Merge branch 'v2'

Author: shps951023

.gitignore

@@ -37,3 +37,4 @@ tests/MiniPdf.Benchmark/__pycache__/
 tests/MiniPdf.Scripts/pdf_test_single/docx_classic01_single_paragraph.pdf
 MiniPdf.Web/publish.zip
 /tests/MiniPdf.Scripts/pdf_output_baseline
+/tests/MiniPdf.Benchmark/minipdf_docx

_latest_bench.txt

@@ -90,9 +90,9 @@ internal PdfPage(float width, float height)
     /// <param name="fontSize">Font size in points (default: 12).</param>
     /// <param name="color">Text color (default: black).</param>
     /// <returns>The current page for chaining.</returns>
-    public PdfPage AddText(string text, float x, float y, float fontSize = 12, PdfColor? color = null, (float, float, float, float)? clipRect = null, float? maxWidth = null)
+    public PdfPage AddText(string text, float x, float y, float fontSize = 12, PdfColor? color = null, (float, float, float, float)? clipRect = null, float? maxWidth = null, bool bold = false, bool underline = false, float charSpacing = 0)
     {
-        _textBlocks.Add(new PdfTextBlock(text, x, y, fontSize, color, clipRect, maxWidth));
+        _textBlocks.Add(new PdfTextBlock(text, x, y, fontSize, color, clipRect, maxWidth, bold, underline, charSpacing));
         return this;
     }
 

src/MiniPdf/DocxReader.cs

@@ -47,7 +47,22 @@ internal sealed class PdfTextBlock
     /// </summary>
     public float? MaxWidth { get; }
 
-    internal PdfTextBlock(string text, float x, float y, float fontSize, PdfColor? color = null, (float, float, float, float)? clipRect = null, float? maxWidth = null)
+    /// <summary>
+    /// Whether to render text using the bold font variant.
+    /// </summary>
+    public bool Bold { get; }
+
+    /// <summary>
+    /// Whether to render an underline below the text.
+    /// </summary>
+    public bool Underline { get; }
+
+    /// <summary>
+    /// Character spacing in points (PDF Tc operator). 0 means default.
+    /// </summary>
+    public float CharSpacing { get; }
+
+    internal PdfTextBlock(string text, float x, float y, float fontSize, PdfColor? color = null, (float, float, float, float)? clipRect = null, float? maxWidth = null, bool bold = false, bool underline = false, float charSpacing = 0)
     {
         Text = text;
         X = x;
@@ -56,5 +71,8 @@ internal PdfTextBlock(string text, float x, float y, float fontSize, PdfColor? c
         Color = color ?? PdfColor.Black;
         ClipRect = clipRect;
         MaxWidth = maxWidth;
+        Bold = bold;
+        Underline = underline;
+        CharSpacing = charSpacing;
     }
 }

src/MiniPdf/DocxToPdfConverter.cs

@@ -215,10 +215,10 @@ internal void Write(PdfDocument document)
             contentStreams.Add(Encoding.Latin1.GetBytes(BuildContentStream(pages[i], embeddedFonts.Count > 0, cpToFontSlot, embeddedFonts)));
 
         // Allocate object numbers.
-        //   1 = Catalog, 2 = Pages, 3 = Font F1 (Helvetica/WinAnsi)
+        //   1 = Catalog, 2 = Pages, 3 = Font F1 (Helvetica/WinAnsi), 4 = Font F1B (Helvetica-Bold/WinAnsi)
         //   Per embedded font: 6 objects (ToUnicode, Descriptor, CIDFont, Type0, FontFile2, CIDToGIDMap)
         //   Per page: content stream obj, N image XObject objs, page obj
-        var nextObj = 4;
+        var nextObj = 5;
 
         // Allocate font objects
         foreach (var ef in embeddedFonts)
@@ -263,6 +263,10 @@ internal void Write(PdfDocument document)
         _objectOffsets[3] = Position;
         WriteRaw("3 0 obj\n<< /Type /Font /Subtype /Type1 /BaseFont /Helvetica /Encoding /WinAnsiEncoding >>\nendobj\n");
 
+        // ── Object 4: Font F1B (Helvetica-Bold, built-in WinAnsiEncoding) ──
+        _objectOffsets[4] = Position;
+        WriteRaw("4 0 obj\n<< /Type /Font /Subtype /Type1 /BaseFont /Helvetica-Bold /Encoding /WinAnsiEncoding >>\nendobj\n");
+
         // ── Per-font objects (F2, F3, …) ───────────────────────────────────
         for (var fi = 0; fi < embeddedFonts.Count; fi++)
         {
@@ -356,8 +360,8 @@ internal void Write(PdfDocument document)
             WriteRaw($"<< /Type /Page /Parent 2 0 R /MediaBox [0 0 {w} {h}]\n");
             WriteRaw($"/Contents {contentObjNums[i]} 0 R\n");
             WriteRaw("/Resources <<\n");
-            // Font dictionary: F1 + Fn for each embedded font
-            WriteRaw("/Font << /F1 3 0 R");
+            // Font dictionary: F1, F1B + Fn for each embedded font
+            WriteRaw("/Font << /F1 3 0 R /F1B 4 0 R");
             for (var fi = 0; fi < embeddedFonts.Count; fi++)
                 WriteRaw($" /F{fi + 2} {embeddedFonts[fi].Type0Obj} 0 R");
             WriteRaw(" >>\n");
@@ -514,15 +518,19 @@ private static string BuildContentStream(PdfPage page, bool hasUnicodeFont, Dict
 
             if (!hasUnicodeFont || !block.Text.Any(c => !IsWinAnsiHandled(c)))
             {
-                // Pure Latin-1 text — use F1 (Helvetica) as before
+                // Pure Latin-1 text — use F1 (Helvetica) or F1B (Helvetica-Bold)
+                var fontName = block.Bold ? "F1B" : "F1";
                 var escapedText = EscapePdfString(block.Text);
                 sb.Append("BT\n");
                 sb.Append(colorCmd);
-                sb.Append($"/F1 {fontSize} Tf\n");
+                sb.Append($"/{fontName} {fontSize} Tf\n");
+                // Apply character spacing (Tc)
+                if (block.CharSpacing != 0)
+                    sb.Append($"{block.CharSpacing.ToString("F2", CultureInfo.InvariantCulture)} Tc\n");
                 // Apply horizontal scaling if text overflows MaxWidth
                 if (block.MaxWidth.HasValue)
                 {
-                    var naturalWidth = MeasureTextWidth(block.Text, block.FontSize);
+                    var naturalWidth = MeasureTextWidth(block.Text, block.FontSize, block.CharSpacing);
                     if (naturalWidth > block.MaxWidth.Value && naturalWidth > 0)
                     {
                         var tzPercent = (block.MaxWidth.Value / naturalWidth) * 100.0;
@@ -544,10 +552,13 @@ private static string BuildContentStream(PdfPage page, bool hasUnicodeFont, Dict
                 // emit each run with the appropriate Fn, using Td to advance.
                 sb.Append("BT\n");
                 sb.Append(colorCmd);
+                // Apply character spacing (Tc)
+                if (block.CharSpacing != 0)
+                    sb.Append($"{block.CharSpacing.ToString("F2", CultureInfo.InvariantCulture)} Tc\n");
                 // Apply horizontal scaling if text overflows MaxWidth
                 if (block.MaxWidth.HasValue)
                 {
-                    var naturalWidth = MeasureTextWidth(block.Text, block.FontSize);
+                    var naturalWidth = MeasureTextWidth(block.Text, block.FontSize, block.CharSpacing);
                     if (naturalWidth > block.MaxWidth.Value && naturalWidth > 0)
                     {
                         var tzPercent = (block.MaxWidth.Value / naturalWidth) * 100.0;
@@ -594,6 +605,25 @@ private static string BuildContentStream(PdfPage page, bool hasUnicodeFont, Dict
             // Restore graphics state after clipping
             if (hasClip)
                 sb.Append("Q\n");
+
+            // Render underline as a line below the text
+            if (block.Underline)
+            {
+                var textWidth = MeasureTextWidth(block.Text, block.FontSize, block.CharSpacing);
+                if (block.MaxWidth.HasValue && textWidth > block.MaxWidth.Value)
+                    textWidth = block.MaxWidth.Value;
+                var ulY = block.Y - block.FontSize * 0.15f; // position below baseline
+                var ulThickness = Math.Max(0.5f, block.FontSize * 0.05f);
+                var x1 = block.X.ToString("F3", CultureInfo.InvariantCulture);
+                var y1 = ulY.ToString("F3", CultureInfo.InvariantCulture);
+                var x2 = (block.X + textWidth).ToString("F3", CultureInfo.InvariantCulture);
+                var lw = ulThickness.ToString("F3", CultureInfo.InvariantCulture);
+                sb.Append($"{block.Color.R.ToString("F3", CultureInfo.InvariantCulture)} " +
+                          $"{block.Color.G.ToString("F3", CultureInfo.InvariantCulture)} " +
+                          $"{block.Color.B.ToString("F3", CultureInfo.InvariantCulture)} RG\n");
+                sb.Append($"{lw} w\n");
+                sb.Append($"{x1} {y1} m {x2} {y1} l S\n");
+            }
         }
 
         return sb.ToString();
@@ -603,7 +633,7 @@ private static string BuildContentStream(PdfPage page, bool hasUnicodeFont, Dict
     /// Measures the natural rendering width of text in Helvetica at the given font size.
     /// Uses the standard Helvetica character width table.
     /// </summary>
-    private static double MeasureTextWidth(string text, float fontSize)
+    private static double MeasureTextWidth(string text, float fontSize, float charSpacing = 0)
     {
         double total = 0;
         foreach (var ch in text)
@@ -633,7 +663,11 @@ private static double MeasureTextWidth(string text, float fontSize)
             };
             total += w;
         }
-        return total * fontSize / 1000.0;
+        var result = total * fontSize / 1000.0;
+        // Tc adds charSpacing points per character (except after the last)
+        if (charSpacing != 0 && text.Length > 1)
+            result += charSpacing * (text.Length - 1);
+        return result;
     }
 
     /// <summary>
@@ -806,12 +840,12 @@ private static string EscapePdfString(string text)
                 '\u20AC' => (char)0x80,                     // euro sign
                 '\u00A0' => ' ',                            // non-breaking space
                 '\u0060' => '\'',                           // backtick → apostrophe
-                '\u00B7' => '*',                            // middle dot → asterisk
-                '\u00D7' => 'x',                            // multiplication sign
-                '\u00F7' => '/',                            // division sign
+                '\u00B7' => '\u00B7',                       // middle dot (already in WinAnsi)
+                '\u00D7' => '\u00D7',                       // multiplication sign (already in WinAnsi)
+                '\u00F7' => '\u00F7',                       // division sign (already in WinAnsi)
                 '\u2264' => "<=",                           // ≤
                 '\u2265' => ">=",                           // ≥
-                '\u00B0' => " deg",                         // degree sign
+                '\u00B0' => '\u00B0',                       // degree sign (already in WinAnsi)
                 '\u00AE' => (char)0xAE,                     // registered trademark (already in WinAnsi)
                 '\u00A3' => '\u00A3',                       // pound sign (already in WinAnsi)
                 '\u00A5' => '\u00A5',                       // yen sign (already in WinAnsi)

src/MiniPdf/PdfPage.cs

@@ -0,0 +1,32 @@
+"""Investigate the spacing issue around checkmarks in SA8000"""
+import fitz
+
+mini = fitz.open(r'D:\git\MiniPdf-v2\tests\Issue_Files\minipdf_docx\SA8000 ch sample.pdf')
+
+# Find the line with item 1 checkmark
+td = mini[0].get_text("dict")
+for block in td["blocks"]:
+    if "lines" not in block:
+        continue
+    for line in block["lines"]:
+        text = "".join(s["text"] for s in line["spans"])
+        if "签定劳动合同" in text:
+            print(f"Line text: {repr(text)}")
+            for i, span in enumerate(line["spans"]):
+                print(f"  Span {i}: x={span['origin'][0]:.1f} y={span['origin'][1]:.1f} "
+                      f"size={span['size']:.1f} text={repr(span['text'][:40])}")
+            # Also show raw character positions
+            chars = mini[0].get_text("rawdict")
+            for rb in chars["blocks"]:
+                if "lines" not in rb:
+                    continue
+                for rl in rb["lines"]:
+                    rt = "".join(s["text"] for s in rl["spans"])
+                    if "签定劳动合同" in rt:
+                        for ri, rs in enumerate(rl["spans"]):
+                            if "（" in rs["text"] or "√" in rs["text"] or "）" in rs["text"]:
+                                print(f"\n  Raw span {ri}: text={repr(rs['text'][:30])}")
+                                if "chars" in rs:
+                                    for ci, ch in enumerate(rs["chars"][:20]):
+                                        print(f"    char[{ci}]: '{ch['c']}' x={ch['origin'][0]:.1f} bbox={tuple(round(v,1) for v in ch['bbox'])}")
+            break

src/MiniPdf/PdfTextBlock.cs

@@ -0,0 +1,15 @@
+import fitz
+
+# Check Y coordinates of text spans for paragraph 1 on page 1
+doc = fitz.open('minipdf_docx/SA8000 ch sample.pdf')
+page = doc[0]
+blocks = page.get_text('dict')['blocks']
+for b in blocks:
+    if 'lines' not in b:
+        continue
+    for line in b['lines']:
+        for span in line['spans']:
+            t = span['text']
+            if '签定' in t or t.strip() in ['√', '）', '（']:
+                bbox = span['bbox']
+                print(f"y0={bbox[1]:.1f} y1={bbox[3]:.1f} x0={bbox[0]:.1f} x1={bbox[2]:.1f} fs={span['size']:.1f} text={repr(t)}")

src/MiniPdf/PdfWriter.cs

@@ -0,0 +1,31 @@
+"""Compare extracted text for first few lines of SA8000 MiniPdf vs Reference"""
+import fitz  # PyMuPDF
+
+mini = fitz.open(r'D:\git\MiniPdf-v2\tests\Issue_Files\minipdf_docx\SA8000 ch sample.pdf')
+ref = fitz.open(r'D:\git\MiniPdf-v2\tests\Issue_Files\reference_docx\SA8000 ch sample.pdf')
+
+print("=== MiniPdf Page 1 first 500 chars ===")
+mt = mini[0].get_text()
+print(repr(mt[:500]))
+
+print("\n=== Reference Page 1 first 500 chars ===")
+rt = ref[0].get_text()
+print(repr(rt[:500]))
+
+# Also check raw text spans for first block
+print("\n=== MiniPdf Page 1 text blocks (first 5) ===")
+for i, block in enumerate(mini[0].get_text("blocks")[:5]):
+    print(f"  B{i}: bbox={block[:4]}: {repr(block[4][:100])}")
+
+print("\n=== Reference Page 1 text blocks (first 5) ===")
+for i, block in enumerate(ref[0].get_text("blocks")[:5]):
+    print(f"  B{i}: bbox={block[:4]}: {repr(block[4][:100])}")
+
+# Check text spans for P1 line
+print("\n=== MiniPdf Page 1 text dict (first 3 blocks) ===")
+td = mini[0].get_text("dict")
+for i, block in enumerate(td["blocks"][:3]):
+    if "lines" in block:
+        for j, line in enumerate(block["lines"]):
+            for k, span in enumerate(line["spans"]):
+                print(f"  B{i}L{j}S{k}: origin={span['origin']}, size={span['size']:.1f}, text={repr(span['text'][:80])}")