1 <?php

2 namespace TYPO3\CMS\IndexedSearch\Controller;

3 

17 use TYPO3\CMS\Backend\Utility\BackendUtility;

18 use TYPO3\CMS\Core\Utility\GeneralUtility;

19 

25 class IndexedPagesController extends \TYPO3\CMS\Backend\Module\AbstractFunctionModule {

26 

27  // Internal, dynamic:

31  public $allPhashListed = array();

32 

33  // phash values accumulations for link to clear all

37  public $external_parsers = array();

38 

39  // External content parsers - objects set here with file extensions as keys.

43  public $iconFileNameCache = array();

44 

45  // File extensions - icon map/cache.

49  public $indexerConfig = array();

50 

51  // Indexer configuration, coming from $GLOBALS['TYPO3_CONF_VARS']['EXT']['extConf']['indexed_search']

55  public $enableMetaphoneSearch = FALSE;

56 

63  public $indexerObj;

64 

68  protected $maxListPerPage;

69 

73  public function __construct() {

74  $GLOBALS['LANG']->includeLLFile('EXT:indexed_search/modfunc1/locallang.xlf');

75  }

76 

83  public function modMenu() {

84  global $LANG;

85  return array(

86  'depth' => array(

87  0 => $LANG->sL('LLL:EXT:lang/locallang_core.xlf:labels.depth_0'),

88  1 => $LANG->sL('LLL:EXT:lang/locallang_core.xlf:labels.depth_1'),

89  2 => $LANG->sL('LLL:EXT:lang/locallang_core.xlf:labels.depth_2'),

90  3 => $LANG->sL('LLL:EXT:lang/locallang_core.xlf:labels.depth_3'),

91  999 => $LANG->sL('LLL:EXT:lang/locallang_core.xlf:labels.depth_infi')

92  ),

93  'type' => array(

94  0 => 'Overview',

95  1 => 'Technical Details',

96  2 => 'Words and content'

97  )

98  );

99  }

100 

107  public function main() {

108  // Initializes the module. Done in this function because we may need to re-initialize if data is submitted!

109  global $LANG, $TYPO3_CONF_VARS;

110  // Return if no page id:

111  if ($this->pObj->id <= 0) {

112  return;

113  }

114  // Indexer configuration from Extension Manager interface:

115  $this->indexerConfig = unserialize($GLOBALS['TYPO3_CONF_VARS']['EXT']['extConf']['indexed_search']);

116  // Workaround: If the extension configuration was not updated yet, the value is not existing

117  $this->enableMetaphoneSearch = isset($this->indexerConfig['enableMetaphoneSearch']) ? ($this->indexerConfig['enableMetaphoneSearch'] ? TRUE : FALSE) : TRUE;

118  // Initialize max-list items

119  $this->maxListPerPage = GeneralUtility::_GP('listAll') ? 100000 : 100;

120  // Processing deletion of phash rows:

121  if (GeneralUtility::_GP('deletePhash')) {

122  $this->removeIndexedPhashRow(GeneralUtility::_GP('deletePhash'));

123  }

124  // Processing stop-words:

125  if (GeneralUtility::_POST('_stopwords')) {

126  $this->processStopWords(GeneralUtility::_POST('stopWord'));

127  }

128  // Processing stop-words:

129  if (GeneralUtility::_POST('_pageKeywords')) {

130  $this->processPageKeywords(GeneralUtility::_POST('pageKeyword'), GeneralUtility::_POST('pageKeyword_pageUid'));

131  }

132  // Initialize external document parsers:

133  // Example configuration, see ext_localconf.php of this file!

134  if (is_array($TYPO3_CONF_VARS['EXTCONF']['indexed_search']['external_parsers'])) {

135  foreach ($TYPO3_CONF_VARS['EXTCONF']['indexed_search']['external_parsers'] as $extension => $_objRef) {

136  $this->external_parsers[$extension] = GeneralUtility::getUserObj($_objRef);

137  // Init parser and if it returns FALSE, unset its entry again:

138  if (!$this->external_parsers[$extension]->softInit($extension)) {

139  unset($this->external_parsers[$extension]);

140  }

141  }

142  }

143  // Initialize indexer if we need it (metaphone display does...)

144  $this->indexerObj = GeneralUtility::makeInstance('TYPO3\\CMS\\IndexedSearch\\Indexer');

145  // Set CSS styles specific for this document:

146  $this->pObj->content = str_replace('/*###POSTCSSMARKER###*/', '

147  TABLE.c-list TR TD { white-space: nowrap; vertical-align: top; }

148  ', $this->pObj->content);

149  // Check if details for a phash record should be shown:

150  if (GeneralUtility::_GET('phash')) {

151  // Show title / function menu:

152  $theOutput .= $this->pObj->doc->spacer(5);

153  $theOutput .= $this->pObj->doc->section('Details for a single result row:', $this->showDetailsForPhash(GeneralUtility::_GET('phash')), 0, 1);

154  } elseif (GeneralUtility::_GET('wid')) {

155  // Show title / function menu:

156  $theOutput .= $this->pObj->doc->spacer(5);

157  $theOutput .= $this->pObj->doc->section('Details for a word:', $this->showDetailsForWord(GeneralUtility::_GET('wid')), 0, 1);

158  } elseif ($this->enableMetaphoneSearch && GeneralUtility::_GET('metaphone')) {

159  // Show title / function menu:

160  $theOutput .= $this->pObj->doc->spacer(5);

161  $theOutput .= $this->pObj->doc->section('Details for metaphone value:', $this->showDetailsForMetaphone(GeneralUtility::_GET('metaphone')), 0, 1);

162  } elseif (GeneralUtility::_GET('reindex')) {

163  // Show title / function menu:

164  $theOutput .= $this->pObj->doc->spacer(5);

165  $theOutput .= $this->pObj->doc->section('Reindexing...', $this->reindexPhash(GeneralUtility::_GET('reindex'), GeneralUtility::_GET('reindex_id')), 0, 1);

166  } else {

167  // Detail listings:

168  // Depth function menu:

169  $h_func = BackendUtility::getFuncMenu($this->pObj->id, 'SET[type]', $this->pObj->MOD_SETTINGS['type'], $this->pObj->MOD_MENU['type']);

170  $h_func .= BackendUtility::getFuncMenu($this->pObj->id, 'SET[depth]', $this->pObj->MOD_SETTINGS['depth'], $this->pObj->MOD_MENU['depth']);

171  // Show title / function menu:

172  $theOutput .= $this->pObj->doc->header($LANG->getLL('title'));

173  $theOutput .= $this->pObj->doc->section('', $h_func, 0, 1);

174  $theOutput .= $this->drawTableOfIndexedPages();

175  }

176  return $theOutput;

177  }

178 

179  /*******************************

180  *

181  * Drawing table of indexed pages

182  *

183  ******************************/

190  public function drawTableOfIndexedPages() {

191  global $BACK_PATH;

192  // Drawing tree:

193  $tree = GeneralUtility::makeInstance('TYPO3\\CMS\\Backend\\Tree\\View\\PageTreeView');

194  $perms_clause = $GLOBALS['BE_USER']->getPagePermsClause(1);

195  $tree->init('AND ' . $perms_clause);

196  $HTML = '<img src="' . $BACK_PATH . \TYPO3\CMS\Backend\Utility\IconUtility::getIcon('pages', $this->pObj->pageinfo) . '" width="18" height="16" align="top" alt="" />';

197  $tree->tree[] = array(

198  'row' => $this->pObj->pageinfo,

199  'HTML' => $HTML

200  );

201  if ($this->pObj->MOD_SETTINGS['depth']) {

202  $tree->getTree($this->pObj->id, $this->pObj->MOD_SETTINGS['depth'], '');

203  }

204  // Traverse page tree:

205  $code = '';

206  foreach ($tree->tree as $data) {

207  $code .= $this->indexed_info($data['row'], $data['HTML'] . $this->showPageDetails(

208  BackendUtility::getRecordTitlePrep($data['row']['title']), $data['row']['uid']));

209  }

210  if ($code) {

211  $code = '<br /><br />

212  <table border="0" cellspacing="1" cellpadding="2" class="c-list">' . $this->printPhashRowHeader() . $code . '</table>';

213  // Create section to output:

214  $theOutput .= $this->pObj->doc->section('', $code, 0, 1);

215  } else {

216  $theOutput .= $this->pObj->doc->section('', '<br /><br />' . $this->pObj->doc->icons(1) . 'There were no indexed pages found in the tree.<br /><br />', 0, 1);

217  }

218  return $theOutput;

219  }

220 

229  public function indexed_info($data, $firstColContent) {

230  // Query:

231  $res = $GLOBALS['TYPO3_DB']->exec_SELECTquery('ISEC.phash_t3, ISEC.rl0, ISEC.rl1, ISEC.rl2, ISEC.page_id, ISEC.uniqid, ' . 'IP.phash, IP.phash_grouping, IP.cHashParams, IP.data_filename, IP.data_page_id, ' . 'IP.data_page_reg1, IP.data_page_type, IP.data_page_mp, IP.gr_list, IP.item_type, ' . 'IP.item_title, IP.item_description, IP.item_mtime, IP.tstamp, IP.item_size, ' . 'IP.contentHash, IP.crdate, IP.parsetime, IP.sys_language_uid, IP.item_crdate, ' . 'IP.externalUrl, IP.recordUid, IP.freeIndexUid, IP.freeIndexSetId, count(*) AS count_val', 'index_phash IP, index_section ISEC', 'IP.phash = ISEC.phash AND ISEC.page_id = ' . (int)$data['uid'], 'IP.phash,IP.phash_grouping,IP.cHashParams,IP.data_filename,IP.data_page_id,IP.data_page_reg1,IP.data_page_type,IP.data_page_mp,IP.gr_list,IP.item_type,IP.item_title,IP.item_description,IP.item_mtime,IP.tstamp,IP.item_size,IP.contentHash,IP.crdate,IP.parsetime,IP.sys_language_uid,IP.item_crdate,ISEC.phash,ISEC.phash_t3,ISEC.rl0,ISEC.rl1,ISEC.rl2,ISEC.page_id,ISEC.uniqid,IP.externalUrl,IP.recordUid,IP.freeIndexUid,IP.freeIndexSetId', 'IP.item_type, IP.tstamp', $this->maxListPerPage + 1);

232  // Initialize variables:

233  $rowCount = 0;

234  $lines = array();

235  // Collecting HTML rows here.

236  $phashAcc = array();

237  // Collecting phash values (to remove local indexing for)

238  $phashAcc[] = 0;

239  // Traverse the result set of phash rows selected:

240  while ($row = $GLOBALS['TYPO3_DB']->sql_fetch_assoc($res)) {

241  if ($rowCount == $this->maxListPerPage) {

242  $rowCount++;

243  // Increase to the extra warning row will appear as well.

244  break;

245  }

246  // Adds a display row:

247  $lines[$row['phash_grouping']][] = $this->printPhashRow($row, isset($lines[$row['phash_grouping']]), $this->getGrListEntriesForPhash($row['phash'], $row['gr_list']));

248  $rowCount++;

249  $phashAcc[] = $row['phash'];

250  $this->allPhashListed[] = $row['phash'];

251  }

252  // Compile rows into the table:

253  $out = '';

254  $cellAttrib = $data['_CSSCLASS'] ? ' class="' . $data['_CSSCLASS'] . '"' : '';

255  if (count($lines)) {

256  $firstColContent = '<td rowspan="' . $rowCount . '"' . $cellAttrib . '>' . $firstColContent . '</td>';

257  foreach ($lines as $rowSet) {

258  foreach ($rowSet as $rows) {

259  $out .= '

260  <tr class="bgColor-20">' . $firstColContent . implode('', $rows) . '</tr>';

261  $firstColContent = '';

262  }

263  }

264  if ($rowCount > $this->maxListPerPage) {

265  // Now checking greater than, because we increased $rowCount before...

266  $out .= '

267  <tr class="bgColor-20">

268  <td>&nbsp;</td>

269  <td colspan="' . ($this->returnNumberOfColumns() - 1) . '">' . $this->pObj->doc->icons(3) . '<span class="">There were more than ' . $this->maxListPerPage . ' rows. <a href="' . htmlspecialchars(BackendUtility::getModuleUrl('web_info', array('id' => $this->pObj->id, 'listAll' => 1))) . '">Click here to list them ALL!</a></span></td>

270  </tr>';

271  }

272  } else {

273  $out .= '

274  <tr class="bgColor-20">

275  <td' . $cellAttrib . '>' . $firstColContent . '</td>

276  <td colspan="' . ($this->returnNumberOfColumns() - 1) . '"><em>Not indexed</em></td>

277  </tr>';

278  }

279  // Checking for phash-rows which are NOT joined with the section table:

280  $res = $GLOBALS['TYPO3_DB']->exec_SELECTquery('IP.*', 'index_phash IP', 'IP.data_page_id = ' . (int)$data['uid'] . ' AND IP.phash NOT IN (' . implode(',', $phashAcc) . ')');

281  while ($row = $GLOBALS['TYPO3_DB']->sql_fetch_assoc($res)) {

282  $out .= '

283  <tr class="typo3-red">

284  <td colspan="' . $this->returnNumberOfColumns() . '"><strong>Warning:</strong> phash-row "' . $row['phash'] . '" didn\'t have a representation in the index_section table!</td>

285  </tr>';

286  $this->allPhashListed[] = $row['phash'];

287  }

288  return $out;

289  }

290 

301  public function printPhashRow($row, $grouping = 0, $extraGrListRows) {

302  $lines = array();

303  // Title cell attributes will highlight TYPO3 pages with a slightly darker color (bgColor4) than attached medias. Also IF there are more than one section record for a phash row it will be red as a warning that something is wrong!

304  $titleCellAttribs = $row['count_val'] != 1 ? ' bgcolor="red"' : ($row['item_type'] === '0' ? ' class="bgColor4"' : '');

305  if ($row['item_type']) {

306  $arr = unserialize($row['cHashParams']);

307  $page = $arr['key'] ? ' [' . $arr['key'] . ']' : '';

308  } else {

309  $page = '';

310  }

311  $elTitle = $this->linkDetails($row['item_title'] ? htmlspecialchars(GeneralUtility::fixed_lgd_cs($row['item_title'], 20) . $page) : '<em>[No Title]</em>', $row['phash']);

312  $cmdLinks = $this->printRemoveIndexed($row['phash'], 'Clear phash-row') . $this->printReindex($row, 'Re-index element');

313  switch ($this->pObj->MOD_SETTINGS['type']) {

314  case 1:

315  // Technical details:

316  // Display icon:

317  if (!$grouping) {

318  $lines[] = '<td>' . $this->makeItemTypeIcon($row['item_type'], ($row['data_filename'] ? $row['data_filename'] : $row['item_title'])) . '</td>';

319  } else {

320  $lines[] = '<td>&nbsp;</td>';

321  }

322  // Title displayed:

323  $lines[] = '<td' . $titleCellAttribs . '>' . $elTitle . '</td>';

324  // Remove-indexing-link:

325  $lines[] = '<td>' . $cmdLinks . '</td>';

326  // Various data:

327  $lines[] = '<td>' . $row['phash'] . '</td>';

328  $lines[] = '<td>' . $row['contentHash'] . '</td>';

329  if ($row['item_type'] === '0') {

330  $lines[] = '<td>' . ($row['data_page_id'] ? $row['data_page_id'] : '&nbsp;') . '</td>';

331  $lines[] = '<td>' . ($row['data_page_type'] ? $row['data_page_type'] : '&nbsp;') . '</td>';

332  $lines[] = '<td>' . ($row['sys_language_uid'] ? $row['sys_language_uid'] : '&nbsp;') . '</td>';

333  $lines[] = '<td>' . ($row['data_page_mp'] ? $row['data_page_mp'] : '&nbsp;') . '</td>';

334  } else {

335  $lines[] = '<td colspan="4">' . htmlspecialchars($row['data_filename']) . '</td>';

336  }

337  $lines[] = '<td>' . $row['gr_list'] . $this->printExtraGrListRows($extraGrListRows) . '</td>';

338  $lines[] = '<td>' . $this->printRootlineInfo($row) . '</td>';

339  $lines[] = '<td>' . ($row['page_id'] ? $row['page_id'] : '&nbsp;') . '</td>';

340  $lines[] = '<td>' . ($row['phash_t3'] != $row['phash'] ? $row['phash_t3'] : '&nbsp;') . '</td>';

341  $lines[] = '<td>' . ($row['freeIndexUid'] ? $row['freeIndexUid'] . ($row['freeIndexSetId'] ? '/' . $row['freeIndexSetId'] : '') : '&nbsp;') . '</td>';

342  $lines[] = '<td>' . ($row['recordUid'] ? $row['recordUid'] : '&nbsp;') . '</td>';

343  // cHash parameters:

344  $arr = unserialize($row['cHashParams']);

345  if (!is_array($arr)) {

346  $arr = array(

347  'cHash' => $GLOBALS['LANG']->sL('LLL:EXT:lang/locallang_general.xlf:LGL.error', TRUE)

348  );

349  }

350  $theCHash = $arr['cHash'];

351  unset($arr['cHash']);

352  if ($row['item_type']) {

353  // pdf...

354  $lines[] = '<td>' . ($arr['key'] ? 'Page ' . $arr['key'] : '') . '&nbsp;</td>';

355  } elseif ($row['item_type'] == 0) {

356  $lines[] = '<td>' . htmlspecialchars(GeneralUtility::implodeArrayForUrl('', $arr)) . '&nbsp;</td>';

357  } else {

358  $lines[] = '<td class="bgColor">&nbsp;</td>';

359  }

360  $lines[] = '<td>' . $theCHash . '</td>';

361  break;

362  case 2:

363  // Words and content:

364  // Display icon:

365  if (!$grouping) {

366  $lines[] = '<td>' . $this->makeItemTypeIcon($row['item_type'], ($row['data_filename'] ? $row['data_filename'] : $row['item_title'])) . '</td>';

367  } else {

368  $lines[] = '<td>&nbsp;</td>';

369  }

370  // Title displayed:

371  $lines[] = '<td' . $titleCellAttribs . '>' . $elTitle . '</td>';

372  // Remove-indexing-link:

373  $lines[] = '<td>' . $cmdLinks . '</td>';

374  // Query:

375  $ftrow = $GLOBALS['TYPO3_DB']->exec_SELECTgetSingleRow('*', 'index_fulltext', 'phash = ' . (int)$row['phash']);

376  $lines[] = '<td style="white-space: normal;">' . htmlspecialchars(GeneralUtility::fixed_lgd_cs($ftrow['fulltextdata'], 3000)) . '<hr/><em>Size: ' . strlen($ftrow['fulltextdata']) . '</em>' . '</td>';

377  // Query:

378  $ftrows = $GLOBALS['TYPO3_DB']->exec_SELECTgetRows('index_words.baseword, index_rel.*', 'index_rel, index_words', 'index_rel.phash = ' . (int)$row['phash'] . ' AND index_words.wid = index_rel.wid', '', '', '', 'baseword');

379  $wordList = '';

380  if (is_array($ftrows)) {

381  $indexed_words = array_keys($ftrows);

382  sort($indexed_words);

383  $wordList = htmlspecialchars(implode(' ', $indexed_words));

384  $wordList .= '<hr/><em>Count: ' . count($indexed_words) . '</em>';

385  }

386  $lines[] = '<td style="white-space: normal;">' . $wordList . '</td>';

387  break;

388  default:

389  // Overview

390  // Display icon:

391  if (!$grouping) {

392  $lines[] = '<td>' . $this->makeItemTypeIcon($row['item_type'], ($row['data_filename'] ? $row['data_filename'] : $row['item_title'])) . '</td>';

393  } else {

394  $lines[] = '<td>&nbsp;</td>';

395  }

396  // Title displayed:

397  $lines[] = '<td' . $titleCellAttribs . '>' . $elTitle . '</td>';

398  // Remove-indexing-link:

399  $lines[] = '<td>' . $cmdLinks . '</td>';

400  $lines[] = '<td style="white-space: normal;">' . htmlspecialchars($row['item_description']) . '...</td>';

401  $lines[] = '<td>' . GeneralUtility::formatSize($row['item_size']) . '</td>';

402  $lines[] = '<td>' . BackendUtility::dateTimeAge($row['tstamp']) . '</td>';

403  }

404  return $lines;

405  }

406 

413  public function printPhashRowHeader() {

414  $lines = array();

415  switch ($this->pObj->MOD_SETTINGS['type']) {

416  case 1:

417  $lines[] = '<td>&nbsp;</td>';

418  $lines[] = '<td>&nbsp;</td>';

419  $lines[] = '<td>Title</td>';

420  $lines[] = '<td bgcolor="red">' . $this->printRemoveIndexed('ALL', 'Clear ALL phash-rows below!') . '</td>';

421  $lines[] = '<td>pHash</td>';

422  $lines[] = '<td>contentHash</td>';

423  $lines[] = '<td>&amp;id</td>';

424  $lines[] = '<td>&amp;type</td>';

425  $lines[] = '<td>&amp;L</td>';

426  $lines[] = '<td>&amp;MP</td>';

427  $lines[] = '<td>grlist</td>';

428  $lines[] = '<td>Rootline</td>';

429  $lines[] = '<td>page_id</td>';

430  $lines[] = '<td>phash_t3</td>';

431  $lines[] = '<td>CfgUid</td>';

432  $lines[] = '<td>RecUid</td>';

433  $lines[] = '<td>GET-parameters</td>';

434  $lines[] = '<td>&amp;cHash</td>';

435  break;

436  case 2:

437  $lines[] = '<td>&nbsp;</td>';

438  $lines[] = '<td>&nbsp;</td>';

439  $lines[] = '<td>Title</td>';

440  $lines[] = '<td bgcolor="red">' . $this->printRemoveIndexed('ALL', 'Clear ALL phash-rows below!') . '</td>';

441  $lines[] = '<td>Content<br />

442  <img src="clear.gif" width="300" height="1" alt="" /></td>';

443  $lines[] = '<td>Words<br />

444  <img src="clear.gif" width="300" height="1" alt="" /></td>';

445  break;

446  default:

447  $lines[] = '<td>&nbsp;</td>';

448  $lines[] = '<td>&nbsp;</td>';

449  $lines[] = '<td>Title</td>';

450  $lines[] = '<td bgcolor="red">' . $this->printRemoveIndexed('ALL', 'Clear ALL phash-rows below!') . '</td>';

451  $lines[] = '<td>Description</td>';

452  $lines[] = '<td>Size</td>';

453  $lines[] = '<td>Indexed:</td>';

454  }

455  $out = '<tr class="tableheader bgColor5">' . implode('', $lines) . '</tr>';

456  return $out;

457  }

458 

465  public function returnNumberOfColumns() {

466  switch ($this->pObj->MOD_SETTINGS['type']) {

467  case 1:

468  return 18;

469  break;

470  case 2:

471  return 6;

472  break;

473  default:

474  return 7;

475  }

476  }

477 

478  /*******************************

479  *

480  * Details display, phash row

481  *

482  *******************************/

490  public function showDetailsForPhash($phash) {

491  $content = '';

492  $phash = (int)$phash;

493  // Selects the result row:

494  $phashRecord = $GLOBALS['TYPO3_DB']->exec_SELECTgetSingleRow('*', 'index_phash', 'phash = ' . $phash);

495  // If found, display:

496  if (is_array($phashRecord)) {

497  $content .= '<h4>phash row content:</h4>' . \TYPO3\CMS\Core\Utility\DebugUtility::viewArray($phashRecord);

498  // Getting debug information if any:

499  $ftrows = $GLOBALS['TYPO3_DB']->exec_SELECTgetRows('*', 'index_debug', 'phash = ' . $phash);

500  if (is_array($ftrows)) {

501  $debugInfo = unserialize($ftrows[0]['debuginfo']);

502  $lexer = $debugInfo['lexer'];

503  unset($debugInfo['lexer']);

504  $content .= '<h3>Debug information:</h3>' . \TYPO3\CMS\Core\Utility\DebugUtility::viewArray($debugInfo);

505  $content .= '<h4>Debug information / lexer splitting:</h4>' . '<hr/><strong>' . $lexer . '</strong><hr/>';

506  }

507  $content .= '<h3>Word statistics</h3>';

508  // Finding all words for this phash:

509  $ftrows = $GLOBALS['TYPO3_DB']->exec_SELECTgetRows('index_words.*, index_rel.*', 'index_rel, index_words', 'index_rel.phash = ' . $phash . ' AND index_words.wid = index_rel.wid', '', 'index_words.baseword', '');

510  $pageRec = BackendUtility::getRecord('pages', $phashRecord['data_page_id']);

511  $showStopWordCheckBox = $GLOBALS['BE_USER']->isAdmin();

512  $content .= $this->listWords($ftrows, 'All words found on page (' . count($ftrows) . '):', $showStopWordCheckBox, $pageRec);

513  if ($this->enableMetaphoneSearch) {

514  // Group metaphone hash:

515  $metaphone = array();

516  foreach ($ftrows as $row) {

517  $metaphone[$row['metaphone']][] = $row['baseword'];

518  }

519  $content .= $this->listMetaphoneStat($metaphone, 'Metaphone stats:');

520  }

521  // Finding top-20 on frequency for this phash:

522  $ftrows = $GLOBALS['TYPO3_DB']->exec_SELECTgetRows('index_words.baseword, index_words.metaphone, index_rel.*', 'index_rel, index_words', 'index_rel.phash = ' . $phash . ' AND index_words.wid = index_rel.wid

523  AND index_words.is_stopword=0', '', 'index_rel.freq DESC', '20');

524  $content .= $this->listWords($ftrows, 'Top-20 words by frequency:', 2);

525  // Finding top-20 on count for this phash:

526  $ftrows = $GLOBALS['TYPO3_DB']->exec_SELECTgetRows('index_words.baseword, index_words.metaphone, index_rel.*', 'index_rel, index_words', 'index_rel.phash = ' . $phash . ' AND index_words.wid = index_rel.wid

527  AND index_words.is_stopword=0', '', 'index_rel.count DESC', '20');

528  $content .= $this->listWords($ftrows, 'Top-20 words by count:', 2);

529  $content .= '<h3>Section records for this phash</h3>';

530  // Finding sections for this record:

531  $ftrows = $GLOBALS['TYPO3_DB']->exec_SELECTgetRows('*', 'index_section', 'index_section.phash = ' . $phash, '', '', '');

532  $content .= \TYPO3\CMS\Core\Utility\DebugUtility::viewArray($ftrows);

533  // Add go-back link:

534  $content = $this->linkList() . $content . $this->linkList();

535  } else {

536  $content .= 'Error: No phash row found';

537  }

538  return $content;

539  }

540 

551  public function listWords($ftrows, $header, $stopWordBoxes = FALSE, $page = '') {

552  // Prepare keywords:

553  $keywords = is_array($page) ? array_flip(GeneralUtility::trimExplode(',', $page['keywords'], TRUE)) : '';

554  // Render list:

555  $trows = '';

556  $trows .= '

557  <tr class="tableheader bgColor5">

558  ' . ($stopWordBoxes ? '<td>' . htmlspecialchars('Stopword:') . '</td>' : '') . '

559  <td>' . htmlspecialchars('Word:') . '</td>

560  <td>' . htmlspecialchars('Count:') . '</td>

561  <td>' . htmlspecialchars('First:') . '</td>

562  <td>' . htmlspecialchars('Frequency:') . '</td>

563  <td>' . htmlspecialchars('Flags:') . '</td>

564  ' . (is_array($keywords) ? '<td>' . htmlspecialchars('Page keyword:') . '</td>' : '') . '

565  </tr>

566  ';

567  foreach ($ftrows as $row) {

568  $hiddenField = $stopWordBoxes != 2 ? '<input type="hidden" name="stopWord[' . $row['wid'] . ']" value="0" />' : '';

569  $trows .= '

570  <tr class="' . ($row['is_stopword'] ? 'bgColor' : 'bgColor4') . '">

571  ' . ($stopWordBoxes ? '<td align="center"' . ($row['is_stopword'] ? ' style="background-color:red;"' : '') . '>' . $hiddenField . '<input type="checkbox" name="stopWord[' . $row['wid'] . ']" value="1"' . ($row['is_stopword'] ? 'checked="checked"' : '') . ' /></td>' : '') . '

572  <td>' . $this->linkWordDetails(htmlspecialchars($row['baseword']), $row['wid']) . '</td>

573  <td>' . htmlspecialchars($row['count']) . '</td>

574  <td>' . htmlspecialchars($row['first']) . '</td>

575  <td>' . htmlspecialchars($row['freq']) . '</td>

576  <td>' . htmlspecialchars($this->flagsMsg($row['flags'])) . '</td>

577  ' . (is_array($keywords) ? '<td align="center"' . (isset($keywords[$row['baseword']]) ? ' class="bgColor2"' : '') . '><input type="hidden" name="pageKeyword[' . $row['baseword'] . ']" value="0" /><input type="checkbox" name="pageKeyword[' . $row['baseword'] . ']" value="1"' . (isset($keywords[$row['baseword']]) ? 'checked="checked"' : '') . ' /></td>' : '') . '

578  </tr>

579  ';

580  }

581  return '<h4>' . htmlspecialchars($header) . '</h4>' . '

582  <table border="0" cellspacing="1" cellpadding="2" class="c-list">

583  ' . $trows . '

584  </table>' . ($stopWordBoxes ? '<input type="submit" value="Change stop-word settings" name="_stopwords" onclick="document.webinfoForm.action=\'' . htmlspecialchars(GeneralUtility::getIndpEnv('REQUEST_URI')) . '\';" />' : '') . (is_array($keywords) ? '<input type="submit" value="Set page keywords" name="_pageKeywords" onclick="document.webinfoForm.action=\'' . htmlspecialchars(GeneralUtility::getIndpEnv('REQUEST_URI')) . '\';" /><input type="hidden" name="pageKeyword_pageUid" value="' . $page['uid'] . '" />' . '<br />Current keywords are: <em>' . htmlspecialchars(implode(', ', array_keys($keywords))) . '</em>' : '');

585  }

586 

595  public function listMetaphoneStat($ftrows, $header) {

596  $trows = '';

597  $trows .= '

598  <tr class="tableheader bgColor5">

599  <td>' . htmlspecialchars('Metaphone:') . '</td>

600  <td>' . htmlspecialchars('Hash:') . '</td>

601  <td>' . htmlspecialchars('Count:') . '</td>

602  <td>' . htmlspecialchars('Words:') . '</td>

603  </tr>

604  ';

605  foreach ($ftrows as $metaphone => $words) {

606  if (count($words) > 1) {

607  $trows .= '

608  <tr class="bgColor4">

609  <td>' . $this->linkMetaPhoneDetails($this->indexerObj->metaphone($words[0], 1), $metaphone) . '</td>

610  <td>' . htmlspecialchars($metaphone) . '</td>

611  <td>' . htmlspecialchars(count($words)) . '</td>

612  <td style="white-space: normal;">' . htmlspecialchars(implode(', ', $words)) . '</td>

613  </tr>

614  ';

615  }

616  }

617  return '<h4>' . htmlspecialchars($header) . '</h4>' . '<table border="0" cellspacing="1" cellpadding="2" class="c-list">

618  ' . $trows . '

619  </table>';

620  }

621 

630  public function linkWordDetails($string, $wid) {

631  return '<a href="' . htmlspecialchars(GeneralUtility::linkThisScript(array('wid' => $wid, 'phash' => ''))) . '">' . $string . '</a>';

632  }

633 

642  public function linkMetaPhoneDetails($string, $metaphone) {

643  return '<a href="' . htmlspecialchars(GeneralUtility::linkThisScript(array('metaphone' => $metaphone, 'wid' => '', 'phash' => ''))) . '">' . $string . '</a>';

644  }

645 

653  public function flagsMsg($flags) {

654  if ($flags > 0) {

655  return ($flags & 128 ? '<title>' : '') . ($flags & 64 ? '<meta/keywords>' : '') . ($flags & 32 ? '<meta/description>' : '') . ' (' . $flags . ')';

656  }

657  }

658 

659  /*******************************

660  *

661  * Details display, words / metaphone

662  *

663  *******************************/

671  public function showDetailsForWord($wid) {

672  // Select references to this word

673  $ftrows = $GLOBALS['TYPO3_DB']->exec_SELECTgetRows('index_phash.*, index_section.*, index_rel.*', 'index_rel, index_section, index_phash', 'index_rel.wid = ' . (int)$wid . ' AND index_rel.phash = index_section.phash' . ' AND index_section.phash = index_phash.phash', '', 'index_rel.freq DESC', '');

674  // Headers:

675  $content = '

676  <tr class="tableheader bgColor5">

677  <td>phash</td>

678  <td>page_id</td>

679  <td>data_filename</td>

680  <td>count</td>

681  <td>first</td>

682  <td>freq</td>

683  <td>flags</td>

684  </tr>';

685  if (is_array($ftrows)) {

686  foreach ($ftrows as $wDat) {

687  $content .= '

688  <tr class="bgColor4">

689  <td>' . $this->linkDetails(htmlspecialchars($wDat['phash']), $wDat['phash']) . '</td>

690  <td>' . htmlspecialchars($wDat['page_id']) . '</td>

691  <td>' . htmlspecialchars($wDat['data_filename']) . '</td>

692  <td>' . htmlspecialchars($wDat['count']) . '</td>

693  <td>' . htmlspecialchars($wDat['first']) . '</td>

694  <td>' . htmlspecialchars($wDat['freq']) . '</td>

695  <td>' . htmlspecialchars($wDat['flags']) . '</td>

696  </tr>';

697  }

698  }

699  // Compile table:

700  $content = '

701  <table border="0" cellspacing="1" cellpadding="2" class="c-list">' . $content . '

702  </table>';

703  // Add go-back link:

704  $content = $content . $this->linkList();

705  return $content;

706  }

707 

715  public function showDetailsForMetaphone($metaphone) {

716  // Finding top-20 on frequency for this phash:

717  $ftrows = $GLOBALS['TYPO3_DB']->exec_SELECTgetRows('index_words.*', 'index_words', 'index_words.metaphone = ' . (int)$metaphone, '', 'index_words.baseword', '');

718  if (count($ftrows)) {

719  $content .= '<h4>Metaphone: ' . $this->indexerObj->metaphone($ftrows[0]['baseword'], 1) . '</h4>';

720  $content .= '

721  <tr class="tableheader bgColor5">

722  <td>Word</td>

723  <td>Is stopword?</td>

724  </tr>';

725  if (is_array($ftrows)) {

726  foreach ($ftrows as $wDat) {

727  $content .= '

728  <tr class="bgColor4">

729  <td>' . $this->linkWordDetails(htmlspecialchars($wDat['baseword']), $wDat['wid']) . '</td>

730  <td>' . htmlspecialchars(($wDat['is_stopword'] ? 'YES' : 'No')) . '</td>

731  </tr>';

732  }

733  }

734  $content = '

735  <table border="0" cellspacing="1" cellpadding="2" class="c-list">' . $content . '

736  </table>';

737  if ($this->indexerObj->metaphone($ftrows[0]['baseword']) != $metaphone) {

738  $content .= 'ERROR: Metaphone string and hash did not match for some reason!?';

739  }

740  // Add go-back link:

741  $content = $content . $this->linkList();

742  }

743  return $content;

744  }

745 

746  /*******************************

747  *

748  * Helper functions

749  *

750  *******************************/

759  public function printRemoveIndexed($phash, $alt) {

760  return '<a href="' . htmlspecialchars(GeneralUtility::linkThisScript(array('deletePhash' => $phash))) . '" title="' . htmlspecialchars($alt) . '">' . \TYPO3\CMS\Backend\Utility\IconUtility::getSpriteIcon('actions-edit-delete') . '</a>';

761  }

762 

771  public function printReindex($resultRow, $alt) {

772  if ($resultRow['item_type'] && $resultRow['item_type'] !== '0') {

773  return '<a href="' . htmlspecialchars(GeneralUtility::linkThisScript(array('reindex' => $resultRow['phash'], 'reindex_id' => $resultRow['page_id']))) . '">' . '<img ' . \TYPO3\CMS\Backend\Utility\IconUtility::skinImg($GLOBALS['BACK_PATH'], 'gfx/refresh_n.gif', 'width="14" height="14"') . ' hspace="1" vspace="2" border="0" title="' . htmlspecialchars($alt) . '" alt="" />' . '</a>';

774  }

775  }

776 

785  public function linkDetails($string, $phash) {

786  return '<a href="' . htmlspecialchars(GeneralUtility::linkThisScript(array('phash' => $phash))) . '">' . $string . '</a>';

787  }

788 

795  public function linkList() {

796  return '<br /><a href="' . htmlspecialchars(BackendUtility::getModuleUrl('web_info', array('id' => $this->pObj->id))) . '">Back to list.</a><br />';

797  }

798 

807  public function showPageDetails($string, $id) {

808  return '<a href="' . htmlspecialchars(BackendUtility::getModuleUrl('web_info', array('id' => $id, 'SET' => array('depth' => 0, 'type' => 1)))) . '">' . $string . '</a>';

809  }

810 

818  public function printExtraGrListRows($extraGrListRows) {

819  if (count($extraGrListRows)) {

820  $lines = array();

821  foreach ($extraGrListRows as $r) {

822  $lines[] = $r['gr_list'];

823  }

824  return '<br />' . $GLOBALS['TBE_TEMPLATE']->dfw(implode('<br />', $lines));

825  }

826  }

827 

835  public function printRootlineInfo($row) {

836  $uidCollection = array();

837  if ($row['rl0']) {

838  $uidCollection[0] = $row['rl0'];

839  if ($row['rl1']) {

840  $uidCollection[1] = $row['rl1'];

841  if ($row['rl2']) {

842  $uidCollection[2] = $row['rl2'];

843  // Additional levels:

844  if (is_array($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['indexed_search']['addRootLineFields'])) {

845  foreach ($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['indexed_search']['addRootLineFields'] as $fieldName => $rootLineLevel) {

846  if ($row[$fieldName]) {

847  $uidCollection[$rootLineLevel] = $row[$fieldName];

848  }

849  }

850  }

851  }

852  }

853  }

854  // Return root line.

855  ksort($uidCollection);

856  return implode('/', $uidCollection);

857  }

858 

867  public function makeItemTypeIcon($it, $alt = '') {

868  if (!isset($this->iconFileNameCache[$it])) {

869  if ($it === '0') {

870  $icon = 'EXT:indexed_search/pi/res/pages.gif';

871  } elseif ($this->external_parsers[$it]) {

872  $icon = $this->external_parsers[$it]->getIcon($it);

873  }

874  $fullPath = GeneralUtility::getFileAbsFileName($icon);

875  if ($fullPath) {

876  $info = @getimagesize($fullPath);

877  $iconPath = $GLOBALS['BACK_PATH'] . '../' . \TYPO3\CMS\Core\Utility\PathUtility::stripPathSitePrefix($fullPath);

878  $this->iconFileNameCache[$it] = is_array($info) ? '<img src="' . $iconPath . '" ' . $info[3] . ' title="###TITLE_ATTRIBUTE###" alt="" />' : '';

879  }

880  }

881  return str_replace('###TITLE_ATTRIBUTE###', htmlspecialchars($it . ': ' . $alt), $this->iconFileNameCache[$it]);

882  }

883 

884  /********************************

885  *

886  * Reindexing

887  *

888  *******************************/

897  public function reindexPhash($phash, $pageId) {

898  // Query:

899  $resultRow = $GLOBALS['TYPO3_DB']->exec_SELECTgetSingleRow('ISEC.*, IP.*', 'index_phash IP, index_section ISEC', 'IP.phash = ISEC.phash

900  AND IP.phash = ' . (int)$phash . '

901  AND ISEC.page_id = ' . (int)$pageId);

902  $content = '';

903  if (is_array($resultRow)) {

904  if ($resultRow['item_type'] && $resultRow['item_type'] !== '0') {

905  // (Re)-Indexing file on page.

906  $indexerObj = GeneralUtility::makeInstance('TYPO3\\CMS\\IndexedSearch\\Indexer');

907  $indexerObj->backend_initIndexer($pageId, 0, 0, '', $this->getUidRootLineForClosestTemplate($pageId));

908  // URL or local file:

909  if ($resultRow['externalUrl']) {

910  $indexerObj->indexExternalUrl($resultRow['data_filename']);

911  } else {

912  $indexerObj->indexRegularDocument($resultRow['data_filename'], TRUE);

913  }

914  if ($indexerObj->file_phash_arr['phash'] != $resultRow['phash']) {

915  $content .= 'ERROR: phash (' . $indexerObj->file_phash_arr['phash'] . ') did NOT match ' . $resultRow['phash'] . ' for strange reasons!';

916  }

917  $content .= '<h4>Log for re-indexing of "' . htmlspecialchars($resultRow['data_filename']) . '":</h4>';

918  $content .= \TYPO3\CMS\Core\Utility\DebugUtility::viewArray($indexerObj->internal_log);

919  $content .= '<h4>Hash-array, page:</h4>';

920  $content .= \TYPO3\CMS\Core\Utility\DebugUtility::viewArray($indexerObj->hash);

921  $content .= '<h4>Hash-array, file:</h4>';

922  $content .= \TYPO3\CMS\Core\Utility\DebugUtility::viewArray($indexerObj->file_phash_arr);

923  }

924  }

925  // Link back to list.

926  $content .= $this->linkList();

927  return $content;

928  }

929 

938  public function getUidRootLineForClosestTemplate($id) {

939  $tmpl = GeneralUtility::makeInstance('TYPO3\\CMS\\Core\\TypoScript\\ExtendedTemplateService');

940  // Defined global here!

941  $tmpl->tt_track = 0;

942  // Do not log time-performance information

943  $tmpl->init();

944  // Gets the rootLine

945  $sys_page = GeneralUtility::makeInstance('TYPO3\\CMS\\Frontend\\Page\\PageRepository');

946  $rootLine = $sys_page->getRootLine($id);

947  $tmpl->runThroughTemplates($rootLine, 0);

948  // This generates the constants/config + hierarchy info for the template.

949  // Root line uids

950  $rootline_uids = array();

951  foreach ($tmpl->rootLine as $rlkey => $rldat) {

952  $rootline_uids[$rlkey] = $rldat['uid'];

953  }

954  return $rootline_uids;

955  }

956 

957  /********************************

958  *

959  * SQL functions

960  *

961  *******************************/

970  public function removeIndexedPhashRow($phashList, $clearPageCache = 1) {

971  // FIXME: This is only a workaround

972  if ($phashList == 'ALL') {

973  $this->drawTableOfIndexedPages();

974  $phashRows = $this->allPhashListed;

975  $this->allPhashListed = array();

976  } else {

977  $phashRows = GeneralUtility::trimExplode(',', $phashList, TRUE);

978  }

979  foreach ($phashRows as $phash) {

980  $phash = (int)$phash;

981  if ($phash > 0) {

982  if ($clearPageCache) {

983  // Clearing page cache:

984  $res = $GLOBALS['TYPO3_DB']->exec_SELECTquery('page_id', 'index_section', 'phash=' . (int)$phash);

985  if ($GLOBALS['TYPO3_DB']->sql_num_rows($res)) {

986  $idList = array();

987  while ($row = $GLOBALS['TYPO3_DB']->sql_fetch_assoc($res)) {

988  $idList[] = (int)$row['page_id'];

989  }

990  $pageCache = GeneralUtility::makeInstance('TYPO3\\CMS\\Core\\Cache\\CacheManager')->getCache('cache_pages');

991  foreach ($idList as $pageId) {

992  $pageCache->flushByTag('pageId_' . $pageId);

993  }

994  }

995  }

996  // Removing old registrations for all tables.

997  $tableArr = explode(',', 'index_phash,index_rel,index_section,index_grlist,index_fulltext,index_debug');

998  foreach ($tableArr as $table) {

999  $GLOBALS['TYPO3_DB']->exec_DELETEquery($table, 'phash=' . (int)$phash);

1000  }

1001  }

1002  }

1003  }

1004 

1013  public function getGrListEntriesForPhash($phash, $gr_list) {

1014  $res = $GLOBALS['TYPO3_DB']->exec_SELECTquery('*', 'index_grlist', 'phash=' . (int)$phash);

1015  $lines = array();

1016  $isRemoved = 0;

1017  while ($row = $GLOBALS['TYPO3_DB']->sql_fetch_assoc($res)) {

1018  if (!$isRemoved && (string)$row['gr_list'] === (string)$gr_list) {

1019  $isRemoved = 1;

1020  } else {

1021  $lines[] = $row;

1022  }

1023  }

1024  return $lines;

1025  }

1026 

1034  public function processStopWords($stopWords) {

1035  if ($GLOBALS['BE_USER']->isAdmin()) {

1036  // Traverse words

1037  foreach ($stopWords as $wid => $state) {

1038  $fieldArray = array(

1039  'is_stopword' => $state

1040  );

1041  $GLOBALS['TYPO3_DB']->exec_UPDATEquery('index_words', 'wid=' . $wid, $fieldArray);

1042  }

1043  }

1044  }

1045 

1054  public function processPageKeywords($pageKeywords, $pageUid) {

1055  // Get pages current keywords

1056  $pageRec = BackendUtility::getRecord('pages', $pageUid);

1057  $keywords = array_flip(GeneralUtility::trimExplode(',', $pageRec['keywords'], TRUE));

1058  // Merge keywords:

1059  foreach ($pageKeywords as $key => $v) {

1060  if ($v) {

1061  $keywords[$key] = 1;

1062  } else {

1063  unset($keywords[$key]);

1064  }

1065  }

1066  // Compile new list:

1067  $data = array();

1068  $data['pages'][$pageUid]['keywords'] = implode(', ', array_keys($keywords));

1069  $tce = GeneralUtility::makeInstance('TYPO3\\CMS\\Core\\DataHandling\\DataHandler');

1070  $tce->stripslashes_values = 0;

1071  $tce->start($data, array());

1072  $tce->process_datamap();

1073  }

1074 

1075 }