Document zip file encoding issues

Signed-off-by: Roberto Rosario <roberto.rosario@mayan-edms.com>

mayan/apps/common/compressed_files.py

@@ -11,6 +11,7 @@ except ImportError:
     COMPRESSION = zipfile.ZIP_STORED
 
 from django.core.files.uploadedfile import SimpleUploadedFile
+from django.utils.encoding import force_text
 
 from mayan.apps.mimetype.api import get_mimetype
 
@@ -136,9 +137,32 @@ class ZipArchive(Archive):
         return self._archive.read(filename)
 
     def members(self):
-        return [
-            filename for filename in self._archive.namelist() if not filename.endswith('/')
-        ]
+        results = []
+        from django.utils.encoding import force_str
+
+        for filename in self._archive.namelist():
+            # Zip files only support UTF-8 and CP437 encodings.
+            # Attempt to decode CP437 to be able to check if it ends
+            # with a slash.
+            # Future improvement that violates the Zip format:
+            # Add chardet.detect to detect the most likely encoding
+            # if other than CP437.
+            try:
+                filename = filename.decode('CP437')
+                is_unicode = False
+            except UnicodeEncodeError:
+                is_unicode = True
+
+            if not filename.endswith('/'):
+                # Re encode in the original encoding
+                if not is_unicode:
+                    filename = filename.encode(
+                        encoding='CP437', errors='strict'
+                    )
+
+                results.append(filename)
+
+        return results
 
     def open_member(self, filename):
         return self._archive.open(filename)

mayan/apps/common/tests/literals.py

@@ -12,6 +12,7 @@ TEST_VIEW_URL = 'test-view-url'
 
 # Filenames
 TEST_ARCHIVE_ZIP_SPECIAL_CHARACTERS_FILENAME_MEMBER = 'test_archvive_with_special_characters_filename_member.zip'
+TEST_ARCHIVE_ZIP_CP437_MEMBER = 'test_archvive_with_cp437_member.zip'
 TEST_FILENAME1 = 'test_file1.txt'
 TEST_FILENAME2 = 'test_file2.txt'
 TEST_FILENAME3 = 'test_file3.txt'
@@ -28,6 +29,10 @@ TEST_ARCHIVE_ZIP_SPECIAL_CHARACTERS_FILENAME_MEMBER_PATH = os.path.join(
     settings.BASE_DIR, 'apps', 'common', 'tests', 'contrib',
     TEST_ARCHIVE_ZIP_SPECIAL_CHARACTERS_FILENAME_MEMBER
 )
+TEST_ARCHIVE_ZIP_CP437_MEMBER_PATH = os.path.join(
+    settings.BASE_DIR, 'apps', 'common', 'tests', 'contrib',
+    TEST_ARCHIVE_ZIP_CP437_MEMBER
+)
 TEST_FILE3_PATH = os.path.join(
     settings.BASE_DIR, 'apps', 'common', 'tests', 'contrib', TEST_FILENAME3
 )

mayan/apps/common/tests/test_compressed_files.py

@@ -5,6 +5,7 @@ from mayan.apps.common.tests import BaseTestCase
 from ..compressed_files import Archive, TarArchive, ZipArchive
 
 from .literals import (
+    TEST_ARCHIVE_ZIP_CP437_MEMBER_PATH,
     TEST_ARCHIVE_ZIP_SPECIAL_CHARACTERS_FILENAME_MEMBER_PATH,
     TEST_COMPRESSED_FILE_CONTENTS, TEST_FILE_CONTENTS_1, TEST_FILE3_PATH,
     TEST_FILENAME1, TEST_FILENAME3, TEST_TAR_BZ2_FILE_PATH,
@@ -64,6 +65,11 @@ class ZipArchiveClassTestCase(TarArchiveClassTestCase):
             archive = Archive.open(file_object=file_object)
             list(archive.get_members())
 
+    def test_open_cp437_member(self):
+        with open(TEST_ARCHIVE_ZIP_CP437_MEMBER_PATH, mode='rb') as file_object:
+            archive = Archive.open(file_object=file_object)
+            list(archive.get_members())
+
 
 class TarGzArchiveClassTestCase(TarArchiveClassTestCase):
     archive_path = TEST_TAR_GZ_FILE_PATH