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 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 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
}