Cannot read/write image in a folder with cyrillic name, on windows

[Fedik]

The library cannot read/write the image from folder that contain cyrillic symbols, on windows, however on Linux it works very well.

Example little thing:

# read
image_path = "C:\Users\TheUser\Desktop\test\тест призначення\electromecánico\good bike.JPG"
img = pyexiv2.Image(image_path)
metadata = img.read_xmp()
print(metadata)

# write
keywords = ["foo", "bar"]
new_meta = {'Xmp.dc.subject': keywords}
img.modify_xmp(new_meta)

Produce error Failed to open the data source: No such file or directory (errno = 2)

Traceback:

Traceback (most recent call last):
  File "C:\py-dev\test-meta\edit-xmp.py", line 15
    img = pyexiv2.Image(image_path)
  File "C:\py-dev\test-meta\env\lib\site-packages\pyexiv2\core.py", line 16, in __init__
    self.img = api.open_image(filename.encode(encoding))
RuntimeError: C:\Users\TheUser\Desktop\test\тест призначення\electromecánico\good bike.JPG: Failed to open the data source: No such file or directory (errno = 2)

api.open_image() does not see the file in some reason. If use path without cyrillic symbols then all works.

[LeoHsiao1]

This error, No such file or directory , is usually reported for the following reasons:

• The image path does not exist.
• The image path could not be decoded correctly.

What you're dealing with is obviously the second case.

Most of the functions provided by pyexiv2 include a default parameter: encoding='utf-8' If you encounter an error because the data contains non-ascii code characters, please try to change the encoding parameter. For example:

img = pyexiv2.Image(image_path, encoding='utf-8')
img = pyexiv2.Image(image_path, encoding='ISO 8859-5')

In addition, the default encoding for your Windows system may not be utf-8, so you cannot decode in utf-8.

[Fedik]

What you're dealing with is obviously the second case.

yeah, I suspect it, but not get why :wink: because

file = open(image_path, "r")

works very well, without any encoding (it detected internally).

Is there a way to give the file object to pyexiv2.Image instance? this could be as work around.

In addition, the default encoding for your Windows system may not be utf-8, so you cannot decode in utf-8.

That also an idea, I will try locale.getpreferredencoding() , will see what it will do, and will write you back.

[LeoHsiao1]

In fact, pyexiv2 calls the C++ API of exiv2. Therefore, only one string representing the path of the file can be passed. Only with the correct encoding can the string be successfully identified.

[Fedik]

okay, I have tried locale.getpreferredencoding() but it does not help, it say default is cp1252, but if I use it then encoding fail with EncoderError

In fact, pyexiv2 calls the C++ API of exiv2

I understood,

However, how does native open() work with same exact path without path encoding. I even tried to use the path from it:

file = open(path, 'r');
img = pyexiv2.Image(file.name);

but still no luck

[LeoHsiao1]

As far as I know, executing chcp in Windows CMD will display a number, which represents some kind of encoding format. As for Python's built-in function open(), it might guess the encoding format automatically. The Python3 interpreter supports Unicode characters very well, but C++ can only recognize ASCII code characters after encoding.

[LeoHsiao1]

The default encoding format in Linux system is utf-8, so you can use utf-8 to decode file paths.

leo@Leo:~$ echo $LANG
C.UTF-8

[Fedik]

hm, with os.path.exists(path.encode('utf-8')) it say True.

I tried default encoding cp1252 but it not able to encode cyrillic, and if use cp1251 then it able to encode cyrillic but pyexiv2.Image stlll fail with exception.

well, okay

[Fedik]

@LeoHsiao1 one more question

Does it possible to implement this method static Image::AutoPtr open (const byte *data, long size) ? To load the image binary data.

So I can do:

the_file = open("in-file", "rb") 
data = the_file.read()

img = pyexiv2.Image(binary_data = data)

I think that would be nice feature.

[LeoHsiao1]

It looks like it can be done, and I'll try to do it in the next week.

[LeoHsiao1]

I've implemented this method and will release a new version in a few days.

[LeoHsiao1]

Finally, I released v2.3.0. Add class ImageData. It is used to open an image from bytes data. I ran into some bugs when transferring bytes data, which caused me to take a week longer than I expected.

[Fedik]

@LeoHsiao1 thanks! works very well, I just tested

[kolt54321]

Hi @LeoHsiao1 ! I'm having a bit of trouble following how to use the new class to solve cyrillic/other non-ascii names. The code right now looks like this:

from pyexiv2 import Image as TaggedImage
xmp_file_obj = TaggedImage(final_image_filename) # Image exists but with non-ascii letters
xmp_file_obj.modify_xmp(info.metadata) # XMP data to be embed

Is it as simple as changing the first line to:

from pyexiv2 import ImageData as TaggedImage

Unfortunately it still doesn't work for me...

[github-actions[bot]]

This issue has been automatically closed because there has been no activity for a month.