Shwuyechaxunhetongdaoqi

查询上海市物业项目合同到期信息，提取招标公告合同期限和中标/评标日期，计算合同到期时间并生成CSV。

Security Scan

The skill's stated purpose (scraping Shanghai property tender announcements and computing contract expiry) matches its code and instructions, but there are several incoherencies and packaging omissions that warrant caution before installing or running it.

Purpose & Capability

The skill claims to be a Python-based scraper/OCR pipeline and includes a Python script and Python library names in SKILL.md, but package.json lists Python libraries (requests, beautifulsoup4, pytesseract, etc.) as npm dependencies and also declares a dependency on another skill ('shwuyeyanjiu'). This mismatch between packaging metadata and runtime language is incoherent and suggests sloppy packaging or incorrect distribution metadata. The code also expects another skill's scripts to exist at a relative path ('../.. /shwuyeyanjiu/scripts') which is not declared as a required skill or dependency in a way the platform enforces.

Instruction Scope

The SKILL.md instructs the agent to download PDFs, run OCR (pytesseract), convert PDFs to images (pdf2image), and parse dates — all reasonable for the stated purpose. However: (1) OCR requires external native binaries (Tesseract) and pdf2image typically needs poppler; the skill declares no required binaries, so required system-level dependencies are omitted. (2) The included script alters sys.path to import code from a sibling skill ('shwuyeyanjiu'), which means it will read/execute code outside this skill's directory — a cross-skill dependency that isn't clearly declared or sandboxed. Both issues expand the runtime scope beyond what's explicitly stated.

Install Mechanism

There is no install spec (instruction-only + small script), which is lower-risk. However, package.json is present and inconsistent with a Python runtime (it lists Python libs under 'dependencies' and references a repository URL). The SKILL.md suggests using an environment/tool ('uv run' and 'uv sync') to install/run dependencies; it's unclear what 'uv' will do and whether it will fetch native binaries. This ambiguity increases operational risk but is not necessarily malicious.

Credentials

The skill does not request any environment variables, credentials, or config paths. Its data sources are publicly listed (a Shanghai government announcements site). There are no obvious requests for unrelated secrets or cloud credentials.

Persistence & Privilege

The skill does not request always:true and is user-invocable only. It does not attempt to modify other skills' configurations in the files provided. The main concern here is cross-skill import (sys.path insertion), which increases the effective trusted surface but is not the same as requesting elevated platform privileges.

What to consider before installing

What to check before installing or running this skill: - Packaging mismatch: package.json lists Python libraries and a dependency on 'shwuyeyanjiu' even though this is a Python-based skill. Confirm how dependencies are actually installed on your platform (what 'uv sync' / 'uv run' does) and whether package.json is meaningful here. - Native binaries: OCR and pdf2image require native tools (Tesseract, poppler). Verify these are available or will be installed and that you trust the source that will install them. - Cross-skill import: the script modifies sys.path to import code from a sibling skill (shwuyeyanjiu). Inspect the code of that other skill before running this one — it could execute additional logic or access data you didn't expect. The repository did not include shwuyeyanjiu, so you should obtain and review it or sandbox execution. - Source provenance: the skill's homepage is 'none' and source is 'unknown'. If you plan to run it, prefer running inside a restricted environment (container or VM) and review the full code of any referenced skill (shwuyeyanjiu) and any install scripts 'uv' would run. - Operational safety: run one district first and inspect CSV output; ensure no unexpected network calls to endpoints outside the listed government site. If you need help, ask the author for a clear install/run guide (how dependencies and native binaries are installed) and for the code of the referenced shwuyeyanjiu skill. Given the packaging and dependency ambiguities, treat this skill as potentially safe in intent but operationally sloppy — review the missing pieces and the other skill it depends on before trusting it with real data or broad permissions.

SKILL.md - 上海物业查询合同到期

📋 技能描述

这个技能用于查询上海市物业项目的合同到期信息，通过分析招标公告、中标公告和评标结果公告，提取合同期限和中标日期，计算合同到期时间。

适用场景：

• 查询某个区域的物业项目合同到期情况
• 识别即将到期的物业项目
• 为物业续约决策提供数据支持

---

🎯 核心功能

1. 数据源

• 上海市物业招标公告网站：https://962121.fgj.sh.gov.cn/wyweb/web/front/common/greenmorelist.jsp
• 公告类型：
  • 招标公告：提取合同期限
  • 中标公告：提取中标日期（优先级1）
  • 评标结果公告：提取中标日期（优先级2）

2. 处理流程

1. 搜索指定区域的物业项目
2. 分类整理招标公告、中标公告、评标结果公告
3. 下载PDF文件
4. OCR识别PDF内容
5. 提取合同期限和中标日期
6. 计算合同到期时间
7. 生成CSV结果文件

3. 优先级规则

• 合同期限来源：招标公告
• 中标日期来源：
  • 优先级1：中标公告
  • 优先级2：评标结果公告（当中标公告不存在时）
• 合同到期时间计算：中标日期 + 合同期限

---

⚠️ 常见错误与解决方案

错误1：大量项目中标日期为空

现象：

• 很多项目有合同期限，但没有中标日期
• 导致无法计算合同到期时间

原因：

• 只处理了中标公告，但很多项目没有中标公告
• 评标结果公告比中标公告多得多（3014 vs 878）

解决方案：

# 增加评标结果公告的处理
projects[project_name] = {
    '招标公告': [],
    '中标公告': [],
    '评标结果公告': []  # 新增
}

# 优先级逻辑
if 中标公告存在:
    使用中标公告的日期
elif 评标结果公告存在:
    使用评标结果公告的日期

效果：

• 提取率从36%提升到73%
• 识别出的2026年内到期项目从1个增加到10个

---

错误2：OCR识别日期格式不匹配

现象：

• OCR成功识别，但日期提取失败
• 不同公告的日期格式不统一

原因：

• 正则表达式模式不够全面
• 日期格式多样（YYYY年MM月DD日、YYYY-MM-DD、MM月DD日等）

解决方案：

# 增加多种日期格式的正则表达式
patterns = [
    r'(\d{4})年(\d{1,2})月(\d{1,2})日',  # 2024年7月2日
    r'(\d{4})-(\d{1,2})-(\d{1,2})',       # 2024-07-02
    r'(\d{1,2})月(\d{1,2})日',             # 7月2日（假设当前年份）
    r'(\d{2})年(\d{1,2})月(\d{1,2})日',   # 24年7月2日
]

---

错误3：PDF下载失败

现象：

• 部分公告的PDF文件下载失败
• 导致无法提取信息

原因：

• 网络问题
• PDF链接失效
• 服务器限流

解决方案：

# 增加重试机制
for attempt in range(3):
    try:
        response = requests.get(pdf_url, timeout=30)
        if response.status_code == 200:
            break
    except:
        time.sleep(2)

---

📊 最佳实践

1. 数据源选择

• 优先使用中标公告：数据最准确
• 评标结果公告作为补充：覆盖更多项目
• 招标公告提取合同期限：必须处理

2. 批量处理

• 使用后台进程处理大量项目
• 定期汇报进度（避免用户长时间等待）
• 生成CSV文件便于后续分析

3. 数据验证

• 检查日期格式的合理性
• 标注数据来源（中标公告/评标结果公告）
• 标注失败原因（PDF下载失败、未找到公告等）

---

🔧 技术实现

核心脚本

• 脚本位置：~/.openclaw/workspace/skills/shwuyeyanjiu/scripts/
• 主要脚本：
  • batch_extract_dates.py：初版脚本
  • batch_extract_dates_v2.py：改进版脚本（增加评标结果公告处理）

依赖库

requests          # HTTP请求
beautifulsoup4    # HTML解析
pdf2image         # PDF转图片
pytesseract       # OCR识别
python-dateutil   # 日期处理

运行方式

cd ~/.openclaw/workspace/skills/shwuyeyanjiu/scripts
uv run --with requests --with beautifulsoup4 --with pdf2image --with pytesseract --with python-dateutil python3 batch_extract_dates_v2.py

---

📈 效果对比

改进前（只处理中标公告）

• 处理项目：96个
• 成功提取中标日期：35个（36%）
• 成功计算合同到期时间：34个（35%）
• 2026年内到期项目：1个

改进后（增加评标结果公告）

• 处理项目：96个
• 成功提取中标日期：约70个（73%）
• 成功计算合同到期时间：约70个（73%）
• 2026年内到期项目：10个

---

💡 关键教训

1. 数据源的重要性

   • 评标结果公告比中标公告多得多
   • 必须同时处理多种公告类型

2. 用户反馈的价值

   • 用户指出了数据不完整的问题
   • 提出了具体的改进建议
   • 这些建议非常有效

3. OCR识别的挑战

   • PDF扫描件需要OCR
   • 日期格式多样，需要多种正则表达式

4. 批量处理的注意事项

   • 需要后台运行
   • 需要定期汇报进度
   • 需要处理失败情况

---

🎯 使用示例

查询静安区2026年内到期项目

# 运行改进版脚本
python3 batch_extract_dates_v2.py

# 结果文件
jingan_contract_dates_v2.csv

结果示例

项目名称,合同期限,中标日期,合同到期时间,备注
微星彭浦公寓,3,2023-04-21,2026-04-30, (日期来源: 评标结果公告)
市北云盛公寓,3,2023-04-08,2026-04-30, (日期来源: 评标结果公告)
闸北区339街坊北上海物流号地块配套商品房,2,2024-07-02,2026-07-31, (日期来源: 中标公告)

---

📝 未来改进方向

1. 提高OCR识别准确率

   • 尝试其他OCR引擎
   • 优化图像预处理

2. 增加数据验证

   • 自动检查日期合理性
   • 标注异常数据

3. 支持更多区域

   • 扩展到其他行政区
   • 支持全市范围查询

4. 实时更新

   • 定期自动更新数据
   • 监控即将到期的项目

---

技能创建时间：2026-04-08 创建者：傲小喵 (Ao Xiao Miao) 🐱