def top_filters(request):
response = {'status': -1}
interface = request.POST.get('interface', OPTIMIZER.INTERFACE.get())
db_tables = json.loads(request.POST.get('dbTables', '[]'))
column_name = request.POST.get('columnName') # Unused
api = get_api(request, interface)
data = api.top_filters(db_tables=db_tables)
if data:
response['status'] = 0
response['values'] = data['results']
else:
response['message'] = 'Optimizer: %s' % data
if OPTIMIZER.APPLY_SENTRY_PERMISSIONS.get():
filtered_filters = []
for result in results['results']:
cols = [
_get_table_name(col['columnName'])
for col in result["popularValues"][0]["group"]
]
if len(cols) == len(list(_secure_results(cols, self.user))):
filtered_filters.append(result)
results['results'] = filtered_filters
return JsonResponse(response)
def getkey(table):
names = _get_table_name(table['name'])
return {
'server': get_hive_sentry_provider(),
'db': names['database'],
'table': names['table']
}
def top_tables(request):
response = {'status': -1}
interface = request.POST.get('interface', OPTIMIZER.INTERFACE.get())
database = request.POST.get('database', 'default')
limit = request.POST.get('len', 1000)
api = get_api(request, interface)
data = api.top_tables(database_name=database, page_size=limit)
if OPTIMIZER.APPLY_SENTRY_PERMISSIONS.get():
checker = get_checker(user=self.user)
action = 'SELECT'
def getkey(table):
names = _get_table_name(table['name'])
return {
'server': get_hive_sentry_provider(),
'db': names['database'],
'table': names['table']
}
data['results'] = list(
checker.filter_objects(data['results'], action, key=getkey))
tables = [{
'eid': table['eid'],
'database': _get_table_name(table['name'])['database'],
'name': _get_table_name(table['name'])['table'],
'popularity': table['workloadPercent'],
'column_count': table['columnCount'],
'patternCount': table['patternCount'],
'total': table['total'],
'is_fact': table['type'] != 'Dimension'
} for table in data['results']]
response['top_tables'] = tables
response['status'] = 0
return JsonResponse(response)
def upload_table_stats(request):
response = {'status': -1}
interface = request.POST.get('interface', OPTIMIZER.INTERFACE.get())
db_tables = json.loads(request.POST.get('db_tables', '[]'))
source_platform = json.loads(request.POST.get('sourcePlatform', '"hive"'))
with_ddl = json.loads(request.POST.get('with_ddl', 'false'))
with_table_stats = json.loads(request.POST.get('with_table', 'false'))
with_columns_stats = json.loads(request.POST.get('with_columns', 'false'))
table_ddls = []
table_stats = []
column_stats = []
if not OPTIMIZER.AUTO_UPLOAD_DDL.get():
with_ddl = False
if not OPTIMIZER.AUTO_UPLOAD_STATS.get():
with_table_stats = with_columns_stats = False
for db_table in db_tables:
path = _get_table_name(db_table)
try:
if with_ddl:
db = _get_db(request.user, source_type=source_platform)
query = hql_query(
'SHOW CREATE TABLE `%(database)s`.`%(table)s`' % path)
handle = db.execute_and_wait(query, timeout_sec=5.0)
if handle:
result = db.fetch(handle, rows=5000)
db.close(handle)
table_ddls.append(
(0, 0, ' '.join([row[0] for row in result.rows()]),
path['database']))
if with_table_stats:
mock_request = MockRequest(user=request.user,
source_platform=source_platform)
full_table_stats = json.loads(
get_table_stats(mock_request,
database=path['database'],
table=path['table']).content)
stats = dict((stat['data_type'], stat['comment'])
for stat in full_table_stats['stats'])
table_stats.append({
'table_name':
'%(database)s.%(table)s' % path, # DB Prefix
'num_rows':
stats.get('numRows', -1),
'last_modified_time':
stats.get('transient_lastDdlTime', -1),
'total_size':
stats.get('totalSize', -1),
'raw_data_size':
stats.get('rawDataSize', -1),
'num_files':
stats.get('numFiles', -1),
'num_partitions':
stats.get('numPartitions', -1),
# bytes_cached
# cache_replication
# format
})
if with_columns_stats:
if source_platform == 'impala':
colum_stats = json.loads(
get_table_stats(mock_request,
database=path['database'],
table=path['table'],
column=-1).content)['stats']
else:
colum_stats = [
json.loads(
get_table_stats(mock_request,
database=path['database'],
table=path['table'],
column=col).content)['stats']
for col in full_table_stats['columns'][:25]
]
raw_column_stats = [
dict([(key, val if val is not None else '')
for col_stat in col
for key, val in col_stat.items()])
for col in colum_stats
]
for col_stats in raw_column_stats:
column_stats.append({
'table_name':
'%(database)s.%(table)s' % path, # DB Prefix
'column_name':
col_stats['col_name'],
'data_type':
col_stats['data_type'],
"num_distinct":
int(col_stats.get('distinct_count'))
if col_stats.get('distinct_count') != '' else -1,
"num_nulls":
int(col_stats['num_nulls'])
if col_stats['num_nulls'] != '' else -1,
"avg_col_len":
int(float(col_stats['avg_col_len']))
if col_stats['avg_col_len'] != '' else -1,
"max_size":
int(float(col_stats['max_col_len']))
if col_stats['max_col_len'] != '' else -1,
"min":
col_stats['min']
if col_stats.get('min', '') != '' else -1,
"max":
col_stats['max']
if col_stats.get('max', '') != '' else -1,
"num_trues":
col_stats['num_trues']
if col_stats.get('num_trues', '') != '' else -1,
"num_falses":
col_stats['num_falses']
if col_stats.get('num_falses', '') != '' else -1,
})
except Exception as e:
LOG.exception('Skipping upload of %s: %s' % (db_table, e))
api = get_api(request, interface)
response['status'] = 0
if table_stats:
response['upload_table_stats'] = api.upload(
data=table_stats,
data_type='table_stats',
source_platform=source_platform)
response['upload_table_stats_status'] = 0 if response[
'upload_table_stats']['status']['state'] in ('WAITING', 'FINISHED',
'IN_PROGRESS') else -1
response['status'] = response['upload_table_stats_status']
if column_stats:
response['upload_cols_stats'] = api.upload(
data=column_stats,
data_type='cols_stats',
source_platform=source_platform)
response['upload_cols_stats_status'] = response['status'] if response[
'upload_cols_stats']['status']['state'] in ('WAITING', 'FINISHED',
'IN_PROGRESS') else -1
if response['upload_cols_stats_status'] != 0:
response['status'] = response['upload_cols_stats_status']
if table_ddls:
response['upload_table_ddl'] = api.upload(
data=table_ddls,
data_type='queries',
source_platform=source_platform)
response['upload_table_ddl_status'] = response['status'] if response[
'upload_table_ddl']['status']['state'] in ('WAITING', 'FINISHED',
'IN_PROGRESS') else -1
if response['upload_table_ddl_status'] != 0:
response['status'] = response['upload_table_ddl_status']
return JsonResponse(response)
