コード例 #1
0
        public OperateResult <string> Extract(string extensionName, byte[] data)
        {
            OperateResult <string> operateResult = new OperateResult <string>();

            try
            {
                if (extensionName.IsNullOrWhiteSpace())
                {
                    operateResult.Status      = OperateStatus.Failure;
                    operateResult.Description = "extension不能为空";
                }
                else if (data == null || data.Length == 0)
                {
                    operateResult.Status      = OperateStatus.Failure;
                    operateResult.Description = "fileData不能为空";
                }
                else
                {
                    ICollection <DocumentExtractor> extractors = ExtractorFactory.GetExtractors(extensionName);
                    if (extractors == null || extractors.Count == 0)
                    {
                        operateResult.Status      = OperateStatus.Failure;
                        operateResult.Description = "没有对应的处理程序";
                    }
                    else
                    {
                        bool flag = false;
                        foreach (DocumentExtractor extractor in extractors)
                        {
                            try
                            {
                                ExtractedResult extractedResult = extractor.Extract(extensionName, data, ExtractOption.Text);
                                if (extractedResult != null && extractedResult.Text != null)
                                {
                                    operateResult.Data = extractedResult.Text;
                                    flag = true;
                                    break;
                                }
                            }
                            catch (Exception exception)
                            {
                                flag = false;
                                LoggerWrapper.Logger.Warn("抽取时发生错误", exception);
                            }
                        }
                        if (!flag)
                        {
                            operateResult.Status      = OperateStatus.Failure;
                            operateResult.Description = "抽取出错";
                        }
                    }
                }
            }
            catch (Exception exception)
            {
                operateResult.Description = "抽取出错";
                LoggerWrapper.Logger.Error("ExtractText", exception);
            }
            return(operateResult);
        }
コード例 #2
0
        public OperateResult <ExtractedResult> Extract(string path, ExtractOption[] options)
        {
            OperateResult <ExtractedResult> operateResult = new OperateResult <ExtractedResult>();
            ExtractOption extractOption = CombineOptions(options);
            string        extension     = Path.GetExtension(path);
            ICollection <DocumentExtractor> extractors = ExtractorFactory.GetExtractors(extension);

            if (extractors == null || extractors.Count == 0)
            {
                operateResult.Status      = OperateStatus.Failure;
                operateResult.Description = "没有对应的处理程序";
                return(operateResult);
            }
            bool flag = false;

            if (!File.Exists(path))
            {
                operateResult.Status      = OperateStatus.Failure;
                operateResult.Description = string.Concat("不存在该文件:", path);
                return(operateResult);
            }
            byte[] numArray = File.ReadAllBytes(path);
            try
            {
                foreach (DocumentExtractor extractor in extractors)
                {
                    operateResult.Data = extractor.Extract(extension, numArray, extractOption);
                    if (operateResult.Data == null)
                    {
                        continue;
                    }
                    flag = true;
                    break;
                }
            }
            catch (Exception exception)
            {
                operateResult.Status      = OperateStatus.Failure;
                operateResult.Description = string.Concat("抽取出错:", exception.Message, Environment.NewLine, exception.StackTrace);
                LoggerWrapper.Logger.Error("ExtractText", exception);
            }
            if (!flag)
            {
                operateResult.Status      = OperateStatus.Failure;
                operateResult.Description = "抽取出错";
            }
            return(operateResult);
        }