/// <summary>
/// Remove any false colleagues from the database
/// </summary>
private void RemoveFalseColleagues_Click(object sender, EventArgs e)
{
this.Enabled = false;
this.Cursor = Cursors.WaitCursor;
Database DB = new Database(DSN.Text);
int Before = DB.GetIntValue("SELECT Count(*) FROM StarColleagues");
if (useAlternateCheckbox.Checked)
{
ColleagueFinder.RemoveFalseColleagues(DB, this, alternateTableName.Text);
}
else
{
ColleagueFinder.RemoveFalseColleagues(DB, this, "PeoplePublications");
}
int After = DB.GetIntValue("SELECT Count(*) FROM StarColleagues");;
int Removed = Before - After;
AddLogEntry("Removed " + Removed.ToString() + " false colleague" +
((Removed == 1) ? "" : "s"));
UpdateStatus();
this.Cursor = Cursors.Default;
this.Enabled = true;
}
/// <summary>
/// Retrieve the publications for found colleagues
/// </summary>
private void RetrieveColleaguePublications_Click(object sender, EventArgs e)
{
if (!SetApiKeyFromFormField())
{
return;
}
this.Enabled = false;
this.Cursor = Cursors.WaitCursor;
// Set the language restriction
string[] Languages = null;
if (LanguageList.Text != "")
{
Languages = LanguageList.Text.Split(',');
foreach (string Language in Languages)
{
AddLogEntry("Adding language restriction: " + Language);
}
}
else
{
AddLogEntry("No language restriction added");
}
Database DB = new Database(DSN.Text);
// Clear any publications left over after an interruption
DB.ExecuteNonQuery(
@"DELETE p.* FROM ColleaguePublications p, Colleagues c
WHERE c.Setnb = p.Setnb
AND c.Harvested = 0"
);
// Clear any errors
int NumberOfErrors = DB.GetIntValue("SELECT Count(*) FROM Colleagues WHERE Error IS NOT NULL");
AddLogEntry("Clearing " + NumberOfErrors.ToString() + " colleagues with errors");
DB.ExecuteNonQuery(
@"DELETE p.* FROM ColleaguePublications p, Colleagues c
WHERE c.Setnb = p.Setnb
AND c.Error IS NOT NULL"
);
DB.ExecuteNonQuery(
@"UPDATE Colleagues
SET Harvested = 0, Error = NULL, ErrorMessage = NULL
WHERE Error IS NOT NULL"
);
NumberOfErrors = DB.GetIntValue("SELECT Count(*) FROM Colleagues WHERE Error = 1");
if (NumberOfErrors != 0)
{
AddLogEntry("WARNING: " + NumberOfErrors + " errors were not cleared!");
}
UpdateStatus();
// Retrieve the publications for each unharvested colleague
NCBI ncbi = new NCBI("Medline");
AddApiKeyLogEntries();
ColleagueFinder finder = new ColleagueFinder(DB, roster, ncbi,
GetPeoplePublicationTableName());
People Colleagues = new People(DB, "Colleagues");
int Total = Colleagues.PersonList.Count;
int Count = 0;
toolStripProgressBar1.Minimum = 0;
toolStripProgressBar1.Maximum = Total;
foreach (Person person in Colleagues.PersonList)
{
Count++;
toolStripProgressBar1.Value = Count;
statusStrip1.Refresh();
if (person.Harvested == false)
{
AddLogEntry("Harvesting publications for colleague " + person.Last + " (" + person.Setnb + "), " + Count.ToString() + " of " + Total.ToString());
UpdateStatus();
// Check for existing publications in ColleaguePublications
ArrayList Parameters = new ArrayList();
Parameters.Add(Database.Parameter(person.Setnb));
int ExistingPublications = DB.GetIntValue(
"SELECT Count(*) FROM ColleaguePublications WHERE Setnb = ?", Parameters);
if (ExistingPublications > 0)
{
DB.ExecuteNonQuery("UPDATE Colleagues SET Harvested = 1 WHERE Setnb = ?", Parameters);
AddLogEntry(ExistingPublications.ToString() + " publications already exist for colleague " + person.Last + " (" + person.Setnb + "), " + Count.ToString() + " of " + Total.ToString());
}
else
{
Person[] personArray = new Person[1];
personArray[0] = person;
try
{
List <int> allowedPubTypes = new List <int>();
foreach (string type in AllowedPubTypeCategories.Text.Split(','))
{
allowedPubTypes.Add(int.Parse(type));
}
finder.GetColleaguePublications(personArray, Languages, allowedPubTypes);
GC.Collect(); // no need to wait for finalizers, because they don't do anything
}
catch (Exception ex)
{
AddLogEntry(ex.Message);
// Set the error for the colleague
Parameters = new ArrayList();
Parameters.Add(Database.Parameter(ex.Message));
Parameters.Add(Database.Parameter(person.Setnb));
DB.ExecuteNonQuery(
@"UPDATE Colleagues
SET Error = 1, ErrorMessage = ?
WHERE Setnb = ?", Parameters);
}
}
UpdateStatus();
}
else
{
AddLogEntry("Already harvested publications for colleague " + person.Last + " (" + person.Setnb + "), " + Count.ToString() + " of " + Total.ToString());
}
}
this.Cursor = Cursors.Default;
this.Enabled = true;
}
public void UpdateStatus()
{
try
{
Database UpdateDB = new Database(DSN.Text);
bool TablesCreated;
int NumPeople;
int NumHarvestedPeople;
int NumPublications;
int NumErrors;
UpdateDB.GetStatus(out TablesCreated, out NumPeople, out NumHarvestedPeople, out NumPublications, out NumErrors);
this.TablesCreated.Text = TablesCreated.ToString();
this.TablesCreated.Refresh();
this.People.Text = NumPeople.ToString();
this.People.Refresh();
this.PeopleHarvested.Text = NumHarvestedPeople.ToString();
this.PeopleHarvested.Refresh();
this.PublicationsFound.Text = NumPublications.ToString();
this.PublicationsFound.Refresh();
// People With Errors textbox -- set it to red if there are errors
this.PeopleWithErrors.Text = NumErrors.ToString();
if (NumErrors > 0)
{
PeopleWithErrors.BackColor = Color.Red;
}
else
{
PeopleWithErrors.BackColor = People.BackColor;
}
this.PeopleNotHarvested.Text = ((int)(NumPeople - NumHarvestedPeople)).ToString();
// People Not Harvested textbox -- set it to red if there are unharvested
// people and the software is not currently processing
if (NumPeople - NumHarvestedPeople > 0)
{
PeopleNotHarvested.BackColor = Color.Red;
}
else
{
PeopleNotHarvested.BackColor = People.BackColor;
}
// Get the colleague status, enable buttons appropriately
int DiadsFound;
int ColleaguesFound;
int ColleaguesHarvested;
int ColleaguePublications;
int StarsWithColleagues;
ColleagueFinder.GetDBStatus(UpdateDB, out ColleaguesFound, out DiadsFound, out ColleaguesHarvested, out ColleaguePublications, out StarsWithColleagues);
this.DiadsFound.Text = DiadsFound.ToString();
this.ColleaguesFound.Text = ColleaguesFound.ToString();
this.ColleaguesHarvested.Text = ColleaguesHarvested.ToString();
this.ColleaguePublications.Text = ColleaguePublications.ToString();
this.StarsWithColleagues.Text = StarsWithColleagues.ToString();
// Colleagues With Errors textbox -- set it to red if there are errors
// (first check ColleaguesFound so we don't issue this SELECT against a database with no Colleagues table)
if (ColleaguesFound > 0)
{
try
{
NumErrors = UpdateDB.GetIntValue("SELECT Count(*) FROM Colleagues WHERE Error IS NOT NULL");
this.ColleaguesWithErrors.Text = NumErrors.ToString();
}
catch (Exception ex)
{
this.ColleaguesWithErrors.Text = ex.Message;
NumErrors = -1;
}
if (NumErrors != 0)
{
ColleaguesWithErrors.BackColor = Color.Red;
}
else
{
ColleaguesWithErrors.BackColor = People.BackColor;
}
}
// Enable/disable buttons
if (ColleaguesFound == 0)
{
RetrieveColleaguePublications.Enabled = false;
CopyPublicationsFromAnotherDB.Enabled = false;
}
else
{
RetrieveColleaguePublications.Enabled = true;
CopyPublicationsFromAnotherDB.Enabled = true;
}
if (ColleaguePublications > 0)
{
RemoveFalseColleagues.Enabled = true;
}
else
{
RemoveFalseColleagues.Enabled = false;
}
// Roster status -- disable buttons if the roster is not loaded
if (roster != null)
{
this.RosterRows.Text = roster.RosterData.Rows.Count.ToString();
FindPotentialColleagues.Enabled = true;
}
else
{
this.RosterRows.Text = "not loaded";
FindPotentialColleagues.Enabled = false;
RetrieveColleaguePublications.Enabled = false;
CopyPublicationsFromAnotherDB.Enabled = false;
}
Application.DoEvents();
}
catch
{
//
}
}
/// <summary>
/// Generate the colleagues using the Colleagues class
/// </summary>
private void FindPotentialColleagues_Click(object sender, EventArgs e)
{
if (!SetApiKeyFromFormField())
{
return;
}
// If the tables aren't populated, don't look for colleagues.
Database UpdateDB = new Database(DSN.Text);
bool TablesCreated;
int NumPeople;
int NumHarvestedPeople;
int NumPublications;
int NumErrors;
UpdateDB.GetStatus(out TablesCreated, out NumPeople, out NumHarvestedPeople, out NumPublications, out NumErrors);
if ((!TablesCreated) || (NumPeople == 0) || (NumHarvestedPeople == 0))
{
MessageBox.Show("Please select a database that the Publication Harvester has been run on.", "Specify a Valid Database", MessageBoxButtons.OK, MessageBoxIcon.Stop);
return;
}
// Reset the database by default -- if the user says "Yes" when asked,
// then continue from the previous search.
bool ResetDatabase = true;
if (DiadsFound.Text != "0")
{
if (MessageBox.Show("Colleagues have already been found. Are you sure you want to re-find them (or continue the previous search)?",
"Re-find Colleagues?", MessageBoxButtons.YesNo, MessageBoxIcon.Hand) == DialogResult.No)
{
return;
}
if (MessageBox.Show("Should the colleague search continue where it was previously left off? (Click 'No' to reset the database and start finding new colleagues.)",
"Continue from previous search?", MessageBoxButtons.YesNo, MessageBoxIcon.Hand) == DialogResult.Yes)
{
ResetDatabase = false;
}
}
this.Enabled = false;
this.Cursor = Cursors.WaitCursor;
Database DB = new Database(DSN.Text);
// Use the ResetDatabase variable to determine whether or not to resume
// a past find. If resuming, get the Setnb's of stars from the StarColleagues
// table so they can be skipped.
if (ResetDatabase)
{
ColleagueFinder.CreateTables(DB);
}
DataTable StarSetnbsResult = DB.ExecuteQuery("SELECT StarSetnb FROM StarColleagues");
ArrayList StarSetnbs = new ArrayList();
foreach (DataRow Row in StarSetnbsResult.Rows)
{
if (!(StarSetnbs.Contains(Row[0].ToString())))
{
StarSetnbs.Add(Row[0].ToString());
}
}
NCBI ncbi = new NCBI("Medline");
AddApiKeyLogEntries();
ColleagueFinder finder = new ColleagueFinder(DB, roster, ncbi,
GetPeoplePublicationTableName());
People Stars = new People(DB);
int NumStars = Stars.PersonList.Count;
toolStripProgressBar1.Minimum = 0;
toolStripProgressBar1.Maximum = NumStars;
int Count = 0;
foreach (Person Star in Stars.PersonList)
{
Count++;
toolStripProgressBar1.Value = Count;
statusStrip1.Refresh();
Application.DoEvents();
if (StarSetnbs.Contains(Star.Setnb))
{
AddLogEntry("Already found colleagues for " + Star.Last + " (" + Star.Setnb + "), " + Count.ToString() + " of " + NumStars.ToString());
}
else
{
AddLogEntry("Finding " + Star.Last + " (" + Star.Setnb + "), " + Count.ToString() + " of " + NumStars.ToString());
UpdateStatus();
Person[] Colleagues = finder.FindPotentialColleagues(Star);
}
}
UpdateStatus();
this.Cursor = Cursors.Default;
this.Enabled = true;
}
/// <summary>
/// Retrieve all of the publications a set of colleagues
/// </summary>
/// <param name="Star">Star whose colleagues are being retrieved</param>
/// <param name="Colleagues">List of colleagues whose publications should be retrieved</param>
public void GetColleaguePublications(Person[] Colleagues, string[] Languages, List <int> AllowedPubTypeCategories)
{
// Keep a list of written Setnbs, just to make sure we don't write the
// same colleage twice
ArrayList WrittenSetnbs = new ArrayList();
// Process each colleague
foreach (Person Colleague in Colleagues)
{
// Only process a colleague that hasn't yet been touched
if (!WrittenSetnbs.Contains(Colleague.Setnb))
{
WrittenSetnbs.Add(Colleague.Setnb);
// Get the colleague's publications
// Search NCBI -- if an error is thrown, write that error to the database
string results;
try
{
results = ncbi.Search(Colleague.MedlineSearch);
}
catch (Exception ex)
{
string Message = "Error reading publications for "
+ Colleague.Last + " (" + Colleague.Setnb + ex.Message;
ColleagueFinder.WriteErrorToDB(Message, DB, Colleague);
throw new Exception(Message);
}
// Turn the results into a set of publications for the colleague
if ((results != null) && (results.Trim() != ""))
{
Publications ColleaguePublications = new Publications(results, pubTypes);
// Write the publications to the database -- but only if they
// actually belong to the colleague.
if (ColleaguePublications.PublicationList != null)
{
foreach (Publication pub in ColleaguePublications.PublicationList)
{
// If the publication has no authors, it's clearly not actually
// a publication that belongs to this colleague.
// Also, since the publication harvester only harvests
// English publications, we exclude any non-English ones as well.
if ((pub.Authors != null) && (pub.Language == "eng") &&
(AllowedPubTypeCategories.Contains(pubTypes.GetCategoryNumber(pub.PubType))))
{
// Add a row to the ColleaguePublications table -- this will
// return False if the publication doesn't actually belong
// to the colleague
bool PubBelongsToColleague = WriteColleaguePublicationsToDB(DB, Colleague, pub, pubTypes, Languages);
if (PubBelongsToColleague)
{
// Make sure the publication doesn't already exist, then write
// it to the database.
if (DB.GetIntValue("SELECT Count(*) FROM Publications WHERE PMID = " + pub.PMID.ToString()) == 0)
{
Publications.WriteToDB(pub, DB, pubTypes, Languages);
}
}
}
}
}
// Update the Harvested column in the Colleagues table
ArrayList Parameters = new ArrayList();
Parameters.Add(Database.Parameter(Colleague.Setnb));
DB.ExecuteNonQuery("UPDATE Colleagues SET Harvested = 1 WHERE Setnb = ?", Parameters);
}
}
}
}
