Add ability to read CSV without header row

pydax/loaders/_table.py

@@ -37,6 +37,8 @@ def load(self, path: Union[_typing.PathLike, Dict[str, str]], options: SchemaDic
                - ``columns`` key specifies the data type of each column. Each data type corresponds to a Pandas'
                  supported dtype. If unspecified, then it is default.
                - ``delimiter`` key specifies the delimiter of the input CSV file.
+               - ``no_header`` key specifies if the first row of the CSV file contains the headers. Defaults to False.
+                 If the value is set to anything "truthy" in Python, the first row of the CSV will be read as data.
                - ``encoding`` key specifies the encoding of the CSV file. Defaults to UTF-8.
         :raises TypeError: ``path`` is not a path object.
         """
@@ -55,9 +57,18 @@ def load(self, path: Union[_typing.PathLike, Dict[str, str]], options: SchemaDic
             else:
                 dtypes[column] = type_
 
+        names = None
+        header = None
+        if options.get('no_header'):
+            # If no header use the columns provided in schema
+            names = [*options.get('columns', {})]
+        else:
+            header = 'infer'
+
         return pd.read_csv(path, dtype=dtypes,
                            # The following line after "if" is for circumventing
                            # https://github.com/pandas-dev/pandas/issues/38489
                            parse_dates=parse_dates if len(parse_dates) > 0 else False,
+                           header=header, names=names,
                            encoding=options.get('encoding', 'utf-8'),
                            delimiter=options.get('delimiter', ','))

tests/test_loaders.py

@@ -243,3 +243,19 @@ def test_csv_pandas_loader_no_encoding(self, tmp_path, noaa_jfk_schema):
 
         del noaa_jfk_schema['subdatasets']['jfk_weather_cleaned']['format']['options']['encoding']
         self.test_csv_pandas_loader(tmp_path, noaa_jfk_schema)
+
+    def test_csv_pandas_header(self, tmp_path, noaa_jfk_schema):
+        "Test CSVPandasLoader header options"
+
+        noaa_jfk_schema['subdatasets']['jfk_weather_cleaned']['format']['options']['no_header'] = True
+        with pytest.raises(ValueError) as exinfo:  # Pandas should error from trying to read string as another dtype
+            Dataset(noaa_jfk_schema, tmp_path, mode=Dataset.InitializationMode.DOWNLOAD_AND_LOAD)
+        assert('could not convert string to float' in str(exinfo.value))
+
+        false_test_cases = [False, '', None]  # These should all be treated as False
+        for case in false_test_cases:
+            noaa_jfk_schema['subdatasets']['jfk_weather_cleaned']['format']['options']['no_header'] = case
+            self.test_csv_pandas_loader(tmp_path, noaa_jfk_schema)
+
+        del noaa_jfk_schema['subdatasets']['jfk_weather_cleaned']['format']['options']['no_header']
+        self.test_csv_pandas_loader(tmp_path, noaa_jfk_schema)