4 #use warnings; FIXME - Bug 2505
7 use VerboseWarnings qw( pedantic_p error_normal warn_normal warn_pedantic );
10 use vars qw($VERSION @ISA @EXPORT @EXPORT_OK %EXPORT_TAGS);
12 ###############################################################################
16 TmplTokenizer.pm - Simple-minded tokenizer class for HTML::Template .tmpl files
20 Because .tmpl files contains HTML::Template directives
21 that tend to confuse real parsers (e.g., HTML::Parse),
22 it might be better to create a customized scanner
23 to scan the template files for tokens.
24 This module is a simple-minded attempt at such a scanner.
28 ###############################################################################
35 use vars qw( $pedantic_attribute_error_in_nonpedantic_mode_p );
36 use vars qw( $pedantic_tmpl_var_use_in_nonpedantic_mode_p );
37 use vars qw( $pedantic_error_markup_in_pcdata_p );
39 ###############################################################################
42 use vars qw( $re_directive $re_tmpl_var $re_tmpl_var_escaped $re_tmpl_include );
43 use vars qw( $re_directive_control $re_tmpl_endif_endloop $re_xsl);
45 # $re_directive must not do any backreferences
46 $re_directive = q{<(?:(?i)(?:!--\s*)?\/?TMPL_(?:VAR|LOOP|INCLUDE|IF|ELSE|UNLESS)(?:\s+(?:[a-zA-Z][-a-zA-Z0-9]*=)?(?:'[^']*'|"[^"]*"|[^\s<>]+))*\s*(?:--)?)>};
47 # TMPL_VAR or TMPL_INCLUDE
48 $re_tmpl_var = q{<(?:(?i)(?:!--\s*)?TMPL_(?:VAR)(?:\s+(?:[a-zA-Z][-a-zA-Z0-9]*=)?(?:'[^']*'|"[^"]*"|[^\s<>]+))*\s*(?:--)?)>};
49 $re_tmpl_include = q{<(?:(?i)(?:!--\s*)?TMPL_(?:INCLUDE)(?:\s+(?:[a-zA-Z][-a-zA-Z0-9]*=)?(?:'[^']*'|"[^"]*"|[^\s<>]+))*\s*(?:--)?)>};
50 # TMPL_VAR ESCAPE=1/HTML/URL
51 $re_xsl = q{<\/?(?:xsl:)(?:[\s\-a-zA-Z0-9"'\/\.\[\]\@\(\):=,$]+)\/?>};
52 $re_tmpl_var_escaped = q{<(?:(?i)(?:!--\s*)?TMPL_(?:VAR|INCLUDE)(?:\s+(?:[a-zA-Z][-a-zA-Z0-9]*=)?(?:'[^']*'|"[^"]*"|[^\s<>]+))\s+ESCAPE=(?:1|HTML|URL)(?:\s+(?:[a-zA-Z][-a-zA-Z0-9]*=)?(?:'[^']*'|"[^"]*"|[^\s<>]+))*\s*(?:--)?)>};
53 # Any control flow directive
54 $re_directive_control = q{<(?:(?i)(?:!--\s*)?\/?TMPL_(?:LOOP|IF|ELSE|UNLESS)(?:\s+(?:[a-zA-Z][-a-zA-Z0-9]*=)?(?:'[^']*'|"[^"]*"|[^\s<>]+))*\s*(?:--)?)>};
55 # /LOOP or /IF or /UNLESS
56 $re_tmpl_endif_endloop = q{<(?:(?i)(?:!--\s*)?\/TMPL_(?:LOOP|IF|UNLESS)(?:\s+(?:[a-zA-Z][-a-zA-Z0-9]*=)?(?:'[^']*'|"[^"]*"|[^\s<>]+))*\s*(?:--)?)>};
59 # Hideous stuff from subst.pl, slightly modified to use the above hideous stuff
60 # Note: The $re_tag's set $1 (<tag), $2 (>), and $3 (rest of string)
61 use vars qw( $re_comment $re_entity_name $re_end_entity $re_etag );
62 use vars qw( $re_tag_strict $re_tag_compat @re_tag );
65 my $etag = $compat? '>': '<>\/';
66 # This is no longer similar to the original regexp in subst.pl :-(
67 # Note that we don't want <> in compat mode; Mozilla knows about <
68 q{(<\/?(?:|(?:"(?:} . $re_directive . q{|[^"])*"|'(?:} . $re_directive . q{|[^'])*'|--(?:(?!--)(?:$re_directive)*.)*--|(?:}
70 . q{|(?!--)[^"'<>} . $etag . q{]))+))([} . $etag . q{]|(?=<))(.*)};
73 $re_comment = '(?:--(?:[^-]|-[^-])*--)';
74 $re_entity_name = '(?:[^&%#;<>\s]+)'; # NOTE: not really correct SGML
75 $re_end_entity = '(?:;|$|(?=\s))'; # semicolon or before-whitespace
76 $re_etag = q{(?:<\/?(?:"[^"]*"|'[^']*'|[^"'>\/])*[>\/])}; # end-tag
77 @re_tag = ($re_tag_strict, $re_tag_compat) = (re_tag(0), re_tag(1));
80 # End of the hideous stuff
82 use vars qw( $serial );
84 ###############################################################################
86 sub FATAL_P () {'fatal-p'}
87 sub SYNTAXERROR_P () {'syntaxerror-p'}
89 sub FILENAME () {'input'}
90 sub HANDLE () {'handle'}
92 sub READAHEAD () {'readahead'}
93 sub LINENUM_START () {'lc_0'}
95 sub CDATA_MODE_P () {'cdata-mode-p'}
96 sub CDATA_CLOSE () {'cdata-close'}
97 sub PCDATA_MODE_P () {'pcdata-mode-p'} # additional submode for CDATA
98 sub JS_MODE_P () {'js-mode-p'} # cdata-mode-p must also be true
100 sub ALLOW_CFORMAT_P () {'allow-cformat-p'}
105 open my $handle,$filename or die "can't open $filename";
107 filename => $filename
113 ###############################################################################
119 return $this->{filename};
124 return $this->{handle};
129 return $this->{+FATAL_P};
134 return $this->{+SYNTAXERROR_P};
137 sub has_readahead_p {
139 return @{$this->{readahead}};
142 sub _peek_readahead {
144 return $this->{readahead}->[$#{$this->{readahead}}];
147 sub line_number_start {
149 return $this->{+LINENUM_START};
154 return $this->{+LINENUM};
159 return $this->{+CDATA_MODE_P};
164 return $this->{+PCDATA_MODE_P};
169 return $this->{+JS_MODE_P};
174 return $this->{+CDATA_CLOSE};
177 sub allow_cformat_p {
179 return $this->{+ALLOW_CFORMAT_P};
186 $this->{+FATAL_P} = $_[0];
190 sub _set_syntaxerror {
192 $this->{+SYNTAXERROR_P} = $_[0];
196 sub _push_readahead {
198 push @{$this->{readahead}}, $_[0];
204 return pop @{$this->{readahead}};
207 sub _append_readahead {
209 $this->{readahead}->[$#{$this->{readahead}}] .= $_[0];
215 $this->{readahead}->[$#{$this->{readahead}}] = $_[0];
219 sub _increment_line_number {
221 $this->{+LINENUM} += 1;
225 sub _set_line_number_start {
227 $this->{+LINENUM_START} = $_[0];
231 sub _set_cdata_mode {
233 $this->{+CDATA_MODE_P} = $_[0];
237 sub _set_pcdata_mode {
239 $this->{+PCDATA_MODE_P} = $_[0];
245 $this->{+JS_MODE_P} = $_[0];
249 sub _set_cdata_close {
251 $this->{+CDATA_CLOSE} = $_[0];
255 sub set_allow_cformat {
257 $this->{+ALLOW_CFORMAT_P} = $_[0];
261 ###############################################################################
263 use vars qw( $js_EscapeSequence );
265 # Perl quoting is really screwed up, but this common subexp is way too long
266 $js_EscapeSequence = q{\\\\(?:['"\\\\bfnrt]|[^0-7xu]|[0-3]?[0-7]{1,2}|x[\da-fA-F]{2}|u[\da-fA-F]{4})};
268 sub parenleft () { '(' }
269 sub parenright () { ')' }
275 if ($s0 =~ /^\s+/s) { # whitespace
278 } elsif ($s0 =~ /^\/\/[^\r\n]*(?:[\r\n]|$)/s) { # C++-style comment
281 } elsif ($s0 =~ /^\/\*(?:(?!\*\/).)*\*\//s) { # C-style comment
284 # Keyword or identifier, ECMA-262 p.13 (section 7.5)
285 } elsif ($s0 =~ /^[A-Z_\$][A-Z\d_\$]*/is) { # IdentifierName
288 # Punctuator, ECMA-262 p.13 (section 7.6)
289 } elsif ($s0 =~ /^(?:[\(\){}\[\];]|>>>=|<<=|>>=|[-\+\*\/\&\|\^\%]=|>>>|<<|>>|--|\+\+|\|\||\&\&|==|<=|>=|!=|[=><,!~\?:\.\-\+\*\/\&\|\^\%])/s) {
292 # DecimalLiteral, ECMA-262 p.14 (section 7.7.3); note: bug in the spec
293 } elsif ($s0 =~ /^(?:0|[1-9]\d+(?:\.\d*(?:[eE][-\+]?\d+)?)?)/s) {
296 # HexIntegerLiteral, ECMA-262 p.15 (section 7.7.3)
297 } elsif ($s0 =~ /^0[xX][\da-fA-F]+/s) {
300 # OctalIntegerLiteral, ECMA-262 p.15 (section 7.7.3)
301 } elsif ($s0 =~ /^0[\da-fA-F]+/s) {
304 # StringLiteral, ECMA-262 p.17 (section 7.7.4)
305 # XXX SourceCharacter doesn't seem to be defined (?)
306 } elsif ($s0 =~ /^(?:"(?:(?!["\\\r\n]).|$js_EscapeSequence)*"|'(?:(?!['\\\r\n]).|$js_EscapeSequence)*')/os) {
309 } elsif ($s0 =~ /^./) { # UNKNOWN TOKEN !!!
317 sub STATE_UNDERSCORE () { 1 }
318 sub STATE_PARENLEFT () { 2 }
319 sub STATE_STRING_LITERAL () { 3 }
321 # XXX This is a crazy hack. I don't want to write an ECMAScript parser.
322 # XXX A scanner is one thing; a parser another thing.
323 sub identify_js_translatables (@) {
326 # We mark a JavaScript translatable string as in C, i.e., _("literal")
327 # For simplicity, we ONLY look for "_" "(" StringLiteral ")"
328 for (my $i = 0, my $state = 0, my($j, $q, $s); $i <= $#input; $i += 1) {
329 my $reset_state_p = 0;
330 push @output, [0, $input[$i]];
331 if ($input[$i] !~ /\S/s) {
333 } elsif ($state == 0) {
334 $state = STATE_UNDERSCORE if $input[$i] eq '_';
335 } elsif ($state == STATE_UNDERSCORE) {
336 $state = $input[$i] eq parenleft ? STATE_PARENLEFT : 0;
337 } elsif ($state == STATE_PARENLEFT) {
338 if ($input[$i] =~ /^(['"])(.*)\1$/s) {
339 ($state, $j, $q, $s) = (STATE_STRING_LITERAL, $#output, $1, $2);
343 } elsif ($state == STATE_STRING_LITERAL) {
344 if ($input[$i] eq parenright) {
345 $output[$j] = [1, $output[$j]->[1], $q, $s];
349 die "identify_js_translatables internal error: Unknown state $state"
355 ###############################################################################
357 sub _extract_attributes ($;$) {
361 $s = $1 if $s =~ /^<(?:(?!$re_directive_control)\S)+(.*)\/\S$/s # XML-style self-closing tags
362 || $s =~ /^<(?:(?!$re_directive_control)\S)+(.*)\S$/s; # SGML-style tags
364 for (my $i = 0; $s =~ /^(?:$re_directive_control)?\s+(?:$re_directive_control)?(?:([a-zA-Z][-a-zA-Z0-9]*)\s*=\s*)?('((?:$re_directive|[^'])*)'|"((?:$re_directive|[^"])*)"|((?:$re_directive|[^\s<>])+))/os;) {
365 my($key, $val, $val_orig, $rest)
366 = ($1, (defined $3? $3: defined $4? $4: $5), $2, $');
368 $attr{+lc($key)} = [$key, $val, $val_orig, $i];
370 if ($val =~ /$re_tmpl_include/os) {
371 warn_normal "TMPL_INCLUDE in attribute: $val_orig\n", $lc;
372 } elsif ($val =~ /$re_tmpl_var/os && $val !~ /$re_tmpl_var_escaped/os) {
373 # XXX: we probably should not warn if key is "onclick" etc
374 # XXX: there's just no reasonable thing to suggest
375 my $suggest = ($key =~ /^(?:action|archive|background|cite|classid|codebase|data|datasrc|for|href|longdesc|profile|src|usemap)$/i? 'URL': 'HTML');
376 undef $suggest if $key =~ /^(?:onblur|onchange|onclick|ondblclick|onfocus|onkeydown|onkeypress|onkeyup|onload|onmousedown|onmousemove|onmouseout|onmouseover|onmouseup|onreset|onselect|onsubmit|onunload)$/i;
378 "Suggest ESCAPE=$suggest for TMPL_VAR in attribute \"$key\""
380 $lc, \$pedantic_tmpl_var_use_in_nonpedantic_mode_p
381 if defined $suggest && (pedantic_p || !$pedantic_tmpl_var_use_in_nonpedantic_mode_p);
382 } elsif ($val_orig !~ /^['"]/) {
383 my $t = $val; $t =~ s/$re_directive_control//os;
385 "Unquoted attribute contains character(s) that should be quoted"
387 $lc, \$pedantic_attribute_error_in_nonpedantic_mode_p
388 if $t =~ /[^-\.A-Za-z0-9]/s;
391 my $s2 = $s; $s2 =~ s/$re_tmpl_endif_endloop//g; # for the next check
392 if ($s2 =~ /\S/s) { # should never happen
393 if ($s =~ /^([^\n]*)\n/s) { # this is even worse
394 error_normal("Completely confused while extracting attributes: $1", $lc);
395 error_normal((scalar(split(/\n/, $s)) - 1) . " more line(s) not shown.", undef);
396 $this->_set_fatal( 1 );
398 # There's something wrong with the attribute syntax.
399 # We might be able to deduce a likely cause by looking more.
400 if ($s =~ /^[a-z0-9]/is && "<foo $s>" =~ /^$re_tag_compat$/s) {
401 warn_normal "Probably missing whitespace before or missing quotation mark near: $s\n", $lc;
403 warn_normal "Strange attribute syntax: $s\n", $lc;
410 sub _next_token_internal {
415 $this->_pop_readahead if $this->has_readahead_p
416 && !ref $this->_peek_readahead
417 && !length $this->_peek_readahead;
418 if (!$this->has_readahead_p) {
419 my $next = scalar <$h>;
420 $eof_p = !defined $next;
422 $this->_increment_line_number;
423 $this->_push_readahead( $next );
426 $this->_set_line_number_start( $this->line_number ); # remember 1st line num
427 if ($this->has_readahead_p && ref $this->_peek_readahead) { # TmplToken obj.
428 ($it, $kind) = ($this->_pop_readahead, undef);
429 } elsif ($eof_p && !$this->has_readahead_p) { # nothing left to do
431 } elsif ($this->_peek_readahead =~ /^\s+/s) { # whitespace
432 ($kind, $it) = (TmplTokenType::TEXT, $&);
433 $this->_set_readahead( $' );
434 # FIXME the following (the [<\s] part) is an unreliable HACK :-(
435 } elsif ($this->_peek_readahead =~ /^(?:[^<]|<[<\s])*(?:[^<\s])/s) { # non-space normal text
436 ($kind, $it) = (TmplTokenType::TEXT, $&);
437 $this->_set_readahead( $' );
438 warn_normal "Unescaped < in $it\n", $this->line_number_start
439 if !$this->cdata_mode_p && $it =~ /</s;
440 } else { # tag/declaration/processing instruction
442 my $bad_comment_p = 0;
443 for (my $cdata_close = $this->cdata_close;;) {
444 if ($this->cdata_mode_p) {
445 my $next = $this->_pop_readahead;
446 if ($next =~ /^$cdata_close/is) {
447 ($kind, $it) = (TmplTokenType::TAG, $&);
448 $this->_push_readahead( $' );
450 } elsif ($next =~ /^((?:(?!$cdata_close).)+)($cdata_close)/is) {
451 ($kind, $it) = (TmplTokenType::TEXT, $1);
452 $this->_push_readahead( "$2$'" );
455 ($kind, $it) = (TmplTokenType::TEXT, $next);
458 } elsif ($this->_peek_readahead =~ /^$re_tag_compat/os) {
459 # If we detect a "closed start tag" but we know that the
460 # following token looks like a TMPL_VAR, don't stop
461 my($head, $tail, $post) = ($1, $2, $3);
462 if ($tail eq '' && $post =~ $re_tmpl_var) {
463 # Don't bother to show the warning if we're too confused
464 # FIXME. There's no method for _closed_start_tag_warning
465 if (!defined $this->{'_closed_start_tag_warning'}
466 || ($this->{'_closed_start_tag_warning'}->[0] eq $head
467 && $this->{'_closed_start_tag_warning'}->[1] != $this->line_number - 1)) {
468 warn_normal "Possible SGML \"closed start tag\" notation: $head<\n", $this->line_number
469 if split(/\n/, $head) < 10;
471 $this->{'_closed_start_tag_warning'} = [$head, $this->line_number];
473 ($kind, $it) = (TmplTokenType::TAG, "$head>");
474 $this->_set_readahead( $post );
476 warn_normal "SGML \"closed start tag\" notation: $head<\n", $this->line_number if $tail eq ''
477 and $head ne '<!DOCTYPE stylesheet ['; # another bit of temporary ugliness for bug 4472
479 } elsif ($this->_peek_readahead =~ /^<!--(?:(?!-->)$re_directive*.)*-->/os) {
480 ($kind, $it) = (TmplTokenType::COMMENT, $&);
481 $this->_set_readahead( $' );
486 my $next = scalar <$h>;
487 $eof_p = !defined $next;
489 $this->_increment_line_number;
490 $this->_append_readahead( $next );
492 if ($kind ne TmplTokenType::TAG) {
494 } elsif ($it =~ /^<!/) {
495 $kind = TmplTokenType::DECL;
496 $kind = TmplTokenType::COMMENT if $it =~ /^<!--(?:(?!-->).)*-->/;
497 if ($kind == TmplTokenType::COMMENT && $it =~ /^<!--\s*#include/s) {
498 warn_normal "Apache #include directive found instead of HTML::Template <TMPL_INCLUDE> directive", $this->line_number_start;
500 } elsif ($it =~ /^<\?/) {
501 $kind = TmplTokenType::PI;
503 if ($it =~ /^$re_directive/ios && !$this->cdata_mode_p) {
504 $kind = TmplTokenType::DIRECTIVE;
505 } elsif ($bad_comment_p) {
506 warn_normal sprintf("Syntax error in comment: %s\n", $it),
507 $this->line_number_start;
508 $this->_set_syntaxerror( 1 );
510 if (!$ok_p && $eof_p) {
511 ($kind, $it) = (TmplTokenType::UNKNOWN, $this->_peek_readahead);
512 $this->_set_readahead, undef;
513 $this->_set_syntaxerror( 1 );
516 warn_normal "Unrecognizable token found: "
517 . (split(/\n/, $it) < 10? $it: '(too confused to show details)')
518 . "\n", $this->line_number_start
519 if $kind == TmplTokenType::UNKNOWN;
520 return defined $it? (ref $it? $it: TmplToken->new($it, $kind, $this->line_number, $this->filename)): undef;
523 sub _next_token_intermediate {
525 my $h = $this->_handle;
527 if (!$this->cdata_mode_p) {
528 $it = $this->_next_token_internal($h);
529 if (defined $it && $it->type == TmplTokenType::TAG) {
530 if ($it->string =~ /^<(script|style|textarea)\b/is) {
531 $this->_set_cdata_mode( 1 );
532 $this->_set_cdata_close( "</$1\\s*>" );
533 $this->_set_pcdata_mode( 0 );
534 $this->_set_js_mode( lc($1) eq 'script' );
535 # } elsif ($it->string =~ /^<(title)\b/is) {
536 # $this->_set_cdata_mode( 1 );
537 # $this->_set_cdata_close( "</$1\\s*>" );
538 # $this->_set_pcdata_mode( 1 );
540 $it->set_attributes( $this->_extract_attributes($it->string, $it->line_number) );
544 for ($it = '', my $cdata_close = $this->cdata_close;;) {
545 my $next = $this->_next_token_internal($h);
546 $eof_p = !defined $next;
548 if (defined $next && $next->string =~ /$cdata_close/is) {
549 $this->_push_readahead( $next ); # push entire TmplToken object
550 $this->_set_cdata_mode( 0 );
552 last unless $this->cdata_mode_p;
553 $it .= $next->string;
557 error_normal "Unexpected end of file while looking for "
559 . "\n", $this->line_number_start;
560 $this->_set_fatal( 1 );
561 $this->_set_syntaxerror( 1 );
563 if ($this->pcdata_mode_p) {
565 $check =~ s/$re_directive//gos;
566 warn_pedantic "Markup found in PCDATA\n", $this->line_number,
567 \$pedantic_error_markup_in_pcdata_p
568 if $check =~ /$re_tag_compat/s;
570 # PCDATA should be treated as text, not CDATA
571 # Actually it should be treated as TEXT_PARAMETRIZED :-(
572 $it = TmplToken->new( $it,
573 ($this->pcdata_mode_p?
574 TmplTokenType::TEXT: TmplTokenType::CDATA),
575 $this->line_number, $this->filename )
577 if ($this->js_mode_p) {
578 my $s0 = $it->string;
581 if ($s0 =~ /^(\s*<!--\s*)(.*)(\s*--\s*>\s*)$/s) {
586 push @head, split_js $s0;
587 $it->set_js_data( identify_js_translatables(@head, @tail) );
589 $this->_set_pcdata_mode, 0;
590 $this->_set_cdata_close, undef unless !defined $it;
595 sub _token_groupable1_p ($) { # as first token, groupable into TEXT_PARAMETRIZED
597 return ($t->type == TmplTokenType::TEXT && $t->string !~ /^[,\.:\|\s]+$/is)
598 || ($t->type == TmplTokenType::DIRECTIVE
599 && $t->string =~ /^(?:$re_tmpl_var)$/os)
600 || ($t->type == TmplTokenType::TAG
601 && ($t->string =~ /^<(?:a|b|em|h[123456]|i|u)\b/is
602 || ($t->string =~ /^<input\b/is
603 && $t->attributes->{'type'}->[1] =~ /^(?:radio|text)$/is)
607 sub _token_groupable2_p ($) { # as other token, groupable into TEXT_PARAMETRIZED
609 return ($t->type == TmplTokenType::TEXT && ($t->string =~ /^\s*$/s || $t->string !~ /^[\|\s]+$/is))
610 || ($t->type == TmplTokenType::DIRECTIVE
611 && $t->string =~ /^(?:$re_tmpl_var)$/os)
612 || ($t->type == TmplTokenType::TAG
613 && ($t->string =~ /^<\/?(?:a|b|em|h[123456]|i|u)\b/is
614 || ($t->string =~ /^<input\b/is
615 && $t->attributes->{'type'}->[1] =~ /^(?:radio|text)$/is)))
618 sub _quote_cformat ($) {
624 sub string_canon ($) {
627 # Fold all whitespace into single blanks
633 sub _formalize_string_cformat ($) {
635 return _quote_cformat string_canon $s;
640 return $t->type == TmplTokenType::DIRECTIVE? '%s':
641 $t->type == TmplTokenType::TEXT?
642 _formalize_string_cformat($t->string):
643 $t->type == TmplTokenType::TAG?
644 ($t->string =~ /^<a\b/is? '<a>':
645 $t->string =~ /^<input\b/is? (
646 lc $t->attributes->{'type'}->[1] eq 'text' ? '%S':
648 _quote_cformat($t->string)):
649 _quote_cformat($t->string);
655 my $undo_trailing_blanks = sub {
656 for (my $i = $#structure; $i >= 0; $i -= 1) {
657 last unless ($structure[$i]->type == TmplTokenType::TEXT && blank_p($structure[$i]->string)) ;#|| ($structure[$i]->type == TmplTokenType::TAG && $structure[$i]->string =~ /^<br\b/is);
658 # Queue element structure: [reanalysis-p, token]
659 push @{$this->{_queue}}, [1, pop @structure];
662 &$undo_trailing_blanks;
663 while (@structure >= 2) {
664 my $something_done_p = 0;
665 # FIXME: If the last token is a close tag but there are no tags
666 # FIXME: before it, drop the close tag back into the queue. This
667 # FIXME: is an ugly hack to get rid of "foo %s</h1>" type mess.
669 && $structure[$#structure]->type == TmplTokenType::TAG
670 && $structure[$#structure]->string =~ /^<\//s) {
671 my $has_other_tags_p = 0;
672 for (my $i = 0; $i < $#structure; $i += 1) {
673 $has_other_tags_p = 1
674 if $structure[$i]->type == TmplTokenType::TAG;
675 last if $has_other_tags_p;
677 if (!$has_other_tags_p) {
678 push @{$this->{_queue}}, [0, pop @structure]
679 &$undo_trailing_blanks;
680 $something_done_p = 1;
683 # FIXME: Do the same ugly hack for the last token being a ( or [
685 && $structure[$#structure]->type == TmplTokenType::TEXT
686 && $structure[$#structure]->string =~ /^[\(\[]$/) { # not )]
687 push @{$this->{_queue}}, [1, pop @structure];
688 &$undo_trailing_blanks;
689 $something_done_p = 1;
691 # FIXME: If the first token is an open tag, but there is no
692 # FIXME: corresponding close tag, "drop the open tag", i.e.,
693 # FIXME: requeue everything for reanalysis, except the frist tag. :-(
695 && $structure[0]->type == TmplTokenType::TAG
696 && $structure[0]->string =~ /^<([a-z0-9]+)/is
697 && (my $tag = $1) !~ /^(?:br|hr|img|input)\b/is
699 my $tag_open_count = 1;
700 for (my $i = 1; $i <= $#structure; $i += 1) {
701 if ($structure[$i]->type == TmplTokenType::TAG) {
702 if ($structure[$i]->string =~ /^<(\/?)$tag\b/is) {
703 $tag_open_count += ($1? -1: +1);
707 if ($tag_open_count > 0) {
708 for (my $i = $#structure; $i; $i -= 1) {
709 push @{$this->{_queue}}, [1, pop @structure];
711 $something_done_p = 1;
714 # FIXME: If the first token is an open tag, the last token is the
715 # FIXME: corresponding close tag, and there are no other close tags
716 # FIXME: inbetween, requeue the tokens from the second token on,
717 # FIXME: flagged as ok for re-analysis
719 && $structure[0]->type == TmplTokenType::TAG
720 && $structure[0]->string =~ /^<([a-z0-9]+)/is && (my $tag = $1)
721 && $structure[$#structure]->type == TmplTokenType::TAG
722 && $structure[$#structure]->string =~ /^<\/$1\s*>$/is) {
723 my $has_other_open_or_close_tags_p = 0;
724 for (my $i = 1; $i < $#structure; $i += 1) {
725 $has_other_open_or_close_tags_p = 1
726 if $structure[$i]->type == TmplTokenType::TAG
727 && $structure[$i]->string =~ /^<\/?$tag\b/is;
728 last if $has_other_open_or_close_tags_p;
730 if (!$has_other_open_or_close_tags_p) {
731 for (my $i = $#structure; $i; $i -= 1) {
732 push @{$this->{_queue}}, [1, pop @structure];
734 $something_done_p = 1;
737 last if !$something_done_p;
742 sub looks_plausibly_like_groupable_text_p (@) {
744 # The text would look plausibly groupable if all open tags are also closed.
747 for (my $i = 0; $i <= $#structure; $i += 1) {
748 if ($structure[$i]->type == TmplTokenType::TAG) {
749 my $form = $structure[$i]->string;
750 if ($form =~ /^<([A-Z0-9]+)/is) {
752 if ($tag !~ /^(?:br|input)$/is && $form !~ /\/>$/is) {
755 } elsif ($form =~ /^<\/([A-Z0-9]+)/is) {
756 if (@tags && lc($1) eq $tags[$#tags]) {
762 } elsif ($structure[$i]->type != TmplTokenType::TEXT) {
767 return !$error_p && !@tags;
772 my $h = $this->_handle;
774 $this->{_queue} = [] unless defined $this->{_queue};
776 # Elements in the queue are ordered pairs. The first in the ordered pair
777 # specifies whether we are allowed to reanalysis; the second is the token.
778 if (@{$this->{_queue}} && !$this->{_queue}->[$#{$this->{_queue}}]->[0]) {
779 $it = (pop @{$this->{_queue}})->[1];
781 if (@{$this->{_queue}}) {
782 $it = (pop @{$this->{_queue}})->[1];
784 $it = $this->_next_token_intermediate($h);
786 if (!$this->cdata_mode_p && $this->allow_cformat_p && defined $it
787 && ($it->type == TmplTokenType::TEXT?
788 !blank_p( $it->string ): _token_groupable1_p( $it ))) {
789 my @structure = ( $it );
792 my($nonblank_text_p, $parametrized_p, $with_anchor_p, $with_input_p) = (0, 0, 0, 0);
793 if ($it->type == TmplTokenType::TEXT) {
794 $nonblank_text_p = 1 if !blank_p( $it->string );
795 } elsif ($it->type == TmplTokenType::DIRECTIVE) {
797 } elsif ($it->type == TmplTokenType::TAG && $it->string =~ /^<([A-Z0-9]+)/is) {
799 push @tags, $tag if $tag !~ /^(?:br|input)$/i;
800 $with_anchor_p = 1 if $tag eq 'a';
801 $with_input_p = 1 if $tag eq 'input';
803 # We hate | and || in msgid strings, so we try to avoid them
804 for (my $i = 1, my $quit_p = 0, my $quit_next_p = ($it->type == TmplTokenType::TEXT && $it->string =~ /^\|+$/s);; $i += 1) {
805 if (@{$this->{_queue}}) {
806 $next = (pop @{$this->{_queue}})->[1];
808 $next = $this->_next_token_intermediate($h);
810 push @structure, $next; # for consistency (with initialization)
811 last unless defined $next && _token_groupable2_p( $next );
812 last if $quit_next_p;
813 if ($next->type == TmplTokenType::TEXT) {
814 $nonblank_text_p = 1 if !blank_p( $next->string );
815 $quit_p = 1 if $next->string =~ /^\|+$/s; # We hate | and ||
816 } elsif ($next->type == TmplTokenType::DIRECTIVE) {
818 } elsif ($next->type == TmplTokenType::TAG) {
819 if ($next->string =~ /^<([A-Z0-9]+)/is) {
821 push @tags, $tag if $tag !~ /^(?:br|input)$/i;
822 $with_anchor_p = 1 if $tag eq 'a';
823 $with_input_p = 1 if $tag eq 'input';
824 } elsif ($next->string =~ /^<\/([A-Z0-9]+)/is) {
826 $quit_p = 1 unless @tags && $close eq $tags[$#tags];
827 $quit_next_p = 1 if $close =~ /^h\d$/;
833 # Undo the last token, allowing reanalysis
834 push @{$this->{_queue}}, [1, pop @structure];
835 # Simply it a bit more
836 @structure = $this->_optimize( @structure );
837 if (@structure < 2) {
840 } elsif ($nonblank_text_p && ($parametrized_p || $with_anchor_p || $with_input_p)) {
841 # Create the corresponding c-format string
842 my $string = join('', map { $_->string } @structure);
843 my $form = join('', map { _formalize $_ } @structure);
844 my($a_counter, $input_counter) = (0, 0);
845 $form =~ s/<a>/ $a_counter += 1, "<a$a_counter>" /egs;
846 $form =~ s/<input>/ $input_counter += 1, "<input$input_counter>" /egs;
847 $it = TmplToken->new($string, TmplTokenType::TEXT_PARAMETRIZED,
848 $it->line_number, $it->pathname);
849 $it->set_form( $form );
850 $it->set_children( @structure );
851 } elsif ($nonblank_text_p
852 && looks_plausibly_like_groupable_text_p( @structure )
853 && $structure[$#structure]->type == TmplTokenType::TEXT) {
854 # Combine the strings
855 my $string = join('', map { $_->string } @structure);
856 $it = TmplToken->new($string, TmplTokenType::TEXT,
857 $it->line_number, $it->pathname);;
859 # Requeue the tokens thus seen for re-emitting, allow reanalysis
861 push @{$this->{_queue}}, [1, pop @structure];
864 $it = (pop @{$this->{_queue}})->[1];
868 if (defined $it && $it->type == TmplTokenType::TEXT) {
869 my $form = string_canon $it->string;
870 $it->set_form( $form );
875 ###############################################################################
877 # Other simple functions (These are not methods)
881 return $s =~ /^(?:\s|\ $re_end_entity|$re_tmpl_var|$re_xsl)*$/os;
888 $s =~ s/^(\s|\ $re_end_entity)+//os; my $l1 = $l0 - length $s;
889 $s =~ s/(\s|\ $re_end_entity)+$//os; my $l2 = $l0 - $l1 - length $s;
890 return wantarray? (substr($s0, 0, $l1), $s, substr($s0, $l0 - $l2)): $s;
895 # Locale::PO->quote is buggy, it doesn't quote newlines :-/
896 $s =~ s/([\\"])/\\\1/gs;
898 #$s =~ s/[\177-\377]/ sprintf("\\%03o", ord($&)) /egs;
902 # Some functions that shouldn't be here... should be moved out some time
903 sub parametrize ($$$$) {
904 my($fmt_0, $cformat_p, $t, $f) = @_;
907 my @params = $t->parameters_and_fields;
908 for (my $n = 0, my $fmt = $fmt_0; length $fmt;) {
909 if ($fmt =~ /^[^%]+/) {
912 } elsif ($fmt =~ /^%%/) {
915 } elsif ($fmt =~ /^%(?:(\d+)\$)?(?:(\d+)(?:\.(\d+))?)?s/s) {
917 my($i, $width, $prec) = ((defined $1? $1: $n), $2, $3);
919 if (defined $width && defined $prec && !$width && !$prec) {
921 } elsif (defined $params[$i - 1]) {
922 my $param = $params[$i - 1];
923 warn_normal "$fmt_0: $&: Expected a TMPL_VAR, but found a "
924 . $param->type->to_string . "\n", undef
925 if $param->type != TmplTokenType::DIRECTIVE;
926 warn_normal "$fmt_0: $&: Unsupported "
927 . "field width or precision\n", undef
928 if defined $width || defined $prec;
929 warn_normal "$fmt_0: $&: Parameter $i not known", undef
930 unless defined $param;
931 $it .= defined $f? &$f( $param ): $param->string;
933 } elsif ($fmt =~ /^%(?:(\d+)\$)?(?:(\d+)(?:\.(\d+))?)?([pS])/s) {
935 my($i, $width, $prec, $conv) = ((defined $1? $1: $n), $2, $3, $4);
938 my $param = $params[$i - 1];
939 if (!defined $param) {
940 warn_normal "$fmt_0: $&: Parameter $i not known", undef;
942 if ($param->type == TmplTokenType::TAG
943 && $param->string =~ /^<input\b/is) {
944 my $type = defined $param->attributes?
945 lc($param->attributes->{'type'}->[1]): undef;
947 warn_normal "$fmt_0: $&: Expected type=text, "
948 . "but found type=$type", undef
949 unless $type eq 'text';
950 } elsif ($conv eq 'p') {
951 warn_normal "$fmt_0: $&: Expected type=radio, "
952 . "but found type=$type", undef
953 unless $type eq 'radio';
956 warn_normal "$&: Expected an INPUT, but found a "
957 . $param->type->to_string . "\n", undef
959 warn_normal "$fmt_0: $&: Unsupported "
960 . "field width or precision\n", undef
961 if defined $width || defined $prec;
962 $it .= defined $f? &$f( $param ): $param->string;
964 } elsif ($fmt =~ /^%[^%a-zA-Z]*[a-zA-Z]/) {
967 die "$&: Unknown or unsupported format specification\n"; #XXX
969 die "$&: Completely confused parametrizing\n";#XXX
973 my @anchors = $t->anchors;
974 for (my $n = 0, my $fmt = $it, $it = ''; length $fmt;) {
975 if ($fmt =~ /^(?:(?!<a\d+>).)+/is) {
978 } elsif ($fmt =~ /^<a(\d+)>/is) {
982 my $anchor = $anchors[$i - 1];
983 warn_normal "$&: Anchor $1 not found for msgid \"$fmt_0\"", undef #FIXME
984 unless defined $anchor;
985 $it .= $anchor->string;
987 die "Completely confused decoding anchors: $fmt\n";#XXX
993 sub charset_canon ($) {
995 $charset = uc($charset);
996 $charset = "$1-$2" if $charset =~ /^(ISO|UTF)(\d.*)/i;
997 $charset = 'Big5' if $charset eq 'BIG5'; # "Big5" must be in mixed case
1001 use vars qw( @latin1_utf8 );
1003 "\302\200", "\302\201", "\302\202", "\302\203", "\302\204", "\302\205",
1004 "\302\206", "\302\207", "\302\210", "\302\211", "\302\212", "\302\213",
1005 "\302\214", "\302\215", undef, undef, "\302\220", "\302\221",
1006 "\302\222", "\302\223", "\302\224", "\302\225", "\302\226", "\302\227",
1007 "\302\230", "\302\231", "\302\232", "\302\233", "\302\234", "\302\235",
1008 "\302\236", "\302\237", "\302\240", "\302\241", "\302\242", "\302\243",
1009 "\302\244", "\302\245", "\302\246", "\302\247", "\302\250", "\302\251",
1010 "\302\252", "\302\253", "\302\254", "\302\255", "\302\256", "\302\257",
1011 "\302\260", "\302\261", "\302\262", "\302\263", "\302\264", "\302\265",
1012 "\302\266", "\302\267", "\302\270", "\302\271", "\302\272", "\302\273",
1013 "\302\274", "\302\275", "\302\276", "\302\277", "\303\200", "\303\201",
1014 "\303\202", "\303\203", "\303\204", "\303\205", "\303\206", "\303\207",
1015 "\303\210", "\303\211", "\303\212", "\303\213", "\303\214", "\303\215",
1016 "\303\216", "\303\217", "\303\220", "\303\221", "\303\222", "\303\223",
1017 "\303\224", "\303\225", "\303\226", "\303\227", "\303\230", "\303\231",
1018 "\303\232", "\303\233", "\303\234", "\303\235", "\303\236", "\303\237",
1019 "\303\240", "\303\241", "\303\242", "\303\243", "\303\244", "\303\245",
1020 "\303\246", "\303\247", "\303\250", "\303\251", "\303\252", "\303\253",
1021 "\303\254", "\303\255", "\303\256", "\303\257", "\303\260", "\303\261",
1022 "\303\262", "\303\263", "\303\264", "\303\265", "\303\266", "\303\267",
1023 "\303\270", "\303\271", "\303\272", "\303\273", "\303\274", "\303\275",
1024 "\303\276", "\303\277" );
1026 sub charset_convert ($$$) {
1027 my($s, $charset_in, $charset_out) = @_;
1028 if ($s !~ /[\200-\377]/s) { # FIXME: don't worry about iso2022 for now
1030 } elsif ($charset_in eq 'ISO-8859-1' && $charset_out eq 'UTF-8') {
1031 $s =~ s/[\200-\377]/ $latin1_utf8[ord($&) - 128] /egs;
1032 } elsif ($charset_in ne $charset_out) {
1033 VerboseWarnings::warn_normal "conversion from $charset_in to $charset_out is not supported\n", undef;
1038 ###############################################################################
1042 In addition to the basic scanning, this class will also perform
1049 Emulation of c-format strings (see below)
1053 Display of warnings for certain things that affects either the
1054 ability of this class to yield correct output, or things that
1055 are known to cause the original template to cause trouble.
1059 Automatic correction of some of the things warned about
1060 (e.g., SGML "closed start tag" notation).
1064 =head2 c-format strings emulation
1066 Because English word order is not universal, a simple extraction
1067 of translatable strings may yield some strings like "Accounts for"
1068 or ambiguous strings like "in". This makes the resulting strings
1069 difficult to translate, but does not affect all languages alike.
1070 For example, Chinese (with a somewhat different word order) would
1071 be hit harder, but French would be relatively unaffected.
1073 To overcome this problem, the scanner can be configured to detect
1074 patterns with <TMPL_VAR> directives (as well as certain HTML tags),
1075 and try to construct a larger pattern that will appear in the PO
1076 file as c-format strings with %s placeholders. This additional
1077 step allows the translator to deal with cases where word order
1078 is different (replacing %s with %1$s, %2$s, etc.), or when certain
1079 words will require certain inflectional suffixes in sentences.
1081 Because this is an incompatible change, this mode must be explicitly
1082 turned on using the set_cformat(1) method call.
1084 =head2 The flag characters
1086 The character % is followed by zero or more of the following flags:
1092 The value comes from HTML <INPUT> elements.
1093 This abuse of the flag character is somewhat reasonable,
1094 since TMPL_VAR and INPUT are both variables, but of different kinds.
1098 =head2 The field width and precision
1100 An optional 0.0 can be specified for %s to specify
1101 that the <TMPL_VAR> should be suppressed.
1103 =head2 The conversion specifier
1109 Specifies any input field that is neither text nor hidden
1110 (which currently mean radio buttons).
1111 The p conversion specifier is chosen because this does not
1112 evoke any certain sensible data type.
1116 Specifies a text input field (<INPUT TYPE=TEXT>).
1117 This use of the S conversion specifier is somewhat reasonable,
1118 since text input fields contain values of undeterminable type,
1119 which can be treated as strings.
1123 Specifies a <TMPL_VAR>.
1124 This use of the o conversion specifier is somewhat reasonable,
1125 since <TMPL_VAR> denotes values of undeterminable type, which
1126 can be treated as strings.
1132 There is no code to save the tag name anywhere in the scanned token.
1134 The use of <AI<i>> to stand for the I<i>th anchor
1135 is not very well thought out.
1136 Some abuse of c-format specifies might have been more appropriate.
1140 This tokenizer is mostly based
1141 on Ambrose's hideous Perl script known as subst.pl.