3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
4 * 2000, 2001, 2002, 2003, 2004, 2005, 2006, 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";
30 static const char commaless_variable_list[] = "comma-less variable list";
32 static void restore_rsfp(pTHX_ void *f);
33 #ifndef PERL_NO_UTF16_FILTER
34 static I32 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen);
35 static I32 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen);
39 # define CURMAD(slot,sv) if (PL_madskills) { curmad(slot,sv); sv = 0; }
40 # define NEXTVAL_NEXTTOKE PL_nexttoke[PL_curforce].next_val
42 # define CURMAD(slot,sv)
43 # define NEXTVAL_NEXTTOKE PL_nextval[PL_nexttoke]
46 #define XFAKEBRACK 128
49 #ifdef USE_UTF8_SCRIPTS
50 # define UTF (!IN_BYTES)
52 # define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || (PL_hints & HINT_UTF8))
55 /* In variables named $^X, these are the legal values for X.
56 * 1999-02-27 mjd-perl-patch@plover.com */
57 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
59 /* On MacOS, respect nonbreaking spaces */
60 #ifdef MACOS_TRADITIONAL
61 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
63 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
66 /* LEX_* are values for PL_lex_state, the state of the lexer.
67 * They are arranged oddly so that the guard on the switch statement
68 * can get by with a single comparison (if the compiler is smart enough).
71 /* #define LEX_NOTPARSING 11 is done in perl.h. */
73 #define LEX_NORMAL 10 /* normal code (ie not within "...") */
74 #define LEX_INTERPNORMAL 9 /* code within a string, eg "$foo[$x+1]" */
75 #define LEX_INTERPCASEMOD 8 /* expecting a \U, \Q or \E etc */
76 #define LEX_INTERPPUSH 7 /* starting a new sublex parse level */
77 #define LEX_INTERPSTART 6 /* expecting the start of a $var */
79 /* at end of code, eg "$x" followed by: */
80 #define LEX_INTERPEND 5 /* ... eg not one of [, { or -> */
81 #define LEX_INTERPENDMAYBE 4 /* ... eg one of [, { or -> */
83 #define LEX_INTERPCONCAT 3 /* expecting anything, eg at start of
84 string or after \E, $foo, etc */
85 #define LEX_INTERPCONST 2 /* NOT USED */
86 #define LEX_FORMLINE 1 /* expecting a format line */
87 #define LEX_KNOWNEXT 0 /* next token known; just return it */
91 static const char* const lex_state_names[] = {
110 #include "keywords.h"
112 /* CLINE is a macro that ensures PL_copline has a sane value */
117 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
120 # define SKIPSPACE0(s) skipspace0(s)
121 # define SKIPSPACE1(s) skipspace1(s)
122 # define SKIPSPACE2(s,tsv) skipspace2(s,&tsv)
123 # define PEEKSPACE(s) skipspace2(s,0)
125 # define SKIPSPACE0(s) skipspace(s)
126 # define SKIPSPACE1(s) skipspace(s)
127 # define SKIPSPACE2(s,tsv) skipspace(s)
128 # define PEEKSPACE(s) skipspace(s)
132 * Convenience functions to return different tokens and prime the
133 * lexer for the next token. They all take an argument.
135 * TOKEN : generic token (used for '(', DOLSHARP, etc)
136 * OPERATOR : generic operator
137 * AOPERATOR : assignment operator
138 * PREBLOCK : beginning the block after an if, while, foreach, ...
139 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
140 * PREREF : *EXPR where EXPR is not a simple identifier
141 * TERM : expression term
142 * LOOPX : loop exiting command (goto, last, dump, etc)
143 * FTST : file test operator
144 * FUN0 : zero-argument function
145 * FUN1 : not used, except for not, which isn't a UNIOP
146 * BOop : bitwise or or xor
148 * SHop : shift operator
149 * PWop : power operator
150 * PMop : pattern-matching operator
151 * Aop : addition-level operator
152 * Mop : multiplication-level operator
153 * Eop : equality-testing operator
154 * Rop : relational operator <= != gt
156 * Also see LOP and lop() below.
159 #ifdef DEBUGGING /* Serve -DT. */
160 # define REPORT(retval) tokereport((I32)retval)
162 # define REPORT(retval) (retval)
165 #define TOKEN(retval) return ( PL_bufptr = s, REPORT(retval))
166 #define OPERATOR(retval) return (PL_expect = XTERM, PL_bufptr = s, REPORT(retval))
167 #define AOPERATOR(retval) return ao((PL_expect = XTERM, PL_bufptr = s, REPORT(retval)))
168 #define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s, REPORT(retval))
169 #define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s, REPORT(retval))
170 #define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s, REPORT(retval))
171 #define TERM(retval) return (CLINE, PL_expect = XOPERATOR, PL_bufptr = s, REPORT(retval))
172 #define LOOPX(f) return (yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)LOOPEX))
173 #define FTST(f) return (yylval.ival=f, PL_expect=XTERMORDORDOR, PL_bufptr=s, REPORT((int)UNIOP))
174 #define FUN0(f) return (yylval.ival=f, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC0))
175 #define FUN1(f) return (yylval.ival=f, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC1))
176 #define BOop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)BITOROP)))
177 #define BAop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)BITANDOP)))
178 #define SHop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)SHIFTOP)))
179 #define PWop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)POWOP)))
180 #define PMop(f) return(yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)MATCHOP))
181 #define Aop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)ADDOP)))
182 #define Mop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)MULOP)))
183 #define Eop(f) return (yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)EQOP))
184 #define Rop(f) return (yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)RELOP))
186 /* This bit of chicanery makes a unary function followed by
187 * a parenthesis into a function with one argument, highest precedence.
188 * The UNIDOR macro is for unary functions that can be followed by the //
189 * operator (such as C<shift // 0>).
191 #define UNI2(f,x) { \
195 PL_last_uni = PL_oldbufptr; \
196 PL_last_lop_op = f; \
198 return REPORT( (int)FUNC1 ); \
200 return REPORT( *s=='(' ? (int)FUNC1 : (int)UNIOP ); \
202 #define UNI(f) UNI2(f,XTERM)
203 #define UNIDOR(f) UNI2(f,XTERMORDORDOR)
205 #define UNIBRACK(f) { \
208 PL_last_uni = PL_oldbufptr; \
210 return REPORT( (int)FUNC1 ); \
212 return REPORT( (*s == '(') ? (int)FUNC1 : (int)UNIOP ); \
215 /* grandfather return to old style */
216 #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
220 /* how to interpret the yylval associated with the token */
224 TOKENTYPE_OPNUM, /* yylval.ival contains an opcode number */
230 static struct debug_tokens {
232 enum token_type type;
234 } const debug_tokens[] =
236 { ADDOP, TOKENTYPE_OPNUM, "ADDOP" },
237 { ANDAND, TOKENTYPE_NONE, "ANDAND" },
238 { ANDOP, TOKENTYPE_NONE, "ANDOP" },
239 { ANONSUB, TOKENTYPE_IVAL, "ANONSUB" },
240 { ARROW, TOKENTYPE_NONE, "ARROW" },
241 { ASSIGNOP, TOKENTYPE_OPNUM, "ASSIGNOP" },
242 { BITANDOP, TOKENTYPE_OPNUM, "BITANDOP" },
243 { BITOROP, TOKENTYPE_OPNUM, "BITOROP" },
244 { COLONATTR, TOKENTYPE_NONE, "COLONATTR" },
245 { CONTINUE, TOKENTYPE_NONE, "CONTINUE" },
246 { DEFAULT, TOKENTYPE_NONE, "DEFAULT" },
247 { DO, TOKENTYPE_NONE, "DO" },
248 { DOLSHARP, TOKENTYPE_NONE, "DOLSHARP" },
249 { DORDOR, TOKENTYPE_NONE, "DORDOR" },
250 { DOROP, TOKENTYPE_OPNUM, "DOROP" },
251 { DOTDOT, TOKENTYPE_IVAL, "DOTDOT" },
252 { ELSE, TOKENTYPE_NONE, "ELSE" },
253 { ELSIF, TOKENTYPE_IVAL, "ELSIF" },
254 { EQOP, TOKENTYPE_OPNUM, "EQOP" },
255 { FOR, TOKENTYPE_IVAL, "FOR" },
256 { FORMAT, TOKENTYPE_NONE, "FORMAT" },
257 { FUNC, TOKENTYPE_OPNUM, "FUNC" },
258 { FUNC0, TOKENTYPE_OPNUM, "FUNC0" },
259 { FUNC0SUB, TOKENTYPE_OPVAL, "FUNC0SUB" },
260 { FUNC1, TOKENTYPE_OPNUM, "FUNC1" },
261 { FUNCMETH, TOKENTYPE_OPVAL, "FUNCMETH" },
262 { GIVEN, TOKENTYPE_IVAL, "GIVEN" },
263 { HASHBRACK, TOKENTYPE_NONE, "HASHBRACK" },
264 { IF, TOKENTYPE_IVAL, "IF" },
265 { LABEL, TOKENTYPE_PVAL, "LABEL" },
266 { LOCAL, TOKENTYPE_IVAL, "LOCAL" },
267 { LOOPEX, TOKENTYPE_OPNUM, "LOOPEX" },
268 { LSTOP, TOKENTYPE_OPNUM, "LSTOP" },
269 { LSTOPSUB, TOKENTYPE_OPVAL, "LSTOPSUB" },
270 { MATCHOP, TOKENTYPE_OPNUM, "MATCHOP" },
271 { METHOD, TOKENTYPE_OPVAL, "METHOD" },
272 { MULOP, TOKENTYPE_OPNUM, "MULOP" },
273 { MY, TOKENTYPE_IVAL, "MY" },
274 { MYSUB, TOKENTYPE_NONE, "MYSUB" },
275 { NOAMP, TOKENTYPE_NONE, "NOAMP" },
276 { NOTOP, TOKENTYPE_NONE, "NOTOP" },
277 { OROP, TOKENTYPE_IVAL, "OROP" },
278 { OROR, TOKENTYPE_NONE, "OROR" },
279 { PACKAGE, TOKENTYPE_NONE, "PACKAGE" },
280 { PMFUNC, TOKENTYPE_OPVAL, "PMFUNC" },
281 { POSTDEC, TOKENTYPE_NONE, "POSTDEC" },
282 { POSTINC, TOKENTYPE_NONE, "POSTINC" },
283 { POWOP, TOKENTYPE_OPNUM, "POWOP" },
284 { PREDEC, TOKENTYPE_NONE, "PREDEC" },
285 { PREINC, TOKENTYPE_NONE, "PREINC" },
286 { PRIVATEREF, TOKENTYPE_OPVAL, "PRIVATEREF" },
287 { REFGEN, TOKENTYPE_NONE, "REFGEN" },
288 { RELOP, TOKENTYPE_OPNUM, "RELOP" },
289 { SHIFTOP, TOKENTYPE_OPNUM, "SHIFTOP" },
290 { SUB, TOKENTYPE_NONE, "SUB" },
291 { THING, TOKENTYPE_OPVAL, "THING" },
292 { UMINUS, TOKENTYPE_NONE, "UMINUS" },
293 { UNIOP, TOKENTYPE_OPNUM, "UNIOP" },
294 { UNIOPSUB, TOKENTYPE_OPVAL, "UNIOPSUB" },
295 { UNLESS, TOKENTYPE_IVAL, "UNLESS" },
296 { UNTIL, TOKENTYPE_IVAL, "UNTIL" },
297 { USE, TOKENTYPE_IVAL, "USE" },
298 { WHEN, TOKENTYPE_IVAL, "WHEN" },
299 { WHILE, TOKENTYPE_IVAL, "WHILE" },
300 { WORD, TOKENTYPE_OPVAL, "WORD" },
301 { 0, TOKENTYPE_NONE, NULL }
304 /* dump the returned token in rv, plus any optional arg in yylval */
307 S_tokereport(pTHX_ I32 rv)
311 const char *name = NULL;
312 enum token_type type = TOKENTYPE_NONE;
313 const struct debug_tokens *p;
314 SV* const report = newSVpvs("<== ");
316 for (p = debug_tokens; p->token; p++) {
317 if (p->token == (int)rv) {
324 Perl_sv_catpv(aTHX_ report, name);
325 else if ((char)rv > ' ' && (char)rv < '~')
326 Perl_sv_catpvf(aTHX_ report, "'%c'", (char)rv);
328 sv_catpvs(report, "EOF");
330 Perl_sv_catpvf(aTHX_ report, "?? %"IVdf, (IV)rv);
333 case TOKENTYPE_GVVAL: /* doesn't appear to be used */
336 Perl_sv_catpvf(aTHX_ report, "(ival=%"IVdf")", (IV)yylval.ival);
338 case TOKENTYPE_OPNUM:
339 Perl_sv_catpvf(aTHX_ report, "(ival=op_%s)",
340 PL_op_name[yylval.ival]);
343 Perl_sv_catpvf(aTHX_ report, "(pval=\"%s\")", yylval.pval);
345 case TOKENTYPE_OPVAL:
347 Perl_sv_catpvf(aTHX_ report, "(opval=op_%s)",
348 PL_op_name[yylval.opval->op_type]);
349 if (yylval.opval->op_type == OP_CONST) {
350 Perl_sv_catpvf(aTHX_ report, " %s",
351 SvPEEK(cSVOPx_sv(yylval.opval)));
356 sv_catpvs(report, "(opval=null)");
359 PerlIO_printf(Perl_debug_log, "### %s\n\n", SvPV_nolen_const(report));
365 /* print the buffer with suitable escapes */
368 S_printbuf(pTHX_ const char* fmt, const char* s)
370 SV* const tmp = newSVpvs("");
371 PerlIO_printf(Perl_debug_log, fmt, pv_display(tmp, s, strlen(s), 0, 60));
380 * This subroutine detects &&=, ||=, and //= and turns an ANDAND, OROR or DORDOR
381 * into an OP_ANDASSIGN, OP_ORASSIGN, or OP_DORASSIGN
385 S_ao(pTHX_ int toketype)
388 if (*PL_bufptr == '=') {
390 if (toketype == ANDAND)
391 yylval.ival = OP_ANDASSIGN;
392 else if (toketype == OROR)
393 yylval.ival = OP_ORASSIGN;
394 else if (toketype == DORDOR)
395 yylval.ival = OP_DORASSIGN;
403 * When Perl expects an operator and finds something else, no_op
404 * prints the warning. It always prints "<something> found where
405 * operator expected. It prints "Missing semicolon on previous line?"
406 * if the surprise occurs at the start of the line. "do you need to
407 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
408 * where the compiler doesn't know if foo is a method call or a function.
409 * It prints "Missing operator before end of line" if there's nothing
410 * after the missing operator, or "... before <...>" if there is something
411 * after the missing operator.
415 S_no_op(pTHX_ const char *what, char *s)
418 char * const oldbp = PL_bufptr;
419 const bool is_first = (PL_oldbufptr == PL_linestart);
425 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
426 if (ckWARN_d(WARN_SYNTAX)) {
428 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
429 "\t(Missing semicolon on previous line?)\n");
430 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
432 for (t = PL_oldoldbufptr; (isALNUM_lazy_if(t,UTF) || *t == ':'); t++)
434 if (t < PL_bufptr && isSPACE(*t))
435 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
436 "\t(Do you need to predeclare %.*s?)\n",
437 (int)(t - PL_oldoldbufptr), PL_oldoldbufptr);
441 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
442 "\t(Missing operator before %.*s?)\n", (int)(s - oldbp), oldbp);
450 * Complain about missing quote/regexp/heredoc terminator.
451 * If it's called with NULL then it cauterizes the line buffer.
452 * If we're in a delimited string and the delimiter is a control
453 * character, it's reformatted into a two-char sequence like ^C.
458 S_missingterm(pTHX_ char *s)
464 char * const nl = strrchr(s,'\n');
470 iscntrl(PL_multi_close)
472 PL_multi_close < 32 || PL_multi_close == 127
476 tmpbuf[1] = (char)toCTRL(PL_multi_close);
481 *tmpbuf = (char)PL_multi_close;
485 q = strchr(s,'"') ? '\'' : '"';
486 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
489 #define FEATURE_IS_ENABLED(name) \
490 ((0 != (PL_hints & HINT_LOCALIZE_HH)) \
491 && S_feature_is_enabled(aTHX_ STR_WITH_LEN(name)))
493 * S_feature_is_enabled
494 * Check whether the named feature is enabled.
497 S_feature_is_enabled(pTHX_ const char *name, STRLEN namelen)
500 HV * const hinthv = GvHV(PL_hintgv);
501 char he_name[32] = "feature_";
502 (void) my_strlcpy(&he_name[8], name, 24);
504 return (hinthv && hv_exists(hinthv, he_name, 8 + namelen));
512 Perl_deprecate(pTHX_ const char *s)
514 if (ckWARN(WARN_DEPRECATED))
515 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED), "Use of %s is deprecated", s);
519 Perl_deprecate_old(pTHX_ const char *s)
521 /* This function should NOT be called for any new deprecated warnings */
522 /* Use Perl_deprecate instead */
524 /* It is here to maintain backward compatibility with the pre-5.8 */
525 /* warnings category hierarchy. The "deprecated" category used to */
526 /* live under the "syntax" category. It is now a top-level category */
527 /* in its own right. */
529 if (ckWARN2(WARN_DEPRECATED, WARN_SYNTAX))
530 Perl_warner(aTHX_ packWARN2(WARN_DEPRECATED, WARN_SYNTAX),
531 "Use of %s is deprecated", s);
535 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
536 * utf16-to-utf8-reversed.
539 #ifdef PERL_CR_FILTER
543 register const char *s = SvPVX_const(sv);
544 register const char * const e = s + SvCUR(sv);
545 /* outer loop optimized to do nothing if there are no CR-LFs */
547 if (*s++ == '\r' && *s == '\n') {
548 /* hit a CR-LF, need to copy the rest */
549 register char *d = s - 1;
552 if (*s == '\r' && s[1] == '\n')
563 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
565 const I32 count = FILTER_READ(idx+1, sv, maxlen);
566 if (count > 0 && !maxlen)
574 * Initialize variables. Uses the Perl save_stack to save its state (for
575 * recursive calls to the parser).
579 Perl_lex_start(pTHX_ SV *line)
585 SAVEI32(PL_lex_dojoin);
586 SAVEI32(PL_lex_brackets);
587 SAVEI32(PL_lex_casemods);
588 SAVEI32(PL_lex_starts);
589 SAVEI32(PL_lex_state);
590 SAVEVPTR(PL_lex_inpat);
591 SAVEI32(PL_lex_inwhat);
593 if (PL_lex_state == LEX_KNOWNEXT) {
594 I32 toke = PL_lasttoke;
595 while (--toke >= 0) {
596 SAVEI32(PL_nexttoke[toke].next_type);
597 SAVEVPTR(PL_nexttoke[toke].next_val);
599 SAVEVPTR(PL_nexttoke[toke].next_mad);
601 SAVEI32(PL_lasttoke);
604 SAVESPTR(PL_thistoken);
605 SAVESPTR(PL_thiswhite);
606 SAVESPTR(PL_nextwhite);
607 SAVESPTR(PL_thisopen);
608 SAVESPTR(PL_thisclose);
609 SAVESPTR(PL_thisstuff);
610 SAVEVPTR(PL_thismad);
611 SAVEI32(PL_realtokenstart);
612 SAVEI32(PL_faketokens);
614 SAVEI32(PL_curforce);
616 if (PL_lex_state == LEX_KNOWNEXT) {
617 I32 toke = PL_nexttoke;
618 while (--toke >= 0) {
619 SAVEI32(PL_nexttype[toke]);
620 SAVEVPTR(PL_nextval[toke]);
622 SAVEI32(PL_nexttoke);
625 SAVECOPLINE(PL_curcop);
628 SAVEPPTR(PL_oldbufptr);
629 SAVEPPTR(PL_oldoldbufptr);
630 SAVEPPTR(PL_last_lop);
631 SAVEPPTR(PL_last_uni);
632 SAVEPPTR(PL_linestart);
633 SAVESPTR(PL_linestr);
634 SAVEGENERICPV(PL_lex_brackstack);
635 SAVEGENERICPV(PL_lex_casestack);
636 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
637 SAVESPTR(PL_lex_stuff);
638 SAVEI32(PL_lex_defer);
639 SAVEI32(PL_sublex_info.sub_inwhat);
640 SAVESPTR(PL_lex_repl);
642 SAVEINT(PL_lex_expect);
644 PL_lex_state = LEX_NORMAL;
648 Newx(PL_lex_brackstack, 120, char);
649 Newx(PL_lex_casestack, 12, char);
651 *PL_lex_casestack = '\0';
663 PL_sublex_info.sub_inwhat = 0;
665 if (SvREADONLY(PL_linestr))
666 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
667 s = SvPV_const(PL_linestr, len);
668 if (!len || s[len-1] != ';') {
669 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
670 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
671 sv_catpvs(PL_linestr, "\n;");
673 SvTEMP_off(PL_linestr);
674 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
675 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
676 PL_last_lop = PL_last_uni = NULL;
682 * Finalizer for lexing operations. Must be called when the parser is
683 * done with the lexer.
690 PL_doextract = FALSE;
695 * This subroutine has nothing to do with tilting, whether at windmills
696 * or pinball tables. Its name is short for "increment line". It
697 * increments the current line number in CopLINE(PL_curcop) and checks
698 * to see whether the line starts with a comment of the form
699 * # line 500 "foo.pm"
700 * If so, it sets the current line number and file to the values in the comment.
704 S_incline(pTHX_ char *s)
712 CopLINE_inc(PL_curcop);
715 while (SPACE_OR_TAB(*s))
717 if (strnEQ(s, "line", 4))
721 if (SPACE_OR_TAB(*s))
725 while (SPACE_OR_TAB(*s))
733 while (SPACE_OR_TAB(*s))
735 if (*s == '"' && (t = strchr(s+1, '"'))) {
745 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
747 if (*e != '\n' && *e != '\0')
748 return; /* false alarm */
754 const char * const cf = CopFILE(PL_curcop);
755 STRLEN tmplen = cf ? strlen(cf) : 0;
756 if (tmplen > 7 && strnEQ(cf, "(eval ", 6)) {
757 /* must copy *{"::_<(eval N)[oldfilename:L]"}
758 * to *{"::_<newfilename"} */
759 char smallbuf[256], smallbuf2[256];
760 char *tmpbuf, *tmpbuf2;
762 STRLEN tmplen2 = strlen(s);
763 if (tmplen + 3 < sizeof smallbuf)
766 Newx(tmpbuf, tmplen + 3, char);
767 if (tmplen2 + 3 < sizeof smallbuf2)
770 Newx(tmpbuf2, tmplen2 + 3, char);
771 tmpbuf[0] = tmpbuf2[0] = '_';
772 tmpbuf[1] = tmpbuf2[1] = '<';
773 memcpy(tmpbuf + 2, cf, ++tmplen);
774 memcpy(tmpbuf2 + 2, s, ++tmplen2);
776 gvp = (GV**)hv_fetch(PL_defstash, tmpbuf, tmplen, FALSE);
778 gv2 = *(GV**)hv_fetch(PL_defstash, tmpbuf2, tmplen2, TRUE);
780 gv_init(gv2, PL_defstash, tmpbuf2, tmplen2, FALSE);
781 /* adjust ${"::_<newfilename"} to store the new file name */
782 GvSV(gv2) = newSVpvn(tmpbuf2 + 2, tmplen2 - 2);
783 GvHV(gv2) = (HV*)SvREFCNT_inc(GvHV(*gvp));
784 GvAV(gv2) = (AV*)SvREFCNT_inc(GvAV(*gvp));
786 if (tmpbuf != smallbuf) Safefree(tmpbuf);
787 if (tmpbuf2 != smallbuf2) Safefree(tmpbuf2);
790 CopFILE_free(PL_curcop);
791 CopFILE_set(PL_curcop, s);
794 CopLINE_set(PL_curcop, atoi(n)-1);
798 /* skip space before PL_thistoken */
801 S_skipspace0(pTHX_ register char *s)
808 PL_thiswhite = newSVpvn("",0);
809 sv_catsv(PL_thiswhite, PL_skipwhite);
810 sv_free(PL_skipwhite);
813 PL_realtokenstart = s - SvPVX(PL_linestr);
817 /* skip space after PL_thistoken */
820 S_skipspace1(pTHX_ register char *s)
822 const char *start = s;
823 I32 startoff = start - SvPVX(PL_linestr);
828 start = SvPVX(PL_linestr) + startoff;
829 if (!PL_thistoken && PL_realtokenstart >= 0) {
830 const char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
831 PL_thistoken = newSVpvn(tstart, start - tstart);
833 PL_realtokenstart = -1;
836 PL_nextwhite = newSVpvn("",0);
837 sv_catsv(PL_nextwhite, PL_skipwhite);
838 sv_free(PL_skipwhite);
845 S_skipspace2(pTHX_ register char *s, SV **svp)
848 const I32 bufptroff = PL_bufptr - SvPVX(PL_linestr);
849 const I32 startoff = s - SvPVX(PL_linestr);
852 PL_bufptr = SvPVX(PL_linestr) + bufptroff;
853 if (!PL_madskills || !svp)
855 start = SvPVX(PL_linestr) + startoff;
856 if (!PL_thistoken && PL_realtokenstart >= 0) {
857 char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
858 PL_thistoken = newSVpvn(tstart, start - tstart);
859 PL_realtokenstart = -1;
863 *svp = newSVpvn("",0);
864 sv_setsv(*svp, PL_skipwhite);
865 sv_free(PL_skipwhite);
875 * Called to gobble the appropriate amount and type of whitespace.
876 * Skips comments as well.
880 S_skipspace(pTHX_ register char *s)
885 int startoff = s - SvPVX(PL_linestr);
888 sv_free(PL_skipwhite);
893 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
894 while (s < PL_bufend && SPACE_OR_TAB(*s))
904 SSize_t oldprevlen, oldoldprevlen;
905 SSize_t oldloplen = 0, oldunilen = 0;
906 while (s < PL_bufend && isSPACE(*s)) {
907 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
912 if (s < PL_bufend && *s == '#') {
913 while (s < PL_bufend && *s != '\n')
917 if (PL_in_eval && !PL_rsfp) {
924 /* only continue to recharge the buffer if we're at the end
925 * of the buffer, we're not reading from a source filter, and
926 * we're in normal lexing mode
928 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
929 PL_lex_state == LEX_FORMLINE)
936 /* try to recharge the buffer */
938 curoff = s - SvPVX(PL_linestr);
941 if ((s = filter_gets(PL_linestr, PL_rsfp,
942 (prevlen = SvCUR(PL_linestr)))) == NULL)
945 if (PL_madskills && curoff != startoff) {
947 PL_skipwhite = newSVpvn("",0);
948 sv_catpvn(PL_skipwhite, SvPVX(PL_linestr) + startoff,
952 /* mustn't throw out old stuff yet if madpropping */
953 SvCUR(PL_linestr) = curoff;
954 s = SvPVX(PL_linestr) + curoff;
956 if (curoff && s[-1] == '\n')
960 /* end of file. Add on the -p or -n magic */
961 /* XXX these shouldn't really be added here, can't set PL_faketokens */
965 ";}continue{print or die qq(-p destination: $!\\n);}");
968 ";}continue{print or die qq(-p destination: $!\\n);}");
970 PL_minus_n = PL_minus_p = 0;
972 else if (PL_minus_n) {
974 sv_catpvn(PL_linestr, ";}", 2);
976 sv_setpvn(PL_linestr, ";}", 2);
982 sv_catpvn(PL_linestr,";", 1);
984 sv_setpvn(PL_linestr,";", 1);
987 /* reset variables for next time we lex */
988 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
994 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
995 PL_last_lop = PL_last_uni = NULL;
997 /* Close the filehandle. Could be from -P preprocessor,
998 * STDIN, or a regular file. If we were reading code from
999 * STDIN (because the commandline held no -e or filename)
1000 * then we don't close it, we reset it so the code can
1001 * read from STDIN too.
1004 if (PL_preprocess && !PL_in_eval)
1005 (void)PerlProc_pclose(PL_rsfp);
1006 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
1007 PerlIO_clearerr(PL_rsfp);
1009 (void)PerlIO_close(PL_rsfp);
1014 /* not at end of file, so we only read another line */
1015 /* make corresponding updates to old pointers, for yyerror() */
1016 oldprevlen = PL_oldbufptr - PL_bufend;
1017 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
1019 oldunilen = PL_last_uni - PL_bufend;
1021 oldloplen = PL_last_lop - PL_bufend;
1022 PL_linestart = PL_bufptr = s + prevlen;
1023 PL_bufend = s + SvCUR(PL_linestr);
1025 PL_oldbufptr = s + oldprevlen;
1026 PL_oldoldbufptr = s + oldoldprevlen;
1028 PL_last_uni = s + oldunilen;
1030 PL_last_lop = s + oldloplen;
1033 /* debugger active and we're not compiling the debugger code,
1034 * so store the line into the debugger's array of lines
1036 if (PERLDB_LINE && PL_curstash != PL_debstash) {
1037 SV * const sv = newSV(0);
1039 sv_upgrade(sv, SVt_PVMG);
1040 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
1043 av_store(CopFILEAVx(PL_curcop),(I32)CopLINE(PL_curcop),sv);
1051 PL_skipwhite = newSVpvn("",0);
1052 curoff = s - SvPVX(PL_linestr);
1053 if (curoff - startoff)
1054 sv_catpvn(PL_skipwhite, SvPVX(PL_linestr) + startoff,
1063 * Check the unary operators to ensure there's no ambiguity in how they're
1064 * used. An ambiguous piece of code would be:
1066 * This doesn't mean rand() + 5. Because rand() is a unary operator,
1067 * the +5 is its argument.
1077 if (PL_oldoldbufptr != PL_last_uni)
1079 while (isSPACE(*PL_last_uni))
1082 while (isALNUM_lazy_if(s,UTF) || *s == '-')
1084 if ((t = strchr(s, '(')) && t < PL_bufptr)
1087 if (ckWARN_d(WARN_AMBIGUOUS)){
1088 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
1089 "Warning: Use of \"%.*s\" without parentheses is ambiguous",
1090 (int)(s - PL_last_uni), PL_last_uni);
1095 * LOP : macro to build a list operator. Its behaviour has been replaced
1096 * with a subroutine, S_lop() for which LOP is just another name.
1099 #define LOP(f,x) return lop(f,x,s)
1103 * Build a list operator (or something that might be one). The rules:
1104 * - if we have a next token, then it's a list operator [why?]
1105 * - if the next thing is an opening paren, then it's a function
1106 * - else it's a list operator
1110 S_lop(pTHX_ I32 f, int x, char *s)
1117 PL_last_lop = PL_oldbufptr;
1118 PL_last_lop_op = (OPCODE)f;
1121 return REPORT(LSTOP);
1124 return REPORT(LSTOP);
1127 return REPORT(FUNC);
1130 return REPORT(FUNC);
1132 return REPORT(LSTOP);
1138 * Sets up for an eventual force_next(). start_force(0) basically does
1139 * an unshift, while start_force(-1) does a push. yylex removes items
1144 S_start_force(pTHX_ int where)
1148 if (where < 0) /* so people can duplicate start_force(PL_curforce) */
1149 where = PL_lasttoke;
1150 assert(PL_curforce < 0 || PL_curforce == where);
1151 if (PL_curforce != where) {
1152 for (i = PL_lasttoke; i > where; --i) {
1153 PL_nexttoke[i] = PL_nexttoke[i-1];
1157 if (PL_curforce < 0) /* in case of duplicate start_force() */
1158 Zero(&PL_nexttoke[where], 1, NEXTTOKE);
1159 PL_curforce = where;
1162 curmad('^', newSVpvn("",0));
1163 CURMAD('_', PL_nextwhite);
1168 S_curmad(pTHX_ char slot, SV *sv)
1174 if (PL_curforce < 0)
1175 where = &PL_thismad;
1177 where = &PL_nexttoke[PL_curforce].next_mad;
1180 sv_setpvn(sv, "", 0);
1183 if (UTF && is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
1185 else if (PL_encoding) {
1186 sv_recode_to_utf8(sv, PL_encoding);
1191 /* keep a slot open for the head of the list? */
1192 if (slot != '_' && *where && (*where)->mad_key == '^') {
1193 (*where)->mad_key = slot;
1194 sv_free((*where)->mad_val);
1195 (*where)->mad_val = (void*)sv;
1198 addmad(newMADsv(slot, sv), where, 0);
1201 # define start_force(where) NOOP
1202 # define curmad(slot, sv) NOOP
1207 * When the lexer realizes it knows the next token (for instance,
1208 * it is reordering tokens for the parser) then it can call S_force_next
1209 * to know what token to return the next time the lexer is called. Caller
1210 * will need to set PL_nextval[] (or PL_nexttoke[].next_val with PERL_MAD),
1211 * and possibly PL_expect to ensure the lexer handles the token correctly.
1215 S_force_next(pTHX_ I32 type)
1219 if (PL_curforce < 0)
1220 start_force(PL_lasttoke);
1221 PL_nexttoke[PL_curforce].next_type = type;
1222 if (PL_lex_state != LEX_KNOWNEXT)
1223 PL_lex_defer = PL_lex_state;
1224 PL_lex_state = LEX_KNOWNEXT;
1225 PL_lex_expect = PL_expect;
1228 PL_nexttype[PL_nexttoke] = type;
1230 if (PL_lex_state != LEX_KNOWNEXT) {
1231 PL_lex_defer = PL_lex_state;
1232 PL_lex_expect = PL_expect;
1233 PL_lex_state = LEX_KNOWNEXT;
1239 S_newSV_maybe_utf8(pTHX_ const char *start, STRLEN len)
1242 SV * const sv = newSVpvn(start,len);
1243 if (UTF && !IN_BYTES && is_utf8_string((const U8*)start, len))
1250 * When the lexer knows the next thing is a word (for instance, it has
1251 * just seen -> and it knows that the next char is a word char, then
1252 * it calls S_force_word to stick the next word into the PL_next lookahead.
1255 * char *start : buffer position (must be within PL_linestr)
1256 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
1257 * int check_keyword : if true, Perl checks to make sure the word isn't
1258 * a keyword (do this if the word is a label, e.g. goto FOO)
1259 * int allow_pack : if true, : characters will also be allowed (require,
1260 * use, etc. do this)
1261 * int allow_initial_tick : used by the "sub" lexer only.
1265 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
1271 start = SKIPSPACE1(start);
1273 if (isIDFIRST_lazy_if(s,UTF) ||
1274 (allow_pack && *s == ':') ||
1275 (allow_initial_tick && *s == '\'') )
1277 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
1278 if (check_keyword && keyword(PL_tokenbuf, len))
1280 start_force(PL_curforce);
1282 curmad('X', newSVpvn(start,s-start));
1283 if (token == METHOD) {
1288 PL_expect = XOPERATOR;
1291 NEXTVAL_NEXTTOKE.opval
1292 = (OP*)newSVOP(OP_CONST,0,
1293 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
1294 NEXTVAL_NEXTTOKE.opval->op_private |= OPpCONST_BARE;
1302 * Called when the lexer wants $foo *foo &foo etc, but the program
1303 * text only contains the "foo" portion. The first argument is a pointer
1304 * to the "foo", and the second argument is the type symbol to prefix.
1305 * Forces the next token to be a "WORD".
1306 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
1310 S_force_ident(pTHX_ register const char *s, int kind)
1314 const STRLEN len = strlen(s);
1315 OP* const o = (OP*)newSVOP(OP_CONST, 0, newSVpvn(s, len));
1316 start_force(PL_curforce);
1317 NEXTVAL_NEXTTOKE.opval = o;
1320 o->op_private = OPpCONST_ENTERED;
1321 /* XXX see note in pp_entereval() for why we forgo typo
1322 warnings if the symbol must be introduced in an eval.
1324 gv_fetchpvn_flags(s, len,
1325 PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL)
1327 kind == '$' ? SVt_PV :
1328 kind == '@' ? SVt_PVAV :
1329 kind == '%' ? SVt_PVHV :
1337 Perl_str_to_version(pTHX_ SV *sv)
1342 const char *start = SvPV_const(sv,len);
1343 const char * const end = start + len;
1344 const bool utf = SvUTF8(sv) ? TRUE : FALSE;
1345 while (start < end) {
1349 n = utf8n_to_uvchr((U8*)start, len, &skip, 0);
1354 retval += ((NV)n)/nshift;
1363 * Forces the next token to be a version number.
1364 * If the next token appears to be an invalid version number, (e.g. "v2b"),
1365 * and if "guessing" is TRUE, then no new token is created (and the caller
1366 * must use an alternative parsing method).
1370 S_force_version(pTHX_ char *s, int guessing)
1376 I32 startoff = s - SvPVX(PL_linestr);
1385 while (isDIGIT(*d) || *d == '_' || *d == '.')
1389 start_force(PL_curforce);
1390 curmad('X', newSVpvn(s,d-s));
1393 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
1395 s = scan_num(s, &yylval);
1396 version = yylval.opval;
1397 ver = cSVOPx(version)->op_sv;
1398 if (SvPOK(ver) && !SvNIOK(ver)) {
1399 SvUPGRADE(ver, SVt_PVNV);
1400 SvNV_set(ver, str_to_version(ver));
1401 SvNOK_on(ver); /* hint that it is a version */
1404 else if (guessing) {
1407 sv_free(PL_nextwhite); /* let next token collect whitespace */
1409 s = SvPVX(PL_linestr) + startoff;
1417 if (PL_madskills && !version) {
1418 sv_free(PL_nextwhite); /* let next token collect whitespace */
1420 s = SvPVX(PL_linestr) + startoff;
1423 /* NOTE: The parser sees the package name and the VERSION swapped */
1424 start_force(PL_curforce);
1425 NEXTVAL_NEXTTOKE.opval = version;
1433 * Tokenize a quoted string passed in as an SV. It finds the next
1434 * chunk, up to end of string or a backslash. It may make a new
1435 * SV containing that chunk (if HINT_NEW_STRING is on). It also
1440 S_tokeq(pTHX_ SV *sv)
1444 register char *send;
1452 s = SvPV_force(sv, len);
1453 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
1456 while (s < send && *s != '\\')
1461 if ( PL_hints & HINT_NEW_STRING ) {
1462 pv = sv_2mortal(newSVpvn(SvPVX_const(pv), len));
1468 if (s + 1 < send && (s[1] == '\\'))
1469 s++; /* all that, just for this */
1474 SvCUR_set(sv, d - SvPVX_const(sv));
1476 if ( PL_hints & HINT_NEW_STRING )
1477 return new_constant(NULL, 0, "q", sv, pv, "q");
1482 * Now come three functions related to double-quote context,
1483 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
1484 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
1485 * interact with PL_lex_state, and create fake ( ... ) argument lists
1486 * to handle functions and concatenation.
1487 * They assume that whoever calls them will be setting up a fake
1488 * join call, because each subthing puts a ',' after it. This lets
1491 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
1493 * (I'm not sure whether the spurious commas at the end of lcfirst's
1494 * arguments and join's arguments are created or not).
1499 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
1501 * Pattern matching will set PL_lex_op to the pattern-matching op to
1502 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
1504 * OP_CONST and OP_READLINE are easy--just make the new op and return.
1506 * Everything else becomes a FUNC.
1508 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
1509 * had an OP_CONST or OP_READLINE). This just sets us up for a
1510 * call to S_sublex_push().
1514 S_sublex_start(pTHX)
1517 register const I32 op_type = yylval.ival;
1519 if (op_type == OP_NULL) {
1520 yylval.opval = PL_lex_op;
1524 if (op_type == OP_CONST || op_type == OP_READLINE) {
1525 SV *sv = tokeq(PL_lex_stuff);
1527 if (SvTYPE(sv) == SVt_PVIV) {
1528 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
1530 const char * const p = SvPV_const(sv, len);
1531 SV * const nsv = newSVpvn(p, len);
1537 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
1538 PL_lex_stuff = NULL;
1539 /* Allow <FH> // "foo" */
1540 if (op_type == OP_READLINE)
1541 PL_expect = XTERMORDORDOR;
1545 PL_sublex_info.super_state = PL_lex_state;
1546 PL_sublex_info.sub_inwhat = op_type;
1547 PL_sublex_info.sub_op = PL_lex_op;
1548 PL_lex_state = LEX_INTERPPUSH;
1552 yylval.opval = PL_lex_op;
1562 * Create a new scope to save the lexing state. The scope will be
1563 * ended in S_sublex_done. Returns a '(', starting the function arguments
1564 * to the uc, lc, etc. found before.
1565 * Sets PL_lex_state to LEX_INTERPCONCAT.
1574 PL_lex_state = PL_sublex_info.super_state;
1575 SAVEI32(PL_lex_dojoin);
1576 SAVEI32(PL_lex_brackets);
1577 SAVEI32(PL_lex_casemods);
1578 SAVEI32(PL_lex_starts);
1579 SAVEI32(PL_lex_state);
1580 SAVEVPTR(PL_lex_inpat);
1581 SAVEI32(PL_lex_inwhat);
1582 SAVECOPLINE(PL_curcop);
1583 SAVEPPTR(PL_bufptr);
1584 SAVEPPTR(PL_bufend);
1585 SAVEPPTR(PL_oldbufptr);
1586 SAVEPPTR(PL_oldoldbufptr);
1587 SAVEPPTR(PL_last_lop);
1588 SAVEPPTR(PL_last_uni);
1589 SAVEPPTR(PL_linestart);
1590 SAVESPTR(PL_linestr);
1591 SAVEGENERICPV(PL_lex_brackstack);
1592 SAVEGENERICPV(PL_lex_casestack);
1594 PL_linestr = PL_lex_stuff;
1595 PL_lex_stuff = NULL;
1597 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1598 = SvPVX(PL_linestr);
1599 PL_bufend += SvCUR(PL_linestr);
1600 PL_last_lop = PL_last_uni = NULL;
1601 SAVEFREESV(PL_linestr);
1603 PL_lex_dojoin = FALSE;
1604 PL_lex_brackets = 0;
1605 Newx(PL_lex_brackstack, 120, char);
1606 Newx(PL_lex_casestack, 12, char);
1607 PL_lex_casemods = 0;
1608 *PL_lex_casestack = '\0';
1610 PL_lex_state = LEX_INTERPCONCAT;
1611 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
1613 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1614 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1615 PL_lex_inpat = PL_sublex_info.sub_op;
1617 PL_lex_inpat = NULL;
1624 * Restores lexer state after a S_sublex_push.
1631 if (!PL_lex_starts++) {
1632 SV * const sv = newSVpvs("");
1633 if (SvUTF8(PL_linestr))
1635 PL_expect = XOPERATOR;
1636 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1640 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1641 PL_lex_state = LEX_INTERPCASEMOD;
1645 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1646 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1647 PL_linestr = PL_lex_repl;
1649 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1650 PL_bufend += SvCUR(PL_linestr);
1651 PL_last_lop = PL_last_uni = NULL;
1652 SAVEFREESV(PL_linestr);
1653 PL_lex_dojoin = FALSE;
1654 PL_lex_brackets = 0;
1655 PL_lex_casemods = 0;
1656 *PL_lex_casestack = '\0';
1658 if (SvEVALED(PL_lex_repl)) {
1659 PL_lex_state = LEX_INTERPNORMAL;
1661 /* we don't clear PL_lex_repl here, so that we can check later
1662 whether this is an evalled subst; that means we rely on the
1663 logic to ensure sublex_done() is called again only via the
1664 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1667 PL_lex_state = LEX_INTERPCONCAT;
1677 PL_endwhite = newSVpvn("",0);
1678 sv_catsv(PL_endwhite, PL_thiswhite);
1682 sv_setpvn(PL_thistoken,"",0);
1684 PL_realtokenstart = -1;
1688 PL_bufend = SvPVX(PL_linestr);
1689 PL_bufend += SvCUR(PL_linestr);
1690 PL_expect = XOPERATOR;
1691 PL_sublex_info.sub_inwhat = 0;
1699 Extracts a pattern, double-quoted string, or transliteration. This
1702 It looks at PL_lex_inwhat and PL_lex_inpat to find out whether it's
1703 processing a pattern (PL_lex_inpat is true), a transliteration
1704 (PL_lex_inwhat == OP_TRANS is true), or a double-quoted string.
1706 Returns a pointer to the character scanned up to. If this is
1707 advanced from the start pointer supplied (i.e. if anything was
1708 successfully parsed), will leave an OP for the substring scanned
1709 in yylval. Caller must intuit reason for not parsing further
1710 by looking at the next characters herself.
1714 double-quoted style: \r and \n
1715 regexp special ones: \D \s
1718 case and quoting: \U \Q \E
1719 stops on @ and $, but not for $ as tail anchor
1721 In transliterations:
1722 characters are VERY literal, except for - not at the start or end
1723 of the string, which indicates a range. If the range is in bytes,
1724 scan_const expands the range to the full set of intermediate
1725 characters. If the range is in utf8, the hyphen is replaced with
1726 a certain range mark which will be handled by pmtrans() in op.c.
1728 In double-quoted strings:
1730 double-quoted style: \r and \n
1732 deprecated backrefs: \1 (in substitution replacements)
1733 case and quoting: \U \Q \E
1736 scan_const does *not* construct ops to handle interpolated strings.
1737 It stops processing as soon as it finds an embedded $ or @ variable
1738 and leaves it to the caller to work out what's going on.
1740 embedded arrays (whether in pattern or not) could be:
1741 @foo, @::foo, @'foo, @{foo}, @$foo, @+, @-.
1743 $ in double-quoted strings must be the symbol of an embedded scalar.
1745 $ in pattern could be $foo or could be tail anchor. Assumption:
1746 it's a tail anchor if $ is the last thing in the string, or if it's
1747 followed by one of "()| \r\n\t"
1749 \1 (backreferences) are turned into $1
1751 The structure of the code is
1752 while (there's a character to process) {
1753 handle transliteration ranges
1754 skip regexp comments /(?#comment)/ and codes /(?{code})/
1755 skip #-initiated comments in //x patterns
1756 check for embedded arrays
1757 check for embedded scalars
1759 leave intact backslashes from leaveit (below)
1760 deprecate \1 in substitution replacements
1761 handle string-changing backslashes \l \U \Q \E, etc.
1762 switch (what was escaped) {
1763 handle \- in a transliteration (becomes a literal -)
1764 handle \132 (octal characters)
1765 handle \x15 and \x{1234} (hex characters)
1766 handle \N{name} (named characters)
1767 handle \cV (control characters)
1768 handle printf-style backslashes (\f, \r, \n, etc)
1770 } (end if backslash)
1771 } (end while character to read)
1776 S_scan_const(pTHX_ char *start)
1779 register char *send = PL_bufend; /* end of the constant */
1780 SV *sv = newSV(send - start); /* sv for the constant */
1781 register char *s = start; /* start of the constant */
1782 register char *d = SvPVX(sv); /* destination for copies */
1783 bool dorange = FALSE; /* are we in a translit range? */
1784 bool didrange = FALSE; /* did we just finish a range? */
1785 I32 has_utf8 = FALSE; /* Output constant is UTF8 */
1786 I32 this_utf8 = UTF; /* The source string is assumed to be UTF8 */
1789 UV literal_endpoint = 0;
1790 bool native_range = TRUE; /* turned to FALSE if the first endpoint is Unicode. */
1793 const char * const leaveit = /* set of acceptably-backslashed characters */
1796 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
1799 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1800 /* If we are doing a trans and we know we want UTF8 set expectation */
1801 has_utf8 = PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF);
1802 this_utf8 = PL_sublex_info.sub_op->op_private & (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1806 while (s < send || dorange) {
1807 /* get transliterations out of the way (they're most literal) */
1808 if (PL_lex_inwhat == OP_TRANS) {
1809 /* expand a range A-Z to the full set of characters. AIE! */
1811 I32 i; /* current expanded character */
1812 I32 min; /* first character in range */
1813 I32 max; /* last character in range */
1824 char * const c = (char*)utf8_hop((U8*)d, -1);
1828 *c = (char)UTF_TO_NATIVE(0xff);
1829 /* mark the range as done, and continue */
1835 i = d - SvPVX_const(sv); /* remember current offset */
1838 SvLEN(sv) + (has_utf8 ?
1839 (512 - UTF_CONTINUATION_MARK +
1842 /* How many two-byte within 0..255: 128 in UTF-8,
1843 * 96 in UTF-8-mod. */
1845 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1847 d = SvPVX(sv) + i; /* refresh d after realloc */
1851 for (j = 0; j <= 1; j++) {
1852 char * const c = (char*)utf8_hop((U8*)d, -1);
1853 const UV uv = utf8n_to_uvchr((U8*)c, d - c, NULL, 0);
1859 max = (U8)0xff; /* only to \xff */
1860 uvmax = uv; /* \x{100} to uvmax */
1862 d = c; /* eat endpoint chars */
1867 d -= 2; /* eat the first char and the - */
1868 min = (U8)*d; /* first char in range */
1869 max = (U8)d[1]; /* last char in range */
1876 "Invalid range \"%c-%c\" in transliteration operator",
1877 (char)min, (char)max);
1881 if (literal_endpoint == 2 &&
1882 ((isLOWER(min) && isLOWER(max)) ||
1883 (isUPPER(min) && isUPPER(max)))) {
1885 for (i = min; i <= max; i++)
1887 *d++ = NATIVE_TO_NEED(has_utf8,i);
1889 for (i = min; i <= max; i++)
1891 *d++ = NATIVE_TO_NEED(has_utf8,i);
1896 for (i = min; i <= max; i++)
1899 const U8 ch = (U8)NATIVE_TO_UTF(i);
1900 if (UNI_IS_INVARIANT(ch))
1903 *d++ = (U8)UTF8_EIGHT_BIT_HI(ch);
1904 *d++ = (U8)UTF8_EIGHT_BIT_LO(ch);
1913 d = (char*)uvchr_to_utf8((U8*)d, 0x100);
1915 *d++ = (char)UTF_TO_NATIVE(0xff);
1917 d = (char*)uvchr_to_utf8((U8*)d, uvmax);
1921 /* mark the range as done, and continue */
1925 literal_endpoint = 0;
1930 /* range begins (ignore - as first or last char) */
1931 else if (*s == '-' && s+1 < send && s != start) {
1933 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
1940 *d++ = (char)UTF_TO_NATIVE(0xff); /* use illegal utf8 byte--see pmtrans */
1950 literal_endpoint = 0;
1951 native_range = TRUE;
1956 /* if we get here, we're not doing a transliteration */
1958 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1959 except for the last char, which will be done separately. */
1960 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1962 while (s+1 < send && *s != ')')
1963 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1965 else if (s[2] == '{' /* This should match regcomp.c */
1966 || ((s[2] == 'p' || s[2] == '?') && s[3] == '{'))
1969 char *regparse = s + (s[2] == '{' ? 3 : 4);
1972 while (count && (c = *regparse)) {
1973 if (c == '\\' && regparse[1])
1981 if (*regparse != ')')
1982 regparse--; /* Leave one char for continuation. */
1983 while (s < regparse)
1984 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1988 /* likewise skip #-initiated comments in //x patterns */
1989 else if (*s == '#' && PL_lex_inpat &&
1990 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1991 while (s+1 < send && *s != '\n')
1992 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1995 /* check for embedded arrays
1996 (@foo, @::foo, @'foo, @{foo}, @$foo, @+, @-)
1998 else if (*s == '@' && s[1]) {
1999 if (isALNUM_lazy_if(s+1,UTF))
2001 if (strchr(":'{$", s[1]))
2003 if (!PL_lex_inpat && (s[1] == '+' || s[1] == '-'))
2004 break; /* in regexp, neither @+ nor @- are interpolated */
2007 /* check for embedded scalars. only stop if we're sure it's a
2010 else if (*s == '$') {
2011 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
2013 if (s + 1 < send && !strchr("()| \r\n\t", s[1]))
2014 break; /* in regexp, $ might be tail anchor */
2017 /* End of else if chain - OP_TRANS rejoin rest */
2020 if (*s == '\\' && s+1 < send) {
2023 /* some backslashes we leave behind */
2024 if (*leaveit && *s && strchr(leaveit, *s)) {
2025 *d++ = NATIVE_TO_NEED(has_utf8,'\\');
2026 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
2030 /* deprecate \1 in strings and substitution replacements */
2031 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
2032 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
2034 if (ckWARN(WARN_SYNTAX))
2035 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "\\%c better written as $%c", *s, *s);
2040 /* string-change backslash escapes */
2041 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
2046 /* if we get here, it's either a quoted -, or a digit */
2049 /* quoted - in transliterations */
2051 if (PL_lex_inwhat == OP_TRANS) {
2058 if ((isALPHA(*s) || isDIGIT(*s)) &&
2060 Perl_warner(aTHX_ packWARN(WARN_MISC),
2061 "Unrecognized escape \\%c passed through",
2063 /* default action is to copy the quoted character */
2064 goto default_action;
2067 /* \132 indicates an octal constant */
2068 case '0': case '1': case '2': case '3':
2069 case '4': case '5': case '6': case '7':
2073 uv = grok_oct(s, &len, &flags, NULL);
2076 goto NUM_ESCAPE_INSERT;
2078 /* \x24 indicates a hex constant */
2082 char* const e = strchr(s, '}');
2083 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
2084 PERL_SCAN_DISALLOW_PREFIX;
2089 yyerror("Missing right brace on \\x{}");
2093 uv = grok_hex(s, &len, &flags, NULL);
2099 I32 flags = PERL_SCAN_DISALLOW_PREFIX;
2100 uv = grok_hex(s, &len, &flags, NULL);
2106 /* Insert oct or hex escaped character.
2107 * There will always enough room in sv since such
2108 * escapes will be longer than any UTF-8 sequence
2109 * they can end up as. */
2111 /* We need to map to chars to ASCII before doing the tests
2114 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(uv))) {
2115 if (!has_utf8 && uv > 255) {
2116 /* Might need to recode whatever we have
2117 * accumulated so far if it contains any
2120 * (Can't we keep track of that and avoid
2121 * this rescan? --jhi)
2125 for (c = (U8 *) SvPVX(sv); c < (U8 *)d; c++) {
2126 if (!NATIVE_IS_INVARIANT(*c)) {
2131 const STRLEN offset = d - SvPVX_const(sv);
2133 d = SvGROW(sv, SvLEN(sv) + hicount + 1) + offset;
2137 while (src >= (const U8 *)SvPVX_const(sv)) {
2138 if (!NATIVE_IS_INVARIANT(*src)) {
2139 const U8 ch = NATIVE_TO_ASCII(*src);
2140 *dst-- = (U8)UTF8_EIGHT_BIT_LO(ch);
2141 *dst-- = (U8)UTF8_EIGHT_BIT_HI(ch);
2151 if (has_utf8 || uv > 255) {
2152 d = (char*)uvchr_to_utf8((U8*)d, uv);
2154 if (PL_lex_inwhat == OP_TRANS &&
2155 PL_sublex_info.sub_op) {
2156 PL_sublex_info.sub_op->op_private |=
2157 (PL_lex_repl ? OPpTRANS_FROM_UTF
2161 if (uv > 255 && !dorange)
2162 native_range = FALSE;
2174 /* \N{LATIN SMALL LETTER A} is a named character */
2178 char* e = strchr(s, '}');
2184 yyerror("Missing right brace on \\N{}");
2188 if (e > s + 2 && s[1] == 'U' && s[2] == '+') {
2190 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
2191 PERL_SCAN_DISALLOW_PREFIX;
2194 uv = grok_hex(s, &len, &flags, NULL);
2196 goto NUM_ESCAPE_INSERT;
2198 res = newSVpvn(s + 1, e - s - 1);
2199 res = new_constant( NULL, 0, "charnames",
2200 res, NULL, "\\N{...}" );
2202 sv_utf8_upgrade(res);
2203 str = SvPV_const(res,len);
2204 #ifdef EBCDIC_NEVER_MIND
2205 /* charnames uses pack U and that has been
2206 * recently changed to do the below uni->native
2207 * mapping, so this would be redundant (and wrong,
2208 * the code point would be doubly converted).
2209 * But leave this in just in case the pack U change
2210 * gets revoked, but the semantics is still
2211 * desireable for charnames. --jhi */
2213 UV uv = utf8_to_uvchr((const U8*)str, 0);
2216 U8 tmpbuf[UTF8_MAXBYTES+1], *d;
2218 d = uvchr_to_utf8(tmpbuf, UNI_TO_NATIVE(uv));
2219 sv_setpvn(res, (char *)tmpbuf, d - tmpbuf);
2220 str = SvPV_const(res, len);
2224 if (!has_utf8 && SvUTF8(res)) {
2225 const char * const ostart = SvPVX_const(sv);
2226 SvCUR_set(sv, d - ostart);
2229 sv_utf8_upgrade(sv);
2230 /* this just broke our allocation above... */
2231 SvGROW(sv, (STRLEN)(send - start));
2232 d = SvPVX(sv) + SvCUR(sv);
2235 if (len > (STRLEN)(e - s + 4)) { /* I _guess_ 4 is \N{} --jhi */
2236 const char * const odest = SvPVX_const(sv);
2238 SvGROW(sv, (SvLEN(sv) + len - (e - s + 4)));
2239 d = SvPVX(sv) + (d - odest);
2243 native_range = FALSE; /* \N{} is guessed to be Unicode */
2245 Copy(str, d, len, char);
2252 yyerror("Missing braces on \\N{}");
2255 /* \c is a control character */
2264 *d++ = NATIVE_TO_NEED(has_utf8,toCTRL(c));
2267 yyerror("Missing control char name in \\c");
2271 /* printf-style backslashes, formfeeds, newlines, etc */
2273 *d++ = NATIVE_TO_NEED(has_utf8,'\b');
2276 *d++ = NATIVE_TO_NEED(has_utf8,'\n');
2279 *d++ = NATIVE_TO_NEED(has_utf8,'\r');
2282 *d++ = NATIVE_TO_NEED(has_utf8,'\f');
2285 *d++ = NATIVE_TO_NEED(has_utf8,'\t');
2288 *d++ = ASCII_TO_NEED(has_utf8,'\033');
2291 *d++ = ASCII_TO_NEED(has_utf8,'\007');
2297 } /* end if (backslash) */
2304 /* If we started with encoded form, or already know we want it
2305 and then encode the next character */
2306 if ((has_utf8 || this_utf8) && !NATIVE_IS_INVARIANT((U8)(*s))) {
2308 const UV nextuv = (this_utf8) ? utf8n_to_uvchr((U8*)s, send - s, &len, 0) : (UV) ((U8) *s);
2309 const STRLEN need = UNISKIP(NATIVE_TO_UNI(nextuv));
2312 /* encoded value larger than old, need extra space (NOTE: SvCUR() not set here) */
2313 const STRLEN off = d - SvPVX_const(sv);
2314 d = SvGROW(sv, SvLEN(sv) + (need-len)) + off;
2316 d = (char*)uvchr_to_utf8((U8*)d, nextuv);
2319 if (uv > 255 && !dorange)
2320 native_range = FALSE;
2324 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
2326 } /* while loop to process each character */
2328 /* terminate the string and set up the sv */
2330 SvCUR_set(sv, d - SvPVX_const(sv));
2331 if (SvCUR(sv) >= SvLEN(sv))
2332 Perl_croak(aTHX_ "panic: constant overflowed allocated space");
2335 if (PL_encoding && !has_utf8) {
2336 sv_recode_to_utf8(sv, PL_encoding);
2342 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
2343 PL_sublex_info.sub_op->op_private |=
2344 (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
2348 /* shrink the sv if we allocated more than we used */
2349 if (SvCUR(sv) + 5 < SvLEN(sv)) {
2350 SvPV_shrink_to_cur(sv);
2353 /* return the substring (via yylval) only if we parsed anything */
2354 if (s > PL_bufptr) {
2355 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
2356 sv = new_constant(start, s - start,
2357 (const char *)(PL_lex_inpat ? "qr" : "q"),
2360 (( PL_lex_inwhat == OP_TRANS
2362 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
2365 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2372 * Returns TRUE if there's more to the expression (e.g., a subscript),
2375 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
2377 * ->[ and ->{ return TRUE
2378 * { and [ outside a pattern are always subscripts, so return TRUE
2379 * if we're outside a pattern and it's not { or [, then return FALSE
2380 * if we're in a pattern and the first char is a {
2381 * {4,5} (any digits around the comma) returns FALSE
2382 * if we're in a pattern and the first char is a [
2384 * [SOMETHING] has a funky algorithm to decide whether it's a
2385 * character class or not. It has to deal with things like
2386 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
2387 * anything else returns TRUE
2390 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
2393 S_intuit_more(pTHX_ register char *s)
2396 if (PL_lex_brackets)
2398 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
2400 if (*s != '{' && *s != '[')
2405 /* In a pattern, so maybe we have {n,m}. */
2422 /* On the other hand, maybe we have a character class */
2425 if (*s == ']' || *s == '^')
2428 /* this is terrifying, and it works */
2429 int weight = 2; /* let's weigh the evidence */
2431 unsigned char un_char = 255, last_un_char;
2432 const char * const send = strchr(s,']');
2433 char tmpbuf[sizeof PL_tokenbuf * 4];
2435 if (!send) /* has to be an expression */
2438 Zero(seen,256,char);
2441 else if (isDIGIT(*s)) {
2443 if (isDIGIT(s[1]) && s[2] == ']')
2449 for (; s < send; s++) {
2450 last_un_char = un_char;
2451 un_char = (unsigned char)*s;
2456 weight -= seen[un_char] * 10;
2457 if (isALNUM_lazy_if(s+1,UTF)) {
2459 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
2460 len = (int)strlen(tmpbuf);
2461 if (len > 1 && gv_fetchpvn_flags(tmpbuf, len, 0, SVt_PV))
2466 else if (*s == '$' && s[1] &&
2467 strchr("[#!%*<>()-=",s[1])) {
2468 if (/*{*/ strchr("])} =",s[2]))
2477 if (strchr("wds]",s[1]))
2479 else if (seen[(U8)'\''] || seen[(U8)'"'])
2481 else if (strchr("rnftbxcav",s[1]))
2483 else if (isDIGIT(s[1])) {
2485 while (s[1] && isDIGIT(s[1]))
2495 if (strchr("aA01! ",last_un_char))
2497 if (strchr("zZ79~",s[1]))
2499 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
2500 weight -= 5; /* cope with negative subscript */
2503 if (!isALNUM(last_un_char)
2504 && !(last_un_char == '$' || last_un_char == '@'
2505 || last_un_char == '&')
2506 && isALPHA(*s) && s[1] && isALPHA(s[1])) {
2511 if (keyword(tmpbuf, d - tmpbuf))
2514 if (un_char == last_un_char + 1)
2516 weight -= seen[un_char];
2521 if (weight >= 0) /* probably a character class */
2531 * Does all the checking to disambiguate
2533 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
2534 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
2536 * First argument is the stuff after the first token, e.g. "bar".
2538 * Not a method if bar is a filehandle.
2539 * Not a method if foo is a subroutine prototyped to take a filehandle.
2540 * Not a method if it's really "Foo $bar"
2541 * Method if it's "foo $bar"
2542 * Not a method if it's really "print foo $bar"
2543 * Method if it's really "foo package::" (interpreted as package->foo)
2544 * Not a method if bar is known to be a subroutine ("sub bar; foo bar")
2545 * Not a method if bar is a filehandle or package, but is quoted with
2550 S_intuit_method(pTHX_ char *start, GV *gv, CV *cv)
2553 char *s = start + (*start == '$');
2554 char tmpbuf[sizeof PL_tokenbuf];
2562 if (SvTYPE(gv) == SVt_PVGV && GvIO(gv))
2566 const char *proto = SvPVX_const(cv);
2577 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
2578 /* start is the beginning of the possible filehandle/object,
2579 * and s is the end of it
2580 * tmpbuf is a copy of it
2583 if (*start == '$') {
2584 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
2587 len = start - SvPVX(PL_linestr);
2591 start = SvPVX(PL_linestr) + len;
2595 return *s == '(' ? FUNCMETH : METHOD;
2597 if (!keyword(tmpbuf, len)) {
2598 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
2602 soff = s - SvPVX(PL_linestr);
2606 indirgv = gv_fetchpvn_flags(tmpbuf, len, 0, SVt_PVCV);
2607 if (indirgv && GvCVu(indirgv))
2609 /* filehandle or package name makes it a method */
2610 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
2612 soff = s - SvPVX(PL_linestr);
2615 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
2616 return 0; /* no assumptions -- "=>" quotes bearword */
2618 start_force(PL_curforce);
2619 NEXTVAL_NEXTTOKE.opval = (OP*)newSVOP(OP_CONST, 0,
2620 newSVpvn(tmpbuf,len));
2621 NEXTVAL_NEXTTOKE.opval->op_private = OPpCONST_BARE;
2623 curmad('X', newSVpvn(start,SvPVX(PL_linestr) + soff - start));
2628 PL_bufptr = SvPVX(PL_linestr) + soff; /* restart before space */
2630 return *s == '(' ? FUNCMETH : METHOD;
2638 * Return a string of Perl code to load the debugger. If PERL5DB
2639 * is set, it will return the contents of that, otherwise a
2640 * compile-time require of perl5db.pl.
2648 const char * const pdb = PerlEnv_getenv("PERL5DB");
2652 SETERRNO(0,SS_NORMAL);
2653 return "BEGIN { require 'perl5db.pl' }";
2659 /* Encoded script support. filter_add() effectively inserts a
2660 * 'pre-processing' function into the current source input stream.
2661 * Note that the filter function only applies to the current source file
2662 * (e.g., it will not affect files 'require'd or 'use'd by this one).
2664 * The datasv parameter (which may be NULL) can be used to pass
2665 * private data to this instance of the filter. The filter function
2666 * can recover the SV using the FILTER_DATA macro and use it to
2667 * store private buffers and state information.
2669 * The supplied datasv parameter is upgraded to a PVIO type
2670 * and the IoDIRP/IoANY field is used to store the function pointer,
2671 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
2672 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
2673 * private use must be set using malloc'd pointers.
2677 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
2683 if (!PL_rsfp_filters)
2684 PL_rsfp_filters = newAV();
2687 SvUPGRADE(datasv, SVt_PVIO);
2688 IoANY(datasv) = FPTR2DPTR(void *, funcp); /* stash funcp into spare field */
2689 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
2690 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
2691 FPTR2DPTR(void *, IoANY(datasv)),
2692 SvPV_nolen(datasv)));
2693 av_unshift(PL_rsfp_filters, 1);
2694 av_store(PL_rsfp_filters, 0, datasv) ;
2699 /* Delete most recently added instance of this filter function. */
2701 Perl_filter_del(pTHX_ filter_t funcp)
2707 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p",
2708 FPTR2DPTR(void*, funcp)));
2710 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
2712 /* if filter is on top of stack (usual case) just pop it off */
2713 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
2714 if (IoANY(datasv) == FPTR2DPTR(void *, funcp)) {
2715 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
2716 IoANY(datasv) = (void *)NULL;
2717 sv_free(av_pop(PL_rsfp_filters));
2721 /* we need to search for the correct entry and clear it */
2722 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
2726 /* Invoke the idxth filter function for the current rsfp. */
2727 /* maxlen 0 = read one text line */
2729 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
2734 /* This API is bad. It should have been using unsigned int for maxlen.
2735 Not sure if we want to change the API, but if not we should sanity
2736 check the value here. */
2737 const unsigned int correct_length
2746 if (!PL_rsfp_filters)
2748 if (idx > AvFILLp(PL_rsfp_filters)) { /* Any more filters? */
2749 /* Provide a default input filter to make life easy. */
2750 /* Note that we append to the line. This is handy. */
2751 DEBUG_P(PerlIO_printf(Perl_debug_log,
2752 "filter_read %d: from rsfp\n", idx));
2753 if (correct_length) {
2756 const int old_len = SvCUR(buf_sv);
2758 /* ensure buf_sv is large enough */
2759 SvGROW(buf_sv, (STRLEN)(old_len + correct_length)) ;
2760 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len,
2761 correct_length)) <= 0) {
2762 if (PerlIO_error(PL_rsfp))
2763 return -1; /* error */
2765 return 0 ; /* end of file */
2767 SvCUR_set(buf_sv, old_len + len) ;
2770 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
2771 if (PerlIO_error(PL_rsfp))
2772 return -1; /* error */
2774 return 0 ; /* end of file */
2777 return SvCUR(buf_sv);
2779 /* Skip this filter slot if filter has been deleted */
2780 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef) {
2781 DEBUG_P(PerlIO_printf(Perl_debug_log,
2782 "filter_read %d: skipped (filter deleted)\n",
2784 return FILTER_READ(idx+1, buf_sv, correct_length); /* recurse */
2786 /* Get function pointer hidden within datasv */
2787 funcp = DPTR2FPTR(filter_t, IoANY(datasv));
2788 DEBUG_P(PerlIO_printf(Perl_debug_log,
2789 "filter_read %d: via function %p (%s)\n",
2790 idx, (void*)datasv, SvPV_nolen_const(datasv)));
2791 /* Call function. The function is expected to */
2792 /* call "FILTER_READ(idx+1, buf_sv)" first. */
2793 /* Return: <0:error, =0:eof, >0:not eof */
2794 return (*funcp)(aTHX_ idx, buf_sv, correct_length);
2798 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
2801 #ifdef PERL_CR_FILTER
2802 if (!PL_rsfp_filters) {
2803 filter_add(S_cr_textfilter,NULL);
2806 if (PL_rsfp_filters) {
2808 SvCUR_set(sv, 0); /* start with empty line */
2809 if (FILTER_READ(0, sv, 0) > 0)
2810 return ( SvPVX(sv) ) ;
2815 return (sv_gets(sv, fp, append));
2819 S_find_in_my_stash(pTHX_ const char *pkgname, I32 len)
2824 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2828 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2829 (gv = gv_fetchpvn_flags(pkgname, len, 0, SVt_PVHV)))
2831 return GvHV(gv); /* Foo:: */
2834 /* use constant CLASS => 'MyClass' */
2835 gv = gv_fetchpvn_flags(pkgname, len, 0, SVt_PVCV);
2836 if (gv && GvCV(gv)) {
2837 SV * const sv = cv_const_sv(GvCV(gv));
2839 pkgname = SvPV_nolen_const(sv);
2842 return gv_stashpv(pkgname, FALSE);
2848 * The intent of this yylex wrapper is to minimize the changes to the
2849 * tokener when we aren't interested in collecting madprops. It remains
2850 * to be seen how successful this strategy will be...
2857 char *s = PL_bufptr;
2859 /* make sure PL_thiswhite is initialized */
2863 /* just do what yylex would do on pending identifier; leave PL_thiswhite alone */
2864 if (PL_pending_ident)
2865 return S_pending_ident(aTHX);
2867 /* previous token ate up our whitespace? */
2868 if (!PL_lasttoke && PL_nextwhite) {
2869 PL_thiswhite = PL_nextwhite;
2873 /* isolate the token, and figure out where it is without whitespace */
2874 PL_realtokenstart = -1;
2878 assert(PL_curforce < 0);
2880 if (!PL_thismad || PL_thismad->mad_key == '^') { /* not forced already? */
2881 if (!PL_thistoken) {
2882 if (PL_realtokenstart < 0 || !CopLINE(PL_curcop))
2883 PL_thistoken = newSVpvn("",0);
2885 char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
2886 PL_thistoken = newSVpvn(tstart, s - tstart);
2889 if (PL_thismad) /* install head */
2890 CURMAD('X', PL_thistoken);
2893 /* last whitespace of a sublex? */
2894 if (optype == ')' && PL_endwhite) {
2895 CURMAD('X', PL_endwhite);
2900 /* if no whitespace and we're at EOF, bail. Otherwise fake EOF below. */
2901 if (!PL_thiswhite && !PL_endwhite && !optype) {
2902 sv_free(PL_thistoken);
2907 /* put off final whitespace till peg */
2908 if (optype == ';' && !PL_rsfp) {
2909 PL_nextwhite = PL_thiswhite;
2912 else if (PL_thisopen) {
2913 CURMAD('q', PL_thisopen);
2915 sv_free(PL_thistoken);
2919 /* Store actual token text as madprop X */
2920 CURMAD('X', PL_thistoken);
2924 /* add preceding whitespace as madprop _ */
2925 CURMAD('_', PL_thiswhite);
2929 /* add quoted material as madprop = */
2930 CURMAD('=', PL_thisstuff);
2934 /* add terminating quote as madprop Q */
2935 CURMAD('Q', PL_thisclose);
2939 /* special processing based on optype */
2943 /* opval doesn't need a TOKEN since it can already store mp */
2954 append_madprops(PL_thismad, yylval.opval, 0);
2962 addmad(newMADsv('p', PL_endwhite), &PL_thismad, 0);
2971 /* remember any fake bracket that lexer is about to discard */
2972 if (PL_lex_brackets == 1 &&
2973 ((expectation)PL_lex_brackstack[0] & XFAKEBRACK))
2976 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
2979 PL_thiswhite = newSVpvn(PL_bufptr, ++s - PL_bufptr);
2980 addmad(newMADsv('#', PL_thiswhite), &PL_thismad, 0);
2983 break; /* don't bother looking for trailing comment */
2992 /* attach a trailing comment to its statement instead of next token */
2996 if (PL_bufptr > PL_oldbufptr && PL_bufptr[-1] == optype) {
2998 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
3000 if (*s == '\n' || *s == '#') {
3001 while (s < PL_bufend && *s != '\n')
3005 PL_thiswhite = newSVpvn(PL_bufptr, s - PL_bufptr);
3006 addmad(newMADsv('#', PL_thiswhite), &PL_thismad, 0);
3023 /* Create new token struct. Note: opvals return early above. */
3024 yylval.tkval = newTOKEN(optype, yylval, PL_thismad);
3031 S_tokenize_use(pTHX_ int is_use, char *s) {
3033 if (PL_expect != XSTATE)
3034 yyerror(Perl_form(aTHX_ "\"%s\" not allowed in expression",
3035 is_use ? "use" : "no"));
3037 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
3038 s = force_version(s, TRUE);
3039 if (*s == ';' || (s = SKIPSPACE1(s), *s == ';')) {
3040 start_force(PL_curforce);
3041 NEXTVAL_NEXTTOKE.opval = NULL;
3044 else if (*s == 'v') {
3045 s = force_word(s,WORD,FALSE,TRUE,FALSE);
3046 s = force_version(s, FALSE);
3050 s = force_word(s,WORD,FALSE,TRUE,FALSE);
3051 s = force_version(s, FALSE);
3053 yylval.ival = is_use;
3057 static const char* const exp_name[] =
3058 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
3059 "ATTRTERM", "TERMBLOCK", "TERMORDORDOR"
3066 Works out what to call the token just pulled out of the input
3067 stream. The yacc parser takes care of taking the ops we return and
3068 stitching them into a tree.
3074 if read an identifier
3075 if we're in a my declaration
3076 croak if they tried to say my($foo::bar)
3077 build the ops for a my() declaration
3078 if it's an access to a my() variable
3079 are we in a sort block?
3080 croak if my($a); $a <=> $b
3081 build ops for access to a my() variable
3082 if in a dq string, and they've said @foo and we can't find @foo
3084 build ops for a bareword
3085 if we already built the token before, use it.
3090 #pragma segment Perl_yylex
3096 register char *s = PL_bufptr;
3101 /* orig_keyword, gvp, and gv are initialized here because
3102 * jump to the label just_a_word_zero can bypass their
3103 * initialization later. */
3104 I32 orig_keyword = 0;
3109 SV* tmp = newSVpvs("");
3110 PerlIO_printf(Perl_debug_log, "### %"IVdf":LEX_%s/X%s %s\n",
3111 (IV)CopLINE(PL_curcop),
3112 lex_state_names[PL_lex_state],
3113 exp_name[PL_expect],
3114 pv_display(tmp, s, strlen(s), 0, 60));
3117 /* check if there's an identifier for us to look at */
3118 if (PL_pending_ident)
3119 return REPORT(S_pending_ident(aTHX));
3121 /* no identifier pending identification */
3123 switch (PL_lex_state) {
3125 case LEX_NORMAL: /* Some compilers will produce faster */
3126 case LEX_INTERPNORMAL: /* code if we comment these out. */
3130 /* when we've already built the next token, just pull it out of the queue */
3134 yylval = PL_nexttoke[PL_lasttoke].next_val;
3136 PL_thismad = PL_nexttoke[PL_lasttoke].next_mad;
3137 PL_nexttoke[PL_lasttoke].next_mad = 0;
3138 if (PL_thismad && PL_thismad->mad_key == '_') {
3139 PL_thiswhite = (SV*)PL_thismad->mad_val;
3140 PL_thismad->mad_val = 0;
3141 mad_free(PL_thismad);
3146 PL_lex_state = PL_lex_defer;
3147 PL_expect = PL_lex_expect;
3148 PL_lex_defer = LEX_NORMAL;
3149 if (!PL_nexttoke[PL_lasttoke].next_type)
3154 yylval = PL_nextval[PL_nexttoke];
3156 PL_lex_state = PL_lex_defer;
3157 PL_expect = PL_lex_expect;
3158 PL_lex_defer = LEX_NORMAL;
3162 /* FIXME - can these be merged? */
3163 return(PL_nexttoke[PL_lasttoke].next_type);
3165 return REPORT(PL_nexttype[PL_nexttoke]);
3168 /* interpolated case modifiers like \L \U, including \Q and \E.
3169 when we get here, PL_bufptr is at the \
3171 case LEX_INTERPCASEMOD:
3173 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
3174 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
3176 /* handle \E or end of string */
3177 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
3179 if (PL_lex_casemods) {
3180 const char oldmod = PL_lex_casestack[--PL_lex_casemods];
3181 PL_lex_casestack[PL_lex_casemods] = '\0';
3183 if (PL_bufptr != PL_bufend
3184 && (oldmod == 'L' || oldmod == 'U' || oldmod == 'Q')) {
3186 PL_lex_state = LEX_INTERPCONCAT;
3189 PL_thistoken = newSVpvn("\\E",2);
3195 while (PL_bufptr != PL_bufend &&
3196 PL_bufptr[0] == '\\' && PL_bufptr[1] == 'E') {
3198 PL_thiswhite = newSVpvn("",0);
3199 sv_catpvn(PL_thiswhite, PL_bufptr, 2);
3203 if (PL_bufptr != PL_bufend)
3206 PL_lex_state = LEX_INTERPCONCAT;
3210 DEBUG_T({ PerlIO_printf(Perl_debug_log,
3211 "### Saw case modifier\n"); });
3213 if (s[1] == '\\' && s[2] == 'E') {
3216 PL_thiswhite = newSVpvn("",0);
3217 sv_catpvn(PL_thiswhite, PL_bufptr, 4);
3220 PL_lex_state = LEX_INTERPCONCAT;
3225 if (!PL_madskills) /* when just compiling don't need correct */
3226 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
3227 tmp = *s, *s = s[2], s[2] = (char)tmp; /* misordered... */
3228 if ((*s == 'L' || *s == 'U') &&
3229 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U'))) {
3230 PL_lex_casestack[--PL_lex_casemods] = '\0';
3233 if (PL_lex_casemods > 10)
3234 Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
3235 PL_lex_casestack[PL_lex_casemods++] = *s;
3236 PL_lex_casestack[PL_lex_casemods] = '\0';
3237 PL_lex_state = LEX_INTERPCONCAT;
3238 start_force(PL_curforce);
3239 NEXTVAL_NEXTTOKE.ival = 0;
3241 start_force(PL_curforce);
3243 NEXTVAL_NEXTTOKE.ival = OP_LCFIRST;
3245 NEXTVAL_NEXTTOKE.ival = OP_UCFIRST;
3247 NEXTVAL_NEXTTOKE.ival = OP_LC;
3249 NEXTVAL_NEXTTOKE.ival = OP_UC;
3251 NEXTVAL_NEXTTOKE.ival = OP_QUOTEMETA;
3253 Perl_croak(aTHX_ "panic: yylex");
3255 SV* const tmpsv = newSVpvn("",0);
3256 Perl_sv_catpvf(aTHX_ tmpsv, "\\%c", *s);
3262 if (PL_lex_starts) {
3268 sv_free(PL_thistoken);
3269 PL_thistoken = newSVpvn("",0);
3272 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
3273 if (PL_lex_casemods == 1 && PL_lex_inpat)
3282 case LEX_INTERPPUSH:
3283 return REPORT(sublex_push());
3285 case LEX_INTERPSTART:
3286 if (PL_bufptr == PL_bufend)
3287 return REPORT(sublex_done());
3288 DEBUG_T({ PerlIO_printf(Perl_debug_log,
3289 "### Interpolated variable\n"); });
3291 PL_lex_dojoin = (*PL_bufptr == '@');
3292 PL_lex_state = LEX_INTERPNORMAL;
3293 if (PL_lex_dojoin) {
3294 start_force(PL_curforce);
3295 NEXTVAL_NEXTTOKE.ival = 0;
3297 start_force(PL_curforce);
3298 force_ident("\"", '$');
3299 start_force(PL_curforce);
3300 NEXTVAL_NEXTTOKE.ival = 0;
3302 start_force(PL_curforce);
3303 NEXTVAL_NEXTTOKE.ival = 0;
3305 start_force(PL_curforce);
3306 NEXTVAL_NEXTTOKE.ival = OP_JOIN; /* emulate join($", ...) */
3309 if (PL_lex_starts++) {
3314 sv_free(PL_thistoken);
3315 PL_thistoken = newSVpvn("",0);
3318 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
3319 if (!PL_lex_casemods && PL_lex_inpat)
3326 case LEX_INTERPENDMAYBE:
3327 if (intuit_more(PL_bufptr)) {
3328 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
3334 if (PL_lex_dojoin) {
3335 PL_lex_dojoin = FALSE;
3336 PL_lex_state = LEX_INTERPCONCAT;
3340 sv_free(PL_thistoken);
3341 PL_thistoken = newSVpvn("",0);
3346 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
3347 && SvEVALED(PL_lex_repl))
3349 if (PL_bufptr != PL_bufend)
3350 Perl_croak(aTHX_ "Bad evalled substitution pattern");
3354 case LEX_INTERPCONCAT:
3356 if (PL_lex_brackets)
3357 Perl_croak(aTHX_ "panic: INTERPCONCAT");
3359 if (PL_bufptr == PL_bufend)
3360 return REPORT(sublex_done());
3362 if (SvIVX(PL_linestr) == '\'') {
3363 SV *sv = newSVsv(PL_linestr);
3366 else if ( PL_hints & HINT_NEW_RE )
3367 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
3368 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3372 s = scan_const(PL_bufptr);
3374 PL_lex_state = LEX_INTERPCASEMOD;
3376 PL_lex_state = LEX_INTERPSTART;
3379 if (s != PL_bufptr) {
3380 start_force(PL_curforce);
3382 curmad('X', newSVpvn(PL_bufptr,s-PL_bufptr));
3384 NEXTVAL_NEXTTOKE = yylval;
3387 if (PL_lex_starts++) {
3391 sv_free(PL_thistoken);
3392 PL_thistoken = newSVpvn("",0);
3395 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
3396 if (!PL_lex_casemods && PL_lex_inpat)
3409 PL_lex_state = LEX_NORMAL;
3410 s = scan_formline(PL_bufptr);
3411 if (!PL_lex_formbrack)
3417 PL_oldoldbufptr = PL_oldbufptr;
3423 sv_free(PL_thistoken);
3426 PL_realtokenstart = s - SvPVX(PL_linestr); /* assume but undo on ws */
3430 if (isIDFIRST_lazy_if(s,UTF))
3432 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
3435 goto fake_eof; /* emulate EOF on ^D or ^Z */
3444 if (PL_lex_brackets) {
3445 yyerror((const char *)
3447 ? "Format not terminated"
3448 : "Missing right curly or square bracket"));
3450 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3451 "### Tokener got EOF\n");
3455 if (s++ < PL_bufend)
3456 goto retry; /* ignore stray nulls */
3459 if (!PL_in_eval && !PL_preambled) {
3460 PL_preambled = TRUE;
3465 sv_setpv(PL_linestr,incl_perldb());
3466 if (SvCUR(PL_linestr))
3467 sv_catpvs(PL_linestr,";");
3469 while(AvFILLp(PL_preambleav) >= 0) {
3470 SV *tmpsv = av_shift(PL_preambleav);
3471 sv_catsv(PL_linestr, tmpsv);
3472 sv_catpvs(PL_linestr, ";");
3475 sv_free((SV*)PL_preambleav);
3476 PL_preambleav = NULL;
3478 if (PL_minus_n || PL_minus_p) {
3479 sv_catpvs(PL_linestr, "LINE: while (<>) {");
3481 sv_catpvs(PL_linestr,"chomp;");
3484 if ((*PL_splitstr == '/' || *PL_splitstr == '\''
3485 || *PL_splitstr == '"')
3486 && strchr(PL_splitstr + 1, *PL_splitstr))
3487 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s);", PL_splitstr);
3489 /* "q\0${splitstr}\0" is legal perl. Yes, even NUL
3490 bytes can be used as quoting characters. :-) */
3491 const char *splits = PL_splitstr;
3492 sv_catpvs(PL_linestr, "our @F=split(q\0");
3495 if (*splits == '\\')
3496 sv_catpvn(PL_linestr, splits, 1);
3497 sv_catpvn(PL_linestr, splits, 1);
3498 } while (*splits++);
3499 /* This loop will embed the trailing NUL of
3500 PL_linestr as the last thing it does before
3502 sv_catpvs(PL_linestr, ");");
3506 sv_catpvs(PL_linestr,"our @F=split(' ');");
3510 sv_catpvs(PL_linestr,"use feature ':5.10';");
3511 sv_catpvs(PL_linestr, "\n");
3512 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3513 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3514 PL_last_lop = PL_last_uni = NULL;
3515 if (PERLDB_LINE && PL_curstash != PL_debstash) {
3516 SV * const sv = newSV(0);
3518 sv_upgrade(sv, SVt_PVMG);
3519 sv_setsv(sv,PL_linestr);
3522 av_store(CopFILEAVx(PL_curcop),(I32)CopLINE(PL_curcop),sv);
3527 bof = PL_rsfp ? TRUE : FALSE;
3528 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == NULL) {
3531 PL_realtokenstart = -1;
3534 if (PL_preprocess && !PL_in_eval)
3535 (void)PerlProc_pclose(PL_rsfp);
3536 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
3537 PerlIO_clearerr(PL_rsfp);
3539 (void)PerlIO_close(PL_rsfp);
3541 PL_doextract = FALSE;
3543 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
3548 sv_setpv(PL_linestr,
3551 ? ";}continue{print;}" : ";}"));
3552 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3553 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3554 PL_last_lop = PL_last_uni = NULL;
3555 PL_minus_n = PL_minus_p = 0;
3558 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3559 PL_last_lop = PL_last_uni = NULL;
3560 sv_setpvn(PL_linestr,"",0);
3561 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
3563 /* If it looks like the start of a BOM or raw UTF-16,
3564 * check if it in fact is. */
3570 #ifdef PERLIO_IS_STDIO
3571 # ifdef __GNU_LIBRARY__
3572 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
3573 # define FTELL_FOR_PIPE_IS_BROKEN
3577 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
3578 # define FTELL_FOR_PIPE_IS_BROKEN
3583 #ifdef FTELL_FOR_PIPE_IS_BROKEN
3584 /* This loses the possibility to detect the bof
3585 * situation on perl -P when the libc5 is being used.
3586 * Workaround? Maybe attach some extra state to PL_rsfp?
3589 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
3591 bof = PerlIO_tell(PL_rsfp) == (Off_t)SvCUR(PL_linestr);
3594 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3595 s = swallow_bom((U8*)s);
3599 /* Incest with pod. */
3602 sv_catsv(PL_thiswhite, PL_linestr);
3604 if (*s == '=' && strnEQ(s, "=cut", 4)) {
3605 sv_setpvn(PL_linestr, "", 0);
3606 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3607 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3608 PL_last_lop = PL_last_uni = NULL;
3609 PL_doextract = FALSE;
3613 } while (PL_doextract);
3614 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
3615 if (PERLDB_LINE && PL_curstash != PL_debstash) {
3616 SV * const sv = newSV(0);
3618 sv_upgrade(sv, SVt_PVMG);
3619 sv_setsv(sv,PL_linestr);
3622 av_store(CopFILEAVx(PL_curcop),(I32)CopLINE(PL_curcop),sv);
3624 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3625 PL_last_lop = PL_last_uni = NULL;
3626 if (CopLINE(PL_curcop) == 1) {
3627 while (s < PL_bufend && isSPACE(*s))
3629 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
3633 PL_thiswhite = newSVpvn(PL_linestart, s - PL_linestart);
3637 if (*s == '#' && *(s+1) == '!')
3639 #ifdef ALTERNATE_SHEBANG
3641 static char const as[] = ALTERNATE_SHEBANG;
3642 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
3643 d = s + (sizeof(as) - 1);
3645 #endif /* ALTERNATE_SHEBANG */
3654 while (*d && !isSPACE(*d))
3658 #ifdef ARG_ZERO_IS_SCRIPT
3659 if (ipathend > ipath) {
3661 * HP-UX (at least) sets argv[0] to the script name,
3662 * which makes $^X incorrect. And Digital UNIX and Linux,
3663 * at least, set argv[0] to the basename of the Perl
3664 * interpreter. So, having found "#!", we'll set it right.
3666 SV * const x = GvSV(gv_fetchpvs("\030", GV_ADD|GV_NOTQUAL,
3668 assert(SvPOK(x) || SvGMAGICAL(x));
3669 if (sv_eq(x, CopFILESV(PL_curcop))) {
3670 sv_setpvn(x, ipath, ipathend - ipath);
3676 const char *bstart = SvPV_const(CopFILESV(PL_curcop),blen);
3677 const char * const lstart = SvPV_const(x,llen);
3679 bstart += blen - llen;
3680 if (strnEQ(bstart, lstart, llen) && bstart[-1] == '/') {
3681 sv_setpvn(x, ipath, ipathend - ipath);
3686 TAINT_NOT; /* $^X is always tainted, but that's OK */
3688 #endif /* ARG_ZERO_IS_SCRIPT */
3693 d = instr(s,"perl -");
3695 d = instr(s,"perl");
3697 /* avoid getting into infinite loops when shebang
3698 * line contains "Perl" rather than "perl" */
3700 for (d = ipathend-4; d >= ipath; --d) {
3701 if ((*d == 'p' || *d == 'P')
3702 && !ibcmp(d, "perl", 4))
3712 #ifdef ALTERNATE_SHEBANG
3714 * If the ALTERNATE_SHEBANG on this system starts with a
3715 * character that can be part of a Perl expression, then if
3716 * we see it but not "perl", we're probably looking at the
3717 * start of Perl code, not a request to hand off to some
3718 * other interpreter. Similarly, if "perl" is there, but
3719 * not in the first 'word' of the line, we assume the line
3720 * contains the start of the Perl program.
3722 if (d && *s != '#') {
3723 const char *c = ipath;
3724 while (*c && !strchr("; \t\r\n\f\v#", *c))
3727 d = NULL; /* "perl" not in first word; ignore */
3729 *s = '#'; /* Don't try to parse shebang line */
3731 #endif /* ALTERNATE_SHEBANG */
3732 #ifndef MACOS_TRADITIONAL
3737 !instr(s,"indir") &&
3738 instr(PL_origargv[0],"perl"))
3745 while (s < PL_bufend && isSPACE(*s))
3747 if (s < PL_bufend) {
3748 Newxz(newargv,PL_origargc+3,char*);
3750 while (s < PL_bufend && !isSPACE(*s))
3753 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
3756 newargv = PL_origargv;
3759 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
3761 Perl_croak(aTHX_ "Can't exec %s", ipath);
3765 while (*d && !isSPACE(*d))
3767 while (SPACE_OR_TAB(*d))
3771 const bool switches_done = PL_doswitches;
3772 const U32 oldpdb = PL_perldb;
3773 const bool oldn = PL_minus_n;
3774 const bool oldp = PL_minus_p;
3777 if (*d == 'M' || *d == 'm' || *d == 'C') {
3778 const char * const m = d;
3779 while (*d && !isSPACE(*d))
3781 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
3784 d = moreswitches(d);
3786 if (PL_doswitches && !switches_done) {
3787 int argc = PL_origargc;
3788 char **argv = PL_origargv;
3791 } while (argc && argv[0][0] == '-' && argv[0][1]);
3792 init_argv_symbols(argc,argv);
3794 if ((PERLDB_LINE && !oldpdb) ||
3795 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
3796 /* if we have already added "LINE: while (<>) {",
3797 we must not do it again */
3799 sv_setpvn(PL_linestr, "", 0);
3800 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3801 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3802 PL_last_lop = PL_last_uni = NULL;
3803 PL_preambled = FALSE;
3805 (void)gv_fetchfile(PL_origfilename);
3812 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
3814 PL_lex_state = LEX_FORMLINE;
3819 #ifdef PERL_STRICT_CR
3820 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
3822 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
3824 case ' ': case '\t': case '\f': case 013:
3825 #ifdef MACOS_TRADITIONAL
3829 PL_realtokenstart = -1;
3838 PL_realtokenstart = -1;
3842 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
3843 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
3844 /* handle eval qq[#line 1 "foo"\n ...] */
3845 CopLINE_dec(PL_curcop);
3848 if (PL_madskills && !PL_lex_formbrack && !PL_in_eval) {
3850 if (!PL_in_eval || PL_rsfp)
3855 while (d < PL_bufend && *d != '\n')
3859 else if (d > PL_bufend) /* Found by Ilya: feed random input to Perl. */
3860 Perl_croak(aTHX_ "panic: input overflow");
3863 PL_thiswhite = newSVpvn(s, d - s);
3868 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
3870 PL_lex_state = LEX_FORMLINE;
3876 if (PL_madskills && CopLINE(PL_curcop) >= 1 && !PL_lex_formbrack) {
3877 if (CopLINE(PL_curcop) == 1 && s[0] == '#' && s[1] == '!') {
3880 TOKEN(PEG); /* make sure any #! line is accessible */
3885 /* if (PL_madskills && PL_lex_formbrack) { */
3887 while (d < PL_bufend && *d != '\n')
3891 else if (d > PL_bufend) /* Found by Ilya: feed random input to Perl. */
3892 Perl_croak(aTHX_ "panic: input overflow");
3893 if (PL_madskills && CopLINE(PL_curcop) >= 1) {
3895 PL_thiswhite = newSVpvn("",0);
3896 if (CopLINE(PL_curcop) == 1) {
3897 sv_setpvn(PL_thiswhite, "", 0);
3900 sv_catpvn(PL_thiswhite, s, d - s);
3914 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
3922 while (s < PL_bufend && SPACE_OR_TAB(*s))
3925 if (strnEQ(s,"=>",2)) {
3926 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
3927 DEBUG_T( { printbuf("### Saw unary minus before =>, forcing word %s\n", s); } );
3928 OPERATOR('-'); /* unary minus */
3930 PL_last_uni = PL_oldbufptr;
3932 case 'r': ftst = OP_FTEREAD; break;
3933 case 'w': ftst = OP_FTEWRITE; break;
3934 case 'x': ftst = OP_FTEEXEC; break;
3935 case 'o': ftst = OP_FTEOWNED; break;
3936 case 'R': ftst = OP_FTRREAD; break;
3937 case 'W': ftst = OP_FTRWRITE; break;
3938 case 'X': ftst = OP_FTREXEC; break;
3939 case 'O': ftst = OP_FTROWNED; break;
3940 case 'e': ftst = OP_FTIS; break;
3941 case 'z': ftst = OP_FTZERO; break;
3942 case 's': ftst = OP_FTSIZE; break;
3943 case 'f': ftst = OP_FTFILE; break;
3944 case 'd': ftst = OP_FTDIR; break;
3945 case 'l': ftst = OP_FTLINK; break;
3946 case 'p': ftst = OP_FTPIPE; break;
3947 case 'S': ftst = OP_FTSOCK; break;
3948 case 'u': ftst = OP_FTSUID; break;
3949 case 'g': ftst = OP_FTSGID; break;
3950 case 'k': ftst = OP_FTSVTX; break;
3951 case 'b': ftst = OP_FTBLK; break;
3952 case 'c': ftst = OP_FTCHR; break;
3953 case 't': ftst = OP_FTTTY; break;
3954 case 'T': ftst = OP_FTTEXT; break;
3955 case 'B': ftst = OP_FTBINARY; break;
3956 case 'M': case 'A': case 'C':
3957 gv_fetchpvs("\024", GV_ADD|GV_NOTQUAL, SVt_PV);
3959 case 'M': ftst = OP_FTMTIME; break;
3960 case 'A': ftst = OP_FTATIME; break;
3961 case 'C': ftst = OP_FTCTIME; break;
3969 PL_last_lop_op = (OPCODE)ftst;
3970 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3971 "### Saw file test %c\n", (int)tmp);
3976 /* Assume it was a minus followed by a one-letter named
3977 * subroutine call (or a -bareword), then. */
3978 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3979 "### '-%c' looked like a file test but was not\n",
3986 const char tmp = *s++;
3989 if (PL_expect == XOPERATOR)
3994 else if (*s == '>') {
3997 if (isIDFIRST_lazy_if(s,UTF)) {
3998 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
4006 if (PL_expect == XOPERATOR)
4009 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
4011 OPERATOR('-'); /* unary minus */
4017 const char tmp = *s++;
4020 if (PL_expect == XOPERATOR)
4025 if (PL_expect == XOPERATOR)
4028 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
4035 if (PL_expect != XOPERATOR) {
4036 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4037 PL_expect = XOPERATOR;
4038 force_ident(PL_tokenbuf, '*');
4051 if (PL_expect == XOPERATOR) {
4055 PL_tokenbuf[0] = '%';
4056 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
4057 if (!PL_tokenbuf[1]) {
4060 PL_pending_ident = '%';
4071 && (PL_expect == XOPERATOR || PL_expect == XTERMORDORDOR)
4072 && FEATURE_IS_ENABLED("~~"))
4079 const char tmp = *s++;
4085 goto just_a_word_zero_gv;
4088 switch (PL_expect) {
4094 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
4096 PL_bufptr = s; /* update in case we back off */
4102 PL_expect = XTERMBLOCK;
4105 stuffstart = s - SvPVX(PL_linestr) - 1;
4109 while (isIDFIRST_lazy_if(s,UTF)) {
4112 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4113 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
4114 if (tmp < 0) tmp = -tmp;
4129 sv = newSVpvn(s, len);
4131 d = scan_str(d,TRUE,TRUE);
4133 /* MUST advance bufptr here to avoid bogus
4134 "at end of line" context messages from yyerror().
4136 PL_bufptr = s + len;
4137 yyerror("Unterminated attribute parameter in attribute list");
4141 return REPORT(0); /* EOF indicator */
4145 sv_catsv(sv, PL_lex_stuff);
4146 attrs = append_elem(OP_LIST, attrs,
4147 newSVOP(OP_CONST, 0, sv));
4148 SvREFCNT_dec(PL_lex_stuff);
4149 PL_lex_stuff = NULL;
4152 if (len == 6 && strnEQ(SvPVX(sv), "unique", len)) {
4154 if (PL_in_my == KEY_our) {
4156 GvUNIQUE_on(cGVOPx_gv(yylval.opval));
4158 /* skip to avoid loading attributes.pm */
4160 deprecate(":unique");
4163 Perl_croak(aTHX_ "The 'unique' attribute may only be applied to 'our' variables");
4166 /* NOTE: any CV attrs applied here need to be part of
4167 the CVf_BUILTIN_ATTRS define in cv.h! */
4168 else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "lvalue", len)) {
4170 CvLVALUE_on(PL_compcv);
4172 else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "locked", len)) {
4174 CvLOCKED_on(PL_compcv);
4176 else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "method", len)) {
4178 CvMETHOD_on(PL_compcv);
4180 else if (!PL_in_my && len == 9 && strnEQ(SvPVX(sv), "assertion", len)) {
4182 CvASSERTION_on(PL_compcv);
4184 /* After we've set the flags, it could be argued that
4185 we don't need to do the attributes.pm-based setting
4186 process, and shouldn't bother appending recognized
4187 flags. To experiment with that, uncomment the
4188 following "else". (Note that's already been
4189 uncommented. That keeps the above-applied built-in
4190 attributes from being intercepted (and possibly
4191 rejected) by a package's attribute routines, but is
4192 justified by the performance win for the common case
4193 of applying only built-in attributes.) */
4195 attrs = append_elem(OP_LIST, attrs,
4196 newSVOP(OP_CONST, 0,
4200 if (*s == ':' && s[1] != ':')
4203 break; /* require real whitespace or :'s */
4204 /* XXX losing whitespace on sequential attributes here */
4208 = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
4209 if (*s != ';' && *s != '}' && *s != tmp
4210 && (tmp != '=' || *s != ')')) {
4211 const char q = ((*s == '\'') ? '"' : '\'');
4212 /* If here for an expression, and parsed no attrs, back
4214 if (tmp == '=' && !attrs) {
4218 /* MUST advance bufptr here to avoid bogus "at end of line"
4219 context messages from yyerror().
4222 yyerror( (const char *)
4224 ? Perl_form(aTHX_ "Invalid separator character "
4225 "%c%c%c in attribute list", q, *s, q)
4226 : "Unterminated attribute list" ) );
4234 start_force(PL_curforce);
4235 NEXTVAL_NEXTTOKE.opval = attrs;
4236 CURMAD('_', PL_nextwhite);
4241 PL_thistoken = newSVpvn(SvPVX(PL_linestr) + stuffstart,
4242 (s - SvPVX(PL_linestr)) - stuffstart);
4250 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
4251 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
4259 const char tmp = *s++;
4264 const char tmp = *s++;
4272 if (PL_lex_brackets <= 0)
4273 yyerror("Unmatched right square bracket");
4276 if (PL_lex_state == LEX_INTERPNORMAL) {
4277 if (PL_lex_brackets == 0) {
4278 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
4279 PL_lex_state = LEX_INTERPEND;
4286 if (PL_lex_brackets > 100) {
4287 Renew(PL_lex_brackstack, PL_lex_brackets + 10, char);
4289 switch (PL_expect) {
4291 if (PL_lex_formbrack) {
4295 if (PL_oldoldbufptr == PL_last_lop)
4296 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
4298 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
4299 OPERATOR(HASHBRACK);
4301 while (s < PL_bufend && SPACE_OR_TAB(*s))
4304 PL_tokenbuf[0] = '\0';
4305 if (d < PL_bufend && *d == '-') {
4306 PL_tokenbuf[0] = '-';
4308 while (d < PL_bufend && SPACE_OR_TAB(*d))
4311 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
4312 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
4314 while (d < PL_bufend && SPACE_OR_TAB(*d))
4317 const char minus = (PL_tokenbuf[0] == '-');
4318 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
4326 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
4331 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
4336 if (PL_oldoldbufptr == PL_last_lop)
4337 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
4339 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
4342 if (PL_expect == XREF && PL_lex_state == LEX_INTERPNORMAL) {
4344 /* This hack is to get the ${} in the message. */
4346 yyerror("syntax error");
4349 OPERATOR(HASHBRACK);
4351 /* This hack serves to disambiguate a pair of curlies
4352 * as being a block or an anon hash. Normally, expectation
4353 * determines that, but in cases where we're not in a
4354 * position to expect anything in particular (like inside
4355 * eval"") we have to resolve the ambiguity. This code
4356 * covers the case where the first term in the curlies is a
4357 * quoted string. Most other cases need to be explicitly
4358 * disambiguated by prepending a "+" before the opening
4359 * curly in order to force resolution as an anon hash.
4361 * XXX should probably propagate the outer expectation
4362 * into eval"" to rely less on this hack, but that could
4363 * potentially break current behavior of eval"".
4367 if (*s == '\'' || *s == '"' || *s == '`') {
4368 /* common case: get past first string, handling escapes */
4369 for (t++; t < PL_bufend && *t != *s;)
4370 if (*t++ == '\\' && (*t == '\\' || *t == *s))
4374 else if (*s == 'q') {
4377 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
4380 /* skip q//-like construct */
4382 char open, close, term;
4385 while (t < PL_bufend && isSPACE(*t))
4387 /* check for q => */
4388 if (t+1 < PL_bufend && t[0] == '=' && t[1] == '>') {
4389 OPERATOR(HASHBRACK);
4393 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
4397 for (t++; t < PL_bufend; t++) {
4398 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
4400 else if (*t == open)
4404 for (t++; t < PL_bufend; t++) {
4405 if (*t == '\\' && t+1 < PL_bufend)
4407 else if (*t == close && --brackets <= 0)
4409 else if (*t == open)
4416 /* skip plain q word */
4417 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
4420 else if (isALNUM_lazy_if(t,UTF)) {
4422 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
4425 while (t < PL_bufend && isSPACE(*t))
4427 /* if comma follows first term, call it an anon hash */
4428 /* XXX it could be a comma expression with loop modifiers */
4429 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
4430 || (*t == '=' && t[1] == '>')))
4431 OPERATOR(HASHBRACK);
4432 if (PL_expect == XREF)
4435 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
4441 yylval.ival = CopLINE(PL_curcop);
4442 if (isSPACE(*s) || *s == '#')
4443 PL_copline = NOLINE; /* invalidate current command line number */
4448 if (PL_lex_brackets <= 0)
4449 yyerror("Unmatched right curly bracket");
4451 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
4452 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
4453 PL_lex_formbrack = 0;
4454 if (PL_lex_state == LEX_INTERPNORMAL) {
4455 if (PL_lex_brackets == 0) {
4456 if (PL_expect & XFAKEBRACK) {
4457 PL_expect &= XENUMMASK;
4458 PL_lex_state = LEX_INTERPEND;
4463 PL_thiswhite = newSVpvn("",0);
4464 sv_catpvn(PL_thiswhite,"}",1);
4467 return yylex(); /* ignore fake brackets */
4469 if (*s == '-' && s[1] == '>')
4470 PL_lex_state = LEX_INTERPENDMAYBE;
4471 else if (*s != '[' && *s != '{')
4472 PL_lex_state = LEX_INTERPEND;
4475 if (PL_expect & XFAKEBRACK) {
4476 PL_expect &= XENUMMASK;
4478 return yylex(); /* ignore fake brackets */
4480 start_force(PL_curforce);
4482 curmad('X', newSVpvn(s-1,1));
4483 CURMAD('_', PL_thiswhite);
4488 PL_thistoken = newSVpvn("",0);
4496 if (PL_expect == XOPERATOR) {
4497 if (PL_bufptr == PL_linestart && ckWARN(WARN_SEMICOLON)
4498 && isIDFIRST_lazy_if(s,UTF))
4500 CopLINE_dec(PL_curcop);
4501 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
4502 CopLINE_inc(PL_curcop);
4507 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4509 PL_expect = XOPERATOR;
4510 force_ident(PL_tokenbuf, '&');
4514 yylval.ival = (OPpENTERSUB_AMPER<<8);
4526 const char tmp = *s++;
4533 if (tmp && isSPACE(*s) && ckWARN(WARN_SYNTAX)
4534 && strchr("+-*/%.^&|<",tmp))
4535 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4536 "Reversed %c= operator",(int)tmp);
4538 if (PL_expect == XSTATE && isALPHA(tmp) &&
4539 (s == PL_linestart+1 || s[-2] == '\n') )
4541 if (PL_in_eval && !PL_rsfp) {
4546 if (strnEQ(s,"=cut",4)) {
4562 PL_thiswhite = newSVpvn("",0);
4563 sv_catpvn(PL_thiswhite, PL_linestart,
4564 PL_bufend - PL_linestart);
4568 PL_doextract = TRUE;
4572 if (PL_lex_brackets < PL_lex_formbrack) {
4574 #ifdef PERL_STRICT_CR
4575 while (SPACE_OR_TAB(*t))
4577 while (SPACE_OR_TAB(*t) || *t == '\r')
4580 if (*t == '\n' || *t == '#') {
4591 const char tmp = *s++;
4593 /* was this !=~ where !~ was meant?
4594 * warn on m:!=~\s+([/?]|[msy]\W|tr\W): */
4596 if (*s == '~' && ckWARN(WARN_SYNTAX)) {
4597 const char *t = s+1;
4599 while (t < PL_bufend && isSPACE(*t))
4602 if (*t == '/' || *t == '?' ||
4603 ((*t == 'm' || *t == 's' || *t == 'y')
4604 && !isALNUM(t[1])) ||
4605 (*t == 't' && t[1] == 'r' && !isALNUM(t[2])))
4606 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4607 "!=~ should be !~");
4617 if (PL_expect != XOPERATOR) {
4618 if (s[1] != '<' && !strchr(s,'>'))
4621 s = scan_heredoc(s);
4623 s = scan_inputsymbol(s);
4624 TERM(sublex_start());
4630 SHop(OP_LEFT_SHIFT);
4644 const char tmp = *s++;
4646 SHop(OP_RIGHT_SHIFT);
4647 else if (tmp == '=')
4656 if (PL_expect == XOPERATOR) {
4657 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
4659 deprecate_old(commaless_variable_list);
4660 return REPORT(','); /* grandfather non-comma-format format */
4664 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
4665 PL_tokenbuf[0] = '@';
4666 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
4667 sizeof PL_tokenbuf - 1, FALSE);
4668 if (PL_expect == XOPERATOR)
4669 no_op("Array length", s);
4670 if (!PL_tokenbuf[1])
4672 PL_expect = XOPERATOR;
4673 PL_pending_ident = '#';
4677 PL_tokenbuf[0] = '$';
4678 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
4679 sizeof PL_tokenbuf - 1, FALSE);
4680 if (PL_expect == XOPERATOR)
4682 if (!PL_tokenbuf[1]) {
4684 yyerror("Final $ should be \\$ or $name");
4688 /* This kludge not intended to be bulletproof. */
4689 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
4690 yylval.opval = newSVOP(OP_CONST, 0,
4691 newSViv(CopARYBASE_get(&PL_compiling)));
4692 yylval.opval->op_private = OPpCONST_ARYBASE;
4698 const char tmp = *s;
4699 if (PL_lex_state == LEX_NORMAL)
4702 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop)
4703 && intuit_more(s)) {
4705 PL_tokenbuf[0] = '@';
4706 if (ckWARN(WARN_SYNTAX)) {
4709 while (isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$')
4712 PL_bufptr = PEEKSPACE(PL_bufptr); /* XXX can realloc */
4713 while (t < PL_bufend && *t != ']')
4715 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4716 "Multidimensional syntax %.*s not supported",
4717 (int)((t - PL_bufptr) + 1), PL_bufptr);
4721 else if (*s == '{') {
4723 PL_tokenbuf[0] = '%';
4724 if (strEQ(PL_tokenbuf+1, "SIG") && ckWARN(WARN_SYNTAX)
4725 && (t = strchr(s, '}')) && (t = strchr(t, '=')))
4727 char tmpbuf[sizeof PL_tokenbuf];
4730 } while (isSPACE(*t));
4731 if (isIDFIRST_lazy_if(t,UTF)) {
4733 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE,
4737 if (*t == ';' && get_cv(tmpbuf, FALSE))
4738 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4739 "You need to quote \"%s\"",
4746 PL_expect = XOPERATOR;
4747 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
4748 const bool islop = (PL_last_lop == PL_oldoldbufptr);
4749 if (!islop || PL_last_lop_op == OP_GREPSTART)
4750 PL_expect = XOPERATOR;
4751 else if (strchr("$@\"'`q", *s))
4752 PL_expect = XTERM; /* e.g. print $fh "foo" */
4753 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
4754 PL_expect = XTERM; /* e.g. print $fh &sub */
4755 else if (isIDFIRST_lazy_if(s,UTF)) {
4756 char tmpbuf[sizeof PL_tokenbuf];
4758 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4759 if ((t2 = keyword(tmpbuf, len))) {
4760 /* binary operators exclude handle interpretations */
4772 PL_expect = XTERM; /* e.g. print $fh length() */
4777 PL_expect = XTERM; /* e.g. print $fh subr() */
4780 else if (isDIGIT(*s))
4781 PL_expect = XTERM; /* e.g. print $fh 3 */
4782 else if (*s == '.' && isDIGIT(s[1]))
4783 PL_expect = XTERM; /* e.g. print $fh .3 */
4784 else if ((*s == '?' || *s == '-' || *s == '+')
4785 && !isSPACE(s[1]) && s[1] != '=')
4786 PL_expect = XTERM; /* e.g. print $fh -1 */
4787 else if (*s == '/' && !isSPACE(s[1]) && s[1] != '='
4789 PL_expect = XTERM; /* e.g. print $fh /.../
4790 XXX except DORDOR operator
4792 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2])
4794 PL_expect = XTERM; /* print $fh <<"EOF" */
4797 PL_pending_ident = '$';
4801 if (PL_expect == XOPERATOR)
4803 PL_tokenbuf[0] = '@';
4804 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
4805 if (!PL_tokenbuf[1]) {
4808 if (PL_lex_state == LEX_NORMAL)
4810 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
4812 PL_tokenbuf[0] = '%';
4814 /* Warn about @ where they meant $. */
4815 if (*s == '[' || *s == '{') {
4816 if (ckWARN(WARN_SYNTAX)) {
4817 const char *t = s + 1;
4818 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
4820 if (*t == '}' || *t == ']') {
4822 PL_bufptr = PEEKSPACE(PL_bufptr); /* XXX can realloc */
4823 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4824 "Scalar value %.*s better written as $%.*s",
4825 (int)(t-PL_bufptr), PL_bufptr,
4826 (int)(t-PL_bufptr-1), PL_bufptr+1);
4831 PL_pending_ident = '@';
4834 case '/': /* may be division, defined-or, or pattern */
4835 if (PL_expect == XTERMORDORDOR && s[1] == '/') {
4839 case '?': /* may either be conditional or pattern */
4840 if(PL_expect == XOPERATOR) {
4848 /* A // operator. */
4858 /* Disable warning on "study /blah/" */
4859 if (PL_oldoldbufptr == PL_last_uni
4860 && (*PL_last_uni != 's' || s - PL_last_uni < 5
4861 || memNE(PL_last_uni, "study", 5)
4862 || isALNUM_lazy_if(PL_last_uni+5,UTF)
4865 s = scan_pat(s,OP_MATCH);
4866 TERM(sublex_start());
4870 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
4871 #ifdef PERL_STRICT_CR
4874 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
4876 && (s == PL_linestart || s[-1] == '\n') )
4878 PL_lex_formbrack = 0;
4882 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
4888 yylval.ival = OPf_SPECIAL;
4894 if (PL_expect != XOPERATOR)
4899 case '0': case '1': case '2': case '3': case '4':
4900 case '5': case '6': case '7': case '8': case '9':
4901 s = scan_num(s, &yylval);
4902 DEBUG_T( { printbuf("### Saw number in %s\n", s); } );
4903 if (PL_expect == XOPERATOR)
4908 s = scan_str(s,!!PL_madskills,FALSE);
4909 DEBUG_T( { printbuf("### Saw string before %s\n", s); } );
4910 if (PL_expect == XOPERATOR) {
4911 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
4913 deprecate_old(commaless_variable_list);
4914 return REPORT(','); /* grandfather non-comma-format format */
4921 yylval.ival = OP_CONST;
4922 TERM(sublex_start());
4925 s = scan_str(s,!!PL_madskills,FALSE);
4926 DEBUG_T( { printbuf("### Saw string before %s\n", s); } );
4927 if (PL_expect == XOPERATOR) {
4928 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
4930 deprecate_old(commaless_variable_list);
4931 return REPORT(','); /* grandfather non-comma-format format */
4938 yylval.ival = OP_CONST;
4939 /* FIXME. I think that this can be const if char *d is replaced by
4940 more localised variables. */
4941 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
4942 if (*d == '$' || *d == '@' || *d == '\\' || !UTF8_IS_INVARIANT((U8)*d)) {
4943 yylval.ival = OP_STRINGIFY;
4947 TERM(sublex_start());
4950 s = scan_str(s,!!PL_madskills,FALSE);
4951 DEBUG_T( { printbuf("### Saw backtick string before %s\n", s); } );
4952 if (PL_expect == XOPERATOR)
4953 no_op("Backticks",s);
4956 yylval.ival = OP_BACKTICK;
4958 TERM(sublex_start());
4962 if (PL_lex_inwhat && isDIGIT(*s) && ckWARN(WARN_SYNTAX))
4963 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),"Can't use \\%c to mean $%c in expression",
4965 if (PL_expect == XOPERATOR)
4966 no_op("Backslash",s);
4970 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
4971 char *start = s + 2;
4972 while (isDIGIT(*start) || *start == '_')
4974 if (*start == '.' && isDIGIT(start[1])) {
4975 s = scan_num(s, &yylval);
4978 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
4979 else if (!isALPHA(*start) && (PL_expect == XTERM
4980 || PL_expect == XREF || PL_expect == XSTATE
4981 || PL_expect == XTERMORDORDOR)) {
4982 /* XXX Use gv_fetchpvn rather than stomping on a const string */
4983 const char c = *start;
4986 gv = gv_fetchpv(s, 0, SVt_PVCV);
4989 s = scan_num(s, &yylval);
4996 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
5038 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
5040 /* Some keywords can be followed by any delimiter, including ':' */
5041 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
5042 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
5043 (PL_tokenbuf[0] == 'q' &&
5044 strchr("qwxr", PL_tokenbuf[1])))));
5046 /* x::* is just a word, unless x is "CORE" */
5047 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
5051 while (d < PL_bufend && isSPACE(*d))
5052 d++; /* no comments skipped here, or s### is misparsed */
5054 /* Is this a label? */
5055 if (!tmp && PL_expect == XSTATE
5056 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
5058 yylval.pval = savepv(PL_tokenbuf);
5063 /* Check for keywords */
5064 tmp = keyword(PL_tokenbuf, len);
5066 /* Is this a word before a => operator? */
5067 if (*d == '=' && d[1] == '>') {
5070 = (OP*)newSVOP(OP_CONST, 0,
5071 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
5072 yylval.opval->op_private = OPpCONST_BARE;
5076 if (tmp < 0) { /* second-class keyword? */
5077 GV *ogv = NULL; /* override (winner) */
5078 GV *hgv = NULL; /* hidden (loser) */
5079 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
5081 if ((gv = gv_fetchpvn_flags(PL_tokenbuf, len, 0, SVt_PVCV)) &&
5084 if (GvIMPORTED_CV(gv))
5086 else if (! CvMETHOD(cv))
5090 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
5091 (gv = *gvp) != (GV*)&PL_sv_undef &&
5092 GvCVu(gv) && GvIMPORTED_CV(gv))
5099 tmp = 0; /* overridden by import or by GLOBAL */
5102 && -tmp==KEY_lock /* XXX generalizable kludge */
5104 && !hv_fetchs(GvHVn(PL_incgv), "Thread.pm", FALSE))
5106 tmp = 0; /* any sub overrides "weak" keyword */
5108 else { /* no override */
5110 if (tmp == KEY_dump && ckWARN(WARN_MISC)) {
5111 Perl_warner(aTHX_ packWARN(WARN_MISC),
5112 "dump() better written as CORE::dump()");
5116 if (hgv && tmp != KEY_x && tmp != KEY_CORE
5117 && ckWARN(WARN_AMBIGUOUS)) /* never ambiguous */
5118 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5119 "Ambiguous call resolved as CORE::%s(), %s",
5120 GvENAME(hgv), "qualify as such or use &");
5127 default: /* not a keyword */
5128 /* Trade off - by using this evil construction we can pull the
5129 variable gv into the block labelled keylookup. If not, then
5130 we have to give it function scope so that the goto from the
5131 earlier ':' case doesn't bypass the initialisation. */
5133 just_a_word_zero_gv:
5141 const char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
5144 SV *nextPL_nextwhite = 0;
5148 /* Get the rest if it looks like a package qualifier */
5150 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
5152 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
5155 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
5156 *s == '\'' ? "'" : "::");
5161 if (PL_expect == XOPERATOR) {
5162 if (PL_bufptr == PL_linestart) {
5163 CopLINE_dec(PL_curcop);
5164 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
5165 CopLINE_inc(PL_curcop);
5168 no_op("Bareword",s);
5171 /* Look for a subroutine with this name in current package,
5172 unless name is "Foo::", in which case Foo is a bearword
5173 (and a package name). */
5175 if (len > 2 && !PL_madskills &&
5176 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
5178 if (ckWARN(WARN_BAREWORD)
5179 && ! gv_fetchpvn_flags(PL_tokenbuf, len, 0, SVt_PVHV))
5180 Perl_warner(aTHX_ packWARN(WARN_BAREWORD),
5181 "Bareword \"%s\" refers to nonexistent package",
5184 PL_tokenbuf[len] = '\0';
5190 /* Mustn't actually add anything to a symbol table.
5191 But also don't want to "initialise" any placeholder
5192 constants that might already be there into full
5193 blown PVGVs with attached PVCV. */
5194 gv = gv_fetchpvn_flags(PL_tokenbuf, len,
5195 GV_NOADD_NOINIT, SVt_PVCV);
5200 /* if we saw a global override before, get the right name */
5203 sv = newSVpvs("CORE::GLOBAL::");
5204 sv_catpv(sv,PL_tokenbuf);
5207 /* If len is 0, newSVpv does strlen(), which is correct.
5208 If len is non-zero, then it will be the true length,
5209 and so the scalar will be created correctly. */
5210 sv = newSVpv(PL_tokenbuf,len);
5213 if (PL_madskills && !PL_thistoken) {
5214 char *start = SvPVX(PL_linestr) + PL_realtokenstart;
5215 PL_thistoken = newSVpv(start,s - start);
5216 PL_realtokenstart = s - SvPVX(PL_linestr);
5220 /* Presume this is going to be a bareword of some sort. */
5223 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
5224 yylval.opval->op_private = OPpCONST_BARE;
5225 /* UTF-8 package name? */
5226 if (UTF && !IN_BYTES &&
5227 is_utf8_string((U8*)SvPVX_const(sv), SvCUR(sv)))
5230 /* And if "Foo::", then that's what it certainly is. */
5235 /* Do the explicit type check so that we don't need to force
5236 the initialisation of the symbol table to have a real GV.
5237 Beware - gv may not really be a PVGV, cv may not really be
5238 a PVCV, (because of the space optimisations that gv_init
5239 understands) But they're true if for this symbol there is
5240 respectively a typeglob and a subroutine.
5242 cv = gv ? ((SvTYPE(gv) == SVt_PVGV)
5243 /* Real typeglob, so get the real subroutine: */
5245 /* A proxy for a subroutine in this package? */
5246 : SvOK(gv) ? (CV *) gv : NULL)
5249 /* See if it's the indirect object for a list operator. */
5251 if (PL_oldoldbufptr &&
5252 PL_oldoldbufptr < PL_bufptr &&
5253 (PL_oldoldbufptr == PL_last_lop
5254 || PL_oldoldbufptr == PL_last_uni) &&
5255 /* NO SKIPSPACE BEFORE HERE! */
5256 (PL_expect == XREF ||
5257 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
5259 bool immediate_paren = *s == '(';
5261 /* (Now we can afford to cross potential line boundary.) */
5262 s = SKIPSPACE2(s,nextPL_nextwhite);
5264 PL_nextwhite = nextPL_nextwhite; /* assume no & deception */
5267 /* Two barewords in a row may indicate method call. */
5269 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') &&
5270 (tmp = intuit_method(s, gv, cv)))
5273 /* If not a declared subroutine, it's an indirect object. */
5274 /* (But it's an indir obj regardless for sort.) */
5275 /* Also, if "_" follows a filetest operator, it's a bareword */
5278 ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
5280 (PL_last_lop_op != OP_MAPSTART &&
5281 PL_last_lop_op != OP_GREPSTART))))
5282 || (PL_tokenbuf[0] == '_' && PL_tokenbuf[1] == '\0'
5283 && ((PL_opargs[PL_last_lop_op] & OA_CLASS_MASK) == OA_FILESTATOP))
5286 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
5291 PL_expect = XOPERATOR;
5294 s = SKIPSPACE2(s,nextPL_nextwhite);
5295 PL_nextwhite = nextPL_nextwhite;
5300 /* Is this a word before a => operator? */
5301 if (*s == '=' && s[1] == '>' && !pkgname) {
5303 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
5304 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
5305 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
5309 /* If followed by a paren, it's certainly a subroutine. */
5314 while (SPACE_OR_TAB(*d))
5316 if (*d == ')' && (sv = gv_const_sv(gv))) {
5320 char *par = SvPVX(PL_linestr) + PL_realtokenstart;
5321 sv_catpvn(PL_thistoken, par, s - par);
5323 sv_free(PL_nextwhite);
5333 PL_nextwhite = PL_thiswhite;
5336 start_force(PL_curforce);
5338 NEXTVAL_NEXTTOKE.opval = yylval.opval;
5339 PL_expect = XOPERATOR;
5342 PL_nextwhite = nextPL_nextwhite;
5343 curmad('X', PL_thistoken);
5344 PL_thistoken = newSVpvn("",0);
5352 /* If followed by var or block, call it a method (unless sub) */
5354 if ((*s == '$' || *s == '{') && (!gv || !cv)) {
5355 PL_last_lop = PL_oldbufptr;
5356 PL_last_lop_op = OP_METHOD;
5360 /* If followed by a bareword, see if it looks like indir obj. */
5363 && (isIDFIRST_lazy_if(s,UTF) || *s == '$')
5364 && (tmp = intuit_method(s, gv, cv)))
5367 /* Not a method, so call it a subroutine (if defined) */
5370 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
5371 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5372 "Ambiguous use of -%s resolved as -&%s()",
5373 PL_tokenbuf, PL_tokenbuf);
5374 /* Check for a constant sub */
5375 if ((sv = gv_const_sv(gv))) {
5377 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
5378 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc_simple(sv);
5379 yylval.opval->op_private = 0;
5383 /* Resolve to GV now. */
5384 if (SvTYPE(gv) != SVt_PVGV) {
5385 gv = gv_fetchpv(PL_tokenbuf, 0, SVt_PVCV);
5386 assert (SvTYPE(gv) == SVt_PVGV);
5387 /* cv must have been some sort of placeholder, so
5388 now needs replacing with a real code reference. */
5392 op_free(yylval.opval);
5393 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
5394 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
5395 PL_last_lop = PL_oldbufptr;
5396 PL_last_lop_op = OP_ENTERSUB;
5397 /* Is there a prototype? */
5404 const char *proto = SvPV_const((SV*)cv, protolen);
5407 if (*proto == '$' && proto[1] == '\0')
5409 while (*proto == ';')
5411 if (*proto == '&' && *s == '{') {
5412 sv_setpv(PL_subname,
5415 "__ANON__" : "__ANON__::__ANON__"));
5422 PL_nextwhite = PL_thiswhite;
5425 start_force(PL_curforce);
5426 NEXTVAL_NEXTTOKE.opval = yylval.opval;
5429 PL_nextwhite = nextPL_nextwhite;
5430 curmad('X', PL_thistoken);
5431 PL_thistoken = newSVpvn("",0);
5438 /* Guess harder when madskills require "best effort". */
5439 if (PL_madskills && (!gv || !GvCVu(gv))) {
5440 int probable_sub = 0;
5441 if (strchr("\"'`$@%0123456789!*+{[<", *s))
5443 else if (isALPHA(*s)) {
5447 d = scan_word(d, tmpbuf, sizeof tmpbuf, TRUE, &tmplen);
5448 if (!keyword(tmpbuf,tmplen))
5451 while (d < PL_bufend && isSPACE(*d))
5453 if (*d == '=' && d[1] == '>')
5458 gv = gv_fetchpv(PL_tokenbuf, TRUE, SVt_PVCV);
5459 op_free(yylval.opval);
5460 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
5461 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
5462 PL_last_lop = PL_oldbufptr;
5463 PL_last_lop_op = OP_ENTERSUB;
5464 PL_nextwhite = PL_thiswhite;
5466 start_force(PL_curforce);
5467 NEXTVAL_NEXTTOKE.opval = yylval.opval;
5469 PL_nextwhite = nextPL_nextwhite;
5470 curmad('X', PL_thistoken);
5471 PL_thistoken = newSVpvn("",0);
5476 NEXTVAL_NEXTTOKE.opval = yylval.opval;
5483 /* Call it a bare word */
5485 if (PL_hints & HINT_STRICT_SUBS)
5486 yylval.opval->op_private |= OPpCONST_STRICT;
5489 if (lastchar != '-') {
5490 if (ckWARN(WARN_RESERVED)) {
5494 if (!*d && !gv_stashpv(PL_tokenbuf,FALSE))
5495 Perl_warner(aTHX_ packWARN(WARN_RESERVED), PL_warn_reserved,
5502 if ((lastchar == '*' || lastchar == '%' || lastchar == '&')
5503 && ckWARN_d(WARN_AMBIGUOUS)) {
5504 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5505 "Operator or semicolon missing before %c%s",
5506 lastchar, PL_tokenbuf);
5507 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5508 "Ambiguous use of %c resolved as operator %c",
5509 lastchar, lastchar);
5515 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
5516 newSVpv(CopFILE(PL_curcop),0));
5520 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
5521 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
5524 case KEY___PACKAGE__:
5525 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
5527 ? newSVhek(HvNAME_HEK(PL_curstash))
5534 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
5535 const char *pname = "main";
5536 if (PL_tokenbuf[2] == 'D')
5537 pname = HvNAME_get(PL_curstash ? PL_curstash : PL_defstash);
5538 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), GV_ADD,
5542 GvIOp(gv) = newIO();
5543 IoIFP(GvIOp(gv)) = PL_rsfp;
5544 #if defined(HAS_FCNTL) && defined(F_SETFD)
5546 const int fd = PerlIO_fileno(PL_rsfp);
5547 fcntl(fd,F_SETFD,fd >= 3);
5550 /* Mark this internal pseudo-handle as clean */
5551 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
5553 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
5554 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
5555 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
5557 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
5558 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
5559 /* if the script was opened in binmode, we need to revert
5560 * it to text mode for compatibility; but only iff it has CRs
5561 * XXX this is a questionable hack at best. */
5562 if (PL_bufend-PL_bufptr > 2
5563 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
5566 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
5567 loc = PerlIO_tell(PL_rsfp);
5568 (void)PerlIO_seek(PL_rsfp, 0L, 0);
5571 if (PerlLIO_setmode(PL_rsfp, O_TEXT) != -1) {
5573 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
5574 #endif /* NETWARE */
5575 #ifdef PERLIO_IS_STDIO /* really? */
5576 # if defined(__BORLANDC__)
5577 /* XXX see note in do_binmode() */
5578 ((FILE*)PL_rsfp)->flags &= ~_F_BIN;
5582 PerlIO_seek(PL_rsfp, loc, 0);
5586 #ifdef PERLIO_LAYERS
5589 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
5590 else if (PL_encoding) {
5597 XPUSHs(PL_encoding);
5599 call_method("name", G_SCALAR);
5603 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL,
5604 Perl_form(aTHX_ ":encoding(%"SVf")",
5613 if (PL_realtokenstart >= 0) {
5614 char *tstart = SvPVX(PL_linestr) + PL_realtokenstart;
5616 PL_endwhite = newSVpvn("",0);
5617 sv_catsv(PL_endwhite, PL_thiswhite);
5619 sv_catpvn(PL_endwhite, tstart, PL_bufend - tstart);
5620 PL_realtokenstart = -1;
5622 while ((s = filter_gets(PL_endwhite, PL_rsfp,
5623 SvCUR(PL_endwhite))) != Nullch) ;
5637 if (PL_expect == XSTATE) {
5644 if (*s == ':' && s[1] == ':') {
5647 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
5648 if (!(tmp = keyword(PL_tokenbuf, len)))
5649 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
5652 else if (tmp == KEY_require || tmp == KEY_do)
5653 /* that's a way to remember we saw "CORE::" */
5666 LOP(OP_ACCEPT,XTERM);
5672 LOP(OP_ATAN2,XTERM);
5678 LOP(OP_BINMODE,XTERM);
5681 LOP(OP_BLESS,XTERM);
5690 /* When 'use switch' is in effect, continue has a dual
5691 life as a control operator. */
5693 if (!FEATURE_IS_ENABLED("switch"))
5696 /* We have to disambiguate the two senses of
5697 "continue". If the next token is a '{' then
5698 treat it as the start of a continue block;
5699 otherwise treat it as a control operator.
5711 (void)gv_fetchpvs("ENV", GV_ADD|GV_NOTQUAL, SVt_PVHV);
5728 if (!PL_cryptseen) {
5729 PL_cryptseen = TRUE;
5733 LOP(OP_CRYPT,XTERM);
5736 LOP(OP_CHMOD,XTERM);
5739 LOP(OP_CHOWN,XTERM);
5742 LOP(OP_CONNECT,XTERM);
5761 s = force_word(s,WORD,TRUE,TRUE,FALSE);
5762 if (orig_keyword == KEY_do) {
5771 PL_hints |= HINT_BLOCK_SCOPE;
5781 gv_fetchpvs("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
5782 LOP(OP_DBMOPEN,XTERM);
5788 s = force_word(s,WORD,TRUE,FALSE,FALSE);
5795 yylval.ival = CopLINE(PL_curcop);
5811 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
5812 UNIBRACK(OP_ENTEREVAL);
5830 case KEY_endhostent:
5836 case KEY_endservent:
5839 case KEY_endprotoent:
5850 yylval.ival = CopLINE(PL_curcop);
5852 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
5855 int soff = s - SvPVX(PL_linestr); /* for skipspace realloc */
5858 if ((PL_bufend - p) >= 3 &&
5859 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
5861 else if ((PL_bufend - p) >= 4 &&
5862 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
5865 if (isIDFIRST_lazy_if(p,UTF)) {
5866 p = scan_ident(p, PL_bufend,
5867 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
5871 Perl_croak(aTHX_ "Missing $ on loop variable");
5873 s = SvPVX(PL_linestr) + soff;
5879 LOP(OP_FORMLINE,XTERM);
5885 LOP(OP_FCNTL,XTERM);
5891 LOP(OP_FLOCK,XTERM);
5900 LOP(OP_GREPSTART, XREF);
5903 s = force_word(s,WORD,TRUE,FALSE,FALSE);
5918 case KEY_getpriority:
5919 LOP(OP_GETPRIORITY,XTERM);
5921 case KEY_getprotobyname:
5924 case KEY_getprotobynumber:
5925 LOP(OP_GPBYNUMBER,XTERM);
5927 case KEY_getprotoent:
5939 case KEY_getpeername:
5940 UNI(OP_GETPEERNAME);
5942 case KEY_gethostbyname:
5945 case KEY_gethostbyaddr:
5946 LOP(OP_GHBYADDR,XTERM);
5948 case KEY_gethostent:
5951 case KEY_getnetbyname:
5954 case KEY_getnetbyaddr:
5955 LOP(OP_GNBYADDR,XTERM);
5960 case KEY_getservbyname:
5961 LOP(OP_GSBYNAME,XTERM);
5963 case KEY_getservbyport:
5964 LOP(OP_GSBYPORT,XTERM);
5966 case KEY_getservent:
5969 case KEY_getsockname:
5970 UNI(OP_GETSOCKNAME);
5972 case KEY_getsockopt:
5973 LOP(OP_GSOCKOPT,XTERM);
5988 yylval.ival = CopLINE(PL_curcop);
5999 yylval.ival = CopLINE(PL_curcop);
6003 LOP(OP_INDEX,XTERM);
6009 LOP(OP_IOCTL,XTERM);
6021 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6053 LOP(OP_LISTEN,XTERM);
6062 s = scan_pat(s,OP_MATCH);
6063 TERM(sublex_start());
6066 LOP(OP_MAPSTART, XREF);
6069 LOP(OP_MKDIR,XTERM);
6072 LOP(OP_MSGCTL,XTERM);
6075 LOP(OP_MSGGET,XTERM);
6078 LOP(OP_MSGRCV,XTERM);
6081 LOP(OP_MSGSND,XTERM);
6088 if (isIDFIRST_lazy_if(s,UTF)) {
6092 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
6093 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
6095 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
6096 if (!PL_in_my_stash) {
6099 my_snprintf(tmpbuf, sizeof(tmpbuf), "No such class %.1000s", PL_tokenbuf);
6103 if (PL_madskills) { /* just add type to declarator token */
6104 sv_catsv(PL_thistoken, PL_nextwhite);
6106 sv_catpvn(PL_thistoken, start, s - start);
6114 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6121 s = tokenize_use(0, s);
6125 if (*s == '(' || (s = SKIPSPACE1(s), *s == '('))
6132 if (isIDFIRST_lazy_if(s,UTF)) {
6134 for (d = s; isALNUM_lazy_if(d,UTF);)
6136 for (t=d; isSPACE(*t);)
6138 if ( *t && strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE)
6140 && !(t[0] == '=' && t[1] == '>')
6142 int parms_len = (int)(d-s);
6143 Perl_warner(aTHX_ packWARN(WARN_PRECEDENCE),
6144 "Precedence problem: open %.*s should be open(%.*s)",
6145 parms_len, s, parms_len, s);
6151 yylval.ival = OP_OR;
6161 LOP(OP_OPEN_DIR,XTERM);
6164 checkcomma(s,PL_tokenbuf,"filehandle");
6168 checkcomma(s,PL_tokenbuf,"filehandle");
6187 s = force_word(s,WORD,FALSE,TRUE,FALSE);
6191 LOP(OP_PIPE_OP,XTERM);
6194 s = scan_str(s,!!PL_madskills,FALSE);
6197 yylval.ival = OP_CONST;
6198 TERM(sublex_start());
6204 s = scan_str(s,!!PL_madskills,FALSE);
6207 PL_expect = XOPERATOR;
6209 if (SvCUR(PL_lex_stuff)) {
6212 d = SvPV_force(PL_lex_stuff, len);
6214 for (; isSPACE(*d) && len; --len, ++d)
6219 if (!warned && ckWARN(WARN_QW)) {
6220 for (; !isSPACE(*d) && len; --len, ++d) {
6222 Perl_warner(aTHX_ packWARN(WARN_QW),
6223 "Possible attempt to separate words with commas");
6226 else if (*d == '#') {
6227 Perl_warner(aTHX_ packWARN(WARN_QW),
6228 "Possible attempt to put comments in qw() list");
6234 for (; !isSPACE(*d) && len; --len, ++d)
6237 sv = newSVpvn(b, d-b);
6238 if (DO_UTF8(PL_lex_stuff))
6240 words = append_elem(OP_LIST, words,
6241 newSVOP(OP_CONST, 0, tokeq(sv)));
6245 start_force(PL_curforce);
6246 NEXTVAL_NEXTTOKE.opval = words;
6251 SvREFCNT_dec(PL_lex_stuff);
6252 PL_lex_stuff = NULL;
6258 s = scan_str(s,!!PL_madskills,FALSE);
6261 yylval.ival = OP_STRINGIFY;
6262 if (SvIVX(PL_lex_stuff) == '\'')
6263 SvIV_set(PL_lex_stuff, 0); /* qq'$foo' should intepolate */
6264 TERM(sublex_start());
6267 s = scan_pat(s,OP_QR);
6268 TERM(sublex_start());
6271 s = scan_str(s,!!PL_madskills,FALSE);
6274 yylval.ival = OP_BACKTICK;
6276 TERM(sublex_start());
6284 s = force_version(s, FALSE);
6286 else if (*s != 'v' || !isDIGIT(s[1])
6287 || (s = force_version(s, TRUE), *s == 'v'))
6289 *PL_tokenbuf = '\0';
6290 s = force_word(s,WORD,TRUE,TRUE,FALSE);
6291 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
6292 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
6294 yyerror("<> should be quotes");
6296 if (orig_keyword == KEY_require) {
6304 PL_last_uni = PL_oldbufptr;
6305 PL_last_lop_op = OP_REQUIRE;
6307 return REPORT( (int)REQUIRE );
6313 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6317 LOP(OP_RENAME,XTERM);
6326 LOP(OP_RINDEX,XTERM);
6336 UNIDOR(OP_READLINE);
6349 LOP(OP_REVERSE,XTERM);
6352 UNIDOR(OP_READLINK);
6360 TERM(sublex_start());
6362 TOKEN(1); /* force error */
6365 checkcomma(s,PL_tokenbuf,"filehandle");
6375 LOP(OP_SELECT,XTERM);
6381 LOP(OP_SEMCTL,XTERM);
6384 LOP(OP_SEMGET,XTERM);
6387 LOP(OP_SEMOP,XTERM);
6393 LOP(OP_SETPGRP,XTERM);
6395 case KEY_setpriority:
6396 LOP(OP_SETPRIORITY,XTERM);
6398 case KEY_sethostent:
6404 case KEY_setservent:
6407 case KEY_setprotoent:
6417 LOP(OP_SEEKDIR,XTERM);
6419 case KEY_setsockopt:
6420 LOP(OP_SSOCKOPT,XTERM);
6426 LOP(OP_SHMCTL,XTERM);
6429 LOP(OP_SHMGET,XTERM);
6432 LOP(OP_SHMREAD,XTERM);
6435 LOP(OP_SHMWRITE,XTERM);
6438 LOP(OP_SHUTDOWN,XTERM);
6447 LOP(OP_SOCKET,XTERM);
6449 case KEY_socketpair:
6450 LOP(OP_SOCKPAIR,XTERM);
6453 checkcomma(s,PL_tokenbuf,"subroutine name");
6455 if (*s == ';' || *s == ')') /* probably a close */
6456 Perl_croak(aTHX_ "sort is now a reserved word");
6458 s = force_word(s,WORD,TRUE,TRUE,FALSE);
6462 LOP(OP_SPLIT,XTERM);
6465 LOP(OP_SPRINTF,XTERM);
6468 LOP(OP_SPLICE,XTERM);
6483 LOP(OP_SUBSTR,XTERM);
6489 char tmpbuf[sizeof PL_tokenbuf];
6490 SSize_t tboffset = 0;
6491 expectation attrful;
6492 bool have_name, have_proto, bad_proto;
6493 const int key = tmp;
6498 char *tstart = SvPVX(PL_linestr) + PL_realtokenstart;
6499 SV *subtoken = newSVpvn(tstart, s - tstart);
6503 s = SKIPSPACE2(s,tmpwhite);
6508 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
6509 (*s == ':' && s[1] == ':'))
6516 attrful = XATTRBLOCK;
6517 /* remember buffer pos'n for later force_word */
6518 tboffset = s - PL_oldbufptr;
6519 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
6522 nametoke = newSVpvn(s, d - s);
6524 if (strchr(tmpbuf, ':'))
6525 sv_setpv(PL_subname, tmpbuf);
6527 sv_setsv(PL_subname,PL_curstname);
6528 sv_catpvs(PL_subname,"::");
6529 sv_catpvn(PL_subname,tmpbuf,len);
6536 CURMAD('X', nametoke);
6537 CURMAD('_', tmpwhite);
6538 (void) force_word(PL_oldbufptr + tboffset, WORD,
6541 s = SKIPSPACE2(d,tmpwhite);
6548 Perl_croak(aTHX_ "Missing name in \"my sub\"");
6549 PL_expect = XTERMBLOCK;
6550 attrful = XATTRTERM;
6551 sv_setpvn(PL_subname,"?",1);
6555 if (key == KEY_format) {
6557 PL_lex_formbrack = PL_lex_brackets + 1;
6559 PL_thistoken = subtoken;
6563 (void) force_word(PL_oldbufptr + tboffset, WORD,
6569 /* Look for a prototype */
6573 s = scan_str(s,!!PL_madskills,FALSE);
6575 Perl_croak(aTHX_ "Prototype not terminated");
6576 /* strip spaces and check for bad characters */
6577 d = SvPVX(PL_lex_stuff);
6580 for (p = d; *p; ++p) {
6583 if (!strchr("$@%*;[]&\\", *p))
6588 if (bad_proto && ckWARN(WARN_SYNTAX))
6589 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
6590 "Illegal character in prototype for %"SVf" : %s",
6591 (void*)PL_subname, d);
6592 SvCUR_set(PL_lex_stuff, tmp);
6597 CURMAD('q', PL_thisopen);
6598 CURMAD('_', tmpwhite);
6599 CURMAD('=', PL_thisstuff);
6600 CURMAD('Q', PL_thisclose);
6601 NEXTVAL_NEXTTOKE.opval =
6602 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
6603 PL_lex_stuff = Nullsv;
6606 s = SKIPSPACE2(s,tmpwhite);
6614 if (*s == ':' && s[1] != ':')
6615 PL_expect = attrful;
6616 else if (*s != '{' && key == KEY_sub) {
6618 Perl_croak(aTHX_ "Illegal declaration of anonymous subroutine");
6620 Perl_croak(aTHX_ "Illegal declaration of subroutine %"SVf, (void*)PL_subname);
6627 curmad('^', newSVpvn("",0));
6628 CURMAD('_', tmpwhite);
6632 PL_thistoken = subtoken;
6635 NEXTVAL_NEXTTOKE.opval =
6636 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
6637 PL_lex_stuff = NULL;
6642 sv_setpv(PL_subname,
6644 (PL_curstash ? "__ANON__" : "__ANON__::__ANON__"));
6648 (void) force_word(PL_oldbufptr + tboffset, WORD,
6658 LOP(OP_SYSTEM,XREF);
6661 LOP(OP_SYMLINK,XTERM);
6664 LOP(OP_SYSCALL,XTERM);
6667 LOP(OP_SYSOPEN,XTERM);
6670 LOP(OP_SYSSEEK,XTERM);
6673 LOP(OP_SYSREAD,XTERM);
6676 LOP(OP_SYSWRITE,XTERM);
6680 TERM(sublex_start());
6701 LOP(OP_TRUNCATE,XTERM);
6713 yylval.ival = CopLINE(PL_curcop);
6717 yylval.ival = CopLINE(PL_curcop);
6721 LOP(OP_UNLINK,XTERM);
6727 LOP(OP_UNPACK,XTERM);
6730 LOP(OP_UTIME,XTERM);
6736 LOP(OP_UNSHIFT,XTERM);
6739 s = tokenize_use(1, s);
6749 yylval.ival = CopLINE(PL_curcop);
6753 yylval.ival = CopLINE(PL_curcop);
6757 PL_hints |= HINT_BLOCK_SCOPE;
6764 LOP(OP_WAITPID,XTERM);
6773 ctl_l[0] = toCTRL('L');
6775 gv_fetchpvn_flags(ctl_l, 1, GV_ADD|GV_NOTQUAL, SVt_PV);
6778 /* Make sure $^L is defined */
6779 gv_fetchpvs("\f", GV_ADD|GV_NOTQUAL, SVt_PV);
6784 if (PL_expect == XOPERATOR)
6790 yylval.ival = OP_XOR;
6795 TERM(sublex_start());
6800 #pragma segment Main
6804 S_pending_ident(pTHX)
6809 /* pit holds the identifier we read and pending_ident is reset */
6810 char pit = PL_pending_ident;
6811 PL_pending_ident = 0;
6813 /* PL_realtokenstart = realtokenend = PL_bufptr - SvPVX(PL_linestr); */
6814 DEBUG_T({ PerlIO_printf(Perl_debug_log,
6815 "### Pending identifier '%s'\n", PL_tokenbuf); });
6817 /* if we're in a my(), we can't allow dynamics here.
6818 $foo'bar has already been turned into $foo::bar, so
6819 just check for colons.
6821 if it's a legal name, the OP is a PADANY.
6824 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
6825 if (strchr(PL_tokenbuf,':'))
6826 yyerror(Perl_form(aTHX_ "No package name allowed for "
6827 "variable %s in \"our\"",
6829 tmp = allocmy(PL_tokenbuf);
6832 if (strchr(PL_tokenbuf,':'))
6833 yyerror(Perl_form(aTHX_ PL_no_myglob,
6834 PL_in_my == KEY_my ? "my" : "state", PL_tokenbuf));
6836 yylval.opval = newOP(OP_PADANY, 0);
6837 yylval.opval->op_targ = allocmy(PL_tokenbuf);
6843 build the ops for accesses to a my() variable.
6845 Deny my($a) or my($b) in a sort block, *if* $a or $b is
6846 then used in a comparison. This catches most, but not
6847 all cases. For instance, it catches
6848 sort { my($a); $a <=> $b }
6850 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
6851 (although why you'd do that is anyone's guess).
6854 if (!strchr(PL_tokenbuf,':')) {
6856 tmp = pad_findmy(PL_tokenbuf);
6857 if (tmp != NOT_IN_PAD) {
6858 /* might be an "our" variable" */
6859 if (PAD_COMPNAME_FLAGS_isOUR(tmp)) {
6860 /* build ops for a bareword */
6861 HV * const stash = PAD_COMPNAME_OURSTASH(tmp);
6862 HEK * const stashname = HvNAME_HEK(stash);
6863 SV * const sym = newSVhek(stashname);
6864 sv_catpvs(sym, "::");
6865 sv_catpv(sym, PL_tokenbuf+1);
6866 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
6867 yylval.opval->op_private = OPpCONST_ENTERED;
6870 ? (GV_ADDMULTI | GV_ADDINEVAL)
6873 ((PL_tokenbuf[0] == '$') ? SVt_PV
6874 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
6879 /* if it's a sort block and they're naming $a or $b */
6880 if (PL_last_lop_op == OP_SORT &&
6881 PL_tokenbuf[0] == '$' &&
6882 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
6885 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
6886 d < PL_bufend && *d != '\n';
6889 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
6890 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
6896 yylval.opval = newOP(OP_PADANY, 0);
6897 yylval.opval->op_targ = tmp;
6903 Whine if they've said @foo in a doublequoted string,
6904 and @foo isn't a variable we can find in the symbol
6907 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
6908 GV *gv = gv_fetchpv(PL_tokenbuf+1, 0, SVt_PVAV);
6909 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
6910 && ckWARN(WARN_AMBIGUOUS))
6912 /* Downgraded from fatal to warning 20000522 mjd */
6913 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
6914 "Possible unintended interpolation of %s in string",
6919 /* build ops for a bareword */
6920 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
6921 yylval.opval->op_private = OPpCONST_ENTERED;
6924 /* If the identifier refers to a stash, don't autovivify it.
6925 * Change 24660 had the side effect of causing symbol table
6926 * hashes to always be defined, even if they were freshly
6927 * created and the only reference in the entire program was
6928 * the single statement with the defined %foo::bar:: test.
6929 * It appears that all code in the wild doing this actually
6930 * wants to know whether sub-packages have been loaded, so
6931 * by avoiding auto-vivifying symbol tables, we ensure that
6932 * defined %foo::bar:: continues to be false, and the existing
6933 * tests still give the expected answers, even though what
6934 * they're actually testing has now changed subtly.
6936 (*PL_tokenbuf == '%' && *(d = PL_tokenbuf + strlen(PL_tokenbuf) - 1) == ':' && d[-1] == ':'
6938 : PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : GV_ADD),
6939 ((PL_tokenbuf[0] == '$') ? SVt_PV
6940 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
6946 * The following code was generated by perl_keyword.pl.
6950 Perl_keyword (pTHX_ const char *name, I32 len)
6955 case 1: /* 5 tokens of length 1 */
6987 case 2: /* 18 tokens of length 2 */
7133 case 3: /* 29 tokens of length 3 */
7137 if (name[1] == 'N' &&
7200 if (name[1] == 'i' &&
7222 return (FEATURE_IS_ENABLED("err") ? -KEY_err : 0);
7240 if (name[1] == 'o' &&
7249 if (name[1] == 'e' &&
7258 if (name[1] == 'n' &&
7267 if (name[1] == 'o' &&
7276 if (name[1] == 'a' &&
7285 if (name[1] == 'o' &&
7347 if (name[1] == 'e' &&
7361 return (FEATURE_IS_ENABLED("say") ? -KEY_say : 0);
7387 if (name[1] == 'i' &&
7396 if (name[1] == 's' &&
7405 if (name[1] == 'e' &&
7414 if (name[1] == 'o' &&
7426 case 4: /* 41 tokens of length 4 */
7430 if (name[1] == 'O' &&
7440 if (name[1] == 'N' &&
7450 if (name[1] == 'i' &&
7460 if (name[1] == 'h' &&
7470 if (name[1] == 'u' &&
7483 if (name[2] == 'c' &&
7492 if (name[2] == 's' &&
7501 if (name[2] == 'a' &&
7537 if (name[1] == 'o' &&
7550 if (name[2] == 't' &&
7559 if (name[2] == 'o' &&
7568 if (name[2] == 't' &&
7577 if (name[2] == 'e' &&
7590 if (name[1] == 'o' &&
7603 if (name[2] == 'y' &&
7612 if (name[2] == 'l' &&
7628 if (name[2] == 's' &&
7637 if (name[2] == 'n' &&
7646 if (name[2] == 'c' &&
7659 if (name[1] == 'e' &&
7669 if (name[1] == 'p' &&
7682 if (name[2] == 'c' &&
7691 if (name[2] == 'p' &&
7700 if (name[2] == 's' &&
7716 if (name[2] == 'n' &&
7786 if (name[2] == 'r' &&
7795 if (name[2] == 'r' &&
7804 if (name[2] == 'a' &&
7820 if (name[2] == 'l' &&
7882 if (name[2] == 'e' &&
7885 return (FEATURE_IS_ENABLED("switch") ? KEY_when : 0);
7898 case 5: /* 39 tokens of length 5 */
7902 if (name[1] == 'E' &&
7913 if (name[1] == 'H' &&
7927 if (name[2] == 'a' &&
7937 if (name[2] == 'a' &&
7954 if (name[2] == 'e' &&
7964 if (name[2] == 'e' &&
7968 return (FEATURE_IS_ENABLED("switch") ? -KEY_break : 0);
7984 if (name[3] == 'i' &&
7993 if (name[3] == 'o' &&
8029 if (name[2] == 'o' &&
8039 if (name[2] == 'y' &&
8053 if (name[1] == 'l' &&
8067 if (name[2] == 'n' &&
8077 if (name[2] == 'o' &&
8091 if (name[1] == 'i' &&
8096 return (FEATURE_IS_ENABLED("switch") ? KEY_given : 0);
8105 if (name[2] == 'd' &&
8115 if (name[2] == 'c' &&
8132 if (name[2] == 'c' &&
8142 if (name[2] == 't' &&
8156 if (name[1] == 'k' &&
8167 if (name[1] == 'r' &&
8181 if (name[2] == 's' &&
8191 if (name[2] == 'd' &&
8208 if (name[2] == 'm' &&
8218 if (name[2] == 'i' &&
8228 if (name[2] == 'e' &&
8238 if (name[2] == 'l' &&
8248 if (name[2] == 'a' &&
8261 if (name[3] == 't' &&
8264 return (FEATURE_IS_ENABLED("state") ? KEY_state : 0);
8270 if (name[3] == 'd' &&
8287 if (name[1] == 'i' &&
8301 if (name[2] == 'a' &&
8314 if (name[3] == 'e' &&
8349 if (name[2] == 'i' &&
8366 if (name[2] == 'i' &&
8376 if (name[2] == 'i' &&
8393 case 6: /* 33 tokens of length 6 */
8397 if (name[1] == 'c' &&
8412 if (name[2] == 'l' &&
8423 if (name[2] == 'r' &&
8438 if (name[1] == 'e' &&
8453 if (name[2] == 's' &&
8458 if(ckWARN_d(WARN_SYNTAX))
8459 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "elseif should be elsif");
8465 if (name[2] == 'i' &&
8483 if (name[2] == 'l' &&
8494 if (name[2] == 'r' &&
8509 if (name[1] == 'm' &&
8524 if (name[2] == 'n' &&
8535 if (name[2] == 's' &&
8550 if (name[1] == 's' &&
8556 if (name[4] == 't' &&
8565 if (name[4] == 'e' &&
8574 if (name[4] == 'c' &&
8583 if (name[4] == 'n' &&
8599 if (name[1] == 'r' &&
8617 if (name[3] == 'a' &&
8627 if (name[3] == 'u' &&
8641 if (name[2] == 'n' &&
8659 if (name[2] == 'a' &&
8673 if (name[3] == 'e' &&
8686 if (name[4] == 't' &&
8695 if (name[4] == 'e' &&
8717 if (name[4] == 't' &&
8726 if (name[4] == 'e' &&
8742 if (name[2] == 'c' &&
8753 if (name[2] == 'l' &&
8764 if (name[2] == 'b' &&
8775 if (name[2] == 's' &&
8798 if (name[4] == 's' &&
8807 if (name[4] == 'n' &&
8820 if (name[3] == 'a' &&
8837 if (name[1] == 'a' &&
8852 case 7: /* 29 tokens of length 7 */
8856 if (name[1] == 'E' &&
8869 if (name[1] == '_' &&
8882 if (name[1] == 'i' &&
8889 return -KEY_binmode;
8895 if (name[1] == 'o' &&
8902 return -KEY_connect;
8911 if (name[2] == 'm' &&
8917 return -KEY_dbmopen;
8928 if (name[4] == 'u' &&
8932 return (FEATURE_IS_ENABLED("switch") ? KEY_default : 0);
8938 if (name[4] == 'n' &&
8959 if (name[1] == 'o' &&
8972 if (name[1] == 'e' &&
8979 if (name[5] == 'r' &&
8982 return -KEY_getpgrp;
8988 if (name[5] == 'i' &&
8991 return -KEY_getppid;
9004 if (name[1] == 'c' &&
9011 return -KEY_lcfirst;
9017 if (name[1] == 'p' &&
9024 return -KEY_opendir;
9030 if (name[1] == 'a' &&
9048 if (name[3] == 'd' &&
9053 return -KEY_readdir;
9059 if (name[3] == 'u' &&
9070 if (name[3] == 'e' &&
9075 return -KEY_reverse;
9094 if (name[3] == 'k' &&
9099 return -KEY_seekdir;
9105 if (name[3] == 'p' &&
9110 return -KEY_setpgrp;
9120 if (name[2] == 'm' &&
9126 return -KEY_shmread;
9132 if (name[2] == 'r' &&
9138 return -KEY_sprintf;
9147 if (name[3] == 'l' &&
9152 return -KEY_symlink;
9161 if (name[4] == 'a' &&
9165 return -KEY_syscall;
9171 if (name[4] == 'p' &&
9175 return -KEY_sysopen;
9181 if (name[4] == 'e' &&
9185 return -KEY_sysread;
9191 if (name[4] == 'e' &&
9195 return -KEY_sysseek;
9213 if (name[1] == 'e' &&
9220 return -KEY_telldir;
9229 if (name[2] == 'f' &&
9235 return -KEY_ucfirst;
9241 if (name[2] == 's' &&
9247 return -KEY_unshift;
9257 if (name[1] == 'a' &&
9264 return -KEY_waitpid;
9273 case 8: /* 26 tokens of length 8 */
9277 if (name[1] == 'U' &&
9285 return KEY_AUTOLOAD;
9296 if (name[3] == 'A' &&
9302 return KEY___DATA__;
9308 if (name[3] == 'I' &&
9314 return -KEY___FILE__;
9320 if (name[3] == 'I' &&
9326 return -KEY___LINE__;
9342 if (name[2] == 'o' &&
9349 return -KEY_closedir;
9355 if (name[2] == 'n' &&
9362 return -KEY_continue;
9372 if (name[1] == 'b' &&
9380 return -KEY_dbmclose;
9386 if (name[1] == 'n' &&
9392 if (name[4] == 'r' &&
9397 return -KEY_endgrent;
9403 if (name[4] == 'w' &&
9408 return -KEY_endpwent;
9421 if (name[1] == 'o' &&
9429 return -KEY_formline;
9435 if (name[1] == 'e' &&
9446 if (name[6] == 'n' &&
9449 return -KEY_getgrent;
9455 if (name[6] == 'i' &&
9458 return -KEY_getgrgid;
9464 if (name[6] == 'a' &&
9467 return -KEY_getgrnam;
9480 if (name[4] == 'o' &&
9485 return -KEY_getlogin;
9496 if (name[6] == 'n' &&
9499 return -KEY_getpwent;
9505 if (name[6] == 'a' &&
9508 return -KEY_getpwnam;
9514 if (name[6] == 'i' &&
9517 return -KEY_getpwuid;
9537 if (name[1] == 'e' &&
9544 if (name[5] == 'i' &&
9551 return -KEY_readline;
9556 return -KEY_readlink;
9567 if (name[5] == 'i' &&
9571 return -KEY_readpipe;
9592 if (name[4] == 'r' &&
9597 return -KEY_setgrent;
9603 if (name[4] == 'w' &&
9608 return -KEY_setpwent;
9624 if (name[3] == 'w' &&
9630 return -KEY_shmwrite;
9636 if (name[3] == 't' &&
9642 return -KEY_shutdown;
9652 if (name[2] == 's' &&
9659 return -KEY_syswrite;
9669 if (name[1] == 'r' &&
9677 return -KEY_truncate;
9686 case 9: /* 8 tokens of length 9 */
9690 if (name[1] == 'n' &&
9699 return -KEY_endnetent;
9705 if (name[1] == 'e' &&
9714 return -KEY_getnetent;
9720 if (name[1] == 'o' &&
9729 return -KEY_localtime;
9735 if (name[1] == 'r' &&
9744 return KEY_prototype;
9750 if (name[1] == 'u' &&
9759 return -KEY_quotemeta;
9765 if (name[1] == 'e' &&
9774 return -KEY_rewinddir;
9780 if (name[1] == 'e' &&
9789 return -KEY_setnetent;
9795 if (name[1] == 'a' &&
9804 return -KEY_wantarray;
9813 case 10: /* 9 tokens of length 10 */
9817 if (name[1] == 'n' &&
9823 if (name[4] == 'o' &&
9830 return -KEY_endhostent;
9836 if (name[4] == 'e' &&
9843 return -KEY_endservent;
9856 if (name[1] == 'e' &&
9862 if (name[4] == 'o' &&
9869 return -KEY_gethostent;
9878 if (name[5] == 'r' &&
9884 return -KEY_getservent;
9890 if (name[5] == 'c' &&
9896 return -KEY_getsockopt;
9921 if (name[4] == 'o' &&
9928 return -KEY_sethostent;
9937 if (name[5] == 'r' &&
9943 return -KEY_setservent;
9949 if (name[5] == 'c' &&
9955 return -KEY_setsockopt;
9972 if (name[2] == 'c' &&
9981 return -KEY_socketpair;
9994 case 11: /* 8 tokens of length 11 */
9998 if (name[1] == '_' &&
10008 { /* __PACKAGE__ */
10009 return -KEY___PACKAGE__;
10015 if (name[1] == 'n' &&
10025 { /* endprotoent */
10026 return -KEY_endprotoent;
10032 if (name[1] == 'e' &&
10041 if (name[5] == 'e' &&
10047 { /* getpeername */
10048 return -KEY_getpeername;
10057 if (name[6] == 'o' &&
10062 { /* getpriority */
10063 return -KEY_getpriority;
10069 if (name[6] == 't' &&
10074 { /* getprotoent */
10075 return -KEY_getprotoent;
10089 if (name[4] == 'o' &&
10096 { /* getsockname */
10097 return -KEY_getsockname;
10110 if (name[1] == 'e' &&
10118 if (name[6] == 'o' &&
10123 { /* setpriority */
10124 return -KEY_setpriority;
10130 if (name[6] == 't' &&
10135 { /* setprotoent */
10136 return -KEY_setprotoent;
10152 case 12: /* 2 tokens of length 12 */
10153 if (name[0] == 'g' &&
10165 if (name[9] == 'd' &&
10168 { /* getnetbyaddr */
10169 return -KEY_getnetbyaddr;
10175 if (name[9] == 'a' &&
10178 { /* getnetbyname */
10179 return -KEY_getnetbyname;
10191 case 13: /* 4 tokens of length 13 */
10192 if (name[0] == 'g' &&
10199 if (name[4] == 'o' &&
10208 if (name[10] == 'd' &&
10211 { /* gethostbyaddr */
10212 return -KEY_gethostbyaddr;
10218 if (name[10] == 'a' &&
10221 { /* gethostbyname */
10222 return -KEY_gethostbyname;
10235 if (name[4] == 'e' &&
10244 if (name[10] == 'a' &&
10247 { /* getservbyname */
10248 return -KEY_getservbyname;
10254 if (name[10] == 'o' &&
10257 { /* getservbyport */
10258 return -KEY_getservbyport;
10277 case 14: /* 1 tokens of length 14 */
10278 if (name[0] == 'g' &&
10292 { /* getprotobyname */
10293 return -KEY_getprotobyname;
10298 case 16: /* 1 tokens of length 16 */
10299 if (name[0] == 'g' &&
10315 { /* getprotobynumber */
10316 return -KEY_getprotobynumber;
10330 S_checkcomma(pTHX_ const char *s, const char *name, const char *what)
10334 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
10335 if (ckWARN(WARN_SYNTAX)) {
10338 for (w = s+2; *w && level; w++) {
10341 else if (*w == ')')
10344 while (isSPACE(*w))
10346 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
10347 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10348 "%s (...) interpreted as function",name);
10351 while (s < PL_bufend && isSPACE(*s))
10355 while (s < PL_bufend && isSPACE(*s))
10357 if (isIDFIRST_lazy_if(s,UTF)) {
10358 const char * const w = s++;
10359 while (isALNUM_lazy_if(s,UTF))
10361 while (s < PL_bufend && isSPACE(*s))
10365 if (keyword(w, s - w))
10368 gv = gv_fetchpvn_flags(w, s - w, 0, SVt_PVCV);
10369 if (gv && GvCVu(gv))
10371 Perl_croak(aTHX_ "No comma allowed after %s", what);
10376 /* Either returns sv, or mortalizes sv and returns a new SV*.
10377 Best used as sv=new_constant(..., sv, ...).
10378 If s, pv are NULL, calls subroutine with one argument,
10379 and type is used with error messages only. */
10382 S_new_constant(pTHX_ const char *s, STRLEN len, const char *key, SV *sv, SV *pv,
10386 HV * const table = GvHV(PL_hintgv); /* ^H */
10390 const char *why1 = "", *why2 = "", *why3 = "";
10392 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
10395 why2 = (const char *)
10396 (strEQ(key,"charnames")
10397 ? "(possibly a missing \"use charnames ...\")"
10399 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
10400 (type ? type: "undef"), why2);
10402 /* This is convoluted and evil ("goto considered harmful")
10403 * but I do not understand the intricacies of all the different
10404 * failure modes of %^H in here. The goal here is to make
10405 * the most probable error message user-friendly. --jhi */
10410 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
10411 (type ? type: "undef"), why1, why2, why3);
10413 yyerror(SvPVX_const(msg));
10417 cvp = hv_fetch(table, key, strlen(key), FALSE);
10418 if (!cvp || !SvOK(*cvp)) {
10421 why3 = "} is not defined";
10424 sv_2mortal(sv); /* Parent created it permanently */
10427 pv = sv_2mortal(newSVpvn(s, len));
10429 typesv = sv_2mortal(newSVpv(type, 0));
10431 typesv = &PL_sv_undef;
10433 PUSHSTACKi(PERLSI_OVERLOAD);
10445 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
10449 /* Check the eval first */
10450 if (!PL_in_eval && SvTRUE(ERRSV)) {
10451 sv_catpvs(ERRSV, "Propagated");
10452 yyerror(SvPV_nolen_const(ERRSV)); /* Duplicates the message inside eval */
10454 res = SvREFCNT_inc_simple(sv);
10458 SvREFCNT_inc_simple_void(res);
10467 why1 = "Call to &{$^H{";
10469 why3 = "}} did not return a defined value";
10477 /* Returns a NUL terminated string, with the length of the string written to
10481 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
10484 register char *d = dest;
10485 register char * const e = d + destlen - 3; /* two-character token, ending NUL */
10488 Perl_croak(aTHX_ ident_too_long);
10489 if (isALNUM(*s)) /* UTF handled below */
10491 else if (allow_package && (*s == '\'') && isIDFIRST_lazy_if(s+1,UTF)) {
10496 else if (allow_package && (s[0] == ':') && (s[1] == ':') && (s[2] != '$')) {
10500 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
10501 char *t = s + UTF8SKIP(s);
10503 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
10507 Perl_croak(aTHX_ ident_too_long);
10508 Copy(s, d, len, char);
10521 S_scan_ident(pTHX_ register char *s, register const char *send, char *dest, STRLEN destlen, I32 ck_uni)
10524 char *bracket = NULL;
10526 register char *d = dest;
10527 register char * const e = d + destlen + 3; /* two-character token, ending NUL */
10532 while (isDIGIT(*s)) {
10534 Perl_croak(aTHX_ ident_too_long);
10541 Perl_croak(aTHX_ ident_too_long);
10542 if (isALNUM(*s)) /* UTF handled below */
10544 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
10549 else if (*s == ':' && s[1] == ':') {
10553 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
10554 char *t = s + UTF8SKIP(s);
10555 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
10557 if (d + (t - s) > e)
10558 Perl_croak(aTHX_ ident_too_long);
10559 Copy(s, d, t - s, char);
10570 if (PL_lex_state != LEX_NORMAL)
10571 PL_lex_state = LEX_INTERPENDMAYBE;
10574 if (*s == '$' && s[1] &&
10575 (isALNUM_lazy_if(s+1,UTF) || s[1] == '$' || s[1] == '{' || strnEQ(s+1,"::",2)) )
10588 if (*d == '^' && *s && isCONTROLVAR(*s)) {
10593 if (isSPACE(s[-1])) {
10595 const char ch = *s++;
10596 if (!SPACE_OR_TAB(ch)) {
10602 if (isIDFIRST_lazy_if(d,UTF)) {
10606 while ((end < send && isALNUM_lazy_if(end,UTF)) || *end == ':') {
10607 end += UTF8SKIP(end);
10608 while (end < send && UTF8_IS_CONTINUED(*end) && is_utf8_mark((U8*)end))
10609 end += UTF8SKIP(end);
10611 Copy(s, d, end - s, char);
10616 while ((isALNUM(*s) || *s == ':') && d < e)
10619 Perl_croak(aTHX_ ident_too_long);
10622 while (s < send && SPACE_OR_TAB(*s))
10624 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
10625 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
10626 const char * const brack =
10628 ((*s == '[') ? "[...]" : "{...}");
10629 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
10630 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
10631 funny, dest, brack, funny, dest, brack);
10634 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
10638 /* Handle extended ${^Foo} variables
10639 * 1999-02-27 mjd-perl-patch@plover.com */
10640 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
10644 while (isALNUM(*s) && d < e) {
10648 Perl_croak(aTHX_ ident_too_long);
10653 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets) {
10654 PL_lex_state = LEX_INTERPEND;
10657 if (PL_lex_state == LEX_NORMAL) {
10658 if (ckWARN(WARN_AMBIGUOUS) &&
10659 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
10663 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
10664 "Ambiguous use of %c{%s} resolved to %c%s",
10665 funny, dest, funny, dest);
10670 s = bracket; /* let the parser handle it */
10674 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
10675 PL_lex_state = LEX_INTERPEND;
10680 Perl_pmflag(pTHX_ U32* pmfl, int ch)
10682 PERL_UNUSED_CONTEXT;
10685 else if (ch == 'g')
10686 *pmfl |= PMf_GLOBAL;
10687 else if (ch == 'c')
10688 *pmfl |= PMf_CONTINUE;
10689 else if (ch == 'o')
10691 else if (ch == 'm')
10692 *pmfl |= PMf_MULTILINE;
10693 else if (ch == 's')
10694 *pmfl |= PMf_SINGLELINE;
10695 else if (ch == 'x')
10696 *pmfl |= PMf_EXTENDED;
10700 S_scan_pat(pTHX_ char *start, I32 type)
10704 char *s = scan_str(start,!!PL_madskills,FALSE);
10705 const char * const valid_flags =
10706 (const char *)((type == OP_QR) ? "iomsx" : "iogcmsx");
10713 const char * const delimiter = skipspace(start);
10717 ? "Search pattern not terminated or ternary operator parsed as search pattern"
10718 : "Search pattern not terminated" ));
10721 pm = (PMOP*)newPMOP(type, 0);
10722 if (PL_multi_open == '?')
10723 pm->op_pmflags |= PMf_ONCE;
10727 while (*s && strchr(valid_flags, *s))
10728 pmflag(&pm->op_pmflags,*s++);
10730 if (PL_madskills && modstart != s) {
10731 SV* tmptoken = newSVpvn(modstart, s - modstart);
10732 append_madprops(newMADPROP('m', MAD_SV, tmptoken, 0), (OP*)pm, 0);
10735 /* issue a warning if /c is specified,but /g is not */
10736 if ((pm->op_pmflags & PMf_CONTINUE) && !(pm->op_pmflags & PMf_GLOBAL)
10737 && ckWARN(WARN_REGEXP))
10739 Perl_warner(aTHX_ packWARN(WARN_REGEXP), "Use of /c modifier is meaningless without /g" );
10742 pm->op_pmpermflags = pm->op_pmflags;
10744 PL_lex_op = (OP*)pm;
10745 yylval.ival = OP_MATCH;
10750 S_scan_subst(pTHX_ char *start)
10761 yylval.ival = OP_NULL;
10763 s = scan_str(start,!!PL_madskills,FALSE);
10766 Perl_croak(aTHX_ "Substitution pattern not terminated");
10768 if (s[-1] == PL_multi_open)
10771 if (PL_madskills) {
10772 CURMAD('q', PL_thisopen);
10773 CURMAD('_', PL_thiswhite);
10774 CURMAD('E', PL_thisstuff);
10775 CURMAD('Q', PL_thisclose);
10776 PL_realtokenstart = s - SvPVX(PL_linestr);
10780 first_start = PL_multi_start;
10781 s = scan_str(s,!!PL_madskills,FALSE);
10783 if (PL_lex_stuff) {
10784 SvREFCNT_dec(PL_lex_stuff);
10785 PL_lex_stuff = NULL;
10787 Perl_croak(aTHX_ "Substitution replacement not terminated");
10789 PL_multi_start = first_start; /* so whole substitution is taken together */
10791 pm = (PMOP*)newPMOP(OP_SUBST, 0);
10794 if (PL_madskills) {
10795 CURMAD('z', PL_thisopen);
10796 CURMAD('R', PL_thisstuff);
10797 CURMAD('Z', PL_thisclose);
10807 else if (strchr("iogcmsx", *s))
10808 pmflag(&pm->op_pmflags,*s++);
10814 if (PL_madskills) {
10816 curmad('m', newSVpvn(modstart, s - modstart));
10817 append_madprops(PL_thismad, (OP*)pm, 0);
10821 if ((pm->op_pmflags & PMf_CONTINUE) && ckWARN(WARN_REGEXP)) {
10822 Perl_warner(aTHX_ packWARN(WARN_REGEXP), "Use of /c modifier is meaningless in s///" );
10826 SV * const repl = newSVpvs("");
10828 PL_sublex_info.super_bufptr = s;
10829 PL_sublex_info.super_bufend = PL_bufend;
10831 pm->op_pmflags |= PMf_EVAL;
10833 sv_catpv(repl, (const char *)(es ? "eval " : "do "));
10834 sv_catpvs(repl, "{");
10835 sv_catsv(repl, PL_lex_repl);
10836 if (strchr(SvPVX(PL_lex_repl), '#'))
10837 sv_catpvs(repl, "\n");
10838 sv_catpvs(repl, "}");
10840 SvREFCNT_dec(PL_lex_repl);
10841 PL_lex_repl = repl;
10844 pm->op_pmpermflags = pm->op_pmflags;
10845 PL_lex_op = (OP*)pm;
10846 yylval.ival = OP_SUBST;
10851 S_scan_trans(pTHX_ char *start)
10864 yylval.ival = OP_NULL;
10866 s = scan_str(start,!!PL_madskills,FALSE);
10868 Perl_croak(aTHX_ "Transliteration pattern not terminated");
10870 if (s[-1] == PL_multi_open)
10873 if (PL_madskills) {
10874 CURMAD('q', PL_thisopen);
10875 CURMAD('_', PL_thiswhite);
10876 CURMAD('E', PL_thisstuff);
10877 CURMAD('Q', PL_thisclose);
10878 PL_realtokenstart = s - SvPVX(PL_linestr);
10882 s = scan_str(s,!!PL_madskills,FALSE);
10884 if (PL_lex_stuff) {
10885 SvREFCNT_dec(PL_lex_stuff);
10886 PL_lex_stuff = NULL;
10888 Perl_croak(aTHX_ "Transliteration replacement not terminated");
10890 if (PL_madskills) {
10891 CURMAD('z', PL_thisopen);
10892 CURMAD('R', PL_thisstuff);
10893 CURMAD('Z', PL_thisclose);
10896 complement = del = squash = 0;
10903 complement = OPpTRANS_COMPLEMENT;
10906 del = OPpTRANS_DELETE;
10909 squash = OPpTRANS_SQUASH;
10918 Newx(tbl, complement&&!del?258:256, short);
10919 o = newPVOP(OP_TRANS, 0, (char*)tbl);
10920 o->op_private &= ~OPpTRANS_ALL;
10921 o->op_private |= del|squash|complement|
10922 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
10923 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
10926 yylval.ival = OP_TRANS;
10929 if (PL_madskills) {
10931 curmad('m', newSVpvn(modstart, s - modstart));
10932 append_madprops(PL_thismad, o, 0);
10941 S_scan_heredoc(pTHX_ register char *s)
10945 I32 op_type = OP_SCALAR;
10949 const char *found_newline;
10953 const int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
10955 I32 stuffstart = s - SvPVX(PL_linestr);
10958 PL_realtokenstart = -1;
10963 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
10967 while (SPACE_OR_TAB(*peek))
10969 if (*peek == '`' || *peek == '\'' || *peek =='"') {
10972 s = delimcpy(d, e, s, PL_bufend, term, &len);
10982 if (!isALNUM_lazy_if(s,UTF))
10983 deprecate_old("bare << to mean <<\"\"");
10984 for (; isALNUM_lazy_if(s,UTF); s++) {
10989 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
10990 Perl_croak(aTHX_ "Delimiter for here document is too long");
10993 len = d - PL_tokenbuf;
10996 if (PL_madskills) {
10997 tstart = PL_tokenbuf + !outer;
10998 PL_thisclose = newSVpvn(tstart, len - !outer);
10999 tstart = SvPVX(PL_linestr) + stuffstart;
11000 PL_thisopen = newSVpvn(tstart, s - tstart);
11001 stuffstart = s - SvPVX(PL_linestr);
11004 #ifndef PERL_STRICT_CR
11005 d = strchr(s, '\r');
11007 char * const olds = s;
11009 while (s < PL_bufend) {
11015 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
11024 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
11031 if ( outer || !(found_newline = (char*)memchr((void*)s, '\n', PL_bufend - s)) ) {
11032 herewas = newSVpvn(s,PL_bufend-s);
11036 herewas = newSVpvn(s-1,found_newline-s+1);
11039 herewas = newSVpvn(s,found_newline-s);
11043 if (PL_madskills) {
11044 tstart = SvPVX(PL_linestr) + stuffstart;
11046 sv_catpvn(PL_thisstuff, tstart, s - tstart);
11048 PL_thisstuff = newSVpvn(tstart, s - tstart);
11051 s += SvCUR(herewas);
11054 stuffstart = s - SvPVX(PL_linestr);
11060 tmpstr = newSV(79);
11061 sv_upgrade(tmpstr, SVt_PVIV);
11062 if (term == '\'') {
11063 op_type = OP_CONST;
11064 SvIV_set(tmpstr, -1);
11066 else if (term == '`') {
11067 op_type = OP_BACKTICK;
11068 SvIV_set(tmpstr, '\\');
11072 PL_multi_start = CopLINE(PL_curcop);
11073 PL_multi_open = PL_multi_close = '<';
11074 term = *PL_tokenbuf;
11075 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
11076 char * const bufptr = PL_sublex_info.super_bufptr;
11077 char * const bufend = PL_sublex_info.super_bufend;
11078 char * const olds = s - SvCUR(herewas);
11079 s = strchr(bufptr, '\n');
11083 while (s < bufend &&
11084 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
11086 CopLINE_inc(PL_curcop);
11089 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11090 missingterm(PL_tokenbuf);
11092 sv_setpvn(herewas,bufptr,d-bufptr+1);
11093 sv_setpvn(tmpstr,d+1,s-d);
11095 sv_catpvn(herewas,s,bufend-s);
11096 Copy(SvPVX_const(herewas),bufptr,SvCUR(herewas) + 1,char);
11103 while (s < PL_bufend &&
11104 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
11106 CopLINE_inc(PL_curcop);
11108 if (s >= PL_bufend) {
11109 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11110 missingterm(PL_tokenbuf);
11112 sv_setpvn(tmpstr,d+1,s-d);
11114 if (PL_madskills) {
11116 sv_catpvn(PL_thisstuff, d + 1, s - d);
11118 PL_thisstuff = newSVpvn(d + 1, s - d);
11119 stuffstart = s - SvPVX(PL_linestr);
11123 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
11125 sv_catpvn(herewas,s,PL_bufend-s);
11126 sv_setsv(PL_linestr,herewas);
11127 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
11128 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11129 PL_last_lop = PL_last_uni = NULL;
11132 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
11133 while (s >= PL_bufend) { /* multiple line string? */
11135 if (PL_madskills) {
11136 tstart = SvPVX(PL_linestr) + stuffstart;
11138 sv_catpvn(PL_thisstuff, tstart, PL_bufend - tstart);
11140 PL_thisstuff = newSVpvn(tstart, PL_bufend - tstart);
11144 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
11145 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11146 missingterm(PL_tokenbuf);
11149 stuffstart = s - SvPVX(PL_linestr);
11151 CopLINE_inc(PL_curcop);
11152 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11153 PL_last_lop = PL_last_uni = NULL;
11154 #ifndef PERL_STRICT_CR
11155 if (PL_bufend - PL_linestart >= 2) {
11156 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
11157 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
11159 PL_bufend[-2] = '\n';
11161 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
11163 else if (PL_bufend[-1] == '\r')
11164 PL_bufend[-1] = '\n';
11166 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
11167 PL_bufend[-1] = '\n';
11169 if (PERLDB_LINE && PL_curstash != PL_debstash) {
11170 SV * const sv = newSV(0);
11172 sv_upgrade(sv, SVt_PVMG);
11173 sv_setsv(sv,PL_linestr);
11174 (void)SvIOK_on(sv);
11176 av_store(CopFILEAVx(PL_curcop), (I32)CopLINE(PL_curcop),sv);
11178 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
11179 STRLEN off = PL_bufend - 1 - SvPVX_const(PL_linestr);
11180 *(SvPVX(PL_linestr) + off ) = ' ';
11181 sv_catsv(PL_linestr,herewas);
11182 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11183 s = SvPVX(PL_linestr) + off; /* In case PV of PL_linestr moved. */
11187 sv_catsv(tmpstr,PL_linestr);
11192 PL_multi_end = CopLINE(PL_curcop);
11193 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
11194 SvPV_shrink_to_cur(tmpstr);
11196 SvREFCNT_dec(herewas);
11198 if (UTF && is_utf8_string((U8*)SvPVX_const(tmpstr), SvCUR(tmpstr)))
11200 else if (PL_encoding)
11201 sv_recode_to_utf8(tmpstr, PL_encoding);
11203 PL_lex_stuff = tmpstr;
11204 yylval.ival = op_type;
11208 /* scan_inputsymbol
11209 takes: current position in input buffer
11210 returns: new position in input buffer
11211 side-effects: yylval and lex_op are set.
11216 <FH> read from filehandle
11217 <pkg::FH> read from package qualified filehandle
11218 <pkg'FH> read from package qualified filehandle
11219 <$fh> read from filehandle in $fh
11220 <*.h> filename glob
11225 S_scan_inputsymbol(pTHX_ char *start)
11228 register char *s = start; /* current position in buffer */
11232 char *d = PL_tokenbuf; /* start of temp holding space */
11233 const char * const e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
11235 end = strchr(s, '\n');
11238 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
11240 /* die if we didn't have space for the contents of the <>,
11241 or if it didn't end, or if we see a newline
11244 if (len >= (I32)sizeof PL_tokenbuf)
11245 Perl_croak(aTHX_ "Excessively long <> operator");
11247 Perl_croak(aTHX_ "Unterminated <> operator");
11252 Remember, only scalar variables are interpreted as filehandles by
11253 this code. Anything more complex (e.g., <$fh{$num}>) will be
11254 treated as a glob() call.
11255 This code makes use of the fact that except for the $ at the front,
11256 a scalar variable and a filehandle look the same.
11258 if (*d == '$' && d[1]) d++;
11260 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
11261 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
11264 /* If we've tried to read what we allow filehandles to look like, and
11265 there's still text left, then it must be a glob() and not a getline.
11266 Use scan_str to pull out the stuff between the <> and treat it
11267 as nothing more than a string.
11270 if (d - PL_tokenbuf != len) {
11271 yylval.ival = OP_GLOB;
11273 s = scan_str(start,!!PL_madskills,FALSE);
11275 Perl_croak(aTHX_ "Glob not terminated");
11279 bool readline_overriden = FALSE;
11282 /* we're in a filehandle read situation */
11285 /* turn <> into <ARGV> */
11287 Copy("ARGV",d,5,char);
11289 /* Check whether readline() is overriden */
11290 gv_readline = gv_fetchpvs("readline", GV_NOTQUAL, SVt_PVCV);
11292 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline))
11294 ((gvp = (GV**)hv_fetchs(PL_globalstash, "readline", FALSE))
11295 && (gv_readline = *gvp) != (GV*)&PL_sv_undef
11296 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline)))
11297 readline_overriden = TRUE;
11299 /* if <$fh>, create the ops to turn the variable into a
11303 /* try to find it in the pad for this block, otherwise find
11304 add symbol table ops
11306 const PADOFFSET tmp = pad_findmy(d);
11307 if (tmp != NOT_IN_PAD) {
11308 if (PAD_COMPNAME_FLAGS_isOUR(tmp)) {
11309 HV * const stash = PAD_COMPNAME_OURSTASH(tmp);
11310 HEK * const stashname = HvNAME_HEK(stash);
11311 SV * const sym = sv_2mortal(newSVhek(stashname));
11312 sv_catpvs(sym, "::");
11313 sv_catpv(sym, d+1);
11318 OP * const o = newOP(OP_PADSV, 0);
11320 PL_lex_op = readline_overriden
11321 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
11322 append_elem(OP_LIST, o,
11323 newCVREF(0, newGVOP(OP_GV,0,gv_readline))))
11324 : (OP*)newUNOP(OP_READLINE, 0, o);
11333 ? (GV_ADDMULTI | GV_ADDINEVAL)
11336 PL_lex_op = readline_overriden
11337 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
11338 append_elem(OP_LIST,
11339 newUNOP(OP_RV2SV, 0, newGVOP(OP_GV, 0, gv)),
11340 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
11341 : (OP*)newUNOP(OP_READLINE, 0,
11342 newUNOP(OP_RV2SV, 0,
11343 newGVOP(OP_GV, 0, gv)));
11345 if (!readline_overriden)
11346 PL_lex_op->op_flags |= OPf_SPECIAL;
11347 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
11348 yylval.ival = OP_NULL;
11351 /* If it's none of the above, it must be a literal filehandle
11352 (<Foo::BAR> or <FOO>) so build a simple readline OP */
11354 GV * const gv = gv_fetchpv(d, GV_ADD, SVt_PVIO);
11355 PL_lex_op = readline_overriden
11356 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
11357 append_elem(OP_LIST,
11358 newGVOP(OP_GV, 0, gv),
11359 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
11360 : (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
11361 yylval.ival = OP_NULL;
11370 takes: start position in buffer
11371 keep_quoted preserve \ on the embedded delimiter(s)
11372 keep_delims preserve the delimiters around the string
11373 returns: position to continue reading from buffer
11374 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
11375 updates the read buffer.
11377 This subroutine pulls a string out of the input. It is called for:
11378 q single quotes q(literal text)
11379 ' single quotes 'literal text'
11380 qq double quotes qq(interpolate $here please)
11381 " double quotes "interpolate $here please"
11382 qx backticks qx(/bin/ls -l)
11383 ` backticks `/bin/ls -l`
11384 qw quote words @EXPORT_OK = qw( func() $spam )
11385 m// regexp match m/this/
11386 s/// regexp substitute s/this/that/
11387 tr/// string transliterate tr/this/that/
11388 y/// string transliterate y/this/that/
11389 ($*@) sub prototypes sub foo ($)
11390 (stuff) sub attr parameters sub foo : attr(stuff)
11391 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
11393 In most of these cases (all but <>, patterns and transliterate)
11394 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
11395 calls scan_str(). s/// makes yylex() call scan_subst() which calls
11396 scan_str(). tr/// and y/// make yylex() call scan_trans() which
11399 It skips whitespace before the string starts, and treats the first
11400 character as the delimiter. If the delimiter is one of ([{< then
11401 the corresponding "close" character )]}> is used as the closing
11402 delimiter. It allows quoting of delimiters, and if the string has
11403 balanced delimiters ([{<>}]) it allows nesting.
11405 On success, the SV with the resulting string is put into lex_stuff or,
11406 if that is already non-NULL, into lex_repl. The second case occurs only
11407 when parsing the RHS of the special constructs s/// and tr/// (y///).
11408 For convenience, the terminating delimiter character is stuffed into
11413 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
11416 SV *sv; /* scalar value: string */
11417 const char *tmps; /* temp string, used for delimiter matching */
11418 register char *s = start; /* current position in the buffer */
11419 register char term; /* terminating character */
11420 register char *to; /* current position in the sv's data */
11421 I32 brackets = 1; /* bracket nesting level */
11422 bool has_utf8 = FALSE; /* is there any utf8 content? */
11423 I32 termcode; /* terminating char. code */
11424 U8 termstr[UTF8_MAXBYTES]; /* terminating string */
11425 STRLEN termlen; /* length of terminating string */
11426 char *last = NULL; /* last position for nesting bracket */
11432 /* skip space before the delimiter */
11438 if (PL_realtokenstart >= 0) {
11439 stuffstart = PL_realtokenstart;
11440 PL_realtokenstart = -1;
11443 stuffstart = start - SvPVX(PL_linestr);
11445 /* mark where we are, in case we need to report errors */
11448 /* after skipping whitespace, the next character is the terminator */
11451 termcode = termstr[0] = term;
11455 termcode = utf8_to_uvchr((U8*)s, &termlen);
11456 Copy(s, termstr, termlen, U8);
11457 if (!UTF8_IS_INVARIANT(term))
11461 /* mark where we are */
11462 PL_multi_start = CopLINE(PL_curcop);
11463 PL_multi_open = term;
11465 /* find corresponding closing delimiter */
11466 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
11467 termcode = termstr[0] = term = tmps[5];
11469 PL_multi_close = term;
11471 /* create a new SV to hold the contents. 79 is the SV's initial length.
11472 What a random number. */
11474 sv_upgrade(sv, SVt_PVIV);
11475 SvIV_set(sv, termcode);
11476 (void)SvPOK_only(sv); /* validate pointer */
11478 /* move past delimiter and try to read a complete string */
11480 sv_catpvn(sv, s, termlen);
11483 tstart = SvPVX(PL_linestr) + stuffstart;
11484 if (!PL_thisopen && !keep_delims) {
11485 PL_thisopen = newSVpvn(tstart, s - tstart);
11486 stuffstart = s - SvPVX(PL_linestr);
11490 if (PL_encoding && !UTF) {
11494 int offset = s - SvPVX_const(PL_linestr);
11495 const bool found = sv_cat_decode(sv, PL_encoding, PL_linestr,
11496 &offset, (char*)termstr, termlen);
11497 const char * const ns = SvPVX_const(PL_linestr) + offset;
11498 char * const svlast = SvEND(sv) - 1;
11500 for (; s < ns; s++) {
11501 if (*s == '\n' && !PL_rsfp)
11502 CopLINE_inc(PL_curcop);
11505 goto read_more_line;
11507 /* handle quoted delimiters */
11508 if (SvCUR(sv) > 1 && *(svlast-1) == '\\') {
11510 for (t = svlast-2; t >= SvPVX_const(sv) && *t == '\\';)
11512 if ((svlast-1 - t) % 2) {
11513 if (!keep_quoted) {
11514 *(svlast-1) = term;
11516 SvCUR_set(sv, SvCUR(sv) - 1);
11521 if (PL_multi_open == PL_multi_close) {
11529 for (t = w = last; t < svlast; w++, t++) {
11530 /* At here, all closes are "was quoted" one,
11531 so we don't check PL_multi_close. */
11533 if (!keep_quoted && *(t+1) == PL_multi_open)
11538 else if (*t == PL_multi_open)
11546 SvCUR_set(sv, w - SvPVX_const(sv));
11549 if (--brackets <= 0)
11554 if (!keep_delims) {
11555 SvCUR_set(sv, SvCUR(sv) - 1);
11561 /* extend sv if need be */
11562 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
11563 /* set 'to' to the next character in the sv's string */
11564 to = SvPVX(sv)+SvCUR(sv);
11566 /* if open delimiter is the close delimiter read unbridle */
11567 if (PL_multi_open == PL_multi_close) {
11568 for (; s < PL_bufend; s++,to++) {
11569 /* embedded newlines increment the current line number */
11570 if (*s == '\n' && !PL_rsfp)
11571 CopLINE_inc(PL_curcop);
11572 /* handle quoted delimiters */
11573 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
11574 if (!keep_quoted && s[1] == term)
11576 /* any other quotes are simply copied straight through */
11580 /* terminate when run out of buffer (the for() condition), or
11581 have found the terminator */
11582 else if (*s == term) {
11585 if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen))
11588 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
11594 /* if the terminator isn't the same as the start character (e.g.,
11595 matched brackets), we have to allow more in the quoting, and
11596 be prepared for nested brackets.
11599 /* read until we run out of string, or we find the terminator */
11600 for (; s < PL_bufend; s++,to++) {
11601 /* embedded newlines increment the line count */
11602 if (*s == '\n' && !PL_rsfp)
11603 CopLINE_inc(PL_curcop);
11604 /* backslashes can escape the open or closing characters */
11605 if (*s == '\\' && s+1 < PL_bufend) {
11606 if (!keep_quoted &&
11607 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
11612 /* allow nested opens and closes */
11613 else if (*s == PL_multi_close && --brackets <= 0)
11615 else if (*s == PL_multi_open)
11617 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
11622 /* terminate the copied string and update the sv's end-of-string */
11624 SvCUR_set(sv, to - SvPVX_const(sv));
11627 * this next chunk reads more into the buffer if we're not done yet
11631 break; /* handle case where we are done yet :-) */
11633 #ifndef PERL_STRICT_CR
11634 if (to - SvPVX_const(sv) >= 2) {
11635 if ((to[-2] == '\r' && to[-1] == '\n') ||
11636 (to[-2] == '\n' && to[-1] == '\r'))
11640 SvCUR_set(sv, to - SvPVX_const(sv));
11642 else if (to[-1] == '\r')
11645 else if (to - SvPVX_const(sv) == 1 && to[-1] == '\r')
11650 /* if we're out of file, or a read fails, bail and reset the current
11651 line marker so we can report where the unterminated string began
11654 if (PL_madskills) {
11655 char * const tstart = SvPVX(PL_linestr) + stuffstart;
11657 sv_catpvn(PL_thisstuff, tstart, PL_bufend - tstart);
11659 PL_thisstuff = newSVpvn(tstart, PL_bufend - tstart);
11663 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
11665 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11671 /* we read a line, so increment our line counter */
11672 CopLINE_inc(PL_curcop);
11674 /* update debugger info */
11675 if (PERLDB_LINE && PL_curstash != PL_debstash) {
11676 SV * const line_sv = newSV(0);
11678 sv_upgrade(line_sv, SVt_PVMG);
11679 sv_setsv(line_sv,PL_linestr);
11680 (void)SvIOK_on(line_sv);
11681 SvIV_set(line_sv, 0);
11682 av_store(CopFILEAVx(PL_curcop), (I32)CopLINE(PL_curcop), line_sv);
11685 /* having changed the buffer, we must update PL_bufend */
11686 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11687 PL_last_lop = PL_last_uni = NULL;
11690 /* at this point, we have successfully read the delimited string */
11692 if (!PL_encoding || UTF) {
11694 if (PL_madskills) {
11695 char * const tstart = SvPVX(PL_linestr) + stuffstart;
11696 const int len = s - start;
11698 sv_catpvn(PL_thisstuff, tstart, len);
11700 PL_thisstuff = newSVpvn(tstart, len);
11701 if (!PL_thisclose && !keep_delims)
11702 PL_thisclose = newSVpvn(s,termlen);
11707 sv_catpvn(sv, s, termlen);
11712 if (PL_madskills) {
11713 char * const tstart = SvPVX(PL_linestr) + stuffstart;
11714 const int len = s - tstart - termlen;
11716 sv_catpvn(PL_thisstuff, tstart, len);
11718 PL_thisstuff = newSVpvn(tstart, len);
11719 if (!PL_thisclose && !keep_delims)
11720 PL_thisclose = newSVpvn(s - termlen,termlen);
11724 if (has_utf8 || PL_encoding)
11727 PL_multi_end = CopLINE(PL_curcop);
11729 /* if we allocated too much space, give some back */
11730 if (SvCUR(sv) + 5 < SvLEN(sv)) {
11731 SvLEN_set(sv, SvCUR(sv) + 1);
11732 SvPV_renew(sv, SvLEN(sv));
11735 /* decide whether this is the first or second quoted string we've read
11748 takes: pointer to position in buffer
11749 returns: pointer to new position in buffer
11750 side-effects: builds ops for the constant in yylval.op
11752 Read a number in any of the formats that Perl accepts:
11754 \d(_?\d)*(\.(\d(_?\d)*)?)?[Ee][\+\-]?(\d(_?\d)*) 12 12.34 12.
11755 \.\d(_?\d)*[Ee][\+\-]?(\d(_?\d)*) .34
11758 0x[0-9A-Fa-f](_?[0-9A-Fa-f])*
11760 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
11763 If it reads a number without a decimal point or an exponent, it will
11764 try converting the number to an integer and see if it can do so
11765 without loss of precision.
11769 Perl_scan_num(pTHX_ const char *start, YYSTYPE* lvalp)
11772 register const char *s = start; /* current position in buffer */
11773 register char *d; /* destination in temp buffer */
11774 register char *e; /* end of temp buffer */
11775 NV nv; /* number read, as a double */
11776 SV *sv = NULL; /* place to put the converted number */
11777 bool floatit; /* boolean: int or float? */
11778 const char *lastub = NULL; /* position of last underbar */
11779 static char const number_too_long[] = "Number too long";
11781 /* We use the first character to decide what type of number this is */
11785 Perl_croak(aTHX_ "panic: scan_num");
11787 /* if it starts with a 0, it could be an octal number, a decimal in
11788 0.13 disguise, or a hexadecimal number, or a binary number. */
11792 u holds the "number so far"
11793 shift the power of 2 of the base
11794 (hex == 4, octal == 3, binary == 1)
11795 overflowed was the number more than we can hold?
11797 Shift is used when we add a digit. It also serves as an "are
11798 we in octal/hex/binary?" indicator to disallow hex characters
11799 when in octal mode.
11804 bool overflowed = FALSE;
11805 bool just_zero = TRUE; /* just plain 0 or binary number? */
11806 static const NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
11807 static const char* const bases[5] =
11808 { "", "binary", "", "octal", "hexadecimal" };
11809 static const char* const Bases[5] =
11810 { "", "Binary", "", "Octal", "Hexadecimal" };
11811 static const char* const maxima[5] =
11813 "0b11111111111111111111111111111111",
11817 const char *base, *Base, *max;
11819 /* check for hex */
11824 } else if (s[1] == 'b') {
11829 /* check for a decimal in disguise */
11830 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
11832 /* so it must be octal */
11839 if (ckWARN(WARN_SYNTAX))
11840 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
11841 "Misplaced _ in number");
11845 base = bases[shift];
11846 Base = Bases[shift];
11847 max = maxima[shift];
11849 /* read the rest of the number */
11851 /* x is used in the overflow test,
11852 b is the digit we're adding on. */
11857 /* if we don't mention it, we're done */
11861 /* _ are ignored -- but warned about if consecutive */
11863 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
11864 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
11865 "Misplaced _ in number");
11869 /* 8 and 9 are not octal */
11870 case '8': case '9':
11872 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
11876 case '2': case '3': case '4':
11877 case '5': case '6': case '7':
11879 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
11882 case '0': case '1':
11883 b = *s++ & 15; /* ASCII digit -> value of digit */
11887 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
11888 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
11889 /* make sure they said 0x */
11892 b = (*s++ & 7) + 9;
11894 /* Prepare to put the digit we have onto the end
11895 of the number so far. We check for overflows.
11901 x = u << shift; /* make room for the digit */
11903 if ((x >> shift) != u
11904 && !(PL_hints & HINT_NEW_BINARY)) {
11907 if (ckWARN_d(WARN_OVERFLOW))
11908 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
11909 "Integer overflow in %s number",
11912 u = x | b; /* add the digit to the end */
11915 n *= nvshift[shift];
11916 /* If an NV has not enough bits in its
11917 * mantissa to represent an UV this summing of
11918 * small low-order numbers is a waste of time
11919 * (because the NV cannot preserve the
11920 * low-order bits anyway): we could just
11921 * remember when did we overflow and in the
11922 * end just multiply n by the right
11930 /* if we get here, we had success: make a scalar value from
11935 /* final misplaced underbar check */
11936 if (s[-1] == '_') {
11937 if (ckWARN(WARN_SYNTAX))
11938 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
11943 if (n > 4294967295.0 && ckWARN(WARN_PORTABLE))
11944 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
11945 "%s number > %s non-portable",
11951 if (u > 0xffffffff && ckWARN(WARN_PORTABLE))
11952 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
11953 "%s number > %s non-portable",
11958 if (just_zero && (PL_hints & HINT_NEW_INTEGER))
11959 sv = new_constant(start, s - start, "integer",
11961 else if (PL_hints & HINT_NEW_BINARY)
11962 sv = new_constant(start, s - start, "binary", sv, NULL, NULL);
11967 handle decimal numbers.
11968 we're also sent here when we read a 0 as the first digit
11970 case '1': case '2': case '3': case '4': case '5':
11971 case '6': case '7': case '8': case '9': case '.':
11974 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
11977 /* read next group of digits and _ and copy into d */
11978 while (isDIGIT(*s) || *s == '_') {
11979 /* skip underscores, checking for misplaced ones
11983 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
11984 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
11985 "Misplaced _ in number");
11989 /* check for end of fixed-length buffer */
11991 Perl_croak(aTHX_ number_too_long);
11992 /* if we're ok, copy the character */
11997 /* final misplaced underbar check */
11998 if (lastub && s == lastub + 1) {
11999 if (ckWARN(WARN_SYNTAX))
12000 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
12003 /* read a decimal portion if there is one. avoid
12004 3..5 being interpreted as the number 3. followed
12007 if (*s == '.' && s[1] != '.') {
12012 if (ckWARN(WARN_SYNTAX))
12013 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12014 "Misplaced _ in number");
12018 /* copy, ignoring underbars, until we run out of digits.
12020 for (; isDIGIT(*s) || *s == '_'; s++) {
12021 /* fixed length buffer check */
12023 Perl_croak(aTHX_ number_too_long);
12025 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
12026 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12027 "Misplaced _ in number");
12033 /* fractional part ending in underbar? */
12034 if (s[-1] == '_') {
12035 if (ckWARN(WARN_SYNTAX))
12036 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12037 "Misplaced _ in number");
12039 if (*s == '.' && isDIGIT(s[1])) {
12040 /* oops, it's really a v-string, but without the "v" */
12046 /* read exponent part, if present */
12047 if ((*s == 'e' || *s == 'E') && strchr("+-0123456789_", s[1])) {
12051 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
12052 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
12054 /* stray preinitial _ */
12056 if (ckWARN(WARN_SYNTAX))
12057 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12058 "Misplaced _ in number");
12062 /* allow positive or negative exponent */
12063 if (*s == '+' || *s == '-')
12066 /* stray initial _ */
12068 if (ckWARN(WARN_SYNTAX))
12069 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12070 "Misplaced _ in number");
12074 /* read digits of exponent */
12075 while (isDIGIT(*s) || *s == '_') {
12078 Perl_croak(aTHX_ number_too_long);
12082 if (((lastub && s == lastub + 1) ||
12083 (!isDIGIT(s[1]) && s[1] != '_'))
12084 && ckWARN(WARN_SYNTAX))
12085 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12086 "Misplaced _ in number");
12093 /* make an sv from the string */
12097 We try to do an integer conversion first if no characters
12098 indicating "float" have been found.
12103 const int flags = grok_number (PL_tokenbuf, d - PL_tokenbuf, &uv);
12105 if (flags == IS_NUMBER_IN_UV) {
12107 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
12110 } else if (flags == (IS_NUMBER_IN_UV | IS_NUMBER_NEG)) {
12111 if (uv <= (UV) IV_MIN)
12112 sv_setiv(sv, -(IV)uv);
12119 /* terminate the string */
12121 nv = Atof(PL_tokenbuf);
12125 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
12126 (PL_hints & HINT_NEW_INTEGER) )
12127 sv = new_constant(PL_tokenbuf,
12130 (floatit ? "float" : "integer"),
12134 /* if it starts with a v, it could be a v-string */
12137 sv = newSV(5); /* preallocate storage space */
12138 s = scan_vstring(s,sv);
12142 /* make the op for the constant and return */
12145 lvalp->opval = newSVOP(OP_CONST, 0, sv);
12147 lvalp->opval = NULL;
12153 S_scan_formline(pTHX_ register char *s)
12156 register char *eol;
12158 SV * const stuff = newSVpvs("");
12159 bool needargs = FALSE;
12160 bool eofmt = FALSE;
12162 char *tokenstart = s;
12165 if (PL_madskills) {
12166 savewhite = PL_thiswhite;
12171 while (!needargs) {
12174 #ifdef PERL_STRICT_CR
12175 while (SPACE_OR_TAB(*t))
12178 while (SPACE_OR_TAB(*t) || *t == '\r')
12181 if (*t == '\n' || t == PL_bufend) {
12186 if (PL_in_eval && !PL_rsfp) {
12187 eol = (char *) memchr(s,'\n',PL_bufend-s);
12192 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
12194 for (t = s; t < eol; t++) {
12195 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
12197 goto enough; /* ~~ must be first line in formline */
12199 if (*t == '@' || *t == '^')
12203 sv_catpvn(stuff, s, eol-s);
12204 #ifndef PERL_STRICT_CR
12205 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
12206 char *end = SvPVX(stuff) + SvCUR(stuff);
12209 SvCUR_set(stuff, SvCUR(stuff) - 1);
12219 if (PL_madskills) {
12221 sv_catpvn(PL_thistoken, tokenstart, PL_bufend - tokenstart);
12223 PL_thistoken = newSVpvn(tokenstart, PL_bufend - tokenstart);
12226 s = filter_gets(PL_linestr, PL_rsfp, 0);
12228 tokenstart = PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
12230 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
12232 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
12233 PL_last_lop = PL_last_uni = NULL;
12242 if (SvCUR(stuff)) {
12245 PL_lex_state = LEX_NORMAL;
12246 start_force(PL_curforce);
12247 NEXTVAL_NEXTTOKE.ival = 0;
12251 PL_lex_state = LEX_FORMLINE;
12253 if (UTF && is_utf8_string((U8*)SvPVX_const(stuff), SvCUR(stuff)))
12255 else if (PL_encoding)
12256 sv_recode_to_utf8(stuff, PL_encoding);
12258 start_force(PL_curforce);
12259 NEXTVAL_NEXTTOKE.opval = (OP*)newSVOP(OP_CONST, 0, stuff);
12261 start_force(PL_curforce);
12262 NEXTVAL_NEXTTOKE.ival = OP_FORMLINE;
12266 SvREFCNT_dec(stuff);
12268 PL_lex_formbrack = 0;
12272 if (PL_madskills) {
12274 sv_catpvn(PL_thistoken, tokenstart, s - tokenstart);
12276 PL_thistoken = newSVpvn(tokenstart, s - tokenstart);
12277 PL_thiswhite = savewhite;
12289 PL_cshlen = strlen(PL_cshname);
12291 #if defined(USE_ITHREADS)
12292 PERL_UNUSED_CONTEXT;
12298 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
12301 const I32 oldsavestack_ix = PL_savestack_ix;
12302 CV* const outsidecv = PL_compcv;
12305 assert(SvTYPE(PL_compcv) == SVt_PVCV);
12307 SAVEI32(PL_subline);
12308 save_item(PL_subname);
12309 SAVESPTR(PL_compcv);
12311 PL_compcv = (CV*)newSV(0);
12312 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
12313 CvFLAGS(PL_compcv) |= flags;
12315 PL_subline = CopLINE(PL_curcop);
12316 CvPADLIST(PL_compcv) = pad_new(padnew_SAVE|padnew_SAVESUB);
12317 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc_simple(outsidecv);
12318 CvOUTSIDE_SEQ(PL_compcv) = PL_cop_seqmax;
12320 return oldsavestack_ix;
12324 #pragma segment Perl_yylex
12327 Perl_yywarn(pTHX_ const char *s)
12330 PL_in_eval |= EVAL_WARNONLY;
12332 PL_in_eval &= ~EVAL_WARNONLY;
12337 Perl_yyerror(pTHX_ const char *s)
12340 const char *where = NULL;
12341 const char *context = NULL;
12345 if (!yychar || (yychar == ';' && !PL_rsfp))
12347 else if (PL_oldoldbufptr && PL_bufptr > PL_oldoldbufptr &&
12348 PL_bufptr - PL_oldoldbufptr < 200 && PL_oldoldbufptr != PL_oldbufptr &&
12349 PL_oldbufptr != PL_bufptr) {
12352 The code below is removed for NetWare because it abends/crashes on NetWare
12353 when the script has error such as not having the closing quotes like:
12354 if ($var eq "value)
12355 Checking of white spaces is anyway done in NetWare code.
12358 while (isSPACE(*PL_oldoldbufptr))
12361 context = PL_oldoldbufptr;
12362 contlen = PL_bufptr - PL_oldoldbufptr;
12364 else if (PL_oldbufptr && PL_bufptr > PL_oldbufptr &&
12365 PL_bufptr - PL_oldbufptr < 200 && PL_oldbufptr != PL_bufptr) {
12368 The code below is removed for NetWare because it abends/crashes on NetWare
12369 when the script has error such as not having the closing quotes like:
12370 if ($var eq "value)
12371 Checking of white spaces is anyway done in NetWare code.
12374 while (isSPACE(*PL_oldbufptr))
12377 context = PL_oldbufptr;
12378 contlen = PL_bufptr - PL_oldbufptr;
12380 else if (yychar > 255)
12381 where = "next token ???";
12382 else if (yychar == -2) { /* YYEMPTY */
12383 if (PL_lex_state == LEX_NORMAL ||
12384 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
12385 where = "at end of line";
12386 else if (PL_lex_inpat)
12387 where = "within pattern";
12389 where = "within string";
12392 SV * const where_sv = sv_2mortal(newSVpvs("next char "));
12394 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
12395 else if (isPRINT_LC(yychar))
12396 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
12398 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
12399 where = SvPVX_const(where_sv);
12401 msg = sv_2mortal(newSVpv(s, 0));
12402 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
12403 OutCopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
12405 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
12407 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
12408 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
12409 Perl_sv_catpvf(aTHX_ msg,
12410 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
12411 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
12414 if (PL_in_eval & EVAL_WARNONLY && ckWARN_d(WARN_SYNTAX))
12415 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "%"SVf, (void*)msg);
12418 if (PL_error_count >= 10) {
12419 if (PL_in_eval && SvCUR(ERRSV))
12420 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
12421 (void*)ERRSV, OutCopFILE(PL_curcop));
12423 Perl_croak(aTHX_ "%s has too many errors.\n",
12424 OutCopFILE(PL_curcop));
12427 PL_in_my_stash = NULL;
12431 #pragma segment Main
12435 S_swallow_bom(pTHX_ U8 *s)
12438 const STRLEN slen = SvCUR(PL_linestr);
12441 if (s[1] == 0xFE) {
12442 /* UTF-16 little-endian? (or UTF32-LE?) */
12443 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
12444 Perl_croak(aTHX_ "Unsupported script encoding UTF32-LE");
12445 #ifndef PERL_NO_UTF16_FILTER
12446 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF16-LE script encoding (BOM)\n");
12449 if (PL_bufend > (char*)s) {
12453 filter_add(utf16rev_textfilter, NULL);
12454 Newx(news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
12455 utf16_to_utf8_reversed(s, news,
12456 PL_bufend - (char*)s - 1,
12458 sv_setpvn(PL_linestr, (const char*)news, newlen);
12460 s = (U8*)SvPVX(PL_linestr);
12461 Copy(news, s, newlen, U8);
12465 SvUTF8_on(PL_linestr);
12466 s = (U8*)SvPVX(PL_linestr);
12468 /* FIXME - is this a general bug fix? */
12471 PL_bufend = SvPVX(PL_linestr) + newlen;
12474 Perl_croak(aTHX_ "Unsupported script encoding UTF16-LE");
12479 if (s[1] == 0xFF) { /* UTF-16 big-endian? */
12480 #ifndef PERL_NO_UTF16_FILTER
12481 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (BOM)\n");
12484 if (PL_bufend > (char *)s) {
12488 filter_add(utf16_textfilter, NULL);
12489 Newx(news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
12490 utf16_to_utf8(s, news,
12491 PL_bufend - (char*)s,
12493 sv_setpvn(PL_linestr, (const char*)news, newlen);
12495 SvUTF8_on(PL_linestr);
12496 s = (U8*)SvPVX(PL_linestr);
12497 PL_bufend = SvPVX(PL_linestr) + newlen;
12500 Perl_croak(aTHX_ "Unsupported script encoding UTF16-BE");
12505 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
12506 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-8 script encoding (BOM)\n");
12507 s += 3; /* UTF-8 */
12513 if (s[2] == 0xFE && s[3] == 0xFF) {
12514 /* UTF-32 big-endian */
12515 Perl_croak(aTHX_ "Unsupported script encoding UTF32-BE");
12518 else if (s[2] == 0 && s[3] != 0) {
12521 * are a good indicator of UTF-16BE. */
12522 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (no BOM)\n");
12528 if (slen > 3 && s[1] == 0x73 && s[2] == 0x66 && s[3] == 0x73) {
12529 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-8 script encoding (BOM)\n");
12530 s += 4; /* UTF-8 */
12536 if (slen > 3 && s[1] == 0 && s[2] != 0 && s[3] == 0) {
12539 * are a good indicator of UTF-16LE. */
12540 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16LE script encoding (no BOM)\n");
12549 * Restore a source filter.
12553 restore_rsfp(pTHX_ void *f)
12556 PerlIO * const fp = (PerlIO*)f;
12558 if (PL_rsfp == PerlIO_stdin())
12559 PerlIO_clearerr(PL_rsfp);
12560 else if (PL_rsfp && (PL_rsfp != fp))
12561 PerlIO_close(PL_rsfp);
12565 #ifndef PERL_NO_UTF16_FILTER
12567 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
12570 const STRLEN old = SvCUR(sv);
12571 const I32 count = FILTER_READ(idx+1, sv, maxlen);
12572 DEBUG_P(PerlIO_printf(Perl_debug_log,
12573 "utf16_textfilter(%p): %d %d (%d)\n",
12574 FPTR2DPTR(void *, utf16_textfilter),
12575 idx, maxlen, (int) count));
12579 Newx(tmps, SvCUR(sv) * 3 / 2 + 1, U8);
12580 Copy(SvPVX_const(sv), tmps, old, char);
12581 utf16_to_utf8((U8*)SvPVX_const(sv) + old, tmps + old,
12582 SvCUR(sv) - old, &newlen);
12583 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
12585 DEBUG_P({sv_dump(sv);});
12590 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
12593 const STRLEN old = SvCUR(sv);
12594 const I32 count = FILTER_READ(idx+1, sv, maxlen);
12595 DEBUG_P(PerlIO_printf(Perl_debug_log,
12596 "utf16rev_textfilter(%p): %d %d (%d)\n",
12597 FPTR2DPTR(void *, utf16rev_textfilter),
12598 idx, maxlen, (int) count));
12602 Newx(tmps, SvCUR(sv) * 3 / 2 + 1, U8);
12603 Copy(SvPVX_const(sv), tmps, old, char);
12604 utf16_to_utf8((U8*)SvPVX_const(sv) + old, tmps + old,
12605 SvCUR(sv) - old, &newlen);
12606 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
12608 DEBUG_P({ sv_dump(sv); });
12614 Returns a pointer to the next character after the parsed
12615 vstring, as well as updating the passed in sv.
12617 Function must be called like
12620 s = scan_vstring(s,sv);
12622 The sv should already be large enough to store the vstring
12623 passed in, for performance reasons.
12628 Perl_scan_vstring(pTHX_ const char *s, SV *sv)
12631 const char *pos = s;
12632 const char *start = s;
12633 if (*pos == 'v') pos++; /* get past 'v' */
12634 while (pos < PL_bufend && (isDIGIT(*pos) || *pos == '_'))
12636 if ( *pos != '.') {
12637 /* this may not be a v-string if followed by => */
12638 const char *next = pos;
12639 while (next < PL_bufend && isSPACE(*next))
12641 if ((PL_bufend - next) >= 2 && *next == '=' && next[1] == '>' ) {
12642 /* return string not v-string */
12643 sv_setpvn(sv,(char *)s,pos-s);
12644 return (char *)pos;
12648 if (!isALPHA(*pos)) {
12649 U8 tmpbuf[UTF8_MAXBYTES+1];
12652 s++; /* get past 'v' */
12654 sv_setpvn(sv, "", 0);
12657 /* this is atoi() that tolerates underscores */
12660 const char *end = pos;
12662 while (--end >= s) {
12664 const UV orev = rev;
12665 rev += (*end - '0') * mult;
12667 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
12668 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
12669 "Integer overflow in decimal number");
12673 if (rev > 0x7FFFFFFF)
12674 Perl_croak(aTHX_ "In EBCDIC the v-string components cannot exceed 2147483647");
12676 /* Append native character for the rev point */
12677 tmpend = uvchr_to_utf8(tmpbuf, rev);
12678 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
12679 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(rev)))
12681 if (pos + 1 < PL_bufend && *pos == '.' && isDIGIT(pos[1]))
12687 while (pos < PL_bufend && (isDIGIT(*pos) || *pos == '_'))
12691 sv_magic(sv,NULL,PERL_MAGIC_vstring,(const char*)start, pos-start);
12699 * c-indentation-style: bsd
12700 * c-basic-offset: 4
12701 * indent-tabs-mode: t
12704 * ex: set ts=8 sts=4 sw=4 noet: