crawlo 1.4.5__py3-none-any.whl → 1.4.7__py3-none-any.whl

crawlo/tools/data_validator.py

@@ -1,181 +0,0 @@
-#!/usr/bin/python
-# -*- coding: UTF-8 -*-
-"""
-# @Time    : 2025-09-10 22:00
-# @Author  : crawl-coder
-# @Desc    : 数据验证工具
-"""
-
-import re
-from typing import Any, Union, Dict, List
-from datetime import datetime
-from urllib.parse import urlparse
-
-
-class DataValidator:
-    """数据验证工具类"""
-
-    @staticmethod
-    def validate_email(email: str) -> bool:
-        """
-        验证邮箱地址格式
-        
-        Args:
-            email (str): 邮箱地址
-            
-        Returns:
-            bool: 验证结果
-        """
-        pattern = r'^[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\.[a-zA-Z]{2,}$'
-        return bool(re.match(pattern, email))
-
-    @staticmethod
-    def validate_phone(phone: str, country_code: str = "CN") -> bool:
-        """
-        验证电话号码格式
-        
-        Args:
-            phone (str): 电话号码
-            country_code (str): 国家代码，默认为"CN"
-            
-        Returns:
-            bool: 验证结果
-        """
-        if country_code == "CN":
-            # 中国手机号码格式
-            pattern = r'^1[3-9]\d{9}$'
-            return bool(re.match(pattern, phone))
-        else:
-            # 通用格式，只检查是否全为数字且长度在7-15之间
-            pattern = r'^\d{7,15}$'
-            return bool(re.match(pattern, phone))
-
-    @staticmethod
-    def validate_url(url: str) -> bool:
-        """
-        验证URL格式
-        
-        Args:
-            url (str): URL地址
-            
-        Returns:
-            bool: 验证结果
-        """
-        try:
-            result = urlparse(url)
-            return all([result.scheme, result.netloc])
-        except Exception:
-            return False
-
-    @staticmethod
-    def validate_chinese_id_card(id_card: str) -> bool:
-        """
-        验证中国身份证号码格式
-        
-        Args:
-            id_card (str): 身份证号码
-            
-        Returns:
-            bool: 验证结果
-        """
-        # 18位身份证号码格式
-        pattern = r'^[1-9]\d{5}(18|19|20)\d{2}((0[1-9])|(1[0-2]))(([0-2][1-9])|10|20|30|31)\d{3}[0-9Xx]$'
-        return bool(re.match(pattern, id_card))
-
-    @staticmethod
-    def validate_date(date_str: str, date_format: str = "%Y-%m-%d") -> bool:
-        """
-        验证日期格式
-        
-        Args:
-            date_str (str): 日期字符串
-            date_format (str): 日期格式，默认为"%Y-%m-%d"
-            
-        Returns:
-            bool: 验证结果
-        """
-        try:
-            datetime.strptime(date_str, date_format)
-            return True
-        except ValueError:
-            return False
-
-    @staticmethod
-    def validate_number_range(value: Union[int, float], min_val: Union[int, float], 
-                             max_val: Union[int, float]) -> bool:
-        """
-        验证数值是否在指定范围内
-        
-        Args:
-            value (Union[int, float]): 要验证的数值
-            min_val (Union[int, float]): 最小值
-            max_val (Union[int, float]): 最大值
-            
-        Returns:
-            bool: 验证结果
-        """
-        return min_val <= value <= max_val
-
-    @staticmethod
-    def check_data_integrity(data: Dict[str, Any], required_fields: List[str]) -> Dict[str, Any]:
-        """
-        检查数据完整性，确保关键字段不为空
-        
-        Args:
-            data (Dict[str, Any]): 要检查的数据
-            required_fields (List[str]): 必需字段列表
-            
-        Returns:
-            Dict[str, Any]: 检查结果，包含缺失字段和空值字段
-        """
-        missing_fields = []
-        empty_fields = []
-        
-        for field in required_fields:
-            if field not in data:
-                missing_fields.append(field)
-            elif data[field] is None or data[field] == "":
-                empty_fields.append(field)
-                
-        return {
-            "is_valid": len(missing_fields) == 0 and len(empty_fields) == 0,
-            "missing_fields": missing_fields,
-            "empty_fields": empty_fields
-        }
-
-
-# 便捷函数
-def validate_email(email: str) -> bool:
-    """验证邮箱地址格式"""
-    return DataValidator.validate_email(email)
-
-
-def validate_phone(phone: str, country_code: str = "CN") -> bool:
-    """验证电话号码格式"""
-    return DataValidator.validate_phone(phone, country_code)
-
-
-def validate_url(url: str) -> bool:
-    """验证URL格式"""
-    return DataValidator.validate_url(url)
-
-
-def validate_chinese_id_card(id_card: str) -> bool:
-    """验证中国身份证号码格式"""
-    return DataValidator.validate_chinese_id_card(id_card)
-
-
-def validate_date(date_str: str, date_format: str = "%Y-%m-%d") -> bool:
-    """验证日期格式"""
-    return DataValidator.validate_date(date_str, date_format)
-
-
-def validate_number_range(value: Union[int, float], min_val: Union[int, float], 
-                         max_val: Union[int, float]) -> bool:
-    """验证数值是否在指定范围内"""
-    return DataValidator.validate_number_range(value, min_val, max_val)
-
-
-def check_data_integrity(data: Dict[str, Any], required_fields: List[str]) -> Dict[str, Any]:
-    """检查数据完整性"""
-    return DataValidator.check_data_integrity(data, required_fields)

crawlo/tools/encoding_converter.py

@@ -1,127 +0,0 @@
-#!/usr/bin/python
-# -*- coding: UTF-8 -*-
-"""
-# @Time    : 2025-09-10 22:00
-# @Author  : crawl-coder
-# @Desc    : 编码转换工具
-"""
-try:
-    import chardet
-
-    HAS_CHARDET = True
-except ImportError:
-    HAS_CHARDET = False
-from typing import Optional, Union
-
-
-class EncodingConverter:
-    """
-    编码转换工具类，提供各种编码转换功能。
-    特别适用于爬虫中处理不同编码的网页内容。
-    """
-
-    @staticmethod
-    def detect_encoding(data: Union[str, bytes]) -> Optional[str]:
-        """
-        检测数据编码
-        
-        :param data: 数据（字符串或字节）
-        :return: 检测到的编码
-        """
-        if isinstance(data, str):
-            # 如果是字符串，直接返回
-            return 'utf-8'
-
-        if not isinstance(data, bytes):
-            return None
-
-        if HAS_CHARDET:
-            try:
-                # 使用chardet检测编码
-                result = chardet.detect(data)
-                return result['encoding']
-            except Exception:
-                return None
-        else:
-            # 如果没有chardet，返回None
-            return None
-
-    @staticmethod
-    def to_utf8(data: Union[str, bytes], source_encoding: Optional[str] = None) -> Optional[str]:
-        """
-        转换为UTF-8编码的字符串
-        
-        :param data: 数据（字符串或字节）
-        :param source_encoding: 源编码（如果为None则自动检测）
-        :return: UTF-8编码的字符串
-        """
-        if isinstance(data, str):
-            # 如果已经是字符串，假设它已经是UTF-8
-            return data
-
-        if not isinstance(data, bytes):
-            return None
-
-        try:
-            if source_encoding is None:
-                # 自动检测编码
-                source_encoding = EncodingConverter.detect_encoding(data)
-                if source_encoding is None:
-                    # 如果检测失败，尝试常见编码
-                    for encoding in ['utf-8', 'gbk', 'gb2312', 'latin1']:
-                        try:
-                            decoded = data.decode(encoding)
-                            return decoded
-                        except UnicodeDecodeError:
-                            continue
-                    return None
-            else:
-                # 使用指定编码
-                return data.decode(source_encoding)
-
-            # 使用检测到的编码解码
-            return data.decode(source_encoding)
-        except Exception:
-            return None
-
-    @staticmethod
-    def convert_encoding(data: Union[str, bytes],
-                         source_encoding: Optional[str] = None,
-                         target_encoding: str = 'utf-8') -> Optional[bytes]:
-        """
-        编码转换
-        
-        :param data: 数据（字符串或字节）
-        :param source_encoding: 源编码（如果为None则自动检测）
-        :param target_encoding: 目标编码
-        :return: 转换后的字节数据
-        """
-        # 先转换为UTF-8字符串
-        utf8_str = EncodingConverter.to_utf8(data, source_encoding)
-        if utf8_str is None:
-            return None
-
-        try:
-            # 再转换为目标编码
-            return utf8_str.encode(target_encoding)
-        except Exception:
-            return None
-
-
-# =======================对外接口=======================
-
-def detect_encoding(data: Union[str, bytes]) -> Optional[str]:
-    """检测数据编码"""
-    return EncodingConverter.detect_encoding(data)
-
-
-def to_utf8(data: Union[str, bytes], source_encoding: Optional[str] = None) -> Optional[str]:
-    """转换为UTF-8编码的字符串"""
-    return EncodingConverter.to_utf8(data, source_encoding)
-
-
-def convert_encoding(data: Union[str, bytes],
-                     source_encoding: Optional[str] = None,
-                     target_encoding: str = 'utf-8') -> Optional[bytes]:
-    """编码转换"""
-    return EncodingConverter.convert_encoding(data, source_encoding, target_encoding)

crawlo/tools/network_diagnostic.py

@@ -1,365 +0,0 @@
-#!/usr/bin/python
-# -*- coding: UTF-8 -*-
-"""
-网络诊断工具
-提供网络连接问题的诊断和解决方案
-"""
-
-import asyncio
-import socket
-import time
-from typing import Dict, List, Optional, Tuple
-from urllib.parse import urlparse
-
-import aiohttp
-from crawlo.utils.log import get_logger
-
-
-class NetworkDiagnostic:
-    """网络诊断工具类"""
-    
-    def __init__(self):
-        self.logger = get_logger(self.__class__.__name__)
-        self._dns_cache: Dict[str, str] = {}
-    
-    async def diagnose_url(self, url: str) -> Dict[str, any]:
-        """
-        诊断URL的网络连接问题
-        
-        Args:
-            url: 要诊断的URL
-            
-        Returns:
-            诊断结果字典
-        """
-        parsed = urlparse(url)
-        hostname = parsed.hostname
-        port = parsed.port or (443 if parsed.scheme == 'https' else 80)
-        
-        result = {
-            'url': url,
-            'hostname': hostname,
-            'port': port,
-            'dns_resolution': None,
-            'tcp_connection': None,
-            'http_response': None,
-            'recommendations': []
-        }
-        
-        # DNS解析测试
-        dns_result = await self._test_dns_resolution(hostname)
-        result['dns_resolution'] = dns_result
-        
-        if dns_result['success']:
-            # TCP连接测试
-            tcp_result = await self._test_tcp_connection(hostname, port)
-            result['tcp_connection'] = tcp_result
-            
-            if tcp_result['success']:
-                # HTTP响应测试
-                http_result = await self._test_http_response(url)
-                result['http_response'] = http_result
-        
-        # 生成建议
-        result['recommendations'] = self._generate_recommendations(result)
-        
-        return result
-    
-    async def _test_dns_resolution(self, hostname: str) -> Dict[str, any]:
-        """测试DNS解析"""
-        try:
-            start_time = time.time()
-            
-            # 使用asyncio的DNS解析
-            loop = asyncio.get_event_loop()
-            addr_info = await loop.getaddrinfo(hostname, None)
-            
-            resolution_time = time.time() - start_time
-            ip_addresses = list(set([addr[4][0] for addr in addr_info]))
-            
-            # 缓存DNS结果
-            if ip_addresses:
-                self._dns_cache[hostname] = ip_addresses[0]
-            
-            return {
-                'success': True,
-                'ip_addresses': ip_addresses,
-                'resolution_time': resolution_time,
-                'error': None
-            }
-            
-        except socket.gaierror as e:
-            return {
-                'success': False,
-                'ip_addresses': [],
-                'resolution_time': None,
-                'error': {
-                    'type': 'DNSError',
-                    'code': e.errno,
-                    'message': str(e)
-                }
-            }
-        except Exception as e:
-            return {
-                'success': False,
-                'ip_addresses': [],
-                'resolution_time': None,
-                'error': {
-                    'type': type(e).__name__,
-                    'message': str(e)
-                }
-            }
-    
-    async def _test_tcp_connection(self, hostname: str, port: int) -> Dict[str, any]:
-        """测试TCP连接"""
-        try:
-            start_time = time.time()
-            
-            # 尝试TCP连接
-            reader, writer = await asyncio.wait_for(
-                asyncio.open_connection(hostname, port),
-                timeout=10.0
-            )
-            
-            connection_time = time.time() - start_time
-            
-            # 关闭连接
-            writer.close()
-            await writer.wait_closed()
-            
-            return {
-                'success': True,
-                'connection_time': connection_time,
-                'error': None
-            }
-            
-        except asyncio.TimeoutError:
-            return {
-                'success': False,
-                'connection_time': None,
-                'error': {
-                    'type': 'TimeoutError',
-                    'message': 'Connection timeout'
-                }
-            }
-        except Exception as e:
-            return {
-                'success': False,
-                'connection_time': None,
-                'error': {
-                    'type': type(e).__name__,
-                    'message': str(e)
-                }
-            }
-    
-    async def _test_http_response(self, url: str) -> Dict[str, any]:
-        """测试HTTP响应"""
-        try:
-            start_time = time.time()
-            
-            timeout = aiohttp.ClientTimeout(total=30, connect=10)
-            async with aiohttp.ClientSession(timeout=timeout) as session:
-                async with session.get(url) as response:
-                    response_time = time.time() - start_time
-                    
-                    return {
-                        'success': True,
-                        'url': url,
-                        'status_code': response.status_code,  # 修复：使用status_code而不是status
-                        'response_time': response_time,
-                        'content_length': len(response.body) if response.body else 0,
-                        'headers': dict(response.headers)
-                    }
-                    
-        except aiohttp.ClientError as e:
-            return {
-                'success': False,
-                'status_code': None,
-                'response_time': None,
-                'headers': {},
-                'error': {
-                    'type': type(e).__name__,
-                    'message': str(e)
-                }
-            }
-        except Exception as e:
-            return {
-                'success': False,
-                'status_code': None,
-                'response_time': None,
-                'headers': {},
-                'error': {
-                    'type': type(e).__name__,
-                    'message': str(e)
-                }
-            }
-    
-    def _generate_recommendations(self, result: Dict[str, any]) -> List[str]:
-        """根据诊断结果生成建议"""
-        recommendations = []
-        
-        dns_result = result.get('dns_resolution', {})
-        tcp_result = result.get('tcp_connection', {})
-        http_result = result.get('http_response', {})
-        
-        # DNS问题建议
-        if not dns_result.get('success'):
-            error = dns_result.get('error', {})
-            if error.get('code') == 8:  # nodename nor servname provided, or not known
-                recommendations.extend([
-                    "DNS解析失败 - 检查域名是否正确",
-                    "检查网络连接是否正常",
-                    "尝试使用不同的DNS服务器（如8.8.8.8或1.1.1.1）",
-                    "检查本地hosts文件是否有相关配置",
-                    "确认域名是否可以从外部访问"
-                ])
-            elif error.get('code') == 2:  # Name or service not known
-                recommendations.extend([
-                    "域名不存在或无法解析",
-                    "检查域名拼写是否正确",
-                    "确认域名是否已注册且配置了DNS记录"
-                ])
-        
-        # TCP连接问题建议
-        elif not tcp_result.get('success'):
-            error = tcp_result.get('error', {})
-            if error.get('type') == 'TimeoutError':
-                recommendations.extend([
-                    "TCP连接超时 - 服务器可能无响应",
-                    "检查防火墙设置是否阻止了连接",
-                    "尝试增加连接超时时间",
-                    "检查代理设置"
-                ])
-        
-        # HTTP问题建议
-        elif not http_result.get('success'):
-            error = http_result.get('error', {})
-            recommendations.extend([
-                f"HTTP请求失败: {error.get('message', 'Unknown error')}",
-                "检查URL是否正确",
-                "确认服务器是否正常运行"
-            ])
-        
-        # 性能建议
-        if dns_result.get('success') and dns_result.get('resolution_time', 0) > 1.0:
-            recommendations.append("DNS解析时间较长，考虑使用DNS缓存或更快的DNS服务器")
-        
-        if tcp_result.get('success') and tcp_result.get('connection_time', 0) > 2.0:
-            recommendations.append("TCP连接时间较长，可能存在网络延迟问题")
-        
-        if http_result.get('success') and http_result.get('response_time', 0) > 5.0:
-            recommendations.append("HTTP响应时间较长，服务器可能负载较高")
-        
-        return recommendations
-    
-    async def batch_diagnose(self, urls: List[str]) -> Dict[str, Dict[str, any]]:
-        """批量诊断多个URL"""
-        tasks = []
-        for url in urls:
-            task = asyncio.create_task(self.diagnose_url(url))
-            tasks.append((url, task))
-        
-        results = {}
-        for url, task in tasks:
-            try:
-                result = await task
-                results[url] = result
-            except Exception as e:
-                results[url] = {
-                    'url': url,
-                    'error': f"诊断过程出错: {e}",
-                    'recommendations': ["诊断工具本身出现问题，请检查网络环境"]
-                }
-        
-        return results
-    
-    def format_diagnostic_report(self, result: Dict[str, any]) -> str:
-        """格式化诊断报告"""
-        lines = [
-            f"=== 网络诊断报告 ===",
-            f"URL: {result['url']}",
-            f"主机: {result['hostname']}:{result['port']}",
-            ""
-        ]
-        
-        # DNS解析结果
-        dns = result.get('dns_resolution', {})
-        if dns.get('success'):
-            lines.extend([
-                "✅ DNS解析: 成功",
-                f"   IP地址: {', '.join(dns['ip_addresses'])}",
-                f"   解析时间: {dns['resolution_time']:.3f}秒"
-            ])
-        else:
-            error = dns.get('error', {})
-            lines.extend([
-                "❌ DNS解析: 失败",
-                f"   错误类型: {error.get('type', 'Unknown')}",
-                f"   错误信息: {error.get('message', 'Unknown error')}"
-            ])
-        
-        lines.append("")
-        
-        # TCP连接结果
-        tcp = result.get('tcp_connection', {})
-        if tcp and tcp.get('success'):
-            lines.extend([
-                "✅ TCP连接: 成功",
-                f"   连接时间: {tcp['connection_time']:.3f}秒"
-            ])
-        elif tcp:
-            error = tcp.get('error', {})
-            lines.extend([
-                "❌ TCP连接: 失败",
-                f"   错误类型: {error.get('type', 'Unknown')}",
-                f"   错误信息: {error.get('message', 'Unknown error')}"
-            ])
-        
-        lines.append("")
-        
-        # HTTP响应结果
-        http = result.get('http_response', {})
-        if http and http.get('success'):
-            lines.extend([
-                "✅ HTTP响应: 成功",
-                f"   状态码: {http['status_code']}",
-                f"   响应时间: {http['response_time']:.3f}秒"
-            ])
-        elif http:
-            error = http.get('error', {})
-            lines.extend([
-                "❌ HTTP响应: 失败",
-                f"   错误类型: {error.get('type', 'Unknown')}",
-                f"   错误信息: {error.get('message', 'Unknown error')}"
-            ])
-        
-        # 建议
-        recommendations = result.get('recommendations', [])
-        if recommendations:
-            lines.extend([
-                "",
-                "🔧 建议:",
-            ])
-            for i, rec in enumerate(recommendations, 1):
-                lines.append(f"   {i}. {rec}")
-        
-        return "\n".join(lines)
-
-
-# 便捷函数
-async def diagnose_url(url: str) -> Dict[str, any]:
-    """诊断单个URL的网络问题"""
-    diagnostic = NetworkDiagnostic()
-    return await diagnostic.diagnose_url(url)
-
-
-async def diagnose_urls(urls: List[str]) -> Dict[str, Dict[str, any]]:
-    """批量诊断URL的网络问题"""
-    diagnostic = NetworkDiagnostic()
-    return await diagnostic.batch_diagnose(urls)
-
-
-def format_report(result: Dict[str, any]) -> str:
-    """格式化诊断报告"""
-    diagnostic = NetworkDiagnostic()
-    return diagnostic.format_diagnostic_report(result)