aggregate and filter rivers

Makefile

@@ -1,11 +1,7 @@
-# HACK HACK
-EMPTY :=
-SPACE := $(EMPTY) $(EMPTY)
-
-RIVERS ?= Visinčia Šalčia Žeimena Lakaja Nemunas
+WHERE ?= name='Visinčia' OR name='Šalčia' OR name='Nemunas'
 
 .faux_filter-rivers: .faux_import-osm
-	./filter-rivers-query.awk $(RIVERS) | ./db -f -
+	./db -v where="$(WHERE)" -f aggregate-rivers.sql
 	touch $@
 
 .faux_import-osm: lithuania-latest.osm.pbf .faux.db

aggregate-rivers.sql

@@ -0,0 +1,37 @@
+/* Aggregates rivers by name and proximity. */
+drop function if exists aggregate_rivers;
+create function aggregate_rivers() returns table(osm_id bigint, name text, way geometry) as $$
+declare
+  c record;
+  cc record;
+  changed boolean;
+begin
+  while (select count(1) from agg_tmp_objects) > 0 loop
+    select * from agg_tmp_objects limit 1 into c;
+    delete from agg_tmp_objects a where a.osm_id = c.osm_id;
+    changed = true;
+    while changed loop
+      changed = false;
+      for cc in (select * from agg_tmp_objects a where a.name = c.name and st_dwithin(a.way, c.way, 500)) loop
+        c.way = st_linemerge(st_union(c.way, cc.way));
+        delete from agg_tmp_objects a where a.osm_id = cc.osm_id;
+        changed = true;
+      end loop;
+    end loop; -- while changed
+    return query select c.osm_id, c.name, c.way;
+  end loop; -- count(1) from agg_tmp_objects > 0
+  return;
+end
+$$ language plpgsql;
+
+create temporary table agg_tmp_objects (osm_id bigint, name text, way geometry);
+create index agg_tmp_objects_id on agg_tmp_objects(osm_id);
+create index agg_tmp_objects_gix on agg_tmp_objects using gist(way) include(name);
+
+insert into agg_tmp_objects
+  select p.osm_id, p.name, p.way from planet_osm_line p
+    where waterway in ('river', 'stream', 'canal') and :where;
+
+drop table if exists agg_rivers;
+create table agg_rivers as (select * from aggregate_rivers());
+drop table agg_tmp_objects;

filter-rivers-query.awk

@@ -1,12 +0,0 @@
-#!/usr/bin/awk -f
-
-BEGIN {
-    print "DROP TABLE IF EXISTS rivers;"
-    printf "CREATE TABLE rivers AS SELECT name,way FROM planet_osm_line WHERE "
-    for (i = 1; i < ARGC; i++) {
-        printf "name='%s'", ARGV[i]
-        if (i != ARGC - 1)
-            printf " OR ";
-    }
-    print ";";
-}