Add Transformation model

sources.yml

@@ -30,6 +30,11 @@ scraper:
             steps:
               - type: "xpath"
                 value: "//li/text()"
+            transform:
+              - name: "replace"
+                parameters:
+                  - " ("
+                  - ""
           language:
             steps:
               - type: "xpath"

src/main/java/com/rak/model/transform/TestRegistry.java

@@ -1,38 +0,0 @@
-package com.rak.model.transform;
-
-import jakarta.enterprise.context.ApplicationScoped;
-
-import java.time.LocalDate;
-import java.util.Map;
-import java.util.concurrent.ConcurrentHashMap;
-import java.util.regex.Matcher;
-import java.util.regex.Pattern;
-
-@ApplicationScoped
-public class TestRegistry {
-    private final Map<String, Transformation> registry = new ConcurrentHashMap<>();
-
-    public TestRegistry() {
-        // Register built-in transformations
-        register("trim", string -> string.trim());
-        register("upper", String::toUpperCase);
-        register("parseInt", s -> Integer.parseInt((String) s));
-        register("parseFloat", s -> Float.parseFloat((String) s));
-        register("parseDate", s -> LocalDate.parse((String) s));
-        register("extract", this::extract);
-    }
-
-    public void register(String name, Transformation transformation) {
-        registry.put(name, transformation);
-    }
-
-    public Transformation get(String name) {
-        return registry.get(name);
-    }
-
-    private Object extract(Object input, String pattern) {
-        Pattern regex = Pattern.compile(pattern);
-        Matcher matcher = regex.matcher((String) input);
-        return matcher.find() ? matcher.group(1) : input;
-    }
-}

src/main/kotlin/com/rak/config/AbstractModelDefinition.kt

@@ -1,6 +1,5 @@
-package com.rak.config.converter
+package com.rak.config
 
-import com.rak.config.Step
 import io.smallrye.config.WithName
 
 interface AbstractModelDefinition {

src/main/kotlin/com/rak/config/RegionalSetDefinition.kt

@@ -1,6 +1,5 @@
 package com.rak.config
 
-import com.rak.config.converter.AbstractModelDefinition
 import io.smallrye.config.WithName
 
 interface RegionalSetDefinition : AbstractModelDefinition {

src/main/kotlin/com/rak/config/SelectorDefinition.kt

@@ -1,5 +1,8 @@
 package com.rak.config
 
+import java.util.*
+
 interface SelectorDefinition {
     fun steps(): Set<Step>
+    fun transform(): Optional<Set<TransformationStep>>
 }

src/main/kotlin/com/rak/config/TransformationStep.kt

@@ -0,0 +1,10 @@
+package com.rak.config
+
+import com.rak.config.converter.EmptyStringConverter
+import io.smallrye.config.WithConverter
+
+interface TransformationStep {
+    fun name(): String
+    @WithConverter(EmptyStringConverter::class)
+    fun parameters(): MutableList<String?>
+}

src/main/kotlin/com/rak/config/converter/EmptyStringConverter.kt

@@ -0,0 +1,11 @@
+package com.rak.config.converter
+
+import jakarta.annotation.Priority
+import org.eclipse.microprofile.config.spi.Converter
+
+@Priority(1)
+class EmptyStringConverter : Converter<String> {
+    override fun convert(value: String): String {
+        return value
+    }
+}

src/main/kotlin/com/rak/model/transform/ParameterizedTransformation.kt

@@ -2,5 +2,5 @@ package com.rak.model.transform
 
 @FunctionalInterface
 fun interface ParameterizedTransformation : AbstractTransformation {
-    fun apply(input: Any, vararg parameters: Any): Any?
+    fun apply(input: String, vararg parameters: String): Any?
 }

src/main/kotlin/com/rak/model/transform/Transformation.kt

@@ -2,5 +2,5 @@ package com.rak.model.transform
 
 @FunctionalInterface
 fun interface Transformation : AbstractTransformation {
-    fun apply(input: Any): Any?
+    fun apply(input: String): String?
 }

src/main/kotlin/com/rak/model/transform/TransformationRegistry.kt

@@ -1,27 +1,56 @@
 package com.rak.model.transform
 
-import jakarta.enterprise.context.ApplicationScoped
+import java.util.concurrent.ConcurrentHashMap
 
-@ApplicationScoped
 class TransformationRegistry {
 
-    private val transformations = hashMapOf<String, (input: Any) -> Any>()
+    private val transformations: ConcurrentHashMap<String, Transformation> = ConcurrentHashMap()
+    private val parameterizedTransformation: ConcurrentHashMap<String, ParameterizedTransformation> =
+        ConcurrentHashMap()
 
     init {
-        register<String>("trim") {
+        register("trim") { it.trim() }
-            (it as String).trim()
+        register("replace") { input, parameters ->
+            require(parameters.size == 2) {
+                "'replace' requires exactly 2 parameters"
+            }
+            input.replace(parameters[0], parameters[1])
+        }
+        register("regexReplace") { input, params ->
+            require(params.size == 2) {
+                "'regexReplace' requires exactly 2 parameters"
+            }
+            input.replace(params[0].toRegex(), params[1])
         }
-        register<String>("replace", { s: Any ->
-            Integer.
-        })
     }
 
-//    fun <T : Any> register(name: String, transformation: (input: Any) -> T) {
-//        transformations[name] = transformation
-//    }
 
-    fun <T : Any> register(name: String, transformation: Transformation) {
+    fun register(name: String, transformation: Transformation) {
+        transformations.put(name, transformation)
+    }
 
+    fun register(name: String, transformation: ParameterizedTransformation) {
+        parameterizedTransformation.put(name, transformation)
+    }
+
+    fun getTransformation(name: String, parameters: List<String>? = null): AbstractTransformation {
+        return when {
+            transformations.containsKey(name) -> {
+                if (!parameters.isNullOrEmpty()) {
+                    throw IllegalArgumentException("'$name' doesn't accept parameters")
+                } else {
+                    transformations[name]!!
+                }
+            }
+            parameterizedTransformation.containsKey(name) -> {
+                if (parameters.isNullOrEmpty()) {
+                    throw IllegalArgumentException("'$name' requires parameters")
+                } else {
+                    parameterizedTransformation[name]!!
+                }
+            }
+            else -> throw IllegalArgumentException("Unknown transformation: '$name'")
+        }
     }
 
 }

src/main/kotlin/com/rak/util/XPathUtil.kt

@@ -26,8 +26,8 @@ class XPathUtil private constructor() {
                 .firstOrNull()?.text()
         }
 
-        fun getNextElement(root: Element, path: String): Element? {
+        fun getNextElement(element: Element, path: String): Element? {
-            return root.selectXpath(path).firstOrNull()
+            return element.selectXpath(path).firstOrNull()
         }
 
         fun extractResult(root: Element, path: String): String? {

src/main/resources/META-INF/services/org.eclipse.microprofile.config.spi.Converter

@@ -1 +1 @@
-com.rak.config.converter.TypeSelectorConverter
+com.rak.config.converter.TypeSelectorConverter

src/main/resources/application.yml

@@ -36,6 +36,11 @@ scraper:
             steps:
               - type: xpath
                 value: "//li/text()"
+            transform:
+              - name: "replace"
+                parameters:
+                  - " ("
+                  - ""
           language:
             steps:
               - type: xpath