Tweaks

floss/language/rust/decode_utf8.py

@@ -27,53 +27,52 @@ def extract_utf8_strings_from_buffer(buf, min_length=MIN_STR_LEN) -> List[List[A
 
     # Reference: https://en.wikipedia.org/wiki/UTF-8
 
+    character_and_index = []
     strings = []
 
     for i in range(0, len(buf)):
         # for 1 byte
         if buf[i] & 0x80 == 0x00:
             character = buf[i].to_bytes(1, "big").decode("utf-8", "ignore")
-            strings.append([character, i])
+            character_and_index.append([character, i, 1])
 
         # for 2 bytes
         elif buf[i] & 0xE0 == 0xC0:
             temp = buf[i] << 8 | buf[i + 1]
             character = temp.to_bytes(2, "big").decode("utf-8", "ignore")
             i += 1
-            strings.append([character, i])
+            character_and_index.append([character, i, 2])
 
         # for 3 bytes
         elif buf[i] & 0xF0 == 0xE0:
             temp = buf[i] << 16 | buf[i + 1] << 8 | buf[i + 2]
             character = temp.to_bytes(3, "big").decode("utf-8", "ignore")
             i += 2
-            strings.append([character, i])
+            character_and_index.append([character, i, 3])
 
         # for 4 bytes
         elif buf[i] & 0xF8 == 0xF0:
             temp = buf[i] << 24 | buf[i + 1] << 16 | buf[i + 2] << 8 | buf[i + 3]
             character = temp.to_bytes(4, "big").decode("utf-8", "ignore")
             i += 3
-            strings.append([character, i])
+            character_and_index.append([character, i, 4])
 
     prev = False
 
-    for i in range(0, len(strings)):
-        if strings[i][0].isprintable() == True:
+    for i in range(0, len(character_and_index)):
+        if character_and_index[i][0].isprintable() == True:
             if prev == False:
-                strings.append([strings[i][0], strings[i][1]])
+                strings.append([character_and_index[i][0], character_and_index[i][1], character_and_index[i][1]])
                 prev = True
             else:
-                strings[-1][0] += strings[i][0]
-                strings[-1][1] = strings[i][1]
+                strings[-1][0] += character_and_index[i][0]
+                strings[-1][2] = character_and_index[i][1]
         else:
             prev = False
 
     # filter strings less than min length
     strings = [string for string in strings if len(string[0]) >= min_length]
 
-    print(strings)
-
     return strings