1 | ###########################################################################
|
---|
2 | #
|
---|
3 | # WebDownload.pm -- base class for all the import plugins
|
---|
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 |
|
---|
26 | package OAIDownload;
|
---|
27 |
|
---|
28 | eval {require bytes};
|
---|
29 |
|
---|
30 | # suppress the annoying "subroutine redefined" warning that various
|
---|
31 | # plugins cause under perl 5.6
|
---|
32 | $SIG{__WARN__} = sub {warn($_[0]) unless ($_[0] =~ /Subroutine\s+\S+\sredefined/)};
|
---|
33 |
|
---|
34 | use strict;
|
---|
35 |
|
---|
36 | use WgetDownload;
|
---|
37 | use XMLParser;
|
---|
38 |
|
---|
39 | use POSIX qw(tmpnam);
|
---|
40 | use util;
|
---|
41 |
|
---|
42 | sub BEGIN {
|
---|
43 | @OAIDownload::ISA = ('WgetDownload');
|
---|
44 | }
|
---|
45 |
|
---|
46 | my $arguments =
|
---|
47 | [ { 'name' => "url",
|
---|
48 | 'disp' => "{OAIDownload.url_disp}",
|
---|
49 | 'desc' => "{OAIDownload.url}",
|
---|
50 | 'type' => "string",
|
---|
51 | 'reqd' => "yes"},
|
---|
52 | { 'name' => "set",
|
---|
53 | 'disp' => "{OAIDownload.set_disp}",
|
---|
54 | 'desc' => "{OAIDownload.set}",
|
---|
55 | 'type' => "string",
|
---|
56 | 'reqd' => "no"},
|
---|
57 | { 'name' => "get_doc",
|
---|
58 | 'disp' => "{OAIDownload.get_doc_disp}",
|
---|
59 | 'desc' => "{OAIDownload.get_doc}",
|
---|
60 | 'type' => "flag",
|
---|
61 | 'reqd' => "no"},
|
---|
62 | { 'name' => "max_records",
|
---|
63 | 'disp' => "{OAIDownload.max_records_disp}",
|
---|
64 | 'desc' => "{OAIDownload.max_records}",
|
---|
65 | 'type' => "int",
|
---|
66 | 'deft' => "500",
|
---|
67 | 'range' => "1,",
|
---|
68 | 'reqd' => "no"} ];
|
---|
69 |
|
---|
70 | my $options = { 'name' => "OAIDownload",
|
---|
71 | 'desc' => "{OAIDownload.desc}",
|
---|
72 | 'abstract' => "no",
|
---|
73 | 'inherits' => "yes",
|
---|
74 | 'args' => $arguments };
|
---|
75 |
|
---|
76 | my $self;
|
---|
77 |
|
---|
78 | my $strWgetOptions="";
|
---|
79 |
|
---|
80 | sub new
|
---|
81 | {
|
---|
82 | my ($class) = shift (@_);
|
---|
83 | my ($getlist,$inputargs,$hashArgOptLists) = @_;
|
---|
84 | push(@$getlist, $class);
|
---|
85 |
|
---|
86 | if(defined $arguments){ push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});}
|
---|
87 | if(defined $options) { push(@{$hashArgOptLists->{"OptList"}},$options)};
|
---|
88 |
|
---|
89 | $self = (defined $hashArgOptLists)? new WgetDownload($getlist,$inputargs,$hashArgOptLists): new WgetDownload($getlist,$inputargs);
|
---|
90 |
|
---|
91 | if ($self->{'info_only'}) {
|
---|
92 | # don't worry about any options etc
|
---|
93 | return bless $self, $class;
|
---|
94 | }
|
---|
95 |
|
---|
96 | my $parser = new XML::Parser('Style' => 'Stream',
|
---|
97 | 'Handlers' => {'Char' => \&Char,
|
---|
98 | 'Start' => \&OAI_StartTag,
|
---|
99 | 'End' => \&OAI_EndTag
|
---|
100 | });
|
---|
101 | $self->{'parser'} = $parser;
|
---|
102 |
|
---|
103 | # make sure the tmp directory that we will use later exists
|
---|
104 | my $tmp_dir = "$ENV{GSDLHOME}/tmp";
|
---|
105 | if (! -e $tmp_dir) {
|
---|
106 | &util::mk_dir($tmp_dir);
|
---|
107 | }
|
---|
108 |
|
---|
109 | return bless $self, $class;
|
---|
110 | }
|
---|
111 |
|
---|
112 | sub download
|
---|
113 | {
|
---|
114 | my ($self) = shift (@_);
|
---|
115 | my ($hashGeneralOptions) = @_;
|
---|
116 |
|
---|
117 | print STDERR "here2";
|
---|
118 |
|
---|
119 | $strWgetOptions = $self->getWgetOptions();
|
---|
120 | my $cmdWget = $strWgetOptions;
|
---|
121 |
|
---|
122 | my $strOutputDir ="";
|
---|
123 | $strOutputDir = $hashGeneralOptions->{"cache_dir"};
|
---|
124 | my $strBasURL = $self->{'url'};
|
---|
125 | my $intMaxRecords = $self->{'max_records'};
|
---|
126 | my $blnDownloadDoc = $self->{'get_doc'};
|
---|
127 |
|
---|
128 | print STDERR "<<Defined Maximum>>\n";
|
---|
129 |
|
---|
130 | my $strIDs = $self->getOAIIDs($strBasURL);
|
---|
131 |
|
---|
132 | if($strIDs eq "")
|
---|
133 | {
|
---|
134 | print STDERR "Error: No ID being found\n";
|
---|
135 | return 0;
|
---|
136 | }
|
---|
137 | my $aryIDs = $self->parseOAIIDs($strIDs);
|
---|
138 | my $intIDs = 0;
|
---|
139 | if($self->{'max_records'} < scalar(@$aryIDs))
|
---|
140 | {
|
---|
141 | $intIDs = $self->{'max_records'};
|
---|
142 | }
|
---|
143 | else
|
---|
144 | {
|
---|
145 | $intIDs = scalar(@$aryIDs);
|
---|
146 | }
|
---|
147 | print STDERR "<<Total number of record(s):$intIDs>>\n";
|
---|
148 |
|
---|
149 | $self->getOAIRecords($aryIDs, $strOutputDir, $strBasURL, $intMaxRecords, $blnDownloadDoc);
|
---|
150 |
|
---|
151 | my $tmp_file = "$ENV{GSDLHOME}/tmp/oai.tmp";
|
---|
152 | &util::rm($tmp_file);
|
---|
153 |
|
---|
154 | return 1;
|
---|
155 | }
|
---|
156 |
|
---|
157 | sub getOAIIDs
|
---|
158 | {
|
---|
159 | my ($self,$strBasURL) = @_;
|
---|
160 | my ($cmdWget);
|
---|
161 |
|
---|
162 | my $wgetOptions = $self->getWgetOptions();
|
---|
163 |
|
---|
164 | $cmdWget = $wgetOptions;
|
---|
165 |
|
---|
166 | print STDERR "Gathering OAI identifiers.....\n";
|
---|
167 |
|
---|
168 | if($self->{'set'} ne "")
|
---|
169 | {
|
---|
170 | $cmdWget .= " -q -O - \"$strBasURL?verb=ListIdentifiers&metadataPrefix=oai_dc&set=$self->{'set'}\" ";
|
---|
171 | }
|
---|
172 | else
|
---|
173 | {
|
---|
174 | $cmdWget .= " -q -O - \"$strBasURL?verb=ListIdentifiers&metadataPrefix=oai_dc\" ";
|
---|
175 | }
|
---|
176 |
|
---|
177 |
|
---|
178 | my $strIDs = $self->useWget($cmdWget);
|
---|
179 |
|
---|
180 | if (!defined $strIDs or $strIDs eq "" ){
|
---|
181 | print STDERR "Server information is unavailable.\n";
|
---|
182 | print STDERR "<<Finished>>\n";
|
---|
183 | return;
|
---|
184 | }
|
---|
185 |
|
---|
186 | print STDERR "<<Download Information>>\n";
|
---|
187 |
|
---|
188 | $self->parse_xml($strIDs);
|
---|
189 |
|
---|
190 | return $strIDs;
|
---|
191 | }
|
---|
192 |
|
---|
193 | sub parseOAIIDs
|
---|
194 | {
|
---|
195 | my ($self,$strIDs) = @_;
|
---|
196 |
|
---|
197 | print STDERR "Parsing OAI identifiers.....\n";
|
---|
198 | $strIDs =~ s/^.*?<identifier>/<identifier>/s;
|
---|
199 | $strIDs =~ s/^(.*<\/identifier>).*$/$1/s;
|
---|
200 |
|
---|
201 | my @aryIDs = ();
|
---|
202 |
|
---|
203 | while ($strIDs =~ m/<identifier>(.*?)<\/identifier>(.*)$/s)
|
---|
204 | {
|
---|
205 | $strIDs = $2;
|
---|
206 | push(@aryIDs,$1);
|
---|
207 | }
|
---|
208 |
|
---|
209 | return \@aryIDs;
|
---|
210 | }
|
---|
211 |
|
---|
212 | sub dirFileSplit
|
---|
213 | {
|
---|
214 | my ($self,$strFile) = @_;
|
---|
215 |
|
---|
216 | my @aryDirs = split("[/\]",$strFile);
|
---|
217 |
|
---|
218 | my $strLocalFile = pop(@aryDirs);
|
---|
219 | my $strSubDirs = join("/",@aryDirs);
|
---|
220 |
|
---|
221 | return ($strSubDirs,$strLocalFile);
|
---|
222 | }
|
---|
223 |
|
---|
224 | sub getOAIDoc
|
---|
225 | {
|
---|
226 | my ($self,$strRecord, $strSubDirPath) = @_;
|
---|
227 |
|
---|
228 | print STDERR "Gathering source documents.....\n";
|
---|
229 | # look out for identifier tag in metadata section
|
---|
230 |
|
---|
231 | if ($strRecord =~ m/<metadata>(.*)<\/metadata>/s)
|
---|
232 | {
|
---|
233 | my $strMetaTag = $1;
|
---|
234 |
|
---|
235 | if ($strMetaTag =~ m/<(dc:)?identifier>(.*?)<\/(dc:)?identifier>/s)
|
---|
236 | {
|
---|
237 | my $strDocURL = $2;
|
---|
238 |
|
---|
239 | my ($unused,$strDocFile) = $self->dirFileSplit($strDocURL);
|
---|
240 |
|
---|
241 | my $strSoureDirPath ="";
|
---|
242 |
|
---|
243 | $strSoureDirPath = &util::filename_cat($strSubDirPath,"srcdocs");
|
---|
244 |
|
---|
245 | &util::mk_dir($strSoureDirPath) if (!-e "$strSoureDirPath");
|
---|
246 |
|
---|
247 | my $strFullDocFilePath = &util::filename_cat($strSoureDirPath,$strDocFile);
|
---|
248 |
|
---|
249 | my $wget_cmd = $strWgetOptions." -q -O \"$strFullDocFilePath\" \"$strDocURL\"";
|
---|
250 |
|
---|
251 | my $strResponse = $self->useWget($wget_cmd,1);
|
---|
252 |
|
---|
253 | if($strResponse ne "")
|
---|
254 | {
|
---|
255 | print STDERR "Error occured while retriving OAI souce documents: $strResponse\n";
|
---|
256 | exit(-1);
|
---|
257 | }
|
---|
258 |
|
---|
259 | $strRecord =~ s/<metadata>(.*?)<(dc:)?identifier>$strDocURL<\/(dc:)?identifier>(.*?)<\/metadata>/<metadata>$1<OrigURL>$strDocURL<\/OrigURL>\n <identifier>srcdocs\/$strDocFile<\/identifier>$4<\/metadata>/s;
|
---|
260 | }
|
---|
261 | else
|
---|
262 | {
|
---|
263 | print STDERR "\tNo souce document URL is specified in the OAI record (No (dc:)?identifier is provided)\n";
|
---|
264 | }
|
---|
265 | }
|
---|
266 | else
|
---|
267 | {
|
---|
268 | print STDERR "\tNo souce document URL is specified in the OAI record (No metadata field is provided)\n";
|
---|
269 | }
|
---|
270 |
|
---|
271 | }
|
---|
272 |
|
---|
273 | sub getOAIRecords
|
---|
274 | {
|
---|
275 | my ($self,$aryIDs, $strOutputDir, $strBasURL, $intMaxRecords, $blnDownloadDoc) = @_;
|
---|
276 |
|
---|
277 | my $intDocCounter = 0;
|
---|
278 |
|
---|
279 | foreach my $strID ( @$aryIDs)
|
---|
280 | {
|
---|
281 | print STDERR "Gathering OAI record with ID:$strID.....\n";
|
---|
282 |
|
---|
283 | my $cmdWget= $strWgetOptions." -q -O - \"$strBasURL?verb=GetRecord&metadataPrefix=oai_dc&identifier=$strID\"";
|
---|
284 |
|
---|
285 | my $strRecord = $self->useWget($cmdWget);
|
---|
286 |
|
---|
287 |
|
---|
288 | my @fileDirs = split(":",$strID);
|
---|
289 |
|
---|
290 | # setup directories
|
---|
291 |
|
---|
292 | $strOutputDir =~ s/"//g; #"
|
---|
293 |
|
---|
294 | my $host =$self->{'url'};
|
---|
295 |
|
---|
296 | $host =~ s/http:\/\///g;
|
---|
297 |
|
---|
298 | $host =~ s/:.*//g;
|
---|
299 |
|
---|
300 | my $midDir = join ("/",@fileDirs);
|
---|
301 | my $strFileURL = "$strOutputDir/$host/".$midDir.".oai";
|
---|
302 |
|
---|
303 | # prepare subdirectory for record (if needed)
|
---|
304 | my ($strSubDirPath,$unused) = ("", "");
|
---|
305 |
|
---|
306 | ($strSubDirPath,$unused) = $self->dirFileSplit($strFileURL);
|
---|
307 |
|
---|
308 | &util::mk_all_dir($strSubDirPath);
|
---|
309 |
|
---|
310 | my $ds = &util::get_dirsep();
|
---|
311 |
|
---|
312 | if($blnDownloadDoc)
|
---|
313 | {
|
---|
314 | $self->getOAIDoc($strRecord,$strSubDirPath);
|
---|
315 | }
|
---|
316 |
|
---|
317 | # save record
|
---|
318 | open (OAIOUT,">$strFileURL")
|
---|
319 | || die "Unable to save oai metadata record: $!\n";
|
---|
320 | print OAIOUT $strRecord;
|
---|
321 | close(OAIOUT);
|
---|
322 |
|
---|
323 | print STDERR "Saving records to $strFileURL\n";
|
---|
324 | print STDERR "<<Done>>\n";
|
---|
325 | $intDocCounter ++;
|
---|
326 | last if ($intDocCounter >= $intMaxRecords);
|
---|
327 | }
|
---|
328 |
|
---|
329 | ($intDocCounter >= $intMaxRecords) ?
|
---|
330 | print STDERR "Reach maximum download records, use -max_records to set the maximum.\n":
|
---|
331 | print STDERR "Complete download meta record from $strBasURL\n";
|
---|
332 |
|
---|
333 | print STDERR "<<Finished>>\n";
|
---|
334 | }
|
---|
335 |
|
---|
336 | sub url_information
|
---|
337 | {
|
---|
338 | my ($self) = shift (@_);
|
---|
339 | if(!defined $self){ die "System Error: No \$self defined for url_information in OAIDownload\n";}
|
---|
340 |
|
---|
341 | my $wgetOptions = $self->getWgetOptions();
|
---|
342 | my $strBaseCMD = $wgetOptions." -q -O - \"$self->{'url'}?_OPTS_\"";
|
---|
343 |
|
---|
344 | my $strIdentify = "verb=Identify";
|
---|
345 | my $strListSets = "verb=ListSets";
|
---|
346 |
|
---|
347 | my $strIdentifyCMD = $strBaseCMD;
|
---|
348 | $strIdentifyCMD =~ s/_OPTS_/$strIdentify/;
|
---|
349 |
|
---|
350 | my $strIdentifyText = $self->useWget($strIdentifyCMD);
|
---|
351 |
|
---|
352 | if (!defined $strIdentifyText or $strIdentifyText eq "" ){
|
---|
353 | print STDERR "Server information is unavailable.\n";
|
---|
354 | print STDERR "<<Finished>>\n";
|
---|
355 | return;
|
---|
356 | }
|
---|
357 |
|
---|
358 | print STDERR "General information:\n";
|
---|
359 | $self->parse_xml($strIdentifyText);
|
---|
360 |
|
---|
361 | my $strListSetCMD = $strBaseCMD;
|
---|
362 | $strListSetCMD =~ s/_OPTS_/$strListSets/;
|
---|
363 | my $strListSetsText = $self->useWget($strListSetCMD);
|
---|
364 |
|
---|
365 |
|
---|
366 | print STDERR "List Information:\n";
|
---|
367 | $self->parse_xml($strListSetsText);
|
---|
368 | }
|
---|
369 |
|
---|
370 | sub parse_xml
|
---|
371 | {
|
---|
372 | my ($self) = shift (@_);
|
---|
373 | my ($strOutputText) = @_;
|
---|
374 |
|
---|
375 | #Open a temporary file to store OAI information, and store the information to the temp file
|
---|
376 | my $name = "$ENV{GSDLHOME}/tmp/oai.tmp";
|
---|
377 |
|
---|
378 | open(*OAIOUT,"> $name");
|
---|
379 |
|
---|
380 | print OAIOUT $strOutputText;
|
---|
381 | close(OAIOUT);
|
---|
382 |
|
---|
383 | $self->{'temp_file_name'} = $name;
|
---|
384 |
|
---|
385 | eval {
|
---|
386 | $self->{'parser'}->parsefile("$name");
|
---|
387 | };
|
---|
388 |
|
---|
389 | if ($@) {
|
---|
390 | die "OAI: $name is not a well formed XML file ($@)\n";
|
---|
391 | }
|
---|
392 | }
|
---|
393 |
|
---|
394 | END{
|
---|
395 | if($self->{'info'})
|
---|
396 | {
|
---|
397 | unlink($self->{'temp_file_name'}) or die "Could not unlink $self->{'temp_file_name'}: $!";
|
---|
398 | }
|
---|
399 | }
|
---|
400 |
|
---|
401 | # This Char function overrides the one in XML::Parser::Stream to overcome a
|
---|
402 | # problem where $expat->{Text} is treated as the return value, slowing
|
---|
403 | # things down significantly in some cases.
|
---|
404 | sub Char {
|
---|
405 | use bytes; # Necessary to prevent encoding issues with XML::Parser 2.31+
|
---|
406 | $_[0]->{'Text'} .= $_[1];
|
---|
407 | if ((defined $self->{'subfield'} && ($self->{'subfield'} ne ""))) {
|
---|
408 | $self->{'text'} .= $_[1];
|
---|
409 | $self->{'text'} =~ s/[\n]|([ ]{2,})//g;
|
---|
410 | if($self->{'text'} ne "")
|
---|
411 | {
|
---|
412 | print STDERR " $self->{'subfield'}:($self->{'text'})\n";
|
---|
413 | }
|
---|
414 | }
|
---|
415 | return undef;
|
---|
416 | }
|
---|
417 |
|
---|
418 | sub OAI_StartTag
|
---|
419 | {
|
---|
420 | my ($expat, $element, %attr) = @_;
|
---|
421 |
|
---|
422 | $self->{'subfield'} = $element;
|
---|
423 |
|
---|
424 | }
|
---|
425 |
|
---|
426 | sub OAI_EndTag
|
---|
427 | {
|
---|
428 | my ($expat, $element) = @_;
|
---|
429 | $self->{'text'} = "";
|
---|
430 | $self->{'subfield'} = "";
|
---|
431 | }
|
---|
432 |
|
---|
433 | sub error
|
---|
434 | {
|
---|
435 | my ($self,$strFunctionName,$strError) = @_;
|
---|
436 | {
|
---|
437 | print "Error occoured in OAIDownload.pm\n".
|
---|
438 | "In Function:".$strFunctionName."\n".
|
---|
439 | "Error Message:".$strError."\n";
|
---|
440 | exit(-1);
|
---|
441 | }
|
---|
442 | }
|
---|
443 |
|
---|
444 |
|
---|
445 |
|
---|
446 | 1;
|
---|