Merge pull request #7 from sanori/dev

Zip password with MBCS encoding

LICENSE.txt

@@ -1,4 +1,4 @@
-Copyright 2016 Joo-Won Jung
+Copyright (c) 2016-2022 Joo-Won Jung
 
 Licensed under the Apache License, Version 2.0 (the "License");
 you may not use this file except in compliance with the License.

README.md

@@ -27,6 +27,8 @@ optional arguments:
   -h, --help            show this help message and exit
   -e ENCODING, --encoding ENCODING
                         character encoding of filename in the .zip
+  -p PASSWORD, --password PASSWORD
+                        password for encrypted .zip
 ```
 
 
@@ -35,11 +37,12 @@ optional arguments:
 Return the information of the files in zip archive `filename`
 with character `encoding`
 
-### extractZip(filename, encoding='utf-8', filters=None)
+### extractZip(filename, encoding='utf-8', filters=None, password=None)
 Extract files in zip archive `filename` on current directory.
 Assume that the file names in zip archive are encoded as `encoding`.
 Only the files prefixed the values of `filters` list are extracted
 if `filters` are provided.
+Use `password` on encrypted zip archive.
 
 ### fixZipFilename(filename, enc)
 Fix `filename` as UNICODE string which is originally encoded as `enc`.
@@ -53,6 +56,6 @@ But, in non-ASCII, non-Western environment, it makes trouble due to filenames.
 Since ZIP format was created too old (1993), there is no standard character encoding about the file name of zip archive entries.
 Most of zip file entries are encoded as legacy character encoding, local charset.
 
-In modern UNICODE based environment or global data processing environment such as Linux, this makes inconvinience, less portability, mangled file names, fail to extract the file, and so on.
+In modern UNICODE based environment or global data processing environment such as Linux, this makes inconvenience, less portability, mangled file names, fail to extract the file, and so on.
 
-This module may mitigate the inconviniences.
+This module may mitigate the inconveniences.

pyproject.toml

@@ -4,13 +4,12 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "unzipmbcs"
-version = "0.1.2"
+version = "0.2.0"
 description = "UnZip for non-UTF8 encoding such as cp949, sjis, gbk, euc-kr, euc-jp, and gb2312"
 readme = "README.md"
-license = { file = "LICENSE.txt" }
 keywords= [
   "unzip", "pkzip", "non-UTF8", "mbcs",
-  "cp949", "sjis", "shift_jis", "gbk", "gb18030"
+  "cp949", "sjis", "shift_jis", "gbk", "gb18030",
 ]
 authors = [
   {name="Joo-Won Jung", email="[email protected]"},

test/test_unzipmbcs.py

@@ -6,33 +6,62 @@
 sys.path.append(os.path.join(os.path.dirname(__file__), '..'))
 import unzipmbcs
 
+def setUpModule():
+    if (sys.getfilesystemencoding().lower() != 'utf-8') and (not os.environ.get('PYTHONIOENCODING')):
+        raise Exception('non-UTF8 filesystem. set PYTHONIOENCODING as your filesystem encoding!')
+
+def clearFiles(fileList):
+    for entry in fileList:
+        if (os.path.isfile(entry)):
+            os.remove(entry)
+        elif (len(os.listdir(entry)) == 0):
+            os.rmdir(entry)
+        entry = os.path.dirname(entry)
+        while entry != '':
+            if len(os.listdir(entry)) > 0:
+                break;
+            os.rmdir(entry)
+            entry = os.path.dirname(entry)
 
 class TestFromZip(unittest.TestCase):
-    filename = 'NewFolder.zip'
+    filename = 'win-default.zip'
     encoding = 'cp949'
-    expected = [u'새 텍스트 문서.txt', u'새 폴더/', u'새 폴더/한글문서.txt']
+    expected = [u'똠방각하.txt', u'한글 디렉토리/새 텍스트 문서.txt']
 
     def testListZip(self):
         result = unzipmbcs.listZip(self.filename, self.encoding)
         self.assertEqual(list(map(lambda x: x[0], result)), self.expected)
 
     def testExtractZip(self):
-        if (sys.getfilesystemencoding() != 'UTF-8') and (not os.environ.get('PYTHONIOENCODING')):
-            print('Warning: non-UTF8 filesystem.',
-                  'set PYTHONIOENCODING as your filesystem encoding!')
-            return
         unzipmbcs.extractZip(self.filename, self.encoding)
         map(lambda x: self.assertTrue(os.path.exists(x), x + ' not exist'),
             self.expected)
+        clearFiles(self.expected)
+
+class TestEncryptedZip(unittest.TestCase):
+    filename = 'lhaplus-zkenc.zip'
+    encoding = 'sjis'
+    expected = [u'ローマ字テキスト.txt', u'秘密/パスワード.txt']
+    password = '全角暗号'
+
+    # ListZip should work without password
+    def testListZip(self):
+        result = unzipmbcs.listZip(self.filename, self.encoding)
+        self.assertEqual(list(map(lambda x: x[0], result)), self.expected)
+
+    def testExtractWithoutPassword(self):
+        with self.assertRaises(RuntimeError):
+            unzipmbcs.extractZip(self.filename, self.encoding)
+
+    def testExtractWithWrongPassword(self):
+        with self.assertRaises(RuntimeError):
+            unzipmbcs.extractZip(self.filename, self.encoding, password='wrongpass')
 
-        # clean-up
-        files = list(self.expected)   # clone the list
-        files.reverse()
-        for f in files:
-            if (os.path.isfile(f)):
-                os.remove(f)
-            else:
-                os.rmdir(f)
+    def testExtractWithPassword(self):
+        unzipmbcs.extractZip(self.filename, self.encoding, password=self.password)
+        map(lambda x: self.assertTrue(os.path.exists(x), x + ' not exist'),
+            self.expected)
+        clearFiles(self.expected)
 
 if __name__ == '__main__':
     unittest.main()

unzipmbcs.py

@@ -1,6 +1,6 @@
 #! python
 ######################## BEGIN LICENSE BLOCK ########################
-# Copyright 2016 Joo-Won Jung
+# Copyright (c) 2016-2022 Joo-Won Jung
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -47,24 +47,35 @@ def fixZipFilename(filename, enc):
             raise e
     return result
 
+def transcodeBytes(str, toEnc, fromEnc='utf-8'):
+    if sys.version_info[0] == 2:
+        return str.decode(fromEnc).encode(toEnc)
+    return bytes(str, toEnc)
 
 def _extractFileFromZip(z, fn, ofn):
     """
     extract a file `fn` in ZipFile `z` as `ofn`
     """
     f = open(ofn, 'wb')
-    f.write(z.read(fn))
+    try:
+        f.write(z.read(fn))
+    except RuntimeError as e:
+        f.close()
+        os.remove(ofn)
+        raise e
     f.close()
 
 
-def extractZip(filename, encoding='utf-8', filters=None):
+def extractZip(filename, encoding='utf-8', filters=None, password=None):
     """
     Extract files in zip archive `filename` on current directory.
     Assume that the file names in zip archive are encoded as `encoding`.
     Only the files prefixed the values of `filters` list are extracted
     if `filters` are provided.
     """
     z = zipfile.ZipFile(filename, 'r')
+    if password:
+        z.setpassword(transcodeBytes(password, encoding))
     l = z.namelist()
     for fn in l:
         if len(fn) == 0 or fn[-1] == '/':
@@ -127,6 +138,9 @@ def _main():
     parser.add_argument('-e', '--encoding',
                         help='character encoding of filename in the .zip',
                         default='utf-8')
+    parser.add_argument('-p', '--password',
+                        help='password for encrypted .zip',
+                        default=None)
     parser.add_argument('zipfile', help='.zip file to unzip')
     parser.add_argument('target', nargs='*',
                         help='file prefix to extract')
@@ -141,7 +155,7 @@ def _main():
                   % tuple([entry[1]] + list(entry[2][:-1]) + [entry[0]]))
     elif args.cmd == 'x':
         extractZip(args.zipfile, encoding=args.encoding,
-                   filters=args.target)
+                   filters=args.target, password=args.password)
     else:
         print('Unknown command:', args.cmd)