/// <summary>
/// 转发消息到从节点
/// </summary>
/// <param name="slaveNode"></param>
/// <param name="soaMsg"></param>
/// <returns></returns>
private IDataContainer TransferMsgToSlave(PeekerClusterNode slaveNode, SoapMessage soaMsg)
{
IDataContainer result = null;
try
{
//string address = "127.0.0.1";
using (var conn = new SoapTcpConnection(slaveNode.IpAddress, slaveNode.Port))
{
if (conn.State == ConnectionState.Closed)
{
conn.Open();
}
//发送soap
result = conn.SendSoapMessage(soaMsg);
}
}
catch (Exception ex)
{
Logger.Error(ex);
}
return(result);
}
/// <summary>
/// 向服务端发送消息，注册登记
/// 并分配端口
/// </summary>
/// <param name="cmd"></param>
/// <param name="data"></param>
public static int RegisterSlaveToMaster(string slaveIdentity)
{
var result = -1;
using (var conn = new SoapTcpConnection("127.0.0.1", GlobalContext.MasterSocketPort))
{
if (conn.State == ConnectionState.Closed)
{
conn.Open();
}
if (conn.Ping() == false)
{
return(result);
}
//发送soap
var paras = new RegisterPortArgument {
SlaveIdentity = slaveIdentity
};
string msg = JsonConvert.SerializeObject(paras);
SoapMessage sopMsg = new SoapMessage()
{
Head = CommandConstants.CMD_RegisterSlavePort,
Body = msg
};
var repResult = conn.SendSoapMessage(sopMsg);
if (repResult.Status == 1)
{
result = repResult.Result.ToInt();
}
}
return(result);
}
/// <summary>
/// 根据关键词 ,筛选条件， 请求对应平台上的返回结果
/// 结果是各自平台上的 商品Item列表
/// </summary>
/// <param name="webArgs"></param>
/// <returns></returns>
public SearchProductViewModel QueryProductsByKeyWords(BaseFetchWebPageArgument webArgs)
{
SearchProductViewModel dataModel = new SearchProductViewModel();
if (webArgs.IsValid() == false)
{
return(dataModel);
}
try
{
//注册搜索词到热词服务
HotWordService.AddWord(webArgs.KeyWord);
//是否开启内容缓存，如果开启，那么从缓存中加载内容
if (true == WorkContext.IsFetchPageCacheaAble)
{
dataModel = WorkContext.GetFetchPageResultFromCache(webArgs);
if (null != dataModel)
{
return(dataModel);
}
}
//工厂模式 获取指定平台的内容解析器
var resolver = ResolverFactory.GetSearchProductResolver(webArgs.Platform);
//尝试解析页面参数的检索地址
var searchUrl = resolver.ResolveSearchUrl(webArgs);
if (null != searchUrl)
{
webArgs.ResolvedUrl = searchUrl;
}
string pageContent = string.Empty;
using (var connMgr = new WebCrawlerConnConfigManager())
{
var connStrConfig = connMgr.Connection;
//;//ConfigHelper.WebCrawlerSection.ConnectionStringCollection["Crawler-Server1"];
webArgs.SystemAttachParas["SoapTcpConnectionString"] = connStrConfig;//register to attach paras
if (searchUrl.IsNeedPreRequest == true)
{
////1 打开tcp 链接
////2 发送参数
////3 解析结果
using (var conn = new SoapTcpConnection(connStrConfig))
{
if (conn.State == ConnectionState.Closed)
{
conn.Open();
}
//发送soap
var soapCmd = new SoapMessage()
{
Head = CommandConstants.CMD_FetchPage
};
soapCmd.Body = webArgs.ToJson();
var dataContainer = conn.SendSoapMessage(soapCmd);
if (null != dataContainer && dataContainer.Status == 1)
{
pageContent = dataContainer.Result;
}
else
{
StringBuilder errMsg = new StringBuilder("抓取网页请求失败！参数：");
errMsg.Append(soapCmd.Body);
if (null != dataContainer && !string.IsNullOrEmpty(dataContainer.ErrorMsg))
{
errMsg.Append("；服务端错误消息：")
.Append(dataContainer.ErrorMsg);
}
throw new Exception(errMsg.ToString());
}
}
}
}
//开始解析内容字符串
//*******注意：针对可以直接进行内容解析的连接，交给内容解析函数进行地址的内容请求和解析*********
if (!string.IsNullOrEmpty(pageContent) || !searchUrl.IsNeedPreRequest)
{
dataModel = resolver.ResolvePageContent(webArgs, pageContent);
if (null != dataModel)
{
dataModel.KeyWord = webArgs.KeyWord;
dataModel.IsNeedResolveHeaderTags = webArgs.IsNeedResolveHeaderTags;
}
}
}
catch (Exception ex)
{
Logger.Error(ex);
}
//如果开启缓存页面结果
if (true == WorkContext.IsFetchPageCacheaAble &&
null != dataModel &&
dataModel.Products.IsNotEmpty())
{
int cacheTime = ConfigHelper.AppSettingsConfiguration.GetConfigInt("FetchPageCacheTime");
if (cacheTime <= 0)
{
cacheTime = 60;//默认缓存页面结果60秒
}
WorkContext.SetFetchPageResultFromCache(webArgs, dataModel, cacheTime);
}
return(dataModel);
}
/// <summary>
/// 执行内容解析
/// </summary>
///<param name="webArgs"></param>
/// <param name="content">要解析的内容</param>
/// <returns>返回需要的字段对应的字典</returns>
public override Dictionary <string, object> ResolveSearchPageContent(BaseFetchWebPageArgument webArgs, string content)
{
var resultBag = new Dictionary <string, object>();
try
{
string jsonData = string.Empty;
if (content.IndexOf("g_page_config") < 0)
{
return(null);//无效的页面结果数据
}
//send request for load other data of first search page
Task <string> tskSilcedJsonpContent = null;
if (webArgs.PageIndex == 0)
{
tskSilcedJsonpContent = Task.Factory.StartNew(() =>
{
string jsonpContent = "";
////1 打开tcp 链接
////2 发送参数
////3 解析结果
if (!webArgs.SystemAttachParas.ContainsKey("SoapTcpConnectionString"))
{
return(jsonpContent);
}
var connStrConfig = webArgs.SystemAttachParas["SoapTcpConnectionString"] as WebCrawlerConnection;
if (null == connStrConfig)
{
return(jsonpContent);
}
//重写解析地址-首页的分片jsonp地址
string urlOfSlicedJsonp = this.ResolveSlicedSearchPageSilcedUrl(webArgs);
webArgs.ResolvedUrl = new ResolvedSearchUrlWithParas {
Url = urlOfSlicedJsonp
};
using (var conn = new SoapTcpConnection(connStrConfig))
{
if (conn.State == ConnectionState.Closed)
{
conn.Open();
}
//发送soap
var soapCmd = new SoapMessage()
{
Head = CommandConstants.CMD_FetchPage
};
soapCmd.Body = JsonConvert.SerializeObject(webArgs);
var dataContainer = conn.SendSoapMessage(soapCmd);
if (null != dataContainer && dataContainer.Status == 1)
{
jsonpContent = dataContainer.Result;
}
else
{
StringBuilder errMsg = new StringBuilder("抓取网页请求失败！参数：");
errMsg.Append(soapCmd.Body);
if (null != dataContainer && !string.IsNullOrEmpty(dataContainer.ErrorMsg))
{
errMsg.Append("；服务端错误消息：")
.Append(dataContainer.ErrorMsg);
}
PluginContext.Logger.Error(errMsg.ToString());
}
}
return(jsonpContent);
});
}
int startPos = content.IndexOf("g_page_config");
int endPos = content.IndexOf("g_srp_loadCss") - startPos;
var secondContent = content.Substring(startPos, endPos);
int secStartPos = secondContent.IndexOf('{');
int secEndPos = secondContent.IndexOf("};") - secStartPos + 1;
jsonData = secondContent.Substring(secStartPos, secEndPos);
TaobaoPageJsonResut pageJsonObj = JsonConvert.DeserializeObject <TaobaoPageJsonResut>(jsonData);
if (null == pageJsonObj)
{
return(null);
}
if (webArgs.IsNeedResolveHeaderTags == true)
{
var navNode = pageJsonObj.mods.nav;
if (null != navNode && null != navNode.data)
{
var commonNode = navNode.data.common;
var advNode = navNode.data.adv;
//解析common节点
if (null != commonNode && commonNode.Any())
{
//1 检测是否有品牌，有的话 解析品牌
#region 品牌解析
var brandNode = commonNode.FirstOrDefault(x => x.text == "品牌" && x.sub != null);
if (null != brandNode && brandNode.sub != null)
{
var lstBrands = new List <BrandTag>();
foreach (var subItem in brandNode.sub)
{
var model = new BrandTag();
model.Platform = SupportPlatformEnum.Taobao;
model.FilterField = "ppath";//使用的过滤字段参数
model.BrandId = subItem.value;
model.BrandName = subItem.text;
model.CharIndex = PinYin.GetFirstLetter(model.BrandName);
lstBrands.Add(model);
}
//解析完毕品牌
resultBag.Add("Brands", lstBrands);
}
#endregion
}
//2其他筛选节点的分析
#region tags 解析
var lstTags = new List <KeyWordTagGroup>();
var otherFilterNode1 = commonNode.Where(x => x.text != "品牌" && x.sub != null);
foreach (var itemNode in otherFilterNode1)
{
//找到归属的组
string groupName = itemNode.text;
ProcessTags(lstTags, itemNode.sub, groupName);
}
////////if (null!= advNode)----高级筛选不要了
////////{
//////// //advNode 的解析
//////// foreach (var itemNode in advNode)
//////// {
//////// //找到归属的组
//////// string groupName = itemNode.text;
//////// ProcessTags(lstTags, itemNode.sub, groupName);
//////// }
////////}
resultBag.Add("Tags", lstTags);
#endregion
}
}
#region products 解析
var lstProducts = new ProductBaseCollection();
resultBag.Add("Products", lstProducts);
var itemListNode = pageJsonObj.mods.itemlist;
if (null != itemListNode && itemListNode.data != null && null != itemListNode.data.auctions)
{
foreach (var itemProduct in itemListNode.data.auctions)
{
TaobaoProduct modelProduct = this.ResolverProductDom(itemProduct);
if (null != modelProduct)
{
lstProducts.Add(modelProduct);
}
}
}
//淘宝的搜索列表 - 第一页的数据是进行了分片的，在加载html ；36条数据， 后续会进行一次jsonp的请求；加载12条数据
if (webArgs.PageIndex == 0 && null != tskSilcedJsonpContent)
{
string jsonpContent = tskSilcedJsonpContent.Result;
if (!string.IsNullOrEmpty(jsonpContent) && jsonpContent.Contains("API.CustomizedApi"))
{
int startIdx = jsonpContent.IndexOf(':') + 1;
int endIdx = jsonpContent.Length - startIdx - 3;
string pureJsonContent = jsonpContent.Substring(startIdx, endIdx);
var slicedJsonpResut = JsonConvert.DeserializeObject <TaobaoSlicedJsonpResut>(pureJsonContent);
if (null != slicedJsonpResut)
{
var itemList = slicedJsonpResut.itemlist;
if (null != itemList && itemList.auctions != null)
{
foreach (var itemProduct in itemList.auctions)
{
TaobaoProduct modelProduct = this.ResolverProductDom(itemProduct);
if (null != modelProduct)
{
lstProducts.Add(modelProduct);
}
}
}
}
}
}
#endregion
}
catch (Exception ex)
{
PluginContext.Logger.Error(ex);
}
return(resultBag);// string.Concat("has process input :" + content);
}