Exclude description sections of automations.yaml (entities noted in comment sections)

[bcutter]

Problem: Currently Watchman also finds and lists entities which are used in the comment sections of automations. This triggers false positives - and makes (based on the intensity of individual documentation/commenting) using this basically absolutely great integration a bit useless. 😞

Feature Request: Ignore all entites from the description: sections of automations.yaml

Example:

+--------------------------------------------------------------+-----------------+--------------------------------------------------------------+
| Entity ID                                                    | State           | Location                                                     |
+--------------------------------------------------------------+-----------------+--------------------------------------------------------------+
| group.anwesenheit_zuhause                                    | missing         | automations.yaml:1575,3896,4151,8725                         |

Actual automations.yaml snippet:

- id: 'XXXXXXXXXXXXXX'
  alias: Automation
  description: '...
    ...
    - Von "group.anwesenheit_zuhause" (state: home|not_home) auf "zone.home" (numeric_state:
    0|!0) umgestellt
    ...'

This should remove false positives massively. Looking at some (former) issues and reports, plenty of those findings for others too.

Manually setting those entities on the ignore list is only workaround - if at all.

[bcutter]

Plus (more improvement to reduce false positives):

• ignore everything with example: domain.entity_id from /config/custom_components (example: as keyword) (again: https://github.com/dummylabs/thewatchman#ignored-files-option-example is just a workaround for this)

[Hiekkaharju]

• ignore everything with example: domain.entity_id from /config/custom_components (example: as keyword) (again: https://github.com/dummylabs/thewatchman#ignored-files-option-example is just a workaround for this)

This part is pretty much same as #56

[bcutter]

You're right. So ignore that additional post for now and let's focus on the original one.

At the same time both share the same target - getting rid of unnecessary false positives.

[bcutter]

Meanwhile I discovered that the integration alreay ignores certain entities, usually when there's a # close or directly before the entity, e. g. this is a text and this is a #sensor.entity.

Unfortunately, that's not enough for the basic issue, as comment sections spread over a certain amount of text. But maybe this "comment section detection" can be extended to also address and solve this issue.

[DataBitz]

This also affects automation blueprints which are not deployed.

input:
power_sensor_1_entity:
name: Power Sensor 1
description: This is the main power sensor that will be used to measure load (e.g. sensor.main_meter_power)
selector:
entity:
filter:
- domain: sensor
device_class: power

[dummylabs]

In version 0.6.4 single-line description: and example: fields will be excluded from parsing. As far as I can see, multiline descriptions are not prohibited in YAML configuration, but they are not encouraged by the Home Assistant UI. For example, in the automations editor, they are collapsed into a single line when displayed or updated via the UI.

Supporting multiline descriptions in parsing requires significant changes, we might address this later.

[bcutter]

I really like the recent work you are putting in this integration.

Looking forward to see the improvement running 0.6.4. Any estimations when it will be released?

[dummylabs]

Thanks for the kind words! It's almost ready, just polishing minor details and running some real-world tests before publishing. There were a lot of changes and refactoring under the hood, so I expect there may be some bugs not covered by the unit tests. Hopefully, this will make future development easier.