add booleans for writing untagged nodes, ways, relations, and areas, …

…set them to true by default
ad-freiburg · Nov 21, 2024 · a032789 · a032789
1 parent 3769c8e
commit a032789
Show file tree

Hide file tree

Showing 4 changed files with 39 additions and 18 deletions.
diff --git a/include/osm2rdf/config/Config.h b/include/osm2rdf/config/Config.h
@@ -75,6 +75,11 @@ struct Config {
   bool addWayNodeSpatialMetadata = false;
   bool skipWikiLinks = false;
 
+  bool addUntaggedNodes = true;
+  bool addUntaggedWays = true;
+  bool addUntaggedRelations = true;
+  bool addUntaggedAreas = true;
+
   int numThreads = std::thread::hardware_concurrency();
 
   // Default settings for data

diff --git a/include/osm2rdf/osm/CountHandler.h b/include/osm2rdf/osm/CountHandler.h
@@ -26,6 +26,7 @@ namespace osm2rdf::osm {
 
 class CountHandler : public osmium::handler::Handler {
  public:
+  CountHandler(const osm2rdf::config::Config& config) : _config(config) {};
   void node(const osmium::Node& node);
   void relation(const osmium::Relation& relation);
   void way(const osmium::Way& way);
@@ -45,6 +46,8 @@ class CountHandler : public osmium::handler::Handler {
   bool _firstPassDone = false;
   size_t _minId = std::numeric_limits<size_t>::max();
   size_t _maxId = 0;
+
+  osm2rdf::config::Config _config;
 };
 }
 

diff --git a/src/osm/CountHandler.cpp b/src/osm/CountHandler.cpp
@@ -16,52 +16,46 @@
 // You should have received a copy of the GNU General Public License
 // along with osm2rdf.  If not, see <https://www.gnu.org/licenses/>.
 
-#include "osm2rdf/osm/CountHandler.h"
-
 #include <iostream>
 
+#include "osm2rdf/osm/CountHandler.h"
+
 // ____________________________________________________________________________
-void osm2rdf::osm::CountHandler::prepare_for_lookup() {
-  _firstPassDone = true;
-}
+void osm2rdf::osm::CountHandler::prepare_for_lookup() { _firstPassDone = true; }
 
 // ____________________________________________________________________________
-void osm2rdf::osm::CountHandler::node(const osmium::Node& node){
+void osm2rdf::osm::CountHandler::node(const osmium::Node& node) {
   if (node.positive_id() < _minId) _minId = node.positive_id();
   if (node.positive_id() > _maxId) _maxId = node.positive_id();
-  if (_firstPassDone)  {
+  if (_firstPassDone || (!_config.addUntaggedNodes && node.tags().empty())) {
     return;
   }
   _numNodes++;
 }
 
 // ____________________________________________________________________________
-void osm2rdf::osm::CountHandler::relation(const osmium::Relation&) {
-  if (_firstPassDone) {
+void osm2rdf::osm::CountHandler::relation(const osmium::Relation& rel) {
+  if (_firstPassDone || (!_config.addUntaggedRelations && rel.tags().empty())) {
     return;
   }
   _numRelations++;
 }
 
 // ____________________________________________________________________________
-void osm2rdf::osm::CountHandler::way(const osmium::Way&) {
-  if (_firstPassDone) {
+void osm2rdf::osm::CountHandler::way(const osmium::Way& way) {
+  if (_firstPassDone || (!_config.addUntaggedWays && way.tags().empty())) {
     return;
   }
   _numWays++;
 }
 
 // ____________________________________________________________________________
-size_t osm2rdf::osm::CountHandler::numNodes() const {
-  return _numNodes;
-}
+size_t osm2rdf::osm::CountHandler::numNodes() const { return _numNodes; }
 
 // ____________________________________________________________________________
 size_t osm2rdf::osm::CountHandler::numRelations() const {
   return _numRelations;
 }
 
 // ____________________________________________________________________________
-size_t osm2rdf::osm::CountHandler::numWays() const {
-  return _numWays;
-}
+size_t osm2rdf::osm::CountHandler::numWays() const { return _numWays; }
diff --git a/src/osm/OsmiumHandler.cpp b/src/osm/OsmiumHandler.cpp
@@ -56,7 +56,7 @@ void osm2rdf::osm::OsmiumHandler<W>::handle() {
     assembler_config.create_empty_areas = false;
     osmium::area::MultipolygonManager<osmium::area::Assembler> mp_manager{
         assembler_config};
-    osm2rdf::osm::CountHandler countHandler;
+    osm2rdf::osm::CountHandler countHandler(_config);
 
     // read relations for areas
     {
@@ -163,6 +163,11 @@ void osm2rdf::osm::OsmiumHandler<W>::handle() {
 template <typename W>
 void osm2rdf::osm::OsmiumHandler<W>::area(const osmium::Area& area) {
   _areasSeen++;
+
+  if (!_config.addUntaggedAreas && area.tags().empty()) {
+    return;
+  }
+
   try {
     auto osmArea = osm2rdf::osm::Area(area);
 #pragma omp task
@@ -187,6 +192,10 @@ template <typename W>
 void osm2rdf::osm::OsmiumHandler<W>::node(const osmium::Node& node) {
   _nodesSeen++;
 
+  if (!_config.addUntaggedNodes && node.tags().empty()) {
+    return;
+  }
+
   try {
     const auto& osmNode = osm2rdf::osm::Node(node);
 #pragma omp task
@@ -224,6 +233,11 @@ template <typename W>
 void osm2rdf::osm::OsmiumHandler<W>::relation(
     const osmium::Relation& relation) {
   _relationsSeen++;
+
+  if (!_config.addUntaggedRelations && relation.tags().empty()) {
+    return;
+  }
+
   try {
     // only task this away if we actually build the relation geometries,
     // otherwise this just adds multithreading overhead for nothing
@@ -267,6 +281,11 @@ void osm2rdf::osm::OsmiumHandler<W>::relation(
 template <typename W>
 void osm2rdf::osm::OsmiumHandler<W>::way(const osmium::Way& way) {
   _waysSeen++;
+
+  if (!_config.addUntaggedWays && way.tags().empty()) {
+    return;
+  }
+
   try {
     auto osmWay = osm2rdf::osm::Way(way);