// USE CASE: Configuring fields for better recognition results
public static void Configuring_fields_for_better_recognition_results(IEngine engine)
{
trace("Create a Document Definition from a FlexiLayout...");
IDocumentDefinition newDefinition = engine.CreateDocumentDefinitionFromAFL(SamplesFolder + "\\SampleMisc\\Invoice_eng.afl", "English");
assert(newDefinition != null);
trace("Configure data types...");
setFieldValueType(newDefinition, "InvoiceDate", FieldValueTypeEnum.FVT_DateTime);
setFieldValueType(newDefinition, "Quantity", FieldValueTypeEnum.FVT_Number);
setFieldValueType(newDefinition, "UnitPrice", FieldValueTypeEnum.FVT_Currency);
setFieldValueType(newDefinition, "Total", FieldValueTypeEnum.FVT_Currency);
setFieldValueType(newDefinition, "TotalAmount", FieldValueTypeEnum.FVT_Currency);
trace("Configure recognition languages for text fields ...");
IFieldDefinition fieldDef = findFieldDef(newDefinition, "InvoiceNumber");
assert(fieldDef != null);
ITextRecognitionParams textParams = fieldDef.RecognitionParams.AsTextParams();
ILanguage newLanguage = textParams.CreateEmbeddedLanguageByDataType(FieldValueTypeEnum.FVT_DateTime);
textParams.Language = newLanguage;
newLanguage = textParams.CreateEmbeddedLanguage(textParams.Language.Type, textParams.Language);
assert(newLanguage != textParams.Language);
assert(newLanguage.LanguageCategory == LanguageCategoryEnum.LC_DataType);
assert(newLanguage.DatatypeCategory == DatatypeCategoryEnum.TC_DateTime);
textParams.Language = newLanguage;
newLanguage = textParams.CreateEmbeddedLanguage(LanguageTypeEnum.LT_Group, null);
newLanguage.AsGroupLanguage().Add(engine.PredefinedLanguages.FindLanguage("English"));
newLanguage.AsGroupLanguage().Add(engine.PredefinedLanguages.FindLanguage("Russian"));
textParams.Language = newLanguage;
assert(textParams.Language.Type == LanguageTypeEnum.LT_Group);
assert(textParams.Language.AsGroupLanguage().Count == 2);
assert(textParams.Language.AsGroupLanguage().Item(0).InternalName == "English");
assert(textParams.Language.AsGroupLanguage().Item(1).InternalName == "Russian");
newLanguage = textParams.CreateEmbeddedLanguage(LanguageTypeEnum.LT_Simple, null);
newLanguage.AsSimpleLanguage().set_LetterSet(LanguageLetterSetEnum.LLS_Alphabet, "ABCDEFGHIJKLMNOPQRSTUVWXYZ");
newLanguage.AsSimpleLanguage().RegularExpression = "[A-Z]{1-}";
textParams.Language = newLanguage;
assert(textParams.Language.AsSimpleLanguage().RegularExpression.Length > 0);
newLanguage = textParams.CreateEmbeddedLanguage(LanguageTypeEnum.LT_Simple, engine.PredefinedLanguages.FindLanguage("English"));
assert(newLanguage.AsSimpleLanguage().UsePredefinedDictionary == true);
assert(newLanguage.AsSimpleLanguage().UseUserDefinedDictionary == false);
assert(newLanguage.AsSimpleLanguage().UserDefinedDictionary == null);
newLanguage.AsSimpleLanguage().UseUserDefinedDictionary = true;
FCEngine.IDictionary dictionary = newLanguage.AsSimpleLanguage().UserDefinedDictionary;
assert(dictionary != null);
assert(dictionary.WordsCount == 0);
dictionary.AddWord("ONE", 1);
dictionary.AddWord("TWO", 1);
dictionary.AddWord("THREE", 1);
assert(dictionary.WordsCount == 3);
IEnumDictionaryWords enumWords = dictionary.EnumWords();
for (int i = 0; i < 10; i++)
{
int confidence = 0;
string word = enumWords.Next(out confidence);
if (confidence == 0)
{
break;
}
trace(word);
}
textParams.Language = newLanguage;
trace("Check the Document Definition...");
assert(newDefinition.Check() == true);
traceBegin("Use the Document Definition in FlexiCaptureProcessor...");
IFlexiCaptureProcessor processor = engine.CreateFlexiCaptureProcessor();
processor.AddDocumentDefinition(newDefinition);
// Add images for a single document
processor.AddImageFile(SamplesFolder + "\\SampleImages\\Invoices_1.tif");
// Recognize the document
IDocument document = processor.RecognizeNextDocument();
assert(document != null);
assert(document.DocumentDefinition != null);
assert(document.Pages.Count == 1);
processor.ExportDocumentEx(document, SamplesFolder + "\\FCEExport", "Invoice", null);
traceEnd("OK");
}
