Merge pull request #43 from seandenigris/enh_ma-manual-column-names

[Enh]: Magritte Importer - Allow Manual Setting of Column Names
svenvc · Oct 20, 2023 · 0ba1afb · 0ba1afb
2 parents 63cbd24 + 30a5df9
commit 0ba1afb
Show file tree

Hide file tree

Showing 2 changed files with 54 additions and 24 deletions.
diff --git a/repository/Neo-CSV-Magritte/MACSVImporter.class.st b/repository/Neo-CSV-Magritte/MACSVImporter.class.st
@@ -5,11 +5,33 @@ Class {
 		'reader',
 		'readerClass',
 		'recordClass',
-		'source'
+		'source',
+		'columnNames'
 	],
 	#category : #'Neo-CSV-Magritte-Visitors'
 }
 
+{ #category : #accessing }
+MACSVImporter >> columnNames [
+	"See setter comment"
+
+	^ columnNames
+]
+
+{ #category : #accessing }
+MACSVImporter >> columnNames: aCollection [
+	"If not supplied, will assume the first row contains them"
+
+	columnNames := aCollection
+]
+
+{ #category : #private }
+MACSVImporter >> configureReaderFor: aStream [
+	self reader on: aStream.
+	self columnNames ifNil: [ self columnNames: self readHeader ].
+	self reader recordClass: self recordClass.
+]
+
 { #category : #accessing }
 MACSVImporter >> execute [
 	^ self source isStream
@@ -33,23 +55,22 @@ MACSVImporter >> fieldReaderPropertyKey [
 
 { #category : #private }
 MACSVImporter >> importStream: aStream [
+	| fields |
+	self configureReaderFor: aStream.
 
-	| fields header |
-	self reader on: aStream.
-	header := self readHeader.
-	self reader recordClass: self recordClass.
 	fields := self recordClass new magritteDescription children.
-	header do: [ :h | 
-		fields
-			detect: [ :f | 
-					f 
-						propertyAt: self fieldNamePropertyKey 
+	self columnNames
+		do: [ :h | 
+			fields
+				detect: [ :f | 
+					f
+						propertyAt: self fieldNamePropertyKey
 						ifPresent: [ :fieldName | fieldName = h asString trimmed ]
 						ifAbsent: [ false ] ]
-			ifFound: [ :e | 
-				self flag: 'need a way to customize the reader here'.
-				self reader addFieldDescribedByMagritte: e ]
-			ifNone: [ self reader addIgnoredField ] ].
+				ifFound: [ :e | 
+					self flag: 'need a way to customize the reader here'.
+					self reader addFieldDescribedByMagritte: e ]
+				ifNone: [ self reader addIgnoredField ] ].
 	^ self reader upToEnd "or do more processing e.g. `select: [ :record | record lastName isNotNil ]`"
 ]
 

diff --git a/repository/Neo-CSV-Magritte/MACSVTwoStageImporter.class.st b/repository/Neo-CSV-Magritte/MACSVTwoStageImporter.class.st
@@ -8,6 +8,16 @@ Class {
 	#category : #'Neo-CSV-Magritte-Visitors'
 }
 
+{ #category : #accessing }
+MACSVTwoStageImporter >> configureReaderFor: aStream [
+
+	super configureReaderFor: aStream.
+
+	self reader
+		emptyFieldValue: #passNil;
+		addFieldsAt: (self columnNames collect: [ :each | each asSymbol ]) "Adapted from NeoCSVReader>>#namedColumnsConfiguration"
+]
+
 { #category : #accessing }
 MACSVTwoStageImporter >> convertToDomainObjects: aCollectionOfDictionaries [
 	^ aCollectionOfDictionaries
@@ -33,11 +43,18 @@ MACSVTwoStageImporter >> domainObjectFromDictionary: aDictionary [
 { #category : #accessing }
 MACSVTwoStageImporter >> importStream: aStream [
 	| rows |
-	self reader on: aStream.
-	rows := self readInput.
+	self configureReaderFor: aStream.
+	rows := self reader upToEnd.
 	^ self convertToDomainObjects: rows
 ]
 
+{ #category : #accessing }
+MACSVTwoStageImporter >> initialize [
+
+	super initialize.
+	self recordClass: Dictionary.
+]
+
 { #category : #accessing }
 MACSVTwoStageImporter >> initializeDomainObject: anObject fromRecord: aDictionary [
 	"We needed an instance-side version because some objects may need configuration during instance creation"
@@ -101,14 +118,6 @@ MACSVTwoStageImporter >> preprocessor: anObject [
 	preprocessor := anObject
 ]
 
-{ #category : #accessing }
-MACSVTwoStageImporter >> readInput [
-	^ self reader
-		emptyFieldValue: #passNil;
-		namedColumnsConfiguration;
-		upToEnd.
-]
-
 { #category : #accessing }
 MACSVTwoStageImporter >> selectBlock [
 	^ selectBlock ifNil: [ #isNotNil ]