add simple tests for postcode import

[nominatim.git] / utils / setup.php

diff --git a/utils/setup.php b/utils/setup.php
index eeba0a28e23ac79d1bedae806e6533c6a680ae75..67c7fd7f80b8be31139dca46684c29fdd34c7e87 100755 (executable)
--- a/utils/setup.php
+++ b/utils/setup.php
@@ -407,9 +407,13 @@ if ($aCMDResult['load-data'] || $aCMDResult['all']) {
 
     $sDatabaseDate = getDatabaseDate($oDB);
     pg_query($oDB->connection, 'TRUNCATE import_status');
-    $sSQL = "INSERT INTO import_status (lastimportdate) VALUES('".$sDatabaseDate."')";
-    pg_query($oDB->connection, $sSQL);
-    echo "Latest data imported from $sDatabaseDate.\n";
+    if ($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 ($aCMDResult['import-tiger-data']) {
@@ -489,24 +493,59 @@ if ($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,address,country_code,geometry) ";
-    $sSQL .= "select 'P',nextval('seq_postcodes'),'place','postcode',";
-    $sSQL .= "hstore('postcode', pc),country_code,";
-    $sSQL .= "ST_SetSRID(ST_Point(x,y),4326) as geometry from (select country_code,";
-    $sSQL .= "address->'postcode' as pc,";
-    $sSQL .= "avg(st_x(st_centroid(geometry))) as x,avg(st_y(st_centroid(geometry))) as y ";
-    $sSQL .= "from placex where address ? 'postcode' group by country_code,pc) 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,address,country_code,geometry) ";
-        $sSQL .= "select 'P',nextval('seq_postcodes'),'place','postcode',";
-        $sSQL .= "hstore('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));
     }
+
+    // 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 (!$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 ";
+    $sSQL .= "(SELECT distinct(postcode) as v FROM location_postcode) p";
+
+    if (!pg_query($oDB->connection, $sSQL)) {
+        fail(pg_last_error($oDB->connection));
+    }
 }
 
 if ($aCMDResult['osmosis-init']) {
@@ -523,6 +562,11 @@ if ($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']) {