def parseHYDetailsMeta(self, text, id):
key_value_dic = GateHuoYuanSource.get_k_v_dic()
"""
{
u"货源名称": "sourceName",
u"货源地址": "sourceAddress",
u"货物名称": "startPlace",
u"货物类型": "goodsType",
u"重量": "weight",
u"数量": "number",
u"外形": "appearance",
u"体积": "volume",
u"所需车型": "vehicleDependent",
u"运送方式": "shippingMethod",
u"运费": "freight",
u"包装方式": "packingManer",
u"注意事项": "attention",
u"货物情况": "cargoCondition",
u"最早起运": "earliestShipment",
u"最晚起运": "latestShipment",
u"是否长期货源": "isLongTerm",
u"出发地": "departure",
u"目的地": "destPlace",
u"发布时间": "postTime",
u"webSiteId": "webSiteId",
u"出发地联系人": "sourceContact",
u"目的地联系人": "destContact",
u"补充说明": "remark",
}
"""
result = {}
soup = BeautifulSoup(text)
soup.prettify()
sourceDetailinfo = soup.find(attrs={"class": "detail_top_content_1"})
sourceName = sourceDetailinfo.findAll("li")[0].find("span").fetchText(text=True)[0].replace("\t", "").strip(
"\t")
sourceName = "".join(sourceName.split())
sourceAdress = sourceDetailinfo.findAll("li")[1].fetchText(text=True)[1].strip()
remark = soup.find(attrs={"class": "detail_left_content_1_4"}).find("li").fetchText(text=True)
if (len(remark) > 1):
remark = "".join(remark[1:])
else:
remark = None
result["remark"] = remark
result["sourceName"] = sourceName
result["sourceAddress"] = sourceAdress
result["webSiteId"] = id
sourceInfo = soup.findAll(attrs={"class": "detail_left_content_1_1"})
infos = list(sourceInfo[0].findAll("li"))
infos.extend(sourceInfo[1].findAll("li"))
allInfo = {}
for info in infos:
key = info.fetchText(text=True)[0].replace("：", "")
if key in ["最早起运", "最晚起运"]:
value = "".join(info.fetchText(text=True))
else:
value = "".join(info.fetchText(text=True))
allInfo[key] = value
for k, v in allInfo.iteritems():
result[key_value_dic.get(k)] = v
contact = self.parseHYContact(id)
result["source"] = contact[0]
result["dest"] = contact[1]
return result
try:
GateHuoYuanSource.objects.get(webSiteId=id)
logger.info("HuoYuan %s already exsists" % id)
return
except:
pass
text = self.httpClient.geturlcon(url)
result = {}
try:
result = self.parseHYDetailsMeta(text, id)
except Exception, e:
traceback.print_exc()
logger.error(e)
if True:
source = GateHuoYuanSource()
try:
for k, v in result.iteritems():
logger.info("gateee " + str(k) + " " + str(v))
source.save_from_a_source(result)
except Exception, e:
traceback.print_exc()
logger.error(e)
else:
logger.info("%s saved" % (id))
else:
logger.error(url + " is a null page")
def getAndSaveCYFromId(self, id):
url = "http://56gate.com/html/cy/cydetail_%s.html" % id
try: