1 | ###########################################################################
|
---|
2 | #
|
---|
3 | # WordPlugin.pm -- plugin for importing Microsoft Word documents
|
---|
4 | # A component of the Greenstone digital library software
|
---|
5 | # from the New Zealand Digital Library Project at the
|
---|
6 | # University of Waikato, New Zealand.
|
---|
7 | #
|
---|
8 | # Copyright (C) 1999 New Zealand Digital Library Project
|
---|
9 | #
|
---|
10 | # This program is free software; you can redistribute it and/or modify
|
---|
11 | # it under the terms of the GNU General Public License as published by
|
---|
12 | # the Free Software Foundation; either version 2 of the License, or
|
---|
13 | # (at your option) any later version.
|
---|
14 | #
|
---|
15 | # This program is distributed in the hope that it will be useful,
|
---|
16 | # but WITHOUT ANY WARRANTY; without even the implied warranty of
|
---|
17 | # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
---|
18 | # GNU General Public License for more details.
|
---|
19 | #
|
---|
20 | # You should have received a copy of the GNU General Public License
|
---|
21 | # along with this program; if not, write to the Free Software
|
---|
22 | # Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
|
---|
23 | #
|
---|
24 | ###########################################################################
|
---|
25 | package WordPlugin;
|
---|
26 |
|
---|
27 | use ConvertBinaryFile;
|
---|
28 |
|
---|
29 |
|
---|
30 | use strict;
|
---|
31 | no strict 'refs'; # allow filehandles to be variables and viceversa
|
---|
32 |
|
---|
33 |
|
---|
34 | # @ISA dynamically configured to be either OpenOfficeConverter or ConvertBinaryFile
|
---|
35 |
|
---|
36 | my $arguments =
|
---|
37 | [ { 'name' => "process_exp",
|
---|
38 | 'desc' => "{BasePlugin.process_exp}",
|
---|
39 | 'type' => "regexp",
|
---|
40 | 'deft' => &get_default_process_exp(),
|
---|
41 | 'reqd' => "no" },
|
---|
42 | { 'name' => "description_tags",
|
---|
43 | 'desc' => "{HTMLPlugin.description_tags}",
|
---|
44 | 'type' => "flag" }
|
---|
45 | ];
|
---|
46 |
|
---|
47 |
|
---|
48 | my $opt_windows_args = [ { 'name' => "windows_scripting",
|
---|
49 | 'desc' => "{WordPlugin.windows_scripting}",
|
---|
50 | 'type' => "flag",
|
---|
51 | 'reqd' => "no" } ];
|
---|
52 |
|
---|
53 | my $opt_office_args = [ { 'name' => "metadata_fields",
|
---|
54 | 'desc' => "{WordPlugin.metadata_fields}",
|
---|
55 | 'type' => "string",
|
---|
56 | 'deft' => "Title" },
|
---|
57 | { 'name' => "level1_header",
|
---|
58 | 'desc' => "{StructuredHTMLPlugin.level1_header}",
|
---|
59 | 'type' => "regexp",
|
---|
60 | 'reqd' => "no",
|
---|
61 | 'deft' => "" },
|
---|
62 | { 'name' => "level2_header",
|
---|
63 | 'desc' => "{StructuredHTMLPlugin.level2_header}",
|
---|
64 | 'type' => "regexp",
|
---|
65 | 'reqd' => "no",
|
---|
66 | 'deft' => "" },
|
---|
67 | { 'name' => "level3_header",
|
---|
68 | 'desc' => "{StructuredHTMLPlugin.level3_header}",
|
---|
69 | 'type' => "regexp",
|
---|
70 | 'reqd' => "no",
|
---|
71 | 'deft' => "" },
|
---|
72 | { 'name' => "title_header",
|
---|
73 | 'desc' => "{StructuredHTMLPlugin.title_header}",
|
---|
74 | 'type' => "regexp",
|
---|
75 | 'reqd' => "no",
|
---|
76 | 'deft' => "" },
|
---|
77 | { 'name' => "delete_toc",
|
---|
78 | 'desc' => "{StructuredHTMLPlugin.delete_toc}",
|
---|
79 | 'type' => "flag",
|
---|
80 | 'reqd' => "no" },
|
---|
81 | { 'name' => "toc_header",
|
---|
82 | 'desc' => "{StructuredHTMLPlugin.toc_header}",
|
---|
83 | 'type' => "regexp",
|
---|
84 | 'reqd' => "no",
|
---|
85 | 'deft' => "" } ];
|
---|
86 |
|
---|
87 |
|
---|
88 | my $options = { 'name' => "WordPlugin",
|
---|
89 | 'desc' => "{WordPlugin.desc}",
|
---|
90 | 'abstract' => "no",
|
---|
91 | 'inherits' => "yes",
|
---|
92 | 'srcreplaceable' => "yes", # Source docs in Word can be replaced with GS-generated html
|
---|
93 | 'args' => $arguments };
|
---|
94 |
|
---|
95 | sub new {
|
---|
96 | my ($class) = shift (@_);
|
---|
97 | my ($pluginlist,$inputargs,$hashArgOptLists) = @_;
|
---|
98 | push(@$pluginlist, $class);
|
---|
99 |
|
---|
100 | my $openoffice_ext_installed;
|
---|
101 | eval("require OpenOfficeConverter");
|
---|
102 | if ($@) {
|
---|
103 | # Useful debugging statement if there is a syntax error in OpenOfficeConverter
|
---|
104 | # print STDERR "$@\n";
|
---|
105 |
|
---|
106 | push(@WordPlugin::ISA,"ConvertBinaryFile");
|
---|
107 | $openoffice_ext_installed = 0;
|
---|
108 | }
|
---|
109 | else {
|
---|
110 | # Successfully found
|
---|
111 | print STDERR "WordPlugin: OpenOffice Extension to Greenstone detected\n";
|
---|
112 | push(@WordPlugin::ISA,"OpenOfficeConverter");
|
---|
113 | $openoffice_ext_installed = 1;
|
---|
114 | }
|
---|
115 |
|
---|
116 | if ($ENV{'GSDLOS'} =~ m/^windows$/i) {
|
---|
117 | push(@$arguments,@$opt_windows_args);
|
---|
118 | push(@$arguments,@$opt_office_args);
|
---|
119 | }
|
---|
120 | elsif ($openoffice_ext_installed) {
|
---|
121 | push(@$arguments,@$opt_office_args);
|
---|
122 | }
|
---|
123 |
|
---|
124 | push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});
|
---|
125 | push(@{$hashArgOptLists->{"OptList"}},$options);
|
---|
126 |
|
---|
127 | my $self = {};
|
---|
128 | my $outhandle;
|
---|
129 |
|
---|
130 | if ($openoffice_ext_installed) {
|
---|
131 | $self = new OpenOfficeConverter($pluginlist, $inputargs, $hashArgOptLists);
|
---|
132 | $outhandle = $self->{'outhandle'};
|
---|
133 |
|
---|
134 | if ($self->{'openoffice_conversion_available'}) {
|
---|
135 | print $outhandle "WordPlugin: OpenOffice scripting functionality available\n";
|
---|
136 |
|
---|
137 | # Override default process expression
|
---|
138 | $self->{'process_exp'} = q^(?i)\.(doc|dot|docx|odt)$^;
|
---|
139 |
|
---|
140 | if ($self->{'openoffice_scripting'}) {
|
---|
141 | print $outhandle "WordPlugin: Activating OpenOffice scripting functionality\n";
|
---|
142 | }
|
---|
143 | }
|
---|
144 | else {
|
---|
145 | print $outhandle "WordPlugin: Unable to run 'soffice'\n";
|
---|
146 | print $outhandle "WordPlugin: Defaulting to ConvertBinaryFile inheritence\n";
|
---|
147 | }
|
---|
148 | }
|
---|
149 | else {
|
---|
150 | $self = new ConvertBinaryFile($pluginlist, $inputargs, $hashArgOptLists);
|
---|
151 | $outhandle = $self->{'outhandle'};
|
---|
152 | }
|
---|
153 |
|
---|
154 | if ($self->{'info_only'}) {
|
---|
155 | # don't worry about any options etc
|
---|
156 | return bless $self, $class;
|
---|
157 | }
|
---|
158 |
|
---|
159 | $self->{'filename_extension'} = "doc";
|
---|
160 | $self->{'file_type'} = "Word";
|
---|
161 |
|
---|
162 | if ($self->{'windows_scripting'}) {
|
---|
163 | $self->{'convert_options'} = "-windows_scripting";
|
---|
164 | $self->{'office_scripting'} = 1;
|
---|
165 | }
|
---|
166 | if ($self->{'openoffice_scripting'}) {
|
---|
167 | if ($self->{'windows_scripting'}) {
|
---|
168 | print $outhandle "Warning: Cannot have -windows_scripting and -openoffice_scripting\n";
|
---|
169 | print $outhandle " on at the same time. Defaulting to -windows_scripting\n";
|
---|
170 | }
|
---|
171 | else {
|
---|
172 | $self->{'convert_options'} = "-openoffice_scripting";
|
---|
173 | $self->{'office_scripting'} = 1;
|
---|
174 | }
|
---|
175 | }
|
---|
176 |
|
---|
177 | # we always save as utf-8
|
---|
178 | # if ($self->{'input_encoding'} eq "auto") {
|
---|
179 | # $self->{'input_encoding'} = "utf8";
|
---|
180 | # }
|
---|
181 |
|
---|
182 | my $secondary_plugin_options = $self->{'secondary_plugin_options'};
|
---|
183 | if (defined $self->{'office_scripting'}) {
|
---|
184 | if (!defined $secondary_plugin_options->{'StructuredHTMLPlugin'}){
|
---|
185 | $secondary_plugin_options->{'StructuredHTMLPlugin'} = [];
|
---|
186 | my $structhtml_options = $secondary_plugin_options->{'StructuredHTMLPlugin'};
|
---|
187 |
|
---|
188 | # Instruct HTMLPlugin (when eventually accessed through read_into_doc_obj)
|
---|
189 | # to extract these metadata fields from the HEAD META fields
|
---|
190 | push (@$structhtml_options, "-metadata_fields","Title,GENERATOR,date,author<Creator>");
|
---|
191 | push (@$structhtml_options, "-title_sub", '^(Page\s+\d+)?(\s*1\s+)?');
|
---|
192 | push (@$structhtml_options, "-description_tags") if $self->{'windows_scripting'}; # What about openoffice???
|
---|
193 | push (@$structhtml_options, "-extract_language") if $self->{'extract_language'};
|
---|
194 | push (@$structhtml_options, "-delete_toc") if $self->{'delete_toc'};
|
---|
195 | push (@$structhtml_options, "-toc_header", $self->{'toc_header'}) if $self->{'toc_header'};
|
---|
196 | push (@$structhtml_options, "-title_header", $self->{'title_header'}) if $self->{'title_header'};
|
---|
197 | push (@$structhtml_options, "-level1_header", $self->{'level1_header'}) if $self->{'level1_header'};
|
---|
198 | push (@$structhtml_options, "-level2_header", $self->{'level2_header'})if $self->{'level2_header'};
|
---|
199 | push (@$structhtml_options, "-level3_header", $self->{'level3_header'}) if $self->{'level3_header'};
|
---|
200 | push (@$structhtml_options, "-metadata_fields", $self->{'metadata_fields'}) if $self->{'metadata_fields'};
|
---|
201 | push (@$structhtml_options, "-metadata_field_separator", $self->{'metadata_field_separator'}) if $self->{'metadata_field_separator'};
|
---|
202 | }
|
---|
203 | }
|
---|
204 | if (!defined $secondary_plugin_options->{'HTMLPlugin'}) {
|
---|
205 | $secondary_plugin_options->{'HTMLPlugin'} = [];
|
---|
206 | }
|
---|
207 | if (!defined $secondary_plugin_options->{'TextPlugin'}) {
|
---|
208 | $secondary_plugin_options->{'TextPlugin'} = [];
|
---|
209 | }
|
---|
210 |
|
---|
211 | my $html_options = $secondary_plugin_options->{'HTMLPlugin'};
|
---|
212 | my $text_options = $secondary_plugin_options->{'TextPlugin'};
|
---|
213 | my $structhtml_options = $secondary_plugin_options->{'StructuredHTMLPlugin'};
|
---|
214 | # tell the secondary plugins that they are processing tmp files
|
---|
215 | push(@$html_options, "-processing_tmp_files");
|
---|
216 | push(@$structhtml_options, "-processing_tmp_files");
|
---|
217 |
|
---|
218 | # wvWare will always produce html files encoded as utf-8, so make sure the secondary HTMLPlugin knows this
|
---|
219 | push(@$html_options,"-input_encoding", "utf8");
|
---|
220 | push(@$html_options,"-extract_language") if $self->{'extract_language'};
|
---|
221 | push(@$html_options, "-description_tags") if $self->{'description_tags'};
|
---|
222 |
|
---|
223 | # Instruct HTMLPlugin (when eventually accessed through read_into_doc_obj)
|
---|
224 | # to extract these metadata fields from the HEAD META fields
|
---|
225 | push(@$html_options,"-metadata_fields","Title,GENERATOR,date,author<Creator>");
|
---|
226 | push(@$html_options , "-title_sub", '^(Page\s+\d+)?(\s*1\s+)?');
|
---|
227 |
|
---|
228 | my $associate_tail_re = $self->{'associate_tail_re'};
|
---|
229 | if ((defined $associate_tail_re) && ($associate_tail_re ne "")) {
|
---|
230 | push(@$html_options, "-associate_tail_re", $associate_tail_re);
|
---|
231 | push(@$text_options, "-associate_tail_re", $associate_tail_re);
|
---|
232 | push(@$structhtml_options, "-associate_tail_re", $associate_tail_re) if defined $structhtml_options;
|
---|
233 | }
|
---|
234 |
|
---|
235 | push(@$html_options, "-file_rename_method", "none");
|
---|
236 | push(@$text_options, "-file_rename_method", "none");
|
---|
237 | push(@$structhtml_options, "-file_rename_method", "none") if defined $structhtml_options;
|
---|
238 |
|
---|
239 | $self = bless $self, $class;
|
---|
240 | $self->load_secondary_plugins($class,$secondary_plugin_options,$hashArgOptLists);
|
---|
241 |
|
---|
242 | return bless $self;
|
---|
243 | }
|
---|
244 |
|
---|
245 | sub init {
|
---|
246 | my $self = shift (@_);
|
---|
247 | my ($verbosity, $outhandle, $failhandle) = @_;
|
---|
248 |
|
---|
249 | $self->SUPER::init($verbosity,$outhandle,$failhandle);
|
---|
250 | }
|
---|
251 |
|
---|
252 | sub deinit {
|
---|
253 | # called only once, after all plugin passes have been done
|
---|
254 | my ($self) = @_;
|
---|
255 |
|
---|
256 | $self->SUPER::deinit();
|
---|
257 | }
|
---|
258 |
|
---|
259 | sub get_default_process_exp {
|
---|
260 | my $self = shift (@_);
|
---|
261 |
|
---|
262 | return q^(?i)\.(doc|dot)$^;
|
---|
263 | }
|
---|
264 |
|
---|
265 | sub convert_post_process_old
|
---|
266 | {
|
---|
267 | my $self = shift (@_);
|
---|
268 | my ($conv_filename) = @_;
|
---|
269 |
|
---|
270 | my $outhandle=$self->{'outhandle'};
|
---|
271 |
|
---|
272 | my ($language, $encoding) = $self->textcat_get_language_encoding ($conv_filename);
|
---|
273 |
|
---|
274 | # read in file ($text will be in utf8)
|
---|
275 | my $text = "";
|
---|
276 | $self->read_file ($conv_filename, $encoding, $language, \$text);
|
---|
277 |
|
---|
278 | # turn any high bytes that aren't valid utf-8 into utf-8.
|
---|
279 | #unicode::ensure_utf8(\$text);
|
---|
280 |
|
---|
281 | # Write it out again!
|
---|
282 | #$self->utf8_write_file (\$text, $conv_filename);
|
---|
283 | }
|
---|
284 |
|
---|
285 | # Modified to cache HTML files for efficieny reasons rather
|
---|
286 | # than delete all. HTML is modified not to use IE's VML.
|
---|
287 | # VML uses WML files, so these can be deleted.
|
---|
288 | sub cleanup_tmp_area {
|
---|
289 | my ($self) = @_;
|
---|
290 | if (defined $self->{'files_dir'}) {
|
---|
291 | my $html_files_dir = $self->{'files_dir'};
|
---|
292 |
|
---|
293 | if (opendir(DIN,$html_files_dir)) {
|
---|
294 | my @wmz_files = grep( /\.wmz$/, readdir(DIN));
|
---|
295 | foreach my $f (@wmz_files) {
|
---|
296 | my $full_f = &util::filename_cat($html_files_dir,$f);
|
---|
297 | &util::rm($full_f);
|
---|
298 | }
|
---|
299 | closedir(DIN);
|
---|
300 | }
|
---|
301 | else {
|
---|
302 | # if HTML file has no supporting images, then no _files dir made
|
---|
303 | # => do nothing
|
---|
304 | }
|
---|
305 | }
|
---|
306 | }
|
---|
307 |
|
---|
308 |
|
---|
309 | 1;
|
---|
310 |
|
---|