X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/4255f5a0d0d7f1bfbb3bbabb03d4627283386916..64661947109e6ed45a1527529b079ddab26d23e5:/website/search.php

diff --git a/website/search.php b/website/search.php
index 2e235ee5..6596fca4 100755
--- a/website/search.php
+++ b/website/search.php
@@ -13,6 +13,7 @@
 	$sOutputFormat = 'html';
 	$aSearchResults = array();
 	$aExcludePlaceIDs = array();
+	$sCountryCodesSQL = false;
 	$sSuggestion = $sSuggestionURL = false;
 	$bDeDupe = isset($_GET['dedupe'])?(bool)$_GET['dedupe']:true;
 	$bReverseInPlan = false;
@@ -37,7 +38,10 @@
 
 	// Prefered language	
 	$aLangPrefOrder = getPrefferedLangauges();
-//	if (isset($aLangPrefOrder['name:de'])) $bReverseInPlan = true;
+	if (isset($aLangPrefOrder['name:de'])) $bReverseInPlan = true;
+	if (isset($aLangPrefOrder['name:ru'])) $bReverseInPlan = true;
+	if (isset($aLangPrefOrder['name:ja'])) $bReverseInPlan = true;
+
 	$sLanguagePrefArraySQL = "ARRAY[".join(',',array_map("getDBQuoted",$aLangPrefOrder))."]";
 
 	if (isset($_GET['exclude_place_ids']) && $_GET['exclude_place_ids'])
@@ -62,10 +66,27 @@
 			break;
 		case 'city':
 			$iMinAddressRank = 14;
-			$iMaxAddressRank = 18;
+			$iMaxAddressRank = 16;
+			break;
+		case 'settlement':
+			$iMinAddressRank = 8;
+			$iMaxAddressRank = 20;
 			break;
 		}
 	}
+
+	if (isset($_GET['countrycodes']))
+	{
+		$aCountryCodes = array();
+		foreach(explode(',',$_GET['countrycodes']) as $sCountryCode)
+		{
+			if (preg_match('/^[a-zA-Z][a-zA-Z]$/', $sCountryCode))
+			{
+				$aCountryCodes[] = "'".strtolower($sCountryCode)."'";
+			}
+		}
+		$sCountryCodesSQL = join(',', $aCountryCodes);
+	}
 		
 	// Search query
 	$sQuery = (isset($_GET['q'])?trim($_GET['q']):'');
@@ -84,7 +105,7 @@
 		$hLog = logStart($oDB, 'search', $sQuery, $aLangPrefOrder);
 
 		// 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, new york', $sQuery);
+                $sQuery = str_ireplace(array('New York, ny','new york, new york', 'New York ny','new york new york'), 'new york city, ny', $sQuery);
 
 		// If we have a view box create the SQL
 		// Small is the actual view box, Large is double (on each axis) that 
@@ -283,6 +304,7 @@
 				var_dump($sSQL, $aDatabaseWords);
 				exit;
 			}
+			$aPossibleMainWordIDs = array();
 			foreach($aDatabaseWords as $aToken)
 			{
 				if (isset($aValidTokens[$aToken['word_token']]))
@@ -293,6 +315,7 @@
 				{
 					$aValidTokens[$aToken['word_token']] = array($aToken);
 				}
+				if ($aToken['word_token'][0]==' ' && !$aToken['class'] && !$aToken['country_code']) $aPossibleMainWordIDs[$aToken['word_id']] = 1;
 			}
 			if (CONST_Debug) var_Dump($aPhrases, $aValidTokens);
 
@@ -452,7 +475,7 @@
 										{
 											if (sizeof($aSearch['aName']))
 											{
-												if (!isset($aValidTokens[$sToken]) || strlen($sToken) < 4)
+												if (!isset($aValidTokens[$sToken]) || strlen($sToken) < 4 || strpos($sToken, ' ') !== false)
 												{
 													$aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
 												}
@@ -470,14 +493,14 @@
 										}
 									}
 								}
-								if (isset($aValidTokens[$sToken]) && strlen($sToken) >= 4)
+								if (isset($aValidTokens[$sToken]))
 								{
 									// Allow searching for a word - but at extra cost
 									foreach($aValidTokens[$sToken] as $aSearchTerm)
 									{
 //var_Dump('<hr>',$aSearch['aName']);
 
-										if (sizeof($aCurrentSearch['aName']))
+										if (sizeof($aCurrentSearch['aName'])  && strlen($sToken) >= 4)
 										{
   										$aSearch = $aCurrentSearch;
 											$aSearch['iSearchRank'] += 1;
@@ -488,7 +511,8 @@
 										if (!sizeof($aCurrentSearch['aName']) || $aCurrentSearch['iNamePhrase'] == $iPhrase)
 										{
   										$aSearch = $aCurrentSearch;
-											$aSearch['iSearchRank'] += 4;
+											$aSearch['iSearchRank'] += 2;
+											if (preg_match('#^[0-9]+$#', $sToken)) $aSearch['iSearchRank'] += 2;
 											$aSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
 											$aSearch['iNamePhrase'] = $iPhrase;
   										if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch;
@@ -511,6 +535,21 @@
 
 						$aNewPhraseSearches = array_merge($aNewPhraseSearches, $aNewWordsetSearches);
 						usort($aNewPhraseSearches, 'bySearchRank');
+
+          $aSearchHash = array();
+          foreach($aNewPhraseSearches as $iSearch => $aSearch)
+          {
+            $sHash = serialize($aSearch);
+            if (isset($aSearchHash[$sHash]))
+            {
+              unset($aNewPhraseSearches[$iSearch]);
+            }
+            else
+            {
+              $aSearchHash[$sHash] = 1;
+            }
+          }
+
 						$aNewPhraseSearches = array_slice($aNewPhraseSearches, 0, 50);
 					}
 
@@ -556,18 +595,24 @@
 				
 				if (CONST_Debug) var_Dump($aGroupedSearches);
 
-				if ($bReverseInPlan && false)
+				if ($bReverseInPlan)
 				{
-					foreach($aGroupedSearches as $iGroup => $aSearches)
+					$aCopyGroupedSearches = $aGroupedSearches;
+					foreach($aCopyGroupedSearches as $iGroup => $aSearches)
 					{
 						foreach($aSearches as $iSearch => $aSearch)
 						{
 							if (sizeof($aSearch['aAddress']))
 							{
-								$aReverseSearch = $aSearch;
 								$iReverseItem = array_pop($aSearch['aAddress']);
-								$aReverseSearch['aName'][$iReverseItem] = $iReverseItem;
-								$aGroupedSearches[$iGroup][] = $aReverseSearch;
+								if (isset($aPossibleMainWordIDs[$iReverseItem]))
+								{
+									$aSearch['aAddress'] = array_merge($aSearch['aAddress'], $aSearch['aName']);
+									$aSearch['aName'] = array($iReverseItem);
+									$aGroupedSearches[$iGroup][] = $aSearch;
+								}
+//								$aReverseSearch['aName'][$iReverseItem] = $iReverseItem;
+	//							$aGroupedSearches[$iGroup][] = $aReverseSearch;
 							}
 						}
 					}
@@ -625,7 +670,9 @@
 								if ($oDB->getOne($sSQL))
 								{
 								$sSQL = "select place_id from place_classtype_".$aSearch['sClass']."_".$aSearch['sType'];								
+								if ($sCountryCodesSQL) $sSQL .= " join placex using (place_id)";
 								$sSQL .= " where st_contains($sViewboxSmallSQL, centroid)";
+								if ($sCountryCodesSQL) $sSQL .= " and country_code in ($sCountryCodesSQL)";								
 								if ($sViewboxCentreSQL)	$sSQL .= " order by st_distance($sViewboxCentreSQL, centroid) asc";
 								$sSQL .= " limit $iLimit";
 								if (CONST_Debug) var_dump($sSQL);
@@ -634,7 +681,9 @@
 								if (!sizeof($aPlaceIDs))
 								{
 									$sSQL = "select place_id from place_classtype_".$aSearch['sClass']."_".$aSearch['sType'];								
+									if ($sCountryCodesSQL) $sSQL .= " join placex using (place_id)";
 									$sSQL .= " where st_contains($sViewboxLargeSQL, centroid)";
+									if ($sCountryCodesSQL) $sSQL .= " and country_code in ($sCountryCodesSQL)";								
 									if ($sViewboxCentreSQL)	$sSQL .= " order by st_distance($sViewboxCentreSQL, centroid) asc";
 									$sSQL .= " limit $iLimit";
 									if (CONST_Debug) var_dump($sSQL);
@@ -645,6 +694,7 @@
 							{
 								$sSQL = "select place_id from placex where class='".$aSearch['sClass']."' and type='".$aSearch['sType']."'";
 								$sSQL .= " and st_contains($sViewboxSmallSQL, centroid)";
+								if ($sCountryCodesSQL) $sSQL .= " and country_code in ($sCountryCodesSQL)";								
 								if ($sViewboxCentreSQL)	$sSQL .= " order by st_distance($sViewboxCentreSQL, centroid) asc";
 								$sSQL .= " limit $iLimit";
 								if (CONST_Debug) var_dump($sSQL);
@@ -674,19 +724,23 @@
 							{
 								$aTerms[] = "place_id not in (".join(',',$aExcludePlaceIDs).")";
 							}
+							if ($sCountryCodesSQL)
+							{
+								$aTerms[] = "country_code in ($sCountryCodesSQL)";
+							}
+
 							if ($bBoundingBoxSearch) $aTerms[] = "centroid && $sViewboxSmallSQL";
 							if ($sNearPointSQL) $aOrder[] = "ST_Distance($sNearPointSQL, centroid) asc";
-							if ($sViewboxSmallSQL) $aOrder[] = "ST_Contains($sViewboxSmallSQL, centroid) desc";
-							if ($sViewboxLargeSQL) $aOrder[] = "ST_Contains($sViewboxLargeSQL, centroid) desc";
-							$aOrder[] = "search_rank ASC";
+
+							$sImportanceSQL = 'case when importance = 0 OR importance IS NULL then 0.92-(search_rank::float/33) else importance end';
+
+							if ($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";
+							$aOrder[] = "$sImportanceSQL DESC";
 						
 							if (sizeof($aTerms))
 							{
 								$sSQL = "select place_id";
-								if ($sViewboxSmallSQL) $sSQL .= ",ST_Contains($sViewboxSmallSQL, centroid) as in_small";
-								else $sSQL .= ",false as in_small";
-								if ($sViewboxLargeSQL) $sSQL .= ",ST_Contains($sViewboxLargeSQL, centroid) as in_large";
-								else $sSQL .= ",false as in_large";
 								$sSQL .= " from search_name";
 								$sSQL .= " where ".join(' and ',$aTerms);
 								$sSQL .= " order by ".join(', ',$aOrder);
@@ -704,19 +758,21 @@
 									var_dump($sSQL, $aViewBoxPlaceIDs);					
 									exit;
 								}
-
+//var_dump($aViewBoxPlaceIDs);
 								// Did we have an viewbox matches?
 								$aPlaceIDs = array();
 								$bViewBoxMatch = false;
 								foreach($aViewBoxPlaceIDs as $aViewBoxRow)
 								{
-									if ($bViewBoxMatch == 1 && $aViewBoxRow['in_small'] == 'f') break;
-									if ($bViewBoxMatch == 2 && $aViewBoxRow['in_large'] == 'f') break;
-									if ($aViewBoxRow['in_small'] == 't') $bViewBoxMatch = 1;
-									else if ($aViewBoxRow['in_large'] == 't') $bViewBoxMatch = 2;
+//									if ($bViewBoxMatch == 1 && $aViewBoxRow['in_small'] == 'f') break;
+//									if ($bViewBoxMatch == 2 && $aViewBoxRow['in_large'] == 'f') break;
+//									if ($aViewBoxRow['in_small'] == 't') $bViewBoxMatch = 1;
+//									else if ($aViewBoxRow['in_large'] == 't') $bViewBoxMatch = 2;
 									$aPlaceIDs[] = $aViewBoxRow['place_id'];
 								}
 							}
+//var_Dump($aPlaceIDs);
+//exit;
 
 							if ($aSearch['sHouseNumber'] && sizeof($aPlaceIDs))
 							{
@@ -760,7 +816,7 @@
 								}
 
 								// Fallback to the road
-								if (!sizeof($aPlaceIDs))
+								if (!sizeof($aPlaceIDs) && preg_match('/^[0-9]+$/', $aSearch['sHouseNumber']))
 								{
 									$aPlaceIDs = $aRoadPlaceIDs;
 								}
@@ -775,6 +831,7 @@
 								{
 									// If they were searching for a named class (i.e. 'Kings Head pub') then we might have an extra match
 									$sSQL = "select place_id from placex where place_id in ($sPlaceIDs) and class='".$aSearch['sClass']."' and type='".$aSearch['sType']."'";
+									if ($sCountryCodesSQL) $sSQL .= " and country_code in ($sCountryCodesSQL)";								
 									$sSQL .= " order by rank_search asc limit $iLimit";
 									if (CONST_Debug) var_dump($sSQL);
 									$aPlaceIDs = $oDB->getCol($sSQL);
@@ -783,6 +840,7 @@
 								if (!$aSearch['sOperator'] || $aSearch['sOperator'] == 'near') // & in
 								{
 									$sSQL = "select rank_search from placex where place_id in ($sPlaceIDs) order by rank_search asc limit 1";
+
 									if (CONST_Debug) var_dump($sSQL);
 									$iMaxRank = ((int)$oDB->getOne($sSQL)) + 5;
 
@@ -791,6 +849,9 @@
 									$aPlaceIDs = $oDB->getCol($sSQL);
 									$sPlaceIDs = join(',',$aPlaceIDs);
 
+									if ($sPlaceIDs)
+									{
+
 									$fRange = 0.01;
 									$sSQL = "select count(*) from pg_tables where tablename = 'place_classtype_".$aSearch['sClass']."_".$aSearch['sType']."'";
 									if ($oDB->getOne($sSQL))
@@ -798,12 +859,14 @@
 										// More efficient - can make the range bigger
   									$fRange = 0.05;
 										$sSQL = "select l.place_id from place_classtype_".$aSearch['sClass']."_".$aSearch['sType']." as l";
+										if ($sCountryCodesSQL) $sSQL .= " join placex as lp using (place_id)";
 										$sSQL .= ",placex as f where ";
 										$sSQL .= "f.place_id in ($sPlaceIDs) and ST_DWithin(l.centroid, st_centroid(f.geometry), $fRange) ";
 										if (sizeof($aExcludePlaceIDs))
 										{
 											$sSQL .= " and l.place_id not in (".join(',',$aExcludePlaceIDs).")";
 										}
+										if ($sCountryCodesSQL) $sSQL .= " and lp.country_code in ($sCountryCodesSQL)";
 										if ($sNearPointSQL) $sSQL .= " order by ST_Distance($sNearPointSQL, l.centroid) ASC";
 										else $sSQL .= " order by ST_Distance(l.centroid, f.geometry) asc";
 										$sSQL .= " limit $iLimit";
@@ -814,18 +877,20 @@
 									{
 										if (isset($aSearch['fRadius']) && $aSearch['fRadius']) $fRange = $aSearch['fRadius'];
 										$sSQL = "select l.place_id from placex as l,placex as f where ";
-										$sSQL .= "f.place_id in ($sPlaceIDs) and ST_DWithin(l.geometry, st_centroid(f.geometry), $fRange) ";
+										$sSQL .= "f.place_id in ( $sPlaceIDs) and ST_DWithin(l.geometry, st_centroid(f.geometry), $fRange) ";
 										$sSQL .= "and l.class='".$aSearch['sClass']."' and l.type='".$aSearch['sType']."' ";
 										if (sizeof($aExcludePlaceIDs))
 										{
 											$sSQL .= " and l.place_id not in (".join(',',$aExcludePlaceIDs).")";
 										}
+										if ($sCountryCodesSQL) $sSQL .= " and l.country_code in ($sCountryCodesSQL)";								
 										if ($sNearPointSQL) $sSQL .= " order by ST_Distance($sNearPointSQL, l.geometry) ASC";
 										else $sSQL .= " order by ST_Distance(l.geometry, f.geometry) asc, l.rank_search ASC";
 										$sSQL .= " limit $iLimit";
 										if (CONST_Debug) var_dump($sSQL);
 										$aPlaceIDs = $oDB->getCol($sSQL);
 									}
+									}
 								}
 							}
 						
@@ -868,10 +933,11 @@
 					$sSQL .= "get_name_by_language(name, $sLanguagePrefArraySQL) as placename,";
 					$sSQL .= "get_name_by_language(name, ARRAY['ref']) as ref,";
 					$sSQL .= "avg(ST_X(ST_Centroid(geometry))) as lon,avg(ST_Y(ST_Centroid(geometry))) as lat, ";
-					$sSQL .= $sOrderSQL." as porder ";
+//					$sSQL .= $sOrderSQL." as porder, ";
+					$sSQL .= "coalesce(importance,0.9-(rank_search::float/30)) as importance ";
 					$sSQL .= "from placex where place_id in ($sPlaceIDs) ";
 					$sSQL .= "and placex.rank_address between $iMinAddressRank and $iMaxAddressRank ";
-					$sSQL .= "group by osm_type,osm_id,class,type,admin_level,rank_search,rank_address,country_code";
+					$sSQL .= "group by osm_type,osm_id,class,type,admin_level,rank_search,rank_address,country_code,importance";
 					if (!$bDeDupe) $sSQL .= ",place_id";
 					$sSQL .= ",get_address_by_language(place_id, $sLanguagePrefArraySQL) ";
 					$sSQL .= ",get_name_by_language(name, $sLanguagePrefArraySQL) ";
@@ -882,24 +948,27 @@
 					$sSQL .= "null as placename,";
 					$sSQL .= "null as ref,";
 					$sSQL .= "avg(ST_X(centroid)) as lon,avg(ST_Y(centroid)) as lat, ";
-					$sSQL .= $sOrderSQL." as porder ";
+//					$sSQL .= $sOrderSQL." as porder, ";
+					$sSQL .= "-0.15 as importance ";
 					$sSQL .= "from location_property_tiger where place_id in ($sPlaceIDs) ";
 					$sSQL .= "and 30 between $iMinAddressRank and $iMaxAddressRank ";
 					$sSQL .= "group by place_id";
 					if (!$bDeDupe) $sSQL .= ",place_id";
 					$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,'us' as country_code,";
+					$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,'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 .= $sOrderSQL." as porder ";
+//					$sSQL .= $sOrderSQL." as porder, ";
+					$sSQL .= "-0.10 as importance ";
 					$sSQL .= "from location_property_aux where place_id in ($sPlaceIDs) ";
 					$sSQL .= "and 30 between $iMinAddressRank and $iMaxAddressRank ";
 					$sSQL .= "group by place_id";
 					if (!$bDeDupe) $sSQL .= ",place_id";
 					$sSQL .= ",get_address_by_language(place_id, $sLanguagePrefArraySQL) ";
-					$sSQL .= "order by rank_search,rank_address,porder asc";
+					$sSQL .= "order by importance desc";
+//					$sSQL .= "order by rank_search,rank_address,porder asc";
 					if (CONST_Debug) var_dump('<hr>',$sSQL);
 					$aSearchResults = $oDB->getAll($sSQL);
 //var_dump($sSQL,$aSearchResults);exit;
@@ -918,9 +987,9 @@
 	{
 		$sSearchResult = 'No Results Found';
 	}
-	
+//var_Dump($aSearchResults);
+//exit;
 	$aClassType = getClassTypesWithImportance();
-
 	foreach($aSearchResults as $iResNum => $aResult)
 	{
 		if (CONST_Search_AreaPolygons || true)
@@ -1033,7 +1102,7 @@
 		}
 
 //if (CONST_Debug) var_dump($aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']);
-
+/*
 		if (isset($aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['importance']) 
 			&& $aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['importance'])
 		{
@@ -1048,15 +1117,15 @@
 		{
 			$aResult['importance'] = 1000000000000000;
 		}
-
+*/
 		$aResult['name'] = $aResult['langaddress'];
 		$aResult['foundorder'] = $iResNum;
 		$aSearchResults[$iResNum] = $aResult;
 	}
-
-//var_dump($aSearchResults);exit;
 	
 	uasort($aSearchResults, 'byImportance');
+
+//var_dump($aSearchResults);exit;
 	
 	$aOSMIDDone = array();
 	$aClassTypeNameDone = array();