods] = '\0';
2444 if (PL_lex_casemods > 10)
2445 Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2446 PL_lex_casestack[PL_lex_casemods++] = *s;
2447 PL_lex_casestack[PL_lex_casemods] = '\0';
2448 PL_lex_state = LEX_INTERPCONCAT;
2449 PL_nextval[PL_nexttoke].ival = 0;
2452 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2454 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2456 PL_nextval[PL_nexttoke].ival = OP_LC;
2458 PL_nextval[PL_nexttoke].ival = OP_UC;
2460 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2462 Perl_croak(aTHX_ "panic: yylex");
2466 if (PL_lex_starts) {
2469 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
2470 if (PL_lex_casemods == 1 && PL_lex_inpat)
2479 case LEX_INTERPPUSH:
2480 return REPORT(sublex_push());
2482 case LEX_INTERPSTART:
2483 if (PL_bufptr == PL_bufend)
2484 return REPORT(sublex_done());
2485 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2486 "### Interpolated variable at '%s'\n", PL_bufptr); });
2488 PL_lex_dojoin = (*PL_bufptr == '@');
2489 PL_lex_state = LEX_INTERPNORMAL;
2490 if (PL_lex_dojoin) {
2491 PL_nextval[PL_nexttoke].ival = 0;
2493 force_ident("\"", '$');
2494 PL_nextval[PL_nexttoke].ival = 0;
2496 PL_nextval[PL_nexttoke].ival = 0;
2498 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2501 if (PL_lex_starts++) {
2503 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
2504 if (!PL_lex_casemods && PL_lex_inpat)
2511 case LEX_INTERPENDMAYBE:
2512 if (intuit_more(PL_bufptr)) {
2513 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2519 if (PL_lex_dojoin) {
2520 PL_lex_dojoin = FALSE;
2521 PL_lex_state = LEX_INTERPCONCAT;
2524 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2525 && SvEVALED(PL_lex_repl))
2527 if (PL_bufptr != PL_bufend)
2528 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2529 PL_lex_repl = Nullsv;
2532 case LEX_INTERPCONCAT:
2534 if (PL_lex_brackets)
2535 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2537 if (PL_bufptr == PL_bufend)
2538 return REPORT(sublex_done());
2540 if (SvIVX(PL_linestr) == '\'') {
2541 SV *sv = newSVsv(PL_linestr);
2544 else if ( PL_hints & HINT_NEW_RE )
2545 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2546 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2550 s = scan_const(PL_bufptr);
2552 PL_lex_state = LEX_INTERPCASEMOD;
2554 PL_lex_state = LEX_INTERPSTART;
2557 if (s != PL_bufptr) {
2558 PL_nextval[PL_nexttoke] = yylval;
2561 if (PL_lex_starts++) {
2562 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
2563 if (!PL_lex_casemods && PL_lex_inpat)
2576 PL_lex_state = LEX_NORMAL;
2577 s = scan_formline(PL_bufptr);
2578 if (!PL_lex_formbrack)
2584 PL_oldoldbufptr = PL_oldbufptr;
2587 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at [%s]\n",
2588 exp_name[PL_expect], s);
2594 if (isIDFIRST_lazy_if(s,UTF))
2596 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2599 goto fake_eof; /* emulate EOF on ^D or ^Z */
2604 if (PL_lex_brackets) {
2605 if (PL_lex_formbrack)
2606 yyerror("Format not terminated");
2608 yyerror("Missing right curly or square bracket");
2610 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2611 "### Tokener got EOF\n");
2615 if (s++ < PL_bufend)
2616 goto retry; /* ignore stray nulls */
2619 if (!PL_in_eval && !PL_preambled) {
2620 PL_preambled = TRUE;
2621 sv_setpv(PL_linestr,incl_perldb());
2622 if (SvCUR(PL_linestr))
2623 sv_catpvn(PL_linestr,";", 1);
2625 while(AvFILLp(PL_preambleav) >= 0) {
2626 SV *tmpsv = av_shift(PL_preambleav);
2627 sv_catsv(PL_linestr, tmpsv);
2628 sv_catpvn(PL_linestr, ";", 1);
2631 sv_free((SV*)PL_preambleav);
2632 PL_preambleav = NULL;
2634 if (PL_minus_n || PL_minus_p) {
2635 sv_catpv(PL_linestr, "LINE: while (<>) {");
2637 sv_catpv(PL_linestr,"chomp;");
2640 if ((*PL_splitstr == '/' || *PL_splitstr == '\''
2641 || *PL_splitstr == '"')
2642 && strchr(PL_splitstr + 1, *PL_splitstr))
2643 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s);", PL_splitstr);
2645 /* "q\0${splitstr}\0" is legal perl. Yes, even NUL
2646 bytes can be used as quoting characters. :-) */
2647 /* The count here deliberately includes the NUL
2648 that terminates the C string constant. This
2649 embeds the opening NUL into the string. */
2650 const char *splits = PL_splitstr;
2651 sv_catpvn(PL_linestr, "our @F=split(q", 15);
2654 if (*splits == '\\')
2655 sv_catpvn(PL_linestr, splits, 1);
2656 sv_catpvn(PL_linestr, splits, 1);
2657 } while (*splits++);
2658 /* This loop will embed the trailing NUL of
2659 PL_linestr as the last thing it does before
2661 sv_catpvn(PL_linestr, ");", 2);
2665 sv_catpv(PL_linestr,"our @F=split(' ');");
2668 sv_catpvn(PL_linestr, "\n", 1);
2669 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2670 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2671 PL_last_lop = PL_last_uni = Nullch;
2672 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2673 SV * const sv = NEWSV(85,0);
2675 sv_upgrade(sv, SVt_PVMG);
2676 sv_setsv(sv,PL_linestr);
2679 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2684 bof = PL_rsfp ? TRUE : FALSE;
2685 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2688 if (PL_preprocess && !PL_in_eval)
2689 (void)PerlProc_pclose(PL_rsfp);
2690 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2691 PerlIO_clearerr(PL_rsfp);
2693 (void)PerlIO_close(PL_rsfp);
2695 PL_doextract = FALSE;
2697 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2698 sv_setpv(PL_linestr,PL_minus_p
2699 ? ";}continue{print;}" : ";}");
2700 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2701 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2702 PL_last_lop = PL_last_uni = Nullch;
2703 PL_minus_n = PL_minus_p = 0;
2706 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2707 PL_last_lop = PL_last_uni = Nullch;
2708 sv_setpvn(PL_linestr,"",0);
2709 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2711 /* If it looks like the start of a BOM or raw UTF-16,
2712 * check if it in fact is. */
2718 #ifdef PERLIO_IS_STDIO
2719 # ifdef __GNU_LIBRARY__
2720 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
2721 # define FTELL_FOR_PIPE_IS_BROKEN
2725 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
2726 # define FTELL_FOR_PIPE_IS_BROKEN
2731 #ifdef FTELL_FOR_PIPE_IS_BROKEN
2732 /* This loses the possibility to detect the bof
2733 * situation on perl -P when the libc5 is being used.
2734 * Workaround? Maybe attach some extra state to PL_rsfp?
2737 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
2739 bof = PerlIO_tell(PL_rsfp) == (Off_t)SvCUR(PL_linestr);
2742 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2743 s = swallow_bom((U8*)s);
2747 /* Incest with pod. */
2748 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2749 sv_setpvn(PL_linestr, "", 0);
2750 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2751 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2752 PL_last_lop = PL_last_uni = Nullch;
2753 PL_doextract = FALSE;
2757 } while (PL_doextract);
2758 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2759 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2760 SV * const sv = NEWSV(85,0);
2762 sv_upgrade(sv, SVt_PVMG);
2763 sv_setsv(sv,PL_linestr);
2766 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2768 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2769 PL_last_lop = PL_last_uni = Nullch;
2770 if (CopLINE(PL_curcop) == 1) {
2771 while (s < PL_bufend && isSPACE(*s))
2773 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2777 if (*s == '#' && *(s+1) == '!')
2779 #ifdef ALTERNATE_SHEBANG
2781 static char const as[] = ALTERNATE_SHEBANG;
2782 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2783 d = s + (sizeof(as) - 1);
2785 #endif /* ALTERNATE_SHEBANG */
2794 while (*d && !isSPACE(*d))
2798 #ifdef ARG_ZERO_IS_SCRIPT
2799 if (ipathend > ipath) {
2801 * HP-UX (at least) sets argv[0] to the script name,
2802 * which makes $^X incorrect. And Digital UNIX and Linux,
2803 * at least, set argv[0] to the basename of the Perl
2804 * interpreter. So, having found "#!", we'll set it right.
2806 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV)); /* $^X */
2807 assert(SvPOK(x) || SvGMAGICAL(x));
2808 if (sv_eq(x, CopFILESV(PL_curcop))) {
2809 sv_setpvn(x, ipath, ipathend - ipath);
2815 const char *bstart = SvPV_const(CopFILESV(PL_curcop),blen);
2816 const char * const lstart = SvPV_const(x,llen);
2818 bstart += blen - llen;
2819 if (strnEQ(bstart, lstart, llen) && bstart[-1] == '/') {
2820 sv_setpvn(x, ipath, ipathend - ipath);
2825 TAINT_NOT; /* $^X is always tainted, but that's OK */
2827 #endif /* ARG_ZERO_IS_SCRIPT */
2832 d = instr(s,"perl -");
2834 d = instr(s,"perl");
2836 /* avoid getting into infinite loops when shebang
2837 * line contains "Perl" rather than "perl" */
2839 for (d = ipathend-4; d >= ipath; --d) {
2840 if ((*d == 'p' || *d == 'P')
2841 && !ibcmp(d, "perl", 4))
2851 #ifdef ALTERNATE_SHEBANG
2853 * If the ALTERNATE_SHEBANG on this system starts with a
2854 * character that can be part of a Perl expression, then if
2855 * we see it but not "perl", we're probably looking at the
2856 * start of Perl code, not a request to hand off to some
2857 * other interpreter. Similarly, if "perl" is there, but
2858 * not in the first 'word' of the line, we assume the line
2859 * contains the start of the Perl program.
2861 if (d && *s != '#') {
2862 const char *c = ipath;
2863 while (*c && !strchr("; \t\r\n\f\v#", *c))
2866 d = Nullch; /* "perl" not in first word; ignore */
2868 *s = '#'; /* Don't try to parse shebang line */
2870 #endif /* ALTERNATE_SHEBANG */
2871 #ifndef MACOS_TRADITIONAL
2876 !instr(s,"indir") &&
2877 instr(PL_origargv[0],"perl"))
2884 while (s < PL_bufend && isSPACE(*s))
2886 if (s < PL_bufend) {
2887 Newxz(newargv,PL_origargc+3,char*);
2889 while (s < PL_bufend && !isSPACE(*s))
2892 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2895 newargv = PL_origargv;
2898 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
2900 Perl_croak(aTHX_ "Can't exec %s", ipath);
2904 const U32 oldpdb = PL_perldb;
2905 const bool oldn = PL_minus_n;
2906 const bool oldp = PL_minus_p;
2908 while (*d && !isSPACE(*d)) d++;
2909 while (SPACE_OR_TAB(*d)) d++;
2912 const bool switches_done = PL_doswitches;
2914 if (*d == 'M' || *d == 'm' || *d == 'C') {
2915 const char * const m = d;
2916 while (*d && !isSPACE(*d)) d++;
2917 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2920 d = moreswitches(d);
2922 if (PL_doswitches && !switches_done) {
2923 int argc = PL_origargc;
2924 char **argv = PL_origargv;
2927 } while (argc && argv[0][0] == '-' && argv[0][1]);
2928 init_argv_symbols(argc,argv);
2930 if ((PERLDB_LINE && !oldpdb) ||
2931 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2932 /* if we have already added "LINE: while (<>) {",
2933 we must not do it again */
2935 sv_setpvn(PL_linestr, "", 0);
2936 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2937 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2938 PL_last_lop = PL_last_uni = Nullch;
2939 PL_preambled = FALSE;
2941 (void)gv_fetchfile(PL_origfilename);
2944 if (PL_doswitches && !switches_done) {
2945 int argc = PL_origargc;
2946 char **argv = PL_origargv;
2949 } while (argc && argv[0][0] == '-' && argv[0][1]);
2950 init_argv_symbols(argc,argv);
2956 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2958 PL_lex_state = LEX_FORMLINE;
2963 #ifdef PERL_STRICT_CR
2964 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2966 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
2968 case ' ': case '\t': case '\f': case 013:
2969 #ifdef MACOS_TRADITIONAL
2976 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2977 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2978 /* handle eval qq[#line 1 "foo"\n ...] */
2979 CopLINE_dec(PL_curcop);
2983 while (s < d && *s != '\n')
2987 else if (s > d) /* Found by Ilya: feed random input to Perl. */
2988 Perl_croak(aTHX_ "panic: input overflow");
2990 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2992 PL_lex_state = LEX_FORMLINE;
3002 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
3009 while (s < PL_bufend && SPACE_OR_TAB(*s))
3012 if (strnEQ(s,"=>",2)) {
3013 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
3014 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3015 "### Saw unary minus before =>, forcing word '%s'\n", s);
3017 OPERATOR('-'); /* unary minus */
3019 PL_last_uni = PL_oldbufptr;
3021 case 'r': ftst = OP_FTEREAD; break;
3022 case 'w': ftst = OP_FTEWRITE; break;
3023 case 'x': ftst = OP_FTEEXEC; break;
3024 case 'o': ftst = OP_FTEOWNED; break;
3025 case 'R': ftst = OP_FTRREAD; break;
3026 case 'W': ftst = OP_FTRWRITE; break;
3027 case 'X': ftst = OP_FTREXEC; break;
3028 case 'O': ftst = OP_FTROWNED; break;
3029 case 'e': ftst = OP_FTIS; break;
3030 case 'z': ftst = OP_FTZERO; break;
3031 case 's': ftst = OP_FTSIZE; break;
3032 case 'f': ftst = OP_FTFILE; break;
3033 case 'd': ftst = OP_FTDIR; break;
3034 case 'l': ftst = OP_FTLINK; break;
3035 case 'p': ftst = OP_FTPIPE; break;
3036 case 'S': ftst = OP_FTSOCK; break;
3037 case 'u': ftst = OP_FTSUID; break;
3038 case 'g': ftst = OP_FTSGID; break;
3039 case 'k': ftst = OP_FTSVTX; break;
3040 case 'b': ftst = OP_FTBLK; break;
3041 case 'c': ftst = OP_FTCHR; break;
3042 case 't': ftst = OP_FTTTY; break;
3043 case 'T': ftst = OP_FTTEXT; break;
3044 case 'B': ftst = OP_FTBINARY; break;
3045 case 'M': case 'A': case 'C':
3046 gv_fetchpv("\024",TRUE, SVt_PV);
3048 case 'M': ftst = OP_FTMTIME; break;
3049 case 'A': ftst = OP_FTATIME; break;
3050 case 'C': ftst = OP_FTCTIME; break;
3058 PL_last_lop_op = (OPCODE)ftst;
3059 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3060 "### Saw file test %c\n", (int)ftst);
3065 /* Assume it was a minus followed by a one-letter named
3066 * subroutine call (or a -bareword), then. */
3067 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3068 "### '-%c' looked like a file test but was not\n",
3077 if (PL_expect == XOPERATOR)
3082 else if (*s == '>') {
3085 if (isIDFIRST_lazy_if(s,UTF)) {
3086 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
3094 if (PL_expect == XOPERATOR)
3097 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
3099 OPERATOR('-'); /* unary minus */
3106 if (PL_expect == XOPERATOR)
3111 if (PL_expect == XOPERATOR)
3114 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
3120 if (PL_expect != XOPERATOR) {
3121 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3122 PL_expect = XOPERATOR;
3123 force_ident(PL_tokenbuf, '*');
3136 if (PL_expect == XOPERATOR) {
3140 PL_tokenbuf[0] = '%';
3141 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
3142 if (!PL_tokenbuf[1]) {
3145 PL_pending_ident = '%';
3164 switch (PL_expect) {
3167 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
3169 PL_bufptr = s; /* update in case we back off */
3175 PL_expect = XTERMBLOCK;
3179 while (isIDFIRST_lazy_if(s,UTF)) {
3180 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3181 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
3182 if (tmp < 0) tmp = -tmp;
3198 d = scan_str(d,TRUE,TRUE);
3200 /* MUST advance bufptr here to avoid bogus
3201 "at end of line" context messages from yyerror().
3203 PL_bufptr = s + len;
3204 yyerror("Unterminated attribute parameter in attribute list");
3207 return REPORT(0); /* EOF indicator */
3211 SV *sv = newSVpvn(s, len);
3212 sv_catsv(sv, PL_lex_stuff);
3213 attrs = append_elem(OP_LIST, attrs,
3214 newSVOP(OP_CONST, 0, sv));
3215 SvREFCNT_dec(PL_lex_stuff);
3216 PL_lex_stuff = Nullsv;
3219 if (len == 6 && strnEQ(s, "unique", len)) {
3220 if (PL_in_my == KEY_our)
3222 GvUNIQUE_on(cGVOPx_gv(yylval.opval));
3224 ; /* skip to avoid loading attributes.pm */
3227 Perl_croak(aTHX_ "The 'unique' attribute may only be applied to 'our' variables");
3230 /* NOTE: any CV attrs applied here need to be part of
3231 the CVf_BUILTIN_ATTRS define in cv.h! */
3232 else if (!PL_in_my && len == 6 && strnEQ(s, "lvalue", len))
3233 CvLVALUE_on(PL_compcv);
3234 else if (!PL_in_my && len == 6 && strnEQ(s, "locked", len))
3235 CvLOCKED_on(PL_compcv);
3236 else if (!PL_in_my && len == 6 && strnEQ(s, "method", len))
3237 CvMETHOD_on(PL_compcv);
3238 else if (!PL_in_my && len == 9 && strnEQ(s, "assertion", len))
3239 CvASSERTION_on(PL_compcv);
3240 /* After we've set the flags, it could be argued that
3241 we don't need to do the attributes.pm-based setting
3242 process, and shouldn't bother appending recognized
3243 flags. To experiment with that, uncomment the
3244 following "else". (Note that's already been
3245 uncommented. That keeps the above-applied built-in
3246 attributes from being intercepted (and possibly
3247 rejected) by a package's attribute routines, but is
3248 justified by the performance win for the common case
3249 of applying only built-in attributes.) */
3251 attrs = append_elem(OP_LIST, attrs,
3252 newSVOP(OP_CONST, 0,
3256 if (*s == ':' && s[1] != ':')
3259 break; /* require real whitespace or :'s */
3261 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
3262 if (*s != ';' && *s != '}' && *s != tmp && (tmp != '=' || *s != ')')) {
3263 const char q = ((*s == '\'') ? '"' : '\'');
3264 /* If here for an expression, and parsed no attrs, back off. */
3265 if (tmp == '=' && !attrs) {
3269 /* MUST advance bufptr here to avoid bogus "at end of line"
3270 context messages from yyerror().
3274 yyerror("Unterminated attribute list");
3276 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
3284 PL_nextval[PL_nexttoke].opval = attrs;
3292 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
3293 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
3310 if (PL_lex_brackets <= 0)
3311 yyerror("Unmatched right square bracket");
3314 if (PL_lex_state == LEX_INTERPNORMAL) {
3315 if (PL_lex_brackets == 0) {
3316 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
3317 PL_lex_state = LEX_INTERPEND;
3324 if (PL_lex_brackets > 100) {
3325 Renew(PL_lex_brackstack, PL_lex_brackets + 10, char);
3327 switch (PL_expect) {
3329 if (PL_lex_formbrack) {
3333 if (PL_oldoldbufptr == PL_last_lop)
3334 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3336 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3337 OPERATOR(HASHBRACK);
3339 while (s < PL_bufend && SPACE_OR_TAB(*s))
3342 PL_tokenbuf[0] = '\0';
3343 if (d < PL_bufend && *d == '-') {
3344 PL_tokenbuf[0] = '-';
3346 while (d < PL_bufend && SPACE_OR_TAB(*d))
3349 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3350 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3352 while (d < PL_bufend && SPACE_OR_TAB(*d))
3355 const char minus = (PL_tokenbuf[0] == '-');
3356 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3364 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3369 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3374 if (PL_oldoldbufptr == PL_last_lop)
3375 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3377 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3380 if (PL_expect == XREF && PL_lex_state == LEX_INTERPNORMAL) {
3382 /* This hack is to get the ${} in the message. */
3384 yyerror("syntax error");
3387 OPERATOR(HASHBRACK);
3389 /* This hack serves to disambiguate a pair of curlies
3390 * as being a block or an anon hash. Normally, expectation
3391 * determines that, but in cases where we're not in a
3392 * position to expect anything in particular (like inside
3393 * eval"") we have to resolve the ambiguity. This code
3394 * covers the case where the first term in the curlies is a
3395 * quoted string. Most other cases need to be explicitly
3396 * disambiguated by prepending a "+" before the opening
3397 * curly in order to force resolution as an anon hash.
3399 * XXX should probably propagate the outer expectation
3400 * into eval"" to rely less on this hack, but that could
3401 * potentially break current behavior of eval"".
3405 if (*s == '\'' || *s == '"' || *s == '`') {
3406 /* common case: get past first string, handling escapes */
3407 for (t++; t < PL_bufend && *t != *s;)
3408 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3412 else if (*s == 'q') {
3415 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3418 /* skip q//-like construct */
3420 char open, close, term;
3423 while (t < PL_bufend && isSPACE(*t))
3425 /* check for q => */
3426 if (t+1 < PL_bufend && t[0] == '=' && t[1] == '>') {
3427 OPERATOR(HASHBRACK);
3431 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3435 for (t++; t < PL_bufend; t++) {
3436 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3438 else if (*t == open)
3442 for (t++; t < PL_bufend; t++) {
3443 if (*t == '\\' && t+1 < PL_bufend)
3445 else if (*t == close && --brackets <= 0)
3447 else if (*t == open)
3454 /* skip plain q word */
3455 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3458 else if (isALNUM_lazy_if(t,UTF)) {
3460 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3463 while (t < PL_bufend && isSPACE(*t))
3465 /* if comma follows first term, call it an anon hash */
3466 /* XXX it could be a comma expression with loop modifiers */
3467 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3468 || (*t == '=' && t[1] == '>')))
3469 OPERATOR(HASHBRACK);
3470 if (PL_expect == XREF)
3473 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3479 yylval.ival = CopLINE(PL_curcop);
3480 if (isSPACE(*s) || *s == '#')
3481 PL_copline = NOLINE; /* invalidate current command line number */
3486 if (PL_lex_brackets <= 0)
3487 yyerror("Unmatched right curly bracket");
3489 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3490 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3491 PL_lex_formbrack = 0;
3492 if (PL_lex_state == LEX_INTERPNORMAL) {
3493 if (PL_lex_brackets == 0) {
3494 if (PL_expect & XFAKEBRACK) {
3495 PL_expect &= XENUMMASK;
3496 PL_lex_state = LEX_INTERPEND;
3498 return yylex(); /* ignore fake brackets */
3500 if (*s == '-' && s[1] == '>')
3501 PL_lex_state = LEX_INTERPENDMAYBE;
3502 else if (*s != '[' && *s != '{')
3503 PL_lex_state = LEX_INTERPEND;
3506 if (PL_expect & XFAKEBRACK) {
3507 PL_expect &= XENUMMASK;
3509 return yylex(); /* ignore fake brackets */
3519 if (PL_expect == XOPERATOR) {
3520 if (PL_bufptr == PL_linestart && ckWARN(WARN_SEMICOLON)
3521 && isIDFIRST_lazy_if(s,UTF))
3523 CopLINE_dec(PL_curcop);
3524 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
3525 CopLINE_inc(PL_curcop);
3530 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3532 PL_expect = XOPERATOR;
3533 force_ident(PL_tokenbuf, '&');
3537 yylval.ival = (OPpENTERSUB_AMPER<<8);
3556 if (tmp && isSPACE(*s) && ckWARN(WARN_SYNTAX) && strchr("+-*/%.^&|<",tmp))
3557 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Reversed %c= operator",(int)tmp);
3559 if (PL_expect == XSTATE && isALPHA(tmp) &&
3560 (s == PL_linestart+1 || s[-2] == '\n') )
3562 if (PL_in_eval && !PL_rsfp) {
3567 if (strnEQ(s,"=cut",4)) {
3581 PL_doextract = TRUE;
3584 if (PL_lex_brackets < PL_lex_formbrack) {
3586 #ifdef PERL_STRICT_CR
3587 for (t = s; SPACE_OR_TAB(*t); t++) ;
3589 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3591 if (*t == '\n' || *t == '#') {
3603 /* was this !=~ where !~ was meant?
3604 * warn on m:!=~\s+([/?]|[msy]\W|tr\W): */
3606 if (*s == '~' && ckWARN(WARN_SYNTAX)) {
3607 const char *t = s+1;
3609 while (t < PL_bufend && isSPACE(*t))
3612 if (*t == '/' || *t == '?' ||
3613 ((*t == 'm' || *t == 's' || *t == 'y') && !isALNUM(t[1])) ||
3614 (*t == 't' && t[1] == 'r' && !isALNUM(t[2])))
3615 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3616 "!=~ should be !~");
3625 if (PL_expect != XOPERATOR) {
3626 if (s[1] != '<' && !strchr(s,'>'))
3629 s = scan_heredoc(s);
3631 s = scan_inputsymbol(s);
3632 TERM(sublex_start());
3637 SHop(OP_LEFT_SHIFT);
3651 SHop(OP_RIGHT_SHIFT);
3660 if (PL_expect == XOPERATOR) {
3661 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3664 return REPORT(','); /* grandfather non-comma-format format */
3668 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3669 PL_tokenbuf[0] = '@';
3670 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3671 sizeof PL_tokenbuf - 1, FALSE);
3672 if (PL_expect == XOPERATOR)
3673 no_op("Array length", s);
3674 if (!PL_tokenbuf[1])
3676 PL_expect = XOPERATOR;
3677 PL_pending_ident = '#';
3681 PL_tokenbuf[0] = '$';
3682 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3683 sizeof PL_tokenbuf - 1, FALSE);
3684 if (PL_expect == XOPERATOR)
3686 if (!PL_tokenbuf[1]) {
3688 yyerror("Final $ should be \\$ or $name");
3692 /* This kludge not intended to be bulletproof. */
3693 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3694 yylval.opval = newSVOP(OP_CONST, 0,
3695 newSViv(PL_compiling.cop_arybase));
3696 yylval.opval->op_private = OPpCONST_ARYBASE;
3702 if (PL_lex_state == LEX_NORMAL)
3705 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3707 PL_tokenbuf[0] = '@';
3708 if (ckWARN(WARN_SYNTAX)) {
3711 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3714 PL_bufptr = skipspace(PL_bufptr);
3715 while (t < PL_bufend && *t != ']')
3717 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3718 "Multidimensional syntax %.*s not supported",
3719 (t - PL_bufptr) + 1, PL_bufptr);
3723 else if (*s == '{') {
3725 PL_tokenbuf[0] = '%';
3726 if (strEQ(PL_tokenbuf+1, "SIG") && ckWARN(WARN_SYNTAX)
3727 && (t = strchr(s, '}')) && (t = strchr(t, '=')))
3729 char tmpbuf[sizeof PL_tokenbuf];
3730 for (t++; isSPACE(*t); t++) ;
3731 if (isIDFIRST_lazy_if(t,UTF)) {
3733 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3734 for (; isSPACE(*t); t++) ;
3735 if (*t == ';' && get_cv(tmpbuf, FALSE))
3736 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3737 "You need to quote \"%s\"", tmpbuf);
3743 PL_expect = XOPERATOR;
3744 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3745 const bool islop = (PL_last_lop == PL_oldoldbufptr);
3746 if (!islop || PL_last_lop_op == OP_GREPSTART)
3747 PL_expect = XOPERATOR;
3748 else if (strchr("$@\"'`q", *s))
3749 PL_expect = XTERM; /* e.g. print $fh "foo" */
3750 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3751 PL_expect = XTERM; /* e.g. print $fh &sub */
3752 else if (isIDFIRST_lazy_if(s,UTF)) {
3753 char tmpbuf[sizeof PL_tokenbuf];
3754 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3755 if ((tmp = keyword(tmpbuf, len))) {
3756 /* binary operators exclude handle interpretations */
3768 PL_expect = XTERM; /* e.g. print $fh length() */
3773 PL_expect = XTERM; /* e.g. print $fh subr() */
3776 else if (isDIGIT(*s))
3777 PL_expect = XTERM; /* e.g. print $fh 3 */
3778 else if (*s == '.' && isDIGIT(s[1]))
3779 PL_expect = XTERM; /* e.g. print $fh .3 */
3780 else if ((*s == '?' || *s == '-' || *s == '+')
3781 && !isSPACE(s[1]) && s[1] != '=')
3782 PL_expect = XTERM; /* e.g. print $fh -1 */
3783 else if (*s == '/' && !isSPACE(s[1]) && s[1] != '=' && s[1] != '/')
3784 PL_expect = XTERM; /* e.g. print $fh /.../
3785 XXX except DORDOR operator */
3786 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3787 PL_expect = XTERM; /* print $fh <<"EOF" */
3789 PL_pending_ident = '$';
3793 if (PL_expect == XOPERATOR)
3795 PL_tokenbuf[0] = '@';
3796 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3797 if (!PL_tokenbuf[1]) {
3800 if (PL_lex_state == LEX_NORMAL)
3802 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3804 PL_tokenbuf[0] = '%';
3806 /* Warn about @ where they meant $. */
3807 if (*s == '[' || *s == '{') {
3808 if (ckWARN(WARN_SYNTAX)) {
3809 const char *t = s + 1;
3810 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3812 if (*t == '}' || *t == ']') {
3814 PL_bufptr = skipspace(PL_bufptr);
3815 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3816 "Scalar value %.*s better written as $%.*s",
3817 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3822 PL_pending_ident = '@';
3825 case '/': /* may be division, defined-or, or pattern */
3826 if (PL_expect == XTERMORDORDOR && s[1] == '/') {
3830 case '?': /* may either be conditional or pattern */
3831 if(PL_expect == XOPERATOR) {
3839 /* A // operator. */
3849 /* Disable warning on "study /blah/" */
3850 if (PL_oldoldbufptr == PL_last_uni
3851 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3852 || memNE(PL_last_uni, "study", 5)
3853 || isALNUM_lazy_if(PL_last_uni+5,UTF)
3856 s = scan_pat(s,OP_MATCH);
3857 TERM(sublex_start());
3861 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3862 #ifdef PERL_STRICT_CR
3865 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3867 && (s == PL_linestart || s[-1] == '\n') )
3869 PL_lex_formbrack = 0;
3873 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3879 yylval.ival = OPf_SPECIAL;
3885 if (PL_expect != XOPERATOR)
3890 case '0': case '1': case '2': case '3': case '4':
3891 case '5': case '6': case '7': case '8': case '9':
3892 s = scan_num(s, &yylval);
3893 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3894 "### Saw number in '%s'\n", s);
3896 if (PL_expect == XOPERATOR)
3901 s = scan_str(s,FALSE,FALSE);
3902 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3903 "### Saw string before '%s'\n", s);
3905 if (PL_expect == XOPERATOR) {
3906 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3909 return REPORT(','); /* grandfather non-comma-format format */
3915 missingterm((char*)0);
3916 yylval.ival = OP_CONST;
3917 TERM(sublex_start());
3920 s = scan_str(s,FALSE,FALSE);
3921 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3922 "### Saw string before '%s'\n", s);
3924 if (PL_expect == XOPERATOR) {
3925 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3928 return REPORT(','); /* grandfather non-comma-format format */
3934 missingterm((char*)0);
3935 yylval.ival = OP_CONST;
3936 /* FIXME. I think that this can be const if char *d is replaced by
3937 more localised variables. */
3938 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3939 if (*d == '$' || *d == '@' || *d == '\\' || !UTF8_IS_INVARIANT((U8)*d)) {
3940 yylval.ival = OP_STRINGIFY;
3944 TERM(sublex_start());
3947 s = scan_str(s,FALSE,FALSE);
3948 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3949 "### Saw backtick string before '%s'\n", s);
3951 if (PL_expect == XOPERATOR)
3952 no_op("Backticks",s);
3954 missingterm((char*)0);
3955 yylval.ival = OP_BACKTICK;
3957 TERM(sublex_start());
3961 if (PL_lex_inwhat && isDIGIT(*s) && ckWARN(WARN_SYNTAX))
3962 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),"Can't use \\%c to mean $%c in expression",
3964 if (PL_expect == XOPERATOR)
3965 no_op("Backslash",s);
3969 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3970 char *start = s + 2;
3971 while (isDIGIT(*start) || *start == '_')
3973 if (*start == '.' && isDIGIT(start[1])) {
3974 s = scan_num(s, &yylval);
3977 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3978 else if (!isALPHA(*start) && (PL_expect == XTERM
3979 || PL_expect == XREF || PL_expect == XSTATE
3980 || PL_expect == XTERMORDORDOR)) {
3981 const char c = *start;
3984 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3987 s = scan_num(s, &yylval);
3994 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
4034 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4036 /* Some keywords can be followed by any delimiter, including ':' */
4037 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
4038 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
4039 (PL_tokenbuf[0] == 'q' &&
4040 strchr("qwxr", PL_tokenbuf[1])))));
4042 /* x::* is just a word, unless x is "CORE" */
4043 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
4047 while (d < PL_bufend && isSPACE(*d))
4048 d++; /* no comments skipped here, or s### is misparsed */
4050 /* Is this a label? */
4051 if (!tmp && PL_expect == XSTATE
4052 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
4054 yylval.pval = savepv(PL_tokenbuf);
4059 /* Check for keywords */
4060 tmp = keyword(PL_tokenbuf, len);
4062 /* Is this a word before a => operator? */
4063 if (*d == '=' && d[1] == '>') {
4066 = (OP*)newSVOP(OP_CONST, 0,
4067 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
4068 yylval.opval->op_private = OPpCONST_BARE;
4072 if (tmp < 0) { /* second-class keyword? */
4073 GV *ogv = Nullgv; /* override (winner) */
4074 GV *hgv = Nullgv; /* hidden (loser) */
4075 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
4077 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
4080 if (GvIMPORTED_CV(gv))
4082 else if (! CvMETHOD(cv))
4086 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
4087 (gv = *gvp) != (GV*)&PL_sv_undef &&
4088 GvCVu(gv) && GvIMPORTED_CV(gv))
4095 tmp = 0; /* overridden by import or by GLOBAL */
4098 && -tmp==KEY_lock /* XXX generalizable kludge */
4100 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
4102 tmp = 0; /* any sub overrides "weak" keyword */
4107 && PL_expect != XOPERATOR
4108 && PL_expect != XTERMORDORDOR)
4110 /* any sub overrides the "err" keyword, except when really an
4111 * operator is expected */
4114 else { /* no override */
4116 if (tmp == KEY_dump && ckWARN(WARN_MISC)) {
4117 Perl_warner(aTHX_ packWARN(WARN_MISC),
4118 "dump() better written as CORE::dump()");
4122 if (hgv && tmp != KEY_x && tmp != KEY_CORE
4123 && ckWARN(WARN_AMBIGUOUS)) /* never ambiguous */
4124 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4125 "Ambiguous call resolved as CORE::%s(), %s",
4126 GvENAME(hgv), "qualify as such or use &");
4133 default: /* not a keyword */
4137 const char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
4139 /* Get the rest if it looks like a package qualifier */
4141 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
4143 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
4146 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
4147 *s == '\'' ? "'" : "::");
4152 if (PL_expect == XOPERATOR) {
4153 if (PL_bufptr == PL_linestart) {
4154 CopLINE_dec(PL_curcop);
4155 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
4156 CopLINE_inc(PL_curcop);
4159 no_op("Bareword",s);
4162 /* Look for a subroutine with this name in current package,
4163 unless name is "Foo::", in which case Foo is a bearword
4164 (and a package name). */
4167 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
4169 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
4170 Perl_warner(aTHX_ packWARN(WARN_BAREWORD),
4171 "Bareword \"%s\" refers to nonexistent package",
4174 PL_tokenbuf[len] = '\0';
4181 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
4184 /* if we saw a global override before, get the right name */
4187 sv = newSVpvn("CORE::GLOBAL::",14);
4188 sv_catpv(sv,PL_tokenbuf);
4191 /* If len is 0, newSVpv does strlen(), which is correct.
4192 If len is non-zero, then it will be the true length,
4193 and so the scalar will be created correctly. */
4194 sv = newSVpv(PL_tokenbuf,len);
4197 /* Presume this is going to be a bareword of some sort. */
4200 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
4201 yylval.opval->op_private = OPpCONST_BARE;
4202 /* UTF-8 package name? */
4203 if (UTF && !IN_BYTES &&
4204 is_utf8_string((U8*)SvPVX_const(sv), SvCUR(sv)))
4207 /* And if "Foo::", then that's what it certainly is. */
4212 /* See if it's the indirect object for a list operator. */
4214 if (PL_oldoldbufptr &&
4215 PL_oldoldbufptr < PL_bufptr &&
4216 (PL_oldoldbufptr == PL_last_lop
4217 || PL_oldoldbufptr == PL_last_uni) &&
4218 /* NO SKIPSPACE BEFORE HERE! */
4219 (PL_expect == XREF ||
4220 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
4222 bool immediate_paren = *s == '(';
4224 /* (Now we can afford to cross potential line boundary.) */
4227 /* Two barewords in a row may indicate method call. */
4229 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
4232 /* If not a declared subroutine, it's an indirect object. */
4233 /* (But it's an indir obj regardless for sort.) */
4235 if ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
4236 ((!gv || !GvCVu(gv)) &&
4237 (PL_last_lop_op != OP_MAPSTART &&
4238 PL_last_lop_op != OP_GREPSTART))))
4240 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
4245 PL_expect = XOPERATOR;
4248 /* Is this a word before a => operator? */
4249 if (*s == '=' && s[1] == '>' && !pkgname) {
4251 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
4252 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
4253 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
4257 /* If followed by a paren, it's certainly a subroutine. */
4260 if (gv && GvCVu(gv)) {
4261 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
4262 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
4267 PL_nextval[PL_nexttoke].opval = yylval.opval;
4268 PL_expect = XOPERATOR;
4274 /* If followed by var or block, call it a method (unless sub) */
4276 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
4277 PL_last_lop = PL_oldbufptr;
4278 PL_last_lop_op = OP_METHOD;
4282 /* If followed by a bareword, see if it looks like indir obj. */
4285 && (isIDFIRST_lazy_if(s,UTF) || *s == '$')
4286 && (tmp = intuit_method(s,gv)))
4289 /* Not a method, so call it a subroutine (if defined) */
4291 if (gv && GvCVu(gv)) {
4293 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
4294 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4295 "Ambiguous use of -%s resolved as -&%s()",
4296 PL_tokenbuf, PL_tokenbuf);
4297 /* Check for a constant sub */
4299 if ((sv = cv_const_sv(cv))) {
4301 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
4302 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
4303 yylval.opval->op_private = 0;
4307 /* Resolve to GV now. */
4308 op_free(yylval.opval);
4309 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
4310 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
4311 PL_last_lop = PL_oldbufptr;
4312 PL_last_lop_op = OP_ENTERSUB;
4313 /* Is there a prototype? */
4316 const char *proto = SvPV_const((SV*)cv, len);
4319 if (*proto == '$' && proto[1] == '\0')
4321 while (*proto == ';')
4323 if (*proto == '&' && *s == '{') {
4324 sv_setpv(PL_subname, PL_curstash ?
4325 "__ANON__" : "__ANON__::__ANON__");
4329 PL_nextval[PL_nexttoke].opval = yylval.opval;
4335 /* Call it a bare word */
4337 if (PL_hints & HINT_STRICT_SUBS)
4338 yylval.opval->op_private |= OPpCONST_STRICT;
4341 if (lastchar != '-') {
4342 if (ckWARN(WARN_RESERVED)) {
4343 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
4344 if (!*d && !gv_stashpv(PL_tokenbuf,FALSE))
4345 Perl_warner(aTHX_ packWARN(WARN_RESERVED), PL_warn_reserved,
4352 if ((lastchar == '*' || lastchar == '%' || lastchar == '&')
4353 && ckWARN_d(WARN_AMBIGUOUS)) {
4354 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4355 "Operator or semicolon missing before %c%s",
4356 lastchar, PL_tokenbuf);
4357 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4358 "Ambiguous use of %c resolved as operator %c",
4359 lastchar, lastchar);
4365 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4366 newSVpv(CopFILE(PL_curcop),0));
4370 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4371 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
4374 case KEY___PACKAGE__:
4375 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4377 ? newSVhek(HvNAME_HEK(PL_curstash))
4384 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
4385 const char *pname = "main";
4386 if (PL_tokenbuf[2] == 'D')
4387 pname = HvNAME_get(PL_curstash ? PL_curstash : PL_defstash);
4388 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
4391 GvIOp(gv) = newIO();
4392 IoIFP(GvIOp(gv)) = PL_rsfp;
4393 #if defined(HAS_FCNTL) && defined(F_SETFD)
4395 const int fd = PerlIO_fileno(PL_rsfp);
4396 fcntl(fd,F_SETFD,fd >= 3);
4399 /* Mark this internal pseudo-handle as clean */
4400 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
4402 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
4403 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
4404 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
4406 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
4407 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
4408 /* if the script was opened in binmode, we need to revert
4409 * it to text mode for compatibility; but only iff it has CRs
4410 * XXX this is a questionable hack at best. */
4411 if (PL_bufend-PL_bufptr > 2
4412 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
4415 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
4416 loc = PerlIO_tell(PL_rsfp);
4417 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4420 if (PerlLIO_setmode(PL_rsfp, O_TEXT) != -1) {
4422 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4423 #endif /* NETWARE */
4424 #ifdef PERLIO_IS_STDIO /* really? */
4425 # if defined(__BORLANDC__)
4426 /* XXX see note in do_binmode() */
4427 ((FILE*)PL_rsfp)->flags &= ~_F_BIN;
4431 PerlIO_seek(PL_rsfp, loc, 0);
4435 #ifdef PERLIO_LAYERS
4438 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
4439 else if (PL_encoding) {
4446 XPUSHs(PL_encoding);
4448 call_method("name", G_SCALAR);
4452 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL,
4453 Perl_form(aTHX_ ":encoding(%"SVf")",
4471 if (PL_expect == XSTATE) {
4478 if (*s == ':' && s[1] == ':') {
4481 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4482 if (!(tmp = keyword(PL_tokenbuf, len)))
4483 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4497 LOP(OP_ACCEPT,XTERM);
4503 LOP(OP_ATAN2,XTERM);
4509 LOP(OP_BINMODE,XTERM);
4512 LOP(OP_BLESS,XTERM);
4521 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4538 if (!PL_cryptseen) {
4539 PL_cryptseen = TRUE;
4543 LOP(OP_CRYPT,XTERM);
4546 LOP(OP_CHMOD,XTERM);
4549 LOP(OP_CHOWN,XTERM);
4552 LOP(OP_CONNECT,XTERM);
4568 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4572 PL_hints |= HINT_BLOCK_SCOPE;
4582 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4583 LOP(OP_DBMOPEN,XTERM);
4589 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4596 yylval.ival = CopLINE(PL_curcop);
4610 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4611 UNIBRACK(OP_ENTEREVAL);
4629 case KEY_endhostent:
4635 case KEY_endservent:
4638 case KEY_endprotoent:
4649 yylval.ival = CopLINE(PL_curcop);
4651 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4653 if ((PL_bufend - p) >= 3 &&
4654 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4656 else if ((PL_bufend - p) >= 4 &&
4657 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4660 if (isIDFIRST_lazy_if(p,UTF)) {
4661 p = scan_ident(p, PL_bufend,
4662 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4666 Perl_croak(aTHX_ "Missing $ on loop variable");
4671 LOP(OP_FORMLINE,XTERM);
4677 LOP(OP_FCNTL,XTERM);
4683 LOP(OP_FLOCK,XTERM);
4692 LOP(OP_GREPSTART, XREF);
4695 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4710 case KEY_getpriority:
4711 LOP(OP_GETPRIORITY,XTERM);
4713 case KEY_getprotobyname:
4716 case KEY_getprotobynumber:
4717 LOP(OP_GPBYNUMBER,XTERM);
4719 case KEY_getprotoent:
4731 case KEY_getpeername:
4732 UNI(OP_GETPEERNAME);
4734 case KEY_gethostbyname:
4737 case KEY_gethostbyaddr:
4738 LOP(OP_GHBYADDR,XTERM);
4740 case KEY_gethostent:
4743 case KEY_getnetbyname:
4746 case KEY_getnetbyaddr:
4747 LOP(OP_GNBYADDR,XTERM);
4752 case KEY_getservbyname:
4753 LOP(OP_GSBYNAME,XTERM);
4755 case KEY_getservbyport:
4756 LOP(OP_GSBYPORT,XTERM);
4758 case KEY_getservent:
4761 case KEY_getsockname:
4762 UNI(OP_GETSOCKNAME);
4764 case KEY_getsockopt:
4765 LOP(OP_GSOCKOPT,XTERM);
4787 yylval.ival = CopLINE(PL_curcop);
4791 LOP(OP_INDEX,XTERM);
4797 LOP(OP_IOCTL,XTERM);
4809 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4841 LOP(OP_LISTEN,XTERM);
4850 s = scan_pat(s,OP_MATCH);
4851 TERM(sublex_start());
4854 LOP(OP_MAPSTART, XREF);
4857 LOP(OP_MKDIR,XTERM);
4860 LOP(OP_MSGCTL,XTERM);
4863 LOP(OP_MSGGET,XTERM);
4866 LOP(OP_MSGRCV,XTERM);
4869 LOP(OP_MSGSND,XTERM);
4875 if (isIDFIRST_lazy_if(s,UTF)) {
4876 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4877 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4879 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
4880 if (!PL_in_my_stash) {
4883 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4891 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4898 s = tokenize_use(0, s);
4902 if (*s == '(' || (s = skipspace(s), *s == '('))
4909 if (isIDFIRST_lazy_if(s,UTF)) {
4911 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4912 for (t=d; *t && isSPACE(*t); t++) ;
4913 if ( *t && strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE)
4915 && !(t[0] == '=' && t[1] == '>')
4917 Perl_warner(aTHX_ packWARN(WARN_PRECEDENCE),
4918 "Precedence problem: open %.*s should be open(%.*s)",
4919 d - s, s, d - s, s);
4925 yylval.ival = OP_OR;
4935 LOP(OP_OPEN_DIR,XTERM);
4938 checkcomma(s,PL_tokenbuf,"filehandle");
4942 checkcomma(s,PL_tokenbuf,"filehandle");
4961 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4965 LOP(OP_PIPE_OP,XTERM);
4968 s = scan_str(s,FALSE,FALSE);
4970 missingterm((char*)0);
4971 yylval.ival = OP_CONST;
4972 TERM(sublex_start());
4978 s = scan_str(s,FALSE,FALSE);
4980 missingterm((char*)0);
4981 PL_expect = XOPERATOR;
4983 if (SvCUR(PL_lex_stuff)) {
4986 d = SvPV_force(PL_lex_stuff, len);
4989 for (; isSPACE(*d) && len; --len, ++d) ;
4992 if (!warned && ckWARN(WARN_QW)) {
4993 for (; !isSPACE(*d) && len; --len, ++d) {
4995 Perl_warner(aTHX_ packWARN(WARN_QW),
4996 "Possible attempt to separate words with commas");
4999 else if (*d == '#') {
5000 Perl_warner(aTHX_ packWARN(WARN_QW),
5001 "Possible attempt to put comments in qw() list");
5007 for (; !isSPACE(*d) && len; --len, ++d) ;
5009 sv = newSVpvn(b, d-b);
5010 if (DO_UTF8(PL_lex_stuff))
5012 words = append_elem(OP_LIST, words,
5013 newSVOP(OP_CONST, 0, tokeq(sv)));
5017 PL_nextval[PL_nexttoke].opval = words;
5022 SvREFCNT_dec(PL_lex_stuff);
5023 PL_lex_stuff = Nullsv;
5029 s = scan_str(s,FALSE,FALSE);
5031 missingterm((char*)0);
5032 yylval.ival = OP_STRINGIFY;
5033 if (SvIVX(PL_lex_stuff) == '\'')
5034 SvIV_set(PL_lex_stuff, 0); /* qq'$foo' should intepolate */
5035 TERM(sublex_start());
5038 s = scan_pat(s,OP_QR);
5039 TERM(sublex_start());
5042 s = scan_str(s,FALSE,FALSE);
5044 missingterm((char*)0);
5045 yylval.ival = OP_BACKTICK;
5047 TERM(sublex_start());
5055 s = force_version(s, FALSE);
5057 else if (*s != 'v' || !isDIGIT(s[1])
5058 || (s = force_version(s, TRUE), *s == 'v'))
5060 *PL_tokenbuf = '\0';
5061 s = force_word(s,WORD,TRUE,TRUE,FALSE);
5062 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
5063 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
5065 yyerror("<> should be quotes");
5073 s = force_word(s,WORD,TRUE,FALSE,FALSE);
5077 LOP(OP_RENAME,XTERM);
5086 LOP(OP_RINDEX,XTERM);
5096 UNIDOR(OP_READLINE);
5109 LOP(OP_REVERSE,XTERM);
5112 UNIDOR(OP_READLINK);
5120 TERM(sublex_start());
5122 TOKEN(1); /* force error */
5131 LOP(OP_SELECT,XTERM);
5137 LOP(OP_SEMCTL,XTERM);
5140 LOP(OP_SEMGET,XTERM);
5143 LOP(OP_SEMOP,XTERM);
5149 LOP(OP_SETPGRP,XTERM);
5151 case KEY_setpriority:
5152 LOP(OP_SETPRIORITY,XTERM);
5154 case KEY_sethostent:
5160 case KEY_setservent:
5163 case KEY_setprotoent:
5173 LOP(OP_SEEKDIR,XTERM);
5175 case KEY_setsockopt:
5176 LOP(OP_SSOCKOPT,XTERM);
5182 LOP(OP_SHMCTL,XTERM);
5185 LOP(OP_SHMGET,XTERM);
5188 LOP(OP_SHMREAD,XTERM);
5191 LOP(OP_SHMWRITE,XTERM);
5194 LOP(OP_SHUTDOWN,XTERM);
5203 LOP(OP_SOCKET,XTERM);
5205 case KEY_socketpair:
5206 LOP(OP_SOCKPAIR,XTERM);
5209 checkcomma(s,PL_tokenbuf,"subroutine name");
5211 if (*s == ';' || *s == ')') /* probably a close */
5212 Perl_croak(aTHX_ "sort is now a reserved word");
5214 s = force_word(s,WORD,TRUE,TRUE,FALSE);
5218 LOP(OP_SPLIT,XTERM);
5221 LOP(OP_SPRINTF,XTERM);
5224 LOP(OP_SPLICE,XTERM);
5239 LOP(OP_SUBSTR,XTERM);
5245 char tmpbuf[sizeof PL_tokenbuf];
5246 SSize_t tboffset = 0;
5247 expectation attrful;
5248 bool have_name, have_proto, bad_proto;
5249 const int key = tmp;
5253 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
5254 (*s == ':' && s[1] == ':'))
5257 attrful = XATTRBLOCK;
5258 /* remember buffer pos'n for later force_word */
5259 tboffset = s - PL_oldbufptr;
5260 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
5261 if (strchr(tmpbuf, ':'))
5262 sv_setpv(PL_subname, tmpbuf);
5264 sv_setsv(PL_subname,PL_curstname);
5265 sv_catpvn(PL_subname,"::",2);
5266 sv_catpvn(PL_subname,tmpbuf,len);
5273 Perl_croak(aTHX_ "Missing name in \"my sub\"");
5274 PL_expect = XTERMBLOCK;
5275 attrful = XATTRTERM;
5276 sv_setpvn(PL_subname,"?",1);
5280 if (key == KEY_format) {
5282 PL_lex_formbrack = PL_lex_brackets + 1;
5284 (void) force_word(PL_oldbufptr + tboffset, WORD,
5289 /* Look for a prototype */
5293 s = scan_str(s,FALSE,FALSE);
5295 Perl_croak(aTHX_ "Prototype not terminated");
5296 /* strip spaces and check for bad characters */
5297 d = SvPVX(PL_lex_stuff);
5300 for (p = d; *p; ++p) {
5303 if (!strchr("$@%*;[]&\\", *p))
5308 if (bad_proto && ckWARN(WARN_SYNTAX))
5309 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5310 "Illegal character in prototype for %"SVf" : %s",
5312 SvCUR_set(PL_lex_stuff, tmp);
5320 if (*s == ':' && s[1] != ':')
5321 PL_expect = attrful;
5322 else if (*s != '{' && key == KEY_sub) {
5324 Perl_croak(aTHX_ "Illegal declaration of anonymous subroutine");
5326 Perl_croak(aTHX_ "Illegal declaration of subroutine %"SVf, PL_subname);
5330 PL_nextval[PL_nexttoke].opval =
5331 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
5332 PL_lex_stuff = Nullsv;
5336 sv_setpv(PL_subname,
5337 PL_curstash ? "__ANON__" : "__ANON__::__ANON__");
5340 (void) force_word(PL_oldbufptr + tboffset, WORD,
5349 LOP(OP_SYSTEM,XREF);
5352 LOP(OP_SYMLINK,XTERM);
5355 LOP(OP_SYSCALL,XTERM);
5358 LOP(OP_SYSOPEN,XTERM);
5361 LOP(OP_SYSSEEK,XTERM);
5364 LOP(OP_SYSREAD,XTERM);
5367 LOP(OP_SYSWRITE,XTERM);
5371 TERM(sublex_start());
5392 LOP(OP_TRUNCATE,XTERM);
5404 yylval.ival = CopLINE(PL_curcop);
5408 yylval.ival = CopLINE(PL_curcop);
5412 LOP(OP_UNLINK,XTERM);
5418 LOP(OP_UNPACK,XTERM);
5421 LOP(OP_UTIME,XTERM);
5427 LOP(OP_UNSHIFT,XTERM);
5430 s = tokenize_use(1, s);
5440 yylval.ival = CopLINE(PL_curcop);
5444 PL_hints |= HINT_BLOCK_SCOPE;
5451 LOP(OP_WAITPID,XTERM);
5460 ctl_l[0] = toCTRL('L');
5462 gv_fetchpv(ctl_l,TRUE, SVt_PV);
5465 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
5470 if (PL_expect == XOPERATOR)
5476 yylval.ival = OP_XOR;
5481 TERM(sublex_start());
5486 #pragma segment Main
5490 S_pending_ident(pTHX)
5493 register I32 tmp = 0;
5494 /* pit holds the identifier we read and pending_ident is reset */
5495 char pit = PL_pending_ident;
5496 PL_pending_ident = 0;
5498 DEBUG_T({ PerlIO_printf(Perl_debug_log,
5499 "### Tokener saw identifier '%s'\n", PL_tokenbuf); });
5501 /* if we're in a my(), we can't allow dynamics here.
5502 $foo'bar has already been turned into $foo::bar, so
5503 just check for colons.
5505 if it's a legal name, the OP is a PADANY.
5508 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
5509 if (strchr(PL_tokenbuf,':'))
5510 yyerror(Perl_form(aTHX_ "No package name allowed for "
5511 "variable %s in \"our\"",
5513 tmp = allocmy(PL_tokenbuf);
5516 if (strchr(PL_tokenbuf,':'))
5517 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
5519 yylval.opval = newOP(OP_PADANY, 0);
5520 yylval.opval->op_targ = allocmy(PL_tokenbuf);
5526 build the ops for accesses to a my() variable.
5528 Deny my($a) or my($b) in a sort block, *if* $a or $b is
5529 then used in a comparison. This catches most, but not
5530 all cases. For instance, it catches
5531 sort { my($a); $a <=> $b }
5533 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
5534 (although why you'd do that is anyone's guess).
5537 if (!strchr(PL_tokenbuf,':')) {
5539 tmp = pad_findmy(PL_tokenbuf);
5540 if (tmp != NOT_IN_PAD) {
5541 /* might be an "our" variable" */
5542 if (PAD_COMPNAME_FLAGS(tmp) & SVpad_OUR) {
5543 /* build ops for a bareword */
5544 HV * const stash = PAD_COMPNAME_OURSTASH(tmp);
5545 HEK * const stashname = HvNAME_HEK(stash);
5546 SV * const sym = newSVhek(stashname);
5547 sv_catpvn(sym, "::", 2);
5548 sv_catpv(sym, PL_tokenbuf+1);
5549 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
5550 yylval.opval->op_private = OPpCONST_ENTERED;
5553 ? (GV_ADDMULTI | GV_ADDINEVAL)
5556 ((PL_tokenbuf[0] == '$') ? SVt_PV
5557 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5562 /* if it's a sort block and they're naming $a or $b */
5563 if (PL_last_lop_op == OP_SORT &&
5564 PL_tokenbuf[0] == '$' &&
5565 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
5568 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
5569 d < PL_bufend && *d != '\n';
5572 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
5573 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
5579 yylval.opval = newOP(OP_PADANY, 0);
5580 yylval.opval->op_targ = tmp;
5586 Whine if they've said @foo in a doublequoted string,
5587 and @foo isn't a variable we can find in the symbol
5590 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
5591 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
5592 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
5593 && ckWARN(WARN_AMBIGUOUS))
5595 /* Downgraded from fatal to warning 20000522 mjd */
5596 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5597 "Possible unintended interpolation of %s in string",
5602 /* build ops for a bareword */
5603 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
5604 yylval.opval->op_private = OPpCONST_ENTERED;
5605 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
5606 ((PL_tokenbuf[0] == '$') ? SVt_PV
5607 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5613 * The following code was generated by perl_keyword.pl.
5617 Perl_keyword (pTHX_ const char *name, I32 len)
5621 case 1: /* 5 tokens of length 1 */
5653 case 2: /* 18 tokens of length 2 */
5799 case 3: /* 28 tokens of length 3 */
5803 if (name[1] == 'N' &&
5866 if (name[1] == 'i' &&
5906 if (name[1] == 'o' &&
5915 if (name[1] == 'e' &&
5924 if (name[1] == 'n' &&
5933 if (name[1] == 'o' &&
5942 if (name[1] == 'a' &&
5951 if (name[1] == 'o' &&
6013 if (name[1] == 'e' &&
6045 if (name[1] == 'i' &&
6054 if (name[1] == 's' &&
6063 if (name[1] == 'e' &&
6072 if (name[1] == 'o' &&
6084 case 4: /* 40 tokens of length 4 */
6088 if (name[1] == 'O' &&
6098 if (name[1] == 'N' &&
6108 if (name[1] == 'i' &&
6118 if (name[1] == 'h' &&
6128 if (name[1] == 'u' &&
6141 if (name[2] == 'c' &&
6150 if (name[2] == 's' &&
6159 if (name[2] == 'a' &&
6195 if (name[1] == 'o' &&
6208 if (name[2] == 't' &&
6217 if (name[2] == 'o' &&
6226 if (name[2] == 't' &&
6235 if (name[2] == 'e' &&
6248 if (name[1] == 'o' &&
6261 if (name[2] == 'y' &&
6270 if (name[2] == 'l' &&
6286 if (name[2] == 's' &&
6295 if (name[2] == 'n' &&
6304 if (name[2] == 'c' &&
6317 if (name[1] == 'e' &&
6327 if (name[1] == 'p' &&
6340 if (name[2] == 'c' &&
6349 if (name[2] == 'p' &&
6358 if (name[2] == 's' &&
6374 if (name[2] == 'n' &&
6444 if (name[2] == 'r' &&
6453 if (name[2] == 'r' &&
6462 if (name[2] == 'a' &&
6478 if (name[2] == 'l' &&
6545 case 5: /* 36 tokens of length 5 */
6549 if (name[1] == 'E' &&
6560 if (name[1] == 'H' &&
6574 if (name[2] == 'a' &&
6584 if (name[2] == 'a' &&
6598 if (name[1] == 'l' &&
6615 if (name[3] == 'i' &&
6624 if (name[3] == 'o' &&
6660 if (name[2] == 'o' &&
6670 if (name[2] == 'y' &&
6684 if (name[1] == 'l' &&
6698 if (name[2] == 'n' &&
6708 if (name[2] == 'o' &&
6725 if (name[2] == 'd' &&
6735 if (name[2] == 'c' &&
6752 if (name[2] == 'c' &&
6762 if (name[2] == 't' &&
6776 if (name[1] == 'k' &&
6787 if (name[1] == 'r' &&
6801 if (name[2] == 's' &&
6811 if (name[2] == 'd' &&
6828 if (name[2] == 'm' &&
6838 if (name[2] == 'i' &&
6848 if (name[2] == 'e' &&
6858 if (name[2] == 'l' &&
6868 if (name[2] == 'a' &&
6878 if (name[2] == 'u' &&
6892 if (name[1] == 'i' &&
6906 if (name[2] == 'a' &&
6919 if (name[3] == 'e' &&
6954 if (name[2] == 'i' &&
6971 if (name[2] == 'i' &&
6981 if (name[2] == 'i' &&
6998 case 6: /* 33 tokens of length 6 */
7002 if (name[1] == 'c' &&
7017 if (name[2] == 'l' &&
7028 if (name[2] == 'r' &&
7043 if (name[1] == 'e' &&
7058 if (name[2] == 's' &&
7063 if(ckWARN_d(WARN_SYNTAX))
7064 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "elseif should be elsif");
7070 if (name[2] == 'i' &&
7088 if (name[2] == 'l' &&
7099 if (name[2] == 'r' &&
7114 if (name[1] == 'm' &&
7129 if (name[2] == 'n' &&
7140 if (name[2] == 's' &&
7155 if (name[1] == 's' &&
7161 if (name[4] == 't' &&
7170 if (name[4] == 'e' &&
7179 if (name[4] == 'c' &&
7188 if (name[4] == 'n' &&
7204 if (name[1] == 'r' &&
7222 if (name[3] == 'a' &&
7232 if (name[3] == 'u' &&
7246 if (name[2] == 'n' &&
7264 if (name[2] == 'a' &&
7278 if (name[3] == 'e' &&
7291 if (name[4] == 't' &&
7300 if (name[4] == 'e' &&
7322 if (name[4] == 't' &&
7331 if (name[4] == 'e' &&
7347 if (name[2] == 'c' &&
7358 if (name[2] == 'l' &&
7369 if (name[2] == 'b' &&
7380 if (name[2] == 's' &&
7403 if (name[4] == 's' &&
7412 if (name[4] == 'n' &&
7425 if (name[3] == 'a' &&
7442 if (name[1] == 'a' &&
7457 case 7: /* 28 tokens of length 7 */
7461 if (name[1] == 'E' &&
7474 if (name[1] == '_' &&
7487 if (name[1] == 'i' &&
7494 return -KEY_binmode;
7500 if (name[1] == 'o' &&
7507 return -KEY_connect;
7516 if (name[2] == 'm' &&
7522 return -KEY_dbmopen;
7528 if (name[2] == 'f' &&
7544 if (name[1] == 'o' &&
7557 if (name[1] == 'e' &&
7564 if (name[5] == 'r' &&
7567 return -KEY_getpgrp;
7573 if (name[5] == 'i' &&
7576 return -KEY_getppid;
7589 if (name[1] == 'c' &&
7596 return -KEY_lcfirst;
7602 if (name[1] == 'p' &&
7609 return -KEY_opendir;
7615 if (name[1] == 'a' &&
7633 if (name[3] == 'd' &&
7638 return -KEY_readdir;
7644 if (name[3] == 'u' &&
7655 if (name[3] == 'e' &&
7660 return -KEY_reverse;
7679 if (name[3] == 'k' &&
7684 return -KEY_seekdir;
7690 if (name[3] == 'p' &&
7695 return -KEY_setpgrp;
7705 if (name[2] == 'm' &&
7711 return -KEY_shmread;
7717 if (name[2] == 'r' &&
7723 return -KEY_sprintf;
7732 if (name[3] == 'l' &&
7737 return -KEY_symlink;
7746 if (name[4] == 'a' &&
7750 return -KEY_syscall;
7756 if (name[4] == 'p' &&
7760 return -KEY_sysopen;
7766 if (name[4] == 'e' &&
7770 return -KEY_sysread;
7776 if (name[4] == 'e' &&
7780 return -KEY_sysseek;
7798 if (name[1] == 'e' &&
7805 return -KEY_telldir;
7814 if (name[2] == 'f' &&
7820 return -KEY_ucfirst;
7826 if (name[2] == 's' &&
7832 return -KEY_unshift;
7842 if (name[1] == 'a' &&
7849 return -KEY_waitpid;
7858 case 8: /* 26 tokens of length 8 */
7862 if (name[1] == 'U' &&
7870 return KEY_AUTOLOAD;
7881 if (name[3] == 'A' &&
7887 return KEY___DATA__;
7893 if (name[3] == 'I' &&
7899 return -KEY___FILE__;
7905 if (name[3] == 'I' &&
7911 return -KEY___LINE__;
7927 if (name[2] == 'o' &&
7934 return -KEY_closedir;
7940 if (name[2] == 'n' &&
7947 return -KEY_continue;
7957 if (name[1] == 'b' &&
7965 return -KEY_dbmclose;
7971 if (name[1] == 'n' &&
7977 if (name[4] == 'r' &&
7982 return -KEY_endgrent;
7988 if (name[4] == 'w' &&
7993 return -KEY_endpwent;
8006 if (name[1] == 'o' &&
8014 return -KEY_formline;
8020 if (name[1] == 'e' &&
8031 if (name[6] == 'n' &&
8034 return -KEY_getgrent;
8040 if (name[6] == 'i' &&
8043 return -KEY_getgrgid;
8049 if (name[6] == 'a' &&
8052 return -KEY_getgrnam;
8065 if (name[4] == 'o' &&
8070 return -KEY_getlogin;
8081 if (name[6] == 'n' &&
8084 return -KEY_getpwent;
8090 if (name[6] == 'a' &&
8093 return -KEY_getpwnam;
8099 if (name[6] == 'i' &&
8102 return -KEY_getpwuid;
8122 if (name[1] == 'e' &&
8129 if (name[5] == 'i' &&
8136 return -KEY_readline;
8141 return -KEY_readlink;
8152 if (name[5] == 'i' &&
8156 return -KEY_readpipe;
8177 if (name[4] == 'r' &&
8182 return -KEY_setgrent;
8188 if (name[4] == 'w' &&
8193 return -KEY_setpwent;
8209 if (name[3] == 'w' &&
8215 return -KEY_shmwrite;
8221 if (name[3] == 't' &&
8227 return -KEY_shutdown;
8237 if (name[2] == 's' &&
8244 return -KEY_syswrite;
8254 if (name[1] == 'r' &&
8262 return -KEY_truncate;
8271 case 9: /* 8 tokens of length 9 */
8275 if (name[1] == 'n' &&
8284 return -KEY_endnetent;
8290 if (name[1] == 'e' &&
8299 return -KEY_getnetent;
8305 if (name[1] == 'o' &&
8314 return -KEY_localtime;
8320 if (name[1] == 'r' &&
8329 return KEY_prototype;
8335 if (name[1] == 'u' &&
8344 return -KEY_quotemeta;
8350 if (name[1] == 'e' &&
8359 return -KEY_rewinddir;
8365 if (name[1] == 'e' &&
8374 return -KEY_setnetent;
8380 if (name[1] == 'a' &&
8389 return -KEY_wantarray;
8398 case 10: /* 9 tokens of length 10 */
8402 if (name[1] == 'n' &&
8408 if (name[4] == 'o' &&
8415 return -KEY_endhostent;
8421 if (name[4] == 'e' &&
8428 return -KEY_endservent;
8441 if (name[1] == 'e' &&
8447 if (name[4] == 'o' &&
8454 return -KEY_gethostent;
8463 if (name[5] == 'r' &&
8469 return -KEY_getservent;
8475 if (name[5] == 'c' &&
8481 return -KEY_getsockopt;
8506 if (name[4] == 'o' &&
8513 return -KEY_sethostent;
8522 if (name[5] == 'r' &&
8528 return -KEY_setservent;
8534 if (name[5] == 'c' &&
8540 return -KEY_setsockopt;
8557 if (name[2] == 'c' &&
8566 return -KEY_socketpair;
8579 case 11: /* 8 tokens of length 11 */
8583 if (name[1] == '_' &&
8594 return -KEY___PACKAGE__;
8600 if (name[1] == 'n' &&
8611 return -KEY_endprotoent;
8617 if (name[1] == 'e' &&
8626 if (name[5] == 'e' &&
8633 return -KEY_getpeername;
8642 if (name[6] == 'o' &&
8648 return -KEY_getpriority;
8654 if (name[6] == 't' &&
8660 return -KEY_getprotoent;
8674 if (name[4] == 'o' &&
8682 return -KEY_getsockname;
8695 if (name[1] == 'e' &&
8703 if (name[6] == 'o' &&
8709 return -KEY_setpriority;
8715 if (name[6] == 't' &&
8721 return -KEY_setprotoent;
8737 case 12: /* 2 tokens of length 12 */
8738 if (name[0] == 'g' &&
8750 if (name[9] == 'd' &&
8753 { /* getnetbyaddr */
8754 return -KEY_getnetbyaddr;
8760 if (name[9] == 'a' &&
8763 { /* getnetbyname */
8764 return -KEY_getnetbyname;
8776 case 13: /* 4 tokens of length 13 */
8777 if (name[0] == 'g' &&
8784 if (name[4] == 'o' &&
8793 if (name[10] == 'd' &&
8796 { /* gethostbyaddr */
8797 return -KEY_gethostbyaddr;
8803 if (name[10] == 'a' &&
8806 { /* gethostbyname */
8807 return -KEY_gethostbyname;
8820 if (name[4] == 'e' &&
8829 if (name[10] == 'a' &&
8832 { /* getservbyname */
8833 return -KEY_getservbyname;
8839 if (name[10] == 'o' &&
8842 { /* getservbyport */
8843 return -KEY_getservbyport;
8862 case 14: /* 1 tokens of length 14 */
8863 if (name[0] == 'g' &&
8877 { /* getprotobyname */
8878 return -KEY_getprotobyname;
8883 case 16: /* 1 tokens of length 16 */
8884 if (name[0] == 'g' &&
8900 { /* getprotobynumber */
8901 return -KEY_getprotobynumber;
8915 S_checkcomma(pTHX_ register char *s, const char *name, const char *what)
8919 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
8920 if (ckWARN(WARN_SYNTAX)) {
8922 for (w = s+2; *w && level; w++) {
8929 for (; *w && isSPACE(*w); w++) ;
8930 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
8931 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
8932 "%s (...) interpreted as function",name);
8935 while (s < PL_bufend && isSPACE(*s))
8939 while (s < PL_bufend && isSPACE(*s))
8941 if (isIDFIRST_lazy_if(s,UTF)) {
8943 while (isALNUM_lazy_if(s,UTF))
8945 while (s < PL_bufend && isSPACE(*s))
8949 *s = '\0'; /* XXX If we didn't do this, we could const a lot of toke.c */
8950 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
8954 Perl_croak(aTHX_ "No comma allowed after %s", what);
8959 /* Either returns sv, or mortalizes sv and returns a new SV*.
8960 Best used as sv=new_constant(..., sv, ...).
8961 If s, pv are NULL, calls subroutine with one argument,
8962 and type is used with error messages only. */
8965 S_new_constant(pTHX_ const char *s, STRLEN len, const char *key, SV *sv, SV *pv,
8969 HV * const table = GvHV(PL_hintgv); /* ^H */
8973 const char *why1 = "", *why2 = "", *why3 = "";
8975 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
8978 why2 = strEQ(key,"charnames")
8979 ? "(possibly a missing \"use charnames ...\")"
8981 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
8982 (type ? type: "undef"), why2);
8984 /* This is convoluted and evil ("goto considered harmful")
8985 * but I do not understand the intricacies of all the different
8986 * failure modes of %^H in here. The goal here is to make
8987 * the most probable error message user-friendly. --jhi */
8992 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
8993 (type ? type: "undef"), why1, why2, why3);
8995 yyerror(SvPVX_const(msg));
8999 cvp = hv_fetch(table, key, strlen(key), FALSE);
9000 if (!cvp || !SvOK(*cvp)) {
9003 why3 = "} is not defined";
9006 sv_2mortal(sv); /* Parent created it permanently */
9009 pv = sv_2mortal(newSVpvn(s, len));
9011 typesv = sv_2mortal(newSVpv(type, 0));
9013 typesv = &PL_sv_undef;
9015 PUSHSTACKi(PERLSI_OVERLOAD);
9027 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
9031 /* Check the eval first */
9032 if (!PL_in_eval && SvTRUE(ERRSV)) {
9033 sv_catpv(ERRSV, "Propagated");
9034 yyerror(SvPV_nolen_const(ERRSV)); /* Duplicates the message inside eval */
9036 res = SvREFCNT_inc(sv);
9040 (void)SvREFCNT_inc(res);
9049 why1 = "Call to &{$^H{";
9051 why3 = "}} did not return a defined value";
9059 /* Returns a NUL terminated string, with the length of the string written to
9063 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
9065 register char *d = dest;
9066 register char * const e = d + destlen - 3; /* two-character token, ending NUL */
9069 Perl_croak(aTHX_ ident_too_long);
9070 if (isALNUM(*s)) /* UTF handled below */
9072 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
9077 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
9081 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
9082 char *t = s + UTF8SKIP(s);
9083 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
9085 if (d + (t - s) > e)
9086 Perl_croak(aTHX_ ident_too_long);
9087 Copy(s, d, t - s, char);
9100 S_scan_ident(pTHX_ register char *s, register const char *send, char *dest, STRLEN destlen, I32 ck_uni)
9104 char *bracket = Nullch;
9110 e = d + destlen - 3; /* two-character token, ending NUL */
9112 while (isDIGIT(*s)) {
9114 Perl_croak(aTHX_ ident_too_long);
9121 Perl_croak(aTHX_ ident_too_long);
9122 if (isALNUM(*s)) /* UTF handled below */
9124 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
9129 else if (*s == ':' && s[1] == ':') {
9133 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
9134 char *t = s + UTF8SKIP(s);
9135 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
9137 if (d + (t - s) > e)
9138 Perl_croak(aTHX_ ident_too_long);
9139 Copy(s, d, t - s, char);
9150 if (PL_lex_state != LEX_NORMAL)
9151 PL_lex_state = LEX_INTERPENDMAYBE;
9154 if (*s == '$' && s[1] &&
9155 (isALNUM_lazy_if(s+1,UTF) || s[1] == '$' || s[1] == '{' || strnEQ(s+1,"::",2)) )
9168 if (*d == '^' && *s && isCONTROLVAR(*s)) {
9173 if (isSPACE(s[-1])) {
9175 const char ch = *s++;
9176 if (!SPACE_OR_TAB(ch)) {
9182 if (isIDFIRST_lazy_if(d,UTF)) {
9186 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
9188 while (e < send && UTF8_IS_CONTINUED(*e) && is_utf8_mark((U8*)e))
9191 Copy(s, d, e - s, char);
9196 while ((isALNUM(*s) || *s == ':') && d < e)
9199 Perl_croak(aTHX_ ident_too_long);
9202 while (s < send && SPACE_OR_TAB(*s)) s++;
9203 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
9204 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
9205 const char *brack = *s == '[' ? "[...]" : "{...}";
9206 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
9207 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
9208 funny, dest, brack, funny, dest, brack);
9211 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
9215 /* Handle extended ${^Foo} variables
9216 * 1999-02-27 mjd-perl-patch@plover.com */
9217 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
9221 while (isALNUM(*s) && d < e) {
9225 Perl_croak(aTHX_ ident_too_long);
9230 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets) {
9231 PL_lex_state = LEX_INTERPEND;
9236 if (PL_lex_state == LEX_NORMAL) {
9237 if (ckWARN(WARN_AMBIGUOUS) &&
9238 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
9240 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
9241 "Ambiguous use of %c{%s} resolved to %c%s",
9242 funny, dest, funny, dest);
9247 s = bracket; /* let the parser handle it */
9251 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
9252 PL_lex_state = LEX_INTERPEND;
9257 Perl_pmflag(pTHX_ U32* pmfl, int ch)
9262 *pmfl |= PMf_GLOBAL;
9264 *pmfl |= PMf_CONTINUE;
9268 *pmfl |= PMf_MULTILINE;
9270 *pmfl |= PMf_SINGLELINE;
9272 *pmfl |= PMf_EXTENDED;
9276 S_scan_pat(pTHX_ char *start, I32 type)
9279 char *s = scan_str(start,FALSE,FALSE);
9282 char * const delimiter = skipspace(start);
9283 Perl_croak(aTHX_ *delimiter == '?'
9284 ? "Search pattern not terminated or ternary operator parsed as search pattern"
9285 : "Search pattern not terminated" );
9288 pm = (PMOP*)newPMOP(type, 0);
9289 if (PL_multi_open == '?')
9290 pm->op_pmflags |= PMf_ONCE;
9292 while (*s && strchr("iomsx", *s))
9293 pmflag(&pm->op_pmflags,*s++);
9296 while (*s && strchr("iogcmsx", *s))
9297 pmflag(&pm->op_pmflags,*s++);
9299 /* issue a warning if /c is specified,but /g is not */
9300 if ((pm->op_pmflags & PMf_CONTINUE) && !(pm->op_pmflags & PMf_GLOBAL)
9301 && ckWARN(WARN_REGEXP))
9303 Perl_warner(aTHX_ packWARN(WARN_REGEXP), c_without_g);
9306 pm->op_pmpermflags = pm->op_pmflags;
9308 PL_lex_op = (OP*)pm;
9309 yylval.ival = OP_MATCH;
9314 S_scan_subst(pTHX_ char *start)
9322 yylval.ival = OP_NULL;
9324 s = scan_str(start,FALSE,FALSE);
9327 Perl_croak(aTHX_ "Substitution pattern not terminated");
9329 if (s[-1] == PL_multi_open)
9332 first_start = PL_multi_start;
9333 s = scan_str(s,FALSE,FALSE);
9336 SvREFCNT_dec(PL_lex_stuff);
9337 PL_lex_stuff = Nullsv;
9339 Perl_croak(aTHX_ "Substitution replacement not terminated");
9341 PL_multi_start = first_start; /* so whole substitution is taken together */
9343 pm = (PMOP*)newPMOP(OP_SUBST, 0);
9349 else if (strchr("iogcmsx", *s))
9350 pmflag(&pm->op_pmflags,*s++);
9355 /* /c is not meaningful with s/// */
9356 if ((pm->op_pmflags & PMf_CONTINUE) && ckWARN(WARN_REGEXP))
9358 Perl_warner(aTHX_ packWARN(WARN_REGEXP), c_in_subst);
9363 PL_sublex_info.super_bufptr = s;
9364 PL_sublex_info.super_bufend = PL_bufend;
9366 pm->op_pmflags |= PMf_EVAL;
9367 repl = newSVpvn("",0);
9369 sv_catpv(repl, es ? "eval " : "do ");
9370 sv_catpvn(repl, "{ ", 2);
9371 sv_catsv(repl, PL_lex_repl);
9372 sv_catpvn(repl, " };", 2);
9374 SvREFCNT_dec(PL_lex_repl);
9378 pm->op_pmpermflags = pm->op_pmflags;
9379 PL_lex_op = (OP*)pm;
9380 yylval.ival = OP_SUBST;
9385 S_scan_trans(pTHX_ char *start)
9394 yylval.ival = OP_NULL;
9396 s = scan_str(start,FALSE,FALSE);
9398 Perl_croak(aTHX_ "Transliteration pattern not terminated");
9399 if (s[-1] == PL_multi_open)
9402 s = scan_str(s,FALSE,FALSE);
9405 SvREFCNT_dec(PL_lex_stuff);
9406 PL_lex_stuff = Nullsv;
9408 Perl_croak(aTHX_ "Transliteration replacement not terminated");
9411 complement = del = squash = 0;
9415 complement = OPpTRANS_COMPLEMENT;
9418 del = OPpTRANS_DELETE;
9421 squash = OPpTRANS_SQUASH;
9430 Newx(tbl, complement&&!del?258:256, short);
9431 o = newPVOP(OP_TRANS, 0, (char*)tbl);
9432 o->op_private &= ~OPpTRANS_ALL;
9433 o->op_private |= del|squash|complement|
9434 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
9435 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
9438 yylval.ival = OP_TRANS;
9443 S_scan_heredoc(pTHX_ register char *s)
9446 I32 op_type = OP_SCALAR;
9450 const char newline[] = "\n";
9451 const char *found_newline;
9455 const int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
9459 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
9462 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
9463 if (*peek == '`' || *peek == '\'' || *peek =='"') {
9466 s = delimcpy(d, e, s, PL_bufend, term, &len);
9476 if (!isALNUM_lazy_if(s,UTF))
9477 deprecate_old("bare << to mean <<\"\"");
9478 for (; isALNUM_lazy_if(s,UTF); s++) {
9483 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
9484 Perl_croak(aTHX_ "Delimiter for here document is too long");
9487 len = d - PL_tokenbuf;
9488 #ifndef PERL_STRICT_CR
9489 d = strchr(s, '\r');
9491 char * const olds = s;
9493 while (s < PL_bufend) {
9499 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
9508 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
9512 if ( outer || !(found_newline = ninstr(s,PL_bufend,newline,newline+1)) ) {
9513 herewas = newSVpvn(s,PL_bufend-s);
9517 herewas = newSVpvn(s,found_newline-s);
9519 s += SvCUR(herewas);
9521 tmpstr = NEWSV(87,79);
9522 sv_upgrade(tmpstr, SVt_PVIV);
9525 SvIV_set(tmpstr, -1);
9527 else if (term == '`') {
9528 op_type = OP_BACKTICK;
9529 SvIV_set(tmpstr, '\\');
9533 PL_multi_start = CopLINE(PL_curcop);
9534 PL_multi_open = PL_multi_close = '<';
9535 term = *PL_tokenbuf;
9536 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
9537 char *bufptr = PL_sublex_info.super_bufptr;
9538 char *bufend = PL_sublex_info.super_bufend;
9539 char * const olds = s - SvCUR(herewas);
9540 s = strchr(bufptr, '\n');
9544 while (s < bufend &&
9545 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
9547 CopLINE_inc(PL_curcop);
9550 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
9551 missingterm(PL_tokenbuf);
9553 sv_setpvn(herewas,bufptr,d-bufptr+1);
9554 sv_setpvn(tmpstr,d+1,s-d);
9556 sv_catpvn(herewas,s,bufend-s);
9557 Copy(SvPVX_const(herewas),bufptr,SvCUR(herewas) + 1,char);
9564 while (s < PL_bufend &&
9565 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
9567 CopLINE_inc(PL_curcop);
9569 if (s >= PL_bufend) {
9570 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
9571 missingterm(PL_tokenbuf);
9573 sv_setpvn(tmpstr,d+1,s-d);
9575 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
9577 sv_catpvn(herewas,s,PL_bufend-s);
9578 sv_setsv(PL_linestr,herewas);
9579 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
9580 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
9581 PL_last_lop = PL_last_uni = Nullch;
9584 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
9585 while (s >= PL_bufend) { /* multiple line string? */
9587 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
9588 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
9589 missingterm(PL_tokenbuf);
9591 CopLINE_inc(PL_curcop);
9592 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
9593 PL_last_lop = PL_last_uni = Nullch;
9594 #ifndef PERL_STRICT_CR
9595 if (PL_bufend - PL_linestart >= 2) {
9596 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
9597 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
9599 PL_bufend[-2] = '\n';
9601 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
9603 else if (PL_bufend[-1] == '\r')
9604 PL_bufend[-1] = '\n';
9606 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
9607 PL_bufend[-1] = '\n';
9609 if (PERLDB_LINE && PL_curstash != PL_debstash) {
9610 SV *sv = NEWSV(88,0);
9612 sv_upgrade(sv, SVt_PVMG);
9613 sv_setsv(sv,PL_linestr);
9616 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
9618 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
9619 STRLEN off = PL_bufend - 1 - SvPVX_const(PL_linestr);
9620 *(SvPVX(PL_linestr) + off ) = ' ';
9621 sv_catsv(PL_linestr,herewas);
9622 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
9623 s = SvPVX(PL_linestr) + off; /* In case PV of PL_linestr moved. */
9627 sv_catsv(tmpstr,PL_linestr);
9632 PL_multi_end = CopLINE(PL_curcop);
9633 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
9634 SvPV_shrink_to_cur(tmpstr);
9636 SvREFCNT_dec(herewas);
9638 if (UTF && is_utf8_string((U8*)SvPVX_const(tmpstr), SvCUR(tmpstr)))
9640 else if (PL_encoding)
9641 sv_recode_to_utf8(tmpstr, PL_encoding);
9643 PL_lex_stuff = tmpstr;
9644 yylval.ival = op_type;
9649 takes: current position in input buffer
9650 returns: new position in input buffer
9651 side-effects: yylval and lex_op are set.
9656 <FH> read from filehandle
9657 <pkg::FH> read from package qualified filehandle
9658 <pkg'FH> read from package qualified filehandle
9659 <$fh> read from filehandle in $fh
9665 S_scan_inputsymbol(pTHX_ char *start)
9667 register char *s = start; /* current position in buffer */
9673 d = PL_tokenbuf; /* start of temp holding space */
9674 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
9675 end = strchr(s, '\n');
9678 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
9680 /* die if we didn't have space for the contents of the <>,
9681 or if it didn't end, or if we see a newline
9684 if (len >= sizeof PL_tokenbuf)
9685 Perl_croak(aTHX_ "Excessively long <> operator");
9687 Perl_croak(aTHX_ "Unterminated <> operator");
9692 Remember, only scalar variables are interpreted as filehandles by
9693 this code. Anything more complex (e.g., <$fh{$num}>) will be
9694 treated as a glob() call.
9695 This code makes use of the fact that except for the $ at the front,
9696 a scalar variable and a filehandle look the same.
9698 if (*d == '$' && d[1]) d++;
9700 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
9701 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
9704 /* If we've tried to read what we allow filehandles to look like, and
9705 there's still text left, then it must be a glob() and not a getline.
9706 Use scan_str to pull out the stuff between the <> and treat it
9707 as nothing more than a string.
9710 if (d - PL_tokenbuf != len) {
9711 yylval.ival = OP_GLOB;
9713 s = scan_str(start,FALSE,FALSE);
9715 Perl_croak(aTHX_ "Glob not terminated");
9719 bool readline_overriden = FALSE;
9720 GV *gv_readline = Nullgv;
9722 /* we're in a filehandle read situation */
9725 /* turn <> into <ARGV> */
9727 Copy("ARGV",d,5,char);
9729 /* Check whether readline() is overriden */
9730 if (((gv_readline = gv_fetchpv("readline", FALSE, SVt_PVCV))
9731 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline))
9733 ((gvp = (GV**)hv_fetch(PL_globalstash, "readline", 8, FALSE))
9734 && (gv_readline = *gvp) != (GV*)&PL_sv_undef
9735 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline)))
9736 readline_overriden = TRUE;
9738 /* if <$fh>, create the ops to turn the variable into a
9744 /* try to find it in the pad for this block, otherwise find
9745 add symbol table ops
9747 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
9748 if (PAD_COMPNAME_FLAGS(tmp) & SVpad_OUR) {
9749 HV *stash = PAD_COMPNAME_OURSTASH(tmp);
9750 HEK *stashname = HvNAME_HEK(stash);
9751 SV *sym = sv_2mortal(newSVhek(stashname));
9752 sv_catpvn(sym, "::", 2);
9758 OP *o = newOP(OP_PADSV, 0);
9760 PL_lex_op = readline_overriden
9761 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
9762 append_elem(OP_LIST, o,
9763 newCVREF(0, newGVOP(OP_GV,0,gv_readline))))
9764 : (OP*)newUNOP(OP_READLINE, 0, o);
9773 ? (GV_ADDMULTI | GV_ADDINEVAL)
9776 PL_lex_op = readline_overriden
9777 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
9778 append_elem(OP_LIST,
9779 newUNOP(OP_RV2SV, 0, newGVOP(OP_GV, 0, gv)),
9780 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
9781 : (OP*)newUNOP(OP_READLINE, 0,
9782 newUNOP(OP_RV2SV, 0,
9783 newGVOP(OP_GV, 0, gv)));
9785 if (!readline_overriden)
9786 PL_lex_op->op_flags |= OPf_SPECIAL;
9787 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
9788 yylval.ival = OP_NULL;
9791 /* If it's none of the above, it must be a literal filehandle
9792 (<Foo::BAR> or <FOO>) so build a simple readline OP */
9794 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
9795 PL_lex_op = readline_overriden
9796 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
9797 append_elem(OP_LIST,
9798 newGVOP(OP_GV, 0, gv),
9799 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
9800 : (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
9801 yylval.ival = OP_NULL;
9810 takes: start position in buffer
9811 keep_quoted preserve \ on the embedded delimiter(s)
9812 keep_delims preserve the delimiters around the string
9813 returns: position to continue reading from buffer
9814 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
9815 updates the read buffer.
9817 This subroutine pulls a string out of the input. It is called for:
9818 q single quotes q(literal text)
9819 ' single quotes 'literal text'
9820 qq double quotes qq(interpolate $here please)
9821 " double quotes "interpolate $here please"
9822 qx backticks qx(/bin/ls -l)
9823 ` backticks `/bin/ls -l`
9824 qw quote words @EXPORT_OK = qw( func() $spam )
9825 m// regexp match m/this/
9826 s/// regexp substitute s/this/that/
9827 tr/// string transliterate tr/this/that/
9828 y/// string transliterate y/this/that/
9829 ($*@) sub prototypes sub foo ($)
9830 (stuff) sub attr parameters sub foo : attr(stuff)
9831 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
9833 In most of these cases (all but <>, patterns and transliterate)
9834 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
9835 calls scan_str(). s/// makes yylex() call scan_subst() which calls
9836 scan_str(). tr/// and y/// make yylex() call scan_trans() which
9839 It skips whitespace before the string starts, and treats the first
9840 character as the delimiter. If the delimiter is one of ([{< then
9841 the corresponding "close" character )]}> is used as the closing
9842 delimiter. It allows quoting of delimiters, and if the string has
9843 balanced delimiters ([{<>}]) it allows nesting.
9845 On success, the SV with the resulting string is put into lex_stuff or,
9846 if that is already non-NULL, into lex_repl. The second case occurs only
9847 when parsing the RHS of the special constructs s/// and tr/// (y///).
9848 For convenience, the terminating delimiter character is stuffed into
9853 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
9855 SV *sv; /* scalar value: string */
9856 char *tmps; /* temp string, used for delimiter matching */
9857 register char *s = start; /* current position in the buffer */
9858 register char term; /* terminating character */
9859 register char *to; /* current position in the sv's data */
9860 I32 brackets = 1; /* bracket nesting level */
9861 bool has_utf8 = FALSE; /* is there any utf8 content? */
9862 I32 termcode; /* terminating char. code */
9863 U8 termstr[UTF8_MAXBYTES]; /* terminating string */
9864 STRLEN termlen; /* length of terminating string */
9865 char *last = NULL; /* last position for nesting bracket */
9867 /* skip space before the delimiter */
9871 /* mark where we are, in case we need to report errors */
9874 /* after skipping whitespace, the next character is the terminator */
9877 termcode = termstr[0] = term;
9881 termcode = utf8_to_uvchr((U8*)s, &termlen);
9882 Copy(s, termstr, termlen, U8);
9883 if (!UTF8_IS_INVARIANT(term))
9887 /* mark where we are */
9888 PL_multi_start = CopLINE(PL_curcop);
9889 PL_multi_open = term;
9891 /* find corresponding closing delimiter */
9892 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
9893 termcode = termstr[0] = term = tmps[5];
9895 PL_multi_close = term;
9897 /* create a new SV to hold the contents. 87 is leak category, I'm
9898 assuming. 79 is the SV's initial length. What a random number. */
9900 sv_upgrade(sv, SVt_PVIV);
9901 SvIV_set(sv, termcode);
9902 (void)SvPOK_only(sv); /* validate pointer */
9904 /* move past delimiter and try to read a complete string */
9906 sv_catpvn(sv, s, termlen);
9909 if (PL_encoding && !UTF) {
9913 int offset = s - SvPVX_const(PL_linestr);
9914 const bool found = sv_cat_decode(sv, PL_encoding, PL_linestr,
9915 &offset, (char*)termstr, termlen);
9916 const char *ns = SvPVX_const(PL_linestr) + offset;
9917 char *svlast = SvEND(sv) - 1;
9919 for (; s < ns; s++) {
9920 if (*s == '\n' && !PL_rsfp)
9921 CopLINE_inc(PL_curcop);
9924 goto read_more_line;
9926 /* handle quoted delimiters */
9927 if (SvCUR(sv) > 1 && *(svlast-1) == '\\') {
9929 for (t = svlast-2; t >= SvPVX_const(sv) && *t == '\\';)
9931 if ((svlast-1 - t) % 2) {
9935 SvCUR_set(sv, SvCUR(sv) - 1);
9940 if (PL_multi_open == PL_multi_close) {
9948 for (t = w = last; t < svlast; w++, t++) {
9949 /* At here, all closes are "was quoted" one,
9950 so we don't check PL_multi_close. */
9952 if (!keep_quoted && *(t+1) == PL_multi_open)
9957 else if (*t == PL_multi_open)
9965 SvCUR_set(sv, w - SvPVX_const(sv));
9968 if (--brackets <= 0)
9974 SvCUR_set(sv, SvCUR(sv) - 1);
9980 /* extend sv if need be */
9981 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
9982 /* set 'to' to the next character in the sv's string */
9983 to = SvPVX(sv)+SvCUR(sv);
9985 /* if open delimiter is the close delimiter read unbridle */
9986 if (PL_multi_open == PL_multi_close) {
9987 for (; s < PL_bufend; s++,to++) {
9988 /* embedded newlines increment the current line number */
9989 if (*s == '\n' && !PL_rsfp)
9990 CopLINE_inc(PL_curcop);
9991 /* handle quoted delimiters */
9992 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
9993 if (!keep_quoted && s[1] == term)
9995 /* any other quotes are simply copied straight through */
9999 /* terminate when run out of buffer (the for() condition), or
10000 have found the terminator */
10001 else if (*s == term) {
10004 if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen))
10007 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
10013 /* if the terminator isn't the same as the start character (e.g.,
10014 matched brackets), we have to allow more in the quoting, and
10015 be prepared for nested brackets.
10018 /* read until we run out of string, or we find the terminator */
10019 for (; s < PL_bufend; s++,to++) {
10020 /* embedded newlines increment the line count */
10021 if (*s == '\n' && !PL_rsfp)
10022 CopLINE_inc(PL_curcop);
10023 /* backslashes can escape the open or closing characters */
10024 if (*s == '\\' && s+1 < PL_bufend) {
10025 if (!keep_quoted &&
10026 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
10031 /* allow nested opens and closes */
10032 else if (*s == PL_multi_close && --brackets <= 0)
10034 else if (*s == PL_multi_open)
10036 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
10041 /* terminate the copied string and update the sv's end-of-string */
10043 SvCUR_set(sv, to - SvPVX_const(sv));
10046 * this next chunk reads more into the buffer if we're not done yet
10050 break; /* handle case where we are done yet :-) */
10052 #ifndef PERL_STRICT_CR
10053 if (to - SvPVX_const(sv) >= 2) {
10054 if ((to[-2] == '\r' && to[-1] == '\n') ||
10055 (to[-2] == '\n' && to[-1] == '\r'))
10059 SvCUR_set(sv, to - SvPVX_const(sv));
10061 else if (to[-1] == '\r')
10064 else if (to - SvPVX_const(sv) == 1 && to[-1] == '\r')
10069 /* if we're out of file, or a read fails, bail and reset the current
10070 line marker so we can report where the unterminated string began
10073 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
10075 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
10078 /* we read a line, so increment our line counter */
10079 CopLINE_inc(PL_curcop);
10081 /* update debugger info */
10082 if (PERLDB_LINE && PL_curstash != PL_debstash) {
10083 SV *sv = NEWSV(88,0);
10085 sv_upgrade(sv, SVt_PVMG);
10086 sv_setsv(sv,PL_linestr);
10087 (void)SvIOK_on(sv);
10089 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
10092 /* having changed the buffer, we must update PL_bufend */
10093 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
10094 PL_last_lop = PL_last_uni = Nullch;
10097 /* at this point, we have successfully read the delimited string */
10099 if (!PL_encoding || UTF) {
10101 sv_catpvn(sv, s, termlen);
10104 if (has_utf8 || PL_encoding)
10107 PL_multi_end = CopLINE(PL_curcop);
10109 /* if we allocated too much space, give some back */
10110 if (SvCUR(sv) + 5 < SvLEN(sv)) {
10111 SvLEN_set(sv, SvCUR(sv) + 1);
10112 SvPV_renew(sv, SvLEN(sv));
10115 /* decide whether this is the first or second quoted string we've read
10128 takes: pointer to position in buffer
10129 returns: pointer to new position in buffer
10130 side-effects: builds ops for the constant in yylval.op
10132 Read a number in any of the formats that Perl accepts:
10134 \d(_?\d)*(\.(\d(_?\d)*)?)?[Ee][\+\-]?(\d(_?\d)*) 12 12.34 12.
10135 \.\d(_?\d)*[Ee][\+\-]?(\d(_?\d)*) .34
10138 0x[0-9A-Fa-f](_?[0-9A-Fa-f])*
10140 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
10143 If it reads a number without a decimal point or an exponent, it will
10144 try converting the number to an integer and see if it can do so
10145 without loss of precision.
10149 Perl_scan_num(pTHX_ const char *start, YYSTYPE* lvalp)
10151 register const char *s = start; /* current position in buffer */
10152 register char *d; /* destination in temp buffer */
10153 register char *e; /* end of temp buffer */
10154 NV nv; /* number read, as a double */
10155 SV *sv = Nullsv; /* place to put the converted number */
10156 bool floatit; /* boolean: int or float? */
10157 const char *lastub = 0; /* position of last underbar */
10158 static char const number_too_long[] = "Number too long";
10160 /* We use the first character to decide what type of number this is */
10164 Perl_croak(aTHX_ "panic: scan_num");
10166 /* if it starts with a 0, it could be an octal number, a decimal in
10167 0.13 disguise, or a hexadecimal number, or a binary number. */
10171 u holds the "number so far"
10172 shift the power of 2 of the base
10173 (hex == 4, octal == 3, binary == 1)
10174 overflowed was the number more than we can hold?
10176 Shift is used when we add a digit. It also serves as an "are
10177 we in octal/hex/binary?" indicator to disallow hex characters
10178 when in octal mode.
10183 bool overflowed = FALSE;
10184 bool just_zero = TRUE; /* just plain 0 or binary number? */
10185 static const NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
10186 static const char* const bases[5] =
10187 { "", "binary", "", "octal", "hexadecimal" };
10188 static const char* const Bases[5] =
10189 { "", "Binary", "", "Octal", "Hexadecimal" };
10190 static const char* const maxima[5] =
10192 "0b11111111111111111111111111111111",
10196 const char *base, *Base, *max;
10198 /* check for hex */
10203 } else if (s[1] == 'b') {
10208 /* check for a decimal in disguise */
10209 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
10211 /* so it must be octal */
10218 if (ckWARN(WARN_SYNTAX))
10219 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10220 "Misplaced _ in number");
10224 base = bases[shift];
10225 Base = Bases[shift];
10226 max = maxima[shift];
10228 /* read the rest of the number */
10230 /* x is used in the overflow test,
10231 b is the digit we're adding on. */
10236 /* if we don't mention it, we're done */
10240 /* _ are ignored -- but warned about if consecutive */
10242 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
10243 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10244 "Misplaced _ in number");
10248 /* 8 and 9 are not octal */
10249 case '8': case '9':
10251 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
10255 case '2': case '3': case '4':
10256 case '5': case '6': case '7':
10258 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
10261 case '0': case '1':
10262 b = *s++ & 15; /* ASCII digit -> value of digit */
10266 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
10267 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
10268 /* make sure they said 0x */
10271 b = (*s++ & 7) + 9;
10273 /* Prepare to put the digit we have onto the end
10274 of the number so far. We check for overflows.
10280 x = u << shift; /* make room for the digit */
10282 if ((x >> shift) != u
10283 && !(PL_hints & HINT_NEW_BINARY)) {
10286 if (ckWARN_d(WARN_OVERFLOW))
10287 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
10288 "Integer overflow in %s number",
10291 u = x | b; /* add the digit to the end */
10294 n *= nvshift[shift];
10295 /* If an NV has not enough bits in its
10296 * mantissa to represent an UV this summing of
10297 * small low-order numbers is a waste of time
10298 * (because the NV cannot preserve the
10299 * low-order bits anyway): we could just
10300 * remember when did we overflow and in the
10301 * end just multiply n by the right
10309 /* if we get here, we had success: make a scalar value from
10314 /* final misplaced underbar check */
10315 if (s[-1] == '_') {
10316 if (ckWARN(WARN_SYNTAX))
10317 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
10322 if (n > 4294967295.0 && ckWARN(WARN_PORTABLE))
10323 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
10324 "%s number > %s non-portable",
10330 if (u > 0xffffffff && ckWARN(WARN_PORTABLE))
10331 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
10332 "%s number > %s non-portable",
10337 if (just_zero && (PL_hints & HINT_NEW_INTEGER))
10338 sv = new_constant(start, s - start, "integer",
10340 else if (PL_hints & HINT_NEW_BINARY)
10341 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
10346 handle decimal numbers.
10347 we're also sent here when we read a 0 as the first digit
10349 case '1': case '2': case '3': case '4': case '5':
10350 case '6': case '7': case '8': case '9': case '.':
10353 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
10356 /* read next group of digits and _ and copy into d */
10357 while (isDIGIT(*s) || *s == '_') {
10358 /* skip underscores, checking for misplaced ones
10362 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
10363 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10364 "Misplaced _ in number");
10368 /* check for end of fixed-length buffer */
10370 Perl_croak(aTHX_ number_too_long);
10371 /* if we're ok, copy the character */
10376 /* final misplaced underbar check */
10377 if (lastub && s == lastub + 1) {
10378 if (ckWARN(WARN_SYNTAX))
10379 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
10382 /* read a decimal portion if there is one. avoid
10383 3..5 being interpreted as the number 3. followed
10386 if (*s == '.' && s[1] != '.') {
10391 if (ckWARN(WARN_SYNTAX))
10392 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10393 "Misplaced _ in number");
10397 /* copy, ignoring underbars, until we run out of digits.
10399 for (; isDIGIT(*s) || *s == '_'; s++) {
10400 /* fixed length buffer check */
10402 Perl_croak(aTHX_ number_too_long);
10404 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
10405 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10406 "Misplaced _ in number");
10412 /* fractional part ending in underbar? */
10413 if (s[-1] == '_') {
10414 if (ckWARN(WARN_SYNTAX))
10415 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10416 "Misplaced _ in number");
10418 if (*s == '.' && isDIGIT(s[1])) {
10419 /* oops, it's really a v-string, but without the "v" */
10425 /* read exponent part, if present */
10426 if ((*s == 'e' || *s == 'E') && strchr("+-0123456789_", s[1])) {
10430 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
10431 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
10433 /* stray preinitial _ */
10435 if (ckWARN(WARN_SYNTAX))
10436 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10437 "Misplaced _ in number");
10441 /* allow positive or negative exponent */
10442 if (*s == '+' || *s == '-')
10445 /* stray initial _ */
10447 if (ckWARN(WARN_SYNTAX))
10448 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10449 "Misplaced _ in number");
10453 /* read digits of exponent */
10454 while (isDIGIT(*s) || *s == '_') {
10457 Perl_croak(aTHX_ number_too_long);
10461 if (((lastub && s == lastub + 1) ||
10462 (!isDIGIT(s[1]) && s[1] != '_'))
10463 && ckWARN(WARN_SYNTAX))
10464 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10465 "Misplaced _ in number");
10472 /* make an sv from the string */
10476 We try to do an integer conversion first if no characters
10477 indicating "float" have been found.
10482 int flags = grok_number (PL_tokenbuf, d - PL_tokenbuf, &uv);
10484 if (flags == IS_NUMBER_IN_UV) {
10486 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
10489 } else if (flags == (IS_NUMBER_IN_UV | IS_NUMBER_NEG)) {
10490 if (uv <= (UV) IV_MIN)
10491 sv_setiv(sv, -(IV)uv);
10498 /* terminate the string */
10500 nv = Atof(PL_tokenbuf);
10504 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
10505 (PL_hints & HINT_NEW_INTEGER) )
10506 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
10507 (floatit ? "float" : "integer"),
10511 /* if it starts with a v, it could be a v-string */
10514 sv = NEWSV(92,5); /* preallocate storage space */
10515 s = scan_vstring(s,sv);
10519 /* make the op for the constant and return */
10522 lvalp->opval = newSVOP(OP_CONST, 0, sv);
10524 lvalp->opval = Nullop;
10530 S_scan_formline(pTHX_ register char *s)
10532 register char *eol;
10534 SV *stuff = newSVpvn("",0);
10535 bool needargs = FALSE;
10536 bool eofmt = FALSE;
10538 while (!needargs) {
10540 #ifdef PERL_STRICT_CR
10541 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
10543 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
10545 if (*t == '\n' || t == PL_bufend) {
10550 if (PL_in_eval && !PL_rsfp) {
10551 eol = (char *) memchr(s,'\n',PL_bufend-s);
10556 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
10558 for (t = s; t < eol; t++) {
10559 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
10561 goto enough; /* ~~ must be first line in formline */
10563 if (*t == '@' || *t == '^')
10567 sv_catpvn(stuff, s, eol-s);
10568 #ifndef PERL_STRICT_CR
10569 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
10570 char *end = SvPVX(stuff) + SvCUR(stuff);
10573 SvCUR_set(stuff, SvCUR(stuff) - 1);
10582 s = filter_gets(PL_linestr, PL_rsfp, 0);
10583 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
10584 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
10585 PL_last_lop = PL_last_uni = Nullch;
10594 if (SvCUR(stuff)) {
10597 PL_lex_state = LEX_NORMAL;
10598 PL_nextval[PL_nexttoke].ival = 0;
10602 PL_lex_state = LEX_FORMLINE;
10604 if (UTF && is_utf8_string((U8*)SvPVX_const(stuff), SvCUR(stuff)))
10606 else if (PL_encoding)
10607 sv_recode_to_utf8(stuff, PL_encoding);
10609 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
10611 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
10615 SvREFCNT_dec(stuff);
10617 PL_lex_formbrack = 0;
10628 PL_cshlen = strlen(PL_cshname);
10633 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
10635 const I32 oldsavestack_ix = PL_savestack_ix;
10636 CV* outsidecv = PL_compcv;
10639 assert(SvTYPE(PL_compcv) == SVt_PVCV);
10641 SAVEI32(PL_subline);
10642 save_item(PL_subname);
10643 SAVESPTR(PL_compcv);
10645 PL_compcv = (CV*)NEWSV(1104,0);
10646 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
10647 CvFLAGS(PL_compcv) |= flags;
10649 PL_subline = CopLINE(PL_curcop);
10650 CvPADLIST(PL_compcv) = pad_new(padnew_SAVE|padnew_SAVESUB);
10651 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
10652 CvOUTSIDE_SEQ(PL_compcv) = PL_cop_seqmax;
10654 return oldsavestack_ix;
10658 #pragma segment Perl_yylex
10661 Perl_yywarn(pTHX_ const char *s)
10663 PL_in_eval |= EVAL_WARNONLY;
10665 PL_in_eval &= ~EVAL_WARNONLY;
10670 Perl_yyerror(pTHX_ const char *s)
10672 const char *where = NULL;
10673 const char *context = NULL;
10677 if (!yychar || (yychar == ';' && !PL_rsfp))
10679 else if (PL_oldoldbufptr && PL_bufptr > PL_oldoldbufptr &&
10680 PL_bufptr - PL_oldoldbufptr < 200 && PL_oldoldbufptr != PL_oldbufptr &&
10681 PL_oldbufptr != PL_bufptr) {
10684 The code below is removed for NetWare because it abends/crashes on NetWare
10685 when the script has error such as not having the closing quotes like:
10686 if ($var eq "value)
10687 Checking of white spaces is anyway done in NetWare code.
10690 while (isSPACE(*PL_oldoldbufptr))
10693 context = PL_oldoldbufptr;
10694 contlen = PL_bufptr - PL_oldoldbufptr;
10696 else if (PL_oldbufptr && PL_bufptr > PL_oldbufptr &&
10697 PL_bufptr - PL_oldbufptr < 200 && PL_oldbufptr != PL_bufptr) {
10700 The code below is removed for NetWare because it abends/crashes on NetWare
10701 when the script has error such as not having the closing quotes like:
10702 if ($var eq "value)
10703 Checking of white spaces is anyway done in NetWare code.
10706 while (isSPACE(*PL_oldbufptr))
10709 context = PL_oldbufptr;
10710 contlen = PL_bufptr - PL_oldbufptr;
10712 else if (yychar > 255)
10713 where = "next token ???";
10714 else if (yychar == -2) { /* YYEMPTY */
10715 if (PL_lex_state == LEX_NORMAL ||
10716 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
10717 where = "at end of line";
10718 else if (PL_lex_inpat)
10719 where = "within pattern";
10721 where = "within string";
10724 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
10726 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
10727 else if (isPRINT_LC(yychar))
10728 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
10730 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
10731 where = SvPVX_const(where_sv);
10733 msg = sv_2mortal(newSVpv(s, 0));
10734 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
10735 OutCopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
10737 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
10739 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
10740 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
10741 Perl_sv_catpvf(aTHX_ msg,
10742 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
10743 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
10746 if (PL_in_eval & EVAL_WARNONLY && ckWARN_d(WARN_SYNTAX))
10747 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "%"SVf, msg);
10750 if (PL_error_count >= 10) {
10751 if (PL_in_eval && SvCUR(ERRSV))
10752 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
10753 ERRSV, OutCopFILE(PL_curcop));
10755 Perl_croak(aTHX_ "%s has too many errors.\n",
10756 OutCopFILE(PL_curcop));
10759 PL_in_my_stash = Nullhv;
10763 #pragma segment Main
10767 S_swallow_bom(pTHX_ U8 *s)
10769 const STRLEN slen = SvCUR(PL_linestr);
10772 if (s[1] == 0xFE) {
10773 /* UTF-16 little-endian? (or UTF32-LE?) */
10774 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
10775 Perl_croak(aTHX_ "Unsupported script encoding UTF32-LE");
10776 #ifndef PERL_NO_UTF16_FILTER
10777 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF16-LE script encoding (BOM)\n");
10780 if (PL_bufend > (char*)s) {
10784 filter_add(utf16rev_textfilter, NULL);
10785 Newx(news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
10786 utf16_to_utf8_reversed(s, news,
10787 PL_bufend - (char*)s - 1,
10789 sv_setpvn(PL_linestr, (const char*)news, newlen);
10791 SvUTF8_on(PL_linestr);
10792 s = (U8*)SvPVX(PL_linestr);
10793 PL_bufend = SvPVX(PL_linestr) + newlen;
10796 Perl_croak(aTHX_ "Unsupported script encoding UTF16-LE");
10801 if (s[1] == 0xFF) { /* UTF-16 big-endian? */
10802 #ifndef PERL_NO_UTF16_FILTER
10803 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (BOM)\n");
10806 if (PL_bufend > (char *)s) {
10810 filter_add(utf16_textfilter, NULL);
10811 Newx(news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
10812 utf16_to_utf8(s, news,
10813 PL_bufend - (char*)s,
10815 sv_setpvn(PL_linestr, (const char*)news, newlen);
10817 SvUTF8_on(PL_linestr);
10818 s = (U8*)SvPVX(PL_linestr);
10819 PL_bufend = SvPVX(PL_linestr) + newlen;
10822 Perl_croak(aTHX_ "Unsupported script encoding UTF16-BE");
10827 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
10828 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-8 script encoding (BOM)\n");
10829 s += 3; /* UTF-8 */
10835 if (s[2] == 0xFE && s[3] == 0xFF) {
10836 /* UTF-32 big-endian */
10837 Perl_croak(aTHX_ "Unsupported script encoding UTF32-BE");
10840 else if (s[2] == 0 && s[3] != 0) {
10843 * are a good indicator of UTF-16BE. */
10844 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (no BOM)\n");
10849 if (slen > 3 && s[1] == 0 && s[2] != 0 && s[3] == 0) {
10852 * are a good indicator of UTF-16LE. */
10853 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16LE script encoding (no BOM)\n");
10862 * Restore a source filter.
10866 restore_rsfp(pTHX_ void *f)
10868 PerlIO *fp = (PerlIO*)f;
10870 if (PL_rsfp == PerlIO_stdin())
10871 PerlIO_clearerr(PL_rsfp);
10872 else if (PL_rsfp && (PL_rsfp != fp))
10873 PerlIO_close(PL_rsfp);
10877 #ifndef PERL_NO_UTF16_FILTER
10879 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
10881 const STRLEN old = SvCUR(sv);
10882 const I32 count = FILTER_READ(idx+1, sv, maxlen);
10883 DEBUG_P(PerlIO_printf(Perl_debug_log,
10884 "utf16_textfilter(%p): %d %d (%d)\n",
10885 utf16_textfilter, idx, maxlen, (int) count));
10889 Newx(tmps, SvCUR(sv) * 3 / 2 + 1, U8);
10890 Copy(SvPVX_const(sv), tmps, old, char);
10891 utf16_to_utf8((U8*)SvPVX_const(sv) + old, tmps + old,
10892 SvCUR(sv) - old, &newlen);
10893 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
10895 DEBUG_P({sv_dump(sv);});
10900 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
10902 const STRLEN old = SvCUR(sv);
10903 const I32 count = FILTER_READ(idx+1, sv, maxlen);
10904 DEBUG_P(PerlIO_printf(Perl_debug_log,
10905 "utf16rev_textfilter(%p): %d %d (%d)\n",
10906 utf16rev_textfilter, idx, maxlen, (int) count));
10910 Newx(tmps, SvCUR(sv) * 3 / 2 + 1, U8);
10911 Copy(SvPVX_const(sv), tmps, old, char);
10912 utf16_to_utf8((U8*)SvPVX_const(sv) + old, tmps + old,
10913 SvCUR(sv) - old, &newlen);
10914 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
10916 DEBUG_P({ sv_dump(sv); });
10922 Returns a pointer to the next character after the parsed
10923 vstring, as well as updating the passed in sv.
10925 Function must be called like
10928 s = scan_vstring(s,sv);
10930 The sv should already be large enough to store the vstring
10931 passed in, for performance reasons.
10936 Perl_scan_vstring(pTHX_ const char *s, SV *sv)
10938 const char *pos = s;
10939 const char *start = s;
10940 if (*pos == 'v') pos++; /* get past 'v' */
10941 while (pos < PL_bufend && (isDIGIT(*pos) || *pos == '_'))
10943 if ( *pos != '.') {
10944 /* this may not be a v-string if followed by => */
10945 const char *next = pos;
10946 while (next < PL_bufend && isSPACE(*next))
10948 if ((PL_bufend - next) >= 2 && *next == '=' && next[1] == '>' ) {
10949 /* return string not v-string */
10950 sv_setpvn(sv,(char *)s,pos-s);
10951 return (char *)pos;
10955 if (!isALPHA(*pos)) {
10957 U8 tmpbuf[UTF8_MAXBYTES+1];
10960 if (*s == 'v') s++; /* get past 'v' */
10962 sv_setpvn(sv, "", 0);
10967 /* this is atoi() that tolerates underscores */
10968 const char *end = pos;
10970 while (--end >= s) {
10975 rev += (*end - '0') * mult;
10977 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
10978 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
10979 "Integer overflow in decimal number");
10983 if (rev > 0x7FFFFFFF)
10984 Perl_croak(aTHX_ "In EBCDIC the v-string components cannot exceed 2147483647");
10986 /* Append native character for the rev point */
10987 tmpend = uvchr_to_utf8(tmpbuf, rev);
10988 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
10989 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(rev)))
10991 if (pos + 1 < PL_bufend && *pos == '.' && isDIGIT(pos[1]))
10997 while (pos < PL_bufend && (isDIGIT(*pos) || *pos == '_'))
11001 sv_magic(sv,NULL,PERL_MAGIC_vstring,(const char*)start, pos-start);
11009 * c-indentation-style: bsd
11010 * c-basic-offset: 4
11011 * indent-tabs-mode: t
11014 * ex: set ts=8 sts=4 sw=4 noet: