Json support

src/BaselineOfDataFrame/BaselineOfDataFrame.class.st

@@ -35,14 +35,17 @@ BaselineOfDataFrame >> baseline: spec [
 		spec
 			baseline: 'NeoCSV'
 			with: [ spec repository: 'github://svenvc/NeoCSV/repository' ].
+		spec
+			baseline: 'NeoJSON'
+			with: [ spec repository: 'github://svenvc/NeoJSON/repository' ].
 
 		"Packages"
 		spec 
 			package: 'DataFrame';
 			package: 'DataFrame-Tests' with: [ spec requires: #('DataFrame') ];
 			package: 'DataFrame-Type' with: [ spec requires: #('DataFrame') ];
 			package: 'DataFrame-Type-Tests' with: [ spec requires: #('DataFrame-Type') ];
-			package: 'DataFrame-IO' with: [ spec requires: #('DataFrame' 'DataFrame-Type' 'NeoCSV') ];
+			package: 'DataFrame-IO' with: [ spec requires: #('DataFrame' 'DataFrame-Type' 'NeoCSV' 'NeoJSON') ];
 			package: 'DataFrame-IO-Tests' with: [ spec requires: #('DataFrame-IO') ]].
 
 		spec

src/DataFrame-IO-Tests/DataFrameJsonReaderTest.class.st

@@ -0,0 +1,148 @@
+Class {
+	#name : #DataFrameJsonReaderTest,
+	#superclass : #TestCase,
+	#instVars : [
+		'directory',
+		'df',
+		'dfWithColNames',
+		'dfWithRowNames',
+		'dfWithRowColNames'
+	],
+	#category : #'DataFrame-IO-Tests'
+}
+
+{ #category : #running }
+DataFrameJsonReaderTest >> createFileNamed: name withContents: aString [
+	| stream |
+	stream := (directory / name) writeStream.
+	stream nextPutAll: aString.
+	stream close.
+]
+
+{ #category : #running }
+DataFrameJsonReaderTest >> setUp [
+	super setUp.
+	directory := FileSystem memory workingDirectory / 'testDataForJson'.
+	directory createDirectory.
+
+	self createFileNamed: 'column.json' withContents: TestJsonStrings columnJsonString.
+	self createFileNamed: 'index.json' withContents: TestJsonStrings indexJsonString.
+	self createFileNamed: 'nonNull1.json' withContents: TestJsonStrings nonNullJsonString1.
+	self createFileNamed: 'records.json' withContents: TestJsonStrings recordsJsonString.
+	self createFileNamed: 'split.json' withContents: TestJsonStrings splitJsonString.
+	self createFileNamed: 'values.json' withContents: TestJsonStrings valuesJsonString.
+
+	df := DataFrame withRows: #(
+		(1 2 nil nil)
+		(nil 2 3 nil)
+		(nil nil nil 5)
+		(1 nil 2 nil)).
+
+	dfWithColNames := df deepCopy.
+	dfWithColNames columnNames: #('col1' 'col2' 'col3' 'col4').
+
+	dfWithRowNames := df deepCopy.
+	dfWithRowNames rowNames: #('row1' 'row2' 'row3' 'row4').
+
+	dfWithRowColNames := df deepCopy.
+	dfWithRowColNames columnNames: #('col1' 'col2' 'col3' 'col4').
+	dfWithRowColNames rowNames: #('row1' 'row2' 'row3' 'row4').
+
+
+
+
+]
+
+{ #category : #running }
+DataFrameJsonReaderTest >> sortByRowColNames: inputDf [
+	"Sorts df according to column and row names"
+
+	| sortedRowNames sortedColNames dfColSorted outputDf |
+	sortedRowNames := inputDf rowNames sorted.
+	sortedColNames := inputDf columnNames sorted.
+
+	dfColSorted := DataFrame withRowNames: inputDf rowNames.
+	sortedColNames do: [ :col | dfColSorted addColumn: (inputDf column: col) asArray named: col ].
+
+	outputDf := DataFrame withColumnNames: dfColSorted columnNames.
+	sortedRowNames do: [ :row | outputDf addRow: (dfColSorted row: row) asArray named: row ].
+
+	^ outputDf
+]
+
+{ #category : #tests }
+DataFrameJsonReaderTest >> testReadFrom [
+	| output |
+
+	output := DataFrameJsonReader new readFrom: directory / 'nonNull1.json'.
+
+	self assert: (self sortByRowColNames: output) equals: dfWithColNames.
+]
+
+{ #category : #tests }
+DataFrameJsonReaderTest >> testReadFromColumns [
+	| output |
+
+	output := DataFrameJsonReader new readFrom: directory / 'column.json'.
+
+	self assert: (self sortByRowColNames: output) equals: dfWithRowColNames.
+]
+
+{ #category : #tests }
+DataFrameJsonReaderTest >> testReadFromIndex [
+	| output |
+
+	output := DataFrameJsonReader new readFrom: directory / 'index.json'.
+
+	self assert: (self sortByRowColNames: output) equals: dfWithRowColNames transposed.
+]
+
+{ #category : #tests }
+DataFrameJsonReaderTest >> testReadFromJson [
+	| output |
+	output := DataFrame readFromJson: (directory / 'nonNull1.json').
+	self assert: (self sortByRowColNames: output) equals: dfWithColNames.
+]
+
+{ #category : #tests }
+DataFrameJsonReaderTest >> testReadFromJsonOrient [
+	| output |
+	output := DataFrame readFromJson: (directory / 'split.json') orient: 'split'.
+	self assert: (self sortByRowColNames: output) equals: dfWithRowColNames.
+]
+
+{ #category : #tests }
+DataFrameJsonReaderTest >> testReadFromRecords [
+	| output |
+
+	output := DataFrameJsonReader new readFrom: directory / 'records.json'.
+
+	self assert: (self sortByRowColNames: output) equals: dfWithColNames.
+]
+
+{ #category : #tests }
+DataFrameJsonReaderTest >> testReadFromSplit [
+	| output |
+
+	output := DataFrameJsonReader new readFrom: directory / 'split.json'.
+
+	self assert: (self sortByRowColNames: output) equals: dfWithRowColNames.
+]
+
+{ #category : #tests }
+DataFrameJsonReaderTest >> testReadFromString [
+	| output |
+
+	output := DataFrameJsonReader new readFrom: TestJsonStrings nonNullJsonString2.
+
+	self assert: (self sortByRowColNames: output) equals: dfWithRowColNames.
+]
+
+{ #category : #tests }
+DataFrameJsonReaderTest >> testReadFromValues [
+	| output |
+
+	output := DataFrameJsonReader new readFrom: directory / 'values.json'.
+
+	self assert: (self sortByRowColNames: output) equals: df.
+]

src/DataFrame-IO-Tests/DataFrameJsonWriterTest.class.st

@@ -0,0 +1,75 @@
+Class {
+	#name : #DataFrameJsonWriterTest,
+	#superclass : #TestCase,
+	#instVars : [
+		'directory',
+		'df'
+	],
+	#category : #'DataFrame-IO-Tests'
+}
+
+{ #category : #running }
+DataFrameJsonWriterTest >> readFile: aFileReference [
+	| stream contents |
+	stream := aFileReference readStream.
+	contents := stream upToEnd.
+	stream close.
+	^ contents
+]
+
+{ #category : #running }
+DataFrameJsonWriterTest >> setUp [
+	super setUp.
+	directory := FileSystem memory workingDirectory / 'testDataForJson'.
+	directory createDirectory.
+
+	df := DataFrame withRows: #(
+		(1 2 nil nil)
+		(nil 2 3 nil)
+		(nil nil nil 5)
+		(1 nil 2 nil))
+	rowNames: #('row1' 'row2' 'row3' 'row4')
+	columnNames: #('col1' 'col2' 'col3' 'col4').
+]
+
+{ #category : #tests }
+DataFrameJsonWriterTest >> testWriteAsString [
+	| actual expected |.
+	actual := DataFrameJsonWriter new writeAsString: df.
+	expected := '' join: (TestJsonStrings recordsJsonString 
+		regex: '[^\s]+' matchesCollect: [:x|x]).
+	self assert: actual equals: expected.
+]
+
+{ #category : #tests }
+DataFrameJsonWriterTest >> testWriteTo [
+	| file actual expected |
+	file := directory / 'output.json'.
+	DataFrameJsonWriter new write: df to: file.
+	actual := self readFile: file.
+	expected := '' join: (TestJsonStrings recordsJsonString 
+		regex: '[^\s]+' matchesCollect: [:x|x]).
+	self assert: actual lines equals: expected lines.
+]
+
+{ #category : #tests }
+DataFrameJsonWriterTest >> testWriteToJson [
+	| file actual expected |
+	file := directory / 'output.json'.
+	df writeToJson: file.
+	actual := self readFile: file.
+	expected := '' join: (TestJsonStrings recordsJsonString 
+		regex: '[^\s]+' matchesCollect: [:x|x]).
+	self assert: actual lines equals: expected lines.
+]
+
+{ #category : #tests }
+DataFrameJsonWriterTest >> testWriteToJsonOrient [
+	| file actual expected |
+	file := directory / 'output.json'.
+	df writeToJson: file orient: 'values'.
+	actual := self readFile: file.
+	expected := '' join: (TestJsonStrings valuesJsonString 
+		regex: '[^\s]+' matchesCollect: [:x|x]).
+	self assert: actual lines equals: expected lines.
+]

src/DataFrame-IO-Tests/TestJsonStrings.class.st

@@ -0,0 +1,60 @@
+Class {
+	#name : #TestJsonStrings,
+	#superclass : #Object,
+	#category : #'DataFrame-IO-Tests'
+}
+
+{ #category : #running }
+TestJsonStrings class >> columnJsonString [
+	^ '{"col1":{"row1":1.0,"row2":null,"row3":null,"row4":1.0},
+"col2":{"row1":2.0,"row2":2.0,"row3":null,"row4":null},
+"col3":{"row1":null,"row2":3.0,"row3":null,"row4":2.0},
+"col4":{"row1":null,"row2":null,"row3":5.0,"row4":null}}'
+]
+
+{ #category : #running }
+TestJsonStrings class >> indexJsonString [
+	^ '{"row1":{"col1":1.0,"col2":2.0,"col3":null,"col4":null},
+"row2":{"col1":null,"col2":2.0,"col3":3.0,"col4":null},
+"row3":{"col1":null,"col2":null,"col3":null,"col4":5.0},
+"row4":{"col1":1.0,"col2":null,"col3":2.0,"col4":null}}'
+]
+
+{ #category : #running }
+TestJsonStrings class >> nonNullJsonString1 [
+	^ '[{"col1":1, "col2":2},
+		 {"col3":3, "col2":2},
+		 {"col4":5},
+		 {"col1":1, "col3":2}]'
+]
+
+{ #category : #running }
+TestJsonStrings class >> nonNullJsonString2 [
+	^ '{
+			"col1": {"row1":1, "row4":1},
+			"col2": {"row1":2, "row2":2},
+			"col3": {"row2":3, "row4":2},
+			"col4": {"row3":5}
+		}'
+]
+
+{ #category : #running }
+TestJsonStrings class >> recordsJsonString [
+	^ '[{"col1":1,"col2":2,"col3":null,"col4":null},
+{"col1":null,"col2":2,"col3":3,"col4":null},
+{"col1":null,"col2":null,"col3":null,"col4":5},
+{"col1":1,"col2":null,"col3":2,"col4":null}]'
+]
+
+{ #category : #running }
+TestJsonStrings class >> splitJsonString [
+	^ '{"columns":["col1","col2","col3","col4"],
+"index":["row1","row2","row3","row4"],
+"data":[[1.0,2.0,null,null],[null,2.0,3.0,null],[null,null,null,5.0],[1.0,null,2.0,null]]}'
+]
+
+{ #category : #running }
+TestJsonStrings class >> valuesJsonString [
+	^ '[[1,2,null,null],[null,2,3,null],
+[null,null,null,5],[1,null,2,null]]'
+]

src/DataFrame-IO/DataFrame.extension.st

@@ -49,6 +49,21 @@ DataFrame class >> readFromCsvWithRowNames: aFileReference separator: aSeparator
 	^ self readFrom: aFileReference using: reader
 ]
 
+{ #category : #'*DataFrame-IO' }
+DataFrame class >> readFromJson: aFileReference [ 
+	| reader |
+	reader := DataFrameJsonReader new.
+	^ self readFrom: aFileReference using: reader.
+]
+
+{ #category : #'*DataFrame-IO' }
+DataFrame class >> readFromJson: aFileReference orient: orient [
+	| reader |
+	reader := DataFrameJsonReader new.
+	reader orient: orient.
+	^ self readFrom: aFileReference using: reader.
+]
+
 { #category : #'*DataFrame-IO' }
 DataFrame >> writeTo: aLocation using: aDataFrameWriter [
 	"Write data frame to a given location using a given DataFrameWriter. Location can be a file reference, a database connection, or something else (depending on the implementation of the writer)"
@@ -69,3 +84,35 @@ DataFrame >> writeToCsv: aFileReference withSeparator: aSeparator [
 	writer separator: aSeparator.
 	self writeTo: aFileReference using: writer.
 ]
+
+{ #category : #'*DataFrame-IO' }
+DataFrame >> writeToJson: aFileReference [
+	| writer |
+	writer := DataFrameJsonWriter new.
+	self writeTo: aFileReference using: writer.
+]
+
+{ #category : #'*DataFrame-IO' }
+DataFrame >> writeToJson: aFileReference orient: orient [
+	| writer |
+	writer := DataFrameJsonWriter new.
+	writer orient: orient.
+	self writeTo: aFileReference using: writer.
+]
+
+{ #category : #'*DataFrame-IO' }
+DataFrame >> writeToPrettyJson: aFileReference [
+	| writer |
+	writer := DataFrameJsonWriter new.
+	writer pretty: true.
+	self writeTo: aFileReference using: writer.
+]
+
+{ #category : #'*DataFrame-IO' }
+DataFrame >> writeToPrettyJson: aFileReference orient: orient [
+	| writer |
+	writer := DataFrameJsonWriter new.
+	writer orient: orient.
+	writer pretty: true.
+	self writeTo: aFileReference using: writer.
+]