def testInsertList(self):
"""Test case - create collection and insert data -"""
try:
with Connection(cfgOb=self.__cfgOb, resourceName=self.__resourceName) as client:
mg = MongoDbUtil(client)
ok = mg.createCollection(self.__dbName, self.__collectionName)
self.assertTrue(ok)
ok = mg.databaseExists(self.__dbName)
self.assertTrue(ok)
ok = mg.collectionExists(self.__dbName, self.__collectionName)
self.assertTrue(ok)
#
dList = []
for ii in range(100):
dList.append(self.__makeDataObj(2, 5, 5, ii))
#
keyName = "DOC_ID"
rIdL = mg.insertList(self.__dbName, self.__collectionName, dList, keyNames=[keyName], salvage=True)
self.assertEqual(len(rIdL), len(dList))
#
# Note that dObj is mutated by additional key '_id' that is added on insert -
#
for ii, rId in enumerate(rIdL):
rObj = mg.fetchOne(self.__dbName, self.__collectionName, "_id", rId)
logger.debug("Return Object %s", pprint.pformat(rObj))
jj = int(rObj["DOC_ID"][4:])
self.assertEqual(len(dList[jj]), len(rObj))
self.assertEqual(dList[jj], rObj)
except Exception as e:
logger.exception("Failing with %s", str(e))
self.fail()
def testReplaceList(self):
"""Test case - create collection and insert document list - replace and upsert document list"""
try:
with Connection(cfgOb=self.__cfgOb, resourceName=self.__resourceName) as client:
nDocs = 10
mg = MongoDbUtil(client)
ok = mg.createCollection(self.__dbName, self.__collectionName)
self.assertTrue(ok)
ok = mg.databaseExists(self.__dbName)
self.assertTrue(ok)
ok = mg.collectionExists(self.__dbName, self.__collectionName)
self.assertTrue(ok)
#
dList = []
for ii in range(nDocs):
dObj = self.__makeDataObj(2, 5, 5, ii)
dList.append(dObj)
#
keyName = "DOC_ID"
rIdL = mg.insertList(self.__dbName, self.__collectionName, dList, keyNames=[keyName], salvage=True)
self.assertEqual(len(rIdL), len(dList))
#
for ii, rId in enumerate(rIdL):
rObj = mg.fetchOne(self.__dbName, self.__collectionName, "_id", rId)
# logger.debug("Return Object %s", pprint.pformat(rObj))
self.assertEqual(len(dList[ii]), len(rObj))
self.assertEqual(dList[ii], rObj)
#
# Replace with 2x the list length - half are duplicates id's
dList = []
for ii in range(nDocs + nDocs):
dObj = self.__makeDataObj(4, 10, 10, ii)
dList.append(dObj)
#
updL = mg.replaceList(self.__dbName, self.__collectionName, dList, ["DOC_ID"], upsertFlag=True)
#
logger.info("Upserted id list length %d", len(updL))
for ii in range(nDocs + nDocs):
kVal = "DOC_%d" % ii
rObj = mg.fetchOne(self.__dbName, self.__collectionName, "DOC_ID", kVal)
if not rObj:
logger.info("Failing to recover doc %s", kVal)
# logger.debug("Return Object %s", pprint.pformat(rObj))
rObj.pop("_id", None)
dList[ii].pop("_id", None)
self.assertEqual(len(dList[ii]), len(rObj))
self.assertEqual(dList[ii], rObj)
except Exception as e:
logger.exception("Failing with %s", str(e))
self.fail()
def testSingleIndex(self):
"""Test case - create collection, create simple single index, insert document list, read check documents"""
try:
with Connection(cfgOb=self.__cfgOb, resourceName=self.__resourceName) as client:
nDocs = 100
mg = MongoDbUtil(client)
ok = mg.createCollection(self.__dbName, self.__collectionName)
self.assertTrue(ok)
ok = mg.databaseExists(self.__dbName)
self.assertTrue(ok)
ok = mg.collectionExists(self.__dbName, self.__collectionName)
self.assertTrue(ok)
#
# Create before insert
ok = mg.createIndex(self.__dbName, self.__collectionName, keyList=["DOC_ID"], indexName="primary", indexType="DESCENDING", uniqueFlag=True)
self.assertTrue(ok)
dList = []
for ii in range(nDocs):
dObj = self.__makeDataObj(2, 5, 5, ii)
dList.append(dObj)
#
keyName = "DOC_ID"
rIdL = mg.insertList(self.__dbName, self.__collectionName, dList, keyNames=[keyName], salvage=True)
self.assertEqual(len(dList), len(rIdL))
#
for ii in range(nDocs):
kVal = "DOC_%d" % ii
rObj = mg.fetchOne(self.__dbName, self.__collectionName, "DOC_ID", kVal)
# logger.debug("Return Object %s" % pprint.pformat(rObj))
rObj.pop("_id", None)
dList[ii].pop("_id", None)
self.assertEqual(len(dList[ii]), len(rObj))
self.assertEqual(dList[ii], rObj)
#
ok = mg.dropIndex(self.__dbName, self.__collectionName, indexName="primary")
self.assertTrue(ok)
ok = mg.createIndex(self.__dbName, self.__collectionName, keyList=["DOC_ID"], indexName="primary", indexType="DESCENDING", uniqueFlag=True)
self.assertTrue(ok)
ok = mg.reIndex(self.__dbName, self.__collectionName)
self.assertTrue(ok)
except Exception as e:
logger.exception("Failing with %s", str(e))
self.fail()
def testSingleIndexSelect(self):
"""Test case - create collection, create simple single index, insert document list, read check documents."""
try:
logger.debug("Starting testSingleIndexSelect")
with Connection(cfgOb=self.__cfgOb, resourceName=self.__resourceName) as client:
nDocs = 100
mg = MongoDbUtil(client)
ok = mg.createCollection(self.__dbName, self.__collectionName)
self.assertTrue(ok)
ok = mg.databaseExists(self.__dbName)
self.assertTrue(ok)
ok = mg.collectionExists(self.__dbName, self.__collectionName)
self.assertTrue(ok)
#
# Create before insert
ok = mg.createIndex(self.__dbName, self.__collectionName, keyList=["DOC_ID"], indexName="primary", indexType="DESCENDING", uniqueFlag=True)
self.assertTrue(ok)
dList = []
nRows = 5
for ii in range(nDocs):
dObj = self.__makeDataObj(2, 5, nRows, ii)
dList.append(dObj)
#
keyName = "DOC_ID"
rIdL = mg.insertList(self.__dbName, self.__collectionName, dList, keyNames=[keyName], salvage=True)
self.assertEqual(len(dList), len(rIdL))
#
for ii in range(nDocs):
kVal = "DOC_%d" % ii
rObj = mg.fetchOne(self.__dbName, self.__collectionName, "DOC_ID", kVal)
# logger.debug("Return Object %s" % pprint.pformat(rObj))
rObj.pop("_id", None)
dList[ii].pop("_id", None)
self.assertEqual(len(dList[ii]), len(rObj))
self.assertEqual(dList[ii], rObj)
#
ok = mg.dropIndex(self.__dbName, self.__collectionName, indexName="primary")
self.assertTrue(ok)
ok = mg.createIndex(self.__dbName, self.__collectionName, keyList=["DOC_ID"], indexName="primary", indexType="DESCENDING", uniqueFlag=True)
self.assertTrue(ok)
ok = mg.reIndex(self.__dbName, self.__collectionName)
self.assertTrue(ok)
#
with Connection(cfgOb=self.__cfgOb, resourceName=self.__resourceName) as client:
mg = MongoDbUtil(client)
ii = mg.count(self.__dbName, self.__collectionName)
logger.debug("collection length %d", ii)
#
dList = mg.fetch(self.__dbName, self.__collectionName, ["DOC_ID"])
self.assertEqual(len(dList), nDocs)
logger.debug("Fetch length %d", len(dList))
for ii, dD in enumerate(dList):
logger.debug("Fetch num %d: %r", ii, dD)
#
dList = mg.fetch(self.__dbName, self.__collectionName, ["category_0.attribute_0"], queryD={"category_0.attribute_0": "val_0_0"})
self.assertEqual(len(dList), nDocs)
logger.debug("Fetch length %d", len(dList))
for ii, dD in enumerate(dList):
logger.debug("Fetch num %d: %r", ii, dD)
atName = "category_0.attribute_0"
vL0 = mg.distinct(self.__dbName, self.__collectionName, atName)
self.assertEqual(len(vL0), nRows + 2)
logger.debug("vL0 %r", vL0)
vL1 = mg.distinct(self.__dbName, self.__collectionName, "category_1.attribute_0")
self.assertEqual(len(vL1), nRows + 2)
for v in vL0:
num = mg.count(self.__dbName, self.__collectionName, countFilter={atName: v})
logger.debug("%s value %s (%d)", atName, v, num)
self.assertGreaterEqual(num, 100)
#
except Exception as e:
logger.exception("Failing with %s", str(e))
self.fail()
def __loadDocuments(self,
dbName,
collectionName,
docList,
loadType="full",
readBackCheck=False,
keyNames=None):
#
# Load database/collection with input document list -
#
failList = []
rIdL = []
successList = []
logger.debug(
"Loading dbName %s collectionName %s with document count %d keynames %r",
dbName, collectionName, len(docList), keyNames)
if keyNames:
# map the document list to some document key if this is provided
indD = {}
indL = []
try:
for ii, doc in enumerate(docList):
dIdTup = self.__getKeyValues(doc, keyNames)
indD[dIdTup] = ii
indL = list(range(len(docList)))
except Exception as e:
logger.exception("Failing ii %d d %r with %s", ii, doc, str(e))
try:
with Connection(cfgOb=self.__cfgOb,
resourceName=self.__resourceName) as client:
mg = MongoDbUtil(client)
#
if loadType == "replace" and keyNames:
dTupL = mg.deleteList(dbName, collectionName, docList,
keyNames)
logger.debug("Deleted document status %r", (dTupL, ))
#
rIdL = mg.insertList(dbName,
collectionName,
docList,
keyNames=keyNames)
logger.debug("Insert returns rIdL length %r", len(rIdL))
# ---
# If there is a failure then determine the specific successes and failures -
#
successList = docList
failList = []
if len(rIdL) != len(docList):
if keyNames:
successIndList = []
for rId in rIdL:
rObj = mg.fetchOne(dbName, collectionName, "_id",
rId)
dIdTup = self.__getKeyValues(rObj, keyNames)
successIndList.append(indD[dIdTup])
failIndList = list(set(indL) - set(successIndList))
failList = [docList[ii] for ii in failIndList]
successList = [docList[ii] for ii in successIndList]
else:
# fail the whole batch if we don't have visibility into each document
failList = docList
successList = []
#
rbStatus = True
if readBackCheck and keyNames:
#
# Note that objects in docList are mutated by the insert operation with the additional key '_id',
# hence, it is possible to compare the fetched object with the input object.
#
for ii, rId in enumerate(rIdL):
rObj = mg.fetchOne(dbName, collectionName, "_id", rId)
dIdTup = self.__getKeyValues(rObj, keyNames)
jj = indD[dIdTup]
if rObj != docList[jj]:
rbStatus = False
break
#
if readBackCheck and not rbStatus:
return False, successList, failList
#
return len(rIdL) == len(docList), successList, failList
except Exception as e:
logger.exception("Failing %r %r (len=%d) %s with %s", dbName,
collectionName, len(docList), keyNames, str(e))
return False, [], docList