We could not synchronize checks via GitHub's checks API since Scrutinizer's GitHub App is not installed for this repository.
1 | <?php |
||
2 | |||
3 | /** |
||
4 | * (c) Kitodo. Key to digital objects e.V. <[email protected]> |
||
5 | * |
||
6 | * This file is part of the Kitodo and TYPO3 projects. |
||
7 | * |
||
8 | * @license GNU General Public License version 3 or later. |
||
9 | * For the full copyright and license information, please read the |
||
10 | * LICENSE.txt file that was distributed with this source code. |
||
11 | */ |
||
12 | |||
13 | namespace Kitodo\Dlf\Command; |
||
14 | |||
15 | use Kitodo\Dlf\Common\AbstractDocument; |
||
16 | use Kitodo\Dlf\Command\BaseCommand; |
||
17 | use Kitodo\Dlf\Common\Indexer; |
||
18 | use Kitodo\Dlf\Domain\Model\Document; |
||
19 | use Symfony\Component\Console\Input\InputInterface; |
||
20 | use Symfony\Component\Console\Input\InputOption; |
||
21 | use Symfony\Component\Console\Output\OutputInterface; |
||
22 | use Symfony\Component\Console\Style\SymfonyStyle; |
||
23 | use TYPO3\CMS\Core\Utility\GeneralUtility; |
||
24 | use TYPO3\CMS\Core\Utility\MathUtility; |
||
25 | use Phpoaipmh\Endpoint; |
||
26 | use Phpoaipmh\Exception\BaseOaipmhException; |
||
27 | |||
28 | /** |
||
29 | * CLI Command for harvesting OAI-PMH interfaces into database and Solr. |
||
30 | * |
||
31 | * @package TYPO3 |
||
32 | * @subpackage dlf |
||
33 | * |
||
34 | * @access public |
||
35 | */ |
||
36 | class HarvestCommand extends BaseCommand |
||
37 | { |
||
38 | /** |
||
39 | * Configure the command by defining the name, options and arguments |
||
40 | * |
||
41 | * @access public |
||
42 | * |
||
43 | * @return void |
||
44 | */ |
||
45 | public function configure(): void |
||
46 | { |
||
47 | $this |
||
48 | ->setDescription('Harvest OAI-PMH contents into database and Solr.') |
||
49 | ->setHelp('') |
||
50 | ->addOption( |
||
51 | 'dry-run', |
||
52 | null, |
||
53 | InputOption::VALUE_NONE, |
||
54 | 'If this option is set, the files will not actually be processed but the location URIs are shown.' |
||
55 | ) |
||
56 | ->addOption( |
||
57 | 'lib', |
||
58 | 'l', |
||
59 | InputOption::VALUE_REQUIRED, |
||
60 | 'UID of the library to harvest.' |
||
61 | ) |
||
62 | ->addOption( |
||
63 | 'pid', |
||
64 | 'p', |
||
65 | InputOption::VALUE_REQUIRED, |
||
66 | 'UID of the page the documents should be added to.' |
||
67 | ) |
||
68 | ->addOption( |
||
69 | 'solr', |
||
70 | 's', |
||
71 | InputOption::VALUE_REQUIRED, |
||
72 | '[UID|index_name] of the Solr core the document should be added to.' |
||
73 | ) |
||
74 | ->addOption( |
||
75 | 'from', |
||
76 | null, |
||
77 | InputOption::VALUE_OPTIONAL, |
||
78 | 'Datestamp (YYYY-MM-DD) to begin harvesting from.' |
||
79 | ) |
||
80 | ->addOption( |
||
81 | 'until', |
||
82 | null, |
||
83 | InputOption::VALUE_OPTIONAL, |
||
84 | 'Datestamp (YYYY-MM-DD) to end harvesting on.' |
||
85 | ) |
||
86 | ->addOption( |
||
87 | 'set', |
||
88 | null, |
||
89 | InputOption::VALUE_OPTIONAL, |
||
90 | 'Name of the set to limit harvesting to.' |
||
91 | ) |
||
92 | ->addOption( |
||
93 | 'softCommit', |
||
94 | null, |
||
95 | InputOption::VALUE_NONE, |
||
96 | 'If this option is set, documents are just added to the index by a soft commit.' |
||
97 | ); |
||
98 | } |
||
99 | |||
100 | /** |
||
101 | * Executes the command to index the given document to DB and SOLR. |
||
102 | * |
||
103 | * @access protected |
||
104 | * |
||
105 | * @param InputInterface $input The input parameters |
||
106 | * @param OutputInterface $output The Symfony interface for outputs on console |
||
107 | * |
||
108 | * @return int |
||
109 | */ |
||
110 | protected function execute(InputInterface $input, OutputInterface $output): int |
||
111 | { |
||
112 | $dryRun = $input->getOption('dry-run') != false ? true : false; |
||
113 | |||
114 | $io = new SymfonyStyle($input, $output); |
||
115 | $io->title($this->getDescription()); |
||
116 | |||
117 | $this->initializeRepositories((int) $input->getOption('pid')); |
||
118 | |||
119 | if ($this->storagePid == 0) { |
||
120 | $io->error('ERROR: No valid PID (' . $this->storagePid . ') given.'); |
||
121 | return BaseCommand::FAILURE; |
||
122 | } |
||
123 | |||
124 | if ( |
||
125 | !empty($input->getOption('solr')) |
||
126 | && !is_array($input->getOption('solr')) |
||
127 | ) { |
||
128 | $allSolrCores = $this->getSolrCores($this->storagePid); |
||
129 | if (MathUtility::canBeInterpretedAsInteger($input->getOption('solr'))) { |
||
130 | $solrCoreUid = MathUtility::forceIntegerInRange((int) $input->getOption('solr'), 0); |
||
131 | } else { |
||
132 | $solrCoreUid = $allSolrCores[$input->getOption('solr')]; |
||
133 | } |
||
134 | // Abort if solrCoreUid is empty or not in the array of allowed solr cores. |
||
135 | if (empty($solrCoreUid) || !in_array($solrCoreUid, $allSolrCores)) { |
||
136 | $outputSolrCores = []; |
||
137 | foreach ($allSolrCores as $indexName => $uid) { |
||
138 | $outputSolrCores[] = $uid . ' : ' . $indexName; |
||
139 | } |
||
140 | if (empty($outputSolrCores)) { |
||
141 | $io->error('ERROR: No valid Solr core ("' . $input->getOption('solr') . '") given. No valid cores found on PID ' . $this->storagePid . ".\n"); |
||
142 | return BaseCommand::FAILURE; |
||
143 | } else { |
||
144 | $io->error('ERROR: No valid Solr core ("' . $input->getOption('solr') . '") given. ' . "Valid cores are (<uid>:<index_name>):\n" . implode("\n", $outputSolrCores) . "\n"); |
||
145 | return BaseCommand::FAILURE; |
||
146 | } |
||
147 | } |
||
148 | } else { |
||
149 | $io->error('ERROR: Required parameter --solr|-s is missing or array.'); |
||
150 | return BaseCommand::FAILURE; |
||
151 | } |
||
152 | |||
153 | if (MathUtility::canBeInterpretedAsInteger($input->getOption('lib'))) { |
||
154 | $this->owner = $this->libraryRepository->findByUid(MathUtility::forceIntegerInRange((int) $input->getOption('lib'), 1)); |
||
155 | } |
||
156 | |||
157 | if ($this->owner) { |
||
158 | $baseUrl = $this->owner->getOaiBase(); |
||
159 | } else { |
||
160 | $io->error('ERROR: Required parameter --lib|-l is not a valid UID.'); |
||
161 | return BaseCommand::FAILURE; |
||
162 | } |
||
163 | if (!GeneralUtility::isValidUrl($baseUrl)) { |
||
164 | $io->error('ERROR: No valid OAI Base URL set for library with given UID ("' . $input->getOption('lib') . '").'); |
||
165 | return BaseCommand::FAILURE; |
||
166 | } else { |
||
167 | try { |
||
168 | $oai = Endpoint::build($baseUrl); |
||
169 | } catch (BaseoaipmhException $e) { |
||
170 | $this->handleOaiError($e, $io); |
||
171 | } |
||
172 | } |
||
173 | |||
174 | if ( |
||
175 | !is_array($input->getOption('from')) |
||
176 | && preg_match('/^\d{4}-\d{2}-\d{2}$/', $input->getOption('from')) |
||
177 | ) { |
||
178 | $from = new \DateTime($input->getOption('from')); |
||
179 | } else { |
||
180 | $from = null; |
||
181 | } |
||
182 | |||
183 | if ( |
||
184 | !is_array($input->getOption('until')) |
||
185 | && preg_match('/^\d{4}-\d{2}-\d{2}$/', $input->getOption('until')) |
||
186 | ) { |
||
187 | $until = new \DateTime($input->getOption('until')); |
||
188 | } else { |
||
189 | $until = null; |
||
190 | } |
||
191 | |||
192 | $set = null; |
||
193 | if ( |
||
194 | !is_array($input->getOption('set')) |
||
195 | && !empty($input->getOption('set')) |
||
196 | && !empty($oai) |
||
197 | ) { |
||
198 | $setsAvailable = $oai->listSets(); |
||
199 | foreach ($setsAvailable as $setAvailable) { |
||
200 | if ((string) $setAvailable->setSpec === $input->getOption('set')) { |
||
201 | $set = $input->getOption('set'); |
||
202 | break; |
||
203 | } |
||
204 | } |
||
205 | if (empty($set)) { |
||
206 | $io->error('ERROR: OAI interface does not provide a set with given setSpec ("' . $input->getOption('set') . '").'); |
||
207 | return BaseCommand::FAILURE; |
||
208 | } |
||
209 | } |
||
210 | |||
211 | $identifiers = []; |
||
212 | // Get OAI record identifiers to process. |
||
213 | try { |
||
214 | if (!empty($oai)) { |
||
215 | $identifiers = $oai->listIdentifiers('mets', $from, $until, $set); |
||
216 | } else { |
||
217 | $io->error('ERROR: OAI interface does not exist.'); |
||
218 | } |
||
219 | } catch (BaseoaipmhException $exception) { |
||
220 | $this->handleOaiError($exception, $io); |
||
221 | } |
||
222 | |||
223 | // Process all identifiers. |
||
224 | $baseLocation = $baseUrl . (parse_url($baseUrl, PHP_URL_QUERY) ? '&' : '?'); |
||
225 | foreach ($identifiers as $identifier) { |
||
226 | // Build OAI GetRecord URL... |
||
227 | $params = [ |
||
228 | 'verb' => 'GetRecord', |
||
229 | 'metadataPrefix' => 'mets', |
||
230 | 'identifier' => (string) $identifier->identifier |
||
231 | ]; |
||
232 | $docLocation = $baseLocation . http_build_query($params); |
||
233 | // ...index the document... |
||
234 | $document = null; |
||
235 | $doc = AbstractDocument::getInstance($docLocation, ['storagePid' => $this->storagePid], true); |
||
236 | |||
237 | if ($doc === null) { |
||
238 | $io->warning('WARNING: Document "' . $docLocation . '" could not be loaded. Skip to next document.'); |
||
239 | continue; |
||
240 | } |
||
241 | |||
242 | if ($doc->recordId) { |
||
243 | $document = $this->documentRepository->findOneByRecordId($doc->recordId); |
||
0 ignored issues
–
show
Bug
introduced
by
Loading history...
|
|||
244 | } |
||
245 | |||
246 | if ($document === null) { |
||
247 | // create new Document object |
||
248 | $document = GeneralUtility::makeInstance(Document::class); |
||
249 | } |
||
250 | |||
251 | $document->setLocation($docLocation); |
||
252 | $document->setSolrcore($solrCoreUid); |
||
253 | |||
254 | if ($dryRun) { |
||
255 | $io->writeln('DRY RUN: Would index ' . $document->getUid() . ' ("' . $document->getLocation() . '") on PID ' . $this->storagePid . ' and Solr core ' . $solrCoreUid . '.'); |
||
256 | } else { |
||
257 | if ($io->isVerbose()) { |
||
258 | $io->writeln(date('Y-m-d H:i:s') . ' Indexing ' . $document->getUid() . ' ("' . $document->getLocation() . '") on PID ' . $this->storagePid . ' and Solr core ' . $solrCoreUid . '.'); |
||
259 | } |
||
260 | $document->setCurrentDocument($doc); |
||
261 | // save to database |
||
262 | $this->saveToDatabase($document, $input->getOption('softCommit')); |
||
263 | // add to index |
||
264 | Indexer::add($document, $this->documentRepository, $input->getOption('softCommit')); |
||
265 | } |
||
266 | } |
||
267 | |||
268 | $io->success('All done!'); |
||
269 | |||
270 | return BaseCommand::SUCCESS; |
||
271 | } |
||
272 | |||
273 | /** |
||
274 | * Handles OAI errors |
||
275 | * |
||
276 | * @access protected |
||
277 | * |
||
278 | * @param BaseoaipmhException $exception Instance of exception thrown |
||
279 | * @param SymfonyStyle $io |
||
280 | * |
||
281 | * @return void |
||
282 | */ |
||
283 | protected function handleOaiError(BaseoaipmhException $exception, SymfonyStyle $io): void |
||
284 | { |
||
285 | $io->error('ERROR: Trying to retrieve data from OAI interface resulted in error:' . "\n " . $exception->getMessage()); |
||
286 | } |
||
287 | } |
||
288 |