fix: crash with empty indexes (#39)

* fix: crash with empty indexes

* no cols no meta

* codecov check

* Update es/basesqlalchemy.py

Co-authored-by: Ville Brofeldt <33317356+villebro@users.noreply.github.com>

* black it

Co-authored-by: Ville Brofeldt <33317356+villebro@users.noreply.github.com>

es/basesqlalchemy.py

@@ -4,6 +4,7 @@
 from __future__ import unicode_literals
 
 import logging
+from typing import List
 
 import es
 from es import exceptions
@@ -145,10 +146,16 @@ def get_schema_names(self, connection, **kwargs):
     def has_table(self, connection, table_name, schema=None):
         return table_name in self.get_table_names(connection, schema)
 
-    def get_table_names(self, connection, schema=None, **kwargs):
+    def get_table_names(self, connection, schema=None, **kwargs) -> List[str]:
         query = "SHOW TABLES"
         result = connection.execute(query)
-        return [row.name for row in result if row.name[0] != "."]
+        # return a list of table names exclude hidden and empty indexes
+        return [
+            table.name
+            for table in result
+            if table.name[0] != "."
+            and len(self.get_columns(connection, table.name)) > 0
+        ]
 
     def get_view_names(self, connection, schema=None, **kwargs):
         return []

es/elastic/api.py

@@ -4,22 +4,22 @@
 from __future__ import unicode_literals
 
 import re
-from typing import Dict
+from typing import Any, Dict, Optional
 
 from elasticsearch import Elasticsearch, exceptions as es_exceptions
 from es import exceptions
 from es.baseapi import apply_parameters, BaseConnection, BaseCursor, check_closed, Type
 
 
 def connect(
-    host="localhost",
-    port=9200,
-    path="",
-    scheme="http",
-    user=None,
-    password=None,
-    context=None,
-    **kwargs,
+    host: str = "localhost",
+    port: int = 9200,
+    path: str = "",
+    scheme: str = "http",
+    user: Optional[str] = None,
+    password: Optional[str] = None,
+    context: Optional[Dict] = None,
+    **kwargs: Any,
 ):
     """
     Constructor for creating a connection to the database.
@@ -156,7 +156,7 @@ def get_array_type_columns(self, table_name: str) -> "Cursor":
         """
         array_columns = []
         try:
-            resp = self.es.search(index=table_name, size=1)
+            response = self.es.search(index=table_name, size=1)
         except es_exceptions.ConnectionError as e:
             raise exceptions.OperationalError(
                 f"Error connecting to {self.url}: {e.info}"
@@ -166,12 +166,15 @@ def get_array_type_columns(self, table_name: str) -> "Cursor":
                 f"Error ({e.error}): {e.info['error']['reason']}"
             )
         try:
-            _source = resp["hits"]["hits"][0]["_source"]
+            if response["hits"]["total"]["value"] == 0:
+                source = {}
+            else:
+                source = response["hits"]["hits"][0]["_source"]
         except KeyError as e:
             raise exceptions.DataError(
                 f"Error inferring array type columns {self.url}: {e}"
             )
-        for col_name, value in _source.items():
+        for col_name, value in source.items():
             # If it's a list (ES Array add to cursor)
             if isinstance(value, list):
                 if len(value) > 0:

es/opendistro/api.py

@@ -5,6 +5,7 @@
 
 import csv
 import re
+from typing import Any, Dict, Optional  # pragma: no cover
 
 from elasticsearch import Elasticsearch
 from es import exceptions
@@ -13,14 +14,14 @@
 
 
 def connect(
-    host="localhost",
-    port=443,
-    path="",
-    scheme="https",
-    user=None,
-    password=None,
-    context=None,
-    **kwargs,
+    host: str = "localhost",
+    port: int = 443,
+    path: str = "",
+    scheme: str = "https",
+    user: Optional[str] = None,
+    password: Optional[str] = None,
+    context: Optional[Dict] = None,
+    **kwargs: Any,
 ):  # pragma: no cover
     """
     Constructor for creating a connection to the database.

es/tests/fixtures/fixtures.py

@@ -103,3 +103,7 @@ def import_flights(base_url):
 def import_data1(base_url):
     path = os.path.join(os.path.dirname(__file__), "data1.json")
     import_file_to_es(base_url, path, "data1")
+
+
+def import_empty_index(base_url):
+    set_index_replica_zero(base_url, "empty_index")

es/tests/test_0_fixtures.py

@@ -1,7 +1,12 @@
 import os
 import unittest
 
-from .fixtures.fixtures import delete_index, import_data1, import_flights
+from .fixtures.fixtures import (
+    delete_index,
+    import_data1,
+    import_empty_index,
+    import_flights,
+)
 
 BASE_URL = "http://localhost:9200"
 
@@ -17,3 +22,7 @@ def test_data_flights(self):
     def test_data_data1(self):
         delete_index(self.base_url, "data1")
         import_data1(self.base_url)
+
+    def test_data_empty_index(self):
+        delete_index(self.base_url, "empty_index")
+        import_empty_index(self.base_url)