]> git.openstreetmap.org Git - nominatim.git/blobdiff - lib/Geocode.php
Make deduplicate work again. Only currently support 9.3 due to new postgresql functi...
[nominatim.git] / lib / Geocode.php
index 9b7b9d6c596d620e539b6dfbf323b8ca6a5bf32f..7563a26d54e326691ccb982b39124832abcf23dd 100644 (file)
@@ -27,6 +27,8 @@
 
                protected $bBoundedSearch = false;
                protected $aViewBox = false;
+               protected $sViewboxSmallSQL = false;
+               protected $sViewboxLargeSQL = false;
                protected $aRoutePoints = false;
 
                protected $iMaxRank = 20;
                        return $this->sQuery;
                }
 
+
+               function loadParamArray($aParams)
+               {
+                       if (isset($aParams['addressdetails'])) $this->bIncludeAddressDetails = (bool)$aParams['addressdetails'];
+                       if (isset($aParams['bounded'])) $this->bBoundedSearch = (bool)$aParams['bounded'];
+                       if (isset($aParams['dedupe'])) $this->bDeDupe = (bool)$aParams['dedupe'];
+
+                       if (isset($aParams['limit'])) $this->setLimit((int)$aParams['limit']);
+                       if (isset($aParams['offset'])) $this->iOffset = (int)$aParams['offset'];
+
+                       if (isset($aParams['fallback'])) $this->bFallback = (bool)$aParams['fallback'];
+
+                       // List of excluded Place IDs - used for more acurate pageing
+                       if (isset($aParams['exclude_place_ids']) && $aParams['exclude_place_ids'])
+                       {
+                               foreach(explode(',',$aParams['exclude_place_ids']) as $iExcludedPlaceID)
+                               {
+                                       $iExcludedPlaceID = (int)$iExcludedPlaceID;
+                                       if ($iExcludedPlaceID) $aExcludePlaceIDs[$iExcludedPlaceID] = $iExcludedPlaceID;
+                               }
+                               $this->aExcludePlaceIDs = $aExcludePlaceIDs;
+                       }
+
+                       // Only certain ranks of feature
+                       if (isset($aParams['featureType'])) $this->setFeatureType($aParams['featureType']);
+                       if (isset($aParams['featuretype'])) $this->setFeatureType($aParams['featuretype']);
+
+                       // Country code list
+                       if (isset($aParams['countrycodes']))
+                       {
+                               $aCountryCodes = array();
+                               foreach(explode(',',$aParams['countrycodes']) as $sCountryCode)
+                               {
+                                       if (preg_match('/^[a-zA-Z][a-zA-Z]$/', $sCountryCode))
+                                       {
+                                               $aCountryCodes[] = strtolower($sCountryCode);
+                                       }
+                               }
+                               $this->aCountryCodes = $aCountryCodes;
+                       }
+
+                       if (isset($aParams['viewboxlbrt']) && $aParams['viewboxlbrt'])
+                       {
+                               $aCoOrdinatesLBRT = explode(',',$aParams['viewboxlbrt']);
+                               $this->setViewBox($aCoOrdinatesLBRT[0], $aCoOrdinatesLBRT[1], $aCoOrdinatesLBRT[2], $aCoOrdinatesLBRT[3]);
+                       }
+                       else if (isset($aParams['viewbox']) && $aParams['viewbox'])
+                       {
+                               $aCoOrdinatesLTRB = explode(',',$aParams['viewbox']);
+                               $this->setViewBox($aCoOrdinatesLTRB[0], $aCoOrdinatesLTRB[3], $aCoOrdinatesLTRB[2], $aCoOrdinatesLTRB[1]);
+                       }
+
+                       if (isset($aParams['route']) && $aParams['route'] && isset($aParams['routewidth']) && $aParams['routewidth'])
+                       {
+                               $aPoints = explode(',',$aParams['route']);
+                               if (sizeof($aPoints) % 2 != 0)
+                               {
+                                       userError("Uneven number of points");
+                                       exit;
+                               }
+                               $fPrevCoord = false;
+                               $aRoute = array();
+                               foreach($aPoints as $i => $fPoint)
+                               {
+                                       if ($i%2)
+                                       {
+                                               $aRoute[] = array((float)$fPoint, $fPrevCoord);
+                                       }
+                                       else
+                                       {
+                                               $fPrevCoord = (float)$fPoint;
+                                       }
+                               }
+                               $this->aRoutePoints = $aRoute;
+                       }
+               }
+
+               function setQueryFromParams($aParams)
+               {
+                       // Search query
+                       $sQuery = (isset($aParams['q'])?trim($aParams['q']):'');
+                       if (!$sQuery)
+                       {
+                               $this->setStructuredQuery(@$aParams['amenity'], @$aParams['street'], @$aParams['city'], @$aParams['county'], @$aParams['state'], @$aParams['country'], @$aParams['postalcode']);
+                               $this->setReverseInPlan(false);
+                       }
+                       else
+                       {
+                               $this->setQuery($sQuery);
+                       }
+               }
+
                function loadStructuredAddressElement($sValue, $sKey, $iNewMinAddressRank, $iNewMaxAddressRank, $aItemListValues)
                {
                        $sValue = trim($sValue);
                        // Get the details for display (is this a redundant extra step?)
                        $sPlaceIDs = join(',',$aPlaceIDs);
 
+                       $sImportanceSQL = '';
+                       if ($this->sViewboxSmallSQL) $sImportanceSQL .= " case when ST_Contains($this->sViewboxSmallSQL, ST_Collect(centroid)) THEN 1 ELSE 0.75 END * ";
+                       if ($this->sViewboxLargeSQL) $sImportanceSQL .= " case when ST_Contains($this->sViewboxLargeSQL, ST_Collect(centroid)) THEN 1 ELSE 0.75 END * ";
+
                        $sSQL = "select osm_type,osm_id,class,type,admin_level,rank_search,rank_address,min(place_id) as place_id, min(parent_place_id) as parent_place_id, calculated_country_code as country_code,";
                        $sSQL .= "get_address_by_language(place_id, $sLanguagePrefArraySQL) as langaddress,";
                        $sSQL .= "get_name_by_language(name, $sLanguagePrefArraySQL) as placename,";
                        $sSQL .= "get_name_by_language(name, ARRAY['ref']) as ref,";
                        $sSQL .= "avg(ST_X(centroid)) as lon,avg(ST_Y(centroid)) as lat, ";
-                       $sSQL .= "coalesce(importance,0.75-(rank_search::float/40)) as importance, ";
+                       $sSQL .= $sImportanceSQL."coalesce(importance,0.75-(rank_search::float/40)) as importance, ";
                        $sSQL .= "(select max(p.importance*(p.rank_address+2)) from place_addressline s, placex p where s.place_id = min(CASE WHEN placex.rank_search < 28 THEN placex.place_id ELSE placex.parent_place_id END) and p.place_id = s.address_place_id and s.isaddress and p.importance is not null) as addressimportance, ";
                        $sSQL .= "(extratags->'place') as extra_place ";
                        $sSQL .= "from placex where place_id in ($sPlaceIDs) ";
                                $sSQL .= "null as placename,";
                                $sSQL .= "null as ref,";
                                $sSQL .= "avg(ST_X(centroid)) as lon,avg(ST_Y(centroid)) as lat, ";
-                               $sSQL .= "-0.15 as importance, ";
+                               $sSQL .= $sImportanceSQL."-1.15 as importance, ";
                                $sSQL .= "(select max(p.importance*(p.rank_address+2)) from place_addressline s, placex p where s.place_id = min(location_property_tiger.parent_place_id) and p.place_id = s.address_place_id and s.isaddress and p.importance is not null) as addressimportance, ";
                                $sSQL .= "null as extra_place ";
                                $sSQL .= "from location_property_tiger where place_id in ($sPlaceIDs) ";
                                $sSQL .= "and 30 between $this->iMinAddressRank and $this->iMaxAddressRank ";
                                $sSQL .= "group by place_id";
-                               if (!$this->bDeDupe) $sSQL .= ",place_id";
+                               if (!$this->bDeDupe) $sSQL .= ",place_id ";
                                $sSQL .= " union ";
                                $sSQL .= "select 'L' as osm_type,place_id as osm_id,'place' as class,'house' as type,null as admin_level,30 as rank_search,30 as rank_address,min(place_id) as place_id, min(parent_place_id) as parent_place_id,'us' as country_code,";
                                $sSQL .= "get_address_by_language(place_id, $sLanguagePrefArraySQL) as langaddress,";
                                $sSQL .= "null as placename,";
                                $sSQL .= "null as ref,";
                                $sSQL .= "avg(ST_X(centroid)) as lon,avg(ST_Y(centroid)) as lat, ";
-                               $sSQL .= "-0.10 as importance, ";
+                               $sSQL .= $sImportanceSQL."-1.10 as importance, ";
                                $sSQL .= "(select max(p.importance*(p.rank_address+2)) from place_addressline s, placex p where s.place_id = min(location_property_aux.parent_place_id) and p.place_id = s.address_place_id and s.isaddress and p.importance is not null) as addressimportance, ";
                                $sSQL .= "null as extra_place ";
                                $sSQL .= "from location_property_aux where place_id in ($sPlaceIDs) ";
                                $sSQL .= ",get_address_by_language(place_id, $sLanguagePrefArraySQL) ";
                        }
 
-                       $sSQL .= "order by importance desc";
+                       $sSQL .= " order by importance desc";
                        if (CONST_Debug) { echo "<hr>"; var_dump($sSQL); }
                        $aSearchResults = $this->oDB->getAll($sSQL);
 
                                $sCountryCodesSQL = join(',', array_map('addQuotes', $this->aCountryCodes));
                        }
 
-                       // Hack to make it handle "new york, ny" (and variants) correctly
-                       $sQuery = str_ireplace(array('New York, ny','new york, new york', 'New York ny','new york new york'), 'new york city, ny', $this->sQuery);
+                       $sQuery = $this->sQuery;
 
                        // Conflicts between US state abreviations and various words for 'the' in different languages
                        if (isset($this->aLangPrefOrder['name:en']))
                        {
-                               $sQuery = preg_replace('/,\s*il\s*(,|$)/',', illinois\1', $sQuery);
-                               $sQuery = preg_replace('/,\s*al\s*(,|$)/',', alabama\1', $sQuery);
-                               $sQuery = preg_replace('/,\s*la\s*(,|$)/',', louisiana\1', $sQuery);
+                               $sQuery = preg_replace('/(^|,)\s*il\s*(,|$)/','\1illinois\2', $sQuery);
+                               $sQuery = preg_replace('/(^|,)\s*al\s*(,|$)/','\1alabama\2', $sQuery);
+                               $sQuery = preg_replace('/(^|,)\s*la\s*(,|$)/','\1louisiana\2', $sQuery);
                        }
 
                        // View Box SQL
-                       $sViewboxCentreSQL = $sViewboxSmallSQL = $sViewboxLargeSQL = false;
+                       $sViewboxCentreSQL = false;
                        $bBoundingBoxSearch = false;
                        if ($this->aViewBox)
                        {
                                $aBigViewBox[1] = $this->aViewBox[1] + $fWidth;
                                $aBigViewBox[3] = $this->aViewBox[3] - $fWidth;
 
-                               $sViewboxSmallSQL = "ST_SetSRID(ST_MakeBox2D(ST_Point(".(float)$this->aViewBox[0].",".(float)$this->aViewBox[1]."),ST_Point(".(float)$this->aViewBox[2].",".(float)$this->aViewBox[3].")),4326)";
-                               $sViewboxLargeSQL = "ST_SetSRID(ST_MakeBox2D(ST_Point(".(float)$aBigViewBox[0].",".(float)$aBigViewBox[1]."),ST_Point(".(float)$aBigViewBox[2].",".(float)$aBigViewBox[3].")),4326)";
+                               $this->sViewboxSmallSQL = "ST_SetSRID(ST_MakeBox2D(ST_Point(".(float)$this->aViewBox[0].",".(float)$this->aViewBox[1]."),ST_Point(".(float)$this->aViewBox[2].",".(float)$this->aViewBox[3].")),4326)";
+                               $this->sViewboxLargeSQL = "ST_SetSRID(ST_MakeBox2D(ST_Point(".(float)$aBigViewBox[0].",".(float)$aBigViewBox[1]."),ST_Point(".(float)$aBigViewBox[2].",".(float)$aBigViewBox[3].")),4326)";
                                $bBoundingBoxSearch = $this->bBoundedSearch;
                        }
 
                                $sViewboxCentreSQL .= ")'::geometry,4326)";
 
                                $sSQL = "select st_buffer(".$sViewboxCentreSQL.",".(float)($_GET['routewidth']/69).")";
-                               $sViewboxSmallSQL = $this->oDB->getOne($sSQL);
-                               if (PEAR::isError($sViewboxSmallSQL))
+                               $this->sViewboxSmallSQL = $this->oDB->getOne($sSQL);
+                               if (PEAR::isError($this->sViewboxSmallSQL))
                                {
-                                       failInternalError("Could not get small viewbox.", $sSQL, $sViewboxSmallSQL);
+                                       failInternalError("Could not get small viewbox.", $sSQL, $this->sViewboxSmallSQL);
                                }
-                               $sViewboxSmallSQL = "'".$sViewboxSmallSQL."'::geometry";
+                               $this->sViewboxSmallSQL = "'".$this->sViewboxSmallSQL."'::geometry";
 
                                $sSQL = "select st_buffer(".$sViewboxCentreSQL.",".(float)($_GET['routewidth']/30).")";
-                               $sViewboxLargeSQL = $this->oDB->getOne($sSQL);
-                               if (PEAR::isError($sViewboxLargeSQL))
+                               $this->sViewboxLargeSQL = $this->oDB->getOne($sSQL);
+                               if (PEAR::isError($this->sViewboxLargeSQL))
                                {
-                                       failInternalError("Could not get large viewbox.", $sSQL, $sViewboxLargeSQL);
+                                       failInternalError("Could not get large viewbox.", $sSQL, $this->sViewboxLargeSQL);
                                }
-                               $sViewboxLargeSQL = "'".$sViewboxLargeSQL."'::geometry";
+                               $this->sViewboxLargeSQL = "'".$this->sViewboxLargeSQL."'::geometry";
                                $bBoundingBoxSearch = $this->bBoundedSearch;
                        }
 
                                                                                                if ($aSearch['sHouseNumber'] === '')
                                                                                                {
                                                                                                        $aSearch['sHouseNumber'] = $sToken;
+                                                                                                       // sanity check: if the housenumber is not mainly made
+                                                                                                       // up of numbers, add a penalty
+                                                                                                       if (preg_match_all("/[^0-9]/", $sToken, $aMatches) > 2) $aSearch['iSearchRank']++;
                                                                                                        if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
                                                                                                        /*
                                                                                                        // Fall back to not searching for this item (better than nothing)
                                                                                                        $aSearch['sType'] = $aSearchTerm['type'];
                                                                                                        if (sizeof($aSearch['aName'])) $aSearch['sOperator'] = 'name';
                                                                                                        else $aSearch['sOperator'] = 'near'; // near = in for the moment
+                                                                                                       if (strlen($aSearchTerm['operator']) == 0) $aSearch['iSearchRank'] += 1;
 
                                                                                                        // Do we have a shortcut id?
                                                                                                        if ($aSearch['sOperator'] == 'name')
                                                                        if (CONST_Debug) var_dump($sSQL);
                                                                        $aPlaceIDs = $this->oDB->getCol($sSQL);
                                                                }
+                                                               else
+                                                               {
+                                                                       $aPlaceIDs = array();
+                                                               }
                                                        }
                                                        else
                                                        {
                                                                {
                                                                        $sSQL = "select place_id from place_classtype_".$aSearch['sClass']."_".$aSearch['sType']." ct";
                                                                        if ($sCountryCodesSQL) $sSQL .= " join placex using (place_id)";
-                                                                       $sSQL .= " where st_contains($sViewboxSmallSQL, ct.centroid)";
+                                                                       $sSQL .= " where st_contains($this->sViewboxSmallSQL, ct.centroid)";
                                                                        if ($sCountryCodesSQL) $sSQL .= " and calculated_country_code in ($sCountryCodesSQL)";
                                                                        if (sizeof($this->aExcludePlaceIDs))
                                                                        {
                                                                        // If excluded place IDs are given, it is fair to assume that
                                                                        // there have been results in the small box, so no further
                                                                        // expansion in that case.
-                                                                       if (!sizeof($aPlaceIDs) && !sizeof($this->aExcludePlaceIDs))
+                                                                       // Also don't expand if bounded results were requested.
+                                                                       if (!sizeof($aPlaceIDs) && !sizeof($this->aExcludePlaceIDs) && !$this->bBoundedSearch)
                                                                        {
                                                                                $sSQL = "select place_id from place_classtype_".$aSearch['sClass']."_".$aSearch['sType']." ct";
                                                                                if ($sCountryCodesSQL) $sSQL .= " join placex using (place_id)";
-                                                                               $sSQL .= " where st_contains($sViewboxLargeSQL, ct.centroid)";
+                                                                               $sSQL .= " where st_contains($this->sViewboxLargeSQL, ct.centroid)";
                                                                                if ($sCountryCodesSQL) $sSQL .= " and calculated_country_code in ($sCountryCodesSQL)";
                                                                                if ($sViewboxCentreSQL) $sSQL .= " order by st_distance($sViewboxCentreSQL, ct.centroid) asc";
                                                                                $sSQL .= " limit $this->iLimit";
                                                                else
                                                                {
                                                                        $sSQL = "select place_id from placex where class='".$aSearch['sClass']."' and type='".$aSearch['sType']."'";
-                                                                       $sSQL .= " and st_contains($sViewboxSmallSQL, geometry) and linked_place_id is null";
+                                                                       $sSQL .= " and st_contains($this->sViewboxSmallSQL, geometry) and linked_place_id is null";
                                                                        if ($sCountryCodesSQL) $sSQL .= " and calculated_country_code in ($sCountryCodesSQL)";
                                                                        if ($sViewboxCentreSQL) $sSQL .= " order by st_distance($sViewboxCentreSQL, centroid) asc";
                                                                        $sSQL .= " limit $this->iLimit";
                                                                $aTerms[] = "country_code in ($sCountryCodesSQL)";
                                                        }
 
-                                                       if ($bBoundingBoxSearch) $aTerms[] = "centroid && $sViewboxSmallSQL";
+                                                       if ($bBoundingBoxSearch) $aTerms[] = "centroid && $this->sViewboxSmallSQL";
                                                        if ($sNearPointSQL) $aOrder[] = "ST_Distance($sNearPointSQL, centroid) asc";
 
-                                                       $sImportanceSQL = '(case when importance = 0 OR importance IS NULL then 0.75-(search_rank::float/40) else importance end)';
-                                                       if ($sViewboxSmallSQL) $sImportanceSQL .= " * case when ST_Contains($sViewboxSmallSQL, centroid) THEN 1 ELSE 0.5 END";
-                                                       if ($sViewboxLargeSQL) $sImportanceSQL .= " * case when ST_Contains($sViewboxLargeSQL, centroid) THEN 1 ELSE 0.5 END";
+                                                       if ($aSearch['sHouseNumber'])
+                                                       {
+                                                               $sImportanceSQL = '- abs(26 - address_rank) + 3';
+                                                       }
+                                                       else
+                                                       {
+                                                               $sImportanceSQL = '(case when importance = 0 OR importance IS NULL then 0.75-(search_rank::float/40) else importance end)';
+                                                       }
+                                                       if ($this->sViewboxSmallSQL) $sImportanceSQL .= " * case when ST_Contains($this->sViewboxSmallSQL, centroid) THEN 1 ELSE 0.5 END";
+                                                       if ($this->sViewboxLargeSQL) $sImportanceSQL .= " * case when ST_Contains($this->sViewboxLargeSQL, centroid) THEN 1 ELSE 0.5 END";
+
                                                        $aOrder[] = "$sImportanceSQL DESC";
                                                        if (sizeof($aSearch['aFullNameAddress']))
                                                        {
                                                                $sPlaceIDs = join(',',$aPlaceIDs);
 
                                                                // Now they are indexed look for a house attached to a street we found
-                                                               $sHouseNumberRegex = '\\\\m'.str_replace(' ','[-,/ ]',$aSearch['sHouseNumber']).'\\\\M';
-                                                               $sSQL = "select place_id from placex where parent_place_id in (".$sPlaceIDs.") and housenumber ~* E'".$sHouseNumberRegex."'";
+                                                               $sHouseNumberRegex = '\\\\m'.$aSearch['sHouseNumber'].'\\\\M';
+                                                               $sSQL = "select place_id from placex where parent_place_id in (".$sPlaceIDs.") and transliteration(housenumber) ~* E'".$sHouseNumberRegex."'";
                                                                if (sizeof($this->aExcludePlaceIDs))
                                                                {
                                                                        $sSQL .= " and place_id not in (".join(',',$this->aExcludePlaceIDs).")";
                                                                                        }
                                                                                        if ($sCountryCodesSQL) $sSQL .= " and lp.calculated_country_code in ($sCountryCodesSQL)";
                                                                                        if ($sOrderBySQL) $sSQL .= "order by ".$sOrderBySQL." asc";
-                                                                                       if ($iOffset) $sSQL .= " offset $iOffset";
+                                                                                       if ($this->iOffset) $sSQL .= " offset $this->iOffset";
                                                                                        $sSQL .= " limit $this->iLimit";
                                                                                        if (CONST_Debug) var_dump($sSQL);
                                                                                        $aClassPlaceIDs = array_merge($aClassPlaceIDs, $this->oDB->getCol($sSQL));
                                                                                        }
                                                                                        if ($sCountryCodesSQL) $sSQL .= " and l.calculated_country_code in ($sCountryCodesSQL)";
                                                                                        if ($sOrderBy) $sSQL .= "order by ".$OrderBysSQL." asc";
-                                                                                       if ($iOffset) $sSQL .= " offset $iOffset";
+                                                                                       if ($this->iOffset) $sSQL .= " offset $this->iOffset";
                                                                                        $sSQL .= " limit $this->iLimit";
                                                                                        if (CONST_Debug) var_dump($sSQL);
                                                                                        $aClassPlaceIDs = array_merge($aClassPlaceIDs, $this->oDB->getCol($sSQL));
                        }
 
                        $aClassType = getClassTypesWithImportance();
-                       $aRecheckWords = preg_split('/\b/u',$sQuery);
+                       $aRecheckWords = preg_split('/\b[\s,\\-]*/u',$sQuery);
                        foreach($aRecheckWords as $i => $sWord)
                        {
                                if (!$sWord) unset($aRecheckWords[$i]);
                                $sAddress = $aResult['langaddress'];
                                foreach($aRecheckWords as $i => $sWord)
                                {
-                                       if (stripos($sAddress, $sWord)!==false) $iCountWords++;
+                                       if (stripos($sAddress, $sWord)!==false)
+                                       {
+                                               $iCountWords++;
+                                               if (preg_match("/(^|,)\s*$sWord\s*(,|$)/", $sAddress)) $iCountWords += 0.1;
+                                       }
                                }
 
                                $aResult['importance'] = $aResult['importance'] + ($iCountWords*0.1); // 0.1 is a completely arbitrary number but something in the range 0.1 to 0.5 would seem right
 
        } // end class
 
-
-/*
-               if (isset($_GET['route']) && $_GET['route'] && isset($_GET['routewidth']) && $_GET['routewidth'])
-               {
-                       $aPoints = explode(',',$_GET['route']);
-                       if (sizeof($aPoints) % 2 != 0)
-                       {
-                               userError("Uneven number of points");
-                               exit;
-                       }
-                       $sViewboxCentreSQL = "ST_SetSRID('LINESTRING(";
-                       $fPrevCoord = false;
-               }
-*/