Archives
/
monolith
mirror of https://github.com/Y2Z/monolith
2
0
Fork 0
Code Issues Projects Releases Wiki Activity

refactor code that processes the DOM

Browse Source
pull/226/head
Sunshine 4 years ago
parent 15d98a7269
commit d89b4d5f5b
No known key found for this signature in database
GPG Key ID: B80CA68703CD8AB1
4 changed files with 541 additions and 623 deletions
Show Stats Download Patch File Download Diff File

647
src/html.rs
Unescape Escape View File

@ -4,7 +4,7 @@ use html5ever::interface::QualName;
use html5ever::parse_document;
use html5ever::rcdom::{Handle, NodeData, RcDom};
use html5ever::serialize::{serialize, SerializeOpts};
use html5ever::tendril::{format_tendril, Tendril, TendrilSink};
use html5ever::tendril::{format_tendril, TendrilSink};
use html5ever::tree_builder::{Attribute, TreeSink};
use html5ever::{local_name, namespace_url, ns, LocalName};
use reqwest::blocking::Client;
@ -493,39 +493,33 @@ pub fn walk_and_embed_assets(
ref attrs,
..
} => {
let attrs_mut = &mut attrs.borrow_mut();
match name.local.as_ref() {
"meta" => {
// Determine type
let mut is_unwanted_meta: bool = false;
for attr in attrs_mut.iter_mut() {
let attr_name: &str = &attr.name.local;
if attr_name.eq_ignore_ascii_case("http-equiv") {
let value: String = attr.value.to_string();
is_unwanted_meta = value.eq_ignore_ascii_case("refresh")
|| value.eq_ignore_ascii_case("location");
}
}
if is_unwanted_meta {
// Strip this node off all its attributes
while attrs_mut.len() > 0 {
attrs_mut.remove(0);
// Remove http-equiv attributes from META nodes if they're able to control the page
if let Some(meta_attr_http_equiv_value) = get_node_attr(node, "http-equiv") {
let meta_attr_http_equiv_value: &str = &meta_attr_http_equiv_value;
if meta_attr_http_equiv_value.eq_ignore_ascii_case("refresh")
|| meta_attr_http_equiv_value.eq_ignore_ascii_case("location")
{
set_node_attr(
&node,
"http-equiv",
Some(format!(
"disabled by monolith ({})",
meta_attr_http_equiv_value
)),
);
}
}
}
"link" => {
// Remove integrity attributes, keep value of the last one
let mut integrity: String = str!();
let mut i = 0;
while i < attrs_mut.len() {
let attr_name: &str = &attrs_mut[i].name.local;
if attr_name.eq_ignore_ascii_case("integrity") {
integrity = str!(attrs_mut.remove(i).value.trim());
} else {
i += 1;
}
// Read and remember integrity attribute value of this LINK node
let link_attr_integrity_value: Option<String> =
get_node_attr(node, "integrity");
// Remove integrity attribute from the LINK node
if link_attr_integrity_value != None {
set_node_attr(node, "integrity", None);
}
enum LinkType {
@ -537,43 +531,27 @@ pub fn walk_and_embed_assets(
}
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) {
if let Some(link_attr_rel_value) = get_node_attr(node, "rel") {
if is_icon(&link_attr_rel_value) {
link_type = LinkType::Icon;
break;
} else if value.eq_ignore_ascii_case("stylesheet") {
} else if link_attr_rel_value.eq_ignore_ascii_case("stylesheet") {
link_type = LinkType::Stylesheet;
break;
} else if value.eq_ignore_ascii_case("preload") {
} else if link_attr_rel_value.eq_ignore_ascii_case("preload") {
link_type = LinkType::Preload;
break;
} else if value.eq_ignore_ascii_case("dns-prefetch") {
} else if link_attr_rel_value.eq_ignore_ascii_case("dns-prefetch") {
link_type = LinkType::DnsPrefetch;
break;
}
}
}
// Shadow the variable (to make it non-mutable)
let link_type = link_type;
match link_type {
LinkType::Icon => {
// Find and remove href attribute(s), keep value of the last found one
let mut link_href: String = str!();
let mut i = 0;
while i < attrs_mut.len() {
let attr_name: &str = &attrs_mut[i].name.local;
if attr_name.eq_ignore_ascii_case("href") {
link_href = str!(attrs_mut.remove(i).value.trim());
} else {
i += 1;
}
}
if !options.no_images && !link_href.is_empty() {
// Find and resolve this LINK node's href attribute
if let Some(link_attr_href_value) = get_node_attr(node, "href") {
if !options.no_images && !link_attr_href_value.is_empty() {
let link_href_full_url =
resolve_url(&url, link_href).unwrap_or_default();
resolve_url(&url, link_attr_href_value).unwrap_or_default();
let link_href_url_fragment =
get_url_fragment(link_href_full_url.clone());
match retrieve_asset(
@ -589,10 +567,21 @@ pub fn walk_and_embed_assets(
link_href_final_url,
link_href_media_type,
)) => {
let mut ok_to_include = true;
// Check integrity
if integrity.is_empty()
|| check_integrity(&link_href_data, &integrity)
if let Some(link_attr_integrity_value) =
link_attr_integrity_value
{
if !link_attr_integrity_value.is_empty() {
ok_to_include = check_integrity(
&link_href_data,
&link_attr_integrity_value,
);
}
}
if ok_to_include {
let link_href_data_url = data_to_data_url(
&link_href_media_type,
&link_href_data,
@ -603,14 +592,7 @@ pub fn walk_and_embed_assets(
link_href_data_url.as_str(),
link_href_url_fragment.as_str(),
);
attrs_mut.push(Attribute {
name: QualName::new(
None,
ns!(),
local_name!("href"),
),
value: Tendril::from_slice(assembled_url.as_ref()),
});
set_node_attr(&node, "href", Some(assembled_url));
}
}
Err(_) => {
@ -620,35 +602,23 @@ pub fn walk_and_embed_assets(
link_href_full_url.as_str(),
link_href_url_fragment.as_str(),
);
attrs_mut.push(Attribute {
name: QualName::new(
None,
ns!(),
local_name!("href"),
),
value: Tendril::from_slice(assembled_url.as_ref()),
});
}
set_node_attr(node, "href", Some(assembled_url));
}
}
}
}
LinkType::Stylesheet => {
// Find and remove href attribute(s), keep value of the last found one
let mut link_href: String = str!();
let mut i = 0;
while i < attrs_mut.len() {
let attr_name: &str = &attrs_mut[i].name.local;
if attr_name.eq_ignore_ascii_case("href") {
link_href = str!(attrs_mut.remove(i).value.trim());
} else {
i += 1;
set_node_attr(node, "href", None);
}
}
}
LinkType::Stylesheet => {
// Find and resolve this LINK node's href attribute
if let Some(link_attr_href_value) = get_node_attr(node, "href") {
set_node_attr(node, "href", None);
if !options.no_css && !link_href.is_empty() {
if !options.no_css && !link_attr_href_value.is_empty() {
let link_href_full_url =
resolve_url(&url, link_href).unwrap_or_default();
resolve_url(&url, link_attr_href_value).unwrap_or_default();
match retrieve_asset(
cache,
client,
@ -662,10 +632,21 @@ pub fn walk_and_embed_assets(
link_href_final_url,
_link_href_media_type,
)) => {
let mut ok_to_include = true;
// Check integrity
if integrity.is_empty()
|| check_integrity(&link_href_data, &integrity)
if let Some(link_attr_integrity_value) =
link_attr_integrity_value
{
if !link_attr_integrity_value.is_empty() {
ok_to_include = check_integrity(
&link_href_data,
&link_attr_integrity_value,
);
}
}
if ok_to_include {
let css: String = embed_css(
cache,
client,
@ -680,52 +661,37 @@ pub fn walk_and_embed_assets(
&link_href_final_url,
);
// Add new data URL href attribute
attrs_mut.push(Attribute {
name: QualName::new(
None,
ns!(),
local_name!("href"),
),
value: Tendril::from_slice(
link_href_data_url.as_ref(),
),
});
set_node_attr(
&node,
"href",
Some(link_href_data_url),
);
}
}
Err(_) => {
// Keep remote reference if unable to retrieve the asset
if is_http_url(link_href_full_url.clone()) {
attrs_mut.push(Attribute {
name: QualName::new(
None,
ns!(),
local_name!("href"),
),
value: Tendril::from_slice(
link_href_full_url.as_ref(),
),
});
set_node_attr(
&node,
"href",
Some(link_href_full_url),
);
}
}
}
}
}
LinkType::Preload | LinkType::DnsPrefetch => {
// Since all resources are embedded as data URL, preloading and prefetching are unnecessary
for _ in 0..attrs_mut.len() {
attrs_mut.remove(0);
}
LinkType::Preload | LinkType::DnsPrefetch => {
// Since all resources are embedded as data URLs, preloading and prefetching are not necessary
set_node_attr(node, "rel", None);
}
LinkType::Unknown => {
// Make sure that all other LINKs' href attributes are full URLs
for attr in attrs_mut.iter_mut() {
let attr_name: &str = &attr.name.local;
if attr_name.eq_ignore_ascii_case("href") {
if let Some(link_attr_href_value) = get_node_attr(node, "href") {
let href_full_url =
resolve_url(&url, attr.value.trim()).unwrap_or_default();
attr.value.clear();
attr.value.push_slice(&href_full_url.as_str());
}
resolve_url(&url, link_attr_href_value).unwrap_or_default();
set_node_attr(node, "href", Some(href_full_url));
}
}
}
@ -733,33 +699,24 @@ pub fn walk_and_embed_assets(
"base" => {
if is_http_url(url) {
// Ensure the BASE node doesn't have a relative URL
for attr in attrs_mut.iter_mut() {
let attr_name: &str = &attr.name.local;
if attr_name.eq_ignore_ascii_case("href") {
if let Some(base_attr_href_value) = get_node_attr(node, "href") {
let href_full_url =
resolve_url(&url, attr.value.trim()).unwrap_or_default();
attr.value.clear();
attr.value.push_slice(&href_full_url.as_str());
}
resolve_url(&url, base_attr_href_value).unwrap_or_default();
set_node_attr(node, "href", Some(href_full_url));
}
}
}
"body" => {
// Find and remove background attribute(s), keep value of the last found one
let mut background: String = str!();
let mut i = 0;
while i < attrs_mut.len() {
let attr_name: &str = &attrs_mut[i].name.local;
if attr_name.eq_ignore_ascii_case("background") {
background = str!(attrs_mut.remove(i).value.trim());
} else {
i += 1;
}
}
if !options.no_images && !background.is_empty() {
let background_full_url = resolve_url(&url, background).unwrap_or_default();
let background_url_fragment = get_url_fragment(background_full_url.clone());
// Read and remember background attribute value of this BODY node
if let Some(body_attr_background_value) = get_node_attr(node, "background") {
// Remove background BODY node attribute by default
set_node_attr(node, "background", None);
if !options.no_images && !body_attr_background_value.is_empty() {
let background_full_url =
resolve_url(&url, body_attr_background_value).unwrap_or_default();
let background_url_fragment =
get_url_fragment(background_full_url.clone());
match retrieve_asset(
cache,
client,
@ -768,21 +725,22 @@ pub fn walk_and_embed_assets(
options,
depth + 1,
) {
Ok((background_data, background_final_url, background_media_type)) => {
Ok((
background_data,
background_final_url,
background_media_type,
)) => {
let background_data_url = data_to_data_url(
&background_media_type,
&background_data,
&background_final_url,
);
// Add new data URL background attribute
// Convert background attribute to data URL
let assembled_url: String = url_with_fragment(
background_data_url.as_str(),
background_url_fragment.as_str(),
);
attrs_mut.push(Attribute {
name: QualName::new(None, ns!(), local_name!("background")),
value: Tendril::from_slice(assembled_url.as_ref()),
});
set_node_attr(node, "background", Some(assembled_url));
}
Err(_) => {
// Keep remote reference if unable to retrieve the asset
@ -791,10 +749,8 @@ pub fn walk_and_embed_assets(
background_full_url.as_str(),
background_url_fragment.as_str(),
);
attrs_mut.push(Attribute {
name: QualName::new(None, ns!(), local_name!("background")),
value: Tendril::from_slice(assembled_url.as_ref()),
});
set_node_attr(node, "background", Some(assembled_url));
}
}
}
}
@ -802,48 +758,43 @@ pub fn walk_and_embed_assets(
}
"img" => {
// Find source attribute(s)
let mut img_data_src: String = str!();
let mut img_src: String = str!();
let mut img_srcset: String = str!();
let mut i = 0;
while i < attrs_mut.len() {
let attr_name: &str = &attrs_mut[i].name.local;
if attr_name.eq_ignore_ascii_case("data-src") {
img_data_src = str!(attrs_mut.remove(i).value.trim());
} else if attr_name.eq_ignore_ascii_case("src") {
img_src = str!(attrs_mut.remove(i).value.trim());
} else if attr_name.eq_ignore_ascii_case("srcset") {
img_srcset = str!(attrs_mut.remove(i).value.trim());
} else {
i += 1;
}
}
let img_attr_src_value: Option<String> = get_node_attr(node, "src");
let img_attr_data_src_value: Option<String> = get_node_attr(node, "data-src");
if options.no_images {
// Add empty image src attribute
attrs_mut.push(Attribute {
name: QualName::new(None, ns!(), local_name!("src")),
value: Tendril::from_slice(empty_image!()),
});
// Put empty images into src and data-src attributes
if img_attr_src_value != None {
set_node_attr(node, "src", Some(str!(empty_image!())));
}
if img_attr_data_src_value != None {
set_node_attr(node, "data-src", Some(str!(empty_image!())));
}
} else {
if img_src.is_empty() && img_data_src.is_empty() {
if img_attr_src_value.clone().unwrap_or_default().is_empty()
&& img_attr_data_src_value
.clone()
.unwrap_or_default()
.is_empty()
{
// Add empty src attribute
attrs_mut.push(Attribute {
name: QualName::new(None, ns!(), local_name!("src")),
value: Tendril::from_slice(""),
});
set_node_attr(node, "src", Some(str!()));
} else {
// Add data URL src attribute
let img_full_url = resolve_url(
&url,
if !img_data_src.is_empty() {
img_data_src
if !img_attr_data_src_value
.clone()
.unwrap_or_default()
.is_empty()
{
img_attr_data_src_value.unwrap_or_default()
} else {
img_src
img_attr_src_value.unwrap_or_default()
},
)
.unwrap_or_default();
let img_url_fragment = get_url_fragment(img_full_url.clone());
match retrieve_asset(
cache,
client,
@ -862,36 +813,32 @@ pub fn walk_and_embed_assets(
img_data_url.as_str(),
img_url_fragment.as_str(),
);
attrs_mut.push(Attribute {
name: QualName::new(None, ns!(), local_name!("src")),
value: Tendril::from_slice(assembled_url.as_ref()),
});
set_node_attr(node, "src", Some(assembled_url));
}
Err(_) => {
// Keep remote reference if unable to retrieve the asset
if is_http_url(img_full_url.clone()) {
// Keep remote reference if unable to retrieve the asset
let assembled_url: String = url_with_fragment(
img_full_url.as_str(),
img_url_fragment.as_str(),
);
attrs_mut.push(Attribute {
name: QualName::new(None, ns!(), local_name!("src")),
value: Tendril::from_slice(assembled_url.as_ref()),
});
set_node_attr(node, "src", Some(assembled_url));
} else {
// Don't keep original reference if it's not a remote target
set_node_attr(node, "src", None);
}
}
}
}
}
// Resolve srcset attribute
if let Some(img_srcset) = get_node_attr(node, "srcset") {
if !img_srcset.is_empty() {
attrs_mut.push(Attribute {
name: QualName::new(None, ns!(), local_name!("srcset")),
value: Tendril::from_slice(
embed_srcset(cache, client, &url, &img_srcset, options, depth)
.as_ref(),
),
});
let resolved_srcset: String =
embed_srcset(cache, client, &url, &img_srcset, options, depth);
set_node_attr(node, "srcset", Some(resolved_srcset));
}
}
}
"svg" => {
@ -900,31 +847,19 @@ pub fn walk_and_embed_assets(
}
}
"input" => {
if let Some(attr_value) = get_node_attr(node, "type") {
if attr_value.to_string().eq_ignore_ascii_case("image") {
let mut input_image_src: String = str!();
let mut i = 0;
while i < attrs_mut.len() {
let attr_name: &str = &attrs_mut[i].name.local;
if attr_name.eq_ignore_ascii_case("src") {
input_image_src = str!(attrs_mut.remove(i).value.trim());
if let Some(input_attr_type_value) = get_node_attr(node, "type") {
if input_attr_type_value.eq_ignore_ascii_case("image") {
if let Some(input_attr_src_value) = get_node_attr(node, "src") {
if options.no_images || input_attr_src_value.is_empty() {
let value = if input_attr_src_value.is_empty() {
str!()
} else {
i += 1;
}
}
if options.no_images || input_image_src.is_empty() {
attrs_mut.push(Attribute {
name: QualName::new(None, ns!(), local_name!("src")),
value: Tendril::from_slice(if input_image_src.is_empty() {
""
} else {
empty_image!()
}),
});
str!(empty_image!())
};
set_node_attr(node, "src", Some(value));
} else {
let input_image_full_url =
resolve_url(&url, input_image_src).unwrap_or_default();
resolve_url(&url, input_attr_src_value).unwrap_or_default();
let input_image_url_fragment =
get_url_fragment(input_image_full_url.clone());
match retrieve_asset(
@ -950,10 +885,7 @@ pub fn walk_and_embed_assets(
input_image_data_url.as_str(),
input_image_url_fragment.as_str(),
);
attrs_mut.push(Attribute {
name: QualName::new(None, ns!(), local_name!("src")),
value: Tendril::from_slice(assembled_url.as_ref()),
});
set_node_attr(node, "src", Some(assembled_url));
}
Err(_) => {
// Keep remote reference if unable to retrieve the asset
@ -962,14 +894,8 @@ pub fn walk_and_embed_assets(
input_image_full_url.as_str(),
input_image_url_fragment.as_str(),
);
attrs_mut.push(Attribute {
name: QualName::new(
None,
ns!(),
local_name!("src"),
),
value: Tendril::from_slice(assembled_url.as_ref()),
});
set_node_attr(node, "src", Some(assembled_url));
}
}
}
}
@ -978,17 +904,18 @@ pub fn walk_and_embed_assets(
}
}
"image" => {
// Find and remove (xlink:)href attribute(s), keep value of the last one
let mut image_href: String = str!();
let mut i = 0;
while i < attrs_mut.len() {
let attr_name: &str = &attrs_mut[i].name.local;
if attr_name.eq_ignore_ascii_case("xlink:href")
|| attr_name.eq_ignore_ascii_case("href")
{
image_href = str!(attrs_mut.remove(i).value.trim());
} else {
i += 1;
if let Some(image_attr_href_value) = get_node_attr(node, "href") {
image_href = image_attr_href_value;
if options.no_images {
set_node_attr(node, "href", None);
}
}
if let Some(image_attr_xlink_href_value) = get_node_attr(node, "xlink:href") {
image_href = image_attr_xlink_href_value;
if options.no_images {
set_node_attr(node, "xlink:href", None);
}
}
@ -1014,10 +941,7 @@ pub fn walk_and_embed_assets(
image_data_url.as_str(),
image_url_fragment.as_str(),
);
attrs_mut.push(Attribute {
name: QualName::new(None, ns!(), local_name!("href")),
value: Tendril::from_slice(assembled_url.as_ref()),
});
set_node_attr(node, "href", Some(assembled_url));
}
Err(_) => {
// Keep remote reference if unable to retrieve the asset
@ -1026,34 +950,27 @@ pub fn walk_and_embed_assets(
image_full_url.as_str(),
image_url_fragment.as_str(),
);
attrs_mut.push(Attribute {
name: QualName::new(None, ns!(), local_name!("href")),
value: Tendril::from_slice(assembled_url.as_ref()),
});
set_node_attr(node, "href", Some(assembled_url));
}
}
}
}
}
"source" => {
for attr in attrs_mut.iter_mut() {
let attr_name: &str = &attr.name.local;
if attr_name.eq_ignore_ascii_case("src") {
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());
} else if attr_name.eq_ignore_ascii_case("srcset") {
if let Some(source_attr_src_value) = get_node_attr(node, "src") {
let src_full_url: String = resolve_url(&url, source_attr_src_value.clone())
.unwrap_or_else(|_| source_attr_src_value.to_string());
set_node_attr(node, "src", Some(src_full_url));
}
if let Some(source_attr_srcset_value) = get_node_attr(node, "srcset") {
if get_node_name(&get_parent_node(&node)) == Some("picture") {
if options.no_images {
attr.value.clear();
attr.value.push_slice(empty_image!());
set_node_attr(node, "srcset", Some(str!(empty_image!())));
} else {
let srcset_full_url =
resolve_url(&url, attr.value.trim()).unwrap_or_default();
let srcset_url_fragment =
get_url_fragment(srcset_full_url.clone());
resolve_url(&url, source_attr_srcset_value).unwrap_or_default();
let srcset_url_fragment = get_url_fragment(srcset_full_url.clone());
match retrieve_asset(
cache,
client,
@ -1068,23 +985,20 @@ pub fn walk_and_embed_assets(
&srcset_data,
&srcset_final_url,
);
attr.value.clear();
let assembled_url: String = url_with_fragment(
srcset_data_url.as_str(),
srcset_url_fragment.as_str(),
);
attr.value.push_slice(assembled_url.as_str());
set_node_attr(node, "srcset", Some(assembled_url));
}
Err(_) => {
// Keep remote reference if unable to retrieve the asset
if is_http_url(srcset_full_url.clone()) {
attr.value.clear();
let assembled_url: String = url_with_fragment(
srcset_full_url.as_str(),
srcset_url_fragment.as_str(),
);
attr.value.push_slice(assembled_url.as_str());
}
set_node_attr(node, "srcset", Some(assembled_url));
}
}
}
@ -1093,50 +1007,47 @@ pub fn walk_and_embed_assets(
}
}
"a" | "area" => {
for attr in attrs_mut.iter_mut() {
let attr_name: &str = &attr.name.local;
if attr_name.eq_ignore_ascii_case("href") {
let attr_value = attr.value.trim();
if options.no_js && attr_value.trim().starts_with("javascript:") {
attr.value.clear();
if let Some(anchor_attr_href_value) = get_node_attr(node, "href") {
if options.no_js
&& anchor_attr_href_value
.clone()
.trim()
.starts_with("javascript:")
{
// Replace with empty JS call to preserve original behavior
attr.value.push_slice("javascript:;");
continue;
}
set_node_attr(node, "href", Some(str!("javascript:;")));
} else if anchor_attr_href_value.clone().starts_with('#')
|| url_has_protocol(anchor_attr_href_value.clone())
{
// Don't touch email links or hrefs which begin with a hash
if attr_value.starts_with('#') || url_has_protocol(attr_value) {
continue;
}
let href_full_url = resolve_url(&url, attr_value).unwrap_or_default();
attr.value.clear();
attr.value.push_slice(href_full_url.as_str());
} else {
let href_full_url =
resolve_url(&url, anchor_attr_href_value).unwrap_or_default();
set_node_attr(node, "href", Some(href_full_url));
}
}
}
"script" => {
// Remove integrity and src attributes, keep values of the last ones
let mut script_integrity: String = str!();
let mut script_src: String = str!();
let mut i = 0;
while i < attrs_mut.len() {
let attr_name: &str = &attrs_mut[i].name.local;
if attr_name.eq_ignore_ascii_case("integrity") {
script_integrity = str!(attrs_mut.remove(i).value.trim());
} else if attr_name.eq_ignore_ascii_case("src") {
script_src = str!(attrs_mut.remove(i).value.trim());
} else {
i += 1;
}
// Read values of integrity and src attributes
let script_attr_integrity: Option<String> = get_node_attr(node, "integrity");
let script_attr_src: Option<String> = get_node_attr(node, "src");
// Wipe integrity attribute
if script_attr_integrity != None {
set_node_attr(node, "integrity", None);
}
if options.no_js {
// Empty inner content (src is already gone)
// Empty inner content
node.children.borrow_mut().clear();
} else if !script_src.is_empty() {
let script_full_url = resolve_url(&url, script_src).unwrap_or_default();
// Remove src attribute
if script_attr_src != None {
set_node_attr(node, "src", None);
}
} else if !script_attr_src.clone().unwrap_or_default().is_empty() {
let script_full_url =
resolve_url(&url, script_attr_src.unwrap_or_default())
.unwrap_or_default();
match retrieve_asset(
cache,
client,
@ -1146,29 +1057,37 @@ pub fn walk_and_embed_assets(
depth + 1,
) {
Ok((script_data, script_final_url, _script_media_type)) => {
let mut ok_to_include = true;
// Check integrity
if let Some(script_attr_integrity_value) = script_attr_integrity {
if !script_attr_integrity_value.is_empty() {
ok_to_include = check_integrity(
&script_data,
&script_attr_integrity_value,
);
}
}
if ok_to_include {
// Only embed if we're able to validate integrity
if script_integrity.is_empty()
|| check_integrity(&script_data, &script_integrity)
{
let script_data_url = data_to_data_url(
"application/javascript",
&script_data,
&script_final_url,
);
// Add new data URL src attribute
attrs_mut.push(Attribute {
name: QualName::new(None, ns!(), local_name!("src")),
value: Tendril::from_slice(script_data_url.as_ref()),
});
set_node_attr(node, "src", Some(script_data_url));
} else {
set_node_attr(node, "src", None);
}
}
Err(_) => {
// Keep remote reference if unable to retrieve the asset
if is_http_url(script_full_url.clone()) {
attrs_mut.push(Attribute {
name: QualName::new(None, ns!(), local_name!("src")),
value: Tendril::from_slice(script_full_url.as_ref()),
});
// Keep remote reference if unable to retrieve the asset
set_node_attr(node, "src", Some(script_full_url));
} else {
// Remove src attribute if target is not remote
set_node_attr(node, "src", None);
}
}
};
@ -1197,38 +1116,27 @@ pub fn walk_and_embed_assets(
}
}
"form" => {
for attr in attrs_mut.iter_mut() {
let attr_name: &str = &attr.name.local;
if attr_name.eq_ignore_ascii_case("action") {
let form_action = attr.value.trim();
if let Some(form_attr_action_value) = get_node_attr(node, "action") {
// Modify action property to ensure it's a full URL
if !is_http_url(form_action) {
if !is_http_url(form_attr_action_value.clone()) {
let form_action_full_url =
resolve_url(&url, form_action).unwrap_or_default();
attr.value.clear();
attr.value.push_slice(form_action_full_url.as_str());
}
resolve_url(&url, form_attr_action_value).unwrap_or_default();
set_node_attr(node, "action", Some(form_action_full_url));
}
}
}
"frame" | "iframe" => {
for attr in attrs_mut.iter_mut() {
let attr_name: &str = &attr.name.local;
if attr_name.eq_ignore_ascii_case("src") {
if let Some(frame_attr_src_value) = get_node_attr(node, "src") {
if options.no_frames {
// Empty the src attribute
attr.value.clear();
continue;
}
let frame_src = attr.value.trim();
// Ignore (i)frames with empty source — they cause infinite loops
if frame_src.is_empty() {
continue;
}
set_node_attr(node, "src", Some(str!()));
} else {
let frame_src = frame_attr_src_value.trim();
let frame_full_url = resolve_url(&url, frame_src).unwrap_or_default();
// Ignore (i)frames with empty source (they cause infinite loops)
if !frame_src.is_empty() {
let frame_full_url =
resolve_url(&url, frame_src).unwrap_or_default();
let frame_url_fragment = get_url_fragment(frame_full_url.clone());
match retrieve_asset(
cache,
@ -1261,22 +1169,21 @@ pub fn walk_and_embed_assets(
&frame_data,
&frame_final_url,
);
attr.value.clear();
let assembled_url: String = url_with_fragment(
frame_data_url.as_str(),
frame_url_fragment.as_str(),
);
attr.value.push_slice(assembled_url.as_str());
set_node_attr(node, "src", Some(assembled_url));
}
Err(_) => {
// Keep remote reference if unable to retrieve the asset
if is_http_url(frame_full_url.clone()) {
attr.value.clear();
let assembled_url: String = url_with_fragment(
frame_full_url.as_str(),
frame_url_fragment.as_str(),
);
attr.value.push_slice(assembled_url.as_str());
set_node_attr(node, "src", Some(assembled_url));
}
}
}
}
@ -1284,23 +1191,14 @@ pub fn walk_and_embed_assets(
}
}
"video" => {
for attr in attrs_mut.iter_mut() {
let attr_name: &str = &attr.name.local;
if attr_name.eq_ignore_ascii_case("poster") {
let video_poster_url = attr.value.trim();
if let Some(video_attr_poster_value) = get_node_attr(node, "poster") {
// Skip posters with empty source
if video_poster_url.is_empty() {
continue;
}
if !video_attr_poster_value.is_empty() {
if options.no_images {
attr.value.clear();
continue;
}
set_node_attr(node, "poster", Some(str!(empty_image!())));
} else {
let video_poster_full_url =
resolve_url(&url, video_poster_url).unwrap_or_default();
resolve_url(&url, video_attr_poster_value).unwrap_or_default();
let video_poster_url_fragment =
get_url_fragment(video_poster_full_url.clone());
match retrieve_asset(
@ -1321,22 +1219,21 @@ pub fn walk_and_embed_assets(
&video_poster_data,
&video_poster_final_url,
);
attr.value.clear();
let assembled_url: String = url_with_fragment(
video_poster_data_url.as_str(),
video_poster_url_fragment.as_str(),
);
attr.value.push_slice(assembled_url.as_str());
set_node_attr(node, "poster", Some(assembled_url));
}
Err(_) => {
// Keep remote reference if unable to retrieve the asset
if is_http_url(video_poster_full_url.clone()) {
attr.value.clear();
let assembled_url: String = url_with_fragment(
video_poster_full_url.as_str(),
video_poster_url_fragment.as_str(),
);
attr.value.push_slice(assembled_url.as_str());
set_node_attr(node, "poster", Some(assembled_url));
}
}
}
}
@ -1349,39 +1246,21 @@ pub fn walk_and_embed_assets(
// Process style attributes
if options.no_css {
// Get rid of style attributes
let mut i = 0;
while i < attrs_mut.len() {
let attr_name: &str = &attrs_mut[i].name.local;
if attr_name.eq_ignore_ascii_case("style") {
attrs_mut.remove(i);
set_node_attr(node, "style", None);
} else {
i += 1;
}
}
} else {
// Otherwise, parse any links found in the attributes
for attribute in attrs_mut
.iter_mut()
.filter(|a| a.name.local.as_ref().eq_ignore_ascii_case("style"))
{
let replacement = embed_css(
cache,
client,
&url,
attribute.value.as_ref(),
options,
depth,
);
// let replacement = str!();
attribute.value.clear();
attribute.value.push_slice(&replacement);
// Embed URLs found within the style attribute of this node
if let Some(node_attr_style_value) = get_node_attr(node, "style") {
let embedded_style =
embed_css(cache, client, &url, &node_attr_style_value, options, depth);
set_node_attr(node, "style", Some(embedded_style));
}
}
if options.no_js {
let attrs_mut = &mut attrs.borrow_mut();
// Get rid of JS event attributes
let mut js_attr_indexes = Vec::new();
for (i, attr) in attrs_mut.iter_mut().enumerate() {
for (i, attr) in attrs_mut.iter().enumerate() {
if attr_is_event_handler(&attr.name.local) {
js_attr_indexes.push(i);
}

4
src/tests/cli/basic.rs
Unescape Escape View File

@ -337,7 +337,7 @@ mod passing {
<title>Local HTML file</title>\n \
<link rel=\"stylesheet\" type=\"text/css\">\n \
<link rel=\"stylesheet\" type=\"text/css\">\n</head>\n\n<body>\n \
<img alt=\"\" src=\"{empty_image}\">\n \
<img src=\"{empty_image}\" alt=\"\">\n \
<a href=\"file://local-file.html/\">Tricky href</a>\n \
<a href=\"https://github.com/Y2Z/monolith\">Remote URL</a>\n \
<script></script>\n\n\n\n\
@ -399,7 +399,7 @@ mod passing {
<title>Local HTML file</title>\n \
<link rel=\"stylesheet\" type=\"text/css\">\n \
<link rel=\"stylesheet\" type=\"text/css\">\n</head>\n\n<body>\n \
<img alt=\"\" src=\"{empty_image}\">\n \
<img src=\"{empty_image}\" alt=\"\">\n \
<a href=\"file://local-file.html/\">Tricky href</a>\n \
<a href=\"https://github.com/Y2Z/monolith\">Remote URL</a>\n \
<script></script>\n\n\n\n\

39
src/tests/html/set_node_attr.rs
Unescape Escape View File

@ -63,4 +63,43 @@ mod passing {
assert_eq!(count, 5);
}
#[test]
fn body_background() {
let html = "<!doctype html><html lang=\"en\"><head></head><body background=\"1\" background=\"2\"></body></html>";
let dom = html::html_to_dom(&html);
let mut count = 0;
fn test_walk(node: &Handle, i: &mut i8) {
*i += 1;
match &node.data {
NodeData::Document => {
// Dig deeper
for child in node.children.borrow().iter() {
test_walk(child, &mut *i);
}
}
NodeData::Element { ref name, .. } => {
let node_name = name.local.as_ref().to_string();
if node_name == "body" {
assert_eq!(html::get_node_attr(node, "background"), Some(str!("1")));
html::set_node_attr(node, "background", None);
assert_eq!(html::get_node_attr(node, "background"), None);
}
for child in node.children.borrow().iter() {
test_walk(child, &mut *i);
}
}
_ => (),
};
}
test_walk(&dom.document, &mut count);
assert_eq!(count, 5);
}
}

4
src/tests/html/walk_and_embed_assets.rs
Unescape Escape View File

@ -319,8 +319,8 @@ mod passing {
buf.iter().map(|&c| c as char).collect::<String>(),
"<html>\
<head>\
<meta>\
<meta>\
<meta http-equiv=\"disabled by monolith (Refresh)\" value=\"20\">\
<meta http-equiv=\"disabled by monolith (Location)\" value=\"https://freebsd.org\">\
</head>\
<body></body>\
</html>"

Write Preview
Loading…
Cancel
Save