Try and decode Exif.Photo.UserComment according to its charset if specified.

4 files changed, 51 insertions, 7 deletions

diff --git a/src/exiv2wrapper.cpp b/src/exiv2wrapper.cpp
index 4004f20..293bb0b 100644
--- a/src/exiv2wrapper.cpp
+++ b/src/exiv2wrapper.cpp
@@ -230,7 +230,7 @@ const ExifTag Image::getExifTag(std::string key)
         throw Exiv2::Error(KEY_NOT_FOUND, key);
     }
 
-    return ExifTag(key, &(*_exifData)[key], _exifData);
+    return ExifTag(key, &(*_exifData)[key], _exifData, _image->byteOrder());
 }
 
 void Image::deleteExifTag(std::string key)
@@ -450,6 +450,12 @@ std::string Image::getDataBuffer() const
     return buffer;
 }
 
+Exiv2::ByteOrder Image::getByteOrder() const
+{
+    CHECK_METADATA_READ
+    return _image->byteOrder();
+}
+
 Exiv2::ExifThumb* Image::_getExifThumbnail()
 {
     CHECK_METADATA_READ
@@ -510,7 +516,10 @@ void Image::setExifThumbnailFromData(const std::string& data)
 }
 
 
-ExifTag::ExifTag(const std::string& key, Exiv2::Exifdatum* datum, Exiv2::ExifData* data): _key(key)
+ExifTag::ExifTag(const std::string& key,
+                 Exiv2::Exifdatum* datum, Exiv2::ExifData* data,
+                 Exiv2::ByteOrder byteOrder):
+    _key(key), _byteOrder(byteOrder)
 {
     if (datum != 0 && data != 0)
     {
@@ -580,6 +589,8 @@ void ExifTag::setParentImage(Image& image)
     delete _datum;
     _datum = &(*_data)[_key.key()];
     _datum->setValue(value);
+
+    _byteOrder = image.getByteOrder();
 }
 
 const std::string ExifTag::getKey()
@@ -627,6 +638,11 @@ const std::string ExifTag::getHumanValue()
     return _datum->print(_data);
 }
 
+int ExifTag::getByteOrder()
+{
+    return _byteOrder;
+}
+
 
 IptcTag::IptcTag(const std::string& key, Exiv2::IptcData* data): _key(key)
 {
diff --git a/src/exiv2wrapper.hpp b/src/exiv2wrapper.hpp
index 5058d56..93b7823 100644
--- a/src/exiv2wrapper.hpp
+++ b/src/exiv2wrapper.hpp
@@ -42,7 +42,9 @@ class ExifTag
 {
 public:
     // Constructor
-    ExifTag(const std::string& key, Exiv2::Exifdatum* datum=0, Exiv2::ExifData* data=0);
+    ExifTag(const std::string& key,
+            Exiv2::Exifdatum* datum=0, Exiv2::ExifData* data=0,
+            Exiv2::ByteOrder byteOrder=Exiv2::invalidByteOrder);
 
     ~ExifTag();
 
@@ -58,6 +60,7 @@ public:
     const std::string getSectionDescription();
     const std::string getRawValue();
     const std::string getHumanValue();
+    int getByteOrder();
 
 private:
     Exiv2::ExifKey _key;
@@ -69,6 +72,7 @@ private:
     std::string _description;
     std::string _sectionName;
     std::string _sectionDescription;
+    int _byteOrder;
 };
 
 
@@ -250,6 +254,8 @@ public:
     Exiv2::IptcData* getIptcData() { return _iptcData; };
     Exiv2::XmpData* getXmpData() { return _xmpData; };
 
+    Exiv2::ByteOrder getByteOrder() const;
+
 private:
     std::string _filename;
     Exiv2::byte* _data;
diff --git a/src/exiv2wrapper_python.cpp b/src/exiv2wrapper_python.cpp
index 5b793f7..a718afe 100644
--- a/src/exiv2wrapper_python.cpp
+++ b/src/exiv2wrapper_python.cpp
@@ -64,6 +64,7 @@ BOOST_PYTHON_MODULE(libexiv2python)
         .def("_getSectionDescription", &ExifTag::getSectionDescription)
         .def("_getRawValue", &ExifTag::getRawValue)
         .def("_getHumanValue", &ExifTag::getHumanValue)
+        .def("_getByteOrder", &ExifTag::getByteOrder)
     ;
 
     class_<IptcTag>("_IptcTag", init<std::string>())
diff --git a/src/pyexiv2/exif.py b/src/pyexiv2/exif.py
index 87ce581..f3e97ac 100644
--- a/src/pyexiv2/exif.py
+++ b/src/pyexiv2/exif.py
@@ -36,6 +36,7 @@ from pyexiv2.utils import Rational, Fraction, \
 
 import time
 import datetime
+import sys
 
 
 class ExifValueError(ValueError):
@@ -264,10 +265,30 @@ class ExifTag(ListenerInterface):
             return value
 
         elif self.type == 'Comment':
-            # There is currently no charset conversion.
-            # TODO: guess the encoding and decode accordingly into unicode
-            # where relevant.
-            return value
+            if value.startswith('charset='):
+                charset, val = value.split(' ', 1)
+                charset = charset.split('=')[1].strip('"')
+                encoding = sys.getdefaultencoding()
+                if charset == 'Ascii':
+                    encoding = 'ascii'
+                elif charset == 'Jis':
+                    encoding = 'shift_jis'
+                elif charset == 'Unicode':
+                    byte_order = self._tag._getByteOrder()
+                    if byte_order == 1:
+                        # little endian (II)
+                        encoding = 'utf-16le'
+                    elif byte_order == 2:
+                        # big endian (MM)
+                        encoding = 'utf-16be'
+                elif charset == 'Undefined':
+                    pass
+                elif charset == 'InvalidCharsetId':
+                    pass
+                return val.decode(encoding, 'replace')
+            else:
+                # No encoding defined.
+                return value
 
         elif self.type in ('Short', 'SShort'):
             try: