Added support for DataFrame init with missing values

src/DataFrame-Tests/DataFrameInternalTest.class.st

@@ -1,4 +1,4 @@
-Class {
+Class { 
 	#name : #DataFrameInternalTest,
 	#superclass : #TestCase,
 	#instVars : [
@@ -370,6 +370,19 @@ DataFrameInternalTest >> testVarSizeInstanceCreation [
 	self assert: dfActual numberOfColumns equals: 8.
 ]
 
+{ #category : #tests }
+DataFrameInternalTest >> testWithColumns [
+
+	| dfi validOutput |
+	dfi := DataFrameInternal withColumns: #((1 2) (1 2 3 4 5) ('a' 3.14 'bcd')).
+	validOutput := Array2D
+		rows: 5
+		columns: 3
+		contents: #(1 1 'a' 2 2 3.14 nil 3 'bcd' nil 4 nil nil 5 nil).
+
+	self assert: dfi asArray2D equals: validOutput.
+]
+
 { #category : #tests }
 DataFrameInternalTest >> testWithIndicesCollect [
 
@@ -399,3 +412,16 @@ DataFrameInternalTest >> testWithIndicesDo [
 	dfInternal withIndicesDo: [ :each :i :j |
 		self assert: each equals: ((i - 1) * 2 + j) ].
 ]
+
+{ #category : #tests }
+DataFrameInternalTest >> testWithRows [
+
+	| dfi validOutput |
+	dfi := DataFrameInternal withRows: #((1 2) (1 2 3 4 5) ('a' 3.14 'bcd')).
+	validOutput := Array2D
+		rows: 3
+		columns: 5
+		contents: #(1 2 nil nil nil 1 2 3 4 5 'a' 3.14 'bcd' nil nil).
+
+	self assert: dfi asArray2D equals: validOutput.
+]

src/DataFrame-Tests/DataFrameTest.class.st

@@ -1081,6 +1081,42 @@ DataFrameTest >> testCreateEmptyDataFrameWithColumnNames [
 	self assert: dataFrame columnNames equals: columnNames.
 ]
 
+{ #category : #tests }
+DataFrameTest >> testCreateEmptyDataFrameWithColumns [
+
+	| dataFrame  |
+	dataFrame := DataFrame withColumns: #().
+
+	self assert: dataFrame class equals: DataFrame.
+]
+
+{ #category : #tests }
+DataFrameTest >> testCreateEmptyDataFrameWithColumnsColumnNames [
+
+	| dataFrame  |
+	dataFrame := DataFrame withColumns: #() columnNames: #().
+
+	self assert: dataFrame class equals: DataFrame.
+]
+
+{ #category : #tests }
+DataFrameTest >> testCreateEmptyDataFrameWithColumnsRowNames [
+
+	| dataFrame  |
+	dataFrame := DataFrame withColumns: #() rowNames: #().
+
+	self assert: dataFrame class equals: DataFrame.
+]
+
+{ #category : #tests }
+DataFrameTest >> testCreateEmptyDataFrameWithColumnsRowNamesColumnNames [
+
+	| dataFrame  |
+	dataFrame := DataFrame withColumns: #() rowNames: #() columnNames: #().
+
+	self assert: dataFrame class equals: DataFrame.
+]
+
 { #category : #tests }
 DataFrameTest >> testCreateEmptyDataFrameWithRowNames [
 	| rowNames dataFrame |
@@ -1111,6 +1147,42 @@ DataFrameTest >> testCreateEmptyDataFrameWithRowNamesColumnNames [
 	self assert: dataFrame columnNames equals: columnNames.
 ]
 
+{ #category : #tests }
+DataFrameTest >> testCreateEmptyDataFrameWithRows [
+
+	| dataFrame  |
+	dataFrame := DataFrame withRows: #().
+
+	self assert: dataFrame class equals: DataFrame.
+]
+
+{ #category : #tests }
+DataFrameTest >> testCreateEmptyDataFrameWithRowsColumnNames [
+
+	| dataFrame  |
+	dataFrame := DataFrame withRows: #() columnNames: #().
+
+	self assert: dataFrame class equals: DataFrame.
+]
+
+{ #category : #tests }
+DataFrameTest >> testCreateEmptyDataFrameWithRowsRowNames [
+
+	| dataFrame  |
+	dataFrame := DataFrame withRows: #() rowNames: #().
+
+	self assert: dataFrame class equals: DataFrame.
+]
+
+{ #category : #tests }
+DataFrameTest >> testCreateEmptyDataFrameWithRowsRowNamesColumnNames [
+
+	| dataFrame  |
+	dataFrame := DataFrame withRows: #() rowNames: #() columnNames: #().
+
+	self assert: dataFrame class equals: DataFrame.
+]
+
 { #category : #tests }
 DataFrameTest >> testCrossTabulation [
 
@@ -1407,6 +1479,19 @@ DataFrameTest >> testReject [
 	self assert: actual equals: expected.
 ]
 
+{ #category : #tests }
+DataFrameTest >> testRejectEntireDataFrame [
+	| actual expected |
+
+	expected := DataFrame
+		withRows: #().
+
+	actual := df reject:
+		[ :row | (row at: #Population) < 10 ].
+
+	self assert: actual equals: expected.
+]
+
 { #category : #tests }
 DataFrameTest >> testRemoveColumn [
 
@@ -1948,6 +2033,18 @@ DataFrameTest >> testSelect [
 	self assert: actual equals: expected.
 ]
 
+{ #category : #tests }
+DataFrameTest >> testSelectEmptyDataFrame [
+	| actual expected |
+
+	expected := DataFrame withRows: #().
+
+	actual := df select:
+		[ :row | (row at: #Population) > 10 ].
+
+	self assert: actual equals: expected.
+]
+
 { #category : #tests }
 DataFrameTest >> testSortBy [
 

src/DataFrame/DataFrame.class.st

@@ -1155,17 +1155,20 @@ DataFrame >> select: aBlock [
 	Collect into a new collection like the receiver, only those elements for 
 	which aBlock evaluates to true. Answer the new collection."
 
-	| rows selectedRows selectedRowNames selectedRowsAsArrays df |
+	| rows selectedRows selectedRowNames selectedColumnNames selectedRowsAsArrays df |
 
 	rows := self asArrayOfRows.
 	selectedRows := rows select: aBlock.
-	selectedRowNames := selectedRows collect: #name.
+	selectedRowNames := (selectedRows isEmpty)
+		ifTrue: [ #() ] ifFalse: [ selectedRows collect: #name ].
+	selectedColumnNames := (selectedRows isEmpty)
+		ifTrue: [ #() ] ifFalse: [ self columnNames ].
 	selectedRowsAsArrays := selectedRows collect: #asArray.
 
 	df := self class
 		withRows: selectedRowsAsArrays
 		rowNames: selectedRowNames
-		columnNames: self columnNames.
+		columnNames: selectedColumnNames.
 
 	^ df
 ]

src/DataFrame/DataFrameInternal.class.st

@@ -36,14 +36,17 @@ DataFrameInternal class >> withColumns: anArrayOfArrays [
 
 	| numberOfRows numberOfColumns dfInternal |
 
-	numberOfRows := anArrayOfArrays first size.
 	numberOfColumns := anArrayOfArrays size.
+	numberOfRows := 0.
+	1 to: numberOfColumns do: [ :index |
+		numberOfRows := numberOfRows max: (anArrayOfArrays at: index) size
+		].
 	dfInternal := self new: (numberOfRows @ numberOfColumns ).
 
 	1 to: numberOfRows do: [ :i |
 		1 to: numberOfColumns do: [ :j |
-			dfInternal at: i at: j put:
-				((anArrayOfArrays at: j) asArray at: i) ] ].
+			dfInternal at: i at: j
+				put: ((anArrayOfArrays at: j) asArray at: i ifAbsent: nil) ] ].
 
 	^ dfInternal
 ]
@@ -54,14 +57,17 @@ DataFrameInternal class >> withRows: anArrayOfArrays [
 	| numberOfRows numberOfColumns dfInternal |
 
 	numberOfRows := anArrayOfArrays size.
-	numberOfColumns := anArrayOfArrays first size.
-	dfInternal := self new: (numberOfRows @ numberOfColumns ).
+	numberOfColumns := 0.
+	1 to: numberOfRows do: [ :index |
+		numberOfColumns := numberOfColumns max: (anArrayOfArrays at: index) size
+		].
+	dfInternal := self new: numberOfRows @ numberOfColumns.
+
+	1 to: numberOfRows do: [ :i | 
+		1 to: numberOfColumns do: [ :j | 
+			dfInternal at: i at: j
+				put: ((anArrayOfArrays at: i) asArray at: j ifAbsent: nil) ] ].
 
-	1 to: numberOfRows do: [ :i |
-		1 to: numberOfColumns do: [ :j |
-			dfInternal at: i at: j put:
-				((anArrayOfArrays at: i) asArray at: j) ] ].
-
 	^ dfInternal
 ]