Use pdfScanner in ReadVerticalDisplacements and fix #693 and return 0…

… in CMap on exception in ReadByte() if useLenientParsing is true and fix #692
UglyToad · Oct 18, 2024 · e10609e · e10609e
1 parent ea95a7a
commit e10609e
Show file tree

Hide file tree

Showing 11 changed files with 74 additions and 34 deletions.
diff --git a/src/UglyToad.PdfPig.Tests/Integration/Documents/cmap-parsing-exception.pdf b/src/UglyToad.PdfPig.Tests/Integration/Documents/cmap-parsing-exception.pdf
diff --git a/src/UglyToad.PdfPig.Tests/Integration/Documents/reference-2-numeric-error.pdf b/src/UglyToad.PdfPig.Tests/Integration/Documents/reference-2-numeric-error.pdf
diff --git a/src/UglyToad.PdfPig.Tests/Integration/GithubIssuesTests.cs b/src/UglyToad.PdfPig.Tests/Integration/GithubIssuesTests.cs
@@ -4,6 +4,36 @@
 
     public class GithubIssuesTests
     {
+        [Fact]
+        public void Issue693()
+        {
+            var doc = IntegrationHelpers.GetDocumentPath("reference-2-numeric-error.pdf");
+
+            using (var document = PdfDocument.Open(doc, new ParsingOptions() { UseLenientParsing = true, SkipMissingFonts = true }))
+            {
+                var page1 = document.GetPage(1);
+                Assert.Equal(1269, page1.Letters.Count);
+            }
+        }
+
+        [Fact]
+        public void Issue692()
+        {
+            var doc = IntegrationHelpers.GetDocumentPath("cmap-parsing-exception.pdf");
+
+            using (var document = PdfDocument.Open(doc, new ParsingOptions() { UseLenientParsing = true, SkipMissingFonts = true }))
+            {
+                var page1 = document.GetPage(1);
+                Assert.Equal(796, page1.Letters.Count);
+            }
+
+            using (var document = PdfDocument.Open(doc, new ParsingOptions() { UseLenientParsing = false, SkipMissingFonts = false }))
+            {
+                var ex = Assert.Throws<InvalidOperationException>(() => document.GetPage(1));
+                Assert.StartsWith("Read byte called on input bytes which was at end of byte set.", ex.Message);
+            }
+        }
+
         [Fact]
         public void Issue874()
         {

diff --git a/src/UglyToad.PdfPig.Tests/Integration/IntegrationDocumentTests.cs b/src/UglyToad.PdfPig.Tests/Integration/IntegrationDocumentTests.cs
@@ -7,7 +7,8 @@ public class IntegrationDocumentTests
         [
             "issue_671.pdf",
             "GHOSTSCRIPT-698363-0.pdf",
-            "ErcotFacts.pdf"
+            "ErcotFacts.pdf",
+            "cmap-parsing-exception.pdf"
         ];
 
         [Theory]

diff --git a/src/UglyToad.PdfPig/Parser/PdfDocumentFactory.cs b/src/UglyToad.PdfPig/Parser/PdfDocumentFactory.cs
@@ -166,7 +166,7 @@ private static PdfDocument OpenDocument(
                 cidFontFactory,
                 filterProvider,
                 pdfScanner,
-                parsingOptions.Logger);
+                parsingOptions);
 
             var type1Handler = new Type1FontHandler(pdfScanner, filterProvider, encodingReader);
 

diff --git a/src/UglyToad.PdfPig/PdfExtensions.cs b/src/UglyToad.PdfPig/PdfExtensions.cs
@@ -1,11 +1,10 @@
 namespace UglyToad.PdfPig
 {
-    using System;
-    using System.Collections.Generic;
+    using System;
     using System.Diagnostics.CodeAnalysis;
     using Core;
     using Filters;
-    using Parser.Parts;
+    using Parser.Parts;
     using Tokenization.Scanner;
     using Tokens;
 

diff --git a/src/UglyToad.PdfPig/PdfFonts/Cmap/CMap.cs b/src/UglyToad.PdfPig/PdfFonts/Cmap/CMap.cs
@@ -10,7 +10,7 @@
     /// The CMap (character code map) maps character codes to character identifiers (CIDs).
     /// The set of characters which a CMap refers to is the "character set" (charset).
     /// </summary>
-    internal class CMap
+    internal sealed class CMap
     {
         public CharacterIdentifierSystemInfo Info { get; }
 
@@ -140,13 +140,12 @@ public int ConvertToCid(int code)
             return 0;
         }
 
-
         public override string ToString()
         {
             return Name;
         }
 
-        public int ReadCode(IInputBytes bytes)
+        public int ReadCode(IInputBytes bytes, bool useLenientParsing)
         {
             if (hasEmptyCodespace)
             {
@@ -166,7 +165,7 @@ public int ReadCode(IInputBytes bytes)
                     break;
                 }
 
-                result[i] = ReadByte(bytes);
+                result[i] = ReadByte(bytes, useLenientParsing);
             }
 
             for (int i = minCodeLength - 1; i < maxCodeLength; i++)
@@ -181,17 +180,23 @@ public int ReadCode(IInputBytes bytes)
                 }
                 if (byteCount < maxCodeLength)
                 {
-                    result[byteCount] = ReadByte(bytes);
+                    result[byteCount] = ReadByte(bytes, useLenientParsing);
                 }
             }
 
             throw new PdfDocumentFormatException($"CMap is invalid, min code length was {minCodeLength}, max was {maxCodeLength}.");
         }
 
-        private static byte ReadByte(IInputBytes bytes)
+        private static byte ReadByte(IInputBytes bytes, bool useLenientParsing)
         {
             if (!bytes.MoveNext())
             {
+                if (useLenientParsing)
+                {
+                    // See issue #692
+                    return 0;
+                }
+
                 throw new InvalidOperationException("Read byte called on input bytes which was at end of byte set. Current offset: " + bytes.CurrentOffset);
             }
 
@@ -208,6 +213,5 @@ private static int ByteArrayToInt(ReadOnlySpan<byte> data)
             }
             return code;
         }
-
     }
 }
diff --git a/src/UglyToad.PdfPig/PdfFonts/Composite/ToUnicodeCMap.cs b/src/UglyToad.PdfPig/PdfFonts/Composite/ToUnicodeCMap.cs
@@ -9,7 +9,7 @@
     /// Defines the information content (actual text) of the font
     /// as opposed to the display format.
     /// </summary>
-    internal class ToUnicodeCMap
+    internal sealed class ToUnicodeCMap
     {
         private readonly CMap? cMap;
 
@@ -45,9 +45,9 @@ public bool TryGet(int code, [NotNullWhen(true)] out string? value)
             return cMap.TryConvertToUnicode(code, out value);
         }
 
-        public int ReadCode(IInputBytes inputBytes)
+        public int ReadCode(IInputBytes inputBytes, bool useLenientParsing)
         {
-            return cMap!.ReadCode(inputBytes);
+            return cMap!.ReadCode(inputBytes, useLenientParsing);
         }
     }
 }
diff --git a/src/UglyToad.PdfPig/PdfFonts/Composite/Type0Font.cs b/src/UglyToad.PdfPig/PdfFonts/Composite/Type0Font.cs
@@ -21,6 +21,8 @@ internal sealed class Type0Font : IFont, IVerticalWritingSupported
         private readonly Dictionary<int, CharacterBoundingBox> boundingBoxCache
             = new Dictionary<int, CharacterBoundingBox>();
 
+        private readonly bool useLenientParsing;
+
         public NameToken Name => BaseFont;
 
         public NameToken BaseFont { get; }
@@ -41,6 +43,7 @@ public Type0Font(
             CMap cmap,
             CMap? toUnicodeCMap,
             CMap? ucs2CMap,
+            ParsingOptions parsingOptions,
             bool isChineseJapaneseOrKorean)
         {
             this.ucs2CMap = ucs2CMap;
@@ -52,13 +55,15 @@ public Type0Font(
             ToUnicode = new ToUnicodeCMap(toUnicodeCMap);
             Details = cidFont.Details?.WithName(Name.Data)
                       ?? FontDetails.GetDefault(Name.Data);
+
+            useLenientParsing = parsingOptions.UseLenientParsing;
         }
 
         public int ReadCharacterCode(IInputBytes bytes, out int codeLength)
         {
             var current = bytes.CurrentOffset;
 
-            var code = CMap.ReadCode(bytes);
+            var code = CMap.ReadCode(bytes, useLenientParsing);
 
             codeLength = (int)(bytes.CurrentOffset - current);
 

diff --git a/src/UglyToad.PdfPig/PdfFonts/Parser/Handlers/Type0FontHandler.cs b/src/UglyToad.PdfPig/PdfFonts/Parser/Handlers/Type0FontHandler.cs
@@ -15,23 +15,25 @@
     using Tokens;
     using Util;
 
-    internal class Type0FontHandler : IFontHandler
+    internal sealed class Type0FontHandler : IFontHandler
     {
         private readonly CidFontFactory cidFontFactory;
         private readonly ILookupFilterProvider filterProvider;
         private readonly IPdfTokenScanner scanner;
         private readonly ILog logger;
+        private readonly ParsingOptions parsingOptions;
 
         public Type0FontHandler(
             CidFontFactory cidFontFactory,
             ILookupFilterProvider filterProvider,
             IPdfTokenScanner scanner,
-            ILog logger)
+            ParsingOptions parsingOptions)
         {
             this.cidFontFactory = cidFontFactory;
             this.filterProvider = filterProvider;
             this.scanner = scanner;
-            this.logger = logger;
+            logger = parsingOptions.Logger;
+            this.parsingOptions = parsingOptions;
         }
 
         public IFont Generate(DictionaryToken dictionary)
@@ -91,7 +93,7 @@ public IFont Generate(DictionaryToken dictionary)
                 }
             }
 
-            var font = new Type0Font(baseFont!, cidFont, cMap, toUnicodeCMap, ucs2CMap, isChineseJapaneseOrKorean);
+            var font = new Type0Font(baseFont!, cidFont, cMap, toUnicodeCMap, ucs2CMap, parsingOptions, isChineseJapaneseOrKorean);
 
             return font;
         }

diff --git a/src/UglyToad.PdfPig/PdfFonts/Parser/Parts/CidFontFactory.cs b/src/UglyToad.PdfPig/PdfFonts/Parser/Parts/CidFontFactory.cs
@@ -17,7 +17,7 @@
     using UglyToad.PdfPig.Logging;
     using Util;
 
-    internal class CidFontFactory
+    internal sealed class CidFontFactory
     {
         private readonly ILookupFilterProvider filterProvider;
         private readonly IPdfTokenScanner pdfScanner;
@@ -46,7 +46,7 @@ public CidFontFactory(ILog log, IPdfTokenScanner pdfScanner, ILookupFilterProvid
                 defaultWidth = defaultWidthToken.Double;
             }
 
-            var verticalWritingMetrics = ReadVerticalDisplacements(dictionary);
+            var verticalWritingMetrics = ReadVerticalDisplacements(dictionary, pdfScanner);
 
             FontDescriptor? descriptor = null;
             if (TryGetFontDescriptor(dictionary, out var descriptorDictionary))
@@ -190,7 +190,7 @@ private IReadOnlyDictionary<int, double> ReadWidths(DictionaryToken dict)
             return widths;
         }
 
-        private static VerticalWritingMetrics ReadVerticalDisplacements(DictionaryToken dict)
+        private static VerticalWritingMetrics ReadVerticalDisplacements(DictionaryToken dict, IPdfTokenScanner pdfScanner)
         {
             var verticalDisplacements = new Dictionary<int, double>();
             var positionVectors = new Dictionary<int, PdfVector>();
@@ -210,22 +210,21 @@ private static VerticalWritingMetrics ReadVerticalDisplacements(DictionaryToken
             }
 
             // vertical metrics for individual CIDs.
-            if (dict.TryGet(NameToken.W2, out var w2Token) && w2Token is ArrayToken w2)
+            if (dict.TryGet(NameToken.W2, pdfScanner, out ArrayToken? w2))
             {
                 for (var i = 0; i < w2.Data.Count; i++)
                 {
-                    var c = (NumericToken)w2.Data[i];
+                    var c = DirectObjectFinder.Get<NumericToken>(w2.Data[i], pdfScanner);
                     var next = w2.Data[++i];
-
-                    if (next is ArrayToken array)
+                    if (DirectObjectFinder.TryGet(next, pdfScanner, out ArrayToken? array))
                     {
                         for (var j = 0; j < array.Data.Count; j++)
                         {
                             var cid = c.Int + j;
                             // ReSharper disable InconsistentNaming
-                            var w1y = (NumericToken)array.Data[j];
-                            var v1x = (NumericToken)array.Data[++j];
-                            var v1y = (NumericToken)array.Data[++j];
+                            var w1y = DirectObjectFinder.Get<NumericToken>(array.Data[j], pdfScanner);
+                            var v1x = DirectObjectFinder.Get<NumericToken>(array.Data[++j], pdfScanner);
+                            var v1y = DirectObjectFinder.Get<NumericToken>(array.Data[++j], pdfScanner);
 
                             verticalDisplacements[cid] = w1y.Double;
 
@@ -236,9 +235,9 @@ private static VerticalWritingMetrics ReadVerticalDisplacements(DictionaryToken
                     {
                         var first = c.Int;
                         var last = ((NumericToken)next).Int;
-                        var w1y = (NumericToken)w2.Data[++i];
-                        var v1x = (NumericToken)w2.Data[++i];
-                        var v1y = (NumericToken)w2.Data[++i];
+                        var w1y = DirectObjectFinder.Get<NumericToken>(w2.Data[++i], pdfScanner);
+                        var v1x = DirectObjectFinder.Get<NumericToken>(w2.Data[++i], pdfScanner);
+                        var v1y = DirectObjectFinder.Get<NumericToken>(w2.Data[++i], pdfScanner);
                         // ReSharper restore InconsistentNaming
 
                         for (var cid = first; cid <= last; cid++)
@@ -250,7 +249,7 @@ private static VerticalWritingMetrics ReadVerticalDisplacements(DictionaryToken
                     }
                 }
             }
-
+            
             return new VerticalWritingMetrics(dw2, verticalDisplacements, positionVectors);
         }