]> git.openstreetmap.org Git - nominatim.git/blobdiff - utils/setup.php
add simple tests for postcode import
[nominatim.git] / utils / setup.php
index 87c1dd0725cc6715f76c0cbc1d177119f083dd1d..67c7fd7f80b8be31139dca46684c29fdd34c7e87 100755 (executable)
@@ -132,9 +132,15 @@ if ($aCMDResult['setup-db'] || $aCMDResult['all']) {
         pgsqlRunScript('ALTER FUNCTION ST_Distance_Spheroid(geometry, geometry, spheroid) RENAME TO ST_DistanceSpheroid');
     }
 
+    if (!file_exists(CONST_ExtraDataPath.'/country_osm_grid.sql.gz')) {
+        echo "Error: you need to download the country_osm_grid first:";
+        echo "\n    wget -O ".CONST_ExtraDataPath."/country_osm_grid.sql.gz http://www.nominatim.org/data/country_grid.sql.gz\n";
+        exit(1);
+    }
+
     pgsqlRunScriptFile(CONST_BasePath.'/data/country_name.sql');
     pgsqlRunScriptFile(CONST_BasePath.'/data/country_naturalearthdata.sql');
-    pgsqlRunScriptFile(CONST_BasePath.'/data/country_osm_grid.sql');
+    pgsqlRunScriptFile(CONST_BasePath.'/data/country_osm_grid.sql.gz');
     pgsqlRunScriptFile(CONST_BasePath.'/data/gb_postcode_table.sql');
     if (file_exists(CONST_BasePath.'/data/gb_postcode_data.sql.gz')) {
         pgsqlRunScriptFile(CONST_BasePath.'/data/gb_postcode_data.sql.gz');
@@ -186,7 +192,7 @@ if ($aCMDResult['import-data'] || $aCMDResult['all']) {
     passthruCheckReturn($osm2pgsql);
 
     $oDB =& getDB();
-    if (!chksql($oDB->getRow('select * from place limit 1'))) {
+    if (!$aCMDResult['ignore-errors'] && !chksql($oDB->getRow('select * from place limit 1'))) {
         fail('No Data');
     }
 }
@@ -360,27 +366,31 @@ if ($aCMDResult['load-data'] || $aCMDResult['all']) {
     }
 
     echo "Load Data\n";
+    $sColumns = 'osm_type, osm_id, class, type, name, admin_level, address, extratags, geometry';
+
     $aDBInstances = array();
     $iLoadThreads = max(1, $iInstances - 1);
     for ($i = 0; $i < $iLoadThreads; $i++) {
         $aDBInstances[$i] =& getDB(true);
-        $sSQL = 'insert into placex (osm_type, osm_id, class, type, name, admin_level, ';
-        $sSQL .= 'housenumber, street, addr_place, isin, postcode, country_code, extratags, ';
-        $sSQL .= 'geometry) select * from place where osm_id % '.$iLoadThreads.' = '.$i;
-        $sSQL .= " and not (class='place' and type='houses' and osm_type='W' and ST_GeometryType(geometry) = 'ST_LineString')";
+        $sSQL = "INSERT INTO placex ($sColumns) SELECT $sColumns FROM place WHERE osm_id % $iLoadThreads = $i";
+        $sSQL .= " and not (class='place' and type='houses' and osm_type='W'";
+        $sSQL .= "          and ST_GeometryType(geometry) = 'ST_LineString')";
+        $sSQL .= " and ST_IsValid(geometry)";
         if ($aCMDResult['verbose']) echo "$sSQL\n";
-        if (!pg_send_query($aDBInstances[$i]->connection, $sSQL)) fail(pg_last_error($oDB->connection));
+        if (!pg_send_query($aDBInstances[$i]->connection, $sSQL)) {
+            fail(pg_last_error($aDBInstances[$i]->connection));
+        }
     }
     // last thread for interpolation lines
     $aDBInstances[$iLoadThreads] =& getDB(true);
     $sSQL = 'insert into location_property_osmline';
-    $sSQL .= ' (osm_id, interpolationtype, street, addr_place,';
-    $sSQL .= '  postcode, calculated_country_code, linegeo)';
-    $sSQL .= ' SELECT osm_id, housenumber, street, addr_place,';
-    $sSQL .= ' postcode, country_code, geometry from place where ';
+    $sSQL .= ' (osm_id, address, linegeo)';
+    $sSQL .= ' SELECT osm_id, address, geometry from place where ';
     $sSQL .= "class='place' and type='houses' and osm_type='W' and ST_GeometryType(geometry) = 'ST_LineString'";
     if ($aCMDResult['verbose']) echo "$sSQL\n";
-    if (!pg_send_query($aDBInstances[$i]->connection, $sSQL)) fail(pg_last_error($oDB->connection));
+    if (!pg_send_query($aDBInstances[$iLoadThreads]->connection, $sSQL)) {
+        fail(pg_last_error($aDBInstances[$iLoadThreads]->connection));
+    }
 
     $bAnyBusy = true;
     while ($bAnyBusy) {
@@ -394,6 +404,16 @@ if ($aCMDResult['load-data'] || $aCMDResult['all']) {
     echo "\n";
     echo "Reanalysing database...\n";
     pgsqlRunScript('ANALYSE');
+
+    $sDatabaseDate = getDatabaseDate($oDB);
+    pg_query($oDB->connection, 'TRUNCATE import_status');
+    if ($sDatabaseDate === false) {
+        echo "WARNING: could not determine database date.\n";
+    } else {
+        $sSQL = "INSERT INTO import_status (lastimportdate) VALUES('".$sDatabaseDate."')";
+        pg_query($oDB->connection, $sSQL);
+        echo "Latest data imported from $sDatabaseDate.\n";
+    }
 }
 
 if ($aCMDResult['import-tiger-data']) {
@@ -473,102 +493,66 @@ if ($aCMDResult['import-tiger-data']) {
 if ($aCMDResult['calculate-postcodes'] || $aCMDResult['all']) {
     $bDidSomething = true;
     $oDB =& getDB();
-    if (!pg_query($oDB->connection, 'DELETE from placex where osm_type=\'P\'')) fail(pg_last_error($oDB->connection));
-    $sSQL = "insert into placex (osm_type,osm_id,class,type,postcode,calculated_country_code,geometry) ";
-    $sSQL .= "select 'P',nextval('seq_postcodes'),'place','postcode',postcode,calculated_country_code,";
-    $sSQL .= "ST_SetSRID(ST_Point(x,y),4326) as geometry from (select calculated_country_code,postcode,";
-    $sSQL .= "avg(st_x(st_centroid(geometry))) as x,avg(st_y(st_centroid(geometry))) as y ";
-    $sSQL .= "from placex where postcode is not null group by calculated_country_code,postcode) as x ";
-    $sSQL .= "where ST_Point(x,y) is not null";
-    if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection));
+    if (!pg_query($oDB->connection, 'TRUNCATE location_postcode')) {
+        fail(pg_last_error($oDB->connection));
+    }
+
+    $sSQL  = "INSERT INTO location_postcode";
+    $sSQL .= " (place_id, indexed_status, country_code, postcode, geometry) ";
+    $sSQL .= "SELECT nextval('seq_place'), 1, country_code,";
+    $sSQL .= "       upper(trim (both ' ' from address->'postcode')) as pc,";
+    $sSQL .= "       ST_Centroid(ST_Collect(ST_Centroid(geometry)))";
+    $sSQL .= "  FROM placex";
+    $sSQL .= " WHERE address ? 'postcode' AND address->'postcode' NOT SIMILAR TO '%(,|;)%'";
+    $sSQL .= "       AND geometry IS NOT null";
+    $sSQL .= " GROUP BY country_code, pc";
+
+    if (!pg_query($oDB->connection, $sSQL)) {
+        fail(pg_last_error($oDB->connection));
+    }
 
     if (CONST_Use_Extra_US_Postcodes) {
-        $sSQL = "insert into placex (osm_type,osm_id,class,type,postcode,calculated_country_code,geometry) ";
-        $sSQL .= "select 'P',nextval('seq_postcodes'),'place','postcode',postcode,'us',";
-        $sSQL .= "ST_SetSRID(ST_Point(x,y),4326) as geometry from us_postcode";
+        // only add postcodes that are not yet available in OSM
+        $sSQL  = "INSERT INTO location_postcode";
+        $sSQL .= " (place_id, indexed_status, country_code, postcode, geometry) ";
+        $sSQL .= "SELECT nextval('seq_place'), 1, 'us', postcode,";
+        $sSQL .= "       ST_SetSRID(ST_Point(x,y),4326)";
+        $sSQL .= "  FROM us_postcode WHERE postcode NOT IN";
+        $sSQL .= "        (SELECT postcode FROM location_postcode";
+        $sSQL .= "          WHERE country_code = 'us')";
+
         if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection));
     }
-}
 
-if ($aCMDResult['osmosis-init'] || ($aCMDResult['all'] && !$aCMDResult['drop'])) { // no use doing osmosis-init when dropping update tables
-    $bDidSomething = true;
-    $oDB =& getDB();
+    // add missing postcodes for GB (if available)
+    $sSQL  = "INSERT INTO location_postcode";
+    $sSQL .= " (place_id, indexed_status, country_code, postcode, geometry) ";
+    $sSQL .= "SELECT nextval('seq_place'), 1, 'gb', postcode, geometry";
+    $sSQL .= "  FROM gb_postcode WHERE postcode NOT IN";
+    $sSQL .= "           (SELECT postcode FROM location_postcode";
+    $sSQL .= "             WHERE country_code = 'gb')";
+    if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection));
 
-    if (!file_exists(CONST_Osmosis_Binary)) {
-        echo "Please download osmosis.\nIf it is already installed, check the path in your local settings (settings/local.php) file.\n";
-        if (!$aCMDResult['all']) {
-            fail("osmosis not found in '".CONST_Osmosis_Binary."'");
-        }
-    } else {
-        if (file_exists(CONST_InstallPath.'/settings/configuration.txt')) {
-            echo "settings/configuration.txt already exists\n";
-        } else {
-            passthru(CONST_Osmosis_Binary.' --read-replication-interval-init '.CONST_InstallPath.'/settings');
-            // update osmosis configuration.txt with our settings
-            passthru("sed -i 's!baseUrl=.*!baseUrl=".CONST_Replication_Url."!' ".CONST_InstallPath.'/settings/configuration.txt');
-            passthru("sed -i 's:maxInterval = .*:maxInterval = ".CONST_Replication_MaxInterval.":' ".CONST_InstallPath.'/settings/configuration.txt');
+    if (!$aCMDResult['all']) {
+        $sSQL = "DELETE FROM word WHERE class='place' and type='postcode'";
+        $sSQL .= "and word NOT IN (SELECT postcode FROM location_postcode)";
+        if (!pg_query($oDB->connection, $sSQL)) {
+            fail(pg_last_error($oDB->connection));
         }
+    }
+    $sSQL = "SELECT count(getorcreate_postcode_id(v)) FROM ";
+    $sSQL .= "(SELECT distinct(postcode) as v FROM location_postcode) p";
 
-        // Find the last node in the DB
-        $iLastOSMID = $oDB->getOne("select max(osm_id) from place where osm_type = 'N'");
-
-        // Lookup the timestamp that node was created (less 3 hours for margin for changsets to be closed)
-        $sLastNodeURL = 'http://www.openstreetmap.org/api/0.6/node/'.$iLastOSMID."/1";
-        $sLastNodeXML = file_get_contents($sLastNodeURL);
-        preg_match('#timestamp="(([0-9]{4})-([0-9]{2})-([0-9]{2})T([0-9]{2}):([0-9]{2}):([0-9]{2})Z)"#', $sLastNodeXML, $aLastNodeDate);
-        $iLastNodeTimestamp = strtotime($aLastNodeDate[1]) - (3*60*60);
-
-        // Search for the correct state file - uses file timestamps so need to sort by date descending
-        $sRepURL = CONST_Replication_Url."/";
-        $sRep = file_get_contents($sRepURL."?C=M;O=D;F=1");
-        // download.geofabrik.de:    <a href="000/">000/</a></td><td align="right">26-Feb-2013 11:53  </td>
-        // planet.openstreetmap.org: <a href="273/">273/</a>                    2013-03-11 07:41    -
-        preg_match_all('#<a href="[0-9]{3}/">([0-9]{3}/)</a>\s*([-0-9a-zA-Z]+ [0-9]{2}:[0-9]{2})#', $sRep, $aRepMatches, PREG_SET_ORDER);
-        if ($aRepMatches) {
-            $aPrevRepMatch = false;
-            foreach ($aRepMatches as $aRepMatch) {
-                if (strtotime($aRepMatch[2]) < $iLastNodeTimestamp) break;
-                $aPrevRepMatch = $aRepMatch;
-            }
-            if ($aPrevRepMatch) $aRepMatch = $aPrevRepMatch;
-
-            $sRepURL .= $aRepMatch[1];
-            $sRep = file_get_contents($sRepURL."?C=M;O=D;F=1");
-            preg_match_all('#<a href="[0-9]{3}/">([0-9]{3}/)</a>\s*([-0-9a-zA-Z]+ [0-9]{2}:[0-9]{2})#', $sRep, $aRepMatches, PREG_SET_ORDER);
-            $aPrevRepMatch = false;
-            foreach ($aRepMatches as $aRepMatch) {
-                if (strtotime($aRepMatch[2]) < $iLastNodeTimestamp) break;
-                $aPrevRepMatch = $aRepMatch;
-            }
-            if ($aPrevRepMatch) $aRepMatch = $aPrevRepMatch;
-
-            $sRepURL .= $aRepMatch[1];
-            $sRep = file_get_contents($sRepURL."?C=M;O=D;F=1");
-            preg_match_all('#<a href="[0-9]{3}.state.txt">([0-9]{3}).state.txt</a>\s*([-0-9a-zA-Z]+ [0-9]{2}:[0-9]{2})#', $sRep, $aRepMatches, PREG_SET_ORDER);
-            $aPrevRepMatch = false;
-            foreach ($aRepMatches as $aRepMatch) {
-                if (strtotime($aRepMatch[2]) < $iLastNodeTimestamp) break;
-                $aPrevRepMatch = $aRepMatch;
-            }
-            if ($aPrevRepMatch) $aRepMatch = $aPrevRepMatch;
-
-            $sRepURL .= $aRepMatch[1].'.state.txt';
-            echo "Getting state file: $sRepURL\n";
-            $sStateFile = file_get_contents($sRepURL);
-            if (!$sStateFile || strlen($sStateFile) > 1000) fail("unable to obtain state file");
-            file_put_contents(CONST_InstallPath.'/settings/state.txt', $sStateFile);
-            echo "Updating DB status\n";
-            pg_query($oDB->connection, 'TRUNCATE import_status');
-            $sSQL = "INSERT INTO import_status VALUES('".$aRepMatch[2]."')";
-            pg_query($oDB->connection, $sSQL);
-        } else {
-            if (!$aCMDResult['all']) {
-                fail("Cannot read state file directory.");
-            }
-        }
+    if (!pg_query($oDB->connection, $sSQL)) {
+        fail(pg_last_error($oDB->connection));
     }
 }
 
+if ($aCMDResult['osmosis-init']) {
+    $bDidSomething = true;
+    echo "Command 'osmosis-init' no longer available, please use utils/update.php --init-updates.\n";
+}
+
 if ($aCMDResult['index'] || $aCMDResult['all']) {
     $bDidSomething = true;
     $sOutputFile = '';
@@ -578,6 +562,11 @@ if ($aCMDResult['index'] || $aCMDResult['all']) {
     passthruCheckReturn($sBaseCmd.' -r 5 -R 25');
     if (!$aCMDResult['index-noanalyse']) pgsqlRunScript('ANALYSE');
     passthruCheckReturn($sBaseCmd.' -r 26');
+
+    echo "Indexing postcodes....\n";
+    $oDB =& getDB();
+    $sSQL = 'UPDATE location_postcode SET indexed_status = 0';
+    if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection));
 }
 
 if ($aCMDResult['create-search-indices'] || $aCMDResult['all']) {