如何通过Java的IMAP客户端解析邮件时检测并过滤邮件签名

Got it, let's break down how to detect and strip email signatures without relying on third-party libraries—since you're concerned about data privacy, this approach keeps everything under your control.

Signatures almost always live at the end of an email, so we’ll work backwards from the bottom of the message to spot and remove them. We’ll use pattern matching and positional checks to target each common signature type:

1. Chinese Business Signatures ("顺颂商祺", etc.) #

These follow predictable structures: a closing greeting followed by name, contact info, or company details.

• Detection Logic:
  • Scan from the end of the message for trigger phrases: 顺颂商祺, 此致敬礼, 祝商祺, 祝好, 谨上
  • Once a trigger is found, check subsequent lines for signature staples like names, phone numbers (use regex like \d{3}-\d{8}|\d{4}-\d{7}), or company names
• Stripping: Cut the message content from the first trigger phrase onwards.

2. Formal English Signatures & Professional Links (LinkedIn, etc.) #

For international or English-language signatures, target standard closings and professional profiles:

• Detection Logic:
  • Trigger phrases: Best regards,, Sincerely,, Kind regards,, Cheers,
  • Match LinkedIn links with regex: https?://(www\.)?linkedin\.com/in/[\w-]+
  • Watch for phone numbers, physical addresses, or branded email footers too
• Stripping: Remove everything from the closing greeting (including the greeting) to the end of the message.

3. Device Auto-Signatures ("发自iPhone", "Sent from my Android") #

These are the easiest to spot thanks to their standardized phrasing:

• Detection Logic:
  • Match exact or partial phrases: 发自iPhone, Sent from my, Sent via Outlook, 使用QQ邮箱发送
  • These almost always appear as the final line(s) of the message
• Stripping: Remove any lines that match these phrases (and any content after them, though they’re usually the end).

4. Image Signatures #

Image signatures require a bit more work, especially for HTML emails:

• Detection Logic:
  • In HTML content, look for <img> tags in the final 10-20% of the message
  • Check alt text for keywords like signature, company logo, or the user’s name
  • Embedded images (not external links) are far more likely to be signatures
• Stripping: Remove the <img> tag and any surrounding container elements (like <div> or <p>) that hold the signature.

Here’s a simplified snippet to handle text-based signatures—you can extend this for HTML using Python’s built-in html.parser:

import re

def strip_signature(email_content):
    lines = email_content.split('\n')
    # Work backwards to find the first signature trigger
    for i in range(len(lines)-1, -1, -1):
        line = lines[i].strip()
        # Check Chinese greeting triggers
        if any(trigger in line for trigger in ['顺颂商祺', '此致敬礼', '祝好']):
            return '\n'.join(lines[:i]).strip()
        # Check device auto-signatures
        if any(trigger in line for trigger in ['发自iPhone', 'Sent from my', '使用QQ邮箱发送']):
            return '\n'.join(lines[:i]).strip()
        # Check English closings
        if any(trigger in line for trigger in ['Best regards,', 'Sincerely,', 'Kind regards,']):
            return '\n'.join(lines[:i]).strip()
        # Check LinkedIn links
        if re.search(r'https?://(www\.)?linkedin\.com/in/[\w-]+', line):
            return '\n'.join(lines[:i]).strip()
    # No signature found, return original content
    return email_content

# Test it out
email_text = """Hi there,

Thanks for your reply. Let's schedule a call next week.

顺颂商祺
张三
联系方式：138-XXXX-XXXX
LinkedIn: https://linkedin.com/in/zhangsan

发自iPhone
"""
cleaned_text = strip_signature(email_text)
print(cleaned_text)

• Update Your Trigger List: Add new phrases as you encounter regional or company-specific signatures (e.g., 祝工作顺利, Sent via Gmail)
• HTML vs. Plain Text: For HTML emails, parse the content into a DOM and apply similar logic to text nodes in the final sections
• Reduce False Positives: Only strip content if multiple signature elements are present (e.g., a closing greeting + a phone number)
• Partial Signatures: If a signature is split across lines, adjust the logic to check 2-3 lines back from a trigger phrase

内容的提问来源于stack exchange，提问作者Yash Jain