modif on joins

Author: Gmousse

examples/titanic_analysis.js

@@ -55,7 +55,7 @@ DataFrame.fromCSV('http://vincentarelbundock.github.io/Rdatasets/csv/COUNT/titan
         const cleanCountByGroup = countByGroup.rename('aggregation', 'count').sortBy('count', true);
 
         // And now show the result
-        cleanCountByGroup.show(100);
+        cleanCountByGroup.show(300);
         // | class     | age       | sex       | survived  | count     |
         // ------------------------------------------------------------
         // | 3rd class | adults    | man       | no        | 387       |

src/dataframe.js

@@ -174,26 +174,33 @@ class DataFrame {
         return [array.map(row => new Row(row, columns)), columns];
     }
 
-    _joinByType(gdf1, gdf2, type) {
-        if (type === 'out' || type === 'in') {
-            const gdf2Hashs = gdf2.listHashs();
-            return gdf1.toCollection().map(({group, hash}) => {
-                const isContained = gdf2Hashs.includes(hash);
-                console.log(gdf2.get(hash).group.listColumns());
-                const filterCondition = (bool) => bool ? group : false;
-                return type === 'out' ? filterCondition(!isContained) : filterCondition(isContained);
-            }).filter(group => group);
-        }
-        return gdf1.toCollection().map(({group}) => group);
+    _joinByType(gdf1, gdf2, type, newColumns) {
+        const gdf2Hashs = gdf2.listHashs();
+        return gdf1.toCollection().map(({group, hash}) => {
+            const isContained = gdf2Hashs.includes(hash);
+            if (gdf2.get(hash)) {
+                const gdf2Collection = gdf2.get(hash).group.toCollection();
+                const combinedGroup = group.toCollection().map(row => {
+                    return gdf2Collection.map(row2 => Object.assign({}, row2, row));
+                }).reduce((p, n) => [...p, ...n], []);
+                group = this.__newInstance__(
+                    combinedGroup,
+                    newColumns
+                );
+            }
+            const filterCondition = (bool) => bool ? group : false;
+            if (type !== 'out' && type !== 'in') return group;
+            return type === 'out' ? filterCondition(!isContained) : filterCondition(isContained);
+        }).filter(group => group);
     }
 
     _join(dfToJoin, on, types) {
         const newColumns = [...new Set([...this.listColumns(), ...dfToJoin.listColumns()])];
         const gdf = this.groupBy(...on);
         const gdfToJoin = dfToJoin.groupBy(...on);
         return [this.__newInstance__([], newColumns), ...iter([
-            ...this._joinByType(gdf, gdfToJoin, types[0]),
-            ...this._joinByType(gdfToJoin, gdf, types[1]),
+            ...this._joinByType(gdf, gdfToJoin, types[0], newColumns),
+            ...this._joinByType(gdfToJoin, gdf, types[1], newColumns),
         ], group => group.restructure(newColumns))].reduce((p, n) => p.union(n));
     }
 

tests/dataframe-test.js

@@ -587,55 +587,53 @@ test('DataFrame rows can be ', (assert) => {
 
     assert.deepEqual(
         df5.innerJoin(df6, 'id').sortBy('id').toArray(), [
-            [1, 0, undefined],
-            [1, undefined, 0],
-            [3, 1, undefined],
-            [3, undefined, 6],
-            [8, 1, undefined],
-            [8, undefined, 2],
+            [1, 0, 0],
+            [3, 1, 6],
+            [8, 1, 2],
         ], 'inner joined.'
     );
 
     assert.deepEqual(
         df5.fullJoin(df6, 'id').sortBy('id').toArray(), [
-            [1, 0, undefined],
-            [1, undefined, 0],
+            [1, 0, 0],
+            [1, 0, 0],
             [2, undefined, 1],
-            [3, 1, undefined],
-            [3, undefined, 6],
+            [3, 1, 6],
+            [3, 1, 6],
             [6, undefined, 1],
-            [8, 1, undefined],
-            [8, undefined, 2],
+            [8, 1, 2],
+            [8, 1, 2],
         ], 'full joined.'
     );
+
     assert.deepEqual(
         df5.outerJoin(df6, 'id').sortBy('id').toArray(), [
             [2, undefined, 1],
             [6, undefined, 1],
         ], 'outer joined.'
     );
 
+    df5.leftJoin(df6, 'id').sortBy('id').show()
+
     assert.deepEqual(
-        df5.leftJoin(df6, 'id').sortBy('id').toArray(), [
-            [1, 0, undefined],
-            [1, undefined, 0],
-            [3, 1, undefined],
-            [3, undefined, 6],
-            [8, 1, undefined],
-            [8, undefined, 2],
+        df6.leftJoin(df5, 'id').sortBy('id').toArray(), [
+            [1, 0, 0],
+            [2, undefined, 1],
+            [3, 1, 6],
+            [6, undefined, 1],
+            [8, 1, 2],
         ], 'left joined.'
     );
 
+    df5.rightJoin(df6, 'id').sortBy('id').show()
+
     assert.deepEqual(
         df5.rightJoin(df6, 'id').sortBy('id').toArray(), [
-            [1, 0, undefined],
-            [1, undefined, 0],
+            [1, 0, 0],
             [2, undefined, 1],
-            [3, 1, undefined],
-            [3, undefined, 6],
+            [3, 1, 6],
             [6, undefined, 1],
-            [8, 1, undefined],
-            [8, undefined, 2],
+            [8, 1, 2],
         ], 'right joined.'
     );