public static IIndexDocument FromFile(string file, IIndexDataSource parent)
{
string s = file.ToLower();
if (s.EndsWith(".pdf"))
{
// PDF
return(new IndexPDFDocument(file, parent));
}
else
if (s.EndsWith(".txt"))
{
return(new IndexedTextFile(file, parent));
}
else
if ((s.EndsWith(".html")) || (s.EndsWith(".html")))
{
using (FileStream fs = File.OpenRead(file))
{
return(WebDataSource.FromHtml(fs, file, parent.Name));
}
}
return(null);
}
public IndexPDFDocument(string fname, Stream data, IIndexDataSource parent) : base(fname, parent)
{
try
{
pdfDocument = PdfReader.Open(data, PdfDocumentOpenMode.ReadOnly);
pdfExtractor = new Extractor(pdfDocument);
}
catch (Exception e)
{
Console.WriteLine($"Error open pdf: {fname}");
}
}
/* Create IIndexDocument instance parsing url using parent as parent of created instance,
* returns one of the known documents: html,txt,pdf, ... depending on server url responce content-Type*/
public static IIndexDocument FromUrl(string url, IIndexDataSource parent)
{
HttpWebRequest req = HttpWebRequest.CreateHttp(url);
req.UserAgent = "DOCODO";
req.Accept = "text/html, text/plain, application/pdf";
req.Method = "GET";
IIndexDocument ret = null;
WebResponse res;
try
{
res = req.GetResponse();
}
catch (WebException e)
{
return(null);
}
if (res.ContentType.ToLower().Equals("application/pdf"))
{
ret = new DocumentsDataSource.IndexPDFDocument(url, res.GetResponseStream(), parent);
}
else
if (res.ContentType.ToLower().Equals("text/plain"))
{
using (StreamReader reader = new StreamReader(res.GetResponseStream()))
{
ret = new IndexPagedTextFile(url.Substring(parent.Path.Length), reader.ReadToEnd(), "Source=" + parent.Name);
}
}
else
{
ret = FromHtml(res.GetResponseStream(), url.Substring(parent.Path.Length), parent.Name);
}
return(ret);
}
/// <summary>
/// Add datasource to main index
/// </summary>
/// <param name="ds">datasource to get data from</param>
public DocodoServiceConfigure AddDataSource(IIndexDataSource ds)
{
_service.getBaseIndex().AddDataSource(ds); return(this);
}
public IndexedTextFile(string fname, IIndexDataSource parent)
{
this.fname = fname;
Name = fname.Substring(parent?.Path.Length ?? 0);// fname.IndexOfAny(new char[] { '\\', '/' }, parent.Path.Length) + 1);
this.parent = parent;
}
