From 257bfe843ee4345420b4f01cb26001f0ff3b0761 Mon Sep 17 00:00:00 2001
From: FooIbar <118464521+FooIbar@users.noreply.github.com>
Date: Sun, 26 Jul 2026 22:45:05 +0800
Subject: [PATCH] Add missing `nbsp` and `times` entities

---
 app/src/main/rust/src/parser/config.rs           |  2 +-
 app/src/main/rust/src/parser/detail.rs           |  2 +-
 app/src/main/rust/src/parser/fav.rs              |  2 +-
 app/src/main/rust/src/parser/list.rs             |  2 +-
 app/src/main/rust/src/parser/profile.rs          |  2 +-
 app/src/main/rust/src/parser/torrent.rs          |  2 +-
 app/src/main/rust/src/parser/utils.rs            | 16 ++++++++++++++++
 .../kotlin/com/ehviewer/core/util/StringUtils.kt |  4 +++-
 8 files changed, 25 insertions(+), 7 deletions(-)

diff --git a/app/src/main/rust/src/parser/config.rs b/app/src/main/rust/src/parser/config.rs
index 454267d5d..2cb8eaf63 100644
--- a/app/src/main/rust/src/parser/config.rs
+++ b/app/src/main/rust/src/parser/config.rs
@@ -1,5 +1,5 @@
+use crate::parser::utils::unescape;
 use crate::regex;
-use quick_xml::escape::unescape;
 
 pub fn parse_fav_cat(html: &str) -> Vec<String> {
     regex!("<input type=\"text\" name=\"favorite_\\d\" value=\"([^\"]+)\"")
diff --git a/app/src/main/rust/src/parser/detail.rs b/app/src/main/rust/src/parser/detail.rs
index 4719181b8..4421d0c7b 100644
--- a/app/src/main/rust/src/parser/detail.rs
+++ b/app/src/main/rust/src/parser/detail.rs
@@ -1,9 +1,9 @@
 use crate::parser::list::{BaseGalleryInfo, get_thumb_key, parse_token_and_gid, to_category_i32};
+use crate::parser::utils::unescape;
 use crate::{EhError, get_first_child, get_first_element_by_class_name, get_node_handle_attr};
 use crate::{get_tag_attr, get_vdom_first_element_by_class_name, regex};
 use anyhow::{Context, Result, bail};
 use chrono::naive::NaiveDateTime;
-use quick_xml::escape::unescape;
 use serde::Serialize;
 use tl::{Bytes, Node, NodeHandle, Parser, VDom};
 
diff --git a/app/src/main/rust/src/parser/fav.rs b/app/src/main/rust/src/parser/fav.rs
index b7159ddbd..c50be544c 100644
--- a/app/src/main/rust/src/parser/fav.rs
+++ b/app/src/main/rust/src/parser/fav.rs
@@ -1,6 +1,6 @@
 use crate::parser::list::{GalleryListResult, parse_info_list};
+use crate::parser::utils::unescape;
 use anyhow::{Result, bail};
-use quick_xml::escape::unescape;
 use serde::Serialize;
 use tl::Parser;
 use tl::VDom;
diff --git a/app/src/main/rust/src/parser/list.rs b/app/src/main/rust/src/parser/list.rs
index fe7d4d952..c0d89d1cf 100644
--- a/app/src/main/rust/src/parser/list.rs
+++ b/app/src/main/rust/src/parser/list.rs
@@ -1,10 +1,10 @@
 use crate::EhError;
+use crate::parser::utils::unescape;
 use crate::{EHGT_PREFIX, EX_PREFIX};
 use crate::{get_element_by_id, get_vdom_first_element_by_class_name};
 use crate::{get_first_element_by_class_name, query_childs_first_match_attr};
 use crate::{get_node_attr, get_node_handle_attr, regex};
 use anyhow::{Context, Result, bail};
-use quick_xml::escape::unescape;
 use serde::Serialize;
 use std::borrow::Cow;
 use std::ops::Index;
diff --git a/app/src/main/rust/src/parser/profile.rs b/app/src/main/rust/src/parser/profile.rs
index 21d0416a4..7dfa1930f 100644
--- a/app/src/main/rust/src/parser/profile.rs
+++ b/app/src/main/rust/src/parser/profile.rs
@@ -1,6 +1,6 @@
+use crate::parser::utils::unescape;
 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};
 
diff --git a/app/src/main/rust/src/parser/torrent.rs b/app/src/main/rust/src/parser/torrent.rs
index 2b3cd9164..e14edc8a9 100644
--- a/app/src/main/rust/src/parser/torrent.rs
+++ b/app/src/main/rust/src/parser/torrent.rs
@@ -1,6 +1,6 @@
+use crate::parser::utils::unescape;
 use crate::regex;
 use anyhow::{Context, Result};
-use quick_xml::escape::unescape;
 use serde::Serialize;
 use tl::{Parser, VDom};
 
diff --git a/app/src/main/rust/src/parser/utils.rs b/app/src/main/rust/src/parser/utils.rs
index ed2c938f5..2a91665ac 100644
--- a/app/src/main/rust/src/parser/utils.rs
+++ b/app/src/main/rust/src/parser/utils.rs
@@ -1,3 +1,19 @@
+use quick_xml::escape::{EscapeError, resolve_xml_entity, unescape_with};
+use std::borrow::Cow;
+
 pub fn parse_int_lenient(s: &str, default: i32) -> i32 {
     s.replace(",", "").parse().unwrap_or(default)
 }
+
+pub fn unescape(raw: &str) -> Result<Cow<'_, str>, EscapeError> {
+    unescape_with(raw, resolve_predefined_entity)
+}
+
+#[inline]
+const fn resolve_predefined_entity(entity: &str) -> Option<&'static str> {
+    match entity.as_bytes() {
+        b"nbsp" => Some(" "),
+        b"times" => Some("×"),
+        _ => resolve_xml_entity(entity),
+    }
+}
diff --git a/core/common/src/commonMain/kotlin/com/ehviewer/core/util/StringUtils.kt b/core/common/src/commonMain/kotlin/com/ehviewer/core/util/StringUtils.kt
index d62601140..3ecfda77c 100644
--- a/core/common/src/commonMain/kotlin/com/ehviewer/core/util/StringUtils.kt
+++ b/core/common/src/commonMain/kotlin/com/ehviewer/core/util/StringUtils.kt
@@ -1,12 +1,14 @@
 package com.ehviewer.core.util
 
-private val EntityRegex = "&(?:amp|lt|gt|quot|#039);".toRegex()
+private val EntityRegex = "&(?:amp|lt|gt|quot|#039|nbsp|times);".toRegex()
 private val EntityMap = mapOf(
     "&amp;" to "&",
     "&lt;" to "<",
     "&gt;" to ">",
     "&quot;" to "\"",
     "&#039;" to "'", // HTML uses &#039; instead of &apos;
+    "&nbsp;" to " ",
+    "&times;" to "×",
 )
 
 fun String.unescapeXml() = replace(EntityRegex) { EntityMap[it.value]!! }