Implemented groupBy:aggregate: and select:groupBy:aggregate:. Wrote tests for them

olekscode · olekscode · commit d0e1ecdebbf5 · 2017-08-28T18:32:31.000+03:00
diff --git a/DataFrame-Core.package/DataFrame.class/class/fromXLSX.worksheetNamed..st b/DataFrame-Core.package/DataFrame.class/class/fromXLSX.worksheetNamed..st
@@ -4,7 +4,7 @@ fromXLSX: pathToFile worksheetNamed: sheetName
 	| file workbook sheet |
 	
 	file := FileStream readOnlyFileNamed: pathToFile.
-	workbook := XLSXImporter import file.
+	workbook := XLSXImporter import: file.
 	sheet := workbook worksheetNamed: sheetName.
 	
 	^ self fromXLSXworksheet: sheet.
diff --git a/DataFrame-Core.package/DataFrame.class/instance/group.by..st b/DataFrame-Core.package/DataFrame.class/instance/group.by..st
@@ -1,4 +1,4 @@
-as yet unclassified
+queries
 group: colNameOrArray by: colName
 
 	| left right |
diff --git a/DataFrame-Core.package/DataFrame.class/instance/groupBy..st b/DataFrame-Core.package/DataFrame.class/instance/groupBy..st
@@ -1,4 +1,4 @@
-as yet unclassified
+queries
 groupBy: colName
 
 	| groupedColNames |
diff --git a/DataFrame-Core.package/DataFrame.class/instance/groupBy.aggregate..st b/DataFrame-Core.package/DataFrame.class/instance/groupBy.aggregate..st
@@ -0,0 +1,4 @@
+queries
+groupBy: colName aggregate: selector
+
+	^ (self groupBy: colName) perform: selector.
diff --git a/DataFrame-Core.package/DataFrame.class/instance/select.groupBy.aggregate..st b/DataFrame-Core.package/DataFrame.class/instance/select.groupBy.aggregate..st
@@ -0,0 +1,10 @@
+queries
+select: colNames groupBy: aColName aggregate: selector
+	
+	| df |
+	
+	df := (colNames includes: aColName)
+		ifTrue: [ self columns: colNames ]
+		ifFalse: [ self columns: (colNames copyWith: aColName) ].
+	
+	^ df groupBy: aColName aggregate: selector.
diff --git a/DataFrame-Core.package/DataFrame.class/instance/select.where..st b/DataFrame-Core.package/DataFrame.class/instance/select.where..st
@@ -1,4 +1,4 @@
-as yet unclassified
+queries
 select: colNames where: aBlock
 	
 	^ (self selectAllWhere: aBlock) columns: colNames
diff --git a/DataFrame-Core.package/DataFrame.class/instance/select.where.groupBy.aggregate..st b/DataFrame-Core.package/DataFrame.class/instance/select.where.groupBy.aggregate..st
@@ -1,5 +1,6 @@
-as yet unclassified
+queries
 select: colNames where: aBlock groupBy: aColName aggregate: selector
 	
-	^ ((self select: colNames where: aBlock)
-		groupBy: aColName) perform: selector.
+	^ (self select: colNames where: aBlock)
+		groupBy: aColName
+		aggregate: selector.
diff --git a/DataFrame-Core.package/DataFrame.class/instance/selectAllWhere..st b/DataFrame-Core.package/DataFrame.class/instance/selectAllWhere..st
@@ -1,4 +1,4 @@
-as yet unclassified
+queries
 selectAllWhere: aBlock
 
 	| testedColumnNames |
diff --git a/DataFrame-Tests.package/DataFrameInternalTests.class/instance/testAddColumn.st b/DataFrame-Tests.package/DataFrameInternalTests.class/instance/testAddColumn.st
@@ -4,11 +4,11 @@ testAddColumn
 	| expected |
 	
 	expected := DataFrameInternal fromRows: #(
-		('Barcelona' 1.609 true 9)
-   		('Dubai' 2.789 true 8)
-   		('London' 8.788 false 7)).
+		('Barcelona' 1.609 true 4)
+   		('Dubai' 2.789 true 3)
+   		('London' 8.788 false 4)).
 		
-	df addColumn: #(9 8 7).
+	df addColumn: #(4 3 4).
 	
 	self assert: df equals: expected.
 	
diff --git a/DataFrame-Tests.package/DataFrameQueriesTests.class/instance/testGroupbyAggregate.st b/DataFrame-Tests.package/DataFrameQueriesTests.class/instance/testGroupbyAggregate.st
@@ -0,0 +1,16 @@
+initialization
+testGroupbyAggregate
+
+	| actual expected |
+	
+	actual := df groupBy: #species aggregate: #sum.
+	
+	expected := DataFrame fromRows: #(
+		(250.3 170.9 73.2 12.2)
+		(296.8 138.5 213.0 66.3)
+		(329.4 148.7 277.6 101.3)).
+		
+	expected columnNames: #(sepal_length sepal_width petal_length petal_width).
+	expected rowNames: #(setosa versicolor virginica).
+	
+	self assert: actual closeTo: expected.
diff --git a/DataFrame-Tests.package/DataFrameQueriesTests.class/instance/testSelectGroupbyAggregate.st b/DataFrame-Tests.package/DataFrameQueriesTests.class/instance/testSelectGroupbyAggregate.st
@@ -0,0 +1,20 @@
+initialization
+testSelectGroupbyAggregate
+
+	| selectedCols actual expected |
+	selectedCols := #(petal_length petal_width).
+	
+	actual := df
+		select: selectedCols
+		groupBy: #species
+		aggregate: #sum.
+	
+	expected := DataFrame fromRows: #(
+		(73.2 12.2)
+		(213.0 66.3)
+		(277.6 101.3)).
+		
+	expected columnNames: selectedCols.
+	expected rowNames: #(setosa versicolor virginica).
+	
+	self assert: actual closeTo: expected.
diff --git a/DataFrame-Tests.package/DataFrameTests.class/instance/setUp.st b/DataFrame-Tests.package/DataFrameTests.class/instance/setUp.st
@@ -7,4 +7,4 @@ setUp
    		('London' 8.788 false)).
 		
 	df rowNames: #(A B C).
-	df columnNames: #(City Population SomeBool).
+	df columnNames: #(City Population BeenThere).
diff --git a/DataFrame-Tests.package/DataFrameTests.class/instance/testAddColumn.st b/DataFrame-Tests.package/DataFrameTests.class/instance/testAddColumn.st
@@ -4,13 +4,13 @@ testAddColumn
 	| expected |
 	
 	expected := DataFrame fromRows: #(
-		('Barcelona' 1.609 true 9)
-   		('Dubai' 2.789 true 8)
-   		('London' 8.788 false 7)).
+		('Barcelona' 1.609 true 4)
+   		('Dubai' 2.789 true 3)
+   		('London' 8.788 false 4)).
 		
 	expected rowNames: #(A B C).
-	expected columnNames: #(City Population SomeBool SomeInt).
+	expected columnNames: #(City Population BeenThere Rating).
 		
-	df addColumn: #(9 8 7) named: #SomeInt.
+	df addColumn: #(4 3 4) named: #Rating.
 	
 	self assert: df equals: expected.
diff --git a/DataFrame-Tests.package/DataFrameTests.class/instance/testAddRow.st b/DataFrame-Tests.package/DataFrameTests.class/instance/testAddRow.st
@@ -10,7 +10,7 @@ testAddRow
 		('Lviv' 0.724 true)).
 		
 	expected rowNames: #(A B C D).
-	expected columnNames: #(City Population SomeBool).
+	expected columnNames: #(City Population BeenThere).
 		
 	df addRow: #('Lviv' 0.724 true) named: #D.
 	
diff --git a/DataFrame-Tests.package/DataFrameTests.class/instance/testColumnAtPut.st b/DataFrame-Tests.package/DataFrameTests.class/instance/testColumnAtPut.st
@@ -9,7 +9,7 @@ testColumnAtPut
    		('Z' 8.788 0)).
 		
 	expected rowNames: #(A B C).
-	expected columnNames: #(City Population SomeBool).
+	expected columnNames: #(City Population BeenThere).
 	
 	actual := df.
 	actual columnAt: 1 put: #('X' 'Y' 'Z').
diff --git a/DataFrame-Tests.package/DataFrameTests.class/instance/testColumnPut.st b/DataFrame-Tests.package/DataFrameTests.class/instance/testColumnPut.st
@@ -9,10 +9,10 @@ testColumnPut
    		('Z' 8.788 0)).
 		
 	expected rowNames: #(A B C).
-	expected columnNames: #(City Population SomeBool).
+	expected columnNames: #(City Population BeenThere).
 	
 	actual := df.
 	actual column: #City put: #('X' 'Y' 'Z').
-	actual column: #SomeBool put: #(0 1 0).
+	actual column: #BeenThere put: #(0 1 0).
 	
 	self assert: actual equals: expected.
diff --git a/DataFrame-Tests.package/DataFrameTests.class/instance/testColumns.st b/DataFrame-Tests.package/DataFrameTests.class/instance/testColumns.st
@@ -3,7 +3,7 @@ testColumns
 
 	| columnNames actualDataFrame expectedDataFrame |
 	
-	columnNames := #(City SomeBool).
+	columnNames := #(City BeenThere).
 	
 	expectedDataFrame := DataFrame fromRows: #(
 		('Barcelona' true)
diff --git a/DataFrame-Tests.package/DataFrameTests.class/instance/testColumnsFromTo.st b/DataFrame-Tests.package/DataFrameTests.class/instance/testColumnsFromTo.st
@@ -9,7 +9,7 @@ testColumnsFromTo
    		(8.788 false)).
 		
 	expectedDataFrame rowNames: df rowNames.
-	expectedDataFrame columnNames: #(Population SomeBool).
+	expectedDataFrame columnNames: #(Population BeenThere).
 	
 	actualDataFrame := df columnsFrom: 2 to: 3.
 	
diff --git a/DataFrame-Tests.package/DataFrameTests.class/instance/testRowAtPut.st b/DataFrame-Tests.package/DataFrameTests.class/instance/testRowAtPut.st
@@ -9,7 +9,7 @@ testRowAtPut
    		('London' 8.788 false)).
 		
 	expected rowNames: #(A B C).
-	expected columnNames: #(City Population SomeBool).
+	expected columnNames: #(City Population BeenThere).
 	
 	actual := df.
 	actual rowAt: 2 put: #('X' 'Y' 'Z').
diff --git a/DataFrame-Tests.package/DataFrameTests.class/instance/testRowPut.st b/DataFrame-Tests.package/DataFrameTests.class/instance/testRowPut.st
@@ -9,7 +9,7 @@ testRowPut
    		('London' 8.788 false)).
 		
 	expected rowNames: #(A B C).
-	expected columnNames: #(City Population SomeBool).
+	expected columnNames: #(City Population BeenThere).
 	
 	actual := df.
 	actual row: #B put: #('X' 'Y' 'Z').

Original file line number	Diff line number	Diff line change
`@@ -1,4 +1,4 @@`
`1`		`-as yet unclassified`
	`1`	`+queries`
`2`	`2`	`group: colNameOrArray by: colName`
`3`	`3`
`4`	`4`	`\| left right \|`
-Original file line number
+Diff line change
@@ @@ -0,0 +1,4 @@ @@
 +queries
 +groupBy: colName aggregate: selector
++
 +	^ (self groupBy: colName) perform: selector.