aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
Diffstat
-rw-r--r--NEWS1
-rw-r--r--serd/serd.h8
-rw-r--r--src/env.c5
-rw-r--r--src/node.c34
-rw-r--r--src/reader.c4
-rw-r--r--src/reader.h3
-rw-r--r--src/serd_internal.h1
-rw-r--r--src/string.c45
-rw-r--r--src/writer.c7
-rw-r--r--tests/serd_test.c49
10 files changed, 58 insertions, 99 deletions
diff --git a/NEWS b/NEWS
index e1009660..446387d6 100644
--- a/NEWS
+++ b/NEWS
@@ -3,6 +3,7 @@ serd (1.0.0) unstable;
* Use SerdBuffer for mutable buffers instead of abusing SerdChunk
* Add serd_node_new_relative_uri()
* Fix construction and comparison of URIs with UTF-8 characters
+ * Remove useless character counting
* Report I/O errors with message and return appropriate status code
* Report missing statement separator errors
* Fix potential out of bounds read
diff --git a/serd/serd.h b/serd/serd.h
index 49ee3c3a..d289ee72 100644
--- a/serd/serd.h
+++ b/serd/serd.h
@@ -217,7 +217,6 @@ typedef uint32_t SerdNodeFlags;
typedef struct {
const uint8_t* buf; /**< Value string */
size_t n_bytes; /**< Size in bytes (not including null) */
- size_t n_chars; /**< Length in characters (not including null)*/
SerdNodeFlags flags; /**< Node flags (e.g. string properties) */
SerdType type; /**< Node type */
} SerdNode;
@@ -296,14 +295,13 @@ serd_strerror(SerdStatus status);
/**
Measure a UTF-8 string.
- @return Length of `str` in characters (except NULL).
+ @return Length of `str` in bytes.
@param str A null-terminated UTF-8 string.
- @param n_bytes (Output) Set to the size of `str` in bytes (except NULL).
@param flags (Output) Set to the applicable flags.
*/
SERD_API
size_t
-serd_strlen(const uint8_t* str, size_t* n_bytes, SerdNodeFlags* flags);
+serd_strlen(const uint8_t* str, SerdNodeFlags* flags);
/**
Parse a string to a double.
@@ -445,7 +443,7 @@ serd_uri_serialise_relative(const SerdURI* uri,
@{
*/
-static const SerdNode SERD_NODE_NULL = { NULL, 0, 0, 0, SERD_NOTHING };
+static const SerdNode SERD_NODE_NULL = { NULL, 0, 0, SERD_NOTHING };
/**
Make a (shallow) node from `str`.
diff --git a/src/env.c b/src/env.c
index d969d3df..cf1c3538 100644
--- a/src/env.c
+++ b/src/env.c
@@ -245,9 +245,8 @@ serd_env_expand_node(const SerdEnv* env,
}
const size_t len = prefix.len + suffix.len;
uint8_t* buf = (uint8_t*)malloc(len + 1);
- SerdNode ret = { buf, len, 0, 0, SERD_URI };
- snprintf((char*)buf, len + 1, "%s%s", prefix.buf, suffix.buf);
- ret.n_chars = serd_strlen(buf, NULL, NULL);
+ SerdNode ret = { buf, len, 0, SERD_URI };
+ snprintf((char*)buf, ret.n_bytes + 1, "%s%s", prefix.buf, suffix.buf);
return ret;
}
case SERD_URI: {
diff --git a/src/node.c b/src/node.c
index 1472ad13..1ea6bd5b 100644
--- a/src/node.c
+++ b/src/node.c
@@ -39,10 +39,9 @@ serd_node_from_string(SerdType type, const uint8_t* str)
return SERD_NODE_NULL;
}
- uint32_t flags = 0;
- size_t buf_n_bytes = 0;
- const size_t buf_n_chars = serd_strlen(str, &buf_n_bytes, &flags);
- SerdNode ret = { str, buf_n_bytes, buf_n_chars, flags, type };
+ uint32_t flags = 0;
+ const size_t n_bytes = serd_strlen(str, &flags);
+ const SerdNode ret = { str, n_bytes, flags, type };
return ret;
}
@@ -54,11 +53,9 @@ serd_node_from_substring(SerdType type, const uint8_t* str, const size_t len)
return SERD_NODE_NULL;
}
- uint32_t flags = 0;
- size_t buf_n_bytes = 0;
- const size_t buf_n_chars = serd_substrlen(str, len, &buf_n_bytes, &flags);
- assert(buf_n_bytes <= len);
- SerdNode ret = { str, buf_n_bytes, buf_n_chars, flags, type };
+ uint32_t flags = 0;
+ const size_t n_bytes = serd_substrlen(str, len, &flags);
+ const SerdNode ret = { str, n_bytes, flags, type };
return ret;
}
@@ -84,7 +81,6 @@ serd_node_equals(const SerdNode* a, const SerdNode* b)
return (a == b)
|| (a->type == b->type
&& a->n_bytes == b->n_bytes
- && a->n_chars == b->n_chars
&& ((a->buf == b->buf) || !memcmp((const char*)a->buf,
(const char*)b->buf,
a->n_bytes + 1)));
@@ -218,13 +214,12 @@ serd_node_new_uri(const SerdURI* uri, const SerdURI* base, SerdURI* out)
const size_t len = serd_uri_string_length(&abs_uri);
uint8_t* buf = (uint8_t*)malloc(len + 1);
- SerdNode node = { buf, 0, 0, 0, SERD_URI };
+ SerdNode node = { buf, len, 0, SERD_URI };
uint8_t* ptr = buf;
const size_t actual_len = serd_uri_serialise(&abs_uri, string_sink, &ptr);
buf[actual_len] = '\0';
node.n_bytes = actual_len;
- node.n_chars = serd_strlen(buf, NULL, NULL);
if (out) {
serd_uri_parse(buf, out); // TODO: cleverly avoid double parse
@@ -243,14 +238,13 @@ serd_node_new_relative_uri(const SerdURI* uri,
const size_t uri_len = serd_uri_string_length(uri);
const size_t base_len = serd_uri_string_length(base);
uint8_t* buf = (uint8_t*)malloc(uri_len + base_len + 1);
- SerdNode node = { buf, 0, 0, 0, SERD_URI };
+ SerdNode node = { buf, 0, 0, SERD_URI };
uint8_t* ptr = buf;
const size_t actual_len = serd_uri_serialise_relative(
uri, base, root, string_sink, &ptr);
buf[actual_len] = '\0';
node.n_bytes = actual_len;
- node.n_chars = serd_strlen(buf, NULL, NULL);
if (out) {
serd_uri_parse(buf, out); // TODO: cleverly avoid double parse
@@ -277,7 +271,7 @@ serd_node_new_decimal(double d, unsigned frac_digits)
const double abs_d = fabs(d);
const unsigned int_digits = serd_digits(abs_d);
char* buf = (char*)calloc(int_digits + frac_digits + 3, 1);
- SerdNode node = { (const uint8_t*)buf, 0, 0, 0, SERD_LITERAL };
+ SerdNode node = { (const uint8_t*)buf, 0, 0, SERD_LITERAL };
const double int_part = floor(abs_d);
// Point s to decimal point location
@@ -300,7 +294,7 @@ serd_node_new_decimal(double d, unsigned frac_digits)
double frac_part = fabs(d - int_part);
if (frac_part < DBL_EPSILON) {
*s++ = '0';
- node.n_bytes = node.n_chars = (s - buf);
+ node.n_bytes = (s - buf);
} else {
uint64_t frac = frac_part * pow(10.0, (int)frac_digits) + 0.5;
s += frac_digits - 1;
@@ -309,7 +303,7 @@ serd_node_new_decimal(double d, unsigned frac_digits)
// Skip trailing zeros
for (; i < frac_digits - 1 && !(frac % 10); ++i, --s, frac /= 10) {}
- node.n_bytes = node.n_chars = (s - buf) + 1;
+ node.n_bytes = (s - buf) + 1;
// Write digits from last trailing zero to decimal point
for (; i < frac_digits; ++i) {
@@ -328,7 +322,7 @@ serd_node_new_integer(int64_t i)
int64_t abs_i = (i < 0) ? -i : i;
const unsigned digits = serd_digits(abs_i);
char* buf = (char*)calloc(digits + 2, 1);
- SerdNode node = { (const uint8_t*)buf, 0, 0, 0, SERD_LITERAL };
+ SerdNode node = { (const uint8_t*)buf, 0, 0, SERD_LITERAL };
// Point s to the end
char* s = buf + digits - 1;
@@ -337,7 +331,7 @@ serd_node_new_integer(int64_t i)
++s;
}
- node.n_bytes = node.n_chars = (s - buf) + 1;
+ node.n_bytes = (s - buf) + 1;
// Write integer part (right to left)
do {
@@ -374,7 +368,7 @@ serd_node_new_blob(const void* buf, size_t size, bool wrap_lines)
{
const size_t len = ((size + 2) / 3) * 4 + (wrap_lines ? (size / 57) : 0);
uint8_t* str = (uint8_t*)calloc(1, len + 2);
- SerdNode node = { str, len, len, 0, SERD_LITERAL };
+ SerdNode node = { str, len, 0, SERD_LITERAL };
for (size_t i = 0, j = 0; i < size; i += 3, j += 4) {
uint8_t in[4] = { 0, 0, 0, 0 };
size_t n_in = MIN(3, size - i);
diff --git a/src/reader.c b/src/reader.c
index d515aaab..4266e128 100644
--- a/src/reader.c
+++ b/src/reader.c
@@ -41,7 +41,7 @@ set_blank_id(SerdReader* reader, Ref ref, size_t buf_size)
{
SerdNode* node = deref(reader, ref);
const char* prefix = reader->bprefix ? (const char*)reader->bprefix : "";
- node->n_bytes = node->n_chars = snprintf(
+ node->n_bytes = snprintf(
(char*)node->buf, buf_size, "%sb%u", prefix, reader->next_id++);
}
@@ -80,7 +80,7 @@ push_node_padded(SerdReader* reader, size_t maxlen,
&reader->stack, sizeof(SerdNode) + maxlen + 1, sizeof(SerdNode));
SerdNode* const node = (SerdNode*)mem;
- node->n_bytes = node->n_chars = n_bytes;
+ node->n_bytes = n_bytes;
node->flags = 0;
node->type = type;
node->buf = NULL;
diff --git a/src/reader.h b/src/reader.h
index 723a1c5a..7dc981af 100644
--- a/src/reader.h
+++ b/src/reader.h
@@ -69,9 +69,6 @@ push_byte(SerdReader* reader, Ref ref, const uint8_t c)
uint8_t* const s = serd_stack_push(&reader->stack, 1);
SerdNode* const node = (SerdNode*)(reader->stack.buf + ref);
++node->n_bytes;
- if (!(c & 0x80)) { // Starts with 0 bit, start of new character
- ++node->n_chars;
- }
*(s - 1) = c;
*s = '\0';
return SERD_SUCCESS;
diff --git a/src/serd_internal.h b/src/serd_internal.h
index cfcdc82c..46b88c83 100644
--- a/src/serd_internal.h
+++ b/src/serd_internal.h
@@ -353,7 +353,6 @@ is_windows_path(const uint8_t* path)
size_t
serd_substrlen(const uint8_t* str,
const size_t len,
- size_t* n_bytes,
SerdNodeFlags* flags);
static inline int
diff --git a/src/string.c b/src/string.c
index dedd0713..ee2ac290 100644
--- a/src/string.c
+++ b/src/string.c
@@ -51,47 +51,32 @@ serd_update_flags(const uint8_t c, SerdNodeFlags* const flags)
size_t
serd_substrlen(const uint8_t* const str,
const size_t len,
- size_t* const n_bytes,
SerdNodeFlags* const flags)
{
- size_t n_chars = 0;
- size_t i = 0;
- SerdNodeFlags f = 0;
- for (; i < len && str[i]; ++i) {
- if ((str[i] & 0xC0) != 0x80) { // Start of new character
- ++n_chars;
- serd_update_flags(str[i], &f);
- }
- }
- if (n_bytes) {
- *n_bytes = i;
- }
if (flags) {
- *flags = f;
+ size_t i = 0;
+ *flags = 0;
+ for (; i < len && str[i]; ++i) {
+ serd_update_flags(str[i], flags);
+ }
+ return i;
}
- return n_chars;
+ return strlen((const char*)str);
}
SERD_API
size_t
-serd_strlen(const uint8_t* str, size_t* n_bytes, SerdNodeFlags* flags)
+serd_strlen(const uint8_t* str, SerdNodeFlags* flags)
{
- size_t n_chars = 0;
- size_t i = 0;
- SerdNodeFlags f = 0;
- for (; str[i]; ++i) {
- if ((str[i] & 0xC0) != 0x80) { // Start of new character
- ++n_chars;
- serd_update_flags(str[i], &f);
- }
- }
- if (n_bytes) {
- *n_bytes = i;
- }
if (flags) {
- *flags = f;
+ size_t i = 0;
+ *flags = 0;
+ for (; str[i]; ++i) {
+ serd_update_flags(str[i], flags);
+ }
+ return i;
}
- return n_chars;
+ return strlen((const char*)str);
}
static inline double
diff --git a/src/writer.c b/src/writer.c
index 26bf5e5a..cecaacf5 100644
--- a/src/writer.c
+++ b/src/writer.c
@@ -28,9 +28,9 @@ typedef struct {
} WriteContext;
static const WriteContext WRITE_CONTEXT_NULL = {
- { 0, 0, 0, 0, SERD_NOTHING },
- { 0, 0, 0, 0, SERD_NOTHING },
- { 0, 0, 0, 0, SERD_NOTHING }
+ { 0, 0, 0, SERD_NOTHING },
+ { 0, 0, 0, SERD_NOTHING },
+ { 0, 0, 0, SERD_NOTHING }
};
typedef enum {
@@ -148,7 +148,6 @@ copy_node(SerdNode* dst, const SerdNode* src)
if (src) {
dst->buf = (uint8_t*)realloc((char*)dst->buf, src->n_bytes + 1);
dst->n_bytes = src->n_bytes;
- dst->n_chars = src->n_chars;
dst->flags = src->flags;
dst->type = src->type;
memcpy((char*)dst->buf, src->buf, src->n_bytes + 1);
diff --git a/tests/serd_test.c b/tests/serd_test.c
index 62621e16..71801c1b 100644
--- a/tests/serd_test.c
+++ b/tests/serd_test.c
@@ -163,9 +163,9 @@ main(void)
node.buf, dbl_test_strs[i]);
}
const size_t len = node.buf ? strlen((const char*)node.buf) : 0;
- if (node.n_bytes != len || node.n_chars != len) {
- FAILF("Length %zu,%zu != %zu\n",
- node.n_bytes, node.n_chars, len);
+ if (node.n_bytes != len) {
+ FAILF("Length %zu != %zu\n",
+ node.n_bytes, len);
}
serd_node_free(&node);
}
@@ -187,9 +187,9 @@ main(void)
node.buf, int_test_strs[i]);
}
const size_t len = strlen((const char*)node.buf);
- if (node.n_bytes != len || node.n_chars != len) {
+ if (node.n_bytes != len) {
FAILF("Length %zu,%zu != %zu\n",
- node.n_bytes, node.n_chars, len);
+ node.n_bytes, len);
}
serd_node_free(&node);
}
@@ -202,12 +202,6 @@ main(void)
}
SerdNode blob = serd_node_new_blob(data, size, size % 5);
-
- if (blob.n_bytes != blob.n_chars) {
- FAILF("Blob %zu bytes != %zu chars\n",
- blob.n_bytes, blob.n_chars);
- }
-
size_t out_size;
uint8_t* out = (uint8_t*)serd_base64_decode(
blob.buf, blob.n_bytes, &out_size);
@@ -230,18 +224,11 @@ main(void)
const uint8_t str[] = { '"', '5', 0xE2, 0x82, 0xAC, '"', '\n', 0 };
- size_t n_bytes;
SerdNodeFlags flags;
- size_t len = serd_strlen(str, &n_bytes, &flags);
- if (len != 5 || n_bytes != 7
- || flags != (SERD_HAS_QUOTE|SERD_HAS_NEWLINE)) {
- FAILF("Bad serd_strlen(%s) len=%zu n_bytes=%zu flags=%u\n",
- str, len, n_bytes, flags);
- }
- len = serd_strlen(str, NULL, &flags);
- if (len != 5) {
- FAILF("Bad serd_strlen(%s) len=%zu flags=%u\n",
- str, len, flags);
+ size_t n_bytes = serd_strlen(str, &flags);
+ if (n_bytes != 7 || flags != (SERD_HAS_QUOTE|SERD_HAS_NEWLINE)) {
+ FAILF("Bad serd_strlen(%s) n_bytes=%zu flags=%u\n",
+ str, n_bytes, flags);
}
// Test serd_strerror
@@ -346,10 +333,10 @@ main(void)
// Test serd_node_from_string
SerdNode node = serd_node_from_string(SERD_LITERAL, (const uint8_t*)"hello\"");
- if (node.n_bytes != 6 || node.n_chars != 6 || node.flags != SERD_HAS_QUOTE
+ if (node.n_bytes != 6 || node.flags != SERD_HAS_QUOTE
|| strcmp((const char*)node.buf, "hello\"")) {
- FAILF("Bad node %s %zu %zu %d %d\n",
- node.buf, node.n_bytes, node.n_chars, node.flags, node.type);
+ FAILF("Bad node %s %zu %d %d\n",
+ node.buf, node.n_bytes, node.flags, node.type);
}
node = serd_node_from_string(SERD_URI, NULL);
@@ -360,22 +347,22 @@ main(void)
// Test serd_node_from_substring
SerdNode empty = serd_node_from_substring(SERD_LITERAL, NULL, 32);
- if (empty.buf || empty.n_bytes || empty.n_chars || empty.flags || empty.type) {
+ if (empty.buf || empty.n_bytes || empty.flags || empty.type) {
FAIL("Successfully created node from NULL substring\n");
}
SerdNode a_b = serd_node_from_substring(SERD_LITERAL, USTR("a\"bc"), 3);
- if (a_b.n_bytes != 3 || a_b.n_chars != 3 || a_b.flags != SERD_HAS_QUOTE
+ if (a_b.n_bytes != 3 || a_b.flags != SERD_HAS_QUOTE
|| strncmp((const char*)a_b.buf, "a\"b", 3)) {
- FAILF("Bad node %s %zu %zu %d %d\n",
- a_b.buf, a_b.n_bytes, a_b.n_chars, a_b.flags, a_b.type);
+ FAILF("Bad node %s %zu %d %d\n",
+ a_b.buf, a_b.n_bytes, a_b.flags, a_b.type);
}
a_b = serd_node_from_substring(SERD_LITERAL, USTR("a\"bc"), 10);
- if (a_b.n_bytes != 4 || a_b.n_chars != 4 || a_b.flags != SERD_HAS_QUOTE
+ if (a_b.n_bytes != 4 || a_b.flags != SERD_HAS_QUOTE
|| strncmp((const char*)a_b.buf, "a\"bc", 4)) {
FAILF("Bad node %s %zu %zu %d %d\n",
- a_b.buf, a_b.n_bytes, a_b.n_chars, a_b.flags, a_b.type);
+ a_b.buf, a_b.n_bytes, a_b.flags, a_b.type);
}
// Test serd_node_new_uri_from_string
generated by cgit v1.2.1 (git 2.18.0) at 2024-06-02 03:48:38 +0000