3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
4 * 2000, 2001, 2002, 2003, 2004, 2005, by Larry Wall and others
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
12 * "It all comes from here, the stench and the peril." --Frodo
16 * This file is the lexer for Perl. It's closely linked to the
19 * The main routine is yylex(), which returns the next token.
23 #define PERL_IN_TOKE_C
26 #define yychar (*PL_yycharp)
27 #define yylval (*PL_yylvalp)
29 static const char ident_too_long[] =
30 "Identifier too long";
31 static const char c_without_g[] =
32 "Use of /c modifier is meaningless without /g";
33 static const char c_in_subst[] =
34 "Use of /c modifier is meaningless in s///";
36 static void restore_rsfp(pTHX_ void *f);
37 #ifndef PERL_NO_UTF16_FILTER
38 static I32 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen);
39 static I32 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen);
42 #define XFAKEBRACK 128
45 #ifdef USE_UTF8_SCRIPTS
46 # define UTF (!IN_BYTES)
48 # define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || (PL_hints & HINT_UTF8))
51 /* In variables named $^X, these are the legal values for X.
52 * 1999-02-27 mjd-perl-patch@plover.com */
53 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
55 /* On MacOS, respect nonbreaking spaces */
56 #ifdef MACOS_TRADITIONAL
57 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
59 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
62 /* LEX_* are values for PL_lex_state, the state of the lexer.
63 * They are arranged oddly so that the guard on the switch statement
64 * can get by with a single comparison (if the compiler is smart enough).
67 /* #define LEX_NOTPARSING 11 is done in perl.h. */
70 #define LEX_INTERPNORMAL 9
71 #define LEX_INTERPCASEMOD 8
72 #define LEX_INTERPPUSH 7
73 #define LEX_INTERPSTART 6
74 #define LEX_INTERPEND 5
75 #define LEX_INTERPENDMAYBE 4
76 #define LEX_INTERPCONCAT 3
77 #define LEX_INTERPCONST 2
78 #define LEX_FORMLINE 1
79 #define LEX_KNOWNEXT 0
82 static const char* const lex_state_names[] = {
101 #include "keywords.h"
103 /* CLINE is a macro that ensures PL_copline has a sane value */
108 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
110 /* According to some strict interpretations of ANSI C89 one cannot
111 * cast void pointers to code pointers or vice versa (as filter_add(),
112 * filter_del(), and filter_read() will want to do). We should still
113 * be able to use a union for sneaky "casting". */
120 * Convenience functions to return different tokens and prime the
121 * lexer for the next token. They all take an argument.
123 * TOKEN : generic token (used for '(', DOLSHARP, etc)
124 * OPERATOR : generic operator
125 * AOPERATOR : assignment operator
126 * PREBLOCK : beginning the block after an if, while, foreach, ...
127 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
128 * PREREF : *EXPR where EXPR is not a simple identifier
129 * TERM : expression term
130 * LOOPX : loop exiting command (goto, last, dump, etc)
131 * FTST : file test operator
132 * FUN0 : zero-argument function
133 * FUN1 : not used, except for not, which isn't a UNIOP
134 * BOop : bitwise or or xor
136 * SHop : shift operator
137 * PWop : power operator
138 * PMop : pattern-matching operator
139 * Aop : addition-level operator
140 * Mop : multiplication-level operator
141 * Eop : equality-testing operator
142 * Rop : relational operator <= != gt
144 * Also see LOP and lop() below.
147 #ifdef DEBUGGING /* Serve -DT. */
148 # define REPORT(retval) tokereport(s,(int)retval)
150 # define REPORT(retval) (retval)
153 #define TOKEN(retval) return ( PL_bufptr = s, REPORT(retval))
154 #define OPERATOR(retval) return (PL_expect = XTERM, PL_bufptr = s, REPORT(retval))
155 #define AOPERATOR(retval) return ao((PL_expect = XTERM, PL_bufptr = s, REPORT(retval)))
156 #define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s, REPORT(retval))
157 #define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s, REPORT(retval))
158 #define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s, REPORT(retval))
159 #define TERM(retval) return (CLINE, PL_expect = XOPERATOR, PL_bufptr = s, REPORT(retval))
160 #define LOOPX(f) return (yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)LOOPEX))
161 #define FTST(f) return (yylval.ival=f, PL_expect=XTERMORDORDOR, PL_bufptr=s, REPORT((int)UNIOP))
162 #define FUN0(f) return (yylval.ival=f, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC0))
163 #define FUN1(f) return (yylval.ival=f, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC1))
164 #define BOop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)BITOROP)))
165 #define BAop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)BITANDOP)))
166 #define SHop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)SHIFTOP)))
167 #define PWop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)POWOP)))
168 #define PMop(f) return(yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)MATCHOP))
169 #define Aop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)ADDOP)))
170 #define Mop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)MULOP)))
171 #define Eop(f) return (yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)EQOP))
172 #define Rop(f) return (yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)RELOP))
174 /* This bit of chicanery makes a unary function followed by
175 * a parenthesis into a function with one argument, highest precedence.
176 * The UNIDOR macro is for unary functions that can be followed by the //
177 * operator (such as C<shift // 0>).
179 #define UNI2(f,x) return ( \
183 PL_last_uni = PL_oldbufptr, \
184 PL_last_lop_op = f, \
186 (*s == '(' || (s = skipspace(s), *s == '(') \
187 ? (int)FUNC1 : (int)UNIOP)))
188 #define UNI(f) UNI2(f,XTERM)
189 #define UNIDOR(f) UNI2(f,XTERMORDORDOR)
191 #define UNIBRACK(f) return ( \
194 PL_last_uni = PL_oldbufptr, \
196 (*s == '(' || (s = skipspace(s), *s == '(') \
197 ? (int)FUNC1 : (int)UNIOP)))
199 /* grandfather return to old style */
200 #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
204 /* how to interpret the yylval associated with the token */
208 TOKENTYPE_OPNUM, /* yylval.ival contains an opcode number */
214 static struct debug_tokens { const int token, type; const char *name; }
215 const debug_tokens[] =
217 { ADDOP, TOKENTYPE_OPNUM, "ADDOP" },
218 { ANDAND, TOKENTYPE_NONE, "ANDAND" },
219 { ANDOP, TOKENTYPE_NONE, "ANDOP" },
220 { ANONSUB, TOKENTYPE_IVAL, "ANONSUB" },
221 { ARROW, TOKENTYPE_NONE, "ARROW" },
222 { ASSIGNOP, TOKENTYPE_OPNUM, "ASSIGNOP" },
223 { BITANDOP, TOKENTYPE_OPNUM, "BITANDOP" },
224 { BITOROP, TOKENTYPE_OPNUM, "BITOROP" },
225 { COLONATTR, TOKENTYPE_NONE, "COLONATTR" },
226 { CONTINUE, TOKENTYPE_NONE, "CONTINUE" },
227 { DO, TOKENTYPE_NONE, "DO" },
228 { DOLSHARP, TOKENTYPE_NONE, "DOLSHARP" },
229 { DORDOR, TOKENTYPE_NONE, "DORDOR" },
230 { DOROP, TOKENTYPE_OPNUM, "DOROP" },
231 { DOTDOT, TOKENTYPE_IVAL, "DOTDOT" },
232 { ELSE, TOKENTYPE_NONE, "ELSE" },
233 { ELSIF, TOKENTYPE_IVAL, "ELSIF" },
234 { EQOP, TOKENTYPE_OPNUM, "EQOP" },
235 { FOR, TOKENTYPE_IVAL, "FOR" },
236 { FORMAT, TOKENTYPE_NONE, "FORMAT" },
237 { FUNC, TOKENTYPE_OPNUM, "FUNC" },
238 { FUNC0, TOKENTYPE_OPNUM, "FUNC0" },
239 { FUNC0SUB, TOKENTYPE_OPVAL, "FUNC0SUB" },
240 { FUNC1, TOKENTYPE_OPNUM, "FUNC1" },
241 { FUNCMETH, TOKENTYPE_OPVAL, "FUNCMETH" },
242 { HASHBRACK, TOKENTYPE_NONE, "HASHBRACK" },
243 { IF, TOKENTYPE_IVAL, "IF" },
244 { LABEL, TOKENTYPE_PVAL, "LABEL" },
245 { LOCAL, TOKENTYPE_IVAL, "LOCAL" },
246 { LOOPEX, TOKENTYPE_OPNUM, "LOOPEX" },
247 { LSTOP, TOKENTYPE_OPNUM, "LSTOP" },
248 { LSTOPSUB, TOKENTYPE_OPVAL, "LSTOPSUB" },
249 { MATCHOP, TOKENTYPE_OPNUM, "MATCHOP" },
250 { METHOD, TOKENTYPE_OPVAL, "METHOD" },
251 { MULOP, TOKENTYPE_OPNUM, "MULOP" },
252 { MY, TOKENTYPE_IVAL, "MY" },
253 { MYSUB, TOKENTYPE_NONE, "MYSUB" },
254 { NOAMP, TOKENTYPE_NONE, "NOAMP" },
255 { NOTOP, TOKENTYPE_NONE, "NOTOP" },
256 { OROP, TOKENTYPE_IVAL, "OROP" },
257 { OROR, TOKENTYPE_NONE, "OROR" },
258 { PACKAGE, TOKENTYPE_NONE, "PACKAGE" },
259 { PMFUNC, TOKENTYPE_OPVAL, "PMFUNC" },
260 { POSTDEC, TOKENTYPE_NONE, "POSTDEC" },
261 { POSTINC, TOKENTYPE_NONE, "POSTINC" },
262 { POWOP, TOKENTYPE_OPNUM, "POWOP" },
263 { PREDEC, TOKENTYPE_NONE, "PREDEC" },
264 { PREINC, TOKENTYPE_NONE, "PREINC" },
265 { PRIVATEREF, TOKENTYPE_OPVAL, "PRIVATEREF" },
266 { REFGEN, TOKENTYPE_NONE, "REFGEN" },
267 { RELOP, TOKENTYPE_OPNUM, "RELOP" },
268 { SHIFTOP, TOKENTYPE_OPNUM, "SHIFTOP" },
269 { SUB, TOKENTYPE_NONE, "SUB" },
270 { THING, TOKENTYPE_OPVAL, "THING" },
271 { UMINUS, TOKENTYPE_NONE, "UMINUS" },
272 { UNIOP, TOKENTYPE_OPNUM, "UNIOP" },
273 { UNIOPSUB, TOKENTYPE_OPVAL, "UNIOPSUB" },
274 { UNLESS, TOKENTYPE_IVAL, "UNLESS" },
275 { UNTIL, TOKENTYPE_IVAL, "UNTIL" },
276 { USE, TOKENTYPE_IVAL, "USE" },
277 { WHILE, TOKENTYPE_IVAL, "WHILE" },
278 { WORD, TOKENTYPE_OPVAL, "WORD" },
279 { 0, TOKENTYPE_NONE, 0 }
282 /* dump the returned token in rv, plus any optional arg in yylval */
285 S_tokereport(pTHX_ const char* s, I32 rv)
288 const char *name = Nullch;
289 enum token_type type = TOKENTYPE_NONE;
290 const struct debug_tokens *p;
291 SV* report = newSVpvn("<== ", 4);
293 for (p = debug_tokens; p->token; p++) {
294 if (p->token == (int)rv) {
301 Perl_sv_catpv(aTHX_ report, name);
302 else if ((char)rv > ' ' && (char)rv < '~')
303 Perl_sv_catpvf(aTHX_ report, "'%c'", (char)rv);
305 Perl_sv_catpv(aTHX_ report, "EOF");
307 Perl_sv_catpvf(aTHX_ report, "?? %"IVdf, (IV)rv);
310 case TOKENTYPE_GVVAL: /* doesn't appear to be used */
313 Perl_sv_catpvf(aTHX_ report, "(ival=%"IVdf")", (IV)yylval.ival);
315 case TOKENTYPE_OPNUM:
316 Perl_sv_catpvf(aTHX_ report, "(ival=op_%s)",
317 PL_op_name[yylval.ival]);
320 Perl_sv_catpvf(aTHX_ report, "(pval=\"%s\")", yylval.pval);
322 case TOKENTYPE_OPVAL:
324 Perl_sv_catpvf(aTHX_ report, "(opval=op_%s)",
325 PL_op_name[yylval.opval->op_type]);
327 Perl_sv_catpv(aTHX_ report, "(opval=null)");
330 Perl_sv_catpvf(aTHX_ report, " at line %"IVdf" [", (IV)CopLINE(PL_curcop));
331 if (s - PL_bufptr > 0)
332 sv_catpvn(report, PL_bufptr, s - PL_bufptr);
334 if (PL_oldbufptr && *PL_oldbufptr)
335 sv_catpv(report, PL_tokenbuf);
337 PerlIO_printf(Perl_debug_log, "### %s]\n", SvPV_nolen(report));
347 * This subroutine detects &&=, ||=, and //= and turns an ANDAND, OROR or DORDOR
348 * into an OP_ANDASSIGN, OP_ORASSIGN, or OP_DORASSIGN
352 S_ao(pTHX_ int toketype)
354 if (*PL_bufptr == '=') {
356 if (toketype == ANDAND)
357 yylval.ival = OP_ANDASSIGN;
358 else if (toketype == OROR)
359 yylval.ival = OP_ORASSIGN;
360 else if (toketype == DORDOR)
361 yylval.ival = OP_DORASSIGN;
369 * When Perl expects an operator and finds something else, no_op
370 * prints the warning. It always prints "<something> found where
371 * operator expected. It prints "Missing semicolon on previous line?"
372 * if the surprise occurs at the start of the line. "do you need to
373 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
374 * where the compiler doesn't know if foo is a method call or a function.
375 * It prints "Missing operator before end of line" if there's nothing
376 * after the missing operator, or "... before <...>" if there is something
377 * after the missing operator.
381 S_no_op(pTHX_ const char *what, char *s)
383 char *oldbp = PL_bufptr;
384 bool is_first = (PL_oldbufptr == PL_linestart);
390 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
391 if (ckWARN_d(WARN_SYNTAX)) {
393 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
394 "\t(Missing semicolon on previous line?)\n");
395 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
397 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy_if(t,UTF) || *t == ':'); t++) ;
398 if (t < PL_bufptr && isSPACE(*t))
399 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
400 "\t(Do you need to predeclare %.*s?)\n",
401 t - PL_oldoldbufptr, PL_oldoldbufptr);
405 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
406 "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp);
414 * Complain about missing quote/regexp/heredoc terminator.
415 * If it's called with (char *)NULL then it cauterizes the line buffer.
416 * If we're in a delimited string and the delimiter is a control
417 * character, it's reformatted into a two-char sequence like ^C.
422 S_missingterm(pTHX_ char *s)
427 char *nl = strrchr(s,'\n');
433 iscntrl(PL_multi_close)
435 PL_multi_close < 32 || PL_multi_close == 127
439 tmpbuf[1] = toCTRL(PL_multi_close);
444 *tmpbuf = (char)PL_multi_close;
448 q = strchr(s,'"') ? '\'' : '"';
449 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
457 Perl_deprecate(pTHX_ const char *s)
459 if (ckWARN(WARN_DEPRECATED))
460 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED), "Use of %s is deprecated", s);
464 Perl_deprecate_old(pTHX_ const char *s)
466 /* This function should NOT be called for any new deprecated warnings */
467 /* Use Perl_deprecate instead */
469 /* It is here to maintain backward compatibility with the pre-5.8 */
470 /* warnings category hierarchy. The "deprecated" category used to */
471 /* live under the "syntax" category. It is now a top-level category */
472 /* in its own right. */
474 if (ckWARN2(WARN_DEPRECATED, WARN_SYNTAX))
475 Perl_warner(aTHX_ packWARN2(WARN_DEPRECATED, WARN_SYNTAX),
476 "Use of %s is deprecated", s);
481 * Deprecate a comma-less variable list.
487 deprecate_old("comma-less variable list");
491 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
492 * utf16-to-utf8-reversed.
495 #ifdef PERL_CR_FILTER
499 register const char *s = SvPVX(sv);
500 register const char *e = s + SvCUR(sv);
501 /* outer loop optimized to do nothing if there are no CR-LFs */
503 if (*s++ == '\r' && *s == '\n') {
504 /* hit a CR-LF, need to copy the rest */
505 register char *d = s - 1;
508 if (*s == '\r' && s[1] == '\n')
519 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
521 const I32 count = FILTER_READ(idx+1, sv, maxlen);
522 if (count > 0 && !maxlen)
530 * Initialize variables. Uses the Perl save_stack to save its state (for
531 * recursive calls to the parser).
535 Perl_lex_start(pTHX_ SV *line)
540 SAVEI32(PL_lex_dojoin);
541 SAVEI32(PL_lex_brackets);
542 SAVEI32(PL_lex_casemods);
543 SAVEI32(PL_lex_starts);
544 SAVEI32(PL_lex_state);
545 SAVEVPTR(PL_lex_inpat);
546 SAVEI32(PL_lex_inwhat);
547 if (PL_lex_state == LEX_KNOWNEXT) {
548 I32 toke = PL_nexttoke;
549 while (--toke >= 0) {
550 SAVEI32(PL_nexttype[toke]);
551 SAVEVPTR(PL_nextval[toke]);
553 SAVEI32(PL_nexttoke);
555 SAVECOPLINE(PL_curcop);
558 SAVEPPTR(PL_oldbufptr);
559 SAVEPPTR(PL_oldoldbufptr);
560 SAVEPPTR(PL_last_lop);
561 SAVEPPTR(PL_last_uni);
562 SAVEPPTR(PL_linestart);
563 SAVESPTR(PL_linestr);
564 SAVEGENERICPV(PL_lex_brackstack);
565 SAVEGENERICPV(PL_lex_casestack);
566 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
567 SAVESPTR(PL_lex_stuff);
568 SAVEI32(PL_lex_defer);
569 SAVEI32(PL_sublex_info.sub_inwhat);
570 SAVESPTR(PL_lex_repl);
572 SAVEINT(PL_lex_expect);
574 PL_lex_state = LEX_NORMAL;
578 New(899, PL_lex_brackstack, 120, char);
579 New(899, PL_lex_casestack, 12, char);
581 *PL_lex_casestack = '\0';
584 PL_lex_stuff = Nullsv;
585 PL_lex_repl = Nullsv;
589 PL_sublex_info.sub_inwhat = 0;
591 if (SvREADONLY(PL_linestr))
592 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
593 s = SvPV(PL_linestr, len);
594 if (!len || s[len-1] != ';') {
595 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
596 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
597 sv_catpvn(PL_linestr, "\n;", 2);
599 SvTEMP_off(PL_linestr);
600 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
601 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
602 PL_last_lop = PL_last_uni = Nullch;
608 * Finalizer for lexing operations. Must be called when the parser is
609 * done with the lexer.
615 PL_doextract = FALSE;
620 * This subroutine has nothing to do with tilting, whether at windmills
621 * or pinball tables. Its name is short for "increment line". It
622 * increments the current line number in CopLINE(PL_curcop) and checks
623 * to see whether the line starts with a comment of the form
624 * # line 500 "foo.pm"
625 * If so, it sets the current line number and file to the values in the comment.
629 S_incline(pTHX_ char *s)
636 CopLINE_inc(PL_curcop);
639 while (SPACE_OR_TAB(*s)) s++;
640 if (strnEQ(s, "line", 4))
644 if (SPACE_OR_TAB(*s))
648 while (SPACE_OR_TAB(*s)) s++;
654 while (SPACE_OR_TAB(*s))
656 if (*s == '"' && (t = strchr(s+1, '"'))) {
661 for (t = s; !isSPACE(*t); t++) ;
664 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
666 if (*e != '\n' && *e != '\0')
667 return; /* false alarm */
672 CopFILE_free(PL_curcop);
673 CopFILE_set(PL_curcop, s);
676 CopLINE_set(PL_curcop, atoi(n)-1);
681 * Called to gobble the appropriate amount and type of whitespace.
682 * Skips comments as well.
686 S_skipspace(pTHX_ register char *s)
688 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
689 while (s < PL_bufend && SPACE_OR_TAB(*s))
695 SSize_t oldprevlen, oldoldprevlen;
696 SSize_t oldloplen = 0, oldunilen = 0;
697 while (s < PL_bufend && isSPACE(*s)) {
698 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
703 if (s < PL_bufend && *s == '#') {
704 while (s < PL_bufend && *s != '\n')
708 if (PL_in_eval && !PL_rsfp) {
715 /* only continue to recharge the buffer if we're at the end
716 * of the buffer, we're not reading from a source filter, and
717 * we're in normal lexing mode
719 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
720 PL_lex_state == LEX_FORMLINE)
723 /* try to recharge the buffer */
724 if ((s = filter_gets(PL_linestr, PL_rsfp,
725 (prevlen = SvCUR(PL_linestr)))) == Nullch)
727 /* end of file. Add on the -p or -n magic */
730 ";}continue{print or die qq(-p destination: $!\\n);}");
731 PL_minus_n = PL_minus_p = 0;
733 else if (PL_minus_n) {
734 sv_setpvn(PL_linestr, ";}", 2);
738 sv_setpvn(PL_linestr,";", 1);
740 /* reset variables for next time we lex */
741 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
743 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
744 PL_last_lop = PL_last_uni = Nullch;
746 /* Close the filehandle. Could be from -P preprocessor,
747 * STDIN, or a regular file. If we were reading code from
748 * STDIN (because the commandline held no -e or filename)
749 * then we don't close it, we reset it so the code can
750 * read from STDIN too.
753 if (PL_preprocess && !PL_in_eval)
754 (void)PerlProc_pclose(PL_rsfp);
755 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
756 PerlIO_clearerr(PL_rsfp);
758 (void)PerlIO_close(PL_rsfp);
763 /* not at end of file, so we only read another line */
764 /* make corresponding updates to old pointers, for yyerror() */
765 oldprevlen = PL_oldbufptr - PL_bufend;
766 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
768 oldunilen = PL_last_uni - PL_bufend;
770 oldloplen = PL_last_lop - PL_bufend;
771 PL_linestart = PL_bufptr = s + prevlen;
772 PL_bufend = s + SvCUR(PL_linestr);
774 PL_oldbufptr = s + oldprevlen;
775 PL_oldoldbufptr = s + oldoldprevlen;
777 PL_last_uni = s + oldunilen;
779 PL_last_lop = s + oldloplen;
782 /* debugger active and we're not compiling the debugger code,
783 * so store the line into the debugger's array of lines
785 if (PERLDB_LINE && PL_curstash != PL_debstash) {
786 SV *sv = NEWSV(85,0);
788 sv_upgrade(sv, SVt_PVMG);
789 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
792 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
799 * Check the unary operators to ensure there's no ambiguity in how they're
800 * used. An ambiguous piece of code would be:
802 * This doesn't mean rand() + 5. Because rand() is a unary operator,
803 * the +5 is its argument.
812 if (PL_oldoldbufptr != PL_last_uni)
814 while (isSPACE(*PL_last_uni))
816 for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ;
817 if ((t = strchr(s, '(')) && t < PL_bufptr)
819 if (ckWARN_d(WARN_AMBIGUOUS)){
822 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
823 "Warning: Use of \"%s\" without parentheses is ambiguous",
830 * LOP : macro to build a list operator. Its behaviour has been replaced
831 * with a subroutine, S_lop() for which LOP is just another name.
834 #define LOP(f,x) return lop(f,x,s)
838 * Build a list operator (or something that might be one). The rules:
839 * - if we have a next token, then it's a list operator [why?]
840 * - if the next thing is an opening paren, then it's a function
841 * - else it's a list operator
845 S_lop(pTHX_ I32 f, int x, char *s)
851 PL_last_lop = PL_oldbufptr;
852 PL_last_lop_op = (OPCODE)f;
854 return REPORT(LSTOP);
861 return REPORT(LSTOP);
866 * When the lexer realizes it knows the next token (for instance,
867 * it is reordering tokens for the parser) then it can call S_force_next
868 * to know what token to return the next time the lexer is called. Caller
869 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
870 * handles the token correctly.
874 S_force_next(pTHX_ I32 type)
876 PL_nexttype[PL_nexttoke] = type;
878 if (PL_lex_state != LEX_KNOWNEXT) {
879 PL_lex_defer = PL_lex_state;
880 PL_lex_expect = PL_expect;
881 PL_lex_state = LEX_KNOWNEXT;
886 S_newSV_maybe_utf8(pTHX_ const char *start, STRLEN len)
888 SV *sv = newSVpvn(start,len);
889 if (UTF && !IN_BYTES && is_utf8_string((const U8*)start, len))
896 * When the lexer knows the next thing is a word (for instance, it has
897 * just seen -> and it knows that the next char is a word char, then
898 * it calls S_force_word to stick the next word into the PL_next lookahead.
901 * char *start : buffer position (must be within PL_linestr)
902 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
903 * int check_keyword : if true, Perl checks to make sure the word isn't
904 * a keyword (do this if the word is a label, e.g. goto FOO)
905 * int allow_pack : if true, : characters will also be allowed (require,
907 * int allow_initial_tick : used by the "sub" lexer only.
911 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
916 start = skipspace(start);
918 if (isIDFIRST_lazy_if(s,UTF) ||
919 (allow_pack && *s == ':') ||
920 (allow_initial_tick && *s == '\'') )
922 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
923 if (check_keyword && keyword(PL_tokenbuf, len))
925 if (token == METHOD) {
930 PL_expect = XOPERATOR;
933 PL_nextval[PL_nexttoke].opval
934 = (OP*)newSVOP(OP_CONST,0,
935 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
936 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
944 * Called when the lexer wants $foo *foo &foo etc, but the program
945 * text only contains the "foo" portion. The first argument is a pointer
946 * to the "foo", and the second argument is the type symbol to prefix.
947 * Forces the next token to be a "WORD".
948 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
952 S_force_ident(pTHX_ register const char *s, int kind)
955 OP* const o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
956 PL_nextval[PL_nexttoke].opval = o;
959 o->op_private = OPpCONST_ENTERED;
960 /* XXX see note in pp_entereval() for why we forgo typo
961 warnings if the symbol must be introduced in an eval.
963 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
964 kind == '$' ? SVt_PV :
965 kind == '@' ? SVt_PVAV :
966 kind == '%' ? SVt_PVHV :
974 Perl_str_to_version(pTHX_ SV *sv)
979 const char *start = SvPVx(sv,len);
980 const char *end = start + len;
981 bool utf = SvUTF8(sv) ? TRUE : FALSE;
982 while (start < end) {
986 n = utf8n_to_uvchr((U8*)start, len, &skip, 0);
991 retval += ((NV)n)/nshift;
1000 * Forces the next token to be a version number.
1001 * If the next token appears to be an invalid version number, (e.g. "v2b"),
1002 * and if "guessing" is TRUE, then no new token is created (and the caller
1003 * must use an alternative parsing method).
1007 S_force_version(pTHX_ char *s, int guessing)
1009 OP *version = Nullop;
1018 while (isDIGIT(*d) || *d == '_' || *d == '.')
1020 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
1022 s = scan_num(s, &yylval);
1023 version = yylval.opval;
1024 ver = cSVOPx(version)->op_sv;
1025 if (SvPOK(ver) && !SvNIOK(ver)) {
1026 (void)SvUPGRADE(ver, SVt_PVNV);
1027 SvNV_set(ver, str_to_version(ver));
1028 SvNOK_on(ver); /* hint that it is a version */
1035 /* NOTE: The parser sees the package name and the VERSION swapped */
1036 PL_nextval[PL_nexttoke].opval = version;
1044 * Tokenize a quoted string passed in as an SV. It finds the next
1045 * chunk, up to end of string or a backslash. It may make a new
1046 * SV containing that chunk (if HINT_NEW_STRING is on). It also
1051 S_tokeq(pTHX_ SV *sv)
1054 register char *send;
1062 s = SvPV_force(sv, len);
1063 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
1066 while (s < send && *s != '\\')
1071 if ( PL_hints & HINT_NEW_STRING ) {
1072 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
1078 if (s + 1 < send && (s[1] == '\\'))
1079 s++; /* all that, just for this */
1084 SvCUR_set(sv, d - SvPVX(sv));
1086 if ( PL_hints & HINT_NEW_STRING )
1087 return new_constant(NULL, 0, "q", sv, pv, "q");
1092 * Now come three functions related to double-quote context,
1093 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
1094 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
1095 * interact with PL_lex_state, and create fake ( ... ) argument lists
1096 * to handle functions and concatenation.
1097 * They assume that whoever calls them will be setting up a fake
1098 * join call, because each subthing puts a ',' after it. This lets
1101 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
1103 * (I'm not sure whether the spurious commas at the end of lcfirst's
1104 * arguments and join's arguments are created or not).
1109 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
1111 * Pattern matching will set PL_lex_op to the pattern-matching op to
1112 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
1114 * OP_CONST and OP_READLINE are easy--just make the new op and return.
1116 * Everything else becomes a FUNC.
1118 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
1119 * had an OP_CONST or OP_READLINE). This just sets us up for a
1120 * call to S_sublex_push().
1124 S_sublex_start(pTHX)
1126 const register I32 op_type = yylval.ival;
1128 if (op_type == OP_NULL) {
1129 yylval.opval = PL_lex_op;
1133 if (op_type == OP_CONST || op_type == OP_READLINE) {
1134 SV *sv = tokeq(PL_lex_stuff);
1136 if (SvTYPE(sv) == SVt_PVIV) {
1137 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
1139 const char *p = SvPV(sv, len);
1140 SV * const nsv = newSVpvn(p, len);
1146 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
1147 PL_lex_stuff = Nullsv;
1148 /* Allow <FH> // "foo" */
1149 if (op_type == OP_READLINE)
1150 PL_expect = XTERMORDORDOR;
1154 PL_sublex_info.super_state = PL_lex_state;
1155 PL_sublex_info.sub_inwhat = op_type;
1156 PL_sublex_info.sub_op = PL_lex_op;
1157 PL_lex_state = LEX_INTERPPUSH;
1161 yylval.opval = PL_lex_op;
1171 * Create a new scope to save the lexing state. The scope will be
1172 * ended in S_sublex_done. Returns a '(', starting the function arguments
1173 * to the uc, lc, etc. found before.
1174 * Sets PL_lex_state to LEX_INTERPCONCAT.
1183 PL_lex_state = PL_sublex_info.super_state;
1184 SAVEI32(PL_lex_dojoin);
1185 SAVEI32(PL_lex_brackets);
1186 SAVEI32(PL_lex_casemods);
1187 SAVEI32(PL_lex_starts);
1188 SAVEI32(PL_lex_state);
1189 SAVEVPTR(PL_lex_inpat);
1190 SAVEI32(PL_lex_inwhat);
1191 SAVECOPLINE(PL_curcop);
1192 SAVEPPTR(PL_bufptr);
1193 SAVEPPTR(PL_bufend);
1194 SAVEPPTR(PL_oldbufptr);
1195 SAVEPPTR(PL_oldoldbufptr);
1196 SAVEPPTR(PL_last_lop);
1197 SAVEPPTR(PL_last_uni);
1198 SAVEPPTR(PL_linestart);
1199 SAVESPTR(PL_linestr);
1200 SAVEGENERICPV(PL_lex_brackstack);
1201 SAVEGENERICPV(PL_lex_casestack);
1203 PL_linestr = PL_lex_stuff;
1204 PL_lex_stuff = Nullsv;
1206 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1207 = SvPVX(PL_linestr);
1208 PL_bufend += SvCUR(PL_linestr);
1209 PL_last_lop = PL_last_uni = Nullch;
1210 SAVEFREESV(PL_linestr);
1212 PL_lex_dojoin = FALSE;
1213 PL_lex_brackets = 0;
1214 New(899, PL_lex_brackstack, 120, char);
1215 New(899, PL_lex_casestack, 12, char);
1216 PL_lex_casemods = 0;
1217 *PL_lex_casestack = '\0';
1219 PL_lex_state = LEX_INTERPCONCAT;
1220 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
1222 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1223 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1224 PL_lex_inpat = PL_sublex_info.sub_op;
1226 PL_lex_inpat = Nullop;
1233 * Restores lexer state after a S_sublex_push.
1240 if (!PL_lex_starts++) {
1241 SV *sv = newSVpvn("",0);
1242 if (SvUTF8(PL_linestr))
1244 PL_expect = XOPERATOR;
1245 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1249 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1250 PL_lex_state = LEX_INTERPCASEMOD;
1254 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1255 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1256 PL_linestr = PL_lex_repl;
1258 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1259 PL_bufend += SvCUR(PL_linestr);
1260 PL_last_lop = PL_last_uni = Nullch;
1261 SAVEFREESV(PL_linestr);
1262 PL_lex_dojoin = FALSE;
1263 PL_lex_brackets = 0;
1264 PL_lex_casemods = 0;
1265 *PL_lex_casestack = '\0';
1267 if (SvEVALED(PL_lex_repl)) {
1268 PL_lex_state = LEX_INTERPNORMAL;
1270 /* we don't clear PL_lex_repl here, so that we can check later
1271 whether this is an evalled subst; that means we rely on the
1272 logic to ensure sublex_done() is called again only via the
1273 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1276 PL_lex_state = LEX_INTERPCONCAT;
1277 PL_lex_repl = Nullsv;
1283 PL_bufend = SvPVX(PL_linestr);
1284 PL_bufend += SvCUR(PL_linestr);
1285 PL_expect = XOPERATOR;
1286 PL_sublex_info.sub_inwhat = 0;
1294 Extracts a pattern, double-quoted string, or transliteration. This
1297 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1298 processing a pattern (PL_lex_inpat is true), a transliteration
1299 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1301 Returns a pointer to the character scanned up to. Iff this is
1302 advanced from the start pointer supplied (ie if anything was
1303 successfully parsed), will leave an OP for the substring scanned
1304 in yylval. Caller must intuit reason for not parsing further
1305 by looking at the next characters herself.
1309 double-quoted style: \r and \n
1310 regexp special ones: \D \s
1312 backrefs: \1 (deprecated in substitution replacements)
1313 case and quoting: \U \Q \E
1314 stops on @ and $, but not for $ as tail anchor
1316 In transliterations:
1317 characters are VERY literal, except for - not at the start or end
1318 of the string, which indicates a range. scan_const expands the
1319 range to the full set of intermediate characters.
1321 In double-quoted strings:
1323 double-quoted style: \r and \n
1325 backrefs: \1 (deprecated)
1326 case and quoting: \U \Q \E
1329 scan_const does *not* construct ops to handle interpolated strings.
1330 It stops processing as soon as it finds an embedded $ or @ variable
1331 and leaves it to the caller to work out what's going on.
1333 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @::foo.
1335 $ in pattern could be $foo or could be tail anchor. Assumption:
1336 it's a tail anchor if $ is the last thing in the string, or if it's
1337 followed by one of ")| \n\t"
1339 \1 (backreferences) are turned into $1
1341 The structure of the code is
1342 while (there's a character to process) {
1343 handle transliteration ranges
1344 skip regexp comments
1345 skip # initiated comments in //x patterns
1346 check for embedded @foo
1347 check for embedded scalars
1349 leave intact backslashes from leave (below)
1350 deprecate \1 in strings and sub replacements
1351 handle string-changing backslashes \l \U \Q \E, etc.
1352 switch (what was escaped) {
1353 handle - in a transliteration (becomes a literal -)
1354 handle \132 octal characters
1355 handle 0x15 hex characters
1356 handle \cV (control V)
1357 handle printf backslashes (\f, \r, \n, etc)
1359 } (end if backslash)
1360 } (end while character to read)
1365 S_scan_const(pTHX_ char *start)
1367 register char *send = PL_bufend; /* end of the constant */
1368 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1369 register char *s = start; /* start of the constant */
1370 register char *d = SvPVX(sv); /* destination for copies */
1371 bool dorange = FALSE; /* are we in a translit range? */
1372 bool didrange = FALSE; /* did we just finish a range? */
1373 I32 has_utf8 = FALSE; /* Output constant is UTF8 */
1374 I32 this_utf8 = UTF; /* The source string is assumed to be UTF8 */
1377 const char *leaveit = /* set of acceptably-backslashed characters */
1379 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxz0123456789[{]} \t\n\r\f\v#"
1382 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1383 /* If we are doing a trans and we know we want UTF8 set expectation */
1384 has_utf8 = PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF);
1385 this_utf8 = PL_sublex_info.sub_op->op_private & (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1389 while (s < send || dorange) {
1390 /* get transliterations out of the way (they're most literal) */
1391 if (PL_lex_inwhat == OP_TRANS) {
1392 /* expand a range A-Z to the full set of characters. AIE! */
1394 I32 i; /* current expanded character */
1395 I32 min; /* first character in range */
1396 I32 max; /* last character in range */
1399 char *c = (char*)utf8_hop((U8*)d, -1);
1403 *c = (char)UTF_TO_NATIVE(0xff);
1404 /* mark the range as done, and continue */
1410 i = d - SvPVX(sv); /* remember current offset */
1411 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1412 d = SvPVX(sv) + i; /* refresh d after realloc */
1413 d -= 2; /* eat the first char and the - */
1415 min = (U8)*d; /* first char in range */
1416 max = (U8)d[1]; /* last char in range */
1420 "Invalid range \"%c-%c\" in transliteration operator",
1421 (char)min, (char)max);
1425 if ((isLOWER(min) && isLOWER(max)) ||
1426 (isUPPER(min) && isUPPER(max))) {
1428 for (i = min; i <= max; i++)
1430 *d++ = NATIVE_TO_NEED(has_utf8,i);
1432 for (i = min; i <= max; i++)
1434 *d++ = NATIVE_TO_NEED(has_utf8,i);
1439 for (i = min; i <= max; i++)
1442 /* mark the range as done, and continue */
1448 /* range begins (ignore - as first or last char) */
1449 else if (*s == '-' && s+1 < send && s != start) {
1451 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
1454 *d++ = (char)UTF_TO_NATIVE(0xff); /* use illegal utf8 byte--see pmtrans */
1466 /* if we get here, we're not doing a transliteration */
1468 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1469 except for the last char, which will be done separately. */
1470 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1472 while (s+1 < send && *s != ')')
1473 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1475 else if (s[2] == '{' /* This should match regcomp.c */
1476 || ((s[2] == 'p' || s[2] == '?') && s[3] == '{'))
1479 char *regparse = s + (s[2] == '{' ? 3 : 4);
1482 while (count && (c = *regparse)) {
1483 if (c == '\\' && regparse[1])
1491 if (*regparse != ')')
1492 regparse--; /* Leave one char for continuation. */
1493 while (s < regparse)
1494 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1498 /* likewise skip #-initiated comments in //x patterns */
1499 else if (*s == '#' && PL_lex_inpat &&
1500 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1501 while (s+1 < send && *s != '\n')
1502 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1505 /* check for embedded arrays
1506 (@foo, @::foo, @'foo, @{foo}, @$foo, @+, @-)
1508 else if (*s == '@' && s[1]
1509 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$+-", s[1])))
1512 /* check for embedded scalars. only stop if we're sure it's a
1515 else if (*s == '$') {
1516 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1518 if (s + 1 < send && !strchr("()| \r\n\t", s[1]))
1519 break; /* in regexp, $ might be tail anchor */
1522 /* End of else if chain - OP_TRANS rejoin rest */
1525 if (*s == '\\' && s+1 < send) {
1528 /* some backslashes we leave behind */
1529 if (*leaveit && *s && strchr(leaveit, *s)) {
1530 *d++ = NATIVE_TO_NEED(has_utf8,'\\');
1531 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1535 /* deprecate \1 in strings and substitution replacements */
1536 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1537 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1539 if (ckWARN(WARN_SYNTAX))
1540 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "\\%c better written as $%c", *s, *s);
1545 /* string-change backslash escapes */
1546 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1551 /* if we get here, it's either a quoted -, or a digit */
1554 /* quoted - in transliterations */
1556 if (PL_lex_inwhat == OP_TRANS) {
1563 if (ckWARN(WARN_MISC) &&
1566 Perl_warner(aTHX_ packWARN(WARN_MISC),
1567 "Unrecognized escape \\%c passed through",
1569 /* default action is to copy the quoted character */
1570 goto default_action;
1573 /* \132 indicates an octal constant */
1574 case '0': case '1': case '2': case '3':
1575 case '4': case '5': case '6': case '7':
1579 uv = grok_oct(s, &len, &flags, NULL);
1582 goto NUM_ESCAPE_INSERT;
1584 /* \x24 indicates a hex constant */
1588 char* e = strchr(s, '}');
1589 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
1590 PERL_SCAN_DISALLOW_PREFIX;
1595 yyerror("Missing right brace on \\x{}");
1599 uv = grok_hex(s, &len, &flags, NULL);
1605 I32 flags = PERL_SCAN_DISALLOW_PREFIX;
1606 uv = grok_hex(s, &len, &flags, NULL);
1612 /* Insert oct or hex escaped character.
1613 * There will always enough room in sv since such
1614 * escapes will be longer than any UTF-8 sequence
1615 * they can end up as. */
1617 /* We need to map to chars to ASCII before doing the tests
1620 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(uv))) {
1621 if (!has_utf8 && uv > 255) {
1622 /* Might need to recode whatever we have
1623 * accumulated so far if it contains any
1626 * (Can't we keep track of that and avoid
1627 * this rescan? --jhi)
1631 for (c = (U8 *) SvPVX(sv); c < (U8 *)d; c++) {
1632 if (!NATIVE_IS_INVARIANT(*c)) {
1637 STRLEN offset = d - SvPVX(sv);
1639 d = SvGROW(sv, SvLEN(sv) + hicount + 1) + offset;
1643 while (src >= (U8 *)SvPVX(sv)) {
1644 if (!NATIVE_IS_INVARIANT(*src)) {
1645 U8 ch = NATIVE_TO_ASCII(*src);
1646 *dst-- = (U8)UTF8_EIGHT_BIT_LO(ch);
1647 *dst-- = (U8)UTF8_EIGHT_BIT_HI(ch);
1657 if (has_utf8 || uv > 255) {
1658 d = (char*)uvchr_to_utf8((U8*)d, uv);
1660 if (PL_lex_inwhat == OP_TRANS &&
1661 PL_sublex_info.sub_op) {
1662 PL_sublex_info.sub_op->op_private |=
1663 (PL_lex_repl ? OPpTRANS_FROM_UTF
1676 /* \N{LATIN SMALL LETTER A} is a named character */
1680 char* e = strchr(s, '}');
1686 yyerror("Missing right brace on \\N{}");
1690 if (e > s + 2 && s[1] == 'U' && s[2] == '+') {
1692 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
1693 PERL_SCAN_DISALLOW_PREFIX;
1696 uv = grok_hex(s, &len, &flags, NULL);
1698 goto NUM_ESCAPE_INSERT;
1700 res = newSVpvn(s + 1, e - s - 1);
1701 res = new_constant( Nullch, 0, "charnames",
1702 res, Nullsv, "\\N{...}" );
1704 sv_utf8_upgrade(res);
1705 str = SvPV(res,len);
1706 #ifdef EBCDIC_NEVER_MIND
1707 /* charnames uses pack U and that has been
1708 * recently changed to do the below uni->native
1709 * mapping, so this would be redundant (and wrong,
1710 * the code point would be doubly converted).
1711 * But leave this in just in case the pack U change
1712 * gets revoked, but the semantics is still
1713 * desireable for charnames. --jhi */
1715 UV uv = utf8_to_uvchr((U8*)str, 0);
1718 U8 tmpbuf[UTF8_MAXBYTES+1], *d;
1720 d = uvchr_to_utf8(tmpbuf, UNI_TO_NATIVE(uv));
1721 sv_setpvn(res, (char *)tmpbuf, d - tmpbuf);
1722 str = SvPV(res, len);
1726 if (!has_utf8 && SvUTF8(res)) {
1727 char *ostart = SvPVX(sv);
1728 SvCUR_set(sv, d - ostart);
1731 sv_utf8_upgrade(sv);
1732 /* this just broke our allocation above... */
1733 SvGROW(sv, (STRLEN)(send - start));
1734 d = SvPVX(sv) + SvCUR(sv);
1737 if (len > (STRLEN)(e - s + 4)) { /* I _guess_ 4 is \N{} --jhi */
1738 char *odest = SvPVX(sv);
1740 SvGROW(sv, (SvLEN(sv) + len - (e - s + 4)));
1741 d = SvPVX(sv) + (d - odest);
1743 Copy(str, d, len, char);
1750 yyerror("Missing braces on \\N{}");
1753 /* \c is a control character */
1762 *d++ = NATIVE_TO_NEED(has_utf8,toCTRL(c));
1765 yyerror("Missing control char name in \\c");
1769 /* printf-style backslashes, formfeeds, newlines, etc */
1771 *d++ = NATIVE_TO_NEED(has_utf8,'\b');
1774 *d++ = NATIVE_TO_NEED(has_utf8,'\n');
1777 *d++ = NATIVE_TO_NEED(has_utf8,'\r');
1780 *d++ = NATIVE_TO_NEED(has_utf8,'\f');
1783 *d++ = NATIVE_TO_NEED(has_utf8,'\t');
1786 *d++ = ASCII_TO_NEED(has_utf8,'\033');
1789 *d++ = ASCII_TO_NEED(has_utf8,'\007');
1795 } /* end if (backslash) */
1798 /* If we started with encoded form, or already know we want it
1799 and then encode the next character */
1800 if ((has_utf8 || this_utf8) && !NATIVE_IS_INVARIANT((U8)(*s))) {
1802 UV uv = (this_utf8) ? utf8n_to_uvchr((U8*)s, send - s, &len, 0) : (UV) ((U8) *s);
1803 STRLEN need = UNISKIP(NATIVE_TO_UNI(uv));
1806 /* encoded value larger than old, need extra space (NOTE: SvCUR() not set here) */
1807 STRLEN off = d - SvPVX(sv);
1808 d = SvGROW(sv, SvLEN(sv) + (need-len)) + off;
1810 d = (char*)uvchr_to_utf8((U8*)d, uv);
1814 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1816 } /* while loop to process each character */
1818 /* terminate the string and set up the sv */
1820 SvCUR_set(sv, d - SvPVX(sv));
1821 if (SvCUR(sv) >= SvLEN(sv))
1822 Perl_croak(aTHX_ "panic: constant overflowed allocated space");
1825 if (PL_encoding && !has_utf8) {
1826 sv_recode_to_utf8(sv, PL_encoding);
1832 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1833 PL_sublex_info.sub_op->op_private |=
1834 (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1838 /* shrink the sv if we allocated more than we used */
1839 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1840 SvPV_shrink_to_cur(sv);
1843 /* return the substring (via yylval) only if we parsed anything */
1844 if (s > PL_bufptr) {
1845 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1846 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1848 ( PL_lex_inwhat == OP_TRANS
1850 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1853 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1860 * Returns TRUE if there's more to the expression (e.g., a subscript),
1863 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1865 * ->[ and ->{ return TRUE
1866 * { and [ outside a pattern are always subscripts, so return TRUE
1867 * if we're outside a pattern and it's not { or [, then return FALSE
1868 * if we're in a pattern and the first char is a {
1869 * {4,5} (any digits around the comma) returns FALSE
1870 * if we're in a pattern and the first char is a [
1872 * [SOMETHING] has a funky algorithm to decide whether it's a
1873 * character class or not. It has to deal with things like
1874 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1875 * anything else returns TRUE
1878 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1881 S_intuit_more(pTHX_ register char *s)
1883 if (PL_lex_brackets)
1885 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1887 if (*s != '{' && *s != '[')
1892 /* In a pattern, so maybe we have {n,m}. */
1909 /* On the other hand, maybe we have a character class */
1912 if (*s == ']' || *s == '^')
1915 /* this is terrifying, and it works */
1916 int weight = 2; /* let's weigh the evidence */
1918 unsigned char un_char = 255, last_un_char;
1919 const char *send = strchr(s,']');
1920 char tmpbuf[sizeof PL_tokenbuf * 4];
1922 if (!send) /* has to be an expression */
1925 Zero(seen,256,char);
1928 else if (isDIGIT(*s)) {
1930 if (isDIGIT(s[1]) && s[2] == ']')
1936 for (; s < send; s++) {
1937 last_un_char = un_char;
1938 un_char = (unsigned char)*s;
1943 weight -= seen[un_char] * 10;
1944 if (isALNUM_lazy_if(s+1,UTF)) {
1945 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1946 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1951 else if (*s == '$' && s[1] &&
1952 strchr("[#!%*<>()-=",s[1])) {
1953 if (/*{*/ strchr("])} =",s[2]))
1962 if (strchr("wds]",s[1]))
1964 else if (seen['\''] || seen['"'])
1966 else if (strchr("rnftbxcav",s[1]))
1968 else if (isDIGIT(s[1])) {
1970 while (s[1] && isDIGIT(s[1]))
1980 if (strchr("aA01! ",last_un_char))
1982 if (strchr("zZ79~",s[1]))
1984 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1985 weight -= 5; /* cope with negative subscript */
1988 if (!isALNUM(last_un_char)
1989 && !(last_un_char == '$' || last_un_char == '@'
1990 || last_un_char == '&')
1991 && isALPHA(*s) && s[1] && isALPHA(s[1])) {
1996 if (keyword(tmpbuf, d - tmpbuf))
1999 if (un_char == last_un_char + 1)
2001 weight -= seen[un_char];
2006 if (weight >= 0) /* probably a character class */
2016 * Does all the checking to disambiguate
2018 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
2019 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
2021 * First argument is the stuff after the first token, e.g. "bar".
2023 * Not a method if bar is a filehandle.
2024 * Not a method if foo is a subroutine prototyped to take a filehandle.
2025 * Not a method if it's really "Foo $bar"
2026 * Method if it's "foo $bar"
2027 * Not a method if it's really "print foo $bar"
2028 * Method if it's really "foo package::" (interpreted as package->foo)
2029 * Not a method if bar is known to be a subroutine ("sub bar; foo bar")
2030 * Not a method if bar is a filehandle or package, but is quoted with
2035 S_intuit_method(pTHX_ char *start, GV *gv)
2037 char *s = start + (*start == '$');
2038 char tmpbuf[sizeof PL_tokenbuf];
2046 if ((cv = GvCVu(gv))) {
2047 const char *proto = SvPVX(cv);
2057 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
2058 /* start is the beginning of the possible filehandle/object,
2059 * and s is the end of it
2060 * tmpbuf is a copy of it
2063 if (*start == '$') {
2064 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
2069 return *s == '(' ? FUNCMETH : METHOD;
2071 if (!keyword(tmpbuf, len)) {
2072 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
2077 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
2078 if (indirgv && GvCVu(indirgv))
2080 /* filehandle or package name makes it a method */
2081 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
2083 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
2084 return 0; /* no assumptions -- "=>" quotes bearword */
2086 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
2087 newSVpvn(tmpbuf,len));
2088 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
2092 return *s == '(' ? FUNCMETH : METHOD;
2100 * Return a string of Perl code to load the debugger. If PERL5DB
2101 * is set, it will return the contents of that, otherwise a
2102 * compile-time require of perl5db.pl.
2109 const char *pdb = PerlEnv_getenv("PERL5DB");
2113 SETERRNO(0,SS_NORMAL);
2114 return "BEGIN { require 'perl5db.pl' }";
2120 /* Encoded script support. filter_add() effectively inserts a
2121 * 'pre-processing' function into the current source input stream.
2122 * Note that the filter function only applies to the current source file
2123 * (e.g., it will not affect files 'require'd or 'use'd by this one).
2125 * The datasv parameter (which may be NULL) can be used to pass
2126 * private data to this instance of the filter. The filter function
2127 * can recover the SV using the FILTER_DATA macro and use it to
2128 * store private buffers and state information.
2130 * The supplied datasv parameter is upgraded to a PVIO type
2131 * and the IoDIRP/IoANY field is used to store the function pointer,
2132 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
2133 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
2134 * private use must be set using malloc'd pointers.
2138 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
2145 if (!PL_rsfp_filters)
2146 PL_rsfp_filters = newAV();
2148 datasv = NEWSV(255,0);
2149 if (!SvUPGRADE(datasv, SVt_PVIO))
2150 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
2152 IoANY(datasv) = u.iop; /* stash funcp into spare field */
2153 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
2154 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
2155 (void*)funcp, SvPV_nolen(datasv)));
2156 av_unshift(PL_rsfp_filters, 1);
2157 av_store(PL_rsfp_filters, 0, datasv) ;
2162 /* Delete most recently added instance of this filter function. */
2164 Perl_filter_del(pTHX_ filter_t funcp)
2169 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", (void*)funcp));
2170 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
2172 /* if filter is on top of stack (usual case) just pop it off */
2173 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
2174 u.iop = IoANY(datasv);
2175 if (u.filter == funcp) {
2176 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
2177 IoANY(datasv) = (void *)NULL;
2178 sv_free(av_pop(PL_rsfp_filters));
2182 /* we need to search for the correct entry and clear it */
2183 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
2187 /* Invoke the idxth filter function for the current rsfp. */
2188 /* maxlen 0 = read one text line */
2190 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
2196 if (!PL_rsfp_filters)
2198 if (idx > AvFILLp(PL_rsfp_filters)) { /* Any more filters? */
2199 /* Provide a default input filter to make life easy. */
2200 /* Note that we append to the line. This is handy. */
2201 DEBUG_P(PerlIO_printf(Perl_debug_log,
2202 "filter_read %d: from rsfp\n", idx));
2206 const int old_len = SvCUR(buf_sv);
2208 /* ensure buf_sv is large enough */
2209 SvGROW(buf_sv, (STRLEN)(old_len + maxlen)) ;
2210 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
2211 if (PerlIO_error(PL_rsfp))
2212 return -1; /* error */
2214 return 0 ; /* end of file */
2216 SvCUR_set(buf_sv, old_len + len) ;
2219 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
2220 if (PerlIO_error(PL_rsfp))
2221 return -1; /* error */
2223 return 0 ; /* end of file */
2226 return SvCUR(buf_sv);
2228 /* Skip this filter slot if filter has been deleted */
2229 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef) {
2230 DEBUG_P(PerlIO_printf(Perl_debug_log,
2231 "filter_read %d: skipped (filter deleted)\n",
2233 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
2235 /* Get function pointer hidden within datasv */
2236 u.iop = IoANY(datasv);
2238 DEBUG_P(PerlIO_printf(Perl_debug_log,
2239 "filter_read %d: via function %p (%s)\n",
2240 idx, (void*)funcp, SvPV_nolen(datasv)));
2241 /* Call function. The function is expected to */
2242 /* call "FILTER_READ(idx+1, buf_sv)" first. */
2243 /* Return: <0:error, =0:eof, >0:not eof */
2244 return (*funcp)(aTHX_ idx, buf_sv, maxlen);
2248 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
2250 #ifdef PERL_CR_FILTER
2251 if (!PL_rsfp_filters) {
2252 filter_add(S_cr_textfilter,NULL);
2255 if (PL_rsfp_filters) {
2257 SvCUR_set(sv, 0); /* start with empty line */
2258 if (FILTER_READ(0, sv, 0) > 0)
2259 return ( SvPVX(sv) ) ;
2264 return (sv_gets(sv, fp, append));
2268 S_find_in_my_stash(pTHX_ const char *pkgname, I32 len)
2272 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2276 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2277 (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV)))
2279 return GvHV(gv); /* Foo:: */
2282 /* use constant CLASS => 'MyClass' */
2283 if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) {
2285 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2286 pkgname = SvPV_nolen(sv);
2290 return gv_stashpv(pkgname, FALSE);
2294 static const char* const exp_name[] =
2295 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2296 "ATTRTERM", "TERMBLOCK", "TERMORDORDOR"
2303 Works out what to call the token just pulled out of the input
2304 stream. The yacc parser takes care of taking the ops we return and
2305 stitching them into a tree.
2311 if read an identifier
2312 if we're in a my declaration
2313 croak if they tried to say my($foo::bar)
2314 build the ops for a my() declaration
2315 if it's an access to a my() variable
2316 are we in a sort block?
2317 croak if my($a); $a <=> $b
2318 build ops for access to a my() variable
2319 if in a dq string, and they've said @foo and we can't find @foo
2321 build ops for a bareword
2322 if we already built the token before, use it.
2327 #pragma segment Perl_yylex
2332 register char *s = PL_bufptr;
2339 I32 orig_keyword = 0;
2342 PerlIO_printf(Perl_debug_log, "### LEX_%s\n",
2343 lex_state_names[PL_lex_state]);
2345 /* check if there's an identifier for us to look at */
2346 if (PL_pending_ident)
2347 return REPORT(S_pending_ident(aTHX));
2349 /* no identifier pending identification */
2351 switch (PL_lex_state) {
2353 case LEX_NORMAL: /* Some compilers will produce faster */
2354 case LEX_INTERPNORMAL: /* code if we comment these out. */
2358 /* when we've already built the next token, just pull it out of the queue */
2361 yylval = PL_nextval[PL_nexttoke];
2363 PL_lex_state = PL_lex_defer;
2364 PL_expect = PL_lex_expect;
2365 PL_lex_defer = LEX_NORMAL;
2367 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2368 "### Next token after '%s' was known, type %"IVdf"\n", PL_bufptr,
2369 (IV)PL_nexttype[PL_nexttoke]); });
2371 return REPORT(PL_nexttype[PL_nexttoke]);
2373 /* interpolated case modifiers like \L \U, including \Q and \E.
2374 when we get here, PL_bufptr is at the \
2376 case LEX_INTERPCASEMOD:
2378 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2379 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2381 /* handle \E or end of string */
2382 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2384 if (PL_lex_casemods) {
2385 const char oldmod = PL_lex_casestack[--PL_lex_casemods];
2386 PL_lex_casestack[PL_lex_casemods] = '\0';
2388 if (PL_bufptr != PL_bufend
2389 && (oldmod == 'L' || oldmod == 'U' || oldmod == 'Q')) {
2391 PL_lex_state = LEX_INTERPCONCAT;
2395 if (PL_bufptr != PL_bufend)
2397 PL_lex_state = LEX_INTERPCONCAT;
2401 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2402 "### Saw case modifier at '%s'\n", PL_bufptr); });
2404 if (s[1] == '\\' && s[2] == 'E') {
2406 PL_lex_state = LEX_INTERPCONCAT;
2410 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2411 tmp = *s, *s = s[2], s[2] = (char)tmp; /* misordered... */
2412 if ((*s == 'L' || *s == 'U') &&
2413 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U'))) {
2414 PL_lex_casestack[--PL_lex_casemods] = '\0';
2417 if (PL_lex_casemods > 10)
2418 Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2419 PL_lex_casestack[PL_lex_casemods++] = *s;
2420 PL_lex_casestack[PL_lex_casemods] = '\0';
2421 PL_lex_state = LEX_INTERPCONCAT;
2422 PL_nextval[PL_nexttoke].ival = 0;
2425 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2427 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2429 PL_nextval[PL_nexttoke].ival = OP_LC;
2431 PL_nextval[PL_nexttoke].ival = OP_UC;
2433 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2435 Perl_croak(aTHX_ "panic: yylex");
2439 if (PL_lex_starts) {
2442 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
2443 if (PL_lex_casemods == 1 && PL_lex_inpat)
2452 case LEX_INTERPPUSH:
2453 return REPORT(sublex_push());
2455 case LEX_INTERPSTART:
2456 if (PL_bufptr == PL_bufend)
2457 return REPORT(sublex_done());
2458 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2459 "### Interpolated variable at '%s'\n", PL_bufptr); });
2461 PL_lex_dojoin = (*PL_bufptr == '@');
2462 PL_lex_state = LEX_INTERPNORMAL;
2463 if (PL_lex_dojoin) {
2464 PL_nextval[PL_nexttoke].ival = 0;
2466 force_ident("\"", '$');
2467 PL_nextval[PL_nexttoke].ival = 0;
2469 PL_nextval[PL_nexttoke].ival = 0;
2471 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2474 if (PL_lex_starts++) {
2476 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
2477 if (!PL_lex_casemods && PL_lex_inpat)
2484 case LEX_INTERPENDMAYBE:
2485 if (intuit_more(PL_bufptr)) {
2486 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2492 if (PL_lex_dojoin) {
2493 PL_lex_dojoin = FALSE;
2494 PL_lex_state = LEX_INTERPCONCAT;
2497 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2498 && SvEVALED(PL_lex_repl))
2500 if (PL_bufptr != PL_bufend)
2501 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2502 PL_lex_repl = Nullsv;
2505 case LEX_INTERPCONCAT:
2507 if (PL_lex_brackets)
2508 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2510 if (PL_bufptr == PL_bufend)
2511 return REPORT(sublex_done());
2513 if (SvIVX(PL_linestr) == '\'') {
2514 SV *sv = newSVsv(PL_linestr);
2517 else if ( PL_hints & HINT_NEW_RE )
2518 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2519 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2523 s = scan_const(PL_bufptr);
2525 PL_lex_state = LEX_INTERPCASEMOD;
2527 PL_lex_state = LEX_INTERPSTART;
2530 if (s != PL_bufptr) {
2531 PL_nextval[PL_nexttoke] = yylval;
2534 if (PL_lex_starts++) {
2535 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
2536 if (!PL_lex_casemods && PL_lex_inpat)
2549 PL_lex_state = LEX_NORMAL;
2550 s = scan_formline(PL_bufptr);
2551 if (!PL_lex_formbrack)
2557 PL_oldoldbufptr = PL_oldbufptr;
2560 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at [%s]\n",
2561 exp_name[PL_expect], s);
2567 if (isIDFIRST_lazy_if(s,UTF))
2569 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2572 goto fake_eof; /* emulate EOF on ^D or ^Z */
2577 if (PL_lex_brackets) {
2578 if (PL_lex_formbrack)
2579 yyerror("Format not terminated");
2581 yyerror("Missing right curly or square bracket");
2583 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2584 "### Tokener got EOF\n");
2588 if (s++ < PL_bufend)
2589 goto retry; /* ignore stray nulls */
2592 if (!PL_in_eval && !PL_preambled) {
2593 PL_preambled = TRUE;
2594 sv_setpv(PL_linestr,incl_perldb());
2595 if (SvCUR(PL_linestr))
2596 sv_catpvn(PL_linestr,";", 1);
2598 while(AvFILLp(PL_preambleav) >= 0) {
2599 SV *tmpsv = av_shift(PL_preambleav);
2600 sv_catsv(PL_linestr, tmpsv);
2601 sv_catpvn(PL_linestr, ";", 1);
2604 sv_free((SV*)PL_preambleav);
2605 PL_preambleav = NULL;
2607 if (PL_minus_n || PL_minus_p) {
2608 sv_catpv(PL_linestr, "LINE: while (<>) {");
2610 sv_catpv(PL_linestr,"chomp;");
2613 if ((*PL_splitstr == '/' || *PL_splitstr == '\''
2614 || *PL_splitstr == '"')
2615 && strchr(PL_splitstr + 1, *PL_splitstr))
2616 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s);", PL_splitstr);
2618 /* "q\0${splitstr}\0" is legal perl. Yes, even NUL
2619 bytes can be used as quoting characters. :-) */
2620 /* The count here deliberately includes the NUL
2621 that terminates the C string constant. This
2622 embeds the opening NUL into the string. */
2623 const char *splits = PL_splitstr;
2624 sv_catpvn(PL_linestr, "our @F=split(q", 15);
2627 if (*splits == '\\')
2628 sv_catpvn(PL_linestr, splits, 1);
2629 sv_catpvn(PL_linestr, splits, 1);
2630 } while (*splits++);
2631 /* This loop will embed the trailing NUL of
2632 PL_linestr as the last thing it does before
2634 sv_catpvn(PL_linestr, ");", 2);
2638 sv_catpv(PL_linestr,"our @F=split(' ');");
2641 sv_catpvn(PL_linestr, "\n", 1);
2642 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2643 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2644 PL_last_lop = PL_last_uni = Nullch;
2645 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2646 SV *sv = NEWSV(85,0);
2648 sv_upgrade(sv, SVt_PVMG);
2649 sv_setsv(sv,PL_linestr);
2652 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2657 bof = PL_rsfp ? TRUE : FALSE;
2658 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2661 if (PL_preprocess && !PL_in_eval)
2662 (void)PerlProc_pclose(PL_rsfp);
2663 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2664 PerlIO_clearerr(PL_rsfp);
2666 (void)PerlIO_close(PL_rsfp);
2668 PL_doextract = FALSE;
2670 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2671 sv_setpv(PL_linestr,PL_minus_p
2672 ? ";}continue{print;}" : ";}");
2673 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2674 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2675 PL_last_lop = PL_last_uni = Nullch;
2676 PL_minus_n = PL_minus_p = 0;
2679 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2680 PL_last_lop = PL_last_uni = Nullch;
2681 sv_setpvn(PL_linestr,"",0);
2682 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2684 /* If it looks like the start of a BOM or raw UTF-16,
2685 * check if it in fact is. */
2691 #ifdef PERLIO_IS_STDIO
2692 # ifdef __GNU_LIBRARY__
2693 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
2694 # define FTELL_FOR_PIPE_IS_BROKEN
2698 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
2699 # define FTELL_FOR_PIPE_IS_BROKEN
2704 #ifdef FTELL_FOR_PIPE_IS_BROKEN
2705 /* This loses the possibility to detect the bof
2706 * situation on perl -P when the libc5 is being used.
2707 * Workaround? Maybe attach some extra state to PL_rsfp?
2710 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
2712 bof = PerlIO_tell(PL_rsfp) == (Off_t)SvCUR(PL_linestr);
2715 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2716 s = swallow_bom((U8*)s);
2720 /* Incest with pod. */
2721 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2722 sv_setpvn(PL_linestr, "", 0);
2723 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2724 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2725 PL_last_lop = PL_last_uni = Nullch;
2726 PL_doextract = FALSE;
2730 } while (PL_doextract);
2731 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2732 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2733 SV *sv = NEWSV(85,0);
2735 sv_upgrade(sv, SVt_PVMG);
2736 sv_setsv(sv,PL_linestr);
2739 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2741 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2742 PL_last_lop = PL_last_uni = Nullch;
2743 if (CopLINE(PL_curcop) == 1) {
2744 while (s < PL_bufend && isSPACE(*s))
2746 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2750 if (*s == '#' && *(s+1) == '!')
2752 #ifdef ALTERNATE_SHEBANG
2754 static char const as[] = ALTERNATE_SHEBANG;
2755 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2756 d = s + (sizeof(as) - 1);
2758 #endif /* ALTERNATE_SHEBANG */
2767 while (*d && !isSPACE(*d))
2771 #ifdef ARG_ZERO_IS_SCRIPT
2772 if (ipathend > ipath) {
2774 * HP-UX (at least) sets argv[0] to the script name,
2775 * which makes $^X incorrect. And Digital UNIX and Linux,
2776 * at least, set argv[0] to the basename of the Perl
2777 * interpreter. So, having found "#!", we'll set it right.
2779 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV)); /* $^X */
2780 assert(SvPOK(x) || SvGMAGICAL(x));
2781 if (sv_eq(x, CopFILESV(PL_curcop))) {
2782 sv_setpvn(x, ipath, ipathend - ipath);
2788 const char *bstart = SvPV(CopFILESV(PL_curcop),blen);
2789 const char *lstart = SvPV(x,llen);
2791 bstart += blen - llen;
2792 if (strnEQ(bstart, lstart, llen) && bstart[-1] == '/') {
2793 sv_setpvn(x, ipath, ipathend - ipath);
2798 TAINT_NOT; /* $^X is always tainted, but that's OK */
2800 #endif /* ARG_ZERO_IS_SCRIPT */
2805 d = instr(s,"perl -");
2807 d = instr(s,"perl");
2809 /* avoid getting into infinite loops when shebang
2810 * line contains "Perl" rather than "perl" */
2812 for (d = ipathend-4; d >= ipath; --d) {
2813 if ((*d == 'p' || *d == 'P')
2814 && !ibcmp(d, "perl", 4))
2824 #ifdef ALTERNATE_SHEBANG
2826 * If the ALTERNATE_SHEBANG on this system starts with a
2827 * character that can be part of a Perl expression, then if
2828 * we see it but not "perl", we're probably looking at the
2829 * start of Perl code, not a request to hand off to some
2830 * other interpreter. Similarly, if "perl" is there, but
2831 * not in the first 'word' of the line, we assume the line
2832 * contains the start of the Perl program.
2834 if (d && *s != '#') {
2835 const char *c = ipath;
2836 while (*c && !strchr("; \t\r\n\f\v#", *c))
2839 d = Nullch; /* "perl" not in first word; ignore */
2841 *s = '#'; /* Don't try to parse shebang line */
2843 #endif /* ALTERNATE_SHEBANG */
2844 #ifndef MACOS_TRADITIONAL
2849 !instr(s,"indir") &&
2850 instr(PL_origargv[0],"perl"))
2857 while (s < PL_bufend && isSPACE(*s))
2859 if (s < PL_bufend) {
2860 Newz(899,newargv,PL_origargc+3,char*);
2862 while (s < PL_bufend && !isSPACE(*s))
2865 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2868 newargv = PL_origargv;
2871 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
2873 Perl_croak(aTHX_ "Can't exec %s", ipath);
2877 const U32 oldpdb = PL_perldb;
2878 const bool oldn = PL_minus_n;
2879 const bool oldp = PL_minus_p;
2881 while (*d && !isSPACE(*d)) d++;
2882 while (SPACE_OR_TAB(*d)) d++;
2885 const bool switches_done = PL_doswitches;
2887 if (*d == 'M' || *d == 'm' || *d == 'C') {
2889 while (*d && !isSPACE(*d)) d++;
2890 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2893 d = moreswitches(d);
2895 if (PL_doswitches && !switches_done) {
2896 int argc = PL_origargc;
2897 char **argv = PL_origargv;
2900 } while (argc && argv[0][0] == '-' && argv[0][1]);
2901 init_argv_symbols(argc,argv);
2903 if ((PERLDB_LINE && !oldpdb) ||
2904 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2905 /* if we have already added "LINE: while (<>) {",
2906 we must not do it again */
2908 sv_setpvn(PL_linestr, "", 0);
2909 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2910 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2911 PL_last_lop = PL_last_uni = Nullch;
2912 PL_preambled = FALSE;
2914 (void)gv_fetchfile(PL_origfilename);
2917 if (PL_doswitches && !switches_done) {
2918 int argc = PL_origargc;
2919 char **argv = PL_origargv;
2922 } while (argc && argv[0][0] == '-' && argv[0][1]);
2923 init_argv_symbols(argc,argv);
2929 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2931 PL_lex_state = LEX_FORMLINE;
2936 #ifdef PERL_STRICT_CR
2937 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2939 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
2941 case ' ': case '\t': case '\f': case 013:
2942 #ifdef MACOS_TRADITIONAL
2949 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2950 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2951 /* handle eval qq[#line 1 "foo"\n ...] */
2952 CopLINE_dec(PL_curcop);
2956 while (s < d && *s != '\n')
2960 else if (s > d) /* Found by Ilya: feed random input to Perl. */
2961 Perl_croak(aTHX_ "panic: input overflow");
2963 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2965 PL_lex_state = LEX_FORMLINE;
2975 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2982 while (s < PL_bufend && SPACE_OR_TAB(*s))
2985 if (strnEQ(s,"=>",2)) {
2986 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2987 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2988 "### Saw unary minus before =>, forcing word '%s'\n", s);
2990 OPERATOR('-'); /* unary minus */
2992 PL_last_uni = PL_oldbufptr;
2994 case 'r': ftst = OP_FTEREAD; break;
2995 case 'w': ftst = OP_FTEWRITE; break;
2996 case 'x': ftst = OP_FTEEXEC; break;
2997 case 'o': ftst = OP_FTEOWNED; break;
2998 case 'R': ftst = OP_FTRREAD; break;
2999 case 'W': ftst = OP_FTRWRITE; break;
3000 case 'X': ftst = OP_FTREXEC; break;
3001 case 'O': ftst = OP_FTROWNED; break;
3002 case 'e': ftst = OP_FTIS; break;
3003 case 'z': ftst = OP_FTZERO; break;
3004 case 's': ftst = OP_FTSIZE; break;
3005 case 'f': ftst = OP_FTFILE; break;
3006 case 'd': ftst = OP_FTDIR; break;
3007 case 'l': ftst = OP_FTLINK; break;
3008 case 'p': ftst = OP_FTPIPE; break;
3009 case 'S': ftst = OP_FTSOCK; break;
3010 case 'u': ftst = OP_FTSUID; break;
3011 case 'g': ftst = OP_FTSGID; break;
3012 case 'k': ftst = OP_FTSVTX; break;
3013 case 'b': ftst = OP_FTBLK; break;
3014 case 'c': ftst = OP_FTCHR; break;
3015 case 't': ftst = OP_FTTTY; break;
3016 case 'T': ftst = OP_FTTEXT; break;
3017 case 'B': ftst = OP_FTBINARY; break;
3018 case 'M': case 'A': case 'C':
3019 gv_fetchpv("\024",TRUE, SVt_PV);
3021 case 'M': ftst = OP_FTMTIME; break;
3022 case 'A': ftst = OP_FTATIME; break;
3023 case 'C': ftst = OP_FTCTIME; break;
3031 PL_last_lop_op = (OPCODE)ftst;
3032 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3033 "### Saw file test %c\n", (int)ftst);
3038 /* Assume it was a minus followed by a one-letter named
3039 * subroutine call (or a -bareword), then. */
3040 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3041 "### '-%c' looked like a file test but was not\n",
3050 if (PL_expect == XOPERATOR)
3055 else if (*s == '>') {
3058 if (isIDFIRST_lazy_if(s,UTF)) {
3059 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
3067 if (PL_expect == XOPERATOR)
3070 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
3072 OPERATOR('-'); /* unary minus */
3079 if (PL_expect == XOPERATOR)
3084 if (PL_expect == XOPERATOR)
3087 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
3093 if (PL_expect != XOPERATOR) {
3094 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3095 PL_expect = XOPERATOR;
3096 force_ident(PL_tokenbuf, '*');
3109 if (PL_expect == XOPERATOR) {
3113 PL_tokenbuf[0] = '%';
3114 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
3115 if (!PL_tokenbuf[1]) {
3118 PL_pending_ident = '%';
3137 switch (PL_expect) {
3140 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
3142 PL_bufptr = s; /* update in case we back off */
3148 PL_expect = XTERMBLOCK;
3152 while (isIDFIRST_lazy_if(s,UTF)) {
3153 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3154 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
3155 if (tmp < 0) tmp = -tmp;
3171 d = scan_str(d,TRUE,TRUE);
3173 /* MUST advance bufptr here to avoid bogus
3174 "at end of line" context messages from yyerror().
3176 PL_bufptr = s + len;
3177 yyerror("Unterminated attribute parameter in attribute list");
3180 return REPORT(0); /* EOF indicator */
3184 SV *sv = newSVpvn(s, len);
3185 sv_catsv(sv, PL_lex_stuff);
3186 attrs = append_elem(OP_LIST, attrs,
3187 newSVOP(OP_CONST, 0, sv));
3188 SvREFCNT_dec(PL_lex_stuff);
3189 PL_lex_stuff = Nullsv;
3192 if (len == 6 && strnEQ(s, "unique", len)) {
3193 if (PL_in_my == KEY_our)
3195 GvUNIQUE_on(cGVOPx_gv(yylval.opval));
3197 ; /* skip to avoid loading attributes.pm */
3200 Perl_croak(aTHX_ "The 'unique' attribute may only be applied to 'our' variables");
3203 /* NOTE: any CV attrs applied here need to be part of
3204 the CVf_BUILTIN_ATTRS define in cv.h! */
3205 else if (!PL_in_my && len == 6 && strnEQ(s, "lvalue", len))
3206 CvLVALUE_on(PL_compcv);
3207 else if (!PL_in_my && len == 6 && strnEQ(s, "locked", len))
3208 CvLOCKED_on(PL_compcv);
3209 else if (!PL_in_my && len == 6 && strnEQ(s, "method", len))
3210 CvMETHOD_on(PL_compcv);
3211 else if (!PL_in_my && len == 9 && strnEQ(s, "assertion", len))
3212 CvASSERTION_on(PL_compcv);
3213 /* After we've set the flags, it could be argued that
3214 we don't need to do the attributes.pm-based setting
3215 process, and shouldn't bother appending recognized
3216 flags. To experiment with that, uncomment the
3217 following "else". (Note that's already been
3218 uncommented. That keeps the above-applied built-in
3219 attributes from being intercepted (and possibly
3220 rejected) by a package's attribute routines, but is
3221 justified by the performance win for the common case
3222 of applying only built-in attributes.) */
3224 attrs = append_elem(OP_LIST, attrs,
3225 newSVOP(OP_CONST, 0,
3229 if (*s == ':' && s[1] != ':')
3232 break; /* require real whitespace or :'s */
3234 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
3235 if (*s != ';' && *s != '}' && *s != tmp && (tmp != '=' || *s != ')')) {
3236 const char q = ((*s == '\'') ? '"' : '\'');
3237 /* If here for an expression, and parsed no attrs, back off. */
3238 if (tmp == '=' && !attrs) {
3242 /* MUST advance bufptr here to avoid bogus "at end of line"
3243 context messages from yyerror().
3247 yyerror("Unterminated attribute list");
3249 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
3257 PL_nextval[PL_nexttoke].opval = attrs;
3265 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
3266 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
3283 if (PL_lex_brackets <= 0)
3284 yyerror("Unmatched right square bracket");
3287 if (PL_lex_state == LEX_INTERPNORMAL) {
3288 if (PL_lex_brackets == 0) {
3289 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
3290 PL_lex_state = LEX_INTERPEND;
3297 if (PL_lex_brackets > 100) {
3298 Renew(PL_lex_brackstack, PL_lex_brackets + 10, char);
3300 switch (PL_expect) {
3302 if (PL_lex_formbrack) {
3306 if (PL_oldoldbufptr == PL_last_lop)
3307 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3309 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3310 OPERATOR(HASHBRACK);
3312 while (s < PL_bufend && SPACE_OR_TAB(*s))
3315 PL_tokenbuf[0] = '\0';
3316 if (d < PL_bufend && *d == '-') {
3317 PL_tokenbuf[0] = '-';
3319 while (d < PL_bufend && SPACE_OR_TAB(*d))
3322 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3323 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3325 while (d < PL_bufend && SPACE_OR_TAB(*d))
3328 const char minus = (PL_tokenbuf[0] == '-');
3329 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3337 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3342 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3347 if (PL_oldoldbufptr == PL_last_lop)
3348 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3350 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3353 if (PL_expect == XREF && PL_lex_state == LEX_INTERPNORMAL) {
3355 /* This hack is to get the ${} in the message. */
3357 yyerror("syntax error");
3360 OPERATOR(HASHBRACK);
3362 /* This hack serves to disambiguate a pair of curlies
3363 * as being a block or an anon hash. Normally, expectation
3364 * determines that, but in cases where we're not in a
3365 * position to expect anything in particular (like inside
3366 * eval"") we have to resolve the ambiguity. This code
3367 * covers the case where the first term in the curlies is a
3368 * quoted string. Most other cases need to be explicitly
3369 * disambiguated by prepending a `+' before the opening
3370 * curly in order to force resolution as an anon hash.
3372 * XXX should probably propagate the outer expectation
3373 * into eval"" to rely less on this hack, but that could
3374 * potentially break current behavior of eval"".
3378 if (*s == '\'' || *s == '"' || *s == '`') {
3379 /* common case: get past first string, handling escapes */
3380 for (t++; t < PL_bufend && *t != *s;)
3381 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3385 else if (*s == 'q') {
3388 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3391 /* skip q//-like construct */
3393 char open, close, term;
3396 while (t < PL_bufend && isSPACE(*t))
3398 /* check for q => */
3399 if (t+1 < PL_bufend && t[0] == '=' && t[1] == '>') {
3400 OPERATOR(HASHBRACK);
3404 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3408 for (t++; t < PL_bufend; t++) {
3409 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3411 else if (*t == open)
3415 for (t++; t < PL_bufend; t++) {
3416 if (*t == '\\' && t+1 < PL_bufend)
3418 else if (*t == close && --brackets <= 0)
3420 else if (*t == open)
3427 /* skip plain q word */
3428 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3431 else if (isALNUM_lazy_if(t,UTF)) {
3433 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3436 while (t < PL_bufend && isSPACE(*t))
3438 /* if comma follows first term, call it an anon hash */
3439 /* XXX it could be a comma expression with loop modifiers */
3440 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3441 || (*t == '=' && t[1] == '>')))
3442 OPERATOR(HASHBRACK);
3443 if (PL_expect == XREF)
3446 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3452 yylval.ival = CopLINE(PL_curcop);
3453 if (isSPACE(*s) || *s == '#')
3454 PL_copline = NOLINE; /* invalidate current command line number */
3459 if (PL_lex_brackets <= 0)
3460 yyerror("Unmatched right curly bracket");
3462 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3463 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3464 PL_lex_formbrack = 0;
3465 if (PL_lex_state == LEX_INTERPNORMAL) {
3466 if (PL_lex_brackets == 0) {
3467 if (PL_expect & XFAKEBRACK) {
3468 PL_expect &= XENUMMASK;
3469 PL_lex_state = LEX_INTERPEND;
3471 return yylex(); /* ignore fake brackets */
3473 if (*s == '-' && s[1] == '>')
3474 PL_lex_state = LEX_INTERPENDMAYBE;
3475 else if (*s != '[' && *s != '{')
3476 PL_lex_state = LEX_INTERPEND;
3479 if (PL_expect & XFAKEBRACK) {
3480 PL_expect &= XENUMMASK;
3482 return yylex(); /* ignore fake brackets */
3492 if (PL_expect == XOPERATOR) {
3493 if (ckWARN(WARN_SEMICOLON)
3494 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3496 CopLINE_dec(PL_curcop);
3497 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
3498 CopLINE_inc(PL_curcop);
3503 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3505 PL_expect = XOPERATOR;
3506 force_ident(PL_tokenbuf, '&');
3510 yylval.ival = (OPpENTERSUB_AMPER<<8);
3529 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3530 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Reversed %c= operator",(int)tmp);
3532 if (PL_expect == XSTATE && isALPHA(tmp) &&
3533 (s == PL_linestart+1 || s[-2] == '\n') )
3535 if (PL_in_eval && !PL_rsfp) {
3540 if (strnEQ(s,"=cut",4)) {
3554 PL_doextract = TRUE;
3557 if (PL_lex_brackets < PL_lex_formbrack) {
3559 #ifdef PERL_STRICT_CR
3560 for (t = s; SPACE_OR_TAB(*t); t++) ;
3562 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3564 if (*t == '\n' || *t == '#') {
3576 /* was this !=~ where !~ was meant?
3577 * warn on m:!=~\s+([/?]|[msy]\W|tr\W): */
3579 if (*s == '~' && ckWARN(WARN_SYNTAX)) {
3580 const char *t = s+1;
3582 while (t < PL_bufend && isSPACE(*t))
3585 if (*t == '/' || *t == '?' ||
3586 ((*t == 'm' || *t == 's' || *t == 'y') && !isALNUM(t[1])) ||
3587 (*t == 't' && t[1] == 'r' && !isALNUM(t[2])))
3588 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3589 "!=~ should be !~");
3598 if (PL_expect != XOPERATOR) {
3599 if (s[1] != '<' && !strchr(s,'>'))
3602 s = scan_heredoc(s);
3604 s = scan_inputsymbol(s);
3605 TERM(sublex_start());
3610 SHop(OP_LEFT_SHIFT);
3624 SHop(OP_RIGHT_SHIFT);
3633 if (PL_expect == XOPERATOR) {
3634 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3637 return REPORT(','); /* grandfather non-comma-format format */
3641 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3642 PL_tokenbuf[0] = '@';
3643 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3644 sizeof PL_tokenbuf - 1, FALSE);
3645 if (PL_expect == XOPERATOR)
3646 no_op("Array length", s);
3647 if (!PL_tokenbuf[1])
3649 PL_expect = XOPERATOR;
3650 PL_pending_ident = '#';
3654 PL_tokenbuf[0] = '$';
3655 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3656 sizeof PL_tokenbuf - 1, FALSE);
3657 if (PL_expect == XOPERATOR)
3659 if (!PL_tokenbuf[1]) {
3661 yyerror("Final $ should be \\$ or $name");
3665 /* This kludge not intended to be bulletproof. */
3666 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3667 yylval.opval = newSVOP(OP_CONST, 0,
3668 newSViv(PL_compiling.cop_arybase));
3669 yylval.opval->op_private = OPpCONST_ARYBASE;
3675 if (PL_lex_state == LEX_NORMAL)
3678 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3681 PL_tokenbuf[0] = '@';
3682 if (ckWARN(WARN_SYNTAX)) {
3684 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3687 PL_bufptr = skipspace(PL_bufptr);
3688 while (t < PL_bufend && *t != ']')
3690 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3691 "Multidimensional syntax %.*s not supported",
3692 (t - PL_bufptr) + 1, PL_bufptr);
3696 else if (*s == '{') {
3697 PL_tokenbuf[0] = '%';
3698 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3699 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3701 char tmpbuf[sizeof PL_tokenbuf];
3702 for (t++; isSPACE(*t); t++) ;
3703 if (isIDFIRST_lazy_if(t,UTF)) {
3705 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3706 for (; isSPACE(*t); t++) ;
3707 if (*t == ';' && get_cv(tmpbuf, FALSE))
3708 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3709 "You need to quote \"%s\"", tmpbuf);
3715 PL_expect = XOPERATOR;
3716 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3717 const bool islop = (PL_last_lop == PL_oldoldbufptr);
3718 if (!islop || PL_last_lop_op == OP_GREPSTART)
3719 PL_expect = XOPERATOR;
3720 else if (strchr("$@\"'`q", *s))
3721 PL_expect = XTERM; /* e.g. print $fh "foo" */
3722 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3723 PL_expect = XTERM; /* e.g. print $fh &sub */
3724 else if (isIDFIRST_lazy_if(s,UTF)) {
3725 char tmpbuf[sizeof PL_tokenbuf];
3726 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3727 if ((tmp = keyword(tmpbuf, len))) {
3728 /* binary operators exclude handle interpretations */
3740 PL_expect = XTERM; /* e.g. print $fh length() */
3745 PL_expect = XTERM; /* e.g. print $fh subr() */
3748 else if (isDIGIT(*s))
3749 PL_expect = XTERM; /* e.g. print $fh 3 */
3750 else if (*s == '.' && isDIGIT(s[1]))
3751 PL_expect = XTERM; /* e.g. print $fh .3 */
3752 else if ((*s == '?' || *s == '-' || *s == '+')
3753 && !isSPACE(s[1]) && s[1] != '=')
3754 PL_expect = XTERM; /* e.g. print $fh -1 */
3755 else if (*s == '/' && !isSPACE(s[1]) && s[1] != '=' && s[1] != '/')
3756 PL_expect = XTERM; /* e.g. print $fh /.../
3757 XXX except DORDOR operator */
3758 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3759 PL_expect = XTERM; /* print $fh <<"EOF" */
3761 PL_pending_ident = '$';
3765 if (PL_expect == XOPERATOR)
3767 PL_tokenbuf[0] = '@';
3768 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3769 if (!PL_tokenbuf[1]) {
3772 if (PL_lex_state == LEX_NORMAL)
3774 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3776 PL_tokenbuf[0] = '%';
3778 /* Warn about @ where they meant $. */
3779 if (ckWARN(WARN_SYNTAX)) {
3780 if (*s == '[' || *s == '{') {
3781 const char *t = s + 1;
3782 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3784 if (*t == '}' || *t == ']') {
3786 PL_bufptr = skipspace(PL_bufptr);
3787 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3788 "Scalar value %.*s better written as $%.*s",
3789 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3794 PL_pending_ident = '@';
3797 case '/': /* may be division, defined-or, or pattern */
3798 if (PL_expect == XTERMORDORDOR && s[1] == '/') {
3802 case '?': /* may either be conditional or pattern */
3803 if(PL_expect == XOPERATOR) {
3811 /* A // operator. */
3821 /* Disable warning on "study /blah/" */
3822 if (PL_oldoldbufptr == PL_last_uni
3823 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3824 || memNE(PL_last_uni, "study", 5)
3825 || isALNUM_lazy_if(PL_last_uni+5,UTF)
3828 s = scan_pat(s,OP_MATCH);
3829 TERM(sublex_start());
3833 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3834 #ifdef PERL_STRICT_CR
3837 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3839 && (s == PL_linestart || s[-1] == '\n') )
3841 PL_lex_formbrack = 0;
3845 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3851 yylval.ival = OPf_SPECIAL;
3857 if (PL_expect != XOPERATOR)
3862 case '0': case '1': case '2': case '3': case '4':
3863 case '5': case '6': case '7': case '8': case '9':
3864 s = scan_num(s, &yylval);
3865 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3866 "### Saw number in '%s'\n", s);
3868 if (PL_expect == XOPERATOR)
3873 s = scan_str(s,FALSE,FALSE);
3874 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3875 "### Saw string before '%s'\n", s);
3877 if (PL_expect == XOPERATOR) {
3878 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3881 return REPORT(','); /* grandfather non-comma-format format */
3887 missingterm((char*)0);
3888 yylval.ival = OP_CONST;
3889 TERM(sublex_start());
3892 s = scan_str(s,FALSE,FALSE);
3893 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3894 "### Saw string before '%s'\n", s);
3896 if (PL_expect == XOPERATOR) {
3897 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3900 return REPORT(','); /* grandfather non-comma-format format */
3906 missingterm((char*)0);
3907 yylval.ival = OP_CONST;
3908 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3909 if (*d == '$' || *d == '@' || *d == '\\' || !UTF8_IS_INVARIANT((U8)*d)) {
3910 yylval.ival = OP_STRINGIFY;
3914 TERM(sublex_start());
3917 s = scan_str(s,FALSE,FALSE);
3918 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3919 "### Saw backtick string before '%s'\n", s);
3921 if (PL_expect == XOPERATOR)
3922 no_op("Backticks",s);
3924 missingterm((char*)0);
3925 yylval.ival = OP_BACKTICK;
3927 TERM(sublex_start());
3931 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3932 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),"Can't use \\%c to mean $%c in expression",
3934 if (PL_expect == XOPERATOR)
3935 no_op("Backslash",s);
3939 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3940 char *start = s + 2;
3941 while (isDIGIT(*start) || *start == '_')
3943 if (*start == '.' && isDIGIT(start[1])) {
3944 s = scan_num(s, &yylval);
3947 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3948 else if (!isALPHA(*start) && (PL_expect == XTERM
3949 || PL_expect == XREF || PL_expect == XSTATE
3950 || PL_expect == XTERMORDORDOR)) {
3951 const char c = *start;
3954 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3957 s = scan_num(s, &yylval);
3964 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
4004 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4006 /* Some keywords can be followed by any delimiter, including ':' */
4007 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
4008 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
4009 (PL_tokenbuf[0] == 'q' &&
4010 strchr("qwxr", PL_tokenbuf[1])))));
4012 /* x::* is just a word, unless x is "CORE" */
4013 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
4017 while (d < PL_bufend && isSPACE(*d))
4018 d++; /* no comments skipped here, or s### is misparsed */
4020 /* Is this a label? */
4021 if (!tmp && PL_expect == XSTATE
4022 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
4024 yylval.pval = savepv(PL_tokenbuf);
4029 /* Check for keywords */
4030 tmp = keyword(PL_tokenbuf, len);
4032 /* Is this a word before a => operator? */
4033 if (*d == '=' && d[1] == '>') {
4036 = (OP*)newSVOP(OP_CONST, 0,
4037 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
4038 yylval.opval->op_private = OPpCONST_BARE;
4042 if (tmp < 0) { /* second-class keyword? */
4043 GV *ogv = Nullgv; /* override (winner) */
4044 GV *hgv = Nullgv; /* hidden (loser) */
4045 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
4047 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
4050 if (GvIMPORTED_CV(gv))
4052 else if (! CvMETHOD(cv))
4056 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
4057 (gv = *gvp) != (GV*)&PL_sv_undef &&
4058 GvCVu(gv) && GvIMPORTED_CV(gv))
4065 tmp = 0; /* overridden by import or by GLOBAL */
4068 && -tmp==KEY_lock /* XXX generalizable kludge */
4070 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
4072 tmp = 0; /* any sub overrides "weak" keyword */
4077 && PL_expect != XOPERATOR
4078 && PL_expect != XTERMORDORDOR)
4080 /* any sub overrides the "err" keyword, except when really an
4081 * operator is expected */
4084 else { /* no override */
4086 if (tmp == KEY_dump && ckWARN(WARN_MISC)) {
4087 Perl_warner(aTHX_ packWARN(WARN_MISC),
4088 "dump() better written as CORE::dump()");
4092 if (ckWARN(WARN_AMBIGUOUS) && hgv
4093 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
4094 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4095 "Ambiguous call resolved as CORE::%s(), %s",
4096 GvENAME(hgv), "qualify as such or use &");
4103 default: /* not a keyword */
4107 const char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
4109 /* Get the rest if it looks like a package qualifier */
4111 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
4113 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
4116 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
4117 *s == '\'' ? "'" : "::");
4122 if (PL_expect == XOPERATOR) {
4123 if (PL_bufptr == PL_linestart) {
4124 CopLINE_dec(PL_curcop);
4125 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
4126 CopLINE_inc(PL_curcop);
4129 no_op("Bareword",s);
4132 /* Look for a subroutine with this name in current package,
4133 unless name is "Foo::", in which case Foo is a bearword
4134 (and a package name). */
4137 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
4139 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
4140 Perl_warner(aTHX_ packWARN(WARN_BAREWORD),
4141 "Bareword \"%s\" refers to nonexistent package",
4144 PL_tokenbuf[len] = '\0';
4151 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
4154 /* if we saw a global override before, get the right name */
4157 sv = newSVpvn("CORE::GLOBAL::",14);
4158 sv_catpv(sv,PL_tokenbuf);
4161 /* If len is 0, newSVpv does strlen(), which is correct.
4162 If len is non-zero, then it will be the true length,
4163 and so the scalar will be created correctly. */
4164 sv = newSVpv(PL_tokenbuf,len);
4167 /* Presume this is going to be a bareword of some sort. */
4170 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
4171 yylval.opval->op_private = OPpCONST_BARE;
4172 /* UTF-8 package name? */
4173 if (UTF && !IN_BYTES &&
4174 is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
4177 /* And if "Foo::", then that's what it certainly is. */
4182 /* See if it's the indirect object for a list operator. */
4184 if (PL_oldoldbufptr &&
4185 PL_oldoldbufptr < PL_bufptr &&
4186 (PL_oldoldbufptr == PL_last_lop
4187 || PL_oldoldbufptr == PL_last_uni) &&
4188 /* NO SKIPSPACE BEFORE HERE! */
4189 (PL_expect == XREF ||
4190 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
4192 bool immediate_paren = *s == '(';
4194 /* (Now we can afford to cross potential line boundary.) */
4197 /* Two barewords in a row may indicate method call. */
4199 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
4202 /* If not a declared subroutine, it's an indirect object. */
4203 /* (But it's an indir obj regardless for sort.) */
4205 if ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
4206 ((!gv || !GvCVu(gv)) &&
4207 (PL_last_lop_op != OP_MAPSTART &&
4208 PL_last_lop_op != OP_GREPSTART))))
4210 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
4215 PL_expect = XOPERATOR;
4218 /* Is this a word before a => operator? */
4219 if (*s == '=' && s[1] == '>' && !pkgname) {
4221 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
4222 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
4223 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
4227 /* If followed by a paren, it's certainly a subroutine. */
4230 if (gv && GvCVu(gv)) {
4231 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
4232 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
4237 PL_nextval[PL_nexttoke].opval = yylval.opval;
4238 PL_expect = XOPERATOR;
4244 /* If followed by var or block, call it a method (unless sub) */
4246 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
4247 PL_last_lop = PL_oldbufptr;
4248 PL_last_lop_op = OP_METHOD;
4252 /* If followed by a bareword, see if it looks like indir obj. */
4255 && (isIDFIRST_lazy_if(s,UTF) || *s == '$')
4256 && (tmp = intuit_method(s,gv)))
4259 /* Not a method, so call it a subroutine (if defined) */
4261 if (gv && GvCVu(gv)) {
4263 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
4264 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4265 "Ambiguous use of -%s resolved as -&%s()",
4266 PL_tokenbuf, PL_tokenbuf);
4267 /* Check for a constant sub */
4269 if ((sv = cv_const_sv(cv))) {
4271 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
4272 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
4273 yylval.opval->op_private = 0;
4277 /* Resolve to GV now. */
4278 op_free(yylval.opval);
4279 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
4280 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
4281 PL_last_lop = PL_oldbufptr;
4282 PL_last_lop_op = OP_ENTERSUB;
4283 /* Is there a prototype? */
4286 char *proto = SvPV((SV*)cv, len);
4289 if (*proto == '$' && proto[1] == '\0')
4291 while (*proto == ';')
4293 if (*proto == '&' && *s == '{') {
4294 sv_setpv(PL_subname, PL_curstash ?
4295 "__ANON__" : "__ANON__::__ANON__");
4299 PL_nextval[PL_nexttoke].opval = yylval.opval;
4305 /* Call it a bare word */
4307 if (PL_hints & HINT_STRICT_SUBS)
4308 yylval.opval->op_private |= OPpCONST_STRICT;
4311 if (ckWARN(WARN_RESERVED)) {
4312 if (lastchar != '-') {
4313 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
4314 if (!*d && !gv_stashpv(PL_tokenbuf,FALSE))
4315 Perl_warner(aTHX_ packWARN(WARN_RESERVED), PL_warn_reserved,
4322 if ((lastchar == '*' || lastchar == '%' || lastchar == '&')
4323 && ckWARN_d(WARN_AMBIGUOUS)) {
4324 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4325 "Operator or semicolon missing before %c%s",
4326 lastchar, PL_tokenbuf);
4327 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4328 "Ambiguous use of %c resolved as operator %c",
4329 lastchar, lastchar);
4335 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4336 newSVpv(CopFILE(PL_curcop),0));
4340 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4341 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
4344 case KEY___PACKAGE__:
4345 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4347 ? newSVpv(HvNAME(PL_curstash), 0)
4356 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
4357 const char *pname = "main";
4358 if (PL_tokenbuf[2] == 'D')
4359 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
4360 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
4363 GvIOp(gv) = newIO();
4364 IoIFP(GvIOp(gv)) = PL_rsfp;
4365 #if defined(HAS_FCNTL) && defined(F_SETFD)
4367 const int fd = PerlIO_fileno(PL_rsfp);
4368 fcntl(fd,F_SETFD,fd >= 3);
4371 /* Mark this internal pseudo-handle as clean */
4372 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
4374 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
4375 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
4376 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
4378 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
4379 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
4380 /* if the script was opened in binmode, we need to revert
4381 * it to text mode for compatibility; but only iff it has CRs
4382 * XXX this is a questionable hack at best. */
4383 if (PL_bufend-PL_bufptr > 2
4384 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
4387 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
4388 loc = PerlIO_tell(PL_rsfp);
4389 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4392 if (PerlLIO_setmode(PL_rsfp, O_TEXT) != -1) {
4394 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4395 #endif /* NETWARE */
4396 #ifdef PERLIO_IS_STDIO /* really? */
4397 # if defined(__BORLANDC__)
4398 /* XXX see note in do_binmode() */
4399 ((FILE*)PL_rsfp)->flags &= ~_F_BIN;
4403 PerlIO_seek(PL_rsfp, loc, 0);
4407 #ifdef PERLIO_LAYERS
4410 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
4411 else if (PL_encoding) {
4418 XPUSHs(PL_encoding);
4420 call_method("name", G_SCALAR);
4424 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL,
4425 Perl_form(aTHX_ ":encoding(%"SVf")",
4443 if (PL_expect == XSTATE) {
4450 if (*s == ':' && s[1] == ':') {
4453 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4454 if (!(tmp = keyword(PL_tokenbuf, len)))
4455 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4469 LOP(OP_ACCEPT,XTERM);
4475 LOP(OP_ATAN2,XTERM);
4481 LOP(OP_BINMODE,XTERM);
4484 LOP(OP_BLESS,XTERM);
4493 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4510 if (!PL_cryptseen) {
4511 PL_cryptseen = TRUE;
4515 LOP(OP_CRYPT,XTERM);
4518 LOP(OP_CHMOD,XTERM);
4521 LOP(OP_CHOWN,XTERM);
4524 LOP(OP_CONNECT,XTERM);
4540 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4544 PL_hints |= HINT_BLOCK_SCOPE;
4554 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4555 LOP(OP_DBMOPEN,XTERM);
4561 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4568 yylval.ival = CopLINE(PL_curcop);
4582 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4583 UNIBRACK(OP_ENTEREVAL);
4601 case KEY_endhostent:
4607 case KEY_endservent:
4610 case KEY_endprotoent:
4621 yylval.ival = CopLINE(PL_curcop);
4623 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4625 if ((PL_bufend - p) >= 3 &&
4626 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4628 else if ((PL_bufend - p) >= 4 &&
4629 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4632 if (isIDFIRST_lazy_if(p,UTF)) {
4633 p = scan_ident(p, PL_bufend,
4634 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4638 Perl_croak(aTHX_ "Missing $ on loop variable");
4643 LOP(OP_FORMLINE,XTERM);
4649 LOP(OP_FCNTL,XTERM);
4655 LOP(OP_FLOCK,XTERM);
4664 LOP(OP_GREPSTART, XREF);
4667 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4682 case KEY_getpriority:
4683 LOP(OP_GETPRIORITY,XTERM);
4685 case KEY_getprotobyname:
4688 case KEY_getprotobynumber:
4689 LOP(OP_GPBYNUMBER,XTERM);
4691 case KEY_getprotoent:
4703 case KEY_getpeername:
4704 UNI(OP_GETPEERNAME);
4706 case KEY_gethostbyname:
4709 case KEY_gethostbyaddr:
4710 LOP(OP_GHBYADDR,XTERM);
4712 case KEY_gethostent:
4715 case KEY_getnetbyname:
4718 case KEY_getnetbyaddr:
4719 LOP(OP_GNBYADDR,XTERM);
4724 case KEY_getservbyname:
4725 LOP(OP_GSBYNAME,XTERM);
4727 case KEY_getservbyport:
4728 LOP(OP_GSBYPORT,XTERM);
4730 case KEY_getservent:
4733 case KEY_getsockname:
4734 UNI(OP_GETSOCKNAME);
4736 case KEY_getsockopt:
4737 LOP(OP_GSOCKOPT,XTERM);
4759 yylval.ival = CopLINE(PL_curcop);
4763 LOP(OP_INDEX,XTERM);
4769 LOP(OP_IOCTL,XTERM);
4781 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4813 LOP(OP_LISTEN,XTERM);
4822 s = scan_pat(s,OP_MATCH);
4823 TERM(sublex_start());
4826 LOP(OP_MAPSTART, XREF);
4829 LOP(OP_MKDIR,XTERM);
4832 LOP(OP_MSGCTL,XTERM);
4835 LOP(OP_MSGGET,XTERM);
4838 LOP(OP_MSGRCV,XTERM);
4841 LOP(OP_MSGSND,XTERM);
4847 if (isIDFIRST_lazy_if(s,UTF)) {
4848 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4849 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4851 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
4852 if (!PL_in_my_stash) {
4855 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4863 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4870 if (PL_expect != XSTATE)
4871 yyerror("\"no\" not allowed in expression");
4872 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4873 s = force_version(s, FALSE);
4878 if (*s == '(' || (s = skipspace(s), *s == '('))
4885 if (isIDFIRST_lazy_if(s,UTF)) {
4887 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4888 for (t=d; *t && isSPACE(*t); t++) ;
4889 if ( *t && strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE)
4891 && !(t[0] == '=' && t[1] == '>')
4893 Perl_warner(aTHX_ packWARN(WARN_PRECEDENCE),
4894 "Precedence problem: open %.*s should be open(%.*s)",
4895 d - s, s, d - s, s);
4901 yylval.ival = OP_OR;
4911 LOP(OP_OPEN_DIR,XTERM);
4914 checkcomma(s,PL_tokenbuf,"filehandle");
4918 checkcomma(s,PL_tokenbuf,"filehandle");
4937 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4941 LOP(OP_PIPE_OP,XTERM);
4944 s = scan_str(s,FALSE,FALSE);
4946 missingterm((char*)0);
4947 yylval.ival = OP_CONST;
4948 TERM(sublex_start());
4954 s = scan_str(s,FALSE,FALSE);
4956 missingterm((char*)0);
4958 if (SvCUR(PL_lex_stuff)) {
4961 d = SvPV_force(PL_lex_stuff, len);
4964 for (; isSPACE(*d) && len; --len, ++d) ;
4967 if (!warned && ckWARN(WARN_QW)) {
4968 for (; !isSPACE(*d) && len; --len, ++d) {
4970 Perl_warner(aTHX_ packWARN(WARN_QW),
4971 "Possible attempt to separate words with commas");
4974 else if (*d == '#') {
4975 Perl_warner(aTHX_ packWARN(WARN_QW),
4976 "Possible attempt to put comments in qw() list");
4982 for (; !isSPACE(*d) && len; --len, ++d) ;
4984 sv = newSVpvn(b, d-b);
4985 if (DO_UTF8(PL_lex_stuff))
4987 words = append_elem(OP_LIST, words,
4988 newSVOP(OP_CONST, 0, tokeq(sv)));
4992 PL_nextval[PL_nexttoke].opval = words;
4997 SvREFCNT_dec(PL_lex_stuff);
4998 PL_lex_stuff = Nullsv;
5004 s = scan_str(s,FALSE,FALSE);
5006 missingterm((char*)0);
5007 yylval.ival = OP_STRINGIFY;
5008 if (SvIVX(PL_lex_stuff) == '\'')
5009 SvIV_set(PL_lex_stuff, 0); /* qq'$foo' should intepolate */
5010 TERM(sublex_start());
5013 s = scan_pat(s,OP_QR);
5014 TERM(sublex_start());
5017 s = scan_str(s,FALSE,FALSE);
5019 missingterm((char*)0);
5020 yylval.ival = OP_BACKTICK;
5022 TERM(sublex_start());
5030 s = force_version(s, FALSE);
5032 else if (*s != 'v' || !isDIGIT(s[1])
5033 || (s = force_version(s, TRUE), *s == 'v'))
5035 *PL_tokenbuf = '\0';
5036 s = force_word(s,WORD,TRUE,TRUE,FALSE);
5037 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
5038 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
5040 yyerror("<> should be quotes");
5048 s = force_word(s,WORD,TRUE,FALSE,FALSE);
5052 LOP(OP_RENAME,XTERM);
5061 LOP(OP_RINDEX,XTERM);
5071 UNIDOR(OP_READLINE);
5084 LOP(OP_REVERSE,XTERM);
5087 UNIDOR(OP_READLINK);
5095 TERM(sublex_start());
5097 TOKEN(1); /* force error */
5106 LOP(OP_SELECT,XTERM);
5112 LOP(OP_SEMCTL,XTERM);
5115 LOP(OP_SEMGET,XTERM);
5118 LOP(OP_SEMOP,XTERM);
5124 LOP(OP_SETPGRP,XTERM);
5126 case KEY_setpriority:
5127 LOP(OP_SETPRIORITY,XTERM);
5129 case KEY_sethostent:
5135 case KEY_setservent:
5138 case KEY_setprotoent:
5148 LOP(OP_SEEKDIR,XTERM);
5150 case KEY_setsockopt:
5151 LOP(OP_SSOCKOPT,XTERM);
5157 LOP(OP_SHMCTL,XTERM);
5160 LOP(OP_SHMGET,XTERM);
5163 LOP(OP_SHMREAD,XTERM);
5166 LOP(OP_SHMWRITE,XTERM);
5169 LOP(OP_SHUTDOWN,XTERM);
5178 LOP(OP_SOCKET,XTERM);
5180 case KEY_socketpair:
5181 LOP(OP_SOCKPAIR,XTERM);
5184 checkcomma(s,PL_tokenbuf,"subroutine name");
5186 if (*s == ';' || *s == ')') /* probably a close */
5187 Perl_croak(aTHX_ "sort is now a reserved word");
5189 s = force_word(s,WORD,TRUE,TRUE,FALSE);
5193 LOP(OP_SPLIT,XTERM);
5196 LOP(OP_SPRINTF,XTERM);
5199 LOP(OP_SPLICE,XTERM);
5214 LOP(OP_SUBSTR,XTERM);
5220 char tmpbuf[sizeof PL_tokenbuf];
5221 SSize_t tboffset = 0;
5222 expectation attrful;
5223 bool have_name, have_proto, bad_proto;
5224 const int key = tmp;
5228 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
5229 (*s == ':' && s[1] == ':'))
5232 attrful = XATTRBLOCK;
5233 /* remember buffer pos'n for later force_word */
5234 tboffset = s - PL_oldbufptr;
5235 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
5236 if (strchr(tmpbuf, ':'))
5237 sv_setpv(PL_subname, tmpbuf);
5239 sv_setsv(PL_subname,PL_curstname);
5240 sv_catpvn(PL_subname,"::",2);
5241 sv_catpvn(PL_subname,tmpbuf,len);
5248 Perl_croak(aTHX_ "Missing name in \"my sub\"");
5249 PL_expect = XTERMBLOCK;
5250 attrful = XATTRTERM;
5251 sv_setpvn(PL_subname,"?",1);
5255 if (key == KEY_format) {
5257 PL_lex_formbrack = PL_lex_brackets + 1;
5259 (void) force_word(PL_oldbufptr + tboffset, WORD,
5264 /* Look for a prototype */
5268 s = scan_str(s,FALSE,FALSE);
5270 Perl_croak(aTHX_ "Prototype not terminated");
5271 /* strip spaces and check for bad characters */
5272 d = SvPVX(PL_lex_stuff);
5275 for (p = d; *p; ++p) {
5278 if (!strchr("$@%*;[]&\\", *p))
5283 if (bad_proto && ckWARN(WARN_SYNTAX))
5284 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5285 "Illegal character in prototype for %"SVf" : %s",
5287 SvCUR_set(PL_lex_stuff, tmp);
5295 if (*s == ':' && s[1] != ':')
5296 PL_expect = attrful;
5297 else if (*s != '{' && key == KEY_sub) {
5299 Perl_croak(aTHX_ "Illegal declaration of anonymous subroutine");
5301 Perl_croak(aTHX_ "Illegal declaration of subroutine %"SVf, PL_subname);
5305 PL_nextval[PL_nexttoke].opval =
5306 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
5307 PL_lex_stuff = Nullsv;
5311 sv_setpv(PL_subname,
5312 PL_curstash ? "__ANON__" : "__ANON__::__ANON__");
5315 (void) force_word(PL_oldbufptr + tboffset, WORD,
5324 LOP(OP_SYSTEM,XREF);
5327 LOP(OP_SYMLINK,XTERM);
5330 LOP(OP_SYSCALL,XTERM);
5333 LOP(OP_SYSOPEN,XTERM);
5336 LOP(OP_SYSSEEK,XTERM);
5339 LOP(OP_SYSREAD,XTERM);
5342 LOP(OP_SYSWRITE,XTERM);
5346 TERM(sublex_start());
5367 LOP(OP_TRUNCATE,XTERM);
5379 yylval.ival = CopLINE(PL_curcop);
5383 yylval.ival = CopLINE(PL_curcop);
5387 LOP(OP_UNLINK,XTERM);
5393 LOP(OP_UNPACK,XTERM);
5396 LOP(OP_UTIME,XTERM);
5402 LOP(OP_UNSHIFT,XTERM);
5405 if (PL_expect != XSTATE)
5406 yyerror("\"use\" not allowed in expression");
5408 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
5409 s = force_version(s, TRUE);
5410 if (*s == ';' || (s = skipspace(s), *s == ';')) {
5411 PL_nextval[PL_nexttoke].opval = Nullop;
5414 else if (*s == 'v') {
5415 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5416 s = force_version(s, FALSE);
5420 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5421 s = force_version(s, FALSE);
5433 yylval.ival = CopLINE(PL_curcop);
5437 PL_hints |= HINT_BLOCK_SCOPE;
5444 LOP(OP_WAITPID,XTERM);
5453 ctl_l[0] = toCTRL('L');
5455 gv_fetchpv(ctl_l,TRUE, SVt_PV);
5458 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
5463 if (PL_expect == XOPERATOR)
5469 yylval.ival = OP_XOR;
5474 TERM(sublex_start());
5479 #pragma segment Main
5483 S_pending_ident(pTHX)
5486 register I32 tmp = 0;
5487 /* pit holds the identifier we read and pending_ident is reset */
5488 char pit = PL_pending_ident;
5489 PL_pending_ident = 0;
5491 DEBUG_T({ PerlIO_printf(Perl_debug_log,
5492 "### Tokener saw identifier '%s'\n", PL_tokenbuf); });
5494 /* if we're in a my(), we can't allow dynamics here.
5495 $foo'bar has already been turned into $foo::bar, so
5496 just check for colons.
5498 if it's a legal name, the OP is a PADANY.
5501 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
5502 if (strchr(PL_tokenbuf,':'))
5503 yyerror(Perl_form(aTHX_ "No package name allowed for "
5504 "variable %s in \"our\"",
5506 tmp = allocmy(PL_tokenbuf);
5509 if (strchr(PL_tokenbuf,':'))
5510 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
5512 yylval.opval = newOP(OP_PADANY, 0);
5513 yylval.opval->op_targ = allocmy(PL_tokenbuf);
5519 build the ops for accesses to a my() variable.
5521 Deny my($a) or my($b) in a sort block, *if* $a or $b is
5522 then used in a comparison. This catches most, but not
5523 all cases. For instance, it catches
5524 sort { my($a); $a <=> $b }
5526 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
5527 (although why you'd do that is anyone's guess).
5530 if (!strchr(PL_tokenbuf,':')) {
5532 tmp = pad_findmy(PL_tokenbuf);
5533 if (tmp != NOT_IN_PAD) {
5534 /* might be an "our" variable" */
5535 if (PAD_COMPNAME_FLAGS(tmp) & SVpad_OUR) {
5536 /* build ops for a bareword */
5537 SV *sym = newSVpv(HvNAME(PAD_COMPNAME_OURSTASH(tmp)), 0);
5538 sv_catpvn(sym, "::", 2);
5539 sv_catpv(sym, PL_tokenbuf+1);
5540 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
5541 yylval.opval->op_private = OPpCONST_ENTERED;
5544 ? (GV_ADDMULTI | GV_ADDINEVAL)
5547 ((PL_tokenbuf[0] == '$') ? SVt_PV
5548 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5553 /* if it's a sort block and they're naming $a or $b */
5554 if (PL_last_lop_op == OP_SORT &&
5555 PL_tokenbuf[0] == '$' &&
5556 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
5559 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
5560 d < PL_bufend && *d != '\n';
5563 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
5564 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
5570 yylval.opval = newOP(OP_PADANY, 0);
5571 yylval.opval->op_targ = tmp;
5577 Whine if they've said @foo in a doublequoted string,
5578 and @foo isn't a variable we can find in the symbol
5581 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
5582 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
5583 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
5584 && ckWARN(WARN_AMBIGUOUS))
5586 /* Downgraded from fatal to warning 20000522 mjd */
5587 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5588 "Possible unintended interpolation of %s in string",
5593 /* build ops for a bareword */
5594 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
5595 yylval.opval->op_private = OPpCONST_ENTERED;
5596 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
5597 ((PL_tokenbuf[0] == '$') ? SVt_PV
5598 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5604 * The following code was generated by perl_keyword.pl.
5608 Perl_keyword (pTHX_ const char *name, I32 len)
5612 case 1: /* 5 tokens of length 1 */
5644 case 2: /* 18 tokens of length 2 */
5790 case 3: /* 28 tokens of length 3 */
5794 if (name[1] == 'N' &&
5857 if (name[1] == 'i' &&
5897 if (name[1] == 'o' &&
5906 if (name[1] == 'e' &&
5915 if (name[1] == 'n' &&
5924 if (name[1] == 'o' &&
5933 if (name[1] == 'a' &&
5942 if (name[1] == 'o' &&
6004 if (name[1] == 'e' &&
6036 if (name[1] == 'i' &&
6045 if (name[1] == 's' &&
6054 if (name[1] == 'e' &&
6063 if (name[1] == 'o' &&
6075 case 4: /* 40 tokens of length 4 */
6079 if (name[1] == 'O' &&
6089 if (name[1] == 'N' &&
6099 if (name[1] == 'i' &&
6109 if (name[1] == 'h' &&
6119 if (name[1] == 'u' &&
6132 if (name[2] == 'c' &&
6141 if (name[2] == 's' &&
6150 if (name[2] == 'a' &&
6186 if (name[1] == 'o' &&
6199 if (name[2] == 't' &&
6208 if (name[2] == 'o' &&
6217 if (name[2] == 't' &&
6226 if (name[2] == 'e' &&
6239 if (name[1] == 'o' &&
6252 if (name[2] == 'y' &&
6261 if (name[2] == 'l' &&
6277 if (name[2] == 's' &&
6286 if (name[2] == 'n' &&
6295 if (name[2] == 'c' &&
6308 if (name[1] == 'e' &&
6318 if (name[1] == 'p' &&
6331 if (name[2] == 'c' &&
6340 if (name[2] == 'p' &&
6349 if (name[2] == 's' &&
6365 if (name[2] == 'n' &&
6435 if (name[2] == 'r' &&
6444 if (name[2] == 'r' &&
6453 if (name[2] == 'a' &&
6469 if (name[2] == 'l' &&
6536 case 5: /* 36 tokens of length 5 */
6540 if (name[1] == 'E' &&
6551 if (name[1] == 'H' &&
6565 if (name[2] == 'a' &&
6575 if (name[2] == 'a' &&
6589 if (name[1] == 'l' &&
6606 if (name[3] == 'i' &&
6615 if (name[3] == 'o' &&
6651 if (name[2] == 'o' &&
6661 if (name[2] == 'y' &&
6675 if (name[1] == 'l' &&
6689 if (name[2] == 'n' &&
6699 if (name[2] == 'o' &&
6716 if (name[2] == 'd' &&
6726 if (name[2] == 'c' &&
6743 if (name[2] == 'c' &&
6753 if (name[2] == 't' &&
6767 if (name[1] == 'k' &&
6778 if (name[1] == 'r' &&
6792 if (name[2] == 's' &&
6802 if (name[2] == 'd' &&
6819 if (name[2] == 'm' &&
6829 if (name[2] == 'i' &&
6839 if (name[2] == 'e' &&
6849 if (name[2] == 'l' &&
6859 if (name[2] == 'a' &&
6869 if (name[2] == 'u' &&
6883 if (name[1] == 'i' &&
6897 if (name[2] == 'a' &&
6910 if (name[3] == 'e' &&
6945 if (name[2] == 'i' &&
6962 if (name[2] == 'i' &&
6972 if (name[2] == 'i' &&
6989 case 6: /* 33 tokens of length 6 */
6993 if (name[1] == 'c' &&
7008 if (name[2] == 'l' &&
7019 if (name[2] == 'r' &&
7034 if (name[1] == 'e' &&
7049 if (name[2] == 's' &&
7054 if(ckWARN_d(WARN_SYNTAX))
7055 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "elseif should be elsif");
7061 if (name[2] == 'i' &&
7079 if (name[2] == 'l' &&
7090 if (name[2] == 'r' &&
7105 if (name[1] == 'm' &&
7120 if (name[2] == 'n' &&
7131 if (name[2] == 's' &&
7146 if (name[1] == 's' &&
7152 if (name[4] == 't' &&
7161 if (name[4] == 'e' &&
7170 if (name[4] == 'c' &&
7179 if (name[4] == 'n' &&
7195 if (name[1] == 'r' &&
7213 if (name[3] == 'a' &&
7223 if (name[3] == 'u' &&
7237 if (name[2] == 'n' &&
7255 if (name[2] == 'a' &&
7269 if (name[3] == 'e' &&
7282 if (name[4] == 't' &&
7291 if (name[4] == 'e' &&
7313 if (name[4] == 't' &&
7322 if (name[4] == 'e' &&
7338 if (name[2] == 'c' &&
7349 if (name[2] == 'l' &&
7360 if (name[2] == 'b' &&
7371 if (name[2] == 's' &&
7394 if (name[4] == 's' &&
7403 if (name[4] == 'n' &&
7416 if (name[3] == 'a' &&
7433 if (name[1] == 'a' &&
7448 case 7: /* 28 tokens of length 7 */
7452 if (name[1] == 'E' &&
7465 if (name[1] == '_' &&
7478 if (name[1] == 'i' &&
7485 return -KEY_binmode;
7491 if (name[1] == 'o' &&
7498 return -KEY_connect;
7507 if (name[2] == 'm' &&
7513 return -KEY_dbmopen;
7519 if (name[2] == 'f' &&
7535 if (name[1] == 'o' &&
7548 if (name[1] == 'e' &&
7555 if (name[5] == 'r' &&
7558 return -KEY_getpgrp;
7564 if (name[5] == 'i' &&
7567 return -KEY_getppid;
7580 if (name[1] == 'c' &&
7587 return -KEY_lcfirst;
7593 if (name[1] == 'p' &&
7600 return -KEY_opendir;
7606 if (name[1] == 'a' &&
7624 if (name[3] == 'd' &&
7629 return -KEY_readdir;
7635 if (name[3] == 'u' &&
7646 if (name[3] == 'e' &&
7651 return -KEY_reverse;
7670 if (name[3] == 'k' &&
7675 return -KEY_seekdir;
7681 if (name[3] == 'p' &&
7686 return -KEY_setpgrp;
7696 if (name[2] == 'm' &&
7702 return -KEY_shmread;
7708 if (name[2] == 'r' &&
7714 return -KEY_sprintf;
7723 if (name[3] == 'l' &&
7728 return -KEY_symlink;
7737 if (name[4] == 'a' &&
7741 return -KEY_syscall;
7747 if (name[4] == 'p' &&
7751 return -KEY_sysopen;
7757 if (name[4] == 'e' &&
7761 return -KEY_sysread;
7767 if (name[4] == 'e' &&
7771 return -KEY_sysseek;
7789 if (name[1] == 'e' &&
7796 return -KEY_telldir;
7805 if (name[2] == 'f' &&
7811 return -KEY_ucfirst;
7817 if (name[2] == 's' &&
7823 return -KEY_unshift;
7833 if (name[1] == 'a' &&
7840 return -KEY_waitpid;
7849 case 8: /* 26 tokens of length 8 */
7853 if (name[1] == 'U' &&
7861 return KEY_AUTOLOAD;
7872 if (name[3] == 'A' &&
7878 return KEY___DATA__;
7884 if (name[3] == 'I' &&
7890 return -KEY___FILE__;
7896 if (name[3] == 'I' &&
7902 return -KEY___LINE__;
7918 if (name[2] == 'o' &&
7925 return -KEY_closedir;
7931 if (name[2] == 'n' &&
7938 return -KEY_continue;
7948 if (name[1] == 'b' &&
7956 return -KEY_dbmclose;
7962 if (name[1] == 'n' &&
7968 if (name[4] == 'r' &&
7973 return -KEY_endgrent;
7979 if (name[4] == 'w' &&
7984 return -KEY_endpwent;
7997 if (name[1] == 'o' &&
8005 return -KEY_formline;
8011 if (name[1] == 'e' &&
8022 if (name[6] == 'n' &&
8025 return -KEY_getgrent;
8031 if (name[6] == 'i' &&
8034 return -KEY_getgrgid;
8040 if (name[6] == 'a' &&
8043 return -KEY_getgrnam;
8056 if (name[4] == 'o' &&
8061 return -KEY_getlogin;
8072 if (name[6] == 'n' &&
8075 return -KEY_getpwent;
8081 if (name[6] == 'a' &&
8084 return -KEY_getpwnam;
8090 if (name[6] == 'i' &&
8093 return -KEY_getpwuid;
8113 if (name[1] == 'e' &&
8120 if (name[5] == 'i' &&
8127 return -KEY_readline;
8132 return -KEY_readlink;
8143 if (name[5] == 'i' &&
8147 return -KEY_readpipe;
8168 if (name[4] == 'r' &&
8173 return -KEY_setgrent;
8179 if (name[4] == 'w' &&
8184 return -KEY_setpwent;
8200 if (name[3] == 'w' &&
8206 return -KEY_shmwrite;
8212 if (name[3] == 't' &&
8218 return -KEY_shutdown;
8228 if (name[2] == 's' &&
8235 return -KEY_syswrite;
8245 if (name[1] == 'r' &&
8253 return -KEY_truncate;
8262 case 9: /* 8 tokens of length 9 */
8266 if (name[1] == 'n' &&
8275 return -KEY_endnetent;
8281 if (name[1] == 'e' &&
8290 return -KEY_getnetent;
8296 if (name[1] == 'o' &&
8305 return -KEY_localtime;
8311 if (name[1] == 'r' &&
8320 return KEY_prototype;
8326 if (name[1] == 'u' &&
8335 return -KEY_quotemeta;
8341 if (name[1] == 'e' &&
8350 return -KEY_rewinddir;
8356 if (name[1] == 'e' &&
8365 return -KEY_setnetent;
8371 if (name[1] == 'a' &&
8380 return -KEY_wantarray;
8389 case 10: /* 9 tokens of length 10 */
8393 if (name[1] == 'n' &&
8399 if (name[4] == 'o' &&
8406 return -KEY_endhostent;
8412 if (name[4] == 'e' &&
8419 return -KEY_endservent;
8432 if (name[1] == 'e' &&
8438 if (name[4] == 'o' &&
8445 return -KEY_gethostent;
8454 if (name[5] == 'r' &&
8460 return -KEY_getservent;
8466 if (name[5] == 'c' &&
8472 return -KEY_getsockopt;
8497 if (name[4] == 'o' &&
8504 return -KEY_sethostent;
8513 if (name[5] == 'r' &&
8519 return -KEY_setservent;
8525 if (name[5] == 'c' &&
8531 return -KEY_setsockopt;
8548 if (name[2] == 'c' &&
8557 return -KEY_socketpair;
8570 case 11: /* 8 tokens of length 11 */
8574 if (name[1] == '_' &&
8585 return -KEY___PACKAGE__;
8591 if (name[1] == 'n' &&
8602 return -KEY_endprotoent;
8608 if (name[1] == 'e' &&
8617 if (name[5] == 'e' &&
8624 return -KEY_getpeername;
8633 if (name[6] == 'o' &&
8639 return -KEY_getpriority;
8645 if (name[6] == 't' &&
8651 return -KEY_getprotoent;
8665 if (name[4] == 'o' &&
8673 return -KEY_getsockname;
8686 if (name[1] == 'e' &&
8694 if (name[6] == 'o' &&
8700 return -KEY_setpriority;
8706 if (name[6] == 't' &&
8712 return -KEY_setprotoent;
8728 case 12: /* 2 tokens of length 12 */
8729 if (name[0] == 'g' &&
8741 if (name[9] == 'd' &&
8744 { /* getnetbyaddr */
8745 return -KEY_getnetbyaddr;
8751 if (name[9] == 'a' &&
8754 { /* getnetbyname */
8755 return -KEY_getnetbyname;
8767 case 13: /* 4 tokens of length 13 */
8768 if (name[0] == 'g' &&
8775 if (name[4] == 'o' &&
8784 if (name[10] == 'd' &&
8787 { /* gethostbyaddr */
8788 return -KEY_gethostbyaddr;
8794 if (name[10] == 'a' &&
8797 { /* gethostbyname */
8798 return -KEY_gethostbyname;
8811 if (name[4] == 'e' &&
8820 if (name[10] == 'a' &&
8823 { /* getservbyname */
8824 return -KEY_getservbyname;
8830 if (name[10] == 'o' &&
8833 { /* getservbyport */
8834 return -KEY_getservbyport;
8853 case 14: /* 1 tokens of length 14 */
8854 if (name[0] == 'g' &&
8868 { /* getprotobyname */
8869 return -KEY_getprotobyname;
8874 case 16: /* 1 tokens of length 16 */
8875 if (name[0] == 'g' &&
8891 { /* getprotobynumber */
8892 return -KEY_getprotobynumber;
8906 S_checkcomma(pTHX_ register char *s, const char *name, const char *what)
8910 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
8911 if (ckWARN(WARN_SYNTAX)) {
8913 for (w = s+2; *w && level; w++) {
8920 for (; *w && isSPACE(*w); w++) ;
8921 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
8922 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
8923 "%s (...) interpreted as function",name);
8926 while (s < PL_bufend && isSPACE(*s))
8930 while (s < PL_bufend && isSPACE(*s))
8932 if (isIDFIRST_lazy_if(s,UTF)) {
8934 while (isALNUM_lazy_if(s,UTF))
8936 while (s < PL_bufend && isSPACE(*s))
8940 *s = '\0'; /* XXX If we didn't do this, we could const a lot of toke.c */
8941 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
8945 Perl_croak(aTHX_ "No comma allowed after %s", what);
8950 /* Either returns sv, or mortalizes sv and returns a new SV*.
8951 Best used as sv=new_constant(..., sv, ...).
8952 If s, pv are NULL, calls subroutine with one argument,
8953 and type is used with error messages only. */
8956 S_new_constant(pTHX_ const char *s, STRLEN len, const char *key, SV *sv, SV *pv,
8960 HV *table = GvHV(PL_hintgv); /* ^H */
8964 const char *why1, *why2, *why3;
8966 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
8969 why2 = strEQ(key,"charnames")
8970 ? "(possibly a missing \"use charnames ...\")"
8972 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
8973 (type ? type: "undef"), why2);
8975 /* This is convoluted and evil ("goto considered harmful")
8976 * but I do not understand the intricacies of all the different
8977 * failure modes of %^H in here. The goal here is to make
8978 * the most probable error message user-friendly. --jhi */
8983 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
8984 (type ? type: "undef"), why1, why2, why3);
8986 yyerror(SvPVX(msg));
8990 cvp = hv_fetch(table, key, strlen(key), FALSE);
8991 if (!cvp || !SvOK(*cvp)) {
8994 why3 = "} is not defined";
8997 sv_2mortal(sv); /* Parent created it permanently */
9000 pv = sv_2mortal(newSVpvn(s, len));
9002 typesv = sv_2mortal(newSVpv(type, 0));
9004 typesv = &PL_sv_undef;
9006 PUSHSTACKi(PERLSI_OVERLOAD);
9018 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
9022 /* Check the eval first */
9023 if (!PL_in_eval && SvTRUE(ERRSV)) {
9025 sv_catpv(ERRSV, "Propagated");
9026 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
9028 res = SvREFCNT_inc(sv);
9032 (void)SvREFCNT_inc(res);
9041 why1 = "Call to &{$^H{";
9043 why3 = "}} did not return a defined value";
9051 /* Returns a NUL terminated string, with the length of the string written to
9055 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
9057 register char *d = dest;
9058 register char *e = d + destlen - 3; /* two-character token, ending NUL */
9061 Perl_croak(aTHX_ ident_too_long);
9062 if (isALNUM(*s)) /* UTF handled below */
9064 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
9069 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
9073 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
9074 char *t = s + UTF8SKIP(s);
9075 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
9077 if (d + (t - s) > e)
9078 Perl_croak(aTHX_ ident_too_long);
9079 Copy(s, d, t - s, char);
9092 S_scan_ident(pTHX_ register char *s, register const char *send, char *dest, STRLEN destlen, I32 ck_uni)
9102 e = d + destlen - 3; /* two-character token, ending NUL */
9104 while (isDIGIT(*s)) {
9106 Perl_croak(aTHX_ ident_too_long);
9113 Perl_croak(aTHX_ ident_too_long);
9114 if (isALNUM(*s)) /* UTF handled below */
9116 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
9121 else if (*s == ':' && s[1] == ':') {
9125 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
9126 char *t = s + UTF8SKIP(s);
9127 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
9129 if (d + (t - s) > e)
9130 Perl_croak(aTHX_ ident_too_long);
9131 Copy(s, d, t - s, char);
9142 if (PL_lex_state != LEX_NORMAL)
9143 PL_lex_state = LEX_INTERPENDMAYBE;
9146 if (*s == '$' && s[1] &&
9147 (isALNUM_lazy_if(s+1,UTF) || s[1] == '$' || s[1] == '{' || strnEQ(s+1,"::",2)) )
9160 if (*d == '^' && *s && isCONTROLVAR(*s)) {
9165 if (isSPACE(s[-1])) {
9167 const char ch = *s++;
9168 if (!SPACE_OR_TAB(ch)) {
9174 if (isIDFIRST_lazy_if(d,UTF)) {
9178 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
9180 while (e < send && UTF8_IS_CONTINUED(*e) && is_utf8_mark((U8*)e))
9183 Copy(s, d, e - s, char);
9188 while ((isALNUM(*s) || *s == ':') && d < e)
9191 Perl_croak(aTHX_ ident_too_long);
9194 while (s < send && SPACE_OR_TAB(*s)) s++;
9195 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
9196 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
9197 const char *brack = *s == '[' ? "[...]" : "{...}";
9198 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
9199 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
9200 funny, dest, brack, funny, dest, brack);
9203 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
9207 /* Handle extended ${^Foo} variables
9208 * 1999-02-27 mjd-perl-patch@plover.com */
9209 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
9213 while (isALNUM(*s) && d < e) {
9217 Perl_croak(aTHX_ ident_too_long);
9222 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets) {
9223 PL_lex_state = LEX_INTERPEND;
9228 if (PL_lex_state == LEX_NORMAL) {
9229 if (ckWARN(WARN_AMBIGUOUS) &&
9230 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
9232 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
9233 "Ambiguous use of %c{%s} resolved to %c%s",
9234 funny, dest, funny, dest);
9239 s = bracket; /* let the parser handle it */
9243 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
9244 PL_lex_state = LEX_INTERPEND;
9249 Perl_pmflag(pTHX_ U32* pmfl, int ch)
9254 *pmfl |= PMf_GLOBAL;
9256 *pmfl |= PMf_CONTINUE;
9260 *pmfl |= PMf_MULTILINE;
9262 *pmfl |= PMf_SINGLELINE;
9264 *pmfl |= PMf_EXTENDED;
9268 S_scan_pat(pTHX_ char *start, I32 type)
9271 char *s = scan_str(start,FALSE,FALSE);
9274 Perl_croak(aTHX_ "Search pattern not terminated");
9276 pm = (PMOP*)newPMOP(type, 0);
9277 if (PL_multi_open == '?')
9278 pm->op_pmflags |= PMf_ONCE;
9280 while (*s && strchr("iomsx", *s))
9281 pmflag(&pm->op_pmflags,*s++);
9284 while (*s && strchr("iogcmsx", *s))
9285 pmflag(&pm->op_pmflags,*s++);
9287 /* issue a warning if /c is specified,but /g is not */
9288 if (ckWARN(WARN_REGEXP) &&
9289 (pm->op_pmflags & PMf_CONTINUE) && !(pm->op_pmflags & PMf_GLOBAL))
9291 Perl_warner(aTHX_ packWARN(WARN_REGEXP), c_without_g);
9294 pm->op_pmpermflags = pm->op_pmflags;
9296 PL_lex_op = (OP*)pm;
9297 yylval.ival = OP_MATCH;
9302 S_scan_subst(pTHX_ char *start)
9310 yylval.ival = OP_NULL;
9312 s = scan_str(start,FALSE,FALSE);
9315 Perl_croak(aTHX_ "Substitution pattern not terminated");
9317 if (s[-1] == PL_multi_open)
9320 first_start = PL_multi_start;
9321 s = scan_str(s,FALSE,FALSE);
9324 SvREFCNT_dec(PL_lex_stuff);
9325 PL_lex_stuff = Nullsv;
9327 Perl_croak(aTHX_ "Substitution replacement not terminated");
9329 PL_multi_start = first_start; /* so whole substitution is taken together */
9331 pm = (PMOP*)newPMOP(OP_SUBST, 0);
9337 else if (strchr("iogcmsx", *s))
9338 pmflag(&pm->op_pmflags,*s++);
9343 /* /c is not meaningful with s/// */
9344 if (ckWARN(WARN_REGEXP) && (pm->op_pmflags & PMf_CONTINUE))
9346 Perl_warner(aTHX_ packWARN(WARN_REGEXP), c_in_subst);
9351 PL_sublex_info.super_bufptr = s;
9352 PL_sublex_info.super_bufend = PL_bufend;
9354 pm->op_pmflags |= PMf_EVAL;
9355 repl = newSVpvn("",0);
9357 sv_catpv(repl, es ? "eval " : "do ");
9358 sv_catpvn(repl, "{ ", 2);
9359 sv_catsv(repl, PL_lex_repl);
9360 sv_catpvn(repl, " };", 2);
9362 SvREFCNT_dec(PL_lex_repl);
9366 pm->op_pmpermflags = pm->op_pmflags;
9367 PL_lex_op = (OP*)pm;
9368 yylval.ival = OP_SUBST;
9373 S_scan_trans(pTHX_ char *start)
9382 yylval.ival = OP_NULL;
9384 s = scan_str(start,FALSE,FALSE);
9386 Perl_croak(aTHX_ "Transliteration pattern not terminated");
9387 if (s[-1] == PL_multi_open)
9390 s = scan_str(s,FALSE,FALSE);
9393 SvREFCNT_dec(PL_lex_stuff);
9394 PL_lex_stuff = Nullsv;
9396 Perl_croak(aTHX_ "Transliteration replacement not terminated");
9399 complement = del = squash = 0;
9403 complement = OPpTRANS_COMPLEMENT;
9406 del = OPpTRANS_DELETE;
9409 squash = OPpTRANS_SQUASH;
9418 New(803, tbl, complement&&!del?258:256, short);
9419 o = newPVOP(OP_TRANS, 0, (char*)tbl);
9420 o->op_private &= ~OPpTRANS_ALL;
9421 o->op_private |= del|squash|complement|
9422 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
9423 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
9426 yylval.ival = OP_TRANS;
9431 S_scan_heredoc(pTHX_ register char *s)
9434 I32 op_type = OP_SCALAR;
9438 const char newline[] = "\n";
9439 const char *found_newline;
9443 const int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
9447 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
9450 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
9451 if (*peek == '`' || *peek == '\'' || *peek =='"') {
9454 s = delimcpy(d, e, s, PL_bufend, term, &len);
9464 if (!isALNUM_lazy_if(s,UTF))
9465 deprecate_old("bare << to mean <<\"\"");
9466 for (; isALNUM_lazy_if(s,UTF); s++) {
9471 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
9472 Perl_croak(aTHX_ "Delimiter for here document is too long");
9475 len = d - PL_tokenbuf;
9476 #ifndef PERL_STRICT_CR
9477 d = strchr(s, '\r');
9481 while (s < PL_bufend) {
9487 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
9496 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
9500 if ( outer || !(found_newline = ninstr(s,PL_bufend,newline,newline+1)) ) {
9501 herewas = newSVpvn(s,PL_bufend-s);
9505 herewas = newSVpvn(s,found_newline-s);
9507 s += SvCUR(herewas);
9509 tmpstr = NEWSV(87,79);
9510 sv_upgrade(tmpstr, SVt_PVIV);
9513 SvIV_set(tmpstr, -1);
9515 else if (term == '`') {
9516 op_type = OP_BACKTICK;
9517 SvIV_set(tmpstr, '\\');
9521 PL_multi_start = CopLINE(PL_curcop);
9522 PL_multi_open = PL_multi_close = '<';
9523 term = *PL_tokenbuf;
9524 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
9525 char *bufptr = PL_sublex_info.super_bufptr;
9526 char *bufend = PL_sublex_info.super_bufend;
9527 char *olds = s - SvCUR(herewas);
9528 s = strchr(bufptr, '\n');
9532 while (s < bufend &&
9533 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
9535 CopLINE_inc(PL_curcop);
9538 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
9539 missingterm(PL_tokenbuf);
9541 sv_setpvn(herewas,bufptr,d-bufptr+1);
9542 sv_setpvn(tmpstr,d+1,s-d);
9544 sv_catpvn(herewas,s,bufend-s);
9545 Copy(SvPVX(herewas),bufptr,SvCUR(herewas) + 1,char);
9552 while (s < PL_bufend &&
9553 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
9555 CopLINE_inc(PL_curcop);
9557 if (s >= PL_bufend) {
9558 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
9559 missingterm(PL_tokenbuf);
9561 sv_setpvn(tmpstr,d+1,s-d);
9563 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
9565 sv_catpvn(herewas,s,PL_bufend-s);
9566 sv_setsv(PL_linestr,herewas);
9567 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
9568 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
9569 PL_last_lop = PL_last_uni = Nullch;
9572 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
9573 while (s >= PL_bufend) { /* multiple line string? */
9575 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
9576 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
9577 missingterm(PL_tokenbuf);
9579 CopLINE_inc(PL_curcop);
9580 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
9581 PL_last_lop = PL_last_uni = Nullch;
9582 #ifndef PERL_STRICT_CR
9583 if (PL_bufend - PL_linestart >= 2) {
9584 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
9585 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
9587 PL_bufend[-2] = '\n';
9589 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
9591 else if (PL_bufend[-1] == '\r')
9592 PL_bufend[-1] = '\n';
9594 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
9595 PL_bufend[-1] = '\n';
9597 if (PERLDB_LINE && PL_curstash != PL_debstash) {
9598 SV *sv = NEWSV(88,0);
9600 sv_upgrade(sv, SVt_PVMG);
9601 sv_setsv(sv,PL_linestr);
9604 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
9606 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
9607 STRLEN off = PL_bufend - 1 - SvPVX(PL_linestr);
9608 *(SvPVX(PL_linestr) + off ) = ' ';
9609 sv_catsv(PL_linestr,herewas);
9610 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
9611 s = SvPVX(PL_linestr) + off; /* In case PV of PL_linestr moved. */
9615 sv_catsv(tmpstr,PL_linestr);
9620 PL_multi_end = CopLINE(PL_curcop);
9621 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
9622 SvPV_shrink_to_cur(tmpstr);
9624 SvREFCNT_dec(herewas);
9626 if (UTF && is_utf8_string((U8*)SvPVX(tmpstr), SvCUR(tmpstr)))
9628 else if (PL_encoding)
9629 sv_recode_to_utf8(tmpstr, PL_encoding);
9631 PL_lex_stuff = tmpstr;
9632 yylval.ival = op_type;
9637 takes: current position in input buffer
9638 returns: new position in input buffer
9639 side-effects: yylval and lex_op are set.
9644 <FH> read from filehandle
9645 <pkg::FH> read from package qualified filehandle
9646 <pkg'FH> read from package qualified filehandle
9647 <$fh> read from filehandle in $fh
9653 S_scan_inputsymbol(pTHX_ char *start)
9655 register char *s = start; /* current position in buffer */
9661 d = PL_tokenbuf; /* start of temp holding space */
9662 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
9663 end = strchr(s, '\n');
9666 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
9668 /* die if we didn't have space for the contents of the <>,
9669 or if it didn't end, or if we see a newline
9672 if (len >= sizeof PL_tokenbuf)
9673 Perl_croak(aTHX_ "Excessively long <> operator");
9675 Perl_croak(aTHX_ "Unterminated <> operator");
9680 Remember, only scalar variables are interpreted as filehandles by
9681 this code. Anything more complex (e.g., <$fh{$num}>) will be
9682 treated as a glob() call.
9683 This code makes use of the fact that except for the $ at the front,
9684 a scalar variable and a filehandle look the same.
9686 if (*d == '$' && d[1]) d++;
9688 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
9689 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
9692 /* If we've tried to read what we allow filehandles to look like, and
9693 there's still text left, then it must be a glob() and not a getline.
9694 Use scan_str to pull out the stuff between the <> and treat it
9695 as nothing more than a string.
9698 if (d - PL_tokenbuf != len) {
9699 yylval.ival = OP_GLOB;
9701 s = scan_str(start,FALSE,FALSE);
9703 Perl_croak(aTHX_ "Glob not terminated");
9707 bool readline_overriden = FALSE;
9708 GV *gv_readline = Nullgv;
9710 /* we're in a filehandle read situation */
9713 /* turn <> into <ARGV> */
9715 Copy("ARGV",d,5,char);
9717 /* Check whether readline() is overriden */
9718 if (((gv_readline = gv_fetchpv("readline", FALSE, SVt_PVCV))
9719 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline))
9721 ((gvp = (GV**)hv_fetch(PL_globalstash, "readline", 8, FALSE))
9722 && (gv_readline = *gvp) != (GV*)&PL_sv_undef
9723 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline)))
9724 readline_overriden = TRUE;
9726 /* if <$fh>, create the ops to turn the variable into a
9732 /* try to find it in the pad for this block, otherwise find
9733 add symbol table ops
9735 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
9736 if (PAD_COMPNAME_FLAGS(tmp) & SVpad_OUR) {
9737 SV *sym = sv_2mortal(
9738 newSVpv(HvNAME(PAD_COMPNAME_OURSTASH(tmp)),0));
9739 sv_catpvn(sym, "::", 2);
9745 OP *o = newOP(OP_PADSV, 0);
9747 PL_lex_op = readline_overriden
9748 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
9749 append_elem(OP_LIST, o,
9750 newCVREF(0, newGVOP(OP_GV,0,gv_readline))))
9751 : (OP*)newUNOP(OP_READLINE, 0, o);
9760 ? (GV_ADDMULTI | GV_ADDINEVAL)
9763 PL_lex_op = readline_overriden
9764 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
9765 append_elem(OP_LIST,
9766 newUNOP(OP_RV2SV, 0, newGVOP(OP_GV, 0, gv)),
9767 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
9768 : (OP*)newUNOP(OP_READLINE, 0,
9769 newUNOP(OP_RV2SV, 0,
9770 newGVOP(OP_GV, 0, gv)));
9772 if (!readline_overriden)
9773 PL_lex_op->op_flags |= OPf_SPECIAL;
9774 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
9775 yylval.ival = OP_NULL;
9778 /* If it's none of the above, it must be a literal filehandle
9779 (<Foo::BAR> or <FOO>) so build a simple readline OP */
9781 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
9782 PL_lex_op = readline_overriden
9783 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
9784 append_elem(OP_LIST,
9785 newGVOP(OP_GV, 0, gv),
9786 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
9787 : (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
9788 yylval.ival = OP_NULL;
9797 takes: start position in buffer
9798 keep_quoted preserve \ on the embedded delimiter(s)
9799 keep_delims preserve the delimiters around the string
9800 returns: position to continue reading from buffer
9801 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
9802 updates the read buffer.
9804 This subroutine pulls a string out of the input. It is called for:
9805 q single quotes q(literal text)
9806 ' single quotes 'literal text'
9807 qq double quotes qq(interpolate $here please)
9808 " double quotes "interpolate $here please"
9809 qx backticks qx(/bin/ls -l)
9810 ` backticks `/bin/ls -l`
9811 qw quote words @EXPORT_OK = qw( func() $spam )
9812 m// regexp match m/this/
9813 s/// regexp substitute s/this/that/
9814 tr/// string transliterate tr/this/that/
9815 y/// string transliterate y/this/that/
9816 ($*@) sub prototypes sub foo ($)
9817 (stuff) sub attr parameters sub foo : attr(stuff)
9818 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
9820 In most of these cases (all but <>, patterns and transliterate)
9821 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
9822 calls scan_str(). s/// makes yylex() call scan_subst() which calls
9823 scan_str(). tr/// and y/// make yylex() call scan_trans() which
9826 It skips whitespace before the string starts, and treats the first
9827 character as the delimiter. If the delimiter is one of ([{< then
9828 the corresponding "close" character )]}> is used as the closing
9829 delimiter. It allows quoting of delimiters, and if the string has
9830 balanced delimiters ([{<>}]) it allows nesting.
9832 On success, the SV with the resulting string is put into lex_stuff or,
9833 if that is already non-NULL, into lex_repl. The second case occurs only
9834 when parsing the RHS of the special constructs s/// and tr/// (y///).
9835 For convenience, the terminating delimiter character is stuffed into
9840 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
9842 SV *sv; /* scalar value: string */
9843 char *tmps; /* temp string, used for delimiter matching */
9844 register char *s = start; /* current position in the buffer */
9845 register char term; /* terminating character */
9846 register char *to; /* current position in the sv's data */
9847 I32 brackets = 1; /* bracket nesting level */
9848 bool has_utf8 = FALSE; /* is there any utf8 content? */
9849 I32 termcode; /* terminating char. code */
9850 U8 termstr[UTF8_MAXBYTES]; /* terminating string */
9851 STRLEN termlen; /* length of terminating string */
9852 char *last = NULL; /* last position for nesting bracket */
9854 /* skip space before the delimiter */
9858 /* mark where we are, in case we need to report errors */
9861 /* after skipping whitespace, the next character is the terminator */
9864 termcode = termstr[0] = term;
9868 termcode = utf8_to_uvchr((U8*)s, &termlen);
9869 Copy(s, termstr, termlen, U8);
9870 if (!UTF8_IS_INVARIANT(term))
9874 /* mark where we are */
9875 PL_multi_start = CopLINE(PL_curcop);
9876 PL_multi_open = term;
9878 /* find corresponding closing delimiter */
9879 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
9880 termcode = termstr[0] = term = tmps[5];
9882 PL_multi_close = term;
9884 /* create a new SV to hold the contents. 87 is leak category, I'm
9885 assuming. 79 is the SV's initial length. What a random number. */
9887 sv_upgrade(sv, SVt_PVIV);
9888 SvIV_set(sv, termcode);
9889 (void)SvPOK_only(sv); /* validate pointer */
9891 /* move past delimiter and try to read a complete string */
9893 sv_catpvn(sv, s, termlen);
9896 if (PL_encoding && !UTF) {
9900 int offset = s - SvPVX(PL_linestr);
9901 bool found = sv_cat_decode(sv, PL_encoding, PL_linestr,
9902 &offset, (char*)termstr, termlen);
9903 char *ns = SvPVX(PL_linestr) + offset;
9904 char *svlast = SvEND(sv) - 1;
9906 for (; s < ns; s++) {
9907 if (*s == '\n' && !PL_rsfp)
9908 CopLINE_inc(PL_curcop);
9911 goto read_more_line;
9913 /* handle quoted delimiters */
9914 if (SvCUR(sv) > 1 && *(svlast-1) == '\\') {
9916 for (t = svlast-2; t >= SvPVX(sv) && *t == '\\';)
9918 if ((svlast-1 - t) % 2) {
9922 SvCUR_set(sv, SvCUR(sv) - 1);
9927 if (PL_multi_open == PL_multi_close) {
9935 for (t = w = last; t < svlast; w++, t++) {
9936 /* At here, all closes are "was quoted" one,
9937 so we don't check PL_multi_close. */
9939 if (!keep_quoted && *(t+1) == PL_multi_open)
9944 else if (*t == PL_multi_open)
9952 SvCUR_set(sv, w - SvPVX(sv));
9955 if (--brackets <= 0)
9961 SvCUR_set(sv, SvCUR(sv) - 1);
9967 /* extend sv if need be */
9968 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
9969 /* set 'to' to the next character in the sv's string */
9970 to = SvPVX(sv)+SvCUR(sv);
9972 /* if open delimiter is the close delimiter read unbridle */
9973 if (PL_multi_open == PL_multi_close) {
9974 for (; s < PL_bufend; s++,to++) {
9975 /* embedded newlines increment the current line number */
9976 if (*s == '\n' && !PL_rsfp)
9977 CopLINE_inc(PL_curcop);
9978 /* handle quoted delimiters */
9979 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
9980 if (!keep_quoted && s[1] == term)
9982 /* any other quotes are simply copied straight through */
9986 /* terminate when run out of buffer (the for() condition), or
9987 have found the terminator */
9988 else if (*s == term) {
9991 if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen))
9994 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
10000 /* if the terminator isn't the same as the start character (e.g.,
10001 matched brackets), we have to allow more in the quoting, and
10002 be prepared for nested brackets.
10005 /* read until we run out of string, or we find the terminator */
10006 for (; s < PL_bufend; s++,to++) {
10007 /* embedded newlines increment the line count */
10008 if (*s == '\n' && !PL_rsfp)
10009 CopLINE_inc(PL_curcop);
10010 /* backslashes can escape the open or closing characters */
10011 if (*s == '\\' && s+1 < PL_bufend) {
10012 if (!keep_quoted &&
10013 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
10018 /* allow nested opens and closes */
10019 else if (*s == PL_multi_close && --brackets <= 0)
10021 else if (*s == PL_multi_open)
10023 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
10028 /* terminate the copied string and update the sv's end-of-string */
10030 SvCUR_set(sv, to - SvPVX(sv));
10033 * this next chunk reads more into the buffer if we're not done yet
10037 break; /* handle case where we are done yet :-) */
10039 #ifndef PERL_STRICT_CR
10040 if (to - SvPVX(sv) >= 2) {
10041 if ((to[-2] == '\r' && to[-1] == '\n') ||
10042 (to[-2] == '\n' && to[-1] == '\r'))
10046 SvCUR_set(sv, to - SvPVX(sv));
10048 else if (to[-1] == '\r')
10051 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
10056 /* if we're out of file, or a read fails, bail and reset the current
10057 line marker so we can report where the unterminated string began
10060 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
10062 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
10065 /* we read a line, so increment our line counter */
10066 CopLINE_inc(PL_curcop);
10068 /* update debugger info */
10069 if (PERLDB_LINE && PL_curstash != PL_debstash) {
10070 SV *sv = NEWSV(88,0);
10072 sv_upgrade(sv, SVt_PVMG);
10073 sv_setsv(sv,PL_linestr);
10074 (void)SvIOK_on(sv);
10076 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
10079 /* having changed the buffer, we must update PL_bufend */
10080 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
10081 PL_last_lop = PL_last_uni = Nullch;
10084 /* at this point, we have successfully read the delimited string */
10086 if (!PL_encoding || UTF) {
10088 sv_catpvn(sv, s, termlen);
10091 if (has_utf8 || PL_encoding)
10094 PL_multi_end = CopLINE(PL_curcop);
10096 /* if we allocated too much space, give some back */
10097 if (SvCUR(sv) + 5 < SvLEN(sv)) {
10098 SvLEN_set(sv, SvCUR(sv) + 1);
10099 SvPV_renew(sv, SvLEN(sv));
10102 /* decide whether this is the first or second quoted string we've read
10115 takes: pointer to position in buffer
10116 returns: pointer to new position in buffer
10117 side-effects: builds ops for the constant in yylval.op
10119 Read a number in any of the formats that Perl accepts:
10121 \d(_?\d)*(\.(\d(_?\d)*)?)?[Ee][\+\-]?(\d(_?\d)*) 12 12.34 12.
10122 \.\d(_?\d)*[Ee][\+\-]?(\d(_?\d)*) .34
10125 0x[0-9A-Fa-f](_?[0-9A-Fa-f])*
10127 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
10130 If it reads a number without a decimal point or an exponent, it will
10131 try converting the number to an integer and see if it can do so
10132 without loss of precision.
10136 Perl_scan_num(pTHX_ const char *start, YYSTYPE* lvalp)
10138 register const char *s = start; /* current position in buffer */
10139 register char *d; /* destination in temp buffer */
10140 register char *e; /* end of temp buffer */
10141 NV nv; /* number read, as a double */
10142 SV *sv = Nullsv; /* place to put the converted number */
10143 bool floatit; /* boolean: int or float? */
10144 const char *lastub = 0; /* position of last underbar */
10145 static char const number_too_long[] = "Number too long";
10147 /* We use the first character to decide what type of number this is */
10151 Perl_croak(aTHX_ "panic: scan_num");
10153 /* if it starts with a 0, it could be an octal number, a decimal in
10154 0.13 disguise, or a hexadecimal number, or a binary number. */
10158 u holds the "number so far"
10159 shift the power of 2 of the base
10160 (hex == 4, octal == 3, binary == 1)
10161 overflowed was the number more than we can hold?
10163 Shift is used when we add a digit. It also serves as an "are
10164 we in octal/hex/binary?" indicator to disallow hex characters
10165 when in octal mode.
10170 bool overflowed = FALSE;
10171 bool just_zero = TRUE; /* just plain 0 or binary number? */
10172 static const NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
10173 static const char* const bases[5] =
10174 { "", "binary", "", "octal", "hexadecimal" };
10175 static const char* const Bases[5] =
10176 { "", "Binary", "", "Octal", "Hexadecimal" };
10177 static const char* const maxima[5] =
10179 "0b11111111111111111111111111111111",
10183 const char *base, *Base, *max;
10185 /* check for hex */
10190 } else if (s[1] == 'b') {
10195 /* check for a decimal in disguise */
10196 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
10198 /* so it must be octal */
10205 if (ckWARN(WARN_SYNTAX))
10206 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10207 "Misplaced _ in number");
10211 base = bases[shift];
10212 Base = Bases[shift];
10213 max = maxima[shift];
10215 /* read the rest of the number */
10217 /* x is used in the overflow test,
10218 b is the digit we're adding on. */
10223 /* if we don't mention it, we're done */
10227 /* _ are ignored -- but warned about if consecutive */
10229 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
10230 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10231 "Misplaced _ in number");
10235 /* 8 and 9 are not octal */
10236 case '8': case '9':
10238 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
10242 case '2': case '3': case '4':
10243 case '5': case '6': case '7':
10245 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
10248 case '0': case '1':
10249 b = *s++ & 15; /* ASCII digit -> value of digit */
10253 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
10254 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
10255 /* make sure they said 0x */
10258 b = (*s++ & 7) + 9;
10260 /* Prepare to put the digit we have onto the end
10261 of the number so far. We check for overflows.
10267 x = u << shift; /* make room for the digit */
10269 if ((x >> shift) != u
10270 && !(PL_hints & HINT_NEW_BINARY)) {
10273 if (ckWARN_d(WARN_OVERFLOW))
10274 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
10275 "Integer overflow in %s number",
10278 u = x | b; /* add the digit to the end */
10281 n *= nvshift[shift];
10282 /* If an NV has not enough bits in its
10283 * mantissa to represent an UV this summing of
10284 * small low-order numbers is a waste of time
10285 * (because the NV cannot preserve the
10286 * low-order bits anyway): we could just
10287 * remember when did we overflow and in the
10288 * end just multiply n by the right
10296 /* if we get here, we had success: make a scalar value from
10301 /* final misplaced underbar check */
10302 if (s[-1] == '_') {
10303 if (ckWARN(WARN_SYNTAX))
10304 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
10309 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
10310 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
10311 "%s number > %s non-portable",
10317 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
10318 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
10319 "%s number > %s non-portable",
10324 if (just_zero && (PL_hints & HINT_NEW_INTEGER))
10325 sv = new_constant(start, s - start, "integer",
10327 else if (PL_hints & HINT_NEW_BINARY)
10328 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
10333 handle decimal numbers.
10334 we're also sent here when we read a 0 as the first digit
10336 case '1': case '2': case '3': case '4': case '5':
10337 case '6': case '7': case '8': case '9': case '.':
10340 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
10343 /* read next group of digits and _ and copy into d */
10344 while (isDIGIT(*s) || *s == '_') {
10345 /* skip underscores, checking for misplaced ones
10349 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
10350 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10351 "Misplaced _ in number");
10355 /* check for end of fixed-length buffer */
10357 Perl_croak(aTHX_ number_too_long);
10358 /* if we're ok, copy the character */
10363 /* final misplaced underbar check */
10364 if (lastub && s == lastub + 1) {
10365 if (ckWARN(WARN_SYNTAX))
10366 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
10369 /* read a decimal portion if there is one. avoid
10370 3..5 being interpreted as the number 3. followed
10373 if (*s == '.' && s[1] != '.') {
10378 if (ckWARN(WARN_SYNTAX))
10379 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10380 "Misplaced _ in number");
10384 /* copy, ignoring underbars, until we run out of digits.
10386 for (; isDIGIT(*s) || *s == '_'; s++) {
10387 /* fixed length buffer check */
10389 Perl_croak(aTHX_ number_too_long);
10391 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
10392 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10393 "Misplaced _ in number");
10399 /* fractional part ending in underbar? */
10400 if (s[-1] == '_') {
10401 if (ckWARN(WARN_SYNTAX))
10402 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10403 "Misplaced _ in number");
10405 if (*s == '.' && isDIGIT(s[1])) {
10406 /* oops, it's really a v-string, but without the "v" */
10412 /* read exponent part, if present */
10413 if ((*s == 'e' || *s == 'E') && strchr("+-0123456789_", s[1])) {
10417 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
10418 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
10420 /* stray preinitial _ */
10422 if (ckWARN(WARN_SYNTAX))
10423 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10424 "Misplaced _ in number");
10428 /* allow positive or negative exponent */
10429 if (*s == '+' || *s == '-')
10432 /* stray initial _ */
10434 if (ckWARN(WARN_SYNTAX))
10435 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10436 "Misplaced _ in number");
10440 /* read digits of exponent */
10441 while (isDIGIT(*s) || *s == '_') {
10444 Perl_croak(aTHX_ number_too_long);
10448 if (ckWARN(WARN_SYNTAX) &&
10449 ((lastub && s == lastub + 1) ||
10450 (!isDIGIT(s[1]) && s[1] != '_')))
10451 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10452 "Misplaced _ in number");
10459 /* make an sv from the string */
10463 We try to do an integer conversion first if no characters
10464 indicating "float" have been found.
10469 int flags = grok_number (PL_tokenbuf, d - PL_tokenbuf, &uv);
10471 if (flags == IS_NUMBER_IN_UV) {
10473 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
10476 } else if (flags == (IS_NUMBER_IN_UV | IS_NUMBER_NEG)) {
10477 if (uv <= (UV) IV_MIN)
10478 sv_setiv(sv, -(IV)uv);
10485 /* terminate the string */
10487 nv = Atof(PL_tokenbuf);
10491 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
10492 (PL_hints & HINT_NEW_INTEGER) )
10493 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
10494 (floatit ? "float" : "integer"),
10498 /* if it starts with a v, it could be a v-string */
10501 sv = NEWSV(92,5); /* preallocate storage space */
10502 s = scan_vstring(s,sv);
10506 /* make the op for the constant and return */
10509 lvalp->opval = newSVOP(OP_CONST, 0, sv);
10511 lvalp->opval = Nullop;
10517 S_scan_formline(pTHX_ register char *s)
10519 register char *eol;
10521 SV *stuff = newSVpvn("",0);
10522 bool needargs = FALSE;
10523 bool eofmt = FALSE;
10525 while (!needargs) {
10528 #ifdef PERL_STRICT_CR
10529 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
10531 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
10533 if (*t == '\n' || t == PL_bufend) {
10538 if (PL_in_eval && !PL_rsfp) {
10539 eol = (char *) memchr(s,'\n',PL_bufend-s);
10544 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
10546 for (t = s; t < eol; t++) {
10547 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
10549 goto enough; /* ~~ must be first line in formline */
10551 if (*t == '@' || *t == '^')
10555 sv_catpvn(stuff, s, eol-s);
10556 #ifndef PERL_STRICT_CR
10557 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
10558 char *end = SvPVX(stuff) + SvCUR(stuff);
10561 SvCUR_set(stuff, SvCUR(stuff) - 1);
10570 s = filter_gets(PL_linestr, PL_rsfp, 0);
10571 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
10572 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
10573 PL_last_lop = PL_last_uni = Nullch;
10582 if (SvCUR(stuff)) {
10585 PL_lex_state = LEX_NORMAL;
10586 PL_nextval[PL_nexttoke].ival = 0;
10590 PL_lex_state = LEX_FORMLINE;
10592 if (UTF && is_utf8_string((U8*)SvPVX(stuff), SvCUR(stuff)))
10594 else if (PL_encoding)
10595 sv_recode_to_utf8(stuff, PL_encoding);
10597 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
10599 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
10603 SvREFCNT_dec(stuff);
10605 PL_lex_formbrack = 0;
10616 PL_cshlen = strlen(PL_cshname);
10621 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
10623 I32 oldsavestack_ix = PL_savestack_ix;
10624 CV* outsidecv = PL_compcv;
10627 assert(SvTYPE(PL_compcv) == SVt_PVCV);
10629 SAVEI32(PL_subline);
10630 save_item(PL_subname);
10631 SAVESPTR(PL_compcv);
10633 PL_compcv = (CV*)NEWSV(1104,0);
10634 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
10635 CvFLAGS(PL_compcv) |= flags;
10637 PL_subline = CopLINE(PL_curcop);
10638 CvPADLIST(PL_compcv) = pad_new(padnew_SAVE|padnew_SAVESUB);
10639 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
10640 CvOUTSIDE_SEQ(PL_compcv) = PL_cop_seqmax;
10642 return oldsavestack_ix;
10646 #pragma segment Perl_yylex
10649 Perl_yywarn(pTHX_ const char *s)
10651 PL_in_eval |= EVAL_WARNONLY;
10653 PL_in_eval &= ~EVAL_WARNONLY;
10658 Perl_yyerror(pTHX_ const char *s)
10660 const char *where = NULL;
10661 const char *context = NULL;
10665 if (!yychar || (yychar == ';' && !PL_rsfp))
10667 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
10668 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
10671 The code below is removed for NetWare because it abends/crashes on NetWare
10672 when the script has error such as not having the closing quotes like:
10673 if ($var eq "value)
10674 Checking of white spaces is anyway done in NetWare code.
10677 while (isSPACE(*PL_oldoldbufptr))
10680 context = PL_oldoldbufptr;
10681 contlen = PL_bufptr - PL_oldoldbufptr;
10683 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
10684 PL_oldbufptr != PL_bufptr) {
10687 The code below is removed for NetWare because it abends/crashes on NetWare
10688 when the script has error such as not having the closing quotes like:
10689 if ($var eq "value)
10690 Checking of white spaces is anyway done in NetWare code.
10693 while (isSPACE(*PL_oldbufptr))
10696 context = PL_oldbufptr;
10697 contlen = PL_bufptr - PL_oldbufptr;
10699 else if (yychar > 255)
10700 where = "next token ???";
10701 else if (yychar == -2) { /* YYEMPTY */
10702 if (PL_lex_state == LEX_NORMAL ||
10703 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
10704 where = "at end of line";
10705 else if (PL_lex_inpat)
10706 where = "within pattern";
10708 where = "within string";
10711 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
10713 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
10714 else if (isPRINT_LC(yychar))
10715 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
10717 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
10718 where = SvPVX(where_sv);
10720 msg = sv_2mortal(newSVpv(s, 0));
10721 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
10722 OutCopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
10724 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
10726 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
10727 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
10728 Perl_sv_catpvf(aTHX_ msg,
10729 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
10730 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
10733 if (PL_in_eval & EVAL_WARNONLY && ckWARN_d(WARN_SYNTAX))
10734 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "%"SVf, msg);
10737 if (PL_error_count >= 10) {
10738 if (PL_in_eval && SvCUR(ERRSV))
10739 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
10740 ERRSV, OutCopFILE(PL_curcop));
10742 Perl_croak(aTHX_ "%s has too many errors.\n",
10743 OutCopFILE(PL_curcop));
10746 PL_in_my_stash = Nullhv;
10750 #pragma segment Main
10754 S_swallow_bom(pTHX_ U8 *s)
10756 const STRLEN slen = SvCUR(PL_linestr);
10759 if (s[1] == 0xFE) {
10760 /* UTF-16 little-endian? (or UTF32-LE?) */
10761 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
10762 Perl_croak(aTHX_ "Unsupported script encoding UTF32-LE");
10763 #ifndef PERL_NO_UTF16_FILTER
10764 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF16-LE script encoding (BOM)\n");
10767 if (PL_bufend > (char*)s) {
10771 filter_add(utf16rev_textfilter, NULL);
10772 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
10773 utf16_to_utf8_reversed(s, news,
10774 PL_bufend - (char*)s - 1,
10776 sv_setpvn(PL_linestr, (const char*)news, newlen);
10778 SvUTF8_on(PL_linestr);
10779 s = (U8*)SvPVX(PL_linestr);
10780 PL_bufend = SvPVX(PL_linestr) + newlen;
10783 Perl_croak(aTHX_ "Unsupported script encoding UTF16-LE");
10788 if (s[1] == 0xFF) { /* UTF-16 big-endian? */
10789 #ifndef PERL_NO_UTF16_FILTER
10790 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (BOM)\n");
10793 if (PL_bufend > (char *)s) {
10797 filter_add(utf16_textfilter, NULL);
10798 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
10799 utf16_to_utf8(s, news,
10800 PL_bufend - (char*)s,
10802 sv_setpvn(PL_linestr, (const char*)news, newlen);
10804 SvUTF8_on(PL_linestr);
10805 s = (U8*)SvPVX(PL_linestr);
10806 PL_bufend = SvPVX(PL_linestr) + newlen;
10809 Perl_croak(aTHX_ "Unsupported script encoding UTF16-BE");
10814 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
10815 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-8 script encoding (BOM)\n");
10816 s += 3; /* UTF-8 */
10822 if (s[2] == 0xFE && s[3] == 0xFF) {
10823 /* UTF-32 big-endian */
10824 Perl_croak(aTHX_ "Unsupported script encoding UTF32-BE");
10827 else if (s[2] == 0 && s[3] != 0) {
10830 * are a good indicator of UTF-16BE. */
10831 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (no BOM)\n");
10836 if (slen > 3 && s[1] == 0 && s[2] != 0 && s[3] == 0) {
10839 * are a good indicator of UTF-16LE. */
10840 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16LE script encoding (no BOM)\n");
10849 * Restore a source filter.
10853 restore_rsfp(pTHX_ void *f)
10855 PerlIO *fp = (PerlIO*)f;
10857 if (PL_rsfp == PerlIO_stdin())
10858 PerlIO_clearerr(PL_rsfp);
10859 else if (PL_rsfp && (PL_rsfp != fp))
10860 PerlIO_close(PL_rsfp);
10864 #ifndef PERL_NO_UTF16_FILTER
10866 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
10868 const STRLEN old = SvCUR(sv);
10869 const I32 count = FILTER_READ(idx+1, sv, maxlen);
10870 DEBUG_P(PerlIO_printf(Perl_debug_log,
10871 "utf16_textfilter(%p): %d %d (%d)\n",
10872 utf16_textfilter, idx, maxlen, (int) count));
10876 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
10877 Copy(SvPVX(sv), tmps, old, char);
10878 utf16_to_utf8((U8*)SvPVX(sv) + old, tmps + old,
10879 SvCUR(sv) - old, &newlen);
10880 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
10882 DEBUG_P({sv_dump(sv);});
10887 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
10889 const STRLEN old = SvCUR(sv);
10890 const I32 count = FILTER_READ(idx+1, sv, maxlen);
10891 DEBUG_P(PerlIO_printf(Perl_debug_log,
10892 "utf16rev_textfilter(%p): %d %d (%d)\n",
10893 utf16rev_textfilter, idx, maxlen, (int) count));
10897 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
10898 Copy(SvPVX(sv), tmps, old, char);
10899 utf16_to_utf8((U8*)SvPVX(sv) + old, tmps + old,
10900 SvCUR(sv) - old, &newlen);
10901 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
10903 DEBUG_P({ sv_dump(sv); });
10909 Returns a pointer to the next character after the parsed
10910 vstring, as well as updating the passed in sv.
10912 Function must be called like
10915 s = scan_vstring(s,sv);
10917 The sv should already be large enough to store the vstring
10918 passed in, for performance reasons.
10923 Perl_scan_vstring(pTHX_ const char *s, SV *sv)
10925 const char *pos = s;
10926 const char *start = s;
10927 if (*pos == 'v') pos++; /* get past 'v' */
10928 while (pos < PL_bufend && (isDIGIT(*pos) || *pos == '_'))
10930 if ( *pos != '.') {
10931 /* this may not be a v-string if followed by => */
10932 const char *next = pos;
10933 while (next < PL_bufend && isSPACE(*next))
10935 if ((PL_bufend - next) >= 2 && *next == '=' && next[1] == '>' ) {
10936 /* return string not v-string */
10937 sv_setpvn(sv,(char *)s,pos-s);
10938 return (char *)pos;
10942 if (!isALPHA(*pos)) {
10944 U8 tmpbuf[UTF8_MAXBYTES+1];
10947 if (*s == 'v') s++; /* get past 'v' */
10949 sv_setpvn(sv, "", 0);
10954 /* this is atoi() that tolerates underscores */
10955 const char *end = pos;
10957 while (--end >= s) {
10962 rev += (*end - '0') * mult;
10964 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
10965 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
10966 "Integer overflow in decimal number");
10970 if (rev > 0x7FFFFFFF)
10971 Perl_croak(aTHX_ "In EBCDIC the v-string components cannot exceed 2147483647");
10973 /* Append native character for the rev point */
10974 tmpend = uvchr_to_utf8(tmpbuf, rev);
10975 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
10976 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(rev)))
10978 if (pos + 1 < PL_bufend && *pos == '.' && isDIGIT(pos[1]))
10984 while (pos < PL_bufend && (isDIGIT(*pos) || *pos == '_'))
10988 sv_magic(sv,NULL,PERL_MAGIC_vstring,(const char*)start, pos-start);
10996 * c-indentation-style: bsd
10997 * c-basic-offset: 4
10998 * indent-tabs-mode: t
11001 * vim: ts=8 sts=4 sw=4 noet: