Python SinaBlogSpiderUtils.divisionTarget Exemples

Exemple #1

Fichier : SinaBlogSpider.py Projet : nicevoice/SpiderKTServer

def dailySinaBlog(link,id):
    startcontext = SinaBlogSpiderUtils.returnStartContext(link)
    blogList = []
    for  i in range(SinaBlogSpiderUtils.findAllTarget(startcontext)):
        targetContext = SinaBlogSpiderUtils.divisionTarget(startcontext, '<div class="articleCell SG_j_linedot1">', '</div>')
        startcontext = targetContext['nextContext']
        currentcontext =  targetContext['targetContext']
        title = SinaBlogSpiderUtils.filterContextByTarget(currentcontext, '<a title="', '" target=')
        linkUrl = SinaBlogSpiderUtils.filterContextByTarget(currentcontext, 'target="_blank" href="', '.html">')+'.html'
        pubDate = SinaBlogSpiderUtils.filtetContextExpertise(currentcontext,'<span class="atc_tm SG_txtc">','</span>')
        imageUrl = SinaBlogSpiderUtils.filtetContextExpertise(currentcontext, 'src="', '" width=')
        blogList.append([id,title,linkUrl,pubDate,imageUrl])
    return blogList

Exemple #2

Fichier : SinaBlogSpider.py Projet : wusezhangserver/kttspider

def dailySinaBlog(link,id):
    startcontext = SinaBlogSpiderUtils.returnStartContext(link)
    blogList = []
    for  i in range(SinaBlogSpiderUtils.findAllTarget(startcontext)):
        targetContext = SinaBlogSpiderUtils.divisionTarget(startcontext, '<div class="articleCell SG_j_linedot1">', '</div>')
        startcontext = targetContext['nextContext']
        currentcontext =  targetContext['targetContext']
        title = SinaBlogSpiderUtils.filterContextByTarget(currentcontext, '<a title="', '" target=')
        linkUrl = SinaBlogSpiderUtils.filterContextByTarget(currentcontext, 'target="_blank" href="', '.html">')+'.html'
        pubDate = SinaBlogSpiderUtils.filtetContextExpertise(currentcontext,'<span class="atc_tm SG_txtc">','</span>')
        imageUrl = SinaBlogSpiderUtils.filtetContextExpertise(currentcontext, 'src="', '" width=')
        blogList.append([id,title,linkUrl,pubDate,imageUrl])
    return blogList