From 4c6d047bb81e76e0c6d181c19a63ac631babc72b Mon Sep 17 00:00:00 2001
From: Uwe Steinmann <steinm@debian.org>
Date: Fri, 16 May 2025 07:46:49 +0200
Subject: [PATCH] copy most of the search functions into own class

---
 inc/inc.Utils.php | 601 +++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 598 insertions(+), 3 deletions(-)

diff --git a/inc/inc.Utils.php b/inc/inc.Utils.php
index 049035396..fbf1b6098 100644
--- a/inc/inc.Utils.php
+++ b/inc/inc.Utils.php
@@ -1399,16 +1399,611 @@ class SeedDMS_FolderTree { /* {{{ */
 class SeedDMS_Search { /* {{{ */
 	protected $dms;
 
+	protected $user;
+
+	protected $fulltextservice;
+
 	protected $settings;
 
-	public function __construct($dms, $settings) {
+	public $searchparams;
+
+	protected $dcount;
+
+	protected $fcount;
+
+	protected $totalPages;
+
+	protected $entries;
+
+	protected $terms;
+
+	protected $searchTime;
+
+	public function __construct($dms, $user, $fulltextservice, $settings) {
 		$this->dms = $dms;
+		$this->user = $user;
+		$this->fulltextservice = $fulltextservice;
 		$this->settings = $settings;
+		$this->searchparams = [];
+		$this->dcount = 0;
+		$this->fcount = 0;
+		$this->totalPages = 0;
+		$this->entries = array();
+		$this->terms = array();
+		$this->searchTime = 0;
 	}
 
-	public function createSearchFromQuery($urlquery) { /* {{{ */
+	protected function getTime() { /* {{{ */
+		if (function_exists('microtime')) {
+			$tm = microtime();
+			$tm = explode(' ', $tm);
+			return (float) sprintf('%f', $tm[1] + $tm[0]);
+		}
+		return time();
 	} /* }}} */
 
-	public function search() { /* {{{ */
+	public function createSearchFromQuery($get) { /* {{{ */
+
+		/* Creation date {{{ */
+		$createstartts = null;
+		$createstartdate = null;
+		$createendts = null;
+		$createenddate = null;
+		$created['from'] = null;
+		$created['to'] = null;
+		if(!empty($get["created"]["from"])) {
+			$createstartts = makeTsFromDate($get["created"]["from"]);
+			$createstartdate = array('year'=>(int)date('Y', $createstartts), 'month'=>(int)date('m', $createstartts), 'day'=>(int)date('d', $createstartts), 'hour'=>0, 'minute'=>0, 'second'=>0);
+			if (!checkdate($createstartdate['month'], $createstartdate['day'], $createstartdate['year'])) {
+				UI::exitError(getMLText("search"),getMLText("invalid_create_date_end"));
+			}
+			$created['from'] = $createstartts;
+		}
+		if(!empty($get["created"]["to"])) {
+			$createendts = makeTsFromDate($get["created"]["to"]);
+			$createenddate = array('year'=>(int)date('Y', $createendts), 'month'=>(int)date('m', $createendts), 'day'=>(int)date('d', $createendts), 'hour'=>23, 'minute'=>59, 'second'=>59);
+			if (!checkdate($createenddate['month'], $createenddate['day'], $createenddate['year'])) {
+				UI::exitError(getMLText("search"),getMLText("invalid_create_date_end"));
+			}
+			$created['to'] = $createendts;
+		}
+		$this->searchparams['created'] = $created;
+		/* }}} */
+
+		/* Modification date {{{ */
+		$modifystartts = null;
+		$modifystartdate = null;
+		$modifyendts = null;
+		$modifyenddate = null;
+		$modified['from'] = null;
+		$modified['to'] = null;
+		if(!empty($get["modified"]["from"])) {
+			$modifystartts = makeTsFromDate($get["modified"]["from"]);
+			$modifystartdate = array('year'=>(int)date('Y', $modifystartts), 'month'=>(int)date('m', $modifystartts), 'day'=>(int)date('d', $modifystartts), 'hour'=>0, 'minute'=>0, 'second'=>0);
+			if (!checkdate($modifystartdate['month'], $modifystartdate['day'], $modifystartdate['year'])) {
+				UI::exitError(getMLText("search"),getMLText("invalid_modification_date_end"));
+			}
+			$modified['from'] = $modifystartts;
+		}
+		if(!empty($get["modified"]["to"])) {
+			$modifyendts = makeTsFromDate($get["modified"]["to"]);
+			$modifyenddate = array('year'=>(int)date('Y', $modifyendts), 'month'=>(int)date('m', $modifyendts), 'day'=>(int)date('d', $modifyendts), 'hour'=>23, 'minute'=>59, 'second'=>59);
+			if (!checkdate($modifyenddate['month'], $modifyenddate['day'], $modifyenddate['year'])) {
+				UI::exitError(getMLText("search"),getMLText("invalid_modification_date_end"));
+			}
+			$modified['to'] = $modifyendts;
+		}
+		$this->searchparams['modified'] = $modified;
+		/* }}} */
+
+		/* Filesize {{{ */
+		$filesizestart = 0;
+		$filesizeend = 0;
+		$filesize['from'] = null;
+		$filesize['to'] = null;
+		if(!empty($get["filesize"]["from"])) {
+			$filesizestart = $get["filesize"]["from"];
+			$filesize['from'] = $get["filesize"]["from"];
+		}
+		if(!empty($get["filesize"]["to"])) {
+			$filesizeend = $get["filesize"]["to"];
+			$filesize['to'] = $get["filesize"]["to"];
+		}
+		$this->searchparams['filesize'] = $filesize;
+		/* }}} */
+
+		// Check to see if the search has been restricted to a particular
+		// document owner.
+		// $get['owner'] can be a name of an array of names or ids {{{
+		$owner = [];
+		$ownernames = []; // Needed by fulltext search
+		$ownerobjs = []; // Needed by database search
+		if(!empty($get["owner"])) {
+			$owner = $get['owner'];
+			if (!is_array($get['owner'])) {
+				if(is_numeric($get['owner']))
+					$o = $dms->getUser($get['owner']);
+				else
+					$o = $dms->getUserByLogin($get['owner']);
+				if($o) {
+					$ownernames[] = $o->getLogin();
+					$ownerobjs[] = $o;
+				}
+			} else {
+				foreach($get["owner"] as $l) {
+					if($l) {
+						if(is_numeric($l))
+							$o = $dms->getUser($l);
+						else
+							$o = $dms->getUserByLogin($l);
+						if($o) {
+							$ownernames[] = $o->getLogin();
+							$ownerobjs[] = $o;
+						}
+					}
+				}
+			}
+		}
+		$this->searchparams['ownernames'] = $ownernames;
+		$this->searchparams['ownerobjs'] = $ownerobjs;
+		/* }}} */
+
+		// category {{{
+		$categories = array();
+		$categorynames = array();
+		$category = array();
+		if(isset($get['category']) && $get['category']) {
+			$category = $get['category'];
+			foreach($get['category'] as $catid) {
+				if($catid) {
+					if(is_numeric($catid)) {
+						if($cat = $dms->getDocumentCategory($catid)) {
+							$categories[] = $cat;
+							$categorynames[] = $cat->getName();
+						}
+					} else {
+						$categorynames[] = $catid;
+					}
+				}
+			}
+		}
+		$this->searchparams['categorynames'] = $categorynames;
+		$this->searchparams['categories'] = $categories;
+		/* }}} */
+
+		if (isset($get["orderby"]) && is_string($get["orderby"])) {
+			$orderby = $get["orderby"];
+		} else {
+			$orderby = "";
+		}
+		$this->searchparams['orderby'] = $orderby;
+
+		$limit = (isset($get["limit"]) && is_numeric($get["limit"])) ? (int) $get['limit'] : 20;
+		$this->searchparams['limit'] = $limit;
+		$fullsearch = ((!isset($get["fullsearch"]) && $this->settings->_defaultSearchMethod == 'fulltext') || !empty($get["fullsearch"])) && $this->settings->_enableFullSearch;
+		$this->searchparams['fullsearch'] = $fullsearch;
+		$facetsearch = !empty($get["facetsearch"]) && $this->settings->_enableFullSearch;
+		$this->searchparams['facetsearch'] = $facetsearch;
+
+		if (isset($get["query"]) && is_string($get["query"])) {
+			$query = $get["query"];
+		} else {
+			$query = "";
+		}
+		$this->searchparams['query'] = $query;
+
+		// Check to see if the search has been restricted to a particular
+		// mimetype. {{{
+		$mimetype = [];
+		if (isset($get["mimetype"])) {
+			if (!is_array($get['mimetype'])) {
+				if(!empty($get['mimetype']))
+					$mimetype[] = $get['mimetype'];
+			} else {
+				foreach($get["mimetype"] as $l) {
+					if($l)
+						$mimetype[] = $l;
+				}
+			}
+		}
+		$this->searchparams['mimetype'] = $mimetype;
+		/* }}} */
+
+		// status
+		$status = isset($get['status']) ? $get['status'] : array();
+		$this->searchparams['status'] = $status;
+
+		// Get the page number to display. If the result set contains more than
+		// 25 entries, it is displayed across multiple pages.
+		//
+		// This requires that a page number variable be used to track which page the
+		// user is interested in, and an extra clause on the select statement.
+		//
+		// Default page to display is always one.
+		$pageNumber=1;
+		if (isset($get["pg"])) {
+			if (is_numeric($get["pg"]) && $get["pg"]>0) {
+				$pageNumber = (int) $get["pg"];
+			}
+			elseif (!strcasecmp($get["pg"], "all")) {
+				$pageNumber = "all";
+			}
+		}
+		$this->searchparams['pageNumber'] = $pageNumber;
+
+		if($fullsearch) {
+			// Search in Fulltext {{{
+
+			// record_type
+			if(isset($get['record_type']))
+				$record_type = $get['record_type'];
+			else
+				$record_type = array();
+			$this->searchparams['record_type'] = $record_type;
+
+			if (isset($get["attributes"]))
+				$attributes = $get["attributes"];
+			else
+				$attributes = array();
+
+			foreach($attributes as $an=>&$av) {
+				if(substr($an, 0, 5) == 'attr_') {
+					$tmp = explode('_', $an);
+					if($attrdef = $dms->getAttributeDefinition($tmp[1])) {
+						switch($attrdef->getType()) {
+						/* Turn dates into timestamps */
+						case SeedDMS_Core_AttributeDefinition::type_date:
+							foreach(['from', 'to'] as $kk)
+								if(!empty($av[$kk])) {
+									if(!is_numeric($av[$kk])) {
+										$av[$kk] = makeTsFromDate($av[$kk]);
+									}
+								}
+							break;
+						}
+					}
+				}
+			}
+			$this->searchparams['attributes'] = $attributes;
+
+			/* Create $order array for fulltext search */
+			$order = ['by'=>'', 'dir'=>''];
+			switch($orderby) {
+			case 'dd':
+				$order = ['by'=>'created', 'dir'=>'desc'];
+				break;
+			case 'd':
+				$order = ['by'=>'created', 'dir'=>'asc'];
+				break;
+			case 'nd':
+				$order = ['by'=>'title', 'dir'=>'desc'];
+				break;
+			case 'n':
+				$order = ['by'=>'title', 'dir'=>'asc'];
+				break;
+			case 'id':
+				$order = ['by'=>'id', 'dir'=>'desc'];
+				break;
+			case 'i':
+				$order = ['by'=>'id', 'dir'=>'asc'];
+				break;
+			default:
+				$order = ['by'=>'', 'dir'=>''];
+			}
+			$this->searchparams['order'] = $order;
+
+			// Check to see if the search has been restricted to a particular sub-tree in
+			// the folder hierarchy.
+			$startFolder = null;
+			if (isset($get["folderfullsearchid"]) && is_numeric($get["folderfullsearchid"]) && $get["folderfullsearchid"]>0) {
+				$targetid = $get["folderfullsearchid"];
+				$startFolder = $this->dms->getFolder($targetid);
+				if (!is_object($startFolder)) {
+					UI::exitError(getMLText("search"),getMLText("invalid_folder_id"));
+				}
+			}
+			$this->searchparams['startFolder'] = $startFolder;
+
+			$rootFolder = $this->dms->getFolder($this->settings->_rootFolderID);
+			$this->searchparams['rootFolder'] = $rootFolder;
+
+			// }}}
+		} else {
+			// Search in Database {{{
+
+			/* Select if only documents (0x01), only folders (0x02) or both (0x03)
+			 * are found
+			 */
+			$resultmode = 0x03;
+			if (isset($get["resultmode"]) && is_numeric($get["resultmode"])) {
+					$resultmode = $get['resultmode'];
+			}
+			$this->searchparams['resultmode'] = $resultmode;
+
+			$mode = "AND";
+			if (isset($get["mode"]) && is_numeric($get["mode"]) && $get["mode"]==0) {
+					$mode = "OR";
+			}
+			$this->searchparams['mode'] = $mode;
+
+			$searchin = array();
+			if (isset($get['searchin']) && is_array($get["searchin"])) {
+				foreach ($get["searchin"] as $si) {
+					if (isset($si) && is_numeric($si)) {
+						switch ($si) {
+							case 1: // keywords
+							case 2: // name
+							case 3: // comment
+							case 4: // attributes
+							case 5: // id
+								$searchin[$si] = $si;
+								break;
+						}
+					}
+				}
+			}
+
+			// if none is checkd search all
+			if (count($searchin)==0) $searchin=array(1, 2, 3, 4, 5);
+			$this->searchparams['searchin'] = $searchin;
+
+			// Check to see if the search has been restricted to a particular sub-tree in
+			// the folder hierarchy.
+			if (isset($get["targetid"]) && is_numeric($get["targetid"]) && $get["targetid"]>0) {
+				$targetid = $get["targetid"];
+				$startFolder = $this->dms->getFolder($targetid);
+			}
+			else {
+				$startFolder = $this->dms->getRootFolder();
+			}
+			if (!is_object($startFolder)) {
+				UI::exitError(getMLText("search"),getMLText("invalid_folder_id"));
+			}
+			$this->searchparams['startFolder'] = $startFolder;
+
+			/* Status date {{{ */
+			$statusstartdate = array();
+			$statusenddate = array();
+			if(!empty($get["statusdatestart"])) {
+				$statusstartts = makeTsFromDate($get["statusdatestart"]);
+				$statusstartdate = array('year'=>(int)date('Y', $statusstartts), 'month'=>(int)date('m', $statusstartts), 'day'=>(int)date('d', $statusstartts), 'hour'=>0, 'minute'=>0, 'second'=>0);
+			}
+			$this->searchparams['statusstartdate'] = $statusstartdate;
+			if ($statusstartdate && !checkdate($statusstartdate['month'], $statusstartdate['day'], $statusstartdate['year'])) {
+				UI::exitError(getMLText("search"),getMLText("invalid_status_date_start"));
+			}
+
+			if(!empty($get["statusdateend"])) {
+				$statusendts = makeTsFromDate($get["statusdateend"]);
+				$statusenddate = array('year'=>(int)date('Y', $statusendts), 'month'=>(int)date('m', $statusendts), 'day'=>(int)date('d', $statusendts), 'hour'=>23, 'minute'=>59, 'second'=>59);
+			}
+			if ($statusenddate && !checkdate($statusenddate['month'], $statusenddate['day'], $statusenddate['year'])) {
+				UI::exitError(getMLText("search"),getMLText("invalid_status_date_end"));
+			}
+			$this->searchparams['statusenddate'] = $statusenddate;
+			/* }}} */
+
+			/* Expiration date {{{ */
+			$expstartdate = array();
+			$expenddate = array();
+			if(!empty($get["expirationstart"])) {
+				$expstartts = makeTsFromDate($get["expirationstart"]);
+				$expstartdate = array('year'=>(int)date('Y', $expstartts), 'month'=>(int)date('m', $expstartts), 'day'=>(int)date('d', $expstartts), 'hour'=>0, 'minute'=>0, 'second'=>0);
+				if (!checkdate($expstartdate['month'], $expstartdate['day'], $expstartdate['year'])) {
+					UI::exitError(getMLText("search"),getMLText("invalid_expiration_date_start"));
+				}
+			}
+			if(!empty($get["expirationend"])) {
+				$expendts = makeTsFromDate($get["expirationend"]);
+				$expenddate = array('year'=>(int)date('Y', $expendts), 'month'=>(int)date('m', $expendts), 'day'=>(int)date('d', $expendts), 'hour'=>23, 'minute'=>59, 'second'=>59);
+				if (!checkdate($expenddate['month'], $expenddate['day'], $expenddate['year'])) {
+					UI::exitError(getMLText("search"),getMLText("invalid_expiration_date_end"));
+				}
+			}
+			/* }}} */
+
+			/* Do not search for folders if result shall be filtered by status.
+			 * If this is not done, unexplainable results will be delivered.
+			 * e.g. a search for expired documents of a given user will list
+			 * also all folders of that user because the status doesn't apply
+			 * to folders.
+			 */
+		//	if($status)
+		//		$resultmode = 0x01;
+
+			if (isset($get["attributes"]))
+				$attributes = $get["attributes"];
+			else
+				$attributes = array();
+
+			foreach($attributes as $attrdefid=>$attribute) {
+				$attrdef = $this->dms->getAttributeDefinition($attrdefid);
+				if($attribute) {
+					if($attrdef->getType() == SeedDMS_Core_AttributeDefinition::type_date) {
+						if(is_array($attribute)) {
+							if(!empty($attributes[$attrdefid]['from']))
+								$attributes[$attrdefid]['from'] = date('Y-m-d', makeTsFromDate($attribute['from']));
+							if(!empty($attributes[$attrdefid]['to']))
+								$attributes[$attrdefid]['to'] = date('Y-m-d', makeTsFromDate($attribute['to']));
+						} else {
+							$attributes[$attrdefid] = date('Y-m-d', makeTsFromDate($attribute));
+						}
+					}
+				}
+			}
+			$this->searchparams['attributes'] = $attributes;
+
+		// }}}
+		}
+		return $this->searchparams;
 	} /* }}} */
+
+	public function setSearchParam($name, $value) {
+		$this->searchparams[$name] = $value;
+	}
+
+	public function search() { /* {{{ */
+		if($this->searchparams['fullsearch']) {
+			if($this->settings->_fullSearchEngine == 'lucene') {
+				Zend_Search_Lucene_Search_QueryParser::setDefaultEncoding('utf-8');
+			}
+
+			$startTime = $this->getTime();
+			$total = 0;
+			$index = $this->fulltextservice->Indexer();
+			if($index) {
+				if(!empty($this->settings->_suggestTerms) && !empty($get['query'])) {
+					$st = preg_split("/[\s,]+/", trim($get['query']));
+					if($lastterm = end($st))
+						$this->terms = $index->terms($lastterm, $this->settings->_suggestTerms);
+				}
+				$limit = $this->searchparams['limit'];
+				$lucenesearch = $this->fulltextservice->Search();
+				$searchresult = $lucenesearch->search($this->searchparams['query'],
+					array(
+						'record_type'=>$this->searchparams['record_type'],
+						'owner'=>$this->searchparams['ownernames'],
+						'status'=>$this->searchparams['status'],
+						'category'=>$this->searchparams['categorynames'],
+						'user'=>$this->user->isAdmin() ? [] : [$this->user->getLogin()],
+						'mimetype'=>$this->searchparams['mimetype'],
+						'startFolder'=>$this->searchparams['startFolder'],
+						'rootFolder'=>$this->searchparams['rootFolder'],
+						'created_start'=>$this->searchparams['created']['from'],
+						'created_end'=>$this->searchparams['created']['to'],
+						'modified_start'=>$this->searchparams['modified']['from'],
+						'modified_end'=>$this->searchparams['modified']['to'],
+						'filesize_start'=>$this->searchparams['filesize']['from'],
+						'filesize_end'=>$this->searchparams['filesize']['to'],
+						'attributes'=>$this->searchparams['attributes']
+					), ($this->searchparams['pageNumber'] == 'all' ? array() : array('limit'=>$limit, 'offset'=>$limit * ($this->searchparams['pageNumber']-1))), $this->searchparams['order']);
+				if($searchresult !== false) {
+					$entries = array();
+					$facets = $searchresult['facets'];
+					$stats = $searchresult['stats'] ?? null;
+					$dcount = 0;
+					$fcount = 0;
+					if($searchresult['hits']) {
+						foreach($searchresult['hits'] as $hit) {
+							if($hit['document_id'][0] == 'D') {
+								if($tmp = $this->dms->getDocument(substr($hit['document_id'], 1))) {
+	//								if($tmp->getAccessMode($user) >= M_READ) {
+										$tmp->verifyLastestContentExpriry();
+										$entries[] = $tmp;
+										$dcount++;
+	//								}
+								}
+							} elseif($hit['document_id'][0] == 'F') {
+								if($tmp = $this->dms->getFolder(substr($hit['document_id'], 1))) {
+	//								if($tmp->getAccessMode($user) >= M_READ) {
+										$entries[] = $tmp;
+										$fcount++;
+	//								}
+								}
+							}
+						}
+						if(isset($facets['record_type'])) {
+							$fcount = isset($facets['record_type']['folder']) ? $facets['record_type']['folder'] : 0;
+							$dcount = isset($facets['record_type']['document']) ? $facets['record_type']['document'] : 0 ;
+						}
+					}
+					$this->fcount = $fcount;
+					$this->dcount = $dcount;
+					$this->stats = $stats;
+					$this->entries = $entries;
+					$this->facets = $facets;
+					$totalPages = 0;
+					if($limit > 0) {
+						if($searchresult['count'] > $limit) {
+							$totalPages = (int) ($searchresult['count']/$limit);
+							if($searchresult['count']%$limit)
+								$totalPages++;
+						} else {
+							$totalPages = 1;
+						}
+					}
+					$this->total = $searchresult['count'];
+					$this->totalPages = $totalPages;
+				}
+				$searchTime = $this->getTime() - $startTime;
+				$this->searchTime = round($searchTime, 2);
+			}
+		} else {
+			// ---------------- Start searching -----------------------------------------
+			$startTime = $this->getTime();
+			$resArr = $this->dms->search(array(
+				'query'=>$query,
+				'limit'=>0,
+				'offset'=>0,
+				'logicalmode'=>$mode,
+				'searchin'=>$searchin,
+				'startFolder'=>$startFolder,
+				'owner'=>$ownerobjs,
+				'status'=>$status,
+				'mimetype'=>$mimetype,
+				'creationstartdate'=>$created['from'],
+				'creationenddate'=>$created['to'],
+				'modificationstartdate'=>$modified['from'],
+				'modificationenddate'=>$modified['to'],
+				'filesizestart'=>$filesize['from'],
+				'filesizeend'=>$filesize['to'],
+				'categories'=>$categories,
+				'attributes'=>$attributes,
+				'mode'=>$resultmode,
+				'expirationstartdate'=>$expstartdate ? $expstartdate : array(),
+				'expirationenddate'=>$expenddate ? $expenddate : array(),
+				'statusstartdate'=>$statusstartdate ? $statusstartdate : array(),
+				'statusenddate'=>$statusenddate ? $statusenddate : array(),
+				'orderby'=>$orderby
+			));
+			$this->total = $resArr['totalDocs'] + $resArr['totalFolders'];
+			$searchTime = $this->getTime() - $startTime;
+			$this->searchTime = round($searchTime, 2);
+
+			$entries = array();
+			$fcount = 0;
+			if($resArr['folders']) {
+				foreach ($resArr['folders'] as $entry) {
+					if ($entry->getAccessMode($user) >= M_READ) {
+						$entries[] = $entry;
+						$fcount++;
+					}
+				}
+			}
+			$this->fcount = $fcount;
+			$dcount = 0;
+			if($resArr['docs']) {
+				foreach ($resArr['docs'] as $entry) {
+					if ($entry->getAccessMode($user) >= M_READ) {
+						if($entry->getLatestContent()) {
+							$entry->verifyLastestContentExpriry();
+							$entries[] = $entry;
+							$dcount++;
+						}
+					}
+				}
+			}
+			$this->dcount = $dcount;
+			$totalPages = (int) (count($entries)/$limit);
+			if(count($entries)%$limit)
+				$totalPages++;
+			if($pageNumber != 'all')
+				$this->entries = array_slice($entries, ($pageNumber-1)*$limit, $limit);
+			$this->totalPages = $totalPages;
+			$this->facets = array();
+			$this->stats = array();
+		}
+	} /* }}} */
+
+	public function getFolderCount() {
+		return $this->fcount;
+	}
+
+	public function getDocumentCount() {
+		return $this->dcount;
+	}
+
+	public function getTotal() {
+		return $this->total;
+	}
 } /* }}} */