source: main/trunk/greenstone2/bin/script/gti.pl@ 31751

Last change on this file since 31751 was 31751, checked in by ak19, 7 years ago

Added check for JRE_HOME if no JAVA_HOME

  • Property svn:executable set to *
  • Property svn:keywords set to Author Date Id Revision
File size: 95.8 KB
RevLine 
[10019]1#!/usr/bin/perl -w
2
3###########################################################################
4#
5# gti.pl
6#
7# A component of the Greenstone digital library software
8# from the New Zealand Digital Library Project at the
9# University of Waikato, New Zealand.
10#
11# Copyright (C) 2005 New Zealand Digital Library Project
12#
13# This program is free software; you can redistribute it and/or modify
14# it under the terms of the GNU General Public License as published by
15# the Free Software Foundation; either version 2 of the License, or
16# (at your option) any later version.
17#
18# This program is distributed in the hope that it will be useful,
19# but WITHOUT ANY WARRANTY; without even the implied warranty of
20# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
21# GNU General Public License for more details.
22#
23# You should have received a copy of the GNU General Public License
24# along with this program; if not, write to the Free Software
25# Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
26#
27###########################################################################
28
29
30BEGIN {
31 die "GSDLHOME not set\n" unless defined $ENV{'GSDLHOME'};
32 unshift (@INC, "$ENV{'GSDLHOME'}/perllib");
33}
34
[24627]35
[10019]36use iso639;
37use strict;
38use util;
[29415]39use FileUtils;
[10019]40
41my $gsdl_root_directory = "$ENV{'GSDLHOME'}";
42my $gti_log_file = &util::filename_cat($gsdl_root_directory, "etc", "gti.log");
[28976]43my $source_language_code = "en"; # This is non-negotiable
[10019]44
45my $gti_translation_files =
[24627]46[ # Greenstone macrofiles
47{ 'key' => "coredm",
[10019]48 'file_type' => "macrofile",
49 'source_file' => "macros/english.dm",
[11224]50 'target_file' => "macros/{bn:bengali;fa:farsi;gd:gaelic;id:indo;lv:latvian;pt-br:port-br;pt-pt:port-pt;zh-tr:chinese-trad;iso_639_1_target_language_name}.dm" },
[24627]51
52{ 'key' => "auxdm",
[10019]53 'file_type' => "macrofile",
54 'source_file' => "macros/english2.dm",
[11224]55 'target_file' => "macros/{bn:bengali;fa:farsi;gd:gaelic;id:indo;lv:latvian;pt-br:port-br;pt-pt:port-pt;zh-tr:chinese-trad;iso_639_1_target_language_name}2.dm" },
[10019]56
[29411]57#{ 'key' => "paperspastdm",
58# 'file_type' => "macrofile",
59# 'source_file' => "macros/paperspast-english.dm",
60# 'target_file' => "macros/paperspast-{bn:bengali;fa:farsi;gd:gaelic;id:indo;lv:latvian;pt-br:port-br;pt-pt:port-pt;zh-tr:chinese-trad;iso_639_1_target_language_name}.dm" },
[25285]61
[24627]62# GLI dictionary
63{ 'key' => "glidict",
[13946]64 'file_type' => "resource_bundle",
65 'source_file' => "gli/classes/dictionary.properties",
66 'target_file' => "gli/classes/dictionary_{target_language_code}.properties" },
[10114]67
[24627]68# GLI help
69{ 'key' => "glihelp",
[13946]70 'file_type' => "greenstone_xml",
71 'source_file' => "gli/help/en/help.xml",
72 'target_file' => "gli/help/{target_language_code}/help.xml" },
[11602]73
[24627]74# Greenstone Perl modules
75{ 'key' => "perlmodules",
[10019]76 'file_type' => "resource_bundle",
[11634]77 'source_file' => "perllib/strings.properties",
78 'target_file' => "perllib/strings_{target_language_code}.properties" },
[10019]79
[28337]80# Greenstone Installer interface
81{ 'key' => "gsinstaller",
82 'file_type' => "resource_bundle",
83 'source_file' => "gsinstaller/LanguagePack.properties",
84 'target_file' => "gsinstaller/LanguagePack_{target_language_code}.properties" },
85
[24627]86# Greenstone tutorial exercises
87# { 'key' => "tutorials",
88# 'file_type' => "greenstone_xml",
89# 'source_file' => "gsdl-documentation/tutorials/xml-source/tutorial_en.xml",
90# 'target_file' => "gsdl-documentation/tutorials/xml-source/tutorial_{target_language_code}.xml" },
[11366]91
[24627]92# new Greenstone.org
93{ 'key' => "greenorg",
94 'file_type' => "resource_bundle",
95 'source_file' => "greenstoneorg/website/classes/Gsc.properties",
96 'target_file' => "greenstoneorg/website/classes/Gsc_{target_language_code}.properties"
[29411]97},
98
[29412]99# greenstone 3 interface files, from http://svn.greenstone.org/main/trunk/greenstone3/web/WEB-INF/classes
[29413]100# check it out as greenstone3
[29411]101{ 'key' => "gs3interface",
[29415]102 'file_type' => "resource_bundle",
103 'source_file' => "greenstone3",
104 'target_file' => "greenstone3"
[30581]105},
106
107# collection config display items of GS3 demo collections. Checked out as gs3-collection-configs
108# from http://svn.greenstone.org/main/trunk/gs3-collection-configs
109{ 'key' => "gs3colcfg",
110 'file_type' => "resource_bundle",
111 'source_file' => "gs3-collection-configs",
112 'target_file' => "gs3-collection-configs"
[24627]113}
114];
[10019]115
[30844]116my @gs3_col_cfg_files = ("lucene-jdbm-demo", "solr-jdbm-demo", "localsite");
[30581]117
[30490]118my @gs3_interface_files = ("interface_default", "ServiceRack", "metadata_names");
119#"AbstractBrowse", "AbstractGS2FieldSearch", "AbstractSearch", "AbstractTextSearch", "Authentication", "CrossCollectionSearch", "GS2LuceneSearch", "LuceneSearch", "MapRetrieve", "MapSearch", "PhindPhraseBrowse", "SharedSoleneGS2FieldSearch");
[11026]120
[26547]121# Auxilliary GS3 interface files. This list is not used at present
122# Combine with above list if generating translation spreadsheet for all interface files
[30490]123my @gs3_aux_interface_files = ("GATEServices","QBRWebServicesHelp", "Visualizer", "IViaSearch", "GS2Construct");
[18460]124
[30490]125my @gs3_other_interface_files = ("interface_default2", "interface_basic", "interface_basic2", "interface_nzdl", "interface_gs2");
126
[26547]127# Not: i18n, log4j
128
[10019]129sub main
130{
131 # Get the command to process, and any arguments
132 my $gti_command = shift(@_);
133 my @gti_command_arguments = @_;
[18460]134 my $module = $_[1];
[26544]135
136 # for GS3, set gsdl_root_dir to GSDL3HOME
[29411]137 #if($module && $module eq "gs3interface"){ # module is empty when the gti-command is create-glihelp-zip-file
138 #if($ENV{'GSDL3SRCHOME'}) {
139 # $gsdl_root_directory = (defined $ENV{'GSDL3HOME'}) ? $ENV{'GSDL3HOME'} : &util::filename_cat($ENV{'GSDL3SRCHOME'}, "web");
140 # $gti_log_file = &util::filename_cat($gsdl_root_directory, "logs", "gti.log");
141 #}
142 #}
[24627]143
[10019]144 # Open the GTI log file for appending, or write to STDERR if that fails
145 if (!open(GTI_LOG, ">>$gti_log_file")) {
[24627]146 open(GTI_LOG, ">&STDERR");
[10019]147 }
[24627]148
[10019]149 # Log the command that launched this script
150 &log_message("Command: $0 @ARGV");
[24627]151
[10019]152 # Check that a command was supplied
153 if (!$gti_command) {
[24627]154 &throw_fatal_error("Missing command.");
[18460]155 }
[24627]156
[10019]157 # Process the command
[13948]158 if ($gti_command =~ /^get-all-chunks$/i) {
[24627]159 # Check that GS3 interface is the target
[30582]160 if ($module =~ m/^gs3/) { # gs3interface, gs3colcfg
[24627]161 print &get_all_chunks_for_gs3(@gti_command_arguments);
162 } else {
163 print &get_all_chunks(@gti_command_arguments);
164 }
[13948]165 }
[24627]166 elsif ($gti_command =~ /^get-first-n-chunks-requiring-work$/i) {
[30582]167 if ($module =~ m/^gs3/) {
[24627]168 print &get_first_n_chunks_requiring_work_for_gs3(@gti_command_arguments);
169 } else {
170 print &get_first_n_chunks_requiring_work(@gti_command_arguments);
171 }
[10019]172 }
[25249]173 elsif ($gti_command =~ /^get-uptodate-chunks$/i) {
[30582]174 if ($module =~ m/^gs3/) {
[25249]175 print &get_uptodate_chunks_for_gs3(@gti_command_arguments);
176 } else {
177 print &get_uptodate_chunks(@gti_command_arguments);
178 }
179 }
[24627]180 elsif ($gti_command =~ /^get-language-status$/i) {
181 print &get_language_status(@gti_command_arguments);
[10019]182 }
[24627]183 elsif ($gti_command =~ /^search-chunks$/i) {
184 print &search_chunks(@gti_command_arguments);
[10019]185 }
[24627]186 elsif ($gti_command =~ /^submit-translations$/i) {
187 # This command cannot produce any output since it reads input
188 &submit_translations(@gti_command_arguments);
[10019]189 }
[24627]190 elsif ($gti_command =~ /^create-glihelp-zip-file$/i) {
191 # This command cannot produce any output since it reads input
192 &create_glihelp_zip_file(@gti_command_arguments);
[18460]193 }
[24627]194 else {
195 # The command was not recognized
196 &throw_fatal_error("Unknown command \"$gti_command\".");
197 }
[10019]198}
199
200
201sub throw_fatal_error
202{
203 my $error_message = shift(@_);
[24627]204
[10019]205 # Write an XML error response
206 print "<?xml version=\"1.0\" encoding=\"UTF-8\" ?>\n";
207 print "<GTIResponse>\n";
208 print " <GTIError time=\"" . time() . "\">" . $error_message . "</GTIError>\n";
209 print "</GTIResponse>\n";
[24627]210
[10019]211 # Log the error message, then die
212 &log_message("Error: $error_message");
213 die "\n";
214}
215
216
217sub log_message
218{
219 my $log_message = shift(@_);
220 print GTI_LOG time() . " -- " . $log_message . "\n";
221}
222
223
[13948]224sub get_all_chunks
225{
226 # The code of the target language (ensure it is lowercase)
227 my $target_language_code = lc(shift(@_));
228 # The key of the file to translate (ensure it is lowercase)
229 my $translation_file_key = lc(shift(@_));
[24627]230
[13948]231 # Check that the necessary arguments were supplied
232 if (!$target_language_code || !$translation_file_key) {
[24627]233 &throw_fatal_error("Missing command argument.");
[13948]234 }
[24627]235
[13948]236 # Get (and check) the translation configuration
237 my ($source_file, $target_file, $translation_file_type)
238 = &get_translation_configuration($target_language_code, $translation_file_key);
[24627]239
[13948]240 # Parse the source language and target language files
241 my $source_file_path = &util::filename_cat($gsdl_root_directory, $source_file);
242 my @source_file_lines = &read_file_lines($source_file_path);
243 my %source_file_key_to_line_mapping = &build_key_to_line_mapping(\@source_file_lines, $translation_file_type);
244
245 my $target_file_path = &util::filename_cat($gsdl_root_directory, $target_file);
246 my @target_file_lines = &read_file_lines($target_file_path);
247 my %target_file_key_to_line_mapping = &build_key_to_line_mapping(\@target_file_lines, $translation_file_type);
[24627]248
[13948]249 # Filter out any automatically translated chunks
250 foreach my $chunk_key (keys(%source_file_key_to_line_mapping)) {
[24627]251 if (&is_chunk_automatically_translated($chunk_key, $translation_file_type)) {
252 delete $source_file_key_to_line_mapping{$chunk_key};
253 delete $target_file_key_to_line_mapping{$chunk_key};
254 }
[13948]255 }
[24627]256
[13948]257 my %source_file_key_to_text_mapping = &build_key_to_text_mapping(\@source_file_lines, \%source_file_key_to_line_mapping, $translation_file_type);
258 my %target_file_key_to_text_mapping = &build_key_to_text_mapping(\@target_file_lines, \%target_file_key_to_line_mapping, $translation_file_type);
259 &log_message("Number of source chunks: " . scalar(keys(%source_file_key_to_text_mapping)));
260 &log_message("Number of target chunks: " . scalar(keys(%target_file_key_to_text_mapping)));
[24627]261
[13948]262 my %source_file_key_to_last_update_date_mapping = &build_key_to_last_update_date_mapping($source_file, \@source_file_lines, \%source_file_key_to_line_mapping, $translation_file_type);
263 my %target_file_key_to_last_update_date_mapping = &build_key_to_last_update_date_mapping($target_file, \@target_file_lines, \%target_file_key_to_line_mapping, $translation_file_type);
[24627]264
[18460]265 my $xml_response = &create_xml_response_for_all_chunks($translation_file_key, $target_file, \%source_file_key_to_text_mapping, \%target_file_key_to_text_mapping, \%source_file_key_to_last_update_date_mapping, \%target_file_key_to_last_update_date_mapping);
[13948]266
267 return $xml_response;
268}
269
270
[25249]271sub get_uptodate_chunks
272{
273 # The code of the target language (ensure it is lowercase)
274 my $target_language_code = lc(shift(@_));
275 # The key of the file to translate (ensure it is lowercase)
276 my $translation_file_key = lc(shift(@_));
277
278 # Check that the necessary arguments were supplied
279 if (!$target_language_code || !$translation_file_key) {
280 &throw_fatal_error("Missing command argument.");
281 }
282
283 # Get (and check) the translation configuration
284 my ($source_file, $target_file, $translation_file_type)
285 = &get_translation_configuration($target_language_code, $translation_file_key);
286
287 # Parse the source language and target language files
288 my $source_file_path = &util::filename_cat($gsdl_root_directory, $source_file);
289 my @source_file_lines = &read_file_lines($source_file_path);
290 my %source_file_key_to_line_mapping = &build_key_to_line_mapping(\@source_file_lines, $translation_file_type);
291
292 my $target_file_path = &util::filename_cat($gsdl_root_directory, $target_file);
293 my @target_file_lines = &read_file_lines($target_file_path);
294 my %target_file_key_to_line_mapping = &build_key_to_line_mapping(\@target_file_lines, $translation_file_type);
295
296 # Filter out any automatically translated chunks
297 foreach my $chunk_key (keys(%source_file_key_to_line_mapping)) {
298 if (&is_chunk_automatically_translated($chunk_key, $translation_file_type)) {
299 delete $source_file_key_to_line_mapping{$chunk_key};
300 delete $target_file_key_to_line_mapping{$chunk_key};
301 }
302 }
303
304 my %source_file_key_to_text_mapping = &build_key_to_text_mapping(\@source_file_lines, \%source_file_key_to_line_mapping, $translation_file_type);
305 my %target_file_key_to_text_mapping = &build_key_to_text_mapping(\@target_file_lines, \%target_file_key_to_line_mapping, $translation_file_type);
306 &log_message("Number of source chunks: " . scalar(keys(%source_file_key_to_text_mapping)));
307 &log_message("Number of target chunks: " . scalar(keys(%target_file_key_to_text_mapping)));
308
309 my %source_file_key_to_last_update_date_mapping = &build_key_to_last_update_date_mapping($source_file, \@source_file_lines, \%source_file_key_to_line_mapping, $translation_file_type);
310 my %target_file_key_to_last_update_date_mapping = &build_key_to_last_update_date_mapping($target_file, \@target_file_lines, \%target_file_key_to_line_mapping, $translation_file_type);
311
312
313 # Chunks needing updating are those in the target file that have been more recently edited in the source file
314 # All others are uptodate (which implies that they have certainly been translated at some point and would not be empty)
315 my @uptodate_target_file_keys = ();
316 foreach my $chunk_key (keys(%source_file_key_to_last_update_date_mapping)) {
317 my $source_chunk_last_update_date = $source_file_key_to_last_update_date_mapping{$chunk_key};
318 my $target_chunk_last_update_date = $target_file_key_to_last_update_date_mapping{$chunk_key};
319
320 # print "key: $chunk_key\nsource date : $source_chunk_last_update_date\ntarget date : $target_chunk_last_update_date\nafter? ". &is_date_after($source_chunk_last_update_date, $target_chunk_last_update_date) . "\n\n";
321
322 if (defined($target_chunk_last_update_date) && !&is_date_after($source_chunk_last_update_date, $target_chunk_last_update_date)) {
323 # &log_message("Chunk with key $chunk_key needs updating.");
324 push(@uptodate_target_file_keys, $chunk_key);
325 }
326 }
327
328 my $xml_response = &create_xml_response_for_uptodate_chunks($translation_file_key, $target_file, \@uptodate_target_file_keys, \%source_file_key_to_text_mapping, \%target_file_key_to_text_mapping, \%source_file_key_to_last_update_date_mapping, \%target_file_key_to_last_update_date_mapping);
329
330 return $xml_response;
331}
332
333
[10019]334sub get_first_n_chunks_requiring_work
335{
336 # The code of the target language (ensure it is lowercase)
337 my $target_language_code = lc(shift(@_));
338 # The key of the file to translate (ensure it is lowercase)
339 my $translation_file_key = lc(shift(@_));
340 # The number of chunks to return (defaults to one if not specified)
341 my $num_chunks_to_return = shift(@_) || "1";
[24627]342
[10019]343 # Check that the necessary arguments were supplied
344 if (!$target_language_code || !$translation_file_key) {
[24627]345 &throw_fatal_error("Missing command argument.");
[10019]346 }
[24627]347
[10019]348 # Get (and check) the translation configuration
349 my ($source_file, $target_file, $translation_file_type)
350 = &get_translation_configuration($target_language_code, $translation_file_key);
[28755]351
[10019]352 # Parse the source language and target language files
353 my $source_file_path = &util::filename_cat($gsdl_root_directory, $source_file);
354 my @source_file_lines = &read_file_lines($source_file_path);
355 my %source_file_key_to_line_mapping = &build_key_to_line_mapping(\@source_file_lines, $translation_file_type);
[14258]356
[10019]357 my $target_file_path = &util::filename_cat($gsdl_root_directory, $target_file);
358 my @target_file_lines = &read_file_lines($target_file_path);
359 my %target_file_key_to_line_mapping = &build_key_to_line_mapping(\@target_file_lines, $translation_file_type);
[24627]360
[10019]361 # Filter out any automatically translated chunks
362 foreach my $chunk_key (keys(%source_file_key_to_line_mapping)) {
[24627]363 if (&is_chunk_automatically_translated($chunk_key, $translation_file_type)) {
364 delete $source_file_key_to_line_mapping{$chunk_key};
365 delete $target_file_key_to_line_mapping{$chunk_key};
366 }
[10019]367 }
[24627]368
[10019]369 my %source_file_key_to_text_mapping = &build_key_to_text_mapping(\@source_file_lines, \%source_file_key_to_line_mapping, $translation_file_type);
370 my %target_file_key_to_text_mapping = &build_key_to_text_mapping(\@target_file_lines, \%target_file_key_to_line_mapping, $translation_file_type);
371 &log_message("Number of source chunks: " . scalar(keys(%source_file_key_to_text_mapping)));
372 &log_message("Number of target chunks: " . scalar(keys(%target_file_key_to_text_mapping)));
[24627]373
[10019]374 # Determine the target file chunks requiring translation
375 my @target_file_keys_requiring_translation = &determine_chunks_requiring_translation(\%source_file_key_to_text_mapping, \%target_file_key_to_text_mapping);
376 &log_message("Number of target chunks requiring translation: " . scalar(@target_file_keys_requiring_translation));
[24627]377
[10019]378 # Determine the target file chunks requiring updating
379 my %source_file_key_to_last_update_date_mapping = &build_key_to_last_update_date_mapping($source_file, \@source_file_lines, \%source_file_key_to_line_mapping, $translation_file_type);
380 my %target_file_key_to_last_update_date_mapping = &build_key_to_last_update_date_mapping($target_file, \@target_file_lines, \%target_file_key_to_line_mapping, $translation_file_type);
381 my @target_file_keys_requiring_updating = &determine_chunks_requiring_updating(\%source_file_key_to_last_update_date_mapping, \%target_file_key_to_last_update_date_mapping);
382 &log_message("Number of target chunks requiring updating: " . scalar(@target_file_keys_requiring_updating));
[24627]383
[18460]384 my $xml_response = &create_xml_response_for_chunks_requiring_work($translation_file_key, $target_file, scalar(keys(%source_file_key_to_text_mapping)), \@target_file_keys_requiring_translation, \@target_file_keys_requiring_updating, $num_chunks_to_return, \%source_file_key_to_text_mapping, \%target_file_key_to_text_mapping, \%source_file_key_to_last_update_date_mapping, \%target_file_key_to_last_update_date_mapping);
385
[10040]386 return $xml_response;
[10019]387}
388
389
390sub get_language_status
391{
392 # The code of the target language (ensure it is lowercase)
393 my $target_language_code = lc(shift(@_));
[24627]394
[10019]395 # Check that the necessary arguments were supplied
396 if (!$target_language_code) {
[24627]397 &throw_fatal_error("Missing command argument.");
[10019]398 }
[24627]399
[10040]400 # Form an XML response to the command
401 my $xml_response = "<?xml version=\"1.0\" encoding=\"UTF-8\" ?>\n";
402 $xml_response .= "<GTIResponse>\n";
403 $xml_response .= " <LanguageStatus code=\"$target_language_code\">\n";
[24627]404
[18460]405 foreach my $translation_file (@$gti_translation_files) {
[24627]406 my ($num_source_chunks, $num_target_chunks, $num_chunks_requiring_translation, $num_chunks_requiring_updating) = 0;
407 my $target_file_name = "";
408
[30582]409 if ($translation_file->{'key'} =~ m/^gs3/) { # gs3interface, gs3colcfg
[24627]410 my (%source_file_key_to_text_mapping, %target_file_key_to_text_mapping, %source_file_key_to_last_update_date_mapping, %target_file_key_to_last_update_date_mapping ) = ();
[30581]411 &build_gs3_configuration($translation_file->{'key'}, $target_language_code, \%source_file_key_to_text_mapping, \%target_file_key_to_text_mapping, \%source_file_key_to_last_update_date_mapping, \%target_file_key_to_last_update_date_mapping );
[24627]412
413 my @target_file_keys_requiring_translation = &determine_chunks_requiring_translation(\%source_file_key_to_text_mapping, \%target_file_key_to_text_mapping);
414 my @target_file_keys_requiring_updating = &determine_chunks_requiring_updating(\%source_file_key_to_last_update_date_mapping, \%target_file_key_to_last_update_date_mapping);
415
416 $num_source_chunks = scalar(keys(%source_file_key_to_text_mapping));
417 $num_target_chunks = scalar(keys(%target_file_key_to_text_mapping));
418 $num_chunks_requiring_translation = scalar(@target_file_keys_requiring_translation);
419 $num_chunks_requiring_updating = scalar(@target_file_keys_requiring_updating);
[18460]420 }
[24627]421 else {
422 # Get (and check) the translation configuration
423 my ($source_file, $target_file, $translation_file_type) = &get_translation_configuration($target_language_code, $translation_file->{'key'});
424 $target_file_name = $target_file;
425
426 # Parse the source language and target language files
427 my $source_file_path = &util::filename_cat($gsdl_root_directory, $source_file);
428 my @source_file_lines = &read_file_lines($source_file_path);
429 my %source_file_key_to_line_mapping = &build_key_to_line_mapping(\@source_file_lines, $translation_file_type);
430
431 my $target_file_path = &util::filename_cat($gsdl_root_directory, $target_file);
432 my @target_file_lines = &read_file_lines($target_file_path);
433 my %target_file_key_to_line_mapping = &build_key_to_line_mapping(\@target_file_lines, $translation_file_type);
434
435 # Filter out any automatically translated chunks
436 foreach my $chunk_key (keys(%source_file_key_to_line_mapping)) {
437 if (&is_chunk_automatically_translated($chunk_key, $translation_file_type)) {
438 delete $source_file_key_to_line_mapping{$chunk_key};
439 delete $target_file_key_to_line_mapping{$chunk_key};
440 }
441 }
442
443 my %source_file_key_to_text_mapping = &build_key_to_text_mapping(\@source_file_lines, \%source_file_key_to_line_mapping, $translation_file_type);
444 my %target_file_key_to_text_mapping = &build_key_to_text_mapping(\@target_file_lines, \%target_file_key_to_line_mapping, $translation_file_type);
445
446 # Determine the target file chunks requiring translation
447 my @target_file_keys_requiring_translation = &determine_chunks_requiring_translation(\%source_file_key_to_text_mapping, \%target_file_key_to_text_mapping);
448
449 # Determine the target file chunks requiring updating
450 my @target_file_keys_requiring_updating = ();
451 if (-e $target_file_path) {
452 my %source_file_key_to_last_update_date_mapping = &build_key_to_last_update_date_mapping($source_file, \@source_file_lines, \%source_file_key_to_line_mapping, $translation_file_type);
453 my %target_file_key_to_last_update_date_mapping = &build_key_to_last_update_date_mapping($target_file, \@target_file_lines, \%target_file_key_to_line_mapping, $translation_file_type);
454 @target_file_keys_requiring_updating = &determine_chunks_requiring_updating(\%source_file_key_to_last_update_date_mapping, \%target_file_key_to_last_update_date_mapping);
455 }
456
457 $num_source_chunks = scalar(keys(%source_file_key_to_text_mapping));
458 $num_target_chunks = scalar(keys(%target_file_key_to_text_mapping));
459 $num_chunks_requiring_translation = scalar(@target_file_keys_requiring_translation);
460 $num_chunks_requiring_updating = scalar(@target_file_keys_requiring_updating);
461 }
462
463 &log_message("Status of " . $translation_file->{'key'});
464 &log_message("Number of source chunks: " . $num_source_chunks);
465 &log_message("Number of target chunks: " . $num_target_chunks);
466 &log_message("Number of target chunks requiring translation: " . $num_chunks_requiring_translation);
467 &log_message("Number of target chunks requiring updating: " . $num_chunks_requiring_updating);
468
469 $xml_response .= " <TranslationFile"
[10019]470 . " key=\"" . $translation_file->{'key'} . "\""
[18460]471 . " target_file_path=\"" . $target_file_name . "\""
472 . " num_chunks_translated=\"" . ($num_source_chunks - $num_chunks_requiring_translation) . "\""
473 . " num_chunks_requiring_translation=\"" . $num_chunks_requiring_translation . "\""
474 . " num_chunks_requiring_updating=\"" . $num_chunks_requiring_updating . "\"\/>\n";
[10019]475 }
[24627]476
[10040]477 $xml_response .= " </LanguageStatus>\n";
[24627]478
[10040]479 $xml_response .= "</GTIResponse>\n";
480 return $xml_response;
[10019]481}
482
483
484sub search_chunks
485{
486 # The code of the target language (ensure it is lowercase)
487 my $target_language_code = lc(shift(@_));
488 # The key of the file to translate (ensure it is lowercase)
489 my $translation_file_key = lc(shift(@_));
490 # The query string
[10020]491 my $query_string = join(' ', @_);
[24627]492
[10019]493 # Check that the necessary arguments were supplied
494 if (!$target_language_code || !$translation_file_key || !$query_string) {
[24627]495 &throw_fatal_error("Missing command argument.");
[10019]496 }
[24627]497
[18460]498 my ($source_file, $target_file, $translation_file_type) = ();
499 my %source_file_key_to_text_mapping = ();
500 my %target_file_key_to_text_mapping = ();
501
502
[30582]503 if ($translation_file_key !~ m/^gs3/) {
[24627]504 # Get (and check) the translation configuration
505 ($source_file, $target_file, $translation_file_type) = &get_translation_configuration($target_language_code, $translation_file_key);
506
507 # Parse the source language and target language files
508 my $source_file_path = &util::filename_cat($gsdl_root_directory, $source_file);
509 my @source_file_lines = &read_file_lines($source_file_path);
510 my %source_file_key_to_line_mapping = &build_key_to_line_mapping(\@source_file_lines, $translation_file_type);
511
512 my $target_file_path = &util::filename_cat($gsdl_root_directory, $target_file);
513 my @target_file_lines = &read_file_lines($target_file_path);
514 my %target_file_key_to_line_mapping = &build_key_to_line_mapping(\@target_file_lines, $translation_file_type);
515
516 # Filter out any automatically translated chunks
517 foreach my $chunk_key (keys(%source_file_key_to_line_mapping)) {
518 if (&is_chunk_automatically_translated($chunk_key, $translation_file_type)) {
519 delete $source_file_key_to_line_mapping{$chunk_key};
520 delete $target_file_key_to_line_mapping{$chunk_key};
521 }
522 }
523
524 %source_file_key_to_text_mapping = &build_key_to_text_mapping(\@source_file_lines, \%source_file_key_to_line_mapping, $translation_file_type);
525 %target_file_key_to_text_mapping = &build_key_to_text_mapping(\@target_file_lines, \%target_file_key_to_line_mapping, $translation_file_type);
[10019]526 }
[18460]527 else {
[24627]528 # Not needed in this case
529 my (%source_file_key_to_gti_command_mapping, %target_file_key_to_gti_command_mapping) = ();
[30581]530 &build_gs3_configuration($translation_file_key, $target_language_code, \%source_file_key_to_text_mapping, \%target_file_key_to_text_mapping,
[30548]531 \%source_file_key_to_gti_command_mapping, \%target_file_key_to_gti_command_mapping, 1);
[18460]532 }
[24627]533
[10019]534 &log_message("Number of source chunks: " . scalar(keys(%source_file_key_to_text_mapping)));
535 &log_message("Number of target chunks: " . scalar(keys(%target_file_key_to_text_mapping)));
[24627]536
[10019]537 # Determine the target file chunks matching the query
538 my @target_file_keys_matching_query = ();
539 foreach my $chunk_key (keys(%target_file_key_to_text_mapping)) {
[24627]540 my $target_file_text = $target_file_key_to_text_mapping{$chunk_key};
541 if ($target_file_text =~ /$query_string/i) {
542 # &log_message("Chunk with key $chunk_key matches query.");
543 push(@target_file_keys_matching_query, $chunk_key);
544 }
[10019]545 }
[24627]546
[10040]547 # Form an XML response to the command
548 my $xml_response = "<?xml version=\"1.0\" encoding=\"UTF-8\" ?>\n";
549 $xml_response .= "<GTIResponse>\n";
[24627]550
[10040]551 $xml_response .= " <ChunksMatchingQuery size=\"" . scalar(@target_file_keys_matching_query) . "\">\n";
[10019]552 foreach my $chunk_key (@target_file_keys_matching_query) {
[24627]553 my $target_file_chunk_text = &make_text_xml_safe($target_file_key_to_text_mapping{$chunk_key});
554
555 $xml_response .= " <Chunk key=\"$chunk_key\">\n";
556 $xml_response .= " <TargetFileText>$target_file_chunk_text</TargetFileText>\n";
557 $xml_response .= " </Chunk>\n";
[10019]558 }
[10040]559 $xml_response .= " </ChunksMatchingQuery>\n";
[24627]560
[10040]561 $xml_response .= "</GTIResponse>\n";
562 return $xml_response;
[10019]563}
564
565
[10050]566sub submit_translations
[10019]567{
568 # The code of the target language (ensure it is lowercase)
569 my $target_language_code = lc(shift(@_));
570 # The key of the file to translate (ensure it is lowercase)
571 my $translation_file_key = lc(shift(@_));
[12484]572 # The username of the translation submitter
573 my $submitter_username = shift(@_);
[10027]574 # Whether to submit a target chunk even if it hasn't changed
575 my $force_submission_flag = shift(@_);
[24627]576
[10019]577 # Check that the necessary arguments were supplied
[12484]578 if (!$target_language_code || !$translation_file_key || !$submitter_username) {
[24627]579 &log_message("Fatal error (but cannot be thrown): Missing command argument.");
580 die "\n";
[10019]581 }
[18460]582
583 my %source_file_key_to_text_mapping = ();
584 my %source_file_key_to_gti_comment_mapping = ();
585 my %target_file_key_to_text_mapping = ();
586 my %target_file_key_to_gti_comment_mapping = ();
[24627]587
[18460]588 my (@source_file_lines, @target_file_lines) = ();
589 my ($source_file, $target_file, $translation_file_type);
[24627]590
[18460]591
[30582]592 if ($translation_file_key !~ m/^gs3/) {
[24627]593 # Get (and check) the translation configuration
594 ($source_file, $target_file, $translation_file_type)
[18460]595 = &get_translation_configuration($target_language_code, $translation_file_key);
[24627]596
597 # Parse the source language and target language files
598 @source_file_lines = &read_file_lines(&util::filename_cat($gsdl_root_directory, $source_file));
599 my %source_file_key_to_line_mapping = &build_key_to_line_mapping(\@source_file_lines, $translation_file_type);
600 %source_file_key_to_text_mapping = &build_key_to_text_mapping(\@source_file_lines, \%source_file_key_to_line_mapping, $translation_file_type);
601 %source_file_key_to_gti_comment_mapping = &build_key_to_gti_comment_mapping(\@source_file_lines, \%source_file_key_to_line_mapping, $translation_file_type);
602
603 @target_file_lines = &read_file_lines(&util::filename_cat($gsdl_root_directory, $target_file));
604 my %target_file_key_to_line_mapping = &build_key_to_line_mapping(\@target_file_lines, $translation_file_type);
605 %target_file_key_to_text_mapping = &build_key_to_text_mapping(\@target_file_lines, \%target_file_key_to_line_mapping, $translation_file_type);
606 %target_file_key_to_gti_comment_mapping = &build_key_to_gti_comment_mapping(\@target_file_lines, \%target_file_key_to_line_mapping, $translation_file_type);
[18460]607 }
608 else {
[30581]609 &build_gs3_configuration($translation_file_key, $target_language_code, \%source_file_key_to_text_mapping, \%target_file_key_to_text_mapping,
[30548]610 \%source_file_key_to_gti_comment_mapping, \%target_file_key_to_gti_comment_mapping, 1);
[18460]611 }
[10019]612 &log_message("Number of source chunks: " . scalar(keys(%source_file_key_to_text_mapping)));
613 &log_message("Number of target chunks: " . scalar(keys(%target_file_key_to_text_mapping)));
[24627]614
[10019]615 # Submission date
616 my $day = (localtime)[3];
617 my $month = ("Jan", "Feb", "Mar", "Apr", "May", "Jun", "Jul", "Aug", "Sep", "Oct", "Nov", "Dec")[(localtime)[4]];
618 my $year = (localtime)[5] + 1900;
619 my $submission_date = "$day-$month-$year";
[24627]620
[10019]621 open(SUBMISSION, "-");
622 my @submission_lines = <SUBMISSION>;
623 close(SUBMISSION);
[24627]624
[10019]625 # Remove any nasty carriage returns
[18460]626 # &log_message("Submission:");
[10019]627 foreach my $submission_line (@submission_lines) {
[24627]628 $submission_line =~ s/\r$//;
629 #&log_message(" $submission_line");
[10019]630 }
[24627]631
[10019]632 my %source_file_key_to_submission_mapping = ();
633 my %target_file_key_to_submission_mapping = ();
634 for (my $i = 0; $i < scalar(@submission_lines); $i++) {
[24627]635 # Read source file part of submission
636 if ($submission_lines[$i] =~ /^\<SourceFileText key=\"(.+)\"\>/) {
637 my $chunk_key = $1;
638
639 # Read the source file text
640 my $source_file_chunk_text = "";
641 $i++;
642 while ($i < scalar(@submission_lines) && $submission_lines[$i] !~ /^\<\/SourceFileText\>/) {
643 $source_file_chunk_text .= $submission_lines[$i];
644 $i++;
645 }
646 $source_file_chunk_text =~ s/\n$//; # Strip the extra newline character added
647 $source_file_chunk_text = &unmake_text_xml_safe($source_file_chunk_text);
648
649 #&log_message("Source file key: $chunk_key");
650 #&log_message("Source file text: $source_file_chunk_text");
651 $source_file_key_to_submission_mapping{$chunk_key} = $source_file_chunk_text;
652 }
653
654 # Read target file part of submission
655 if ($submission_lines[$i] =~ /^\<TargetFileText key=\"(.+)\"\>/) {
656 my $chunk_key = $1;
657
658 # Read the target file text
659 my $target_file_chunk_text = "";
660 $i++;
661 while ($i < scalar(@submission_lines) && $submission_lines[$i] !~ /^\<\/TargetFileText\>/) {
662 $target_file_chunk_text .= $submission_lines[$i];
663 $i++;
664 }
665 $target_file_chunk_text =~ s/\n$//; # Strip the extra newline character added
666 $target_file_chunk_text = &unmake_text_xml_safe($target_file_chunk_text);
667
668 #&log_message("Target file key: $chunk_key");
669 #&log_message("Target file text: $target_file_chunk_text");
670 $target_file_key_to_submission_mapping{$chunk_key} = $target_file_chunk_text;
671 }
[10019]672 }
[24627]673
[10019]674 # -----------------------------------------
675 # Validate the translation submissions
676 # -----------------------------------------
[24627]677
[10019]678 # Check that the translations are valid
679 foreach my $chunk_key (keys(%source_file_key_to_submission_mapping)) {
[30687]680
[30719]681 # Kathy introduced escaped colons ("\:") into chunk keys in properties files (greenstone3/metadata_names),
682 # but they're not escaped in the submitted XML versions, nor are they escaped in memory (in the $chunk_key)
[30687]683
[24627]684 # Make sure the submitted chunk still exists in the source file
[30719]685 if (!defined($source_file_key_to_text_mapping{$chunk_key})) {
[24627]686 &log_message("Warning: Source chunk $chunk_key no longer exists (ignoring submission).");
687 delete $source_file_key_to_submission_mapping{$chunk_key};
688 delete $target_file_key_to_submission_mapping{$chunk_key};
689 next;
690 }
691
692 # Make sure the submitted source chunk matches the source file chunk
[30719]693 if ($source_file_key_to_submission_mapping{$chunk_key} ne &unmake_text_xml_safe($source_file_key_to_text_mapping{$chunk_key})) {
[28518]694 #if (&unmake_text_xml_safe($source_file_key_to_submission_mapping{$chunk_key}) ne &unmake_text_xml_safe($source_file_key_to_text_mapping{$chunk_key})) {
[28503]695 #print STDERR "**** $source_file_key_to_submission_mapping{$chunk_key}\n";
696 #print STDERR "**** " . &unmake_text_xml_safe($source_file_key_to_text_mapping{$chunk_key}) ."\n";
697
[30719]698 &log_message("Warning: Source chunk $chunk_key has changed (ignoring submission).");
[30687]699 &log_message("Submission source: |$source_file_key_to_submission_mapping{$chunk_key}|");
[30719]700 &log_message(" Source text: |$source_file_key_to_text_mapping{$chunk_key}|");
[24627]701 delete $source_file_key_to_submission_mapping{$chunk_key};
702 delete $target_file_key_to_submission_mapping{$chunk_key};
703 next;
704 }
[10027]705 }
[24627]706
[10027]707 # Apply the submitted translations
708 foreach my $chunk_key (keys(%target_file_key_to_submission_mapping)) {
[24627]709 # Only apply the submission if it is a change, unless -force_submission has been specified
[30719]710 if ($force_submission_flag || !defined($target_file_key_to_text_mapping{$chunk_key}) || $target_file_key_to_submission_mapping{$chunk_key} ne $target_file_key_to_text_mapping{$chunk_key}) {
711 $target_file_key_to_text_mapping{$chunk_key} = $target_file_key_to_submission_mapping{$chunk_key};
712 $target_file_key_to_gti_comment_mapping{$chunk_key} = "Updated $submission_date by $submitter_username";
[24627]713 }
[10019]714 }
[18460]715
[30582]716 if ($translation_file_key !~ m/^gs3/) {
[24627]717 eval "&write_translated_${translation_file_type}(\$source_file, \\\@source_file_lines, \\\%source_file_key_to_text_mapping, \$target_file, \\\@target_file_lines, \\\%target_file_key_to_text_mapping, \\\%target_file_key_to_gti_comment_mapping, \$target_language_code)";
[18460]718 } else {
[30581]719 eval "&write_translated_gs3interface(\$translation_file_key, \\\%source_file_key_to_text_mapping, \\\%target_file_key_to_text_mapping, \\\%target_file_key_to_gti_comment_mapping, \$target_language_code)";
[18460]720 }
721}
[10019]722
[18460]723
724sub create_glihelp_zip_file
725{
726 my $target_language_code = shift(@_);
727 my $translation_file_key = "glihelp";
728
729 &log_message("Creating GLI Help zip file for $target_language_code");
[24627]730
[18460]731 my ($source_file, $target_file, $translation_file_type) = &get_translation_data_for($target_language_code, $translation_file_key);
732
733 my $classpath = &util::filename_cat($gsdl_root_directory, "gti-lib");
[28977]734 my $oldclasspath = $classpath;
[18460]735 if ( ! -e $classpath) {
[28977]736 $classpath = &util::filename_cat($gsdl_root_directory, "gli", "shared");
737 }
738 if ( ! -e $classpath) {
739 &throw_fatal_error("$classpath doesn't exist! (Neither does $oldclasspath.) Need the files in this directory (ApplyXLST and its related files) to create the zip file for GLI Help");
740 }
741
[28991]742
743 my $perllib_path = &util::filename_cat($gsdl_root_directory, "perllib"); # strings.properties
744 my $gliclasses_path = &util::filename_cat($gsdl_root_directory, "gli", "classes"); # dictionary.properties
745 my $os = $^O;
746 my $path_separator = ($^O =~ m/mswin/i) ? ";" : ":";
747 my $xalan_path = &util::filename_cat($classpath, "xalan.jar");
748 $classpath = "$perllib_path$path_separator$gliclasses_path$path_separator$classpath$path_separator$xalan_path";
749
[18460]750 my $gli_help_directory = &util::filename_cat($gsdl_root_directory, "gli");
751 $gli_help_directory = &util::filename_cat($gli_help_directory, "help");
752
753 my $gen_many_html_xsl_filepath = &util::filename_cat($gli_help_directory, "gen-many-html.xsl");
754 if ( ! -e $gen_many_html_xsl_filepath) {
[24627]755 &throw_fatal_error("$gen_many_html_xsl_filepath doesn't exist! Need this file to create the zip file for GLI Help");
[18460]756 }
[24627]757
[18460]758 my $gen_index_xml_xsl_filepath = &util::filename_cat($gli_help_directory, "gen-index-xml.xsl");
759 my $split_script_filepath = &util::filename_cat($gli_help_directory, "splithelpdocument.pl");
760
761 my $target_file_directory = &util::filename_cat($gli_help_directory, $target_language_code);
762 $target_file_directory = $target_file_directory."/";
[24627]763
[18460]764 my $target_filepath = &util::filename_cat($gsdl_root_directory, $target_file);
[28991]765
766 # if gli/help/nl doesn't exist, create it by copying over gli/help/en/help.xml, then process the copied file
767 my ($tailname, $glihelp_lang_dir, $suffix) = &File::Basename::fileparse($target_filepath, "\\.[^\\.]+\$");
768 if(!&FileUtils::directoryExists($glihelp_lang_dir)) {
769
770 # copy across the gli/help/en/help.xml into a new folder for the new language gli/help/<newlang>
771 my $en_glihelp_dir = &util::filename_cat($gli_help_directory, "en");
[28992]772 my $en_helpxml_file = &util::filename_cat($en_glihelp_dir, "$tailname$suffix"); #$tailname$suffix="help.xml"
[28991]773 &FileUtils::copyFilesRecursiveNoSVN($en_helpxml_file, $glihelp_lang_dir);
774
775 # In gli/help/<newlang>/help.xml, replace all occurrences of
776 # <Text id="1">This text in en will be removed for new langcode</Text>
777 # with <!-- Missing translation: 1 -->
778 open(FIN,"<$target_filepath") or &throw_fatal_error("Could not open $target_filepath for READING after creating it");
779 my $help_xml_contents;
780 # Read in the entire contents of the file in one hit
781 sysread(FIN, $help_xml_contents, -s FIN);
782 close(FIN);
[24627]783
[28991]784 $help_xml_contents =~ s@<Text id="([^"]+?)">(.*?)</Text>@<!-- Missing translation: $1 -->@sg;
785
786 open(FOUT, ">$target_filepath") or &throw_fatal_error("Could not open $target_filepath for WRITING after creating it");
787 print FOUT $help_xml_contents;
788 close(FOUT);
789 }
790
[24627]791 my $perl_exec = &util::get_perl_exec();
792 my $java_exec = "java";
793 if(defined($ENV{'JAVA_HOME'}) && $ENV{'JAVA_HOME'} ne ""){
794 $java_exec = &util::filename_cat($ENV{'JAVA_HOME'}, "bin", "java");
[31751]795 } elsif(defined($ENV{'JRE_HOME'}) && $ENV{'JRE_HOME'} ne ""){
796 $java_exec = &util::filename_cat($ENV{'JRE_HOME'}, "bin", "java");
[24627]797 }
[28991]798
799 #my $cmd = "$java_exec -cp $classpath:$classpath/xalan.jar ApplyXSLT $target_language_code $gen_many_html_xsl_filepath $target_filepath | \"$perl_exec\" -S $split_script_filepath $target_file_directory";
800 my $cmd = "$java_exec -DGSDLHOME=$gsdl_root_directory -cp $classpath ApplyXSLT $target_language_code $gen_many_html_xsl_filepath $target_filepath | \"$perl_exec\" -S $split_script_filepath $target_file_directory";
801 #&throw_fatal_error("RAN gti command: $cmd");
[18460]802 my $response = `$cmd`;
[28991]803
804 #$cmd = "$java_exec -cp $classpath:$classpath/xalan.jar ApplyXSLT $target_language_code $gen_index_xml_xsl_filepath $target_filepath > " . $target_file_directory . "help_index.xml"; # 2>/dev/null";
805 $cmd = "$java_exec -cp $classpath -DGSDLHOME=$gsdl_root_directory ApplyXSLT $target_language_code $gen_index_xml_xsl_filepath $target_filepath > " . $target_file_directory . "help_index.xml"; # 2>/dev/null";
[18460]806 $response = `$cmd`;
[28991]807
808 # create a gti/tmp folder, if one doesn't already exist, and store the downloadable zip file in there
809 my $tmpdir = &util::filename_cat($gsdl_root_directory, "tmp");
810 if(!&FileUtils::directoryExists($tmpdir)) {
811 &FileUtils::makeDirectory($tmpdir);
812 }
[28977]813 #my $zip_file_path = "/greenstone/custom/gti/" . $target_language_code . "_GLIHelp.zip";
[28991]814 my $zip_file_path = &util::filename_cat($tmpdir, $target_language_code . "_GLIHelp.zip");
[18460]815 $cmd = "zip -rj $zip_file_path $target_file_directory -i \*.htm \*.xml";
[28991]816
[18460]817 $response = `$cmd`;
[10019]818}
819
820
821sub get_translation_configuration
822{
823 # Get the code of the target language
824 my $target_language_code = shift(@_);
825 # Get the key of the file to translate
826 my $translation_file_key = shift(@_);
[24627]827
[10019]828 # Read the translation data from the gti.cfg file
829 my ($source_file, $target_file, $translation_file_type) =
830 &get_translation_data_for($target_language_code, $translation_file_key);
[24627]831
[10019]832 # Check that the file to translate is defined in the gti.cfg file
833 if (!$source_file || !$target_file || !$translation_file_type) {
[24627]834 &throw_fatal_error("Missing or incomplete specification for translation file \"$translation_file_key\" in gti.pl.");
[10019]835 }
[24627]836
[10019]837 # Check that the source file exists
838 my $source_file_path = &util::filename_cat($gsdl_root_directory, $source_file);
839 if (!-e $source_file_path) {
[24627]840 &throw_fatal_error("Source file $source_file_path does not exist.");
[10019]841 }
[24627]842
[10019]843 # Check that the source file is up to date
[11104]844 # The "2>/dev/null" is very important! If it is missing this will never return when run from the receptionist
[10019]845 # unless ($translation_file_is_not_in_cvs) {
[14258]846 #my $source_file_cvs_status = `cd $gsdl_root_directory; cvs -d $anonymous_cvs_root update $source_file 2>/dev/null`;
[24627]847 my $source_file_cvs_status = `cd $gsdl_root_directory; svn status $source_file 2>/dev/null`;
[10019]848 if ($source_file_cvs_status =~ /^C /) {
849 &throw_fatal_error("Source file $source_file_path conflicts with the repository.");
850 }
851 if ($source_file_cvs_status =~ /^M /) {
852 &throw_fatal_error("Source file $source_file_path contains uncommitted changes.");
853 }
854 # }
[24627]855
[10019]856 return ($source_file, $target_file, $translation_file_type);
857}
858
859
860sub get_translation_data_for
861{
862 my ($target_language_code, $translation_file_key) = @_;
[24627]863
[10019]864 foreach my $translation_file (@$gti_translation_files) {
[24627]865 # If this isn't the correct translation file, move onto the next one
866 next if ($translation_file_key ne $translation_file->{'key'});
867
868 # Resolve the target language file
869 my $target_language_file = $translation_file->{'target_file'};
870 if ($target_language_file =~ /(\{.+\;.+\})/) {
871 my $unresolved_target_language_file_part = $1;
872
873 # Check for a special case for the target language code
874 if ($unresolved_target_language_file_part =~ /(\{|\;)$target_language_code:([^\;]+)(\;|\})/) {
875 my $resolved_target_language_file_part = $2;
876 $target_language_file =~ s/$unresolved_target_language_file_part/$resolved_target_language_file_part/;
877 }
878 # Otherwise use the last part as the default value
879 else {
880 my ($default_target_language_file_part) = $unresolved_target_language_file_part =~ /([^\;]+)\}/;
881 $target_language_file =~ s/$unresolved_target_language_file_part/\{$default_target_language_file_part\}/;
[10019]882 }
[10041]883 }
[24627]884
[10041]885 # Resolve instances of {iso_639_1_target_language_name}
886 my $iso_639_1_target_language_name = $iso639::fromiso639{$target_language_code};
887 $iso_639_1_target_language_name =~ tr/A-Z/a-z/ if $iso_639_1_target_language_name;
888 $target_language_file =~ s/\{iso_639_1_target_language_name\}/$iso_639_1_target_language_name/g;
[24627]889
[10041]890 # Resolve instances of {target_language_code}
891 $target_language_file =~ s/\{target_language_code\}/$target_language_code/g;
[24627]892
[10041]893 return ($translation_file->{'source_file'}, $target_language_file, $translation_file->{'file_type'});
[24627]894}
[10019]895
[24627]896return ();
[10019]897}
898
899
900sub read_file_lines
901{
902 my ($file_path) = @_;
[24627]903
[10019]904 if (!open(FILE_IN, "<$file_path")) {
[24627]905 &log_message("Note: Could not open file $file_path.");
906 return ();
[10019]907 }
908 my @file_lines = <FILE_IN>;
909 close(FILE_IN);
[24627]910
[10019]911 return @file_lines;
912}
913
914
915sub build_key_to_line_mapping
916{
917 my ($file_lines, $translation_file_type) = @_;
918 eval "return &build_key_to_line_mapping_for_${translation_file_type}(\@\$file_lines)";
919}
920
921
922sub build_key_to_text_mapping
923{
924 my ($file_lines, $key_to_line_mapping, $translation_file_type) = @_;
[24627]925
[10019]926 my %key_to_text_mapping = ();
927 foreach my $chunk_key (keys(%$key_to_line_mapping)) {
[24627]928 my $chunk_starting_line = (split(/-/, $key_to_line_mapping->{$chunk_key}))[0];
929 my $chunk_finishing_line = (split(/-/, $key_to_line_mapping->{$chunk_key}))[1];
930
931 my $chunk_text = @$file_lines[$chunk_starting_line];
932 for (my $l = ($chunk_starting_line + 1); $l <= $chunk_finishing_line; $l++) {
933 $chunk_text .= @$file_lines[$l];
934 }
935
936 # Map from chunk key to text
937 eval "\$key_to_text_mapping{\${chunk_key}} = &import_chunk_from_${translation_file_type}(\$chunk_text)";
[30681]938
939 #if($chunk_key =~ m/document\\/) {
940 #&log_message("Submission source: $source_file_key_to_submission_mapping{$chunk_key}");
941 #&log_message("@@@ chunk key: $chunk_key");
942 #}
943
[10019]944 }
[24627]945
[10019]946 return %key_to_text_mapping;
947}
948
949
950sub build_key_to_last_update_date_mapping
951{
952 my ($file, $file_lines, $key_to_line_mapping, $translation_file_type) = @_;
[24627]953
[10019]954 # If the files aren't in CVS then we can't tell anything about what needs updating
955 # return () if ($translation_file_is_not_in_cvs);
[24627]956
[10019]957 # Build a mapping from key to CVS date
958 # Need to be careful with this mapping because the chunk keys won't necessarily all be valid
959 my %key_to_cvs_date_mapping = &build_key_to_cvs_date_mapping($file, $translation_file_type);
[24627]960
[10019]961 # Build a mapping from key to comment date
[12483]962 my %key_to_gti_comment_mapping = &build_key_to_gti_comment_mapping($file_lines, $key_to_line_mapping, $translation_file_type);
[24627]963
[10019]964 # Build a mapping from key to last update date (the latter of the CVS date and comment date)
965 my %key_to_last_update_date_mapping = ();
966 foreach my $chunk_key (keys(%$key_to_line_mapping)) {
[24627]967 # Use the CVS date as a starting point
968 my $chunk_cvs_date = $key_to_cvs_date_mapping{$chunk_key};
969 $key_to_last_update_date_mapping{$chunk_key} = $chunk_cvs_date;
970
971 # If a comment date exists and it is after the CVS date, use that instead
[14258]972 # need to convert the comment date format to SVN format
[24627]973 my $chunk_gti_comment = $key_to_gti_comment_mapping{$chunk_key};
974 if (defined($chunk_gti_comment) && $chunk_gti_comment =~ /(\d?\d-\D\D\D-\d\d\d\d)/) {
975 my $chunk_comment_date = $1;
976 if ((!defined($chunk_cvs_date) || &is_date_after($chunk_comment_date, $chunk_cvs_date))) {
977 $key_to_last_update_date_mapping{$chunk_key} = $chunk_comment_date;
978 }
979 }
[10019]980 }
[24627]981
[10019]982 return %key_to_last_update_date_mapping;
983}
984
985
986sub build_key_to_cvs_date_mapping
987{
988 my ($filename, $translation_file_type) = @_;
[24627]989
990 # Use SVN to annotate each line of the file with the date it was last edited
[10019]991 # The "2>/dev/null" is very important! If it is missing this will never return when run from the receptionist
[24627]992 my $cvs_annotated_file = `cd $gsdl_root_directory; svn annotate -v --force $filename 2>/dev/null`;
[14258]993
[10019]994 my @cvs_annotated_file_lines = split(/\n/, $cvs_annotated_file);
[24627]995
[10019]996 my @cvs_annotated_file_lines_date = ();
997 foreach my $cvs_annotated_file_line (@cvs_annotated_file_lines) {
[24627]998 # Extract the date from the SVN annotation at the front
999 # svn format : 2007-07-16
[14258]1000 $cvs_annotated_file_line =~ s/^\s+\S+\s+\S+\s(\S+)//;
1001
1002 push(@cvs_annotated_file_lines_date, $1);
1003
1004 # trim extra date information in svn annotation format
1005 # 15:42:49 +1200 (Wed, 21 Jun 2006)
1006 $cvs_annotated_file_line =~ s/^\s+\S+\s\S+\s\((.+?)\)\s//;
1007 }
1008
[10019]1009 # Build a key to line mapping for the CVS annotated file, for matching the chunk key to the CVS date
1010 my %key_to_line_mapping = &build_key_to_line_mapping(\@cvs_annotated_file_lines, $translation_file_type);
[24627]1011
[10019]1012 my %key_to_cvs_date_mapping = ();
1013 foreach my $chunk_key (keys(%key_to_line_mapping)) {
[24627]1014 my $chunk_starting_line = (split(/-/, $key_to_line_mapping{$chunk_key}))[0];
1015 my $chunk_finishing_line = (split(/-/, $key_to_line_mapping{$chunk_key}))[1];
1016
1017 # Find the date this chunk was last edited, from the CVS annotation
1018 my $chunk_date = $cvs_annotated_file_lines_date[$chunk_starting_line];
1019 for (my $l = ($chunk_starting_line + 1); $l <= $chunk_finishing_line; $l++) {
1020 if (&is_date_after($cvs_annotated_file_lines_date[$l], $chunk_date)) {
1021 # This part of the chunk has been updated more recently
1022 $chunk_date = $cvs_annotated_file_lines_date[$l];
1023
1024 }
1025 }
1026
1027 # Map from chunk key to CVS date
1028 $key_to_cvs_date_mapping{$chunk_key} = $chunk_date;
[10019]1029 }
[24627]1030
[10019]1031 return %key_to_cvs_date_mapping;
1032}
1033
1034
[12483]1035sub build_key_to_gti_comment_mapping
[10019]1036{
1037 my ($file_lines, $key_to_line_mapping, $translation_file_type) = @_;
[24627]1038
[12483]1039 my %key_to_gti_comment_mapping = ();
[10019]1040 foreach my $chunk_key (keys(%$key_to_line_mapping)) {
[24627]1041 my $chunk_starting_line = (split(/-/, $key_to_line_mapping->{$chunk_key}))[0];
1042 my $chunk_finishing_line = (split(/-/, $key_to_line_mapping->{$chunk_key}))[1];
1043
1044 my $chunk_text = @$file_lines[$chunk_starting_line];
1045 for (my $l = ($chunk_starting_line + 1); $l <= $chunk_finishing_line; $l++) {
1046 $chunk_text .= @$file_lines[$l];
1047 }
1048
1049 # Map from chunk key to GTI comment
1050 my $chunk_gti_comment;
1051 eval "\$chunk_gti_comment = &get_${translation_file_type}_chunk_gti_comment(\$chunk_text)";
1052 $key_to_gti_comment_mapping{$chunk_key} = $chunk_gti_comment if (defined($chunk_gti_comment));
[10019]1053 }
[24627]1054
[12483]1055 return %key_to_gti_comment_mapping;
[10019]1056}
1057
1058
1059sub determine_chunks_requiring_translation
1060{
1061 my $source_file_key_to_text_mapping = shift(@_);
1062 my $target_file_key_to_text_mapping = shift(@_);
[24627]1063
[10019]1064 # Chunks needing translation are those in the source file with no translation in the target file
1065 my @target_file_keys_requiring_translation = ();
1066 foreach my $chunk_key (keys(%$source_file_key_to_text_mapping)) {
[24627]1067 if ($source_file_key_to_text_mapping->{$chunk_key} && !$target_file_key_to_text_mapping->{$chunk_key}) {
1068 # &log_message("Chunk with key $chunk_key needs translating.");
1069 push(@target_file_keys_requiring_translation, $chunk_key);
1070 }
[10019]1071 }
[24627]1072
[10019]1073 return @target_file_keys_requiring_translation;
1074}
1075
1076
1077sub determine_chunks_requiring_updating
1078{
1079 my $source_file_key_to_last_update_date_mapping = shift(@_);
1080 my $target_file_key_to_last_update_date_mapping = shift(@_);
[24627]1081
[10019]1082 # Chunks needing updating are those in the target file that have been more recently edited in the source file
1083 my @target_file_keys_requiring_updating = ();
1084 foreach my $chunk_key (keys(%$source_file_key_to_last_update_date_mapping)) {
[24627]1085 my $source_chunk_last_update_date = $source_file_key_to_last_update_date_mapping->{$chunk_key};
1086 my $target_chunk_last_update_date = $target_file_key_to_last_update_date_mapping->{$chunk_key};
[14258]1087
[30548]1088 # print "key: $chunk_key\nsource date : $source_chunk_last_update_date\ntarget date : $target_chunk_last_update_date\nafter? ". &is_date_after($source_chunk_last_update_date, $target_chunk_last_update_date) . "\n\n";
1089
[14258]1090 if (defined($target_chunk_last_update_date) && &is_date_after($source_chunk_last_update_date, $target_chunk_last_update_date)) {
[24627]1091 # &log_message("Chunk with key $chunk_key needs updating.");
[30548]1092 # &log_message("key: $chunk_key\nsource date : $source_chunk_last_update_date\ntarget date : $target_chunk_last_update_date\nafter? ". &is_date_after($source_chunk_last_update_date, $target_chunk_last_update_date) . "\n\n");
[24627]1093 push(@target_file_keys_requiring_updating, $chunk_key);
1094 }
[10019]1095 }
[24627]1096
[10019]1097 return @target_file_keys_requiring_updating;
1098}
1099
1100
1101sub is_chunk_automatically_translated
1102{
1103 my ($chunk_key, $translation_file_type) = @_;
1104 eval "return &is_${translation_file_type}_chunk_automatically_translated(\$chunk_key)";
1105}
1106
1107
1108sub make_text_xml_safe
1109{
1110 my $text = shift(@_);
[11452]1111 $text =~ s/\&/\&amp\;/g;
1112 $text =~ s/\&amp\;lt\;/\&amp\;amp\;lt\;/g;
1113 $text =~ s/\&amp\;gt\;/\&amp\;amp\;gt\;/g;
[11498]1114 $text =~ s/\&amp\;rarr\;/\&amp\;amp\;rarr\;/g;
1115 $text =~ s/\&amp\;mdash\;/\&amp\;amp\;mdash\;/g;
[10019]1116 $text =~ s/</\&lt\;/g;
1117 $text =~ s/>/\&gt\;/g;
1118 return $text;
1119}
1120
1121
[11448]1122sub unmake_text_xml_safe
1123{
1124 my $text = shift(@_);
1125 $text =~ s/\&lt\;/</g;
1126 $text =~ s/\&gt\;/>/g;
1127 $text =~ s/\&amp\;/\&/g;
1128 return $text;
1129}
1130
1131
[10019]1132# Returns 1 if $date1 is after $date2, 0 otherwise
[14258]1133sub is_date_after_cvs
[10019]1134{
1135 my ($date1, $date2) = @_;
1136 my %months = ("Jan", 1, "Feb", 2, "Mar", 3, "Apr", 4, "May", 5, "Jun", 6,
[24627]1137 "Jul", 7, "Aug", 8, "Sep", 9, "Oct", 10, "Nov", 11, "Dec", 12);
1138
[14258]1139 if(!defined $date1) {
1140 return 1;
1141 }
[24627]1142
[10019]1143 my @date1parts = split(/-/, $date1);
1144 my @date2parts = split(/-/, $date2);
[24627]1145
[10019]1146 # Compare year - nasty because we have rolled over into a new century
1147 my $year1 = $date1parts[2];
1148 if ($year1 < 80) {
[14258]1149 $year1 += 2000;
[10019]1150 }
1151 my $year2 = $date2parts[2];
1152 if ($year2 < 80) {
[14258]1153 $year2 += 2000;
[10019]1154 }
[24627]1155
[10019]1156 # Compare year
1157 if ($year1 > $year2) {
[24627]1158 return 1;
[10019]1159 }
1160 elsif ($year1 == $year2) {
[24627]1161 # Year is the same, so compare month
1162 if ($months{$date1parts[1]} > $months{$date2parts[1]}) {
1163 return 1;
1164 }
1165 elsif ($months{$date1parts[1]} == $months{$date2parts[1]}) {
1166 # Month is the same, so compare day
1167 if ($date1parts[0] > $date2parts[0]) {
1168 return 1;
1169 }
1170 }
[10019]1171 }
[24627]1172
[10019]1173 return 0;
1174}
1175
[14258]1176sub is_date_after
1177{
1178 my ($date1, $date2) = @_;
1179
1180 if(!defined $date1) {
[24627]1181 return 1;
[14258]1182 }
1183 if(!defined $date2) {
[24627]1184 return 0;
[14258]1185 }
1186
1187 # 16-Aug-2006
1188 if($date1=~ /(\d+?)-(\S\S\S)-(\d\d\d\d)/){
[24627]1189 my %months = ("Jan", "01", "Feb", "02", "Mar", "03", "Apr", "04", "May", "05", "Jun", "06",
1190 "Jul", "07", "Aug", "08", "Sep", "09", "Oct", "10", "Nov", "11", "Dec", "12");
1191 $date1=$3 . "-" . $months{$2} . "-" . $1;
1192 # print "** converted date1: $date1\n";
[14258]1193 }
1194 if($date2=~ /(\d+?)-(\S\S\S)-(\d\d\d\d)/){
[24627]1195 my %months = ("Jan", "01", "Feb", "02", "Mar", "03", "Apr", "04", "May", "05", "Jun", "06",
1196 "Jul", "07", "Aug", "08", "Sep", "09", "Oct", "10", "Nov", "11", "Dec", "12");
1197 $date2=$3 . "-" . $months{$2} . "-" . $1;
1198 # print "** converted date2: $date2\n";
[14258]1199 }
1200
1201
1202 # 2006-08-16
1203 my @date1parts = split(/-/, $date1);
1204 my @date2parts = split(/-/, $date2);
1205
1206 # Compare year
1207 if ($date1parts[0] > $date2parts[0]) {
[24627]1208 return 1;
[14258]1209 }
1210 elsif ($date1parts[0] == $date2parts[0]) {
[24627]1211 # Year is the same, so compare month
1212 if ($date1parts[1] > $date2parts[1]) {
1213 return 1;
1214 }
1215 elsif ($date1parts[1] == $date2parts[1]) {
1216 # Month is the same, so compare day
1217 if ($date1parts[2] > $date2parts[2]) {
1218 return 1;
1219 }
1220 }
[14258]1221 }
1222
1223 return 0;
1224}
[10019]1225
[14258]1226
[18460]1227sub create_xml_response_for_chunks_requiring_work
1228{
1229 my ($translation_file_key, $target_file, $total_num_chunks, $target_files_keys_requiring_translation, $target_files_keys_requiring_updating, $num_chunks_to_return, $source_files_key_to_text_mapping, $target_files_key_to_text_mapping, $source_files_key_to_last_update_date_mapping, $target_files_key_to_last_update_date_mapping) = @_;
[24627]1230
[18460]1231 # Form an XML response to the command
1232 my $xml_response = "<?xml version=\"1.0\" encoding=\"UTF-8\" ?>\n";
1233 $xml_response .= "<GTIResponse>\n";
1234 $xml_response .= " <TranslationFile"
1235 . " key=\"" . $translation_file_key . "\""
1236 . " target_file_path=\"" . $target_file . "\""
1237 . " num_chunks_translated=\"" . ($total_num_chunks - scalar(@$target_files_keys_requiring_translation)) . "\""
1238 . " num_chunks_requiring_translation=\"" . scalar(@$target_files_keys_requiring_translation) . "\""
1239 . " num_chunks_requiring_updating=\"" . scalar(@$target_files_keys_requiring_updating) . "\"\/>\n";
[24627]1240
[18460]1241 # Do chunks requiring translation first
1242 if ($num_chunks_to_return > scalar(@$target_files_keys_requiring_translation)) {
[24627]1243 $xml_response .= " <ChunksRequiringTranslation size=\"" . scalar(@$target_files_keys_requiring_translation) . "\">\n";
[18460]1244 }
1245 else {
[24627]1246 $xml_response .= " <ChunksRequiringTranslation size=\"" . $num_chunks_to_return . "\">\n";
[18460]1247 }
[24627]1248
[18460]1249 my @sorted_chunk_keys = sort (@$target_files_keys_requiring_translation);
1250 foreach my $chunk_key (@sorted_chunk_keys) {
[24627]1251 last if ($num_chunks_to_return == 0);
1252
1253 my $source_file_chunk_date = $source_files_key_to_last_update_date_mapping->{$chunk_key} || "";
1254 my $source_file_chunk_text = &make_text_xml_safe($source_files_key_to_text_mapping->{$chunk_key});
1255
1256 $xml_response .= " <Chunk key=\"" . &make_text_xml_safe($chunk_key) . "\">\n";
1257 $xml_response .= " <SourceFileText date=\"$source_file_chunk_date\">$source_file_chunk_text</SourceFileText>\n";
1258 $xml_response .= " <TargetFileText></TargetFileText>\n";
1259 $xml_response .= " </Chunk>\n";
1260
1261 $num_chunks_to_return--;
1262 }
[18460]1263
1264 $xml_response .= " </ChunksRequiringTranslation>\n";
[24627]1265
[18460]1266 # Then do chunks requiring updating
1267 if ($num_chunks_to_return > scalar(@$target_files_keys_requiring_updating)) {
[24627]1268 $xml_response .= " <ChunksRequiringUpdating size=\"" . scalar(@$target_files_keys_requiring_updating) . "\">\n";
[18460]1269 }
1270 else {
[24627]1271 $xml_response .= " <ChunksRequiringUpdating size=\"" . $num_chunks_to_return . "\">\n";
[18460]1272 }
[24627]1273
[18460]1274 # foreach my $chunk_key (@target_file_keys_requiring_updating) {
1275 @sorted_chunk_keys = sort (@$target_files_keys_requiring_updating);
1276 foreach my $chunk_key (@sorted_chunk_keys) {
[24627]1277 last if ($num_chunks_to_return == 0);
1278
1279 my $source_file_chunk_date = $source_files_key_to_last_update_date_mapping->{$chunk_key} || "";
1280 my $source_file_chunk_text = &make_text_xml_safe($source_files_key_to_text_mapping->{$chunk_key});
1281 my $target_file_chunk_date = $target_files_key_to_last_update_date_mapping->{$chunk_key} || "";
1282 my $target_file_chunk_text = &make_text_xml_safe($target_files_key_to_text_mapping->{$chunk_key});
1283
1284 $xml_response .= " <Chunk key=\"" . &make_text_xml_safe($chunk_key) . "\">\n";
1285 $xml_response .= " <SourceFileText date=\"$source_file_chunk_date\">$source_file_chunk_text</SourceFileText>\n";
1286 $xml_response .= " <TargetFileText date=\"$target_file_chunk_date\">$target_file_chunk_text</TargetFileText>\n";
1287 $xml_response .= " </Chunk>\n";
1288
1289 $num_chunks_to_return--;
[18460]1290 }
[24627]1291
[18460]1292 $xml_response .= " </ChunksRequiringUpdating>\n";
[24627]1293
[18460]1294 $xml_response .= "</GTIResponse>\n";
[24627]1295
[18460]1296 return $xml_response;
1297}
1298
[25249]1299sub create_xml_response_for_uptodate_chunks
1300{
1301 my ($translation_file_key, $target_file, $uptodate_target_files_keys, $source_files_key_to_text_mapping, $target_files_key_to_text_mapping, $source_files_key_to_last_update_date_mapping, $target_files_key_to_last_update_date_mapping) = @_;
1302
1303 # Form an XML response to the command
1304 my $xml_response = "<?xml version=\"1.0\" encoding=\"UTF-8\" ?>\n";
1305 $xml_response .= "<GTIResponse>\n";
1306 $xml_response .= " <TranslationFile"
1307 . " key=\"" . $translation_file_key . "\""
1308 . " target_file_path=\"" . $target_file . "\""
1309 . " num_chunks_uptodate=\"" . scalar(@$uptodate_target_files_keys) . "\"\/>\n";
1310
1311
1312 # Then do chunks requiring updating
1313 $xml_response .= " <UptodateChunks size=\"" . scalar(@$uptodate_target_files_keys) . "\">\n";
1314
1315
1316 # foreach my $chunk_key (@uptodate_target_file_keys) {
1317 my @sorted_chunk_keys = sort (@$uptodate_target_files_keys);
1318 foreach my $chunk_key (@sorted_chunk_keys) {
1319
1320 my $source_file_chunk_date = $source_files_key_to_last_update_date_mapping->{$chunk_key} || "";
1321 my $source_file_chunk_text = &make_text_xml_safe($source_files_key_to_text_mapping->{$chunk_key});
1322 my $target_file_chunk_date = $target_files_key_to_last_update_date_mapping->{$chunk_key} || "";
1323 my $target_file_chunk_text = &make_text_xml_safe($target_files_key_to_text_mapping->{$chunk_key});
1324
1325 $xml_response .= " <Chunk key=\"" . &make_text_xml_safe($chunk_key) . "\">\n";
1326 $xml_response .= " <SourceFileText date=\"$source_file_chunk_date\">$source_file_chunk_text</SourceFileText>\n";
1327 $xml_response .= " <TargetFileText date=\"$target_file_chunk_date\">$target_file_chunk_text</TargetFileText>\n";
1328 $xml_response .= " </Chunk>\n";
[18460]1329
[25249]1330 }
1331
1332 $xml_response .= " </UptodateChunks>\n";
1333
1334 $xml_response .= "</GTIResponse>\n";
1335
1336 return $xml_response;
1337}
1338
[18460]1339sub create_xml_response_for_all_chunks
1340{
1341 my ($translation_file_key, $target_file, $source_file_key_to_text_mapping, $target_file_key_to_text_mapping, $source_file_key_to_last_update_date_mapping, $target_file_key_to_last_update_date_mapping) = @_;
[24627]1342
[18460]1343 # Form an XML response to the command
1344 my $xml_response = "<?xml version=\"1.0\" encoding=\"UTF-8\" ?>\n";
1345 $xml_response .= "<GTIResponse>\n";
1346 $xml_response .= " <TranslationFile"
1347 . " key=\"" . $translation_file_key . "\""
1348 . " target_file_path=\"" . $target_file . "\"\/>\n";
1349
1350 # Do all the chunks
1351 $xml_response .= " <Chunks size=\"" . scalar(keys(%$source_file_key_to_text_mapping)) . "\">\n";
[24627]1352
[18460]1353 my @sorted_chunk_keys = sort (keys(%$source_file_key_to_text_mapping));
1354 foreach my $chunk_key (@sorted_chunk_keys) {
[24627]1355 my $source_file_chunk_date = $source_file_key_to_last_update_date_mapping->{$chunk_key} || "";
1356 my $source_file_chunk_text = &make_text_xml_safe($source_file_key_to_text_mapping->{$chunk_key});
1357
1358 $xml_response .= " <Chunk key=\"" . &make_text_xml_safe($chunk_key) . "\">\n";
1359 $xml_response .= " <SourceFileText date=\"$source_file_chunk_date\">$source_file_chunk_text</SourceFileText>\n";
1360 if (defined($target_file_key_to_text_mapping->{$chunk_key})) {
1361 my $target_file_chunk_date = $target_file_key_to_last_update_date_mapping->{$chunk_key} || "";
1362 my $target_file_chunk_text = &make_text_xml_safe($target_file_key_to_text_mapping->{$chunk_key});
1363 $xml_response .= " <TargetFileText date=\"$target_file_chunk_date\">$target_file_chunk_text</TargetFileText>\n";
1364 }
1365 else {
1366 $xml_response .= " <TargetFileText></TargetFileText>\n";
1367 }
1368
1369 $xml_response .= " </Chunk>\n";
[18460]1370 }
1371 $xml_response .= " </Chunks>\n";
1372
1373 $xml_response .= "</GTIResponse>\n";
1374 return $xml_response;
1375}
1376
1377
1378
[10019]1379# ==========================================================================================
1380# MACROFILE FUNCTIONS
1381
1382sub build_key_to_line_mapping_for_macrofile
1383{
1384 my (@file_lines) = @_;
[24627]1385
[10019]1386 my $macro_package;
1387 my %chunk_key_to_line_mapping = ();
1388 # Process the contents of the file, line by line
1389 for (my $i = 0; $i < scalar(@file_lines); $i++) {
[24627]1390 my $line = $file_lines[$i];
1391 $line =~ s/(\s*)$//; # Remove any nasty whitespace, carriage returns etc.
1392
1393 # Check if a new package is being defined
1394 if ($line =~ m/^package\s+(.+)/) {
1395 $macro_package = $1;
[10041]1396 }
[24627]1397
1398 # Line contains a macro name
1399 elsif ($line =~ m/^(_\w+_)/) {
1400 my $macro_key = $1;
1401 $line =~ s/\s*([^\\]\#[^\}]+)?$//; # Remove any comments and nasty whitespace
1402
1403 # While there is still text of the macro to go...
1404 my $startline = $i;
1405 while ($line !~ /\}$/) {
1406 $i++;
1407 if ($i == scalar(@file_lines)) {
1408 &throw_fatal_error("Could not find end of macro $macro_key.");
1409 }
1410 $line = $file_lines[$i];
1411 $line =~ s/\s*([^\\]\#[^\}]+)?$//; # Remove any comments and nasty whitespace
1412 }
1413
[10019]1414 # The chunk key consists of the package name and the macro key
1415 my $chunk_key = $macro_package . "." . $macro_key;
1416 # Map from chunk key to line
1417 $chunk_key_to_line_mapping{$chunk_key} = $startline . "-" . $i;
1418 }
[24627]1419
[10019]1420 # Icon: line in format ## "image text" ## image_type ## macro_name ##
1421 elsif ($line =~ m/^\#\# .* \#\# .* \#\# (.*) \#\#/) {
[24627]1422 # The chunk key consists of package name and macro key
1423 my $chunk_key = $macro_package . "." . $1;
1424 # Map from chunk key to line
1425 $chunk_key_to_line_mapping{$chunk_key} = $i . "-" . $i;
1426}
1427}
[10019]1428
[24627]1429return %chunk_key_to_line_mapping;
[10019]1430}
1431
1432
1433sub import_chunk_from_macrofile
1434{
1435 my ($chunk_text) = @_;
[24627]1436
[10019]1437 # Is this an icon macro??
1438 if ($chunk_text =~ /^\#\# (.*)/) {
[24627]1439 # Extract image macro text
1440 $chunk_text =~ /^\#\#\s+([^\#]+)\s+\#\#/;
1441 $chunk_text = $1;
1442
[10019]1443 # Remove enclosing quotes
1444 $chunk_text =~ s/^\"//;
1445 $chunk_text =~ s/\"$//;
[30562]1446 }
[10019]1447
[30562]1448 # No, so it must be a text macro
1449 else {
[10019]1450 # Remove macro key
1451 $chunk_text =~ s/^_([^_]+)_(\s*)//;
[24627]1452
[10019]1453 # Remove language specifier
[30562]1454 $chunk_text =~ s/^\[l=[^\]]*\](\s*)//; # only remove until first closing square bracket, ]
[24627]1455
[10019]1456 # Remove braces enclosing text
1457 $chunk_text =~ s/^{(\s*)((.|\n)*)}(\s*)(\#.+\s*)?/$2/;
[30562]1458 }
[10019]1459
[30562]1460 return $chunk_text;
[10019]1461}
1462
1463
[12483]1464sub get_macrofile_chunk_gti_comment
[10019]1465{
1466 my ($chunk_text) = @_;
[24627]1467
[10019]1468 # Check for an "Updated DD-MMM-YYYY" comment at the end of the chunk
[12484]1469 if ($chunk_text =~ /\#\s+(Updated\s+\d?\d-\D\D\D-\d\d\d\d.*)\s*$/i) {
[24627]1470 return $1;
1471}
[10019]1472
[24627]1473return undef;
[10019]1474}
1475
1476
1477sub is_macrofile_chunk_automatically_translated
1478{
1479 my ($chunk_key) = @_;
[24627]1480
[10019]1481 # The _httpiconX_, _widthX_ and _heightX_ image macros are automatically translated
1482 if ($chunk_key =~ /\._(httpicon|width|height)/) {
[24627]1483 return 1;
[10019]1484 }
[24627]1485
[10019]1486 return 0;
1487}
1488
1489
1490# Use the source file to generate a target file that is formatted the same
1491sub write_translated_macrofile
1492{
1493 my $source_file = shift(@_); # Not used
[11321]1494 my @source_file_lines = @{shift(@_)};
[10019]1495 my $source_file_key_to_text_mapping = shift(@_);
1496 my $target_file = shift(@_);
[11321]1497 my @target_file_lines = @{shift(@_)};
[10019]1498 my $target_file_key_to_text_mapping = shift(@_);
[12483]1499 my $target_file_key_to_gti_comment_mapping = shift(@_);
[10019]1500 my $target_language_code = shift(@_);
[24627]1501
[10126]1502 # Build a mapping from source file line to chunk key
[11321]1503 my %source_file_key_to_line_mapping = &build_key_to_line_mapping_for_macrofile(@source_file_lines);
[10019]1504 my %source_file_line_to_key_mapping = ();
1505 foreach my $chunk_key (keys(%source_file_key_to_line_mapping)) {
[24627]1506 $source_file_line_to_key_mapping{$source_file_key_to_line_mapping{$chunk_key}} = $chunk_key;
[10019]1507 }
[10126]1508 my @source_file_line_keys = (sort sort_by_line (keys(%source_file_line_to_key_mapping)));
1509 my $source_file_line_number = 0;
[24627]1510
[10126]1511 # Build a mapping from target file line to chunk key
[11321]1512 my %target_file_key_to_line_mapping = &build_key_to_line_mapping_for_macrofile(@target_file_lines);
[10126]1513 my %target_file_line_to_key_mapping = ();
1514 foreach my $chunk_key (keys(%target_file_key_to_line_mapping)) {
[24627]1515 $target_file_line_to_key_mapping{$target_file_key_to_line_mapping{$chunk_key}} = $chunk_key;
[10126]1516 }
1517 my @target_file_line_keys = (sort sort_by_line (keys(%target_file_line_to_key_mapping)));
[24627]1518
[10019]1519 # Write the new target file
1520 my $target_file_path = &util::filename_cat($gsdl_root_directory, $target_file);
1521 if (!open(TARGET_FILE, ">$target_file_path")) {
[24627]1522 &throw_fatal_error("Could not write target file $target_file_path.");
[10019]1523 }
[24627]1524
[10126]1525 # Use the header from the target file, to keep language and author information
1526 if (scalar(@target_file_line_keys) > 0) {
[24627]1527 my $target_file_line_number = 0;
1528 my $target_file_chunk_starting_line_number = (split(/-/, $target_file_line_keys[0]))[0];
1529 while ($target_file_line_number < $target_file_chunk_starting_line_number) {
1530 my $target_file_line = $target_file_lines[$target_file_line_number];
1531 last if ($target_file_line =~ /^\# -- Missing translation: /); # We don't want to get into the macros
1532 print TARGET_FILE $target_file_line;
1533 $target_file_line_number++;
1534 }
1535
1536 $source_file_line_number = (split(/-/, $source_file_line_keys[0]))[0];
[10126]1537 }
[24627]1538
[10019]1539 # Model the new target file on the source file, with the target file translations
[10126]1540 foreach my $line_key (@source_file_line_keys) {
[24627]1541 # Fill in the gaps before this chunk starts
1542 my $source_file_chunk_starting_line_number = (split(/-/, $line_key))[0];
1543 my $source_file_chunk_finishing_line_number = (split(/-/, $line_key))[1];
1544 while ($source_file_line_number < $source_file_chunk_starting_line_number) {
1545 print TARGET_FILE $source_file_lines[$source_file_line_number];
1546 $source_file_line_number++;
1547 }
1548 $source_file_line_number = $source_file_chunk_finishing_line_number + 1;
1549
1550 my $chunk_key = $source_file_line_to_key_mapping{$line_key};
1551 my $source_file_chunk_text = $source_file_key_to_text_mapping->{$chunk_key};
1552 my $target_file_chunk_text = $target_file_key_to_text_mapping->{$chunk_key} || "";
1553
1554 my $macrofile_key = $chunk_key;
1555 $macrofile_key =~ s/^(.+?)\.//;
1556
1557 # If no translation exists for this chunk, show this, and move on
1558 if ($source_file_chunk_text ne "" && $target_file_chunk_text eq "") {
1559 print TARGET_FILE "# -- Missing translation: $macrofile_key\n";
1560 next;
1561 }
1562
1563 # Grab the source chunk text
1564 my $source_file_chunk = $source_file_lines[$source_file_chunk_starting_line_number];
1565 for (my $l = ($source_file_chunk_starting_line_number + 1); $l <= $source_file_chunk_finishing_line_number; $l++) {
1566 $source_file_chunk .= $source_file_lines[$l];
1567 }
1568
1569 # Is this an icon macro??
1570 if ($source_file_chunk =~ /^\#\# (.*)/) {
1571 # Escape any newline and question mark characters so the source text is replaced correctly
1572 $source_file_chunk_text =~ s/\\/\\\\/g;
[10019]1573 $source_file_chunk_text =~ s/\?/\\\?/g;
[24627]1574
[10019]1575 # Build the new target chunk from the source chunk
1576 my $target_file_chunk = $source_file_chunk;
1577 $target_file_chunk =~ s/$source_file_chunk_text/$target_file_chunk_text/;
[10050]1578 $target_file_chunk =~ s/(\s)*$//;
[10019]1579 print TARGET_FILE "$target_file_chunk";
1580 }
[24627]1581
[10019]1582 # No, it is just a normal text macro
1583 else {
1584 print TARGET_FILE "$macrofile_key [l=$target_language_code] {$target_file_chunk_text}";
1585 }
[24627]1586
[12483]1587 # Add the "updated" comment, if one exists
1588 if ($target_file_key_to_gti_comment_mapping->{$chunk_key}) {
1589 print TARGET_FILE " # " . $target_file_key_to_gti_comment_mapping->{$chunk_key};
[10050]1590 }
1591 print TARGET_FILE "\n";
[24627]1592}
[10019]1593
[24627]1594close(TARGET_FILE);
[10019]1595}
1596
1597
1598sub sort_by_line
1599{
1600 return ((split(/-/, $a))[0] <=> (split(/-/, $b))[0]);
1601}
1602
1603
1604# ==========================================================================================
1605# RESOURCE BUNDLE FUNCTIONS
1606
[29456]1607# need to handle multi-line properties. A multiline ends on \ if it continues over the next line
[10019]1608sub build_key_to_line_mapping_for_resource_bundle
1609{
1610 my (@file_lines) = @_;
[24627]1611
[11321]1612 my %chunk_key_to_line_mapping = ();
[29456]1613
1614 my $chunk_key;
1615 my $startindex = -1;
1616
[10019]1617 for (my $i = 0; $i < scalar(@file_lines); $i++) {
[24627]1618 my $line = $file_lines[$i];
1619 $line =~ s/(\s*)$//; # Remove any nasty whitespace, carriage returns etc.
1620
[30719]1621 # a property line has a colon/equals sign as separator that is NOT escaped with a backslash (both keys and values
1622 # can use the colon or = sign. But in the key, such a char is always escaped. Unfortunately, they've not always been
1623 # escaped in the values. So we get the left most occurrence by not doing a greedy match (use ? to not be greedy).
1624 # So find the first :/= char not preceded by \. That will be the true separator of a chunk_key and its value chunk_text
[30687]1625
[30735]1626 if ($line =~ m/^(\S*?[^\\])[:|=](.*)$/) {
[30719]1627 # Line contains a dictionary string
[30687]1628
[30719]1629 # Unused but useful: http://stackoverflow.com/questions/87380/how-can-i-find-the-location-of-a-regex-match-in-perl
1630 # http://perldoc.perl.org/perlvar.html
[30687]1631
[30719]1632 $chunk_key = $1;
1633 # remove the escaping of any :/= property separator from the chunk_key in memory,
1634 # to make comparison with its unescaped version during submissions easier. Will write out with escaping.
1635 $chunk_key =~ s/\\([:=])/$1/g;
1636
[29456]1637 $startindex = $i;
1638 }
1639 if ($startindex != -1) {
1640 if($line !~ m/\\$/) { # line finished
1641 # $i keeps track of the line at which this property (chunk_key) finishes
1642
[24627]1643 # Map from chunk key to line
[29456]1644 $chunk_key_to_line_mapping{$chunk_key} = $startindex . "-" . $i;
1645 $startindex = -1;
1646 $chunk_key = "";
1647 }
[30687]1648 }
[10019]1649 }
[24627]1650
[11321]1651 return %chunk_key_to_line_mapping;
[10019]1652}
1653
1654
1655sub import_chunk_from_resource_bundle
1656{
1657 my ($chunk_text) = @_;
[24627]1658
[30719]1659 # Simple: just remove string key.
1660 # But key can contain an escaped separator (\: or \=).
1661 # So just as in the previous subroutine, find the first (leftmost) : or = char not preceded by \.
1662 # That will be the true separator of a chunk_key and its value chunk_text
[30735]1663 $chunk_text =~ s/^(\S*?[^\\])[:|=](\s*)//s;
[30687]1664
[29456]1665 $chunk_text =~ s/(\s*)$//s; # Remove any nasty whitespace, carriage returns etc.
1666 $chunk_text =~ s/(\s*)\#\s+Updated\s+(\d?\d-\D\D\D-\d\d\d\d.*)\s*$//is;
[24627]1667
[10019]1668 return $chunk_text;
1669}
1670
1671
[12483]1672sub get_resource_bundle_chunk_gti_comment
[10019]1673{
1674 my ($chunk_text) = @_;
[24627]1675
[10019]1676 # Check for an "Updated DD-MMM-YYYY" comment at the end of the chunk
[12484]1677 if ($chunk_text =~ /\#\s+(Updated\s+\d?\d-\D\D\D-\d\d\d\d.*)\s*$/i) {
[24627]1678 return $1;
[30719]1679 }
[10019]1680
[30719]1681 return undef;
[10019]1682}
1683
1684
1685sub is_resource_bundle_chunk_automatically_translated
1686{
1687 # No resource bundle chunks are automatically translated
1688 return 0;
1689}
1690
1691
1692sub write_translated_resource_bundle
1693{
1694 my $source_file = shift(@_); # Not used
[11321]1695 my @source_file_lines = @{shift(@_)};
[10019]1696 my $source_file_key_to_text_mapping = shift(@_);
1697 my $target_file = shift(@_);
[11321]1698 my @target_file_lines = @{shift(@_)}; # Not used
[10019]1699 my $target_file_key_to_text_mapping = shift(@_);
[12483]1700 my $target_file_key_to_gti_comment_mapping = shift(@_);
[10019]1701 my $target_language_code = shift(@_); # Not used
[24627]1702
[10019]1703 # Build a mapping from chunk key to source file line, and from source file line to chunk key
[11321]1704 my %source_file_key_to_line_mapping = &build_key_to_line_mapping_for_resource_bundle(@source_file_lines);
[10019]1705 my %source_file_line_to_key_mapping = ();
1706 foreach my $chunk_key (keys(%source_file_key_to_line_mapping)) {
[24627]1707 $source_file_line_to_key_mapping{$source_file_key_to_line_mapping{$chunk_key}} = $chunk_key;
[10019]1708 }
[24627]1709
[10019]1710 # Write the new target file
1711 my $target_file_path = &util::filename_cat($gsdl_root_directory, $target_file);
1712 if (!open(TARGET_FILE, ">$target_file_path")) {
[24627]1713 &throw_fatal_error("Could not write target file $target_file_path.");
[10019]1714 }
[24627]1715
[10019]1716 # Model the new target file on the source file, with the target file translations
[10091]1717 my $source_file_line_number = 0;
[10019]1718 foreach my $line_key (sort sort_by_line (keys(%source_file_line_to_key_mapping))) {
[24627]1719 # Fill in the gaps before this chunk starts
1720 my $source_file_chunk_starting_line_number = (split(/-/, $line_key))[0];
1721 my $source_file_chunk_finishing_line_number = (split(/-/, $line_key))[1];
1722 while ($source_file_line_number < $source_file_chunk_starting_line_number) {
1723 print TARGET_FILE $source_file_lines[$source_file_line_number];
1724 $source_file_line_number++;
1725 }
1726 $source_file_line_number = $source_file_chunk_finishing_line_number + 1;
1727
1728 my $chunk_key = $source_file_line_to_key_mapping{$line_key};
1729 my $source_file_chunk_text = $source_file_key_to_text_mapping->{$chunk_key};
1730 my $target_file_chunk_text = $target_file_key_to_text_mapping->{$chunk_key} || "";
1731
[30719]1732 # make sure any : or = sign in the chunk key is escaped again (with \) when written out
1733 # since the key-value separator in a property resource bundle file is : or =
1734 my $escaped_chunk_key = $chunk_key;
1735 $escaped_chunk_key =~ s/(:|=)/\\$1/g; #$escaped_chunk_key =~ s/([^\\])(:|=)/\\$1$2/g;
1736
[24627]1737 # If no translation exists for this chunk, show this, and move on
1738 if ($source_file_chunk_text ne "" && $target_file_chunk_text eq "") {
[30719]1739 print TARGET_FILE "# -- Missing translation: $escaped_chunk_key\n";
[24627]1740 next;
1741 }
[30719]1742
1743 print TARGET_FILE "$escaped_chunk_key:$target_file_chunk_text";
[24627]1744 if ($target_file_key_to_gti_comment_mapping->{$chunk_key}) {
1745 print TARGET_FILE " # " . $target_file_key_to_gti_comment_mapping->{$chunk_key};
1746 }
1747 print TARGET_FILE "\n";
[10019]1748 }
[24627]1749
[10019]1750 close(TARGET_FILE);
1751}
1752
1753
[11321]1754# ==========================================================================================
1755# GREENSTONE XML FUNCTIONS
1756
1757sub build_key_to_line_mapping_for_greenstone_xml
1758{
1759 my (@file_lines) = @_;
[24627]1760
[11321]1761 my %chunk_key_to_line_mapping = ();
1762 for (my $i = 0; $i < scalar(@file_lines); $i++) {
[24627]1763 my $line = $file_lines[$i];
1764 $line =~ s/(\s*)$//; # Remove any nasty whitespace, carriage returns etc.
1765
1766 # Line contains a string to translate
1767 if ($line =~ /^\s*<Text id=\"(.*?)\">/) {
1768 my $chunk_key = $1;
1769 $line =~ s/\s*$//; # Remove any nasty whitespace
1770 $line =~ s/<Updated date=\"\d?\d-\D\D\D-\d\d\d\d.*\"\/>$//;
1771
1772 # While there is still text of the string to go...
1773 my $startline = $i;
1774 while ($line !~ /<\/Text>$/) {
1775 $i++;
1776 if ($i == scalar(@file_lines)) {
1777 &throw_fatal_error("Could not find end of string $chunk_key.");
1778 }
1779 $line = $file_lines[$i];
1780 $line =~ s/\s*$//; # Remove any nasty whitespace
1781 $line =~ s/<Updated date=\"\d?\d-\D\D\D-\d\d\d\d.*\"\/>$//;
1782 }
1783
1784 # Map from chunk key to line
1785 if (!defined($chunk_key_to_line_mapping{$chunk_key})) {
1786 $chunk_key_to_line_mapping{$chunk_key} = $startline . "-" . $i;
1787 }
1788 else {
1789 &throw_fatal_error("Duplicate key $chunk_key.");
1790 }
[11321]1791 }
1792 }
[24627]1793
[11321]1794 return %chunk_key_to_line_mapping;
1795}
1796
1797
1798sub import_chunk_from_greenstone_xml
1799{
1800 my ($chunk_text) = @_;
[24627]1801
[11321]1802 # Simple: just remove the Text tags
[11447]1803 $chunk_text =~ s/^\s*<Text id=\"(.*?)\">(\s*)//;
[12484]1804 $chunk_text =~ s/<Updated date=\"\d?\d-\D\D\D-\d\d\d\d.*\"\/>$//;
[11321]1805 $chunk_text =~ s/<\/Text>$//;
[24627]1806
[11321]1807 return $chunk_text;
1808}
1809
1810
[12483]1811sub get_greenstone_xml_chunk_gti_comment
[11321]1812{
1813 my ($chunk_text) = @_;
[24627]1814
[11321]1815 # Check for an "Updated DD-MMM-YYYY" comment at the end of the chunk
[12484]1816 if ($chunk_text =~ /<Updated date=\"(\d?\d-\D\D\D-\d\d\d\d.*)\"\/>$/i) {
[24627]1817 return $1;
[11487]1818 }
[24627]1819
[11321]1820 return undef;
1821}
1822
1823
1824sub is_greenstone_xml_chunk_automatically_translated
1825{
1826 # No greenstone XML chunks are automatically translated
1827 return 0;
1828}
1829
1830
1831sub write_translated_greenstone_xml
1832{
1833 my $source_file = shift(@_); # Not used
1834 my @source_file_lines = @{shift(@_)};
1835 my $source_file_key_to_text_mapping = shift(@_);
1836 my $target_file = shift(@_);
1837 my @target_file_lines = @{shift(@_)}; # Not used
1838 my $target_file_key_to_text_mapping = shift(@_);
[12483]1839 my $target_file_key_to_gti_comment_mapping = shift(@_);
[11321]1840 my $target_language_code = shift(@_); # Not used
[24627]1841
[11321]1842 # Build a mapping from chunk key to source file line, and from source file line to chunk key
1843 my %source_file_key_to_line_mapping = &build_key_to_line_mapping_for_greenstone_xml(@source_file_lines);
1844 my %source_file_line_to_key_mapping = ();
1845 foreach my $chunk_key (keys(%source_file_key_to_line_mapping)) {
[24627]1846 $source_file_line_to_key_mapping{$source_file_key_to_line_mapping{$chunk_key}} = $chunk_key;
[11321]1847 }
[24627]1848
[11321]1849 # Write the new target file
1850 my $target_file_path = &util::filename_cat($gsdl_root_directory, $target_file);
1851 if (!open(TARGET_FILE, ">$target_file_path")) {
[24627]1852 &throw_fatal_error("Could not write target file $target_file_path.");
[11321]1853 }
[24627]1854
[11321]1855 # Model the new target file on the source file, with the target file translations
1856 my $source_file_line_number = 0;
1857 foreach my $line_key (sort sort_by_line (keys(%source_file_line_to_key_mapping))) {
[24627]1858 # Fill in the gaps before this chunk starts
1859 my $source_file_chunk_starting_line_number = (split(/-/, $line_key))[0];
1860 my $source_file_chunk_finishing_line_number = (split(/-/, $line_key))[1];
1861 while ($source_file_line_number < $source_file_chunk_starting_line_number) {
1862 print TARGET_FILE $source_file_lines[$source_file_line_number];
1863 $source_file_line_number++;
1864 }
1865 $source_file_line_number = $source_file_chunk_finishing_line_number + 1;
1866
1867 my $chunk_key = $source_file_line_to_key_mapping{$line_key};
1868 my $source_file_chunk_text = $source_file_key_to_text_mapping->{$chunk_key};
1869 my $target_file_chunk_text = $target_file_key_to_text_mapping->{$chunk_key} || "";
1870 $target_file_chunk_text =~ s/(\n)*$//g;
1871
1872 # If no translation exists for this chunk, show this, and move on
1873 if ($source_file_chunk_text ne "" && $target_file_chunk_text eq "") {
1874 print TARGET_FILE "<!-- Missing translation: $chunk_key -->\n";
1875 next;
1876 }
1877
1878 print TARGET_FILE "<Text id=\"$chunk_key\">$target_file_chunk_text</Text>";
1879 if ($target_file_key_to_gti_comment_mapping->{$chunk_key}) {
1880 my $chunk_gti_comment = $target_file_key_to_gti_comment_mapping->{$chunk_key};
1881 $chunk_gti_comment =~ s/^Updated //;
1882 print TARGET_FILE "<Updated date=\"" . $chunk_gti_comment . "\"\/>";
1883 }
1884 print TARGET_FILE "\n";
[11321]1885 }
[24627]1886
[11321]1887 # Fill in the end of the file
1888 while ($source_file_line_number < scalar(@source_file_lines)) {
[24627]1889 print TARGET_FILE $source_file_lines[$source_file_line_number];
1890 $source_file_line_number++;
[11321]1891 }
[24627]1892
[11321]1893 close(TARGET_FILE);
1894}
1895
1896
[18460]1897# ==========================================================================================
1898# GREENSTONE3 FUNCTIONS
1899
1900sub get_all_chunks_for_gs3
1901{
1902 # The code of the target language (ensure it is lowercase)
1903 my $target_language_code = lc(shift(@_));
1904 my $translation_file_key = lc(shift(@_));
[24627]1905
[18460]1906 # Check that the necessary arguments were supplied
1907 if (!$target_language_code) {
[24627]1908 &throw_fatal_error("Missing command argument.");
[18460]1909 }
[24627]1910
[18460]1911 # Get (and check) the translation configuration
1912 # my ($source_file_dir, $target_file, $translation_file_type) = &get_translation_configuration($target_language_code, $translation_file_key);
1913
1914 my %source_files_key_to_text_mapping = ();
1915 my %target_files_key_to_text_mapping = ();
1916 my %source_files_key_to_last_update_date_mapping = ();
1917 my %target_files_key_to_last_update_date_mapping = ();
[24627]1918
[30581]1919 &build_gs3_configuration($translation_file_key, $target_language_code, \%source_files_key_to_text_mapping, \%target_files_key_to_text_mapping, \%source_files_key_to_last_update_date_mapping, \%target_files_key_to_last_update_date_mapping);
[24627]1920
[18460]1921 &log_message("Total number of source chunks: " . scalar(keys(%source_files_key_to_text_mapping)));
1922 &log_message("Total number of target chunks: " . scalar(keys(%target_files_key_to_text_mapping)));
[24627]1923
[18460]1924 my $xml_response = &create_xml_response_for_all_chunks($translation_file_key, "", \%source_files_key_to_text_mapping, \%target_files_key_to_text_mapping, \%source_files_key_to_last_update_date_mapping, \%target_files_key_to_last_update_date_mapping);
1925 return $xml_response;
1926}
1927
1928
1929sub get_first_n_chunks_requiring_work_for_gs3
1930{
1931 # The code of the target language (ensure it is lowercase)
1932 my $target_language_code = lc(shift(@_));
1933 # The key of the file to translate (ensure it is lowercase)
1934 my $translation_file_key = lc(shift(@_));
1935 # The number of chunks to return (defaults to one if not specified)
1936 my $num_chunks_to_return = shift(@_) || "1";
1937
1938 # Check that the necessary arguments were supplied
1939 if (!$target_language_code || !$translation_file_key) {
[24627]1940 &throw_fatal_error("Missing command argument.");
[18460]1941 }
[28755]1942
[18460]1943 my %source_files_key_to_text_mapping = ();
1944 my %target_files_key_to_text_mapping = ();
1945 my %source_files_key_to_last_update_date_mapping = ();
1946 my %target_files_key_to_last_update_date_mapping = ();
[24627]1947
[30581]1948 &build_gs3_configuration($translation_file_key, $target_language_code, \%source_files_key_to_text_mapping, \%target_files_key_to_text_mapping,
[24627]1949 \%source_files_key_to_last_update_date_mapping, \%target_files_key_to_last_update_date_mapping);
[18460]1950
1951 # Determine the target file chunks requiring translation
1952 my @target_files_keys_requiring_translation = &determine_chunks_requiring_translation(\%source_files_key_to_text_mapping, \%target_files_key_to_text_mapping);
1953 # Determine the target file chunks requiring updating
1954 my @target_files_keys_requiring_updating = &determine_chunks_requiring_updating(\%source_files_key_to_last_update_date_mapping, \%target_files_key_to_last_update_date_mapping);
1955 &log_message("Total number of target chunks requiring translation: " . scalar(@target_files_keys_requiring_translation));
1956 &log_message("Total number of target chunks requiring updating: " . scalar(@target_files_keys_requiring_updating));
[29415]1957
1958 my $download_target_filepath = "";
1959
1960
1961 # ****** DOWNLOADING LANGUAGE FILES WAS NOT YET IMPLEMENTED FOR GS3. RUDIMENTARY VERSION ****** #
1962
1963 # if there is no copy of the language files for download, there's also no link to the spreadsheet
1964 # for translating offline. So GS3's download option, we will zip up all the relevant greenstone 3
1965 # interface *.properties files,and link to that zip as the file for offline translation.
1966 # Selecting only properties files for English and the language they're working on (if the last exists)
1967
1968 # tar -cvzf gs3interface.tar.gz greenstone3/AbstractBrowse.properties greenstone3/AbstractBrowse_nl.properties
1969 # will generate a tar file containing a folder called "greenstone3" with the specified *.properties files
1970
1971 my $zip = &FileUtils::filenameConcatenate("tmp", "gs3interface_".$target_language_code.".tar.gz");
1972 my $tar_cmd = "tar -cvzf $zip";
1973
1974
1975 # store cur dir and cd to gsdlhome to generate the correct path in the zip file
1976 my $curdir = `pwd`;
1977 chdir $gsdl_root_directory;
1978
[30581]1979 $tar_cmd .= " " . &get_gs3_zip_file_listing($target_language_code, "greenstone3", \@gs3_interface_files);
1980 $tar_cmd .= " " . &get_gs3_zip_file_listing($target_language_code, "gs3-collection-configs", \@gs3_col_cfg_files);
[29415]1981
1982 # tar command will overwrite the previous version, but want to check we've created it
1983 if(&FileUtils::fileExists($zip)) {
1984 &FileUtils::removeFiles($zip);
1985 }
1986
1987 #my $tar_result = system($tar_cmd); # works but then interface breaks
1988 `$tar_cmd`;
1989 my $tar_result = $?;
1990
1991 if(&FileUtils::fileExists($zip)) { ## if($tar_result == 0) {, # breaks the interface
1992 $download_target_filepath = $zip;
1993 } else {
1994 &log_message("Unable to generate zip containing gs3interface files " . $download_target_filepath . "$!");
1995 }
1996
1997 # change back to original working directory (cgi-bin/linux probably)
1998 chdir $curdir;
1999
2000 # ************** END RUDIMENTARY VERSION OF DOWNLOADING LANGUAGE FILES FOR GS3 ************* #
2001
2002
2003 my $xml_response = &create_xml_response_for_chunks_requiring_work($translation_file_key, $download_target_filepath, scalar(keys(%source_files_key_to_text_mapping)),
[24627]2004 \@target_files_keys_requiring_translation, \@target_files_keys_requiring_updating,
2005 $num_chunks_to_return, \%source_files_key_to_text_mapping, \%target_files_key_to_text_mapping,
2006 \%source_files_key_to_last_update_date_mapping, \%target_files_key_to_last_update_date_mapping);
2007
[18460]2008 return $xml_response;
2009}
2010
[30581]2011# helper function
2012# gets the listing of gs3 files for a gs3 interface module (gs3interface, gs3colcfg)
2013# formatted correctly to go into a zip file
2014sub get_gs3_zip_file_listing
2015{
2016 my $target_language_code = shift(@_);
2017 my $sourcedir = shift(@_);
2018 my $files_array = shift(@_); # reference to an array of the interfaces files for the gs3 module
2019
2020 my $filelisting = "";
2021 foreach my $interface_file (@$files_array) {
2022
2023 my $source_filepath = &FileUtils::filenameConcatenate($sourcedir, $interface_file.".properties");
2024 my $target_filepath = &FileUtils::filenameConcatenate($sourcedir, $interface_file."_".$target_language_code.".properties");
2025
2026 $filelisting = "$filelisting $source_filepath";
2027 if(&FileUtils::fileExists($target_filepath)) {
2028 $filelisting = "$filelisting $target_filepath";
2029 }
2030 }
2031
2032 return $filelisting;
2033}
2034
[25249]2035sub get_uptodate_chunks_for_gs3
2036{
2037 # The code of the target language (ensure it is lowercase)
2038 my $target_language_code = lc(shift(@_));
2039 # The key of the file to translate (ensure it is lowercase)
2040 my $translation_file_key = lc(shift(@_));
2041 # The number of chunks to return (defaults to one if not specified)
2042 my $num_chunks_to_return = shift(@_) || "1";
2043
2044 # Check that the necessary arguments were supplied
2045 if (!$target_language_code || !$translation_file_key) {
2046 &throw_fatal_error("Missing command argument.");
2047 }
2048
2049 my %source_files_key_to_text_mapping = ();
2050 my %target_files_key_to_text_mapping = ();
2051 my %source_files_key_to_last_update_date_mapping = ();
2052 my %target_files_key_to_last_update_date_mapping = ();
2053
[30581]2054 &build_gs3_configuration($translation_file_key, $target_language_code, \%source_files_key_to_text_mapping, \%target_files_key_to_text_mapping,
[25249]2055 \%source_files_key_to_last_update_date_mapping, \%target_files_key_to_last_update_date_mapping);
2056
[18460]2057
[25249]2058 # Chunks needing updating are those in the target file that have been more recently edited in the source file
2059 # All others are uptodate (which implies that they have certainly been translated at some point and would not be empty)
2060 my @uptodate_target_file_keys = ();
2061 foreach my $chunk_key (keys(%source_files_key_to_last_update_date_mapping)) {
2062 my $source_chunk_last_update_date = $source_files_key_to_last_update_date_mapping{$chunk_key};
2063 my $target_chunk_last_update_date = $target_files_key_to_last_update_date_mapping{$chunk_key};
2064
2065 # print "key: $chunk_key\nsource date : $source_chunk_last_update_date\ntarget date : $target_chunk_last_update_date\nafter? ". &is_date_after($source_chunk_last_update_date, $target_chunk_last_update_date) . "\n\n";
2066
2067 if (defined($target_chunk_last_update_date) && !&is_date_after($source_chunk_last_update_date, $target_chunk_last_update_date)) {
2068 # &log_message("Chunk with key $chunk_key needs updating.");
2069 push(@uptodate_target_file_keys, $chunk_key);
2070 }
2071 }
[18460]2072
[25249]2073 my $xml_response = &create_xml_response_for_uptodate_chunks($translation_file_key, "", \@uptodate_target_file_keys, \%source_files_key_to_text_mapping, \%target_files_key_to_text_mapping, \%source_files_key_to_last_update_date_mapping, \%target_files_key_to_last_update_date_mapping);
2074
2075 return $xml_response;
2076}
2077
2078
[18460]2079sub build_gs3_configuration
2080{
[30581]2081 my ($translation_file_key, $target_language_code, $source_files_key_to_text_mapping, $target_files_key_to_text_mapping,
[30548]2082 $source_files_key_to_gti_comment_or_last_updated_mapping, $target_files_key_to_gti_comment_or_last_updated_mapping, $get_gti_comments_not_last_updated) = @_;
[18460]2083
[28755]2084 my $source_file_directory = "greenstone3"; # my $source_file_directory = &util::filename_cat("WEB-INF","classes");
[30581]2085 my $files_array = \@gs3_interface_files;
2086
2087 if($translation_file_key eq "gs3colcfg") {
2088 $source_file_directory = "gs3-collection-configs";
2089 $files_array = \@gs3_col_cfg_files;
2090 }
[18460]2091 my $translation_file_type = "resource_bundle";
[24627]2092
[30581]2093 foreach my $interface_file_key (@$files_array) {
[24627]2094
2095 &log_message("Greenstone 3 interface file: " . $interface_file_key);
2096
2097 # Parse the source language and target language files
2098 my $source_file = &util::filename_cat($source_file_directory, $interface_file_key.".properties");
2099 my @source_file_lines = &read_file_lines(&util::filename_cat($gsdl_root_directory, $source_file));
2100 my %source_file_key_to_line_mapping = &build_key_to_line_mapping(\@source_file_lines, $translation_file_type);
2101 my %source_file_key_to_text_mapping = &build_key_to_text_mapping(\@source_file_lines, \%source_file_key_to_line_mapping, $translation_file_type);
[30548]2102 #my %source_file_key_to_gti_comment_mapping = &build_key_to_gti_comment_mapping(\@source_file_lines, \%source_file_key_to_line_mapping, $translation_file_type);
[24627]2103
[30548]2104 my %source_file_key_to_gti_comment_or_last_updated_mapping;
2105 if($get_gti_comments_not_last_updated) {
2106 %source_file_key_to_gti_comment_or_last_updated_mapping = &build_key_to_gti_comment_mapping(\@source_file_lines, \%source_file_key_to_line_mapping, $translation_file_type);
2107 } else {
2108 %source_file_key_to_gti_comment_or_last_updated_mapping = &build_key_to_last_update_date_mapping($source_file, \@source_file_lines, \%source_file_key_to_line_mapping, $translation_file_type);
2109 }
2110
[24627]2111 my $target_file = &util::filename_cat($source_file_directory, $interface_file_key."_".$target_language_code.".properties");
2112 my @target_file_lines = &read_file_lines(&util::filename_cat($gsdl_root_directory, $target_file));
2113 my %target_file_key_to_line_mapping = &build_key_to_line_mapping(\@target_file_lines, $translation_file_type);
2114 my %target_file_key_to_text_mapping = &build_key_to_text_mapping(\@target_file_lines, \%target_file_key_to_line_mapping, $translation_file_type);
[30548]2115 #my %target_file_key_to_gti_comment_mapping = &build_key_to_gti_comment_mapping(\@target_file_lines, \%target_file_key_to_line_mapping, $translation_file_type);
[24627]2116
[30548]2117 my %target_file_key_to_gti_comment_or_last_updated_mapping;
2118 if($get_gti_comments_not_last_updated) {
2119 %target_file_key_to_gti_comment_or_last_updated_mapping = &build_key_to_gti_comment_mapping(\@target_file_lines, \%target_file_key_to_line_mapping, $translation_file_type);
2120 } else {
2121 %target_file_key_to_gti_comment_or_last_updated_mapping = &build_key_to_last_update_date_mapping($target_file, \@target_file_lines, \%target_file_key_to_line_mapping, $translation_file_type);
2122 }
[24627]2123
[30548]2124
[24627]2125 # Filter out any automatically translated chunks
2126 foreach my $chunk_key (keys(%source_file_key_to_line_mapping)) {
2127 if (&is_chunk_automatically_translated($chunk_key, $translation_file_type)) {
2128 delete $source_file_key_to_line_mapping{$chunk_key};
2129 delete $target_file_key_to_line_mapping{$chunk_key};
2130 }
2131 }
2132
2133 &log_message("Number of source chunks: " . scalar(keys(%source_file_key_to_text_mapping)));
2134 &log_message("Number of target chunks: " . scalar(keys(%target_file_key_to_text_mapping)));
2135
2136 foreach my $chunk_key (keys(%source_file_key_to_text_mapping)) {
2137 my $global_chunk_key = "$interface_file_key.$chunk_key";
2138 $source_files_key_to_text_mapping->{$global_chunk_key} = $source_file_key_to_text_mapping{$chunk_key};
[30548]2139 $source_files_key_to_gti_comment_or_last_updated_mapping->{$global_chunk_key} = $source_file_key_to_gti_comment_or_last_updated_mapping{$chunk_key};
[24627]2140
2141 if (defined $target_file_key_to_text_mapping{$chunk_key}) {
2142 $target_files_key_to_text_mapping->{$global_chunk_key} = $target_file_key_to_text_mapping{$chunk_key};
[30548]2143 $target_files_key_to_gti_comment_or_last_updated_mapping->{$global_chunk_key} = $target_file_key_to_gti_comment_or_last_updated_mapping{$chunk_key};
[24627]2144 }
[29456]2145 }
[18460]2146 }
2147}
2148
2149
2150sub write_translated_gs3interface
2151{
[30581]2152 my $translation_file_key = shift(@_);
[18460]2153 my $source_file_key_to_text_mapping = shift(@_);
2154 my $target_file_key_to_text_mapping = shift(@_);
2155 my $target_file_key_to_gti_comment_mapping = shift(@_);
2156 my $target_language_code = shift(@_);
2157
2158 my @sorted_chunk_keys = sort (keys(%$source_file_key_to_text_mapping));
[24627]2159
[18460]2160 my %translated_interface_file_keys = ();
2161 foreach my $chunk_key (keys(%$target_file_key_to_text_mapping)) {
[24627]2162 $chunk_key =~ /^([^\.]+)?\.(.*)$/;
2163 if (!defined $translated_interface_file_keys{$1}) {
2164 &log_message("Updated interface file: " . $1);
2165 $translated_interface_file_keys{$1}="";
2166 }
[18460]2167 }
2168 &log_message("Updated interface files: " . scalar(keys(%translated_interface_file_keys)));
2169
2170 my $source_file_directory = "greenstone3";
[30581]2171 $source_file_directory = "gs3-collection-configs" if $translation_file_key eq "gs3colcfg";
2172
[18460]2173 foreach my $interface_file_key (keys(%translated_interface_file_keys)) {
[24627]2174
2175 # Build a mapping from chunk key to source file line, and from source file line to chunk key
2176 my $source_file = &util::filename_cat($source_file_directory, "$interface_file_key.properties");
2177 my @source_file_lines = &read_file_lines(&util::filename_cat($gsdl_root_directory, $source_file));
2178 my %source_file_key_to_line_mapping = &build_key_to_line_mapping_for_resource_bundle(@source_file_lines);
2179 my %source_file_line_to_key_mapping = ();
2180 foreach my $chunk_key (keys(%source_file_key_to_line_mapping)) {
2181 $source_file_line_to_key_mapping{$source_file_key_to_line_mapping{$chunk_key}} = $chunk_key;
[18460]2182 }
[24627]2183
2184 # Write the new target file
2185 my $target_file = &util::filename_cat($source_file_directory, $interface_file_key . "_" . $target_language_code . ".properties");
2186 my $target_file_path = &util::filename_cat($gsdl_root_directory, $target_file);
2187 if (!open(TARGET_FILE, ">$target_file_path")) {
2188 &throw_fatal_error("Could not write target file $target_file_path.");
2189 }
2190
2191 # Model the new target file on the source file, with the target file translations
2192 my $source_file_line_number = 0;
2193 foreach my $line_key (sort sort_by_line (keys(%source_file_line_to_key_mapping))) {
2194 # Fill in the gaps before this chunk starts
2195 my $source_file_chunk_starting_line_number = (split(/-/, $line_key))[0];
2196 my $source_file_chunk_finishing_line_number = (split(/-/, $line_key))[1];
2197 while ($source_file_line_number < $source_file_chunk_starting_line_number) {
2198 print TARGET_FILE $source_file_lines[$source_file_line_number];
2199 $source_file_line_number++;
2200 }
2201 $source_file_line_number = $source_file_chunk_finishing_line_number + 1;
2202
2203 my $chunk_key = $source_file_line_to_key_mapping{$line_key};
2204 my $global_chunk_key = "$interface_file_key.$chunk_key";
2205 my $source_file_chunk_text = $source_file_key_to_text_mapping->{$global_chunk_key};
2206 my $target_file_chunk_text = $target_file_key_to_text_mapping->{$global_chunk_key} || "";
2207
[30719]2208 # make sure any : or = sign in the chunk key is escaped again (with \) when written out
2209 # since the key-value separator in a property resource bundle file is : or =
2210 my $escaped_chunk_key = $chunk_key;
2211 $escaped_chunk_key =~ s/(:|=)/\\$1/g; #$escaped_chunk_key =~ s/([^\\])(:|=)/\\$1$2/g;
2212
[24627]2213 # If no translation exists for this chunk, show this, and move on
2214 if ($source_file_chunk_text ne "" && $target_file_chunk_text eq "") {
[30719]2215 print TARGET_FILE "# -- Missing translation: $escaped_chunk_key\n";
[24627]2216 next;
2217 }
2218
[30719]2219 print TARGET_FILE "$escaped_chunk_key:$target_file_chunk_text";
[24627]2220 if ($target_file_key_to_gti_comment_mapping->{$global_chunk_key}) {
2221 print TARGET_FILE " # " . $target_file_key_to_gti_comment_mapping->{$global_chunk_key};
2222 }
2223 print TARGET_FILE "\n";
2224 }
2225
2226 close(TARGET_FILE);
[18460]2227 }
2228}
2229
[10019]2230&main(@ARGV);
Note: See TracBrowser for help on using the repository browser.