const Tesseract = require('tesseract.js'); const path = require('path'); const logger = require('./logger'); // 彩票类型识别规则 (模块级别常量,避免每次调用时重建) const LOTTERY_PATTERNS = { ssq: { name: '双色球', type: 'welfare', pattern: /双\s*色\s*球|SSQ/i, price: 2 }, fc3d: { name: '福彩3D', type: 'welfare', pattern: /福\s*彩\s*3D|3D/i, price: 2 }, qlc: { name: '七乐彩', type: 'welfare', pattern: /七\s*乐\s*彩/i, price: 2 }, dlt: { name: '大乐透', type: 'sports', pattern: /大\s*乐\s*透|超级\s*大\s*乐\s*透/i, price: 2 }, pl3: { name: '排列3', type: 'sports', pattern: /排列\s*3|排列\s*三/i, price: 2 }, pl5: { name: '排列5', type: 'sports', pattern: /排列\s*5|排列\s*五/i, price: 2 }, qxc: { name: '七星彩', type: 'sports', pattern: /七\s*星\s*彩/i, price: 2 }, }; // 简单的内存任务队列 (有界队列防止内存溢出) const MAX_QUEUE_SIZE = 100; const taskQueue = []; const processingTasks = new Set(); const completedTasks = new Map(); // taskId -> result const TASK_TIMEOUT = 5 * 60 * 1000; // 5分钟超时 // OCR处理函数 async function processOCR(task) { const { taskId, imagePath } = task; try { logger.info(`OCR任务开始: ${taskId}`); const text = await parseWithLocalTesseract(imagePath); const result = parseTicketText(text); completedTasks.set(taskId, { success: true, data: result }); // 5分钟后清理已完成的任务 (只设置一次) setTimeout(() => { completedTasks.delete(taskId); logger.debug(`OCR任务已清理: ${taskId}`); }, 5 * 60 * 1000); logger.info(`OCR任务完成: ${taskId}`); } catch (err) { logger.error(`OCR任务失败: ${taskId}`, { error: err.message }); completedTasks.set(taskId, { success: false, error: err.message }); setTimeout(() => { completedTasks.delete(taskId); }, 5 * 60 * 1000); } finally { processingTasks.delete(taskId); processNextTask(); } } // 处理下一个任务 function processNextTask() { if (processingTasks.size >= 2) return; // 最多同时处理2个 const nextTask = taskQueue.shift(); if (nextTask) { processingTasks.add(nextTask.taskId); processOCR(nextTask); } } // 提交OCR任务 (同步返回taskId) function submitOCRTask(imagePath) { if (taskQueue.length >= MAX_QUEUE_SIZE) { throw new Error('任务队列已满,请稍后再试'); } const taskId = `ocr_${Date.now()}_${Math.random().toString(36).slice(2)}`; taskQueue.push({ taskId, imagePath, submittedAt: Date.now() }); logger.info(`OCR任务已提交: ${taskId}, 队列长度: ${taskQueue.length}`); processNextTask(); return taskId; } // 查询任务状态 function getOCRTaskStatus(taskId) { if (completedTasks.has(taskId)) { return { completed: true, ...completedTasks.get(taskId) }; } // 检查是否超时 - O(n) scan (可接受,队列通常很小) const task = taskQueue.find(t => t.taskId === taskId); if (task && Date.now() - task.submittedAt > TASK_TIMEOUT) { return { completed: true, success: false, error: '任务处理超时' }; } if (processingTasks.has(taskId)) { return { completed: false, processing: true }; } return { completed: false }; } // 使用本地tesseract.js-core进行OCR识别 async function parseWithLocalTesseract(imagePath) { const corePath = path.join(__dirname, '../node_modules/tesseract.js-core'); const { data } = await Tesseract.recognize(imagePath, 'chi_sim+eng', { corePath: corePath, logger: m => logger.debug(`OCR: ${m.status} ${m.progress || ''}`) }); return data.text; } // 解析彩票图片 (改为异步提交) function parseTicketImage(imagePath) { const taskId = submitOCRTask(imagePath); return { taskId, status: 'pending' }; } // 查询OCR结果 (轮询) function getOCRResult(taskId) { return getOCRTaskStatus(taskId); } // 解析彩票文本 function parseTicketText(text) { const result = { type: null, game: null, issue: null, numbers: [], bet_count: 1, multiple: 1, is_additional: false, cost: 0, raw_text: text }; // 识别彩票类型 (使用模块级别的 LOTTERY_PATTERNS) for (const [key, lottery] of Object.entries(LOTTERY_PATTERNS)) { if (lottery.pattern.test(text)) { result.type = lottery.type; result.game = lottery.name; break; } } // 提取期号 const issueMatch = text.match(/期\s*号\s*[::]\s*(\d{5,})/); if (issueMatch) { result.issue = issueMatch[1]; } else { const issueMatch2 = text.match(/第?\s*(\d{5,})\s*期/); if (issueMatch2) { result.issue = issueMatch2[1]; } } // 提取注数 const betMatch = text.match(/(\d+)\s*注/); if (betMatch) { result.bet_count = parseInt(betMatch[1]); } // 提取倍数 const multipleMatch = text.match(/(\d+)\s*倍/); if (multipleMatch) { result.multiple = parseInt(multipleMatch[1]); } // 检测是否追加 result.is_additional = /追加|追加投注/.test(text); // 提取金额 const costMatch = text.match(/金额\s*[::]\s*(\d+(?:[.,]\d+)?)\s*元|合计[::]\s*(\d+(?:\.\d+)?)\s*元|(\d+(?:\.\d+)?)\s*元/); if (costMatch) { const costStr = (costMatch[1] || costMatch[2] || costMatch[3]).replace(',', '.'); result.cost = parseFloat(costStr); } // 提取号码 result.numbers = extractNumbers(text, result.game); if (result.numbers.length > 0) { result.bet_count = result.numbers.length; } if (!result.cost && result.game && result.numbers.length > 0) { const lottery = Object.values(LOTTERY_PATTERNS).find(l => l.name === result.game); if (lottery) { let basePrice = lottery.price * result.bet_count * result.multiple; if (result.is_additional) basePrice += result.bet_count * result.multiple; result.cost = basePrice; } } return result; } // 提取号码 function extractNumbers(text, game) { const numbers = []; let cleanText = text .replace(/[oO]/g, '0') .replace(/[D]/g, '0') .replace(/[lI|]/g, '1') .replace(/[S$§]/g, '5') .replace(/[@®»«]/g, '') .replace(/[::]/g, ' ') .replace(/[g]/g, '9'); cleanText = cleanText.replace(/\b(\d{6,})\b/g, (match) => { return match.match(/.{1,2}/g).join(' '); }); cleanText = cleanText.replace(/\b(\d{4})\b/g, (match) => { return match.slice(0, 2) + ' ' + match.slice(2); }); cleanText = cleanText.split('\n') .filter(line => !line.includes('开奖') && !line.includes('号码')) .join('\n'); if (game === '双色球') { const patterns = [ /(\d{2})\s+(\d{2})\s+(\d{2})\s+(\d{2})\s+(\d{2})\s+(\d{2})[-+](\d{2})/g, /(\d{2})\s+(\d{2})\s+(\d{2})\s+(\d{2})\s+(\d{2})\s+(\d{2})\s*[+|]\s*(\d{2})/g, ]; for (const pattern of patterns) { const matches = cleanText.matchAll(pattern); for (const match of matches) { numbers.push({ red: [match[1], match[2], match[3], match[4], match[5], match[6]], blue: match[7] }); } if (numbers.length > 0) break; } } else if (game === '大乐透') { const patterns = [ /(\d{2})\s+(\d{2})\s+(\d{2})\s+(\d{2})\s+(\d{2})[-+](\d{2})\s+(\d{2})/g, /(\d{2})\s+(\d{2})\s+(\d{2})\s+(\d{2})\s+(\d{2})\s*[+|]\s*(\d{2})\s+(\d{2})/g, /(\d{2})(\d{2})(\d{2})(\d{2})(\d{2})[-+](\d{2})(\d{2})/g, /(\d{2})(\d{2})(\d{2})(\d{2})(\d{2})[+|](\d{2})(\d{2})/g, ]; for (const pattern of patterns) { const matches = cleanText.matchAll(pattern); for (const match of matches) { numbers.push({ front: [match[1], match[2], match[3], match[4], match[5]], back: [match[6], match[7]] }); } if (numbers.length > 0) break; } } else if (game === '福彩3D' || game === '排列3') { const matches = cleanText.matchAll(/[^\d](\d)\s*(\d)\s*(\d)[^\d]/g); for (const match of matches) { numbers.push([match[1], match[2], match[3]]); } } return numbers; } module.exports = { parseTicketImage, parseTicketText, getOCRResult };