Options for untagged entities

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

include/osm2rdf/config/Constants.h

@@ -175,6 +175,38 @@ const static inline std::string NO_WAY_GEOM_RELATIONS_OPTION_LONG =
 const static inline std::string NO_WAY_GEOM_RELATIONS_OPTION_HELP =
     "Do not dump way geometric relations";
 
+const static inline std::string NO_UNTAGGED_NODES_INFO =
+    "Do not output untagged nodes";
+const static inline std::string NO_UNTAGGED_NODES_OPTION_SHORT = "";
+const static inline std::string NO_UNTAGGED_NODES_OPTION_LONG =
+    "no-untagged-nodes";
+const static inline std::string NO_UNTAGGED_NODES_OPTION_HELP =
+    "Do not output untagged nodes";
+
+const static inline std::string NO_UNTAGGED_WAYS_INFO =
+    "Do not output untagged ways";
+const static inline std::string NO_UNTAGGED_WAYS_OPTION_SHORT = "";
+const static inline std::string NO_UNTAGGED_WAYS_OPTION_LONG =
+    "no-untagged-ways";
+const static inline std::string NO_UNTAGGED_WAYS_OPTION_HELP =
+    "Do not output untagged ways";
+
+const static inline std::string NO_UNTAGGED_RELATIONS_INFO =
+    "Do not output untagged relations";
+const static inline std::string NO_UNTAGGED_RELATIONS_OPTION_SHORT = "";
+const static inline std::string NO_UNTAGGED_RELATIONS_OPTION_LONG =
+    "no-untagged-relations";
+const static inline std::string NO_UNTAGGED_RELATIONS_OPTION_HELP =
+    "Do not output untagged relations";
+
+const static inline std::string NO_UNTAGGED_AREAS_INFO =
+    "Do not output untagged areas";
+const static inline std::string NO_UNTAGGED_AREAS_OPTION_SHORT = "";
+const static inline std::string NO_UNTAGGED_AREAS_OPTION_LONG =
+    "no-untagged-areas";
+const static inline std::string NO_UNTAGGED_AREAS_OPTION_HELP =
+    "Do not output untagged areas";
+
 const static inline std::string ADD_AREA_WAY_LINESTRINGS_INFO =
     "Adding linestrings for ways which form areas";
 const static inline std::string ADD_AREA_WAY_LINESTRINGS_OPTION_SHORT = "";

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;
 };
 }
 

src/config/Config.cpp

@@ -17,12 +17,12 @@
 // 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/config/Config.h"
-
 #include <filesystem>
 #include <iostream>
 #include <string>
 
+#include "osm2rdf/config/Config.h"
+
 #if defined(_OPENMP)
 #include "omp.h"
 #endif
@@ -87,6 +87,22 @@ std::string osm2rdf::config::Config::getInfo(std::string_view prefix) const {
             << osm2rdf::config::constants::ADD_AREA_WAY_LINESTRINGS_INFO;
       }
     }
+    if (!addUntaggedNodes) {
+      oss << "\n"
+          << prefix << osm2rdf::config::constants::NO_UNTAGGED_NODES_INFO;
+    }
+    if (!addUntaggedWays) {
+      oss << "\n"
+          << prefix << osm2rdf::config::constants::NO_UNTAGGED_WAYS_INFO;
+    }
+    if (!addUntaggedRelations) {
+      oss << "\n"
+          << prefix << osm2rdf::config::constants::NO_UNTAGGED_RELATIONS_INFO;
+    }
+    if (!addUntaggedAreas) {
+      oss << "\n"
+          << prefix << osm2rdf::config::constants::NO_UNTAGGED_AREAS_INFO;
+    }
     if (simplifyWKT > 0) {
       oss << "\n" << prefix << osm2rdf::config::constants::SIMPLIFY_WKT_INFO;
       oss << "\n"
@@ -145,13 +161,12 @@ std::string osm2rdf::config::Config::getInfo(std::string_view prefix) const {
     }
   }
   oss << "\n" << prefix << osm2rdf::config::constants::SECTION_MISCELLANEOUS;
-  oss << "\n"
-      << prefix << "Num Threads: " << numThreads;
+  oss << "\n" << prefix << "Num Threads: " << numThreads;
 
   if (!storeLocations.empty()) {
     oss << "\n"
-        << prefix << osm2rdf::config::constants::STORE_LOCATIONS_INFO
-        << " " << storeLocations;
+        << prefix << osm2rdf::config::constants::STORE_LOCATIONS_INFO << " "
+        << storeLocations;
   }
 
   if (writeRDFStatistics) {
@@ -265,6 +280,30 @@ void osm2rdf::config::Config::fromArgs(int argc, char** argv) {
           osm2rdf::config::constants::ADD_AREA_WAY_LINESTRINGS_OPTION_LONG,
           osm2rdf::config::constants::ADD_AREA_WAY_LINESTRINGS_OPTION_HELP);
 
+  auto noUntaggedNodesOp =
+      parser.add<popl::Switch, popl::Attribute::expert>(
+          osm2rdf::config::constants::NO_UNTAGGED_NODES_OPTION_SHORT,
+          osm2rdf::config::constants::NO_UNTAGGED_NODES_OPTION_LONG,
+          osm2rdf::config::constants::NO_UNTAGGED_NODES_OPTION_HELP);
+
+  auto noUntaggedWaysOp =
+      parser.add<popl::Switch, popl::Attribute::expert>(
+          osm2rdf::config::constants::NO_UNTAGGED_WAYS_OPTION_SHORT,
+          osm2rdf::config::constants::NO_UNTAGGED_WAYS_OPTION_LONG,
+          osm2rdf::config::constants::NO_UNTAGGED_WAYS_OPTION_HELP);
+
+  auto noUntaggedRelationsOp =
+      parser.add<popl::Switch, popl::Attribute::expert>(
+          osm2rdf::config::constants::NO_UNTAGGED_RELATIONS_OPTION_SHORT,
+          osm2rdf::config::constants::NO_UNTAGGED_RELATIONS_OPTION_LONG,
+          osm2rdf::config::constants::NO_UNTAGGED_RELATIONS_OPTION_HELP);
+
+  auto noUntaggedAreasOp =
+      parser.add<popl::Switch, popl::Attribute::expert>(
+          osm2rdf::config::constants::NO_UNTAGGED_AREAS_OPTION_SHORT,
+          osm2rdf::config::constants::NO_UNTAGGED_AREAS_OPTION_LONG,
+          osm2rdf::config::constants::NO_UNTAGGED_AREAS_OPTION_HELP);
+
   auto addWayMetadataOp = parser.add<popl::Switch>(
       osm2rdf::config::constants::ADD_WAY_METADATA_OPTION_SHORT,
       osm2rdf::config::constants::ADD_WAY_METADATA_OPTION_LONG,
@@ -288,11 +327,10 @@ void osm2rdf::config::Config::fromArgs(int argc, char** argv) {
           osm2rdf::config::constants::AUX_GEO_FILES_OPTION_LONG,
           osm2rdf::config::constants::AUX_GEO_FILES_OPTION_HELP);
 
-  auto numThreadsOp =
-      parser.add<popl::Value<int>, popl::Attribute::advanced>(
-          osm2rdf::config::constants::NUM_THREADS_OPTION_SHORT,
-          osm2rdf::config::constants::NUM_THREADS_OPTION_LONG,
-          osm2rdf::config::constants::NUM_THREADS_OPTION_HELP, numThreads);
+  auto numThreadsOp = parser.add<popl::Value<int>, popl::Attribute::advanced>(
+      osm2rdf::config::constants::NUM_THREADS_OPTION_SHORT,
+      osm2rdf::config::constants::NUM_THREADS_OPTION_LONG,
+      osm2rdf::config::constants::NUM_THREADS_OPTION_HELP, numThreads);
 
   auto semicolonTagKeysOp =
       parser.add<popl::Value<std::string>, popl::Attribute::advanced>(
@@ -444,6 +482,11 @@ void osm2rdf::config::Config::fromArgs(int argc, char** argv) {
     wktDeviation = wktDeviationOp->value();
     wktPrecision = wktPrecisionOp->value();
 
+    addUntaggedNodes = !noUntaggedNodesOp->is_set();
+    addUntaggedWays = !noUntaggedWaysOp->is_set();
+    addUntaggedRelations = !noUntaggedRelationsOp->is_set();
+    addUntaggedAreas = !noUntaggedAreasOp->is_set();
+
     addWayNodeOrder |= addWayNodeSpatialMetadata;
 
     if (semicolonTagKeysOp->is_set()) {
@@ -464,7 +507,9 @@ void osm2rdf::config::Config::fromArgs(int argc, char** argv) {
     // Output
     output = outputOp->value();
     outputFormat = outputFormatOp->value();
-    outputCompress = outputCompressOp->value() == "none" ? NONE : (outputCompressOp->value() == "gz" ? GZ : BZ2);
+    outputCompress = outputCompressOp->value() == "none"
+                         ? NONE
+                         : (outputCompressOp->value() == "gz" ? GZ : BZ2);
     outputKeepFiles = outputKeepFilesOp->is_set();
     if (output.empty()) {
       outputCompress = NONE;

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; }

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);