def crawDailyMetalComments(link,webNet):
currentList = []
startContext = SilverMetalNetSpiderUtils.returnStartContext(link, '<div class="lb_you left">')
startContext = SilverMetalNetSpiderUtils.filterContextByTarget(startContext,'<ul>','</ul>')
len = SilverMetalNetSpiderUtils.findAllTarget(startContext,'<li>')
for i in range(len):
targetContext = SilverMetalNetSpiderUtils.divisionTarget(startContext,'<li>','</li>')
startContext = targetContext['nextContext']
currentContext = targetContext['targetContext']
linkUrl = webNet+ SilverMetalNetSpiderUtils.filterContextByTarget(currentContext,'href="','" target')
title = SilverMetalNetSpiderUtils.filterContextByTarget(currentContext,'target=_blank>','</a>')
title = SilverMetalNetSpiderUtils.removeSpecialCharacter(title)
pubDate = SilverMetalNetSpiderUtils.filterContextByTarget(currentContext,'</a>','</li>')
currentTime = time.strftime("%Y-%m-%d",time.localtime())
filterCurrentTime = pubDate[0:10]
if currentTime ==filterCurrentTime:
descriptContext = crawDailyDescriptContext(linkUrl)
currentList.append([str(uuid.uuid1()),linkUrl,title,pubDate,descriptContext,'METAL','SILVERNET'])
else :
break
return currentList
def crawDailyDescriptContext(linkUrl):
startContext = SilverMetalNetSpiderUtils.returnStartContext(linkUrl,'<div class="zhaiyao">')
startContext = SilverMetalNetSpiderUtils.filterAfterContext(startContext,'<span class="intro">')
filterContext =SilverMetalNetSpiderUtils.filterContextByTarget(startContext,'', '</span>')
filterContext =SilverMetalNetSpiderUtils.removeSpecialCharacter(filterContext)
return filterContext