Skip to content
A utility library for Clojure of functions and macros that complement clojure.core
Branch: develop
Clone or download
Latest commit 81ad34e Jan 12, 2020
Type Name Latest commit message Commit time
Failed to load latest commit information.
src/missing add backoff Jan 13, 2020
test/missing - fix merge sort bug producing nil value when the colls are uneven / … Dec 22, 2019
.gitignore update .gitignore May 30, 2018
LICENSE prep for open source distribution Sep 7, 2018 update docs Apr 6, 2019
project.clj - bump version Dec 31, 2019

Build Status Maven metadata URL


A utility library for Clojure of functions and macros that complement clojure.core.


[com.vodori/missing "0.1.22"]

Sample usages

Below are some examples of functions available in this library. For a full indication of what the library supports, please see the tests.

Comparable operators for more than numbers and O(n) scans

The existing <, >, <=, >= unfortunately only work on numbers. Use these for a more general alternative with similar semantics. If all you're looking for is the smallest or largest item, you shouldn't have to sort your sequence.

(require '[missing.core :refer :all])

(lt "a" "b") ;=> true
(lte "a" "a") ;=> true
(gt "b" "a") ;=> true
(gte "b" "b") ;=> true

(least ["a" "b" "c"]) ;=> "a"
(greatest ["a" "b" "c"]) ;=> "c"
(least-by :count [{:count 0} {:count 1} {:count -1}]) ;=> {:count -1}
(greatest-by :count [{:count 0} {:count 1} {:count -1}]) ;=> {:count 1}

Concurrent execution of forms

A little macro sugar on top of future to execute multiple independent tasks concurrently and return a sequence of their results.

(require '[missing.core :refer :all])

(zipmap [:users :posts] (together (get-users) (get-posts)))


(require '[missing.core :refer :all])

(subsets #{1 2 3}) 
;=> #{#{} #{1} #{2} #{3} #{1 2} #{2 3} #{1 2 3}}

(submaps {:a 1 :b 2 :c 3}) 
;=> {} {:a 1} {:c 3} {:b 2} {:c 3, :b 2} {:b 2, :a 1} {:c 3, :a 1} {:c 3, :b 2, :a 1}

(intersect? #{1 2 3} #{1 3} #{1}) 
;=> true

(exclusive? #{1 2 3} #{4 5} #{6}) 
;=> true

Lazy merge sort

Create a single sorted sequence by lazily interleaving already sorted sequences.

(require '[missing.core :refer :all])

(def x [:a :c :d :e :i])
(def y [:b :f :g :h])
(merge-sort [x y]) ;=> [:a :b :c :d :e :f :g :h :i]

Indexing collections

Use these when you're building lookup tables to efficiently perform batch operations.

(require '[missing.core :refer :all])

(def users 
    (index-by :username 
        [{:username "fred" :email ""} 
         {:username "greg" :email ""}])

(get users "fred") 
;=> {:username "fred" :email ""}

(def resources 
  [{:meta {:app "sso" :version "2018-10" :stage "dev"}}
   {:meta {:app "db" :version "2018-07" :stage "prod"}}
   {:meta {:app "api" :version "2018-07" :stage "dev"}}])
(def table (group-by-labels :meta resources))

(get table {:version "2018-10"})
;=> [{:meta {:app "sso" :version "2018-10" :stage "dev"}}]

(get table {:stage "dev"}) 
;=> [{:meta {:app "sso" :version "2018-10" :stage "dev"}}
;    {:meta {:app "api" :version "2018-07" :stage "dev"}}]

(get table {:version "2018-07" :stage "dev"}) 
;=> [{:meta {:app "api" :version "2018-07" :stage "dev"}}]

(let [pk1           ""
      pk2           ""
      person-info  [{:id pk1 :name "Paul" :age 26}
                    {:id pk2 :name "Raul" :age 62}]
      account-info [{:email pk1 :year-joined 2018}
                    {:email pk2 :year-joined 8102}]]
  (collate [[:id person-info] [:email account-info]])))
;=> {""
;    [{:id "", :name "Raul", :age 62}
;      {:email "", :year-joined 8102}],
;    ""
;    [{:id "", :name "Paul", :age 26}
;     {:email "", :year-joined 2018}]}

Locking by value

Clojure has lots of great ways to deal with state. Reference locking is probably least among them but if the use case is isolated and the exclusive operation involves side effects (can't run in a CAS loop) sometimes it's an appropriate choice. Missing provides reentrant locks that lock on values so you can ensure different parts of your program never act on behalf of the same value (often an identifier) at the same time.

(require '[missing.locks :as locks])

(locks/locking "user-id" (update-user user))

; you can also subdivide the exclusive scope by wrapping 
; evaluation with your own lock tables (a map in an atom)

(def profile-locks (atom {}))

(locks/with-locks profile-locks
  (locks/locking "user-id" (update-user user)))

Transducers: distinct-by and dedupe-by

The xxx-by transducers are, in my opinion, always preferable to a xxx transducer. The reason being that xxx-by degrades into xxx when f is identity and so is equivalent but more powerful.

(require '[missing.core :refer :all])

(distinct-by :x [{:x 1} {:x 2} {:x 1}]) ;=> [{:x 1} {:x 2}]

(dedupe-by :x [{:x 1} {:x 1} {:x 2} {:x 1}]) ;=> [{:x 1} {:x 2} {:x 1}]

Transducer: contiguous-by

Sometimes you have a sequence of elements that are potentially overlapping / abutting. You might want to merge these segments into one item instead. This transducer just wraps partition-by with one-dimensional overlap tracking to chunk the sequence into contiguous segments. It works on any comparables and is lazy.

(require '[missing.core :refer :all])

(def flat [{:x1 0 :x2 4} {:x1 1 :x2 5} {:x1 5 :x2 6} {:x1 10 :x2 12}])

(contiguous-by :x1 :x2 flat)
;=> [[{:x1 0 :x2 4} {:x1 1 :x2 5} {:x1 5 :x2 6}] [{:x1 10 :x2 12}]]

; after this you'll probably perform a map to merge the
; contiguous items in each partition into one element


Sometimes you want to bail on further computation if you happen to discover a result early but you don't want to restructure your code to plan for the early termination. Preemptables offer a generic solution to this problem by utilizing exceptions as a pseudo-continuation.

(require '[missing.core :refer :all])

; preemptable marks the place that preempt can provide a value for.
; preempt provides a value and throws to unwind the stack rather than
; continue the computation. If you never call preempt then the return
; value is equal to the result of the entire evaluation.

 (dotimes [x 1000]
   (if (and (pos? x) (even? x))
     (preempt x)
; => 2

Structural Selections

Sometimes you want flat data that describes the structure of nested data. Missing provides a path protocol and implementations to extract a path to each leaf in a piece of data. There's even a function that will "select" a given structure from another piece of data by example.

(require '[missing.core :refer :all])

(index-values-by-paths {:a [:b {:c :d}]})
; => {[:a 0] :b, [:a 1 :c] :d}

  ; data to extract from
  {:a [1 {:test [4 [{:one 5}]]}] :extra :stuff}
  ; example structure to use for the extraction
  {:a [:b {:test [:thing [{:one :two :three 4}]]} :d]})
; => {:a [1 {:test [4 [{:one 5 :three nil}]]} nil]}

Graph Functions

The missing.topology namespace contains simple implementations of many basic graph functions. Every function that takes a graph accepts just a plain adjacency map (map of node to list of nodes it has edges to). If your nodes don't have value semantics then you should just use a graph of identifiers and perform lookups back to your objects.

Perhaps most generally useful are the topological sorts:

(require '[missing.topology :refer :all])

; define an adjacency graph with edges indicating
; left 'is a requirement of' right
; a is a requirement of b and c
; b is a requirement of d
(def g {:a [:b :c] :b [:d]})

; sort them into phases where items
; in each phase can be resolved 
; concurrently
(topological-sort-with-grouping g) 
; => [#{:a} #{:c :b} #{:d}]

; If you'd rather, you could define
; an adjacency graph with edges indicating
; left 'depends on' right
; b depends on a
; c depends on a
; d depends on b

(def g' {:b [:a] :c [:a] :d [:b]})

; and just invert it before sorting:
(topological-sort-with-grouping (inverse g')) 
; => [#{:a} #{:c :b} #{:d}]


This project is licensed under MIT license.

You can’t perform that action at this time.