public DocInfo ParseInfoDoc(HtmlDocument doc, string url)
{
DocInfo info = new DocInfo();
var trNodes = doc.DocumentNode.SelectNodes("//table[@class='tableproperties']//tr");
if (trNodes != null)
{
foreach (var VARIABLE in trNodes)
{
var tdNodes = VARIABLE.SelectNodes("./td");
if (tdNodes != null)
{
for (int i = 0; i < tdNodes.Count; i++)
{
if (tdNodes[i].Attributes.Contains("class") &&
tdNodes[i].GetAttributeValue("class", "") == "headerproperties")
{
if (tdNodes[i + 1].GetAttributeValue("class", "") == "contentproperties")
{
string file = tdNodes[i].InnerText.Trim().ToLower();
string properties = tdNodes[i + 1].InnerText.Trim();
ParseData(file, properties, info);
}
}
}
}
}
return(info);
}
else
{
return(null);
}
}
public void UpdateDocData(DocInfo docInfo)
{
string query = @"Update Documents Set
DateEnable = @DateEnable,
DateNoEnable = @DateNoEnable,
DatePost = @DatePost,
DatePublish = @DatePublish,
DateUsed = @DateUsed,
DocNumber = @DocNumber,
DocRef = @DocRef,
Enable = @Enable,
PartNoEnable = @PartNoEnable,
ReasonNoEnable = @ReasonNoEnable,
Source = @Source,
TypeDoc = @TypeDoc,
DocFrom = @DocFrom,
Scope = @Scope,
LastUpdate = Getdate()
Where Id = @Id";
this.sqlDb.RunQuery(query, CommandType.Text, new SqlParameter[]
{
SqlDb.CreateParamteterSQL("Id", docInfo.Id, SqlDbType.BigInt),
SqlDb.CreateParamteterSQL("DateEnable", docInfo.DateEnable, SqlDbType.NVarChar),
SqlDb.CreateParamteterSQL("DateNoEnable", docInfo.DateNoEnable, SqlDbType.NVarChar),
SqlDb.CreateParamteterSQL("DatePost", docInfo.DatePost, SqlDbType.NVarChar),
SqlDb.CreateParamteterSQL("DatePublish", docInfo.DatePublish, SqlDbType.NVarChar),
SqlDb.CreateParamteterSQL("DateUsed", docInfo.DateUsed, SqlDbType.NVarChar),
SqlDb.CreateParamteterSQL("DocNumber", docInfo.DocNumber, SqlDbType.NVarChar),
SqlDb.CreateParamteterSQL("DocRef", docInfo.DocRef, SqlDbType.NVarChar),
SqlDb.CreateParamteterSQL("Enable", docInfo.Enable, SqlDbType.NVarChar),
SqlDb.CreateParamteterSQL("PartNoEnable", docInfo.PartNoEnable, SqlDbType.NVarChar),
SqlDb.CreateParamteterSQL("ReasonNoEnable", docInfo.ReasonNoEnable, SqlDbType.NVarChar),
SqlDb.CreateParamteterSQL("Source", docInfo.Source, SqlDbType.NVarChar),
SqlDb.CreateParamteterSQL("TypeDoc", docInfo.TypeDoc, SqlDbType.NVarChar),
SqlDb.CreateParamteterSQL("DocFrom", docInfo.DocFrom, SqlDbType.NVarChar),
SqlDb.CreateParamteterSQL("Scope", docInfo.Scope, SqlDbType.NVarChar),
});
}
private static void Main(string[] args)
{
string str = "1. PushLinkParse. 2. RunWorkerProcess. 3.PushDocInfo. 4.WorkerDocInfo";
Console.WriteLine(str);
int a = Convert.ToInt32(Console.ReadLine());
switch (a)
{
case 1: Test t = new Test();
t.PushQueueAs();
break;
case 3:
Test t1 = new Test();
t1.PushQueueAsDocInfo();
break;
case 2:
var v = new WorkerCrawler();
v.StartConsume();
break;
case 4:
var v1 = new WorkerDocInfo();
v1.StartConsume();
break;
}
return;
//var v = new WorkerCrawler();
//v.StartConsume();
//return;
//// Test t = new Test();
//// t.PushQueueAs();
DocManAdapter docManAdapter = new DocManAdapter();
string url = @"http://moj.gov.vn/vbpq/Lists/Vn%20bn%20php%20lut/View_Detail.aspx?ItemID=30517";
url = @"http://moj.gov.vn/vbpq/Lists/Vn%20bn%20php%20lut/View_Detail.aspx?ItemID=6527";
//url = @"http://moj.gov.vn/vbpq/Lists/Vn%20bn%20php%20lut/View_Detail.aspx?ItemID=20516#Chuong_I";
url = @"http://moj.gov.vn/vbpq/Pages/View_Propertes.aspx?ItemID=8070";
string html = System.Web.HttpUtility.HtmlDecode(GABIZ.Base.HtmlUrl.HTMLTransmitter.getHTML(url, 42, 2));
HtmlDocument htmlDocument = new HtmlDocument();
html = Common.ChuanHoaTextOfHtml(html);
htmlDocument.LoadHtml(html);
htmlDocument.DocumentNode.Descendants()
.Where(n => n.Name == "script" || n.Name == "style")
.ToList()
.ForEach(n => n.Remove());
Documet document = new Documet();
//string urlInfo = @"http://moj.gov.vn/vbpq/Pages/View_Propertes.aspx?ItemID=3001";
//HtmlDocument htmlDocumentInfo = new HtmlDocument();
//htmlDocumentInfo.LoadHtml(GABIZ.Base.HtmlUrl.HTMLTransmitter.getHTML(url, 42, 2));
ParserData p = new ParserData();
DocInfo docInfo = new DocInfo();
//p.Parse(ref document, htmlDocument, url);
DocInfo di = p.ParseInfoDoc(htmlDocument, url);
Console.Write(di);
if (document.IsValidData())
{
docManAdapter.InsertData(document);
}
}
public void ParseData(string nameProperties, string valueProperties, DocInfo docinfo)
{
if (nameProperties == "số, ký hiệu")
{
docinfo.DocNumber = valueProperties;
}
else if (nameProperties == "ngày ban hành")
{
docinfo.DatePublish = valueProperties;
}
else if (nameProperties == "loại văn bản")
{
docinfo.TypeDoc = valueProperties;
}
else if (nameProperties == "nguồn trích")
{
docinfo.Source = valueProperties;
}
else if (nameProperties == "phạm vi")
{
docinfo.Scope = valueProperties;
}
else if (nameProperties == "ngày đăng công báo")
{
docinfo.DatePost = valueProperties;
}
else if (nameProperties == "tình trạng hiệu lực")
{
docinfo.Enable = valueProperties;
}
else if (nameProperties == "ngày có hiệu lực")
{
docinfo.DateEnable = valueProperties;
}
else if (nameProperties == "ngày hết hiệu lực")
{
docinfo.DateNoEnable = valueProperties;
}
else if (nameProperties == "lí do hết hiệu lực")
{
docinfo.ReasonNoEnable = valueProperties;
}
else if (nameProperties == "phần hết hiệu lực")
{
docinfo.PartNoEnable = valueProperties;
}
else if (nameProperties == "ngày áp dụng")
{
docinfo.DateUsed = valueProperties;
}
else if (nameProperties == "văn bản dẫn chiếu")
{
docinfo.DocRef = valueProperties;
}
else if (nameProperties == "văn bản căn cứ")
{
docinfo.DocFrom = valueProperties;
}
else if (nameProperties == "văn bản bị thay thế")
{
docinfo.DocIsReplate = valueProperties;
}
else if (nameProperties == "văn bản bị sửa đổi bổ sung")
{
docinfo.DocIsAlter = valueProperties;
}
else
{
logInfo.Info(nameProperties);
}
}