A3Manga: decrypt page URLs for whole multisrc, update domains (#19160)

A3Manga: decrypt page URLs for whole multisrc
This commit is contained in:
Vetle Ledaal 2023-12-02 15:41:35 +00:00 committed by GitHub
parent a17eee9d8c
commit 10a45da680
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
3 changed files with 39 additions and 57 deletions

View File

@ -1,52 +0,0 @@
package eu.kanade.tachiyomi.extension.vi.ocumeo
import eu.kanade.tachiyomi.multisrc.a3manga.A3Manga
import eu.kanade.tachiyomi.source.model.Page
import org.jsoup.Jsoup
import org.jsoup.nodes.Document
class OCuMeo : A3Manga("Ổ Cú Mèo", "https://www.ocumoe.com", "vi") {
override fun pageListParse(document: Document): List<Page> {
val imgListHtml = decodeImgList(document)
return Jsoup.parseBodyFragment(imgListHtml).select("img").mapIndexed { idx, element ->
val encryptedUrl = element.attributes().find { it.key.startsWith("data") }?.value
Page(idx, imageUrl = encryptedUrl?.decodeUrl())
}
}
private fun String.decodeUrl(): String {
// We expect the URL to start with `https://`, where the last 3 characters are encoded.
// The length of the encoded character is not known, but it is the same across all.
// Essentially we are looking for the two encoded slashes, which tells us the length.
val patternIdx = patternsLengthCheck.indexOfFirst { pattern ->
val matchResult = pattern.find(this)
val g1 = matchResult?.groupValues?.get(1)
val g2 = matchResult?.groupValues?.get(2)
g1 == g2 && g1 != null
}
if (patternIdx == -1) {
throw Exception("Failed to decrypt URL")
}
// With a known length we can predict all the encoded characters.
// This is a slightly more expensive pattern, hence the separation.
val matchResult = patternsSubstitution[patternIdx].find(this)
return matchResult?.destructured?.let { (colon, slash, period) ->
this
.replace(colon, ":")
.replace(slash, "/")
.replace(period, ".")
} ?: throw Exception("Failed to reconstruct URL")
}
companion object {
private val patternsLengthCheck: List<Regex> = (20 downTo 1).map { i ->
"""^https.{$i}(.{$i})(.{$i})""".toRegex()
}
private val patternsSubstitution: List<Regex> = (20 downTo 1).map { i ->
"""^https(.{$i})(.{$i}).*(.{$i})(?:webp|jpeg|tiff|.{3})$""".toRegex()
}
}
}

View File

@ -181,8 +181,35 @@ open class A3Manga(
override fun pageListParse(document: Document): List<Page> { override fun pageListParse(document: Document): List<Page> {
val imgListHtml = decodeImgList(document) val imgListHtml = decodeImgList(document)
return Jsoup.parseBodyFragment(imgListHtml).select("img").mapIndexed { idx, it -> return Jsoup.parseBodyFragment(imgListHtml).select("img").mapIndexed { idx, element ->
Page(idx, imageUrl = it.attr("abs:src")) val encryptedUrl = element.attributes().find { it.key.startsWith("data") }?.value
val effectiveUrl = encryptedUrl?.decodeUrl() ?: element.attr("abs:src")
Page(idx, imageUrl = effectiveUrl)
}
}
private fun String.decodeUrl(): String? {
// We expect the URL to start with `https://`, where the last 3 characters are encoded.
// The length of the encoded character is not known, but it is the same across all.
// Essentially we are looking for the two encoded slashes, which tells us the length.
val patternIdx = patternsLengthCheck.indexOfFirst { pattern ->
val matchResult = pattern.find(this)
val g1 = matchResult?.groupValues?.get(1)
val g2 = matchResult?.groupValues?.get(2)
g1 == g2 && g1 != null
}
if (patternIdx == -1) {
return null
}
// With a known length we can predict all the encoded characters.
// This is a slightly more expensive pattern, hence the separation.
val matchResult = patternsSubstitution[patternIdx].find(this)
return matchResult?.destructured?.let { (colon, slash, period) ->
this
.replace(colon, ":")
.replace(slash, "/")
.replace(period, ".")
} }
} }
@ -209,5 +236,12 @@ open class A3Manga(
val dateFormat = SimpleDateFormat("dd/MM/yyyy", Locale.US).apply { val dateFormat = SimpleDateFormat("dd/MM/yyyy", Locale.US).apply {
timeZone = TimeZone.getTimeZone("Asia/Ho_Chi_Minh") timeZone = TimeZone.getTimeZone("Asia/Ho_Chi_Minh")
} }
private val patternsLengthCheck: List<Regex> = (20 downTo 1).map { i ->
"""^https.{$i}(.{$i})(.{$i})""".toRegex()
}
private val patternsSubstitution: List<Regex> = (20 downTo 1).map { i ->
"""^https(.{$i})(.{$i}).*(.{$i})(?:webp|jpeg|tiff|.{3})$""".toRegex()
}
} }
} }

View File

@ -9,11 +9,11 @@ class A3MangaGenerator : ThemeSourceGenerator {
override val themeClass = "A3Manga" override val themeClass = "A3Manga"
override val baseVersionCode: Int = 1 override val baseVersionCode: Int = 2
override val sources = listOf( override val sources = listOf(
SingleLang("A3 Manga", "https://www.a3mnga.com", "vi"), SingleLang("A3 Manga", "https://www.a3manga.info", "vi"),
SingleLang("Team Lanh Lung", "https://teamlanhlung.vip", "vi", sourceName = "Team Lạnh Lùng", overrideVersionCode = 1), SingleLang("Team Lanh Lung", "https://teamlanhlung.me", "vi", sourceName = "Team Lạnh Lùng", overrideVersionCode = 1),
SingleLang("Ngon Phong", "https://www.ngonphong.com", "vi", sourceName = "Ngôn Phong", overrideVersionCode = 1), SingleLang("Ngon Phong", "https://www.ngonphong.com", "vi", sourceName = "Ngôn Phong", overrideVersionCode = 1),
SingleLang("O Cu Meo", "https://www.ocumoe.com", "vi", sourceName = "Ổ Cú Mèo", overrideVersionCode = 1), SingleLang("O Cu Meo", "https://www.ocumoe.com", "vi", sourceName = "Ổ Cú Mèo", overrideVersionCode = 1),
) )