promote local media assets to file attachments

src/files.ts

@@ -9,6 +9,7 @@ import {relativeUrl, resolvePath} from "./url.js";
 // A path is local if it doesn’t go outside the the root.
 export function getLocalPath(sourcePath: string, name: string): string | null {
   if (/^\w+:/.test(name)) return null; // URL
+  if (name.startsWith("#")) return null; // anchor tag
   const path = resolvePath(sourcePath, name);
   if (path.startsWith("../")) return null; // goes above root
   return path;

src/markdown.ts

@@ -17,7 +17,7 @@ import {computeHash} from "./hash.js";
 import {parseInfo} from "./info.js";
 import {type FileReference, type ImportReference, type Transpile, transpileJavaScript} from "./javascript.js";
 import {transpileTag} from "./tag.js";
-import {relativeUrl, resolvePath} from "./url.js";
+import {resolvePath} from "./url.js";
 
 export interface ReadMarkdownResult {
   contents: string;
@@ -316,22 +316,63 @@ function renderIntoPieces(renderer: Renderer, root: string, sourcePath: string):
     }
     let result = "";
     for (const piece of context.pieces) {
-      result += piece.html = normalizePieceHtml(piece.html, root, sourcePath, context);
+      result += piece.html = normalizePieceHtml(piece.html, sourcePath, context);
     }
     return result;
   };
 }
 
-function normalizePieceHtml(html: string, root: string, sourcePath: string, context: ParseContext): string {
+const SUPPORTED_PROPERTIES: readonly {query: string; src: "href" | "src" | "srcset"}[] = Object.freeze([
+  {query: "audio[src]", src: "src"},
+  {query: "audio source[src]", src: "src"},
+  {query: "img[src]", src: "src"},
+  {query: "img[srcset]", src: "srcset"},
+  {query: "link[href]", src: "href"},
+  {query: "picture source[srcset]", src: "srcset"},
+  {query: "video[src]", src: "src"},
+  {query: "video source[src]", src: "src"}
+]);
+export function normalizePieceHtml(html: string, sourcePath: string, context: ParseContext): string {
   const {document} = parseHTML(html);
 
   // Extracting references to files (such as from linked stylesheets).
-  for (const element of document.querySelectorAll("link[href]") as any as Iterable<Element>) {
-    const href = element.getAttribute("href")!;
-    const path = getLocalPath(sourcePath, href);
-    if (path) {
-      context.files.push(fileReference(href, sourcePath));
-      element.setAttribute("href", relativeUrl(sourcePath, join("_file", path)));
+  const filePaths = new Set<FileReference["path"]>();
+  for (const {query, src} of SUPPORTED_PROPERTIES) {
+    for (const element of document.querySelectorAll(query) as any as Iterable<Element>) {
+      if (src === "srcset") {
+        const srcset = element.getAttribute(src);
+        const paths =
+          srcset &&
+          srcset
+            .split(",")
+            .map((p) => {
+              const parts = p.trim().split(/\s+/);
+              const source = parts[0];
+              const path = getLocalPath(sourcePath, source);
+              if (path) {
+                const file = fileReference(source, sourcePath);
+                if (!filePaths.has(file.path)) {
+                  filePaths.add(file.path);
+                  context.files.push(file);
+                }
+                return `${file.path} ${parts.slice(1).join(" ")}`.trim();
+              }
+              return parts.join(" ");
+            })
+            .filter((p) => !!p);
+        if (paths && paths.length > 0) element.setAttribute(src, paths.join(", "));
+      } else {
+        const source = element.getAttribute(src);
+        const path = getLocalPath(sourcePath, source!);
+        if (path) {
+          const file = fileReference(source!, sourcePath);
+          if (!filePaths.has(file.path)) {
+            filePaths.add(file.path);
+            context.files.push(file);
+          }
+          element.setAttribute(src, file.path);
+        }
+      }
     }
   }
 

test/promote-file-attachment-test.ts

@@ -0,0 +1,204 @@
+import assert from "node:assert";
+import {normalizePieceHtml} from "../src/markdown.js";
+
+const html = (strings, ...values) => String.raw({raw: strings}, ...values);
+const mockContext = () => ({files: [], imports: [], pieces: [], startLine: 0, currentLine: 0});
+
+describe("file attachments", () => {
+  describe("added", () => {
+    const sourcePath = "/attachments.md";
+
+    it("img[src]", () => {
+      const htmlStr = html`<img src="./test.png">`;
+      const expected = html`<img src="./_file/test.png">`;
+      const context = mockContext();
+      const actual = normalizePieceHtml(htmlStr, sourcePath, context);
+
+      assert.equal(actual, expected);
+      assert.deepEqual(context.files, [
+        {
+          mimeType: "image/png",
+          name: "./test.png",
+          path: "./_file/test.png"
+        }
+      ]);
+    });
+
+    it("img[srcset]", () => {
+      const htmlStr = html`
+        <img
+          srcset="small.jpg 480w, large.jpg 800w"
+          sizes="(max-width: 600px) 480px,
+                800px"
+          src="large.jpg"
+          alt="Image for testing"
+        />
+      `;
+      const expected = html`
+        <img srcset="./_file/small.jpg 480w, ./_file/large.jpg 800w" sizes="(max-width: 600px) 480px,
+                800px" src="./_file/large.jpg" alt="Image for testing">
+      `;
+      const context = mockContext();
+      const actual = normalizePieceHtml(htmlStr, sourcePath, context);
+
+      assert.equal(actual, expected);
+      assert.deepEqual(context.files, [
+        {
+          mimeType: "image/jpeg",
+          name: "large.jpg",
+          path: "./_file/large.jpg"
+        },
+        {
+          mimeType: "image/jpeg",
+          name: "small.jpg",
+          path: "./_file/small.jpg"
+        }
+      ]);
+    });
+
+    it("video[src]", () => {
+      const htmlStr = html`<video src="observable.mov" controls>
+      Your browser doesn't support HTML video.
+      </video>`;
+      const expected = html`<video src="./_file/observable.mov" controls>
+      Your browser doesn't support HTML video.
+      </video>`;
+      const context = mockContext();
+      const actual = normalizePieceHtml(htmlStr, sourcePath, context);
+
+      assert.equal(actual, expected);
+      assert.deepEqual(context.files, [
+        {
+          mimeType: "video/quicktime",
+          name: "observable.mov",
+          path: "./_file/observable.mov"
+        }
+      ]);
+    });
+
+    it("video source[src]", () => {
+      const htmlStr = html`<video width="320" height="240" controls>
+      <source src="observable.mp4" type="video/mp4">
+      <source src="observable.mov" type="video/mov">
+      Your browser doesn't support HTML video.
+      </video>`;
+
+      const expected = html`<video width="320" height="240" controls>
+      <source src="./_file/observable.mp4" type="video/mp4">
+      <source src="./_file/observable.mov" type="video/mov">
+      Your browser doesn't support HTML video.
+      </video>`;
+
+      const context = mockContext();
+      const actual = normalizePieceHtml(htmlStr, sourcePath, context);
+
+      assert.equal(actual, expected);
+      assert.deepEqual(context.files, [
+        {
+          mimeType: "video/mp4",
+          name: "observable.mp4",
+          path: "./_file/observable.mp4"
+        },
+        {
+          mimeType: "video/quicktime",
+          name: "observable.mov",
+          path: "./_file/observable.mov"
+        }
+      ]);
+    });
+
+    it("picture source[srcset]", () => {
+      const htmlStr = html`<picture>
+      <source srcset="observable-logo-wide.png" media="(min-width: 600px)"/>
+      <img src="observable-logo-narrow.png" />
+    </picture>`;
+
+      const expected = html`<picture>
+      <source srcset="./_file/observable-logo-wide.png" media="(min-width: 600px)">
+      <img src="./_file/observable-logo-narrow.png">
+    </picture>`;
+
+      const context = mockContext();
+      const actual = normalizePieceHtml(htmlStr, sourcePath, context);
+
+      assert.equal(actual, expected);
+      assert.deepEqual(context.files, [
+        {
+          mimeType: "image/png",
+          name: "observable-logo-narrow.png",
+          path: "./_file/observable-logo-narrow.png"
+        },
+        {
+          mimeType: "image/png",
+          name: "observable-logo-wide.png",
+          path: "./_file/observable-logo-wide.png"
+        }
+      ]);
+    });
+  });
+
+  describe("not added", () => {
+    const sourcePath = "/attachments.md";
+
+    it("img[src] only adds local files", () => {
+      const htmlStr = html`<img src="https://upload.wikimedia.org/wikipedia/commons/thumb/0/0c/American_Shorthair.jpg/900px-American_Shorthair.jpg">`;
+      const expected = html`<img src="https://upload.wikimedia.org/wikipedia/commons/thumb/0/0c/American_Shorthair.jpg/900px-American_Shorthair.jpg">`;
+      const context = mockContext();
+      const actual = normalizePieceHtml(htmlStr, sourcePath, context);
+
+      assert.equal(actual, expected);
+      assert.deepEqual(context.files, []);
+    });
+
+    it("img[srcset] only adds local files", () => {
+      const htmlStr = html`
+        <img
+          srcset="small.jpg 480w, https://upload.wikimedia.org/900px-American_Shorthair.jpg 900w"
+          sizes="(max-width: 600px) 480px, 900px"
+          src="https://upload.wikimedia.org/900px-American_Shorthair.jpg"
+          alt="Cat image for testing"
+        />
+      `;
+      const expected = html`
+        <img srcset="./_file/small.jpg 480w, https://upload.wikimedia.org/900px-American_Shorthair.jpg 900w" sizes="(max-width: 600px) 480px, 900px" src="https://upload.wikimedia.org/900px-American_Shorthair.jpg" alt="Cat image for testing">
+      `;
+      const context = mockContext();
+      const actual = normalizePieceHtml(htmlStr, sourcePath, context);
+
+      assert.equal(actual, expected);
+      assert.deepEqual(context.files, [
+        {
+          mimeType: "image/jpeg",
+          name: "small.jpg",
+          path: "./_file/small.jpg"
+        }
+      ]);
+    });
+
+    it("video source[src] only adds local files", () => {
+      const htmlStr = html`<video width="320" height="240" controls>
+      <source src="https://www.youtube.com/watch?v=SsFyayu5csc" type="video/youtube"/>
+      <source src="observable.mov" type="video/mov">
+      Your browser doesn't support HTML video.
+      </video>`;
+
+      const expected = html`<video width="320" height="240" controls>
+      <source src="https://www.youtube.com/watch?v=SsFyayu5csc" type="video/youtube">
+      <source src="./_file/observable.mov" type="video/mov">
+      Your browser doesn't support HTML video.
+      </video>`;
+
+      const context = mockContext();
+      const actual = normalizePieceHtml(htmlStr, sourcePath, context);
+
+      assert.equal(actual, expected);
+      assert.deepEqual(context.files, [
+        {
+          mimeType: "video/quicktime",
+          name: "observable.mov",
+          path: "./_file/observable.mov"
+        }
+      ]);
+    });
+  });
+});