Clean up and rework minimize implementation

lambdaisland · Mar 30, 2023 · 41d3153 · 41d3153
1 parent 5793228
commit 41d3153
Show file tree

Hide file tree

Showing 9 changed files with 155 additions and 70 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -3,7 +3,8 @@
 ## Added
 
 - Enable print tests in babashka
-- Add namespace `lambdaisland.deep-diff2.manipulate` ns with `remove-unchanged` API
+- Add a `lambdaisland.deep-diff2/minimize` function, which removes any items
+  that haven't changed from the diff.
 
 ## Fixed
 

diff --git a/README.md b/README.md
@@ -88,6 +88,21 @@ For fine grained control you can create a custom Puget printer, and supply it to
 
 For more advanced uses like incorporating diffs into your own Fipp documents, see `lambdaisland.deep-diff2.printer/format-doc`, `lambdaisland.deep-diff2.printer/print-doc`.
 
+### Minimizing
+
+If you are only interested in the changes, and not in any values that haven't
+changed, then you can use `ddiff/minimize` to return a more compact diff.
+
+This is especially useful for potentially large nested data structures, for
+example a JSON response coming from a web service.
+
+```clj
+(-> (ddiff/diff {:a "apple" :b "pear"} {:a "apple" :b "banana"})
+    ddiff/minimize
+    ddiff/pretty-print)
+;; {:b -"pear" +"banana"}
+```
+
 ### Print handlers for custom or built-in types
 
 In recent versions deep-diff2 initializes its internal copy of Puget with

diff --git a/package.json b/package.json
@@ -4,6 +4,6 @@
   "dependencies": {
     "react": "^16.13.1",
     "react-dom": "^16.13.1",
-    "ws": "^8.11.0"
+    "ws": "^8.13.0"
   }
 }
diff --git a/repl_sessions/poke.clj b/repl_sessions/poke.clj
@@ -0,0 +1,62 @@
+(ns repl-sessions.poke
+  (:require [lambdaisland.deep-diff2 :as ddiff]))
+
+(seq #{{:foo 1M} {:bar 2}}) ;; => ({:foo 1M} {:bar 2})
+(seq #{{:foo 1} {:bar 2}})  ;; => ({:bar 2} {:foo 1})
+
+(def d1 {{:foo 1M} {:bar 2}})
+(def d2 {{:foo 1} {:bar 2}})
+(ddiff/pretty-print (ddiff/diff d1 d2))
+;; #{+{:foo 1} -{:foo 1M} {:bar 2}}
+
+(def d1 #{{:foo 1M}})
+(def d2 #{{:foo 1}})
+(ddiff/pretty-print (ddiff/diff d1 d2))
+
+(-> (ddiff/diff {:a "apple" :b "pear"} {:a "apple" :b "banana"})
+    ddiff/minimize
+    ddiff/pretty-print)
+;; {:b -"pear" +"banana"}
+
+;; {:b -2 +3}
+
+[#{1.1197369622161879e-14 1.3019822841584656e-21 0.6875
+   #uuid "a907a7fe-d2eb-482d-b1cc-3acfc12daf55"
+   -30
+   :X/*!1:3
+   :u7*A/p?2IG5d*!Nl
+   :**d7ws
+   "ý"
+   "ÔB*àñS�¬ÚûV¡ç�¯±·á£H�
+                            �û?'V$ëY;CL�k-oOV"
+   !U-h_C*A7/x0_n1
+   A-*wn./o_?4w18-!
+   "ìêÜ¼à4�^¤mÐðkt�ê1_ò�· À�4\n@J\"2�9)cd-\t®"
+   y3W-2
+   #uuid "6d507164-f8b9-401d-8c44-d6b0e310c248"
+   "M"
+   :cy7-3
+   :w4/R.-s?9V5
+   #uuid "1bcb00c9-88b9-4eae-9fea-60600dfaefa0"
+   -20
+   #uuid "269ab6f9-f19d-4c9d-a0cb-51150e52e9f7"
+   -235024979
+   :O:m_9.9+A/N+usPa6.HA*G
+   228944.657438457
+   :x/w?
+   :__+o+sut9!t/?0l
+   "�â��«"
+   false
+   #uuid "b6295f83-8176-47b5-946e-466f74226629"
+   e3zQ!E*5
+   :T5rb
+   :++y:2
+   -7364
+   zG/ex23
+   "¡"
+   -4318364480
+   :D+?2?!/Hrc!jA7z_2
+   :z-I/!8Uq+d?
+   -0.5588235294117647
+   -0.5925925925925926
+   -0.8108108108108109}]
diff --git a/src/lambdaisland/deep_diff2.cljc b/src/lambdaisland/deep_diff2.cljc
@@ -1,6 +1,8 @@
 (ns lambdaisland.deep-diff2
-  (:require [lambdaisland.deep-diff2.diff-impl :as diff-impl]
-            [lambdaisland.deep-diff2.printer-impl :as printer-impl]))
+  (:require
+   [lambdaisland.deep-diff2.diff-impl :as diff-impl]
+   [lambdaisland.deep-diff2.printer-impl :as printer-impl]
+   [lambdaisland.deep-diff2.minimize-impl :as minimize]))
 
 (defn diff
   "Compare two values recursively.
@@ -41,3 +43,8 @@
    (-> diff
        (printer-impl/format-doc printer)
        (printer-impl/print-doc printer))))
+
+(defn minimize
+  "Return a minimal diff, removing any values that haven't changed."
+  [diff]
+  (minimize/minimize diff))
diff --git a/src/lambdaisland/deep_diff2/manipulate.cljc b/src/lambdaisland/deep_diff2/manipulate.cljc
diff --git a/src/lambdaisland/deep_diff2/minimize_impl.cljc b/src/lambdaisland/deep_diff2/minimize_impl.cljc
@@ -0,0 +1,47 @@
+(ns lambdaisland.deep-diff2.minimize-impl
+  "Provide API for manipulate the diff structure data "
+  (:require [clojure.walk :refer [postwalk]]
+            #?(:clj [lambdaisland.deep-diff2.diff-impl]
+               :cljs [lambdaisland.deep-diff2.diff-impl :refer [Mismatch Deletion Insertion]]))
+  #?(:clj (:import [lambdaisland.deep_diff2.diff_impl Mismatch Deletion Insertion])))
+
+(defn diff-item?
+  "Checks if x is a Mismatch, Deletion, or Insertion"
+  [x]
+  (or (instance? Mismatch x)
+      (instance? Deletion x)
+      (instance? Insertion x)))
+
+(defn has-diff-item?
+  "Checks if there are any diff items in x or sub-tree of x"
+  [x]
+  (or (diff-item? x)
+      (and (map? x) (some #(or (has-diff-item? (key %))
+                               (has-diff-item? (val %))) x))
+      (and (coll? x) (some has-diff-item? x))))
+
+(defn minimize
+  "Postwalk diff, removing values that are unchanged"
+  [diff]
+  (let [y (postwalk
+           (fn [x]
+             (cond
+               (map-entry? x)
+               ;; Either k or v of a map-entry contains/is? diff-item,
+               ;; keep the map-entry. Otherwise, remove it.
+               (when (or (has-diff-item? (key x))
+                         (has-diff-item? (val x)))
+                 x)
+
+               (map? x)
+               x
+
+               (coll? x)
+               (into (empty x) (filter has-diff-item?) x)
+
+               :else
+               x))
+           diff)]
+    (cond
+      (coll? y) y
+      :else     nil)))
diff --git a/...bdaisland/deep_diff2/manipulate_test.cljc → ...ambdaisland/deep_diff2/minimize_test.cljc b/...bdaisland/deep_diff2/manipulate_test.cljc → ...ambdaisland/deep_diff2/minimize_test.cljc
@@ -1,15 +1,14 @@
-(ns lambdaisland.deep-diff2.manipulate-test
+(ns lambdaisland.deep-diff2.minimize-test
   (:require [clojure.test :refer [deftest testing is are]]
             [lambdaisland.deep-diff2.diff-test :as diff-test]
             [clojure.test.check.clojure-test :refer [defspec]]
             [clojure.test.check.generators :as gen]
             [clojure.test.check.properties :as prop]
             [lambdaisland.deep-diff2.diff-impl :as diff]
-            [lambdaisland.deep-diff2.manipulate :as manipulate]
             [lambdaisland.deep-diff2 :as ddiff]))
 
 (deftest basic-strip-test
-  (testing "diff without remove-unchanged"
+  (testing "diff without minimize"
     (let [x {:a 1 :b 2 :d {:e 1} :g [:e [:k 14 :g 15]]}
           y {:a 1 :c 3 :d {:e 15} :g [:e [:k 14 :g 15]]}]
       (is (= (ddiff/diff x y)
@@ -18,62 +17,62 @@
               :d {:e (diff/->Mismatch 1 15)}
               :g [:e [:k 14 :g 15]]
               (diff/->Insertion :c) 3}))))
-  (testing "diff with remove-unchanged"
+  (testing "diff with minimize"
     (let [x {:a 1 :b 2 :d {:e 1} :g [:e [:k 14 :g 15]]}
           y {:a 1 :c 3 :d {:e 15} :g [:e [:k 14 :g 15]]}]
-      (is (= (manipulate/remove-unchanged (ddiff/diff x y))
+      (is (= (ddiff/minimize (ddiff/diff x y))
              {(diff/->Deletion :b) 2
               :d {:e (diff/->Mismatch 1 15)}
               (diff/->Insertion :c) 3})))))
 
-(deftest remove-unchanged-on-diff-test
+(deftest minimize-on-diff-test
   (testing "diffing atoms"
     (testing "when different"
-      (is (= (manipulate/remove-unchanged
+      (is (= (ddiff/minimize
               (ddiff/diff :a :b))
              (diff/->Mismatch :a :b))))
 
     (testing "when equal"
-      (is (= (manipulate/remove-unchanged
+      (is (= (ddiff/minimize
               (ddiff/diff :a :a))
              nil))))
 
   (testing "diffing collections"
     (testing "when different collection types"
-      (is (= (manipulate/remove-unchanged
+      (is (= (ddiff/minimize
               (ddiff/diff [:a :b] #{:a :b}))
              (diff/->Mismatch [:a :b] #{:a :b}))))
 
     (testing "when equal with clojure set"
-      (is (= (manipulate/remove-unchanged
+      (is (= (ddiff/minimize
               (ddiff/diff #{:a :b} #{:a :b}))
              #{})))
 
     (testing "when different with clojure set"
-      (is (= (manipulate/remove-unchanged
+      (is (= (ddiff/minimize
               (ddiff/diff #{:a :b :c} #{:a :b :d}))
              #{(diff/->Insertion :d) (diff/->Deletion :c)})))
 
     (testing "when equal with clojure vector"
-      (is (= (manipulate/remove-unchanged
+      (is (= (ddiff/minimize
               (ddiff/diff [:a :b] [:a :b]))
              [])))
 
     (testing "when equal with clojure hashmap"
-      (is (= (manipulate/remove-unchanged
+      (is (= (ddiff/minimize
               (ddiff/diff {:a 1} {:a 1}))
              {})))
 
     (testing "when equal with clojure nesting vector"
-      (is (= (manipulate/remove-unchanged
+      (is (= (ddiff/minimize
               (ddiff/diff [:a [:b :c :d]] [:a [:b :c :d]]))
              [])))))
 
-;; "diff itself and remove-unchanged yields empty"
+;; "diff itself and minimize yields empty"
 (defspec diff-itself 100
   (prop/for-all
    [x diff-test/gen-any-except-NaN]
    (if (coll? x)
-     (= (manipulate/remove-unchanged (ddiff/diff x x))
+     (= (ddiff/minimize (ddiff/diff x x))
         (empty x))
-     (nil? (manipulate/remove-unchanged (ddiff/diff x x))))))
+     (nil? (ddiff/minimize (ddiff/diff x x))))))
diff --git a/tests.edn b/tests.edn
@@ -1,4 +1,6 @@
 #kaocha/v1
 {:tests [{:id :clj}
          {:id :cljs
-          :type :kaocha.type/cljs}]}
+          :type :kaocha.type/cljs}]
+ :kaocha/bindings {kaocha.stacktrace/*stacktrace-filters* []}
+ }