update

PDFWorkflowManager/PDFWorkflowManager/SystemFontResolver.cs

@@ -0,0 +1,18 @@
+using PdfSharp.Fonts;
+using System.IO;
+
+public class SystemFontResolver : IFontResolver
+{
+    public FontResolverInfo ResolveTypeface(string familyName, bool isBold, bool isItalic)
+    {
+        // Map any requested font to Arial
+        return new FontResolverInfo("Arial#");
+    }
+
+    public byte[] GetFont(string faceName)
+    {
+        // Use system fonts directory
+        string fontPath = Path.Combine(System.Environment.GetFolderPath(System.Environment.SpecialFolder.Fonts), "arial.ttf");
+        return File.ReadAllBytes(fontPath);
+    }
+}

PDFWorkflowManager/PDFWorkflowManager/TiffToPdfConverter.cs

@@ -1,146 +1,173 @@
 using System;
 using System.IO;
 using System.Drawing;
-using BitMiracle.LibTiff.Classic;
+using System.Threading.Tasks;
+using System.Windows.Forms;
 using PdfSharp.Pdf;
 using PdfSharp.Drawing;
 using Tesseract;
-using System.Windows.Forms;
-using System.Linq;
+using ImageMagick;
 
 namespace RCEU_PDFWorkflowManager
 {
     public class TiffToPdfConverter
     {
-        public void ConvertTiffToPdfAWithOcr(
+        /// <summary>
+        /// Maximum number of threads to use for page processing
+        /// </summary>
+        public int MaxDegreeOfParallelism { get; set; } = Environment.ProcessorCount;
+
+        /// <summary>
+        /// Converts all TIFF files in a directory to a searchable PDF.
+        /// </summary>
+        public async Task ConvertTiffToPdfWithOcrAsync(
             string workOutDir,
-            string outputPdfPath,
-            string ocrLanguage,
+            string outputPdfFileName, // e.g., "output.pdf"
+            string selectedLanguage,
             ToolStripProgressBar progressBar,
-            ToolStripStatusLabel statusLabel)
+            ToolStripStatusLabel statusLabel,
+            Form mainForm,
+            string bannerPdfPath = null) // optional banner PDF
         {
             string[] tiffFiles = Directory.GetFiles(workOutDir, "*.tif");
 
+            if (tiffFiles.Length == 0)
+                throw new FileNotFoundException("No TIFF files found in the directory.");
+
             // Count total pages for progress bar
-            int totalPages = tiffFiles.Sum(file => Tiff.Open(file, "r")?.NumberOfDirectories() ?? 0);
-            progressBar.Maximum = totalPages;
-            progressBar.Value = 0;
-            statusLabel.Text = "Converting TIFFs to PDF...";
+            int totalPages = 0;
+            foreach (var file in tiffFiles)
+                using (var collection = new MagickImageCollection(file))
+                    totalPages += collection.Count;
 
             PdfDocument pdf = new PdfDocument();
             pdf.Info.Title = "Converted TIFF to PDF/A";
             pdf.Info.Creator = "RCEU_PDFWorkflowManager";
 
-            foreach (var tiffFile in tiffFiles)
+            mainForm.Invoke((MethodInvoker)(() =>
             {
-                using (Tiff image = Tiff.Open(tiffFile, "r"))
+                progressBar.Maximum = totalPages;
+                progressBar.Value = 0;
+                statusLabel.Text = "Converting TIFFs to PDF...";
+            }));
+
+            bool success = true;
+
+            try
+            {
+                await Task.Run(() =>
                 {
-                    if (image == null) continue;
-
-                    int pageCount = image.NumberOfDirectories();
-                    for (int pageIndex = 0; pageIndex < pageCount; pageIndex++)
+                    foreach (var tiffFile in tiffFiles)
                     {
-                        image.SetDirectory((short)pageIndex);
-
-                        int width = image.GetField(TiffTag.IMAGEWIDTH)[0].ToInt();
-                        int height = image.GetField(TiffTag.IMAGELENGTH)[0].ToInt();
-
-                        int[] raster = new int[width * height];
-                        image.ReadRGBAImage(width, height, raster);
-
-                        using (var bmp = new Bitmap(width, height, System.Drawing.Imaging.PixelFormat.Format32bppArgb))
+                        using (var collection = new MagickImageCollection(tiffFile))
                         {
-                            var bmpData = bmp.LockBits(
-                                new Rectangle(0, 0, width, height),
-                                System.Drawing.Imaging.ImageLockMode.WriteOnly,
-                                bmp.PixelFormat);
-
-                            System.Runtime.InteropServices.Marshal.Copy(raster, 0, bmpData.Scan0, raster.Length);
-                            bmp.UnlockBits(bmpData);
-
-                            PdfPage pagePdf = pdf.AddPage();
-                            pagePdf.Width = XUnit.FromPoint(width);
-                            pagePdf.Height = XUnit.FromPoint(height);
-
-                            using (XGraphics gfx = XGraphics.FromPdfPage(pagePdf))
+                            Parallel.ForEach(collection, new ParallelOptions { MaxDegreeOfParallelism = MaxDegreeOfParallelism }, magickImage =>
                             {
-                                // Save temp PNG
-                                string tempPath = Path.Combine(Path.GetTempPath(), Guid.NewGuid() + ".png");
-                                bmp.Save(tempPath, System.Drawing.Imaging.ImageFormat.Png);
-
-                                using (XImage ximg = XImage.FromFile(tempPath))
+                                using (Bitmap bmp = magickImage.ToBitmap())
                                 {
-                                    gfx.DrawImage(ximg, 0, 0, width, height);
+                                    string extractedText = "";
+                                    try
+                                    {
+                                        extractedText = PerformOcr(bmp, selectedLanguage);
+                                    }
+                                    catch
+                                    {
+                                        // If OCR fails, continue but mark success false
+                                        success = false;
+                                    }
+
+                                    lock (pdf)
+                                    {
+                                        PdfPage page = pdf.AddPage();
+                                        page.Width = XUnit.FromPoint(bmp.Width).Point;
+                                        page.Height = XUnit.FromPoint(bmp.Height).Point;
+
+                                        using (var gfx = XGraphics.FromPdfPage(page))
+                                        using (var ms = new MemoryStream())
+                                        {
+                                            bmp.Save(ms, System.Drawing.Imaging.ImageFormat.Png);
+                                            ms.Position = 0;
+                                            using (var ximg = XImage.FromStream(ms))
+                                                gfx.DrawImage(ximg,
+                                                    XUnit.FromPoint(0),
+                                                    XUnit.FromPoint(0),
+                                                    XUnit.FromPoint(bmp.Width),
+                                                    XUnit.FromPoint(bmp.Height));
+                                        }
+
+                                        OverlayTextOntoPdfPage(page, extractedText);
+                                    }
+
+                                    mainForm.Invoke((MethodInvoker)(() =>
+                                    {
+                                        progressBar.Value++;
+                                        statusLabel.Text = $"Processing page {progressBar.Value} of {progressBar.Maximum}";
+                                    }));
                                 }
-
-                                File.Delete(tempPath);
-                            }
-
-                            // OCR and overlay text
-                            string extractedText = PerformOcr(tiffFile, pageIndex, ocrLanguage);
-                            OverlayTextOntoPdfPage(pdf, pagePdf, extractedText);
-
-                            // Update progress bar safely
-                            if (progressBar.InvokeRequired)
-                            {
-                                progressBar.Invoke((MethodInvoker)delegate
-                                {
-                                    progressBar.Value++;
-                                });
-                            }
-                            else
-                            {
-                                progressBar.Value++;
-                            }
+                            });
                         }
                     }
-                }
-            }
 
-            pdf.Save(outputPdfPath);
-            statusLabel.Text = "Conversion complete!";
-        }
-
-        // OCR method
-        private string PerformOcr(string tiffFile, int pageIndex, string language)
-        {
-            using (Tiff tiff = Tiff.Open(tiffFile, "r"))
-            {
-                tiff.SetDirectory((short)pageIndex);
-                int width = tiff.GetField(TiffTag.IMAGEWIDTH)[0].ToInt();
-                int height = tiff.GetField(TiffTag.IMAGELENGTH)[0].ToInt();
-
-                int[] raster = new int[width * height];
-                tiff.ReadRGBAImage(width, height, raster);
-
-                using (Bitmap bmp = new Bitmap(width, height, System.Drawing.Imaging.PixelFormat.Format32bppArgb))
-                {
-                    for (int y = 0; y < height; y++)
-                        for (int x = 0; x < width; x++)
-                        {
-                            int rgba = raster[y * width + x];
-                            int r = rgba & 0xFF;
-                            int g = (rgba >> 8) & 0xFF;
-                            int b = (rgba >> 16) & 0xFF;
-                            bmp.SetPixel(x, height - y - 1, Color.FromArgb(255, r, g, b));
-                        }
-
-                    using (var engine = new TesseractEngine(@"./tessdata", language, EngineMode.Default))
-                    using (var page = engine.Process(bmp))
+                    // Add banner page if selected
+                    if (!string.IsNullOrEmpty(bannerPdfPath) && File.Exists(bannerPdfPath))
                     {
-                        return page.GetText();
+                        lock (pdf)
+                        {
+                            using (PdfDocument bannerDoc = PdfSharp.Pdf.IO.PdfReader.Open(bannerPdfPath, PdfSharp.Pdf.IO.PdfDocumentOpenMode.Import))
+                            {
+                                foreach (PdfPage bannerPage in bannerDoc.Pages)
+                                {
+                                    pdf.AddPage(bannerPage);
+                                }
+                            }
+                        }
                     }
+                });
+
+                // Only save if all pages processed successfully
+                if (success)
+                {
+                    string outputPdfPath = Path.Combine(AppDomain.CurrentDomain.BaseDirectory, outputPdfFileName);
+                    pdf.Save(outputPdfPath);
+                    mainForm.Invoke((MethodInvoker)(() => statusLabel.Text = $"PDF saved: {outputPdfPath}"));
                 }
+                else
+                {
+                    mainForm.Invoke((MethodInvoker)(() => statusLabel.Text = "Conversion incomplete, PDF not saved."));
+                }
+            }
+            catch (Exception ex)
+            {
+                mainForm.Invoke((MethodInvoker)(() => statusLabel.Text = $"Error during conversion: {ex.Message}"));
             }
         }
 
-        // Overlay OCR text (invisible but searchable)
-        private void OverlayTextOntoPdfPage(PdfDocument pdf, PdfPage page, string text)
+        private string PerformOcr(Bitmap bmp, string language)
         {
-            using (XGraphics gfx = XGraphics.FromPdfPage(page, XGraphicsPdfPageOptions.Prepend))
+            string tessDataPath = Path.Combine(AppDomain.CurrentDomain.BaseDirectory, "tessdata");
+
+            if (!Directory.Exists(tessDataPath))
+                throw new DirectoryNotFoundException($"Tessdata folder not found: {tessDataPath}");
+
+            using (var engine = new TesseractEngine(tessDataPath, language, EngineMode.Default))
+            using (var page = engine.Process(bmp))
             {
-                XFont font = new XFont("Arial", 10); // regular font
-                gfx.DrawString(text, font, XBrushes.Transparent, new XRect(0, 0, page.Width, page.Height));
+                return page.GetText();
+            }
+        }
+
+        private void OverlayTextOntoPdfPage(PdfPage page, string text)
+        {
+            using (var gfx = XGraphics.FromPdfPage(page, XGraphicsPdfPageOptions.Prepend))
+            {
+                XFont font = new XFont("Arial", 10);
+                gfx.DrawString(text, font, XBrushes.Transparent,
+                    new XRect(
+                        XUnit.FromPoint(0),
+                        XUnit.FromPoint(0),
+                        XUnit.FromPoint(page.Width),
+                        XUnit.FromPoint(page.Height)),
+                    XStringFormats.TopLeft);
             }
         }
     }

PDFWorkflowManager/PDFWorkflowManager/tessdata/configs/alto

@@ -0,0 +1 @@
+tessedit_create_alto 1

PDFWorkflowManager/PDFWorkflowManager/tessdata/configs/ambigs.train

@@ -0,0 +1,7 @@
+tessedit_ambigs_training	1
+load_freq_dawg	0
+load_punc_dawg	0
+load_system_dawg	0
+load_number_dawg	0
+ambigs_debug_level	3
+load_fixed_length_dawgs	0

PDFWorkflowManager/PDFWorkflowManager/tessdata/configs/api_config

@@ -0,0 +1 @@
+tessedit_zero_rejection T

PDFWorkflowManager/PDFWorkflowManager/tessdata/configs/bigram

@@ -0,0 +1,5 @@
+load_bigram_dawg	True
+tessedit_enable_bigram_correction	True
+tessedit_bigram_debug	3
+save_raw_choices	True
+save_alt_choices	True

PDFWorkflowManager/PDFWorkflowManager/tessdata/configs/box.train

@@ -0,0 +1,12 @@
+disable_character_fragments T
+file_type                   .bl
+textord_fast_pitch_test	T
+tessedit_zero_rejection T
+tessedit_minimal_rejection F
+tessedit_write_rep_codes F
+edges_children_fix F
+edges_childarea 0.65
+edges_boxarea 0.9
+tessedit_resegment_from_boxes T
+tessedit_train_from_boxes T
+textord_no_rejects T

PDFWorkflowManager/PDFWorkflowManager/tessdata/configs/box.train.stderr

@@ -0,0 +1,13 @@
+file_type .bl
+#tessedit_use_nn F
+textord_fast_pitch_test T
+tessedit_zero_rejection T
+tessedit_minimal_rejection F
+tessedit_write_rep_codes F
+edges_children_fix F
+edges_childarea 0.65
+edges_boxarea 0.9
+tessedit_resegment_from_boxes T
+tessedit_train_from_boxes T
+#textord_repeat_extraction F
+textord_no_rejects T

PDFWorkflowManager/PDFWorkflowManager/tessdata/configs/digits

@@ -0,0 +1 @@
+tessedit_char_whitelist 0123456789-.

PDFWorkflowManager/PDFWorkflowManager/tessdata/configs/get.images

@@ -0,0 +1 @@
+tessedit_write_images T

PDFWorkflowManager/PDFWorkflowManager/tessdata/configs/hocr

@@ -0,0 +1,2 @@
+tessedit_create_hocr 1
+hocr_font_info 0

PDFWorkflowManager/PDFWorkflowManager/tessdata/configs/inter

@@ -0,0 +1,2 @@
+interactive_display_mode				T
+tessedit_display_outwords		T

PDFWorkflowManager/PDFWorkflowManager/tessdata/configs/kannada

@@ -0,0 +1,4 @@
+textord_skewsmooth_offset 8
+textord_skewsmooth_offset2 8
+textord_merge_desc 0.5
+textord_no_rejects 1

PDFWorkflowManager/PDFWorkflowManager/tessdata/configs/linebox

@@ -0,0 +1,2 @@
+tessedit_resegment_from_line_boxes 1
+tessedit_make_boxes_from_boxes 1

PDFWorkflowManager/PDFWorkflowManager/tessdata/configs/logfile

@@ -0,0 +1 @@
+debug_file tesseract.log

PDFWorkflowManager/PDFWorkflowManager/tessdata/configs/lstm.train

@@ -0,0 +1,11 @@
+file_type                   .bl
+textord_fast_pitch_test	T
+tessedit_zero_rejection T
+tessedit_minimal_rejection F
+tessedit_write_rep_codes F
+edges_children_fix F
+edges_childarea 0.65
+edges_boxarea 0.9
+tessedit_train_line_recognizer T
+textord_no_rejects T
+tessedit_init_config_only T

PDFWorkflowManager/PDFWorkflowManager/tessdata/configs/lstmbox

@@ -0,0 +1 @@
+tessedit_create_lstmbox 1

PDFWorkflowManager/PDFWorkflowManager/tessdata/configs/lstmdebug

@@ -0,0 +1,4 @@
+stopper_debug_level 1
+classify_debug_level 1
+segsearch_debug_level 1
+language_model_debug_level 3

PDFWorkflowManager/PDFWorkflowManager/tessdata/configs/makebox

@@ -0,0 +1 @@
+tessedit_create_boxfile 1

PDFWorkflowManager/PDFWorkflowManager/tessdata/configs/page

@@ -0,0 +1,3 @@
+tessedit_create_page_xml 1
+# page_xml_polygon 1
+# page_xml_level 0

PDFWorkflowManager/PDFWorkflowManager/tessdata/configs/pdf

@@ -0,0 +1 @@
+tessedit_create_pdf 1

PDFWorkflowManager/PDFWorkflowManager/tessdata/configs/quiet

@@ -0,0 +1 @@
+debug_file /dev/null

PDFWorkflowManager/PDFWorkflowManager/tessdata/configs/rebox

@@ -0,0 +1,2 @@
+tessedit_resegment_from_boxes 1
+tessedit_make_boxes_from_boxes 1

PDFWorkflowManager/PDFWorkflowManager/tessdata/configs/strokewidth

@@ -0,0 +1,12 @@
+textord_show_blobs 0
+textord_debug_tabfind 3
+textord_tabfind_show_partitions 1
+textord_tabfind_show_initial_partitions 1
+textord_tabfind_show_columns 1
+textord_tabfind_show_blocks 1
+textord_tabfind_show_initialtabs 1
+textord_tabfind_show_finaltabs 1
+textord_tabfind_show_strokewidths 1
+textord_tabfind_show_vlines 0
+textord_tabfind_show_images 1
+tessedit_dump_pageseg_images 0

PDFWorkflowManager/PDFWorkflowManager/tessdata/configs/tsv

@@ -0,0 +1 @@
+tessedit_create_tsv 1

PDFWorkflowManager/PDFWorkflowManager/tessdata/configs/txt

@@ -0,0 +1,3 @@
+# This config file should be used with other config files which create renderers.
+# usage example: tesseract eurotext.tif eurotext txt hocr pdf
+tessedit_create_txt 1

PDFWorkflowManager/PDFWorkflowManager/tessdata/configs/unlv

@@ -0,0 +1,2 @@
+tessedit_write_unlv 1
+unlv_tilde_crunching T

PDFWorkflowManager/PDFWorkflowManager/tessdata/configs/wordstrbox

@@ -0,0 +1 @@
+tessedit_create_wordstrbox 1