1 package net.pterodactylus.rhynodge.webpages.weather.wetterde
3 import net.pterodactylus.rhynodge.Filter
4 import net.pterodactylus.rhynodge.State
5 import net.pterodactylus.rhynodge.states.FailedState
6 import net.pterodactylus.rhynodge.states.HtmlState
7 import net.pterodactylus.rhynodge.webpages.weather.WindDirection
8 import net.pterodactylus.rhynodge.webpages.weather.toWindDirection
9 import org.jsoup.nodes.Document
10 import org.jsoup.nodes.Element
11 import java.time.DateTimeException
12 import java.time.LocalDate
13 import java.time.LocalTime
14 import java.time.ZoneId
15 import java.time.ZonedDateTime
16 import java.time.format.DateTimeFormatter
17 import java.util.Locale
20 * HTML parser for [wetter.de](https://www.wetter.de/).
22 * @author [David ‘Bombe’ Roden](mailto:bombe@pterodactylus.net)
24 class WetterDeFilter : Filter {
27 val parser = DateTimeFormatter.ofPattern("dd. MMM yyyy").withLocale(Locale.GERMAN).withZone(ZoneId.of("Europe/Berlin"))
30 override fun filter(state: State): State {
31 if (state.success().not()) {
32 return FailedState.from(state)
34 state as? HtmlState ?: throw IllegalArgumentException("state is not an HTML state")
36 return parseWetterDeState(state)
39 private fun parseWetterDeState(htmlState: HtmlState): WetterDeState {
40 val dateTime = parseDate(htmlState.document()) ?: throw IllegalArgumentException("date can not be parsed")
41 val wetterDeState = WetterDeState(dateTime)
42 parseHourStates(htmlState.document()).forEach { wetterDeState += it }
46 private fun parseDate(document: Document): ZonedDateTime? {
47 return document.select(".forecast-detail-headline").text()
48 .extract(Regex(".*?([0-9]{1,2}\\. [^ ]+ [0-9]{4}).*?$"))?.toDate() ?: return null
51 private fun String.extract(regex: Regex): String? {
52 return regex.find(this)?.groups?.get(1)?.value
55 private fun String.toDate(): ZonedDateTime? {
57 ZonedDateTime.of(DateParser.parser.parse(this, LocalDate::from), LocalTime.of(0, 0), ZoneId.of("Europe/Berlin"))
58 } catch (e: DateTimeException) {
64 private fun parseHourStates(document: Document): Collection<HourState> {
65 return document.select(".forecast-detail-column-1h").mapIndexed { index, element -> parseHourState(index, element) }
68 private fun parseHourState(index: Int, element: Element): HourState {
71 element.select("span.temperature").text().extract(Regex("([0-9]+).*$"))?.toInt() ?: 0,
72 element.select("span.temperature-wind-chill").text().extract(Regex("[^0-9]*?([0-9]+)[^0-9]*?"))?.toInt() ?: 0,
73 (element.select(".forecast-icon-rain-fill").attr("style").extract(Regex("[^0-9]*?([0-9]+)[^0-9]*?"))?.toInt() ?: 0) / 100.0,
74 element.select(".forecast-rain span:eq(4)").text().replace(',', '.').extract(Regex("[^0-9]*?([0-9\\.]+)[^0-9]*?"))?.toDouble() ?: 0.0,
75 element.select(".forecast-wind-text").text().extract(Regex(".*?(Nord((nord)?(ost|west))?|Ost((nord|süd)ost)?|Süd((süd)?(ost|west))?|West((nord|süd)west)?).*?"))?.toWindDirection() ?: WindDirection.NONE,
76 element.select(".forecast-wind-text span:eq(1)").text().extract(Regex(".*?([0-9]+) km/h.*?"))?.toInt() ?: 0,
77 element.select(".forecast-wind-text span:eq(4)").text().extract(Regex(".*?([0-9]+) km/h.*?"))?.toInt() ?: 0,
78 (element.select(".forecast-humidity-text span:eq(0)").text().extract(Regex("([0-9]+)[^0-9]*?"))?.toInt() ?: 0) / 100.0,
79 element.select("span.temperature-condition").text(),
80 element.select(".forecast-image img").attr("src").prefixProtocol()
84 private fun String.prefixProtocol(): String {
85 return if (startsWith("//")) "http:" + this else this