Peter M. Groen / knowledgetree

Commit 81865212adce4d390cca58695d3874c703c1a509

Authored by kevin_fourie
1 parent 319ec2a8

Merged in from DEV trunk... 

KTS-673
"The search algorithm needs some work"
Updated. Integrated with DocumentConverter.py

Committed By: Conrad Vermeulen
Reviewed By: Kevin Fourie

KTS-2547
"Test RTF Indexer"
Implemented. 

Committed By: Conrad Vermeulen
Reviewed By: Kevin Fourie

KTS-673
"The search algorithm needs some work"
Updated.

Committed By: Conrad Vermeulen
Reviewed By: Kevin Fourie

KTS-673
"The search algorithm needs some work"
Updated.  MailMime is not tested... will implement again when have more time.

Committed By: Conrad Vermeulen
Reviewed By: Kevin Fourie

KTS-673
"The search algorithm needs some work"
Updated. Sometimes the extractor may change the target filename. 

Committed By: Conrad Vermeulen
Reviewed By: Kevin Fourie

KTS-1753
" Implement Disk Usage Plugin"
Updated. Licensing.

Committed By: Conrad Vermeulen
Reviewed By: Kevin Fourie

KTS-673
"The search algorithm needs some work"
Updated. Search requires php, python and java.

Committed By: Conrad Vermeulen
Reviewed By: Kevin Fourie


git-svn-id: https://kt-dms.svn.sourceforge.net/svnroot/kt-dms/STABLE/trunk@7477 c91229c3-7414-0410-bfa2-8a42b809f60b
Inline Side-by-side
Showing 13 changed files with 269 additions and 84 deletions
config/config.ini
  View file @8186521
... ... @@ -242,7 +242,10 @@ pdftotext = pdftotext
242 242 catppt = catppt
243 243 pstotext = pstotext
244 244 catdoc = catdoc
245   -antiword = antiword.exe
  245 +antiword = antiword
  246 +python = python
  247 +java = java
  248 +php = php
246 249  
247 250 [search]
248 251 ; The number of results per page
... ...
plugins/housekeeper/DiskUsageDashlet.inc.php
  View file @8186521
1 1 <?php
2 2  
3 3 /**
  4 + * $Id
4 5 *
5   - * Copyright (c) 2007 Jam Warehouse http://www.jamwarehouse.com
  6 + * The contents of this file are subject to the KnowledgeTree Public
  7 + * License Version 1.1.2 ("License"); You may not use this file except in
  8 + * compliance with the License. You may obtain a copy of the License at
  9 + * http://www.knowledgetree.com/KPL
6 10 *
7   - * This program is free software; you can redistribute it and/or modify
8   - * it under the terms of the GNU General Public License as published by
9   - * the Free Software Foundation; using version 2 of the License.
  11 + * Software distributed under the License is distributed on an "AS IS"
  12 + * basis, WITHOUT WARRANTY OF ANY KIND, either express or implied.
  13 + * See the License for the specific language governing rights and
  14 + * limitations under the License.
10 15 *
11   - * This program is distributed in the hope that it will be useful,
12   - * but WITHOUT ANY WARRANTY; without even the implied warranty of
13   - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14   - * GNU General Public License for more details.
  16 + * All copies of the Covered Code must include on each user interface screen:
  17 + * (i) the "Powered by KnowledgeTree" logo and
  18 + * (ii) the KnowledgeTree copyright notice
  19 + * in the same form as they appear in the distribution. See the License for
  20 + * requirements.
15 21 *
16   - * You should have received a copy of the GNU General Public License
17   - * along with this program; if not, write to the Free Software
18   - * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
  22 + * The Original Code is: KnowledgeTree Open Source
19 23 *
20   - * -------------------------------------------------------------------------
21   - *
22   - * You can contact the copyright owner regarding licensing via the contact
23   - * details that can be found on the KnowledgeTree web site:
24   - *
25   - * http://www.knowledgetree.com/
  24 + * The Initial Developer of the Original Code is The Jam Warehouse Software
  25 + * (Pty) Ltd, trading as KnowledgeTree.
  26 + * Portions created by The Jam Warehouse Software (Pty) Ltd are Copyright
  27 + * (C) 2007 The Jam Warehouse Software (Pty) Ltd;
  28 + * All Rights Reserved.
  29 + * Contributor( s): ______________________________________
26 30 */
27 31  
28 32 class DiskUsageDashlet extends KTBaseDashlet
... ... @@ -34,7 +38,7 @@ class DiskUsageDashlet extends KTBaseDashlet
34 38  
35 39 function DiskUsageDashlet()
36 40 {
37   - $this->sTitle = _kt('Disk Usage');
  41 + $this->sTitle = _kt('Storage Utilization');
38 42 $this->sClass = "ktInfo";
39 43 }
40 44  
... ...
plugins/housekeeper/FolderUsageDashlet.inc.php
  View file @8186521
1 1 <?php
2 2  
3 3 /**
  4 + * $Id
4 5 *
5   - * Copyright (c) 2007 Jam Warehouse http://www.jamwarehouse.com
  6 + * The contents of this file are subject to the KnowledgeTree Public
  7 + * License Version 1.1.2 ("License"); You may not use this file except in
  8 + * compliance with the License. You may obtain a copy of the License at
  9 + * http://www.knowledgetree.com/KPL
6 10 *
7   - * This program is free software; you can redistribute it and/or modify
8   - * it under the terms of the GNU General Public License as published by
9   - * the Free Software Foundation; using version 2 of the License.
  11 + * Software distributed under the License is distributed on an "AS IS"
  12 + * basis, WITHOUT WARRANTY OF ANY KIND, either express or implied.
  13 + * See the License for the specific language governing rights and
  14 + * limitations under the License.
10 15 *
11   - * This program is distributed in the hope that it will be useful,
12   - * but WITHOUT ANY WARRANTY; without even the implied warranty of
13   - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14   - * GNU General Public License for more details.
  16 + * All copies of the Covered Code must include on each user interface screen:
  17 + * (i) the "Powered by KnowledgeTree" logo and
  18 + * (ii) the KnowledgeTree copyright notice
  19 + * in the same form as they appear in the distribution. See the License for
  20 + * requirements.
15 21 *
16   - * You should have received a copy of the GNU General Public License
17   - * along with this program; if not, write to the Free Software
18   - * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
  22 + * The Original Code is: KnowledgeTree Open Source
19 23 *
20   - * -------------------------------------------------------------------------
21   - *
22   - * You can contact the copyright owner regarding licensing via the contact
23   - * details that can be found on the KnowledgeTree web site:
24   - *
25   - * http://www.knowledgetree.com/
  24 + * The Initial Developer of the Original Code is The Jam Warehouse Software
  25 + * (Pty) Ltd, trading as KnowledgeTree.
  26 + * Portions created by The Jam Warehouse Software (Pty) Ltd are Copyright
  27 + * (C) 2007 The Jam Warehouse Software (Pty) Ltd;
  28 + * All Rights Reserved.
  29 + * Contributor( s): ______________________________________
26 30 */
27 31  
28 32 class FolderUsageDashlet extends KTBaseDashlet
... ... @@ -31,7 +35,7 @@ class FolderUsageDashlet extends KTBaseDashlet
31 35  
32 36 function FolderUsageDashlet()
33 37 {
34   - $this->sTitle = _kt('System Folder Usage');
  38 + $this->sTitle = _kt('System Folder Utilization');
35 39 $this->sClass = "ktInfo";
36 40 }
37 41  
... ...
plugins/housekeeper/HouseKeeperDispatcher.php
  View file @8186521
1 1 <?php
2 2  
  3 +/**
  4 + * $Id
  5 + *
  6 + * The contents of this file are subject to the KnowledgeTree Public
  7 + * License Version 1.1.2 ("License"); You may not use this file except in
  8 + * compliance with the License. You may obtain a copy of the License at
  9 + * http://www.knowledgetree.com/KPL
  10 + *
  11 + * Software distributed under the License is distributed on an "AS IS"
  12 + * basis, WITHOUT WARRANTY OF ANY KIND, either express or implied.
  13 + * See the License for the specific language governing rights and
  14 + * limitations under the License.
  15 + *
  16 + * All copies of the Covered Code must include on each user interface screen:
  17 + * (i) the "Powered by KnowledgeTree" logo and
  18 + * (ii) the KnowledgeTree copyright notice
  19 + * in the same form as they appear in the distribution. See the License for
  20 + * requirements.
  21 + *
  22 + * The Original Code is: KnowledgeTree Open Source
  23 + *
  24 + * The Initial Developer of the Original Code is The Jam Warehouse Software
  25 + * (Pty) Ltd, trading as KnowledgeTree.
  26 + * Portions created by The Jam Warehouse Software (Pty) Ltd are Copyright
  27 + * (C) 2007 The Jam Warehouse Software (Pty) Ltd;
  28 + * All Rights Reserved.
  29 + * Contributor( s): ______________________________________
  30 + */
  31 +
3 32 session_start();
4 33  
5 34 require_once("../../config/dmsDefaults.php");
... ... @@ -88,4 +117,4 @@ class HouseKeeperDispatcher extends KTStandardDispatcher
88 117 $oDispatcher = new HouseKeeperDispatcher();
89 118 $oDispatcher->dispatch();
90 119  
91   -?>
92 120 \ No newline at end of file
  121 +?>
... ...
plugins/housekeeper/HouseKeeperPlugin.php
  View file @8186521
1 1 <?php
2 2  
  3 +/**
  4 + * $Id
  5 + *
  6 + * The contents of this file are subject to the KnowledgeTree Public
  7 + * License Version 1.1.2 ("License"); You may not use this file except in
  8 + * compliance with the License. You may obtain a copy of the License at
  9 + * http://www.knowledgetree.com/KPL
  10 + *
  11 + * Software distributed under the License is distributed on an "AS IS"
  12 + * basis, WITHOUT WARRANTY OF ANY KIND, either express or implied.
  13 + * See the License for the specific language governing rights and
  14 + * limitations under the License.
  15 + *
  16 + * All copies of the Covered Code must include on each user interface screen:
  17 + * (i) the "Powered by KnowledgeTree" logo and
  18 + * (ii) the KnowledgeTree copyright notice
  19 + * in the same form as they appear in the distribution. See the License for
  20 + * requirements.
  21 + *
  22 + * The Original Code is: KnowledgeTree Open Source
  23 + *
  24 + * The Initial Developer of the Original Code is The Jam Warehouse Software
  25 + * (Pty) Ltd, trading as KnowledgeTree.
  26 + * Portions created by The Jam Warehouse Software (Pty) Ltd are Copyright
  27 + * (C) 2007 The Jam Warehouse Software (Pty) Ltd;
  28 + * All Rights Reserved.
  29 + * Contributor( s): ______________________________________
  30 + */
  31 +
3 32 class HouseKeeperPlugin extends KTPlugin
4 33 {
5 34 var $autoRegister = true;
... ... @@ -93,4 +122,4 @@ class HouseKeeperPlugin extends KTPlugin
93 122 $oPluginRegistry =& KTPluginRegistry::getSingleton();
94 123 $oPluginRegistry->registerPlugin('HouseKeeperPlugin', 'ktcore.housekeeper.plugin', __FILE__);
95 124  
96   -?>
97 125 \ No newline at end of file
  126 +?>
... ...
search2/indexing/extractorCore.inc.php
  View file @8186521
... ... @@ -124,10 +124,22 @@ abstract class DocumentExtractor
124 124 }
125 125 $classname=get_class($this);
126 126  
  127 + $sql = "select id as extractor_id from mime_extractors WHERE name='$classname'";
  128 + $rs = DBUtil::getResultArray($sql);
  129 + if (count($rs) == 0)
  130 + {
  131 + $extractor_id = DBUtil::autoInsert('mime_extractors', array('name'=>$classname, 'active'=>1));
  132 + }
  133 + else
  134 + {
  135 + $extractor_id = $rs[0]['extractor_id'];
  136 + }
  137 +
  138 +
127 139 foreach($types as $type)
128 140 {
129   - $sql = "update mime_types set extractor='$classname' where mimetypes='$type' and extractor is null";
130   - DBUtil::runQuery($sql);
  141 + $sql = "update mime_types set extractor_id=$extractor_id where mimetypes='$type' and extractor_id is null";
  142 + $rs = DBUtil::runQuery($sql);
131 143 }
132 144 }
133 145  
... ... @@ -510,15 +522,17 @@ abstract class CompositeExtractor extends DocumentExtractor
510 522 public function extractTextContent()
511 523 {
512 524 $intermediateFile = $this->targetfile . '.' . $this->targetExtension;
  525 + touch($intermediateFile);
513 526  
514 527 $this->sourceExtractor->setSourceFile($this->sourcefile);
515 528 $this->sourceExtractor->setTargetFile($intermediateFile);
516 529 $this->sourceExtractor->setMimeType($this->mimetype);
517 530 $this->sourceExtractor->setExtension($this->extension);
518   - if ($this->sourceExtractor->extractTextContent())
  531 + if (!$this->sourceExtractor->extractTextContent())
519 532 {
520 533 return false;
521 534 }
  535 + $intermediateFile = $this->sourceExtractor->getTargetFile();
522 536  
523 537 $this->targetExtractor->setSourceFile($intermediateFile);
524 538 $this->targetExtractor->setTargetFile($this->targetfile);
... ...
search2/indexing/extractors/OOPresentationExtractor.inc.php
  View file @8186521
1 1 <?php
2 2  
3   -require_once('OOPDFTextExtractor.inc.php');
  3 +require_once('OOTextExtractor.inc.php');
  4 +require_once('PDFExtractor.inc.php');
4 5  
5   -class OOPresentationExtractor extends OOPDFTextExtractor
  6 +class OOPresentationExtractor extends CompositeExtractor
6 7 {
  8 + public function __construct()
  9 + {
  10 + $sourceExtractor = new OOPresentationToPDF();
  11 + $targetExtractor = new PDFExtractor();
  12 + parent::__construct($sourceExtractor, 'pdf', 'application/pdf', $targetExtractor, true);
  13 + }
  14 +
7 15 public function getDisplayName()
8 16 {
9 17 return _kt('OpenOffice Presentation Extractor');
... ... @@ -18,4 +26,18 @@ class OOPresentationExtractor extends OOPDFTextExtractor
18 26 }
19 27 }
20 28  
  29 +class OOPresentationToPDF extends OOTextExtractor
  30 +{
  31 + public function __construct()
  32 + {
  33 + parent::__construct('pdf');
  34 + $this->documentConverter = KT_DIR . '/bin/openoffice/pdfgen.py';
  35 + if (!is_file($this->documentConverter))
  36 + {
  37 + $this->documentConverter = false;
  38 + }
  39 + }
  40 +}
  41 +
  42 +
21 43 ?>
22 44 \ No newline at end of file
... ...
search2/indexing/extractors/OOSpreadsheetExtractor.inc.php
  View file @8186521
1 1 <?php
2 2  
3   -require_once('OOPDFTextExtractor.inc.php');
  3 +require_once('OOTextExtractor.inc.php');
4 4  
5   -class OOSpreadsheetExtractor extends OOPDFTextExtractor
  5 +class OOSpreadsheetExtractor extends OOTextExtractor
6 6 {
7 7 public function getDisplayName()
8 8 {
... ...
search2/indexing/extractors/OOTextExtractor.inc.php
  View file @8186521
... ... @@ -2,22 +2,27 @@
2 2  
3 3 class OOTextExtractor extends ExternalDocumentExtractor
4 4 {
5   - private $converter;
6   - private $javaPath;
7   - private $ooHost;
8   - private $ooPort;
9   - private $targetMimeType;
  5 + protected $python;
  6 + protected $documentConverter;
  7 + protected $ooHost;
  8 + protected $ooPort;
  9 + protected $targetExtension;
10 10  
11   - public function __construct($targetMimeType='plain/text')
  11 + public function __construct($targetExtension='html')
12 12 {
13 13 parent::__construct();
  14 + $this->targetExtension = $targetExtension;
14 15 $config =& KTConfig::getSingleton();
15 16  
16   - $this->converter = KTUtil::findCommand('extractors/jodconverter', 'jodconverter');
17   - $this->javaPath = KTUtil::findCommand('extractors/java', 'java');
18   - $this->ooHost = $config->get('openoffice/host', 'localhost');
19   - $this->ooPort = $config->get('openoffice/port', 8100);
20   - $this->targetMimeType = $targetMimeType;
  17 + $this->python = KTUtil::findCommand('externalBinary/python');
  18 + $this->ooHost = $config->get('openoffice/host');
  19 + $this->ooPort = $config->get('openoffice/port');
  20 +
  21 + $this->documentConverter = KT_DIR . '/bin/openoffice/DocumentConverter.py';
  22 + if (!is_file($this->documentConverter))
  23 + {
  24 + $this->documentConverter = false;
  25 + }
21 26 }
22 27  
23 28 public function getDisplayName()
... ... @@ -28,53 +33,78 @@ class OOTextExtractor extends ExternalDocumentExtractor
28 33 public function getSupportedMimeTypes()
29 34 {
30 35 return array(
31   - 'text/rtf',
32   - 'application/vnd.oasis.opendocument.text',
33   - 'application/vnd.oasis.opendocument.text-template',
34   - 'application/vnd.oasis.opendocument.text-web',
35   - 'application/vnd.oasis.opendocument.text-master',
36   - 'application/vnd.sun.xml.writer',
37   - 'application/vnd.sun.xml.writer.template',
38   - 'application/vnd.sun.xml.writer.global',
  36 +
39 37 );
40 38 }
41 39  
42 40 public function needsIntermediateSourceFile()
43 41 {
44 42 // we need the intermediate file because it
45   - // has the correct extension. jodconverter uses the extension to determine mimetype
  43 + // has the correct extension. documentConverter uses the extension to determine mimetype
46 44 return true;
47 45 }
48 46  
49 47 protected function getCommandLine()
50 48 {
51   - $cmdline = "$this->javaPath -jar $this->converter $this->sourcefile $this->mimetype $this->targetfile $this->targetMimeType $this->ooHost $this->ooPort";
  49 + $sourcefile = escapeshellcmd($this->sourcefile);
  50 + unlink($this->targetfile);
  51 + $this->targetfile .= '.' . $this->targetExtension;
  52 + $targetfile = escapeshellcmd($this->targetfile);
  53 +
  54 + $escape = OS_WINDOWS?'"':'\'';
  55 +
  56 + $cmdline = "{$this->python} {$escape}{$this->documentConverter}{$escape} {$escape}{$sourcefile}{$escape} {$escape}{$targetfile}{$escape} {$this->ooHost} {$this->ooPort}";
52 57 return $cmdline;
53 58 }
54 59  
55   - public function diagnose()
  60 + protected function filter($text)
56 61 {
57   - if (false === $this->converter)
  62 + $text = preg_replace ("@(</?[^>]*>)+@", '', $text);
  63 +
  64 + do
  65 + {
  66 + $old = $text;
  67 +
  68 + $text= preg_replace("@([\r\n])[\s]+@",'\1', $text);
  69 +
  70 + $text = preg_replace('@\ \ @',' ', $text);
  71 + $text = preg_replace("@\n\n@","\n", $text);
  72 + }
  73 + while ($old != $text);
  74 +
  75 + return $text;
  76 + }
  77 +
  78 + public function extractTextContent()
  79 + {
  80 + if (false === parent::extractTextContent())
58 81 {
59   - return _kt('Cannot locate jodconverter');
  82 + return false;
60 83 }
61 84  
62   - if (false === $this->javaPath)
  85 + if ($this->targetExtension != 'html')
63 86 {
64   - return _kt('Cannot locate java');
  87 + return true;
65 88 }
  89 + $content = file_get_contents($this->targetfile);
  90 + return file_put_contents($this->targetfile, $this->filter($content));
66 91  
  92 + }
67 93  
68 94  
69   - $connection = @fsockopen($this->ooHost, $this->ooPort,$errno, $errstr,5 );
70   - if (false === $connection)
  95 + public function diagnose()
  96 + {
  97 + if (false === $this->python)
71 98 {
72   - return _kt('Cannot connect to openoffice host');
  99 + return _kt('Cannot locate python');
73 100 }
74   - fclose($connection);
75 101  
  102 + if (false === $this->documentConverter)
  103 + {
  104 + return _kt('Cannot locate DocumentConverter.py');
  105 + }
76 106  
77   - return null;
  107 + return SearchHelper::checkOpenOfficeAvailablity();
78 108 }
79 109 }
80 110  
... ...
search2/indexing/extractors/PDFExtractor.inc.php
  View file @8186521
... ... @@ -4,7 +4,7 @@ class PDFExtractor extends ApplicationExtractor
4 4 {
5 5 public function __construct()
6 6 {
7   - parent::__construct('externalBinary','pdftotext','pdftotext',_kt('PDF Text Extractor'),'-nopgbrk -enc UTF-8 {source} {target}');
  7 + parent::__construct('externalBinary','pdftotext','pdftotext',_kt('PDF Text Extractor'),'-nopgbrk -enc UTF-8 \'{source}\' \'{target}\'');
8 8 }
9 9  
10 10 public function getSupportedMimeTypes()
... ...
search2/indexing/extractors/MailMimeExtractor.inc.php renamed to search2/indexing/extractors/RTFExtractor.inc.php
  View file @8186521
1 1 <?php
2 2  
3   -class MailMimeExtractor extends TextExtractor
  3 +require_once('OOTextExtractor.inc.php');
  4 +
  5 +class RTFExtractor extends OOTextExtractor
4 6 {
5 7 public function getDisplayName()
6 8 {
7   - return _kt('Mail Mime Extractor');
  9 + return _kt('RTF Extractor');
8 10 }
9 11  
10 12 public function getSupportedMimeTypes()
11 13 {
12   - return array('text/msg');
  14 + return array(
  15 + 'text/rtf'
  16 + );
13 17 }
14   -
15 18 }
16 19  
  20 +
17 21 ?>
18 22 \ No newline at end of file
... ...
search2/indexing/extractors/XMLExtractor.inc.php
  View file @8186521
... ... @@ -9,7 +9,7 @@ class XMLExtractor extends TextExtractor
9 9  
10 10 public function getSupportedMimeTypes()
11 11 {
12   - return array('text/xml','application/xml','text/html');
  12 + return array('text/xml','application/xml','text/html','text/enriched');
13 13 }
14 14  
15 15 protected function filter($text)
... ...
search2/indexing/indexerCore.inc.php
  View file @8186521
... ... @@ -330,7 +330,11 @@ abstract class Indexer
330 330 public function clearExtractors()
331 331 {
332 332 global $default;
333   - $sql = "update mime_types set extractor=null";
  333 +
  334 + $sql = "update mime_types set extractor_id=null";
  335 + DBUtil::runQuery($sql);
  336 +
  337 + $sql = "delete from mime_extractors";
334 338 DBUtil::runQuery($sql);
335 339  
336 340 $default->log->debug('clearExtractors');
... ... @@ -616,6 +620,43 @@ abstract class Indexer
616 620 }
617 621  
618 622 /**
  623 + * This does the initial mime type association between mime types and text extractors
  624 + *
  625 + */
  626 + public function checkForRegisteredTypes()
  627 + {
  628 + global $default;
  629 +
  630 + // we are only doing this once!
  631 + $initRegistered = KTUtil::getSystemSetting('mimeTypesRegistered', false);
  632 + if ($initRegistered)
  633 + {
  634 + return;
  635 + }
  636 + $default->log->info('checkForRegisteredTypes: start');
  637 +
  638 + $this->registerTypes(true);
  639 +
  640 +
  641 + $disable = array(
  642 + OS_WINDOWS=>array('PSExtractor'),
  643 + OS_UNIX => array()
  644 +
  645 + );
  646 +
  647 + foreach($disable[OS_WINDOWS] as $extractor)
  648 + {
  649 + $sql = "UPDATE mime_extractors SET active=0 WHERE name='$extractor'";
  650 + DBUtil::runQuery($sql);
  651 + $default->log->info("checkForRegisteredTypes: disabled '$extractor'");
  652 + }
  653 +
  654 + $default->log->info('checkForRegisteredTypes: done');
  655 + KTUtil::setSystemSetting('mimeTypesRegistered', true);
  656 + }
  657 +
  658 +
  659 + /**
619 660 * The main function that may be called repeatedly to index documents.
620 661 *
621 662 * @param int $max Default 20
... ... @@ -624,6 +665,8 @@ abstract class Indexer
624 665 {
625 666 global $default;
626 667  
  668 + $this->checkForRegisteredTypes();
  669 +
627 670 $default->log->info('indexDocuments: start');
628 671 if (!$this->doesDiagnosticsPass())
629 672 {
... ... @@ -798,6 +841,9 @@ abstract class Indexer
798 841  
799 842 if ($extractor->extractTextContent())
800 843 {
  844 + // the extractor may need to create another target file
  845 + $targetFile = $extractor->getTargetFile();
  846 +
801 847 $extractor->setExtractionStatus(true);
802 848 $this->executeHook($extractor, 'pre_index');
803 849 $this->executeHook($extractor, 'pre_index', $mimeType);
... ...