public void GivenRegions_WhenCallingGetAsPlainText_ThenTextOfAllWordsAreReturned()
{
//Arrange
var regions = new[]
{
new Region
{
Lines = new List <Line>()
{
new Line {
Words = new List <Word>()
{
new Word {
Text = "Test A"
}, new Word {
Text = "Test B"
}
}
},
new Line {
Words = new List <Word>()
{
new Word {
Text = "Test C"
}
}
}
}
},
new Region
{
Lines = new List <Line>()
{
new Line {
Words = new List <Word>()
{
new Word {
Text = "Test D"
}, new Word {
Text = "Test E"
}
}
},
}
}
};
var target = new RawAzureOcrResult()
{
Regions = regions
};
//Act
var result = target.GetAsPlainText();
//Assert
result.Should().Be("Test A Test B Test C Test D Test E");
}
public List <Line> Execute(RawAzureOcrResult ocrOutput)
{
if (ocrOutput == null)
{
throw new ArgumentNullException(nameof(ocrOutput));
}
return(ocrOutput.Regions?.ToList().SelectMany(x => x.Lines).OrderBy(x => x.Rectangle.Top).ToList());
}
public void GivenEmptyRawResult_WhenInvokingExecute_ThenNullIsReturned()
{
//Arrange
var ocrResult = new RawAzureOcrResult();
//Act
var orderedLines = _target.Execute(ocrResult);
//Assert
orderedLines.Should().BeNull();
}
public void GivenNoRegions_WhenCallingGetAsPlainText_ThenEmptyStringIsReturned()
{
//Arrange
var target = RawAzureOcrResult.CreateFrom(_resultMock.Object);
//Act
var result = target.GetAsPlainText();
//Assert
result.Should().BeEmpty();
}
public void GivenNoRegions_WhenCallingTextFound_ThenFalseIsReturned()
{
//Arrange
var target = RawAzureOcrResult.CreateFrom(_resultMock.Object);
//Act
var result = target.TextFound();
//Assert
result.Should().BeFalse();
}
public void GivenAProjectOxfordOcrResult_WhenInvokingCreateFrom_ThenPropertiesAreSetCorrectly()
{
//Arrange / Act
var result = RawAzureOcrResult.CreateFrom(_resultMock.Object);
//Assert
Assert.AreEqual(_language, result.Language);
Assert.AreEqual(_orientation, result.Orientation);
Assert.AreEqual(_regions, result.Regions);
Assert.AreEqual(_textAngle, result.TextAngle);
}
public void GivenRegions_WhenCallingTextFound_ThenTrueIsReturned()
{
//Arrange
_resultMock.Object.Regions = new [] { new Region() };
var target = RawAzureOcrResult.CreateFrom(_resultMock.Object);
//Act
var result = target.TextFound();
//Assert
result.Should().BeTrue();
}
public void GivenEmptyOcrResult_WhenInvokingExecute_ThenGetLinesAndSortIntoLogicalLinesAndTransformIntoSentencesAreCalledAnImageContentIsReturned()
{
//Arrange
var rawResult = new RawAzureOcrResult();
int height = 100, width = 100;
_getLinesOrderedByTopPosition.Setup(x => x.Execute(rawResult)).Returns((List <Line>)null);
//Act
var imageContent = _target.Execute(rawResult, height, width);
//Assert
imageContent.Should().NotBeNull();
imageContent.Sentences.Should().BeEmpty();
}
public void GivenRegions_WhenCallingGetAsPlainText_ThenTextOfAllWordsAreReturned()
{
//Arrange
_resultMock.Object.Regions = new[]
{
new Region
{
Lines = new[]
{
new Line {
Words = new[] { new Word {
Text = "Test A"
}, new Word {
Text = "Test B"
} }
},
new Line {
Words = new[] { new Word {
Text = "Test C"
} }
}
}
},
new Region
{
Lines = new[]
{
new Line {
Words = new[] { new Word {
Text = "Test D"
}, new Word {
Text = "Test E"
} }
},
}
}
};
var target = RawAzureOcrResult.CreateFrom(_resultMock.Object);
//Act
var result = target.GetAsPlainText();
//Assert
result.Should().Be("Test A Test B Test C Test D Test E");
}
public ImageContent Execute(RawAzureOcrResult ocrOutput, int height, int width)
{
if (ocrOutput == null)
{
throw new ArgumentNullException(nameof(ocrOutput));
}
var azureLines = _getLinesOrderedByTopPosition.Execute(ocrOutput);
if (azureLines == null)
{
return(new ImageContent(new List <ISentence>()));
}
var lines = _sortIntoLogicalLines.Execute(azureLines);
var sentences = _transformLinesIntoSentences.Execute(height, width, lines);
return(new ImageContent(sentences));
}
public void GivenRawResult_WhenInvokingExecute_ThenOrderedLinesAreReturned()
{
//Arrange
var topLine1 = new Line {
BoundingBox = "0,5,0,0"
};
var topLine2 = new Line {
BoundingBox = "0,200,0,0"
};
var topLine3 = new Line {
BoundingBox = "0,201,0,0"
};
var ocrResult = new RawAzureOcrResult()
{
Regions = new[]
{
new Region {
Lines = new List <Line>()
{
topLine2, topLine3
}
},
new Region {
Lines = new List <Line>()
{
topLine1
}
}
}
};
//Act
var orderedLines = _target.Execute(ocrResult);
//Assert
orderedLines.Should().NotBeNull();
orderedLines.Count.Should().Be(3);
orderedLines[0].Should().Be(topLine1);
orderedLines[1].Should().Be(topLine2);
orderedLines[2].Should().Be(topLine3);
}
private async Task <AzureOcrResult> DoOcr(Stream imageAsStream, FileFormatEnum fileFormatEnum, DateTime start)
{
try
{
var preprocessedResult = _ocrPreProcessing.AjustOrientationAndSize(imageAsStream, fileFormatEnum);
using (var stream = preprocessedResult.ImageFileStream)
{
ValidateImageProportions(preprocessedResult.NewImageHeight, preprocessedResult.NewImageWidth);
var entries = await _azureVisionApi.Execute(stream);
var rawAzureOcrResult = RawAzureOcrResult.CreateFrom(entries);
var content = _azureOcrParser.Execute(rawAzureOcrResult, preprocessedResult.NewImageHeight,
preprocessedResult.NewImageWidth);
return(AzureOcrResult.CreateSuccesResult(DateTime.Now.Subtract(start), content, rawAzureOcrResult));
}
}
catch (Exception e)
{
return(AzureOcrResult.CreateErrorResult(DateTime.Now.Subtract(start), e));
}
}
public void GivenOcrResult_WhenInvokingExecute_ThenGetLinesAndSortIntoLogicalLinesAndTransformIntoSentencesAreCalledAnImageContentIsReturned()
{
//Arrange
var rawResult = new RawAzureOcrResult();
int height = 100, width = 100;
var azureLines = new List <Line>();
_getLinesOrderedByTopPosition.Setup(x => x.Execute(rawResult)).Returns(azureLines);
var logicalLines = new Mock <IOrderedEnumerable <KeyValuePair <Point, List <Line> > > >().Object;
_sortIntoLogicalLines.Setup(x => x.Execute(azureLines)).Returns(logicalLines);
var sentences = new List <ISentence>();
_transformLinesIntoSenteces.Setup(x => x.Execute(height, width, logicalLines)).Returns(sentences);
//Act
var imageContent = _target.Execute(rawResult, height, width);
//Assert
Assert.AreEqual(sentences, imageContent.Sentences);
}
public void Setup()
{
_timespan = new TimeSpan();
_rawResult = new Mock <RawAzureOcrResult>().Object;
_imageContent = new Mock <IImageContent>();
}
