first commit

.venv/Lib/site-packages/gitdb/test/__init__.py

@@ -0,0 +1,4 @@
+# Copyright (C) 2010, 2011 Sebastian Thiel (byronimo@gmail.com) and contributors
+#
+# This module is part of GitDB and is released under
+# the New BSD License: http://www.opensource.org/licenses/bsd-license.php

.venv/Lib/site-packages/gitdb/test/lib.py

@@ -0,0 +1,207 @@
+# Copyright (C) 2010, 2011 Sebastian Thiel (byronimo@gmail.com) and contributors
+#
+# This module is part of GitDB and is released under
+# the New BSD License: http://www.opensource.org/licenses/bsd-license.php
+"""Utilities used in ODB testing"""
+from gitdb import OStream
+
+import sys
+import random
+from array import array
+
+from io import BytesIO
+
+import glob
+import unittest
+import tempfile
+import shutil
+import os
+import gc
+import logging
+from functools import wraps
+
+
+#{ Bases
+
+class TestBase(unittest.TestCase):
+    """Base class for all tests
+
+    TestCase providing access to readonly repositories using the following member variables.
+
+    * gitrepopath
+
+     * read-only base path of the git source repository, i.e. .../git/.git
+    """
+
+    #{ Invvariants
+    k_env_git_repo = "GITDB_TEST_GIT_REPO_BASE"
+    #} END invariants
+
+    @classmethod
+    def setUpClass(cls):
+        try:
+            super(TestBase, cls).setUpClass()
+        except AttributeError:
+            pass
+
+        cls.gitrepopath = os.environ.get(cls.k_env_git_repo)
+        if not cls.gitrepopath:
+            logging.info(
+                "You can set the %s environment variable to a .git repository of your choice - defaulting to the gitdb repository", cls.k_env_git_repo)
+            ospd = os.path.dirname
+            cls.gitrepopath = os.path.join(ospd(ospd(ospd(__file__))), '.git')
+        # end assure gitrepo is set
+        assert cls.gitrepopath.endswith('.git')
+
+
+#} END bases
+
+#{ Decorators
+
+def skip_on_travis_ci(func):
+    """All tests decorated with this one will raise SkipTest when run on travis ci.
+    Use it to workaround difficult to solve issues
+    NOTE: copied from bcore (https://github.com/Byron/bcore)"""
+    @wraps(func)
+    def wrapper(self, *args, **kwargs):
+        if 'TRAVIS' in os.environ:
+            import nose
+            raise nose.SkipTest("Cannot run on travis-ci")
+        # end check for travis ci
+        return func(self, *args, **kwargs)
+    # end wrapper
+    return wrapper
+
+
+def with_rw_directory(func):
+    """Create a temporary directory which can be written to, remove it if the
+    test succeeds, but leave it otherwise to aid additional debugging"""
+
+    def wrapper(self):
+        path = tempfile.mktemp(prefix=func.__name__)
+        os.mkdir(path)
+        keep = False
+        try:
+            try:
+                return func(self, path)
+            except Exception:
+                sys.stderr.write("Test {}.{} failed, output is at {!r}\n".format(type(self).__name__, func.__name__, path))
+                keep = True
+                raise
+        finally:
+            # Need to collect here to be sure all handles have been closed. It appears
+            # a windows-only issue. In fact things should be deleted, as well as
+            # memory maps closed, once objects go out of scope. For some reason
+            # though this is not the case here unless we collect explicitly.
+            if not keep:
+                gc.collect()
+                shutil.rmtree(path)
+        # END handle exception
+    # END wrapper
+
+    wrapper.__name__ = func.__name__
+    return wrapper
+
+
+def with_packs_rw(func):
+    """Function that provides a path into which the packs for testing should be
+    copied. Will pass on the path to the actual function afterwards"""
+
+    def wrapper(self, path):
+        src_pack_glob = fixture_path('packs/*')
+        copy_files_globbed(src_pack_glob, path, hard_link_ok=True)
+        return func(self, path)
+    # END wrapper
+
+    wrapper.__name__ = func.__name__
+    return wrapper
+
+#} END decorators
+
+#{ Routines
+
+
+def fixture_path(relapath=''):
+    """:return: absolute path into the fixture directory
+    :param relapath: relative path into the fixtures directory, or ''
+        to obtain the fixture directory itself"""
+    return os.path.join(os.path.dirname(__file__), 'fixtures', relapath)
+
+
+def copy_files_globbed(source_glob, target_dir, hard_link_ok=False):
+    """Copy all files found according to the given source glob into the target directory
+    :param hard_link_ok: if True, hard links will be created if possible. Otherwise
+        the files will be copied"""
+    for src_file in glob.glob(source_glob):
+        if hard_link_ok and hasattr(os, 'link'):
+            target = os.path.join(target_dir, os.path.basename(src_file))
+            try:
+                os.link(src_file, target)
+            except OSError:
+                shutil.copy(src_file, target_dir)
+            # END handle cross device links ( and resulting failure )
+        else:
+            shutil.copy(src_file, target_dir)
+        # END try hard link
+    # END for each file to copy
+
+
+def make_bytes(size_in_bytes, randomize=False):
+    """:return: string with given size in bytes
+    :param randomize: try to produce a very random stream"""
+    actual_size = size_in_bytes // 4
+    producer = range(actual_size)
+    if randomize:
+        producer = list(producer)
+        random.shuffle(producer)
+    # END randomize
+    a = array('i', producer)
+    return a.tobytes()
+
+
+def make_object(type, data):
+    """:return: bytes resembling an uncompressed object"""
+    odata = "blob %i\0" % len(data)
+    return odata.encode("ascii") + data
+
+
+def make_memory_file(size_in_bytes, randomize=False):
+    """:return: tuple(size_of_stream, stream)
+    :param randomize: try to produce a very random stream"""
+    d = make_bytes(size_in_bytes, randomize)
+    return len(d), BytesIO(d)
+
+#} END routines
+
+#{ Stream Utilities
+
+
+class DummyStream(object):
+
+    def __init__(self):
+        self.was_read = False
+        self.bytes = 0
+        self.closed = False
+
+    def read(self, size):
+        self.was_read = True
+        self.bytes = size
+
+    def close(self):
+        self.closed = True
+
+    def _assert(self):
+        assert self.was_read
+
+
+class DeriveTest(OStream):
+
+    def __init__(self, sha, type, size, stream, *args, **kwargs):
+        self.myarg = kwargs.pop('myarg')
+        self.args = args
+
+    def _assert(self):
+        assert self.args
+        assert self.myarg
+
+#} END stream utilitiess

.venv/Lib/site-packages/gitdb/test/test_base.py

@@ -0,0 +1,105 @@
+# Copyright (C) 2010, 2011 Sebastian Thiel (byronimo@gmail.com) and contributors
+#
+# This module is part of GitDB and is released under
+# the New BSD License: http://www.opensource.org/licenses/bsd-license.php
+"""Test for object db"""
+from gitdb.test.lib import (
+    TestBase,
+    DummyStream,
+    DeriveTest,
+)
+
+from gitdb import (
+    OInfo,
+    OPackInfo,
+    ODeltaPackInfo,
+    OStream,
+    OPackStream,
+    ODeltaPackStream,
+    IStream
+)
+from gitdb.util import (
+    NULL_BIN_SHA
+)
+
+from gitdb.typ import (
+    str_blob_type
+)
+
+
+class TestBaseTypes(TestBase):
+
+    def test_streams(self):
+        # test info
+        sha = NULL_BIN_SHA
+        s = 20
+        blob_id = 3
+
+        info = OInfo(sha, str_blob_type, s)
+        assert info.binsha == sha
+        assert info.type == str_blob_type
+        assert info.type_id == blob_id
+        assert info.size == s
+
+        # test pack info
+        # provides type_id
+        pinfo = OPackInfo(0, blob_id, s)
+        assert pinfo.type == str_blob_type
+        assert pinfo.type_id == blob_id
+        assert pinfo.pack_offset == 0
+
+        dpinfo = ODeltaPackInfo(0, blob_id, s, sha)
+        assert dpinfo.type == str_blob_type
+        assert dpinfo.type_id == blob_id
+        assert dpinfo.delta_info == sha
+        assert dpinfo.pack_offset == 0
+
+        # test ostream
+        stream = DummyStream()
+        ostream = OStream(*(info + (stream, )))
+        assert ostream.stream is stream
+        ostream.read(15)
+        stream._assert()
+        assert stream.bytes == 15
+        ostream.read(20)
+        assert stream.bytes == 20
+
+        # test packstream
+        postream = OPackStream(*(pinfo + (stream, )))
+        assert postream.stream is stream
+        postream.read(10)
+        stream._assert()
+        assert stream.bytes == 10
+
+        # test deltapackstream
+        dpostream = ODeltaPackStream(*(dpinfo + (stream, )))
+        dpostream.stream is stream
+        dpostream.read(5)
+        stream._assert()
+        assert stream.bytes == 5
+
+        # derive with own args
+        DeriveTest(sha, str_blob_type, s, stream, 'mine', myarg=3)._assert()
+
+        # test istream
+        istream = IStream(str_blob_type, s, stream)
+        assert istream.binsha == None
+        istream.binsha = sha
+        assert istream.binsha == sha
+
+        assert len(istream.binsha) == 20
+        assert len(istream.hexsha) == 40
+
+        assert istream.size == s
+        istream.size = s * 2
+        istream.size == s * 2
+        assert istream.type == str_blob_type
+        istream.type = "something"
+        assert istream.type == "something"
+        assert istream.stream is stream
+        istream.stream = None
+        assert istream.stream is None
+
+        assert istream.error is None
+        istream.error = Exception()
+        assert isinstance(istream.error, Exception)

.venv/Lib/site-packages/gitdb/test/test_example.py

@@ -0,0 +1,43 @@
+# Copyright (C) 2010, 2011 Sebastian Thiel (byronimo@gmail.com) and contributors
+#
+# This module is part of GitDB and is released under
+# the New BSD License: http://www.opensource.org/licenses/bsd-license.php
+"""Module with examples from the tutorial section of the docs"""
+import os
+from gitdb.test.lib import TestBase
+from gitdb import IStream
+from gitdb.db import LooseObjectDB
+
+from io import BytesIO
+
+
+class TestExamples(TestBase):
+
+    def test_base(self):
+        ldb = LooseObjectDB(os.path.join(self.gitrepopath, 'objects'))
+
+        for sha1 in ldb.sha_iter():
+            oinfo = ldb.info(sha1)
+            ostream = ldb.stream(sha1)
+            assert oinfo[:3] == ostream[:3]
+
+            assert len(ostream.read()) == ostream.size
+            assert ldb.has_object(oinfo.binsha)
+        # END for each sha in database
+        # assure we close all files
+        try:
+            del(ostream)
+            del(oinfo)
+        except UnboundLocalError:
+            pass
+        # END ignore exception if there are no loose objects
+
+        data = "my data".encode("ascii")
+        istream = IStream("blob", len(data), BytesIO(data))
+
+        # the object does not yet have a sha
+        assert istream.binsha is None
+        ldb.store(istream)
+        # now the sha is set
+        assert len(istream.binsha) == 20
+        assert ldb.has_object(istream.binsha)

.venv/Lib/site-packages/gitdb/test/test_pack.py

@@ -0,0 +1,249 @@
+# Copyright (C) 2010, 2011 Sebastian Thiel (byronimo@gmail.com) and contributors
+#
+# This module is part of GitDB and is released under
+# the New BSD License: http://www.opensource.org/licenses/bsd-license.php
+"""Test everything about packs reading and writing"""
+from gitdb.test.lib import (
+    TestBase,
+    with_rw_directory,
+    fixture_path
+)
+
+from gitdb.stream import DeltaApplyReader
+
+from gitdb.pack import (
+    PackEntity,
+    PackIndexFile,
+    PackFile
+)
+
+from gitdb.base import (
+    OInfo,
+    OStream,
+)
+
+from gitdb.fun import delta_types
+from gitdb.exc import UnsupportedOperation
+from gitdb.util import to_bin_sha
+
+from nose import SkipTest
+
+import os
+import tempfile
+
+
+#{ Utilities
+def bin_sha_from_filename(filename):
+    return to_bin_sha(os.path.splitext(os.path.basename(filename))[0][5:])
+#} END utilities
+
+
+class TestPack(TestBase):
+
+    packindexfile_v1 = (fixture_path('packs/pack-c0438c19fb16422b6bbcce24387b3264416d485b.idx'), 1, 67)
+    packindexfile_v2 = (fixture_path('packs/pack-11fdfa9e156ab73caae3b6da867192221f2089c2.idx'), 2, 30)
+    packindexfile_v2_3_ascii = (fixture_path('packs/pack-a2bf8e71d8c18879e499335762dd95119d93d9f1.idx'), 2, 42)
+    packfile_v2_1 = (fixture_path('packs/pack-c0438c19fb16422b6bbcce24387b3264416d485b.pack'), 2, packindexfile_v1[2])
+    packfile_v2_2 = (fixture_path('packs/pack-11fdfa9e156ab73caae3b6da867192221f2089c2.pack'), 2, packindexfile_v2[2])
+    packfile_v2_3_ascii = (
+        fixture_path('packs/pack-a2bf8e71d8c18879e499335762dd95119d93d9f1.pack'), 2, packindexfile_v2_3_ascii[2])
+
+    def _assert_index_file(self, index, version, size):
+        assert index.packfile_checksum() != index.indexfile_checksum()
+        assert len(index.packfile_checksum()) == 20
+        assert len(index.indexfile_checksum()) == 20
+        assert index.version() == version
+        assert index.size() == size
+        assert len(index.offsets()) == size
+
+        # get all data of all objects
+        for oidx in range(index.size()):
+            sha = index.sha(oidx)
+            assert oidx == index.sha_to_index(sha)
+
+            entry = index.entry(oidx)
+            assert len(entry) == 3
+
+            assert entry[0] == index.offset(oidx)
+            assert entry[1] == sha
+            assert entry[2] == index.crc(oidx)
+
+            # verify partial sha
+            for l in (4, 8, 11, 17, 20):
+                assert index.partial_sha_to_index(sha[:l], l * 2) == oidx
+
+        # END for each object index in indexfile
+        self.assertRaises(ValueError, index.partial_sha_to_index, "\0", 2)
+
+    def _assert_pack_file(self, pack, version, size):
+        assert pack.version() == 2
+        assert pack.size() == size
+        assert len(pack.checksum()) == 20
+
+        num_obj = 0
+        for obj in pack.stream_iter():
+            num_obj += 1
+            info = pack.info(obj.pack_offset)
+            stream = pack.stream(obj.pack_offset)
+
+            assert info.pack_offset == stream.pack_offset
+            assert info.type_id == stream.type_id
+            assert hasattr(stream, 'read')
+
+            # it should be possible to read from both streams
+            assert obj.read() == stream.read()
+
+            streams = pack.collect_streams(obj.pack_offset)
+            assert streams
+
+            # read the stream
+            try:
+                dstream = DeltaApplyReader.new(streams)
+            except ValueError:
+                # ignore these, old git versions use only ref deltas,
+                # which we havent resolved ( as we are without an index )
+                # Also ignore non-delta streams
+                continue
+            # END get deltastream
+
+            # read all
+            data = dstream.read()
+            assert len(data) == dstream.size
+
+            # test seek
+            dstream.seek(0)
+            assert dstream.read() == data
+
+            # read chunks
+            # NOTE: the current implementation is safe, it basically transfers
+            # all calls to the underlying memory map
+
+        # END for each object
+        assert num_obj == size
+
+    def test_pack_index(self):
+        # check version 1 and 2
+        for indexfile, version, size in (self.packindexfile_v1, self.packindexfile_v2):
+            index = PackIndexFile(indexfile)
+            self._assert_index_file(index, version, size)
+        # END run tests
+
+    def test_pack(self):
+        # there is this special version 3, but apparently its like 2 ...
+        for packfile, version, size in (self.packfile_v2_3_ascii, self.packfile_v2_1, self.packfile_v2_2):
+            pack = PackFile(packfile)
+            self._assert_pack_file(pack, version, size)
+        # END for each pack to test
+
+    @with_rw_directory
+    def test_pack_entity(self, rw_dir):
+        pack_objs = list()
+        for packinfo, indexinfo in ((self.packfile_v2_1, self.packindexfile_v1),
+                                    (self.packfile_v2_2, self.packindexfile_v2),
+                                    (self.packfile_v2_3_ascii, self.packindexfile_v2_3_ascii)):
+            packfile, version, size = packinfo
+            indexfile, version, size = indexinfo
+            entity = PackEntity(packfile)
+            assert entity.pack().path() == packfile
+            assert entity.index().path() == indexfile
+            pack_objs.extend(entity.stream_iter())
+
+            count = 0
+            for info, stream in zip(entity.info_iter(), entity.stream_iter()):
+                count += 1
+                assert info.binsha == stream.binsha
+                assert len(info.binsha) == 20
+                assert info.type_id == stream.type_id
+                assert info.size == stream.size
+
+                # we return fully resolved items, which is implied by the sha centric access
+                assert not info.type_id in delta_types
+
+                # try all calls
+                assert len(entity.collect_streams(info.binsha))
+                oinfo = entity.info(info.binsha)
+                assert isinstance(oinfo, OInfo)
+                assert oinfo.binsha is not None
+                ostream = entity.stream(info.binsha)
+                assert isinstance(ostream, OStream)
+                assert ostream.binsha is not None
+
+                # verify the stream
+                try:
+                    assert entity.is_valid_stream(info.binsha, use_crc=True)
+                except UnsupportedOperation:
+                    pass
+                # END ignore version issues
+                assert entity.is_valid_stream(info.binsha, use_crc=False)
+            # END for each info, stream tuple
+            assert count == size
+
+        # END for each entity
+
+        # pack writing - write all packs into one
+        # index path can be None
+        pack_path1 = tempfile.mktemp('', "pack1", rw_dir)
+        pack_path2 = tempfile.mktemp('', "pack2", rw_dir)
+        index_path = tempfile.mktemp('', 'index', rw_dir)
+        iteration = 0
+
+        def rewind_streams():
+            for obj in pack_objs:
+                obj.stream.seek(0)
+        # END utility
+        for ppath, ipath, num_obj in zip((pack_path1, pack_path2),
+                                         (index_path, None),
+                                         (len(pack_objs), None)):
+            iwrite = None
+            if ipath:
+                ifile = open(ipath, 'wb')
+                iwrite = ifile.write
+            # END handle ip
+
+            # make sure we rewind the streams ... we work on the same objects over and over again
+            if iteration > 0:
+                rewind_streams()
+            # END rewind streams
+            iteration += 1
+
+            with open(ppath, 'wb') as pfile:
+                pack_sha, index_sha = PackEntity.write_pack(pack_objs, pfile.write, iwrite, object_count=num_obj)
+            assert os.path.getsize(ppath) > 100
+
+            # verify pack
+            pf = PackFile(ppath)
+            assert pf.size() == len(pack_objs)
+            assert pf.version() == PackFile.pack_version_default
+            assert pf.checksum() == pack_sha
+            pf.close()
+
+            # verify index
+            if ipath is not None:
+                ifile.close()
+                assert os.path.getsize(ipath) > 100
+                idx = PackIndexFile(ipath)
+                assert idx.version() == PackIndexFile.index_version_default
+                assert idx.packfile_checksum() == pack_sha
+                assert idx.indexfile_checksum() == index_sha
+                assert idx.size() == len(pack_objs)
+                idx.close()
+            # END verify files exist
+        # END for each packpath, indexpath pair
+
+        # verify the packs thoroughly
+        rewind_streams()
+        entity = PackEntity.create(pack_objs, rw_dir)
+        count = 0
+        for info in entity.info_iter():
+            count += 1
+            for use_crc in range(2):
+                assert entity.is_valid_stream(info.binsha, use_crc)
+            # END for each crc mode
+        # END for each info
+        assert count == len(pack_objs)
+        entity.close()
+        
+    def test_pack_64(self):
+        # TODO: hex-edit a pack helping us to verify that we can handle 64 byte offsets
+        # of course without really needing such a huge pack
+        raise SkipTest()

.venv/Lib/site-packages/gitdb/test/test_stream.py

@@ -0,0 +1,164 @@
+# Copyright (C) 2010, 2011 Sebastian Thiel (byronimo@gmail.com) and contributors
+#
+# This module is part of GitDB and is released under
+# the New BSD License: http://www.opensource.org/licenses/bsd-license.php
+"""Test for object db"""
+
+from gitdb.test.lib import (
+    TestBase,
+    DummyStream,
+    make_bytes,
+    make_object,
+    fixture_path
+)
+
+from gitdb import (
+    DecompressMemMapReader,
+    FDCompressedSha1Writer,
+    LooseObjectDB,
+    Sha1Writer,
+    MemoryDB,
+    IStream,
+)
+from gitdb.util import hex_to_bin
+
+import zlib
+from gitdb.typ import (
+    str_blob_type
+)
+
+import tempfile
+import os
+from io import BytesIO
+
+
+class TestStream(TestBase):
+
+    """Test stream classes"""
+
+    data_sizes = (15, 10000, 1000 * 1024 + 512)
+
+    def _assert_stream_reader(self, stream, cdata, rewind_stream=lambda s: None):
+        """Make stream tests - the orig_stream is seekable, allowing it to be
+        rewound and reused
+        :param cdata: the data we expect to read from stream, the contents
+        :param rewind_stream: function called to rewind the stream to make it ready
+            for reuse"""
+        ns = 10
+        assert len(cdata) > ns - 1, "Data must be larger than %i, was %i" % (ns, len(cdata))
+
+        # read in small steps
+        ss = len(cdata) // ns
+        for i in range(ns):
+            data = stream.read(ss)
+            chunk = cdata[i * ss:(i + 1) * ss]
+            assert data == chunk
+        # END for each step
+        rest = stream.read()
+        if rest:
+            assert rest == cdata[-len(rest):]
+        # END handle rest
+
+        if isinstance(stream, DecompressMemMapReader):
+            assert len(stream.data()) == stream.compressed_bytes_read()
+        # END handle special type
+
+        rewind_stream(stream)
+
+        # read everything
+        rdata = stream.read()
+        assert rdata == cdata
+
+        if isinstance(stream, DecompressMemMapReader):
+            assert len(stream.data()) == stream.compressed_bytes_read()
+        # END handle special type
+
+    def test_decompress_reader(self):
+        for close_on_deletion in range(2):
+            for with_size in range(2):
+                for ds in self.data_sizes:
+                    cdata = make_bytes(ds, randomize=False)
+
+                    # zdata = zipped actual data
+                    # cdata = original content data
+
+                    # create reader
+                    if with_size:
+                        # need object data
+                        zdata = zlib.compress(make_object(str_blob_type, cdata))
+                        typ, size, reader = DecompressMemMapReader.new(zdata, close_on_deletion)
+                        assert size == len(cdata)
+                        assert typ == str_blob_type
+
+                        # even if we don't set the size, it will be set automatically on first read
+                        test_reader = DecompressMemMapReader(zdata, close_on_deletion=False)
+                        assert test_reader._s == len(cdata)
+                    else:
+                        # here we need content data
+                        zdata = zlib.compress(cdata)
+                        reader = DecompressMemMapReader(zdata, close_on_deletion, len(cdata))
+                        assert reader._s == len(cdata)
+                    # END get reader
+
+                    self._assert_stream_reader(reader, cdata, lambda r: r.seek(0))
+
+                    # put in a dummy stream for closing
+                    dummy = DummyStream()
+                    reader._m = dummy
+
+                    assert not dummy.closed
+                    del(reader)
+                    assert dummy.closed == close_on_deletion
+                # END for each datasize
+            # END whether size should be used
+        # END whether stream should be closed when deleted
+
+    def test_sha_writer(self):
+        writer = Sha1Writer()
+        assert 2 == writer.write("hi".encode("ascii"))
+        assert len(writer.sha(as_hex=1)) == 40
+        assert len(writer.sha(as_hex=0)) == 20
+
+        # make sure it does something ;)
+        prev_sha = writer.sha()
+        writer.write("hi again".encode("ascii"))
+        assert writer.sha() != prev_sha
+
+    def test_compressed_writer(self):
+        for ds in self.data_sizes:
+            fd, path = tempfile.mkstemp()
+            ostream = FDCompressedSha1Writer(fd)
+            data = make_bytes(ds, randomize=False)
+
+            # for now, just a single write, code doesn't care about chunking
+            assert len(data) == ostream.write(data)
+            ostream.close()
+
+            # its closed already
+            self.assertRaises(OSError, os.close, fd)
+
+            # read everything back, compare to data we zip
+            fd = os.open(path, os.O_RDONLY | getattr(os, 'O_BINARY', 0))
+            written_data = os.read(fd, os.path.getsize(path))
+            assert len(written_data) == os.path.getsize(path)
+            os.close(fd)
+            assert written_data == zlib.compress(data, 1)   # best speed
+
+            os.remove(path)
+        # END for each os
+
+    def test_decompress_reader_special_case(self):
+        odb = LooseObjectDB(fixture_path('objects'))
+        mdb = MemoryDB()
+        for sha in (b'888401851f15db0eed60eb1bc29dec5ddcace911',
+                    b'7bb839852ed5e3a069966281bb08d50012fb309b',):
+            ostream = odb.stream(hex_to_bin(sha))
+
+            # if there is a bug, we will be missing one byte exactly !
+            data = ostream.read()
+            assert len(data) == ostream.size
+
+            # Putting it back in should yield nothing new - after all, we have
+            dump = mdb.store(IStream(ostream.type, ostream.size, BytesIO(data)))
+            assert dump.hexsha == sha
+        # end for each loose object sha to test

.venv/Lib/site-packages/gitdb/test/test_util.py

@@ -0,0 +1,100 @@
+# Copyright (C) 2010, 2011 Sebastian Thiel (byronimo@gmail.com) and contributors
+#
+# This module is part of GitDB and is released under
+# the New BSD License: http://www.opensource.org/licenses/bsd-license.php
+"""Test for object db"""
+import tempfile
+import os
+
+from gitdb.test.lib import TestBase
+from gitdb.util import (
+    to_hex_sha,
+    to_bin_sha,
+    NULL_HEX_SHA,
+    LockedFD
+)
+
+
+class TestUtils(TestBase):
+
+    def test_basics(self):
+        assert to_hex_sha(NULL_HEX_SHA) == NULL_HEX_SHA
+        assert len(to_bin_sha(NULL_HEX_SHA)) == 20
+        assert to_hex_sha(to_bin_sha(NULL_HEX_SHA)) == NULL_HEX_SHA.encode("ascii")
+
+    def _cmp_contents(self, file_path, data):
+        # raise if data from file at file_path
+        # does not match data string
+        with open(file_path, "rb") as fp:
+            assert fp.read() == data.encode("ascii")
+
+    def test_lockedfd(self):
+        my_file = tempfile.mktemp()
+        orig_data = "hello"
+        new_data = "world"
+        with open(my_file, "wb") as my_file_fp:
+            my_file_fp.write(orig_data.encode("ascii"))
+
+        try:
+            lfd = LockedFD(my_file)
+            lockfilepath = lfd._lockfilepath()
+
+            # cannot end before it was started
+            self.assertRaises(AssertionError, lfd.rollback)
+            self.assertRaises(AssertionError, lfd.commit)
+
+            # open for writing
+            assert not os.path.isfile(lockfilepath)
+            wfd = lfd.open(write=True)
+            assert lfd._fd is wfd
+            assert os.path.isfile(lockfilepath)
+
+            # write data and fail
+            os.write(wfd, new_data.encode("ascii"))
+            lfd.rollback()
+            assert lfd._fd is None
+            self._cmp_contents(my_file, orig_data)
+            assert not os.path.isfile(lockfilepath)
+
+            # additional call doesn't fail
+            lfd.commit()
+            lfd.rollback()
+
+            # test reading
+            lfd = LockedFD(my_file)
+            rfd = lfd.open(write=False)
+            assert os.read(rfd, len(orig_data)) == orig_data.encode("ascii")
+
+            assert os.path.isfile(lockfilepath)
+            # deletion rolls back
+            del(lfd)
+            assert not os.path.isfile(lockfilepath)
+
+            # write data - concurrently
+            lfd = LockedFD(my_file)
+            olfd = LockedFD(my_file)
+            assert not os.path.isfile(lockfilepath)
+            wfdstream = lfd.open(write=True, stream=True)       # this time as stream
+            assert os.path.isfile(lockfilepath)
+            # another one fails
+            self.assertRaises(IOError, olfd.open)
+
+            wfdstream.write(new_data.encode("ascii"))
+            lfd.commit()
+            assert not os.path.isfile(lockfilepath)
+            self._cmp_contents(my_file, new_data)
+
+            # could test automatic _end_writing on destruction
+        finally:
+            os.remove(my_file)
+        # END final cleanup
+
+        # try non-existing file for reading
+        lfd = LockedFD(tempfile.mktemp())
+        try:
+            lfd.open(write=False)
+        except OSError:
+            assert not os.path.exists(lfd._lockfilepath())
+        else:
+            self.fail("expected OSError")
+        # END handle exceptions