fix: handle null values in data (#636)

* fix: handle null values in Flux data

* test: add tests for null value handling and extension dtypes

* test: fix failures with empty warnings cases

* test: comment out dtype some extra assertion until solved

* test: skip extension dtypes test on pythn 3.7

* fix: single place of dtypes conversion

* fix: bump pandas dependency version

* docs: update CHANGELOG

* chore(build): trigger CI/CD pipeline

* fix: add use_extension_dtypes also to async query API methods

CHANGELOG.md

@@ -1,5 +1,8 @@
 ## 1.41.0 [unreleased]
 
+### Bug Fixes
+1. [#636](https://github.com/influxdata/influxdb-client-python/pull/636): Handle missing data in data frames
+
 ## 1.40.0 [2024-01-30]
 
 ### Features

influxdb_client/client/_base.py

@@ -277,23 +277,27 @@ async def _to_flux_record_stream_async(self, response, query_options=None, respo
         return (await _parser.__aenter__()).generator_async()
 
     def _to_data_frame_stream(self, data_frame_index, response, query_options=None,
-                              response_metadata_mode: FluxResponseMetadataMode = FluxResponseMetadataMode.full):
+                              response_metadata_mode: FluxResponseMetadataMode = FluxResponseMetadataMode.full,
+                              use_extension_dtypes=False):
         """
         Parse HTTP response to DataFrame stream.
 
         :param response: HTTP response from an HTTP client. Expected type: `urllib3.response.HTTPResponse`.
         """
-        _parser = self._to_data_frame_stream_parser(data_frame_index, query_options, response, response_metadata_mode)
+        _parser = self._to_data_frame_stream_parser(data_frame_index, query_options, response, response_metadata_mode,
+                                                    use_extension_dtypes)
         return _parser.generator()
 
     async def _to_data_frame_stream_async(self, data_frame_index, response, query_options=None, response_metadata_mode:
-                                          FluxResponseMetadataMode = FluxResponseMetadataMode.full):
+                                          FluxResponseMetadataMode = FluxResponseMetadataMode.full,
+                                          use_extension_dtypes=False):
         """
         Parse HTTP response to DataFrame stream.
 
         :param response: HTTP response from an HTTP client. Expected type: `aiohttp.client_reqrep.ClientResponse`.
         """
-        _parser = self._to_data_frame_stream_parser(data_frame_index, query_options, response, response_metadata_mode)
+        _parser = self._to_data_frame_stream_parser(data_frame_index, query_options, response, response_metadata_mode,
+                                                    use_extension_dtypes)
         return (await _parser.__aenter__()).generator_async()
 
     def _to_tables_parser(self, response, query_options, response_metadata_mode):
@@ -304,10 +308,12 @@ def _to_flux_record_stream_parser(self, query_options, response, response_metada
         return FluxCsvParser(response=response, serialization_mode=FluxSerializationMode.stream,
                              query_options=query_options, response_metadata_mode=response_metadata_mode)
 
-    def _to_data_frame_stream_parser(self, data_frame_index, query_options, response, response_metadata_mode):
+    def _to_data_frame_stream_parser(self, data_frame_index, query_options, response, response_metadata_mode,
+                                     use_extension_dtypes):
         return FluxCsvParser(response=response, serialization_mode=FluxSerializationMode.dataFrame,
                              data_frame_index=data_frame_index, query_options=query_options,
-                             response_metadata_mode=response_metadata_mode)
+                             response_metadata_mode=response_metadata_mode,
+                             use_extension_dtypes=use_extension_dtypes)
 
     def _to_data_frames(self, _generator):
         """Parse stream of DataFrames into expected type."""

influxdb_client/client/flux_csv_parser.py

@@ -64,7 +64,8 @@ class FluxCsvParser(object):
 
     def __init__(self, response, serialization_mode: FluxSerializationMode,
                  data_frame_index: List[str] = None, query_options=None,
-                 response_metadata_mode: FluxResponseMetadataMode = FluxResponseMetadataMode.full) -> None:
+                 response_metadata_mode: FluxResponseMetadataMode = FluxResponseMetadataMode.full,
+                 use_extension_dtypes=False) -> None:
         """
         Initialize defaults.
 
@@ -75,6 +76,7 @@ def __init__(self, response, serialization_mode: FluxSerializationMode,
         self.tables = TableList()
         self._serialization_mode = serialization_mode
         self._response_metadata_mode = response_metadata_mode
+        self._use_extension_dtypes = use_extension_dtypes
         self._data_frame_index = data_frame_index
         self._data_frame_values = []
         self._profilers = query_options.profilers if query_options is not None else None
@@ -211,7 +213,7 @@ def _parse_flux_response_row(self, metadata, csv):
                         pass
                 else:
 
-                    # to int converions todo
+                    # to int conversions todo
                     current_id = int(csv[2])
                     if metadata.table_id == -1:
                         metadata.table_id = current_id
@@ -253,7 +255,11 @@ def _prepare_data_frame(self):
             _temp_df = _temp_df.set_index(self._data_frame_index)
 
         # Append data
-        return pd.concat([self._data_frame.astype(_temp_df.dtypes), _temp_df])
+        df = pd.concat([self._data_frame.astype(_temp_df.dtypes), _temp_df])
+
+        if self._use_extension_dtypes:
+            return df.convert_dtypes()
+        return df
 
     def parse_record(self, table_index, table, csv):
         """Parse one record."""
@@ -273,8 +279,10 @@ def _to_value(self, str_val, column):
             default_value = column.default_value
             if default_value == '' or default_value is None:
                 if self._serialization_mode is FluxSerializationMode.dataFrame:
-                    from ..extras import np
-                    return self._to_value(np.nan, column)
+                    if self._use_extension_dtypes:
+                        from ..extras import pd
+                        return pd.NA
+                    return None
                 return None
             return self._to_value(default_value, column)
 

influxdb_client/client/query_api.py

@@ -222,7 +222,8 @@ def query_stream(self, query: str, org=None, params: dict = None) -> Generator['
                                               async_req=False, _preload_content=False, _return_http_data_only=False)
         return self._to_flux_record_stream(response, query_options=self._get_query_options())
 
-    def query_data_frame(self, query: str, org=None, data_frame_index: List[str] = None, params: dict = None):
+    def query_data_frame(self, query: str, org=None, data_frame_index: List[str] = None, params: dict = None,
+                         use_extension_dtypes: bool = False):
         """
         Execute synchronous Flux query and return Pandas DataFrame.
 
@@ -234,6 +235,11 @@ def query_data_frame(self, query: str, org=None, data_frame_index: List[str] = N
                                       If not specified the default value from ``InfluxDBClient.org`` is used.
         :param data_frame_index: the list of columns that are used as DataFrame index
         :param params: bind parameters
+        :param use_extension_dtypes: set to ``True`` to use panda's extension data types.
+                                     Useful for queries with ``pivot`` function.
+                                     When data has missing values, column data type may change (to ``object`` or ``float64``).
+                                     Nullable extension types (``Int64``, ``Float64``, ``boolean``) support ``panda.NA`` value.
+                                     For more info, see https://pandas.pydata.org/docs/user_guide/missing_data.html.
         :return: :class:`~DataFrame` or :class:`~List[DataFrame]`
 
         .. warning:: For the optimal processing of the query results use the ``pivot() function`` which align results as a table.
@@ -250,10 +256,12 @@ def query_data_frame(self, query: str, org=None, data_frame_index: List[str] = N
                 - https://docs.influxdata.com/flux/latest/stdlib/universe/pivot/
                 - https://docs.influxdata.com/flux/latest/stdlib/influxdata/influxdb/schema/fieldsascols/
         """  # noqa: E501
-        _generator = self.query_data_frame_stream(query, org=org, data_frame_index=data_frame_index, params=params)
+        _generator = self.query_data_frame_stream(query, org=org, data_frame_index=data_frame_index, params=params,
+                                                  use_extension_dtypes=use_extension_dtypes)
         return self._to_data_frames(_generator)
 
-    def query_data_frame_stream(self, query: str, org=None, data_frame_index: List[str] = None, params: dict = None):
+    def query_data_frame_stream(self, query: str, org=None, data_frame_index: List[str] = None, params: dict = None,
+                                use_extension_dtypes: bool = False):
         """
         Execute synchronous Flux query and return stream of Pandas DataFrame as a :class:`~Generator[DataFrame]`.
 
@@ -265,6 +273,11 @@ def query_data_frame_stream(self, query: str, org=None, data_frame_index: List[s
                                       If not specified the default value from ``InfluxDBClient.org`` is used.
         :param data_frame_index: the list of columns that are used as DataFrame index
         :param params: bind parameters
+        :param use_extension_dtypes: set to ``True`` to use panda's extension data types.
+                                     Useful for queries with ``pivot`` function.
+                                     When data has missing values, column data type may change (to ``object`` or ``float64``).
+                                     Nullable extension types (``Int64``, ``Float64``, ``boolean``) support ``panda.NA`` value.
+                                     For more info, see https://pandas.pydata.org/docs/user_guide/missing_data.html.
         :return: :class:`~Generator[DataFrame]`
 
         .. warning:: For the optimal processing of the query results use the ``pivot() function`` which align results as a table.
@@ -289,7 +302,8 @@ def query_data_frame_stream(self, query: str, org=None, data_frame_index: List[s
 
         return self._to_data_frame_stream(data_frame_index=data_frame_index,
                                           response=response,
-                                          query_options=self._get_query_options())
+                                          query_options=self._get_query_options(),
+                                          use_extension_dtypes=use_extension_dtypes)
 
     def __del__(self):
         """Close QueryAPI."""

influxdb_client/client/query_api_async.py

@@ -120,7 +120,8 @@ async def query_stream(self, query: str, org=None, params: dict = None) -> Async
 
         return await self._to_flux_record_stream_async(response, query_options=self._get_query_options())
 
-    async def query_data_frame(self, query: str, org=None, data_frame_index: List[str] = None, params: dict = None):
+    async def query_data_frame(self, query: str, org=None, data_frame_index: List[str] = None, params: dict = None,
+                               use_extension_dtypes: bool = False):
         """
         Execute asynchronous Flux query and return :class:`~pandas.core.frame.DataFrame`.
 
@@ -132,6 +133,11 @@ async def query_data_frame(self, query: str, org=None, data_frame_index: List[st
                                       If not specified the default value from ``InfluxDBClientAsync.org`` is used.
         :param data_frame_index: the list of columns that are used as DataFrame index
         :param params: bind parameters
+        :param use_extension_dtypes: set to ``True`` to use panda's extension data types.
+                                     Useful for queries with ``pivot`` function.
+                                     When data has missing values, column data type may change (to ``object`` or ``float64``).
+                                     Nullable extension types (``Int64``, ``Float64``, ``boolean``) support ``panda.NA`` value.
+                                     For more info, see https://pandas.pydata.org/docs/user_guide/missing_data.html.
         :return: :class:`~DataFrame` or :class:`~List[DataFrame]`
 
         .. warning:: For the optimal processing of the query results use the ``pivot() function`` which align results as a table.
@@ -149,7 +155,7 @@ async def query_data_frame(self, query: str, org=None, data_frame_index: List[st
                 - https://docs.influxdata.com/flux/latest/stdlib/influxdata/influxdb/schema/fieldsascols/
         """  # noqa: E501
         _generator = await self.query_data_frame_stream(query, org=org, data_frame_index=data_frame_index,
-                                                        params=params)
+                                                        params=params, use_extension_dtypes=use_extension_dtypes)
 
         dataframes = []
         async for dataframe in _generator:
@@ -158,7 +164,7 @@ async def query_data_frame(self, query: str, org=None, data_frame_index: List[st
         return self._to_data_frames(dataframes)
 
     async def query_data_frame_stream(self, query: str, org=None, data_frame_index: List[str] = None,
-                                      params: dict = None):
+                                      params: dict = None, use_extension_dtypes: bool = False):
         """
         Execute asynchronous Flux query and return stream of :class:`~pandas.core.frame.DataFrame` as an AsyncGenerator[:class:`~pandas.core.frame.DataFrame`].
 
@@ -170,6 +176,11 @@ async def query_data_frame_stream(self, query: str, org=None, data_frame_index:
                                       If not specified the default value from ``InfluxDBClientAsync.org`` is used.
         :param data_frame_index: the list of columns that are used as DataFrame index
         :param params: bind parameters
+        :param use_extension_dtypes: set to ``True`` to use panda's extension data types.
+                                     Useful for queries with ``pivot`` function.
+                                     When data has missing values, column data type may change (to ``object`` or ``float64``).
+                                     Nullable extension types (``Int64``, ``Float64``, ``boolean``) support ``panda.NA`` value.
+                                     For more info, see https://pandas.pydata.org/docs/user_guide/missing_data.html.
         :return: :class:`AsyncGenerator[:class:`DataFrame`]`
 
         .. warning:: For the optimal processing of the query results use the ``pivot() function`` which align results as a table.
@@ -192,7 +203,8 @@ async def query_data_frame_stream(self, query: str, org=None, data_frame_index:
                                                                             dataframe_query=True))
 
         return await self._to_data_frame_stream_async(data_frame_index=data_frame_index, response=response,
-                                                      query_options=self._get_query_options())
+                                                      query_options=self._get_query_options(),
+                                                      use_extension_dtypes=use_extension_dtypes)
 
     async def query_raw(self, query: str, org=None, dialect=_BaseQueryApi.default_dialect, params: dict = None):
         """

setup.py

@@ -31,7 +31,7 @@
 ]
 
 extra_requires = [
-    'pandas>=0.25.3',
+    'pandas>=1.0.0',
     'numpy'
 ]