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[] = "Identifier too long";
31 static void restore_rsfp(pTHX_ void *f);
32 #ifndef PERL_NO_UTF16_FILTER
33 static I32 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen);
34 static I32 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen);
37 #define XFAKEBRACK 128
40 #ifdef USE_UTF8_SCRIPTS
41 # define UTF (!IN_BYTES)
43 # define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || (PL_hints & HINT_UTF8))
46 /* In variables named $^X, these are the legal values for X.
47 * 1999-02-27 mjd-perl-patch@plover.com */
48 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
50 /* On MacOS, respect nonbreaking spaces */
51 #ifdef MACOS_TRADITIONAL
52 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
54 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
57 /* LEX_* are values for PL_lex_state, the state of the lexer.
58 * They are arranged oddly so that the guard on the switch statement
59 * can get by with a single comparison (if the compiler is smart enough).
62 /* #define LEX_NOTPARSING 11 is done in perl.h. */
64 #define LEX_NORMAL 10 /* normal code (ie not within "...") */
65 #define LEX_INTERPNORMAL 9 /* code within a string, eg "$foo[$x+1]" */
66 #define LEX_INTERPCASEMOD 8 /* expecting a \U, \Q or \E etc */
67 #define LEX_INTERPPUSH 7 /* starting a new sublex parse level */
68 #define LEX_INTERPSTART 6 /* expecting the start of a $var */
70 /* at end of code, eg "$x" followed by: */
71 #define LEX_INTERPEND 5 /* ... eg not one of [, { or -> */
72 #define LEX_INTERPENDMAYBE 4 /* ... eg one of [, { or -> */
74 #define LEX_INTERPCONCAT 3 /* expecting anything, eg at start of
75 string or after \E, $foo, etc */
76 #define LEX_INTERPCONST 2 /* NOT USED */
77 #define LEX_FORMLINE 1 /* expecting a format line */
78 #define LEX_KNOWNEXT 0 /* next token known; just return it */
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((I32)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 { DEFAULT, TOKENTYPE_NONE, "DEFAULT" },
223 { DO, TOKENTYPE_NONE, "DO" },
224 { DOLSHARP, TOKENTYPE_NONE, "DOLSHARP" },
225 { DORDOR, TOKENTYPE_NONE, "DORDOR" },
226 { DOROP, TOKENTYPE_OPNUM, "DOROP" },
227 { DOTDOT, TOKENTYPE_IVAL, "DOTDOT" },
228 { ELSE, TOKENTYPE_NONE, "ELSE" },
229 { ELSIF, TOKENTYPE_IVAL, "ELSIF" },
230 { EQOP, TOKENTYPE_OPNUM, "EQOP" },
231 { FOR, TOKENTYPE_IVAL, "FOR" },
232 { FORMAT, TOKENTYPE_NONE, "FORMAT" },
233 { FUNC, TOKENTYPE_OPNUM, "FUNC" },
234 { FUNC0, TOKENTYPE_OPNUM, "FUNC0" },
235 { FUNC0SUB, TOKENTYPE_OPVAL, "FUNC0SUB" },
236 { FUNC1, TOKENTYPE_OPNUM, "FUNC1" },
237 { FUNCMETH, TOKENTYPE_OPVAL, "FUNCMETH" },
238 { GIVEN, TOKENTYPE_IVAL, "GIVEN" },
239 { HASHBRACK, TOKENTYPE_NONE, "HASHBRACK" },
240 { IF, TOKENTYPE_IVAL, "IF" },
241 { LABEL, TOKENTYPE_PVAL, "LABEL" },
242 { LOCAL, TOKENTYPE_IVAL, "LOCAL" },
243 { LOOPEX, TOKENTYPE_OPNUM, "LOOPEX" },
244 { LSTOP, TOKENTYPE_OPNUM, "LSTOP" },
245 { LSTOPSUB, TOKENTYPE_OPVAL, "LSTOPSUB" },
246 { MATCHOP, TOKENTYPE_OPNUM, "MATCHOP" },
247 { METHOD, TOKENTYPE_OPVAL, "METHOD" },
248 { MULOP, TOKENTYPE_OPNUM, "MULOP" },
249 { MY, TOKENTYPE_IVAL, "MY" },
250 { MYSUB, TOKENTYPE_NONE, "MYSUB" },
251 { NOAMP, TOKENTYPE_NONE, "NOAMP" },
252 { NOTOP, TOKENTYPE_NONE, "NOTOP" },
253 { OROP, TOKENTYPE_IVAL, "OROP" },
254 { OROR, TOKENTYPE_NONE, "OROR" },
255 { PACKAGE, TOKENTYPE_NONE, "PACKAGE" },
256 { PMFUNC, TOKENTYPE_OPVAL, "PMFUNC" },
257 { POSTDEC, TOKENTYPE_NONE, "POSTDEC" },
258 { POSTINC, TOKENTYPE_NONE, "POSTINC" },
259 { POWOP, TOKENTYPE_OPNUM, "POWOP" },
260 { PREDEC, TOKENTYPE_NONE, "PREDEC" },
261 { PREINC, TOKENTYPE_NONE, "PREINC" },
262 { PRIVATEREF, TOKENTYPE_OPVAL, "PRIVATEREF" },
263 { REFGEN, TOKENTYPE_NONE, "REFGEN" },
264 { RELOP, TOKENTYPE_OPNUM, "RELOP" },
265 { SHIFTOP, TOKENTYPE_OPNUM, "SHIFTOP" },
266 { SUB, TOKENTYPE_NONE, "SUB" },
267 { THING, TOKENTYPE_OPVAL, "THING" },
268 { UMINUS, TOKENTYPE_NONE, "UMINUS" },
269 { UNIOP, TOKENTYPE_OPNUM, "UNIOP" },
270 { UNIOPSUB, TOKENTYPE_OPVAL, "UNIOPSUB" },
271 { UNLESS, TOKENTYPE_IVAL, "UNLESS" },
272 { UNTIL, TOKENTYPE_IVAL, "UNTIL" },
273 { USE, TOKENTYPE_IVAL, "USE" },
274 { WHEN, TOKENTYPE_IVAL, "WHEN" },
275 { WHILE, TOKENTYPE_IVAL, "WHILE" },
276 { WORD, TOKENTYPE_OPVAL, "WORD" },
277 { 0, TOKENTYPE_NONE, 0 }
280 /* dump the returned token in rv, plus any optional arg in yylval */
283 S_tokereport(pTHX_ I32 rv)
286 const char *name = Nullch;
287 enum token_type type = TOKENTYPE_NONE;
288 const struct debug_tokens *p;
289 SV* const report = newSVpvn("<== ", 4);
291 for (p = debug_tokens; p->token; p++) {
292 if (p->token == (int)rv) {
299 Perl_sv_catpv(aTHX_ report, name);
300 else if ((char)rv > ' ' && (char)rv < '~')
301 Perl_sv_catpvf(aTHX_ report, "'%c'", (char)rv);
303 Perl_sv_catpv(aTHX_ report, "EOF");
305 Perl_sv_catpvf(aTHX_ report, "?? %"IVdf, (IV)rv);
308 case TOKENTYPE_GVVAL: /* doesn't appear to be used */
311 Perl_sv_catpvf(aTHX_ report, "(ival=%"IVdf")", (IV)yylval.ival);
313 case TOKENTYPE_OPNUM:
314 Perl_sv_catpvf(aTHX_ report, "(ival=op_%s)",
315 PL_op_name[yylval.ival]);
318 Perl_sv_catpvf(aTHX_ report, "(pval=\"%s\")", yylval.pval);
320 case TOKENTYPE_OPVAL:
322 Perl_sv_catpvf(aTHX_ report, "(opval=op_%s)",
323 PL_op_name[yylval.opval->op_type]);
324 if (yylval.opval->op_type == OP_CONST) {
325 Perl_sv_catpvf(aTHX_ report, " %s",
326 SvPEEK(cSVOPx_sv(yylval.opval)));
331 Perl_sv_catpv(aTHX_ report, "(opval=null)");
334 PerlIO_printf(Perl_debug_log, "### %s\n\n", SvPV_nolen_const(report));
340 /* print the buffer with suitable escapes */
343 S_printbuf(pTHX_ const char* fmt, const char* s)
345 SV* const tmp = newSVpvn("", 0);
346 PerlIO_printf(Perl_debug_log, fmt, pv_display(tmp, s, strlen(s), 0, 60));
355 * This subroutine detects &&=, ||=, and //= and turns an ANDAND, OROR or DORDOR
356 * into an OP_ANDASSIGN, OP_ORASSIGN, or OP_DORASSIGN
360 S_ao(pTHX_ int toketype)
362 if (*PL_bufptr == '=') {
364 if (toketype == ANDAND)
365 yylval.ival = OP_ANDASSIGN;
366 else if (toketype == OROR)
367 yylval.ival = OP_ORASSIGN;
368 else if (toketype == DORDOR)
369 yylval.ival = OP_DORASSIGN;
377 * When Perl expects an operator and finds something else, no_op
378 * prints the warning. It always prints "<something> found where
379 * operator expected. It prints "Missing semicolon on previous line?"
380 * if the surprise occurs at the start of the line. "do you need to
381 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
382 * where the compiler doesn't know if foo is a method call or a function.
383 * It prints "Missing operator before end of line" if there's nothing
384 * after the missing operator, or "... before <...>" if there is something
385 * after the missing operator.
389 S_no_op(pTHX_ const char *what, char *s)
391 char * const oldbp = PL_bufptr;
392 const bool is_first = (PL_oldbufptr == PL_linestart);
398 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
399 if (ckWARN_d(WARN_SYNTAX)) {
401 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
402 "\t(Missing semicolon on previous line?)\n");
403 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
405 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy_if(t,UTF) || *t == ':'); t++) ;
406 if (t < PL_bufptr && isSPACE(*t))
407 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
408 "\t(Do you need to predeclare %.*s?)\n",
409 (int)(t - PL_oldoldbufptr), PL_oldoldbufptr);
413 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
414 "\t(Missing operator before %.*s?)\n", (int)(s - oldbp), oldbp);
422 * Complain about missing quote/regexp/heredoc terminator.
423 * If it's called with (char *)NULL then it cauterizes the line buffer.
424 * If we're in a delimited string and the delimiter is a control
425 * character, it's reformatted into a two-char sequence like ^C.
430 S_missingterm(pTHX_ char *s)
435 char * const nl = strrchr(s,'\n');
441 iscntrl(PL_multi_close)
443 PL_multi_close < 32 || PL_multi_close == 127
447 tmpbuf[1] = (char)toCTRL(PL_multi_close);
452 *tmpbuf = (char)PL_multi_close;
456 q = strchr(s,'"') ? '\'' : '"';
457 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
460 #define FEATURE_IS_ENABLED(name, namelen) \
461 ((0 != (PL_hints & HINT_LOCALIZE_HH)) \
462 && feature_is_enabled(name, namelen))
464 * S_feature_is_enabled
465 * Check whether the named feature is enabled.
468 S_feature_is_enabled(pTHX_ char *name, STRLEN namelen)
470 HV * const hinthv = GvHV(PL_hintgv);
471 char he_name[32] = "feature_";
472 (void) strncpy(&he_name[8], name, 24);
474 return (hinthv && hv_exists(hinthv, he_name, 8 + namelen));
482 Perl_deprecate(pTHX_ const char *s)
484 if (ckWARN(WARN_DEPRECATED))
485 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED), "Use of %s is deprecated", s);
489 Perl_deprecate_old(pTHX_ const char *s)
491 /* This function should NOT be called for any new deprecated warnings */
492 /* Use Perl_deprecate instead */
494 /* It is here to maintain backward compatibility with the pre-5.8 */
495 /* warnings category hierarchy. The "deprecated" category used to */
496 /* live under the "syntax" category. It is now a top-level category */
497 /* in its own right. */
499 if (ckWARN2(WARN_DEPRECATED, WARN_SYNTAX))
500 Perl_warner(aTHX_ packWARN2(WARN_DEPRECATED, WARN_SYNTAX),
501 "Use of %s is deprecated", s);
506 * Deprecate a comma-less variable list.
512 deprecate_old("comma-less variable list");
516 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
517 * utf16-to-utf8-reversed.
520 #ifdef PERL_CR_FILTER
524 register const char *s = SvPVX_const(sv);
525 register const char * const e = s + SvCUR(sv);
526 /* outer loop optimized to do nothing if there are no CR-LFs */
528 if (*s++ == '\r' && *s == '\n') {
529 /* hit a CR-LF, need to copy the rest */
530 register char *d = s - 1;
533 if (*s == '\r' && s[1] == '\n')
544 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
546 const I32 count = FILTER_READ(idx+1, sv, maxlen);
547 if (count > 0 && !maxlen)
555 * Initialize variables. Uses the Perl save_stack to save its state (for
556 * recursive calls to the parser).
560 Perl_lex_start(pTHX_ SV *line)
565 SAVEI32(PL_lex_dojoin);
566 SAVEI32(PL_lex_brackets);
567 SAVEI32(PL_lex_casemods);
568 SAVEI32(PL_lex_starts);
569 SAVEI32(PL_lex_state);
570 SAVEVPTR(PL_lex_inpat);
571 SAVEI32(PL_lex_inwhat);
572 if (PL_lex_state == LEX_KNOWNEXT) {
573 I32 toke = PL_nexttoke;
574 while (--toke >= 0) {
575 SAVEI32(PL_nexttype[toke]);
576 SAVEVPTR(PL_nextval[toke]);
578 SAVEI32(PL_nexttoke);
580 SAVECOPLINE(PL_curcop);
583 SAVEPPTR(PL_oldbufptr);
584 SAVEPPTR(PL_oldoldbufptr);
585 SAVEPPTR(PL_last_lop);
586 SAVEPPTR(PL_last_uni);
587 SAVEPPTR(PL_linestart);
588 SAVESPTR(PL_linestr);
589 SAVEGENERICPV(PL_lex_brackstack);
590 SAVEGENERICPV(PL_lex_casestack);
591 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
592 SAVESPTR(PL_lex_stuff);
593 SAVEI32(PL_lex_defer);
594 SAVEI32(PL_sublex_info.sub_inwhat);
595 SAVESPTR(PL_lex_repl);
597 SAVEINT(PL_lex_expect);
599 PL_lex_state = LEX_NORMAL;
603 Newx(PL_lex_brackstack, 120, char);
604 Newx(PL_lex_casestack, 12, char);
606 *PL_lex_casestack = '\0';
609 PL_lex_stuff = Nullsv;
610 PL_lex_repl = Nullsv;
614 PL_sublex_info.sub_inwhat = 0;
616 if (SvREADONLY(PL_linestr))
617 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
618 s = SvPV_const(PL_linestr, len);
619 if (!len || s[len-1] != ';') {
620 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
621 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
622 sv_catpvn(PL_linestr, "\n;", 2);
624 SvTEMP_off(PL_linestr);
625 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
626 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
627 PL_last_lop = PL_last_uni = Nullch;
633 * Finalizer for lexing operations. Must be called when the parser is
634 * done with the lexer.
640 PL_doextract = FALSE;
645 * This subroutine has nothing to do with tilting, whether at windmills
646 * or pinball tables. Its name is short for "increment line". It
647 * increments the current line number in CopLINE(PL_curcop) and checks
648 * to see whether the line starts with a comment of the form
649 * # line 500 "foo.pm"
650 * If so, it sets the current line number and file to the values in the comment.
654 S_incline(pTHX_ char *s)
661 CopLINE_inc(PL_curcop);
664 while (SPACE_OR_TAB(*s)) s++;
665 if (strnEQ(s, "line", 4))
669 if (SPACE_OR_TAB(*s))
673 while (SPACE_OR_TAB(*s)) s++;
679 while (SPACE_OR_TAB(*s))
681 if (*s == '"' && (t = strchr(s+1, '"'))) {
686 for (t = s; !isSPACE(*t); t++) ;
689 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
691 if (*e != '\n' && *e != '\0')
692 return; /* false alarm */
698 const char * const cf = CopFILE(PL_curcop);
699 STRLEN tmplen = cf ? strlen(cf) : 0;
700 if (tmplen > 7 && strnEQ(cf, "(eval ", 6)) {
701 /* must copy *{"::_<(eval N)[oldfilename:L]"}
702 * to *{"::_<newfilename"} */
703 char smallbuf[256], smallbuf2[256];
704 char *tmpbuf, *tmpbuf2;
706 STRLEN tmplen2 = strlen(s);
707 if (tmplen + 3 < sizeof smallbuf)
710 Newx(tmpbuf, tmplen + 3, char);
711 if (tmplen2 + 3 < sizeof smallbuf2)
714 Newx(tmpbuf2, tmplen2 + 3, char);
715 tmpbuf[0] = tmpbuf2[0] = '_';
716 tmpbuf[1] = tmpbuf2[1] = '<';
717 memcpy(tmpbuf + 2, cf, ++tmplen);
718 memcpy(tmpbuf2 + 2, s, ++tmplen2);
720 gvp = (GV**)hv_fetch(PL_defstash, tmpbuf, tmplen, FALSE);
722 gv2 = *(GV**)hv_fetch(PL_defstash, tmpbuf2, tmplen2, TRUE);
724 gv_init(gv2, PL_defstash, tmpbuf2, tmplen2, FALSE);
725 /* adjust ${"::_<newfilename"} to store the new file name */
726 GvSV(gv2) = newSVpvn(tmpbuf2 + 2, tmplen2 - 2);
727 GvHV(gv2) = (HV*)SvREFCNT_inc(GvHV(*gvp));
728 GvAV(gv2) = (AV*)SvREFCNT_inc(GvAV(*gvp));
730 if (tmpbuf != smallbuf) Safefree(tmpbuf);
731 if (tmpbuf2 != smallbuf2) Safefree(tmpbuf2);
734 CopFILE_free(PL_curcop);
735 CopFILE_set(PL_curcop, s);
738 CopLINE_set(PL_curcop, atoi(n)-1);
743 * Called to gobble the appropriate amount and type of whitespace.
744 * Skips comments as well.
748 S_skipspace(pTHX_ register char *s)
750 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
751 while (s < PL_bufend && SPACE_OR_TAB(*s))
757 SSize_t oldprevlen, oldoldprevlen;
758 SSize_t oldloplen = 0, oldunilen = 0;
759 while (s < PL_bufend && isSPACE(*s)) {
760 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
765 if (s < PL_bufend && *s == '#') {
766 while (s < PL_bufend && *s != '\n')
770 if (PL_in_eval && !PL_rsfp) {
777 /* only continue to recharge the buffer if we're at the end
778 * of the buffer, we're not reading from a source filter, and
779 * we're in normal lexing mode
781 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
782 PL_lex_state == LEX_FORMLINE)
785 /* try to recharge the buffer */
786 if ((s = filter_gets(PL_linestr, PL_rsfp,
787 (prevlen = SvCUR(PL_linestr)))) == Nullch)
789 /* end of file. Add on the -p or -n magic */
792 ";}continue{print or die qq(-p destination: $!\\n);}");
793 PL_minus_n = PL_minus_p = 0;
795 else if (PL_minus_n) {
796 sv_setpvn(PL_linestr, ";}", 2);
800 sv_setpvn(PL_linestr,";", 1);
802 /* reset variables for next time we lex */
803 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
805 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
806 PL_last_lop = PL_last_uni = Nullch;
808 /* Close the filehandle. Could be from -P preprocessor,
809 * STDIN, or a regular file. If we were reading code from
810 * STDIN (because the commandline held no -e or filename)
811 * then we don't close it, we reset it so the code can
812 * read from STDIN too.
815 if (PL_preprocess && !PL_in_eval)
816 (void)PerlProc_pclose(PL_rsfp);
817 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
818 PerlIO_clearerr(PL_rsfp);
820 (void)PerlIO_close(PL_rsfp);
825 /* not at end of file, so we only read another line */
826 /* make corresponding updates to old pointers, for yyerror() */
827 oldprevlen = PL_oldbufptr - PL_bufend;
828 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
830 oldunilen = PL_last_uni - PL_bufend;
832 oldloplen = PL_last_lop - PL_bufend;
833 PL_linestart = PL_bufptr = s + prevlen;
834 PL_bufend = s + SvCUR(PL_linestr);
836 PL_oldbufptr = s + oldprevlen;
837 PL_oldoldbufptr = s + oldoldprevlen;
839 PL_last_uni = s + oldunilen;
841 PL_last_lop = s + oldloplen;
844 /* debugger active and we're not compiling the debugger code,
845 * so store the line into the debugger's array of lines
847 if (PERLDB_LINE && PL_curstash != PL_debstash) {
848 SV * const sv = NEWSV(85,0);
850 sv_upgrade(sv, SVt_PVMG);
851 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
854 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
861 * Check the unary operators to ensure there's no ambiguity in how they're
862 * used. An ambiguous piece of code would be:
864 * This doesn't mean rand() + 5. Because rand() is a unary operator,
865 * the +5 is its argument.
874 if (PL_oldoldbufptr != PL_last_uni)
876 while (isSPACE(*PL_last_uni))
878 for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ;
879 if ((t = strchr(s, '(')) && t < PL_bufptr)
881 if (ckWARN_d(WARN_AMBIGUOUS)){
884 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
885 "Warning: Use of \"%s\" without parentheses is ambiguous",
892 * LOP : macro to build a list operator. Its behaviour has been replaced
893 * with a subroutine, S_lop() for which LOP is just another name.
896 #define LOP(f,x) return lop(f,x,s)
900 * Build a list operator (or something that might be one). The rules:
901 * - if we have a next token, then it's a list operator [why?]
902 * - if the next thing is an opening paren, then it's a function
903 * - else it's a list operator
907 S_lop(pTHX_ I32 f, int x, char *s)
913 PL_last_lop = PL_oldbufptr;
914 PL_last_lop_op = (OPCODE)f;
916 return REPORT(LSTOP);
923 return REPORT(LSTOP);
928 * When the lexer realizes it knows the next token (for instance,
929 * it is reordering tokens for the parser) then it can call S_force_next
930 * to know what token to return the next time the lexer is called. Caller
931 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
932 * handles the token correctly.
936 S_force_next(pTHX_ I32 type)
938 PL_nexttype[PL_nexttoke] = type;
940 if (PL_lex_state != LEX_KNOWNEXT) {
941 PL_lex_defer = PL_lex_state;
942 PL_lex_expect = PL_expect;
943 PL_lex_state = LEX_KNOWNEXT;
948 S_newSV_maybe_utf8(pTHX_ const char *start, STRLEN len)
950 SV * const sv = newSVpvn(start,len);
951 if (UTF && !IN_BYTES && is_utf8_string((const U8*)start, len))
958 * When the lexer knows the next thing is a word (for instance, it has
959 * just seen -> and it knows that the next char is a word char, then
960 * it calls S_force_word to stick the next word into the PL_next lookahead.
963 * char *start : buffer position (must be within PL_linestr)
964 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
965 * int check_keyword : if true, Perl checks to make sure the word isn't
966 * a keyword (do this if the word is a label, e.g. goto FOO)
967 * int allow_pack : if true, : characters will also be allowed (require,
969 * int allow_initial_tick : used by the "sub" lexer only.
973 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
978 start = skipspace(start);
980 if (isIDFIRST_lazy_if(s,UTF) ||
981 (allow_pack && *s == ':') ||
982 (allow_initial_tick && *s == '\'') )
984 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
985 if (check_keyword && keyword(PL_tokenbuf, len))
987 if (token == METHOD) {
992 PL_expect = XOPERATOR;
995 PL_nextval[PL_nexttoke].opval
996 = (OP*)newSVOP(OP_CONST,0,
997 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
998 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
1006 * Called when the lexer wants $foo *foo &foo etc, but the program
1007 * text only contains the "foo" portion. The first argument is a pointer
1008 * to the "foo", and the second argument is the type symbol to prefix.
1009 * Forces the next token to be a "WORD".
1010 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
1014 S_force_ident(pTHX_ register const char *s, int kind)
1017 OP* const o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
1018 PL_nextval[PL_nexttoke].opval = o;
1021 o->op_private = OPpCONST_ENTERED;
1022 /* XXX see note in pp_entereval() for why we forgo typo
1023 warnings if the symbol must be introduced in an eval.
1025 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : GV_ADD,
1026 kind == '$' ? SVt_PV :
1027 kind == '@' ? SVt_PVAV :
1028 kind == '%' ? SVt_PVHV :
1036 Perl_str_to_version(pTHX_ SV *sv)
1041 const char *start = SvPV_const(sv,len);
1042 const char * const end = start + len;
1043 const bool utf = SvUTF8(sv) ? TRUE : FALSE;
1044 while (start < end) {
1048 n = utf8n_to_uvchr((U8*)start, len, &skip, 0);
1053 retval += ((NV)n)/nshift;
1062 * Forces the next token to be a version number.
1063 * If the next token appears to be an invalid version number, (e.g. "v2b"),
1064 * and if "guessing" is TRUE, then no new token is created (and the caller
1065 * must use an alternative parsing method).
1069 S_force_version(pTHX_ char *s, int guessing)
1071 OP *version = Nullop;
1080 while (isDIGIT(*d) || *d == '_' || *d == '.')
1082 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
1084 s = scan_num(s, &yylval);
1085 version = yylval.opval;
1086 ver = cSVOPx(version)->op_sv;
1087 if (SvPOK(ver) && !SvNIOK(ver)) {
1088 SvUPGRADE(ver, SVt_PVNV);
1089 SvNV_set(ver, str_to_version(ver));
1090 SvNOK_on(ver); /* hint that it is a version */
1097 /* NOTE: The parser sees the package name and the VERSION swapped */
1098 PL_nextval[PL_nexttoke].opval = version;
1106 * Tokenize a quoted string passed in as an SV. It finds the next
1107 * chunk, up to end of string or a backslash. It may make a new
1108 * SV containing that chunk (if HINT_NEW_STRING is on). It also
1113 S_tokeq(pTHX_ SV *sv)
1116 register char *send;
1124 s = SvPV_force(sv, len);
1125 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
1128 while (s < send && *s != '\\')
1133 if ( PL_hints & HINT_NEW_STRING ) {
1134 pv = sv_2mortal(newSVpvn(SvPVX_const(pv), len));
1140 if (s + 1 < send && (s[1] == '\\'))
1141 s++; /* all that, just for this */
1146 SvCUR_set(sv, d - SvPVX_const(sv));
1148 if ( PL_hints & HINT_NEW_STRING )
1149 return new_constant(NULL, 0, "q", sv, pv, "q");
1154 * Now come three functions related to double-quote context,
1155 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
1156 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
1157 * interact with PL_lex_state, and create fake ( ... ) argument lists
1158 * to handle functions and concatenation.
1159 * They assume that whoever calls them will be setting up a fake
1160 * join call, because each subthing puts a ',' after it. This lets
1163 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
1165 * (I'm not sure whether the spurious commas at the end of lcfirst's
1166 * arguments and join's arguments are created or not).
1171 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
1173 * Pattern matching will set PL_lex_op to the pattern-matching op to
1174 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
1176 * OP_CONST and OP_READLINE are easy--just make the new op and return.
1178 * Everything else becomes a FUNC.
1180 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
1181 * had an OP_CONST or OP_READLINE). This just sets us up for a
1182 * call to S_sublex_push().
1186 S_sublex_start(pTHX)
1188 register const I32 op_type = yylval.ival;
1190 if (op_type == OP_NULL) {
1191 yylval.opval = PL_lex_op;
1195 if (op_type == OP_CONST || op_type == OP_READLINE) {
1196 SV *sv = tokeq(PL_lex_stuff);
1198 if (SvTYPE(sv) == SVt_PVIV) {
1199 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
1201 const char *p = SvPV_const(sv, len);
1202 SV * const nsv = newSVpvn(p, len);
1208 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
1209 PL_lex_stuff = Nullsv;
1210 /* Allow <FH> // "foo" */
1211 if (op_type == OP_READLINE)
1212 PL_expect = XTERMORDORDOR;
1216 PL_sublex_info.super_state = PL_lex_state;
1217 PL_sublex_info.sub_inwhat = op_type;
1218 PL_sublex_info.sub_op = PL_lex_op;
1219 PL_lex_state = LEX_INTERPPUSH;
1223 yylval.opval = PL_lex_op;
1233 * Create a new scope to save the lexing state. The scope will be
1234 * ended in S_sublex_done. Returns a '(', starting the function arguments
1235 * to the uc, lc, etc. found before.
1236 * Sets PL_lex_state to LEX_INTERPCONCAT.
1245 PL_lex_state = PL_sublex_info.super_state;
1246 SAVEI32(PL_lex_dojoin);
1247 SAVEI32(PL_lex_brackets);
1248 SAVEI32(PL_lex_casemods);
1249 SAVEI32(PL_lex_starts);
1250 SAVEI32(PL_lex_state);
1251 SAVEVPTR(PL_lex_inpat);
1252 SAVEI32(PL_lex_inwhat);
1253 SAVECOPLINE(PL_curcop);
1254 SAVEPPTR(PL_bufptr);
1255 SAVEPPTR(PL_bufend);
1256 SAVEPPTR(PL_oldbufptr);
1257 SAVEPPTR(PL_oldoldbufptr);
1258 SAVEPPTR(PL_last_lop);
1259 SAVEPPTR(PL_last_uni);
1260 SAVEPPTR(PL_linestart);
1261 SAVESPTR(PL_linestr);
1262 SAVEGENERICPV(PL_lex_brackstack);
1263 SAVEGENERICPV(PL_lex_casestack);
1265 PL_linestr = PL_lex_stuff;
1266 PL_lex_stuff = Nullsv;
1268 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1269 = SvPVX(PL_linestr);
1270 PL_bufend += SvCUR(PL_linestr);
1271 PL_last_lop = PL_last_uni = Nullch;
1272 SAVEFREESV(PL_linestr);
1274 PL_lex_dojoin = FALSE;
1275 PL_lex_brackets = 0;
1276 Newx(PL_lex_brackstack, 120, char);
1277 Newx(PL_lex_casestack, 12, char);
1278 PL_lex_casemods = 0;
1279 *PL_lex_casestack = '\0';
1281 PL_lex_state = LEX_INTERPCONCAT;
1282 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
1284 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1285 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1286 PL_lex_inpat = PL_sublex_info.sub_op;
1288 PL_lex_inpat = Nullop;
1295 * Restores lexer state after a S_sublex_push.
1302 if (!PL_lex_starts++) {
1303 SV * const sv = newSVpvn("",0);
1304 if (SvUTF8(PL_linestr))
1306 PL_expect = XOPERATOR;
1307 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1311 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1312 PL_lex_state = LEX_INTERPCASEMOD;
1316 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1317 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1318 PL_linestr = PL_lex_repl;
1320 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1321 PL_bufend += SvCUR(PL_linestr);
1322 PL_last_lop = PL_last_uni = Nullch;
1323 SAVEFREESV(PL_linestr);
1324 PL_lex_dojoin = FALSE;
1325 PL_lex_brackets = 0;
1326 PL_lex_casemods = 0;
1327 *PL_lex_casestack = '\0';
1329 if (SvEVALED(PL_lex_repl)) {
1330 PL_lex_state = LEX_INTERPNORMAL;
1332 /* we don't clear PL_lex_repl here, so that we can check later
1333 whether this is an evalled subst; that means we rely on the
1334 logic to ensure sublex_done() is called again only via the
1335 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1338 PL_lex_state = LEX_INTERPCONCAT;
1339 PL_lex_repl = Nullsv;
1345 PL_bufend = SvPVX(PL_linestr);
1346 PL_bufend += SvCUR(PL_linestr);
1347 PL_expect = XOPERATOR;
1348 PL_sublex_info.sub_inwhat = 0;
1356 Extracts a pattern, double-quoted string, or transliteration. This
1359 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1360 processing a pattern (PL_lex_inpat is true), a transliteration
1361 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1363 Returns a pointer to the character scanned up to. Iff this is
1364 advanced from the start pointer supplied (ie if anything was
1365 successfully parsed), will leave an OP for the substring scanned
1366 in yylval. Caller must intuit reason for not parsing further
1367 by looking at the next characters herself.
1371 double-quoted style: \r and \n
1372 regexp special ones: \D \s
1374 backrefs: \1 (deprecated in substitution replacements)
1375 case and quoting: \U \Q \E
1376 stops on @ and $, but not for $ as tail anchor
1378 In transliterations:
1379 characters are VERY literal, except for - not at the start or end
1380 of the string, which indicates a range. scan_const expands the
1381 range to the full set of intermediate characters.
1383 In double-quoted strings:
1385 double-quoted style: \r and \n
1387 backrefs: \1 (deprecated)
1388 case and quoting: \U \Q \E
1391 scan_const does *not* construct ops to handle interpolated strings.
1392 It stops processing as soon as it finds an embedded $ or @ variable
1393 and leaves it to the caller to work out what's going on.
1395 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @::foo.
1397 $ in pattern could be $foo or could be tail anchor. Assumption:
1398 it's a tail anchor if $ is the last thing in the string, or if it's
1399 followed by one of ")| \n\t"
1401 \1 (backreferences) are turned into $1
1403 The structure of the code is
1404 while (there's a character to process) {
1405 handle transliteration ranges
1406 skip regexp comments
1407 skip # initiated comments in //x patterns
1408 check for embedded @foo
1409 check for embedded scalars
1411 leave intact backslashes from leave (below)
1412 deprecate \1 in strings and sub replacements
1413 handle string-changing backslashes \l \U \Q \E, etc.
1414 switch (what was escaped) {
1415 handle - in a transliteration (becomes a literal -)
1416 handle \132 octal characters
1417 handle 0x15 hex characters
1418 handle \cV (control V)
1419 handle printf backslashes (\f, \r, \n, etc)
1421 } (end if backslash)
1422 } (end while character to read)
1427 S_scan_const(pTHX_ char *start)
1429 register char *send = PL_bufend; /* end of the constant */
1430 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1431 register char *s = start; /* start of the constant */
1432 register char *d = SvPVX(sv); /* destination for copies */
1433 bool dorange = FALSE; /* are we in a translit range? */
1434 bool didrange = FALSE; /* did we just finish a range? */
1435 I32 has_utf8 = FALSE; /* Output constant is UTF8 */
1436 I32 this_utf8 = UTF; /* The source string is assumed to be UTF8 */
1439 UV literal_endpoint = 0;
1442 const char *leaveit = /* set of acceptably-backslashed characters */
1444 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxz0123456789[{]} \t\n\r\f\v#"
1447 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1448 /* If we are doing a trans and we know we want UTF8 set expectation */
1449 has_utf8 = PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF);
1450 this_utf8 = PL_sublex_info.sub_op->op_private & (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1454 while (s < send || dorange) {
1455 /* get transliterations out of the way (they're most literal) */
1456 if (PL_lex_inwhat == OP_TRANS) {
1457 /* expand a range A-Z to the full set of characters. AIE! */
1459 I32 i; /* current expanded character */
1460 I32 min; /* first character in range */
1461 I32 max; /* last character in range */
1464 char * const c = (char*)utf8_hop((U8*)d, -1);
1468 *c = (char)UTF_TO_NATIVE(0xff);
1469 /* mark the range as done, and continue */
1475 i = d - SvPVX_const(sv); /* remember current offset */
1476 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1477 d = SvPVX(sv) + i; /* refresh d after realloc */
1478 d -= 2; /* eat the first char and the - */
1480 min = (U8)*d; /* first char in range */
1481 max = (U8)d[1]; /* last char in range */
1485 "Invalid range \"%c-%c\" in transliteration operator",
1486 (char)min, (char)max);
1490 if (literal_endpoint == 2 &&
1491 ((isLOWER(min) && isLOWER(max)) ||
1492 (isUPPER(min) && isUPPER(max)))) {
1494 for (i = min; i <= max; i++)
1496 *d++ = NATIVE_TO_NEED(has_utf8,i);
1498 for (i = min; i <= max; i++)
1500 *d++ = NATIVE_TO_NEED(has_utf8,i);
1505 for (i = min; i <= max; i++)
1508 /* mark the range as done, and continue */
1512 literal_endpoint = 0;
1517 /* range begins (ignore - as first or last char) */
1518 else if (*s == '-' && s+1 < send && s != start) {
1520 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
1523 *d++ = (char)UTF_TO_NATIVE(0xff); /* use illegal utf8 byte--see pmtrans */
1533 literal_endpoint = 0;
1538 /* if we get here, we're not doing a transliteration */
1540 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1541 except for the last char, which will be done separately. */
1542 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1544 while (s+1 < send && *s != ')')
1545 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1547 else if (s[2] == '{' /* This should match regcomp.c */
1548 || ((s[2] == 'p' || s[2] == '?') && s[3] == '{'))
1551 char *regparse = s + (s[2] == '{' ? 3 : 4);
1554 while (count && (c = *regparse)) {
1555 if (c == '\\' && regparse[1])
1563 if (*regparse != ')')
1564 regparse--; /* Leave one char for continuation. */
1565 while (s < regparse)
1566 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1570 /* likewise skip #-initiated comments in //x patterns */
1571 else if (*s == '#' && PL_lex_inpat &&
1572 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1573 while (s+1 < send && *s != '\n')
1574 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1577 /* check for embedded arrays
1578 (@foo, @::foo, @'foo, @{foo}, @$foo, @+, @-)
1580 else if (*s == '@' && s[1]
1581 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$+-", s[1])))
1584 /* check for embedded scalars. only stop if we're sure it's a
1587 else if (*s == '$') {
1588 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1590 if (s + 1 < send && !strchr("()| \r\n\t", s[1]))
1591 break; /* in regexp, $ might be tail anchor */
1594 /* End of else if chain - OP_TRANS rejoin rest */
1597 if (*s == '\\' && s+1 < send) {
1600 /* some backslashes we leave behind */
1601 if (*leaveit && *s && strchr(leaveit, *s)) {
1602 *d++ = NATIVE_TO_NEED(has_utf8,'\\');
1603 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1607 /* deprecate \1 in strings and substitution replacements */
1608 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1609 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1611 if (ckWARN(WARN_SYNTAX))
1612 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "\\%c better written as $%c", *s, *s);
1617 /* string-change backslash escapes */
1618 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1623 /* if we get here, it's either a quoted -, or a digit */
1626 /* quoted - in transliterations */
1628 if (PL_lex_inwhat == OP_TRANS) {
1638 Perl_warner(aTHX_ packWARN(WARN_MISC),
1639 "Unrecognized escape \\%c passed through",
1641 /* default action is to copy the quoted character */
1642 goto default_action;
1645 /* \132 indicates an octal constant */
1646 case '0': case '1': case '2': case '3':
1647 case '4': case '5': case '6': case '7':
1651 uv = grok_oct(s, &len, &flags, NULL);
1654 goto NUM_ESCAPE_INSERT;
1656 /* \x24 indicates a hex constant */
1660 char* const e = strchr(s, '}');
1661 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
1662 PERL_SCAN_DISALLOW_PREFIX;
1667 yyerror("Missing right brace on \\x{}");
1671 uv = grok_hex(s, &len, &flags, NULL);
1677 I32 flags = PERL_SCAN_DISALLOW_PREFIX;
1678 uv = grok_hex(s, &len, &flags, NULL);
1684 /* Insert oct or hex escaped character.
1685 * There will always enough room in sv since such
1686 * escapes will be longer than any UTF-8 sequence
1687 * they can end up as. */
1689 /* We need to map to chars to ASCII before doing the tests
1692 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(uv))) {
1693 if (!has_utf8 && uv > 255) {
1694 /* Might need to recode whatever we have
1695 * accumulated so far if it contains any
1698 * (Can't we keep track of that and avoid
1699 * this rescan? --jhi)
1703 for (c = (U8 *) SvPVX(sv); c < (U8 *)d; c++) {
1704 if (!NATIVE_IS_INVARIANT(*c)) {
1709 const STRLEN offset = d - SvPVX_const(sv);
1711 d = SvGROW(sv, SvLEN(sv) + hicount + 1) + offset;
1715 while (src >= (const U8 *)SvPVX_const(sv)) {
1716 if (!NATIVE_IS_INVARIANT(*src)) {
1717 const U8 ch = NATIVE_TO_ASCII(*src);
1718 *dst-- = (U8)UTF8_EIGHT_BIT_LO(ch);
1719 *dst-- = (U8)UTF8_EIGHT_BIT_HI(ch);
1729 if (has_utf8 || uv > 255) {
1730 d = (char*)uvchr_to_utf8((U8*)d, uv);
1732 if (PL_lex_inwhat == OP_TRANS &&
1733 PL_sublex_info.sub_op) {
1734 PL_sublex_info.sub_op->op_private |=
1735 (PL_lex_repl ? OPpTRANS_FROM_UTF
1748 /* \N{LATIN SMALL LETTER A} is a named character */
1752 char* e = strchr(s, '}');
1758 yyerror("Missing right brace on \\N{}");
1762 if (e > s + 2 && s[1] == 'U' && s[2] == '+') {
1764 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
1765 PERL_SCAN_DISALLOW_PREFIX;
1768 uv = grok_hex(s, &len, &flags, NULL);
1770 goto NUM_ESCAPE_INSERT;
1772 res = newSVpvn(s + 1, e - s - 1);
1773 res = new_constant( Nullch, 0, "charnames",
1774 res, Nullsv, "\\N{...}" );
1776 sv_utf8_upgrade(res);
1777 str = SvPV_const(res,len);
1778 #ifdef EBCDIC_NEVER_MIND
1779 /* charnames uses pack U and that has been
1780 * recently changed to do the below uni->native
1781 * mapping, so this would be redundant (and wrong,
1782 * the code point would be doubly converted).
1783 * But leave this in just in case the pack U change
1784 * gets revoked, but the semantics is still
1785 * desireable for charnames. --jhi */
1787 UV uv = utf8_to_uvchr((const U8*)str, 0);
1790 U8 tmpbuf[UTF8_MAXBYTES+1], *d;
1792 d = uvchr_to_utf8(tmpbuf, UNI_TO_NATIVE(uv));
1793 sv_setpvn(res, (char *)tmpbuf, d - tmpbuf);
1794 str = SvPV_const(res, len);
1798 if (!has_utf8 && SvUTF8(res)) {
1799 const char * const ostart = SvPVX_const(sv);
1800 SvCUR_set(sv, d - ostart);
1803 sv_utf8_upgrade(sv);
1804 /* this just broke our allocation above... */
1805 SvGROW(sv, (STRLEN)(send - start));
1806 d = SvPVX(sv) + SvCUR(sv);
1809 if (len > (STRLEN)(e - s + 4)) { /* I _guess_ 4 is \N{} --jhi */
1810 const char * const odest = SvPVX_const(sv);
1812 SvGROW(sv, (SvLEN(sv) + len - (e - s + 4)));
1813 d = SvPVX(sv) + (d - odest);
1815 Copy(str, d, len, char);
1822 yyerror("Missing braces on \\N{}");
1825 /* \c is a control character */
1834 *d++ = NATIVE_TO_NEED(has_utf8,toCTRL(c));
1837 yyerror("Missing control char name in \\c");
1841 /* printf-style backslashes, formfeeds, newlines, etc */
1843 *d++ = NATIVE_TO_NEED(has_utf8,'\b');
1846 *d++ = NATIVE_TO_NEED(has_utf8,'\n');
1849 *d++ = NATIVE_TO_NEED(has_utf8,'\r');
1852 *d++ = NATIVE_TO_NEED(has_utf8,'\f');
1855 *d++ = NATIVE_TO_NEED(has_utf8,'\t');
1858 *d++ = ASCII_TO_NEED(has_utf8,'\033');
1861 *d++ = ASCII_TO_NEED(has_utf8,'\007');
1867 } /* end if (backslash) */
1874 /* If we started with encoded form, or already know we want it
1875 and then encode the next character */
1876 if ((has_utf8 || this_utf8) && !NATIVE_IS_INVARIANT((U8)(*s))) {
1878 const UV uv = (this_utf8) ? utf8n_to_uvchr((U8*)s, send - s, &len, 0) : (UV) ((U8) *s);
1879 const STRLEN need = UNISKIP(NATIVE_TO_UNI(uv));
1882 /* encoded value larger than old, need extra space (NOTE: SvCUR() not set here) */
1883 const STRLEN off = d - SvPVX_const(sv);
1884 d = SvGROW(sv, SvLEN(sv) + (need-len)) + off;
1886 d = (char*)uvchr_to_utf8((U8*)d, uv);
1890 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1892 } /* while loop to process each character */
1894 /* terminate the string and set up the sv */
1896 SvCUR_set(sv, d - SvPVX_const(sv));
1897 if (SvCUR(sv) >= SvLEN(sv))
1898 Perl_croak(aTHX_ "panic: constant overflowed allocated space");
1901 if (PL_encoding && !has_utf8) {
1902 sv_recode_to_utf8(sv, PL_encoding);
1908 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1909 PL_sublex_info.sub_op->op_private |=
1910 (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1914 /* shrink the sv if we allocated more than we used */
1915 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1916 SvPV_shrink_to_cur(sv);
1919 /* return the substring (via yylval) only if we parsed anything */
1920 if (s > PL_bufptr) {
1921 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1922 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1924 ( PL_lex_inwhat == OP_TRANS
1926 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1929 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1936 * Returns TRUE if there's more to the expression (e.g., a subscript),
1939 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1941 * ->[ and ->{ return TRUE
1942 * { and [ outside a pattern are always subscripts, so return TRUE
1943 * if we're outside a pattern and it's not { or [, then return FALSE
1944 * if we're in a pattern and the first char is a {
1945 * {4,5} (any digits around the comma) returns FALSE
1946 * if we're in a pattern and the first char is a [
1948 * [SOMETHING] has a funky algorithm to decide whether it's a
1949 * character class or not. It has to deal with things like
1950 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1951 * anything else returns TRUE
1954 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1957 S_intuit_more(pTHX_ register char *s)
1959 if (PL_lex_brackets)
1961 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1963 if (*s != '{' && *s != '[')
1968 /* In a pattern, so maybe we have {n,m}. */
1985 /* On the other hand, maybe we have a character class */
1988 if (*s == ']' || *s == '^')
1991 /* this is terrifying, and it works */
1992 int weight = 2; /* let's weigh the evidence */
1994 unsigned char un_char = 255, last_un_char;
1995 const char * const send = strchr(s,']');
1996 char tmpbuf[sizeof PL_tokenbuf * 4];
1998 if (!send) /* has to be an expression */
2001 Zero(seen,256,char);
2004 else if (isDIGIT(*s)) {
2006 if (isDIGIT(s[1]) && s[2] == ']')
2012 for (; s < send; s++) {
2013 last_un_char = un_char;
2014 un_char = (unsigned char)*s;
2019 weight -= seen[un_char] * 10;
2020 if (isALNUM_lazy_if(s+1,UTF)) {
2021 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
2022 if ((int)strlen(tmpbuf) > 1
2023 && gv_fetchpv(tmpbuf, 0, SVt_PV))
2028 else if (*s == '$' && s[1] &&
2029 strchr("[#!%*<>()-=",s[1])) {
2030 if (/*{*/ strchr("])} =",s[2]))
2039 if (strchr("wds]",s[1]))
2041 else if (seen['\''] || seen['"'])
2043 else if (strchr("rnftbxcav",s[1]))
2045 else if (isDIGIT(s[1])) {
2047 while (s[1] && isDIGIT(s[1]))
2057 if (strchr("aA01! ",last_un_char))
2059 if (strchr("zZ79~",s[1]))
2061 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
2062 weight -= 5; /* cope with negative subscript */
2065 if (!isALNUM(last_un_char)
2066 && !(last_un_char == '$' || last_un_char == '@'
2067 || last_un_char == '&')
2068 && isALPHA(*s) && s[1] && isALPHA(s[1])) {
2073 if (keyword(tmpbuf, d - tmpbuf))
2076 if (un_char == last_un_char + 1)
2078 weight -= seen[un_char];
2083 if (weight >= 0) /* probably a character class */
2093 * Does all the checking to disambiguate
2095 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
2096 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
2098 * First argument is the stuff after the first token, e.g. "bar".
2100 * Not a method if bar is a filehandle.
2101 * Not a method if foo is a subroutine prototyped to take a filehandle.
2102 * Not a method if it's really "Foo $bar"
2103 * Method if it's "foo $bar"
2104 * Not a method if it's really "print foo $bar"
2105 * Method if it's really "foo package::" (interpreted as package->foo)
2106 * Not a method if bar is known to be a subroutine ("sub bar; foo bar")
2107 * Not a method if bar is a filehandle or package, but is quoted with
2112 S_intuit_method(pTHX_ char *start, GV *gv)
2114 char *s = start + (*start == '$');
2115 char tmpbuf[sizeof PL_tokenbuf];
2123 if ((cv = GvCVu(gv))) {
2124 const char *proto = SvPVX_const(cv);
2134 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
2135 /* start is the beginning of the possible filehandle/object,
2136 * and s is the end of it
2137 * tmpbuf is a copy of it
2140 if (*start == '$') {
2141 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
2146 return *s == '(' ? FUNCMETH : METHOD;
2148 if (!keyword(tmpbuf, len)) {
2149 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
2154 indirgv = gv_fetchpv(tmpbuf, 0, SVt_PVCV);
2155 if (indirgv && GvCVu(indirgv))
2157 /* filehandle or package name makes it a method */
2158 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
2160 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
2161 return 0; /* no assumptions -- "=>" quotes bearword */
2163 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
2164 newSVpvn(tmpbuf,len));
2165 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
2169 return *s == '(' ? FUNCMETH : METHOD;
2177 * Return a string of Perl code to load the debugger. If PERL5DB
2178 * is set, it will return the contents of that, otherwise a
2179 * compile-time require of perl5db.pl.
2186 const char * const pdb = PerlEnv_getenv("PERL5DB");
2190 SETERRNO(0,SS_NORMAL);
2191 return "BEGIN { require 'perl5db.pl' }";
2197 /* Encoded script support. filter_add() effectively inserts a
2198 * 'pre-processing' function into the current source input stream.
2199 * Note that the filter function only applies to the current source file
2200 * (e.g., it will not affect files 'require'd or 'use'd by this one).
2202 * The datasv parameter (which may be NULL) can be used to pass
2203 * private data to this instance of the filter. The filter function
2204 * can recover the SV using the FILTER_DATA macro and use it to
2205 * store private buffers and state information.
2207 * The supplied datasv parameter is upgraded to a PVIO type
2208 * and the IoDIRP/IoANY field is used to store the function pointer,
2209 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
2210 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
2211 * private use must be set using malloc'd pointers.
2215 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
2220 if (!PL_rsfp_filters)
2221 PL_rsfp_filters = newAV();
2223 datasv = NEWSV(255,0);
2224 SvUPGRADE(datasv, SVt_PVIO);
2225 IoANY(datasv) = FPTR2DPTR(void *, funcp); /* stash funcp into spare field */
2226 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
2227 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
2228 IoANY(datasv), SvPV_nolen(datasv)));
2229 av_unshift(PL_rsfp_filters, 1);
2230 av_store(PL_rsfp_filters, 0, datasv) ;
2235 /* Delete most recently added instance of this filter function. */
2237 Perl_filter_del(pTHX_ filter_t funcp)
2242 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", FPTR2DPTR(XPVIO *, funcp)));
2244 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
2246 /* if filter is on top of stack (usual case) just pop it off */
2247 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
2248 if (IoANY(datasv) == FPTR2DPTR(void *, funcp)) {
2249 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
2250 IoANY(datasv) = (void *)NULL;
2251 sv_free(av_pop(PL_rsfp_filters));
2255 /* we need to search for the correct entry and clear it */
2256 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
2260 /* Invoke the idxth filter function for the current rsfp. */
2261 /* maxlen 0 = read one text line */
2263 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
2268 if (!PL_rsfp_filters)
2270 if (idx > AvFILLp(PL_rsfp_filters)) { /* Any more filters? */
2271 /* Provide a default input filter to make life easy. */
2272 /* Note that we append to the line. This is handy. */
2273 DEBUG_P(PerlIO_printf(Perl_debug_log,
2274 "filter_read %d: from rsfp\n", idx));
2278 const int old_len = SvCUR(buf_sv);
2280 /* ensure buf_sv is large enough */
2281 SvGROW(buf_sv, (STRLEN)(old_len + maxlen)) ;
2282 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
2283 if (PerlIO_error(PL_rsfp))
2284 return -1; /* error */
2286 return 0 ; /* end of file */
2288 SvCUR_set(buf_sv, old_len + len) ;
2291 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
2292 if (PerlIO_error(PL_rsfp))
2293 return -1; /* error */
2295 return 0 ; /* end of file */
2298 return SvCUR(buf_sv);
2300 /* Skip this filter slot if filter has been deleted */
2301 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef) {
2302 DEBUG_P(PerlIO_printf(Perl_debug_log,
2303 "filter_read %d: skipped (filter deleted)\n",
2305 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
2307 /* Get function pointer hidden within datasv */
2308 funcp = DPTR2FPTR(filter_t, IoANY(datasv));
2309 DEBUG_P(PerlIO_printf(Perl_debug_log,
2310 "filter_read %d: via function %p (%s)\n",
2311 idx, datasv, SvPV_nolen_const(datasv)));
2312 /* Call function. The function is expected to */
2313 /* call "FILTER_READ(idx+1, buf_sv)" first. */
2314 /* Return: <0:error, =0:eof, >0:not eof */
2315 return (*funcp)(aTHX_ idx, buf_sv, maxlen);
2319 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
2321 #ifdef PERL_CR_FILTER
2322 if (!PL_rsfp_filters) {
2323 filter_add(S_cr_textfilter,NULL);
2326 if (PL_rsfp_filters) {
2328 SvCUR_set(sv, 0); /* start with empty line */
2329 if (FILTER_READ(0, sv, 0) > 0)
2330 return ( SvPVX(sv) ) ;
2335 return (sv_gets(sv, fp, append));
2339 S_find_in_my_stash(pTHX_ const char *pkgname, I32 len)
2343 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2347 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2348 (gv = gv_fetchpv(pkgname, 0, SVt_PVHV)))
2350 return GvHV(gv); /* Foo:: */
2353 /* use constant CLASS => 'MyClass' */
2354 if ((gv = gv_fetchpv(pkgname, 0, SVt_PVCV))) {
2356 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2357 pkgname = SvPV_nolen_const(sv);
2361 return gv_stashpv(pkgname, FALSE);
2365 S_tokenize_use(pTHX_ int is_use, char *s) {
2366 if (PL_expect != XSTATE)
2367 yyerror(Perl_form(aTHX_ "\"%s\" not allowed in expression",
2368 is_use ? "use" : "no"));
2370 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
2371 s = force_version(s, TRUE);
2372 if (*s == ';' || (s = skipspace(s), *s == ';')) {
2373 PL_nextval[PL_nexttoke].opval = Nullop;
2376 else if (*s == 'v') {
2377 s = force_word(s,WORD,FALSE,TRUE,FALSE);
2378 s = force_version(s, FALSE);
2382 s = force_word(s,WORD,FALSE,TRUE,FALSE);
2383 s = force_version(s, FALSE);
2385 yylval.ival = is_use;
2389 static const char* const exp_name[] =
2390 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2391 "ATTRTERM", "TERMBLOCK", "TERMORDORDOR"
2398 Works out what to call the token just pulled out of the input
2399 stream. The yacc parser takes care of taking the ops we return and
2400 stitching them into a tree.
2406 if read an identifier
2407 if we're in a my declaration
2408 croak if they tried to say my($foo::bar)
2409 build the ops for a my() declaration
2410 if it's an access to a my() variable
2411 are we in a sort block?
2412 croak if my($a); $a <=> $b
2413 build ops for access to a my() variable
2414 if in a dq string, and they've said @foo and we can't find @foo
2416 build ops for a bareword
2417 if we already built the token before, use it.
2422 #pragma segment Perl_yylex
2427 register char *s = PL_bufptr;
2433 SV* tmp = newSVpvn("", 0);
2434 PerlIO_printf(Perl_debug_log, "### %"IVdf":LEX_%s/X%s %s\n",
2435 (IV)CopLINE(PL_curcop),
2436 lex_state_names[PL_lex_state],
2437 exp_name[PL_expect],
2438 pv_display(tmp, s, strlen(s), 0, 60));
2441 /* check if there's an identifier for us to look at */
2442 if (PL_pending_ident)
2443 return REPORT(S_pending_ident(aTHX));
2445 /* no identifier pending identification */
2447 switch (PL_lex_state) {
2449 case LEX_NORMAL: /* Some compilers will produce faster */
2450 case LEX_INTERPNORMAL: /* code if we comment these out. */
2454 /* when we've already built the next token, just pull it out of the queue */
2457 yylval = PL_nextval[PL_nexttoke];
2459 PL_lex_state = PL_lex_defer;
2460 PL_expect = PL_lex_expect;
2461 PL_lex_defer = LEX_NORMAL;
2463 return REPORT(PL_nexttype[PL_nexttoke]);
2465 /* interpolated case modifiers like \L \U, including \Q and \E.
2466 when we get here, PL_bufptr is at the \
2468 case LEX_INTERPCASEMOD:
2470 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2471 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2473 /* handle \E or end of string */
2474 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2476 if (PL_lex_casemods) {
2477 const char oldmod = PL_lex_casestack[--PL_lex_casemods];
2478 PL_lex_casestack[PL_lex_casemods] = '\0';
2480 if (PL_bufptr != PL_bufend
2481 && (oldmod == 'L' || oldmod == 'U' || oldmod == 'Q')) {
2483 PL_lex_state = LEX_INTERPCONCAT;
2487 if (PL_bufptr != PL_bufend)
2489 PL_lex_state = LEX_INTERPCONCAT;
2493 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2494 "### Saw case modifier\n"); });
2496 if (s[1] == '\\' && s[2] == 'E') {
2498 PL_lex_state = LEX_INTERPCONCAT;
2503 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2504 tmp = *s, *s = s[2], s[2] = (char)tmp; /* misordered... */
2505 if ((*s == 'L' || *s == 'U') &&
2506 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U'))) {
2507 PL_lex_casestack[--PL_lex_casemods] = '\0';
2510 if (PL_lex_casemods > 10)
2511 Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2512 PL_lex_casestack[PL_lex_casemods++] = *s;
2513 PL_lex_casestack[PL_lex_casemods] = '\0';
2514 PL_lex_state = LEX_INTERPCONCAT;
2515 PL_nextval[PL_nexttoke].ival = 0;
2518 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2520 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2522 PL_nextval[PL_nexttoke].ival = OP_LC;
2524 PL_nextval[PL_nexttoke].ival = OP_UC;
2526 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2528 Perl_croak(aTHX_ "panic: yylex");
2532 if (PL_lex_starts) {
2535 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
2536 if (PL_lex_casemods == 1 && PL_lex_inpat)
2545 case LEX_INTERPPUSH:
2546 return REPORT(sublex_push());
2548 case LEX_INTERPSTART:
2549 if (PL_bufptr == PL_bufend)
2550 return REPORT(sublex_done());
2551 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2552 "### Interpolated variable\n"); });
2554 PL_lex_dojoin = (*PL_bufptr == '@');
2555 PL_lex_state = LEX_INTERPNORMAL;
2556 if (PL_lex_dojoin) {
2557 PL_nextval[PL_nexttoke].ival = 0;
2559 force_ident("\"", '$');
2560 PL_nextval[PL_nexttoke].ival = 0;
2562 PL_nextval[PL_nexttoke].ival = 0;
2564 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2567 if (PL_lex_starts++) {
2569 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
2570 if (!PL_lex_casemods && PL_lex_inpat)
2577 case LEX_INTERPENDMAYBE:
2578 if (intuit_more(PL_bufptr)) {
2579 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2585 if (PL_lex_dojoin) {
2586 PL_lex_dojoin = FALSE;
2587 PL_lex_state = LEX_INTERPCONCAT;
2590 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2591 && SvEVALED(PL_lex_repl))
2593 if (PL_bufptr != PL_bufend)
2594 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2595 PL_lex_repl = Nullsv;
2598 case LEX_INTERPCONCAT:
2600 if (PL_lex_brackets)
2601 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2603 if (PL_bufptr == PL_bufend)
2604 return REPORT(sublex_done());
2606 if (SvIVX(PL_linestr) == '\'') {
2607 SV *sv = newSVsv(PL_linestr);
2610 else if ( PL_hints & HINT_NEW_RE )
2611 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2612 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2616 s = scan_const(PL_bufptr);
2618 PL_lex_state = LEX_INTERPCASEMOD;
2620 PL_lex_state = LEX_INTERPSTART;
2623 if (s != PL_bufptr) {
2624 PL_nextval[PL_nexttoke] = yylval;
2627 if (PL_lex_starts++) {
2628 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
2629 if (!PL_lex_casemods && PL_lex_inpat)
2642 PL_lex_state = LEX_NORMAL;
2643 s = scan_formline(PL_bufptr);
2644 if (!PL_lex_formbrack)
2650 PL_oldoldbufptr = PL_oldbufptr;
2656 if (isIDFIRST_lazy_if(s,UTF))
2658 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2661 goto fake_eof; /* emulate EOF on ^D or ^Z */
2666 if (PL_lex_brackets) {
2667 yyerror(PL_lex_formbrack
2668 ? "Format not terminated"
2669 : "Missing right curly or square bracket");
2671 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2672 "### Tokener got EOF\n");
2676 if (s++ < PL_bufend)
2677 goto retry; /* ignore stray nulls */
2680 if (!PL_in_eval && !PL_preambled) {
2681 PL_preambled = TRUE;
2682 sv_setpv(PL_linestr,incl_perldb());
2683 if (SvCUR(PL_linestr))
2684 sv_catpvn(PL_linestr,";", 1);
2686 while(AvFILLp(PL_preambleav) >= 0) {
2687 SV *tmpsv = av_shift(PL_preambleav);
2688 sv_catsv(PL_linestr, tmpsv);
2689 sv_catpvn(PL_linestr, ";", 1);
2692 sv_free((SV*)PL_preambleav);
2693 PL_preambleav = NULL;
2695 if (PL_minus_n || PL_minus_p) {
2696 sv_catpv(PL_linestr, "LINE: while (<>) {");
2698 sv_catpv(PL_linestr,"chomp;");
2701 if ((*PL_splitstr == '/' || *PL_splitstr == '\''
2702 || *PL_splitstr == '"')
2703 && strchr(PL_splitstr + 1, *PL_splitstr))
2704 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s);", PL_splitstr);
2706 /* "q\0${splitstr}\0" is legal perl. Yes, even NUL
2707 bytes can be used as quoting characters. :-) */
2708 /* The count here deliberately includes the NUL
2709 that terminates the C string constant. This
2710 embeds the opening NUL into the string. */
2711 const char *splits = PL_splitstr;
2712 sv_catpvn(PL_linestr, "our @F=split(q", 15);
2715 if (*splits == '\\')
2716 sv_catpvn(PL_linestr, splits, 1);
2717 sv_catpvn(PL_linestr, splits, 1);
2718 } while (*splits++);
2719 /* This loop will embed the trailing NUL of
2720 PL_linestr as the last thing it does before
2722 sv_catpvn(PL_linestr, ");", 2);
2726 sv_catpv(PL_linestr,"our @F=split(' ');");
2729 sv_catpvn(PL_linestr, "\n", 1);
2730 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2731 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2732 PL_last_lop = PL_last_uni = Nullch;
2733 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2734 SV * const sv = NEWSV(85,0);
2736 sv_upgrade(sv, SVt_PVMG);
2737 sv_setsv(sv,PL_linestr);
2740 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2745 bof = PL_rsfp ? TRUE : FALSE;
2746 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2749 if (PL_preprocess && !PL_in_eval)
2750 (void)PerlProc_pclose(PL_rsfp);
2751 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2752 PerlIO_clearerr(PL_rsfp);
2754 (void)PerlIO_close(PL_rsfp);
2756 PL_doextract = FALSE;
2758 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2759 sv_setpv(PL_linestr,PL_minus_p
2760 ? ";}continue{print;}" : ";}");
2761 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2762 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2763 PL_last_lop = PL_last_uni = Nullch;
2764 PL_minus_n = PL_minus_p = 0;
2767 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2768 PL_last_lop = PL_last_uni = Nullch;
2769 sv_setpvn(PL_linestr,"",0);
2770 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2772 /* If it looks like the start of a BOM or raw UTF-16,
2773 * check if it in fact is. */
2779 #ifdef PERLIO_IS_STDIO
2780 # ifdef __GNU_LIBRARY__
2781 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
2782 # define FTELL_FOR_PIPE_IS_BROKEN
2786 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
2787 # define FTELL_FOR_PIPE_IS_BROKEN
2792 #ifdef FTELL_FOR_PIPE_IS_BROKEN
2793 /* This loses the possibility to detect the bof
2794 * situation on perl -P when the libc5 is being used.
2795 * Workaround? Maybe attach some extra state to PL_rsfp?
2798 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
2800 bof = PerlIO_tell(PL_rsfp) == (Off_t)SvCUR(PL_linestr);
2803 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2804 s = swallow_bom((U8*)s);
2808 /* Incest with pod. */
2809 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2810 sv_setpvn(PL_linestr, "", 0);
2811 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2812 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2813 PL_last_lop = PL_last_uni = Nullch;
2814 PL_doextract = FALSE;
2818 } while (PL_doextract);
2819 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2820 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2821 SV * const sv = NEWSV(85,0);
2823 sv_upgrade(sv, SVt_PVMG);
2824 sv_setsv(sv,PL_linestr);
2827 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2829 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2830 PL_last_lop = PL_last_uni = Nullch;
2831 if (CopLINE(PL_curcop) == 1) {
2832 while (s < PL_bufend && isSPACE(*s))
2834 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2838 if (*s == '#' && *(s+1) == '!')
2840 #ifdef ALTERNATE_SHEBANG
2842 static char const as[] = ALTERNATE_SHEBANG;
2843 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2844 d = s + (sizeof(as) - 1);
2846 #endif /* ALTERNATE_SHEBANG */
2855 while (*d && !isSPACE(*d))
2859 #ifdef ARG_ZERO_IS_SCRIPT
2860 if (ipathend > ipath) {
2862 * HP-UX (at least) sets argv[0] to the script name,
2863 * which makes $^X incorrect. And Digital UNIX and Linux,
2864 * at least, set argv[0] to the basename of the Perl
2865 * interpreter. So, having found "#!", we'll set it right.
2868 = GvSV(gv_fetchpv("\030", GV_ADD, SVt_PV)); /* $^X */
2869 assert(SvPOK(x) || SvGMAGICAL(x));
2870 if (sv_eq(x, CopFILESV(PL_curcop))) {
2871 sv_setpvn(x, ipath, ipathend - ipath);
2877 const char *bstart = SvPV_const(CopFILESV(PL_curcop),blen);
2878 const char * const lstart = SvPV_const(x,llen);
2880 bstart += blen - llen;
2881 if (strnEQ(bstart, lstart, llen) && bstart[-1] == '/') {
2882 sv_setpvn(x, ipath, ipathend - ipath);
2887 TAINT_NOT; /* $^X is always tainted, but that's OK */
2889 #endif /* ARG_ZERO_IS_SCRIPT */
2894 d = instr(s,"perl -");
2896 d = instr(s,"perl");
2898 /* avoid getting into infinite loops when shebang
2899 * line contains "Perl" rather than "perl" */
2901 for (d = ipathend-4; d >= ipath; --d) {
2902 if ((*d == 'p' || *d == 'P')
2903 && !ibcmp(d, "perl", 4))
2913 #ifdef ALTERNATE_SHEBANG
2915 * If the ALTERNATE_SHEBANG on this system starts with a
2916 * character that can be part of a Perl expression, then if
2917 * we see it but not "perl", we're probably looking at the
2918 * start of Perl code, not a request to hand off to some
2919 * other interpreter. Similarly, if "perl" is there, but
2920 * not in the first 'word' of the line, we assume the line
2921 * contains the start of the Perl program.
2923 if (d && *s != '#') {
2924 const char *c = ipath;
2925 while (*c && !strchr("; \t\r\n\f\v#", *c))
2928 d = Nullch; /* "perl" not in first word; ignore */
2930 *s = '#'; /* Don't try to parse shebang line */
2932 #endif /* ALTERNATE_SHEBANG */
2933 #ifndef MACOS_TRADITIONAL
2938 !instr(s,"indir") &&
2939 instr(PL_origargv[0],"perl"))
2946 while (s < PL_bufend && isSPACE(*s))
2948 if (s < PL_bufend) {
2949 Newxz(newargv,PL_origargc+3,char*);
2951 while (s < PL_bufend && !isSPACE(*s))
2954 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2957 newargv = PL_origargv;
2960 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
2962 Perl_croak(aTHX_ "Can't exec %s", ipath);
2966 const U32 oldpdb = PL_perldb;
2967 const bool oldn = PL_minus_n;
2968 const bool oldp = PL_minus_p;
2970 while (*d && !isSPACE(*d)) d++;
2971 while (SPACE_OR_TAB(*d)) d++;
2974 const bool switches_done = PL_doswitches;
2976 if (*d == 'M' || *d == 'm' || *d == 'C') {
2977 const char * const m = d;
2978 while (*d && !isSPACE(*d)) d++;
2979 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2982 d = moreswitches(d);
2984 if (PL_doswitches && !switches_done) {
2985 int argc = PL_origargc;
2986 char **argv = PL_origargv;
2989 } while (argc && argv[0][0] == '-' && argv[0][1]);
2990 init_argv_symbols(argc,argv);
2992 if ((PERLDB_LINE && !oldpdb) ||
2993 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2994 /* if we have already added "LINE: while (<>) {",
2995 we must not do it again */
2997 sv_setpvn(PL_linestr, "", 0);
2998 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2999 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3000 PL_last_lop = PL_last_uni = Nullch;
3001 PL_preambled = FALSE;
3003 (void)gv_fetchfile(PL_origfilename);
3006 if (PL_doswitches && !switches_done) {
3007 int argc = PL_origargc;
3008 char **argv = PL_origargv;
3011 } while (argc && argv[0][0] == '-' && argv[0][1]);
3012 init_argv_symbols(argc,argv);
3018 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
3020 PL_lex_state = LEX_FORMLINE;
3025 #ifdef PERL_STRICT_CR
3026 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
3028 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
3030 case ' ': case '\t': case '\f': case 013:
3031 #ifdef MACOS_TRADITIONAL
3038 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
3039 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
3040 /* handle eval qq[#line 1 "foo"\n ...] */
3041 CopLINE_dec(PL_curcop);
3045 while (s < d && *s != '\n')
3049 else if (s > d) /* Found by Ilya: feed random input to Perl. */
3050 Perl_croak(aTHX_ "panic: input overflow");
3052 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
3054 PL_lex_state = LEX_FORMLINE;
3064 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
3072 while (s < PL_bufend && SPACE_OR_TAB(*s))
3075 if (strnEQ(s,"=>",2)) {
3076 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
3077 DEBUG_T( { S_printbuf(aTHX_
3078 "### Saw unary minus before =>, forcing word %s\n", s);
3080 OPERATOR('-'); /* unary minus */
3082 PL_last_uni = PL_oldbufptr;
3084 case 'r': ftst = OP_FTEREAD; break;
3085 case 'w': ftst = OP_FTEWRITE; break;
3086 case 'x': ftst = OP_FTEEXEC; break;
3087 case 'o': ftst = OP_FTEOWNED; break;
3088 case 'R': ftst = OP_FTRREAD; break;
3089 case 'W': ftst = OP_FTRWRITE; break;
3090 case 'X': ftst = OP_FTREXEC; break;
3091 case 'O': ftst = OP_FTROWNED; break;
3092 case 'e': ftst = OP_FTIS; break;
3093 case 'z': ftst = OP_FTZERO; break;
3094 case 's': ftst = OP_FTSIZE; break;
3095 case 'f': ftst = OP_FTFILE; break;
3096 case 'd': ftst = OP_FTDIR; break;
3097 case 'l': ftst = OP_FTLINK; break;
3098 case 'p': ftst = OP_FTPIPE; break;
3099 case 'S': ftst = OP_FTSOCK; break;
3100 case 'u': ftst = OP_FTSUID; break;
3101 case 'g': ftst = OP_FTSGID; break;
3102 case 'k': ftst = OP_FTSVTX; break;
3103 case 'b': ftst = OP_FTBLK; break;
3104 case 'c': ftst = OP_FTCHR; break;
3105 case 't': ftst = OP_FTTTY; break;
3106 case 'T': ftst = OP_FTTEXT; break;
3107 case 'B': ftst = OP_FTBINARY; break;
3108 case 'M': case 'A': case 'C':
3109 gv_fetchpv("\024",GV_ADD, SVt_PV);
3111 case 'M': ftst = OP_FTMTIME; break;
3112 case 'A': ftst = OP_FTATIME; break;
3113 case 'C': ftst = OP_FTCTIME; break;
3121 PL_last_lop_op = (OPCODE)ftst;
3122 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3123 "### Saw file test %c\n", (int)tmp);
3128 /* Assume it was a minus followed by a one-letter named
3129 * subroutine call (or a -bareword), then. */
3130 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3131 "### '-%c' looked like a file test but was not\n",
3138 const char tmp = *s++;
3141 if (PL_expect == XOPERATOR)
3146 else if (*s == '>') {
3149 if (isIDFIRST_lazy_if(s,UTF)) {
3150 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
3158 if (PL_expect == XOPERATOR)
3161 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
3163 OPERATOR('-'); /* unary minus */
3169 const char tmp = *s++;
3172 if (PL_expect == XOPERATOR)
3177 if (PL_expect == XOPERATOR)
3180 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
3187 if (PL_expect != XOPERATOR) {
3188 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3189 PL_expect = XOPERATOR;
3190 force_ident(PL_tokenbuf, '*');
3203 if (PL_expect == XOPERATOR) {
3207 PL_tokenbuf[0] = '%';
3208 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
3209 if (!PL_tokenbuf[1]) {
3212 PL_pending_ident = '%';
3223 && (PL_expect == XOPERATOR || PL_expect == XTERMORDORDOR)
3224 && FEATURE_IS_ENABLED("~~", 2))
3231 const char tmp = *s++;
3237 goto just_a_word_zero_gv;
3240 switch (PL_expect) {
3243 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
3245 PL_bufptr = s; /* update in case we back off */
3251 PL_expect = XTERMBLOCK;
3255 while (isIDFIRST_lazy_if(s,UTF)) {
3257 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3258 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
3259 if (tmp < 0) tmp = -tmp;
3275 d = scan_str(d,TRUE,TRUE);
3277 /* MUST advance bufptr here to avoid bogus
3278 "at end of line" context messages from yyerror().
3280 PL_bufptr = s + len;
3281 yyerror("Unterminated attribute parameter in attribute list");
3284 return REPORT(0); /* EOF indicator */
3288 SV *sv = newSVpvn(s, len);
3289 sv_catsv(sv, PL_lex_stuff);
3290 attrs = append_elem(OP_LIST, attrs,
3291 newSVOP(OP_CONST, 0, sv));
3292 SvREFCNT_dec(PL_lex_stuff);
3293 PL_lex_stuff = Nullsv;
3296 if (len == 6 && strnEQ(s, "unique", len)) {
3297 if (PL_in_my == KEY_our)
3299 GvUNIQUE_on(cGVOPx_gv(yylval.opval));
3301 ; /* skip to avoid loading attributes.pm */
3304 Perl_croak(aTHX_ "The 'unique' attribute may only be applied to 'our' variables");
3307 /* NOTE: any CV attrs applied here need to be part of
3308 the CVf_BUILTIN_ATTRS define in cv.h! */
3309 else if (!PL_in_my && len == 6 && strnEQ(s, "lvalue", len))
3310 CvLVALUE_on(PL_compcv);
3311 else if (!PL_in_my && len == 6 && strnEQ(s, "locked", len))
3312 CvLOCKED_on(PL_compcv);
3313 else if (!PL_in_my && len == 6 && strnEQ(s, "method", len))
3314 CvMETHOD_on(PL_compcv);
3315 else if (!PL_in_my && len == 9 && strnEQ(s, "assertion", len))
3316 CvASSERTION_on(PL_compcv);
3317 /* After we've set the flags, it could be argued that
3318 we don't need to do the attributes.pm-based setting
3319 process, and shouldn't bother appending recognized
3320 flags. To experiment with that, uncomment the
3321 following "else". (Note that's already been
3322 uncommented. That keeps the above-applied built-in
3323 attributes from being intercepted (and possibly
3324 rejected) by a package's attribute routines, but is
3325 justified by the performance win for the common case
3326 of applying only built-in attributes.) */
3328 attrs = append_elem(OP_LIST, attrs,
3329 newSVOP(OP_CONST, 0,
3333 if (*s == ':' && s[1] != ':')
3336 break; /* require real whitespace or :'s */
3340 = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
3341 if (*s != ';' && *s != '}' && *s != tmp
3342 && (tmp != '=' || *s != ')')) {
3343 const char q = ((*s == '\'') ? '"' : '\'');
3344 /* If here for an expression, and parsed no attrs, back
3346 if (tmp == '=' && !attrs) {
3350 /* MUST advance bufptr here to avoid bogus "at end of line"
3351 context messages from yyerror().
3355 ? Perl_form(aTHX_ "Invalid separator character "
3356 "%c%c%c in attribute list", q, *s, q)
3357 : "Unterminated attribute list" );
3365 PL_nextval[PL_nexttoke].opval = attrs;
3373 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
3374 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
3382 const char tmp = *s++;
3387 const char tmp = *s++;
3395 if (PL_lex_brackets <= 0)
3396 yyerror("Unmatched right square bracket");
3399 if (PL_lex_state == LEX_INTERPNORMAL) {
3400 if (PL_lex_brackets == 0) {
3401 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
3402 PL_lex_state = LEX_INTERPEND;
3409 if (PL_lex_brackets > 100) {
3410 Renew(PL_lex_brackstack, PL_lex_brackets + 10, char);
3412 switch (PL_expect) {
3414 if (PL_lex_formbrack) {
3418 if (PL_oldoldbufptr == PL_last_lop)
3419 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3421 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3422 OPERATOR(HASHBRACK);
3424 while (s < PL_bufend && SPACE_OR_TAB(*s))
3427 PL_tokenbuf[0] = '\0';
3428 if (d < PL_bufend && *d == '-') {
3429 PL_tokenbuf[0] = '-';
3431 while (d < PL_bufend && SPACE_OR_TAB(*d))
3434 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3435 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3437 while (d < PL_bufend && SPACE_OR_TAB(*d))
3440 const char minus = (PL_tokenbuf[0] == '-');
3441 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3449 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3454 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3459 if (PL_oldoldbufptr == PL_last_lop)
3460 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3462 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3465 if (PL_expect == XREF && PL_lex_state == LEX_INTERPNORMAL) {
3467 /* This hack is to get the ${} in the message. */
3469 yyerror("syntax error");
3472 OPERATOR(HASHBRACK);
3474 /* This hack serves to disambiguate a pair of curlies
3475 * as being a block or an anon hash. Normally, expectation
3476 * determines that, but in cases where we're not in a
3477 * position to expect anything in particular (like inside
3478 * eval"") we have to resolve the ambiguity. This code
3479 * covers the case where the first term in the curlies is a
3480 * quoted string. Most other cases need to be explicitly
3481 * disambiguated by prepending a "+" before the opening
3482 * curly in order to force resolution as an anon hash.
3484 * XXX should probably propagate the outer expectation
3485 * into eval"" to rely less on this hack, but that could
3486 * potentially break current behavior of eval"".
3490 if (*s == '\'' || *s == '"' || *s == '`') {
3491 /* common case: get past first string, handling escapes */
3492 for (t++; t < PL_bufend && *t != *s;)
3493 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3497 else if (*s == 'q') {
3500 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3503 /* skip q//-like construct */
3505 char open, close, term;
3508 while (t < PL_bufend && isSPACE(*t))
3510 /* check for q => */
3511 if (t+1 < PL_bufend && t[0] == '=' && t[1] == '>') {
3512 OPERATOR(HASHBRACK);
3516 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3520 for (t++; t < PL_bufend; t++) {
3521 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3523 else if (*t == open)
3527 for (t++; t < PL_bufend; t++) {
3528 if (*t == '\\' && t+1 < PL_bufend)
3530 else if (*t == close && --brackets <= 0)
3532 else if (*t == open)
3539 /* skip plain q word */
3540 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3543 else if (isALNUM_lazy_if(t,UTF)) {
3545 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3548 while (t < PL_bufend && isSPACE(*t))
3550 /* if comma follows first term, call it an anon hash */
3551 /* XXX it could be a comma expression with loop modifiers */
3552 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3553 || (*t == '=' && t[1] == '>')))
3554 OPERATOR(HASHBRACK);
3555 if (PL_expect == XREF)
3558 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3564 yylval.ival = CopLINE(PL_curcop);
3565 if (isSPACE(*s) || *s == '#')
3566 PL_copline = NOLINE; /* invalidate current command line number */
3571 if (PL_lex_brackets <= 0)
3572 yyerror("Unmatched right curly bracket");
3574 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3575 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3576 PL_lex_formbrack = 0;
3577 if (PL_lex_state == LEX_INTERPNORMAL) {
3578 if (PL_lex_brackets == 0) {
3579 if (PL_expect & XFAKEBRACK) {
3580 PL_expect &= XENUMMASK;
3581 PL_lex_state = LEX_INTERPEND;
3583 return yylex(); /* ignore fake brackets */
3585 if (*s == '-' && s[1] == '>')
3586 PL_lex_state = LEX_INTERPENDMAYBE;
3587 else if (*s != '[' && *s != '{')
3588 PL_lex_state = LEX_INTERPEND;
3591 if (PL_expect & XFAKEBRACK) {
3592 PL_expect &= XENUMMASK;
3594 return yylex(); /* ignore fake brackets */
3603 if (PL_expect == XOPERATOR) {
3604 if (PL_bufptr == PL_linestart && ckWARN(WARN_SEMICOLON)
3605 && isIDFIRST_lazy_if(s,UTF))
3607 CopLINE_dec(PL_curcop);
3608 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
3609 CopLINE_inc(PL_curcop);
3614 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3616 PL_expect = XOPERATOR;
3617 force_ident(PL_tokenbuf, '&');
3621 yylval.ival = (OPpENTERSUB_AMPER<<8);
3633 const char tmp = *s++;
3640 if (tmp && isSPACE(*s) && ckWARN(WARN_SYNTAX)
3641 && strchr("+-*/%.^&|<",tmp))
3642 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3643 "Reversed %c= operator",(int)tmp);
3645 if (PL_expect == XSTATE && isALPHA(tmp) &&
3646 (s == PL_linestart+1 || s[-2] == '\n') )
3648 if (PL_in_eval && !PL_rsfp) {
3653 if (strnEQ(s,"=cut",4)) {
3667 PL_doextract = TRUE;
3671 if (PL_lex_brackets < PL_lex_formbrack) {
3673 #ifdef PERL_STRICT_CR
3674 for (t = s; SPACE_OR_TAB(*t); t++) ;
3676 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3678 if (*t == '\n' || *t == '#') {
3689 const char tmp = *s++;
3691 /* was this !=~ where !~ was meant?
3692 * warn on m:!=~\s+([/?]|[msy]\W|tr\W): */
3694 if (*s == '~' && ckWARN(WARN_SYNTAX)) {
3695 const char *t = s+1;
3697 while (t < PL_bufend && isSPACE(*t))
3700 if (*t == '/' || *t == '?' ||
3701 ((*t == 'm' || *t == 's' || *t == 'y')
3702 && !isALNUM(t[1])) ||
3703 (*t == 't' && t[1] == 'r' && !isALNUM(t[2])))
3704 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3705 "!=~ should be !~");
3715 if (PL_expect != XOPERATOR) {
3716 if (s[1] != '<' && !strchr(s,'>'))
3719 s = scan_heredoc(s);
3721 s = scan_inputsymbol(s);
3722 TERM(sublex_start());
3728 SHop(OP_LEFT_SHIFT);
3742 const char tmp = *s++;
3744 SHop(OP_RIGHT_SHIFT);
3754 if (PL_expect == XOPERATOR) {
3755 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3758 return REPORT(','); /* grandfather non-comma-format format */
3762 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3763 PL_tokenbuf[0] = '@';
3764 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3765 sizeof PL_tokenbuf - 1, FALSE);
3766 if (PL_expect == XOPERATOR)
3767 no_op("Array length", s);
3768 if (!PL_tokenbuf[1])
3770 PL_expect = XOPERATOR;
3771 PL_pending_ident = '#';
3775 PL_tokenbuf[0] = '$';
3776 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3777 sizeof PL_tokenbuf - 1, FALSE);
3778 if (PL_expect == XOPERATOR)
3780 if (!PL_tokenbuf[1]) {
3782 yyerror("Final $ should be \\$ or $name");
3786 /* This kludge not intended to be bulletproof. */
3787 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3788 yylval.opval = newSVOP(OP_CONST, 0,
3789 newSViv(PL_compiling.cop_arybase));
3790 yylval.opval->op_private = OPpCONST_ARYBASE;
3796 const char tmp = *s;
3797 if (PL_lex_state == LEX_NORMAL)
3800 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop)
3801 && intuit_more(s)) {
3803 PL_tokenbuf[0] = '@';
3804 if (ckWARN(WARN_SYNTAX)) {
3807 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3810 PL_bufptr = skipspace(PL_bufptr);
3811 while (t < PL_bufend && *t != ']')
3813 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3814 "Multidimensional syntax %.*s not supported",
3815 (t - PL_bufptr) + 1, PL_bufptr);
3819 else if (*s == '{') {
3821 PL_tokenbuf[0] = '%';
3822 if (strEQ(PL_tokenbuf+1, "SIG") && ckWARN(WARN_SYNTAX)
3823 && (t = strchr(s, '}')) && (t = strchr(t, '=')))
3825 char tmpbuf[sizeof PL_tokenbuf];
3826 for (t++; isSPACE(*t); t++) ;
3827 if (isIDFIRST_lazy_if(t,UTF)) {
3829 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE,
3831 for (; isSPACE(*t); t++) ;
3832 if (*t == ';' && get_cv(tmpbuf, FALSE))
3833 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3834 "You need to quote \"%s\"",
3841 PL_expect = XOPERATOR;
3842 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3843 const bool islop = (PL_last_lop == PL_oldoldbufptr);
3844 if (!islop || PL_last_lop_op == OP_GREPSTART)
3845 PL_expect = XOPERATOR;
3846 else if (strchr("$@\"'`q", *s))
3847 PL_expect = XTERM; /* e.g. print $fh "foo" */
3848 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3849 PL_expect = XTERM; /* e.g. print $fh &sub */
3850 else if (isIDFIRST_lazy_if(s,UTF)) {
3851 char tmpbuf[sizeof PL_tokenbuf];
3853 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3854 if ((t2 = keyword(tmpbuf, len))) {
3855 /* binary operators exclude handle interpretations */
3867 PL_expect = XTERM; /* e.g. print $fh length() */
3872 PL_expect = XTERM; /* e.g. print $fh subr() */
3875 else if (isDIGIT(*s))
3876 PL_expect = XTERM; /* e.g. print $fh 3 */
3877 else if (*s == '.' && isDIGIT(s[1]))
3878 PL_expect = XTERM; /* e.g. print $fh .3 */
3879 else if ((*s == '?' || *s == '-' || *s == '+')
3880 && !isSPACE(s[1]) && s[1] != '=')
3881 PL_expect = XTERM; /* e.g. print $fh -1 */
3882 else if (*s == '/' && !isSPACE(s[1]) && s[1] != '='
3884 PL_expect = XTERM; /* e.g. print $fh /.../
3885 XXX except DORDOR operator
3887 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2])
3889 PL_expect = XTERM; /* print $fh <<"EOF" */
3892 PL_pending_ident = '$';
3896 if (PL_expect == XOPERATOR)
3898 PL_tokenbuf[0] = '@';
3899 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3900 if (!PL_tokenbuf[1]) {
3903 if (PL_lex_state == LEX_NORMAL)
3905 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3907 PL_tokenbuf[0] = '%';
3909 /* Warn about @ where they meant $. */
3910 if (*s == '[' || *s == '{') {
3911 if (ckWARN(WARN_SYNTAX)) {
3912 const char *t = s + 1;
3913 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3915 if (*t == '}' || *t == ']') {
3917 PL_bufptr = skipspace(PL_bufptr);
3918 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3919 "Scalar value %.*s better written as $%.*s",
3920 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3925 PL_pending_ident = '@';
3928 case '/': /* may be division, defined-or, or pattern */
3929 if (PL_expect == XTERMORDORDOR && s[1] == '/') {
3933 case '?': /* may either be conditional or pattern */
3934 if(PL_expect == XOPERATOR) {
3942 /* A // operator. */
3952 /* Disable warning on "study /blah/" */
3953 if (PL_oldoldbufptr == PL_last_uni
3954 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3955 || memNE(PL_last_uni, "study", 5)
3956 || isALNUM_lazy_if(PL_last_uni+5,UTF)
3959 s = scan_pat(s,OP_MATCH);
3960 TERM(sublex_start());
3964 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3965 #ifdef PERL_STRICT_CR
3968 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3970 && (s == PL_linestart || s[-1] == '\n') )
3972 PL_lex_formbrack = 0;
3976 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3982 yylval.ival = OPf_SPECIAL;
3988 if (PL_expect != XOPERATOR)
3993 case '0': case '1': case '2': case '3': case '4':
3994 case '5': case '6': case '7': case '8': case '9':
3995 s = scan_num(s, &yylval);
3996 DEBUG_T( { S_printbuf(aTHX_ "### Saw number in %s\n", s); } );
3997 if (PL_expect == XOPERATOR)
4002 s = scan_str(s,FALSE,FALSE);
4003 DEBUG_T( { S_printbuf(aTHX_ "### Saw string before %s\n", s); } );
4004 if (PL_expect == XOPERATOR) {
4005 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
4008 return REPORT(','); /* grandfather non-comma-format format */
4014 missingterm((char*)0);
4015 yylval.ival = OP_CONST;
4016 TERM(sublex_start());
4019 s = scan_str(s,FALSE,FALSE);
4020 DEBUG_T( { S_printbuf(aTHX_ "### Saw string before %s\n", s); } );
4021 if (PL_expect == XOPERATOR) {
4022 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
4025 return REPORT(','); /* grandfather non-comma-format format */
4031 missingterm((char*)0);
4032 yylval.ival = OP_CONST;
4033 /* FIXME. I think that this can be const if char *d is replaced by
4034 more localised variables. */
4035 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
4036 if (*d == '$' || *d == '@' || *d == '\\' || !UTF8_IS_INVARIANT((U8)*d)) {
4037 yylval.ival = OP_STRINGIFY;
4041 TERM(sublex_start());
4044 s = scan_str(s,FALSE,FALSE);
4045 DEBUG_T( { S_printbuf(aTHX_ "### Saw backtick string before %s\n", s); } );
4046 if (PL_expect == XOPERATOR)
4047 no_op("Backticks",s);
4049 missingterm((char*)0);
4050 yylval.ival = OP_BACKTICK;
4052 TERM(sublex_start());
4056 if (PL_lex_inwhat && isDIGIT(*s) && ckWARN(WARN_SYNTAX))
4057 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),"Can't use \\%c to mean $%c in expression",
4059 if (PL_expect == XOPERATOR)
4060 no_op("Backslash",s);
4064 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
4065 char *start = s + 2;
4066 while (isDIGIT(*start) || *start == '_')
4068 if (*start == '.' && isDIGIT(start[1])) {
4069 s = scan_num(s, &yylval);
4072 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
4073 else if (!isALPHA(*start) && (PL_expect == XTERM
4074 || PL_expect == XREF || PL_expect == XSTATE
4075 || PL_expect == XTERMORDORDOR)) {
4076 const char c = *start;
4079 gv = gv_fetchpv(s, 0, SVt_PVCV);
4082 s = scan_num(s, &yylval);
4089 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
4125 I32 orig_keyword = 0;
4130 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4132 /* Some keywords can be followed by any delimiter, including ':' */
4133 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
4134 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
4135 (PL_tokenbuf[0] == 'q' &&
4136 strchr("qwxr", PL_tokenbuf[1])))));
4138 /* x::* is just a word, unless x is "CORE" */
4139 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
4143 while (d < PL_bufend && isSPACE(*d))
4144 d++; /* no comments skipped here, or s### is misparsed */
4146 /* Is this a label? */
4147 if (!tmp && PL_expect == XSTATE
4148 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
4150 yylval.pval = savepv(PL_tokenbuf);
4155 /* Check for keywords */
4156 tmp = keyword(PL_tokenbuf, len);
4158 /* Is this a word before a => operator? */
4159 if (*d == '=' && d[1] == '>') {
4162 = (OP*)newSVOP(OP_CONST, 0,
4163 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
4164 yylval.opval->op_private = OPpCONST_BARE;
4168 if (tmp < 0) { /* second-class keyword? */
4169 GV *ogv = Nullgv; /* override (winner) */
4170 GV *hgv = Nullgv; /* hidden (loser) */
4171 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
4173 if ((gv = gv_fetchpv(PL_tokenbuf, 0, SVt_PVCV)) &&
4176 if (GvIMPORTED_CV(gv))
4178 else if (! CvMETHOD(cv))
4182 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
4183 (gv = *gvp) != (GV*)&PL_sv_undef &&
4184 GvCVu(gv) && GvIMPORTED_CV(gv))
4191 tmp = 0; /* overridden by import or by GLOBAL */
4194 && -tmp==KEY_lock /* XXX generalizable kludge */
4196 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
4198 tmp = 0; /* any sub overrides "weak" keyword */
4203 && PL_expect != XOPERATOR
4204 && PL_expect != XTERMORDORDOR)
4206 /* any sub overrides the "err" keyword, except when really an
4207 * operator is expected */
4210 else { /* no override */
4212 if (tmp == KEY_dump && ckWARN(WARN_MISC)) {
4213 Perl_warner(aTHX_ packWARN(WARN_MISC),
4214 "dump() better written as CORE::dump()");
4218 if (hgv && tmp != KEY_x && tmp != KEY_CORE
4219 && ckWARN(WARN_AMBIGUOUS)) /* never ambiguous */
4220 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4221 "Ambiguous call resolved as CORE::%s(), %s",
4222 GvENAME(hgv), "qualify as such or use &");
4229 default: /* not a keyword */
4230 /* Trade off - by using this evil construction we can pull the
4231 variable gv into the block labelled keylookup. If not, then
4232 we have to give it function scope so that the goto from the
4233 earlier ':' case doesn't bypass the initialisation. */
4235 just_a_word_zero_gv:
4242 const char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
4245 /* Get the rest if it looks like a package qualifier */
4247 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
4249 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
4252 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
4253 *s == '\'' ? "'" : "::");
4258 if (PL_expect == XOPERATOR) {
4259 if (PL_bufptr == PL_linestart) {
4260 CopLINE_dec(PL_curcop);
4261 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
4262 CopLINE_inc(PL_curcop);
4265 no_op("Bareword",s);
4268 /* Look for a subroutine with this name in current package,
4269 unless name is "Foo::", in which case Foo is a bearword
4270 (and a package name). */
4273 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
4275 if (ckWARN(WARN_BAREWORD)
4276 && ! gv_fetchpv(PL_tokenbuf, 0, SVt_PVHV))
4277 Perl_warner(aTHX_ packWARN(WARN_BAREWORD),
4278 "Bareword \"%s\" refers to nonexistent package",
4281 PL_tokenbuf[len] = '\0';
4288 gv = gv_fetchpv(PL_tokenbuf, 0, SVt_PVCV);
4291 /* if we saw a global override before, get the right name */
4294 sv = newSVpvn("CORE::GLOBAL::",14);
4295 sv_catpv(sv,PL_tokenbuf);
4298 /* If len is 0, newSVpv does strlen(), which is correct.
4299 If len is non-zero, then it will be the true length,
4300 and so the scalar will be created correctly. */
4301 sv = newSVpv(PL_tokenbuf,len);
4304 /* Presume this is going to be a bareword of some sort. */
4307 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
4308 yylval.opval->op_private = OPpCONST_BARE;
4309 /* UTF-8 package name? */
4310 if (UTF && !IN_BYTES &&
4311 is_utf8_string((U8*)SvPVX_const(sv), SvCUR(sv)))
4314 /* And if "Foo::", then that's what it certainly is. */
4319 /* Do the explicit type check so that we don't need to force
4320 the initialisation of the symbol table to have a real GV.
4321 Beware - gv may not really be a PVGV, cv may not really be
4322 a PVCV, (because of the space optimisations that gv_init
4323 understands) But they're true if for this symbol there is
4324 respectively a typeglob and a subroutine.
4326 cv = gv ? ((SvTYPE(gv) == SVt_PVGV)
4327 /* Real typeglob, so get the real subroutine: */
4329 /* A proxy for a subroutine in this package? */
4330 : SvOK(gv) ? (CV *) gv : NULL)
4333 /* See if it's the indirect object for a list operator. */
4335 if (PL_oldoldbufptr &&
4336 PL_oldoldbufptr < PL_bufptr &&
4337 (PL_oldoldbufptr == PL_last_lop
4338 || PL_oldoldbufptr == PL_last_uni) &&
4339 /* NO SKIPSPACE BEFORE HERE! */
4340 (PL_expect == XREF ||
4341 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
4343 bool immediate_paren = *s == '(';
4345 /* (Now we can afford to cross potential line boundary.) */
4348 /* Two barewords in a row may indicate method call. */
4350 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
4353 /* If not a declared subroutine, it's an indirect object. */
4354 /* (But it's an indir obj regardless for sort.) */
4355 /* Also, if "_" follows a filetest operator, it's a bareword */
4358 ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
4360 (PL_last_lop_op != OP_MAPSTART &&
4361 PL_last_lop_op != OP_GREPSTART))))
4362 || (PL_tokenbuf[0] == '_' && PL_tokenbuf[1] == '\0'
4363 && ((PL_opargs[PL_last_lop_op] & OA_CLASS_MASK) == OA_FILESTATOP))
4366 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
4371 PL_expect = XOPERATOR;
4374 /* Is this a word before a => operator? */
4375 if (*s == '=' && s[1] == '>' && !pkgname) {
4377 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
4378 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
4379 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
4383 /* If followed by a paren, it's certainly a subroutine. */
4387 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
4388 if (*d == ')' && (sv = cv_const_sv(cv))) {
4393 PL_nextval[PL_nexttoke].opval = yylval.opval;
4394 PL_expect = XOPERATOR;
4400 /* If followed by var or block, call it a method (unless sub) */
4402 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
4403 PL_last_lop = PL_oldbufptr;
4404 PL_last_lop_op = OP_METHOD;
4408 /* If followed by a bareword, see if it looks like indir obj. */
4411 && (isIDFIRST_lazy_if(s,UTF) || *s == '$')
4412 && (tmp = intuit_method(s,gv)))
4415 /* Not a method, so call it a subroutine (if defined) */
4418 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
4419 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4420 "Ambiguous use of -%s resolved as -&%s()",
4421 PL_tokenbuf, PL_tokenbuf);
4422 /* Check for a constant sub */
4423 if ((sv = cv_const_sv(cv))) {
4425 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
4426 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
4427 yylval.opval->op_private = 0;
4431 /* Resolve to GV now. */
4432 op_free(yylval.opval);
4433 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
4434 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
4435 PL_last_lop = PL_oldbufptr;
4436 PL_last_lop_op = OP_ENTERSUB;
4437 /* Is there a prototype? */
4440 const char *proto = SvPV_const((SV*)cv, len);
4443 if (*proto == '$' && proto[1] == '\0')
4445 while (*proto == ';')
4447 if (*proto == '&' && *s == '{') {
4448 sv_setpv(PL_subname, PL_curstash ?
4449 "__ANON__" : "__ANON__::__ANON__");
4453 PL_nextval[PL_nexttoke].opval = yylval.opval;
4459 /* Call it a bare word */
4461 if (PL_hints & HINT_STRICT_SUBS)
4462 yylval.opval->op_private |= OPpCONST_STRICT;
4465 if (lastchar != '-') {
4466 if (ckWARN(WARN_RESERVED)) {
4467 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
4468 if (!*d && !gv_stashpv(PL_tokenbuf,FALSE))
4469 Perl_warner(aTHX_ packWARN(WARN_RESERVED), PL_warn_reserved,
4476 if ((lastchar == '*' || lastchar == '%' || lastchar == '&')
4477 && ckWARN_d(WARN_AMBIGUOUS)) {
4478 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4479 "Operator or semicolon missing before %c%s",
4480 lastchar, PL_tokenbuf);
4481 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4482 "Ambiguous use of %c resolved as operator %c",
4483 lastchar, lastchar);
4489 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4490 newSVpv(CopFILE(PL_curcop),0));
4494 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4495 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
4498 case KEY___PACKAGE__:
4499 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4501 ? newSVhek(HvNAME_HEK(PL_curstash))
4508 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
4509 const char *pname = "main";
4510 if (PL_tokenbuf[2] == 'D')
4511 pname = HvNAME_get(PL_curstash ? PL_curstash : PL_defstash);
4512 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), GV_ADD,
4516 GvIOp(gv) = newIO();
4517 IoIFP(GvIOp(gv)) = PL_rsfp;
4518 #if defined(HAS_FCNTL) && defined(F_SETFD)
4520 const int fd = PerlIO_fileno(PL_rsfp);
4521 fcntl(fd,F_SETFD,fd >= 3);
4524 /* Mark this internal pseudo-handle as clean */
4525 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
4527 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
4528 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
4529 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
4531 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
4532 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
4533 /* if the script was opened in binmode, we need to revert
4534 * it to text mode for compatibility; but only iff it has CRs
4535 * XXX this is a questionable hack at best. */
4536 if (PL_bufend-PL_bufptr > 2
4537 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
4540 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
4541 loc = PerlIO_tell(PL_rsfp);
4542 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4545 if (PerlLIO_setmode(PL_rsfp, O_TEXT) != -1) {
4547 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4548 #endif /* NETWARE */
4549 #ifdef PERLIO_IS_STDIO /* really? */
4550 # if defined(__BORLANDC__)
4551 /* XXX see note in do_binmode() */
4552 ((FILE*)PL_rsfp)->flags &= ~_F_BIN;
4556 PerlIO_seek(PL_rsfp, loc, 0);
4560 #ifdef PERLIO_LAYERS
4563 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
4564 else if (PL_encoding) {
4571 XPUSHs(PL_encoding);
4573 call_method("name", G_SCALAR);
4577 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL,
4578 Perl_form(aTHX_ ":encoding(%"SVf")",
4596 if (PL_expect == XSTATE) {
4603 if (*s == ':' && s[1] == ':') {
4606 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4607 if (!(tmp = keyword(PL_tokenbuf, len)))
4608 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4611 else if (tmp == KEY_require || tmp == KEY_do)
4612 /* that's a way to remember we saw "CORE::" */
4625 LOP(OP_ACCEPT,XTERM);
4631 LOP(OP_ATAN2,XTERM);
4637 LOP(OP_BINMODE,XTERM);
4640 LOP(OP_BLESS,XTERM);
4649 /* When 'use switch' is in effect, continue has a dual
4650 life as a control operator. */
4652 if (!FEATURE_IS_ENABLED("switch", 6))
4655 /* We have to disambiguate the two senses of
4656 "continue". If the next token is a '{' then
4657 treat it as the start of a continue block;
4658 otherwise treat it as a control operator.
4669 (void)gv_fetchpv("ENV", GV_ADD, SVt_PVHV); /* may use HOME */
4686 if (!PL_cryptseen) {
4687 PL_cryptseen = TRUE;
4691 LOP(OP_CRYPT,XTERM);
4694 LOP(OP_CHMOD,XTERM);
4697 LOP(OP_CHOWN,XTERM);
4700 LOP(OP_CONNECT,XTERM);
4719 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4720 if (orig_keyword == KEY_do) {
4729 PL_hints |= HINT_BLOCK_SCOPE;
4739 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4740 LOP(OP_DBMOPEN,XTERM);
4746 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4753 yylval.ival = CopLINE(PL_curcop);
4767 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4768 UNIBRACK(OP_ENTEREVAL);
4786 case KEY_endhostent:
4792 case KEY_endservent:
4795 case KEY_endprotoent:
4806 yylval.ival = CopLINE(PL_curcop);
4808 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4810 if ((PL_bufend - p) >= 3 &&
4811 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4813 else if ((PL_bufend - p) >= 4 &&
4814 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4817 if (isIDFIRST_lazy_if(p,UTF)) {
4818 p = scan_ident(p, PL_bufend,
4819 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4823 Perl_croak(aTHX_ "Missing $ on loop variable");
4828 LOP(OP_FORMLINE,XTERM);
4834 LOP(OP_FCNTL,XTERM);
4840 LOP(OP_FLOCK,XTERM);
4849 LOP(OP_GREPSTART, XREF);
4852 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4867 case KEY_getpriority:
4868 LOP(OP_GETPRIORITY,XTERM);
4870 case KEY_getprotobyname:
4873 case KEY_getprotobynumber:
4874 LOP(OP_GPBYNUMBER,XTERM);
4876 case KEY_getprotoent:
4888 case KEY_getpeername:
4889 UNI(OP_GETPEERNAME);
4891 case KEY_gethostbyname:
4894 case KEY_gethostbyaddr:
4895 LOP(OP_GHBYADDR,XTERM);
4897 case KEY_gethostent:
4900 case KEY_getnetbyname:
4903 case KEY_getnetbyaddr:
4904 LOP(OP_GNBYADDR,XTERM);
4909 case KEY_getservbyname:
4910 LOP(OP_GSBYNAME,XTERM);
4912 case KEY_getservbyport:
4913 LOP(OP_GSBYPORT,XTERM);
4915 case KEY_getservent:
4918 case KEY_getsockname:
4919 UNI(OP_GETSOCKNAME);
4921 case KEY_getsockopt:
4922 LOP(OP_GSOCKOPT,XTERM);
4937 yylval.ival = CopLINE(PL_curcop);
4948 yylval.ival = CopLINE(PL_curcop);
4952 LOP(OP_INDEX,XTERM);
4958 LOP(OP_IOCTL,XTERM);
4970 s = force_word(s,WORD,TRUE,FALSE,FALSE);
5002 LOP(OP_LISTEN,XTERM);
5011 s = scan_pat(s,OP_MATCH);
5012 TERM(sublex_start());
5015 LOP(OP_MAPSTART, XREF);
5018 LOP(OP_MKDIR,XTERM);
5021 LOP(OP_MSGCTL,XTERM);
5024 LOP(OP_MSGGET,XTERM);
5027 LOP(OP_MSGRCV,XTERM);
5030 LOP(OP_MSGSND,XTERM);
5036 if (isIDFIRST_lazy_if(s,UTF)) {
5037 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
5038 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
5040 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
5041 if (!PL_in_my_stash) {
5044 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
5052 s = force_word(s,WORD,TRUE,FALSE,FALSE);
5059 s = tokenize_use(0, s);
5063 if (*s == '(' || (s = skipspace(s), *s == '('))
5070 if (isIDFIRST_lazy_if(s,UTF)) {
5072 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
5073 for (t=d; *t && isSPACE(*t); t++) ;
5074 if ( *t && strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE)
5076 && !(t[0] == '=' && t[1] == '>')
5078 int len = (int)(d-s);
5079 Perl_warner(aTHX_ packWARN(WARN_PRECEDENCE),
5080 "Precedence problem: open %.*s should be open(%.*s)",
5087 yylval.ival = OP_OR;
5097 LOP(OP_OPEN_DIR,XTERM);
5100 checkcomma(s,PL_tokenbuf,"filehandle");
5104 checkcomma(s,PL_tokenbuf,"filehandle");
5123 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5127 LOP(OP_PIPE_OP,XTERM);
5130 s = scan_str(s,FALSE,FALSE);
5132 missingterm((char*)0);
5133 yylval.ival = OP_CONST;
5134 TERM(sublex_start());
5140 s = scan_str(s,FALSE,FALSE);
5142 missingterm((char*)0);
5143 PL_expect = XOPERATOR;
5145 if (SvCUR(PL_lex_stuff)) {
5148 d = SvPV_force(PL_lex_stuff, len);
5151 for (; isSPACE(*d) && len; --len, ++d) ;
5154 if (!warned && ckWARN(WARN_QW)) {
5155 for (; !isSPACE(*d) && len; --len, ++d) {
5157 Perl_warner(aTHX_ packWARN(WARN_QW),
5158 "Possible attempt to separate words with commas");
5161 else if (*d == '#') {
5162 Perl_warner(aTHX_ packWARN(WARN_QW),
5163 "Possible attempt to put comments in qw() list");
5169 for (; !isSPACE(*d) && len; --len, ++d) ;
5171 sv = newSVpvn(b, d-b);
5172 if (DO_UTF8(PL_lex_stuff))
5174 words = append_elem(OP_LIST, words,
5175 newSVOP(OP_CONST, 0, tokeq(sv)));
5179 PL_nextval[PL_nexttoke].opval = words;
5184 SvREFCNT_dec(PL_lex_stuff);
5185 PL_lex_stuff = Nullsv;
5191 s = scan_str(s,FALSE,FALSE);
5193 missingterm((char*)0);
5194 yylval.ival = OP_STRINGIFY;
5195 if (SvIVX(PL_lex_stuff) == '\'')
5196 SvIV_set(PL_lex_stuff, 0); /* qq'$foo' should intepolate */
5197 TERM(sublex_start());
5200 s = scan_pat(s,OP_QR);
5201 TERM(sublex_start());
5204 s = scan_str(s,FALSE,FALSE);
5206 missingterm((char*)0);
5207 yylval.ival = OP_BACKTICK;
5209 TERM(sublex_start());
5217 s = force_version(s, FALSE);
5219 else if (*s != 'v' || !isDIGIT(s[1])
5220 || (s = force_version(s, TRUE), *s == 'v'))
5222 *PL_tokenbuf = '\0';
5223 s = force_word(s,WORD,TRUE,TRUE,FALSE);
5224 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
5225 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
5227 yyerror("<> should be quotes");
5229 if (orig_keyword == KEY_require) {
5237 PL_last_uni = PL_oldbufptr;
5238 PL_last_lop_op = OP_REQUIRE;
5240 return REPORT( (int)REQUIRE );
5246 s = force_word(s,WORD,TRUE,FALSE,FALSE);
5250 LOP(OP_RENAME,XTERM);
5259 LOP(OP_RINDEX,XTERM);
5269 UNIDOR(OP_READLINE);
5282 LOP(OP_REVERSE,XTERM);
5285 UNIDOR(OP_READLINK);
5293 TERM(sublex_start());
5295 TOKEN(1); /* force error */
5298 checkcomma(s,PL_tokenbuf,"filehandle");
5308 LOP(OP_SELECT,XTERM);
5314 LOP(OP_SEMCTL,XTERM);
5317 LOP(OP_SEMGET,XTERM);
5320 LOP(OP_SEMOP,XTERM);
5326 LOP(OP_SETPGRP,XTERM);
5328 case KEY_setpriority:
5329 LOP(OP_SETPRIORITY,XTERM);
5331 case KEY_sethostent:
5337 case KEY_setservent:
5340 case KEY_setprotoent:
5350 LOP(OP_SEEKDIR,XTERM);
5352 case KEY_setsockopt:
5353 LOP(OP_SSOCKOPT,XTERM);
5359 LOP(OP_SHMCTL,XTERM);
5362 LOP(OP_SHMGET,XTERM);
5365 LOP(OP_SHMREAD,XTERM);
5368 LOP(OP_SHMWRITE,XTERM);
5371 LOP(OP_SHUTDOWN,XTERM);
5380 LOP(OP_SOCKET,XTERM);
5382 case KEY_socketpair:
5383 LOP(OP_SOCKPAIR,XTERM);
5386 checkcomma(s,PL_tokenbuf,"subroutine name");
5388 if (*s == ';' || *s == ')') /* probably a close */
5389 Perl_croak(aTHX_ "sort is now a reserved word");
5391 s = force_word(s,WORD,TRUE,TRUE,FALSE);
5395 LOP(OP_SPLIT,XTERM);
5398 LOP(OP_SPRINTF,XTERM);
5401 LOP(OP_SPLICE,XTERM);
5416 LOP(OP_SUBSTR,XTERM);
5422 char tmpbuf[sizeof PL_tokenbuf];
5423 SSize_t tboffset = 0;
5424 expectation attrful;
5425 bool have_name, have_proto, bad_proto;
5426 const int key = tmp;
5430 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
5431 (*s == ':' && s[1] == ':'))
5434 attrful = XATTRBLOCK;
5435 /* remember buffer pos'n for later force_word */
5436 tboffset = s - PL_oldbufptr;
5437 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
5438 if (strchr(tmpbuf, ':'))
5439 sv_setpv(PL_subname, tmpbuf);
5441 sv_setsv(PL_subname,PL_curstname);
5442 sv_catpvn(PL_subname,"::",2);
5443 sv_catpvn(PL_subname,tmpbuf,len);
5450 Perl_croak(aTHX_ "Missing name in \"my sub\"");
5451 PL_expect = XTERMBLOCK;
5452 attrful = XATTRTERM;
5453 sv_setpvn(PL_subname,"?",1);
5457 if (key == KEY_format) {
5459 PL_lex_formbrack = PL_lex_brackets + 1;
5461 (void) force_word(PL_oldbufptr + tboffset, WORD,
5466 /* Look for a prototype */
5470 s = scan_str(s,FALSE,FALSE);
5472 Perl_croak(aTHX_ "Prototype not terminated");
5473 /* strip spaces and check for bad characters */
5474 d = SvPVX(PL_lex_stuff);
5477 for (p = d; *p; ++p) {
5480 if (!strchr("$@%*;[]&\\", *p))
5485 if (bad_proto && ckWARN(WARN_SYNTAX))
5486 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5487 "Illegal character in prototype for %"SVf" : %s",
5489 SvCUR_set(PL_lex_stuff, tmp);
5497 if (*s == ':' && s[1] != ':')
5498 PL_expect = attrful;
5499 else if (*s != '{' && key == KEY_sub) {
5501 Perl_croak(aTHX_ "Illegal declaration of anonymous subroutine");
5503 Perl_croak(aTHX_ "Illegal declaration of subroutine %"SVf, PL_subname);
5507 PL_nextval[PL_nexttoke].opval =
5508 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
5509 PL_lex_stuff = Nullsv;
5513 sv_setpv(PL_subname,
5514 PL_curstash ? "__ANON__" : "__ANON__::__ANON__");
5517 (void) force_word(PL_oldbufptr + tboffset, WORD,
5526 LOP(OP_SYSTEM,XREF);
5529 LOP(OP_SYMLINK,XTERM);
5532 LOP(OP_SYSCALL,XTERM);
5535 LOP(OP_SYSOPEN,XTERM);
5538 LOP(OP_SYSSEEK,XTERM);
5541 LOP(OP_SYSREAD,XTERM);
5544 LOP(OP_SYSWRITE,XTERM);
5548 TERM(sublex_start());
5569 LOP(OP_TRUNCATE,XTERM);
5581 yylval.ival = CopLINE(PL_curcop);
5585 yylval.ival = CopLINE(PL_curcop);
5589 LOP(OP_UNLINK,XTERM);
5595 LOP(OP_UNPACK,XTERM);
5598 LOP(OP_UTIME,XTERM);
5604 LOP(OP_UNSHIFT,XTERM);
5607 s = tokenize_use(1, s);
5617 yylval.ival = CopLINE(PL_curcop);
5621 yylval.ival = CopLINE(PL_curcop);
5625 PL_hints |= HINT_BLOCK_SCOPE;
5632 LOP(OP_WAITPID,XTERM);
5641 ctl_l[0] = toCTRL('L');
5643 gv_fetchpv(ctl_l, GV_ADD, SVt_PV);
5646 gv_fetchpv("\f", GV_ADD, SVt_PV); /* Make sure $^L is defined */
5651 if (PL_expect == XOPERATOR)
5657 yylval.ival = OP_XOR;
5662 TERM(sublex_start());
5667 #pragma segment Main
5671 S_pending_ident(pTHX)
5674 register I32 tmp = 0;
5675 /* pit holds the identifier we read and pending_ident is reset */
5676 char pit = PL_pending_ident;
5677 PL_pending_ident = 0;
5679 DEBUG_T({ PerlIO_printf(Perl_debug_log,
5680 "### Pending identifier '%s'\n", PL_tokenbuf); });
5682 /* if we're in a my(), we can't allow dynamics here.
5683 $foo'bar has already been turned into $foo::bar, so
5684 just check for colons.
5686 if it's a legal name, the OP is a PADANY.
5689 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
5690 if (strchr(PL_tokenbuf,':'))
5691 yyerror(Perl_form(aTHX_ "No package name allowed for "
5692 "variable %s in \"our\"",
5694 tmp = allocmy(PL_tokenbuf);
5697 if (strchr(PL_tokenbuf,':'))
5698 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
5700 yylval.opval = newOP(OP_PADANY, 0);
5701 yylval.opval->op_targ = allocmy(PL_tokenbuf);
5707 build the ops for accesses to a my() variable.
5709 Deny my($a) or my($b) in a sort block, *if* $a or $b is
5710 then used in a comparison. This catches most, but not
5711 all cases. For instance, it catches
5712 sort { my($a); $a <=> $b }
5714 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
5715 (although why you'd do that is anyone's guess).
5718 if (!strchr(PL_tokenbuf,':')) {
5720 tmp = pad_findmy(PL_tokenbuf);
5721 if (tmp != NOT_IN_PAD) {
5722 /* might be an "our" variable" */
5723 if (PAD_COMPNAME_FLAGS(tmp) & SVpad_OUR) {
5724 /* build ops for a bareword */
5725 HV * const stash = PAD_COMPNAME_OURSTASH(tmp);
5726 HEK * const stashname = HvNAME_HEK(stash);
5727 SV * const sym = newSVhek(stashname);
5728 sv_catpvn(sym, "::", 2);
5729 sv_catpv(sym, PL_tokenbuf+1);
5730 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
5731 yylval.opval->op_private = OPpCONST_ENTERED;
5734 ? (GV_ADDMULTI | GV_ADDINEVAL)
5737 ((PL_tokenbuf[0] == '$') ? SVt_PV
5738 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5743 /* if it's a sort block and they're naming $a or $b */
5744 if (PL_last_lop_op == OP_SORT &&
5745 PL_tokenbuf[0] == '$' &&
5746 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
5749 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
5750 d < PL_bufend && *d != '\n';
5753 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
5754 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
5760 yylval.opval = newOP(OP_PADANY, 0);
5761 yylval.opval->op_targ = tmp;
5767 Whine if they've said @foo in a doublequoted string,
5768 and @foo isn't a variable we can find in the symbol
5771 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
5772 GV *gv = gv_fetchpv(PL_tokenbuf+1, 0, SVt_PVAV);
5773 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
5774 && ckWARN(WARN_AMBIGUOUS))
5776 /* Downgraded from fatal to warning 20000522 mjd */
5777 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5778 "Possible unintended interpolation of %s in string",
5783 /* build ops for a bareword */
5784 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
5785 yylval.opval->op_private = OPpCONST_ENTERED;
5789 ? (GV_ADDMULTI | GV_ADDINEVAL)
5790 /* If the identifier refers to a stash, don't autovivify it.
5791 * Change 24660 had the side effect of causing symbol table
5792 * hashes to always be defined, even if they were freshly
5793 * created and the only reference in the entire program was
5794 * the single statement with the defined %foo::bar:: test.
5795 * It appears that all code in the wild doing this actually
5796 * wants to know whether sub-packages have been loaded, so
5797 * by avoiding auto-vivifying symbol tables, we ensure that
5798 * defined %foo::bar:: continues to be false, and the existing
5799 * tests still give the expected answers, even though what
5800 * they're actually testing has now changed subtly.
5802 : !(*PL_tokenbuf == '%' && *(d = PL_tokenbuf + strlen(PL_tokenbuf) - 1) == ':' && d[-1] == ':'),
5803 ((PL_tokenbuf[0] == '$') ? SVt_PV
5804 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5810 * The following code was generated by perl_keyword.pl.
5814 Perl_keyword (pTHX_ const char *name, I32 len)
5818 case 1: /* 5 tokens of length 1 */
5850 case 2: /* 18 tokens of length 2 */
5996 case 3: /* 29 tokens of length 3 */
6000 if (name[1] == 'N' &&
6063 if (name[1] == 'i' &&
6103 if (name[1] == 'o' &&
6112 if (name[1] == 'e' &&
6121 if (name[1] == 'n' &&
6130 if (name[1] == 'o' &&
6139 if (name[1] == 'a' &&
6148 if (name[1] == 'o' &&
6210 if (name[1] == 'e' &&
6224 return (FEATURE_IS_ENABLED("say", 3) ? -KEY_say : 0);
6250 if (name[1] == 'i' &&
6259 if (name[1] == 's' &&
6268 if (name[1] == 'e' &&
6277 if (name[1] == 'o' &&
6289 case 4: /* 41 tokens of length 4 */
6293 if (name[1] == 'O' &&
6303 if (name[1] == 'N' &&
6313 if (name[1] == 'i' &&
6323 if (name[1] == 'h' &&
6333 if (name[1] == 'u' &&
6346 if (name[2] == 'c' &&
6355 if (name[2] == 's' &&
6364 if (name[2] == 'a' &&
6400 if (name[1] == 'o' &&
6413 if (name[2] == 't' &&
6422 if (name[2] == 'o' &&
6431 if (name[2] == 't' &&
6440 if (name[2] == 'e' &&
6453 if (name[1] == 'o' &&
6466 if (name[2] == 'y' &&
6475 if (name[2] == 'l' &&
6491 if (name[2] == 's' &&
6500 if (name[2] == 'n' &&
6509 if (name[2] == 'c' &&
6522 if (name[1] == 'e' &&
6532 if (name[1] == 'p' &&
6545 if (name[2] == 'c' &&
6554 if (name[2] == 'p' &&
6563 if (name[2] == 's' &&
6579 if (name[2] == 'n' &&
6649 if (name[2] == 'r' &&
6658 if (name[2] == 'r' &&
6667 if (name[2] == 'a' &&
6683 if (name[2] == 'l' &&
6745 if (name[2] == 'e' &&
6748 return (FEATURE_IS_ENABLED("switch", 6) ? KEY_when : 0);
6761 case 5: /* 38 tokens of length 5 */
6765 if (name[1] == 'E' &&
6776 if (name[1] == 'H' &&
6790 if (name[2] == 'a' &&
6800 if (name[2] == 'a' &&
6817 if (name[2] == 'e' &&
6827 if (name[2] == 'e' &&
6831 return (FEATURE_IS_ENABLED("switch", 6) ? -KEY_break : 0);
6847 if (name[3] == 'i' &&
6856 if (name[3] == 'o' &&
6892 if (name[2] == 'o' &&
6902 if (name[2] == 'y' &&
6916 if (name[1] == 'l' &&
6930 if (name[2] == 'n' &&
6940 if (name[2] == 'o' &&
6954 if (name[1] == 'i' &&
6959 return (FEATURE_IS_ENABLED("switch", 6) ? KEY_given : 0);
6968 if (name[2] == 'd' &&
6978 if (name[2] == 'c' &&
6995 if (name[2] == 'c' &&
7005 if (name[2] == 't' &&
7019 if (name[1] == 'k' &&
7030 if (name[1] == 'r' &&
7044 if (name[2] == 's' &&
7054 if (name[2] == 'd' &&
7071 if (name[2] == 'm' &&
7081 if (name[2] == 'i' &&
7091 if (name[2] == 'e' &&
7101 if (name[2] == 'l' &&
7111 if (name[2] == 'a' &&
7121 if (name[2] == 'u' &&
7135 if (name[1] == 'i' &&
7149 if (name[2] == 'a' &&
7162 if (name[3] == 'e' &&
7197 if (name[2] == 'i' &&
7214 if (name[2] == 'i' &&
7224 if (name[2] == 'i' &&
7241 case 6: /* 33 tokens of length 6 */
7245 if (name[1] == 'c' &&
7260 if (name[2] == 'l' &&
7271 if (name[2] == 'r' &&
7286 if (name[1] == 'e' &&
7301 if (name[2] == 's' &&
7306 if(ckWARN_d(WARN_SYNTAX))
7307 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "elseif should be elsif");
7313 if (name[2] == 'i' &&
7331 if (name[2] == 'l' &&
7342 if (name[2] == 'r' &&
7357 if (name[1] == 'm' &&
7372 if (name[2] == 'n' &&
7383 if (name[2] == 's' &&
7398 if (name[1] == 's' &&
7404 if (name[4] == 't' &&
7413 if (name[4] == 'e' &&
7422 if (name[4] == 'c' &&
7431 if (name[4] == 'n' &&
7447 if (name[1] == 'r' &&
7465 if (name[3] == 'a' &&
7475 if (name[3] == 'u' &&
7489 if (name[2] == 'n' &&
7507 if (name[2] == 'a' &&
7521 if (name[3] == 'e' &&
7534 if (name[4] == 't' &&
7543 if (name[4] == 'e' &&
7565 if (name[4] == 't' &&
7574 if (name[4] == 'e' &&
7590 if (name[2] == 'c' &&
7601 if (name[2] == 'l' &&
7612 if (name[2] == 'b' &&
7623 if (name[2] == 's' &&
7646 if (name[4] == 's' &&
7655 if (name[4] == 'n' &&
7668 if (name[3] == 'a' &&
7685 if (name[1] == 'a' &&
7700 case 7: /* 29 tokens of length 7 */
7704 if (name[1] == 'E' &&
7717 if (name[1] == '_' &&
7730 if (name[1] == 'i' &&
7737 return -KEY_binmode;
7743 if (name[1] == 'o' &&
7750 return -KEY_connect;
7759 if (name[2] == 'm' &&
7765 return -KEY_dbmopen;
7776 if (name[4] == 'u' &&
7780 return (FEATURE_IS_ENABLED("switch", 6) ? KEY_default : 0);
7786 if (name[4] == 'n' &&
7807 if (name[1] == 'o' &&
7820 if (name[1] == 'e' &&
7827 if (name[5] == 'r' &&
7830 return -KEY_getpgrp;
7836 if (name[5] == 'i' &&
7839 return -KEY_getppid;
7852 if (name[1] == 'c' &&
7859 return -KEY_lcfirst;
7865 if (name[1] == 'p' &&
7872 return -KEY_opendir;
7878 if (name[1] == 'a' &&
7896 if (name[3] == 'd' &&
7901 return -KEY_readdir;
7907 if (name[3] == 'u' &&
7918 if (name[3] == 'e' &&
7923 return -KEY_reverse;
7942 if (name[3] == 'k' &&
7947 return -KEY_seekdir;
7953 if (name[3] == 'p' &&
7958 return -KEY_setpgrp;
7968 if (name[2] == 'm' &&
7974 return -KEY_shmread;
7980 if (name[2] == 'r' &&
7986 return -KEY_sprintf;
7995 if (name[3] == 'l' &&
8000 return -KEY_symlink;
8009 if (name[4] == 'a' &&
8013 return -KEY_syscall;
8019 if (name[4] == 'p' &&
8023 return -KEY_sysopen;
8029 if (name[4] == 'e' &&
8033 return -KEY_sysread;
8039 if (name[4] == 'e' &&
8043 return -KEY_sysseek;
8061 if (name[1] == 'e' &&
8068 return -KEY_telldir;
8077 if (name[2] == 'f' &&
8083 return -KEY_ucfirst;
8089 if (name[2] == 's' &&
8095 return -KEY_unshift;
8105 if (name[1] == 'a' &&
8112 return -KEY_waitpid;
8121 case 8: /* 26 tokens of length 8 */
8125 if (name[1] == 'U' &&
8133 return KEY_AUTOLOAD;
8144 if (name[3] == 'A' &&
8150 return KEY___DATA__;
8156 if (name[3] == 'I' &&
8162 return -KEY___FILE__;
8168 if (name[3] == 'I' &&
8174 return -KEY___LINE__;
8190 if (name[2] == 'o' &&
8197 return -KEY_closedir;
8203 if (name[2] == 'n' &&
8210 return -KEY_continue;
8220 if (name[1] == 'b' &&
8228 return -KEY_dbmclose;
8234 if (name[1] == 'n' &&
8240 if (name[4] == 'r' &&
8245 return -KEY_endgrent;
8251 if (name[4] == 'w' &&
8256 return -KEY_endpwent;
8269 if (name[1] == 'o' &&
8277 return -KEY_formline;
8283 if (name[1] == 'e' &&
8294 if (name[6] == 'n' &&
8297 return -KEY_getgrent;
8303 if (name[6] == 'i' &&
8306 return -KEY_getgrgid;
8312 if (name[6] == 'a' &&
8315 return -KEY_getgrnam;
8328 if (name[4] == 'o' &&
8333 return -KEY_getlogin;
8344 if (name[6] == 'n' &&
8347 return -KEY_getpwent;
8353 if (name[6] == 'a' &&
8356 return -KEY_getpwnam;
8362 if (name[6] == 'i' &&
8365 return -KEY_getpwuid;
8385 if (name[1] == 'e' &&
8392 if (name[5] == 'i' &&
8399 return -KEY_readline;
8404 return -KEY_readlink;
8415 if (name[5] == 'i' &&
8419 return -KEY_readpipe;
8440 if (name[4] == 'r' &&
8445 return -KEY_setgrent;
8451 if (name[4] == 'w' &&
8456 return -KEY_setpwent;
8472 if (name[3] == 'w' &&
8478 return -KEY_shmwrite;
8484 if (name[3] == 't' &&
8490 return -KEY_shutdown;
8500 if (name[2] == 's' &&
8507 return -KEY_syswrite;
8517 if (name[1] == 'r' &&
8525 return -KEY_truncate;
8534 case 9: /* 8 tokens of length 9 */
8538 if (name[1] == 'n' &&
8547 return -KEY_endnetent;
8553 if (name[1] == 'e' &&
8562 return -KEY_getnetent;
8568 if (name[1] == 'o' &&
8577 return -KEY_localtime;
8583 if (name[1] == 'r' &&
8592 return KEY_prototype;
8598 if (name[1] == 'u' &&
8607 return -KEY_quotemeta;
8613 if (name[1] == 'e' &&
8622 return -KEY_rewinddir;
8628 if (name[1] == 'e' &&
8637 return -KEY_setnetent;
8643 if (name[1] == 'a' &&
8652 return -KEY_wantarray;
8661 case 10: /* 9 tokens of length 10 */
8665 if (name[1] == 'n' &&
8671 if (name[4] == 'o' &&
8678 return -KEY_endhostent;
8684 if (name[4] == 'e' &&
8691 return -KEY_endservent;
8704 if (name[1] == 'e' &&
8710 if (name[4] == 'o' &&
8717 return -KEY_gethostent;
8726 if (name[5] == 'r' &&
8732 return -KEY_getservent;
8738 if (name[5] == 'c' &&
8744 return -KEY_getsockopt;
8769 if (name[4] == 'o' &&
8776 return -KEY_sethostent;
8785 if (name[5] == 'r' &&
8791 return -KEY_setservent;
8797 if (name[5] == 'c' &&
8803 return -KEY_setsockopt;
8820 if (name[2] == 'c' &&
8829 return -KEY_socketpair;
8842 case 11: /* 8 tokens of length 11 */
8846 if (name[1] == '_' &&
8857 return -KEY___PACKAGE__;
8863 if (name[1] == 'n' &&
8874 return -KEY_endprotoent;
8880 if (name[1] == 'e' &&
8889 if (name[5] == 'e' &&
8896 return -KEY_getpeername;
8905 if (name[6] == 'o' &&
8911 return -KEY_getpriority;
8917 if (name[6] == 't' &&
8923 return -KEY_getprotoent;
8937 if (name[4] == 'o' &&
8945 return -KEY_getsockname;
8958 if (name[1] == 'e' &&
8966 if (name[6] == 'o' &&
8972 return -KEY_setpriority;
8978 if (name[6] == 't' &&
8984 return -KEY_setprotoent;
9000 case 12: /* 2 tokens of length 12 */
9001 if (name[0] == 'g' &&
9013 if (name[9] == 'd' &&
9016 { /* getnetbyaddr */
9017 return -KEY_getnetbyaddr;
9023 if (name[9] == 'a' &&
9026 { /* getnetbyname */
9027 return -KEY_getnetbyname;
9039 case 13: /* 4 tokens of length 13 */
9040 if (name[0] == 'g' &&
9047 if (name[4] == 'o' &&
9056 if (name[10] == 'd' &&
9059 { /* gethostbyaddr */
9060 return -KEY_gethostbyaddr;
9066 if (name[10] == 'a' &&
9069 { /* gethostbyname */
9070 return -KEY_gethostbyname;
9083 if (name[4] == 'e' &&
9092 if (name[10] == 'a' &&
9095 { /* getservbyname */
9096 return -KEY_getservbyname;
9102 if (name[10] == 'o' &&
9105 { /* getservbyport */
9106 return -KEY_getservbyport;
9125 case 14: /* 1 tokens of length 14 */
9126 if (name[0] == 'g' &&
9140 { /* getprotobyname */
9141 return -KEY_getprotobyname;
9146 case 16: /* 1 tokens of length 16 */
9147 if (name[0] == 'g' &&
9163 { /* getprotobynumber */
9164 return -KEY_getprotobynumber;
9178 S_checkcomma(pTHX_ register char *s, const char *name, const char *what)
9182 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
9183 if (ckWARN(WARN_SYNTAX)) {
9185 for (w = s+2; *w && level; w++) {
9192 for (; *w && isSPACE(*w); w++) ;
9193 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
9194 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
9195 "%s (...) interpreted as function",name);
9198 while (s < PL_bufend && isSPACE(*s))
9202 while (s < PL_bufend && isSPACE(*s))
9204 if (isIDFIRST_lazy_if(s,UTF)) {
9206 while (isALNUM_lazy_if(s,UTF))
9208 while (s < PL_bufend && isSPACE(*s))
9212 *s = '\0'; /* XXX If we didn't do this, we could const a lot of toke.c */
9213 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
9217 Perl_croak(aTHX_ "No comma allowed after %s", what);
9222 /* Either returns sv, or mortalizes sv and returns a new SV*.
9223 Best used as sv=new_constant(..., sv, ...).
9224 If s, pv are NULL, calls subroutine with one argument,
9225 and type is used with error messages only. */
9228 S_new_constant(pTHX_ const char *s, STRLEN len, const char *key, SV *sv, SV *pv,
9232 HV * const table = GvHV(PL_hintgv); /* ^H */
9236 const char *why1 = "", *why2 = "", *why3 = "";
9238 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
9241 why2 = strEQ(key,"charnames")
9242 ? "(possibly a missing \"use charnames ...\")"
9244 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
9245 (type ? type: "undef"), why2);
9247 /* This is convoluted and evil ("goto considered harmful")
9248 * but I do not understand the intricacies of all the different
9249 * failure modes of %^H in here. The goal here is to make
9250 * the most probable error message user-friendly. --jhi */
9255 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
9256 (type ? type: "undef"), why1, why2, why3);
9258 yyerror(SvPVX_const(msg));
9262 cvp = hv_fetch(table, key, strlen(key), FALSE);
9263 if (!cvp || !SvOK(*cvp)) {
9266 why3 = "} is not defined";
9269 sv_2mortal(sv); /* Parent created it permanently */
9272 pv = sv_2mortal(newSVpvn(s, len));
9274 typesv = sv_2mortal(newSVpv(type, 0));
9276 typesv = &PL_sv_undef;
9278 PUSHSTACKi(PERLSI_OVERLOAD);
9290 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
9294 /* Check the eval first */
9295 if (!PL_in_eval && SvTRUE(ERRSV)) {
9296 sv_catpv(ERRSV, "Propagated");
9297 yyerror(SvPV_nolen_const(ERRSV)); /* Duplicates the message inside eval */
9299 res = SvREFCNT_inc(sv);
9303 (void)SvREFCNT_inc(res);
9312 why1 = "Call to &{$^H{";
9314 why3 = "}} did not return a defined value";
9322 /* Returns a NUL terminated string, with the length of the string written to
9326 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
9328 register char *d = dest;
9329 register char * const e = d + destlen - 3; /* two-character token, ending NUL */
9332 Perl_croak(aTHX_ ident_too_long);
9333 if (isALNUM(*s)) /* UTF handled below */
9335 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
9340 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
9344 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
9345 char *t = s + UTF8SKIP(s);
9346 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
9348 if (d + (t - s) > e)
9349 Perl_croak(aTHX_ ident_too_long);
9350 Copy(s, d, t - s, char);
9363 S_scan_ident(pTHX_ register char *s, register const char *send, char *dest, STRLEN destlen, I32 ck_uni)
9367 char *bracket = Nullch;
9373 e = d + destlen - 3; /* two-character token, ending NUL */
9375 while (isDIGIT(*s)) {
9377 Perl_croak(aTHX_ ident_too_long);
9384 Perl_croak(aTHX_ ident_too_long);
9385 if (isALNUM(*s)) /* UTF handled below */
9387 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
9392 else if (*s == ':' && s[1] == ':') {
9396 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
9397 char *t = s + UTF8SKIP(s);
9398 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
9400 if (d + (t - s) > e)
9401 Perl_croak(aTHX_ ident_too_long);
9402 Copy(s, d, t - s, char);
9413 if (PL_lex_state != LEX_NORMAL)
9414 PL_lex_state = LEX_INTERPENDMAYBE;
9417 if (*s == '$' && s[1] &&
9418 (isALNUM_lazy_if(s+1,UTF) || s[1] == '$' || s[1] == '{' || strnEQ(s+1,"::",2)) )
9431 if (*d == '^' && *s && isCONTROLVAR(*s)) {
9436 if (isSPACE(s[-1])) {
9438 const char ch = *s++;
9439 if (!SPACE_OR_TAB(ch)) {
9445 if (isIDFIRST_lazy_if(d,UTF)) {
9449 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
9451 while (e < send && UTF8_IS_CONTINUED(*e) && is_utf8_mark((U8*)e))
9454 Copy(s, d, e - s, char);
9459 while ((isALNUM(*s) || *s == ':') && d < e)
9462 Perl_croak(aTHX_ ident_too_long);
9465 while (s < send && SPACE_OR_TAB(*s)) s++;
9466 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
9467 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
9468 const char *brack = *s == '[' ? "[...]" : "{...}";
9469 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
9470 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
9471 funny, dest, brack, funny, dest, brack);
9474 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
9478 /* Handle extended ${^Foo} variables
9479 * 1999-02-27 mjd-perl-patch@plover.com */
9480 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
9484 while (isALNUM(*s) && d < e) {
9488 Perl_croak(aTHX_ ident_too_long);
9493 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets) {
9494 PL_lex_state = LEX_INTERPEND;
9499 if (PL_lex_state == LEX_NORMAL) {
9500 if (ckWARN(WARN_AMBIGUOUS) &&
9501 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
9503 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
9504 "Ambiguous use of %c{%s} resolved to %c%s",
9505 funny, dest, funny, dest);
9510 s = bracket; /* let the parser handle it */
9514 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
9515 PL_lex_state = LEX_INTERPEND;
9520 Perl_pmflag(pTHX_ U32* pmfl, int ch)
9525 *pmfl |= PMf_GLOBAL;
9527 *pmfl |= PMf_CONTINUE;
9531 *pmfl |= PMf_MULTILINE;
9533 *pmfl |= PMf_SINGLELINE;
9535 *pmfl |= PMf_EXTENDED;
9539 S_scan_pat(pTHX_ char *start, I32 type)
9542 char *s = scan_str(start,FALSE,FALSE);
9545 char * const delimiter = skipspace(start);
9546 Perl_croak(aTHX_ *delimiter == '?'
9547 ? "Search pattern not terminated or ternary operator parsed as search pattern"
9548 : "Search pattern not terminated" );
9551 pm = (PMOP*)newPMOP(type, 0);
9552 if (PL_multi_open == '?')
9553 pm->op_pmflags |= PMf_ONCE;
9555 while (*s && strchr("iomsx", *s))
9556 pmflag(&pm->op_pmflags,*s++);
9559 while (*s && strchr("iogcmsx", *s))
9560 pmflag(&pm->op_pmflags,*s++);
9562 /* issue a warning if /c is specified,but /g is not */
9563 if ((pm->op_pmflags & PMf_CONTINUE) && !(pm->op_pmflags & PMf_GLOBAL)
9564 && ckWARN(WARN_REGEXP))
9566 Perl_warner(aTHX_ packWARN(WARN_REGEXP), "Use of /c modifier is meaningless without /g" );
9569 pm->op_pmpermflags = pm->op_pmflags;
9571 PL_lex_op = (OP*)pm;
9572 yylval.ival = OP_MATCH;
9577 S_scan_subst(pTHX_ char *start)
9585 yylval.ival = OP_NULL;
9587 s = scan_str(start,FALSE,FALSE);
9590 Perl_croak(aTHX_ "Substitution pattern not terminated");
9592 if (s[-1] == PL_multi_open)
9595 first_start = PL_multi_start;
9596 s = scan_str(s,FALSE,FALSE);
9599 SvREFCNT_dec(PL_lex_stuff);
9600 PL_lex_stuff = Nullsv;
9602 Perl_croak(aTHX_ "Substitution replacement not terminated");
9604 PL_multi_start = first_start; /* so whole substitution is taken together */
9606 pm = (PMOP*)newPMOP(OP_SUBST, 0);
9612 else if (strchr("iogcmsx", *s))
9613 pmflag(&pm->op_pmflags,*s++);
9618 if ((pm->op_pmflags & PMf_CONTINUE) && ckWARN(WARN_REGEXP)) {
9619 Perl_warner(aTHX_ packWARN(WARN_REGEXP), "Use of /c modifier is meaningless in s///" );
9624 PL_sublex_info.super_bufptr = s;
9625 PL_sublex_info.super_bufend = PL_bufend;
9627 pm->op_pmflags |= PMf_EVAL;
9628 repl = newSVpvn("",0);
9630 sv_catpv(repl, es ? "eval " : "do ");
9631 sv_catpvn(repl, "{ ", 2);
9632 sv_catsv(repl, PL_lex_repl);
9633 sv_catpvn(repl, " };", 2);
9635 SvREFCNT_dec(PL_lex_repl);
9639 pm->op_pmpermflags = pm->op_pmflags;
9640 PL_lex_op = (OP*)pm;
9641 yylval.ival = OP_SUBST;
9646 S_scan_trans(pTHX_ char *start)
9655 yylval.ival = OP_NULL;
9657 s = scan_str(start,FALSE,FALSE);
9659 Perl_croak(aTHX_ "Transliteration pattern not terminated");
9660 if (s[-1] == PL_multi_open)
9663 s = scan_str(s,FALSE,FALSE);
9666 SvREFCNT_dec(PL_lex_stuff);
9667 PL_lex_stuff = Nullsv;
9669 Perl_croak(aTHX_ "Transliteration replacement not terminated");
9672 complement = del = squash = 0;
9676 complement = OPpTRANS_COMPLEMENT;
9679 del = OPpTRANS_DELETE;
9682 squash = OPpTRANS_SQUASH;
9691 Newx(tbl, complement&&!del?258:256, short);
9692 o = newPVOP(OP_TRANS, 0, (char*)tbl);
9693 o->op_private &= ~OPpTRANS_ALL;
9694 o->op_private |= del|squash|complement|
9695 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
9696 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
9699 yylval.ival = OP_TRANS;
9704 S_scan_heredoc(pTHX_ register char *s)
9707 I32 op_type = OP_SCALAR;
9711 const char newline[] = "\n";
9712 const char *found_newline;
9716 const int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
9720 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
9723 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
9724 if (*peek == '`' || *peek == '\'' || *peek =='"') {
9727 s = delimcpy(d, e, s, PL_bufend, term, &len);
9737 if (!isALNUM_lazy_if(s,UTF))
9738 deprecate_old("bare << to mean <<\"\"");
9739 for (; isALNUM_lazy_if(s,UTF); s++) {
9744 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
9745 Perl_croak(aTHX_ "Delimiter for here document is too long");
9748 len = d - PL_tokenbuf;
9749 #ifndef PERL_STRICT_CR
9750 d = strchr(s, '\r');
9752 char * const olds = s;
9754 while (s < PL_bufend) {
9760 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
9769 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
9773 if ( outer || !(found_newline = ninstr(s,PL_bufend,newline,newline+1)) ) {
9774 herewas = newSVpvn(s,PL_bufend-s);
9778 herewas = newSVpvn(s,found_newline-s);
9780 s += SvCUR(herewas);
9782 tmpstr = NEWSV(87,79);
9783 sv_upgrade(tmpstr, SVt_PVIV);
9786 SvIV_set(tmpstr, -1);
9788 else if (term == '`') {
9789 op_type = OP_BACKTICK;
9790 SvIV_set(tmpstr, '\\');
9794 PL_multi_start = CopLINE(PL_curcop);
9795 PL_multi_open = PL_multi_close = '<';
9796 term = *PL_tokenbuf;
9797 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
9798 char *bufptr = PL_sublex_info.super_bufptr;
9799 char *bufend = PL_sublex_info.super_bufend;
9800 char * const olds = s - SvCUR(herewas);
9801 s = strchr(bufptr, '\n');
9805 while (s < bufend &&
9806 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
9808 CopLINE_inc(PL_curcop);
9811 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
9812 missingterm(PL_tokenbuf);
9814 sv_setpvn(herewas,bufptr,d-bufptr+1);
9815 sv_setpvn(tmpstr,d+1,s-d);
9817 sv_catpvn(herewas,s,bufend-s);
9818 Copy(SvPVX_const(herewas),bufptr,SvCUR(herewas) + 1,char);
9825 while (s < PL_bufend &&
9826 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
9828 CopLINE_inc(PL_curcop);
9830 if (s >= PL_bufend) {
9831 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
9832 missingterm(PL_tokenbuf);
9834 sv_setpvn(tmpstr,d+1,s-d);
9836 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
9838 sv_catpvn(herewas,s,PL_bufend-s);
9839 sv_setsv(PL_linestr,herewas);
9840 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
9841 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
9842 PL_last_lop = PL_last_uni = Nullch;
9845 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
9846 while (s >= PL_bufend) { /* multiple line string? */
9848 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
9849 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
9850 missingterm(PL_tokenbuf);
9852 CopLINE_inc(PL_curcop);
9853 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
9854 PL_last_lop = PL_last_uni = Nullch;
9855 #ifndef PERL_STRICT_CR
9856 if (PL_bufend - PL_linestart >= 2) {
9857 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
9858 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
9860 PL_bufend[-2] = '\n';
9862 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
9864 else if (PL_bufend[-1] == '\r')
9865 PL_bufend[-1] = '\n';
9867 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
9868 PL_bufend[-1] = '\n';
9870 if (PERLDB_LINE && PL_curstash != PL_debstash) {
9871 SV *sv = NEWSV(88,0);
9873 sv_upgrade(sv, SVt_PVMG);
9874 sv_setsv(sv,PL_linestr);
9877 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
9879 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
9880 STRLEN off = PL_bufend - 1 - SvPVX_const(PL_linestr);
9881 *(SvPVX(PL_linestr) + off ) = ' ';
9882 sv_catsv(PL_linestr,herewas);
9883 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
9884 s = SvPVX(PL_linestr) + off; /* In case PV of PL_linestr moved. */
9888 sv_catsv(tmpstr,PL_linestr);
9893 PL_multi_end = CopLINE(PL_curcop);
9894 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
9895 SvPV_shrink_to_cur(tmpstr);
9897 SvREFCNT_dec(herewas);
9899 if (UTF && is_utf8_string((U8*)SvPVX_const(tmpstr), SvCUR(tmpstr)))
9901 else if (PL_encoding)
9902 sv_recode_to_utf8(tmpstr, PL_encoding);
9904 PL_lex_stuff = tmpstr;
9905 yylval.ival = op_type;
9910 takes: current position in input buffer
9911 returns: new position in input buffer
9912 side-effects: yylval and lex_op are set.
9917 <FH> read from filehandle
9918 <pkg::FH> read from package qualified filehandle
9919 <pkg'FH> read from package qualified filehandle
9920 <$fh> read from filehandle in $fh
9926 S_scan_inputsymbol(pTHX_ char *start)
9928 register char *s = start; /* current position in buffer */
9934 d = PL_tokenbuf; /* start of temp holding space */
9935 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
9936 end = strchr(s, '\n');
9939 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
9941 /* die if we didn't have space for the contents of the <>,
9942 or if it didn't end, or if we see a newline
9945 if (len >= sizeof PL_tokenbuf)
9946 Perl_croak(aTHX_ "Excessively long <> operator");
9948 Perl_croak(aTHX_ "Unterminated <> operator");
9953 Remember, only scalar variables are interpreted as filehandles by
9954 this code. Anything more complex (e.g., <$fh{$num}>) will be
9955 treated as a glob() call.
9956 This code makes use of the fact that except for the $ at the front,
9957 a scalar variable and a filehandle look the same.
9959 if (*d == '$' && d[1]) d++;
9961 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
9962 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
9965 /* If we've tried to read what we allow filehandles to look like, and
9966 there's still text left, then it must be a glob() and not a getline.
9967 Use scan_str to pull out the stuff between the <> and treat it
9968 as nothing more than a string.
9971 if (d - PL_tokenbuf != len) {
9972 yylval.ival = OP_GLOB;
9974 s = scan_str(start,FALSE,FALSE);
9976 Perl_croak(aTHX_ "Glob not terminated");
9980 bool readline_overriden = FALSE;
9981 GV *gv_readline = Nullgv;
9983 /* we're in a filehandle read situation */
9986 /* turn <> into <ARGV> */
9988 Copy("ARGV",d,5,char);
9990 /* Check whether readline() is overriden */
9991 if (((gv_readline = gv_fetchpv("readline", 0, SVt_PVCV))
9992 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline))
9994 ((gvp = (GV**)hv_fetch(PL_globalstash, "readline", 8, FALSE))
9995 && (gv_readline = *gvp) != (GV*)&PL_sv_undef
9996 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline)))
9997 readline_overriden = TRUE;
9999 /* if <$fh>, create the ops to turn the variable into a
10005 /* try to find it in the pad for this block, otherwise find
10006 add symbol table ops
10008 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
10009 if (PAD_COMPNAME_FLAGS(tmp) & SVpad_OUR) {
10010 HV *stash = PAD_COMPNAME_OURSTASH(tmp);
10011 HEK *stashname = HvNAME_HEK(stash);
10012 SV *sym = sv_2mortal(newSVhek(stashname));
10013 sv_catpvn(sym, "::", 2);
10014 sv_catpv(sym, d+1);
10019 OP *o = newOP(OP_PADSV, 0);
10021 PL_lex_op = readline_overriden
10022 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
10023 append_elem(OP_LIST, o,
10024 newCVREF(0, newGVOP(OP_GV,0,gv_readline))))
10025 : (OP*)newUNOP(OP_READLINE, 0, o);
10034 ? (GV_ADDMULTI | GV_ADDINEVAL)
10037 PL_lex_op = readline_overriden
10038 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
10039 append_elem(OP_LIST,
10040 newUNOP(OP_RV2SV, 0, newGVOP(OP_GV, 0, gv)),
10041 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
10042 : (OP*)newUNOP(OP_READLINE, 0,
10043 newUNOP(OP_RV2SV, 0,
10044 newGVOP(OP_GV, 0, gv)));
10046 if (!readline_overriden)
10047 PL_lex_op->op_flags |= OPf_SPECIAL;
10048 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
10049 yylval.ival = OP_NULL;
10052 /* If it's none of the above, it must be a literal filehandle
10053 (<Foo::BAR> or <FOO>) so build a simple readline OP */
10055 GV *gv = gv_fetchpv(d, GV_ADD, SVt_PVIO);
10056 PL_lex_op = readline_overriden
10057 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
10058 append_elem(OP_LIST,
10059 newGVOP(OP_GV, 0, gv),
10060 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
10061 : (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
10062 yylval.ival = OP_NULL;
10071 takes: start position in buffer
10072 keep_quoted preserve \ on the embedded delimiter(s)
10073 keep_delims preserve the delimiters around the string
10074 returns: position to continue reading from buffer
10075 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
10076 updates the read buffer.
10078 This subroutine pulls a string out of the input. It is called for:
10079 q single quotes q(literal text)
10080 ' single quotes 'literal text'
10081 qq double quotes qq(interpolate $here please)
10082 " double quotes "interpolate $here please"
10083 qx backticks qx(/bin/ls -l)
10084 ` backticks `/bin/ls -l`
10085 qw quote words @EXPORT_OK = qw( func() $spam )
10086 m// regexp match m/this/
10087 s/// regexp substitute s/this/that/
10088 tr/// string transliterate tr/this/that/
10089 y/// string transliterate y/this/that/
10090 ($*@) sub prototypes sub foo ($)
10091 (stuff) sub attr parameters sub foo : attr(stuff)
10092 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
10094 In most of these cases (all but <>, patterns and transliterate)
10095 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
10096 calls scan_str(). s/// makes yylex() call scan_subst() which calls
10097 scan_str(). tr/// and y/// make yylex() call scan_trans() which
10100 It skips whitespace before the string starts, and treats the first
10101 character as the delimiter. If the delimiter is one of ([{< then
10102 the corresponding "close" character )]}> is used as the closing
10103 delimiter. It allows quoting of delimiters, and if the string has
10104 balanced delimiters ([{<>}]) it allows nesting.
10106 On success, the SV with the resulting string is put into lex_stuff or,
10107 if that is already non-NULL, into lex_repl. The second case occurs only
10108 when parsing the RHS of the special constructs s/// and tr/// (y///).
10109 For convenience, the terminating delimiter character is stuffed into
10114 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
10116 SV *sv; /* scalar value: string */
10117 char *tmps; /* temp string, used for delimiter matching */
10118 register char *s = start; /* current position in the buffer */
10119 register char term; /* terminating character */
10120 register char *to; /* current position in the sv's data */
10121 I32 brackets = 1; /* bracket nesting level */
10122 bool has_utf8 = FALSE; /* is there any utf8 content? */
10123 I32 termcode; /* terminating char. code */
10124 U8 termstr[UTF8_MAXBYTES]; /* terminating string */
10125 STRLEN termlen; /* length of terminating string */
10126 char *last = NULL; /* last position for nesting bracket */
10128 /* skip space before the delimiter */
10132 /* mark where we are, in case we need to report errors */
10135 /* after skipping whitespace, the next character is the terminator */
10138 termcode = termstr[0] = term;
10142 termcode = utf8_to_uvchr((U8*)s, &termlen);
10143 Copy(s, termstr, termlen, U8);
10144 if (!UTF8_IS_INVARIANT(term))
10148 /* mark where we are */
10149 PL_multi_start = CopLINE(PL_curcop);
10150 PL_multi_open = term;
10152 /* find corresponding closing delimiter */
10153 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
10154 termcode = termstr[0] = term = tmps[5];
10156 PL_multi_close = term;
10158 /* create a new SV to hold the contents. 87 is leak category, I'm
10159 assuming. 79 is the SV's initial length. What a random number. */
10161 sv_upgrade(sv, SVt_PVIV);
10162 SvIV_set(sv, termcode);
10163 (void)SvPOK_only(sv); /* validate pointer */
10165 /* move past delimiter and try to read a complete string */
10167 sv_catpvn(sv, s, termlen);
10170 if (PL_encoding && !UTF) {
10174 int offset = s - SvPVX_const(PL_linestr);
10175 const bool found = sv_cat_decode(sv, PL_encoding, PL_linestr,
10176 &offset, (char*)termstr, termlen);
10177 const char *ns = SvPVX_const(PL_linestr) + offset;
10178 char *svlast = SvEND(sv) - 1;
10180 for (; s < ns; s++) {
10181 if (*s == '\n' && !PL_rsfp)
10182 CopLINE_inc(PL_curcop);
10185 goto read_more_line;
10187 /* handle quoted delimiters */
10188 if (SvCUR(sv) > 1 && *(svlast-1) == '\\') {
10190 for (t = svlast-2; t >= SvPVX_const(sv) && *t == '\\';)
10192 if ((svlast-1 - t) % 2) {
10193 if (!keep_quoted) {
10194 *(svlast-1) = term;
10196 SvCUR_set(sv, SvCUR(sv) - 1);
10201 if (PL_multi_open == PL_multi_close) {
10209 for (t = w = last; t < svlast; w++, t++) {
10210 /* At here, all closes are "was quoted" one,
10211 so we don't check PL_multi_close. */
10213 if (!keep_quoted && *(t+1) == PL_multi_open)
10218 else if (*t == PL_multi_open)
10226 SvCUR_set(sv, w - SvPVX_const(sv));
10229 if (--brackets <= 0)
10234 if (!keep_delims) {
10235 SvCUR_set(sv, SvCUR(sv) - 1);
10241 /* extend sv if need be */
10242 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
10243 /* set 'to' to the next character in the sv's string */
10244 to = SvPVX(sv)+SvCUR(sv);
10246 /* if open delimiter is the close delimiter read unbridle */
10247 if (PL_multi_open == PL_multi_close) {
10248 for (; s < PL_bufend; s++,to++) {
10249 /* embedded newlines increment the current line number */
10250 if (*s == '\n' && !PL_rsfp)
10251 CopLINE_inc(PL_curcop);
10252 /* handle quoted delimiters */
10253 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
10254 if (!keep_quoted && s[1] == term)
10256 /* any other quotes are simply copied straight through */
10260 /* terminate when run out of buffer (the for() condition), or
10261 have found the terminator */
10262 else if (*s == term) {
10265 if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen))
10268 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
10274 /* if the terminator isn't the same as the start character (e.g.,
10275 matched brackets), we have to allow more in the quoting, and
10276 be prepared for nested brackets.
10279 /* read until we run out of string, or we find the terminator */
10280 for (; s < PL_bufend; s++,to++) {
10281 /* embedded newlines increment the line count */
10282 if (*s == '\n' && !PL_rsfp)
10283 CopLINE_inc(PL_curcop);
10284 /* backslashes can escape the open or closing characters */
10285 if (*s == '\\' && s+1 < PL_bufend) {
10286 if (!keep_quoted &&
10287 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
10292 /* allow nested opens and closes */
10293 else if (*s == PL_multi_close && --brackets <= 0)
10295 else if (*s == PL_multi_open)
10297 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
10302 /* terminate the copied string and update the sv's end-of-string */
10304 SvCUR_set(sv, to - SvPVX_const(sv));
10307 * this next chunk reads more into the buffer if we're not done yet
10311 break; /* handle case where we are done yet :-) */
10313 #ifndef PERL_STRICT_CR
10314 if (to - SvPVX_const(sv) >= 2) {
10315 if ((to[-2] == '\r' && to[-1] == '\n') ||
10316 (to[-2] == '\n' && to[-1] == '\r'))
10320 SvCUR_set(sv, to - SvPVX_const(sv));
10322 else if (to[-1] == '\r')
10325 else if (to - SvPVX_const(sv) == 1 && to[-1] == '\r')
10330 /* if we're out of file, or a read fails, bail and reset the current
10331 line marker so we can report where the unterminated string began
10334 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
10336 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
10339 /* we read a line, so increment our line counter */
10340 CopLINE_inc(PL_curcop);
10342 /* update debugger info */
10343 if (PERLDB_LINE && PL_curstash != PL_debstash) {
10344 SV *sv = NEWSV(88,0);
10346 sv_upgrade(sv, SVt_PVMG);
10347 sv_setsv(sv,PL_linestr);
10348 (void)SvIOK_on(sv);
10350 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
10353 /* having changed the buffer, we must update PL_bufend */
10354 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
10355 PL_last_lop = PL_last_uni = Nullch;
10358 /* at this point, we have successfully read the delimited string */
10360 if (!PL_encoding || UTF) {
10362 sv_catpvn(sv, s, termlen);
10365 if (has_utf8 || PL_encoding)
10368 PL_multi_end = CopLINE(PL_curcop);
10370 /* if we allocated too much space, give some back */
10371 if (SvCUR(sv) + 5 < SvLEN(sv)) {
10372 SvLEN_set(sv, SvCUR(sv) + 1);
10373 SvPV_renew(sv, SvLEN(sv));
10376 /* decide whether this is the first or second quoted string we've read
10389 takes: pointer to position in buffer
10390 returns: pointer to new position in buffer
10391 side-effects: builds ops for the constant in yylval.op
10393 Read a number in any of the formats that Perl accepts:
10395 \d(_?\d)*(\.(\d(_?\d)*)?)?[Ee][\+\-]?(\d(_?\d)*) 12 12.34 12.
10396 \.\d(_?\d)*[Ee][\+\-]?(\d(_?\d)*) .34
10399 0x[0-9A-Fa-f](_?[0-9A-Fa-f])*
10401 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
10404 If it reads a number without a decimal point or an exponent, it will
10405 try converting the number to an integer and see if it can do so
10406 without loss of precision.
10410 Perl_scan_num(pTHX_ const char *start, YYSTYPE* lvalp)
10412 register const char *s = start; /* current position in buffer */
10413 register char *d; /* destination in temp buffer */
10414 register char *e; /* end of temp buffer */
10415 NV nv; /* number read, as a double */
10416 SV *sv = Nullsv; /* place to put the converted number */
10417 bool floatit; /* boolean: int or float? */
10418 const char *lastub = 0; /* position of last underbar */
10419 static char const number_too_long[] = "Number too long";
10421 /* We use the first character to decide what type of number this is */
10425 Perl_croak(aTHX_ "panic: scan_num");
10427 /* if it starts with a 0, it could be an octal number, a decimal in
10428 0.13 disguise, or a hexadecimal number, or a binary number. */
10432 u holds the "number so far"
10433 shift the power of 2 of the base
10434 (hex == 4, octal == 3, binary == 1)
10435 overflowed was the number more than we can hold?
10437 Shift is used when we add a digit. It also serves as an "are
10438 we in octal/hex/binary?" indicator to disallow hex characters
10439 when in octal mode.
10444 bool overflowed = FALSE;
10445 bool just_zero = TRUE; /* just plain 0 or binary number? */
10446 static const NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
10447 static const char* const bases[5] =
10448 { "", "binary", "", "octal", "hexadecimal" };
10449 static const char* const Bases[5] =
10450 { "", "Binary", "", "Octal", "Hexadecimal" };
10451 static const char* const maxima[5] =
10453 "0b11111111111111111111111111111111",
10457 const char *base, *Base, *max;
10459 /* check for hex */
10464 } else if (s[1] == 'b') {
10469 /* check for a decimal in disguise */
10470 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
10472 /* so it must be octal */
10479 if (ckWARN(WARN_SYNTAX))
10480 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10481 "Misplaced _ in number");
10485 base = bases[shift];
10486 Base = Bases[shift];
10487 max = maxima[shift];
10489 /* read the rest of the number */
10491 /* x is used in the overflow test,
10492 b is the digit we're adding on. */
10497 /* if we don't mention it, we're done */
10501 /* _ are ignored -- but warned about if consecutive */
10503 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
10504 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10505 "Misplaced _ in number");
10509 /* 8 and 9 are not octal */
10510 case '8': case '9':
10512 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
10516 case '2': case '3': case '4':
10517 case '5': case '6': case '7':
10519 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
10522 case '0': case '1':
10523 b = *s++ & 15; /* ASCII digit -> value of digit */
10527 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
10528 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
10529 /* make sure they said 0x */
10532 b = (*s++ & 7) + 9;
10534 /* Prepare to put the digit we have onto the end
10535 of the number so far. We check for overflows.
10541 x = u << shift; /* make room for the digit */
10543 if ((x >> shift) != u
10544 && !(PL_hints & HINT_NEW_BINARY)) {
10547 if (ckWARN_d(WARN_OVERFLOW))
10548 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
10549 "Integer overflow in %s number",
10552 u = x | b; /* add the digit to the end */
10555 n *= nvshift[shift];
10556 /* If an NV has not enough bits in its
10557 * mantissa to represent an UV this summing of
10558 * small low-order numbers is a waste of time
10559 * (because the NV cannot preserve the
10560 * low-order bits anyway): we could just
10561 * remember when did we overflow and in the
10562 * end just multiply n by the right
10570 /* if we get here, we had success: make a scalar value from
10575 /* final misplaced underbar check */
10576 if (s[-1] == '_') {
10577 if (ckWARN(WARN_SYNTAX))
10578 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
10583 if (n > 4294967295.0 && ckWARN(WARN_PORTABLE))
10584 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
10585 "%s number > %s non-portable",
10591 if (u > 0xffffffff && ckWARN(WARN_PORTABLE))
10592 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
10593 "%s number > %s non-portable",
10598 if (just_zero && (PL_hints & HINT_NEW_INTEGER))
10599 sv = new_constant(start, s - start, "integer",
10601 else if (PL_hints & HINT_NEW_BINARY)
10602 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
10607 handle decimal numbers.
10608 we're also sent here when we read a 0 as the first digit
10610 case '1': case '2': case '3': case '4': case '5':
10611 case '6': case '7': case '8': case '9': case '.':
10614 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
10617 /* read next group of digits and _ and copy into d */
10618 while (isDIGIT(*s) || *s == '_') {
10619 /* skip underscores, checking for misplaced ones
10623 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
10624 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10625 "Misplaced _ in number");
10629 /* check for end of fixed-length buffer */
10631 Perl_croak(aTHX_ number_too_long);
10632 /* if we're ok, copy the character */
10637 /* final misplaced underbar check */
10638 if (lastub && s == lastub + 1) {
10639 if (ckWARN(WARN_SYNTAX))
10640 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
10643 /* read a decimal portion if there is one. avoid
10644 3..5 being interpreted as the number 3. followed
10647 if (*s == '.' && s[1] != '.') {
10652 if (ckWARN(WARN_SYNTAX))
10653 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10654 "Misplaced _ in number");
10658 /* copy, ignoring underbars, until we run out of digits.
10660 for (; isDIGIT(*s) || *s == '_'; s++) {
10661 /* fixed length buffer check */
10663 Perl_croak(aTHX_ number_too_long);
10665 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
10666 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10667 "Misplaced _ in number");
10673 /* fractional part ending in underbar? */
10674 if (s[-1] == '_') {
10675 if (ckWARN(WARN_SYNTAX))
10676 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10677 "Misplaced _ in number");
10679 if (*s == '.' && isDIGIT(s[1])) {
10680 /* oops, it's really a v-string, but without the "v" */
10686 /* read exponent part, if present */
10687 if ((*s == 'e' || *s == 'E') && strchr("+-0123456789_", s[1])) {
10691 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
10692 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
10694 /* stray preinitial _ */
10696 if (ckWARN(WARN_SYNTAX))
10697 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10698 "Misplaced _ in number");
10702 /* allow positive or negative exponent */
10703 if (*s == '+' || *s == '-')
10706 /* stray initial _ */
10708 if (ckWARN(WARN_SYNTAX))
10709 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10710 "Misplaced _ in number");
10714 /* read digits of exponent */
10715 while (isDIGIT(*s) || *s == '_') {
10718 Perl_croak(aTHX_ number_too_long);
10722 if (((lastub && s == lastub + 1) ||
10723 (!isDIGIT(s[1]) && s[1] != '_'))
10724 && ckWARN(WARN_SYNTAX))
10725 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10726 "Misplaced _ in number");
10733 /* make an sv from the string */
10737 We try to do an integer conversion first if no characters
10738 indicating "float" have been found.
10743 int flags = grok_number (PL_tokenbuf, d - PL_tokenbuf, &uv);
10745 if (flags == IS_NUMBER_IN_UV) {
10747 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
10750 } else if (flags == (IS_NUMBER_IN_UV | IS_NUMBER_NEG)) {
10751 if (uv <= (UV) IV_MIN)
10752 sv_setiv(sv, -(IV)uv);
10759 /* terminate the string */
10761 nv = Atof(PL_tokenbuf);
10765 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
10766 (PL_hints & HINT_NEW_INTEGER) )
10767 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
10768 (floatit ? "float" : "integer"),
10772 /* if it starts with a v, it could be a v-string */
10775 sv = NEWSV(92,5); /* preallocate storage space */
10776 s = scan_vstring(s,sv);
10780 /* make the op for the constant and return */
10783 lvalp->opval = newSVOP(OP_CONST, 0, sv);
10785 lvalp->opval = Nullop;
10791 S_scan_formline(pTHX_ register char *s)
10793 register char *eol;
10795 SV *stuff = newSVpvn("",0);
10796 bool needargs = FALSE;
10797 bool eofmt = FALSE;
10799 while (!needargs) {
10801 #ifdef PERL_STRICT_CR
10802 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
10804 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
10806 if (*t == '\n' || t == PL_bufend) {
10811 if (PL_in_eval && !PL_rsfp) {
10812 eol = (char *) memchr(s,'\n',PL_bufend-s);
10817 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
10819 for (t = s; t < eol; t++) {
10820 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
10822 goto enough; /* ~~ must be first line in formline */
10824 if (*t == '@' || *t == '^')
10828 sv_catpvn(stuff, s, eol-s);
10829 #ifndef PERL_STRICT_CR
10830 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
10831 char *end = SvPVX(stuff) + SvCUR(stuff);
10834 SvCUR_set(stuff, SvCUR(stuff) - 1);
10843 s = filter_gets(PL_linestr, PL_rsfp, 0);
10844 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
10845 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
10846 PL_last_lop = PL_last_uni = Nullch;
10855 if (SvCUR(stuff)) {
10858 PL_lex_state = LEX_NORMAL;
10859 PL_nextval[PL_nexttoke].ival = 0;
10863 PL_lex_state = LEX_FORMLINE;
10865 if (UTF && is_utf8_string((U8*)SvPVX_const(stuff), SvCUR(stuff)))
10867 else if (PL_encoding)
10868 sv_recode_to_utf8(stuff, PL_encoding);
10870 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
10872 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
10876 SvREFCNT_dec(stuff);
10878 PL_lex_formbrack = 0;
10889 PL_cshlen = strlen(PL_cshname);
10894 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
10896 const I32 oldsavestack_ix = PL_savestack_ix;
10897 CV* outsidecv = PL_compcv;
10900 assert(SvTYPE(PL_compcv) == SVt_PVCV);
10902 SAVEI32(PL_subline);
10903 save_item(PL_subname);
10904 SAVESPTR(PL_compcv);
10906 PL_compcv = (CV*)NEWSV(1104,0);
10907 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
10908 CvFLAGS(PL_compcv) |= flags;
10910 PL_subline = CopLINE(PL_curcop);
10911 CvPADLIST(PL_compcv) = pad_new(padnew_SAVE|padnew_SAVESUB);
10912 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
10913 CvOUTSIDE_SEQ(PL_compcv) = PL_cop_seqmax;
10915 return oldsavestack_ix;
10919 #pragma segment Perl_yylex
10922 Perl_yywarn(pTHX_ const char *s)
10924 PL_in_eval |= EVAL_WARNONLY;
10926 PL_in_eval &= ~EVAL_WARNONLY;
10931 Perl_yyerror(pTHX_ const char *s)
10933 const char *where = NULL;
10934 const char *context = NULL;
10938 if (!yychar || (yychar == ';' && !PL_rsfp))
10940 else if (PL_oldoldbufptr && PL_bufptr > PL_oldoldbufptr &&
10941 PL_bufptr - PL_oldoldbufptr < 200 && PL_oldoldbufptr != PL_oldbufptr &&
10942 PL_oldbufptr != PL_bufptr) {
10945 The code below is removed for NetWare because it abends/crashes on NetWare
10946 when the script has error such as not having the closing quotes like:
10947 if ($var eq "value)
10948 Checking of white spaces is anyway done in NetWare code.
10951 while (isSPACE(*PL_oldoldbufptr))
10954 context = PL_oldoldbufptr;
10955 contlen = PL_bufptr - PL_oldoldbufptr;
10957 else if (PL_oldbufptr && PL_bufptr > PL_oldbufptr &&
10958 PL_bufptr - PL_oldbufptr < 200 && PL_oldbufptr != PL_bufptr) {
10961 The code below is removed for NetWare because it abends/crashes on NetWare
10962 when the script has error such as not having the closing quotes like:
10963 if ($var eq "value)
10964 Checking of white spaces is anyway done in NetWare code.
10967 while (isSPACE(*PL_oldbufptr))
10970 context = PL_oldbufptr;
10971 contlen = PL_bufptr - PL_oldbufptr;
10973 else if (yychar > 255)
10974 where = "next token ???";
10975 else if (yychar == -2) { /* YYEMPTY */
10976 if (PL_lex_state == LEX_NORMAL ||
10977 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
10978 where = "at end of line";
10979 else if (PL_lex_inpat)
10980 where = "within pattern";
10982 where = "within string";
10985 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
10987 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
10988 else if (isPRINT_LC(yychar))
10989 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
10991 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
10992 where = SvPVX_const(where_sv);
10994 msg = sv_2mortal(newSVpv(s, 0));
10995 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
10996 OutCopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
10998 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
11000 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
11001 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
11002 Perl_sv_catpvf(aTHX_ msg,
11003 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
11004 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
11007 if (PL_in_eval & EVAL_WARNONLY && ckWARN_d(WARN_SYNTAX))
11008 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "%"SVf, msg);
11011 if (PL_error_count >= 10) {
11012 if (PL_in_eval && SvCUR(ERRSV))
11013 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
11014 ERRSV, OutCopFILE(PL_curcop));
11016 Perl_croak(aTHX_ "%s has too many errors.\n",
11017 OutCopFILE(PL_curcop));
11020 PL_in_my_stash = NULL;
11024 #pragma segment Main
11028 S_swallow_bom(pTHX_ U8 *s)
11030 const STRLEN slen = SvCUR(PL_linestr);
11033 if (s[1] == 0xFE) {
11034 /* UTF-16 little-endian? (or UTF32-LE?) */
11035 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
11036 Perl_croak(aTHX_ "Unsupported script encoding UTF32-LE");
11037 #ifndef PERL_NO_UTF16_FILTER
11038 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF16-LE script encoding (BOM)\n");
11041 if (PL_bufend > (char*)s) {
11045 filter_add(utf16rev_textfilter, NULL);
11046 Newx(news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
11047 utf16_to_utf8_reversed(s, news,
11048 PL_bufend - (char*)s - 1,
11050 sv_setpvn(PL_linestr, (const char*)news, newlen);
11052 SvUTF8_on(PL_linestr);
11053 s = (U8*)SvPVX(PL_linestr);
11054 PL_bufend = SvPVX(PL_linestr) + newlen;
11057 Perl_croak(aTHX_ "Unsupported script encoding UTF16-LE");
11062 if (s[1] == 0xFF) { /* UTF-16 big-endian? */
11063 #ifndef PERL_NO_UTF16_FILTER
11064 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (BOM)\n");
11067 if (PL_bufend > (char *)s) {
11071 filter_add(utf16_textfilter, NULL);
11072 Newx(news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
11073 utf16_to_utf8(s, news,
11074 PL_bufend - (char*)s,
11076 sv_setpvn(PL_linestr, (const char*)news, newlen);
11078 SvUTF8_on(PL_linestr);
11079 s = (U8*)SvPVX(PL_linestr);
11080 PL_bufend = SvPVX(PL_linestr) + newlen;
11083 Perl_croak(aTHX_ "Unsupported script encoding UTF16-BE");
11088 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
11089 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-8 script encoding (BOM)\n");
11090 s += 3; /* UTF-8 */
11096 if (s[2] == 0xFE && s[3] == 0xFF) {
11097 /* UTF-32 big-endian */
11098 Perl_croak(aTHX_ "Unsupported script encoding UTF32-BE");
11101 else if (s[2] == 0 && s[3] != 0) {
11104 * are a good indicator of UTF-16BE. */
11105 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (no BOM)\n");
11110 if (slen > 3 && s[1] == 0 && s[2] != 0 && s[3] == 0) {
11113 * are a good indicator of UTF-16LE. */
11114 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16LE script encoding (no BOM)\n");
11123 * Restore a source filter.
11127 restore_rsfp(pTHX_ void *f)
11129 PerlIO * const fp = (PerlIO*)f;
11131 if (PL_rsfp == PerlIO_stdin())
11132 PerlIO_clearerr(PL_rsfp);
11133 else if (PL_rsfp && (PL_rsfp != fp))
11134 PerlIO_close(PL_rsfp);
11138 #ifndef PERL_NO_UTF16_FILTER
11140 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
11142 const STRLEN old = SvCUR(sv);
11143 const I32 count = FILTER_READ(idx+1, sv, maxlen);
11144 DEBUG_P(PerlIO_printf(Perl_debug_log,
11145 "utf16_textfilter(%p): %d %d (%d)\n",
11146 utf16_textfilter, idx, maxlen, (int) count));
11150 Newx(tmps, SvCUR(sv) * 3 / 2 + 1, U8);
11151 Copy(SvPVX_const(sv), tmps, old, char);
11152 utf16_to_utf8((U8*)SvPVX_const(sv) + old, tmps + old,
11153 SvCUR(sv) - old, &newlen);
11154 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
11156 DEBUG_P({sv_dump(sv);});
11161 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
11163 const STRLEN old = SvCUR(sv);
11164 const I32 count = FILTER_READ(idx+1, sv, maxlen);
11165 DEBUG_P(PerlIO_printf(Perl_debug_log,
11166 "utf16rev_textfilter(%p): %d %d (%d)\n",
11167 utf16rev_textfilter, idx, maxlen, (int) count));
11171 Newx(tmps, SvCUR(sv) * 3 / 2 + 1, U8);
11172 Copy(SvPVX_const(sv), tmps, old, char);
11173 utf16_to_utf8((U8*)SvPVX_const(sv) + old, tmps + old,
11174 SvCUR(sv) - old, &newlen);
11175 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
11177 DEBUG_P({ sv_dump(sv); });
11183 Returns a pointer to the next character after the parsed
11184 vstring, as well as updating the passed in sv.
11186 Function must be called like
11189 s = scan_vstring(s,sv);
11191 The sv should already be large enough to store the vstring
11192 passed in, for performance reasons.
11197 Perl_scan_vstring(pTHX_ const char *s, SV *sv)
11199 const char *pos = s;
11200 const char *start = s;
11201 if (*pos == 'v') pos++; /* get past 'v' */
11202 while (pos < PL_bufend && (isDIGIT(*pos) || *pos == '_'))
11204 if ( *pos != '.') {
11205 /* this may not be a v-string if followed by => */
11206 const char *next = pos;
11207 while (next < PL_bufend && isSPACE(*next))
11209 if ((PL_bufend - next) >= 2 && *next == '=' && next[1] == '>' ) {
11210 /* return string not v-string */
11211 sv_setpvn(sv,(char *)s,pos-s);
11212 return (char *)pos;
11216 if (!isALPHA(*pos)) {
11217 U8 tmpbuf[UTF8_MAXBYTES+1];
11219 if (*s == 'v') s++; /* get past 'v' */
11221 sv_setpvn(sv, "", 0);
11227 /* this is atoi() that tolerates underscores */
11228 const char *end = pos;
11230 while (--end >= s) {
11235 rev += (*end - '0') * mult;
11237 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
11238 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
11239 "Integer overflow in decimal number");
11243 if (rev > 0x7FFFFFFF)
11244 Perl_croak(aTHX_ "In EBCDIC the v-string components cannot exceed 2147483647");
11246 /* Append native character for the rev point */
11247 tmpend = uvchr_to_utf8(tmpbuf, rev);
11248 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
11249 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(rev)))
11251 if (pos + 1 < PL_bufend && *pos == '.' && isDIGIT(pos[1]))
11257 while (pos < PL_bufend && (isDIGIT(*pos) || *pos == '_'))
11261 sv_magic(sv,NULL,PERL_MAGIC_vstring,(const char*)start, pos-start);
11269 * c-indentation-style: bsd
11270 * c-basic-offset: 4
11271 * indent-tabs-mode: t
11274 * ex: set ts=8 sts=4 sw=4 noet: