def __init__(self, name=None, **kwargs):
super(WXSourceSpider, self).__init__(name=None, **kwargs)
self.count = 0
self.wxSourceDao = WxSourceDao()
self.currIp = ''
self.logDao = LogDao(self.logger, 'weixin_source_catch')
self.dataMonitor = DataMonitorDao()
def __init__(self, name=None, **kwargs):
super(SinaSpider, self).__init__(name=None, **kwargs)
self.count = 0
self.logDao = LogDao(self.logger, 'sina_detail')
self.checkDao = CheckDao()
# 用于缓存css
self.css = {'hash': 'style'}
self.dataMonitor = DataMonitorDao()
def __init__(self, name=None, **kwargs):
super(WXDetailSpider, self).__init__(name=None, **kwargs)
self.count = 0
self.wxSourceDao = WxSourceDao()
self.logDao = LogDao(self.logger, 'weixin_list_detail')
self.checkDao = CheckDao()
self.dataMonitor = DataMonitorDao()
self.wxSources = []
self.brokenAccounts = [] # 当前被禁止了的账号，下次抓取优先抓取
def __init__(self, name=None, **kwargs):
super(WYDetailSpider, self).__init__(name=None, **kwargs)
self.count = 0
self.request_stop = False
self.request_stop_time = 0
self.logDao = LogDao(self.logger, 'wangyi_list_detail')
self.checkDao = CheckDao()
# 用于缓存css
self.css = {'hash': 'style'}
self.dataMonitor = DataMonitorDao()
def __init__(self, name=None, **kwargs):
super(SinaSpider, self).__init__(name=None, **kwargs)
self.count = 0
self.request_stop = False
self.request_stop_time = 0
self.logDao = LogDao(self.logger, 'sina_detail')
class SinaSpider(scrapy.Spider):
name = 'sina'
download_delay = 20 # 基础间隔 0.5*download_delay --- 1.5*download_delays之间的随机数
handle_httpstatus_list = [301, 302, 204, 206, 403, 404,
500] # 可以处理重定向及其他错误码导致的 页面无法获取解析的问题
def __init__(self, name=None, **kwargs):
super(SinaSpider, self).__init__(name=None, **kwargs)
self.count = 0
self.request_stop = False
self.request_stop_time = 0
self.logDao = LogDao(self.logger, 'sina_detail')
def start_requests(self):
# while True:
# 检测网络
if not NetworkUtil.checkNetWork():
# 20s检测一次
TimerUtil.sleep(20)
self.logDao.warn(u'检测网络不可行')
# continue
# 检测服务器
if not NetworkUtil.checkService():
# 20s检测一次
TimerUtil.sleep(20)
self.logDao.warn(u'检测服务器不可行')
# continue
if self.request_stop:
# 拨号生效时间不定，所以需要间隔一段时间再重试
timeSpace = time.time() - self.request_stop_time
if timeSpace / 60 <= 2:
# 当时间间隔小于 2分钟 就不请求
# continue
pass
else:
self.request_stop = False
# 进行爬虫
url = 'http://roll.news.sina.com.cn/interface/rollnews_ch_out_interface.php?col=30&spec=&type=&ch=05&k' \
'=&offset_page=0&offset_num=0&num=60&asc=&page='
for page in range(0, 11):
if self.request_stop:
self.logDao.warn(u'出现被绊或者出现网络异常，退出循环')
# 当网络出现被绊的情况，就需要停止所有的请求等待IP更换
break
r = random.uniform(0, 1)
newUrl = url + str(page)
newUrl += ('&r=' + str(r))
self.logDao.info(u"开始抓取列表：" + newUrl)
yield scrapy.Request(url=newUrl,
meta={'url': newUrl},
callback=self.parseList)
# 跑空线程2秒
TimerUtil.sleep(2)
if self.request_stop:
# 需要发起通知 进行重新拨号
self.logDao.warn(u'发送重新拨号信号，请等待2分钟会尝试重新抓取')
self.request_stop_time = time.time()
pass
else:
# 正常抓好之后，当前跑空线程10分钟，不影响一些还没请求完成的request
self.logDao.info(u'请求了一轮了，但是可能还有没有请求完成，睡一会10分钟')
TimerUtil.sleep(10 * 60)
pass
# TODO。。还没有找到被禁止的情况
def parseList(self, response):
url = response.meta['url']
data = response.body.decode('gbk')
data = data.lstrip('var jsonData = ').rstrip(';')
# 格式化
data = demjson.decode(data) or {}
list = data['list'] or []
self.logDao.info(u"解析列表：" + url)
for item in list:
itemTime = item['time'] or 0
contentItem = ContentItem()
channel = item['channel'] or {}
channel_name = channel['title']
contentItem['channel_name'] = channel_name
contentItem['title'] = item['title']
contentItem['source_url'] = item['url']
# 暂时知道 两种不同的文章界面
if 'http://tech.sina.com.cn/zl/' in item['url']:
callback = self.parseDetail2
else:
callback = self.parseDetail
self.logDao.info(u"开始抓取文章：" + item['url'])
yield scrapy.Request(url=item['url'],
meta={
'contentItem': contentItem,
'source_url': item['url']
},
callback=callback)
def parseDetail2(self, response):
source_url = response.meta['source_url']
contentItem = response.meta['contentItem']
selector = Selector(text=response.body)
pl_main_content = selector.xpath(
'//*[@id="J_Article_Wrap"]').extract_first()
title = selector.xpath(
'//*[@id="artibodyTitle"]/text()').extract_first() or ''
post_date = selector.xpath(
'//*[@id="pub_date"]/text()').extract_first() or ''
post_date = post_date.replace('\r\n', '').strip(' ')
post_user = selector.xpath(
'//*[@id="media_name"]/a[1]/text()').extract_first()
tags = selector.xpath(
'//p[@class="art_keywords"]/a/text()').extract() or []
tags = ','.join(tags)
main = {
'title': title,
'post_date': post_date,
'post_user': post_user,
'pl_main_content': pl_main_content,
'tags': tags
}
m2 = hashlib.md5()
m2.update(source_url.encode('utf8'))
urlHash = m2.hexdigest()
self.saveFile(urlHash,
json.dumps(main, encoding="utf8", ensure_ascii=False))
contentChilds = selector.xpath(
'//*[@id="artibody"]/child::*').extract()
image_url = ''
content = ''
type = ''
image_hash = ''
contents = []
image_urls = []
for child in contentChilds:
image_url = ''
content = ''
type = ''
image_hash = ''
curSelector = Selector(text=child)
# 特别的网站 http://tech.sina.com.cn/d/2017-06-28/doc-ifyhmtrw4294617.shtml
# http://tech.sina.com.cn/zl/post/detail/i/2017-06-28/pid_8511506.htm
if 'img_wrapper' in child or 'img' in child:
# 有的页面没有 img_wrapper,只有img
# 图片形
# 获取图片摘要，下载图片，替换图片名称
type = 'img'
image_url = curSelector.xpath('//img/@src').extract_first()
content = curSelector.xpath('//span/text()').extract_first()
# image_url = image_url[0] if image_url and len(image_url) else ''
m2 = hashlib.md5()
m2.update(image_url)
image_hash = m2.hexdigest()
image_urls.append({'url': image_url, 'hash': image_hash})
elif 'strong' in child:
# 标题形
type = 'title'
content = curSelector.xpath('//strong/text()').extract_first()
elif 'gb2312, simkai;' in child:
# 小描述形
type = 'shortInfo'
content = curSelector.xpath('//span/text()').extract_first()
elif '"pictext" align="center"' in child:
# 小描述形
type = 'centerContent'
content = curSelector.xpath('//p/text()').extract_first()
else:
# 默认
type = 'normalContent'
content = curSelector.xpath('//p').xpath('string(.)').extract()
contents.append({
'type': type,
'image_url': image_url,
'content': content,
'image_hash': image_hash
})
contentItem['title'] = title
contentItem['post_date'] = post_date
contentItem['post_user'] = post_user
contentItem['image_urls'] = image_urls
contentItem['page_content'] = contents
contentItem['tags'] = tags
return contentItem
def parseDetail(self, response):
source_url = response.meta['source_url']
contentItem = response.meta['contentItem']
selector = Selector(text=response.body)
pl_main_content = selector.xpath(
'//*[@id="pl_main_content"]').extract_first()
title = selector.xpath(
'//*[@id="main_title"]/text()').extract_first() or ''
post_date = selector.xpath(
'//*[@id="page-tools"]/span/span[1]/text()').extract_first() or ''
post_user = selector.xpath(
'//*[@id="page-tools"]/span/span[2]/text() | //*[@id="page-tools"]/span/span[2]/a/text()'
).extract()
tags = selector.xpath(
'//p[@class="art_keywords"]/a/text()').extract() or []
tags = ','.join(tags)
if len(post_user):
post_user = ''.join(post_user)
else:
post_user = ''
main = {
'title': title,
'post_date': post_date,
'post_user': post_user,
'pl_main_content': pl_main_content,
'tags': tags
}
m2 = hashlib.md5()
m2.update(source_url.encode('utf8'))
urlHash = m2.hexdigest()
self.saveFile(urlHash,
json.dumps(main, encoding="utf8", ensure_ascii=False))
contentChilds = selector.xpath(
'//*[@id="artibody"]/child::*').extract()
image_url = ''
content = ''
type = ''
image_hash = ''
contents = []
image_urls = []
for child in contentChilds:
image_url = ''
content = ''
type = ''
image_hash = ''
curSelector = Selector(text=child)
# 特别的网站 http://tech.sina.com.cn/d/2017-06-28/doc-ifyhmtrw4294617.shtml
# http://tech.sina.com.cn/zl/post/detail/i/2017-06-28/pid_8511506.htm
if 'img_wrapper' in child or 'img' in child:
# 有的页面没有 img_wrapper,只有img
# 图片形
# 获取图片摘要，下载图片，替换图片名称
type = 'img'
image_url = curSelector.xpath('//img/@src').extract_first()
content = curSelector.xpath('//span/text()').extract_first()
# image_url = image_url[0] if image_url and len(image_url) else ''
m2 = hashlib.md5()
m2.update(image_url)
image_hash = m2.hexdigest()
image_urls.append({'url': image_url, 'hash': image_hash})
elif 'strong' in child:
# 标题形
type = 'title'
content = curSelector.xpath('//strong/text()').extract_first()
elif 'font-family: KaiTi_GB2312, KaiTi;' in child:
# 小描述形
type = 'shortInfo'
content = curSelector.xpath('//span/text()').extract_first()
elif '"pictext" align="center"' in child:
# 小描述形
type = 'centerContent'
content = curSelector.xpath('//p/text()').extract_first()
else:
# 默认
type = 'normalContent'
content = curSelector.xpath('//p').xpath('string(.)').extract()
contents.append({
'type': type,
'image_url': image_url,
'content': content,
'image_hash': image_hash
})
contentItem['title'] = title
contentItem['post_date'] = post_date
contentItem['post_user'] = post_user
contentItem['image_urls'] = image_urls
contentItem['page_content'] = contents
contentItem['tags'] = tags
return contentItem
def saveFile(self, title, content):
filename = 'html/%s.json' % title
with open(filename, 'wb') as f:
f.write(content.encode('utf8'))
self.log('Saved file %s' % filename)
class DetailSpider(scrapy.Spider):
name = 'demoName_detail'
download_delay = 2.5 # 基础间隔 0.5*download_delay --- 1.5*download_delays之间的随机数
handle_httpstatus_list = [301, 302, 204, 206, 403, 404,
500] # 可以处理重定向及其他错误码导致的 页面无法获取解析的问题
def __init__(self, name=None, **kwargs):
super(DetailSpider, self).__init__(name=None, **kwargs)
self.count = 0
self.request_stop = False
self.request_stop_time = 0
self.logDao = LogDao(self.logger, 'demoName_list_detail')
self.checkDao = CheckDao()
# 用于缓存css
self.css = {'hash': 'style'}
self.dataMonitor = DataMonitorDao()
self.isRunningStop = False
def close(spider, reason):
if not spider.isRunningStop:
# 如果启动爬虫时候，还有未完成的抓取，此时不应该设置状态为停止，反之
spider.saveStatus('stop')
# spider.dataMonitor.updateTotal('demoName_total')
pass
def start_requests(self):
# 如果正在爬，就不请求
status = self.getStatus()
if status == 'running':
self.isRunningStop = True
return
self.saveStatus('running')
# 检测网络
while not NetworkUtil.checkNetWork():
# 20s检测一次
TimerUtil.sleep(20)
self.logDao.warn(u'检测网络不可行')
# 检测服务器
while not NetworkUtil.checkService():
# 20s检测一次
TimerUtil.sleep(20)
self.logDao.warn(u'检测服务器不可行')
# 进行页面访问
newUrl = 'http://tech.qq.com/l/scroll.htm'
self.logDao.warn(u'进行抓取列表:' + newUrl)
yield scrapy.Request(url=newUrl,
meta={
'request_type': 'demoName_page_list',
'url': newUrl
},
callback=self.parseArticleList,
dont_filter=True)
# TODO...还没有遇到被禁止的情况
def parseArticleList(self, response):
body = EncodeUtil.toUnicode(response.body)
if False:
self.logDao.info(u'访问过多被禁止')
else:
self.logDao.info(u'开始解析列表')
selector = Selector(text=body)
articles = selector.xpath('//div[@class="mod newslist"]//li')
for article in articles:
source_url = article.xpath('a/@href').extract_first('')
title = article.xpath('a/text()').extract_first('')
post_date = article.xpath('span/text()').extract_first('')
post_date = time.strftime('%Y', time.localtime(
time.time())) + u'年' + post_date
if not source_url:
continue
# 如果存在则不抓取
if self.checkDao.checkExist(source_url):
self.logDao.info(u'文章已经存在' + title + ':' + post_date +
':' + source_url)
continue
self.logDao.info(u'抓取文章' + title + ':' + post_date + ':' +
source_url)
yield scrapy.Request(url=source_url,
meta={
'request_type': 'demoName_detail',
"title": title,
'post_date': post_date,
"source_url": source_url
},
callback=self.parseArticle)
def parseArticle(self, response):
body = EncodeUtil.toUnicode(response.body)
if False:
self.logDao.info(u'访问过多被禁止')
else:
title = response.meta['title']
post_date = response.meta['post_date']
source_url = response.meta['source_url']
self.logDao.info(u'开始解析文章:' + title + ':' + post_date + ':' +
source_url)
selector = Selector(text=body)
# 得到样式
styleUrls = selector.xpath(
'//link[@rel="stylesheet"]/@href').extract()
styleList = []
for styleUrl in styleUrls:
if styleUrl.startswith('//'):
styleUrl = 'http:' + styleUrl
# 得到hash作为key
styleUrlHash = EncryptUtil.md5(styleUrl)
if not self.css.get(styleUrlHash):
# 不存在则去下载 并保存
self.css[styleUrlHash] = CssUtil.downLoad(styleUrl)
styleList.append(self.css[styleUrlHash])
styles = CssUtil.compressCss(styleList).replace('\'', '"').replace(
'\\', '\\\\')
# 替换样式里面的链接
styles = CssUtil.clearUrl(styles)
category = selector.xpath(
'//*[@class="a_catalog"]/a/text()|//*[@class="a_catalog"]/text()'
).extract_first('')
post_user = selector.xpath(
'//*[@class="a_author"]/text() | //*[@class="where"]/text()| //*[@class="where"]/a/text()'
).extract_first('')
src_ref = selector.xpath(
'//*[@class="a_source"]/text() | //*[@class="a_source"]/a/text()'
).extract_first('')
a_time = selector.xpath(
'//*[@class="a_time"]/text() | //*[@class="pubTime"]/text()'
).extract_first('')
if a_time:
post_date = a_time
else:
post_date = (post_date or '')
post_date = post_date.replace(u'年', '-').replace(
u'月', '-').replace(u'日', u' ').replace(u'\xa0', u' ')
try:
post_date = time.strftime(
"%Y-%m-%d %H:%M:%S",
time.strptime(post_date, "%Y-%m-%d %H:%M"))
except Exception:
pass
content_html = selector.xpath('//div[@id="Cnt-Main-Article-QQ"]')
if not len(content_html):
self.logDao.info(u'不存在内容：' + source_url)
return
# 去除内部不需要的标签
# 完整案例：content_html.xpath('*[not(boolean(@class="entHdPic" or @class="ep-source cDGray")) and not(name(.)="script")]').extract()
content_items = content_html.xpath(
'*[not(name(.)="script") and not(name(.)="style") and not(name(.)="iframe") and not(boolean(@class="rv-root-v2 rv-js-root"))]'
)
if not len(content_items):
self.logDao.info(u'不存在内容：' + source_url)
return
# 得到纯文本
content_txt = []
for item in content_items:
# 文本
allTxt = item.xpath('.//text()').extract()
allTxt = ''.join(allTxt).replace('\t', '')
# 加入
content_txt.append(allTxt)
content_txt = '\n'.join(content_txt)
# 组装新的内容标签
outHtml = """<div id="Cnt-Main-Article-QQ" class="Cnt-Main-Article-QQ" bosszone="content">${++content++}</div>"""
content_items = content_items.extract()
content_items = ''.join(content_items)
content_html = outHtml.replace('${++content++}', content_items)
selector = Selector(text=content_html)
# 解析文档中的所有图片url，然后替换成标识
image_urls = []
imgs = selector.xpath('descendant::img')
for img in imgs:
# 图片可能放在src 或者data-src
image_url_base = img.xpath('@src').extract_first('')
if image_url_base.startswith('//'):
image_url = 'http:' + image_url_base
else:
image_url = image_url_base
if image_url and image_url.startswith('http'):
self.logDao.info(u'得到图片：' + image_url)
image_urls.append({
'url': image_url,
})
content_html = content_html.replace(
image_url_base, image_url)
urlHash = EncryptUtil.md5(source_url.encode('utf8'))
self.saveFile(urlHash, body)
# 得到hashCode
hash_code = self.checkDao.getHashCode(source_url)
# 去除 image 的 alt title
selector = Selector(text=content_html)
imgAltTitles = selector.xpath('//img/@alt|//img/@title').extract()
# 处理提示块img的 alt title, 关注//img/@alt|//img/@title
for imgAltTitle in imgAltTitles:
if imgAltTitle.strip(' '):
content_html = content_html.replace(imgAltTitle, '')
contentItem = ContentItem()
contentItem['content_txt'] = content_txt
contentItem['image_urls'] = image_urls
contentItem['title'] = title
contentItem['source_url'] = source_url
contentItem['post_date'] = post_date
contentItem['sub_channel'] = category
contentItem['post_user'] = post_user
contentItem['tags'] = ''
contentItem['styles'] = styles
contentItem['content_html'] = content_html
contentItem['hash_code'] = hash_code
contentItem['info_type'] = 1
contentItem['src_source_id'] = 3
# contentItem['src_account_id'] = 0
contentItem['src_channel'] = '腾讯科技'
contentItem['src_ref'] = src_ref
return contentItem
def saveFile(self, title, content):
# TODO..暂时不保存，考虑保存下来复用效果不佳
return
# filename = 'html/%s.html' % title
# with open(filename, 'wb') as f:
# f.write(content.encode("utf8"))
# self.log('Saved file %s' % filename)
def getStatus(self):
loadF = None
try:
with open("catchStatus.json", 'r') as loadF:
aa = json.load(loadF)
return aa.get('status')
finally:
if loadF:
loadF.close()
def saveStatus(self, status):
loadF = None
try:
with open("catchStatus.json", "w") as loadF:
json.dump({'status': status}, loadF)
finally:
if loadF:
loadF.close()
class WXSourceSpider(scrapy.Spider):
name = 'wx_source'
download_delay = 20 # 基础间隔 0.5*download_delay --- 1.5*download_delays之间的随机数
handle_httpstatus_list = [301, 302, 204, 206, 403, 404,
500] # 可以处理重定向及其他错误码导致的 页面无法获取解析的问题
def __init__(self, name=None, **kwargs):
super(WXSourceSpider, self).__init__(name=None, **kwargs)
self.count = 0
self.wxSourceDao = WxSourceDao()
self.currIp = ''
self.logDao = LogDao(self.logger, 'weixin_source_catch')
self.dataMonitor = DataMonitorDao()
def close(spider, reason):
spider.saveStatus('stop')
spider.dataMonitor.updateTotal('weixin_source_total')
def start_requests(self):
# 如果正在爬，就不请求
status = self.getStatus()
if status == 'running':
return
self.saveStatus('running')
# 检测网络
while not NetworkUtil.checkNetWork():
# 20s检测一次
TimerUtil.sleep(20)
self.logDao.warn(u'检测网络不可行')
# 检测服务器
while not NetworkUtil.checkService():
# 20s检测一次
TimerUtil.sleep(20)
self.logDao.warn(u'检测服务器不可行')
# 进行爬虫
# 获取源 可用的，且（是更新失败的，或者最新的同时更新时间跟当前相比大于40分钟）
sources = self.wxSourceDao.queryEnable(isRandom=True)
for source in sources:
# 更新当前条状态为 更新中，如果更新失败或者被绊则更新为更新失败，更新成功之后设置为成功
(wx_name, wx_account, wx_url, wx_avatar, update_status, is_enable,
update_time) = source
# 更新状态为更新中
self.wxSourceDao.updateStatus(wx_account, 'updating')
# 进行页面访问
url = 'http://weixin.sogou.com/weixin?type=1&s_from=input&ie=utf8&_sug_=n&_sug_type_=&query='
newUrl = url + wx_account
self.logDao.warn(u'进行抓取:' + newUrl)
yield scrapy.Request(url=newUrl,
meta={
'request_type': 'weixin_source',
'url': newUrl,
'wx_account': wx_account,
'source': source
},
callback=self.parseList,
dont_filter=True)
def parseList(self, response):
source = response.meta['source']
wx_account = response.meta['wx_account']
url = response.meta['url']
body = EncodeUtil.toUnicode(response.body)
# 判断被禁止 提示需要重启路由 清理cookie
if response.status == 302:
# 更新状态为更新失败
self.logDao.warn(u'您的访问过于频繁,重新拨号')
self.wxSourceDao.updateStatus(wx_account, 'updateFail')
# 获取Ip # 同时空线程30s
NetworkUtil.getNewIp()
TimerUtil.sleep(30)
else:
self.logDao.info(u'开始解析:' + wx_account)
# 进行解析
selector = Selector(text=body)
results = selector.xpath('//*[@id="main"]/div[4]/ul/li')
self.logDao.info(u'列表长度:' + str(len(results)))
hasCatch = False
for result in results:
wx_name = result.xpath(
'//*[@id="sogou_vr_11002301_box_0"]/div/div[2]/p[1]/a/text()'
).extract_first()
wx_account_ = result.xpath(
'//p[@class="info"]/label/text()').extract_first()
wx_url = result.xpath(
'//p[@class="tit"]/a/@href').extract_first()
if wx_account_ == wx_account:
self.logDao.info(u'成功抓取:' + wx_account_)
self.wxSourceDao.updateSource(wx_account, wx_name, wx_url,
'last')
hasCatch = True
break
if not hasCatch:
self.logDao.info(u'没有抓到:' + wx_account_)
self.wxSourceDao.updateStatus(wx_account, 'none')
pass
def getStatus(self):
try:
with open("catchStatus.json", 'r') as load_f:
aa = json.load(load_f)
return aa.get('status')
finally:
if load_f:
load_f.close()
def saveStatus(self, status):
try:
with open("catchStatus.json", "w") as f:
json.dump({'status': status}, f)
finally:
if f:
f.close()
class DetailSpider(scrapy.Spider):
name = 'jiemian_detail'
download_delay = 2.5 # 基础间隔 0.5*download_delay --- 1.5*download_delays之间的随机数
handle_httpstatus_list = [301, 302, 204, 206, 403, 404,
500] # 可以处理重定向及其他错误码导致的 页面无法获取解析的问题
def __init__(self, name=None, **kwargs):
super(DetailSpider, self).__init__(name=None, **kwargs)
self.count = 0
self.request_stop = False
self.request_stop_time = 0
self.logDao = LogDao(self.logger, 'jiemian_list_detail')
self.checkDao = CheckDao()
# 用于缓存css
self.css = {'hash': 'style'}
self.dataMonitor = DataMonitorDao()
self.isRunningStop = False
def close(spider, reason):
if not spider.isRunningStop:
# 如果启动爬虫时候，还有未完成的抓取，此时不应该设置状态为停止，反之
spider.saveStatus('stop')
# spider.dataMonitor.updateTotal('jiemian_total')
pass
def start_requests(self):
# 如果正在爬，就不请求
status = self.getStatus()
if status == 'running':
self.isRunningStop = True
return
self.saveStatus('running')
# 检测网络
while not NetworkUtil.checkNetWork():
# 20s检测一次
TimerUtil.sleep(20)
self.logDao.warn(u'检测网络不可行')
# 检测服务器
while not NetworkUtil.checkService():
# 20s检测一次
TimerUtil.sleep(20)
self.logDao.warn(u'检测服务器不可行')
# 必读 玩物 产品榜 快报 游戏要闻 单品 盘点 花边要闻 游戏快报
cids = [{
'src_channel': u'界面科技',
'sub_channel': u'必读',
'num': '6'
}, {
'src_channel': u'界面科技',
'sub_channel': u'玩物',
'num': '66'
}, {
'src_channel': u'界面科技',
'sub_channel': u'产品榜',
'num': '73'
}, {
'src_channel': u'界面科技',
'sub_channel': u'快报',
'num': '84'
}, {
'src_channel': u'界面游戏',
'sub_channel': u'游戏要闻',
'num': '100'
}, {
'src_channel': u'界面游戏',
'sub_channel': u'单品',
'num': '119'
}, {
'src_channel': u'界面游戏',
'sub_channel': u'盘点',
'num': '120'
}, {
'src_channel': u'界面游戏',
'sub_channel': u'花边要闻',
'num': '121'
}, {
'src_channel': u'界面游戏',
'sub_channel': u'游戏快报',
'num': '122'
}]
# 必读
url = 'https://a.jiemian.com/index.php?m=lists&a=ajaxlist&callback=&_=1502103362598&page='
for cid in cids:
for page in range(1, 2):
cidNum = cid.get('num')
src_channel = cid.get('src_channel')
sub_channel = cid.get('sub_channel')
newUrl = url + str(page) + ('&cid=' + cidNum)
self.logDao.warn(u'进行抓取列表:' + newUrl)
yield scrapy.Request(url=newUrl,
meta={
'request_type': 'jiemian_page_list',
'url': newUrl,
'src_channel': src_channel,
'sub_channel': sub_channel
},
callback=self.parseArticleList,
dont_filter=True)
# TODO...还没有遇到被禁止的情况
def parseArticleList(self, response):
body = EncodeUtil.toUnicode(response.body)
if False:
self.logDao.info(u'访问过多被禁止')
else:
# 格式化
jsonStr = demjson.decode(body.lstrip('(').rstrip(')')) or {}
rst = jsonStr.get('rst', '')
if not rst:
self.logDao.info(u'不存在内容')
return
self.logDao.info(u'开始解析列表')
src_channel = response.meta['src_channel']
sub_channel = response.meta['sub_channel']
selector = Selector(text=rst)
articles = selector.xpath(
'//div[boolean(contains(@class,"news-view"))]')
for article in articles:
source_url = article.xpath(
'.//div[@class="news-header"]//a/@href').extract_first('')
title = article.xpath(
'.//div[@class="news-header"]//a/@title | .//div[@class="news-header"]//a/text()'
).extract_first('')
post_date = article.xpath(
'.//div[@class="news-footer"]//span[@class="date"]/text()'
).extract_first('')
tags = article.xpath(
'.//div[@class="news-tag"]/a/text()').extract()
if not source_url:
self.logDao.info(u'文章不存在' + title + ':' + source_url +
':' + post_date)
continue
# 如果存在则不抓取
if self.checkDao.checkExist(source_url):
self.logDao.info(u'文章已经存在' + title + ':' + source_url +
':' + post_date)
continue
self.logDao.info(u'抓取文章' + title + ':' + source_url + ':' +
post_date)
yield scrapy.Request(url=source_url,
meta={
'request_type': 'jiemian_detail',
"title": title,
'post_date': post_date,
'sub_channel': sub_channel,
'src_channel': src_channel,
'tags': tags,
'source_url': source_url
},
callback=self.parseArticle)
def parseArticle(self, response):
body = EncodeUtil.toUnicode(response.body)
if False:
self.logDao.info(u'访问过多被禁止')
else:
title = response.meta['title']
post_date = response.meta['post_date']
source_url = response.meta['source_url']
sub_channel = response.meta['sub_channel']
src_channel = response.meta['src_channel']
tags = response.meta['tags']
self.logDao.info(u'开始解析文章:' + title + ':' + post_date + ':' +
source_url)
selector = Selector(text=body)
# 得到样式
styleUrls = selector.xpath(
'//link[@rel="stylesheet"]/@href').extract()
styleList = []
for styleUrl in styleUrls:
if styleUrl.startswith('//'):
styleUrl = 'http:' + styleUrl
# 得到hash作为key
styleUrlHash = EncryptUtil.md5(styleUrl)
if not self.css.get(styleUrlHash):
# 不存在则去下载 并保存
self.css[styleUrlHash] = CssUtil.downLoad(styleUrl)
styleList.append(self.css[styleUrlHash])
styles = CssUtil.compressCss(styleList).replace('\'', '"').replace(
'\\', '\\\\')
# 替换样式里面的链接
styles = CssUtil.clearUrl(styles)
styles = CssUtil.clearBackgroundColor(styles, ['#f5f5f5'])
post_user = selector.xpath(
'//div[@class="article-info"]//span[@class="author"]//text()'
).extract_first('')
src_ref = src_channel
post_date = selector.xpath(
'//div[@class="article-info"]//span[@class="date"]//text()'
).extract_first('')
try:
post_date = time.strftime(
"%Y-%m-%d %H:%M:%S",
time.strptime(post_date, "%Y/%m/%d %H:%M"))
except Exception:
pass
tags_ = selector.xpath(
'//div[@class="article-info"]//*[@class="tags"]//text()'
).extract()
tags = tags + tags_
tags = ','.join(tags)
"""
article-main
article-img
article-content
p
article-source
p:来源
p:点击下载“界面新闻”APP 不抓
"""
# 得到article-img
article_img = selector.xpath(
'//div[@class="article-main"]/div[@class="article-img"]'
).extract_first('')
# 得到article-content
article_content = selector.xpath(
'//div[@class="article-main"]/div[@class="article-content"]'
).extract_first('')
if not article_content:
self.logDao.info(u'文章不存在' + title + ':' + source_url + ':' +
post_date)
return
contentSelector = Selector(text=article_content)
content_items = contentSelector.xpath(
'//div[@class="article-content"]/*[not(name(.)="script") and not('
'name(.)="iframe") and not(name(.)="style") and not(boolean( '
'contains(a//@href,"?m=app"))) and not(boolean(@class="share-view" '
'or @class="article-source"))]')
# 得到来源 做替换
contentSource = contentSelector.xpath(
'//div[@class="article-content"]/div[@class="article-source"]/p/text()'
).extract_first('')
if contentSource:
contentSource = contentSource.replace(u'来源：', u'')
src_ref = contentSource
# 得到纯文本
content_txt = []
for item in content_items:
# 文本
allTxt = item.xpath('.//text()').extract()
allTxt = ''.join(allTxt).replace('\t', '')
# 加入
content_txt.append(allTxt)
content_txt = '\n'.join(content_txt)
# 组装新的内容标签
outHtml = u"""<div class="article-main">${++articleImg++}<div class="article-content" style="font-family:
'Microsoft YaHei', 黑体;">${++content++}</div></div> """
content_items = content_items.extract()
content_items = ''.join(content_items)
content_html = outHtml.replace('${++articleImg++}',
article_img).replace(
'${++content++}', content_items)
selector = Selector(text=content_html)
# 解析文档中的所有图片url，然后替换成标识
image_urls = []
imgs = selector.xpath('descendant::img')
for img in imgs:
# 图片可能放在src 或者data-src
image_url_base = img.xpath('@src').extract_first('')
if image_url_base.startswith('//'):
image_url = 'http:' + image_url_base
else:
image_url = image_url_base
if image_url and image_url.startswith('http'):
self.logDao.info(u'得到图片：' + image_url)
image_urls.append({
'url': image_url,
})
content_html = content_html.replace(
image_url_base, image_url)
urlHash = EncryptUtil.md5(source_url.encode('utf8'))
self.saveFile(urlHash, body)
# 得到hashCode
hash_code = self.checkDao.getHashCode(source_url)
# 去除 image 的 alt title
selector = Selector(text=content_html)
imgAltTitles = selector.xpath('//img/@alt|//img/@title').extract()
# 处理提示块img的 alt title, 关注//img/@alt|//img/@title
for imgAltTitle in imgAltTitles:
if imgAltTitle.strip(' '):
content_html = content_html.replace(imgAltTitle, '')
contentItem = ContentItem()
contentItem['content_txt'] = content_txt
contentItem['image_urls'] = image_urls
contentItem['title'] = title
contentItem['source_url'] = source_url
contentItem['post_date'] = post_date
contentItem['sub_channel'] = sub_channel
contentItem['post_user'] = post_user
contentItem['tags'] = tags
contentItem['styles'] = styles
contentItem['content_html'] = content_html
contentItem['hash_code'] = hash_code
contentItem['info_type'] = 1
contentItem['src_source_id'] = 5
# contentItem['src_account_id'] = 0
contentItem['src_channel'] = src_channel
contentItem['src_ref'] = src_ref
return contentItem
def saveFile(self, title, content):
# TODO..暂时不保存，考虑保存下来复用效果不佳
return
# filename = 'html/%s.html' % title
# with open(filename, 'wb') as f:
# f.write(content.encode("utf8"))
# self.log('Saved file %s' % filename)
def getStatus(self):
loadF = None
try:
with open("catchStatus.json", 'r') as loadF:
aa = json.load(loadF)
return aa.get('status')
finally:
if loadF:
loadF.close()
def saveStatus(self, status):
loadF = None
try:
with open("catchStatus.json", "w") as loadF:
json.dump({'status': status}, loadF)
finally:
if loadF:
loadF.close()
class SinaSpider(scrapy.Spider):
name = 'sina2'
download_delay = 2.5 # 基础间隔 0.5*download_delay --- 1.5*download_delays之间的随机数
handle_httpstatus_list = [301, 302, 204, 206, 403, 404, 500] # 可以处理重定向及其他错误码导致的 页面无法获取解析的问题
# 错误码 [scrapy.downloadermiddlewares.retry] DEBUG: Retrying <GET http://tech.sina.com.cn/i/2017-07-18/doc-ifyiakur9086112.shtml> (failed 1 times): TCP connection timed out: 10060: �������ӷ���һ��ʱ���û����ȷ�����ӵ�����û�з�Ӧ�����ӳ���ʧ�ܡ�.
# [scrapy.downloadermiddlewares.retry] DEBUG: Retrying <GET http://tech.sina.com.cn/i/2017-07-17/doc-ifyiakwa4300270.shtml> (failed 1 times): User timeout caused connection failure: Getting http://tech.sina.com.cn/i/2017-07-17/doc-ifyiakwa4300270.shtml took longer than 180.0 seconds..
def __init__(self, name=None, **kwargs):
super(SinaSpider, self).__init__(name=None, **kwargs)
self.count = 0
self.logDao = LogDao(self.logger, 'sina_detail')
self.checkDao = CheckDao()
# 用于缓存css
self.css = {
'hash': 'style'
}
self.dataMonitor = DataMonitorDao()
self.logger.info(u'重走init')
def close(spider, reason):
spider.dataMonitor.updateTotal('sina_total')
def start_requests(self):
# while True:
# 检测网络
while not NetworkUtil.checkNetWork():
# 20s检测一次
TimerUtil.sleep(20)
self.logDao.warn(u'检测网络不可行')
# continue
# 检测服务器
while not NetworkUtil.checkService():
# 20s检测一次
TimerUtil.sleep(20)
self.logDao.warn(u'检测服务器不可行')
# continue
# 进行爬虫
url = 'http://roll.news.sina.com.cn/interface/rollnews_ch_out_interface.php?col=96&spec=&type=&ch=01&k=&offset_page=0&offset_num=0&num=220&asc=&page=1'
# url = 'http://tech.sina.com.cn/t/2017-07-24/doc-ifyihrit1274195.shtml'
r = random.uniform(0, 1)
newUrl = url + ('&r=' + str(r))
self.logDao.info(u"开始抓取列表：" + newUrl)
yield scrapy.Request(url=newUrl, meta={'request_type': 'sina_list', 'url': newUrl}, callback=self.parseList)
# 补缺补漏
url = 'http://feed.mix.sina.com.cn/api/roll/get?pageid=372&lid=2431&k=&num=50&page=1&callback=&_=1501148356254'
r = random.uniform(0, 1)
newUrl = url + ('&r=' + str(r))
self.logDao.info(u"开始抓取列表：" + newUrl)
yield scrapy.Request(url=newUrl, meta={'request_type': 'sina_list', 'url': newUrl}, callback=self.parseList2)
# TODO。。还没有找到被禁止的情况
def parseList2(self, response):
data = EncodeUtil.toUnicode(response.body)
if False:
self.logDao.info(u'访问过多被禁止')
else:
url = response.meta['url']
self.logDao.info(u'开始解析列表' + url)
# 格式化
data = demjson.decode(data) or {}
result = data.get('result', {})
list = result.get('data', [])
for item in list:
channel_name = u'科技'
title = item.get('title', '')
source_url = item.get('url', '')
callback = self.parseDetail2
if self.checkDao.checkExist(source_url):
self.logDao.info(u'文章已经存在：' + title + source_url)
continue
self.logDao.info(u"开始抓取文章：" + source_url)
# item['url'] = "http://tech.sina.com.cn/d/f/2017-07-31/doc-ifyinvwu3872514.shtml"
yield scrapy.Request(url=item['url'],
meta={'request_type': 'sina_detail', 'category': channel_name,
'title': title, 'source_url': source_url},
callback=callback)
# TODO。。还没有找到被禁止的情况
def parseList(self, response):
data = EncodeUtil.toUnicode(response.body)
if False:
self.logDao.info(u'访问过多被禁止')
else:
url = response.meta['url']
self.logDao.info(u'开始解析列表' + url)
data = data.lstrip('var jsonData = ').rstrip(';')
# 格式化
data = demjson.decode(data) or {}
list = data.get('list', [])
for item in list:
channel = item.get('channel', {})
channel_name = channel.get('title', '')
title = item.get('title', '')
source_url = item.get('url', '')
callback = self.parseDetail2
if self.checkDao.checkExist(source_url):
self.logDao.info(u'文章已经存在：' + title + source_url)
continue
self.logDao.info(u"开始抓取文章：" + item['url'])
# item['url'] = "http://tech.sina.com.cn/d/f/2017-07-31/doc-ifyinvwu3872514.shtml"
yield scrapy.Request(url=item['url'],
meta={'request_type': 'sina_detail', 'category': channel_name,
'title': title, 'source_url': source_url},
callback=callback)
def parseDetail2(self, response):
body = EncodeUtil.toUnicode(response.body)
if False:
self.logDao.info(u'访问过多被禁止')
else:
category = response.meta['category']
title = response.meta['title']
source_url = response.meta['source_url']
self.logDao.info(u'开始解析文章:' + title + ':' + category + ':' + source_url)
selector = Selector(text=body)
# 得到样式
styleUrls = selector.xpath('//link[@rel="stylesheet"]/@href').extract()
styleList = []
for styleUrl in styleUrls:
# 得到hash作为key
styleUrlHash = EncryptUtil.md5(styleUrl)
if not self.css.get(styleUrlHash):
# 不存在则去下载 并保存
self.css[styleUrlHash] = CssUtil.downLoad(styleUrl)
styleList.append(self.css[styleUrlHash])
styles = CssUtil.compressCss(styleList).replace('\'', '"').replace('\\', '\\\\')
styles = CssUtil.clearUrl(styles)
styles = styles.replace('overflow-x:hidden', '').replace('overflow:hidden', '')
post_date = selector.xpath('//*[@id="pub_date"]/text() | //*[@class="titer"]/text()').extract_first('')
post_date = post_date.replace('\r\n', '').strip(' ').replace(u'年', '-').replace(u'月', '-').replace(u'日', ' ')
try:
post_date = time.strftime("%Y-%m-%d %H:%M:%S", time.strptime(post_date, "%Y-%m-%d %H:%M"))
except Exception:
pass
src_ref = selector.xpath(
'//*[@id="media_name"]/a[1]/text() | //*[@class="source"]/a/text() | //*[@class="source"]/text()').extract_first(
'')
post_user = selector.xpath('//*[@id="author_ename"]/a/text()').extract_first('')
tags = selector.xpath('//p[@class="art_keywords"]/a/text()').extract() or []
tags = ','.join(tags)
content_html = selector.xpath('//*[@id="artibody"][1]')
if not len(content_html):
self.logDao.info(u'不存在内容：' + source_url)
return
# 去除内部不需要的标签 2017-07-24 19:23
# 完整案例：content_html.xpath('*[not(boolean(@class="entHdPic" or @class="ep-source cDGray")) and not(name(.)="script")]').extract()
content_items = content_html.xpath('*[not(boolean(@class="entHdPic")) and not(name(.)="script")]')
if not len(content_items):
self.logDao.info(u'不存在内容：' + source_url)
return
# 得到纯文本
content_txt = []
for item in content_items:
# 文本
# TODO...之后处理 取出标题类型
allTxt = item.xpath('.//text()').extract()
allTxt = ''.join(allTxt).replace('\t', '')
if u'来源：' in allTxt and len(allTxt) < 25:
# 说明这是真正的来源
if not post_user:
# 先替换作者 ，如果不存在的话
post_user = src_ref
src_ref = allTxt.replace(u'来源：', '').strip(u' ')
# 加入
content_txt.append(allTxt)
content_txt = '\n'.join(content_txt)
# 组装新的内容标签
outHtml = """<div class="content_wrappr_left"><div class="content"><div class="BSHARE_POP blkContainerSblkCon clearfix blkContainerSblkCon_16" id="artibody">${++content++}</div></div></div>"""
content_items = content_items.extract()
content_items = ''.join(content_items)
content_html = outHtml.replace('${++content++}', content_items)
selector = Selector(text=content_html)
# 解析文档中的所有图片url，然后替换成标识
image_urls = []
imgs = selector.xpath('descendant::img')
for img in imgs:
# 图片可能放在src
image_url = img.xpath('@src').extract_first()
if image_url and image_url.startswith('http'):
self.logDao.info(u'得到图片：' + image_url)
image_urls.append({
'url': image_url,
})
urlHash = EncryptUtil.md5(source_url.encode('utf8'))
self.saveFile(urlHash, body)
# 得到hashCode
hash_code = self.checkDao.getHashCode(source_url)
# 去除 image 的 alt title
selector = Selector(text=content_html)
imgAltTitles = selector.xpath('//img/@alt|//img/@title').extract()
# 处理提示块img的 alt title, 关注//img/@alt|//img/@title
for imgAltTitle in imgAltTitles:
if imgAltTitle.strip(' '):
content_html = content_html.replace(imgAltTitle, '')
contentItem = ContentItem()
contentItem['content_txt'] = content_txt
contentItem['image_urls'] = image_urls
contentItem['title'] = title
contentItem['source_url'] = source_url
contentItem['post_date'] = post_date
contentItem['sub_channel'] = category
contentItem['post_user'] = post_user
contentItem['tags'] = tags
contentItem['styles'] = styles
contentItem['content_html'] = content_html
contentItem['hash_code'] = hash_code
contentItem['info_type'] = 1
contentItem['src_source_id'] = 2
# contentItem['src_account_id'] = 0
contentItem['src_channel'] = '新浪科技'
contentItem['src_ref'] = src_ref
return contentItem
def saveFile(self, title, content):
filename = 'html/%s.html' % title
with open(filename, 'wb') as f:
f.write(content.encode("utf8"))
self.log('Saved file %s' % filename)
class DetailSpider(scrapy.Spider):
name = 'fenghuang_detail'
# 基础间隔 0.5*download_delay --- 1.5*download_delays之间的随机数
download_delay = 2.5
# 可以处理重定向及其他错误码导致的 页面无法获取解析的问题
handle_httpstatus_list = [301, 302, 204, 206, 403, 404, 500]
def __init__(self, name=None, **kwargs):
super(DetailSpider, self).__init__(name=None, **kwargs)
self.count = 0
self.request_stop = False
self.request_stop_time = 0
self.logDao = LogDao(self.logger, 'fenghuang_list_detail')
self.checkDao = CheckDao()
# 用于缓存css
self.css = {'hash': 'style'}
self.dataMonitor = DataMonitorDao()
self.isRunningStop = False
def close(spider, reason):
if not spider.isRunningStop:
# 如果启动爬虫时候，还有未完成的抓取，此时不应该设置状态为停止，反之
spider.saveStatus('stop')
# spider.dataMonitor.updateTotal('fenghuang_total')
pass
def start_requests(self):
# 如果正在爬，就不请求
status = self.getStatus()
if status == 'running':
self.isRunningStop = True
return
self.saveStatus('running')
# 检测网络
while not NetworkUtil.checkNetWork():
# 20s检测一次
TimerUtil.sleep(20)
self.logDao.warn(u'检测网络不可行')
# 检测服务器
while not NetworkUtil.checkService():
# 20s检测一次
TimerUtil.sleep(20)
self.logDao.warn(u'检测服务器不可行')
src_channel = u'凤凰财经'
sub_channel = u'电子竞技'
url = 'http://games.ifeng.com/listpage/17886/1/list.shtml'
styleUrlDefault = [
'http://p2.ifengimg.com/a/2016/0523/esports.css',
'http://y1.ifengimg.com/package/t_20130820__15953/css/pl_detail_v8.css'
]
newUrl = url
self.logDao.warn(u'进行抓取列表:' + newUrl)
yield scrapy.Request(url=newUrl,
meta={
'request_type': 'fenghuang_page_list',
'url': newUrl,
'src_channel': src_channel,
'sub_channel': sub_channel,
'styleUrlDefault': styleUrlDefault
},
callback=self.parseArticleList3,
dont_filter=True)
sub_channel = u'产品资讯'
url = 'http://games.ifeng.com/listpage/27456/1/list.shtml'
styleUrlDefault = [
'http://p0.ifengimg.com/fe/responsiveDetail/styles/pc_c38f5a0e.css'
]
newUrl = url
self.logDao.warn(u'进行抓取列表:' + newUrl)
yield scrapy.Request(url=newUrl,
meta={
'request_type': 'fenghuang_page_list',
'url': newUrl,
'src_channel': src_channel,
'sub_channel': sub_channel,
'styleUrlDefault': styleUrlDefault
},
callback=self.parseArticleList,
dont_filter=True)
sub_channel = u'热点资讯'
url = 'http://games.ifeng.com/listpage/27455/1/list.shtml'
styleUrlDefault = [
'http://p0.ifengimg.com/fe/responsiveDetail/styles/pc_c38f5a0e.css'
]
newUrl = url
self.logDao.warn(u'进行抓取列表:' + newUrl)
yield scrapy.Request(url=newUrl,
meta={
'request_type': 'fenghuang_page_list',
'url': newUrl,
'src_channel': src_channel,
'sub_channel': sub_channel,
'styleUrlDefault': styleUrlDefault
},
callback=self.parseArticleList,
dont_filter=True)
src_channel = u'凤凰科技'
sub_channel = u'资讯'
url = 'http://tech.ifeng.com/listpage/800/0/1/rtlist.shtml'
styleUrlDefault = [
'http://p0.ifengimg.com/fe/responsiveDetail/styles/pc_c38f5a0e.css'
]
newUrl = url
self.logDao.warn(u'进行抓取列表:' + newUrl)
yield scrapy.Request(url=newUrl,
meta={
'request_type': 'fenghuang_page_list',
'url': newUrl,
'src_channel': src_channel,
'sub_channel': sub_channel,
'styleUrlDefault': styleUrlDefault
},
callback=self.parseArticleList2,
dont_filter=True)
def parseArticleList2(self, response):
body = EncodeUtil.toUnicode(response.body)
if False:
self.logDao.info(u'访问过多被禁止')
else:
src_channel = response.meta['src_channel']
sub_channel = response.meta['sub_channel']
styleUrlDefault = response.meta['styleUrlDefault']
self.logDao.info(u'开始解析列表')
selector = Selector(text=body)
articles = selector.xpath('//div[@class="zheng_list pl10 box"]')
for article in articles:
source_url = article.xpath('./h1/a/@href').extract_first(
'').lstrip('%20').lstrip(' ')
title = article.xpath('./h1/a/text()').extract_first('')
if not source_url:
continue
# 如果存在则不抓取
if self.checkDao.checkExist(source_url):
self.logDao.info(u'文章已经存在' + title + ':' + source_url)
continue
self.logDao.info(u'抓取文章' + title + ':' + source_url)
yield scrapy.Request(url=source_url,
meta={
'request_type': 'fenghuang_detail',
"title": title,
"source_url": source_url,
'src_channel': src_channel,
'sub_channel': sub_channel,
'styleUrlDefault': styleUrlDefault
},
callback=self.parseArticle)
def parseArticleList3(self, response):
body = EncodeUtil.toUnicode(response.body)
if False:
self.logDao.info(u'访问过多被禁止')
else:
src_channel = response.meta['src_channel']
sub_channel = response.meta['sub_channel']
styleUrlDefault = response.meta['styleUrlDefault']
self.logDao.info(u'开始解析列表')
selector = Selector(text=body)
articles = selector.xpath(
'//div[boolean(contains(@class, "box_list"))]')
for article in articles:
source_url = article.xpath('./h2/a/@href').extract_first(
'').lstrip('%20').lstrip(' ')
title = article.xpath('./h2/a/text()').extract_first('')
if not source_url:
continue
# 如果存在则不抓取
if self.checkDao.checkExist(source_url):
self.logDao.info(u'文章已经存在' + title + ':' + source_url)
continue
self.logDao.info(u'抓取文章' + title + ':' + source_url)
yield scrapy.Request(url=source_url,
meta={
'request_type': 'fenghuang_detail',
"title": title,
"source_url": source_url,
'src_channel': src_channel,
'sub_channel': sub_channel,
'styleUrlDefault': styleUrlDefault
},
callback=self.parseArticle)
# TODO...还没有遇到被禁止的情况
def parseArticleList(self, response):
body = EncodeUtil.toUnicode(response.body)
if False:
self.logDao.info(u'访问过多被禁止')
else:
src_channel = response.meta['src_channel']
sub_channel = response.meta['sub_channel']
styleUrlDefault = response.meta['styleUrlDefault']
self.logDao.info(u'开始解析列表')
selector = Selector(text=body)
articles = selector.xpath('//div[@class="newsList"]//li')
for article in articles:
source_url = article.xpath('./a/@href').extract_first(
'').lstrip('%20').lstrip(' ')
title = article.xpath('./a/text()').extract_first('')
# 如果存在则不抓取
if self.checkDao.checkExist(source_url):
self.logDao.info(u'文章已经存在' + title + ':' + source_url)
continue
self.logDao.info(u'抓取文章' + title + ':' + source_url)
yield scrapy.Request(url=source_url,
meta={
'request_type': 'fenghuang_detail',
"title": title,
"source_url": source_url,
'src_channel': src_channel,
'sub_channel': sub_channel,
'styleUrlDefault': styleUrlDefault
},
callback=self.parseArticle)
def parseArticle(self, response):
body = EncodeUtil.toUnicode(response.body)
if False:
self.logDao.info(u'访问过多被禁止')
else:
src_channel = response.meta['src_channel']
sub_channel = response.meta['sub_channel']
styleUrlDefault = response.meta['styleUrlDefault']
title = response.meta['title']
source_url = response.meta['source_url']
self.logDao.info(u'开始解析文章:' + title + ':' + source_url)
selector = Selector(text=body)
# 得到样式
styleUrls = selector.xpath(
'//link[@rel="stylesheet"]/@href').extract()
styleUrls = styleUrls + styleUrlDefault
styleList = []
for styleUrl in styleUrls:
if styleUrl.startswith('//'):
styleUrl = 'http:' + styleUrl
# 得到hash作为key
styleUrlHash = EncryptUtil.md5(styleUrl)
if not self.css.get(styleUrlHash):
# 不存在则去下载 并保存
self.css[styleUrlHash] = CssUtil.downLoad(styleUrl)
styleList.append(self.css[styleUrlHash])
styles = CssUtil.compressCss(styleList).replace('\'', '"').replace(
'\\', '\\\\')
# 替换样式里面的链接
styles = CssUtil.clearUrl(styles)
styles = CssUtil.clearBackgroundColor(styles, ['#eaeaea'])
tags = selector.xpath(
'//meta[@name="keywords"]/@content').extract_first('')
category = selector.xpath(
'//meta[boolean(contains(@name, "og:category"))]/@content'
).extract_first('')
if category:
sub_channel = sub_channel + ',' + category
src_ref = selector.xpath(
'//span[@class="ss03"]//text()').extract_first('')
if not src_ref.replace('\n', '').replace(' ', ''):
src_ref = selector.xpath(
'//div[@id="artical_sth"]/p/text()').extract()
src_ref = ''.join(src_ref).replace('\n', '').replace(
u'来源：', '').replace(' ', '')
post_date = selector.xpath(
'//meta[@name="og:time"]/@content').extract_first('')
post_date = post_date.replace(u'年', '-').replace(
u'月', '-').replace(u'日', u' ').replace(u'\xa0', u' ')
try:
post_date = time.strftime(
"%Y-%m-%d %H:%M:%S",
time.strptime(post_date, "%Y-%m-%d %H:%M:%S"))
except Exception as e:
try:
post_date = time.strftime(
"%Y-%m-%d %H:%M:%S",
time.strptime(post_date, "%Y-%m-%d %H:%M"))
except Exception as e:
self.logDao.warn(e.message)
pass
pass
content_html = selector.xpath('//div[@id="main_content"]')
logoHtml = selector.xpath(
'//span[@class="ifengLogo"]').extract_first('')
if not len(content_html):
self.logDao.info(u'不存在内容：' + source_url)
return
# 去除内部不需要的标签
# 完整案例：content_html.xpath('*[not(boolean(@class="entHdPic" or @class="ep-source cDGray")) and not(name(.)="script")]').extract()
content_items = content_html.xpath(
'*[not(name(.)="script") and not(name(.)="style") and not(name(.)="iframe")]'
)
if not len(content_items):
self.logDao.info(u'不存在内容：' + source_url)
return
# 得到纯文本
content_txt = []
for item in content_items:
# 文本
allTxt = item.xpath('.//text()').extract()
allTxt = ''.join(allTxt).replace('\t', '')
# 加入
content_txt.append(allTxt)
content_txt = '\n'.join(content_txt)
# 组装新的内容标签
outHtml = """<div id="artical_real" class="js_img_share_area"><div id="main_content" class="js_selection_area" bosszone="content">${++content++}</div></div>"""
content_items = content_items.extract()
content_items = ''.join(content_items)
content_html = outHtml.replace('${++content++}', content_items)
content_html = content_html.replace(logoHtml, '')
selector = Selector(text=content_html)
# 解析文档中的所有图片url，然后替换成标识
image_urls = []
imgs = selector.xpath('descendant::img')
for img in imgs:
# 图片可能放在src 或者data-src
image_url_base = img.xpath('@src').extract_first('')
if image_url_base.startswith('//'):
image_url = 'http:' + image_url_base
else:
image_url = image_url_base
if image_url and image_url.startswith('http'):
self.logDao.info(u'得到图片：' + image_url)
image_urls.append({
'url': image_url,
})
content_html = content_html.replace(
image_url_base, image_url)
urlHash = EncryptUtil.md5(source_url.encode('utf8'))
self.saveFile(urlHash, body)
# 得到hashCode
hash_code = self.checkDao.getHashCode(source_url)
# 去除 image 的 alt title
selector = Selector(text=content_html)
imgAltTitles = selector.xpath('//img/@alt|//img/@title').extract()
# 处理提示块img的 alt title, 关注//img/@alt|//img/@title
for imgAltTitle in imgAltTitles:
if imgAltTitle.strip(' '):
content_html = content_html.replace(imgAltTitle, '')
contentItem = ContentItem()
contentItem['content_txt'] = content_txt
contentItem['image_urls'] = image_urls
contentItem['title'] = title
contentItem['source_url'] = source_url
contentItem['post_date'] = post_date
contentItem['sub_channel'] = sub_channel
contentItem['post_user'] = ''
contentItem['tags'] = tags
contentItem['styles'] = styles
contentItem['content_html'] = content_html
contentItem['hash_code'] = hash_code
contentItem['info_type'] = 1
contentItem['src_source_id'] = 8
# contentItem['src_account_id'] = 0
contentItem['src_channel'] = src_channel
contentItem['src_ref'] = src_ref
return contentItem
def saveFile(self, title, content):
# TODO..暂时不保存，考虑保存下来复用效果不佳
return
# filename = 'html/%s.html' % title
# with open(filename, 'wb') as f:
# f.write(content.encode("utf8"))
# self.log('Saved file %s' % filename)
def getStatus(self):
loadF = None
try:
with open("catchStatus.json", 'r') as loadF:
aa = json.load(loadF)
return aa.get('status')
finally:
if loadF:
loadF.close()
def saveStatus(self, status):
loadF = None
try:
with open("catchStatus.json", "w") as loadF:
json.dump({'status': status}, loadF)
finally:
if loadF:
loadF.close()
class TXDetailSpider(scrapy.Spider):
name = 'jiemian_detail'
download_delay = 2.5 # 基础间隔 0.5*download_delay --- 1.5*download_delays之间的随机数
handle_httpstatus_list = [301, 302, 204, 206, 403, 404,
500] # 可以处理重定向及其他错误码导致的 页面无法获取解析的问题
def __init__(self, name=None, **kwargs):
super(TXDetailSpider, self).__init__(name=None, **kwargs)
self.count = 0
self.request_stop = False
self.request_stop_time = 0
self.logDao = LogDao(self.logger, 'jiemian_list_detail')
self.checkDao = CheckDao()
# 用于缓存css
self.css = {'hash': 'style'}
self.dataMonitor = DataMonitorDao()
self.logger.info(u'重走init')
def close(spider, reason):
# spider.dataMonitor.updateTotal('jiemian_total')
pass
def start_requests(self):
# 检测网络
while not NetworkUtil.checkNetWork():
# 20s检测一次
TimerUtil.sleep(20)
self.logDao.warn(u'检测网络不可行')
# 检测服务器
while not NetworkUtil.checkService():
# 20s检测一次
TimerUtil.sleep(20)
self.logDao.warn(u'检测服务器不可行')
# 必读 玩物 产品榜 快报 游戏要闻 单品 盘点 花边要闻 游戏快报
cids = ['6', '66', '73', '84', '100', '119', '120', '121', '122']
# 必读
url = 'https://a.jiemian.com/index.php?m=lists&a=ajaxlist&callback=&_=1502103362598&page='
for cid in cids:
for page in range(1, 2):
newUrl = url + str(page) + ('&cid=' + str(cid))
self.logDao.warn(u'进行抓取列表:' + newUrl)
yield scrapy.Request(url=newUrl,
meta={
'request_type': 'jiemian_page_list',
'url': newUrl
},
callback=self.parseArticleList,
dont_filter=True)
# TODO...还没有遇到被禁止的情况
def parseArticleList(self, response):
body = EncodeUtil.toUnicode(response.body)
if False:
self.logDao.info(u'访问过多被禁止')
else:
# 格式化
jsonStr = demjson.decode(body.lstrip('(').rstrip(')')) or {}
rst = jsonStr.get('rst', '')
if not rst:
self.logDao.info(u'不存在内容')
return
self.logDao.info(u'开始解析列表')
selector = Selector(text=rst)
articles = selector.xpath('//div[@class="news-img"]/a')
for article in articles:
source_url = article.xpath('@href').extract_first('')
title = article.xpath('@title').extract_first('')
# 如果存在则不抓取
if self.checkDao.checkExist(source_url):
self.logDao.info(u'文章已经存在' + title + ':' + source_url)
continue
if not source_url:
self.logDao.info(u'文章不存在' + title + ':' + source_url)
continue
self.logDao.info(u'抓取文章' + title + ':' + source_url)
yield scrapy.Request(url=source_url,
meta={
'request_type': 'jiemian_detail',
"title": title,
"source_url": source_url
},
callback=self.parseArticle)
def parseArticle(self, response):
body = EncodeUtil.toUnicode(response.body)
if False:
self.logDao.info(u'访问过多被禁止')
else:
title = response.meta['title']
post_date = response.meta['post_date']
source_url = response.meta['source_url']
self.logDao.info(u'开始解析文章:' + title + ':' + post_date + ':' +
source_url)
selector = Selector(text=body)
# 得到样式
styleUrls = selector.xpath(
'//link[@rel="stylesheet"]/@href').extract()
styleList = []
for styleUrl in styleUrls:
# 得到hash作为key
styleUrlHash = EncryptUtil.md5(styleUrl)
if not self.css.get(styleUrlHash):
# 不存在则去下载 并保存
self.css[styleUrlHash] = CssUtil.downLoad(styleUrl)
styleList.append(self.css[styleUrlHash])
styles = CssUtil.compressCss(styleList).replace('\'', '"').replace(
'\\', '\\\\')
# 替换样式里面的链接
styles = CssUtil.clearUrl(styles)
category = selector.xpath(
'//*[@class="a_catalog"]/a/text()|//*[@class="a_catalog"]/text()'
).extract_first('')
post_user = selector.xpath(
'//*[@class="a_author"]/text() | //*[@class="where"]/text()| //*[@class="where"]/a/text()'
).extract_first('')
src_ref = selector.xpath(
'//*[@class="a_source"]/text() | //*[@class="a_source"]/a/text()'
).extract_first('')
a_time = selector.xpath(
'//*[@class="a_time"]/text() | //*[@class="pubTime"]/text()'
).extract_first('')
if a_time:
post_date = a_time
else:
post_date = (post_date or '')
post_date = post_date.replace(u'年', '-').replace(
u'月', '-').replace(u'日', u' ').replace(u'\xa0', u' ')
try:
post_date = time.strftime(
"%Y-%m-%d %H:%M:%S",
time.strptime(post_date, "%Y-%m-%d %H:%M"))
except Exception:
pass
content_html = selector.xpath('//div[@id="Cnt-Main-Article-QQ"]')
if not len(content_html):
self.logDao.info(u'不存在内容：' + source_url)
return
# 去除内部不需要的标签
# 完整案例：content_html.xpath('*[not(boolean(@class="entHdPic" or @class="ep-source cDGray")) and not(name(.)="script")]').extract()
content_items = content_html.xpath(
'*[not(name(.)="script") and not(name(.)="style") and not(name(.)="iframe") and not(boolean(@class="rv-root-v2 rv-js-root"))]'
)
if not len(content_items):
self.logDao.info(u'不存在内容：' + source_url)
return
# 得到纯文本
content_txt = []
for item in content_items:
# 文本
allTxt = item.xpath('.//text()').extract()
allTxt = ''.join(allTxt).replace('\t', '')
# 加入
content_txt.append(allTxt)
content_txt = '\n'.join(content_txt)
# 组装新的内容标签
outHtml = """<div id="Cnt-Main-Article-QQ" class="Cnt-Main-Article-QQ" bosszone="content">${++content++}</div>"""
content_items = content_items.extract()
content_items = ''.join(content_items)
content_html = outHtml.replace('${++content++}', content_items)
selector = Selector(text=content_html)
# 解析文档中的所有图片url，然后替换成标识
image_urls = []
imgs = selector.xpath('descendant::img')
for img in imgs:
# 图片可能放在src 或者data-src
image_url = img.xpath('@src').extract_first()
if image_url and image_url.startswith('http'):
self.logDao.info(u'得到图片：' + image_url)
image_urls.append({
'url': image_url,
})
urlHash = EncryptUtil.md5(source_url.encode('utf8'))
self.saveFile(urlHash, body)
# 得到hashCode
hash_code = self.checkDao.getHashCode(source_url)
# 去除 image 的 alt title
selector = Selector(text=content_html)
imgAltTitles = selector.xpath('//img/@alt|//img/@title').extract()
# 处理提示块img的 alt title, 关注//img/@alt|//img/@title
for imgAltTitle in imgAltTitles:
if imgAltTitle.strip(' '):
content_html = content_html.replace(imgAltTitle, '')
contentItem = ContentItem()
contentItem['content_txt'] = content_txt
contentItem['image_urls'] = image_urls
contentItem['title'] = title
contentItem['source_url'] = source_url
contentItem['post_date'] = post_date
contentItem['sub_channel'] = category
contentItem['post_user'] = post_user
contentItem['tags'] = ''
contentItem['styles'] = styles
contentItem['content_html'] = content_html
contentItem['hash_code'] = hash_code
contentItem['info_type'] = 1
contentItem['src_source_id'] = 3
# contentItem['src_account_id'] = 0
contentItem['src_channel'] = '腾讯科技'
contentItem['src_ref'] = src_ref
return contentItem
def saveFile(self, title, content):
filename = 'html/%s.html' % title
with open(filename, 'wb') as f:
f.write(content.encode("utf8"))
self.log('Saved file %s' % filename)
class WYDetailSpider(scrapy.Spider):
name = 'wangyi_detail'
download_delay = 2.5 # 基础间隔 0.5*download_delay --- 1.5*download_delays之间的随机数
handle_httpstatus_list = [301, 302, 204, 206, 403, 404,
500] # 可以处理重定向及其他错误码导致的 页面无法获取解析的问题
def __init__(self, name=None, **kwargs):
super(WYDetailSpider, self).__init__(name=None, **kwargs)
self.count = 0
self.request_stop = False
self.request_stop_time = 0
self.logDao = LogDao(self.logger, 'wangyi_list_detail')
self.checkDao = CheckDao()
# 用于缓存css
self.css = {'hash': 'style'}
self.dataMonitor = DataMonitorDao()
self.logger.info(u'重走init')
def close(spider, reason):
spider.dataMonitor.updateTotal('wangyi_total')
def start_requests(self):
# 检测网络
while not NetworkUtil.checkNetWork():
# 20s检测一次
TimerUtil.sleep(20)
self.logDao.warn(u'检测网络不可行')
# 检测服务器
while not NetworkUtil.checkService():
# 20s检测一次
TimerUtil.sleep(20)
self.logDao.warn(u'检测服务器不可行')
# 进行页面访问
newUrl = 'http://tech.163.com/special/00094IHV/news_json.js?' + str(
random.uniform(0, 1))
self.logDao.warn(u'进行抓取列表:' + newUrl)
yield scrapy.Request(url=newUrl,
meta={
'request_type': 'wangyi_page_list',
'url': newUrl
},
callback=self.parseArticleList,
dont_filter=True)
# TODO...还没有遇到被禁止的情况
def parseArticleList(self, response):
url = response.meta['url']
body = EncodeUtil.toUnicode(response.body)
if False:
self.logDao.info(u'访问过多被禁止')
else:
self.logDao.info(u'开始解析列表')
body = body.lstrip('var data=').rstrip(';')
# 格式化
jsonStr = demjson.decode(body) or {}
articles = jsonStr.get('news') or []
categoryList = jsonStr.get('category') or []
for article_ins in articles:
for article in article_ins:
source_url = article.get('l', '')
title = article.get('t', '')
timeStr = article.get('p', '')
# 如果存在则不抓取
if self.checkDao.checkExist(source_url):
self.logDao.info(u'文章已经存在' + title + timeStr +
source_url)
continue
categoryIndex = article.get('c')
category = ''
if 0 <= categoryIndex < len(categoryList):
category = categoryList[categoryIndex].get('n')
post_date = article.get('p')
self.logDao.info(u'抓取文章' + title + ':' + post_date + ':' +
source_url)
yield scrapy.Request(url=source_url,
meta={
'request_type': 'wangyi_detail',
"title": title,
'category': category,
'post_date': post_date,
"source_url": source_url
},
callback=self.parseArticle)
def parseArticle(self, response):
body = EncodeUtil.toUnicode(response.body)
if False:
self.logDao.info(u'访问过多被禁止')
else:
category = response.meta['category']
title = response.meta['title']
post_date = response.meta['post_date']
source_url = response.meta['source_url']
self.logDao.info(u'开始解析文章:' + title + ':' + post_date + ':' +
source_url)
selector = Selector(text=body)
# 得到样式
styleUrls = selector.xpath(
'//link[@rel="stylesheet"]/@href').extract()
styleList = []
for styleUrl in styleUrls:
# 得到hash作为key
styleUrlHash = EncryptUtil.md5(styleUrl)
if not self.css.get(styleUrlHash):
# 不存在则去下载 并保存
self.css[styleUrlHash] = CssUtil.downLoad(styleUrl)
styleList.append(self.css[styleUrlHash])
styles = CssUtil.compressCss(styleList).replace('\'', '"').replace(
'\\', '\\\\')
styles = CssUtil.clearUrl(styles)
src_ref = selector.xpath(
'//*[@id="ne_article_source"]/text()').extract_first()
content_html = selector.xpath('//*[@id="endText"]')
if not len(content_html):
self.logDao.info(u'不存在内容：' + source_url)
return
# 去除内部不需要的标签
# 完整案例：content_html.xpath('*[not(boolean(@class="entHdPic" or @class="ep-source cDGray")) and not(name(.)="script")]').extract()
content_items = content_html.xpath(
'*[not(boolean(@class="gg200x300" or @class="ep-source cDGray")) and not(name(.)="script")]'
)
if not len(content_items):
self.logDao.info(u'不存在内容：' + source_url)
return
# 得到纯文本
content_txt = []
for item in content_items:
# 文本
allTxt = item.xpath('.//text()').extract()
allTxt = ''.join(allTxt).replace('\t', '')
# 加入
content_txt.append(allTxt)
content_txt = '\n'.join(content_txt)
# 组装新的内容标签
outHtml = """<div class="post_text" id="endText" style="border-top:1px solid #ddd;" jcid="5611">${++content++}</div>"""
content_items = content_items.extract()
content_items = ''.join(content_items)
content_html = outHtml.replace('${++content++}', content_items)
selector = Selector(text=content_html)
# 解析文档中的所有图片url，然后替换成标识
image_urls = []
imgs = selector.xpath('descendant::img')
for img in imgs:
# 图片可能放在src 或者data-src
image_url = img.xpath('@src').extract_first()
if image_url and image_url.startswith('http'):
self.logDao.info(u'得到图片：' + image_url)
image_urls.append({
'url': image_url,
})
urlHash = EncryptUtil.md5(source_url.encode('utf8'))
self.saveFile(urlHash, body)
# 得到hashCode
hash_code = self.checkDao.getHashCode(source_url)
# 去除 image 的 alt title
selector = Selector(text=content_html)
imgAltTitles = selector.xpath('//img/@alt|//img/@title').extract()
# 处理提示块img的 alt title, 关注//img/@alt|//img/@title
for imgAltTitle in imgAltTitles:
if imgAltTitle.strip(' '):
content_html = content_html.replace(imgAltTitle, '')
contentItem = ContentItem()
contentItem['content_txt'] = content_txt
contentItem['image_urls'] = image_urls
contentItem['title'] = title
contentItem['source_url'] = source_url
contentItem['post_date'] = post_date
contentItem['sub_channel'] = category
contentItem['post_user'] = ''
contentItem['tags'] = ''
contentItem['styles'] = styles
contentItem['content_html'] = content_html
contentItem['hash_code'] = hash_code
contentItem['info_type'] = 1
contentItem['src_source_id'] = 4
# contentItem['src_account_id'] = 0
contentItem['src_channel'] = '网易科技'
contentItem['src_ref'] = src_ref
return contentItem
def saveFile(self, title, content):
filename = 'html/%s.html' % title
with open(filename, 'wb') as f:
f.write(content.encode("utf8"))
self.log('Saved file %s' % filename)
class WXDetailSpider(scrapy.Spider):
name = 'wx_detail'
download_delay = 15 # 基础间隔 0.5*download_delay --- 1.5*download_delays之间的随机数
handle_httpstatus_list = [301, 302, 204, 206, 403, 404,
500] # 可以处理重定向及其他错误码导致的 页面无法获取解析的问题
def __init__(self, name=None, **kwargs):
super(WXDetailSpider, self).__init__(name=None, **kwargs)
self.count = 0
self.wxSourceDao = WxSourceDao()
self.logDao = LogDao(self.logger, 'weixin_list_detail')
self.checkDao = CheckDao()
self.dataMonitor = DataMonitorDao()
self.wxSources = []
self.brokenAccounts = [] # 当前被禁止了的账号，下次抓取优先抓取
def close(spider, reason):
# 存被禁止的微信账号
spider.saveBrokenAccounts(spider.brokenAccounts)
# 缓存状态
spider.saveStatus('stop')
spider.dataMonitor.updateTotal('weixin_total')
for source in spider.wxSources:
(id, wx_name, wx_account, wx_url, wx_avatar, update_status,
is_enable, update_time) = source
spider.dataMonitor.updateTotal('weixin_account_total',
account=wx_account)
def start_requests(self):
# 如果在晚上12点到早上6点不爬
hour = datetime.datetime.now().hour
if 0 <= hour <= 6:
self.logDao.info(u'这个时间不爬。0-6点')
return
# 如果正在爬，就不请求
status = self.getStatus()
if status == 'running':
return
self.saveStatus('running')
# 检测网络
while not NetworkUtil.checkNetWork():
# 20s检测一次
TimerUtil.sleep(20)
self.logDao.warn(u'检测网络不可行')
# 检测服务器
while not NetworkUtil.checkService():
# 20s检测一次
TimerUtil.sleep(20)
self.logDao.warn(u'检测服务器不可行')
# 获取源 可用有值
sources = self.wxSourceDao.queryWxUrl(isRandom=True)
# 排序优先
update_time, brokenAccounts = self.getBrokenAccounts()
firstGroup = []
secondGroup = []
for source in sources:
(id, wx_name, wx_account, wx_url, wx_avatar, update_status,
is_enable, update_time) = source
if wx_account in brokenAccounts:
firstGroup.append(source)
else:
secondGroup.append(source)
sources = firstGroup + secondGroup
self.wxSources = sources
for source in sources:
(id, wx_name, wx_account, wx_url, wx_avatar, update_status,
is_enable, update_time) = source
# 进行页面访问
newUrl = wx_url
self.logDao.warn(u'进行抓取:' + newUrl)
yield scrapy.Request(url=newUrl,
meta={
'request_type': 'weixin_page_list',
'source_url': newUrl,
'wx_account': wx_account,
'source': source,
'wx_account_id': id
},
callback=self.parseArticleList,
dont_filter=True)
def parseArticleList(self, response):
body = EncodeUtil.toUnicode(response.body)
selector = Selector(text=body)
source_url = response.meta['source_url']
wx_account = response.meta['wx_account']
title = selector.xpath('//title/text()').extract_first('').strip(u' ')
isN = u"请输入验证码" == title
if isN or response.status == 302:
self.logDao.info(u'访问过多被禁止,重新拨号')
# 存起来
self.brokenAccounts.append(wx_account)
# 获取Ip # 同时空线程30s
NetworkUtil.getNewIp()
TimerUtil.sleep(80)
NetworkUtil.openWebbrowser(source_url)
else:
source = response.meta['source']
wx_account_id = response.meta['wx_account_id']
self.logDao.info(u'开始解析列表:' + wx_account)
# 进行解析
articleJS = selector.xpath('//script/text()').extract()
for js in articleJS:
if 'var msgList = ' in js:
p8 = re.compile('var\s*msgList\s*=.*;')
matchList = p8.findall(js)
for match in matchList:
match = match.lstrip('var msgList = ').rstrip(';')
# 格式化
articles = demjson.decode(match) or {}
articles = articles['list'] or []
self.logDao.info(u'匹配到文章列表' + wx_account)
for article in articles:
app_msg_ext_info = article.get(
'app_msg_ext_info') or {}
desc = app_msg_ext_info.get('digest') or ''
title = app_msg_ext_info.get('title') or ''
# 如果存在则不抓取
if self.checkDao.checkExist(title, wx_account, 1):
self.logDao.info(u'已经存在' + wx_account + ':' +
title)
continue
detailUrl = app_msg_ext_info['content_url'] or ''
detailUrl = "http://mp.weixin.qq.com" + detailUrl
detailUrl = detailUrl.replace("amp;", "")
self.logDao.info(u'抓取' + wx_account + ':' + title +
':' + detailUrl)
if not detailUrl:
continue
yield scrapy.Request(url=detailUrl,
meta={
'request_type':
'weixin_detail',
'wx_account': wx_account,
"source": source,
"title": title,
'wx_account_id':
wx_account_id,
"source_url": detailUrl
},
callback=self.parseArticle)
def parseArticle(self, response):
body = EncodeUtil.toUnicode(response.body)
selector = Selector(text=body)
title = selector.xpath('//title/text()').extract_first('').strip(u' ')
source_url = response.meta['source_url']
wx_account = response.meta['wx_account']
isN = u"请输入验证码" == title
if isN or response.status == 302:
self.logDao.info(u'访问过多被禁止,重新拨号')
# 存起来
self.brokenAccounts.append(wx_account)
# 获取Ip # 同时空线程30s
NetworkUtil.getNewIp()
TimerUtil.sleep(80)
NetworkUtil.openWebbrowser(source_url)
else:
title = response.meta['title']
source_url = response.meta['source_url']
wx_account_id = response.meta['wx_account_id']
self.logDao.info(u'开始解析文章' + wx_account + ':' + title + ':' +
source_url)
self.logDao.info(u'开始解析文章：' + source_url)
# 进行解析
post_date = selector.xpath(
'//*[@id="post-date"]/text()').extract_first('')
try:
post_date = time.strftime("%Y-%m-%d %H:%M:%S",
time.strptime(post_date, "%Y-%m-%d"))
except Exception:
pass
styles = selector.xpath('//style/text()').extract()
styles = CssUtil.compressCss(styles).replace('\'', '"').replace(
'\\', '\\\\')
styles = CssUtil.clearUrl(styles)
styles = CssUtil.clearBackgroundColor(styles, ['#f3f3f3'])
post_user = selector.xpath(
'//*[@id="post-user"]/text()').extract_first('')
content_html = selector.xpath('//*[@id="js_content"]')
if not len(content_html):
self.logDao.info(u'不存在内容：' + source_url)
return
# 去除内部不需要的标签
content_items = content_html.xpath('*')
if not len(content_items):
self.logDao.info(u'不存在内容：' + source_url)
return
# content_items_new = []
# for item in content_items:
# itemStr = item.extract()
# if u'订阅微信' in itemStr:
# continue
# content_items_new.append(item)
# content_items = content_items_new
# 得到纯文本
content_txt = []
for item in content_items:
# 文本
allTxt = item.xpath('.//text()').extract()
allTxt = ''.join(allTxt).replace('\t', '')
# 加入
content_txt.append(allTxt)
content_txt = '\n'.join(content_txt)
# 组装新的内容标签
outHtml = """<div class="rich_media_content " id="js_content">${++content++}</div>"""
content_items = content_items.extract()
content_items = ''.join(content_items)
content_html = outHtml.replace('${++content++}', content_items)
selector = Selector(text=content_html)
# 解析文档中的所有图片url，然后替换成标识
image_urls = []
imgs = selector.xpath('descendant::img')
for img in imgs:
# 图片可能放在src 或者data-src
image_url = img.xpath('@src | @data-src').extract_first('')
if image_url and image_url.startswith('http'):
self.logDao.info(u'得到图片：' + image_url)
image_urls.append({
'url': image_url,
})
self.logDao.info(wx_account + u'得到文章：' + title + ":" + post_date +
':' + post_user)
self.logDao.info(u'得到文章：' + source_url)
# 得到hashCode1
hash_code = self.checkDao.getHashCode(title, wx_account, 1)
self.saveFile(hash_code, body)
# 去除 image 的 alt title
selector = Selector(text=content_html)
imgAltTitles = selector.xpath('//img/@alt|//img/@title').extract()
# 处理提示块img的 alt title, 关注//img/@alt|//img/@title
for imgAltTitle in imgAltTitles:
if imgAltTitle.strip(' '):
content_html = content_html.replace(imgAltTitle, '')
contentItem = ContentItem()
contentItem['content_txt'] = content_txt
contentItem['image_urls'] = image_urls
contentItem['title'] = title
contentItem['source_url'] = source_url
contentItem['post_date'] = post_date
contentItem['sub_channel'] = ''
contentItem['post_user'] = post_user
contentItem['tags'] = ''
contentItem['styles'] = styles
contentItem['content_html'] = content_html
contentItem['hash_code'] = hash_code
contentItem['info_type'] = 1
contentItem['src_source_id'] = 1
contentItem['src_account_id'] = wx_account_id
contentItem['src_channel'] = '微信公众号'
contentItem['src_ref'] = ''
contentItem['wx_account'] = wx_account
return contentItem
def saveFile(self, title, content):
# TODO..暂时不保存，考虑保存下来复用效果不佳
return
# filename = 'html/%s.html' % title
# with open(filename, 'wb') as f:
# f.write(content.encode("utf8"))
# self.log('Saved file %s' % filename)
def getBrokenAccounts(self):
loadF = None
try:
with open("brokenAccount.json", 'r') as loadF:
aa = json.load(loadF)
return aa.get('update_time', ''), aa.get('accounts', [])
finally:
if loadF:
loadF.close()
def saveBrokenAccounts(self, accounts):
loadF = None
try:
with open("brokenAccount.json", "w") as loadF:
update_time = time.strftime('%Y-%m-%d %H:%M:%S',
time.localtime(time.time()))
json.dump({
'update_time': update_time,
'accounts': accounts
}, loadF)
finally:
if loadF:
loadF.close()
def getStatus(self):
loadF = None
try:
with open("catchStatus.json", 'r') as loadF:
aa = json.load(loadF)
return aa.get('status')
finally:
if loadF:
loadF.close()
def saveStatus(self, status):
loadF = None
try:
with open("catchStatus.json", "w") as loadF:
json.dump({'status': status}, loadF)
finally:
if loadF:
loadF.close()
class DetailSpider(scrapy.Spider):
name = 'sohu_detail'
download_delay = 2.5 # 基础间隔 0.5*download_delay --- 1.5*download_delays之间的随机数
handle_httpstatus_list = [301, 302, 204, 206, 403, 404,
500] # 可以处理重定向及其他错误码导致的 页面无法获取解析的问题
def __init__(self, name=None, **kwargs):
super(DetailSpider, self).__init__(name=None, **kwargs)
self.count = 0
self.request_stop = False
self.request_stop_time = 0
self.logDao = LogDao(self.logger, 'sohu_list_detail')
self.checkDao = CheckDao()
# 用于缓存css
self.css = {'hash': 'style'}
self.dataMonitor = DataMonitorDao()
self.isRunningStop = False
def close(spider, reason):
if not spider.isRunningStop:
# 如果启动爬虫时候，还有未完成的抓取，此时不应该设置状态为停止，反之
spider.saveStatus('stop')
# spider.dataMonitor.updateTotal('sohu_total')
pass
def start_requests(self):
# 如果正在爬，就不请求
status = self.getStatus()
if status == 'running':
self.isRunningStop = True
return
self.saveStatus('running')
# 检测网络
while not NetworkUtil.checkNetWork():
# 20s检测一次
TimerUtil.sleep(20)
self.logDao.warn(u'检测网络不可行')
# 检测服务器
while not NetworkUtil.checkService():
# 20s检测一次
TimerUtil.sleep(20)
self.logDao.warn(u'检测服务器不可行')
src_channel = u'搜狐科技'
sub_channel = u'科技'
for page in range(1, 4):
# 进行页面访问
url = 'http://v2.sohu.com/public-api/feed?scene=CHANNEL&sceneId=30&size=20&callback=&_=1502075449669&page='
newUrl = url + str(page)
self.logDao.warn(u'进行抓取列表:' + newUrl)
yield scrapy.Request(url=newUrl,
meta={
'request_type': 'sohu_page_list',
'url': newUrl,
'src_channel': src_channel,
'sub_channel': sub_channel
},
callback=self.parseArticleList,
dont_filter=True)
# TODO...还没有遇到被禁止的情况
def parseArticleList(self, response):
body = EncodeUtil.toUnicode(response.body)
if False:
self.logDao.info(u'访问过多被禁止')
else:
src_channel = response.meta['src_channel']
sub_channel = response.meta['sub_channel']
# 格式化
articles = demjson.decode(
body.lstrip('/**/').lstrip('(').rstrip(';').rstrip(')')) or []
if not articles:
self.logDao.info(u'不存在内容')
return
self.logDao.info(u'开始解析列表')
for article in articles:
id = article.get('id', '')
authorId = article.get('authorId', '')
if not id or not authorId:
continue
pageId = str(id) + '_' + str(authorId)
source_url = 'http://www.sohu.com/a/' + pageId + '?loc=1&focus_pic=0'
title = article.get('title', '')
post_user = article.get('authorName', '')
tags = article.get('tags', [])
tagsStr = []
for tag in tags:
tagsStr.append(tag.get('name', ''))
publicTime = article.get('publicTime', time.time() * 1000)
post_date = time.strftime('%Y-%m-%d %H:%M:%S',
time.localtime(publicTime / 1000))
# 如果存在则不抓取
if self.checkDao.checkExist(source_url):
self.logDao.info(u'文章已经存在' + title + ':' + post_date +
':' + source_url)
continue
self.logDao.info(u'抓取文章' + title + ':' + post_date + ':' +
source_url)
yield scrapy.Request(url=source_url,
meta={
'request_type': 'sohu_detail',
'title': title,
'post_date': post_date,
'post_user': post_user,
'tags': tagsStr,
'source_url': source_url,
'src_channel': src_channel,
'sub_channel': sub_channel
},
callback=self.parseArticle)
def parseArticle(self, response):
body = EncodeUtil.toUnicode(response.body)
if False:
self.logDao.info(u'访问过多被禁止')
else:
src_channel = response.meta['src_channel']
sub_channel = response.meta['sub_channel']
title = response.meta['title']
post_user = response.meta['post_user']
tags = response.meta['tags']
post_date = response.meta['post_date']
source_url = response.meta['source_url']
self.logDao.info(u'开始解析文章:' + title + ':' + post_date + ':' +
source_url)
selector = Selector(text=body)
# 得到样式
styleUrls = selector.xpath(
'//link[@rel="stylesheet"]/@href').extract()
styleList = []
for styleUrl in styleUrls:
if styleUrl.startswith('//'):
styleUrl = 'http:' + styleUrl
# 得到hash作为key
styleUrlHash = EncryptUtil.md5(styleUrl)
if not self.css.get(styleUrlHash):
# 不存在则去下载 并保存
self.css[styleUrlHash] = CssUtil.downLoad(styleUrl)
styleList.append(self.css[styleUrlHash])
styles = CssUtil.compressCss(styleList).replace('\'', '"').replace(
'\\', '\\\\')
# 替换样式里面的链接
styles = CssUtil.clearUrl(styles)
content_html = selector.xpath('//*[@class="article"]')
backHtml = selector.xpath('//*[@id="backsohucom"]').extract_first(
'')
if not len(content_html):
self.logDao.info(u'不存在内容：' + source_url)
return
# 去除内部不需要的标签u'<p data-role="editor-name">责任编辑：<span></span></p>'
# 完整案例：content_html.xpath('*[not(boolean(@class="entHdPic" or @class="ep-source cDGray")) and not(name(.)="script")]').extract()
content_items = content_html.xpath(
'*[not(name(.)="script") and not(name(.)="style") and not(name(.)="iframe") and not(boolean(@data-role="editor-name"))]'
)
if not len(content_items):
self.logDao.info(u'不存在内容：' + source_url)
return
# 得到纯文本
content_txt = []
for item in content_items:
# 文本
allTxt = item.xpath('.//text()').extract()
allTxt = ''.join(allTxt).replace('\t', '')
# 加入
content_txt.append(allTxt)
content_txt = '\n'.join(content_txt)
# 组装新的内容标签
outHtml = """<div class="article-page"><article class="article">${++content++}</article></div>"""
content_items = content_items.extract()
content_items = ''.join(content_items)
content_html = outHtml.replace('${++content++}', content_items)
content_html = content_html.replace(backHtml, '')
selector = Selector(text=content_html)
# 解析文档中的所有图片url，然后替换成标识
image_urls = []
imgs = selector.xpath('descendant::img')
for img in imgs:
# 图片可能放在src 或者data-src
image_url_base = img.xpath('@src').extract_first('')
if image_url_base.startswith('//'):
image_url = 'http:' + image_url_base
else:
image_url = image_url_base
if image_url and image_url.startswith('http'):
self.logDao.info(u'得到图片：' + image_url)
image_urls.append({
'url': image_url,
})
content_html = content_html.replace(
image_url_base, image_url)
urlHash = EncryptUtil.md5(source_url.encode('utf8'))
self.saveFile(urlHash, body)
# 得到hashCode
hash_code = self.checkDao.getHashCode(source_url)
# 去除 image 的 alt title
selector = Selector(text=content_html)
imgAltTitles = selector.xpath('//img/@alt|//img/@title').extract()
# 处理提示块img的 alt title, 关注//img/@alt|//img/@title
for imgAltTitle in imgAltTitles:
if imgAltTitle.strip(' '):
content_html = content_html.replace(imgAltTitle, '')
contentItem = ContentItem()
contentItem['content_txt'] = content_txt
contentItem['image_urls'] = image_urls
contentItem['title'] = title
contentItem['source_url'] = source_url
contentItem['post_date'] = post_date
contentItem['sub_channel'] = sub_channel
contentItem['post_user'] = post_user
contentItem['tags'] = ','.join(tags)
contentItem['styles'] = styles
contentItem['content_html'] = content_html
contentItem['hash_code'] = hash_code
contentItem['info_type'] = 1
contentItem['src_source_id'] = 7
# contentItem['src_account_id'] = 0
contentItem['src_channel'] = src_channel
contentItem['src_ref'] = '搜狐科技'
return contentItem
def saveFile(self, title, content):
# TODO..暂时不保存，考虑保存下来复用效果不佳
return
# filename = 'html/%s.html' % title
# with open(filename, 'wb') as f:
# f.write(content.encode("utf8"))
# self.log('Saved file %s' % filename)
def getStatus(self):
loadF = None
try:
with open("catchStatus.json", 'r') as loadF:
aa = json.load(loadF)
return aa.get('status')
finally:
if loadF:
loadF.close()
def saveStatus(self, status):
loadF = None
try:
with open("catchStatus.json", "w") as loadF:
json.dump({'status': status}, loadF)
finally:
if loadF:
loadF.close()
def __init__(self, name=None, **kwargs):
super(WXDetailSpecialOneSpider, self).__init__(name=None, **kwargs)
self.count = 0
self.wxSourceDao = WxSourceDao()
self.logDao = LogDao(self.logger, 'weixin_list_detail')
self.checkDao = CheckDao()