--- /home/shenliang/Desktop/apachesolr_attachments.admin.inc 2010-12-02 16:03:57.000000000 +0800 +++ /var/www/drupal_commons/sites/all/modules/apachesolr_attachments/apachesolr_attachments.admin.inc 2010-12-26 15:34:27.000000000 +0800 @@ -1,5 +1,5 @@ t("The name of the tika CLI application jar file, e.g. tika-0.3.jar or tika-app-0.4.jar."), '#default_value' => variable_get('apachesolr_attachments_tika_jar', 'tika-0.3.jar'), ); + $form['apachesolr_attachments_poppler'] = array( + '#type' => 'textfield', + '#title' => t('Path to pdftotext file'), + '#description' => t('Use Poppler pdftotext command to parse pdf file'), + '#default_value' => variable_get('apachesolr_attachments_poppler', '/usr/bin/pdftotext'), + ); + $form['apachesolr_attachments_catdoc'] = array( + '#type' => 'textfield', + '#title' => t('Path to catdoc'), + '#description' => t('Use catdoc command to parse doc file'), + '#default_value' => variable_get('apachesolr_attachments_catdoc', '/usr/bin/catdoc'), + ); + $form['apachesolr_attachments_catppt'] = array( + '#type' => 'textfield', + '#title' => t('Path to catppt'), + '#description' => t('Use catppt command to parse ppt file'), + '#default_value' => variable_get('apachesolr_attachments_catppt', '/usr/bin/catppt'), + ); + $form['apachesolr_attachments_xls2csv'] = array( + '#type' => 'textfield', + '#title' => t('Path to xls2csv'), + '#description' => t('Use xls2csv command to parse xls file'), + '#default_value' => variable_get('apachesolr_attachments_xls2csv', '/usr/bin/xls2csv'), + ); $form['apachesolr_attachments-cron-settings'] = array( '#type' => 'fieldset', '#title' => t('Cron settings'), @@ -208,8 +232,7 @@ function apachesolr_attachements_delete_ return TRUE; } catch (Exception $e) { - // Shortened project name because the watchdog limits type to 16 characters. - watchdog('ApacheSolrAttach', nl2br(check_plain($e->getMessage())), NULL, WATCHDOG_ERROR); + watchdog('Apache Solr Attachments', nl2br(check_plain($e->getMessage())), NULL, WATCHDOG_ERROR); } return FALSE; } @@ -272,13 +295,6 @@ function apachesolr_attachments_add_docu $document->promote = $node->promote; $document->uid = $node->uid; $document->name = $node->name; - if (empty($node->language)) { - // 'und' is the language-neutral code in Drupal 7. - $document->language = 'und'; - } - else { - $document->language = $node->language; - } $document->body = apachesolr_clean_text($file->description) .' '. $text; $document->ss_filemime = $file->filemime; @@ -297,8 +313,7 @@ function apachesolr_attachments_add_docu $documents[] = $document; } else { - // Shortened project name because the watchdog limits type to 16 characters. - watchdog('ApacheSolrAttach', 'Could not extract any indexable text from %filepath', array('%filepath' => $file->filepath), WATCHDOG_WARNING); + watchdog('Apache Solr Attachments', 'Could not extract any indexable text from %filepath', array('%filepath' => $file->filepath), WATCHDOG_WARNING); } } } @@ -416,6 +431,36 @@ function apachesolr_attachments_get_atta return $cached['body']; } + switch ($file->filemime){ + case 'application/pdf': + $text = apachesolr_attachments_extract_using_poppler($filepath); + $text = trim(apachesolr_clean_text($text)); + db_query("UPDATE {apachesolr_attachments_files} SET sha1 = '%s', body = '%s' WHERE fid = %d", $sha1, $text, $file->fid); + return $text; + case 'application/msword': + $text = apachesolr_attachments_extract_using_catdoc($filepath); + $text = trim(apachesolr_clean_text($text)); + db_query("UPDATE {apachesolr_attachments_files} SET sha1 = '%s', body = '%s' WHERE fid = %d", $sha1, $text, $file->fid); + return $text; + case 'application/vnd.ms-powerpoint': + $text = apachesolr_attachments_extract_using_catppt($filepath); + $text = trim(apachesolr_clean_text($text)); + db_query("UPDATE {apachesolr_attachments_files} SET sha1 = '%s', body = '%s' WHERE fid = %d", $sha1, $text, $file->fid); + return $text; + case 'application/vnd.ms-excel': + $text = apachesolr_attachments_extract_using_xls2csv($filepath); + $text = trim(apachesolr_clean_text($text)); + db_query("UPDATE {apachesolr_attachments_files} SET sha1 = '%s', body = '%s' WHERE fid = %d", $sha1, $text, $file->fid); + return $text; + } + + if (array_pop(explode('.', $file->filename)) == 'epub') { + $text = apachesolr_attachments_extract_using_ebooklib($filepath); + $text = trim(apachesolr_clean_text($text)); + db_query("UPDATE {apachesolr_attachments_files} SET sha1 = '%s', body = '%s' WHERE fid = %d", $sha1, $text, $file->fid); + return $text; + } + if (variable_get('apachesolr_attachment_extract_using', 'tika') == 'tika') { $text = apachesolr_attachments_extract_using_tika($filepath); } @@ -426,8 +471,7 @@ function apachesolr_attachments_get_atta } catch (Exception $e) { // Exceptions from Solr may be transient, or indicate a problem with a specific file. - // Shortened project name because the watchdog limits type to 16 characters. - watchdog('ApacheSolrAttach', "Exception occured sending %filepath to Solr\n!message", array('%filepath' => $file->filepath, '!message' => nl2br(check_plain($e->getMessage()))), WATCHDOG_ERROR); + watchdog('Apache Solr Attachments', "Exception occured sending %filepath to Solr\n!message", array('%filepath' => $file->filepath, '!message' => nl2br(check_plain($e->getMessage()))), WATCHDOG_ERROR); return ''; } } @@ -494,3 +538,60 @@ function apachesolr_attachments_extract_ return array($response->extracted, $response->extracted_metadata); } +function apachesolr_attachments_extract_using_ebooklib($filepath) { + $ebooklib = variable_get('ebooklib_path', 'sites/all/libraries/eBookLib'); + if (file_exists($ebooklib .'/ebookData.php') && file_exists($ebooklib .'/ebookRead.php')) { + require_once($ebooklib .'/ebookData.php'); + require_once($ebooklib .'/ebookRead.php'); + $ebook = new ebookRead($filepath); + $spine = $ebook->getSpine(); + $content =''; + foreach ($spine as $index => $id) { + $content .= $ebook->getContentById($id); + } + return $content; + } + else { + return 'FALSE'; + } +} + +function apachesolr_attachments_extract_using_poppler($filepath) { + if (file_exists(variable_get('apachesolr_attachments_poppler', '/usr/bin/pdftotext'))){ + $cmd = 'pdftotext "' . $filepath . '" -'; + return shell_exec($cmd); + } + else { + return FALSE; + } +} + +function apachesolr_attachments_extract_using_catdoc($filepath) { + if (file_exists(variable_get('apachesolr_attachments_catdoc', '/usr/bin/catdoc'))){ + $cmd = 'LANG=en_US.UTF-8; catdoc -w "' . $filepath . '"'; + return shell_exec($cmd); + } + else { + return FALSE; + } +} + +function apachesolr_attachments_extract_using_catppt($filepath) { + if (file_exists(variable_get('apachesolr_attachments_catppt', '/usr/bin/catppt'))){ + $cmd = 'LANG=en_US.UTF-8; catppt "' . $filepath . '"'; + return shell_exec($cmd); + } + else { + return FALSE; + } +} + +function apachesolr_attachments_extract_using_xls2csv($filepath) { + if (file_exists(variable_get('apachesolr_attachments_xls2csv', '/usr/bin/xls2csv'))){ + $cmd = 'LANG=en_US.UTF-8; xls2csv "' . $filepath . '"'; + return shell_exec($cmd); + } + else { + return FALSE; + } +}