/// <summary>
/// Retrieve the publications for a person and write them to the database
/// </summary>
/// <param name="ncbi">NCBI web query object</param>
/// <param name="pubTypes">PublicationTypes object</param>
/// <param name="person">Person to retrieve publications for</param>
/// <param name="StatusCallback">Callback function to return status</param>
/// <param name="MessageCallback">Callback function to send messages</param>
/// <param name="AverageMilliseconds">Average time (in milliseconds) of each publication write</param>
/// <returns>Number of publications written</returns>
public int GetPublications(NCBI ncbi, PublicationTypes pubTypes, Person person,
GetPublicationsStatus StatusCallback, GetPublicationsMessage MessageCallback,
CheckForInterrupt InterruptCallback, out double AverageMilliseconds)
{
ArrayList Parameters;
DateTime StartTime;
DateTime EndTime;
double TotalMilliseconds = 0;
AverageMilliseconds = 0;
int numberFound = 0;
int numberWritten = 0;
// Double-check that the person is really unharvested. If we try to
// write publications for a person who already has publications, it will
// cause an error -- and that could happen if this person was already
// written from a duplicate person earlier.
Parameters = new ArrayList();
Parameters.Add(Database.Parameter(person.Setnb));
int HarvestedCount = DB.GetIntValue("SELECT Count(*) FROM People WHERE Setnb = ? AND Harvested = 1", Parameters);
if (HarvestedCount > 0)
{
MessageCallback("Already harvested publications for " + person.Last + " (" + person.Setnb + ")", false);
Parameters = new ArrayList();
Parameters.Add(Database.Parameter(person.Setnb));
return(DB.GetIntValue("SELECT Count(*) FROM PeoplePublications WHERE Setnb = ?", Parameters));
}
MessageCallback("Retrieving data from NCBI", true);
// Find any other people with the same names and search criteria.
// Any publications found for this person should also be found
// for them, so when we write the rows to PeoplePublications later
// we'll also write them for the other people as well.
// Look in the database for any other people with the same
// values for name1, name2, name3, name4, name5, name6, and MedlineSearch.
// Write their PeoplePublications as well.
string NamesClause = "";
Parameters = new ArrayList();
for (int i = 0; i < 6; i++)
{
if (i < person.Names.Length)
{
Parameters.Add(Database.Parameter(person.Names[i]));
NamesClause += " Name" + ((int)(i + 1)).ToString() + " = ? AND ";
}
else
{
NamesClause += " Name" + ((int)(i + 1)).ToString() + " IS NULL AND ";
}
}
Parameters.Add(Database.Parameter(person.MedlineSearch));
Parameters.Add(Database.Parameter(person.Setnb));
DataTable Results = DB.ExecuteQuery("SELECT " + Database.PEOPLE_COLUMNS +
@"FROM People
WHERE Harvested = 0 AND "
+ NamesClause +
@" MedlineSearch = ?
AND Setnb <> ?", Parameters
);
ArrayList DuplicatePeople = new ArrayList();
foreach (DataRow Row in Results.Rows)
{
Person dupe = new Person(Row, Results.Columns);
DuplicatePeople.Add(dupe);
MessageCallback("Also writing publications for " + dupe.Last + " (" + dupe.Setnb + ") with same names and search criteria", false);
}
// Search NCBI -- if an error is thrown, write that error to the database
string results;
try
{
results = ncbi.Search(person.MedlineSearch);
if (results.Substring(0, 100).Contains("Error occurred"))
{
// NCBI returns an HTML error page in the results
//
// <html>
// <body>
// <br/><h2>Error occurred: Unable to obtain query #1</h2><br/>
// ...
//
// If NCBI returns an empty result set with no publications, it will give the error:
// Error occurred: Empty result - nothing todo
//
// That error should generate a warning and mark the person as harvested in the database.
// Any other error should be written to the database as an error.
string Error = results.Substring(results.IndexOf("Error occurred"));
if (results.Contains("<"))
{
Error = Error.Substring(0, Error.IndexOf("<"));
}
string Message;
if (Error.ToLower().Contains("empty result"))
{
Message = "Warning for "
+ person.Last + " (" + person.Setnb + "): no publications found (NCBI returned empty results)";
person.Harvested = true;
person.WriteToDB(DB);
}
else
{
Message = "Error reading publications for "
+ person.Last + " (" + person.Setnb + "): NCBI returned '" + Error + "'";
person.WriteErrorToDB(DB, Message);
}
MessageCallback(Message, false);
return(0);
}
}
catch (Exception ex)
{
string Message = "Error reading publications for "
+ person.Last + " (" + person.Setnb + "): " + ex.Message;
person.WriteErrorToDB(DB, Message);
MessageCallback(Message, false);
return(0);
}
Publications mpr = new Publications(results, pubTypes);
if (mpr.PublicationList != null)
{
foreach (Publication publication in mpr.PublicationList)
{
numberFound++;
// Exit immediately if the user interrupted the harvest
if (InterruptCallback())
{
return(numberWritten);
}
try
{
// Calculate the average time, to return in the callback status function
StartTime = DateTime.Now;
// Add the publication to PeoplePublications
// First find the author position and calculate the position type
int AuthorPosition = 0;
for (int i = 1; (publication.Authors != null) && (AuthorPosition == 0) && (i <= publication.Authors.Length); i++)
{
foreach (string name in person.Names)
{
if (StringComparer.CurrentCultureIgnoreCase.Equals(
publication.Authors[i - 1], name //.ToUpper()
))
{
AuthorPosition = i;
}
else if (name == "*")
{
AuthorPosition = -1;
}
}
}
// If the PMID is 0, we don't have a way to process the publication
// and it was probably a Medline search result error.
if (publication.PMID == int.MinValue)
{
string errorMessage = "Found an invalid publication";
if (!string.IsNullOrEmpty(publication.Title))
{
errorMessage += " (Title = '" + publication.Title + "')";
}
person.WriteErrorToDB(DB, errorMessage);
MessageCallback(errorMessage, false);
}
else if (publication.PMID == 0)
{
string errorMessage = "WARNING: Found a publication with PMID = 0, not marking this as an error";
if (!string.IsNullOrEmpty(publication.Title))
{
errorMessage += " (Title = '" + publication.Title + "')";
}
MessageCallback(errorMessage, false);
}
// If for some reason the author doesn't exist in the publication, send a message back
else if (AuthorPosition == 0)
{
MessageCallback("Publication " + publication.PMID + " does not contain author " + person.Setnb, false);
}
else
{
// Write the publication to the database
if (Publications.WriteToDB(publication, DB, pubTypes, Languages))
{
// Exit immediately if the user interrupted the harvest
if (InterruptCallback())
{
return(numberWritten);
}
// Only increment the publication count if the publication
// is actually written or already in the database
numberWritten++;
// Only add the row to PeoplePublications if the publication
// was written, or was already in the database. (For example,
// if the publication is not in English, it won't be written.)
Publications.WritePeoplePublicationsToDB(DB, person, publication);
// Write the publication for each of the other people
foreach (Person dupe in DuplicatePeople)
{
Publications.WritePeoplePublicationsToDB(DB, dupe, publication);
}
// Calculate the average time per publication in milliseconds
EndTime = DateTime.Now;
TimeSpan Difference = EndTime - StartTime;
TotalMilliseconds += Difference.TotalMilliseconds;
AverageMilliseconds = TotalMilliseconds / numberWritten;
}
}
}
catch (Exception ex)
{
person.WriteErrorToDB(DB, ex.Message);
MessageCallback("Error writing publication " + publication.PMID.ToString() + ": " + ex.Message, false);
}
StatusCallback(numberFound, mpr.PublicationList.Length, (int)AverageMilliseconds);
}
}
// Make sure each of the people with the same names and search query
// are marked as harvested and have their errors cleared
foreach (Person dupe in DuplicatePeople)
{
Parameters = new ArrayList();
Parameters.Add(Database.Parameter(dupe.Setnb));
DB.ExecuteNonQuery(
@"UPDATE People
SET Harvested = 1, Error = NULL, ErrorMessage = NULL
WHERE Setnb = ?", Parameters);
}
// Once the publications are all read, updated People.Harvested, as part of
// the fault-tolerance scheme -- PeoplePublications rows are only "final" when
// this value is updated for the person. Any others can be cleared using
// ClearDataAfterInterruption().
Parameters = new ArrayList();
Parameters.Add(Database.Parameter(person.Setnb));
DB.ExecuteNonQuery(@"UPDATE People
SET Harvested = 1
WHERE Setnb = ?", Parameters);
return(numberWritten);
}
