cc @mattklein123 @htuch

this is a first attempt at a script to cleanup text/code files in the repo that have language-independent linting issues

so far it does not:

• allow exceptions
• automate a fix
• fix the files

just looking at autofix - to automate untabify it will require hard-coding what the indentation should be for any given file type

re untabify - this situation is made worse as .sh scripts currently use a mix of indents sometimes within the same file

my vote is defo to use 4-space indents but im struggling to find anything (reliable) that will properly lint indents in shell scripts (no matter what the number of spaces is)

@phlax I'd probably prefer Python here TBH. Bash becomes hard to write if we want to extend it.

ill have a look at reimplementing in python.

my concern is that it will make it much slower and probs require quite a bit more code. Also some of the things (like finding the text files) are going to be much harder to implement in python (bash still has some uses 8/) so we may end up with the ~same bash lines just wrapped in python

I think pathlib globbing is pretty good for this. But, I agree on performance. I guess my question is how complex do we anticipate this growing? The current level of complexity is fine and somewhat maintainable. I would hate to see something that reached 2x the size though..

I think pathlib globbing is pretty good for this.

its not the globbing thats the issue - its easy enough to find and filter files - this is using git to filter what it sees as a "text" file

CC @envoyproxy/api-shepherds: Your approval is needed for changes made to (api/envoy[\w/]*/(v1alpha\d?|v1|v2alpha\d?|v2))|(api/envoy/type/(matcher/)?\w+.proto).
CC @envoyproxy/api-shepherds: Your approval is needed for changes made to api/envoy/.
API shepherd assignee is @mattklein123
CC @envoyproxy/api-watchers: FYI only for changes made to api/envoy/.

+1 to write it with python because it would help with having cross platform tooling

apologies - the conversation continued in slack - and for the reason that bash really is the best tool for these it has been done this way for now

using python would add a huge amount of complexity and depending on how much complexity was added would potentially be a lot slower

also at very least (without a lot of working around) it would need to include a github dependency - so it would require a lot more of users (ie python + git package)

@mattklein123 this should be ready for final review hopefully

i have excluded all of the *_corpus/ folders so hopefully this doesnt mess with any test data that it shouldnt

current exclude is

 NOLINT_RE="\.patch$|^test/.*_corpus/|^tools/.*_corpus/|password_protected_password.txt"

/retest

Retrying Azure Pipelines:
Retried failed jobs in: envoy-presubmit