add Risk samples #1411
add Risk samples #1411
Conversation
dlp/risk.py
Outdated
| # See the License for the specific language governing permissions and | ||
| # limitations under the License. | ||
|
|
||
| """Sample app that uses the Data Loss Prevent API to """ |
There was a problem hiding this comment.
Looks like this is unfinished. Maybe "to perform risk anaylsis"?
dlp/risk.py
Outdated
| def numerical_risk_analysis(project, table_project_id, dataset_id, table_id, | ||
| column_name, topic_id, subscription_id, | ||
| timeout=180): | ||
| """Uses the Data Loss Prevention API to computes risk metrics of a column |
dlp/risk.py
Outdated
| print('Value Range: [{}, {}]'.format( | ||
| results.min_value.integer_value, | ||
| results.max_value.integer_value)) | ||
| print('25% quantile value: {}\n' # NOTE: change this to unique |
There was a problem hiding this comment.
This is a note to ask you about switching this to be more like the NodeJS samples. The NodeJS samples only print the quantiles with unique values. Here I am summarizing by providing the quartiles.
There was a problem hiding this comment.
I don't have a strong opinion on the output, since I don't really understand the functionality of the endpoint very well. = ) It should be fine as it is.
dlp/risk.py
Outdated
| .categorical_stats_result | ||
| .value_frequency_histogram_buckets) | ||
| # Print bucket stats | ||
| for i in range(len(histogram_buckets)): |
There was a problem hiding this comment.
Generally in python you would do "for bucket in histogram_buckets" instead of indexing them by number. Likewise in the examples below.
dlp/risk.py
Outdated
|
|
||
| def k_map_estimate_analysis(project, table_project_id, dataset_id, table_id, | ||
| topic_id, subscription_id, quasi_ids, info_types, | ||
| region_code='USA', timeout=180): |
There was a problem hiding this comment.
The default for region_code also needs to be added as "default=" to the corresponding command-line flag. Otherwise, the flag's default value of None will override this default when it's actually called.
Alternatively, if region_code can be safely omitted, maybe this should default to None in the first place?
| GCLOUD_PROJECT = os.getenv('GCLOUD_PROJECT') | ||
| TOPIC_ID = 'dlp-test' | ||
| SUBSCRIPTION_ID = 'dlp-test-subscription' | ||
| DATASET_ID = 'integration_tests_dlp' |
There was a problem hiding this comment.
Are these datasets public? (So, will they work for our test runner as well?)
No description provided.