Coverage increased (+1.5%) to 67.225% when pulling 113d922 on svenbluege:master into b044158 on lsolesen:master.

Coverage increased (+1.5%) to 67.225% when pulling 113d922 on svenbluege:master into b044158 on lsolesen:master.

Coverage increased (+1.5%) to 67.225% when pulling 9547b7d on svenbluege:master into 7dadd2a on lsolesen:master.

Coverage increased (+1.6%) to 67.281% when pulling 392d5d6 on svenbluege:master into 511b159 on lsolesen:master.

I have updated PEL not to fail on a position where it throws an exception. But it's really hard to debug the issue itself.
When somebody finds a solution we should merge the test into PEL.

The issue is, I think, more general as it affects any tag that is represented by a string, For example 'Make', 'ImageDescription', 'Software', 'Artist' all have the last character trimmed. See dump below

You are right. I have had a look at it some weeks ago. Unfortunately the parsing process is really complex and I could not find the problem. Currently I don't have time to debug that issue...

My main question is: is the image broken? Other tools seam to be able to read the information properly. I did some debugging but I ended up with no clue. At least (think) I saw that the length of content is specified one character lower than what is the expected content length.

Some generators could place a null terminator at the end of the string while others don't. Maybe that lenght-1 at the end should first check the last character?
I didn't check that but it sounds reasonable.

I was coming to same conclusion. In this test image there's no null terminator. Patch soon.

I made a PR, but probably to the wrong repo, svenbluege#1

Sorry I am not familiar with github PR process.

The fix seems to be pretty easy, always take full size of the tag value for PelEntryAscii.

Can anybody replicate the proposed change to this repo and see if tests pass?

Dump (partial, only ifd0) with patch applied:

Coverage increased (+1.6%) to 67.281% when pulling c83792e on svenbluege:master into 958d94e on lsolesen:master.

Coverage increased (+1.6%) to 67.281% when pulling 9a91732 on svenbluege:master into c19abee on lsolesen:master.

Thanks a lot for the test - I'll merge #77 is solved.

Coverage increased (+1.6%) to 67.281% when pulling 9a91732 on svenbluege:master into c19abee on lsolesen:master.