Welcome to mirror list, hosted at ThFree Co, Russian Federation.

git.blender.org/blender.git - Unnamed repository; edit this file 'description' to name the repository.
summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
Diffstat (limited to 'source/blender/blenlib/intern/string_utf8.c')
-rw-r--r--source/blender/blenlib/intern/string_utf8.c32
1 files changed, 16 insertions, 16 deletions
diff --git a/source/blender/blenlib/intern/string_utf8.c b/source/blender/blenlib/intern/string_utf8.c
index 16a5f03095d..bf98f2ae77c 100644
--- a/source/blender/blenlib/intern/string_utf8.c
+++ b/source/blender/blenlib/intern/string_utf8.c
@@ -394,10 +394,10 @@ int BLI_str_utf8_size_safe(const char *p)
/* was g_utf8_get_char */
/**
* BLI_str_utf8_as_unicode:
- * @p a pointer to Unicode character encoded as UTF-8
+ * \param p a pointer to Unicode character encoded as UTF-8
*
* Converts a sequence of bytes encoded as UTF-8 to a Unicode character.
- * If @p does not point to a valid UTF-8 encoded character, results are
+ * If \a p does not point to a valid UTF-8 encoded character, results are
* undefined. If you are not sure that the bytes are complete
* valid Unicode characters, you should use g_utf8_get_char_validated()
* instead.
@@ -536,14 +536,14 @@ size_t BLI_str_utf8_from_unicode(unsigned int c, char *outbuf)
/* was g_utf8_find_prev_char */
/**
* BLI_str_find_prev_char_utf8:
- * @str: pointer to the beginning of a UTF-8 encoded string
- * @p pointer to some position within @str
+ * \param str pointer to the beginning of a UTF-8 encoded string
+ * \param p pointer to some position within \a str
*
- * Given a position @p with a UTF-8 encoded string @str, find the start
- * of the previous UTF-8 character starting before. @p Returns %NULL if no
- * UTF-8 characters are present in @str before @p
+ * Given a position \a p with a UTF-8 encoded string \a str, find the start
+ * of the previous UTF-8 character starting before. \a p Returns %NULL if no
+ * UTF-8 characters are present in \a str before \a p
*
- * @p does not have to be at the beginning of a UTF-8 character. No check
+ * \a p does not have to be at the beginning of a UTF-8 character. No check
* is made to see if the character found is actually valid other than
* it starts with an appropriate byte.
*
@@ -562,13 +562,13 @@ char * BLI_str_find_prev_char_utf8(const char *str, const char *p)
/* was g_utf8_find_next_char */
/**
* BLI_str_find_next_char_utf8:
- * @p a pointer to a position within a UTF-8 encoded string
- * @end a pointer to the byte following the end of the string,
+ * \param p a pointer to a position within a UTF-8 encoded string
+ * \param end a pointer to the byte following the end of the string,
* or %NULL to indicate that the string is nul-terminated.
*
- * Finds the start of the next UTF-8 character in the string after @p
+ * Finds the start of the next UTF-8 character in the string after \a p
*
- * @p does not have to be at the beginning of a UTF-8 character. No check
+ * \a p does not have to be at the beginning of a UTF-8 character. No check
* is made to see if the character found is actually valid other than
* it starts with an appropriate byte.
*
@@ -594,13 +594,13 @@ char *BLI_str_find_next_char_utf8(const char *p, const char *end)
/* was g_utf8_prev_char */
/**
* BLI_str_prev_char_utf8:
- * @p a pointer to a position within a UTF-8 encoded string
+ * \param p a pointer to a position within a UTF-8 encoded string
*
- * Finds the previous UTF-8 character in the string before @p
+ * Finds the previous UTF-8 character in the string before \a p
*
- * @p does not have to be at the beginning of a UTF-8 character. No check
+ * \a p does not have to be at the beginning of a UTF-8 character. No check
* is made to see if the character found is actually valid other than
- * it starts with an appropriate byte. If @p might be the first
+ * it starts with an appropriate byte. If \a p might be the first
* character of the string, you must use g_utf8_find_prev_char() instead.
*
* Return value: a pointer to the found character.