Links diff: links not shown as changed

[Mr0grog]

In this diff: https://monitoring.envirodatagov.org/page/1de9a11d-330b-4a87-9926-6c6357b6f668/36442e96-71da-4a54-899b-e3c193e5d5fd..5ca0de1e-c4c9-472b-8f66-83cbf18069c8

The two links near the top appear to have been removed and added because their text changed enough that a different link moved from before them in the list to after them.

Ideally, we should find a way for these links to correctly show that their text was changed, rather than that the links were added or removed.

I think this probably requires some deep re-thinking of how we diff links. Right now, we create a list, then we diff the list on “rough similarity,” then diff the internals of similar links (these are the links that show up as changed, rather than added or removed). The problem here is diffing a list, where order matters. In reality, the links diff is about diffing a set.

We should probably start by grouping roughly similar links from the old and new list together, tagging each link with sorting properties that come exclusively from either the new version of the link (so both lists sort the same). Then do the diffing.

[Mr0grog]

Here’s another example. Simpler (only one change) and much more obvious (because the text changed so much): https://monitoring.envirodatagov.org/page/f2e7706b-59ab-4261-b9e3-d50eef8562d3/c4fbc0d7-7b86-4e62-82b4-07d37433ded8..db1a8fb6-7a76-4a81-b962-c4f08c4807b7

[Mr0grog]

Additional thought: when grouping links, we should probably look at text + some canonicalized form of the URL (e.g. SURT) instead of text + URL.