]> git.openstreetmap.org Git - nominatim.git/blobdiff - lib/Geocode.php
Merge remote-tracking branch 'upstream/master'
[nominatim.git] / lib / Geocode.php
index cc846f54d1896293143c3a5a452ca70c7229549b..5451c3820dee27bafbccba637151aa8e8598689d 100644 (file)
@@ -15,7 +15,7 @@
 
                protected $aExcludePlaceIDs = array();
                protected $bDeDupe = true;
-               protected $bReverseInPlan = false;
+               protected $bReverseInPlan = true;
 
                protected $iLimit = 20;
                protected $iFinalLimit = 10;
                                $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,";
                                $sSQL .= "group by place_id";
                                if (!$this->bDeDupe) $sSQL .= ",place_id";
                                $sSQL .= ",get_address_by_language(place_id, $sLanguagePrefArraySQL) ";
+                               */
                        }
 
                        $sSQL .= " order by importance desc";
                                                                                {
                                                                                        if (isset($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;
                                                                                                                $aSearch['aAddress'][$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'])
                                                                                                        else
                                                                                                        {
                                                                                                                $aSearch['aAddressNonSearch'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
+                                                                                                               if (preg_match('#^[0-9]+$#', $sToken)) $aSearch['iSearchRank'] += 2;
                                                                                                                if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
                                                                                                        }
                                                                                                }
                                                                                                if (!sizeof($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']] = $aSearchTerm['word_id'];
                                                                        if (CONST_Debug) var_dump($sSQL);
                                                                        $aPlaceIDs = $this->oDB->getCol($sSQL);
                                                                }
+                                                               else
+                                                               {
+                                                                       $aPlaceIDs = array();
+                                                               }
                                                        }
                                                        else
                                                        {
                                                        // TODO: filter out the pointless search terms (2 letter name tokens and less)
                                                        // they might be right - but they are just too darned expensive to run
                                                        if (sizeof($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'] != $aSearch['aAddress'])
                                                        {
                                                                // For infrequent name terms disable index usage for address
                                                                                sizeof($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'],",")."]";
-                                                                       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'])."'";
                                                                $sPlaceIDs = join(',',$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).")";
                                                                $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($aSearch['sHouseNumber'])."'";
                                                                        if (CONST_Debug) var_dump($sSQL);
                                                                        $aPlaceIDs = $this->oDB->getCol($sSQL);
                                                                }
+                                                               */
 
                                                                if (!sizeof($aPlaceIDs))
                                                                {
                                                                                        }
                                                                                        if ($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]);
                                                        {
                                                                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;
                                $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