Merge pull request 'OZG-4367 Umlaute-in-Feldnamen-aus-FC-uebertragen' (#100)...

Merge pull request 'OZG-4367 Umlaute-in-Feldnamen-aus-FC-uebertragen' (#100) from OZG-4367-Umlaute-in-Feldnamen-aus-FC-uebertragen into master

Reviewed-on: https://git.ozg-sh.de/ozgcloud-app/eingang-manager/pulls/100

formcycle-adapter/formcycle-adapter-impl/pom.xml

@@ -38,19 +38,19 @@
 
 	<properties>
 		<formcycle-interface.version>${project.version}</formcycle-interface.version>
+		<jsoup.version>1.17.2</jsoup.version>
 	</properties>
 
 	<dependencies>
 		<!--own project-->
 		<dependency>
-			<groupId>de.ozgcloud.vorgang</groupId>
-			<artifactId>vorgang-manager-utils</artifactId>
+			<groupId>de.ozgcloud.eingang</groupId>
+			<artifactId>formcycle-adapter-interface</artifactId>
+			<version>${formcycle-interface.version}</version>
 		</dependency>
 		<dependency>
-			<groupId>de.ozgcloud.vorgang</groupId>
-			<artifactId>vorgang-manager-utils</artifactId>
-			<type>test-jar</type>
-			<scope>test</scope>
+			<groupId>de.ozgcloud.eingang</groupId>
+			<artifactId>semantik-adapter</artifactId>
 		</dependency>
 		<dependency>
 			<groupId>de.ozgcloud.eingang</groupId>
@@ -58,6 +58,16 @@
 			<type>test-jar</type>
 			<scope>test</scope>
 		</dependency>
+		<dependency>
+			<groupId>de.ozgcloud.vorgang</groupId>
+			<artifactId>vorgang-manager-utils</artifactId>
+		</dependency>
+		<dependency>
+			<groupId>de.ozgcloud.vorgang</groupId>
+			<artifactId>vorgang-manager-utils</artifactId>
+			<type>test-jar</type>
+			<scope>test</scope>
+		</dependency>
 
 		<!--spring-->
 		<dependency>
@@ -69,15 +79,10 @@
 			<artifactId>spring-boot-starter-actuator</artifactId>
 		</dependency>
 
-
 		<dependency>
-			<groupId>de.ozgcloud.eingang</groupId>
-			<artifactId>formcycle-adapter-interface</artifactId>
-			<version>${formcycle-interface.version}</version>
-		</dependency>
-		<dependency>
-			<groupId>de.ozgcloud.eingang</groupId>
-			<artifactId>semantik-adapter</artifactId>
+			<groupId>org.jsoup</groupId>
+			<artifactId>jsoup</artifactId>
+			<version>${jsoup.version}</version>
 		</dependency>
 	</dependencies>
 

formcycle-adapter/formcycle-adapter-impl/src/main/java/de/ozgcloud/eingang/formcycle/FormDataController.java

@@ -65,6 +65,7 @@ class FormDataController {
 	private final FormCycleFormDataMapper mapper;
 	private final SemantikAdapter semantikAdapter;
 	private final VorgangNummerSupplier vorgangNummerSupplier;
+	private final FormDataHtmlCleaner formDataHtmlCleaner;
 
 	@PostMapping(consumes = "multipart/form-data", produces = HTTP_TYPE_PROTOBUF)
 	public FormCycleConfirmationResponse receiveFormData(@RequestPart FormCycleFormData formData,
@@ -72,6 +73,7 @@ class FormDataController {
 			@RequestPart(required = false) Optional<Collection<MultipartFile>> attachments) {
 
 		FormData mappedFormData = mapper.toFormData(formData);
+		mappedFormData = formDataHtmlCleaner.clean(mappedFormData);
 		mappedFormData = addRepresentations(representations, mappedFormData);
 		mappedFormData = addFiles(formData, attachments, mappedFormData);
 		mappedFormData = addServiceKonto(formData, mappedFormData);

formcycle-adapter/formcycle-adapter-impl/src/main/java/de/ozgcloud/eingang/formcycle/FormDataHtmlCleaner.java

+/*
+ * Copyright (C) 2023 Das Land Schleswig-Holstein vertreten durch den
+ * Ministerpräsidenten des Landes Schleswig-Holstein
+ * Staatskanzlei
+ * Abteilung Digitalisierung und zentrales IT-Management der Landesregierung
+ *
+ * Lizenziert unter der EUPL, Version 1.2 oder - sobald
+ * diese von der Europäischen Kommission genehmigt wurden -
+ * Folgeversionen der EUPL ("Lizenz");
+ * Sie dürfen dieses Werk ausschließlich gemäß
+ * dieser Lizenz nutzen.
+ * Eine Kopie der Lizenz finden Sie hier:
+ *
+ * https://joinup.ec.europa.eu/collection/eupl/eupl-text-eupl-12
+ *
+ * Sofern nicht durch anwendbare Rechtsvorschriften
+ * gefordert oder in schriftlicher Form vereinbart, wird
+ * die unter der Lizenz verbreitete Software "so wie sie
+ * ist", OHNE JEGLICHE GEWÄHRLEISTUNG ODER BEDINGUNGEN -
+ * ausdrücklich oder stillschweigend - verbreitet.
+ * Die sprachspezifischen Genehmigungen und Beschränkungen
+ * unter der Lizenz sind dem Lizenztext zu entnehmen.
+ */
+package de.ozgcloud.eingang.formcycle;
+
+import java.util.Collection;
+import java.util.Map;
+import java.util.stream.Collectors;
+
+import org.jsoup.Jsoup;
+import org.jsoup.nodes.Document;
+import org.jsoup.safety.Safelist;
+import org.springframework.stereotype.Component;
+
+import de.ozgcloud.eingang.common.formdata.FormData;
+
+@Component
+public class FormDataHtmlCleaner {
+
+	public FormData clean(FormData formData) {
+		return formData.toBuilder().formData(cleanFormData(formData.getFormData())).build();
+	}
+
+	Map<String, Object> cleanFormData(Map<String, Object> formData) {
+		return formData.entrySet().stream().collect(Collectors.toMap(Map.Entry::getKey, entry -> cleanValue(entry.getValue())));
+	}
+
+	@SuppressWarnings("unchecked")
+	Object cleanValue(Object value) {
+		if (value instanceof Map) {
+			return cleanFormData((Map<String, Object>) value);
+		} else if (value instanceof Collection<?> values) {
+			return values.stream().map(this::cleanValue).toList();
+		} else if (value instanceof String valueString) {
+			return parseHtml(valueString);
+		}
+		return value;
+	}
+
+	Object parseHtml(String html) {
+		var jsoupDocument = Jsoup.parse(html);
+		var outputSettings = new Document.OutputSettings(); // keep new lines
+		outputSettings.prettyPrint(false);
+		jsoupDocument.outputSettings(outputSettings);
+		var innerHtml = jsoupDocument.html().replace("\\\\n", "\n");
+		return Jsoup.clean(innerHtml, "", Safelist.none(), outputSettings);
+	}
+
+}

formcycle-adapter/formcycle-adapter-impl/src/test/java/de/ozgcloud/eingang/formcycle/FormDataControllerTest.java

@@ -71,6 +71,8 @@ class FormDataControllerTest {
 	private SemantikAdapter semantikAdapter;
 	@Mock
 	private VorgangNummerSupplier vorgangNummerSupplier;
+	@Mock
+	private FormDataHtmlCleaner htmlCleaner;
 
 	private MockMvc mockMvc;
 
@@ -93,6 +95,7 @@ class FormDataControllerTest {
 
 		@BeforeEach
 		void init() {
+			when(htmlCleaner.clean(any())).thenReturn(mappedFormData);
 			when(mapper.toFormData(any())).thenReturn(mappedFormData);
 			when(vorgangNummerSupplier.get()).thenReturn(VORGANG_NUMMER);
 		}
@@ -111,7 +114,7 @@ class FormDataControllerTest {
 
 		@Test
 		@SneakyThrows
-		void shouldRespondeWithVorgangNummer() {
+		void shouldResponseWithVorgangNummer() {
 
 			var confirmation = FormCycleConfirmationResponse.parseFrom(
 					doPostRequest().andReturn().getResponse().getContentAsByteArray());
@@ -127,6 +130,13 @@ class FormDataControllerTest {
 			assertThat(formDataCaptor.getValue().getHeader().getRequestId()).isEqualTo(VORGANG_NUMMER);
 		}
 
+		@Test
+		void shouldCallHtmlCleaner() {
+			doPostRequest();
+
+			verify(htmlCleaner).clean(any());
+		}
+
 		@Test
 		void shouldCallMapper() {
 			doPostRequest();
@@ -157,8 +167,9 @@ class FormDataControllerTest {
 
 			@BeforeEach
 			void init() {
-				when(mapper.toFormData(any()))
-						.thenReturn(FormDataTestFactory.createBuilder().clearRepresentations().numberOfRepresentations(0).build());
+				var formData = FormDataTestFactory.createBuilder().clearRepresentations().numberOfRepresentations(0).build();
+				when(mapper.toFormData(any())).thenReturn(formData);
+				when(htmlCleaner.clean(any())).thenReturn(formData);
 			}
 
 			@Test
@@ -195,8 +206,9 @@ class FormDataControllerTest {
 		class Attachments {
 			@BeforeEach
 			void init() {
-				when(mapper.toFormData(any()))
-						.thenReturn(FormDataTestFactory.createBuilder().clearAttachments().numberOfAttachments(0).build());
+				var formData = FormDataTestFactory.createBuilder().clearAttachments().numberOfAttachments(0).build();
+				when(mapper.toFormData(any())).thenReturn(formData);
+				when(htmlCleaner.clean(any())).thenReturn(formData);
 			}
 
 			@Test

formcycle-adapter/formcycle-adapter-impl/src/test/java/de/ozgcloud/eingang/formcycle/FormDataHtmlCleanerTest.java

+/*
+ * Copyright (C) 2023 Das Land Schleswig-Holstein vertreten durch den
+ * Ministerpräsidenten des Landes Schleswig-Holstein
+ * Staatskanzlei
+ * Abteilung Digitalisierung und zentrales IT-Management der Landesregierung
+ *
+ * Lizenziert unter der EUPL, Version 1.2 oder - sobald
+ * diese von der Europäischen Kommission genehmigt wurden -
+ * Folgeversionen der EUPL ("Lizenz");
+ * Sie dürfen dieses Werk ausschließlich gemäß
+ * dieser Lizenz nutzen.
+ * Eine Kopie der Lizenz finden Sie hier:
+ *
+ * https://joinup.ec.europa.eu/collection/eupl/eupl-text-eupl-12
+ *
+ * Sofern nicht durch anwendbare Rechtsvorschriften
+ * gefordert oder in schriftlicher Form vereinbart, wird
+ * die unter der Lizenz verbreitete Software "so wie sie
+ * ist", OHNE JEGLICHE GEWÄHRLEISTUNG ODER BEDINGUNGEN -
+ * ausdrücklich oder stillschweigend - verbreitet.
+ * Die sprachspezifischen Genehmigungen und Beschränkungen
+ * unter der Lizenz sind dem Lizenztext zu entnehmen.
+ */
+package de.ozgcloud.eingang.formcycle;
+
+import static org.assertj.core.api.Assertions.*;
+import static org.mockito.Mockito.*;
+
+import java.util.List;
+import java.util.Map;
+
+import org.assertj.core.data.MapEntry;
+import org.junit.jupiter.api.Nested;
+import org.junit.jupiter.api.Test;
+import org.mockito.InjectMocks;
+import org.mockito.Spy;
+
+import de.ozgcloud.eingang.common.formdata.FormData;
+
+class FormDataHtmlCleanerTest {
+
+	private static final String KEY = "key";
+	private static final Object VALUE = "value";
+
+	@Spy
+	@InjectMocks
+	private FormDataHtmlCleaner cleaner;
+
+	@Nested
+	class TestClean {
+
+		private final static Map<String, Object> FORM_DATA_MAP = Map.of(KEY, VALUE);
+		private final static FormData FORM_DATA = FormData.builder().formData(FORM_DATA_MAP).build();
+
+		@Test
+		void shouldCallCleanFormData() {
+			cleaner.clean(FORM_DATA);
+
+			verify(cleaner).cleanFormData(FORM_DATA_MAP);
+		}
+
+		@Test
+		void shouldSetCleanedFormData() {
+			var cleanedFormData = Map.of(KEY, VALUE);
+			doReturn(cleanedFormData).when(cleaner).cleanFormData(anyMap());
+
+			var result = cleaner.clean(FORM_DATA);
+
+			assertThat(result.getFormData()).isSameAs(cleanedFormData);
+		}
+	}
+
+	@Nested
+	class TestCleanFormData {
+
+		@Test
+		void shouldCallCleanValue() {
+			cleaner.cleanFormData(Map.of(KEY, VALUE));
+
+			verify(cleaner).cleanValue(VALUE);
+		}
+
+		@Test
+		void shouldReturnCleanedMap() {
+			var cleanedValue = "noHtml";
+			doReturn(cleanedValue).when(cleaner).cleanValue(any());
+
+			var result = cleaner.cleanFormData(Map.of(KEY, VALUE));
+
+			assertThat(result).containsOnly(MapEntry.entry(KEY, cleanedValue));
+		}
+	}
+
+	@Nested
+	class TestCleanValue {
+
+		@Nested
+		class TestCleanMap {
+
+			@Test
+			void shouldCallCleanFormData() {
+				var expectedMap = Map.of(KEY, VALUE);
+
+				cleaner.cleanValue(expectedMap);
+
+				verify(cleaner).cleanFormData(expectedMap);
+			}
+
+			@Test
+			void shouldReturnValue() {
+				var expectedMap = Map.of(KEY, VALUE);
+				doReturn(expectedMap).when(cleaner).cleanFormData(anyMap());
+
+				var result = cleaner.cleanValue(Map.of("a", "b"));
+
+				assertThat(result).isSameAs(expectedMap);
+			}
+		}
+
+		@Nested
+		class TestCleanCollection {
+
+			@Test
+			void shouldCallCleanValue() {
+				cleaner.cleanValue(List.of(VALUE));
+
+				verify(cleaner).cleanValue(VALUE);
+			}
+
+			@Test
+			void shouldReturnValue() {
+				doReturn(List.of(VALUE)).when(cleaner).cleanValue(any());
+
+				var result = cleaner.cleanValue(List.of("a"));
+
+				assertThat(result).isInstanceOf(List.class).asList().containsExactly(VALUE);
+			}
+		}
+
+		@Nested
+		class TestCleanString {
+
+			@Test
+			void shouldCallParseHtml() {
+				var stringValue = VALUE.toString();
+
+				cleaner.cleanValue(VALUE);
+
+				verify(cleaner).parseHtml(stringValue);
+			}
+
+			@Test
+			void shouldReturnValue() {
+				var cleanedValue = "noHtml";
+				doReturn(cleanedValue).when(cleaner).parseHtml(anyString());
+
+				var result = cleaner.cleanValue(VALUE);
+
+				assertThat(result).isEqualTo(cleanedValue);
+			}
+		}
+
+		@Test
+		void shouldReturnUnmodifiedValue() {
+			var value = 1;
+
+			var result = cleaner.cleanValue(value);
+
+			assertThat(result).isEqualTo(1);
+			verify(cleaner, never()).parseHtml(any());
+		}
+	}
+
+	@Nested
+	class TestHtmlCleaner {
+
+		static final String KEY_LABEL = "label";
+		static final String KEY_VALUE = "value";
+
+		static final Map<String, Object> FORM_DATA_MAP = Map.of("tf1", Map.of(
+						KEY_LABEL, "<p><em>&Auml;</em></p>",
+						KEY_VALUE, "Ä - Wert"),
+				"tf2", Map.of(
+						KEY_LABEL, "<p><strong>&Ouml;</strong></p>",
+						KEY_VALUE, "Ö - Wert"),
+				"fs1", Map.of(
+						KEY_LABEL, "Ü",
+						KEY_VALUE, Map.of(
+								"tf3", Map.of(
+										KEY_LABEL, " <p><s>Label mit</s> &szlig;</p>",
+										KEY_VALUE, "ein Text mit ß und <html><body><h1>Hello</h1><body><html>")),
+						"tf4", Map.of(
+								KEY_LABEL, "<p><span style=\"background-color:#1abc9c;\">&auml;</span></p>",
+								KEY_VALUE, "Text"),
+						"ed1", Map.of(
+								KEY_LABEL,
+								"<ol>\n\t<li><em><strong><u>&ouml;</u></strong></em></li>\n\t<li><span style=\"color:#e74c3c;\">&uuml;</span></li>\n</ol>",
+								KEY_VALUE, "TExt\nmit\n    Leerzeichen\nund\n  Umbrüchen"
+						)));
+
+		static final Map<String, Object> EXPECTED_MAP = Map.of("tf1", Map.of(
+						KEY_LABEL, "Ä",
+						KEY_VALUE, "Ä - Wert"),
+				"tf2", Map.of(
+						KEY_LABEL, "Ö",
+						KEY_VALUE, "Ö - Wert"),
+				"fs1", Map.of(
+						KEY_LABEL, "Ü",
+						KEY_VALUE, Map.of(
+								"tf3", Map.of(
+										KEY_LABEL, "Label mit ß",
+										KEY_VALUE, "ein Text mit ß und Hello")),
+						"tf4", Map.of(
+								KEY_LABEL, "ä",
+								KEY_VALUE, "Text"),
+						"ed1", Map.of(
+								KEY_LABEL,
+								"\n\tö\n\tü\n",
+								KEY_VALUE, "TExt\nmit\n    Leerzeichen\nund\n  Umbrüchen"
+						)));
+
+		@Test
+		void shouldCleanHtml() {
+			var result = cleaner.clean(FormData.builder().formData(FORM_DATA_MAP).build());
+
+			assertThat(result.getFormData()).isEqualTo(EXPECTED_MAP);
+		}
+	}
+}
\ No newline at end of file