Fix MangaHost not working (#3801)

* Fix MangaHost not working due to website redesign.

* Change OkHttpClient to Cloudflare.

src/pt/mangahost/build.gradle

@@ -5,7 +5,7 @@ ext {
     extName = 'MangaHost'
     pkgNameSuffix = 'pt.mangahost'
     extClass = '.MangaHost'
-    extVersionCode = 9
+    extVersionCode = 10
     libVersion = '1.2'
 }
 

src/pt/mangahost/src/eu/kanade/tachiyomi/extension/pt/mangahost/MangaHost.kt

@@ -11,6 +11,7 @@ import java.text.SimpleDateFormat
 import java.util.Locale
 import okhttp3.Headers
 import okhttp3.HttpUrl
+import okhttp3.OkHttpClient
 import okhttp3.Request
 import org.jsoup.Jsoup
 import org.jsoup.nodes.Document
@@ -30,15 +31,17 @@ class MangaHost : ParsedHttpSource() {
 
     override val supportsLatest = true
 
+    override val client: OkHttpClient = network.cloudflareClient
+
     override fun headersBuilder(): Headers.Builder = Headers.Builder()
         .add("User-Agent", USER_AGENT)
         .add("Referer", baseUrl)
 
-    private fun genericMangaFromElement(element: Element, lazy: Boolean = true): SManga =
+    private fun genericMangaFromElement(element: Element): SManga =
         SManga.create().apply {
             title = element.attr("title").withoutLanguage()
-            thumbnail_url = element.select("img.manga")
+            thumbnail_url = element.select("img")
-                .attr(if (lazy) "data-path" else "src")
+                .attr("data-path")
                 .toLargeUrl()
             setUrlWithoutDomain(element.attr("href").substringBeforeLast("-mh"))
         }
@@ -52,7 +55,7 @@ class MangaHost : ParsedHttpSource() {
         return GET("$baseUrl/mangas/mais-visualizados$pageStr", newHeaders)
     }
 
-    override fun popularMangaSelector(): String = "div.thumbnail div a.pull-left"
+    override fun popularMangaSelector(): String = "div#dados div.manga-block div.manga-block-left a"
 
     override fun popularMangaFromElement(element: Element): SManga = genericMangaFromElement(element)
 
@@ -67,9 +70,9 @@ class MangaHost : ParsedHttpSource() {
         return GET("$baseUrl/lancamentos$pageStr", newHeaders)
     }
 
-    override fun latestUpdatesSelector() = "table.table-lancamentos > tbody > tr > td:eq(0) > a"
+    override fun latestUpdatesSelector() = "div#dados div.line-lancamentos div.column-img a"
 
-    override fun latestUpdatesFromElement(element: Element): SManga = genericMangaFromElement(element, false)
+    override fun latestUpdatesFromElement(element: Element): SManga = genericMangaFromElement(element)
 
     override fun latestUpdatesNextPageSelector() = popularMangaNextPageSelector()
 
@@ -87,56 +90,32 @@ class MangaHost : ParsedHttpSource() {
     override fun searchMangaNextPageSelector(): String? = null
 
     override fun mangaDetailsParse(document: Document): SManga {
-        val infoElement = document.select("div#page > section > div > div.pull-left")
+        val infoElement = document.select("div.box-content div.w-row div.w-col:eq(1) article")
 
         return SManga.create().apply {
-            author = infoElement.select("li:contains(Autor:)").textWithoutLabel()
+            author = infoElement.select("div.text li div:contains(Autor:)").textWithoutLabel()
-            artist = infoElement.select("li:contains(Desenho (Art):)").textWithoutLabel()
+            artist = infoElement.select("div.text li div:contains(Arte:)").textWithoutLabel()
-            genre = infoElement.select("li:contains(Categoria(s):)").textWithoutLabel()
+            genre = infoElement.select("h3.subtitle + div.tags a").joinToString { it.text() }
-            description = infoElement.select("article").first()?.text()
+            description = infoElement.select("div.text div.paragraph").first()?.text()
                 ?.substringBefore("Relacionados:")
-            status = parseStatus(infoElement.select("li:contains(Status:)").text().orEmpty())
+            status = infoElement.select("div.text li div:contains(Status:)").text().toStatus()
-            thumbnail_url = document.select("div#page > section > div > img.thumbnail")
+            thumbnail_url = document.select("div.box-content div.w-row div.w-col:eq(0) div.widget img")
                 .attr("src")
         }
     }
 
-    private fun parseStatus(status: String) = when {
-        status.contains("Ativo") -> SManga.ONGOING
-        status.contains("Completo") -> SManga.COMPLETED
-        else -> SManga.UNKNOWN
-    }
-
     override fun chapterListSelector(): String =
-        "ul.list_chapters li a, table.table-hover:not(.table-mangas) > tbody > tr"
+        "article.article > section.clearfix div.chapters div.cap div.card.pop"
 
-    override fun chapterFromElement(element: Element): SChapter {
+    override fun chapterFromElement(element: Element): SChapter = SChapter.create().apply {
-        val isNewLayout = element.tagName() == "a"
+        name = element.select("div.pop-title").text().withoutLanguage()
-
+        scanlator = element.select("div.pop-content small strong").text()
-        if (isNewLayout) {
+        date_upload = element.select("small.clearfix").text()
-            val content = Jsoup.parse(element.attr("data-content"))
+            .substringAfter("Adicionado em ")
-            val date = content.select("small.clearfix").text()
+            .let { DATE_FORMAT.tryParseTime(it) }
-                .substringAfter("Adicionado em ")
+        chapter_number = element.select("div.pop-title span.btn-caps").text()
-
+            .toFloatOrNull() ?: 1f
-            return SChapter.create().apply {
+        setUrlWithoutDomain(element.select("div.tags a").attr("href"))
-                name = element.attr("data-original-title").withoutLanguage()
-                scanlator = content.select("small.clearfix strong").text()
-                date_upload = DATE_FORMAT_NEW.tryParseTime(date)
-                chapter_number = element.text().toFloatOrNull() ?: 1f
-                setUrlWithoutDomain(content.select("div.clearfix a").attr("href"))
-            }
-        }
-
-        val firstColumn = element.select("td:eq(0)")
-        val secondColumn = element.select("td:eq(1)")
-        val thirdColumn = element.select("td:eq(2)")
-
-        return SChapter.create().apply {
-            name = firstColumn.select("a").text().withoutLanguage()
-            scanlator = secondColumn.text()
-            date_upload = DATE_FORMAT_OLD.tryParseTime(thirdColumn.text())
-            setUrlWithoutDomain(firstColumn.select("a").attr("href"))
-        }
     }
 
     override fun pageListRequest(chapter: SChapter): Request {
@@ -149,17 +128,17 @@ class MangaHost : ParsedHttpSource() {
     }
 
     override fun pageListParse(document: Document): List<Page> {
-        val documentStr = document.toString()
+        val imagesHtml = document.select("script:containsData(var images)").first()!!
-        val images = documentStr
+            .data()
-            .substringAfter(SCRIPT_BEGIN)
+            .substringAfter("var images = [")
-            .substringBefore(SCRIPT_END)
+            .substringBefore("];")
             .replace(SCRIPT_REGEX, "")
 
-        val newDocument = Jsoup.parse(images)
+        return Jsoup.parse(imagesHtml)
-        val referer = document.select("link[rel='canonical']").first()
+            .select("a img")
-
+            .mapIndexed { i, el ->
-        return newDocument.select("a img")
+                Page(i, document.location(), el.attr("src"))
-            .mapIndexed { i, el -> Page(i, referer.attr("href"), el.attr("src")) }
+            }
     }
 
     override fun imageUrlParse(document: Document) = ""
@@ -172,9 +151,15 @@ class MangaHost : ParsedHttpSource() {
         return GET(page.imageUrl!!, newHeaders)
     }
 
+    private fun String.toStatus() = when {
+        contains("Ativo") -> SManga.ONGOING
+        contains("Completo") -> SManga.COMPLETED
+        else -> SManga.UNKNOWN
+    }
+
     private fun SimpleDateFormat.tryParseTime(date: String): Long {
         return try {
-            parse(date).time
+            parse(date)!!.time
         } catch (e: ParseException) {
             0L
         }
@@ -184,19 +169,16 @@ class MangaHost : ParsedHttpSource() {
 
     private fun String.toLargeUrl(): String = replace(IMAGE_REGEX, "_large.")
 
-    private fun Elements.textWithoutLabel(): String = text()!!.substringAfter(":")
+    private fun Elements.textWithoutLabel(): String = text()!!.substringAfter(":").trim()
 
     companion object {
-        private const val USER_AGENT = "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/80.0.3987.163 Safari/537.36"
+        private const val USER_AGENT = "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/84.0.4147.89 Safari/537.36"
 
         private val LANG_REGEX = "( )?\\((PT-)?BR\\)".toRegex()
-        private val IMAGE_REGEX = "_(small|medium)\\.".toRegex()
+        private val IMAGE_REGEX = "_(small|medium|xmedium)\\.".toRegex()
 
-        private val DATE_FORMAT_OLD by lazy { SimpleDateFormat("dd/MM/yyyy", Locale.ENGLISH) }
+        private val DATE_FORMAT by lazy { SimpleDateFormat("MMM dd, yyyy", Locale.ENGLISH) }
-        private val DATE_FORMAT_NEW by lazy { SimpleDateFormat("MMM d, yyyy", Locale.ENGLISH) }
 
-        private const val SCRIPT_BEGIN = "var images = ["
-        private const val SCRIPT_END = "];"
         private val SCRIPT_REGEX = "[\",]".toRegex()
     }
 }