[libexif] Added character set conversion to UserComment

closes xbmc#12628

Author: fetzerch

lib/libexif/ExifParse.cpp

@@ -451,11 +451,32 @@ void CExifParse::ProcessDir(const unsigned char* const DirStart,
 
       case TAG_USERCOMMENT:
       {
-        const int EXIF_COMMENT_HDR_LENGTH = 8;        // All comment tags have 8 bytes of header info
-        int length = max(ByteCount - EXIF_COMMENT_HDR_LENGTH, 0);
-        length = min(length, MAX_COMMENT);
-        strncpy(m_ExifInfo->Comments, (char *)ValuePtr+EXIF_COMMENT_HDR_LENGTH, length);
-//        FixComment(comment);                          // Ensure comment is printable
+        // The UserComment allows comments without the charset limitations of ImageDescription.
+        // Therefore the UserComment field is prefixed by a CharacterCode field (8 Byte):
+        //  - ASCII:         'ASCII\0\0\0'
+        //  - Unicode:       'UNICODE\0'
+        //  - JIS X208-1990: 'JIS\0\0\0\0\0'
+        //  - Unknown:       '\0\0\0\0\0\0\0\0' (application specific)
+
+        m_ExifInfo->CommentsCharset = EXIF_COMMENT_CHARSET_UNKNOWN;
+
+        const int EXIF_COMMENT_CHARSET_LENGTH = 8;
+        if (ByteCount >= EXIF_COMMENT_CHARSET_LENGTH)
+        {
+          // As some implementations use spaces instead of \0 for the padding,
+          // we're not so strict and check only the prefix.
+          if (memcmp(ValuePtr, "ASCII", 5) == 0)
+            m_ExifInfo->CommentsCharset = EXIF_COMMENT_CHARSET_ASCII;
+          else if (memcmp(ValuePtr, "UNICODE", 7) == 0)
+            m_ExifInfo->CommentsCharset = EXIF_COMMENT_CHARSET_UNICODE;
+          else if (memcmp(ValuePtr, "JIS", 3) == 0)
+            m_ExifInfo->CommentsCharset = EXIF_COMMENT_CHARSET_JIS;
+
+          int length = ByteCount - EXIF_COMMENT_CHARSET_LENGTH;
+          length = min(length, MAX_COMMENT);
+          memcpy(m_ExifInfo->Comments, ValuePtr + EXIF_COMMENT_CHARSET_LENGTH, length);
+//          FixComment(comment);                          // Ensure comment is printable
+        }
       }
       break;
 

lib/libexif/libexif.h

@@ -75,6 +75,12 @@ typedef struct {
   char ImageType[MAX_IPTC_STRING];
 } IPTCInfo_t;
 
+#define EXIF_COMMENT_CHARSET_CONVERTED -1 // Comments contains converted data
+#define EXIF_COMMENT_CHARSET_UNKNOWN    0 // Exif: Unknown
+#define EXIF_COMMENT_CHARSET_ASCII      2 // Exif: Ascii
+#define EXIF_COMMENT_CHARSET_UNICODE    3 // Exif: Unicode (UTF-16)
+#define EXIF_COMMENT_CHARSET_JIS        4 // Exif: JIS X208-1990
+
 #define MAX_COMMENT 2000
 #define MAX_DATE_COPIES 10
 
@@ -101,6 +107,7 @@ typedef struct {
     int   ExposureMode;
     int   ISOequivalent;
     int   LightSource;
+    int   CommentsCharset;             // EXIF_COMMENT_CHARSET_*
     char  Comments[MAX_COMMENT];
     char  Description[MAX_COMMENT];
 

xbmc/pictures/PictureInfoTag.cpp

@@ -22,6 +22,7 @@
 #include "XBDateTime.h"
 #include "Util.h"
 #include "utils/Variant.h"
+#include "utils/CharsetConverter.h"
 
 using namespace std;
 
@@ -64,7 +65,7 @@ void CPictureInfoTag::Archive(CArchive& ar)
     ar << CStdString(m_exifInfo.CameraMake);
     ar << CStdString(m_exifInfo.CameraModel);
     ar << m_exifInfo.CCDWidth;
-    ar << CStdString(m_exifInfo.Comments);
+    ar << GetInfo(SLIDE_EXIF_COMMENT); // Store and restore the comment charset converted
     ar << CStdString(m_exifInfo.Description);
     ar << CStdString(m_exifInfo.DateTime);
     for (int i = 0; i < 10; i++)
@@ -128,6 +129,7 @@ void CPictureInfoTag::Archive(CArchive& ar)
     GetStringFromArchive(ar, m_exifInfo.CameraModel, sizeof(m_exifInfo.CameraModel));
     ar >> m_exifInfo.CCDWidth;
     GetStringFromArchive(ar, m_exifInfo.Comments, sizeof(m_exifInfo.Comments));
+    m_exifInfo.CommentsCharset = EXIF_COMMENT_CHARSET_CONVERTED; // Store and restore the comment charset converted
     GetStringFromArchive(ar, m_exifInfo.Description, sizeof(m_exifInfo.Description));
     GetStringFromArchive(ar, m_exifInfo.DateTime, sizeof(m_exifInfo.DateTime));
     for (int i = 0; i < 10; i++)
@@ -191,7 +193,7 @@ void CPictureInfoTag::Serialize(CVariant& value)
   value["cameramake"] = CStdString(m_exifInfo.CameraMake);
   value["cameramodel"] = CStdString(m_exifInfo.CameraModel);
   value["ccdwidth"] = m_exifInfo.CCDWidth;
-  value["comments"] = CStdString(m_exifInfo.Comments);
+  value["comments"] = GetInfo(SLIDE_EXIF_COMMENT); // Charset conversion
   value["description"] = CStdString(m_exifInfo.Description);
   value["datetime"] = CStdString(m_exifInfo.DateTime);
   for (int i = 0; i < 10; i++)
@@ -302,7 +304,20 @@ const CStdString CPictureInfoTag::GetInfo(int info) const
     break;
   case SLIDE_COMMENT:
   case SLIDE_EXIF_COMMENT:
-    value = m_exifInfo.Comments;
+    // The charset used for the UserComment is stored in CommentsCharset:
+    // Ascii, Unicode (UCS2), JIS (X208-1990), Unknown (application specific)
+    if (m_exifInfo.CommentsCharset == EXIF_COMMENT_CHARSET_UNICODE)
+    {
+      g_charsetConverter.ucs2ToUTF8(CStdString16((uint16_t*)m_exifInfo.Comments), value);
+    }
+    else
+    {
+      // Ascii doesn't need to be converted (EXIF_COMMENT_CHARSET_ASCII)
+      // Archived data is already converted (EXIF_COMMENT_CHARSET_CONVERTED)
+      // Unknown data can't be converted as it could be any codec (EXIF_COMMENT_CHARSET_UNKNOWN)
+      // JIS data can't be converted as CharsetConverter and iconv lacks support (EXIF_COMMENT_CHARSET_JIS)
+      value = m_exifInfo.Comments;
+    }
     break;
   case SLIDE_EXIF_DATE_TIME:
   case SLIDE_EXIF_DATE: