def update_bookmark(self, state, value):
current_bookmark = self.get_bookmark(state)
if value and utils.strptime_with_tz(value) > current_bookmark:
singer.write_bookmark(state, self.name, self.replication_key,
value)
def sync_records(ns, catalog_entry, state, counter):
chunked_bookmark = singer_utils.strptime_with_tz(
ns.get_start_date(state, catalog_entry))
stream = catalog_entry['stream']
schema = catalog_entry['schema']
stream_alias = catalog_entry.get('stream_alias')
catalog_metadata = metadata.to_map(catalog_entry['metadata'])
replication_key = catalog_metadata.get((), {}).get('replication-key')
stream_version = get_stream_version(catalog_entry, state)
activate_version_message = singer.ActivateVersionMessage(
stream=(stream_alias or stream), version=stream_version)
start_time = singer_utils.now()
LOGGER.info('Syncing NetSuite data for stream %s', stream)
previous_max_replication_key = None
query_func = ns.query
for rec in query_func(ns, catalog_entry, state):
counter.increment()
with Transformer(
pre_hook=transform_data_hook(ns, stream)) as transformer:
rec = transformer.transform(rec, schema)
singer.write_message(
singer.RecordMessage(stream=(stream_alias or stream),
record=rec,
version=stream_version,
time_extracted=start_time))
if replication_key:
_rec = rec.get(replication_key, None)
original_replication_key_value = ""
replication_key_value = None
if replication_key and _rec is not None:
original_replication_key_value = _rec
replication_key_value = singer_utils.strptime_with_tz(
original_replication_key_value)
# Before writing a bookmark, make sure Quickbooks has not given us a
# record with one outside our range
if previous_max_replication_key is None or (
replication_key_value
and replication_key_value <= start_time
and replication_key_value > previous_max_replication_key):
state = singer.write_bookmark(state,
catalog_entry['tap_stream_id'],
replication_key,
original_replication_key_value)
previous_max_replication_key = replication_key_value
# Tables with no replication_key will send an
# activate_version message for the next sync
if not replication_key:
singer.write_message(activate_version_message)
state = singer.write_bookmark(state, catalog_entry['tap_stream_id'],
'version', None)
def test_run(self):
# Connect to stitch service.
runner.run_check_job_and_check_status(self)
# Get and check streams.
self.found_catalogs = menagerie.get_catalogs(self.conn_id)
self.assertEqual(len(self.found_catalogs), 9, msg="unable to locate schemas for connection {}".format(self.conn_id))
# Match streams.
our_catalogs = [c for c in self.found_catalogs if c.get('tap_stream_id') in self.expected_sync_streams()]
for c in our_catalogs:
c_annotated = menagerie.get_annotated_schema(self.conn_id, c['stream_id'])
c_metadata = metadata.to_map(c_annotated['metadata'])
connections.select_catalog_and_fields_via_metadata(self.conn_id, c, c_annotated, [], [])
# Clear state before our run
menagerie.set_state(self.conn_id, {})
# Run a sync job using orchestrator, verify tap and target exit codes
# and verify actual rows were synced.
first_sync_record_count = self.run_sync(self.conn_id)
replicated_row_count = reduce(lambda accum, c : accum + c, first_sync_record_count.values())
self.assertGreater(replicated_row_count, 0, msg="failed to replicate any data: {}".format(first_sync_record_count))
print("total replicated row count: {}".format(replicated_row_count))
# Get incremental vs. non-incremental streams.
non_incremental_streams = {key for key, value in self.expected_replication_method().items() if value != 'INCREMENTAL'}
incremental_streams = {key for key, value in self.expected_replication_method().items() if value == 'INCREMENTAL'}
# Get bookmark and state data for first sync, excluding full table streams.
first_sync_state = menagerie.get_state(self.conn_id)
first_sync_records = runner.get_records_from_target_output()
for v in non_incremental_streams:
first_sync_records.pop(v, None)
first_max_bookmarks = self.max_bookmarks_by_stream(first_sync_records)
first_min_bookmarks = self.min_bookmarks_by_stream(first_sync_records)
# Run a second sync job using orchestrator.
second_sync_record_count = self.run_sync(self.conn_id)
# Get data about rows synced, excluding full table streams.
second_sync_records = runner.get_records_from_target_output()
for v in non_incremental_streams:
second_sync_records.pop(v, None)
second_min_bookmarks = self.min_bookmarks_by_stream(second_sync_records)
for stream in incremental_streams:
if stream in {'tasks', 'groups'}:
continue
with self.subTest(stream=stream):
# get bookmark values from state and target data
stream_bookmark_key = self.expected_rks().get(stream, set())
assert len(stream_bookmark_key) == 1 # There shouldn't be a compound replication key
stream_bookmark_key = stream_bookmark_key.pop()
if not first_sync_state.get("bookmarks", {}).get(stream, None):
# Some streams require more than a free tier plan (tasks)
continue
state_value = first_sync_state.get("bookmarks", {}).get(
stream, {None: None}).get(stream_bookmark_key)
target_value = first_max_bookmarks.get(
stream, {None: None}).get(stream_bookmark_key)
target_min_value = first_min_bookmarks.get(
stream, {None: None}).get(stream_bookmark_key)
# Convert everything to datetime.
state_value = utils.strptime_with_tz(state_value)
target_value = utils.strptime_with_tz(target_value)
target_min_value = utils.strptime_with_tz(target_min_value)
# verify that there is data with different bookmark values - setup necessary
self.assertTrue(target_value >= target_min_value, msg="Data isn't set up to be able to test bookmarks")
# verify state agrees with target data after 1st sync
self.assertEqual(state_value, target_value, msg="The bookmark value isn't correct based on target data")
# verify that you get less data the 2nd time around
self.assertGreater(
first_sync_record_count.get(stream, 0),
second_sync_record_count.get(stream, 0),
msg="second sync didn't have less records, bookmark usage not verified")
if len(second_sync_records) > 0 and len(second_min_bookmarks) > 0:
# verify all data from 2nd sync >= 1st bookmark
target_value = second_min_bookmarks.get(stream, {None: None}).get(stream_bookmark_key)
target_value = utils.strptime_with_tz(target_value)
# verify that the minimum bookmark sent to the target for the second sync
# is greater than or equal to the bookmark from the first sync
self.assertTrue(target_value >= state_value)
def get_attribution_window_bookmark(customer_id, stream_name):
mid_bk_value = bookmarks.get_bookmark(
STATE, state_key_name(customer_id, stream_name),
'last_attribution_window_date')
return utils.strptime_with_tz(mid_bk_value) if mid_bk_value else None
def is_session_bookmark_old(self, value):
if self.session_bookmark is None:
return True
return utils.strptime_with_tz(value) > utils.strptime_with_tz(
self.session_bookmark)
def sync_deals(STATE, ctx):
catalog = ctx.get_catalog_from_id(singer.get_currently_syncing(STATE))
mdata = metadata.to_map(catalog.get('metadata'))
bookmark_key = 'hs_lastmodifieddate'
start = utils.strptime_with_tz(get_start(STATE, "deals", bookmark_key))
max_bk_value = start
LOGGER.info("sync_deals from %s", start)
most_recent_modified_time = start
params = {'count': 250, 'includeAssociations': False, 'properties': []}
schema = load_schema("deals")
singer.write_schema("deals", schema, ["dealId"], [bookmark_key],
catalog.get('stream_alias'))
# Check if we should include associations
for key in mdata.keys():
if 'associations' in key:
assoc_mdata = mdata.get(key)
if (assoc_mdata.get('selected')
and assoc_mdata.get('selected') == True):
params['includeAssociations'] = True
v3_fields = None
has_selected_properties = mdata.get(('properties', 'properties'),
{}).get('selected')
if has_selected_properties or has_selected_custom_field(mdata):
# On 2/12/20, hubspot added a lot of additional properties for
# deals, and appending all of them to requests ended up leading to
# 414 (url-too-long) errors. Hubspot recommended we use the
# `includeAllProperties` and `allpropertiesFetchMode` params
# instead.
params['includeAllProperties'] = True
params['allPropertiesFetchMode'] = 'latest_version'
# Grab selected `hs_date_entered/exited` fields to call the v3 endpoint with
v3_fields = [
x[1].replace('property_', '') for x, y in mdata.items()
if x and (y.get('selected') == True or has_selected_properties) and
('hs_date_entered' in x[1] or 'hs_date_exited' in x[1])
]
url = get_url('deals_all')
with Transformer(UNIX_MILLISECONDS_INTEGER_DATETIME_PARSING) as bumble_bee:
for row in gen_request(STATE,
'deals',
url,
params,
'deals',
"hasMore", ["offset"], ["offset"],
v3_fields=v3_fields):
row_properties = row['properties']
modified_time = None
if bookmark_key in row_properties:
# Hubspot returns timestamps in millis
timestamp_millis = row_properties[bookmark_key][
'timestamp'] / 1000.0
modified_time = datetime.datetime.fromtimestamp(
timestamp_millis, datetime.timezone.utc)
elif 'createdate' in row_properties:
# Hubspot returns timestamps in millis
timestamp_millis = row_properties['createdate'][
'timestamp'] / 1000.0
modified_time = datetime.datetime.fromtimestamp(
timestamp_millis, datetime.timezone.utc)
if modified_time and modified_time >= max_bk_value:
max_bk_value = modified_time
if not modified_time or modified_time >= start:
record = bumble_bee.transform(
lift_properties_and_versions(row), schema, mdata)
singer.write_record("deals",
record,
catalog.get('stream_alias'),
time_extracted=utils.now())
STATE = singer.write_bookmark(STATE, 'deals', bookmark_key,
utils.strftime(max_bk_value))
singer.write_state(STATE)
return STATE
def sync_deals(STATE, ctx):
catalog = ctx.get_catalog_from_id(singer.get_currently_syncing(STATE))
mdata = metadata.to_map(catalog.get('metadata'))
bookmark_key = 'hs_lastmodifieddate'
start = utils.strptime_with_tz(get_start(STATE, "deals", bookmark_key))
max_bk_value = start
LOGGER.info("sync_deals from %s", start)
most_recent_modified_time = start
params = {'limit': 100}
schema = load_schema("deals")
singer.write_schema("deals", schema, ["dealId"], [bookmark_key],
catalog.get('stream_alias'))
# We fetch all the deals properties
deals_v3_custom_schema = get_v3_schema("deals")
properties = []
for key, value in deals_v3_custom_schema.items():
properties.append(key)
# Splitting properties into chunks of max 100 properties
# to avoid asking for too many properties at once
# as properties names are passed in the URL
# URL have a safe length limit of 2000 chars, so 100 properties max should do it
property_chunks = []
while len(properties) > 100:
head, tail = head_100(properties)
property_chunks.append(head)
properties = tail
property_chunks.append(properties)
# TODO: Make it configurable through the singer catalog metadata selection pattern
associations = ["contacts"]
url = get_url('deals_v3_all')
with Transformer(UNIX_MILLISECONDS_INTEGER_DATETIME_PARSING) as bumble_bee:
for row in gen_request_v3(STATE,
"deals",
url,
params,
"results",
custom_properties_chunks=property_chunks,
associations=associations):
modified_time = None
if 'updatedAt' in row:
# Hubspot returns timestamps in ISO 8601
modified_time = dateutil.parser.isoparse(row['updatedAt'])
if modified_time and modified_time >= max_bk_value:
max_bk_value = modified_time
if not modified_time or modified_time >= start:
record = bumble_bee.transform(
lift_properties_and_versions_v3(row), schema, mdata)
singer.write_record("deals",
record,
catalog.get('stream_alias'),
time_extracted=utils.now())
STATE = singer.write_bookmark(STATE, 'deals', bookmark_key,
utils.strftime(max_bk_value))
singer.write_state(STATE)
return STATE
def test_run(self):
conn_id = connections.ensure_connection(self)
#run in check mode
check_job_name = runner.run_check_mode(self, conn_id)
#verify check exit codes
exit_status = menagerie.get_exit_status(conn_id, check_job_name)
menagerie.verify_check_exit_status(self, exit_status, check_job_name)
found_catalogs = menagerie.get_catalogs(conn_id)
self.assertGreater(
len(found_catalogs),
0,
msg="unable to locate schemas for connection {}".format(conn_id))
found_catalog_names = set(
map(lambda c: c['tap_stream_id'], found_catalogs))
diff = self.expected_check_streams().symmetric_difference(
found_catalog_names)
self.assertEqual(
len(diff),
0,
msg="discovered schemas do not match: {}".format(diff))
print("discovered schemas are kosher")
# Select all Catalogs
for catalog in found_catalogs:
connections.select_catalog_and_fields_via_metadata(
conn_id, catalog,
menagerie.get_annotated_schema(conn_id, catalog['stream_id']))
#clear state
menagerie.set_state(conn_id, {})
sync_job_name = runner.run_sync_mode(self, conn_id)
#verify tap and target exit codes
exit_status = menagerie.get_exit_status(conn_id, sync_job_name)
menagerie.verify_sync_exit_status(self, exit_status, sync_job_name)
record_count_by_stream = runner.examine_target_output_file(
self, conn_id, self.expected_sync_streams(), self.expected_pks())
replicated_row_count = reduce(lambda accum, c: accum + c,
record_count_by_stream.values())
self.assertGreater(replicated_row_count,
0,
msg="failed to replicate any data: {}".format(
record_count_by_stream))
print("total replicated row count: {}".format(replicated_row_count))
max_bookmarks_from_records = runner.get_most_recent_records_from_target(
self, self.expected_bookmarks(),
self.get_properties()['start_date'])
start_of_today = utils.strftime(
datetime.datetime(datetime.datetime.utcnow().year,
datetime.datetime.utcnow().month,
datetime.datetime.utcnow().day, 0, 0, 0, 0,
datetime.timezone.utc))
max_bookmarks_from_records['subscription_changes'] = start_of_today
max_bookmarks_from_records['email_events'] = start_of_today
#if we didn't replicate data, the bookmark should be the start_date
for k in self.expected_bookmarks().keys():
if max_bookmarks_from_records.get(k) is None:
max_bookmarks_from_records[k] = utils.strftime(
datetime.datetime(2017, 5, 1, 0, 0, 0, 0,
datetime.timezone.utc))
state = menagerie.get_state(conn_id)
bookmarks = state.get('bookmarks')
bookmark_streams = set(state.get('bookmarks').keys())
#verify bookmarks and offsets
for k, v in sorted(list(self.expected_bookmarks().items())):
for w in v:
bk_value = bookmarks.get(k, {}).get(w)
self.assertEqual(
utils.strptime_with_tz(bk_value),
utils.strptime_with_tz(max_bookmarks_from_records[k]),
"Bookmark {} ({}) for stream {} should have been updated to {}"
.format(bk_value, w, k, max_bookmarks_from_records[k]))
print("bookmark {}({}) updated to {} from max record value {}".
format(k, w, bk_value, max_bookmarks_from_records[k]))
for k, v in self.expected_offsets().items():
self.assertEqual(
bookmarks.get(k, {}).get('offset', {}),
v,
msg="unexpected offset found for stream {} {}. state: {}".
format(k, v, state))
print("offsets {} cleared".format(k))
diff = bookmark_streams.difference(self.acceptable_bookmarks())
self.assertEqual(
len(diff),
0,
msg="Unexpected bookmarks: {} Expected: {} Actual: {}".format(
diff, self.acceptable_bookmarks(), bookmarks))
self.assertEqual(
state.get('currently_syncing'), None,
"Unexpected `currently_syncing` bookmark value: {} Expected: None".
format(state.get('currently_syncing')))
def _datetime_string_to_epoch(self, datetime_string):
return utils.strptime_with_tz(datetime_string).timestamp() * 1000
def sync_deals(STATE, ctx):
catalog = ctx.get_catalog_from_id(singer.get_currently_syncing(STATE))
mdata = metadata.to_map(catalog.get('metadata'))
bookmark_key = 'hs_lastmodifieddate'
start = utils.strptime_with_tz(get_start(STATE, "deals", bookmark_key))
max_bk_value = start
LOGGER.info("sync_deals from %s", start)
most_recent_modified_time = start
params = {'count': 250, 'includeAssociations': False, 'properties': []}
schema = load_schema("deals")
singer.write_schema("deals", schema, ["dealId"], [bookmark_key],
catalog.get('stream_alias'))
# Check if we should include associations
for key in mdata.keys():
if 'associations' in key:
assoc_mdata = mdata.get(key)
if (assoc_mdata.get('selected')
and assoc_mdata.get('selected') == True):
params['includeAssociations'] = True
# Append all the properties fields for deals to the request if
# properties is selectedOB
if mdata.get(('properties', 'properties'), {}).get('selected'):
additional_properties = schema.get("properties").get("properties").get(
"properties")
for key in additional_properties.keys():
params['properties'].append(key)
url = get_url('deals_all')
with Transformer(UNIX_MILLISECONDS_INTEGER_DATETIME_PARSING) as bumble_bee:
for row in gen_request(STATE, 'deals', url, params, 'deals', "hasMore",
["offset"], ["offset"]):
row_properties = row['properties']
modified_time = None
if bookmark_key in row_properties:
# Hubspot returns timestamps in millis
timestamp_millis = row_properties[bookmark_key][
'timestamp'] / 1000.0
modified_time = datetime.datetime.fromtimestamp(
timestamp_millis, datetime.timezone.utc)
elif 'createdate' in row_properties:
# Hubspot returns timestamps in millis
timestamp_millis = row_properties['createdate'][
'timestamp'] / 1000.0
modified_time = datetime.datetime.fromtimestamp(
timestamp_millis, datetime.timezone.utc)
if modified_time and modified_time >= max_bk_value:
max_bk_value = modified_time
if not modified_time or modified_time >= start:
record = bumble_bee.transform(row, schema, mdata)
singer.write_record("deals",
record,
catalog.get('stream_alias'),
time_extracted=utils.now())
STATE = singer.write_bookmark(STATE, 'deals', bookmark_key,
utils.strftime(max_bk_value))
singer.write_state(STATE)
return STATE
def _query_recur(self,
query,
catalog_entry,
start_date_str,
end_date=None,
retries=MAX_RETRIES):
params = {"q": query}
url = "{}/services/data/v41.0/queryAll".format(self.sf.instance_url)
headers = self.sf._get_standard_headers()
if end_date is None:
end_date = singer_utils.now()
if retries == 0:
raise TapSalesforceException(
"Ran out of retries attempting to query Salesforce Object {}".
format(catalog_entry['stream']))
retryable = False
try:
while True:
resp = self.sf._make_request('GET',
url,
headers=headers,
params=params)
resp_json = resp.json()
for rec in resp_json.get('records'):
yield rec
next_records_url = resp_json.get('nextRecordsUrl')
if next_records_url is None:
break
else:
url = "{}{}".format(self.sf.instance_url, next_records_url)
except HTTPError as ex:
response = ex.response.json()
if isinstance(
response,
list) and response[0].get("errorCode") == "QUERY_TIMEOUT":
start_date = singer_utils.strptime_with_tz(start_date_str)
day_range = (end_date - start_date).days
LOGGER.info(
"Salesforce returned QUERY_TIMEOUT querying %d days of %s",
day_range, catalog_entry['stream'])
retryable = True
else:
raise ex
if retryable:
start_date = singer_utils.strptime_with_tz(start_date_str)
half_day_range = (end_date - start_date) // 2
end_date = end_date - half_day_range
if half_day_range.days == 0:
raise TapSalesforceException(
"Attempting to query by 0 day range, this would cause infinite looping."
)
query = self.sf._build_query_string(
catalog_entry, singer_utils.strftime(start_date),
singer_utils.strftime(end_date))
for record in self._query_recur(query, catalog_entry,
start_date_str, end_date,
retries - 1):
yield record
def get_end_date():
if CONFIG.get('end_date'):
return utils.strptime_with_tz(CONFIG.get('end_date'))
return utils.now()
def get_start_for_stream(customer_id, stream_name):
bk_value = bookmarks.get_bookmark(STATE,
state_key_name(customer_id, stream_name),
'date')
bk_start_date = utils.strptime_with_tz(bk_value or CONFIG['start_date'])
return bk_start_date
def sync_deals(state: State):
bookmark_key = 'hs_lastmodifieddate'
start = utils.strptime_with_tz(get_start(state, "deals", bookmark_key))
max_bk_value = start
logger.info("sync_deals: start from %s", start)
params = {
'limit': 100,
'includeAssociations': True,
# On 2/12/20, hubspot added a lot of additional properties for
# deals, and appending all of them to requests ended up leading to
# 414 (url-too-long) errors. Hubspot recommended we use the
# `includeAllProperties` and `allpropertiesFetchMode` params
# instead.
'includeAllProperties': True,
'allPropertiesFetchMode': 'latest_version',
}
schema = load_schema("deals")
singer.write_schema("hubspot_deals", schema, ["dealId"], [bookmark_key])
v3_fields = None
# Grab selected `hs_date_entered/exited` fields to call the v3 endpoint with
v3_fields = []
for field_name in schema['properties']:
hs_field_name = field_name.replace('property_', '')
if any(hs_field_name.startswith(prefix) for prefix in V3_PREFIXES):
v3_fields.append(hs_field_name)
url = get_url('deals_all')
for row in gen_request(state,
'deals',
url,
params,
'deals',
"hasMore", ["offset"], ["offset"],
v3_fields=v3_fields):
row_properties = row['properties']
modified_time = None
if bookmark_key in row_properties:
# Hubspot returns timestamps in millis
timestamp_millis = row_properties[bookmark_key][
'timestamp'] / 1000.0
modified_time = datetime.datetime.fromtimestamp(
timestamp_millis, datetime.timezone.utc)
elif 'createdate' in row_properties:
# Hubspot returns timestamps in millis
timestamp_millis = row_properties['createdate'][
'timestamp'] / 1000.0
modified_time = datetime.datetime.fromtimestamp(
timestamp_millis, datetime.timezone.utc)
if modified_time and modified_time >= max_bk_value:
max_bk_value = modified_time
if not modified_time or modified_time >= start:
record = build_record(row, schema)
write_record('hubspot_deals', record)
state = singer.write_bookmark(state, 'hubspot_deals', bookmark_key,
utils.strftime(max_bk_value))
singer.write_state(state)
return state
def _query_recur(self,
query,
catalog_entry,
start_date_str,
end_date=None,
retries=MAX_RETRIES):
params = {"q": query}
url = "{}/services/data/v41.0/queryAll".format(self.sf.instance_url)
headers = self.sf._get_standard_headers()
sync_start = singer_utils.now()
if end_date is None:
end_date = sync_start
if retries == 0:
raise TapSalesforceException(
"Ran out of retries attempting to query Salesforce Object {}".
format(catalog_entry['stream']))
retryable = False
try:
for rec in self._sync_records(url, headers, params):
yield rec
# If the date range was chunked (an end_date was passed), sync
# from the end_date -> now
if end_date < sync_start:
next_start_date_str = singer_utils.strftime(end_date)
query = self.sf._build_query_string(catalog_entry,
next_start_date_str)
for record in self._query_recur(query,
catalog_entry,
next_start_date_str,
retries=retries):
yield record
except HTTPError as ex:
response = ex.response.json()
if isinstance(
response,
list) and response[0].get("errorCode") == "QUERY_TIMEOUT":
start_date = singer_utils.strptime_with_tz(start_date_str)
day_range = (end_date - start_date).days
LOGGER.info(
"Salesforce returned QUERY_TIMEOUT querying %d days of %s",
day_range, catalog_entry['stream'])
retryable = True
else:
raise ex
if retryable:
start_date = singer_utils.strptime_with_tz(start_date_str)
half_day_range = (end_date - start_date) // 2
end_date = end_date - half_day_range
if half_day_range.days == 0:
raise TapSalesforceException(
"Attempting to query by 0 day range, this would cause infinite looping."
)
query = self.sf._build_query_string(
catalog_entry, singer_utils.strftime(start_date),
singer_utils.strftime(end_date))
for record in self._query_recur(query, catalog_entry,
start_date_str, end_date,
retries - 1):
yield record
def sync_entity_chunked(state: State, entity_name, key_properties, path):
schema = load_schema(entity_name)
bookmark_key = 'startTimestamp'
singer.write_schema('hubspot_' + entity_name, schema, key_properties,
[bookmark_key])
start = get_start(state, entity_name, bookmark_key)
logger.info("sync_%s from %s", entity_name, start)
now = datetime.datetime.utcnow().replace(tzinfo=pytz.UTC)
now_ts = int(now.timestamp() * 1000)
start_ts = int(utils.strptime_with_tz(start).timestamp() * 1000)
url = get_url(entity_name)
if entity_name == 'email_events':
window_size = int(CONFIG['email_chunk_size'])
elif entity_name == 'subscription_changes':
window_size = int(CONFIG['subscription_chunk_size'])
with metrics.record_counter(entity_name) as counter:
while start_ts < now_ts:
end_ts = start_ts + window_size
params = {
'startTimestamp': start_ts,
'endTimestamp': end_ts,
'limit': 1000,
}
while True:
our_offset = singer.get_offset(state, entity_name)
if bool(our_offset) and our_offset.get('offset') is not None:
params[StateFields.offset] = our_offset.get('offset')
data = request(url, params).json()
if data.get(path) is None:
raise RuntimeError(
"Unexpected API response: {} not in {}".format(
path, data.keys()))
for row in data[path]:
counter.increment()
record = build_record(row, schema)
write_record(entity_name, record)
if data.get('hasMore'):
state = singer.set_offset(state, entity_name, 'offset',
data['offset'])
singer.write_state(state)
else:
state = singer.clear_offset(state, entity_name)
singer.write_state(state)
break
state = singer.write_bookmark(state, "hubspot_" + entity_name, 'startTimestamp', utils.strftime(
datetime.datetime.fromtimestamp((start_ts / 1000),
datetime.timezone.utc))) # pylint: disable=line-too-long
singer.write_state(state)
start_ts = end_ts
state = singer.clear_offset(state, entity_name)
singer.write_state(state)
return state
def get_start_for_stream(config, state, advertiser_ids, stream_name):
"""Get start date for stream sync."""
bk_value = bookmarks.get_bookmark(
state, state_key_name(advertiser_ids, stream_name), "date")
bk_start_date = utils.strptime_with_tz(bk_value or config["start_date"])
return bk_start_date
def append_times_to_dates(item, date_fields):
if date_fields:
for date_field in date_fields:
if item.get(date_field):
item[date_field] = utils.strftime(
utils.strptime_with_tz(item[date_field]))
def get_end_date(config):
"""Get end date from config file."""
if config.get("end_date"):
return utils.strptime_with_tz(config.get("end_date"))
return utils.now()
def is_bookmark_old(self, state, value, name=None):
current_bookmark = self.get_bookmark(state, name)
return utils.strptime_with_tz(value) > utils.strptime_with_tz(
current_bookmark)
def sync_records(qb, catalog_entry, state, counter, state_passed):
chunked_bookmark = singer_utils.strptime_with_tz(qb.get_start_date(state, catalog_entry))
stream = catalog_entry['stream']
schema = catalog_entry['schema']
stream_alias = catalog_entry.get('stream_alias')
catalog_metadata = metadata.to_map(catalog_entry['metadata'])
replication_key = catalog_metadata.get((), {}).get('replication-key')
stream_version = get_stream_version(catalog_entry, state)
activate_version_message = singer.ActivateVersionMessage(stream=(stream_alias or stream),
version=stream_version)
start_time = singer_utils.now()
LOGGER.info('Syncing Quickbooks data for stream %s', stream)
previous_max_replication_key = None;
query_func = qb.query
if stream.endswith("Report"):
query_func = qb.query_report
for rec in query_func(catalog_entry, state, state_passed):
counter.increment()
with Transformer(pre_hook=transform_data_hook) as transformer:
rec = transformer.transform(rec, schema)
singer.write_message(
singer.RecordMessage(
stream=(
stream_alias or stream),
record=rec,
version=stream_version,
time_extracted=start_time))
if replication_key:
jsonpath_expression = parse(f"$.{replication_key}")
_rec = {'MetaData': json.loads(rec.get('MetaData', {}))}
match = jsonpath_expression.find(_rec)
original_replication_key_value = ""
if replication_key and len(match) > 0:
original_replication_key_value = match[0].value
replication_key_value = singer_utils.strptime_with_tz(original_replication_key_value)
# Before writing a bookmark, make sure Quickbooks has not given us a
# record with one outside our range
if previous_max_replication_key is None or (
replication_key_value and replication_key_value <= start_time and replication_key_value > previous_max_replication_key
):
state = singer.write_bookmark(
state,
catalog_entry['tap_stream_id'],
replication_key,
original_replication_key_value)
previous_max_replication_key = replication_key_value
# Tables with no replication_key will send an
# activate_version message for the next sync
if not replication_key:
singer.write_message(activate_version_message)
state = singer.write_bookmark(
state, catalog_entry['tap_stream_id'], 'version', None)
def sync_entity_chunked(STATE, catalog, entity_name, key_properties, path):
schema = load_schema(entity_name)
bookmark_key = 'startTimestamp'
singer.write_schema(entity_name, schema, key_properties, [bookmark_key],
catalog.get('stream_alias'))
start = get_start(STATE, entity_name, bookmark_key)
LOGGER.info("sync_%s from %s", entity_name, start)
now = datetime.datetime.utcnow().replace(tzinfo=pytz.UTC)
now_ts = int(now.timestamp() * 1000)
start_ts = int(utils.strptime_with_tz(start).timestamp() * 1000)
url = get_url(entity_name)
mdata = metadata.to_map(catalog.get('metadata'))
if entity_name == 'email_events':
window_size = int(CONFIG['email_chunk_size'])
elif entity_name == 'subscription_changes':
window_size = int(CONFIG['subscription_chunk_size'])
with metrics.record_counter(entity_name) as counter:
while start_ts < now_ts:
end_ts = start_ts + window_size
params = {
'startTimestamp': start_ts,
'endTimestamp': end_ts,
'limit': 1000,
}
with Transformer(
UNIX_MILLISECONDS_INTEGER_DATETIME_PARSING) as bumble_bee:
while True:
our_offset = singer.get_offset(STATE, entity_name)
if bool(our_offset) and our_offset.get('offset') != None:
params[StateFields.offset] = our_offset.get('offset')
data = request(url, params).json()
time_extracted = utils.now()
for row in data[path]:
counter.increment()
record = bumble_bee.transform(
lift_properties_and_versions(row), schema, mdata)
singer.write_record(entity_name,
record,
catalog.get('stream_alias'),
time_extracted=time_extracted)
if data.get('hasMore'):
STATE = singer.set_offset(STATE, entity_name, 'offset',
data['offset'])
singer.write_state(STATE)
else:
STATE = singer.clear_offset(STATE, entity_name)
singer.write_state(STATE)
break
STATE = singer.write_bookmark(STATE, entity_name, 'startTimestamp', utils.strftime(datetime.datetime.fromtimestamp((start_ts / 1000), datetime.timezone.utc))) # pylint: disable=line-too-long
singer.write_state(STATE)
start_ts = end_ts
STATE = singer.clear_offset(STATE, entity_name)
singer.write_state(STATE)
return STATE
def get_bookmark(self, state):
return utils.strptime_with_tz(
singer.get_bookmark(state, self.name, self.replication_key))
def sync_records(sf, catalog_entry, state, counter):
chunked_bookmark = singer_utils.strptime_with_tz(sf.get_start_date(state, catalog_entry))
stream = catalog_entry['stream']
schema = catalog_entry['schema']
stream_alias = catalog_entry.get('stream_alias')
catalog_metadata = metadata.to_map(catalog_entry['metadata'])
replication_key = catalog_metadata.get((), {}).get('replication-key')
stream_version = get_stream_version(catalog_entry, state)
activate_version_message = singer.ActivateVersionMessage(stream=(stream_alias or stream),
version=stream_version)
start_time = singer_utils.now()
LOGGER.info('Syncing Salesforce data for stream %s', stream)
for rec in sf.query(catalog_entry, state):
counter.increment()
with Transformer(pre_hook=transform_bulk_data_hook) as transformer:
rec = transformer.transform(rec, schema)
rec = fix_record_anytype(rec, schema)
singer.write_message(
singer.RecordMessage(
stream=(
stream_alias or stream),
record=rec,
version=stream_version,
time_extracted=start_time))
replication_key_value = replication_key and singer_utils.strptime_with_tz(rec[replication_key])
if sf.pk_chunking:
if replication_key_value and replication_key_value <= start_time and replication_key_value > chunked_bookmark:
# Replace the highest seen bookmark and save the state in case we need to resume later
chunked_bookmark = singer_utils.strptime_with_tz(rec[replication_key])
state = singer.write_bookmark(
state,
catalog_entry['tap_stream_id'],
'JobHighestBookmarkSeen',
singer_utils.strftime(chunked_bookmark))
singer.write_state(state)
# Before writing a bookmark, make sure Salesforce has not given us a
# record with one outside our range
elif replication_key_value and replication_key_value <= start_time:
state = singer.write_bookmark(
state,
catalog_entry['tap_stream_id'],
replication_key,
rec[replication_key])
singer.write_state(state)
# Tables with no replication_key will send an
# activate_version message for the next sync
if not replication_key:
singer.write_message(activate_version_message)
state = singer.write_bookmark(
state, catalog_entry['tap_stream_id'], 'version', None)
# If pk_chunking is set, only write a bookmark at the end
if sf.pk_chunking:
# Write a bookmark with the highest value we've seen
state = singer.write_bookmark(
state,
catalog_entry['tap_stream_id'],
replication_key,
singer_utils.strftime(chunked_bookmark))
def _get_end_datetime(self, startDateTime):
endDateTime = utils.strptime_with_tz(startDateTime) + timedelta(
self.api_window_in_days)
return endDateTime.strftime("%Y-%m-%d %H:%M:%S")
