Export json with id and add 'from_json' func

[dgercho]

Changes:

1. Added the node identifier (not presented when exporting to json format).
2. Added the ability to load a json blob into a 'Tree' structure.
3. Update the constant json outputs as part of the 'test_shallow_paste'.

[dgercho]

@liamlundy @caesar0301
Hi, any updates regarding this PR?

[cmoliverio]

When is this PR going to be integrated? Been half a year and this addition is quite small, but IMO critical. to_json needs to exist with a from_json counterpart, even if 1.8.0 rolls into 2.0.0 IMO. There is even a duplicate PR (#105) , and another PR that wants to add the identifier to the to_json, #123, so people clearly want this. . . I understand those are relatively easy to implement but this seems like a slam dunk that hasn't been dunking. @liamlundy

[liamlundy]

In order for this to not be a breaking change, can you change the default to_dict behavior to not return nid? Make sure all tests pass and this should be good to merge

[cmoliverio]

Hey @liamlundy , I believe I have come up with a solution that is the best of both worlds. It adds the optional parameter with_identifier to the functon to_dict and to_json. This makes it backwards compatible. This makes it fairly trivial to implement the from_json function that @dgercho has made but with a few changes.

Please see #244

[dgercho]

@liamlundy Updated the PR as requested and added a new test for this functionality.

[dgercho]

@liamlundy Can you re-run the CI?

[cmoliverio]

Hey @dgercho you can re-run the CI/CD by force pushing.

Also, the reason 3.9 fails is because the workflow needs to be fixed. Here is the fix: https://github.com/caesar0301/treelib/pull/244/changes

You could add that change and then make a new commit and it should pass :)

[dgercho]

@liamlundy Done 🎉

[liamlundy]

@cmoliverio @dgercho this looks good. It looks like some code is duped between this and #244 but they also have unique patches. I'm happy to approve both but let me know how you want to approach the merge order.

[cmoliverio]

Respectfully, I believe the from_json() from this PR is inferior in one way: it uses the tag for both the tag and identifier when creating a new code.

This PR:

tree.create_node(tag=tag, identifier=tag, parent=parent_id, data=node_data)

#244:

tree.create_node(tag=tag, identifier=node_id, parent=parent_id, data=node_data)

I believe we need to consider the use case where a user needs to identify a node with two different names/methods.
This is why it becomes important to give the user the option to export to JSON with the identifier, which is also implemented in my PR. Otherwise we should just let the library handle the generation of a UUID (like normal). (I personally need this functionality in a project so I really want this)

@dgercho What are your thoughts on this slight improvement?

[dgercho]

Go ahead @cmoliverio

[cmoliverio]

Apologies, do you mean you are on board with liam merging my PR?

[liamlundy]

Let's do this: can you @cmoliverio merge this branch into yours? The main component from this PR that I want to include is the unit test

[cmoliverio]

dgercho's branch was merged. #244 is ready to go.

[liamlundy]

Thanks @dgercho for your work on this! I approved #244