Expected Behavior

The CSVs which can be downloaded from search/browse views should should appear the same as what appears in all pages the search results view. This means that all columns and rows should match.

Current Behavior

Currently csvs are generated directly from what solr returns. This means that any column that does not directly correspond to data that appears in the index will not show up in the csv. Having direct correspondence between what data exists in the index and what data shows up to the user is not possible since what the user should see is expected to be different based on language settings or permissions.

On a related note, the search filters are also not applied to CSV's as they should be (noted in ANW-1056.

Possible Solution

Possibly csv's could be generated in a way similar to how report csvs are generated. At the very least, it should be possible to modify and enrich data that comes from solr before the csv is generated. Ideally, we would reuse as much of the code for rendering search/browse views as possible, perhaps by rendering the html and then using some sort of html to plain text converter.

Steps to Reproduce (for bugs)

For example, the found in column which appears in several search/browse views will not appear in the corresponding CSV. The contents of this column can not simply be added to the index since non-admin users should not be able to see suppressed parent records while admin users should be able to see these.

Context

In my work on browse search preferences, I repeatedly hit roadblocks in adding new functionality due to the inflexible way csvs are generated.

Your Environment

Not environment specific.