Add RLE

analyzer/main.py

@@ -106,9 +106,52 @@ def _encode_binary(binary: Binary) -> bytes:
     _write_u8(0)
 
     _write_u32(len(binary.starting_state))
-    # TODO: No RLE for now
-    for b in binary.starting_state:
-        _write_u8(b)
+
+    # == encode RLE ==
+    def _pull_repeats(start: int, data: bytes) -> int | None:
+        i = 0
+        first_byte = data[start + i]
+        while True:
+            if i >= 255:
+                break
+            if start + i >= len(data):
+                break 
+            if data[start + i] != first_byte:
+                break
+
+            i += 1
+
+        if i >= 2:
+            _write_u8(i)
+            _write_u8(first_byte)
+            return start + i
+        
+        return None
+
+    def _pull_non_repeats(start: int, data: bytes):
+        i = 0
+        while True:
+            if i >= 255:
+                break
+            if start + i >= len(data):
+                break
+            if i >= 1 and data[start + i] == data[start + i - 1]:
+                break
+
+            i += 1
+        
+        _write_u8(0)
+        _write_u8(i)
+        for j in range(i):
+            _write_u8(data[start + j])
+        return start + i
+
+    i = 0
+    while i < len(binary.starting_state):
+        if new_i := _pull_repeats(i, binary.starting_state):
+            i = new_i
+            continue
+        i = _pull_non_repeats(i, binary.starting_state)
 
     return out.getbuffer()