Prefer libarchive for Rar extraction

Signed-off-by: Philippe Ombredanne <[email protected]>

Author: pombredanne

src/extractcode/archive.py

@@ -411,7 +411,7 @@ def try_to_extract(location, target_dir, extractor):
 extract_springboot = functools.partial(try_to_extract, extractor=extract_zip)
 
 extract_iso = sevenzip.extract
-extract_rar = sevenzip.extract
+extract_rar = libarchive2.extract
 extract_rpm = sevenzip.extract
 extract_xz = sevenzip.extract
 extract_lzma = sevenzip.extract
@@ -703,7 +703,7 @@ def try_to_extract(location, target_dir, extractor):
     extensions=('.rar',),
     kind=regular,
     extractors=[extract_rar],
-    strict=False
+    strict=True
 )
 
 CabHandler = Handler(

src/extractcode/bin/linux-32/bin/Codecs/Rar29.so

@@ -114,6 +114,17 @@ def list_extracted_7z_files(stdout):
     return get_file_list(stdout)
 
 
+def is_rar(location):
+    """
+    Return True if the file at location is a RAR archive.
+    """
+    if not os.path.exists(location):
+        return
+    from typecode import contenttype
+    T = contenttype.get_type(location)
+    return T.filetype_file.lower().startswith('rar archive')
+
+
 def extract(location, target_dir, arch_type='*'):
     """
     Extract all files from a 7zip-supported archive file at location in the
@@ -128,6 +139,9 @@ def extract(location, target_dir, arch_type='*'):
     abs_location = os.path.abspath(os.path.expanduser(location))
     abs_target_dir = os.path.abspath(os.path.expanduser(target_dir))
 
+    if is_rar(location):
+        raise ExtractErrorFailedToExtract('RAR extraction disactivated')
+
     # note: there are some issues with the extraction of debian .deb ar files
     # see sevenzip bug http://sourceforge.net/p/sevenzip/bugs/1472/
 
@@ -189,6 +203,10 @@ def list_entries(location, arch_type='*'):
     """
     assert location
     abs_location = os.path.abspath(os.path.expanduser(location))
+
+    if is_rar(location):
+        return []
+
     # 7z arguments
     listing = 'l'
 

src/extractcode/bin/linux-64/bin/Codecs/Rar29.so

@@ -1596,7 +1596,7 @@ def test_extract_rar_with_trailing_data(self):
     def test_extract_rar_broken(self):
         test_file = self.get_test_loc('archive/rar/broken.rar')
         test_dir = self.get_temp_dir()
-        expected = Exception('Unknown extraction error')
+        expected = Exception('Header CRC error')
         self.assertRaisesInstance(expected, archive.extract_rar, test_file, test_dir)
 
     def test_extract_rar_with_relative_path(self):
@@ -1627,21 +1627,19 @@ def test_extract_rar_with_absolute_path(self):
     def test_extract_rar_with_password(self):
         test_file = self.get_test_loc('archive/rar/rar_password.rar')
         test_dir = self.get_temp_dir()
-        expected = Exception('Password protected archive, unable to extract')
+        expected = Exception('RAR encryption support unavailable.')
         self.assertRaisesInstance(expected, archive.extract_rar,
                                   test_file, test_dir)
 
     def test_extract_rar_with_non_ascii_path(self):
         test_file = self.get_test_loc('archive/rar/non_ascii_corrupted.rar')
-        # The bug only occurs if the path was given as Unicode !
+        # The bug only occurs if the path was given as Unicode
         test_file = unicode(test_file)
         test_dir = self.get_temp_dir()
         # raise an exception but still extracts some
-        expected = Exception('Unknown extraction error')
-        self.assertRaisesInstance(expected, archive.extract_rar,
-                                  test_file, test_dir)
-        result = os.path.join(test_dir, 'EdoProject_java/WebContent'
-                               '/WEB-INF/lib/cos.jar')
+        expected = Exception('Prefix found')
+        self.assertRaisesInstance(expected, archive.extract_rar, test_file, test_dir)
+        result = os.path.join(test_dir, 'EdoProject_java/WebContent/WEB-INF/lib/cos.jar')
         assert os.path.exists(result)
 
 
@@ -2024,21 +2022,22 @@ class TestCbr(BaseArchiveTestCase):
     def test_get_extractor_cbr(self):
         test_file = self.get_test_loc('archive/cbr/t.cbr')
         result = archive.get_extractor(test_file)
-        expected = archive.extract_rar
+        # we do not handle these rare extensions (this is a RAR)
+        expected = None  # archive.extract_rar
         assert expected == result
 
     def test_extract_cbr_basic(self):
         test_file = self.get_test_loc('archive/cbr/t.cbr')
         test_dir = self.get_temp_dir()
-        archive.extract_cab(test_file, test_dir)
+        libarchive2.extract(test_file, test_dir)
         extracted = self.collect_extracted_path(test_dir)
         expected = ['/t/', '/t/t.txt']
         assert expected == extracted
 
     def test_extract_cbr_basic_with_weird_filename_extension(self):
         test_file = self.get_test_loc('archive/cbr/t.cbr.foo')
         test_dir = self.get_temp_dir()
-        archive.extract_cab(test_file, test_dir)
+        libarchive2.extract(test_file, test_dir)
         extracted = self.collect_extracted_path(test_dir)
         expected = ['/t/', '/t/t.txt']
         assert expected == extracted
@@ -2300,6 +2299,7 @@ def test_extract_iso_with_weird_filenames_with_sevenzip(self):
         test_file = self.get_test_loc('archive/weird_names/weird_names.iso')
         self.check_extract(sevenzip.extract, test_file, expected_warnings=[], expected_suffix='7zip')
 
+    @expectedFailure  # not a problem: we now use libarchive for these
     def test_extract_rar_with_weird_filenames_with_sevenzip(self):
         test_file = self.get_test_loc('archive/weird_names/weird_names.rar')
         self.check_extract(sevenzip.extract, test_file, expected_warnings=[], expected_suffix='7zip')