tags working now. rust parser works pretty good. next is cache

markdown_backend/src/markdown.rs

@@ -15,6 +15,7 @@ use std::time::Duration;
 use syntect::highlighting::{ThemeSet, Style};
 use syntect::parsing::SyntaxSet;
 use syntect::html::{highlighted_html_for_string, IncludeBackground};
+use once_cell::sync::Lazy;
 
 #[derive(Debug, Deserialize, Clone, serde::Serialize)]
 pub struct PostFrontmatter {
@@ -77,6 +78,39 @@ fn process_anchor_links(content: &str) -> String {
     }).to_string()
 }
 
+// Helper function to strip emojis from a string
+fn strip_emojis(s: &str) -> String {
+    // Remove all characters in the Emoji Unicode ranges
+    // This is a simple approach and may not cover all emojis, but works for most cases
+    // Unicode emoji ranges: https://unicode.org/Public/emoji/15.0/emoji-data.txt
+    s.chars()
+        .filter(|c| {
+            let c = *c as u32;
+            // Basic Emoji ranges
+            !( (c >= 0x1F600 && c <= 0x1F64F)  // Emoticons
+            || (c >= 0x1F300 && c <= 0x1F5FF)  // Misc Symbols and Pictographs
+            || (c >= 0x1F680 && c <= 0x1F6FF)  // Transport and Map
+            || (c >= 0x2600 && c <= 0x26FF)    // Misc symbols
+            || (c >= 0x2700 && c <= 0x27BF)    // Dingbats
+            || (c >= 0x1F900 && c <= 0x1F9FF)  // Supplemental Symbols and Pictographs
+            || (c >= 0x1FA70 && c <= 0x1FAFF)  // Symbols and Pictographs Extended-A
+            || (c >= 0x1F1E6 && c <= 0x1F1FF)  // Regional Indicator Symbols
+            )
+        })
+        .collect()
+}
+
+static AMMONIA: Lazy<ammonia::Builder<'static>> = Lazy::new(|| {
+    let mut builder = ammonia::Builder::default();
+    builder.add_tag_attributes("h1", &["id"]);
+    builder.add_tag_attributes("h2", &["id"]);
+    builder.add_tag_attributes("h3", &["id"]);
+    builder.add_tag_attributes("h4", &["id"]);
+    builder.add_tag_attributes("h5", &["id"]);
+    builder.add_tag_attributes("h6", &["id"]);
+    builder
+});
+
 pub fn get_post_by_slug(slug: &str) -> Result<Post, Box<dyn std::error::Error>> {
     let posts_dir = get_posts_directory();
     let file_path = posts_dir.join(format!("{}.md", slug));
@@ -122,7 +156,9 @@ pub fn get_post_by_slug(slug: &str) -> Result<Post, Box<dyn std::error::Error>>
             },
             Event::End(Tag::Heading(_, _, _)) => {
                 in_heading = false;
-                let id = slugify(&heading_text);
+                // Strip emojis before slugifying for the id
+                let heading_no_emoji = strip_emojis(&heading_text);
+                let id = slugify(&heading_no_emoji);
                 events.push(Event::Html(CowStr::Boxed(format!("<h{lvl} id=\"{id}\">", lvl=heading_level, id=id).into_boxed_str())));
                 events.push(Event::Text(CowStr::Boxed(heading_text.clone().into_boxed_str())));
                 events.push(Event::Html(CowStr::Boxed(format!("</h{lvl}>", lvl=heading_level).into_boxed_str())));
@@ -164,7 +200,7 @@ pub fn get_post_by_slug(slug: &str) -> Result<Post, Box<dyn std::error::Error>>
     }
     html::push_html(&mut html_output, events.into_iter());
 
-    let sanitized_html = clean(&html_output);
+    let sanitized_html = AMMONIA.clean(&html_output).to_string();
 
     Ok(Post {
         slug: slug.to_string(),