From a9acfa31e3a6e7f491a553129d4b321b619687d9 Mon Sep 17 00:00:00 2001
From: Jakub Melka <Jakub.Melka@seznam.cz>
Date: Sun, 20 Sep 2020 11:53:46 +0200
Subject: [PATCH] Bugfix - find inline image stream length

---
 .../sources/pdfpagecontentprocessor.cpp       | 18 +++++-
 PdfForQtLib/sources/pdfstreamfilters.cpp      | 56 +++++++++++++++++++
 PdfForQtLib/sources/pdfstreamfilters.h        | 16 ++++++
 3 files changed, 88 insertions(+), 2 deletions(-)

diff --git a/PdfForQtLib/sources/pdfpagecontentprocessor.cpp b/PdfForQtLib/sources/pdfpagecontentprocessor.cpp
index 4f5f718..9e678a5 100644
--- a/PdfForQtLib/sources/pdfpagecontentprocessor.cpp
+++ b/PdfForQtLib/sources/pdfpagecontentprocessor.cpp
@@ -21,6 +21,7 @@
 #include "pdfimage.h"
 #include "pdfpattern.h"
 #include "pdfexecutionpolicy.h"
+#include "pdfstreamfilters.h"
 
 #include <QPainterPathStroker>
 
@@ -548,8 +549,21 @@ void PDFPageContentProcessor::processContent(const QByteArray& content)
                         }
                         else if (dictionary->hasKey("Filter"))
                         {
-                            // We will use EI operator position to determine stream length
-                            dataLength = operatorEIPosition - startDataPosition;
+                            dataLength = -1;
+
+                            // We will try to use stream filter hint
+                            PDFDocumentDataLoaderDecorator loader(m_document);
+                            QByteArray filterName = loader.readNameFromDictionary(dictionary, "Filter");
+                            if (!filterName.isEmpty())
+                            {
+                                dataLength = PDFStreamFilterStorage::getStreamDataLength(content, filterName, startDataPosition);
+                            }
+
+                            if (dataLength == -1)
+                            {
+                                // We will use EI operator position to determine stream length
+                                dataLength = operatorEIPosition - startDataPosition;
+                            }
                         }
                         else
                         {
diff --git a/PdfForQtLib/sources/pdfstreamfilters.cpp b/PdfForQtLib/sources/pdfstreamfilters.cpp
index 15f4995..20acefe 100644
--- a/PdfForQtLib/sources/pdfstreamfilters.cpp
+++ b/PdfForQtLib/sources/pdfstreamfilters.cpp
@@ -457,6 +457,44 @@ QByteArray PDFFlateDecodeFilter::recompress(const QByteArray& data)
     return result;
 }
 
+PDFInteger PDFFlateDecodeFilter::getStreamDataLength(const QByteArray& data, PDFInteger offset) const
+{
+    if (offset < 0 || offset >= data.size())
+    {
+        return -1;
+    }
+
+    z_stream stream = { };
+    stream.next_in = const_cast<Bytef*>(convertByteArrayToUcharPtr(data) + offset);
+    stream.avail_in = data.size() - offset;
+
+    std::array<Bytef, 1024> outputBuffer = { };
+
+    int error = inflateInit(&stream);
+    if (error != Z_OK)
+    {
+        return -1;
+    }
+
+    do
+    {
+        stream.next_out = outputBuffer.data();
+        stream.avail_out = static_cast<uInt>(outputBuffer.size());
+
+        error = inflate(&stream, Z_NO_FLUSH);
+    } while (error == Z_OK);
+
+    PDFInteger dataLength = stream.total_in;
+    inflateEnd(&stream);
+
+    if (error == Z_STREAM_END)
+    {
+        return dataLength;
+    }
+
+    return -1;
+}
+
 QByteArray PDFFlateDecodeFilter::uncompress(const QByteArray& data)
 {
     QByteArray result;
@@ -676,6 +714,16 @@ QByteArray PDFStreamFilterStorage::getDecodedStream(const PDFStream* stream, con
     return getDecodedStream(stream, [](const PDFObject& object) -> const PDFObject& { return object; }, securityHandler);
 }
 
+PDFInteger PDFStreamFilterStorage::getStreamDataLength(const QByteArray& data, const QByteArray& filterName, PDFInteger offset)
+{
+    if (const PDFStreamFilter* filter = getFilter(filterName))
+    {
+        return filter->getStreamDataLength(data, offset);
+    }
+
+    return -1;
+}
+
 PDFStreamFilterStorage::PDFStreamFilterStorage()
 {
     // Initialize map with the filters
@@ -931,4 +979,12 @@ QByteArray PDFCryptFilter::apply(const QByteArray& data,
     return securityHandler->decryptByFilter(data, cryptFilterName, objectReference);
 }
 
+PDFInteger PDFStreamFilter::getStreamDataLength(const QByteArray& data, PDFInteger offset) const
+{
+    Q_UNUSED(data);
+    Q_UNUSED(offset);
+
+    return -1;
+}
+
 }   // namespace pdf
diff --git a/PdfForQtLib/sources/pdfstreamfilters.h b/PdfForQtLib/sources/pdfstreamfilters.h
index 26d5879..bcbbe2b 100644
--- a/PdfForQtLib/sources/pdfstreamfilters.h
+++ b/PdfForQtLib/sources/pdfstreamfilters.h
@@ -54,6 +54,14 @@ public:
     /// \param securityHandler Security handler for Crypt filters
     static QByteArray getDecodedStream(const PDFStream* stream, const PDFSecurityHandler* securityHandler);
 
+    /// Tries to find stream data length using given filter. Stream will
+    /// start at given \p offset in \p data. If stream length cannot be determined,
+    /// then -1 is returned.
+    /// \param data Buffer data
+    /// \param filterName Filter name
+    /// \param offset Offset to buffer, at which stream data starts
+    static PDFInteger getStreamDataLength(const QByteArray& data, const QByteArray& filterName, PDFInteger offset);
+
     struct StreamFilters
     {
         bool valid = true;
@@ -148,6 +156,12 @@ public:
     {
         return apply(data, [](const PDFObject& object) -> const PDFObject& { return object; }, parameters, securityHandler);
     }
+
+    /// Tries to find stream data length. Stream will start at given \p offset in \p data.
+    /// If stream length cannot be determined, then -1 is returned.
+    /// \param data Buffer data
+    /// \param offset Offset to buffer, at which stream data starts
+    virtual PDFInteger getStreamDataLength(const QByteArray& data, PDFInteger offset) const;
 };
 
 class PDFFORQTLIBSHARED_EXPORT PDFAsciiHexDecodeFilter : public PDFStreamFilter
@@ -197,6 +211,8 @@ public:
                              const PDFObject& parameters,
                              const PDFSecurityHandler* securityHandler) const override;
 
+    virtual PDFInteger getStreamDataLength(const QByteArray& data, PDFInteger offset) const;
+
     /// Recompresses data. So, first, data are decompressed, and then
     /// recompressed again with maximal compress ratio possible.
     /// \param data Compressed data to be recompressed