separate closing from deferred closing for streams.. make context work...

don't leak file in SeekableArchive..

libarchive/__init__.py

@@ -24,6 +24,7 @@
 # SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 
 import os
+import pathlib
 import stat
 import sys
 import time
@@ -200,7 +201,7 @@ class EntryReadStream(object):
         return self
 
     def __exit__(self, *args):
-        return
+        self.close()
 
     def __iter__(self):
         if self.closed:
@@ -438,17 +439,20 @@ class Archive(object):
         self.encoding = encoding
         self.blocksize = blocksize
         self.password = password
+        if isinstance(f, pathlib.PurePath):
+            f = str(f)
         if isinstance(f, str):
             self.filename = f
             f = open(f, mode)
             # Only close it if we opened it...
-            self._defer_close = True
+            self._doclose = True
         elif hasattr(f, 'fileno'):
             self.filename = getattr(f, 'name', None)
             # Leave the fd alone, caller should manage it...
-            self._defer_close = False
+            self._doclose = False
         else:
             raise Exception('Provided file is not path or open file.')
+        self._defer_close = False
         self.f = f
         self.mode = mode
         # Guess the format/filter from file name (if not provided)
@@ -493,7 +497,7 @@ class Archive(object):
         return self
 
     def __exit__(self, type, value, traceback):
-        self.denit()
+        self.close()
 
     def __del__(self):
         self.close()
@@ -563,7 +567,7 @@ class Archive(object):
                 if hasattr(self.f, "fileno"):
                     os.fsync(self.f.fileno())
             # and then close it, if we opened it...
-            if getattr(self, '_close', None):
+            if self._doclose and getattr(self.f, 'close', None):
                 self.f.close()
 
     @property
@@ -665,8 +669,6 @@ class SeekableArchive(Archive):
         self._stream = None
         # Convert file to open file. We need this to reopen the archive.
         mode = kwargs.setdefault('mode', 'r')
-        if isinstance(f, str):
-            f = open(f, mode)
         super(SeekableArchive, self).__init__(f, **kwargs)
         self.entries = []
         self.eof = False

tests.py

@@ -27,7 +27,9 @@
 # SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 
 import os, unittest, tempfile, random, string, sys
+import hashlib
 import io
+import pathlib
 import shutil
 import zipfile
 
@@ -368,6 +370,51 @@ class TestHighLevelAPI(unittest.TestCase, MakeTempMixIn):
             with io.FileIO(zf.fileno(), mode='r', closefd=False) as f:
                 self._test_listing_content(f)
 
+_defaulthash = 'sha512'
+
+def _readfp(fp):
+    while True:
+        r = fp.read(64*1024)
+        # libarchive returns None on EOF
+        if r == b'' or r is None:
+            return
+
+        yield r
+
+def _hashfp(fp):
+    hash = getattr(hashlib, _defaulthash)()
+    for r in _readfp(fp):
+        hash.update(r)
+
+    return '%s:%s' % (_defaulthash, hash.hexdigest())
+
+
+class TestArchive(unittest.TestCase):
+    def setUp(self):
+        self.fixtures = pathlib.Path(__file__).parent / 'fixtures'
+
+    def test_closed(self):
+        fname = self.fixtures / 'testfile.tar.gz'
+
+        with Archive(fname) as arch:
+            origfp = arch.f
+
+            hashes = []
+
+            for i in arch:
+                if not i.isfile():
+                    continue
+
+                with arch.readstream(i.size) as fp:
+                    hashes.append(_hashfp(fp))
+
+                self.assertTrue(fp.closed)
+                self.assertIsNone(arch._stream)
+
+            self.assertEqual(hashes, [ 'sha512:90f8342520f0ac57fb5a779f5d331c2fa87aa40f8799940257f9ba619940951e67143a8d746535ed0284924b2b7bc1478f095198800ba96d01847d7b56ca465c', 'sha512:7d5768d47b6bc27dc4fa7e9732cfa2de506ca262a2749cb108923e5dddffde842bbfee6cb8d692fb43aca0f12946c521cce2633887914ca1f96898478d10ad3f' ])
+
+        self.assertTrue(arch.f.closed)
+
 
 if __name__ == '__main__':
     unittest.main()