public void TestLanguages()
{
// GetPublicationsFromInput1XLS_Using_MockNCBI reads four people from
// Input1.xls and writes their publications to the datbase using
// MockNCBI. MockNCBI gets its test data from TestHarvester\*.dat.
// There are 22 English publications that belong to the four people
string[] Languages = { "eng" };
int ExpectedPublications = 22;
TestHarvester.GetPublicationsFromInput1XLS_Using_MockNCBI(false, Languages, ExpectedPublications);
// There's one Russian publication
Languages = new string[] { "rus" };
ExpectedPublications = 1;
TestHarvester.GetPublicationsFromInput1XLS_Using_MockNCBI(false, Languages, ExpectedPublications);
// There's one Spanish publication
Languages = new string[] { "spa" };
ExpectedPublications = 1;
TestHarvester.GetPublicationsFromInput1XLS_Using_MockNCBI(false, Languages, ExpectedPublications);
// There's one Russian and one Spanish publication -- two total
Languages = new string[] { "spa", "rus" };
ExpectedPublications = 2;
TestHarvester.GetPublicationsFromInput1XLS_Using_MockNCBI(false, Languages, ExpectedPublications);
// There are no French or Portugese publications
Languages = new string[] { "fre", "por" };
ExpectedPublications = 0;
TestHarvester.GetPublicationsFromInput1XLS_Using_MockNCBI(false, Languages, ExpectedPublications);
// There are 22 English and one Russian publication
Languages = new string[] { "eng", "rus" };
ExpectedPublications = 23;
TestHarvester.GetPublicationsFromInput1XLS_Using_MockNCBI(false, Languages, ExpectedPublications);
Languages = null;
ExpectedPublications = 24;
TestHarvester.GetPublicationsFromInput1XLS_Using_MockNCBI(false, Languages, ExpectedPublications);
}
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 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 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;
}
}
}
}
}
public void TestTwoPeopleWithSameNames()
{
Database DB = new Database("Publication Harvester Unit Test");
// Set up the database
TestHarvester.GetPublicationsFromInput1XLS_Using_MockNCBI(false, new string[] { "eng" }, 22);
// Add two people to the database with the same names and search criteria
// (where the search should make MockNCBI use OtherPeople.dat)
string[] names = new string[2];
names[0] = "Guy JF";
names[1] = "Guy J";
Person Joe = new Person("A1234567", "JOE", "FIRST", "GUY",
false, names, "Special query for OtherPeople.dat");
Joe.WriteToDB(DB);
Person Jane = new Person("Z7654321", "JANE", "FIFTH", "GUY",
false, names, "Special query for OtherPeople.dat");
Jane.WriteToDB(DB);
// Also add Jim, but give him an error so we can make sure it's cleared
Person Jim = new Person("Q2222222", "JIM", "FOURTEENTH", "GUY",
false, names, "Special query for OtherPeople.dat");
Jim.WriteToDB(DB);
Jim.WriteErrorToDB(DB, "This is an error message");
PublicationTypes ptc = new PublicationTypes(
AppDomain.CurrentDomain.BaseDirectory + "\\Unit Tests\\TestPublicationTypes",
"PublicationTypes.csv"
);
// 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
int MessageCallbacks = 0;
Harvester.GetPublicationsMessage MessageCallback = delegate(string Message, bool StatusBarOnly)
{
// Only increment MessageCallbacks if the message contains Joe's Setnb
// and the word "same"
if ((Message.Contains("A1234567") || (Message.Contains("Q2222222")) &&
Message.Contains("same")))
{
MessageCallbacks++;
}
};
// Make an anonymous callback function to return false for CheckForInterrupt
Harvester.CheckForInterrupt InterruptCallback = delegate()
{
return(false);
};
// More stuff for the harvester
Harvester harvester = new Harvester(DB);
MockNCBI mockNCBI = new MockNCBI("medline");
double AverageMilliseconds;
// Harvest the people
harvester.GetPublications(mockNCBI, ptc, Jane, StatusCallback, MessageCallback, InterruptCallback, out AverageMilliseconds);
// Make sure the harvester got Jane's publications
DataTable Results = DB.ExecuteQuery("SELECT PMID FROM PeoplePublications WHERE Setnb = 'Z7654321'");
Assert.AreEqual(Results.Rows.Count, 3);
foreach (DataRow Row in Results.Rows)
{
Assert.IsTrue(
(Row["PMID"].ToString() == "2417121") ||
(Row["PMID"].ToString() == "12679283") ||
(Row["PMID"].ToString() == "14653276"));
}
ArrayList Parameters = new ArrayList();
Parameters.Add(Database.Parameter(Jane.Setnb));
Results = DB.ExecuteQuery("SELECT Harvested, Error, ErrorMessage FROM People WHERE Setnb = ?", Parameters);
Assert.AreEqual(Results.Rows[0]["Harvested"], true);
Assert.AreEqual(Results.Rows[0]["Error"], DBNull.Value);
Assert.AreEqual(Results.Rows[0]["ErrorMessage"].ToString(), "");
// It should also get Joe's publications. It should call MessageCallback()
// twice to let us know Joe'and Jim's s publications were found, and it
// should add the appropriate rows to PeoplePublications.
Assert.AreEqual(MessageCallbacks, 2);
Results = DB.ExecuteQuery("SELECT PMID FROM PeoplePublications WHERE Setnb = 'A1234567'");
Assert.AreEqual(Results.Rows.Count, 3);
foreach (DataRow Row in Results.Rows)
{
Assert.IsTrue(
(Row["PMID"].ToString() == "2417121") ||
(Row["PMID"].ToString() == "12679283") ||
(Row["PMID"].ToString() == "14653276"));
}
Parameters = new ArrayList();
Parameters.Add(Database.Parameter(Joe.Setnb));
Results = DB.ExecuteQuery("SELECT " + Database.PEOPLE_COLUMNS + " FROM People WHERE Setnb = ?", Parameters);
bool boolValue; // needed for GetBoolValue workaround for bit field bug in MySQL
Assert.IsTrue(Database.GetBoolValue(Results.Rows[0]["Harvested"], out boolValue));
Assert.IsTrue(boolValue);
Assert.IsTrue(Database.GetBoolValue(Results.Rows[0]["Error"], out boolValue));
Assert.IsFalse(boolValue);
Assert.AreEqual(Results.Rows[0]["Error"], DBNull.Value);
Assert.AreEqual(Results.Rows[0]["ErrorMessage"].ToString(), "");
// It should also get Jim's publications -- and it should also clear his error.
Assert.AreEqual(MessageCallbacks, 2);
Results = DB.ExecuteQuery("SELECT PMID FROM PeoplePublications WHERE Setnb = 'A1234567'");
Assert.AreEqual(Results.Rows.Count, 3);
foreach (DataRow Row in Results.Rows)
{
Assert.IsTrue(
(Row["PMID"].ToString() == "2417121") ||
(Row["PMID"].ToString() == "12679283") ||
(Row["PMID"].ToString() == "14653276"));
}
Parameters = new ArrayList();
Parameters.Add(Database.Parameter(Jim.Setnb));
Results = DB.ExecuteQuery("SELECT " + Database.PEOPLE_COLUMNS + "FROM People WHERE Setnb = ?", Parameters);
Assert.IsTrue(Database.GetBoolValue(Results.Rows[0]["Harvested"], out boolValue));
Assert.AreEqual(boolValue, true);
Assert.AreEqual(Results.Rows[0]["Error"], DBNull.Value);
Assert.AreEqual(Results.Rows[0]["ErrorMessage"].ToString(), "");
}
public void TestNCBIError()
{
// Set up the database -- tell MockNCBI to throw an error
// This function does all of the necessary testing.
TestHarvester.GetPublicationsFromInput1XLS_Using_MockNCBI(false, new string[] { "eng" }, 22);
}
public void TestClearingDataAfterInterruption()
{
// Set up the database
TestHarvester.GetPublicationsFromInput1XLS_Using_MockNCBI(false, new string[] { "eng" }, 22);
Database DB = new Database("Publication Harvester Unit Test");
Harvester harvester = new Harvester(DB);
// Add a grant for publication 12462241, since it doesn't have one
ArrayList Parameters = new ArrayList();
Parameters.Add(Database.Parameter(12462241));
Parameters.Add(Database.Parameter("Fake grant ID"));
DB.ExecuteNonQuery("INSERT INTO PublicationGrants (PMID, GrantID) VALUES ( ? , ? )", Parameters);
// Verify that InterruptedDataExists and UnharvestedPeopleExist both return false
Assert.IsFalse(harvester.InterruptedDataExists());
Assert.IsFalse(harvester.UnharvestedPeopleExist());
// Verify that there are publications for Tobian (setnb A5401532)
// (there should be 5 publications)
Assert.IsTrue(DB.GetIntValue(
@"SELECT Count(*)
FROM PeoplePublications
WHERE Setnb = 'A5401532'") == 5);
int TotalPublicationsBeforeClear
= DB.GetIntValue("SELECT Count(*) FROM PeoplePublications");
// Verify that there are authors and MeSH headings for publication 12462241
// (which belongs to Guillemin, and should have 82 authors and 9 headings)
Assert.IsTrue(DB.GetIntValue(
@"SELECT Count(*)
FROM PublicationMeSHHeadings
WHERE PMID = 12462241") == 9);
int TotalHeadingsBeforeClear
= DB.GetIntValue("SELECT Count(*) FROM PublicationMeSHHeadings");
Assert.IsTrue(DB.GetIntValue(
@"SELECT Count(*)
FROM PublicationAuthors
WHERE PMID = 12462241") == 82);
int TotalAuthorsBeforeClear
= DB.GetIntValue("SELECT Count(*) FROM PublicationAuthors");
Assert.IsTrue(DB.GetIntValue(
@"SELECT Count(*)
FROM PublicationGrants
WHERE PMID = 12462241") == 1);
int TotalGrantsBeforeClear
= DB.GetIntValue("SELECT Count(*) FROM PublicationGrants");
// Set Tobian's Harvested to 0
DB.ExecuteNonQuery("UPDATE People SET Harvested = 0 WHERE Setnb = 'A5401532'");
// Remove publication 12462241
DB.ExecuteNonQuery("DELETE FROM Publications WHERE PMID = 12462241");
// Verify that InterruptedDataExists and UnharvestedPeopleExist both return true
Assert.IsTrue(harvester.InterruptedDataExists());
Assert.IsTrue(harvester.UnharvestedPeopleExist());
// Execute ClearDataAfterInterruption
harvester.ClearDataAfterInterruption();
// Verify that InterruptedDataExists returns false again, now that
// the intterrupted data has been cleared, but that there are still
// unharvested people
Assert.IsFalse(harvester.InterruptedDataExists());
Assert.IsTrue(harvester.UnharvestedPeopleExist());
// Verify that only Tobian's publications have been removed
Assert.IsTrue(
DB.GetIntValue("SELECT Count(*) FROM PeoplePublications")
== TotalPublicationsBeforeClear - 5);
// Verify that only publication 12462241's grants, headings and authors have been removed
Assert.IsTrue(
DB.GetIntValue("SELECT Count(*) FROM PublicationMeSHHeadings")
== TotalHeadingsBeforeClear - 9);
Assert.IsTrue(
DB.GetIntValue("SELECT Count(*) FROM PublicationAuthors")
== TotalAuthorsBeforeClear - 82);
Assert.IsTrue(
DB.GetIntValue("SELECT Count(*) FROM PublicationGrants")
== TotalGrantsBeforeClear - 1);
}
public void GetPublications()
{
// Set up the database
TestHarvester.GetPublicationsFromInput1XLS_Using_MockNCBI(false, new string[] { "eng" }, 22);
Database DB = new Database("Publication Harvester Unit Test");
PublicationTypes ptc = new PublicationTypes(
AppDomain.CurrentDomain.BaseDirectory + "\\Unit Tests\\TestPublicationTypes",
"PublicationTypes.csv"
);
// Verify the correct publications were written (including publication type and author position)
People people = new People(DB);
Assert.IsTrue(people.PersonList.Count == 4);
int FoundPublications = 0;
foreach (Person person in people.PersonList)
{
Publications pubs = new Publications(DB, person, false);
switch (person.Setnb)
{
case "A6009400": // Van Eys
Assert.IsTrue(pubs.PublicationList.Length == 8);
if (pubs.PublicationList != null)
{
foreach (Publication pub in pubs.PublicationList)
{
switch (pub.PMID)
{
case 9876482:
FoundPublications++;
Assert.IsTrue(pub.Title == "Benefits of nutritional intervention on nutritional status, quality of life and survival.");
Assert.IsTrue(pub.Pages == "66-8");
Assert.IsTrue(pub.Year == 1998);
Assert.IsTrue(pub.Month == null);
Assert.IsTrue(pub.Day == null);
Assert.IsTrue(pub.Journal == "Int J Cancer Suppl");
Assert.IsTrue(pub.Volume == "11");
Assert.IsTrue(pub.Issue == null);
Assert.IsTrue(pub.Authors.Length == 1);
Assert.IsTrue(pub.Authors[0] == "Van Eys J");
// Verify publication type
Assert.IsTrue(pub.PubType == "Journal Article");
Assert.IsTrue(ptc.GetCategoryNumber(pub.PubType) == 3);
// Verify MeSH headings
Assert.IsTrue(pub.MeSHHeadings.Count == 8);
Assert.IsTrue(pub.MeSHHeadings.Contains("Child"));
Assert.IsTrue(pub.MeSHHeadings.Contains("Nutrition Disorders/*complications/*therapy"));
Assert.IsTrue(pub.MeSHHeadings.Contains("Survival Rate"));
// Verify position type
Harvester.AuthorPositions PositionType;
int AuthorPosition = Publications.GetAuthorPosition(DB, pub.PMID, person, out PositionType, "PeoplePublications");
Assert.IsTrue(AuthorPosition == 1);
Assert.IsTrue(PositionType == Harvester.AuthorPositions.First);
AuthorPosition = person.GetAuthorPosition(DB, pub, out PositionType);
Assert.IsTrue(AuthorPosition == 1);
Assert.IsTrue(PositionType == Harvester.AuthorPositions.First);
break;
case 8403744:
FoundPublications++;
Assert.IsTrue(pub.Title == "Early hospital discharge and the timing of newborn metabolic screening.");
Assert.IsTrue(pub.Pages == "463-6");
Assert.IsTrue(pub.Year == 1993);
Assert.IsTrue(pub.Month == "Aug");
Assert.IsTrue(pub.Day == null);
Assert.IsTrue(pub.Journal == "Clin Pediatr (Phila)");
Assert.IsTrue(pub.Volume == "32");
Assert.IsTrue(pub.Issue == "8");
Assert.IsTrue(pub.Authors.Length == 4);
Assert.IsTrue(pub.Authors[0] == "Coody D");
Assert.IsTrue(pub.Authors[1] == "Yetman RJ");
Assert.IsTrue(pub.Authors[2] == "Montgomery D");
Assert.IsTrue(pub.Authors[3] == "van Eys J");
// Verify publication type
Assert.IsTrue(pub.PubType == "Consensus Development Conference, NIH");
Assert.IsTrue(ptc.GetCategoryNumber(pub.PubType) == 1);
// Verify MeSH headings
Assert.IsTrue(pub.MeSHHeadings.Count == 15);
Assert.IsTrue(pub.MeSHHeadings.Contains("Cesarean Section"));
Assert.IsTrue(pub.MeSHHeadings.Contains("Hospitals, Private"));
Assert.IsTrue(pub.MeSHHeadings.Contains("*Insurance, Health"));
Assert.IsTrue(pub.MeSHHeadings.Contains("United States"));
// Verify position type
AuthorPosition = Publications.GetAuthorPosition(DB, pub.PMID, person, out PositionType, "PeoplePublications");
Assert.IsTrue(AuthorPosition == 4);
Assert.IsTrue(PositionType == Harvester.AuthorPositions.Last);
AuthorPosition = person.GetAuthorPosition(DB, pub, out PositionType);
Assert.IsTrue(AuthorPosition == 4);
Assert.IsTrue(PositionType == Harvester.AuthorPositions.Last);
break;
}
}
}
break;
case "A5401532": // Tobian
Assert.IsTrue(pubs.PublicationList.Length == 5);
if (pubs.PublicationList != null)
{
foreach (Publication pub in pubs.PublicationList)
{
switch (pub.PMID)
{
case 9931073:
FoundPublications++;
Assert.IsTrue(pub.Title == "Story of the birth of the journal called Hypertension.");
Assert.IsTrue(pub.Pages == "7");
Assert.IsTrue(pub.Year == 1999);
Assert.IsTrue(pub.Month == "Jan");
Assert.IsTrue(pub.Day == null);
Assert.IsTrue(pub.Volume == "33");
Assert.IsTrue(pub.Issue == "1");
Assert.IsTrue(pub.Authors.Length == 1);
Assert.IsTrue(pub.Authors[0] == "Tobian L");
// Verify publication type
Assert.IsTrue(pub.PubType == "Historical Article");
Assert.IsTrue(ptc.GetCategoryNumber(pub.PubType) == 0);
// Verify MeSH headings
Assert.IsTrue(pub.MeSHHeadings.Count == 5);
Assert.IsTrue(pub.MeSHHeadings.Contains("American Heart Association/*history"));
Assert.IsTrue(pub.MeSHHeadings.Contains("*Hypertension"));
Assert.IsTrue(pub.MeSHHeadings.Contains("United States"));
// Verify position type
Harvester.AuthorPositions PositionType;
int AuthorPosition = Publications.GetAuthorPosition(DB, pub.PMID, person, out PositionType, "PeoplePublications");
Assert.IsTrue(AuthorPosition == 1);
Assert.IsTrue(PositionType == Harvester.AuthorPositions.First);
AuthorPosition = person.GetAuthorPosition(DB, pub, out PositionType);
Assert.IsTrue(AuthorPosition == 1);
Assert.IsTrue(PositionType == Harvester.AuthorPositions.First);
break;
}
}
}
break;
case "A5501586": // Reemtsma
Assert.IsTrue(pubs.PublicationList.Length == 3);
if (pubs.PublicationList != null)
{
foreach (Publication pub in pubs.PublicationList)
{
switch (pub.PMID)
{
case 11528018:
FoundPublications++;
Assert.IsTrue(pub.Title == "Xenotransplantation: A Historical Perspective.");
Assert.IsTrue(pub.Pages == "9-12");
Assert.IsTrue(pub.Year == 1995);
Assert.IsTrue(pub.Month == null);
Assert.IsTrue(pub.Day == null);
Assert.IsTrue(pub.Volume == "37");
Assert.IsTrue(pub.Issue == "1");
Assert.IsTrue(pub.Authors.Length == 1);
Assert.IsTrue(pub.Authors[0] == "Reemtsma K");
// Verify publication type
Assert.IsTrue(pub.PubType == "JOURNAL ARTICLE");
Assert.IsTrue(ptc.GetCategoryNumber(pub.PubType) == 3);
// Verify MeSH headings
Assert.IsTrue(pub.MeSHHeadings == null);
// Verify position type
Harvester.AuthorPositions PositionType;
int AuthorPosition = Publications.GetAuthorPosition(DB, pub.PMID, person, out PositionType, "PeoplePublications");
Assert.IsTrue(AuthorPosition == 1);
Assert.IsTrue(PositionType == Harvester.AuthorPositions.First);
AuthorPosition = person.GetAuthorPosition(DB, pub, out PositionType);
Assert.IsTrue(AuthorPosition == 1);
Assert.IsTrue(PositionType == Harvester.AuthorPositions.First);
break;
}
}
}
break;
case "A5702471": // Guillemin
Assert.IsTrue(pubs.PublicationList.Length == 6);
if (pubs.PublicationList != null)
{
foreach (Publication pub in pubs.PublicationList)
{
switch (pub.PMID)
{
case 15642779:
// For this publication, we're just concerned that
// the publication type is "Review" -- even though
// it's the second publication type in the citation,
// it's flagged as an "override first pubtype"
// in PublicationTypes.csv
Assert.IsTrue(pub.PubType == "Review");
break;
// NOTE: The title has a quote (laureates') that gets stripped off
case 12462241:
FoundPublications++;
Assert.IsTrue(pub.Title == "Nobel laureates letter to President Bush.");
Assert.IsTrue(pub.Pages == "A02");
Assert.IsTrue(pub.Year == 2001);
Assert.IsTrue(pub.Month == "Feb");
Assert.IsTrue(pub.Day == "22");
Assert.IsTrue(pub.Journal == "Washington Post");
Assert.IsTrue(pub.Volume == null);
Assert.IsTrue(pub.Issue == null);
Assert.IsTrue(pub.Authors.Length == 82);
Assert.IsTrue(pub.Authors[0] == "Arrow KJ");
Assert.IsTrue(pub.Authors[26] == "Guillemin R");
Assert.IsTrue(pub.Authors[81] == "Wilson RW");
// Verify publication type
Assert.IsTrue(pub.PubType == "Newspaper Article");
Assert.IsTrue(ptc.GetCategoryNumber(pub.PubType) == 0);
// Verify MeSH headings
Assert.IsTrue(pub.MeSHHeadings.Count == 9);
Assert.IsTrue(pub.MeSHHeadings.Contains("Embryo Disposition"));
Assert.IsTrue(pub.MeSHHeadings.Contains("National Institutes of Health (U.S.)"));
Assert.IsTrue(pub.MeSHHeadings.Contains("United States"));
// Verify position type
Harvester.AuthorPositions PositionType;
int AuthorPosition = Publications.GetAuthorPosition(DB, pub.PMID, person, out PositionType, "PeoplePublications");
Assert.IsTrue(AuthorPosition == 27);
Assert.IsTrue(PositionType == Harvester.AuthorPositions.Middle);
AuthorPosition = person.GetAuthorPosition(DB, pub, out PositionType);
Assert.IsTrue(AuthorPosition == 27);
Assert.IsTrue(PositionType == Harvester.AuthorPositions.Middle);
break;
case 3086749:
// This publication was altered to contain six GrantIDs in order to
// test the GrantID column length in the database
FoundPublications++;
Assert.IsTrue(pub.Title == "Pituitary FSH is released by a heterodimer of the beta-subunits from the two forms of inhibin.");
Assert.IsTrue(pub.Pages == "779-82");
Assert.IsTrue(pub.Year == 1986);
Assert.IsTrue(pub.Month == "Jun");
Assert.IsTrue(pub.Day == "19-25");
Assert.IsTrue(pub.Journal == "Nature");
Assert.IsTrue(pub.Volume == "321");
Assert.IsTrue(pub.Issue == "6072");
Assert.IsTrue(pub.Authors.Length == 7);
Assert.IsTrue(pub.Authors[0] == "Ling N");
Assert.IsTrue(pub.Authors[4] == "Esch F");
Assert.IsTrue(pub.Authors[6] == "Guillemin R");
// Verify publication type
Assert.IsTrue(pub.PubType == "Journal Article");
Assert.IsTrue(ptc.GetCategoryNumber(pub.PubType) == 3);
// Verify MeSH headings
Assert.IsTrue(pub.MeSHHeadings.Count == 14);
Assert.IsTrue(pub.MeSHHeadings.Contains("Amino Acid Sequence"));
Assert.IsTrue(pub.MeSHHeadings.Contains("Follicle Stimulating Hormone/*secretion"));
Assert.IsTrue(pub.MeSHHeadings.Contains("Swine"));
// Verify position type
AuthorPosition = Publications.GetAuthorPosition(DB, pub.PMID, person, out PositionType, "PeoplePublications");
Assert.IsTrue(AuthorPosition == 7);
Assert.IsTrue(PositionType == Harvester.AuthorPositions.Last);
AuthorPosition = person.GetAuthorPosition(DB, pub, out PositionType);
Assert.IsTrue(AuthorPosition == 7);
Assert.IsTrue(PositionType == Harvester.AuthorPositions.Last);
break;
}
}
}
break;
}
}
Assert.IsTrue(FoundPublications == 6);
// Verify that People.Harvested has been updated for each person
DataTable Results = DB.ExecuteQuery("SELECT Setnb, Harvested FROM People");
Assert.IsTrue(Results.Rows.Count == 4);
foreach (DataRow Row in Results.Rows)
{
Assert.IsTrue((bool)Row["Harvested"] == true);
}
}
