Unverified Commit 49043a2e authored by Marius Göcke's avatar Marius Göcke
Browse files

fix

parent 31f6b546

Other/Metrics/LinesOfCode.csv

+1 −1
Original line number Diff line number Diff line 
v1.3.1;3180
 
 No newline at end of file
 
v1.3.1;3156
 
 No newline at end of file

SimpleOCRCLI/SimpleOCRCLI/Runner/RunWithArgumentsFromCLI.cs

+1 −1
Original line number Diff line number Diff line
@@ -19,7 +19,7 @@ namespace SimpleOCR.CLI.Core.Runner 


        protected override string RunImplementation()

        {

            SimpleOCR.Library.Core.IOCRService ocrService = new SimpleOCR.Library.Core.OCRService(this._Options.OCRDataFolder, GeneralLogger.CreateUsingConsole());

            SimpleOCR.Library.Core.IOCRService ocrService = new SimpleOCR.Library.Core.OCRService(new SimpleOCR.Library.Core.OCRServiceConfiguration() { DataFolder = this._Options.OCRDataFolder }, GeneralLogger.CreateUsingConsole());

            ocrService.Initialize();

            (FileType fileType, byte[] content,string mimeType) = SimpleOCR.Library.Core.Misc.Utilities.LoadFile(_Options.File);

            content = fileType.Accept(new ToPictureVisitor(content,mimeType));

SimpleOCRLibrary/SimpleOCRLibrary/IOCRServiceConfiguration.cs

0 → 100644
+7 −0
Original line number Diff line number Diff line 
namespace SimpleOCR.Library.Core

{

    public interface IOCRServiceConfiguration

    {

        public string DataFolder { get; }

    }

}

SimpleOCRLibrary/SimpleOCRLibrary/OCRService.cs

+9 −10
Original line number Diff line number Diff line 
using GRYLibrary.Core.Exceptions;

using GRYLibrary.Core.ExecutePrograms;

using GRYLibrary.Core.Logging.GRYLogger;

using GRYLibrary.Core.Misc;

using SimpleOCR.Library.Core.VIsitors;
@@ -16,14 +15,14 @@ namespace SimpleOCR.Library.Core 
    {



        public bool IsInitialized { get; private set; }

        private readonly string _DataFolder;

        private readonly IOCRServiceConfiguration  _Configuration;

        private readonly IGRYLog _Log;



        /// <param name="dataFolder">Location where OCRService is allowed to store downloaded OCR-data.</param>

        /// <remarks><see cref="OCRService"/> takes care of the download itself. The only requirement is that git is available as command.</remarks>

        public OCRService(string dataFolder, IGRYLog log)

        public OCRService(IOCRServiceConfiguration oCRServiceConfiguration, IGRYLog log)

        {

            this._DataFolder = dataFolder.Replace('\\', '/');

            this._Configuration = oCRServiceConfiguration;

            this._Log = log;

        }

        public string GetOCRContent(byte[] fileContent, ISet<string> languages)
@@ -36,7 +35,7 @@ namespace SimpleOCR.Library.Core 
            var supportedLanguages = this.GetSupportedLanguages();

            foreach (string language in languages)

            {

                if (language.Count() != 3)

                if (language.Length != 3)

                {

                    throw new BadRequestException($"Language '{language}' is not a valid ISO639-3 identifier because its length is not equal to 3.");

                }
@@ -45,7 +44,7 @@ namespace SimpleOCR.Library.Core 
                    throw new BadRequestException($"Language '{language}' is not supported.");

                }

            }

            string dataPath = _DataFolder;

            string dataPath = _Configuration.DataFolder;

            using TesseractEngine engine = new TesseractEngine(dataPath, languagesConcatenated, EngineMode.Default);

            using MemoryStream ms = new MemoryStream(fileContent);

            using Pix img = Pix.LoadFromMemory(ms.ToArray());
@@ -64,7 +63,7 @@ namespace SimpleOCR.Library.Core 
        {

            var result = new HashSet<string>();

            string pattern = @"([a-z][a-z][a-z])\.traineddata";

            string tessdataFolder = _DataFolder;

            string tessdataFolder = _Configuration.DataFolder;

            var files = Directory.GetFiles(tessdataFolder);

            foreach (var file in files)

            {
@@ -82,12 +81,12 @@ namespace SimpleOCR.Library.Core 
        {

            try

            {

                string tessdataFolder = _DataFolder;

                string tessdataFolder = _Configuration.DataFolder;

                GRYLibrary.Core.Misc.Utilities.AssertCondition(!string.IsNullOrEmpty(tessdataFolder), "No OCR-data-folder set.");

                this._Log.Log($"OCRFolder: {tessdataFolder}");

                if (!this.IsInitialized)

                {

                    var traineddataFiles = Directory.GetFiles(_DataFolder).Where(file => file.EndsWith(".traineddata"));

                    var traineddataFiles = Directory.GetFiles(_Configuration.DataFolder).Where(file => file.EndsWith(".traineddata"));

                    GRYLibrary.Core.Misc.Utilities.AssertCondition(1 < traineddataFiles.Count(), "Expected multiple *.traineddata-files.");

                    this.IsInitialized = true;

                }
@@ -101,7 +100,7 @@ namespace SimpleOCR.Library.Core 


        public void ReInitialize()

        {

            GRYLibrary.Core.Misc.Utilities.EnsureDirectoryDoesNotExist(_DataFolder);

            GRYLibrary.Core.Misc.Utilities.EnsureDirectoryDoesNotExist(_Configuration.DataFolder);

            this.IsInitialized = false;

            this.Initialize();

        }

SimpleOCRLibrary/SimpleOCRLibrary/OCRServiceConfiguration.cs

0 → 100644
+7 −0
Original line number Diff line number Diff line 
namespace SimpleOCR.Library.Core

{

    public class OCRServiceConfiguration : IOCRServiceConfiguration

    {

        public string DataFolder { get; set; }

    }

}