aboutsummaryrefslogtreecommitdiffstats
path: root/src
diff options
context:
space:
mode:
Diffstat (limited to 'src')
-rw-r--r--src/n3.c5
-rw-r--r--src/node.c141
-rw-r--r--src/nodes.c19
-rw-r--r--src/read_ntriples.c22
-rw-r--r--src/string.c51
-rw-r--r--src/string_utils.h5
-rw-r--r--src/writer.c3
7 files changed, 88 insertions, 158 deletions
diff --git a/src/n3.c b/src/n3.c
index 5940c010..6813bccc 100644
--- a/src/n3.c
+++ b/src/n3.c
@@ -114,7 +114,7 @@ read_STRING_LITERAL_LONG(SerdReader* const reader,
eat_byte_safe(reader, q3);
break;
}
- ref->flags |= SERD_HAS_QUOTE;
+
if (!(st = push_byte(reader, ref, c))) {
st = read_character(reader, ref, (uint8_t)q2);
}
@@ -151,7 +151,10 @@ read_String(SerdReader* const reader, SerdNode* const node)
return SERD_SUCCESS;
}
+ // Long string
eat_byte_safe(reader, q3);
+ node->flags |= SERD_IS_LONG;
+
return read_STRING_LITERAL_LONG(reader, node, (uint8_t)q1);
}
diff --git a/src/node.c b/src/node.c
index d9cfbe93..9468325b 100644
--- a/src/node.c
+++ b/src/node.c
@@ -173,8 +173,8 @@ serd_new_simple_node(const SerdNodeType type, const SerdStringView str)
return NULL;
}
- SerdNodeFlags flags = 0;
- const size_t length = str.buf ? serd_strlen(str.buf, &flags) : 0;
+ SerdNodeFlags flags = 0u;
+ const size_t length = str.buf ? str.len : 0u;
SerdNode* node = serd_node_malloc(length, flags, type);
if (node) {
@@ -193,87 +193,93 @@ serd_new_simple_node(const SerdNodeType type, const SerdStringView str)
SerdNode*
serd_new_string(const SerdStringView str)
{
- SerdNodeFlags flags = 0;
- const size_t length = serd_substrlen(str.buf, str.len, &flags);
- SerdNode* node = serd_node_malloc(length, flags, SERD_LITERAL);
-
- memcpy(serd_node_buffer(node), str.buf, str.len);
- node->length = length;
-
- serd_node_check_padding(node);
- return node;
-}
-
-/// Internal pre-measured implementation of serd_new_plain_literal
-static SerdNode*
-serd_new_plain_literal_i(const SerdStringView str,
- SerdNodeFlags flags,
- const SerdStringView lang)
-{
- assert(str.len);
- assert(lang.len);
-
- flags |= SERD_HAS_LANGUAGE;
-
- const size_t len = serd_node_pad_length(str.len);
- const size_t total_len = len + sizeof(SerdNode) + lang.len;
+ SerdNodeFlags flags = 0u;
+ SerdNode* node = serd_node_malloc(str.len, flags, SERD_LITERAL);
- SerdNode* node = serd_node_malloc(total_len, flags, SERD_LITERAL);
- memcpy(serd_node_buffer(node), str.buf, str.len);
- node->length = str.len;
+ if (node) {
+ if (str.buf && str.len) {
+ memcpy(serd_node_buffer(node), str.buf, str.len);
+ }
- 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.buf, lang.len);
- serd_node_check_padding(lang_node);
+ node->length = str.len;
+ serd_node_check_padding(node);
+ }
- serd_node_check_padding(node);
return node;
}
-SerdNode*
-serd_new_plain_literal(const SerdStringView str, const SerdStringView lang)
+SERD_PURE_FUNC
+static bool
+is_langtag(const SerdStringView string)
{
- if (!lang.len) {
- return serd_new_string(str);
+ // First character must be a letter
+ size_t i = 0;
+ if (!string.len || !is_alpha(string.buf[i])) {
+ return false;
}
- SerdNodeFlags flags = 0;
- serd_strlen(str.buf, &flags);
+ // First component must be all letters
+ while (++i < string.len && string.buf[i] && string.buf[i] != '-') {
+ if (!is_alpha(string.buf[i])) {
+ return false;
+ }
+ }
+
+ // Following components can have letters and digits
+ while (i < string.len && string.buf[i] == '-') {
+ while (++i < string.len && string.buf[i] && string.buf[i] != '-') {
+ const char c = string.buf[i];
+ if (!is_alpha(c) && !is_digit(c)) {
+ return false;
+ }
+ }
+ }
- return serd_new_plain_literal_i(str, flags, lang);
+ return true;
}
SerdNode*
-serd_new_typed_literal(const SerdStringView str,
- const SerdStringView datatype_uri)
+serd_new_literal(const SerdStringView string,
+ const SerdNodeFlags flags,
+ const SerdStringView meta)
{
- if (!datatype_uri.len) {
- return serd_new_string(str);
+ if (!(flags & (SERD_HAS_DATATYPE | SERD_HAS_LANGUAGE))) {
+ SerdNode* node = serd_node_malloc(string.len, flags, SERD_LITERAL);
+
+ memcpy(serd_node_buffer(node), string.buf, string.len);
+ node->length = string.len;
+ serd_node_check_padding(node);
+ return node;
}
- if (!strcmp(datatype_uri.buf, NS_RDF "langString")) {
+ if ((flags & SERD_HAS_DATATYPE) && (flags & SERD_HAS_LANGUAGE)) {
return NULL;
}
- SerdNodeFlags flags = 0u;
- serd_strlen(str.buf, &flags);
+ if (!meta.len) {
+ return NULL;
+ }
- flags |= SERD_HAS_DATATYPE;
+ if (((flags & SERD_HAS_DATATYPE) &&
+ (!serd_uri_string_has_scheme(meta.buf) ||
+ !strcmp(meta.buf, NS_RDF "langString"))) ||
+ ((flags & SERD_HAS_LANGUAGE) && !is_langtag(meta))) {
+ return NULL;
+ }
- const size_t len = serd_node_pad_length(str.len);
- const size_t total_len = len + sizeof(SerdNode) + datatype_uri.len;
+ const size_t len = serd_node_pad_length(string.len);
+ const size_t meta_len = serd_node_pad_length(meta.len);
+ const size_t meta_size = sizeof(SerdNode) + meta_len;
- SerdNode* node = serd_node_malloc(total_len, flags, SERD_LITERAL);
- memcpy(serd_node_buffer(node), str.buf, str.len);
- node->length = str.len;
+ SerdNode* node = serd_node_malloc(len + meta_size, flags, SERD_LITERAL);
+ memcpy(serd_node_buffer(node), string.buf, string.len);
+ node->length = string.len;
- SerdNode* datatype_node = node + 1 + (len / sizeof(SerdNode));
- datatype_node->length = datatype_uri.len;
- datatype_node->type = SERD_URI;
- memcpy(serd_node_buffer(datatype_node), datatype_uri.buf, datatype_uri.len);
- serd_node_check_padding(datatype_node);
+ SerdNode* meta_node = node + 1u + (len / sizeof(SerdNode));
+ meta_node->length = meta.len;
+ meta_node->type = (flags & SERD_HAS_DATATYPE) ? SERD_URI : SERD_LITERAL;
+ memcpy(serd_node_buffer(meta_node), meta.buf, meta.len);
+ serd_node_check_padding(meta_node);
serd_node_check_padding(node);
return node;
@@ -564,8 +570,9 @@ serd_new_double(const double d)
const ExessResult r = exess_write_double(d, sizeof(buf), buf);
return r.status ? NULL
- : serd_new_typed_literal(SERD_SUBSTRING(buf, r.count),
- SERD_STRING(EXESS_XSD_URI "double"));
+ : serd_new_literal(SERD_SUBSTRING(buf, r.count),
+ SERD_HAS_DATATYPE,
+ SERD_STRING(EXESS_XSD_URI "double"));
}
SerdNode*
@@ -576,15 +583,17 @@ serd_new_float(const float f)
const ExessResult r = exess_write_float(f, sizeof(buf), buf);
return r.status ? NULL
- : serd_new_typed_literal(SERD_SUBSTRING(buf, r.count),
- SERD_STRING(EXESS_XSD_URI "float"));
+ : serd_new_literal(SERD_SUBSTRING(buf, r.count),
+ SERD_HAS_DATATYPE,
+ SERD_STRING(EXESS_XSD_URI "float"));
}
SerdNode*
serd_new_boolean(bool b)
{
- return serd_new_typed_literal(b ? SERD_STRING("true") : SERD_STRING("false"),
- serd_node_string_view(&serd_xsd_boolean.node));
+ return serd_new_literal(b ? SERD_STRING("true") : SERD_STRING("false"),
+ SERD_HAS_DATATYPE,
+ serd_node_string_view(&serd_xsd_boolean.node));
}
SerdNode*
diff --git a/src/nodes.c b/src/nodes.c
index 412d0d24..51f354bb 100644
--- a/src/nodes.c
+++ b/src/nodes.c
@@ -156,20 +156,13 @@ serd_nodes_string(SerdNodes* const nodes, const SerdStringView string)
return serd_nodes_manage(nodes, serd_new_string(string));
}
-const SerdNode*
-serd_nodes_plain_literal(SerdNodes* const nodes,
- const SerdStringView string,
- const SerdStringView language)
-{
- return serd_nodes_manage(nodes, serd_new_plain_literal(string, language));
-}
-
-const SerdNode*
-serd_nodes_typed_literal(SerdNodes* const nodes,
- const SerdStringView string,
- const SerdStringView datatype_uri)
+const SerdNode* SERD_ALLOCATED
+serd_nodes_literal(SerdNodes* const nodes,
+ const SerdStringView string,
+ const SerdNodeFlags flags,
+ const SerdStringView meta)
{
- return serd_nodes_manage(nodes, serd_new_typed_literal(string, datatype_uri));
+ return serd_nodes_manage(nodes, serd_new_literal(string, flags, meta));
}
const SerdNode*
diff --git a/src/read_ntriples.c b/src/read_ntriples.c
index 78e46634..017c4dcf 100644
--- a/src/read_ntriples.c
+++ b/src/read_ntriples.c
@@ -206,24 +206,8 @@ read_IRI(SerdReader* const reader, SerdNode** const dest)
SerdStatus
read_character(SerdReader* const reader, SerdNode* const dest, const uint8_t c)
{
- if (!(c & 0x80)) {
- switch (c) {
- case 0xA:
- case 0xD:
- dest->flags |= SERD_HAS_NEWLINE;
- break;
- case '"':
- case '\'':
- dest->flags |= SERD_HAS_QUOTE;
- break;
- default:
- break;
- }
-
- return push_byte(reader, dest, c);
- }
-
- return read_utf8_continuation(reader, dest, c);
+ return !(c & 0x80) ? push_byte(reader, dest, c)
+ : read_utf8_continuation(reader, dest, c);
}
/// [9] STRING_LITERAL_QUOTE
@@ -432,11 +416,9 @@ read_ECHAR(SerdReader* const reader, SerdNode* const dest)
eat_byte_safe(reader, 'b');
return push_byte(reader, dest, '\b');
case 'n':
- dest->flags |= SERD_HAS_NEWLINE;
eat_byte_safe(reader, 'n');
return push_byte(reader, dest, '\n');
case 'r':
- dest->flags |= SERD_HAS_NEWLINE;
eat_byte_safe(reader, 'r');
return push_byte(reader, dest, '\r');
case 'f':
diff --git a/src/string.c b/src/string.c
index 97c1432b..2d25fb80 100644
--- a/src/string.c
+++ b/src/string.c
@@ -14,13 +14,9 @@
OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
*/
-#include "string_utils.h"
-
#include "serd/serd.h"
-#include <assert.h>
#include <stdlib.h>
-#include <string.h>
void
serd_free(void* const ptr)
@@ -66,50 +62,3 @@ serd_strerror(const SerdStatus status)
return "Unknown error";
}
-
-static void
-serd_update_flags(const char c, SerdNodeFlags* const flags)
-{
- switch (c) {
- case '\r':
- case '\n':
- *flags |= SERD_HAS_NEWLINE;
- break;
- case '"':
- *flags |= SERD_HAS_QUOTE;
- break;
- default:
- break;
- }
-}
-
-size_t
-serd_substrlen(const char* const str,
- const size_t len,
- SerdNodeFlags* const flags)
-{
- assert(flags);
-
- size_t i = 0;
- *flags = 0;
- for (; i < len && str[i]; ++i) {
- serd_update_flags(str[i], flags);
- }
-
- return i;
-}
-
-size_t
-serd_strlen(const char* const str, SerdNodeFlags* const flags)
-{
- if (flags) {
- size_t i = 0;
- *flags = 0;
- for (; str[i]; ++i) {
- serd_update_flags(str[i], flags);
- }
- return i;
- }
-
- return strlen(str);
-}
diff --git a/src/string_utils.h b/src/string_utils.h
index a411b90d..54f7877c 100644
--- a/src/string_utils.h
+++ b/src/string_utils.h
@@ -17,8 +17,6 @@
#ifndef SERD_STRING_UTILS_H
#define SERD_STRING_UTILS_H
-#include "serd/serd.h"
-
#include <stdbool.h>
#include <stddef.h>
#include <stdint.h>
@@ -90,9 +88,6 @@ is_windows_path(const char* path)
(path[2] == '/' || path[2] == '\\');
}
-size_t
-serd_substrlen(const char* str, size_t len, SerdNodeFlags* flags);
-
static inline char
serd_to_upper(const char c)
{
diff --git a/src/writer.c b/src/writer.c
index c6a91d78..5416d144 100644
--- a/src/writer.c
+++ b/src/writer.c
@@ -737,8 +737,7 @@ write_literal(SerdWriter* const writer,
}
SerdStatus st = SERD_SUCCESS;
- if (supports_abbrev(writer) &&
- (node->flags & (SERD_HAS_NEWLINE | SERD_HAS_QUOTE))) {
+ if (supports_abbrev(writer) && (node->flags & SERD_IS_LONG)) {
TRY(st, esink("\"\"\"", 3, writer));
TRY(st, write_text(writer, WRITE_LONG_STRING, node_str, node->length));
TRY(st, esink("\"\"\"", 3, writer));
generated by cgit v1.2.1 (git 2.18.0) at 2024-07-13 17:08:29 +0000