Merge lp:~jelmer/brz-git/check into lp:brz-git

[Jelmer Vernooij]

Add really basic check implementation.

Add really basic check implementation.

[Jelmer Vernooij (jelmer)]

Rubberstamp! Proposer approves of own proposal.

review: Approve

=== modified file 'cache.py'
--- cache.py	2018-04-08 01:37:56 +0000
+++ cache.py	2018-04-08 17:07:41 +0000
@@ -862,12 +862,12 @@
 
     def _add_node(self, key, value):
         try:
+            self._get_entry(key)
+        except KeyError:
             self._builder.add_node(key, value)
-        except _mod_index.BadIndexDuplicateKey:
-            # Multiple bzr objects can have the same contents
+            return False
+        else:
             return True
-        else:
-            return False
 
     def _get_entry(self, key):
         entries = self._index.iter_entries([key])
@@ -918,9 +918,12 @@
         value = self._get_entry(("git", sha, "X"))
         data = value.split(" ", 3)
         if data[0] == "commit":
-            if data[3]:
-                verifiers = {"testament3-sha1": data[3]}
-            else:
+            try:
+                if data[3]:
+                    verifiers = {"testament3-sha1": data[3]}
+                else:
+                    verifiers = {}
+            except IndexError:
                 verifiers = {}
             yield ("commit", (data[1], data[2], verifiers))
         else:
@@ -953,13 +956,7 @@
 formats.register(IndexGitCacheFormat().get_format_string(),
     IndexGitCacheFormat())
 # In the future, this will become the default:
-# formats.register('default', IndexGitCacheFormat())
-try:
-    import tdb
-except ImportError:
-    formats.register('default', SqliteGitCacheFormat())
-else:
-    formats.register('default', TdbGitCacheFormat())
+formats.register('default', IndexGitCacheFormat())
 
 
 

=== modified file 'repository.py'
--- repository.py	2018-03-30 21:45:13 +0000
+++ repository.py	2018-04-08 17:07:41 +0000
@@ -26,6 +26,7 @@
     lock,
     repository,
     revision as _mod_revision,
+    trace,
     transactions,
     ui,
     version_info as breezy_version,
@@ -93,17 +94,37 @@
 
     def __init__(self, repository, check_repo=True):
         self.repository = repository
+        self.check_repo = check_repo
         self.checked_rev_cnt = 0
+        self.object_count = None
+        self.problems = []
 
     def check(self, callback_refs=None, check_repo=True):
         if callback_refs is None:
             callback_refs = {}
-        with self.repository.lock_read():
-            # TODO(jelmer): Check some things
-            pass
+        with self.repository.lock_read(), ui.ui_factory.nested_progress_bar() as self.progress:
+            shas = set(self.repository._git.object_store)
+            self.object_count = len(shas)
+            # TODO(jelmer): Check more things
+            for i, sha in enumerate(shas):
+                self.progress.update('checking objects', i, self.object_count)
+                o = self.repository._git.object_store[sha]
+                try:
+                    o.check()
+                except Exception as e:
+                    self.problems.append((sha, e))
+
+    def _report_repo_results(self, verbose):
+        trace.note('checked repository {0} format {1}'.format(
+            self.repository.user_url,
+            self.repository._format))
+        trace.note('%6d objects', self.object_count)
+        for sha, problem in self.problems:
+            trace.note('%s: %s', sha, problem)
 
     def report_results(self, verbose):
-        pass
+        if self.check_repo:
+            self._report_repo_results(verbose)
 
 
 _optimisers_loaded = False

=== modified file 'tests/test_blackbox.py'
--- tests/test_blackbox.py	2018-03-21 03:06:49 +0000
+++ tests/test_blackbox.py	2018-04-08 17:07:41 +0000
@@ -287,3 +287,13 @@
         self.build_tree_contents([("bzrb/foo", "hello from bzr")])
         self.run_bzr(["commit", "-m", "msg", "bzrb"])
         self.run_bzr(["dpush", "-d", "bzrb", "gitr"])
+
+    def test_check(self):
+        r = GitRepo.init("gitr", mkdir=True)
+        self.build_tree_contents([("gitr/foo", "hello from git")])
+        r.stage("foo")
+        r.do_commit("message", committer="Somebody <[e-mail]>")
+        out, err = self.run_bzr(["check", "gitr"])
+        self.maxDiff = None
+        self.assertMultiLineEqual(out, '')
+        self.assertTrue(err.endswith, '3 objects\n')

=== modified file 'workingtree.py'
--- workingtree.py	2018-04-02 21:58:43 +0000
+++ workingtree.py	2018-04-08 17:07:41 +0000
@@ -1166,6 +1166,16 @@
                                 show_base=show_base)
             return count
 
+    def _get_check_refs(self):
+        """Return the references needed to perform a check of this tree.
+
+        The default implementation returns no refs, and is only suitable for
+        trees that have no local caching and can commit on ghosts at any time.
+
+        :seealso: breezy.check for details about check_refs.
+        """
+        return []
+
 
 class GitWorkingTreeFormat(workingtree.WorkingTreeFormat):