File hash missing when iati-activities element lacks version attribute about ckanext-iati HOT 4 OPEN

@andreaszenasidi Could you please provide more context and the specific need, because could not be a bug that need a fix, instead, be a feature that's not in place anymore

from ckanext-iati.

Hi @adrianoamaral , thanks for taking a look at this.

There are several services, including the Unified Platform, that use the file hash from the Registry to uniquely identify files by content. The hashing on the Registry allows for our services to tell when the contents of a file have actually changed, without needing to re-download each and every file. In several cases, we've found the hash has not been updating. Sometimes it's due to a missing version number, as above, but other times we cannot identify why the Registry is not assigning a unique hash to the file (e.g. https://iatiregistry.org/dataset/ec-intpa-tz). Would it be possible to de-couple the creation and assigning of a file hash in the Registry from the process that checks the version-number/count of activities? If not, is there another method you could suggest for the Unified Platform to identify when the contents of a file changes without re-downloading the file and doing the hashing ourselves?

from ckanext-iati.

Here is the recent example: https://iatiregistry.org/api/3/action/package_show?id=pont-activities

from ckanext-iati.

@akmiller01 and I took a quick look at the code responsible for the version in the archiver:

If the version list returned by the xpath is length 0, this line log.info(version[0]) will drop an IndexError

from ckanext-iati.