feat(reader): support v3 DjVu primary via image stream and text layer

validateReadingManifest accepts primary.kind=djvu (source-scoped objects/<sha>/<ns>/document.djvu in pdf-pages-v2); pdfPrimarySource returns '' for non-PDF primaries so PDF.js never loads DjVu; isBucketPath accepts document.djvu; assetFields rejects .djvu sidecar links.

static/reader/core/reader-contract.js

@@ -305,7 +305,7 @@
   function isBucketPath(path, versioned = false, bucket = "vomebook/pdf-pages-v2") {
     if (bucket === "vomebook/reader-assets-v2") return staticBucketPathPattern.test(path) || v2ChapterPathPattern.test(path) || v2ObjectPathPattern.test(path);
     if (bucket !== "vomebook/pdf-pages-v2") return false;
-    return /^objects\/[0-9a-f]{2}\/[0-9a-f]{64}\/[0-9a-f]{16}\/document\.pdf$/.test(path) ||
+    return /^objects\/[0-9a-f]{2}\/[0-9a-f]{64}\/[0-9a-f]{16}\/document\.(?:pdf|djvu)$/.test(path) ||
       (versioned ? versionedBucketPathPattern : bucketPathPattern).test(path) ||
       v3BucketPathPattern.test(path) ||
       staticBucketPathPattern.test(path) ||
@@ -403,6 +403,7 @@
     if (
       !asset ||
       asset.s !== 2 ||
+      path.endsWith(".djvu") ||
       !Object.prototype.hasOwnProperty.call(assetModes, asset.m) ||
        !(asset.b ? bucket : bucket || assetPrimaryPattern.test(path)) ||
         (asset?.b && bucketName !== "vomebook/pdf-pages-v2" && bucketName !== "vomebook/reader-assets-v2") ||

static/reader/pdf/reader-pdf-renderer.mjs

@@ -114,7 +114,9 @@ export function createReaderPdfRenderer(readerContext) {
 
   function pdfPrimarySource() {
     if (readerContext.capability.mode !== "pdf-pages") return "";
-    if (readerContext.pdfV3Manifest) return readerContext.v3ResourceUrl(readerContext.pdfV3Manifest.primary.resource);
+    // A non-PDF (DjVu) primary keeps the image stream plus v3 text layer.
+    if (readerContext.pdfV3Manifest)
+      return readerContext.pdfV3Manifest.primary.kind === "pdf" ? readerContext.v3ResourceUrl(readerContext.pdfV3Manifest.primary.resource) : "";
     const candidate = readerContext.normalizeSourceUrl(readerContext.params.get("pdf_document") || readerContext.resolvedReaderData?.pdf_document ||
       (readerContext.resolvedReaderData?.original_extension === "pdf" || /\.pdf(?:$|[?#])/i.test(readerContext.downloadUrl)
         ? readerContext.downloadUrl : ""));

static/reader/pdf/reader-v3.mjs

@@ -177,10 +177,15 @@ export function validateReadingManifest(manifest, maxPages) {
   if (!manifest || manifest.version !== 3 || manifest.kind !== "pdf-reading" ||
       !sha(manifest.source_sha256) || !sha(manifest.generation) || !integer(manifest.page_count) ||
       manifest.page_count > maxPages || typeof manifest.source_key !== "string" ||
-      manifest.primary?.kind !== "pdf" || manifest.primary.status !== "ready" ||
+      !["pdf", "djvu"].includes(manifest.primary?.kind) || manifest.primary.status !== "ready" ||
       typeof manifest.preview?.complete !== "boolean" || manifest.preview.page_count !== manifest.page_count) invalid();
   validateResource(manifest.primary.resource);
-  if (!manifest.primary.resource.path.endsWith(".pdf")) invalid();
+  // Browsers cannot render DjVu: such books read from the preview stream and text layer.
+  if (manifest.primary.kind === "djvu"
+    ? manifest.primary.resource.bucket !== "vomebook/pdf-pages-v2" ||
+      !manifest.primary.resource.path.startsWith(`objects/${manifest.source_sha256.slice(0, 2)}/${manifest.source_sha256}/`) ||
+      !manifest.primary.resource.path.endsWith("/document.djvu")
+    : !manifest.primary.resource.path.endsWith(".pdf")) invalid();
   validateResource(manifest.page_map, manifest.source_sha256);
   if (!manifest.page_map.path.endsWith("/page-map.json.gz")) invalid();
   partitions(manifest.preview.partitions, manifest.page_count, manifest.preview.complete, manifest.source_sha256);

tests/test_reader_pdf_scroll_scheduling.mjs

@@ -1,5 +1,7 @@
 import assert from "node:assert/strict";
+import fs from "node:fs";
 import path from "node:path";
+import vm from "node:vm";
 import { pathToFileURL, fileURLToPath } from "node:url";
 
 const root = path.resolve(process.argv[2] || path.join(path.dirname(fileURLToPath(import.meta.url)), ".."));
@@ -140,4 +142,53 @@ for (const outcome of ["promoted", "pdf", "cancelled"]) {
     }
   } finally {Object.defineProperty(globalThis,'performance',{configurable:true,value:nativePerformance});}
 }
+{
+  // DjVu primary: the manifest validates, its primary is never handed to PDF.js,
+  // and only the exact versioned document.djvu bucket path is accepted.
+  const sandbox = { self: {}, TextEncoder, URLSearchParams, URL };
+  vm.runInNewContext(fs.readFileSync(path.join(root, "static/reader/core/reader-contract.js"), "utf8"), sandbox);
+  const contract = globalThis.VoiceOfMLReader = sandbox.self.VoiceOfMLReader;
+  const { validateReadingManifest } = await load("pdf/reader-v3.mjs");
+  const source = "a".repeat(64), base = `objects/aa/${source}/`;
+  const ref = (p, bucket = "vomebook/pdf-pages-v2") => ({ bucket, path: base + p, sha256: "b".repeat(64), bytes: 10, role: "runtime" });
+  const manifest = (kind, primaryPath) => ({
+    version: 3, kind: "pdf-reading", source_sha256: source, source_key: "repo\0book.djvu", generation: "c".repeat(64),
+    page_count: 2, page_map: ref("d".repeat(16) + "/page-map.json.gz"),
+    primary: { kind, status: "ready", resource: ref(primaryPath) },
+    preview: { page_count: 2, complete: true, ready_pages: 2, partitions: [
+      { start: 1, end: 2, complete: true, resource: ref("d".repeat(16) + "/preview/partition-000001-000128-manifest.json") }] },
+    text_layer: ref("e".repeat(16) + "/text-layer-manifest.json"),
+    components: { document: "ready", preview: "ready", text: "ready", correction: "pending", searchable_pdf: "skipped" }
+  });
+  const djvuPath = "f".repeat(16) + "/document.djvu";
+  assert.equal(validateReadingManifest(manifest("djvu", djvuPath), 100).primary.kind, "djvu");
+  assert.equal(validateReadingManifest(manifest("pdf", "f".repeat(16) + "/document.pdf"), 100).primary.kind, "pdf");
+  for (const [kind, primaryPath] of [["pdf", djvuPath], ["djvu", "f".repeat(16) + "/document.pdf"], ["djvu", "document.djvu"],
+      ["djvu", "f".repeat(16) + "/book.djvu"], ["epub", djvuPath]])
+    assert.throws(() => validateReadingManifest(manifest(kind, primaryPath), 100), /PDF_V3_INVALID/, `${kind} ${primaryPath}`);
+  const foreign = manifest("djvu", djvuPath);
+  foreign.primary.resource.path = `objects/bb/${"b".repeat(64)}/${djvuPath}`;
+  assert.throws(() => validateReadingManifest(foreign, 100), /PDF_V3_INVALID/, "DjVu primary must belong to its source");
+  const otherBucket = manifest("djvu", djvuPath);
+  otherBucket.primary.resource.bucket = "vomebook/reader-assets-v2";
+  assert.throws(() => validateReadingManifest(otherBucket, 100), /PDF_V3_INVALID/);
+
+  assert.equal(contract.isBucketPath(base + djvuPath, true), true);
+  assert.equal(contract.isBucketPath(base + "f".repeat(16) + "/document.pdf", true), true);
+  for (const bad of [base + "document.djvu", base + "f".repeat(16) + "/x/document.djvu", base + "f".repeat(16) + "/book.djvu",
+      base + "f".repeat(16) + "/document.djvu.pdf", base + "f".repeat(16) + "/../document.djvu", `objects/aa/${source}.djvu`])
+    assert.equal(contract.isBucketPath(bad, true), false, bad);
+  assert.equal(contract.isBucketPath(base + djvuPath, true, "vomebook/reader-assets-v2"), false);
+  assert.equal(contract.isBucketPath(base + djvuPath, true, "vomebook/pdf-optimized"), false);
+  assert.equal(contract.assetFields({ s: 2, m: "p", p: base + djvuPath, b: "vomebook/pdf-pages-v2" }, "/api/reader-bucket-resource"), null,
+    "a DjVu original is never a sidecar PDF Reader link");
+
+  const reader = context(false);
+  Object.assign(reader, { capability: { mode: "pdf-pages" }, v3ResourceUrl: r => "/api/reader-bucket-resource?path=" + r.path });
+  reader.pdfV3Manifest = manifest("djvu", djvuPath);
+  assert.equal(reader.pdfPrimarySource(), "", "DjVu primary must not reach PDF.js");
+  reader.pdfV3Manifest = manifest("pdf", "f".repeat(16) + "/document.pdf");
+  assert.match(reader.pdfPrimarySource(), /document\.pdf$/);
+  delete globalThis.VoiceOfMLReader;
+}
 console.log("PDF scroll scheduling: bounded reads, live demand/fallback promotion, motion grace, cancellation and directional neighbors passed", root);