Fixes #259, Fixes #311

CHANGELOG.md

@@ -1,4 +1,8 @@
 # Changelog
+# 8.093
+ * Fixes for issue [259](https://github.com/techascent/tech.ml.dataset/issues/259/), which is same as new issue 311.  `:key-fn` should only be applied once per column and does not have to
+   be idemptotent.
+
 # 8.092
  * Fixes for issues [312](https://github.com/techascent/tech.ml.dataset/issues/312/), [315](https://github.com/techascent/tech.ml.dataset/issues/315/), [316](https://github.com/techascent/tech.ml.dataset/issues/316/)
 

src/tech/v3/dataset/io/context.clj

@@ -104,7 +104,8 @@
           (mapv (fn [{:keys [column-name column-parser]}]
                   (assoc (column-parsers/finalize! column-parser row-count)
                          :tech.v3.dataset/name column-name)))
-          (ds-impl/new-dataset options))))
+          ;;key-fn has already been applied
+          (ds-impl/new-dataset (assoc options :key-fn nil)))))
   ([options parsers]
    (parsers->dataset options parsers
                      (apply max 0 (map (comp dtype/ecount :column-parser)

src/tech/v3/dataset/io/mapseq_colmap.clj

@@ -44,7 +44,8 @@
                       (recur (.hasNext iter)
                              (unchecked-inc row-idx)))
                     row-idx))]
-     (parse-context/parsers->dataset options parsers n-rows)))
+     ;;key-fn has already been applied
+     (parse-context/parsers->dataset (assoc options :key-fn nil) parsers n-rows)))
   ([mapseq]
    (mapseq->dataset {} mapseq)))
 

test/tech/v3/dataset_test.clj

@@ -1512,6 +1512,24 @@
                             (ds/drop-rows (ds/->dataset [{:a 1 :c3 2}]) [0]))))))
 
 
+(deftest issue-259
+  (let [ds (ds/->dataset [{"a o" 1 "b o" 2} {"a o" 5 "b o" 3}]
+                         {:key-fn #(keyword (clojure.string/replace % " " "-"))})]
+    (is (= #{:b-o :a-o} (set (map (comp :name meta) (vals ds))))))
+  (let [ds (ds/->dataset {"a o" [1 5] "b o" [2 3]}
+                         {:key-fn #(keyword (clojure.string/replace % " " "-"))})]
+    (is (= #{:b-o :a-o} (set (map (comp :name meta) (vals ds))))))
+  (let [ds (ds/->dataset [{"Foo" 1 , "Bar" 2}]
+                         {:key-fn #(keyword (.toLowerCase %))})]
+    (is (= #{:foo :bar}
+           (set (map (comp :name meta) (vals ds))))))
+  (let [ds (ds/->dataset (java.io.ByteArrayInputStream. (.getBytes "Foo,Bar\n1,2"))
+                         {:key-fn #(keyword (.toLowerCase %))
+                          :file-type :csv})]
+    (is (= #{:foo :bar}
+           (set (map (comp :name meta) (vals ds)))))))
+
+
 (comment
 
   (def test-ds (ds/->dataset