public void TestHyphens()
{
// Retrieve the publications for each person in input1.xls using GetPublications()
People PeopleFromFile = new People(
AppDomain.CurrentDomain.BaseDirectory + "\\Unit Tests\\TestPeople",
"test hyphens.xls");
Assert.IsTrue(PeopleFromFile.PersonList.Count == 1);
Person PersonToWrite = PeopleFromFile.PersonList[0];
Assert.IsTrue(PersonToWrite.Names.Length == 3);
Assert.IsTrue(PersonToWrite.Names[0] == "wassertheil-smoller s");
Assert.IsTrue(PersonToWrite.Names[1] == "wassertheil s");
Assert.IsTrue(PersonToWrite.Names[2] == "wassertheil-smoller sm");
// Write them to the database and read them back
Database DB = new Database("Publication Harvester Unit Test");
Harvester harvester = new Harvester(DB);
harvester.CreateTables();
PersonToWrite.WriteToDB(DB);
People PeopleFromDB = new People(DB);
Assert.IsTrue(PeopleFromDB.PersonList.Count == 1);
Person PersonToRead = PeopleFromDB.PersonList[0];
Assert.IsTrue(PersonToRead.Names.Length == 3);
Assert.IsTrue(PersonToRead.Names[0] == "wassertheil-smoller s");
Assert.IsTrue(PersonToRead.Names[1] == "wassertheil s");
Assert.IsTrue(PersonToRead.Names[2] == "wassertheil-smoller sm");
}
public void WriteAndClearError()
{
string[] Names = { "a", "b", "c" };
Person PersonToWrite = new Person("1234ABCD", "First", "Middle", "Last",
false, Names, "Medline search query");
// Write the person to the database
Database DB = new Database("Publication Harvester Unit Test");
Harvester harvester = new Harvester(DB);
harvester.CreateTables();
PersonToWrite.WriteToDB(DB);
// Write an error
PersonToWrite.WriteErrorToDB(DB, "This is the error message");
DataTable Results = DB.ExecuteQuery("SELECT Error, ErrorMessage FROM People WHERE Setnb = '1234ABCD'");
Assert.IsTrue(Results.Rows[0]["Error"].Equals(true));
Assert.IsTrue(Results.Rows[0]["ErrorMessage"].ToString() == "This is the error message");
// Clear the error -- note that we need to use GetBoolValue() to get around the MySQL bug
PersonToWrite.ClearErrorInDB(DB);
Results = DB.ExecuteQuery("SELECT " + Database.PEOPLE_COLUMNS + " FROM People WHERE Setnb = '1234ABCD'");
bool boolValue;
Assert.IsTrue(Database.GetBoolValue(Results.Rows[0]["Error"], out boolValue));
Assert.IsTrue(boolValue == false);
Assert.IsTrue(Results.Rows[0]["ErrorMessage"].ToString() == "");
}
public void ResetDatabase()
{
// Import "TestPeopleMaintenance/input1 plus testhyphens.xls" into the People table
People PeopleFromFile = new People(
AppDomain.CurrentDomain.BaseDirectory + "\\Unit Tests\\TestPeopleMaintenance",
"input1 plus testhypens.xls");
DB = new Database("Publication Harvester Unit Test");
Harvester harvester = new Harvester(DB);
harvester.CreateTables();
MockNCBI mockNCBI = new MockNCBI("medline");
mockNCBI.SearchThrowsAnError = false;
PublicationTypes ptc = new PublicationTypes(
AppDomain.CurrentDomain.BaseDirectory + "\\Unit Tests\\TestPublicationTypes",
"PublicationTypes.csv"
);
ptc.WriteToDB(DB);
// Anonymous callback functions for GetPublications
Harvester.GetPublicationsStatus StatusCallback = delegate(int number, int total, int averageTime)
{
//
};
Harvester.GetPublicationsMessage MessageCallback = delegate(string Message, bool StatusBarOnly)
{
//
};
Harvester.CheckForInterrupt InterruptCallback = delegate()
{
return(false);
};
// Write the people, then "harvest" the publications using MockNCBI
double AverageMilliseconds;
foreach (Person person in PeopleFromFile.PersonList)
{
person.WriteToDB(DB);
harvester.GetPublications(mockNCBI, ptc, person, StatusCallback, MessageCallback, InterruptCallback, out AverageMilliseconds);
}
People PeopleFromDB = new People(DB);
Assert.AreEqual(PeopleFromDB.PersonList.Count, 4);
}
public void ReadAndWritePublicationTypes()
{
// Read the publication types from the CSV file
PublicationTypes ptc = new PublicationTypes(
AppDomain.CurrentDomain.BaseDirectory + "\\Unit Tests\\TestPublicationTypes",
"PublicationTypes.csv"
);
Assert.AreEqual(ptc.Categories.Count, 52);
Assert.AreEqual(ptc.GetCategoryNumber("Legislation"), 0);
Assert.AreEqual(ptc.GetCategoryNumber("Consensus Development Conference, NIH"), 1);
Assert.AreEqual(ptc.GetCategoryNumber("Review, Multicase"), 2);
Assert.AreEqual(ptc.GetCategoryNumber("Technical Report"), 3);
Assert.AreEqual(ptc.GetCategoryNumber("Comment"), 4);
// Verify OverrideFirstCategory values
Assert.IsTrue(ptc.OverrideFirstCategory.ContainsKey("Review"));
Assert.IsTrue(ptc.OverrideFirstCategory.ContainsKey("Review, Multicase"));
Assert.AreEqual(ptc.OverrideFirstCategory.ContainsKey("Comment"), false);
// First recreate the database, then write the publication types to it
Database DB = new Database("Publication Harvester Unit Test");
Harvester harvester = new Harvester(DB);
harvester.CreateTables();
ptc.WriteToDB(DB);
// Read the publication types from the database
PublicationTypes ptcFromDB =
new PublicationTypes(DB);
Assert.AreEqual(ptcFromDB.Categories.Count, 52);
Assert.AreEqual(ptcFromDB.GetCategoryNumber("Overall"), 0);
Assert.AreEqual(ptcFromDB.GetCategoryNumber("Clinical Trial, Phase II"), 1);
Assert.AreEqual(ptcFromDB.GetCategoryNumber("Review of Reported Cases"), 2);
Assert.AreEqual(ptcFromDB.GetCategoryNumber("Technical Report"), 3);
Assert.AreEqual(ptcFromDB.GetCategoryNumber("Letter"), 4);
Assert.AreEqual(ptcFromDB.GetCategoryNumber("Comment"), 4);
// Verify OverrideFirstCategory values
Assert.IsTrue(ptcFromDB.OverrideFirstCategory.ContainsKey("Review"));
Assert.IsTrue(ptcFromDB.OverrideFirstCategory.ContainsKey("Review, Multicase"));
Assert.AreEqual(ptcFromDB.OverrideFirstCategory.ContainsKey("Comment"), false);
}
public void TestCreateFromDatabase()
{
// Add a publication to the database
Database DB = new Database("Publication Harvester Unit Test");
Harvester harvester = new Harvester(DB);
harvester.CreateTables();
Publication write = CreateTestPublication();
PublicationTypes ptc = new PublicationTypes(
AppDomain.CurrentDomain.BaseDirectory + "\\Unit Tests\\TestPublicationTypes",
"PublicationTypes.csv"
);
Publications.WriteToDB(write, DB, ptc, new string[] { "eng" });
// Read it back
Publication read;
Assert.IsTrue(Publications.GetPublication(DB, 15904469, out read, false));
TestCreatedPublication(read);
}
/// <summary>
/// Verify that input1.* was read properly
/// </summary>
/// <param name="PeopleFromFile">People object that contains contents of input1.* file</param>
private static void VerifyInput1File(People PeopleFromFile)
{
int Count = 0;
// Verify that each PersonToWrite was read properly -- note that the order people
// are returned is not guaranteed
foreach (Person person in PeopleFromFile.PersonList)
{
Count++;
TestInput1People(person);
}
// Verify that all four people were read from input1.xls
Assert.AreEqual(Count, 4);
// Write the people to the database -- first initialize it
Database DB = new Database("Publication Harvester Unit Test");
Harvester harvester = new Harvester(DB);
harvester.CreateTables();
foreach (Person person in PeopleFromFile.PersonList)
{
person.WriteToDB(DB);
}
// Read the people back from the database
People PeopleFromDB = new People(DB);
Count = 0;
// Verify that each PersonToWrite was read properly -- note that the order people
// are returned is not guaranteed
foreach (Person person in PeopleFromDB.PersonList)
{
Count++;
TestInput1People(person);
}
// Verify that all four people were read from input1.xls
Assert.AreEqual(Count, 4);
}
public void TestPeopleReportRows()
{
// Set up the database with test publications (and don't forget to add the
// publication types!)
DB = new Database("Publication Harvester Unit Test");
Harvester harvester = new Harvester(DB);
harvester.CreateTables();
PublicationTypes PubTypes = new PublicationTypes(
AppDomain.CurrentDomain.BaseDirectory + "\\Unit Tests\\TestPublicationTypes",
"PublicationTypes.csv"
);
PubTypes.WriteToDB(DB);
reports = new Reports(DB, AppDomain.CurrentDomain.BaseDirectory + "\\Unit Tests\\TestReports\\pubmed_jifs.xls");
Assert.IsTrue(reports.Weights.Count == 10);
TestHarvester.GetPublicationsFromInput1XLS_Using_MockNCBI(false, new string[] { "eng" }, 22);
people = new People(DB);
foreach (Person person in people.PersonList)
{
Publications pubs = new Publications(DB, person, false);
switch (person.Setnb)
{
case "A6009400":
// Van Eys has two publications in 1998, both have zero weight
DataRow Row = WriteAndReadBackCSVRow(reports.ReportRow(person, pubs, 1998));
Assert.IsTrue(Row.ItemArray.Length == 74);
Assert.IsTrue(Row.ItemArray[0].ToString() == "A6009400");
Assert.IsTrue(Row.ItemArray[1].ToString() == "1998");
// Verify that all values are zero, except for pubcount (#3),
// pubcount_pos1 (#5), 123pubcount (#15), 123pubcount_pos1 (#17),
// 3pubcount (#51), 3pubcount_pos1 (#53)
for (int i = 2; i <= 73; i++)
{
if ((i == 2) || (i == 4) || (i == 14) || (i == 16) ||
(i == 50) || (i == 52))
{
Assert.IsTrue(Row.ItemArray[i].ToString() == "2", "Failed at i == " + i.ToString());
}
else
{
Assert.IsTrue(Row.ItemArray[i].ToString() == "0", "Failed at i == " + i.ToString());
}
}
break;
case "A5401532":
// Tobian has two publications in 1997 of type 3 with a
// combined weight of 4.602
Row = WriteAndReadBackCSVRow(reports.ReportRow(person, pubs, 1997));
Assert.IsTrue(Row.ItemArray.Length == 74);
Assert.IsTrue(Row.ItemArray[0].ToString() == "A5401532");
Assert.IsTrue(Row.ItemArray[1].ToString() == "1997");
// Verify that all values are zero, except for pubcount (#3),
// pubcount_pos1 (#5), 123pubcount (#15), 123pubcount_pos1 (#17),
// 3pubcount (#51), 3pubcount_pos1 (#53), which should be 2
//
// and wghtd_pubcount (#4), wghtd_pubcount_pos1 (#6),
// wghtd_123pubcount (#16), wghtd_123pubcount_pos1 (#18),
// wghtd_3pubcount (#52), wghtd_3pubcount_pos1 (#54),
// which should be 4.602
for (int i = 2; i <= 73; i++)
{
if ((i == 2) || (i == 4) || (i == 14) || (i == 16) ||
(i == 50) || (i == 52))
{
Assert.IsTrue(Row.ItemArray[i].ToString() == "2", "Failed at i == " + i.ToString());
}
else if ((i == 3) || (i == 5) || (i == 15) || (i == 17) ||
(i == 51) || (i == 53))
{
Assert.IsTrue(Row.ItemArray[i].ToString() == "4.602", "Failed at i == " + i.ToString());
}
else
{
Assert.IsTrue(Row.ItemArray[i].ToString() == "0", "Failed at i == " + i.ToString());
}
}
break;
}
}
}
public void TestSkipSetnbs()
{
// Set up the database with test publications (and don't forget to add the
// publication types!)
DB = new Database("Publication Harvester Unit Test");
Harvester harvester = new Harvester(DB);
harvester.CreateTables();
PublicationTypes PubTypes = new PublicationTypes(
AppDomain.CurrentDomain.BaseDirectory + "\\Unit Tests\\TestPublicationTypes",
"PublicationTypes.csv"
);
PubTypes.WriteToDB(DB);
reports = new Reports(DB, AppDomain.CurrentDomain.BaseDirectory + "\\Unit Tests\\TestReports\\pubmed_jifs.xls");
Assert.IsTrue(reports.Weights.Count == 10);
TestHarvester.GetPublicationsFromInput1XLS_Using_MockNCBI(false, new string[] { "eng" }, 22);
people = new People(DB);
Reports.ReportStatus StatusCallback = delegate(int number, int total, Person person, bool ProgressBarOnly)
{
//
};
Reports.ReportMessage MessageCallback = delegate(string Message)
{
//
};
// Set up the Setnbs to skip
ArrayList SetnbsToSkip = new ArrayList();
SetnbsToSkip.Add("A5401532");
SetnbsToSkip.Add("A6009400");
// Verify that the people report skips the setnbs
StreamWriter writer = new StreamWriter(AppDomain.CurrentDomain.BaseDirectory + "\\TestSkipSetnbs.csv");
reports.PeopleReport(SetnbsToSkip, writer, StatusCallback, MessageCallback);
writer.Close();
StreamReader reader = new StreamReader(AppDomain.CurrentDomain.BaseDirectory + "\\TestSkipSetnbs.csv");
string Line = reader.ReadLine(); // skip the header row
while ((Line = reader.ReadLine()) != null)
{
Assert.IsFalse(Line.StartsWith("A5401532"));
Assert.IsFalse(Line.StartsWith("A6009400"));
}
reader.Close();
// Verify that the people report skips the setnbs
writer = new StreamWriter(AppDomain.CurrentDomain.BaseDirectory + "\\TestSkipSetnbs.csv", false);
reports.PubsReport(SetnbsToSkip, writer, StatusCallback, MessageCallback);
writer.Close();
reader = new StreamReader(AppDomain.CurrentDomain.BaseDirectory + "\\TestSkipSetnbs.csv");
Line = reader.ReadLine(); // skip the header row
while ((Line = reader.ReadLine()) != null)
{
Assert.IsFalse(Line.StartsWith("A5401532"));
Assert.IsFalse(Line.StartsWith("A6009400"));
}
reader.Close();
File.Delete(AppDomain.CurrentDomain.BaseDirectory + "\\TestSkipSetnbs.csv");
}
public void TestEntireReport()
{
string Setnb = "";
// Set up the database with test publications (and don't forget to add the
// publication types!)
DB = new Database("Publication Harvester Unit Test");
Harvester harvester = new Harvester(DB);
harvester.CreateTables();
PublicationTypes PubTypes = new PublicationTypes(
AppDomain.CurrentDomain.BaseDirectory + "\\Unit Tests\\TestPublicationTypes",
"PublicationTypes.csv"
);
PubTypes.WriteToDB(DB);
reports = new Reports(DB, AppDomain.CurrentDomain.BaseDirectory + "\\Unit Tests\\TestReports\\pubmed_jifs.xls");
Assert.IsTrue(reports.Weights.Count == 10);
TestHarvester.GetPublicationsFromInput1XLS_Using_MockNCBI(false, new string[] { "eng" }, 22);
people = new People(DB);
// Write the report
StreamWriter writer = new StreamWriter(AppDomain.CurrentDomain.BaseDirectory + "\\TestEntireReport.csv");
Reports.ReportStatus StatusCallback = delegate(int number, int total, Person person, bool ProgressBarOnly) {
//
};
Reports.ReportMessage MessageCallback = delegate(string Message)
{
//
};
reports.PeopleReport(null, writer, StatusCallback, MessageCallback);
writer.Close();
// Read the report into an array
var lines = File.ReadAllLines($"{AppDomain.CurrentDomain.BaseDirectory}\\TestEntireReport.csv")
.Select(line => line.Split(new char[] { ',' }));
var header = lines.First();
var data = lines.Skip(1).ToList();
Assert.AreEqual(85, data.Count);
string ReportData(string setnb, string year, string column)
{
var index = Array.IndexOf(header, column);
var row = data.Where(line => line[0] == setnb && line[1] == year);
Assert.AreEqual(1, row.Count(), $"Unable to find setnb={setnb} year={year} in TestEntireReport.csv");
return(row.First()[index]);
}
var q = ReportData("A5401532", "1997", "wghtd_pubcount_pos1");
// Read the report file that was generated by hand (TestEntireReport_Data.xls)
// and check each value against the report that was generated by Reports()
string[] Columns =
{
"setnb", "year", "pubcount", "wghtd_pubcount", "pubcount_pos1",
"wghtd_pubcount_pos1", "pubcount_posN", "wghtd_pubcount_posN",
"pubcount_posM", "wghtd_pubcount_posM", "pubcount_posNTL",
"wghtd_pubcount_posNTL", "pubcount_pos2", "wghtd_pubcount_pos2",
"123pubcount", "wghtd_123pubcount", "123pubcount_pos1",
"wghtd_123pubcount_pos1", "123pubcount_posN", "wghtd_123pubcount_posN",
"123pubcount_posM", "wghtd_123pubcount_posM", "123pubcount_posNTL",
"wghtd_123pubcount_posNTL", "123pubcount_pos2", "wghtd_123pubcount_pos2",
"1pubcount", "wghtd_1pubcount", "1pubcount_pos1", "wghtd_1pubcount_pos1",
"1pubcount_posN", "wghtd_1pubcount_posN", "1pubcount_posM",
"wghtd_1pubcount_posM", "1pubcount_posNTL", "wghtd_1pubcount_posNTL",
"1pubcount_pos2", "wghtd_1pubcount_pos2", "2pubcount", "wghtd_2pubcount",
"2pubcount_pos1", "wghtd_2pubcount_pos1", "2pubcount_posN",
"wghtd_2pubcount_posN", "2pubcount_posM", "wghtd_2pubcount_posM",
"2pubcount_posNTL", "wghtd_2pubcount_posNTL", "2pubcount_pos2",
"wghtd_2pubcount_pos2", "3pubcount", "wghtd_3pubcount", "3pubcount_pos1",
"wghtd_3pubcount_pos1", "3pubcount_posN", //"wghtd_3pubcount_posN",
"3pubcount_posM", "wghtd_3pubcount_posM", "3pubcount_posNTL",
"wghtd_3pubcount_posNTL", "3pubcount_pos2", "wghtd_3pubcount_pos2",
"4pubcount", "wghtd_4pubcount", "4pubcount_pos1", "wghtd_4pubcount_pos1",
"4pubcount_posN", "wghtd_4pubcount_posN", "4pubcount_posM",
"wghtd_4pubcount_posM", "4pubcount_posNTL", "wghtd_4pubcount_posNTL",
"4pubcount_pos2", "wghtd_4pubcount_pos2"
};
DataTable HandGeneratedData = NpoiHelper.ReadExcelFileToDataTable(AppDomain.CurrentDomain.BaseDirectory +
"\\Unit Tests\\TestReports", "TestEntireReport_Data.xls");
Assert.AreEqual(HandGeneratedData.Rows.Count, 85);
var valuesChecked = 0;
for (int RowNum = 0; RowNum < HandGeneratedData.Rows.Count; RowNum++)
{
// Find the rows in the hand-generated data and the report
DataRow HandGeneratedRow = HandGeneratedData.Rows[RowNum];
Setnb = HandGeneratedRow[0].ToString();
int Year = Convert.ToInt32(HandGeneratedRow[1]);
for (int i = 2; i < Columns.Length; i++)
{
valuesChecked++;
String columnName = Columns[i];
var actualValue = ReportData(Setnb, Year.ToString(), columnName);
string expectedValue = HandGeneratedRow[columnName].ToString();
Assert.AreEqual(expectedValue, actualValue, Setnb + "/" + Year.ToString() + "/" + columnName + " -- hand generated has " + expectedValue + ", report has" + actualValue);
}
}
Assert.AreEqual((HandGeneratedData.Rows.Count) * (Columns.Length - 2), valuesChecked);
// Use BackupReportAndGetSetnbs to back up the report -- check that it
// returns the correct list of Setnbs and removes the last one from
// the file. The last Setnb should still be in Setnb.
ArrayList Setnbs = Reports.BackupReportAndGetSetnbs(AppDomain.CurrentDomain.BaseDirectory + "\\TestEntireReport.csv");
// Read the backup file that was created, make sure that the last setnb in the
// file isn't contained and the others are
StreamReader reader = new StreamReader(AppDomain.CurrentDomain.BaseDirectory + "\\TestEntireReport.csv.bak");
string Line = reader.ReadLine(); //skip the header row
while ((Line = reader.ReadLine()) != null) // Find the last setnb in the original file
{
Setnb = Line.Substring(0, 8);
}
string RemovedSetnb = Setnb;
Assert.IsFalse(Setnbs.Contains(RemovedSetnb));
reader.Close();
// Verify that the new file contains only the other setnbs
Assert.IsTrue(Setnbs.Count == 3);
Assert.IsFalse(Setnbs.Contains(RemovedSetnb));
reader = new StreamReader(AppDomain.CurrentDomain.BaseDirectory + "\\TestEntireReport.csv");
Line = reader.ReadLine(); //skip the header row
while ((Line = reader.ReadLine()) != null)
{
Setnb = Line.Substring(0, 8);
Assert.IsTrue(Setnbs.Contains(Setnb));
Assert.IsFalse(Setnb == RemovedSetnb);
}
reader.Close();
// Delete the temporary files
File.Delete(AppDomain.CurrentDomain.BaseDirectory + "\\TestEntireReport.csv");
File.Delete(AppDomain.CurrentDomain.BaseDirectory + "\\TestEntireReport.csv.bak");
}
public void TestPubsReportRow()
{
// Set up the database with test publications (and don't forget to add the
// publication types!)
DB = new Database("Publication Harvester Unit Test");
Harvester harvester = new Harvester(DB);
harvester.CreateTables();
PublicationTypes PubTypes = new PublicationTypes(
AppDomain.CurrentDomain.BaseDirectory + "\\Unit Tests\\TestPublicationTypes",
"PublicationTypes.csv"
);
PubTypes.WriteToDB(DB);
reports = new Reports(DB, AppDomain.CurrentDomain.BaseDirectory + "\\Unit Tests\\TestReports\\pubmed_jifs.xls");
Assert.IsTrue(reports.Weights.Count == 10);
TestHarvester.GetPublicationsFromInput1XLS_Using_MockNCBI(false, new string[] { "eng" }, 22);
people = new People(DB);
foreach (Person person in people.PersonList)
{
Publications pubs = new Publications(DB, person, false);
if (pubs.PublicationList != null)
{
foreach (Publication pub in pubs.PublicationList)
{
DataRow Row = WriteAndReadBackCSVRow(reports.PubsReportRow(person, pub));
switch (pub.PMID)
{
case 15249795:
// 0. setnb
Assert.IsTrue(Row.ItemArray[0].ToString() == "A5401532");
// 1. pmid
Assert.IsTrue(Row.ItemArray[1].ToString() == "15249795");
// 2. journal_name
Assert.IsTrue(Row.ItemArray[2].ToString() == "J Clin Hypertens (Greenwich)");
// 3. year
Assert.IsTrue(Row.ItemArray[3].ToString() == "2004");
// 4. Month
Assert.IsTrue(Row.ItemArray[4].ToString() == "Jul");
// 5. day
Assert.IsTrue(Row.ItemArray[5].ToString() == "");
// 6. title
Assert.IsTrue(Row.ItemArray[6].ToString() == "Interview with Louis Tobian, MD. Interview by Marvin Moser.");
// 7. Volume
Assert.IsTrue(Row.ItemArray[7].ToString() == "6");
// 8. issue
Assert.IsTrue(Row.ItemArray[8].ToString() == "7");
// 9. position
Assert.IsTrue(Row.ItemArray[9].ToString() == "1");
// 10. nbauthors
Assert.IsTrue(Row.ItemArray[10].ToString() == "1");
// 11. Bin
Assert.IsTrue(Row.ItemArray[11].ToString() == "0");
// 12. Pages
Assert.IsTrue(Row.ItemArray[12].ToString() == "391-2");
// 13. Publication_type
Assert.IsTrue(Row.ItemArray[13].ToString() == "Historical Article");
break;
}
}
}
}
}
/// <summary>
/// Set up the database with data from Input1.XLS using the Mock NCBI object
/// (this is also called from TestReports())
/// </summary>
/// <param name="NCBISearchThrowsAnError">True if the MockNCBI object is supposed to throw an error</param>
public static void GetPublicationsFromInput1XLS_Using_MockNCBI(bool NCBISearchThrowsAnError, string[] Languages, int ExpectedPublications)
{
bool TablesCreated;
int NumPeople;
int NumHarvestedPeople;
int NumPublications;
int NumErrors;
Database DB = new Database("Publication Harvester Unit Test");
// Drop all tables and make sure the database reports as empty
foreach (string Table in new string[] {
"meshheadings", "people", "peoplepublications", "publicationauthors",
"publicationmeshheadings", "publications", "pubtypecategories"
})
{
DB.ExecuteNonQuery("DROP TABLE IF EXISTS " + Table);
DB.GetStatus(out TablesCreated, out NumPeople, out NumHarvestedPeople, out NumPublications, out NumErrors);
Assert.IsFalse(TablesCreated);
Assert.AreEqual(NumPeople, 0);
Assert.AreEqual(NumHarvestedPeople, 0);
Assert.AreEqual(NumPublications, 0);
Assert.AreEqual(NumErrors, 0);
}
// Create and populate the tables
Harvester harvester = new Harvester(DB);
harvester.Languages = Languages;
MockNCBI mockNCBI = new MockNCBI("medline");
mockNCBI.SearchThrowsAnError = NCBISearchThrowsAnError;
PublicationTypes ptc = new PublicationTypes(
AppDomain.CurrentDomain.BaseDirectory + "\\Unit Tests\\TestPublicationTypes",
"PublicationTypes.csv"
);
// Reinitialize the database
harvester.CreateTables();
ptc.WriteToDB(DB);
// Retrieve the publications for each person in input1.xls using GetPublications()
People PeopleFromFile = new People(
AppDomain.CurrentDomain.BaseDirectory + "\\Unit Tests\\TestPeople",
"input1.xls");
// Make an anonymous callback function that keeps track of the callback data
int Callbacks = 0; // this will count all of the publications
Harvester.GetPublicationsStatus StatusCallback = delegate(int number, int total, int averageTime)
{
Callbacks++;
};
// Make an anonymous callback function to do nothing for GetPublicationsMessage
Harvester.GetPublicationsMessage MessageCallback = delegate(string Message, bool StatusBarOnly)
{
//
};
// Make an anonymous callback function to return false for CheckForInterrupt
Harvester.CheckForInterrupt InterruptCallback = delegate()
{
return(false);
};
// Verify that the database was created and populated properly
DB.GetStatus(out TablesCreated, out NumPeople, out NumHarvestedPeople, out NumPublications, out NumErrors);
Assert.IsTrue(TablesCreated);
Assert.AreEqual(NumPeople, 0);
Assert.AreEqual(NumHarvestedPeople, 0);
Assert.AreEqual(NumPublications, 0);
Assert.AreEqual(NumErrors, 0);
int PeopleCount = 0;
int HarvestedCount = 0;
int PubCount = 0;
foreach (Person person in PeopleFromFile.PersonList)
{
double AverageMilliseconds;
// First write the person to the database
person.WriteToDB(DB);
PeopleCount++;
// Check that the database status is updated properly
DB.GetStatus(out TablesCreated, out NumPeople, out NumHarvestedPeople, out NumPublications, out NumErrors);
Assert.IsTrue(TablesCreated);
Assert.AreEqual(NumPeople, PeopleCount);
if (!NCBISearchThrowsAnError)
{
Assert.AreEqual(NumHarvestedPeople, HarvestedCount);
}
else
{
Assert.AreEqual(NumHarvestedPeople, 0);
}
Assert.AreEqual(NumPublications, PubCount);
if (!NCBISearchThrowsAnError)
{
Assert.AreEqual(NumErrors, 0);
}
else
{
Assert.AreEqual(NumErrors, PeopleCount - 1);
}
// Harvest the person's publications
PubCount += harvester.GetPublications(mockNCBI, ptc, person, StatusCallback, MessageCallback, InterruptCallback, out AverageMilliseconds);
HarvestedCount++;
// Check the status again after the people were harvested
DB.GetStatus(out TablesCreated, out NumPeople, out NumHarvestedPeople, out NumPublications, out NumErrors);
Assert.IsTrue(TablesCreated);
Assert.AreEqual(NumPeople, PeopleCount);
if (!NCBISearchThrowsAnError)
{
Assert.AreEqual(NumHarvestedPeople, HarvestedCount);
}
else
{
Assert.AreEqual(NumHarvestedPeople, 0);
}
Assert.AreEqual(NumPublications, PubCount);
if (!NCBISearchThrowsAnError)
{
Assert.AreEqual(NumErrors, 0);
}
else
{
Assert.AreEqual(NumErrors, PeopleCount);
}
}
// Verify that the database was written properly
if (!NCBISearchThrowsAnError)
{
Assert.IsTrue(Callbacks == 24);
}
else
{
Assert.IsTrue(Callbacks == 0);
}
DB.GetStatus(out TablesCreated, out NumPeople, out NumHarvestedPeople, out NumPublications, out NumErrors);
Assert.IsTrue(TablesCreated);
Assert.AreEqual(NumPeople, 4);
if (!NCBISearchThrowsAnError)
{
Assert.AreEqual(NumHarvestedPeople, 4);
Assert.AreEqual(NumPublications, ExpectedPublications);
Assert.AreEqual(NumErrors, 0);
}
else
{
Assert.AreEqual(NumHarvestedPeople, 0);
Assert.AreEqual(NumPublications, 0);
Assert.AreEqual(NumErrors, 4);
}
}
/// <summary>
/// Harvest each of the publications in the people file
/// </summary>
/// <param name="PeopleFile">Filename of the people file</param>
/// <param name="PublicationTypeFile">Filename of publication type file</param>
/// <param name="ContinueFromInterruption">True if continuing from a previously interrupted harvest</param>
public void Harvest(string PeopleFile, string PublicationTypeFile, bool ContinueFromInterruption)
{
// First verify that the files exist
if (!File.Exists(PeopleFile))
{
MessageBox.Show("The People file '" + PeopleFile + "' does not exist", "People file not found", MessageBoxButtons.OK, MessageBoxIcon.Warning);
return;
}
if (!File.Exists(PublicationTypeFile))
{
MessageBox.Show("The Publication Type file '" + PublicationTypeFile + "' does not exist", "Publication Type file not found", MessageBoxButtons.OK, MessageBoxIcon.Warning);
return;
}
UpdateDatabaseStatus();
if (ContinueFromInterruption)
{
AddLogEntry("Continuing interrupted harvest");
}
else
{
AddLogEntry("Beginning harvesting");
}
// Reset lastDSNSelected to make sure that the next check for interrupted data is NOT skipped
lastDSNSelected = "";
// Initialize the harvester
Harvester harvester;
Database DB;
// Initialize objects
try
{
DB = new Database(DSN.Text);
harvester = new Harvester(DB);
// Set the language restriction
string[] Languages;
if (LanguageList.Text != "")
{
Languages = LanguageList.Text.Split(',');
harvester.Languages = Languages;
foreach (string Language in Languages)
{
AddLogEntry("Adding language restriction: " + Language);
}
}
else
{
AddLogEntry("No language restriction added");
}
}
catch (Exception ex)
{
AddLogEntryWithErrorBox(ex.Message, "Unable to begin harvesting");
return;
}
// Initializethe database
try
{
if (!ContinueFromInterruption)
{
AddLogEntry("Initializing the database");
harvester.CreateTables();
UpdateDatabaseStatus();
}
}
catch (Exception ex)
{
AddLogEntryWithErrorBox(ex.Message, "Unable to initialize database");
return;
}
PublicationTypes pubTypes;
if (ContinueFromInterruption)
{
// If we're continuing, read the publication types from the databse
try
{
AddLogEntry("Reading publication types from the database");
pubTypes = new PublicationTypes(DB);
}
catch (Exception ex)
{
AddLogEntryWithErrorBox(ex.Message, "Unable to read publication types");
return;
}
// Remove any data left over from the interruption
if (ContinueFromInterruption)
{
AddLogEntry("Removing any data left over from the previous interruption");
harvester.ClearDataAfterInterruption();
}
UpdateDatabaseStatus();
}
else
{
// Read the publication types from the file and write them to the database
try
{
AddLogEntry("Writing publication types to database");
pubTypes = new PublicationTypes(Path.GetDirectoryName(PublicationTypeFile), Path.GetFileName(PublicationTypeFile));
pubTypes.WriteToDB(DB);
UpdateDatabaseStatus();
}
catch (Exception ex)
{
AddLogEntryWithErrorBox(ex.Message, "Unable to read publication types");
return;
}
// Read the people
try
{
AddLogEntry("Reading people from " + Path.GetFileName(PeopleFile) + " and writing them to the database");
harvester.ImportPeople(PeopleFile);
UpdateDatabaseStatus();
}
catch (Exception ex)
{
AddLogEntryWithErrorBox(ex.Message, "Unable to read the people from " + Path.GetFileName(PeopleFile));
return;
}
}
// Make an anonymous callback function that keeps track of the callback data
Harvester.GetPublicationsStatus StatusCallback = delegate(int number, int total, int averageTime)
{
// No need to update the progress bar for this -- it leads to a messy-looking UI because it's also updated for the person total
// toolStripProgressBar1.Minimum = 0;
// toolStripProgressBar1.Maximum = total;
// toolStripProgressBar1.Value = number;
toolStripStatusLabel1.Text = "Reading publication " + number.ToString() + " of " + total.ToString() + " (" + averageTime.ToString() + " ms average)";
UpdateDatabaseStatus();
Application.DoEvents();
};
// Make an anonymous callback function that logs any messages passed back
Harvester.GetPublicationsMessage MessageCallback = delegate(string Message, bool StatusBarOnly)
{
if (StatusBarOnly)
{
toolStripStatusLabel1.Text = Message;
//this.Refresh();
//statusStrip1.Refresh();
Application.DoEvents();
}
else
{
AddLogEntry(Message);
}
};
// Make an anonymous callback function to return the value of Interrupt for CheckForInterrupt
Harvester.CheckForInterrupt InterruptCallback = delegate()
{
return(InterruptClicked);
};
// Get each person's publications and write them to the database
NCBI ncbi = new NCBI("medline");
if (NCBI.ApiKeyExists)
{
AddLogEntry("Using API key: " + NCBI.ApiKeyPath);
}
else
{
AddLogEntry("Performance is limited to under 3 requests per second.");
AddLogEntry("Consider pasting an API key into " + NCBI.ApiKeyPath);
AddLogEntry("Or set the NCBI_API_KEY_FILE environemnt variable to the API key file path");
AddLogEntry("For more information, see https://ncbiinsights.ncbi.nlm.nih.gov/2017/11/02/new-api-keys-for-the-e-utilities/");
}
People people = new People(DB);
int totalPeopleInPersonList = people.PersonList.Count;
int numberOfPeopleProcessed = 0;
toolStripProgressBar1.Minimum = 0;
toolStripProgressBar1.Maximum = totalPeopleInPersonList;
foreach (Person person in people.PersonList)
{
numberOfPeopleProcessed++;
try
{
// If continuing from interruption, only harvest unharvested people
if ((!ContinueFromInterruption) || (!person.Harvested))
{
AddLogEntry("Getting publications for " + person.Last + " (" + person.Setnb + "), number " + numberOfPeopleProcessed.ToString() + " of " + totalPeopleInPersonList.ToString());
toolStripProgressBar1.Value = numberOfPeopleProcessed;
double AverageMilliseconds;
int NumPublications = harvester.GetPublications(ncbi, pubTypes, person, StatusCallback, MessageCallback, InterruptCallback, out AverageMilliseconds);
if (InterruptClicked)
{
AddLogEntry("Publication harvesting was interrupted");
UpdateDatabaseStatus();
return;
}
AddLogEntry("Wrote " + NumPublications.ToString() + " publications, average write time " + Convert.ToString(Math.Round(AverageMilliseconds, 1)) + " ms");
UpdateDatabaseStatus();
}
else
{
AddLogEntry("Already retrieved publications for " + person.Last + " (" + person.Setnb + ")");
}
}
catch (Exception ex)
{
AddLogEntry("An error occurred while reading publications for " + person.Last + " (" + person.Setnb + "): " + ex.Message);
}
}
AddLogEntry("Finished reading publications");
UpdateDatabaseStatus();
}
public void TestMiddleAuthor()
{
string MedlineData = @"PMID- 14332364
OWN - NLM
STAT- OLDMEDLINE
DA - 19651101
DCOM- 19961201
LR - 20051116
PUBM- Print
IS - 0002-922X (Print)
VI - 110
DP - 1965 Sep
TI - CONGENITAL SCALP DEFECTS IN TWIN SISTERS.
PG - 293-5
FAU - HODGMAN, J E
AU - HODGMAN JE
FAU - MATHIES, A W Jr
AU - MATHIES AW Jr
FAU - LEVAN, N E
AU - LEVAN NE
LA - eng
PT - Journal Article
PL - UNITED STATES
TA - Am J Dis Child
JT - American journal of diseases of children (1960)
JID - 0370471
SB - OM
MH - *Abnormalities
MH - *Diseases in Twins
MH - *Infant, Newborn
MH - *Scalp
OTO - NLM
OT - *ABNORMALITIES
OT - *DISEASES IN TWINS
OT - *INFANT, NEWBORN
OT - *SCALP
EDAT- 1965/09/01
MHDA- 1965/09/01 00:01
PST - ppublish
SO - Am J Dis Child. 1965 Sep;110:293-5.";
// Add a person to the database
Database DB = new Database("Publication Harvester Unit Test");
Person Mathies = new Person("A5703161", "Allan", "W", "MATHIES", false,
new String[] { "MATHIES AW Jr" }, "MATHIES AW Jr[au]");
// Create the publication
PublicationTypes ptc = new PublicationTypes(
AppDomain.CurrentDomain.BaseDirectory + "\\Unit Tests\\TestPublicationTypes",
"PublicationTypes.csv"
);
Publications Pubs = new Publications(MedlineData, ptc);
Assert.IsTrue(Pubs.PublicationList.Length == 1);
Publication Pub = Pubs.PublicationList[0];
Assert.IsTrue(Pub.Authors[1] == "MATHIES AW Jr");
// Clear the database
Harvester harvester = new Harvester(DB);
harvester.CreateTables();
// Write the person and publication to the database
// NOTE: The author position is determined in WritePeoplePublicationsToDB()
PublicationTypes PubTypes = new PublicationTypes(
AppDomain.CurrentDomain.BaseDirectory + "\\Unit Tests\\TestPublicationTypes",
"PublicationTypes.csv"
);
Mathies.WriteToDB(DB);
Publications.WriteToDB(Pub, DB, PubTypes, new string[] { "eng" });
Publications.WritePeoplePublicationsToDB(DB, Mathies, Pub);
// Verify that Mathies is the middle author, not the second author
Publications PubsFromDB = new Publications(DB, Mathies, false);
Assert.IsTrue(PubsFromDB.PublicationList.Length == 1);
Assert.IsTrue(PubsFromDB.PublicationList[0].PMID == 14332364);
Harvester.AuthorPositions PositionType;
Assert.IsTrue(Publications.GetAuthorPosition(DB, 14332364, Mathies, out PositionType, "PeoplePublications") == 2);
Assert.IsTrue(PositionType == Harvester.AuthorPositions.Middle);
}
public void TestWriteToDB()
{
// Create a new database and write the publication to it
Database DB = new Database("Publication Harvester Unit Test");
Harvester h = new Harvester(DB);
h.CreateTables();
Publication p = CreateTestPublication();
PublicationTypes ptc = new PublicationTypes(
AppDomain.CurrentDomain.BaseDirectory + "\\Unit Tests\\TestPublicationTypes",
"PublicationTypes.csv"
);
Publications.WriteToDB(p, DB, ptc, new string[] { "eng" });
// Verify the publication has been added
DataTable Results = DB.ExecuteQuery(
@"SELECT Journal, Year, Authors, Month, Day, Title, Volume,
Issue, Pages, PubType, PubTypeCategoryID
FROM Publications
WHERE PMID = 15904469"
);
Assert.IsTrue(Results.Rows[0]["Year"].Equals(2005));
Assert.IsTrue(Results.Rows[0]["Authors"].Equals(4));
Assert.IsTrue(Results.Rows[0]["Year"].Equals(2005));
Assert.IsTrue(Results.Rows[0]["Month"].Equals("May"));
Assert.IsTrue(Results.Rows[0]["Day"].Equals(DBNull.Value));
Assert.IsTrue(Results.Rows[0]["Title"].Equals("Charcoal cigarette filters and lung cancer risk in Aichi Prefecture, Japan."));
Assert.IsTrue(Results.Rows[0]["Volume"].Equals("96"));
Assert.IsTrue(Results.Rows[0]["Issue"].Equals("5"));
Assert.IsTrue(Results.Rows[0]["Pages"].Equals("283-7"));
Assert.IsTrue(Results.Rows[0]["PubType"].Equals("Clinical Trial"));
Assert.IsTrue(Results.Rows[0]["PubTypeCategoryID"].Equals((short)1));
// Verify the authors
Assert.IsTrue(DB.GetIntValue("SELECT Count(*) FROM PublicationAuthors") == 4);
Results = DB.ExecuteQuery("SELECT PMID, Position, Author, First, Last FROM PublicationAuthors ORDER BY Position");
Assert.IsTrue(Results.Rows[0]["PMID"].Equals(15904469));
Assert.IsTrue(Results.Rows[1]["PMID"].Equals(15904469));
Assert.IsTrue(Results.Rows[2]["PMID"].Equals(15904469));
Assert.IsTrue(Results.Rows[3]["PMID"].Equals(15904469));
Assert.IsTrue(Results.Rows[0]["Position"].Equals(1));
Assert.IsTrue(Results.Rows[1]["Position"].Equals(2));
Assert.IsTrue(Results.Rows[2]["Position"].Equals(3));
Assert.IsTrue(Results.Rows[3]["Position"].Equals(4));
Assert.IsTrue(Results.Rows[0]["Author"].ToString() == "Muscat JE");
Assert.IsTrue(Results.Rows[1]["Author"].ToString() == "Takezaki T");
Assert.IsTrue(Results.Rows[2]["Author"].ToString() == "Tajima K");
Assert.IsTrue(Results.Rows[3]["Author"].ToString() == "Stellman SD");
Assert.IsTrue(Results.Rows[0]["First"].Equals((short)1));
Assert.IsTrue(Results.Rows[1]["First"].Equals((short)0));
Assert.IsTrue(Results.Rows[2]["First"].Equals((short)0));
Assert.IsTrue(Results.Rows[3]["First"].Equals((short)0));
Assert.IsTrue(Results.Rows[0]["Last"].Equals((short)0));
Assert.IsTrue(Results.Rows[1]["Last"].Equals((short)0));
Assert.IsTrue(Results.Rows[2]["Last"].Equals((short)0));
Assert.IsTrue(Results.Rows[3]["Last"].Equals((short)1));
// Verifying the MeSH headings
// Note the join to PublicationMeSHHeadings to make sure it was populated properly
Assert.IsTrue(DB.GetIntValue("SELECT Count(*) FROM PublicationMeSHHeadings WHERE PMID = " + p.PMID.ToString()) == 17);
Results = DB.ExecuteQuery(
@"SELECT m.Heading
FROM MeSHHeadings m, PublicationMeSHHeadings p
WHERE p.MeSHHeadingID = m.ID
AND p.PMID = " + p.PMID.ToString()
);
Assert.IsTrue(Results.Rows.Count == 17);
Assert.IsTrue(Results.Rows[0]["Heading"].ToString() == "Adult");
Assert.IsTrue(Results.Rows[5]["Heading"].ToString() == "Female");
Assert.IsTrue(Results.Rows[9]["Heading"].ToString() == "Lung Neoplasms/*etiology/pathology");
Assert.IsTrue(Results.Rows[16]["Heading"].ToString() == "Smoking/*adverse effects");
// Verify the grants
Assert.IsTrue(DB.GetIntValue("SELECT Count(*) FROM PublicationGrants WHERE PMID = " + p.PMID.ToString()) == 2);
Results = DB.ExecuteQuery(
@"SELECT GrantID
FROM PublicationGrants
WHERE PMID = " + p.PMID.ToString() + " ORDER BY GrantID ASC"
);
Assert.IsTrue(Results.Rows[0]["GrantID"].ToString() == "CA-17613/CA/NCI");
Assert.IsTrue(Results.Rows[1]["GrantID"].ToString() == "CA-68387/CA/NCI");
}
public void TestOtherPeople()
{
// Note: Publication 14560782 was added to OtherPeople.dat to verify
// that the software handles the situation where a publication has
// no authors listed.
Database DB = new Database("Publication Harvester Unit Test");
Harvester harvester = new Harvester(DB);
MockNCBI mockNCBI = new MockNCBI("medline");
PublicationTypes ptc = new PublicationTypes(
AppDomain.CurrentDomain.BaseDirectory + "\\Unit Tests\\TestPublicationTypes",
"PublicationTypes.csv"
);
// Reinitialize the database
harvester.CreateTables();
// Make an anonymous callback function that keeps track of the callback data
int Callbacks = 0; // this will count all of the publications
Harvester.GetPublicationsStatus StatusCallback = delegate(int number, int total, int averageTime)
{
Callbacks++;
};
// Make an anonymous callback function to do nothing for GetPublicationsMessage
Harvester.GetPublicationsMessage MessageCallback = delegate(string Message, bool StatusBarOnly)
{
//
};
// Make an anonymous callback function to return false for CheckForInterrupt
Harvester.CheckForInterrupt InterruptCallback = delegate()
{
return(false);
};
// Create a new person to test
string[] names = new string[2];
names[0] = "Klein RG";
names[1] = "Guillemin R";
Person person = new Person("A1234567", "FIRST", "MIDDLE", "LAST",
false, names, "Special query for OtherPeople.dat");
person.WriteToDB(DB);
double AverageMilliseconds;
harvester.GetPublications(mockNCBI, ptc, person, StatusCallback, MessageCallback, InterruptCallback, out AverageMilliseconds);
// Verify that the data was written properly
int FoundPublications = 0;
Publications pubs = new Publications(DB, person, false);
if (pubs.PublicationList != null)
{
foreach (Publication pub in pubs.PublicationList)
{
FoundPublications++;
switch (pub.PMID)
{
case 12679283:
// The weird part of this publication is the second MeSH heading, which is very long
Assert.IsTrue(pub.MeSHHeadings.Count == 23);
Assert.IsTrue(pub.MeSHHeadings.Contains(
"Attention Deficit and Disruptive Behavior Disorders/etiology/*prevention & control/psychology"));
break;
case 2417121:
// The weird part of this publication is the date, which has a weird format that causes
// the day to be long
Assert.IsTrue(pub.Day == "19-1986 Jan 1");
break;
case 6148773:
// One of the headers is long
Assert.IsTrue(pub.MeSHHeadings.Contains("Peptide Fragments/antagonists & inhibitors/chemical synthesis/diagnostic use/isolation & purification/pharmacology/*physiology"));
break;
case 16291338:
// One of the authors is long
Assert.IsTrue(pub.Authors.Length == 8);
Assert.IsTrue(pub.Authors[7] == "For The Michigan Alliance For The National Children's Study");
break;
case 15451956:
// Volume is long
Assert.IsTrue(pub.Volume == "Suppl Web Exclusives");
break;
case 14653276:
// Issue is long
Assert.IsTrue(pub.Issue == "5 Suppl Nitric Oxide");
break;
case 9965612:
// Journal name is long
Assert.IsTrue(pub.Journal == "PHYSICAL REVIEW. E. STATISTICAL PHYSICS, PLASMAS, FLUIDS, AND RELATED INTERDISCIPLINARY TOPICS");
break;
case 9469584:
// Title is long
Assert.IsTrue(pub.Title ==
Database.Left("Down-regulation of cholesterol biosynthesis in sitosterolemia: diminished activities of acetoacetyl-CoA thiolase, 3-hydroxy-3-methylglutaryl-CoA synthase, reductase, squalene synthase, and 7-dehydrocholesterol delta7-reductase in liver and mononuclear leukocytes."
, 244));
break;
case 2545230:
// Month is long
Assert.IsTrue(pub.Month == "Spring-Summer");
break;
default:
break;
}
}
}
Assert.IsTrue(FoundPublications == 13);
}
public void TestMeSHHeadingReport()
{
// Set up the database with test publications (and don't forget to add the
// publication types!)
DB = new Database("Publication Harvester Unit Test");
Harvester harvester = new Harvester(DB);
harvester.CreateTables();
PublicationTypes PubTypes = new PublicationTypes(
AppDomain.CurrentDomain.BaseDirectory + "\\Unit Tests\\TestPublicationTypes",
"PublicationTypes.csv"
);
PubTypes.WriteToDB(DB);
reports = new Reports(DB, AppDomain.CurrentDomain.BaseDirectory + "\\Unit Tests\\TestReports\\pubmed_jifs.xls");
Assert.IsTrue(reports.Weights.Count == 10);
TestHarvester.GetPublicationsFromInput1XLS_Using_MockNCBI(false, new string[] { "eng" }, 22);
// Write the MeSH Heading report
StreamWriter writer = new StreamWriter(AppDomain.CurrentDomain.BaseDirectory + "\\MeSHHeadingReport.csv");
Reports.ReportStatus StatusCallback = delegate(int number, int total, Person person, bool ProgressBarOnly)
{
//
};
Reports.ReportMessage MessageCallback = delegate(string Message)
{
//
};
reports.MeSHHeadingReport(writer, StatusCallback, MessageCallback);
writer.Close();
// Verify that the MeSH headings were written properly
// Read the rows back from the file
string ConnectionString =
"Driver={Microsoft Text Driver (*.txt; *.csv)};Dbq="
+ AppDomain.CurrentDomain.BaseDirectory + ";";
OdbcConnection Connection = new OdbcConnection(ConnectionString);
OdbcDataAdapter DataAdapter = new OdbcDataAdapter
("SELECT * FROM [MeSHHeadingReport.csv]", Connection);
DataTable Results = new DataTable();
int Rows = DataAdapter.Fill(Results);
Connection.Close();
int numChecked = 0;
// Check a few selected results
foreach (DataRow Row in Results.Rows)
{
string Setnb = Row[0].ToString();
int Year = Convert.ToInt32(Row[1]);
string Heading = Row[2].ToString();
int Count = Convert.ToInt32(Row[3]);
switch (Setnb)
{
case "A6009400": // Van Eys
if ((Year == 1998) && (Heading == "Humans"))
{
Assert.IsTrue(Count == 2);
numChecked++;
}
if ((Year == 1998) && (Heading == "Child"))
{
Assert.IsTrue(Count == 1);
numChecked++;
}
if ((Year == 2001) && (Heading == "Humans"))
{
Assert.IsTrue(Count == 1);
numChecked++;
}
break;
case "A5702471": // Guillemin
if ((Year == 2005) && (Heading == "Hypothalamic Hormones/*physiology"))
{
Assert.IsTrue(Count == 1);
numChecked++;
}
break;
}
}
Assert.IsTrue(numChecked == 4);
File.Delete(AppDomain.CurrentDomain.BaseDirectory + "\\MeSHHeadingReport.csv");
}
public void TestGrantsReport()
{
// Set up the database with test publications (and don't forget to add the
// publication types!)
DB = new Database("Publication Harvester Unit Test");
Harvester harvester = new Harvester(DB);
harvester.CreateTables();
PublicationTypes PubTypes = new PublicationTypes(
AppDomain.CurrentDomain.BaseDirectory + "\\Unit Tests\\TestPublicationTypes",
"PublicationTypes.csv"
);
PubTypes.WriteToDB(DB);
reports = new Reports(DB, AppDomain.CurrentDomain.BaseDirectory + "\\Unit Tests\\TestReports\\pubmed_jifs.xls");
Assert.IsTrue(reports.Weights.Count == 10);
TestHarvester.GetPublicationsFromInput1XLS_Using_MockNCBI(false, new string[] { "eng" }, 22);
// Write the grants report
StreamWriter writer = new StreamWriter(AppDomain.CurrentDomain.BaseDirectory + "\\GrantsReport.csv");
reports.GrantsReport(writer);
writer.Close();
// Verify that the grants were written properly
// Read the rows back from the file
string ConnectionString =
"Driver={Microsoft Text Driver (*.txt; *.csv)};Dbq="
+ AppDomain.CurrentDomain.BaseDirectory + ";";
OdbcConnection Connection = new OdbcConnection(ConnectionString);
OdbcDataAdapter DataAdapter = new OdbcDataAdapter
("SELECT * FROM [GrantsReport.csv]", Connection);
DataTable Results = new DataTable();
int Rows = DataAdapter.Fill(Results);
Connection.Close();
int numChecked = 0;
// Check a few selected results
foreach (DataRow Row in Results.Rows)
{
int Year = Convert.ToInt32(Row[0]);
int PMID = Convert.ToInt32(Row[1]);
string GrantID = Row[2].ToString();
switch (PMID)
{
case 3086749: // Guillemin
numChecked++;
Assert.IsTrue((Year == 1986) && (
(GrantID == "AM-18811/AM/NIADDK") ||
(GrantID == "HD-09690/HD/NICHD") ||
(GrantID == "MH-00663/MH/NIMH") ||
(GrantID == "AG 03106/AG/NIA") ||
(GrantID == "DK-26741/DK/NIDDK")));
break;
case 9049886: // Van Eys
numChecked++;
Assert.IsTrue((Year == 1997) && (GrantID == "RO1-CA33097/CA/NCI"));
break;
}
}
Assert.IsTrue(numChecked == 6);
File.Delete(AppDomain.CurrentDomain.BaseDirectory + "\\GrantsReport.csv");
}
public void WriteAPersonTwice()
{
string[] Names = { "a", "b", "c" };
Person PersonToWrite = new Person("1234ABCD", "First", "Middle", "Last",
false, Names, "Medline search query");
Assert.IsTrue(PersonToWrite.Setnb == "1234ABCD");
Assert.IsTrue(PersonToWrite.First == "First");
Assert.IsTrue(PersonToWrite.Middle == "Middle");
Assert.IsTrue(PersonToWrite.Last == "Last");
Assert.IsTrue(PersonToWrite.Harvested == false);
Assert.IsTrue(PersonToWrite.Names.Length == 3);
Assert.IsTrue(PersonToWrite.Names[0] == "a");
Assert.IsTrue(PersonToWrite.Names[1] == "b");
Assert.IsTrue(PersonToWrite.Names[2] == "c");
Assert.IsTrue(PersonToWrite.MedlineSearch == "Medline search query");
// Write the person to the database
Database DB = new Database("Publication Harvester Unit Test");
Harvester harvester = new Harvester(DB);
harvester.CreateTables();
PersonToWrite.WriteToDB(DB);
// Read the person back from the database
DataTable PeopleFileData = new DataTable();
PeopleFileData = DB.ExecuteQuery("SELECT " + Database.PEOPLE_COLUMNS + " FROM People WHERE Setnb = '1234ABCD'");
Assert.IsTrue(PeopleFileData.Rows.Count == 1);
Person PersonToRead = new Person(PeopleFileData.Rows[0], PeopleFileData.Columns);
Assert.IsTrue(PersonToRead.Setnb == "1234ABCD");
Assert.IsTrue(PersonToRead.First == "First");
Assert.IsTrue(PersonToRead.Middle == "Middle");
Assert.IsTrue(PersonToRead.Last == "Last");
Assert.IsTrue(PersonToRead.Harvested == false);
Assert.IsTrue(PersonToRead.Names.Length == 3);
Assert.IsTrue(PersonToRead.Names[0] == "a");
Assert.IsTrue(PersonToRead.Names[1] == "b");
Assert.IsTrue(PersonToRead.Names[2] == "c");
Assert.IsTrue(PersonToRead.MedlineSearch == "Medline search query");
// Update the database, read him back again, and make sure there's still
// one row in People
PersonToRead.First = "NewFirst";
PersonToRead.Middle = "NewMiddle";
PersonToRead.Last = "NewLast";
PersonToRead.Names = new string[] { "d", "e", "f", "g" };
PersonToRead.MedlineSearch = "new query";
PersonToRead.Harvested = true;
PersonToRead.WriteToDB(DB);
DataTable Results = DB.ExecuteQuery("SELECT " + Database.PEOPLE_COLUMNS + " FROM People WHERE Setnb = '1234ABCD'");
Assert.IsTrue(Results.Rows.Count == 1);
Assert.IsTrue(Results.Rows[0]["Setnb"].ToString() == "1234ABCD");
Assert.IsTrue(Results.Rows[0]["First"].ToString() == "NewFirst");
Assert.IsTrue(Results.Rows[0]["Middle"].ToString() == "NewMiddle");
Assert.IsTrue(Results.Rows[0]["Last"].ToString() == "NewLast");
Assert.IsTrue(Results.Rows[0]["Name1"].ToString() == "d");
Assert.IsTrue(Results.Rows[0]["Name2"].ToString() == "e");
Assert.IsTrue(Results.Rows[0]["Name3"].ToString() == "f");
Assert.IsTrue(Results.Rows[0]["Name4"].ToString() == "g");
Assert.IsTrue(Results.Rows[0]["MedlineSearch"].ToString() == "new query");
bool boolValue;
Assert.IsTrue(Database.GetBoolValue(Results.Rows[0]["Harvested"], out boolValue));
Assert.IsTrue(boolValue);
// Make sure that Name2 through Name4 are updated properly when nulls are inserted
PersonToRead.Names = new string[] { "a name" };
PersonToRead.WriteToDB(DB);
Results = DB.ExecuteQuery("SELECT " + Database.PEOPLE_COLUMNS + " FROM People WHERE Setnb = '1234ABCD'");
Assert.IsTrue(Results.Rows[0]["Name1"].ToString() == "a name");
Assert.IsTrue(Results.Rows[0]["Name2"].Equals(DBNull.Value));
Assert.IsTrue(Results.Rows[0]["Name3"].Equals(DBNull.Value));
Assert.IsTrue(Results.Rows[0]["Name4"].Equals(DBNull.Value));
}
public void TestColleaguesSetUp()
{
// Create the AAMC roster object
roster = new Roster(AppDomain.CurrentDomain.BaseDirectory + "\\Test Data\\TestRoster\\testroster.csv");
// Stuff for GetPublications()
// Make an anonymous callback function that keeps track of the callback data
Harvester.GetPublicationsStatus StatusCallback = delegate(int number, int total, int averageTime)
{
//
};
// Make an anonymous callback function to do nothing for GetPublicationsMessage
Harvester.GetPublicationsMessage MessageCallback = delegate(string Message, bool StatusBarOnly)
{
//
};
// Make an anonymous callback function to return false for CheckForInterrupt
Harvester.CheckForInterrupt InterruptCallback = delegate()
{
return(false);
};
double AverageMilliseconds;
// Read the people file
People PeopleFromFile = new People(
AppDomain.CurrentDomain.BaseDirectory + "\\Test Data\\TestColleagues",
"PeopleFile.xls");
// Drop all tables from the test database
DB = new Database("Colleague Generator Unit Test");
foreach (string Table in new string[]
{
"colleaguepublications", "colleagues", "meshheadings",
"people", "peoplepublications", "publicationauthors",
"publicationgrants", "publicationmeshheadings", "publications",
"pubtypecategories", "starcolleagues"
}
)
{
DB.ExecuteNonQuery("DROP TABLE IF EXISTS " + Table + ";");
}
// Create the test database
harvester = new Harvester(DB);
harvester.CreateTables();
ColleagueFinder.CreateTables(DB);
// Populate it using the Mock NCBI object
ncbi = new MockNCBI("Medline");
PubTypes = new PublicationTypes(
AppDomain.CurrentDomain.BaseDirectory + "\\Test Data\\TestColleagues",
"PublicationTypes.csv"
);
// Write each person and his publications to the database
foreach (Person person in PeopleFromFile.PersonList)
{
person.WriteToDB(DB);
harvester.GetPublications(ncbi, PubTypes, person, StatusCallback, MessageCallback, InterruptCallback, out AverageMilliseconds);
}
}
public void GetPeopleFromInputXLS()
{
// Import input1.xls into the database
Database DB = new Database("Publication Harvester Unit Test");
Harvester harvester = new Harvester(DB);
harvester.CreateTables();
harvester.ImportPeople(AppDomain.CurrentDomain.BaseDirectory
+ "\\Unit Tests\\TestPeople\\input1.xls");
DataTable Results = DB.ExecuteQuery(
@"SELECT Setnb, First, Middle, Last, Name1, Name2, Name3, Name4, Name5, Name6,
MedlineSearch, Harvested, Error, ErrorMessage
FROM People"
);
// Test each person
for (int Row = 0; Row < Results.Rows.Count; Row++)
{
Person person = new Person(Results.Rows[Row], Results.Columns);
switch (person.Setnb)
{
case "A6009400":
Assert.IsTrue(person.First == "Jan");
Assert.IsTrue(person.Middle == "");
Assert.IsTrue(person.Last == "Van Eys");
Assert.IsTrue(person.Names.Length == 3);
Assert.IsTrue(person.Names[0] == "van eys j");
Assert.IsTrue(person.Names[1] == "vaneys j");
Assert.IsTrue(person.Names[2] == "eys jv");
Assert.IsTrue(person.MedlineSearch == "(\"van eys j\"[au] OR \"vaneys j\"[au] OR \"eys jv\"[au])");
break;
case "A5401532":
Assert.IsTrue(person.First == "Louis");
Assert.IsTrue(person.Middle == "");
Assert.IsTrue(person.Last == "Tobian");
Assert.IsTrue(person.Names.Length == 3);
Assert.IsTrue(person.Names[0] == "tobian l");
Assert.IsTrue(person.Names[1] == "tobian l jr");
Assert.IsTrue(person.Names[2] == "tobian lj");
Assert.IsTrue(person.MedlineSearch == "(\"tobian l\"[au] OR \"tobian l jr\"[au] OR \"tobian lj\"[au])");
break;
case "A5501586":
Assert.IsTrue(person.First == "Keith");
Assert.IsTrue(person.Middle == "B");
Assert.IsTrue(person.Last == "Reemtsma");
Assert.IsTrue(person.Names.Length == 6);
Assert.IsTrue(person.Names[0] == "reemtsma k");
Assert.IsTrue(person.Names[1] == "reemtsma kb");
Assert.IsTrue(person.Names[2] == "test data");
Assert.IsTrue(person.Names[3] == "more test data");
Assert.IsTrue(person.Names[4] == "test data name 5");
Assert.IsTrue(person.Names[5] == "test data name 6");
Assert.IsTrue(person.MedlineSearch == "((\"reemtsma k\"[au] OR \"reemtsma kb\"[au]) AND 1956:2000[dp])");
break;
case "A5702471":
Assert.IsTrue(person.First == "Roger");
Assert.IsTrue(person.Middle == "");
Assert.IsTrue(person.Last == "Guillemin");
Assert.IsTrue(person.Names.Length == 2);
Assert.IsTrue(person.Names[0] == "guillemin r");
Assert.IsTrue(person.Names[1] == "guillemin rc");
Assert.IsTrue(person.MedlineSearch ==
"(\"guillemin rc\"[au] OR (\"guillemin r\"[au] NOT (Electrodiagn Ther[ta] OR Phys Rev Lett[ta] OR vegas[ad] OR lindle[au])))"
);
break;
}
}
}
