我有一个带有地址列的大型DataFrame：
data addr
0 0.617964 IN,Krishnagiri,635115
1 0.635428 IN,Chennai,600005
2 0.630125 IN,Karnal,132001
3 0.981282 IN,Jaipur,302021
4 0.715813 IN,Chennai,600005
...
并且我编写了以下函数，用地址的经度和纬度坐标替换地址：
from geopy.geocoders import Nominatim
geo_locator = Nominatim(user_agent="MY_APP_ID")
def get_coordinates(addr):
location = geo_locator.geocode(addr)
if location is not None:
return pd.Series({'lat': location.latitude, 'lon': location.longitude})
location = geo_locator.geocode(addr.split(',')[0])
if location is not None:
return pd.Series({'lat': location.latitude, 'lon': location.longitude})
return pd.Series({'lat': -1, 'lon': -1})
然后在地址列上调用pandas apply方法，并将结果隐藏到DF的末尾，而不是地址列：
df = pd.concat([df, df.addr.apply(get_coordinates)], axis=1).drop(['addr'], axis=1)
但是，由于get_coordinates调用了第三方API，因此它失败了：geopy.exc.GeocoderTimedOut: Service timed out
在继续下一个值之前，如何限制请求以确保得到响应？
更新：
为了进一步改进，我只想对唯一值调用API，即：如果地址IN,Krishnagiri,635115在我的DataFrame中出现20次，我只想调用一次并将结果应用于所有20次出现。
更新2：
日志+堆栈跟踪，用于@Andrew Lavers代码：
...
Fetched Gandipet, Khanapur, Rangareddy District, Telangana, 500075, India
Fetched Jaipur Municipal Corporation, Jaipur, Rajasthan, 302015, India
Fetched Chennai, Chennai district, Tamil Nadu, India
Exception from geolocator: Fake exception for testing
Backing off for 1 seconds.
Exception from geolocator: Fake exception for testing
Backing off for 3 seconds.
Fetched None
Traceback (most recent call last):
File "/usr/local/Cellar/python3/3.6.1/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/geopy/geocoders/base.py", line 344, in _call_geocoder
page = requester(req, timeout=timeout, **kwargs)
File "/usr/local/Cellar/python3/3.6.1/Frameworks/Python.framework/Versions/3.6/lib/python3.6/urllib/request.py", line 526, in open
response = self._open(req, data)
File "/usr/local/Cellar/python3/3.6.1/Frameworks/Python.framework/Versions/3.6/lib/python3.6/urllib/request.py", line 544, in _open
'_open', req)
File "/usr/local/Cellar/python3/3.6.1/Frameworks/Python.framework/Versions/3.6/lib/python3.6/urllib/request.py", line 504, in _call_chain
result = func(*args)
File "/usr/local/Cellar/python3/3.6.1/Frameworks/Python.framework/Versions/3.6/lib/python3.6/urllib/request.py", line 1361, in https_open
context=self._context, check_hostname=self._check_hostname)
File "/usr/local/Cellar/python3/3.6.1/Frameworks/Python.framework/Versions/3.6/lib/python3.6/urllib/request.py", line 1321, in do_open
r = h.getresponse()
File "/usr/local/Cellar/python3/3.6.1/Frameworks/Python.framework/Versions/3.6/lib/python3.6/http/client.py", line 1331, in getresponse
response.begin()
File "/usr/local/Cellar/python3/3.6.1/Frameworks/Python.framework/Versions/3.6/lib/python3.6/http/client.py", line 297, in begin
version, status, reason = self._read_status()
File "/usr/local/Cellar/python3/3.6.1/Frameworks/Python.framework/Versions/3.6/lib/python3.6/http/client.py", line 258, in _read_status
line = str(self.fp.readline(_MAXLINE + 1), "iso-8859-1")
File "/usr/local/Cellar/python3/3.6.1/Frameworks/Python.framework/Versions/3.6/lib/python3.6/socket.py", line 586, in readinto
return self._sock.recv_into(b)
File "/usr/local/Cellar/python3/3.6.1/Frameworks/Python.framework/Versions/3.6/lib/python3.6/ssl.py", line 1002, in recv_into
return self.read(nbytes, buffer)
File "/usr/local/Cellar/python3/3.6.1/Frameworks/Python.framework/Versions/3.6/lib/python3.6/ssl.py", line 865, in read
return self._sslobj.read(len, buffer)
File "/usr/local/Cellar/python3/3.6.1/Frameworks/Python.framework/Versions/3.6/lib/python3.6/ssl.py", line 625, in read
v = self._sslobj.read(len, buffer)
socket.timeout: The read operation timed out
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/Users/...//tmp.py", line 89, in <module>
df.addr.apply(get_coordinates)
File "/usr/local/Cellar/python3/3.6.1/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/pandas/core/series.py", line 3194, in apply
mapped = lib.map_infer(values, f, convert=convert_dtype)
File "pandas/_libs/src/inference.pyx", line 1472, in pandas._libs.lib.map_infer
File "/Users/...//tmp.py", line 76, in get_coordinates
location = geo_locator.geocode(addr.split(',')[0])
File "/usr/local/Cellar/python3/3.6.1/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/geopy/geocoders/osm.py", line 307, in geocode
self._call_geocoder(url, timeout=timeout), exactly_one
File "/usr/local/Cellar/python3/3.6.1/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/geopy/geocoders/base.py", line 371, in _call_geocoder
raise GeocoderTimedOut('Service timed out')
geopy.exc.GeocoderTimedOut: Service timed out
Process finished with exit code 1
答案 0 :(得分：2)
这里有一些 testing 代码可能会有所帮助。 1）对Api指定的速率进行简单的限制（Nominatum似乎是每秒1，但我获得的成功低至0.1秒）。 2）简单地将结果缓存在字典中，可通过参数进行控制以进行测试3）重试循环，并具有乘法减慢和线性加速功能。 （速度变慢，速度变慢）4）测试伪造错误的异常
我无法复制您遇到的问题-可能是由于您使用API的路径。
更健壮的策略可能是建立本地持久性缓存并继续重试，直到建立完整的批处理为止。缓存可以是写为csv的pandas数据帧到文件。整个伪代码类似。
repeat until all addresses are in the cache
cache = pd.read_csv("cache.csv)
addressess_to_get = addresses in df that are not in cache
for batch of n addresses in addresses_to_get:
cache.add(get_location(addr))
cache.write_csv("cache.csv")
这是经过测试的代码
import datetime
import time
import pandas as pd
from geopy.geocoders import Nominatim
geo_locator = Nominatim(user_agent="notarealemail@gmail.com")
# Define the rate limit function and associated global variable
last_time = datetime.datetime.now()
backoff_time = 0
def rate_limit(min_interval_seconds = .1):
global last_time
sleep = min_interval_seconds - (datetime.datetime.now() - last_time).total_seconds()
if sleep > 0 :
print(f'Sleeping for {sleep} seconds')
time.sleep(sleep)
last_time = datetime.datetime.now()
# make a cache dictionary keyed by address
geo_cache = {}
backoff_seconds = 0
def get_coordinates_with_retry(addr):
# Return coords from global cache if it exists
global backoff_seconds
# set the backoff intital values and factors
max_backoff_seconds = 60
backoff_exponential = 2
backoff_linear = 2
# rate limited API call
rate_limit()
# Retry until max_back_seconds is reached
while backoff_seconds < max_backoff_seconds: # backoff up to this time
if backoff_seconds > 0:
print(f"Backing off for {backoff_seconds} seconds.")
time.sleep(backoff_seconds)
try:
location = geo_locator.geocode(addr)
# REMOVE THIS: fake an error for testing
#import random
#if random.random() < .3:
# raise(Exception("Fake exception for testing"))
# Success - so reduce the backoff linearly
print (f"Fetched {location} for address {addr}")
backoff_seconds = backoff_seconds - backoff_linear if backoff_seconds > backoff_linear else 0
break
except Exception as e:
print(f"Exception from geolocator: {e}")
# Backoff exponentially
backoff_seconds = 1 + backoff_seconds * backoff_exponential
if backoff_seconds > max_backoff_seconds:
raise Exception("Max backoff reached\n")
return(location)
def get_coordinates(addr, useCache = True):
# Return from cache if previously loaded
global geo_cache
if addr in geo_cache:
return geo_cache[addr]
# Attempt using the full address
location = get_coordinates_with_retry(addr)
# Attempt using the first part only if None found
if location is not None:
result = pd.Series({'lat': location.latitude, 'lon': location.longitude})
else :
print (f"Trying split address for address {addr}")
location = get_coordinates_with_retry(addr.split(',')[0])
if location is not None:
result = pd.Series({'lat': location.latitude, 'lon': location.longitude})
else:
result = pd.Series({'lat': -1, 'lon': -1})
# assign to cache
if useCache:
geo_cache[addr] = result
return(result)
# Use the test data
df = pd.DataFrame({'addr' : [
'IN,Krishnagiri,635115',
'IN,Chennai,600005',
'IN,Karnal,132001',
'IN,Jaipur,302021',
'IN,Chennai,600005']})
# repeat the test data to make alarger set
df = pd.concat([df, df, df, df, df, df, df, df, df, df])
df.addr.apply(get_coordinates)
print(f"Address cache contains {len(geo_cache)} address locations.")