Feature list for next-gen EXFOR-JSON code

[brown170]

Requirements for EXFOR-JSON

• Respect the data model of original EXFOR file
• POINTERS
  + Unravel or not?
• DATA & COMMON
  + use JSON arrangement that plays nice with PANDAS
  + Unwrap wrap-around columns and column headers
  + Units translated into SI names ("EV" -> "eV")
  + Scheme for angles and angle units vs. scattering cosine
• metadata:
  + Version number of format
  + Version number of library
  + original EXFOR ENTRY/SUBENT designator
• BIB:
  + Standardized bibliographic schema with NSR, CINDA, EXFOR, ENDF, DataCite
  + URL/DOI of original file
  + DOI's of referenced articles
  + Unravel EXFOR tautologies in various fields
• REACTION:
  + Tree-like structure for math in EXFOR REACTION string
  + The core part (a "process" in x4i) of a REACTION string gets broken down into target, projectile, residuals
  + Possibly salvage coding from x4i
• EXFOR Dicts
  + If expanding, keep original coding
• Bonus items:
  + ENDF MF, MT (if known or computable)
  + Nuclear level (if known or computable) of target, project, and/or residual or other as applicable
  + Round trip translation

[brown170]

What about YAML instead of JSON for the main hierarchy and CSV for the data tables ?

[julia-sprenger]

Yaml has some advantages indeed, in particular some users might consider it more readable, comments are supported, more data types exist, but I would still argue for json as better format, as it has a more mature schema support, which I think should be a key feature of a new format.

In general the mixed approach (hierarchical format for metadata, tabular format for data) is a common approach and sounds like a good fit also for the exfor data here.