Merge remote-tracking branch 'upstream/master'

[nominatim.git] / lib / lib.php

diff --git a/lib/lib.php b/lib/lib.php
index 1517b317102502d75b175ee39949c96af8ce9434..b8f8d8b0b07397f5e21199799a89f657c99ef4a1 100644 (file)
--- a/lib/lib.php
+++ b/lib/lib.php
@@ -45,7 +45,7 @@
                if (!$sUserError) $sUserError = $sError;
                error_log('ERROR: '.$sError);
                echo $sUserError."\n";
                if (!$sUserError) $sUserError = $sError;
                error_log('ERROR: '.$sError);
                echo $sUserError."\n";

-               exit;
+               exit(-1);

        }
 
 
        }
 
 


@@ -94,7 +94,8 @@
 
        function bySearchRank($a, $b)
        {
 
        function bySearchRank($a, $b)
        {

-               if ($a['iSearchRank'] == $b['iSearchRank']) return 0;
+               if ($a['iSearchRank'] == $b['iSearchRank'])
+            return strlen($a['sOperator']) + strlen($a['sHouseNumber']) - strlen($b['sOperator']) - strlen($b['sHouseNumber']);

                return ($a['iSearchRank'] < $b['iSearchRank']?-1:1);
        }
 
                return ($a['iSearchRank'] < $b['iSearchRank']?-1:1);
        }
 


@@ -117,18 +118,26 @@
        }
 
 
        }
 
 

-       function getPreferredLanguages()
+       function getPreferredLanguages($sLangString=false)

        {
        {

-               // If we have been provided the value in $_GET it overrides browser value
-               if (isset($_GET['accept-language']) && $_GET['accept-language'])
+               if (!$sLangString)

                {
                {

-                       $_SERVER["HTTP_ACCEPT_LANGUAGE"] = $_GET['accept-language'];
+                       // If we have been provided the value in $_GET it overrides browser value
+                       if (isset($_GET['accept-language']) && $_GET['accept-language'])
+                       {
+                               $_SERVER["HTTP_ACCEPT_LANGUAGE"] = $_GET['accept-language'];
+                               $sLangString = $_GET['accept-language'];
+                       }
+                       else if (isset($_SERVER["HTTP_ACCEPT_LANGUAGE"]))
+                       {
+                               $sLangString = $_SERVER["HTTP_ACCEPT_LANGUAGE"];
+                       }

                }
 
                $aLanguages = array();
                }
 
                $aLanguages = array();

-               if (isset($_SERVER["HTTP_ACCEPT_LANGUAGE"]))
+               if ($sLangString)

                {
                {

-                       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))
+                       if (preg_match_all('/(([a-z]{1,8})(-[a-z]{1,8})?)\s*(;\s*q\s*=\s*(1|0\.[0-9]+))?/i', $sLangString, $aLanguagesParse, PREG_SET_ORDER))

                        {
                                foreach($aLanguagesParse as $iLang => $aLanguage)
                                {
                        {
                                foreach($aLanguagesParse as $iLang => $aLanguage)
                                {


@@ -169,7 +178,7 @@
        {
                $aResult = array(array(join(' ',$aWords)));
                $sFirstToken = '';
        {
                $aResult = array(array(join(' ',$aWords)));
                $sFirstToken = '';

-               if ($iDepth < 8) {
+               if ($iDepth < 7) {

                        while(sizeof($aWords) > 1)
                        {
                                $sWord = array_shift($aWords);
                        while(sizeof($aWords) > 1)
                        {
                                $sWord = array_shift($aWords);


@@ -184,6 +193,26 @@
                return $aResult;
        }
 
                return $aResult;
        }
 

+       function getInverseWordSets($aWords, $iDepth)
+       {
+               $aResult = array(array(join(' ',$aWords)));
+               $sFirstToken = '';
+               if ($iDepth < 8)
+               {
+                       while(sizeof($aWords) > 1)
+                       {
+                               $sWord = array_pop($aWords);
+                               $sFirstToken = $sWord.($sFirstToken?' ':'').$sFirstToken;
+                               $aRest = getInverseWordSets($aWords, $iDepth+1);
+                               foreach($aRest as $aSet)
+                               {
+                                       $aResult[] = array_merge(array($sFirstToken),$aSet);
+                               }
+                       }
+               }
+               return $aResult;
+       }
+

 
        function getTokensFromSets($aSets)
        {
 
        function getTokensFromSets($aSets)
        {


@@ -246,7 +275,13 @@
 
                if (sizeof($aNearPostcodes))
                {
 
                if (sizeof($aNearPostcodes))
                {

-                       return array(array('lat' => $aNearPostcodes[0]['lat'], 'lon' => $aNearPostcodes[0]['lon'], 'radius' => 0.005));
+                       $aPostcodes = array();
+                       foreach($aNearPostcodes as $aPostcode)
+                       {
+                               $aPostcodes[] = array('lat' => $aPostcode['lat'], 'lon' => $aPostcode['lon'], 'radius' => 0.005);
+                       }
+
+                       return $aPostcodes;

                }
 
                return false;
                }
 
                return false;


@@ -343,7 +378,7 @@
  'landuse:farm' => array('label'=>'Farm','frequency'=>1201,'icon'=>'', 'defdiameter' => 0.02,),
  'place:farm' => array('label'=>'Farm','frequency'=>1162,'icon'=>'', 'defdiameter' => 0.02,),
 
  'landuse:farm' => array('label'=>'Farm','frequency'=>1201,'icon'=>'', 'defdiameter' => 0.02,),
  'place:farm' => array('label'=>'Farm','frequency'=>1162,'icon'=>'', 'defdiameter' => 0.02,),
 

- 'highway:motorway_junction' => array('label'=>'Motorway Junction','frequency'=>1126,'icon'=>'','simplelabel'=>'Road',),
+ 'highway:motorway_junction' => array('label'=>'Motorway Junction','frequency'=>1126,'icon'=>'','simplelabel'=>'Junction',),

  'highway:motorway' => array('label'=>'Motorway','frequency'=>4627,'icon'=>'','simplelabel'=>'Road',),
  'highway:trunk' => array('label'=>'Trunk','frequency'=>23084,'icon'=>'','simplelabel'=>'Road',),
  'highway:primary' => array('label'=>'Primary','frequency'=>32138,'icon'=>'','simplelabel'=>'Road',),
  'highway:motorway' => array('label'=>'Motorway','frequency'=>4627,'icon'=>'','simplelabel'=>'Road',),
  'highway:trunk' => array('label'=>'Trunk','frequency'=>23084,'icon'=>'','simplelabel'=>'Road',),
  'highway:primary' => array('label'=>'Primary','frequency'=>32138,'icon'=>'','simplelabel'=>'Road',),


@@ -371,6 +406,8 @@
  'landuse:commercial' => array('label'=>'Commercial','frequency'=>657,'icon'=>'',),
 
  'place:airport' => array('label'=>'Airport','frequency'=>36,'icon'=>'transport_airport2', 'defdiameter' => 0.03,),
  'landuse:commercial' => array('label'=>'Commercial','frequency'=>657,'icon'=>'',),
 
  'place:airport' => array('label'=>'Airport','frequency'=>36,'icon'=>'transport_airport2', 'defdiameter' => 0.03,),

+ 'aeroway:aerodrome' => array('label'=>'Aerodrome','frequency'=>36,'icon'=>'transport_airport2', 'defdiameter' => 0.03,),
+ 'aeroway' => array('label'=>'Aeroway','frequency'=>36,'icon'=>'transport_airport2', 'defdiameter' => 0.03,),

  'railway:station' => array('label'=>'Station','frequency'=>3431,'icon'=>'transport_train_station2', 'defdiameter' => 0.01,),
  'amenity:place_of_worship' => array('label'=>'Place Of Worship','frequency'=>9049,'icon'=>'place_of_worship_unknown3',),
  'amenity:pub' => array('label'=>'Pub','frequency'=>18969,'icon'=>'food_pub',),
  'railway:station' => array('label'=>'Station','frequency'=>3431,'icon'=>'transport_train_station2', 'defdiameter' => 0.01,),
  'amenity:place_of_worship' => array('label'=>'Place Of Worship','frequency'=>9049,'icon'=>'place_of_worship_unknown3',),
  'amenity:pub' => array('label'=>'Pub','frequency'=>18969,'icon'=>'food_pub',),


@@ -443,7 +480,7 @@
  'shop:electronics' => array('label'=>'Electronics','frequency'=>96,'icon'=>'',),
  'shop:department_store' => array('label'=>'Department Store','frequency'=>86,'icon'=>'',),
  'shop:books' => array('label'=>'Books','frequency'=>85,'icon'=>'',),
  'shop:electronics' => array('label'=>'Electronics','frequency'=>96,'icon'=>'',),
  'shop:department_store' => array('label'=>'Department Store','frequency'=>86,'icon'=>'',),
  'shop:books' => array('label'=>'Books','frequency'=>85,'icon'=>'',),

- 'shop:yes' => array('label'=>'Yes','frequency'=>68,'icon'=>'',),
+ 'shop:yes' => array('label'=>'Shop','frequency'=>68,'icon'=>'',),

  'shop:outdoor' => array('label'=>'Outdoor','frequency'=>67,'icon'=>'',),
  'shop:mall' => array('label'=>'Mall','frequency'=>63,'icon'=>'',),
  'shop:florist' => array('label'=>'Florist','frequency'=>61,'icon'=>'',),
  'shop:outdoor' => array('label'=>'Outdoor','frequency'=>67,'icon'=>'',),
  'shop:mall' => array('label'=>'Mall','frequency'=>63,'icon'=>'',),
  'shop:florist' => array('label'=>'Florist','frequency'=>61,'icon'=>'',),


@@ -523,7 +560,6 @@
  'natural:beach' => array('label'=>'Beach','frequency'=>121,'icon'=>'tourist_beach',),
  'place:moor' => array('label'=>'Moor','frequency'=>118,'icon'=>'',),
  'amenity:grave_yard' => array('label'=>'Grave Yard','frequency'=>110,'icon'=>'',),
  'natural:beach' => array('label'=>'Beach','frequency'=>121,'icon'=>'tourist_beach',),
  'place:moor' => array('label'=>'Moor','frequency'=>118,'icon'=>'',),
  'amenity:grave_yard' => array('label'=>'Grave Yard','frequency'=>110,'icon'=>'',),

- 'waterway:derelict_canal' => array('label'=>'Derelict Canal','frequency'=>109,'icon'=>'',),

  'waterway:drain' => array('label'=>'Drain','frequency'=>108,'icon'=>'',),
  'landuse:grass' => array('label'=>'Grass','frequency'=>106,'icon'=>'',),
  'landuse:village_green' => array('label'=>'Village Green','frequency'=>106,'icon'=>'',),
  'waterway:drain' => array('label'=>'Drain','frequency'=>108,'icon'=>'',),
  'landuse:grass' => array('label'=>'Grass','frequency'=>106,'icon'=>'',),
  'landuse:village_green' => array('label'=>'Village Green','frequency'=>106,'icon'=>'',),


@@ -589,7 +625,7 @@
  'historic:roman_road' => array('label'=>'Roman Road','frequency'=>27,'icon'=>'',),
  'historic:fort' => array('label'=>'Fort','frequency'=>26,'icon'=>'',),
  'railway:subway_entrance' => array('label'=>'Subway Entrance','frequency'=>26,'icon'=>'',),
  'historic:roman_road' => array('label'=>'Roman Road','frequency'=>27,'icon'=>'',),
  'historic:fort' => array('label'=>'Fort','frequency'=>26,'icon'=>'',),
  'railway:subway_entrance' => array('label'=>'Subway Entrance','frequency'=>26,'icon'=>'',),

- 'historic:yes' => array('label'=>'Yes','frequency'=>25,'icon'=>'',),
+ 'historic:yes' => array('label'=>'Historic','frequency'=>25,'icon'=>'',),

  'highway:gate' => array('label'=>'Gate','frequency'=>25,'icon'=>'',),
  'leisure:fishing' => array('label'=>'Fishing','frequency'=>24,'icon'=>'',),
  'historic:museum' => array('label'=>'Museum','frequency'=>24,'icon'=>'',),
  'highway:gate' => array('label'=>'Gate','frequency'=>25,'icon'=>'',),
  'leisure:fishing' => array('label'=>'Fishing','frequency'=>24,'icon'=>'',),
  'historic:museum' => array('label'=>'Museum','frequency'=>24,'icon'=>'',),


@@ -611,7 +647,7 @@
 
  'place:houses' => array('label'=>'Houses','frequency'=>85,'icon'=>'',),
  'railway:preserved' => array('label'=>'Preserved','frequency'=>227,'icon'=>'',),
 
  'place:houses' => array('label'=>'Houses','frequency'=>85,'icon'=>'',),
  'railway:preserved' => array('label'=>'Preserved','frequency'=>227,'icon'=>'',),

- 'waterway:derelict canal' => array('label'=>'Derelict Canal','frequency'=>21,'icon'=>'',),
+ 'waterway:derelict_canal' => array('label'=>'Derelict Canal','frequency'=>21,'icon'=>'',),

  'amenity:dead_pub' => array('label'=>'Dead Pub','frequency'=>20,'icon'=>'',),
  'railway:disused_station' => array('label'=>'Disused Station','frequency'=>114,'icon'=>'',),
  'railway:abandoned' => array('label'=>'Abandoned','frequency'=>641,'icon'=>'',),
  'amenity:dead_pub' => array('label'=>'Dead Pub','frequency'=>20,'icon'=>'',),
  'railway:disused_station' => array('label'=>'Disused Station','frequency'=>114,'icon'=>'',),
  'railway:abandoned' => array('label'=>'Abandoned','frequency'=>641,'icon'=>'',),


@@ -774,7 +810,7 @@
                        {
                                $sTypeLabel = strtolower(isset($aTypeLabel['simplelabel'])?$aTypeLabel['simplelabel']:$aTypeLabel['label']);
                                $sTypeLabel = str_replace(' ','_',$sTypeLabel);
                        {
                                $sTypeLabel = strtolower(isset($aTypeLabel['simplelabel'])?$aTypeLabel['simplelabel']:$aTypeLabel['label']);
                                $sTypeLabel = str_replace(' ','_',$sTypeLabel);

-                               if (!isset($aAddress[$sTypeLabel]) || (isset($aFallback[$sTypeLabel]) && $aFallback[$sTypeLabel]))
+                               if (!isset($aAddress[$sTypeLabel]) || (isset($aFallback[$sTypeLabel]) && $aFallback[$sTypeLabel]) || $aLine['class'] == 'place')

                                {
                                        $aAddress[$sTypeLabel] = $aLine['localname']?$aLine['localname']:$aLine['housenumber'];
                                }
                                {
                                        $aAddress[$sTypeLabel] = $aLine['localname']?$aLine['localname']:$aLine['housenumber'];
                                }


@@ -846,12 +882,12 @@
                        $sSQL .= ' ORDER BY ST_distance('.$sPointSQL.', geometry) ASC limit 1';
                        //var_dump($sSQL);
                        $aPlace = $oDB->getRow($sSQL);
                        $sSQL .= ' ORDER BY ST_distance('.$sPointSQL.', geometry) ASC limit 1';
                        //var_dump($sSQL);
                        $aPlace = $oDB->getRow($sSQL);

-                       $iPlaceID = $aPlace['place_id'];
-                       if (PEAR::IsError($iPlaceID))
+                       if (PEAR::IsError($aPlace))

                        {
                        {

-                               var_Dump($sSQL, $iPlaceID);
+                               var_Dump($sSQL, $aPlace);

                                exit;
                        }
                                exit;
                        }

+                       $iPlaceID = $aPlace['place_id'];

                }
 
                // The point we found might be too small - use the address to find what it is a child of
                }
 
                // The point we found might be too small - use the address to find what it is a child of


@@ -884,16 +920,97 @@
                return $iPlaceID;
        }
 
                return $iPlaceID;
        }
 

-       function loadStructuredAddressElement(&$aStructuredQuery, &$iMinAddressRank, &$iMaxAddressRank, $aParams, $sKey, $iNewMinAddressRank, $iNewMaxAddressRank)
+       function addQuotes($s)
+       {
+               return "'".$s."'";
+       }
+
+       // returns boolean
+       function validLatLon($fLat,$fLon)

        {
        {

-               if (!isset($_GET[$sKey])) return false;
-               $sValue = trim($_GET[$sKey]);
-               if (!$sValue) return false;
-               $aStructuredQuery[$sKey] = $sValue;
-               if ($iMinAddressRank == 0 && $iMaxAddressRank == 30)
+               return ($fLat <= 90.1 && $fLat >= -90.1 && $fLon <= 180.1 && $fLon >= -180.1);
+       }
+
+       // Do we have anything that looks like a lat/lon pair?
+       // returns array(lat,lon,query_with_lat_lon_removed)
+       // or null
+       function looksLikeLatLonPair($sQuery)
+       {
+               $sFound    = null;
+               $fQueryLat = null;
+               $fQueryLon = null;
+
+               // degrees decimal minutes
+               // N 40 26.767, W 79 58.933
+               // N 40°26.767′, W 79°58.933′
+               //                  1         2                   3                  4         5            6
+               if (preg_match('/\\b([NS])[ ]+([0-9]+[0-9.]*)[° ]+([0-9.]+)?[′\']*[, ]+([EW])[ ]+([0-9]+)[° ]+([0-9]+[0-9.]*)[′\']*?\\b/', $sQuery, $aData))

                {
                {

-                       $iMinAddressRank = $iNewMinAddressRank;
-                       $iMaxAddressRank = $iNewMaxAddressRank;
+                       $sFound    = $aData[0];
+                       $fQueryLat = ($aData[1]=='N'?1:-1) * ($aData[2] + $aData[3]/60);
+                       $fQueryLon = ($aData[4]=='E'?1:-1) * ($aData[5] + $aData[6]/60);

                }
                }

-               return true;
+               // degrees decimal minutes
+               // 40 26.767 N, 79 58.933 W
+               // 40° 26.767′ N 79° 58.933′ W
+               //                      1             2                      3          4            5                    6
+               elseif (preg_match('/\\b([0-9]+)[° ]+([0-9]+[0-9.]*)?[′\']*[ ]+([NS])[, ]+([0-9]+)[° ]+([0-9]+[0-9.]*)?[′\' ]+([EW])\\b/', $sQuery, $aData))
+               {
+                       $sFound    = $aData[0];
+                       $fQueryLat = ($aData[3]=='N'?1:-1) * ($aData[1] + $aData[2]/60);
+                       $fQueryLon = ($aData[6]=='E'?1:-1) * ($aData[4] + $aData[5]/60);
+               }
+               // degrees decimal seconds
+               // N 40 26 46 W 79 58 56
+               // N 40° 26′ 46″ W, 79° 58′ 56″
+               //                      1        2            3            4                5        6            7            8
+               elseif (preg_match('/\\b([NS])[ ]([0-9]+)[° ]+([0-9]+)[′\' ]+([0-9]+)[″"]*[, ]+([EW])[ ]([0-9]+)[° ]+([0-9]+)[′\' ]+([0-9]+)[″"]*\\b/', $sQuery, $aData))
+               {
+                       $sFound    = $aData[0];
+                       $fQueryLat = ($aData[1]=='N'?1:-1) * ($aData[2] + $aData[3]/60 + $aData[4]/3600);
+                       $fQueryLon = ($aData[5]=='E'?1:-1) * ($aData[6] + $aData[7]/60 + $aData[8]/3600);
+               }
+               // degrees decimal seconds
+               // 40 26 46 N 79 58 56 W
+               // 40° 26′ 46″ N, 79° 58′ 56″ W
+               //                      1            2            3            4          5            6            7            8
+               elseif (preg_match('/\\b([0-9]+)[° ]+([0-9]+)[′\' ]+([0-9]+)[″" ]+([NS])[, ]+([0-9]+)[° ]+([0-9]+)[′\' ]+([0-9]+)[″" ]+([EW])\\b/', $sQuery, $aData))
+               {
+                       $sFound    = $aData[0];
+                       $fQueryLat = ($aData[4]=='N'?1:-1) * ($aData[1] + $aData[2]/60 + $aData[3]/3600);
+                       $fQueryLon = ($aData[8]=='E'?1:-1) * ($aData[5] + $aData[6]/60 + $aData[7]/3600);
+               }
+               // degrees decimal
+               // N 40.446° W 79.982°
+               //                      1        2                               3        4
+               elseif (preg_match('/\\b([NS])[ ]([0-9]+[0-9]*\\.[0-9]+)[°]*[, ]+([EW])[ ]([0-9]+[0-9]*\\.[0-9]+)[°]*\\b/', $sQuery, $aData))
+               {
+                       $sFound    = $aData[0];
+                       $fQueryLat = ($aData[1]=='N'?1:-1) * ($aData[2]);
+                       $fQueryLon = ($aData[3]=='E'?1:-1) * ($aData[4]);
+               }
+               // degrees decimal
+               // 40.446° N 79.982° W
+               //                      1                           2          3                           4
+               elseif (preg_match('/\\b([0-9]+[0-9]*\\.[0-9]+)[° ]+([NS])[, ]+([0-9]+[0-9]*\\.[0-9]+)[° ]+([EW])\\b/', $sQuery, $aData))
+               {
+                       $sFound    = $aData[0];
+                       $fQueryLat = ($aData[2]=='N'?1:-1) * ($aData[1]);
+                       $fQueryLon = ($aData[4]=='E'?1:-1) * ($aData[3]);
+               }
+               // degrees decimal
+               // 12.34, 56.78
+               // [12.456,-78.90]
+               //                   1          2                             3                        4
+               elseif (preg_match('/(\\[|^|\\b)(-?[0-9]+[0-9]*\\.[0-9]+)[, ]+(-?[0-9]+[0-9]*\\.[0-9]+)(\\]|$|\\b)/', $sQuery, $aData))
+               {
+                       $sFound    = $aData[0];
+                       $fQueryLat = $aData[2];
+                       $fQueryLon = $aData[3];
+               }
+
+               if (!validLatLon($fQueryLat, $fQueryLon)) return;
+               $sQuery = trim(str_replace($sFound, ' ', $sQuery));
+
+               return array('lat' => $fQueryLat, 'lon' => $fQueryLon, 'query' => $sQuery);

        }
        }