The best way to control Claude Code, Codex and more from any computer or phone
Mirrored from https://github.com/Camo67/t3code.git
Synced Nov 22, 17:00 UTC
feat(web): preview opendocument, epub, rich text, and notebooks (#21469)
maria committed
Oct 26, 2026 at 02:17 UTC
34ead566d44881a29e7438b949a421636742ab20
28 files changed
+3093
-92
apps/server/src/assets/AssetAccess.test.ts
+32
@@ -221,6 +221,12 @@ describe("AssetAccess", () => {
221
["recording.webm", "video/webm"],
222
["report.html", "text/html"],
223
["report.pdf", "application/pdf"],
224
+ ["notes.odt", "application/vnd.oasis.opendocument.text"],
225
+ ["budget.ods", "application/vnd.oasis.opendocument.spreadsheet"],
226
+ ["slides.odp", "application/vnd.oasis.opendocument.presentation"],
227
+ ["notes.rtf", "application/rtf"],
228
+ ["book.epub", "application/epub+zip"],
229
+ ["analysis.ipynb", "application/x-ipynb+json"],
230
["tone.wav", "audio/wav"],
231
["tone.webm-audio", "audio/webm"],
232
["archive.zip", "application/zip"],
@@ -944,6 +950,20 @@ describe("AssetAccess", () => {
950
expect(yield* resolveAsset(token, ".env")).toBeNull();
951
expect(yield* resolveAsset(`${token}tampered`, "report.html")).toBeNull();
952
for (const extension of [
953
+ "odt",
954
+ "ods",
955
+ "odp",
956
+ "rtf",
957
+ "epub",
958
+ "ipynb",
959
+ "zip",
960
+ "jar",
961
+ "tar",
962
+ "gz",
963
+ "tgz",
964
+ "sqlite",
965
+ "sqlite3",
966
+ "db",
967
"docx",
968
"xlsx",
969
"pptx",
@@ -1018,6 +1038,12 @@ describe("AssetAccess", () => {
1038
);
1039
for (const extension of [
1040
"pdf",
1041
+ "odt",
1042
+ "ods",
1043
+ "odp",
1044
+ "rtf",
1045
+ "epub",
1046
+ "ipynb",
1047
"png",
1048
"bmp",
1049
"heic",
@@ -1082,6 +1108,12 @@ describe("AssetAccess", () => {
1108
"report.docx",
1109
"workbook.xlsx",
1110
"slides.pptx",
1111
+ "notes.odt",
1112
+ "budget.ods",
1113
+ "slides.odp",
1114
+ "notes.rtf",
1115
+ "book.epub",
1116
+ "analysis.ipynb",
1117
"specimen.ttf",
1118
"specimen.otf",
1119
"specimen.woff",
apps/web/package.json
+2
@@ -61,6 +61,7 @@
61
"culori": "^4.0.2",
62
"dompurify": "^3.4.16",
63
"effect": "catalog:",
64
+ "entities": "8.1.0",
65
"fast-xml-parser": "5.7.3",
66
"hast-util-to-html": "^9.0.5",
67
"hast-util-to-jsx-runtime": "^2.3.6",
@@ -72,6 +73,7 @@
73
"lucide": "^0.564.0",
74
"lucide-react": "^0.564.0",
75
"morphicons": "^1.7.1",
76
+ "pdfjs-dist": "6.4.299",
77
"pele": "^1.8.0",
78
"react": "catalog:",
79
"react-dom": "catalog:",
apps/web/src/components/ComposerPromptEditorTiptap.tsx
+4
@@ -1,3 +1,4 @@
1
+import { documentIconColorForFileName } from "~/pierre-icons";
2
import { isProviderSkillUserInvocable } from "@t3tools/client-runtime/providerSkills";
3
import { formatProviderSkillDisplayName } from "@t3tools/shared/inlineSkills";
4
import {
@@ -45,6 +46,7 @@ import {
46
useLayoutEffect,
47
useRef,
48
useState,
49
+ type CSSProperties,
50
type KeyboardEvent as ReactKeyboardEvent,
51
type RefObject,
52
} from "react";
@@ -260,9 +262,11 @@ const ComposerMentionExtension = Node.create({
262
function ComposerMentionNodeView({ node }: NodeViewProps) {
263
const actions = use(ComposerContextActionsContext);
264
const path = (node.attrs.path as string) ?? "";
265
+ const accent = documentIconColorForFileName(path);
266
const chip = (
267
<ContextChip
268
kind="mention"
269
+ style={accent ? ({ "--context-chip-accent": accent } as CSSProperties) : undefined}
270
dir="ltr"
271
render={<button type="button" />}
272
onClick={() => actions.openMention(path)}
apps/web/src/components/chat/MarkdownFileLink.tsx
+6
-1
@@ -5,12 +5,13 @@ import {
5
type AtomCommandResult,
6
} from "@t3tools/client-runtime/state/runtime";
7
import type { EnvironmentId, ScopedThreadRef } from "@t3tools/contracts";
8
-import { memo, type MouseEvent as ReactMouseEvent } from "react";
8
+import { memo, type CSSProperties, type MouseEvent as ReactMouseEvent } from "react";
9
10
import { readLocalApi } from "../../localApi";
11
import { cn } from "../../lib/utils";
12
import { shouldOpenMarkdownFileLinkInEditor } from "../../markdown-links";
13
import type { markdownFileCopyPaths } from "../../markdownFileHostPath";
14
+import { documentIconColorForFileName } from "../../pierre-icons";
15
import { ContextChip } from "../ContextChip";
16
import { stackedThreadToast, toastManager } from "../ui/toast";
17
import { Tooltip, TooltipPopup, TooltipTrigger } from "../ui/tooltip";
@@ -320,6 +321,8 @@ export const MarkdownFileLink = memo(function MarkdownFileLink({
321
canOpenInBrowser,
322
canOpenInPanel,
323
});
324
+ const accent = documentIconColorForFileName(iconPath);
325
+ const chipStyle = accent ? ({ "--context-chip-accent": accent } as CSSProperties) : undefined;
326
327
return (
328
<Tooltip>
@@ -331,6 +334,7 @@ export const MarkdownFileLink = memo(function MarkdownFileLink({
334
dir="ltr"
335
render={<a href={href} />}
336
className={MARKDOWN_FILE_LINK_CLASS_NAME}
337
+ style={chipStyle}
338
data-markdown-copy={copyMarkdown}
339
onClick={(event) => {
340
event.preventDefault();
@@ -362,6 +366,7 @@ export const MarkdownFileLink = memo(function MarkdownFileLink({
366
aria-label={`File options for ${label}`}
367
aria-haspopup="menu"
368
className={cn(MARKDOWN_FILE_LINK_CLASS_NAME, "select-text")}
369
+ style={chipStyle}
370
data-markdown-copy={copyMarkdown}
371
onClick={handleContextMenu}
372
onContextMenu={handleContextMenu}
apps/web/src/components/chat/UserMessageContextChips.tsx
+4
-1
@@ -1,3 +1,4 @@
1
+import { documentIconColorForFileName } from "~/pierre-icons";
2
import { isSvgAttachment } from "@t3tools/shared/image";
3
import { useRenderedAppearance } from "../../hooks/useTheme";
4
import { PullRequestGlyph } from "../pullRequest/pullRequestIcons";
@@ -5,7 +6,7 @@ import { ReadOnlySourcePreview } from "../files/AttachmentFilePreview";
6
import { useRightPanelStore } from "~/rightPanelStore";
7
import { COMPOSER_CONTEXT_KINDS } from "@t3tools/contracts";
8
import { formatAttachmentSize } from "@t3tools/client-runtime/state/attachments";
8
-import { use, type ReactNode } from "react";
9
+import { use, type CSSProperties, type ReactNode } from "react";
10
import { FileDiff } from "@pierre/diffs/react";
11
import { DiffWorkerPoolProvider } from "../DiffWorkerPoolProvider";
12
import {
@@ -61,12 +62,14 @@ function UserMessageMentionChip(props: {
62
copyMarkdown: string;
63
}) {
64
const ctx = use(TimelineRowCtx);
65
+ const accent = documentIconColorForFileName(props.record.path);
66
return (
67
<Tooltip>
68
<TooltipTrigger
69
render={
70
<ContextChip
71
kind="mention"
72
+ style={accent ? ({ "--context-chip-accent": accent } as CSSProperties) : undefined}
73
render={<button type="button" />}
74
aria-label={`Preview ${props.record.path}`}
75
data-markdown-copy={props.copyMarkdown}
apps/web/src/components/contextChipParts.tsx
+3
@@ -9,6 +9,7 @@ import {
9
type ReactNode,
10
} from "react";
11
import { useAssetUrlState } from "~/assets/assetUrls";
12
+import { documentIconColorForFileName } from "~/pierre-icons";
13
14
import { PULL_REQUEST_STATE_PRESENTATION } from "~/components/pullRequest/pullRequestIcons";
15
import type { PullRequestContextDisplayState } from "~/lib/composerContextRecords";
@@ -242,6 +243,7 @@ export function FileChip(props: {
243
}) {
244
if (props.svgPreview && !props.unresolved) return <SvgFileChip {...props} />;
245
const content = <FileChipContent {...props} />;
246
+ const accent = props.isVideo ? undefined : documentIconColorForFileName(props.name);
247
const state = props.unresolved
248
? ("unresolved" as const)
249
: props.error
@@ -249,6 +251,7 @@ export function FileChip(props: {
251
: null;
252
const attributes = {
253
kind: props.isVideo ? "video" : "file",
254
+ style: accent ? ({ "--context-chip-accent": accent } as CSSProperties) : undefined,
255
...(state ? { state } : {}),
256
"aria-label": [props.accessibleLabel, props.suffix].filter(Boolean).join(", "),
257
"data-markdown-copy": props.copyMarkdown,
apps/web/src/components/files/AttachmentFilePreview.browser.test.tsx
+464
@@ -1,3 +1,4 @@
1
+import DocumentPages, { sanitizeDocumentHtml } from "./DocumentPages";
2
import { act, render, stubGlobal, type Screen } from "../../../test/render";
3
import { EnvironmentId, ThreadId } from "@t3tools/contracts";
4
import { describe, expect, it, vi } from "vite-plus/test";
@@ -2057,6 +2058,40 @@ describe("attachment HTML preview recovery", () => {
2058
});
2059
});
2060
2061
+describe("document output sanitization", () => {
2062
+ it("fits saved HTML outputs to their content and resizes when the content changes", async () => {
2063
+ await using screen = await render(
2064
+ <DocumentPages html="<table><tr><td>42</td></tr></table>" name="Output" output />,
2065
+ );
2066
+ const frame = screen.container.querySelector("iframe")!;
2067
+ await vi.waitFor(() => {
2068
+ expect(frame.contentDocument?.querySelector("td")?.textContent).toBe("42");
2069
+ expect(frame.clientHeight).toBeGreaterThan(32);
2070
+ expect(frame.clientHeight).toBeLessThan(120);
2071
+ });
2072
+ const initialHeight = frame.clientHeight;
2073
+ await screen.rerender(
2074
+ <DocumentPages html={"<p>More saved output</p>".repeat(10)} name="Output" output />,
2075
+ );
2076
+ await vi.waitFor(() => {
2077
+ expect(frame.contentDocument?.querySelectorAll("p")).toHaveLength(10);
2078
+ expect(frame.clientHeight).toBeGreaterThan(initialHeight);
2079
+ expect(frame.contentDocument!.body.scrollHeight).toBeLessThanOrEqual(frame.clientHeight);
2080
+ });
2081
+ });
2082
+
2083
+ it("preserves table content and raster data while removing scripts and network resources", () => {
2084
+ const html = sanitizeDocumentHtml(
2085
+ '<script>alert(1)</script><style>@import "https://example.com/style.css";</style><iframe src="https://example.com"></iframe><img src="https://example.com/tracker.png" onerror="alert(1)"><img src="data:image/png;base64,iVBORw0KGgo="><table><tr><td>Research</td></tr></table><a href="https://example.com">Results</a>',
2086
+ );
2087
+ const doc = new DOMParser().parseFromString(html, "text/html");
2088
+ expect(doc.querySelector("td")?.textContent).toBe("Research");
2089
+ expect(doc.images).toHaveLength(1);
2090
+ expect(doc.images[0]!.getAttribute("src")).toBe("data:image/png;base64,iVBORw0KGgo=");
2091
+ expect(doc.querySelector("script,style,iframe,a,[onerror]")).toBeNull();
2092
+ expect(html).not.toContain("https://");
2093
+ });
2094
+});
2095
it("opens a known empty attachment even when its range response is inaccessible", async () => {
2096
refresh.mockResolvedValue("https://unreachable.invalid/empty.txt");
2097
await using screen = await render(
@@ -2295,3 +2330,432 @@ it("toggles a Mermaid attachment between diagram and source through the lazy pre
2330
);
2331
expect(screen.container.querySelector('[role="alert"]')).toBeNull();
2332
});
2333
+
2334
+it("navigates, searches, zooms, and selects mixed-height PDF pages", async () => {
2335
+ const objects = [
2336
+ "<< /Type /Catalog /Pages 2 0 R >>",
2337
+ "<< /Type /Pages /Count 3 /Kids [3 0 R 5 0 R 7 0 R] >>",
2338
+ ];
2339
+ for (const [index, title] of [
2340
+ "Quarterly outlook",
2341
+ "Delivery milestones",
2342
+ "Budget and risks",
2343
+ ].entries()) {
2344
+ const text = `/GS1 gs BT /F1 24 Tf 30 ${index === 0 ? 70 : 700} Td ${index === 0 ? "(Quarter) Tj /F2 24 Tf (ly outlook) Tj" : `(${title}) Tj`} ET`;
2345
+ objects.push(
2346
+ `<< /Type /Page /Parent 2 0 R /MediaBox [0 0 612 ${index === 0 ? 140 : 792}] /Rotate ${index === 2 ? 90 : 0} /Resources << /Font << /F1 9 0 R /F2 10 0 R >> /ExtGState << /GS1 << /TR << /FunctionType 2 /Domain [0 1] /C0 [0] /C1 [1] /N 2 >> >> >> >> /Contents ${4 + index * 2} 0 R >>`,
2347
+ `<< /Length ${text.length} >>\nstream\n${text}\nendstream`,
2348
+ );
2349
+ }
2350
+ objects.push("<< /Type /Font /Subtype /Type1 /BaseFont /Helvetica >>");
2351
+ objects.push("<< /Type /Font /Subtype /Type1 /BaseFont /Helvetica-Bold >>");
2352
+ let source = "%PDF-1.4\n";
2353
+ const offsets = [0];
2354
+ for (const [index, object] of objects.entries()) {
2355
+ offsets.push(source.length);
2356
+ source += `${index + 1} 0 obj\n${object}\nendobj\n`;
2357
+ }
2358
+ const xref = source.length;
2359
+ source += `xref\n0 11\n0000000000 65535 f \n${offsets
2360
+ .slice(1)
2361
+ .map((offset) => `${String(offset).padStart(10, "0")} 00000 n \n`)
2362
+ .join("")}trailer\n<< /Size 11 /Root 1 0 R >>\nstartxref\n${xref}\n%%EOF`;
2363
+ const file = new Blob([source], { type: "application/pdf" });
2364
+ const { default: PdfViewer } = await import("./PdfViewer");
2365
+ const url = URL.createObjectURL(file);
2366
+ try {
2367
+ await using screen = await render(
2368
+ <div style={{ display: "flex", height: 600, width: 320 }}>
2369
+ <PdfViewer src={url} title="Planning PDF" />
2370
+ </div>,
2371
+ );
2372
+ await vi.waitFor(() => expect(screen.container.textContent).toContain("1 / 3 pages"), {
2373
+ timeout: 15_000,
2374
+ });
2375
+ const region = screen.container.querySelector<HTMLElement>('[role="region"]')!;
2376
+ await vi.waitFor(() => {
2377
+ expect(region.scrollWidth).toBe(region.clientWidth);
2378
+ expect(region.querySelector("canvas")!.getBoundingClientRect().width).toBeLessThan(
2379
+ region.clientWidth,
2380
+ );
2381
+ expect(screen.container.textContent).toContain("Fit width");
2382
+ });
2383
+ await act(() => {
2384
+ (screen.container.firstElementChild as HTMLElement).style.width = "500px";
2385
+ });
2386
+ await vi.waitFor(() =>
2387
+ expect(region.querySelector("canvas")!.getBoundingClientRect().width).toBe(468),
2388
+ );
2389
+ await screen.getByRole("button", { name: "Next page", exact: true }).click();
2390
+ await vi.waitFor(() => expect(screen.container.textContent).toContain("2 / 3 pages"));
2391
+ await screen.getByRole("button", { name: "Previous page", exact: true }).click();
2392
+ await vi.waitFor(() => expect(screen.container.textContent).toContain("1 / 3 pages"));
2393
+ await screen.getByRole("button", { name: "Zoom in", exact: true }).click();
2394
+ await vi.waitFor(() => expect(region.scrollWidth).toBeGreaterThan(region.clientWidth));
2395
+ await screen.getByRole("button", { name: "Fit width", exact: true }).click();
2396
+ await vi.waitFor(() => expect(region.scrollWidth).toBe(region.clientWidth));
2397
+ await screen.getByRole("searchbox", { name: "Find in PDF" }).fill("Quarterly outlook");
2398
+ await vi.waitFor(() => {
2399
+ expect(screen.container.querySelector('[role="status"]')?.textContent).toBe("1/1");
2400
+ expect(
2401
+ Array.from(region.querySelectorAll(".pdf-find-match")).map((span) => span.textContent),
2402
+ ).toEqual(["Quarter", "ly outlook"]);
2403
+ });
2404
+ await screen.getByRole("searchbox", { name: "Find in PDF" }).fill("Budget");
2405
+ await vi.waitFor(() => expect(screen.container.textContent).toContain("3 / 3 pages"));
2406
+ await screen.getByRole("button", { name: "Zoom in", exact: true }).click();
2407
+ await vi.waitFor(() => expect(screen.container.textContent).toContain("3 / 3 pages"));
2408
+ await vi.waitFor(() => {
2409
+ const span = Array.from(screen.container.querySelectorAll("[data-pdf-text-layer] span")).find(
2410
+ (node) => node.textContent === "Budget and risks",
2411
+ );
2412
+ expect(span).toBeDefined();
2413
+ const textBounds = span!.getBoundingClientRect();
2414
+ const pageBounds = span!.parentElement!.getBoundingClientRect();
2415
+ expect(textBounds.width).toBeGreaterThan(0);
2416
+ expect(textBounds.height).toBeGreaterThan(0);
2417
+ expect(textBounds.left).toBeGreaterThanOrEqual(pageBounds.left - 1);
2418
+ expect(textBounds.top).toBeGreaterThanOrEqual(pageBounds.top - 1);
2419
+ expect(textBounds.right).toBeLessThanOrEqual(pageBounds.right + 1);
2420
+ expect(textBounds.bottom).toBeLessThanOrEqual(pageBounds.bottom + 1);
2421
+ const range = document.createRange();
2422
+ range.selectNodeContents(span!);
2423
+ const selection = window.getSelection()!;
2424
+ selection.removeAllRanges();
2425
+ selection.addRange(range);
2426
+ expect(selection.toString()).toBe("Budget and risks");
2427
+ expect(getComputedStyle(span!).userSelect).toBe("text");
2428
+ selection.removeAllRanges();
2429
+ });
2430
+ await screen.getByRole("button", { name: "Fit width", exact: true }).click();
2431
+ await vi.waitFor(() => expect(screen.container.textContent).toContain("3 / 3 pages"));
2432
+ await screen.getByRole("searchbox", { name: "Find in PDF" }).fill(" ");
2433
+ await vi.waitFor(() =>
2434
+ expect(screen.container.querySelector('[role="status"]')?.textContent).toBe("0/0"),
2435
+ );
2436
+ } finally {
2437
+ URL.revokeObjectURL(url);
2438
+ }
2439
+});
2440
+it("mounts a bounded set of saved notebook outputs as the reader scrolls", async () => {
2441
+ const notebook = {
2442
+ nbformat: 4,
2443
+ metadata: {},
2444
+ cells: Array.from({ length: 20 }, (_, cell) => ({
2445
+ cell_type: "code",
2446
+ source: [],
2447
+ metadata: {},
2448
+ outputs: Array.from({ length: 100 }, (_, output) => ({
2449
+ output_type: "display_data",
2450
+ data: { "text/html": `<p>saved ${cell}-${output}</p>` },
2451
+ })),
2452
+ })),
2453
+ };
2454
+ const file = new Blob([JSON.stringify(notebook)]);
2455
+ await using screen = await render(
2456
+ <div style={{ display: "flex", height: 400, width: 600 }}>
2457
+ <AttachmentFilePreview
2458
+ name="observations.ipynb"
2459
+ mimeType="application/json"
2460
+ sizeBytes={file.size}
2461
+ file={file}
2462
+ />
2463
+ </div>,
2464
+ );
2465
+ await vi.waitFor(() =>
2466
+ expect(screen.container.querySelectorAll("iframe").length).toBeGreaterThan(0),
2467
+ );
2468
+ expect(screen.container.querySelectorAll("iframe").length).toBeLessThanOrEqual(16);
2469
+ const first = screen.container.querySelector("iframe")!;
2470
+ expect(first.srcdoc).toContain("saved 0-0");
2471
+ const viewport = screen.container.querySelector<HTMLElement>(".scrollbar-gutter-both")!;
2472
+ await act(() => {
2473
+ viewport.scrollTop = viewport.scrollHeight;
2474
+ });
2475
+ await vi.waitFor(() => {
2476
+ expect(
2477
+ [...screen.container.querySelectorAll("iframe")].some((frame) =>
2478
+ frame.srcdoc.includes("saved 19-99"),
2479
+ ),
2480
+ ).toBe(true);
2481
+ expect(first.isConnected).toBe(false);
2482
+ });
2483
+ expect(screen.container.querySelectorAll("iframe").length).toBeLessThanOrEqual(16);
2484
+ await act(() => {
2485
+ viewport.scrollTop = 0;
2486
+ });
2487
+ await vi.waitFor(() =>
2488
+ expect(screen.container.querySelector("iframe")?.srcdoc).toContain("saved 0-0"),
2489
+ );
2490
+});
2491
+
2492
+it.each([
2493
+ [String.raw`{\rtf1\uc1\u945{\b bold}Hello}`, "αboldHello"],
2494
+ [String.raw`{\rtf1\uc1{\u945}Hello}`, "αHello"],
2495
+ [
2496
+ String.raw`{\rtf1\ansi{\fonttbl{\f0\fcharset204 Cyrillic;}{\f1\fcharset0 Latin;}}\f0\'c0{\f1\'c0}\'c1}`,
2497
+ "АÀБ",
2498
+ ],
2499
+ [String.raw`{\rtf1\ansi\deff1{\fonttbl{\f1\fcharset204 Cyrillic;}}\'c0\plain\'c1}`, "АБ"],
2500
+])("retains RTF grouped Unicode and font text in %s", async (source, expected) => {
2501
+ const { openRtf } = await import("./extendedDocumentPreview");
2502
+ const result = openRtf(new TextEncoder().encode(source).buffer);
2503
+ if (result.type !== "document") throw new Error("Expected rich text document");
2504
+ const html = result.pages.map((page) => page.html).join("");
2505
+ expect(new DOMParser().parseFromString(html, "text/html").body.textContent).toBe(expected);
2506
+});
2507
+
2508
+it.each(["odt", "odp"] as const)("preserves rendered %s spaces and tabs", async (kind) => {
2509
+ const archive = new JSZip().file(
2510
+ "content.xml",
2511
+ `<office:document>${kind === "odt" ? "<office:text>" : "<draw:page>"}<text:p>A<text:s text:c="3"/>B<text:tab/>C</text:p>${kind === "odt" ? "</office:text>" : "</draw:page>"}</office:document>`,
2512
+ );
2513
+ const bytes = await archive.generateAsync({ type: "arraybuffer" });
2514
+ const worker = new Worker(new URL("./officePreview.worker.ts", import.meta.url), {
2515
+ type: "module",
2516
+ });
2517
+ try {
2518
+ const result = await new Promise<OfficeWorkerResult>((resolve) => {
2519
+ worker.addEventListener(
2520
+ "message",
2521
+ (event: MessageEvent<OfficeWorkerResult>) => resolve(event.data),
2522
+ { once: true },
2523
+ );
2524
+ worker.postMessage({ type: "open", kind, bytes }, [bytes]);
2525
+ });
2526
+ if (result.type !== "document") throw new Error("Expected document");
2527
+ await using screen = await render(
2528
+ <DocumentPages html={result.pages[0]!.html} name="spacing" output />,
2529
+ );
2530
+ await vi.waitFor(() => {
2531
+ const doc = screen.container.querySelector("iframe")?.contentDocument;
2532
+ const spans = doc?.querySelectorAll<HTMLElement>("span[data-odf-whitespace]");
2533
+ expect(doc?.querySelector("p")?.textContent).toBe("A B\tC");
2534
+ expect(spans).toHaveLength(2);
2535
+ for (const span of spans!) {
2536
+ expect(doc!.defaultView!.getComputedStyle(span).whiteSpace).toBe("pre-wrap");
2537
+ expect(span.getBoundingClientRect().width).toBeGreaterThan(0);
2538
+ }
2539
+ expect(spans![0]!.getBoundingClientRect().width).toBeGreaterThan(10);
2540
+ });
2541
+ } finally {
2542
+ worker.terminate();
2543
+ }
2544
+});
2545
+
2546
+it.each(["rtf", "epub", "ods"] as const)(
2547
+ "preserves %s document text through the worker and attachment viewer",
2548
+ async (kind) => {
2549
+ const archive = new JSZip();
2550
+ if (kind === "epub") {
2551
+ archive.file(
2552
+ "META-INF/container.xml",
2553
+ '<container><rootfiles><rootfile full-path="book.opf"/></rootfiles></container>',
2554
+ );
2555
+ archive.file(
2556
+ "book.opf",
2557
+ '<package><manifest><item id="chapter" href="chapter.xhtml" media-type="application/xhtml+xml"/></manifest><spine><itemref idref="chapter"/></spine></package>',
2558
+ );
2559
+ archive.file(
2560
+ "chapter.xhtml",
2561
+ "<html><head><title>Handbook</title></head><body><p>first<br/>second<br/>third</p><hr/></body></html>",
2562
+ );
2563
+ } else if (kind === "ods") {
2564
+ archive.file(
2565
+ "content.xml",
2566
+ '<office:document><table:table table:name="Forecast"><table:table-row><table:table-cell><text:p>North<office:annotation><text:p>nested note</text:p></office:annotation></text:p><office:annotation><text:p>private note</text:p></office:annotation></table:table-cell><table:table-cell office:value="150"><office:annotation><text:p>revised from 120</text:p></office:annotation></table:table-cell><table:table-cell table:number-columns-repeated="1024"/></table:table-row><table:table-row table:number-rows-repeated="1048570"><table:table-cell table:number-columns-repeated="1024"/></table:table-row></table:table></office:document>',
2567
+ );
2568
+ }
2569
+ const file =
2570
+ kind === "rtf"
2571
+ ? new Blob([
2572
+ String.raw`{\rtf1\ansi 10\endash 20 \emdash \bullet \lquote a\rquote \ldblquote b\rdblquote\par Name\cell Value\cell\row 12\cell 34\cell\row{\*\private hidden\endash\cell\row}Unicode:\u8211\endash end}`,
2573
+ ])
2574
+ : await archive.generateAsync({ type: "blob" });
2575
+ const bytes = await file.arrayBuffer();
2576
+ const worker = new Worker(new URL("./officePreview.worker.ts", import.meta.url), {
2577
+ type: "module",
2578
+ });
2579
+ try {
2580
+ const result = await new Promise<OfficeWorkerResult>((resolve) => {
2581
+ worker.addEventListener(
2582
+ "message",
2583
+ (event: MessageEvent<OfficeWorkerResult>) => resolve(event.data),
2584
+ { once: true },
2585
+ );
2586
+ worker.postMessage({ type: "open", kind, bytes }, [bytes]);
2587
+ });
2588
+ if (kind === "ods") {
2589
+ expect(result.type).toBe("ods");
2590
+ if (result.type !== "ods") throw new Error("Expected spreadsheet");
2591
+ expect(result.sheets[0]!.rows).toEqual([["North", "150"]]);
2592
+ } else {
2593
+ expect(result.type).toBe("document");
2594
+ if (result.type !== "document") throw new Error("Expected document");
2595
+ const body = new DOMParser().parseFromString(result.pages[0]!.html, "text/html").body;
2596
+ if (kind === "epub") {
2597
+ expect(body.querySelectorAll("br")).toHaveLength(2);
2598
+ expect(body.querySelectorAll("hr")).toHaveLength(1);
2599
+ } else {
2600
+ expect(body.textContent).toContain("10–20 —•‘a’“b”");
2601
+ expect(body.textContent).toContain("Name\tValue\t12\t34\t");
2602
+ expect(body.textContent).toContain("Unicode:–end");
2603
+ expect(body.textContent).not.toContain("hidden");
2604
+ expect(body.querySelectorAll("br")).toHaveLength(3);
2605
+ }
2606
+ }
2607
+ } finally {
2608
+ worker.terminate();
2609
+ }
2610
+ await using screen = await render(
2611
+ <div style={{ display: "flex", height: 400, width: 600 }}>
2612
+ <AttachmentFilePreview
2613
+ name={`report.${kind}`}
2614
+ mimeType="application/octet-stream"
2615
+ sizeBytes={file.size}
2616
+ file={file}
2617
+ />
2618
+ </div>,
2619
+ );
2620
+ await vi.waitFor(() => {
2621
+ if (kind === "ods") {
2622
+ expect(screen.container.textContent).toContain("North");
2623
+ expect(screen.container.textContent).toContain("150");
2624
+ expect(screen.container.textContent).not.toContain("private note");
2625
+ expect(screen.container.textContent).toContain("1 rows · 2 columns");
2626
+ expect(screen.container.textContent).not.toContain("Limited to 500");
2627
+ } else {
2628
+ const frame = screen.container.querySelector("iframe");
2629
+ expect(frame?.srcdoc).toContain(kind === "rtf" ? "10–20" : "first<br>second<br>third");
2630
+ }
2631
+ });
2632
+ expect(screen.container.querySelector('[role="alert"]')).toBeNull();
2633
+ },
2634
+);
2635
+
2636
+it.each(["epub", "odt", "ipynb", "html", "encoded-html"] as const)(
2637
+ "bounds %s image pixels before returning browser markup",
2638
+ async (kind) => {
2639
+ const canvas = new OffscreenCanvas(2, 3);
2640
+ canvas.getContext("2d");
2641
+ const blob = await canvas.convertToBlob({ type: "image/png" });
2642
+ for (const width of [2, 12000]) {
2643
+ const image = await blob.arrayBuffer();
2644
+ new DataView(image).setUint32(16, width);
2645
+ new DataView(image).setUint32(20, width === 2 ? 3 : 12000);
2646
+ const archive = new JSZip();
2647
+ archive.file("cover.png", image);
2648
+ if (kind === "epub") {
2649
+ archive.file(
2650
+ "META-INF/container.xml",
2651
+ '<container><rootfiles><rootfile full-path="book.opf"/></rootfiles></container>',
2652
+ );
2653
+ archive.file(
2654
+ "book.opf",
2655
+ '<package><manifest><item id="chapter" href="chapter.xhtml" media-type="application/xhtml+xml"/></manifest><spine><itemref idref="chapter"/></spine></package>',
2656
+ );
2657
+ archive.file(
2658
+ "chapter.xhtml",
2659
+ '<html><body><p>retained caption</p><img src="cover.png"/></body></html>',
2660
+ );
2661
+ } else
2662
+ archive.file(
2663
+ "content.xml",
2664
+ '<office:document><office:text><text:p>retained caption</text:p><draw:image xlink:href="cover.png"/></office:text></office:document>',
2665
+ );
2666
+ const base64 = btoa(String.fromCharCode(...new Uint8Array(image)));
2667
+ const bytes =
2668
+ kind === "ipynb" || kind === "html" || kind === "encoded-html"
2669
+ ? new TextEncoder().encode(
2670
+ JSON.stringify({
2671
+ nbformat: 4,
2672
+ cells: [
2673
+ {
2674
+ cell_type: "code",
2675
+ source: "plot()",
2676
+ outputs: [
2677
+ {
2678
+ data: {
2679
+ ...(kind === "html" || kind === "encoded-html"
2680
+ ? {
2681
+ "text/html": `<table><tr><td>retained caption</td></tr></table><img src="${kind === "encoded-html" ? "data" : "data"}:image/png;base64,${base64}">`,
2682
+ }
2683
+ : { "image/png": base64 }),
2684
+ "text/plain": "retained caption",
2685
+ },
2686
+ },
2687
+ ],
2688
+ },
2689
+ ],
2690
+ }),
2691
+ ).buffer
2692
+ : await archive.generateAsync({ type: "arraybuffer" });
2693
+ const worker = new Worker(new URL("./officePreview.worker.ts", import.meta.url), {
2694
+ type: "module",
2695
+ });
2696
+ try {
2697
+ const result = await new Promise<OfficeWorkerResult>((resolve) => {
2698
+ worker.addEventListener(
2699
+ "message",
2700
+ (event: MessageEvent<OfficeWorkerResult>) => resolve(event.data),
2701
+ { once: true },
2702
+ );
2703
+ worker.postMessage(
2704
+ {
2705
+ type: "open",
2706
+ kind: kind === "html" || kind === "encoded-html" ? "ipynb" : kind,
2707
+ bytes,
2708
+ },
2709
+ [bytes],
2710
+ );
2711
+ });
2712
+ if (result.type === "notebook") {
2713
+ const output = result.cells[0]!.outputs[0]!;
2714
+ expect(output.kind).toBe(
2715
+ width === 2 ? (kind === "html" || kind === "encoded-html" ? "html" : "image") : "text",
2716
+ );
2717
+ if (output.kind === "html") {
2718
+ const rendered = new DOMParser().parseFromString(
2719
+ sanitizeDocumentHtml(output.text),
2720
+ "text/html",
2721
+ );
2722
+ expect(rendered.images).toHaveLength(1);
2723
+ expect(rendered.querySelector("td")!.textContent).toBe("retained caption");
2724
+ }
2725
+ if (width !== 2) expect(result.cells[0]!.outputs[0]!.text).toBe("retained caption");
2726
+ } else {
2727
+ expect(result.type).toBe("document");
2728
+ if (result.type !== "document") throw new Error("Expected document");
2729
+ expect(result.pages[0]!.html.includes("<img")).toBe(width === 2);
2730
+ expect(result.pages[0]!.html).toContain("retained caption");
2731
+ }
2732
+ } finally {
2733
+ worker.terminate();
2734
+ }
2735
+ }
2736
+ },
2737
+);
2738
+
2739
+it.each([
2740
+ ["standardFontDataUrl", "FoxitSerif.pfb", "standard_fonts"],
2741
+ ["cMapUrl", "78-EUC-H.bcmap", "cmaps"],
2742
+ ["wasmUrl", "openjpeg.wasm", "wasm"],
2743
+])("validates PDF %s assets before decoding", async (kind, filename, directory) => {
2744
+ const { BinaryDataFactory } = await import("./pdfDocument");
2745
+ const factory = new BinaryDataFactory();
2746
+ const fetchAsset = vi.fn(
2747
+ async () =>
2748
+ new Response("<html>app</html>", {
2749
+ headers: { "content-type": "text/html; charset=utf-8" },
2750
+ }),
2751
+ );
2752
+ using _fetch = stubGlobal("fetch", fetchAsset);
2753
+ await expect(factory.fetch({ kind, filename })).rejects.toThrow(
2754
+ `PDF ${directory} asset unavailable.`,
2755
+ );
2756
+ fetchAsset.mockImplementation(async () => new Response(new Uint8Array([1, 2, 3])));
2757
+ await expect(factory.fetch({ kind, filename })).resolves.toEqual(new Uint8Array([1, 2, 3]));
2758
+ await expect(factory.fetch({ kind, filename: "missing" })).rejects.toThrow(
2759
+ `Unsupported PDF ${directory} asset.`,
2760
+ );
2761
+});
apps/web/src/components/files/BrowserDocumentFrame.tsx
+38
-9
@@ -22,7 +22,17 @@ import {
22
matchesKeybindingShortcut,
23
normalizeEventKey,
24
} from "@t3tools/shared/keybindings";
25
-import { useEffect, useEffectEvent, useLayoutEffect, useRef, useState } from "react";
25
+import {
26
+ lazy,
27
+ Suspense,
28
+ useEffect,
29
+ useEffectEvent,
30
+ useLayoutEffect,
31
+ useRef,
32
+ useState,
33
+} from "react";
34
+import { isElectron } from "~/env";
35
+import { FileSurfaceLoading } from "./fileSurfaceChrome";
36
37
import { useOpenLink } from "~/browser/useOpenLink";
38
import { ExpandedImageDialog } from "~/components/chat/ExpandedImageDialog";
@@ -61,9 +71,34 @@ const HTML_RENDER_SHORTCUT_COMMANDS: ReadonlySet<KeybindingCommand> = new Set([
71
export const isPdfPreviewFile = (path: string): boolean =>
72
/\.pdf$/i.test(path.split(/[?#]/, 1)[0] ?? "");
73
74
+const PdfViewer = lazy(() => import("./PdfViewer"));
75
+
76
+function PdfDocument(props: { src: string; title: string }) {
77
+ const [src] = useState(props.src);
78
+ // The Electron shell does not enable PDF plugins. WebKit embeds on iOS
79
+ // expose only the first page, even when pdfViewerEnabled reports support.
80
+ const native =
81
+ !isElectron &&
82
+ navigator.pdfViewerEnabled &&
83
+ !/iPad|iPhone|iPod/.test(navigator.userAgent) &&
84
+ !(navigator.platform === "MacIntel" && navigator.maxTouchPoints > 1);
85
+ return native ? (
86
+ // oxlint-disable-next-line react/iframe-missing-sandbox -- Chromium's PDF plugin requires an unsandboxed frame; src is an exact-file asset grant.
87
+ <iframe
88
+ src={`${src.split("#", 1)[0]}#navpanes=0&view=FitH`}
89
+ title={props.title}
90
+ className="block h-full min-h-0 w-full flex-1 border-0 bg-white"
91
+ />
92
+ ) : (
93
+ <Suspense fallback={<FileSurfaceLoading />}>
94
+ <PdfViewer src={src} title={props.title} />
95
+ </Suspense>
96
+ );
97
+}
98
+
99
/**
100
* HTML runs in a sandboxed frame with an opaque origin, so a page cannot reach
66
- * the app's session or storage. PDFs use the browser's native viewer.
101
+ * the app's session or storage. PDFs use a native viewer or the lazy PDF.js fallback.
102
*/
103
export function BrowserDocumentFrame(props: {
104
readonly src: string;
@@ -76,13 +111,7 @@ export function BrowserDocumentFrame(props: {
111
}) {
112
const key = props.documentKey ?? props.src;
113
return props.pdf ? (
79
- <embed
80
- key={props.src}
81
- src={props.src}
82
- type="application/pdf"
83
- title={props.title}
84
- className="min-h-0 w-full flex-1 border-0"
85
- />
114
+ <PdfDocument key={key} src={props.src} title={props.title} />
115
) : props.htmlRender ? (
116
<HtmlRenderDocument
117
key={key}
apps/web/src/components/files/DocumentPages.tsx
new
+145
@@ -0,0 +1,145 @@
1
+import DOMPurify from "dompurify";
2
+import { MinusIcon, PlusIcon, RotateCcwIcon } from "lucide-react";
3
+import { FileSurfaceAction } from "./fileSurfaceChrome";
4
+import { useEffect, useRef, useState } from "react";
5
+import { useTheme } from "~/hooks/useTheme";
6
+import { observeResize } from "~/lib/observeResize";
7
+
8
+export function sanitizeDocumentHtml(source: string): string {
9
+ const fragment = DOMPurify.sanitize(source, {
10
+ RETURN_DOM_FRAGMENT: true,
11
+ ALLOWED_TAGS: [
12
+ "p",
13
+ "h1",
14
+ "h2",
15
+ "h3",
16
+ "h4",
17
+ "h5",
18
+ "h6",
19
+ "div",
20
+ "span",
21
+ "strong",
22
+ "b",
23
+ "em",
24
+ "i",
25
+ "u",
26
+ "s",
27
+ "ul",
28
+ "ol",
29
+ "li",
30
+ "blockquote",
31
+ "pre",
32
+ "code",
33
+ "table",
34
+ "thead",
35
+ "tbody",
36
+ "tr",
37
+ "th",
38
+ "td",
39
+ "caption",
40
+ "figure",
41
+ "figcaption",
42
+ "br",
43
+ "hr",
44
+ "sub",
45
+ "sup",
46
+ "dl",
47
+ "dt",
48
+ "dd",
49
+ "img",
50
+ ],
51
+ ALLOWED_ATTR: ["src", "alt", "colspan", "rowspan", "data-odf-whitespace"],
52
+ });
53
+ for (const image of fragment.querySelectorAll("img")) {
54
+ if (
55
+ !/^data:image\/(png|jpeg|gif|webp);base64,[a-z\d+/]*={0,2}$/i.test(
56
+ image.getAttribute("src") ?? "",
57
+ )
58
+ )
59
+ image.remove();
60
+ }
61
+ const container = document.createElement("div");
62
+ container.append(fragment);
63
+ return container.innerHTML;
64
+}
65
+
66
+function documentColors() {
67
+ const style = getComputedStyle(document.documentElement);
68
+ return ["background", "foreground", "card", "border"].map((token) =>
69
+ style.getPropertyValue(`--${token}`),
70
+ );
71
+}
72
+
73
+export default function DocumentPages(props: {
74
+ readonly html: string;
75
+ readonly name: string;
76
+ readonly output?: boolean;
77
+ readonly details?: string;
78
+}) {
79
+ const { theme, resolvedTheme, themeHalves } = useTheme();
80
+ const [colors, setColors] = useState(documentColors);
81
+ // oxlint-disable-next-line t3code/react-effect-budget -- Read the host DOM's applied theme tokens so the sandboxed document receives the current palette.
82
+ useEffect(() => {
83
+ // The iframe cannot inherit the host palette across its sandbox boundary.
84
+ // oxlint-disable-next-line react/set-state-in-effect -- Read tokens after useTheme applies the selected palette.
85
+ setColors(documentColors());
86
+ // oxlint-disable-next-line react/exhaustive-effect-dependencies -- Theme values invalidate the palette read from the host DOM.
87
+ }, [theme, resolvedTheme, themeHalves]);
88
+ const [zoom, setZoom] = useState(1);
89
+ const [outputHeight, setOutputHeight] = useState(0);
90
+ const stopObserving = useRef<(() => void) | null>(null);
91
+ // oxlint-disable-next-line t3code/react-effect-budget -- Disconnect the iframe body's resize observer when the document unmounts.
92
+ useEffect(() => () => stopObserving.current?.(), []);
93
+ const html = sanitizeDocumentHtml(props.html);
94
+ const srcDoc = `<!doctype html><html><head><meta charset="utf-8"><meta http-equiv="Content-Security-Policy" content="default-src 'none'; img-src data:; style-src 'unsafe-inline'; base-uri 'none'; form-action 'none'"><style>:root{color-scheme:${resolvedTheme}}html{background:${colors[0]};color:${colors[1]};font:${15 * zoom}px/1.65 system-ui,sans-serif}body{box-sizing:border-box;margin:${props.output ? "0" : "20px auto"};padding:${props.output ? "16px" : "32px"};width:${props.output ? "100%" : "calc(100% - 40px)"};max-width:${860 * zoom}px;min-height:${props.output ? "0" : "calc(100vh - 40px)"};display:flow-root;background:${colors[2]};overflow-wrap:anywhere}h1,h2,h3{line-height:1.25}img{max-width:100%;height:auto}table{border-collapse:collapse;min-width:100%;max-width:100%}td,th{border:1px solid ${colors[3]};padding:6px 12px;text-align:left}th{background:${colors[0]}}tr:nth-child(even){background:${colors[0]}}pre,span[data-odf-whitespace]{white-space:pre-wrap}*{scrollbar-width:thin}${props.output ? "" : "@media(max-width:420px){body{padding:20px;width:calc(100% - 24px);margin:12px auto}}"}</style></head><body>${html}</body></html>`;
95
+ return (
96
+ <>
97
+ {!props.output ? (
98
+ <div className="flex min-h-10 shrink-0 flex-wrap items-center gap-1 border-b border-border/60 px-3 py-1 text-xs text-muted-foreground">
99
+ <span className="min-w-0 flex-1 text-muted-foreground tabular-nums" role="status">
100
+ {props.details}
101
+ </span>
102
+ <FileSurfaceAction
103
+ label="Decrease reading size"
104
+ disabled={zoom <= 0.5}
105
+ onPress={() => setZoom((value) => Math.max(0.5, value - 0.25))}
106
+ >
107
+ <MinusIcon className="size-3.5" />
108
+ </FileSurfaceAction>
109
+ <span className="min-w-10 text-center tabular-nums">{Math.round(zoom * 100)}%</span>
110
+ <FileSurfaceAction
111
+ label="Increase reading size"
112
+ disabled={zoom >= 2}
113
+ onPress={() => setZoom((value) => Math.min(2, value + 0.25))}
114
+ >
115
+ <PlusIcon className="size-3.5" />
116
+ </FileSurfaceAction>
117
+ <FileSurfaceAction
118
+ label="Reset reading size"
119
+ disabled={zoom === 1}
120
+ onPress={() => setZoom(1)}
121
+ >
122
+ <RotateCcwIcon className="size-3.5" />
123
+ </FileSurfaceAction>
124
+ </div>
125
+ ) : null}
126
+ <iframe
127
+ title={props.name}
128
+ srcDoc={srcDoc}
129
+ // Same-origin access lets the parent measure saved output; scripts remain blocked.
130
+ sandbox={props.output ? "allow-same-origin" : ""}
131
+ onLoad={(event) => {
132
+ if (!props.output) return;
133
+ stopObserving.current?.();
134
+ const body = event.currentTarget.contentDocument?.body;
135
+ if (!body) return;
136
+ const measure = () => setOutputHeight(Math.ceil(body.getBoundingClientRect().height));
137
+ stopObserving.current = observeResize(body, measure);
138
+ measure();
139
+ }}
140
+ style={props.output ? { height: outputHeight } : undefined}
141
+ className={props.output ? "block w-full border-0" : "min-h-0 flex-1 border-0"}
142
+ />
143
+ </>
144
+ );
145
+}
apps/web/src/components/files/FilePreviewPanel.test.ts
+255
-1
@@ -1,7 +1,9 @@
1
+import JSZip from "jszip";
2
+import { openRtf, openNotebook, openDocument, openEpub } from "./extendedDocumentPreview";
3
import type { ScopedThreadRef } from "@t3tools/contracts";
4
import { ProjectReadFileError } from "@t3tools/contracts";
5
import * as Schema from "effect/Schema";
4
-import { describe, expect, it } from "vite-plus/test";
6
+import { describe, expect, it, vi } from "vite-plus/test";
7
8
import { remapFileCommentAnnotations } from "./fileCommentAnnotations";
9
import {
@@ -144,3 +146,255 @@ describe("workspaceAssetResource", () => {
146
).toEqual({ _tag: "draft-workspace-file", cwd: "/repo", path: "/tmp/screenshots/shot.png" });
147
});
148
});
149
+
150
+describe("document preview contents", () => {
151
+ const bytes = (text: string) => new TextEncoder().encode(text).buffer;
152
+
153
+ it.each(["?", String.raw`\'3f`, String.raw`\\`])(
154
+ "consumes skipped rich text Unicode fallback %s",
155
+ (fallback) => {
156
+ const result = openRtf(bytes(String.raw`{\rtf1{\*\unknown\u945` + fallback + "}Hello}"));
157
+ if (result.type !== "document") throw new Error("Expected document");
158
+ expect(result.pages[0]!.html).toBe("<p>Hello</p>");
159
+ },
160
+ );
161
+
162
+ it("bounds EPUB markup made entirely of empty allowed elements", async () => {
163
+ const archive = new JSZip()
164
+ .file(
165
+ "META-INF/container.xml",
166
+ '<container><rootfiles><rootfile full-path="book.opf"/></rootfiles></container>',
167
+ )
168
+ .file(
169
+ "book.opf",
170
+ '<package><manifest><item id="a" href="chapter.xhtml" media-type="application/xhtml+xml"/></manifest><spine>' +
171
+ '<itemref idref="a"/>'.repeat(30) +
172
+ "</spine></package>",
173
+ )
174
+ .file("chapter.xhtml", "<html><body>" + "<figcaption/>".repeat(30_000) + "</body></html>");
175
+ await expect(openEpub(archive)).rejects.toThrow("too large");
176
+ });
177
+
178
+ it("decodes rich text formatting and Unicode without showing hidden destinations", () => {
179
+ const result = openRtf(
180
+ bytes(
181
+ String.raw`{\rtf1\ansi{\fonttbl{\f0 Hidden font;}}Hello \b bold\b0\par Unicode: \u945?{\*\unknown hidden}{\field{\*\fldinst HYPERLINK "https://example.com"}{\fldrslt visible}}}`,
182
+ ),
183
+ );
184
+ expect(result.type).toBe("document");
185
+ if (result.type !== "document") throw new Error("Expected document");
186
+ const html = result.pages[0]!.html;
187
+ expect(html.replaceAll(/<[^>]*>/g, "")).toBe("Hello boldUnicode: αvisible");
188
+ expect(html).toContain("<strong>");
189
+ expect(html).not.toContain("https:");
190
+ });
191
+
192
+ it("preserves rich text page breaks and formatting without splitting hidden destinations", () => {
193
+ const result = openRtf(
194
+ bytes(String.raw`{\rtf1\b First\page Second\b0{\*\unknown \page hidden}\page Third\page}`),
195
+ );
196
+ if (result.type !== "document") throw new Error("Expected document");
197
+ expect(result.navigation).toBe("page");
198
+ expect(result.pages.map((page) => page.html)).toEqual([
199
+ "<p><strong>First</strong></p>",
200
+ "<p><strong>Second</strong></p>",
201
+ "<p>Third</p>",
202
+ ]);
203
+ expect(() => openRtf(bytes(String.raw`{\rtf1 ` + String.raw`\page`.repeat(501) + "}"))).toThrow(
204
+ "500 pages",
205
+ );
206
+ });
207
+
208
+ it("decodes declared rich text code pages for raw bytes and escaped byte runs", () => {
209
+ const input = new TextEncoder().encode(
210
+ String.raw`{\rtf1\ansi\ansicpg1251 A \'c0\par\ansicpg932\'82\'a0}`,
211
+ );
212
+ input[input.indexOf(65)] = 0xc0;
213
+ const result = openRtf(input.buffer);
214
+ if (result.type !== "document") throw new Error("Expected document");
215
+ expect(result.pages[0]!.html).toBe("<p>А А<br>あ</p>");
216
+ expect(() => openRtf(bytes(String.raw`{\rtf1\ansicpg99999 text}`))).toThrow("code page");
217
+ });
218
+
219
+ it("resolves scoped ODF namespaces and explicit spreadsheet whitespace", async () => {
220
+ const archive = new JSZip().file(
221
+ "content.xml",
222
+ '<o:document-content xmlns:o="urn:oasis:names:tc:opendocument:xmlns:office:1.0" xmlns:t="urn:oasis:names:tc:opendocument:xmlns:table:1.0"><o:body><o:spreadsheet><t:table t:name="Accounts & names"><t:table-row><t:table-cell><p xmlns="urn:oasis:names:tc:opendocument:xmlns:text:1.0" xmlns:x="urn:oasis:names:tc:opendocument:xmlns:text:1.0">Account<s/>Name<s x:c="3"/>A<tab/>B<line-break/>C</p></t:table-cell></t:table-row></t:table></o:spreadsheet></o:body></o:document-content>',
223
+ );
224
+ const result = await openDocument(archive, "ods");
225
+ if (result.type !== "ods") throw new Error("Expected workbook");
226
+ expect(result.sheets[0]!.name).toBe("Accounts & names");
227
+ expect(result.sheets[0]!.rows[0]).toEqual(["Account Name A\tB\nC"]);
228
+ });
229
+
230
+ it("stops repeated slide images before encoding beyond the output budget", async () => {
231
+ const pixels = new Uint8Array(4 * 1024 * 1024);
232
+ pixels.set(
233
+ Buffer.from(
234
+ "iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAYAAAAfFcSJAAAADUlEQVQIHWP4z8DwHwAFgAI/ScLbtAAAAABJRU5ErkJggg==",
235
+ "base64",
236
+ ),
237
+ );
238
+ const archive = new JSZip()
239
+ .file(
240
+ "content.xml",
241
+ '<o:document-content xmlns:o="urn:oasis:names:tc:opendocument:xmlns:office:1.0" xmlns:d="urn:oasis:names:tc:opendocument:xmlns:drawing:1.0" xmlns:l="http://www.w3.org/1999/xlink"><o:body>' +
242
+ '<d:page><d:image l:href="picture.png"/></d:page>'.repeat(500) +
243
+ "</o:body></o:document-content>",
244
+ )
245
+ .file("picture.png", pixels);
246
+ const image = archive.file("picture.png")!;
247
+ const read = vi.spyOn(image, "async");
248
+ await expect(openDocument(archive, "odp")).rejects.toThrow("too large");
249
+ expect(read.mock.calls.filter(([type]) => type === "base64")).toHaveLength(2);
250
+ expect(read).toHaveBeenCalledTimes(5);
251
+ });
252
+
253
+ it("reads saved notebook cells and preferred outputs without executing code", () => {
254
+ const result = openNotebook(
255
+ bytes(
256
+ JSON.stringify({
257
+ nbformat: 4,
258
+ metadata: { language_info: { name: "python" } },
259
+ cells: [
260
+ { cell_type: "markdown", source: ["# Results"] },
261
+ {
262
+ cell_type: "code",
263
+ source: ["print(42)"],
264
+ outputs: [
265
+ { output_type: "stream", text: ["\n "] },
266
+ { output_type: "display_data", data: {} },
267
+ { data: { "text/html": [""] } },
268
+ { data: { "text/html": [], "text/plain": "array fallback" } },
269
+ { data: { "text/html": "", "text/plain": "string fallback" } },
270
+ { data: { "text/html": " ", "text/plain": "space fallback" } },
271
+ { output_type: "stream", text: ["42\n"] },
272
+ { data: { "text/html": ["<b>Results</b>"], "text/plain": ["Results"] } },
273
+ ],
274
+ },
275
+ ],
276
+ }),
277
+ ),
278
+ );
279
+ expect(result).toEqual({
280
+ type: "notebook",
281
+ language: "python",
282
+ cells: [
283
+ { kind: "markdown", source: "# Results", outputs: [] },
284
+ {
285
+ kind: "code",
286
+ source: "print(42)",
287
+ outputs: [
288
+ { kind: "text", text: "array fallback" },
289
+ { kind: "text", text: "string fallback" },
290
+ { kind: "text", text: "space fallback" },
291
+ { kind: "text", text: "42\n" },
292
+ { kind: "html", text: "<b>Results</b>" },
293
+ ],
294
+ },
295
+ ],
296
+ });
297
+ expect(() => openNotebook(bytes('{"nbformat":3,"cells":[]}'))).toThrow("version 4");
298
+ expect(() =>
299
+ openNotebook(
300
+ bytes(JSON.stringify({ nbformat: 4, cells: Array.from({ length: 501 }, () => ({})) })),
301
+ ),
302
+ ).toThrow("500 cells");
303
+ });
304
+
305
+ it("bounds repeated OpenDocument spreadsheet cells and keeps sheet names and values", async () => {
306
+ const archive = new JSZip().file(
307
+ "content.xml",
308
+ '<office:document-content xmlns:office="urn:oasis:names:tc:opendocument:xmlns:office:1.0" xmlns:table="urn:oasis:names:tc:opendocument:xmlns:table:1.0" xmlns:text="urn:oasis:names:tc:opendocument:xmlns:text:1.0"><office:body><office:spreadsheet><table:table table:name="Budget"><table:table-row table:number-rows-repeated="999999999"><table:table-cell table:number-columns-repeated="999999999"><text:p>Research & design</text:p></table:table-cell></table:table-row></table:table></office:spreadsheet></office:body></office:document-content>',
309
+ );
310
+ const result = await openDocument(archive, "ods");
311
+ expect(result.type).toBe("ods");
312
+ if (result.type !== "ods") throw new Error("Expected workbook");
313
+ expect(result.sheets[0]!.name).toBe("Budget");
314
+ expect(result.sheets[0]!.truncated).toBe(true);
315
+ expect(result.sheets[0]!.rows).toHaveLength(500);
316
+ expect(result.sheets[0]!.rows[0]).toHaveLength(50);
317
+ expect(result.sheets[0]!.rows[0]![0]).toBe("Research & design");
318
+ });
319
+
320
+ it.each([1, 500])("ignores styled ODS filler after %s populated rows", async (count) => {
321
+ const archive = new JSZip().file(
322
+ "content.xml",
323
+ `<office:document><table:table><table:table-row table:number-rows-repeated="${count}"><table:table-cell><text:p>North</text:p></table:table-cell><table:table-cell table:number-columns-repeated="1024"/></table:table-row><table:table-row table:number-rows-repeated="1048570"><table:table-cell table:number-columns-repeated="1024"/></table:table-row></table:table></office:document>`,
324
+ );
325
+ const result = await openDocument(archive, "ods");
326
+ if (result.type !== "ods") throw new Error("Expected workbook");
327
+ expect(result.sheets[0]).toMatchObject({ rowCount: count, columns: 1, truncated: false });
328
+ expect(result.sheets[0]!.rows).toEqual(Array.from({ length: count }, () => ["North"]));
329
+ });
330
+
331
+ it.each([2, 999999])(
332
+ "preserves internal ODS gaps of %s rows and reports real overflow",
333
+ async (gap) => {
334
+ const archive = new JSZip().file(
335
+ "content.xml",
336
+ `<office:document><table:table><table:table-row><table:table-cell><text:p>First</text:p></table:table-cell></table:table-row><table:table-row table:number-rows-repeated="${gap}"><table:table-cell/></table:table-row><table:table-row><table:table-cell table:number-columns-repeated="2"/><table:table-cell><text:p>Last</text:p></table:table-cell></table:table-row></table:table></office:document>`,
337
+ );
338
+ const result = await openDocument(archive, "ods");
339
+ if (result.type !== "ods") throw new Error("Expected workbook");
340
+ expect(result.sheets[0]).toMatchObject({
341
+ rowCount: Math.min(gap + 2, 500),
342
+ columns: gap === 2 ? 3 : 1,
343
+ truncated: gap !== 2,
344
+ });
345
+ expect(result.sheets[0]!.rows[1]).toEqual([]);
346
+ if (gap === 2) expect(result.sheets[0]!.rows[3]).toEqual(["", "", "Last"]);
347
+ },
348
+ );
349
+
350
+ it.each(["odt", "odp"] as const)("reads %s text and rejects XML entities", async (kind) => {
351
+ const archive = new JSZip().file(
352
+ "content.xml",
353
+ `<office:document-content xmlns:office="urn:oasis:names:tc:opendocument:xmlns:office:1.0" xmlns:text="urn:oasis:names:tc:opendocument:xmlns:text:1.0" xmlns:draw="urn:oasis:names:tc:opendocument:xmlns:drawing:1.0"><office:body>${kind === "odt" ? "<office:text><text:p>Field notes</text:p></office:text>" : '<draw:page draw:name="Overview"><text:h>Launch plan</text:h></draw:page>'}</office:body></office:document-content>`,
354
+ );
355
+ const result = await openDocument(archive, kind);
356
+ expect(result.type).toBe("document");
357
+ if (result.type !== "document") throw new Error("Expected document");
358
+ expect(result.pages[0]!.html).toContain(kind === "odt" ? "Field notes" : "Launch plan");
359
+ archive.file(
360
+ "content.xml",
361
+ '<!DOCTYPE x [<!ENTITY external SYSTEM "file:///etc/passwd">]><x>&external;</x>',
362
+ );
363
+ await expect(openDocument(archive, kind)).rejects.toThrow("unsupported XML");
364
+ });
365
+
366
+ it("reads EPUB chapters in spine order and removes active and remote content", async () => {
367
+ const archive = new JSZip()
368
+ .file(
369
+ "META-INF/container.xml",
370
+ '<container><rootfiles><rootfile full-path="Book/book.opf"/></rootfiles></container>',
371
+ )
372
+ .file(
373
+ "Book/book.opf",
374
+ '<package><manifest><item id="a" href="one.xhtml" media-type="application/xhtml+xml"/><item id="b" href="two&&&.xhtml" media-type="application/xhtml+xml"/></manifest><spine><itemref idref="b"/><itemref idref="a"/></spine></package>',
375
+ )
376
+ .file(
377
+ "Book/one.xhtml",
378
+ '<html><head><title>First</title></head><body><h1>First chapter</h1><img src="cover.png" alt="Cover"/><img src="https://example.com/tracker.png"/><script>alert(1)</script><style>@import "https://example.com/style.css";</style></body></html>',
379
+ )
380
+ .file(
381
+ "Book/two&&&.xhtml",
382
+ '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.1//EN" "http://www.w3.org/TR/xhtml11/DTD/xhtml11.dtd"><html><head><title>Second</title></head><body><h1>Second chapter</h1><p>A B © — ≂̸ &lt;</p><img src="cover.png" alt="© ≂̸"/></body></html>',
383
+ )
384
+ .file(
385
+ "Book/cover.png",
386
+ Buffer.from(
387
+ "iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAYAAAAfFcSJAAAADUlEQVQIHWP4z8DwHwAFgAI/ScLbtAAAAABJRU5ErkJggg==",
388
+ "base64",
389
+ ),
390
+ );
391
+ const result = await openEpub(archive);
392
+ expect(result.type).toBe("document");
393
+ if (result.type !== "document") throw new Error("Expected document");
394
+ expect(result.pages.map((page) => page.title)).toEqual(["Second", "First"]);
395
+ expect(result.pages[0]!.html).toContain("A\u00a0B © — ≂̸ &lt;");
396
+ expect(result.pages[0]!.html).toContain('alt="© ≂̸"');
397
+ expect(result.pages[1]!.html).toContain('src="data:image/png;base64,');
398
+ expect(result.pages[1]!.html).not.toMatch(/https:|script|alert/);
399
+ });
400
+});
apps/web/src/components/files/NotebookPreview.tsx
new
+182
@@ -0,0 +1,182 @@
1
+/* oxlint-disable react/no-array-index-key -- Notebook cells and outputs retain their saved source order. */
2
+import { formatAttachmentSize } from "@t3tools/client-runtime/state/attachments";
3
+import { useEffect, useRef, useState } from "react";
4
+import type { Root, RootContent } from "mdast";
5
+import ChatMarkdown from "~/components/ChatMarkdown";
6
+import { Button } from "~/components/ui/button";
7
+import ReadOnlySourcePreview from "./ReadOnlySourcePreview";
8
+import DocumentPages from "./DocumentPages";
9
+import { FileSurfaceNotice } from "./fileSurfaceChrome";
10
+import type { OfficeWorkerResult } from "./officePreviewProtocol";
11
+
12
+// Saved notebooks must not load URLs or turn authored links into app actions.
13
+function inertNotebookMarkdown() {
14
+ function clean(nodes: RootContent[]): RootContent[] {
15
+ return nodes.flatMap((node): RootContent[] => {
16
+ if (node.type === "html" || node.type === "definition") return [];
17
+ if (node.type === "image" || node.type === "imageReference")
18
+ return [{ type: "text", value: node.alt ?? "" }];
19
+ if (node.type === "link" || node.type === "linkReference") return clean(node.children);
20
+ if (node.type === "inlineCode") return [{ type: "text", value: node.value }];
21
+ if (node.type === "code" && node.lang === "mermaid") return [{ ...node, lang: "text" }];
22
+ if ("children" in node) node.children = clean(node.children) as typeof node.children;
23
+ return [node];
24
+ });
25
+ }
26
+ return (tree: Root) => {
27
+ tree.children = clean(tree.children) as Root["children"];
28
+ };
29
+}
30
+const NOTEBOOK_PLUGINS = [inertNotebookMarkdown];
31
+
32
+function NotebookSource(props: { readonly name: string; readonly text: string }) {
33
+ const [height, setHeight] = useState(Math.min(360, props.text.split("\n").length * 20 + 16));
34
+ const measure = (node: HTMLElement) => {
35
+ setHeight(Math.min(360, Math.ceil(node.getBoundingClientRect().height)));
36
+ };
37
+ return (
38
+ <div className="flex min-h-0 flex-col" style={{ height }}>
39
+ <ReadOnlySourcePreview {...props} onPostRender={measure} />
40
+ </div>
41
+ );
42
+}
43
+
44
+export default function NotebookPreview(props: {
45
+ readonly notebook: Extract<OfficeWorkerResult, { type: "notebook" }>;
46
+ readonly sizeBytes?: number | undefined;
47
+}) {
48
+ const [visible, setVisible] = useState(20);
49
+ const scroll = useRef<HTMLDivElement>(null);
50
+ const [outputView, setOutputView] = useState(() => ({
51
+ mounted: new Set<string>(),
52
+ heights: new Map<string, number>(),
53
+ }));
54
+ const outputCount = props.notebook.cells
55
+ .slice(0, visible)
56
+ .reduce(
57
+ (count, cell) => count + cell.outputs.filter((output) => output.kind === "html").length,
58
+ 0,
59
+ );
60
+ // oxlint-disable-next-line t3code/react-effect-budget -- Observe output visibility and measure departing outputs to bound iframe lifetimes while preserving scroll height.
61
+ useEffect(() => {
62
+ const root = scroll.current;
63
+ if (!root || !outputCount) return;
64
+ const candidates = new Map<Element, string>();
65
+ let active: Element[] = [];
66
+ const observer = new IntersectionObserver(
67
+ (entries) => {
68
+ for (const entry of entries) {
69
+ if (entry.isIntersecting)
70
+ candidates.set(entry.target, (entry.target as HTMLElement).dataset.notebookOutput!);
71
+ else candidates.delete(entry.target);
72
+ }
73
+ // Bound browsing contexts even when a cell contains hundreds of tiny HTML outputs.
74
+ const next = [...candidates.keys()].slice(0, 16);
75
+ const measured = active
76
+ .filter((node) => !next.includes(node))
77
+ .map(
78
+ (node) =>
79
+ [
80
+ (node as HTMLElement).dataset.notebookOutput!,
81
+ node.getBoundingClientRect().height,
82
+ ] as const,
83
+ );
84
+ active = next;
85
+ setOutputView((old) => ({
86
+ mounted: new Set(next.map((node) => candidates.get(node)!)),
87
+ heights: new Map([...old.heights, ...measured]),
88
+ }));
89
+ },
90
+ { root, rootMargin: "160px" },
91
+ );
92
+ for (const node of root.querySelectorAll("[data-notebook-output]")) observer.observe(node);
93
+ return () => observer.disconnect();
94
+ }, [outputCount]);
95
+ return (
96
+ <>
97
+ <div
98
+ className="border-b border-border/60 px-3 py-2 text-xs text-muted-foreground tabular-nums"
99
+ role="status"
100
+ >
101
+ {props.notebook.cells.length.toLocaleString()} cells · {props.notebook.language}
102
+ {props.sizeBytes !== undefined
103
+ ? ` · ${props.sizeBytes === 0 ? "0 bytes" : formatAttachmentSize(props.sizeBytes)}`
104
+ : ""}
105
+ </div>
106
+ <FileSurfaceNotice>Saved outputs only. Code is never executed.</FileSurfaceNotice>
107
+ <div ref={scroll} className="scrollbar-gutter-both min-h-0 flex-1 overflow-auto">
108
+ <div className="mx-auto max-w-4xl space-y-5 px-5 py-6">
109
+ {props.notebook.cells.slice(0, visible).map((cell, index) => (
110
+ <section
111
+ key={index}
112
+ aria-label={`Cell ${index + 1}`}
113
+ className="min-w-0 overflow-hidden rounded-md border border-border/60"
114
+ >
115
+ <div className="border-b border-border/60 bg-muted/30 px-3 py-1.5 text-xs text-muted-foreground">
116
+ {cell.kind === "code" ? props.notebook.language : cell.kind} · {index + 1}
117
+ </div>
118
+ {cell.kind === "markdown" ? (
119
+ <ChatMarkdown
120
+ text={cell.source}
121
+ cwd={undefined}
122
+ parseRawHtml={false}
123
+ extraRemarkPlugins={NOTEBOOK_PLUGINS}
124
+ className="px-4 py-3"
125
+ />
126
+ ) : cell.source.trim() ? (
127
+ <NotebookSource
128
+ name={`cell.${cell.kind === "code" ? (({ python: "py", javascript: "js", typescript: "ts", julia: "jl" } as Record<string, string>)[props.notebook.language] ?? props.notebook.language) : "txt"}`}
129
+ text={cell.source}
130
+ />
131
+ ) : null}
132
+ {cell.outputs.map((output, outputIndex) => (
133
+ <div
134
+ key={outputIndex}
135
+ data-notebook-output={
136
+ output.kind === "html" ? `${index}-${outputIndex}` : undefined
137
+ }
138
+ className="border-t border-border/60"
139
+ style={{
140
+ minHeight:
141
+ output.kind !== "html" || outputView.mounted.has(`${index}-${outputIndex}`)
142
+ ? undefined
143
+ : (outputView.heights.get(`${index}-${outputIndex}`) ?? 80),
144
+ }}
145
+ >
146
+ {output.kind === "html" ? (
147
+ !outputView.mounted.has(`${index}-${outputIndex}`) ? null : (
148
+ <DocumentPages
149
+ html={output.text}
150
+ name={`Cell ${index + 1} HTML output`}
151
+ output
152
+ />
153
+ )
154
+ ) : output.kind === "image" ? (
155
+ <img
156
+ src={output.text}
157
+ alt={`Cell ${index + 1} output`}
158
+ loading="lazy"
159
+ className="max-h-96 max-w-full p-4 object-contain"
160
+ />
161
+ ) : (
162
+ <pre className="overflow-x-auto p-4 text-xs whitespace-pre-wrap break-words">
163
+ {output.text}
164
+ </pre>
165
+ )}
166
+ </div>
167
+ ))}
168
+ </section>
169
+ ))}
170
+ {props.notebook.cells.length === 0 ? (
171
+ <p className="text-sm text-muted-foreground">This notebook is empty.</p>
172
+ ) : null}
173
+ {visible < props.notebook.cells.length ? (
174
+ <Button variant="outline" size="sm" onClick={() => setVisible((count) => count + 20)}>
175
+ Show more cells ({props.notebook.cells.length - visible} remaining)
176
+ </Button>
177
+ ) : null}
178
+ </div>
179
+ </div>
180
+ </>
181
+ );
182
+}
apps/web/src/components/files/OfficeViewer.tsx
+229
-50
@@ -1,15 +1,29 @@
1
-import { useStableCallback } from "~/hooks/useStableCallback";
1
+/* oxlint-disable react/no-array-index-key -- Spreadsheet rows and columns retain their source positions. */
2
+import { formatAttachmentSize } from "@t3tools/client-runtime/state/attachments";
3
import type { OfficePreviewKind } from "@t3tools/shared/filePreview";
3
-import { lazy, Suspense, useEffect, useState } from "react";
4
+import { ChevronLeftIcon, ChevronRightIcon } from "lucide-react";
5
+import { lazy, Suspense, useEffect, useRef, useState } from "react";
6
+
7
+import VirtualTablePreview from "./VirtualTablePreview";
8
+
9
import {
5
- FileSurfaceLoading,
6
- FileSurfaceNotice,
7
- FILE_SURFACE_SUBHEADER_CLASS,
8
-} from "./fileSurfaceChrome";
10
+ Select,
11
+ SelectItem,
12
+ SelectPopup,
13
+ SelectTrigger,
14
+ SelectValue,
15
+} from "~/components/ui/select";
16
+
17
+import { FileSurfaceAction, FileSurfaceLoading, FileSurfaceNotice } from "./fileSurfaceChrome";
18
import { OFFICE_PREVIEW_MAX_BYTES, type OfficeWorkerResult } from "./officePreviewProtocol";
10
-const WordDocument = lazy(() => import("./office/WordDocument"));
11
-const Spreadsheet = lazy(() => import("./office/Spreadsheet"));
12
-const Presentation = lazy(() => import("./office/Presentation"));
19
+
20
+const DocumentPages = lazy(() => import("./DocumentPages"));
21
+const NotebookPreview = lazy(() => import("./NotebookPreview"));
22
+
23
+type Workbook = Extract<OfficeWorkerResult, { type: "ods" }>;
24
+
25
+const OpenXmlOfficeViewer = lazy(() => import("./OpenXmlOfficeViewer"));
26
+
27
async function readOfficeFile(src: string, signal: AbortSignal) {
28
const response = await fetch(src, { signal });
29
if (!response.ok) throw new Error("The file could not be loaded. Close it and try again.");
@@ -38,59 +52,108 @@ export default function OfficeViewer(props: {
52
readonly name: string;
53
readonly kind: OfficePreviewKind;
54
readonly sizeBytes?: number;
55
+}) {
56
+ if (props.kind === "docx" || props.kind === "xlsx" || props.kind === "pptx")
57
+ return (
58
+ <Suspense fallback={<FileSurfaceLoading />}>
59
+ <OpenXmlOfficeViewer {...props} kind={props.kind} />
60
+ </Suspense>
61
+ );
62
+ return <ExtendedOfficeViewer {...props} />;
63
+}
64
+
65
+function ExtendedOfficeViewer(props: {
66
+ readonly src: string;
67
+ readonly name: string;
68
+ readonly kind: OfficePreviewKind;
69
+ readonly sizeBytes?: number;
70
}) {
71
const [error, setError] = useState<string | null>(null);
43
- const [prepared, setPrepared] = useState<Exclude<OfficeWorkerResult, { type: "error" }>>();
44
- const [details, setDetails] = useState("");
45
- const fail = useStableCallback((cause: unknown) =>
46
- setError(cause instanceof Error ? cause.message : "Could not render this Office document."),
47
- );
72
+ const [sheetPages, setSheetPages] = useState(new Map<string, string[][]>());
73
+ const pendingPages = useRef(new Set<string>());
74
+ const [previewDocument, setDocument] = useState<Extract<
75
+ OfficeWorkerResult,
76
+ { type: "document" }
77
+ > | null>(null);
78
+ const [notebook, setNotebook] = useState<Extract<
79
+ OfficeWorkerResult,
80
+ { type: "notebook" }
81
+ > | null>(null);
82
+ const [workbook, setWorkbook] = useState<Workbook | null>(null);
83
+ const [count, setCount] = useState(0);
84
+ const [fileSize, setFileSize] = useState(props.sizeBytes);
85
+ const [page, setPage] = useState(0);
86
+ const workerRef = useRef<Worker | null>(null);
87
+ const timerRef = useRef<ReturnType<typeof setTimeout> | null>(null);
88
// oxlint-disable-next-line t3code/react-effect-budget -- Own the Office fetch, preparation worker, and timeout for this source and release them on cleanup.
89
useEffect(() => {
90
+ const controller = new AbortController();
91
if ((props.sizeBytes ?? 0) > OFFICE_PREVIEW_MAX_BYTES) {
51
- // Metadata bounds are checked before fetching or starting an engine.
52
- // oxlint-disable-next-line react/set-state-in-effect
92
+ // oxlint-disable-next-line react/set-state-in-effect -- The file metadata can reject the preview before loading bytes or engines.
93
setError("Preview limited to files under 50 MB.");
94
return;
95
}
56
- const controller = new AbortController();
96
const worker = new Worker(new URL("./officePreview.worker.ts", import.meta.url), {
97
type: "module",
98
});
60
- const failure = (cause: unknown) => {
61
- if (!controller.signal.aborted) fail(cause);
99
+ workerRef.current = worker;
100
+ const fail = (message: string) => {
101
+ if (!controller.signal.aborted) setError(message);
102
controller.abort();
103
worker.terminate();
104
+ if (timerRef.current) clearTimeout(timerRef.current);
105
};
65
- const timer = setTimeout(
66
- () => failure(new Error("This document took too long to preview. Open it in another app.")),
67
- 30000,
106
+ timerRef.current = setTimeout(
107
+ () => fail("This document took too long to preview. Open it in another app."),
108
+ 30_000,
109
);
69
- worker.addEventListener("error", () => failure(new Error("Could not load the Office viewer.")));
70
- worker.addEventListener("messageerror", () =>
71
- failure(new Error("Could not read the Office viewer response.")),
110
+ worker.addEventListener("error", () =>
111
+ fail("Could not load the Office viewer. Close the file and try again."),
112
);
113
worker.addEventListener("message", (event: MessageEvent<OfficeWorkerResult>) => {
114
if (controller.signal.aborted) return;
75
- clearTimeout(timer);
76
- if (event.data.type === "error") failure(new Error(event.data.message));
77
- else {
78
- setPrepared(event.data);
79
- worker.terminate();
115
+ const result = event.data;
116
+ if (result.type === "error") fail(result.message);
117
+ else if (result.type === "document") {
118
+ if (timerRef.current) clearTimeout(timerRef.current);
119
+ setDocument(result);
120
+ setCount(result.pages.length > 1 ? result.pages.length : 0);
121
+ } else if (result.type === "notebook") {
122
+ if (timerRef.current) clearTimeout(timerRef.current);
123
+ setNotebook(result);
124
+ } else if (result.type === "ods") {
125
+ if (timerRef.current) clearTimeout(timerRef.current);
126
+ setWorkbook(result);
127
+ setCount(result.sheets.length);
128
+ } else if (result.type === "sheet-page") {
129
+ const key = `${result.sheet}:${result.index}`;
130
+ pendingPages.current.delete(key);
131
+ setSheetPages((old) => {
132
+ const next = new Map(old);
133
+ next.set(key, result.rows);
134
+ if (next.size > 12) next.delete(next.keys().next().value!);
135
+ return next;
136
+ });
137
}
138
});
139
void readOfficeFile(props.src, controller.signal)
140
.then((bytes) => {
84
- if (!controller.signal.aborted)
141
+ if (!controller.signal.aborted) {
142
+ setFileSize(bytes.byteLength);
143
worker.postMessage({ type: "open", kind: props.kind, bytes }, [bytes]);
144
+ }
145
})
87
- .catch(failure);
146
+ .catch((cause: unknown) =>
147
+ fail(cause instanceof Error ? cause.message : "Could not read this file."),
148
+ );
149
return () => {
150
controller.abort();
151
worker.terminate();
91
- clearTimeout(timer);
152
+ workerRef.current = null;
153
+ if (timerRef.current) clearTimeout(timerRef.current);
154
};
93
- }, [props.src, props.kind, props.sizeBytes, fail]);
155
+ }, [props.src, props.kind, props.sizeBytes]);
156
+ const selectPage = setPage;
157
if (error)
158
return (
159
<div
@@ -103,27 +166,143 @@ export default function OfficeViewer(props: {
166
</p>
167
</div>
168
);
106
- if (!prepared) return <FileSurfaceLoading />;
107
- const Viewer =
108
- props.kind === "docx" ? WordDocument : props.kind === "xlsx" ? Spreadsheet : Presentation;
169
+ if (notebook)
170
+ return (
171
+ <Suspense fallback={<FileSurfaceLoading />}>
172
+ <NotebookPreview notebook={notebook} sizeBytes={fileSize} />
173
+ </Suspense>
174
+ );
175
+ const sheet = workbook?.sheets[page];
176
+ const documentPage = previewDocument?.pages[page];
177
+ const navigation = workbook ? "sheet" : (previewDocument?.navigation ?? "slide");
178
+ const documentCount = previewDocument?.pages.length ?? 0;
179
+ const documentDetails =
180
+ navigation === "chapter"
181
+ ? `Chapter ${page + 1} of ${documentCount}`
182
+ : `${documentCount} ${navigation}${documentCount === 1 ? "" : "s"}`;
183
return (
184
<div className="flex min-h-0 flex-1 flex-col">
111
- {details ? (
112
- <div className={FILE_SURFACE_SUBHEADER_CLASS}>
113
- <span className="text-xs text-muted-foreground" aria-live="polite">
114
- {details}
115
- {props.kind !== "xlsx" ? " · Fit to width" : ""}
116
- </span>
185
+ {count > 0 ? (
186
+ <div className="flex h-10 shrink-0 items-center gap-2 border-b border-border/60 px-3 text-xs">
187
+ <FileSurfaceAction
188
+ label={`Previous ${navigation}`}
189
+ disabled={page === 0}
190
+ onPress={() => selectPage(page - 1)}
191
+ >
192
+ <ChevronLeftIcon className="size-3.5" />
193
+ </FileSurfaceAction>
194
+ {workbook ? (
195
+ <div className="min-w-0 flex-1">
196
+ <Select
197
+ value={page}
198
+ onValueChange={(value) => {
199
+ if (value !== null) selectPage(value);
200
+ }}
201
+ >
202
+ <SelectTrigger aria-label="Sheet" size="xs">
203
+ <SelectValue>{sheet?.name}</SelectValue>
204
+ </SelectTrigger>
205
+ <SelectPopup>
206
+ {workbook.sheets.map((item, index) => (
207
+ <SelectItem key={index} value={index}>
208
+ {item.name}
209
+ </SelectItem>
210
+ ))}
211
+ </SelectPopup>
212
+ </Select>
213
+ </div>
214
+ ) : previewDocument?.navigation === "chapter" ? (
215
+ <div className="min-w-0 flex-1">
216
+ <Select
217
+ value={page}
218
+ onValueChange={(value) => {
219
+ if (value !== null) selectPage(value);
220
+ }}
221
+ >
222
+ <SelectTrigger aria-label="Chapter" size="xs">
223
+ <SelectValue>{documentPage?.title}</SelectValue>
224
+ </SelectTrigger>
225
+ <SelectPopup>
226
+ {previewDocument.pages.map((item, index) => (
227
+ <SelectItem key={index} value={index}>
228
+ {item.title}
229
+ </SelectItem>
230
+ ))}
231
+ </SelectPopup>
232
+ </Select>
233
+ </div>
234
+ ) : (
235
+ <span className="flex-1 text-center tabular-nums" aria-live="polite">
236
+ {navigation === "slide" ? "Slide" : "Page"} {page + 1} of {count}
237
+ </span>
238
+ )}
239
+ <FileSurfaceAction
240
+ label={`Next ${navigation}`}
241
+ disabled={page + 1 === count}
242
+ onPress={() => selectPage(page + 1)}
243
+ >
244
+ <ChevronRightIcon className="size-3.5" />
245
+ </FileSurfaceAction>
246
</div>
247
) : null}
119
- {prepared.truncated ? (
120
- <FileSurfaceNotice>
121
- Preview limited. Save it or open it in another app to read the full document.
122
- </FileSurfaceNotice>
248
+ {props.kind === "ods" && sheet ? (
249
+ <div
250
+ className="flex min-h-10 shrink-0 flex-wrap items-center gap-x-1 border-b border-border/60 px-3 py-2 text-xs text-muted-foreground tabular-nums"
251
+ role="status"
252
+ >
253
+ Sheet {page + 1} of {count} · {sheet.rowCount.toLocaleString()} rows ·{" "}
254
+ {sheet.columns.toLocaleString()} columns
255
+ {fileSize !== undefined
256
+ ? ` · ${fileSize === 0 ? "0 bytes" : formatAttachmentSize(fileSize)}`
257
+ : ""}
258
+ </div>
259
) : null}
124
- <Suspense fallback={<FileSurfaceLoading />}>
125
- <Viewer bytes={prepared.bytes} onDetails={setDetails} onError={fail} />
126
- </Suspense>
260
+ {documentPage ? (
261
+ <>
262
+ {props.kind !== "epub" ? (
263
+ <FileSurfaceNotice>
264
+ Text and images. Layout and embedded objects may differ from the original.
265
+ </FileSurfaceNotice>
266
+ ) : null}
267
+ <Suspense fallback={<FileSurfaceLoading />}>
268
+ <DocumentPages
269
+ html={documentPage.html}
270
+ name={`${props.name}: ${documentPage.title}`}
271
+ details={`${documentDetails}${fileSize !== undefined ? ` · ${fileSize === 0 ? "0 bytes" : formatAttachmentSize(fileSize)}` : ""}`}
272
+ />
273
+ </Suspense>
274
+ </>
275
+ ) : sheet ? (
276
+ <>
277
+ <FileSurfaceNotice>
278
+ Cell values only. Charts, images, and formatting are not shown.
279
+ {sheet.truncated
280
+ ? props.kind === "ods"
281
+ ? " Limited to 500 rows and 50 columns."
282
+ : " Preview limited to 250,000 rows, 1,024 columns, and bounded cell text."
283
+ : ""}
284
+ </FileSurfaceNotice>
285
+ {sheet.rowCount === 0 ? (
286
+ <p className="p-6 text-xs text-muted-foreground">This sheet is empty.</p>
287
+ ) : (
288
+ <VirtualTablePreview
289
+ key={page}
290
+ name={`${props.name}: ${sheet.name}`}
291
+ count={sheet.rowCount}
292
+ columns={sheet.columns}
293
+ getRows={(index) => (index === 0 ? sheet.rows : sheetPages.get(`${page}:${index}`))}
294
+ loadPage={(index) => {
295
+ const key = `${page}:${index}`;
296
+ if (pendingPages.current.has(key)) return;
297
+ pendingPages.current.add(key);
298
+ workerRef.current?.postMessage({ type: "sheet-page", sheet: page, index });
299
+ }}
300
+ />
301
+ )}
302
+ </>
303
+ ) : (
304
+ <FileSurfaceLoading />
305
+ )}
306
</div>
307
);
308
}
apps/web/src/components/files/OpenXmlOfficeViewer.tsx
new
+129
@@ -0,0 +1,129 @@
1
+import { useStableCallback } from "~/hooks/useStableCallback";
2
+import type { OfficePreviewKind } from "@t3tools/shared/filePreview";
3
+import { lazy, Suspense, useEffect, useState } from "react";
4
+import {
5
+ FileSurfaceLoading,
6
+ FileSurfaceNotice,
7
+ FILE_SURFACE_SUBHEADER_CLASS,
8
+} from "./fileSurfaceChrome";
9
+import { OFFICE_PREVIEW_MAX_BYTES, type OfficeWorkerResult } from "./officePreviewProtocol";
10
+const WordDocument = lazy(() => import("./office/WordDocument"));
11
+const Spreadsheet = lazy(() => import("./office/Spreadsheet"));
12
+const Presentation = lazy(() => import("./office/Presentation"));
13
+async function readOfficeFile(src: string, signal: AbortSignal) {
14
+ const response = await fetch(src, { signal });
15
+ if (!response.ok) throw new Error("The file could not be loaded. Close it and try again.");
16
+ if (Number(response.headers.get("content-length")) > OFFICE_PREVIEW_MAX_BYTES)
17
+ throw new Error("Preview limited to files under 50 MB.");
18
+ const reader = response.body?.getReader();
19
+ if (!reader) throw new Error("The file is unavailable.");
20
+ const chunks: Uint8Array<ArrayBuffer>[] = [];
21
+ let size = 0;
22
+ try {
23
+ while (true) {
24
+ const { value, done } = await reader.read();
25
+ if (done) break;
26
+ size += value.length;
27
+ if (size > OFFICE_PREVIEW_MAX_BYTES) throw new Error("Preview limited to files under 50 MB.");
28
+ chunks.push(value);
29
+ }
30
+ } finally {
31
+ await reader.cancel();
32
+ }
33
+ return new Blob(chunks).arrayBuffer();
34
+}
35
+
36
+export default function OfficeViewer(props: {
37
+ readonly src: string;
38
+ readonly name: string;
39
+ readonly kind: Extract<OfficePreviewKind, "docx" | "xlsx" | "pptx">;
40
+ readonly sizeBytes?: number;
41
+}) {
42
+ const [error, setError] = useState<string | null>(null);
43
+ const [prepared, setPrepared] = useState<Extract<OfficeWorkerResult, { bytes: ArrayBuffer }>>();
44
+ const [details, setDetails] = useState("");
45
+ const fail = useStableCallback((cause: unknown) =>
46
+ setError(cause instanceof Error ? cause.message : "Could not render this Office document."),
47
+ );
48
+ // oxlint-disable-next-line t3code/react-effect-budget -- Own the Open XML fetch, preparation worker, and timeout for this source and release them on cleanup.
49
+ useEffect(() => {
50
+ if ((props.sizeBytes ?? 0) > OFFICE_PREVIEW_MAX_BYTES) {
51
+ // Metadata bounds are checked before fetching or starting an engine.
52
+ // oxlint-disable-next-line react/set-state-in-effect
53
+ setError("Preview limited to files under 50 MB.");
54
+ return;
55
+ }
56
+ const controller = new AbortController();
57
+ const worker = new Worker(new URL("./officePreview.worker.ts", import.meta.url), {
58
+ type: "module",
59
+ });
60
+ const failure = (cause: unknown) => {
61
+ if (!controller.signal.aborted) fail(cause);
62
+ controller.abort();
63
+ worker.terminate();
64
+ };
65
+ const timer = setTimeout(
66
+ () => failure(new Error("This document took too long to preview. Open it in another app.")),
67
+ 30000,
68
+ );
69
+ worker.addEventListener("error", () => failure(new Error("Could not load the Office viewer.")));
70
+ worker.addEventListener("messageerror", () =>
71
+ failure(new Error("Could not read the Office viewer response.")),
72
+ );
73
+ worker.addEventListener("message", (event: MessageEvent<OfficeWorkerResult>) => {
74
+ if (controller.signal.aborted) return;
75
+ clearTimeout(timer);
76
+ if (event.data.type === "error") failure(new Error(event.data.message));
77
+ else if ("bytes" in event.data) {
78
+ setPrepared(event.data);
79
+ worker.terminate();
80
+ }
81
+ });
82
+ void readOfficeFile(props.src, controller.signal)
83
+ .then((bytes) => {
84
+ if (!controller.signal.aborted)
85
+ worker.postMessage({ type: "open", kind: props.kind, bytes }, [bytes]);
86
+ })
87
+ .catch(failure);
88
+ return () => {
89
+ controller.abort();
90
+ worker.terminate();
91
+ clearTimeout(timer);
92
+ };
93
+ }, [props.src, props.kind, props.sizeBytes, fail]);
94
+ if (error)
95
+ return (
96
+ <div
97
+ role="alert"
98
+ className="flex min-h-0 flex-1 flex-col items-center justify-center gap-1 px-6 text-center"
99
+ >
100
+ <p className="text-sm font-medium">No preview for this file</p>
101
+ <p className="max-w-sm text-xs leading-relaxed text-muted-foreground">
102
+ {error} Save it or open it in another app.
103
+ </p>
104
+ </div>
105
+ );
106
+ if (!prepared) return <FileSurfaceLoading />;
107
+ const Viewer =
108
+ props.kind === "docx" ? WordDocument : props.kind === "xlsx" ? Spreadsheet : Presentation;
109
+ return (
110
+ <div className="flex min-h-0 flex-1 flex-col">
111
+ {details ? (
112
+ <div className={FILE_SURFACE_SUBHEADER_CLASS}>
113
+ <span className="text-xs text-muted-foreground" aria-live="polite">
114
+ {details}
115
+ {props.kind !== "xlsx" ? " · Fit to width" : ""}
116
+ </span>
117
+ </div>
118
+ ) : null}
119
+ {prepared.truncated ? (
120
+ <FileSurfaceNotice>
121
+ Preview limited. Save it or open it in another app to read the full document.
122
+ </FileSurfaceNotice>
123
+ ) : null}
124
+ <Suspense fallback={<FileSurfaceLoading />}>
125
+ <Viewer bytes={prepared.bytes} onDetails={setDetails} onError={fail} />
126
+ </Suspense>
127
+ </div>
128
+ );
129
+}
apps/web/src/components/files/PdfViewer.tsx
new
+431
@@ -0,0 +1,431 @@
1
+import {
2
+ useEffect,
3
+ useEffectEvent,
4
+ useLayoutEffect,
5
+ useRef,
6
+ useState,
7
+ type CSSProperties,
8
+} from "react";
9
+import { TextLayer, type PDFDocumentProxy } from "pdfjs-dist";
10
+import { ChevronDownIcon, ChevronUpIcon, MinusIcon, PlusIcon, ScanIcon } from "lucide-react";
11
+import { Input } from "~/components/ui/input";
12
+import { observeResize } from "~/lib/observeResize";
13
+import { FileSurfaceAction, FileSurfaceFailure, FileSurfaceLoading } from "./fileSurfaceChrome";
14
+import { readBoundedDocument } from "./readBoundedDocument";
15
+import { openPdf } from "./pdfDocument";
16
+import "./pdfViewer.css";
17
+
18
+const MAX_PAGES = 2000;
19
+const MAX_FIND_TEXT = 4 * 1024 * 1024;
20
+
21
+function PdfPage(props: {
22
+ document: PDFDocumentProxy;
23
+ index: number;
24
+ width: number;
25
+ query: string;
26
+ onSize: (index: number, ratio: number) => void;
27
+ onError: (error: string) => void;
28
+}) {
29
+ const canvas = useRef<HTMLCanvasElement>(null);
30
+ const text = useRef<HTMLDivElement>(null);
31
+ const { document: pdf, index, width, onSize, onError } = props;
32
+ // oxlint-disable-next-line t3code/react-effect-budget -- Render the PDF page canvas and text layer, cancelling tasks and releasing pdf.js page resources on cleanup.
33
+ useEffect(() => {
34
+ let cancelled = false;
35
+ let paint: ReturnType<Awaited<ReturnType<PDFDocumentProxy["getPage"]>>["render"]> | undefined;
36
+ let layer: TextLayer | undefined;
37
+ let loadedPage: Awaited<ReturnType<PDFDocumentProxy["getPage"]>> | undefined;
38
+ void (async () => {
39
+ const page = await pdf.getPage(index + 1);
40
+ loadedPage = page;
41
+ if (cancelled || !canvas.current || !text.current) {
42
+ page.cleanup();
43
+ return;
44
+ }
45
+ const natural = page.getViewport({ scale: 1 });
46
+ onSize(index, natural.height / natural.width);
47
+ const viewport = page.getViewport({ scale: width / natural.width });
48
+ const ratio = Math.min(
49
+ 2,
50
+ window.devicePixelRatio || 1,
51
+ Math.sqrt((8 * 1024 * 1024) / (viewport.width * viewport.height)),
52
+ );
53
+ canvas.current.width = Math.ceil(viewport.width * ratio);
54
+ canvas.current.height = Math.ceil(viewport.height * ratio);
55
+ paint = page.render({
56
+ canvas: canvas.current,
57
+ viewport,
58
+ transform: [ratio, 0, 0, ratio, 0, 0],
59
+ background: "white",
60
+ });
61
+ text.current.replaceChildren();
62
+ text.current.style.setProperty("--total-scale-factor", String(viewport.scale));
63
+ layer = new TextLayer({
64
+ container: text.current,
65
+ viewport,
66
+ textContentSource: page.streamTextContent(),
67
+ });
68
+ await Promise.all([paint.promise, layer.render()]);
69
+ })().catch((cause: unknown) => {
70
+ if (!cancelled)
71
+ onError(cause instanceof Error ? cause.message : "This PDF page could not render.");
72
+ });
73
+ return () => {
74
+ cancelled = true;
75
+ paint?.cancel();
76
+ layer?.cancel();
77
+ // Release operator lists when virtual pages leave the viewport.
78
+ void Promise.allSettled([paint?.promise]).then(() => loadedPage?.cleanup());
79
+ };
80
+ }, [pdf, index, width, onSize, onError]);
81
+ // oxlint-disable-next-line t3code/react-effect-budget -- Highlight pdf.js-owned text nodes and observe asynchronous text layer mutations for the current search query.
82
+ useEffect(() => {
83
+ const node = text.current;
84
+ if (!node) return;
85
+ const mark = () => {
86
+ const runs = Array.from(node.querySelectorAll('span[role="presentation"], br')).map(
87
+ (element) => ({
88
+ element,
89
+ text: element.tagName === "BR" ? "\n" : (element.textContent ?? "").toLowerCase(),
90
+ }),
91
+ );
92
+ const text = runs.map((run) => run.text).join("");
93
+ const query = props.query.toLowerCase();
94
+ let match = query.trim() ? text.indexOf(query) : -1;
95
+ let offset = 0;
96
+ for (const run of runs) {
97
+ while (match !== -1 && match + query.length <= offset)
98
+ match = text.indexOf(query, match + query.length);
99
+ run.element.classList.toggle(
100
+ "pdf-find-match",
101
+ match !== -1 && match < offset + run.text.length,
102
+ );
103
+ offset += run.text.length;
104
+ }
105
+ };
106
+ mark();
107
+ const observer = new MutationObserver(mark);
108
+ observer.observe(node, { childList: true, subtree: true });
109
+ return () => observer.disconnect();
110
+ }, [props.query]);
111
+ return (
112
+ <>
113
+ <canvas ref={canvas} aria-label={`Page ${index + 1}`} className="block h-full w-full" />
114
+ <div ref={text} data-pdf-text-layer />
115
+ </>
116
+ );
117
+}
118
+
119
+export default function PdfViewer(props: { src: string; title: string }) {
120
+ const [pdf, setPdf] = useState<PDFDocumentProxy | null>(null);
121
+ const [error, setError] = useState<string | null>(null);
122
+ const [revision, setRevision] = useState(0);
123
+ const [ratios, setRatios] = useState<number[]>([]);
124
+ const [zoom, setZoom] = useState<number | null>(null);
125
+ const [pageWidth, setPageWidth] = useState(612);
126
+ const [viewport, setViewport] = useState({ width: 600, height: 800 });
127
+ const [scrollTop, setScrollTop] = useState(0);
128
+ const [query, setQuery] = useState("");
129
+ const [matches, setMatches] = useState<number[]>([]);
130
+ const [finding, setFinding] = useState(false);
131
+ const [findError, setFindError] = useState<string | null>(null);
132
+ const [match, setMatch] = useState(0);
133
+ const scroll = useRef<HTMLDivElement>(null);
134
+ const find = useRef<HTMLInputElement>(null);
135
+ const zoomAnchor = useRef<{ page: number; offset: number; width: number } | null>(null);
136
+ // oxlint-disable-next-line t3code/react-effect-budget -- Own the PDF fetch, load timeout, and pdf.js document task across source changes, retries, and unmounts.
137
+ useEffect(() => {
138
+ const controller = new AbortController();
139
+ let disposed = false;
140
+ let task: ReturnType<typeof openPdf> | undefined;
141
+ const timeout = setTimeout(() => {
142
+ setError("This PDF took too long to load. Try again.");
143
+ controller.abort();
144
+ void task?.destroy();
145
+ }, 30_000);
146
+ void readBoundedDocument(
147
+ props.src,
148
+ controller.signal,
149
+ 50 * 1024 * 1024,
150
+ revision > 0 ? "reload" : "default",
151
+ )
152
+ .then(async (bytes) => {
153
+ controller.signal.throwIfAborted();
154
+ task = openPdf(bytes);
155
+ const document = await task.promise;
156
+ if (document.numPages > MAX_PAGES)
157
+ throw new Error("Preview limited to 2,000 pages. Save this PDF to read it in full.");
158
+ const first = (await document.getPage(1)).getViewport({ scale: 1 });
159
+ if (controller.signal.aborted) return;
160
+ setPageWidth(first.width);
161
+ setRatios(Array.from({ length: document.numPages }, () => first.height / first.width));
162
+ setPdf(document);
163
+ })
164
+ .catch((cause: unknown) => {
165
+ if (!disposed && !controller.signal.aborted)
166
+ setError(cause instanceof Error ? cause.message : "Unable to open PDF.");
167
+ })
168
+ .finally(() => clearTimeout(timeout));
169
+ return () => {
170
+ disposed = true;
171
+ controller.abort();
172
+ clearTimeout(timeout);
173
+ void task?.destroy();
174
+ };
175
+ }, [props.src, revision]);
176
+ // oxlint-disable-next-line t3code/react-effect-budget -- Measure the PDF viewport before paint and observe resizes for fit-width sizing and visible-page selection.
177
+ useLayoutEffect(() => {
178
+ const node = scroll.current;
179
+ if (!node) return;
180
+ const update = () => setViewport({ width: node.clientWidth, height: node.clientHeight });
181
+ update();
182
+ return observeResize(node, update);
183
+ }, []);
184
+ const fitWidth = Math.max(1, viewport.width - 32);
185
+ const scale = zoom ?? fitWidth / pageWidth;
186
+ const width = zoom === null ? fitWidth : pageWidth * zoom;
187
+ const offsets = (() => {
188
+ const result = [16];
189
+ for (const ratio of ratios) result.push(result[result.length - 1]! + width * ratio + 16);
190
+ return result;
191
+ })();
192
+ const current = Math.max(
193
+ 0,
194
+ ratios.findIndex((_, index) => offsets[index + 1]! > scrollTop + 17),
195
+ );
196
+ const visible = ratios.flatMap((ratio, index) =>
197
+ offsets[index + 1]! >= scrollTop - viewport.height &&
198
+ offsets[index]! <= scrollTop + viewport.height * 2
199
+ ? [{ index, ratio }]
200
+ : [],
201
+ );
202
+ // Stable callbacks let page canvases survive scroll and find updates.
203
+ const [onSize] = useState(
204
+ () => (index: number, ratio: number) =>
205
+ setRatios((old) =>
206
+ old[index] === ratio ? old : old.map((value, i) => (i === index ? ratio : value)),
207
+ ),
208
+ );
209
+ const jumpToPage = useEffectEvent((index: number) => {
210
+ scroll.current?.scrollTo({ top: Math.max(0, (offsets[index] ?? 16) - 16) });
211
+ });
212
+ const showPage = (index: number) => {
213
+ scroll.current?.scrollTo({ top: Math.max(0, (offsets[index] ?? 16) - 16) });
214
+ };
215
+ const changeZoom = (value: number | null) => {
216
+ zoomAnchor.current = {
217
+ page: current,
218
+ offset: (scroll.current?.scrollTop ?? 0) - (offsets[current]! - 16),
219
+ width,
220
+ };
221
+ setZoom(value);
222
+ };
223
+ // oxlint-disable-next-line t3code/react-effect-budget -- Restore the DOM scroll anchor after the new zoom width has committed to page layout.
224
+ useLayoutEffect(() => {
225
+ const anchor = zoomAnchor.current;
226
+ zoomAnchor.current = null;
227
+ if (!anchor || anchor.width === width) return;
228
+ scroll.current?.scrollTo({
229
+ top: offsets[anchor.page]! - 16 + (anchor.offset * width) / anchor.width,
230
+ });
231
+ });
232
+ // oxlint-disable-next-line t3code/react-effect-budget -- Scroll to the first asynchronous search result after its measured page offsets have committed.
233
+ useLayoutEffect(() => {
234
+ if (matches.length) jumpToPage(matches[0]!);
235
+ }, [matches]);
236
+ // oxlint-disable-next-line t3code/react-effect-budget -- Debounce pdf.js text extraction and cancel obsolete searches when the query or document changes.
237
+ useEffect(() => {
238
+ if (!pdf || !query.trim()) {
239
+ // oxlint-disable-next-line react/set-state-in-effect -- A blank query cancels the previous search.
240
+ setFinding(false);
241
+ return;
242
+ }
243
+ let cancelled = false;
244
+ const timer = setTimeout(() => {
245
+ setFinding(true);
246
+ setFindError(null);
247
+ setMatches([]);
248
+ setMatch(0);
249
+ void (async () => {
250
+ const found: number[] = [];
251
+ const sizes: number[] = [];
252
+ let size = 0;
253
+ for (let index = 0; index < pdf.numPages; index++) {
254
+ if (cancelled) return;
255
+ const page = await pdf.getPage(index + 1);
256
+ const pageSize = page.getViewport({ scale: 1 });
257
+ sizes.push(pageSize.height / pageSize.width);
258
+ const content = await page.getTextContent();
259
+ const text = content.items
260
+ .map((item) => ("str" in item ? item.str + (item.hasEOL ? "\n" : "") : ""))
261
+ .join("");
262
+ size += text.length;
263
+ if (size > MAX_FIND_TEXT)
264
+ throw new Error(
265
+ "This PDF has too much text to search. Save it to search in another app.",
266
+ );
267
+ if (text.toLowerCase().includes(query.toLowerCase())) found.push(index);
268
+ }
269
+ if (cancelled) return;
270
+ setRatios(sizes);
271
+ setMatches(found);
272
+ })()
273
+ .catch((cause: unknown) => {
274
+ if (!cancelled)
275
+ setFindError(cause instanceof Error ? cause.message : "Could not search this PDF.");
276
+ })
277
+ .finally(() => {
278
+ if (!cancelled) setFinding(false);
279
+ });
280
+ }, 250);
281
+ return () => {
282
+ cancelled = true;
283
+ clearTimeout(timer);
284
+ };
285
+ }, [pdf, query]);
286
+ return (
287
+ <div className="flex min-h-0 min-w-0 flex-1 flex-col overflow-hidden">
288
+ <div className="flex min-h-10 shrink-0 flex-wrap items-center gap-1 border-b border-border/60 px-2 py-1 text-xs">
289
+ <span className="mr-auto px-1 tabular-nums" aria-live="polite">
290
+ {pdf ? `${current + 1} / ${pdf.numPages} pages` : "PDF"}
291
+ </span>
292
+ <FileSurfaceAction
293
+ label="Previous page"
294
+ disabled={!pdf || current === 0}
295
+ onPress={() => showPage(current - 1)}
296
+ >
297
+ <ChevronUpIcon className="size-3.5" />
298
+ </FileSurfaceAction>
299
+ <FileSurfaceAction
300
+ label="Next page"
301
+ disabled={!pdf || current + 1 >= pdf.numPages}
302
+ onPress={() => showPage(current + 1)}
303
+ >
304
+ <ChevronDownIcon className="size-3.5" />
305
+ </FileSurfaceAction>
306
+ <FileSurfaceAction
307
+ label="Zoom out"
308
+ disabled={scale <= 0.25}
309
+ onPress={() => changeZoom(Math.max(0.25, scale - 0.25))}
310
+ >
311
+ <MinusIcon className="size-3.5" />
312
+ </FileSurfaceAction>
313
+ <span className="min-w-14 text-center tabular-nums">
314
+ {zoom === null ? "Fit width" : `${Math.round(scale * 100)}%`}
315
+ </span>
316
+ <FileSurfaceAction
317
+ label="Zoom in"
318
+ disabled={scale >= 3}
319
+ onPress={() => changeZoom(Math.min(3, scale + 0.25))}
320
+ >
321
+ <PlusIcon className="size-3.5" />
322
+ </FileSurfaceAction>
323
+ <FileSurfaceAction
324
+ label="Fit width"
325
+ onPress={() => {
326
+ changeZoom(null);
327
+ scroll.current?.scrollTo({ left: 0 });
328
+ }}
329
+ >
330
+ <ScanIcon className="size-3.5" />
331
+ </FileSurfaceAction>
332
+ <div className="w-28">
333
+ <Input
334
+ ref={find}
335
+ size="compact"
336
+ type="search"
337
+ aria-label="Find in PDF"
338
+ placeholder="Find in PDF"
339
+ value={query}
340
+ onChange={(event) => {
341
+ setQuery(event.target.value);
342
+ setMatches([]);
343
+ setFindError(null);
344
+ }}
345
+ />
346
+ </div>
347
+ {query ? (
348
+ <>
349
+ <span role="status">
350
+ {finding
351
+ ? "Searching…"
352
+ : (findError ?? `${matches.length ? match + 1 : 0}/${matches.length}`)}
353
+ </span>
354
+ <FileSurfaceAction
355
+ label="Next matching page"
356
+ disabled={!matches.length}
357
+ onPress={() => {
358
+ const next = (match + 1) % matches.length;
359
+ setMatch(next);
360
+ showPage(matches[next]!);
361
+ }}
362
+ >
363
+ <ChevronDownIcon className="size-3.5" />
364
+ </FileSurfaceAction>
365
+ </>
366
+ ) : null}
367
+ </div>
368
+ <div
369
+ ref={scroll}
370
+ role="region"
371
+ aria-label={props.title}
372
+ tabIndex={0}
373
+ className="relative min-h-0 min-w-0 w-full flex-1 overflow-auto bg-muted/40"
374
+ onScroll={(event) => setScrollTop(event.currentTarget.scrollTop)}
375
+ onKeyDown={(event) => {
376
+ if ((event.ctrlKey || event.metaKey) && event.key === "f") {
377
+ event.preventDefault();
378
+ event.stopPropagation();
379
+ find.current?.focus();
380
+ }
381
+ }}
382
+ >
383
+ {error ? (
384
+ <FileSurfaceFailure
385
+ message={error}
386
+ onRetry={() => {
387
+ setError(null);
388
+ setPdf(null);
389
+ setRevision((value) => value + 1);
390
+ }}
391
+ />
392
+ ) : !pdf ? (
393
+ <FileSurfaceLoading />
394
+ ) : (
395
+ <div
396
+ className="relative"
397
+ style={{
398
+ height: `max(${offsets.at(-1) ?? 0}px, calc(100% + ${(offsets.at(-2) ?? 16) - 16}px))`,
399
+ minWidth: width + 32,
400
+ }}
401
+ >
402
+ {visible.map(({ index, ratio }) => (
403
+ <div
404
+ key={index}
405
+ className="absolute bg-white shadow-sm"
406
+ style={
407
+ {
408
+ top: offsets[index],
409
+ left: Math.max(16, (viewport.width - width) / 2),
410
+ width,
411
+ height: width * ratio,
412
+ "--total-scale-factor": 1,
413
+ } as CSSProperties
414
+ }
415
+ >
416
+ <PdfPage
417
+ document={pdf}
418
+ index={index}
419
+ width={width}
420
+ query={query}
421
+ onSize={onSize}
422
+ onError={setError}
423
+ />
424
+ </div>
425
+ ))}
426
+ </div>
427
+ )}
428
+ </div>
429
+ </div>
430
+ );
431
+}
apps/web/src/components/files/extendedDocumentPreview.ts
new
+672
@@ -0,0 +1,672 @@
1
+import {
2
+ IMAGE_DIMENSIONS_HEADER_BYTES,
3
+ readImageDimensions,
4
+} from "@t3tools/shared/imageDimensions";
5
+import { XMLParser, XMLValidator } from "fast-xml-parser";
6
+import { decodeHTMLAttribute, decodeHTMLStrict as decodeEntities } from "entities";
7
+import type JSZip from "jszip";
8
+
9
+import { type OfficeWorkerResult, type NotebookCell } from "./officePreviewProtocol";
10
+
11
+const ODS_PREVIEW_MAX_ROWS = 500;
12
+const ODS_PREVIEW_MAX_COLUMNS = 50;
13
+
14
+type XmlNode = { name: string; attrs: Record<string, string>; children: XmlNode[]; text: string };
15
+type OrderedXml = { [key: string]: OrderedXml[] | Record<string, string> | string };
16
+const escape = (value: string) =>
17
+ value
18
+ .replaceAll("&", "&")
19
+ .replaceAll("<", "<")
20
+ .replaceAll(">", ">")
21
+ .replaceAll('"', """);
22
+
23
+const namespacePrefixes: Record<string, string> = {
24
+ "urn:oasis:names:tc:opendocument:xmlns:office:1.0": "office",
25
+ "urn:oasis:names:tc:opendocument:xmlns:table:1.0": "table",
26
+ "urn:oasis:names:tc:opendocument:xmlns:text:1.0": "text",
27
+ "urn:oasis:names:tc:opendocument:xmlns:drawing:1.0": "draw",
28
+ "http://www.w3.org/1999/xlink": "xlink",
29
+ "http://www.w3.org/1999/xhtml": "",
30
+ "http://www.idpf.org/2007/opf": "",
31
+ "urn:oasis:names:tc:opendocument:xmlns:container": "",
32
+};
33
+function parseXml(source: string): XmlNode {
34
+ // EPUB XHTML commonly declares an external DTD. Discard it without resolving it.
35
+ source = source.replaceAll(/<!DOCTYPE\s+[^<>[\]]*>/gi, "");
36
+ if (
37
+ source.length > 8 * 1024 * 1024 ||
38
+ /<!\s*(DOCTYPE|ENTITY)/i.test(source) ||
39
+ XMLValidator.validate(source) !== true
40
+ )
41
+ throw new Error("This document contains invalid or unsupported XML.");
42
+ const parser = new XMLParser({
43
+ preserveOrder: true,
44
+ ignoreAttributes: false,
45
+ attributeNamePrefix: "",
46
+ processEntities: false,
47
+ trimValues: false,
48
+ parseTagValue: false,
49
+ });
50
+ const ordered = parser.parse(source) as OrderedXml[];
51
+ let count = 0;
52
+ function convert(
53
+ items: OrderedXml[],
54
+ depth: number,
55
+ inherited: Record<string, string>,
56
+ ): XmlNode[] {
57
+ if (depth > 100) throw new Error("This document is nested too deeply to preview.");
58
+ return items.map((item) => {
59
+ if (++count > 100_000) throw new Error("This document has too much content to preview.");
60
+ const name = Object.keys(item).find((key) => key !== ":@") ?? "#text";
61
+ const value = item[name];
62
+ const attrs = Object.fromEntries(
63
+ Object.entries((item[":@"] ?? {}) as Record<string, string>).map(([key, value]) => [
64
+ key,
65
+ decodeEntities(value),
66
+ ]),
67
+ );
68
+ const namespaces = Object.create(inherited) as Record<string, string>;
69
+ for (const [key, value] of Object.entries(attrs)) {
70
+ if (key === "xmlns") namespaces[""] = value;
71
+ else if (key.startsWith("xmlns:")) namespaces[key.slice(6)] = value;
72
+ }
73
+ const expanded = (name: string, attribute = false) => {
74
+ if (name.startsWith("#")) return name;
75
+ const colon = name.indexOf(":");
76
+ if (colon < 0 && attribute) return name;
77
+ const uri = namespaces[colon < 0 ? "" : name.slice(0, colon)];
78
+ if (!uri) return name;
79
+ const prefix = Object.hasOwn(namespacePrefixes, uri) ? namespacePrefixes[uri] : undefined;
80
+ const local = name.slice(colon + 1);
81
+ return prefix === undefined
82
+ ? "{" + uri + "}" + local
83
+ : prefix
84
+ ? prefix + ":" + local
85
+ : local;
86
+ };
87
+ return {
88
+ name: expanded(name),
89
+ attrs: Object.fromEntries(
90
+ Object.entries(attrs).map(([key, value]) => [expanded(key, true), value]),
91
+ ),
92
+ children: Array.isArray(value) ? convert(value, depth + 1, namespaces) : [],
93
+ text: typeof value === "string" ? value : "",
94
+ };
95
+ });
96
+ }
97
+ return {
98
+ name: "root",
99
+ attrs: {},
100
+ children: convert(ordered, 0, Object.create(null) as Record<string, string>),
101
+ text: "",
102
+ };
103
+}
104
+function find(node: XmlNode, name: string, skip?: string): XmlNode[] {
105
+ return node.children.flatMap((child) =>
106
+ child.name === skip
107
+ ? []
108
+ : child.name === name || (!name.includes(":") && child.name.endsWith(`:${name}`))
109
+ ? [child]
110
+ : find(child, name, skip),
111
+ );
112
+}
113
+function text(node: XmlNode): string {
114
+ if (node.name === "office:annotation") return "";
115
+ if (node.name === "text:s") return " ".repeat(repeat(node.attrs["text:c"], 100));
116
+ if (node.name === "text:tab") return "\t";
117
+ if (node.name === "text:line-break") return "\n";
118
+ return node.name === "#text" ? decodeEntities(node.text) : node.children.map(text).join("");
119
+}
120
+function archivePath(base: string, href: string): string | null {
121
+ if (/^(?:[a-z][a-z\d+.-]*:|\/|\\)/i.test(href)) return null;
122
+ const parts = base.split("/").slice(0, -1);
123
+ let decoded: string;
124
+ try {
125
+ decoded = decodeURIComponent(href.split(/[?#]/, 1)[0] ?? "");
126
+ } catch {
127
+ return null;
128
+ }
129
+ for (const part of decoded.split("/")) {
130
+ if (part === "..") {
131
+ if (!parts.length) return null;
132
+ parts.pop();
133
+ } else if (part && part !== ".") parts.push(part);
134
+ }
135
+ return parts.join("/");
136
+}
137
+function imageFits(bytes: Uint8Array) {
138
+ const size = readImageDimensions(bytes.subarray(0, IMAGE_DIMENSIONS_HEADER_BYTES));
139
+ return size !== null && size.width <= 16_777_216 / size.height;
140
+}
141
+function base64ImageFits(base64: string) {
142
+ return (
143
+ /^[a-z\d+/]*={0,2}$/i.test(base64) &&
144
+ base64.length % 4 !== 1 &&
145
+ (!base64.endsWith("=") || base64.length % 4 === 0) &&
146
+ imageFits(
147
+ Uint8Array.from(
148
+ atob(base64.slice(0, 4 * Math.ceil(IMAGE_DIMENSIONS_HEADER_BYTES / 3))),
149
+ (char) => char.charCodeAt(0),
150
+ ),
151
+ )
152
+ );
153
+}
154
+function htmlImagesFit(html: string) {
155
+ // HTML attribute entities can obscure a data URL until the client parses it.
156
+ for (const match of decodeHTMLAttribute(html).matchAll(
157
+ /data:image\/(?:png|jpeg|gif|webp);base64,([a-z\d+/]*={0,2})/gi,
158
+ ))
159
+ if (!base64ImageFits(match[1]!)) return false;
160
+ return true;
161
+}
162
+async function imageData(archive: JSZip, path: string | null): Promise<string> {
163
+ if (!path) return "";
164
+ const mime = (
165
+ { png: "png", jpg: "jpeg", jpeg: "jpeg", gif: "gif", webp: "webp" } as Record<string, string>
166
+ )[path.split(".").at(-1)?.toLowerCase() ?? ""];
167
+ const entry = archive.file(path);
168
+ if (!mime || !entry) return "";
169
+ const bytes = await entry.async("uint8array");
170
+ if (bytes.length > 4 * 1024 * 1024 || !imageFits(bytes)) return "";
171
+ reserveMarkup(4 * Math.ceil(bytes.length / 3) + 100);
172
+ return `data:image/${mime};base64,${await entry.async("base64")}`;
173
+}
174
+let markupBytes = 0;
175
+function reserveMarkup(size: number) {
176
+ markupBytes += size;
177
+ if (markupBytes > 16 * 1024 * 1024)
178
+ throw new Error("The rendered document is too large to preview.");
179
+}
180
+function boundedMarkup(value: string): string {
181
+ reserveMarkup(value.length);
182
+ return value;
183
+}
184
+const HTML_TAGS = new Set([
185
+ "p",
186
+ "h1",
187
+ "h2",
188
+ "h3",
189
+ "h4",
190
+ "h5",
191
+ "h6",
192
+ "div",
193
+ "span",
194
+ "strong",
195
+ "b",
196
+ "em",
197
+ "i",
198
+ "u",
199
+ "s",
200
+ "ul",
201
+ "ol",
202
+ "li",
203
+ "blockquote",
204
+ "pre",
205
+ "code",
206
+ "table",
207
+ "thead",
208
+ "tbody",
209
+ "tr",
210
+ "th",
211
+ "td",
212
+ "caption",
213
+ "figure",
214
+ "figcaption",
215
+ "br",
216
+ "hr",
217
+ "sub",
218
+ "sup",
219
+ "dl",
220
+ "dt",
221
+ "dd",
222
+]);
223
+async function epubHtml(node: XmlNode, archive: JSZip, base: string): Promise<string> {
224
+ if (node.name === "#text") return boundedMarkup(escape(text(node)));
225
+ const tag =
226
+ node.name
227
+ .replace(/^\{[^}]*\}/, "")
228
+ .split(":")
229
+ .at(-1) ?? "";
230
+ if (["head", "script", "style", "iframe", "object", "embed", "svg", "math", "form"].includes(tag))
231
+ return "";
232
+ if (tag === "br" || tag === "hr") return boundedMarkup(`<${tag}>`);
233
+ if (tag === "img") {
234
+ const src = await imageData(archive, archivePath(base, node.attrs.src ?? ""));
235
+ return src ? `<img src="${src}" alt="${boundedMarkup(escape(node.attrs.alt ?? ""))}">` : "";
236
+ }
237
+ let children = "";
238
+ for (const child of node.children) children += await epubHtml(child, archive, base);
239
+ if (!HTML_TAGS.has(tag)) return children;
240
+ const spans =
241
+ tag === "td" || tag === "th"
242
+ ? ["colspan", "rowspan"]
243
+ .map((name) =>
244
+ /^\d{1,4}$/.test(node.attrs[name] ?? "") ? ` ${name}="${node.attrs[name]}"` : "",
245
+ )
246
+ .join("")
247
+ : "";
248
+ reserveMarkup(tag.length * 2 + spans.length + 5);
249
+ return `<${tag}${spans}>${children}</${tag}>`;
250
+}
251
+async function odfHtml(node: XmlNode, archive: JSZip): Promise<string> {
252
+ if (node.name === "#text") return boundedMarkup(escape(text(node)));
253
+ if (node.name === "text:s" || node.name === "text:tab")
254
+ return boundedMarkup(
255
+ `<span data-odf-whitespace>${node.name === "text:tab" ? "	" : " ".repeat(repeat(node.attrs["text:c"], 100))}</span>`,
256
+ );
257
+ if (node.name === "text:line-break") return boundedMarkup("<br>");
258
+ if (node.name === "draw:image") {
259
+ const src = await imageData(
260
+ archive,
261
+ archivePath("content.xml", node.attrs["xlink:href"] ?? ""),
262
+ );
263
+ return src ? `<img src="${src}" alt="">` : "";
264
+ }
265
+ if (
266
+ [
267
+ "office:annotation",
268
+ "text:tracked-changes",
269
+ "office:scripts",
270
+ "draw:object",
271
+ "draw:object-ole",
272
+ ].includes(node.name)
273
+ )
274
+ return "";
275
+ let children = "";
276
+ for (const child of node.children) children += await odfHtml(child, archive);
277
+ const tags: Record<string, string> = {
278
+ "text:p": "p",
279
+ "text:h": "h2",
280
+ "text:span": "span",
281
+ "text:list": "ul",
282
+ "text:list-item": "li",
283
+ "table:table": "table",
284
+ "table:table-row": "tr",
285
+ "table:table-cell": "td",
286
+ };
287
+ const tag = tags[node.name];
288
+ if (!tag) return children;
289
+ reserveMarkup(tag.length * 2 + 5);
290
+ return `<${tag}>${children}</${tag}>`;
291
+}
292
+function repeat(value: string | undefined, max: number) {
293
+ const count = Number(value ?? 1);
294
+ if (!Number.isSafeInteger(count) || count < 1)
295
+ throw new Error("This spreadsheet contains invalid repeated cells.");
296
+ return Math.min(count, max);
297
+}
298
+export async function openDocument(
299
+ archive: JSZip,
300
+ kind: "odt" | "ods" | "odp",
301
+): Promise<OfficeWorkerResult> {
302
+ markupBytes = 0;
303
+ const source = await archive.file("content.xml")?.async("string");
304
+ if (!source) throw new Error("This OpenDocument file has no readable content.");
305
+ const root = parseXml(source);
306
+ if (kind === "ods") {
307
+ const tables = find(root, "table:table");
308
+ if (!tables.length || tables.length > 100)
309
+ throw new Error("This workbook has no sheets or too many sheets to preview.");
310
+ let cellBytes = 0;
311
+ return {
312
+ type: "ods",
313
+ sheets: tables.map((table, index) => {
314
+ const rows: string[][] = [];
315
+ let truncated = false;
316
+ let pendingEmptyRows = 0;
317
+ for (const row of find(table, "table:table-row")) {
318
+ const cells: string[] = [];
319
+ for (const cell of row.children.filter((child) =>
320
+ ["table:table-cell", "table:covered-table-cell"].includes(child.name),
321
+ )) {
322
+ const value =
323
+ find(cell, "text:p", "office:annotation").map(text).join("\n") ||
324
+ cell.attrs["office:value"] ||
325
+ cell.attrs["office:date-value"] ||
326
+ cell.attrs["office:boolean-value"] ||
327
+ cell.attrs["office:string-value"] ||
328
+ "";
329
+ const count = repeat(
330
+ cell.attrs["table:number-columns-repeated"],
331
+ ODS_PREVIEW_MAX_COLUMNS + 1,
332
+ );
333
+ if (value && cells.length + count > ODS_PREVIEW_MAX_COLUMNS) truncated = true;
334
+ for (let i = 0; i < count && cells.length < ODS_PREVIEW_MAX_COLUMNS; i++) {
335
+ if (value.length > 10_000)
336
+ throw new Error("A spreadsheet cell is too large to preview.");
337
+ cells.push(value);
338
+ }
339
+ }
340
+ while (cells.at(-1) === "") cells.pop();
341
+ const count = repeat(row.attrs["table:number-rows-repeated"], ODS_PREVIEW_MAX_ROWS + 1);
342
+ if (!cells.length) {
343
+ pendingEmptyRows = Math.min(ODS_PREVIEW_MAX_ROWS + 1, pendingEmptyRows + count);
344
+ continue;
345
+ }
346
+ while (pendingEmptyRows > 0 && rows.length < ODS_PREVIEW_MAX_ROWS) {
347
+ rows.push([]);
348
+ pendingEmptyRows--;
349
+ }
350
+ pendingEmptyRows = 0;
351
+ if (rows.length + count > ODS_PREVIEW_MAX_ROWS) truncated = true;
352
+ if (rows.length >= ODS_PREVIEW_MAX_ROWS) break;
353
+ for (let i = 0; i < count && rows.length < ODS_PREVIEW_MAX_ROWS; i++) {
354
+ cellBytes += cells.reduce((size, cell) => size + cell.length, 0);
355
+ if (cellBytes > 8 * 1024 * 1024)
356
+ throw new Error("This spreadsheet has too much text to preview.");
357
+ rows.push(cells);
358
+ }
359
+ }
360
+ return {
361
+ name: table.attrs["table:name"] || `Sheet ${index + 1}`,
362
+ rows,
363
+ rowCount: rows.length,
364
+ columns: rows.reduce((columns, row) => Math.max(columns, row.length), 1),
365
+ truncated,
366
+ };
367
+ }),
368
+ };
369
+ }
370
+ const pages = kind === "odp" ? find(root, "draw:page") : find(root, "office:text");
371
+ if (!pages.length || pages.length > 500)
372
+ throw new Error("This document has no pages or too many pages to preview.");
373
+ const rendered = [];
374
+ for (const [index, page] of pages.entries()) {
375
+ rendered.push({
376
+ title: page.attrs["draw:name"] || `Page ${index + 1}`,
377
+ html: await odfHtml(page, archive),
378
+ });
379
+ }
380
+ return { type: "document", navigation: kind === "odp" ? "slide" : "page", pages: rendered };
381
+}
382
+export async function openEpub(archive: JSZip): Promise<OfficeWorkerResult> {
383
+ markupBytes = 0;
384
+ const container = parseXml((await archive.file("META-INF/container.xml")?.async("string")) ?? "");
385
+ const rootfile = find(container, "rootfile")[0]?.attrs["full-path"];
386
+ if (!rootfile || !archive.file(rootfile)) throw new Error("This EPUB has no readable package.");
387
+ const packageRoot = parseXml(await archive.file(rootfile)!.async("string"));
388
+ const manifest = new Map(find(packageRoot, "item").map((item) => [item.attrs.id, item]));
389
+ const spine = find(packageRoot, "itemref");
390
+ if (!spine.length || spine.length > 500)
391
+ throw new Error("This EPUB has no chapters or too many chapters to preview.");
392
+ const pages = [];
393
+ for (const itemref of spine) {
394
+ const item = manifest.get(itemref.attrs.idref);
395
+ if (!item || item.attrs["media-type"] !== "application/xhtml+xml") continue;
396
+ const path = archivePath(rootfile, item.attrs.href ?? "");
397
+ const entry = path ? archive.file(path) : null;
398
+ if (!entry || !path) continue;
399
+ const chapter = parseXml(await entry.async("string"));
400
+ pages.push({
401
+ title:
402
+ text(find(chapter, "title")[0] ?? find(chapter, "h1")[0] ?? chapter).slice(0, 120) ||
403
+ `Chapter ${pages.length + 1}`,
404
+ html: await epubHtml(chapter, archive, path),
405
+ });
406
+ }
407
+ if (!pages.length) throw new Error("This EPUB has no readable XHTML chapters.");
408
+ return { type: "document", navigation: "chapter", pages };
409
+}
410
+function record(value: unknown): Record<string, unknown> {
411
+ return value !== null && typeof value === "object" && !Array.isArray(value)
412
+ ? (value as Record<string, unknown>)
413
+ : {};
414
+}
415
+function notebookText(value: unknown, limit = 256_000): string {
416
+ const result =
417
+ typeof value === "string"
418
+ ? value
419
+ : Array.isArray(value)
420
+ ? value.filter((part) => typeof part === "string").join("")
421
+ : "";
422
+ if (result.length > limit) throw new Error("A notebook cell is too large to preview.");
423
+ return result;
424
+}
425
+export function openNotebook(bytes: ArrayBuffer): OfficeWorkerResult {
426
+ if (bytes.byteLength > 8 * 1024 * 1024)
427
+ throw new Error("Notebook preview limited to files under 8 MB.");
428
+ const notebook = record(JSON.parse(new TextDecoder("utf-8", { fatal: true }).decode(bytes)));
429
+ if (notebook.nbformat !== 4 || !Array.isArray(notebook.cells) || notebook.cells.length > 500)
430
+ throw new Error("Preview supports version 4 notebooks with up to 500 cells.");
431
+ const language = String(record(record(notebook.metadata).language_info).name ?? "text")
432
+ .replaceAll(/[^a-z\d+#-]/gi, "")
433
+ .slice(0, 40);
434
+ const cells: NotebookCell[] = notebook.cells.map((value) => {
435
+ const cell = record(value);
436
+ const outputs = Array.isArray(cell.outputs) ? cell.outputs : [];
437
+ if (outputs.length > 100) throw new Error("A notebook cell has too many outputs to preview.");
438
+ return {
439
+ kind: cell.cell_type === "markdown" ? "markdown" : cell.cell_type === "code" ? "code" : "raw",
440
+ source: notebookText(cell.source),
441
+ outputs: outputs
442
+ .map((value) => {
443
+ const output = record(value);
444
+ const data = record(output.data);
445
+ for (const mime of ["image/png", "image/jpeg", "image/gif", "image/webp"]) {
446
+ if (data[mime]) {
447
+ const base64 = notebookText(data[mime], 4 * 1024 * 1024).replaceAll(/\s/g, "");
448
+ if (base64ImageFits(base64))
449
+ return { kind: "image" as const, text: `data:${mime};base64,${base64}` };
450
+ }
451
+ }
452
+ const html = notebookText(data["text/html"]);
453
+ if (html.trim() && htmlImagesFit(html)) return { kind: "html" as const, text: html };
454
+ return {
455
+ kind: "text" as const,
456
+ text: notebookText(output.text ?? data["text/plain"] ?? output.traceback),
457
+ };
458
+ })
459
+ .filter((output) => output.text.trim().length > 0),
460
+ };
461
+ });
462
+ return { type: "notebook", language, cells };
463
+}
464
+
465
+export function openRtf(bytes: ArrayBuffer): OfficeWorkerResult {
466
+ if (bytes.byteLength > 8 * 1024 * 1024)
467
+ throw new Error("Rich text preview limited to files under 8 MB.");
468
+ // Keep byte offsets intact for binary destinations and multibyte code pages.
469
+ const input = new Uint8Array(bytes);
470
+ let source = "";
471
+ for (let offset = 0; offset < input.length; offset += 8192)
472
+ source += String.fromCharCode(...input.subarray(offset, offset + 8192));
473
+ if (!source.startsWith("{\\rtf")) throw new Error("This is not a valid rich text document.");
474
+ let state = {
475
+ skip: false,
476
+ bold: false,
477
+ italic: false,
478
+ underline: false,
479
+ uc: 1,
480
+ font: 0,
481
+ fontTable: false,
482
+ decoder: new TextDecoder("windows-1252"),
483
+ };
484
+ const fonts = new Map<number, TextDecoder>();
485
+ let defaultFont = 0;
486
+ const stack: (typeof state)[] = [];
487
+ let html = "";
488
+ const pages: { title: string; html: string }[] = [];
489
+ let completedLength = 0;
490
+ const finishPage = () => {
491
+ if (pages.length >= 500) throw new Error("Rich text preview limited to 500 pages.");
492
+ pages.push({ title: `Page ${pages.length + 1}`, html: `<p>${html}</p>` });
493
+ completedLength += html.length;
494
+ html = "";
495
+ };
496
+ let run = "";
497
+ let fallback = 0;
498
+ const encoded = new Uint8Array(bytes.byteLength);
499
+ let encodedLength = 0;
500
+ const flush = () => {
501
+ if (encodedLength) {
502
+ run += escape(
503
+ (fonts.get(state.font) ?? state.decoder).decode(encoded.subarray(0, encodedLength)),
504
+ );
505
+ encodedLength = 0;
506
+ }
507
+ if (completedLength + html.length + run.length > 8 * 1024 * 1024)
508
+ throw new Error("This rich text document has too much content to preview.");
509
+ if (!run) return;
510
+ let span = run;
511
+ if (state.bold) span = `<strong>${span}</strong>`;
512
+ if (state.italic) span = `<em>${span}</em>`;
513
+ if (state.underline) span = `<u>${span}</u>`;
514
+ html += span;
515
+ run = "";
516
+ };
517
+ const emit = (value: string) => {
518
+ if (fallback > 0) {
519
+ fallback--;
520
+ return;
521
+ }
522
+ if (state.skip) return;
523
+ run += escape(value);
524
+ if (completedLength + html.length + run.length > 8 * 1024 * 1024)
525
+ throw new Error("This rich text document has too much content to preview.");
526
+ };
527
+ const emitByte = (value: number) => {
528
+ if (fallback > 0) fallback--;
529
+ else if (!state.skip) encoded[encodedLength++] = value;
530
+ };
531
+ for (let i = 0; i < source.length;) {
532
+ const char = source[i++];
533
+ if (char === "{") {
534
+ flush();
535
+ fallback = 0;
536
+ if (stack.length > 100) throw new Error("This rich text document is nested too deeply.");
537
+ stack.push({ ...state });
538
+ } else if (char === "}") {
539
+ flush();
540
+ fallback = 0;
541
+ const parent = stack.pop();
542
+ if (!parent) throw new Error("This rich text document is damaged.");
543
+ state = parent;
544
+ } else if (char === "\\") {
545
+ const next = source[i];
546
+ if (next === "'") {
547
+ const hex = source.slice(i + 1, i + 3);
548
+ if (!/^[\da-f]{2}$/i.test(hex)) throw new Error("Invalid rich text byte escape.");
549
+ i += 3;
550
+ emitByte(Number.parseInt(hex, 16));
551
+ continue;
552
+ }
553
+ flush();
554
+ if (next === "\\" || next === "{" || next === "}") {
555
+ i++;
556
+ emit(next);
557
+ continue;
558
+ }
559
+ if (next === "*") {
560
+ state.skip = true;
561
+ i++;
562
+ continue;
563
+ }
564
+ const match = /^([a-z]+)(-?\d+)? ?/i.exec(source.slice(i));
565
+ if (!match) {
566
+ if (next === "~") emit("\u00a0");
567
+ i++;
568
+ continue;
569
+ }
570
+ i += match[0].length;
571
+ const word = match[1];
572
+ const arg = match[2] === undefined ? 1 : Number(match[2]);
573
+ if (word === "fonttbl") {
574
+ state.fontTable = true;
575
+ state.skip = true;
576
+ } else if (word === "deff") state.font = defaultFont = arg;
577
+ else if (word === "f") state.font = arg;
578
+ else if (word === "fcharset" && state.fontTable) {
579
+ const label = (
580
+ {
581
+ 0: state.decoder.encoding,
582
+ 1: state.decoder.encoding,
583
+ 77: "macintosh",
584
+ 128: "shift_jis",
585
+ 129: "euc-kr",
586
+ 134: "gbk",
587
+ 136: "big5",
588
+ 161: "windows-1253",
589
+ 162: "windows-1254",
590
+ 163: "windows-1258",
591
+ 177: "windows-1255",
592
+ 178: "windows-1256",
593
+ 186: "windows-1257",
594
+ 204: "windows-1251",
595
+ 222: "windows-874",
596
+ 238: "windows-1250",
597
+ 255: "ibm866",
598
+ } as Record<number, string>
599
+ )[arg];
600
+ if (label) fonts.set(state.font, new TextDecoder(label));
601
+ } else if (
602
+ [
603
+ "colortbl",
604
+ "stylesheet",
605
+ "info",
606
+ "pict",
607
+ "object",
608
+ "fldinst",
609
+ "header",
610
+ "headerl",
611
+ "headerr",
612
+ "headerf",
613
+ "footer",
614
+ "footerl",
615
+ "footerr",
616
+ "footerf",
617
+ "filetbl",
618
+ "datastore",
619
+ "themedata",
620
+ ].includes(word!)
621
+ )
622
+ state.skip = true;
623
+ else if (word === "bin") {
624
+ if (!Number.isSafeInteger(arg) || arg < 0 || arg > source.length - i)
625
+ throw new Error("Invalid rich text binary data.");
626
+ i += arg;
627
+ } else if (word === "ansicpg") {
628
+ const label =
629
+ (
630
+ {
631
+ 65001: "utf-8",
632
+ 932: "shift_jis",
633
+ 936: "gbk",
634
+ 949: "euc-kr",
635
+ 950: "big5",
636
+ 874: "windows-874",
637
+ } as Record<number, string>
638
+ )[arg] ?? `windows-${arg}`;
639
+ try {
640
+ state.decoder = new TextDecoder(label);
641
+ } catch {
642
+ throw new Error("This rich text code page is not supported.");
643
+ }
644
+ } else if (word === "b") state.bold = arg !== 0;
645
+ else if (word === "i") state.italic = arg !== 0;
646
+ else if (word === "ul") state.underline = arg !== 0;
647
+ else if (word === "ulnone") state.underline = false;
648
+ else if (word === "plain")
649
+ state = { ...state, bold: false, italic: false, underline: false, font: defaultFont };
650
+ else if (word === "uc") state.uc = Math.max(0, Math.min(10, arg));
651
+ else if (word === "u") {
652
+ fallback = 0;
653
+ emit(String.fromCharCode(arg < 0 ? arg + 65536 : arg));
654
+ fallback = state.uc;
655
+ } else if (!state.skip && word === "page") finishPage();
656
+ else if (!state.skip && (word === "par" || word === "line" || word === "row")) html += "<br>";
657
+ else if (word === "tab" || word === "cell") emit("\t");
658
+ else if (word === "endash") emit("–");
659
+ // oxlint-disable-next-line t3code/no-em-dash-copy -- Preserve the RTF document's punctuation.
660
+ else if (word === "emdash") emit("—");
661
+ else if (word === "bullet") emit("•");
662
+ else if (word === "lquote") emit("‘");
663
+ else if (word === "rquote") emit("’");
664
+ else if (word === "ldblquote") emit("“");
665
+ else if (word === "rdblquote") emit("”");
666
+ } else if (char !== "\r" && char !== "\n" && char) emitByte(char.charCodeAt(0));
667
+ }
668
+ if (stack.length) throw new Error("This rich text document is damaged.");
669
+ flush();
670
+ if (html || !pages.length) finishPage();
671
+ return { type: "document", navigation: "page", pages };
672
+}
apps/web/src/components/files/officeArchive.ts
+6
-2
@@ -225,6 +225,8 @@ function boundedWorksheet(
225
let explicitRow = false;
226
let rowTag = "";
227
let column = 0;
228
+ let maxRow = 0;
229
+ let maxColumn = 0;
230
let cells = new Map<number, string>();
231
let cell: {
232
start: number;
@@ -299,7 +301,9 @@ function boundedWorksheet(
301
truncated = true;
302
return;
303
}
302
- if (++budget.cells > 2_000_000 || row * column > 8_000_000)
304
+ maxRow = Math.max(maxRow, row);
305
+ maxColumn = Math.max(maxColumn, column);
306
+ if (++budget.cells > 2_000_000 || maxRow * maxColumn > 8_000_000)
307
throw new Error(
308
"This workbook exceeds the preview limit: 2 million cells or 8 million grid positions.",
309
);
@@ -934,7 +938,7 @@ async function wordExpansion(
938
939
// Stream each entry before handing it to an engine. A tiny compressed archive
940
// must not turn into an unbounded allocation on the UI thread.
937
-async function boundedArchive(input: Extract<OfficeWorkerRequest, { type: "open" }>) {
941
+export async function boundedArchive(input: Extract<OfficeWorkerRequest, { type: "open" }>) {
942
const { bytes } = input;
943
const spreadsheet = input.kind === "xlsx";
944
if (bytes.byteLength > OFFICE_PREVIEW_MAX_BYTES)
apps/web/src/components/files/officePreview.worker.ts
+49
-17
@@ -1,20 +1,52 @@
1
-import { prepareOffice } from "./officeArchive";
1
+import { boundedArchive, prepareOffice } from "./officeArchive";
2
+import { PREVIEW_PAGE_ROWS } from "./largeTextProtocol";
3
import type { OfficeWorkerRequest, OfficeWorkerResult } from "./officePreviewProtocol";
3
-self.addEventListener("message", (event: MessageEvent<OfficeWorkerRequest>) => {
4
- if (event.data.type !== "open") return;
5
- const input = event.data;
6
- void prepareOffice(input.bytes, input.kind)
7
- .then((result) => {
8
- const reply: OfficeWorkerResult = { type: input.kind, ...result };
9
- self.postMessage(reply, { transfer: [result.bytes] });
10
- })
11
- .catch((cause: unknown) => {
12
- self.postMessage(
13
- {
14
- type: "error",
15
- message: cause instanceof Error ? cause.message : "Could not parse this file.",
16
- } satisfies OfficeWorkerResult,
17
- { transfer: [] },
18
- );
4
+
5
+let workbook: string[][][] = [];
6
+function reply(result: OfficeWorkerResult) {
7
+ self.postMessage(result, { transfer: "bytes" in result ? [result.bytes] : [] });
8
+}
9
+async function handle(input: OfficeWorkerRequest) {
10
+ if (input.type === "sheet-page") {
11
+ const rows = workbook[input.sheet];
12
+ if (!rows) throw new Error("This sheet is unavailable.");
13
+ reply({
14
+ type: "sheet-page",
15
+ sheet: input.sheet,
16
+ index: input.index,
17
+ rows: rows.slice(input.index * PREVIEW_PAGE_ROWS, (input.index + 1) * PREVIEW_PAGE_ROWS),
18
+ });
19
+ return;
20
+ }
21
+ if (input.kind === "docx" || input.kind === "xlsx" || input.kind === "pptx") {
22
+ reply({ type: input.kind, ...(await prepareOffice(input.bytes, input.kind)) });
23
+ return;
24
+ }
25
+ const { openDocument, openEpub, openNotebook, openRtf } =
26
+ await import("./extendedDocumentPreview");
27
+ if (input.kind === "ipynb" || input.kind === "rtf") {
28
+ reply(input.kind === "ipynb" ? openNotebook(input.bytes) : openRtf(input.bytes));
29
+ return;
30
+ }
31
+ const { archive } = await boundedArchive(input);
32
+ const result =
33
+ input.kind === "epub" ? await openEpub(archive) : await openDocument(archive, input.kind);
34
+ if (result.type === "ods") {
35
+ workbook = result.sheets.map((sheet) => sheet.rows);
36
+ reply({
37
+ ...result,
38
+ sheets: result.sheets.map((sheet) => ({
39
+ ...sheet,
40
+ rows: sheet.rows.slice(0, PREVIEW_PAGE_ROWS),
41
+ })),
42
});
43
+ } else reply(result);
44
+}
45
+self.addEventListener("message", (event: MessageEvent<OfficeWorkerRequest>) => {
46
+ void handle(event.data).catch((cause: unknown) =>
47
+ reply({
48
+ type: "error",
49
+ message: cause instanceof Error ? cause.message : "Could not parse this file.",
50
+ }),
51
+ );
52
});
apps/web/src/components/files/officePreviewProtocol.ts
+34
-5
@@ -1,16 +1,45 @@
1
import type { OfficePreviewKind } from "@t3tools/shared/filePreview";
2
+
3
export const OFFICE_PREVIEW_MAX_BYTES = 50 * 1024 * 1024;
4
export const OFFICE_PREVIEW_MAX_ROWS = 250_000;
5
export const OFFICE_PREVIEW_MAX_COLUMNS = 1024;
5
-export type OfficeWorkerRequest = {
6
- readonly type: "open";
7
- readonly kind: OfficePreviewKind;
8
- readonly bytes: ArrayBuffer;
6
+
7
+export type OfficeWorkerRequest =
8
+ | { readonly type: "open"; readonly kind: OfficePreviewKind; readonly bytes: ArrayBuffer }
9
+ | { readonly type: "sheet-page"; readonly sheet: number; readonly index: number };
10
+
11
+export type NotebookCell = {
12
+ readonly kind: "markdown" | "code" | "raw";
13
+ readonly source: string;
14
+ readonly outputs: { readonly kind: "text" | "image" | "html"; readonly text: string }[];
15
};
16
+
17
export type OfficeWorkerResult =
18
+ | {
19
+ readonly type: "document";
20
+ readonly navigation: "page" | "slide" | "chapter";
21
+ readonly pages: { title: string; html: string }[];
22
+ }
23
+ | { readonly type: "notebook"; readonly language: string; readonly cells: NotebookCell[] }
24
| { readonly type: "error"; readonly message: string }
25
| {
13
- readonly type: OfficePreviewKind;
26
+ readonly type: "docx" | "xlsx" | "pptx";
27
readonly bytes: ArrayBuffer;
28
readonly truncated: boolean;
29
+ }
30
+ | {
31
+ readonly type: "ods";
32
+ readonly sheets: {
33
+ name: string;
34
+ rows: string[][];
35
+ truncated: boolean;
36
+ rowCount: number;
37
+ columns: number;
38
+ }[];
39
+ }
40
+ | {
41
+ readonly type: "sheet-page";
42
+ readonly sheet: number;
43
+ readonly index: number;
44
+ readonly rows: string[][];
45
};
apps/web/src/components/files/pdfDocument.ts
new
+45
@@ -0,0 +1,45 @@
1
+import { getDocument, GlobalWorkerOptions } from "pdfjs-dist";
2
+import workerUrl from "pdfjs-dist/build/pdf.worker.min.mjs?url";
3
+import fallbackUrl from "pdfjs-dist/wasm/openjpeg_nowasm_fallback.js?url";
4
+
5
+GlobalWorkerOptions.workerSrc = workerUrl;
6
+
7
+const standardFonts = import.meta.glob<string>(
8
+ "../../../node_modules/pdfjs-dist/standard_fonts/*",
9
+ { query: "?url", import: "default", eager: true },
10
+);
11
+const cMaps = import.meta.glob<string>("../../../node_modules/pdfjs-dist/cmaps/*.bcmap", {
12
+ query: "?url",
13
+ import: "default",
14
+ eager: true,
15
+});
16
+const wasm = import.meta.glob<string>("../../../node_modules/pdfjs-dist/wasm/*.{wasm,js}", {
17
+ query: "?url",
18
+ import: "default",
19
+ eager: true,
20
+});
21
+export class BinaryDataFactory {
22
+ async fetch({ kind, filename }: { kind: string; filename: string }) {
23
+ const assets = kind === "cMapUrl" ? cMaps : kind === "wasmUrl" ? wasm : standardFonts;
24
+ const directory = kind === "cMapUrl" ? "cmaps" : kind === "wasmUrl" ? "wasm" : "standard_fonts";
25
+ const url = assets[`../../../node_modules/pdfjs-dist/${directory}/${filename}`];
26
+ if (!url) throw new Error(`Unsupported PDF ${directory} asset.`);
27
+ const response = await fetch(url);
28
+ if (!response.ok || response.headers.get("content-type")?.startsWith("text/html"))
29
+ throw new Error(`PDF ${directory} asset unavailable.`);
30
+ return new Uint8Array(await response.arrayBuffer());
31
+ }
32
+}
33
+
34
+export function openPdf(bytes: ArrayBuffer) {
35
+ return getDocument({
36
+ data: new Uint8Array(bytes),
37
+ useSystemFonts: true,
38
+ useWorkerFetch: false,
39
+ BinaryDataFactory,
40
+ // PDF.js imports JavaScript decoders directly, bypassing BinaryDataFactory.
41
+ wasmUrl: new URL(".", new URL(fallbackUrl, window.location.href)).href,
42
+ canvasMaxAreaInBytes: 32 * 1024 * 1024,
43
+ useWasm: true,
44
+ });
45
+}
apps/web/src/components/files/pdfViewer.css
new
+52
@@ -0,0 +1,52 @@
1
+/* TextLayer geometry follows PDF.js' text layer contract (Apache-2.0). */
2
+[data-pdf-text-layer] {
3
+ position: absolute;
4
+ inset: 0;
5
+ overflow: clip;
6
+ line-height: 1;
7
+ text-align: initial;
8
+ letter-spacing: normal;
9
+ word-spacing: normal;
10
+ text-size-adjust: none;
11
+ forced-color-adjust: none;
12
+ transform-origin: 0 0;
13
+ --min-font-size: 1;
14
+ --text-scale-factor: calc(var(--total-scale-factor) * var(--min-font-size));
15
+ --min-font-size-inv: calc(1 / var(--min-font-size));
16
+ --scale-round-x: 1px;
17
+ --scale-round-y: 1px;
18
+}
19
+[data-pdf-text-layer][data-main-rotation="90"] {
20
+ transform: rotate(90deg) translateY(-100%);
21
+}
22
+[data-pdf-text-layer][data-main-rotation="180"] {
23
+ transform: rotate(180deg) translate(-100%, -100%);
24
+}
25
+[data-pdf-text-layer][data-main-rotation="270"] {
26
+ transform: rotate(270deg) translateX(-100%);
27
+}
28
+[data-pdf-text-layer] :is(span, br) {
29
+ color: transparent;
30
+ position: absolute;
31
+ white-space: pre;
32
+ cursor: text;
33
+ transform-origin: 0 0;
34
+ user-select: text;
35
+}
36
+[data-pdf-text-layer] > :not(.markedContent),
37
+[data-pdf-text-layer] .markedContent span:not(.markedContent) {
38
+ --font-height: 0;
39
+ --scale-x: 1;
40
+ --rotate: 0deg;
41
+ font-size: calc(var(--text-scale-factor) * var(--font-height));
42
+ transform: rotate(var(--rotate)) scaleX(var(--scale-x)) scale(var(--min-font-size-inv));
43
+}
44
+[data-pdf-text-layer] .markedContent {
45
+ display: contents;
46
+}
47
+[data-pdf-text-layer] ::selection {
48
+ background: #4a90e266;
49
+}
50
+[data-pdf-text-layer] .pdf-find-match {
51
+ background: #ffc80066;
52
+}
apps/web/src/components/files/readBoundedDocument.ts
new
+26
@@ -0,0 +1,26 @@
1
+export async function readBoundedDocument(
2
+ src: string,
3
+ signal: AbortSignal,
4
+ limit: number,
5
+ cache: RequestCache = "default",
6
+) {
7
+ const response = await fetch(src, { signal, cache });
8
+ if (!response.ok || Number(response.headers.get("content-length")) > limit)
9
+ throw new Error("This file is unavailable or too large to preview.");
10
+ const reader = response.body?.getReader();
11
+ if (!reader) throw new Error("This file is unavailable.");
12
+ const chunks: Uint8Array<ArrayBuffer>[] = [];
13
+ let size = 0;
14
+ try {
15
+ while (true) {
16
+ const { done, value } = await reader.read();
17
+ if (done) break;
18
+ size += value.length;
19
+ if (size > limit) throw new Error("This file is too large to preview.");
20
+ chunks.push(value);
21
+ }
22
+ } finally {
23
+ await reader.cancel();
24
+ }
25
+ return new Blob(chunks).arrayBuffer();
26
+}
apps/web/src/pierre-icons.test.ts
+22
@@ -31,6 +31,28 @@ describe("Pierre file icons", () => {
31
);
32
});
33
34
+ it.each([
35
+ ["pdf", "pdf"],
36
+ ["doc", "word"],
37
+ ["docx", "word"],
38
+ ["odt", "word"],
39
+ ["rtf", "word"],
40
+ ["xls", "spreadsheet"],
41
+ ["xlsx", "spreadsheet"],
42
+ ["ods", "spreadsheet"],
43
+ ["csv", "spreadsheet"],
44
+ ["tsv", "spreadsheet"],
45
+ ["ppt", "slides"],
46
+ ["pptx", "slides"],
47
+ ["odp", "slides"],
48
+ ["epub", "epub"],
49
+ ])("uses the document family icon for %s", (extension, family) => {
50
+ for (const path of [`report.${extension}`, `folder/REPORT.${extension.toUpperCase()}`]) {
51
+ assert.equal(resolvePierreIconForEntry(path, "file")?.name, `t3-file-icon-${family}`);
52
+ assert.isTrue(hasSpecificPierreIconForFileName(path));
53
+ }
54
+ });
55
+
56
it("ships every custom icon referenced by the extended resolver", () => {
57
const customIconNames = new Set(
58
[
apps/web/src/pierre-icons.ts
+69
@@ -13,6 +13,14 @@ export interface PierreIconResolution {
13
14
const PIERRE_ICON_SPRITE_ID = "t3code-pierre-file-icon-sprite";
15
16
+const DOCUMENT_ICON_COLORS: Readonly<Record<string, string>> = {
17
+ "t3-file-icon-pdf": "light-dark(#d52c36, #ff6762)",
18
+ "t3-file-icon-word": "light-dark(#1a85d4, #69b1ff)",
19
+ "t3-file-icon-spreadsheet": "light-dark(#199f43, #5ecc71)",
20
+ "t3-file-icon-slides": "light-dark(#d47628, #ffa359)",
21
+ "t3-file-icon-epub": "light-dark(#693acf, #9d6afb)",
22
+};
23
+
24
const T3_FILE_ICON_SPRITE = `
25
<svg xmlns="http://www.w3.org/2000/svg" width="0" height="0" aria-hidden="true">
26
<!-- Lucide Film icon, ISC license. -->
@@ -41,6 +49,48 @@ const T3_FILE_ICON_SPRITE = `
49
<symbol id="t3-file-icon-xml" viewBox="0 0 24 24">
50
<path fill="none" style="stroke: light-dark(#d47628, #ffa359)" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" d="m18 16 4-4-4-4M6 8l-4 4 4 4M14.5 4l-5 16" />
51
</symbol>
52
+ <!-- Lucide File Type, File Text, File Spreadsheet, Presentation and Book Open icons, ISC license. -->
53
+ <symbol id="t3-file-icon-pdf" viewBox="0 0 24 24">
54
+ <g fill="none" style="stroke: ${DOCUMENT_ICON_COLORS["t3-file-icon-pdf"]}" stroke-width="2" stroke-linecap="round" stroke-linejoin="round">
55
+ <path d="M6 22a2 2 0 0 1-2-2V4a2 2 0 0 1 2-2h8a2.4 2.4 0 0 1 1.704.706l3.588 3.588A2.4 2.4 0 0 1 20 8v12a2 2 0 0 1-2 2z" />
56
+ <path d="M14 2v5a1 1 0 0 0 1 1h5" />
57
+ <path d="M11 18h2" />
58
+ <path d="M12 12v6" />
59
+ <path d="M9 13v-.5a.5.5 0 0 1 .5-.5h5a.5.5 0 0 1 .5.5v.5" />
60
+ </g>
61
+ </symbol>
62
+ <symbol id="t3-file-icon-word" viewBox="0 0 24 24">
63
+ <g fill="none" style="stroke: ${DOCUMENT_ICON_COLORS["t3-file-icon-word"]}" stroke-width="2" stroke-linecap="round" stroke-linejoin="round">
64
+ <path d="M6 22a2 2 0 0 1-2-2V4a2 2 0 0 1 2-2h8a2.4 2.4 0 0 1 1.704.706l3.588 3.588A2.4 2.4 0 0 1 20 8v12a2 2 0 0 1-2 2z" />
65
+ <path d="M14 2v5a1 1 0 0 0 1 1h5" />
66
+ <path d="M10 9H8" />
67
+ <path d="M16 13H8" />
68
+ <path d="M16 17H8" />
69
+ </g>
70
+ </symbol>
71
+ <symbol id="t3-file-icon-spreadsheet" viewBox="0 0 24 24">
72
+ <g fill="none" style="stroke: ${DOCUMENT_ICON_COLORS["t3-file-icon-spreadsheet"]}" stroke-width="2" stroke-linecap="round" stroke-linejoin="round">
73
+ <path d="M6 22a2 2 0 0 1-2-2V4a2 2 0 0 1 2-2h8a2.4 2.4 0 0 1 1.704.706l3.588 3.588A2.4 2.4 0 0 1 20 8v12a2 2 0 0 1-2 2z" />
74
+ <path d="M14 2v5a1 1 0 0 0 1 1h5" />
75
+ <path d="M8 13h2" />
76
+ <path d="M14 13h2" />
77
+ <path d="M8 17h2" />
78
+ <path d="M14 17h2" />
79
+ </g>
80
+ </symbol>
81
+ <symbol id="t3-file-icon-slides" viewBox="0 0 24 24">
82
+ <g fill="none" style="stroke: ${DOCUMENT_ICON_COLORS["t3-file-icon-slides"]}" stroke-width="2" stroke-linecap="round" stroke-linejoin="round">
83
+ <path d="M2 3h20" />
84
+ <path d="M21 3v11a2 2 0 0 1-2 2H5a2 2 0 0 1-2-2V3" />
85
+ <path d="m7 21 5-5 5 5" />
86
+ </g>
87
+ </symbol>
88
+ <symbol id="t3-file-icon-epub" viewBox="0 0 24 24">
89
+ <g fill="none" style="stroke: ${DOCUMENT_ICON_COLORS["t3-file-icon-epub"]}" stroke-width="2" stroke-linecap="round" stroke-linejoin="round">
90
+ <path d="M12 7v14" />
91
+ <path d="M3 18a1 1 0 0 1-1-1V4a1 1 0 0 1 1-1h5a4 4 0 0 1 4 4 4 4 0 0 1 4-4h5a1 1 0 0 1 1 1v13a1 1 0 0 1-1 1h-6a3 3 0 0 0-3 3 3 3 0 0 0-3-3z" />
92
+ </g>
93
+ </symbol>
94
<!-- Language logos from Simple Icons, CC0. -->
95
<symbol id="t3-file-icon-csharp" viewBox="0 0 24 24">
96
<path style="fill: light-dark(#693acf, #9d6afb)" d="M1.194 7.543v8.913c0 1.103.588 2.122 1.544 2.674l7.718 4.456a3.086 3.086 0 0 0 3.088 0l7.718-4.456a3.087 3.087 0 0 0 1.544-2.674V7.543a3.084 3.084 0 0 0-1.544-2.673L13.544.414a3.086 3.086 0 0 0-3.088 0L2.738 4.87a3.085 3.085 0 0 0-1.544 2.673Zm5.403 2.914v3.087a.77.77 0 0 0 .772.772.773.773 0 0 0 .772-.772.773.773 0 0 1 1.317-.546.775.775 0 0 1 .226.546 2.314 2.314 0 1 1-4.631 0v-3.087c0-.615.244-1.203.679-1.637a2.312 2.312 0 0 1 3.274 0c.434.434.678 1.023.678 1.637a.769.769 0 0 1-.226.545.767.767 0 0 1-1.091 0 .77.77 0 0 1-.226-.545.77.77 0 0 0-.772-.772.771.771 0 0 0-.772.772Zm12.35 3.087a.77.77 0 0 1-.772.772h-.772v.772a.773.773 0 0 1-1.544 0v-.772h-1.544v.772a.773.773 0 0 1-1.317.546.775.775 0 0 1-.226-.546v-.772H12a.771.771 0 1 1 0-1.544h.772v-1.543H12a.77.77 0 1 1 0-1.544h.772v-.772a.773.773 0 0 1 1.317-.546.775.775 0 0 1 .226.546v.772h1.544v-.772a.773.773 0 0 1 1.544 0v.772h.772a.772.772 0 0 1 0 1.544h-.772v1.543h.772a.776.776 0 0 1 .772.772Zm-3.088-2.315h-1.544v1.543h1.544v-1.543Z" />
@@ -95,6 +145,20 @@ export const T3_PIERRE_ICONS = {
145
...Object.fromEntries(
146
VIDEO_FILE_EXTENSIONS.map((extension) => [extension, "t3-file-icon-video"]),
147
),
148
+ pdf: "t3-file-icon-pdf",
149
+ doc: "t3-file-icon-word",
150
+ docx: "t3-file-icon-word",
151
+ odt: "t3-file-icon-word",
152
+ rtf: "t3-file-icon-word",
153
+ xls: "t3-file-icon-spreadsheet",
154
+ xlsx: "t3-file-icon-spreadsheet",
155
+ ods: "t3-file-icon-spreadsheet",
156
+ csv: "t3-file-icon-spreadsheet",
157
+ tsv: "t3-file-icon-spreadsheet",
158
+ ppt: "t3-file-icon-slides",
159
+ pptx: "t3-file-icon-slides",
160
+ odp: "t3-file-icon-slides",
161
+ epub: "t3-file-icon-epub",
162
cs: "t3-file-icon-csharp",
163
csx: "t3-file-icon-csharp",
164
dart: "t3-file-icon-dart",
@@ -180,6 +244,11 @@ export function resolvePierreIconForEntry(
244
: icon;
245
}
246
247
+export function documentIconColorForFileName(fileName: string) {
248
+ const icon = resolvePierreIconForEntry(fileName, "file");
249
+ return icon ? DOCUMENT_ICON_COLORS[icon.name] : undefined;
250
+}
251
+
252
export function hasSpecificPierreIconForFileName(fileName: string): boolean {
253
return resolvePierreIconForEntry(fileName, "file")?.token !== "default";
254
}
apps/web/vite.config.ts
+13
@@ -11,6 +11,7 @@ import "vite-plus/test/config";
11
import { defineConfig, type Connect, type Plugin } from "vite-plus";
12
import * as Hash from "effect/Hash";
13
import pkg from "./package.json" with { type: "json" };
14
+import pdfjsPkg from "pdfjs-dist/package.json" with { type: "json" };
15
16
import {
17
DEV_PROXIED_ORIGIN_PRESERVING_PREFIXES,
@@ -112,9 +113,11 @@ const browserTestProject = {
113
"@betteroffice/xlsx",
114
"@betteroffice/xlsx/headless",
115
"fast-xml-parser",
116
+ "entities",
117
"dompurify",
118
"heic-to/next",
119
"jszip",
120
+ "pdfjs-dist",
121
"sql.js",
122
"tiff",
123
],
@@ -351,9 +354,19 @@ export default defineConfig(() => {
354
sourcemap: buildSourcemap,
355
rolldownOptions: {
356
output: {
357
+ // PDF.js resolves decoder modules and ICC data relative to wasmUrl.
358
+ assetFileNames: (asset) =>
359
+ asset.originalFileNames.some((path) => /[/\\]pdfjs-dist[/\\]wasm[/\\]/.test(path))
360
+ ? `assets/pdfjs-${pdfjsPkg.version}/[name][extname]`
361
+ : "assets/[name]-[hash][extname]",
362
codeSplitting: {
363
// Preview imports must not fold shared file icons back into main.
364
groups: [
365
+ {
366
+ name: "react",
367
+ test: /\/node_modules\/(?:react|react-dom|scheduler)\//,
368
+ includeDependenciesRecursively: false,
369
+ },
370
{
371
name: "icons",
372
test: /\/(?:components\/chat\/PierreEntryIcon\.tsx|pierre-icons\.ts|mediaPreview\.ts)$/,
packages/shared/src/filePreview.test.ts
+14
-3
@@ -68,6 +68,12 @@ describe("workspace file previews", () => {
68
"pptx",
69
"application/vnd.openxmlformats-officedocument.presentationml.presentation",
70
],
71
+ ["notes.odt", "odt", "application/vnd.oasis.opendocument.text"],
72
+ ["budget.ods", "ods", "application/vnd.oasis.opendocument.spreadsheet"],
73
+ ["deck.odp", "odp", "application/vnd.oasis.opendocument.presentation"],
74
+ ["notes.rtf", "rtf", "application/rtf"],
75
+ ["guide.epub", "epub", "application/epub+zip"],
76
+ ["analysis.ipynb", "ipynb", "application/x-ipynb+json"],
77
])("recognizes Office assets at literal path %s", (name, kind, mimeType) => {
78
expect(officePreviewKind({ name })).toBe(kind);
79
expect(filePreviewKind({ name, mimeType: "application/octet-stream" })).toBe(kind);
@@ -85,6 +91,9 @@ describe("workspace file previews", () => {
91
"report.xlsx?download=1",
92
"slides%2Epptx",
93
"report.docm",
94
+ "guide.epub?download=1",
95
+ "analysis.ipynb#notes.txt",
96
+ "report.odt.ts",
97
"slides.ppt",
98
])("rejects non-preview path %s", (path) => {
99
expect(isWorkspacePreviewEntryPath(path)).toBe(false);
@@ -145,8 +154,10 @@ describe("attachment preview classification", () => {
154
["diagram.MERMAID", "text/plain", "mermaid"],
155
["download", "text/vnd.mermaid", "mermaid"],
156
["source-code.otf", "application/vnd.oasis.opendocument.formula-template", "font"],
148
- ["notes.rtf", "application/rtf", "text"],
149
- ["notes.rtf", "", "text"],
157
+ ["notes.rtf", "application/rtf", "rtf"],
158
+ ["notes.rtf", "", "rtf"],
159
+ ["download", "text/rtf", "rtf"],
160
+ ["analysis.IPYNB", "application/json", "ipynb"],
161
["clip.weba", "", "audio"],
162
["image.bmp", "application/octet-stream", "image"],
163
["photo.heic", "application/octet-stream", "image"],
@@ -165,7 +176,7 @@ describe("attachment preview classification", () => {
176
expect(() => decodeFilePreviewText(new Uint8Array([65, 0, 66]))).toThrow("binary");
177
expect(() => decodeFilePreviewText(new Uint8Array([255]))).toThrow("UTF-8");
178
});
168
- it.each(["doc", "xls", "ppt", "odt", "ods", "odp", "epub", "7z", "rar"])(
179
+ it.each(["doc", "xls", "ppt", "7z", "rar"])(
180
"retains the binary fallback for %s, which opencode does not render",
181
(extension) => {
182
expect(filePreviewKind({ name: `sample.${extension}` })).toBe("unsupported");
packages/shared/src/filePreview.ts
+18
-3
@@ -25,6 +25,12 @@ export type FilePreviewKind =
25
| "docx"
26
| "xlsx"
27
| "pptx"
28
+ | "odt"
29
+ | "ods"
30
+ | "odp"
31
+ | "rtf"
32
+ | "epub"
33
+ | "ipynb"
34
| "text"
35
| "unsupported";
36
@@ -54,7 +60,6 @@ export function filePreviewKind(
60
mime === "application/vnd.oasis.opendocument.formula-template"
61
)
62
return "font";
57
- if (richDocuments && mime === "application/rtf") return "text";
63
if (mime.startsWith("image/")) return "image";
64
if (mime.startsWith("video/")) return "video";
65
if (mime.startsWith("audio/")) return "audio";
@@ -132,6 +137,12 @@ const OFFICE_MIME_TYPES = {
137
docx: "application/vnd.openxmlformats-officedocument.wordprocessingml.document",
138
xlsx: "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
139
pptx: "application/vnd.openxmlformats-officedocument.presentationml.presentation",
140
+ odt: "application/vnd.oasis.opendocument.text",
141
+ ods: "application/vnd.oasis.opendocument.spreadsheet",
142
+ odp: "application/vnd.oasis.opendocument.presentation",
143
+ rtf: "application/rtf",
144
+ epub: "application/epub+zip",
145
+ ipynb: "application/x-ipynb+json",
146
} as const;
147
148
export type OfficePreviewKind = keyof typeof OFFICE_MIME_TYPES;
@@ -141,10 +152,14 @@ export function officePreviewKind(file: {
152
readonly mimeType?: string;
153
}): OfficePreviewKind | null {
154
const mime = file.mimeType?.split(";", 1)[0]?.trim().toLowerCase() ?? "";
144
- for (const kind of ["docx", "xlsx", "pptx"] as const) {
155
+ if (mime === "text/rtf") return "rtf";
156
+ for (const kind of Object.keys(OFFICE_MIME_TYPES) as OfficePreviewKind[]) {
157
if (mime === OFFICE_MIME_TYPES[kind]) return kind;
158
if (
147
- (!mime || mime === "application/octet-stream" || mime === "text/plain") &&
159
+ (!mime ||
160
+ mime === "application/octet-stream" ||
161
+ mime === "text/plain" ||
162
+ (kind === "ipynb" && mime === "application/json")) &&
163
file.name.toLowerCase().endsWith(`.${kind}`)
164
)
165
return kind;
pnpm-lock.yaml
+137
@@ -1136,6 +1136,9 @@ importers:
1136
effect:
1137
specifier: 4.0.5
1138
version: 4.0.5
1139
+ entities:
1140
+ specifier: 8.1.0
1141
+ version: 8.1.0
1142
fast-xml-parser:
1143
specifier: 5.7.3
1144
version: 5.7.3
@@ -1169,6 +1172,9 @@ importers:
1172
morphicons:
1173
specifier: ^1.7.1
1174
version: 1.7.1(react-native-svg@15.16.0(react-native@0.88.0(@babel/core@7.29.7(supports-color@8.1.1))(@react-native/metro-config@0.88.0(@babel/core@7.29.7(supports-color@8.1.1))(bufferutil@4.1.0)(supports-color@8.1.1)(utf-8-validate@6.0.6))(@types/react@19.3.0)(bufferutil@4.1.0)(react@19.3.0)(supports-color@8.1.1)(utf-8-validate@6.0.6))(react@19.3.0))(react-native@0.88.0(@babel/core@7.29.7(supports-color@8.1.1))(@react-native/metro-config@0.88.0(@babel/core@7.29.7(supports-color@8.1.1))(bufferutil@4.1.0)(supports-color@8.1.1)(utf-8-validate@6.0.6))(@types/react@19.3.0)(bufferutil@4.1.0)(react@19.3.0)(supports-color@8.1.1)(utf-8-validate@6.0.6))(react@19.3.0)
1175
+ pdfjs-dist:
1176
+ specifier: 6.4.299
1177
+ version: 6.4.299
1178
pele:
1179
specifier: ^1.8.0
1180
version: 1.8.0
@@ -4558,6 +4564,81 @@ packages:
4564
'@types/node':
4565
optional: true
4566
4567
+ '@napi-rs/canvas-android-arm64@1.0.13':
4568
+ resolution: {integrity: sha512-trn7AaDzQ+yeeA0ezhkC7cB1ZDPtp4UYk0EvB5F8qtYL9jEVWtVrU30WOpI52DszoTxaTsCgf5PP4ZVaOlOIHQ==}
4569
+ engines: {node: '>= 10'}
4570
+ cpu: [arm64]
4571
+ os: [android]
4572
+
4573
+ '@napi-rs/canvas-darwin-arm64@1.0.13':
4574
+ resolution: {integrity: sha512-6Y2djEetL8copxfNK6hg7oig/xqTwOchv19hBOm+zBNdKc/02Z3kjZN5wrLzv/l5O04/bAQS6I/AlQpwBlXgPg==}
4575
+ engines: {node: '>= 10'}
4576
+ cpu: [arm64]
4577
+ os: [darwin]
4578
+
4579
+ '@napi-rs/canvas-darwin-x64@1.0.13':
4580
+ resolution: {integrity: sha512-OZoPwwu4TMMg8GTCFBrg8UFExwQTU/55i3uTC/izDat6PHE+ro9dimyQoOboFf91+NY7Pnn70ScR8f8I8+eZXw==}
4581
+ engines: {node: '>= 10'}
4582
+ cpu: [x64]
4583
+ os: [darwin]
4584
+
4585
+ '@napi-rs/canvas-linux-arm-gnueabihf@1.0.13':
4586
+ resolution: {integrity: sha512-dZSYaKROuswU2eUWG5aBKndNVe68elo1YX93xtRax7q658tj8rYE+HwduovWJgs6qwv1r1/CLR60nDeBbyjvyg==}
4587
+ engines: {node: '>= 10'}
4588
+ cpu: [arm]
4589
+ os: [linux]
4590
+
4591
+ '@napi-rs/canvas-linux-arm64-gnu@1.0.13':
4592
+ resolution: {integrity: sha512-W6FzYE8ADHI93In1vi80KBL6btDtAMuj50Ch5ptkWS8xuntyzbR4h5JDUmXoakxZn2yMInksudTEqVSWoFqFGg==}
4593
+ engines: {node: '>= 10'}
4594
+ cpu: [arm64]
4595
+ os: [linux]
4596
+ libc: [glibc]
4597
+
4598
+ '@napi-rs/canvas-linux-arm64-musl@1.0.13':
4599
+ resolution: {integrity: sha512-+6fQ+rR49hkAa4voh2stLIwQ+4t+gmoseM8Bc53/DlMXb73IKymDPWK9A/NtWeqTW4ijWppYuyuxS2hPI8tzNA==}
4600
+ engines: {node: '>= 10'}
4601
+ cpu: [arm64]
4602
+ os: [linux]
4603
+ libc: [musl]
4604
+
4605
+ '@napi-rs/canvas-linux-riscv64-gnu@1.0.13':
4606
+ resolution: {integrity: sha512-17XEMO3t9RW/TtHtSFf2AkaSWuCWdzbPeidz7zrOFawHxoLSJCnpTI1hN5aUvrZNnkNNmt/vZVJCh0ryUaik3A==}
4607
+ engines: {node: '>= 10'}
4608
+ cpu: [riscv64]
4609
+ os: [linux]
4610
+ libc: [glibc]
4611
+
4612
+ '@napi-rs/canvas-linux-x64-gnu@1.0.13':
4613
+ resolution: {integrity: sha512-lc48IwknVHBsNzXn9Xs5OkUsRX38Ml3Z1aOybQ51mMEzZckm8+4A0qROlyYHEoBPzr1kL3IU8ZUAqDb7wtdX8w==}
4614
+ engines: {node: '>= 10'}
4615
+ cpu: [x64]
4616
+ os: [linux]
4617
+ libc: [glibc]
4618
+
4619
+ '@napi-rs/canvas-linux-x64-musl@1.0.13':
4620
+ resolution: {integrity: sha512-bGY1r26S2bs8UNFJcPWcCSNdCFbKep5PATsvtgtC9A5u6SRgr2/jJE/RlhHL5m9ZpixNDhVkLY/5h16Kx5V9YA==}
4621
+ engines: {node: '>= 10'}
4622
+ cpu: [x64]
4623
+ os: [linux]
4624
+ libc: [musl]
4625
+
4626
+ '@napi-rs/canvas-win32-arm64-msvc@1.0.13':
4627
+ resolution: {integrity: sha512-2vRPBZoaIvJrUkDhZTE2Hn0YrnpJRCWVM3EJACoBKjuWI6WgPxGtUoRMsxRviCsyks8bKTSFdyw14CellnG2qA==}
4628
+ engines: {node: '>= 10'}
4629
+ cpu: [arm64]
4630
+ os: [win32]
4631
+
4632
+ '@napi-rs/canvas-win32-x64-msvc@1.0.13':
4633
+ resolution: {integrity: sha512-DUlW3lZPz63kYsf8mxgyyZuevhlH0T3/BGP9l+NazhB0F1+gL6D4hqZdnhZpw24DlWmTpATDcyYSZlJOi+DdfQ==}
4634
+ engines: {node: '>= 10'}
4635
+ cpu: [x64]
4636
+ os: [win32]
4637
+
4638
+ '@napi-rs/canvas@1.0.13':
4639
+ resolution: {integrity: sha512-b87ugZpQ5ViwqVBkjervxpwa0GmJPsKHLBNoZN6GpjWeHzTbDUGryriq+2DYW2XbB+xCqSIINQysC0/I87nPbw==}
4640
+ engines: {node: '>= 10'}
4641
+
4642
'@napi-rs/keyring-darwin-arm64@1.3.0':
4643
resolution: {integrity: sha512-pl76hJvdYUBn6I24bXiOBMA9nbDapo3I5B+f3OorjDU4dUMSypXeKbOVehJe8fhgTiH24flMyTS3aAIy43xegQ==}
4644
engines: {node: '>= 10'}
@@ -10375,6 +10456,10 @@ packages:
10456
pause-stream@0.0.11:
10457
resolution: {integrity: sha512-e3FBlXLmN/D1S+zHzanP4E/4Z60oFAa3O051qt1pxa7DEJWKAyil6upYVXCWadEnuoqa4Pkc9oUx9zsxYeRv8A==}
10458
10459
+ pdfjs-dist@6.4.299:
10460
+ resolution: {integrity: sha512-AVl138zALtfaAPvADulE0PZThbYzCBS79nL4pOSL/6Sm/4AH5A21BD9VHt97OlCuzJuCpmeZtAtkinisF4Vb1g==}
10461
+ engines: {node: '>=22.13.0 || >=24'}
10462
+
10463
pe-library@0.4.1:
10464
resolution: {integrity: sha512-eRWB5LBz7PpDu4PUlwT0PhnQfTQJlDDdPa35urV4Osrm0t0AqQFGn+UIkU3klZvwJ8KPO3VbBFsXquA6p6kqZw==}
10465
engines: {node: '>=12', npm: '>=6'}
@@ -14793,6 +14878,54 @@ snapshots:
14878
optionalDependencies:
14879
'@types/node': 24.12.4
14880
14881
+ '@napi-rs/canvas-android-arm64@1.0.13':
14882
+ optional: true
14883
+
14884
+ '@napi-rs/canvas-darwin-arm64@1.0.13':
14885
+ optional: true
14886
+
14887
+ '@napi-rs/canvas-darwin-x64@1.0.13':
14888
+ optional: true
14889
+
14890
+ '@napi-rs/canvas-linux-arm-gnueabihf@1.0.13':
14891
+ optional: true
14892
+
14893
+ '@napi-rs/canvas-linux-arm64-gnu@1.0.13':
14894
+ optional: true
14895
+
14896
+ '@napi-rs/canvas-linux-arm64-musl@1.0.13':
14897
+ optional: true
14898
+
14899
+ '@napi-rs/canvas-linux-riscv64-gnu@1.0.13':
14900
+ optional: true
14901
+
14902
+ '@napi-rs/canvas-linux-x64-gnu@1.0.13':
14903
+ optional: true
14904
+
14905
+ '@napi-rs/canvas-linux-x64-musl@1.0.13':
14906
+ optional: true
14907
+
14908
+ '@napi-rs/canvas-win32-arm64-msvc@1.0.13':
14909
+ optional: true
14910
+
14911
+ '@napi-rs/canvas-win32-x64-msvc@1.0.13':
14912
+ optional: true
14913
+
14914
+ '@napi-rs/canvas@1.0.13':
14915
+ optionalDependencies:
14916
+ '@napi-rs/canvas-android-arm64': 1.0.13
14917
+ '@napi-rs/canvas-darwin-arm64': 1.0.13
14918
+ '@napi-rs/canvas-darwin-x64': 1.0.13
14919
+ '@napi-rs/canvas-linux-arm-gnueabihf': 1.0.13
14920
+ '@napi-rs/canvas-linux-arm64-gnu': 1.0.13
14921
+ '@napi-rs/canvas-linux-arm64-musl': 1.0.13
14922
+ '@napi-rs/canvas-linux-riscv64-gnu': 1.0.13
14923
+ '@napi-rs/canvas-linux-x64-gnu': 1.0.13
14924
+ '@napi-rs/canvas-linux-x64-musl': 1.0.13
14925
+ '@napi-rs/canvas-win32-arm64-msvc': 1.0.13
14926
+ '@napi-rs/canvas-win32-x64-msvc': 1.0.13
14927
+ optional: true
14928
+
14929
'@napi-rs/keyring-darwin-arm64@1.3.0':
14930
optional: true
14931
@@ -20858,6 +20991,10 @@ snapshots:
20991
dependencies:
20992
through: 2.3.8
20993
20994
+ pdfjs-dist@6.4.299:
20995
+ optionalDependencies:
20996
+ '@napi-rs/canvas': 1.0.13
20997
+
20998
pe-library@0.4.1: {}
20999
21000
pele@1.8.0: {}
third-party-licenses.config.json
+12
@@ -229,6 +229,12 @@
229
}
230
],
231
"packageOverrides": [
232
+ {
233
+ "name": "entities",
234
+ "version": "8.1.0",
235
+ "license": "BSD-2-Clause",
236
+ "sourceUrl": "https://github.com/fb55/entities"
237
+ },
238
{
239
"name": "@nodable/entities",
240
"version": "2.1.1",
@@ -304,6 +310,12 @@
310
]
311
}
312
},
313
+ {
314
+ "name": "pdfjs-dist",
315
+ "version": "6.4.299",
316
+ "license": "Apache-2.0",
317
+ "sourceUrl": "https://github.com/mozilla/pdf.js"
318
+ },
319
{
320
"name": "@betteroffice/fonts",
321
"version": "0.4.0",