fix: restore current blog RSS feed

apps/docs/__tests__/Meta_.test.res

@@ -1,5 +1,18 @@
 open Vitest
 
+test("advertises the blog RSS feed in page metadata", async () => {
+  let _screen = await render(
+    <ReactRouter.MemoryRouter initialEntries=["/blog"]>
+      <Meta />
+    </ReactRouter.MemoryRouter>,
+  )
+  let feed =
+    document->WebAPI.Document.querySelector("link[rel='alternate'][type='application/rss+xml']")
+  expect(feed->Null.toOption->Option.map(WebAPI.Element.getAttribute(_, "href")))->toBe(
+    Some(Null.make("/blog/feed.xml")),
+  )
+})
+
 let getMetaContent = name => {
   switch document->WebAPI.Document.querySelector(`meta[name='${name}']`) {
   | Value(element) =>

apps/docs/scripts/__tests__/BlogFeed_.test.res

@@ -0,0 +1,90 @@
+open ScriptTest
+
+@val @scope("process") external chdir: string => unit = "chdir"
+
+let fixture = (title, date) =>
+  `---
+author: rescript-team
+title: ${title}
+date: "${date}"
+---
+Article content.
+`
+
+let blogDirectory = async () => {
+  let directory = await temporaryDirectory("blog-feed-")
+  let blog = join([directory, "markdown-pages", "blog"])
+  await mkdir(join([blog, "archived"]), {recursive: true})
+  await write(join([blog, "z-old.mdx"]), fixture("Old post", "2019-11-28"))
+  await write(join([blog, "a-new.mdx"]), fixture("New post", "2026-04-28"))
+  await write(join([blog, "archived", "middle.mdx"]), fixture("Archived post", "2020-01-01"))
+  directory
+}
+
+let latestFrom = (directory, ~max=10) => {
+  let previousDirectory = cwd()
+  chdir(directory)
+  let items = try {
+    BlogApi.RssFeed.getLatest(~max, ())
+  } catch {
+  | exn =>
+    chdir(previousDirectory)
+    throw(exn)
+  }
+  chdir(previousDirectory)
+  items
+}
+
+test("RSS selects posts by publication date before applying the limit", async () => {
+  let directory = await blogDirectory()
+  let items = latestFrom(directory, ~max=2)
+  expect(items->Array.map(item => item.title))->toStrictEqual(["New post", "Archived post"])
+  expect(latestFrom(directory, ~max=0))->toStrictEqual([])
+})
+
+test("RSS links match article routes, including archived posts", async () => {
+  let directory = await blogDirectory()
+  let items = latestFrom(directory)
+  expect(items->Array.map(item => item.href))->toStrictEqual([
+    "https://rescript-lang.org/blog/a-new",
+    "https://rescript-lang.org/blog/archived/middle",
+    "https://rescript-lang.org/blog/z-old",
+  ])
+  let routes = MdxFile.scanPaths(~dir=join([directory, "markdown-pages", "blog"]), ~alias="blog")
+  items->Array.forEach(item => {
+    let route = item.href->String.replace("https://rescript-lang.org/", "")
+    expect(routes->Array.includes(route))->toBe(true)
+  })
+})
+
+test("blog slugs preserve route filenames and existing slugs", async () => {
+  expect(BlogApi.blogPathToSlug("release-12-0-0.mdx"))->toBe("release-12-0-0")
+  expect(BlogApi.blogPathToSlug("archived/whats-new-in-7-pt2.mdx"))->toBe(
+    "archived/whats-new-in-7-pt2",
+  )
+  expect(BlogApi.blogPathToSlug("archived/2020-01-01-post.mdx"))->toBe("archived/2020-01-01-post")
+  expect(BlogApi.blogPathToSlug("archived/middle"))->toBe("archived/middle")
+  expect(BlogApi.blogPathToSlug("archived\\middle.mdx"))->toBe("archived/middle")
+})
+
+test("RSS dates use midnight UTC and XML serialization leaves dates unchanged", async () => {
+  let directory = await blogDirectory()
+  let items = latestFrom(directory)
+  let before = items->Array.map(item => item.pubDate->Date.getTime)
+  let xml = items->BlogApi.RssFeed.toXmlString
+  expect(xml)->toContain("<lastBuildDate>Tue, 28 Apr 2026 00:00:00 GMT</lastBuildDate>")
+  expect(xml)->toContain("<pubDate>Thu, 28 Nov 2019 00:00:00 GMT</pubDate>")
+  expect(items->Array.map(item => item.pubDate->Date.getTime))->toStrictEqual(before)
+  expect(items->BlogApi.RssFeed.toXmlString)->toBe(xml)
+})
+
+test("the current blog feed contains ten recent posts with existing routes", async () => {
+  let items = BlogApi.RssFeed.getLatest()
+  let routes = MdxFile.scanPaths(~dir="markdown-pages/blog", ~alias="blog")
+  expect(items->Array.length)->toBe(10)
+  items->Array.forEach(item => {
+    expect(
+      routes->Array.includes(item.href->String.replace("https://rescript-lang.org/", "")),
+    )->toBe(true)
+  })
+})

apps/docs/src/common/DateStr.res

@@ -6,4 +6,7 @@ let parse = (dateStr: string): Date.t => dateStr->String.replaceRegExp(/-/g, "/"
 let fromDate = date => Date.toString(date)
 let toDate = dateStr => parse(dateStr)
 
+// ISO date-only strings represent midnight UTC, independent of the build timezone.
+let toUTCDate = dateStr => Date.fromString(dateStr)
+
 external fromString: string => t = "%identity"

apps/docs/src/common/DateStr.resi

@@ -5,3 +5,4 @@ external fromString: string => t = "%identity"
 
 let fromDate: Date.t => t
 let toDate: t => Date.t
+let toUTCDate: t => Date.t

apps/docs/src/data/BlogApi.res

@@ -1,27 +1,5 @@
-// Same design as in the official nextjs blog starter template
-// https://github.com/zeit/next.js/blob/canary/examples/blog-starter/lib/api.js
-
-/*
-    BlogApi gives access to the data within the markdown-pages/blog directory.
-
-    The markdown-pages/blog content is treated as a flat directory. There is a
-    set of whitelisted directories (for each category) to help declutter
-    huge file masses.
-
-    Every filename within markdown-pages/blog must be unique, even when nested in
-    different directories. The filename will be sluggified, that means that e.g.
-    for the filepath "markdown-pages/blog/compiler/2021-03-12-my-post.mdx", slug = "my-post"
-
-    The reason for the flat hierarchy is that we want to keep a flat URL
-    hierarchy with `https://rescript-lang.org/blog/[some-slug]`, without carrying
-    over the full subdirectory hierarchy. This allows us to restructure
-    internally more easily without breaking permalinks.
-
-    This module offers functionality to query all slug information, and even
-    gives access to fully parsed blog posts with frontmatter data etc.
-
-    See `pages/blog.re` for more context on why we need this API.
- */
+// Blog paths are relative to markdown-pages/blog. Article routes preserve the
+// filename and any archived/ prefix, removing only the .mdx extension.
 
 type post = {
   path: string,
@@ -30,7 +8,7 @@ type post = {
 }
 
 let blogPathToSlug = path => {
-  path->String.replaceRegExp(/^(archive\/)?\d\d\d\d-\d\d-\d\d-(.+)\.mdx$/, "$2")
+  path->String.replaceAll("\\", "/")->String.replaceRegExp(/\.mdx$/, "")
 }
 
 let mdxFiles = dir => {
@@ -68,7 +46,8 @@ let getAllPosts = () => {
   })
 
   Array.concat(nonArchivedPosts, archivedPosts)->Array.toSorted((a, b) =>
-    String.compare(Node.Path.basename(b.path), Node.Path.basename(a.path))
+    b.frontmatter.date->DateStr.toUTCDate->Date.getTime -.
+      a.frontmatter.date->DateStr.toUTCDate->Date.getTime
   )
 }
 
@@ -116,56 +95,36 @@ let getArchivedPosts = () => {
 }
 
 module RssFeed = {
-  // Module inspired by
-  // https://gist.github.com/fredrikbergqvist/36704828353ebf5379a5c08c7583fe2d
-
-  // Example reference example for RSS 2.0:
-  // http://static.userland.com/gems/backend/rssTwoExample2.xml
-
-  // TODO:
-  // In case the pubDate formatting doesn't work with toUTCString,
-  // we can port something like this:
-  // https://github.com/tjconcept/js-rfc822-date
-
   type item = {
     title: string,
     href: string,
     description: string,
     pubDate: Date.t,
   }
 
-  // TODO: This is yet again a dirty approach to prevent UTC to substract too many
-  //       hours of my local timezone so it does end up on another day, so we set the hours
-  //       to 15 o clock. We need to reconsider the way we parse blog article dates,
-  //       since the dates should always be parsed from a single timezone perspective
-  let dateToUTCString = date => {
-    date->Date.setHours(15)->ignore
-    date->Date.toUTCString
-  }
-
   // Retrieves the most recent [max] blog post feed items
   let getLatest = (~max=10, ~baseUrl="https://rescript-lang.org", ()): array<item> => {
     let items =
       getAllPosts()
+      ->Array.slice(~start=0, ~end=max)
       ->Array.map(post => {
         let fm = post.frontmatter
         let description = fm.description->Nullable.getOr("")
         {
           title: fm.title,
           href: baseUrl ++ "/blog/" ++ blogPathToSlug(post.path),
           description,
-          pubDate: DateStr.toDate(fm.date),
+          pubDate: DateStr.toUTCDate(fm.date),
         }
       })
-      ->Array.slice(~start=0, ~end=max)
     items
   }
 
   let toXmlString = (~siteTitle="ReScript Blog", ~siteDescription="", items: array<item>) => {
     let latestPubDateElement =
       items[0]
       ->Option.map(item => {
-        let latestPubDateStr = item.pubDate->dateToUTCString
+        let latestPubDateStr = item.pubDate->Date.toUTCString
         `<lastBuildDate>${latestPubDateStr}</lastBuildDate>`
       })
       ->Option.getOr("")
@@ -181,8 +140,7 @@ module RssFeed = {
           </description>`
         }
 
-        // TODO: convert pubdate to string
-        let dateStr = pubDate->dateToUTCString
+        let dateStr = pubDate->Date.toUTCString
         `
         <item>
           <title> <![CDATA[${title}]]></title>
@@ -204,7 +162,7 @@ module RssFeed = {
       ${latestPubDateElement}
 ${itemsStr}
     </channel>
-  </rss>` //rescript-lang.org</link>
+  </rss>`
 
     ret
   }