def migrate(muri, dbname, huri): "Migrate data from MongoDB (muri) to HDFS (huri)" mstg = MongoStorage(muri, dbname) hstg = HdfsStorage(huri) # read data from MongoDB query = {'stype': mstg.stype} mdocs = mstg.read(query) mids = [d['wmaid'] for d in mdocs] # do nothing if no documents is found if not len(mdocs): return # store data to HDFS wmaid = hstg.write(mdocs) # read data from HDFS hdocs = hstg.read(wmaid) # now we can compare MongoDB docs with HDFS docs, a la cross-check for mdoc, hdoc in zip(mdocs, hdocs): # drop WMArchive keys for key in ['stype', 'wmaid']: if key in mdoc: del mdoc[key] if key in hdoc: del hdoc[key] if mdoc != hdoc: print("ERROR", mdoc, hdoc) sys.exit(1) # update status attributes of docs in MongoDB query = {'$set': {'stype': hstg.stype}} mstg.update(mids, query)
def migrate(muri, huri): "Migrate data from MongoDB (muri) to HDFS (huri)" mstg = MongoStorage(muri) hstg = HdfsStorage(huri) # read data from MongoDB query = {'stype': mstg.stype} mdocs = mstg.read(query) mids = [d['wmaid'] for d in mdocs] # do nothing if no documents is found if not len(mdocs): return # store data to HDFS wmaid = hstg.write(mdocs) # read data from HDFS hdocs = hstg.read(wmaid) # now we can compare MongoDB docs with HDFS docs, a la cross-check for mdoc, hdoc in zip(mdocs, hdocs): # drop WMArchive keys for key in ['stype', 'wmaid']: if key in mdoc: del mdoc[key] if key in hdoc: del hdoc[key] if mdoc != hdoc: print("ERROR", mdoc, hdoc) sys.exit(1) # update status attributes of docs in MongoDB query = {'$set' : {'stype': hstg.stype}} mstg.update(mids, query)
class MongoStorageTest(unittest.TestCase): def setUp(self): uri = os.environ.get("WMA_MONGODB", "mongodb://localhost:8230") self.dbname = "test_fwjr" try: self.mgr = MongoStorage(uri, dbname=self.dbname) self.mgr.remove() except: self.mgr = None print("WARNING: cannot connect to %s" % uri) data = { "int": 1, "float": 1.2, "list": [1, 2, 3], "dict": {"dname": "foo", "dval": 1}, "listdict": [{"lname": "foo"}], "str": "string", } self.bare_data = dict(data) data["wmaid"] = wmaHash(data) data["stype"] = "mongodb" self.data = data def tearDown(self): "Tear down content of temp dir" self.mgr.remove() self.mgr.dropdb(self.dbname) def test_write(self): "Test write functionality" if self.mgr: wmaids = self.mgr.write(self.data) self.assertEqual(len(wmaids), 1) data = self.mgr.read(wmaids[0]) record = data[0] for key in ["wmaid", "stype"]: if key in record: del record[key] self.assertEqual(record, self.bare_data) data = self.mgr.read(wmaids[0], ["dict"]) self.assertEqual(1, len(data)) self.assertEqual(data[0]["dict"], self.bare_data["dict"])
def migrate(muri): "Write and read data to MongoDB" mstg = MongoStorage(muri, dbname='test_fwjr') doc = {"test": 1, 'wmaid': 1} mstg.write([doc]) # read data from MongoDB query = {} mdocs = mstg.read(query) for doc in mdocs: print(doc) if '_id' in doc: print("found _id in doc") print(doc) break
def migrate(muri): "Write and read data to MongoDB" mstg = MongoStorage(muri, dbname='test_fwjr') doc = {"test":1, 'wmaid':1} mstg.write([doc]) # read data from MongoDB query = {} mdocs = mstg.read(query) for doc in mdocs: print(doc) if '_id' in doc: print("found _id in doc") print(doc) break