データアセットを検索する | Data Catalog Documentation | Google CloudClass SearchCatalogRequest (3.23.0) | Python client library | Google Cloudサンプルで仕様書のAPIを使っているがqueryが空白刻みで入れる等の使い方が分かる
■BQスキーマ+ポリシータグ取得
from google.cloud import bigquery
def get_policy_tags_from_bq_table(project_id, dataset_id, table_id):
print("################ bigquery.Client.get_table().schema start ################")
print(f"Target table: {project_id}.{dataset_id}.{table_id}")
bq_client = bigquery.Client()
table = bq_client.get_table(f"{project_id}.{dataset_id}.{table_id}")
schema = table.schema
policy_tags = []
for field in schema:
print(f"Column: {field.name}")
if field.policy_tags:
tags = [tag for tag in field.policy_tags.names]
policy_tags.extend(tags)
print(f"Policy Tags: {tags}")
else:
print("> No Policy Tags assigned.")
return policy_tags
PROJECT_ID = "prj"
DATASET_ID = "ds"
TABLE_ID = "test001"
policy_tags = get_policy_tags_from_bq_table(PROJECT_ID, DATASET_ID, TABLE_ID)
print("Collected Policy Tags:", policy_tags)
■ポリシータグ設定
from google.cloud import datacatalog_v1
from google.cloud import bigquery
PROJECT_ID = "prj"
DATASET_ID = "ds"
TABLE_ID = "tbl01"
COLUMN_NAME = "aaa"
POLICY_TAG_PROJECT = "prj"
POLICY_TAG_NAME = "projects/prj/locations/us/taxonomies/83893110/policyTags/11089383"
def list_taxonomy_and_policy_tag():
print("############# Start #############")
list_policy_tags = []
client = datacatalog_v1.PolicyTagManagerClient()
request = datacatalog_v1.ListTaxonomiesRequest(
parent=f"projects/{POLICY_TAG_PROJECT}/locations/us"
)
try:
page_result = client.list_taxonomies(request=request)
except google.api_core.exceptions.PermissionDenied as e:
print(f"Skipping project {POLICY_TAG_PROJECT} due to PermissionDenied error: {e}")
return []
except Exception as e:
print(f"An error occurred for project {POLICY_TAG_PROJECT}: {e}")
return []
for taxonomy in page_result:
print(f"############ Taxonomy display_name: {taxonomy.display_name} #############")
print(f"############ Taxonomy name: {taxonomy.name} #############")
request_tag = datacatalog_v1.ListPolicyTagsRequest(parent=taxonomy.name)
try:
page_result_tag = client.list_policy_tags(request=request_tag)
except Exception as e:
print(f"Error on {request_tag}: {e}")
break
for policy_tag in page_result_tag:
print("Policy tag:")
print(policy_tag)
list_policy_tags.append({
"project_id": POLICY_TAG_PROJECT,
"taxonomy_display_name": taxonomy.display_name,
"taxonomy_name": taxonomy.name,
"policy_tag_name": policy_tag.name,
"policy_tag_display_name": policy_tag.display_name,
})
return list_policy_tags
def update_table_schema_with_policy_tag(list_policy_tags):
for policy_tag in list_policy_tags:
if policy_tag['policy_tag_name'] == POLICY_TAG_NAME:
print(
f"Target policy tag:\n"
f" Project ID: {policy_tag['project_id']}\n"
f" Taxonomy Display Name: {policy_tag['taxonomy_display_name']}\n"
f" Taxonomy Name: {policy_tag['taxonomy_name']}\n"
f" Policy Tag Name: {policy_tag['policy_tag_name']}\n"
f" Policy Tag Display Name: {policy_tag['policy_tag_display_name']}"
)
client = bigquery.Client()
table_ref = f"{PROJECT_ID}.{DATASET_ID}.{TABLE_ID}"
table = client.get_table(table_ref)
new_schema = []
for field in table.schema:
if field.name == COLUMN_NAME:
new_schema.append(
bigquery.SchemaField(
name=field.name,
field_type=field.field_type, # Keep original field type
mode=field.mode, # Keep original mode
description=field.description,
policy_tags=bigquery.PolicyTagList([POLICY_TAG_NAME]),
)
)
else:
new_schema.append(field)
table.schema = new_schema
updated_table = client.update_table(table, ["schema"])
print(
f"Updated table {updated_table.project}.{updated_table.dataset_id}.{updated_table.table_id} schema\n"
f"with policy_tag {POLICY_TAG_NAME} on the column {COLUMN_NAME} successfully."
)
if __name__ == "__main__":
list_policy_tags = list_taxonomy_and_policy_tag()
update_table_schema_with_policy_tag(list_policy_tags)
