Added support for IPTC data found in an Exif Photoshop IRB tag of a TIFF/RAW image

src/jpgimage.cpp

@@ -48,6 +48,7 @@ EXIV2_RCSID("@(#) $Id$");
 
 // *****************************************************************************
 // class member definitions
+
 namespace Exiv2 {
 
     const byte JpegBase::sos_    = 0xda;
@@ -56,11 +57,8 @@ namespace Exiv2 {
     const byte JpegBase::app1_   = 0xe1;
     const byte JpegBase::app13_  = 0xed;
     const byte JpegBase::com_    = 0xfe;
-    const uint16_t JpegBase::iptc_ = 0x0404;
     const char JpegBase::exifId_[] = "Exif\0\0";
     const char JpegBase::jfifId_[] = "JFIF\0";
-    const char JpegBase::ps3Id_[]  = "Photoshop 3.0\0";
-    const char JpegBase::bimId_[]  = "8BIM";
 
     JpegBase::JpegBase(BasicIo::AutoPtr io, bool create,
                        const byte initData[], long dataSize)
@@ -187,7 +185,7 @@ namespace Exiv2 {
                 if (exifData_.load(rawExif.pData_, sizeExifData)) throw Error(36, "Exif");
                 --search;
             }
-            else if (marker == app13_ && memcmp(buf.pData_ + 2, ps3Id_, 14) == 0) {
+            else if (marker == app13_ && memcmp(buf.pData_ + 2, Photoshop::ps3Id, 14) == 0) {
                 if (size < 16) throw Error(15);
                 // Read the rest of the APP13 segment
                 // needed if bufMinSize!=16: io_->seek(16-bufRead, BasicIo::cur);
@@ -233,7 +231,6 @@ namespace Exiv2 {
         }
     } // JpegBase::readMetadata
 
-
     // Operates on raw data (rather than file streams) to simplify reuse
     int JpegBase::locateIptcData(const byte *pPsData,
                                  long sizePsData,
@@ -241,41 +238,8 @@ namespace Exiv2 {
                                  uint16_t *const sizeHdr,
                                  uint16_t *const sizeIptc) const
     {
-        assert(record);
+        return locate8BimData(pPsData, sizePsData, Photoshop::iptc, record, sizeHdr, sizeIptc);
-        assert(sizeHdr);
+    }
-        assert(sizeIptc);
-        // Used for error checking
-        long position = 0;
-
-        // Data should follow Photoshop format, if not exit
-        while (position <= (sizePsData - 14) &&
-                memcmp(pPsData + position, bimId_, 4)==0) {
-            const byte *hrd = pPsData + position;
-            position += 4;
-            uint16_t type = getUShort(pPsData+ position, bigEndian);
-            position += 2;
-
-            // Pascal string is padded to have an even size (including size byte)
-            byte psSize = pPsData[position] + 1;
-            psSize += (psSize & 1);
-            position += psSize;
-            if (position >= sizePsData) return -2;
-
-            // Data is also padded to be even
-            long dataSize = getULong(pPsData + position, bigEndian);
-            position += 4;
-            if (dataSize > sizePsData - position) return -2;
-
-            if (type == iptc_) {
-                *sizeIptc = static_cast<uint16_t>(dataSize);
-                *sizeHdr = psSize + 10;
-                *record = hrd;
-                return 0;
-            }
-            position += dataSize + (dataSize & 1);
-        }
-        return 3;
-    } // JpegBase::locateIptcData
 
     void JpegBase::writeMetadata()
     {
@@ -341,7 +305,7 @@ namespace Exiv2 {
                 ++search;
                 if (io_->seek(size-bufRead, BasicIo::cur)) throw Error(22);
             }
-            else if (marker == app13_ && memcmp(buf.pData_ + 2, ps3Id_, 14) == 0) {
+            else if (marker == app13_ && memcmp(buf.pData_ + 2, Photoshop::ps3Id, 14) == 0) {
                 if (size < 16) throw Error(22);
                 skipApp13Ps3 = count;
                 ++search;
@@ -439,7 +403,7 @@ namespace Exiv2 {
                     us2Data(tmpBuf + 2,
                             static_cast<uint16_t>(psData.size_-sizeOldData+sizeNewData+16),
                             bigEndian);
-                    memcpy(tmpBuf + 4, ps3Id_, 14);
+                    memcpy(tmpBuf + 4, Photoshop::ps3Id, 14);
                     if (outIo.write(tmpBuf, 18) != 18) throw Error(21);
                     if (outIo.error()) throw Error(21);
 
@@ -450,8 +414,8 @@ namespace Exiv2 {
 
                     // write new iptc record if we have it
                     if (iptcData_.count() > 0) {
-                        memcpy(tmpBuf, bimId_, 4);
+                        memcpy(tmpBuf, Photoshop::bimId, 4);
-                        us2Data(tmpBuf+4, iptc_, bigEndian);
+                        us2Data(tmpBuf+4, Photoshop::iptc, bigEndian);
                         tmpBuf[6] = 0;
                         tmpBuf[7] = 0;
                         ul2Data(tmpBuf + 8, rawIptc.size_, bigEndian);
@@ -623,4 +587,50 @@ namespace Exiv2 {
         return result;
     }
 
+    // Todo: Generalised from JpegBase::locateIptcData without really understanding
+    //       the format (in particular the header). So it remains to be confirmed 
+    //       if this also makes sense for psTag != Photoshop::iptc
+    int locate8BimData(const byte *pPsData,
+                       long sizePsData,
+                       uint16_t psTag,
+                       const byte **record,
+                       uint16_t *const sizeHdr,
+                       uint16_t *const sizeData)
+    {
+        assert(record);
+        assert(sizeHdr);
+        assert(sizeData);
+        // Used for error checking
+        long position = 0;
+
+        // Data should follow Photoshop format, if not exit
+        while (   position <= sizePsData - 14
+               && memcmp(pPsData + position, Photoshop::bimId, 4) == 0) {
+            const byte *hrd = pPsData + position;
+            position += 4;
+            uint16_t type = getUShort(pPsData + position, bigEndian);
+            position += 2;
+
+            // Pascal string is padded to have an even size (including size byte)
+            byte psSize = pPsData[position] + 1;
+            psSize += (psSize & 1);
+            position += psSize;
+            if (position >= sizePsData) return -2;
+
+            // Data is also padded to be even
+            long dataSize = getULong(pPsData + position, bigEndian);
+            position += 4;
+            if (dataSize > sizePsData - position) return -2;
+
+            if (type == psTag) {
+                *sizeData = static_cast<uint16_t>(dataSize);
+                *sizeHdr = psSize + 10;
+                *record = hrd;
+                return 0;
+            }
+            position += dataSize + (dataSize & 1);
+        }
+        return 3;
+    } // locate8BimData
+
 }                                       // namespace Exiv2

src/jpgimage.hpp

@@ -56,6 +56,13 @@ namespace Exiv2 {
         const int exv  = 2;         //!< Exv image type (see class ExvImage)
     }
 
+    //! %Photoshop constants
+    namespace Photoshop {
+        const char     ps3Id[]  = "Photoshop 3.0\0"; //!< %Photoshop marker
+        const char     bimId[]  = "8BIM";            //!< %Photoshop marker
+        const uint16_t iptc     = 0x0404;            //!< %Photoshop IPTC marker
+    }
+
     /*!
       @brief Abstract helper base class to access JPEG images.
      */
@@ -156,9 +163,6 @@ namespace Exiv2 {
         static const byte com_;                 //!< JPEG Comment marker
         static const char exifId_[];            //!< Exif identifier
         static const char jfifId_[];            //!< JFIF identifier
-        static const char ps3Id_[];             //!< Photoshop marker
-        static const char bimId_[];             //!< Photoshop marker
-        static const uint16_t iptc_;              //!< Photoshop Iptc marker
 
     private:
         // DATA
@@ -353,6 +357,29 @@ namespace Exiv2 {
     Image::AutoPtr newExvInstance(BasicIo::AutoPtr io, bool create);
     //! Check if the file iIo is an EXV file
     bool isExvType(BasicIo& iIo, bool advance);
+    /*!
+      @brief Locates the data for a Photoshop tag in a Photoshop formated memory
+          buffer. Operates on raw data to simplify reuse.
+      @param pPsData Pointer to buffer containing entire payload of
+          Photoshop formated APP13 Jpeg segment.
+      @param sizePsData Size in bytes of pPsData.
+      @param psTag Tag number of the block to look for.
+      @param record Output value that is set to the start of the
+          data block within pPsData (may not be null).
+      @param sizeHdr Output value that is set to the size of the header
+          within the data block pointed to by record (may not be null).
+      @param sizeData Output value that is set to the size of the actual
+          data within the data block pointed to by record (may not be null).
+      @return 0 if successful;<BR>
+              3 if no data for psTag was found in pPsData;<BR>
+             -2 if the pPsData buffer does not contain valid data.
+     */
+    int locate8BimData(const byte *pPsData,
+                       long sizePsData,
+                       uint16_t psTag,
+                       const byte **record,
+                       uint16_t *const sizeHdr,
+                       uint16_t *const sizeData);
 
 }                                       // namespace Exiv2
 

src/mrwimage.cpp

@@ -55,7 +55,7 @@ EXIV2_RCSID("@(#) $Id$");
 namespace Exiv2 {
 
     MrwImage::MrwImage(BasicIo::AutoPtr io, bool create)
-        : Image(mdExif), io_(io)
+        : Image(mdExif | mdIptc), io_(io)
     {
         if (create) {
             IoCloser closer(*io_);
@@ -92,12 +92,12 @@ namespace Exiv2 {
 
     void MrwImage::clearIptcData()
     {
-        // not supported
+        iptcData_.clear();
     }
 
-    void MrwImage::setIptcData(const IptcData& /*iptcData*/)
+    void MrwImage::setIptcData(const IptcData& iptcData)
     {
-        // not supported
+        iptcData_ = iptcData;
     }
 
     void MrwImage::clearComment()

src/mrwimage.hpp

@@ -52,8 +52,8 @@ namespace Exiv2 {
     }
 
     /*!
-      @brief Class to access raw MRW images. Only Exif metadata is currently
+      @brief Class to access raw MRW images. Exif metadata is supported
-          supported.
+             directly, IPTC is read from the Exif data, if present.
      */
     class MrwImage : public Image {
         friend bool isMrwType(BasicIo& iIo, bool advance);
@@ -99,15 +99,7 @@ namespace Exiv2 {
         void            writeMetadata();
         void            setExifData(const ExifData& exifData);
         void            clearExifData();
-        /*!
-          @brief Not supported. MRW format does not contain IPTC metadata.
-                 Calling this function will do nothing.
-         */
         void            setIptcData(const IptcData& iptcData);
-        /*!
-          @brief Not supported. MRW format does not contain IPTC metadata.
-                 Calling this function will do nothing.
-         */
         void            clearIptcData();
         /*!
           @brief Not supported. MRW format does not contain a comment.

src/tiffimage.cpp

@@ -54,7 +54,7 @@ EXIV2_RCSID("@(#) $Id$");
 namespace Exiv2 {
 
     TiffImage::TiffImage(BasicIo::AutoPtr io, bool create)
-        : Image(mdExif | mdComment), io_(io)
+        : Image(mdExif | mdIptc | mdComment), io_(io)
     {
         if (create) {
             IoCloser closer(*io_);
@@ -93,12 +93,12 @@ namespace Exiv2 {
 
     void TiffImage::clearIptcData()
     {
-        // not supported
+        iptcData_.clear();
     }
 
-    void TiffImage::setIptcData(const IptcData& /*iptcData*/)
+    void TiffImage::setIptcData(const IptcData& iptcData)
     {
-        // not supported
+        iptcData_ = iptcData;
     }
 
     void TiffImage::clearComment()

src/tiffimage.hpp

@@ -51,8 +51,8 @@ namespace Exiv2 {
     }
 
     /*!
-      @brief Class to access raw TIFF images. Only Exif metadata and a comment
+      @brief Class to access raw TIFF images. Exif metadata and a comment
-          are supported. TIFF format does not contain IPTC metadata.
+             are supported directly, IPTC is read from the Exif data, if present.
      */
     class TiffImage : public Image {
         friend bool isTiffType(BasicIo& iIo, bool advance);
@@ -98,15 +98,7 @@ namespace Exiv2 {
         void            writeMetadata();
         void            setExifData(const ExifData& exifData);
         void            clearExifData();
-        /*!
-          @brief Not supported. TIFF format does not contain IPTC metadata.
-                 Calling this function will do nothing.
-         */
         void            setIptcData(const IptcData& iptcData);
-        /*!
-          @brief Not supported. TIFF format does not contain IPTC metadata.
-                 Calling this function will do nothing.
-         */
         void            clearIptcData();
         void            setComment(const std::string& comment);
         void            clearComment();

src/tiffvisitor.cpp

@@ -40,8 +40,10 @@ EXIV2_RCSID("@(#) $Id$");
 #include "tiffcomposite.hpp"
 #include "makernote2.hpp"
 #include "exif.hpp"
+#include "iptc.hpp"
 #include "value.hpp"
 #include "image.hpp"
+#include "jpgimage.hpp"
 
 // + standard includes
 #include <string>
@@ -59,7 +61,8 @@ namespace Exiv2 {
         { "OLYMPUS",   0x0100, Group::olympmn, &TiffMetadataDecoder::decodeOlympThumb },
         { "*",         0x014a, Group::ifd0,    0 }, // Todo: Controversial, causes problems with Exiftool
         { "*",       Tag::all, Group::sub0_0,  &TiffMetadataDecoder::decodeSubIfd     },
-        { "*",       Tag::all, Group::sub0_1,  &TiffMetadataDecoder::decodeSubIfd     }
+        { "*",       Tag::all, Group::sub0_1,  &TiffMetadataDecoder::decodeSubIfd     },
+        { "*",         0x8649, Group::ifd0,    &TiffMetadataDecoder::decodeIrbIptc    }
     };
 
     bool TiffDecoderInfo::operator==(const TiffDecoderInfo::Key& key) const
@@ -194,6 +197,37 @@ namespace Exiv2 {
         }
     }
 
+    void TiffMetadataDecoder::decodeIrbIptc(const TiffEntryBase* object)
+    {
+        assert(object != 0);
+        assert(pImage_ != 0);
+        if (!object->pData()) return;
+        byte const* record = 0;
+        uint16_t sizeHdr = 0;
+        uint16_t sizeData = 0;
+        if (0 != locate8BimData(object->pData(), 
+                                object->size(),
+                                Photoshop::iptc, 
+                                &record,
+                                &sizeHdr,
+                                &sizeData)) {
+            return;
+        }
+        if (0 != pImage_->iptcData().load(record + sizeHdr, sizeData)) {
+#ifndef SUPPRESS_WARNINGS
+            std::cerr << "Warning: Failed to decode IPTC block found in "
+                      << "Directory " << object->groupName()
+                      << ", entry 0x" << std::setw(4)
+                      << std::setfill('0') << std::hex << object->tag()
+                      << "\n";
+#endif
+            // Todo: ExifKey should have an appropriate c'tor, it should not be 
+            //       necessary to use groupName here
+            ExifKey key(object->tag(), object->groupName());
+            setExifTag(key, object->pValue());
+        }
+    } // TiffMetadataDecoder::decodeIrbIptc
+
     void TiffMetadataDecoder::decodeSubIfd(const TiffEntryBase* object)
     {
         assert(object);
@@ -396,7 +430,7 @@ namespace Exiv2 {
                            TiffRwState::AutoPtr state)
         : pData_(pData),
           size_(size),
-          pLast_(pData + size - 1),
+          pLast_(pData + size),
           pRoot_(pRoot),
           pState_(state.release()),
           pOrigState_(pState_)
@@ -724,8 +758,8 @@ namespace Exiv2 {
             if (object->pData() + object->size() > pLast_) {
 #ifndef SUPPRESS_WARNINGS
                 std::cerr << "Warning: Upper boundary of data for "
-                          << "directory " << object->groupName() << ", "
+                          << "directory " << object->groupName()
-                          << " entry 0x" << std::setw(4)
+                          << ", entry 0x" << std::setw(4)
                           << std::setfill('0') << std::hex << object->tag()
                           << " is out of bounds:\n"
                           << "Offset = 0x" << std::setw(8)

src/tiffvisitor.hpp

@@ -282,6 +282,8 @@ namespace Exiv2 {
         void decodeOlympThumb(const TiffEntryBase* object);
         //! Decode SubIFD contents to Image group if it contains primary image data
         void decodeSubIfd(const TiffEntryBase* object);
+        //! Decode IPTC data from a Photoshop IRB tag
+        void decodeIrbIptc(const TiffEntryBase* object);
         //@}
 
     private: