4 #use warnings; FIXME - Bug 2505
8 use VerboseWarnings qw( pedantic_p error_normal warn_normal warn_pedantic );
11 use vars qw($VERSION @ISA @EXPORT @EXPORT_OK %EXPORT_TAGS);
13 ###############################################################################
17 TmplTokenizer.pm - Simple-minded wrapper class for TTParser
21 A wrapper for the functionality found in TTParser to allow an easier transition to Template Toolkit
25 ###############################################################################
32 use vars qw( $pedantic_attribute_error_in_nonpedantic_mode_p );
33 use vars qw( $pedantic_tmpl_var_use_in_nonpedantic_mode_p );
34 use vars qw( $pedantic_error_markup_in_pcdata_p );
36 ###############################################################################
39 use vars qw( $re_xsl $re_end_entity $re_tmpl_var);
41 $re_tmpl_var = q{\[%\s*[get|set|default]?\s*[\w\.]+\s*[|.*?]?\s*%\]};
42 $re_xsl = q{<\/?(?:xsl:)(?:[\s\-a-zA-Z0-9"'\/\.\[\]\@\(\):=,$]+)\/?>};
43 $re_end_entity = '(?:;|$|(?=\s))'; # semicolon or before-whitespace
45 # End of the hideous stuff
47 use vars qw( $serial );
49 ###############################################################################
51 sub FATAL_P () {'fatal-p'}
52 sub SYNTAXERROR_P () {'syntaxerror-p'}
54 sub FILENAME () {'input'}
55 #sub HANDLE () {'handle'}
57 #sub READAHEAD () {'readahead'}
58 sub LINENUM_START () {'lc_0'}
60 sub CDATA_MODE_P () {'cdata-mode-p'}
61 sub CDATA_CLOSE () {'cdata-close'}
62 #sub PCDATA_MODE_P () {'pcdata-mode-p'} # additional submode for CDATA
63 sub JS_MODE_P () {'js-mode-p'} # cdata-mode-p must also be true
65 sub ALLOW_CFORMAT_P () {'allow-cformat-p'}
70 #open my $handle,$filename or die "can't open $filename";
71 my $parser = TTParser->new;
72 $parser->build_tokens( $filename );
74 filename => $filename,
81 ###############################################################################
87 return $this->{filename};
92 return $this->{+FATAL_P};
95 # work around, currently not implemented
98 # return $this->{+SYNTAXERROR_P};
104 return $this->{+JS_MODE_P};
107 sub allow_cformat_p {
109 return $this->{+ALLOW_CFORMAT_P};
116 $this->{+FATAL_P} = $_[0];
122 $this->{+JS_MODE_P} = $_[0];
126 #used in xgettext, tmpl_process3 and text-extract2
127 sub set_allow_cformat {
129 $this->{+ALLOW_CFORMAT_P} = $_[0];
133 ###############################################################################
135 use vars qw( $js_EscapeSequence );
137 # Perl quoting is really screwed up, but this common subexp is way too long
138 $js_EscapeSequence = q{\\\\(?:['"\\\\bfnrt]|[^0-7xu]|[0-3]?[0-7]{1,2}|x[\da-fA-F]{2}|u[\da-fA-F]{4})};
140 sub parenleft () { '(' }
141 sub parenright () { ')' }
147 if ($s0 =~ /^\s+/s) { # whitespace
150 } elsif ($s0 =~ /^\/\/[^\r\n]*(?:[\r\n]|$)/s) { # C++-style comment
153 } elsif ($s0 =~ /^\/\*(?:(?!\*\/).)*\*\//s) { # C-style comment
156 # Keyword or identifier, ECMA-262 p.13 (section 7.5)
157 } elsif ($s0 =~ /^[A-Z_\$][A-Z\d_\$]*/is) { # IdentifierName
160 # Punctuator, ECMA-262 p.13 (section 7.6)
161 } elsif ($s0 =~ /^(?:[\(\){}\[\];]|>>>=|<<=|>>=|[-\+\*\/\&\|\^\%]=|>>>|<<|>>|--|\+\+|\|\||\&\&|==|<=|>=|!=|[=><,!~\?:\.\-\+\*\/\&\|\^\%])/s) {
164 # DecimalLiteral, ECMA-262 p.14 (section 7.7.3); note: bug in the spec
165 } elsif ($s0 =~ /^(?:0|[1-9]\d+(?:\.\d*(?:[eE][-\+]?\d+)?)?)/s) {
168 # HexIntegerLiteral, ECMA-262 p.15 (section 7.7.3)
169 } elsif ($s0 =~ /^0[xX][\da-fA-F]+/s) {
172 # OctalIntegerLiteral, ECMA-262 p.15 (section 7.7.3)
173 } elsif ($s0 =~ /^0[\da-fA-F]+/s) {
176 # StringLiteral, ECMA-262 p.17 (section 7.7.4)
177 # XXX SourceCharacter doesn't seem to be defined (?)
178 } elsif ($s0 =~ /^(?:"(?:(?!["\\\r\n]).|$js_EscapeSequence)*"|'(?:(?!['\\\r\n]).|$js_EscapeSequence)*')/os) {
181 } elsif ($s0 =~ /^./) { # UNKNOWN TOKEN !!!
189 sub STATE_UNDERSCORE () { 1 }
190 sub STATE_PARENLEFT () { 2 }
191 sub STATE_STRING_LITERAL () { 3 }
193 # XXX This is a crazy hack. I don't want to write an ECMAScript parser.
194 # XXX A scanner is one thing; a parser another thing.
195 sub _identify_js_translatables (@) {
198 # We mark a JavaScript translatable string as in C, i.e., _("literal")
199 # For simplicity, we ONLY look for "_" "(" StringLiteral ")"
200 for (my $i = 0, my $state = 0, my($j, $q, $s); $i <= $#input; $i += 1) {
201 my $reset_state_p = 0;
202 push @output, [0, $input[$i]];
203 if ($input[$i] !~ /\S/s) {
205 } elsif ($state == 0) {
206 $state = STATE_UNDERSCORE if $input[$i] eq '_';
207 } elsif ($state == STATE_UNDERSCORE) {
208 $state = $input[$i] eq parenleft ? STATE_PARENLEFT : 0;
209 } elsif ($state == STATE_PARENLEFT) {
210 if ($input[$i] =~ /^(['"])(.*)\1$/s) {
211 ($state, $j, $q, $s) = (STATE_STRING_LITERAL, $#output, $1, $2);
215 } elsif ($state == STATE_STRING_LITERAL) {
216 if ($input[$i] eq parenright) {
217 $output[$j] = [1, $output[$j]->[1], $q, $s];
221 die "identify_js_translatables internal error: Unknown state $state"
227 ###############################################################################
229 sub string_canon ($) {
231 # Fold all whitespace into single blanks
236 # safer version used internally, preserves new lines
237 sub string_canon_safe ($) {
239 # fold tabs and spaces into single spaces
240 $s =~ s/[\ \t]+/ /gs;
251 sub _formalize_string_cformat{
253 return _quote_cformat( string_canon_safe $s );
258 if( $t->type == TmplTokenType::DIRECTIVE ){
260 } elsif( $t->type == TmplTokenType::TEXT ){
261 return _formalize_string_cformat( $t->string );
262 } elsif( $t->type == TmplTokenType::TAG ){
263 if( $t->string =~ m/^a\b/is ){
265 } elsif( $t->string =~ m/^input\b/is ){
266 if( lc $t->attributes->{'type'}->[1] eq 'text' ){
272 return _quote_cformat $t->string;
275 return _quote_cformat $t->string;
279 # internal parametization, used within next_token
280 # method that takes in an array of TEXT and DIRECTIVE tokens (DIRECTIVEs must be GET) and return a TmplTokenType::TEXT_PARAMETRIZED
281 sub _parametrize_internal{
285 # for my $item (@parts){
286 # if( $item->type == TmplTokenType::TEXT ){
287 # $s .= $item->string;
289 # #must be a variable directive
293 my $s = join( "", map { _formalize $_ } @parts );
294 # should both the string and form be $s? maybe only the later? posibly the former....
295 # used line number from first token, should suffice
296 my $t = TmplToken->new( $s, TmplTokenType::TEXT_PARAMETRIZED, $parts[0]->line_number, $this->filename );
297 $t->set_children(@parts);
305 # parts that make up a text_parametrized (future children of the token)
308 $next = $self->{_parser}->next_token;
309 return undef unless defined $next;
310 # if cformat mode is off, dont bother parametrizing, just return them as they come
311 return $next unless $self->allow_cformat_p;
312 if( $next->type == TmplTokenType::TEXT ){
314 } elsif( $next->type == TmplTokenType::DIRECTIVE && $next->string =~ m/\[%\s*\w+\s*%\]/ ){
317 # if there is nothing in parts, return this token
318 return $next unless @parts;
319 # OTHERWISE, put this token back and return the parametrized string of @parts
320 $self->{_parser}->unshift_token($next);
321 return $self->_parametrize_internal(@parts);
326 ###############################################################################
328 # function taken from old version
329 # used by tmpl_process3
330 sub parametrize ($$$$) {
331 my($fmt_0, $cformat_p, $t, $f) = @_;
334 my @params = $t->parameters_and_fields;
335 for (my $n = 0, my $fmt = $fmt_0; length $fmt;) {
336 if ($fmt =~ /^[^%]+/) {
339 } elsif ($fmt =~ /^%%/) {
342 } elsif ($fmt =~ /^%(?:(\d+)\$)?(?:(\d+)(?:\.(\d+))?)?s/s) {
344 my($i, $width, $prec) = ((defined $1? $1: $n), $2, $3);
346 if (defined $width && defined $prec && !$width && !$prec) {
348 } elsif (defined $params[$i - 1]) {
349 my $param = $params[$i - 1];
350 warn_normal "$fmt_0: $&: Expected a TMPL_VAR, but found a "
351 . $param->type->to_string . "\n", undef
352 if $param->type != TmplTokenType::DIRECTIVE;
353 warn_normal "$fmt_0: $&: Unsupported "
354 . "field width or precision\n", undef
355 if defined $width || defined $prec;
356 warn_normal "$fmt_0: $&: Parameter $i not known", undef
357 unless defined $param;
358 $it .= defined $f? &$f( $param ): $param->string;
360 } elsif ($fmt =~ /^%(?:(\d+)\$)?(?:(\d+)(?:\.(\d+))?)?([pS])/s) {
362 my($i, $width, $prec, $conv) = ((defined $1? $1: $n), $2, $3, $4);
365 my $param = $params[$i - 1];
366 if (!defined $param) {
367 warn_normal "$fmt_0: $&: Parameter $i not known", undef;
369 if ($param->type == TmplTokenType::TAG
370 && $param->string =~ /^<input\b/is) {
371 my $type = defined $param->attributes?
372 lc($param->attributes->{'type'}->[1]): undef;
374 warn_normal "$fmt_0: $&: Expected type=text, "
375 . "but found type=$type", undef
376 unless $type eq 'text';
377 } elsif ($conv eq 'p') {
378 warn_normal "$fmt_0: $&: Expected type=radio, "
379 . "but found type=$type", undef
380 unless $type eq 'radio';
383 warn_normal "$&: Expected an INPUT, but found a "
384 . $param->type->to_string . "\n", undef
386 warn_normal "$fmt_0: $&: Unsupported "
387 . "field width or precision\n", undef
388 if defined $width || defined $prec;
389 $it .= defined $f? &$f( $param ): $param->string;
391 } elsif ($fmt =~ /^%[^%a-zA-Z]*[a-zA-Z]/) {
394 die "$&: Unknown or unsupported format specification\n"; #XXX
396 die "$&: Completely confused parametrizing\n";#XXX
400 my @anchors = $t->anchors;
401 for (my $n = 0, my $fmt = $it, $it = ''; length $fmt;) {
402 if ($fmt =~ /^(?:(?!<a\d+>).)+/is) {
405 } elsif ($fmt =~ /^<a(\d+)>/is) {
409 my $anchor = $anchors[$i - 1];
410 warn_normal "$&: Anchor $1 not found for msgid \"$fmt_0\"", undef #FIXME
411 unless defined $anchor;
412 $it .= $anchor->string;
414 die "Completely confused decoding anchors: $fmt\n";#XXX
421 # Other simple functions (These are not methods)
425 return $s =~ /^(?:\s|\ $re_end_entity|$re_tmpl_var|$re_xsl)*$/osi;
432 $s =~ s/^(\s|\ $re_end_entity)+//os; my $l1 = $l0 - length $s;
433 $s =~ s/(\s|\ $re_end_entity)+$//os; my $l2 = $l0 - $l1 - length $s;
434 return wantarray? (substr($s0, 0, $l1), $s, substr($s0, $l0 - $l2)): $s;
439 # Locale::PO->quote is buggy, it doesn't quote newlines :-/
440 $s =~ s/([\\"])/\\\1/gs;
442 #$s =~ s/[\177-\377]/ sprintf("\\%03o", ord($&)) /egs;
446 sub charset_canon ($) {
448 $charset = uc($charset);
449 $charset = "$1-$2" if $charset =~ /^(ISO|UTF)(\d.*)/i;
450 $charset = 'Big5' if $charset eq 'BIG5'; # "Big5" must be in mixed case
454 use vars qw( @latin1_utf8 );
456 "\302\200", "\302\201", "\302\202", "\302\203", "\302\204", "\302\205",
457 "\302\206", "\302\207", "\302\210", "\302\211", "\302\212", "\302\213",
458 "\302\214", "\302\215", undef, undef, "\302\220", "\302\221",
459 "\302\222", "\302\223", "\302\224", "\302\225", "\302\226", "\302\227",
460 "\302\230", "\302\231", "\302\232", "\302\233", "\302\234", "\302\235",
461 "\302\236", "\302\237", "\302\240", "\302\241", "\302\242", "\302\243",
462 "\302\244", "\302\245", "\302\246", "\302\247", "\302\250", "\302\251",
463 "\302\252", "\302\253", "\302\254", "\302\255", "\302\256", "\302\257",
464 "\302\260", "\302\261", "\302\262", "\302\263", "\302\264", "\302\265",
465 "\302\266", "\302\267", "\302\270", "\302\271", "\302\272", "\302\273",
466 "\302\274", "\302\275", "\302\276", "\302\277", "\303\200", "\303\201",
467 "\303\202", "\303\203", "\303\204", "\303\205", "\303\206", "\303\207",
468 "\303\210", "\303\211", "\303\212", "\303\213", "\303\214", "\303\215",
469 "\303\216", "\303\217", "\303\220", "\303\221", "\303\222", "\303\223",
470 "\303\224", "\303\225", "\303\226", "\303\227", "\303\230", "\303\231",
471 "\303\232", "\303\233", "\303\234", "\303\235", "\303\236", "\303\237",
472 "\303\240", "\303\241", "\303\242", "\303\243", "\303\244", "\303\245",
473 "\303\246", "\303\247", "\303\250", "\303\251", "\303\252", "\303\253",
474 "\303\254", "\303\255", "\303\256", "\303\257", "\303\260", "\303\261",
475 "\303\262", "\303\263", "\303\264", "\303\265", "\303\266", "\303\267",
476 "\303\270", "\303\271", "\303\272", "\303\273", "\303\274", "\303\275",
477 "\303\276", "\303\277" );
479 sub charset_convert ($$$) {
480 my($s, $charset_in, $charset_out) = @_;
481 if ($s !~ /[\200-\377]/s) { # FIXME: don't worry about iso2022 for now
483 } elsif ($charset_in eq 'ISO-8859-1' && $charset_out eq 'UTF-8') {
484 $s =~ s/[\200-\377]/ $latin1_utf8[ord($&) - 128] /egs;
485 } elsif ($charset_in ne $charset_out) {
486 VerboseWarnings::warn_normal "conversion from $charset_in to $charset_out is not supported\n", undef;
491 ###############################################################################
495 In addition to the basic scanning, this class will also perform
502 Emulation of c-format strings (see below)
506 Display of warnings for certain things that affects either the
507 ability of this class to yield correct output, or things that
508 are known to cause the original template to cause trouble.
512 Automatic correction of some of the things warned about
513 (e.g., SGML "closed start tag" notation).
517 =head2 c-format strings emulation
519 Because English word order is not universal, a simple extraction
520 of translatable strings may yield some strings like "Accounts for"
521 or ambiguous strings like "in". This makes the resulting strings
522 difficult to translate, but does not affect all languages alike.
523 For example, Chinese (with a somewhat different word order) would
524 be hit harder, but French would be relatively unaffected.
526 To overcome this problem, the scanner can be configured to detect
527 patterns with <TMPL_VAR> directives (as well as certain HTML tags),
528 and try to construct a larger pattern that will appear in the PO
529 file as c-format strings with %s placeholders. This additional
530 step allows the translator to deal with cases where word order
531 is different (replacing %s with %1$s, %2$s, etc.), or when certain
532 words will require certain inflectional suffixes in sentences.
534 Because this is an incompatible change, this mode must be explicitly
535 turned on using the set_allow_cformat(1) method call.
537 =head2 The flag characters
539 The character % is followed by zero or more of the following flags:
545 The value comes from HTML <INPUT> elements.
546 This abuse of the flag character is somewhat reasonable,
547 since TMPL_VAR and INPUT are both variables, but of different kinds.
551 =head2 The field width and precision
553 An optional 0.0 can be specified for %s to specify
554 that the <TMPL_VAR> should be suppressed.
556 =head2 The conversion specifier
562 Specifies any input field that is neither text nor hidden
563 (which currently mean radio buttons).
564 The p conversion specifier is chosen because this does not
565 evoke any certain sensible data type.
569 Specifies a text input field (<INPUT TYPE=TEXT>).
570 This use of the S conversion specifier is somewhat reasonable,
571 since text input fields contain values of undeterminable type,
572 which can be treated as strings.
576 Specifies a <TMPL_VAR>.
577 This use of the o conversion specifier is somewhat reasonable,
578 since <TMPL_VAR> denotes values of undeterminable type, which
579 can be treated as strings.
585 There is no code to save the tag name anywhere in the scanned token.
587 The use of <AI<i>> to stand for the I<i>th anchor
588 is not very well thought out.
589 Some abuse of c-format specifies might have been more appropriate.
593 This tokenizer is mostly based
594 on Ambrose's hideous Perl script known as subst.pl.