"Sync" Synapse fileview model with schematic model

[anngvu]

Originating from nf-osi/nfportalutils#35 (comment). Basically, sometimes the initial fileview needs to be updated according to changes in the source data model.

However, the schematic model currently does not contain enough info to specify a fileview. For example, if a string column, what is the size allowed? (Though the job should be able to calculate string sizes based if based on enums). If a list column, what is the list length limit?

Since our (nf.jsonld) schema is dynamic, I wonder if we could have a, say, weekly job that does the following:

• check for changes to jsonld schema since last run, if no changes, stop
• if changes, iterate over each project view schema:
• for each view, first run syn._get_annotation_view_columns() to get all annotations on files within scope of view
• then, for all annotation keys that are defined parents in the jsonld schema:
• Set as STRINGLIST if lists are permitted in jsonld schema.
• Set as other column types as necessary (maybe we need to define these in the jsonld?)
• modify the maximumSize to the max length valid value if a string/stringlist.
• set as max length, say, 50, if string type but valid values are not restricted.
• for all columns not in the jsonld schema, include them in the view unchanged (ie use the schema that syn._get_annotation_view_columns() returns)
• if total number columns >152, skip (and send log via email?)
• repeat on all NF Portal study views