//private const string SegmentReadResults = "readResults";
//private const string SegmentLines = "lines";
//private const string SegmentWords = "words";
private static FormContent ResolveTextReference(ReadResult_internal readResult, string reference)
{
// TODO: Add additional validations here.
// https://github.com/Azure/azure-sdk-for-net/issues/10363
// Example: the following should result in LineIndex = 7, WordIndex = 12
// "#/readResults/3/lines/7/words/12"
string[] segments = reference.Split('/');
var lineIndex = int.Parse(segments[4], CultureInfo.InvariantCulture);
var wordIndex = int.Parse(segments[6], CultureInfo.InvariantCulture);
// TODO: Support case where text reference is lines only, without word segment
// https://github.com/Azure/azure-sdk-for-net/issues/10364
return(new FormWord(readResult.Lines.ToList()[lineIndex].Words.ToList()[wordIndex], readResult.Page));
// Code from Chris Stone below
//if (!string.IsNullOrEmpty(reference) && reference.Length > 2 && reference[0] == '#')
//{
// // offset by 2 to skip the '#/' prefix
// var segments = reference.Substring(2).Split('/');
// // must have an even number of segments
// if (segments.Length % 2 == 0)
// {
// int offset;
// for (var i = 0; i < segments.Length; i += 2)
// {
// // the next segment must be an integer
// if (int.TryParse(segments[i + 1], out offset))
// {
// var segment = segments[i];
// // We assume we're already on the correct page element
// //// this is the root page element
// //if (segment == SegmentReadResults)
// //{
// // readResult = results[offset];
// //}
//
// // this is a text element
// if (readResult != default)
// {
// if (segment == SegmentLines)
// {
// textElement = new RawExtractedLine(readResult.Lines.ToList()[offset]);
// }
// else if (segment == SegmentWords && textElement is RawExtractedLine)
// {
// textElement = (textElement as RawExtractedLine).Words[offset];
// }
// }
// }
// }
// }
//}
}
// TODO: Refactor to move OCR code to a common file, rather than it living in this file.
internal static IReadOnlyList <FormContent> ConvertTextReferences(ReadResult_internal readResult, IReadOnlyList <string> references)
{
List <FormContent> extractedTexts = new List <FormContent>();
foreach (var reference in references)
{
extractedTexts.Add(ResolveTextReference(readResult, reference));
}
return(extractedTexts);
}
// TODO: Refactor to move OCR code to a common file, rather than it living in this file.
internal static IReadOnlyList <RawExtractedItem> ConvertTextReferences(ReadResult_internal readResult, ICollection <string> references)
{
List <RawExtractedItem> extractedTexts = new List <RawExtractedItem>();
foreach (var reference in references)
{
extractedTexts.Add(ResolveTextReference(readResult, reference));
}
return(extractedTexts);
}
// Unsupervised
internal ExtractedPage(PageResult_internal pageResult, ReadResult_internal readResult)
{
PageNumber = pageResult.Page;
FormTypeId = pageResult.ClusterId;
Fields = ConvertFields(pageResult.KeyValuePairs, readResult);
Tables = ExtractedLayoutPage.ConvertTables(pageResult.Tables, readResult);
if (readResult != null)
{
RawExtractedPage = new RawExtractedPage(readResult);
}
}
internal ExtractedField(KeyValuePair_internal field, ReadResult_internal readResult)
{
Confidence = field.Confidence;
Name = field.Key.Text;
NameBoundingBox = field.Key.BoundingBox == null ? null : new BoundingBox(field.Key.BoundingBox);
if (field.Key.Elements != null)
{
NameRawExtractedItems = ConvertTextReferences(readResult, field.Key.Elements);
}
Value = field.Value.Text;
ValueBoundingBox = field.Value.BoundingBox == null ? null : new BoundingBox(field.Value.BoundingBox);
if (field.Value.Elements != null)
{
ValueRawExtractedItems = ConvertTextReferences(readResult, field.Value.Elements);
}
}
#pragma warning disable CA1801
internal FormField(KeyValuePair_internal field, ReadResult_internal readResult)
{
#pragma warning restore CA1801
//Confidence = field.Confidence;
//Name = field.Key.Text;
//NameBoundingBox = new BoundingBox(field.Key.BoundingBox);
//if (field.Key.Elements != null)
//{
// NameTextElements = ConvertTextReferences(readResult, field.Key.Elements);
//}
//Value = field.Value.Text;
//ValueBoundingBox = new BoundingBox(field.Value.BoundingBox);
//if (field.Value.Elements != null)
//{
// ValueTextElements = ConvertTextReferences(readResult, field.Value.Elements);
//}
}
internal ExtractedLabeledTable(DataTable_internal table, ReadResult_internal readResult)
: base(table, readResult)
{
}
private static IReadOnlyList <ExtractedField> ConvertFields(IReadOnlyList <KeyValuePair_internal> keyValuePairs, ReadResult_internal readResult)
{
List <ExtractedField> fields = new List <ExtractedField>();
foreach (var kvp in keyValuePairs)
{
ExtractedField field = new ExtractedField(kvp, readResult);
fields.Add(field);
}
return(fields);
}
internal ExtractedLabeledTable(DataTable_internal table, ReadResult_internal readResult, int pageNumber)
: base(table, readResult)
{
PageNumber = pageNumber;
}
