]> git.openstreetmap.org Git - nominatim.git/blobdiff - lib/lib.php
Merge branch 'master' of github.com:twain47/Nominatim
[nominatim.git] / lib / lib.php
index e384949e7bcb8d503be6f27831730e951ab61b32..1564a83cceffa45dc4a1f65ef29cce3d87cb0608 100644 (file)
@@ -1,10 +1,35 @@
 <?php
 
+       function failInternalError($sError, $sSQL = false, $vDumpVar = false) 
+       {
+               header('HTTP/1.0 500 Internal Server Error');
+               header('Content-type: text/html; charset=utf-8');
+               echo "<html><body><h1>Internal Server Error</h1>";
+               echo '<p>Nominatim has encountered an internal error while processing your request. This is most likely because of a bug in the software.</p>';
+               echo "<p><b>Details:</b> ".$sError,"</p>";
+               echo '<p>Feel free to report the bug in the <a href="http://trac.openstreetmap.org">OSM bug database</a>. Please include the error message above an the URL you used.</p>';
+               if (CONST_Debug)
+               {
+                       echo "<hr><h2>Debugging Information</h2><br>";
+                       if ($sSQL) {
+                               echo "<h3>SQL query</h3><code>".$sSQL."</code>";
+                       }
+                       if ($vDumpVar) {
+                               echo "<h3>Result</h3> <code>";
+                               var_dump($vDumpVar);
+                               echo "</code>";
+                       }
+               }
+               echo "\n</body></html>\n";
+               exit;
+
+       }
+
        function fail($sError, $sUserError = false)
        {
                if (!$sUserError) $sUserError = $sError;
-               log('ERROR:'.$sError);
-               echo $sUserError;
+               error_log('ERROR: '.$sError);
+               echo $sUserError."\n";
                exit;
        }
 
                 $aLoadAverage = explode(' ',$sLoadAverage);
                return (int)$aLoadAverage[0];
        }
-       
+
+       function getProcessorCount()
+       {
+               $sCPU = file_get_contents('/proc/cpuinfo');
+               preg_match_all('#processor      : [0-9]+#', $sCPU, $aMatches);
+               return sizeof($aMatches[0]);
+       }
+
        function bySearchRank($a, $b)
        {
                if ($a['iSearchRank'] == $b['iSearchRank']) return 0;
 
        function byImportance($a, $b)
        {
+               if ($a['importance'] != $b['importance'])
+                       return ($a['importance'] > $b['importance']?-1:1);
+/*
                if ($a['aPointPolygon']['numfeatures'] != $b['aPointPolygon']['numfeatures'])
                        return ($a['aPointPolygon']['numfeatures'] > $b['aPointPolygon']['numfeatures']?-1:1);
                if ($a['aPointPolygon']['area'] != $b['aPointPolygon']['area'])
                        return ($a['aPointPolygon']['area'] > $b['aPointPolygon']['area']?-1:1);
-               if ($a['importance'] != $b['importance'])
-                       return ($a['importance'] < $b['importance']?-1:1);
+//             if ($a['levenshtein'] != $b['levenshtein'])
+//                     return ($a['levenshtein'] < $b['levenshtein']?-1:1);
+               if ($a['rank_search'] != $b['rank_search'])
+                       return ($a['rank_search'] < $b['rank_search']?-1:1);
+*/
                return ($a['foundorder'] < $b['foundorder']?-1:1);
        }
 
@@ -49,7 +87,7 @@
                {
                        $_SERVER["HTTP_ACCEPT_LANGUAGE"] = $_GET['accept-language'];
                }
-               
+
                $aLanguages = array();
                if (preg_match_all('/(([a-z]{1,8})(-[a-z]{1,8})?)\s*(;\s*q\s*=\s*(1|0\.[0-9]+))?/i', $_SERVER['HTTP_ACCEPT_LANGUAGE'], $aLanguagesParse, PREG_SET_ORDER))
                {
        function gbPostcodeCalculate($sPostcode, $sPostcodeSector, $sPostcodeEnd, &$oDB)
        {
                // Try an exact match on the gb_postcode table
-               $sSQL = 'select \'AA\', ST_X(ST_Centroid(geometry)) as lon,ST_Y(ST_Centroid(geometry)) as lat from gb_postcode where upper(postcode) = \''.$sPostcode.'\'';
+               $sSQL = 'select \'AA\', ST_X(ST_Centroid(geometry)) as lon,ST_Y(ST_Centroid(geometry)) as lat from gb_postcode where postcode = \''.$sPostcode.'\'';
                $aNearPostcodes = $oDB->getAll($sSQL);
                if (PEAR::IsError($aNearPostcodes))
                {
                        var_dump($sSQL, $aNearPostcodes);
                        exit;
                }
-
-               if (!sizeof($aNearPostcodes))
-               {
-                       $sSQL = 'select substring(upper(postcode) from \'^[A-Z][A-Z]?[0-9][0-9A-Z]? [0-9]([A-Z][A-Z])$\'),ST_X(ST_Centroid(geometry)) as lon,ST_Y(ST_Centroid(geometry)) as lat from placex where country_code::text = \'gb\'::text AND substring(upper(postcode) from \'^([A-Z][A-Z]?[0-9][0-9A-Z]? [0-9])[A-Z][A-Z]$\') = \''.$sPostcodeSector.'\' and class=\'place\' and type=\'postcode\' ';
-                       $sSQL .= ' union ';
-                       $sSQL .= 'select substring(upper(postcode) from \'^[A-Z][A-Z]?[0-9][0-9A-Z]? [0-9]([A-Z][A-Z])$\'),ST_X(ST_Centroid(geometry)) as lon,ST_Y(ST_Centroid(geometry)) as lat from gb_postcode where substring(upper(postcode) from \'^([A-Z][A-Z]?[0-9][0-9A-Z]? [0-9])[A-Z][A-Z]$\') = \''.$sPostcodeSector.'\'';
-                       $aNearPostcodes = $oDB->getAll($sSQL);
-                       if (PEAR::IsError($aNearPostcodes))
-                       {
-                               var_dump($sSQL, $aNearPostcodes);
-                               exit;
-                       }
-               }
-
-               if (!sizeof($aNearPostcodes))
+               
+               if (sizeof($aNearPostcodes))
                {
-                       return false;
+                       return array(array('lat' => $aNearPostcodes[0]['lat'], 'lon' => $aNearPostcodes[0]['lon'], 'radius' => 0.005));
                }
 
-               $fTotalLat = 0;
-               $fTotalLon = 0;
-               $fTotalFac = 0;
-               foreach($aNearPostcodes as $aPostcode)
-               {
-                       $iDiff = gbPostcodeAlphaDifference($sPostcodeEnd, $aPostcode['substring'])*2 + 1;
-                       if ($iDiff == 0)
-                               $fFac = 1;
-                       else
-                               $fFac = 1/($iDiff*$iDiff);
-                       
-                       $fTotalFac += $fFac;
-                       $fTotalLat += $aPostcode['lat'] * $fFac;
-                       $fTotalLon += $aPostcode['lon'] * $fFac;
-               }
-               if ($fTotalFac)
-               {
-                       $fLat = $fTotalLat / $fTotalFac;
-                       $fLon = $fTotalLon / $fTotalFac;
-                       $fRadius = min(0.1 / $fTotalFac, 0.02);
-                       return array(array('lat' => $fLat, 'lon' => $fLon, 'radius' => $fRadius));
-               }
                return false;
-
-               /*
-                       $fTotalFac is a suprisingly good indicator of accuracy
-                       $iZoom = 18 + round(log($fTotalFac,32));
-                       $iZoom = max(13,min(18,$iZoom));
-               */
        }
 
        function usPostcodeCalculate($sPostcode, &$oDB)
        function getClassTypes()
        {
                return array(
- 'boundary:adminitrative:2' => array('label'=>'Country','frequency'=>0,'icon'=>'poi_boundary_administrative', 'defdiameter' => 0.32,),
- 'boundary:adminitrative:4' => array('label'=>'State','frequency'=>0,'icon'=>'poi_boundary_administrative', 'defdiameter' => 0.32,),
- 'boundary:adminitrative:5' => array('label'=>'State District','frequency'=>0,'icon'=>'poi_boundary_administrative', 'defdiameter' => 0.32,),
- 'boundary:adminitrative:6' => array('label'=>'County','frequency'=>0,'icon'=>'poi_boundary_administrative', 'defdiameter' => 0.32,),
- 'boundary:adminitrative:8' => array('label'=>'City','frequency'=>0,'icon'=>'poi_boundary_administrative', 'defdiameter' => 0.32,),
- 'boundary:adminitrative:9' => array('label'=>'City District','frequency'=>0,'icon'=>'poi_boundary_administrative', 'defdiameter' => 0.32,),
- 'boundary:adminitrative:10' => array('label'=>'Suburb','frequency'=>0,'icon'=>'poi_boundary_administrative', 'defdiameter' => 0.32,),
- 'boundary:adminitrative:11' => array('label'=>'Neighbourhood','frequency'=>0,'icon'=>'poi_boundary_administrative', 'defdiameter' => 0.32,),
- 'place:city' => array('label'=>'City','frequency'=>66,'icon'=>'poi_place_city','defzoom'=>12, 'defdiameter' => 0.32,),
+ 'boundary:administrative:2' => array('label'=>'Country','frequency'=>0,'icon'=>'poi_boundary_administrative', 'defdiameter' => 0.32,),
  'place:country' => array('label'=>'Country','frequency'=>0,'icon'=>'poi_boundary_administrative','defzoom'=>6, 'defdiameter' => 15,),
+ 'boundary:administrative:4' => array('label'=>'State','frequency'=>0,'icon'=>'poi_boundary_administrative', 'defdiameter' => 0.32,),
  'place:state' => array('label'=>'State','frequency'=>0,'icon'=>'poi_boundary_administrative','defzoom'=>8, 'defdiameter' => 5.12,),
- 'place:region' => array('label'=>'State','frequency'=>0,'icon'=>'poi_boundary_administrative','defzoom'=>8, 'defdiameter' => 5.12,),
- 'place:island' => array('label'=>'Island','frequency'=>288,'icon'=>'','defzoom'=>11, 'defdiameter' => 0.64,),
+ 'boundary:administrative:5' => array('label'=>'State District','frequency'=>0,'icon'=>'poi_boundary_administrative', 'defdiameter' => 0.32,),
+ 'boundary:administrative:6' => array('label'=>'County','frequency'=>0,'icon'=>'poi_boundary_administrative', 'defdiameter' => 0.32,),
  'place:county' => array('label'=>'County','frequency'=>108,'icon'=>'poi_boundary_administrative','defzoom'=>10, 'defdiameter' => 1.28,),
- 'boundary:adminitrative' => array('label'=>'Administrative','frequency'=>413,'icon'=>'poi_boundary_administrative', 'defdiameter' => 0.32,),
+ 'boundary:administrative:8' => array('label'=>'City','frequency'=>0,'icon'=>'poi_boundary_administrative', 'defdiameter' => 0.32,),
+ 'place:city' => array('label'=>'City','frequency'=>66,'icon'=>'poi_place_city','defzoom'=>12, 'defdiameter' => 0.32,),
+ 'boundary:administrative:9' => array('label'=>'City District','frequency'=>0,'icon'=>'poi_boundary_administrative', 'defdiameter' => 0.32,),
+ 'boundary:administrative:10' => array('label'=>'Suburb','frequency'=>0,'icon'=>'poi_boundary_administrative', 'defdiameter' => 0.32,),
+ 'boundary:administrative:11' => array('label'=>'Neighbourhood','frequency'=>0,'icon'=>'poi_boundary_administrative', 'defdiameter' => 0.32,),
+ 'place:region' => array('label'=>'Region','frequency'=>0,'icon'=>'poi_boundary_administrative','defzoom'=>8, 'defdiameter' => 5.12,),
+ 'place:island' => array('label'=>'Island','frequency'=>288,'icon'=>'','defzoom'=>11, 'defdiameter' => 0.64,),
+ 'boundary:administrative' => array('label'=>'Administrative','frequency'=>413,'icon'=>'poi_boundary_administrative', 'defdiameter' => 0.32,),
  'place:town' => array('label'=>'Town','frequency'=>1497,'icon'=>'poi_place_town','defzoom'=>14, 'defdiameter' => 0.08,),
  'place:village' => array('label'=>'Village','frequency'=>11230,'icon'=>'poi_place_village','defzoom'=>15, 'defdiameter' => 0.04,),
  'place:hamlet' => array('label'=>'Hamlet','frequency'=>7075,'icon'=>'poi_place_village','defzoom'=>15, 'defdiameter' => 0.04,),
  'amenity:shop' => array('label'=>'Shop','frequency'=>61,'icon'=>'',),
 
  'place:house' => array('label'=>'House','frequency'=>2086,'icon'=>'','defzoom'=>18,),
+ 'place:house_name' => array('label'=>'House','frequency'=>2086,'icon'=>'','defzoom'=>18,),
+ 'place:house_number' => array('label'=>'House Number','frequency'=>2086,'icon'=>'','defzoom'=>18,),
+ 'place:country_code' => array('label'=>'Country Code','frequency'=>2086,'icon'=>'','defzoom'=>18,),
 
 //
 
  'amenity:shopping' => array('label'=>'Shopping','frequency'=>21,'icon'=>'',),
  'natural:scrub' => array('label'=>'Scrub','frequency'=>20,'icon'=>'',),
  'natural:fen' => array('label'=>'Fen','frequency'=>20,'icon'=>'',),
+ 'building:yes' => array('label'=>'Building','frequency'=>200,'icon'=>'',),
 
  'amenity:parking' => array('label'=>'Parking','frequency'=>3157,'icon'=>'',),
  'highway:bus_stop' => array('label'=>'Bus Stop','frequency'=>35777,'icon'=>'transport_bus_stop2',),
                 else
                 {
                         if (is_bool($xVal)) return $xVal?'true':'false';
-                       if (is_numeric($xVal)) return $xVal;
+//                     if (is_numeric($xVal)) return $xVal;
                         return '"'.str_replace('>','\\>',str_replace(array("\n","\r"),'\\n',str_replace(array("\n\r","\r\n"),'\\n',str_replace('"','\\"',$xVal)))).'"';
                 }
         }
        function _debugDumpGroupedSearches($aData, $aTokens)
        {
                $aWordsIDs = array();
-               foreach($aTokens as $sToken => $aWords)
+               if ($aTokens)
                {
-                       foreach($aWords as $aToken)
+                       foreach($aTokens as $sToken => $aWords)
                        {
-                               $aWordsIDs[$aToken['word_id']] = $sToken.'('.$aToken['word_id'].')';
+                               if ($aWords)
+                               {
+                                       foreach($aWords as $aToken)
+                                       {
+                                               $aWordsIDs[$aToken['word_id']] = $sToken.'('.$aToken['word_id'].')';
+                                       }
+                               }
                        }
                }
                echo "<table border=\"1\">";
                                $sSep = '';
                                foreach($aRow['aAddress'] as $iWordID)
                                {
-//                                     if (!isset($aRow['aName'][$iWordID]))
-                                       {
-                                               echo $sSep.'#'.$aWordsIDs[$iWordID].'#';
-                                               $sSep = ', ';
-                                       }
+                                       echo $sSep.'#'.$aWordsIDs[$iWordID].'#';
+                                       $sSep = ', ';
                                }
                                echo "</td>";
 
 
        function getAddressDetails(&$oDB, $sLanguagePrefArraySQL, $iPlaceID, $sCountryCode = false, $bRaw = false)
        {
-               $aHouseNumber = $oDB->getRow('select housenumber, get_name_by_language(name,ARRAY[\'addr:housename\']) as housename,rank_search from placex where place_id = '.$iPlaceID);
+               $sSQL = "select *,get_name_by_language(name,$sLanguagePrefArraySQL) as localname from get_addressdata($iPlaceID)";
+               IF (!$bRaw) $sSQL .= " WHERE isaddress OR type = 'country_code'";
+               $sSQL .= " order by rank_address desc,isaddress desc";
+               $aAddressLines = $oDB->getAll($sSQL);
+               if (PEAR::IsError($aAddressLines))
+               {
+                       var_dump($aAddressLines);
+                       exit;
+               }
+               if ($bRaw) return $aAddressLines;
+//echo "<pre>";
+//var_dump($aAddressLines);
+               $aAddress = array();
+               $aClassType = getClassTypes();
+               foreach($aAddressLines as $aLine)
+               {
+                       $aTypeLabel = false;
+                       if (isset($aClassType[$aLine['class'].':'.$aLine['type'].':'.$aLine['admin_level']])) $aTypeLabel = $aClassType[$aLine['class'].':'.$aLine['type'].':'.$aLine['admin_level']];
+                       elseif (isset($aClassType[$aLine['class'].':'.$aLine['type']])) $aTypeLabel = $aClassType[$aLine['class'].':'.$aLine['type']];
+                       else $aTypeLabel = array('simplelabel'=>$aLine['class']);
+                       if ($aTypeLabel && ($aLine['localname'] || $aLine['housenumber']))
+                       {
+                               $sTypeLabel = strtolower(isset($aTypeLabel['simplelabel'])?$aTypeLabel['simplelabel']:$aTypeLabel['label']);
+                               $sTypeLabel = str_replace(' ','_',$sTypeLabel);
+                               if (!isset($aAddress[$sTypeLabel]) && $aLine['localname']) $aAddress[$sTypeLabel] = $aLine['localname']?$aLine['localname']:$aLine['housenumber'];
+                       }
+               }
+//var_dump($aAddress);
+//exit;
+               return $aAddress;
+
+               $aHouseNumber = $oDB->getRow('select housenumber, get_name_by_language(name,ARRAY[\'addr:housename\']) as housename,rank_search,postcode from placex where place_id = '.$iPlaceID);
                $sHouseNumber = $aHouseNumber['housenumber'];
                $sHouseName = $aHouseNumber['housename'];
+               $sPostcode = $aHouseNumber['postcode'];
                $iRank = $aHouseNumber['rank_search'];
 
                // Address
                $sSQL .= " length(name::text) as namelength ";
                $sSQL .= " from place_addressline join placex on (address_place_id = placex.place_id)";
                $sSQL .= " where place_addressline.place_id = $iPlaceID and (rank_address > 0 OR address_place_id = $iPlaceID)";
-// and isaddress";
-               if ($sCountryCode)
-               {
-                       $sSQL .= " and (placex.country_code IS NULL OR placex.country_code = '".$sCountryCode."' OR rank_address < 4)";
-               }
-               $sSQL .= " order by cached_rank_address desc,fromarea desc,distance asc,rank_search desc,namelength desc";
+               if (!$bRaw) $sSQL .= " and isaddress";
+               $sSQL .= " order by cached_rank_address desc,isaddress desc,fromarea desc,distance asc,rank_search desc,namelength desc";
 //var_dump($sSQL);
                $aAddressLines = $oDB->getAll($sSQL);
                if (PEAR::IsError($aAddressLines))
                                if ($aTypeLabel && ($aLine['localname'] || $aLine['housenumber']))
                                {
                                        $sTypeLabel = strtolower(isset($aTypeLabel['simplelabel'])?$aTypeLabel['simplelabel']:$aTypeLabel['label']);
+                                       $sTypeLabel = str_replace(' ','_',$sTypeLabel);
                                        if (!isset($aAddress[$sTypeLabel]) && $aLine['localname']) $aAddress[$sTypeLabel] = $aLine['localname']?$aLine['localname']:$aLine['housenumber'];
                                }
                                $iMinRank = $aLine['rank_address'];
                        }
                }
+
+               if ($sPostcode)
+               {
+                       $aAddress['postcode'] = $sPostcode;
+               }
+
                if ($iMinRank > 4 && $sCountryCode)
                {
                        $sSQL = "select get_name_by_language(country_name.name,$sLanguagePrefArraySQL) as name";
                $aSimilar = $oDB->getAll($sSQL);
                return $aSimilar;
        }
+
+       function geocodeReverse($fLat, $fLon, $iZoom=18)
+       {
+               $oDB =& getDB();
+
+               $sPointSQL = "ST_SetSRID(ST_Point($fLon,$fLat),4326)";
+
+               // Zoom to rank, this could probably be calculated but a lookup gives fine control
+               $aZoomRank = array(
+                       0 => 2, // Continent / Sea
+                       1 => 2,
+                       2 => 2,
+                       3 => 4, // Country
+                       4 => 4,
+                       5 => 8, // State
+                       6 => 10, // Region
+                       7 => 10, 
+                       8 => 12, // County
+                       9 => 12,  
+                       10 => 17, // City
+                       11 => 17, 
+                       12 => 18, // Town / Village
+                       13 => 18, 
+                       14 => 22, // Suburb
+                       15 => 22,
+                       16 => 26, // Street, TODO: major street?
+                       17 => 26, 
+                       18 => 30, // or >, Building
+                       19 => 30, // or >, Building
+                       );
+               $iMaxRank = isset($aZoomRank[$iZoom])?$aZoomRank[$iZoom]:28;
+
+               // Find the nearest point
+               $fSearchDiam = 0.0001;
+               $iPlaceID = null;
+               $aArea = false;
+               $fMaxAreaDistance = 1;
+               while(!$iPlaceID && $fSearchDiam < $fMaxAreaDistance)
+               {
+                       $fSearchDiam = $fSearchDiam * 2;
+
+                       // If we have to expand the search area by a large amount then we need a larger feature
+                       // then there is a limit to how small the feature should be
+                       if ($fSearchDiam > 2 && $iMaxRank > 4) $iMaxRank = 4;
+                       if ($fSearchDiam > 1 && $iMaxRank > 9) $iMaxRank = 8;
+                       if ($fSearchDiam > 0.8 && $iMaxRank > 10) $iMaxRank = 10;
+                       if ($fSearchDiam > 0.6 && $iMaxRank > 12) $iMaxRank = 12;
+                       if ($fSearchDiam > 0.2 && $iMaxRank > 17) $iMaxRank = 17;
+                       if ($fSearchDiam > 0.1 && $iMaxRank > 18) $iMaxRank = 18;
+                       if ($fSearchDiam > 0.008 && $iMaxRank > 22) $iMaxRank = 22;
+                       if ($fSearchDiam > 0.001 && $iMaxRank > 26) $iMaxRank = 26;
+
+                       $sSQL = 'select place_id,parent_place_id from placex';
+                       $sSQL .= ' WHERE ST_DWithin('.$sPointSQL.', geometry, '.$fSearchDiam.')';
+                       $sSQL .= ' and rank_search != 28 and rank_search >= '.$iMaxRank;
+                       $sSQL .= ' and (name is not null or housenumber is not null)';
+                       $sSQL .= ' and class not in (\'waterway\')';
+                       $sSQL .= ' and (ST_GeometryType(geometry) not in (\'ST_Polygon\',\'ST_MultiPolygon\') ';
+                       $sSQL .= ' OR ST_DWithin('.$sPointSQL.', ST_Centroid(geometry), '.$fSearchDiam.'))';
+                       $sSQL .= ' ORDER BY ST_distance('.$sPointSQL.', geometry) ASC limit 1';
+//var_dump($sSQL);
+                       $aPlace = $oDB->getRow($sSQL);
+                       $iPlaceID = $aPlace['place_id'];
+                       if (PEAR::IsError($iPlaceID))
+                       {
+                               var_Dump($sSQL, $iPlaceID); 
+                               exit;
+                       }
+               }
+
+               // The point we found might be too small - use the address to find what it is a child of
+               if ($iPlaceID)
+               {
+                       $sSQL = "select address_place_id from place_addressline where cached_rank_address <= $iMaxRank and place_id = $iPlaceID order by cached_rank_address desc,isaddress desc,distance desc limit 1";
+                       $iPlaceID = $oDB->getOne($sSQL);
+                       if (PEAR::IsError($iPlaceID))
+                       {
+                               var_Dump($sSQL, $iPlaceID); 
+                               exit;
+                       }
+
+                       if ($iPlaceID && $aPlace['place_id'] && $iMaxRank < 28)
+                       {
+                               $sSQL = "select address_place_id from place_addressline where cached_rank_address <= $iMaxRank and place_id = ".$aPlace['place_id']." order by cached_rank_address desc,isaddress desc,distance desc";
+                               $iPlaceID = $oDB->getOne($sSQL);
+                               if (PEAR::IsError($iPlaceID))
+                               {
+                                       var_Dump($sSQL, $iPlaceID); 
+                                       exit;
+                               }
+                       }
+                       if (!$iPlaceID)
+                       {
+                               $iPlaceID = $aPlace['place_id'];
+                       }
+               }
+
+               return $iPlaceID;
+       }