There is an edge case in the current implementation of read-mime-multipart that causes it to miss boundaries when the data is split in the middle of a boundary and the next boundary is in the buffer after the split. I've updated the tests to randomly chunk the fixtures, which made the error apparent, and fixed the implementation.
There is an edge case in the current implementation of
read-mime-multipart
that causes it to miss boundaries when the data is split in the middle of a boundary and the next boundary is in the buffer after the split. I've updated the tests to randomly chunk the fixtures, which made the error apparent, and fixed the implementation.