def test_prechunk(source, target, old_docs, new_docs):
builder = CopyBuilder(source, target, delete_orphans=True)
source.update(old_docs)
source.update(new_docs)
chunk_queries = list(builder.prechunk(2))
assert len(chunk_queries) == 2
assert chunk_queries[0] == {"query": {"k": {"$in": [0, 1, 2, 3, 4, 5, 6, 7, 8, 9]}}}
def test_query(source, target, old_docs, new_docs):
builder = CopyBuilder(source, target)
builder.query = {"k": {"$gt": 5}}
source.update(old_docs)
source.update(new_docs)
builder.run()
all_docs = list(target.query(criteria={}))
assert len(all_docs) == 14
assert min([d["k"] for d in all_docs]) == 6
def test_run(source, target, old_docs, new_docs):
source.update(old_docs)
source.update(new_docs)
target.update(old_docs)
builder = CopyBuilder(source, target)
builder.run()
assert target.query_one(criteria={"k": 0})["v"] == "new"
assert target.query_one(criteria={"k": 10})["v"] == "old"
def test_delete_orphans(source, target, old_docs, new_docs):
builder = CopyBuilder(source, target, delete_orphans=True)
source.update(old_docs)
source.update(new_docs)
target.update(old_docs)
deletion_criteria = {"k": {"$in": list(range(5))}}
source.collection.delete_many(deletion_criteria)
builder.run()
assert target.collection.count_documents(deletion_criteria) == 0
assert target.query_one(criteria={"k": 5})["v"] == "new"
assert target.query_one(criteria={"k": 10})["v"] == "old"
def test_delete_orphans(self):
self.builder = CopyBuilder(self.source, self.target, delete_orphans=True)
self.source.collection.insert_many(self.old_docs)
self.source.update(self.new_docs, update_lu=False)
self.target.collection.insert_many(self.old_docs)
deletion_criteria = {"k": {"$in": list(range(5))}}
self.source.collection.delete_many(deletion_criteria)
runner = Runner([self.builder])
runner.run()
self.assertEqual(self.target.collection.count_documents(deletion_criteria), 0)
self.assertEqual(self.target.query_one(criteria={"k": 5})["v"], "new")
self.assertEqual(self.target.query_one(criteria={"k": 10})["v"], "old")
def test_incremental_false(self):
tic = datetime.now()
toc = tic + timedelta(seconds=1)
keys = list(range(20))
earlier = [{"lu": tic, "k": k, "v": "val"} for k in keys]
later = [{"lu": toc, "k": k, "v": "val"} for k in keys]
self.source.collection.insert_many(earlier)
self.target.collection.insert_many(later)
query = {"k": {"$gt": 5}}
self.builder = CopyBuilder(self.source, self.target, incremental=False, query=query)
Runner([self.builder]).run()
docs = sorted(self.target.query(), key=lambda d: d["k"])
self.assertTrue(all(d["lu"] == tic) for d in docs[5:])
self.assertTrue(all(d["lu"] == toc) for d in docs[:5])
def test_reporting(mongostore, reporting_store):
memorystore = MemoryStore("temp")
builder = CopyBuilder(mongostore, memorystore)
mongostore.update([{
mongostore.key: i,
mongostore.last_updated_field: datetime.utcnow()
} for i in range(10)])
runner = CliRunner()
with runner.isolated_filesystem():
dumpfn(builder, "test_builder.json")
dumpfn(reporting_store, "test_reporting_store.json")
result = runner.invoke(
run,
["-v", "test_builder.json", "-r", "test_reporting_store.json"])
assert result.exit_code == 0
report_docs = list(reporting_store.query())
assert len(report_docs) == 3
start_doc = next(d for d in report_docs
if d["event"] == "BUILD_STARTED")
assert "sources" in start_doc
assert "targets" in start_doc
end_doc = next(d for d in report_docs if d["event"] == "BUILD_ENDED")
assert "errors" in end_doc
assert "warnings" in end_doc
update_doc = next(d for d in report_docs if d["event"] == "UPDATE")
assert "items" in update_doc
def test_run_builder(mongostore):
memorystore = MemoryStore("temp")
builder = CopyBuilder(mongostore, memorystore)
mongostore.update([{
mongostore.key: i,
mongostore.last_updated_field: datetime.utcnow()
} for i in range(10)])
runner = CliRunner()
with runner.isolated_filesystem():
dumpfn(builder, "test_builder.json")
result = runner.invoke(run, ["-v", "test_builder.json"])
assert result.exit_code == 0
assert "CopyBuilder" in result.output
assert "SerialProcessor" in result.output
result = runner.invoke(run, ["-vvv", "--no_bars", "test_builder.json"])
assert result.exit_code == 0
assert "Get" not in result.output
assert "Update" not in result.output
result = runner.invoke(run, ["-v", "-n", "2", "test_builder.json"])
assert result.exit_code == 0
assert "CopyBuilder" in result.output
assert "MultiProcessor" in result.output
result = runner.invoke(
run, ["-vvv", "-n", "2", "--no_bars", "test_builder.json"])
assert result.exit_code == 0
assert "Get" not in result.output
assert "Update" not in result.output
def test_simple_runner(self):
builder = CopyBuilder(self.source, self.target)
runner = Runner([builder])
dumpfn(runner, self.runner_filename)
p = subprocess.run("python -m maggma.cli.mrun {}".format(
self.runner_filename).split(),
timeout=15)
self.assertEqual(p.returncode, 0)
def setUp(self):
tic = datetime.now()
toc = tic + timedelta(seconds=1)
keys = list(range(20))
self.old_docs = [{"lu": tic, "k": k, "v": "old"} for k in keys]
self.new_docs = [{"lu": toc, "k": k, "v": "new"} for k in keys[:10]]
kwargs = dict(key="k", lu_field="lu")
self.source = MongoStore(self.dbname, "source", **kwargs)
self.target = MongoStore(self.dbname, "target", **kwargs)
self.builder = CopyBuilder(self.source, self.target)
self.source.connect()
self.source.ensure_index(self.source.key)
self.source.ensure_index(self.source.lu_field)
self.target.connect()
self.target.ensure_index(self.target.key)
self.target.ensure_index(self.target.lu_field)
def test_get_items(source, target, old_docs):
builder = CopyBuilder(source, target)
source.update(old_docs)
assert len(list(builder.get_items())) == len(old_docs)
target.update(old_docs)
assert len(list(builder.get_items())) == 0
builder = CopyBuilder(source, target, projection=["k"])
target.remove_docs({})
assert len(list(builder.get_items())) == len(old_docs)
assert all("v" not in d for d in builder.get_items())
class TestCopyBuilder(TestCase):
@classmethod
def setUpClass(cls):
cls.dbname = "test_" + uuid4().hex
s = MongoStore(cls.dbname, "test")
s.connect()
cls.client = s.collection.database.client
@classmethod
def tearDownClass(cls):
cls.client.drop_database(cls.dbname)
def setUp(self):
tic = datetime.now()
toc = tic + timedelta(seconds=1)
keys = list(range(20))
self.old_docs = [{"lu": tic, "k": k, "v": "old"} for k in keys]
self.new_docs = [{"lu": toc, "k": k, "v": "new"} for k in keys[:10]]
kwargs = dict(key="k", lu_field="lu")
self.source = MongoStore(self.dbname, "source", **kwargs)
self.target = MongoStore(self.dbname, "target", **kwargs)
self.builder = CopyBuilder(self.source, self.target)
self.source.connect()
self.source.ensure_index(self.source.key)
self.source.ensure_index(self.source.lu_field)
self.target.connect()
self.target.ensure_index(self.target.key)
self.target.ensure_index(self.target.lu_field)
def tearDown(self):
self.source.collection.drop()
self.target.collection.drop()
def test_get_items(self):
self.source.collection.insert_many(self.old_docs)
self.assertEqual(len(list(self.builder.get_items())),
len(self.old_docs))
self.target.collection.insert_many(self.old_docs)
self.assertEqual(len(list(self.builder.get_items())), 0)
self.source.update(self.new_docs, update_lu=False)
self.assertEqual(len(list(self.builder.get_items())),
len(self.new_docs))
def test_process_item(self):
self.source.collection.insert_many(self.old_docs)
items = list(self.builder.get_items())
self.assertCountEqual(items, map(self.builder.process_item, items))
def test_update_targets(self):
self.source.collection.insert_many(self.old_docs)
self.source.update(self.new_docs, update_lu=False)
self.target.collection.insert_many(self.old_docs)
items = list(map(self.builder.process_item, self.builder.get_items()))
self.builder.update_targets(items)
self.assertEqual(self.target.query_one(criteria={"k": 0})["v"], "new")
self.assertEqual(self.target.query_one(criteria={"k": 10})["v"], "old")
@unittest.skip(
"Have to refactor how we force read-only so a warning will get thrown")
def test_index_warning(self):
"""Should log warning when recommended store indexes are not present."""
self.source.collection.drop_index([(self.source.key, 1)])
with self.assertLogs(level=logging.WARNING) as cm:
list(self.builder.get_items())
self.assertIn("Ensure indices", "\n".join(cm.output))
def test_run(self):
self.source.collection.insert_many(self.old_docs)
self.source.update(self.new_docs, update_lu=False)
self.target.collection.insert_many(self.old_docs)
self.builder.run()
self.assertEqual(self.target.query_one(criteria={"k": 0})["v"], "new")
self.assertEqual(self.target.query_one(criteria={"k": 10})["v"], "old")
def test_query(self):
self.builder.query = {"k": {"$gt": 5}}
self.source.collection.insert_many(self.old_docs)
self.source.update(self.new_docs, update_lu=False)
self.builder.run()
all_docs = list(self.target.query(criteria={}))
self.assertEqual(len(all_docs), 14)
self.assertTrue(min([d['k'] for d in all_docs]), 6)
def test_delete_orphans(self):
self.builder = CopyBuilder(self.source,
self.target,
delete_orphans=True)
self.source.collection.insert_many(self.old_docs)
self.source.update(self.new_docs, update_lu=False)
self.target.collection.insert_many(self.old_docs)
deletion_criteria = {"k": {"$in": list(range(5))}}
self.source.collection.delete_many(deletion_criteria)
self.builder.run()
self.assertEqual(
self.target.collection.count_documents(deletion_criteria), 0)
self.assertEqual(self.target.query_one(criteria={"k": 5})["v"], "new")
self.assertEqual(self.target.query_one(criteria={"k": 10})["v"], "old")
def test_incremental_false(self):
tic = datetime.now()
toc = tic + timedelta(seconds=1)
keys = list(range(20))
earlier = [{"lu": tic, "k": k, "v": "val"} for k in keys]
later = [{"lu": toc, "k": k, "v": "val"} for k in keys]
self.source.collection.insert_many(earlier)
self.target.collection.insert_many(later)
query = {"k": {"$gt": 5}}
self.builder = CopyBuilder(self.source,
self.target,
incremental=False,
query=query)
self.builder.run()
docs = sorted(self.target.query(), key=lambda d: d["k"])
self.assertTrue(all(d["lu"] == tic) for d in docs[5:])
self.assertTrue(all(d["lu"] == toc) for d in docs[:5])
def test_update_targets(source, target, old_docs, new_docs):
builder = CopyBuilder(source, target)
builder.update_targets(old_docs)
builder.update_targets(new_docs)
assert target.query_one(criteria={"k": 0})["v"] == "new"
assert target.query_one(criteria={"k": 10})["v"] == "old"
def test_process_item(source, target, old_docs):
builder = CopyBuilder(source, target)
source.update(old_docs)
items = list(builder.get_items())
assert len(items) == len(list(map(builder.process_item, items)))
def test_get_items(source, target, old_docs, some_failed_old_docs):
builder = CopyBuilder(source, target)
source.update(old_docs)
assert len(list(builder.get_items())) == len(old_docs)
target.update(old_docs)
assert len(list(builder.get_items())) == 0
builder = CopyBuilder(source, target, projection=["k"])
target.remove_docs({})
assert len(list(builder.get_items())) == len(old_docs)
assert all("v" not in d for d in builder.get_items())
source.update(some_failed_old_docs)
target.update(old_docs)
target.update(some_failed_old_docs)
builder = CopyBuilder(source, target)
assert len(list(builder.get_items())) == 0
builder = CopyBuilder(source, target, retry_failed=True)
assert len(list(builder.get_items())) == len(some_failed_old_docs)
builder = CopyBuilder(source, target, query={"k": {"$lt": 11}})
assert len(list(builder.get_items())) == 0
builder = CopyBuilder(source,
target,
retry_failed=True,
query={"k": {
"$lt": 11
}})
assert len(list(builder.get_items())) == 3