Custom equality in Clojure distinct - clojure

Question

In a Clojure program I've an array composed by maps containing peoples' names and emails.
e.g.
[
{ :name "John" :email "john#gmail.com" }
{ :name "Batman" :email "batman#gmail.com" }
{ :name "John Doe" :email "john#gmail.com" }
]
I'd like to remove the duplicate entries considering, for comparison purposes, pairs having the same e-mail to be equals. In the example above the output would be:
[
{ :name "John" :email "john#gmail.com" }
{ :name "Batman" :email "batman#gmail.com" }
]
What's the best way to achieve this in Clojure? Is there a way to let distinct knows what equals function to use?
Thanks.

Answer 1

yet another way to do it, kinda more idiomatic, i guess:
(let [items [{ :name "John" :email "john#gmail.com" }
{ :name "Batman" :email "batman#gmail.com" }
{ :name "John Doe" :email "john#gmail.com" }]]
(map first (vals (group-by :email items))))
output:
({:name "John", :email "john#gmail.com"}
{:name "Batman", :email "batman#gmail.com"})
that is how it works:
(group-by :email items) makes a map, whose keys are emails, and values are groups of records with this email
{"john#gmail.com" [{:name "John", :email "john#gmail.com"}
{:name "John Doe", :email "john#gmail.com"}],
"batman#gmail.com" [{:name "Batman", :email "batman#gmail.com"}]}
then you just need to take its vals (groups of records) and select firsts from them.
And another way is to create a sorted set by email, so it will treat all the records with equal emails as equal records:
(let [items [{ :name "John" :email "john#gmail.com" }
{ :name "Batman" :email "batman#gmail.com" }
{ :name "John Doe" :email "john#gmail.com" }]]
(into (sorted-set-by #(compare (:email %1) (:email %2))) items))
output:
#{{:name "Batman", :email "batman#gmail.com"}
{:name "John", :email "john#gmail.com"}}
don't really know which of them is more idiomatic and has a better performance. But i bet on the first one.

Answer 2

This would do it: https://crossclj.info/fun/medley.core/distinct-by.html.
The function in the link goes through every value lazily and stores everything it's seen. If the value in the coll is already seen, it does not add it.
You could then call this as: (distinct-by #(% :email) maps), where maps is your vector of people-maps.

Answer 3

A distinct-by can easily be implemented as
(defn distinct-by [f coll]
(let [groups (group-by f coll)]
(map #(first (groups %)) (distinct (map f coll)))))
For the example case this can be used like
(distinct-by :email
[{:name "John" :email "john#gmail.com"}
{:name "Batman" :email "batman#gmail.com"}
{:name "John Doe" :email "john#gmail.com"}])