minimize-tools/tests/minimize/engine.test.ts

import { describe, expect, it } from "vitest";
import { minimizeInput } from "../../src/minimize/engine";
import {
  createPredicate,
  type RegexEvaluator,
} from "../../src/minimize/predicates";
import {
  bundlePredicateAdapter,
  minimizeBundle,
} from "../../src/minimize/bundle";

const unusedRegex: RegexEvaluator = async () => ({
  matched: false,
  elapsedMs: 0,
  timedOut: false,
});

describe("minimization engine", () => {
  it("removes XML structure while preserving a literal marker", async () => {
    const input = `<case><noise a="1">discard</noise><payload><message>BOOM</message><also>discard</also></payload></case>`;
    const predicate = createPredicate(
      { kind: "contains", needle: "BOOM" },
      unusedRegex,
    );
    const result = await minimizeInput(
      input,
      { structure: "xml", maxTests: 300, maxSeconds: 5 },
      predicate,
      new AbortController().signal,
    );
    expect(result.minimized).toContain("BOOM");
    expect(result.minimized.length).toBeLessThan(input.length);
    expect(
      new DOMParser()
        .parseFromString(result.minimized, "application/xml")
        .querySelector("parsererror"),
    ).toBeNull();
    expect(result.steps.length).toBeGreaterThan(0);
  });

  it("keeps the original JSON Schema failure signature", async () => {
    const input = JSON.stringify({
      request: { id: 0, label: "noise" },
      unrelated: [1, 2, 3],
    });
    const predicate = createPredicate(
      {
        kind: "json-schema-fails",
        preserveFailureSignature: true,
        schema: JSON.stringify({
          type: "object",
          required: ["request"],
          properties: {
            request: {
              type: "object",
              required: ["id"],
              properties: { id: { type: "integer", minimum: 1 } },
            },
          },
        }),
      },
      unusedRegex,
    );
    const result = await minimizeInput(
      input,
      { structure: "json", maxTests: 500, maxSeconds: 5 },
      predicate,
      new AbortController().signal,
    );
    expect(JSON.parse(result.minimized)).toEqual({ request: { id: 0 } });
  });

  it("minimizes malformed syntax and reports budget exhaustion", async () => {
    const predicate = createPredicate({ kind: "invalid-json" }, unusedRegex);
    const result = await minimizeInput(
      '{"large": [1,2,}',
      { structure: "json", maxTests: 1, maxSeconds: 5 },
      predicate,
      new AbortController().signal,
    );
    expect(result.exhausted).toBe(true);
    expect(result.tests).toBe(1);
  });

  it("rejects a baseline that does not reproduce and honours cancellation", async () => {
    await expect(
      minimizeInput(
        "safe",
        { structure: "text", maxTests: 20, maxSeconds: 2 },
        createPredicate({ kind: "contains", needle: "BOOM" }, unusedRegex),
        new AbortController().signal,
      ),
    ).rejects.toThrow(/does not satisfy/u);
    const controller = new AbortController();
    controller.abort();
    await expect(
      minimizeInput(
        "BOOM",
        { structure: "text", maxTests: 20, maxSeconds: 2 },
        createPredicate({ kind: "contains", needle: "BOOM" }, unusedRegex),
        controller.signal,
      ),
    ).rejects.toMatchObject({ name: "AbortError" });
  });

  it("uses isolated regex outcomes and fails closed on unsafe schemas/XSLT", async () => {
    const slow = createPredicate(
      { kind: "regex-slow", pattern: "a+", flags: "u", thresholdMs: 20 },
      async () => ({ matched: false, elapsedMs: 20, timedOut: false }),
    );
    expect(await slow("aaaa", new AbortController().signal)).toBe(true);
    expect(() =>
      createPredicate(
        {
          kind: "json-schema-fails",
          schema: '{"$ref":"https://invalid/schema.json"}',
        },
        unusedRegex,
      ),
    ).toThrow(/unsupported.*\$ref/iu);
    expect(() =>
      createPredicate(
        {
          kind: "xslt-throws",
          stylesheet:
            '<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform"><xsl:include href="https://invalid/x.xsl"/></xsl:stylesheet>',
        },
        unusedRegex,
      ),
    ).toThrow(/include/u);
  });

  it("samples a flaky predicate according to an explicit stability policy", async () => {
    const observations = new Map<string, number>();
    const result = await minimizeInput(
      "noise BOOM noise",
      {
        structure: "text",
        maxTests: 300,
        maxSeconds: 5,
        stability: { samples: 3, requiredPasses: 2 },
      },
      async (candidate) => {
        const count = (observations.get(candidate) ?? 0) + 1;
        observations.set(candidate, count);
        return candidate.includes("BOOM") && count % 3 !== 0;
      },
      new AbortController().signal,
    );
    expect(result.minimized).toContain("BOOM");
    expect(result.stability).toEqual({ samples: 3, requiredPasses: 2 });
    expect(result.unstableCandidates).toBeGreaterThan(0);
  });

  it("removes files and then minimizes retained bundle content", async () => {
    const predicate = bundlePredicateAdapter(
      createPredicate({ kind: "contains", needle: "BOOM" }, unusedRegex),
      "concatenated",
    );
    const result = await minimizeBundle(
      [
        { path: "noise.txt", content: "discard all of this" },
        { path: "case/input.txt", content: "prefix BOOM suffix" },
      ],
      { maxTests: 400, maxSeconds: 5 },
      predicate,
      new AbortController().signal,
    );
    expect(result.minimized).toHaveLength(1);
    expect(result.minimized[0]).toMatchObject({ path: "case/input.txt" });
    expect(result.minimized[0]?.content).toContain("BOOM");
    expect(result.steps.some((step) => step.stage === "files")).toBe(true);
    expect(result.steps.some((step) => step.stage === "content")).toBe(true);
  });
});