proper prefixes for OSM changesets and user, also output OSM uid

ad-freiburg · Jan 10, 2025 · 0dc87ed · 0dc87ed
1 parent 2501399
commit 0dc87ed
Show file tree

Hide file tree

Showing 10 changed files with 63 additions and 9 deletions.
diff --git a/include/osm2rdf/osm/Node.h b/include/osm2rdf/osm/Node.h
@@ -37,6 +37,7 @@ class Node {
   [[nodiscard]] osm2rdf::osm::generic::changeset_id_t changeset() const noexcept;
   [[nodiscard]] std::time_t timestamp() const noexcept;
   [[nodiscard]] std::string user() const noexcept;
+  [[nodiscard]] id_t uid() const noexcept;
   [[nodiscard]] osm2rdf::osm::generic::version_t version() const noexcept;
   [[nodiscard]] bool visible() const noexcept;
   [[nodiscard]] const ::util::geo::DBox envelope() const noexcept;
@@ -51,6 +52,7 @@ class Node {
   osm2rdf::osm::generic::changeset_id_t _changeset;
   std::time_t _timestamp;
   std::string _user;
+  id_t _uid;
   osm2rdf::osm::generic::version_t _version;
   bool _visible;
   ::util::geo::DPoint _geom;

diff --git a/include/osm2rdf/osm/Relation.h b/include/osm2rdf/osm/Relation.h
@@ -41,6 +41,7 @@ class Relation {
   [[nodiscard]] osm2rdf::osm::generic::changeset_id_t changeset() const noexcept;
   [[nodiscard]] std::time_t timestamp() const noexcept;
   [[nodiscard]] std::string user() const noexcept;
+  [[nodiscard]] id_t uid() const noexcept;
   [[nodiscard]] osm2rdf::osm::generic::version_t version() const noexcept;
   [[nodiscard]] bool visible() const noexcept;
   [[nodiscard]] const std::vector<osm2rdf::osm::RelationMember>& members()
@@ -65,6 +66,7 @@ class Relation {
   osm2rdf::osm::generic::changeset_id_t _changeset;
   std::time_t _timestamp;
   std::string _user;
+  id_t _uid;
   osm2rdf::osm::generic::version_t _version;
   bool _visible;
   std::vector<osm2rdf::osm::RelationMember> _members;

diff --git a/include/osm2rdf/osm/Way.h b/include/osm2rdf/osm/Way.h
@@ -40,6 +40,7 @@ class Way {
   [[nodiscard]] osm2rdf::osm::generic::changeset_id_t changeset() const noexcept;
   [[nodiscard]] std::time_t timestamp() const noexcept;
   [[nodiscard]] std::string user() const noexcept;
+  [[nodiscard]] id_t uid() const noexcept;
   [[nodiscard]] osm2rdf::osm::generic::version_t version() const noexcept;
   [[nodiscard]] bool visible() const noexcept;
   [[nodiscard]] bool closed() const noexcept;
@@ -61,6 +62,7 @@ class Way {
   osm2rdf::osm::generic::changeset_id_t _changeset;
   std::time_t _timestamp;
   std::string _user;
+  id_t _uid;
   osm2rdf::osm::generic::version_t _version;
   bool _visible;
   std::vector<osm2rdf::osm::Node> _nodes;

diff --git a/include/osm2rdf/ttl/Constants.h b/include/osm2rdf/ttl/Constants.h
@@ -28,12 +28,16 @@ const static inline std::string NAMESPACE__GEOSPARQL = "geo";
 const static inline std::string NAMESPACE__OHM_NODE = "ohmnode";
 const static inline std::string NAMESPACE__OHM_RELATION = "ohmrel";
 const static inline std::string NAMESPACE__OHM_WAY = "ohmway";
+const static inline std::string NAMESPACE__OHM_USER = "ohmuser";
+const static inline std::string NAMESPACE__OHM_CHANGESET = "ohmchangeset";
 const static inline std::string NAMESPACE__OHM = "ohm";
 const static inline std::string NAMESPACE__OPENGIS = "ogc";
 const static inline std::string NAMESPACE__OSM_NODE = "osmnode";
 const static inline std::string NAMESPACE__OSM_RELATION = "osmrel";
 const static inline std::string NAMESPACE__OSM_TAG = "osmkey";
 const static inline std::string NAMESPACE__OSM_WAY = "osmway";
+const static inline std::string NAMESPACE__OSM_USER = "osmuser";
+const static inline std::string NAMESPACE__OSM_CHANGESET = "osmchangeset";
 const static inline std::string NAMESPACE__OSM_META = "osmmeta";
 const static inline std::string NAMESPACE__OSM = "osm";
 const static inline std::string NAMESPACE__OSM2RDF = "osm2rdf";
@@ -74,6 +78,7 @@ inline std::string IRI__OSMMETA_CHANGESET;
 inline std::string IRI__OSM2RDF__LENGTH;
 inline std::string IRI__OSMMETA_TIMESTAMP;
 inline std::string IRI__OSMMETA_USER;
+inline std::string IRI__OSMMETA_UID;
 inline std::string IRI__OSMMETA_VERSION;
 inline std::string IRI__OSMMETA_VISIBLE;
 inline std::string IRI__OSMWAY_IS_CLOSED;
@@ -87,6 +92,8 @@ inline std::string IRI__OSM_NODE;
 inline std::string IRI__OSM_RELATION;
 inline std::string IRI__OSM_TAG;
 inline std::string IRI__OSM_WAY;
+inline std::string IRI__OSM_USER;
+inline std::string IRI__OSM_CHANGESET;
 
 inline std::string IRI__RDF_TYPE;
 
@@ -111,6 +118,10 @@ const static inline std::vector<std::string> RELATION_NAMESPACE = {
     NAMESPACE__OSM_RELATION, NAMESPACE__OHM_RELATION};
 const static inline std::vector<std::string> WAY_NAMESPACE = {
     NAMESPACE__OSM_WAY, NAMESPACE__OHM_WAY};
+const static inline std::vector<std::string> CHANGESET_NAMESPACE = {
+    NAMESPACE__OSM_CHANGESET, NAMESPACE__OHM_CHANGESET};
+const static inline std::vector<std::string> USER_NAMESPACE = {
+    NAMESPACE__OSM_USER, NAMESPACE__OHM_USER};
 
 }  // namespace osm2rdf::ttl::constants
 

diff --git a/src/osm/FactHandler.cpp b/src/osm/FactHandler.cpp
@@ -33,6 +33,7 @@
 
 using osm2rdf::osm::constants::AREA_PRECISION;
 using osm2rdf::osm::constants::BASE_SIMPLIFICATION_FACTOR;
+using osm2rdf::ttl::constants::CHANGESET_NAMESPACE;
 using osm2rdf::ttl::constants::DATASET_ID;
 using osm2rdf::ttl::constants::IRI__GEOSPARQL__AS_WKT;
 using osm2rdf::ttl::constants::IRI__GEOSPARQL__HAS_CENTROID;
@@ -46,12 +47,15 @@ using osm2rdf::ttl::constants::IRI__OSM2RDF_GEOM__OBB;
 using osm2rdf::ttl::constants::IRI__OSM2RDF_MEMBER__ID;
 using osm2rdf::ttl::constants::IRI__OSM2RDF_MEMBER__POS;
 using osm2rdf::ttl::constants::IRI__OSM2RDF_MEMBER__ROLE;
+using osm2rdf::ttl::constants::IRI__OSM_CHANGESET;
 using osm2rdf::ttl::constants::IRI__OSM_NODE;
 using osm2rdf::ttl::constants::IRI__OSM_RELATION;
 using osm2rdf::ttl::constants::IRI__OSM_TAG;
+using osm2rdf::ttl::constants::IRI__OSM_USER;
 using osm2rdf::ttl::constants::IRI__OSM_WAY;
 using osm2rdf::ttl::constants::IRI__OSMMETA_CHANGESET;
 using osm2rdf::ttl::constants::IRI__OSMMETA_TIMESTAMP;
+using osm2rdf::ttl::constants::IRI__OSMMETA_UID;
 using osm2rdf::ttl::constants::IRI__OSMMETA_USER;
 using osm2rdf::ttl::constants::IRI__OSMMETA_VERSION;
 using osm2rdf::ttl::constants::IRI__OSMMETA_VISIBLE;
@@ -82,6 +86,7 @@ using osm2rdf::ttl::constants::NAMESPACE__OSM_WAY;
 using osm2rdf::ttl::constants::NAMESPACE__WIKIDATA_ENTITY;
 using osm2rdf::ttl::constants::NODE_NAMESPACE;
 using osm2rdf::ttl::constants::RELATION_NAMESPACE;
+using osm2rdf::ttl::constants::USER_NAMESPACE;
 using osm2rdf::ttl::constants::WAY_NAMESPACE;
 
 // ____________________________________________________________________________
@@ -397,16 +402,25 @@ void osm2rdf::osm::FactHandler<W>::writeMeta(const std::string& subj,
   if (object.changeset() != 0) {
     _writer->writeTriple(
         subj, IRI__OSMMETA_CHANGESET,
-        _writer->generateLiteralUnsafe(std::to_string(object.changeset()),
-                                       "^^" + IRI__XSD_INTEGER));
+        _writer->generateIRI(CHANGESET_NAMESPACE[_config.sourceDataset],
+                             object.changeset()));
   }
 
   writeSecondsAsISO(subj, IRI__OSMMETA_TIMESTAMP, object.timestamp());
 
   // avoid writing empty users, drop entire triple
   if (!object.user().empty()) {
-    _writer->writeTriple(subj, IRI__OSMMETA_USER,
-                         _writer->generateLiteral(object.user(), ""));
+    _writer->writeTriple(
+        subj, IRI__OSMMETA_USER,
+        _writer->generateIRI(USER_NAMESPACE[_config.sourceDataset],
+                             object.user()));
+  }
+
+  // avoid writing empty user IDs, drop entire triple
+  if (object.uid() != 0) {
+    _writer->writeTriple(subj, IRI__OSMMETA_UID,
+                         _writer->generateLiteral(std::to_string(object.uid()),
+                                                  "^^" + IRI__XSD_INTEGER));
   }
 
   _writer->writeTriple(

diff --git a/src/osm/Node.cpp b/src/osm/Node.cpp
@@ -17,7 +17,6 @@
 // along with osm2rdf.  If not, see <https://www.gnu.org/licenses/>.
 
 #include "osm2rdf/osm/Node.h"
-
 #include "osm2rdf/osm/TagList.h"
 #include "osmium/osm/node.hpp"
 #include "osmium/osm/node_ref.hpp"
@@ -34,6 +33,7 @@ osm2rdf::osm::Node::Node(const osmium::Node& node) {
   _timestamp = node.timestamp().seconds_since_epoch();
   _changeset = node.changeset();
   _user = node.user();
+  _uid = node.uid();
   _version = node.version();
   _visible = node.visible();
   const auto& loc = node.location();
@@ -65,6 +65,11 @@ std::time_t osm2rdf::osm::Node::timestamp() const noexcept {
 // ____________________________________________________________________________
 std::string osm2rdf::osm::Node::user() const noexcept { return _user; }
 
+// ____________________________________________________________________________
+osm2rdf::osm::Node::id_t osm2rdf::osm::Node::uid() const noexcept {
+  return _uid;
+}
+
 // ____________________________________________________________________________
 osm2rdf::osm::generic::version_t osm2rdf::osm::Node::version() const noexcept {
   return _version;

diff --git a/src/osm/Relation.cpp b/src/osm/Relation.cpp
@@ -38,6 +38,7 @@ osm2rdf::osm::Relation::Relation(const osmium::Relation& relation) {
   _timestamp = relation.timestamp().seconds_since_epoch();
   _changeset = relation.changeset();
   _user = relation.user();
+  _uid = relation.uid();
   _version = relation.version();
   _visible = relation.visible();
   _tags = std::move(osm2rdf::osm::convertTagList(relation.tags()));
@@ -71,6 +72,9 @@ std::time_t osm2rdf::osm::Relation::timestamp() const noexcept {
 // ____________________________________________________________________________
 std::string osm2rdf::osm::Relation::user() const noexcept { return _user; }
 
+// ____________________________________________________________________________
+id_t osm2rdf::osm::Relation::uid() const noexcept { return _uid; }
+
 // ____________________________________________________________________________
 osm2rdf::osm::generic::version_t osm2rdf::osm::Relation::version()
     const noexcept {

diff --git a/src/osm/Way.cpp b/src/osm/Way.cpp
@@ -36,6 +36,7 @@ osm2rdf::osm::Way::Way(const osmium::Way& way) {
   _timestamp = way.timestamp().seconds_since_epoch();
   _changeset = way.changeset();
   _user = way.user();
+  _uid = way.uid();
   _version = way.version();
   _visible = way.visible();
   _tags = std::move(osm2rdf::osm::convertTagList(way.tags()));
@@ -95,6 +96,9 @@ std::time_t osm2rdf::osm::Way::timestamp() const noexcept { return _timestamp; }
 // ____________________________________________________________________________
 std::string osm2rdf::osm::Way::user() const noexcept { return _user; }
 
+// ____________________________________________________________________________
+id_t osm2rdf::osm::Way::uid() const noexcept { return _uid; }
+
 // ____________________________________________________________________________
 osm2rdf::osm::generic::version_t osm2rdf::osm::Way::version() const noexcept {
   return _version;

diff --git a/src/ttl/Writer.cpp b/src/ttl/Writer.cpp
@@ -76,6 +76,10 @@ osm2rdf::ttl::Writer<T>::Writer(const osm2rdf::config::Config& config,
        "https://www.openstreetmap.org/relation/"},
       {osm2rdf::ttl::constants::NAMESPACE__OSM_WAY,
        "https://www.openstreetmap.org/way/"},
+      {osm2rdf::ttl::constants::NAMESPACE__OSM_USER,
+       "https://www.openstreetmap.org/user/"},
+      {osm2rdf::ttl::constants::NAMESPACE__OSM_CHANGESET,
+       "https://www.openstreetmap.org/changeset/"},
       // ohm prefixes
       {osm2rdf::ttl::constants::NAMESPACE__OHM,
        "https://www.openhistoricalmap.org/"},
@@ -141,6 +145,8 @@ osm2rdf::ttl::Writer<T>::Writer(const osm2rdf::config::Config& config,
       generateIRI(osm2rdf::ttl::constants::NAMESPACE__OSM_META, "timestamp");
   osm2rdf::ttl::constants::IRI__OSMMETA_USER =
       generateIRI(osm2rdf::ttl::constants::NAMESPACE__OSM_META, "user");
+  osm2rdf::ttl::constants::IRI__OSMMETA_UID =
+      generateIRI(osm2rdf::ttl::constants::NAMESPACE__OSM_META, "uid");
   osm2rdf::ttl::constants::IRI__OSMMETA_VERSION =
       generateIRI(osm2rdf::ttl::constants::NAMESPACE__OSM_META, "version");
   osm2rdf::ttl::constants::IRI__OSMMETA_VISIBLE =
@@ -159,6 +165,10 @@ osm2rdf::ttl::Writer<T>::Writer(const osm2rdf::config::Config& config,
       osm2rdf::ttl::constants::NAMESPACE__OSM_WAY, "uniqueNodeCount");
   osm2rdf::ttl::constants::IRI__OSM_NODE =
       generateIRI(osm2rdf::ttl::constants::NAMESPACE__OSM, "node");
+  osm2rdf::ttl::constants::IRI__OSM_USER =
+      generateIRI(osm2rdf::ttl::constants::NAMESPACE__OSM, "user");
+  osm2rdf::ttl::constants::IRI__OSM_CHANGESET =
+      generateIRI(osm2rdf::ttl::constants::NAMESPACE__OSM, "changeset");
   osm2rdf::ttl::constants::IRI__OSM_RELATION =
       generateIRI(osm2rdf::ttl::constants::NAMESPACE__OSM, "relation");
   osm2rdf::ttl::constants::IRI__OSM_TAG =

diff --git a/tests/ttl/Writer.cpp b/tests/ttl/Writer.cpp
@@ -704,8 +704,8 @@ TEST(TTL_WriterTTL, writeStatisticJson) {
   statsBuffer << statsIFStream.rdbuf();
 
   ASSERT_THAT(statsBuffer.str(), ::testing::HasSubstr("\"blankNodes\": 3"));
-  ASSERT_THAT(statsBuffer.str(), ::testing::HasSubstr("\"header\": 20"));
-  ASSERT_THAT(statsBuffer.str(), ::testing::HasSubstr("\"lines\": 25"));
+  ASSERT_THAT(statsBuffer.str(), ::testing::HasSubstr("\"header\": 22"));
+  ASSERT_THAT(statsBuffer.str(), ::testing::HasSubstr("\"lines\": 27"));
   ASSERT_THAT(statsBuffer.str(), ::testing::HasSubstr("\"triples\": 5"));
 
   // Cleanup
@@ -766,8 +766,8 @@ TEST(TTL_WriterQLEVER, writeStatisticJson) {
   statsBuffer << statsIFStream.rdbuf();
 
   ASSERT_THAT(statsBuffer.str(), ::testing::HasSubstr("\"blankNodes\": 3"));
-  ASSERT_THAT(statsBuffer.str(), ::testing::HasSubstr("\"header\": 20"));
-  ASSERT_THAT(statsBuffer.str(), ::testing::HasSubstr("\"lines\": 25"));
+  ASSERT_THAT(statsBuffer.str(), ::testing::HasSubstr("\"header\": 22"));
+  ASSERT_THAT(statsBuffer.str(), ::testing::HasSubstr("\"lines\": 27"));
   ASSERT_THAT(statsBuffer.str(), ::testing::HasSubstr("\"triples\": 5"));
 
   // Cleanup