api: add argument "length" in function utf8_is_valid()

3 files changed, 15 insertions, 8 deletions

diff --git a/src/core/wee-string.c b/src/core/wee-string.c
index e27603512..2f1adb7d9 100644
--- a/src/core/wee-string.c
+++ b/src/core/wee-string.c
@@ -2307,7 +2307,7 @@ string_iconv_to_internal (const char *charset, const char *string)
     if (local_utf8 && (!charset || !charset[0]))
         return input;
 
-    if (utf8_has_8bits (input) && utf8_is_valid (input, NULL))
+    if (utf8_has_8bits (input) && utf8_is_valid (input, -1, NULL))
         return input;
 
     output = string_iconv (0,
diff --git a/src/core/wee-utf8.c b/src/core/wee-utf8.c
index afa64d951..7cfb8a875 100644
--- a/src/core/wee-utf8.c
+++ b/src/core/wee-utf8.c
@@ -70,18 +70,24 @@ utf8_has_8bits (const char *string)
 /*
  * Checks if a string is UTF-8 valid.
  *
+ * If length is <= 0, checks whole string.
+ * If length is > 0, checks only this number of chars (not bytes).
+ *
  * Returns:
  *   1: string is UTF-8 valid
- *   0: string it not UTF-8 valid, and then if error is not NULL, it is set with
- *      first non valid UTF-8 char in string
+ *   0: string it not UTF-8 valid, and then if error is not NULL, it is set
+ *      with first non valid UTF-8 char in string
  */
 
 int
-utf8_is_valid (const char *string, char **error)
+utf8_is_valid (const char *string, int length, char **error)
 {
-    int code_point;
+    int code_point, current_char;
 
-    while (string && string[0])
+    current_char = 0;
+
+    while (string && string[0]
+           && ((length <= 0) || (current_char < length)))
     {
         /*
          * UTF-8, 2 bytes, should be: 110vvvvv 10vvvvvv
@@ -142,6 +148,7 @@ utf8_is_valid (const char *string, char **error)
             goto invalid;
         else
             string++;
+        current_char++;
     }
     if (error)
         *error = NULL;
@@ -165,7 +172,7 @@ utf8_normalize (char *string, char replacement)
 
     while (string && string[0])
     {
-        if (utf8_is_valid (string, &error))
+        if (utf8_is_valid (string, -1, &error))
             return;
         error[0] = replacement;
         string = error + 1;
diff --git a/src/core/wee-utf8.h b/src/core/wee-utf8.h
index 8e256c9f7..abdf96b2b 100644
--- a/src/core/wee-utf8.h
+++ b/src/core/wee-utf8.h
@@ -30,7 +30,7 @@ extern int local_utf8;
 
 extern void utf8_init ();
 extern int utf8_has_8bits (const char *string);
-extern int utf8_is_valid (const char *string, char **error);
+extern int utf8_is_valid (const char *string, int length, char **error);
 extern void utf8_normalize (char *string, char replacement);
 extern const char *utf8_prev_char (const char *string_start,
                                    const char *string);