Feature/parse xml

CMakeLists.txt

@@ -23,6 +23,7 @@ find_package(highway REQUIRED)
 find_package(dice-hash REQUIRED)
 find_package(dice-sparse-map REQUIRED)
 find_package(dice-template-library REQUIRED)
+find_package(libxml2 REQUIRED)
 
 configure_file(${CMAKE_CURRENT_SOURCE_DIR}/cmake/version.hpp.in ${CMAKE_CURRENT_SOURCE_DIR}/src/rdf4cpp/version.hpp)
 
@@ -130,6 +131,7 @@ add_library(rdf4cpp
         src/rdf4cpp/namespaces/RDF.cpp
         src/rdf4cpp/parser/IStreamQuadIterator.cpp
         src/rdf4cpp/parser/RDFFileParser.cpp
+        src/rdf4cpp/parser/XMLParser.cpp
         src/rdf4cpp/query/QuadPattern.cpp
         src/rdf4cpp/query/Solution.cpp
         src/rdf4cpp/query/TriplePattern.cpp
@@ -177,6 +179,7 @@ target_link_libraries(rdf4cpp
         OpenSSL::Crypto
         uni-algo::uni-algo
         highway::highway
+        LibXml2::LibXml2
         )
 
 set_target_properties(rdf4cpp PROPERTIES

conanfile.py

@@ -39,6 +39,8 @@ def requirements(self):
         self.requires("dice-hash/0.4.11", transitive_headers=True)
         self.requires("dice-sparse-map/0.2.9", transitive_headers=True)
         self.requires("dice-template-library/1.13.0", transitive_headers=True)
+        self.requires("libxml2/2.14.5")
+        self.requires("zlib/1.3.1", force=True)
 
         if self.options.with_test_deps:
             self.test_requires("doctest/2.4.11")

src/rdf4cpp/IRIFactory.cpp

@@ -292,5 +292,9 @@ IRIFactoryError IRIFactory::set_base(std::string_view b) noexcept {
     base_parts_cache = IRIView{base}.all_parts();
     return IRIFactoryError::Ok;
 }
+void IRIFactory::set_base_unchecked(std::string_view b) noexcept {
+    base = b;
+    base_parts_cache = IRIView{base}.all_parts();
+}
 
 }  // namespace rdf4cpp

src/rdf4cpp/IRIFactory.hpp

@@ -105,6 +105,12 @@ struct IRIFactory {
      * @return
      */
     [[nodiscard]] IRIFactoryError set_base(std::string_view b) noexcept;
+    /**
+     * Changes the base IRI. Skips validating the new base IRI.
+     * @param b
+     * @return
+     */
+    void set_base_unchecked(std::string_view b) noexcept;
 
     /**
      * validates the given IRI and creates it in the given node storage, if valid.

src/rdf4cpp/parser/XMLParser.hpp

@@ -0,0 +1,85 @@
+#ifndef RDF4CPP_XMLPARSER_HPP
+#define RDF4CPP_XMLPARSER_HPP
+
+#include <iterator>
+#include <memory>
+
+#include <rdf4cpp/Expected.hpp>
+
+#include <rdf4cpp/Quad.hpp>
+
+#include <rdf4cpp/parser/ParsingError.hpp>
+#include <rdf4cpp/parser/ParsingFlags.hpp>
+#include <rdf4cpp/parser/ParsingState.hpp>
+#include <rdf4cpp/IRIFactory.hpp>
+
+namespace rdf4cpp::parser {
+    struct XMLQuadIterator {
+        using flags_type = ParsingFlags;
+        using state_type = ParsingState;
+        using ok_type = Quad;
+        using error_type = ParsingError;
+
+        using value_type = nonstd::expected<ok_type, error_type>;
+        using reference = value_type const &;
+        using pointer = value_type const *;
+        using difference_type = std::ptrdiff_t;
+        using iterator_category = std::input_iterator_tag;
+        using istream_type = std::istream;
+
+        /**
+         * Identical semantics to fread.
+         * Uses stream to read at most count elements of size element_size into buffer.
+         *
+         * @param buffer pointer to buffer with at least count elements of size elem_size
+         * @param elem_size sizeof each element
+         * @param count number of elements to read
+         * @param stream pointer to any object.
+         * @return number of elements read
+         */
+        using ReadFunc = size_t (*)(void *buffer, size_t elem_size, size_t count, void *stream);
+
+        /**
+         * Identical semantics to ferror.
+         *
+         * @param stream pointer to any object
+         * @return nonzero value if there is an error in stream, zero value otherwise
+         */
+        using ErrorFunc = int (*)(void *stream);
+
+        /**
+         * Identical semantics to feof.
+         *
+         *
+         * @param stream pointer to any object
+         * @return nonzero value if there is an error in stream, zero value otherwise
+         */
+        using EOFFunc = int (*)(void *stream);
+
+    private:
+        struct Impl;
+
+        std::unique_ptr<Impl> impl_;
+        std::optional<nonstd::expected<ok_type, error_type>> cur_;
+
+    public:
+        XMLQuadIterator(void *stream, ReadFunc read, ErrorFunc error, EOFFunc eof);
+        explicit XMLQuadIterator(std::istream& stream);
+
+        XMLQuadIterator(XMLQuadIterator&&) noexcept = delete;
+        XMLQuadIterator& operator=(XMLQuadIterator&&) noexcept = delete;
+
+        XMLQuadIterator(XMLQuadIterator const &) = delete;
+        XMLQuadIterator& operator=(XMLQuadIterator const &) = delete;
+
+        ~XMLQuadIterator() noexcept;
+
+        reference operator*() const noexcept;
+        pointer operator->() const noexcept;
+        XMLQuadIterator &operator++();
+
+        bool operator==(std::default_sentinel_t) const noexcept;
+    };
+}
+
+#endif  //RDF4CPP_XMLPARSER_H

tests/CMakeLists.txt

@@ -354,6 +354,13 @@ target_link_libraries(tests_Serialization
 )
 add_test(NAME tests_Serialization COMMAND tests_Serialization)
 
+add_executable(tests_XMLParser parser/tests_XMLParser.cpp)
+target_link_libraries(tests_XMLParser
+        doctest::doctest
+        rdf4cpp
+)
+add_test(NAME tests_XMLParser COMMAND tests_XMLParser)
+
 if(NOT EXISTS "${CMAKE_CURRENT_BINARY_DIR}/test_swdf/swdf.nt")
         file(DOWNLOAD "https://files.tentris.dev/swdf.zip" "${CMAKE_CURRENT_BINARY_DIR}/test_swdf/swdf.zip")
         execute_process(COMMAND unzip "${CMAKE_CURRENT_BINARY_DIR}/test_swdf/swdf.zip" -d "${CMAKE_CURRENT_BINARY_DIR}/test_swdf")