Example #1
0
def crawStockAccountDataCenter(link, webNet):
    currentList = []
    target = '<tbody>'
    startContext = StockAccountDataNetSpiderUtils.returnStartContext(
        link, target)
    startContext = StockAccountDataNetSpiderUtils.filterContextByTarget(
        startContext, '<tbody>', '</tbody>')
    startContext = StockAccountDataNetSpiderUtils.filterAfterContext(
        startContext, '</tr>')
    startContext = StockAccountDataNetSpiderUtils.filterAfterContext(
        startContext, '</tr>')
    len = StockAccountDataNetSpiderUtils.findAllTarget(startContext, '<tr')
    for i in range(len):
        targetContext = StockAccountDataNetSpiderUtils.divisionTarget(
            startContext, '<tr', '</tr>')
        startContext = targetContext['nextContext']
        currentContext = targetContext['targetContext']
        currentContext = StockAccountDataNetSpiderUtils.removeSpecialCharacter(
            currentContext)
        currentDate = StockAccountDataNetSpiderUtils.filterContextByTarget(
            currentContext, '<td>', '</td>').replace('.', '-')
        startDate = currentDate[0:10]
        endDate = currentDate[0:4] + '-' + currentDate[13:]
        crawUrl = webNet + StockAccountDataNetSpiderUtils.filterContextByTarget(
            currentContext, '<ahref="', '"target=')
        dataList = crawFinalDateSource(crawUrl, startDate, endDate)
        currentList.append(dataList)
    return currentList
def crawStockAccountDataCenter(link,webNet):
    currentList = []
    target ='<tbody>'
    startContext = StockAccountDataNetSpiderUtils.returnStartContext(link,target)
    startContext = StockAccountDataNetSpiderUtils.filterContextByTarget(startContext,'<tbody>','</tbody>')
    startContext = StockAccountDataNetSpiderUtils.filterAfterContext(startContext,'</tr>')
    startContext = StockAccountDataNetSpiderUtils.filterAfterContext(startContext,'</tr>')
    len = StockAccountDataNetSpiderUtils.findAllTarget(startContext,'<tr')
    for i in range(len):
        targetContext = StockAccountDataNetSpiderUtils.divisionTarget(startContext,'<tr','</tr>')
        startContext = targetContext['nextContext']
        currentContext =  targetContext['targetContext']
        currentContext = StockAccountDataNetSpiderUtils.removeSpecialCharacter(currentContext)
        currentDate = StockAccountDataNetSpiderUtils.filterContextByTarget(currentContext,'<td>','</td>').replace('.','-')
        startDate = currentDate[0:10]
        endDate = currentDate[0:4]+'-'+currentDate[13:]
        crawUrl = webNet+StockAccountDataNetSpiderUtils.filterContextByTarget(currentContext,'<ahref="','"target=')
        dataList = crawFinalDateSource(crawUrl,startDate,endDate)
        currentList.append(dataList)
    return currentList