EAD structural analysis

[mikesname]

We would like to be able to compare two EAD files structurally with the following (approximate) logic:

• Do they have common documentary units, e.g. is the archdesc/unitid the same? If yes, has the structure changed?
  • Has the total count of doc units changed?
  • Is the combination of unitid, index, and path the same? If yes, we can update the EAD in place. Otherwise:
  • Have only new units been added?
  • Have only items been deleted (and which ones?)
  • Has the hierarchy been altered in some other way? New intermediate levels added, removed? Items moved between intermediate levels?

[VladimirAlexiev]

cc @boyan-simeonov and @lindareijnhoudt

Hi Mike! Guess this is related to Synchronization. Why not delete all objects ingested from the EAD, and create them anew?

• Is it difficult to circumscribe (determine the boundary) that neo4j subgraph?
• We must ensure that any manual links of old nodes are preserved, assuming their EAD IDs are also preserved. I think this would be possible by remembering the links, and after the delete/insert, updating their targets to point to the new nodes, finding by EAD ID

We need to figure out how to do this, since it's an explicit requirement for USHMM reingest, see https://docs.google.com/document/d/1tsprKbISLteIO6sbKMNcnIz14CrKjAoP1mSkXW56dts/edit#heading=h.bpgb083jmll1: "preserve links between parallel descriptions manually added in the portal, and those ingested from USHMM"

About your original question, I'd study features of https://www.oxygenxml.com/xml_diff_and_merge.html and http://www.altova.com/diffdog/xml-diff.html, then look for an open source library that implements part of these features.