public void GeoCoderCodeAll_PerfsTests_TimeToCodeAll()
{
const string dbLocation1 = @"TestGeoLocationTool.sdf";
connection = DBHelper.GetDbConnection(dbLocation1);
GeoCoder geoCoder = new GeoCoder(connection);
geoCoder.LoadGazetteerFile(@"TestGaz1.csv");
Stopwatch watch = new Stopwatch();
geoCoder.SetGazetteerColumns(
new GazetteerColumnHeaders
{
Level1Code = "ID_1",
Level2Code = "ID_2",
Level3Code = "ID_3",
Level1Name = "NAME_1",
Level2Name = "NAME_2",
Level3Name = "NAME_3",
Level1AltName = "VARNAME_1",
Level2AltName = "VARNAME_2",
Level3AltName = "VARNAME_3"
},
false);
geoCoder.LoadInputFileCsv("TestInput1.csv");
geoCoder.SetInputColumns(geoCoder.DefaultInputColumnHeaders());
watch.Start();
geoCoder.AddAllLocationCodes();
Debug.WriteLine(
geoCoder.InputData.Rows.Count + " input lines: " +
watch.Elapsed.TotalSeconds);
// Example results
// 25 input lines: 0.072016
}
public void GeoCoder_PerfsTestsUsingMatchedNamesCache()
{
const string dbLocation1 = @"TestGeoLocationTool.sdf";
connection = DBHelper.GetDbConnection(dbLocation1);
GeoCoder geoCoder = new GeoCoder(connection);
geoCoder.LoadGazetteerFile(@"TestGaz1.csv");
Stopwatch watch = new Stopwatch();
geoCoder.SetGazetteerColumns(
new GazetteerColumnHeaders
{
Level1Code = "ID_1",
Level2Code = "ID_2",
Level3Code = "ID_3",
Level1Name = "NAME_1",
Level2Name = "NAME_2",
Level3Name = "NAME_3"
},
false);
foreach (
string inputFile in
new[]
{@"TestInput1000.csv", @"TestInput10000.csv", @"TestInput50000.csv"})
{
geoCoder.LoadInputFileCsv(inputFile);
geoCoder.SetInputColumns(geoCoder.DefaultInputColumnHeaders());
// use cache
watch.Restart();
InputData.UseMatchedNamesCache = true;
geoCoder.AddAllLocationCodes();
var elapsed = watch.Elapsed.TotalSeconds;
// don't use cache
geoCoder.LoadInputFileCsv(inputFile);
geoCoder.SetInputColumns(geoCoder.DefaultInputColumnHeaders());
watch.Restart();
InputData.UseMatchedNamesCache = false;
geoCoder.AddAllLocationCodes();
Debug.WriteLine(
"input file: " + inputFile + " cached: " + elapsed +
" vs " + "non cached: " + watch.Elapsed.TotalSeconds);
}
// Example results
// input file: TestInput1000.csv cached: 0.0089728 vs non cached: 0.3835499
// input file: TestInput10000.csv cached: 0.073879 vs non cached: 2.8980328
// input file: TestInput50000.csv cached: 0.3852506 vs non cached: 14.464649
}
public void CodeAll_CorrectAndMissSpeltInputWithSavedMatchesForMissSpellings_AllCodesAdded()
{
// arrange
GeoCoder geoCoder =
new GeoCoder(MockRepository.GenerateStub<IDbConnection>() as DbConnection);
InputColumnHeaders inputColumnHeaders = InputColumnNames();
GazetteerColumnHeaders gazetteerColumnHeaders = GazetteerColumnNames();
// create input test data with
// line 1, all names correct
// line 2, all names correct
// line 3, all names miss-spelt
string[] names1 = {"P1", "T1", "V1"};
string[] names2 = {"P2", "T2", "V2"};
string[] names3 = {"P1x", "T1x", "V1x"};
string[] codes1 = {"1", "10", "100"};
string[] codes2 = {"2", "20", "200"};
InputTestData inputTestData = new InputTestData();
inputTestData.AddLine(names1);
inputTestData.AddLine(names2);
inputTestData.AddLine(names3);
geoCoder.SetInputData(inputTestData.Data(inputColumnHeaders));
geoCoder.SetInputColumns(inputColumnHeaders);
// create gazetteer data
GazetteerRecords gazetteerRecords = new GazetteerRecords();
gazetteerRecords.AddLine(names1, codes1);
gazetteerRecords.AddLine(names2, codes2);
geoCoder.SetGazetteerData(gazetteerRecords.Data(gazetteerColumnHeaders));
// add records matching input line 3 names to gazetteer names 1
MatchProviderTestData matchProviderTestData = new MatchProviderTestData();
matchProviderTestData.AddLevel1(names3, names1);
matchProviderTestData.AddLevel2(names3, names1);
matchProviderTestData.AddLevel3(names3, names1);
MatchProviderStub matchProviderStub = new MatchProviderStub(matchProviderTestData);
geoCoder.SetMatchProvider(matchProviderStub.MatchProvider());
geoCoder.SetGazetteerColumns(gazetteerColumnHeaders, false);
// act
geoCoder.AddAllLocationCodes();
// assert
var columns = geoCoder.LocationCodeColumnHeaders();
//line 1 - should contain codes 1
DataRow line1 = geoCoder.InputData.Rows[0];
Assert.AreEqual(codes1[0], line1[columns.Level1]);
Assert.AreEqual(codes1[1], line1[columns.Level2]);
Assert.AreEqual(codes1[2], line1[columns.Level3]);
//line 2 - should contain codes 2
DataRow line2 = geoCoder.InputData.Rows[1];
Assert.AreEqual(codes2[0], line2[columns.Level1]);
Assert.AreEqual(codes2[1], line2[columns.Level2]);
Assert.AreEqual(codes2[2], line2[columns.Level3]);
//line 3 - should contain codes 1
DataRow line3 = geoCoder.InputData.Rows[2];
Assert.AreEqual(codes1[0], line3[columns.Level1]);
Assert.AreEqual(codes1[1], line3[columns.Level2]);
Assert.AreEqual(codes1[2], line3[columns.Level3]);
}
public void GeoCoder_PerfsTestsUsingDictionaries()
{
connection = DBHelper.GetDbConnection(dbLocation);
connection.InitializeDB();
GeoCoder geoCoder = new GeoCoder(connection);
geoCoder.LoadGazetteerFile(@"PHL_adm3.csv");
//You need to copy this file manually
Stopwatch watch = new Stopwatch();
watch.Start();
geoCoder.SetGazetteerColumns(
new GazetteerColumnHeaders
{
Level1Code = "ID_1",
Level2Code = "ID_2",
Level3Code = "ID_3",
Level1Name = "NAME_1",
Level2Name = "NAME_2",
Level3Name = "NAME_3"
},
false);
Debug.WriteLine("Time to create dictionaries: " + watch.Elapsed.TotalSeconds);
foreach (int linesCount in new[] {500, 1000, 2000})
{
geoCoder.LoadInputFileCsv(GenerateInputFile(linesCount));
geoCoder.SetInputColumns(geoCoder.DefaultInputColumnHeaders());
watch.Restart();
geoCoder.AddAllLocationCodes();
var elapsed = watch.Elapsed.TotalSeconds;
//LocationCodes.useDictionaries = !LocationCodes.useDictionaries;
geoCoder.LoadInputFileCsv(GenerateInputFile(linesCount));
geoCoder.SetInputColumns(geoCoder.DefaultInputColumnHeaders());
watch.Restart();
geoCoder.AddAllLocationCodes();
Debug.WriteLine(
linesCount + " input lines: " + elapsed + " vs " +
watch.Elapsed.TotalSeconds);
// LocationCodes.useDictionaries = !LocationCodes.useDictionaries;
foreach (var row in geoCoder.InputData.AsEnumerable())
{
var elems = row.ItemArray;
Assert.IsFalse(elems[5] is DBNull);
Assert.IsFalse(elems[6] is DBNull);
Assert.IsFalse(elems[7] is DBNull);
}
}
}
public void CodeAll_InputContainMissSpellingsWithDifferentCasingToSavedMatches_AllCodesAdded()
{
// arrange
GeoCoder geoCoder =
new GeoCoder(MockRepository.GenerateStub<IDbConnection>() as DbConnection);
InputColumnHeaders inputColumnHeaders = InputColumnNames();
GazetteerColumnHeaders gazetteerColumnHeaders = GazetteerColumnNames();
//gazetteer data
string[] names1 = { "P1", "T1", "V1" };
string[] codes1 = { "1", "10", "100" };
// saved matched names data
string[] names2 = { "P1x", "T1x", "V1x" };
// input data
// line 1, all names miss-spelt with different casing
string[] names3 = {"p1x", "t1x", "v1x"};
InputTestData inputTestData = new InputTestData();
inputTestData.AddLine(names3);
geoCoder.SetInputData(inputTestData.Data(inputColumnHeaders));
geoCoder.SetInputColumns(inputColumnHeaders);
// create gazetteer data
GazetteerRecords gazetteerRecords = new GazetteerRecords();
gazetteerRecords.AddLine(names1, codes1);
geoCoder.SetGazetteerData(gazetteerRecords.Data(gazetteerColumnHeaders));
// add records matched names records
MatchProviderTestData matchProviderTestData = new MatchProviderTestData();
// add records matching saved matched names to gazetteer names
matchProviderTestData.AddLevel1(names2, names1);
matchProviderTestData.AddLevel2(names2, names1);
matchProviderTestData.AddLevel3(names2, names1);
MatchProviderStub matchProviderStub = new MatchProviderStub(matchProviderTestData);
geoCoder.SetMatchProvider(matchProviderStub.MatchProvider());
geoCoder.SetGazetteerColumns(gazetteerColumnHeaders, false);
// act
geoCoder.AddAllLocationCodes();
// assert
var columns = geoCoder.LocationCodeColumnHeaders();
//line 1 - should contain codes 1
DataRow line1 = geoCoder.InputData.Rows[0];
Assert.AreEqual(codes1[0], line1[columns.Level1]);
Assert.AreEqual(codes1[1], line1[columns.Level2]);
Assert.AreEqual(codes1[2], line1[columns.Level3]);
}