/// <summary>
/// Classifies a file.
/// </summary>
/// <param name="file">The file to be classified.</param>
/// <param name="parameters">File classification parameters.</param>
/// <returns>Результат классификации файла.</returns>
public ClassificationResult ClassifyFile(FileContainer file, ClassificationParameters parameters)
{
if (file == null)
throw new ArgumentNullException("file");
if (parameters == null)
throw new ArgumentNullException("parameters");
ComprenoTaskSettings settings = convertToTaskSettings(parameters);
byte[] content = _tasksApi.RunTaskAndWaitResult(file, settings);
return _resultParser.Parse(content, parameters.CategoriesFilter);
}
/// <summary>
/// Classifies a byte array instance.
/// </summary>
/// <param name="api">An object that implements <see cref="IClassifierApiProvider"/>.</param>
/// <param name="content">The byte array instance to be classified.</param>
/// <param name="extension">Extension.</param>
/// <param name="parameters">Byte array classification parameters.</param>
/// <returns>Результат классификации a byte array instance.</returns>
public static ClassificationResult ClassifyByteArray(this IClassifierApiProvider api,
byte[] content, string extension, ClassificationParameters parameters)
{
if (api == null)
throw new ArgumentNullException("api");
FileContainer container = new FileContainer
{
Content = content,
Extension = extension
};
return api.ClassifyFile(container, parameters);
}
/// <summary>
/// Classifies files.
/// </summary>
/// <param name="files">The collection of files to be classified.</param>
/// <param name="parameters">File classification parameters.</param>
/// <returns>The results of an classification performed on files.</returns>
public IEnumerable<ClassifiedFile> ClassifyFiles(IEnumerable<string> files, ClassificationParameters parameters)
{
if (files == null)
throw new ArgumentNullException("files");
IEnumerable<BatchFileContainer> batch = files.Select(f => new BatchFileContainer(f));
return ClassifyFiles(batch, parameters);
}
/// <summary>
/// Classifies a file.
/// </summary>
/// <param name="file">The file to be classified.</param>
/// <param name="parameters">File classification parameters.</param>
/// <returns>Результат классификации файла.</returns>
public ClassificationResult ClassifyFile(string file, ClassificationParameters parameters)
{
return ClassifyFile(new FileContainer(file), parameters);
}
/// <summary>
/// Performs files classification.
/// </summary>
/// <param name="files">The collection of files to be classified.</param>
/// <param name="parameters">File classification parameters.</param>
/// <param name="progressChangedHandler">A handler for changes in the progress of files processing operation.</param>
/// <returns>The results of an classification performed on files.</returns>
private IEnumerable<ClassifiedFile> classifyFiles(IEnumerable<BatchFileContainer> files, ClassificationParameters parameters, Action<int> progressChangedHandler)
{
if (!files.Any())
return Enumerable.Empty<ClassifiedFile>();
// Получаем настройки для задания.
ComprenoTaskSettings settings = convertToTaskSettings(parameters);
// Создаем пакет из файлов на классификацию.
IComprenoTasksBatch tasksBatch = _tasksApi.CreateTasksBatch(files, settings);
// Устанавливаем обработчик прогресса классификации, если передан.
if (progressChangedHandler != null)
tasksBatch.WithProgressChangedHandler(progressChangedHandler);
// Запускаем пакет на выполнение и возвращаем результаты.
return tasksBatch.RunAndWaitResults()
.Select(r => convertToClassifiedFile(r, parameters.CategoriesFilter))
.ToList();
}
/// <summary>
/// Converts the document classification parameters into task settings.
/// </summary>
/// <param name="parameters">Document classification parameters.</param>
/// <returns>Task settings for the ABBYY Compreno REST API.</returns>
private static ComprenoTaskSettings convertToTaskSettings(ClassificationParameters parameters)
{
return new ComprenoTaskSettings
{
SourceLanguage = parameters.SourceLanguage == null ? null : parameters.SourceLanguage.Name,
ProcessingParameters = new ProcessingParameters
{
HtmlParsingMode = parameters.RemoveHtmlTags ? HtmlPlainTextMode.Html : HtmlPlainTextMode.None
},
Operations = new OperationsSettings
{
Classification = new ClassificationOperation
{
CategoriesConfidenceLevel = convertToConfidenceLevel(parameters.CategoriesFilter),
MultipleModelNames = parameters.Models,
ReturnFeatures = parameters.ReturnFeatures
}
}
};
}
/// <summary>
/// Classifies files in the folder and subfolders.
/// </summary>
/// <param name="folder">The folder to be classified.</param>
/// <param name="parameters">File classification parameters.</param>
/// <param name="progressChangedHandler">A handler for changes in the progress of a folder processing operation.</param>
/// <returns>The results of an classification performed on files.</returns>
public IEnumerable<ClassifiedFile> ClassifyFolder(string folder, ClassificationParameters parameters, Action<int> progressChangedHandler)
{
return ClassifyFiles(Directory.GetFiles(folder, "*", SearchOption.AllDirectories), parameters, progressChangedHandler);
}
/// <summary>
/// Classifies files in the folder and subfolders.
/// </summary>
/// <param name="folder">The folder to be classified.</param>
/// <param name="parameters">File classification parameters.</param>
/// <returns>The results of an classification performed on files.</returns>
public IEnumerable<ClassifiedFile> ClassifyFolder(string folder, ClassificationParameters parameters)
{
return ClassifyFiles(Directory.GetFiles(folder, "*", SearchOption.AllDirectories), parameters);
}
/// <summary>
/// Classifies files.
/// </summary>
/// <param name="files">The collection of files to be classified.</param>
/// <param name="parameters">File classification parameters.</param>
/// <param name="progressChangedHandler">A handler for changes in the progress of files processing operation.</param>
/// <returns>The results of an classification performed on files.</returns>
public IEnumerable<ClassifiedFile> ClassifyFiles(IEnumerable<BatchFileContainer> files, ClassificationParameters parameters, Action<int> progressChangedHandler)
{
if (files == null)
throw new ArgumentNullException("files");
if (parameters == null)
throw new ArgumentNullException("parameters");
if (progressChangedHandler == null)
throw new ArgumentNullException("progressChangedHandler");
return classifyFiles(files, parameters, progressChangedHandler);
}
/// <summary>
/// Classifies files.
/// </summary>
/// <param name="files">The collection of files to be classified.</param>
/// <param name="parameters">File classification parameters.</param>
/// <returns>The results of an classification performed on files.</returns>
public IEnumerable<ClassifiedFile> ClassifyFiles(IEnumerable<BatchFileContainer> files, ClassificationParameters parameters)
{
if (files == null)
throw new ArgumentNullException("files");
if (parameters == null)
throw new ArgumentNullException("parameters");
return classifyFiles(files, parameters, null);
}
/// <summary>
/// Classifies a UTF-8 string instance.
/// </summary>
/// <param name="api">An object that implements <see cref="IClassifierApiProvider"/>.</param>
/// <param name="content">The string instance to be classified.</param>
/// <param name="extension">Extension.</param>
/// <param name="parameters">String classification parameters.</param>
/// <returns>Результат классификации a string instance.</returns>
public static ClassificationResult ClassifyString(this IClassifierApiProvider api,
string content, string extension, ClassificationParameters parameters)
{
if (api == null)
throw new ArgumentNullException("api");
FileContainer container = new FileContainer
{
Content = Encoding.UTF8.GetBytes(content),
Extension = extension
};
return api.ClassifyFile(container, parameters);
}
