Add more Fix functions.

[blackwinter]

We're going to need some more Fix functions that aren't implemented yet. E.g.:

Morph	Fix
replace	replace_all
split	split_field
unique	uniq

I'm just going to go ahead and implement those plus a bunch more from the Catmandu Cheat Sheet.

Functional review: @TobiasNx
Code review: @fsteeg

[TobiasNx]

In Catmandu replace_all does not seem to support regexp as in metafacture. That would be nice.

unique seem to process only arrays in Catmandu. Metafacture is here also more powerful.

split_field outputs arrays and differs from split-behavour in Metafacture.

Bonus: Unique/uniq objects not just strings would be awesome.

[blackwinter]

In Catmandu replace_all does not seem to support regexp as in metafacture. That would be nice.

What makes you think that? The cheat sheet says: replace_all(title,'[au]','X') -> title: cXtmXndX. And the tests confirm this behaviour.

unique seem to process only arrays in Catmandu. Metafacture is here also more powerful. [Bonus: Unique/uniq objects not just strings would be awesome.]

Well, yes, it seems that Metamorph's unique function is a little more powerful. But I don't think we've ever made use of that.

split_field outputs arrays and differs from split-behavour in Metafacture.

Hm, other than the absence of arrays in Metamorph, I don't really see a difference. It just emits a repeated field instead of an actual array object.

But thanks for your input! Let's wait and see what our hybrid "better than both worlds" beast will look like in the end ;)

[blackwinter]

@TobiasNx: Sorry, forgot to assign you for functional review.

[TobiasNx]

This will take some time to review since this are a lot of functions

[blackwinter]

That's understandable ;) Thanks for your diligence!

In effect, this means that #103 and #105 have passed code review, but won't be merged until both functional reviews are done - unless we decide otherwise.

[fsteeg]

In effect, this means that #103 and #105 have passed code review, but won't be merged until both functional reviews are done - unless we decide otherwise.

+1 for merging, we can always open issues/PRs for things that come up in reviews (cc @dr0i for #14).

[blackwinter]

Fine with me, but we should probably not make a habit of it :)

[blackwinter]

If nobody objects I'll merge #103 and #105 tomorrow.

[TobiasNx]

rename i am not sure anymore if with rename: the sublevels should be change too as default or if that should be an option or done by the each-bind.

[blackwinter]

random still seems to have some troubles with arrays

These are expected (see #110; requires set_array(...)).

[TobiasNx]

random still seems to have some troubles with arrays

These are expected (see #110; requires set_array(...)).

you are right, i didn't think about that.

[blackwinter]

sort_field: All options do not work yet.

See comment above (use "true").

sort_field: Selecting arrays with * does not work

I'll have to investigate. (see next comment)

[blackwinter]

In general, for inserting/transforming values with wildcards, see #102.

[blackwinter]

reverse("animals[]"): These seem not to work perhapse due to (JSON-)Array. Either way does not work.

Result shows ["zebra", "cat", "dog"], which is the reverse of ["dog", "cat", "zebra"]. See also my previous comment.

[blackwinter]

rename i am not sure anymore if with rename: the sublevels should be change too as default or if that should be an option or done by the each-bind.

Catmandu does it by default (can't be controlled by an option).

[blackwinter]

split_field: This results in empty array:

Without move_field(...) I get the indexed hash:

"OTHERS" : [ {
  "tools" : {
    "1" : "hammer",
    "2" : "saw",
    "3" : "bow"
  }
} ]

So I would ascribe this to our array handling, not the split_field function itself.

Can you confirm?

[TobiasNx]

Yes that is true and I can confirm that.

I have set up a separate test for moving arrays in object-arrays:

https://github.com/TobiasNx/fix-FunctionalReview-Testing/tree/master/data/testing/move_field_in_object_array

The problem seems to be something if you have an array in an object of an array and if you move this.

[TobiasNx]

All in all, as discussed off board the function seem to work. All problems that are open are not related to the functions themselves:

1. Order of elements are changed.
2. Problems with index-wildcard in paths.

Therefore:
+1