ZLLentz
commented
2 years ago Travis is failing on the pre-commit isort style check
isort....................................................................Failed
- hook id: isort
- files were modified by this hook
Fixing /home/travis/build/pcdshub/happi/happi/utils.py
=====================================================================
WARNING: One or more pre-commit checks have failed!
This means you do not have pre-commit set up in your local checkout!
See https://github.com/pcdshub/pre-commit-hooks/blob/master/README.md
or use the above output and the following diff to fix the issues
manually:
=====================================================================
diff --git a/happi/utils.py b/happi/utils.py
index 6b491c3..8b0363b 100644
--- a/happi/utils.py
+++ b/happi/utils.py
@@ -4,8 +4,8 @@ Basic module utilities
import functools
import keyword
import logging
-import warnings
import os
+import warnings
from typing import Callable
from .errors import EnforceError
~/build/pcdshub/happi
The command "bash $CI_HELPER_PATH/travis/pre_commit.sh" exited with 1.
JJL772
Paths to the json db can now be relative to the happi config, unless they're already absolute.
Resolves #188
Description
This allows paths to the json db to be relative to the happi config.
For the
device_configrepo, itshappi.cfgcan look like this, instead of containing an absolute path:Other relative paths will work fine too, ie
../device_config/db.jsonm./db.json. Absolute paths are not transformed at all by the json backend when looking for the db, so/reg/g/pcds/happi.cfgwill still mean/reg/g/pcds/happi.cfg.The solution I came up with for this isn't as clean as I'd like. Since each backend accepts arbitrary args from the happi config, there's no good way to process paths before sending them to the backend. Ultimately, it's up to the backend implementation to process any paths such that they're correct. Luckily, the only backend that relies on paths from the config is the json backend.
Motivation and Context
Our happi configs all have absolute paths in them, which makes them annoying to use outside of our LCLS environment. For example, I needed to modify the happi.cfg in our
device_configrepo so I could play around with the LCLS device db on my local machine.How Has This Been Tested?
One test was modified to account for the new handling, and another test was added to specifically test configs with absolute paths for the json db.
Where Has This Been Documented?
Relevant comments have been modified or created in the code changed by this PR. I didn't see anything in the happi documentation that would need to be changed.