Repository: marmotta Updated Branches: refs/heads/develop 6be5dc757 -> 5491d5fd0
Ostrich: remove URI compression code, as it is currently not fully implemented Project: http://git-wip-us.apache.org/repos/asf/marmotta/repo Commit: http://git-wip-us.apache.org/repos/asf/marmotta/commit/5491d5fd Tree: http://git-wip-us.apache.org/repos/asf/marmotta/tree/5491d5fd Diff: http://git-wip-us.apache.org/repos/asf/marmotta/diff/5491d5fd Branch: refs/heads/develop Commit: 5491d5fd0aedfbde6cce00d7b41fe40b626b1431 Parents: 6be5dc7 Author: Sebastian Schaffert <[email protected]> Authored: Sun Nov 20 14:47:42 2016 +0100 Committer: Sebastian Schaffert <[email protected]> Committed: Sun Nov 20 14:47:42 2016 +0100 ---------------------------------------------------------------------- .../ostrich/backend/model/rdf_namespaces.cc | 26 ++++ .../ostrich/backend/model/rdf_namespaces.h | 120 +++++++++++++++++++ .../backend/persistence/base_persistence.cc | 28 ----- .../backend/persistence/base_persistence.h | 118 ------------------ .../backend/persistence/leveldb_persistence.cc | 1 - libraries/ostrich/backend/test/CMakeLists.txt | 2 +- libraries/ostrich/backend/test/NamespaceTest.cc | 65 ++++++++++ .../ostrich/backend/test/PersistenceTest.cc | 51 -------- 8 files changed, 212 insertions(+), 199 deletions(-) ---------------------------------------------------------------------- http://git-wip-us.apache.org/repos/asf/marmotta/blob/5491d5fd/libraries/ostrich/backend/model/rdf_namespaces.cc ---------------------------------------------------------------------- diff --git a/libraries/ostrich/backend/model/rdf_namespaces.cc b/libraries/ostrich/backend/model/rdf_namespaces.cc index 3f54017..ba54dc1 100644 --- a/libraries/ostrich/backend/model/rdf_namespaces.cc +++ b/libraries/ostrich/backend/model/rdf_namespaces.cc @@ -37,6 +37,32 @@ const std::map<std::string, std::string>& NamespacesByPrefix() { return kNamespacePrefixes; } +// Apply prefix substitution for well-known URIs to save disk space. +// Modifies the string passed as argument. +void EncodeWellknownURI(std::string* uri) { + for (auto& ns : NamespacesByPrefix()) { + if (uri->compare(0, ns.second.size(), ns.second) == 0) { + std::string tmp = ns.first; + tmp += uri->substr(ns.second.size()); + uri->swap(tmp); + return; + } + } +} + +// Unapply prefix substitution for well-known URIs. +// Modifies the string passed as argument. +void DecodeWellknownURI(std::string* uri) { + for (auto& ns : NamespacesByPrefix()) { + if (uri->compare(0, ns.first.size(), ns.first) == 0) { + std::string tmp = ns.second; + tmp += uri->substr(ns.first.size()); + uri->swap(tmp); + return; + } + } +} + } // namespace rdf } // namespace marmotta http://git-wip-us.apache.org/repos/asf/marmotta/blob/5491d5fd/libraries/ostrich/backend/model/rdf_namespaces.h ---------------------------------------------------------------------- diff --git a/libraries/ostrich/backend/model/rdf_namespaces.h b/libraries/ostrich/backend/model/rdf_namespaces.h index 793ac2f..126dc2b 100644 --- a/libraries/ostrich/backend/model/rdf_namespaces.h +++ b/libraries/ostrich/backend/model/rdf_namespaces.h @@ -21,6 +21,8 @@ #include <map> #include <string> +#include "model/rdf_model.h" + // Contains maps of well-known default namespaces. namespace marmotta { namespace rdf { @@ -29,6 +31,124 @@ namespace rdf { // namespace URI. const std::map<std::string, std::string>& NamespacesByPrefix(); +// Apply prefix substitution for well-known URIs to save disk space. +// Modifies the string passed as argument. +void EncodeWellknownURI(std::string* uri); + +// Apply prefix substitution for well-known URIs to save disk space. +// Replaces the uri string of the URI with the encoded one +inline void EncodeWellknownURI(proto::URI* value){ + EncodeWellknownURI(value->mutable_uri()); +} + +// Apply prefix substitution for well-known URIs to save disk space. +// Replaces the uri string of the type URI with the encoded one +inline void EncodeWellknownURI(proto::DatatypeLiteral* value) { + EncodeWellknownURI(value->mutable_datatype()); +} + +// Apply prefix substitution for well-known URIs to save disk space. +// Cases: +// - value is a URI: replace the uri string with the encoded one +// - otherwise: do nothing +inline void EncodeWellknownURI(proto::Resource* value) { + if (value->has_uri()) { + EncodeWellknownURI(value->mutable_uri()); + } +} + +// Apply prefix substitution for well-known URIs to save disk space. +// Cases: +// - value is a URI: replace the uri string with the encoded one +// - value is a DatatypeLiteral: replace type URI with encoded one +// - otherwise: do nothing +inline void EncodeWellknownURI(proto::Value* value) { + if (value->has_resource()) { + EncodeWellknownURI(value->mutable_resource()); + } else if (value->has_literal() && value->mutable_literal()->has_dataliteral()) { + EncodeWellknownURI(value->mutable_literal()->mutable_dataliteral()); + } +} + +// Apply prefix substitution for well-known URIs to save disk space. +// Performs prefix substitution for subject, predicate, object and context. +inline void EncodeWellknownURI(proto::Statement* stmt) { + if (stmt->has_subject()) { + EncodeWellknownURI(stmt->mutable_subject()); + } + if (stmt->has_predicate()) { + EncodeWellknownURI(stmt->mutable_predicate()); + } + if (stmt->has_object()) { + EncodeWellknownURI(stmt->mutable_object()); + } + if (stmt->has_context()) { + EncodeWellknownURI(stmt->mutable_context()); + } +} + +// Compatibility placeholder, does nothing for namespaces. +inline void EncodeWellknownURI(proto::Namespace* ns) {} + +// Unapply prefix substitution for well-known URIs. +// Modifies the string passed as argument. +void DecodeWellknownURI(std::string* uri); + +// Unapply prefix substitution for well-known URIs. +// Replaces the uri string of the URI with the decoded one +inline void DecodeWellknownURI(proto::URI* value){ + DecodeWellknownURI(value->mutable_uri()); +} + +// Unapply prefix substitution for well-known URIs. +// Replaces the uri string of the type URI with the decoded one +inline void DecodeWellknownURI(proto::DatatypeLiteral* value) { + DecodeWellknownURI(value->mutable_datatype()); +} + +// Unapply prefix substitution for well-known URIs. +// Cases: +// - value is a URI: replace the uri string with the decoded one +// - otherwise: do nothing +inline void DecodeWellknownURI(proto::Resource* value) { + if (value->has_uri()) { + DecodeWellknownURI(value->mutable_uri()); + } +} + +// Unapply prefix substitution for well-known URIs. +// Cases: +// - value is a URI: replace the uri string with the decoded one +// - value is a DatatypeLiteral: replace type URI with decoded one +// - otherwise: do nothing +inline void DecodeWellknownURI(proto::Value* value) { + if (value->has_resource()) { + DecodeWellknownURI(value->mutable_resource()); + } else if (value->has_literal() && value->mutable_literal()->has_dataliteral()) { + DecodeWellknownURI(value->mutable_literal()->mutable_dataliteral()); + } +} + +// Apply prefix substitution for well-known URIs to save disk space. +// Performs prefix substitution for subject, predicate, object and context. +inline void DecodeWellknownURI(proto::Statement* stmt) { + if (stmt->has_subject()) { + DecodeWellknownURI(stmt->mutable_subject()); + } + if (stmt->has_predicate()) { + DecodeWellknownURI(stmt->mutable_predicate()); + } + if (stmt->has_object()) { + DecodeWellknownURI(stmt->mutable_object()); + } + if (stmt->has_context()) { + DecodeWellknownURI(stmt->mutable_context()); + } +} + +// Compatibility placeholder, does nothing for namespaces. +inline void DecodeWellknownURI(proto::Namespace* ns) {} + } // namespace rdf } // namespace marmotta http://git-wip-us.apache.org/repos/asf/marmotta/blob/5491d5fd/libraries/ostrich/backend/persistence/base_persistence.cc ---------------------------------------------------------------------- diff --git a/libraries/ostrich/backend/persistence/base_persistence.cc b/libraries/ostrich/backend/persistence/base_persistence.cc index ce1727b..1358763 100644 --- a/libraries/ostrich/backend/persistence/base_persistence.cc +++ b/libraries/ostrich/backend/persistence/base_persistence.cc @@ -188,34 +188,6 @@ bool Matches(const Statement& pattern, const Statement& stmt) { } return !(pattern.has_object() && stmt.object() != pattern.object()); } - - -// Apply prefix substitution for well-known URIs to save disk space. -// Modifies the string passed as argument. -void EncodeWellknownURI(std::string* uri) { - for (auto& ns : rdf::NamespacesByPrefix()) { - if (uri->compare(0, ns.second.size(), ns.second) == 0) { - std::string tmp = ns.first; - tmp += uri->substr(ns.second.size()); - uri->swap(tmp); - return; - } - } -} - -// Unapply prefix substitution for well-known URIs. -// Modifies the string passed as argument. -void DecodeWellknownURI(std::string* uri) { - for (auto& ns : rdf::NamespacesByPrefix()) { - if (uri->compare(0, ns.first.size(), ns.first) == 0) { - std::string tmp = ns.second; - tmp += uri->substr(ns.first.size()); - uri->swap(tmp); - return; - } - } -} - } // namespace persistence } // namespace marmotta http://git-wip-us.apache.org/repos/asf/marmotta/blob/5491d5fd/libraries/ostrich/backend/persistence/base_persistence.h ---------------------------------------------------------------------- diff --git a/libraries/ostrich/backend/persistence/base_persistence.h b/libraries/ostrich/backend/persistence/base_persistence.h index f1cc494..d182518 100644 --- a/libraries/ostrich/backend/persistence/base_persistence.h +++ b/libraries/ostrich/backend/persistence/base_persistence.h @@ -27,123 +27,6 @@ namespace marmotta { namespace persistence { -// Apply prefix substitution for well-known URIs to save disk space. -// Modifies the string passed as argument. -void EncodeWellknownURI(std::string* uri); - -// Apply prefix substitution for well-known URIs to save disk space. -// Replaces the uri string of the URI with the encoded one -inline void EncodeWellknownURI(rdf::proto::URI* value){ - EncodeWellknownURI(value->mutable_uri()); -} - -// Apply prefix substitution for well-known URIs to save disk space. -// Replaces the uri string of the type URI with the encoded one -inline void EncodeWellknownURI(rdf::proto::DatatypeLiteral* value) { - EncodeWellknownURI(value->mutable_datatype()); -} - -// Apply prefix substitution for well-known URIs to save disk space. -// Cases: -// - value is a URI: replace the uri string with the encoded one -// - otherwise: do nothing -inline void EncodeWellknownURI(rdf::proto::Resource* value) { - if (value->has_uri()) { - EncodeWellknownURI(value->mutable_uri()); - } -} - -// Apply prefix substitution for well-known URIs to save disk space. -// Cases: -// - value is a URI: replace the uri string with the encoded one -// - value is a DatatypeLiteral: replace type URI with encoded one -// - otherwise: do nothing -inline void EncodeWellknownURI(rdf::proto::Value* value) { - if (value->has_resource()) { - EncodeWellknownURI(value->mutable_resource()); - } else if (value->has_literal() && value->mutable_literal()->has_dataliteral()) { - EncodeWellknownURI(value->mutable_literal()->mutable_dataliteral()); - } -} - -// Apply prefix substitution for well-known URIs to save disk space. -// Performs prefix substitution for subject, predicate, object and context. -inline void EncodeWellknownURI(rdf::proto::Statement* stmt) { - if (stmt->has_subject()) { - EncodeWellknownURI(stmt->mutable_subject()); - } - if (stmt->has_predicate()) { - EncodeWellknownURI(stmt->mutable_predicate()); - } - if (stmt->has_object()) { - EncodeWellknownURI(stmt->mutable_object()); - } - if (stmt->has_context()) { - EncodeWellknownURI(stmt->mutable_context()); - } -} - -// Compatibility placeholder, does nothing for namespaces. -inline void EncodeWellknownURI(rdf::proto::Namespace* ns) {} - -// Unapply prefix substitution for well-known URIs. -// Modifies the string passed as argument. -void DecodeWellknownURI(std::string* uri); - -// Unapply prefix substitution for well-known URIs. -// Replaces the uri string of the URI with the decoded one -inline void DecodeWellknownURI(rdf::proto::URI* value){ - DecodeWellknownURI(value->mutable_uri()); -} - -// Unapply prefix substitution for well-known URIs. -// Replaces the uri string of the type URI with the decoded one -inline void DecodeWellknownURI(rdf::proto::DatatypeLiteral* value) { - DecodeWellknownURI(value->mutable_datatype()); -} - -// Unapply prefix substitution for well-known URIs. -// Cases: -// - value is a URI: replace the uri string with the decoded one -// - otherwise: do nothing -inline void DecodeWellknownURI(rdf::proto::Resource* value) { - if (value->has_uri()) { - DecodeWellknownURI(value->mutable_uri()); - } -} - -// Unapply prefix substitution for well-known URIs. -// Cases: -// - value is a URI: replace the uri string with the decoded one -// - value is a DatatypeLiteral: replace type URI with decoded one -// - otherwise: do nothing -inline void DecodeWellknownURI(rdf::proto::Value* value) { - if (value->has_resource()) { - DecodeWellknownURI(value->mutable_resource()); - } else if (value->has_literal() && value->mutable_literal()->has_dataliteral()) { - DecodeWellknownURI(value->mutable_literal()->mutable_dataliteral()); - } -} - -// Apply prefix substitution for well-known URIs to save disk space. -// Performs prefix substitution for subject, predicate, object and context. -inline void DecodeWellknownURI(rdf::proto::Statement* stmt) { - if (stmt->has_subject()) { - DecodeWellknownURI(stmt->mutable_subject()); - } - if (stmt->has_predicate()) { - DecodeWellknownURI(stmt->mutable_predicate()); - } - if (stmt->has_object()) { - DecodeWellknownURI(stmt->mutable_object()); - } - if (stmt->has_context()) { - DecodeWellknownURI(stmt->mutable_context()); - } -} - -// Compatibility placeholder, does nothing for namespaces. -inline void DecodeWellknownURI(rdf::proto::Namespace* ns) {} // Length of key in bytes per field S, P, O and C. constexpr int kKeyLength = 16; @@ -307,7 +190,6 @@ class DBIterator : public util::CloseableIterator<T> { const T& next() override { // Parse current position, then iterate to next position for next call. proto.ParseFromString(it->value().ToString()); - DecodeWellknownURI(&proto); it->Next(); return proto; }; http://git-wip-us.apache.org/repos/asf/marmotta/blob/5491d5fd/libraries/ostrich/backend/persistence/leveldb_persistence.cc ---------------------------------------------------------------------- diff --git a/libraries/ostrich/backend/persistence/leveldb_persistence.cc b/libraries/ostrich/backend/persistence/leveldb_persistence.cc index a14542b..c84c57c 100644 --- a/libraries/ostrich/backend/persistence/leveldb_persistence.cc +++ b/libraries/ostrich/backend/persistence/leveldb_persistence.cc @@ -498,7 +498,6 @@ int64_t LevelDBPersistence::RemoveStatements( int64_t count = 0; - std::string bufs, bufp, bufo, bufc; GetStatements(pattern, [&](const Statement stmt) -> bool { Key key(stmt); http://git-wip-us.apache.org/repos/asf/marmotta/blob/5491d5fd/libraries/ostrich/backend/test/CMakeLists.txt ---------------------------------------------------------------------- diff --git a/libraries/ostrich/backend/test/CMakeLists.txt b/libraries/ostrich/backend/test/CMakeLists.txt index a24e395..8aaf419 100644 --- a/libraries/ostrich/backend/test/CMakeLists.txt +++ b/libraries/ostrich/backend/test/CMakeLists.txt @@ -6,7 +6,7 @@ include_directories(${RAPTOR_INCLUDE_DIR}/raptor2) add_library(gtest STATIC gtest/gtest.h gmock/gmock.h gmock-gtest-all.cc) -add_executable(model_tests StatementTest.cc main.cc) +add_executable(model_tests NamespaceTest.cc StatementTest.cc main.cc) target_link_libraries(model_tests gtest marmotta_model ${GLOG_LIBRARY}) add_executable(sparql_tests SparqlTest.cc main.cc) http://git-wip-us.apache.org/repos/asf/marmotta/blob/5491d5fd/libraries/ostrich/backend/test/NamespaceTest.cc ---------------------------------------------------------------------- diff --git a/libraries/ostrich/backend/test/NamespaceTest.cc b/libraries/ostrich/backend/test/NamespaceTest.cc new file mode 100644 index 0000000..9fc37e3 --- /dev/null +++ b/libraries/ostrich/backend/test/NamespaceTest.cc @@ -0,0 +1,65 @@ +// +// Created by wastl on 18.04.15. +// + +#include "gtest/gtest.h" +#include "model/rdf_model.h" +#include "model/rdf_operators.h" +#include "model/rdf_namespaces.h" + +namespace marmotta { +namespace rdf { + +TEST(NamespaceTest, EncodeURI) { + std::string uri1 = "http://www.w3.org/2002/07/owl#sameAs";; + std::string uri2 = "http://marmotta.apache.org/test/uri1";; + + EncodeWellknownURI(&uri1); + EXPECT_EQ("owl:sameAs", uri1); + + EncodeWellknownURI(&uri2); + EXPECT_EQ("http://marmotta.apache.org/test/uri1";, uri2); +} + +TEST(NamespaceTest, EncodeURIProto) { + rdf::URI uri1 = "http://www.w3.org/2002/07/owl#sameAs";; + rdf::URI uri2 = "http://marmotta.apache.org/test/uri1";; + + rdf::proto::URI msg1 = uri1.getMessage(); + rdf::proto::URI msg2 = uri2.getMessage(); + + EncodeWellknownURI(&msg1); + EXPECT_EQ("owl:sameAs", msg1.uri()); + + EncodeWellknownURI(&msg2); + EXPECT_EQ("http://marmotta.apache.org/test/uri1";, msg2.uri()); +} + + +TEST(NamespaceTest, DecodeURI) { + std::string uri1 = "owl:sameAs"; + std::string uri2 = "http://marmotta.apache.org/test/uri1";; + + DecodeWellknownURI(&uri1); + EXPECT_EQ("http://www.w3.org/2002/07/owl#sameAs";, uri1); + + DecodeWellknownURI(&uri2); + EXPECT_EQ("http://marmotta.apache.org/test/uri1";, uri2); +} + +TEST(NamespaceTest, DecodeURIProto) { + rdf::URI uri1 = "owl:sameAs"; + rdf::URI uri2 = "http://marmotta.apache.org/test/uri1";; + + rdf::proto::URI msg1 = uri1.getMessage(); + rdf::proto::URI msg2 = uri2.getMessage(); + + DecodeWellknownURI(&msg1); + EXPECT_EQ("http://www.w3.org/2002/07/owl#sameAs";, msg1.uri()); + + DecodeWellknownURI(&msg2); + EXPECT_EQ("http://marmotta.apache.org/test/uri1";, msg2.uri()); +} + +} // namespace rdf +} // namespace marmotta \ No newline at end of file http://git-wip-us.apache.org/repos/asf/marmotta/blob/5491d5fd/libraries/ostrich/backend/test/PersistenceTest.cc ---------------------------------------------------------------------- diff --git a/libraries/ostrich/backend/test/PersistenceTest.cc b/libraries/ostrich/backend/test/PersistenceTest.cc index 7f37d92..5626adb 100644 --- a/libraries/ostrich/backend/test/PersistenceTest.cc +++ b/libraries/ostrich/backend/test/PersistenceTest.cc @@ -60,57 +60,6 @@ TEST(KeyTest, BoundsDiffer) { } } -TEST(URITest, EncodeURI) { - std::string uri1 = "http://www.w3.org/2002/07/owl#sameAs";; - std::string uri2 = "http://marmotta.apache.org/test/uri1";; - - EncodeWellknownURI(&uri1); - EXPECT_EQ("owl:sameAs", uri1); - - EncodeWellknownURI(&uri2); - EXPECT_EQ("http://marmotta.apache.org/test/uri1";, uri2); -} - -TEST(URITest, EncodeURIProto) { - rdf::URI uri1 = "http://www.w3.org/2002/07/owl#sameAs";; - rdf::URI uri2 = "http://marmotta.apache.org/test/uri1";; - - rdf::proto::URI msg1 = uri1.getMessage(); - rdf::proto::URI msg2 = uri2.getMessage(); - - EncodeWellknownURI(&msg1); - EXPECT_EQ("owl:sameAs", msg1.uri()); - - EncodeWellknownURI(&msg2); - EXPECT_EQ("http://marmotta.apache.org/test/uri1";, msg2.uri()); -} - - -TEST(URITest, DecodeURI) { - std::string uri1 = "owl:sameAs"; - std::string uri2 = "http://marmotta.apache.org/test/uri1";; - - DecodeWellknownURI(&uri1); - EXPECT_EQ("http://www.w3.org/2002/07/owl#sameAs";, uri1); - - DecodeWellknownURI(&uri2); - EXPECT_EQ("http://marmotta.apache.org/test/uri1";, uri2); -} - -TEST(URITest, DecodeURIProto) { - rdf::URI uri1 = "owl:sameAs"; - rdf::URI uri2 = "http://marmotta.apache.org/test/uri1";; - - rdf::proto::URI msg1 = uri1.getMessage(); - rdf::proto::URI msg2 = uri2.getMessage(); - - DecodeWellknownURI(&msg1); - EXPECT_EQ("http://www.w3.org/2002/07/owl#sameAs";, msg1.uri()); - - DecodeWellknownURI(&msg2); - EXPECT_EQ("http://marmotta.apache.org/test/uri1";, msg2.uri()); -} - } // namespace test } // namespace persistence
