Handle nested query parameters in ApiClient.do() (#249)

## Changes
The Query History list API filter_by query parameter is modeled by a
dictionary, rather than a primitive type, defining the allowed filters
including `query_start_time_range`, `statuses`, `user_ids`, and
`warehouse_ids`. To be compatible with gRPC transcoding, query
parameters modeled by message types as opposed to primitives need to be
separated into one query parameter per nested field, where the key is
the path to that query parameter. For example:

```
{'filter_by': {'user_ids': [123, 456]}}
```
becomes
```
filter_by.user_ids=123&filter_by.user_ids=456
```
For this to be compatible with the requests library we use today, we
need to convert the first dictionary to
```
{'filter_by.user_ids': [123, 456]}
```

This resolves one of the problems from #99. The issue with the conflict
between filter_by and next_page_token will be resolved by the backend
service.

## Tests
Added an integration test that covers query history listing with a
filter_by parameter.

- [x] `make test` run locally
- [x] `make fmt` applied
- [x] relevant integration tests applied

databricks/sdk/core.py

@@ -14,7 +14,7 @@
 import urllib.parse
 from datetime import datetime
 from json import JSONDecodeError
-from typing import Callable, Dict, Iterable, List, Optional, Union
+from typing import Any, Callable, Dict, Iterable, List, Optional, Union
 
 import requests
 import requests.auth
@@ -911,7 +911,27 @@ def _fix_query_string(query: Optional[dict] = None) -> Optional[dict]:
         # See: https://github.com/databricks/databricks-sdk-py/issues/142
         if query is None:
             return None
-        return {k: v if type(v) != bool else ('true' if v else 'false') for k, v in query.items()}
+        with_fixed_bools = {k: v if type(v) != bool else ('true' if v else 'false') for k, v in query.items()}
+
+        # Query parameters may be nested, e.g.
+        # {'filter_by': {'user_ids': [123, 456]}}
+        # The HTTP-compatible representation of this is
+        # filter_by.user_ids=123&filter_by.user_ids=456
+        # To achieve this, we convert the above dictionary to
+        # {'filter_by.user_ids': [123, 456]}
+        # See the following for more information:
+        # https://cloud.google.com/endpoints/docs/grpc-service-config/reference/rpc/google.api#google.api.HttpRule
+        def flatten_dict(d: Dict[str, Any]) -> Dict[str, Any]:
+            for k1, v1 in d.items():
+                if isinstance(v1, dict):
+                    v1 = dict(flatten_dict(v1))
+                    for k2, v2 in v1.items():
+                        yield f"{k1}.{k2}", v2
+                else:
+                    yield k1, v1
+
+        flattened = dict(flatten_dict(with_fixed_bools))
+        return flattened
 
     def do(self,
            method: str,

tests/integration/test_sql.py

@@ -0,0 +1,15 @@
+from datetime import datetime
+
+from databricks.sdk.service.sql import QueryFilter, TimeRange
+
+
+def test_query_history_list_with_filter(w):
+
+    def date_to_ms(date):
+        return int(datetime.strptime(date, '%Y-%m-%d').timestamp() * 1000)
+
+    filter = QueryFilter(query_start_time_range=TimeRange(start_time_ms=date_to_ms('2023-01-01'),
+                                                          end_time_ms=date_to_ms('2023-01-02')))
+    queries = w.query_history.list(filter_by=filter)
+    for q in queries:
+        print(q)