private static async Task TestLocal(string endpoint, string apiKey, string modelId, string filePath)
{
FormRecognizerClient client = new FormRecognizerClient(new Uri(endpoint), new AzureKeyCredential(apiKey));
using var stream = new FileStream(filePath, FileMode.Open);
RecognizeCustomFormsOperation operation = await client.StartRecognizeCustomFormsAsync(modelId, stream);
Response <RecognizedFormCollection> operationResponse = await operation.WaitForCompletionAsync();
RecognizedFormCollection forms = operationResponse.Value;
foreach (RecognizedForm form in forms)
{
Console.WriteLine($"Form of type: {form.FormType}");
Console.WriteLine($"Form was analyzed with model with ID: {form.ModelId}");
foreach (FormField field in form.Fields.Values)
{
Console.WriteLine($"Field '{field.Name}': ");
if (field.LabelData != null)
{
Console.WriteLine($" Label: '{field.LabelData.Text}'");
}
Console.WriteLine($" Value: '{field.ValueData.Text}'");
Console.WriteLine($" Confidence: '{field.Confidence}'");
}
}
}
public async Task OutputModelsTrainedWithoutLabels()
{
string endpoint = TestEnvironment.Endpoint;
string apiKey = TestEnvironment.ApiKey;
string trainingFileUrl = TestEnvironment.BlobContainerSasUrl;
string formFilePath = FormRecognizerTestEnvironment.CreatePath("Form_1.jpg");
FormRecognizerClient client = new FormRecognizerClient(new Uri(endpoint), new AzureKeyCredential(apiKey));
FormTrainingClient trainingClient = new FormTrainingClient(new Uri(endpoint), new AzureKeyCredential(apiKey));
// Model trained without labels
CustomFormModel modelTrainedWithoutLabels = await trainingClient.StartTrainingAsync(new Uri(trainingFileUrl), useTrainingLabels : false).WaitForCompletionAsync();
using (FileStream stream = new FileStream(formFilePath, FileMode.Open))
{
RecognizedFormCollection forms = await client.StartRecognizeCustomFormsAsync(modelTrainedWithoutLabels.ModelId, stream).WaitForCompletionAsync();
// With a form recognized by a model trained without labels, the 'field.Name' property will be denoted
// by a numeric index. To look for the labels identified during the training step,
// use the `field.LabelText` property.
Console.WriteLine("---------Recognizing forms using models trained without labels---------");
foreach (RecognizedForm form in forms)
{
Console.WriteLine($"Form of type: {form.FormType}");
foreach (FormField field in form.Fields.Values)
{
Console.WriteLine($"Field {field.Name}: ");
if (field.LabelData != null)
{
Console.WriteLine($" Label: '{field.LabelData.Text}");
}
Console.WriteLine($" Value: '{field.ValueData.Text}");
Console.WriteLine($" Confidence: '{field.Confidence}");
}
}
// Find the value of unlabeled fields.
foreach (RecognizedForm form in forms)
{
// Find the value of a specific unlabeled field.
Console.WriteLine("Find the value for a specific unlabeled field:");
foreach (FormField field in form.Fields.Values)
{
if (field.LabelData != null && field.LabelData.Text == "Vendor Name:")
{
Console.WriteLine($"The Vendor Name is {field.ValueData.Text}");
}
}
// Find the value of unlabeled fields with specific words
Console.WriteLine("Find the value for labeled field with specific words:");
form.Fields.Values.Where(field => field.LabelData.Text.StartsWith("Ven"))
.ToList().ForEach(v => Console.WriteLine($"{v.LabelData.Text} is {v.ValueData.Text}"));
form.Fields.Values.Where(field => field.LabelData.Text.Contains("Name"))
.ToList().ForEach(v => Console.WriteLine($"{v.LabelData.Text} is {v.ValueData.Text}"));
}
}
}
public async Task RecognizeCustomFormsFromFile()
{
string endpoint = TestEnvironment.Endpoint;
string apiKey = TestEnvironment.ApiKey;
string trainingFileUrl = TestEnvironment.BlobContainerSasUrl;
// Firstly, create a trained model we can use to recognize the custom form. Please note that
// models can also be trained using a graphical user interface such as the Form Recognizer
// Labeling Tool found here:
// https://docs.microsoft.com/azure/cognitive-services/form-recognizer/quickstarts/label-tool
FormTrainingClient trainingClient = new FormTrainingClient(new Uri(endpoint), new AzureKeyCredential(apiKey));
CustomFormModel model = await trainingClient.StartTrainingAsync(new Uri(trainingFileUrl), useTrainingLabels : false, "My Model").WaitForCompletionAsync();
// Proceed with the custom form recognition.
FormRecognizerClient client = new FormRecognizerClient(new Uri(endpoint), new AzureKeyCredential(apiKey));
#region Snippet:FormRecognizerRecognizeCustomFormsFromFile
#if SNIPPET
string modelId = "<modelId>";
string filePath = "<filePath>";
#else
string filePath = FormRecognizerTestEnvironment.CreatePath("Form_1.jpg");
string modelId = model.ModelId;
#endif
using var stream = new FileStream(filePath, FileMode.Open);
RecognizeCustomFormsOperation operation = await client.StartRecognizeCustomFormsAsync(modelId, stream);
Response <RecognizedFormCollection> operationResponse = await operation.WaitForCompletionAsync();
RecognizedFormCollection forms = operationResponse.Value;
foreach (RecognizedForm form in forms)
{
Console.WriteLine($"Form of type: {form.FormType}");
Console.WriteLine($"Form was analyzed with model with ID: {form.ModelId}");
foreach (FormField field in form.Fields.Values)
{
Console.WriteLine($"Field '{field.Name}': ");
if (field.LabelData != null)
{
Console.WriteLine($" Label: '{field.LabelData.Text}'");
}
Console.WriteLine($" Value: '{field.ValueData.Text}'");
Console.WriteLine($" Confidence: '{field.Confidence}'");
}
}
#endregion
// Delete the model on completion to clean environment.
await trainingClient.DeleteModelAsync(model.ModelId);
}
public async Task <List <dynamic> > ScanLocal(string classificationKey, string filePath)
{
string classificationModelId = await _distributedCache.GetStringAsync(classificationKey);
using var stream = new FileStream(filePath, FileMode.Open);
RecognizeCustomFormsOperation operation = await _client.StartRecognizeCustomFormsAsync(classificationModelId, stream);
return(await Evaluate(operation));
}
public async Task OutputModelsTrainedWithLabels()
{
string endpoint = TestEnvironment.Endpoint;
string apiKey = TestEnvironment.ApiKey;
string trainingFileUrl = TestEnvironment.BlobContainerSasUrl;
string formFilePath = FormRecognizerTestEnvironment.CreatePath("Form_1.jpg");
FormRecognizerClient client = new FormRecognizerClient(new Uri(endpoint), new AzureKeyCredential(apiKey));
FormTrainingClient trainingClient = new FormTrainingClient(new Uri(endpoint), new AzureKeyCredential(apiKey));
// Model trained with labels
CustomFormModel modelTrainedWithLabels = await trainingClient.StartTrainingAsync(new Uri(trainingFileUrl), useTrainingLabels : true, new TrainingOptions()
{
ModelDisplayName = "My Model with labels"
}).WaitForCompletionAsync();
using (FileStream stream = new FileStream(formFilePath, FileMode.Open))
{
RecognizedFormCollection forms = await client.StartRecognizeCustomFormsAsync(modelTrainedWithLabels.ModelId, stream).WaitForCompletionAsync();
// With a form recognized by a model trained with labels, the 'field.Name' key will be the label
// that you gave it at training time.
// Note that Label data is not returned for model trained with labels, as the trained model
// contains this information and therefore the service returns the value of the recognized label.
Console.WriteLine("---------Recognizing forms using models trained with labels---------");
foreach (RecognizedForm form in forms)
{
Console.WriteLine($"Form of type: {form.FormType}");
Console.WriteLine($"Form has form type confidence: {form.FormTypeConfidence.Value}");
Console.WriteLine($"Form was analyzed with model with ID: {form.ModelId}");
foreach (FormField field in form.Fields.Values)
{
Console.WriteLine($"Field {field.Name}: ");
Console.WriteLine($" Value: '{field.ValueData.Text}");
Console.WriteLine($" Confidence: '{field.Confidence}");
}
}
// Find labeled field.
foreach (RecognizedForm form in forms)
{
// Find the specific labeled field.
Console.WriteLine("Find the value for a specific labeled field:");
if (form.Fields.TryGetValue("VendorName", out FormField field))
{
Console.WriteLine($"VendorName is {field.ValueData.Text}");
}
// Find labeled fields with specific words
Console.WriteLine("Find the value for labeled field with specific words:");
form.Fields.Where(kv => kv.Key.StartsWith("Ven"))
.ToList().ForEach(v => Console.WriteLine($"{v.Key} is {v.Value.ValueData.Text}"));
form.Fields.Where(kv => kv.Key.Contains("Name"))
.ToList().ForEach(v => Console.WriteLine($"{v.Key} is {v.Value.ValueData.Text}"));
}
}
}
public async Task RecognizeCustomFormsFromFile()
{
string endpoint = TestEnvironment.Endpoint;
string apiKey = TestEnvironment.ApiKey;
string trainingFileUrl = TestEnvironment.BlobContainerSasUrl;
// Firstly, create a trained model we can use to recognize the custom form. Please note that
// models can also be trained using a graphical user interface such as the Form Recognizer
// Labeling Tool found here:
// https://docs.microsoft.com/azure/cognitive-services/form-recognizer/label-tool?tabs=v2-1
FormTrainingClient trainingClient = new FormTrainingClient(new Uri(endpoint), new AzureKeyCredential(apiKey));
CustomFormModel model = await trainingClient.StartTrainingAsync(new Uri(trainingFileUrl), useTrainingLabels : false, "My Model").WaitForCompletionAsync();
// Proceed with the custom form recognition.
FormRecognizerClient client = new FormRecognizerClient(new Uri(endpoint), new AzureKeyCredential(apiKey));
#region Snippet:FormRecognizerRecognizeCustomFormsFromFile
#if SNIPPET
string modelId = "<modelId>";
string filePath = "<filePath>";
#else
string filePath = FormRecognizerTestEnvironment.CreatePath("Form_1.jpg");
string modelId = model.ModelId;
#endif
using var stream = new FileStream(filePath, FileMode.Open);
var options = new RecognizeCustomFormsOptions()
{
IncludeFieldElements = true
};
RecognizeCustomFormsOperation operation = await client.StartRecognizeCustomFormsAsync(modelId, stream, options);
Response <RecognizedFormCollection> operationResponse = await operation.WaitForCompletionAsync();
RecognizedFormCollection forms = operationResponse.Value;
foreach (RecognizedForm form in forms)
{
Console.WriteLine($"Form of type: {form.FormType}");
Console.WriteLine($"Form was analyzed with model with ID: {form.ModelId}");
foreach (FormField field in form.Fields.Values)
{
Console.WriteLine($"Field '{field.Name}': ");
if (field.LabelData != null)
{
Console.WriteLine($" Label: '{field.LabelData.Text}'");
}
Console.WriteLine($" Value: '{field.ValueData.Text}'");
Console.WriteLine($" Confidence: '{field.Confidence}'");
}
// Iterate over tables, lines, and selection marks on each page
foreach (var page in form.Pages)
{
for (int i = 0; i < page.Tables.Count; i++)
{
Console.WriteLine($"Table {i+1} on page {page.Tables[i].PageNumber}");
foreach (var cell in page.Tables[i].Cells)
{
Console.WriteLine($" Cell[{cell.RowIndex}][{cell.ColumnIndex}] has text '{cell.Text}' with confidence {cell.Confidence}");
}
}
Console.WriteLine($"Lines found on page {page.PageNumber}");
foreach (var line in page.Lines)
{
Console.WriteLine($" Line {line.Text}");
}
if (page.SelectionMarks.Count != 0)
{
Console.WriteLine($"Selection marks found on page {page.PageNumber}");
foreach (var selectionMark in page.SelectionMarks)
{
Console.WriteLine($" Selection mark is '{selectionMark.State}' with confidence {selectionMark.Confidence}");
}
}
}
}
#endregion
// Delete the model on completion to clean environment.
await trainingClient.DeleteModelAsync(model.ModelId);
}
public async Task CreateComposedModel()
{
string endpoint = TestEnvironment.Endpoint;
string apiKey = TestEnvironment.ApiKey;
string trainingFileUrl = TestEnvironment.BlobContainerSasUrl;
FormTrainingClient client = new FormTrainingClient(new Uri(endpoint), new AzureKeyCredential(apiKey));
string purchaseOrderOfficeSuppliesUrl = trainingFileUrl;
string purchaseOrderOfficeEquipmentUrl = trainingFileUrl;
string purchaseOrderFurnitureUrl = trainingFileUrl;
string purchaseOrderCleaningSuppliesUrl = trainingFileUrl;
#region Snippet:FormRecognizerSampleTrainVariousModels
// For this sample, you can use the training forms found in the `trainingFiles` folder.
// Upload the forms to your storage container and then generate a container SAS URL.
// For instructions on setting up forms for training in an Azure Storage Blob Container, see
// https://docs.microsoft.com/azure/cognitive-services/form-recognizer/build-training-data-set#upload-your-training-data
//@@ string purchaseOrderOfficeSuppliesUrl = "<purchaseOrderOfficeSupplies>";
//@@ string purchaseOrderOfficeEquipmentUrl = "<purchaseOrderOfficeEquipment>";
//@@ string purchaseOrderFurnitureUrl = "<purchaseOrderFurniture>";
//@@ string purchaseOrderCleaningSuppliesUrl = "<purchaseOrderCleaningSupplies>";
CustomFormModel purchaseOrderOfficeSuppliesModel = (await client.StartTrainingAsync(new Uri(purchaseOrderOfficeSuppliesUrl), useTrainingLabels: true, "Purchase order - Office supplies").WaitForCompletionAsync()).Value;
CustomFormModel purchaseOrderOfficeEquipmentModel = (await client.StartTrainingAsync(new Uri(purchaseOrderOfficeEquipmentUrl), useTrainingLabels: true, "Purchase order - Office Equipment").WaitForCompletionAsync()).Value;
CustomFormModel purchaseOrderFurnitureModel = (await client.StartTrainingAsync(new Uri(purchaseOrderFurnitureUrl), useTrainingLabels: true, "Purchase order - Furniture").WaitForCompletionAsync()).Value;
CustomFormModel purchaseOrderCleaningSuppliesModel = (await client.StartTrainingAsync(new Uri(purchaseOrderCleaningSuppliesUrl), useTrainingLabels: true, "Purchase order - Cleaning Supplies").WaitForCompletionAsync()).Value;
#endregion
#region Snippet:FormRecognizerSampleCreateComposedModel
List <string> modelIds = new List <string>()
{
purchaseOrderOfficeSuppliesModel.ModelId,
purchaseOrderOfficeEquipmentModel.ModelId,
purchaseOrderFurnitureModel.ModelId,
purchaseOrderCleaningSuppliesModel.ModelId
};
CustomFormModel purchaseOrderModel = (await client.StartCreateComposedModelAsync(modelIds, "Composed Purchase order").WaitForCompletionAsync()).Value;
Console.WriteLine($"Purchase Order Model Info:");
Console.WriteLine($" Is composed model: {purchaseOrderModel.Properties.IsComposedModel}");
Console.WriteLine($" Model Id: {purchaseOrderModel.ModelId}");
Console.WriteLine($" Model name: {purchaseOrderModel.ModelName}");
Console.WriteLine($" Model Status: {purchaseOrderModel.Status}");
Console.WriteLine($" Create model started on: {purchaseOrderModel.TrainingStartedOn}");
Console.WriteLine($" Create model completed on: {purchaseOrderModel.TrainingCompletedOn}");
#endregion
#region Snippet:FormRecognizerSampleSubmodelsInComposedModel
Dictionary <string, List <TrainingDocumentInfo> > trainingDocsPerModel = purchaseOrderModel.TrainingDocuments.GroupBy(doc => doc.ModelId).ToDictionary(g => g.Key, g => g.ToList());
Console.WriteLine($"The purchase order model is based on {purchaseOrderModel.Submodels.Count} model{(purchaseOrderModel.Submodels.Count > 1 ? "s" : "")}.");
foreach (CustomFormSubmodel model in purchaseOrderModel.Submodels)
{
Console.WriteLine($" Model Id: {model.ModelId}");
Console.WriteLine(" The documents used to trained the model are: ");
foreach (var doc in trainingDocsPerModel[model.ModelId])
{
Console.WriteLine($" {doc.Name}");
}
}
#endregion
string purchaseOrderFilePath = FormRecognizerTestEnvironment.CreatePath("Form_1.jpg");
#region Snippet:FormRecognizerSampleRecognizeCustomFormWithComposedModel
//@@ string purchaseOrderFilePath = "<purchaseOrderFilePath>";
FormRecognizerClient formRecognizerClient = client.GetFormRecognizerClient();
RecognizedFormCollection forms;
using (FileStream stream = new FileStream(purchaseOrderFilePath, FileMode.Open))
{
forms = await formRecognizerClient.StartRecognizeCustomFormsAsync(purchaseOrderModel.ModelId, stream).WaitForCompletionAsync();
// Find labeled field.
foreach (RecognizedForm form in forms)
{
// Setting an arbitrary confidence level
if (form.FormTypeConfidence.Value > 0.9)
{
if (form.Fields.TryGetValue("Total", out FormField field))
{
Console.WriteLine($"Total value in the form `{form.FormType}` is `{field.ValueData.Text}`");
}
}
else
{
Console.WriteLine("Unable to recognize form.");
}
}
}
#endregion
// Delete the models on completion to clean environment.
await client.DeleteModelAsync(purchaseOrderOfficeSuppliesModel.ModelId).ConfigureAwait(false);
await client.DeleteModelAsync(purchaseOrderOfficeEquipmentModel.ModelId).ConfigureAwait(false);
await client.DeleteModelAsync(purchaseOrderFurnitureModel.ModelId).ConfigureAwait(false);
await client.DeleteModelAsync(purchaseOrderCleaningSuppliesModel.ModelId).ConfigureAwait(false);
await client.DeleteModelAsync(purchaseOrderModel.ModelId).ConfigureAwait(false);
}
public async Task OutputModelsTrainedWithDynamicRowsTables()
{
string endpoint = TestEnvironment.Endpoint;
string apiKey = TestEnvironment.ApiKey;
string trainingFileUrl = TestEnvironment.TableDynamicRowsContainerSasUrlV2;
string formFilePath = FormRecognizerTestEnvironment.CreatePath("label_table_dynamic_rows1.pdf");
FormRecognizerClient client = new FormRecognizerClient(new Uri(endpoint), new AzureKeyCredential(apiKey));
FormTrainingClient trainingClient = new FormTrainingClient(new Uri(endpoint), new AzureKeyCredential(apiKey));
// Model trained with labeled table with dynamic rows.
CustomFormModel modelTrainedWithLabels = await trainingClient.StartTrainingAsync(new Uri(trainingFileUrl), useTrainingLabels : true).WaitForCompletionAsync();
using (FileStream stream = new FileStream(formFilePath, FileMode.Open))
{
RecognizedFormCollection forms = await client.StartRecognizeCustomFormsAsync(modelTrainedWithLabels.ModelId, stream).WaitForCompletionAsync();
Console.WriteLine("---------Recognized labeled table with dynamic rows---------");
foreach (RecognizedForm form in forms)
{
foreach (FormField field in form.Fields.Values)
{
/// Substitute "table" for the label given to the table tag during training
/// (if different than sample training docs).
if (field.Name == "table")
{
if (field.Value.ValueType == FieldValueType.List)
{
var table = field.Value.AsList();
//columns
for (int i = 0; i < table.Count; i++)
{
Console.WriteLine($"Row {i+1}:");
var row = table[i].Value.AsDictionary();
foreach (var colValues in row)
{
Console.Write($" Col {colValues.Key}. Value: ");
if (colValues.Value != null)
{
var rowContent = colValues.Value;
if (rowContent.Value.ValueType == FieldValueType.String)
{
Console.WriteLine($"'{rowContent.Value.AsString()}'");
}
}
else
{
Console.WriteLine("Empty cell.");
}
}
}
}
}
else
{
Console.WriteLine($"Field {field.Name}: ");
Console.WriteLine($" Value: '{field.ValueData.Text}");
Console.WriteLine($" Confidence: '{field.Confidence}");
}
}
}
}
}
public async Task <RecognizedFormCollection> AnalyzeCustomFormFromStream(Stream form, string modelId) => await client.StartRecognizeCustomFormsAsync(modelId, form).WaitForCompletionAsync();
public async Task <RecognizedFormCollection> AnalyzeCustomFormFromStream(Stream form, string modelId)
{
RecognizedFormCollection collection = await _frClient.StartRecognizeCustomFormsAsync(modelId, form).WaitForCompletionAsync();
return(collection);
}
