gnh1201/paperless
1
0
Fork 0
mirror of https://codeberg.org/SeedDMS/paperless synced 2025-02-06 06:55:28 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

improve search for similar documents (extend or reduce the list of search terms)

Browse Source
This commit is contained in:
Uwe Steinmann 2023-03-15 18:05:28 +01:00
parent 16a3083f33
commit 2580c56a09
1 changed files with 35 additions and 13 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

48
class.paperless.php
View File

@ -584,8 +584,9 @@ class SeedDMS_ExtPaperless_RestAPI_Controller { /* {{{ */
* to the classification. * to the classification.
* The '.' is added as valid character in a word, because solr's * The '.' is added as valid character in a word, because solr's
* standard tokenizer treats it as a valid char as well. * standard tokenizer treats it as a valid char as well.
* But sqlitefts treats '.' as a separator
*/ */
$wordcount = self::mb_word_count($shortcontent, MB_CASE_LOWER, '.'); $wordcount = self::mb_word_count($shortcontent, MB_CASE_LOWER, '');
arsort($wordcount); arsort($wordcount);
$newquery = []; $newquery = [];
foreach($wordcount as $word=>$n) { foreach($wordcount as $word=>$n) {
@ -594,21 +595,42 @@ class SeedDMS_ExtPaperless_RestAPI_Controller { /* {{{ */
} }
// echo implode(' ', $newquery); // echo implode(' ', $newquery);
$logger->log("Query for '".implode(' ', $newquery)."'", PEAR_LOG_DEBUG); $logger->log("Query for '".implode(' ', $newquery)."'", PEAR_LOG_DEBUG);
$searchresult = $lucenesearch->search(implode(' ', $newquery), array('record_type'=>['document'], 'status'=>[2], 'user'=>[$userobj->getLogin()], 'startFolder'=>$startfolder, 'rootFolder'=>$rootfolder), array('limit'=>$limit, 'offset'=>$offset), $order); /* $newquery is empty if the document doesn't have a fulltext.
if($searchresult) { * In that case it makes no sense to search for similar documents
$recs = array(); * Otherwise search for documents with newquery, but if doesn't yield
if($searchresult['hits']) { * a result, short the newquery by the last term and try again until
$allids = ''; * newquery is void
foreach($searchresult['hits'] as $hit) { */
if($hit['document_id'][0] == 'D') { while($newquery) {
if($tmp = $dms->getDocument((int) substr($hit['document_id'], 1))) { $searchresult = $lucenesearch->search(implode(' ', $newquery), array('record_type'=>['document'], 'status'=>[2], 'user'=>[$userobj->getLogin()], 'startFolder'=>$startfolder, 'rootFolder'=>$rootfolder), array('limit'=>$limit, 'offset'=>$offset), $order);
$allids .= $hit['document_id'].' '; if($searchresult) {
$recs[] = $this->__getDocumentData($tmp, $truncate_content); $recs = array();
if($searchresult['hits']) {
$allids = '';
foreach($searchresult['hits'] as $hit) {
if(($hit['document_id'][0] == 'D') && ($hit['document_id'] != 'D'.((int)$params['more_like_id']))) {
if($tmp = $dms->getDocument((int) substr($hit['document_id'], 1))) {
$allids .= $hit['document_id'].' ';
$recs[] = $this->__getDocumentData($tmp, $truncate_content);
}
} else {
$searchresult['count']--;
} }
} }
$logger->log('Result is '.$allids, PEAR_LOG_DEBUG);
if($recs)
return $response->withJson(array('count'=>$searchresult['count'], 'next'=>null, 'previous'=>null, 'offset'=>$offset, 'limit'=>$limit, 'results'=>$recs), 200);
else {
/* Still nothing found, so try a shorter query */
array_pop($newquery);
}
} else {
/* Still nothing found, so try a shorter query */
array_pop($newquery);
} }
$logger->log('Result is '.$allids, PEAR_LOG_DEBUG); } else {
return $response->withJson(array('count'=>$searchresult['count'], 'next'=>null, 'previous'=>null, 'offset'=>$offset, 'limit'=>$limit, 'results'=>$recs), 200); /* Quit the while loop right away, if the search failed */
$newquery = false;
} }
} }
} }