def gethtml(url,**dict_args):
try:
html = get_html_urllib(url, 2)
except Exception,e:
html = 'timeout'
def get_dict_info(self,hc,**dict_args):
''' 冲详细信息页面的代码中获取数据 {} '''
fun = 'function get_dict_info of %s' % self.name
dict_info = {}
hc = filter_html(hc)
''' 此时 hc 是否为详细信息页面的代码 '''
iip = dict_args.get('is_info_page', False)
if iip:
''' 处在详细信息页面 '''
res_imgurl = p_imgurl_iip.findall(hc)
res_partno = p_partno_iip.findall(hc)
res_mfr = p_mfr_iip.findall(hc)
res_desc = p_desc_iip.findall(hc)
res_stock = p_stock_iip.findall(hc)
else:
res_imgurl = p_imgurl.findall(hc)
res_partno = p_partno.findall(hc)
res_mfr = p_mfr.findall(hc)
res_desc = p_desc.findall(hc)
res_stock = p_stock.findall(hc)
if res_imgurl:
dict_info[keys_mouser_tt[28][1]] = p_jkh.sub('',res_imgurl[0]).replace('"','').strip()
if res_partno:
dict_info[keys_mouser_tt[1][1]] = p_jkh.sub('',res_partno[0]).strip()
if res_mfr:
dict_info[keys_mouser_tt[2][1]] = p_jkh.sub('',res_mfr[0]).strip()
if res_desc:
dict_info[keys_mouser_tt[3][1]] = p_jkh.sub('',res_desc[0]).strip()
if res_stock:
dict_info[keys_mouser_tt[5][1]] = p_jkh.sub('',res_stock[0]).strip()
''' 处理区间价格 [('1-24','$82.08'),.('25 +','$61.56')] >>> ['1:$82.08','1000:$61.56'] >>> '1:$82.08|||25:$61.56' '''
if iip:
''' 详细信息页面 '''
res_priceinfo = p_priceinfo_iip.findall(hc)
pc_pp_new = ['%s:%s' % (p_invalidnum.sub('',pc).strip(), pp) for pc,pp in res_priceinfo]
pc_pp_finall = '|||'.join(pc_pp_new)
else:
''' 现在为表格页面模式 获取该<tr>对应的详细信息页面url '''
res_seemore = p_seemore.findall(hc)
res_priceinfo = []#默认设置空
if res_seemore:
''' 需要进入详细信息页面获取完整价格 '''
res_uip = p_uip.findall(hc)
if res_uip:
dict_info[keys_mouser_tt[27][1]] = res_uip[0]
hc_more = filter_html(get_html_urllib(res_uip[0], 2))
if hc_more != 'timeout':
''' 正常获取了 详细信息页面的价格信息 '''
res_priceinfo = p_priceinfo_iip.findall(hc_more)
else:
''' 只需从<tr>行字符串获取价格 即可 '''
res_priceinfo = p_priceinfo.findall(hc)
pc_pp_new = ['%s:%s' % (p_invalidnum.sub('',pc).strip(), pp) for pc,pp in res_priceinfo]
pc_pp_finall = '|||'.join(pc_pp_new)
if pc_pp_finall:dict_info[keys_mouser_tt[10][1]] = pc_pp_finall
''' 加上型号搜索页面url 详细页面url 详细页面html代码 '''
dict_info[keys_mouser_tt[19][1]] = self.url_search
return dict_info
