root/main/trunk/greenstone2/bin/script/gti.pl @ 30582

Revision 30582, 93.6 KB (checked in by ak19, 4 years ago)

Final fixes to get changes to GTI working with new gs3colcfg (gs3-collection-configs) module: 1. gti.pl uses regex to check if module is a GS3 specific module rather than a GS2 specific module. 2. gs3-collection-configs properties files no longer have the collection name prefixed to their properties, as gti already adds this in.

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