[Madara] Refactoring

src/main/kotlin/org/koitharu/kotatsu/parsers/site/TuMangaOnlineParser.kt

@@ -95,13 +95,13 @@ class TuMangaOnlineParser(context: MangaLoaderContext) : PagedMangaParser(
 					oneShotChapterFromElement(item)
 				}
 			} else {
-				val chapters = mutableListOf<MangaChapter>()
+				val chapters = ChaptersListBuilder(10)
 				doc.select(regularChapterListSelector()).reversed().forEachIndexed { i, item ->
 					val chaptername = item.select("div.col-10.text-truncate").text().replace("&nbsp;", " ").trim()
 					val scanelement = item.select("ul.chapter-list > li")
 					scanelement.forEach { chapters.add(regularChapterFromElement(it, chaptername, i)) }
 				}
-				chapters
+				chapters.toList()
 			},
 		)
 	}

src/main/kotlin/org/koitharu/kotatsu/parsers/site/madara/MadaraParser.kt

@@ -8,15 +8,12 @@ import org.jsoup.nodes.Element
 import org.koitharu.kotatsu.parsers.MangaLoaderContext
 import org.koitharu.kotatsu.parsers.PagedMangaParser
 import org.koitharu.kotatsu.parsers.config.ConfigKey
-import org.koitharu.kotatsu.parsers.exception.ParseException
 import org.koitharu.kotatsu.parsers.model.*
 import org.koitharu.kotatsu.parsers.util.*
-import org.koitharu.kotatsu.parsers.util.cryptoaes.CryptoAES
 import java.text.DateFormat
 import java.text.SimpleDateFormat
 import java.util.*
 
-
 internal abstract class MadaraParser(
 	context: MangaLoaderContext,
 	source: MangaSource,
@@ -396,13 +393,11 @@ internal abstract class MadaraParser(
 		val fullUrl = chapter.url.toAbsoluteUrl(domain)
 		val doc = webClient.httpGet(fullUrl).parseHtml()
 
-		val chapterProtector = doc.selectFirst("#chapter-protector-data")
+		val chapterProtector = doc.getElementById("chapter-protector-data")
-
 		if (chapterProtector == null) {
-			val root = doc.body().selectFirst("div.main-col-inner")?.selectFirst("div.reading-content")
+			val root = doc.body().selectFirstOrThrow("div.main-col-inner").selectFirstOrThrow("div.reading-content")
-				?: throw ParseException("Root not found", fullUrl)
 			return root.select("div.page-break").map { div ->
-				val img = div.selectFirst("img") ?: div.parseFailed("Page image not found")
+				val img = div.selectFirstOrThrow("img")
 				val url = img.src()?.toRelativeUrl(domain) ?: div.parseFailed("Image src not found")
 				MangaPage(
 					id = generateUid(url),
@@ -412,8 +407,6 @@ internal abstract class MadaraParser(
 				)
 			}
 		} else {
-
-
 			val chapterProtectorHtml = chapterProtector.html()
 			val password = chapterProtectorHtml.substringAfter("wpmangaprotectornonce='").substringBefore("';")
 			val chapterData = JSONObject(
@@ -423,166 +416,74 @@ internal abstract class MadaraParser(
 			val salt = chapterData.getString("s").toString().decodeHex()
 			val ciphertext = SALTED + salt + unsaltedCiphertext
 
-			val rawImgArray = CryptoAES.decrypt(Base64.getEncoder().encodeToString(ciphertext), password)
+			val rawImgArray = CryptoAES(context).decrypt(context.encodeBase64(ciphertext), password)
-			val imgArrayString = rawImgArray
+			val imgArrayString = rawImgArray.filterNot { c -> c == '[' || c == ']' || c == '\\' || c == '"' }
-				.replace("[", "")
-				.replace("]", "")
-				.replace("\\", "")
-				.replace("\"", "")
-
 
 			return imgArrayString.split(",").map { url ->
 				MangaPage(
-					id = generateUid(url.toString()),
+					id = generateUid(url),
-					url = url.toString(),
+					url = url,
 					preview = null,
 					source = source,
 				)
 			}
 
 		}
-
-
-	}
-
-	fun String.decodeHex(): ByteArray {
-		check(length % 2 == 0) { "Must have an even length" }
-
-		return chunked(2)
-			.map { it.toInt(16).toByte() }
-			.toByteArray()
-	}
-
-	companion object {
-		const val URL_SEARCH_PREFIX = "slug:"
-		val SALTED = "Salted__".toByteArray(Charsets.UTF_8)
 	}
 
 	protected fun parseChapterDate(dateFormat: DateFormat, date: String?): Long {
-		date ?: return 0
+		// Clean date (e.g. 5th December 2019 to 5 December 2019) before parsing it
+		val d = date?.lowercase() ?: return 0
 		return when {
-			date.endsWith(" ago", ignoreCase = true) -> {
+			d.endsWith(" ago") || d.endsWith(" atrás") || // Handle translated 'ago' in Portuguese.
-				parseRelativeDate(date)
+				d.startsWith("há ") || // other translated 'ago' in Portuguese.
-			}
+				d.endsWith(" hace") || // other translated 'ago' in Spanish
-			// Handle translated 'ago' in Portuguese.
+				d.endsWith(" önce") || // Handle translated 'ago' in Turkish.
-			date.endsWith(" atrás", ignoreCase = true) -> {
+				d.endsWith(" trước") || // Handle translated 'ago' in Viêt Nam.
-				parseRelativeDate(date)
+				d.startsWith("il y a") || // Handle translated 'ago' in French.
-			}
-			// other translated 'ago' in Portuguese.
-			date.startsWith("há ", ignoreCase = true) -> {
-				parseRelativeDate(date)
-			}
-			// other translated 'ago' in Spanish
-			date.endsWith(" hace", ignoreCase = true) -> {
-				parseRelativeDate(date)
-			}
-			// Handle translated 'ago' in Turkish.
-			date.endsWith(" önce", ignoreCase = true) -> {
-				parseRelativeDate(date)
-			}
-			// Handle translated 'ago' in Viêt Nam.
-			date.endsWith(" trước", ignoreCase = true) -> {
-				parseRelativeDate(date)
-			}
-			// Handle translated 'ago' in French.
-			date.startsWith("il y a", ignoreCase = true) -> {
-				parseRelativeDate(date)
-			}
 				//If there is no ago but just a motion of time
-
 				// short Hours
-			date.endsWith(" h", ignoreCase = true) -> {
+				d.endsWith(" h") ||
-				parseRelativeDate(date)
-			}
 				// short Day
-			date.endsWith(" d", ignoreCase = true) -> {
+				d.endsWith(" d") ||
-				parseRelativeDate(date)
-			}
-
 				// Day in Portuguese
-			date.endsWith(" días", ignoreCase = true) -> {
+				d.endsWith(" días") || d.endsWith(" día") ||
-				parseRelativeDate(date)
-			}
-
-			date.endsWith(" día", ignoreCase = true) -> {
-				parseRelativeDate(date)
-			}
 				// Day in French
-			date.endsWith(" jour", ignoreCase = true) -> {
+				d.endsWith(" jour") || d.endsWith(" jours") ||
-				parseRelativeDate(date)
-			}
-
-			date.endsWith(" jours", ignoreCase = true) -> {
-				parseRelativeDate(date)
-			}
 				// Hours in Portuguese
-			date.endsWith(" horas", ignoreCase = true) -> {
+				d.endsWith(" horas") || d.endsWith(" hora") ||
-				parseRelativeDate(date)
-			}
-
-			date.endsWith(" hora", ignoreCase = true) -> {
-				parseRelativeDate(date)
-			}
 				// Hours in french
-			date.endsWith(" heure", ignoreCase = true) -> {
+				d.endsWith(" heure") || d.endsWith(" heures") ||
-				parseRelativeDate(date)
-			}
-
-			date.endsWith(" heures", ignoreCase = true) -> {
-				parseRelativeDate(date)
-			}
 				// Minutes in English
-			date.endsWith(" mins", ignoreCase = true) -> {
+				d.endsWith(" mins") ||
-				parseRelativeDate(date)
-			}
-
 				// Minutes in Portuguese
-			date.endsWith(" minutos", ignoreCase = true) -> {
+				d.endsWith(" minutos") || d.endsWith(" minuto") ||
-				parseRelativeDate(date)
-			}
-
-			date.endsWith(" minuto", ignoreCase = true) -> {
-				parseRelativeDate(date)
-			}
 				//Minutes in French
-			date.endsWith(" minute", ignoreCase = true) -> {
+				d.endsWith(" minute") || d.endsWith(" minutes") -> parseRelativeDate(date)
-				parseRelativeDate(date)
-			}
-
-			date.endsWith(" minutes", ignoreCase = true) -> {
-				parseRelativeDate(date)
-			}
 
 			// Handle 'yesterday' and 'today', using midnight
-			date.startsWith("year", ignoreCase = true) -> {
+			d.startsWith("year") -> Calendar.getInstance().apply {
-				Calendar.getInstance().apply {
 				add(Calendar.DAY_OF_MONTH, -1) // yesterday
 				set(Calendar.HOUR_OF_DAY, 0)
 				set(Calendar.MINUTE, 0)
 				set(Calendar.SECOND, 0)
 				set(Calendar.MILLISECOND, 0)
 			}.timeInMillis
-			}
 
-			date.startsWith("today", ignoreCase = true) -> {
+			d.startsWith("today") -> Calendar.getInstance().apply {
-				Calendar.getInstance().apply {
 				set(Calendar.HOUR_OF_DAY, 0)
 				set(Calendar.MINUTE, 0)
 				set(Calendar.SECOND, 0)
 				set(Calendar.MILLISECOND, 0)
 			}.timeInMillis
-			}
 
-			date.contains(Regex("""\d(st|nd|rd|th)""")) -> {
+			date.contains(Regex("""\d(st|nd|rd|th)""")) -> date.split(" ").map {
-				// Clean date (e.g. 5th December 2019 to 5 December 2019) before parsing it
-				date.split(" ").map {
 				if (it.contains(Regex("""\d\D\D"""))) {
 					it.replace(Regex("""\D"""), "")
 				} else {
 					it
 				}
 			}.let { dateFormat.tryParse(it.joinToString(" ")) }
-			}
 
 			else -> dateFormat.tryParse(date)
 		}
@@ -649,13 +550,27 @@ internal abstract class MadaraParser(
 		return result.ifEmpty { null }
 	}
 
+	private companion object {
+
 		private fun createRequestTemplate() =
-		("action=madara_load_more&page=1&template=madara-core%2Fcontent%2Fcontent-search&vars%5Bs%5D=&vars%5B" + "orderby%5D=meta_value_num&vars%5Bpaged%5D=1&vars%5Btemplate%5D=search&vars%5Bmeta_query" + "%5D%5B0%5D%5Brelation%5D=AND&vars%5Bmeta_query%5D%5Brelation%5D=OR&vars%5Bpost_type" + "%5D=wp-manga&vars%5Bpost_status%5D=publish&vars%5Bmeta_key%5D=_latest_update&vars%5Border" + "%5D=desc&vars%5Bmanga_archives_item_layout%5D=default").split(
+			("action=madara_load_more&page=1&template=madara-core%2Fcontent%2Fcontent-search&vars%5Bs%5D=&vars%5B" +
+				"orderby%5D=meta_value_num&vars%5Bpaged%5D=1&vars%5Btemplate%5D=search&vars%5Bmeta_query" +
+				"%5D%5B0%5D%5Brelation%5D=AND&vars%5Bmeta_query%5D%5Brelation%5D=OR&vars%5Bpost_type" +
+				"%5D=wp-manga&vars%5Bpost_status%5D=publish&vars%5Bmeta_key%5D=_latest_update&vars%5Border" +
+				"%5D=desc&vars%5Bmanga_archives_item_layout%5D=default").split(
 				'&',
 			).map {
 				val pos = it.indexOf('=')
 				it.substring(0, pos) to it.substring(pos + 1)
 			}.toMutableMap()
 
+		fun String.decodeHex(): ByteArray {
+			check(length % 2 == 0) { "Must have an even length" }
+
+			return chunked(2).map { it.toInt(16).toByte() }.toByteArray()
+		}
 
+		const val URL_SEARCH_PREFIX = "slug:"
+		val SALTED = "Salted__".toByteArray(Charsets.UTF_8)
+	}
 }

src/main/kotlin/org/koitharu/kotatsu/parsers/site/madara/pt/LimaScans.kt

@@ -6,12 +6,7 @@ import org.koitharu.kotatsu.parsers.MangaSourceParser
 import org.koitharu.kotatsu.parsers.model.MangaChapter
 import org.koitharu.kotatsu.parsers.model.MangaSource
 import org.koitharu.kotatsu.parsers.site.madara.MadaraParser
-import org.koitharu.kotatsu.parsers.util.attrAsRelativeUrlOrNull
+import org.koitharu.kotatsu.parsers.util.*
-import org.koitharu.kotatsu.parsers.util.domain
-import org.koitharu.kotatsu.parsers.util.generateUid
-import org.koitharu.kotatsu.parsers.util.mapChapters
-import org.koitharu.kotatsu.parsers.util.parseFailed
-import org.koitharu.kotatsu.parsers.util.parseHtml
 import java.text.SimpleDateFormat
 
 @MangaSourceParser("LIMASCANS", "Lima Scans", "pt")
@@ -33,8 +28,8 @@ internal class LimaScans(context: MangaLoaderContext) :
 		val dateFormat = SimpleDateFormat(datePattern, sourceLocale)
 
 		return doc.select(selectchapter).mapChapters(reversed = true) { i, li ->
-			val a = li.selectFirst("a")
+			val a = li.selectFirstOrThrow("a")
-			val href = a?.attrAsRelativeUrlOrNull("href") ?: li.parseFailed("Link is missing")
+			val href = a.attrAsRelativeUrl("href")
 			val link = href + stylepage
 			val dateText = li.selectFirst("a.c-new-tag")?.attr("title") ?: li.selectFirst(selectdate)?.text()
 			val name = a.selectFirst("p")?.text() ?: a.ownText()
@@ -44,10 +39,7 @@ internal class LimaScans(context: MangaLoaderContext) :
 				name = name,
 				number = i + 1,
 				branch = null,
-				uploadDate = parseChapterDate(
+				uploadDate = parseChapterDate(dateFormat, dateText),
-					dateFormat,
-					dateText,
-				),
 				scanlator = null,
 				source = source,
 			)

src/main/kotlin/org/koitharu/kotatsu/parsers/util/CryptoAES.kt

@@ -1,24 +1,23 @@
-package org.koitharu.kotatsu.parsers.util.cryptoaes
+package org.koitharu.kotatsu.parsers.util
 
+import org.koitharu.kotatsu.parsers.MangaLoaderContext
 import java.security.MessageDigest
-import java.util.Arrays
+import java.util.*
-import java.util.Base64
 import javax.crypto.Cipher
 import javax.crypto.spec.IvParameterSpec
 import javax.crypto.spec.SecretKeySpec
 
-/**
- * Conforming with CryptoJS AES method
- */
-@Suppress("unused", "FunctionName")
-object CryptoAES {
-
-	private const val KEY_SIZE = 256
-	private const val IV_SIZE = 128
 private const val HASH_CIPHER = "AES/CBC/PKCS7PADDING"
 private const val AES = "AES"
 private const val KDF_DIGEST = "MD5"
 
+/**
+ * Conforming with CryptoJS AES method
+ */
+class CryptoAES(
+	private val context: MangaLoaderContext,
+) {
+
 	/**
 	 * Decrypt using CryptoJS defaults compatible method.
 	 * Uses KDF equivalent to OpenSSL's EVP_BytesToKey function
@@ -27,19 +26,18 @@ object CryptoAES {
 	 * @param cipherText base64 encoded ciphertext
 	 * @param password passphrase
 	 */
+	@Throws(Exception::class)
 	fun decrypt(cipherText: String, password: String): String {
-		try {
+		val ctBytes = context.decodeBase64(cipherText)
-			val ctBytes = Base64.getDecoder().decode(cipherText)
 		val saltBytes = Arrays.copyOfRange(ctBytes, 8, 16)
 		val cipherTextBytes = Arrays.copyOfRange(ctBytes, 16, ctBytes.size)
-			val md5: MessageDigest = MessageDigest.getInstance("MD5")
+		val md5: MessageDigest = MessageDigest.getInstance(KDF_DIGEST)
 		val keyAndIV = generateKeyAndIV(32, 16, 1, saltBytes, password.toByteArray(Charsets.UTF_8), md5)
-			return decryptAES(cipherTextBytes,
+		return decryptAES(
-				keyAndIV?.get(0) ?: ByteArray(32),
+			cipherTextBytes,
-				keyAndIV?.get(1) ?: ByteArray(16))
+			keyAndIV.getOrNull(0) ?: ByteArray(32),
-		} catch (e: Exception) {
+			keyAndIV.getOrNull(1) ?: ByteArray(16),
-			return ""
+		)
-		}
 	}
 
 	/**
@@ -49,13 +47,10 @@ object CryptoAES {
 	 * @param keyBytes key as a bytearray
 	 * @param ivBytes iv as a bytearray
 	 */
+	@Throws(Exception::class)
 	fun decrypt(cipherText: String, keyBytes: ByteArray, ivBytes: ByteArray): String {
-		return try {
+		val cipherTextBytes = context.decodeBase64(cipherText)
-			val cipherTextBytes = Base64.getDecoder().decode(cipherText)
+		return decryptAES(cipherTextBytes, keyBytes, ivBytes)
-			decryptAES(cipherTextBytes, keyBytes, ivBytes)
-		} catch (e: Exception) {
-			""
-		}
 	}
 
 	/**
@@ -65,15 +60,12 @@ object CryptoAES {
 	 * @param keyBytes key as a bytearray
 	 * @param ivBytes iv as a bytearray
 	 */
+	@Throws(Exception::class)
 	private fun decryptAES(cipherTextBytes: ByteArray, keyBytes: ByteArray, ivBytes: ByteArray): String {
-		return try {
 		val cipher = Cipher.getInstance(HASH_CIPHER)
 		val keyS = SecretKeySpec(keyBytes, AES)
 		cipher.init(Cipher.DECRYPT_MODE, keyS, IvParameterSpec(ivBytes))
-			cipher.doFinal(cipherTextBytes).toString(Charsets.UTF_8)
+		return cipher.doFinal(cipherTextBytes).toString(Charsets.UTF_8)
-		} catch (e: Exception) {
-			""
-		}
 	}
 
 	/**
@@ -92,12 +84,20 @@ object CryptoAES {
 	 * @param md the message digest algorithm to use
 	 * @return an two-element array with the generated key and IV
 	 */
-	private fun generateKeyAndIV(keyLength: Int, ivLength: Int, iterations: Int, salt: ByteArray, password: ByteArray, md: MessageDigest): Array<ByteArray?>? {
+	@Suppress("SameParameterValue")
+	@Throws(Exception::class)
+	private fun generateKeyAndIV(
+		keyLength: Int,
+		ivLength: Int,
+		iterations: Int,
+		salt: ByteArray?,
+		password: ByteArray,
+		md: MessageDigest,
+	): Array<ByteArray?> {
 		val digestLength = md.digestLength
 		val requiredLength = (keyLength + ivLength + digestLength - 1) / digestLength * digestLength
 		val generatedData = ByteArray(requiredLength)
 		var generatedLength = 0
-		return try {
 		md.reset()
 
 		// Repeat process until sufficient data has been generated
@@ -121,12 +121,6 @@ object CryptoAES {
 		val result = arrayOfNulls<ByteArray>(2)
 		result[0] = generatedData.copyOfRange(0, keyLength)
 		if (ivLength > 0) result[1] = generatedData.copyOfRange(keyLength, keyLength + ivLength)
-			result
+		return result
-		} catch (e: Exception) {
-			throw e
-		} finally {
-			// Clean out temporary data
-			Arrays.fill(generatedData, 0.toByte())
-		}
 	}
 }