Make minimum number of tags/relations of some type needed for some statistics configurable.

5 files changed, 21 insertions, 7 deletions

diff --git a/bin/taginfo-config.rb b/bin/taginfo-config.rb
index fbeb163..96d8a43 100755
--- a/bin/taginfo-config.rb
+++ b/bin/taginfo-config.rb
@@ -1,6 +1,6 @@
 #!/usr/bin/ruby
 #
-#  taginfo-config.rb [KEY]
+#  taginfo-config.rb KEY [DEFAULT]
 #
 
 require 'rubygems'
@@ -10,7 +10,7 @@ require File.expand_path(File.dirname(__FILE__)) + '/../web/lib/config.rb'
 
 TaginfoConfig.read
 
-value = TaginfoConfig.get(ARGV[0])
+value = TaginfoConfig.get(ARGV[0], ARGV[1])
 if value.nil?
     puts ''
     exit 1
diff --git a/sources/master/interesting_relation_types.sql b/sources/master/interesting_relation_types.sql
index 18aab2d..e2e7954 100644
--- a/sources/master/interesting_relation_types.sql
+++ b/sources/master/interesting_relation_types.sql
@@ -17,8 +17,9 @@ CREATE TABLE interesting_relation_types (
     rtype TEXT
 );
 
+-- MIN_COUNT_RELATIONS_PER_TYPE setting: sources.master.min_count_relations_per_type
 INSERT INTO interesting_relation_types (rtype)
-    SELECT value FROM db.tags WHERE key='type' AND count_relations > 100;
+    SELECT value FROM db.tags WHERE key='type' AND count_relations > __MIN_COUNT_RELATIONS_PER_TYPE__;
 
 ANALYZE interesting_relation_types;
 
diff --git a/sources/master/interesting_tags.sql b/sources/master/interesting_tags.sql
index 160e1fb..406b525 100644
--- a/sources/master/interesting_tags.sql
+++ b/sources/master/interesting_tags.sql
@@ -21,10 +21,11 @@ CREATE TABLE interesting_tags (
     value TEXT
 );
 
+-- MIN_COUNT_TAGS setting: sources.master.min_count_tags
 INSERT INTO interesting_tags (key, value)
-    SELECT DISTINCT key, NULL FROM db.keys WHERE count_all > 10000
+    SELECT DISTINCT key, NULL FROM db.keys WHERE count_all > __MIN_COUNT_TAGS__
     UNION
-    SELECT key, value FROM db.tags WHERE count_all > 10000;
+    SELECT key, value FROM db.tags WHERE count_all > __MIN_COUNT_TAGS__;
 
 DELETE FROM interesting_tags WHERE key IN ('created_by', 'ele', 'height', 'is_in', 'lanes', 'layer', 'maxspeed', 'name', 'ref', 'width') AND value IS NOT NULL;
 DELETE FROM interesting_tags WHERE value IS NOT NULL AND key LIKE '%:%';
diff --git a/sources/master/update.sh b/sources/master/update.sh
index d030134..63814a1 100755
--- a/sources/master/update.sh
+++ b/sources/master/update.sh
@@ -35,10 +35,12 @@ $EXEC_RUBY -pe "\$_.sub!(/__DIR__/, '$DIR')" search.sql | sqlite3 $DIR/taginfo-s
 rm -f $DATABASE
 
 echo "`$DATECMD` Create master database..."
+min_count_tags=`../../bin/taginfo-config.rb sources.master.min_count_tags 10000`
+min_count_relations_per_type=`../../bin/taginfo-config.rb sources.master.min_count_relations_per_type 100`
 sqlite3 $DATABASE <languages.sql
 $EXEC_RUBY -pe "\$_.sub!(/__DIR__/, '$DIR')" master.sql | sqlite3 $DATABASE
-$EXEC_RUBY -pe "\$_.sub!(/__DIR__/, '$DIR')" interesting_tags.sql | sqlite3 $DATABASE
-$EXEC_RUBY -pe "\$_.sub!(/__DIR__/, '$DIR')" interesting_relation_types.sql | sqlite3 $DATABASE
+$EXEC_RUBY -pe "\$_.sub!(/__DIR__/, '$DIR').sub!(/__MIN_COUNT_TAGS__/, '$min_count_tags')" interesting_tags.sql | sqlite3 $DATABASE
+$EXEC_RUBY -pe "\$_.sub!(/__DIR__/, '$DIR').sub!(/__MIN_COUNT_RELATIONS_PER_TYPE__/, '$min_count_relations_per_type')" interesting_relation_types.sql | sqlite3 $DATABASE
 
 echo "`$DATECMD` Updating history database..."
 if [ ! -e $HISTORYDB ]; then
diff --git a/taginfo-config-example.json b/taginfo-config-example.json
index fab2cb7..9795904 100644
--- a/taginfo-config-example.json
+++ b/taginfo-config-example.json
@@ -58,6 +58,16 @@
             "tagstats": "./tagstats",
             // The OSM planet file or extract where we read the OSM data from.
             "planetfile": "/osm/planet/var/current-planet.osm.pbf"
+        },
+        "master": {
+            // Minimum number of occurances of a tag to make it
+            // potentially "interesting", ie. create tag combination
+            // statistics for it.
+            "min_count_tags": 10000,
+            // Minimum number of relations per type to make this
+            // relation type "interesting", ie. to make it show
+            // up as a relation type.
+            "min_count_relations_per_type": 100
         }
     },
     "logging": {