public void CreateDatabaseEntry(string fileidentifier)
{
DescriptionScraper ds = new DescriptionScraper(fileidentifier, false);
Entry e = ds.Scrape();
if (e.Identifier != "ERROR")
{
System.Console.WriteLine("Entry Added For " + e.Identifier);
this.db.AddEntryObject(e);
}
else
{
System.Console.WriteLine("An Error Occurred.");
}
}
public Entry Scrape()
{
HtmlWeb hw = new HtmlWeb();
HtmlDocument doc = hw.Load(url);
string[] substrings;
string title = "";
string abstr = "";
string[] tags = null;
if (doc == null)
{
return new Entry {
Identifier = "ERROR"
}
}
;
try
{
foreach (HtmlNode node in doc.DocumentNode.SelectNodes("//p"))
{
substrings = node.InnerText.ToLower().Split(':');
substrings[0] = substrings[0].Substring(0, substrings[0].Length - 1);
if (substrings[0].Contains("title"))
{
title = substrings[1].Trim().Replace(" ", "").Replace("*", "");
}
if (substrings[0].Contains("abstract"))
{
abstr = substrings[1].Trim().Replace(" ", "").Replace("*", "");
}
if (substrings[0].Contains("descriptors"))
{
substrings[1] = substrings[1].Trim().ToLower().Replace(" ", "").Replace("*", "");
tags = substrings[1].Split(',');
for (int i = 0; i < substrings.Length; i++)
{
substrings[i] = substrings[i].Trim();
}
}
}
if (title != null && abstr != null && tags != null)
{
return(new Entry {
Identifier = identifier, Title = title, Abstr = abstr, Tags = tags
});
}
else
{
return(new Entry {
Identifier = "ERROR"
});
}
}
catch (Exception e)
{
if (!retry)
{
DescriptionScraper retry = new DescriptionScraper(identifier, true);
return(retry.Scrape());
}
else
{
return(new Entry {
Identifier = "ERROR"
});
}
}
}
}
