[AIRFLOW-3905] Allow using "parameters" in SqlSensor

[XD-DENG]

Jira

• https://issues.apache.org/jira/browse/AIRFLOW-3905

Description

In most SQL-related operators/sensors, argument parameters is available to help render SQL command conveniently, like MySqlOperator (https://github.com/apache/airflow/blob/master/airflow/operators/mysql_operator.py#L25) and PostgresOperator (https://github.com/apache/airflow/blob/master/airflow/operators/postgres_operator.py#L24).

But this is not available in SqlSensor yet.

Tests

• My PR adds the following unit tests OR does not need testing for this extremely good reason:

Tests passed.

Commits

• My commits all reference Jira issues in their subject lines, and I have squashed multiple commits if they address the same issue. In addition, my commits follow the guidelines from "How to write a good git commit message":
  1. Subject is separated from body by a blank line
  2. Subject is limited to 50 characters (not including Jira issue reference)
  3. Subject does not end with a period
  4. Subject uses the imperative mood ("add", not "adding")
  5. Body wraps at 72 characters
  6. Body explains "what" and "why", not "how"

Documentation

• In case of new functionality, my PR adds documentation that describes how to use it.
  • When adding new operators/hooks/sensors, the autoclass documentation generation needs to be added.
  • All the public functions and the classes in the PR contain docstrings that explain what it does

Code Quality

• Passes flake8

[codecov-io]

Codecov Report

Merging #4723 into master will increase coverage by <.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master    #4723      +/-   ##
==========================================
+ Coverage   74.65%   74.65%   +<.01%     
==========================================
  Files         430      430              
  Lines       27990    27992       +2     
==========================================
+ Hits        20896    20898       +2     
  Misses       7094     7094

Impacted Files	Coverage Δ
airflow/sensors/sql_sensor.py	100% <100%> (ø)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update b71a4b6...c63f008. Read the comment docs.

[zhongjiajie]

Please find my two cents.

[XD-DENG]

@feng-tao @ashb PTAL. Thanks.

[feng-tao]

Do all the sql type hooks(druid, mysql, postgres etc) are all inherited from dbapi hook? If that's case, the change LGTM

[feng-tao]

If not, we should change that hook to inherit from dbapi hook.

[XD-DENG]

Yes, hooks of druid, mysql, mssql, Postgres, oracle etc are all inherited from dbapi hook.

reference:

• https://github.com/apache/airflow/blob/master/airflow/hooks/mssql_hook.py#L25
• https://github.com/apache/airflow/blob/master/airflow/hooks/mysql_hook.py#L28
• https://github.com/apache/airflow/blob/master/airflow/hooks/oracle_hook.py#L29
• https://github.com/apache/airflow/blob/master/airflow/hooks/postgres_hook.py#L28

For SqlSensor, the get_hook() in BaseHook.get_connection(self.conn_id).get_hook() will decide which exact hook to use based on the connection type (reference: https://github.com/apache/airflow/blob/master/airflow/models/connection.py#L184).

[feng-tao]

But I checkCloudSqlDatabaseHook (

airflow/airflow/contrib/hooks/gcp_sql_hook.py

Line 669 in 6b38649

class CloudSqlDatabaseHook(BaseHook):

) which doesn't seem to inherit from dbapi hook.

But the I assume the sqlSensor could use CloudSqlDatabasehook? Given there are many connection hooks defined, a safer approach would be check if the hook is an instance of dbapi hook, use the records = hook.get_records(self.sql, self.parameters), otherwise fall back to records = hook.get_records(self.sql). what you think?

[XD-DENG]

Actually SqlSensor can not support CloudSqlDatabaseHook.

Eventually SqlSensor uses get_records method to retrieve the records from DB, while get_records method is not available in CloudSqlDatabaseHook at all.

[feng-tao]

I see, but would a check on hook type be safer?

[XD-DENG]

Get what you mean.

Given the limitation in https://github.com/apache/airflow/blob/master/airflow/models/connection.py#L184 and the implementation of each hook, only the connection types below are supported by SqlSensor:

• 'google_cloud_platform'
• 'jdbc'
• 'mssql'
• 'mysql'
• 'oracle'
• 'postgres'
• 'presto'
• 'sqlite'
• 'vertica'

I will add a check.

[feng-tao]

LGTM

[XD-DENG]

Thanks @feng-tao

[pingzh]

@XD-DENG @feng-tao it breaks the MetastorePartitionSensor

Traceback (most recent call last):
  File "/usr/local/lib/python2.7/dist-packages/airflow/models/taskinstance.py", line 913, in _run_raw_task
    result = task_copy.execute(context=context)
  File "/usr/local/lib/python2.7/dist-packages/airflow/sensors/base_sensor_operator.py", line 107, in execute
    while not self.poke(context):
  File "/usr/local/lib/python2.7/dist-packages/airflow/sensors/metastore_partition_sensor.py", line 83, in poke
    return super(MetastorePartitionSensor, self).poke(context)
  File "/usr/local/lib/python2.7/dist-packages/airflow/sensors/sql_sensor.py", line 82, in poke
    self.log.info('Poking: %s (with parameters %s)', self.sql, self.parameters)
AttributeError: 'MetastorePartitionSensor' object has no attribute 'parameters

As thecontructor of the sqlSensor is not called in the MetastorePartitionSensor

[malthe]

Shouldn't the parameters be templated?

Or perhaps templated_parameters due to the native vs string templating issue.