Add support to repositories for storing run metadata

This commit adds repository APIs around storing and accessing run
metadata strings. The concept being that you can store a metadata string
to identify a run by some other identifier then the run id. The example
use case in mind for this feature is using it to store a VCS commit hash
(like a git sha1) or identifier. The APIs in this commit allow storing
metadata at run creation time, accessing the metadata for a run, and
finding run_ids by a metadata value. This is implemented for both the
file and sql repository types. The memory repository type does not have
this implemented, as it is more limited.

Related to: #224

stestr/repository/abstract.py

@@ -67,7 +67,7 @@ def get_failing(self):
         """
         raise NotImplementedError(self.get_failing)
 
-    def get_inserter(self, partial=False, run_id=None):
+    def get_inserter(self, partial=False, run_id=None, metadata=None):
         """Get an inserter that will insert a test run into the repository.
 
         Repository implementations should implement _get_inserter.
@@ -83,9 +83,9 @@ def get_inserter(self, partial=False, run_id=None):
             that testtools 0.9.2 and above offer. The startTestRun and
             stopTestRun methods in particular must be called.
         """
-        return self._get_inserter(partial, run_id)
+        return self._get_inserter(partial, run_id, metadata)
 
-    def _get_inserter(self, partial=False, run_id=None):
+    def _get_inserter(self, partial=False, run_id=None, metadata=None):
         """Get an inserter for get_inserter.
 
         The result is decorated with an AutoTimingTestResultDecorator.
@@ -156,6 +156,14 @@ def gather(test_dict):
             result.stopTestRun()
         return ids
 
+    def find_metadata(self, metadata):
+        """Return the list of run_ids for a given metadata string.
+
+        :param: metadata: the metadata string to search for.
+        :return: a list of any test_ids that have that metadata value.
+        """
+        raise NotImplementedError(self.find_metadata)
+
 
 class AbstractTestRun(object):
     """A test run that has been stored in a repository.
@@ -186,6 +194,13 @@ def get_test(self):
         """
         raise NotImplementedError(self.get_test)
 
+    def get_metadata(self):
+        """Get the metadata value for the test run.
+
+        :return: A string of the metadata or None if it doesn't exist.
+        """
+        raise NotImplementedError(self.get_metadata)
+
 
 class RepositoryNotFound(Exception):
     """Raised when we try to open a repository that isn't there."""

stestr/repository/file.py

@@ -115,6 +115,15 @@ def get_failing(self):
                 raise
         return _DiskRun(None, run_subunit_content)
 
+    def _get_metadata(self, run_id):
+        db = my_dbm.open(self._path('meta.dbm'), 'c')
+        try:
+            print([x for x in db])
+            metadata = db.get(str(run_id))
+        finally:
+            db.close()
+        return metadata
+
     def get_test_run(self, run_id):
         try:
             with open(os.path.join(self.base, str(run_id)), 'rb') as fp:
@@ -124,10 +133,11 @@ def get_test_run(self, run_id):
                 raise KeyError("No such run.")
             else:
                 raise
-        return _DiskRun(run_id, run_subunit_content)
+        metadata = self._get_metadata(run_id)
+        return _DiskRun(run_id, run_subunit_content, metadata=metadata)
 
-    def _get_inserter(self, partial, run_id=None):
-        return _Inserter(self, partial, run_id)
+    def _get_inserter(self, partial, run_id=None, metadata=None):
+        return _Inserter(self, partial, run_id, metadata=metadata)
 
     def _get_test_times(self, test_ids):
         # May be too slow, but build and iterate.
@@ -167,15 +177,27 @@ def _write_next_stream(self, value):
             stream.write('%d\n' % value)
         atomicish_rename(prefix + '.new', prefix)
 
+    def find_metadata(self, metadata):
+        run_ids = []
+        db = my_dbm.open(self._path('meta.dbm'), 'c')
+        try:
+            for run_id in db:
+                if db.get(run_id) == metadata:
+                    run_ids.append(run_id)
+        finally:
+            db.close()
+        return run_ids
+
 
 class _DiskRun(repository.AbstractTestRun):
     """A test run that was inserted into the repository."""
 
-    def __init__(self, run_id, subunit_content):
+    def __init__(self, run_id, subunit_content, metadata=None):
         """Create a _DiskRun with the content subunit_content."""
         self._run_id = run_id
         self._content = subunit_content
         assert type(subunit_content) is bytes
+        self._metadata = metadata
 
     def get_id(self):
         return self._run_id
@@ -211,21 +233,31 @@ def wrap_result(result):
             case, wrap_result, methodcaller('startTestRun'),
             methodcaller('stopTestRun'))
 
+    def get_metadata(self):
+        return self._metadata
+
 
 class _SafeInserter(object):
 
-    def __init__(self, repository, partial=False, run_id=None):
+    def __init__(self, repository, partial=False, run_id=None, metadata=None):
         # XXX: Perhaps should factor into a decorator and use an unaltered
         # TestProtocolClient.
         self._repository = repository
         self._run_id = run_id
+        self._metadata = metadata
         if not self._run_id:
             fd, name = tempfile.mkstemp(dir=self._repository.base)
             self.fname = name
             stream = os.fdopen(fd, 'wb')
         else:
             self.fname = os.path.join(self._repository.base, self._run_id)
             stream = open(self.fname, 'ab')
+        if self._metadata:
+            db = my_dbm.open(self._repository._path('meta.dbm'), 'c')
+            try:
+                db[str(run_id)] = self._metadata
+            finally:
+                db.close()
         self.partial = partial
         # The time take by each test, flushed at the end.
         self._times = {}

stestr/repository/memory.py

@@ -71,7 +71,7 @@ def latest_id(self):
             raise KeyError("No tests in repository")
         return result
 
-    def _get_inserter(self, partial, run_id=None):
+    def _get_inserter(self, partial, run_id=None, metadata=None):
         return _Inserter(self, partial, run_id)
 
     def _get_test_times(self, test_ids):
@@ -127,13 +127,14 @@ def run(self, result):
 class _Inserter(repository.AbstractTestRun):
     """Insert test results into a memory repository."""
 
-    def __init__(self, repository, partial, run_id=None):
+    def __init__(self, repository, partial, run_id=None, metadata=None):
         self._repository = repository
         self._partial = partial
         self._tests = []
         # Subunit V2 stream for get_subunit_stream
         self._subunit = None
         self._run_id = run_id
+        self._metadata = metadata
 
     def startTestRun(self):
         self._subunit = BytesIO()

stestr/repository/sql.py

@@ -116,8 +116,8 @@ def get_failing(self):
     def get_test_run(self, run_id):
         return _Subunit2SqlRun(self.base, run_id)
 
-    def _get_inserter(self, partial, run_id=None):
-        return _SqlInserter(self, partial, run_id)
+    def _get_inserter(self, partial, run_id=None, metadata=None):
+        return _SqlInserter(self, partial, run_id, metadata)
 
     def _get_test_times(self, test_ids):
         result = {}
@@ -136,6 +136,12 @@ def _get_test_times(self, test_ids):
         session.close()
         return result
 
+    def find_metadata(self, metadata):
+        session = self.session_factory()
+        runs = db_api.get_runs_by_key_value('stestr_run_meta', metadata,
+                                            session=session)
+        return [x.uuid for x in runs]
+
 
 class _Subunit2SqlRun(repository.AbstractTestRun):
     """A test run that was inserted into the repository."""
@@ -177,15 +183,25 @@ def get_test(self):
         case = subunit.ByteStreamToStreamResult(stream)
         return case
 
+    def get_metadata(self):
+        if self._run_id:
+            session = self.session_factory()
+            metadata = db_api.get_run_metadata(self._run_id, session=session)
+            for meta in metadata:
+                if meta.key == 'stestr_run_meta':
+                    return meta.value
+        return None
+
 
 class _SqlInserter(repository.AbstractTestRun):
     """Insert test results into a sql repository."""
 
-    def __init__(self, repository, partial=False, run_id=None):
+    def __init__(self, repository, partial=False, run_id=None, metadata=None):
         self._repository = repository
         self.partial = partial
         self._subunit = None
         self._run_id = run_id
+        self._metadata = metadata
         # Create a new session factory
         self.engine = sqlalchemy.create_engine(self._repository.base)
         self.session_factory = orm.sessionmaker(bind=self.engine,
@@ -202,6 +218,9 @@ def startTestRun(self):
         session = self.session_factory()
         if not self._run_id:
             self.run = db_api.create_run(session=session)
+            if self._metadata:
+                db_api.add_run_metadata({'stestr_run_meta': self._metadata},
+                                        self.run.id, session=session)
             self._run_id = self.run.uuid
         else:
             int_id = db_api.get_run_id_from_uuid(self._run_id, session=session)

stestr/tests/repository/test_file.py

@@ -117,3 +117,23 @@ def test_get_test_run_unexpected_ioerror_errno(self):
         self.assertTrue(os.path.isfile(os.path.join(repo.base, '0')))
         os.chmod(os.path.join(repo.base, '0'), 0000)
         self.assertRaises(IOError, repo.get_test_run, '0')
+
+    def test_get_metadata(self):
+        repo = self.useFixture(FileRepositoryFixture()).repo
+        result = repo.get_inserter(metadata='fun')
+        result.startTestRun()
+        result.stopTestRun()
+        run = repo.get_test_run(result.get_id())
+        self.assertEqual('fun', run.get_metadata())
+
+    def test_find_metadata(self):
+        repo = self.useFixture(FileRepositoryFixture()).repo
+        result = repo.get_inserter(metadata='fun')
+        result.startTestRun()
+        result.stopTestRun()
+        result_bad = repo.get_inserter(metadata='not_fun')
+        result_bad.startTestRun()
+        result_bad.stopTestRun()
+        run_ids = repo.find_metadata('fun')
+        self.assertIn(result.get_id(), run_ids)
+        self.assertNotIn(bad_result.get_id(), run_ids)

stestr/tests/repository/test_sql.py

@@ -80,3 +80,23 @@ def test_run_get_subunit_stream(self):
         self.assertIsNotNone(stream)
         self.assertTrue(stream.readable())
         self.assertEqual([], stream.readlines())
+
+    def test_get_metadata(self):
+        repo = self.useFixture(SqlRepositoryFixture(url=self.url)).repo    
+        result = repo.get_inserter(metadata='fun')
+        result.startTestRun()
+        result.stopTestRun()
+        run = repo.get_test_run(result.get_id())
+        self.assertEqual('fun', run.get_metadata())
+
+    def test_find_metadata(self):
+        repo = self.useFixture(SqlRepositoryFixture(url=self.url)).repo
+        result = repo.get_inserter(metadata='fun')
+        result.startTestRun()
+        result.stopTestRun()
+        result_bad = repo.get_inserter(metadata='not_fun')
+        result_bad.startTestRun()
+        result_bad.stopTestRun()
+        run_ids = repo.find_metadata('fun')
+        self.assertIn(result.get_id(), run_ids)
+        self.assertNotIn(result_bad.get_id(), run_ids)