file-tools/tests/file/core.test.ts

import { describe, expect, it } from "vitest";
import { manifestCsv, manifestJson } from "../../src/file/manifest";
import {
  checksumManifest,
  joinBlobs,
  planBatchRename,
  splitBlob,
} from "../../src/file/operations";
import {
  detectKnownSignature,
  extensionMismatch,
} from "../../src/file/signatures";
import {
  extractStrings,
  looksLikeText,
  shannonEntropy,
} from "../../src/file/text";

describe("file signatures", () => {
  it("detects PNG without claiming parser validation", () =>
    expect(
      detectKnownSignature(
        Uint8Array.of(0x89, 0x50, 0x4e, 0x47, 0x0d, 0x0a, 0x1a, 0x0a),
      ),
    ).toMatchObject({ extension: "png", mime: "image/png" }));
  it("accounts for container extension aliases", () => {
    expect(extensionMismatch("book.epub", "zip")).toBeUndefined();
    expect(extensionMismatch("image.txt", "png")).toMatch(/does not match/u);
  });
});

describe("bounded text helpers", () => {
  it("extracts ASCII and UTF-16LE strings", () =>
    expect(
      extractStrings(
        Uint8Array.from([65, 66, 67, 68, 0, 90, 0, 89, 0, 88, 0, 87, 0]),
      ),
    ).toEqual(expect.arrayContaining(["ABCD", "ZYXW"])));
  it("distinguishes NUL-containing binary data", () =>
    expect(looksLikeText(Uint8Array.of(65, 0, 66))).toBe(false));
  it("measures a uniform byte distribution", () =>
    expect(
      shannonEntropy(Uint8Array.from({ length: 256 }, (_, index) => index)),
    ).toBeCloseTo(8));
});

describe("manifests", () => {
  const records = [
    {
      name: "=formula",
      path: "=formula",
      size: 1,
      type: "text/plain",
      lastModified: 0,
    },
  ];
  it("is deterministic", () =>
    expect(manifestJson(records)).toBe(manifestJson(records)));
  it("neutralises spreadsheet-leading formulas", () =>
    expect(manifestCsv(records)).toContain("'=formula"));
  it("reports truthful inspection coverage and per-file failures", () => {
    const text = manifestJson([
      {
        ...records[0]!,
        inspectionStatus: "inspected",
        inspection: {
          evidence: [],
          entropy: 0,
          sampleBytes: 0,
          strings: [],
          findings: [],
        },
      },
      {
        ...records[0]!,
        path: "broken.bin",
        inspectionStatus: "error",
        inspectionError: "Timed out",
      },
    ]);
    const manifest = JSON.parse(text) as {
      inspectionCoverage: { inspected: number; failed: number };
      files: Array<{
        inspectionStatus: string;
        inspectionError: string | null;
      }>;
    };
    expect(manifest.inspectionCoverage).toMatchObject({
      inspected: 1,
      failed: 1,
    });
    expect(
      manifest.files.find((file) => file.inspectionStatus === "error"),
    ).toMatchObject({ inspectionError: "Timed out" });
    expect(
      manifestCsv([{ ...records[0]!, inspectionStatus: "cancelled" }]),
    ).toContain('"cancelled"');
  });
});

describe("bounded file operations", () => {
  it("plans sanitized collision-safe rename downloads deterministically", () => {
    const files = [
      { name: "first.txt", blob: new Blob(["one"]) },
      { name: "second.txt", blob: new Blob(["two"]) },
    ];
    const plan = planBatchRename(files, "same.{ext}");
    expect(plan.map((item) => item.filename)).toEqual([
      "same.txt",
      "same (2).txt",
    ]);
    expect(plan.map((item) => item.originalName)).toEqual([
      "first.txt",
      "second.txt",
    ]);
  });

  it("splits and rejoins byte-exact Blob parts with an offset manifest", async () => {
    const source = new Blob([new Uint8Array(150_000).map((_, index) => index)]);
    const split = splitBlob(source, "payload.bin", 65_536);
    expect(split.parts.map((part) => part.size)).toEqual([
      65_536, 65_536, 18_928,
    ]);
    expect(split.manifest.parts[1]).toMatchObject({ offset: 65_536, index: 2 });
    const joined = joinBlobs(split.parts.map((part) => part.blob));
    expect(new Uint8Array(await joined.arrayBuffer())).toEqual(
      new Uint8Array(await source.arrayBuffer()),
    );
  });

  it("refuses unbounded split/join and emits sorted checksum files", () => {
    expect(() => splitBlob(new Blob(["x"]), "x", 1)).toThrow(/Chunk size/u);
    expect(() =>
      joinBlobs([new Blob(["abc"]), new Blob(["def"])], "", 5),
    ).toThrow(/exceeds/u);
    const output = checksumManifest(
      [
        {
          name: "b",
          path: "b",
          size: 0,
          type: "",
          lastModified: 0,
          sha256: "b".repeat(64),
        },
        {
          name: "a",
          path: "a",
          size: 0,
          type: "",
          lastModified: 0,
          sha256: "a".repeat(64),
        },
      ],
      "sha256",
    );
    expect(output.split("\n")[0]).toBe(`${"a".repeat(64)} *a`);
  });
});