Fix crash when page URL points to huge file

This commit is contained in:
Daniel García 2019-12-19 00:37:16 +01:00
parent f09996a21d
commit 2545469713
No known key found for this signature in database
GPG key ID: FC8A7D14C3CD543A
2 changed files with 33 additions and 2 deletions

View file

@ -213,7 +213,7 @@ fn get_icon_url(domain: &str) -> Result<(Vec<Icon>, String), Error> {
let mut cookie_str = String::new();
let resp = get_page(&ssldomain).or_else(|_| get_page(&httpdomain));
if let Ok(content) = resp {
if let Ok(mut content) = resp {
// Extract the URL from the respose in case redirects occured (like @ gitlab.com)
let url = content.url().clone();
@ -233,7 +233,11 @@ fn get_icon_url(domain: &str) -> Result<(Vec<Icon>, String), Error> {
// Add the default favicon.ico to the list with the domain the content responded from.
iconlist.push(Icon::new(35, url.join("/favicon.ico").unwrap().into_string()));
let soup = Soup::from_reader(content)?;
// 512KB should be more than enough for the HTML, though as we only really need
// the HTML header, it could potentially be reduced even further
let limited_reader = crate::util::LimitedReader::new(&mut content, 512 * 1024);
let soup = Soup::from_reader(limited_reader)?;
// Search for and filter
let favicons = soup
.tag("link")

View file

@ -216,6 +216,33 @@ pub fn delete_file(path: &str) -> IOResult<()> {
res
}
pub struct LimitedReader<'a> {
reader: &'a mut dyn std::io::Read,
limit: usize, // In bytes
count: usize,
}
impl<'a> LimitedReader<'a> {
pub fn new(reader: &'a mut dyn std::io::Read, limit: usize) -> LimitedReader<'a> {
LimitedReader {
reader,
limit,
count: 0,
}
}
}
impl<'a> std::io::Read for LimitedReader<'a> {
fn read(&mut self, buf: &mut [u8]) -> std::io::Result<usize> {
self.count += buf.len();
if self.count > self.limit {
Ok(0) // End of the read
} else {
self.reader.read(buf)
}
}
}
const UNITS: [&str; 6] = ["bytes", "KB", "MB", "GB", "TB", "PB"];
pub fn get_display_size(size: i32) -> String {