Python OplogThread.dump_collection方法代码示例

# 需要导入模块: from mongo_connector.oplog_manager import OplogThread [as 别名]
# 或者: from mongo_connector.oplog_manager.OplogThread import dump_collection [as 别名]
    def test_dump_collection(self):
        """Test the dump_collection method

        Cases:

        1. empty oplog
        2. non-empty oplog, with gridfs collections
        3. non-empty oplog, specified a namespace-set, none of the oplog
           entries are for collections in the namespace-set
        """

        # Test with empty oplog
        self.opman.oplog = self.primary_conn["test"]["emptycollection"]
        last_ts = self.opman.dump_collection()
        self.assertEqual(last_ts, None)

        # Test with non-empty oplog with gridfs collections
        self.opman.oplog = self.primary_conn["local"]["oplog.rs"]
        # Insert 10 gridfs files
        for i in range(10):
            fs = gridfs.GridFS(self.primary_conn["gridfs"],
                               collection="test" + str(i))
            fs.put(b"hello world")
        # Insert 1000 documents
        for i in range(1000):
            self.primary_conn["test"]["test"].insert_one({
                "i": i + 500
            })
        last_ts = self.opman.get_last_oplog_timestamp()
        self.assertEqual(last_ts, self.opman.dump_collection())
        self.assertEqual(len(self.opman.doc_managers[0]._search()), 1010)

        # Case 3
        # 1MB oplog so that we can rollover quickly
        repl_set = ReplicaSetSingle(oplogSize=1).start()
        conn = repl_set.client()
        opman = OplogThread(
            primary_client=conn,
            doc_managers=(DocManager(),),
            oplog_progress_dict=LockingDict(),
            namespace_config=NamespaceConfig(namespace_set=["test.test"]),
        )
        # Insert a document into an included collection
        conn["test"]["test"].insert_one({"test": 1})
        # Cause the oplog to rollover on a non-included collection
        while conn["local"]["oplog.rs"].find_one({"ns": "test.test"}):
            conn["test"]["ignored"].insert_many(
                [{"test": "1" * 1024} for _ in range(1024)])
        last_ts = opman.get_last_oplog_timestamp()
        self.assertEqual(last_ts, opman.dump_collection())
        self.assertEqual(len(opman.doc_managers[0]._search()), 1)
        conn.close()
        repl_set.stop()

# 需要导入模块: from mongo_connector.oplog_manager import OplogThread [as 别名]
# 或者: from mongo_connector.oplog_manager.OplogThread import dump_collection [as 别名]
    def test_dump_collection(self):
        """Test the dump_collection method

        Cases:

        1. empty oplog
        2. non-empty oplog
        3. non-empty oplog, specified a namespace-set, none of the oplog
           entries are for collections in the namespace-set
        """

        # Test with empty oplog
        self.opman.oplog = self.primary_conn["test"]["emptycollection"]
        last_ts = self.opman.dump_collection()
        self.assertEqual(last_ts, None)

        # Test with non-empty oplog
        self.opman.oplog = self.primary_conn["local"]["oplog.rs"]
        for i in range(1000):
            self.primary_conn["test"]["test"].insert_one({
                "i": i + 500
            })
        last_ts = self.opman.get_last_oplog_timestamp()
        self.assertEqual(last_ts, self.opman.dump_collection())
        self.assertEqual(len(self.opman.doc_managers[0]._search()), 1000)

        # Case 3
        # 1MB oplog so that we can rollover quickly
        repl_set = ReplicaSetSingle(oplogSize=1).start()
        conn = repl_set.client()
        dest_mapping_stru = DestMapping(["test.test"], [], {})
        opman = OplogThread(
            primary_client=conn,
            doc_managers=(DocManager(),),
            oplog_progress_dict=LockingDict(),
            dest_mapping_stru=dest_mapping_stru,
            ns_set=set(["test.test"])
        )
        # Insert a document into a ns_set collection
        conn["test"]["test"].insert_one({"test": 1})
        # Cause the oplog to rollover on a non-ns_set collection
        while conn["local"]["oplog.rs"].find_one({"ns": "test.test"}):
            conn["test"]["ignored"].insert_many(
                [{"test": "1" * 1024} for _ in range(1024)])
        last_ts = opman.get_last_oplog_timestamp()
        self.assertEqual(last_ts, opman.dump_collection())
        self.assertEqual(len(opman.doc_managers[0]._search()), 1)
        conn.close()
        repl_set.stop()

# 需要导入模块: from mongo_connector.oplog_manager import OplogThread [as 别名]
# 或者: from mongo_connector.oplog_manager.OplogThread import dump_collection [as 别名]

#.........这里部分代码省略.........
        # get_oplog_cursor fast-forwards *one doc beyond* the given timestamp
        doc1 = self.mongos_conn["test"]["mcsharded"].find_one(
            {"_id": next(cursor1)["o"]["_id"]})
        doc2 = self.mongos_conn["test"]["mcsharded"].find_one(
            {"_id": next(cursor2)["o"]["_id"]})
        self.assertEqual(doc1["i"], self.opman1.retrieve_doc(pivot1)["i"] + 1)
        self.assertEqual(doc2["i"], self.opman2.retrieve_doc(pivot2)["i"] + 1)

    def test_get_last_oplog_timestamp(self):
        """Test the get_last_oplog_timestamp method"""

        # "empty" the oplog
        self.opman1.oplog = self.shard1_conn["test"]["emptycollection"]
        self.opman2.oplog = self.shard2_conn["test"]["emptycollection"]
        self.assertEqual(self.opman1.get_last_oplog_timestamp(), None)
        self.assertEqual(self.opman2.get_last_oplog_timestamp(), None)

        # Test non-empty oplog
        self.opman1.oplog = self.shard1_conn["local"]["oplog.rs"]
        self.opman2.oplog = self.shard2_conn["local"]["oplog.rs"]
        for i in range(1000):
            self.mongos_conn["test"]["mcsharded"].insert({
                "i": i + 500
            })
        oplog1 = self.shard1_conn["local"]["oplog.rs"]
        oplog1 = oplog1.find().sort("$natural", pymongo.DESCENDING).limit(1)[0]
        oplog2 = self.shard2_conn["local"]["oplog.rs"]
        oplog2 = oplog2.find().sort("$natural", pymongo.DESCENDING).limit(1)[0]
        self.assertEqual(self.opman1.get_last_oplog_timestamp(),
                         oplog1["ts"])
        self.assertEqual(self.opman2.get_last_oplog_timestamp(),
                         oplog2["ts"])

    def test_dump_collection(self):
        """Test the dump_collection method

        Cases:

        1. empty oplog
        2. non-empty oplog
        """

        # Test with empty oplog
        self.opman1.oplog = self.shard1_conn["test"]["emptycollection"]
        self.opman2.oplog = self.shard2_conn["test"]["emptycollection"]
        last_ts1 = self.opman1.dump_collection()
        last_ts2 = self.opman2.dump_collection()
        self.assertEqual(last_ts1, None)
        self.assertEqual(last_ts2, None)

        # Test with non-empty oplog
        self.opman1.oplog = self.shard1_conn["local"]["oplog.rs"]
        self.opman2.oplog = self.shard2_conn["local"]["oplog.rs"]
        for i in range(1000):
            self.mongos_conn["test"]["mcsharded"].insert({
                "i": i + 500
            })
        last_ts1 = self.opman1.get_last_oplog_timestamp()
        last_ts2 = self.opman2.get_last_oplog_timestamp()
        self.assertEqual(last_ts1, self.opman1.dump_collection())
        self.assertEqual(last_ts2, self.opman2.dump_collection())
        self.assertEqual(len(self.opman1.doc_managers[0]._search()), 1000)

    def test_init_cursor(self):
        """Test the init_cursor method

# 需要导入模块: from mongo_connector.oplog_manager import OplogThread [as 别名]
# 或者: from mongo_connector.oplog_manager.OplogThread import dump_collection [as 别名]
class TestOplogManager(unittest.TestCase):
    """Defines all the testing methods, as well as a method that sets up the
        cluster
    """

    def setUp(self):
        _, _, self.primary_p = start_replica_set("test-oplog-manager")
        self.primary_conn = pymongo.MongoClient(mongo_host, self.primary_p)
        self.oplog_coll = self.primary_conn.local["oplog.rs"]
        self.opman = OplogThread(
            primary_conn=self.primary_conn,
            main_address="%s:%d" % (mongo_host, self.primary_p),
            oplog_coll=self.oplog_coll,
            is_sharded=False,
            doc_manager=DocManager(),
            oplog_progress_dict=LockingDict(),
            namespace_set=None,
            auth_key=None,
            auth_username=None,
            repl_set="test-oplog-manager",
        )

    def tearDown(self):
        try:
            self.opman.join()
        except RuntimeError:
            pass  # OplogThread may not have been started
        self.primary_conn.close()
        kill_replica_set("test-oplog-manager")

    def test_get_oplog_cursor(self):
        """Test the get_oplog_cursor method"""

        # timestamp is None - all oplog entries are returned.
        cursor = self.opman.get_oplog_cursor(None)
        self.assertEqual(cursor.count(), self.primary_conn["local"]["oplog.rs"].count())

        # earliest entry is the only one at/after timestamp
        doc = {"ts": bson.Timestamp(1000, 0), "i": 1}
        self.primary_conn["test"]["test"].insert(doc)
        latest_timestamp = self.opman.get_last_oplog_timestamp()
        cursor = self.opman.get_oplog_cursor(latest_timestamp)
        self.assertNotEqual(cursor, None)
        self.assertEqual(cursor.count(), 1)
        next_entry_id = next(cursor)["o"]["_id"]
        retrieved = self.primary_conn.test.test.find_one(next_entry_id)
        self.assertEqual(retrieved, doc)

        # many entries before and after timestamp
        self.primary_conn["test"]["test"].insert({"i": i} for i in range(2, 1002))
        oplog_cursor = self.oplog_coll.find(sort=[("ts", pymongo.ASCENDING)])

        # startup + insert + 1000 inserts
        self.assertEqual(oplog_cursor.count(), 2 + 1000)
        pivot = oplog_cursor.skip(400).limit(1)[0]

        goc_cursor = self.opman.get_oplog_cursor(pivot["ts"])
        self.assertEqual(goc_cursor.count(), 2 + 1000 - 400)

    def test_get_last_oplog_timestamp(self):
        """Test the get_last_oplog_timestamp method"""

        # "empty" the oplog
        self.opman.oplog = self.primary_conn["test"]["emptycollection"]
        self.assertEqual(self.opman.get_last_oplog_timestamp(), None)

        # Test non-empty oplog
        self.opman.oplog = self.primary_conn["local"]["oplog.rs"]
        for i in range(1000):
            self.primary_conn["test"]["test"].insert({"i": i + 500})
        oplog = self.primary_conn["local"]["oplog.rs"]
        oplog = oplog.find().sort("$natural", pymongo.DESCENDING).limit(1)[0]
        self.assertEqual(self.opman.get_last_oplog_timestamp(), oplog["ts"])

    def test_dump_collection(self):
        """Test the dump_collection method

        Cases:

        1. empty oplog
        2. non-empty oplog
        """

        # Test with empty oplog
        self.opman.oplog = self.primary_conn["test"]["emptycollection"]
        last_ts = self.opman.dump_collection()
        self.assertEqual(last_ts, None)

        # Test with non-empty oplog
        self.opman.oplog = self.primary_conn["local"]["oplog.rs"]
        for i in range(1000):
            self.primary_conn["test"]["test"].insert({"i": i + 500})
        last_ts = self.opman.get_last_oplog_timestamp()
        self.assertEqual(last_ts, self.opman.dump_collection())
        self.assertEqual(len(self.opman.doc_managers[0]._search()), 1000)

    def test_dump_collection_with_error(self):
        """Test the dump_collection method with invalid documents.

        Cases:
#.........这里部分代码省略.........

# 需要导入模块: from mongo_connector.oplog_manager import OplogThread [as 别名]
# 或者: from mongo_connector.oplog_manager.OplogThread import dump_collection [as 别名]

#.........这里部分代码省略.........
        pivot1 = oplog1.skip(400).limit(-1)[0]
        pivot2 = oplog2.skip(400).limit(-1)[0]

        cursor1 = self.opman1.get_oplog_cursor(pivot1["ts"])
        cursor2 = self.opman2.get_oplog_cursor(pivot2["ts"])
        self.assertEqual(cursor1.count(), oplog1_count - 400)
        self.assertEqual(cursor2.count(), oplog2_count - 400)

    def test_get_last_oplog_timestamp(self):
        """Test the get_last_oplog_timestamp method"""

        # "empty" the oplog
        self.opman1.oplog = self.shard1_conn["test"]["emptycollection"]
        self.opman2.oplog = self.shard2_conn["test"]["emptycollection"]
        self.assertEqual(self.opman1.get_last_oplog_timestamp(), None)
        self.assertEqual(self.opman2.get_last_oplog_timestamp(), None)

        # Test non-empty oplog
        self.opman1.oplog = self.shard1_conn["local"]["oplog.rs"]
        self.opman2.oplog = self.shard2_conn["local"]["oplog.rs"]
        for i in range(1000):
            self.mongos_conn["test"]["mcsharded"].insert_one({
                "i": i + 500
            })
        oplog1 = self.shard1_conn["local"]["oplog.rs"]
        oplog1 = oplog1.find().sort("$natural", pymongo.DESCENDING).limit(-1)[0]
        oplog2 = self.shard2_conn["local"]["oplog.rs"]
        oplog2 = oplog2.find().sort("$natural", pymongo.DESCENDING).limit(-1)[0]
        self.assertEqual(self.opman1.get_last_oplog_timestamp(),
                         oplog1["ts"])
        self.assertEqual(self.opman2.get_last_oplog_timestamp(),
                         oplog2["ts"])

    def test_dump_collection(self):
        """Test the dump_collection method

        Cases:

        1. empty oplog
        2. non-empty oplog
        """

        # Test with empty oplog
        self.opman1.oplog = self.shard1_conn["test"]["emptycollection"]
        self.opman2.oplog = self.shard2_conn["test"]["emptycollection"]
        last_ts1 = self.opman1.dump_collection()
        last_ts2 = self.opman2.dump_collection()
        self.assertEqual(last_ts1, None)
        self.assertEqual(last_ts2, None)

        # Test with non-empty oplog
        self.opman1.oplog = self.shard1_conn["local"]["oplog.rs"]
        self.opman2.oplog = self.shard2_conn["local"]["oplog.rs"]
        for i in range(1000):
            self.mongos_conn["test"]["mcsharded"].insert_one({
                "i": i + 500
            })
        last_ts1 = self.opman1.get_last_oplog_timestamp()
        last_ts2 = self.opman2.get_last_oplog_timestamp()
        self.assertEqual(last_ts1, self.opman1.dump_collection())
        self.assertEqual(last_ts2, self.opman2.dump_collection())
        self.assertEqual(len(self.opman1.doc_managers[0]._search()), 1000)

    def test_init_cursor(self):
        """Test the init_cursor method

# 需要导入模块: from mongo_connector.oplog_manager import OplogThread [as 别名]
# 或者: from mongo_connector.oplog_manager.OplogThread import dump_collection [as 别名]
class TestOplogManager(unittest.TestCase):
    """Defines all the testing methods, as well as a method that sets up the
        cluster
    """

    def setUp(self):
        self.repl_set = ReplicaSetSingle().start()
        self.primary_conn = self.repl_set.client()
        self.oplog_coll = self.primary_conn.local["oplog.rs"]
        self.opman = OplogThread(
            primary_client=self.primary_conn,
            doc_managers=(DocManager(),),
            oplog_progress_dict=LockingDict(),
            namespace_config=NamespaceConfig(
                namespace_options={"test.*": True, "gridfs.*": {"gridfs": True}}
            ),
        )

    def tearDown(self):
        try:
            self.opman.join()
        except RuntimeError:
            pass  # OplogThread may not have been started
        self.primary_conn.drop_database("test")
        close_client(self.primary_conn)
        self.repl_set.stop()

    def test_get_oplog_cursor(self):
        """Test the get_oplog_cursor method"""

        # timestamp is None - all oplog entries excluding no-ops are returned.
        cursor = self.opman.get_oplog_cursor(None)
        self.assertEqual(
            cursor.count(),
            self.primary_conn["local"]["oplog.rs"].find({"op": {"$ne": "n"}}).count(),
        )

        # earliest entry is the only one at/after timestamp
        doc = {"ts": bson.Timestamp(1000, 0), "i": 1}
        self.primary_conn["test"]["test"].insert_one(doc)
        latest_timestamp = self.opman.get_last_oplog_timestamp()
        cursor = self.opman.get_oplog_cursor(latest_timestamp)
        self.assertNotEqual(cursor, None)
        self.assertEqual(cursor.count(), 1)
        next_entry_id = next(cursor)["o"]["_id"]
        retrieved = self.primary_conn.test.test.find_one(next_entry_id)
        self.assertEqual(retrieved, doc)

        # many entries before and after timestamp
        self.primary_conn["test"]["test"].insert_many(
            [{"i": i} for i in range(2, 1002)]
        )
        oplog_cursor = self.oplog_coll.find(
            {"op": {"$ne": "n"}, "ns": {"$not": re.compile(r"\.(system|\$cmd)")}},
            sort=[("ts", pymongo.ASCENDING)],
        )

        # initial insert + 1000 more inserts
        self.assertEqual(oplog_cursor.count(), 1 + 1000)
        pivot = oplog_cursor.skip(400).limit(-1)[0]

        goc_cursor = self.opman.get_oplog_cursor(pivot["ts"])
        self.assertEqual(goc_cursor.count(), 1 + 1000 - 400)

    def test_get_last_oplog_timestamp(self):
        """Test the get_last_oplog_timestamp method"""

        # "empty" the oplog
        self.opman.oplog = self.primary_conn["test"]["emptycollection"]
        self.assertEqual(self.opman.get_last_oplog_timestamp(), None)

        # Test non-empty oplog
        self.opman.oplog = self.primary_conn["local"]["oplog.rs"]
        for i in range(1000):
            self.primary_conn["test"]["test"].insert_one({"i": i + 500})
        oplog = self.primary_conn["local"]["oplog.rs"]
        oplog = oplog.find().sort("$natural", pymongo.DESCENDING).limit(-1)[0]
        self.assertEqual(self.opman.get_last_oplog_timestamp(), oplog["ts"])

    def test_dump_collection(self):
        """Test the dump_collection method

        Cases:

        1. empty oplog
        2. non-empty oplog, with gridfs collections
        3. non-empty oplog, specified a namespace-set, none of the oplog
           entries are for collections in the namespace-set
        """

        # Test with empty oplog
        self.opman.oplog = self.primary_conn["test"]["emptycollection"]
        last_ts = self.opman.dump_collection()
        self.assertEqual(last_ts, None)

        # Test with non-empty oplog with gridfs collections
        self.opman.oplog = self.primary_conn["local"]["oplog.rs"]
        # Insert 10 gridfs files
        for i in range(10):
            fs = gridfs.GridFS(self.primary_conn["gridfs"], collection="test" + str(i))
#.........这里部分代码省略.........

# 需要导入模块: from mongo_connector.oplog_manager import OplogThread [as 别名]
# 或者: from mongo_connector.oplog_manager.OplogThread import dump_collection [as 别名]
class TestFilterFields(unittest.TestCase):

    @classmethod
    def setUpClass(cls):
        cls.repl_set = ReplicaSetSingle().start()
        cls.primary_conn = cls.repl_set.client()
        cls.oplog_coll = cls.primary_conn.local['oplog.rs']

    @classmethod
    def tearDownClass(cls):
        cls.primary_conn.drop_database("test")
        close_client(cls.primary_conn)
        cls.repl_set.stop()

    def setUp(self):
        self.dest_mapping_stru = DestMapping([], [], {})
        self.opman = OplogThread(
            primary_client=self.primary_conn,
            doc_managers=(DocManager(),),
            oplog_progress_dict=LockingDict(),
            dest_mapping_stru=self.dest_mapping_stru
        )

    def tearDown(self):
        try:
            self.opman.join()
        except RuntimeError:
            # OplogThread may not have been started
            pass

    def _check_fields(self, opman, fields, exclude_fields, projection):
        if fields:
            self.assertEqual(sorted(opman.fields), sorted(fields))
            self.assertEqual(opman._fields, set(fields))
        else:
            self.assertEqual(opman.fields, None)
            self.assertEqual(opman._fields, set([]))
        if exclude_fields:
            self.assertEqual(sorted(opman.exclude_fields),
                             sorted(exclude_fields))
            self.assertEqual(opman._exclude_fields, set(exclude_fields))
        else:
            self.assertEqual(opman.exclude_fields, None)
            self.assertEqual(opman._exclude_fields, set([]))

        self.assertEqual(opman._projection, projection)

    def test_filter_fields(self):
        docman = self.opman.doc_managers[0]
        conn = self.opman.primary_client

        include_fields = ["a", "b", "c"]
        exclude_fields = ["d", "e", "f"]

        # Set fields to care about
        self.opman.fields = include_fields
        # Documents have more than just these fields
        doc = {
            "a": 1, "b": 2, "c": 3,
            "d": 4, "e": 5, "f": 6,
            "_id": 1
        }
        db = conn['test']['test']
        db.insert_one(doc)
        assert_soon(lambda: db.count() == 1)
        self.opman.dump_collection()

        result = docman._search()[0]
        keys = result.keys()
        for inc, exc in zip(include_fields, exclude_fields):
            self.assertIn(inc, keys)
            self.assertNotIn(exc, keys)

    def test_filter_exclude_oplog_entry(self):
        # Test oplog entries: these are callables, since
        # filter_oplog_entry modifies the oplog entry in-place
        insert_op = lambda: {
            "op": "i",
            "o": {
                "_id": 0,
                "a": 1,
                "b": 2,
                "c": 3
            }
        }
        update_op = lambda: {
            "op": "u",
            "o": {
                "$set": {
                    "a": 4,
                    "b": 5
                },
                "$unset": {
                    "c": True
                }
            },
            "o2": {
                "_id": 1
            }
        }
#.........这里部分代码省略.........

# 需要导入模块: from mongo_connector.oplog_manager import OplogThread [as 别名]
# 或者: from mongo_connector.oplog_manager.OplogThread import dump_collection [as 别名]

#.........这里部分代码省略.........
            sort=[("ts", pymongo.ASCENDING)]
        )

        # startup + insert + 1000 inserts
        self.assertEqual(oplog_cursor.count(), 2 + 1000)
        pivot = oplog_cursor.skip(400).limit(1)[0]

        goc_cursor = self.opman.get_oplog_cursor(pivot["ts"])
        self.assertEqual(goc_cursor.count(), 2 + 1000 - 400)

        # get_oplog_cursor fast-forwards *one doc beyond* the given timestamp
        doc = self.primary_conn["test"]["test"].find_one(
            {"_id": next(goc_cursor)["o"]["_id"]})
        self.assertEqual(doc["i"], self.opman.retrieve_doc(pivot)["i"] + 1)

    def test_get_last_oplog_timestamp(self):
        """Test the get_last_oplog_timestamp method"""

        # "empty" the oplog
        self.opman.oplog = self.primary_conn["test"]["emptycollection"]
        self.assertEqual(self.opman.get_last_oplog_timestamp(), None)

        # Test non-empty oplog
        self.opman.oplog = self.primary_conn["local"]["oplog.rs"]
        for i in range(1000):
            self.primary_conn["test"]["test"].insert({
                "i": i + 500
            })
        oplog = self.primary_conn["local"]["oplog.rs"]
        oplog = oplog.find().sort("$natural", pymongo.DESCENDING).limit(1)[0]
        self.assertEqual(self.opman.get_last_oplog_timestamp(),
                         oplog["ts"])

    def test_dump_collection(self):
        """Test the dump_collection method

        Cases:

        1. empty oplog
        2. non-empty oplog
        """

        # Test with empty oplog
        self.opman.oplog = self.primary_conn["test"]["emptycollection"]
        last_ts = self.opman.dump_collection()
        self.assertEqual(last_ts, None)

        # Test with non-empty oplog
        self.opman.oplog = self.primary_conn["local"]["oplog.rs"]
        for i in range(1000):
            self.primary_conn["test"]["test"].insert({
                "i": i + 500
            })
        last_ts = self.opman.get_last_oplog_timestamp()
        self.assertEqual(last_ts, self.opman.dump_collection())
        self.assertEqual(len(self.opman.doc_managers[0]._search()), 1000)

    def test_init_cursor(self):
        """Test the init_cursor method

        Cases:

        1. no last checkpoint, no collection dump
        2. no last checkpoint, collection dump ok and stuff to dump
        3. no last checkpoint, nothing to dump, stuff in oplog
        4. no last checkpoint, nothing to dump, nothing in oplog

# 需要导入模块: from mongo_connector.oplog_manager import OplogThread [as 别名]
# 或者: from mongo_connector.oplog_manager.OplogThread import dump_collection [as 别名]

#.........这里部分代码省略.........
             'ns': {'$not': re.compile(r'\.(system|\$cmd)')}},
            sort=[("ts", pymongo.ASCENDING)])

        # initial insert + 1000 more inserts
        self.assertEqual(oplog_cursor.count(), 11 + 1000)
        pivot = oplog_cursor.skip(400).limit(-1)[0]

        goc_cursor = self.opman.get_oplog_cursor(pivot["ts"])
        self.assertEqual(goc_cursor.count(), 11 + 1000 - 400)

    def test_get_last_oplog_timestamp(self):
        """Test the get_last_oplog_timestamp method"""

        # empty oplog case has been tested in test_oplog_manager.py,
        # skip that here.

        # Put something in the dbs
        self.init_dbs()

        # Test non-empty oplog
        self.reset_opman(["includedb1.*", "includedb2.includecol1"], [], {})
        self.opman.oplog = self.primary_conn["local"]["oplog.rs"]
        for i in range(1000):
            self.primary_conn["includedb1"]["includecol1"].insert_one({
                "idb1col1": i + 500
            })
        oplog = self.primary_conn["local"]["oplog.rs"]
        oplog = oplog.find(
                {'op': {'$ne': 'n'}}).sort(
                "$natural", pymongo.DESCENDING).limit(-1)[0]
        self.assertEqual(self.opman.get_last_oplog_timestamp(),
                         oplog["ts"])

    def test_dump_collection(self):
        """Test the dump_collection method

        Cases:

        1. no namespace set is set
        2. include namespace set is set
        3. exclude namespace set is set

        empty oplog case has been tested in test_oplog_manager.py,
        skip that here.
        """

        # Put something in the dbs
        self.init_dbs()

        # no namespace set is set
        self.reset_opman([], [], {})
        self.opman.oplog = self.primary_conn["local"]["oplog.rs"]
        last_ts = self.opman.get_last_oplog_timestamp()
        self.assertEqual(last_ts, self.opman.dump_collection())
        self.assertEqual(len(self.opman.doc_managers[0]._search()), 10)

        # include namespace set is set
        self.reset_opman(["includedb1.*", "includedb2.includecol1"], [], {})
        self.opman.oplog = self.primary_conn["local"]["oplog.rs"]
        last_ts = self.opman.get_last_oplog_timestamp()
        self.assertEqual(last_ts, self.opman.dump_collection())
        self.assertEqual(len(self.opman.doc_managers[0]._search()), 6)

        # exclude namespace set is set
        self.reset_opman([], ["includedb2.excludecol2", "excludedb3.*"], {})
        self.opman.oplog = self.primary_conn["local"]["oplog.rs"]

# 需要导入模块: from mongo_connector.oplog_manager import OplogThread [as 别名]
# 或者: from mongo_connector.oplog_manager.OplogThread import dump_collection [as 别名]
class TestOplogManager(unittest.TestCase):
    """Defines all the testing methods, as well as a method that sets up the
        cluster
    """

    def setUp(self):
        self.repl_set = ReplicaSet().start()
        self.primary_conn = self.repl_set.client()
        self.oplog_coll = self.primary_conn.local['oplog.rs']
        self.opman = OplogThread(
            primary_client=self.primary_conn,
            doc_managers=(DocManager(),),
            oplog_progress_dict=LockingDict()
        )

    def tearDown(self):
        try:
            self.opman.join()
        except RuntimeError:
            pass                # OplogThread may not have been started
        self.primary_conn.drop_database("test")
        close_client(self.primary_conn)
        self.repl_set.stop()

    def test_get_oplog_cursor(self):
        '''Test the get_oplog_cursor method'''

        # timestamp is None - all oplog entries are returned.
        cursor = self.opman.get_oplog_cursor(None)
        self.assertEqual(cursor.count(),
                         self.primary_conn["local"]["oplog.rs"].count())

        # earliest entry is the only one at/after timestamp
        doc = {"ts": bson.Timestamp(1000, 0), "i": 1}
        self.primary_conn["test"]["test"].insert_one(doc)
        latest_timestamp = self.opman.get_last_oplog_timestamp()
        cursor = self.opman.get_oplog_cursor(latest_timestamp)
        self.assertNotEqual(cursor, None)
        self.assertEqual(cursor.count(), 1)
        next_entry_id = next(cursor)['o']['_id']
        retrieved = self.primary_conn.test.test.find_one(next_entry_id)
        self.assertEqual(retrieved, doc)

        # many entries before and after timestamp
        self.primary_conn["test"]["test"].insert_many(
            [{"i": i} for i in range(2, 1002)])
        oplog_cursor = self.oplog_coll.find(
            {'op': {'$ne': 'n'},
             'ns': {'$not': re.compile(r'\.(system|\$cmd)')}},
            sort=[("ts", pymongo.ASCENDING)]
        )

        # initial insert + 1000 more inserts
        self.assertEqual(oplog_cursor.count(), 1 + 1000)
        pivot = oplog_cursor.skip(400).limit(-1)[0]

        goc_cursor = self.opman.get_oplog_cursor(pivot["ts"])
        self.assertEqual(goc_cursor.count(), 1 + 1000 - 400)

    def test_get_last_oplog_timestamp(self):
        """Test the get_last_oplog_timestamp method"""

        # "empty" the oplog
        self.opman.oplog = self.primary_conn["test"]["emptycollection"]
        self.assertEqual(self.opman.get_last_oplog_timestamp(), None)

        # Test non-empty oplog
        self.opman.oplog = self.primary_conn["local"]["oplog.rs"]
        for i in range(1000):
            self.primary_conn["test"]["test"].insert_one({
                "i": i + 500
            })
        oplog = self.primary_conn["local"]["oplog.rs"]
        oplog = oplog.find().sort("$natural", pymongo.DESCENDING).limit(-1)[0]
        self.assertEqual(self.opman.get_last_oplog_timestamp(),
                         oplog["ts"])

    def test_dump_collection(self):
        """Test the dump_collection method

        Cases:

        1. empty oplog
        2. non-empty oplog
        """

        # Test with empty oplog
        self.opman.oplog = self.primary_conn["test"]["emptycollection"]
        last_ts = self.opman.dump_collection()
        self.assertEqual(last_ts, None)

        # Test with non-empty oplog
        self.opman.oplog = self.primary_conn["local"]["oplog.rs"]
        for i in range(1000):
            self.primary_conn["test"]["test"].insert_one({
                "i": i + 500
            })
        last_ts = self.opman.get_last_oplog_timestamp()
        self.assertEqual(last_ts, self.opman.dump_collection())
        self.assertEqual(len(self.opman.doc_managers[0]._search()), 1000)
#.........这里部分代码省略.........

# 需要导入模块: from mongo_connector.oplog_manager import OplogThread [as 别名]
# 或者: from mongo_connector.oplog_manager.OplogThread import dump_collection [as 别名]
class TestFilterFields(unittest.TestCase):
    @classmethod
    def setUpClass(cls):
        cls.repl_set = ReplicaSetSingle().start()
        cls.primary_conn = cls.repl_set.client()
        cls.oplog_coll = cls.primary_conn.local["oplog.rs"]

    @classmethod
    def tearDownClass(cls):
        cls.primary_conn.drop_database("test")
        close_client(cls.primary_conn)
        cls.repl_set.stop()

    def setUp(self):
        self.namespace_config = NamespaceConfig()
        self.opman = OplogThread(
            primary_client=self.primary_conn,
            doc_managers=(DocManager(),),
            oplog_progress_dict=LockingDict(),
            namespace_config=self.namespace_config,
        )

    def tearDown(self):
        try:
            self.opman.join()
        except RuntimeError:
            # OplogThread may not have been started
            pass

    def reset_include_fields(self, fields):
        self.opman.namespace_config = NamespaceConfig(include_fields=fields)

    def reset_exclude_fields(self, fields):
        self.opman.namespace_config = NamespaceConfig(exclude_fields=fields)

    def test_filter_fields(self):
        docman = self.opman.doc_managers[0]
        conn = self.opman.primary_client

        include_fields = ["a", "b", "c"]
        exclude_fields = ["d", "e", "f"]

        # Set fields to care about
        self.reset_include_fields(include_fields)
        # Documents have more than just these fields
        doc = {"a": 1, "b": 2, "c": 3, "d": 4, "e": 5, "f": 6, "_id": 1}
        db = conn["test"]["test"]
        db.insert_one(doc)
        assert_soon(lambda: db.count() == 1)
        self.opman.dump_collection()

        result = docman._search()[0]
        keys = result.keys()
        for inc, exc in zip(include_fields, exclude_fields):
            self.assertIn(inc, keys)
            self.assertNotIn(exc, keys)

    def test_filter_exclude_oplog_entry(self):
        # Test oplog entries: these are callables, since
        # filter_oplog_entry modifies the oplog entry in-place
        def insert_op():
            return {"op": "i", "o": {"_id": 0, "a": 1, "b": 2, "c": 3}}

        def update_op():
            return {
                "op": "u",
                "o": {"$set": {"a": 4, "b": 5}, "$unset": {"c": True}},
                "o2": {"_id": 1},
            }

        def filter_doc(document, fields):
            if fields and "_id" in fields:
                fields.remove("_id")
            return self.opman.filter_oplog_entry(document, exclude_fields=fields)

        # Case 0: insert op, no fields provided
        filtered = filter_doc(insert_op(), None)
        self.assertEqual(filtered, insert_op())

        # Case 1: insert op, fields provided
        filtered = filter_doc(insert_op(), ["c"])
        self.assertEqual(filtered["o"], {"_id": 0, "a": 1, "b": 2})

        # Case 2: insert op, fields provided, doc becomes empty except for _id
        filtered = filter_doc(insert_op(), ["a", "b", "c"])
        self.assertEqual(filtered["o"], {"_id": 0})

        # Case 3: update op, no fields provided
        filtered = filter_doc(update_op(), None)
        self.assertEqual(filtered, update_op())

        # Case 4: update op, fields provided
        filtered = filter_doc(update_op(), ["b"])
        self.assertNotIn("b", filtered["o"]["$set"])
        self.assertIn("a", filtered["o"]["$set"])
        self.assertEqual(filtered["o"]["$unset"], update_op()["o"]["$unset"])

        # Case 5: update op, fields provided, empty $set
        filtered = filter_doc(update_op(), ["a", "b"])
        self.assertNotIn("$set", filtered["o"])
#.........这里部分代码省略.........

# 需要导入模块: from mongo_connector.oplog_manager import OplogThread [as 别名]
# 或者: from mongo_connector.oplog_manager.OplogThread import dump_collection [as 别名]
class TestFilterFields(unittest.TestCase):

    @classmethod
    def setUpClass(cls):
        cls.repl_set = ReplicaSetSingle().start()
        cls.primary_conn = cls.repl_set.client()
        cls.oplog_coll = cls.primary_conn.local['oplog.rs']

    @classmethod
    def tearDownClass(cls):
        cls.primary_conn.drop_database("test")
        close_client(cls.primary_conn)
        cls.repl_set.stop()

    def setUp(self):
        self.namespace_config = NamespaceConfig()
        self.opman = OplogThread(
            primary_client=self.primary_conn,
            doc_managers=(DocManager(),),
            oplog_progress_dict=LockingDict(),
            namespace_config=self.namespace_config
        )

    def tearDown(self):
        try:
            self.opman.join()
        except RuntimeError:
            # OplogThread may not have been started
            pass

    def reset_include_fields(self, fields):
        self.opman.namespace_config = NamespaceConfig(include_fields=fields)

    def reset_exclude_fields(self, fields):
        self.opman.namespace_config = NamespaceConfig(exclude_fields=fields)

    def test_filter_fields(self):
        docman = self.opman.doc_managers[0]
        conn = self.opman.primary_client

        include_fields = ["a", "b", "c"]
        exclude_fields = ["d", "e", "f"]

        # Set fields to care about
        self.reset_include_fields(include_fields)
        # Documents have more than just these fields
        doc = {
            "a": 1, "b": 2, "c": 3,
            "d": 4, "e": 5, "f": 6,
            "_id": 1
        }
        db = conn['test']['test']
        db.insert_one(doc)
        assert_soon(lambda: db.count() == 1)
        self.opman.dump_collection()

        result = docman._search()[0]
        keys = result.keys()
        for inc, exc in zip(include_fields, exclude_fields):
            self.assertIn(inc, keys)
            self.assertNotIn(exc, keys)

    def test_filter_exclude_oplog_entry(self):
        # Test oplog entries: these are callables, since
        # filter_oplog_entry modifies the oplog entry in-place
        insert_op = lambda: {
            "op": "i",
            "o": {
                "_id": 0,
                "a": 1,
                "b": 2,
                "c": 3
            }
        }
        update_op = lambda: {
            "op": "u",
            "o": {
                "$set": {
                    "a": 4,
                    "b": 5
                },
                "$unset": {
                    "c": True
                }
            },
            "o2": {
                "_id": 1
            }
        }

        def filter_doc(document, fields):
            if fields and '_id' in fields:
                fields.remove('_id')
            return self.opman.filter_oplog_entry(
                document, exclude_fields=fields)

        # Case 0: insert op, no fields provided
        filtered = filter_doc(insert_op(), None)
        self.assertEqual(filtered, insert_op())

#.........这里部分代码省略.........