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

Merge pull request #101 from rhysd/ignore-preload

Browse source 
Improve handling preload links and white spaces in attribute values
This commit is contained in:
Sunshine 2020-01-13 17:51:25 -05:00 committed by GitHub
commit c1dc798ded
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
1 changed files with 109 additions and 81 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

92
src/html.rs
View file

@ -82,8 +82,6 @@ pub fn walk_and_embed_assets(
match name.local.as_ref() {
"link" => {
let mut link_type: &str = "";
// Remove integrity attributes
let mut i = 0;
while i < attrs_mut.len() {
@ -95,26 +93,44 @@ pub fn walk_and_embed_assets(
}
}
for attr in attrs_mut.iter_mut() {
if &attr.name.local == "rel" {
if is_icon(attr.value.as_ref()) {
link_type = "icon";
break;
} else if attr.value.as_ref() == "stylesheet" {
link_type = "stylesheet";
break;
}
}
enum LinkType {
Icon,
Stylesheet,
Preload,
DnsPrefetch,
Unknown,
}
if link_type == "icon" {
let mut link_type = LinkType::Unknown;
for attr in attrs_mut.iter_mut() {
if &attr.name.local == "rel" {
let value = attr.value.trim();
if is_icon(value) {
link_type = LinkType::Icon;
break;
} else if value.eq_ignore_ascii_case("stylesheet") {
link_type = LinkType::Stylesheet;
break;
} else if value.eq_ignore_ascii_case("preload") {
link_type = LinkType::Preload;
break;
} else if value.eq_ignore_ascii_case("dns-prefetch") {
link_type = LinkType::DnsPrefetch;
break;
}
}
}
let link_type = link_type;
match link_type {
LinkType::Icon => {
for attr in attrs_mut.iter_mut() {
if &attr.name.local == "href" {
if opt_no_images {
attr.value.clear();
} else {
let href_full_url =
resolve_url(&url, attr.value.as_ref()).unwrap_or_default();
let href_full_url = resolve_url(&url, attr.value.as_ref())
.unwrap_or_default();
let (favicon_dataurl, _) = retrieve_asset(
cache,
client,
@ -129,14 +145,15 @@ pub fn walk_and_embed_assets(
}
}
}
} else if link_type == "stylesheet" {
}
LinkType::Stylesheet => {
for attr in attrs_mut.iter_mut() {
if &attr.name.local == "href" {
if opt_no_css {
attr.value.clear();
} else {
let href_full_url =
resolve_url(&url, &attr.value.as_ref()).unwrap_or_default();
let href_full_url = resolve_url(&url, &attr.value.as_ref())
.unwrap_or_default();
let replacement_text = match retrieve_asset(
cache,
client,
@ -170,7 +187,16 @@ pub fn walk_and_embed_assets(
}
}
}
} else {
}
LinkType::Preload | LinkType::DnsPrefetch => {
// Since all resources are embedded as data URL, preloading and prefetching are unnecessary
if let Some(attr) =
attrs_mut.iter_mut().find(|a| &a.name.local == "href")
{
attr.value.clear();
}
}
LinkType::Unknown => {
for attr in attrs_mut.iter_mut() {
if &attr.name.local == "href" {
let href_full_url =
@ -181,6 +207,7 @@ pub fn walk_and_embed_assets(
}
}
}
}
"img" => {
// Find source tags
let mut found_src: Option<Attribute> = None;
@ -203,10 +230,10 @@ pub fn walk_and_embed_assets(
name: QualName::new(None, ns!(), local_name!("src")),
value: Tendril::from_slice(TRANSPARENT_PIXEL),
});
} else if let Some((dataurl, _)) = (&found_datasrc)
.into_iter()
} else if let Some((dataurl, _)) = found_datasrc
.iter()
.chain(&found_src) // Give dataurl priority
.map(|attr| &attr.value)
.map(|attr| attr.value.trim())
.filter(|src| !src.is_empty()) // Ignore empty srcs
.next()
.and_then(|src| resolve_url(&url, src).ok()) // Make absolute
@ -232,7 +259,7 @@ pub fn walk_and_embed_assets(
let attr_name: &str = &attr.name.local;
if attr_name == "src" {
let src_full_url = resolve_url(&url, attr.value.as_ref())
let src_full_url = resolve_url(&url, attr.value.trim())
.unwrap_or_else(|_| attr.value.to_string());
attr.value.clear();
attr.value.push_slice(src_full_url.as_str());
@ -243,7 +270,7 @@ pub fn walk_and_embed_assets(
attr.value.push_slice(TRANSPARENT_PIXEL);
} else {
let srcset_full_url =
resolve_url(&url, attr.value.as_ref()).unwrap_or_default();
resolve_url(&url, attr.value.trim()).unwrap_or_default();
let (source_dataurl, _) = retrieve_asset(
cache,
client,
@ -263,13 +290,13 @@ pub fn walk_and_embed_assets(
"a" => {
for attr in attrs_mut.iter_mut() {
if &attr.name.local == "href" {
let attr_value = attr.value.trim();
// Don't touch email links or hrefs which begin with a hash sign
if attr.value.starts_with('#') || url_has_protocol(&attr.value) {
if attr_value.starts_with('#') || url_has_protocol(attr_value) {
continue;
}
let href_full_url =
resolve_url(&url, attr.value.as_ref()).unwrap_or_default();
let href_full_url = resolve_url(&url, attr_value).unwrap_or_default();
attr.value.clear();
attr.value.push_slice(href_full_url.as_str());
}
@ -299,7 +326,7 @@ pub fn walk_and_embed_assets(
for attr in attrs_mut.iter_mut() {
if &attr.name.local == "src" {
let src_full_url =
resolve_url(&url, attr.value.as_ref()).unwrap_or_default();
resolve_url(&url, attr.value.trim()).unwrap_or_default();
let (js_dataurl, _) = retrieve_asset(
cache,
client,
@ -341,10 +368,11 @@ pub fn walk_and_embed_assets(
"form" => {
for attr in attrs_mut.iter_mut() {
if &attr.name.local == "action" {
let attr_value = attr.value.trim();
// Modify action to be a full URL
if !is_valid_url(&attr.value) {
if !is_valid_url(attr_value) {
let href_full_url =
resolve_url(&url, attr.value.as_ref()).unwrap_or_default();
resolve_url(&url, attr_value).unwrap_or_default();
attr.value.clear();
attr.value.push_slice(href_full_url.as_str());
}
@ -360,7 +388,7 @@ pub fn walk_and_embed_assets(
continue;
}
let iframe_src = attr.value.as_ref();
let iframe_src = attr.value.trim();
// Ignore iframes with empty source (they cause infinite loops)
if iframe_src.is_empty() {
@ -400,7 +428,7 @@ pub fn walk_and_embed_assets(
"video" => {
for attr in attrs_mut.iter_mut() {
if &attr.name.local == "poster" {
let video_poster = attr.value.as_ref();
let video_poster = attr.value.trim();
// Skip posters with empty source
if video_poster.is_empty() {