def parse(self, response):
try:
self.task.crawl_count += 1
j = json.loads(response.body)
name = j['data']['card']['name']
mid = j['data']['card']['mid']
# 刷新redis数据缓存
self.redis_connection.delete("author_detail::{}".format(mid))
sex = j['data']['card']['sex']
face = j['data']['card']['face']
fans = j['data']['card']['fans']
attention = j['data']['card']['attention']
level = j['data']['card']['level_info']['current_level']
official = j['data']['card']['Official']['title']
archive = j['data']['archive_count']
article = j['data']['article_count']
face = j['data']['card']['face']
item = AuthorItem()
item['mid'] = int(mid)
item['name'] = name
item['face'] = face
item['official'] = official
item['sex'] = sex
item['level'] = int(level)
item['data'] = {
'fans': int(fans),
'attention': int(attention),
'archive': int(archive),
'article': int(article),
'datetime': datetime.datetime.now()
}
item['c_fans'] = int(fans)
item['c_attention'] = int(attention)
item['c_archive'] = int(archive)
item['c_article'] = int(article)
url_list = response.url.split('&')
if len(url_list) == 2:
item['object_id'] = url_list[1]
else:
item['object_id'] = None
yield Request(
"https://api.bilibili.com/x/space/upstat?mid={mid}".format(
mid=str(mid)),
meta={'item': item},
method='GET',
callback=self.parse_view)
except Exception as error:
# 出现错误时打印错误日志
self.task.crawl_failed += 1
mailer.send(
to=["*****@*****.**"],
subject="BiliobSpiderError",
body="{}\n{}\n{}".format(item, response.url, error),
)
logging.error("视频爬虫在解析时发生错误")
logging.error(response.url)
logging.error(error)
def detailParse(self, response):
try:
j = json.loads(response.body)
name = j['data']['card']['name']
mid = j['data']['card']['mid']
sex = j['data']['card']['sex']
face = j['data']['card']['face']
fans = j['data']['card']['fans']
attention = j['data']['card']['attention']
level = j['data']['card']['level_info']['current_level']
official = j['data']['card']['Official']['title']
archive = j['data']['archive_count']
article = j['data']['article_count']
face = j['data']['card']['face']
item = AuthorItem()
# 粉丝数大于1000才加入
if int(fans) > 1000:
item['c_fans'] = int(fans)
item['c_attention'] = int(attention)
item['c_archive'] = int(archive)
item['c_article'] = int(article)
item['mid'] = int(mid)
item['name'] = name
item['face'] = face
item['official'] = official
item['sex'] = sex
item['focus'] = True
item['level'] = int(level)
item['data'] = {
'fans':
int(fans),
'attention':
int(attention),
'archive':
int(archive),
'article':
int(article),
'datetime':
datetime.datetime.utcnow() + datetime.timedelta(hours=8)
}
yield item
except Exception as error:
# 出现错误时打印错误日志
self.task.crawl_failed += 1
logging.error("视频爬虫在解析时发生错误")
logging.error(response.url)
logging.error(error)
def parse(self, response):
try:
j = json.loads(response.body)
name = j['data']['card']['name']
mid = j['data']['card']['mid']
sex = j['data']['card']['sex']
face = j['data']['card']['face']
fans = j['data']['card']['fans']
attention = j['data']['card']['attention']
level = j['data']['card']['level_info']['current_level']
official = j['data']['card']['Official']['title']
archive = j['data']['archive_count']
article = j['data']['article_count']
face = j['data']['card']['face']
item = AuthorItem()
item['mid'] = int(mid)
item['name'] = name
item['face'] = face
item['official'] = official
item['sex'] = sex
item['level'] = int(level)
item['data'] = {
'fans': int(fans),
'attention': int(attention),
'archive': int(archive),
'article': int(article),
'datetime': datetime.datetime.now()
}
item['c_fans'] = int(fans)
item['c_attention'] = int(attention)
item['c_archive'] = int(archive)
item['c_article'] = int(article)
yield Request(
"https://api.bilibili.com/x/space/upstat?mid={mid}".format(
mid=str(mid)),
meta={'item': item},
method='GET',
callback=self.parse_view)
except Exception as error:
# 出现错误时打印错误日志
mailer.send(
to=["*****@*****.**"],
subject="BiliobSpiderError",
body="{}\n{}\n{}".format(item, response.url, error),
)
logging.error("视频爬虫在解析时发生错误")
logging.error(response.url)
logging.error(error)
def parse(self, response):
try:
j = json.loads(response.body)
name = j['data']['card']['name']
mid = j['data']['card']['mid']
# 刷新redis数据缓存
self.redis_connection.delete("author_detail::{}".format(mid))
sex = j['data']['card']['sex']
face = j['data']['card']['face']
fans = j['data']['card']['fans']
attention = j['data']['card']['attention']
level = j['data']['card']['level_info']['current_level']
official = j['data']['card']['Official']['title']
archive = j['data']['archive_count']
article = j['data']['article_count']
face = j['data']['card']['face']
item = AuthorItem()
item['mid'] = int(mid)
item['name'] = name
item['face'] = face
item['official'] = official
item['sex'] = sex
item['level'] = int(level)
item['data'] = {
'fans': int(fans),
'attention': int(attention),
'archive': int(archive),
'article': int(article),
'datetime': datetime.datetime.now()
}
item['c_fans'] = int(fans)
item['c_attention'] = int(attention)
item['c_archive'] = int(archive)
item['c_article'] = int(article)
yield Request(
"https://api.bilibili.com/x/space/upstat?mid={mid}".format(
mid=str(mid)),
meta={'item': item},
method='GET',
callback=self.parse_view)
except Exception as error:
# 出现错误时打印错误日志
logging.error("视频爬虫在解析时发生错误")
logging.error(response.url)
logging.error(error)
def detailParse(self, response):
try:
j = json.loads(response.body)
name = j['data']['card']['name']
mid = j['data']['card']['mid']
sex = j['data']['card']['sex']
face = j['data']['card']['face']
fans = j['data']['card']['fans']
attention = j['data']['card']['attention']
level = j['data']['card']['level_info']['current_level']
official = j['data']['card']['Official']['title']
archive = j['data']['archive_count']
article = j['data']['article_count']
face = j['data']['card']['face']
item = AuthorItem()
# 粉丝数大于1000才加入
if int(fans) > 1000:
item['c_fans'] = int(fans)
item['c_attention'] = int(attention)
item['c_archive'] = int(archive)
item['c_article'] = int(article)
item['mid'] = int(mid)
item['name'] = name
item['face'] = face
item['official'] = official
item['sex'] = sex
item['focus'] = True
item['level'] = int(level)
item['data'] = {
'fans': int(fans),
'attention': int(attention),
'archive': int(archive),
'article': int(article),
'datetime': datetime.datetime.now()
}
yield item
except Exception as error:
# 出现错误时打印错误日志
self.task.crawl_failed += 1
mailer.send(
to=["*****@*****.**"],
subject="BiliobSpiderError",
body="{}\n{}".format(response.url, error),
)
logging.error("视频爬虫在解析时发生错误")
logging.error(response.url)
logging.error(error)
