Output parse errors for the Rust part of the build step

This fixes #290, by outputting the parse errors encountered by the Rust build step's parser. Previously they were being stored in the RcDom instance's errors vector, and ignored. Now they are threaded through to the final io::Result, and then output by main(). The hardest part of this was adding line numbers to the errors. Doing this necessitated creating a wrapper for RcDom, called RcDomWithLineNumbers, which implements TreeSink with two methods parse_error() and set_current_line() given custom behavior, while the other many methods just delegate to RcDom's implementation. Additionally, this enables exact_errors as a parser option, which provides slightly more information in a couple of cases related to character references.
whatwg · Apr 25, 2024 · 06326f7 · 06326f7
1 parent 6757955
commit 06326f7
Show file tree

Hide file tree

Showing 10 changed files with 303 additions and 42 deletions.
diff --git a/Cargo.lock b/Cargo.lock
diff --git a/Cargo.toml b/Cargo.toml
@@ -11,6 +11,7 @@ tokio = { version = "1", features = ["full"] }
 html5ever = "0.26.0"
 markup5ever_rcdom = "0.2.0"
 regex = "1"
+delegate = "0.12.0"
 
 [dev-dependencies]
 tempfile = "3"
diff --git a/src/annotate_attributes.rs b/src/annotate_attributes.rs
@@ -313,6 +313,7 @@ mod tests {
         // reordered in the HTML spec).
         let document = parse_document_async(
             r#"
+<!DOCTYPE html>
 <h3>The a element</h3>
 <dl class="element">
     <dt>Categories
@@ -338,7 +339,7 @@ mod tests {
         assert_eq!(
             serialize_for_test(&[document]),
             r#"
-<html><head></head><body><h3>The a element</h3>
+<!DOCTYPE html><html><head></head><body><h3>The a element</h3>
 <dl class="element">
     <dt>Categories
     </dt><dd>Flow content
@@ -369,6 +370,7 @@ mod tests {
         // i.e., the variant description is used where requested
         let document = parse_document_async(
             r#"
+<!DOCTYPE html>
 <h3>The a element</h3>
 <dl class="element">
     <dt><span data-x="concept-element-attributes">Content attributes</span>
@@ -390,7 +392,7 @@ mod tests {
         assert_eq!(
             serialize_for_test(&[document]),
             r#"
-<html><head></head><body><h3>The a element</h3>
+<!DOCTYPE html><html><head></head><body><h3>The a element</h3>
 <dl class="element">
     <dt><span data-x="concept-element-attributes">Content attributes</span>
     </dt><dd><code data-x="attr-a-href">href</code>
@@ -415,6 +417,7 @@ mod tests {
         // Checks that the special rules for using : instead of an em dash work.
         let document = parse_document_async(
             r#"
+<!DOCTYPE html>
 <h3>The a element</h3>
 <dl class="element">
     <dt><span data-x="concept-element-attributes">Content attributes</span>
@@ -431,7 +434,7 @@ mod tests {
         assert_eq!(
             serialize_for_test(&[document]),
             r#"
-<html><head></head><body><h3>The a element</h3>
+<!DOCTYPE html><html><head></head><body><h3>The a element</h3>
 <dl class="element">
     <dt><span data-x="concept-element-attributes">Content attributes</span>
     </dt><dd>Also, the <code data-x="attr-a-name">name</code> attribute <span data-x="attr-a-name">has special semantics</span> on this element: Anchor name
@@ -450,6 +453,7 @@ mod tests {
         // Checks that the special rules for joining any special semantics with a ; work.
         let document = parse_document_async(
             r#"
+<!DOCTYPE html>
 <h3>The a element</h3>
 <dl class="element">
     <dt><span data-x="concept-element-attributes">Content attributes</span>
@@ -467,7 +471,7 @@ mod tests {
         assert_eq!(
             serialize_for_test(&[document]),
             r#"
-<html><head></head><body><h3>The a element</h3>
+<!DOCTYPE html><html><head></head><body><h3>The a element</h3>
 <dl class="element">
     <dt><span data-x="concept-element-attributes">Content attributes</span>
     </dt><dd>Also, the <code data-x="attr-a-name">name</code> attribute <span data-x="attr-a-name">has special semantics</span> on this element: Anchor name; Name of the anchor
@@ -488,6 +492,7 @@ mod tests {
         // repeating the description.
         let document = parse_document_async(
             r#"
+<!DOCTYPE html>
 <h3>The img element</h3>
 <dl class="element">
     <dt><span data-x="concept-element-attributes">Content attributes</span>
@@ -509,7 +514,7 @@ mod tests {
         assert_eq!(
             serialize_for_test(&[document]),
             r#"
-<html><head></head><body><h3>The img element</h3>
+<!DOCTYPE html><html><head></head><body><h3>The img element</h3>
 <dl class="element">
     <dt><span data-x="concept-element-attributes">Content attributes</span>
     </dt><dd><code data-x="attr-dim-width">width</code>

diff --git a/src/boilerplate.rs b/src/boilerplate.rs
@@ -170,14 +170,16 @@ mod tests {
             "<tr><td>en<td>English",
         )
         .await?;
-        let document =
-            parse_document_async("<table><!--BOILERPLATE languages-->".as_bytes()).await?;
+        let document = parse_document_async(
+            "<!DOCTYPE html><table><!--BOILERPLATE languages--></table>".as_bytes(),
+        )
+        .await?;
         let mut proc = Processor::new(boilerplate_dir.path(), Path::new("."));
         dom_utils::scan_dom(&document, &mut |h| proc.visit(h));
         proc.apply().await?;
         assert_eq!(
             serialize_for_test(&[document]),
-            "<html><head></head><body><table><tbody><tr><td>en</td><td>English</td></tr></tbody></table></body></html>");
+            "<!DOCTYPE html><html><head></head><body><table><tbody><tr><td>en</td><td>English</td></tr></tbody></table></body></html>");
         Ok(())
     }
 
@@ -189,15 +191,16 @@ mod tests {
             "data:text/html,Hello, world!",
         )
         .await?;
-        let document =
-            parse_document_async("<a href=\"<!--BOILERPLATE data.url-->\">hello</a>".as_bytes())
-                .await?;
+        let document = parse_document_async(
+            "<!DOCTYPE html><a href=\"<!--BOILERPLATE data.url-->\">hello</a>".as_bytes(),
+        )
+        .await?;
         let mut proc = Processor::new(boilerplate_dir.path(), Path::new("."));
         dom_utils::scan_dom(&document, &mut |h| proc.visit(h));
         proc.apply().await?;
         assert_eq!(
             serialize_for_test(&[document]),
-            "<html><head></head><body><a href=\"data:text/html,Hello, world!\">hello</a></body></html>");
+            "<!DOCTYPE html><html><head></head><body><a href=\"data:text/html,Hello, world!\">hello</a></body></html>");
         Ok(())
     }
 
@@ -208,23 +211,23 @@ mod tests {
         tokio::fs::write(example_dir.path().join("ex2"), "second").await?;
         tokio::fs::write(example_dir.path().join("ignored"), "bad").await?;
         let document =
-            parse_document_async("<pre>EXAMPLE ex1</pre><pre><code class=html>\nEXAMPLE ex2  </code></pre><p>EXAMPLE ignored</p>".as_bytes())
+            parse_document_async("<!DOCTYPE html><pre>EXAMPLE ex1</pre><pre><code class=html>\nEXAMPLE ex2  </code></pre><p>EXAMPLE ignored</p>".as_bytes())
                 .await?;
         let mut proc = Processor::new(Path::new("."), example_dir.path());
         dom_utils::scan_dom(&document, &mut |h| proc.visit(h));
         proc.apply().await?;
         assert_eq!(
             serialize_for_test(&[document]),
-            "<html><head></head><body><pre>first</pre><pre><code class=\"html\">second</code></pre><p>EXAMPLE ignored</p></body></html>" );
+            "<!DOCTYPE html><html><head></head><body><pre>first</pre><pre><code class=\"html\">second</code></pre><p>EXAMPLE ignored</p></body></html>" );
         Ok(())
     }
 
     #[tokio::test]
     async fn test_errors_unsafe_paths() -> io::Result<()> {
         let bad_path_examples = [
-            "<body><!--BOILERPLATE /etc/passwd-->",
-            "<body><pre data-x=\"<!--BOILERPLATE src/../../foo-->\"></pre>",
-            "<body><pre>EXAMPLE ../foo</pre>",
+            "<!DOCTYPE html><body><!--BOILERPLATE /etc/passwd-->",
+            "<!DOCTYPE html><body><pre data-x=\"<!--BOILERPLATE src/../../foo-->\"></pre>",
+            "<!DOCTYPE html><body><pre>EXAMPLE ../foo</pre>",
         ];
         for example in bad_path_examples {
             let document = parse_document_async(example.as_bytes()).await?;

diff --git a/src/interface_index.rs b/src/interface_index.rs
@@ -188,6 +188,7 @@ mod tests {
     async fn test_two_interfaces_in_one_block() -> io::Result<()> {
         let document = parse_document_async(
             r#"
+<!DOCTYPE html>
 <pre><code class=idl>
 interface <dfn interface>HTMLMarqueeElement</dfn> { ... }
 interface <dfn interface>HTMLBlinkElement</dfn> { ... }
@@ -204,7 +205,7 @@ INSERT INTERFACES HERE
         assert_eq!(
             serialize_for_test(&[document]),
             r#"
-<html><head></head><body><pre><code class="idl">
+<!DOCTYPE html><html><head></head><body><pre><code class="idl">
 interface <dfn interface="">HTMLMarqueeElement</dfn> { ... }
 interface <dfn interface="">HTMLBlinkElement</dfn> { ... }
 </code></pre>
@@ -217,6 +218,7 @@ interface <dfn interface="">HTMLBlinkElement</dfn> { ... }
     async fn test_two_interfaces_in_separate_blocks() -> io::Result<()> {
         let document = parse_document_async(
             r#"
+<!DOCTYPE html>
 <pre><code class=idl>
 interface <dfn interface>HTMLMarqueeElement</dfn> { ... }
 </code></pre>
@@ -235,7 +237,7 @@ INSERT INTERFACES HERE
         assert_eq!(
             serialize_for_test(&[document]),
             r#"
-<html><head></head><body><pre><code class="idl">
+<!DOCTYPE html><html><head></head><body><pre><code class="idl">
 interface <dfn interface="">HTMLMarqueeElement</dfn> { ... }
 </code></pre>
 <pre><code class="idl">
@@ -250,6 +252,7 @@ interface <dfn interface="">HTMLBlinkElement</dfn> { ... }
     async fn interface_with_partial() -> io::Result<()> {
         let document = parse_document_async(
             r#"
+<!DOCTYPE html>
 <pre><code class=idl>
 interface <dfn interface>HTMLMarqueeElement</dfn> { ... }
 </code></pre>
@@ -268,7 +271,7 @@ INSERT INTERFACES HERE
         assert_eq!(
             serialize_for_test(&[document]),
             r##"
-<html><head></head><body><pre><code class="idl">
+<!DOCTYPE html><html><head></head><body><pre><code class="idl">
 interface <dfn interface="">HTMLMarqueeElement</dfn> { ... }
 </code></pre>
 <pre><code class="idl">
@@ -283,6 +286,7 @@ partial interface <span id="HTMLMarqueeElement-partial">HTMLMarqueeElement</span
     async fn interface_with_two_partials() -> io::Result<()> {
         let document = parse_document_async(
             r#"
+<!DOCTYPE html>
 <pre><code class=idl>
 interface <dfn interface>HTMLMarqueeElement</dfn> { ... }
 partial interface <span id=HTMLMarqueeElement-partial>HTMLMarqueeElement</span> { ... }
@@ -300,7 +304,7 @@ INSERT INTERFACES HERE
         assert_eq!(
             serialize_for_test(&[document]),
             r##"
-<html><head></head><body><pre><code class="idl">
+<!DOCTYPE html><html><head></head><body><pre><code class="idl">
 interface <dfn interface="">HTMLMarqueeElement</dfn> { ... }
 partial interface <span id="HTMLMarqueeElement-partial">HTMLMarqueeElement</span> { ... }
 partial interface <span id="HTMLMarqueeElement-partial-2">HTMLMarqueeElement</span> { ... }
@@ -314,6 +318,7 @@ partial interface <span id="HTMLMarqueeElement-partial-2">HTMLMarqueeElement</sp
     async fn only_partials() -> io::Result<()> {
         let document = parse_document_async(
             r#"
+<!DOCTYPE html>
 <pre><code class=idl>
 partial interface <span id=HTMLMarqueeElement-partial>HTMLMarqueeElement</span> { ... }
 partial interface <span id=HTMLMarqueeElement-partial-2>HTMLMarqueeElement</span> { ... }
@@ -330,7 +335,7 @@ INSERT INTERFACES HERE
         assert_eq!(
             serialize_for_test(&[document]),
             r##"
-<html><head></head><body><pre><code class="idl">
+<!DOCTYPE html><html><head></head><body><pre><code class="idl">
 partial interface <span id="HTMLMarqueeElement-partial">HTMLMarqueeElement</span> { ... }
 partial interface <span id="HTMLMarqueeElement-partial-2">HTMLMarqueeElement</span> { ... }
 </code></pre>
@@ -343,6 +348,7 @@ partial interface <span id="HTMLMarqueeElement-partial-2">HTMLMarqueeElement</sp
     async fn marker_before() -> io::Result<()> {
         let document = parse_document_async(
             r#"
+<!DOCTYPE html>
 INSERT INTERFACES HERE
 <pre><code class=idl>
 interface <dfn interface>HTMLMarqueeElement</dfn> { ... }
@@ -357,20 +363,20 @@ interface <dfn interface>HTMLMarqueeElement</dfn> { ... }
         proc.apply()?;
         assert_eq!(
             serialize_for_test(&[document]),
-            r##"
-<html><head></head><body><ul class="brief"><li><code>HTMLMarqueeElement</code></li></ul>
+            r#"
+<!DOCTYPE html><html><head></head><body><ul class="brief"><li><code>HTMLMarqueeElement</code></li></ul>
 <pre><code class="idl">
 interface <dfn interface="">HTMLMarqueeElement</dfn> { ... }
 </code></pre></body></html>
-            "##
+            "#
             .trim()
         );
         Ok(())
     }
 
     #[tokio::test]
     async fn no_marker() -> io::Result<()> {
-        let document = parse_document_async("".as_bytes()).await?;
+        let document = parse_document_async("<!DOCTYPE html>".as_bytes()).await?;
         let mut proc = Processor::new();
         dom_utils::scan_dom(&document, &mut |h| proc.visit(h));
         let result = proc.apply();
@@ -381,7 +387,8 @@ interface <dfn interface="">HTMLMarqueeElement</dfn> { ... }
     #[tokio::test]
     async fn duplicate_marker() -> io::Result<()> {
         let document = parse_document_async(
-            "<div>INSERT INTERFACES HERE</div><div>INSERT INTERFACES HERE</div>".as_bytes(),
+            "<!DOCTYPE html><div>INSERT INTERFACES HERE</div><div>INSERT INTERFACES HERE</div>"
+                .as_bytes(),
         )
         .await?;
         let mut proc = Processor::new();
@@ -395,6 +402,7 @@ interface <dfn interface="">HTMLMarqueeElement</dfn> { ... }
     async fn duplicate_dfn() -> io::Result<()> {
         let document = parse_document_async(
             r#"
+<!DOCTYPE html>
 <pre><code class=idl>
 interface <dfn interface>HTMLMarqueeElement</dfn> { ... }
 interface <dfn interface>HTMLMarqueeElement</dfn> { ... }

diff --git a/src/main.rs b/src/main.rs
@@ -14,11 +14,21 @@ mod dom_utils;
 mod interface_index;
 mod io_utils;
 mod parser;
+mod rcdom_with_line_numbers;
 mod represents;
 mod tag_omission;
 
 #[tokio::main]
 async fn main() -> io::Result<()> {
+    // This gives slightly prettier error-printing.
+    if let Err(e) = run().await {
+        eprintln!("{}", e);
+        std::process::exit(1);
+    }
+    Ok(())
+}
+
+async fn run() -> io::Result<()> {
     // Since we're using Rc in the DOM implementation, we must ensure that tasks
     // which act on it are confined to this thread.