docs: link multipage version of HTML spec

1 files changed, 47 insertions, 0 deletions

diff --git a/tests/misc.rs b/tests/misc.rs
new file mode 100644
index 0000000..416e506
--- /dev/null
+++ b/tests/misc.rs
@@ -0,0 +1,47 @@
+use similar_asserts::assert_eq;
+use walkdir::{DirEntry, WalkDir};
+
+#[test]
+fn links_to_html_spec_use_multipage_version() {
+    for entry in WalkDir::new(".")
+        .min_depth(1)
+        .into_iter()
+        .filter_entry(is_source_file)
+        .flatten()
+    {
+        if !entry.file_type().is_file() {
+            continue;
+        }
+
+        let actual = match std::fs::read_to_string(entry.path()) {
+            Ok(content) => content,
+            Err(err) => panic!("invalid UTF-8 in file content: {:?}: {}", entry.path(), err),
+        };
+
+        let expected = actual.replace(
+            concat!("://html.spec.whatwg.org/", "#"),
+            concat!("://html.spec.whatwg.org/multipage/???.html#"),
+        );
+
+        assert_eq!(
+            actual,
+            expected,
+            "Found a link to the one-page version of the HTML spec, which is huge and takes long to load. We want to link the multipage version instead."
+        );
+    }
+}
+
+fn is_source_file(entry: &DirEntry) -> bool {
+    let Some(filename) = entry.file_name().to_str() else {
+        panic!("invalid UTF-8 in filename: {:?}", entry.path())
+    };
+
+    if entry.depth() == 1 && filename == "target" || filename == "Cargo.lock" {
+        return false; // cargo files
+    }
+    if filename == "html5lib-tests" {
+        return false; // git submodule
+    }
+
+    !filename.starts_with('.') // .git, etc.
+}