Void-Homelab/tests/repos/search.test.js

import { describe, it, expect, beforeAll, beforeEach, afterEach, vi } from 'vitest';
import { resetDb } from '../helpers/db.js';
import { migrateUp } from '../../lib/db/migrate.js';
import { pool } from '../../lib/db/pool.js';
import * as search from '../../lib/db/repos/search.js';
import * as spacesRepo from '../../lib/db/repos/spaces.js';
import * as pagesRepo from '../../lib/db/repos/pages.js';
import * as refsRepo from '../../lib/db/repos/refs.js';
import * as resourcesRepo from '../../lib/db/repos/resources.js';
import * as sourceDocsRepo from '../../lib/db/repos/source_docs.js';
import * as conversationsRepo from '../../lib/db/repos/conversations.js';
import * as messagesRepo from '../../lib/db/repos/messages.js';

const owner = { kind: 'user', id: null };
let space, otherSpace;

beforeAll(async () => { await resetDb(); await migrateUp(); });
beforeEach(async () => {
  await resetDb();
  await migrateUp();
  space = await spacesRepo.create({ slug: 's-main', name: 'Main' }, owner);
  otherSpace = await spacesRepo.create({ slug: 's-other', name: 'Other' }, owner);
  // Default: pretend Ollama is down so the vector branch is skipped and
  // existing FTS-only assertions still hold deterministically.
  global.fetch = vi.fn(async () => { throw new Error('Ollama unreachable (test default)'); });
});
afterEach(() => { vi.restoreAllMocks(); });

async function seedAll(word) {
  await pagesRepo.create(
    { space_id: space.id, slug: 'pg-search', title: `${word} page`, body_md: `body about ${word}` }, owner
  );
  await refsRepo.create(
    {
      space_id: space.id, kind: 'url', source_url: 'https://example.com/x',
      title: `${word} reference`, body_text: `text mentioning ${word}`
    },
    owner
  );
  const res = await resourcesRepo.create(
    { space_id: space.id, slug: 'r-search', name: 'Res', runtime_type: 'lxc' }, owner
  );
  await sourceDocsRepo.create(
    {
      resource_id: res.id, name: `${word} source doc`,
      upstream_url: 'https://example.com/sd', body_text: `doc body about ${word}`
    },
    owner
  );
  const conv = await conversationsRepo.create({ title: 'chat' }, owner);
  await messagesRepo.append(conv.id, { role: 'user', body: `let's talk about ${word}` });
}

describe('search repo', () => {
  it('fts returns 4 hits across all kinds for the query word', async () => {
    await seedAll('blackflame');
    const hits = await search.fts({ q: 'blackflame' });
    expect(hits.length).toBe(4);
    const kinds = new Set(hits.map(h => h.kind));
    expect(kinds).toEqual(new Set(['page','ref','source_doc','message']));
    for (const h of hits) {
      expect(typeof h.id).toBe('string');
      expect(typeof h.title_or_snippet).toBe('string');
      expect(typeof h.rank).toBe('number');
    }
  });

  it('kinds filter narrows to requested branches', async () => {
    await seedAll('blackflame');
    const hits = await search.fts({ q: 'blackflame', kinds: ['page','ref'] });
    expect(hits.length).toBe(2);
    expect(new Set(hits.map(h => h.kind))).toEqual(new Set(['page','ref']));
  });

  it('space_id filter scopes pages/refs/source_docs (messages excluded)', async () => {
    await seedAll('blackflame');
    await pagesRepo.create(
      { space_id: otherSpace.id, slug: 'pg-other', title: 'blackflame other', body_md: '' }, owner
    );
    const hits = await search.fts({ q: 'blackflame', space_id: space.id });
    // page+ref+source_doc from main space; no messages (no space); no other-space page
    expect(hits.length).toBe(3);
    expect(hits.every(h => !h.kind.includes('message'))).toBe(true);
    expect(hits.every(h => h.space_id === space.id)).toBe(true);
  });

  it('orders by rank desc and respects limit + offset', async () => {
    await seedAll('blackflame');
    const all = await search.fts({ q: 'blackflame', limit: 100 });
    const limited = await search.fts({ q: 'blackflame', limit: 2 });
    expect(limited.length).toBe(2);
    const ranks = all.map(h => h.rank);
    const sorted = [...ranks].sort((a, b) => b - a);
    expect(ranks).toEqual(sorted);
    const off = await search.fts({ q: 'blackflame', limit: 2, offset: 2 });
    expect(off.length).toBe(2);
    expect(off[0].id).not.toBe(limited[0].id);
  });

  it('returns empty when nothing matches', async () => {
    await seedAll('blackflame');
    const hits = await search.fts({ q: 'whitelight' });
    expect(hits).toEqual([]);
  });

  it('vector branch surfaces an FTS-miss when embedding is close to the query', async () => {
    // Page text does not include "blackflame", but its hand-crafted vector
    // is close to the query vector, so the vector branch should surface it.
    const page = await pagesRepo.create(
      { space_id: space.id, slug: 'vec-only', title: 'Unrelated', body_md: 'nothing about it' },
      owner
    );
    const v = '[' + new Array(1024).fill(0.5).join(',') + ']';
    await pool.query('UPDATE pages SET embedding=$1::vector WHERE id=$2', [v, page.id]);
    global.fetch = vi.fn(async () => new Response(
      JSON.stringify({ embedding: new Array(768).fill(0.5) }),
      { status: 200, headers: { 'content-type': 'application/json' } }
    ));
    const hits = await search.fts({ q: 'whatever' });
    expect(hits.find(h => h.id === page.id)).toBeTruthy();
  });

  it('Ollama down → FTS-only fallback still returns FTS hits', async () => {
    await pagesRepo.create(
      { space_id: space.id, slug: 'fb', title: 'blackflame palette', body_md: '' },
      owner
    );
    // Default mock already throws — that simulates Ollama being unreachable.
    const hits = await search.fts({ q: 'blackflame' });
    expect(hits.length).toBeGreaterThan(0);
  });

  it('RRF fuses FTS and vector for the same row (higher rank than FTS alone)', async () => {
    const page = await pagesRepo.create(
      { space_id: space.id, slug: 'rrf', title: 'cradle blackflame motif', body_md: 'blackflame essay' },
      owner
    );
    const v = '[' + new Array(1024).fill(0.5).join(',') + ']';
    await pool.query('UPDATE pages SET embedding=$1::vector WHERE id=$2', [v, page.id]);

    // FTS-only run (vector branch errors)
    const ftsOnly = await search.fts({ q: 'blackflame' });
    const ftsRank = ftsOnly.find(h => h.id === page.id)?.rank;
    expect(ftsRank).toBeGreaterThan(0);

    // FTS + vector (query embedding matches the row's vector)
    global.fetch = vi.fn(async () => new Response(
      JSON.stringify({ embedding: new Array(768).fill(0.5) }),
      { status: 200, headers: { 'content-type': 'application/json' } }
    ));
    const hybrid = await search.fts({ q: 'blackflame' });
    const hybridRank = hybrid.find(h => h.id === page.id)?.rank;
    expect(hybridRank).toBeGreaterThan(ftsRank);
  });
});