🐛 Honor the path component of --base

deno.json

@@ -45,6 +45,7 @@
       "main.ts",
       "staticalize.ts",
       "downloader.ts",
+      "rebase.ts",
       "task-buffer.ts",
       "config.ts",
       "progress.ts",

downloader.ts

@@ -11,6 +11,7 @@ import { fromHtml } from "hast-util-from-html";
 import { toHtml } from "hast-util-to-html";
 import { selectAll } from "hast-util-select";
 import { useTaskBuffer } from "./task-buffer.ts";
+import { rebase } from "./rebase.ts";
 import { createApi } from "@effectionx/context-api";
 
 export interface Downloader extends Operation<void> {
@@ -69,11 +70,7 @@ export const DownloadApi = createApi("@staticalize/download", {
 
           // replace self-referencing absolute urls with the destination site
           if (href.startsWith(host.origin)) {
-            let url = new URL(href);
-            url.host = base.host;
-            url.port = base.port;
-            url.protocol = base.protocol;
-            link.properties.href = url.href;
+            link.properties.href = rebase(new URL(href), base).href;
           }
         }
 
@@ -85,11 +82,7 @@ export const DownloadApi = createApi("@staticalize/download", {
 
           // replace self-referencing absolute urls with the destination site
           if (src.startsWith(host.origin)) {
-            let url = new URL(src);
-            url.host = base.host;
-            url.port = base.port;
-            url.protocol = base.protocol;
-            element.properties.src = url.href;
+            element.properties.src = rebase(new URL(src), base).href;
           }
         }
 
@@ -98,11 +91,7 @@ export const DownloadApi = createApi("@staticalize/download", {
           let attr = String(element.properties.content);
           if (attr.startsWith(host.origin)) {
             yield* downloader.download(attr, source);
-            let url = new URL(attr);
-            url.host = base.host;
-            url.port = base.port;
-            url.protocol = base.protocol;
-            element.properties.content = url.href;
+            element.properties.content = rebase(new URL(attr), base).href;
           }
         }
 

rebase.ts

@@ -0,0 +1,22 @@
+/**
+ * Map a url on the site being crawled onto the public base url.
+ *
+ * Everything the base says about where the site lives is honored, including
+ * its path, so a site published under a subpath such as
+ * `https://frontside.com/effection` gets that prefix on every url rewritten
+ * into it. The source's own path, query and fragment are preserved.
+ *
+ * @param source url on the crawled site
+ * @param base public base url of the site, path included
+ * @returns the equivalent url on the base
+ */
+export function rebase(source: URL, base: URL): URL {
+  let url = new URL(source.href);
+  url.protocol = base.protocol;
+  url.host = base.host;
+  url.port = base.port;
+  // `source.pathname` always begins with a `/`, so the prefix must not end
+  // with one. A base without a path contributes `/`, and therefore nothing.
+  url.pathname = `${base.pathname.replace(/\/$/, "")}${source.pathname}`;
+  return url;
+}

staticalize.ts

@@ -10,6 +10,7 @@ import { ensureDir } from "@std/fs/ensure-dir";
 import { stringify } from "@libs/xml/stringify";
 import { parse } from "@libs/xml/parse";
 import { useDownloader } from "./downloader.ts";
+import { rebase } from "./rebase.ts";
 
 export interface StaticalizeOptions {
   host: URL;
@@ -93,13 +94,9 @@ export function useStaticalizer(
           let xml = stringify({
             urlset: {
               "@xmlns": "http://www.sitemaps.org/schemas/sitemap/0.9",
-              "url": [...urls].map((url) => {
-                let loc = new URL(url);
-                loc.host = base.host;
-                loc.port = base.port;
-                loc.protocol = base.protocol;
-                return { loc: { "#text": loc } };
-              }),
+              "url": [...urls].map((url) => ({
+                loc: { "#text": rebase(new URL(url), base) },
+              })),
             },
           });
           yield* call(() =>

test/staticalize.test.ts

@@ -278,6 +278,52 @@ describe("staticalize", () => {
       `<meta property="og:url" content="https://fs.com/image.png">`,
     );
   });
+
+  it("carries the path of the base url into every rewritten url", async () => {
+    app.get(
+      "/",
+      (c) =>
+        c.html(`
+<html>
+  <head>
+    <link rel="canonical" href="${host}"/>
+    <meta property="og:url" content="${host}image.png"/>
+    <script src="${host}main.js"></script>
+  </head>
+  <body></body>
+</html>
+`),
+    )
+      .get("/about", (c) => c.html("<h1>About</h1>"))
+      .get("/image.png", (c) => c.text(""))
+      .get("/main.js", (c) => c.text("console.log('hi')"))
+      .get(...sitemap(["/", "/about"]));
+
+    await staticalize({
+      base: new URL("https://frontside.com/effection"),
+      host,
+      dir: "test/dist",
+    });
+
+    let index = await content("test/dist/index.html");
+    expect(index).toContain(
+      `<link rel="canonical" href="https://frontside.com/effection/">`,
+    );
+    expect(index).toContain(
+      `<script src="https://frontside.com/effection/main.js">`,
+    );
+    expect(index).toContain(
+      `<meta property="og:url" content="https://frontside.com/effection/image.png">`,
+    );
+
+    let sitemapXML = await Deno.readTextFile("test/dist/sitemap.xml");
+    expect(sitemapXML).toContain(
+      "<loc>https://frontside.com/effection/</loc>",
+    );
+    expect(sitemapXML).toContain(
+      "<loc>https://frontside.com/effection/about</loc>",
+    );
+  });
 });
 
 async function content(path: string): Promise<string> {