Check for positive polygon simplification threshold

[nominatim.git] / lib / Geocode.php

diff --git a/lib/Geocode.php b/lib/Geocode.php
index e441e8869c479a7cc351e77290018054de310365..6781ceecb9993bccdf89b746795ab9a7d8c68763 100644 (file)
--- a/lib/Geocode.php
+++ b/lib/Geocode.php
@@ -12,6 +12,7 @@
                protected $bIncludePolygonAsGeoJSON = false;
                protected $bIncludePolygonAsKML = false;
                protected $bIncludePolygonAsSVG = false;
                protected $bIncludePolygonAsGeoJSON = false;
                protected $bIncludePolygonAsKML = false;
                protected $bIncludePolygonAsSVG = false;

+               protected $fPolygonSimplificationThreshold = 0.0;

 
                protected $aExcludePlaceIDs = array();
                protected $bDeDupe = true;
 
                protected $aExcludePlaceIDs = array();
                protected $bDeDupe = true;


@@ -102,6 +103,11 @@
                        $this->bIncludePolygonAsSVG = $b;
                }
 
                        $this->bIncludePolygonAsSVG = $b;
                }
 

+               function setPolygonSimplificationThreshold($f)
+               {
+                       $this->fPolygonSimplificationThreshold = $f;
+               }
+

                function setDeDupe($bDeDupe = true)
                {
                        $this->bDeDupe = (bool)$bDeDupe;
                function setDeDupe($bDeDupe = true)
                {
                        $this->bDeDupe = (bool)$bDeDupe;


@@ -222,9 +228,12 @@
                                foreach(explode(',',$aParams['exclude_place_ids']) as $iExcludedPlaceID)
                                {
                                        $iExcludedPlaceID = (int)$iExcludedPlaceID;
                                foreach(explode(',',$aParams['exclude_place_ids']) as $iExcludedPlaceID)
                                {
                                        $iExcludedPlaceID = (int)$iExcludedPlaceID;

-                                       if ($iExcludedPlaceID) $aExcludePlaceIDs[$iExcludedPlaceID] = $iExcludedPlaceID;
+                                       if ($iExcludedPlaceID)
+                                               $aExcludePlaceIDs[$iExcludedPlaceID] = $iExcludedPlaceID;

                                }
                                }

-                               $this->aExcludePlaceIDs = $aExcludePlaceIDs;
+
+                               if (isset($aExcludePlaceIDs))
+                                       $this->aExcludePlaceIDs = $aExcludePlaceIDs;

                        }
 
                        // Only certain ranks of feature
                        }
 
                        // Only certain ranks of feature


@@ -526,7 +535,7 @@
 
                                                                                        if (sizeof($aSearch['aName']))
                                                                                        {
 
                                                                                        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'];
                                                                                                }
                                                                                                {
                                                                                                        $aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
                                                                                                }


@@ -553,6 +562,8 @@
                                                                                        // 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']++;
                                                                                        // 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[] = $aSearch;
                                                                                        /*
                                                                                        // Fall back to not searching for this item (better than nothing)
                                                                                        if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
                                                                                        /*
                                                                                        // Fall back to not searching for this item (better than nothing)


@@ -592,7 +603,7 @@
                                                                        {
                                                                                if (sizeof($aSearch['aName']))
                                                                                {
                                                                        {
                                                                                if (sizeof($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'];
                                                                                        }
                                                                                        {
                                                                                                $aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
                                                                                        }


@@ -630,7 +641,7 @@
                                                                                                $aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
                                                                                                if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
                                                                                        }
                                                                                                $aSearch['aAddress'][$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;
                                                                                        {
                                                                                                $aSearch['aAddressNonSearch'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
                                                                                                $aSearch['iSearchRank'] += 1;


@@ -866,10 +877,10 @@
 
                                preg_match_all('/\\[([\\w ]*)\\]/u', $sQuery, $aSpecialTermsRaw, PREG_SET_ORDER);
                                $aSpecialTerms = array();
 
                                preg_match_all('/\\[([\\w ]*)\\]/u', $sQuery, $aSpecialTermsRaw, PREG_SET_ORDER);
                                $aSpecialTerms = array();

-                               if (isset($aStructuredQuery['amenity']) && $aStructuredQuery['amenity'])
+                               if (isset($this->aStructuredQuery['amenity']) && $this->aStructuredQuery['amenity'])

                                {
                                {

-                                       $aSpecialTermsRaw[] = array('['.$aStructuredQuery['amenity'].']', $aStructuredQuery['amenity']);
-                                       unset($aStructuredQuery['amenity']);
+                                       $aSpecialTermsRaw[] = array('['.$this->aStructuredQuery['amenity'].']', $this->aStructuredQuery['amenity']);
+                                       unset($this->aStructuredQuery['amenity']);

                                }
                                foreach($aSpecialTermsRaw as $aSpecialTerm)
                                {
                                }
                                foreach($aSpecialTermsRaw as $aSpecialTerm)
                                {


@@ -967,8 +978,8 @@
                                        foreach($aDatabaseWords as $aToken)
                                        {
                                                // Very special case - require 2 letter country param to match the country code found
                                        foreach($aDatabaseWords as $aToken)
                                        {
                                                // Very special case - require 2 letter country param to match the country code found

-                                               if ($bStructuredPhrases && $aToken['country_code'] && !empty($aStructuredQuery['country'])
-                                                               && strlen($aStructuredQuery['country']) == 2 && strtolower($aStructuredQuery['country']) != $aToken['country_code'])
+                                               if ($bStructuredPhrases && $aToken['country_code'] && !empty($this->aStructuredQuery['country'])
+                                                               && strlen($this->aStructuredQuery['country']) == 2 && strtolower($this->aStructuredQuery['country']) != $aToken['country_code'])

                                                {
                                                        continue;
                                                }
                                                {
                                                        continue;
                                                }


@@ -1092,7 +1103,7 @@
 
                                if (CONST_Debug) var_Dump($aGroupedSearches);
 
 
                                if (CONST_Debug) var_Dump($aGroupedSearches);
 

-                               if (CONST_Search_TryDroppedAddressTerms && sizeof($aStructuredQuery) > 0)
+                               if (CONST_Search_TryDroppedAddressTerms && sizeof($this->aStructuredQuery) > 0)

                                {
                                        $aCopyGroupedSearches = $aGroupedSearches;
                                        foreach($aCopyGroupedSearches as $iGroup => $aSearches)
                                {
                                        $aCopyGroupedSearches = $aGroupedSearches;
                                        foreach($aCopyGroupedSearches as $iGroup => $aSearches)


@@ -1238,6 +1249,12 @@
                                                        $aTerms = array();
                                                        $aOrder = array();
 
                                                        $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'],",")."]";
                                                        // 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 +1335,7 @@
                                                                $sSQL .= " where ".join(' and ',$aTerms);
                                                                $sSQL .= " order by ".join(', ',$aOrder);
                                                                if ($aSearch['sHouseNumber'] || $aSearch['sClass'])
                                                                $sSQL .= " where ".join(' and ',$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'])
                                                                        $sSQL .= " limit 1";
                                                                else
                                                                elseif (!sizeof($aSearch['aName']) && !sizeof($aSearch['aAddress']) && $aSearch['sClass'])
                                                                        $sSQL .= " limit 1";
                                                                else


@@ -1583,9 +1600,11 @@
                        $aRecheckWords = preg_split('/\b[\s,\\-]*/u',$sQuery);
                        foreach($aRecheckWords as $i => $sWord)
                        {
                        $aRecheckWords = preg_split('/\b[\s,\\-]*/u',$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 => $aResult)
                        {
                                if (CONST_Search_AreaPolygons)
                        foreach($aSearchResults as $iResNum => $aResult)
                        {
                                if (CONST_Search_AreaPolygons)


@@ -1593,13 +1612,22 @@
                                        // 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,";
                                        // 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) $sSQL .= ",ST_AsText(geometry) as astext";
                                        if ($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\'';
+                                       $sFrom = " from placex where place_id = ".$aResult['place_id'];
+                                       if ($this->fPolygonSimplificationThreshold > 0)
+                                       {
+                                               $sSQL .= " from (select place_id,centroid,ST_SimplifyPreserveTopology(geometry,".$this->fPolygonSimplificationThreshold.") as geometry".$sFrom.") as plx";
+                                       }
+                                       else
+                                       {
+                                               $sSQL .= $sFrom;
+                                       }
+

                                        $aPointPolygon = $this->oDB->getRow($sSQL);
                                        if (PEAR::IsError($aPointPolygon))
                                        {
                                        $aPointPolygon = $this->oDB->getRow($sSQL);
                                        if (PEAR::IsError($aPointPolygon))
                                        {


@@ -1621,7 +1649,7 @@
 
                                                if ($this->bIncludePolygonAsPoints)
                                                {
 
                                                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);
                                                        if (preg_match('#POLYGON\\(\\(([- 0-9.,]+)#',$aPointPolygon['astext'],$aMatch))
                                                        {
                                                                preg_match_all('/(-?[0-9.]+) (-?[0-9.]+)/',$aMatch[1],$aPolyPoints,PREG_SET_ORDER);


@@ -1744,7 +1772,7 @@
                                        if (stripos($sAddress, $sWord)!==false)
                                        {
                                                $iCountWords++;
                                        if (stripos($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;

                                        }
                                }
 
                                        }
                                }