public string FindFullNameClc(HtmlNode BodyNodeS, Fund fund)
{
bool previous = false;
string name = null;
HtmlNode tableFile = BodyNodeS.SelectSingleNode(".//table[@class='tableFile']");
if (tableFile != null)
{
foreach (HtmlNode tr in tableFile.SelectNodes(".//tr"))
{
HtmlNodeCollection tds = BodyNodeS.SelectNodes(".//td");
if (tds != null)
{
foreach (HtmlNode td in tds)
{
if (td.GetAttributeValue("class", "") == "seriesCell" && tools.ClearSpace(td.InnerText) != "")
{
name = td.InnerText;
previous = true;
}
if (previous == true)
{
if (td.GetAttributeValue("class", "") == "classContract" && name != null)
{
HtmlNode a = td.SelectSingleNode(".//a");
if (a != null && a.InnerText == fund.classcontract)
{
return(name);
}
}
}
}
}
}
}
if (fund.nameFromEdgar != null)
{
name = fund.nameFromEdgar;
}
return(name);
}
public void MakeFund(Fund fund)
{
//Console.WriteLine("Ticker: {0} - {1}", fund.ticker, _I);
_I++;
int currentNum = 1;
int _cNum = currentNum;
string urlForms = baseLink + "/cgi-bin/browse-edgar?CIK=" + fund.ticker + "&owner=exclude&action=getcompany&count=200&Find=Search";
//string urlForms = "http://www.sec.gov/cgi-bin/browse-edgar?company=blackrock+global+allocation&owner=exclude&action=getcompany";
// Console.WriteLine(urlForms);
//*Вход в edgar на страницу с отчетами для этого фонда
HtmlNode body = tools.BodyNodeS(tools.Load(urlForms), htmlMakeAll);
if (body == null)
{
Console.WriteLine("{0} - NOT FOUND IN EDGAR", fund.ticker);
fund.makeFailXml("NOT FOUND IN EDGAR", null, xmlPathR);
//errors.Error("0008");
fund.status = 1;
}
else
{
fund.classcontract = takeClassContract(body);
do
{
//*Ищем необходимую форму
string urlForm = FindNQ(fund, body, currentNum);
fund.companyName = FindCompanyName(body);
try
{
body = LoadFormAll(fund, body, urlForm, true);
}
catch (OutOfMemoryException e)
{
Console.WriteLine("{0} - OutOfMemory", fund.ticker);
}
currentNum++;
}while (currentNum <= numberForms);
}
}
//*Функция, которая ищет тиккеры, по которым появились новые отчеты
public List <Fund> selectUTickers(HtmlNode body)
{
List <Fund> tickers = new List <Fund>();
HtmlNodeCollection tds = body.SelectNodes(".//div[@id='seriesDiv']//tr[@class='contractRow']//td[position() = 4]");
if (tds != null)
{
foreach (HtmlNode td in tds)
{
var tval = td.InnerText.Trim().ToLower();
Fund fund = funds.Find(x => x.ticker.ToLower() == tval);
if (fund != null)
{
tickers.Add(fund);
}
}
}
return(tickers);
}
public void openFile(string _path, Fund fund = null, Func <string, Fund, string> lineMake = null)
{
string path = Directory.GetCurrentDirectory();
string fileName = String.Format(_path);
int counter = 0;
string line;
System.IO.StreamReader file = new System.IO.StreamReader(System.IO.Path.Combine(path, fileName));
while ((line = file.ReadLine()) != null)
{
if (lineMake != null)
{
lineMake(line, fund);
}
counter++;
}
file.Close();
System.Console.ReadLine();
}
//*Функция, которая ищет названия всех фондов и заполняет список заголовков
public void GetFullNames(HtmlNode BodyNodeS, Fund fund, List <string> names, List <Caption> captions)
{
HtmlNodeCollection nms = BodyNodeS.SelectNodes(".//td[@class='seriesCell'][normalize-space()!=' ']");
if (nms != null)
{
foreach (HtmlNode n in nms)
{
bool first = false;
if (fund.nameFromEdgar == n.InnerText)
{
first = true;
}
string captiontext = n.InnerText;
names.Add(n.InnerText);
//*Добавляем заголовок
captions.Add(new Caption(n.InnerText, tools.ClearSpaceCaption(captiontext).ToLower(), first));
}
}
}
//*Функция, которая обрезает одинаковую часть у всех заголовков
public void zNames(List <Caption> captions, Fund fund)
{
int i = zNamesFindI(captions, captions.First().compressednamenozip);
if (i == 0)
{
zNamesFindI2(captions, tools.ClearSpaceCaption(fund.companyName).ToLower());
}
foreach (Caption caption in captions)
{
if (caption.izip != 0)
{
caption.compressedname = caption.compressedname.Substring(caption.izip);
}
if (caption.compressedname.Length > 10)
{
caption.compressedname = Regex.Replace(caption.compressedname, @"fund$", @"", RegexOptions.IgnoreCase);
}
caption.compressedname = Regex.Replace(caption.compressedname, @"inc$", @"", RegexOptions.IgnoreCase);
}
}
public void setName(Fund fund, string name, string _suffix)
{
string suffix = Regex.Replace(_suffix, @"class|shares|" + name, @"", RegexOptions.IgnoreCase).Trim();
suffix = Regex.Replace(suffix, @"[ ]{2,}", @" ", RegexOptions.IgnoreCase).Trim();
List <string> aname = name.Split(' ').ToList();
List <string> asuffix = suffix.Split(' ').ToList();
List <string> newsuffix = new List <string>();
foreach (string psuffix in asuffix)
{
if (aname.Find(x => x == psuffix) == null)
{
newsuffix.Add(psuffix);
}
else
{
newsuffix.Clear();
}
}
fund.name = name + ";" + Regex.Replace(string.Join(" ", newsuffix), @"-|", @"", RegexOptions.IgnoreCase).Trim();
}
public List <Fund> getFundsFromServer()
{
List <Fund> funds = new List <Fund>();
DateTime now = DateTime.Now;
string postData = "Action=GETALLFUNDS&Login=maximgrishkov%40yandex.ru&Password=3333";
HttpWebRequest wr = (HttpWebRequest)WebRequest.Create("https://rixtrema.net/RixtremaWS401k/AJAXFCT.aspx");
wr.Timeout = int.MaxValue;
wr.Method = "POST";
wr.ContentType = "application/x-www-form-urlencoded";
wr.ContentLength = postData.Length;
StreamWriter writer = new StreamWriter(wr.GetRequestStream());
try
{
writer.Write(postData);
}
catch (Exception ex) { throw (ex); }
finally { writer.Close(); }
HttpWebResponse response = (HttpWebResponse)wr.GetResponse();
StreamReader reader = new StreamReader(response.GetResponseStream());
string respText = reader.ReadToEnd();
double seconds = (DateTime.Now - now).TotalSeconds;
Console.WriteLine(seconds);
if (!respText.ToLower().Contains("success"))
{
throw new Exception(respText);
}
else
{
try
{
dynamic js = JsonConvert.DeserializeObject(respText);
if (js.FCT == null)
{
return(null);
}
if (js.FCT.fund == null)
{
return(null);
}
int i = 0;
foreach (dynamic jsfund in js.FCT.fund)
{
i++;
if (jsfund.Ticker != null && jsfund.Name != null)
{
Fund fund = new Fund();
fund.ticker = jsfund.Ticker;
fund.name = jsfund.Name;
fund.fname = jsfund.FamilyName;
funds.Add(fund);
}
}
Console.WriteLine(i);
return(funds);
}
catch
{
return(null);
}
}
}
//*Функция, которая находит нужную форму, num - номер формы (считаются только нужные типы отчетов) от начала таблицы с формами
public string FindNQ(Fund fund, HtmlNode BodyNodeS, int num = 1)
{
int _num = 1;
bool next = false;
string url = null;
HtmlNode series = BodyNodeS.SelectSingleNode(".//div[@id='seriesDiv']");
if (series != null)
{
foreach (HtmlNode table in series.SelectNodes(".//table"))
{
foreach (HtmlNode tr in table.SelectNodes(".//tr"))
{
HtmlNodeCollection tds = BodyNodeS.SelectNodes(".//td");
if (tds != null)
{
foreach (HtmlNode td in tds)
{
if (next == true)
{
HtmlNode a = td.SelectSingleNode(".//a");
if (url != null && td.GetAttributeValue("class", "") == "")
{
fund.forms++;
return(url);
}
if (a != null)
{
if (tools.ClearStrNameNew(a.InnerText) == "Documents")
{
url = baseLink + a.GetAttributeValue("href", "");
}
}
}
else
{
if (td.InnerText == "N-Q")
{
fund.formName = "N-Q";
if (_num == num)
{
next = true;
}
_num++;
}
if (td.InnerText == "N-CSRS")
{
fund.formName = "N-CSRS";
if (_num == num)
{
next = true;
}
_num++;
}
if (td.InnerText == "N-CSR")
{
fund.formName = "N-CSR";
if (_num == num)
{
next = true;
}
_num++;
}
}
}
}
}
}
}
return(null);
}
public HtmlNode LoadFormAll(Fund fund, HtmlNode body, string urlForm, bool makeXmlAll = false)
{
// Console.WriteLine("LoadForm");
fund.netAssetsFromEdgar = 0;
List <string> names = new List <string>();
List <Caption> captions = new List <Caption>();
//*Вторая страница Edgar для конкретного отчета.
HtmlNode bodyForm = tools.BodyNodeS(tools.Load(urlForm), htmlMakeAll);
if (bodyForm == null)
{
return(body);
}
else
{
//*Берем обязательно даты, они нужны для сохранения
fund.periodOfReport = FindPeriodOfReport(bodyForm);
fund.fillingDate = FindFillingDate(bodyForm);
fund.dateEffectiveness = FindDateEffectiveness(bodyForm);
fund.source = FindUrlDocument(bodyForm, fund.formName);
fund.nameFromEdgar = FindFullNameClc(bodyForm, fund);
/*if (FindFullName(bodyForm, fund) == fund.nameFromEdgar)
* {
* Console.WriteLine("{0} - CHECKED", fund.ticker);
* //fund.makeFailXml("NO FAIL", captions, xmlPathR);
* return null;
* }*/
Console.WriteLine("FUND : {0}", fund.ticker);
//*Берем названия фондов, которые так же находятся в данном отчете
GetFullNames(bodyForm, fund, names, captions);
if (fund.source != null)
{
if (fund.nameFromEdgar != null)
{
//*Документ, который нужно распарсить
string doc = tools.Load(fund.source);
//*Определяем тип файла
string[] url = fund.source.Split('/');
string[] nameOfFile = url[url.Length - 1].Split('.');
string ext = nameOfFile[nameOfFile.Length - 1];
//Console.WriteLine(ext);
if (ext == "html" || ext == "htm")
{
// fund.saveSource(s, xmlPathR);
HtmlNode bodyDocument = tools.BodyNodeS(doc, htmlMakeDocs);
if (bodyDocument != null)
{
//*Сумма всех позиций с сайта morningstar
fund.TotalAssetsMorning();
Document document = new Document(bodyDocument);
document.Parse(fund, captions);
//*Вторая попытка
if (document.holdings.Count == 0)
{
//Console.WriteLine("Try Again");
//*Сжимаем заголовки
zNames(captions, fund);
document.Parse(fund, captions);
}
if (document.holdings.Count > 0)
{
//*Трансормируем набор позиций из словаря в список и назначаем фонду
tools.transformToList(fund, document.holdings);
fund.sumHoldings();
//*Сохранение
fund.MakeXml(xmlPathR);
}
else
{
Console.WriteLine("{0} - HOLDINGS FAIL", fund.ticker);
fund.makeFailXml("HOLDINGS FAIL", captions, xmlPathR);
}
fund.remove();
}
else
{
Console.WriteLine("{0} - HTML FAIL", fund.ticker);
fund.makeFailXml("HTML FAIL", captions, xmlPathR);
//errors.Error("0012");
//*ошибка с html функция_для_кривого_html(fund, captions);
//captions - полные названия фондов в этом документе. Взято с предыдущей страницы edgar
//fund.sumHoldings(); fund.MakeXml(xmlPathR); - для сохранения
}
}
else
{
Console.WriteLine("{0} - TXT FAIL", fund.ticker);
fund.makeFailXml("TXT FAIL", captions, xmlPathR);
//NOTE - файл(doc) - txt. функция_для_txt(fund, captions);
}
}
else
{
}
}
else
{
Console.Write("0013");
fund.status = 2;
}
}
return(body);
}
public void MakeFamily(Fund fund)
{
string urlForms = baseLink + "/cgi-bin/browse-edgar?CIK=" + fund.ticker + "&owner=exclude&action=getcompany&count=200&Find=Search";
HtmlNode body = tools.BodyNodeS(tools.Load(urlForms), htmlMakeAll);
string familyName = "";
bool finded = false;
if (body == null)
{
Console.WriteLine("{0} - NOT FOUND IN EDGAR", fund.ticker);
fund.makeFailXml("NOT FOUND IN EDGAR", null, xmlPathR);
//errors.Error("0008");
fund.status = 1;
}
else
{
fund.classcontract = takeClassContract(body);
HtmlNode a = body.SelectSingleNode(".//p[@class='identInfo']/a[position()=2]");
if (a != null)
{
string link = baseLink + a.GetAttributeValue("href", "");
Console.WriteLine(link);
HtmlNode bodyL = tools.BodyNodeS(tools.LoadWeb(link), htmlMakeAll);
if (bodyL != null)
{
HtmlNodeCollection trs = bodyL.SelectNodes(".//table[@summary]//tr");
if (trs != null)
{
foreach (HtmlNode tr in trs)
{
HtmlNodeCollection tds = tr.SelectNodes(".//td");
if (tds != null)
{
int t = 0;
foreach (HtmlNode td in tds)
{
HtmlNode aa = td.SelectSingleNode("./a");
if (!finded)
{
if (aa != null)
{
//Console.WriteLine(aa.GetAttributeValue("class", ""));
if (aa.GetAttributeValue("class", "") == "")
{
familyName = aa.InnerText;
}
}
if (td.InnerText == fund.classcontract)
{
finded = true;
if (familyName != "" && tds[t + 1] != null)
{
setName(fund, familyName, tds[t + 1].InnerText);
}
}
}
if (t == 4)
{
Fund fundBonus = funds.Find(x => x.ticker.ToLower() == tds[t].InnerText.ToLower().Trim());
if (fundBonus != null && familyName != "")
{
setName(fundBonus, familyName, tds[t - 1].InnerText);
}
}
t++;
}
}
}
}
}
}
}
}