correctly quote regular expression

[nominatim.git] / lib / Geocode.php

diff --git a/lib/Geocode.php b/lib/Geocode.php
index 22ad43327adcfd7cf3851257886cd9c1e226d118..dfb3af2ae823c60eabe816bcc3af2d6c7bb651f1 100644 (file)
--- a/lib/Geocode.php
+++ b/lib/Geocode.php
@@ -526,7 +526,7 @@
 
                                                                                        if (sizeof($aSearch['aName']))
                                                                                        {
-                                                                                               if ((!$bStructuredPhrases || $iPhrase > 0) && $sPhraseType != 'country' && (!isset($aValidTokens[$sToken]) || strlen($sToken) < 4 || strpos($sToken, ' ') !== false))
+                                                                                               if ((!$bStructuredPhrases || $iPhrase > 0) && $sPhraseType != 'country' && (!isset($aValidTokens[$sToken]) || strpos($sToken, ' ') !== false))
                                                                                                {
                                                                                                        $aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
                                                                                                }
@@ -553,6 +553,8 @@
                                                                                        // sanity check: if the housenumber is not mainly made
                                                                                        // up of numbers, add a penalty
                                                                                        if (preg_match_all("/[^0-9]/", $sToken, $aMatches) > 2) $aSearch['iSearchRank']++;
+                                                                                       // also housenumbers should appear in the first or second phrase
+                                                                                       if ($iPhrase > 1) $aSearch['iSearchRank'] += 1;
                                                                                        if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
                                                                                        /*
                                                                                        // Fall back to not searching for this item (better than nothing)
@@ -592,7 +594,7 @@
                                                                        {
                                                                                if (sizeof($aSearch['aName']))
                                                                                {
-                                                                                       if ((!$bStructuredPhrases || $iPhrase > 0) && $sPhraseType != 'country' && (!isset($aValidTokens[$sToken]) || strlen($sToken) < 4 || strpos($sToken, ' ') !== false))
+                                                                                       if ((!$bStructuredPhrases || $iPhrase > 0) && $sPhraseType != 'country' && (!isset($aValidTokens[$sToken]) || strpos($sToken, ' ') !== false))
                                                                                        {
                                                                                                $aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
                                                                                        }
@@ -630,7 +632,7 @@
                                                                                                $aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
                                                                                                if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
                                                                                        }
-                                                                                       elseif (isset($aValidTokens[' '.$sToken]) && strlen($sToken) >= 4) // revert to the token version?
+                                                                                       elseif (isset($aValidTokens[' '.$sToken])) // revert to the token version?
                                                                                        {
                                                                                                $aSearch['aAddressNonSearch'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
                                                                                                $aSearch['iSearchRank'] += 1;
@@ -1054,7 +1056,7 @@
                                                if (sizeof($aPhrases) > 1)
                                                {
                                                        $aFinalPhrase = end($aPhrases);
-                                                       $aFinalPhrase['wordsets'] = getInverseWordSets($aFinalPhrase['words'], 0);
+                                                       $aPhrases[sizeof($aPhrases)-1]['wordsets'] = getInverseWordSets($aFinalPhrase['words'], 0);
                                                }
                                                $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, null, $aPhrases, $aValidTokens, $aWordFrequencyScores, false);
 
@@ -1238,6 +1240,12 @@
                                                        $aTerms = array();
                                                        $aOrder = array();
 
+                                                       if ($aSearch['sHouseNumber'] && sizeof($aSearch['aAddress']))
+                                                       {
+                                                               $sHouseNumberRegex = '\\\\m'.$aSearch['sHouseNumber'].'\\\\M';
+                                                               $aOrder[] = "exists(select place_id from placex where parent_place_id = search_name.place_id and transliteration(housenumber) ~* E'".$sHouseNumberRegex."' limit 1) desc";
+                                                       }
+
                                                        // TODO: filter out the pointless search terms (2 letter name tokens and less)
                                                        // they might be right - but they are just too darned expensive to run
                                                        if (sizeof($aSearch['aName'])) $aTerms[] = "name_vector @> ARRAY[".join($aSearch['aName'],",")."]";
@@ -1318,7 +1326,7 @@
                                                                $sSQL .= " where ".join(' and ',$aTerms);
                                                                $sSQL .= " order by ".join(', ',$aOrder);
                                                                if ($aSearch['sHouseNumber'] || $aSearch['sClass'])
-                                                                       $sSQL .= " limit 50";
+                                                                       $sSQL .= " limit 20";
                                                                elseif (!sizeof($aSearch['aName']) && !sizeof($aSearch['aAddress']) && $aSearch['sClass'])
                                                                        $sSQL .= " limit 1";
                                                                else
@@ -1583,9 +1591,11 @@
                        $aRecheckWords = preg_split('/\b[\s,\\-]*/u',$sQuery);
                        foreach($aRecheckWords as $i => $sWord)
                        {
-                               if (!$sWord) unset($aRecheckWords[$i]);
+                               if (!preg_match('/\pL/', $sWord)) unset($aRecheckWords[$i]);
                        }
 
+            if (CONST_Debug) { echo '<i>Recheck words:<\i>'; var_dump($aRecheckWords); }
+
                        foreach($aSearchResults as $iResNum => $aResult)
                        {
                                if (CONST_Search_AreaPolygons)
@@ -1593,13 +1603,13 @@
                                        // Get the bounding box and outline polygon
                                        $sSQL = "select place_id,0 as numfeatures,st_area(geometry) as area,";
                                        $sSQL .= "ST_Y(centroid) as centrelat,ST_X(centroid) as centrelon,";
-                                       $sSQL .= "ST_Y(ST_PointN(ST_ExteriorRing(Box2D(geometry)),4)) as minlat,ST_Y(ST_PointN(ST_ExteriorRing(Box2D(geometry)),2)) as maxlat,";
-                                       $sSQL .= "ST_X(ST_PointN(ST_ExteriorRing(Box2D(geometry)),1)) as minlon,ST_X(ST_PointN(ST_ExteriorRing(Box2D(geometry)),3)) as maxlon";
+                                       $sSQL .= "ST_YMin(geometry) as minlat,ST_YMax(geometry) as maxlat,";
+                                       $sSQL .= "ST_XMin(geometry) as minlon,ST_XMax(geometry) as maxlon";
                                        if ($this->bIncludePolygonAsGeoJSON) $sSQL .= ",ST_AsGeoJSON(geometry) as asgeojson";
                                        if ($this->bIncludePolygonAsKML) $sSQL .= ",ST_AsKML(geometry) as askml";
                                        if ($this->bIncludePolygonAsSVG) $sSQL .= ",ST_AsSVG(geometry) as assvg";
                                        if ($this->bIncludePolygonAsText || $this->bIncludePolygonAsPoints) $sSQL .= ",ST_AsText(geometry) as astext";
-                                       $sSQL .= " from placex where place_id = ".$aResult['place_id'].' and st_geometrytype(Box2D(geometry)) = \'ST_Polygon\'';
+                                       $sSQL .= " from placex where place_id = ".$aResult['place_id'];
                                        $aPointPolygon = $this->oDB->getRow($sSQL);
                                        if (PEAR::IsError($aPointPolygon))
                                        {
@@ -1621,7 +1631,7 @@
 
                                                if ($this->bIncludePolygonAsPoints)
                                                {
-                                                       // Translate geometary string to point array
+                                                       // Translate geometry string to point array
                                                        if (preg_match('#POLYGON\\(\\(([- 0-9.,]+)#',$aPointPolygon['astext'],$aMatch))
                                                        {
                                                                preg_match_all('/(-?[0-9.]+) (-?[0-9.]+)/',$aMatch[1],$aPolyPoints,PREG_SET_ORDER);
@@ -1744,7 +1754,7 @@
                                        if (stripos($sAddress, $sWord)!==false)
                                        {
                                                $iCountWords++;
-                                               if (preg_match("/(^|,)\s*$sWord\s*(,|$)/", $sAddress)) $iCountWords += 0.1;
+                                               if (preg_match("/(^|,)\s*".preg_quote($sWord, '/')."\s*(,|$)/", $sAddress)) $iCountWords += 0.1;
                                        }
                                }