mdbook-slide-evaluator: ignore redirect pages to speed up the evaluation when there are many redirects

michael-kerscher · michael-kerscher · commit 7d2507ec6050 · 2024-10-30T11:58:56.000+01:00
diff --git a/mdbook-slide-evaluator/src/main.rs b/mdbook-slide-evaluator/src/main.rs
@@ -61,6 +61,9 @@ struct Args {
     violations_only: bool,
     /// directory of the book that is evaluated
     source_dir: PathBuf,
+    /// ignore HTML pages that redirect to canonical pages
+    #[arg(long, default_value_t = false)]
+    ignore_redirects: bool,
 }
 
 #[tokio::main]
@@ -71,7 +74,8 @@ async fn main() -> anyhow::Result<()> {
     let args = Args::parse();
 
     // gather information about the book from the filesystem
-    let book = Book::from_html_slides(args.source_dir.clone())?;
+    let book =
+        Book::from_html_slides(args.source_dir.clone(), args.ignore_redirects)?;
 
     // create a new webclient that is used by the evaluator
     let webclient: fantoccini::Client =
diff --git a/mdbook-slide-evaluator/src/slides.rs b/mdbook-slide-evaluator/src/slides.rs
@@ -12,9 +12,12 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
+use std::fs::File;
+use std::io::Read;
 use std::path::{Path, PathBuf};
 use std::sync::Arc;
 
+use anyhow::Ok;
 use log::debug;
 
 /// a slide is a page in the book
@@ -33,17 +36,26 @@ pub struct Book {
 
 impl Book {
     /// create a book from all html files in the source_dir
-    pub fn from_html_slides(source_dir: PathBuf) -> anyhow::Result<Book> {
+    pub fn from_html_slides(
+        source_dir: PathBuf,
+        ignore_redirects: bool,
+    ) -> anyhow::Result<Book> {
         let mut slides = vec![];
         let files = glob::glob(&format!(
             "{}/**/*.html",
             source_dir.to_str().expect("invalid path")
         ))?;
         for file in files {
-            let slide = Slide { filename: file?.into() };
+            let file = file?;
+            if ignore_redirects && file_is_redirect(&file)? {
+                debug!("slide {file:?} is a redirect page");
+                continue;
+            }
+            let slide = Slide { filename: file.into() };
             debug!("add {:?}", slide);
             slides.push(slide);
         }
+        debug!("processing {} slides", slides.len());
         Ok(Book { _source_dir: source_dir, slides })
     }
 
@@ -52,3 +64,20 @@ impl Book {
         &self.slides
     }
 }
+
+const HTML_REDIRECT_PAGE: &str = r#"<!DOCTYPE html>
+<html lang="en">
+  <head>
+    <meta charset="utf-8">
+    <title>Redirecting...</title>"#;
+
+/// check if the file is starting with the mdbook redirect page.
+/// This method is optimized to not read the entire file but only the start
+fn file_is_redirect(filename: &PathBuf) -> anyhow::Result<bool> {
+    let mut file = File::open(filename)?;
+    // create a buffer with the exact length of the text that is checked
+    let mut file_start_buffer = [0u8; HTML_REDIRECT_PAGE.len()];
+    // read only the part that is relevant
+    file.read_exact(&mut file_start_buffer)?;
+    Ok(file_start_buffer.eq(HTML_REDIRECT_PAGE.as_bytes()))
+}
