]> git.openstreetmap.org Git - nominatim.git/blobdiff - utils/setup.php
projects / nominatim.git / blobdiff
? search:
summary | shortlog | log | commit | commitdiff | tree
raw | inline | side by side
Merge remote-tracking branch 'upstream/master'
[nominatim.git] / utils / setup.php
diff --git a/utils/setup.php b/utils/setup.php
index 352aa943edc620bfa744de92993c8835fb1b84b7..edf0b14d460f27156569b65f872f2bb8aa934e6d 100755 (executable)
--- a/utils/setup.php
+++ b/utils/setup.php
@@ -18,23 +18,53 @@
                array('create-db', '', 0, 1, 0, 0, 'bool', 'Create nominatim db'),
                array('setup-db', '', 0, 1, 0, 0, 'bool', 'Build a blank nominatim db'),
                array('import-data', '', 0, 1, 0, 0, 'bool', 'Import a osm file'),
                array('create-db', '', 0, 1, 0, 0, 'bool', 'Create nominatim db'),
                array('setup-db', '', 0, 1, 0, 0, 'bool', 'Build a blank nominatim db'),
                array('import-data', '', 0, 1, 0, 0, 'bool', 'Import a osm file'),
+               array('osm2pgsql-cache', '', 0, 1, 1, 1, 'int', 'Cache size used by osm2pgsql'),
                array('create-functions', '', 0, 1, 0, 0, 'bool', 'Create functions'),
                array('create-functions', '', 0, 1, 0, 0, 'bool', 'Create functions'),
+               array('enable-diff-updates', '', 0, 1, 0, 0, 'bool', 'Turn on the code required to make diff updates work'),
+               array('enable-debug-statements', '', 0, 1, 0, 0, 'bool', 'Include debug warning statements in pgsql commands'),
+               array('create-minimal-tables', '', 0, 1, 0, 0, 'bool', 'Create minimal main tables'),
                array('create-tables', '', 0, 1, 0, 0, 'bool', 'Create main tables'),
                array('create-tables', '', 0, 1, 0, 0, 'bool', 'Create main tables'),
-               array('create-partitions', '', 0, 1, 0, 0, 'bool', 'Create required partition tables and triggers'),
+               array('create-partition-tables', '', 0, 1, 0, 0, 'bool', 'Create required partition tables'),
+               array('create-partition-functions', '', 0, 1, 0, 0, 'bool', 'Create required partition triggers'),
+               array('no-partitions', '', 0, 1, 0, 0, 'bool', "Do not partition search indices (speeds up import of single country extracts)"),
+               array('import-wikipedia-articles', '', 0, 1, 0, 0, 'bool', 'Import wikipedia article dump'),
                array('load-data', '', 0, 1, 0, 0, 'bool', 'Copy data to live tables from import table'),
                array('load-data', '', 0, 1, 0, 0, 'bool', 'Copy data to live tables from import table'),
+               array('disable-token-precalc', '', 0, 1, 0, 0, 'bool', 'Disable name precalculation (EXPERT)'),
                array('import-tiger-data', '', 0, 1, 0, 0, 'bool', 'Import tiger data (not included in \'all\')'),
                array('calculate-postcodes', '', 0, 1, 0, 0, 'bool', 'Calculate postcode centroids'),
                array('import-tiger-data', '', 0, 1, 0, 0, 'bool', 'Import tiger data (not included in \'all\')'),
                array('calculate-postcodes', '', 0, 1, 0, 0, 'bool', 'Calculate postcode centroids'),
-               array('create-roads', '', 0, 1, 0, 0, 'bool', 'Calculate postcode centroids'),
+               array('create-roads', '', 0, 1, 0, 0, 'bool', ''),
                array('osmosis-init', '', 0, 1, 0, 0, 'bool', 'Generate default osmosis configuration'),
                array('osmosis-init', '', 0, 1, 0, 0, 'bool', 'Generate default osmosis configuration'),
-               array('osmosis-init-date', '', 0, 1, 1, 1, 'string', 'Generate default osmosis configuration'),
                array('index', '', 0, 1, 0, 0, 'bool', 'Index the data'),
                array('index', '', 0, 1, 0, 0, 'bool', 'Index the data'),
+               array('index-noanalyse', '', 0, 1, 0, 0, 'bool', 'Do not perform analyse operations during index (EXPERT)'),
                array('index-output', '', 0, 1, 1, 1, 'string', 'File to dump index information to'),
                array('index-output', '', 0, 1, 1, 1, 'string', 'File to dump index information to'),
+               array('create-search-indices', '', 0, 1, 0, 0, 'bool', 'Create additional indices required for search and update'),
+               array('create-website', '', 0, 1, 1, 1, 'realpath', 'Create symlinks to setup web directory'),
        );
        getCmdOpt($_SERVER['argv'], $aCMDOptions, $aCMDResult, true, true);
 
        $bDidSomething = false;
 
        );
        getCmdOpt($_SERVER['argv'], $aCMDOptions, $aCMDResult, true, true);
 
        $bDidSomething = false;
 
-       // This is a pretty hard core defult - the number of processors in the box - 1
+       // Check if osm-file is set and points to a valid file if --all or --import-data is given
+       if ($aCMDResult['import-data'] || $aCMDResult['all'])
+       {
+               if (!isset($aCMDResult['osm-file']))
+               {
+                       fail('missing --osm-file for data import');
+               }
+
+               if (!file_exists($aCMDResult['osm-file']))
+               {
+                       fail('the path supplied to --osm-file does not exist');
+               }
+
+               if (!is_readable($aCMDResult['osm-file']))
+               {
+                       fail('osm-file "'.$aCMDResult['osm-file'].'" not readable');
+               }
+       }
+
+
+       // This is a pretty hard core default - the number of processors in the box - 1
        $iInstances = isset($aCMDResult['threads'])?$aCMDResult['threads']:(getProcessorCount()-1);
        if ($iInstances < 1)
        {
        $iInstances = isset($aCMDResult['threads'])?$aCMDResult['threads']:(getProcessorCount()-1);
        if ($iInstances < 1)
        {
@@ -46,17 +76,18 @@
                $iInstances = getProcessorCount();
                echo "WARNING: resetting threads to $iInstances\n";
        }
                $iInstances = getProcessorCount();
                echo "WARNING: resetting threads to $iInstances\n";
        }
-       if (isset($aCMDResult['osm-file']) && !isset($aCMDResult['osmosis-init-date']))
+
+       // Assume we can steal all the cache memory in the box (unless told otherwise)
+       $iCacheMemory = (isset($aCMDResult['osm2pgsql-cache'])?$aCMDResult['osm2pgsql-cache']:getCacheMemoryMB());
+       if ($iCacheMemory > getTotalMemoryMB())
        {
        {
-               $sBaseFile = basename($aCMDResult['osm-file']);
-               if (preg_match('#^planet-([0-9]{2})([0-9]{2})([0-9]{2})[.]#', $sBaseFile, $aMatch))
-               {
-                       $iTime = mktime(0, 0, 0, $aMatch[2], $aMatch[3], '20'.$aMatch[1]);
-                       $iTime -= (60*60*24);
-                       $aCMDResult['osmosis-init-date'] = date('Y-m-d', $iTime).'T22:00:00Z';
-               }
+               $iCacheMemory = getCacheMemoryMB();
+               echo "WARNING: resetting cache memory to $iCacheMemory\n";
        }
 
        }
 
+       $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
+       if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
+
        if ($aCMDResult['create-db'] || $aCMDResult['all'])
        {
                echo "Create DB\n";
        if ($aCMDResult['create-db'] || $aCMDResult['all'])
        {
                echo "Create DB\n";
@@ -64,31 +95,82 @@
                $oDB =& DB::connect(CONST_Database_DSN, false);
                if (!PEAR::isError($oDB))
                {
                $oDB =& DB::connect(CONST_Database_DSN, false);
                if (!PEAR::isError($oDB))
                {
-                       fail('database already exists');
+                       fail('database already exists ('.CONST_Database_DSN.')');
                }
                }
-               passthru('createdb nominatim');
+               passthruCheckReturn('createdb -E UTF-8 -p '.$aDSNInfo['port'].' '.$aDSNInfo['database']);
        }
 
        }
 
-       if ($aCMDResult['create-db'] || $aCMDResult['all'])
+       if ($aCMDResult['setup-db'] || $aCMDResult['all'])
        {
        {
-               echo "Create DB (2)\n";
+               echo "Setup DB\n";
                $bDidSomething = true;
                // TODO: path detection, detection memory, etc.
 
                $oDB =& getDB();
                $bDidSomething = true;
                // TODO: path detection, detection memory, etc.
 
                $oDB =& getDB();
-               passthru('createlang plpgsql nominatim');
-               pgsqlRunScriptFile(CONST_Path_Postgresql_Contrib.'/_int.sql');
-               pgsqlRunScriptFile(CONST_Path_Postgresql_Contrib.'/hstore.sql');
-               pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/postgis.sql');
-               pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/spatial_ref_sys.sql');
+
+               $sVersionString = $oDB->getOne('select version()');
+               preg_match('#PostgreSQL ([0-9]+)[.]([0-9]+)[.]([0-9]+) #', $sVersionString, $aMatches);
+               if (CONST_Postgresql_Version != $aMatches[1].'.'.$aMatches[2])
+               {
+                       echo "ERROR: PostgreSQL version is not correct.  Expected ".CONST_Postgresql_Version." found ".$aMatches[1].'.'.$aMatches[2]."\n";
+                       exit;
+               }
+
+               passthru('createlang plpgsql -p '.$aDSNInfo['port'].' '.$aDSNInfo['database']);
+               $pgver = (float) CONST_Postgresql_Version;
+               if ($pgver < 9.1) {
+                       pgsqlRunScriptFile(CONST_Path_Postgresql_Contrib.'/hstore.sql');
+                       pgsqlRunScriptFile(CONST_BasePath.'/sql/hstore_compatability_9_0.sql');
+               } else {
+                       pgsqlRunScript('CREATE EXTENSION hstore');
+               }
+
+               $fPostgisVersion = (float) CONST_Postgis_Version;
+               if ($fPostgisVersion < 2.0) {
+                       pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/postgis.sql');
+                       pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/spatial_ref_sys.sql');
+               } else {
+                       pgsqlRunScript('CREATE EXTENSION postgis');
+               }
+               if ($fPostgisVersion < 2.1) {
+                       // Function was renamed in 2.1 and throws an annoying deprecation warning
+                       pgsqlRunScript('ALTER FUNCTION st_line_interpolate_point(geometry, double precision) RENAME TO ST_LineInterpolatePoint');
+               }
+               $sVersionString = $oDB->getOne('select postgis_full_version()');
+               preg_match('#POSTGIS="([0-9]+)[.]([0-9]+)[.]([0-9]+)( r([0-9]+))?"#', $sVersionString, $aMatches);
+               if (CONST_Postgis_Version != $aMatches[1].'.'.$aMatches[2])
+               {
+                       echo "ERROR: PostGIS version is not correct.  Expected ".CONST_Postgis_Version." found ".$aMatches[1].'.'.$aMatches[2]."\n";
+                       exit;
+               }
+
                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_name.sql');
                pgsqlRunScriptFile(CONST_BasePath.'/data/country_naturalearthdata.sql');
                pgsqlRunScriptFile(CONST_BasePath.'/data/country_osm_grid.sql');
-               pgsqlRunScriptFile(CONST_BasePath.'/data/gb_postcode.sql');
+               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');
+               }
+               else
+               {
+                       echo "WARNING: external UK postcode table not found.\n";
+               }
                pgsqlRunScriptFile(CONST_BasePath.'/data/us_statecounty.sql');
                pgsqlRunScriptFile(CONST_BasePath.'/data/us_state.sql');
                pgsqlRunScriptFile(CONST_BasePath.'/data/us_postcode.sql');
                pgsqlRunScriptFile(CONST_BasePath.'/data/us_statecounty.sql');
                pgsqlRunScriptFile(CONST_BasePath.'/data/us_state.sql');
                pgsqlRunScriptFile(CONST_BasePath.'/data/us_postcode.sql');
-               pgsqlRunScriptFile(CONST_BasePath.'/data/worldboundaries.sql');
+
+               if ($aCMDResult['no-partitions'])
+               {
+                       pgsqlRunScript('update country_name set partition = 0');
+               }
+
+               // the following will be needed by create_functions later but
+               // is only defined in the subsequently called create_tables.
+               // Create dummies here that will be overwritten by the proper
+               // versions in create-tables.
+               pgsqlRunScript('CREATE TABLE place_boundingbox ()');
+               pgsqlRunScript('create type wikipedia_article_match as ()');
        }
 
        if ($aCMDResult['import-data'] || $aCMDResult['all'])
        }
 
        if ($aCMDResult['import-data'] || $aCMDResult['all'])
@@ -96,12 +178,30 @@
                echo "Import\n";
                $bDidSomething = true;
 
                echo "Import\n";
                $bDidSomething = true;
 
-               if (!file_exists(CONST_BasePath.'/osm2pgsql/osm2pgsql')) fail("please download and build osm2pgsql");
-               passthru(CONST_BasePath.'/osm2pgsql/osm2pgsql -lsc -O gazetteer -C 10000 --hstore -d nominatim '.$aCMDResult['osm-file']);
+               $osm2pgsql = CONST_Osm2pgsql_Binary;
+               if (!file_exists($osm2pgsql))
+               {
+                       echo "Please download and build osm2pgsql.\nIf it is already installed, check the path in your local settings (settings/local.php) file.\n";
+                       fail("osm2pgsql not found in '$osm2pgsql'");
+               }
+
+               if (!is_null(CONST_Osm2pgsql_Flatnode_File))
+               {
+                       $osm2pgsql .= ' --flat-nodes '.CONST_Osm2pgsql_Flatnode_File;
+               }
+               $osm2pgsql .= ' --tablespace-slim-index ssd --tablespace-main-index ssd --tablespace-main-data ssd --tablespace-slim-data data';
+               $osm2pgsql .= ' -lsc -O gazetteer --hstore';
+               $osm2pgsql .= ' -C 18000';
+               $osm2pgsql .= ' -P '.$aDSNInfo['port'];
+               $osm2pgsql .= ' -d '.$aDSNInfo['database'].' '.$aCMDResult['osm-file'];
+               passthruCheckReturn($osm2pgsql);
 
                $oDB =& getDB();
                $x = $oDB->getRow('select * from place limit 1');
 
                $oDB =& getDB();
                $x = $oDB->getRow('select * from place limit 1');
-               if (!$x || PEAR::isError($x)) fail('No Data');
+               if (PEAR::isError($x)) {
+                       fail($x->getMessage());
+               }
+               if (!$x) fail('No Data');
        }
 
        if ($aCMDResult['create-functions'] || $aCMDResult['all'])
        }
 
        if ($aCMDResult['create-functions'] || $aCMDResult['all'])
@@ -110,10 +210,43 @@
                $bDidSomething = true;
                if (!file_exists(CONST_BasePath.'/module/nominatim.so')) fail("nominatim module not built");
                $sTemplate = file_get_contents(CONST_BasePath.'/sql/functions.sql');
                $bDidSomething = true;
                if (!file_exists(CONST_BasePath.'/module/nominatim.so')) fail("nominatim module not built");
                $sTemplate = file_get_contents(CONST_BasePath.'/sql/functions.sql');
-               $sTemplate = str_replace('{modulepath}',CONST_BasePath.'/module', $sTemplate);
+               $sTemplate = str_replace('{modulepath}', CONST_BasePath.'/module', $sTemplate);
+               if ($aCMDResult['enable-diff-updates']) $sTemplate = str_replace('RETURN NEW; -- @DIFFUPDATES@', '--', $sTemplate);
+               if ($aCMDResult['enable-debug-statements']) $sTemplate = str_replace('--DEBUG:', '', $sTemplate);
+               if (CONST_Limit_Reindexing) $sTemplate = str_replace('--LIMIT INDEXING:', '', $sTemplate);
                pgsqlRunScript($sTemplate);
        }
 
                pgsqlRunScript($sTemplate);
        }
 
+       if ($aCMDResult['create-minimal-tables'])
+       {
+               echo "Minimal Tables\n";
+               $bDidSomething = true;
+               pgsqlRunScriptFile(CONST_BasePath.'/sql/tables-minimal.sql');
+
+               $sScript = '';
+
+               // Backstop the import process - easliest possible import id
+               $sScript .= "insert into import_npi_log values (18022);\n";
+
+               $hFile = @fopen(CONST_BasePath.'/settings/partitionedtags.def', "r");
+               if (!$hFile) fail('unable to open list of partitions: '.CONST_BasePath.'/settings/partitionedtags.def');
+
+               while (($sLine = fgets($hFile, 4096)) !== false && $sLine && substr($sLine,0,1) !='#')
+               {
+                       list($sClass, $sType) = explode(' ', trim($sLine));
+                       $sScript .= "create table place_classtype_".$sClass."_".$sType." as ";
+                       $sScript .= "select place_id as place_id,geometry as centroid from placex limit 0;\n";
+
+                       $sScript .= "CREATE INDEX idx_place_classtype_".$sClass."_".$sType."_centroid ";
+                       $sScript .= "ON place_classtype_".$sClass."_".$sType." USING GIST (centroid);\n";
+
+                       $sScript .= "CREATE INDEX idx_place_classtype_".$sClass."_".$sType."_place_id ";
+                       $sScript .= "ON place_classtype_".$sClass."_".$sType." USING btree(place_id);\n";
+               }
+               fclose($hFile);
+               pgsqlRunScript($sScript);
+       }
+
        if ($aCMDResult['create-tables'] || $aCMDResult['all'])
        {
                echo "Tables\n";
        if ($aCMDResult['create-tables'] || $aCMDResult['all'])
        {
                echo "Tables\n";
@@ -126,20 +259,20 @@
                pgsqlRunScript($sTemplate);
        }
 
                pgsqlRunScript($sTemplate);
        }
 
-       if ($aCMDResult['create-partitions'] || $aCMDResult['all'])
+       if ($aCMDResult['create-partition-tables'] || $aCMDResult['all'])
        {
        {
-               echo "Partitions\n";
+               echo "Partition Tables\n";
                $bDidSomething = true;
                $oDB =& getDB();
                $bDidSomething = true;
                $oDB =& getDB();
-               $sSQL = 'select partition from country_name order by country_code';
+               $sSQL = 'select distinct partition from country_name';
                $aPartitions = $oDB->getCol($sSQL);
                if (PEAR::isError($aPartitions))
                {
                        fail($aPartitions->getMessage());
                }
                $aPartitions = $oDB->getCol($sSQL);
                if (PEAR::isError($aPartitions))
                {
                        fail($aPartitions->getMessage());
                }
-               $aPartitions[] = 0;
+               if (!$aCMDResult['no-partitions']) $aPartitions[] = 0;
 
 
-               $sTemplate = file_get_contents(CONST_BasePath.'/sql/partitions.src.sql');
+               $sTemplate = file_get_contents(CONST_BasePath.'/sql/partition-tables.src.sql');
                preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER);
                foreach($aMatches as $aMatch)
                {
                preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER);
                foreach($aMatches as $aMatch)
                {
@@ -154,9 +287,66 @@
                pgsqlRunScript($sTemplate);
        }
 
                pgsqlRunScript($sTemplate);
        }
 
+
+       if ($aCMDResult['create-partition-functions'] || $aCMDResult['all'])
+       {
+               echo "Partition Functions\n";
+               $bDidSomething = true;
+               $oDB =& getDB();
+               $sSQL = 'select distinct partition from country_name';
+               $aPartitions = $oDB->getCol($sSQL);
+               if (PEAR::isError($aPartitions))
+               {
+                       fail($aPartitions->getMessage());
+               }
+               if (!$aCMDResult['no-partitions']) $aPartitions[] = 0;
+
+               $sTemplate = file_get_contents(CONST_BasePath.'/sql/partition-functions.src.sql');
+               preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER);
+               foreach($aMatches as $aMatch)
+               {
+                       $sResult = '';
+                       foreach($aPartitions as $sPartitionName)
+                       {
+                               $sResult .= str_replace('-partition-', $sPartitionName, $aMatch[1]);
+                       }
+                       $sTemplate = str_replace($aMatch[0], $sResult, $sTemplate);
+               }
+
+               pgsqlRunScript($sTemplate);
+       }
+
+       if ($aCMDResult['import-wikipedia-articles'] || $aCMDResult['all'])
+       {
+               $bDidSomething = true;
+               $sWikiArticlesFile = CONST_BasePath.'/data/wikipedia_article.sql.bin';
+               $sWikiRedirectsFile = CONST_BasePath.'/data/wikipedia_redirect.sql.bin';
+               if (file_exists($sWikiArticlesFile))
+               {
+                       echo "Importing wikipedia articles...";
+                       pgsqlRunDropAndRestore($sWikiArticlesFile);
+                       echo "...done\n";
+               }
+               else
+               {
+                       echo "WARNING: wikipedia article dump file not found - places will have default importance\n";
+               }
+               if (file_exists($sWikiRedirectsFile))
+               {
+                       echo "Importing wikipedia redirects...";
+                       pgsqlRunDropAndRestore($sWikiRedirectsFile);
+                       echo "...done\n";
+               }
+               else
+               {
+                       echo "WARNING: wikipedia redirect dump file not found - some place importance values may be missing\n";
+               }
+       }
+
+
        if ($aCMDResult['load-data'] || $aCMDResult['all'])
        {
        if ($aCMDResult['load-data'] || $aCMDResult['all'])
        {
-               echo "Load Data\n";
+               echo "Drop old Data\n";
                $bDidSomething = true;
 
                $oDB =& getDB();
                $bDidSomething = true;
 
                $oDB =& getDB();
@@ -179,12 +369,37 @@
                if (!pg_query($oDB->connection, 'CREATE SEQUENCE seq_place start 100000')) fail(pg_last_error($oDB->connection));
                echo '.';
 
                if (!pg_query($oDB->connection, 'CREATE SEQUENCE seq_place start 100000')) fail(pg_last_error($oDB->connection));
                echo '.';
 
+               $sSQL = 'select distinct partition from country_name';
+               $aPartitions = $oDB->getCol($sSQL);
+               if (PEAR::isError($aPartitions))
+               {
+                       fail($aPartitions->getMessage());
+               }
+               if (!$aCMDResult['no-partitions']) $aPartitions[] = 0;
+               foreach($aPartitions as $sPartition)
+               {
+                       if (!pg_query($oDB->connection, 'TRUNCATE location_road_'.$sPartition)) fail(pg_last_error($oDB->connection));
+                       echo '.';
+               }
+
+               // used by getorcreate_word_id to ignore frequent partial words
+               if (!pg_query($oDB->connection, 'CREATE OR REPLACE FUNCTION get_maxwordfreq() RETURNS integer AS $$ SELECT '.CONST_Max_Word_Frequency.' as maxwordfreq; $$ LANGUAGE SQL IMMUTABLE')) fail(pg_last_error($oDB->connection));
+               echo ".\n";
+
+               // pre-create the word list
+               if (!$aCMDResult['disable-token-precalc'])
+               {
+                       echo "Loading word list\n";
+                       pgsqlRunScriptFile(CONST_BasePath.'/data/words.sql');
+               }
+
+               echo "Load Data\n";
                $aDBInstances = array();
                for($i = 0; $i < $iInstances; $i++)
                {
                        $aDBInstances[$i] =& getDB(true);
                        $sSQL = 'insert into placex (osm_type, osm_id, class, type, name, admin_level, ';
                $aDBInstances = array();
                for($i = 0; $i < $iInstances; $i++)
                {
                        $aDBInstances[$i] =& getDB(true);
                        $sSQL = 'insert into placex (osm_type, osm_id, class, type, name, admin_level, ';
-                       $sSQL .= 'housenumber, street, isin, postcode, country_code, extratags, ';
+                       $sSQL .= 'housenumber, street, addr_place, isin, postcode, country_code, extratags, ';
                        $sSQL .= 'geometry) select * from place where osm_id % '.$iInstances.' = '.$i;
                        if ($aCMDResult['verbose']) echo "$sSQL\n";
                        if (!pg_send_query($aDBInstances[$i]->connection, $sSQL)) fail(pg_last_error($oDB->connection));
                        $sSQL .= 'geometry) select * from place where osm_id % '.$iInstances.' = '.$i;
                        if ($aCMDResult['verbose']) echo "$sSQL\n";
                        if (!pg_send_query($aDBInstances[$i]->connection, $sSQL)) fail(pg_last_error($oDB->connection));
@@ -201,6 +416,8 @@
                        echo '.';
                }
                echo "\n";
                        echo '.';
                }
                echo "\n";
+               echo "Reanalysing database...\n";
+               pgsqlRunScript('ANALYSE');
        }
 
        if ($aCMDResult['create-roads'])
        }
 
        if ($aCMDResult['create-roads'])
@@ -213,7 +430,7 @@
                {
                        $aDBInstances[$i] =& getDB(true);
                        if (!pg_query($aDBInstances[$i]->connection, 'set enable_bitmapscan = off')) fail(pg_last_error($oDB->connection));
                {
                        $aDBInstances[$i] =& getDB(true);
                        if (!pg_query($aDBInstances[$i]->connection, 'set enable_bitmapscan = off')) fail(pg_last_error($oDB->connection));
-                       $sSQL = 'select count(*) from (select insertLocationRoad(partition, place_id, country_code, geometry) from ';
+                       $sSQL = 'select count(*) from (select insertLocationRoad(partition, place_id, calculated_country_code, geometry) from ';
                        $sSQL .= 'placex where osm_id % '.$iInstances.' = '.$i.' and rank_search between 26 and 27 and class = \'highway\') as x ';
                        if ($aCMDResult['verbose']) echo "$sSQL\n";
                        if (!pg_send_query($aDBInstances[$i]->connection, $sSQL)) fail(pg_last_error($oDB->connection));
                        $sSQL .= 'placex where osm_id % '.$iInstances.' = '.$i.' and rank_search between 26 and 27 and class = \'highway\') as x ';
                        if ($aCMDResult['verbose']) echo "$sSQL\n";
                        if (!pg_send_query($aDBInstances[$i]->connection, $sSQL)) fail(pg_last_error($oDB->connection));
@@ -236,13 +453,15 @@
        {
                $bDidSomething = true;
 
        {
                $bDidSomething = true;
 
+               pgsqlRunScriptFile(CONST_BasePath.'/sql/tiger_import_start.sql');
+
                $aDBInstances = array();
                for($i = 0; $i < $iInstances; $i++)
                {
                        $aDBInstances[$i] =& getDB(true);
                }
 
                $aDBInstances = array();
                for($i = 0; $i < $iInstances; $i++)
                {
                        $aDBInstances[$i] =& getDB(true);
                }
 
-               foreach(glob(CONST_BasePath.'/data/tiger2009/*.sql') as $sFile)
+               foreach(glob(CONST_BasePath.'/data/tiger2011/*.sql') as $sFile)
                {
                        echo $sFile.': ';
                        $hFile = fopen($sFile, "r");
                {
                        echo $sFile.': ';
                        $hFile = fopen($sFile, "r");
@@ -271,7 +490,7 @@
                        }
 
                        fclose($hFile);
                        }
 
                        fclose($hFile);
-       
+
                        $bAnyBusy = true;
                        while($bAnyBusy)
                        {
                        $bAnyBusy = true;
                        while($bAnyBusy)
                        {
@@ -284,39 +503,111 @@
                        }
                        echo "\n";
                }
                        }
                        echo "\n";
                }
+
+               echo "Creating indexes\n";
+               pgsqlRunScriptFile(CONST_BasePath.'/sql/tiger_import_finish.sql');
        }
 
        if ($aCMDResult['calculate-postcodes'] || $aCMDResult['all'])
        {
        }
 
        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));
                $oDB =& getDB();
                if (!pg_query($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,country_code,geometry) ";
-               $sSQL .= "select 'P',nextval('seq_postcodes'),'place','postcode',postcode,country_code,";
-               $sSQL .= "ST_SetSRID(ST_Point(x,y),4326) as geometry from (select country_code,postcode,";
+               $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 .= "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 country_code,postcode) as x";
+               $sSQL .= "from placex where postcode is not null and calculated_country_code not in ('ie') group by calculated_country_code,postcode) as x";
                if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection));
 
                if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection));
 
-               $sSQL = "insert into placex (osm_type,osm_id,class,type,postcode,country_code,geometry) ";
+               $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',";
                $sSQL .= "ST_SetSRID(ST_Point(x,y),4326) as geometry from us_postcode";
                if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection));
        }
 
                $sSQL .= "select 'P',nextval('seq_postcodes'),'place','postcode',postcode,'us',";
                $sSQL .= "ST_SetSRID(ST_Point(x,y),4326) as geometry from us_postcode";
                if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection));
        }
 
-       if (($aCMDResult['osmosis-init'] || $aCMDResult['all']) && isset($aCMDResult['osmosis-init-date']))
+       if ($aCMDResult['osmosis-init'] || $aCMDResult['all'])
        {
                $bDidSomething = true;
        {
                $bDidSomething = true;
+               $oDB =& getDB();
+
+               if (!file_exists(CONST_Osmosis_Binary))
+               {
+                       echo "Please download osmosis.\nIf it is already installed, check the path in your local settings (settings/local.php) file.\n";
+                       if (!$aCMDResult['all'])
+                       {
+                               fail("osmosis not found in '".CONST_Osmosis_Binary."'");
+                       }
+               }
+               else
+               {
+                       if (file_exists(CONST_BasePath.'/settings/configuration.txt'))
+                       {
+                               echo "settings/configuration.txt already exists\n";
+                       }
+                       else
+                       {
+                               passthru(CONST_Osmosis_Binary.' --read-replication-interval-init '.CONST_BasePath.'/settings');
+                               // update osmosis configuration.txt with our settings
+                               passthru("sed -i 's!baseUrl=.*!baseUrl=".CONST_Replication_Url."!' ".CONST_BasePath.'/settings/configuration.txt');
+                               passthru("sed -i 's:maxInterval = .*:maxInterval = ".CONST_Replication_MaxInterval.":' ".CONST_BasePath.'/settings/configuration.txt');
+                       }
+
+                       // Find the last node in the DB
+                       $iLastOSMID = $oDB->getOne("select max(id) from planet_osm_nodes");
+
+                       // 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");
+                       // 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>                    22-Mar-2013 07:41    -
+                       preg_match_all('#<a href="[0-9]{3}/">([0-9]{3}/)</a>.*(([0-9]{2})-([A-z]{3})-([0-9]{4}) ([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;
 
 
-               if (!file_exists(CONST_BasePath.'/osmosis-0.38/bin/osmosis')) fail("please download osmosis");
-               if (file_exists(CONST_BasePath.'/settings/configuration.txt')) echo "settings/configuration.txt already exists\n";
-               else passthru(CONST_BasePath.'/osmosis-0.38/bin/osmosis --read-replication-interval-init '.CONST_BasePath.'/settings');
+                       $sRepURL .= $aRepMatch[1];
+                       $sRep = file_get_contents($sRepURL."?C=M;O=D");
+                       preg_match_all('#<a href="[0-9]{3}/">([0-9]{3}/)</a>.*(([0-9]{2})-([A-z]{3})-([0-9]{4}) ([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;
 
 
-               $sDate = $aCMDResult['osmosis-init-date'];
-               $sURL = 'http://toolserver.org/~mazder/replicate-sequences/?'.$sDate;
-               echo "Getting state file: $sURL\n";
-               $sStateFile = file_get_contents($sURL);
-               if (!$sStateFile || strlen($sStateFile) > 1000) fail("unable to obtain state file");
-               file_put_contents(CONST_BasePath.'/settings/state.txt', $sStateFile);
+                       $sRepURL .= $aRepMatch[1];
+                       $sRep = file_get_contents($sRepURL."?C=M;O=D");
+                       preg_match_all('#<a href="[0-9]{3}.state.txt">([0-9]{3}).state.txt</a>.*(([0-9]{2})-([A-z]{3})-([0-9]{4}) ([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_BasePath.'/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);
+               }
        }
 
        if ($aCMDResult['index'] || $aCMDResult['all'])
        }
 
        if ($aCMDResult['index'] || $aCMDResult['all'])
@@ -324,7 +615,70 @@
                $bDidSomething = true;
                $sOutputFile = '';
                if (isset($aCMDResult['index-output'])) $sOutputFile = ' -F '.$aCMDResult['index-output'];
                $bDidSomething = true;
                $sOutputFile = '';
                if (isset($aCMDResult['index-output'])) $sOutputFile = ' -F '.$aCMDResult['index-output'];
-               passthru(CONST_BasePath.'/nominatim/nominatim -i -d nominatim -t '.$iInstances.$sOutputFile);
+               $sBaseCmd = CONST_BasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'].' -t '.$iInstances.$sOutputFile;
+               passthruCheckReturn($sBaseCmd.' -R 4');
+               if (!$aCMDResult['index-noanalyse']) pgsqlRunScript('ANALYSE');
+               passthruCheckReturn($sBaseCmd.' -r 5 -R 25');
+               if (!$aCMDResult['index-noanalyse']) pgsqlRunScript('ANALYSE');
+               passthruCheckReturn($sBaseCmd.' -r 26');
+       }
+
+       if ($aCMDResult['create-search-indices'] || $aCMDResult['all'])
+       {
+               echo "Search indices\n";
+               $bDidSomething = true;
+               $oDB =& getDB();
+               $sSQL = 'select distinct partition from country_name';
+               $aPartitions = $oDB->getCol($sSQL);
+               if (PEAR::isError($aPartitions))
+               {
+                       fail($aPartitions->getMessage());
+               }
+               if (!$aCMDResult['no-partitions']) $aPartitions[] = 0;
+
+               $sTemplate = file_get_contents(CONST_BasePath.'/sql/indices.src.sql');
+               preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER);
+               foreach($aMatches as $aMatch)
+               {
+                       $sResult = '';
+                       foreach($aPartitions as $sPartitionName)
+                       {
+                               $sResult .= str_replace('-partition-', $sPartitionName, $aMatch[1]);
+                       }
+                       $sTemplate = str_replace($aMatch[0], $sResult, $sTemplate);
+               }
+
+               pgsqlRunScript($sTemplate);
+       }
+
+       if (isset($aCMDResult['create-website']))
+       {
+               $bDidSomething = true;
+               $sTargetDir = $aCMDResult['create-website'];
+               if (!is_dir($sTargetDir))
+               {
+                       echo "You must create the website directory before calling this function.\n";
+                       fail("Target directory does not exist.");
+               }
+
+               @symlink(CONST_BasePath.'/website/details.php', $sTargetDir.'/details.php');
+               @symlink(CONST_BasePath.'/website/reverse.php', $sTargetDir.'/reverse.php');
+               @symlink(CONST_BasePath.'/website/search.php', $sTargetDir.'/search.php');
+               @symlink(CONST_BasePath.'/website/search.php', $sTargetDir.'/index.php');
+               @symlink(CONST_BasePath.'/website/deletable.php', $sTargetDir.'/deletable.php');
+               @symlink(CONST_BasePath.'/website/polygons.php', $sTargetDir.'/polygons.php');
+               @symlink(CONST_BasePath.'/website/status.php', $sTargetDir.'/status.php');
+               @symlink(CONST_BasePath.'/website/images', $sTargetDir.'/images');
+               @symlink(CONST_BasePath.'/website/js', $sTargetDir.'/js');
+               @symlink(CONST_BasePath.'/website/css', $sTargetDir.'/css');
+               echo "Symlinks created\n";
+
+               $sTestFile = @file_get_contents(CONST_Website_BaseURL.'js/tiles.js');
+               if (!$sTestFile)
+               {
+                       echo "\nWARNING: Unable to access the website at ".CONST_Website_BaseURL."\n";
+                       echo "You may want to update settings/local.php with @define('CONST_Website_BaseURL', 'http://[HOST]/[PATH]/');\n";
+               }
        }
 
        if (!$bDidSomething)
        }
 
        if (!$bDidSomething)
@@ -336,9 +690,95 @@
        {
                if (!file_exists($sFilename)) fail('unable to find '.$sFilename);
 
        {
                if (!file_exists($sFilename)) fail('unable to find '.$sFilename);
 
-               // Convert database DSN to psql paramaters
+               // Convert database DSN to psql parameters
+               $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
+               if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
+               $sCMD = 'psql -p '.$aDSNInfo['port'].' -d '.$aDSNInfo['database'];
+
+               $ahGzipPipes = null;
+               if (preg_match('/\\.gz$/', $sFilename))
+               {
+                       $aDescriptors = array(
+                               0 => array('pipe', 'r'),
+                               1 => array('pipe', 'w'),
+                               2 => array('file', '/dev/null', 'a')
+                       );
+                       $hGzipProcess = proc_open('zcat '.$sFilename, $aDescriptors, $ahGzipPipes);
+                       if (!is_resource($hGzipProcess)) fail('unable to start zcat');
+                       $aReadPipe = $ahGzipPipes[1];
+                       fclose($ahGzipPipes[0]);
+               }
+               else
+               {
+                       $sCMD .= ' -f '.$sFilename;
+                       $aReadPipe = array('pipe', 'r');
+               }
+
+               $aDescriptors = array(
+                       0 => $aReadPipe,
+                       1 => array('pipe', 'w'),
+                       2 => array('file', '/dev/null', 'a')
+               );
+               $ahPipes = null;
+               $hProcess = proc_open($sCMD, $aDescriptors, $ahPipes);
+               if (!is_resource($hProcess)) fail('unable to start pgsql');
+
+
+               // TODO: error checking
+               while(!feof($ahPipes[1]))
+               {
+                       echo fread($ahPipes[1], 4096);
+               }
+               fclose($ahPipes[1]);
+
+               $iReturn = proc_close($hProcess);
+               if ($iReturn > 0)
+               {
+                       fail("pgsql returned with error code ($iReturn)");
+               }
+               if ($ahGzipPipes)
+               {
+                       fclose($ahGzipPipes[1]);
+                       proc_close($hGzipProcess);
+               }
+
+       }
+
+       function pgsqlRunScript($sScript)
+       {
+               // Convert database DSN to psql parameters
+               $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
+               if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
+               $sCMD = 'psql -p '.$aDSNInfo['port'].' -d '.$aDSNInfo['database'];
+               $sCMD .= ' -v ON_ERROR_STOP=1';
+               $aDescriptors = array(
+                       0 => array('pipe', 'r'),
+                       1 => STDOUT, 
+                       2 => STDERR
+               );
+               $ahPipes = null;
+               $hProcess = @proc_open($sCMD, $aDescriptors, $ahPipes);
+               if (!is_resource($hProcess)) fail('unable to start pgsql');
+
+               while(strlen($sScript))
+               {
+                       $written = fwrite($ahPipes[0], $sScript);
+                       $sScript = substr($sScript, $written);
+               }
+               fclose($ahPipes[0]);
+               $iReturn = proc_close($hProcess);
+               if ($iReturn > 0)
+               {
+                       fail("pgsql returned with error code ($iReturn)");
+               }
+       }
+
+       function pgsqlRunRestoreData($sDumpFile)
+       {
+               // Convert database DSN to psql parameters
                $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
                $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
-               $sCMD = 'psql -f '.$sFilename.' '.$aDSNInfo['database'];
+               if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
+               $sCMD = 'pg_restore -p '.$aDSNInfo['port'].' -d '.$aDSNInfo['database'].' -Fc -a '.$sDumpFile;
 
                $aDescriptors = array(
                        0 => array('pipe', 'r'),
 
                $aDescriptors = array(
                        0 => array('pipe', 'r'),
@@ -347,7 +787,7 @@
                );
                $ahPipes = null;
                $hProcess = proc_open($sCMD, $aDescriptors, $ahPipes);
                );
                $ahPipes = null;
                $hProcess = proc_open($sCMD, $aDescriptors, $ahPipes);
-               if (!is_resource($hProcess)) fail('unable to start pgsql');
+               if (!is_resource($hProcess)) fail('unable to start pg_restore');
 
                fclose($ahPipes[0]);
 
 
                fclose($ahPipes[0]);
 
@@ -358,15 +798,20 @@
                }
                fclose($ahPipes[1]);
 
                }
                fclose($ahPipes[1]);
 
-               proc_close($hProcess);
+               $iReturn = proc_close($hProcess);
+               if ($iReturn > 0)
+               {
+                       fail("pgsql returned with error code ($iReturn)");
+               }
        }
 
        }
 
-       function pgsqlRunScript($sScript)
+       function pgsqlRunDropAndRestore($sDumpFile)
        {
        {
-               // Convert database DSN to psql paramaters
+               // Convert database DSN to psql parameters
                $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
                if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
                $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
                if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
-               $sCMD = 'psql -p '.$aDSNInfo['port'].' '.$aDSNInfo['database'];
+               $sCMD = 'pg_restore -p '.$aDSNInfo['port'].' -d '.$aDSNInfo['database'].' -Fc --clean '.$sDumpFile;
+
                $aDescriptors = array(
                        0 => array('pipe', 'r'),
                        1 => array('pipe', 'w'),
                $aDescriptors = array(
                        0 => array('pipe', 'r'),
                        1 => array('pipe', 'w'),
@@ -374,9 +819,8 @@
                );
                $ahPipes = null;
                $hProcess = proc_open($sCMD, $aDescriptors, $ahPipes);
                );
                $ahPipes = null;
                $hProcess = proc_open($sCMD, $aDescriptors, $ahPipes);
-               if (!is_resource($hProcess)) fail('unable to start pgsql');
+               if (!is_resource($hProcess)) fail('unable to start pg_restore');
 
 
-               fwrite($ahPipes[0], $sScript);
                fclose($ahPipes[0]);
 
                // TODO: error checking
                fclose($ahPipes[0]);
 
                // TODO: error checking
@@ -386,5 +830,16 @@
                }
                fclose($ahPipes[1]);
 
                }
                fclose($ahPipes[1]);
 
-               proc_close($hProcess);
+               $iReturn = proc_close($hProcess);
+               if ($iReturn > 0)
+               {
+                       fail("pgsql returned with error code ($iReturn)");
+               }
+       }
+
+       function passthruCheckReturn($cmd)
+       {
+               $result = -1;
+               passthru($cmd, $result);
+               if ($result != 0) fail('Error executing external command: '.$cmd);
        }
        }
Open Source search based on OpenStreetMap data