🗓 Nov 26, 2023 11:03:28 AM

✨ lz77 compress/decompress
🧪 tests added/updated

TODO

@@ -26,6 +26,7 @@ New ideas:
   ☐ ✨ amf encode/decode
   ☐ ✨ aes cmac
   ☐ ✨ whitespace encoding https://www.dcode.fr/whitespace-language
+  beaufort
 
 Bug:
 

chepy/modules/compression.py

@@ -9,6 +9,7 @@
 import stat
 from typing import TypeVar
 import lazy_import
+from .internal.helpers import LZ77Compressor
 
 LZ4 = lazy_import.lazy_module("lz4.frame")
 
@@ -25,7 +26,7 @@ def __tar_modes(self, mode):  # pragma: no cover
         assert mode in ["gz", "bz2", "xz", ""], "Valid modes are gz, bz2, xz"
 
     @ChepyDecorators.call_stack
-    def fix_zip_header(self) -> CompressionT:
+    def fix_zip_header(self: CompressionT) -> CompressionT:
         """Fix the first 4 bytes of a zip file
 
         Returns:
@@ -443,3 +444,35 @@ def lz4_decompress(self) -> CompressionT:
         """
         self.state = LZ4.decompress(self._convert_to_bytes())
         return self
+
+    @ChepyDecorators.call_stack
+    def lz77_compress(self, window_size: int = 13, lookahead_buffer_size: int = 6):
+        """To LZ77 compression
+
+        Args:
+            window_size (int, optional): Window size. Defaults to 13.
+            lookahead_buffer_size (int, optional): Lookahead. Defaults to 6.
+
+        Returns:
+            Chepy: The Chepy object.
+        """
+        self.state = LZ77Compressor(window_size, lookahead_buffer_size).compress(
+            self._convert_to_str()
+        )
+        return self
+
+    def lz77_decompress(self, window_size: int = 13, lookahead_buffer_size: int = 6):
+        """From LZ77 compression
+
+        Args:
+            window_size (int, optional): Window size. Defaults to 13.
+            lookahead_buffer_size (int, optional): Lookahead. Defaults to 6.
+
+        Returns:
+            Chepy: The Chepy object.
+        """
+        assert isinstance(self.state, list), "State is not a list"
+        self.state = LZ77Compressor(window_size, lookahead_buffer_size).decompress(
+            self.state
+        )
+        return self

chepy/modules/compression.pyi

@@ -30,3 +30,5 @@ class Compression(ChepyCore):
     def raw_deflate(self: CompressionT) -> CompressionT: ...
     def lz4_compress(self: CompressionT) -> CompressionT: ...
     def lz4_decompress(self: CompressionT) -> CompressionT: ...
+    def lz77_compress(self: CompressionT, window_size: int = 13, lookahead_buffer_size: int = 6) -> CompressionT: ...
+    def lz77_decompress(self: CompressionT, window_size: int = 13, lookahead_buffer_size: int = 6) -> CompressionT: ...

chepy/modules/dataformat.py

@@ -2047,7 +2047,7 @@ def to_uuencode(self, header: str = "-") -> DataFormatT:
         return self
 
     @ChepyDecorators.call_stack
-    def from_uuencode(self, header: str = "-") -> DataFormatT:
+    def from_uuencode(self: DataFormatT, header: str = "-") -> DataFormatT:
         """From UUEncode
 
         Args:
@@ -2058,39 +2058,3 @@ def from_uuencode(self, header: str = "-") -> DataFormatT:
         """
         self.state = UUEncoderDecoder(self._convert_to_bytes(), header).uudecode()
         return self
-
-    @ChepyDecorators.call_stack
-    def to_lz77(
-        self, window_size: int = 13, lookahead_buffer_size: int = 6
-    ) -> DataFormatT:
-        """To LZ77 compression
-
-        Args:
-            window_size (int, optional): Window size. Defaults to 13.
-            lookahead_buffer_size (int, optional): Lookahead. Defaults to 6.
-
-        Returns:
-            Chepy: The Chepy object.
-        """
-        self.state = LZ77Compressor(window_size, lookahead_buffer_size).compress(
-            self._convert_to_str()
-        )
-        return self
-
-    def from_lz77(
-        self, window_size: int = 13, lookahead_buffer_size: int = 6
-    ) -> DataFormatT:
-        """From LZ77 compression
-
-        Args:
-            window_size (int, optional): Window size. Defaults to 13.
-            lookahead_buffer_size (int, optional): Lookahead. Defaults to 6.
-
-        Returns:
-            Chepy: The Chepy object.
-        """
-        assert isinstance(self.state, list), "State is not a list"
-        self.state = LZ77Compressor(window_size, lookahead_buffer_size).decompress(
-            self.state
-        )
-        return self

chepy/modules/dataformat.pyi

@@ -110,5 +110,3 @@ class DataFormat(ChepyCore):
     def from_utf21(self: DataFormatT) -> DataFormatT: ...
     def to_uuencode(self: DataFormatT, header: str='-') -> DataFormatT: ...
     def from_uuencode(self: DataFormatT, header: str='-') -> DataFormatT: ...
-    def to_lz77(self: DataFormatT, window_size: int = 13, lookahead_buffer_size: int = 6) -> DataFormatT: ...
-    def from_lz77(self: DataFormatT, window_size: int = 13, lookahead_buffer_size: int = 6) -> DataFormatT: ...

tests/test_compression.py

@@ -1,4 +1,5 @@
 from chepy import Chepy
+import re
 
 
 def test_fix_zip_header():
@@ -180,3 +181,17 @@ def test_lz4_decompress():
         .o
         == b"data"
     )
+
+def test_lz77():
+    input_str = "(0,0,O)(0,0,M)(0,0,G)(1,1,G)(3,3, )(0,0,Y)(10,1,U)(4,1,A)(0,0,R)(0,0,E)(4,1,C)(0,0,L)(9,1,S)(6,2,T)(5,1, )(3,1,H)(7,2,F)(13,1,A)(1,1,A)(2,2,G)(36,7,C)(28,5,C)(6,5,W)(3,1,L)(1,1, )(0,0,N)(10,1,W)(40,3,I)(15,1, )(3,3,T)(48,6,G)(5,1,E)(0,0,K)(22,1,{)(25,1,I)(38,1,E)(1,1,E)(3,3,E)(7,7,E)(15,15,_)(38,3,O)(2,2,O)(5,5,O)(11,11,O)(3,3,_)(63,23,})"
+    array_of_arrays = []
+    regex = r"\((\d+),(\d+),([A-Z\s_{}]+)\)"
+    matches = re.findall(regex, input_str)
+
+    for match in matches:
+        param1, param2, param3 = match
+        array_of_arrays.append([int(param1), int(param2), param3])
+
+    assert b'EKO{' in Chepy(array_of_arrays).lz77_decompress().o
+
+    assert Chepy('OMGGGGGG').lz77_compress(1).o[1] == [0, 0, 'M']

tests/test_dataformat.py

@@ -1,5 +1,4 @@
 from chepy import Chepy
-import re
 
 
 def test_eval():
@@ -735,17 +734,3 @@ def test_uuencode():
         b"EKO{UUENC0DED_ENCRYPTED?}"
         in Chepy(data).to_uuencode().from_uuencode().remove_nullbytes().o
     )
-
-def test_lz77():
-    input_str = "(0,0,O)(0,0,M)(0,0,G)(1,1,G)(3,3, )(0,0,Y)(10,1,U)(4,1,A)(0,0,R)(0,0,E)(4,1,C)(0,0,L)(9,1,S)(6,2,T)(5,1, )(3,1,H)(7,2,F)(13,1,A)(1,1,A)(2,2,G)(36,7,C)(28,5,C)(6,5,W)(3,1,L)(1,1, )(0,0,N)(10,1,W)(40,3,I)(15,1, )(3,3,T)(48,6,G)(5,1,E)(0,0,K)(22,1,{)(25,1,I)(38,1,E)(1,1,E)(3,3,E)(7,7,E)(15,15,_)(38,3,O)(2,2,O)(5,5,O)(11,11,O)(3,3,_)(63,23,})"
-    array_of_arrays = []
-    regex = r"\((\d+),(\d+),([A-Z\s_{}]+)\)"
-    matches = re.findall(regex, input_str)
-
-    for match in matches:
-        param1, param2, param3 = match
-        array_of_arrays.append([int(param1), int(param2), param3])
-
-    assert b'EKO{' in Chepy(array_of_arrays).from_lz77().o
-
-    assert Chepy('OMGGGGGG').to_lz77(1).o[1] == [0, 0, 'M']