data-daq/service-base/src/main/java/com/simuwang/base/common/util/EmailUtil.java

package com.simuwang.base.common.util;

import cn.hutool.core.collection.CollUtil;
import cn.hutool.core.date.DateUtil;
import cn.hutool.core.exceptions.ExceptionUtil;
import cn.hutool.core.map.MapUtil;
import cn.hutool.core.util.StrUtil;
import cn.hutool.extra.mail.JakartaUserPassAuthenticator;
import com.simuwang.base.common.conts.DateConst;
import com.simuwang.base.common.conts.EmailTypeConst;
import com.simuwang.base.pojo.dto.EmailContentInfoDTO;
import com.simuwang.base.pojo.dto.MailboxInfoDTO;
import com.sun.mail.imap.IMAPStore;
import jakarta.mail.Message;
import jakarta.mail.MessagingException;
import jakarta.mail.Session;
import jakarta.mail.Store;
import jakarta.mail.internet.MimeBodyPart;
import jakarta.mail.internet.MimeMultipart;
import jakarta.mail.internet.MimeUtility;
import org.apache.commons.io.FileUtils;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;

import java.io.File;
import java.util.*;

/**
 * @author mozuwen
 * @date 2024-09-04
 * @description 邮件解析工具
 */
public class EmailUtil {

    private static final Logger logger = LoggerFactory.getLogger(EmailUtil.class);

    private static final String POP3 = "pop3";
    private static final String IMAP = "imap";

    /**
     * 采集邮件(多消息体)信息
     *
     * @param message      邮件
     * @param emailAddress 邮箱地址
     * @param path         存储路径
     * @return 从邮箱采集到的信息
     * @throws Exception 异常信息
     */
    public static List<EmailContentInfoDTO> collectMimeMultipart(Message message, String emailAddress, String path) throws Exception {
        List<EmailContentInfoDTO> emailContentInfoDTOList = CollUtil.newArrayList();
        String emailTitle = message.getSubject();
        String emailDate = DateUtil.format(message.getSentDate(), DateConst.YYYYMMDDHHMMSS24);
        String emailDateStr = DateUtil.format(message.getSentDate(), DateConst.YYYYMMDD);
        String filePath = path + "/" + emailAddress + "/" + emailDateStr + "/";

        MimeMultipart mimeMultipart = (MimeMultipart) message.getContent();
        int length = mimeMultipart.getCount();
        // 遍历邮件消息体
        for (int i = 0; i < length; i++) {
            EmailContentInfoDTO emailContentInfoDTO = new EmailContentInfoDTO();
            MimeBodyPart part = (MimeBodyPart) mimeMultipart.getBodyPart(i);
            Object partContent = part.getContent();
            String contentClass = part.getContent().getClass().getSimpleName();
            // 1.邮件正文
            if ("String".equals(contentClass)) {
                // 文件名 = 邮件主题 + 邮件日期
                String fileName = emailTitle + "_" + emailDate + ".html";
                emailContentInfoDTO = collectTextPart(part, partContent, filePath, fileName);
            } else if ("BASE64DecoderStream".equals(contentClass)) {
                if (StrUtil.isNotBlank(part.getFileName())) {
                    String fileName = MimeUtility.decodeText(part.getFileName());
                    emailContentInfoDTO.setFileName(fileName);

                    String realPath = filePath + emailDate + fileName;
                    File saveFile = new File(realPath);
                    if (!saveFile.exists()) {
                        if (!saveFile.getParentFile().exists()) {
                            saveFile.getParentFile().mkdirs();
                        }
                        FileUtil.saveFile(saveFile, part);
                    } else {
                        FileUtils.deleteQuietly(saveFile);
                        FileUtil.saveFile(saveFile, part);
                    }
                    emailContentInfoDTO.setFilePath(realPath);
                }
            } else if ("MimeMultipart".equals(contentClass)) {
                MimeMultipart contentPart = (MimeMultipart) partContent;
                int length2 = contentPart.getCount();
                for (int i2 = 0; i2 < length2; i2++) {
                    part = (MimeBodyPart) contentPart.getBodyPart(i2);
                    partContent = part.getContent();
                    contentClass = partContent.getClass().getSimpleName();
                    if ("String".equals(contentClass)) {
                        // 文件名 = 邮件主题 + 邮件日期
                        String fileName = emailTitle + "_" + emailDate + ".html";
                        emailContentInfoDTO = collectTextPart(part, partContent, filePath, fileName);
                    }
                }
            }
            if (emailContentInfoDTO.getEmailContent() == null && emailContentInfoDTO.getFilePath() == null) {
                continue;
            }
            emailContentInfoDTO.setEmailAddress(emailAddress);
            emailContentInfoDTO.setEmailTitle(emailTitle);
            emailContentInfoDTO.setEmailDate(DateUtil.format(message.getSentDate(), DateConst.YYYY_MM_DD_HH_MM_SS));
            emailContentInfoDTOList.add(emailContentInfoDTO);
        }

        return emailContentInfoDTOList;
    }

    /**
     * 根据日期过滤邮件
     *
     * @param messages  采集到的邮件
     * @param startDate 邮件起始日期
     * @param endDate   邮件截止日期
     * @return 符合日期的邮件
     */
    public static List<Message> filterMessage(Message[] messages, Date startDate, Date endDate) {
        long startTime = System.currentTimeMillis();
        List<Message> messageList = CollUtil.newArrayList();
        if (messages == null) {
            return messageList;
        }
        for (Message message : messages) {
            try {
                if (message.getSentDate().compareTo(startDate) >= 0 && message.getSentDate().compareTo(endDate) <= 0) {
                    messageList.add(message);
                }
            } catch (MessagingException e) {
                throw new RuntimeException(e);
            }
        }
        logger.info("根据日期过滤邮件耗时 -> {}ms", (System.currentTimeMillis() - startTime));
        return messageList;
    }

    /**
     * 采集邮件正文
     *
     * @param part        邮件消息体
     * @param partContent 邮件消息内筒
     * @param filePath    文件路径
     * @param fileName    文件名
     * @return 采集到邮件正文(html格式包含table标签)
     */
    public static EmailContentInfoDTO collectTextPart(MimeBodyPart part, Object partContent, String filePath, String fileName) {
        EmailContentInfoDTO emailContentInfoDTO = new EmailContentInfoDTO();
        try {
            if ((part.getContentType().contains("text/html") || part.getContentType().contains("TEXT/HTML"))) {
                if (partContent.toString().contains("<table")) {
                    emailContentInfoDTO.setEmailContent(partContent.toString());
                    String savePath = filePath + fileName;
                    File saveFile = new File(savePath);
                    if (!saveFile.exists()) {
                        if (!saveFile.getParentFile().exists()) {
                            saveFile.getParentFile().mkdirs();
                            saveFile.getParentFile().setExecutable(true);
                        }
                        FileUtil.writeFile(savePath, partContent.toString());
                    }
                    emailContentInfoDTO.setFileName(fileName);
                    emailContentInfoDTO.setFilePath(savePath);
                }
            }
        } catch (MessagingException e) {
            logger.info("邮件正文采集失败 -> 文件名:{}, 报错堆栈:{}", fileName, ExceptionUtil.stacktraceToString(e));
            return emailContentInfoDTO;
        }
        return emailContentInfoDTO;
    }

    /**
     * 判断邮件是否符合解析条件
     *
     * @param subject      邮件主题
     * @param emailTypeMap 邮件类型识别规则映射表
     * @return 邮件类型:1-净值,2-估值表,3-定期报告 -> 兜底为净值类型
     */
    public static Integer getEmailTypeBySubject(String subject, Map<Integer, List<String>> emailTypeMap) {
        if (MapUtil.isEmpty(emailTypeMap)) {
            return EmailTypeConst.NAV_EMAIL_TYPE;
        }
        for (Map.Entry<Integer, List<String>> emailTypeEntry : emailTypeMap.entrySet()) {
            for (String field : emailTypeEntry.getValue()) {
                if (subject.contains(field)) {
                    return emailTypeEntry.getKey();
                }
            }
        }
        return EmailTypeConst.NAV_EMAIL_TYPE;
    }

    public static Store getStoreNew(MailboxInfoDTO mailboxInfoDTO) {
        // 配置连接邮件服务器参数
        Properties props = getMailProps(mailboxInfoDTO);
        // 创建Session实例对象
        Session session = Session.getInstance(props, new JakartaUserPassAuthenticator(mailboxInfoDTO.getAccount(), mailboxInfoDTO.getPassword()));
        Store store;
        try {
            String protocol = mailboxInfoDTO.getProtocol().equals(IMAP) ? "imaps" : "pop3";
            if (mailboxInfoDTO.getProtocol().contains(IMAP)) {
                IMAPStore imapStore = (IMAPStore) session.getStore(protocol);
                imapStore.connect(mailboxInfoDTO.getHost(), mailboxInfoDTO.getAccount(), mailboxInfoDTO.getPassword());
                // 网易邮箱需要带上身份标识,详情请看:https://www.hmail163.com/content/?404.html
                Map<String, String> clientParams = new HashMap<>(2);
                clientParams.put("name", "my-imap");
                clientParams.put("version", "1.0");
                imapStore.id(clientParams);
                return imapStore;
            } else {
                store = session.getStore(protocol);
                store.connect(mailboxInfoDTO.getHost(), mailboxInfoDTO.getAccount(), mailboxInfoDTO.getPassword());
                return store;
            }
        } catch (Exception e) {
            logger.error("邮箱信息:{},服务器参数:{}", mailboxInfoDTO, props);
            logger.error("连接邮箱报错堆栈信息:{}", ExceptionUtil.stacktraceToString(e));
            return null;
        }
    }

    public static Properties getMailProps(MailboxInfoDTO mailboxInfoDTO) {
        Properties props = new Properties();
        if (mailboxInfoDTO.getProtocol().equalsIgnoreCase(POP3)) {
            props.put("mail.pop3.host", mailboxInfoDTO.getHost());
            props.put("mail.pop3.user", mailboxInfoDTO.getAccount());
            props.put("mail.pop3.socketFactory", mailboxInfoDTO.getPort());
            props.put("mail.pop3.socketFactory.class", "javax.net.ssl.SSLSocketFactory");
            props.put("mail.pop3.port", mailboxInfoDTO.getPort());
            props.put("mail.store.protocol", mailboxInfoDTO.getProtocol());
        }
        if (mailboxInfoDTO.getProtocol().equalsIgnoreCase(IMAP)) {
            props.put("mail.store.protocol", "imaps");
            props.put("mail.imap.host", mailboxInfoDTO.getHost());
            props.put("mail.imap.port", mailboxInfoDTO.getPort());
            props.put("mail.imaps.ssl.enable", "true");
            props.put("mail.imaps.ssl.trust", "*");
            props.put("mail.imap.auth", "true");
            props.put("mail.imap.starttls.enable", "true");
            props.put("mail.imap.socketFactory.class", "javax.net.ssl.SSLSocketFactory");
            props.put("mail.imap.socketFactory.fallback", "false");
        }
        return props;
    }
}