Fuck hitomi

This commit is contained in:
tom5079
2022-02-01 11:39:26 +09:00
parent 2608796929
commit 117d6dcd2b
21 changed files with 743 additions and 445 deletions

View File

@@ -16,31 +16,81 @@
package xyz.quaver.pupil.hitomi
import android.util.Log
import android.webkit.WebView
import kotlinx.coroutines.*
import kotlinx.coroutines.flow.Flow
import kotlinx.coroutines.flow.first
import kotlinx.coroutines.flow.transformWhile
import kotlinx.serialization.ExperimentalSerializationApi
import kotlinx.coroutines.Job
import kotlinx.coroutines.asCoroutineDispatcher
import kotlinx.coroutines.runBlocking
import kotlinx.coroutines.withContext
import kotlinx.serialization.Serializable
import kotlinx.serialization.decodeFromString
import kotlinx.serialization.encodeToString
import kotlinx.serialization.json.Json
import xyz.quaver.pupil.*
import java.util.*
import kotlin.coroutines.resume
import kotlin.coroutines.suspendCoroutine
import okhttp3.Request
import xyz.quaver.pupil.client
import xyz.quaver.pupil.runtime
import java.io.IOException
import java.net.URL
import java.util.concurrent.Executors
const val protocol = "https:"
val evaluationContext = Dispatchers.Main + Job()
@Serializable
data class Artist(
val artist: String,
val url: String
)
@Serializable
data class Group(
val group: String,
val url: String
)
@Serializable
data class Parody(
val parody: String,
val url: String
)
@Serializable
data class Character(
val character: String,
val url: String
)
@Serializable
data class Tag(
val tag: String,
val url: String,
val female: String? = null,
val male: String? = null
)
@Serializable
data class Language(
val galleryid: String,
val url: String,
val language_localname: String,
val name: String
)
@Serializable
data class GalleryInfo(
val id: String,
val title: String,
val japanese_title: String? = null,
val language: String? = null,
val type: String,
val date: String,
val artists: List<Artist>? = null,
val groups: List<Group>? = null,
val parodys: List<Parody>? = null,
val tags: List<Tag>? = null,
val related: List<Int>,
val languages: List<Language>,
val characters: List<Character>? = null,
val scene_indexes: List<Int>,
val files: List<GalleryFiles>
)
/**
* kotlinx.serialization.json.Json object for global use
* properties should not be changed
*
* @see [https://kotlin.github.io/kotlinx.serialization/kotlinx-serialization-core/kotlinx-serialization-core/kotlinx.serialization.json/-json/index.html]
*/
val json = Json {
isLenient = true
ignoreUnknownKeys = true
@@ -48,89 +98,103 @@ val json = Json {
useArrayPolymorphism = true
}
suspend inline fun <reified T> WebView.evaluate(script: String): T = coroutineScope { withTimeout(60000) {
var result: String? = null
typealias HeaderSetter = (Request.Builder) -> Request.Builder
fun URL.readText(settings: HeaderSetter? = null): String {
val request = Request.Builder()
.url(this).let {
settings?.invoke(it) ?: it
}.build()
while (result == null) {
try {
while (!oldWebView && !webViewReady) delay(1000)
return client.newCall(request).execute().also{ if (it.code() != 200) throw IOException("CODE ${it.code()}") }.body()?.use { it.string() } ?: throw IOException()
}
result = if (oldWebView)
"null"
else withContext(evaluationContext) {
suspendCoroutine { continuation ->
evaluateJavascript(script) {
continuation.resume(it)
}
}
fun URL.readBytes(settings: HeaderSetter? = null): ByteArray {
val request = Request.Builder()
.url(this).let {
settings?.invoke(it) ?: it
}.build()
}
} catch (e: CancellationException) {
if (e.message != "reload") result = "null"
}
}
json.decodeFromString(result)
} }
@OptIn(ExperimentalCoroutinesApi::class)
suspend inline fun <reified T> WebView.evaluatePromise(
script: String,
then: String = ".then(result => Callback.onResult(%uid, JSON.stringify(result))).catch(err => Callback.onError(%uid, String.raw`$script`, err.message, err.stack))"
): T = coroutineScope { withTimeout(60000) {
var result: String? = null
while (result == null) {
try {
while (!oldWebView && !webViewReady) delay(1000)
result = if (oldWebView)
"null"
else withContext(evaluationContext) {
val uid = UUID.randomUUID().toString()
val flow: Flow<Pair<String, String?>> = webViewFlow.transformWhile { (currentUid, result) ->
if (currentUid == uid) {
emit(currentUid to result)
}
currentUid != uid
}
launch {
evaluateJavascript((script + then).replace("%uid", "'$uid'"), null)
}
flow.first().second
}
} catch (e: CancellationException) {
if (e.message != "reload") result = "null"
}
}
json.decodeFromString(result)
} }
return client.newCall(request).execute().also { if (it.code() != 200) throw IOException("CODE ${it.code()}") }.body()?.use { it.bytes() } ?: throw IOException()
}
@Suppress("EXPERIMENTAL_API_USAGE")
suspend fun getGalleryInfo(galleryID: Int): GalleryInfo =
webView.evaluatePromise("get_gallery_info($galleryID)")
fun getGalleryInfo(galleryID: Int) =
json.decodeFromString<GalleryInfo>(
URL("$protocol//$domain/galleries/$galleryID.js").readText()
.replace("var galleryinfo = ", "")
)
//common.js
const val domain = "ltn.hitomi.la"
const val galleryblockextension = ".html"
const val galleryblockdir = "galleryblock"
const val nozomiextension = ".nozomi"
val String?.js: String
get() = if (this == null) "null" else "'$this'"
val evaluationContext = Executors.newSingleThreadExecutor().asCoroutineDispatcher() + Job()
@OptIn(ExperimentalSerializationApi::class)
suspend fun urlFromUrlFromHash(galleryID: Int, image: GalleryFiles, dir: String? = null, ext: String? = null, base: String? = null): String =
webView.evaluate(
"""
url_from_url_from_hash(
${galleryID.toString().js},
${Json.encodeToString(image)},
${dir.js}, ${ext.js}, ${base.js}
)
""".trimIndent()
)
object gg {
suspend fun m(g: Int): Int = withContext(evaluationContext) {
runtime.evaluate("gg.m($g)").toString().toInt()
}
suspend fun b(): String = withContext(evaluationContext) {
runtime.evaluate("gg.b").toString()
}
suspend fun s(h: String): String = withContext(evaluationContext) {
runtime.evaluate("gg.s('$h')").toString()
}
}
suspend fun subdomainFromURL(url: String, base: String? = null) : String {
var retval = "b"
if (!base.isNullOrBlank())
retval = base
val b = 16
val r = Regex("""/[0-9a-f]{61}([0-9a-f]{2})([0-9a-f])""")
val m = r.find(url) ?: return "a"
val g = m.groupValues.let { it[2]+it[1] }.toIntOrNull(b)
if (g != null) {
retval = (97+ gg.m(g)).toChar().toString() + retval
}
return retval
}
suspend fun urlFromUrl(url: String, base: String? = null) : String {
return url.replace(Regex("""//..?\.hitomi\.la/"""), "//${subdomainFromURL(url, base)}.hitomi.la/")
}
suspend fun fullPathFromHash(hash: String) : String =
"${gg.b()}${gg.s(hash)}/$hash"
fun realFullPathFromHash(hash: String): String =
hash.replace(Regex("""^.*(..)(.)$"""), "$2/$1/$hash")
suspend fun urlFromHash(galleryID: Int, image: GalleryFiles, dir: String? = null, ext: String? = null) : String {
val ext = ext ?: dir ?: image.name.takeLastWhile { it != '.' }
val dir = dir ?: "images"
return "https://a.hitomi.la/$dir/${fullPathFromHash(image.hash)}.$ext"
}
suspend fun urlFromUrlFromHash(galleryID: Int, image: GalleryFiles, dir: String? = null, ext: String? = null, base: String? = null) =
if (base == "tn")
urlFromUrl("https://a.hitomi.la/$dir/${realFullPathFromHash(image.hash)}.$ext", base)
else
urlFromUrl(urlFromHash(galleryID, image, dir, ext), base)
suspend fun rewriteTnPaths(html: String) {
html.replace(Regex("""//tn\.hitomi\.la/[^/]+/[0-9a-f]/[0-9a-f]{2}/[0-9a-f]{64}""")) { url ->
runBlocking {
urlFromUrl(url.value, "tn")
}
}
}
suspend fun imageUrlFromImage(galleryID: Int, image: GalleryFiles, noWebp: Boolean) : String {
return when {

View File

@@ -17,14 +17,11 @@
package xyz.quaver.pupil.hitomi
import kotlinx.serialization.Serializable
import kotlinx.serialization.decodeFromString
import kotlinx.serialization.json.Json
import xyz.quaver.pupil.webView
@Serializable
data class Gallery(
val related: List<Int>,
val langList: Map<String, String>,
val langList: List<Pair<String, String>>,
val cover: String,
val title: String,
val artists: List<String>,
@@ -36,5 +33,22 @@ data class Gallery(
val tags: List<String>,
val thumbnails: List<String>
)
suspend fun getGallery(galleryID: Int) : Gallery =
webView.evaluatePromise("get_gallery($galleryID)")
suspend fun getGallery(galleryID: Int) : Gallery {
val info = getGalleryInfo(galleryID)
return Gallery(
info.related,
info.languages.map { it.name to it.galleryid },
urlFromUrlFromHash(galleryID, info.files.first(), "webpbigtn", "webp", "tn"),
info.title,
info.artists?.map { it.artist }.orEmpty(),
info.groups?.map { it.group }.orEmpty(),
info.type,
info.language.orEmpty(),
info.parodys?.map { it.parody }.orEmpty(),
info.characters?.map { it.character }.orEmpty(),
info.tags?.map { "${if (it.female.isNullOrEmpty()) "" else "female:"}${if (it.male.isNullOrEmpty()) "" else "male:"}${it.tag}" }.orEmpty(),
info.files.map { urlFromUrlFromHash(galleryID, it, "webpsmalltn", "webp", "tn") }
)
}

View File

@@ -17,7 +17,48 @@
package xyz.quaver.pupil.hitomi
import kotlinx.serialization.Serializable
import xyz.quaver.pupil.webView
import org.jsoup.Jsoup
import java.net.URL
import java.net.URLDecoder
import java.nio.ByteBuffer
import java.nio.ByteOrder
import javax.net.ssl.HttpsURLConnection
import kotlin.io.readText
//galleryblock.js
fun fetchNozomi(area: String? = null, tag: String = "index", language: String = "all", start: Int = -1, count: Int = -1) : Pair<List<Int>, Int> {
val url = when(area) {
null -> "$protocol//$domain/$tag-$language$nozomiextension"
else -> "$protocol//$domain/$area/$tag-$language$nozomiextension"
}
with(URL(url).openConnection() as HttpsURLConnection) {
requestMethod = "GET"
if (start != -1 && count != -1) {
val startByte = start*4
val endByte = (start+count)*4-1
setRequestProperty("Range", "bytes=$startByte-$endByte")
}
connect()
val totalItems = getHeaderField("Content-Range")
.replace(Regex("^[Bb]ytes \\d+-\\d+/"), "").toInt() / 4
val nozomi = ArrayList<Int>()
val arrayBuffer = ByteBuffer
.wrap(inputStream.readBytes())
.order(ByteOrder.BIG_ENDIAN)
while (arrayBuffer.hasRemaining())
nozomi.add(arrayBuffer.int)
return Pair(nozomi, totalItems)
}
}
@Serializable
data class GalleryBlock(
@@ -28,9 +69,24 @@ data class GalleryBlock(
val artists: List<String>,
val series: List<String>,
val type: String,
val language: String?,
val relatedTags: List<String>
val language: String,
val relatedTags: List<String>,
val groups: List<String>
)
suspend fun getGalleryBlock(galleryID: Int) : GalleryBlock =
webView.evaluatePromise("get_gallery_block($galleryID)")
suspend fun getGalleryBlock(galleryID: Int) : GalleryBlock {
val info = getGalleryInfo(galleryID)
return GalleryBlock(
galleryID,
"",
listOf(urlFromUrlFromHash(galleryID, info.files.first(), "webpbigtn", "webp", "tn")),
info.title,
info.artists?.map { it.artist }.orEmpty(),
info.parodys?.map { it.parody }.orEmpty(),
info.type,
info.language.orEmpty(),
info.tags?.map { "${if (it.female.isNullOrEmpty()) "" else "female:"}${if (it.male.isNullOrEmpty()) "" else "male:"}${it.tag}" }.orEmpty(),
info.groups?.map { it.group }.orEmpty()
)
}

View File

@@ -17,19 +17,8 @@
package xyz.quaver.pupil.hitomi
import kotlinx.serialization.Serializable
fun getReferer(galleryID: Int) = "https://hitomi.la/reader/$galleryID.html"
@Serializable
data class GalleryInfo(
val language_localname: String? = null,
val language: String? = null,
val date: String? = null,
val files: List<GalleryFiles>,
val id: Int? = null,
val type: String? = null,
val title: String? = null
)
import xyz.quaver.pupil.hitomi.GalleryInfo
import xyz.quaver.pupil.hitomi.getGalleryInfo
@Serializable
data class GalleryFiles(
@@ -44,6 +33,6 @@ data class GalleryFiles(
//Set header `Referer` to reader url to avoid 403 error
@Deprecated("", replaceWith = ReplaceWith("getGalleryInfo"))
suspend fun getReader(galleryID: Int) : GalleryInfo {
fun getReader(galleryID: Int) : GalleryInfo {
return getGalleryInfo(galleryID)
}

View File

@@ -16,42 +16,313 @@
package xyz.quaver.pupil.hitomi
import android.util.Log
import kotlinx.serialization.ExperimentalSerializationApi
import kotlinx.serialization.Serializable
import kotlinx.serialization.decodeFromString
import kotlinx.serialization.json.Json
import xyz.quaver.pupil.webView
import okhttp3.Request
import xyz.quaver.pupil.client
import java.net.URL
import java.nio.ByteBuffer
import java.nio.ByteOrder
import java.security.MessageDigest
import kotlin.math.min
//searchlib.js
const val separator = "-"
const val extension = ".html"
const val index_dir = "tagindex"
const val galleries_index_dir = "galleriesindex"
const val max_node_size = 464
const val B = 16
const val compressed_nozomi_prefix = "n"
@OptIn(ExperimentalSerializationApi::class)
suspend fun getGalleryIDsForQuery(query: String) : Set<Int> =
webView.evaluatePromise("get_galleryids_for_query('$query')")
val tag_index_version: String by lazy { getIndexVersion("tagindex") }
val galleries_index_version: String by lazy { getIndexVersion("galleriesindex") }
@Serializable
data class Suggestion(val s: String, val t: Int, val u: String, val n: String)
fun sha256(data: ByteArray) : ByteArray {
return MessageDigest.getInstance("SHA-256").digest(data)
}
@OptIn(ExperimentalSerializationApi::class)
suspend fun getSuggestionsForQuery(query: String) : List<Suggestion> =
webView.evaluatePromise(
"get_suggestions_for_query('$query', ++search_serial)",
then = """
.then(r => {
let [results, results_serial] = r;
if (search_serial !== results_serial) {
Callback.onResult(%uid, '[]');
} else {
Callback.onResult(%uid, JSON.stringify(results));
@OptIn(ExperimentalUnsignedTypes::class)
fun hashTerm(term: String) : UByteArray {
return sha256(term.toByteArray()).toUByteArray().sliceArray(0 until 4)
}
fun sanitize(input: String) : String {
return input.replace(Regex("[/#]"), "")
}
fun getIndexVersion(name: String) =
URL("$protocol//$domain/$name/version?_=${System.currentTimeMillis()}").readText()
//search.js
fun getGalleryIDsForQuery(query: String) : Set<Int> {
query.replace("_", " ").let {
if (it.indexOf(':') > -1) {
val sides = it.split(":")
val ns = sides[0]
var tag = sides[1]
var area : String? = ns
var language = "all"
when (ns) {
"female", "male" -> {
area = "tag"
tag = it
}
});
""".trimIndent()
)
"language" -> {
area = null
language = tag
tag = "index"
}
}
@OptIn(ExperimentalSerializationApi::class)
suspend fun getGalleryIDsFromNozomi(area: String?, tag: String, language: String) : Set<Int> {
val jsArea = if (area == null) "null" else "'$area'"
return getGalleryIDsFromNozomi(area, tag, language)
}
return webView.evaluatePromise("""get_galleryids_from_nozomi($jsArea, '$tag', '$language')""")
val key = hashTerm(it)
val field = "galleries"
val node = getNodeAtAddress(field, 0) ?: return emptySet()
val data = bSearch(field, key, node)
if (data != null)
return getGalleryIDsFromData(data)
return emptySet()
}
}
fun getSuggestionsForQuery(query: String) : List<Suggestion> {
query.replace('_', ' ').let {
var field = "global"
var term = it
if (term.indexOf(':') > -1) {
val sides = it.split(':')
field = sides[0]
term = sides[1]
}
val key = hashTerm(term)
val node = getNodeAtAddress(field, 0) ?: return emptyList()
val data = bSearch(field, key, node)
if (data != null)
return getSuggestionsFromData(field, data)
return emptyList()
}
}
data class Suggestion(val s: String, val t: Int, val u: String, val n: String)
fun getSuggestionsFromData(field: String, data: Pair<Long, Int>) : List<Suggestion> {
val url = "$protocol//$domain/$index_dir/$field.$tag_index_version.data"
val (offset, length) = data
if (length > 10000 || length <= 0)
throw Exception("length $length is too long")
val inbuf = getURLAtRange(url, offset.until(offset+length))
val suggestions = ArrayList<Suggestion>()
val buffer = ByteBuffer
.wrap(inbuf)
.order(ByteOrder.BIG_ENDIAN)
val numberOfSuggestions = buffer.int
if (numberOfSuggestions > 100 || numberOfSuggestions <= 0)
throw Exception("number of suggestions $numberOfSuggestions is too long")
for (i in 0.until(numberOfSuggestions)) {
var top = buffer.int
val ns = inbuf.sliceArray(buffer.position().until(buffer.position()+top)).toString(charset("UTF-8"))
buffer.position(buffer.position()+top)
top = buffer.int
val tag = inbuf.sliceArray(buffer.position().until(buffer.position()+top)).toString(charset("UTF-8"))
buffer.position(buffer.position()+top)
val count = buffer.int
val tagname = sanitize(tag)
val u =
when(ns) {
"female", "male" -> "/tag/$ns:$tagname${separator}1$extension"
"language" -> "/index-$tagname${separator}1$extension"
else -> "/$ns/$tagname${separator}all${separator}1$extension"
}
suggestions.add(Suggestion(tag, count, u, ns))
}
return suggestions
}
fun getGalleryIDsFromNozomi(area: String?, tag: String, language: String) : Set<Int> {
val nozomiAddress =
when(area) {
null -> "$protocol//$domain/$compressed_nozomi_prefix/$tag-$language$nozomiextension"
else -> "$protocol//$domain/$compressed_nozomi_prefix/$area/$tag-$language$nozomiextension"
}
val bytes = try {
URL(nozomiAddress).readBytes()
} catch (e: Exception) {
return emptySet()
}
val nozomi = mutableSetOf<Int>()
val arrayBuffer = ByteBuffer
.wrap(bytes)
.order(ByteOrder.BIG_ENDIAN)
while (arrayBuffer.hasRemaining())
nozomi.add(arrayBuffer.int)
return nozomi
}
fun getGalleryIDsFromData(data: Pair<Long, Int>) : Set<Int> {
val url = "$protocol//$domain/$galleries_index_dir/galleries.$galleries_index_version.data"
val (offset, length) = data
if (length > 100000000 || length <= 0)
throw Exception("length $length is too long")
val inbuf = getURLAtRange(url, offset.until(offset+length))
val galleryIDs = mutableSetOf<Int>()
val buffer = ByteBuffer
.wrap(inbuf)
.order(ByteOrder.BIG_ENDIAN)
val numberOfGalleryIDs = buffer.int
val expectedLength = numberOfGalleryIDs*4+4
if (numberOfGalleryIDs > 10000000 || numberOfGalleryIDs <= 0)
throw Exception("number_of_galleryids $numberOfGalleryIDs is too long")
else if (inbuf.size != expectedLength)
throw Exception("inbuf.byteLength ${inbuf.size} != expected_length $expectedLength")
for (i in 0.until(numberOfGalleryIDs))
galleryIDs.add(buffer.int)
return galleryIDs
}
fun getNodeAtAddress(field: String, address: Long) : Node? {
val url =
when(field) {
"galleries" -> "$protocol//$domain/$galleries_index_dir/galleries.$galleries_index_version.index"
"languages" -> "$protocol//$domain/$galleries_index_dir/languages.$galleries_index_version.index"
"nozomiurl" -> "$protocol//$domain/$galleries_index_dir/nozomiurl.$galleries_index_version.index"
else -> "$protocol//$domain/$index_dir/$field.$tag_index_version.index"
}
val nodedata = getURLAtRange(url, address.until(address+ max_node_size))
return decodeNode(nodedata)
}
fun getURLAtRange(url: String, range: LongRange) : ByteArray {
val request = Request.Builder()
.url(url)
.header("Range", "bytes=${range.first}-${range.last}")
.build()
return client.newCall(request).execute().body()?.use { it.bytes() } ?: byteArrayOf()
}
@OptIn(ExperimentalUnsignedTypes::class)
data class Node(val keys: List<UByteArray>, val datas: List<Pair<Long, Int>>, val subNodeAddresses: List<Long>)
@OptIn(ExperimentalUnsignedTypes::class)
fun decodeNode(data: ByteArray) : Node {
val buffer = ByteBuffer
.wrap(data)
.order(ByteOrder.BIG_ENDIAN)
val uData = data.toUByteArray()
val numberOfKeys = buffer.int
val keys = ArrayList<UByteArray>()
for (i in 0.until(numberOfKeys)) {
val keySize = buffer.int
if (keySize == 0 || keySize > 32)
throw Exception("fatal: !keySize || keySize > 32")
keys.add(uData.sliceArray(buffer.position().until(buffer.position()+keySize)))
buffer.position(buffer.position()+keySize)
}
val numberOfDatas = buffer.int
val datas = ArrayList<Pair<Long, Int>>()
for (i in 0.until(numberOfDatas)) {
val offset = buffer.long
val length = buffer.int
datas.add(Pair(offset, length))
}
val numberOfSubNodeAddresses = B +1
val subNodeAddresses = ArrayList<Long>()
for (i in 0.until(numberOfSubNodeAddresses)) {
val subNodeAddress = buffer.long
subNodeAddresses.add(subNodeAddress)
}
return Node(keys, datas, subNodeAddresses)
}
@OptIn(ExperimentalUnsignedTypes::class)
fun bSearch(field: String, key: UByteArray, node: Node) : Pair<Long, Int>? {
fun compareArrayBuffers(dv1: UByteArray, dv2: UByteArray) : Int {
val top = min(dv1.size, dv2.size)
for (i in 0.until(top)) {
if (dv1[i] < dv2[i])
return -1
else if (dv1[i] > dv2[i])
return 1
}
return 0
}
fun locateKey(key: UByteArray, node: Node) : Pair<Boolean, Int> {
for (i in node.keys.indices) {
val cmpResult = compareArrayBuffers(key, node.keys[i])
if (cmpResult <= 0)
return Pair(cmpResult==0, i)
}
return Pair(false, node.keys.size)
}
fun isLeaf(node: Node) : Boolean {
for (subnode in node.subNodeAddresses)
if (subnode != 0L)
return false
return true
}
if (node.keys.isEmpty())
return null
val (there, where) = locateKey(key, node)
if (there)
return node.datas[where]
else if (isLeaf(node))
return null
val nextNode = getNodeAtAddress(field, node.subNodeAddresses[where]) ?: return null
return bSearch(field, key, nextNode)
}