googlealerts.py 文件源码

def createSpreadsheet():
      wb = xlwt.Workbook()
      sheet = wb.add_sheet("Google Alerts")
      style = xlwt.easyxf('font: bold 1')
      sheet.write(0, 3, 'Headline', style)
      sheet.write(0, 1, 'Company', style)
      sheet.write(0, 4, 'URL', style)
      sheet.write(0, 0, 'Date', style)

      cur_row = 1

      for url in LA_HONDA_ALERTS_URLS:
            print 'Processing google alerts for ' + LA_HONDA_ALERTS_URLS[url] + '...'
            r = requests.get(url)
            xml = r.text
            soup = BeautifulSoup(xml)

            for title, link, date in zip(soup.findAll('title')[1:], soup.findAll('link')[1:], soup.findAll('published')):
                  title = cleanTitle(title)
                  link = cleanLink(link)
                  date = cleanDate(date)

                  writeToSheet(sheet, title, LA_HONDA_ALERTS_URLS[url], link, date, cur_row)
                  cur_row = cur_row + 1

      processSheet(sheet)
      savewb(wb)