Release Privacy Tools 0.1.0

src/privacy/archive.ts

@@ -0,0 +1,121 @@
+import {
+  encodeText,
+  sanitizeDownloadFilename,
+  stableStringify,
+} from "@add-ideas/toolbox-helpers";
+import { zipSync } from "fflate";
+
+import { APP_VERSION } from "../version";
+import {
+  assertLimit,
+  DEFAULT_PRIVACY_LIMITS,
+  PrivacyLimitError,
+} from "./limits";
+import type {
+  BatchReport,
+  ImageScanResult,
+  SanitizedAsset,
+  SanitizationReport,
+} from "./model";
+
+export function createBatchReport(
+  files: readonly ImageScanResult[],
+  assets: readonly SanitizedAsset[],
+  generatedAt = new Date().toISOString(),
+): BatchReport {
+  return {
+    schemaVersion: 1,
+    generatedAt,
+    application: { name: "Privacy Tools", version: APP_VERSION },
+    files: [...files],
+    sanitizations: assets.map((asset) => asset.report),
+    warnings: [
+      "This report may itself contain sensitive source metadata and filenames; store and share it deliberately.",
+      "A successful re-scan is not an anonymity guarantee; see each sanitization disclaimer.",
+    ],
+  };
+}
+
+export function serializeReport(
+  report: BatchReport | SanitizationReport,
+): string {
+  return stableStringify(report, 2, {
+    maxDepth: 64,
+    maxNodes: 500_000,
+    maxTextChars: 2_000_000,
+  });
+}
+
+export async function createBatchArchive(
+  files: readonly ImageScanResult[],
+  assets: readonly SanitizedAsset[],
+  generatedAt = new Date().toISOString(),
+): Promise<Blob> {
+  assertLimit(
+    files.length,
+    DEFAULT_PRIVACY_LIMITS.maxFiles,
+    "Report file count",
+  );
+  assertLimit(
+    assets.length,
+    DEFAULT_PRIVACY_LIMITS.maxFiles,
+    "Archive image count",
+  );
+  const entries: Record<string, Uint8Array> = Object.create(null) as Record<
+    string,
+    Uint8Array
+  >;
+  const names = new Set<string>();
+  let total = 0;
+  for (const asset of assets) {
+    const name = uniqueName(asset.report.outputName, names);
+    const bytes = new Uint8Array(await asset.blob.arrayBuffer());
+    total += bytes.byteLength;
+    if (total > DEFAULT_PRIVACY_LIMITS.maxZipBytes)
+      throw new PrivacyLimitError(
+        "Batch archive input size",
+        total,
+        DEFAULT_PRIVACY_LIMITS.maxZipBytes,
+      );
+    entries[`images/${name}`] = bytes;
+  }
+  const report = encodeText(
+    serializeReport(createBatchReport(files, assets, generatedAt)),
+  );
+  total += report.byteLength;
+  if (total > DEFAULT_PRIVACY_LIMITS.maxZipBytes)
+    throw new PrivacyLimitError(
+      "Batch archive input size",
+      total,
+      DEFAULT_PRIVACY_LIMITS.maxZipBytes,
+    );
+  entries["privacy-tools-report.json"] = report;
+  const archive = zipSync(entries, {
+    level: 0,
+    mtime: new Date("1980-01-01T00:00:00.000Z"),
+  });
+  if (archive.byteLength > DEFAULT_PRIVACY_LIMITS.maxZipBytes)
+    throw new PrivacyLimitError(
+      "Batch archive output size",
+      archive.byteLength,
+      DEFAULT_PRIVACY_LIMITS.maxZipBytes,
+    );
+  const ownedArchive = archive.slice().buffer as ArrayBuffer;
+  return new Blob([ownedArchive], { type: "application/zip" });
+}
+
+function uniqueName(input: string, used: Set<string>): string {
+  const safe = sanitizeDownloadFilename(input, "image.clean");
+  if (!used.has(safe)) {
+    used.add(safe);
+    return safe;
+  }
+  const dot = safe.lastIndexOf(".");
+  const stem = dot > 0 ? safe.slice(0, dot) : safe;
+  const extension = dot > 0 ? safe.slice(dot) : "";
+  let counter = 2;
+  while (used.has(`${stem}-${counter}${extension}`)) counter += 1;
+  const candidate = `${stem}-${counter}${extension}`;
+  used.add(candidate);
+  return candidate;
+}