Mirror/monolith
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

use proper charset detection for linked assets

Browse source
This commit is contained in:
Sunshine 2021-06-08 12:25:19 -10:00
parent 125aeeec3b
commit 5effa38392
No known key found for this signature in database
GPG key ID: B80CA68703CD8AB1
2 changed files with 21 additions and 25 deletions

34
src/html.rs
View file

@ -606,7 +606,7 @@ pub fn retrieve_and_embed_asset(
options,
depth + 1,
) {
Ok((data, final_url, mut media_type, _charset)) => {
Ok((data, final_url, mut media_type, charset)) => {
let node_name: &str = get_node_name(&node).unwrap();
// Check integrity if it's a LINK or SCRIPT element
@ -624,23 +624,25 @@ pub fn retrieve_and_embed_asset(
}
if ok_to_include {
let s: String;
if let Some(encoding) = Encoding::for_label(charset.as_bytes()) {
let (string, _, _) = encoding.decode(&data);
s = string.to_string();
} else {
s = String::from_utf8_lossy(&data).to_string();
}
if node_name == "link" && determine_link_node_type(node) == "stylesheet" {
// Stylesheet LINK elements require special treatment
let css: String = embed_css(
cache,
client,
&final_url,
&String::from_utf8_lossy(&data),
options,
depth + 1,
);
let css: String = embed_css(cache, client, &final_url, &s, options, depth + 1);
// Create and embed data URL
let css_data_url = create_data_url("text/css", css.as_bytes(), &final_url);
// TODO: use charset
let css_data_url = create_data_url(&media_type, css.as_bytes(), &final_url);
set_node_attr(&node, attr_name, Some(css_data_url.to_string()));
} else if node_name == "frame" || node_name == "iframe" {
// (I)FRAMEs are also quite different from conventional resources
let frame_dom = html_to_dom(&data, "utf-8".to_string());
let frame_dom = html_to_dom(&data, charset);
walk_and_embed_assets(
cache,
client,
@ -679,6 +681,7 @@ pub fn retrieve_and_embed_asset(
}
// Create and embed data URL
// TODO: use charset
let mut data_url = create_data_url(&media_type, &data, &final_url);
data_url.set_fragment(resolved_url.fragment());
set_node_attr(node, attr_name, Some(data_url.to_string()));
@ -725,14 +728,7 @@ pub fn walk_and_embed_assets(
|| meta_attr_http_equiv_value.eq_ignore_ascii_case("location")
{
// Remove http-equiv attributes from META nodes if they're able to control the page
set_node_attr(
&node,
"http-equiv",
Some(format!(
"disabled by monolith ({})",
meta_attr_http_equiv_value
)),
);
set_node_attr(&node, "http-equiv", None);
}
}
}

12
src/tests/html/walk_and_embed_assets.rs
View file

@ -290,7 +290,7 @@ mod passing {
}
#[test]
fn keeps_integrity_for_linked_assets() {
fn keeps_integrity_for_unfamiliar_links() {
let html = "<title>Has integrity</title>\
<link integrity=\"sha384-12345\" rel=\"something\" href=\"https://some-site.com/some-file.ext\" />";
let dom = html::html_to_dom(&html.as_bytes().to_vec(), str!());
@ -322,7 +322,7 @@ mod passing {
}
#[test]
fn discards_integrity_for_linked_assets_nojs_nocss() {
fn discards_integrity_for_known_links_nojs_nocss() {
let html = "\
<title>No integrity</title>\
<link integrity=\"\" rel=\"stylesheet\" href=\"data:;\"/>\
@ -403,8 +403,8 @@ mod passing {
let html = "\
<html>\
<head>\
<meta http-equiv=\"Refresh\" value=\"20\"/>\
<meta http-equiv=\"Location\" value=\"https://freebsd.org\"/>\
<meta http-equiv=\"Refresh\" content=\"2\"/>\
<meta http-equiv=\"Location\" content=\"https://freebsd.org\"/>\
</head>\
<body>\
</body>\
@ -433,8 +433,8 @@ mod passing {
"\
<html>\
<head>\
<meta http-equiv=\"disabled by monolith (Refresh)\" value=\"20\">\
<meta http-equiv=\"disabled by monolith (Location)\" value=\"https://freebsd.org\">\
<meta content=\"2\">\
<meta content=\"https://freebsd.org\">\
</head>\
<body>\
</body>\