feat(utf-8 string): implemented validation of utf-8 string

Include/Fledasty/Strings/UTF8String.h

@@ -23,7 +23,6 @@
  * Author: Mineplay
  * -----------------------------------------------------------------------------
  */
-
 #include <stddef.h>
 #include <stdint.h>
 #include <stdbool.h>
@@ -38,7 +37,7 @@ typedef struct {
 FledastyError fledasty_utf8_string_initialize(FledastyUtf8String *new_string, unsigned char *character_string, size_t character_string_length);
 FledastyError fledasty_utf8_string_destroy(FledastyUtf8String *current_string);
 
-FledastyUtf8String fledasty_utf8_string_encode(uint32_t *unicode, size_t size);
+FledastyUtf8String fledasty_utf8_string_encode(uint32_t *unicode, const size_t size);
 uint32_t *fledasty_utf8_string_decode(FledastyUtf8String *current_string, size_t *unicode_string_length);
 
-FledastyError fledasty_utf8_string_append(FledastyUtf8String *current_string, unsigned char *character_string);
+bool fledasty_utf8_string_validate(unsigned char *character_string, const size_t character_string_length);

Include/Fledasty/Utils/Error.h

@@ -30,6 +30,7 @@ typedef enum {
     FLEDASTY_ERROR_INDEX_OUT_OF_RANGE = 3,
     FLEDASTY_ERROR_VALUE_NOT_FOUND = 4,
     FLEDASTY_ERROR_KEY_NOT_FOUND = 5,
+    FLEDASTY_ERROR_INVALID_VALUE = 6,
 } FledastyError;
 
 #endif

Src/Strings/UTF8String.c

@@ -35,12 +35,16 @@ FledastyError fledasty_utf8_string_initialize(FledastyUtf8String *new_string, un
         return FLEDASTY_ERROR_INVALID_POINTER;
     }
 
-    if (character_string == NULL) {
+    if (character_string == NULL || character_string_length == 0) {
         new_string->size = 0;
         new_string->capacity = 10;
 
         new_string->character_string = hallocy_malloc(new_string->capacity);
     } else {
+        if (!fledasty_utf8_string_validate(character_string, character_string_length)) {
+            return FLEDASTY_ERROR_INVALID_VALUE;
+        }
+
         new_string->size = character_string_length;
         new_string->capacity = new_string->size + new_string->size;
 
@@ -68,7 +72,7 @@ FledastyError fledasty_utf8_string_destroy(FledastyUtf8String *current_string) {
     return FLEDASTY_ERROR_NONE;
 }
 
-FledastyUtf8String fledasty_utf8_string_encode(uint32_t *unicode, size_t size) {
+FledastyUtf8String fledasty_utf8_string_encode(uint32_t *unicode, const size_t size) {
     FledastyUtf8String utf8_string;
     fledasty_utf8_string_initialize(&utf8_string, NULL, 0);
 
@@ -130,7 +134,7 @@ FledastyUtf8String fledasty_utf8_string_encode(uint32_t *unicode, size_t size) {
 }
 
 uint32_t *fledasty_utf8_string_decode(FledastyUtf8String *current_string, size_t *unicode_string_length) {
-    if (current_string == NULL) {
+    if (current_string == NULL || unicode_string_length == NULL) {
         return NULL;
     }
 
@@ -157,3 +161,54 @@ uint32_t *fledasty_utf8_string_decode(FledastyUtf8String *current_string, size_t
 
     return unicode_string;
 }
+
+bool fledasty_utf8_string_validate(unsigned char *character_string, const size_t character_string_length) {
+    if (character_string == NULL) {
+        return false;
+    }
+
+    size_t index = 0;
+    while (index < character_string_length) {
+        if ((character_string[index] & 0xF0) == 0xF0) {
+            if (index + 3 >= character_string_length) {
+                return false;
+            }
+
+            if ((character_string[index + 1] & 0xC0) != 0x80) {
+                return false;
+            } else if ((character_string[index + 2] & 0xC0) != 0x80) {
+                return false;
+            } else if ((character_string[index + 3] & 0xC0) != 0x80) {
+                return false;
+            }
+            
+            index += 4;
+        } else if ((character_string[index] & 0xE0) == 0xC0) {
+            if (index + 2 >= character_string_length) {
+                return false;
+            }
+
+            if ((character_string[index + 1] & 0xC0) != 0x80) {
+                return false;
+            } else if ((character_string[index + 2] & 0xC0) != 0x80) {
+                return false;
+            }
+
+            index += 3;
+        } else if ((character_string[index] & 0xC0) == 0xC0) {
+            if (index + 1 >= character_string_length) {
+                return false;
+            }
+
+            if ((character_string[index + 1] & 0xC0) != 0x80) {
+                return false;
+            }
+
+            index += 2;
+        } else {
+            index += 1;
+        }
+    }
+
+    return true;
+}

Tests/Main.c

@@ -235,6 +235,22 @@ int main() {
 
     printf("%s\n", encoded_string.character_string);
     
+    if (fledasty_utf8_string_validate(test_utf8_string.character_string, encoded_string.size)) {
+        printf("UTF-8 test string is valid!\n");
+    }
+
+    if (fledasty_utf8_string_validate(encoded_string.character_string, encoded_string.size)) {
+        printf("UTF-8 encoded string is valid!\n");
+    }
+
+    unsigned char *invalid_utf8 = (unsigned char*)hallocy_malloc(2 * sizeof(unsigned char));
+    invalid_utf8[0] = 0xDF;
+    invalid_utf8[1] = 0xFF;
+    if (!fledasty_utf8_string_validate(invalid_utf8, 2)) {
+        printf("UTF-8 invalid string is invalid!\n");
+    }
+
+    hallocy_free(invalid_utf8);
     hallocy_free(unicode);
     fledasty_utf8_string_destroy(&encoded_string);