private static async Task <AnnotatedPage> CombineMetadata(OcrResult ocr, OcrResult hw, AnalysisResult vis, EntityLink[] cia, ImageReference img)
{
// The handwriting result also included OCR text but OCR will produce better results on typed documents
// so take the result that produces the most text. Consider combining them by region to take the best of each.
var result = hw.Text.Length > ocr.Text.Length ? ocr : hw;
// create metadata for the vision caption and tags
var captionLines = vis.Description.Captions.Select(caption => new lineResult()
{
words = caption
.Text
.Split(' ')
.Select(word => new WordResult()
{
text = word
})
.ToArray()
});
var tagLines = new[] { new lineResult()
{
words = new[] { "(" }
.Concat(vis.Tags.Select(t => t.Name))
.Concat(new[] { ")" })
.Select(t => new WordResult()
{
text = t
}).ToArray()
} };
var newResult = new OcrResult()
{
lines = result
.lines
.Concat(captionLines)
.Concat(tagLines).ToArray()
};
// rotate the image if needed
var pageImg = ocr.Orientation == "Up" || ocr.Orientation == "NotDetected"
? img
: await img.GetImage().Rotate(ocr.Orientation).UploadMedia(blobContainer);
// TODO: merge annotations for Linked Entities and Cryptonyms
return(new AnnotatedPage(newResult, pageImg));
}
