Make print page (print.html) links link to anchors on the print page

Let all the anchors id on the print page to have a path id prefix to
help locate.

e.g. bar/foo.md#abc -> #bar-foo-abc

Also append a dummy div to the start of the original page to make sure
that original page links without an anchor can also be located.

Fix to remove all the `./` in the normalized path id so that for
"./foo/bar.html#abc" we still get "#foo-bar-abc"

Add support for redirect link anchors in print page so that anchors can
also be redirected, also handle URL redirect links on print page

Handle all the elements id to add a path prefix, also make path id to
all be the lower case

Fix for print page footnote links by adding the path id prefix

Signed-off-by: Hollow Man <[email protected]>

src/renderer/html_handlebars/hbs_renderer.rs

@@ -57,16 +57,38 @@ impl HtmlHandlebars {
         let content = ch.content.clone();
         let content = utils::render_markdown(&content, ctx.html_config.curly_quotes);
 
-        let fixed_content =
-            utils::render_markdown_with_path(&ch.content, ctx.html_config.curly_quotes, Some(path));
+        let fixed_content = utils::render_markdown_with_path(
+            &ch.content,
+            ctx.html_config.curly_quotes,
+            Some(path),
+            ctx.html_config.redirect,
+        );
         if !ctx.is_index && ctx.html_config.print.page_break {
             // Add page break between chapters
             // See https://developer.mozilla.org/en-US/docs/Web/CSS/break-before and https://developer.mozilla.org/en-US/docs/Web/CSS/page-break-before
             // Add both two CSS properties because of the compatibility issue
             print_content
                 .push_str(r#"<div style="break-before: page; page-break-before: always;"></div>"#);
         }
-        print_content.push_str(&fixed_content);
+        let path_id = {
+            let mut base = path.display().to_string();
+            if base.ends_with(".md") {
+                base.replace_range(base.len() - 3.., "");
+            }
+            &base
+                .replace("/", "-")
+                .replace("\\", "-")
+                .to_ascii_lowercase()
+        };
+
+        // We have to build header links in advance so that we can know the ranges
+        // for the headers in one page.
+        // Insert a dummy div to make sure that we can locate the specific page.
+        print_content.push_str(&(format!(r#"<div id="{}"></div>"#, &path_id)));
+        print_content.push_str(&build_header_links(
+            &build_print_element_id(&fixed_content, &path_id),
+            Some(path_id),
+        ));
 
         // Update the context with data for this file
         let ctx_path = path
@@ -211,7 +233,7 @@ impl HtmlHandlebars {
         code_config: &Code,
         edition: Option<RustEdition>,
     ) -> String {
-        let rendered = build_header_links(&rendered);
+        let rendered = build_header_links(&rendered, None);
         let rendered = fix_code_blocks(&rendered);
         let rendered = add_playground_pre(&rendered, playground_config, edition);
         let rendered = hide_lines(&rendered, code_config);
@@ -809,9 +831,43 @@ fn make_data(
     Ok(data)
 }
 
+/// Goes through part of the rendered print page HTML,
+/// add path id prefix to all the elements id as well as footnote links.
+fn build_print_element_id(html: &str, path_id: &str) -> String {
+    let all_id = Regex::new(r#"(<[^>]*?id=")([^"]+?)""#).unwrap();
+    let footnote_id = Regex::new(
+        r##"(<sup [^>]*?class="footnote-reference"[^>]*?>[^<]*?<a [^>]*?href="#)([^"]+?)""##,
+    )
+    .unwrap();
+
+    if path_id.is_empty() {
+        return html.to_string();
+    }
+
+    let temp_html = all_id
+        .replace_all(html, |caps: &Captures<'_>| {
+            let mut fixed = String::new();
+            fixed.push_str(&path_id);
+            fixed.push_str("-");
+            fixed.push_str(&caps[2]);
+            format!("{}{}\"", &caps[1], fixed)
+        })
+        .into_owned();
+
+    footnote_id
+        .replace_all(&temp_html, |caps: &Captures<'_>| {
+            let mut fixed = String::new();
+            fixed.push_str(&path_id);
+            fixed.push_str("-");
+            fixed.push_str(&caps[2]);
+            format!("{}{}\"", &caps[1], fixed)
+        })
+        .into_owned()
+}
+
 /// Goes through the rendered HTML, making sure all header tags have
 /// an anchor respectively so people can link to sections directly.
-fn build_header_links(html: &str) -> String {
+fn build_header_links(html: &str, path_id: Option<&str>) -> String {
     static BUILD_HEADER_LINKS: Lazy<Regex> = Lazy::new(|| {
         Regex::new(r#"<h(\d)(?: id="([^"]+)")?(?: class="([^"]+)")?>(.*?)</h\d>"#).unwrap()
     });
@@ -840,21 +896,29 @@ fn build_header_links(html: &str) -> String {
                 caps.get(2).map(|x| x.as_str().to_string()),
                 caps.get(3).map(|x| x.as_str().to_string()),
                 &mut id_counter,
+                path_id,
             )
         })
         .into_owned()
 }
 
 /// Insert a sinle link into a header, making sure each link gets its own
 /// unique ID by appending an auto-incremented number (if necessary).
+///
+/// For `print.html`, we will add a path id prefix.
 fn insert_link_into_header(
     level: usize,
     content: &str,
     id: Option<String>,
     classes: Option<String>,
     id_counter: &mut HashMap<String, usize>,
+    path_id: Option<&str>,
 ) -> String {
-    let id = id.unwrap_or_else(|| utils::unique_id_from_content(content, id_counter));
+    let id = if let Some(path_id) = path_id {
+        id.unwrap_or_else(|| utils::unique_id_from_content_with_path(content, id_counter, path_id))
+    } else {
+        id.unwrap_or_else(|| utils::unique_id_from_content(content, id_counter))
+    };
     let classes = classes
         .map(|s| format!(" class=\"{s}\""))
         .unwrap_or_default();
@@ -1142,7 +1206,7 @@ mod tests {
         ];
 
         for (src, should_be) in inputs {
-            let got = build_header_links(src);
+            let got = build_header_links(src, None);
             assert_eq!(got, should_be);
         }
     }