OutputStorageConfig(mapping=None, *, ignore_unknown_fields=False, **kwargs)
Cloud repository for storing output.
.. _oneof: https://proto-plus-python.readthedocs.io/en/stable/fields.html#oneofs-mutually-exclusive-fields
Attributes | |
---|---|
Name | Description |
table |
google.cloud.dlp_v2.types.BigQueryTable
Store findings in an existing table or a new table in an existing dataset. If table_id is not set a new one will be generated for you with the following format: dlp_googleapis_yyyy_mm_dd_[dlp_job_id]. Pacific time zone will be used for generating the date details. For Inspect, each column in an existing output table must have the same name, type, and mode of a field in the Finding object.
For Risk, an existing output table should be the output of a
previous Risk analysis job run on the same source table,
with the same privacy metric and quasi-identifiers. Risk
jobs that analyze the same table but compute a different
privacy metric, or use different sets of quasi-identifiers,
cannot store their results in the same table.
This field is a member of oneof _ type .
|
output_schema |
google.cloud.dlp_v2.types.OutputStorageConfig.OutputSchema
Schema used for writing the findings for Inspect jobs. This field is only used for Inspect and must be unspecified for Risk jobs. Columns are derived from the Finding object.
If appending to an existing table, any columns from the
predefined schema that are missing will be added. No columns
in the existing table will be deleted.
If unspecified, then all available columns will be used for
a new table or an (existing) table with no schema, and no
changes will be made to an existing table that has a schema.
Only for use with external storage.
|
Classes
OutputSchema
OutputSchema(value)
Predefined schemas for storing findings. Only for use with external storage.
Values:
OUTPUT_SCHEMA_UNSPECIFIED (0):
Unused.
BASIC_COLUMNS (1):
Basic schema including only info_type
, quote
,
certainty
, and timestamp
.
GCS_COLUMNS (2):
Schema tailored to findings from scanning
Cloud Storage.
DATASTORE_COLUMNS (3):
Schema tailored to findings from scanning
Google Datastore.
BIG_QUERY_COLUMNS (4):
Schema tailored to findings from scanning
Google BigQuery.
ALL_COLUMNS (5):
Schema containing all columns.