lt
289 lines
12 KiB
Python
289 lines
12 KiB
Python
#!/usr/bin/env python3
|
||
"""
|
||
内容合规审查模块
|
||
检查文章是否符合法律法规、平台规则、品牌规范
|
||
"""
|
||
|
||
import re
|
||
from typing import Dict, List, Tuple
|
||
|
||
# 敏感词库（示例，需要持续更新）
|
||
SENSITIVE_WORDS = {
|
||
"政治敏感": ["国家主席", "政治局", "常委", "军委", "统战部", "颠覆国家", "分裂主义", "台独", "疆独", "藏独"],
|
||
"违禁内容": ["赌博", "毒品", "迷药", "枪支", "炸药", "色情", "低俗", "反动", "邪教"],
|
||
"不实信息": [" guaranteed 赚钱", "一夜暴富", "100%有效", "包治百病", "绝对正确"],
|
||
"领导人相关": ["主席", "总理", "总书记", "国家领导人"] # 需上下文判断
|
||
}
|
||
|
||
# 平台规则限制
|
||
PLATFORM_RULES = {
|
||
"zhihu": {
|
||
"max_title_len": 100,
|
||
"min_word_count": 1000,
|
||
"allowed_tags": ["科技", "生活", "职场", "教育", "可持续", "AI", "远程工作", "个人成长"],
|
||
"forbidden_patterns": [r"加微信", r"私聊", r"付费咨询", r"点击领取"]
|
||
},
|
||
"wechat": {
|
||
"max_title_len": 32,
|
||
"min_word_count": 800,
|
||
"allowed_tags": ["科技", "生活", "职场", "教育", "可持续", "AI", "远程工作", "个人成长"],
|
||
"forbidden_patterns": [r"诱导分享", r"朋友圈", r"转发群"]
|
||
},
|
||
"xiaohongshu": {
|
||
"max_title_len": 50,
|
||
"min_word_count": 400,
|
||
"allowed_tags": ["生活方式", "可持续", "AI", "个人成长", "极简", "环保"],
|
||
"forbidden_patterns": [r"私信", r"加群", r"导流"]
|
||
}
|
||
}
|
||
|
||
class ComplianceChecker:
|
||
"""合规审查器"""
|
||
|
||
def __init__(self):
|
||
self.issues = []
|
||
|
||
def check_text(self, text: str, platform: str, topic_data: Dict = None) -> Dict:
|
||
"""执行全面合规检查"""
|
||
self.issues = []
|
||
|
||
# 1. 敏感词检查
|
||
self._check_sensitive_words(text)
|
||
|
||
# 2. 平台规则检查
|
||
self._check_platform_rules(text, platform)
|
||
|
||
# 3. 法律法规检查
|
||
self._check_legal_compliance(text)
|
||
|
||
# 4. 品牌调性检查
|
||
self._check_brand_guidelines(text)
|
||
|
||
# 5. 内容事实性检查（如有主题数据）
|
||
if topic_data:
|
||
self._check_factual_consistency(text, topic_data)
|
||
|
||
# 6. 最小字数检查
|
||
self._check_min_length(text, platform)
|
||
|
||
# 7. 结构完整性检查（必须包含关键章节）
|
||
self._check_required_sections(text)
|
||
self._check_inline_images(text)
|
||
self._check_timeliness(text)
|
||
|
||
return {
|
||
"passed": len(self.issues) == 0,
|
||
"issues": self.issues,
|
||
"score": max(0, 100 - len(self.issues) * 10)
|
||
}
|
||
|
||
def _check_sensitive_words(self, text: str):
|
||
"""检查敏感词"""
|
||
for category, words in SENSITIVE_WORDS.items():
|
||
for word in words:
|
||
if word in text:
|
||
self.issues.append({
|
||
"type": "敏感词",
|
||
"category": category,
|
||
"word": word,
|
||
"suggestion": f"删除或替换'{word}'"
|
||
})
|
||
|
||
def _check_platform_rules(self, text: str, platform: str):
|
||
"""检查平台特定规则"""
|
||
rules = PLATFORM_RULES.get(platform, {})
|
||
|
||
# 标题长度（从HTML中提取）
|
||
title_match = re.search(r'<title>([^<]+)</title>', text) or re.search(r'<h1[^>]*>([^<]+)</h1>', text)
|
||
if title_match and rules.get("max_title_len"):
|
||
title_len = len(title_match.group(1))
|
||
if title_len > rules["max_title_len"]:
|
||
self.issues.append({
|
||
"type": "平台规则",
|
||
"category": "标题长度",
|
||
"detail": f"标题{title_len}字，超过{platform}限制{rules['max_title_len']}字",
|
||
"suggestion": "缩短标题"
|
||
})
|
||
|
||
# 禁止的模式匹配
|
||
for pattern in rules.get("forbidden_patterns", []):
|
||
if re.search(pattern, text):
|
||
self.issues.append({
|
||
"type": "平台规则",
|
||
"category": "禁止内容",
|
||
"pattern": pattern,
|
||
"suggestion": "移除违规内容或联系方式"
|
||
})
|
||
|
||
# 标签检查（只匹配 #话题 格式，排除颜色码如 #1a1a1a）
|
||
# 标签模式：#开头，后跟字母数字，长度2-10，不全是十六进制字符
|
||
tags = re.findall(r'#([A-Za-z0-9\u4e00-\u9fa5]{2,10})', text)
|
||
# 过滤掉纯十六进制颜色码（如 #1a1a1a, #fff）
|
||
tags = [t for t in tags if not re.fullmatch(r'[0-9a-fA-F]{3,6}', t)]
|
||
allowed = rules.get("allowed_tags", [])
|
||
if allowed:
|
||
for tag in tags:
|
||
if tag not in allowed:
|
||
self.issues.append({
|
||
"type": "平台规则",
|
||
"category": "标签合规",
|
||
"tag": tag,
|
||
"suggestion": f"使用平台允许的标签，如{', '.join(allowed[:3])}"
|
||
})
|
||
|
||
def _check_legal_compliance(self, text: str):
|
||
"""检查法律法规合规性"""
|
||
# 检查是否涉及国家秘密、国家安全
|
||
if re.search(r'国家机密|军事秘密|绝密|机密', text):
|
||
self.issues.append({
|
||
"type": "法律法规",
|
||
"category": "国家秘密",
|
||
"suggestion": "立即删除涉密内容"
|
||
})
|
||
|
||
# 检查是否宣传迷信、邪教
|
||
if re.search(r'算命|看相|测八字|跳大神|法轮功', text):
|
||
self.issues.append({
|
||
"type": "法律法规",
|
||
"category": "封建迷信",
|
||
"suggestion": "删除迷信内容"
|
||
})
|
||
|
||
# 检查是否赌博相关
|
||
if re.search(r'赌|博彩|下注|时时彩|六合彩', text):
|
||
self.issues.append({
|
||
"type": "法律法规",
|
||
"category": "赌博违法",
|
||
"suggestion": "删除赌博相关内容"
|
||
})
|
||
|
||
# 检查版权问题（是否使用未授权素材）
|
||
if re.search(r'版权声明.*?未经授权|转载请联系|盗用', text, re.IGNORECASE):
|
||
self.issues.append({
|
||
"type": "法律法规",
|
||
"category": "版权风险",
|
||
"suggestion": "确保所有引用已标注来源或获得授权"
|
||
})
|
||
|
||
def _check_brand_guidelines(self, text: str):
|
||
"""检查品牌调性（宇之然）"""
|
||
# 检查是否使用第一人称"我"
|
||
first_person_count = len(re.findall(r'^(我|本人|笔者)\b', text, re.MULTILINE))
|
||
if first_person_count > 2: # 允许少量情感连接
|
||
self.issues.append({
|
||
"type": "品牌规范",
|
||
"category": "人称使用",
|
||
"detail": f"发现{first_person_count}处第一人称，建议使用客观叙事",
|
||
"suggestion": "改为'实践者'、'本专栏'等客观表述"
|
||
})
|
||
|
||
# 检查是否有商业推广倾向
|
||
if re.search(r'强烈推荐|必买|最好的|最赚钱|独家', text):
|
||
self.issues.append({
|
||
"type": "品牌规范",
|
||
"category": "过度推广",
|
||
"suggestion": "使用更中立的表达，避免绝对化用语"
|
||
})
|
||
|
||
# 检查是否提及具体品牌（需模糊化）
|
||
known_brands = ["米家", "花帮主", "园艺助手", "Aerogarden"]
|
||
for brand in known_brands:
|
||
if brand in text:
|
||
self.issues.append({
|
||
"type": "品牌规范",
|
||
"category": "品牌露出",
|
||
"brand": brand,
|
||
"suggestion": f"将'{brand}'改为'一些第三方工具'或'智能设备'"
|
||
})
|
||
|
||
def _check_factual_consistency(self, text: str, topic_data: Dict):
|
||
"""检查内容与选题的一致性"""
|
||
topic = topic_data.get("topic", {})
|
||
expected_title = topic.get("title", "")
|
||
expected_field = topic.get("field", "")
|
||
|
||
# 检查标题是否出现在文章中
|
||
if expected_title and expected_title[:5] not in text:
|
||
self.issues.append({
|
||
"type": "内容质量",
|
||
"category": "主题一致性",
|
||
"detail": f"文章可能偏离选题'{expected_title}'",
|
||
"suggestion": "确认内容围绕选题展开"
|
||
})
|
||
|
||
# 检查是否有核心观点
|
||
core_concept = topic.get("core_concept", "")
|
||
if core_concept and len(core_concept) > 10:
|
||
# 核心概念应出现在前1/3内容
|
||
first_third = text[:len(text)//3]
|
||
if core_concept[:10] not in first_third:
|
||
self.issues.append({
|
||
"type": "内容质量",
|
||
"category": "核心观点",
|
||
"suggestion": "在文章前1/3部分明确阐述核心观点"
|
||
})
|
||
|
||
def _check_min_length(self, text: str, platform: str):
|
||
"""检查文章最小字数（去除HTML标签）"""
|
||
# 简单去除HTML标签
|
||
plain = re.sub(r'<[^>]+>', '', text)
|
||
word_count = len(plain.strip())
|
||
min_words = PLATFORM_RULES.get(platform, {}).get("min_word_count", 1000)
|
||
if word_count < min_words:
|
||
self.issues.append({
|
||
"type": "内容完整度",
|
||
"category": "字数不足",
|
||
"detail": f"当前{word_count}字，低于平台要求{min_words}字",
|
||
"suggestion": "扩写内容至最低要求"
|
||
})
|
||
|
||
def _check_required_sections(self, text: str):
|
||
"""检查是否包含必要章节（如引言、核心观点、总结等）"""
|
||
required_headings = [
|
||
"引言", "核心观点", "受众痛点", "总结", "行动指南"
|
||
]
|
||
missing = []
|
||
for heading in required_headings:
|
||
# 检查 h2 或 h3 中是否出现 heading
|
||
if not re.search(r'<h[23][^>]*>.*' + re.escape(heading) + r'.*</h[23]>', text, re.IGNORECASE):
|
||
missing.append(heading)
|
||
if missing:
|
||
self.issues.append({
|
||
"type": "结构完整",
|
||
"category": "章节缺失",
|
||
"detail": f"缺少必要章节：{', '.join(missing)}",
|
||
"suggestion": "补充缺失章节"
|
||
})
|
||
def _check_inline_images(self, html: str):
|
||
"""检查图片是否以内联方式嵌入（data:image）"""
|
||
# 提取所有 img 标签的 src 属性值
|
||
srcs = re.findall(r'<img\b[^>]*src=[\'"]([^\'"]+)[\'"]', html, re.IGNORECASE)
|
||
for src in srcs:
|
||
if not src.startswith('data:image/'):
|
||
self.issues.append({
|
||
"type": "资源合规",
|
||
"category": "图片内联",
|
||
"detail": f"图片未内联: {src[:50]}... 需手动修复"
|
||
})
|
||
|
||
def _check_timeliness(self, text: str):
|
||
years = re.findall(r'(19\d{2}|20[0-4]\d)', text)
|
||
outdated = {y for y in years if int(y) < 2025}
|
||
if outdated:
|
||
self.issues.append({
|
||
"type": "平台规则",
|
||
"category": "时效性",
|
||
"detail": f"使用过时年份: {', '.join(sorted(outdated))}，需更新为2025年及以后的数据",
|
||
"suggestion": "替换为最新数据，或使用'近期'等模糊表述"
|
||
})
|
||
def check_article(html_content: str, platform: str, topic_data: Dict = None) -> Dict:
|
||
"""便捷函数：执行完整合规检查"""
|
||
checker = ComplianceChecker()
|
||
return checker.check_text(html_content, platform, topic_data)
|
||
|
||
if __name__ == "__main__":
|
||
# 测试
|
||
test_html = "<html><body><h1>测试</h1>内容涉及赌博网站</body></html>"
|
||
result = check_article(test_html, "zhihu")
|
||
print(json.dumps(result, ensure_ascii=False, indent=2))
|