Fix csv download. (#2036)

superset/sql_lab.py

@@ -101,9 +101,10 @@ def get_sql_results(self, query_id, return_results=True, store_results=False):
         logging.exception(e)
         msg = "Template rendering failed: " + utils.error_msg_from_exception(e)
         handle_error(msg)
+
+    query.executed_sql = executed_sql
+    logging.info("Running query: \n{}".format(executed_sql))
     try:
-        query.executed_sql = executed_sql
-        logging.info("Running query: \n{}".format(executed_sql))
         result_proxy = engine.execute(query.executed_sql, schema=query.schema)
     except Exception as e:
         logging.exception(e)

superset/sql_parse.py

@@ -1,6 +1,6 @@
 import sqlparse
 from sqlparse.sql import IdentifierList, Identifier
-from sqlparse.tokens import Keyword, Name
+from sqlparse.tokens import DML, Keyword, Name
 
 RESULT_OPERATIONS = {'UNION', 'INTERSECT', 'EXCEPT'}
 PRECEDES_TABLE_NAME = {'FROM', 'JOIN', 'DESC', 'DESCRIBE', 'WITH'}
@@ -9,7 +9,6 @@ PRECEDES_TABLE_NAME = {'FROM', 'JOIN', 'DESC', 'DESCRIBE', 'WITH'}
 # TODO: some sql_lab logic here.
 class SupersetQuery(object):
     def __init__(self, sql_statement):
-        self._tokens = []
         self.sql = sql_statement
         self._table_names = set()
         self._alias_names = set()
@@ -23,9 +22,8 @@ class SupersetQuery(object):
     def tables(self):
         return self._table_names
 
-    # TODO: use sqlparse for this check.
     def is_select(self):
-        return self.sql.upper().startswith('SELECT')
+        return self._parsed[0].get_type() == 'SELECT'
 
     def stripped(self):
         sql = self.sql

superset/views.py

@@ -6,6 +6,7 @@ from __future__ import unicode_literals
 from datetime import datetime, timedelta
 import json
 import logging
+import pandas as pd
 import pickle
 import re
 import sys
@@ -2563,11 +2564,19 @@ class Superset(BaseSupersetView):
         if rejected_tables:
             flash(get_datasource_access_error_msg('{}'.format(rejected_tables)))
             return redirect('/')
-
-        sql = query.select_sql or query.sql
-        df = query.database.get_df(sql, query.schema)
-        # TODO(bkyryliuk): add compression=gzip for big files.
-        csv = df.to_csv(index=False, encoding='utf-8')
+        blob = None
+        if results_backend and query.results_key:
+            blob = results_backend.get(query.results_key)
+        if blob:
+            json_payload = zlib.decompress(blob)
+            obj = json.loads(json_payload)
+            df = pd.DataFrame.from_records(obj['data'])
+            csv = df.to_csv(index=False, encoding='utf-8')
+        else:
+            sql = query.select_sql or query.executed_sql
+            df = query.database.get_df(sql, query.schema)
+            # TODO(bkyryliuk): add compression=gzip for big files.
+            csv = df.to_csv(index=False, encoding='utf-8')
         response = Response(csv, mimetype='text/csv')
         response.headers['Content-Disposition'] = (
             'attachment; filename={}.csv'.format(query.name))

tests/core_tests.py

@@ -357,7 +357,7 @@ class CoreTests(SupersetTestCase):
             WHERE first_name='admin'
         """
         client_id = "{}".format(random.getrandbits(64))[:10]
-        self.run_sql(sql, client_id)
+        self.run_sql(sql, client_id, raise_on_error=True)
 
         resp = self.get_resp('/superset/csv/{}'.format(client_id))
         data = csv.reader(io.StringIO(resp))