From 27863bbea230ef473832656ad4c26fcf878995d1 Mon Sep 17 00:00:00 2001 From: Sam Thursfield Date: Tue, 6 Jan 2015 17:56:21 +0000 Subject: Add a mechanism for extracting all files from a given commit to a dir This is nice because it's fast. We don't have to copy all the Git history along with it like we do with a clone. And it doesn't touch any files in the cached repo. --- morphlib/cachedrepo.py | 27 ++++++++++++++++++++++++++- morphlib/cachedrepo_tests.py | 30 +++++++++++++++++++++++++++++- morphlib/gitindex.py | 22 +++++++++++++++++++--- morphlib/gitindex_tests.py | 16 +++++++++++++++- 4 files changed, 89 insertions(+), 6 deletions(-) diff --git a/morphlib/cachedrepo.py b/morphlib/cachedrepo.py index aa2b5af1..c9a6bb6b 100644 --- a/morphlib/cachedrepo.py +++ b/morphlib/cachedrepo.py @@ -1,4 +1,4 @@ -# Copyright (C) 2012-2014 Codethink Limited +# Copyright (C) 2012-2015 Codethink Limited # # This program is free software; you can redistribute it and/or modify # it under the terms of the GNU General Public License as published by @@ -15,7 +15,9 @@ import cliapp + import os +import tempfile import morphlib @@ -169,6 +171,29 @@ class CachedRepo(object): self._checkout_ref_in_clone(ref, target_dir) + def extract_commit(self, ref, target_dir): + # FIXME: surely this should do what 'checkout' does, and 'checkout' + # should be called something else? Or maybe there's no speed benefit + # to this function and it shouldn't exist in the first place? Or the + # other 'checkout' shouldn't exist? + '''Extract files from a given commit into target_dir. + + This is different to a 'checkout': a checkout assumes a working tree + associated with a repository. Here, the repository is immutable (it's + in the cache) and we just want to look at the files in a quick way + (quicker than going 'git cat-file everything'). + + This is really fast so it's cool really. + + ''' + if not os.path.exists(target_dir): + os.makedirs(target_dir) + + with tempfile.NamedTemporaryFile() as index_file: + index = self._gitdir.get_index(index_file=index_file.name) + index.set_to_tree(ref) + index.checkout(working_tree=target_dir) + def requires_update_for_ref(self, ref): '''Returns False if there's no need to update this cached repo. diff --git a/morphlib/cachedrepo_tests.py b/morphlib/cachedrepo_tests.py index 6f87bfdd..6fe69ef5 100644 --- a/morphlib/cachedrepo_tests.py +++ b/morphlib/cachedrepo_tests.py @@ -1,4 +1,4 @@ -# Copyright (C) 2012-2014 Codethink Limited +# Copyright (C) 2012-2015 Codethink Limited # # This program is free software; you can redistribute it and/or modify # it under the terms of the GNU General Public License as published by @@ -33,6 +33,21 @@ class FakeApplication(object): } +class FakeIndex(object): + + def __init__(self, index_file): + self.index_file = index_file + self.ref = None + + def set_to_tree(self, ref): + self.ref = ref + + def checkout(self, working_tree=None): + if working_tree: + with open(os.path.join(working_tree, 'foo.morph'), 'w') as f: + f.write('contents of foo.morph') + + class CachedRepoTests(unittest.TestCase): known_commit = 'a4da32f5a81c8bc6d660404724cedc3bc0914a75' @@ -77,6 +92,9 @@ class CachedRepoTests(unittest.TestCase): def update_with_failure(self, **kwargs): raise cliapp.AppException('git remote update origin') + def get_index(self, index_file=None): + return FakeIndex(index_file) + def setUp(self): self.repo_name = 'foo' self.repo_url = 'git://foo.bar/foo.git' @@ -141,6 +159,16 @@ class CachedRepoTests(unittest.TestCase): morph_filename = os.path.join(unpack_dir, 'foo.morph') self.assertTrue(os.path.exists(morph_filename)) + def test_extract_commit_into_new_directory(self): + self.repo._gitdir.get_index = self.get_index + unpack_dir = self.tempfs.getsyspath('unpack-dir') + self.repo.extract_commit('e28a23812eadf2fce6583b8819b9c5dbd36b9fb9', + unpack_dir) + self.assertTrue(os.path.exists(unpack_dir)) + + morph_filename = os.path.join(unpack_dir, 'foo.morph') + self.assertTrue(os.path.exists(morph_filename)) + def test_successful_update(self): self.repo._gitdir.update_remotes = self.update_successfully self.repo.update() diff --git a/morphlib/gitindex.py b/morphlib/gitindex.py index e22f6225..c5c07bd6 100644 --- a/morphlib/gitindex.py +++ b/morphlib/gitindex.py @@ -1,4 +1,4 @@ -# Copyright (C) 2013-2014 Codethink Limited +# Copyright (C) 2013-2015 Codethink Limited # # This program is free software; you can redistribute it and/or modify # it under the terms of the GNU General Public License as published by @@ -48,8 +48,16 @@ class GitIndex(object): def _run_git(self, *args, **kwargs): if self._index_file is not None: - kwargs['env'] = kwargs.get('env', dict(os.environ)) - kwargs['env']['GIT_INDEX_FILE'] = self._index_file + extra_env = kwargs.get('extra_env', {}) + extra_env['GIT_INDEX_FILE'] = self._index_file + kwargs['extra_env'] = extra_env + + if 'extra_env' in kwargs: + env = kwargs.get('env', dict(os.environ)) + env.update(kwargs['extra_env']) + kwargs['env'] = env + del kwargs['extra_env'] + return morphlib.git.gitcmd(self._gd._runcmd, *args, **kwargs) def _get_status(self): @@ -159,3 +167,11 @@ class GitIndex(object): def write_tree(self): '''Transform the index into a tree in the object store.''' return self._run_git('write-tree').strip() + + def checkout(self, working_tree=None): + '''Copy files from the index to the working tree.''' + if working_tree: + extra_env = {'GIT_WORK_TREE': working_tree} + else: + extra_env = {} + self._run_git('checkout-index', '--all', extra_env=extra_env) diff --git a/morphlib/gitindex_tests.py b/morphlib/gitindex_tests.py index 32d40a8c..3f9ff303 100644 --- a/morphlib/gitindex_tests.py +++ b/morphlib/gitindex_tests.py @@ -1,4 +1,4 @@ -# Copyright (C) 2013-2014 Codethink Limited +# Copyright (C) 2013-2015 Codethink Limited # # This program is free software; you can redistribute it and/or modify # it under the terms of the GNU General Public License as published by @@ -38,6 +38,8 @@ class GitIndexTests(unittest.TestCase): self.mirror = os.path.join(self.tempdir, 'mirror') morphlib.git.gitcmd(gd._runcmd, 'clone', '--mirror', self.dirname, self.mirror) + self.working_dir = os.path.join(self.tempdir, 'bar') + os.makedirs(self.working_dir) def tearDown(self): shutil.rmtree(self.tempdir) @@ -91,3 +93,15 @@ class GitIndexTests(unittest.TestCase): gd = morphlib.gitdir.GitDirectory(self.dirname) idx = gd.get_index() self.assertEqual(idx.write_tree(), gd.resolve_ref_to_tree(gd.HEAD)) + + def test_checkout(self): + gd = morphlib.gitdir.GitDirectory(self.dirname) + idx = gd.get_index() + idx.checkout(working_tree=self.working_dir) + self.assertTrue(os.path.exists(os.path.join(self.working_dir, 'foo'))) + + def test_checkout_without_working_dir(self): + gd = morphlib.gitdir.GitDirectory(self.dirname) + idx = gd.get_index() + idx.checkout() + self.assertTrue(os.path.exists(os.path.join(self.dirname, 'foo'))) -- cgit v1.2.1