diff --git a/crates/uv-client/src/html.rs b/crates/uv-client/src/html.rs
index 738a9521f..397cc0142 100644
--- a/crates/uv-client/src/html.rs
+++ b/crates/uv-client/src/html.rs
@@ -108,7 +108,8 @@ impl SimpleHtml {
.ok_or(Error::MissingHref)?;
let href = std::str::from_utf8(href.as_bytes())?;
- let (path, hashes) = if let Some((path, fragment)) = href.split_once('#') {
+ let decoded = html_escape::decode_html_entities(href);
+ let (path, hashes) = if let Some((path, fragment)) = decoded.split_once('#') {
// Extract the hash, which should be in the fragment.
(path, Self::parse_hash(fragment)?)
} else {
@@ -316,6 +317,59 @@ mod tests {
"###);
}
+ #[test]
+ fn parse_escaped_fragment() {
+ let text = r#"
+
+
+
+ Links for jinja2
+ Jinja2-3.1.2+233fca715f49-py3-none-any.whl
+
+
+
+ "#;
+ let base = Url::parse("https://download.pytorch.org/whl/jinja2/").unwrap();
+ let result = SimpleHtml::parse(text, &base).unwrap();
+ insta::assert_debug_snapshot!(result, @r###"
+ SimpleHtml {
+ base: BaseUrl(
+ Url {
+ scheme: "https",
+ cannot_be_a_base: false,
+ username: "",
+ password: None,
+ host: Some(
+ Domain(
+ "download.pytorch.org",
+ ),
+ ),
+ port: None,
+ path: "/whl/jinja2/",
+ query: None,
+ fragment: None,
+ },
+ ),
+ files: [
+ File {
+ dist_info_metadata: None,
+ filename: "Jinja2-3.1.2+233fca715f49-py3-none-any.whl",
+ hashes: Hashes {
+ sha256: Some(
+ "6088930bfe239f0e6710546ab9c19c9ef35e29792895fed6e6e31a023a182a61",
+ ),
+ },
+ requires_python: None,
+ size: None,
+ upload_time: None,
+ url: "/whl/Jinja2-3.1.2+233fca715f49-py3-none-any.whl#sha256=6088930bfe239f0e6710546ab9c19c9ef35e29792895fed6e6e31a023a182a61",
+ yanked: None,
+ },
+ ],
+ }
+ "###);
+ }
+
#[test]
fn parse_missing_hash() {
let text = r#"