S9 (a1): slip reader — zero-API OCR

tesseract.js (self-hosted WASM, Apache-2.0) + pure per-book layout parsers (DK/FD/MGM/Caesars) with per-field confidence and needs_review honesty — the reader never guesses. POST /api/slips/parse (auth, free 1/day paid 10/day, 4MB cap) + Next proxy. Gated /slip page: upload or paste, manual-correct UI, per-leg grades through the normal engine (refusals render honestly), add-all to Parlay Lab, share card. Vision model upgrade logged post-revenue. 2574 -> 2608 tests. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-07-11 19:32:17 -04:00
parent aaafc3e0f2
commit 4e49ee0990
18 changed files with 1914 additions and 1 deletions
@@ -0,0 +1,220 @@
+// A1 S9 — Slip Reader parser suite. PURE parsers on realistic OCR-text
+// fixtures per book. The contract under test: correct extraction on the
+// rigid slip grammars, per-field confidence, and the NEVER-GUESS rule
+// (unreadable → null + needs_review, never a fabricated value).
+
+const fs = require('fs');
+const path = require('path');
+const slipReader = require('../../src/services/slipReader');
+const { parseSlipText, detectBook, CONFIDENCE_THRESHOLD } = slipReader;
+const { normalizeStat, splitPlayerStat, parseOdds, buildLeg } = slipReader.__internals;
+
+const fixture = (book) =>
+  fs.readFileSync(path.join(__dirname, '..', 'fixtures', 'slips', `${book}.txt`), 'utf8');
+
+describe('slipReader — book detection', () => {
+  it.each([
+    ['draftkings'], ['fanduel'], ['betmgm'], ['caesars'],
+  ])('detects %s from its fixture', (book) => {
+    expect(detectBook(fixture(book))).toBe(book);
+  });
+
+  it('returns null for unbranded text', () => {
+    expect(detectBook('Aaron Judge Over 1.5 Total Bases')).toBeNull();
+  });
+});
+
+describe('slipReader — DraftKings layout', () => {
+  const result = parseSlipText(fixture('draftkings'));
+
+  it('extracts all three legs completely', () => {
+    expect(result.book).toBe('draftkings');
+    expect(result.legs).toHaveLength(3);
+    expect(result.needs_review).toBe(false);
+    expect(result.source).toBe('user_slip');
+  });
+
+  it('extracts player/stat/line/side/odds on leg 1', () => {
+    const leg = result.legs[0];
+    expect(leg.player).toBe('Aaron Judge');
+    expect(leg.player_key).toBe('aaron judge');
+    expect(leg.stat).toBe('total_bases');
+    expect(leg.line).toBe(1.5);
+    expect(leg.side).toBe('over');
+    expect(leg.odds).toBe(-115);
+    expect(leg.needs_review).toBe(false);
+  });
+
+  it('normalizes book market labels through the stat vocabulary', () => {
+    // "Strikeouts Thrown" (DK pitcher label) → strikeouts
+    expect(result.legs[1].stat).toBe('strikeouts');
+    expect(result.legs[1].line).toBe(7.5);
+    // plus-odds leg
+    expect(result.legs[2].stat).toBe('home_runs');
+    expect(result.legs[2].odds).toBe(320);
+  });
+
+  it('carries per-field confidences at or above threshold on clean legs', () => {
+    for (const leg of result.legs) {
+      for (const field of ['player', 'stat', 'line', 'side', 'odds']) {
+        expect(leg.confidence[field]).toBeGreaterThanOrEqual(CONFIDENCE_THRESHOLD);
+      }
+    }
+  });
+});
+
+describe('slipReader — FanDuel layout', () => {
+  const result = parseSlipText(fixture('fanduel'));
+
+  it('parses To Record N+ / Any Time / Over grammars', () => {
+    expect(result.book).toBe('fanduel');
+    expect(result.legs).toHaveLength(3);
+    // "To Record 2+ Total Bases" → line 1.5 over
+    expect(result.legs[0]).toMatchObject({ player: 'Aaron Judge', stat: 'total_bases', line: 1.5, side: 'over' });
+    // "Any Time Home Run" → home_runs 0.5 over
+    expect(result.legs[1]).toMatchObject({ player: 'Giancarlo Stanton', stat: 'home_runs', line: 0.5, side: 'over' });
+    // plain Over grammar
+    expect(result.legs[2]).toMatchObject({ player: 'Gerrit Cole', stat: 'strikeouts', line: 6.5, side: 'over' });
+  });
+
+  it('SGP legs without per-leg odds → odds null + needs_review (never guessed)', () => {
+    for (const leg of result.legs) {
+      expect(leg.odds).toBeNull();
+      expect(leg.confidence.odds).toBeLessThan(CONFIDENCE_THRESHOLD);
+      expect(leg.needs_review).toBe(true);
+    }
+    expect(result.needs_review).toBe(true);
+  });
+});
+
+describe('slipReader — BetMGM layout', () => {
+  const result = parseSlipText(fixture('betmgm'));
+
+  it('parses inline "@ odds" and trailing-odds legs', () => {
+    expect(result.book).toBe('betmgm');
+    expect(result.legs).toHaveLength(3);
+    expect(result.legs[0]).toMatchObject({ player: 'Aaron Judge', stat: 'total_bases', line: 1.5, side: 'over', odds: -115 });
+    expect(result.legs[1]).toMatchObject({ player: 'Shohei Ohtani', stat: 'strikeouts', line: 7.5, side: 'under', odds: 105 });
+    expect(result.legs[2]).toMatchObject({ player: 'Mookie Betts', stat: 'hits', line: 1.5, side: 'over', odds: 140 });
+    expect(result.needs_review).toBe(false);
+  });
+});
+
+describe('slipReader — Caesars layout', () => {
+  const result = parseSlipText(fixture('caesars'));
+
+  it('splits the ambiguous player/stat head on known stat aliases', () => {
+    expect(result.book).toBe('caesars');
+    expect(result.legs).toHaveLength(3);
+    // "Pete Alonso Home Runs" must NOT become player "Pete Alonso Home" + stat "runs"
+    expect(result.legs[1]).toMatchObject({ player: 'Pete Alonso', stat: 'home_runs', line: 0.5, side: 'over', odds: 340 });
+    expect(result.legs[0]).toMatchObject({ player: 'Aaron Judge', stat: 'total_bases', odds: -115 });
+    expect(result.legs[2]).toMatchObject({ player: 'Freddie Freeman', stat: 'hits', side: 'under', odds: -105 });
+  });
+
+  it('unknown stat head → stat AND player null (boundary unknowable), needs_review', () => {
+    const r = parseSlipText('Caesars\nAaron Judge Fantasy Score Over 32.5 (-115)');
+    expect(r.legs).toHaveLength(1);
+    expect(r.legs[0].stat).toBeNull();
+    expect(r.legs[0].player).toBeNull(); // boundary uncertain → below threshold → nulled
+    expect(r.legs[0].needs_review).toBe(true);
+    // The readable fields still come through — nothing over-nulled.
+    expect(r.legs[0].line).toBe(32.5);
+    expect(r.legs[0].odds).toBe(-115);
+  });
+});
+
+describe('slipReader — never guess', () => {
+  it('garbage text → zero legs, needs_review envelope, no fabrication', () => {
+    const r = parseSlipText('completely unrelated text\nnothing to see 12345\nlorem ipsum');
+    expect(r.legs).toEqual([]);
+    expect(r.needs_review).toBe(true);
+    expect(r.book).toBeNull();
+  });
+
+  it('empty/nullish input is safe', () => {
+    expect(parseSlipText('').legs).toEqual([]);
+    expect(parseSlipText(null).legs).toEqual([]);
+    expect(parseSlipText(undefined).legs).toEqual([]);
+  });
+
+  it('a below-threshold field is nulled, not passed through', () => {
+    const leg = buildLeg({
+      player: 'judge', // lowercase single token → low confidence
+      statLabel: 'Total Bases',
+      line: 1.5,
+      side: 'over',
+      odds: -115,
+    });
+    expect(leg.player).toBeNull();
+    expect(leg.player_key).toBeNull();
+    expect(leg.needs_review).toBe(true);
+    expect(leg.stat).toBe('total_bases'); // readable fields survive
+  });
+
+  it('book hint routes to the hinted layout parser', () => {
+    const dk = fixture('draftkings');
+    const hinted = parseSlipText(dk, 'draftkings');
+    expect(hinted.book).toBe('draftkings');
+    expect(hinted.legs).toHaveLength(3);
+  });
+
+  it('unknown book → best layout wins without inventing values', () => {
+    const unbranded = fixture('draftkings').replace(/DraftKings Sportsbook\n/, '');
+    const r = parseSlipText(unbranded);
+    expect(r.legs.length).toBeGreaterThanOrEqual(3);
+    for (const leg of r.legs) {
+      expect(leg.player).not.toBeNull();
+      expect(leg.stat).not.toBeNull();
+    }
+  });
+});
+
+describe('slipReader — vocabulary + helpers', () => {
+  it('normalizeStat maps slip labels to canonical stat_types', () => {
+    expect(normalizeStat('Total Bases')).toBe('total_bases');
+    expect(normalizeStat('Strikeouts Thrown')).toBe('strikeouts');
+    expect(normalizeStat('Alt Total Bases')).toBe('total_bases');
+    expect(normalizeStat('Total Bases O/U')).toBe('total_bases');
+    expect(normalizeStat('Pts + Reb + Ast')).toBe('pra');
+    expect(normalizeStat('3 Pointers Made')).toBe('threes');
+    expect(normalizeStat('Runs Batted In')).toBe('rbi');
+    expect(normalizeStat('Fantasy Score')).toBeNull(); // unknown → null, never a guess
+  });
+
+  it('canonical stat_types match the scan-route vocabulary exactly', () => {
+    // Mirror of VALID_STAT_TYPES in src/routes/scan.js — a drifted alias
+    // here would 400 at the grade gate.
+    const VALID = new Set([
+      'points', 'rebounds', 'assists', 'threes', 'blocks', 'steals', 'pra', 'turnovers',
+      'goals', 'shots_on_target', 'shots', 'tackles', 'cards', 'corners', 'saves',
+      'goals_conceded', 'passes', 'clean_sheet',
+      'strikeouts', 'hits', 'home_runs', 'rbi', 'total_bases', 'walks', 'runs',
+      'earned_runs', 'innings_pitched', 'hits_allowed', 'stolen_bases', 'doubles', 'outs',
+    ]);
+    for (const stat of Object.values(slipReader.__internals.STAT_ALIASES)) {
+      expect(VALID.has(stat)).toBe(true);
+    }
+  });
+
+  it('splitPlayerStat prefers the longest alias suffix', () => {
+    expect(splitPlayerStat('Pete Alonso Home Runs')).toMatchObject({ player: 'Pete Alonso', statLabel: 'Home Runs' });
+    expect(splitPlayerStat('Aaron Judge Total Bases')).toMatchObject({ player: 'Aaron Judge', statLabel: 'Total Bases' });
+    expect(splitPlayerStat('Aaron Judge Alt Total Bases')).toMatchObject({ player: 'Aaron Judge', statLabel: 'Total Bases' });
+  });
+
+  it('parseOdds normalizes OCR minus glyphs and rejects non-odds', () => {
+    expect(parseOdds('-115')).toBe(-115);
+    expect(parseOdds('−115')).toBe(-115); // unicode minus
+    expect(parseOdds('+320')).toBe(320);
+    expect(parseOdds('-15')).toBeNull(); // |odds| < 100 is a line, not odds
+    expect(parseOdds('banana')).toBeNull();
+    expect(parseOdds(null)).toBeNull();
+  });
+
+  it('player names normalize through playerName (display + key)', () => {
+    const r = parseSlipText('DraftKings\nA.J. Ewing Over 1.5\nTotal Bases\n-115');
+    expect(r.legs[0].player).toBe('AJ Ewing');
+    expect(r.legs[0].player_key).toBe('aj ewing');
+  });
+});