Funktionierende Version des Decoders

P3/main.py

@@ -2,8 +2,8 @@
 import math
 
 ELEMENTS = []
-INTERVALLS = []
-N = 8  # länge eines Codeworts
+N = 50  # länge eines Codeworts
+EOF = "$"
 
 
 class Element:
@@ -11,7 +11,6 @@ class Element:
     count = 0
     p_x_i = 0
     I_x_i = 0
-    bin = ""
 
     def __init__(self, id):
         self.id = id
@@ -34,12 +33,8 @@ def read_text_file():
 
 
 def Z_statistik(message):
-    # Ergebnisse mit https://de.planetcalc.com/2476/ prüfbar
-    # message = read_text_file()
-
-    m = len(message)
-
     global ELEMENTS
+    m = len(message)
 
     for char in message:
         i, entry = find_element(ELEMENTS, char)
@@ -48,11 +43,6 @@ def Z_statistik(message):
         else:
             ELEMENTS.append(Element(char))
 
-    calculate(m, message)
-
-
-def calculate(m, message):
-    global ELEMENTS
     h = 0
     for element in ELEMENTS:
         element.p_x_i = element.count / m
@@ -61,31 +51,53 @@ def calculate(m, message):
 
     # Ausgabe
     ELEMENTS.sort(key=lambda element: element.count, reverse=True)
-    """for element in elements:
-        print(f"{element.count:3.0f} | {element.p_x_i:10.7f} | {element.I_x_i:10.7f} | »{element.id}«")
-
-    print("Entropie = " + h.__str__() + "\n\n")"""
 
 
 # --- Aus P1 ---
 
-def charToBin():
-    global ELEMENTS
-    for element in ELEMENTS:
-        element.bin = format(ord(element.id), 'b')
+def binary_to_decimal(nkst_bits):
+    result = 0.0
+    for i, bit in enumerate(nkst_bits):
+        if bit == '1':
+            result += 2 ** (-(i + 1))
+    return result
 
 
-def calcBitCount(block):
+def decimal_to_binary(decimal_val, n_bits, min_val, max_val):
+    nkst_bits = []
+    val = decimal_val
+
+    for _ in range(n_bits):
+        val *= 2
+        if val >= 1:
+            nkst_bits.append('1')
+            val -= 1
+        else:
+            nkst_bits.append('0')
+
+    converted_val = binary_to_decimal(nkst_bits)
+
+    # !!! Führt bei zu kleinen Intervallen zu Ungenauigkeiten
+    if not (min_val <= converted_val < max_val):
+        if nkst_bits[-1] == '1':
+            nkst_bits[-1] = '0'
+        else:
+            nkst_bits[-1] = '1'
+
+    return "".join(nkst_bits)
+
+
+def calc_bit_count(block):
     p = 1
     for char in block:
         i, element = find_element(ELEMENTS, char)
         p *= element.p_x_i
 
-    N = round(-math.log(p), 0)
+    N = math.ceil(-math.log(p, 2))
     return N
 
 
-def calcIntervalls():
+def calc_intervalls():
     global INTERVALLS
     current_limit = 0
     for element in ELEMENTS:
@@ -94,90 +106,163 @@ def calcIntervalls():
         INTERVALLS.append(new_limit)
 
 
-def createBlocks(message):
-    blocks = []
-
+def calc_block_size():
+    last_element = ELEMENTS[-2]
     block = []
 
-    for char in message:
-        if len(block) == 0:
-            block.append(char)
-            continue
-        if calcBitCount(block) <= N:
-            block.append(char)
+    while True:
+        tmp = block.copy()
+        tmp.append(last_element.id)
+        if calc_bit_count(tmp) <= N:
+            block.append(last_element.id)
         else:
-            blocks.append(block)
-            block = [char]
+            break
 
-    blocks.append(block)
-    # TODO EOF Symbol setzen und auf Länge des Blocks dabei achten
+    return len(block)
+
+
+def create_blocks(message):
+    blocks = []
+    block = []
+    block_size = calc_block_size()
+
+    for i in range(0, len(message), block_size):
+        block = message[i:i + block_size]
+        blocks.append(block)
 
     return blocks
 
 
-def createCodeWord(block):
-    intervalls = []
+def create_code_word(block, verbose=False):
+    current_low = 0.0
+    current_high = 1.0
 
-    i, element = find_element(ELEMENTS, block[0])
-    print(element.id)
+    for position, char in enumerate(block):
+        target_element = find_element(ELEMENTS, char)
 
-    if i == 0:
-        low = 0
-    else:
-        low = INTERVALLS[i - 1]
-    high = INTERVALLS[i]
-    range = high - low
+        if target_element is None:
+            return None
 
-    # Normierung
-    for i, element in enumerate(ELEMENTS):
-        new_high = low + range * element.p_x_i
-        intervalls.append((low, new_high))
-        low = new_high
+        # Erstelle Intervalle für aktuelle Position
+        char_intervals = []
+        current_range = current_high - current_low
+        low = current_low
 
-    print(intervalls)
+        for element in ELEMENTS:
+            # Intervall für dieses Zeichen: [low, high)
+            high = low + current_range * element.p_x_i
+            char_intervals.append((element.id, (low, high)))
+            low = high
+
+        # Finde das Intervall für das aktuelle Zeichen
+        found_interval = None
+        for element_char, interval in char_intervals:
+            if element_char == char:
+                found_interval = interval
+                break
+
+        if found_interval is None:
+            return None
+
+        interval_low, interval_high = found_interval
+        current_low = interval_low
+        current_high = interval_high
+
+    final_value = current_low + (current_high - current_low) / 2
+    binary_representation = decimal_to_binary(final_value, N, current_low, current_high)
+
+    return binary_representation
 
 
-    for char in block[1:]:
-        i, element = find_element(ELEMENTS, char)
-        print(element.id)
+def decode_single_block(decimal_value, max_chars=N):
+    if not ELEMENTS:
+        return None
 
-        low, high = intervalls[i]
+    decoded_chars = []
+    current_value = decimal_value
 
-        range = high - low
+    for position in range(max_chars):
+        char_intervals = []
+        low = 0.0
 
-        intervalls = []
+        for element in ELEMENTS:
+            high = low + element.p_x_i
+            char_intervals.append((element.id, (low, high)))
+            low = high
 
-        for i, element in enumerate(ELEMENTS):
-            new_high = low + range * element.p_x_i
-            intervalls.append((low, new_high))
-            low = new_high
+        found_char = None
+        found_interval = None
 
-        print(intervalls)
+        for char, interval in char_intervals:
+            interval_low, interval_high = interval
+            if interval_low <= current_value < interval_high:
+                found_char = char
+                found_interval = interval
+                break
 
-    low, x = intervalls[0]
-    x, high = intervalls[-1]
-    print(f"[{low}, {high})")
+        # Spezialfall: Wert liegt genau auf der oberen Grenze des letzten Intervalls
+        if found_char is None and current_value == 1.0:
+            found_char = char_intervals[-1][0]
+            found_interval = char_intervals[-1][1]
+
+        if found_char is None or found_char == EOF:
+            break
+
+        decoded_chars.append(found_char)
+
+        interval_low, interval_high = found_interval
+        interval_range = interval_high - interval_low
+
+        if interval_range == 0:
+            break
+
+        # Normalisierung: Transformiere den Wert vom aktuellen Intervall auf [0, 1)
+        current_value = (current_value - interval_low) / interval_range
+
+    return ''.join(decoded_chars)
 
 
-def AC_Encoder():
-    global ELEMENTS
-    global INTERVALLS
+def AC_Decoder(code):
+    decoded = ""
+    l = calc_block_size()
+
+    for i in range(0, len(code), N):
+        if i + N > len(code):
+            block = code[i:]
+        else:
+            block = code[i:i + N]
+
+        decoded += decode_single_block(binary_to_decimal(block), l)
+
+    return decoded
+
+
+def AC_Encoder(message):
+    Z_statistik(message)
+
+    blocks = create_blocks(message)
+
+    code = ""
+
+    for block in blocks:
+        code += create_code_word(block)
+
+    return code
+
+
+def main():
     message = read_text_file()
 
-    l = len(message)
+    encoded = AC_Encoder(message)
+    decoded = AC_Decoder(encoded)
 
-    l_bin = l * 8
+    print(f"Original: {len(message)} Zeichen = {len(message) * 8} Bits")
+    print(f"Komprimiert: {len(encoded)} Bits")
+    print(f"Maximale Blockgröße: {calc_block_size()}")
+    print(f"Anzahl verschiedener Zeichen: {len(ELEMENTS)}")
+    print(f"Zweitseltenste: '{ELEMENTS[-2].id}' kommt {ELEMENTS[-2].count}x vor")
 
-    Z_statistik(message)
-    # charToBin() # unnötig, da nur das Codewort in binär umgewandelt werden muss
-    calcIntervalls()
-    # INTERVALLS = [0.1, 0.8, 1]
-
-    blocks = createBlocks(message)
-
-    print(INTERVALLS)
-
-    createCodeWord(blocks[0])
+    print(f"\nIst die Nachricht richtig en- & dekodiert worden? {message[:-1] == decoded}")
 
 
-AC_Encoder()
+main()

P3/text.txt

@@ -1 +1 @@
-SWISS
+qwewqwqwewqwewqwewwweeqeqwqwqewqwqwqeqweqwqweqewqwweqeqwwqeeqwqwwqeeqwwwqeeeeqewweqeeewwwwqeeqwwwqeeqewwqwwwqeeqeeeeeqwwwwqwwqeqwqwqwqwwwq$