Add previewBuffer type as a workaround for Sony preview images

include/exiv2/types.hpp

@@ -114,6 +114,7 @@ namespace Exiv2 {
         unsignedLongLong   =16, //!< Exif LONG LONG type, 64-bit (8-byte) unsigned integer.
         signedLongLong     =17, //!< Exif LONG LONG type, 64-bit (8-byte) signed integer.
         tiffIfd8           =18, //!< TIFF IFD type, 64-bit (8-byte) unsigned integer.
+        previewBuffer  =0xffff, //!< Used when the metadata contains a reference to a preview image that is not stored with the metadata, but somewhere else in the file. (Sony does this. See https://github.com/Exiv2/exiv2/issues/2001)
         string        =0x10000, //!< IPTC string type.
         date          =0x10001, //!< IPTC date type.
         time          =0x10002, //!< IPTC time type.

include/exiv2/value.hpp

@@ -1187,6 +1187,93 @@ namespace Exiv2 {
 
     }; // class TimeValue
 
+    /*!
+     @brief %Value for preview images
+
+     This class is used to handle Sony's preview images, which are not
+     stored in the metadata but instead refer to a location in the main
+     file. Since the actual preview is not available while we're parsing
+     the metadata, we create this placeholder value instead.
+
+     See: https://github.com/Exiv2/exiv2/issues/2001
+     */
+    class EXIV2API PreviewBufferValue : public Value {
+    public:
+        //! Shortcut for a %PreviewBufferValue auto pointer.
+        typedef std::unique_ptr<PreviewBufferValue> UniquePtr;
+
+        //! @name Creators
+        //@{
+        //! Default constructor.
+        PreviewBufferValue();
+        //! Virtual destructor.
+        ~PreviewBufferValue() override = default;
+        //@}
+
+        //! @name Manipulators
+        //@{
+        /*!
+          @brief Read the value from a character buffer.
+
+          @note The byte order is required by the interface but not used by this
+                method, so just use the default.
+
+          @param buf Pointer to the data buffer to read from
+          @param len Number of bytes in the data buffer
+          @param byteOrder Byte order. Not needed.
+
+          @return 0 if successful<BR>
+                  1 in case of an unsupported time format
+         */
+        int read(const byte* buf, long len, ByteOrder byteOrder = invalidByteOrder) override;
+        /*!
+          @brief Set the value to that of the string buf.
+
+          @param buf String containing the preview.
+
+          @return 0 if successful<BR>
+                  1 in case of an unsupported time format
+         */
+        int read(const std::string& buf) override;
+        //@}
+
+        //! @name Accessors
+        //@{
+        /*!
+          @brief Write value to a character data buffer.
+
+          The user must ensure that the buffer has enough memory. Otherwise
+          the call results in undefined behaviour.
+
+          @note The byte order is required by the interface but not used by this
+                method, so just use the default.
+
+          @param buf Data buffer to write to.
+          @param byteOrder Byte order. Not used.
+          @return Number of characters written.
+        */
+        long copy(byte* buf, ByteOrder byteOrder = invalidByteOrder) const override;
+        long count() const override;
+        long size() const override;
+        std::ostream& write(std::ostream& os) const override;
+        //! Returns number of seconds in the day in UTC.
+        long toLong(long n = 0) const override;
+        //! Returns number of seconds in the day in UTC converted to float.
+        float toFloat(long n = 0) const override;
+        //! Returns number of seconds in the day in UTC converted to Rational.
+        Rational toRational(long n = 0) const override;
+        //@}
+
+    private:
+        //! @name Accessors
+        //@{
+        //! Internal virtual copy constructor.
+        PreviewBufferValue* clone_() const override;
+        //@}
+
+        size_t size_;
+    }; // class PreviewBufferValue
+
     //! Template to determine the TypeId for a type T
     template<typename T> TypeId getType();
 

src/tiffvisitor_int.cpp

@@ -1548,6 +1548,12 @@ namespace Exiv2 {
                 // #1143
                 if ( object->tag() == 0x2001 && std::string(groupName(object->group())) == "Sony1" ) {
                     isize=size;
+
+                    // This a workaround for Sony preview images, which are stored in the body
+                    // of the main file, not in the metadata. Since we don't have access to the
+                    // rest of the file here, we just store a placeholder (of type PreviewBufferValue).
+                    // See https://github.com/Exiv2/exiv2/issues/2001
+                    typeId = previewBuffer;
                 } else {
 #ifndef SUPPRESS_WARNINGS
             EXV_ERROR << "Offset of directory " << groupName(object->group())
@@ -1595,17 +1601,16 @@ namespace Exiv2 {
         }
         Value::UniquePtr v = Value::create(typeId);
         enforce(v.get() != nullptr, kerCorruptedMetadata);
-        if ( !isize ) {
+        if (typeId != previewBuffer) {
             v->read(pData, size, byteOrder());
         } else {
-            // Prevent large memory allocations: https://github.com/Exiv2/exiv2/issues/1881
-            enforce(isize <= 1024 * 1024, kerCorruptedMetadata);
-
             // #1143 Write a "hollow" buffer for the preview image
             //       Sadly: we don't know the exact location of the image in the source (it's near offset)
             //       And neither TiffReader nor TiffEntryBase have access to the BasicIo object being processed
-            std::vector<byte> buffer(isize);
-            v->read(buffer.data() ,isize, byteOrder());
+            //
+            // typeId == previewBuffer, so v is a PreviewBufferValue, which just stores the size without
+            // attempting to read the bytes.
+            v->read(0, isize, byteOrder());
         }
 
         object->setValue(std::move(v));

src/types.cpp

@@ -81,6 +81,7 @@ namespace {
         { Exiv2::tiffFloat,        "Float",       4 },
         { Exiv2::tiffDouble,       "Double",      8 },
         { Exiv2::tiffIfd,          "Ifd",         4 },
+        { Exiv2::previewBuffer,    "PreviewBuffer", 1 },
         { Exiv2::string,           "String",      1 },
         { Exiv2::date,             "Date",        8 },
         { Exiv2::time,             "Time",       11 },

src/value.cpp

@@ -110,6 +110,9 @@ namespace Exiv2 {
         case langAlt:
             value = UniquePtr(new LangAltValue);
             break;
+        case previewBuffer:
+            value = UniquePtr(new PreviewBufferValue);
+            break;
         default:
             value = UniquePtr(new DataValue(typeId));
             break;
@@ -1205,4 +1208,63 @@ namespace Exiv2 {
         return {toLong(n), 1};
     }
 
+    PreviewBufferValue::PreviewBufferValue()
+        : Value(previewBuffer)
+    {
+    }
+
+    int PreviewBufferValue::read(const byte*, long len, ByteOrder /*byteOrder*/)
+    {
+        if (len < 0) {
+          return 1;
+        }
+        size_ = static_cast<size_t>(len);
+        return 0;
+    }
+
+    int PreviewBufferValue::read(const std::string& buf)
+    {
+        size_ = buf.size();
+        return 0;
+    }
+
+    long PreviewBufferValue::copy(byte* /*buf*/, ByteOrder /*byteOrder*/) const
+    {
+        return size();
+    }
+
+    long PreviewBufferValue::count() const
+    {
+        return size();
+    }
+
+    long PreviewBufferValue::size() const
+    {
+        return static_cast<long>(size_);
+    }
+
+    PreviewBufferValue* PreviewBufferValue::clone_() const
+    {
+        return new PreviewBufferValue(*this);
+    }
+
+    std::ostream& PreviewBufferValue::write(std::ostream& os) const
+    {
+        return os;
+    }
+
+    long PreviewBufferValue::toLong(long /*n*/) const
+    {
+        return 0;
+    }
+
+    float PreviewBufferValue::toFloat(long n) const
+    {
+        return static_cast<float>(toLong(n));
+    }
+
+    Rational PreviewBufferValue::toRational(long n) const
+    {
+        return {toLong(n), 1};
+    }
 }  // namespace Exiv2

tests/bugfixes/github/test_issue_1881.py

@@ -14,9 +14,5 @@ class SonyPreviewImageLargeAllocation(metaclass=CaseMeta):
     filename2 = path("$tmp_path/issue_1881_coverage.jpg")
     commands = ["$exiv2 -q -d I rm $filename1", "$exiv2 -q -d I rm $filename2"]
     stdout = ["",""]
-    stderr = [
-"""Exiv2 exception in erase action for file $filename1:
-$kerCorruptedMetadata
-""",
-""]
-    retval = [1,0]
+    stderr = ["",""]
+    retval = [0,0]