implemented importer für Bandhaus Leipzug

Signed-off-by: Stephan Richter <s.richter@srsoftware.de>
This commit is contained in:
2025-01-02 21:16:38 +01:00
parent c561be4b9c
commit 6b0d3a6bc4
4 changed files with 221 additions and 19 deletions

View File

@@ -47,11 +47,6 @@ public class Psychochor extends SinglePageImporter {
return attributeEquals("itemprop","description");
}
@Override
protected Predicate<Tag> extractDescriptionFilter() {
return attributeEquals("itemprop","description");
}
@Override
protected Result<Coords> extractCoords(Tag eventTag) {
var list = eventTag.find(ofType("iframe"));
@@ -74,6 +69,11 @@ public class Psychochor extends SinglePageImporter {
return error("Failed to parse coords from %s",s);
}
@Override
protected Predicate<Tag> extractDescriptionFilter() {
return attributeEquals("itemprop","description");
}
@Override
protected Predicate<Tag> extractEndDateFilter() {
return null;

View File

@@ -0,0 +1,180 @@
package de.srsoftware.cal.importer.leipzig;
import de.srsoftware.cal.BaseImporter;
import de.srsoftware.cal.Util;
import de.srsoftware.cal.api.Coords;
import de.srsoftware.tools.Payload;
import de.srsoftware.tools.Result;
import de.srsoftware.tools.Tag;
import java.security.NoSuchAlgorithmException;
import java.time.LocalDate;
import java.time.LocalTime;
import java.util.List;
import java.util.Objects;
import java.util.Optional;
import java.util.function.Predicate;
import java.util.stream.Collectors;
import java.util.stream.Stream;
import static de.srsoftware.cal.Util.parseGermanTime;
import static de.srsoftware.tools.Error.error;
import static de.srsoftware.tools.Optionals.nullable;
import static de.srsoftware.tools.Result.transform;
import static de.srsoftware.tools.Tag.*;
import static de.srsoftware.tools.TagFilter.*;
public class Bandhaus extends BaseImporter {
private static final Coords COORDS_BANDHAUS = new Coords(51.32498, 12.31578);
public Bandhaus() throws NoSuchAlgorithmException {
super();
}
@Override
protected String baseUrl() {
return "https://bandcommunity-leipzig.org";
}
@Override
public String description() {
return "Importer für Veranstaltungen des Bandhaus Leipzig";
}
@Override
protected Predicate<Tag> extractAttachmentsFilter() {
return attributeHas(CLASS,"event_description");
}
@Override
protected Predicate<Tag> extractDescriptionFilter() {
return attributeEquals("itemprop","description");
}
@Override
protected Result<Coords> extractCoords(Tag eventTag) {
var list = eventTag.find(attributeHas(CLASS,"evo_location_address"));
if (list.isEmpty()) return error("Failed to locate address → failed to locate coords");
var loc = list.getFirst().strip();
var lower = loc.toLowerCase();
if (lower.contains("saarländer")) return Payload.of(COORDS_BANDHAUS);
return error("Unknown location: %s",loc);
}
@Override
protected Predicate<Tag> extractEndDateFilter() {
return null;
}
@Override
protected Predicate<Tag> extractEndTimeFilter() {
return null;
}
@Override
protected Predicate<Tag> extractEventTagFilter() {
return attributeStartsWith(ID, "event_");
}
@Override
protected Result<List<String>> extractEventUrls(Result<Tag> programPage) {
var opt = programPage.optional();
if (opt.isEmpty())return transform(programPage);
var list = opt.get().find(attributeEquals(ID, "evcal_list")).stream()
.flatMap(tag -> tag.find(IS_ANCHOR).stream())
.map(anchor -> anchor.get(HREF))
.filter(Objects::nonNull).filter(link-> link.contains("/events/"))
.map(link -> link.contains("://") ? link : baseUrl()+link)
.distinct().toList();
return Payload.of(list);
}
@Override
protected Predicate<Tag> extractLinksFilter() {
return attributeEquals("itemprop","description");
}
protected Result<Tag> extractLocationTag(Tag eventTag){
var list = eventTag.find(extractLocationFilter()).stream()
.flatMap(tag -> tag.find(IS_PARAGRAPH).stream()).toList();
if (list.isEmpty()) return error("Failed to find location tag");
return Payload.of(Tag.of(DIV).addAll(list));
}
protected Result<String> extractLocation(Tag eventTag) {
var location = eventTag.find(extractLocationFilter()).stream()
.flatMap(tag -> tag.find(IS_PARAGRAPH).stream())
.map(Tag::strip).collect(Collectors.joining(", "));
return location.isBlank() ? error("Failed to find location!") : Payload.of(location);
}
@Override
protected Predicate<Tag> extractLocationFilter() {
return attributeContains(CLASS,"data_cell").and(tag -> !tag.find(attributeHas(CLASS,"evo_location_address")).isEmpty());
}
@Override
protected Result<LocalDate> extractStartDate(Tag eventTag) {
Result<Tag> startDateTag = extractStartDateTag(eventTag);
var opt = startDateTag.optional();
if (opt.isEmpty()) return transform(startDateTag);
var dateTag = opt.get();
Result<Integer> year = dateTag.find(withAttribute("data-syr")).stream()
.map(tag -> tag.get("data-syr"))
.map(Util::parseInt).findAny().orElseGet(() -> error("Failed to find start date"));
Result<Integer> day = dateTag.find(attributeEquals(CLASS, "date")).stream()
.map(Tag::strip).map(Util::parseInt).findAny().orElseGet(() -> error("Failed to find start date"));
Result<Integer> month = dateTag.find(attributeEquals(CLASS, "month")).stream()
.map(Tag::strip).map(Util::toNumericMonth).findAny().orElseGet(() -> error("Failed to find start date"));
if (year.optional().isEmpty()) return transform(year);
if (month.optional().isEmpty()) return transform(month);
if (day.optional().isEmpty()) return transform(day);
return Payload.of(LocalDate.of(year.optional().get(),month.optional().get(),day.optional().get()));
}
@Override
protected Predicate<Tag> extractStartDateFilter() {
return withAttribute("data-syr").and(tag -> !tag.find(attributeContains(CLASS,"evo_start")).isEmpty());
}
@Override
protected Predicate<Tag> extractStartTimeFilter() {
return attributeContains(CLASS,"evo_start");
}
@Override
protected List<String> extractTags(Tag eventTag) {
return List.of("Bandhaus","Leipzig");
}
@Override
protected Predicate<Tag> extractTitleFilter() {
return attributeHas(CLASS,"evcal_event_title");
}
@Override
protected Result<LocalDate> parseEndDate(String string) {
return null;
}
@Override
protected Result<LocalTime> parseEndTime(String string) {
return null;
}
@Override
protected Result<LocalDate> parseStartDate(String string) {
return null;
}
@Override
protected Result<LocalTime> parseStartTime(String string) {
return parseGermanTime(string+" ");
}
@Override
protected String programURL() {
return baseUrl()+"/veranstaltungen";
}
}