首先所需要的环境:(我用的是Python2的,可以选择python3,具体遇到的问题自行解决,目前我这边几百万的数据量爬取)
环境:
Python 2.7.10
Scrapy Scrapy 1.5.0
第三方库:
PyMySQL==0.8.0
Scrapy==1.5.0
pytesseract==0.2.0
pip==10.0.1
Pillow==5.1.0
logger==1.4
bs4==0.0.1
requests==2.18.4
创建项目
scrapy startproject mytest
创建爬虫程序
cd mytest
scrapy genspider name XXX.com
直接贴代码具体需要注意的特殊颜色标出有注释
import scrapy
import pytesseract #验证码识别库
from PIL import Image #验证码图片处理
from scrapy.http import Request
from yishi.items import YishiItem #items定义爬取字段
from yishi.settings import MYSQL_HOST, MYSQL_DBNAME, MYSQL_USER, MYSQL_PASSWD #settings数据库配置
import pymysql #连接数据库
import logging #打印日志
#设置日志
log_filename = '../static/data/info.log'
logging.basicConfig(filename=log_filename, filemode='a', level=logging.INFO)
class CreditSpider(scrapy.Spider):
name = 'name'
baseURL = 'https://xxx.com'
#start_urls = ''
#设置headers,打开网页直接看请求headers复制进去就可以了
headers = {
'Accept': 'text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,image/apng,*/*;q=0.8',
'Accept-Encoding': 'gzip, deflate, br',
'Connection': 'keep-alive',
'Host': 'xxx',
'Upgrade-Insecure-Requests': 1,
'User-Agent': 'xxx',
}
#数据库
connect = pymysql.connect(
host=MYSQL_HOST,
db=MYSQL_DBNAME,
user=MYSQL_USER,
passwd=MYSQL_PASSWD,
charset='utf8',
use_unicode=True)
#重写start_requests
def start_requests(self):
return [Request(self.baseURL+'xxx',
headers=self.headers,
callback=self.parse,
dont_filter=True, #scrapy会对request的URL去重(RFPDupeFilter),加上dont_filter则告诉它这个URL不参与去重
)
]
#首先需要请求一次网页
def parse(self, response):
#每次查询1条数据,搜索列表所需要的条件
cursor = self.connect.cursor()
sql = 'select id,xxx,xxx,xxx from xxx where xxx order by id limit 1'
cursor.execute(sql)
res = cursor.fetchall()
if res:
#请求网站所需要的参数,搜索条件
data = {
"xxx": res[0][1],
"xxx": '',
"xxx": '',
"xxx": res[0][2],
"xxx": '',
"xxx": '',
"xxx": '',
}
cursor.close()
return scrapy.Request(self.baseURL + '/xxx/captcha', #验证码图片地址
headers=self.headers,
meta={'data': data, 'dr_id': res[0][0], 'static': res[0][3], 'len': len(res)}, #第一次请求的参数传给下次请求,可以保存cookie之类的
callback=self.creditRes,
dont_filter=True
)
else:
#数据表中条件没有的时候结束爬虫,每次爬取要更新下条件表
print '执行完毕!'
pass
#再次请求存验证码图片
def creditRes(self, response):
#保存验证码
captchaFile = '../static/images/code/captcha.png'
with open(captchaFile, 'wb') as f:
f.write(response.body)
try:
#pytesseract识别验证码
image = Image.open(captchaFile)
captcha_value = pytesseract.image_to_string(image)
print '验证码为:'+captcha_value
except:
#验证码失败 重新请求
logging.info('验证码获取失败')
return self.start_urls
#识别后的验证码作为参数使用
data = response.meta.get("data")
data["validCode"] = captcha_value
return [scrapy.FormRequest(
url=self.baseURL+'xxx', #带上全部参数再次请求取数据
formdata=data,
method='GET',
meta={'dr_id': response.meta.get("dr_id"), 'static': response.meta.get("static"), 'len': response.meta.get("len"),
'captcha_value': captcha_value}, #带上部分参数保存或更新状态用
headers=self.headers,
callback=self.creditdata,
dont_filter=True,
)]
def creditdata(self, response):
#获取验证码错误内容,识别验证是否成功
code_data = response.xpath("//span[@class='error']")
if code_data:
code = code_data.xpath(".//text()").extract()[0].decode('UTF-8')
logging.info('验证码校验失败,验证码:'+str(response.meta.get("captcha_value")))
else:
code = ''
#验证码错误时不更新状态,继续重复爬取
dr_id = response.meta.get("dr_id")
#不存在验证码识别更新状态,插入数据
if code.strip() not in ('验证码错误', '验证码不能为空'):
cursor = self.connect.cursor()
sql = 'update xxx set status=%s where id=%s' % (1, dr_id)
cursor.execute(sql)
self.connect.commit()
cursor.close()
else:
#验证码失败不更新状态
logging.info('验证码错误')
node_list = response.xpath("//table[@id='formresult']/tbody/tr")
# 更新状态 0还未抓取数据 1已经抓取
logging.info('当前执行条件表id为'+ str(dr_id))
if node_list:
for node in node_list:
item = YishiItem()
item['xxx'] = dr_id
item['xxx'] = node.xpath(".//td[1]/text()").extract()[0].decode('UTF-8')
item['xxx'] = node.xpath(".//td[2]/text()").extract()[0].decode('UTF-8')
item['xxx'] = node.xpath(".//td[3]/text()").extract()[0].decode('UTF-8')
item['xxx'] = node.xpath(".//td[4]/text()").extract()[0].decode('UTF-8')
item['xxx'] = node.xpath(".//td[5]/text()").extract()[0].decode('UTF-8')
item['xxx'] = node.xpath(".//td[6]/text()").extract()[0].decode('UTF-8')
item['xxx'] = node.xpath(".//td[7]/text()").extract()[0].decode('UTF-8')
yield item
#分页数据,根据下一页爬取,可获取下页按钮状态去爬取分页数据
nextPage = response.xpath("//a[@class='disable' and @class='next']")
if nextPage:
if not len(nextPage):
#下一页a标签url
url = response.xpath("//a[@class='disable' and @class='next']/@href").extract()[0]
yield scrapy.Request(self.baseURL+'/'+url, callback=self.creditdata)
# 根据状态status=0判断是否继续爬取数据
len = response.meta.get("len")
if not len == 0:
yield scrapy.Request(self.baseURL+'xxx',
headers=self.headers,
callback=self.parse,
dont_filter=True)
items设置:
xxx = scrapy.Field()
xxx = scrapy.Field()
...
pipelines存数据库这个就不说了根据自己的业务
注:目前我网站验证码比较简单可以直接使用pytesseract,识别率95%以上,也可以用别的方式CNN等方式识别
个人感觉用 requests.get() 方式写要简单一些,本地已测试过,根据业务需求用scrapy完成的。
requests.get() 主要问题就是 session = requests.session() 这句是重点