Merge remote-tracking branch 'upstream/master'

[nominatim.git] / lib / Geocode.php

diff --git a/lib/Geocode.php b/lib/Geocode.php
index 7f1d373319afcfe717a2b233e497630f34e61eb0..5451c3820dee27bafbccba637151aa8e8598689d 100644 (file)
--- a/lib/Geocode.php
+++ b/lib/Geocode.php
@@ -15,7 +15,7 @@
 
                protected $aExcludePlaceIDs = array();
                protected $bDeDupe = true;
 
                protected $aExcludePlaceIDs = array();
                protected $bDeDupe = true;

-               protected $bReverseInPlan = false;
+               protected $bReverseInPlan = true;

 
                protected $iLimit = 20;
                protected $iFinalLimit = 10;
 
                protected $iLimit = 20;
                protected $iFinalLimit = 10;


@@ -413,6 +413,7 @@
                                $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 .= " 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 .= " 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,";


@@ -427,6 +428,7 @@
                                $sSQL .= "group by place_id";
                                if (!$this->bDeDupe) $sSQL .= ",place_id";
                                $sSQL .= ",get_address_by_language(place_id, $sLanguagePrefArraySQL) ";
                                $sSQL .= "group by place_id";
                                if (!$this->bDeDupe) $sSQL .= ",place_id";
                                $sSQL .= ",get_address_by_language(place_id, $sLanguagePrefArraySQL) ";

+                               */

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


@@ -890,6 +892,9 @@
                                                                                                if ($aSearch['sHouseNumber'] === '')
                                                                                                {
                                                                                                        $aSearch['sHouseNumber'] = $sToken;
                                                                                                if ($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[] = $aSearch;
                                                                                                        /*
                                                                                                        // Fall back to not searching for this item (better than nothing)
                                                                                                        if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
                                                                                                        /*
                                                                                                        // Fall back to not searching for this item (better than nothing)


@@ -908,6 +913,7 @@
                                                                                                        $aSearch['sType'] = $aSearchTerm['type'];
                                                                                                        if (sizeof($aSearch['aName'])) $aSearch['sOperator'] = 'name';
                                                                                                        else $aSearch['sOperator'] = 'near'; // near = in for the moment
                                                                                                        $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')
 
                                                                                                        // Do we have a shortcut id?
                                                                                                        if ($aSearch['sOperator'] == 'name')


@@ -954,7 +960,7 @@
                                                                                {
                                                                                        if (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id'])
                                                                                        {
                                                                                {
                                                                                        if (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id'])
                                                                                        {

-                                                                                               if ((!$bStructuredPhrases || $iPhrase > 0) && sizeof($aCurrentSearch['aName']) && strlen($sToken) >= 4)
+                                                                                               if ((!$bStructuredPhrases || $iPhrase > 0) && sizeof($aCurrentSearch['aName']) && strpos($sToken, ' ') === false)

                                                                                                {
                                                                                                        $aSearch = $aCurrentSearch;
                                                                                                        $aSearch['iSearchRank'] += 1;
                                                                                                {
                                                                                                        $aSearch = $aCurrentSearch;
                                                                                                        $aSearch['iSearchRank'] += 1;


@@ -963,8 +969,11 @@
                                                                                                                $aSearch['aAddress'][$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])) // revert to the token version?
+                                                                                                       elseif (isset($aValidTokens[' '.$sToken]) && strlen($sToken) >= 4) // revert to the token version?

                                                                                                        {
                                                                                                        {

+                                                                                                               $aSearch['aAddressNonSearch'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
+                                                                                                               $aSearch['iSearchRank'] += 1;
+                                                                                                               if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;

                                                                                                                foreach($aValidTokens[' '.$sToken] as $aSearchTermToken)
                                                                                                                {
                                                                                                                        if (empty($aSearchTermToken['country_code'])
                                                                                                                foreach($aValidTokens[' '.$sToken] as $aSearchTermToken)
                                                                                                                {
                                                                                                                        if (empty($aSearchTermToken['country_code'])


@@ -981,6 +990,7 @@
                                                                                                        else
                                                                                                        {
                                                                                                                $aSearch['aAddressNonSearch'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
                                                                                                        else
                                                                                                        {
                                                                                                                $aSearch['aAddressNonSearch'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];

+                                                                                                               if (preg_match('#^[0-9]+$#', $sToken)) $aSearch['iSearchRank'] += 2;

                                                                                                                if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
                                                                                                        }
                                                                                                }
                                                                                                                if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
                                                                                                        }
                                                                                                }


@@ -988,7 +998,8 @@
                                                                                                if (!sizeof($aCurrentSearch['aName']) || $aCurrentSearch['iNamePhrase'] == $iPhrase)
                                                                                                {
                                                                                                        $aSearch = $aCurrentSearch;
                                                                                                if (!sizeof($aCurrentSearch['aName']) || $aCurrentSearch['iNamePhrase'] == $iPhrase)
                                                                                                {
                                                                                                        $aSearch = $aCurrentSearch;

-                                                                                                       $aSearch['iSearchRank'] += 2;
+                                                                                                       $aSearch['iSearchRank'] += 1;
+                                                                                                       if (!sizeof($aCurrentSearch['aName'])) $aSearch['iSearchRank'] += 1;

                                                                                                        if (preg_match('#^[0-9]+$#', $sToken)) $aSearch['iSearchRank'] += 2;
                                                                                                        if ($aWordFrequencyScores[$aSearchTerm['word_id']] < CONST_Max_Word_Frequency)
                                                                                                                $aSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
                                                                                                        if (preg_match('#^[0-9]+$#', $sToken)) $aSearch['iSearchRank'] += 2;
                                                                                                        if ($aWordFrequencyScores[$aSearchTerm['word_id']] < CONST_Max_Word_Frequency)
                                                                                                                $aSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];


@@ -1177,6 +1188,10 @@
                                                                        if (CONST_Debug) var_dump($sSQL);
                                                                        $aPlaceIDs = $this->oDB->getCol($sSQL);
                                                                }
                                                                        if (CONST_Debug) var_dump($sSQL);
                                                                        $aPlaceIDs = $this->oDB->getCol($sSQL);
                                                                }

+                                                               else
+                                                               {
+                                                                       $aPlaceIDs = array();
+                                                               }

                                                        }
                                                        else
                                                        {
                                                        }
                                                        else
                                                        {


@@ -1201,7 +1216,8 @@
                                                                        // 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 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)";
                                                                        {
                                                                                $sSQL = "select place_id from place_classtype_".$aSearch['sClass']."_".$aSearch['sType']." ct";
                                                                                if ($sCountryCodesSQL) $sSQL .= " join placex using (place_id)";


@@ -1236,7 +1252,7 @@
                                                        // 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'],",")."]";

-                                                       if (sizeof($aSearch['aNameNonSearch'])) $aTerms[] = "array_cat(name_vector,ARRAY[]::integer[]) @> ARRAY[".join($aSearch['aNameNonSearch'],",")."]";
+                                                       //if (sizeof($aSearch['aNameNonSearch'])) $aTerms[] = "array_cat(name_vector,ARRAY[]::integer[]) @> ARRAY[".join($aSearch['aNameNonSearch'],",")."]";

                                                        if (sizeof($aSearch['aAddress']) && $aSearch['aName'] != $aSearch['aAddress'])
                                                        {
                                                                // For infrequent name terms disable index usage for address
                                                        if (sizeof($aSearch['aAddress']) && $aSearch['aName'] != $aSearch['aAddress'])
                                                        {
                                                                // For infrequent name terms disable index usage for address


@@ -1244,12 +1260,13 @@
                                                                                sizeof($aSearch['aName']) == 1 &&
                                                                                $aWordFrequencyScores[$aSearch['aName'][reset($aSearch['aName'])]] < CONST_Search_NameOnlySearchFrequencyThreshold)
                                                                {
                                                                                sizeof($aSearch['aName']) == 1 &&
                                                                                $aWordFrequencyScores[$aSearch['aName'][reset($aSearch['aName'])]] < CONST_Search_NameOnlySearchFrequencyThreshold)
                                                                {

-                                                                       $aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join(array_merge($aSearch['aAddress'],$aSearch['aAddressNonSearch']),",")."]";
+                                                                       //$aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join(array_merge($aSearch['aAddress'],$aSearch['aAddressNonSearch']),",")."]";
+                                                                       $aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join($aSearch['aAddress'],",")."]";

                                                                }
                                                                else
                                                                {
                                                                        $aTerms[] = "nameaddress_vector @> ARRAY[".join($aSearch['aAddress'],",")."]";
                                                                }
                                                                else
                                                                {
                                                                        $aTerms[] = "nameaddress_vector @> ARRAY[".join($aSearch['aAddress'],",")."]";

-                                                                       if (sizeof($aSearch['aAddressNonSearch'])) $aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join($aSearch['aAddressNonSearch'],",")."]";
+                                                                       //if (sizeof($aSearch['aAddressNonSearch'])) $aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join($aSearch['aAddressNonSearch'],",")."]";

                                                                }
                                                        }
                                                        if ($aSearch['sCountryCode']) $aTerms[] = "country_code = '".pg_escape_string($aSearch['sCountryCode'])."'";
                                                                }
                                                        }
                                                        if ($aSearch['sCountryCode']) $aTerms[] = "country_code = '".pg_escape_string($aSearch['sCountryCode'])."'";


@@ -1334,8 +1351,8 @@
                                                                $sPlaceIDs = join(',',$aPlaceIDs);
 
                                                                // Now they are indexed look for a house attached to a street we found
                                                                $sPlaceIDs = join(',',$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 (sizeof($this->aExcludePlaceIDs))
                                                                {
                                                                        $sSQL .= " and place_id not in (".join(',',$this->aExcludePlaceIDs).")";


@@ -1345,6 +1362,7 @@
                                                                $aPlaceIDs = $this->oDB->getCol($sSQL);
 
                                                                // If not try the aux fallback table
                                                                $aPlaceIDs = $this->oDB->getCol($sSQL);
 
                                                                // If not try the aux fallback table

+                                                               /*

                                                                if (!sizeof($aPlaceIDs))
                                                                {
                                                                        $sSQL = "select place_id from location_property_aux where parent_place_id in (".$sPlaceIDs.") and housenumber = '".pg_escape_string($aSearch['sHouseNumber'])."'";
                                                                if (!sizeof($aPlaceIDs))
                                                                {
                                                                        $sSQL = "select place_id from location_property_aux where parent_place_id in (".$sPlaceIDs.") and housenumber = '".pg_escape_string($aSearch['sHouseNumber'])."'";


@@ -1356,6 +1374,7 @@
                                                                        if (CONST_Debug) var_dump($sSQL);
                                                                        $aPlaceIDs = $this->oDB->getCol($sSQL);
                                                                }
                                                                        if (CONST_Debug) var_dump($sSQL);
                                                                        $aPlaceIDs = $this->oDB->getCol($sSQL);
                                                                }

+                                                               */

 
                                                                if (!sizeof($aPlaceIDs))
                                                                {
 
                                                                if (!sizeof($aPlaceIDs))
                                                                {


@@ -1458,7 +1477,7 @@
                                                                                        }
                                                                                        if ($sCountryCodesSQL) $sSQL .= " and lp.calculated_country_code in ($sCountryCodesSQL)";
                                                                                        if ($sOrderBySQL) $sSQL .= "order by ".$sOrderBySQL." asc";
                                                                                        }
                                                                                        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));
                                                                                        $sSQL .= " limit $this->iLimit";
                                                                                        if (CONST_Debug) var_dump($sSQL);
                                                                                        $aClassPlaceIDs = array_merge($aClassPlaceIDs, $this->oDB->getCol($sSQL));


@@ -1480,7 +1499,7 @@
                                                                                        }
                                                                                        if ($sCountryCodesSQL) $sSQL .= " and l.calculated_country_code in ($sCountryCodesSQL)";
                                                                                        if ($sOrderBy) $sSQL .= "order by ".$OrderBysSQL." asc";
                                                                                        }
                                                                                        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));
                                                                                        $sSQL .= " limit $this->iLimit";
                                                                                        if (CONST_Debug) var_dump($sSQL);
                                                                                        $aClassPlaceIDs = array_merge($aClassPlaceIDs, $this->oDB->getCol($sSQL));


@@ -1561,7 +1580,7 @@
                        }
 
                        $aClassType = getClassTypesWithImportance();
                        }
 
                        $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]);
                        foreach($aRecheckWords as $i => $sWord)
                        {
                                if (!$sWord) unset($aRecheckWords[$i]);


@@ -1607,10 +1626,12 @@
                                                        {
                                                                preg_match_all('/(-?[0-9.]+) (-?[0-9.]+)/',$aMatch[1],$aPolyPoints,PREG_SET_ORDER);
                                                        }
                                                        {
                                                                preg_match_all('/(-?[0-9.]+) (-?[0-9.]+)/',$aMatch[1],$aPolyPoints,PREG_SET_ORDER);
                                                        }

+                            /*

                                                        elseif (preg_match('#MULTIPOLYGON\\(\\(\\(([- 0-9.,]+)#',$aPointPolygon['astext'],$aMatch))
                                                        {
                                                                preg_match_all('/(-?[0-9.]+) (-?[0-9.]+)/',$aMatch[1],$aPolyPoints,PREG_SET_ORDER);
                                                        }
                                                        elseif (preg_match('#MULTIPOLYGON\\(\\(\\(([- 0-9.,]+)#',$aPointPolygon['astext'],$aMatch))
                                                        {
                                                                preg_match_all('/(-?[0-9.]+) (-?[0-9.]+)/',$aMatch[1],$aPolyPoints,PREG_SET_ORDER);
                                                        }

+                            */

                                                        elseif (preg_match('#POINT\\((-?[0-9.]+) (-?[0-9.]+)\\)#',$aPointPolygon['astext'],$aMatch))
                                                        {
                                                                $fRadius = 0.01;
                                                        elseif (preg_match('#POINT\\((-?[0-9.]+) (-?[0-9.]+)\\)#',$aPointPolygon['astext'],$aMatch))
                                                        {
                                                                $fRadius = 0.01;


@@ -1722,7 +1743,11 @@
                                $sAddress = $aResult['langaddress'];
                                foreach($aRecheckWords as $i => $sWord)
                                {
                                $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
                                }
 
                                $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


@@ -1786,17 +1811,3 @@
 
        } // end class
 
 
        } // 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;
-               }
-*/