Reputation: 10709
I've hash-map in clojure:
{"key1" "value1"} {"key2" "value2"} {"key3" "value1"}
and i need to convert it into hash map of
{"value1" {"key1" "key3"}} {"value2" {"key2"}}
Any clojure way of doing this?
clojure.set/map-invert will not work as if it overrides repeated values.
Upvotes: 3
Views: 496
Reputation: 236094
Give this a try:
(def m {"key1" "value1" "key2" "value2" "key3" "value1"})
(reduce (fn [a x] (assoc a (second x) (conj (a (second x)) (first x)))) {} m)
=> {"value2" ("key2"), "value1" ("key3" "key1")}
Notice that the (possibly) repeated values end up in a list. Or, as suggested by @A.Webb in the comments, the above can be concisely written like this:
(reduce (fn [a [k v]] (update-in a [v] conj k)) {} m)
=> {"value2" ("key2"), "value1" ("key3" "key1")}
Upvotes: 2
Reputation: 34840
(def m {"key1" "value1" "key2" "value2" "key3" "value1"})
(let [g (group-by val m)
vals (map #(map first %) (vals g))]
(zipmap (keys g) vals))
;;=> {"value2" ("key2"), "value1" ("key1" "key3")}
Upvotes: 5
Reputation: 4833
Here's another alternative:
(def maps [{"key1" "value1"} {"key2" "value2"} {"key3" "value1"}])
(into {}
(for [[k v] (group-by #(val (first %)) maps)]
[k (apply concat (map keys v))]))
;=> {"value1" ("key1" "key3"), "value2" ("key2")}
Upvotes: 1
Reputation: 32478
Try the following, I don't have REPL to test, it may slightly differ from the original answer, but, you can use it to get what you want.
(group-by #(val (first %)) {"key1" "value1"} {"key2" "value2"} {"key3" "value1"})
Upvotes: 1
Reputation: 33657
Your question is really confusing. First you said you have hash-map, what you actually have as example is many hash-maps or if you really mean a single hash-map then your example is incorrect as it should be {"key1" "value1" "key2" "value2" "key3" "value1"}
. Second, the required output is weird in the sense that the values in the hash-map are represented as maps themselves but the last value is just a single value in a hash-map which is obviously not possible, so I guess you mean the values are set (not map) as
{"value1" #{"key1" "key3"}} {"value2" #{"key2"}}
Based on all these assumptions one possible solution would be:
(->> {"key1" "value1" "key2" "value2" "key3" "value1"}
(group-by second)
(map #(-> [(%1 0) (into #{} (map first (%1 1)))]))
(into {}))
Upvotes: 1