feat: Describe schemas in the DSL (#13) and Tool to infer initial sch…

…ema of dataset (#12)
Safe-DS · Jun 22, 2022 · fdc0c10 · fdc0c10
1 parent 6c255b0
commit fdc0c10
Show file tree

Hide file tree

Showing 36 changed files with 576 additions and 52 deletions.
diff --git a/DSL/build.gradle.kts b/DSL/build.gradle.kts
@@ -158,7 +158,7 @@ tasks.register("generateXtextLanguage") {
 
                 standardLanguage {
                     setName("com.larsreimann.safeds.SafeDS")
-                    setFileExtensions("sdsflow,sdsstub,sdstest")
+                    setFileExtensions("sdsflow,sdsschema,sdsstub,sdstest")
                     addReferencedResource("platform:/resource/com.larsreimann.safeds/model/SafeDS.genmodel")
 
                     setFormatter(

diff --git a/DSL/com.larsreimann.safeds.vscode/package.json b/DSL/com.larsreimann.safeds.vscode/package.json
@@ -30,6 +30,7 @@
                 ],
                 "extensions": [
                     ".sdsflow",
+                    ".sdsschema",
                     ".sdsstub",
                     ".sdstest"
                 ],

diff --git a/DSL/com.larsreimann.safeds/build.gradle.kts b/DSL/com.larsreimann.safeds/build.gradle.kts
@@ -97,6 +97,8 @@ dependencies {
     api(platform("org.eclipse.xtext:xtext-dev-bom:$xtextVersion"))
     implementation("org.eclipse.xtext:org.eclipse.xtext:$xtextVersion")
 
+    implementation("tech.tablesaw:tablesaw-core:0.43.1")
+
     testRuntimeOnly("org.junit.jupiter:junit-jupiter-engine")
     testImplementation("org.junit.jupiter:junit-jupiter-api:5.8.2")
     testImplementation("org.junit.jupiter:junit-jupiter-params:5.8.2")

diff --git a/DSL/com.larsreimann.safeds/model/SafeDS.ecore b/DSL/com.larsreimann.safeds/model/SafeDS.ecore
@@ -33,7 +33,7 @@
         <eStructuralFeatures xsi:type="ecore:EReference" name="type" eType="#//SdsAbstractType" containment="true" />
     </eClassifiers>
     <!-- Class -->
-    <eClassifiers xsi:type="ecore:EClass" name="SdsClass" abstract="false" eSuperTypes="#//SdsAbstractCallable #//SdsAbstractClassMember #//SdsAbstractCompilationUnitMember #//SdsAbstractNamedTypeDeclaration">
+    <eClassifiers xsi:type="ecore:EClass" name="SdsClass" eSuperTypes="#//SdsAbstractCallable #//SdsAbstractClassMember #//SdsAbstractCompilationUnitMember #//SdsAbstractNamedTypeDeclaration">
         <eStructuralFeatures xsi:type="ecore:EReference" name="typeParameterList" eType="#//SdsTypeParameterList" containment="true" />
         <eStructuralFeatures xsi:type="ecore:EReference" name="parentTypeList" eType="#//SdsParentTypeList" containment="true" />
         <eStructuralFeatures xsi:type="ecore:EReference" name="body" eType="#//SdsClassBody" containment="true" />
@@ -183,6 +183,17 @@
     <eClassifiers xsi:type="ecore:EClass" name="SdsFunctionBody" eSuperTypes="#//SdsAbstractObject">
         <eStructuralFeatures xsi:type="ecore:EReference" name="statements" upperBound="-1" eType="#//SdsAbstractObject" containment="true" />
     </eClassifiers>
+    <!-- Schema -->
+    <eClassifiers xsi:type="ecore:EClass" name="SdsSchema" eSuperTypes="#//SdsAbstractCompilationUnitMember">
+        <eStructuralFeatures xsi:type="ecore:EReference" name="columnList" eType="#//SdsColumnList" containment="true" />
+    </eClassifiers>
+    <eClassifiers xsi:type="ecore:EClass" name="SdsColumnList" eSuperTypes="#//SdsAbstractObject">
+        <eStructuralFeatures xsi:type="ecore:EReference" name="columns" upperBound="-1" eType="#//SdsColumn" containment="true" />
+    </eClassifiers>
+    <eClassifiers xsi:type="ecore:EClass" name="SdsColumn" eSuperTypes="#//SdsAbstractObject">
+        <eStructuralFeatures xsi:type="ecore:EReference" name="columnName" eType="#//SdsString" containment="true" />
+        <eStructuralFeatures xsi:type="ecore:EReference" name="columnType" eType="#//SdsAbstractType" containment="true" />
+    </eClassifiers>
     <!-- Step -->
     <eClassifiers xsi:type="ecore:EClass" name="SdsStep" eSuperTypes="#//SdsAbstractCallable #//SdsAbstractCompilationUnitMember">
         <eStructuralFeatures xsi:type="ecore:EAttribute" name="visibility" eType="ecore:EDataType http://www.eclipse.org/emf/2002/Ecore#//EString" />

diff --git a/DSL/com.larsreimann.safeds/model/SafeDS.genmodel b/DSL/com.larsreimann.safeds/model/SafeDS.genmodel
@@ -180,6 +180,17 @@
         <genClasses ecoreClass="SafeDS.ecore#//SdsFunctionBody">
             <genFeatures property="None" children="true" createChild="true" ecoreFeature="ecore:EReference SafeDS.ecore#//SdsFunctionBody/statements" />
         </genClasses>
+        <!-- Schema -->
+        <genClasses ecoreClass="SafeDS.ecore#//SdsSchema">
+             <genFeatures property="None" children="true" createChild="true" ecoreFeature="ecore:EReference SafeDS.ecore#//SdsSchema/columnList" />
+        </genClasses>
+        <genClasses ecoreClass="SafeDS.ecore#//SdsColumnList">
+             <genFeatures property="None" children="true" createChild="true" ecoreFeature="ecore:EReference SafeDS.ecore#//SdsColumnList/columns" />
+        </genClasses>
+        <genClasses ecoreClass="SafeDS.ecore#//SdsColumn">
+             <genFeatures property="None" children="true" createChild="true" ecoreFeature="ecore:EReference SafeDS.ecore#//SdsColumn/columnName" />
+             <genFeatures property="None" children="true" createChild="true" ecoreFeature="ecore:EReference SafeDS.ecore#//SdsColumn/columnType" />
+        </genClasses>
         <!-- Step -->
         <genClasses ecoreClass="SafeDS.ecore#//SdsStep">
             <genFeatures createChild="false" ecoreFeature="ecore:EAttribute SafeDS.ecore#//SdsStep/visibility" />

diff --git a/DSL/com.larsreimann.safeds/src/main/kotlin/com/larsreimann/safeds/SafeDS.xtext b/DSL/com.larsreimann.safeds/src/main/kotlin/com/larsreimann/safeds/SafeDS.xtext
@@ -49,6 +49,16 @@ SdsCompilationUnitMember returns SdsAbstractAnnotatedObject
           resultList=SdsResultList?
           body=SdsFunctionBody?
 
+    | {SdsPredicate.annotationCallList=current}
+        'predicate' name=ID
+        parameterList=SdsParameterList
+        resultList=SdsResultList?
+        goalList=SdsGoalList
+
+    | {SdsSchema.annotationCallList=current}
+        'schema' name=ID
+        columnList=SdsColumnList
+
     | {SdsStep.annotationCallList=current}
           visibility=('internal'|'private')?
           'step' name=ID
@@ -59,12 +69,6 @@ SdsCompilationUnitMember returns SdsAbstractAnnotatedObject
     | {SdsWorkflow.annotationCallList=current}
           'workflow' name=ID
           body=SdsBlock
-
-    | {SdsPredicate.annotationCallList=current}
-          'predicate' name=ID
-          parameterList=SdsParameterList
-          resultList=SdsResultList?
-          goalList=SdsGoalList
     )
     ;
 
@@ -516,7 +520,7 @@ SdsTemplateStringEnd
 
 
 /**********************************************************************************************************************
- * Predicates
+ * Predicates / Goals
  **********************************************************************************************************************/
 
 SdsGoalList
@@ -576,6 +580,19 @@ SdsParenthesizedGoalExpression
     ;
 
 
+/**********************************************************************************************************************
+ * Schemas
+ **********************************************************************************************************************/
+
+SdsColumnList
+: {SdsColumnList} '{' ( columns+=SdsColumn (',' columns+=SdsColumn)* )? '}'
+;
+
+SdsColumn
+: columnName=SdsString ":" columnType=SdsType
+;
+
+
 /**********************************************************************************************************************
  * Names
  **********************************************************************************************************************/

diff --git a/...om.larsreimann.safeds/src/main/kotlin/com/larsreimann/safeds/constant/SdsFileExtension.kt b/...om.larsreimann.safeds/src/main/kotlin/com/larsreimann/safeds/constant/SdsFileExtension.kt
@@ -18,6 +18,14 @@ enum class SdsFileExtension(val extension: String) {
      */
     Flow("sdsflow"),
 
+    /**
+     * Marks the file as a schema file.
+     *
+     * @see isInSchemaFile
+     * @see isSchemaFile
+     */
+    Schema("sdsschema"),
+
     /**
      * Marks the file as a stub file, which describes an external API.
      *
@@ -45,6 +53,11 @@ enum class SdsFileExtension(val extension: String) {
  */
 fun EObject.isInFlowFile() = this.eResource().isFlowFile()
 
+/**
+ * Returns whether the object is contained in schema file.
+ */
+fun EObject.isInSchemaFile() = this.eResource().isSchemaFile()
+
 /**
  * Returns whether the object is contained in stub file.
  */
@@ -60,6 +73,11 @@ fun EObject.isInTestFile() = this.eResource().isTestFile()
  */
 fun Resource.isFlowFile() = this.hasExtension(SdsFileExtension.Flow)
 
+/**
+ * Returns whether the resource represents a schema file.
+ */
+fun Resource.isSchemaFile() = this.hasExtension(SdsFileExtension.Schema)
+
 /**
  * Returns whether the resource represents a stub file.
  */

diff --git a/DSL/com.larsreimann.safeds/src/main/kotlin/com/larsreimann/safeds/constant/SdsKind.kt b/DSL/com.larsreimann.safeds/src/main/kotlin/com/larsreimann/safeds/constant/SdsKind.kt
@@ -29,3 +29,5 @@ fun SdsTypeParameter.kind(): SdsKind {
     return SdsKind.values().firstOrNull { it.kind == this.kind }
         ?: throw IllegalArgumentException("Unknown kind '$kind'.")
 }
+
+fun SdsTypeParameter.hasSchemaKind() = this.kind() == SdsKind.SchemaKind
diff --git a/DSL/com.larsreimann.safeds/src/main/kotlin/com/larsreimann/safeds/emf/Creators.kt b/DSL/com.larsreimann.safeds/src/main/kotlin/com/larsreimann/safeds/emf/Creators.kt
@@ -44,6 +44,7 @@ import com.larsreimann.safeds.safeDS.SdsBoolean
 import com.larsreimann.safeds.safeDS.SdsCall
 import com.larsreimann.safeds.safeDS.SdsCallableType
 import com.larsreimann.safeds.safeDS.SdsClass
+import com.larsreimann.safeds.safeDS.SdsColumn
 import com.larsreimann.safeds.safeDS.SdsCompilationUnit
 import com.larsreimann.safeds.safeDS.SdsConstraint
 import com.larsreimann.safeds.safeDS.SdsEnum
@@ -90,6 +91,7 @@ import com.larsreimann.safeds.safeDS.SdsProtocolTokenClass
 import com.larsreimann.safeds.safeDS.SdsReference
 import com.larsreimann.safeds.safeDS.SdsResult
 import com.larsreimann.safeds.safeDS.SdsResultList
+import com.larsreimann.safeds.safeDS.SdsSchema
 import com.larsreimann.safeds.safeDS.SdsStarProjection
 import com.larsreimann.safeds.safeDS.SdsStep
 import com.larsreimann.safeds.safeDS.SdsString
@@ -548,6 +550,19 @@ fun createSdsConstraint(goals: List<SdsAbstractConstraintGoal>): SdsConstraint {
     }
 }
 
+/**
+ * Returns a new object of class [SdsColumn].
+ */
+fun createSdsColumn(
+    columnName: SdsString,
+    columnType: SdsAbstractType
+): SdsColumn {
+    return factory.createSdsColumn().apply {
+        this.columnName = columnName
+        this.columnType = columnType
+    }
+}
+
 /**
  * Returns a new object of class [SdsEnum].
  */
@@ -1295,6 +1310,48 @@ fun createSdsStarProjection(): SdsStarProjection {
     return factory.createSdsStarProjection()
 }
 
+/**
+ * Returns a new object of class [SdsSchema].
+ */
+fun createSdsSchema(
+    name: String,
+    annotationCalls: List<SdsAnnotationCall> = emptyList(),
+    columns: List<SdsColumn> = emptyList()
+): SdsSchema {
+    return factory.createSdsSchema().apply {
+        this.name = name
+        this.annotationCallList = createSdsAnnotationCallList(annotationCalls)
+        columns.forEach { addColumn(it) }
+    }
+}
+
+/**
+ * Adds a new object of class [SdsSchema] to the receiver.
+ */
+fun SdsCompilationUnit.SdsSchema(
+    name: String,
+    annotationCalls: List<SdsAnnotationCall> = emptyList(),
+    columns: List<SdsColumn> = emptyList()
+) {
+    this.addMember(
+        createSdsSchema(
+            name,
+            annotationCalls,
+            columns
+        )
+    )
+}
+
+/**
+ * Adds a new column to the receiver.
+ */
+private fun SdsSchema.addColumn(column: SdsColumn) {
+    if (this.columnList == null) {
+        this.columnList = factory.createSdsColumnList()
+    }
+    this.columnList.columns += column
+}
+
 /**
  * Returns a new object of class [SdsStep].
  */

diff --git a/DSL/com.larsreimann.safeds/src/main/kotlin/com/larsreimann/safeds/emf/SimpleShortcuts.kt b/DSL/com.larsreimann.safeds/src/main/kotlin/com/larsreimann/safeds/emf/SimpleShortcuts.kt
@@ -7,6 +7,7 @@
 
 package com.larsreimann.safeds.emf
 
+import com.larsreimann.safeds.constant.hasSchemaKind
 import com.larsreimann.safeds.safeDS.SdsAbstractAssignee
 import com.larsreimann.safeds.safeDS.SdsAbstractCallable
 import com.larsreimann.safeds.safeDS.SdsAbstractClassMember
@@ -39,6 +40,8 @@ import com.larsreimann.safeds.safeDS.SdsEnumVariant
 import com.larsreimann.safeds.safeDS.SdsExpressionLambda
 import com.larsreimann.safeds.safeDS.SdsFunction
 import com.larsreimann.safeds.safeDS.SdsFunctionBody
+import com.larsreimann.safeds.safeDS.SdsGoalArgument
+import com.larsreimann.safeds.safeDS.SdsGoalCall
 import com.larsreimann.safeds.safeDS.SdsImport
 import com.larsreimann.safeds.safeDS.SdsNamedType
 import com.larsreimann.safeds.safeDS.SdsParameter
@@ -177,6 +180,10 @@ fun SdsClass?.typeParametersOrEmpty(): List<SdsTypeParameter> {
     return this?.typeParameterList?.typeParameters.orEmpty()
 }
 
+fun SdsClass?.hasSchema(): Boolean {
+    return this?.typeParametersOrEmpty()?.any { it.hasSchemaKind() } ?: false
+}
+
 fun SdsClass?.parentTypesOrEmpty(): List<SdsAbstractType> {
     return this?.parentTypeList?.parentTypes.orEmpty()
 }
@@ -247,6 +254,12 @@ fun SdsFunction?.constraintsOrEmpty(): List<SdsAbstractConstraintGoal> {
         .orEmpty()
 }
 
+// SdsGoalCall -------------------------------------------------------------------------------------
+
+fun SdsGoalCall?.argumentsOrEmpty(): List<SdsGoalArgument> {
+    return this?.argumentList?.arguments.orEmpty()
+}
+
 // SdsImport ---------------------------------------------------------------------------------------
 
 fun SdsImport.aliasNameOrNull(): String? {

diff --git a/....larsreimann.safeds/src/main/kotlin/com/larsreimann/safeds/formatting2/SafeDSFormatter.kt b/....larsreimann.safeds/src/main/kotlin/com/larsreimann/safeds/formatting2/SafeDSFormatter.kt
@@ -37,6 +37,8 @@ import com.larsreimann.safeds.safeDS.SdsCall
 import com.larsreimann.safeds.safeDS.SdsCallableType
 import com.larsreimann.safeds.safeDS.SdsClass
 import com.larsreimann.safeds.safeDS.SdsClassBody
+import com.larsreimann.safeds.safeDS.SdsColumn
+import com.larsreimann.safeds.safeDS.SdsColumnList
 import com.larsreimann.safeds.safeDS.SdsCompilationUnit
 import com.larsreimann.safeds.safeDS.SdsConstraint
 import com.larsreimann.safeds.safeDS.SdsEnum
@@ -80,6 +82,7 @@ import com.larsreimann.safeds.safeDS.SdsProtocolSubterm
 import com.larsreimann.safeds.safeDS.SdsProtocolSubtermList
 import com.larsreimann.safeds.safeDS.SdsResult
 import com.larsreimann.safeds.safeDS.SdsResultList
+import com.larsreimann.safeds.safeDS.SdsSchema
 import com.larsreimann.safeds.safeDS.SdsStep
 import com.larsreimann.safeds.safeDS.SdsTemplateString
 import com.larsreimann.safeds.safeDS.SdsTypeArgument
@@ -901,6 +904,67 @@ class SafeDSFormatter : AbstractFormatter2() {
                 }
             }
 
+           /**********************************************************************************************************
+             * Schema
+             **********************************************************************************************************/
+
+            is SdsSchema -> {
+
+                // Features "annotations"
+                doc.formatAnnotations(obj)
+
+                // Keyword "schema"
+                if (obj.annotationCallsOrEmpty().isEmpty()) {
+                    doc.formatKeyword(obj, "schema", noSpace, oneSpace)
+                } else {
+                    doc.formatKeyword(obj, "schema", oneSpace, oneSpace)
+                }
+
+                // Feature "name"
+                doc.formatFeature(obj, SDS_ABSTRACT_DECLARATION__NAME, null, oneSpace)
+
+                // EObject "columnList"
+                doc.formatObject(obj.columnList, oneSpace, null)
+            }
+            is SdsColumnList -> {
+
+                // Keyword "{"
+                val openingBrace = obj.regionForKeyword("{")
+                if (obj.columns.isEmpty()) {
+                    doc.append(openingBrace, noSpace)
+                } else {
+                    doc.append(openingBrace, newLine)
+                }
+
+                // Feature "columns"
+                obj.columns.forEach {
+                    doc.formatObject(it, newLine, noSpace)
+                }
+
+                // Keywords ","
+                doc.formatKeyword(obj, ",", noSpace, newLine)
+
+                // Keyword "}"
+                val closingBrace = obj.regionForKeyword("}")
+                if (obj.columns.isEmpty()) {
+                    doc.prepend(closingBrace, noSpace)
+                } else {
+                    doc.prepend(closingBrace, newLine)
+                }
+                doc.interior(openingBrace, closingBrace, indent)
+            }
+            is SdsColumn -> {
+
+                // EObject "columnName"
+                doc.formatObject(obj.columnName, null, oneSpace)
+
+                // Keyword ":"
+                doc.formatKeyword(obj, ":", oneSpace, oneSpace)
+
+                // EObject "columnType"
+                doc.formatObject(obj.columnType)
+            }
+
             /**********************************************************************************************************
              * Statements
              **********************************************************************************************************/