Add option to avoid blank nodes

include/osm2rdf/config/Config.h

@@ -85,6 +85,8 @@ struct Config {
   std::string iriPrefixForUntaggedNodes =
       osm2rdf::ttl::constants::IRI_PREFIX__OSM_NODE_UNTAGGED;
 
+  bool noBlankNodes = false;
+
   int numThreads = std::thread::hardware_concurrency();
 
   // Default settings for data

include/osm2rdf/config/Constants.h

@@ -382,6 +382,13 @@ const static inline std::string WKT_PRECISION_OPTION_LONG = "wkt-precision";
 const static inline std::string WKT_PRECISION_OPTION_HELP =
     "Precision (number of decimal digits) for WKT coordinates";
 
+const static inline std::string BLANK_NODES_INFO =
+    "Blank nodes are masked";
+const static inline std::string BLANK_NODES_OPTION_SHORT = "";
+const static inline std::string BLANK_NODES_OPTION_LONG = "no-blank-nodes";
+const static inline std::string BLANK_NODES_OPTION_HELP =
+    "Avoid blank nodes by using a unique identifier for each member";
+
 const static inline std::string WRITE_RDF_STATISTICS_INFO =
     "Storing RDF statistics as .stats.json";
 const static inline std::string WRITE_RDF_STATISTICS_OPTION_SHORT = "";

include/osm2rdf/ttl/Constants.h

@@ -49,6 +49,7 @@ const static inline std::string NAMESPACE__OSM2RDF_TAG = "osm2rdfkey";
 const static inline std::string NAMESPACE__RDF = "rdf";
 const static inline std::string NAMESPACE__WIKIDATA_ENTITY = "wd";
 const static inline std::string NAMESPACE__XML_SCHEMA = "xsd";
+const static inline std::string NAMESPACE__GENID = "genid";
 
 // IRI Prefixes
 const static inline std::string IRI_PREFIX__OSM_NODE_TAGGED = "https://www.openstreetmap.org/node/";

include/osm2rdf/ttl/Writer.h

@@ -168,6 +168,26 @@ class Writer {
   // generateBlankNode creates a new unique identifier for a blank node.
   std::string generateBlankNode();
 
+  // generateSkolem creates a unique identifier for an osm object member.
+  std::string generateSkolem(const std::string& id);
+
+  // generateSkolemForRelationMember creates a unique identifier for a member of
+  // a relation by combining the id of the relation and member with a letter
+  // indicating the object type ('r', 'w', or 'n'). The relative position of the
+  // object in the relation is appended behind 'p'. Example: "r1234w5678p3"
+  std::string generateSkolemForRelationMember(const uint64_t& relationId,
+                                              const uint64_t& memberId,
+                                              const std::string& memberType,
+                                              const size_t& relPos);
+
+  // generateSkolemForWayMember creates a unique identifier for a member of
+  // a way by combining the id of the way and member with a letter
+  // indicating the object type ('r', 'w', or 'n'). The relative position of the node in the way is
+  // appended behind 'p'. Example: "w1234n5678p0"
+  std::string generateSkolemForWayMember(const uint64_t& wayId,
+                                         const uint64_t& nodeId,
+                                         const size_t& relPos);
+
   // Creates a IRI from given prefix p and string value v.
   // Assumes that both p and v are "safe", that is, they can be used
   // directly in the TTL

src/config/Config.cpp

@@ -175,6 +175,12 @@ 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;
 
+  if (noBlankNodes) {
+    oss << "\n"
+        << prefix
+        << osm2rdf::config::constants::BLANK_NODES_INFO;
+  }
+
   if (!storeLocations.empty()) {
     oss << "\n"
         << prefix << osm2rdf::config::constants::STORE_LOCATIONS_INFO << " "
@@ -408,6 +414,12 @@ void osm2rdf::config::Config::fromArgs(int argc, char** argv) {
           osm2rdf::config::constants::WKT_PRECISION_OPTION_LONG,
           osm2rdf::config::constants::WKT_PRECISION_OPTION_HELP, wktPrecision);
 
+  auto noBlankNodesOp =
+      parser.add<popl::Switch, popl::Attribute::advanced>(
+          osm2rdf::config::constants::BLANK_NODES_OPTION_SHORT,
+          osm2rdf::config::constants::BLANK_NODES_OPTION_LONG,
+          osm2rdf::config::constants::BLANK_NODES_OPTION_HELP);
+
   auto writeRDFStatisticsOp =
       parser.add<popl::Switch, popl::Attribute::advanced>(
           osm2rdf::config::constants::WRITE_RDF_STATISTICS_OPTION_SHORT,
@@ -535,6 +547,8 @@ void osm2rdf::config::Config::fromArgs(int argc, char** argv) {
 
     addSpatialRelsForUntaggedNodes = untaggedNodesSpatialRelsOp->is_set();
 
+    noBlankNodes = noBlankNodesOp->is_set();
+
     addUntaggedNodes = !noUntaggedNodesOp->is_set();
     if (iriPrefixForUntaggedNodesOp->is_set() &&
         iriPrefixForUntaggedNodesOp->value().size() > 0) {

src/osm/FactHandler.cpp

@@ -277,16 +277,22 @@ void osm2rdf::osm::FactHandler<W>::relation(
 
       std::string role = member.role();
       if (role.empty()) role = "member";
-      const std::string& blankNode = _writer->generateBlankNode();
+
+      const std::string memberIRI = _config.noBlankNodes
+                          ? _writer->generateSkolemForRelationMember(
+                              relation.id(), member.positive_ref(),
+                              type, inRelPos)
+                          : _writer->generateBlankNode();
+
       _writer->writeTriple(
           subj, _writer->generateIRIUnsafe(NAMESPACE__OSM_RELATION, "member"),
-          blankNode);
+          memberIRI);
 
-      _writer->writeTriple(blankNode, IRI__OSMREL__MEMBER_ID,
+      _writer->writeTriple(memberIRI, IRI__OSMREL__MEMBER_ID,
                            _writer->generateIRI(type, member.positive_ref()));
-      _writer->writeTriple(blankNode, IRI__OSMREL__MEMBER_ROLE,
+      _writer->writeTriple(memberIRI, IRI__OSMREL__MEMBER_ROLE,
                            _writer->generateLiteral(role));
-      _writer->writeLiteralTripleUnsafe(blankNode, IRI__OSMREL__MEMBER_POS,
+      _writer->writeLiteralTripleUnsafe(memberIRI, IRI__OSMREL__MEMBER_POS,
                                         std::to_string(inRelPos++),
                                         _iriXSDInteger);
     }
@@ -348,11 +354,15 @@ void osm2rdf::osm::FactHandler<W>::way(const osm2rdf::osm::Way& way) {
 
   if (_config.addMemberTriples && way.nodes().size()) {
     size_t wayOrder = 0;
-    std::string lastBlankNode;
+    std::string lastMemberIRI;
     auto lastNode = way.nodes().front();
     for (const auto& node : way.nodes()) {
-      const std::string& blankNode = _writer->generateBlankNode();
-      _writer->writeTriple(subj, IRI__OSMWAY__NODE, blankNode);
+      const std::string memberIRI = _config.noBlankNodes
+                                ? _writer->generateSkolemForWayMember(
+                                    way.id(), node.positive_ref(), wayOrder)
+                                : _writer->generateBlankNode();
+
+      _writer->writeTriple(subj, IRI__OSMWAY__NODE, memberIRI);
 
       std::string nodeNamespace;
       if (_config.iriPrefixForUntaggedNodes ==
@@ -365,17 +375,17 @@ void osm2rdf::osm::FactHandler<W>::way(const osm2rdf::osm::Way& way) {
       }
 
       _writer->writeTriple(
-          blankNode, osm2rdf::ttl::constants::IRI__OSMWAY__MEMBER_ID,
+          memberIRI, osm2rdf::ttl::constants::IRI__OSMWAY__MEMBER_ID,
           _writer->generateIRI(nodeNamespace, node.positive_ref()));
 
       _writer->writeLiteralTripleUnsafe(
-          blankNode, osm2rdf::ttl::constants::IRI__OSMWAY__MEMBER_POS,
+          memberIRI, osm2rdf::ttl::constants::IRI__OSMWAY__MEMBER_POS,
           std::to_string(wayOrder++), _iriXSDInteger);
 
-      if (_config.addWayNodeSpatialMetadata && !lastBlankNode.empty() &&
+      if (_config.addWayNodeSpatialMetadata && !lastMemberIRI.empty() &&
           node.location().valid() && lastNode.location().valid()) {
         _writer->writeTriple(
-            lastBlankNode, IRI__OSMWAY__NEXT_NODE,
+            lastMemberIRI, IRI__OSMWAY__NEXT_NODE,
             _writer->generateIRI(nodeNamespace, node.positive_ref()));
         // Haversine distance
         const double distanceLat = (node.location().lat_without_check() -
@@ -393,10 +403,10 @@ void osm2rdf::osm::FactHandler<W>::way(const osm2rdf::osm::Way& way) {
                                 osm2rdf::osm::constants::METERS_IN_KM * 2 *
                                 asin(sqrt(haversine));
         _writer->writeLiteralTripleUnsafe(
-            lastBlankNode, IRI__OSMWAY__NEXT_NODE_DISTANCE,
+            lastMemberIRI, IRI__OSMWAY__NEXT_NODE_DISTANCE,
             std::to_string(distance), "^^" + IRI__XSD__DECIMAL);
       }
-      lastBlankNode = blankNode;
+      lastMemberIRI = memberIRI;
       lastNode = node;
     }
   }

src/ttl/Writer.cpp

@@ -61,6 +61,8 @@ osm2rdf::ttl::Writer<T>::Writer(const osm2rdf::config::Config& config,
        "https://osm2rdf.cs.uni-freiburg.de/rdf/key#"},
       {osm2rdf::ttl::constants::NAMESPACE__OSM2RDF_META,
        "https://osm2rdf.cs.uni-freiburg.de/rdf/meta#"},
+      {osm2rdf::ttl::constants::NAMESPACE__GENID,
+       "http://osm2rdf.cs.uni-freiburg.de/.well-known/genid/"},
       // https://wiki.openstreetmap.org/wiki/Sophox#How_OSM_data_is_stored
       // https://github.com/Sophox/sophox/blob/master/osm2rdf/osmutils.py#L35-L39
       // osm prefixes
@@ -388,7 +390,8 @@ void osm2rdf::ttl::Writer<T>::writeMetadata() {
   writeOptionTriple(
       osm2rdf::config::constants::UNTAGGED_NODES_SPATIAL_RELS_OPTION_LONG,
       generateBooleanLiteral(_config.addSpatialRelsForUntaggedNodes));
-
+  writeOptionTriple(osm2rdf::config::constants::BLANK_NODES_OPTION_LONG,
+                    generateBooleanLiteral(_config.noBlankNodes));
   writeOptionTriple(osm2rdf::config::constants::NO_UNTAGGED_NODES_OPTION_LONG,
                     generateBooleanLiteral(!_config.addUntaggedNodes));
   writeOptionTriple(osm2rdf::config::constants::NO_UNTAGGED_WAYS_OPTION_LONG,
@@ -426,6 +429,39 @@ std::string osm2rdf::ttl::Writer<T>::generateBlankNode() {
          std::to_string(_blankNodeCount[threadId]++);
 }
 
+// ____________________________________________________________________________
+template <typename T>
+std::string osm2rdf::ttl::Writer<T>::generateSkolem(const std::string& id) {
+  return generateIRIUnsafe(osm2rdf::ttl::constants::NAMESPACE__GENID, id);
+}
+
+// ____________________________________________________________________________
+template <typename T>
+std::string osm2rdf::ttl::Writer<T>::generateSkolemForRelationMember(
+    const uint64_t& relationId,
+    const uint64_t& memberId,
+    const std::string& memberType,
+    const size_t& relPos) {
+  const std::string skolemId = "r" + std::to_string(relationId) +
+                               //Extract the first relevant char to determine
+                               //the type of the osm/ohm object
+                               memberType.at(3) + std::to_string(memberId) +
+                               "p" + std::to_string(relPos);
+  return generateSkolem(skolemId);
+}
+
+// ____________________________________________________________________________
+template <typename T>
+std::string osm2rdf::ttl::Writer<T>::generateSkolemForWayMember(
+    const uint64_t& wayId,
+    const uint64_t& nodeId,
+    const size_t& relPos) {
+  const std::string skolemId = "w" + std::to_string(wayId) +
+                               "n" + std::to_string(nodeId) +
+                               "p" + std::to_string(relPos);
+  return generateSkolem(skolemId);
+}
+
 // ____________________________________________________________________________
 template <typename T>
 void osm2rdf::ttl::Writer<T>::writeIRI(std::string_view p, uint64_t v,

tests/config/Config.cpp

@@ -51,6 +51,8 @@ void assertDefaultConfig(const osm2rdf::config::Config& config) {
 
   ASSERT_FALSE(config.writeRDFStatistics);
 
+  ASSERT_FALSE(config.noBlankNodes);
+
   ASSERT_EQ(0, config.simplifyGeometries);
   ASSERT_EQ(0, config.simplifyWKT);
   ASSERT_EQ(5, config.wktDeviation);
@@ -658,6 +660,21 @@ TEST(CONFIG_Config, fromArgsSimplifyWKTLong) {
   ASSERT_EQ(25, config.simplifyWKT);
 }
 
+// ____________________________________________________________________________
+TEST(CONFIG_Config, fromArgsNoBlankNodesLong) {
+  osm2rdf::config::Config config;
+  assertDefaultConfig(config);
+  osm2rdf::util::CacheFile cf("/tmp/dummyInput");
+
+  const auto arg = "--" + osm2rdf::config::constants::BLANK_NODES_OPTION_LONG;
+  const int argc = 3;
+  char* argv[argc] = {const_cast<char*>(""), const_cast<char*>(arg.c_str()),
+                      const_cast<char*>("/tmp/dummyInput")};
+  config.fromArgs(argc, argv);
+  ASSERT_EQ("", config.output.string());
+  ASSERT_TRUE(config.noBlankNodes);
+}
+
 // ____________________________________________________________________________
 TEST(CONFIG_Config, fromArgsSimplifyWKTDeviationLong) {
   osm2rdf::config::Config config;
@@ -928,6 +945,17 @@ TEST(CONFIG_Config, getInfoSimplifyWKT) {
       res, ::testing::HasSubstr(osm2rdf::config::constants::SIMPLIFY_WKT_INFO));
 }
 
+// ____________________________________________________________________________
+TEST(CONFIG_Config, getInfoNoBlankNodes) {
+  osm2rdf::config::Config config;
+  assertDefaultConfig(config);
+  config.noBlankNodes = true;
+
+  const std::string res = config.getInfo("");
+  ASSERT_THAT(
+      res, ::testing::HasSubstr(osm2rdf::config::constants::BLANK_NODES_INFO));
+}
+
 // ____________________________________________________________________________
 TEST(CONFIG_Config, getInfoSkipWikiLinks) {
   osm2rdf::config::Config config;

tests/ttl/Writer.cpp

@@ -252,6 +252,60 @@ TEST(TTL_WriterQLEVER, generateBlankNode) {
   }
 }
 
+// ____________________________________________________________________________
+TEST(TTL_WriterQLEVER, generateSkolem) {
+  osm2rdf::config::Config config;
+  osm2rdf::ttl::Writer<osm2rdf::ttl::format::QLEVER> w{config, nullptr};
+  {
+    const std::string res = w.generateSkolem("1");
+    ASSERT_STREQ("genid:1", res.c_str());
+  }
+  {
+    const std::string res = w.generateSkolem("2");
+    ASSERT_STREQ("genid:2", res.c_str());
+  }
+  {
+    const std::string res = w.generateSkolem("3");
+    ASSERT_STREQ("genid:3", res.c_str());
+  }
+}
+
+// ____________________________________________________________________________
+TEST(TTL_WriterQLEVER, generateSkolemForRelationMember) {
+  osm2rdf::config::Config config;
+  osm2rdf::ttl::Writer<osm2rdf::ttl::format::QLEVER> w{config, nullptr};
+  {
+    const std::string res = w.generateSkolemForRelationMember(1, 2, "osmnode", 3);
+    ASSERT_STREQ("genid:r1n2p3", res.c_str());
+  }
+  {
+    const std::string res = w.generateSkolemForRelationMember(1, 2, "osmway", 3);
+    ASSERT_STREQ("genid:r1w2p3", res.c_str());
+  }
+  {
+    const std::string res = w.generateSkolemForRelationMember(1, 2, "ohmnode", 3);
+    ASSERT_STREQ("genid:r1n2p3", res.c_str());
+  }
+}
+
+// ____________________________________________________________________________
+TEST(TTL_WriterQLEVER, generateSkolemForWayMember) {
+  osm2rdf::config::Config config;
+  osm2rdf::ttl::Writer<osm2rdf::ttl::format::QLEVER> w{config, nullptr};
+  {
+    const std::string res = w.generateSkolemForWayMember(1, 1, 1);
+    ASSERT_STREQ("genid:w1n1p1", res.c_str());
+  }
+  {
+    const std::string res = w.generateSkolemForWayMember(2, 3, 4);
+    ASSERT_STREQ("genid:w2n3p4", res.c_str());
+  }
+  {
+    const std::string res = w.generateSkolemForWayMember(3, 5, 7);
+    ASSERT_STREQ("genid:w3n5p7", res.c_str());
+  }
+}
+
 // ____________________________________________________________________________
 TEST(TTL_WriterNT, generateIRI_ID) {
   osm2rdf::config::Config config;