# -*- coding:UTF-8 -*-
# @Time : 2021/1/4 13:39
# @File : crawler_xiaohongshu.py
# @email : litao@igengmei.com
# @author : litao
import copy
import redis
import requests
import json
import datetime
import re
# from . import bulk_write_into_es
import hashlib
import time
from selenium import webdriver
from crawler.crawler_sys.utils.output_results import retry_get_url
from crawler_sys.framework.video_fields_std import Std_fields_video
from crawler_sys.utils.output_results import output_result
from crawler.gm_upload.gm_upload import upload, upload_file
from selenium.webdriver import ActionChains
from selenium import webdriver
try:
from crawler_sys.framework.func_get_releaser_id import *
except:
from func_get_releaser_id import *
from crawler.crawler_sys.proxy_pool.func_get_proxy_form_kuaidaili import get_proxy
# from selenium.webdriver.common.desired_capabilities import DesiredCapabilities
from crawler.crawler_sys.utils.trans_str_play_count_to_int import trans_play_count
import random, urllib
# from crawler.crawler_sys.utils.rpc_data_to_answer import post_data
rds = redis.StrictRedis(host='172.18.51.10', port=6379, db=17, decode_responses=True)
user_id_list = [29865245,
36426151,
36426142,
36427666,
36427661,
36427657,
36427655,
36427634,
33524762,
33524779,
33524697,
30963358,
31293584,
31358392,
31358396,
31358397,
31358419,
31358448,
31358610,
31358658,
]
class Crawler_xiaohongshu():
def __init__(self, timeout=None, platform='xiaohongshu'):
if timeout == None:
self.timeout = 10
else:
self.timeout = timeout
self.platform = platform
self.TotalVideo_num = None
self.midstepurl = None
std_fields = Std_fields_video()
self.video_data = std_fields.video_data
self.video_data['platform'] = self.platform
self.chrome_options = webdriver.ChromeOptions()
self.chrome_options.add_argument('--headless')
self.chrome_options.add_argument('--disable-gpu')
self.chrome_options.add_argument("--no-sandbox")
# proxies_dic = get_proxy(1)
# proxies_dic_list = proxies_dic["http"].split(":")
# proxy_server = '--proxy-server=http://{ip}:{port}'.format(ip=proxies_dic_list[1].replace("/",""), port=int(proxies_dic_list[2]))
# print(proxy_server)
# self.chrome_options.add_argument(proxy_server)
# self.chrome_options.add_argument('User-Agent="Mozilla/5.0 (iPhone; CPU iPhone OS 13_2_3 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/13.0.3 Mobile/15E148 Safari/604.1"')
# self.chrome_options.add_argument('accept="text/html,application/xhtml+xml,application/xml;q=0.9,image/avif,image/webp,image/apng,*/*;q=0.8,application/signed-exchange;v=b3;q=0.9"')
# self.chrome_options.add_argument('accept-encoding="gzip, deflate"')
# self.chrome_options.add_argument('accept-language="zh-CN,zh;q=0.9"')
# self.chrome_options.add_argument('cache-control="max-age=0"')
# self.chrome_options.add_argument('referer="https://www.xiaohongshu.com/explore"')
# self.chrome_options.add_argument('sec-fetch-dest="document"')
# self.chrome_options.add_argument('sec-fetch-mode="navigate"')
# self.chrome_options.add_argument('sec-fetch-user="?1"')
# self.chrome_options.add_argument('upgrade-insecure-requests="1"')
self.chrome_options.add_experimental_option('excludeSwitches', ['enable-automation'])
self.timestamp = str(datetime.datetime.now().timestamp() * 1e3)
prefs = {"profile.managed_default_content_settings.images": 2}
self.chrome_options.add_experimental_option("prefs", prefs)
# self.driver = webdriver.Chrome(options=self.chrome_options)
def __exit__(self):
self.driver.close()
def get_one_page(self, page_id, proxies=0,cookies={}):
url = "https://www.xiaohongshu.com/discovery/item/%s" % page_id
headers = {
"accept": "text/html,application/xhtml+xml,application/xml;q=0.9,image/avif,image/webp,image/apng,*/*;q=0.8,application/signed-exchange;v=b3;q=0.9",
"accept-encoding": "gzip, deflate, br",
"accept-language": "zh-CN,zh;q=0.9",
"cache-control": "max-age=0",
# "cookie": "timestamp2=202101062497d4bed842476b2618e0ea;",
"referer": "https://www.xiaohongshu.com/explore",
"sec-fetch-dest": "document",
"sec-fetch-mode": "navigate",
"sec-fetch-site": "same-origin",
"sec-fetch-user": "?1",
"upgrade-insecure-requests": "1",
"user-agent": "Mozilla/5.0 (iPhone; CPU iPhone OS 13_2_3 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/13.0.3 Mobile/15E148 Safari/604.1",
}
res = retry_get_url(url, headers=headers, proxies=proxies,cookies=cookies)
res_text = res.text
res_json_text = re.findall('__INITIAL_SSR_STATE__=(.*?)</script>',res_text, flags=re.DOTALL)[0]
# scope = {}
json_data = json.loads(res_json_text.replace("undefined","null"))
# exec("res_json =" + res_json_text.strip("\n"),scope)
qiniu_img_list = []
# for img_url in json_data["NoteView"]["content"]["imageList"]:
# try:
# img_wb = retry_get_url(img_url["url"]).content
# res = upload(img_wb,img_type=99)
# print(res)
# img_info = retry_get_url(res + "-imageinfo")
# img_info_json = img_info.json()
# qiniu_img_list.append((res + "-w", img_info_json))
# except Exception as e:
# print("down load img error %s" % e)
# return {}
# json_data["NoteView"]["content"]["qiniu_img_list"] = qiniu_img_list
return json_data
def get_releaser_id(self, releaserUrl):
return get_releaser_id(platform="xiaohongshu", releaserUrl=releaserUrl)
def releaser_page(self, releaserUrl,
output_to_file=False,
filepath=None,
releaser_page_num_max=30,
output_to_es_raw=False,
es_index=None,
doc_type=None,
output_to_es_register=False,
push_to_redis=False, proxies_num=None, **kwargs):
# self.driver.get("https://www.xiaohongshu.com/")
# time.sleep(1)
# self.driver.implicitly_wait(2)
# self.driver.add_cookie(cookie_dict={'name': 'timestamp2', 'value': '2021010899964852bd70ca4c0c991c6c'})
# page_element = self.driver.find_element_by_xpath("//*").get_attribute("outerHTML")
# # print(page_element)
# self.driver.get("https://www.xiaohongshu.com/explore")
# self.driver.implicitly_wait(2)
# time.sleep(1)
# page_element = self.driver.find_element_by_xpath("//*").get_attribute("outerHTML")
# # print(page_element)
#
# self.driver.find_element_by_xpath("/html[1]/body[1]/div[1]/div[1]/div[1]/div[1]/div[2]/div[1]/div[1]/div[1]/div[1]/div[1]/div[1]/div[1]").click()
# self.driver.implicitly_wait(2)
# time.sleep(1)
# page_element = self.driver.find_element_by_xpath("//*").get_attribute("outerHTML")
# print(page_element)
# cookie = self.driver.get_cookies()
#
# return True
# print(self.driver.get_log("performance"))
# cookie_dic={}
# for k in cookie:
# cookie_dic[k["name"]] = k["value"]
# print(cookie_dic)
headers = {
"host":"www.xiaohongshu.com",
"accept": "text/html,application/xhtml+xml,application/xml;q=0.9,image/avif,image/webp,image/apng,*/*;q=0.8,application/signed-exchange;v=b3;q=0.9",
"accept-encoding": "gzip, deflate, br",
"accept-language": "zh-CN,zh;q=0.9",
"cache-control": "no-cache",
# "cookie": "xhsTrackerId=a81077f9-661a-4731-c790-ac6fbbeaa44b; extra_exp_ids=gif_clt1,ques_exp2; xhsuid=2EFsw5qOMk70l1we; timestamp2=2021010899964852bd70ca4c0c991c6c; timestamp2.sig=Lj3xTHgJ-JO20IUULPRnAhACddlzUtd7AsUzrlJQbWc; xhs_spses.5dde=*; xhsTracker=url=index&searchengine=baidu; xhs_spid.5dde=4dc700089fbdde46.1610082780.1.1610083480.1610082780.d70776d0-eac9-4684-912e-130f0cdb86a1",
"pragma": "no-cache",
"sec-ch-ua": '"Google Chrome";v="87", " Not;A Brand";v="99", "Chromium";v="87"',
"sec-ch-ua-mobile": "?0",
"sec-fetch-dest": "document",
"sec-fetch-mode": "navigate",
"sec-fetch-site": "none",
"sec-fetch-user": "?1",
"upgrade-insecure-requests": "1",
"user-agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/87.0.4280.88 Safari/537.36",
}
count = 1
# has_more = True
retry_time = 0
result_list = []
releaser_id = self.get_releaser_id(releaserUrl)
releaserUrl = 'http://www.xiaohongshu.com/user/profile/%s' % releaser_id
pcursor = 0
cookie_dic = {'timestamp2': '2021010899964852bd70ca4c0c991c6c'}
# proxies = {'http': 'http://hanye:i9mmu0a3@58.55.159.141:16085/', 'https': 'http://hanye:i9mmu0a3@58.55.159.141:16085/'}
while count <= releaser_page_num_max and count <= 1:
try:
print(releaserUrl)
res = retry_get_url(releaserUrl, headers=headers, proxies=proxies_num, cookies=cookie_dic)
except:
continue
# print(get_page.content)
# time.sleep(random.randint(1, 2))
page_text = res.text
# print(page_text)
data_list = re.findall("window.__INITIAL_SSR_STATE__=(.*?)</script>", page_text)[0]
# print(data_list)
data_json = json.loads(data_list.replace("undefined","null"))
# # print(data_list)
if data_json:
print("get data at releaser: %s page: %s" % (releaser_id, count))
count += 1
for info_dic in data_json["Main"]["notesDetail"]:
video_dic = {}
page_id = info_dic["id"]
title = info_dic["title"]
desc = info_dic.get("desc","")
time_ts = datetime.datetime.strptime(info_dic["time"],'%Y-%m-%d %H:%M').timestamp()
if info_dic["type"] != "normal":
continue
page_data = self.get_one_page(page_id,proxies=proxies_num,cookies=cookie_dic)
print(page_data)
title = title
anwser = desc
try:
pid = page_data["NoteView"]["content"]["id"]
except:
print("get pid error")
continue
if rds.hexists("xiaohongshu",pid):
continue
rds.hset("xiaohongshu",key=pid,value=json.dumps(page_data))
# video_dic["platform"] = "9"
# video_dic["platform_id"] = pid
# video_dic["title"] = title
# # video_dic["content"] = anwser
# video_dic["user_id"] = random.choice(user_id_list)
# video_dic["create_time"] = time_ts
# rpc_res = post_data(video_dic,"cims/question/batch_create")
print(res)
# break
if __name__ == '__main__':
test = Crawler_xiaohongshu()
releaserurl = 'https://www.xiaohongshu.com/user/profile/5abbb57211be1027a0c880ae'
url_list =[
"https://www.xiaohongshu.com/user/profile/5abbb57211be1027a0c880ae",
"https://www.xiaohongshu.com/user/profile/5ea6909900000000010057a3",
"https://www.xiaohongshu.com/user/profile/5a03b1f4b1da1412dd070a86",
"https://www.xiaohongshu.com/user/profile/5b6e76419276ee0001bd5740",
"https://www.xiaohongshu.com/user/profile/5c4140500000000006006cb7",
"https://www.xiaohongshu.com/user/profile/5bd2beff7da0890001b5408a",
"https://www.xiaohongshu.com/user/profile/5b5edc5211be1044bcce7824",
"https://www.xiaohongshu.com/user/profile/5b35cce84eacab52fbe15c0b",
"https://www.xiaohongshu.com/user/profile/5efec35c000000000101d75a",
"https://www.xiaohongshu.com/user/profile/5f91428a000000000101d909",
"https://www.xiaohongshu.com/user/profile/5ed49f1200000000010017f0",
"https://www.xiaohongshu.com/user/profile/5ae3f47b11be105fae4b854c",
"https://www.xiaohongshu.com/user/profile/5a9e10fb11be1006adc5b9d5",
"https://www.xiaohongshu.com/user/profile/5d0c3b900000000012013409",
"https://www.xiaohongshu.com/user/profile/5f1013a70000000001005b16",
"https://www.xiaohongshu.com/user/profile/5f5c6d860000000001001787",
"https://www.xiaohongshu.com/user/profile/5eeb18e600000000010062b6",
"https://www.xiaohongshu.com/user/profile/5bab62e9ee80fc0001505980",
"https://www.xiaohongshu.com/user/profile/5f262a610000000001004ea9",
"https://www.xiaohongshu.com/user/profile/5eb6779300000000010045f5",
"https://www.xiaohongshu.com/user/profile/5c855374000000001202ef0c",
"https://www.xiaohongshu.com/user/profile/5ecb6d7300000000010016a4",
"https://www.xiaohongshu.com/user/profile/5f100b2d000000000100138d",
"https://www.xiaohongshu.com/user/profile/5c14ae400000000006016f5d",
"https://www.xiaohongshu.com/user/profile/5bbd28de4c26220001881cbd",
"https://www.xiaohongshu.com/user/profile/5f86b6fc000000000100a5d8",
"https://www.xiaohongshu.com/user/profile/5db16ca20000000001004c02",
"https://www.xiaohongshu.com/user/profile/5ad553bb4eacab34ee9f7d4a",
"https://www.xiaohongshu.com/user/profile/5f12cffd000000000101da61",
"https://www.xiaohongshu.com/user/profile/596d7e4f5e87e722ff1bfd32",
"https://www.xiaohongshu.com/user/profile/5ef17ad00000000001005e1c",
"https://www.xiaohongshu.com/user/profile/5f75a5700000000001007679",
"https://www.xiaohongshu.com/user/profile/5c639f59000000001000c731",
"https://www.xiaohongshu.com/user/profile/5f865cbd0000000001002f01",
"https://www.xiaohongshu.com/user/profile/5eccc58f000000000100753e",
"https://www.xiaohongshu.com/user/profile/5fbe05b4000000000101c88d",
"https://www.xiaohongshu.com/user/profile/5b7d1da7e8ac2b471ee6fef3",
"https://www.xiaohongshu.com/user/profile/5a11b22211be101018ba7125",
"https://www.xiaohongshu.com/user/profile/5a76c3c611be107f08bd35b3",
"https://www.xiaohongshu.com/user/profile/5ecb6d7300000000010016a4",
"https://www.xiaohongshu.com/user/profile/5f2539e80000000001009d9e",
"https://www.xiaohongshu.com/user/profile/561b1fd8e4b1cf0295755d05",
"https://www.xiaohongshu.com/user/profile/5beeba1ff7e8b93bc0405234",
"https://www.xiaohongshu.com/user/profile/5c87785f000000001000ed51",
"https://www.xiaohongshu.com/user/profile/5efdba65000000000101c79c",
"https://www.xiaohongshu.com/user/profile/5507e7dfa46e9616260827f6",
"https://www.xiaohongshu.com/user/profile/567573470bf90c27957dd73c",
"https://www.xiaohongshu.com/user/profile/5fd1821b000000000100381a",
"https://www.xiaohongshu.com/user/profile/5f5f6b1c00000000010064dc",
"https://www.xiaohongshu.com/user/profile/5aea4d31e8ac2b4a44e1d2d4",
"https://www.xiaohongshu.com/user/profile/5f39eabb00000000010076ca",
"https://www.xiaohongshu.com/user/profile/5cda11d7000000001703780c",
"https://www.xiaohongshu.com/user/profile/5cbc3e9f000000001701d7bf",
"https://www.xiaohongshu.com/user/profile/5e7886930000000001003f7f",
"https://www.xiaohongshu.com/user/profile/566fbc3550c4b435f51f637b",
"https://www.xiaohongshu.com/user/profile/5e86cb34000000000100a223",
"https://www.xiaohongshu.com/user/profile/558e15b2f5a263490c65cdaa",
"https://www.xiaohongshu.com/user/profile/5d9eef320000000001001615",
"https://www.xiaohongshu.com/user/profile/5a6ba3214eacab4eee8e627a",
"https://www.xiaohongshu.com/user/profile/5f58cacb000000000100bdf5",
"https://www.xiaohongshu.com/user/profile/5f954030000000000100780c",
"https://www.xiaohongshu.com/user/profile/5f5745bf000000000100351d",
"https://www.xiaohongshu.com/user/profile/5c74a2b9000000001002e667",
"https://www.xiaohongshu.com/user/profile/595ee5b882ec397553103dd3",
"https://www.xiaohongshu.com/user/profile/5a5e20324eacab30f03654fb",
"https://www.xiaohongshu.com/user/profile/55743bedc2bdeb1a16844741",
"https://www.xiaohongshu.com/user/profile/5f0d523800000000010056de",
"https://www.xiaohongshu.com/user/profile/59d5b03e44363b61a050532f",
"https://www.xiaohongshu.com/user/profile/5ebdd5f40000000001002a67",
"https://www.xiaohongshu.com/user/profile/5f1c1b7b0000000001006cbf",
"https://www.xiaohongshu.com/user/profile/5ae404944eacab794dfb95b1",
"https://www.xiaohongshu.com/user/profile/5d26276a0000000012017538",
"https://www.xiaohongshu.com/user/profile/5ed5aa8f0000000001001f1e",
"https://www.xiaohongshu.com/user/profile/5f92cf4f000000000100a846",
"https://www.xiaohongshu.com/user/profile/5a75d42011be10344b917ffe",
"https://www.xiaohongshu.com/user/profile/5ccea0ff000000001002b753",
"https://www.xiaohongshu.com/user/profile/5c4418750000000005006717",
"https://www.xiaohongshu.com/user/profile/5ec582d60000000001005315",
"https://www.xiaohongshu.com/user/profile/594a93835e87e72f3e2ded11",
"https://www.xiaohongshu.com/user/profile/5b8ab07606311b000184195a",
"https://www.xiaohongshu.com/user/profile/54e7413ea46e96122dab7674",
"https://www.xiaohongshu.com/user/profile/5f3657900000000001002181",
"https://www.xiaohongshu.com/user/profile/5a65d6554eacab6864e2749e",
"https://www.xiaohongshu.com/user/profile/5a745dc911be101d9ceab748",
"https://www.xiaohongshu.com/user/profile/59b2033550c4b45e5d43c3d9",
"https://www.xiaohongshu.com/user/profile/59a97aaa5e87e760e012dcd0",
"https://www.xiaohongshu.com/user/profile/5a5de03611be100219719b0f",
"https://www.xiaohongshu.com/user/profile/5f40a5170000000001008577",
"https://www.xiaohongshu.com/user/profile/597e82aa5e87e73c4915db81",
"https://www.xiaohongshu.com/user/profile/580e0bc36a6a69043935369d",
"https://www.xiaohongshu.com/user/profile/5d1a17670000000012021d8e",
"https://www.xiaohongshu.com/user/profile/59a830be82ec39155146f421",
"https://www.xiaohongshu.com/user/profile/55efc1b73397db0e969c8fbd",
"https://www.xiaohongshu.com/user/profile/5c8c55220000000010005810",
"https://www.xiaohongshu.com/user/profile/5f337df2000000000101e2b2",
"https://www.xiaohongshu.com/user/profile/5f2111500000000001009b7b",
"https://www.xiaohongshu.com/user/profile/59c840ff44363b497f335cd4",
"https://www.xiaohongshu.com/user/profile/5f8e8508000000000101d70e",
"https://www.xiaohongshu.com/user/profile/5a163e3511be10234e1abffd",
"https://www.xiaohongshu.com/user/profile/5e71f6870000000001005e52",
'https://www.xiaohongshu.com/user/profile/5cca9b3700000000120314c9',
'https://www.xiaohongshu.com/user/profile/5aa0f7bae8ac2b65bfcdaf0e',
'https://www.xiaohongshu.com/user/profile/5c20dd200000000007027c07',
'https://www.xiaohongshu.com/user/profile/5fe1c1ba0000000001006e65',
]
for url in url_list:
print(url)
res = test.releaser_page(url,proxies_num=0)