utf8: add conversion with size and refactor names

Add functions to use convert a string with length, instead of assuming
NUL termination.

In addition, move the utf8 to 16 conversion routines into the `git_utf8`
namespace instead of using namespaceless `git__` prefixed names.

1 files changed, 81 insertions, 14 deletions

diff --git a/src/util/win32/utf-conv.h b/src/util/win32/utf-conv.h
index 120d647ef..301f5a6d3 100644
--- a/src/util/win32/utf-conv.h
+++ b/src/util/win32/utf-conv.h
@@ -16,14 +16,45 @@
 #endif
 
 /**
+ * Converts a NUL-terminated UTF-8 string to wide characters. This is a
+ * convenience function for `git_utf8_to_16_with_len`.
+ *
+ * @param dest The buffer to receive the wide string.
+ * @param dest_size The size of the buffer, in characters.
+ * @param src The UTF-8 string to convert.
+ * @return The length of the wide string, in characters
+ *         (not counting the NULL terminator), or < 0 for failure
+ */
+int git_utf8_to_16(wchar_t *dest, size_t dest_size, const char *src);
+
+/**
  * Converts a UTF-8 string to wide characters.
  *
  * @param dest The buffer to receive the wide string.
  * @param dest_size The size of the buffer, in characters.
  * @param src The UTF-8 string to convert.
- * @return The length of the wide string, in characters (not counting the NULL terminator), or < 0 for failure
+ * @param src_len The length of the string to convert.
+ * @return The length of the wide string, in characters
+ *         (not counting the NULL terminator), or < 0 for failure
+ */
+int git_utf8_to_16_with_len(
+	wchar_t *dest,
+	size_t dest_size,
+	const char *src,
+	int src_len);
+
+/**
+ * Converts a NUL-terminated wide string to UTF-8. This is a convenience
+ * function for `git_utf8_from_16_with_len`.
+ *
+ * @param dest The buffer to receive the UTF-8 string.
+ * @param dest_size The size of the buffer, in bytes.
+ * @param src The wide string to convert.
+ * @param src_len The length of the string to convert.
+ * @return The length of the UTF-8 string, in bytes
+ *         (not counting the NULL terminator), or < 0 for failure
  */
-int git__utf8_to_16(wchar_t *dest, size_t dest_size, const char *src);
+int git_utf8_from_16(char *dest, size_t dest_size, const wchar_t *src);
 
 /**
  * Converts a wide string to UTF-8.
@@ -31,30 +62,66 @@ int git__utf8_to_16(wchar_t *dest, size_t dest_size, const char *src);
  * @param dest The buffer to receive the UTF-8 string.
  * @param dest_size The size of the buffer, in bytes.
  * @param src The wide string to convert.
- * @return The length of the UTF-8 string, in bytes (not counting the NULL terminator), or < 0 for failure
+ * @param src_len The length of the string to convert.
+ * @return The length of the UTF-8 string, in bytes
+ *         (not counting the NULL terminator), or < 0 for failure
  */
-int git__utf16_to_8(char *dest, size_t dest_size, const wchar_t *src);
+int git_utf8_from_16_with_len(char *dest, size_t dest_size, const wchar_t *src, int src_len);
 
 /**
- * Converts a UTF-8 string to wide characters.
- * Memory is allocated to hold the converted string.
- * The caller is responsible for freeing the string with git__free.
+ * Converts a UTF-8 string to wide characters. Memory is allocated to hold
+ * the converted string. The caller is responsible for freeing the string
+ * with git__free.
  *
  * @param dest Receives a pointer to the wide string.
  * @param src The UTF-8 string to convert.
- * @return The length of the wide string, in characters (not counting the NULL terminator), or < 0 for failure
+ * @return The length of the wide string, in characters
+ *         (not counting the NULL terminator), or < 0 for failure
  */
-int git__utf8_to_16_alloc(wchar_t **dest, const char *src);
+int git_utf8_to_16_alloc(wchar_t **dest, const char *src);
 
 /**
- * Converts a wide string to UTF-8.
- * Memory is allocated to hold the converted string.
- * The caller is responsible for freeing the string with git__free.
+ * Converts a UTF-8 string to wide characters. Memory is allocated to hold
+ * the converted string. The caller is responsible for freeing the string
+ * with git__free.
+ *
+ * @param dest Receives a pointer to the wide string.
+ * @param src The UTF-8 string to convert.
+ * @param src_len The length of the string.
+ * @return The length of the wide string, in characters
+ *         (not counting the NULL terminator), or < 0 for failure
+ */
+int git_utf8_to_16_alloc_with_len(
+	wchar_t **dest,
+	const char *src,
+	int src_len);
+
+/**
+ * Converts a wide string to UTF-8. Memory is allocated to hold the
+ * converted string. The caller is responsible for freeing the string
+ * with git__free.
+ *
+ * @param dest Receives a pointer to the UTF-8 string.
+ * @param src The wide string to convert.
+ * @return The length of the UTF-8 string, in bytes
+ *         (not counting the NULL terminator), or < 0 for failure
+ */
+int git_utf8_from_16_alloc(char **dest, const wchar_t *src);
+
+/**
+ * Converts a wide string to UTF-8. Memory is allocated to hold the
+ * converted string. The caller is responsible for freeing the string
+ * with git__free.
  *
  * @param dest Receives a pointer to the UTF-8 string.
  * @param src The wide string to convert.
- * @return The length of the UTF-8 string, in bytes (not counting the NULL terminator), or < 0 for failure
+ * @param src_len The length of the wide string.
+ * @return The length of the UTF-8 string, in bytes
+ *         (not counting the NULL terminator), or < 0 for failure
  */
-int git__utf16_to_8_alloc(char **dest, const wchar_t *src);
+int git_utf8_from_16_alloc_with_len(
+	char **dest,
+	const wchar_t *src,
+	int src_len);
 
 #endif