public static List <DoubanHouseInfo> GetDataFromAPI(string groupID, string cityName, int pageIndex)
{
List <DoubanHouseInfo> lstHouseInfo = new List <DoubanHouseInfo>();
var apiURL = $"https://api.douban.com/v2/group/{groupID}/topics?start={pageIndex * 50}";
var doubanTopic = WebAPIHelper.GetAPIResult <DoubanTopic>(apiURL);
if (doubanTopic != null && doubanTopic.topics != null)
{
foreach (var topic in doubanTopic.topics)
{
if (DataContent.DoubanHouseInfos.Any(h => h.HouseOnlineURL == topic.share_url))
{
continue;
}
var housePrice = JiebaTools.GetHousePrice(topic.content);
var house = new DoubanHouseInfo()
{
HouseLocation = topic.title,
HouseTitle = topic.title,
HouseOnlineURL = topic.share_url,
HouseText = topic.content,
HousePrice = JiebaTools.GetHousePrice(topic.content),
IsAnalyzed = true,
DisPlayPrice = housePrice > 0 ? $"{housePrice}元" : "",
Source = ConstConfigurationName.Douban,
LocationCityName = cityName,
Status = 1,
PubTime = DateTime.Parse(topic.created),
DataCreateTime = DateTime.Now,
};
lstHouseInfo.Add(house);
}
}
return(lstHouseInfo);
}
public override List <DBHouse> ParseHouses(DBConfig config, string data)
{
var houses = new List <DBHouse>();
var resultJObject = JsonConvert.DeserializeObject <JObject>(data);
var city = config.City;
foreach (var topic in resultJObject["topics"])
{
var housePrice = JiebaTools.GetHousePrice(topic["content"].ToString());
var photos = topic["photos"]?.Select(photo => photo["alt"].ToString()).ToList();
var house = new DBHouse()
{
Id = Tools.GetUUId(),
Location = topic["title"].ToString(),
Title = topic["title"].ToString(),
OnlineURL = topic["share_url"].ToString(),
Text = topic["content"].ToString(),
JsonData = topic.ToString(),
Price = (int)housePrice,
Source = SourceEnum.Douban.GetSourceName(),
City = city,
RentType = GetRentType(topic["content"].ToString()),
PicURLs = JsonConvert.SerializeObject(photos),
PubTime = topic["created"].ToObject <DateTime>(),
};
houses.Add(house);
}
return(houses);
}
public override List <BaseHouseInfo> ParseHouses(JToken config, string data)
{
var houses = new List <BaseHouseInfo>();
var resultJObject = JsonConvert.DeserializeObject <JObject>(data);
var cityName = config["cityname"].ToString();
foreach (var topic in resultJObject["topics"])
{
var housePrice = JiebaTools.GetHousePrice(topic["content"].ToString());
var photos = topic["photos"]?.Select(photo => photo["alt"].ToString()).ToList();
var house = new BaseHouseInfo()
{
HouseLocation = topic["title"].ToString(),
HouseTitle = topic["title"].ToString(),
HouseOnlineURL = topic["share_url"].ToString(),
HouseText = topic["content"].ToString(),
HousePrice = housePrice,
IsAnalyzed = true,
DisPlayPrice = housePrice > 0 ? $"{housePrice}元" : "",
Source = ConstConfigName.Douban,
LocationCityName = cityName,
Status = 1,
PicURLs = JsonConvert.SerializeObject(photos),
PubTime = topic["created"].ToObject <DateTime>()
};
houses.Add(house);
}
return(houses);
}
private static int GetHousePrice(DBHouse house)
{
var price = JiebaTools.GetHousePrice(house.Title);
if (price == 0 && !string.IsNullOrEmpty(house.Text))
{
price = JiebaTools.GetHousePrice(house.Text);
}
return(price);
}
public static void AnalyzeDoubanHouseContent()
{
LogHelper.Info("AnalyzeDoubanHouseContent Start...");
int index = 0;
try
{
var lstHouse = dataContent.HouseInfos.Where(h =>
h.Source == ConstConfigurationName.Douban && h.IsAnalyzed == false).Take(100).ToList();
foreach (var houseInfo in lstHouse)
{
var housePrice = JiebaTools.GetHousePrice(houseInfo.HouseText);
string houseTextContent = string.Empty;
if (housePrice == 0)
{
var htmlResult = HTTPHelper.GetHTML(houseInfo.HouseOnlineURL);
if (string.IsNullOrEmpty(htmlResult))
{
continue;
}
var page = htmlParser.Parse(htmlResult);
var topicContent = page.QuerySelector("div.topic-content");
if (topicContent == null)
{
continue;
}
var houseDescription = topicContent.QuerySelector("p");
if (houseDescription == null)
{
continue;
}
houseTextContent = houseDescription.TextContent;
housePrice = JiebaTools.GetHousePrice(houseDescription.TextContent);
}
if (housePrice != 0 || !string.IsNullOrEmpty(houseTextContent))
{
index++;
houseInfo.IsAnalyzed = true;
}
houseInfo.HouseText = houseTextContent;
houseInfo.HousePrice = housePrice;
}
dataContent.SaveChanges();
}
catch (Exception ex)
{
LogHelper.Error("AnalyzeDoubanHouseContent Exception", ex);
}
LogHelper.Info("AnalyzeDoubanHouseContent Finish,Update Count:" + index);
}
public static void AnalyzeDoubanHouseContentAll(bool isSleep = false)
{
System.Diagnostics.Stopwatch sw = new System.Diagnostics.Stopwatch();
sw.Start();
LogHelper.Info("AnalyzeDoubanHouseContent Start...");
Console.WriteLine("AnalyzeDoubanHouseContent Start...");
int index = 0;
try
{
var dal = new DBHouseInfoDAL();
var lstHouse = dal.LoadUnAnalyzeList();
foreach (var houseInfo in lstHouse)
{
var housePrice = JiebaTools.GetHousePrice(houseInfo.HouseText);
string houseTextContent = string.Empty;
if (housePrice == 0)
{
AnalyzeFromWebPage(houseInfo, ref housePrice, ref houseTextContent);
}
else
{
houseInfo.Status = 1;
houseInfo.DisPlayPrice = housePrice.ToString(CultureInfo.InvariantCulture);
houseInfo.HousePrice = housePrice;
}
houseInfo.IsAnalyzed = true;
dal.UpdateHouseInfo(houseInfo);
index++;
if (index % 100 == 0 && isSleep)
{
System.Threading.Thread.Sleep(1000 * 120);
}
Console.WriteLine("HouseInfo:" + Newtonsoft.Json.JsonConvert.SerializeObject(houseInfo));
}
}
catch (Exception ex)
{
LogHelper.Error("AnalyzeDoubanHouseContent Exception", ex);
}
sw.Stop();
var copyTime = sw.Elapsed.TotalSeconds.ToString(CultureInfo.InvariantCulture);
LogHelper.Info("AnalyzeDoubanHouseContent Finish,Update Count:" + index + ";花费时间：" + copyTime);
Console.WriteLine("AnalyzeDoubanHouseContent Finish,Update Count:" + index + ";花费时间：" + copyTime);
}
private static List <BizHouseInfo> GetDataFromOnlineWeb(string groupID, string cityName, int pageIndex)
{
List <BizHouseInfo> lstHouseInfo = new List <BizHouseInfo>();
var url = $"https://www.douban.com/group/{groupID}/discussion?start={pageIndex * 25}";
var htmlResult = DoubanHTTPHelper.GetHTMLForDouban(url);
if (string.IsNullOrEmpty(htmlResult))
{
return(lstHouseInfo);
}
var page = HtmlParser.Parse(htmlResult);
var tableElement = page.QuerySelector("table.olt");
if (tableElement == null)
{
return(lstHouseInfo);
}
foreach (var trItem in tableElement.QuerySelectorAll("tr"))
{
var titleItem = trItem.QuerySelector("td.title");
if (titleItem == null || DataContent.HouseInfos.Find(titleItem.QuerySelector("a").GetAttribute("href")) != null)
{
continue;
}
var houseTitle = titleItem.QuerySelector("a").GetAttribute("title");
var housePrice = JiebaTools.GetHousePrice(houseTitle);
var houseInfo = new BizHouseInfo()
{
HouseTitle = houseTitle,
HouseOnlineURL = titleItem.QuerySelector("a").GetAttribute("href"),
HouseLocation = houseTitle,
HouseText = houseTitle,
DataCreateTime = DateTime.Now,
PubTime = titleItem.QuerySelector("td.time") != null
? DateTime.Parse(DateTime.Now.ToString("yyyy-") + titleItem.QuerySelector("td.time").InnerHtml)
: DateTime.Now,
DisPlayPrice = housePrice > 0 ? $"{housePrice}元":"",
Source = ConstConfigurationName.Douban,
HousePrice = housePrice,
LocationCityName = cityName,
IsAnalyzed = housePrice > 0,
Status = housePrice > 0 ? 1 : 0,
};
lstHouseInfo.Add(houseInfo);
}
return(lstHouseInfo);
}
public override List <DBHouse> ParseHouses(DBConfig config, string data)
{
var houses = new List <DBHouse>();
var jsonData = JToken.Parse(config.Json);
var city = config.City;
var htmlDoc = htmlParser.Parse(data);
var topics = htmlDoc?.QuerySelector("#TopicsNode")?.QuerySelectorAll("div");
if (topics == null || topics.Count() == 0)
{
return(houses);
}
foreach (var topic in topics)
{
var title = topic.QuerySelector("span.item_title")?.QuerySelector("a");
if (CheckTopic(title))
{
var house = new DBHouse();
house.Title = title?.TextContent;
house.Location = title?.TextContent;
var path = title.GetAttribute("href");
if (path.Contains("#"))
{
path = path.Split("#").First();
}
house.OnlineURL = $"https://www.v2ex.com{path}";
house.JsonData = Newtonsoft.Json.JsonConvert.SerializeObject(new { html = topic.OuterHtml });
house.Source = SourceEnum.V2ex.GetSourceName();
house.Price = JiebaTools.GetHousePrice(title?.TextContent);
house.Id = Tools.GetGuid();
house.City = city;
house.PicURLs = Tools.GetPicURLs("");
house.PubTime = DateTime.Now;
houses.Add(house);
}
}
FillGoodHouseLocation("478aea7d5ba0cfb604106db0a33c7119", city, houses);
return(houses);
}
public static List <BaseHouseInfo> GetHouseData(string groupID, string cityName, int pageIndex)
{
List <BaseHouseInfo> lstHouseInfo = new List <BaseHouseInfo>();
var apiURL = $"https://api.douban.com/v2/group/{groupID}/topics?start={pageIndex * 50}&count=50";
LogHelper.Debug($"url:{apiURL},groupID:{groupID}, city:{cityName}");
var result = GetAPIResult(apiURL);
if (string.IsNullOrEmpty(result))
{
return(lstHouseInfo);
}
var resultJObject = JsonConvert.DeserializeObject <JObject>(result);
foreach (var topic in resultJObject["topics"])
{
var housePrice = JiebaTools.GetHousePrice(topic["content"].ToString());
var photos = topic["photos"]?.Select(photo => photo["alt"].ToString()).ToList();
var house = new BaseHouseInfo()
{
HouseLocation = topic["title"].ToString(),
HouseTitle = topic["title"].ToString(),
HouseOnlineURL = topic["share_url"].ToString(),
HouseText = topic["content"].ToString(),
HousePrice = housePrice,
IsAnalyzed = true,
DisPlayPrice = housePrice > 0 ? $"{housePrice}元" : "",
Source = ConstConfigName.Douban,
LocationCityName = cityName,
Status = 1,
PicURLs = JsonConvert.SerializeObject(photos),
PubTime = topic["created"].ToObject <DateTime>()
};
lstHouseInfo.Add(house);
}
return(lstHouseInfo);
}
private static void AnalyzeFromWebPage(Web.Model.DBHouseInfo houseInfo,
ref decimal housePrice, ref string houseTextContent)
{
var htmlResult = DoubanHTTPHelper.GetHTMLForDouban(houseInfo.HouseOnlineURL);
//没有页面信息
if (string.IsNullOrEmpty(htmlResult))
{
//404页面
houseInfo.Status = 2;
}
else
{
var page = HtmlParser.Parse(htmlResult);
var topicContent = page.QuerySelector("div.topic-content");
//没有帖子内容
if (topicContent == null || topicContent.QuerySelector("p") == null || topicContent.QuerySelector("p") == null)
{
houseInfo.Status = 3;
}
else
{
//获取帖子内容
houseTextContent = topicContent.QuerySelector("p").TextContent;
//获取价格信息
housePrice = JiebaTools.GetHousePrice(houseTextContent);
if (housePrice != 0 || !string.IsNullOrEmpty(houseTextContent))
{
houseInfo.Status = 1;
}
houseInfo.DisPlayPrice = housePrice.ToString(CultureInfo.InvariantCulture);
houseInfo.HousePrice = housePrice;
houseInfo.HouseText = houseTextContent;
}
}
}
