expose font details on individual letters

also fixes a regression for image extraction
2025-09-18 18:27:55 +08:00 · 2020-04-25 17:15:26 +01:00
parent 98dd736f94
commit 09b951f667
21 changed files with 204 additions and 13 deletions
--- a/src/UglyToad.PdfPig.Fonts/CompactFontFormat/CompactFontFormatFont.cs
+++ b/src/UglyToad.PdfPig.Fonts/CompactFontFormat/CompactFontFormatFont.cs
@@ -29,6 +29,16 @@
        /// </summary>
        public TransformationMatrix FontMatrix => TopDictionary.FontMatrix;

+        /// <summary>
+        /// The value of Weight from the top dictionary or <see langword="null"/>.
+        /// </summary>
+        public string Weight => TopDictionary?.Weight;
+
+        /// <summary>
+        /// The value of Italic Angle from the top dictionary or 0.
+        /// </summary>
+        public decimal ItalicAngle => TopDictionary?.ItalicAngle ?? 0;
+
        internal CompactFontFormatFont(CompactFontFormatTopLevelDictionary topDictionary, CompactFontFormatPrivateDictionary privateDictionary,
            ICompactFontFormatCharset charset,
            Union<Type1CharStrings, Type2CharStrings> charStrings, Encoding fontEncoding)
--- a/src/UglyToad.PdfPig.Tests/PublicApiScannerTests.cs
+++ b/src/UglyToad.PdfPig.Tests/PublicApiScannerTests.cs
@@ -91,6 +91,7 @@
                "UglyToad.PdfPig.PdfFonts.DescriptorFontFile",
                "UglyToad.PdfPig.PdfFonts.FontDescriptor",
                "UglyToad.PdfPig.PdfFonts.FontDescriptorFlags",
+                "UglyToad.PdfPig.PdfFonts.FontDetails",
                "UglyToad.PdfPig.PdfFonts.FontStretch",
                "UglyToad.PdfPig.Geometry.GeometryExtensions",
                "UglyToad.PdfPig.Graphics.Colors.CMYKColor",
--- a/src/UglyToad.PdfPig/Content/Letter.cs
+++ b/src/UglyToad.PdfPig/Content/Letter.cs
@@ -2,6 +2,7 @@
 {
    using Core;
    using Graphics.Colors;
+    using PdfFonts;

    /// <summary>
    /// A glyph or combination of glyphs (characters) drawn by a PDF content stream.
@@ -53,7 +54,12 @@
        /// <summary>
        /// The name of the font.
        /// </summary>
-        public string FontName { get; }
+        public string FontName => Font?.Name;
+
+        /// <summary>
+        /// Details about the font for this letter.
+        /// </summary>
+        public FontDetails Font { get; }

        /// <summary>
        /// The color of the letter.
@@ -79,7 +85,7 @@
            PdfPoint endBaseLine,
            double width,
            double fontSize,
-            string fontName, 
+            FontDetails font, 
            IColor color,
            double pointSize,
            int textSequence)
@@ -90,7 +96,7 @@
            EndBaseLine = endBaseLine;
            Width = width;
            FontSize = fontSize;
-            FontName = fontName;
+            Font = font;
            Color = color ?? GrayColor.Black;
            PointSize = pointSize;
            TextSequence = textSequence;
--- a/src/UglyToad.PdfPig/Graphics/ContentStreamProcessor.cs
+++ b/src/UglyToad.PdfPig/Graphics/ContentStreamProcessor.cs
@@ -251,7 +251,7 @@
                    transformedPdfBounds.BottomRight,
                    transformedPdfBounds.Width,
                    fontSize,
-                    font.Name.Data,
+                    font.Details,
                    color,
                    pointSize,
                    textSequence);
--- a/src/UglyToad.PdfPig/PdfFonts/CidFonts/ICidFont.cs
+++ b/src/UglyToad.PdfPig/PdfFonts/CidFonts/ICidFont.cs
@@ -34,6 +34,8 @@
        /// </summary>
        CharacterIdentifierSystemInfo SystemInfo { get; }

+        FontDetails Details { get; }
+
        TransformationMatrix FontMatrix { get; }

        CidFontType CidFontType { get; }
--- a/src/UglyToad.PdfPig/PdfFonts/CidFonts/ICidFontProgram.cs
+++ b/src/UglyToad.PdfPig/PdfFonts/CidFonts/ICidFontProgram.cs
@@ -8,6 +8,8 @@
    /// </summary>
    internal interface ICidFontProgram
    {
+        FontDetails Details { get; }
+
        bool TryGetBoundingBox(int characterIdentifier, out PdfRectangle boundingBox);

        bool TryGetBoundingBox(int characterIdentifier, Func<int, int?> characterCodeToGlyphId, out PdfRectangle boundingBox);
--- a/src/UglyToad.PdfPig/PdfFonts/CidFonts/PdfCidCompactFontFormatFont.cs
+++ b/src/UglyToad.PdfPig/PdfFonts/CidFonts/PdfCidCompactFontFormatFont.cs
@@ -8,9 +8,36 @@
    {
        private readonly CompactFontFormatFontCollection fontCollection;

+        public FontDetails Details { get; }
+
        public PdfCidCompactFontFormatFont(CompactFontFormatFontCollection fontCollection)
        {
            this.fontCollection = fontCollection;
+            Details = GetDetails(fontCollection?.FirstFont);
+        }
+
+        private static FontDetails GetDetails(CompactFontFormatFont font)
+        {
+            if (font == null)
+            {
+                return FontDetails.GetDefault();
+            }
+
+            FontDetails WithWeightValues(bool isbold, int weight) => new FontDetails(null, isbold, weight, font.ItalicAngle != 0);
+
+            switch (font.Weight?.ToLowerInvariant())
+            {
+                case "light":
+                    return WithWeightValues(false, 300);
+                case "semibold":
+                    return WithWeightValues(true, 600);
+                case "bold":
+                    return WithWeightValues(true, FontDetails.BoldWeight);
+                case "black":
+                    return WithWeightValues(true, 900);
+                default:
+                    return WithWeightValues(false, FontDetails.DefaultWeight);
+            }
        }

        public TransformationMatrix GetFontTransformationMatrix() => fontCollection.GetFirstTransformationMatrix();
--- a/src/UglyToad.PdfPig/PdfFonts/CidFonts/PdfCidTrueTypeFont.cs
+++ b/src/UglyToad.PdfPig/PdfFonts/CidFonts/PdfCidTrueTypeFont.cs
@@ -3,14 +3,23 @@
    using System;
    using Core;
    using Fonts.TrueType;
+    using Fonts.TrueType.Tables;

    internal class PdfCidTrueTypeFont : ICidFontProgram
    {
        private readonly TrueTypeFont font;
+
+        public FontDetails Details { get; }
        
        public PdfCidTrueTypeFont(TrueTypeFont font)
        {
            this.font = font ?? throw new ArgumentNullException(nameof(font));
+
+            var header = font.TableRegister.HeaderTable;
+            var isBold = header.MacStyle.HasFlag(HeaderTable.HeaderMacStyle.Bold);
+            Details = new FontDetails(font.Name, isBold,
+                isBold ? FontDetails.BoldWeight : FontDetails.DefaultWeight,
+                header.MacStyle.HasFlag(HeaderTable.HeaderMacStyle.Italic));
        }

        public bool TryGetBoundingBox(int characterIdentifier, out PdfRectangle boundingBox) => TryGetBoundingBox(characterIdentifier, null, out boundingBox);
--- a/src/UglyToad.PdfPig/PdfFonts/CidFonts/Type0CidFont.cs
+++ b/src/UglyToad.PdfPig/PdfFonts/CidFonts/Type0CidFont.cs
@@ -25,6 +25,9 @@

        public CharacterIdentifierSystemInfo SystemInfo { get; }

+        public FontDetails Details => fontProgram?.Details ?? Descriptor?.ToDetails(BaseFont?.Data) 
+                                      ?? FontDetails.GetDefault(BaseFont?.Data);
+
        public TransformationMatrix FontMatrix { get; }

        public CidFontType CidFontType => CidFontType.Type0;
--- a/src/UglyToad.PdfPig/PdfFonts/CidFonts/Type2CidFont.cs
+++ b/src/UglyToad.PdfPig/PdfFonts/CidFonts/Type2CidFont.cs
@@ -32,6 +32,9 @@

        public FontDescriptor Descriptor { get; }

+        public FontDetails Details => fontProgram?.Details ?? Descriptor?.ToDetails(BaseFont?.Data)
+                                      ?? FontDetails.GetDefault(BaseFont?.Data);
+
        public Type2CidFont(NameToken type, NameToken subType, NameToken baseFont, CharacterIdentifierSystemInfo systemInfo,
            FontDescriptor descriptor, ICidFontProgram fontProgram,
            VerticalWritingMetrics verticalWritingMetrics,
--- a/src/UglyToad.PdfPig/PdfFonts/Composite/Type0Font.cs
+++ b/src/UglyToad.PdfPig/PdfFonts/Composite/Type0Font.cs
@@ -36,6 +36,8 @@

        public bool IsVertical => CMap.WritingMode == WritingMode.Vertical;

+        public FontDetails Details { get; }
+
        public Type0Font(NameToken baseFont, ICidFont cidFont, CMap cmap, CMap toUnicodeCMap,
            CMap ucs2CMap,
            bool isChineseJapaneseOrKorean)
@@ -47,6 +49,8 @@
            CidFont = cidFont ?? throw new ArgumentNullException(nameof(cidFont));
            CMap = cmap ?? throw new ArgumentNullException(nameof(cmap));
            ToUnicode = new ToUnicodeCMap(toUnicodeCMap);
+            Details = cidFont.Details?.WithName(Name.Data) 
+                      ?? FontDetails.GetDefault(Name.Data);
        }

        public int ReadCharacterCode(IInputBytes bytes, out int codeLength)
--- a/src/UglyToad.PdfPig/PdfFonts/FontDescriptor.cs
+++ b/src/UglyToad.PdfPig/PdfFonts/FontDescriptor.cs
@@ -179,6 +179,14 @@
            CharSet = builder.CharSet;
        }

+        internal FontDetails ToDetails(string name = null)
+        {
+            return new FontDetails(name ?? FontName ?? string.Empty,
+                FontWeight > 500,
+                (int)FontWeight,
+                Flags.HasFlag(FontDescriptorFlags.Italic));
+        }
+
        /// <summary>
        /// Provides a mutable way to construct a <see cref="FontDescriptor"/>.
        /// </summary>
--- a/src/UglyToad.PdfPig/PdfFonts/FontDetails.cs
+++ b/src/UglyToad.PdfPig/PdfFonts/FontDetails.cs
@@ -0,0 +1,66 @@
+namespace UglyToad.PdfPig.PdfFonts
+{
+    /// <summary>
+    /// Summary details of the font used to draw a glyph.
+    /// </summary>
+    public class FontDetails
+    {
+        /// <summary>
+        /// The normal weight for a font.
+        /// </summary>
+        public const int DefaultWeight = 500;
+
+        /// <summary>
+        /// The bold weight for a font.
+        /// </summary>
+        public const int BoldWeight = 700;
+        
+        /// <summary>
+        /// The font name.
+        /// </summary>
+        public string Name { get; }
+
+        /// <summary>
+        /// Whether the font is bold.
+        /// </summary>
+        public bool IsBold { get; }
+
+        /// <summary>
+        /// The font weight, values above 500 represent bold.
+        /// </summary>
+        public int Weight { get; }
+        
+        /// <summary>
+        /// Whether the font is italic.
+        /// </summary>
+        public bool IsItalic { get; }
+
+        /// <summary>
+        /// Create a new <see cref="FontDetails"/>.
+        /// </summary>
+        public FontDetails(string name, bool isBold, int weight, bool isItalic)
+        {
+            Name = name ?? string.Empty;
+            IsBold = isBold;
+            Weight = weight;
+            IsItalic = isItalic;
+        }
+
+        internal static FontDetails GetDefault(string name = null) => new FontDetails(name ?? string.Empty,
+            false,
+            DefaultWeight,
+            false);
+
+        internal FontDetails WithName(string name) => name != null
+            ? new FontDetails(name, IsBold, Weight, IsItalic)
+            : this;
+
+        /// <inheritdoc />
+        public override string ToString()
+        {
+            var boldString = IsBold ? " (bold)" : string.Empty;
+            var italicString = IsItalic ? " (italic)" : string.Empty;
+            return $"{Name}{boldString}{italicString}";
+        }
+    }
+}
--- a/src/UglyToad.PdfPig/PdfFonts/IFont.cs
+++ b/src/UglyToad.PdfPig/PdfFonts/IFont.cs
@@ -9,6 +9,8 @@
        
        bool IsVertical { get; }

+        FontDetails Details { get; }
+
        int ReadCharacterCode(IInputBytes bytes, out int codeLength);

        bool TryGetUnicode(int characterCode, out string value);
--- a/src/UglyToad.PdfPig/PdfFonts/Simple/TrueTypeSimpleFont.cs
+++ b/src/UglyToad.PdfPig/PdfFonts/Simple/TrueTypeSimpleFont.cs
@@ -35,6 +35,8 @@

        public bool IsVertical { get; }

+        public FontDetails Details { get; }
+
        [NotNull]
        public ToUnicodeCMap ToUnicode { get; set; }

@@ -55,6 +57,8 @@
            Name = name;
            IsVertical = false;
            ToUnicode = new ToUnicodeCMap(toUnicodeCMap);
+            Details = descriptor?.ToDetails(Name?.Data) 
+                      ?? FontDetails.GetDefault(Name?.Data);
        }

        public int ReadCharacterCode(IInputBytes bytes, out int codeLength)
--- a/src/UglyToad.PdfPig/PdfFonts/Simple/TrueTypeStandard14FallbackSimpleFont.cs
+++ b/src/UglyToad.PdfPig/PdfFonts/Simple/TrueTypeStandard14FallbackSimpleFont.cs
@@ -26,6 +26,8 @@

        public bool IsVertical { get; } = false;

+        public FontDetails Details { get; set; }
+
        public TrueTypeStandard14FallbackSimpleFont(NameToken name, AdobeFontMetrics fontMetrics, Encoding encoding, TrueTypeFont font,
            MetricOverrides overrides)
        {
@@ -34,6 +36,10 @@
            this.font = font;
            this.overrides = overrides;
            Name = name;
+            Details = fontMetrics == null ? FontDetails.GetDefault(Name?.Data) : new FontDetails(Name?.Data,
+                fontMetrics.Weight == "Bold",
+                fontMetrics.Weight == "Bold" ? 700 : FontDetails.DefaultWeight,
+                fontMetrics.ItalicAngle != 0);
        }

        public int ReadCharacterCode(IInputBytes bytes, out int codeLength)
--- a/src/UglyToad.PdfPig/PdfFonts/Simple/Type1FontSimple.cs
+++ b/src/UglyToad.PdfPig/PdfFonts/Simple/Type1FontSimple.cs
@@ -41,6 +41,8 @@

        public bool IsVertical { get; } = false;

+        public FontDetails Details { get; }
+
        public Type1FontSimple(NameToken name, int firstChar, int lastChar, double[] widths, FontDescriptor fontDescriptor, Encoding encoding, 
            CMap toUnicodeCMap,
            Union<Type1Font, CompactFontFormatFontCollection> fontProgram)
@@ -70,6 +72,8 @@
            fontMatrix = matrix;

            Name = name;
+            Details = fontDescriptor?.ToDetails(name?.Data) 
+                      ?? FontDetails.GetDefault(name?.Data);
        }

        public int ReadCharacterCode(IInputBytes bytes, out int codeLength)
--- a/src/UglyToad.PdfPig/PdfFonts/Simple/Type1Standard14Font.cs
+++ b/src/UglyToad.PdfPig/PdfFonts/Simple/Type1Standard14Font.cs
@@ -17,8 +17,11 @@ namespace UglyToad.PdfPig.PdfFonts.Simple
        private readonly Encoding encoding;

        public NameToken Name { get; }
+
        public bool IsVertical { get; }

+        public FontDetails Details { get; }
+
        private readonly TransformationMatrix fontMatrix = TransformationMatrix.FromValues(0.001, 0, 0, 0.001, 0, 0);

        public Type1Standard14Font(AdobeFontMetrics standardFontMetrics, Encoding overrideEncoding = null)
@@ -29,6 +32,10 @@ namespace UglyToad.PdfPig.PdfFonts.Simple
            Name = NameToken.Create(standardFontMetrics.FontName);
            
            IsVertical = false;
+            Details = new FontDetails(Name.Data,
+                standardFontMetrics.Weight == "Bold",
+                standardFontMetrics.Weight == "Bold" ? 700 : FontDetails.DefaultWeight,
+                standardFontMetrics.ItalicAngle != 0);
        }

        public int ReadCharacterCode(IInputBytes bytes, out int codeLength)
--- a/src/UglyToad.PdfPig/PdfFonts/Simple/Type3Font.cs
+++ b/src/UglyToad.PdfPig/PdfFonts/Simple/Type3Font.cs
@@ -24,6 +24,8 @@

        public bool IsVertical { get; } = false;

+        public FontDetails Details { get; }
+
        public Type3Font(NameToken name, PdfRectangle boundingBox, TransformationMatrix fontMatrix,
            Encoding encoding, int firstChar, int lastChar, double[] widths,
            CMap toUnicodeCMap)
@@ -37,6 +39,7 @@
            this.lastChar = lastChar;
            this.widths = widths;
            this.toUnicodeCMap = new ToUnicodeCMap(toUnicodeCMap);
+            Details = FontDetails.GetDefault(name?.Data);
        }

        public int ReadCharacterCode(IInputBytes bytes, out int codeLength)
--- a/src/UglyToad.PdfPig/Writer/PdfPageBuilder.cs
+++ b/src/UglyToad.PdfPig/Writer/PdfPageBuilder.cs
@@ -16,8 +16,9 @@
    using System;
    using System.Collections.Generic;
    using System.IO;
+    using PdfFonts;
    using Tokens;
-    using UglyToad.PdfPig.Graphics.Operations.PathPainting;
+    using Graphics.Operations.PathPainting;

    /// <summary>
    /// A builder used to add construct a page in a PDF document.
@@ -388,7 +389,7 @@

                var documentSpace = textMatrix.Transform(renderingMatrix.Transform(fontMatrix.Transform(rect)));

-                var letter = new Letter(c.ToString(), documentSpace, advanceRect.BottomLeft, advanceRect.BottomRight, width, (double)fontSize, font.Name,
+                var letter = new Letter(c.ToString(), documentSpace, advanceRect.BottomLeft, advanceRect.BottomRight, width, (double)fontSize, FontDetails.GetDefault(font.Name),
                    GrayColor.Black,
                    (double)fontSize,
                    textSequence);
--- a/src/UglyToad.PdfPig/XObjects/XObjectFactory.cs
+++ b/src/UglyToad.PdfPig/XObjects/XObjectFactory.cs
@@ -1,4 +1,6 @@
-namespace UglyToad.PdfPig.XObjects
+using UglyToad.PdfPig.Parser.Parts;
+
+namespace UglyToad.PdfPig.XObjects
 {
    using System;
    using System.Collections.Generic;
@@ -66,14 +68,35 @@
            var interpolate = dictionary.TryGet(NameToken.Interpolate, pdfScanner, out BooleanToken interpolateToken)
                              && interpolateToken.Data;

-            var filters = filterProvider.GetFilters(xObject.Stream.StreamDictionary);
-            var supportsFilters = true;
-            foreach (var filter in filters)
+            DictionaryToken filterDictionary = xObject.Stream.StreamDictionary;
+            if (xObject.Stream.StreamDictionary.TryGet(NameToken.Filter, out var filterToken)
+                && filterToken is IndirectReferenceToken)
            {
-                if (!filter.IsSupported)
+                if (filterDictionary.TryGet(NameToken.Filter, pdfScanner, out ArrayToken filterArray))
                {
-                    supportsFilters = false;
-                    break;
+                    filterDictionary = filterDictionary.With(NameToken.Filter, filterArray);
+                }
+                else if (filterDictionary.TryGet(NameToken.Filter, pdfScanner, out NameToken filterNameToken))
+                {
+                    filterDictionary = filterDictionary.With(NameToken.Filter, filterNameToken);
+                }
+                else
+                {
+                    filterDictionary = null;
+                }
+            }
+
+            var supportsFilters = filterDictionary != null;
+            if (filterDictionary != null)
+            {
+                var filters = filterProvider.GetFilters(filterDictionary);
+                foreach (var filter in filters)
+                {
+                    if (!filter.IsSupported)
+                    {
+                        supportsFilters = false;
+                        break;
+                    }
                }
            }