Remove jsoup usage and clean up

README.md

@@ -98,7 +98,6 @@ Here is the libraries
 - [Coil](https://coil-kt.github.io/coil/)
 - [Compose Destinations](https://composedestinations.rafaelcosta.xyz/)
 - [libarchive](https://www.libarchive.org/)
-- [jsoup](https://jsoup.org/)
 
 # License
 

app/build.gradle.kts

@@ -240,7 +240,6 @@ dependencies {
 
     implementation(libs.kotlinx.coroutines.android)
     implementation(libs.kotlinx.datetime)
-    implementation(libs.jsoup)
 
     coreLibraryDesugaring(libs.desugar)
 

app/src/main/kotlin/com/hippo/ehviewer/client/EhEngine.kt

@@ -178,7 +178,8 @@ object EhEngine {
         location.takeIf { "bounce_login" !in it } ?: throw NotLoggedInException()
     }
 
-    suspend fun getTorrentList(url: String, gid: Long, token: String): TorrentResult {
+    suspend fun getTorrentList(gid: Long, token: String): TorrentResult {
+        val url = EhUrl.getTorrentUrl(gid, token)
         val referer = EhUrl.getGalleryDetailUrl(gid, token)
         return ehRequest(url, referer).fetchUsingAsByteBuffer(TorrentParser::parse)
     }
@@ -188,7 +189,8 @@ object EhEngine {
         "${id}x$key"
     }
 
-    suspend fun getArchiveList(url: String, gid: Long, token: String): ArchiveParser.Result {
+    suspend fun getArchiveList(gid: Long, token: String): ArchiveParser.Result {
+        val url = EhUrl.getArchiveUrl(gid, token)
         val funds = if (EhUtils.isExHentai) getFunds() else null
         return ehRequest(url, EhUrl.getGalleryDetailUrl(gid, token))
             .fetchUsingAsByteBuffer(ArchiveParser::parse.partially2(funds))
@@ -304,7 +306,7 @@ object EhEngine {
         .fetchUsingAsText(FavoritesParser::parseNote)
 
     suspend fun downloadArchive(gid: Long, token: String, res: String, isHath: Boolean): String? {
-        val url = EhUrl.getDownloadArchive(gid, token)
+        val url = EhUrl.getArchiveUrl(gid, token)
         val referer = EhUrl.getGalleryDetailUrl(gid, token)
         val request = ehRequest(url, referer, EhUrl.origin) {
             formBody {

app/src/main/kotlin/com/hippo/ehviewer/client/EhUrl.kt

@@ -125,7 +125,12 @@ object EhUrl {
         addQueryParameter("act", "addfav")
     }.buildString()
 
-    fun getDownloadArchive(gid: Long, token: String) = ehUrl("archiver.php") {
+    fun getArchiveUrl(gid: Long, token: String) = ehUrl("archiver.php") {
+        addQueryParameter("gid", "$gid")
+        addQueryParameter("token", token)
+    }.buildString()
+
+    fun getTorrentUrl(gid: Long, token: String) = ehUrl("gallerytorrents.php") {
         addQueryParameter("gid", "$gid")
         addQueryParameter("token", token)
     }.buildString()

app/src/main/kotlin/com/hippo/ehviewer/client/data/GalleryDetail.kt

@@ -44,8 +44,6 @@ data class GalleryDetail(
     var apiUid: Long = -1L,
     var apiKey: String? = null,
     var torrentCount: Int = 0,
-    var torrentUrl: String? = null,
-    var archiveUrl: String? = null,
     var parent: String? = null,
     var newerVersions: List<BaseGalleryInfo> = emptyList(),
     var visible: String? = null,

app/src/main/kotlin/com/hippo/ehviewer/client/parser/HomeParser.kt

@@ -25,9 +25,9 @@ object HomeParser {
         ?: throw ParseException("Parse torrent key error")
 
     fun parseFunds(body: String): Funds {
-        FundsRegex.find(body)?.groupValues?.run {
-            val fundsC = ParserUtils.parseInt(get(1), 0)
-            val fundsGP = ParserUtils.parseInt(get(2), 0)
+        FundsRegex.find(body)?.run {
+            val fundsC = ParserUtils.parseInt(groupValues[1], 0)
+            val fundsGP = ParserUtils.parseInt(groupValues[2], 0)
             return Funds(fundsGP, fundsC)
         }
         throw ParseException("Parse funds error")

app/src/main/kotlin/com/hippo/ehviewer/client/parser/ParserUtils.kt

@@ -18,8 +18,6 @@ package com.hippo.ehviewer.client.parser
 import com.hippo.ehviewer.util.toEpochMillis
 import com.hippo.ehviewer.util.toIntOrDefault
 import com.hippo.ehviewer.util.toLocalDateTime
-import com.hippo.ehviewer.util.toLongOrDefault
-import com.hippo.ehviewer.util.unescapeXml
 import kotlinx.datetime.LocalDateTime
 import kotlinx.datetime.format.char
 
@@ -41,9 +39,5 @@ object ParserUtils {
 
     fun parseDate(str: String): Long = formatter.parse(str).toEpochMillis()
 
-    fun trim(str: String?): String = str?.unescapeXml()?.trim().orEmpty()
-
-    fun parseInt(str: String?, defValue: Int): Int = trim(str).replace(",", "").toIntOrDefault(defValue)
-
-    fun parseLong(str: String?, defValue: Long): Long = trim(str).replace(",", "").toLongOrDefault(defValue)
+    fun parseInt(str: String, defValue: Int): Int = str.replace(",", "").toIntOrDefault(defValue)
 }

app/src/main/kotlin/com/hippo/ehviewer/ui/GalleryInfoBottomSheet.kt

@@ -58,7 +58,7 @@ private fun GalleryDetail.content() = arrayOf(
     R.string.key_rating_count to ratingCount.toString(),
     R.string.key_rating to rating.toString(),
     R.string.key_torrents to torrentCount.toString(),
-    R.string.key_torrent_url to torrentUrl,
+    R.string.key_torrent_url to EhUrl.getTorrentUrl(gid, token),
     R.string.favorite_name to favoriteName,
 )
 

app/src/main/kotlin/com/hippo/ehviewer/ui/screen/GalleryDetailContent.kt

@@ -551,7 +551,7 @@ fun BelowHeader(galleryDetail: GalleryDetail, voteTag: VoteTag) {
         val archiveResult = remember(galleryDetail) {
             async(Dispatchers.IO + Job(), CoroutineStart.LAZY) {
                 with(galleryDetail) {
-                    EhEngine.getArchiveList(archiveUrl!!, gid, token)
+                    EhEngine.getArchiveList(gid, token)
                 }
             }
         }
@@ -598,7 +598,7 @@ fun BelowHeader(galleryDetail: GalleryDetail, voteTag: VoteTag) {
         val torrentResult = remember(galleryDetail) {
             async(Dispatchers.IO + Job(), CoroutineStart.LAZY) {
                 parZip(
-                    { EhEngine.getTorrentList(galleryDetail.torrentUrl!!, galleryDetail.gid, galleryDetail.token) },
+                    { EhEngine.getTorrentList(galleryDetail.gid, galleryDetail.token) },
                     { EhEngine.getTorrentKey() },
                     { list, key -> list to key },
                 )

app/src/main/kotlin/com/hippo/ehviewer/util/StringUtils.kt

@@ -1,15 +1,16 @@
 package com.hippo.ehviewer.util
 
-import org.jsoup.parser.Parser
+private val EntityRegex = "&(?:amp|lt|gt|quot|#039);".toRegex()
+private val EntityMap = mapOf(
+    "&" to "&",
+    "&lt;" to "<",
+    "&gt;" to ">",
+    "&quot;" to "\"",
+    "&#039;" to "'", // HTML uses &#039; instead of &apos;
+)
 
-fun String.unescapeXml(): String = Parser.unescapeEntities(this, true)
-
-inline infix fun <T> CharSequence.trimAnd(block: CharSequence.() -> T): T = block(trim())
+fun String.unescapeXml() = replace(EntityRegex) { EntityMap[it.value]!! }
 
 fun String.toIntOrDefault(defaultValue: Int): Int = toIntOrNull() ?: defaultValue
 
-fun String.toLongOrDefault(defaultValue: Long): Long = toLongOrNull() ?: defaultValue
-
-fun String.toFloatOrDefault(defaultValue: Float): Float = toFloatOrNull() ?: defaultValue
-
 fun String?.containsIgnoreCase(other: String) = this?.contains(other, true) == true

app/src/main/rust/src/parser/detail.rs

@@ -14,8 +14,6 @@ pub struct GalleryDetail {
     pub apiUid: i64,
     pub apiKey: Option<String>,
     pub torrentCount: i32,
-    pub torrentUrl: Option<String>,
-    pub archiveUrl: Option<String>,
     pub parent: Option<String>,
     pub newerVersions: Vec<BaseGalleryInfo>,
     pub visible: Option<String>,
@@ -148,22 +146,16 @@ fn parse_detail(gd: &mut GalleryDetail, dom: &VDom, parser: &Parser, body: &str)
     gd.apiKey = Some(caps[4].to_string());
     gd.galleryInfo.rating = caps[5].parse().ok()?;
 
-    let torrent_regex = regex!(
-        r#"<a[^<>]*onclick="return popUp\('([^']+)'[^)]+\)">Torrent Download \((\d+)\)</a>"#
-    );
-    let caps = torrent_regex.captures(body)?;
-    gd.torrentUrl = unescape(&caps[1]).ok().map(|s| s.to_string());
-    gd.torrentCount = caps[2].parse().ok()?;
-
-    let archive_regex =
-        regex!(r#"<a[^<>]*onclick="return popUp\('([^']+)'[^)]+\)">Archive Download</a>"#);
-    let caps = archive_regex.captures(body)?;
-    gd.archiveUrl = unescape(&caps[1]).ok().map(|s| s.to_string());
+    let regex = regex!(r#"[^(]+\(([^)]+)"#);
+    let gd5 = dom.get_element_by_id("gd5")?.get(parser)?.as_tag()?;
+    let node = gd5.query_selector(parser, "a[onclick]")?.nth(1)?;
+    let text = node.get(parser)?.inner_text(parser);
+    let caps = regex.captures(&text)?;
+    gd.torrentCount = caps[1].parse().ok()?;
 
     let gd1 = dom.get_element_by_id("gd1")?.get(parser)?.as_tag()?;
     let style = get_tag_attr(get_first_child(gd1, parser)?, "style")?;
-    let cover_regex = regex!(r#"[^(]+\(([^)]+)"#);
-    let caps = cover_regex.captures(style)?;
+    let caps = regex.captures(style)?;
     gd.galleryInfo.thumbKey = get_thumb_key(&caps[1]);
 
     let title = dom.get_element_by_id("gn")?.get(parser)?.inner_text(parser);

app/src/main/rust/src/parser/list.rs

@@ -96,11 +96,8 @@ fn parse_uploader_and_pages(str: &str) -> (Option<String>, bool, i32) {
         regex!(r#"<a href="https://e[x-]hentai.org/uploader/.*?">(.*?)</a>|(\(Disowned\))"#)
             .captures(str)
             .map(|grp| {
-                grp.get(1)
-                    .or_else(|| grp.get(2))
-                    .unwrap()
-                    .as_str()
-                    .to_string()
+                let str = grp.get(1).or_else(|| grp.get(2)).unwrap().as_str();
+                unescape(str).as_deref().unwrap_or(str).to_string()
             });
     let pages = match regex!(r"<div>(\d+) pages</div>").captures(str) {
         None => 0,
@@ -155,7 +152,8 @@ fn parse_gallery_info(node: &Node, parser: &Parser) -> Option<BaseGalleryInfo> {
             None => ("".to_string(), None),
             Some(node) => (
                 node.inner_text(parser).trim().to_string(),
-                get_node_attr(node, "title").map(str::to_string),
+                get_node_attr(node, "title")
+                    .map(|s| unescape(s).as_deref().unwrap_or(s).to_string()),
             ),
         };
     let ir = get_first_element_by_class_name(node, parser, "ir")?
@@ -168,8 +166,11 @@ fn parse_gallery_info(node: &Node, parser: &Parser) -> Option<BaseGalleryInfo> {
             None => (None, false, 0),
             Some(node) => parse_uploader_and_pages(&node.get(parser)?.inner_html(parser)),
         };
-    let favorite_note = get_element_by_id(node, parser, format!("favnote_{gid}").as_str())
-        .map(|e| e.inner_text(parser).to_string());
+    let favorite_note =
+        get_element_by_id(node, parser, format!("favnote_{gid}").as_str()).map(|e| {
+            let str = e.inner_text(parser);
+            unescape(&str).as_deref().unwrap_or(&str).to_string()
+        });
     Some(BaseGalleryInfo {
         gid,
         token,

app/src/main/rust/src/parser/profile.rs

@@ -1,5 +1,6 @@
 use crate::{EhError, get_node_handle_attr, get_tag_attr, select_first};
 use anyhow::{Context, Result};
+use quick_xml::escape::unescape;
 use serde::Serialize;
 use tl::{Parser, VDom};
 
@@ -26,7 +27,8 @@ pub fn parse_profile(dom: &VDom, parser: &Parser) -> Result<Profile> {
         .next()
         .and_then(|n| {
             let mut iter = n.get(parser)?.as_tag()?.query_selector(parser, "div")?;
-            let display_name = iter.next()?.get(parser)?.inner_text(parser).to_string();
+            let str = iter.next()?.get(parser)?.inner_text(parser);
+            let display_name = unescape(&str).as_deref().unwrap_or(&str).to_string();
             let avatar = iter.next().and_then(|n| {
                 let img = select_first(n.get(parser)?.as_tag()?, parser, "img")?;
                 get_tag_attr(img, "src").map(|src| {

docs/README/ja.md

@@ -98,7 +98,6 @@
 - [Coil](https://coil-kt.github.io/coil/)
 - [Compose Destinations](https://composedestinations.rafaelcosta.xyz/)
 - [libarchive](https://www.libarchive.org/)
-- [jsoup](https://jsoup.org/)
 
 # ライセンス
 

docs/README/zh-cn.md

@@ -97,7 +97,6 @@
 - [Coil](https://coil-kt.github.io/coil/)
 - [Compose Destinations](https://composedestinations.rafaelcosta.xyz/)
 - [libarchive](https://www.libarchive.org/)
-- [jsoup](https://jsoup.org/)
 
 # 许可证
 

docs/README/zh-tw.md

@@ -95,7 +95,6 @@
 - [Coil](https://coil-kt.github.io/coil/)
 - [Compose Destinations](https://composedestinations.rafaelcosta.xyz/)
 - [libarchive](https://www.libarchive.org/)
-- [jsoup](https://jsoup.org/)
 
 # 許可證書
 

gradle/libs.versions.toml

@@ -85,7 +85,6 @@ material-kolor = "com.materialkolor:material-kolor:2.1.1"
 desugar = "com.android.tools:desugar_jdk_libs:2.1.5"
 
 diff = "io.github.petertrr:kotlin-multiplatform-diff:1.1.0"
-jsoup = "org.jsoup:jsoup:1.20.1"
 
 kotlinx-coroutines-android = { module = "org.jetbrains.kotlinx:kotlinx-coroutines-android", version.ref = "kotlinx-coroutines" }
 kotlinx-datetime = { module = "org.jetbrains.kotlinx:kotlinx-datetime", version.ref = "kotlinx-datetime" }