Implement geocoding and reverse geocoding with tiger interpolation lines instead...

[nominatim.git] / lib / Geocode.php

diff --git a/lib/Geocode.php b/lib/Geocode.php
index 2f0d085710a1b089499c36170303a4833346a7b9..ed8a4f375a054ea2712aa72cf90335891bc10ce8 100644 (file)
--- a/lib/Geocode.php
+++ b/lib/Geocode.php
@@ -46,10 +46,6 @@
 
                protected $sQuery = false;
                protected $aStructuredQuery = false;
 
                protected $sQuery = false;
                protected $aStructuredQuery = false;

-        
-        //for Tiger housenumber interpolation
-        protected $searchedHousenumber=-1;
-        protected $housenumberFound=false;

 
                function Geocode(&$oDB)
                {
 
                function Geocode(&$oDB)
                {


@@ -397,19 +393,20 @@
 
                function getDetails($aPlaceIDs)
                {
 
                function getDetails($aPlaceIDs)
                {

+            //$aPlaceIDs is an array with key: placeID and value: tiger-housenumber, if found, else -1

                        if (sizeof($aPlaceIDs) == 0)  return array();
 
                        $sLanguagePrefArraySQL = "ARRAY[".join(',',array_map("getDBQuoted",$this->aLangPrefOrder))."]";
 
                        // Get the details for display (is this a redundant extra step?)
                        if (sizeof($aPlaceIDs) == 0)  return array();
 
                        $sLanguagePrefArraySQL = "ARRAY[".join(',',array_map("getDBQuoted",$this->aLangPrefOrder))."]";
 
                        // Get the details for display (is this a redundant extra step?)

-                       $sPlaceIDs = join(',',$aPlaceIDs);
+                       $sPlaceIDs = join(',',array_keys($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,";
 
                        $sImportanceSQL = '';
                        if ($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_address_by_language(place_id, -1, $sLanguagePrefArraySQL) as langaddress,";

                        $sSQL .= "get_name_by_language(name, $sLanguagePrefArraySQL) as placename,";
                        $sSQL .= "get_name_by_language(name, ARRAY['ref']) as ref,";
                        if ($this->bIncludeExtraTags) $sSQL .= "hstore_to_json(extratags)::text as extra,";
                        $sSQL .= "get_name_by_language(name, $sLanguagePrefArraySQL) as placename,";
                        $sSQL .= "get_name_by_language(name, ARRAY['ref']) as ref,";
                        if ($this->bIncludeExtraTags) $sSQL .= "hstore_to_json(extratags)::text as extra,";


@@ -436,32 +433,42 @@
 
                        if (30 >= $this->iMinAddressRank && 30 <= $this->iMaxAddressRank)
                        {
 
                        if (30 >= $this->iMinAddressRank && 30 <= $this->iMaxAddressRank)
                        {

-                               //query also location_property_tiger_line and location_property_aux
-                //Tiger search only if it was searched for a housenumber (searchedHousenumber >=0) and if it was found (housenumberFound = true)
+                               //query also location_property_tiger and location_property_aux
+                //Tiger search only if a housenumber was searched and if it was found (i.e. aPlaceIDs[placeID] = housenumber != -1) (realized through a join)

                 //only Tiger housenumbers need to be interpolated, because they are saved as lines with start- and endnumber, the common osm housenumbers are usually saved as points
                 //only Tiger housenumbers need to be interpolated, because they are saved as lines with start- and endnumber, the common osm housenumbers are usually saved as points

-                if($this->searchedHousenumber>=0 && $this->housenumberFound){
-                    $sSQL .= "union ";
-                    $sSQL .= "select 'T' 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";
-                    $sSQL .= ", null as ref";
-                    if ($this->bIncludeExtraTags) $sSQL .= ", null as extra";
-                    if ($this->bIncludeNameDetails) $sSQL .= ", null as names";
-                    $sSQL .= ", avg(st_x(point)) as lon, avg(st_y(point)) as lat";
-                    $sSQL .= $sImportanceSQL.", -1.15 as importance ";
-                    $sSQL .= ", 1.0 as addressimportance "; //not sure how the addressimportance is/should be calculated for Tiger data
-                    $sSQL .= ", null as extra_place ";
-                    $sSQL .= " from (select place_id";
-                    //interpolate the Tiger housenumbers here
-                    $sSQL .= ",ST_LineInterpolatePoint(linegeo, ($this->searchedHousenumber::float-startnumber::float)/(endnumber-startnumber)::float) as point, parent_place_id ";
-                    $sSQL .= "from location_property_tiger_line where place_id in ($sPlaceIDs) ";
-                    $sSQL .= "and 30 between $this->iMinAddressRank and $this->iMaxAddressRank) as blub"; //postgres wants an alias here
-                    $sSQL .= " group by place_id"; //why group by place_id, isnt place_id unique?
-                    if (!$this->bDeDupe) $sSQL .= ",place_id ";
+                $sHousenumbers = "";
+                $i=0;
+                $length=count($aPlaceIDs);
+                foreach($aPlaceIDs as $placeID => $housenumber){
+                    $i++;
+                    $sHousenumbers .= "(".$placeID.",".$housenumber.")";
+                    if($i<$length)
+                        $sHousenumbers .= ",";

                 }
                 }

+                
+                $sSQL .= "union ";
+                $sSQL .= "select 'T' 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, housenumber_for_place, $sLanguagePrefArraySQL) as langaddress ";
+                $sSQL .= ", null as placename";
+                $sSQL .= ", null as ref";
+                if ($this->bIncludeExtraTags) $sSQL .= ", null as extra";
+                if ($this->bIncludeNameDetails) $sSQL .= ", null as names";
+                $sSQL .= ", avg(st_x(point)) as lon, avg(st_y(point)) as lat";
+                $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(blub.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 (select place_id";
+                //interpolate the Tiger housenumbers here
+                $sSQL .= ",ST_LineInterpolatePoint(linegeo, (housenumber_for_place-startnumber::float)/(endnumber-startnumber)::float) as point, parent_place_id, housenumber_for_place ";
+                $sSQL .= "from (location_property_tiger ";
+                $sSQL .= " join (values ".$sHousenumbers.") as housenumbers(place_id,housenumber_for_place) using(place_id)) ";
+                $sSQL .= " where housenumber_for_place>=0 and 30 between $this->iMinAddressRank and $this->iMaxAddressRank) as blub"; //postgres wants an alias here
+                $sSQL .= " group by place_id, housenumber_for_place"; //is this group by really needed?, place_id + housenumber (in combination) are unique
+                if (!$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 .= " 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 .= "get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) as langaddress,";

                                $sSQL .= "null as placename,";
                                $sSQL .= "null as ref,";
                                if ($this->bIncludeExtraTags) $sSQL .= "null as extra,";
                                $sSQL .= "null as placename,";
                                $sSQL .= "null as ref,";
                                if ($this->bIncludeExtraTags) $sSQL .= "null as extra,";


@@ -474,7 +481,7 @@
                                $sSQL .= "and 30 between $this->iMinAddressRank and $this->iMaxAddressRank ";
                                $sSQL .= "group by place_id";
                                if (!$this->bDeDupe) $sSQL .= ",place_id";
                                $sSQL .= "and 30 between $this->iMinAddressRank and $this->iMaxAddressRank ";
                                $sSQL .= "group by place_id";
                                if (!$this->bDeDupe) $sSQL .= ",place_id";

-                               $sSQL .= ",get_address_by_language(place_id, $sLanguagePrefArraySQL) ";
+                               $sSQL .= ",get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) ";

                        }
 
                        $sSQL .= " order by importance desc";
                        }
 
                        $sSQL .= " order by importance desc";


@@ -1091,6 +1098,7 @@
                                        // TODO: suggestions
 
                                        // Start the search process
                                        // TODO: suggestions
 
                                        // Start the search process

+                    // array with: placeid => -1 | tiger-housenumber

                                        $aResultPlaceIDs = array();
 
                                        $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases);
                                        $aResultPlaceIDs = array();
 
                                        $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases);


@@ -1208,6 +1216,7 @@
                                        foreach($aSearches as $aSearch)
                                        {
                                                $iQueryLoop++;
                                        foreach($aSearches as $aSearch)
                                        {
                                                $iQueryLoop++;

+                        $searchedHousenumber=-1;

 
                                                if (CONST_Debug) { echo "<hr><b>Search Loop, group $iGroupLoop, loop $iQueryLoop</b>"; }
                                                if (CONST_Debug) _debugDumpGroupedSearches(array($iGroupedRank => array($aSearch)), $aValidTokens);
 
                                                if (CONST_Debug) { echo "<hr><b>Search Loop, group $iGroupLoop, loop $iQueryLoop</b>"; }
                                                if (CONST_Debug) _debugDumpGroupedSearches(array($iGroupedRank => array($aSearch)), $aValidTokens);


@@ -1433,18 +1442,18 @@
                                                                        if (CONST_Debug) var_dump($sSQL);
                                                                        $aPlaceIDs = $this->oDB->getCol($sSQL);
                                                                }
                                                                        if (CONST_Debug) var_dump($sSQL);
                                                                        $aPlaceIDs = $this->oDB->getCol($sSQL);
                                                                }

-                                //if nothing was found in placex or location_property_aux, then search in Tiger data for this housenumber(location_property_tiger_line)
+                                //if nothing was found in placex or location_property_aux, then search in Tiger data for this housenumber(location_property_tiger)
+                                $searchedHousenumber = intval($aSearch['sHouseNumber']);

                                                                if (!sizeof($aPlaceIDs))
                                                                {
                                                                        //$sSQL = "select place_id from location_property_tiger where parent_place_id in (".$sPlaceIDs.") and housenumber = '".pg_escape_string($aSearch['sHouseNumber'])."'";
                                                                        //new query for lines, not housenumbers anymore
                                                                if (!sizeof($aPlaceIDs))
                                                                {
                                                                        //$sSQL = "select place_id from location_property_tiger where parent_place_id in (".$sPlaceIDs.") and housenumber = '".pg_escape_string($aSearch['sHouseNumber'])."'";
                                                                        //new query for lines, not housenumbers anymore

-                                    $this->searchedHousenumber = intval($aSearch['sHouseNumber']);
-                                    if($this->searchedHousenumber%2==0){
+                                    if($searchedHousenumber%2==0){

                                         //if housenumber is even, look for housenumber in streets with interpolationtype even or all
                                         //if housenumber is even, look for housenumber in streets with interpolationtype even or all

-                                        $sSQL = "select distinct place_id from location_property_tiger_line where parent_place_id in (".$sPlaceIDs.") and (interpolationtype='even' or interpolationtype='all') and ".$this->searchedHousenumber.">=startnumber and ".$this->searchedHousenumber."<=endnumber";
+                                        $sSQL = "select distinct place_id from location_property_tiger where parent_place_id in (".$sPlaceIDs.") and (interpolationtype='even' or interpolationtype='all') and ".$searchedHousenumber.">=startnumber and ".$searchedHousenumber."<=endnumber";

                                     }else{
                                         //look for housenumber in streets with interpolationtype odd or all
                                     }else{
                                         //look for housenumber in streets with interpolationtype odd or all

-                                        $sSQL = "select distinct place_id from location_property_tiger_line where parent_place_id in (".$sPlaceIDs.") and (interpolationtype='odd' or interpolationtype='all') and ".$this->searchedHousenumber.">=startnumber and ".$this->searchedHousenumber."<=endnumber";
+                                        $sSQL = "select distinct place_id from location_property_tiger where parent_place_id in (".$sPlaceIDs.") and (interpolationtype='odd' or interpolationtype='all') and ".$searchedHousenumber.">=startnumber and ".$searchedHousenumber."<=endnumber";

                                     }
                     
                                     if (sizeof($this->aExcludePlaceIDs))
                                     }
                     
                                     if (sizeof($this->aExcludePlaceIDs))


@@ -1461,11 +1470,10 @@
                                                                if (!sizeof($aPlaceIDs) && preg_match('/[0-9]+/', $aSearch['sHouseNumber']))
                                                                {
                                                                        $aPlaceIDs = $aRoadPlaceIDs;
                                                                if (!sizeof($aPlaceIDs) && preg_match('/[0-9]+/', $aSearch['sHouseNumber']))
                                                                {
                                                                        $aPlaceIDs = $aRoadPlaceIDs;

-                                    //set to false, if no housenumbers were found
-                                    $this->housenumberFound=false;
+                                    //set to -1, if no housenumbers were found
+                                    $searchedHousenumber=-1;

                                                                }else{
                                                                }else{

-                                    //housenumber was found
-                                    $this->housenumberFound=true;
+                                    //housenumber was found, remains saved in searchedHousenumber

                                 }
                                                        }
 
                                 }
                                                        }
 


@@ -1596,7 +1604,8 @@
 
                                                foreach($aPlaceIDs as $iPlaceID)
                                                {
 
                                                foreach($aPlaceIDs as $iPlaceID)
                                                {

-                                                       $aResultPlaceIDs[$iPlaceID] = $iPlaceID;
+                            // array for placeID => -1 | Tiger housenumber
+                                                       $aResultPlaceIDs[$iPlaceID] = $searchedHousenumber;

                                                }
                                                if ($iQueryLoop > 20) break;
                                        }
                                                }
                                                if ($iQueryLoop > 20) break;
                                        }


@@ -1604,16 +1613,22 @@
                                        if (isset($aResultPlaceIDs) && sizeof($aResultPlaceIDs) && ($this->iMinAddressRank != 0 || $this->iMaxAddressRank != 30))
                                        {
                                                // Need to verify passes rank limits before dropping out of the loop (yuk!)
                                        if (isset($aResultPlaceIDs) && sizeof($aResultPlaceIDs) && ($this->iMinAddressRank != 0 || $this->iMaxAddressRank != 30))
                                        {
                                                // Need to verify passes rank limits before dropping out of the loop (yuk!)

-                                               $sSQL = "select place_id from placex where place_id in (".join(',',$aResultPlaceIDs).") ";
+                        // reduces the number of place id, like a filter
+                                               $sSQL = "select place_id from placex where place_id in (".join(',',array_keys($aResultPlaceIDs)).") ";

                                                $sSQL .= "and (placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
                                                if (14 >= $this->iMinAddressRank && 14 <= $this->iMaxAddressRank) $sSQL .= " OR (extratags->'place') = 'city'";
                                                if ($this->aAddressRankList) $sSQL .= " OR placex.rank_address in (".join(',',$this->aAddressRankList).")";
                                                $sSQL .= "and (placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
                                                if (14 >= $this->iMinAddressRank && 14 <= $this->iMaxAddressRank) $sSQL .= " OR (extratags->'place') = 'city'";
                                                if ($this->aAddressRankList) $sSQL .= " OR placex.rank_address in (".join(',',$this->aAddressRankList).")";

-                                               $sSQL .= ") UNION select place_id from location_property_tiger where place_id in (".join(',',$aResultPlaceIDs).") ";
+                                               $sSQL .= ") UNION select place_id from location_property_tiger where place_id in (".join(',',array_keys($aResultPlaceIDs)).") ";

                                                $sSQL .= "and (30 between $this->iMinAddressRank and $this->iMaxAddressRank ";
                                                if ($this->aAddressRankList) $sSQL .= " OR 30 in (".join(',',$this->aAddressRankList).")";
                                                $sSQL .= ")";
                                                if (CONST_Debug) var_dump($sSQL);
                                                $sSQL .= "and (30 between $this->iMinAddressRank and $this->iMaxAddressRank ";
                                                if ($this->aAddressRankList) $sSQL .= " OR 30 in (".join(',',$this->aAddressRankList).")";
                                                $sSQL .= ")";
                                                if (CONST_Debug) var_dump($sSQL);

-                                               $aResultPlaceIDs = $this->oDB->getCol($sSQL);
+                                               $aFilteredPlaceIDs = $this->oDB->getCol($sSQL);
+                        $tempIDs = array();
+                        foreach($aFilteredPlaceIDs as $placeID){
+                            $tempIDs[$placeID]= $aResultPlaceIDs[$placeID];  //assign housenumber to placeID
+                        }
+                        $aResultPlaceIDs=$tempIDs;

                                        }
 
                                        //exit;
                                        }
 
                                        //exit;


@@ -1705,17 +1720,16 @@
                                {
                                        $aResult['label'] = $aClassType[$aResult['class'].':'.$aResult['type']]['label'];
                                }
                                {
                                        $aResult['label'] = $aClassType[$aResult['class'].':'.$aResult['type']]['label'];
                                }

-                /* Implement this function later. if tag '&addressdetails=1' is set in query
+                // if tag '&addressdetails=1' is set in query

                                if ($this->bIncludeAddressDetails)
                                {
                                if ($this->bIncludeAddressDetails)
                                {

-                *   getAddressDetails() is defined in lib.php and uses the SQL function get_addressdata in functions.sql
-                                       $aResult['address'] = getAddressDetails($this->oDB, $sLanguagePrefArraySQL, $aResult['place_id'], $aResult['country_code']);
+                    // getAddressDetails() is defined in lib.php and uses the SQL function get_addressdata in functions.sql
+                                       $aResult['address'] = getAddressDetails($this->oDB, $sLanguagePrefArraySQL, $aResult['place_id'], $aResult['country_code'], $aResultPlaceIDs[$aResult['place_id']]);

                                        if ($aResult['extra_place'] == 'city' && !isset($aResult['address']['city']))
                                        {
                                                $aResult['address'] = array_merge(array('city' => array_shift(array_values($aResult['address']))), $aResult['address']);
                                        }
                                }
                                        if ($aResult['extra_place'] == 'city' && !isset($aResult['address']['city']))
                                        {
                                                $aResult['address'] = array_merge(array('city' => array_shift(array_values($aResult['address']))), $aResult['address']);
                                        }
                                }

-                */

                                if ($this->bIncludeExtraTags)
                                {
                                        if ($aResult['extra'])
                                if ($this->bIncludeExtraTags)
                                {
                                        if ($aResult['extra'])