Reduce Allocations (#821)

* Introduce ValueStringBuilder * Make NumericTokenizer and PlanTextTokenizer thread-safe * Replace ListPool with ArrayPoolBufferWriter * Seal ITokenizer classes * Eliminate array allocation in Type1ArrayTokenizer * Eliminate array allocation in AcroFormFactory * Eliminate StringBuilder allocation in Page.GetText * Optimize PdfSubpath.ToLines * Eliminate various allocations when parsing CompactFontFormat * Remove unused FromOctalInt helper * Ensure Pdf.Content is not null * Write ASCII values directly to stream (avoiding allocations) * Avoid encoding additional ASCII values * Eliminate allocations in TokenWriter.WriteName * Eliminate allocation in TokenWriter.WriteNumber * Add System.Memory reference to Fonts
UglyToad · Apr 28, 2024 · 7f42a8d · 7f42a8d
1 parent 1ef2e12
commit 7f42a8d
Show file tree

Hide file tree

Showing 32 changed files with 515 additions and 470 deletions.
diff --git a/src/UglyToad.PdfPig.Core/OctalHelpers.cs b/src/UglyToad.PdfPig.Core/OctalHelpers.cs
@@ -42,14 +42,6 @@ public static int FromOctalDigits(ReadOnlySpan<short> octal)
             return sum;
         }
 
-        /// <summary>
-        /// Interpret an int as octal.
-        /// </summary>
-        public static int FromOctalInt(int input)
-        {
-            return Convert.ToInt32($"{input}", 8);
-        }
-
         private static int QuickPower(int x, int pow)
         {
             int ret = 1;

diff --git a/src/UglyToad.PdfPig.Core/PdfSubpath.cs b/src/UglyToad.PdfPig.Core/PdfSubpath.cs
@@ -634,15 +634,15 @@ public override IReadOnlyList<Line> ToLines(int n)
                     throw new ArgumentException("BezierCurve.ToLines(): n must be greater than 0.");
                 }
 
-                List<Line> lines = new List<Line>();
+                var lines = new Line[n];
                 var previousPoint = StartPoint;
 
                 for (int p = 1; p <= n; p++)
                 {
                     double t = p / (double)n;
                     var currentPoint = new PdfPoint(ValueWithT(StartPoint.X, ControlPoint.X, EndPoint.X, t),
                         ValueWithT(StartPoint.Y, ControlPoint.Y, EndPoint.Y, t));
-                    lines.Add(new Line(previousPoint, currentPoint));
+                    lines[p - 1] = new Line(previousPoint, currentPoint);
                     previousPoint = currentPoint;
                 }
 
@@ -793,15 +793,15 @@ public override IReadOnlyList<Line> ToLines(int n)
                     throw new ArgumentException("BezierCurve.ToLines(): n must be greater than 0.");
                 }
 
-                List<Line> lines = new List<Line>();
+                var lines = new Line[n];
                 var previousPoint = StartPoint;
 
                 for (int p = 1; p <= n; p++)
                 {
                     double t = p / (double)n;
                     var currentPoint = new PdfPoint(ValueWithT(StartPoint.X, FirstControlPoint.X, SecondControlPoint.X, EndPoint.X, t),
                                                     ValueWithT(StartPoint.Y, FirstControlPoint.Y, SecondControlPoint.Y, EndPoint.Y, t));
-                    lines.Add(new Line(previousPoint, currentPoint));
+                    lines[p - 1] = new Line(previousPoint, currentPoint);
                     previousPoint = currentPoint;
                 }
                 return lines;

diff --git a/src/UglyToad.PdfPig.Fonts/CompactFontFormat/Charsets/CompactFontFormatCharset.cs b/src/UglyToad.PdfPig.Fonts/CompactFontFormat/Charsets/CompactFontFormatCharset.cs
@@ -10,13 +10,8 @@ internal abstract class CompactFontFormatCharset : ICompactFontFormatCharset
 
         public bool IsCidCharset { get; } = false;
 
-        protected CompactFontFormatCharset(IReadOnlyList<(int glyphId, int stringId, string name)> data)
+        protected CompactFontFormatCharset(ReadOnlySpan<(int glyphId, int stringId, string name)> data)
         {
-            if (data == null)
-            {
-                throw new ArgumentNullException(nameof(data));
-            }
-
             var dictionary = new Dictionary<int, (int stringId, string name)>
             {
                 {0, (0, ".notdef")}

diff --git a/src/UglyToad.PdfPig.Fonts/CompactFontFormat/Charsets/CompactFontFormatFormat0Charset.cs b/src/UglyToad.PdfPig.Fonts/CompactFontFormat/Charsets/CompactFontFormatFormat0Charset.cs
@@ -1,14 +1,14 @@
 namespace UglyToad.PdfPig.Fonts.CompactFontFormat.Charsets
 {
-    using System.Collections.Generic;
+    using System;
 
     /// <summary>
     /// A Charset from a Compact Font Format font file best for fonts with relatively unordered string ids.
     /// </summary>
-    internal class CompactFontFormatFormat0Charset : CompactFontFormatCharset
+    internal sealed class CompactFontFormatFormat0Charset : CompactFontFormatCharset
     {
-        public CompactFontFormatFormat0Charset(IReadOnlyList<(int glyphId, int stringId, string name)> data)
-        :base(data)
+        public CompactFontFormatFormat0Charset(ReadOnlySpan<(int glyphId, int stringId, string name)> data)
+            : base(data)
         {
         }
     }

diff --git a/src/UglyToad.PdfPig.Fonts/CompactFontFormat/Charsets/CompactFontFormatFormat1Charset.cs b/src/UglyToad.PdfPig.Fonts/CompactFontFormat/Charsets/CompactFontFormatFormat1Charset.cs
@@ -1,13 +1,13 @@
 namespace UglyToad.PdfPig.Fonts.CompactFontFormat.Charsets
 {
-    using System.Collections.Generic;
+    using System;
 
     /// <summary>
     /// A Charset from a Compact Font Format font file best for fonts with well ordered string ids.
     /// </summary>
-    internal class CompactFontFormatFormat1Charset : CompactFontFormatCharset
+    internal sealed class CompactFontFormatFormat1Charset : CompactFontFormatCharset
     {
-        public CompactFontFormatFormat1Charset(IReadOnlyList<(int glyphId, int stringId, string name)> data)
+        public CompactFontFormatFormat1Charset(ReadOnlySpan<(int glyphId, int stringId, string name)> data)
             : base(data)
         {
         }

diff --git a/src/UglyToad.PdfPig.Fonts/CompactFontFormat/Charsets/CompactFontFormatFormat2Charset.cs b/src/UglyToad.PdfPig.Fonts/CompactFontFormat/Charsets/CompactFontFormatFormat2Charset.cs
@@ -1,13 +1,13 @@
 namespace UglyToad.PdfPig.Fonts.CompactFontFormat.Charsets
 {
-    using System.Collections.Generic;
+    using System;
 
     /// <summary>
     /// A Charset from a Compact Font Format font file best for fonts with a large number of well ordered string ids.
     /// </summary>
     internal class CompactFontFormatFormat2Charset : CompactFontFormatCharset
     {
-        public CompactFontFormatFormat2Charset(IReadOnlyList<(int glyphId, int stringId, string name)> data)
+        public CompactFontFormatFormat2Charset(ReadOnlySpan<(int glyphId, int stringId, string name)> data)
             : base(data)
         {
         }

diff --git a/src/UglyToad.PdfPig.Fonts/CompactFontFormat/CompactFontFormatEncodingReader.cs b/src/UglyToad.PdfPig.Fonts/CompactFontFormat/CompactFontFormatEncodingReader.cs
@@ -5,6 +5,7 @@
     using Charsets;
     using Encodings;
     using Fonts;
+    using UglyToad.PdfPig.Core;
 
     internal static class CompactFontFormatEncodingReader
     {
@@ -36,13 +37,13 @@ private static CompactFontFormatFormat0Encoding ReadFormat0Encoding(CompactFontF
         {
             var numberOfCodes = data.ReadCard8();
 
-            var values = new List<(int code, int sid, string str)>();
+            using var values = new ArrayPoolBufferWriter<(int code, int sid, string str)>();
             for (var i = 1; i <= numberOfCodes; i++)
             {
                 var code = data.ReadCard8();
                 var sid = charset.GetStringIdByGlyphId(i);
                 var str = ReadString(sid, stringIndex);
-                values.Add((code, sid, str));
+                values.Write((code, sid, str));
             }
 
             IReadOnlyList<CompactFontFormatBuiltInEncoding.Supplement> supplements = [];
@@ -51,7 +52,7 @@ private static CompactFontFormatFormat0Encoding ReadFormat0Encoding(CompactFontF
                 supplements = ReadSupplement(data, stringIndex);
             }
 
-            return new CompactFontFormatFormat0Encoding(values, supplements);
+            return new CompactFontFormatFormat0Encoding(values.WrittenSpan, supplements);
         }
 
         private static CompactFontFormatFormat1Encoding ReadFormat1Encoding(CompactFontFormatData data, ICompactFontFormatCharset charset, ReadOnlySpan<string> stringIndex, byte format)

diff --git a/src/UglyToad.PdfPig.Fonts/CompactFontFormat/CompactFontFormatFormat0Encoding.cs b/src/UglyToad.PdfPig.Fonts/CompactFontFormat/CompactFontFormatFormat0Encoding.cs
@@ -1,10 +1,12 @@
 namespace UglyToad.PdfPig.Fonts.CompactFontFormat
 {
+    using System;
     using System.Collections.Generic;
 
     internal sealed class CompactFontFormatFormat0Encoding : CompactFontFormatBuiltInEncoding
     {
-        public CompactFontFormatFormat0Encoding(IReadOnlyList<(int code, int sid, string str)> values,
+        public CompactFontFormatFormat0Encoding(
+            ReadOnlySpan<(int code, int sid, string str)> values,
             IReadOnlyList<Supplement> supplements) : base(supplements)
         {
             Add(0, 0, NotDefined);

diff --git a/src/UglyToad.PdfPig.Fonts/CompactFontFormat/CompactFontFormatIndividualFontParser.cs b/src/UglyToad.PdfPig.Fonts/CompactFontFormat/CompactFontFormatIndividualFontParser.cs
@@ -139,42 +139,42 @@ private static ICompactFontFormatCharset ReadCharset(CompactFontFormatData data,
             {
                 case 0:
                     {
-                        var glyphToNamesAndStringId = new List<(int glyphId, int stringId, string name)>();
+                        using var glyphToNamesAndStringId = new ArrayPoolBufferWriter<(int glyphId, int stringId, string name)>();
 
                         for (var glyphId = 1; glyphId < charStringIndex.Count; glyphId++)
                         {
                             var stringId = data.ReadSid();
-                            glyphToNamesAndStringId.Add((glyphId, stringId, ReadString(stringId, stringIndex)));
+                            glyphToNamesAndStringId.Write((glyphId, stringId, ReadString(stringId, stringIndex)));
                         }
 
-                        return new CompactFontFormatFormat0Charset(glyphToNamesAndStringId);
+                        return new CompactFontFormatFormat0Charset(glyphToNamesAndStringId.WrittenSpan);
                     }
                 case 1:
                 case 2:
                     {
-                        var glyphToNamesAndStringId = new List<(int glyphId, int stringId, string name)>();
+                        using var glyphToNamesAndStringId = new ArrayPoolBufferWriter<(int glyphId, int stringId, string name)>();
 
                         for (var glyphId = 1; glyphId < charStringIndex.Count; glyphId++)
                         {
                             var firstSid = data.ReadSid();
                             var numberInRange = format == 1 ? data.ReadCard8() : data.ReadCard16();
 
-                            glyphToNamesAndStringId.Add((glyphId, firstSid, ReadString(firstSid, stringIndex)));
+                            glyphToNamesAndStringId.Write((glyphId, firstSid, ReadString(firstSid, stringIndex)));
                             for (var i = 0; i < numberInRange; i++)
                             {
                                 glyphId++;
                                 var sid = firstSid + i + 1;
-                                glyphToNamesAndStringId.Add((glyphId, sid, ReadString(sid, stringIndex)));
+                                glyphToNamesAndStringId.Write((glyphId, sid, ReadString(sid, stringIndex)));
                             }
                         }
 
                         if (format == 1)
                         {
 
-                            return new CompactFontFormatFormat1Charset(glyphToNamesAndStringId);
+                            return new CompactFontFormatFormat1Charset(glyphToNamesAndStringId.WrittenSpan);
                         }
 
-                        return new CompactFontFormatFormat2Charset(glyphToNamesAndStringId);
+                        return new CompactFontFormatFormat2Charset(glyphToNamesAndStringId.WrittenSpan);
                     }
                 default:
                     throw new InvalidOperationException($"Unrecognized format for the Charset table in a CFF font. Got: {format}.");

diff --git a/src/UglyToad.PdfPig.Fonts/Type1/Parser/Type1ArrayTokenizer.cs b/src/UglyToad.PdfPig.Fonts/Type1/Parser/Type1ArrayTokenizer.cs
@@ -9,11 +9,13 @@
     using Tokenization;
 
     /// <inheritdoc />
-    public class Type1ArrayTokenizer : ITokenizer
+    public sealed class Type1ArrayTokenizer : ITokenizer
     {
         /// <inheritdoc />
         public bool ReadsNextByte { get; } = false;
 
+        private static readonly string[] Space = [" "];
+
         /// <inheritdoc />
         public bool TryTokenize(byte currentByte, IInputBytes inputBytes, out IToken token)
         {
@@ -36,7 +38,7 @@ public bool TryTokenize(byte currentByte, IInputBytes inputBytes, out IToken tok
                 builder.Append((char) inputBytes.CurrentByte);
             }
 
-            var parts = builder.ToString().Split(new[] {" "}, StringSplitOptions.RemoveEmptyEntries);
+            var parts = builder.ToString().Split(Space, StringSplitOptions.RemoveEmptyEntries);
 
             var tokens = new List<IToken>();
 
@@ -50,7 +52,7 @@ public bool TryTokenize(byte currentByte, IInputBytes inputBytes, out IToken tok
                     }
                     else
                     {
-                        tokens.Add(OperatorToken.Create(part));
+                        tokens.Add(OperatorToken.Create(part.AsSpan()));
                     }
 
                     continue;
@@ -68,8 +70,7 @@ public bool TryTokenize(byte currentByte, IInputBytes inputBytes, out IToken tok
                     continue;
                 }
 
-                tokens.Add(OperatorToken.Create(part));
-
+                tokens.Add(OperatorToken.Create(part.AsSpan()));
             }
 
             token = new ArrayToken(tokens);

diff --git a/src/UglyToad.PdfPig.Fonts/UglyToad.PdfPig.Fonts.csproj b/src/UglyToad.PdfPig.Fonts/UglyToad.PdfPig.Fonts.csproj
@@ -38,4 +38,7 @@
   <ItemGroup Condition="'$(TargetFramework)'=='net462'">
     <PackageReference Include="System.ValueTuple" Version="4.5.0" />
   </ItemGroup>
+  <ItemGroup Condition="'$(TargetFramework)'=='netstandard2.0' or '$(TargetFramework)'=='net462' or '$(TargetFramework)'=='net471'">
+    <PackageReference Include="System.Memory" Version="4.5.5" />
+  </ItemGroup>
 </Project>
diff --git a/src/UglyToad.PdfPig.Tests/Util/OctalHelpersTests.cs b/src/UglyToad.PdfPig.Tests/Util/OctalHelpersTests.cs
diff --git a/src/UglyToad.PdfPig.Tokenization/ArrayTokenizer.cs b/src/UglyToad.PdfPig.Tokenization/ArrayTokenizer.cs
@@ -5,7 +5,7 @@
     using Scanner;
     using Tokens;
 
-    internal class ArrayTokenizer : ITokenizer
+    internal sealed class ArrayTokenizer : ITokenizer
     {
         private readonly bool usePdfDocEncoding;
 

diff --git a/src/UglyToad.PdfPig.Tokenization/CommentTokenizer.cs b/src/UglyToad.PdfPig.Tokenization/CommentTokenizer.cs
@@ -1,10 +1,10 @@
 namespace UglyToad.PdfPig.Tokenization
 {
-    using System.Text;
     using Core;
+    using System.Text;
     using Tokens;
 
-    internal class CommentTokenizer : ITokenizer
+    internal sealed class CommentTokenizer : ITokenizer
     {
         public bool ReadsNextByte { get; } = true;
 
@@ -17,7 +17,7 @@ public bool TryTokenize(byte currentByte, IInputBytes inputBytes, out IToken tok
                 return false;
             }
 
-            var builder = new StringBuilder();
+            using var builder = new ValueStringBuilder();
 
             while (inputBytes.MoveNext() && !ReadHelper.IsEndOfLine(inputBytes.CurrentByte))
             {

diff --git a/src/UglyToad.PdfPig.Tokenization/EndOfLineTokenizer.cs b/src/UglyToad.PdfPig.Tokenization/EndOfLineTokenizer.cs
@@ -6,7 +6,7 @@
     /// <summary>
     /// Read an <see cref="EndOfLineToken"/>.
     /// </summary>
-    public class EndOfLineTokenizer : ITokenizer
+    public sealed class EndOfLineTokenizer : ITokenizer
     {
         /// <inheritdoc />
         public bool ReadsNextByte { get; } = false;

diff --git a/src/UglyToad.PdfPig.Tokenization/ListPool.cs b/src/UglyToad.PdfPig.Tokenization/ListPool.cs
diff --git a/src/UglyToad.PdfPig.Tokenization/NameTokenizer.cs b/src/UglyToad.PdfPig.Tokenization/NameTokenizer.cs
@@ -9,7 +9,7 @@
     using System.Text.Unicode;
 #endif
 
-    internal class NameTokenizer : ITokenizer
+    internal sealed class NameTokenizer : ITokenizer
     {
         static NameTokenizer()
         {