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 <hollowman@opensuse.org>

Author: HollowMan6

crates/mdbook-core/src/utils/mod.rs

@@ -55,7 +55,7 @@ pub fn normalize_id(content: &str) -> String {
 
 /// Generate an ID for use with anchors which is derived from a "normalised"
 /// string.
-fn id_from_content(content: &str) -> String {
+pub fn id_from_content(content: &str) -> String {
     let mut content = content.to_string();
 
     // Skip any tags or html-encoded stuff

crates/mdbook-html/src/html_handlebars/hbs_renderer.rs

@@ -67,7 +67,25 @@ impl HtmlHandlebars {
             print_content
                 .push_str(r#"<div style="break-before: page; page-break-before: always;"></div>"#);
         }
-        print_content.push_str(&fixed_content);
+        let print_page_id = {
+            let mut base = path.display().to_string();
+            if base.ends_with(".md") {
+                base.truncate(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="{print_page_id}"></div>"#)));
+        print_content.push_str(&build_header_links(
+            &build_print_element_id(&fixed_content, &print_page_id),
+            Some(print_page_id),
+        ));
 
         // Update the context with data for this file
         let ctx_path = path
@@ -239,7 +257,23 @@ impl HtmlHandlebars {
         code_config: &Code,
         edition: Option<RustEdition>,
     ) -> String {
-        let rendered = build_header_links(&rendered);
+        let rendered = build_header_links(&rendered, None);
+        let rendered = self.post_process_common(rendered, &playground_config, code_config, edition);
+
+        rendered
+    }
+
+    /// Applies some post-processing to the HTML to apply some adjustments.
+    ///
+    /// This common function is used for both normal chapters (via
+    /// `post_process`) and the combined print page.
+    fn post_process_common(
+        &self,
+        rendered: String,
+        playground_config: &Playground,
+        code_config: &Code,
+        edition: Option<RustEdition>,
+    ) -> String {
         let rendered = fix_code_blocks(&rendered);
         let rendered = add_playground_pre(&rendered, playground_config, edition);
         let rendered = hide_lines(&rendered, code_config);
@@ -497,7 +531,7 @@ impl Renderer for HtmlHandlebars {
             debug!("Render template");
             let rendered = handlebars.render("index", &data)?;
 
-            let rendered = self.post_process(
+            let rendered = self.post_process_common(
                 rendered,
                 &html_config.playground,
                 &html_config.code,
@@ -698,9 +732,35 @@ fn make_data(
     Ok(data)
 }
 
+/// Go through 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, print_page_id: &str) -> String {
+    static ALL_ID: LazyLock<Regex> =
+        LazyLock::new(|| Regex::new(r#"(<[^>]*?id=")([^"]+?)""#).unwrap());
+    static FOOTNOTE_ID: LazyLock<Regex> = LazyLock::new(|| {
+        Regex::new(
+            r##"(<sup [^>]*?class="footnote-reference"[^>]*?>[^<]*?<a [^>]*?href="#)([^"]+?)""##,
+        )
+        .unwrap()
+    });
+
+    let temp_html = ALL_ID.replace_all(html, |caps: &Captures<'_>| {
+        format!("{}{}-{}\"", &caps[1], print_page_id, &caps[2])
+    });
+
+    FOOTNOTE_ID
+        .replace_all(&temp_html, |caps: &Captures<'_>| {
+            format!("{}{}-{}\"", &caps[1], print_page_id, &caps[2])
+        })
+        .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 {
+///
+/// `print_page_id` should be set to the print page ID prefix when adjusting the
+/// print page.
+fn build_header_links(html: &str, print_page_id: Option<&str>) -> String {
     static_regex!(
         BUILD_HEADER_LINKS,
         r#"<h(\d)(?: id="([^"]+)")?(?: class="([^"]+)")?>(.*?)</h\d>"#
@@ -730,21 +790,34 @@ 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,
+                print_page_id,
             )
         })
         .into_owned()
 }
 
 /// Insert a single 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>,
+    print_page_id: Option<&str>,
 ) -> String {
-    let id = id.unwrap_or_else(|| utils::unique_id_from_content(content, id_counter));
+    let id = if let Some(print_page_id) = print_page_id {
+        let content_id = {
+            #[allow(deprecated)]
+            utils::id_from_content(content)
+        };
+        let with_prefix = format!("{} {}", print_page_id, content_id);
+        id.unwrap_or_else(|| utils::unique_id_from_content(&with_prefix, id_counter))
+    } else {
+        id.unwrap_or_else(|| utils::unique_id_from_content(content, id_counter))
+    };
     let classes = classes
         .map(|s| format!(" class=\"{s}\""))
         .unwrap_or_default();
@@ -1125,7 +1198,7 @@ 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);
         }
     }

crates/mdbook-html/src/html_handlebars/search.rs

@@ -134,7 +134,8 @@ fn render_item(
         .with_context(|| "Could not convert HTML path to str")?;
     let anchor_base = utils::fs::normalize_path(filepath);
 
-    let options = HtmlRenderOptions::new(&chapter_path);
+    let redirect = HashMap::new();
+    let options = HtmlRenderOptions::new(&chapter_path, &redirect);
     let mut p = new_cmark_parser(&chapter.content, &options.markdown_options).peekable();
 
     let mut in_heading = false;

crates/mdbook-html/src/lib.rs

@@ -11,9 +11,9 @@ use std::path::Path;
 /// Creates an [`HtmlRenderOptions`] from the given config.
 pub fn html_render_options_from_config<'a>(
     path: &'a Path,
-    config: &HtmlConfig,
+    config: &'a HtmlConfig,
 ) -> HtmlRenderOptions<'a> {
-    let mut options = HtmlRenderOptions::new(path);
+    let mut options = HtmlRenderOptions::new(path, &config.redirect);
     options.markdown_options.smart_punctuation = config.smart_punctuation;
     options
 }