Intersection and difference

felixhao28 commented 7 years ago

Currently we have + for union operation to combine two or more collections together. This is very helpful in the way you can split your desired results into sub-sections and write different query paths for each. In addition to +, in certain use cases, more collection operators will prove their usefulness (borrowing from set theory).

I am proposing 2 new collection operators: intersection and difference. These operations require operands to have non duplicate elements. So for these operations, there will be two modes: by-ref and by-value. A and B are by-ref equal if they have the excact same path. A and B are by-value equal if they have the same primitive value or all fields of them are by-value equal.

An array can be a set as long as there are no two elements equal to each other. If otherwise, duplicates are removed before performing any of the set operations.

Intersection `^` (by-ref) and `.^` (by-value)

A ^ B is an intersection of A and A iff every element in A ^ B is contained by both A and B.

poems:
    - name: The Bronze Horseman
      author: Pushkin
    - name: Devils
      author: Pushkin
    - name: Oh, Laziness, Come...
      author: Pushkin
    - name: Gitanjali
      author: Tagore
    - name: Manasi
      author: Tagore
    - name: Sonar Tori
      author: Tagore
novels:
    - name: Gitanjali
      author: Tagore
    - name: Noukadubi
      author: Tagore
    - name: Les Misérables
      author: Victor Hugo

By-value intersection $.poems.author .^ $.novels.author
- $.poems.author => Pushkin, Pushkin, Pushkin, Tagore, Tagore, Tagore
- $.novels.author => Tagore, Tagore, Victor Hugo
- Remove dups in $.poems.author => Pushkin, Tagore
- $.poems.author .^ $.novels.author => Tagore
By-ref intersection $.poems.author ^ $.novels.author
- $.poems.author => <poems[0].author>, ..., <poems[5].author>
- $.novels.author => <novels[0].author>, ..., <novels[2].author>
- $.poems.author ^ $.novels.author => <None>
By-ref intersection $..*[@.author is 'Tagore'] ^ $.novels
- ($..*[@.author is 'Tagore'] => <poems[3]>, <poem[4]>, <poem[5]>, <novels[0]>, <novels[1]>
- $.novels => <novels[0]>, ..., <novels[5]>
- $..*[@.author is 'Tagore'] ^ $.novels => <novels[0]>, <novels[1]>
By-value intersection $.poems .^ $.novels => {name: Gitanjali, author: Tagore}

Difference/complement `-` and `.-`

A - B is a complement of B in A (or difference of A and B) iff every element in A - B is contained by A but not by B.

$.poems.author - $.novels.author => Pushkin, Pushkin, Pushkin, Tagore, Tagore, Tagore
$.poems.author .- $.novels.author => Pushkin
($..*[@.author is 'Tagore'] - $.novels).name => Gitanjali, Manasi, Sonar Tori

For performance concerns, A - B should only evaluate B within results of A.

adriank commented 6 years ago

I'm closing it because I didn't have time to think about it for a year. If you'd like to add this functionality to ObjectPath, create PR.

felixhao28 commented 6 years ago

I have created a completely new DSL to solve my problem and then moved on from the previous project. Now I don't have the incentive anymore.

adriank / ObjectPath