Python批量API请求处理：数据整合、限流与错误管理

import requests
import json
import pandas as pd
from contextlib import contextmanager
from time import sleep
import http

# 定义全局变量用于API调用计数，以便后续限流使用
api_calls_counter = 0

def get_driving_distance(lat1, lon1, lat2, lon2):
    """
    通过OSRM API获取两点之间的驾驶距离。

    参数:
    lat1 (float): 起点纬度
    lon1 (float): 起点经度
    lat2 (float): 终点纬度
    lon2 (float): 终点经度

    返回:
    float: 驾驶距离 (英里)
    """
    # OSRM API的经纬度顺序是 经度,纬度
    api_url = f"http://router.project-osrm.org/route/v1/car/{lon1},{lat1};{lon2},{lat2}?overview=false"
    try:
        response = requests.get(api_url)
        # 检查HTTP响应状态码，如果不是2xx，则抛出HTTPError
        response.raise_for_status()

        routes = json.loads(response.content)
        # 确保API返回了有效的路线数据
        if routes and "routes" in routes and len(routes["routes"]) > 0:
            route_info = routes["routes"][0]
            driving_distance_meters = route_info.get('distance')
            if driving_distance_meters is not None:
                # 将米转换为英里 (1英里 = 1609.34米)
                return driving_distance_meters / 1609.34
            else:
                print(f"警告: API响应中未找到距离信息 for {lat1},{lon1};{lat2},{lon2}")
                return None
        else:
            print(f"警告: API响应中未找到有效路线 for {lat1},{lon1};{lat2},{lon2}. 响应: {routes}")
            return None
    except requests.exceptions.HTTPError as e:
        print(f"HTTP错误: {e.response.status_code} - {e.response.text} for {lat1},{lon1};{lat2},{lon2}")
        return None
    except requests.exceptions.ConnectionError as e:
        print(f"连接错误: 无法连接到API服务 for {lat1},{lon1};{lat2},{lon2}. 错误: {e}")
        return None
    except requests.exceptions.Timeout as e:
        print(f"请求超时: API请求超时 for {lat1},{lon1};{lat2},{lon2}. 错误: {e}")
        return None
    except json.JSONDecodeError as e:
        print(f"JSON解析错误: 无法解析API响应 for {lat1},{lon1};{lat2},{lon2}. 错误: {e}")
        return None
    except Exception as e:
        print(f"发生未知错误: {e} for {lat1},{lon1};{lat2},{lon2}")
        return None

@contextmanager
def rate_limited(limit=500, delay=5):
    """
    一个上下文管理器，用于限制API调用速率。
    在达到指定调用次数后，暂停一段时间。

    参数:
    limit (int): 在暂停前允许的最大API调用次数。
    delay (int): 暂停的秒数。
    """
    global api_calls_counter
    # 检查是否即将达到限制，如果是，则暂停并重置计数器
    if api_calls_counter + 1 >= limit:
        print(f"达到API调用上限 {limit} 次，暂停 {delay} 秒...")
        sleep(delay)
        api_calls_counter = 0  # 重置计数器，或者可以 api_calls_counter -= limit

    # 增加API调用计数
    api_calls_counter += 1
    yield # 执行被包装的代码块

# 更新后的 get_driving_distance 函数，包含速率限制
def get_driving_distance_with_rate_limit(lat1, lon1, lat2, lon2):
    """
    通过OSRM API获取两点之间的驾驶距离，并集成速率限制。
    """
    with rate_limited(): # 使用速率限制上下文管理器
        api_url = f"http://router.project-osrm.org/route/v1/car/{lon1},{lat1};{lon2},{lat2}?overview=false"
        try:
            response = requests.get(api_url)
            response.raise_for_status() # 检查HTTP状态码

            routes = json.loads(response.content)
            if routes and "routes" in routes and len(routes["routes"]) > 0:
                route_info = routes["routes"][0]
                driving_distance_meters = route_info.get('distance')
                if driving_distance_meters is not None:
                    return driving_distance_meters / 1609.34
                else:
                    print(f"警告: API响应中未找到距离信息 for {lat1},{lon1};{lat2},{lon2}")
                    return None
            else:
                print(f"警告: API响应中未找到有效路线 for {lat1},{lon1};{lat2},{lon2}. 响应: {routes}")
                return None
        except requests.exceptions.HTTPError as e:
            print(f"HTTP错误: {e.response.status_code} - {e.response.text} for {lat1},{lon1};{lat2},{lon2}")
            return None
        except requests.exceptions.RequestException as e: # 捕获所有requests相关的异常
            print(f"请求错误: {e} for {lat1},{lon1};{lat2},{lon2}")
            return None
        except json.JSONDecodeError as e:
            print(f"JSON解析错误: 无法解析API响应 for {lat1},{lon1};{lat2},{lon2}. 错误: {e}")
            return None
        except Exception as e:
            print(f"发生未知错误: {e} for {lat1},{lon1};{lat2},{lon2}")
            return None

# 示例数据
location_latitudes = [51.5074, 51.5100, 51.5200]
location_longitudes = [-0.1278, -0.1300, -0.1400]
station_latitudes = [51.5000, 51.5050, 51.5150]
station_longitudes = [-0.1000, -0.1100, -0.1200]

# 确保所有列表长度一致
if not (len(location_latitudes) == len(location_longitudes) == 
        len(station_latitudes) == len(station_longitudes)):
    raise ValueError("所有输入列表的长度必须一致。")

# 存储计算结果
driving_distances = []

# 遍历所有数据对并调用API
for i in range(len(location_latitudes)):
    lat1 = location_latitudes[i]
    lon1 = location_longitudes[i]
    lat2 = station_latitudes[i]
    lon2 = station_longitudes[i]

    distance = get_driving_distance_with_rate_limit(lat1, lon1, lat2, lon2)
    driving_distances.append(distance)

# 创建DataFrame
data = {
    'Location_Latitude': location_latitudes,
    'Location_Longitude': location_longitudes,
    'Station_Latitude': station_latitudes,
    'Station_Longitude': station_longitudes,
    'Driving_Distance_Miles': driving_distances
}

df = pd.DataFrame(data)
print("\n最终结果DataFrame:")
print(df)