Fix wrong title parsing at Golden Mangás. (#3082)

This commit is contained in:
Alessandro Jean 2020-05-09 12:29:51 -03:00 committed by GitHub
parent d418ae93b2
commit 7cfd2858c9
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
2 changed files with 37 additions and 25 deletions

View File

@ -5,7 +5,7 @@ ext {
appName = 'Tachiyomi: Golden Mangás' appName = 'Tachiyomi: Golden Mangás'
pkgNameSuffix = 'pt.goldenmangas' pkgNameSuffix = 'pt.goldenmangas'
extClass = '.GoldenMangas' extClass = '.GoldenMangas'
extVersionCode = 3 extVersionCode = 4
libVersion = '1.2' libVersion = '1.2'
} }

View File

@ -24,7 +24,7 @@ class GoldenMangas : ParsedHttpSource() {
override val name = "Golden Mangás" override val name = "Golden Mangás"
override val baseUrl = "https://goldenmanga.top" override val baseUrl = "https://goldenmangas.online"
override val lang = "pt-BR" override val lang = "pt-BR"
@ -43,11 +43,11 @@ class GoldenMangas : ParsedHttpSource() {
override fun popularMangaRequest(page: Int): Request = GET(baseUrl, headers) override fun popularMangaRequest(page: Int): Request = GET(baseUrl, headers)
override fun popularMangaSelector(): String = "div.itemmanga" override fun popularMangaSelector(): String = "div#maisLidos div.itemmanga"
override fun popularMangaFromElement(element: Element): SManga = SManga.create().apply { override fun popularMangaFromElement(element: Element): SManga = SManga.create().apply {
title = removeLanguage(element.select("h3").first().text()) title = element.select("h3").first().text().withoutLanguage()
thumbnail_url = baseUrl + element.select("img").first()?.attr("src") thumbnail_url = element.select("img").first()?.attr("abs:src")
url = element.attr("href") url = element.attr("href")
} }
@ -62,10 +62,10 @@ class GoldenMangas : ParsedHttpSource() {
override fun latestUpdatesFromElement(element: Element): SManga = SManga.create().apply { override fun latestUpdatesFromElement(element: Element): SManga = SManga.create().apply {
val infoElement = element.select("div.col-sm-10.col-xs-8 h3").first() val infoElement = element.select("div.col-sm-10.col-xs-8 h3").first()
val thumb = element.select("a:first-child div img").first().attr("src") val thumb = element.select("a:first-child div img").first().attr("abs:src")
title = removeLanguage(infoElement.text()) title = infoElement.text().withoutLanguage()
thumbnail_url = baseUrl + thumb.replace("w=80&h=120", "w=380&h=600") thumbnail_url = thumb.replace("w=80&h=120", "w=380&h=600")
url = element.select("a:first-child").attr("href") url = element.select("a:first-child").attr("href")
} }
@ -85,8 +85,8 @@ class GoldenMangas : ParsedHttpSource() {
override fun searchMangaSelector() = "div.mangas.col-lg-2 a" override fun searchMangaSelector() = "div.mangas.col-lg-2 a"
override fun searchMangaFromElement(element: Element): SManga = SManga.create().apply { override fun searchMangaFromElement(element: Element): SManga = SManga.create().apply {
title = removeLanguage(element.select("h3").first().text()) title = element.select("h3").first().text().withoutLanguage()
thumbnail_url = baseUrl + element.select("img").first().attr("src") thumbnail_url = element.select("img").first().attr("abs:src")
url = element.attr("href") url = element.attr("href")
} }
@ -97,15 +97,15 @@ class GoldenMangas : ParsedHttpSource() {
val firstColumn = infoElement.select("div.col-sm-4.text-right > img").first() val firstColumn = infoElement.select("div.col-sm-4.text-right > img").first()
val secondColumn = infoElement.select("div.col-sm-8").first() val secondColumn = infoElement.select("div.col-sm-8").first()
title = removeLanguage(secondColumn.select("h2:eq(1)").text()) title = secondColumn.select("h2:eq(0)").text().withoutLanguage()
author = removeLabel(secondColumn.select("h5:eq(3)").text()) author = secondColumn.select("h5:eq(3)")!!.text().withoutLabel()
artist = removeLabel(secondColumn.select("h5:eq(4)").text()) artist = secondColumn.select("h5:eq(4)")!!.text().withoutLabel()
genre = secondColumn.select("h5:eq(2) a") genre = secondColumn.select("h5:eq(2) a")
.filter { it.text().isNotEmpty() } .filter { it.text().isNotEmpty() }
.joinToString { it.text() } .joinToString { it.text() }
status = parseStatus(secondColumn.select("h5:eq(5) a").text().orEmpty()) status = parseStatus(secondColumn.select("h5:eq(5) a").text().orEmpty())
description = document.select("#manga_capitulo_descricao").text() description = document.select("#manga_capitulo_descricao").text()
thumbnail_url = baseUrl + firstColumn.attr("src") thumbnail_url = firstColumn.attr("abs:src")
} }
private fun parseStatus(status: String) = when { private fun parseStatus(status: String) = when {
@ -120,36 +120,48 @@ class GoldenMangas : ParsedHttpSource() {
val firstColumn = element.select("a > div.col-sm-5") val firstColumn = element.select("a > div.col-sm-5")
val secondColumn = element.select("div.col-sm-5.text-right a[href^='http']") val secondColumn = element.select("div.col-sm-5.text-right a[href^='http']")
name = firstColumn.select("div.col-sm-5").first().text().substringBefore("(").trim() name = firstColumn.select("div.col-sm-5").first().text()
.substringBefore("(").trim()
scanlator = secondColumn?.joinToString { it.text() } scanlator = secondColumn?.joinToString { it.text() }
date_upload = parseChapterDate(firstColumn.select("div.col-sm-5 span[style]").first().text()) date_upload = DATE_FORMATTER.tryParseTime(firstColumn.select("div.col-sm-5 span[style]").first().text())
url = element.select("a").first().attr("href") url = element.select("a").first().attr("href")
} }
override fun pageListParse(document: Document): List<Page> { override fun pageListParse(document: Document): List<Page> {
val chapImages = document.select("div.col-sm-12[id^='capitulos_images']").first() val chapterUrl = document.location()
val pages = chapImages.select("img[pag]") val chapterImages = document.select("div.col-sm-12[id^='capitulos_images']").first()
return pages return chapterImages.select("img[pag]")
.mapIndexed { i, element -> Page(i, "", baseUrl + element.attr("src")) } .mapIndexed { i, element -> Page(i, chapterUrl, element.attr("abs:src")) }
} }
override fun imageUrlParse(document: Document) = "" override fun imageUrlParse(document: Document) = ""
private fun removeLanguage(text: String): String = text.replace(FLAG_REGEX, "").trim() override fun imageRequest(page: Page): Request {
val newHeaders = headersBuilder()
.set("Referer", page.url)
.build()
private fun removeLabel(text: String?): String = text!!.substringAfter(":").trim() return GET(page.imageUrl!!, newHeaders)
}
private fun parseChapterDate(date: String): Long { private fun String.withoutLanguage(): String = replace(FLAG_REGEX, "").trim()
private fun String.withoutLabel(): String = substringAfter(":").trim()
private fun SimpleDateFormat.tryParseTime(date: String): Long {
return try { return try {
SimpleDateFormat("(dd/MM/yyyy)", Locale.ENGLISH).parse(date).time parse(date).time
} catch (e: ParseException) { } catch (e: ParseException) {
0L 0L
} }
} }
companion object { companion object {
private const val USER_AGENT = "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/81.0.4044.122 Safari/537.36" private const val USER_AGENT = "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/81.0.4044.138 Safari/537.36"
private val FLAG_REGEX = "\\((Pt[-/]br|Scan)\\)".toRegex(RegexOption.IGNORE_CASE) private val FLAG_REGEX = "\\((Pt[-/]br|Scan)\\)".toRegex(RegexOption.IGNORE_CASE)
private val DATE_FORMATTER by lazy { SimpleDateFormat("(dd/MM/yyyy)", Locale.ENGLISH) }
} }
} }