import { describe, expect, it } from "vitest";
import { profileJson, guessFieldSemantics, shapeSignature } from "./SchemaProfiler.ts";
import { mineEntities, classifyResponse, type CapturedResponse } from "./ResponseClassifier.ts";
import { youtubeAdapter } from "@src/connectors";

// Fixture shaped like an innertube search response (property names are *not* assumed by the profiler).
const youtubeLike = {
  contents: {
    sectionList: {
      items: [
        { videoRenderer: { videoId: "dQw4w9WgXcQ", title: { runs: [{ text: "IA au Québec — table ronde" }] }, ownerText: { runs: [{ text: "Chaîne Exemple" }] }, lengthText: { simpleText: "12:34" }, viewCountText: { simpleText: "1,2 M de vues" }, thumbnail: { thumbnails: [{ url: "https://i.ytimg.com/vi/dQw4w9WgXcQ/hqdefault.jpg", width: 480, height: 360 }] } } },
        { videoRenderer: { videoId: "abcdefghijk", title: { runs: [{ text: "Montréal AI meetup" }] }, ownerText: { runs: [{ text: "Autre Chaîne" }] }, lengthText: { simpleText: "1:02:03" }, viewCountText: { simpleText: "48 k vues" }, thumbnail: { thumbnails: [{ url: "https://i.ytimg.com/vi/abcdefghijk/hqdefault.jpg", width: 480, height: 360 }] } } },
        { channelRenderer: { channelId: "UC1234567890abcdefghij", title: { simpleText: "Québec IA" }, subscriberCountText: { simpleText: "12 k abonnés" } } },
      ],
    },
    continuation: { token: "EpMBEgVoZWxsbxqJAVNCU0NBUXFsY1ZvVFVIZG9kMlV5UmtsbGRuZG1kMkkyYjBOaGRIZG9UbGxSVUZGVmxWMFVrMWFTV2xhVjFwWGRGUlY" },
  },
};

describe("SchemaProfiler", () => {
  it("infers field semantics from value shapes and key names", () => {
    expect(guessFieldSemantics("videoId", ["dQw4w9WgXcQ", "abcdefghijk"])[0]!.kind).toBe("identifier");
    expect(guessFieldSemantics("url", ["https://i.ytimg.com/vi/x/hq.jpg"])[0]!.kind).toBe("thumbnail_url");
    expect(guessFieldSemantics("simpleText", ["12:34", "1:02:03"])[0]!.kind).toBe("duration");
    expect(guessFieldSemantics("handle", ["@simon", "@bob"])[0]!.kind).toBe("username");
    expect(guessFieldSemantics("created_utc", [1700000000, 1700000500])[0]!.kind).toBe("timestamp");
  });

  it("produces a stable shape hash independent of values", () => {
    const a = shapeSignature({ id: "1", name: "a", tags: ["x"] });
    const b = shapeSignature({ name: "zzz", id: "999", tags: ["y", "z"] });
    expect(a).toBe(b);
  });

  it("finds repeated objects and candidate entity lists", () => {
    const p = profileJson(youtubeLike);
    expect(p.repeated_object_paths).toContain("contents.sectionList.items");
    expect(p.fields.some((f) => f.path.endsWith("videoId") && f.semantic[0]!.kind === "identifier")).toBe(true);
    expect(p.candidate_entity_types.length).toBeGreaterThan(0);
  });
});

describe("mineEntities (generic JSON entity miner)", () => {
  it("extracts videos and channels with collapsed rich text and provenance", () => {
    const ents = mineEntities(youtubeLike, youtubeAdapter.classifierHints);
    const videos = ents.filter((e) => e.type === "video");
    const channels = ents.filter((e) => e.type === "channel");
    expect(videos).toHaveLength(2);
    expect(channels).toHaveLength(1);
    const v = videos.find((e) => e.platform_id === "dQw4w9WgXcQ")!;
    expect(v.name).toBe("IA au Québec — table ronde");
    expect(v.author).toBe("Chaîne Exemple");
    expect(v.media?.duration_s).toBe(12 * 60 + 34);
    expect(v.metrics?.views).toBe(1_200_000);
    expect(v.url).toBe("https://youtube.com/watch?v=dQw4w9WgXcQ");
    expect(v.provenance[0]!.surface).toBe("network");
    expect(Object.keys(v.fields)).toContain("platform_id");
  });

  it("classifies a JSON response end-to-end", () => {
    const r: CapturedResponse = { request_id: "req_1", url: "https://www.youtube.com/youtubei/v1/search?prettyPrint=false", method: "POST", status: 200, content_type: "application/json; charset=UTF-8", resource_type: "xhr", body_size: 100, body: JSON.stringify(youtubeLike), captured_at: new Date().toISOString() };
    const c = classifyResponse(r, youtubeAdapter.classifierHints);
    expect(c.kind).toBe("json");
    expect(c.entities.length).toBe(3);
    expect(c.fingerprint.observed_entity_types).toContain("video");
    expect(c.fingerprint.path_pattern).toBe("/youtubei/v1/search");
    expect(c.confidence).toBeGreaterThan(0.8);
  });
});