timmy-talking-turd/tests/image-dataurl.test.js

// Canonical base64 data-URL grammar contract.
//
// The client always produces a canonical `data:<mime>;base64,<canonical b64>`
// URL (canvas.toDataURL). Anything else is a hand-crafted request, so ingress
// must require the canonical grammar and a byte-exact round trip rather than
// silently repairing missing padding, embedded newlines, or non-zero trailing
// pad bits. The exact supported MIME policy is unchanged: JPEG/PNG/WebP only.
import test from 'node:test';
import assert from 'node:assert/strict';
import { readFile } from 'node:fs/promises';
import { fileURLToPath } from 'node:url';

import { decodeImageDataUrl, validateImageIngress } from '../src/image-ingress.js';

const fixture = (name) => fileURLToPath(new URL(`../tests/fixtures/${name}`, import.meta.url));
const clean = await readFile(fixture('ingress-clean.jpg'));
const canonical = clean.toString('base64');

const ingest = (imageDataUrl) => validateImageIngress({ consent: true, imageDataUrl });

test('canonical base64 data URLs round-trip byte-exactly', () => {
  const decoded = decodeImageDataUrl(`data:image/jpeg;base64,${canonical}`);
  assert.ok(decoded, 'canonical data URL must decode');
  assert.equal(decoded.mime, 'image/jpeg');
  assert.ok(decoded.bytes.equals(clean), 'decoded bytes must equal the source bytes exactly');
});

test('missing base64 padding is rejected', async () => {
  const unpadded = canonical.replace(/=+$/, '');
  assert.notEqual(unpadded, canonical, 'fixture must actually require padding');
  assert.equal(decodeImageDataUrl(`data:image/jpeg;base64,${unpadded}`), null);
  await assert.rejects(() => ingest(`data:image/jpeg;base64,${unpadded}`), /not a supported image/i);
});

test('excess base64 padding is rejected', async () => {
  assert.equal(decodeImageDataUrl(`data:image/jpeg;base64,${canonical}===`), null);
  await assert.rejects(() => ingest(`data:image/jpeg;base64,${canonical}====`), /not a supported image/i);
});

test('CR and LF inside the base64 payload are rejected', async () => {
  const withCrlf = `${canonical.slice(0, 40)}\r\n${canonical.slice(40)}`;
  const withLf = `${canonical.slice(0, 40)}\n${canonical.slice(40)}`;
  const withCr = `${canonical.slice(0, 40)}\r${canonical.slice(40)}`;
  for (const payload of [withCrlf, withLf, withCr]) {
    assert.equal(decodeImageDataUrl(`data:image/jpeg;base64,${payload}`), null);
    await assert.rejects(() => ingest(`data:image/jpeg;base64,${payload}`), /not a supported image/i);
  }
});

test('non-canonical encodings are rejected: non-zero trailing pad bits, whitespace, URL-safe alphabet', async () => {
  const nonZeroPadBits = `${canonical.slice(0, -2)}/=`;
  const withSpace = `${canonical.slice(0, 20)} ${canonical.slice(20)}`;
  const urlSafe = canonical.replace(/\+/g, '-').replace(/\//g, '_');
  for (const payload of [nonZeroPadBits, withSpace]) {
    assert.equal(decodeImageDataUrl(`data:image/jpeg;base64,${payload}`), null);
    await assert.rejects(() => ingest(`data:image/jpeg;base64,${payload}`), /not a supported image/i);
  }
  if (urlSafe !== canonical) {
    assert.equal(decodeImageDataUrl(`data:image/jpeg;base64,${urlSafe}`), null);
  }
});

test('the supported MIME policy is exact and unchanged', async () => {
  for (const mime of ['image/jpeg', 'image/png', 'image/webp']) {
    assert.ok(decodeImageDataUrl(`data:${mime};base64,${canonical}`), `${mime} must remain supported`);
  }
  for (const mime of ['image/gif', 'image/svg+xml', 'text/html', 'application/octet-stream', '']) {
    assert.equal(decodeImageDataUrl(`data:${mime};base64,${canonical}`), null, `${mime} must not be supported`);
    await assert.rejects(() => ingest(`data:${mime};base64,${canonical}`), /not a supported image/i);
  }
});

test('malformed data-URL envelopes are rejected without repair', async () => {
  const malformed = [
    `data:image/jpeg,${canonical}`,               // no base64 token
    `data:image/jpeg;base64;${canonical}`,        // wrong separator
    `DATA:image/jpeg;base64,${canonical}`,        // scheme casing is not canonical here
    `data:image/jpeg;base64,`,                    // empty payload
    ` data:image/jpeg;base64,${canonical}`,       // leading whitespace
    `data:image/jpeg;base64,${canonical} `,       // trailing whitespace
    `data:image/jpeg;charset=utf-8;base64,${canonical}`, // extra parameters
    canonical,                                    // bare base64, no envelope
  ];
  for (const value of malformed) {
    assert.equal(decodeImageDataUrl(value), null, `must reject: ${value.slice(0, 42)}`);
    await assert.rejects(() => ingest(value), /not a supported image/i);
  }
});

test('a canonical supported data URL still completes ingress end to end', async () => {
  const result = await ingest(`data:image/jpeg;base64,${canonical}`);
  assert.equal(result.format, 'jpeg');
  assert.equal(result.mime, 'image/jpeg');
  assert.match(result.imageDataUrl, /^data:image\/jpeg;base64,[A-Za-z0-9+/]+={0,2}$/);
});