def txfr_blob(filename: str,
bq: BigQueryer = PostBigQueryer(),
cs: CloudStorager = CloudStorager()):
"""
uploads the blob to bigquery. This would probably be better as a shell script
:param cs:
:param bq:
:param bucket:
:param filename:
:return:
"""
tm = current_time_ms(
) # pain in the ass to get nanotime in python apparently
objname = 'api-update-blob-{}'.format(tm)
blob = Blob(objname, cs.get_cloud_storage_bucket())
logger.info("Uploading file (this will take a long time)... ")
blob.upload_from_filename(filename)
# change this to change table
table = bq.get_bigquery_table()
uri = 'gs://' + cs.bucket + "/" + objname
logger.info("Loading file to BQ...")
# insert into tmp table
# tmptable = bq.client.dataset('forums').table(objname)
job = LoadTableFromStorageJob('api-job-{}'.format(tm),
table, [uri],
client=bq.client)
job.write_disposition = 'WRITE_APPEND'
job.source_format = 'AVRO'
job.begin()
wait_for_job(job)
logger.info("Cleaning up...")
blob.delete(cs.client)
def load_table_from_storage(self, job_name, destination, *source_uris):
"""Construct a job for loading data into a table from CloudStorage.
See:
https://cloud.google.com/bigquery/docs/reference/v2/jobs#configuration.load
:type job_name: str
:param job_name: Name of the job.
:type destination: :class:`google.cloud.bigquery.table.Table`
:param destination: Table into which data is to be loaded.
:type source_uris: sequence of string
:param source_uris: URIs of data files to be loaded; in format
``gs://<bucket_name>/<object_name_or_glob>``.
:rtype: :class:`google.cloud.bigquery.job.LoadTableFromStorageJob`
:returns: a new ``LoadTableFromStorageJob`` instance
"""
return LoadTableFromStorageJob(job_name, destination, source_uris,
client=self)
def job_from_resource(self, resource):
"""Detect correct job type from resource and instantiate.
:type resource: dict
:param resource: one job resource from API response
:rtype: One of:
:class:`google.cloud.bigquery.job.LoadTableFromStorageJob`,
:class:`google.cloud.bigquery.job.CopyJob`,
:class:`google.cloud.bigquery.job.ExtractTableToStorageJob`,
:class:`google.cloud.bigquery.job.QueryJob`,
:class:`google.cloud.bigquery.job.RunSyncQueryJob`
:returns: the job instance, constructed via the resource
"""
config = resource['configuration']
if 'load' in config:
return LoadTableFromStorageJob.from_api_repr(resource, self)
elif 'copy' in config:
return CopyJob.from_api_repr(resource, self)
elif 'extract' in config:
return ExtractTableToStorageJob.from_api_repr(resource, self)
elif 'query' in config:
return QueryJob.from_api_repr(resource, self)
raise ValueError('Cannot parse job resource')
def job_from_resource(self, resource):
"""Detect correct job type from resource and instantiate.
:type resource: dict
:param resource: one job resource from API response
:rtype: One of:
:class:`google.cloud.bigquery.job.LoadTableFromStorageJob`,
:class:`google.cloud.bigquery.job.CopyJob`,
:class:`google.cloud.bigquery.job.ExtractTableToStorageJob`,
:class:`google.cloud.bigquery.job.QueryJob`,
:class:`google.cloud.bigquery.job.RunSyncQueryJob`
:returns: the job instance, constructed via the resource
"""
config = resource['configuration']
if 'load' in config:
return LoadTableFromStorageJob.from_api_repr(resource, self)
elif 'copy' in config:
return CopyJob.from_api_repr(resource, self)
elif 'extract' in config:
return ExtractTableToStorageJob.from_api_repr(resource, self)
elif 'query' in config:
return QueryJob.from_api_repr(resource, self)
raise ValueError('Cannot parse job resource')
