added hash to appointment, implemented KassaBlanca parser

Signed-off-by: Stephan Richter <s.richter@srsoftware.de>

de.srsoftware.cal.api/src/main/java/de/srsoftware/cal/api/Appointment.java

@@ -34,6 +34,12 @@ public interface Appointment {
 	 */
 	Optional<LocalDateTime> end();
 
+	/**
+	 * create a unique identifier based on the event content
+	 * @return
+	 */
+	String hash();
+
 	/**
 	 * ID of the appointment – unique within this system
 	 * @return the appointment`s id

de.srsoftware.cal.app/src/main/java/de/srsoftware/cal/app/Application.java

@@ -1,7 +1,8 @@
 /* © SRSoftware 2024 */
 package de.srsoftware.cal.app;
 
-import de.srsoftware.cal.importer.jena.Rosenkeller;
+import de.srsoftware.cal.importer.jena.Kassablanca;
+import java.security.NoSuchAlgorithmException;
 
 /**
  * Test application
@@ -14,9 +15,9 @@ public class Application {
 	 * sandbox
 	 * @param args default
 	 */
-	public static void main(String[] args) {
+	public static void main(String[] args) throws NoSuchAlgorithmException {
-		var rosenkeller	 = new Rosenkeller();
+		var importer	 = new Kassablanca();
-		var appointments = rosenkeller.fetch();
+		var appointments = importer.fetch();
-		appointments.forEach(System.err::println);
+		appointments.forEach(System.out::println);
 	}
 }

de.srsoftware.cal.importer/build.gradle.kts

@@ -3,6 +3,6 @@ description = "OpenCloudCal : Importers"
 dependencies {
     implementation(project(":de.srsoftware.cal.api"))
     implementation("de.srsoftware:tools.optionals:1.0.0")
-    implementation("de.srsoftware:tools.util:1.1.2")
+    implementation("de.srsoftware:tools.util:1.1.3")
-    implementation("de.srsoftware:tools.web:1.3.2")
+    implementation("de.srsoftware:tools.web:1.3.3")
 }

de.srsoftware.cal.importer/src/main/java/de/srsoftware/cal/importer/BaseAppointment.java

@@ -17,6 +17,7 @@ public class BaseAppointment implements Appointment {
 	private final long          id;
 	private final String        title, description;
 	private final LocalDateTime end, start;
+	private final String        hash;
 	private Coords	            coords        = null;
 	private final Set<Attachment> attachments = new HashSet<>();
 	private final Set<String> tags	          = new HashSet<>();
@@ -32,9 +33,10 @@ public class BaseAppointment implements Appointment {
 	 * @param end set the end date
 	 * @param location set the location
 	 */
-	public BaseAppointment(long id, String title, String description, LocalDateTime start, LocalDateTime end, String location) {
+	public BaseAppointment(long id, String title, String description, LocalDateTime start, LocalDateTime end, String location, String hash) {
 		this.description = description;
 		this.end	 = end;
+		this.hash	 = hash;
 		this.id	 = id;
 		this.location	 = location;
 		this.start	 = start;
@@ -131,6 +133,11 @@ public class BaseAppointment implements Appointment {
 		return nullable(end);
 	}
 
+	@Override
+	public String hash() {
+		return hash;
+	}
+
 	@Override
 	public long id() {
 		return id;

de.srsoftware.cal.importer/src/main/java/de/srsoftware/cal/importer/BaseImporter.java

@@ -1,7 +1,9 @@
 /* © SRSoftware 2024 */
 package de.srsoftware.cal.importer;
 
+import static de.srsoftware.tools.Strings.hex;
 import static de.srsoftware.tools.TagFilter.ofType;
+import static java.nio.charset.StandardCharsets.UTF_8;
 
 import de.srsoftware.cal.api.*;
 import de.srsoftware.tools.*;
@@ -12,6 +14,8 @@ import java.net.MalformedURLException;
 import java.net.URI;
 import java.net.URISyntaxException;
 import java.net.URL;
+import java.security.MessageDigest;
+import java.security.NoSuchAlgorithmException;
 import java.time.LocalDateTime;
 import java.util.ArrayList;
 import java.util.List;
@@ -20,6 +24,13 @@ import java.util.Optional;
 import java.util.stream.Stream;
 
 public abstract class BaseImporter implements Importer {
+	private static final String SHA256 = "SHA-256";
+	private final MessageDigest digest;
+
+	protected BaseImporter() throws NoSuchAlgorithmException {
+		digest = MessageDigest.getInstance(SHA256);
+	}
+
 	protected abstract String baseUrl();
 
 	@Override
@@ -47,7 +58,7 @@ public abstract class BaseImporter implements Importer {
 	}
 
 
-	protected Result<String> extractDescription(Tag eventTag){
+	protected Result<String> extractDescription(Tag eventTag) {
 		Result<Tag> titleTag = extractDescriptionTag(eventTag);
 		if (titleTag.optional().isEmpty()) return transform(titleTag);
 		var inner = titleTag.optional().flatMap(tag -> tag.inner(2));
@@ -91,7 +102,9 @@ public abstract class BaseImporter implements Importer {
 		if (locationResult.optional().isEmpty()) return transform(locationResult);
 		var location = locationResult.optional().get();
 
-		var event = new BaseAppointment(id, title, description, start, end, location)  //
+		var hash = hash("%s@%s".formatted(start, location));
+
+		var event = new BaseAppointment(id, title, description, start, end, location, hash)  //
 			.add(extractAttachments(eventTag))
 			.addLinks(extractLinks(eventTag))
 			.tags(extractTags(eventTag));
@@ -139,7 +152,7 @@ public abstract class BaseImporter implements Importer {
 
 	public abstract Result<List<Tag>> extractLinkAnchors(Result<Tag> tagResult);
 
-	private Result<Tag> extractLinksTag(Tag eventTag) {
+	protected Result<Tag> extractLinksTag(Tag eventTag) {
 		return extractDescriptionTag(eventTag);
 	}
 
@@ -153,9 +166,9 @@ public abstract class BaseImporter implements Importer {
 
 
 	protected Result<LocalDateTime> extractStart(Tag eventTag) {
-		Result<Tag> endTag = extractStartTag(eventTag);
+		Result<Tag> startTag = extractStartTag(eventTag);
-		if (endTag.optional().isEmpty()) return transform(endTag);
+		if (startTag.optional().isEmpty()) return transform(startTag);
-		return parseStartDate(endTag.optional().get().toString(0));
+		return parseStartDate(startTag.optional().get().strip());
 	}
 
 	protected abstract Result<Tag> extractStartTag(Tag eventTag);
@@ -164,9 +177,9 @@ public abstract class BaseImporter implements Importer {
 	protected abstract List<String> extractTags(Tag eventTag);
 
 	protected Result<String> extractTitle(Tag eventTag) {
-		Result<Tag> locationTag = extractTitleTag(eventTag);
+		Result<Tag> titleTag = extractTitleTag(eventTag);
-		if (locationTag.optional().isEmpty()) return transform(locationTag);
+		if (titleTag.optional().isEmpty()) return transform(titleTag);
-		var inner = locationTag.optional().flatMap(tag -> tag.inner(2));
+		var inner = titleTag.optional().flatMap(tag -> tag.inner(2));
 		if (inner.isPresent()) return Payload.of(inner.get());
 		return Error.of("No title found");
 	}
@@ -186,9 +199,21 @@ public abstract class BaseImporter implements Importer {
 		return stream  //
 				.map(this::url)
 				.map(this::loadEvent)
+				.peek(e -> {
+					if (e instanceof Error<Appointment> err) System.err.println(err);
+				})
 				.flatMap(result -> result.optional().stream());
 	}
 
+	/**
+	 * create a hash from a text
+	 * @param plain the plain text
+	 * @return the hash of the plain text
+	 */
+	protected String hash(String plain){
+		return hex(digest.digest(plain.getBytes(UTF_8)));
+	}
+
 	protected static <T> Result<T> invalidParameter(Result<?> result) {
 		return Error.format("Invalid parameter: %s", result.getClass().getSimpleName());
 	}

de.srsoftware.cal.importer/src/main/java/de/srsoftware/cal/importer/jena/Kassablanca.java

@@ -0,0 +1,126 @@
+/* © SRSoftware 2024 */
+package de.srsoftware.cal.importer.jena;
+
+import static de.srsoftware.tools.TagFilter.*;
+
+import de.srsoftware.cal.importer.BaseImporter;
+import de.srsoftware.tools.*;
+import de.srsoftware.tools.Error;
+import java.security.NoSuchAlgorithmException;
+import java.time.LocalDateTime;
+import java.util.List;
+import java.util.regex.Pattern;
+
+public class Kassablanca extends BaseImporter {
+	public static final String   BASE_URL	= "https://www.kassablanca.de";
+	private static final String  APPOINTMENT_TAG_ID = "entry-content";
+	private static final Pattern START_DATE_PATTERN = Pattern.compile("(\\d+).(\\d+).(\\d+).*Beginn\\s*(\\d+):(\\d+)\\s*Uhr");
+	private static final String  LOCATION	= "Kassablanca e.V., Felsenkellerstr. 13a, 07745 Jena";
+
+	public Kassablanca() throws NoSuchAlgorithmException {
+		super();
+	}
+
+	@Override
+	protected String baseUrl() {
+		return BASE_URL;
+	}
+
+	@Override
+	protected Result<Tag> extractDescriptionTag(Tag eventTag) {
+		var list = eventTag.find(attributeHas("class", "se-content"));
+		if (list.size() == 1) return Payload.of(list.getFirst());
+		return Error.of("Failed to find description tag");
+	}
+
+	@Override
+	protected Result<Tag> extractEndTag(Tag eventTag) {
+		return Error.format("end date not supported");
+	}
+
+	@Override
+	protected Result<Tag> extractEventTag(Result<Tag> pageResult) {
+		if (pageResult.optional().isEmpty()) return transform(pageResult);
+		var list = pageResult.optional().get().find(attributeEquals("class", APPOINTMENT_TAG_ID));
+		if (list.size() == 1) return Payload.of(list.getFirst());
+		return Error.format("Could not find tag with id \"%s\"", APPOINTMENT_TAG_ID);
+	}
+
+	@Override
+	protected Result<List<String>> extractEventUrls(Result<Tag> programPage) {
+		if (programPage.optional().isEmpty()) return transform(programPage);
+		List<String> list = programPage.optional()
+			        .get()	//
+			        .find(attributeHas("class", "eventrow"))
+			        .stream()
+			        .flatMap(t -> t.find(ofType("h3")).stream())
+			        .map(t -> t.find(ofType("a")))
+			        .flatMap(List::stream)
+			        .map(t -> t.get("href"))
+			        .toList();
+		return Payload.of(list);
+	}
+
+	@Override
+	public Result<List<Tag>> extractLinkAnchors(Result<Tag> tagResult) {
+		if (tagResult.optional().isEmpty()) return transform(tagResult);
+		var tag = tagResult.optional().get();
+		tag.find(attributeEquals("id", "filterbar")).stream().findAny().ifPresent(Tag::remove);  // remove div with unrelated links
+		var anchors = tag.find(withAttribute("href"));
+		return Payload.of(anchors);
+	}
+
+	@Override
+	protected Result<Tag> extractLinksTag(Tag eventTag) {
+		return Payload.of(eventTag);
+	}
+
+	@Override
+	protected Result<Tag> extractLocationTag(Tag eventTag) {
+		return Payload.of(new Text(LOCATION));
+	}
+
+	@Override
+	protected Result<Tag> extractStartTag(Tag eventTag) {
+		List<Tag> tags = eventTag.find(attributeEquals("class", "se-header"));
+		if (tags.size() == 1) return Payload.of(tags.getFirst());
+		return Error.of("Failed to find event time information");
+	}
+
+	@Override
+	protected List<String> extractTags(Tag eventTag) {
+		return List.of("Kassablanca");
+	}
+
+	@Override
+	protected Result<Tag> extractTitleTag(Tag eventTag) {
+		var list = eventTag.find(ofType("h1"));
+		if (list.size() == 1) return Payload.of(list.getFirst());
+		return Error.of("Failed to find title tag");
+	}
+
+	@Override
+	protected Result<LocalDateTime> parseEndDate(String string) {
+		return null;
+	}
+
+	@Override
+	protected Result<LocalDateTime> parseStartDate(String string) {
+		var matcher = START_DATE_PATTERN.matcher(string);
+		if (matcher.find()) {
+			var day    = Integer.parseInt(matcher.group(1));
+			var month  = Integer.parseInt(matcher.group(2));
+			var year   = Integer.parseInt(matcher.group(3));
+			var hour   = Integer.parseInt(matcher.group(4));
+			var minute = Integer.parseInt(matcher.group(5));
+			var date   = LocalDateTime.of(year, month, day, hour, minute);
+			return Payload.of(date);
+		}
+		return Error.of("Could not recognize start date/time");
+	}
+
+	@Override
+	protected String programURL() {
+		return BASE_URL + "/programm";
+	}
+}

de.srsoftware.cal.importer/src/main/java/de/srsoftware/cal/importer/jena/Rosenkeller.java

@@ -9,6 +9,7 @@ import de.srsoftware.tools.Error;
 import de.srsoftware.tools.Payload;
 import de.srsoftware.tools.Result;
 import de.srsoftware.tools.Tag;
+import java.security.NoSuchAlgorithmException;
 import java.time.LocalDateTime;
 import java.util.List;
 import java.util.regex.Pattern;
@@ -19,6 +20,10 @@ public class Rosenkeller extends BaseImporter {
 	private static final Pattern DATE_PATTERN	= Pattern.compile("(\\d+) (\\w+)(\\W+(\\d+):(\\d+))?");
 	private static final String  DEFAULT_LOCATION	= "Rosenkeller, Johannisstr. 13, 07743 Jena";
 
+	public Rosenkeller() throws NoSuchAlgorithmException {
+		super();
+	}
+
 	@Override
 	protected String baseUrl() {
 		return BASE_URL;