SegmentInfoTest.php 33 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562
  1. <?php
  2. /**
  3. * Zend Framework
  4. *
  5. * LICENSE
  6. *
  7. * This source file is subject to the new BSD license that is bundled
  8. * with this package in the file LICENSE.txt.
  9. * It is also available through the world-wide-web at this URL:
  10. * http://framework.zend.com/license/new-bsd
  11. * If you did not receive a copy of the license and are unable to
  12. * obtain it through the world-wide-web, please send an email
  13. * to license@zend.com so we can send you a copy immediately.
  14. *
  15. * @category Zend
  16. * @package Zend_Search_Lucene
  17. * @subpackage UnitTests
  18. * @copyright Copyright (c) 2005-2015 Zend Technologies USA Inc. (http://www.zend.com)
  19. * @license http://framework.zend.com/license/new-bsd New BSD License
  20. * @version $Id$
  21. */
  22. /**
  23. * Zend_Search_Lucene_Index_SegmentInfo
  24. */
  25. require_once 'Zend/Search/Lucene/Index/SegmentInfo.php';
  26. /**
  27. * @category Zend
  28. * @package Zend_Search_Lucene
  29. * @subpackage UnitTests
  30. * @copyright Copyright (c) 2005-2015 Zend Technologies USA Inc. (http://www.zend.com)
  31. * @license http://framework.zend.com/license/new-bsd New BSD License
  32. * @group Zend_Search_Lucene
  33. */
  34. class Zend_Search_Lucene_Index_SegmentInfoTest extends PHPUnit_Framework_TestCase
  35. {
  36. public function testCreate()
  37. {
  38. $directory = new Zend_Search_Lucene_Storage_Directory_Filesystem(dirname(__FILE__) . '/_source/_files');
  39. $segmentInfo = new Zend_Search_Lucene_Index_SegmentInfo($directory, '_1', 2);
  40. $this->assertTrue($segmentInfo instanceof Zend_Search_Lucene_Index_SegmentInfo);
  41. }
  42. public function testOpenCompoundFile()
  43. {
  44. $directory = new Zend_Search_Lucene_Storage_Directory_Filesystem(dirname(__FILE__) . '/_source/_files');
  45. $segmentInfo = new Zend_Search_Lucene_Index_SegmentInfo($directory, '_1', 2);
  46. $file1 = $segmentInfo->openCompoundFile('.fnm');
  47. $this->assertTrue($file1 instanceof Zend_Search_Lucene_Storage_File);
  48. $file2 = $segmentInfo->openCompoundFile('.tii');
  49. $file3 = $segmentInfo->openCompoundFile('.tii');
  50. $file4 = $segmentInfo->openCompoundFile('.tii', false);
  51. $this->assertTrue($file2 instanceof Zend_Search_Lucene_Storage_File);
  52. $this->assertTrue($file2 === $file3);
  53. $this->assertTrue($file4 instanceof Zend_Search_Lucene_Storage_File);
  54. $this->assertTrue($file2 !== $file4);
  55. }
  56. public function testCompoundFileLength()
  57. {
  58. $directory = new Zend_Search_Lucene_Storage_Directory_Filesystem(dirname(__FILE__) . '/_source/_files');
  59. $segmentInfo = new Zend_Search_Lucene_Index_SegmentInfo($directory, '_1', 2);
  60. $this->assertEquals($segmentInfo->compoundFileLength('.tii'), 58);
  61. }
  62. public function testGetFieldNum()
  63. {
  64. $directory = new Zend_Search_Lucene_Storage_Directory_Filesystem(dirname(__FILE__) . '/_source/_files');
  65. $segmentInfo = new Zend_Search_Lucene_Index_SegmentInfo($directory, '_1', 2);
  66. $this->assertEquals($segmentInfo->getFieldNum('contents'), 2);
  67. $this->assertEquals($segmentInfo->getFieldNum('non-presented-field'), -1);
  68. }
  69. public function testGetField()
  70. {
  71. $directory = new Zend_Search_Lucene_Storage_Directory_Filesystem(dirname(__FILE__) . '/_source/_files');
  72. $segmentInfo = new Zend_Search_Lucene_Index_SegmentInfo($directory, '_1', 2);
  73. $fieldInfo = $segmentInfo->getField(2);
  74. $this->assertEquals($fieldInfo->name, 'contents');
  75. $this->assertTrue((boolean)$fieldInfo->isIndexed);
  76. $this->assertEquals($fieldInfo->number, 2);
  77. $this->assertFalse((boolean)$fieldInfo->storeTermVector);
  78. }
  79. public function testGetFields()
  80. {
  81. $directory = new Zend_Search_Lucene_Storage_Directory_Filesystem(dirname(__FILE__) . '/_source/_files');
  82. $segmentInfo = new Zend_Search_Lucene_Index_SegmentInfo($directory, '_1', 2);
  83. $this->assertTrue($segmentInfo->getFields() == array('path' => 'path', 'modified' => 'modified', 'contents' => 'contents'));
  84. $this->assertTrue($segmentInfo->getFields(true) == array('path' => 'path', 'modified' => 'modified', 'contents' => 'contents'));
  85. }
  86. public function testGetFieldInfos()
  87. {
  88. $directory = new Zend_Search_Lucene_Storage_Directory_Filesystem(dirname(__FILE__) . '/_source/_files');
  89. $segmentInfo = new Zend_Search_Lucene_Index_SegmentInfo($directory, '_1', 2);
  90. $fieldInfos = $segmentInfo->getFieldInfos();
  91. $this->assertEquals($fieldInfos[0]->name, 'path');
  92. $this->assertTrue((boolean)$fieldInfos[0]->isIndexed);
  93. $this->assertEquals($fieldInfos[0]->number, 0);
  94. $this->assertFalse((boolean)$fieldInfos[0]->storeTermVector);
  95. $this->assertEquals($fieldInfos[1]->name, 'modified');
  96. $this->assertTrue((boolean)$fieldInfos[1]->isIndexed);
  97. $this->assertEquals($fieldInfos[1]->number, 1);
  98. $this->assertFalse((boolean)$fieldInfos[1]->storeTermVector);
  99. $this->assertEquals($fieldInfos[2]->name, 'contents');
  100. $this->assertTrue((boolean)$fieldInfos[2]->isIndexed);
  101. $this->assertEquals($fieldInfos[2]->number, 2);
  102. $this->assertFalse((boolean)$fieldInfos[2]->storeTermVector);
  103. }
  104. public function testCount()
  105. {
  106. $directory = new Zend_Search_Lucene_Storage_Directory_Filesystem(dirname(__FILE__) . '/_source/_files');
  107. $segmentInfo = new Zend_Search_Lucene_Index_SegmentInfo($directory, '_1', 2);
  108. $this->assertEquals($segmentInfo->count(), 2);
  109. }
  110. public function testNumDocs()
  111. {
  112. $directory = new Zend_Search_Lucene_Storage_Directory_Filesystem(dirname(__FILE__) . '/_source/_files');
  113. $segmentInfo = new Zend_Search_Lucene_Index_SegmentInfo($directory, '_3', 2);
  114. $this->assertEquals($segmentInfo->count(), 2);
  115. $this->assertEquals($segmentInfo->numDocs(), 1);
  116. }
  117. public function testGetName()
  118. {
  119. $directory = new Zend_Search_Lucene_Storage_Directory_Filesystem(dirname(__FILE__) . '/_source/_files');
  120. $segmentInfo = new Zend_Search_Lucene_Index_SegmentInfo($directory, '_1', 2);
  121. $this->assertEquals($segmentInfo->getName(), '_1');
  122. }
  123. public function testGetTermInfo()
  124. {
  125. $directory = new Zend_Search_Lucene_Storage_Directory_Filesystem(dirname(__FILE__) . '/_source/_files');
  126. $segmentInfo = new Zend_Search_Lucene_Index_SegmentInfo($directory, '_1', 2);
  127. $termInfo = $segmentInfo->getTermInfo(new Zend_Search_Lucene_Index_Term('apart', 'contents'));
  128. $this->assertEquals($termInfo->docFreq, 1);
  129. $this->assertEquals($termInfo->freqPointer, 29);
  130. $this->assertEquals($termInfo->proxPointer, 119);
  131. $this->assertEquals($termInfo->skipOffset, 0);
  132. $this->assertEquals($termInfo->indexPointer, null);
  133. $termInfo1 = $segmentInfo->getTermInfo(new Zend_Search_Lucene_Index_Term('apart', 'contents'));
  134. // test for requesting cached information
  135. $this->assertTrue($termInfo === $termInfo1);
  136. // request for non-existing term
  137. $this->assertTrue($segmentInfo->getTermInfo(new Zend_Search_Lucene_Index_Term('nonusedterm', 'contents')) === null);
  138. }
  139. public function testTermFreqs()
  140. {
  141. $directory = new Zend_Search_Lucene_Storage_Directory_Filesystem(dirname(__FILE__) . '/_source/_files');
  142. $segmentInfo = new Zend_Search_Lucene_Index_SegmentInfo($directory, '_1', 2);
  143. $termPositions = $segmentInfo->termFreqs(new Zend_Search_Lucene_Index_Term('bgcolor', 'contents'));
  144. $this->assertTrue($termPositions == array(0 => 3, 1 => 1));
  145. $termPositions = $segmentInfo->termFreqs(new Zend_Search_Lucene_Index_Term('bgcolor', 'contents'), 10);
  146. $this->assertTrue($termPositions == array(10 => 3, 11 => 1));
  147. }
  148. public function testTermPositions()
  149. {
  150. $directory = new Zend_Search_Lucene_Storage_Directory_Filesystem(dirname(__FILE__) . '/_source/_files');
  151. $segmentInfo = new Zend_Search_Lucene_Index_SegmentInfo($directory, '_1', 2);
  152. $termPositions = $segmentInfo->termPositions(new Zend_Search_Lucene_Index_Term('bgcolor', 'contents'));
  153. $this->assertTrue($termPositions == array(0 => array(69, 239, 370),
  154. 1 => array(58)
  155. ));
  156. $termPositions = $segmentInfo->termPositions(new Zend_Search_Lucene_Index_Term('bgcolor', 'contents'), 10);
  157. $this->assertTrue($termPositions == array(10 => array(69, 239, 370),
  158. 11 => array(58)
  159. ));
  160. }
  161. public function testNorm()
  162. {
  163. $directory = new Zend_Search_Lucene_Storage_Directory_Filesystem(dirname(__FILE__) . '/_source/_files');
  164. $segmentInfo = new Zend_Search_Lucene_Index_SegmentInfo($directory, '_1', 2);
  165. $this->assertTrue(abs($segmentInfo->norm(1, 'contents') - 0.0546875) < 0.000001);
  166. }
  167. public function testNormVector()
  168. {
  169. $directory = new Zend_Search_Lucene_Storage_Directory_Filesystem(dirname(__FILE__) . '/_source/_files');
  170. $segmentInfo = new Zend_Search_Lucene_Index_SegmentInfo($directory, '_1', 2);
  171. $this->assertEquals($segmentInfo->normVector('contents'), "\x69\x6B");
  172. }
  173. public function testHasDeletions()
  174. {
  175. $directory = new Zend_Search_Lucene_Storage_Directory_Filesystem(dirname(__FILE__) . '/_source/_files');
  176. $segmentInfo = new Zend_Search_Lucene_Index_SegmentInfo($directory, '_1', 2);
  177. $this->assertFalse($segmentInfo->hasDeletions());
  178. $segmentInfo1 = new Zend_Search_Lucene_Index_SegmentInfo($directory, '_3', 2);
  179. $this->assertTrue($segmentInfo1->hasDeletions());
  180. }
  181. public function testDelete()
  182. {
  183. $directory = new Zend_Search_Lucene_Storage_Directory_Filesystem(dirname(__FILE__) . '/_source/_files');
  184. $segmentInfo = new Zend_Search_Lucene_Index_SegmentInfo($directory, '_1', 2, 0 /* search for _1.del file */);
  185. $this->assertFalse($segmentInfo->hasDeletions());
  186. $segmentInfo->delete(0);
  187. $this->assertTrue($segmentInfo->hasDeletions());
  188. $delGen = $segmentInfo->getDelGen();
  189. // don't write changes
  190. unset($segmentInfo);
  191. $segmentInfo1 = new Zend_Search_Lucene_Index_SegmentInfo($directory, '_1', 2, $delGen);
  192. // Changes wasn't written, segment still has no deletions
  193. $this->assertFalse($segmentInfo1->hasDeletions());
  194. $segmentInfo1->delete(0);
  195. $segmentInfo1->writeChanges();
  196. $delGen = $segmentInfo1->getDelGen();
  197. unset($segmentInfo1);
  198. $segmentInfo2 = new Zend_Search_Lucene_Index_SegmentInfo($directory, '_1', 2, $delGen);
  199. $this->assertTrue($segmentInfo2->hasDeletions());
  200. unset($segmentInfo2);
  201. $directory->deleteFile('_1_' . base_convert($delGen, 10, 36) . '.del');
  202. $segmentInfo3 = new Zend_Search_Lucene_Index_SegmentInfo($directory, '_1', 2, -1 /* no detetions file */);
  203. $this->assertFalse($segmentInfo3->hasDeletions());
  204. }
  205. public function testIsDeleted()
  206. {
  207. $directory = new Zend_Search_Lucene_Storage_Directory_Filesystem(dirname(__FILE__) . '/_source/_files');
  208. $segmentInfo = new Zend_Search_Lucene_Index_SegmentInfo($directory, '_2', 2);
  209. $this->assertFalse($segmentInfo->isDeleted(0));
  210. $segmentInfo1 = new Zend_Search_Lucene_Index_SegmentInfo($directory, '_3', 2);
  211. $this->assertTrue($segmentInfo1->isDeleted(0));
  212. $this->assertFalse($segmentInfo1->isDeleted(1));
  213. }
  214. public function testTermStreamStyleReading()
  215. {
  216. $directory = new Zend_Search_Lucene_Storage_Directory_Filesystem(dirname(__FILE__) . '/_source/_files');
  217. $segmentInfo = new Zend_Search_Lucene_Index_SegmentInfo($directory, '_3', 2);
  218. $this->assertEquals($segmentInfo->resetTermsStream(6, Zend_Search_Lucene_Index_SegmentInfo::SM_FULL_INFO), 8);
  219. $terms = array();
  220. $terms[] = $segmentInfo->currentTerm();
  221. $firstTermPositions = $segmentInfo->currentTermPositions();
  222. $this->assertEquals(count($firstTermPositions), 1);
  223. reset($firstTermPositions); // go to the first element
  224. $this->assertEquals(key($firstTermPositions), 7);
  225. $this->assertTrue(current($firstTermPositions) ==
  226. array(105, 113, 130, 138, 153, 168, 171, 216, 243, 253, 258, 265, 302, 324,
  227. 331, 351, 359, 366, 370, 376, 402, 410, 418, 425, 433, 441, 460, 467));
  228. while (($term = $segmentInfo->nextTerm()) != null) {
  229. $terms[] = $term;
  230. }
  231. $this->assertTrue($terms ==
  232. array(new Zend_Search_Lucene_Index_Term('a', 'contents'),
  233. new Zend_Search_Lucene_Index_Term('about', 'contents'),
  234. new Zend_Search_Lucene_Index_Term('accesskey', 'contents'),
  235. new Zend_Search_Lucene_Index_Term('align', 'contents'),
  236. new Zend_Search_Lucene_Index_Term('alink', 'contents'),
  237. new Zend_Search_Lucene_Index_Term('already', 'contents'),
  238. new Zend_Search_Lucene_Index_Term('and', 'contents'),
  239. new Zend_Search_Lucene_Index_Term('are', 'contents'),
  240. new Zend_Search_Lucene_Index_Term('at', 'contents'),
  241. new Zend_Search_Lucene_Index_Term('b', 'contents'),
  242. new Zend_Search_Lucene_Index_Term('be', 'contents'),
  243. new Zend_Search_Lucene_Index_Term('been', 'contents'),
  244. new Zend_Search_Lucene_Index_Term('bgcolor', 'contents'),
  245. new Zend_Search_Lucene_Index_Term('body', 'contents'),
  246. new Zend_Search_Lucene_Index_Term('border', 'contents'),
  247. new Zend_Search_Lucene_Index_Term('bottom', 'contents'),
  248. new Zend_Search_Lucene_Index_Term('bug', 'contents'),
  249. new Zend_Search_Lucene_Index_Term('bugs', 'contents'),
  250. new Zend_Search_Lucene_Index_Term('can', 'contents'),
  251. new Zend_Search_Lucene_Index_Term('care', 'contents'),
  252. new Zend_Search_Lucene_Index_Term('cellpadding', 'contents'),
  253. new Zend_Search_Lucene_Index_Term('cellspacing', 'contents'),
  254. new Zend_Search_Lucene_Index_Term('center', 'contents'),
  255. new Zend_Search_Lucene_Index_Term('chapter', 'contents'),
  256. new Zend_Search_Lucene_Index_Term('charset', 'contents'),
  257. new Zend_Search_Lucene_Index_Term('check', 'contents'),
  258. new Zend_Search_Lucene_Index_Term('class', 'contents'),
  259. new Zend_Search_Lucene_Index_Term('click', 'contents'),
  260. new Zend_Search_Lucene_Index_Term('colspan', 'contents'),
  261. new Zend_Search_Lucene_Index_Term('contains', 'contents'),
  262. new Zend_Search_Lucene_Index_Term('content', 'contents'),
  263. new Zend_Search_Lucene_Index_Term('contributing', 'contents'),
  264. new Zend_Search_Lucene_Index_Term('developers', 'contents'),
  265. new Zend_Search_Lucene_Index_Term('div', 'contents'),
  266. new Zend_Search_Lucene_Index_Term('docbook', 'contents'),
  267. new Zend_Search_Lucene_Index_Term('documentation', 'contents'),
  268. new Zend_Search_Lucene_Index_Term('does', 'contents'),
  269. new Zend_Search_Lucene_Index_Term('don', 'contents'),
  270. new Zend_Search_Lucene_Index_Term('double', 'contents'),
  271. new Zend_Search_Lucene_Index_Term('easiest', 'contents'),
  272. new Zend_Search_Lucene_Index_Term('equiv', 'contents'),
  273. new Zend_Search_Lucene_Index_Term('existing', 'contents'),
  274. new Zend_Search_Lucene_Index_Term('explanations', 'contents'),
  275. new Zend_Search_Lucene_Index_Term('ff', 'contents'),
  276. new Zend_Search_Lucene_Index_Term('ffffff', 'contents'),
  277. new Zend_Search_Lucene_Index_Term('fill', 'contents'),
  278. new Zend_Search_Lucene_Index_Term('find', 'contents'),
  279. new Zend_Search_Lucene_Index_Term('fixed', 'contents'),
  280. new Zend_Search_Lucene_Index_Term('footer', 'contents'),
  281. new Zend_Search_Lucene_Index_Term('for', 'contents'),
  282. new Zend_Search_Lucene_Index_Term('form', 'contents'),
  283. new Zend_Search_Lucene_Index_Term('found', 'contents'),
  284. new Zend_Search_Lucene_Index_Term('generator', 'contents'),
  285. new Zend_Search_Lucene_Index_Term('guide', 'contents'),
  286. new Zend_Search_Lucene_Index_Term('h', 'contents'),
  287. new Zend_Search_Lucene_Index_Term('hasn', 'contents'),
  288. new Zend_Search_Lucene_Index_Term('have', 'contents'),
  289. new Zend_Search_Lucene_Index_Term('head', 'contents'),
  290. new Zend_Search_Lucene_Index_Term('header', 'contents'),
  291. new Zend_Search_Lucene_Index_Term('hesitate', 'contents'),
  292. new Zend_Search_Lucene_Index_Term('home', 'contents'),
  293. new Zend_Search_Lucene_Index_Term('homepage', 'contents'),
  294. new Zend_Search_Lucene_Index_Term('how', 'contents'),
  295. new Zend_Search_Lucene_Index_Term('hr', 'contents'),
  296. new Zend_Search_Lucene_Index_Term('href', 'contents'),
  297. new Zend_Search_Lucene_Index_Term('html', 'contents'),
  298. new Zend_Search_Lucene_Index_Term('http', 'contents'),
  299. new Zend_Search_Lucene_Index_Term('if', 'contents'),
  300. new Zend_Search_Lucene_Index_Term('in', 'contents'),
  301. new Zend_Search_Lucene_Index_Term('index', 'contents'),
  302. new Zend_Search_Lucene_Index_Term('information', 'contents'),
  303. new Zend_Search_Lucene_Index_Term('is', 'contents'),
  304. new Zend_Search_Lucene_Index_Term('iso', 'contents'),
  305. new Zend_Search_Lucene_Index_Term('it', 'contents'),
  306. new Zend_Search_Lucene_Index_Term('latest', 'contents'),
  307. new Zend_Search_Lucene_Index_Term('left', 'contents'),
  308. new Zend_Search_Lucene_Index_Term('link', 'contents'),
  309. new Zend_Search_Lucene_Index_Term('list', 'contents'),
  310. new Zend_Search_Lucene_Index_Term('manual', 'contents'),
  311. new Zend_Search_Lucene_Index_Term('meet', 'contents'),
  312. new Zend_Search_Lucene_Index_Term('meta', 'contents'),
  313. new Zend_Search_Lucene_Index_Term('modular', 'contents'),
  314. new Zend_Search_Lucene_Index_Term('more', 'contents'),
  315. new Zend_Search_Lucene_Index_Term('n', 'contents'),
  316. new Zend_Search_Lucene_Index_Term('name', 'contents'),
  317. new Zend_Search_Lucene_Index_Term('navfooter', 'contents'),
  318. new Zend_Search_Lucene_Index_Term('navheader', 'contents'),
  319. new Zend_Search_Lucene_Index_Term('navigation', 'contents'),
  320. new Zend_Search_Lucene_Index_Term('net', 'contents'),
  321. new Zend_Search_Lucene_Index_Term('new', 'contents'),
  322. new Zend_Search_Lucene_Index_Term('newpackage', 'contents'),
  323. new Zend_Search_Lucene_Index_Term('next', 'contents'),
  324. new Zend_Search_Lucene_Index_Term('of', 'contents'),
  325. new Zend_Search_Lucene_Index_Term('on', 'contents'),
  326. new Zend_Search_Lucene_Index_Term('out', 'contents'),
  327. new Zend_Search_Lucene_Index_Term('p', 'contents'),
  328. new Zend_Search_Lucene_Index_Term('package', 'contents'),
  329. new Zend_Search_Lucene_Index_Term('packages', 'contents'),
  330. new Zend_Search_Lucene_Index_Term('page', 'contents'),
  331. new Zend_Search_Lucene_Index_Term('patches', 'contents'),
  332. new Zend_Search_Lucene_Index_Term('pear', 'contents'),
  333. new Zend_Search_Lucene_Index_Term('persists', 'contents'),
  334. new Zend_Search_Lucene_Index_Term('php', 'contents'),
  335. new Zend_Search_Lucene_Index_Term('please', 'contents'),
  336. new Zend_Search_Lucene_Index_Term('prev', 'contents'),
  337. new Zend_Search_Lucene_Index_Term('previous', 'contents'),
  338. new Zend_Search_Lucene_Index_Term('proper', 'contents'),
  339. new Zend_Search_Lucene_Index_Term('quote', 'contents'),
  340. new Zend_Search_Lucene_Index_Term('read', 'contents'),
  341. new Zend_Search_Lucene_Index_Term('rel', 'contents'),
  342. new Zend_Search_Lucene_Index_Term('report', 'contents'),
  343. new Zend_Search_Lucene_Index_Term('reported', 'contents'),
  344. new Zend_Search_Lucene_Index_Term('reporting', 'contents'),
  345. new Zend_Search_Lucene_Index_Term('requirements', 'contents'),
  346. new Zend_Search_Lucene_Index_Term('right', 'contents'),
  347. new Zend_Search_Lucene_Index_Term('sect', 'contents'),
  348. new Zend_Search_Lucene_Index_Term('span', 'contents'),
  349. new Zend_Search_Lucene_Index_Term('still', 'contents'),
  350. new Zend_Search_Lucene_Index_Term('stylesheet', 'contents'),
  351. new Zend_Search_Lucene_Index_Term('submitting', 'contents'),
  352. new Zend_Search_Lucene_Index_Term('summary', 'contents'),
  353. new Zend_Search_Lucene_Index_Term('system', 'contents'),
  354. new Zend_Search_Lucene_Index_Term('t', 'contents'),
  355. new Zend_Search_Lucene_Index_Term('table', 'contents'),
  356. new Zend_Search_Lucene_Index_Term('take', 'contents'),
  357. new Zend_Search_Lucene_Index_Term('target', 'contents'),
  358. new Zend_Search_Lucene_Index_Term('td', 'contents'),
  359. new Zend_Search_Lucene_Index_Term('text', 'contents'),
  360. new Zend_Search_Lucene_Index_Term('th', 'contents'),
  361. new Zend_Search_Lucene_Index_Term('that', 'contents'),
  362. new Zend_Search_Lucene_Index_Term('the', 'contents'),
  363. new Zend_Search_Lucene_Index_Term('think', 'contents'),
  364. new Zend_Search_Lucene_Index_Term('this', 'contents'),
  365. new Zend_Search_Lucene_Index_Term('tips', 'contents'),
  366. new Zend_Search_Lucene_Index_Term('title', 'contents'),
  367. new Zend_Search_Lucene_Index_Term('to', 'contents'),
  368. new Zend_Search_Lucene_Index_Term('top', 'contents'),
  369. new Zend_Search_Lucene_Index_Term('tr', 'contents'),
  370. new Zend_Search_Lucene_Index_Term('translating', 'contents'),
  371. new Zend_Search_Lucene_Index_Term('type', 'contents'),
  372. new Zend_Search_Lucene_Index_Term('u', 'contents'),
  373. new Zend_Search_Lucene_Index_Term('unable', 'contents'),
  374. new Zend_Search_Lucene_Index_Term('up', 'contents'),
  375. new Zend_Search_Lucene_Index_Term('using', 'contents'),
  376. new Zend_Search_Lucene_Index_Term('valign', 'contents'),
  377. new Zend_Search_Lucene_Index_Term('version', 'contents'),
  378. new Zend_Search_Lucene_Index_Term('vlink', 'contents'),
  379. new Zend_Search_Lucene_Index_Term('way', 'contents'),
  380. new Zend_Search_Lucene_Index_Term('which', 'contents'),
  381. new Zend_Search_Lucene_Index_Term('width', 'contents'),
  382. new Zend_Search_Lucene_Index_Term('will', 'contents'),
  383. new Zend_Search_Lucene_Index_Term('with', 'contents'),
  384. new Zend_Search_Lucene_Index_Term('writing', 'contents'),
  385. new Zend_Search_Lucene_Index_Term('you', 'contents'),
  386. new Zend_Search_Lucene_Index_Term('your', 'contents'),
  387. new Zend_Search_Lucene_Index_Term('1178009946', 'modified'),
  388. new Zend_Search_Lucene_Index_Term('bugs', 'path'),
  389. new Zend_Search_Lucene_Index_Term('contributing', 'path'),
  390. new Zend_Search_Lucene_Index_Term('html', 'path'),
  391. new Zend_Search_Lucene_Index_Term('indexsource', 'path'),
  392. new Zend_Search_Lucene_Index_Term('newpackage', 'path'),
  393. ));
  394. unset($segmentInfo);
  395. $segmentInfo1 = new Zend_Search_Lucene_Index_SegmentInfo($directory, '_3', 2);
  396. $this->assertEquals($segmentInfo1->resetTermsStream(6, Zend_Search_Lucene_Index_SegmentInfo::SM_MERGE_INFO), 7);
  397. }
  398. public function testTermStreamStyleReadingSkipTo()
  399. {
  400. $directory = new Zend_Search_Lucene_Storage_Directory_Filesystem(dirname(__FILE__) . '/_source/_files');
  401. $segmentInfo = new Zend_Search_Lucene_Index_SegmentInfo($directory, '_3', 2);
  402. $this->assertEquals($segmentInfo->resetTermsStream(6, Zend_Search_Lucene_Index_SegmentInfo::SM_FULL_INFO), 8);
  403. $segmentInfo->skipTo(new Zend_Search_Lucene_Index_Term('prefetch', 'contents'));
  404. $terms = array();
  405. $terms[] = $segmentInfo->currentTerm();
  406. $firstTermPositions = $segmentInfo->currentTermPositions();
  407. $this->assertEquals(count($firstTermPositions), 1);
  408. reset($firstTermPositions); // go to the first element
  409. $this->assertEquals(key($firstTermPositions), 7);
  410. $this->assertTrue(current($firstTermPositions) == array(112, 409));
  411. while (($term = $segmentInfo->nextTerm()) != null) {
  412. $terms[] = $term;
  413. }
  414. $this->assertTrue($terms ==
  415. array(new Zend_Search_Lucene_Index_Term('prev', 'contents'),
  416. new Zend_Search_Lucene_Index_Term('previous', 'contents'),
  417. new Zend_Search_Lucene_Index_Term('proper', 'contents'),
  418. new Zend_Search_Lucene_Index_Term('quote', 'contents'),
  419. new Zend_Search_Lucene_Index_Term('read', 'contents'),
  420. new Zend_Search_Lucene_Index_Term('rel', 'contents'),
  421. new Zend_Search_Lucene_Index_Term('report', 'contents'),
  422. new Zend_Search_Lucene_Index_Term('reported', 'contents'),
  423. new Zend_Search_Lucene_Index_Term('reporting', 'contents'),
  424. new Zend_Search_Lucene_Index_Term('requirements', 'contents'),
  425. new Zend_Search_Lucene_Index_Term('right', 'contents'),
  426. new Zend_Search_Lucene_Index_Term('sect', 'contents'),
  427. new Zend_Search_Lucene_Index_Term('span', 'contents'),
  428. new Zend_Search_Lucene_Index_Term('still', 'contents'),
  429. new Zend_Search_Lucene_Index_Term('stylesheet', 'contents'),
  430. new Zend_Search_Lucene_Index_Term('submitting', 'contents'),
  431. new Zend_Search_Lucene_Index_Term('summary', 'contents'),
  432. new Zend_Search_Lucene_Index_Term('system', 'contents'),
  433. new Zend_Search_Lucene_Index_Term('t', 'contents'),
  434. new Zend_Search_Lucene_Index_Term('table', 'contents'),
  435. new Zend_Search_Lucene_Index_Term('take', 'contents'),
  436. new Zend_Search_Lucene_Index_Term('target', 'contents'),
  437. new Zend_Search_Lucene_Index_Term('td', 'contents'),
  438. new Zend_Search_Lucene_Index_Term('text', 'contents'),
  439. new Zend_Search_Lucene_Index_Term('th', 'contents'),
  440. new Zend_Search_Lucene_Index_Term('that', 'contents'),
  441. new Zend_Search_Lucene_Index_Term('the', 'contents'),
  442. new Zend_Search_Lucene_Index_Term('think', 'contents'),
  443. new Zend_Search_Lucene_Index_Term('this', 'contents'),
  444. new Zend_Search_Lucene_Index_Term('tips', 'contents'),
  445. new Zend_Search_Lucene_Index_Term('title', 'contents'),
  446. new Zend_Search_Lucene_Index_Term('to', 'contents'),
  447. new Zend_Search_Lucene_Index_Term('top', 'contents'),
  448. new Zend_Search_Lucene_Index_Term('tr', 'contents'),
  449. new Zend_Search_Lucene_Index_Term('translating', 'contents'),
  450. new Zend_Search_Lucene_Index_Term('type', 'contents'),
  451. new Zend_Search_Lucene_Index_Term('u', 'contents'),
  452. new Zend_Search_Lucene_Index_Term('unable', 'contents'),
  453. new Zend_Search_Lucene_Index_Term('up', 'contents'),
  454. new Zend_Search_Lucene_Index_Term('using', 'contents'),
  455. new Zend_Search_Lucene_Index_Term('valign', 'contents'),
  456. new Zend_Search_Lucene_Index_Term('version', 'contents'),
  457. new Zend_Search_Lucene_Index_Term('vlink', 'contents'),
  458. new Zend_Search_Lucene_Index_Term('way', 'contents'),
  459. new Zend_Search_Lucene_Index_Term('which', 'contents'),
  460. new Zend_Search_Lucene_Index_Term('width', 'contents'),
  461. new Zend_Search_Lucene_Index_Term('will', 'contents'),
  462. new Zend_Search_Lucene_Index_Term('with', 'contents'),
  463. new Zend_Search_Lucene_Index_Term('writing', 'contents'),
  464. new Zend_Search_Lucene_Index_Term('you', 'contents'),
  465. new Zend_Search_Lucene_Index_Term('your', 'contents'),
  466. new Zend_Search_Lucene_Index_Term('1178009946', 'modified'),
  467. new Zend_Search_Lucene_Index_Term('bugs', 'path'),
  468. new Zend_Search_Lucene_Index_Term('contributing', 'path'),
  469. new Zend_Search_Lucene_Index_Term('html', 'path'),
  470. new Zend_Search_Lucene_Index_Term('indexsource', 'path'),
  471. new Zend_Search_Lucene_Index_Term('newpackage', 'path'),
  472. ));
  473. unset($segmentInfo);
  474. $segmentInfo1 = new Zend_Search_Lucene_Index_SegmentInfo($directory, '_3', 2);
  475. $this->assertEquals($segmentInfo1->resetTermsStream(6, Zend_Search_Lucene_Index_SegmentInfo::SM_MERGE_INFO), 7);
  476. }
  477. }