feat: adds datetime_format as an option

Author: chalmerlowe

google/cloud/bigquery/external_config.py

@@ -862,6 +862,20 @@ def date_format(self) -> Optional[str]:
     def date_format(self, value: Optional[str]):
         self._properties["dateFormat"] = value
 
+    @property
+    def datetime_format(self) -> Optional[str]:
+        """Optional[str]: Date format used for parsing DATETIME values.
+        (Valid for CSV and NEWLINE_DELIMITED_JSON)
+        See:
+        https://cloud.google.com/bigquery/docs/reference/rest/v2/tables#ExternalDataConfiguration.FIELDS.datetime_format
+        """
+        result = self._properties.get("datetimeFormat")
+        return typing.cast(str, result)
+
+    @datetime_format.setter
+    def datetime_format(self, value: Optional[str]):
+        self._properties["datetimeFormat"] = value
+
     @property
     def time_zone(self) -> Optional[str]:
         """Optional[str]: Time zone used when parsing timestamp values that do not

google/cloud/bigquery/job/load.py

@@ -561,6 +561,19 @@ def date_format(self) -> Optional[str]:
     def date_format(self, value: Optional[str]):
         self._set_sub_prop("dateFormat", value)
 
+    @property
+    def datetime_format(self) -> Optional[str]:
+        """Optional[str]: Date format used for parsing DATETIME values.
+        This option is valid for CSV and JSON sources.
+        See:
+        https://cloud.google.com/bigquery/docs/reference/rest/v2/Job#JobConfigurationLoad.FIELDS.datetime_format
+        """
+        return self._get_sub_prop("datetimeFormat")
+
+    @datetime_format.setter
+    def datetime_format(self, value: Optional[str]):
+        self._set_sub_prop("datetimeFormat", value)
+
     @property
     def time_zone(self) -> Optional[str]:
         """Optional[str]: Default time zone that will apply when parsing timestamp
@@ -923,6 +936,13 @@ def date_format(self):
         """
         return self.configuration.date_format
 
+    @property
+    def datetime_format(self):
+        """See
+        :attr:`google.cloud.bigquery.job.LoadJobConfig.datetime_format`.
+        """
+        return self.configuration.datetime_format
+
     @property
     def time_zone(self):
         """See

tests/unit/job/test_load.py

@@ -38,13 +38,15 @@ def _setUpConstants(self):
         self.OUTPUT_ROWS = 345
         self.REFERENCE_FILE_SCHEMA_URI = "gs://path/to/reference"
         self.DATE_FORMAT = "%Y-%m-%d"
+        self.DATETIME_FORMAT = "%Y-%m-%dT%H:%M:%S"
         self.TIME_ZONE = "UTC"
 
     def _make_resource(self, started=False, ended=False):
         resource = super(TestLoadJob, self)._make_resource(started, ended)
         config = resource["configuration"]["load"]
         config["sourceUris"] = [self.SOURCE1]
         config["dateFormat"] = self.DATE_FORMAT
+        config["datetimeFormat"] = self.DATETIME_FORMAT
         config["timeZone"] = self.TIME_ZONE
         config["destinationTable"] = {
             "projectId": self.PROJECT,
@@ -159,6 +161,10 @@ def _verifyResourceProperties(self, job, resource):
             self.assertEqual(job.date_format, config["dateFormat"])
         else:
             self.assertIsNone(job.date_format)
+        if "datetimeFormat" in config:
+            self.assertEqual(job.datetime_format, config["datetimeFormat"])
+        else:
+            self.assertIsNone(job.datetime_format)
         if "timeZone" in config:
             self.assertEqual(job.time_zone, config["timeZone"])
         else:
@@ -206,6 +212,7 @@ def test_ctor(self):
         self.assertIsNone(job.schema_update_options)
         self.assertIsNone(job.reference_file_schema_uri)
         self.assertIsNone(job.date_format)
+        self.assertIsNone(job.datetime_format)
         self.assertIsNone(job.time_zone)
 
     def test_ctor_w_config(self):
@@ -603,6 +610,7 @@ def test_begin_w_alternate_client(self):
             },
             "schemaUpdateOptions": [SchemaUpdateOption.ALLOW_FIELD_ADDITION],
             "dateFormat": self.DATE_FORMAT,
+            "datetimeFormat": self.DATETIME_FORMAT,
             "timeZone": self.TIME_ZONE,
         }
         RESOURCE["configuration"]["load"] = LOAD_CONFIGURATION
@@ -633,6 +641,7 @@ def test_begin_w_alternate_client(self):
         config.schema_update_options = [SchemaUpdateOption.ALLOW_FIELD_ADDITION]
         config.reference_file_schema_uri = "gs://path/to/reference"
         config.date_format = self.DATE_FORMAT
+        config.datetime_format = self.DATETIME_FORMAT
         config.time_zone = self.TIME_ZONE
 
         with mock.patch(

tests/unit/job/test_load_config.py

@@ -844,6 +844,22 @@ def test_date_format_setter(self):
         config.date_format = date_format
         self.assertEqual(config._properties["load"]["dateFormat"], date_format)
 
+    def test_datetime_format_missing(self):
+        config = self._get_target_class()()
+        self.assertIsNone(config.datetime_format)
+
+    def test_datetime_format_hit(self):
+        datetime_format = "%Y-%m-%dT%H:%M:%S"
+        config = self._get_target_class()()
+        config._properties["load"]["datetimeFormat"] = datetime_format
+        self.assertEqual(config.datetime_format, datetime_format)
+
+    def test_datetime_format_setter(self):
+        datetime_format = "YYYY/MM/DD HH24:MI:SS"
+        config = self._get_target_class()()
+        config.datetime_format = datetime_format
+        self.assertEqual(config._properties["load"]["datetimeFormat"], datetime_format)
+
     def test_time_zone_missing(self):
         config = self._get_target_class()()
         self.assertIsNone(config.time_zone)

tests/unit/test_external_config.py

@@ -26,6 +26,7 @@
 class TestExternalConfig(unittest.TestCase):
     SOURCE_URIS = ["gs://foo", "gs://bar"]
     DATE_FORMAT = "MM/DD/YYYY"
+    DATETIME_FORMAT = "MM/DD/YYYY HH24:MI:SS"
     TIME_ZONE = "America/Los_Angeles"
 
     BASE_RESOURCE = {
@@ -36,6 +37,7 @@ class TestExternalConfig(unittest.TestCase):
         "ignoreUnknownValues": False,
         "compression": "compression",
         "dateFormat": DATE_FORMAT,
+        "datetimeFormat": DATETIME_FORMAT,
         "timeZone": TIME_ZONE,
     }
 
@@ -84,6 +86,7 @@ def test_to_api_repr_base(self):
         ec.schema = [schema.SchemaField("full_name", "STRING", mode="REQUIRED")]
 
         ec.date_format = self.DATE_FORMAT
+        ec.datetime_format = self.DATETIME_FORMAT
         ec.time_zone = self.TIME_ZONE
         exp_schema = {
             "fields": [{"name": "full_name", "type": "STRING", "mode": "REQUIRED"}]
@@ -99,6 +102,7 @@ def test_to_api_repr_base(self):
             "connectionId": "path/to/connection",
             "schema": exp_schema,
             "dateFormat": self.DATE_FORMAT,
+            "datetimeFormat": self.DATETIME_FORMAT,
             "timeZone": self.TIME_ZONE,
         }
         self.assertEqual(got_resource, exp_resource)
@@ -136,6 +140,7 @@ def _verify_base(self, ec):
         self.assertEqual(ec.max_bad_records, 17)
         self.assertEqual(ec.source_uris, self.SOURCE_URIS)
         self.assertEqual(ec.date_format, self.DATE_FORMAT)
+        self.assertEqual(ec.datetime_format, self.DATETIME_FORMAT)
         self.assertEqual(ec.time_zone, self.TIME_ZONE)
 
     def test_to_api_repr_source_format(self):