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]);
4244 /* Get the rest if it looks like a package qualifier */
4246 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
4248 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
4251 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
4252 *s == '\'' ? "'" : "::");
4257 if (PL_expect == XOPERATOR) {
4258 if (PL_bufptr == PL_linestart) {
4259 CopLINE_dec(PL_curcop);
4260 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
4261 CopLINE_inc(PL_curcop);
4264 no_op("Bareword",s);
4267 /* Look for a subroutine with this name in current package,
4268 unless name is "Foo::", in which case Foo is a bearword
4269 (and a package name). */
4272 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
4274 if (ckWARN(WARN_BAREWORD)
4275 && ! gv_fetchpv(PL_tokenbuf, 0, SVt_PVHV))
4276 Perl_warner(aTHX_ packWARN(WARN_BAREWORD),
4277 "Bareword \"%s\" refers to nonexistent package",
4280 PL_tokenbuf[len] = '\0';
4287 gv = gv_fetchpv(PL_tokenbuf, 0, SVt_PVCV);
4290 /* if we saw a global override before, get the right name */
4293 sv = newSVpvn("CORE::GLOBAL::",14);
4294 sv_catpv(sv,PL_tokenbuf);
4297 /* If len is 0, newSVpv does strlen(), which is correct.
4298 If len is non-zero, then it will be the true length,
4299 and so the scalar will be created correctly. */
4300 sv = newSVpv(PL_tokenbuf,len);
4303 /* Presume this is going to be a bareword of some sort. */
4306 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
4307 yylval.opval->op_private = OPpCONST_BARE;
4308 /* UTF-8 package name? */
4309 if (UTF && !IN_BYTES &&
4310 is_utf8_string((U8*)SvPVX_const(sv), SvCUR(sv)))
4313 /* And if "Foo::", then that's what it certainly is. */
4318 /* See if it's the indirect object for a list operator. */
4320 if (PL_oldoldbufptr &&
4321 PL_oldoldbufptr < PL_bufptr &&
4322 (PL_oldoldbufptr == PL_last_lop
4323 || PL_oldoldbufptr == PL_last_uni) &&
4324 /* NO SKIPSPACE BEFORE HERE! */
4325 (PL_expect == XREF ||
4326 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
4328 bool immediate_paren = *s == '(';
4330 /* (Now we can afford to cross potential line boundary.) */
4333 /* Two barewords in a row may indicate method call. */
4335 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
4338 /* If not a declared subroutine, it's an indirect object. */
4339 /* (But it's an indir obj regardless for sort.) */
4340 /* Also, if "_" follows a filetest operator, it's a bareword */
4343 ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
4344 ((!gv || !GvCVu(gv)) &&
4345 (PL_last_lop_op != OP_MAPSTART &&
4346 PL_last_lop_op != OP_GREPSTART))))
4347 || (PL_tokenbuf[0] == '_' && PL_tokenbuf[1] == '\0'
4348 && ((PL_opargs[PL_last_lop_op] & OA_CLASS_MASK) == OA_FILESTATOP))
4351 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
4356 PL_expect = XOPERATOR;
4359 /* Is this a word before a => operator? */
4360 if (*s == '=' && s[1] == '>' && !pkgname) {
4362 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
4363 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
4364 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
4368 /* If followed by a paren, it's certainly a subroutine. */
4371 if (gv && GvCVu(gv)) {
4372 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
4373 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
4378 PL_nextval[PL_nexttoke].opval = yylval.opval;
4379 PL_expect = XOPERATOR;
4385 /* If followed by var or block, call it a method (unless sub) */
4387 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
4388 PL_last_lop = PL_oldbufptr;
4389 PL_last_lop_op = OP_METHOD;
4393 /* If followed by a bareword, see if it looks like indir obj. */
4396 && (isIDFIRST_lazy_if(s,UTF) || *s == '$')
4397 && (tmp = intuit_method(s,gv)))
4400 /* Not a method, so call it a subroutine (if defined) */
4402 if (gv && GvCVu(gv)) {
4404 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
4405 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4406 "Ambiguous use of -%s resolved as -&%s()",
4407 PL_tokenbuf, PL_tokenbuf);
4408 /* Check for a constant sub */
4410 if ((sv = cv_const_sv(cv))) {
4412 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
4413 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
4414 yylval.opval->op_private = 0;
4418 /* Resolve to GV now. */
4419 op_free(yylval.opval);
4420 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
4421 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
4422 PL_last_lop = PL_oldbufptr;
4423 PL_last_lop_op = OP_ENTERSUB;
4424 /* Is there a prototype? */
4427 const char *proto = SvPV_const((SV*)cv, len);
4430 if (*proto == '$' && proto[1] == '\0')
4432 while (*proto == ';')
4434 if (*proto == '&' && *s == '{') {
4435 sv_setpv(PL_subname, PL_curstash ?
4436 "__ANON__" : "__ANON__::__ANON__");
4440 PL_nextval[PL_nexttoke].opval = yylval.opval;
4446 /* Call it a bare word */
4448 if (PL_hints & HINT_STRICT_SUBS)
4449 yylval.opval->op_private |= OPpCONST_STRICT;
4452 if (lastchar != '-') {
4453 if (ckWARN(WARN_RESERVED)) {
4454 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
4455 if (!*d && !gv_stashpv(PL_tokenbuf,FALSE))
4456 Perl_warner(aTHX_ packWARN(WARN_RESERVED), PL_warn_reserved,
4463 if ((lastchar == '*' || lastchar == '%' || lastchar == '&')
4464 && ckWARN_d(WARN_AMBIGUOUS)) {
4465 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4466 "Operator or semicolon missing before %c%s",
4467 lastchar, PL_tokenbuf);
4468 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4469 "Ambiguous use of %c resolved as operator %c",
4470 lastchar, lastchar);
4476 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4477 newSVpv(CopFILE(PL_curcop),0));
4481 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4482 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
4485 case KEY___PACKAGE__:
4486 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4488 ? newSVhek(HvNAME_HEK(PL_curstash))
4495 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
4496 const char *pname = "main";
4497 if (PL_tokenbuf[2] == 'D')
4498 pname = HvNAME_get(PL_curstash ? PL_curstash : PL_defstash);
4499 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), GV_ADD,
4503 GvIOp(gv) = newIO();
4504 IoIFP(GvIOp(gv)) = PL_rsfp;
4505 #if defined(HAS_FCNTL) && defined(F_SETFD)
4507 const int fd = PerlIO_fileno(PL_rsfp);
4508 fcntl(fd,F_SETFD,fd >= 3);
4511 /* Mark this internal pseudo-handle as clean */
4512 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
4514 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
4515 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
4516 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
4518 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
4519 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
4520 /* if the script was opened in binmode, we need to revert
4521 * it to text mode for compatibility; but only iff it has CRs
4522 * XXX this is a questionable hack at best. */
4523 if (PL_bufend-PL_bufptr > 2
4524 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
4527 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
4528 loc = PerlIO_tell(PL_rsfp);
4529 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4532 if (PerlLIO_setmode(PL_rsfp, O_TEXT) != -1) {
4534 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4535 #endif /* NETWARE */
4536 #ifdef PERLIO_IS_STDIO /* really? */
4537 # if defined(__BORLANDC__)
4538 /* XXX see note in do_binmode() */
4539 ((FILE*)PL_rsfp)->flags &= ~_F_BIN;
4543 PerlIO_seek(PL_rsfp, loc, 0);
4547 #ifdef PERLIO_LAYERS
4550 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
4551 else if (PL_encoding) {
4558 XPUSHs(PL_encoding);
4560 call_method("name", G_SCALAR);
4564 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL,
4565 Perl_form(aTHX_ ":encoding(%"SVf")",
4583 if (PL_expect == XSTATE) {
4590 if (*s == ':' && s[1] == ':') {
4593 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4594 if (!(tmp = keyword(PL_tokenbuf, len)))
4595 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4598 else if (tmp == KEY_require || tmp == KEY_do)
4599 /* that's a way to remember we saw "CORE::" */
4612 LOP(OP_ACCEPT,XTERM);
4618 LOP(OP_ATAN2,XTERM);
4624 LOP(OP_BINMODE,XTERM);
4627 LOP(OP_BLESS,XTERM);
4636 /* When 'use switch' is in effect, continue has a dual
4637 life as a control operator. */
4639 if (!FEATURE_IS_ENABLED("switch", 6))
4642 /* We have to disambiguate the two senses of
4643 "continue". If the next token is a '{' then
4644 treat it as the start of a continue block;
4645 otherwise treat it as a control operator.
4656 (void)gv_fetchpv("ENV", GV_ADD, SVt_PVHV); /* may use HOME */
4673 if (!PL_cryptseen) {
4674 PL_cryptseen = TRUE;
4678 LOP(OP_CRYPT,XTERM);
4681 LOP(OP_CHMOD,XTERM);
4684 LOP(OP_CHOWN,XTERM);
4687 LOP(OP_CONNECT,XTERM);
4706 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4707 if (orig_keyword == KEY_do) {
4716 PL_hints |= HINT_BLOCK_SCOPE;
4726 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4727 LOP(OP_DBMOPEN,XTERM);
4733 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4740 yylval.ival = CopLINE(PL_curcop);
4754 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4755 UNIBRACK(OP_ENTEREVAL);
4773 case KEY_endhostent:
4779 case KEY_endservent:
4782 case KEY_endprotoent:
4793 yylval.ival = CopLINE(PL_curcop);
4795 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4797 if ((PL_bufend - p) >= 3 &&
4798 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4800 else if ((PL_bufend - p) >= 4 &&
4801 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4804 if (isIDFIRST_lazy_if(p,UTF)) {
4805 p = scan_ident(p, PL_bufend,
4806 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4810 Perl_croak(aTHX_ "Missing $ on loop variable");
4815 LOP(OP_FORMLINE,XTERM);
4821 LOP(OP_FCNTL,XTERM);
4827 LOP(OP_FLOCK,XTERM);
4836 LOP(OP_GREPSTART, XREF);
4839 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4854 case KEY_getpriority:
4855 LOP(OP_GETPRIORITY,XTERM);
4857 case KEY_getprotobyname:
4860 case KEY_getprotobynumber:
4861 LOP(OP_GPBYNUMBER,XTERM);
4863 case KEY_getprotoent:
4875 case KEY_getpeername:
4876 UNI(OP_GETPEERNAME);
4878 case KEY_gethostbyname:
4881 case KEY_gethostbyaddr:
4882 LOP(OP_GHBYADDR,XTERM);
4884 case KEY_gethostent:
4887 case KEY_getnetbyname:
4890 case KEY_getnetbyaddr:
4891 LOP(OP_GNBYADDR,XTERM);
4896 case KEY_getservbyname:
4897 LOP(OP_GSBYNAME,XTERM);
4899 case KEY_getservbyport:
4900 LOP(OP_GSBYPORT,XTERM);
4902 case KEY_getservent:
4905 case KEY_getsockname:
4906 UNI(OP_GETSOCKNAME);
4908 case KEY_getsockopt:
4909 LOP(OP_GSOCKOPT,XTERM);
4924 yylval.ival = CopLINE(PL_curcop);
4935 yylval.ival = CopLINE(PL_curcop);
4939 LOP(OP_INDEX,XTERM);
4945 LOP(OP_IOCTL,XTERM);
4957 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4989 LOP(OP_LISTEN,XTERM);
4998 s = scan_pat(s,OP_MATCH);
4999 TERM(sublex_start());
5002 LOP(OP_MAPSTART, XREF);
5005 LOP(OP_MKDIR,XTERM);
5008 LOP(OP_MSGCTL,XTERM);
5011 LOP(OP_MSGGET,XTERM);
5014 LOP(OP_MSGRCV,XTERM);
5017 LOP(OP_MSGSND,XTERM);
5023 if (isIDFIRST_lazy_if(s,UTF)) {
5024 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
5025 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
5027 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
5028 if (!PL_in_my_stash) {
5031 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
5039 s = force_word(s,WORD,TRUE,FALSE,FALSE);
5046 s = tokenize_use(0, s);
5050 if (*s == '(' || (s = skipspace(s), *s == '('))
5057 if (isIDFIRST_lazy_if(s,UTF)) {
5059 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
5060 for (t=d; *t && isSPACE(*t); t++) ;
5061 if ( *t && strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE)
5063 && !(t[0] == '=' && t[1] == '>')
5065 int len = (int)(d-s);
5066 Perl_warner(aTHX_ packWARN(WARN_PRECEDENCE),
5067 "Precedence problem: open %.*s should be open(%.*s)",
5074 yylval.ival = OP_OR;
5084 LOP(OP_OPEN_DIR,XTERM);
5087 checkcomma(s,PL_tokenbuf,"filehandle");
5091 checkcomma(s,PL_tokenbuf,"filehandle");
5110 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5114 LOP(OP_PIPE_OP,XTERM);
5117 s = scan_str(s,FALSE,FALSE);
5119 missingterm((char*)0);
5120 yylval.ival = OP_CONST;
5121 TERM(sublex_start());
5127 s = scan_str(s,FALSE,FALSE);
5129 missingterm((char*)0);
5130 PL_expect = XOPERATOR;
5132 if (SvCUR(PL_lex_stuff)) {
5135 d = SvPV_force(PL_lex_stuff, len);
5138 for (; isSPACE(*d) && len; --len, ++d) ;
5141 if (!warned && ckWARN(WARN_QW)) {
5142 for (; !isSPACE(*d) && len; --len, ++d) {
5144 Perl_warner(aTHX_ packWARN(WARN_QW),
5145 "Possible attempt to separate words with commas");
5148 else if (*d == '#') {
5149 Perl_warner(aTHX_ packWARN(WARN_QW),
5150 "Possible attempt to put comments in qw() list");
5156 for (; !isSPACE(*d) && len; --len, ++d) ;
5158 sv = newSVpvn(b, d-b);
5159 if (DO_UTF8(PL_lex_stuff))
5161 words = append_elem(OP_LIST, words,
5162 newSVOP(OP_CONST, 0, tokeq(sv)));
5166 PL_nextval[PL_nexttoke].opval = words;
5171 SvREFCNT_dec(PL_lex_stuff);
5172 PL_lex_stuff = Nullsv;
5178 s = scan_str(s,FALSE,FALSE);
5180 missingterm((char*)0);
5181 yylval.ival = OP_STRINGIFY;
5182 if (SvIVX(PL_lex_stuff) == '\'')
5183 SvIV_set(PL_lex_stuff, 0); /* qq'$foo' should intepolate */
5184 TERM(sublex_start());
5187 s = scan_pat(s,OP_QR);
5188 TERM(sublex_start());
5191 s = scan_str(s,FALSE,FALSE);
5193 missingterm((char*)0);
5194 yylval.ival = OP_BACKTICK;
5196 TERM(sublex_start());
5204 s = force_version(s, FALSE);
5206 else if (*s != 'v' || !isDIGIT(s[1])
5207 || (s = force_version(s, TRUE), *s == 'v'))
5209 *PL_tokenbuf = '\0';
5210 s = force_word(s,WORD,TRUE,TRUE,FALSE);
5211 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
5212 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
5214 yyerror("<> should be quotes");
5216 if (orig_keyword == KEY_require) {
5224 PL_last_uni = PL_oldbufptr;
5225 PL_last_lop_op = OP_REQUIRE;
5227 return REPORT( (int)REQUIRE );
5233 s = force_word(s,WORD,TRUE,FALSE,FALSE);
5237 LOP(OP_RENAME,XTERM);
5246 LOP(OP_RINDEX,XTERM);
5256 UNIDOR(OP_READLINE);
5269 LOP(OP_REVERSE,XTERM);
5272 UNIDOR(OP_READLINK);
5280 TERM(sublex_start());
5282 TOKEN(1); /* force error */
5285 checkcomma(s,PL_tokenbuf,"filehandle");
5295 LOP(OP_SELECT,XTERM);
5301 LOP(OP_SEMCTL,XTERM);
5304 LOP(OP_SEMGET,XTERM);
5307 LOP(OP_SEMOP,XTERM);
5313 LOP(OP_SETPGRP,XTERM);
5315 case KEY_setpriority:
5316 LOP(OP_SETPRIORITY,XTERM);
5318 case KEY_sethostent:
5324 case KEY_setservent:
5327 case KEY_setprotoent:
5337 LOP(OP_SEEKDIR,XTERM);
5339 case KEY_setsockopt:
5340 LOP(OP_SSOCKOPT,XTERM);
5346 LOP(OP_SHMCTL,XTERM);
5349 LOP(OP_SHMGET,XTERM);
5352 LOP(OP_SHMREAD,XTERM);
5355 LOP(OP_SHMWRITE,XTERM);
5358 LOP(OP_SHUTDOWN,XTERM);
5367 LOP(OP_SOCKET,XTERM);
5369 case KEY_socketpair:
5370 LOP(OP_SOCKPAIR,XTERM);
5373 checkcomma(s,PL_tokenbuf,"subroutine name");
5375 if (*s == ';' || *s == ')') /* probably a close */
5376 Perl_croak(aTHX_ "sort is now a reserved word");
5378 s = force_word(s,WORD,TRUE,TRUE,FALSE);
5382 LOP(OP_SPLIT,XTERM);
5385 LOP(OP_SPRINTF,XTERM);
5388 LOP(OP_SPLICE,XTERM);
5403 LOP(OP_SUBSTR,XTERM);
5409 char tmpbuf[sizeof PL_tokenbuf];
5410 SSize_t tboffset = 0;
5411 expectation attrful;
5412 bool have_name, have_proto, bad_proto;
5413 const int key = tmp;
5417 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
5418 (*s == ':' && s[1] == ':'))
5421 attrful = XATTRBLOCK;
5422 /* remember buffer pos'n for later force_word */
5423 tboffset = s - PL_oldbufptr;
5424 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
5425 if (strchr(tmpbuf, ':'))
5426 sv_setpv(PL_subname, tmpbuf);
5428 sv_setsv(PL_subname,PL_curstname);
5429 sv_catpvn(PL_subname,"::",2);
5430 sv_catpvn(PL_subname,tmpbuf,len);
5437 Perl_croak(aTHX_ "Missing name in \"my sub\"");
5438 PL_expect = XTERMBLOCK;
5439 attrful = XATTRTERM;
5440 sv_setpvn(PL_subname,"?",1);
5444 if (key == KEY_format) {
5446 PL_lex_formbrack = PL_lex_brackets + 1;
5448 (void) force_word(PL_oldbufptr + tboffset, WORD,
5453 /* Look for a prototype */
5457 s = scan_str(s,FALSE,FALSE);
5459 Perl_croak(aTHX_ "Prototype not terminated");
5460 /* strip spaces and check for bad characters */
5461 d = SvPVX(PL_lex_stuff);
5464 for (p = d; *p; ++p) {
5467 if (!strchr("$@%*;[]&\\", *p))
5472 if (bad_proto && ckWARN(WARN_SYNTAX))
5473 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5474 "Illegal character in prototype for %"SVf" : %s",
5476 SvCUR_set(PL_lex_stuff, tmp);
5484 if (*s == ':' && s[1] != ':')
5485 PL_expect = attrful;
5486 else if (*s != '{' && key == KEY_sub) {
5488 Perl_croak(aTHX_ "Illegal declaration of anonymous subroutine");
5490 Perl_croak(aTHX_ "Illegal declaration of subroutine %"SVf, PL_subname);
5494 PL_nextval[PL_nexttoke].opval =
5495 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
5496 PL_lex_stuff = Nullsv;
5500 sv_setpv(PL_subname,
5501 PL_curstash ? "__ANON__" : "__ANON__::__ANON__");
5504 (void) force_word(PL_oldbufptr + tboffset, WORD,
5513 LOP(OP_SYSTEM,XREF);
5516 LOP(OP_SYMLINK,XTERM);
5519 LOP(OP_SYSCALL,XTERM);
5522 LOP(OP_SYSOPEN,XTERM);
5525 LOP(OP_SYSSEEK,XTERM);
5528 LOP(OP_SYSREAD,XTERM);
5531 LOP(OP_SYSWRITE,XTERM);
5535 TERM(sublex_start());
5556 LOP(OP_TRUNCATE,XTERM);
5568 yylval.ival = CopLINE(PL_curcop);
5572 yylval.ival = CopLINE(PL_curcop);
5576 LOP(OP_UNLINK,XTERM);
5582 LOP(OP_UNPACK,XTERM);
5585 LOP(OP_UTIME,XTERM);
5591 LOP(OP_UNSHIFT,XTERM);
5594 s = tokenize_use(1, s);
5604 yylval.ival = CopLINE(PL_curcop);
5608 yylval.ival = CopLINE(PL_curcop);
5612 PL_hints |= HINT_BLOCK_SCOPE;
5619 LOP(OP_WAITPID,XTERM);
5628 ctl_l[0] = toCTRL('L');
5630 gv_fetchpv(ctl_l, GV_ADD, SVt_PV);
5633 gv_fetchpv("\f", GV_ADD, SVt_PV); /* Make sure $^L is defined */
5638 if (PL_expect == XOPERATOR)
5644 yylval.ival = OP_XOR;
5649 TERM(sublex_start());
5654 #pragma segment Main
5658 S_pending_ident(pTHX)
5661 register I32 tmp = 0;
5662 /* pit holds the identifier we read and pending_ident is reset */
5663 char pit = PL_pending_ident;
5664 PL_pending_ident = 0;
5666 DEBUG_T({ PerlIO_printf(Perl_debug_log,
5667 "### Pending identifier '%s'\n", PL_tokenbuf); });
5669 /* if we're in a my(), we can't allow dynamics here.
5670 $foo'bar has already been turned into $foo::bar, so
5671 just check for colons.
5673 if it's a legal name, the OP is a PADANY.
5676 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
5677 if (strchr(PL_tokenbuf,':'))
5678 yyerror(Perl_form(aTHX_ "No package name allowed for "
5679 "variable %s in \"our\"",
5681 tmp = allocmy(PL_tokenbuf);
5684 if (strchr(PL_tokenbuf,':'))
5685 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
5687 yylval.opval = newOP(OP_PADANY, 0);
5688 yylval.opval->op_targ = allocmy(PL_tokenbuf);
5694 build the ops for accesses to a my() variable.
5696 Deny my($a) or my($b) in a sort block, *if* $a or $b is
5697 then used in a comparison. This catches most, but not
5698 all cases. For instance, it catches
5699 sort { my($a); $a <=> $b }
5701 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
5702 (although why you'd do that is anyone's guess).
5705 if (!strchr(PL_tokenbuf,':')) {
5707 tmp = pad_findmy(PL_tokenbuf);
5708 if (tmp != NOT_IN_PAD) {
5709 /* might be an "our" variable" */
5710 if (PAD_COMPNAME_FLAGS(tmp) & SVpad_OUR) {
5711 /* build ops for a bareword */
5712 HV * const stash = PAD_COMPNAME_OURSTASH(tmp);
5713 HEK * const stashname = HvNAME_HEK(stash);
5714 SV * const sym = newSVhek(stashname);
5715 sv_catpvn(sym, "::", 2);
5716 sv_catpv(sym, PL_tokenbuf+1);
5717 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
5718 yylval.opval->op_private = OPpCONST_ENTERED;
5721 ? (GV_ADDMULTI | GV_ADDINEVAL)
5724 ((PL_tokenbuf[0] == '$') ? SVt_PV
5725 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5730 /* if it's a sort block and they're naming $a or $b */
5731 if (PL_last_lop_op == OP_SORT &&
5732 PL_tokenbuf[0] == '$' &&
5733 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
5736 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
5737 d < PL_bufend && *d != '\n';
5740 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
5741 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
5747 yylval.opval = newOP(OP_PADANY, 0);
5748 yylval.opval->op_targ = tmp;
5754 Whine if they've said @foo in a doublequoted string,
5755 and @foo isn't a variable we can find in the symbol
5758 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
5759 GV *gv = gv_fetchpv(PL_tokenbuf+1, 0, SVt_PVAV);
5760 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
5761 && ckWARN(WARN_AMBIGUOUS))
5763 /* Downgraded from fatal to warning 20000522 mjd */
5764 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5765 "Possible unintended interpolation of %s in string",
5770 /* build ops for a bareword */
5771 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
5772 yylval.opval->op_private = OPpCONST_ENTERED;
5776 ? (GV_ADDMULTI | GV_ADDINEVAL)
5777 /* If the identifier refers to a stash, don't autovivify it.
5778 * Change 24660 had the side effect of causing symbol table
5779 * hashes to always be defined, even if they were freshly
5780 * created and the only reference in the entire program was
5781 * the single statement with the defined %foo::bar:: test.
5782 * It appears that all code in the wild doing this actually
5783 * wants to know whether sub-packages have been loaded, so
5784 * by avoiding auto-vivifying symbol tables, we ensure that
5785 * defined %foo::bar:: continues to be false, and the existing
5786 * tests still give the expected answers, even though what
5787 * they're actually testing has now changed subtly.
5789 : !(*PL_tokenbuf == '%' && *(d = PL_tokenbuf + strlen(PL_tokenbuf) - 1) == ':' && d[-1] == ':'),
5790 ((PL_tokenbuf[0] == '$') ? SVt_PV
5791 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5797 * The following code was generated by perl_keyword.pl.
5801 Perl_keyword (pTHX_ const char *name, I32 len)
5805 case 1: /* 5 tokens of length 1 */
5837 case 2: /* 18 tokens of length 2 */
5983 case 3: /* 29 tokens of length 3 */
5987 if (name[1] == 'N' &&
6050 if (name[1] == 'i' &&
6090 if (name[1] == 'o' &&
6099 if (name[1] == 'e' &&
6108 if (name[1] == 'n' &&
6117 if (name[1] == 'o' &&
6126 if (name[1] == 'a' &&
6135 if (name[1] == 'o' &&
6197 if (name[1] == 'e' &&
6211 return (FEATURE_IS_ENABLED("say", 3) ? -KEY_say : 0);
6237 if (name[1] == 'i' &&
6246 if (name[1] == 's' &&
6255 if (name[1] == 'e' &&
6264 if (name[1] == 'o' &&
6276 case 4: /* 41 tokens of length 4 */
6280 if (name[1] == 'O' &&
6290 if (name[1] == 'N' &&
6300 if (name[1] == 'i' &&
6310 if (name[1] == 'h' &&
6320 if (name[1] == 'u' &&
6333 if (name[2] == 'c' &&
6342 if (name[2] == 's' &&
6351 if (name[2] == 'a' &&
6387 if (name[1] == 'o' &&
6400 if (name[2] == 't' &&
6409 if (name[2] == 'o' &&
6418 if (name[2] == 't' &&
6427 if (name[2] == 'e' &&
6440 if (name[1] == 'o' &&
6453 if (name[2] == 'y' &&
6462 if (name[2] == 'l' &&
6478 if (name[2] == 's' &&
6487 if (name[2] == 'n' &&
6496 if (name[2] == 'c' &&
6509 if (name[1] == 'e' &&
6519 if (name[1] == 'p' &&
6532 if (name[2] == 'c' &&
6541 if (name[2] == 'p' &&
6550 if (name[2] == 's' &&
6566 if (name[2] == 'n' &&
6636 if (name[2] == 'r' &&
6645 if (name[2] == 'r' &&
6654 if (name[2] == 'a' &&
6670 if (name[2] == 'l' &&
6732 if (name[2] == 'e' &&
6735 return (FEATURE_IS_ENABLED("switch", 6) ? KEY_when : 0);
6748 case 5: /* 38 tokens of length 5 */
6752 if (name[1] == 'E' &&
6763 if (name[1] == 'H' &&
6777 if (name[2] == 'a' &&
6787 if (name[2] == 'a' &&
6804 if (name[2] == 'e' &&
6814 if (name[2] == 'e' &&
6818 return (FEATURE_IS_ENABLED("switch", 6) ? -KEY_break : 0);
6834 if (name[3] == 'i' &&
6843 if (name[3] == 'o' &&
6879 if (name[2] == 'o' &&
6889 if (name[2] == 'y' &&
6903 if (name[1] == 'l' &&
6917 if (name[2] == 'n' &&
6927 if (name[2] == 'o' &&
6941 if (name[1] == 'i' &&
6946 return (FEATURE_IS_ENABLED("switch", 6) ? KEY_given : 0);
6955 if (name[2] == 'd' &&
6965 if (name[2] == 'c' &&
6982 if (name[2] == 'c' &&
6992 if (name[2] == 't' &&
7006 if (name[1] == 'k' &&
7017 if (name[1] == 'r' &&
7031 if (name[2] == 's' &&
7041 if (name[2] == 'd' &&
7058 if (name[2] == 'm' &&
7068 if (name[2] == 'i' &&
7078 if (name[2] == 'e' &&
7088 if (name[2] == 'l' &&
7098 if (name[2] == 'a' &&
7108 if (name[2] == 'u' &&
7122 if (name[1] == 'i' &&
7136 if (name[2] == 'a' &&
7149 if (name[3] == 'e' &&
7184 if (name[2] == 'i' &&
7201 if (name[2] == 'i' &&
7211 if (name[2] == 'i' &&
7228 case 6: /* 33 tokens of length 6 */
7232 if (name[1] == 'c' &&
7247 if (name[2] == 'l' &&
7258 if (name[2] == 'r' &&
7273 if (name[1] == 'e' &&
7288 if (name[2] == 's' &&
7293 if(ckWARN_d(WARN_SYNTAX))
7294 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "elseif should be elsif");
7300 if (name[2] == 'i' &&
7318 if (name[2] == 'l' &&
7329 if (name[2] == 'r' &&
7344 if (name[1] == 'm' &&
7359 if (name[2] == 'n' &&
7370 if (name[2] == 's' &&
7385 if (name[1] == 's' &&
7391 if (name[4] == 't' &&
7400 if (name[4] == 'e' &&
7409 if (name[4] == 'c' &&
7418 if (name[4] == 'n' &&
7434 if (name[1] == 'r' &&
7452 if (name[3] == 'a' &&
7462 if (name[3] == 'u' &&
7476 if (name[2] == 'n' &&
7494 if (name[2] == 'a' &&
7508 if (name[3] == 'e' &&
7521 if (name[4] == 't' &&
7530 if (name[4] == 'e' &&
7552 if (name[4] == 't' &&
7561 if (name[4] == 'e' &&
7577 if (name[2] == 'c' &&
7588 if (name[2] == 'l' &&
7599 if (name[2] == 'b' &&
7610 if (name[2] == 's' &&
7633 if (name[4] == 's' &&
7642 if (name[4] == 'n' &&
7655 if (name[3] == 'a' &&
7672 if (name[1] == 'a' &&
7687 case 7: /* 29 tokens of length 7 */
7691 if (name[1] == 'E' &&
7704 if (name[1] == '_' &&
7717 if (name[1] == 'i' &&
7724 return -KEY_binmode;
7730 if (name[1] == 'o' &&
7737 return -KEY_connect;
7746 if (name[2] == 'm' &&
7752 return -KEY_dbmopen;
7763 if (name[4] == 'u' &&
7767 return (FEATURE_IS_ENABLED("switch", 6) ? KEY_default : 0);
7773 if (name[4] == 'n' &&
7794 if (name[1] == 'o' &&
7807 if (name[1] == 'e' &&
7814 if (name[5] == 'r' &&
7817 return -KEY_getpgrp;
7823 if (name[5] == 'i' &&
7826 return -KEY_getppid;
7839 if (name[1] == 'c' &&
7846 return -KEY_lcfirst;
7852 if (name[1] == 'p' &&
7859 return -KEY_opendir;
7865 if (name[1] == 'a' &&
7883 if (name[3] == 'd' &&
7888 return -KEY_readdir;
7894 if (name[3] == 'u' &&
7905 if (name[3] == 'e' &&
7910 return -KEY_reverse;
7929 if (name[3] == 'k' &&
7934 return -KEY_seekdir;
7940 if (name[3] == 'p' &&
7945 return -KEY_setpgrp;
7955 if (name[2] == 'm' &&
7961 return -KEY_shmread;
7967 if (name[2] == 'r' &&
7973 return -KEY_sprintf;
7982 if (name[3] == 'l' &&
7987 return -KEY_symlink;
7996 if (name[4] == 'a' &&
8000 return -KEY_syscall;
8006 if (name[4] == 'p' &&
8010 return -KEY_sysopen;
8016 if (name[4] == 'e' &&
8020 return -KEY_sysread;
8026 if (name[4] == 'e' &&
8030 return -KEY_sysseek;
8048 if (name[1] == 'e' &&
8055 return -KEY_telldir;
8064 if (name[2] == 'f' &&
8070 return -KEY_ucfirst;
8076 if (name[2] == 's' &&
8082 return -KEY_unshift;
8092 if (name[1] == 'a' &&
8099 return -KEY_waitpid;
8108 case 8: /* 26 tokens of length 8 */
8112 if (name[1] == 'U' &&
8120 return KEY_AUTOLOAD;
8131 if (name[3] == 'A' &&
8137 return KEY___DATA__;
8143 if (name[3] == 'I' &&
8149 return -KEY___FILE__;
8155 if (name[3] == 'I' &&
8161 return -KEY___LINE__;
8177 if (name[2] == 'o' &&
8184 return -KEY_closedir;
8190 if (name[2] == 'n' &&
8197 return -KEY_continue;
8207 if (name[1] == 'b' &&
8215 return -KEY_dbmclose;
8221 if (name[1] == 'n' &&
8227 if (name[4] == 'r' &&
8232 return -KEY_endgrent;
8238 if (name[4] == 'w' &&
8243 return -KEY_endpwent;
8256 if (name[1] == 'o' &&
8264 return -KEY_formline;
8270 if (name[1] == 'e' &&
8281 if (name[6] == 'n' &&
8284 return -KEY_getgrent;
8290 if (name[6] == 'i' &&
8293 return -KEY_getgrgid;
8299 if (name[6] == 'a' &&
8302 return -KEY_getgrnam;
8315 if (name[4] == 'o' &&
8320 return -KEY_getlogin;
8331 if (name[6] == 'n' &&
8334 return -KEY_getpwent;
8340 if (name[6] == 'a' &&
8343 return -KEY_getpwnam;
8349 if (name[6] == 'i' &&
8352 return -KEY_getpwuid;
8372 if (name[1] == 'e' &&
8379 if (name[5] == 'i' &&
8386 return -KEY_readline;
8391 return -KEY_readlink;
8402 if (name[5] == 'i' &&
8406 return -KEY_readpipe;
8427 if (name[4] == 'r' &&
8432 return -KEY_setgrent;
8438 if (name[4] == 'w' &&
8443 return -KEY_setpwent;
8459 if (name[3] == 'w' &&
8465 return -KEY_shmwrite;
8471 if (name[3] == 't' &&
8477 return -KEY_shutdown;
8487 if (name[2] == 's' &&
8494 return -KEY_syswrite;
8504 if (name[1] == 'r' &&
8512 return -KEY_truncate;
8521 case 9: /* 8 tokens of length 9 */
8525 if (name[1] == 'n' &&
8534 return -KEY_endnetent;
8540 if (name[1] == 'e' &&
8549 return -KEY_getnetent;
8555 if (name[1] == 'o' &&
8564 return -KEY_localtime;
8570 if (name[1] == 'r' &&
8579 return KEY_prototype;
8585 if (name[1] == 'u' &&
8594 return -KEY_quotemeta;
8600 if (name[1] == 'e' &&
8609 return -KEY_rewinddir;
8615 if (name[1] == 'e' &&
8624 return -KEY_setnetent;
8630 if (name[1] == 'a' &&
8639 return -KEY_wantarray;
8648 case 10: /* 9 tokens of length 10 */
8652 if (name[1] == 'n' &&
8658 if (name[4] == 'o' &&
8665 return -KEY_endhostent;
8671 if (name[4] == 'e' &&
8678 return -KEY_endservent;
8691 if (name[1] == 'e' &&
8697 if (name[4] == 'o' &&
8704 return -KEY_gethostent;
8713 if (name[5] == 'r' &&
8719 return -KEY_getservent;
8725 if (name[5] == 'c' &&
8731 return -KEY_getsockopt;
8756 if (name[4] == 'o' &&
8763 return -KEY_sethostent;
8772 if (name[5] == 'r' &&
8778 return -KEY_setservent;
8784 if (name[5] == 'c' &&
8790 return -KEY_setsockopt;
8807 if (name[2] == 'c' &&
8816 return -KEY_socketpair;
8829 case 11: /* 8 tokens of length 11 */
8833 if (name[1] == '_' &&
8844 return -KEY___PACKAGE__;
8850 if (name[1] == 'n' &&
8861 return -KEY_endprotoent;
8867 if (name[1] == 'e' &&
8876 if (name[5] == 'e' &&
8883 return -KEY_getpeername;
8892 if (name[6] == 'o' &&
8898 return -KEY_getpriority;
8904 if (name[6] == 't' &&
8910 return -KEY_getprotoent;
8924 if (name[4] == 'o' &&
8932 return -KEY_getsockname;
8945 if (name[1] == 'e' &&
8953 if (name[6] == 'o' &&
8959 return -KEY_setpriority;
8965 if (name[6] == 't' &&
8971 return -KEY_setprotoent;
8987 case 12: /* 2 tokens of length 12 */
8988 if (name[0] == 'g' &&
9000 if (name[9] == 'd' &&
9003 { /* getnetbyaddr */
9004 return -KEY_getnetbyaddr;
9010 if (name[9] == 'a' &&
9013 { /* getnetbyname */
9014 return -KEY_getnetbyname;
9026 case 13: /* 4 tokens of length 13 */
9027 if (name[0] == 'g' &&
9034 if (name[4] == 'o' &&
9043 if (name[10] == 'd' &&
9046 { /* gethostbyaddr */
9047 return -KEY_gethostbyaddr;
9053 if (name[10] == 'a' &&
9056 { /* gethostbyname */
9057 return -KEY_gethostbyname;
9070 if (name[4] == 'e' &&
9079 if (name[10] == 'a' &&
9082 { /* getservbyname */
9083 return -KEY_getservbyname;
9089 if (name[10] == 'o' &&
9092 { /* getservbyport */
9093 return -KEY_getservbyport;
9112 case 14: /* 1 tokens of length 14 */
9113 if (name[0] == 'g' &&
9127 { /* getprotobyname */
9128 return -KEY_getprotobyname;
9133 case 16: /* 1 tokens of length 16 */
9134 if (name[0] == 'g' &&
9150 { /* getprotobynumber */
9151 return -KEY_getprotobynumber;
9165 S_checkcomma(pTHX_ register char *s, const char *name, const char *what)
9169 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
9170 if (ckWARN(WARN_SYNTAX)) {
9172 for (w = s+2; *w && level; w++) {
9179 for (; *w && isSPACE(*w); w++) ;
9180 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
9181 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
9182 "%s (...) interpreted as function",name);
9185 while (s < PL_bufend && isSPACE(*s))
9189 while (s < PL_bufend && isSPACE(*s))
9191 if (isIDFIRST_lazy_if(s,UTF)) {
9193 while (isALNUM_lazy_if(s,UTF))
9195 while (s < PL_bufend && isSPACE(*s))
9199 *s = '\0'; /* XXX If we didn't do this, we could const a lot of toke.c */
9200 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
9204 Perl_croak(aTHX_ "No comma allowed after %s", what);
9209 /* Either returns sv, or mortalizes sv and returns a new SV*.
9210 Best used as sv=new_constant(..., sv, ...).
9211 If s, pv are NULL, calls subroutine with one argument,
9212 and type is used with error messages only. */
9215 S_new_constant(pTHX_ const char *s, STRLEN len, const char *key, SV *sv, SV *pv,
9219 HV * const table = GvHV(PL_hintgv); /* ^H */
9223 const char *why1 = "", *why2 = "", *why3 = "";
9225 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
9228 why2 = strEQ(key,"charnames")
9229 ? "(possibly a missing \"use charnames ...\")"
9231 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
9232 (type ? type: "undef"), why2);
9234 /* This is convoluted and evil ("goto considered harmful")
9235 * but I do not understand the intricacies of all the different
9236 * failure modes of %^H in here. The goal here is to make
9237 * the most probable error message user-friendly. --jhi */
9242 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
9243 (type ? type: "undef"), why1, why2, why3);
9245 yyerror(SvPVX_const(msg));
9249 cvp = hv_fetch(table, key, strlen(key), FALSE);
9250 if (!cvp || !SvOK(*cvp)) {
9253 why3 = "} is not defined";
9256 sv_2mortal(sv); /* Parent created it permanently */
9259 pv = sv_2mortal(newSVpvn(s, len));
9261 typesv = sv_2mortal(newSVpv(type, 0));
9263 typesv = &PL_sv_undef;
9265 PUSHSTACKi(PERLSI_OVERLOAD);
9277 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
9281 /* Check the eval first */
9282 if (!PL_in_eval && SvTRUE(ERRSV)) {
9283 sv_catpv(ERRSV, "Propagated");
9284 yyerror(SvPV_nolen_const(ERRSV)); /* Duplicates the message inside eval */
9286 res = SvREFCNT_inc(sv);
9290 (void)SvREFCNT_inc(res);
9299 why1 = "Call to &{$^H{";
9301 why3 = "}} did not return a defined value";
9309 /* Returns a NUL terminated string, with the length of the string written to
9313 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
9315 register char *d = dest;
9316 register char * const e = d + destlen - 3; /* two-character token, ending NUL */
9319 Perl_croak(aTHX_ ident_too_long);
9320 if (isALNUM(*s)) /* UTF handled below */
9322 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
9327 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
9331 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
9332 char *t = s + UTF8SKIP(s);
9333 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
9335 if (d + (t - s) > e)
9336 Perl_croak(aTHX_ ident_too_long);
9337 Copy(s, d, t - s, char);
9350 S_scan_ident(pTHX_ register char *s, register const char *send, char *dest, STRLEN destlen, I32 ck_uni)
9354 char *bracket = Nullch;
9360 e = d + destlen - 3; /* two-character token, ending NUL */
9362 while (isDIGIT(*s)) {
9364 Perl_croak(aTHX_ ident_too_long);
9371 Perl_croak(aTHX_ ident_too_long);
9372 if (isALNUM(*s)) /* UTF handled below */
9374 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
9379 else if (*s == ':' && s[1] == ':') {
9383 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
9384 char *t = s + UTF8SKIP(s);
9385 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
9387 if (d + (t - s) > e)
9388 Perl_croak(aTHX_ ident_too_long);
9389 Copy(s, d, t - s, char);
9400 if (PL_lex_state != LEX_NORMAL)
9401 PL_lex_state = LEX_INTERPENDMAYBE;
9404 if (*s == '$' && s[1] &&
9405 (isALNUM_lazy_if(s+1,UTF) || s[1] == '$' || s[1] == '{' || strnEQ(s+1,"::",2)) )
9418 if (*d == '^' && *s && isCONTROLVAR(*s)) {
9423 if (isSPACE(s[-1])) {
9425 const char ch = *s++;
9426 if (!SPACE_OR_TAB(ch)) {
9432 if (isIDFIRST_lazy_if(d,UTF)) {
9436 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
9438 while (e < send && UTF8_IS_CONTINUED(*e) && is_utf8_mark((U8*)e))
9441 Copy(s, d, e - s, char);
9446 while ((isALNUM(*s) || *s == ':') && d < e)
9449 Perl_croak(aTHX_ ident_too_long);
9452 while (s < send && SPACE_OR_TAB(*s)) s++;
9453 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
9454 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
9455 const char *brack = *s == '[' ? "[...]" : "{...}";
9456 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
9457 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
9458 funny, dest, brack, funny, dest, brack);
9461 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
9465 /* Handle extended ${^Foo} variables
9466 * 1999-02-27 mjd-perl-patch@plover.com */
9467 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
9471 while (isALNUM(*s) && d < e) {
9475 Perl_croak(aTHX_ ident_too_long);
9480 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets) {
9481 PL_lex_state = LEX_INTERPEND;
9486 if (PL_lex_state == LEX_NORMAL) {
9487 if (ckWARN(WARN_AMBIGUOUS) &&
9488 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
9490 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
9491 "Ambiguous use of %c{%s} resolved to %c%s",
9492 funny, dest, funny, dest);
9497 s = bracket; /* let the parser handle it */
9501 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
9502 PL_lex_state = LEX_INTERPEND;
9507 Perl_pmflag(pTHX_ U32* pmfl, int ch)
9512 *pmfl |= PMf_GLOBAL;
9514 *pmfl |= PMf_CONTINUE;
9518 *pmfl |= PMf_MULTILINE;
9520 *pmfl |= PMf_SINGLELINE;
9522 *pmfl |= PMf_EXTENDED;
9526 S_scan_pat(pTHX_ char *start, I32 type)
9529 char *s = scan_str(start,FALSE,FALSE);
9532 char * const delimiter = skipspace(start);
9533 Perl_croak(aTHX_ *delimiter == '?'
9534 ? "Search pattern not terminated or ternary operator parsed as search pattern"
9535 : "Search pattern not terminated" );
9538 pm = (PMOP*)newPMOP(type, 0);
9539 if (PL_multi_open == '?')
9540 pm->op_pmflags |= PMf_ONCE;
9542 while (*s && strchr("iomsx", *s))
9543 pmflag(&pm->op_pmflags,*s++);
9546 while (*s && strchr("iogcmsx", *s))
9547 pmflag(&pm->op_pmflags,*s++);
9549 /* issue a warning if /c is specified,but /g is not */
9550 if ((pm->op_pmflags & PMf_CONTINUE) && !(pm->op_pmflags & PMf_GLOBAL)
9551 && ckWARN(WARN_REGEXP))
9553 Perl_warner(aTHX_ packWARN(WARN_REGEXP), "Use of /c modifier is meaningless without /g" );
9556 pm->op_pmpermflags = pm->op_pmflags;
9558 PL_lex_op = (OP*)pm;
9559 yylval.ival = OP_MATCH;
9564 S_scan_subst(pTHX_ char *start)
9572 yylval.ival = OP_NULL;
9574 s = scan_str(start,FALSE,FALSE);
9577 Perl_croak(aTHX_ "Substitution pattern not terminated");
9579 if (s[-1] == PL_multi_open)
9582 first_start = PL_multi_start;
9583 s = scan_str(s,FALSE,FALSE);
9586 SvREFCNT_dec(PL_lex_stuff);
9587 PL_lex_stuff = Nullsv;
9589 Perl_croak(aTHX_ "Substitution replacement not terminated");
9591 PL_multi_start = first_start; /* so whole substitution is taken together */
9593 pm = (PMOP*)newPMOP(OP_SUBST, 0);
9599 else if (strchr("iogcmsx", *s))
9600 pmflag(&pm->op_pmflags,*s++);
9605 if ((pm->op_pmflags & PMf_CONTINUE) && ckWARN(WARN_REGEXP)) {
9606 Perl_warner(aTHX_ packWARN(WARN_REGEXP), "Use of /c modifier is meaningless in s///" );
9611 PL_sublex_info.super_bufptr = s;
9612 PL_sublex_info.super_bufend = PL_bufend;
9614 pm->op_pmflags |= PMf_EVAL;
9615 repl = newSVpvn("",0);
9617 sv_catpv(repl, es ? "eval " : "do ");
9618 sv_catpvn(repl, "{ ", 2);
9619 sv_catsv(repl, PL_lex_repl);
9620 sv_catpvn(repl, " };", 2);
9622 SvREFCNT_dec(PL_lex_repl);
9626 pm->op_pmpermflags = pm->op_pmflags;
9627 PL_lex_op = (OP*)pm;
9628 yylval.ival = OP_SUBST;
9633 S_scan_trans(pTHX_ char *start)
9642 yylval.ival = OP_NULL;
9644 s = scan_str(start,FALSE,FALSE);
9646 Perl_croak(aTHX_ "Transliteration pattern not terminated");
9647 if (s[-1] == PL_multi_open)
9650 s = scan_str(s,FALSE,FALSE);
9653 SvREFCNT_dec(PL_lex_stuff);
9654 PL_lex_stuff = Nullsv;
9656 Perl_croak(aTHX_ "Transliteration replacement not terminated");
9659 complement = del = squash = 0;
9663 complement = OPpTRANS_COMPLEMENT;
9666 del = OPpTRANS_DELETE;
9669 squash = OPpTRANS_SQUASH;
9678 Newx(tbl, complement&&!del?258:256, short);
9679 o = newPVOP(OP_TRANS, 0, (char*)tbl);
9680 o->op_private &= ~OPpTRANS_ALL;
9681 o->op_private |= del|squash|complement|
9682 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
9683 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
9686 yylval.ival = OP_TRANS;
9691 S_scan_heredoc(pTHX_ register char *s)
9694 I32 op_type = OP_SCALAR;
9698 const char newline[] = "\n";
9699 const char *found_newline;
9703 const int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
9707 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
9710 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
9711 if (*peek == '`' || *peek == '\'' || *peek =='"') {
9714 s = delimcpy(d, e, s, PL_bufend, term, &len);
9724 if (!isALNUM_lazy_if(s,UTF))
9725 deprecate_old("bare << to mean <<\"\"");
9726 for (; isALNUM_lazy_if(s,UTF); s++) {
9731 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
9732 Perl_croak(aTHX_ "Delimiter for here document is too long");
9735 len = d - PL_tokenbuf;
9736 #ifndef PERL_STRICT_CR
9737 d = strchr(s, '\r');
9739 char * const olds = s;
9741 while (s < PL_bufend) {
9747 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
9756 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
9760 if ( outer || !(found_newline = ninstr(s,PL_bufend,newline,newline+1)) ) {
9761 herewas = newSVpvn(s,PL_bufend-s);
9765 herewas = newSVpvn(s,found_newline-s);
9767 s += SvCUR(herewas);
9769 tmpstr = NEWSV(87,79);
9770 sv_upgrade(tmpstr, SVt_PVIV);
9773 SvIV_set(tmpstr, -1);
9775 else if (term == '`') {
9776 op_type = OP_BACKTICK;
9777 SvIV_set(tmpstr, '\\');
9781 PL_multi_start = CopLINE(PL_curcop);
9782 PL_multi_open = PL_multi_close = '<';
9783 term = *PL_tokenbuf;
9784 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
9785 char *bufptr = PL_sublex_info.super_bufptr;
9786 char *bufend = PL_sublex_info.super_bufend;
9787 char * const olds = s - SvCUR(herewas);
9788 s = strchr(bufptr, '\n');
9792 while (s < bufend &&
9793 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
9795 CopLINE_inc(PL_curcop);
9798 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
9799 missingterm(PL_tokenbuf);
9801 sv_setpvn(herewas,bufptr,d-bufptr+1);
9802 sv_setpvn(tmpstr,d+1,s-d);
9804 sv_catpvn(herewas,s,bufend-s);
9805 Copy(SvPVX_const(herewas),bufptr,SvCUR(herewas) + 1,char);
9812 while (s < PL_bufend &&
9813 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
9815 CopLINE_inc(PL_curcop);
9817 if (s >= PL_bufend) {
9818 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
9819 missingterm(PL_tokenbuf);
9821 sv_setpvn(tmpstr,d+1,s-d);
9823 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
9825 sv_catpvn(herewas,s,PL_bufend-s);
9826 sv_setsv(PL_linestr,herewas);
9827 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
9828 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
9829 PL_last_lop = PL_last_uni = Nullch;
9832 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
9833 while (s >= PL_bufend) { /* multiple line string? */
9835 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
9836 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
9837 missingterm(PL_tokenbuf);
9839 CopLINE_inc(PL_curcop);
9840 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
9841 PL_last_lop = PL_last_uni = Nullch;
9842 #ifndef PERL_STRICT_CR
9843 if (PL_bufend - PL_linestart >= 2) {
9844 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
9845 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
9847 PL_bufend[-2] = '\n';
9849 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
9851 else if (PL_bufend[-1] == '\r')
9852 PL_bufend[-1] = '\n';
9854 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
9855 PL_bufend[-1] = '\n';
9857 if (PERLDB_LINE && PL_curstash != PL_debstash) {
9858 SV *sv = NEWSV(88,0);
9860 sv_upgrade(sv, SVt_PVMG);
9861 sv_setsv(sv,PL_linestr);
9864 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
9866 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
9867 STRLEN off = PL_bufend - 1 - SvPVX_const(PL_linestr);
9868 *(SvPVX(PL_linestr) + off ) = ' ';
9869 sv_catsv(PL_linestr,herewas);
9870 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
9871 s = SvPVX(PL_linestr) + off; /* In case PV of PL_linestr moved. */
9875 sv_catsv(tmpstr,PL_linestr);
9880 PL_multi_end = CopLINE(PL_curcop);
9881 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
9882 SvPV_shrink_to_cur(tmpstr);
9884 SvREFCNT_dec(herewas);
9886 if (UTF && is_utf8_string((U8*)SvPVX_const(tmpstr), SvCUR(tmpstr)))
9888 else if (PL_encoding)
9889 sv_recode_to_utf8(tmpstr, PL_encoding);
9891 PL_lex_stuff = tmpstr;
9892 yylval.ival = op_type;
9897 takes: current position in input buffer
9898 returns: new position in input buffer
9899 side-effects: yylval and lex_op are set.
9904 <FH> read from filehandle
9905 <pkg::FH> read from package qualified filehandle
9906 <pkg'FH> read from package qualified filehandle
9907 <$fh> read from filehandle in $fh
9913 S_scan_inputsymbol(pTHX_ char *start)
9915 register char *s = start; /* current position in buffer */
9921 d = PL_tokenbuf; /* start of temp holding space */
9922 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
9923 end = strchr(s, '\n');
9926 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
9928 /* die if we didn't have space for the contents of the <>,
9929 or if it didn't end, or if we see a newline
9932 if (len >= sizeof PL_tokenbuf)
9933 Perl_croak(aTHX_ "Excessively long <> operator");
9935 Perl_croak(aTHX_ "Unterminated <> operator");
9940 Remember, only scalar variables are interpreted as filehandles by
9941 this code. Anything more complex (e.g., <$fh{$num}>) will be
9942 treated as a glob() call.
9943 This code makes use of the fact that except for the $ at the front,
9944 a scalar variable and a filehandle look the same.
9946 if (*d == '$' && d[1]) d++;
9948 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
9949 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
9952 /* If we've tried to read what we allow filehandles to look like, and
9953 there's still text left, then it must be a glob() and not a getline.
9954 Use scan_str to pull out the stuff between the <> and treat it
9955 as nothing more than a string.
9958 if (d - PL_tokenbuf != len) {
9959 yylval.ival = OP_GLOB;
9961 s = scan_str(start,FALSE,FALSE);
9963 Perl_croak(aTHX_ "Glob not terminated");
9967 bool readline_overriden = FALSE;
9968 GV *gv_readline = Nullgv;
9970 /* we're in a filehandle read situation */
9973 /* turn <> into <ARGV> */
9975 Copy("ARGV",d,5,char);
9977 /* Check whether readline() is overriden */
9978 if (((gv_readline = gv_fetchpv("readline", 0, SVt_PVCV))
9979 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline))
9981 ((gvp = (GV**)hv_fetch(PL_globalstash, "readline", 8, FALSE))
9982 && (gv_readline = *gvp) != (GV*)&PL_sv_undef
9983 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline)))
9984 readline_overriden = TRUE;
9986 /* if <$fh>, create the ops to turn the variable into a
9992 /* try to find it in the pad for this block, otherwise find
9993 add symbol table ops
9995 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
9996 if (PAD_COMPNAME_FLAGS(tmp) & SVpad_OUR) {
9997 HV *stash = PAD_COMPNAME_OURSTASH(tmp);
9998 HEK *stashname = HvNAME_HEK(stash);
9999 SV *sym = sv_2mortal(newSVhek(stashname));
10000 sv_catpvn(sym, "::", 2);
10001 sv_catpv(sym, d+1);
10006 OP *o = newOP(OP_PADSV, 0);
10008 PL_lex_op = readline_overriden
10009 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
10010 append_elem(OP_LIST, o,
10011 newCVREF(0, newGVOP(OP_GV,0,gv_readline))))
10012 : (OP*)newUNOP(OP_READLINE, 0, o);
10021 ? (GV_ADDMULTI | GV_ADDINEVAL)
10024 PL_lex_op = readline_overriden
10025 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
10026 append_elem(OP_LIST,
10027 newUNOP(OP_RV2SV, 0, newGVOP(OP_GV, 0, gv)),
10028 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
10029 : (OP*)newUNOP(OP_READLINE, 0,
10030 newUNOP(OP_RV2SV, 0,
10031 newGVOP(OP_GV, 0, gv)));
10033 if (!readline_overriden)
10034 PL_lex_op->op_flags |= OPf_SPECIAL;
10035 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
10036 yylval.ival = OP_NULL;
10039 /* If it's none of the above, it must be a literal filehandle
10040 (<Foo::BAR> or <FOO>) so build a simple readline OP */
10042 GV *gv = gv_fetchpv(d, GV_ADD, SVt_PVIO);
10043 PL_lex_op = readline_overriden
10044 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
10045 append_elem(OP_LIST,
10046 newGVOP(OP_GV, 0, gv),
10047 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
10048 : (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
10049 yylval.ival = OP_NULL;
10058 takes: start position in buffer
10059 keep_quoted preserve \ on the embedded delimiter(s)
10060 keep_delims preserve the delimiters around the string
10061 returns: position to continue reading from buffer
10062 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
10063 updates the read buffer.
10065 This subroutine pulls a string out of the input. It is called for:
10066 q single quotes q(literal text)
10067 ' single quotes 'literal text'
10068 qq double quotes qq(interpolate $here please)
10069 " double quotes "interpolate $here please"
10070 qx backticks qx(/bin/ls -l)
10071 ` backticks `/bin/ls -l`
10072 qw quote words @EXPORT_OK = qw( func() $spam )
10073 m// regexp match m/this/
10074 s/// regexp substitute s/this/that/
10075 tr/// string transliterate tr/this/that/
10076 y/// string transliterate y/this/that/
10077 ($*@) sub prototypes sub foo ($)
10078 (stuff) sub attr parameters sub foo : attr(stuff)
10079 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
10081 In most of these cases (all but <>, patterns and transliterate)
10082 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
10083 calls scan_str(). s/// makes yylex() call scan_subst() which calls
10084 scan_str(). tr/// and y/// make yylex() call scan_trans() which
10087 It skips whitespace before the string starts, and treats the first
10088 character as the delimiter. If the delimiter is one of ([{< then
10089 the corresponding "close" character )]}> is used as the closing
10090 delimiter. It allows quoting of delimiters, and if the string has
10091 balanced delimiters ([{<>}]) it allows nesting.
10093 On success, the SV with the resulting string is put into lex_stuff or,
10094 if that is already non-NULL, into lex_repl. The second case occurs only
10095 when parsing the RHS of the special constructs s/// and tr/// (y///).
10096 For convenience, the terminating delimiter character is stuffed into
10101 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
10103 SV *sv; /* scalar value: string */
10104 char *tmps; /* temp string, used for delimiter matching */
10105 register char *s = start; /* current position in the buffer */
10106 register char term; /* terminating character */
10107 register char *to; /* current position in the sv's data */
10108 I32 brackets = 1; /* bracket nesting level */
10109 bool has_utf8 = FALSE; /* is there any utf8 content? */
10110 I32 termcode; /* terminating char. code */
10111 U8 termstr[UTF8_MAXBYTES]; /* terminating string */
10112 STRLEN termlen; /* length of terminating string */
10113 char *last = NULL; /* last position for nesting bracket */
10115 /* skip space before the delimiter */
10119 /* mark where we are, in case we need to report errors */
10122 /* after skipping whitespace, the next character is the terminator */
10125 termcode = termstr[0] = term;
10129 termcode = utf8_to_uvchr((U8*)s, &termlen);
10130 Copy(s, termstr, termlen, U8);
10131 if (!UTF8_IS_INVARIANT(term))
10135 /* mark where we are */
10136 PL_multi_start = CopLINE(PL_curcop);
10137 PL_multi_open = term;
10139 /* find corresponding closing delimiter */
10140 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
10141 termcode = termstr[0] = term = tmps[5];
10143 PL_multi_close = term;
10145 /* create a new SV to hold the contents. 87 is leak category, I'm
10146 assuming. 79 is the SV's initial length. What a random number. */
10148 sv_upgrade(sv, SVt_PVIV);
10149 SvIV_set(sv, termcode);
10150 (void)SvPOK_only(sv); /* validate pointer */
10152 /* move past delimiter and try to read a complete string */
10154 sv_catpvn(sv, s, termlen);
10157 if (PL_encoding && !UTF) {
10161 int offset = s - SvPVX_const(PL_linestr);
10162 const bool found = sv_cat_decode(sv, PL_encoding, PL_linestr,
10163 &offset, (char*)termstr, termlen);
10164 const char *ns = SvPVX_const(PL_linestr) + offset;
10165 char *svlast = SvEND(sv) - 1;
10167 for (; s < ns; s++) {
10168 if (*s == '\n' && !PL_rsfp)
10169 CopLINE_inc(PL_curcop);
10172 goto read_more_line;
10174 /* handle quoted delimiters */
10175 if (SvCUR(sv) > 1 && *(svlast-1) == '\\') {
10177 for (t = svlast-2; t >= SvPVX_const(sv) && *t == '\\';)
10179 if ((svlast-1 - t) % 2) {
10180 if (!keep_quoted) {
10181 *(svlast-1) = term;
10183 SvCUR_set(sv, SvCUR(sv) - 1);
10188 if (PL_multi_open == PL_multi_close) {
10196 for (t = w = last; t < svlast; w++, t++) {
10197 /* At here, all closes are "was quoted" one,
10198 so we don't check PL_multi_close. */
10200 if (!keep_quoted && *(t+1) == PL_multi_open)
10205 else if (*t == PL_multi_open)
10213 SvCUR_set(sv, w - SvPVX_const(sv));
10216 if (--brackets <= 0)
10221 if (!keep_delims) {
10222 SvCUR_set(sv, SvCUR(sv) - 1);
10228 /* extend sv if need be */
10229 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
10230 /* set 'to' to the next character in the sv's string */
10231 to = SvPVX(sv)+SvCUR(sv);
10233 /* if open delimiter is the close delimiter read unbridle */
10234 if (PL_multi_open == PL_multi_close) {
10235 for (; s < PL_bufend; s++,to++) {
10236 /* embedded newlines increment the current line number */
10237 if (*s == '\n' && !PL_rsfp)
10238 CopLINE_inc(PL_curcop);
10239 /* handle quoted delimiters */
10240 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
10241 if (!keep_quoted && s[1] == term)
10243 /* any other quotes are simply copied straight through */
10247 /* terminate when run out of buffer (the for() condition), or
10248 have found the terminator */
10249 else if (*s == term) {
10252 if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen))
10255 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
10261 /* if the terminator isn't the same as the start character (e.g.,
10262 matched brackets), we have to allow more in the quoting, and
10263 be prepared for nested brackets.
10266 /* read until we run out of string, or we find the terminator */
10267 for (; s < PL_bufend; s++,to++) {
10268 /* embedded newlines increment the line count */
10269 if (*s == '\n' && !PL_rsfp)
10270 CopLINE_inc(PL_curcop);
10271 /* backslashes can escape the open or closing characters */
10272 if (*s == '\\' && s+1 < PL_bufend) {
10273 if (!keep_quoted &&
10274 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
10279 /* allow nested opens and closes */
10280 else if (*s == PL_multi_close && --brackets <= 0)
10282 else if (*s == PL_multi_open)
10284 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
10289 /* terminate the copied string and update the sv's end-of-string */
10291 SvCUR_set(sv, to - SvPVX_const(sv));
10294 * this next chunk reads more into the buffer if we're not done yet
10298 break; /* handle case where we are done yet :-) */
10300 #ifndef PERL_STRICT_CR
10301 if (to - SvPVX_const(sv) >= 2) {
10302 if ((to[-2] == '\r' && to[-1] == '\n') ||
10303 (to[-2] == '\n' && to[-1] == '\r'))
10307 SvCUR_set(sv, to - SvPVX_const(sv));
10309 else if (to[-1] == '\r')
10312 else if (to - SvPVX_const(sv) == 1 && to[-1] == '\r')
10317 /* if we're out of file, or a read fails, bail and reset the current
10318 line marker so we can report where the unterminated string began
10321 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
10323 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
10326 /* we read a line, so increment our line counter */
10327 CopLINE_inc(PL_curcop);
10329 /* update debugger info */
10330 if (PERLDB_LINE && PL_curstash != PL_debstash) {
10331 SV *sv = NEWSV(88,0);
10333 sv_upgrade(sv, SVt_PVMG);
10334 sv_setsv(sv,PL_linestr);
10335 (void)SvIOK_on(sv);
10337 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
10340 /* having changed the buffer, we must update PL_bufend */
10341 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
10342 PL_last_lop = PL_last_uni = Nullch;
10345 /* at this point, we have successfully read the delimited string */
10347 if (!PL_encoding || UTF) {
10349 sv_catpvn(sv, s, termlen);
10352 if (has_utf8 || PL_encoding)
10355 PL_multi_end = CopLINE(PL_curcop);
10357 /* if we allocated too much space, give some back */
10358 if (SvCUR(sv) + 5 < SvLEN(sv)) {
10359 SvLEN_set(sv, SvCUR(sv) + 1);
10360 SvPV_renew(sv, SvLEN(sv));
10363 /* decide whether this is the first or second quoted string we've read
10376 takes: pointer to position in buffer
10377 returns: pointer to new position in buffer
10378 side-effects: builds ops for the constant in yylval.op
10380 Read a number in any of the formats that Perl accepts:
10382 \d(_?\d)*(\.(\d(_?\d)*)?)?[Ee][\+\-]?(\d(_?\d)*) 12 12.34 12.
10383 \.\d(_?\d)*[Ee][\+\-]?(\d(_?\d)*) .34
10386 0x[0-9A-Fa-f](_?[0-9A-Fa-f])*
10388 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
10391 If it reads a number without a decimal point or an exponent, it will
10392 try converting the number to an integer and see if it can do so
10393 without loss of precision.
10397 Perl_scan_num(pTHX_ const char *start, YYSTYPE* lvalp)
10399 register const char *s = start; /* current position in buffer */
10400 register char *d; /* destination in temp buffer */
10401 register char *e; /* end of temp buffer */
10402 NV nv; /* number read, as a double */
10403 SV *sv = Nullsv; /* place to put the converted number */
10404 bool floatit; /* boolean: int or float? */
10405 const char *lastub = 0; /* position of last underbar */
10406 static char const number_too_long[] = "Number too long";
10408 /* We use the first character to decide what type of number this is */
10412 Perl_croak(aTHX_ "panic: scan_num");
10414 /* if it starts with a 0, it could be an octal number, a decimal in
10415 0.13 disguise, or a hexadecimal number, or a binary number. */
10419 u holds the "number so far"
10420 shift the power of 2 of the base
10421 (hex == 4, octal == 3, binary == 1)
10422 overflowed was the number more than we can hold?
10424 Shift is used when we add a digit. It also serves as an "are
10425 we in octal/hex/binary?" indicator to disallow hex characters
10426 when in octal mode.
10431 bool overflowed = FALSE;
10432 bool just_zero = TRUE; /* just plain 0 or binary number? */
10433 static const NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
10434 static const char* const bases[5] =
10435 { "", "binary", "", "octal", "hexadecimal" };
10436 static const char* const Bases[5] =
10437 { "", "Binary", "", "Octal", "Hexadecimal" };
10438 static const char* const maxima[5] =
10440 "0b11111111111111111111111111111111",
10444 const char *base, *Base, *max;
10446 /* check for hex */
10451 } else if (s[1] == 'b') {
10456 /* check for a decimal in disguise */
10457 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
10459 /* so it must be octal */
10466 if (ckWARN(WARN_SYNTAX))
10467 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10468 "Misplaced _ in number");
10472 base = bases[shift];
10473 Base = Bases[shift];
10474 max = maxima[shift];
10476 /* read the rest of the number */
10478 /* x is used in the overflow test,
10479 b is the digit we're adding on. */
10484 /* if we don't mention it, we're done */
10488 /* _ are ignored -- but warned about if consecutive */
10490 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
10491 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10492 "Misplaced _ in number");
10496 /* 8 and 9 are not octal */
10497 case '8': case '9':
10499 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
10503 case '2': case '3': case '4':
10504 case '5': case '6': case '7':
10506 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
10509 case '0': case '1':
10510 b = *s++ & 15; /* ASCII digit -> value of digit */
10514 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
10515 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
10516 /* make sure they said 0x */
10519 b = (*s++ & 7) + 9;
10521 /* Prepare to put the digit we have onto the end
10522 of the number so far. We check for overflows.
10528 x = u << shift; /* make room for the digit */
10530 if ((x >> shift) != u
10531 && !(PL_hints & HINT_NEW_BINARY)) {
10534 if (ckWARN_d(WARN_OVERFLOW))
10535 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
10536 "Integer overflow in %s number",
10539 u = x | b; /* add the digit to the end */
10542 n *= nvshift[shift];
10543 /* If an NV has not enough bits in its
10544 * mantissa to represent an UV this summing of
10545 * small low-order numbers is a waste of time
10546 * (because the NV cannot preserve the
10547 * low-order bits anyway): we could just
10548 * remember when did we overflow and in the
10549 * end just multiply n by the right
10557 /* if we get here, we had success: make a scalar value from
10562 /* final misplaced underbar check */
10563 if (s[-1] == '_') {
10564 if (ckWARN(WARN_SYNTAX))
10565 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
10570 if (n > 4294967295.0 && ckWARN(WARN_PORTABLE))
10571 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
10572 "%s number > %s non-portable",
10578 if (u > 0xffffffff && ckWARN(WARN_PORTABLE))
10579 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
10580 "%s number > %s non-portable",
10585 if (just_zero && (PL_hints & HINT_NEW_INTEGER))
10586 sv = new_constant(start, s - start, "integer",
10588 else if (PL_hints & HINT_NEW_BINARY)
10589 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
10594 handle decimal numbers.
10595 we're also sent here when we read a 0 as the first digit
10597 case '1': case '2': case '3': case '4': case '5':
10598 case '6': case '7': case '8': case '9': case '.':
10601 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
10604 /* read next group of digits and _ and copy into d */
10605 while (isDIGIT(*s) || *s == '_') {
10606 /* skip underscores, checking for misplaced ones
10610 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
10611 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10612 "Misplaced _ in number");
10616 /* check for end of fixed-length buffer */
10618 Perl_croak(aTHX_ number_too_long);
10619 /* if we're ok, copy the character */
10624 /* final misplaced underbar check */
10625 if (lastub && s == lastub + 1) {
10626 if (ckWARN(WARN_SYNTAX))
10627 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
10630 /* read a decimal portion if there is one. avoid
10631 3..5 being interpreted as the number 3. followed
10634 if (*s == '.' && s[1] != '.') {
10639 if (ckWARN(WARN_SYNTAX))
10640 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10641 "Misplaced _ in number");
10645 /* copy, ignoring underbars, until we run out of digits.
10647 for (; isDIGIT(*s) || *s == '_'; s++) {
10648 /* fixed length buffer check */
10650 Perl_croak(aTHX_ number_too_long);
10652 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
10653 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10654 "Misplaced _ in number");
10660 /* fractional part ending in underbar? */
10661 if (s[-1] == '_') {
10662 if (ckWARN(WARN_SYNTAX))
10663 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10664 "Misplaced _ in number");
10666 if (*s == '.' && isDIGIT(s[1])) {
10667 /* oops, it's really a v-string, but without the "v" */
10673 /* read exponent part, if present */
10674 if ((*s == 'e' || *s == 'E') && strchr("+-0123456789_", s[1])) {
10678 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
10679 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
10681 /* stray preinitial _ */
10683 if (ckWARN(WARN_SYNTAX))
10684 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10685 "Misplaced _ in number");
10689 /* allow positive or negative exponent */
10690 if (*s == '+' || *s == '-')
10693 /* stray initial _ */
10695 if (ckWARN(WARN_SYNTAX))
10696 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10697 "Misplaced _ in number");
10701 /* read digits of exponent */
10702 while (isDIGIT(*s) || *s == '_') {
10705 Perl_croak(aTHX_ number_too_long);
10709 if (((lastub && s == lastub + 1) ||
10710 (!isDIGIT(s[1]) && s[1] != '_'))
10711 && ckWARN(WARN_SYNTAX))
10712 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10713 "Misplaced _ in number");
10720 /* make an sv from the string */
10724 We try to do an integer conversion first if no characters
10725 indicating "float" have been found.
10730 int flags = grok_number (PL_tokenbuf, d - PL_tokenbuf, &uv);
10732 if (flags == IS_NUMBER_IN_UV) {
10734 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
10737 } else if (flags == (IS_NUMBER_IN_UV | IS_NUMBER_NEG)) {
10738 if (uv <= (UV) IV_MIN)
10739 sv_setiv(sv, -(IV)uv);
10746 /* terminate the string */
10748 nv = Atof(PL_tokenbuf);
10752 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
10753 (PL_hints & HINT_NEW_INTEGER) )
10754 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
10755 (floatit ? "float" : "integer"),
10759 /* if it starts with a v, it could be a v-string */
10762 sv = NEWSV(92,5); /* preallocate storage space */
10763 s = scan_vstring(s,sv);
10767 /* make the op for the constant and return */
10770 lvalp->opval = newSVOP(OP_CONST, 0, sv);
10772 lvalp->opval = Nullop;
10778 S_scan_formline(pTHX_ register char *s)
10780 register char *eol;
10782 SV *stuff = newSVpvn("",0);
10783 bool needargs = FALSE;
10784 bool eofmt = FALSE;
10786 while (!needargs) {
10788 #ifdef PERL_STRICT_CR
10789 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
10791 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
10793 if (*t == '\n' || t == PL_bufend) {
10798 if (PL_in_eval && !PL_rsfp) {
10799 eol = (char *) memchr(s,'\n',PL_bufend-s);
10804 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
10806 for (t = s; t < eol; t++) {
10807 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
10809 goto enough; /* ~~ must be first line in formline */
10811 if (*t == '@' || *t == '^')
10815 sv_catpvn(stuff, s, eol-s);
10816 #ifndef PERL_STRICT_CR
10817 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
10818 char *end = SvPVX(stuff) + SvCUR(stuff);
10821 SvCUR_set(stuff, SvCUR(stuff) - 1);
10830 s = filter_gets(PL_linestr, PL_rsfp, 0);
10831 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
10832 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
10833 PL_last_lop = PL_last_uni = Nullch;
10842 if (SvCUR(stuff)) {
10845 PL_lex_state = LEX_NORMAL;
10846 PL_nextval[PL_nexttoke].ival = 0;
10850 PL_lex_state = LEX_FORMLINE;
10852 if (UTF && is_utf8_string((U8*)SvPVX_const(stuff), SvCUR(stuff)))
10854 else if (PL_encoding)
10855 sv_recode_to_utf8(stuff, PL_encoding);
10857 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
10859 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
10863 SvREFCNT_dec(stuff);
10865 PL_lex_formbrack = 0;
10876 PL_cshlen = strlen(PL_cshname);
10881 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
10883 const I32 oldsavestack_ix = PL_savestack_ix;
10884 CV* outsidecv = PL_compcv;
10887 assert(SvTYPE(PL_compcv) == SVt_PVCV);
10889 SAVEI32(PL_subline);
10890 save_item(PL_subname);
10891 SAVESPTR(PL_compcv);
10893 PL_compcv = (CV*)NEWSV(1104,0);
10894 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
10895 CvFLAGS(PL_compcv) |= flags;
10897 PL_subline = CopLINE(PL_curcop);
10898 CvPADLIST(PL_compcv) = pad_new(padnew_SAVE|padnew_SAVESUB);
10899 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
10900 CvOUTSIDE_SEQ(PL_compcv) = PL_cop_seqmax;
10902 return oldsavestack_ix;
10906 #pragma segment Perl_yylex
10909 Perl_yywarn(pTHX_ const char *s)
10911 PL_in_eval |= EVAL_WARNONLY;
10913 PL_in_eval &= ~EVAL_WARNONLY;
10918 Perl_yyerror(pTHX_ const char *s)
10920 const char *where = NULL;
10921 const char *context = NULL;
10925 if (!yychar || (yychar == ';' && !PL_rsfp))
10927 else if (PL_oldoldbufptr && PL_bufptr > PL_oldoldbufptr &&
10928 PL_bufptr - PL_oldoldbufptr < 200 && PL_oldoldbufptr != PL_oldbufptr &&
10929 PL_oldbufptr != PL_bufptr) {
10932 The code below is removed for NetWare because it abends/crashes on NetWare
10933 when the script has error such as not having the closing quotes like:
10934 if ($var eq "value)
10935 Checking of white spaces is anyway done in NetWare code.
10938 while (isSPACE(*PL_oldoldbufptr))
10941 context = PL_oldoldbufptr;
10942 contlen = PL_bufptr - PL_oldoldbufptr;
10944 else if (PL_oldbufptr && PL_bufptr > PL_oldbufptr &&
10945 PL_bufptr - PL_oldbufptr < 200 && PL_oldbufptr != PL_bufptr) {
10948 The code below is removed for NetWare because it abends/crashes on NetWare
10949 when the script has error such as not having the closing quotes like:
10950 if ($var eq "value)
10951 Checking of white spaces is anyway done in NetWare code.
10954 while (isSPACE(*PL_oldbufptr))
10957 context = PL_oldbufptr;
10958 contlen = PL_bufptr - PL_oldbufptr;
10960 else if (yychar > 255)
10961 where = "next token ???";
10962 else if (yychar == -2) { /* YYEMPTY */
10963 if (PL_lex_state == LEX_NORMAL ||
10964 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
10965 where = "at end of line";
10966 else if (PL_lex_inpat)
10967 where = "within pattern";
10969 where = "within string";
10972 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
10974 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
10975 else if (isPRINT_LC(yychar))
10976 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
10978 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
10979 where = SvPVX_const(where_sv);
10981 msg = sv_2mortal(newSVpv(s, 0));
10982 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
10983 OutCopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
10985 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
10987 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
10988 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
10989 Perl_sv_catpvf(aTHX_ msg,
10990 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
10991 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
10994 if (PL_in_eval & EVAL_WARNONLY && ckWARN_d(WARN_SYNTAX))
10995 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "%"SVf, msg);
10998 if (PL_error_count >= 10) {
10999 if (PL_in_eval && SvCUR(ERRSV))
11000 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
11001 ERRSV, OutCopFILE(PL_curcop));
11003 Perl_croak(aTHX_ "%s has too many errors.\n",
11004 OutCopFILE(PL_curcop));
11007 PL_in_my_stash = NULL;
11011 #pragma segment Main
11015 S_swallow_bom(pTHX_ U8 *s)
11017 const STRLEN slen = SvCUR(PL_linestr);
11020 if (s[1] == 0xFE) {
11021 /* UTF-16 little-endian? (or UTF32-LE?) */
11022 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
11023 Perl_croak(aTHX_ "Unsupported script encoding UTF32-LE");
11024 #ifndef PERL_NO_UTF16_FILTER
11025 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF16-LE script encoding (BOM)\n");
11028 if (PL_bufend > (char*)s) {
11032 filter_add(utf16rev_textfilter, NULL);
11033 Newx(news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
11034 utf16_to_utf8_reversed(s, news,
11035 PL_bufend - (char*)s - 1,
11037 sv_setpvn(PL_linestr, (const char*)news, newlen);
11039 SvUTF8_on(PL_linestr);
11040 s = (U8*)SvPVX(PL_linestr);
11041 PL_bufend = SvPVX(PL_linestr) + newlen;
11044 Perl_croak(aTHX_ "Unsupported script encoding UTF16-LE");
11049 if (s[1] == 0xFF) { /* UTF-16 big-endian? */
11050 #ifndef PERL_NO_UTF16_FILTER
11051 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (BOM)\n");
11054 if (PL_bufend > (char *)s) {
11058 filter_add(utf16_textfilter, NULL);
11059 Newx(news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
11060 utf16_to_utf8(s, news,
11061 PL_bufend - (char*)s,
11063 sv_setpvn(PL_linestr, (const char*)news, newlen);
11065 SvUTF8_on(PL_linestr);
11066 s = (U8*)SvPVX(PL_linestr);
11067 PL_bufend = SvPVX(PL_linestr) + newlen;
11070 Perl_croak(aTHX_ "Unsupported script encoding UTF16-BE");
11075 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
11076 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-8 script encoding (BOM)\n");
11077 s += 3; /* UTF-8 */
11083 if (s[2] == 0xFE && s[3] == 0xFF) {
11084 /* UTF-32 big-endian */
11085 Perl_croak(aTHX_ "Unsupported script encoding UTF32-BE");
11088 else if (s[2] == 0 && s[3] != 0) {
11091 * are a good indicator of UTF-16BE. */
11092 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (no BOM)\n");
11097 if (slen > 3 && s[1] == 0 && s[2] != 0 && s[3] == 0) {
11100 * are a good indicator of UTF-16LE. */
11101 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16LE script encoding (no BOM)\n");
11110 * Restore a source filter.
11114 restore_rsfp(pTHX_ void *f)
11116 PerlIO * const fp = (PerlIO*)f;
11118 if (PL_rsfp == PerlIO_stdin())
11119 PerlIO_clearerr(PL_rsfp);
11120 else if (PL_rsfp && (PL_rsfp != fp))
11121 PerlIO_close(PL_rsfp);
11125 #ifndef PERL_NO_UTF16_FILTER
11127 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
11129 const STRLEN old = SvCUR(sv);
11130 const I32 count = FILTER_READ(idx+1, sv, maxlen);
11131 DEBUG_P(PerlIO_printf(Perl_debug_log,
11132 "utf16_textfilter(%p): %d %d (%d)\n",
11133 utf16_textfilter, idx, maxlen, (int) count));
11137 Newx(tmps, SvCUR(sv) * 3 / 2 + 1, U8);
11138 Copy(SvPVX_const(sv), tmps, old, char);
11139 utf16_to_utf8((U8*)SvPVX_const(sv) + old, tmps + old,
11140 SvCUR(sv) - old, &newlen);
11141 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
11143 DEBUG_P({sv_dump(sv);});
11148 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
11150 const STRLEN old = SvCUR(sv);
11151 const I32 count = FILTER_READ(idx+1, sv, maxlen);
11152 DEBUG_P(PerlIO_printf(Perl_debug_log,
11153 "utf16rev_textfilter(%p): %d %d (%d)\n",
11154 utf16rev_textfilter, idx, maxlen, (int) count));
11158 Newx(tmps, SvCUR(sv) * 3 / 2 + 1, U8);
11159 Copy(SvPVX_const(sv), tmps, old, char);
11160 utf16_to_utf8((U8*)SvPVX_const(sv) + old, tmps + old,
11161 SvCUR(sv) - old, &newlen);
11162 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
11164 DEBUG_P({ sv_dump(sv); });
11170 Returns a pointer to the next character after the parsed
11171 vstring, as well as updating the passed in sv.
11173 Function must be called like
11176 s = scan_vstring(s,sv);
11178 The sv should already be large enough to store the vstring
11179 passed in, for performance reasons.
11184 Perl_scan_vstring(pTHX_ const char *s, SV *sv)
11186 const char *pos = s;
11187 const char *start = s;
11188 if (*pos == 'v') pos++; /* get past 'v' */
11189 while (pos < PL_bufend && (isDIGIT(*pos) || *pos == '_'))
11191 if ( *pos != '.') {
11192 /* this may not be a v-string if followed by => */
11193 const char *next = pos;
11194 while (next < PL_bufend && isSPACE(*next))
11196 if ((PL_bufend - next) >= 2 && *next == '=' && next[1] == '>' ) {
11197 /* return string not v-string */
11198 sv_setpvn(sv,(char *)s,pos-s);
11199 return (char *)pos;
11203 if (!isALPHA(*pos)) {
11204 U8 tmpbuf[UTF8_MAXBYTES+1];
11206 if (*s == 'v') s++; /* get past 'v' */
11208 sv_setpvn(sv, "", 0);
11214 /* this is atoi() that tolerates underscores */
11215 const char *end = pos;
11217 while (--end >= s) {
11222 rev += (*end - '0') * mult;
11224 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
11225 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
11226 "Integer overflow in decimal number");
11230 if (rev > 0x7FFFFFFF)
11231 Perl_croak(aTHX_ "In EBCDIC the v-string components cannot exceed 2147483647");
11233 /* Append native character for the rev point */
11234 tmpend = uvchr_to_utf8(tmpbuf, rev);
11235 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
11236 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(rev)))
11238 if (pos + 1 < PL_bufend && *pos == '.' && isDIGIT(pos[1]))
11244 while (pos < PL_bufend && (isDIGIT(*pos) || *pos == '_'))
11248 sv_magic(sv,NULL,PERL_MAGIC_vstring,(const char*)start, pos-start);
11256 * c-indentation-style: bsd
11257 * c-basic-offset: 4
11258 * indent-tabs-mode: t
11261 * ex: set ts=8 sts=4 sw=4 noet: