Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
95.20% |
238 / 250 |
|
77.78% |
7 / 9 |
CRAP | |
0.00% |
0 / 1 |
| HarvesterCommand | |
95.20% |
238 / 250 |
|
77.78% |
7 / 9 |
35 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
4 | |||
| configure | |
100.00% |
160 / 160 |
|
100.00% |
1 / 1 |
1 | |||
| updateSettingsWithConsoleOptions | |
100.00% |
8 / 8 |
|
100.00% |
1 / 1 |
6 | |||
| execute | |
74.42% |
32 / 43 |
|
0.00% |
0 / 1 |
15.83 | |||
| getHarvestRoot | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getHttpClient | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getSettingsFromIni | |
100.00% |
10 / 10 |
|
100.00% |
1 / 1 |
4 | |||
| getSettings | |
90.91% |
10 / 11 |
|
0.00% |
0 / 1 |
4.01 | |||
| harvestSingleRepository | |
100.00% |
11 / 11 |
|
100.00% |
1 / 1 |
1 | |||
| 1 | <?php |
| 2 | |
| 3 | /** |
| 4 | * OAI-PMH Harvest Tool (Symfony Console Command) |
| 5 | * |
| 6 | * PHP version 7 |
| 7 | * |
| 8 | * Copyright (c) Demian Katz 2016. |
| 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 version 2, |
| 12 | * as published by the Free Software Foundation. |
| 13 | * |
| 14 | * This program is distributed in the hope that it will be useful, |
| 15 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 16 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 17 | * GNU General Public License for more details. |
| 18 | * |
| 19 | * You should have received a copy of the GNU General Public License |
| 20 | * along with this program; if not, write to the Free Software |
| 21 | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
| 22 | * |
| 23 | * @category VuFind |
| 24 | * @package Harvest_Tools |
| 25 | * @author Demian Katz <demian.katz@villanova.edu> |
| 26 | * @license http://opensource.org/licenses/gpl-2.0.php GNU General Public License |
| 27 | * @link https://vufind.org/wiki/indexing:oai-pmh Wiki |
| 28 | */ |
| 29 | |
| 30 | namespace VuFindHarvest\OaiPmh; |
| 31 | |
| 32 | use Laminas\Http\Client; |
| 33 | use Symfony\Component\Console\Attribute\AsCommand; |
| 34 | use Symfony\Component\Console\Command\Command; |
| 35 | use Symfony\Component\Console\Input\InputArgument; |
| 36 | use Symfony\Component\Console\Input\InputInterface; |
| 37 | use Symfony\Component\Console\Input\InputOption; |
| 38 | use Symfony\Component\Console\Output\OutputInterface; |
| 39 | use VuFindHarvest\ConsoleOutput\ConsoleWriter; |
| 40 | use VuFindHarvest\ConsoleOutput\WriterAwareTrait; |
| 41 | use VuFindHarvest\Exception\OaiException; |
| 42 | |
| 43 | /** |
| 44 | * OAI-PMH Harvest Tool (Symfony Console Command) |
| 45 | * |
| 46 | * @category VuFind |
| 47 | * @package Harvest_Tools |
| 48 | * @author Demian Katz <demian.katz@villanova.edu> |
| 49 | * @license http://opensource.org/licenses/gpl-2.0.php GNU General Public License |
| 50 | * @link https://vufind.org/wiki/indexing:oai-pmh Wiki |
| 51 | */ |
| 52 | #[AsCommand( |
| 53 | name: 'harvest/harvest_oai', |
| 54 | description: 'OAI-PMH harvester' |
| 55 | )] |
| 56 | class HarvesterCommand extends Command |
| 57 | { |
| 58 | use WriterAwareTrait; |
| 59 | |
| 60 | /** |
| 61 | * The name of the command |
| 62 | * |
| 63 | * @var string |
| 64 | */ |
| 65 | protected static $defaultName = 'harvest/harvest_oai'; |
| 66 | |
| 67 | /** |
| 68 | * HTTP client |
| 69 | * |
| 70 | * @var Client |
| 71 | */ |
| 72 | protected $client; |
| 73 | |
| 74 | /** |
| 75 | * Root directory for harvesting |
| 76 | * |
| 77 | * @var string |
| 78 | */ |
| 79 | protected $harvestRoot; |
| 80 | |
| 81 | /** |
| 82 | * Harvester factory |
| 83 | * |
| 84 | * @var HarvesterFactory |
| 85 | */ |
| 86 | protected $factory; |
| 87 | |
| 88 | /** |
| 89 | * Silent mode |
| 90 | * |
| 91 | * @var bool |
| 92 | */ |
| 93 | protected $silent; |
| 94 | |
| 95 | /** |
| 96 | * Constructor |
| 97 | * |
| 98 | * @param Client $client HTTP client (omit for default) |
| 99 | * @param string $harvestRoot Root directory for harvesting (omit for |
| 100 | * default) |
| 101 | * @param HarvesterFactory $factory Harvester factory (omit for default) |
| 102 | * @param bool $silent Should we suppress output? |
| 103 | * @param string|null $name The name of the command; passing null |
| 104 | * means it must be set in configure() |
| 105 | */ |
| 106 | public function __construct( |
| 107 | $client = null, |
| 108 | $harvestRoot = null, |
| 109 | HarvesterFactory $factory = null, |
| 110 | $silent = false, |
| 111 | $name = null |
| 112 | ) { |
| 113 | $this->client = $client ?: new Client(); |
| 114 | $this->harvestRoot = $harvestRoot ?: getcwd(); |
| 115 | $this->factory = $factory ?: new HarvesterFactory(); |
| 116 | $this->silent = $silent; |
| 117 | parent::__construct($name); |
| 118 | } |
| 119 | |
| 120 | /** |
| 121 | * Configure the command. |
| 122 | * |
| 123 | * @return void |
| 124 | * |
| 125 | * @SuppressWarnings(PHPMD.ExcessiveMethodLength) |
| 126 | */ |
| 127 | protected function configure() |
| 128 | { |
| 129 | $this |
| 130 | ->setHelp('Harvests metadata using the OAI-PMH protocol.') |
| 131 | ->addArgument( |
| 132 | 'target', |
| 133 | InputArgument::OPTIONAL, |
| 134 | 'the name of a section of the configuration specified by the ini ' |
| 135 | . "option,\nor a directory to harvest into if no .ini file is used. " |
| 136 | . "If <target> is\nomitted, all .ini sections will be processed." |
| 137 | )->addOption( |
| 138 | 'from', |
| 139 | null, |
| 140 | InputOption::VALUE_REQUIRED, |
| 141 | 'Harvest start date' |
| 142 | )->addOption( |
| 143 | 'until', |
| 144 | null, |
| 145 | InputOption::VALUE_REQUIRED, |
| 146 | 'Harvest end date' |
| 147 | )->addOption( |
| 148 | 'ini', |
| 149 | null, |
| 150 | InputOption::VALUE_REQUIRED, |
| 151 | '.ini file to load; if you set other more specific options, they' |
| 152 | . " will\noverride equivalent settings loaded from the .ini file." |
| 153 | )->addOption( |
| 154 | 'url', |
| 155 | null, |
| 156 | InputOption::VALUE_REQUIRED, |
| 157 | 'Base URL of OAI-PMH server' |
| 158 | )->addOption( |
| 159 | 'httpUser', |
| 160 | null, |
| 161 | InputOption::VALUE_REQUIRED, |
| 162 | 'Username to access url' |
| 163 | )->addOption( |
| 164 | 'httpPass', |
| 165 | null, |
| 166 | InputOption::VALUE_REQUIRED, |
| 167 | 'Password to access url' |
| 168 | )->addOption( |
| 169 | 'set', |
| 170 | null, |
| 171 | InputOption::VALUE_REQUIRED, |
| 172 | 'Set name to harvest' |
| 173 | )->addOption( |
| 174 | 'metadataPrefix', |
| 175 | null, |
| 176 | InputOption::VALUE_REQUIRED, |
| 177 | 'Metadata prefix to harvest' |
| 178 | )->addOption( |
| 179 | 'timeout', |
| 180 | null, |
| 181 | InputOption::VALUE_REQUIRED, |
| 182 | 'HTTP timeout (in seconds)' |
| 183 | )->addOption( |
| 184 | 'combineRecords', |
| 185 | null, |
| 186 | InputOption::VALUE_NONE, |
| 187 | 'Turn off "one record per file" mode' |
| 188 | )->addOption( |
| 189 | 'combineRecordsTag', |
| 190 | null, |
| 191 | InputOption::VALUE_REQUIRED, |
| 192 | 'Specify the XML tag wrapped around multiple records in ' |
| 193 | . "combineRecords\nmode (default = <collection> if this " |
| 194 | . 'option is omitted)' |
| 195 | )->addOption( |
| 196 | 'globalSearch', |
| 197 | null, |
| 198 | InputOption::VALUE_REQUIRED, |
| 199 | 'Regular expression to replace in raw XML' |
| 200 | )->addOption( |
| 201 | 'globalReplace', |
| 202 | null, |
| 203 | InputOption::VALUE_REQUIRED, |
| 204 | 'String to replace globalSearch regex matches' |
| 205 | )->addOption( |
| 206 | 'injectDate', |
| 207 | null, |
| 208 | InputOption::VALUE_REQUIRED, |
| 209 | 'Inject date from header into specified tag' |
| 210 | )->addOption( |
| 211 | 'injectId', |
| 212 | null, |
| 213 | InputOption::VALUE_REQUIRED, |
| 214 | 'Inject ID from header into specified tag' |
| 215 | )->addOption( |
| 216 | 'injectSetName', |
| 217 | null, |
| 218 | InputOption::VALUE_REQUIRED, |
| 219 | 'Inject setName from header into specified tag' |
| 220 | )->addOption( |
| 221 | 'injectSetSpec', |
| 222 | null, |
| 223 | InputOption::VALUE_REQUIRED, |
| 224 | 'Inject setSpec from header into specified tag' |
| 225 | )->addOption( |
| 226 | 'idSearch', |
| 227 | null, |
| 228 | InputOption::VALUE_REQUIRED, |
| 229 | 'Regular expression to replace in ID' |
| 230 | . ' (only relevant when injectId is on)' |
| 231 | )->addOption( |
| 232 | 'idReplace', |
| 233 | null, |
| 234 | InputOption::VALUE_REQUIRED, |
| 235 | 'String to replace idSearch regex matches' |
| 236 | )->addOption( |
| 237 | 'dateGranularity', |
| 238 | null, |
| 239 | InputOption::VALUE_REQUIRED, |
| 240 | '"YYYY-MM-DDThh:mm:ssZ," "YYYY-MM-DD" or "auto" (default)' |
| 241 | )->addOption( |
| 242 | 'harvestedIdLog', |
| 243 | null, |
| 244 | InputOption::VALUE_REQUIRED, |
| 245 | 'Filename (relative to harvest directory)' |
| 246 | . ' to store log of harvested IDs.' |
| 247 | )->addOption( |
| 248 | 'autosslca', |
| 249 | null, |
| 250 | InputOption::VALUE_NONE, |
| 251 | 'Attempt to autodetect SSL certificate file/path' |
| 252 | )->addOption( |
| 253 | 'sslcapath', |
| 254 | null, |
| 255 | InputOption::VALUE_REQUIRED, |
| 256 | 'Path to SSL certificate authority directory' |
| 257 | )->addOption( |
| 258 | 'sslcafile', |
| 259 | null, |
| 260 | InputOption::VALUE_REQUIRED, |
| 261 | 'Path to SSL certificate authority file' |
| 262 | )->addOption( |
| 263 | 'nosslverifypeer', |
| 264 | null, |
| 265 | InputOption::VALUE_NONE, |
| 266 | 'Disable SSL verification' |
| 267 | )->addOption( |
| 268 | 'sanitize', |
| 269 | null, |
| 270 | InputOption::VALUE_NONE, |
| 271 | 'Strip illegal characters from XML' |
| 272 | )->addOption( |
| 273 | 'sanitizeRegex', |
| 274 | null, |
| 275 | InputOption::VALUE_REQUIRED, |
| 276 | 'Optional regular expression defining XML characters to remove' |
| 277 | )->addOption( |
| 278 | 'badXMLLog', |
| 279 | null, |
| 280 | InputOption::VALUE_REQUIRED, |
| 281 | 'Filename (relative to harvest directory) to log' |
| 282 | . ' XML fixed by sanitize setting' |
| 283 | )->addOption( |
| 284 | 'stopAfter', |
| 285 | null, |
| 286 | InputOption::VALUE_NONE, |
| 287 | 'an option to stop harvesting after the first n records of each set.' |
| 288 | ); |
| 289 | } |
| 290 | |
| 291 | /** |
| 292 | * Use command-line switches to add/override settings found in the .ini |
| 293 | * file, if necessary. |
| 294 | * |
| 295 | * @param InputInterface $input Input object |
| 296 | * @param array $settings Incoming settings |
| 297 | * |
| 298 | * @return array |
| 299 | */ |
| 300 | protected function updateSettingsWithConsoleOptions( |
| 301 | InputInterface $input, |
| 302 | $settings |
| 303 | ) { |
| 304 | $directMapSettings = [ |
| 305 | 'url', 'set', 'metadataPrefix', 'timeout', 'combineRecordsTag', |
| 306 | 'injectDate', 'injectId', 'injectSetName', 'injectSetSpec', |
| 307 | 'idSearch', 'idReplace', 'dateGranularity', 'harvestedIdLog', |
| 308 | 'badXMLLog', 'httpUser', 'httpPass', 'sslcapath', 'sslcafile', |
| 309 | 'sanitizeRegex', |
| 310 | ]; |
| 311 | foreach ($directMapSettings as $setting) { |
| 312 | if ($value = $input->getOption($setting)) { |
| 313 | $settings[$setting] = $value; |
| 314 | } |
| 315 | } |
| 316 | $flagSettings = [ |
| 317 | 'combineRecords' => ['combineRecords', true], |
| 318 | 'verbose' => ['verbose', true], |
| 319 | 'autosslca' => ['autosslca', true], |
| 320 | 'nosslverifypeer' => ['sslverifypeer', false], |
| 321 | 'sanitize' => ['sanitize', true], |
| 322 | ]; |
| 323 | foreach ($flagSettings as $in => $details) { |
| 324 | if ($input->hasOption($in) && $input->getOption($in)) { |
| 325 | [$out, $val] = $details; |
| 326 | $settings[$out] = $val; |
| 327 | } |
| 328 | } |
| 329 | return $settings; |
| 330 | } |
| 331 | |
| 332 | /** |
| 333 | * Run the command. |
| 334 | * |
| 335 | * @param InputInterface $input Input object |
| 336 | * @param OutputInterface $output Output object |
| 337 | * |
| 338 | * @return int 0 for success |
| 339 | */ |
| 340 | protected function execute(InputInterface $input, OutputInterface $output) |
| 341 | { |
| 342 | // Only set up output writer if not in silent mode: |
| 343 | if (!$this->silent) { |
| 344 | $this->setOutputWriter(new ConsoleWriter($output)); |
| 345 | } |
| 346 | |
| 347 | if (!$allSettings = $this->getSettings($input)) { |
| 348 | return 1; |
| 349 | } |
| 350 | |
| 351 | // Loop through all the settings and perform harvests: |
| 352 | $processed = $skipped = $errors = 0; |
| 353 | foreach ($allSettings as $target => $baseSettings) { |
| 354 | $settings = $this->updateSettingsWithConsoleOptions( |
| 355 | $input, |
| 356 | $baseSettings |
| 357 | ); |
| 358 | if (empty($target) || empty($settings)) { |
| 359 | $skipped++; |
| 360 | continue; |
| 361 | } |
| 362 | $this->writeLine("Processing {$target}..."); |
| 363 | try { |
| 364 | $this->harvestSingleRepository($input, $output, $target, $settings); |
| 365 | } catch (\Exception $e) { |
| 366 | if ( |
| 367 | $e instanceof OaiException |
| 368 | && strtolower($e->getOaiCode()) == 'norecordsmatch' |
| 369 | ) { |
| 370 | $this->writeLine('No new records found.'); |
| 371 | } else { |
| 372 | $this->writeLine($e->getMessage()); |
| 373 | $errors++; |
| 374 | } |
| 375 | } |
| 376 | $processed++; |
| 377 | } |
| 378 | |
| 379 | // All done. |
| 380 | if (isset($settings['stopAfter'])) { |
| 381 | $this->writeLine( |
| 382 | 'stopAfter option set; ' |
| 383 | . 'all sources may not have been fully harvested.' |
| 384 | ); |
| 385 | } |
| 386 | if ($processed == 0 && $skipped > 0) { |
| 387 | $this->writeLine( |
| 388 | 'No valid settings found; ' |
| 389 | . 'please set url and metadataPrefix at minimum.' |
| 390 | ); |
| 391 | return 1; |
| 392 | } |
| 393 | if ($errors > 0) { |
| 394 | $this->writeLine( |
| 395 | "Completed with {$errors} error(s) -- " |
| 396 | . "{$processed} source(s) processed." |
| 397 | ); |
| 398 | return 1; |
| 399 | } |
| 400 | $this->writeLine( |
| 401 | "Completed without errors -- {$processed} source(s) processed." |
| 402 | ); |
| 403 | return 0; |
| 404 | } |
| 405 | |
| 406 | /** |
| 407 | * Get the target directory for writing harvested files. |
| 408 | * |
| 409 | * @return string |
| 410 | */ |
| 411 | protected function getHarvestRoot() |
| 412 | { |
| 413 | return $this->harvestRoot; |
| 414 | } |
| 415 | |
| 416 | /** |
| 417 | * Get an HTTP client. |
| 418 | * |
| 419 | * @return Client |
| 420 | */ |
| 421 | protected function getHttpClient() |
| 422 | { |
| 423 | return $this->client; |
| 424 | } |
| 425 | |
| 426 | /** |
| 427 | * Load configuration from an .ini file (or return false on error) |
| 428 | * |
| 429 | * @param string $ini Configuration file to load |
| 430 | * @param string|bool $section Section of .ini to load (or false for all) |
| 431 | * |
| 432 | * @return array|bool |
| 433 | */ |
| 434 | protected function getSettingsFromIni($ini, $section) |
| 435 | { |
| 436 | $oaiSettings = @parse_ini_file($ini, true); |
| 437 | if (empty($oaiSettings)) { |
| 438 | $this->writeLine("Please add OAI-PMH settings to {$ini}."); |
| 439 | return false; |
| 440 | } |
| 441 | if ($section) { |
| 442 | if (!isset($oaiSettings[$section])) { |
| 443 | $this->writeLine("$section not found in $ini."); |
| 444 | return false; |
| 445 | } |
| 446 | $oaiSettings = [$section => $oaiSettings[$section]]; |
| 447 | } |
| 448 | return $oaiSettings; |
| 449 | } |
| 450 | |
| 451 | /** |
| 452 | * Load the harvest settings. Return false on error. |
| 453 | * |
| 454 | * @param InputInterface $input Input object |
| 455 | * |
| 456 | * @return array|bool |
| 457 | */ |
| 458 | protected function getSettings(InputInterface $input) |
| 459 | { |
| 460 | $ini = $input->getOption('ini'); |
| 461 | $section = $input->getArgument('target'); |
| 462 | if (!$ini && !$section) { |
| 463 | $this->writeLine( |
| 464 | 'Please specify an .ini file with the --ini flag' |
| 465 | . ' or a target directory with the first parameter.' |
| 466 | ); |
| 467 | return false; |
| 468 | } |
| 469 | return $ini |
| 470 | ? $this->getSettingsFromIni($ini, $section) |
| 471 | : [$section => []]; |
| 472 | } |
| 473 | |
| 474 | /** |
| 475 | * Harvest a single repository. |
| 476 | * |
| 477 | * @param InputInterface $input Input object |
| 478 | * @param OutputInterface $output Output object |
| 479 | * @param string $target Name of repo (used for target directory) |
| 480 | * @param array $settings Settings for the harvester. |
| 481 | * |
| 482 | * @return void |
| 483 | * @throws \Exception |
| 484 | */ |
| 485 | protected function harvestSingleRepository( |
| 486 | InputInterface $input, |
| 487 | OutputInterface $output, |
| 488 | $target, |
| 489 | $settings |
| 490 | ) { |
| 491 | $settings['from'] = $input->getOption('from'); |
| 492 | $settings['until'] = $input->getOption('until'); |
| 493 | $settings['silent'] = false; |
| 494 | $harvest = $this->factory->getHarvester( |
| 495 | $target, |
| 496 | $this->getHarvestRoot(), |
| 497 | $this->getHttpClient(), |
| 498 | $settings, |
| 499 | $output |
| 500 | ); |
| 501 | $harvest->launch(); |
| 502 | } |
| 503 | } |