feat: generate sitemap and robots.txt

Cargo.toml

@@ -32,3 +32,4 @@ form_urlencoded = "1.2.1"
 anyhow = "1.0.86"
 thiserror = "1.0.61"
 fs_extra = "1.3.0"
+xml = "0.8.20"

configuration.yaml

@@ -19,3 +19,10 @@ markdown:
 
 # Copy everything in this directory to the output directory
 static_path: "static"
+
+# Baseurl
+base_url: "http://vincent.jousse.org"
+
+author: "Vincent Jousse"
+
+website_title: "Vince's"

src/config/mod.rs

@@ -13,6 +13,9 @@ pub struct Settings {
     pub create_index_for: Vec<String>,
     pub markdown: markup::Markdown,
     pub static_path: String,
+    pub base_url: String,
+    pub author: String,
+    pub website_title: String,
 }
 
 pub fn get_configuration() -> Result<Settings, config::ConfigError> {

src/content.rs

@@ -4,6 +4,7 @@
 use crate::post::{FrontMatter, Post, PostStatus};
 use crate::rss::write_atom_for_posts;
 use crate::site::Site;
+use crate::sitemap::write_sitemap_for_posts;
 use anyhow::Result;
 use form_urlencoded::byte_serialize;
 use gray_matter::engine::YAML;
@@ -50,11 +51,21 @@
 
     write_posts_html(&pages, site);
 
+    let mut posts_and_pages: Vec<Post> = Vec::new();
+    posts_and_pages.extend(posts.clone());
+    posts_and_pages.extend(pages.clone());
+
+    write_sitemap_for_posts(
+        &posts_and_pages,
+        &site.settings.base_url.as_str(),
+        Path::new(&format!("{}/sitemap.xml", &site.settings.output_path)[..]),
+    )?;
+
     write_atom_for_posts(
         &posts,
-        "https://vincent.jousse.org",
-        "Vincent Jousse",
-        "Vince's",
+        &site.settings.base_url.as_str(),
+        &site.settings.author.as_str(),
+        &site.settings.website_title.as_str(),
         Path::new(&format!("{}/atom.xml", &site.settings.output_path)[..]),
         &site.settings,
     )?;
@@ -187,17 +198,17 @@
     for post in posts {
         for directory in &post.ancestor_directories_paths {
             if let Some(dir_index) = directory.strip_prefix(&site.settings.posts_path) {
                 if dir_index != "" {
                     if site
                         .settings
                         .create_index_for
                         .contains(&dir_index.to_string())
                     {
                         let key = dir_index.to_string();
                         let posts = indexes_to_create.entry(key).or_default();
                         posts.push(post);
                     }
                 }
             };
         }
     }

src/lib.rs

@@ -5,6 +5,7 @@ pub mod errors;
 pub mod post;
 pub mod rss;
 pub mod site;
+pub mod sitemap;
 pub mod startup;
 pub mod templates;
 

src/sitemap.rs

@@ -0,0 +1,69 @@
+use crate::post::Post;
+
+use anyhow::Result;
+use std::fs::File;
+use std::path::Path;
+
+use xml::writer::{EmitterConfig, XmlEvent};
+
+pub fn write_sitemap_for_posts(posts: &[Post], base: &str, path: &Path) -> Result<()> {
+    let file = File::create(path).expect("Unable to create file");
+    let mut writer = EmitterConfig::new()
+        .perform_indent(true)
+        .create_writer(file);
+
+    let start_event: XmlEvent = XmlEvent::start_element("urlset")
+        .attr("xmlns", "http://www.sitemaps.org/schemas/sitemap/0.9")
+        .into();
+    writer.write(start_event)?;
+
+    let mut sorted_posts = posts.to_vec();
+    sorted_posts.sort_by(|p1, p2| p1.front_matter.date.cmp(&p2.front_matter.date));
+
+    for post in sorted_posts {
+        let start_event: XmlEvent = XmlEvent::start_element("url").into();
+        writer.write(start_event)?;
+
+        // <loc>
+        let start_event: XmlEvent = XmlEvent::start_element("loc").into();
+        writer.write(start_event)?;
+
+        // Avoid //
+        let url = format!(
+            "{}/{}",
+            base,
+            if let "/" = post.url_path.as_str() {
+                ""
+            } else {
+                post.url_path.as_str()
+            }
+        );
+        let event = xml::writer::XmlEvent::Characters(url.as_str());
+        writer.write(event)?;
+
+        let end_event: XmlEvent = XmlEvent::end_element().into();
+        writer.write(end_event)?;
+
+        // </loc>
+
+        // <lastmod>
+        let start_event: XmlEvent = XmlEvent::start_element("lastmod").into();
+        writer.write(start_event)?;
+
+        let event = xml::writer::XmlEvent::Characters(post.date_rfc3339.as_str());
+        writer.write(event)?;
+
+        let end_event: XmlEvent = XmlEvent::end_element().into();
+        writer.write(end_event)?;
+
+        // </loc>
+        let end_event: XmlEvent = XmlEvent::end_element().into();
+        writer.write(end_event)?;
+        // </lastmod>
+    }
+
+    let end_event: XmlEvent = XmlEvent::end_element().into();
+    writer.write(end_event)?;
+
+    Ok(())
+}

static/robots.txt

@@ -0,0 +1,4 @@
+User-agent: *
+Allow: /
+
+Sitemap: https://vincent.jousse.org/sitemap.xml