LuceneTest.php 15 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487
  1. <?php
  2. /**
  3. * @category Zend
  4. * @package Zend_Search_Lucene
  5. * @subpackage UnitTests
  6. */
  7. /**
  8. * Zend_Search_Lucene
  9. */
  10. require_once 'Zend/Search/Lucene.php';
  11. /**
  12. * PHPUnit test case
  13. */
  14. require_once 'PHPUnit/Framework/TestCase.php';
  15. /**
  16. * @category Zend
  17. * @package Zend_Search_Lucene
  18. * @subpackage UnitTests
  19. */
  20. class Zend_Search_Lucene_LuceneTest extends PHPUnit_Framework_TestCase
  21. {
  22. private function _clearDirectory($dirName)
  23. {
  24. // remove files from temporary direcytory
  25. $dir = opendir($dirName);
  26. while (($file = readdir($dir)) !== false) {
  27. if (!is_dir($dirName . '/' . $file)) {
  28. @unlink($dirName . '/' . $file);
  29. }
  30. }
  31. closedir($dir);
  32. }
  33. public function testCreate()
  34. {
  35. $index = Zend_Search_Lucene::create(dirname(__FILE__) . '/_index/_files');
  36. $this->assertTrue($index instanceof Zend_Search_Lucene_Interface);
  37. $this->_clearDirectory(dirname(__FILE__) . '/_index/_files');
  38. }
  39. public function testOpen()
  40. {
  41. $index = Zend_Search_Lucene::open(dirname(__FILE__) . '/_indexSample/_files');
  42. $this->assertTrue($index instanceof Zend_Search_Lucene_Interface);
  43. }
  44. public function testOpenNonCompound()
  45. {
  46. $index = Zend_Search_Lucene::open(dirname(__FILE__) . '/_indexSample/_nonCompoundIndexFiles');
  47. $this->assertTrue($index instanceof Zend_Search_Lucene_Interface);
  48. }
  49. public function testDefaultSearchField()
  50. {
  51. $currentDefaultSearchField = Zend_Search_Lucene::getDefaultSearchField();
  52. $this->assertEquals($currentDefaultSearchField, null);
  53. Zend_Search_Lucene::setDefaultSearchField('anotherField');
  54. $this->assertEquals(Zend_Search_Lucene::getDefaultSearchField(), 'anotherField');
  55. Zend_Search_Lucene::setDefaultSearchField($currentDefaultSearchField);
  56. }
  57. public function testCount()
  58. {
  59. $index = Zend_Search_Lucene::open(dirname(__FILE__) . '/_indexSample/_files');
  60. $this->assertEquals($index->count(), 10);
  61. }
  62. public function testMaxDoc()
  63. {
  64. $index = Zend_Search_Lucene::open(dirname(__FILE__) . '/_indexSample/_files');
  65. $this->assertEquals($index->maxDoc(), 10);
  66. }
  67. public function testNumDocs()
  68. {
  69. $index = Zend_Search_Lucene::open(dirname(__FILE__) . '/_indexSample/_files');
  70. $this->assertEquals($index->numDocs(), 9);
  71. }
  72. public function testIsDeleted()
  73. {
  74. $index = Zend_Search_Lucene::open(dirname(__FILE__) . '/_indexSample/_files');
  75. $this->assertFalse($index->isDeleted(3));
  76. $this->assertTrue($index->isDeleted(6));
  77. }
  78. public function testMaxBufferedDocs()
  79. {
  80. $index = Zend_Search_Lucene::open(dirname(__FILE__) . '/_indexSample/_files');
  81. $currentMaxBufferedDocs = $index->getMaxBufferedDocs();
  82. $index->setMaxBufferedDocs(234);
  83. $this->assertEquals($index->getMaxBufferedDocs(), 234);
  84. $index->setMaxBufferedDocs($currentMaxBufferedDocs);
  85. }
  86. public function testMaxMergeDocs()
  87. {
  88. $index = Zend_Search_Lucene::open(dirname(__FILE__) . '/_indexSample/_files');
  89. $currentMaxMergeDocs = $index->getMaxMergeDocs();
  90. $index->setMaxMergeDocs(34);
  91. $this->assertEquals($index->getMaxMergeDocs(), 34);
  92. $index->setMaxMergeDocs($currentMaxMergeDocs);
  93. }
  94. public function testMergeFactor()
  95. {
  96. $index = Zend_Search_Lucene::open(dirname(__FILE__) . '/_indexSample/_files');
  97. $currentMergeFactor = $index->getMergeFactor();
  98. $index->setMergeFactor(113);
  99. $this->assertEquals($index->getMergeFactor(), 113);
  100. $index->setMergeFactor($currentMergeFactor);
  101. }
  102. public function testFind()
  103. {
  104. $index = Zend_Search_Lucene::open(dirname(__FILE__) . '/_indexSample/_files');
  105. $hits = $index->find('submitting');
  106. $this->assertEquals(count($hits), 3);
  107. }
  108. public function testGetFieldNames()
  109. {
  110. $index = Zend_Search_Lucene::open(dirname(__FILE__) . '/_indexSample/_files');
  111. $this->assertTrue(array_values($index->getFieldNames()) == array('path', 'modified', 'contents'));
  112. }
  113. public function testGetDocument()
  114. {
  115. $index = Zend_Search_Lucene::open(dirname(__FILE__) . '/_indexSample/_files');
  116. $doc = $index->getDocument(3);
  117. $this->assertTrue($doc instanceof Zend_Search_Lucene_Document);
  118. $this->assertEquals($doc->path, 'IndexSource/about-pear.html');
  119. }
  120. public function testHasTerm()
  121. {
  122. $index = Zend_Search_Lucene::open(dirname(__FILE__) . '/_indexSample/_files');
  123. $this->assertTrue($index->hasTerm(new Zend_Search_Lucene_Index_Term('packages', 'contents')));
  124. $this->assertFalse($index->hasTerm(new Zend_Search_Lucene_Index_Term('nonusedword', 'contents')));
  125. }
  126. public function testTermDocs()
  127. {
  128. $index = Zend_Search_Lucene::open(dirname(__FILE__) . '/_indexSample/_files');
  129. $this->assertTrue(array_values( $index->termDocs(new Zend_Search_Lucene_Index_Term('packages', 'contents')) ) ==
  130. array(0, 2, 6, 7, 8));
  131. }
  132. public function testTermPositions()
  133. {
  134. $index = Zend_Search_Lucene::open(dirname(__FILE__) . '/_indexSample/_files');
  135. $this->assertTrue($index->termPositions(new Zend_Search_Lucene_Index_Term('packages', 'contents')) ==
  136. array(0 => array(174),
  137. 2 => array(40, 742),
  138. 6 => array(6, 156, 163),
  139. 7 => array(194),
  140. 8 => array(55, 190, 405)));
  141. }
  142. public function testDocFreq()
  143. {
  144. $index = Zend_Search_Lucene::open(dirname(__FILE__) . '/_indexSample/_files');
  145. $this->assertEquals($index->docFreq(new Zend_Search_Lucene_Index_Term('packages', 'contents')), 5);
  146. }
  147. public function testGetSimilarity()
  148. {
  149. $index = Zend_Search_Lucene::open(dirname(__FILE__) . '/_indexSample/_files');
  150. $this->assertTrue($index->getSimilarity() instanceof Zend_Search_Lucene_Search_Similarity);
  151. }
  152. public function testNorm()
  153. {
  154. $index = Zend_Search_Lucene::open(dirname(__FILE__) . '/_indexSample/_files');
  155. $this->assertTrue(abs($index->norm(3, 'contents') - 0.054688) < 0.000001);
  156. }
  157. public function testHasDeletions()
  158. {
  159. $index = Zend_Search_Lucene::open(dirname(__FILE__) . '/_indexSample/_files');
  160. $this->assertTrue($index->hasDeletions());
  161. }
  162. public function testDelete()
  163. {
  164. // Copy index sample into _files directory
  165. $sampleIndexDir = dirname(__FILE__) . '/_indexSample/_files';
  166. $tempIndexDir = dirname(__FILE__) . '/_files';
  167. $this->_clearDirectory($tempIndexDir);
  168. $indexDir = opendir($sampleIndexDir);
  169. while (($file = readdir($indexDir)) !== false) {
  170. if (!is_dir($sampleIndexDir . '/' . $file)) {
  171. copy($sampleIndexDir . '/' . $file, $tempIndexDir . '/' . $file);
  172. }
  173. }
  174. closedir($indexDir);
  175. $index = Zend_Search_Lucene::open($tempIndexDir);
  176. $this->assertFalse($index->isDeleted(2));
  177. $index->delete(2);
  178. $this->assertTrue($index->isDeleted(2));
  179. $index->commit();
  180. unset($index);
  181. $index1 = Zend_Search_Lucene::open($tempIndexDir);
  182. $this->assertTrue($index1->isDeleted(2));
  183. unset($index1);
  184. $this->_clearDirectory($tempIndexDir);
  185. }
  186. public function testAddDocument()
  187. {
  188. $index = Zend_Search_Lucene::create(dirname(__FILE__) . '/_index/_files');
  189. $indexSourceDir = dirname(__FILE__) . '/_indexSource/_files';
  190. $dir = opendir($indexSourceDir);
  191. while (($file = readdir($dir)) !== false) {
  192. if (is_dir($indexSourceDir . '/' . $file)) {
  193. continue;
  194. }
  195. if (strcasecmp(substr($file, strlen($file)-5), '.html') != 0) {
  196. continue;
  197. }
  198. // Create new Document from a file
  199. $doc = new Zend_Search_Lucene_Document();
  200. $doc->addField(Zend_Search_Lucene_Field::Text('path', 'IndexSource/' . $file));
  201. $doc->addField(Zend_Search_Lucene_Field::Keyword( 'modified', filemtime($indexSourceDir . '/' . $file) ));
  202. $f = fopen($indexSourceDir . '/' . $file,'rb');
  203. $byteCount = filesize($indexSourceDir . '/' . $file);
  204. $data = '';
  205. while ( $byteCount > 0 && ($nextBlock = fread($f, $byteCount)) != false ) {
  206. $data .= $nextBlock;
  207. $byteCount -= strlen($nextBlock);
  208. }
  209. fclose($f);
  210. $doc->addField(Zend_Search_Lucene_Field::Text('contents', $data, 'ISO-8859-1'));
  211. // Add document to the index
  212. $index->addDocument($doc);
  213. }
  214. closedir($dir);
  215. unset($index);
  216. $index1 = Zend_Search_Lucene::open(dirname(__FILE__) . '/_index/_files');
  217. $this->assertTrue($index1 instanceof Zend_Search_Lucene_Interface);
  218. $this->_clearDirectory(dirname(__FILE__) . '/_index/_files');
  219. }
  220. public function testOptimize()
  221. {
  222. $index = Zend_Search_Lucene::create(dirname(__FILE__) . '/_index/_files');
  223. $index->setMaxBufferedDocs(2);
  224. $indexSourceDir = dirname(__FILE__) . '/_indexSource/_files';
  225. $dir = opendir($indexSourceDir);
  226. while (($file = readdir($dir)) !== false) {
  227. if (is_dir($indexSourceDir . '/' . $file)) {
  228. continue;
  229. }
  230. if (strcasecmp(substr($file, strlen($file)-5), '.html') != 0) {
  231. continue;
  232. }
  233. // Create new Document from a file
  234. $doc = new Zend_Search_Lucene_Document();
  235. $doc->addField(Zend_Search_Lucene_Field::Keyword('path', 'IndexSource/' . $file));
  236. $doc->addField(Zend_Search_Lucene_Field::Keyword( 'modified', filemtime($indexSourceDir . '/' . $file) ));
  237. $f = fopen($indexSourceDir . '/' . $file,'rb');
  238. $byteCount = filesize($indexSourceDir . '/' . $file);
  239. $data = '';
  240. while ( $byteCount > 0 && ($nextBlock = fread($f, $byteCount)) != false ) {
  241. $data .= $nextBlock;
  242. $byteCount -= strlen($nextBlock);
  243. }
  244. fclose($f);
  245. $doc->addField(Zend_Search_Lucene_Field::Text('contents', $data, 'ISO-8859-1'));
  246. // Add document to the index
  247. $index->addDocument($doc);
  248. }
  249. closedir($dir);
  250. unset($index);
  251. $index1 = Zend_Search_Lucene::open(dirname(__FILE__) . '/_index/_files');
  252. $this->assertTrue($index1 instanceof Zend_Search_Lucene_Interface);
  253. $pathTerm = new Zend_Search_Lucene_Index_Term('IndexSource/contributing.html', 'path');
  254. $contributingDocs = $index1->termDocs($pathTerm);
  255. foreach ($contributingDocs as $id) {
  256. $index1->delete($id);
  257. }
  258. $index1->optimize();
  259. unset($index1);
  260. $index2 = Zend_Search_Lucene::open(dirname(__FILE__) . '/_index/_files');
  261. $this->assertTrue($index2 instanceof Zend_Search_Lucene_Interface);
  262. $hits = $index2->find('submitting');
  263. $this->assertEquals(count($hits), 3);
  264. $this->_clearDirectory(dirname(__FILE__) . '/_index/_files');
  265. }
  266. public function testTerms()
  267. {
  268. $index = Zend_Search_Lucene::open(dirname(__FILE__) . '/_indexSample/_files');
  269. $this->assertEquals(count($index->terms()), 607);
  270. }
  271. public function testTermsStreamInterface()
  272. {
  273. $index = Zend_Search_Lucene::open(dirname(__FILE__) . '/_indexSample/_files');
  274. $terms = array();
  275. $index->resetTermsStream();
  276. while ($index->currentTerm() !== null) {
  277. $terms[] = $index->currentTerm();
  278. $index->nextTerm();
  279. }
  280. $this->assertEquals(count($terms), 607);
  281. }
  282. public function testTermsStreamInterfaceSkipTo()
  283. {
  284. $index = Zend_Search_Lucene::open(dirname(__FILE__) . '/_indexSample/_files');
  285. $terms = array();
  286. $index->resetTermsStream();
  287. $index->skipTo(new Zend_Search_Lucene_Index_Term('one', 'contents'));
  288. while ($index->currentTerm() !== null) {
  289. $terms[] = $index->currentTerm();
  290. $index->nextTerm();
  291. }
  292. $this->assertEquals(count($terms), 244);
  293. }
  294. public function testTermsStreamInterfaceSkipToTermsRetrieving()
  295. {
  296. $index = Zend_Search_Lucene::open(dirname(__FILE__) . '/_indexSample/_files');
  297. $terms = array();
  298. $index->resetTermsStream();
  299. $index->skipTo(new Zend_Search_Lucene_Index_Term('one', 'contents'));
  300. $terms[] = $index->currentTerm();
  301. $terms[] = $index->nextTerm();
  302. $terms[] = $index->nextTerm();
  303. $index->closeTermsStream();
  304. $this->assertTrue($terms ==
  305. array(new Zend_Search_Lucene_Index_Term('one', 'contents'),
  306. new Zend_Search_Lucene_Index_Term('only', 'contents'),
  307. new Zend_Search_Lucene_Index_Term('open', 'contents'),
  308. ));
  309. }
  310. public function testTermsStreamInterfaceSkipToTermsRetrievingZeroTermsCase()
  311. {
  312. $index = Zend_Search_Lucene::create(dirname(__FILE__) . '/_index/_files');
  313. // Zero terms
  314. $doc = new Zend_Search_Lucene_Document();
  315. $doc->addField(Zend_Search_Lucene_Field::Text('contents', ''));
  316. $index->addDocument($doc);
  317. unset($index);
  318. $index = Zend_Search_Lucene::open(dirname(__FILE__) . '/_index/_files');
  319. $index->resetTermsStream();
  320. $index->skipTo(new Zend_Search_Lucene_Index_Term('term', 'contents'));
  321. $this->assertTrue($index->currentTerm() === null);
  322. $index->closeTermsStream();
  323. $this->_clearDirectory(dirname(__FILE__) . '/_index/_files');
  324. }
  325. public function testTermsStreamInterfaceSkipToTermsRetrievingOneTermsCase()
  326. {
  327. $index = Zend_Search_Lucene::create(dirname(__FILE__) . '/_index/_files');
  328. // Zero terms
  329. $doc = new Zend_Search_Lucene_Document();
  330. $doc->addField(Zend_Search_Lucene_Field::Text('contents', 'someterm'));
  331. $index->addDocument($doc);
  332. unset($index);
  333. $index = Zend_Search_Lucene::open(dirname(__FILE__) . '/_index/_files');
  334. $index->resetTermsStream();
  335. $index->skipTo(new Zend_Search_Lucene_Index_Term('term', 'contents'));
  336. $this->assertTrue($index->currentTerm() === null);
  337. $index->closeTermsStream();
  338. $this->_clearDirectory(dirname(__FILE__) . '/_index/_files');
  339. }
  340. public function testTermsStreamInterfaceSkipToTermsRetrievingTwoTermsCase()
  341. {
  342. $index = Zend_Search_Lucene::create(dirname(__FILE__) . '/_index/_files');
  343. // Zero terms
  344. $doc = new Zend_Search_Lucene_Document();
  345. $doc->addField(Zend_Search_Lucene_Field::Text('contents', 'someterm word'));
  346. $index->addDocument($doc);
  347. unset($index);
  348. $index = Zend_Search_Lucene::open(dirname(__FILE__) . '/_index/_files');
  349. $index->resetTermsStream();
  350. $index->skipTo(new Zend_Search_Lucene_Index_Term('term', 'contents'));
  351. $this->assertTrue($index->currentTerm() == new Zend_Search_Lucene_Index_Term('word', 'contents'));
  352. $index->closeTermsStream();
  353. $this->_clearDirectory(dirname(__FILE__) . '/_index/_files');
  354. }
  355. }