Fix incorrect ResponseBody parsing (#12893)

* Fix incorrect ResponseBody parsing

* enable rate limit

src/en/koushoku/build.gradle

@@ -5,7 +5,7 @@ ext {
     extName = 'Koushoku'
     pkgNameSuffix = 'en.koushoku'
     extClass = '.Koushoku'
-    extVersionCode = 12
+    extVersionCode = 13
     isNsfw = true
 }
 

src/en/koushoku/src/eu/kanade/tachiyomi/extension/en/koushoku/Koushoku.kt

@@ -13,6 +13,7 @@ import eu.kanade.tachiyomi.source.online.ParsedHttpSource
 import eu.kanade.tachiyomi.util.asJsoup
 import okhttp3.Headers
 import okhttp3.HttpUrl
+import okhttp3.HttpUrl.Companion.toHttpUrl
 import okhttp3.HttpUrl.Companion.toHttpUrlOrNull
 import okhttp3.OkHttpClient
 import okhttp3.Request
@@ -20,15 +21,15 @@ import okhttp3.Response
 import org.jsoup.nodes.Document
 import org.jsoup.nodes.Element
 import rx.Observable
-import java.net.URL
 
 class Koushoku : ParsedHttpSource() {
     companion object {
         const val PREFIX_ID_SEARCH = "id:"
 
-        val archiveRegex = "/archive/(\\d+)".toRegex()
+        const val thumbnailSelector = "figure img"
-        const val thumbnailSelector = ".thumbnail img"
+        const val magazinesSelector = ".metadata a[href^='/magazines/']"
-        const val magazinesSelector = ".metadata .magazines a"
+
+        private val PATTERN_IMAGES = "(.+/)(\\d+)(.*)".toRegex()
     }
 
     override val baseUrl = "https://koushoku.org"
@@ -42,17 +43,17 @@ class Koushoku : ParsedHttpSource() {
         .build()
 
     override fun headersBuilder(): Headers.Builder = super.headersBuilder()
-        .set("User-Agent", "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/103.0.0.0 Safari/537.36")
+        .set("User-Agent", "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/106.0.0.0 Safari/537.36")
         .add("Referer", "$baseUrl/")
 
     override fun latestUpdatesRequest(page: Int) = GET("$baseUrl/?page=$page", headers)
     override fun latestUpdatesSelector() = "#archives.feed .entries > .entry"
-    override fun latestUpdatesNextPageSelector() = "#archives.feed .pagination .next"
+    override fun latestUpdatesNextPageSelector() = "footer nav li:has(a.active) + li:not(:last-child) > a"
 
     override fun latestUpdatesFromElement(element: Element) = SManga.create().apply {
-        setUrlWithoutDomain(element.select("a").attr("href"))
+        setUrlWithoutDomain(element.selectFirst("a").attr("href"))
-        title = element.select(".title").text()
+        title = element.selectFirst("[title]").attr("title")
-        thumbnail_url = element.select(thumbnailSelector).attr("src")
+        thumbnail_url = element.selectFirst(thumbnailSelector).absUrl("src")
     }
 
     private fun searchMangaByIdRequest(id: String) = GET("$baseUrl/archive/$id", headers)
@@ -131,13 +132,13 @@ class Koushoku : ParsedHttpSource() {
     }
 
     override fun mangaDetailsParse(document: Document) = SManga.create().apply {
-        title = document.selectFirst(".metadata .title").text()
+        title = document.selectFirst(".metadata > h1").text()
 
         // Reuse cover from browse
-        thumbnail_url = document.selectFirst(thumbnailSelector).attr("src")
+        thumbnail_url = document.selectFirst(thumbnailSelector).absUrl("src")
             .replace(Regex("/\\d+\\.webp\$"), "/288.webp")
 
-        artist = document.select(".metadata .artists a, .metadata .circles a")
+        artist = document.select(".metadata a[href^='/artists/'], .metadata a[href^='/circles/']")
             .joinToString { it.text() }
         author = artist
         genre = document.select(".metadata .tags a, $magazinesSelector")
@@ -153,7 +154,7 @@ class Koushoku : ParsedHttpSource() {
             SChapter.create().apply {
                 setUrlWithoutDomain(response.request.url.encodedPath)
                 name = "Chapter"
-                date_upload = document.select(".metadata .published td:nth-child(2)")
+                date_upload = document.select("tr > td:first-child:contains(Uploaded Date) + td")
                     .attr("data-unix").toLong() * 1000
             }
         )
@@ -171,22 +172,26 @@ class Koushoku : ParsedHttpSource() {
         if (totalPages == 0)
             throw UnsupportedOperationException("Error: Empty pages (try Webview)")
 
-        val id = archiveRegex.find(document.location())?.groups?.get(1)?.value
+        val url = document.selectFirst(".main img, main img").absUrl("src")
-        if (id.isNullOrEmpty())
+        val match = PATTERN_IMAGES.find(url)!!
-            throw UnsupportedOperationException("Error: Unknown archive id")
+        val prefix = match.groupValues[1]
-
+        val suffix = match.groupValues[3]
-        val url = URL(document.selectFirst(".main img, main img").attr("src"))
-        val origin = "${url.protocol}://${url.host}"
 
         return (1..totalPages).map {
-            Page(it, "", "$origin/data/$id/$it.jpg")
+            Page(it, "", "$prefix$it$suffix")
         }
     }
 
     override fun imageRequest(page: Page): Request {
-        val newHeaders = headersBuilder()
+        val url = page.imageUrl!!.toHttpUrl()
-            .add("Origin", baseUrl)
+
-            .build()
+        val newHeaders = if (baseUrl.toHttpUrl().host != url.host) {
+            headersBuilder()
+                .add("Origin", baseUrl)
+                .build()
+        } else {
+            headers
+        }
 
         return GET(page.imageUrl!!, newHeaders)
     }
@@ -257,17 +262,17 @@ class Koushoku : ParsedHttpSource() {
             append("\n")
         }
 
-        val parodies = document.select(".metadata .parodies a")
+        val parodies = document.select(".metadata a[href^='/parodies/']")
         if (parodies.isNotEmpty()) {
             append("Parodies: ")
             append(parodies.joinToString { it.text() })
             append("\n")
         }
 
-        val pages = document.selectFirst(".metadata .pages td:nth-child(2)")
+        val pages = document.selectFirst("tr > td:first-child:contains(Pages) + td")
         append("Pages: ").append(pages.text()).append("\n")
 
-        val size = document.selectFirst(".metadata .size td:nth-child(2)")
+        val size = document.selectFirst("tr > td:first-child:contains(Size) + td")
         append("Size: ").append(size.text())
     }
 }

src/en/koushoku/src/eu/kanade/tachiyomi/extension/en/koushoku/KoushokuWebViewInterceptor.kt

@@ -1,6 +1,5 @@
 package eu.kanade.tachiyomi.extension.en.koushoku
 
-import android.annotation.SuppressLint
 import android.app.Application
 import android.os.Handler
 import android.os.Looper
@@ -21,46 +20,48 @@ class KoushokuWebViewInterceptor : Interceptor {
         val response = chain.proceed(request)
 
         if (response.headers("Content-Type").any { it.contains("text/html") }) {
-            val responseBody = response.peekBody(1 * 1024 * 1024).toString()
+            val responseBody = response.peekBody(1 * 1024 * 1024).string()
-            val document = Jsoup.parse(responseBody)
+            if (response.code == 403) {
-
+                val document = Jsoup.parse(responseBody)
-            if (document.selectFirst("h1")?.text()?.contains(Regex("banned$")) == true) {
+                if (document.selectFirst("h1")?.text()?.contains(Regex("banned$")) == true) {
-                throw IOException("You have been banned. Check WebView for details.")
+                    throw IOException("You have been banned. Check WebView for details.")
+                }
             }
 
-            try {
+            if (response.networkResponse != null) {
-                proceedWithWebView(response)
+                try {
-            } catch (e: Exception) {
+                    proceedWithWebView(response, responseBody)
-                throw IOException(e)
+                } catch (e: Exception) {
+                    throw IOException(e)
+                }
             }
         }
 
         return response
     }
 
-    @SuppressLint("SetJavaScriptEnabled")
+    private fun proceedWithWebView(response: Response, responseBody: String) {
-    private fun proceedWithWebView(response: Response) {
         val latch = CountDownLatch(1)
-
         val handler = Handler(Looper.getMainLooper())
 
         handler.post {
-            val webview = WebView(Injekt.get<Application>())
+            val webView = WebView(Injekt.get<Application>())
-            with(webview.settings) {
+            with(webView.settings) {
-                javaScriptEnabled = true
                 loadsImagesAutomatically = false
                 userAgentString = response.request.header("User-Agent")
             }
 
-            webview.webViewClient = object : WebViewClient() {
+            webView.webViewClient = object : WebViewClient() {
                 override fun onPageFinished(view: WebView, url: String) {
+                    webView.stopLoading()
+                    webView.destroy()
                     latch.countDown()
                 }
             }
 
-            webview.loadDataWithBaseURL(
+            webView.loadDataWithBaseURL(
                 response.request.url.toString(),
-                response.peekBody(1 * 1024 * 1024).toString(),
+                responseBody,
                 "text/html",
                 "utf-8",
                 null