source: gsdl/trunk/perllib/plugins/ISISPlugin.pm@ 17479

Last change on this file since 17479 was 17479, checked in by kjdon, 16 years ago

put this back to using block expression for now - on windows sets up xx.FDT to be blocked, but then if the actual file is xx.fdt, it doesn't get blocked, and is unrecognised
  • Property svn:keywords set to Author Date Id Revision
File size: 14.1 KB
Line 
1###########################################################################
2#
3# ISISPlugin.pm -- A plugin for CDS/ISIS databases
4#
5# A component of the Greenstone digital library software
6# from the New Zealand Digital Library Project at the
7# University of Waikato, New Zealand.
8#
9# Copyright 1999-2004 New Zealand Digital Library Project
10#
11# This program is free software; you can redistribute it and/or modify
12# it under the terms of the GNU General Public License as published by
13# the Free Software Foundation; either version 2 of the License, or
14# (at your option) any later version.
15#
16# This program is distributed in the hope that it will be useful,
17# but WITHOUT ANY WARRANTY; without even the implied warranty of
18# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
19# GNU General Public License for more details.
20#
21# You should have received a copy of the GNU General Public License
22# along with this program; if not, write to the Free Software
23# Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
24#
25###########################################################################
26
27package ISISPlugin;
28
29
30use multiread;
31use SplitTextFile;
32
33use strict;
34no strict 'refs'; # allow filehandles to be variables and viceversa
35
36# ISISPlugin is a sub-class of SplitTextFile.
37sub BEGIN {
38 @ISISPlugin::ISA = ('SplitTextFile');
39}
40
41
42my $arguments =
43 [ { 'name' => "process_exp",
44 'desc' => "{BasePlugin.process_exp}",
45 'type' => "regexp",
46 'reqd' => "no",
47 'deft' => &get_default_process_exp() },
48 { 'name' => "block_exp",
49 'desc' => "{BasePlugin.block_exp}",
50 'type' => "regexp",
51 'reqd' => "no",
52 'deft' => &get_default_block_exp(),
53 'hiddengli' => "yes" },
54 { 'name' => "split_exp",
55 'desc' => "{SplitTextFile.split_exp}",
56 'type' => "regexp",
57 'reqd' => "no",
58 'deft' => &get_default_split_exp(),
59 'hiddengli' => "yes" },
60
61 # The interesting options
62 { 'name' => "entry_separator",
63 'desc' => "{ISISPlugin.entry_separator}",
64 'type' => "string",
65 'reqd' => "no",
66 'deft' => "<br>" },
67 { 'name' => "subfield_separator",
68 'desc' => "{ISISPlugin.subfield_separator}",
69 'type' => "string",
70 'reqd' => "no",
71 'deft' => ", " }
72 ];
73
74my $options = { 'name' => "ISISPlugin",
75 'desc' => "{ISISPlugin.desc}",
76 'abstract' => "no",
77 'inherits' => "yes",
78 'explodes' => "yes",
79 'args' => $arguments };
80
81
82# This plugin processes files with the suffix ".mst"
83sub get_default_process_exp {
84 return q^(?i)(\.mst)$^;
85}
86
87
88# This plugin blocks files with the suffix ".fdt" and ".xrf"
89sub get_default_block_exp {
90 return q^(?i)(\.fdt|\.xrf)$^;
91 #return "";
92}
93
94
95# This plugin splits the input text at the "----------" lines
96sub get_default_split_exp {
97 return q^\r?\n----------\r?\n^;
98}
99
100
101sub new
102{
103 my ($class) = shift (@_);
104 my ($pluginlist,$inputargs,$hashArgOptLists) = @_;
105 push(@$pluginlist, $class);
106
107 push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});
108 push(@{$hashArgOptLists->{"OptList"}},$options);
109
110 my $self = new SplitTextFile($pluginlist, $inputargs, $hashArgOptLists);
111
112 if ($self->{'info_only'}) {
113 # don't worry about any options etc
114 return bless $self, $class;
115 }
116
117 # isis plug doesn't care about encoding - it assumes ascii unless the user
118 # has specified an encoding
119 if ($self->{'input_encoding'} eq "auto") {
120 $self->{'input_encoding'} = "ascii";
121 }
122 return bless $self, $class;
123}
124
125# we block the corresponding fdt and xrf
126# a pain on windows. blocks xxx.FDT, but if actual file is xx.fdt then
127# complains that no plugin can process it. Have put it back to using
128# block exp for now
129sub store_block_files_tmp {
130
131 my $self =shift (@_);
132 my ($filename_full_path, $block_hash) = @_;
133 print STDERR "in store block files\n";
134 $self->check_auxiliary_files($filename_full_path);
135 if (-e $self->{'fdt_file_path'}) {
136 print STDERR "$self->{'fdt_file_path'}\n";
137 my $fdt_file = $self->{'fdt_file_path'};
138 $block_hash->{'file_blocks'}->{$fdt_file} = 1;
139 }
140 if (-e $self->{'xrf_file_path'}) {
141 print STDERR "$self->{'xrf_file_path'}\n";
142 my $xrf_file = $self->{'xrf_file_path'};
143 $block_hash->{'file_blocks'}->{$xrf_file} = 1;
144 }
145
146
147}
148
149sub check_auxiliary_files {
150 my $self = shift (@_);
151 my ($filename) = @_;
152
153 my ($database_file_path_root) = ($filename =~ /(.*)\.mst$/i);
154 # Check the associated .fdt and .xrf files exist
155 $self->{'fdt_file_path'} = $database_file_path_root . ".FDT";
156 if (!-e $self->{'fdt_file_path'}) {
157 $self->{'fdt_file_path'} = $database_file_path_root . ".fdt";
158 }
159 $self->{'xrf_file_path'} = $database_file_path_root . ".XRF";
160 if (!-e $self->{'xrf_file_path'}) {
161 $self->{'xrf_file_path'} = $database_file_path_root . ".xrf";
162 }
163}
164
165
166sub read_file
167{
168 my $self = shift (@_);
169 my ($filename, $encoding, $language, $textref) = @_;
170 my $outhandle = $self->{'outhandle'};
171
172 my ($database_file_path_root) = ($filename =~ /(.*)\.mst$/i);
173 my $mst_file_path_relative = $filename;
174 $mst_file_path_relative =~ s/^.+import.(.*?)$/$1/;
175
176 # Check the associated .fdt and .xrf files exist
177 $self->check_auxiliary_files($filename);
178
179 if (!-e $self->{'fdt_file_path'}) {
180 print STDERR "<ProcessingError n='$mst_file_path_relative' r='Could not find ISIS FDT file $self->{'fdt_file_path'}'>\n" if ($self->{'gli'});
181 print $outhandle "Error: Could not find ISIS FDT file " . $self->{'fdt_file_path'} . ".\n";
182 return;
183 }
184 if (!-e $self->{'xrf_file_path'}) {
185 print STDERR "<ProcessingError n='$mst_file_path_relative' r='Could not find ISIS XRF file $self->{'xrf_file_path'}'>\n" if ($self->{'gli'});
186 print $outhandle "Error: Could not find ISIS XRF file " . $self->{'xrf_file_path'} . ".\n";
187 return;
188 }
189
190 # The text to split is exported from the database by the IsisGdl program
191 open(FILE, "IsisGdl \"$filename\" |");
192
193 my $reader = new multiread();
194 $reader->set_handle('ISISPlugin::FILE');
195 $reader->set_encoding($encoding);
196 $reader->read_file($textref);
197
198 close(FILE);
199
200 # Parse the associated ISIS database Field Definition Table file (.fdt)
201 my %fdt_mapping = &parse_field_definition_table($self->{'fdt_file_path'}, $encoding);
202 $self->{'fdt_mapping'} = \%fdt_mapping;
203
204 # Remove the line at the start, and any blank lines, so the data is split and processed properly
205 $$textref =~ s/^----------\n//;
206 $$textref =~ s/\n\n/\n/g;
207}
208
209
210sub process
211{
212 my $self = shift (@_);
213 my ($textref, $pluginfo, $base_dir, $file, $metadata, $doc_obj, $gli) = @_;
214 my $outhandle = $self->{'outhandle'};
215
216 my $section = $doc_obj->get_top_section();
217 my $fdt_mapping = $self->{'fdt_mapping'};
218 my $subfield_separator = $self->{'subfield_separator'};
219 my $entry_separator = $self->{'entry_separator'};
220 my $isis_record_html_metadata_value = "<table cellpadding=\"4\" cellspacing=\"0\">";
221
222 # Process each line of the ISIS record, one at a time
223 foreach my $line (split(/\n/, $$textref)) {
224 $line =~ s/(\s*)$//; # Remove any nasty whitespace (very important for Windows)
225 $line =~ /^tag=(.*) data=(.+)$/;
226 my $tag = $1;
227 my $tag_data = $2;
228 # print STDERR "\nTag: $tag, Data: $tag_data\n";
229
230 # Convert the tag number into a name, and remove any invalid characters
231 my $raw_metadata_name = $fdt_mapping->{$tag}{'name'} || "";
232 $raw_metadata_name =~ s/[,&\#\.\-\/]/ /g;
233 next if ($raw_metadata_name eq "");
234
235 # Metadata field names: title case, then remove spaces
236 my $metadata_name = "";
237 foreach my $word (split(/\s+/, $raw_metadata_name)) {
238 substr($word, 0, 1) =~ tr/a-z/A-Z/;
239 $metadata_name .= $word;
240 }
241
242 my $all_metadata_name = $metadata_name . "^all";
243 my $all_metadata_value = "";
244
245 # Handle repeatable fields
246 if ($fdt_mapping->{$tag}{'repeatable'}) {
247 # Multiple values are separated using the '%' character
248 foreach my $raw_metadata_value (split(/%/, $tag_data)) {
249 my $metadata_value = "";
250
251 # Handle subfields
252 while ($raw_metadata_value ne "") {
253 # If there is a subfield specifier, parse it off
254 my $sub_metadata_name = $metadata_name;
255 if ($raw_metadata_value =~ s/^\^// && $raw_metadata_value =~ s/^([a-z])//) {
256 $sub_metadata_name .= "^$1";
257 }
258
259 # Parse the value off and add it as metadata
260 $raw_metadata_value =~ s/^([^\^]*)//;
261 my $sub_metadata_value = &escape_metadata_value($1);
262
263 # print STDERR "Sub metadata name: $sub_metadata_name, value: $sub_metadata_value\n";
264 if ($sub_metadata_name ne $metadata_name) {
265 $doc_obj->add_utf8_metadata($section, $sub_metadata_name, $sub_metadata_value);
266 }
267
268 # If this tag has subfields and this is the first, use the value for the CDS/ISIS ^* field
269 if ($fdt_mapping->{$tag}{'subfields'} ne "" && $metadata_value eq "") {
270 $doc_obj->add_utf8_metadata($section, $metadata_name . "^*", $sub_metadata_value);
271 }
272
273 $metadata_value .= $subfield_separator unless ($metadata_value eq "");
274 $metadata_value .= $sub_metadata_value;
275 }
276
277 # Add the metadata value
278 # print STDERR "Metadata name: $metadata_name, value: $metadata_value\n";
279 $doc_obj->add_utf8_metadata($section, $metadata_name, $metadata_value);
280
281 $all_metadata_value .= $entry_separator unless ($all_metadata_value eq "");
282 $all_metadata_value .= $metadata_value;
283 }
284 }
285
286 # Handle non-repeatable fields
287 else {
288 my $raw_metadata_value = $tag_data;
289 my $metadata_value = "";
290
291 # Handle subfields
292 while ($raw_metadata_value ne "") {
293 # If there is a subfield specifier, parse it off
294 my $sub_metadata_name = $metadata_name;
295 if ($raw_metadata_value =~ s/^\^// && $raw_metadata_value =~ s/^([a-z])//) {
296 $sub_metadata_name .= "^$1";
297 }
298
299 # Parse the value off and add it as metadata
300 $raw_metadata_value =~ s/^([^\^]*)//;
301 my $sub_metadata_value = $1;
302
303 # Deal with the case when multiple values are specified using <...>
304 if ($sub_metadata_value =~ /\<(.+)\>/) {
305 my $sub_sub_metadata_name = $sub_metadata_name . "^sub";
306 my $tmp_sub_metadata_value = $sub_metadata_value;
307 while ($tmp_sub_metadata_value =~ s/\<(.+?)\>//) {
308 my $sub_sub_metadata_value = $1;
309 $doc_obj->add_utf8_metadata($section, $sub_sub_metadata_name, $sub_sub_metadata_value);
310 }
311 }
312 # Deal with the legacy case when multiple values are specified using /.../
313 elsif ($sub_metadata_value =~ /\/(.+)\//) {
314 my $sub_sub_metadata_name = $sub_metadata_name . "^sub";
315 my $tmp_sub_metadata_value = $sub_metadata_value;
316 while ($tmp_sub_metadata_value =~ s/\/(.+?)\///) {
317 my $sub_sub_metadata_value = $1;
318 $doc_obj->add_utf8_metadata($section, $sub_sub_metadata_name, $sub_sub_metadata_value);
319 }
320 }
321
322 # Escape the metadata value so it appears correctly in the final collection
323 $sub_metadata_value = &escape_metadata_value($sub_metadata_value);
324
325 # print STDERR "Sub metadata name: $sub_metadata_name, value: $sub_metadata_value\n";
326 if ($sub_metadata_name ne $metadata_name) {
327 $doc_obj->add_utf8_metadata($section, $sub_metadata_name, $sub_metadata_value);
328 }
329
330 # If this tag has subfields and this is the first, use the value for the CDS/ISIS ^* field
331 if ($fdt_mapping->{$tag}{'subfields'} ne "" && $metadata_value eq "") {
332 $doc_obj->add_utf8_metadata($section, $metadata_name . "^*", $sub_metadata_value);
333 }
334
335 $metadata_value .= $subfield_separator unless ($metadata_value eq "");
336 $metadata_value .= $sub_metadata_value;
337 }
338
339 # Add the metadata value
340 # print STDERR "Metadata name: $metadata_name, value: $metadata_value\n";
341 $doc_obj->add_utf8_metadata($section, $metadata_name, $metadata_value);
342
343 $all_metadata_value .= $entry_separator unless ($all_metadata_value eq "");
344 $all_metadata_value .= $metadata_value;
345 }
346
347 # Add the "^all" metadata value
348 # print STDERR "All metadata name: $all_metadata_name, value: $all_metadata_value\n";
349 $doc_obj->add_utf8_metadata($section, $all_metadata_name, $all_metadata_value);
350
351 $isis_record_html_metadata_value .= "<tr><td valign=top><nobr><b>" . $fdt_mapping->{$tag}{'name'} . "</b></nobr></td><td valign=top>" . $all_metadata_value . "</td></tr>";
352 }
353
354 # Add a reasonably formatted HTML table view of the record as the document text
355 $isis_record_html_metadata_value .= "</table>";
356 $doc_obj->add_utf8_text($section, $isis_record_html_metadata_value);
357
358 # Add the full raw record as metadata
359 my $isis_raw_record_metadata_value = &escape_metadata_value($$textref);
360 $doc_obj->add_utf8_metadata($section, "ISISRawRecord", $isis_raw_record_metadata_value);
361
362 # Add FileFormat metadata
363 $doc_obj->add_utf8_metadata($section, "FileFormat", "CDS/ISIS");
364
365 # Record was processed successfully
366 return 1;
367}
368
369
370sub parse_field_definition_table
371{
372 my $fdtfilename = shift(@_);
373 my $encoding = shift(@_);
374
375 my %fdtmapping = ();
376
377 open(FDT_FILE, "<$fdtfilename") || die "Error: Could not open file $fdtfilename.\n";
378
379 my $fdtfiletext = "";
380 my $reader = new multiread();
381 $reader->set_handle('ISISPlugin::FDT_FILE');
382 $reader->set_encoding($encoding);
383 $reader->read_file($fdtfiletext);
384
385 my $amongstdefinitions = 0;
386 foreach my $fdtfileline (split(/\n/, $$fdtfiletext)) {
387 $fdtfileline =~ s/(\s*)$//; # Remove any nasty spaces at the end of the lines
388
389 if ($amongstdefinitions) {
390 my $fieldname = &unicode::substr($fdtfileline, 0, 30);
391 my $fieldsubfields = &unicode::substr($fdtfileline, 30, 20);
392 my $fieldspecs = &unicode::substr($fdtfileline, 50, 50);
393
394 # Remove extra spaces
395 $fieldname =~ s/(\s*)$//;
396 $fieldsubfields =~ s/(\s*)$//;
397 $fieldspecs =~ s/(\s*)$//;
398
399 # Map from tag number to metadata field title, subfields, and repeatability
400 my $fieldtag = (split(/ /, $fieldspecs))[0];
401 my $fieldrepeatable = (split(/ /, $fieldspecs))[3];
402 $fdtmapping{$fieldtag} = { 'name' => $fieldname,
403 'subfields' => $fieldsubfields,
404 'repeatable' => $fieldrepeatable };
405 }
406 elsif ($fdtfileline eq "***") {
407 $amongstdefinitions = 1;
408 }
409 }
410
411 close(FDT_FILE);
412
413 return %fdtmapping;
414}
415
416
417sub escape_metadata_value
418{
419 my $value = shift(@_);
420 $value =~ s/\</&lt;/g;
421 $value =~ s/\>/&gt;/g;
422 $value =~ s/\\/\\\\/g;
423 return $value;
424}
425
426
427sub clean_up_after_exploding
428{
429 my $self = shift(@_);
430
431 # Delete the FDT and XRF files too
432 &util::rm($self->{'fdt_file_path'});
433 &util::rm($self->{'xrf_file_path'});
434}
435
436
4371;
Note: See TracBrowser for help on using the repository browser.