multiple issues with git backend

pyvcs/backends/bzr.py

@@ -26,12 +26,16 @@ def __init__(self, *args, **kwargs):
     def _rev_to_commit(self, rev):
         # TODO: this doesn't yet handle the case of multiple parent revisions
         current = self._branch.repository.revision_tree(rev.revision_id)
-        prev = self._branch.repository.revision_tree(rev.parent_ids[0])
+        if len(rev.parent_ids):
+            prev = self._branch.repository.revision_tree(rev.parent_ids[0])
+        else:
+            prev = self._branch.repository.revision_tree('null:')
 
         delta = current.changes_from(prev)
         files = [f[0] for f in delta.added + delta.removed + delta.renamed + delta.kind_changed + delta.modified]
 
         diff_file = StringIO.StringIO()
+
         diff_tree = diff.DiffTree(prev, current, diff_file)
 
         self._branch.lock_read()
@@ -85,9 +89,7 @@ def get_recent_commits(self, since=None):
 
     def list_directory(self, path, revision=None):
         path = path.rstrip(os.path.sep)
-
         tree = self._get_tree(revision)
-
         dir_iter = tree.walkdirs(path)
         try:
             entries = dir_iter.next()

pyvcs/backends/git.py

@@ -4,6 +4,7 @@
 
 from dulwich.repo import Repo
 from dulwich import objects
+from dulwich.errors import NotCommitError
 
 from pyvcs.commit import Commit
 from pyvcs.exceptions import CommitDoesNotExist, FileDoesNotExist, FolderDoesNotExist
@@ -63,41 +64,48 @@ def __init__(self, *args, **kwargs):
 
     def _get_commit(self, commit_id):
         try:
-            return self._repo.commit(commit_id)
+            return self._repo[commit_id]
         except Exception, e:
             raise CommitDoesNotExist("%s is not a commit" % commit_id)
 
     def _get_obj(self, sha):
         return self._repo.get_object(sha)
 
     def _diff_files(self, commit_id1, commit_id2):
+        if commit_id1 == 'NULL':
+            commit_id1 = None
+        if commit_id2 == 'NULL':
+            commit_id2 = None
+        tree1 = self._get_obj(self._get_obj(commit_id1).tree) if commit_id1 else None
+        tree2 = self._get_obj(self._get_obj(commit_id2).tree) if commit_id2 else None
         return sorted(get_differing_files(
             self._repo,
-            self._get_obj(self._get_obj(commit_id1).tree),
-            self._get_obj(self._get_obj(commit_id2).tree),
+            tree1,
+            tree2,
         ))
 
     def get_commit_by_id(self, commit_id):
         commit = self._get_commit(commit_id)
-        files = self._diff_files(commit.id, commit.parents[0])
+        parent = commit.parents[0] if len(commit.parents) else 'NULL'
+        files = self._diff_files(commit.id, parent)
         return Commit(commit.id, commit.committer,
             datetime.fromtimestamp(commit.commit_time), commit.message, files,
-            lambda: generate_unified_diff(self, files, commit.parents[0], commit.id))
+            lambda: generate_unified_diff(self, files, parent, commit.id))
 
     def get_recent_commits(self, since=None):
         if since is None:
-            since = datetime.fromtimestamp(self._repo.commit(self._repo.head()).commit_time) - timedelta(days=5)
+            #since = datetime.fromtimestamp(self._repo.commit(self._repo.head()).commit_time) - timedelta(days=5)
+            since = datetime.fromtimestamp(self._repo[self._repo.head()].commit_time) - timedelta(days=5)
         pending_commits = self._repo.get_refs().values()#[self._repo.head()]
         history = {}
         while pending_commits:
             head = pending_commits.pop(0)
             try:
-                commit = self._get_obj(head)
+                commit = self._repo[head]
             except KeyError:
                 raise CommitDoesNotExist
-            if not hasattr(commit, 'commit_time'):
-                continue
-            if commit.id in history or datetime.fromtimestamp(commit.commit_time) <= since:
+            if not isinstance(commit, objects.Commit) or commit.id in history or\
+               datetime.fromtimestamp(commit.commit_time) <= since:
                 continue
             history[commit.id] = commit
             pending_commits.extend(commit.parents)
@@ -106,16 +114,21 @@ def get_recent_commits(self, since=None):
         return sorted(commits, key=attrgetter('time'), reverse=True)
 
     def list_directory(self, path, revision=None):
-        commit = self._get_commit(revision or self._repo.head())
-        tree = self._repo.tree(commit.tree)
+        if revision is None:
+            commit = self._get_commit(self._repo.head())
+        elif revision is 'NULL':
+            return ([],[])
+        else:
+            commit = self._get_commit(revision)
+        tree = self._repo[commit.tree]
         path = filter(bool, path.split(os.path.sep))
         while path:
             part = path.pop(0)
             found = False
-            for mode, name, hexsha in self._repo.tree(tree.id).entries():
+            for mode, name, hexsha in self._repo[tree.id].entries():
                 if part == name:
                     found = True
-                    tree = self._repo.tree(hexsha)
+                    tree = self._repo[hexsha]
                     break
             if not found:
                 raise FolderDoesNotExist
@@ -128,17 +141,22 @@ def list_directory(self, path, revision=None):
         return files, folders
 
     def file_contents(self, path, revision=None):
-        commit = self._get_commit(revision or self._repo.head())
-        tree = self._repo.tree(commit.tree)
+        if revision is None:
+            commit = self._get_commit(self._repo.head())
+        elif revision is 'NULL':
+            return ''
+        else:
+            commit = self._get_commit(revision)
+        tree = self._repo[commit.tree]
         path = path.split(os.path.sep)
         path, filename = path[:-1], path[-1]
         while path:
             part = path.pop(0)
-            for mode, name, hexsha in self._repo.tree(tree.id).entries():
+            for mode, name, hexsha in self._repo[tree.id].entries():
                 if part == name:
-                    tree = self._repo.tree(hexsha)
+                    tree = self._repo[hexsha]
                     break
         for mode, name, hexsha in tree.entries():
             if name == filename:
-                return self._repo.get_object(hexsha).as_pretty_string()
+                return self._repo[hexsha].as_pretty_string()
         raise FileDoesNotExist

tests/setup_git_test.sh

@@ -0,0 +1,18 @@
+#!/bin/sh
+
+#NOTE: this script depends on git being installed and having access to the /tmp directory...
+
+echo Setting up the directory for the git repository
+mkdir /tmp/pyvcs-test
+mkdir /tmp/pyvcs-test/git-test
+
+cd /tmp/pyvcs-test/git-test
+git init
+echo this is a test README file for a mock project > README
+echo print 'hello, world!' > hello_world.py
+git add README
+git add hello_world.py
+git commit -m "initial add of files to repo"
+echo this is a new line added to the README >> README
+git commit -a -m "slight change to the README"
+

tests/simple_git_tests.py

@@ -0,0 +1,50 @@
+#!/usr/bin/env python
+from datetime import datetime
+import unittest
+import os
+import subprocess
+
+from pyvcs.backends import get_backend
+from pyvcs.exceptions import FileDoesNotExist, FolderDoesNotExist, CommitDoesNotExist
+
+class GitSimpleTest(unittest.TestCase):
+
+    def setUp(self):
+        git = get_backend('git')
+        ret = subprocess.call('./setup_git_test.sh')
+        self.repo = git.Repository('/tmp/pyvcs-test/git-test/')
+
+    def tearDown(self):
+        ret = subprocess.call('./teardown_git_test.sh')
+
+    def test_recent_commits(self):
+        recent_commits = self.repo.get_recent_commits()
+        self.assertEqual(len(recent_commits),2)
+
+    def test_commits(self):
+        recent_commits = self.repo.get_recent_commits()
+        commit = self.repo.get_commit_by_id(recent_commits[1].commit_id)
+        self.assert_(commit.message.startswith('initial add of files'))
+        self.assertEqual(commit.time.date(), datetime.today().date())
+        self.assertEqual(commit.files, ['README', 'hello_world.py'])
+        self.assert_('this is a test README file for a mock project' in commit.diff)
+        self.assertRaises(CommitDoesNotExist,self.repo.get_commit_by_id,'crap')
+
+    def test_list_directory(self):
+        files, folders = self.repo.list_directory('')
+        self.assertEqual(files, ['README', 'hello_world.py'])
+        self.assertEqual(folders, [])
+        self.assertRaises(FolderDoesNotExist, self.repo.list_directory, 'tests/awesometests/')
+
+    def test_file_contents(self):
+        contents = self.repo.file_contents('hello_world.py')
+        self.assertEqual(contents,'print hello, world!\n')
+        self.assertRaises(FileDoesNotExist, self.repo.file_contents, 'waybettertest.py')
+
+    def test_diffs(self):
+        recent_commits = self.repo.get_recent_commits()
+        self.assertEqual(self.repo._diff_files(recent_commits[0].commit_id,recent_commits[1].commit_id),['README'])
+
+
+if __name__ == '__main__':
+    unittest.main()

tests/teardown_git_test.sh

@@ -0,0 +1,2 @@
+#!/bin/sh
+rm -r -f /tmp/pyvcs-test