Fix tests

deadlinks · Nov 10, 2020 · 945b680 · 945b680
1 parent 3420cb8
commit 945b680
Showing 1 changed file with 13 additions and 24 deletions.
diff --git a/src/parse.rs b/src/parse.rs
@@ -11,14 +11,19 @@ pub fn parse_html_file(root_dir: &Path, path: &Path) -> HashSet<Url> {
     let html = std::fs::read_to_string(path)
         .unwrap_or_else(|e| panic!("{} did not contain valid UTF8: {}", path.display(), e));
 
-    // base_url is the relative file path. For `target/dir/crate_x/y`, it's `crate_x/y`.
-    let base_url = Url::from_file_path(path).unwrap();
     // root_url is absolute *relative to* the documentation directory. For `target/dir/crate_x/y`, it's `crate_x`.
     let root_url = Url::from_directory_path(root_dir).unwrap();
+    // base_url is the relative file path. For `target/dir/crate_x/y`, it's `crate_x/y`.
+    let base_url = Url::from_file_path(path).unwrap();
+
+    parse_a_hrefs(&html, root_url, base_url)
+}
 
+/// This is a pure function, unlike `parse_html_file`, allowing it to be easily tested.
+fn parse_a_hrefs(html: &str, root_url: Url, base_url: Url) -> HashSet<Url> {
     let mut urls = HashSet::new();
     lol_html::rewrite_str(
-        &html,
+        html,
         RewriteStrSettings {
             element_content_handlers: vec![element!("a[href]", |el| {
                 let href = el.get_attribute("href").unwrap();
@@ -48,25 +53,9 @@ pub fn parse_html_file(root_dir: &Path, path: &Path) -> HashSet<Url> {
 
 #[cfg(test)]
 mod test {
-    use html5ever::parse_document;
-    use html5ever::{rcdom::RcDom, tendril::TendrilSink};
-    use std::collections::HashSet;
+    use super::parse_a_hrefs as gather_urls;
     use url::Url;
 
-    use super::parse_a_hrefs;
-
-    fn gather_urls(html: &str, root: &Url, url: &Url) -> HashSet<Url> {
-        let dom = parse_document(RcDom::default(), Default::default())
-            .from_utf8()
-            .read_from(&mut html.as_bytes())
-            .unwrap();
-
-        let mut urls = HashSet::new();
-        parse_a_hrefs(&dom.document, &root, &url, &mut urls);
-
-        return urls;
-    }
-
     #[test]
     fn test_parse_a_hrefs() {
         let html = r#"
@@ -80,8 +69,8 @@ mod test {
 
         let urls = gather_urls(
             html,
-            &Url::from_directory_path("/base").unwrap(),
-            &Url::from_file_path("/base/test.html").unwrap(),
+            Url::from_directory_path("/base").unwrap(),
+            Url::from_file_path("/base/test.html").unwrap(),
         );
 
         assert!(urls.contains(&Url::from_file_path("/base/a.html").unwrap()));
@@ -102,8 +91,8 @@ mod test {
 
         let urls = gather_urls(
             html,
-            &Url::from_directory_path("/root").unwrap(),
-            &Url::from_file_path("/root/base/test.html").unwrap(),
+            Url::from_directory_path("/root").unwrap(),
+            Url::from_file_path("/root/base/test.html").unwrap(),
         );
 
         assert!(urls.contains(&Url::from_file_path("/root/base/a.html").unwrap()));