private string ParseTextDefinition(TextDefinition textDefinition, OcrResult ocrResult)
{
string text = string.Empty;
switch (textDefinition.GetAs)
{
case GetTextAs.Text:
text = ocrResult.AsString();
break;
case GetTextAs.Continuous:
text = ocrResult.AsContinuousText();
break;
}
if (textDefinition.Replace != null && textDefinition.Replace.Length > 0)
{
foreach (var replaceDefinition in textDefinition.Replace)
{
text = ParseReplaceDefinition(replaceDefinition, text);
}
}
return(text);
}
public IEnumerable <AnalyzedText> Analyze(OcrResult ocrResult)
{
var matchedText = Constants.TextAnalysisConfiguration.LettersRegexes
.Select(x => Regex.Matches(ocrResult.AsString(), x))
.SelectMany(y => y.Cast <Match>()
.Select(x => x.Value));
return(matchedText.Select(x => new AnalyzedText()
{
Text = x,
TextType = TextType.Letters.ToString(),
BoundingBox = ocrResult.BoundingBox
}));
}
private string[] ParseFromTextTypeDefintion(TextTypeDefinition textTypeDefinition, OcrResult ocrResult)
{
if (string.IsNullOrWhiteSpace(textTypeDefinition.Name))
{
throw new Exception("Could not analyze custom text type definition, a definition is missing a name");
}
if (textTypeDefinition.Regexes.Length == 0)
{
return(new string[] { });
}
if (ocrResult.AsString() is var text && textTypeDefinition.Text != null)
{
text = ParseTextDefinition(textTypeDefinition.Text, ocrResult);
}
var matches = textTypeDefinition.Regexes
.Select(x => Regex.Matches(text, x, RegexOptions.IgnoreCase))
.SelectMany(x => x.Cast <Match>())
.Select(x => x.Value).ToArray();
return(matches);
}
public static IEnumerable <string> AllWords(this OcrResult ocrResult)
{
return(Regex.Split(ocrResult.AsString().Replace(',', ' '), @"\s+"));
}