feat(utf-8 string): implemented encoding and decoding of utf-8 string with unicode

Include/Fledasty/Strings/UTF8String.h

@@ -25,6 +25,7 @@
  */
 
 #include <stddef.h>
+#include <stdint.h>
 #include <stdbool.h>
  
 #include "../Utils/Error.h"
@@ -37,4 +38,7 @@ 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);
+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);

Src/Strings/UTF8String.c

@@ -28,13 +28,14 @@
 #include <Hallocy/Core/Allocator.h>
 #include <Hallocy/Core/Memory.h>
 #include <Hallocy/Utils/Error.h>
+#include <stdint.h>
 
 FledastyError fledasty_utf8_string_initialize(FledastyUtf8String *new_string, unsigned char *character_string, size_t character_string_length) {
     if (new_string == NULL) {
         return FLEDASTY_ERROR_INVALID_POINTER;
     }
 
-    if (character_string == NULL || character_string_length == 0) {
+    if (character_string == NULL) {
         new_string->size = 0;
         new_string->capacity = 10;
 
@@ -66,3 +67,93 @@ FledastyError fledasty_utf8_string_destroy(FledastyUtf8String *current_string) {
     current_string->character_string = NULL;
     return FLEDASTY_ERROR_NONE;
 }
+
+FledastyUtf8String fledasty_utf8_string_encode(uint32_t *unicode, size_t size) {
+    FledastyUtf8String utf8_string;
+    fledasty_utf8_string_initialize(&utf8_string, NULL, 0);
+
+    if (unicode == NULL) {
+        return utf8_string;
+    }
+
+    size_t string_index = 0;
+    for (size_t index = 0; index < size; index += 1) {
+        if (unicode[index] <= 0x00007F) {
+            if (utf8_string.capacity <= string_index) {
+                utf8_string.capacity += utf8_string.capacity;
+                utf8_string.character_string = hallocy_realloc(utf8_string.character_string, utf8_string.capacity);
+            }
+
+            utf8_string.character_string[string_index] = unicode[index];
+            string_index += 1;
+        } else if (unicode[index] <= 0x0007FF) {          
+            if (utf8_string.capacity <= string_index + 2) {
+                utf8_string.capacity += utf8_string.capacity;
+                utf8_string.character_string = hallocy_realloc(utf8_string.character_string, utf8_string.capacity);
+            }
+
+            utf8_string.character_string[string_index] = 0xC0 | ((unicode[index] >> 6) & 0x07);
+            utf8_string.character_string[string_index + 1] = 0x80 | (unicode[index] & 0x3F);
+            string_index += 2;
+        } else if (unicode[index] <= 0x00FFFF) {
+            if (utf8_string.capacity <= string_index + 3) {
+                utf8_string.capacity += utf8_string.capacity;
+                utf8_string.character_string = hallocy_realloc(utf8_string.character_string, utf8_string.capacity);
+            }
+
+            utf8_string.character_string[string_index] = 0xE0 | ((unicode[index] >> 12) & 0x07);
+            utf8_string.character_string[string_index + 1] = 0x80 | ((unicode[index] >> 6) & 0x3F);
+            utf8_string.character_string[string_index + 2] = 0x80 | (unicode[index] & 0x3F);
+            string_index += 3;
+        } else if (unicode[index] <= 0x10FFFF) {
+            if (utf8_string.capacity <= string_index + 4) {
+                utf8_string.capacity += utf8_string.capacity;
+                utf8_string.character_string = hallocy_realloc(utf8_string.character_string, utf8_string.capacity);
+            }
+
+            utf8_string.character_string[string_index] = 0xF0 | ((unicode[index] >> 18) & 0x07);
+            utf8_string.character_string[string_index + 1] = 0x80 | ((unicode[index] >> 12) & 0x3F);
+            utf8_string.character_string[string_index + 2] = 0x80 | ((unicode[index] >> 6) & 0x3F);
+            utf8_string.character_string[string_index + 3] = 0x80 | (unicode[index] & 0x3F);
+            string_index += 4;
+        }
+    }
+
+    utf8_string.size = string_index;
+    if (utf8_string.capacity <= utf8_string.size + 1) {
+        utf8_string.capacity += utf8_string.capacity;
+        utf8_string.character_string = hallocy_realloc(utf8_string.character_string, utf8_string.capacity);
+    }
+
+    utf8_string.character_string[utf8_string.size] = '\0';
+    return utf8_string;
+}
+
+uint32_t *fledasty_utf8_string_decode(FledastyUtf8String *current_string, size_t *unicode_string_length) {
+    if (current_string == NULL) {
+        return NULL;
+    }
+
+    (*unicode_string_length) = 0;
+    size_t index = 0;
+    uint32_t *unicode_string = hallocy_malloc(current_string->size * sizeof(uint32_t));
+    while (index < current_string->size) {
+        if ((current_string->character_string[index] & 0xF0) == 0xF0) {
+            unicode_string[*unicode_string_length] = ((current_string->character_string[index] & 0x07) << 18) | ((current_string->character_string[index + 1] & 0x3F) << 12) | ((current_string->character_string[index + 2] & 0x3F) << 6) | (current_string->character_string[index + 3] & 0x3F);
+            index += 4;
+        } else if ((current_string->character_string[index] & 0xE0) == 0xE0) {
+            unicode_string[*unicode_string_length] = ((current_string->character_string[index] & 0x0F) << 12) | ((current_string->character_string[index + 1] & 0x3F) << 6) | (current_string->character_string[index + 2] & 0x3F);
+            index += 3;
+        } else if ((current_string->character_string[index] & 0xC0) == 0xC0) {
+            unicode_string[*unicode_string_length] = ((current_string->character_string[index] & 0x1F) << 6) | (current_string->character_string[index + 1] & 0x3F);
+            index += 2;
+        } else {
+            unicode_string[*unicode_string_length] = current_string->character_string[index];
+            index += 1;
+        }
+
+        (*unicode_string_length) += 1;
+    }
+
+    return unicode_string;
+}

Tests/Main.c

@@ -20,6 +20,7 @@
  * -----------------------------------------------------------------------------
  */
 #include <stdio.h>
+#include <Hallocy/Core/Allocator.h>
 #include <Fledasty/Core/Queue.h>
 #include <Fledasty/Core/Stack.h>
 #include <Fledasty/Core/DynamicArray.h>
@@ -224,10 +225,19 @@ int main() {
     fledasty_hash_table_destroy(&test_hash_table);
     FledastyUtf8String test_utf8_string;
     unsigned char *test_string = (unsigned char*)"😀€Testing";
-    fledasty_utf8_string_initialize(&test_utf8_string, test_string, 11);
+    fledasty_utf8_string_initialize(&test_utf8_string, test_string, 15);
     printf("%s\n", test_string);
     printf("%s\n", test_utf8_string.character_string);
 
+    size_t unicode_length = 0;
+    uint32_t *unicode = fledasty_utf8_string_decode(&test_utf8_string, &unicode_length);
+    FledastyUtf8String encoded_string = fledasty_utf8_string_encode(unicode, unicode_length);
+
+    printf("%s\n", encoded_string.character_string);
+    
+    hallocy_free(unicode);
+    fledasty_utf8_string_destroy(&encoded_string);
+
     fledasty_utf8_string_destroy(&test_utf8_string);
     printf("Done\n");
     return 0;