private void ParseInproceedings(int year)
{
int global_conference_id = 0;
dicInproceedings.Clear();
dicConferences.Clear();
dicStringConferences.Clear();
GC.Collect();
GC.WaitForPendingFinalizers();
string prefex_saved = "\\";
Parallel.ForEach(fileInproceedings, x =>
{
int fileindex = x.Name.LastIndexOf('.');
fileindex = Convert.ToInt32(x.Name.Substring(fileindex - 1, 1));
// Parse the file
IEnumerable<string> lines;
lines = File.ReadLines(Path.GetFullPath(x.DirectoryName + prefex_saved + x.Name));
int curline = 0;
Parallel.ForEach(lines, l =>
{
InproceedingsDBLP inproceeding = InproceedingsDBLP.CreateFromLine(l, curline, fileindex, year);
if (inproceeding != null)
{
lock (lock_inproceeding)
{
dicInproceedings.Add(inproceeding.Id, inproceeding);
}
if (chkSaveConference.Checked)
{
lock (lock_conference)
{
string[] keys = inproceeding.Key.Split('/');
string conf_key = keys.Length > 1 ? keys[1] : keys[0];
if (dicStringConferences.ContainsKey(conf_key))
{
ConferenceDBLP conference = dicStringConferences[conf_key];
conference.InproceedingsID += "|" + inproceeding.Id;
conference.CountInproceedings++;
}
else
{
inproceeding.ConferenceID = global_conference_id;
ConferenceDBLP conference = new ConferenceDBLP(conf_key, inproceeding.Conference, 1, inproceeding.Id.ToString(), global_conference_id);
global_conference_id++;
dicConferences.Add(conference.Id, conference);
dicStringConferences.Add(conf_key, conference);
}
}
}
}
curline++;
if (curline % 100000 == 0)
{
this.Invoke((MethodInvoker)delegate
{
txtMsgLog.Text += (string.Format("Line Index= {0} - File index={1} - {2} \r\n", curline, fileindex, x.Name));
txtMsgLog.SelectionStart = txtMsgLog.Text.Length;
txtMsgLog.ScrollToCaret();
if (txtMsgLog.Text.Length > 1000)
{
txtMsgLog.Text = "";
}
});
}
});
});
}
private void LoadDataFromDB()
{
using (dblpData dataDBLP = new dblpData())
{
List<dblp_author> listAuthors = dataDBLP.dblp_author.ToList();
List<dblp_conference> listConferences = dataDBLP.dblp_conference.ToList();
List<dblp_inproceedings> listInproceedings = dataDBLP.dblp_inproceedings.ToList();
listAs.Clear();
listIs.Clear();
listCs.Clear();
AuthorDBLP aDBLP;
foreach (dblp_author a in listAuthors)
{
aDBLP = new AuthorDBLP()
{
OldValue = 0,
CurrentValue = Convert.ToInt32(txtStartValue.Text),
Id = a.id,
Name = a.name,
CountInproceedings = a.count == null ? 0 : (int)a.count,
Inproceedings = new List<int>()
};
foreach (string id in a.inproceedings.TrimStart('|').Split('|'))
{
aDBLP.Inproceedings.Add(Convert.ToInt32(id));
}
aDBLP.InproceedingsID = aDBLP.Inproceedings.Aggregate("", (cur, next) => cur + "|" + next);
listAs.Add(aDBLP);
}
InproceedingsDBLP iDBLP;
foreach (dblp_inproceedings i in listInproceedings)
{
iDBLP = new InproceedingsDBLP()
{
Id = i.id,
Title = i.title,
Key = i.key,
Year = i.year,
Pages = i.pages,
Crossref = i.crossref,
Conference = i.conference,
AuthorsID = "",
ConferenceID = i.conference_id == null ? 0 : (int)i.conference_id,
CountAuthors = i.count == null ? 0 : (int)i.count,
Authors = new List<int>()
};
foreach (string id in i.authors.TrimStart('|').Split('|'))
{
iDBLP.Authors.Add(Convert.ToInt32(id));
}
iDBLP.AuthorsID = iDBLP.Authors.Aggregate("", (cur, next) => cur + "|" + next);
listIs.Add(iDBLP);
}
ConferenceDBLP cDBLP;
foreach (dblp_conference c in listConferences)
{
cDBLP = new ConferenceDBLP()
{
Id = c.id,
Name = c.name,
CountInproceedings = c.count == null ? 0 : (int)c.count,
Inproceedings = new List<int>()
};
foreach (string id in c.inproceedings.TrimStart('|').Split('|'))
{
cDBLP.Inproceedings.Add(Convert.ToInt32(id));
}
cDBLP.InproceedingsID = cDBLP.Inproceedings.Aggregate("", (cur, next) => cur + "|" + next);
listCs.Add(cDBLP);
}
}
}
