Merge remote-tracking branch 'origin/master' into osmline

[nominatim.git] / utils / setup.php

diff --git a/utils/setup.php b/utils/setup.php
index 60fe5dace6734689bfc03aa8e45bfc99f25ceef5..caaf0f9126141dba717782a8804cdfa03becc712 100755 (executable)
--- a/utils/setup.php
+++ b/utils/setup.php
@@ -156,9 +156,10 @@
                {
                        echo "WARNING: external UK postcode table not found.\n";
                }
                {
                        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');
+               if (CONST_Use_Extra_US_Postcodes)
+               {
+                       pgsqlRunScriptFile(CONST_BasePath.'/data/us_postcode.sql');
+               }

 
                if ($aCMDResult['no-partitions'])
                {
 
                if ($aCMDResult['no-partitions'])
                {


@@ -216,12 +217,7 @@
                echo "Functions\n";
                $bDidSomething = true;
                if (!file_exists(CONST_InstallPath.'/module/nominatim.so')) fail("nominatim module not built");
                echo "Functions\n";
                $bDidSomething = true;
                if (!file_exists(CONST_InstallPath.'/module/nominatim.so')) fail("nominatim module not built");

-               $sTemplate = file_get_contents(CONST_BasePath.'/sql/functions.sql');
-               $sTemplate = str_replace('{modulepath}', CONST_InstallPath.'/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);
+               create_sql_functions($aCMDResult);

        }
 
        if ($aCMDResult['create-tables'] || $aCMDResult['all'])
        }
 
        if ($aCMDResult['create-tables'] || $aCMDResult['all'])


@@ -247,10 +243,7 @@
 
                // re-run the functions
                echo "Functions\n";
 
                // re-run the functions
                echo "Functions\n";

-               $sTemplate = file_get_contents(CONST_BasePath.'/sql/functions.sql');
-               $sTemplate = str_replace('{modulepath}',
-                                            CONST_InstallPath.'/module', $sTemplate);
-               pgsqlRunScript($sTemplate);
+               create_sql_functions($aCMDResult);

        }
 
        if ($aCMDResult['create-partition-tables'] || $aCMDResult['all'])
        }
 
        if ($aCMDResult['create-partition-tables'] || $aCMDResult['all'])


@@ -360,6 +353,8 @@
                echo '.';
                if (!pg_query($oDB->connection, 'TRUNCATE placex')) fail(pg_last_error($oDB->connection));
                echo '.';
                echo '.';
                if (!pg_query($oDB->connection, 'TRUNCATE placex')) fail(pg_last_error($oDB->connection));
                echo '.';

+               if (!pg_query($oDB->connection, 'TRUNCATE location_property_osmline')) fail(pg_last_error($oDB->connection));
+               echo '.';

                if (!pg_query($oDB->connection, 'TRUNCATE place_addressline')) fail(pg_last_error($oDB->connection));
                echo '.';
                if (!pg_query($oDB->connection, 'TRUNCATE place_boundingbox')) fail(pg_last_error($oDB->connection));
                if (!pg_query($oDB->connection, 'TRUNCATE place_addressline')) fail(pg_last_error($oDB->connection));
                echo '.';
                if (!pg_query($oDB->connection, 'TRUNCATE place_boundingbox')) fail(pg_last_error($oDB->connection));


@@ -401,27 +396,70 @@
 
                echo "Load Data\n";
                $aDBInstances = array();
 
                echo "Load Data\n";
                $aDBInstances = array();

+        
+        $aQueriesPlacex = array();
+        $aQueriesOsmline = array();
+        // the query is divided into parcels, so that the work between the processes, i.e. the DBInstances, will be evenly distributed
+        $iNumberOfParcels = 100;
+        for($i = 0; $i < $iNumberOfParcels; $i++)
+        {
+                       $sSQL = 'insert into placex (osm_type, osm_id, class, type, name, admin_level, ';
+                       $sSQL .= 'housenumber, street, addr_place, isin, postcode, country_code, extratags, ';
+                       $sSQL .= 'geometry) select * from place where osm_id % '.$iNumberOfParcels.' = '.$i.' and not ';
+                       $sSQL .= '(class=\'place\' and type=\'houses\' and osm_type=\'W\' and ST_GeometryType(geometry) = \'ST_LineString\');';
+                       array_push($aQueriesPlacex, $sSQL);
+                       $sSQL = 'select insert_osmline (osm_id, housenumber, street, addr_place, postcode, country_code, ';
+                       $sSQL .= 'geometry) from place where osm_id % '.$iNumberOfParcels.' = '.$i.' and ';
+                       $sSQL .= 'class=\'place\' and type=\'houses\' and osm_type=\'W\' and ST_GeometryType(geometry) = \'ST_LineString\'';
+                       array_push($aQueriesOsmline, $sSQL);
+               }
+        

                for($i = 0; $i < $iInstances; $i++)
                {
                        $aDBInstances[$i] =& getDB(true);
                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, 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));

                }
                }

-               $bAnyBusy = true;
-               while($bAnyBusy)
+               // now execute the query blocks, in the first round for placex, then for osmline, 
+               // because insert_osmline depends on the placex table
+               echo 'Inserting from place to placex.';
+               $aQueries = $aQueriesPlacex;
+               for($j = 0; $j < 2; $j++)

                {
                {

-                       $bAnyBusy = false;
-                       for($i = 0; $i < $iInstances; $i++)
+                       $bAnyBusy = true;
+                       while($bAnyBusy)

                        {
                        {

-                               if (pg_connection_busy($aDBInstances[$i]->connection)) $bAnyBusy = true;
+                               $bAnyBusy = false;
+
+                               for($i = 0; $i < $iInstances; $i++)
+                               {
+                                       if (pg_connection_busy($aDBInstances[$i]->connection)) 
+                                       {
+                                               $bAnyBusy = true;
+                                       }
+                                       else if (count($aQueries) > 0)
+                                       {
+                                               $query = array_pop($aQueries);
+                                               if (!pg_send_query($aDBInstances[$i]->connection, $query))
+                                               {
+                                                       fail(pg_last_error($oDB->connection));
+                                               }
+                                               else
+                                               {
+                                                       pg_get_result($aDBInstances[$i]->connection);
+                                                       $bAnyBusy = true;
+                                               }
+                                       }
+                               }
+                               sleep(1);
+                               echo '.';
+                       }
+                       echo "\n";
+                       if ($j == 0)  //for the second round with osmline
+                       {
+                               echo 'Inserting from place to osmline.';
+                               $aQueries = $aQueriesOsmline;

                        }
                        }

-                       sleep(1);
-                       echo '.';

                }
                }

-               echo "\n";
+               

                echo "Reanalysing database...\n";
                pgsqlRunScript('ANALYSE');
        }
                echo "Reanalysing database...\n";
                pgsqlRunScript('ANALYSE');
        }


@@ -509,10 +547,13 @@
                $sSQL .= "from placex where postcode is not null group by calculated_country_code,postcode) as x";
                if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection));
 
                $sSQL .= "from placex where postcode is not null group by calculated_country_code,postcode) as x";
                if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($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,'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 (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',";
+                       $sSQL .= "ST_SetSRID(ST_Point(x,y),4326) as geometry from us_postcode";
+                       if (!pg_query($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
        }
 
        if ($aCMDResult['osmosis-init'] || ($aCMDResult['all'] && !$aCMDResult['drop'])) // no use doing osmosis-init when dropping update tables


@@ -625,14 +666,6 @@
        {
                echo "Search indices\n";
                $bDidSomething = true;
        {
                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');
                $sTemplate = replace_tablespace('{ts:address-index}',
 
                $sTemplate = file_get_contents(CONST_BasePath.'/sql/indices.src.sql');
                $sTemplate = replace_tablespace('{ts:address-index}',


@@ -641,16 +674,6 @@
                                                CONST_Tablespace_Search_Index, $sTemplate);
                $sTemplate = replace_tablespace('{ts:aux-index}',
                                                CONST_Tablespace_Aux_Index, $sTemplate);
                                                CONST_Tablespace_Search_Index, $sTemplate);
                $sTemplate = replace_tablespace('{ts:aux-index}',
                                                CONST_Tablespace_Aux_Index, $sTemplate);

-               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);
        }
 
                pgsqlRunScript($sTemplate);
        }


@@ -922,3 +945,31 @@
                return $sSql;
        }
 
                return $sSql;
        }
 

+       function create_sql_functions($aCMDResult)
+       {
+               $sTemplate = file_get_contents(CONST_BasePath.'/sql/functions.sql');
+               $sTemplate = str_replace('{modulepath}', CONST_InstallPath.'/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);
+               }
+               if (!CONST_Use_US_Tiger_Data)
+               {
+                       $sTemplate = str_replace('-- %NOTIGERDATA% ', '', $sTemplate);
+               }
+               if (!CONST_Use_Aux_Location_data)
+               {
+                       $sTemplate = str_replace('-- %NOAUXDATA% ', '', $sTemplate);
+               }
+               pgsqlRunScript($sTemplate);
+
+       }
+