private static void ProcessError(string vsErrorType, string result)
{
cFile oFile = new cFile("error.txt");
oFile.Write("___________");
oFile.Write(vsErrorType + " Failed got back: " + result);
oFile.Write("___________");
}
public List <string> IdentifyLinks(string vsfileName)
{
cFile oFile = new cFile("ExtractedLinks");
List <string> oLinks = new List <string>();
StreamReader sr = new StreamReader(vsfileName);
string input;
string pattern = @"<A[^>]*?HREF\s*=\s*""([^""]+)""[^>]*?>([\s\S]*?)<\/A>";
Regex rgx = new Regex(pattern, RegexOptions.IgnoreCase);
while (sr.Peek() >= 0)
{
input = sr.ReadLine();
MatchCollection matches = rgx.Matches(input);
if (matches.Count > 0)
{
foreach (Match match in matches)
{
oLinks.Add(match.Value);
oFile.Write(match.Value);
}
}
}
sr.Close();
return(oLinks);
}
public bool WriteLinks()
{
cFile oFile = new cFile(msExportedLinkPath);
try
{
string sOutput = "" + msUrlList[0];
//oFile.FileName = msExportedLinkPath;
for (int i = 1; i < msUrlList.Count; i++)
{
sOutput += Environment.NewLine + msUrlList[i];
}
oFile.Write(sOutput);
Parallel.For(0, sFileList.Count, x =>
{
oFile.FileName = sFileList[x];
oFile.Delete();
});
return(true);
}
catch (Exception ex)
{
oFile.WriteLog(ex.Message, "WriteLinksError.log");
return(false);
throw ex;
}
}
public bool test3(string vsURL)
{
cWeb oWeb = new cWeb();
string htmlContent = oWeb.GrabPageToString(vsURL);
// Obtain the document interface
IHTMLDocument2 htmlDocument = (IHTMLDocument2) new mshtml.HTMLDocument();
// Construct the document
htmlDocument.write(htmlContent);
//htmlDocument.
List <IHTMLElement> oOut = new List <IHTMLElement>();
// Extract all elements
IHTMLElementCollection allElements = htmlDocument.all;
cFile oFile = new cFile("ele.txt");
// Iterate all the elements and display tag names
foreach (IHTMLElement element in allElements)
{
oFile.Write(element.tagName);
}
return(false);
}
public bool ReadLinksFromHtmlFiles()
{
foreach (var sFile in oFilesList)
{
try
{
ReadHtmlFile(sFile);
}
catch (Exception ex)
{
cFile oFile = new cFile("error.log");
oFile.Write(sFile);
oFile.Write(ex.Message);
}
}
return(true);
}
public void SaveLinks()
{
cFile oFile = new cFile(output);
foreach (var sUrl in oUrListList)
{
oFile.Write(sUrl);
}
}
public bool Save(string vsFileName)
{
try {
cFile oFiler = new cFile(vsFileName);
oFiler.Write(mstrName + "|" + mstrFileName + "|" + mstrLinkDefine);
return(true);
} catch (Exception ex) {
throw ex;
}
}
private void GetSummary(HtmlAgilityPack.HtmlDocument document)
{
//this simply works because InnerText is iterative for all child nodes
HtmlNodeCollection nodes = document.DocumentNode.SelectNodes("html/body/div[4]/div/div/table/tbody/tr/td/table/tbody/tr[1]/td/div[1]/text()");
//but to be more accurate you can use the next line instead
//HtmlNodeCollection nodes = doc.DocumentNode.SelectNodes("//td[@class='title']/a");
string result = "";
result = "";
if (GetData(nodes, ref result))
{
oFFdotNetFic.Summary = result;
}
else
{
cFile oFile = new cFile("error.txt");
oFile.Write("___________");
oFile.Write("Summary Failed got back: " + result);
oFile.Write("___________");
}
}
public bool LoadLinksFromFileToList(string vsPath)
{
try
{
cFile oLogger = new cFile("Error.txt");
oLogger.Write("++++++++++++++++++++++++++");
oLogger.Write(vsPath);
oLogger.Write("++++++++++++++++++++++++++");
cFile oFile = new cFile(vsPath);
//oFile.FileName = ;
string pattern = "\r\n|\n";
String sNewUrlList = oFile.Read();
string[] sTs = Regex.Split(sNewUrlList, pattern);
foreach (string sT in sTs)
{
}
for (int iCnt1 = 0; iCnt1 < sTs.Count(); iCnt1++)
{
oLogger.Write("-----" + iCnt1.ToString() + "-----");
string sUrl1 = sTs[iCnt1];
oLogger.Write(sUrl1);
oLogger.Write("-----" + iCnt1.ToString() + "-----");
CheckLink(sUrl1);
}
return(true);
}
catch (Exception ex)
{
throw ex;
}
//throw new NotImplementedException();
}
public bool LoadLinksFromFile(string vsPath)
{
try
{
cFile oLogger = new cFile("Error.txt");
oLogger.Write("++++++++++++++++++++++++++");
oLogger.Write(vsPath);
oLogger.Write("++++++++++++++++++++++++++");
cFile oFile = new cFile(vsPath);
//oFile.FileName = ;
List <String> sNewUrlList = oFile.ReadList();
for (int iCnt1 = 0; iCnt1 < sNewUrlList.Count(); iCnt1++)
{
oLogger.Write("-----" + iCnt1.ToString() + "-----");
string sUrl1 = sNewUrlList[iCnt1];
oLogger.Write(sUrl1);
oLogger.Write("-----" + iCnt1.ToString() + "-----");
if (sUrl1.Length >= 28)
{
sUrl1 = FixFFdotNet(sUrl1);
}
if (sUrl1.Length >= 5)
{
msUrlList.Add(sUrl1);;
}
}
return(true);
}
catch (Exception ex)
{
throw ex;
}
//throw new NotImplementedException();
}
public bool Load()
{
if (File.Exists(msFileName))
{
string[] oLinkCatRecord = null;
cFile oFile = new cFile(msFileName);
try
{
string sInput = oFile.Read();
// oLinkCatRecord = sInput.Split(Convert.ToChar(Constants.vbCrLf));
oLinkCatRecord = sInput.Split(new String[] { "\r\n" }, StringSplitOptions.RemoveEmptyEntries);
foreach (string linkRow in oLinkCatRecord)
{
if (linkRow.Trim().Length > 3)
{
// Process a student record
cLinkCategory oLinkCategory = new cLinkCategory();
string[] sLinkFields = null;
sLinkFields = linkRow.Split('|');
var _with3 = oLinkCategory;
_with3.Name = sLinkFields[0];
_with3.FileName = sLinkFields[1];
_with3.LinkDefine = sLinkFields[2];
oCatagories.Add(oLinkCategory);
// Add(oLinkCategory);
}
}
return(true);
}
catch (Exception ex)
{
//MessageBox.Show(ex.Message);
cFile oLog = new cFile("error.log");
oLog.Write(ex.Message);
}
}
else
{
File.Create("LinkCategories.txt");
Load();
}
return(false);
}
private void GetAuthor(HtmlAgilityPack.HtmlDocument document)
{
throw new NotImplementedException();
//this simply works because InnerText is iterative for all child nodes
HtmlAgilityPack.HtmlNodeCollection nodes = document.DocumentNode.SelectNodes("Date XPath");
//but to be more accurate you can use the next line instead
//HtmlNodeCollection nodes = doc.DocumentNode.SelectNodes("//td[@class='title']/a");
string result = "";
result = "";
if (GetData(nodes, ref result))
{
//oFFdotNetFic.ChapterText
oFFdotNetFic.LastUpdated1 = result;
}
else
{
cFile oFile = new cFile("error.txt");
oFile.Write("___________");
oFile.Write("LastUpdated1 Failed got back: " + result);
oFile.Write("___________");
}
}
private bool GetData(HtmlNodeCollection nodes, ref string result)
{
List <string> results = new List <string>();
foreach (HtmlNode item in nodes)
{
results.Add(item.InnerText);
result = item.InnerText;
cFile oFile = new cFile("nodes.txt");
oFile.Write("___________");
oFile.Write(item.XPath.ToString());
oFile.Write(result);
oFile.Write("___________");
}
if (results.Count == 1)
{
result = results[0].ToString();
return(true);
}
return(false);
throw new NotImplementedException();
}
private async Task GetLinkValues(string folder, cFile oFile)
{
List <Task> oTasks = new List <Task>();
foreach (string fileName in Directory.GetFiles(folder, "*.txt", SearchOption.AllDirectories))
{
Task oTask = ProcessFile(fileName);
oTask.Start();
lblOutput.Text = fileName;
oTasks.Add(oTask);
oFile.Write(fileName);
//SQLGenerator.GenerateSQLTransactions(Path.GetFullPath(fileName));
}
await oTasks[0];
}
public bool SaveLinks(string vsFilePath)
{
try
{
var oFile = new cFile(vsFilePath);
foreach (var sUrl in oUrListList)
{
oFile.Write(sUrl);
}
}
catch (Exception)
{
return(false);
}
return(true);
}
private void importSortTextFileLinks()
{
cFile oFile = new cFile("FilesLoaded");
try
{
string startupPath = Application.StartupPath;
using (FolderBrowserDialog dialog = new FolderBrowserDialog())
{
dialog.Description = "Open a folder which contains the xml output";
// dialog.ShowNewFolderButton = false;
// dialog.RootFolder = Environment.SpecialFolder.MyComputer;
if (dialog.ShowDialog() == DialogResult.OK)
{
string folder = dialog.SelectedPath;
lblOutput.Text = dialog.SelectedPath;
foreach (string fileName in Directory.GetFiles(folder, "*.txt", SearchOption.AllDirectories))
{
oTool.ExportedLinkPath = "Input.txt";
List <string> sUrls = oTool.GetLinksFromFile(fileName);
oTool.WriteLinks(sUrls);
oTool.AddUrlList(sUrls);
oFile.Write(fileName);
//SQLGenerator.GenerateSQLTransactions(Path.GetFullPath(fileName));
}
oTool.Sort();
}
}
//using (OpenFileDialog dialog = new OpenFileDialog())
//{
// dialog.Filter = "xml files (*.xml)|*.xml";
// dialog.Multiselect = false;
// dialog.InitialDirectory = ".";
// dialog.Title = "Select file (only in XML format)";
// if (dialog.ShowDialog() == DialogResult.OK)
// {
// SQLGenerator.GenerateSQLTransactions(Application.StartupPath + Settings.Default.xmlFile);
// }
//}
lblOutput.Text = "done";
}
catch (Exception exc)
{
MessageBox.Show("Import failed because " + exc.Message + " , please try again later.");
}
}
public void writeChildren(HtmlNode vNode)
{
cFile oFile = new cFile("nodes.txt");
foreach (HtmlNode link in vNode.ChildNodes)
{
oFile.Write("--");
oFile.Write("-----------------------------");
oFile.Write("-----------------------------");
oFile.Write("--");
oFile.Write(link.Name.ToString());
oFile.Write(link.XPath.ToString());
oFile.Write(link.InnerHtml);
// writeChildren(link);
}
}
public bool WriteLinks()
{
try
{
cFile oFile = new cFile(msExportedLinkPath);
//oFile.FileName = msExportedLinkPath;
for (int iCnt = 0; iCnt < msUrlList.Count(); iCnt++)
{
oFile.Write(msUrlList[iCnt]);
}
return(true);
}
catch (Exception ex)
{
return(false);
throw ex;
}
}
private Task ProcessFile(string fileName)
{
cFile oFile = new cFile("log");
// use await here, like so
Task task = new Task(() =>
{
try
{
cUrlTool voTool = new cUrlTool();
voTool.GetLinksFromFile(fileName);
voTool.WriteLinks();
}
catch (Exception ex)
{
oFile.Write(ex.Message);
}
});
//Task<List<String>> voTask = new Task<List<String>>(() => GetLinksFromFile(vsFilepath));
return(task);
}
public bool ReadWriteHtmlFile(string vsFile)
{
//Read bookmarks from file
try
{
cFile oFile = new cFile(vsFile);
cFile oWrite = new cFile("out.txt");
cFile oLog = new cFile("log.log");
var bookmarks = oHtmlReader.Read(oFile.ReadFileStream());
foreach (var b in bookmarks.AllLinks)
{
oWrite.Write(b.Url);
}
oLog.Write(vsFile);
}
catch (Exception)
{
return(false);
}
return(true);
}
