Introducing type induction for data frame

olekscode · olekscode · commit 4dab9a265531 · 2017-08-06T00:24:08.000+03:00
diff --git a/DataFrame-Core.package/DataFrame.class/instance/^equals.st b/DataFrame-Core.package/DataFrame.class/instance/^equals.st
@@ -8,6 +8,10 @@ comparing
 	"This is the fastest way for two data frames with different dimensions"
 	aDataFrame dimensions = self dimensions
 		ifFalse: [ ^ false ].
+		
+	"If column types are different, we don't need to iterate through row and column names"
+	aDataFrame columnTypes = self columnTypes
+		ifFalse: [ ^ false ].
 	
 	"If the names are different we don't need to iterate through values"	
 	(aDataFrame rowNames = self rowNames
diff --git a/DataFrame-Core.package/DataFrame.class/instance/at.at.put..st b/DataFrame-Core.package/DataFrame.class/instance/at.at.put..st
@@ -1,4 +1,12 @@
 accessing
 at: rowNumber at: columnNumber put: value
 
-	contents at: rowNumber at: columnNumber put: value.
+	| typeInductor |
+	
+	contents at: rowNumber at: columnNumber put: value.
+
+	typeInductor := DataTypeInductor new.
+	columnTypes at: columnNumber put:
+		(typeInductor
+			leastCommonSuperclassOf: value class
+			and: (columnTypes at: columnNumber)).
diff --git a/DataFrame-Core.package/DataFrame.class/instance/columnAt.put..st b/DataFrame-Core.package/DataFrame.class/instance/columnAt.put..st
@@ -1,7 +1,13 @@
 accessing
 columnAt: aNumber put: anArray
 
+	| typeInductor type |
+	
 	anArray size = self numberOfRows
 		ifFalse: [ SizeMismatch signal ].
 		
-	contents columnAt: aNumber put: anArray.
+	typeInductor := DataTypeInductor new.
+	type := typeInductor typeOf: anArray.	
+	
+	contents columnAt: aNumber put: anArray.
+	columnTypes at: aNumber put: type.
diff --git a/DataFrame-Core.package/DataFrame.class/instance/columnTypes.st b/DataFrame-Core.package/DataFrame.class/instance/columnTypes.st
@@ -0,0 +1,4 @@
+accessing
+columnTypes
+
+	^ columnTypes.
diff --git a/DataFrame-Core.package/DataFrame.class/instance/induceColumnTypes.st b/DataFrame-Core.package/DataFrame.class/instance/induceColumnTypes.st
@@ -0,0 +1,8 @@
+as yet unclassified
+induceColumnTypes
+"Indeces the least common superclass for all the values in each column of a data frame"
+
+	| inductor |
+	inductor := DataTypeInductor new.
+	
+	columnTypes := inductor columnTypesOf: self.
diff --git a/DataFrame-Core.package/DataFrame.class/instance/initialize..st b/DataFrame-Core.package/DataFrame.class/instance/initialize..st
@@ -4,4 +4,5 @@ initialize: aPoint
 	super initialize.
 	
 	contents := DataFrameInternal new: aPoint.
-	self setDefaultRowColumnNames.
+	self setDefaultRowColumnNames.
+	self induceColumnTypes.
diff --git a/DataFrame-Core.package/DataFrame.class/instance/initialize.st b/DataFrame-Core.package/DataFrame.class/instance/initialize.st
@@ -4,4 +4,5 @@ initialize
 	super initialize.
 	
 	contents := DataFrameInternal new.
-	self setDefaultRowColumnNames.
+	self setDefaultRowColumnNames.
+	self induceColumnTypes.
diff --git a/DataFrame-Core.package/DataFrame.class/instance/initializeColumns..st b/DataFrame-Core.package/DataFrame.class/instance/initializeColumns..st
@@ -2,4 +2,5 @@ initialization
 initializeColumns: anArrayOfArrays
 
 	contents := DataFrameInternal fromColumns: anArrayOfArrays.
-	self setDefaultRowColumnNames.
+	self setDefaultRowColumnNames.
+	self induceColumnTypes.
diff --git a/DataFrame-Core.package/DataFrame.class/instance/initializeContents.rowNames.columnNames..st b/DataFrame-Core.package/DataFrame.class/instance/initializeContents.rowNames.columnNames..st
@@ -5,4 +5,5 @@ initializeContents: aDataFrameInternal rowNames: rows columnNames: columns
 	
 	contents := aDataFrameInternal.
 	self rowNames: rows.
-	self columnNames: columns.
+	self columnNames: columns.
+	self induceColumnTypes.
diff --git a/DataFrame-Core.package/DataFrame.class/instance/initializeMatrix..st b/DataFrame-Core.package/DataFrame.class/instance/initializeMatrix..st
@@ -2,4 +2,5 @@ initialization
 initializeMatrix: aMatrix
 
 	contents := DataFrameInternal fromMatrix: aMatrix.
-	self setDefaultRowColumnNames.
+	self setDefaultRowColumnNames.
+	self induceColumnTypes.
diff --git a/DataFrame-Core.package/DataFrame.class/instance/initializeRows..st b/DataFrame-Core.package/DataFrame.class/instance/initializeRows..st
@@ -2,4 +2,5 @@ initialization
 initializeRows: anArrayOfArrays
 
 	contents := DataFrameInternal fromRows: anArrayOfArrays.
-	self setDefaultRowColumnNames.
+	self setDefaultRowColumnNames.
+	self induceColumnTypes.
diff --git a/DataFrame-Core.package/DataFrame.class/instance/rowAt.put..st b/DataFrame-Core.package/DataFrame.class/instance/rowAt.put..st
@@ -1,7 +1,14 @@
 accessing
 rowAt: aNumber put: anArray
 
+	| typeInductor |
 	anArray size = self numberOfColumns
 		ifFalse: [ SizeMismatch signal ].
 		
-	contents rowAt: aNumber put: anArray.
+	contents rowAt: aNumber put: anArray.
+	
+	typeInductor := DataTypeInductor new.
+	columnTypes := (1 to: self numberOfColumns) collect: [ :j |
+		typeInductor
+			leastCommonSuperclassOf: (columnTypes at: j)
+			and: (anArray at: j) class. ].
diff --git a/DataFrame-Core.package/DataTypeInductor.class/instance/columnTypesOf..st b/DataFrame-Core.package/DataTypeInductor.class/instance/columnTypesOf..st
@@ -0,0 +1,13 @@
+as yet unclassified
+columnTypesOf: aDataFrame
+"Induces the column types of a data frame"
+
+	| allClasses |
+	
+	^ (1 to: aDataFrame numberOfColumns) collect: [ :j |
+		allClasses := Set new.
+		
+		1 to: aDataFrame numberOfRows do: [ :i | 
+			allClasses add: (aDataFrame at: i at: j) class ].
+		
+		self leastCommonSuperclassOf: allClasses asArray. ].
diff --git a/DataFrame-Tests.package/DataFrameTests.class/instance/testColumnAtPut.st b/DataFrame-Tests.package/DataFrameTests.class/instance/testColumnAtPut.st
@@ -12,7 +12,7 @@ testColumnAtPut
 	expected columnNames: #(City Population SomeBool).
 	
 	actual := df.
-	actual columnAt: 1 put: #(X Y Z).
+	actual columnAt: 1 put: #('X' 'Y' 'Z').
 	actual columnAt: 3 put: #(0 1 0).
 	
 	self assert: actual equals: expected.
diff --git a/DataFrame-Tests.package/DataFrameTests.class/instance/testColumnPut.st b/DataFrame-Tests.package/DataFrameTests.class/instance/testColumnPut.st
@@ -12,7 +12,7 @@ testColumnPut
 	expected columnNames: #(City Population SomeBool).
 	
 	actual := df.
-	actual column: #City put: #(X Y Z).
+	actual column: #City put: #('X' 'Y' 'Z').
 	actual column: #SomeBool put: #(0 1 0).
 	
 	self assert: actual equals: expected.
diff --git a/DataFrame-Tests.package/DataFrameTests.class/instance/testRowAtPut.st b/DataFrame-Tests.package/DataFrameTests.class/instance/testRowAtPut.st
@@ -12,6 +12,6 @@ testRowAtPut
 	expected columnNames: #(City Population SomeBool).
 	
 	actual := df.
-	actual rowAt: 2 put: #(X Y Z).
+	actual rowAt: 2 put: #('X' 'Y' 'Z').
 	
 	self assert: actual equals: expected.
diff --git a/DataFrame-Tests.package/DataFrameTests.class/instance/testRowPut.st b/DataFrame-Tests.package/DataFrameTests.class/instance/testRowPut.st
@@ -12,6 +12,6 @@ testRowPut
 	expected columnNames: #(City Population SomeBool).
 	
 	actual := df.
-	actual row: #B put: #(X Y Z).
+	actual row: #B put: #('X' 'Y' 'Z').
 	
 	self assert: actual equals: expected.
