Alternative for matchyaml() for analysing an entire file?

[Rylon]

Hi,

In the older version of Ansible Lint, I was using matchyaml() to look for usage comment somewhere in the file, something we found useful when introducing Ansible to our team, so people could always see right away how a playbook was intended to be run. I was using this:

def matchyaml(self, file, text):
        text_lines = text.split('\n')

        # If this isn't a playbook, skip.
        if not file['type'] == 'playbook':
            return []

        for line in text_lines:
            if re.match("^# Usage: ansible-playbook.*$", line ):
                return []

        return [Match(len(text_lines), ''.join("Usage comment not found in the playbook."), file['path'], self, self.shortdesc)]

Since the signature of matchyaml() has changed, and the introduction of Yamllint, I'm not sure how to accomplish this anymore.

Is it possible to receive the entire text for a particular file, to look for a specific line somewhere in the file?

Thanks!

[ssbarnea]

I am sorry that there is no clear documentation on the AnsibleLintRule methods, they did suffer a lot of refactoring and there will still be a little bit more to do. Your best friend is searching for current usage in internal rules.

Basically you can still use the same method now, the difference is that Lintable object has few attributes as path, kind and content, make use of them. Look for class PlaybookExtension and you will know how to rewrite your custom rule.

Keep in mind that regardless the name of the method, it will be called for any file kind, regardless if is yaml or not.

My desire was to rename it but there were already too many breaking changes in v5, so it was postponed.

I would also like to mention that I am very open to receive new rule contributions to the linter itself. Your rule can easily be modified to be generic purpose and opt-in, by adding a config option like playbook_content_require_regex: .... If user does not add option to the lint config, the rule will not run, if a value is defined, the rule will require the text to be present.

[Rylon]

Hi, yeah general purpose sounds good, it's effectively looking for a particular string somewhere in the playbook that can be defined with a regex. I put the upgrade on hold because too many things broke and we're stretched a bit thin right now, but I'll try your suggestion again in the future.

[Rylon]

An update for anyone else who had this issue, thanks to @ssbarnea pointing me in the right direction I was able to get our custom rule working again. Here's a simpler version of it to demonstrate the concept:

import re
from typing import List

from ansiblelint.errors import MatchError
from ansiblelint.file_utils import Lintable
from ansiblelint.rules import AnsibleLintRule


class MatchStringInLineRule(AnsibleLintRule):
    id = '<id goes here>'
    shortdesc = '<short description goes here>'
    description = '<description goes here>'
    severity = 'LOW'
    tags = ['formatting']

    done: List[str] = []

    def matchyaml(self, file: Lintable) -> List[MatchError]:
        # No errors initially.
        result: List[MatchError] = []

        # Return the empty result set (no errors) if this is not a playbook.
        if file.kind != 'playbook':
            return result

        text_lines = file._content.split('\n')

        for line_num, line in enumerate(text_lines):
            if re.match("<regex goes here>", line):
                # We have a match! Return the empty result set (no errors).
                return result
                
        # If we make it here, we didn't find any matches, so add a matcherror to the results set and return.
        error_message = 'Required text not found anywhere in the playbook.'
        result.append(self.create_matcherror(linenumber=len(text_lines), filename=file_path, details=error_message))

        self.done.append(file_path)
        return result