Release Format Lab 0.2.0

2026-09-02 07:16:08 +02:00
parent cfc525a88e
commit c40a2c4b2a
29 changed files with 972 additions and 64 deletions
@@ -0,0 +1,47 @@
+import { describe, expect, it } from "vitest";
+import { exportCorpusFiles } from "../../src/lab/corpus";
+import {
+  formatSignatures,
+  identifyBySignature,
+} from "../../src/lab/signatures";
+
+describe("format signatures and corpus", () => {
+  it("distinguishes masked RIFF forms and container-only evidence", () => {
+    const wav = Uint8Array.of(
+      0x52,
+      0x49,
+      0x46,
+      0x46,
+      0x24,
+      0,
+      0,
+      0,
+      0x57,
+      0x41,
+      0x56,
+      0x45,
+    );
+    expect(identifyBySignature(wav).map((item) => item.format)).toContain(
+      "wav",
+    );
+    expect(identifyBySignature(wav).map((item) => item.format)).not.toContain(
+      "webp",
+    );
+    expect(
+      identifyBySignature(Uint8Array.of(0x50, 0x4b, 0x03, 0x04))[0],
+    ).toMatchObject({ format: "zip", confidence: "container" });
+  });
+
+  it("ships a broad, deterministic structured regression corpus", () => {
+    expect(formatSignatures.length).toBeGreaterThanOrEqual(20);
+    const first = exportCorpusFiles();
+    expect(first).toEqual(exportCorpusFiles());
+    expect(first.map((file) => file.path)).toEqual(
+      expect.arrayContaining([
+        "structured/typed.json",
+        "structured/text-cells.csv",
+        "corpus-manifest.json",
+      ]),
+    );
+  });
+});