Add OBF reader and writer

.github/workflows/ci.yaml

@@ -29,6 +29,8 @@ jobs:
       - name: Configure CMake
         run: >
           CC=clang cmake -B ${{env.build_dir}}
+          -DCOWL_READERS='functional;obf'
+          -DCOWL_WRITERS='functional;obf'
           -DCOWL_LTO=OFF
           -DCOWL_CLANG_TIDY=ON
           -DCMAKE_COMPILE_WARNING_AS_ERROR=ON
@@ -72,6 +74,8 @@ jobs:
           -DULIB_SANITIZERS=ON
           -DULIB_LIBRARY_TYPE=${{matrix.lib_type}}
           -DCOWL_LIBRARY_TYPE=${{matrix.lib_type}}
+          -DCOWL_READERS='functional;obf'
+          -DCOWL_WRITERS='functional;obf'
           -DCOWL_ENTITY_IDS=ON
           -DCOWL_CPP_TESTS=ON
           -DCMAKE_COMPILE_WARNING_AS_ERROR=ON

include/cowl_reader.h

@@ -62,6 +62,19 @@ CowlReader cowl_reader_functional(void);
 
 #endif // COWL_READER_FUNCTIONAL
 
+#ifdef COWL_READER_OBF
+
+/**
+ * Returns the OBF syntax reader.
+ *
+ * @return OBF syntax reader.
+ */
+COWL_API
+COWL_CONST
+CowlReader cowl_reader_obf(void);
+
+#endif // COWL_READER_OBF
+
 /// @}
 
 COWL_END_DECLS

include/cowl_writer.h

@@ -124,6 +124,19 @@ CowlWriter cowl_writer_functional(void);
 
 #endif // COWL_WRITER_FUNCTIONAL
 
+#ifdef COWL_WRITER_OBF
+
+/**
+ * Returns the OBF syntax writer.
+ *
+ * @return OBF syntax writer.
+ */
+COWL_API
+COWL_CONST
+CowlWriter cowl_writer_obf(void);
+
+#endif // COWL_WRITER_OBF
+
 /**
  * Checks whether the writer supports stream writing.
  *

src/formats/obf/cowl_obf_constants.h

@@ -0,0 +1,42 @@
+/**
+ * @author Valerio Di Ceglie
+ *
+ * @copyright Copyright (c) 2024 SisInf Lab, Polytechnic University of Bari
+ * @copyright <http://swot.sisinflab.poliba.it>
+ * @copyright SPDX-License-Identifier: EPL-2.0
+ *
+ * @file
+ */
+
+#ifndef COWL_OBF_CONSTANTS_H
+#define COWL_OBF_CONSTANTS_H
+
+typedef enum CowlOBFLiteralType {
+    COWL_OLT_TYPED_LITERAL = 0,
+    COWL_OLT_STRING_LITERAL,
+    COWL_OLT_STRING_LITERAL_LANGUAGE
+} CowlOBFLiteralType;
+
+#define COWL_OBF_FORMAT_VERSION 1U
+
+#define COWL_OBF_HEADER_IMPORT_OFFSET 0U
+#define COWL_OBF_HEADER_ANNOTATION_OFFSET 1U
+#define COWL_OBF_HEADER_VERSION_OFFSET 2U
+#define COWL_OBF_HEADER_IRI_OFFSET 3U
+#define COWL_OBF_HEADER_FORMAT_OFFSET 4U
+#define COWL_OBF_HEADER_ANNOTATIONS_BITMASK (1U << COWL_OBF_HEADER_ANNOTATION_OFFSET)
+#define COWL_OBF_HEADER_VERSION_BITMASK (1U << COWL_OBF_HEADER_VERSION_OFFSET)
+#define COWL_OBF_HEADER_IRI_BITMASK (1U << COWL_OBF_HEADER_IRI_OFFSET)
+#define COWL_OBF_HEADER_IMPORT_BITMASK (1U)
+
+#define COWL_OBF_AXIOM_TYPE_SIZE 6U
+#define COWL_OBF_AXIOM_ANNOT_OFFSET COWL_OBF_AXIOM_TYPE_SIZE
+#define COWL_OBF_AXIOM_UTLITY_OFFSET (COWL_OBF_AXIOM_ANNOT_OFFSET + 1U)
+#define COWL_OBF_AXIOM_TYPE_BITMASK ((1U << COWL_OBF_AXIOM_TYPE_SIZE) - 1U)
+#define COWL_OBF_AXIOM_ANNOT_BITMASK (1U << COWL_OBF_AXIOM_ANNOT_OFFSET)
+#define COWL_OBF_AXIOM_UTILITY_BITMASK (1U << COWL_OBF_AXIOM_UTLITY_OFFSET)
+
+#define COWL_OBF_LITERAL_TYPE_BITMASK 0x03
+#define COWL_OBF_STRING_COMPRESSION_OFFSET 2U
+
+#endif // COWL_OBF_CONSTANTS_H

src/formats/obf/cowl_obf_map.c

@@ -0,0 +1,168 @@
+/**
+ * @author Valerio Di Ceglie
+ *
+ * @copyright Copyright (c) 2024 SisInf Lab, Polytechnic University of Bari
+ * @copyright <http://swot.sisinflab.poliba.it>
+ * @copyright SPDX-License-Identifier: EPL-2.0
+ *
+ * @file
+ */
+
+#include "cowl_obf_map.h"
+#include "cowl.h"
+#include "ulib.h"
+#include <stddef.h>
+#include <stdint.h>
+
+static inline void *uintptr_to_void(uintptr_t ptr) {
+    // NOLINTBEGIN(performance-no-int-to-ptr)
+    return (void *)ptr;
+    // NOLINTEND(performance-no-int-to-ptr)
+}
+
+static cowl_ret populate_standard_prefixes(CowlOBFMap *map) {
+    const CowlVocab *vocab = cowl_vocab();
+    // Update COWL_OBF_STANDARD_PREFIX_COUNT if you change the number of standard prefixes.
+    CowlString *ns[] = {
+        vocab->rdf->ns, vocab->rdfs->ns, vocab->xsd->ns, vocab->owl->ns, vocab->xml->ns,
+    };
+    CowlString *prefix[] = {
+        vocab->rdf->prefix, vocab->rdfs->prefix, vocab->xsd->prefix,
+        vocab->owl->prefix, vocab->xml->prefix,
+    };
+    cowl_ret ret = COWL_OK;
+    for (ulib_uint i = 0; i < COWL_OBF_STANDARD_PREFIX_COUNT; ++i) {
+        if ((ret = cowl_obf_map_add_prefix(map, ns[i], prefix[i]))) break;
+    }
+    return ret;
+}
+
+cowl_ret cowl_obf_map_init(CowlOBFMap *map) {
+    map->_idx_to_prefix = uvec(CowlObjectPtr);
+    map->_idx_to_ns = uvec(CowlObjectPtr);
+    map->_prefix_to_idx = uhmap(CowlObjectTable);
+    map->_idx_to_id = uvec(CowlObjectPtr);
+    map->_id_to_idx = uhmap(CowlObjectTable);
+    cowl_ret ret = populate_standard_prefixes(map);
+    if (ret) cowl_obf_map_deinit(map);
+    return ret;
+}
+
+void cowl_obf_map_deinit(CowlOBFMap *map) {
+    uvec_foreach (CowlObjectPtr, &map->_idx_to_prefix, e) {
+        cowl_release(*e.item);
+    }
+    uvec_foreach (CowlObjectPtr, &map->_idx_to_ns, e) {
+        cowl_release(*e.item);
+    }
+    uvec_foreach (CowlObjectPtr, &map->_idx_to_id, e) {
+        cowl_release(*e.item);
+    }
+    uvec_deinit(CowlObjectPtr, &map->_idx_to_id);
+    uvec_deinit(CowlObjectPtr, &map->_idx_to_prefix);
+    uvec_deinit(CowlObjectPtr, &map->_idx_to_ns);
+    uhash_deinit(CowlObjectTable, &map->_prefix_to_idx);
+    uhash_deinit(CowlObjectTable, &map->_id_to_idx);
+}
+
+UVec(CowlObjectPtr) const *cowl_obf_map_get_ids(CowlOBFMap const *map) {
+    return &map->_idx_to_id;
+}
+
+UVec(CowlObjectPtr) const *cowl_obf_map_get_prefixes(CowlOBFMap const *map) {
+    return &map->_idx_to_prefix;
+}
+
+UVec(CowlObjectPtr) const *cowl_obf_map_get_namespaces(CowlOBFMap const *map) {
+    return &map->_idx_to_ns;
+}
+
+CowlString *cowl_obf_map_get_ns(CowlOBFMap const *map, ulib_uint idx) {
+    return uvec_get(CowlObjectPtr, &map->_idx_to_ns, idx);
+}
+
+ulib_uint cowl_obf_map_get_ns_idx(CowlOBFMap const *map, CowlString *namespace) {
+    uintptr_t idx = (uintptr_t)uhmap_get(CowlObjectTable, &map->_prefix_to_idx, namespace,
+                                         uintptr_to_void(COWL_OBF_NOT_FOUND));
+    return (ulib_uint)idx;
+}
+
+CowlIRIOrAnonInd *cowl_obf_map_get_id(CowlOBFMap const *map, ulib_uint idx) {
+    return uvec_get(CowlObjectPtr, &map->_idx_to_id, idx);
+}
+
+ulib_uint cowl_obf_map_get_id_idx(CowlOBFMap const *map, CowlIRIOrAnonInd *id) {
+    uintptr_t idx = (uintptr_t)uhmap_get(CowlObjectTable, &map->_id_to_idx, id,
+                                         uintptr_to_void(COWL_OBF_NOT_FOUND));
+    return (ulib_uint)idx;
+}
+
+cowl_ret cowl_obf_map_add_prefix(CowlOBFMap *map, CowlString *ns, CowlString *prefix) {
+    ulib_uint tbl_idx;
+    uhash_ret ret;
+    ret = uhash_put(CowlObjectTable, &map->_prefix_to_idx, ns, &tbl_idx);
+    if (ret == UHASH_ERR) return COWL_ERR_MEM;
+    if (ret == UHASH_PRESENT) return COWL_OK;
+    if (uvec_push(CowlObjectPtr, &map->_idx_to_prefix, prefix)) goto err_prefix;
+    if (uvec_push(CowlObjectPtr, &map->_idx_to_ns, ns)) goto err_ns;
+
+    ulib_uint index = uvec_count(CowlObjectPtr, &map->_idx_to_prefix) - 1;
+    uhash_value(CowlObjectTable, &map->_prefix_to_idx, tbl_idx) = uintptr_to_void(index);
+    cowl_retain(ns);
+    if (prefix) cowl_retain(prefix);
+
+    return COWL_OK;
+
+err_ns:
+    uvec_pop(CowlObjectPtr, &map->_idx_to_prefix, NULL);
+err_prefix:
+    uhash_delete(CowlObjectTable, &map->_prefix_to_idx, tbl_idx);
+    return COWL_ERR_MEM;
+}
+
+static inline cowl_ret add_iri_ns(CowlOBFMap *map, CowlIRI *iri) {
+    return cowl_obf_map_add_prefix(map, cowl_iri_get_ns(iri), NULL);
+}
+
+cowl_ret cowl_obf_map_add_primitive_fast(CowlOBFMap *map, CowlAnyPrimitive *primitive) {
+    CowlIRIOrAnonInd *id = cowl_get_iri(primitive);
+
+    if (id) {
+        if (add_iri_ns(map, id)) return COWL_ERR_MEM;
+    } else {
+        id = primitive;
+    }
+
+    if (uvec_push(CowlObjectPtr, &map->_idx_to_id, id)) {
+        return COWL_ERR_MEM;
+    }
+
+    cowl_retain(id);
+    return COWL_OK;
+}
+
+cowl_ret cowl_obf_map_add_primitive(CowlOBFMap *map, CowlAnyPrimitive *primitive) {
+    CowlIRIOrAnonInd *id = cowl_get_iri(primitive);
+    bool id_is_iri = true;
+
+    if (!id) {
+        id_is_iri = false;
+        id = primitive;
+    }
+
+    ulib_uint tbl_idx;
+    uhash_ret ret = uhash_put(CowlObjectTable, &map->_id_to_idx, id, &tbl_idx);
+    if (ret == UHASH_ERR) return COWL_ERR_MEM;
+    if (ret == UHASH_PRESENT) return COWL_OK;
+
+    if (uvec_push(CowlObjectPtr, &map->_idx_to_id, id)) {
+        uhash_delete(CowlObjectTable, &map->_id_to_idx, tbl_idx);
+        return COWL_ERR_MEM;
+    }
+
+    ulib_uint index = uvec_count(CowlObjectPtr, &map->_idx_to_id) - 1;
+    uhash_value(CowlObjectTable, &map->_id_to_idx, tbl_idx) = uintptr_to_void(index);
+    cowl_retain(id);
+
+    return id_is_iri ? add_iri_ns(map, id) : COWL_OK;
+}

src/formats/obf/cowl_obf_map.h

@@ -0,0 +1,49 @@
+/**
+ * @author Valerio Di Ceglie
+ *
+ * @copyright Copyright (c) 2024 SisInf Lab, Polytechnic University of Bari
+ * @copyright <http://swot.sisinflab.poliba.it>
+ * @copyright SPDX-License-Identifier: EPL-2.0
+ *
+ * @file
+ */
+
+#ifndef COWL_OBF_MAP_H
+#define COWL_OBF_MAP_H
+
+#include "cowl.h"
+#include "ulib.h"
+
+typedef struct CowlOBFMap {
+    UVec(CowlObjectPtr) _idx_to_prefix;
+    UVec(CowlObjectPtr) _idx_to_ns;
+    UHash(CowlObjectTable) _prefix_to_idx;
+    UVec(CowlObjectPtr) _idx_to_id;
+    UHash(CowlObjectTable) _id_to_idx;
+} CowlOBFMap;
+
+typedef void CowlIRIOrAnonInd;
+
+#define COWL_OBF_NOT_FOUND ((ulib_uint)(-1))
+#define COWL_OBF_STANDARD_PREFIX_COUNT 5
+
+cowl_ret cowl_obf_map_init(CowlOBFMap *map);
+void cowl_obf_map_deinit(CowlOBFMap *map);
+
+UVec(CowlObjectPtr) const *cowl_obf_map_get_ids(CowlOBFMap const *map);
+UVec(CowlObjectPtr) const *cowl_obf_map_get_prefixes(CowlOBFMap const *map);
+UVec(CowlObjectPtr) const *cowl_obf_map_get_namespaces(CowlOBFMap const *map);
+
+CowlString *cowl_obf_map_get_ns(CowlOBFMap const *map, ulib_uint idx);
+ulib_uint cowl_obf_map_get_ns_idx(CowlOBFMap const *map, CowlString *namespace);
+
+CowlIRIOrAnonInd *cowl_obf_map_get_id(CowlOBFMap const *map, ulib_uint idx);
+ulib_uint cowl_obf_map_get_id_idx(CowlOBFMap const *map, CowlIRIOrAnonInd *id);
+
+cowl_ret cowl_obf_map_add_primitive(CowlOBFMap *map, CowlAnyPrimitive *primitive);
+cowl_ret cowl_obf_map_add_prefix(CowlOBFMap *map, CowlString *ns, CowlString *prefix);
+
+// Only use when decoding, this won't populate the primitive -> index map.
+cowl_ret cowl_obf_map_add_primitive_fast(CowlOBFMap *map, CowlAnyPrimitive *primitive);
+
+#endif // COWL_OBF_MAP_H

src/formats/obf/reader/CMakeLists.txt

@@ -0,0 +1,13 @@
+# Sources
+
+set(COWL_READER_SRC_DIR "${CMAKE_CURRENT_SOURCE_DIR}")
+file(GLOB COWL_READER_SOURCES CONFIGURE_DEPENDS "${COWL_READER_SRC_DIR}/*.c")
+list(APPEND COWL_SOURCES ${COWL_READER_SOURCES})
+
+cmake_path(GET COWL_READER_SRC_DIR PARENT_PATH COWL_READER_SRC_DIR)
+file(GLOB COWL_READER_SOURCES CONFIGURE_DEPENDS "${COWL_READER_SRC_DIR}/*.c")
+list(APPEND COWL_SOURCES ${COWL_READER_SOURCES})
+
+# Propagate changes in the parent project
+
+set(COWL_SOURCES ${COWL_SOURCES} PARENT_SCOPE)