perf: add rate limiting, validation, logging, and error handling
- Add express-rate-limit with tiered rate limits per endpoint - Add express-validator for input validation on all API routes - Add winston + morgan for structured logging - Add node-cache for lottery result caching (5min TTL, 1000 max keys) - Convert OCR to async task queue with max 2 concurrent jobs - Add global error handler middleware - Add React ErrorBoundary for client-side error handling - Fix memory leaks in OCR task queue and lottery cache - Add per-item error handling in batch ticket updates - Fix SQL UPDATE to include user_id in WHERE clause Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
This commit is contained in:
Developer
co-authored by
Claude Opus 4.6
Claude Opus 4.6
parent
7b1073e23b
commit
335f0fb8dc
+108
-53
@@ -1,53 +1,119 @@
|
||||
const Tesseract = require('tesseract.js');
|
||||
const path = require('path');
|
||||
const logger = require('./logger');
|
||||
|
||||
// 彩票类型识别规则
|
||||
// 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 = {
|
||||
// 福利彩票
|
||||
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: /福彩\s*3D|3D/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 },
|
||||
// 体育彩票
|
||||
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 },
|
||||
};
|
||||
|
||||
// 使用本地tesseract.js-core进行OCR识别
|
||||
async function parseWithLocalTesseract(imagePath) {
|
||||
// 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 {
|
||||
// 使用本地core文件
|
||||
const corePath = path.join(__dirname, '../node_modules/tesseract.js-core');
|
||||
|
||||
const { data } = await Tesseract.recognize(imagePath, 'chi_sim+eng', {
|
||||
corePath: corePath,
|
||||
logger: m => console.log('OCR:', m.status, m.progress)
|
||||
});
|
||||
|
||||
return data.text;
|
||||
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) {
|
||||
console.error('tesseract.js识别失败:', err.message);
|
||||
throw err;
|
||||
logger.error(`OCR任务失败: ${taskId}`, { error: err.message });
|
||||
completedTasks.set(taskId, { success: false, error: err.message });
|
||||
setTimeout(() => {
|
||||
completedTasks.delete(taskId);
|
||||
}, 5 * 60 * 1000);
|
||||
} finally {
|
||||
processingTasks.delete(taskId);
|
||||
processNextTask();
|
||||
}
|
||||
}
|
||||
|
||||
// 解析彩票图片
|
||||
async function parseTicketImage(imagePath) {
|
||||
let text = '';
|
||||
|
||||
try {
|
||||
text = await parseWithLocalTesseract(imagePath);
|
||||
console.log('OCR识别完成, 文本长度:', text.length);
|
||||
} catch (err) {
|
||||
console.error('OCR识别失败:', err);
|
||||
throw new Error('图片识别失败');
|
||||
// 处理下一个任务
|
||||
function processNextTask() {
|
||||
if (processingTasks.size >= 2) return; // 最多同时处理2个
|
||||
|
||||
const nextTask = taskQueue.shift();
|
||||
if (nextTask) {
|
||||
processingTasks.add(nextTask.taskId);
|
||||
processOCR(nextTask);
|
||||
}
|
||||
}
|
||||
|
||||
// 提交OCR任务 (同步返回taskId)
|
||||
function submitOCRTask(imagePath) {
|
||||
if (taskQueue.length >= MAX_QUEUE_SIZE) {
|
||||
throw new Error('任务队列已满，请稍后再试');
|
||||
}
|
||||
|
||||
// 调试：打印识别的文本
|
||||
console.log('识别的文本:', text.substring(0, 500));
|
||||
const taskId = `ocr_${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}`);
|
||||
|
||||
return parseTicketText(text);
|
||||
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
|
||||
@@ -64,7 +130,7 @@ function parseTicketText(text) {
|
||||
raw_text: text
|
||||
};
|
||||
|
||||
// 识别彩票类型
|
||||
// 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;
|
||||
@@ -73,7 +139,7 @@ function parseTicketText(text) {
|
||||
}
|
||||
}
|
||||
|
||||
// 提取期号 - 优先匹配"期号:xxx"格式
|
||||
// 提取期号
|
||||
const issueMatch = text.match(/期\s*号\s*[:：]\s*(\d{5,})/);
|
||||
if (issueMatch) {
|
||||
result.issue = issueMatch[1];
|
||||
@@ -99,22 +165,20 @@ function parseTicketText(text) {
|
||||
// 检测是否追加
|
||||
result.is_additional = /追加|追加投注/.test(text);
|
||||
|
||||
// 提取金额 - 支持多种格式
|
||||
// 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 - 根据不同彩票类型
|
||||
// 提取号码
|
||||
result.numbers = extractNumbers(text, result.game);
|
||||
|
||||
// 根据识别到的号码数量更新注数
|
||||
if (result.numbers.length > 0) {
|
||||
result.bet_count = result.numbers.length;
|
||||
}
|
||||
|
||||
// 如果没有识别到金额，根据注数和倍数计算
|
||||
if (!result.cost && result.game && result.numbers.length > 0) {
|
||||
const lottery = Object.values(LOTTERY_PATTERNS).find(l => l.name === result.game);
|
||||
if (lottery) {
|
||||
@@ -131,32 +195,27 @@ function parseTicketText(text) {
|
||||
function extractNumbers(text, game) {
|
||||
const numbers = [];
|
||||
|
||||
// 预处理文本：修复常见OCR错误
|
||||
let cleanText = text
|
||||
.replace(/[oO]/g, '0') // o和O经常被误识别为0
|
||||
.replace(/[D]/g, '0') // D经常被误识别
|
||||
.replace(/[lI|]/g, '1') // l和I和|经常被误识别为1
|
||||
.replace(/[S\$§]/g, '5') // S和$和§经常被误识别为5
|
||||
.replace(/[@®»«]/g, '') // 去掉这些特殊字符
|
||||
.replace(/[:：]/g, ' ') // 冒号替换为空格
|
||||
.replace(/[g]/g, '9'); // g经常被误识别为9
|
||||
.replace(/[oO]/g, '0')
|
||||
.replace(/[D]/g, '0')
|
||||
.replace(/[lI|]/g, '1')
|
||||
.replace(/[S$§]/g, '5')
|
||||
.replace(/[@®»«]/g, '')
|
||||
.replace(/[:：]/g, ' ')
|
||||
.replace(/[g]/g, '9');
|
||||
|
||||
// 处理连在一起的数字：将6位以上连续数字按每2位分割
|
||||
cleanText = cleanText.replace(/\b(\d{6,})\b/g, (match) => {
|
||||
return match.match(/.{1,2}/g).join(' ');
|
||||
});
|
||||
// 处理4位连续数字
|
||||
cleanText = cleanText.replace(/\b(\d{4})\b/g, (match) => {
|
||||
return match.slice(0, 2) + ' ' + match.slice(2);
|
||||
});
|
||||
|
||||
// 移除开奖号码行（包含"开奖号码"的行）
|
||||
cleanText = cleanText.split('\n')
|
||||
.filter(line => !line.includes('开奖') && !line.includes('号码'))
|
||||
.join('\n');
|
||||
|
||||
if (game === '双色球') {
|
||||
// 双色球: 6红+1蓝，支持多种格式
|
||||
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,
|
||||
@@ -173,12 +232,9 @@ function extractNumbers(text, game) {
|
||||
if (numbers.length > 0) break;
|
||||
}
|
||||
} else if (game === '大乐透') {
|
||||
// 大乐透: 5前区+2后区
|
||||
const patterns = [
|
||||
// 带空格的标准格式
|
||||
/(\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,
|
||||
// 连在一起无空格的格式（第一注可能这样）
|
||||
/(\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,
|
||||
];
|
||||
@@ -194,7 +250,6 @@ function extractNumbers(text, game) {
|
||||
if (numbers.length > 0) break;
|
||||
}
|
||||
} else if (game === '福彩3D' || game === '排列3') {
|
||||
// 3位数
|
||||
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]]);
|
||||
@@ -204,4 +259,4 @@ function extractNumbers(text, game) {
|
||||
return numbers;
|
||||
}
|
||||
|
||||
module.exports = { parseTicketImage, parseTicketText };
|
||||
module.exports = { parseTicketImage, parseTicketText, getOCRResult };
|
||||
|
||||
Reference in New Issue
Block a user