Add support for various BCJ filters.

fancycode · Mar 24, 2019 · dec3b88 · dec3b88
1 parent 5d50ae2
commit dec3b88
Show file tree

Hide file tree

Showing 8 changed files with 84 additions and 0 deletions.
diff --git a/py7zlib.py b/py7zlib.py
@@ -141,6 +141,12 @@ def bytes(s, encoding):
 COMPRESSION_METHOD_MISC_BZIP     = unhexlify('0402')  # '\x04\x02'
 COMPRESSION_METHOD_7Z_AES256_SHA256 = unhexlify('06f10701')  # '\x06\xf1\x07\x01'
 COMPRESSION_METHOD_LZMA2         = unhexlify('21')  # '\x21'
+COMPRESSION_METHOD_BCJ           = unhexlify('03030103')  # '\x03\x03\x01\x03'
+COMPRESSION_METHOD_BCJ_PPC       = unhexlify('03030205')  # '\x03\x03\x02\x05'
+COMPRESSION_METHOD_BCJ_IA64      = unhexlify('03030401')  # '\x03\x03\x03\x01'
+COMPRESSION_METHOD_BCJ_ARM       = unhexlify('03030501')  # '\x03\x03\x05\x01'
+COMPRESSION_METHOD_BCJ_ARMT      = unhexlify('03030701')  # '\x03\x03\x07\x01'
+COMPRESSION_METHOD_BCJ_SPARC     = unhexlify('03030805')  # '\x03\x03\x08\x05'
 COMPRESSION_METHOD_BCJ2          = unhexlify('0303011B')  # '\x03\x03\x01\x1B'
 
 FILE_ATTRIBUTE_DIRECTORY = 0x10
@@ -604,6 +610,12 @@ def __init__(self, info, start, src_start, folder, archive, maxsize=None):
             COMPRESSION_METHOD_MISC_ZIP: '_read_zip',
             COMPRESSION_METHOD_MISC_BZIP: '_read_bzip',
             COMPRESSION_METHOD_7Z_AES256_SHA256: '_read_7z_aes256_sha256',
+            COMPRESSION_METHOD_BCJ: '_read_bcj',
+            COMPRESSION_METHOD_BCJ_PPC: '_read_bcj_ppc',
+            COMPRESSION_METHOD_BCJ_IA64: '_read_bcj_ia64',
+            COMPRESSION_METHOD_BCJ_ARM: '_read_bcj_arm',
+            COMPRESSION_METHOD_BCJ_ARMT: '_read_bcj_armt',
+            COMPRESSION_METHOD_BCJ_SPARC: '_read_bcj_sparc',
             COMPRESSION_METHOD_BCJ2: '_read_bcj2',
         }
 
@@ -784,6 +796,54 @@ def _read_bcj2(self, coder, input, level, num_coders):
         rc_data = self._file.read(self._packsizes[3])
         return pylzma.bcj2_decode(main_data, call_data, jump_data, rc_data, self.uncompressed)
 
+    def _read_bcj(self, coder, input, level, num_coders):
+        size = self._uncompressed[level]
+        if not input:
+            self._file.seek(self._src_start)
+            input = self._file.read(size)
+        data = pylzma.bcj_x86_convert(input)
+        return data[self._start:self._start+size]
+
+    def _read_bcj_ppc(self, coder, input, level, num_coders):
+        size = self._uncompressed[level]
+        if not input:
+            self._file.seek(self._src_start)
+            input = self._file.read(size)
+        data = pylzma.bcj_ppc_convert(input)
+        return data[self._start:self._start+size]
+
+    def _read_bcj_ia64(self, coder, input, level, num_coders):
+        size = self._uncompressed[level]
+        if not input:
+            self._file.seek(self._src_start)
+            input = self._file.read(size)
+        data = pylzma.bcj_ia64_convert(input)
+        return data[self._start:self._start+size]
+
+    def _read_bcj_arm(self, coder, input, level, num_coders):
+        size = self._uncompressed[level]
+        if not input:
+            self._file.seek(self._src_start)
+            input = self._file.read(size)
+        data = pylzma.bcj_arm_convert(input)
+        return data[self._start:self._start+size]
+
+    def _read_bcj_armt(self, coder, input, level, num_coders):
+        size = self._uncompressed[level]
+        if not input:
+            self._file.seek(self._src_start)
+            input = self._file.read(size)
+        data = pylzma.bcj_armt_convert(input)
+        return data[self._start:self._start+size]
+
+    def _read_bcj_sparc(self, coder, input, level, num_coders):
+        size = self._uncompressed[level]
+        if not input:
+            self._file.seek(self._src_start)
+            input = self._file.read(size)
+        data = pylzma.bcj_sparc_convert(input)
+        return data[self._start:self._start+size]
+
     def checkcrc(self):
         if self.digest is None:
             return True

diff --git a/tests/data/lzma_bcj.7z b/tests/data/lzma_bcj.7z
diff --git a/tests/data/lzma_bcj_arm.7z b/tests/data/lzma_bcj_arm.7z
diff --git a/tests/data/lzma_bcj_armt.7z b/tests/data/lzma_bcj_armt.7z
diff --git a/tests/data/lzma_bcj_ia64.7z b/tests/data/lzma_bcj_ia64.7z
diff --git a/tests/data/lzma_bcj_ppc.7z b/tests/data/lzma_bcj_ppc.7z
diff --git a/tests/data/lzma_bcj_sparc.7z b/tests/data/lzma_bcj_sparc.7z
diff --git a/tests/test_7zfiles.py b/tests/test_7zfiles.py
@@ -201,6 +201,30 @@ def test_copy(self):
         # test loading of copy compressed files
         self._test_archive('copy.7z')
 
+    def test_lzma_bcj(self):
+        # test loading of lzma compressed, filtered through bcj
+        self._test_archive('lzma_bcj.7z')
+
+    def test_lzma_bcj_arm(self):
+        # test loading of lzma compressed, filtered through bcj / ARM
+        self._test_archive('lzma_bcj_arm.7z')
+
+    def test_lzma_bcj_armt(self):
+        # test loading of lzma compressed, filtered through bcj / ARMT
+        self._test_archive('lzma_bcj_armt.7z')
+
+    def test_lzma_bcj_ia64(self):
+        # test loading of lzma compressed, filtered through bcj / IA64
+        self._test_archive('lzma_bcj_ia64.7z')
+
+    def test_lzma_bcj_ppc(self):
+        # test loading of lzma compressed, filtered through bcj / PPC
+        self._test_archive('lzma_bcj_ppc.7z')
+
+    def test_lzma_bcj_sparc(self):
+        # test loading of lzma compressed, filtered through bcj / SPARC
+        self._test_archive('lzma_bcj_sparc.7z')
+
     def test_regress_1(self):
         # prevent regression bug #1 reported by mail
         fp = self._open_file(os.path.join(ROOT, 'data', 'regress_1.7z'), 'rb')