Merge datatype/language into node

This moves closer to the sord API, and is more convenient in most cases.

5 files changed, 171 insertions, 75 deletions

diff --git a/src/n3.c b/src/n3.c
index a4745733..9fb256af 100644
--- a/src/n3.c
+++ b/src/n3.c
@@ -854,10 +854,11 @@ read_0_9(SerdReader* const reader, const Ref str, const bool at_least_one)
 }
 
 static SerdStatus
-read_number(SerdReader* const reader,
-            Ref* const        dest,
-            Ref* const        datatype,
-            bool* const       ate_dot)
+read_number(SerdReader* const    reader,
+            Ref* const           dest,
+            Ref* const           datatype,
+            SerdNodeFlags* const flags,
+            bool* const          ate_dot)
 {
 #define XSD_DECIMAL NS_XSD "decimal"
 #define XSD_DOUBLE NS_XSD "double"
@@ -916,6 +917,10 @@ read_number(SerdReader* const reader,
       push_node(reader, SERD_URI, XSD_INTEGER, sizeof(XSD_INTEGER) - 1);
   }
 
+  if (*datatype) {
+    *flags |= SERD_HAS_DATATYPE;
+  }
+
   return SERD_SUCCESS;
 }
 
@@ -950,6 +955,7 @@ read_literal(SerdReader* const    reader,
   switch (peek_byte(reader)) {
   case '@':
     eat_byte_safe(reader, '@');
+    *flags |= SERD_HAS_LANGUAGE;
     if ((st = read_LANGTAG(reader, lang))) {
       *datatype = pop_node(reader, *datatype);
       *lang     = pop_node(reader, *lang);
@@ -960,6 +966,7 @@ read_literal(SerdReader* const    reader,
   case '^':
     eat_byte_safe(reader, '^');
     eat_byte_check(reader, '^');
+    *flags |= SERD_HAS_DATATYPE;
     if ((st = read_iri(reader, datatype, ate_dot))) {
       *datatype = pop_node(reader, *datatype);
       *lang     = pop_node(reader, *lang);
@@ -968,7 +975,6 @@ read_literal(SerdReader* const    reader,
     }
     break;
   }
-
   return SERD_SUCCESS;
 }
 
@@ -1079,7 +1085,7 @@ read_anon(SerdReader* const reader,
 
   SerdStatus st = SERD_SUCCESS;
   if (ctx.subject) {
-    TRY(st, emit_statement(reader, ctx, *dest, 0, 0));
+    TRY(st, emit_statement(reader, ctx, *dest));
   }
 
   ctx.subject = *dest;
@@ -1088,21 +1094,17 @@ read_anon(SerdReader* const reader,
     if (!subject) {
       *ctx.flags |= SERD_ANON_CONT;
     }
-
     bool ate_dot_in_list = false;
     read_predicateObjectList(reader, ctx, &ate_dot_in_list);
     if (ate_dot_in_list) {
       return r_err(reader, SERD_ERR_BAD_SYNTAX, "`.' inside blank\n");
     }
-
     read_ws_star(reader);
     if (reader->end_func) {
       reader->end_func(reader->handle, deref(reader, *dest));
     }
-
     *ctx.flags = old_flags;
   }
-
   return (eat_byte_check(reader, ']') == ']') ? SERD_SUCCESS
                                               : SERD_ERR_BAD_SYNTAX;
 }
@@ -1175,7 +1177,7 @@ read_object(SerdReader* const  reader,
   case '7':
   case '8':
   case '9':
-    ret = read_number(reader, &o, &datatype, ate_dot);
+    ret = read_number(reader, &o, &datatype, &flags, ate_dot);
     break;
   case '\"':
   case '\'':
@@ -1191,6 +1193,7 @@ read_object(SerdReader* const  reader,
     node = deref(reader, o);
     if ((node->length == 4 && !memcmp(serd_node_string(node), "true", 4)) ||
         (node->length == 5 && !memcmp(serd_node_string(node), "false", 5))) {
+      flags      = flags | SERD_HAS_DATATYPE;
       node->type = SERD_LITERAL;
       datatype   = push_node(reader, SERD_URI, XSD_BOOLEAN, XSD_BOOLEAN_LEN);
       ret        = SERD_SUCCESS;
@@ -1210,7 +1213,7 @@ read_object(SerdReader* const  reader,
   }
 
   if (!ret && emit && simple) {
-    ret = emit_statement(reader, *ctx, o, datatype, lang);
+    ret = emit_statement(reader, *ctx, o);
   } else if (!ret && !emit) {
     ctx->object   = o;
     ctx->datatype = datatype;
@@ -1306,14 +1309,12 @@ read_collection(SerdReader* const reader, ReadContext ctx, Ref* const dest)
 {
   SerdStatus st = SERD_SUCCESS;
   eat_byte_safe(reader, '(');
-
   bool end = peek_delim(reader, ')');
-
-  *dest = end ? reader->rdf_nil : blank_id(reader);
+  *dest    = end ? reader->rdf_nil : blank_id(reader);
   if (ctx.subject) {
     // subject predicate _:head
     *ctx.flags |= (end ? 0 : SERD_LIST_O_BEGIN);
-    TRY(st, emit_statement(reader, ctx, *dest, 0, 0));
+    TRY(st, emit_statement(reader, ctx, *dest));
     *ctx.flags |= SERD_LIST_CONT;
   } else {
     *ctx.flags |= (end ? 0 : SERD_LIST_S_BEGIN);
@@ -1352,7 +1353,7 @@ read_collection(SerdReader* const reader, ReadContext ctx, Ref* const dest)
     // _:node rdf:rest _:rest
     *ctx.flags |= SERD_LIST_CONT;
     ctx.predicate = reader->rdf_rest;
-    TRY(st, emit_statement(reader, ctx, (end ? reader->rdf_nil : rest), 0, 0));
+    TRY(st, emit_statement(reader, ctx, (end ? reader->rdf_nil : rest)));
 
     ctx.subject = rest;        // _:node = _:rest
     rest        = node;        // _:rest = (old)_:node
@@ -1733,7 +1734,7 @@ read_nquadsDoc(SerdReader* const reader)
       }
     }
 
-    TRY(st, emit_statement(reader, ctx, ctx.object, ctx.datatype, ctx.lang));
+    TRY(st, emit_statement(reader, ctx, ctx.object));
 
     pop_node(reader, ctx.graph);
     pop_node(reader, ctx.lang);
diff --git a/src/node.c b/src/node.c
index 854772ac..edfa28b5 100644
--- a/src/node.c
+++ b/src/node.c
@@ -42,6 +42,8 @@
 
 static const size_t serd_node_align = 2 * sizeof(uint64_t);
 
+static const SerdNodeFlags meta_mask = (SERD_HAS_DATATYPE | SERD_HAS_LANGUAGE);
+
 static size_t
 serd_uri_string_length(const SerdURIView* const uri)
 {
@@ -77,11 +79,25 @@ serd_node_pad_size(const size_t n_bytes)
   return n_bytes + 2 + pad;
 }
 
+static const SerdNode*
+serd_node_meta_c(const SerdNode* const node)
+{
+  return node + 1 + (serd_node_pad_size(node->length) / sizeof(SerdNode));
+}
+
+static const SerdNode*
+serd_node_maybe_get_meta_c(const SerdNode* const node)
+{
+  return (node->flags & meta_mask) ? serd_node_meta_c(node) : NULL;
+}
+
 static SERD_PURE_FUNC
 size_t
 serd_node_total_size(const SerdNode* const node)
 {
-  return node ? (sizeof(SerdNode) + serd_node_pad_size(node->length)) : 0;
+  return node ? (sizeof(SerdNode) + serd_node_pad_size(node->length) +
+                 serd_node_total_size(serd_node_maybe_get_meta_c(node)))
+              : 0;
 }
 
 SerdNode*
@@ -96,6 +112,7 @@ serd_node_malloc(const size_t        length,
   node->flags  = flags;
   node->type   = type;
 
+  assert((intptr_t)node % serd_node_align == 0);
   return node;
 }
 
@@ -115,7 +132,7 @@ serd_node_set(SerdNode** const dst, const SerdNode* const src)
   }
 
   assert(*dst);
-  memcpy(*dst, src, sizeof(SerdNode) + src->length + 1);
+  memcpy(*dst, src, size);
 }
 
 SerdNode*
@@ -124,8 +141,10 @@ serd_new_string(SerdNodeType type, const char* str)
   SerdNodeFlags flags  = 0;
   const size_t  length = serd_strlen(str, &flags);
   SerdNode*     node   = serd_node_malloc(length, flags, type);
+
   memcpy(serd_node_buffer(node), str, length);
   node->length = length;
+
   return node;
 }
 
@@ -143,6 +162,49 @@ serd_new_substring(const SerdNodeType type,
 }
 
 SerdNode*
+serd_new_literal(const char* const str,
+                 const char* const datatype,
+                 const char* const lang)
+{
+  SerdNodeFlags flags  = 0;
+  const size_t  length = serd_strlen(str, &flags);
+  const size_t  len    = serd_node_pad_size(length);
+
+  SerdNode* node = NULL;
+  if (lang) {
+    flags |= SERD_HAS_LANGUAGE;
+    const size_t lang_len  = strlen(lang);
+    const size_t total_len = len + sizeof(SerdNode) + lang_len;
+    node                   = serd_node_malloc(total_len, flags, SERD_LITERAL);
+    memcpy(serd_node_buffer(node), str, length);
+    node->length = length;
+
+    SerdNode* lang_node = node + 1 + (len / sizeof(SerdNode));
+    lang_node->type     = SERD_LITERAL;
+    lang_node->length   = lang_len;
+    memcpy(serd_node_buffer(lang_node), lang, lang_len);
+  } else if (datatype) {
+    flags |= SERD_HAS_DATATYPE;
+    const size_t datatype_len = strlen(datatype);
+    const size_t total_len    = len + sizeof(SerdNode) + datatype_len;
+    node = serd_node_malloc(total_len, flags, SERD_LITERAL);
+    memcpy(serd_node_buffer(node), str, length);
+    node->length = length;
+
+    SerdNode* datatype_node = node + 1 + (len / sizeof(SerdNode));
+    datatype_node->type     = SERD_URI;
+    datatype_node->length   = datatype_len;
+    memcpy(serd_node_buffer(datatype_node), datatype, datatype_len);
+  } else {
+    node = serd_node_malloc(length, flags, SERD_LITERAL);
+    memcpy(serd_node_buffer(node), str, length);
+    node->length = length;
+  }
+
+  return node;
+}
+
+SerdNode*
 serd_node_copy(const SerdNode* node)
 {
   if (!node) {
@@ -158,9 +220,30 @@ serd_node_copy(const SerdNode* node)
 bool
 serd_node_equals(const SerdNode* const a, const SerdNode* const b)
 {
-  return (a == b) ||
-         (a && b && a->type == b->type && a->length == b->length &&
-          !memcmp(serd_node_string(a), serd_node_string(b), a->length));
+  if (a == b) {
+    return true;
+  }
+
+  if (!a || !b || a->length != b->length || a->flags != b->flags ||
+      a->type != b->type) {
+    return false;
+  }
+
+  const size_t length = a->length;
+  if (!!memcmp(serd_node_string(a), serd_node_string(b), length)) {
+    return false;
+  }
+
+  const SerdNodeFlags flags = a->flags;
+  if (flags & meta_mask) {
+    const SerdNode* const am = serd_node_meta_c(a);
+    const SerdNode* const bm = serd_node_meta_c(b);
+
+    return am->length == bm->length && am->type == bm->type &&
+           !memcmp(serd_node_string(am), serd_node_string(bm), am->length);
+  }
+
+  return true;
 }
 
 SerdNode*
@@ -444,6 +527,32 @@ serd_node_uri_view(const SerdNode* const node)
                                   : SERD_URI_NULL;
 }
 
+const SerdNode*
+serd_node_datatype(const SerdNode* const node)
+{
+  if (!node || !(node->flags & SERD_HAS_DATATYPE)) {
+    return NULL;
+  }
+
+  const size_t          len      = serd_node_pad_size(node->length);
+  const SerdNode* const datatype = node + 1 + (len / sizeof(SerdNode));
+  assert(datatype->type == SERD_URI || datatype->type == SERD_CURIE);
+  return datatype;
+}
+
+const SerdNode*
+serd_node_language(const SerdNode* const node)
+{
+  if (!node || !(node->flags & SERD_HAS_LANGUAGE)) {
+    return NULL;
+  }
+
+  const size_t          len  = serd_node_pad_size(node->length);
+  const SerdNode* const lang = node + 1 + (len / sizeof(SerdNode));
+  assert(lang->type == SERD_LITERAL);
+  return lang;
+}
+
 SerdNodeFlags
 serd_node_flags(const SerdNode* const node)
 {
diff --git a/src/reader.c b/src/reader.c
index ff53ab46..d0cc8a97 100644
--- a/src/reader.c
+++ b/src/reader.c
@@ -146,11 +146,7 @@ pop_node(SerdReader* const reader, const Ref ref)
 }
 
 SerdStatus
-emit_statement(SerdReader* const reader,
-               const ReadContext ctx,
-               const Ref         o,
-               const Ref         d,
-               const Ref         l)
+emit_statement(SerdReader* const reader, const ReadContext ctx, const Ref o)
 {
   SerdNode* graph = deref(reader, ctx.graph);
   if (!graph && reader->default_graph) {
@@ -164,9 +160,7 @@ emit_statement(SerdReader* const reader,
                                                    graph,
                                                    deref(reader, ctx.subject),
                                                    deref(reader, ctx.predicate),
-                                                   deref(reader, o),
-                                                   deref(reader, d),
-                                                   deref(reader, l));
+                                                   deref(reader, o));
 
   *ctx.flags &= SERD_ANON_CONT | SERD_LIST_CONT; // Preserve only cont flags
   return st;
diff --git a/src/reader.h b/src/reader.h
index bb8f0b30..60a52e9d 100644
--- a/src/reader.h
+++ b/src/reader.h
@@ -118,7 +118,7 @@ Ref
 pop_node(SerdReader* reader, Ref ref);
 
 SerdStatus
-emit_statement(SerdReader* reader, ReadContext ctx, Ref o, Ref d, Ref l);
+emit_statement(SerdReader* reader, ReadContext ctx, Ref o);
 
 SerdStatus
 read_n3_statement(SerdReader* reader);
diff --git a/src/writer.c b/src/writer.c
index f0bd11eb..b4917052 100644
--- a/src/writer.c
+++ b/src/writer.c
@@ -116,8 +116,6 @@ typedef enum { WRITE_STRING, WRITE_LONG_STRING } TextContext;
 static bool
 write_node(SerdWriter*        writer,
            const SerdNode*    node,
-           const SerdNode*    datatype,
-           const SerdNode*    lang,
            Field              field,
            SerdStatementFlags flags);
 
@@ -511,14 +509,14 @@ is_inline_start(const SerdWriter* writer, Field field, SerdStatementFlags flags)
 }
 
 static bool
-write_literal(SerdWriter*        writer,
-              const SerdNode*    node,
-              const SerdNode*    datatype,
-              const SerdNode*    lang,
-              SerdStatementFlags flags)
+write_literal(SerdWriter* const        writer,
+              const SerdNode* const    node,
+              const SerdStatementFlags flags)
 {
-  const char* node_str = serd_node_string(node);
-  const char* type_uri = datatype ? serd_node_string(datatype) : NULL;
+  const SerdNode* datatype = serd_node_datatype(node);
+  const SerdNode* lang     = serd_node_language(node);
+  const char*     node_str = serd_node_string(node);
+  const char*     type_uri = datatype ? serd_node_string(datatype) : NULL;
   if (supports_abbrev(writer) && type_uri) {
     if (!strncmp(type_uri, NS_XSD, sizeof(NS_XSD) - 1) &&
         (!strcmp(type_uri + sizeof(NS_XSD) - 1, "boolean") ||
@@ -554,7 +552,7 @@ write_literal(SerdWriter*        writer,
     sink(serd_node_string(lang), lang->length, writer);
   } else if (type_uri) {
     sink("^^", 2, writer);
-    return write_node(writer, datatype, NULL, NULL, FIELD_NONE, flags);
+    return write_node(writer, datatype, FIELD_NONE, flags);
   }
   return true;
 }
@@ -713,17 +711,15 @@ write_blank(SerdWriter* const        writer,
 }
 
 static bool
-write_node(SerdWriter*        writer,
-           const SerdNode*    node,
-           const SerdNode*    datatype,
-           const SerdNode*    lang,
-           Field              field,
-           SerdStatementFlags flags)
+write_node(SerdWriter* const        writer,
+           const SerdNode* const    node,
+           const Field              field,
+           const SerdStatementFlags flags)
 {
   bool ret = false;
   switch (node->type) {
   case SERD_LITERAL:
-    ret = write_literal(writer, node, datatype, lang, flags);
+    ret = write_literal(writer, node, flags);
     break;
   case SERD_URI:
     ret = write_uri_node(writer, node, field);
@@ -748,18 +744,16 @@ is_resource(const SerdNode* node)
 static void
 write_pred(SerdWriter* writer, SerdStatementFlags flags, const SerdNode* pred)
 {
-  write_node(writer, pred, NULL, NULL, FIELD_PREDICATE, flags);
+  write_node(writer, pred, FIELD_PREDICATE, flags);
   write_sep(writer, SEP_P_O);
   serd_node_set(&writer->context.predicate, pred);
 }
 
 static bool
-write_list_obj(SerdWriter*        writer,
-               SerdStatementFlags flags,
-               const SerdNode*    predicate,
-               const SerdNode*    object,
-               const SerdNode*    datatype,
-               const SerdNode*    lang)
+write_list_obj(SerdWriter* const        writer,
+               const SerdStatementFlags flags,
+               const SerdNode* const    predicate,
+               const SerdNode* const    object)
 {
   if (!strcmp(serd_node_string(object), NS_RDF "nil")) {
     --writer->indent;
@@ -769,21 +763,19 @@ write_list_obj(SerdWriter*        writer,
 
   if (!strcmp(serd_node_string(predicate), NS_RDF "first")) {
     write_sep(writer, SEP_LIST_SEP);
-    write_node(writer, object, datatype, lang, FIELD_OBJECT, flags);
+    write_node(writer, object, FIELD_OBJECT, flags);
   }
 
   return false;
 }
 
 SerdStatus
-serd_writer_write_statement(SerdWriter*        writer,
-                            SerdStatementFlags flags,
-                            const SerdNode*    graph,
-                            const SerdNode*    subject,
-                            const SerdNode*    predicate,
-                            const SerdNode*    object,
-                            const SerdNode*    datatype,
-                            const SerdNode*    lang)
+serd_writer_write_statement(SerdWriter* const        writer,
+                            const SerdStatementFlags flags,
+                            const SerdNode* const    graph,
+                            const SerdNode* const    subject,
+                            const SerdNode* const    predicate,
+                            const SerdNode* const    object)
 {
   if (!is_resource(subject) || !is_resource(predicate) || !object) {
     return SERD_ERR_BAD_ARG;
@@ -797,14 +789,14 @@ serd_writer_write_statement(SerdWriter*        writer,
   } while (0)
 
   if (writer->syntax == SERD_NTRIPLES || writer->syntax == SERD_NQUADS) {
-    TRY(write_node(writer, subject, NULL, NULL, FIELD_SUBJECT, flags));
+    TRY(write_node(writer, subject, FIELD_SUBJECT, flags));
     sink(" ", 1, writer);
-    TRY(write_node(writer, predicate, NULL, NULL, FIELD_PREDICATE, flags));
+    TRY(write_node(writer, predicate, FIELD_PREDICATE, flags));
     sink(" ", 1, writer);
-    TRY(write_node(writer, object, datatype, lang, FIELD_OBJECT, flags));
+    TRY(write_node(writer, object, FIELD_OBJECT, flags));
     if (writer->syntax == SERD_NQUADS && graph) {
       sink(" ", 1, writer);
-      TRY(write_node(writer, graph, datatype, lang, FIELD_GRAPH, flags));
+      TRY(write_node(writer, graph, FIELD_GRAPH, flags));
     }
     sink(" .\n", 3, writer);
     return SERD_SUCCESS;
@@ -822,7 +814,7 @@ serd_writer_write_statement(SerdWriter*        writer,
 
     reset_context(writer, true);
     if (graph) {
-      TRY(write_node(writer, graph, datatype, lang, FIELD_GRAPH, flags));
+      TRY(write_node(writer, graph, FIELD_GRAPH, flags));
       ++writer->indent;
       write_sep(writer, SEP_GRAPH_BEGIN);
       serd_node_set(&writer->context.graph, graph);
@@ -830,7 +822,7 @@ serd_writer_write_statement(SerdWriter*        writer,
   }
 
   if ((flags & SERD_LIST_CONT)) {
-    if (write_list_obj(writer, flags, predicate, object, datatype, lang)) {
+    if (write_list_obj(writer, flags, predicate, object)) {
       // Reached end of list
       if (--writer->list_depth == 0 && writer->list_subj) {
         reset_context(writer, false);
@@ -847,7 +839,7 @@ serd_writer_write_statement(SerdWriter*        writer,
         ++writer->indent;
       }
       write_sep(writer, SEP_END_O);
-      write_node(writer, object, datatype, lang, FIELD_OBJECT, flags);
+      write_node(writer, object, FIELD_OBJECT, flags);
       if (!(flags & SERD_ANON_O_BEGIN)) {
         --writer->indent;
       }
@@ -856,7 +848,7 @@ serd_writer_write_statement(SerdWriter*        writer,
       Sep sep = ctx(writer, FIELD_PREDICATE) ? SEP_END_P : SEP_S_P;
       write_sep(writer, sep);
       write_pred(writer, flags, predicate);
-      write_node(writer, object, datatype, lang, FIELD_OBJECT, flags);
+      write_node(writer, object, FIELD_OBJECT, flags);
     }
   } else {
     // No abbreviation
@@ -871,7 +863,7 @@ serd_writer_write_statement(SerdWriter*        writer,
     }
 
     if (!(flags & SERD_ANON_CONT)) {
-      write_node(writer, subject, NULL, NULL, FIELD_SUBJECT, flags);
+      write_node(writer, subject, FIELD_SUBJECT, flags);
       ++writer->indent;
       write_sep(writer, SEP_S_P);
     } else {
@@ -885,7 +877,7 @@ serd_writer_write_statement(SerdWriter*        writer,
       write_pred(writer, flags, predicate);
     }
 
-    write_node(writer, object, datatype, lang, FIELD_OBJECT, flags);
+    write_node(writer, object, FIELD_OBJECT, flags);
   }
 
   if (flags & (SERD_ANON_S_BEGIN | SERD_ANON_O_BEGIN)) {