Improve parsing of the ADF15 metadata for H-like ions. #425
Conversation
| if not config: | ||
| raise RuntimeError("Unable to parse ADF15 metadata.") | ||
|
|
There was a problem hiding this comment.
I like this. More obvious failure modes are good.
cherab/openadas/parse/adf15.py
Outdated
| if not config: # try hydrogen header (works for 'bnd' files) | ||
| config = _scrape_metadata_hydrogen(file, element, charge) |
There was a problem hiding this comment.
Is there a risk of this inadvertently parsing a different header file incorrectly if it fails on the hydrogen-like parser? I think it would be better to explicitly handle the 'bnd' files here as a special case (with a comment about why it's a special case), and for anything else let it fall through and error until we can manually verify that the same workaround can be applied.
There was a problem hiding this comment.
I have not found any files in ADAS other than those with the "bnd" suffix affected by this issue. Now I did an explicit check on the file name, but since ADAS may update the "bnd" files in the future, the parser tries to read the metadata in "hydrogen-like" format first, and if that fails, then for "bnd" files, it reads the metadata in "hydrogen" format.
This PR addresses the issue raised in #424. Some ADAS ADF15 (PEC) data files for hydrogen-like ions actually have "hydrogen" and not "hydrogen-like" format of the metadata.
While the user can specify the header format in
install_adf15()andparse_adf15(), theparse_adf15()can also try to parse the metadata in "hydrogen" format, if parsing in "hydrogen-like" format fails and theheader_formatis not specified by the user.Also, if parsing of the ADF15 metadata fails,
parse_adf15()does not raise an error andinstall_adf15()runs silently without updating the repository. This behaviour is confusing to the user and should be fixed, which is done in this PR.