class ZhiDaoGenerator:
""" 百度知道关键词抓取，关键词有两部分，1.问题内容包含的关键词，2.问题所属的关键词 """
def __init__(self):
self.logger = LogHandler('zhidao_crawl')
try:
# 知道请求头，获取Cookie的BAIDUID，否则抓取不到数据
self.zhidao_headers = {
'User-Agent':
'Mozilla/5.0 (Windows NT 6.1; Win64; x64) AppleWebKit/537.36 16 (KHTML, like Gecko) Chrome/67.0.3396.99 Safari/537.36',
'Cookie':
'BAIDUID=%s' % requests.get('https://zhidao.baidu.com/browse/',
timeout=3).cookies['BAIDUID']
}
except Exception as e:
raise e
def __zd_question_keywords(self, url):
""" 获取知道问题内容的关键词 """
try:
req = requests.get(url, timeout=3)
req.encoding = req.apparent_encoding
kws = []
for kw_tag in BeautifulSoup(req.text, "html.parser").find_all(
'li',
class_=lambda class_: class_ and ('word grid' in class_)):
kw = kw_tag.find(class_="word-text")
if kw is not None:
kws.append(kw.string)
return kws
except RequestException as re:
self.logger.warning(re)
return []
except Exception as e:
raise e
@timethis
def crawl_zhidao_words(self, save_keyword):
""" 百度知道抓取 """
try:
req = requests.get(
url=
'https://zhidao.baidu.com/list?_pjax=%23j-question-list-pjax-container',
headers=self.zhidao_headers,
timeout=3)
req.encoding = req.apparent_encoding
for qs in BeautifulSoup(req.text, "html.parser").find_all(
'div', class_='question-title-section'):
# 问题所属领域关键词提取
for qt in map(lambda x: x.string.replace('\n', ''),
qs.find_all('a', class_='tag-item')):
save_keyword(qt)
# 问题内容包含关键词提取
for qm in self.__zd_question_keywords(qs.a.get('href')):
save_keyword(str(qm))
except RequestException as re:
self.logger.warn(re)
except Exception as e:
raise e
class ProxyManager(object):
"""
ProxyManager
"""
def __init__(self):
self.db = ProxyDBClient('proxy')
self.raw_proxy_queue = 'raw_proxy'
self.log = LogHandler('proxy_manager')
self.useful_proxy_queue = 'useful_proxy'
def refresh(self):
"""
fetch proxy into Db by ProxyGetter/getFreeProxy.py
:return:
"""
self.db.changeTable(self.raw_proxy_queue)
for proxyGetter in conf.proxy_getter_functions:
# fetch
try:
self.log.info(
"{func}: fetch proxy start".format(func=proxyGetter))
for proxy in getattr(GetFreeProxy, proxyGetter.strip())():
# 直接存储代理, 不用在代码中排重, hash 结构本身具有排重功能
proxy = proxy.strip()
if proxy and verifyProxyFormat(proxy):
self.log.info('{func}: fetch proxy {proxy}'.format(
func=proxyGetter, proxy=proxy))
self.db.put(proxy)
else:
self.log.error(
'{func}: fetch proxy {proxy} error'.format(
func=proxyGetter, proxy=proxy))
except Exception as e:
self.log.warn(
"{func}: fetch proxy fail".format(func=proxyGetter))
continue
def get(self):
"""
return a useful proxy
:return:
"""
self.db.changeTable(self.useful_proxy_queue)
item_dict = self.db.getAll()
if item_dict:
if EnvUtil.PY3:
return random.choice(list(item_dict.keys()))
else:
return random.choice(item_dict.keys())
return None
# return self.db.pop()
def delete(self, proxy):
"""
delete proxy from pool
:param proxy:
:return:
"""
self.db.changeTable(self.useful_proxy_queue)
self.db.delete(proxy)
def getAll(self):
"""
get all proxy from pool as list
:return:
"""
self.db.changeTable(self.useful_proxy_queue)
item_dict = self.db.getAll()
if EnvUtil.PY3:
return list(item_dict.keys()) if item_dict else list()
return item_dict.keys() if item_dict else list()
def getNumber(self):
self.db.changeTable(self.raw_proxy_queue)
total_raw_proxy = self.db.getNumber()
self.db.changeTable(self.useful_proxy_queue)
total_useful_queue = self.db.getNumber()
return {
'raw_proxy': total_raw_proxy,
'useful_proxy': total_useful_queue
}
class QuestionsGenerator:
""" 获取话题内的精华和讨论问题"""
top_activity_url = "https://www.zhihu.com/api/v4/topics/%u/feeds/top_activity?limit=10&offset=%u" # 讨论问题
essence_url = "https://www.zhihu.com/api/v4/topics/%u/feeds/essence?limit=10&offset=%u" # 精华问题
def __init__(self):
self.p_receiver = ProxiesReceiver()
self.essence_dict = dict()
self.activity_dict = dict()
self.session = requests.Session()
# 先从话题id中筛选出未获取问题的id放入待获取队列
QuestionsGenerator.save_new_topic_id()
self.logger = LogHandler("question_generator")
@staticmethod
def badge_parse(badge):
res_array = []
for b in badge:
res_array.append({
'author_type': b['type'],
'author_description': b['description']
})
return res_array
def topic_sticky_parse(self, topic_sticky):
res_array = []
try:
for t in topic_sticky:
if t['target'].get('title'):
res_array.append({
'id':
t['target']['id'],
'title':
t['target']['title'],
'author_name':
t['target']['author']['name'],
'author_gender':
t['target']['author']['gender'],
'author_headline':
t['target']['author']['headline'],
'author_url_token':
t['target']['author']['url_token'],
'excerpt':
t['target']['excerpt'],
'updated_time':
t['target']['updated'],
'contents':
t['target']['comment_count'],
'likes':
t['target']['voteup_count'],
'created_time':
t['target']['created']
})
else:
res_array.append({
'id':
t['target']['id'], # 热门回答
'title':
t['target']['question']['title'],
'author_name':
t['target']['author']['name'],
'author_gender':
t['target']['author']['gender'],
'author_headline':
t['target']['author']['headline'],
'author_url_token':
t['target']['author']['url_token'],
'excerpt':
t['target']['excerpt'],
'updated_time':
t['target']['updated_time'],
'created':
t['target']['question']['created'],
'created_time':
t['target']['created_time']
})
except KeyError as ke:
self.logger.error(ke, topic_sticky)
except Exception as e:
raise e
return res_array
@timethis
def crawl_topic_message(self, tid, q_url, num, message_dict):
"""
话题对应问题/文章获取,最终结果存放:
message_dict: key-类型（文章/问题），value-{key-id, value-{'question/title':问题/文章名称, author':作者,
'gender':性别,'author_badge':作者标签, 'author_headline':作者签名, 'author_url_token':作者url标识，excerpt':摘录，
'created_time':创建时间, 'updated_time':最后更新时间, 'comment_count':'评论数', likes':点赞数}}
:param tid:话题ID
:param q_url:请求url
:param num 抓取数量
:param message_dict:存储字典
:return:
"""
session = requests.session()
sticky_num = 0
for offset in range(0, num, 10):
url = q_url % (tid, offset)
try:
ques = session.get(url=url,
headers=headers,
proxies=self.p_receiver.one_random,
timeout=3)
except Exception as re:
self.logger.warn((re, url))
continue
try:
q_json = ques.json() if ques else {}
except JSONDecodeError as je:
self.logger.error((je, url, ques))
continue
for q in q_json.get('data', []):
target = q.get('target', {})
question_type = str(target.get('type', 'none_type')).lower()
if question_type == 'none_type':
continue
elif question_type == 'answer':
# 问题回答是双id,使用元组转成字符串
addtwodimdict(
message_dict, 'answer',
str((target['question']['id'], target['id'])), {
'question':
target['question']['title'],
'author_name':
target['author']['name'],
'author_gender':
target['author']['gender'],
'author_badge':
QuestionsGenerator.badge_parse(
target['author']['badge']),
'author_headline':
target['author']['headline'],
'author_url_token':
target['author']['url_token'],
'excerpt':
target['excerpt'],
'created_time':
target['created_time'],
'updated_time':
target['updated_time'],
'contents':
target['comment_count'],
'likes':
target['voteup_count']
})
elif question_type == 'article':
# 文章是单id
addtwodimdict(
message_dict, 'article', target['id'], {
'title':
target['title'],
'author_name':
target['author']['name'],
'author_gender':
target['author']['gender'],
'author_badge':
QuestionsGenerator.badge_parse(
target['author']['badge']),
'author_headline':
target['author']['headline'],
'author_url_token':
target['author']['url_token'],
'excerpt':
target['excerpt'],
'created_time':
target['created'],
'updated_time':
target['updated'],
'contents':
target['comment_count'],
'likes':
target['voteup_count']
})
elif question_type == 'question':
pass # 目前抓取到的是一些没有回答的问题，这里过滤掉
elif question_type == 'topic_sticky_module': # 热门置顶
addtwodimdict(
message_dict, 'topic_sticky_module', sticky_num, {
'title': target['title'],
'data': self.topic_sticky_parse(target['data'])
})
sticky_num += 1
else:
self.logger.error(
"There was a new type:{}!\n".format(question_type))
if str(q_json.get('paging', {}).get('is_end',
'none')).lower() == 'true':
return
def process(self):
"""
zhTopicQuestions内包含精华(点赞比较多)和讨论(最新比较热)的问题和文章
对应结构essence_dict和activity_dict
:return:
"""
tid = int(redis_cli.block_pop('zhNewTopicID'))
self.crawl_topic_message(tid, QuestionsGenerator.essence_url,
conf.essence_nums, self.essence_dict)
self.crawl_topic_message(tid, QuestionsGenerator.top_activity_url,
conf.top_activity_nums, self.activity_dict)
with open(os.path.join(OUTPUT_PATH, "{}.json".format(int(tid))),
"w",
encoding="GB18030") as f:
json.dump(
{
'essence': self.essence_dict,
'top_activity': self.activity_dict
},
f,
indent=4,
ensure_ascii=False)
@staticmethod
def save_new_topic_id():
""" output目录下不要手动放其他名称文件 """
for nid in set(redis_cli.hkeys('zhTopicMessage')) - set(
{x.split('.')[0]
for x in os.listdir(OUTPUT_PATH)}):
redis_cli.sadd('zhNewTopicID', nid)
class ZhihuTopicGenerator:
""" 分为两个过程:id获取和扩展 """
def __init__(self):
# 代理ip
self.p_receiver = ProxiesReceiver()
# 建立会话，设置requests重连次数和重连等待时间
self.session = requests.Session()
retry = Retry(connect=3, backoff_factor=0.5)
adapter = HTTPAdapter(max_retries=retry)
self.session.mount('https://', adapter)
self.logger = LogHandler('topics_generator')
logging.getLogger("urllib3").setLevel(logging.ERROR)
def __get_topic_message(self, tid):
"""
话题信息获取,最终结果存放:
zhTopicMessage哈希表内，key-话题id，value-{'name':名称, 'introduction': 简介, 'questions_count':问题数,
'best_answers_count':精华问题数, 'followers_count':关注人数, 'best_answerers_count':优秀回答者人数}
"""
try:
j_rst = self.session.get(url=topic_message_url % tid,
headers=headers,
proxies=self.p_receiver.one_random,
timeout=3).json()
if redis_cli.hset(
'zhTopicMessage', tid,
str({
"name":
j_rst.get("name"),
'introduction':
j_rst.get("introduction"),
"questions_count":
j_rst.get("questions_count"),
"best_answers_count":
j_rst.get("best_answers_count"),
'followers_count':
j_rst.get("followers_count"),
"best_answerers_count":
j_rst.get("best_answerers_count")
})):
# 待获取相关信息
redis_cli.sadd('zhNewTopicID', tid)
self.logger.info("zhNewTopicID:%d", tid)
return True
except RequestException as re:
self.logger.warn(re)
except Exception as e:
raise e
return False
def __get_hot_topics(self):
""" 搜索zhTemporaryWords内关键词，从其结果中得到相关话题id和名称 """
tw = redis_cli.block_pop('zhTemporaryWords').decode('utf-8') # pop
# 不断翻页至最后,最大获取1000条
for offset in range(0, 1000, 10):
try:
url = zh_search_url % (tw, offset)
j_topics = self.session.get(url=url,
headers=headers,
proxies=self.p_receiver.one_random,
timeout=3).json()
topics = j_topics.get('data', None) if j_topics else None
if not topics: # 已到最后
return
# 每一页获取话题相关详细信息
for t in topics:
if t.get('object') and t.get('object').get('id'):
try:
tid = int(t['object']['id'])
except ValueError as ve:
self.logger.warning(ve, t['object']['id'])
continue
if self.__get_topic_message(tid):
yield tid
else:
break
except RequestException as re:
self.logger.warn((re, url))
except ReadTimeout as rte:
self.logger.warn((rte, url))
except KeyError as ke:
self.logger.warn((ke, url))
except Exception as e:
raise e
@staticmethod
def __save_to_dag(child_topic_id, parent_topic_id):
""" 按其结构保存为有向无环图 """
ids = redis_cli.hget('zhTopicDAG', parent_topic_id)
if not ids or ids.decode() == "None":
redis_cli.hset('zhTopicDAG', parent_topic_id,
str({child_topic_id}))
else:
new_ids = eval(ids)
new_ids.add(child_topic_id)
redis_cli.hset('zhTopicDAG', parent_topic_id, str(new_ids))
def __add_topics(self, url, topic_id, func):
try:
req = self.session.get(url=url % int(topic_id),
headers=headers,
proxies=self.p_receiver.one_random,
timeout=3)
if not req: # 获取子父话题有可能不存在
return
for p in req.json()['data']:
expand_topic_id = int(p['id'])
func(topic_id, expand_topic_id)
self.__get_topic_message(expand_topic_id)
except RequestException as re:
self.logger.warn(re)
except ReadTimeout as rte:
self.logger.warn(rte)
except Exception as e:
raise e
def __expand_topics(self, tid):
""" 话题扩展，分别向父子话题不断扩展 """
self.__add_topics(parent_url, tid,
lambda a, b: self.__save_to_dag(a, b))
self.__add_topics(child_url, tid,
lambda a, b: self.__save_to_dag(b, a))
@timethis
def process(self):
for tid in self.__get_hot_topics():
self.__expand_topics(tid)