Merge branch 'feat-sitemap'

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "blog-tools"
-version = "0.1.2"
+version = "0.2.0"
 edition = "2021"
 authors = ["Indigo Curnick <indigocurnick@gmail.com>"]
 description = "A collection of tools that helps make blogs in Rust"
@@ -17,6 +17,7 @@ walkdir = "2.3.2"
 chrono = { version = "0.4.23", features = ["serde"] }
 markdown = "1.0.0-alpha.11"
 tl = "0.7.8"
+xml = "0.8.20"
 
 
 [dev-dependencies]

README.md

@@ -17,7 +17,7 @@ next to a `blog.json`.
 
 The JSON must conform to the following schema
 
-```rust,ignore
+```json,ignore
 {
 "title": String,
 "date": ISO 8601 Date i.e. YYYY-MM-DD,
@@ -27,7 +27,9 @@ The JSON must conform to the following schema
 "keywords": Optional<[String]>,
 "canonical_link": Optional<String>,
 "author_name": Optional<String>,
-"author_webpage": Optional<String>
+"author_webpage": Optional<String>,
+"last_modified": Optional<Date>, (ISO 8601)
+"priority": Optional<float>
 }
 ```
 
@@ -36,7 +38,6 @@ The JSON must conform to the following schema
 In `blog-tools` all slugs are /{date}/{sub-slug}.
 
 Make sure the "slug" filed in the JSON is *just* the final sub-slug
-- `blog-tools` will automatically handle the date for you
 
 ## How This Crate is Organised
 
@@ -56,7 +57,7 @@ the best way to do this is using a lazy static like so
 ```rust,ignore
 lazy_static! {
     pub static ref STATIC_BLOG_ENTRIES: HighBlog =
-        get_high_blog(PathBuf::from(BLOG_ROOT), None, None);
+        get_high_blog(PathBuf::from(BLOG_ROOT), None, None, URL, &SitemapOptions::default());
     }
 ```
 
@@ -66,7 +67,7 @@ blog posts themselves. These will need to be rendered when requested
 ```rust,ignore
 lazy_static! {
     pub static ref STATIC_BLOG_ENTRIES: MediumBlog =
-        get_medium_blog(PathBuf::from(BLOG_ROOT), None, None);
+        get_medium_blog(PathBuf::from(BLOG_ROOT), None, None, URL, &SitemapOptions::default());
     }
 
 let this_blog = match all_blogs.hash.get(&complete_slug) {
@@ -88,6 +89,7 @@ everything at runtime.
 let preview = preview_blogs(PathBuf::from_str(BLOG_ROOT).unwrap(), 2, None);
 let tags = get_blog_tag_list(PathBuf::from_str(BLOG_ROOT).unwrap());
 let blog_post = render_blog_post(PathBuf::from_str(BLOG_ROOT).unwrap(), date, slug, None).unwrap();
+let sitemap = create_sitemap(BLOG_ROOT, URL, &SitemapOptions::default());
 ```
 
 This method can have serious runtime performance implecations, but might be

examples/high/main.rs

@@ -1,10 +1,13 @@
+use std::fs;
 use std::path::PathBuf;
 
 use blog_tools::high::{get_high_blog, HighBlog, HighBlogEntry};
+use blog_tools::sitemap::SitemapOptions;
+use blog_tools::Blog;
 use lazy_static::lazy_static;
 use rocket::{
     fs::{relative, FileServer},
-    response::Redirect,
+    response::{content::RawXml, Redirect},
     Request, Route,
 };
 use rocket_dyn_templates::Template;
@@ -41,6 +44,13 @@ fn blog_index() -> Option<Template> {
     Some(Template::render("blog_index", context.into_json()))
 }
 
+#[get("/sitemap.xml")]
+fn sitemap() -> RawXml<String> {
+    let blog = get_blog_context();
+
+    return RawXml(blog.sitemap.clone());
+}
+
 #[get("/blog/<date>/<slug>", rank = 2)]
 fn blog_article(date: String, slug: String) -> Option<Template> {
     let mut context = rocket_dyn_templates::tera::Context::new();
@@ -62,7 +72,7 @@ fn tag_page(slug: String) -> Option<Template> {
     let mut these_blogs: Vec<&HighBlogEntry> = vec![];
 
     for blog in &all_blogs.entries {
-        if blog.tags.contains(&slug) {
+        if blog.get_tags().contains(&slug) {
             these_blogs.push(&blog);
         }
     }
@@ -87,14 +97,25 @@ async fn error(req: &Request<'_>) -> Redirect {
 }
 
 fn get_all_routes() -> Vec<Route> {
-    return routes![blog_index, blog_article, tag_page];
+    return routes![blog_index, blog_article, tag_page, sitemap];
 }
 
 pub static BLOG_ROOT: &str = "examples/blog";
+pub static URL: &str = "www.example.xyz";
 
 lazy_static! {
-    pub static ref STATIC_BLOG_ENTRIES: HighBlog =
-        get_high_blog(PathBuf::from(BLOG_ROOT), None, None).unwrap();
+    pub static ref STATIC_BLOG_ENTRIES: HighBlog = get_high_blog(
+        PathBuf::from(BLOG_ROOT),
+        None,
+        None,
+        &URL.to_string(),
+        &SitemapOptions {
+            include_tags: true,
+            sitemap_base: Some(fs::read_to_string("examples/xml/sitemap.xml").unwrap()),
+            ..Default::default()
+        }
+    )
+    .unwrap();
 }
 
 fn get_blog_context() -> &'static HighBlog {

examples/low/main.rs

@@ -1,14 +1,17 @@
-use blog_tools::low::{
-    get_blog_tag_list, preview_blogs, preview_blogs_tagged, render_blog_post, PreviewBlogEntry,
+use blog_tools::{
+    low::{
+        get_blog_tag_list, preview_blogs, preview_blogs_tagged, render_blog_post, PreviewBlogEntry,
+    },
+    sitemap::{create_sitemap, SitemapOptions},
 };
 use rocket::{
     fs::{relative, FileServer},
-    response::Redirect,
+    response::{content::RawXml, Redirect},
     Request, Route,
 };
 use rocket_dyn_templates::Template;
 use serde::{Deserialize, Serialize};
-use std::{path::PathBuf, str::FromStr};
+use std::{fs, path::PathBuf, str::FromStr};
 
 pub static BLOG_ROOT: &str = "examples/blog";
 
@@ -82,6 +85,24 @@ fn tag_page(slug: String) -> Option<Template> {
     Some(Template::render("tags", context.into_json()))
 }
 
+pub static URL: &str = "www.example.xyz";
+
+#[get("/sitemap.xml")]
+fn sitemap() -> RawXml<String> {
+    let base_sitemap = fs::read_to_string("examples/xml/sitemap.xml").unwrap();
+    let sitemap = create_sitemap(
+        BLOG_ROOT,
+        &URL.to_string(),
+        &SitemapOptions {
+            include_tags: true,
+            sitemap_base: Some(base_sitemap),
+            ..Default::default()
+        },
+    );
+
+    return RawXml(sitemap.unwrap());
+}
+
 #[catch(404)]
 async fn not_found(req: &Request<'_>) -> Redirect {
     let mut context = rocket_dyn_templates::tera::Context::new();
@@ -97,5 +118,5 @@ async fn error(req: &Request<'_>) -> Redirect {
 }
 
 fn get_all_routes() -> Vec<Route> {
-    return routes![blog_index, blog_article, tag_page];
+    return routes![blog_index, blog_article, tag_page, sitemap];
 }

examples/medium/main.rs

@@ -1,10 +1,14 @@
-use std::{path::PathBuf, str::FromStr};
+use std::{fs, path::PathBuf, str::FromStr};
 
-use blog_tools::medium::{get_medium_blog, MediumBlog, MediumBlogEntry};
+use blog_tools::{
+    medium::{get_medium_blog, MediumBlog, MediumBlogEntry},
+    sitemap::SitemapOptions,
+    Blog,
+};
 use lazy_static::lazy_static;
 use rocket::{
     fs::{relative, FileServer},
-    response::Redirect,
+    response::{content::RawXml, Redirect},
     Request, Route,
 };
 use rocket_dyn_templates::Template;
@@ -34,6 +38,13 @@ async fn main() {
     }
 }
 
+#[get("/sitemap.xml")]
+fn sitemap() -> RawXml<String> {
+    let blog = get_blog_context();
+
+    return RawXml(blog.sitemap.clone());
+}
+
 #[get("/blog")]
 fn blog_index() -> Option<Template> {
     let mut context = rocket_dyn_templates::tera::Context::new();
@@ -68,7 +79,7 @@ fn tag_page(slug: String) -> Option<Template> {
     let mut these_blogs: Vec<&MediumBlogEntry> = vec![];
 
     for blog in &all_blogs.entries {
-        if blog.tags.contains(&slug) {
+        if blog.get_tags().contains(&slug) {
             these_blogs.push(&blog);
         }
     }
@@ -93,14 +104,25 @@ async fn error(req: &Request<'_>) -> Redirect {
 }
 
 fn get_all_routes() -> Vec<Route> {
-    return routes![blog_index, blog_article, tag_page];
+    return routes![blog_index, blog_article, tag_page, sitemap];
 }
 
 pub static BLOG_ROOT: &str = "examples/blog";
+pub static URL: &str = "www.example.xyz";
 
 lazy_static! {
-    pub static ref STATIC_BLOG_ENTRIES: MediumBlog =
-        get_medium_blog(PathBuf::from(BLOG_ROOT), None, None).unwrap();
+    pub static ref STATIC_BLOG_ENTRIES: MediumBlog = get_medium_blog(
+        PathBuf::from(BLOG_ROOT),
+        None,
+        None,
+        &URL.to_string(),
+        &SitemapOptions {
+            include_tags: true,
+            sitemap_base: Some(fs::read_to_string("examples/xml/sitemap.xml").unwrap()),
+            ..Default::default()
+        }
+    )
+    .unwrap();
 }
 
 fn get_blog_context() -> &'static MediumBlog {

examples/xml/sitemap.xml

@@ -0,0 +1,7 @@
+<urlset>
+<url>
+<loc>www.example.xyz</loc>
+<lastmod>19-08-2023</lastmod>
+<priority>1.0</priority>
+</url>
+</urlset>