deckhearth/lib/scan-vision.js

const GATEWAY_URL = 'https://ai-gateway.vercel.sh/v1/chat/completions';
const DEFAULT_VISION_MODEL =
  process.env.SCAN_VISION_MODEL || 'google/gemini-2.5-flash-lite';

const CARD_PROMPT = `You are a specialized trading card recognition system. Analyze this image and determine if it contains a trading card (Magic: The Gathering, Pokemon, Yu-Gi-Oh, Lorcana, etc.).

CRITICAL: Only respond with card data if you can clearly identify a TRADING CARD in the image. Ignore random objects, books, papers, phone screens, screenshots, blurry images, and non-card gaming items.

If you detect a trading card, extract information in this JSON format:
{
  "isCard": true,
  "cardName": "exact card name as printed on the card",
  "setName": "set name if visible",
  "setCode": "set code/symbol if visible",
  "cardNumber": "collector number if visible",
  "game": "mtg, pokemon, or lorcana (lowercase)",
  "cardType": "creature, instant, sorcery, trainer, etc.",
  "rarity": "common, uncommon, rare, mythic, etc.",
  "manaCost": "mana cost if visible",
  "hp": "HP or power if visible",
  "abilities": ["list of abilities or attacks if clearly readable"],
  "confidence": 85,
  "rawText": "all text visible on the card"
}

If NO trading card is clearly visible, respond with:
{
  "isCard": false,
  "confidence": 0,
  "reason": "No trading card detected in image"
}

Be conservative — only extract data you can clearly read. Quality over quantity.`;

function parseVisionJson(content) {
  const cleanContent = content.replace(/```json\n?/g, '').replace(/```\n?/g, '').trim();
  try {
    return JSON.parse(cleanContent);
  } catch {
    const cardNameMatch = content.match(/card.*?name.*?[:"]\s*([^"'\n,}]+)/i);
    return {
      isCard: !!cardNameMatch,
      cardName: cardNameMatch ? cardNameMatch[1].trim() : null,
      confidence: 30,
      rawText: content,
      reason: 'Failed to parse structured response',
    };
  }
}

function normalizeGame(game) {
  if (!game) return null;
  const value = String(game).trim().toLowerCase();
  if (value === 'mtg' || value.includes('magic')) return 'mtg';
  if (value.includes('pokemon') || value.includes('pokémon')) return 'pokemon';
  if (value.includes('lorcana')) return 'lorcana';
  return value;
}

/**
 * Server-side vision analysis via Vercel AI Gateway. Requires AI_GATEWAY_API_KEY.
 * @param {string} imageDataUrl - data:image/jpeg;base64,... capture from scanner
 */
export async function analyzeCardImage(imageDataUrl) {
  const apiKey = process.env.AI_GATEWAY_API_KEY;
  if (!apiKey) {
    throw new Error('AI_GATEWAY_API_KEY is not configured on the server');
  }

  if (!imageDataUrl?.includes(',')) {
    throw new Error('Invalid image data format');
  }

  const response = await fetch(GATEWAY_URL, {
    method: 'POST',
    headers: {
      'Content-Type': 'application/json',
      Authorization: `Bearer ${apiKey}`,
    },
    body: JSON.stringify({
      model: DEFAULT_VISION_MODEL,
      temperature: 0.1,
      messages: [{
        role: 'user',
        content: [
          { type: 'text', text: CARD_PROMPT },
          { type: 'image_url', image_url: { url: imageDataUrl } },
        ],
      }],
    }),
  });

  if (!response.ok) {
    const errorData = await response.json().catch(() => ({}));
    const message = errorData.error?.message || errorData.message || 'Unknown error';
    throw new Error(`Vision API error: ${response.status} - ${message}`);
  }

  const data = await response.json();
  const content = data.choices?.[0]?.message?.content;
  if (!content) {
    throw new Error('No response from vision model');
  }

  const result = parseVisionJson(content);

  return {
    isCard: result.isCard || false,
    cardName: result.cardName || null,
    setName: result.setName || null,
    setCode: result.setCode || null,
    cardNumber: result.cardNumber || null,
    game: normalizeGame(result.game),
    cardType: result.cardType || null,
    rarity: result.rarity || null,
    manaCost: result.manaCost || null,
    hp: result.hp || null,
    abilities: result.abilities || [],
    confidence: result.confidence || 0,
    rawText: result.rawText || content,
    reason: result.reason || null,
  };
}