Merge pull request #652 from mvantzet/PreventCorruptionWhenRemovingText-538

Prevent PDF corruption when removing text (#538)

src/UglyToad.PdfPig.Tests/Writer/PdfDocumentBuilderTests.cs

@@ -1279,6 +1279,7 @@
         public int Tokens { get; private set; }
         public int Objects { get; private set; }
         public bool WroteCrossReferenceTable { get; private set; }
+        public bool WritingPageContents { get; set; }
 
         public void WriteToken(IToken token, Stream outputStream)
         {

src/UglyToad.PdfPig/Writer/IPdfStreamWriter.cs

@@ -1,9 +1,7 @@
 namespace UglyToad.PdfPig.Writer
 {
     using System;
-    using System.Collections.Generic;
     using System.IO;
-    using System.Text;
     using Tokens;
 
     internal interface IPdfStreamWriter : IDisposable
@@ -20,6 +18,11 @@
         /// </summary>
         Stream Stream { get; }
 
+        /// <summary>
+        /// Hints that the stream writer is used for writing page contents.
+        /// </summary>
+        bool WritingPageContents { get; set; }
+
         /// <summary>
         /// Writes a single token to the stream.
         /// </summary>

src/UglyToad.PdfPig/Writer/ITokenWriter.cs

@@ -37,5 +37,10 @@
             IReadOnlyDictionary<IndirectReference, long> objectOffsets,
             IndirectReference catalogToken, Stream outputStream,
             IndirectReference? documentInformationReference);
+
+        /// <summary>
+        /// Hints to the token writer that we are currently writing page contents.
+        /// </summary>
+        bool WritingPageContents { get; set; }
     }
 }

src/UglyToad.PdfPig/Writer/NoTextTokenWriter.cs

@@ -18,6 +18,11 @@ namespace UglyToad.PdfPig.Writer
     /// </summary>
     internal class NoTextTokenWriter : TokenWriter
     {
+        /// <summary>
+        /// Set this value prior to processing page to get the right page number in log messages
+        /// </summary>
+        internal int Page { get; set; }
+
         /// <summary>
         /// Write stream without <see cref="ShowText"/> or <see cref="ShowTextsWithPositioning"/> operations
         /// </summary>
@@ -25,10 +30,16 @@ namespace UglyToad.PdfPig.Writer
         /// <param name="outputStream"></param>
         protected override void WriteStream(StreamToken streamToken, Stream outputStream)
         {
-            if (!TryGetStreamWithoutText(streamToken, out var outputStreamToken))
+            StreamToken outputStreamToken;
+            if (!WritingPageContents && !IsFormStream(streamToken))
             {
                 outputStreamToken = streamToken;
             }
+            else if (!TryGetStreamWithoutText(streamToken, out outputStreamToken))
+            {
+                outputStreamToken = streamToken;
+            }
+
             WriteDictionary(outputStreamToken.StreamDictionary, outputStream);
             WriteLineBreak(outputStream);
             outputStream.Write(StreamStart, 0, StreamStart.Length);
@@ -38,6 +49,12 @@ namespace UglyToad.PdfPig.Writer
             outputStream.Write(StreamEnd, 0, StreamEnd.Length);
         }
 
+        private bool IsFormStream(StreamToken streamToken)
+        {
+            return streamToken.StreamDictionary.Data.TryGetValue(NameToken.Subtype.Data, out var value)
+                   && (NameToken)value == NameToken.Form;
+        }
+
         /// <summary>
         /// Try get a stream without <see cref="ShowText"/> or <see cref="ShowTextsWithPositioning"/> operations.
         /// </summary>
@@ -63,7 +80,7 @@ namespace UglyToad.PdfPig.Writer
             IReadOnlyList<IGraphicsStateOperation> operations;
             try
             {
-                operations = pageContentParser.Parse(1, new ByteArrayInputBytes(bytes), new NoOpLog());
+                operations = pageContentParser.Parse(Page, new ByteArrayInputBytes(bytes), new NoOpLog());
             }
             catch (Exception)
             {
@@ -76,7 +93,9 @@ namespace UglyToad.PdfPig.Writer
                 var haveText = false;
                 foreach (var op in operations)
                 {
-                    if (op.Operator == ShowText.Symbol || op.Operator == ShowTextsWithPositioning.Symbol)
+                    if (op.Operator == ShowText.Symbol 
+                        || op.Operator == ShowTextsWithPositioning.Symbol
+                        || op.Operator == MoveToNextLineShowText.Symbol)
                     {
                         haveText = true;
                         continue;
@@ -89,7 +108,22 @@ namespace UglyToad.PdfPig.Writer
                     return false;
                 }
                 outputStreamT.Seek(0, SeekOrigin.Begin);
-                outputStreamToken = DataCompresser.CompressToStream(outputStreamT.ToArray());
+
+                var compressedBytes = DataCompresser.CompressBytes(outputStreamT.ToArray());
+                var outputStreamDictionary = new Dictionary<NameToken, IToken>()
+                {
+                    { NameToken.Length, new NumericToken(compressedBytes.Length) },
+                    { NameToken.Filter, NameToken.FlateDecode }
+                };
+                foreach (var kv in streamToken.StreamDictionary.Data)
+                {
+                    var key = NameToken.Create(kv.Key);
+                    if (!outputStreamDictionary.ContainsKey(key))
+                    {
+                        outputStreamDictionary[key] = kv.Value;
+                    }
+                };
+                outputStreamToken = new StreamToken(new DictionaryToken(outputStreamDictionary), compressedBytes);
                 return true;
             }
         }

src/UglyToad.PdfPig/Writer/PdfDocumentBuilder.cs

@@ -360,6 +360,7 @@ namespace UglyToad.PdfPig.Writer
                 // dedup if on to avoid issues
                 var prev = context.AttemptDeduplication;
                 context.AttemptDeduplication = false;
+                context.WritingPageContents = true;
                 if (contentsToken is ArrayToken array)
                 {
                     foreach (var item in array.Data)
@@ -378,6 +379,7 @@ namespace UglyToad.PdfPig.Writer
                         WriterUtil.CopyToken(context, ir, document.Structure.TokenScanner, refs) as IndirectReferenceToken));
                 }
                 context.AttemptDeduplication = prev;
+                context.WritingPageContents = false;
             }
 
             // manually copy page dict / resources as we need to modify some
@@ -406,7 +408,6 @@ namespace UglyToad.PdfPig.Writer
                 }
             }
 
-
             foreach (var kvp in pageInfo.Page.Data)
             {
                 if (kvp.Key == NameToken.Contents || kvp.Key == NameToken.Parent || kvp.Key == NameToken.Type)

src/UglyToad.PdfPig/Writer/PdfStreamWriter.cs

@@ -25,6 +25,12 @@
 
         public bool AttemptDeduplication { get; set; } = true;
 
+        public bool WritingPageContents
+        {
+            get => TokenWriter.WritingPageContents;
+            set => TokenWriter.WritingPageContents = value;
+        }
+
         internal PdfStreamWriter(
             Stream baseStream,
             bool disposeStream = true,

src/UglyToad.PdfPig/Writer/PdfTextRemover.cs

@@ -83,19 +83,22 @@ namespace UglyToad.PdfPig.Writer
         /// </summary>
         public static void RemoveText(PdfDocument file, Stream output, IReadOnlyList<int> pagesBundle = null)
         {
-            using (var document = new PdfDocumentBuilder(output, false, PdfWriterType.Default, file.Version, tokenWriter: new NoTextTokenWriter()))
+            var tokenWriter = new NoTextTokenWriter();
+            using (var document = new PdfDocumentBuilder(output, false, PdfWriterType.Default, file.Version, tokenWriter: tokenWriter))
             {
                 if (pagesBundle == null)
                 {
                     for (var i = 1; i <= file.NumberOfPages; i++)
                     {
+                        tokenWriter.Page = i;
                         document.AddPage(file, i);
                     }
-                } 
+                }
                 else
                 {
                     foreach (var i in pagesBundle)
                     {
+                        tokenWriter.Page = i;
                         document.AddPage(file, i);
                     }
                 }

src/UglyToad.PdfPig/Writer/TokenWriter.cs

@@ -276,6 +276,12 @@
             outputStream.Write(Eof, 0, Eof.Length);
         }
 
+        /// <summary>
+        /// Indicates that we are writing page contents.
+        /// Can be used by a derived class.
+        /// </summary>
+        public bool WritingPageContents { get; set; }
+
         /// <inheritdoc cref="ITokenWriter.WriteObject" />
         public void WriteObject(long objectNumber, int generation, byte[] data, Stream outputStream)
         {