Merge pull request #251 from miurahr/backport-lzma-bcj-249

Backport BCJ filter fix

Changelog.rst

@@ -25,6 +25,14 @@ Removed
 Security
 --------
 
+`v0.9.9`_
+=========
+
+Fixed
+-----
+
+* (backport) Fix BCJ filter issue failing a certain data when LZMA+BCJ compression(#240, #250)
+
 `v0.9.8`_
 =========
 
@@ -774,7 +782,8 @@ Changed
 
 
 .. History links
-.. _Unreleased: https://github.com/miurahr/py7zr/compare/v0.9.8...HEAD
+.. _Unreleased: https://github.com/miurahr/py7zr/compare/v0.9.9...HEAD
+.. _v0.9.9: https://github.com/miurahr/py7zr/compare/v0.9.8...v0.9.9
 .. _v0.9.8: https://github.com/miurahr/py7zr/compare/v0.9.7...v0.9.8
 .. _v0.9.7: https://github.com/miurahr/py7zr/compare/v0.9.6...v0.9.7
 .. _v0.9.6: https://github.com/miurahr/py7zr/compare/v0.9.5...v0.9.6

py7zr/compressor.py

@@ -379,6 +379,11 @@ def _arm_code(self) -> int:
         return i
 
     def _x86_code(self) -> int:
+        """
+        The code algorithm from liblzma/simple/x86.c
+        It is slightly different from LZMA-SDK's bra86.c
+        :return: buffer position
+        """
         size = len(self.buffer)
         if size < 5:
             return 0
@@ -429,6 +434,8 @@ def _x86_code(self) -> int:
             else:
                 buffer_pos += 1
                 self.prev_mask |= 1
+                if self.buffer[buffer_pos + 3] in [0, 0xff]:
+                    self.prev_mask |= 0x10
         self.current_position += buffer_pos
         return buffer_pos
 

tests/test_misc.py

@@ -0,0 +1,15 @@
+import os
+import pathlib
+
+import pytest
+
+import py7zr
+
+testdata_path = pathlib.Path(os.path.dirname(__file__)).joinpath('data')
+os.umask(0o022)
+
+
+@pytest.mark.files
+def test_bcj_file(tmp_path):
+    with py7zr.SevenZipFile(testdata_path.joinpath('copy_bcj_1.7z').open(mode='rb')) as ar:
+        ar.extractall(tmp_path)