Use uint8_t for API char type for UTF-8 friendliness.

Fix literal languages.


git-svn-id: http://svn.drobilla.net/sord/trunk@20 3d64ff67-21c5-427c-a301-fe4f08042e5a

6 files changed, 74 insertions, 67 deletions

diff --git a/sord/sord.h b/sord/sord.h
index 2d702a6..09fcc5a 100644
--- a/sord/sord.h
+++ b/sord/sord.h
@@ -160,24 +160,24 @@ sord_tuple_load(Sord      sord,
  */
 SORD_API
 SordID
-sord_get_uri(Sord sord, bool create, const char* str);
+sord_get_uri(Sord sord, bool create, const uint8_t* str);
 
 /** Find a URI, creating a new one if necessary iff @a create is true. */
 SORD_API
 SordID
-sord_get_uri_counted(Sord sord, bool create, const char* str, int str_len);
+sord_get_uri_counted(Sord sord, bool create, const uint8_t* str, int str_len);
 
 /** Find a blank, creating a new one if necessary iff @a create is true
  * Use sord_get_blank_counted instead if the length of @a str is known.
  */
 SORD_API
 SordID
-sord_get_blank(Sord sord, bool create, const char* str);
+sord_get_blank(Sord sord, bool create, const uint8_t* str);
 
 /** Find a blank, creating a new one if necessary iff @a create is true. */
 SORD_API
 SordID
-sord_get_blank_counted(Sord sord, bool create, const char* str, int str_len);
+sord_get_blank_counted(Sord sord, bool create, const uint8_t* str, int str_len);
 
 /** Find a literal, creating a new one if necessary iff @a create is true.
  * Use sord_get_literal_counted instead if the length of @a str is known.
@@ -185,14 +185,14 @@ sord_get_blank_counted(Sord sord, bool create, const char* str, int str_len);
 SORD_API
 SordID
 sord_get_literal(Sord sord, bool create, SordID type,
-                 const char* str, const char* lang);
+                 const uint8_t* str, const char* lang);
 
 /** Find a literal, creating a new one if necessary iff @a create is true. */
 SORD_API
 SordID
 sord_get_literal_counted(Sord sord, bool create, SordID type,
-                         const char* str,  int     str_len,
-                         const char* lang, uint8_t lang_len);
+                         const uint8_t* str,  int     str_len,
+                         const char*    lang, uint8_t lang_len);
 
 
 /** @} */
@@ -208,11 +208,11 @@ sord_node_get_type(SordNode node);
 
 /** Return the string value of a node. */
 SORD_API
-const char*
+const uint8_t*
 sord_node_get_string(SordNode node);
 
 SORD_API
-const char*
+const uint8_t*
 sord_node_get_string_counted(SordNode node, size_t* len);
 
 /** Set an opaque user pointer on a node.
diff --git a/src/sord.c b/src/sord.c
index 3e98095..8519bb3 100644
--- a/src/sord.c
+++ b/src/sord.c
@@ -148,7 +148,7 @@ static unsigned
 sord_literal_hash(const void* n)
 {
 	SordNode node = (SordNode)n;
-	return g_str_hash(node->buf) + g_str_hash(node->lang);
+	return g_str_hash(node->buf) + (node->lang ? g_str_hash(node->lang) : 0);
 }
 		
 static gboolean
@@ -170,10 +170,11 @@ sord_node_compare(const SordNode a, const SordNode b)
 	switch ((SordNodeType)a->type) {
 	case SORD_URI:
 	case SORD_BLANK:
-		return strcmp(a->buf, b->buf);
+		return strcmp((const char*)a->buf, (const char*)b->buf);
 	case SORD_LITERAL:
 		// TODO: lang, type
-		return strcmp(sord_node_get_string(a), sord_node_get_string(b));
+		return strcmp((const char*)sord_node_get_string(a),
+		              (const char*)sord_node_get_string(b));
 	}
 	assert(false);
 	return 0;
@@ -790,13 +791,13 @@ sord_find(Sord sord, const SordTuple pat)
 }
 
 static SordID
-sord_lookup_name(Sord sord, const char* str, int str_len)
+sord_lookup_name(Sord sord, const uint8_t* str, size_t str_len)
 {
 	return g_hash_table_lookup(sord->names, str);
 }
 
 static SordNode
-sord_new_node(SordNodeType type, const char* data, size_t n_bytes)
+sord_new_node(SordNodeType type, const uint8_t* data, size_t n_bytes)
 {
 	SordNode node = malloc(sizeof(struct _SordNode));
 	node->type      = type;
@@ -805,25 +806,25 @@ sord_new_node(SordNodeType type, const char* data, size_t n_bytes)
 	node->user_data = 0;
 	node->datatype  = 0;
 	node->lang      = 0;
-	node->buf       = g_strdup(data); // TODO: add no-copy option
+	node->buf       = (uint8_t*)g_strdup((const char*)data); // TODO: no-copy
 	return node;
 }
 
 static SordNode
 sord_new_literal_node(Sord sord, SordNode datatype,
-                      const char* str,  int str_len,
-                      const char* lang, uint8_t lang_len)
+                      const uint8_t* str,  int str_len,
+                      const char*    lang, uint8_t lang_len)
 {
 	SordNode node = sord_new_node(SORD_LITERAL, str, str_len + 1);
 	node->datatype = datatype;
-	node->lang     = g_intern_string(str);
+	node->lang     = lang ? g_intern_string(lang) : NULL;
 	return node;
 }
 
 static SordNode
 sord_lookup_literal(Sord sord, SordNode type,
-                    const char* str,  int     str_len,
-                    const char* lang, uint8_t lang_len)
+                    const uint8_t* str,  int     str_len,
+                    const char*    lang, uint8_t lang_len)
 {
 	SordNode node = sord_new_literal_node(sord, type, str, str_len, lang, lang_len);
 	SordNode id   = g_hash_table_lookup(sord->literals, node);
@@ -847,13 +848,13 @@ sord_node_get_type(SordNode ref)
 	return ref->type;
 }
 
-const char*
+const uint8_t*
 sord_node_get_string(SordNode ref)
 {
-	return ref->buf;
+	return (const uint8_t*)ref->buf;
 }
 
-const char*
+const uint8_t*
 sord_node_get_string_counted(SordNode ref, size_t* n_bytes)
 {
 	*n_bytes = ref->n_bytes;
@@ -892,53 +893,53 @@ sord_add_node(Sord sord, SordNode node)
 }
 
 SordID
-sord_get_uri_counted(Sord sord, bool create, const char* str, int str_len)
+sord_get_uri_counted(Sord sord, bool create, const uint8_t* str, int str_len)
 {
 	SordID id = sord_lookup_name(sord, str, str_len);
 	if (id || !create)
 		return id;
 
-	id = sord_new_node(SORD_URI, str, str_len + 1);
+	id = sord_new_node(SORD_URI, (const uint8_t*)str, str_len + 1);
 
 	assert(id);
-	g_hash_table_insert(sord->names, (void*)g_strdup(str), (void*)id);
+	g_hash_table_insert(sord->names, (void*)g_strdup((const char*)str), (void*)id);
 	sord_add_node(sord, id);
 	
 	return id;
 }
 
 SordID
-sord_get_uri(Sord sord, bool create, const char* str)
+sord_get_uri(Sord sord, bool create, const uint8_t* str)
 {
-	return sord_get_uri_counted(sord, create, str, strlen(str));
+	return sord_get_uri_counted(sord, create, str, strlen((const char*)str));
 }
 
 SordID
-sord_get_blank_counted(Sord sord, bool create, const char* str, int str_len)
+sord_get_blank_counted(Sord sord, bool create, const uint8_t* str, int str_len)
 {
 	SordID id = sord_lookup_name(sord, str, str_len);
 	if (id || !create)
 		return id;
 
-	id = sord_new_node(SORD_BLANK, str, str_len + 1);
+	id = sord_new_node(SORD_BLANK, (const uint8_t*)str, str_len + 1);
 
 	assert(id);
-	g_hash_table_insert(sord->names, (void*)g_strdup(str), (void*)id);
+	g_hash_table_insert(sord->names, (void*)g_strdup((const char*)str), (void*)id);
 	sord_add_node(sord, id);
 	
 	return id;
 }
 
 SordID
-sord_get_blank(Sord sord, bool create, const char* str)
+sord_get_blank(Sord sord, bool create, const uint8_t* str)
 {
-	return sord_get_blank_counted(sord, create, str, strlen(str));
+	return sord_get_blank_counted(sord, create, str, strlen((const char*)str));
 }
 
 SordID
 sord_get_literal_counted(Sord sord, bool create, SordID type,
-                         const char* str,  int     str_len,
-                         const char* lang, uint8_t lang_len)
+                         const uint8_t* str,  int     str_len,
+                         const char*    lang, uint8_t lang_len)
 {
 	SordID id = sord_lookup_literal(sord, type, str, str_len, lang, lang_len);
 	if (id || !create)
@@ -953,10 +954,12 @@ sord_get_literal_counted(Sord sord, bool create, SordID type,
 }
 
 SordID
-sord_get_literal(Sord sord, bool create, SordID type, const char* str, const char* lang)
+sord_get_literal(Sord sord, bool create, SordID type,
+                 const uint8_t* str, const char* lang)
 {
-	return sord_get_literal_counted(sord, create, type, str, strlen(str),
-			lang, lang ? strlen(lang) : 0);
+	return sord_get_literal_counted(sord, create, type,
+	                                str, strlen((const char*)str),
+	                                lang, lang ? strlen(lang) : 0);
 }
 
 static inline bool
diff --git a/src/sord_internal.h b/src/sord_internal.h
index b7a3398..dc7ac26 100644
--- a/src/sord_internal.h
+++ b/src/sord_internal.h
@@ -18,6 +18,9 @@
 #ifndef SORD_INTERNAL_H
 #define SORD_INTERNAL_H
 
+#include <stddef.h>
+#include <stdint.h>
+
 #include "sord/sord.h"
 
 /** Node */
@@ -28,7 +31,7 @@ struct _SordNode {
 	void*        user_data;  ///< Opaque user data
 	SordNode     datatype;   ///< Literal data type (ID of a URI node, or 0)
 	const char*  lang;       ///< Literal language (interned string)
-	char*        buf;        ///< Value (string)
+	uint8_t*     buf;        ///< Value (string)
 };
 
 #endif // SORD_INTERNAL_H
diff --git a/src/sord_test.c b/src/sord_test.c
index 97b7b4d..4a3dda7 100644
--- a/src/sord_test.c
+++ b/src/sord_test.c
@@ -27,6 +27,8 @@ static const int MAX_NUM = 999;
 
 typedef struct { SordTuple query; int expected_num_results; } QueryTest;
 
+#define USTR(s) ((const uint8_t*)(s))
+
 static SordID
 uri(Sord sord, int num)
 {
@@ -35,9 +37,9 @@ uri(Sord sord, int num)
 
 	char         uri[]   = "eg:000";
 	const size_t uri_len = 3 + DIGITS;
-	char*        uri_num = uri + 3;  // First `0'
+	char*        uri_num = uri + 3; // First `0'
 	snprintf(uri_num, DIGITS + 1, "%0*d", DIGITS, num);
-	return sord_get_uri_counted(sord, true, uri, uri_len);
+	return sord_get_uri_counted(sord, true, (const uint8_t*)uri, uri_len);
 }
 
 void
@@ -64,16 +66,16 @@ generate(Sord sord, size_t n_tuples, size_t n_objects_per)
 	SordTuple tup;
 	tup[0] = uri(sord, 98);
 	tup[1] = uri(sord, 4);
-	tup[2] = sord_get_literal(sord, true, 0, "hello", NULL);
+	tup[2] = sord_get_literal(sord, true, 0, (const uint8_t*)"hello", NULL);
 	tup[3] = 0;
 	sord_add(sord, tup);
-	tup[2] = sord_get_literal(sord, true, 0, "hi", NULL);
+	tup[2] = sord_get_literal(sord, true, 0, USTR("hi"), NULL);
 	sord_add(sord, tup);
 
 	tup[0] = uri(sord, 14);
-	tup[2] = sord_get_literal(sord, true, 0, "bonjour", "fr");
+	tup[2] = sord_get_literal(sord, true, 0, USTR("bonjour"), "fr");
 	sord_add(sord, tup);
-	tup[2] = sord_get_literal(sord, true, 0, "salut", "fr");
+	tup[2] = sord_get_literal(sord, true, 0, USTR("salut"), "fr");
 	sord_add(sord, tup);
 
 	// Attempt to add some duplicates
@@ -81,7 +83,7 @@ generate(Sord sord, size_t n_tuples, size_t n_objects_per)
 	sord_add(sord, tup);
 
 	// Add a blank node subject
-	tup[0] = sord_get_blank(sord, true, "ablank");
+	tup[0] = sord_get_blank(sord, true, USTR("ablank"));
 	sord_add(sord, tup);
 
 	tup[1] = uri(sord, 6);
@@ -99,11 +101,11 @@ test_fail()
 #define TUP_FMT "(%6s %6s %6s)"
 #define TUP_FMT_ARGS(t) \
 	(sord_node_load(sord, (t)[0]) \
-		? sord_node_get_string(sord_node_load(sord, (t)[0])) : "*"), \
+	 ? sord_node_get_string(sord_node_load(sord, (t)[0])) : USTR("*")), \
 	(sord_node_load(sord, (t)[1]) \
-		? sord_node_get_string(sord_node_load(sord, (t)[1])) : "*"), \
+	 ? sord_node_get_string(sord_node_load(sord, (t)[1])) : USTR("*")), \
 	(sord_node_load(sord, (t)[2]) \
-		? sord_node_get_string(sord_node_load(sord, (t)[2])) : "*")
+	 ? sord_node_get_string(sord_node_load(sord, (t)[2])) : USTR("*"))
 
 int
 test_read(Sord sord, const size_t n_tuples, const int n_objects_per)
@@ -169,7 +171,7 @@ test_read(Sord sord, const size_t n_tuples, const int n_objects_per)
 	}
 
 	// Query blank node subject
-	SordTuple pat = { sord_get_blank(sord, true, "ablank"), 0, 0 };
+	SordTuple pat = { sord_get_blank(sord, true, USTR("ablank")), 0, 0 };
 	if (!pat[0]) {
 		fprintf(stderr, "Blank node subject lost\n");
 		return test_fail();
@@ -277,13 +279,13 @@ main(int argc, char** argv)
 	}
 
 	// Check interning merges equivalent values
-	SordID uri_id   = sord_get_uri(sord, true, "http://example.org");
-	SordID blank_id = sord_get_uri(sord, true, "testblank");
-	SordID lit_id   = sord_get_literal(sord, true, uri_id, "hello", NULL);
+	SordID uri_id   = sord_get_uri(sord, true, USTR("http://example.org"));
+	SordID blank_id = sord_get_uri(sord, true, USTR("testblank"));
+	SordID lit_id   = sord_get_literal(sord, true, uri_id, USTR("hello"), NULL);
 	//sord_clear_cache(write);
-	SordID uri_id2   = sord_get_uri(sord, false, "http://example.org");
-	SordID blank_id2 = sord_get_uri(sord, false, "testblank");
-	SordID lit_id2   = sord_get_literal(sord, false, uri_id, "hello", NULL);
+	SordID uri_id2   = sord_get_uri(sord, false, USTR("http://example.org"));
+	SordID blank_id2 = sord_get_uri(sord, false, USTR("testblank"));
+	SordID lit_id2   = sord_get_literal(sord, false, uri_id, USTR("hello"), NULL);
 	if (uri_id2 != uri_id) {
 		fprintf(stderr, "Fail: URI interning failed (duplicates)\n");
 		goto fail;
@@ -296,9 +298,9 @@ main(int argc, char** argv)
 	}
 
 	// Check interning doesn't clash non-equivalent values
-	SordID uri_id3   = sord_get_uri(sord, false, "http://example.orgX");
-	SordID blank_id3 = sord_get_uri(sord, false, "testblankX");
-	SordID lit_id3   = sord_get_literal(sord, false, uri_id, "helloX", NULL);
+	SordID uri_id3   = sord_get_uri(sord, false, USTR("http://example.orgX"));
+	SordID blank_id3 = sord_get_uri(sord, false, USTR("testblankX"));
+	SordID lit_id3   = sord_get_literal(sord, false, uri_id, USTR("helloX"), NULL);
 	if (uri_id3 == uri_id) {
 		fprintf(stderr, "Fail: URI interning failed (clash)\n");
 		goto fail;
diff --git a/src/sordi.c b/src/sordi.c
index f304b9e..4a7454b 100644
--- a/src/sordi.c
+++ b/src/sordi.c
@@ -65,9 +65,9 @@ file_sink(const void* buf, size_t len, void* stream)
 static inline SerdNode
 serd_node_from_sord_node(const SordNode n)
 {
-	size_t      n_bytes = 0;
-	const char* buf     = sord_node_get_string_counted(n, &n_bytes);
-	SerdNode    sn      = { SERD_NOTHING, n_bytes, n_bytes - 1, (const uint8_t*)buf };
+	size_t         n_bytes = 0;
+	const uint8_t* buf     = sord_node_get_string_counted(n, &n_bytes);
+	SerdNode       sn      = { SERD_NOTHING, n_bytes, n_bytes - 1, (const uint8_t*)buf };
 	// FIXME: UTF-8
 	switch (sord_node_get_type(n)) {
 	case SORD_URI:
diff --git a/src/syntax.c b/src/syntax.c
index e7560c5..3afda17 100644
--- a/src/syntax.c
+++ b/src/syntax.c
@@ -101,8 +101,7 @@ sord_node_from_serd_node(ReadState* state, const SerdNode* sn)
 	case SERD_NOTHING:
 		return NULL;
 	case SERD_LITERAL:
-		return sord_get_literal(state->sord, true, NULL,
-		                        (const char*)sn->buf, NULL);
+		return sord_get_literal(state->sord, true, NULL, sn->buf, NULL);
 	case SERD_URI: {
 		SerdURI uri;
 		if (!serd_uri_parse(sn->buf, &uri)) {
@@ -114,7 +113,7 @@ sord_node_from_serd_node(ReadState* state, const SerdNode* sn)
 		}
 		SerdURI ignored;
 		SerdNode abs_uri_node = serd_node_new_uri(&abs_uri, &ignored);
-		SordID ret = sord_get_uri(state->sord, true, (const char*)abs_uri_node.buf);
+		SordID ret = sord_get_uri(state->sord, true, abs_uri_node.buf);
 		serd_node_free(&abs_uri_node);
 		return ret;
 	}
@@ -126,7 +125,7 @@ sord_node_from_serd_node(ReadState* state, const SerdNode* sn)
 			return NULL;
 		}
 		const size_t uri_len = uri_prefix.len + uri_suffix.len;
-		char*        buf     = malloc(uri_len + 1);
+		uint8_t*     buf     = malloc(uri_len + 1);
 		memcpy(buf,                  uri_prefix.buf, uri_prefix.len);
 		memcpy(buf + uri_prefix.len, uri_suffix.buf, uri_suffix.len);
 		buf[uri_len] = '\0';
@@ -138,7 +137,7 @@ sord_node_from_serd_node(ReadState* state, const SerdNode* sn)
 	case SERD_BLANK_ID:
 	case SERD_ANON_BEGIN:
 	case SERD_ANON:
-		return sord_get_blank(state->sord, true, (const char*)sn->buf);
+		return sord_get_blank(state->sord, true, sn->buf);
 	}
 	return NULL;
 }