2 # This file is part of Koha
3 # Parts copyright 2003-2004 Paul Poulain
4 # Parts copyright 2003-2004 Jerome Vizcaino
5 # Parts copyright 2004 Ambrose Li
9 tmpl_process3.pl - Experimental version of tmpl_process.pl
10 using gettext-compatible translation files
17 use File::Temp qw( :POSIX );
19 use VerboseWarnings qw( error_normal warn_normal );
21 ###############################################################################
23 use vars qw( @in_files $in_dir $str_file $out_dir );
24 use vars qw( @excludes $exclude_regex );
25 use vars qw( $recursive_p );
26 use vars qw( $pedantic_p );
28 use vars qw( $type ); # file extension (DOS form without the dot) to match
29 use vars qw( $charset_in $charset_out );
31 ###############################################################################
33 sub find_translation ($) {
37 $key = TmplTokenizer::string_canon($key);
38 $key = TmplTokenizer::charset_convert($key, $charset_in, $charset_out);
39 $key = TmplTokenizer::quote_po($key);
41 return defined $href->{$key}
42 && !$href->{$key}->fuzzy
43 && length Locale::PO->dequote($href->{$key}->msgstr)?
44 Locale::PO->dequote($href->{$key}->msgstr): $s;
47 sub text_replace_tag ($$) {
50 # value [tag=input], meta
51 my $tag = lc($1) if $t =~ /^<(\S+)/s;
53 for my $a ('alt', 'content', 'title', 'value') {
55 next if $a eq 'content' && $tag ne 'meta';
56 next if $a eq 'value' && ($tag ne 'input'
57 || (ref $attr->{'type'} && $attr->{'type'}->[1] =~ /^(?:hidden|radio)$/)); # FIXME
58 my($key, $val, $val_orig, $order) = @{$attr->{$a}}; #FIXME
60 my $s = find_translation($val);
61 if ($attr->{$a}->[1] ne $s) { #FIXME
62 $attr->{$a}->[1] = $s; # FIXME
63 $attr->{$a}->[2] = ($s =~ /"/s)? "'$s'": "\"$s\""; #FIXME
72 sprintf(' %s=%s', $_, $attr->{$_}->[2]) #FIXME
74 $attr->{$a}->[3] <=> $attr->{$b}->[3] #FIXME
83 sub text_replace (**) {
86 my $s = TmplTokenizer::next_token $h;
87 last unless defined $s;
88 my($kind, $t, $attr) = ($s->type, $s->string, $s->attributes);
89 if ($kind eq TmplTokenType::TEXT) {
90 print $output find_translation($t);
91 } elsif ($kind eq TmplTokenType::TEXT_PARAMETRIZED) {
92 my $fmt = find_translation($s->form);
93 print $output TmplTokenizer::parametrize($fmt, [ map {
94 my($kind, $t, $attr) = ($_->type, $_->string, $_->attributes);
95 $kind == TmplTokenType::TAG && %$attr?
96 text_replace_tag($t, $attr): $t } $s->parameters ], [ $s->anchors ]);
97 } elsif ($kind eq TmplTokenType::TAG && %$attr) {
98 print $output text_replace_tag($t, $attr);
99 } elsif (defined $t) {
106 my($dir, $type) = @_;
108 if (opendir(DIR, $dir)) {
109 my @dirent = readdir DIR; # because DIR is shared when recursing
111 for my $dirent (@dirent) {
112 my $path = "$dir/$dirent";
113 if ($dirent =~ /^\./ || $dirent eq 'CVS' || $dirent eq 'RCS'
114 || (defined $exclude_regex && $dirent =~ /^(?:$exclude_regex)$/)) {
117 push @it, $path if !defined $type || $dirent =~ /\.(?:$type)$/;
118 } elsif (-d $path && $recursive_p) {
119 push @it, listfiles($path, $type);
123 warn_normal "$dir: $!", undef;
128 ###############################################################################
130 sub usage_error (;$) {
131 for my $msg (split(/\n/, $_[0])) {
132 print STDERR "$msg\n";
134 print STDERR "Try `$0 --help' for more information.\n";
138 ###############################################################################
141 'input|i=s' => \@in_files,
142 'outputdir|o=s' => \$out_dir,
143 'recursive|r' => \$recursive_p,
144 'str-file|s=s' => \$str_file,
145 'exclude|x=s' => \@excludes,
146 'pedantic-warnings|pedantic' => sub { $pedantic_p = 1 },
149 VerboseWarnings::set_application_name $0;
150 VerboseWarnings::set_pedantic_mode $pedantic_p;
152 # try to make sure .po files are backed up (see BUGS)
153 $ENV{VERSION_CONTROL} = 't';
155 # keep the buggy Locale::PO quiet if it says stupid things
156 $SIG{__WARN__} = sub {
158 print STDERR $s unless $s =~ /^Strange line in [^:]+: #~/s
161 my $action = shift or usage_error('You must specify an ACTION.');
162 usage_error('You must at least specify input and string list filenames.')
163 if !@in_files || !defined $str_file;
165 # Type match defaults to *.tmpl plus *.inc if not specified
166 $type = "tmpl|inc" if !defined($type);
168 # Check the inputs for being files or directories
169 for my $input (@in_files) {
170 usage_error("$input: Input must be a file or directory.\n"
171 . "(Symbolic links are not supported at the moment)")
172 unless -d $input || -f $input;;
175 # Generates the global exclude regular expression
176 $exclude_regex = '(?:'.join('|', @excludes).')' if @excludes;
178 # Generate the list of input files if a directory is specified
179 if (-d $in_files[0]) {
180 die "If you specify a directory as input, you must specify only it.\n"
183 # input is a directory, generates list of files to process
184 $in_dir = $in_files[0];
185 $in_dir =~ s/\/$//; # strips the trailing / if any
186 @in_files = listfiles($in_dir, $type);
188 for my $input (@in_files) {
189 die "You cannot specify input files and directories at the same time.\n"
194 # restores the string list from file
195 $href = Locale::PO->load_file_ashash($str_file);
197 # guess the charsets. HTML::Templates defaults to iso-8859-1
199 $charset_out = TmplTokenizer::charset_canon $2
200 if $href->{'""'}->msgstr =~ /\bcharset=(["']?)([^;\s"'\\]+)\1/;
201 for my $msgid (keys %$href) {
202 if ($msgid =~ /\bcharset=(["']?)([^;\s"'\\]+)\1/) {
203 my $candidate = TmplTokenizer::charset_canon $2;
204 die "Conflicting charsets in msgid: $charset_in vs $candidate\n"
205 if defined $charset_in && $charset_in ne $candidate;
206 $charset_in = $candidate;
210 if (!defined $charset_in) {
211 $charset_in = TmplTokenizer::charset_canon 'iso8859-1';
212 warn "Warning: Can't determine original templates' charset, defaulting to $charset_in\n";
215 if ($action eq 'create') {
216 # updates the list. As the list is empty, every entry will be added
217 die "$str_file: Output file already exists" if -f $str_file;
218 my($tmph, $tmpfile) = tmpnam();
219 # Generate the temporary file that acts as <MODULE>/POTFILES.in
220 for my $input (@in_files) {
221 print $tmph "$input\n";
224 # Generate the specified po file ($str_file)
225 system ('xgettext.pl', '-s', '-f', $tmpfile, '-o', $str_file);
226 unlink $tmpfile || warn_normal "$tmpfile: unlink failed: $!\n", undef;
228 } elsif ($action eq 'update') {
229 my($tmph1, $tmpfile1) = tmpnam();
230 my($tmph2, $tmpfile2) = tmpnam();
231 close $tmph2; # We just want a name
232 # Generate the temporary file that acts as <MODULE>/POTFILES.in
233 for my $input (@in_files) {
234 print $tmph1 "$input\n";
237 # Generate the temporary file that acts as <MODULE>/<LANG>.pot
238 system('./xgettext.pl', '-s', '-f', $tmpfile1, '-o', $tmpfile2,
239 (defined $charset_in? ('-I', $charset_in): ()),
240 (defined $charset_out? ('-O', $charset_out): ()));
241 # Merge the temporary "pot file" with the specified po file ($str_file)
242 # FIXME: msgmerge(1) is a Unix dependency
243 # FIXME: need to check the return value
244 system('msgmerge', '-U', '-s', $str_file, $tmpfile2);
245 unlink $tmpfile1 || warn_normal "$tmpfile1: unlink failed: $!\n", undef;
246 unlink $tmpfile2 || warn_normal "$tmpfile2: unlink failed: $!\n", undef;
248 } elsif ($action eq 'install') {
249 if(!defined($out_dir)) {
250 usage_error("You must specify an output directory when using the install method.");
253 if ($in_dir eq $out_dir) {
254 warn "You must specify a different input and output directory.\n";
258 # Make sure the output directory exists
259 # (It will auto-create it, but for compatibility we should not)
260 -d $out_dir || die "$out_dir: The directory does not exist\n";
262 # Try to open the file, because Locale::PO doesn't check :-/
263 open(INPUT, "<$str_file") || die "$str_file: $!\n";
266 # creates the new tmpl file using the new translation
267 for my $input (@in_files) {
268 die "Assertion failed"
269 unless substr($input, 0, length($in_dir) + 1) eq "$in_dir/";
271 my $h = TmplTokenizer->new( $input );
272 $h->set_allow_cformat( 1 );
273 VerboseWarnings::set_input_file_name $input;
275 my $target = $out_dir . substr($input, length($in_dir));
276 my $targetdir = $` if $target =~ /[^\/]+$/s;
277 if (!-d $targetdir) {
278 print STDERR "Making directory $targetdir...";
279 # creates with rwxrwxr-x permissions
280 mkdir($targetdir, 0775) || warn_normal "$targetdir: $!", undef;
282 print STDERR "Creating $target...\n";
283 open( OUTPUT, ">$target" ) || die "$target: $!\n";
284 text_replace( $h, *OUTPUT );
289 usage_error('Unknown action specified.');
293 ###############################################################################
297 ./tmpl_process3.pl [ I<tmpl_process.pl options> ]
301 This is an experimental version of the tmpl_process.pl script,
302 using standard gettext-style PO files. Note that the behaviour
303 of this script should still be considered unstable.
305 Currently, the create, update, and install actions have all been
306 reimplemented and seem to work.
308 The create action calls xgettext.pl to do the actual work;
309 the update action calls xgettext.pl and msgmerge(1) to do the
312 The script can detect <TMPL_VAR> directives embedded inside what
313 appears to be a full sentence (this actual work being done by
314 TmplTokenizer(3)); these larger patterns appear in the translation
315 file as c-format strings with %s.
317 Whitespace in extracted strings are folded to single blanks, in
318 order to prevent new strings from appearing when minor changes in
319 the original templates occur, and to prevent overly difficult to
320 read strings in the PO file.
324 The --help option has not been implemented yet.
326 xgettext.pl must be present in the current directory; the
327 msgmerge(1) command must also be present in the search path.
328 The script currently does not check carefully whether these
329 dependent commands are present.
331 If xgettext.pl is interrupted by the user, a corrupted po file
332 will result. This is very seriously wrong.
334 Locale::PO(3) has a lot of bugs. It can neither parse nor
335 generate GNU PO files properly; a couple of workarounds have
336 been written in TmplTokenizer and more is likely to be needed
337 (e.g., to get rid of the "Strange line" warning for #~).
339 There are probably some other bugs too, since this has not been