private void Generate()
{
int lowestWidth = LowestWidth.IntValue;
int highestWidth = HighestWidth.IntValue;
int aspectRatioDeviation = AspectRatioDeviation.IntValue;
int lowestNbrImages = LowestNbrImages.IntValue;
int highestNbrImages = HighestNbrImages.IntValue;
int nbrTestsPerNbrImages = NbrTestsPerNbrImages.IntValue;
_showGenerationDetailsOfFailures = chkShowGenerationDetailsFailuresToo.Checked;
bool showGenerationDetails = _showGenerationDetailsOfFailures || chkShowGenerationDetails.Checked;
bool showCanvasImages = chkShowCanvasImages.Checked;
bool useFixedTests = chkFixed.Checked;
bool useRandomTests = chkRandom.Checked;
_cutoffEfficiency = CutoffEfficiency.FloatValue;
_maximumNbrCandidates = MaxCandidates.IntValue;
RandomSizeGenerator randomSizeGenerator =
new RandomSizeGenerator(lowestWidth, highestWidth, aspectRatioDeviation);
// ----------------------------------------
hlCanvasImages.Visible = showGenerationDetails;
phGenerationDetails.Visible = showGenerationDetails;
hlGraphs.Visible = true;
h1TestResults.Visible = true;
// ----------------------------------------
// add random tests
List<SingleTest> tests = new List<SingleTest>();
if (useRandomTests)
{
int testNbr = 1;
for (int nbrImages = lowestNbrImages; nbrImages <= highestNbrImages; nbrImages++)
{
for (int i = 0; i < nbrTestsPerNbrImages; i++)
{
SingleTest singleTest = new SingleTest();
singleTest.Heading = string.Format("Random test {0}, using {1} images", testNbr, nbrImages);
for (int j = 0; j < nbrImages; j++)
{
singleTest.ImageInfos.Add(randomSizeGenerator.Next());
}
tests.Add(singleTest);
testNbr++;
}
}
}
// ---------------------------------------------
// Add fixed tests with Mapper test cases
if (useFixedTests)
{
AddFixedTests(tests);
}
// -------------------------------------------
// Show generation details
rptrTests.Visible = showGenerationDetails;
if (showGenerationDetails)
{
_imageZoom = Convert.ToInt32(ImageZoom.IntValue);
_canvas = new CanvasWritingStepImages(_imageZoom, showCanvasImages);
rptrTests.DataSource = tests;
rptrTests.DataBind();
}
// -----------------------------------------
// Show aggregated statistics for multiple mappers
// Run all the test using a standard canvas rather than a canvas that writes images.
// That way, the timings are not distorted by the generation of the images.
ICanvas standardCanvas = new Canvas();
IMapper<Sprite>[] mappers =
{
new MapperHorizontalOnly<Sprite>(),
new MapperVerticalOnly<Sprite>(),
new MapperOptimalEfficiency<Sprite>(standardCanvas, _cutoffEfficiency, _maximumNbrCandidates)
};
// First run a few tests without recording results. The first few test runs always take way more
// ticks than normal, so doing this will keep bad data out of the results.
int nbrTests = tests.Count; if (nbrTests > 3) { nbrTests = 4; }
for (int i = 0; i < nbrTests; i++)
{
foreach (IMapper<Sprite> mapper in mappers)
{
SingleTestStats singleTestStats = new SingleTestStats();
TestUtils.RunTest<Sprite>(mapper, tests[i].ImageInfos, singleTestStats);
}
}
// Run all the tests. Use all mappers as listed above in the mappers declaration to run the tests.
// Dump the results to a file as they become available.
List<SingleTestStats> testResults = new List<SingleTestStats>();
string testResultsFilePath = Server.MapPath("~/testresults.csv");
using (StreamWriter outfile =
new StreamWriter(testResultsFilePath))
{
outfile.WriteLine(SingleTestStats.CsvHeader());
foreach (SingleTest singleTest in tests)
{
foreach(IMapper<Sprite> mapper in mappers)
{
SingleTestStats singleTestStats = new SingleTestStats();
TestUtils.RunTest<Sprite>(mapper, singleTest.ImageInfos, singleTestStats);
testResults.Add(singleTestStats);
outfile.WriteLine(singleTestStats.CsvDataLine());
outfile.Flush();
}
}
}
ltPathToCSV.Text =
string.Format(
@"A CSV file with all test results is at <a href=""{0}"">{0}</a>",
TestUtils.FilePathToUrl(testResultsFilePath));
// -----------------------------------------------
// Show overall results per mapper
var resultsPerMapper =
from t in testResults
group t by t.MapperType into g
select new {
Mapper = SingleTestStats.FriendlyMapperTypeName(g.Key),
AvgCandidateSpriteFails = g.Average(p => p.CandidateSpriteFails),
AvgCandidateSpritesGenerated = g.Average(p => p.CandidateSpritesGenerated),
AvgCanvasRectangleAddAttempts = g.Average(p => p.CanvasRectangleAddAttempts),
AvgCanvasNbrCellsGenerated = g.Average(p => p.CanvasNbrCellsGenerated),
AvgEfficiency = g.Average(p => p.Efficiency),
AvgElapsedTicks = g.Average(p => p.ElapsedTicks)
};
gvOverallResults.DataSource = resultsPerMapper;
gvOverallResults.DataBind();
// -----------------------------------------------
// Create graphs with the results
int statisticTypeIdx = 0;
foreach (int statisticTypeValue in Enum.GetValues(typeof(SingleTestStats.StatisticType)))
{
// For each statistic type (such as number of images, standard deviation of widths of the images in the sprite, etc.),
// we'll generate 2 charts, one for efficiency and one for elapsed ticks.
string statisticTypeName = Enum.GetName(typeof(SingleTestStats.StatisticType), statisticTypeIdx);
statisticTypeIdx++;
// ---------
string formattedStatisticTypeName = statisticTypeName.Replace('_', ' ');
Literal statisticTypeHeader = new Literal();
statisticTypeHeader.Text = string.Format("<h2>Results by {0}</h2>", formattedStatisticTypeName);
phCharts.Controls.Add(statisticTypeHeader);
string explanation = "";
SingleTestStats.StatisticType statisticType = (SingleTestStats.StatisticType) Enum.Parse(typeof(SingleTestStats.StatisticType), statisticTypeValue.ToString());
switch(statisticType)
{
case SingleTestStats.StatisticType.Nbr_Images:
explanation = "Shows speed and efficiency of the algorithms against the number of images in the sprite";
break;
case SingleTestStats.StatisticType.Width_Standard_Deviation:
explanation = "Shows speed and efficiency of the algorithms against the variability of the width of the images";
break;
case SingleTestStats.StatisticType.Heigth_Standard_Deviation:
explanation = "Shows speed and efficiency of the algorithms against the variability of the height of the images";
break;
case SingleTestStats.StatisticType.Area_Standard_Deviation:
explanation = "Shows speed and efficiency of the algorithms against the variability of the area of the images";
break;
case SingleTestStats.StatisticType.Aspect_Ratio_Standard_Deviation:
explanation = "Shows speed and efficiency of the algorithms against the variability of the aspect ratio of the images";
break;
}
Literal statisticExplanation = new Literal();
statisticExplanation.Text = string.Format("<p><small>{0}</small></p>", explanation);
phCharts.Controls.Add(statisticExplanation);
foreach(string resultName in SingleTestStats.ResultNames())
{
// Create a chart for each result - efficiency and elapsed ticks
Chart chart = new Chart();
chart.ImageType = ChartImageType.Png;
chart.ImageLocation="~/canvasimages/ChartPic_#SEQ(300,3)";
chart.Width = new Unit(1200, UnitType.Pixel);
chart.Height = new Unit(500, UnitType.Pixel);
chart.Legends.Add("Default");
chart.ChartAreas.Clear();
chart.ChartAreas.Add("ChartArea1");
chart.ChartAreas[0].AxisX.Title = formattedStatisticTypeName;
chart.ChartAreas[0].AxisY.Title = resultName;
chart.Series.Clear();
phCharts.Controls.Add(chart);
int seriesIdx = 0;
foreach(IMapper<Sprite> mapper in mappers)
{
Type mapperType = mapper.GetType();
var series =
from r in testResults
where r.MapperType == mapperType
group r by r.ImageStats[statisticTypeValue] into g
select new
{
Efficiency = g.Average(r => r.Efficiency),
ElapsedTicks = g.Average(r => r.ElapsedTicks),
XValue = g.Key
};
chart.Series.Add(SingleTestStats.FriendlyMapperTypeName(mapperType)); // Ensure the legend shows the type name
chart.Series[seriesIdx].MarkerSize=4;
chart.Series[seriesIdx].ChartType = SeriesChartType.Point;
chart.Series[seriesIdx].Points.DataBind(series, "XValue", resultName, "");
seriesIdx++;
}
}
}
}
public SeleniumTestViewModel(SingleTestStats stats)
{
TestStats = stats;
DataSetsJson = $"[{string.Join(",", TestStats.GetDatasetsJson())}]";
}
private void Generate()
{
int lowestWidth = LowestWidth.IntValue;
int highestWidth = HighestWidth.IntValue;
int aspectRatioDeviation = AspectRatioDeviation.IntValue;
int lowestNbrImages = LowestNbrImages.IntValue;
int highestNbrImages = HighestNbrImages.IntValue;
int nbrTestsPerNbrImages = NbrTestsPerNbrImages.IntValue;
_showGenerationDetailsOfFailures = chkShowGenerationDetailsFailuresToo.Checked;
bool showGenerationDetails = _showGenerationDetailsOfFailures || chkShowGenerationDetails.Checked;
bool showCanvasImages = chkShowCanvasImages.Checked;
bool useFixedTests = chkFixed.Checked;
bool useRandomTests = chkRandom.Checked;
_cutoffEfficiency = CutoffEfficiency.FloatValue;
_maximumNbrCandidates = MaxCandidates.IntValue;
RandomSizeGenerator randomSizeGenerator =
new RandomSizeGenerator(lowestWidth, highestWidth, aspectRatioDeviation);
// ----------------------------------------
hlCanvasImages.Visible = showGenerationDetails;
phGenerationDetails.Visible = showGenerationDetails;
hlGraphs.Visible = true;
h1TestResults.Visible = true;
// ----------------------------------------
// add random tests
List <SingleTest> tests = new List <SingleTest>();
if (useRandomTests)
{
int testNbr = 1;
for (int nbrImages = lowestNbrImages; nbrImages <= highestNbrImages; nbrImages++)
{
for (int i = 0; i < nbrTestsPerNbrImages; i++)
{
SingleTest singleTest = new SingleTest();
singleTest.Heading = string.Format("Random test {0}, using {1} images", testNbr, nbrImages);
for (int j = 0; j < nbrImages; j++)
{
singleTest.ImageInfos.Add(randomSizeGenerator.Next());
}
tests.Add(singleTest);
testNbr++;
}
}
}
// ---------------------------------------------
// Add fixed tests with Mapper test cases
if (useFixedTests)
{
AddFixedTests(tests);
}
// -------------------------------------------
// Show generation details
rptrTests.Visible = showGenerationDetails;
if (showGenerationDetails)
{
_imageZoom = Convert.ToInt32(ImageZoom.IntValue);
_canvas = new CanvasWritingStepImages(_imageZoom, showCanvasImages);
rptrTests.DataSource = tests;
rptrTests.DataBind();
}
// -----------------------------------------
// Show aggregated statistics for multiple mappers
// Run all the test using a standard canvas rather than a canvas that writes images.
// That way, the timings are not distorted by the generation of the images.
ICanvas standardCanvas = new Canvas();
IMapper <Sprite>[] mappers =
{
new MapperHorizontalOnly <Sprite>(),
new MapperVerticalOnly <Sprite>(),
new MapperOptimalEfficiency <Sprite>(standardCanvas,_cutoffEfficiency, _maximumNbrCandidates)
};
// First run a few tests without recording results. The first few test runs always take way more
// ticks than normal, so doing this will keep bad data out of the results.
int nbrTests = tests.Count; if (nbrTests > 3)
{
nbrTests = 4;
}
for (int i = 0; i < nbrTests; i++)
{
foreach (IMapper <Sprite> mapper in mappers)
{
SingleTestStats singleTestStats = new SingleTestStats();
TestUtils.RunTest <Sprite>(mapper, tests[i].ImageInfos, singleTestStats);
}
}
// Run all the tests. Use all mappers as listed above in the mappers declaration to run the tests.
// Dump the results to a file as they become available.
List <SingleTestStats> testResults = new List <SingleTestStats>();
string testResultsFilePath = Server.MapPath("~/testresults.csv");
using (StreamWriter outfile =
new StreamWriter(testResultsFilePath))
{
outfile.WriteLine(SingleTestStats.CsvHeader());
foreach (SingleTest singleTest in tests)
{
foreach (IMapper <Sprite> mapper in mappers)
{
SingleTestStats singleTestStats = new SingleTestStats();
TestUtils.RunTest <Sprite>(mapper, singleTest.ImageInfos, singleTestStats);
testResults.Add(singleTestStats);
outfile.WriteLine(singleTestStats.CsvDataLine());
outfile.Flush();
}
}
}
ltPathToCSV.Text =
string.Format(
@"A CSV file with all test results is at <a href=""{0}"">{0}</a>",
TestUtils.FilePathToUrl(testResultsFilePath));
// -----------------------------------------------
// Show overall results per mapper
var resultsPerMapper =
from t in testResults
group t by t.MapperType into g
select new {
Mapper = SingleTestStats.FriendlyMapperTypeName(g.Key),
AvgCandidateSpriteFails = g.Average(p => p.CandidateSpriteFails),
AvgCandidateSpritesGenerated = g.Average(p => p.CandidateSpritesGenerated),
AvgCanvasRectangleAddAttempts = g.Average(p => p.CanvasRectangleAddAttempts),
AvgCanvasNbrCellsGenerated = g.Average(p => p.CanvasNbrCellsGenerated),
AvgEfficiency = g.Average(p => p.Efficiency),
AvgElapsedTicks = g.Average(p => p.ElapsedTicks)
};
gvOverallResults.DataSource = resultsPerMapper;
gvOverallResults.DataBind();
// -----------------------------------------------
// Create graphs with the results
int statisticTypeIdx = 0;
foreach (int statisticTypeValue in Enum.GetValues(typeof(SingleTestStats.StatisticType)))
{
// For each statistic type (such as number of images, standard deviation of widths of the images in the sprite, etc.),
// we'll generate 2 charts, one for efficiency and one for elapsed ticks.
string statisticTypeName = Enum.GetName(typeof(SingleTestStats.StatisticType), statisticTypeIdx);
statisticTypeIdx++;
// ---------
string formattedStatisticTypeName = statisticTypeName.Replace('_', ' ');
Literal statisticTypeHeader = new Literal();
statisticTypeHeader.Text = string.Format("<h2>Results by {0}</h2>", formattedStatisticTypeName);
phCharts.Controls.Add(statisticTypeHeader);
string explanation = "";
SingleTestStats.StatisticType statisticType = (SingleTestStats.StatisticType)Enum.Parse(typeof(SingleTestStats.StatisticType), statisticTypeValue.ToString());
switch (statisticType)
{
case SingleTestStats.StatisticType.Nbr_Images:
explanation = "Shows speed and efficiency of the algorithms against the number of images in the sprite";
break;
case SingleTestStats.StatisticType.Width_Standard_Deviation:
explanation = "Shows speed and efficiency of the algorithms against the variability of the width of the images";
break;
case SingleTestStats.StatisticType.Heigth_Standard_Deviation:
explanation = "Shows speed and efficiency of the algorithms against the variability of the height of the images";
break;
case SingleTestStats.StatisticType.Area_Standard_Deviation:
explanation = "Shows speed and efficiency of the algorithms against the variability of the area of the images";
break;
case SingleTestStats.StatisticType.Aspect_Ratio_Standard_Deviation:
explanation = "Shows speed and efficiency of the algorithms against the variability of the aspect ratio of the images";
break;
}
Literal statisticExplanation = new Literal();
statisticExplanation.Text = string.Format("<p><small>{0}</small></p>", explanation);
phCharts.Controls.Add(statisticExplanation);
foreach (string resultName in SingleTestStats.ResultNames())
{
// Create a chart for each result - efficiency and elapsed ticks
Chart chart = new Chart();
chart.ImageType = ChartImageType.Png;
chart.ImageLocation = "~/canvasimages/ChartPic_#SEQ(300,3)";
chart.Width = new Unit(1200, UnitType.Pixel);
chart.Height = new Unit(500, UnitType.Pixel);
chart.Legends.Add("Default");
chart.ChartAreas.Clear();
chart.ChartAreas.Add("ChartArea1");
chart.ChartAreas[0].AxisX.Title = formattedStatisticTypeName;
chart.ChartAreas[0].AxisY.Title = resultName;
chart.Series.Clear();
phCharts.Controls.Add(chart);
int seriesIdx = 0;
foreach (IMapper <Sprite> mapper in mappers)
{
Type mapperType = mapper.GetType();
var series =
from r in testResults
where r.MapperType == mapperType
group r by r.ImageStats[statisticTypeValue] into g
select new
{
Efficiency = g.Average(r => r.Efficiency),
ElapsedTicks = g.Average(r => r.ElapsedTicks),
XValue = g.Key
};
chart.Series.Add(SingleTestStats.FriendlyMapperTypeName(mapperType)); // Ensure the legend shows the type name
chart.Series[seriesIdx].MarkerSize = 4;
chart.Series[seriesIdx].ChartType = SeriesChartType.Point;
chart.Series[seriesIdx].Points.DataBind(series, "XValue", resultName, "");
seriesIdx++;
}
}
}
}
