Fix Burmese PDF rendering

Dockerfile

@@ -26,7 +26,7 @@ ENV ASPNETCORE_ENVIRONMENT=Production
 EXPOSE 7860
 
 RUN apt-get update \
-    && apt-get install -y --no-install-recommends nginx ca-certificates \
+    && apt-get install -y --no-install-recommends nginx ca-certificates fonts-noto-core \
     && rm -rf /var/lib/apt/lists/*
 
 COPY --from=build /app/publish/api /app/api

TaskTrackingSystem.Shared/SimplePdfReportBuilder.cs

@@ -1,18 +1,23 @@
-using System.Text;
+using System.Globalization;
+using SkiaSharp;
+using SkiaSharp.HarfBuzz;
 
 namespace TaskTrackingSystem.Shared;
 
 public static class SimplePdfReportBuilder
 {
-    private const int PageWidth = 612;
-    private const int PageHeight = 792;
-    private const int LeftMargin = 48;
-    private const int TopMargin = 52;
-    private const int BottomMargin = 48;
-    private const int TitleFontSize = 16;
-    private const int BodyFontSize = 10;
-    private const int LineSpacing = 14;
+    private const float PageWidth = 612;
+    private const float PageHeight = 792;
+    private const float LeftMargin = 48;
+    private const float TopMargin = 52;
+    private const float RightMargin = 48;
+    private const float TitleFontSize = 16;
+    private const float BodyFontSize = 10;
+    private const float LineSpacing = 14;
     private const int BodyLinesPerPage = 42;
+    private const string FontPathEnvironmentVariable = "TASKTRACKING_PDF_FONT_PATH";
+
+    private static readonly string[] RequiredGlyphSamples = ["ABC0123|:-", "မြန်မာ"];
 
     public static byte[] BuildTableReport(
         string title,
@@ -24,22 +29,16 @@ public static class SimplePdfReportBuilder
 
         if (summaryLines != null)
         {
-            foreach (var line in summaryLines.Where(line => !string.IsNullOrWhiteSpace(line)))
-            {
-                lines.AddRange(WrapLine(line.Trim()));
-            }
+            lines.AddRange(summaryLines.Where(line => !string.IsNullOrWhiteSpace(line)).Select(line => line.Trim()));
         }
 
         if (headers.Count > 0)
         {
             lines.Add(string.Join(" | ", headers));
-            lines.Add(new string('-', Math.Min(96, Math.Max(24, headers.Sum(h => h.Length + 3)))));
+            lines.Add(new string('-', 72));
         }
 
-        foreach (var row in rows)
-        {
-            lines.AddRange(WrapLine(string.Join(" | ", row.Select(NormalizeCell))));
-        }
+        lines.AddRange(rows.Select(row => string.Join(" | ", row.Select(NormalizeCell))));
 
         if (lines.Count == 0)
         {
@@ -51,8 +50,13 @@ public static class SimplePdfReportBuilder
 
     public static byte[] BuildTextReport(string title, IEnumerable<string> bodyLines)
     {
+        using var typeface = LoadReportTypeface();
+        using var shaper = new SKShaper(typeface);
+        using var bodyFont = new SKFont(typeface, BodyFontSize);
+
+        var availableWidth = PageWidth - LeftMargin - RightMargin;
         var normalizedLines = bodyLines
-            .SelectMany(line => WrapLine(line))
+            .SelectMany(line => WrapLine(line, availableWidth, shaper, bodyFont))
             .ToList();
 
         var pages = normalizedLines
@@ -62,128 +66,185 @@ public static class SimplePdfReportBuilder
 
         if (pages.Count == 0)
         {
-            pages.Add(new List<string> { "No data available." });
+            pages.Add(["No data available."]);
         }
 
-        var objects = new List<string>();
-        objects.Add("1 0 obj\n<< /Type /Catalog /Pages 2 0 R >>\nendobj\n");
-
-        var pageObjectIds = new List<int>();
-        var contentObjectIds = new List<int>();
-        var nextObjectId = 4;
+        using var output = new MemoryStream();
+        using var document = SKDocument.CreatePdf(output)
+            ?? throw new InvalidOperationException("The PDF renderer could not be initialized.");
 
-        foreach (var _ in pages)
+        for (var pageIndex = 0; pageIndex < pages.Count; pageIndex++)
         {
-            pageObjectIds.Add(nextObjectId++);
-            contentObjectIds.Add(nextObjectId++);
+            using var canvas = document.BeginPage(PageWidth, PageHeight);
+            DrawPage(canvas, shaper, typeface, title, pages[pageIndex], pageIndex + 1, pages.Count);
+            document.EndPage();
         }
 
-        var pagesKids = string.Join(" ", pageObjectIds.Select(id => $"{id} 0 R"));
-        objects.Add($"2 0 obj\n<< /Type /Pages /Kids [{pagesKids}] /Count {pages.Count} >>\nendobj\n");
-        objects.Add("3 0 obj\n<< /Type /Font /Subtype /Type1 /BaseFont /Helvetica >>\nendobj\n");
+        document.Close();
+        return output.ToArray();
+    }
 
-        for (var pageIndex = 0; pageIndex < pages.Count; pageIndex++)
+    private static void DrawPage(
+        SKCanvas canvas,
+        SKShaper shaper,
+        SKTypeface typeface,
+        string title,
+        IReadOnlyList<string> lines,
+        int pageNumber,
+        int totalPages)
+    {
+        canvas.Clear(SKColors.White);
+
+        using var titleFont = new SKFont(typeface, TitleFontSize);
+        using var bodyFont = new SKFont(typeface, BodyFontSize);
+        using var paint = new SKPaint
         {
-            var pageObjectId = pageObjectIds[pageIndex];
-            var contentObjectId = contentObjectIds[pageIndex];
-            var pageLines = pages[pageIndex];
-            var content = BuildContentStream(title, pageLines, pageIndex + 1, pages.Count);
-            var contentBytes = Encoding.ASCII.GetBytes(content);
-
-            objects.Add(
-                $"{pageObjectId} 0 obj\n" +
-                $"<< /Type /Page /Parent 2 0 R /MediaBox [0 0 {PageWidth} {PageHeight}] /Resources << /Font << /F1 3 0 R >> >> /Contents {contentObjectId} 0 R >>\n" +
-                "endobj\n");
-
-            objects.Add(
-                $"{contentObjectId} 0 obj\n" +
-                $"<< /Length {contentBytes.Length} >>\n" +
-                "stream\n" +
-                content +
-                "\nendstream\nendobj\n");
-        }
+            Color = SKColors.Black,
+            IsAntialias = true
+        };
+
+        canvas.DrawShapedText(shaper, title, LeftMargin, TopMargin, SKTextAlign.Left, titleFont, paint);
+        canvas.DrawShapedText(
+            shaper,
+            $"Page {pageNumber} of {totalPages}",
+            LeftMargin,
+            TopMargin + 22,
+            SKTextAlign.Left,
+            bodyFont,
+            paint);
+
+        var baseline = TopMargin + 22 + LineSpacing;
+        foreach (var line in lines)
+        {
+            baseline += LineSpacing;
 
-        var buffer = new StringBuilder();
-        buffer.Append("%PDF-1.4\n");
+            if (!string.IsNullOrEmpty(line))
+            {
+                canvas.DrawShapedText(shaper, line, LeftMargin, baseline, SKTextAlign.Left, bodyFont, paint);
+            }
+        }
+    }
 
-        var offsets = new List<int> { 0 };
-        foreach (var obj in objects)
+    private static IEnumerable<string> WrapLine(string? line, float maxWidth, SKShaper shaper, SKFont font)
+    {
+        if (string.IsNullOrWhiteSpace(line))
         {
-            offsets.Add(Encoding.ASCII.GetByteCount(buffer.ToString()));
-            buffer.Append(obj);
+            yield return string.Empty;
+            yield break;
         }
 
-        var xrefStart = Encoding.ASCII.GetByteCount(buffer.ToString());
-        var xref = new StringBuilder();
-        xref.Append("xref\n");
-        xref.Append($"0 {objects.Count + 1}\n");
-        xref.Append("0000000000 65535 f \n");
-
-        foreach (var offset in offsets.Skip(1))
+        var remaining = line.Trim();
+        while (MeasureText(remaining, shaper, font) > maxWidth)
         {
-            xref.Append($"{offset:D10} 00000 n \n");
-        }
+            var breakIndex = FindBreakIndex(remaining, maxWidth, shaper, font);
+            var wrappedLine = remaining[..breakIndex].TrimEnd();
+
+            if (wrappedLine.Length == 0)
+            {
+                break;
+            }
 
-        xref.Append("trailer\n");
-        xref.Append($"<< /Size {objects.Count + 1} /Root 1 0 R >>\n");
-        xref.Append("startxref\n");
-        xref.Append($"{xrefStart}\n");
-        xref.Append("%%EOF");
+            yield return wrappedLine;
+            remaining = remaining[breakIndex..].TrimStart();
+        }
 
-        buffer.Append(xref);
-        return Encoding.ASCII.GetBytes(buffer.ToString());
+        if (remaining.Length > 0)
+        {
+            yield return remaining;
+        }
     }
 
-    private static string BuildContentStream(string title, IReadOnlyList<string> lines, int pageNumber, int totalPages)
+    private static int FindBreakIndex(string text, float maxWidth, SKShaper shaper, SKFont font)
     {
-        var content = new StringBuilder();
-        content.AppendLine("BT");
-        content.AppendLine("/F1 16 Tf");
-        content.AppendLine($"{LeftMargin} {PageHeight - TopMargin} Td");
-        content.AppendLine($"({EscapeText(title)}) Tj");
-        content.AppendLine("/F1 10 Tf");
-        content.AppendLine($"0 -22 Td");
-        content.AppendLine($"({EscapeText($"Page {pageNumber} of {totalPages}")}) Tj");
+        var textElementIndexes = StringInfo.ParseCombiningCharacters(text);
+        var low = 1;
+        var high = textElementIndexes.Length;
+        var fittingElementCount = 1;
 
-        foreach (var line in lines)
+        while (low <= high)
         {
-            content.AppendLine($"0 -{LineSpacing} Td");
-            content.AppendLine($"({EscapeText(line)}) Tj");
+            var middle = low + ((high - low) / 2);
+            var endIndex = middle < textElementIndexes.Length ? textElementIndexes[middle] : text.Length;
+
+            if (MeasureText(text[..endIndex], shaper, font) <= maxWidth)
+            {
+                fittingElementCount = middle;
+                low = middle + 1;
+            }
+            else
+            {
+                high = middle - 1;
+            }
         }
 
-        content.AppendLine("ET");
-        return content.ToString();
+        var fittingEndIndex = fittingElementCount < textElementIndexes.Length
+            ? textElementIndexes[fittingElementCount]
+            : text.Length;
+        var whitespaceIndex = text.LastIndexOf(' ', Math.Max(0, fittingEndIndex - 1), fittingEndIndex);
+
+        return whitespaceIndex > 0 ? whitespaceIndex : fittingEndIndex;
     }
 
-    private static IEnumerable<string> WrapLine(string? line)
+    private static float MeasureText(string text, SKShaper shaper, SKFont font)
+        => string.IsNullOrEmpty(text) ? 0 : shaper.Shape(text, font).Width;
+
+    private static SKTypeface LoadReportTypeface()
     {
-        if (string.IsNullOrWhiteSpace(line))
+        foreach (var path in GetFontCandidates())
         {
-            yield return string.Empty;
-            yield break;
-        }
+            if (!File.Exists(path))
+            {
+                continue;
+            }
 
-        const int maxWidth = 92;
-        var remaining = line.Trim();
+            var typeface = SKTypeface.FromFile(path);
+            if (typeface != null && SupportsReportGlyphs(typeface))
+            {
+                return typeface;
+            }
+
+            typeface?.Dispose();
+        }
 
-        while (remaining.Length > maxWidth)
+        foreach (var family in new[] { "Noto Sans Myanmar", "Myanmar Text", "Nirmala UI", "Padauk" })
         {
-            var breakIndex = remaining.LastIndexOf(' ', maxWidth);
-            if (breakIndex <= 0)
+            var typeface = SKTypeface.FromFamilyName(family);
+            if (typeface != null && SupportsReportGlyphs(typeface))
             {
-                breakIndex = maxWidth;
+                return typeface;
             }
 
-            yield return remaining[..breakIndex].Trim();
-            remaining = remaining[breakIndex..].TrimStart();
+            typeface?.Dispose();
         }
 
-        if (!string.IsNullOrWhiteSpace(remaining))
+        throw new InvalidOperationException(
+            "No Unicode font capable of rendering Burmese was found. " +
+            $"Install Noto Sans Myanmar or set {FontPathEnvironmentVariable} to a compatible .ttf file.");
+    }
+
+    private static IEnumerable<string> GetFontCandidates()
+    {
+        var configuredPath = Environment.GetEnvironmentVariable(FontPathEnvironmentVariable);
+        if (!string.IsNullOrWhiteSpace(configuredPath))
         {
-            yield return remaining;
+            yield return configuredPath;
         }
+
+        var windowsDirectory = Environment.GetFolderPath(Environment.SpecialFolder.Windows);
+        if (!string.IsNullOrWhiteSpace(windowsDirectory))
+        {
+            yield return Path.Combine(windowsDirectory, "Fonts", "mmrtext.ttf");
+            yield return Path.Combine(windowsDirectory, "Fonts", "Nirmala.ttf");
+        }
+
+        yield return "/usr/share/fonts/truetype/noto/NotoSansMyanmar-Regular.ttf";
+        yield return "/usr/share/fonts/truetype/padauk/Padauk-Regular.ttf";
+        yield return "/usr/share/fonts/opentype/noto/NotoSansMyanmar-Regular.ttf";
     }
 
+    private static bool SupportsReportGlyphs(SKTypeface typeface)
+        => RequiredGlyphSamples.All(sample => typeface.GetGlyphs(sample).All(glyph => glyph != 0));
+
     private static string NormalizeCell(string? value)
     {
         if (string.IsNullOrWhiteSpace(value))
@@ -193,13 +254,4 @@ public static class SimplePdfReportBuilder
 
         return value.Replace("\r", " ").Replace("\n", " ").Trim();
     }
-
-    private static string EscapeText(string value)
-    {
-        var ascii = new string(value.Select(ch => ch is >= ' ' and <= '~' ? ch : '?').ToArray());
-        return ascii
-            .Replace("\\", "\\\\")
-            .Replace("(", "\\(")
-            .Replace(")", "\\)");
-    }
 }

TaskTrackingSystem.Shared/TaskTrackingSystem.Shared.csproj

@@ -18,4 +18,10 @@
     <Folder Include="Models\Role\" />
   </ItemGroup>
 
+  <ItemGroup>
+    <PackageReference Include="HarfBuzzSharp.NativeAssets.Linux" Version="8.3.1.5" />
+    <PackageReference Include="SkiaSharp.HarfBuzz" Version="3.119.4" />
+    <PackageReference Include="SkiaSharp.NativeAssets.Linux.NoDependencies" Version="3.119.4" />
+  </ItemGroup>
+
 </Project>

Tests/TaskTrackingSystem.Tests/SimplePdfReportBuilderTests.cs

@@ -0,0 +1,21 @@
+using System.Text;
+using TaskTrackingSystem.Shared;
+using Xunit;
+
+namespace TaskTrackingSystem.Tests;
+
+public sealed class SimplePdfReportBuilderTests
+{
+    [Fact]
+    public void BuildTableReport_GeneratesPdfForBurmeseText()
+    {
+        var pdf = SimplePdfReportBuilder.BuildTableReport(
+            "လုပ်ငန်း အစီရင်ခံစာ",
+            ["ရှာဖွေရန်: လုပ်ငန်းအားလုံး"],
+            ["လုပ်ငန်း", "စီမံကိန်း", "အခြေအနေ"],
+            [["မြန်မာစာ စမ်းသပ်လုပ်ငန်း", "စီမံကိန်းတစ်", "လုပ်ဆောင်ဆဲ"]]);
+
+        Assert.True(pdf.Length > 1_000);
+        Assert.Equal("%PDF", Encoding.ASCII.GetString(pdf, 0, 4));
+    }
+}