upgrade html5ever to 0.26.0

Y2Z · Jan 14, 2024 · f7e5527 · f7e5527
1 parent f7dd09d
commit f7e5527
Show file tree

Hide file tree

Showing 9 changed files with 233 additions and 257 deletions.
diff --git a/Cargo.lock b/Cargo.lock
diff --git a/Cargo.toml b/Cargo.toml
@@ -28,7 +28,8 @@ chrono = "0.4.31"  # Used for formatting creation timestamp
 clap = "3.2.25"  # Used for processing CLI arguments
 cssparser = "0.33.0"  # Used for dealing with CSS
 encoding_rs = "0.8.33"  # Used for parsing and converting document charsets
-html5ever = "0.24.1"  # Used for all things DOM
+html5ever = "0.26.0"  # Used for all things DOM
+markup5ever_rcdom = "0.2.0"
 percent-encoding = "2.3.1"
 sha2 = "0.10.8"  # Used for calculating checksums during integrity checks
 url = "2.5.0"

diff --git a/src/html.rs b/src/html.rs
@@ -3,11 +3,11 @@ use chrono::prelude::*;
 use encoding_rs::Encoding;
 use html5ever::interface::QualName;
 use html5ever::parse_document;
-use html5ever::rcdom::{Handle, NodeData, RcDom};
 use html5ever::serialize::{serialize, SerializeOpts};
 use html5ever::tendril::{format_tendril, TendrilSink};
 use html5ever::tree_builder::{Attribute, TreeSink};
 use html5ever::{local_name, namespace_url, ns, LocalName};
+use markup5ever_rcdom::{Handle, NodeData, RcDom, SerializableHandle};
 use regex::Regex;
 use reqwest::blocking::Client;
 use reqwest::Url;
@@ -32,8 +32,12 @@ const ICON_VALUES: &'static [&str] = &["icon", "shortcut icon"];
 
 pub fn add_favicon(document: &Handle, favicon_data_url: String) -> RcDom {
     let mut buf: Vec<u8> = Vec::new();
-    serialize(&mut buf, document, SerializeOpts::default())
-        .expect("unable to serialize DOM into buffer");
+    serialize(
+        &mut buf,
+        &SerializableHandle::from(document.clone()),
+        SerializeOpts::default(),
+    )
+    .expect("unable to serialize DOM into buffer");
 
     let mut dom = html_to_dom(&buf, "utf-8".to_string());
     let doc = dom.get_document();
@@ -428,8 +432,12 @@ pub fn is_icon(attr_value: &str) -> bool {
 
 pub fn set_base_url(document: &Handle, desired_base_href: String) -> RcDom {
     let mut buf: Vec<u8> = Vec::new();
-    serialize(&mut buf, document, SerializeOpts::default())
-        .expect("unable to serialize DOM into buffer");
+    serialize(
+        &mut buf,
+        &SerializableHandle::from(document.clone()),
+        SerializeOpts::default(),
+    )
+    .expect("unable to serialize DOM into buffer");
 
     let mut dom = html_to_dom(&buf, "utf-8".to_string());
     let doc = dom.get_document();
@@ -534,7 +542,7 @@ pub fn set_node_attr(node: &Handle, attr_name: &str, attr_value: Option<String>)
 
 pub fn serialize_document(mut dom: RcDom, document_encoding: String, options: &Options) -> Vec<u8> {
     let mut buf: Vec<u8> = Vec::new();
-    let doc = dom.get_document();
+    let document = dom.get_document();
 
     if options.isolate
         || options.no_css
@@ -544,7 +552,7 @@ pub fn serialize_document(mut dom: RcDom, document_encoding: String, options: &O
         || options.no_images
     {
         // Take care of CSP
-        if let Some(html) = get_child_node_by_name(&doc, "html") {
+        if let Some(html) = get_child_node_by_name(&document, "html") {
             if let Some(head) = get_child_node_by_name(&html, "head") {
                 let meta = dom.create_element(
                     QualName::new(None, ns!(), local_name!("meta")),
@@ -570,8 +578,12 @@ pub fn serialize_document(mut dom: RcDom, document_encoding: String, options: &O
         }
     }
 
-    serialize(&mut buf, &doc, SerializeOpts::default())
-        .expect("Unable to serialize DOM into buffer");
+    serialize(
+        &mut buf,
+        &SerializableHandle::from(document.clone()),
+        SerializeOpts::default(),
+    )
+    .expect("Unable to serialize DOM into buffer");
 
     // Unwrap NOSCRIPT elements
     if options.unwrap_noscript {
@@ -660,7 +672,7 @@ pub fn retrieve_and_embed_asset(
                     let mut frame_data: Vec<u8> = Vec::new();
                     serialize(
                         &mut frame_data,
-                        &frame_dom.document,
+                        &SerializableHandle::from(frame_dom.document.clone()),
                         SerializeOpts::default(),
                     )
                     .unwrap();
@@ -1198,8 +1210,12 @@ pub fn walk_and_embed_assets(
                                 {
                                     if let Some(body) = get_child_node_by_name(&html, "body") {
                                         let mut buf: Vec<u8> = Vec::new();
-                                        serialize(&mut buf, &body, SerializeOpts::default())
-                                            .expect("Unable to serialize DOM into buffer");
+                                        serialize(
+                                            &mut buf,
+                                            &SerializableHandle::from(body.clone()),
+                                            SerializeOpts::default(),
+                                        )
+                                        .expect("Unable to serialize DOM into buffer");
                                         let result = String::from_utf8_lossy(&buf);
                                         noscript_contents.push_slice(&result);
                                     }

diff --git a/src/main.rs b/src/main.rs
@@ -1,5 +1,5 @@
 use encoding_rs::Encoding;
-use html5ever::rcdom::RcDom;
+use markup5ever_rcdom::RcDom;
 use reqwest::blocking::Client;
 use reqwest::header::{HeaderMap, HeaderValue, USER_AGENT};
 use std::collections::HashMap;

diff --git a/tests/html/add_favicon.rs b/tests/html/add_favicon.rs
@@ -8,6 +8,7 @@
 #[cfg(test)]
 mod passing {
     use html5ever::serialize::{serialize, SerializeOpts};
+    use markup5ever_rcdom::SerializableHandle;
 
     use monolith::html;
 
@@ -19,7 +20,12 @@ mod passing {
         dom = html::add_favicon(&dom.document, "I_AM_A_FAVICON_DATA_URL".to_string());
 
         let mut buf: Vec<u8> = Vec::new();
-        serialize(&mut buf, &dom.document, SerializeOpts::default()).unwrap();
+        serialize(
+            &mut buf,
+            &SerializableHandle::from(dom.document.clone()),
+            SerializeOpts::default(),
+        )
+        .unwrap();
 
         assert_eq!(
             buf.iter().map(|&c| c as char).collect::<String>(),

diff --git a/tests/html/get_node_attr.rs b/tests/html/get_node_attr.rs
@@ -7,7 +7,7 @@
 
 #[cfg(test)]
 mod passing {
-    use html5ever::rcdom::{Handle, NodeData};
+    use markup5ever_rcdom::{Handle, NodeData};
 
     use monolith::html;
 

diff --git a/tests/html/get_node_name.rs b/tests/html/get_node_name.rs
@@ -7,7 +7,7 @@
 
 #[cfg(test)]
 mod passing {
-    use html5ever::rcdom::{Handle, NodeData};
+    use markup5ever_rcdom::{Handle, NodeData};
 
     use monolith::html;
 

diff --git a/tests/html/set_node_attr.rs b/tests/html/set_node_attr.rs
@@ -7,7 +7,7 @@
 
 #[cfg(test)]
 mod passing {
-    use html5ever::rcdom::{Handle, NodeData};
+    use markup5ever_rcdom::{Handle, NodeData};
 
     use monolith::html;
 

diff --git a/tests/html/walk_and_embed_assets.rs b/tests/html/walk_and_embed_assets.rs
@@ -8,6 +8,7 @@
 #[cfg(test)]
 mod passing {
     use html5ever::serialize::{serialize, SerializeOpts};
+    use markup5ever_rcdom::SerializableHandle;
     use reqwest::blocking::Client;
     use std::collections::HashMap;
     use url::Url;
@@ -32,7 +33,12 @@ mod passing {
         html::walk_and_embed_assets(cache, &client, &url, &dom.document, &options, 0);
 
         let mut buf: Vec<u8> = Vec::new();
-        serialize(&mut buf, &dom.document, SerializeOpts::default()).unwrap();
+        serialize(
+            &mut buf,
+            &SerializableHandle::from(dom.document.clone()),
+            SerializeOpts::default(),
+        )
+        .unwrap();
 
         assert_eq!(
             buf.iter().map(|&c| c as char).collect::<String>(),
@@ -55,7 +61,12 @@ mod passing {
         html::walk_and_embed_assets(cache, &client, &url, &dom.document, &options, 0);
 
         let mut buf: Vec<u8> = Vec::new();
-        serialize(&mut buf, &dom.document, SerializeOpts::default()).unwrap();
+        serialize(
+            &mut buf,
+            &SerializableHandle::from(dom.document.clone()),
+            SerializeOpts::default(),
+        )
+        .unwrap();
 
         assert_eq!(
             buf.iter().map(|&c| c as char).collect::<String>(),
@@ -78,7 +89,12 @@ mod passing {
         html::walk_and_embed_assets(cache, &client, &url, &dom.document, &options, 0);
 
         let mut buf: Vec<u8> = Vec::new();
-        serialize(&mut buf, &dom.document, SerializeOpts::default()).unwrap();
+        serialize(
+            &mut buf,
+            &SerializableHandle::from(dom.document.clone()),
+            SerializeOpts::default(),
+        )
+        .unwrap();
 
         assert_eq!(
             buf.iter().map(|&c| c as char).collect::<String>(),
@@ -107,7 +123,12 @@ mod passing {
         html::walk_and_embed_assets(cache, &client, &url, &dom.document, &options, 0);
 
         let mut buf: Vec<u8> = Vec::new();
-        serialize(&mut buf, &dom.document, SerializeOpts::default()).unwrap();
+        serialize(
+            &mut buf,
+            &SerializableHandle::from(dom.document.clone()),
+            SerializeOpts::default(),
+        )
+        .unwrap();
 
         assert_eq!(
             buf.iter().map(|&c| c as char).collect::<String>(),
@@ -143,7 +164,12 @@ mod passing {
         html::walk_and_embed_assets(cache, &client, &url, &dom.document, &options, 0);
 
         let mut buf: Vec<u8> = Vec::new();
-        serialize(&mut buf, &dom.document, SerializeOpts::default()).unwrap();
+        serialize(
+            &mut buf,
+            &SerializableHandle::from(dom.document.clone()),
+            SerializeOpts::default(),
+        )
+        .unwrap();
 
         assert_eq!(
             buf.iter().map(|&c| c as char).collect::<String>(),
@@ -180,7 +206,12 @@ mod passing {
         html::walk_and_embed_assets(cache, &client, &url, &dom.document, &options, 0);
 
         let mut buf: Vec<u8> = Vec::new();
-        serialize(&mut buf, &dom.document, SerializeOpts::default()).unwrap();
+        serialize(
+            &mut buf,
+            &SerializableHandle::from(dom.document.clone()),
+            SerializeOpts::default(),
+        )
+        .unwrap();
 
         assert_eq!(
             buf.iter().map(|&c| c as char).collect::<String>(),
@@ -204,7 +235,12 @@ mod passing {
         html::walk_and_embed_assets(cache, &client, &url, &dom.document, &options, 0);
 
         let mut buf: Vec<u8> = Vec::new();
-        serialize(&mut buf, &dom.document, SerializeOpts::default()).unwrap();
+        serialize(
+            &mut buf,
+            &SerializableHandle::from(dom.document.clone()),
+            SerializeOpts::default(),
+        )
+        .unwrap();
 
         assert_eq!(
             buf.iter().map(|&c| c as char).collect::<String>(),
@@ -236,7 +272,12 @@ mod passing {
         html::walk_and_embed_assets(cache, &client, &url, &dom.document, &options, 0);
 
         let mut buf: Vec<u8> = Vec::new();
-        serialize(&mut buf, &dom.document, SerializeOpts::default()).unwrap();
+        serialize(
+            &mut buf,
+            &SerializableHandle::from(dom.document.clone()),
+            SerializeOpts::default(),
+        )
+        .unwrap();
 
         assert_eq!(
             buf.iter().map(|&c| c as char).collect::<String>(),
@@ -272,7 +313,12 @@ mod passing {
         html::walk_and_embed_assets(cache, &client, &url, &dom.document, &options, 0);
 
         let mut buf: Vec<u8> = Vec::new();
-        serialize(&mut buf, &dom.document, SerializeOpts::default()).unwrap();
+        serialize(
+            &mut buf,
+            &SerializableHandle::from(dom.document.clone()),
+            SerializeOpts::default(),
+        )
+        .unwrap();
 
         assert_eq!(
             buf.iter().map(|&c| c as char).collect::<String>(),
@@ -306,7 +352,12 @@ mod passing {
         html::walk_and_embed_assets(cache, &client, &url, &dom.document, &options, 0);
 
         let mut buf: Vec<u8> = Vec::new();
-        serialize(&mut buf, &dom.document, SerializeOpts::default()).unwrap();
+        serialize(
+            &mut buf,
+            &SerializableHandle::from(dom.document.clone()),
+            SerializeOpts::default(),
+        )
+        .unwrap();
 
         assert_eq!(
             buf.iter().map(|&c| c as char).collect::<String>(),
@@ -343,7 +394,12 @@ mod passing {
         html::walk_and_embed_assets(cache, &client, &url, &dom.document, &options, 0);
 
         let mut buf: Vec<u8> = Vec::new();
-        serialize(&mut buf, &dom.document, SerializeOpts::default()).unwrap();
+        serialize(
+            &mut buf,
+            &SerializableHandle::from(dom.document.clone()),
+            SerializeOpts::default(),
+        )
+        .unwrap();
 
         assert_eq!(
             buf.iter().map(|&c| c as char).collect::<String>(),
@@ -381,7 +437,12 @@ mod passing {
         html::walk_and_embed_assets(cache, &client, &url, &dom.document, &options, 0);
 
         let mut buf: Vec<u8> = Vec::new();
-        serialize(&mut buf, &dom.document, SerializeOpts::default()).unwrap();
+        serialize(
+            &mut buf,
+            &SerializableHandle::from(dom.document.clone()),
+            SerializeOpts::default(),
+        )
+        .unwrap();
 
         assert_eq!(
             buf.iter().map(|&c| c as char).collect::<String>(),
@@ -427,7 +488,12 @@ mod passing {
         html::walk_and_embed_assets(cache, &client, &url, &dom.document, &options, 0);
 
         let mut buf: Vec<u8> = Vec::new();
-        serialize(&mut buf, &dom.document, SerializeOpts::default()).unwrap();
+        serialize(
+            &mut buf,
+            &SerializableHandle::from(dom.document.clone()),
+            SerializeOpts::default(),
+        )
+        .unwrap();
 
         assert_eq!(
             buf.iter().map(|&c| c as char).collect::<String>(),
@@ -466,7 +532,12 @@ mod passing {
         html::walk_and_embed_assets(cache, &client, &url, &dom.document, &options, 0);
 
         let mut buf: Vec<u8> = Vec::new();
-        serialize(&mut buf, &dom.document, SerializeOpts::default()).unwrap();
+        serialize(
+            &mut buf,
+            &SerializableHandle::from(dom.document.clone()),
+            SerializeOpts::default(),
+        )
+        .unwrap();
 
         assert_eq!(
             buf.iter().map(|&c| c as char).collect::<String>(),
@@ -501,7 +572,12 @@ mod passing {
         html::walk_and_embed_assets(cache, &client, &url, &dom.document, &options, 0);
 
         let mut buf: Vec<u8> = Vec::new();
-        serialize(&mut buf, &dom.document, SerializeOpts::default()).unwrap();
+        serialize(
+            &mut buf,
+            &SerializableHandle::from(dom.document.clone()),
+            SerializeOpts::default(),
+        )
+        .unwrap();
 
         assert_eq!(
             buf.iter().map(|&c| c as char).collect::<String>(),