search

ocropus / hocr-tools

Tools for manipulating and evaluating the hOCR format for representing multi-lingual OCR results by embedding them into HTML.
Other
359 stars 78 forks source link

AttributeError: module 'base64' has no attribute 'decodestring' #169

Open ghost opened 2 years ago

ghost commented 2 years ago

Hello!

I've install hocr-tools on both mac and ubuntu and get the same error. All images are jpg and the hocr files were created by tesseract.

I'm wondering if I have the wrong versions of supporting libraries installed.

Traceback (most recent call last): File "/opt/homebrew/bin/hocr-pdf", line 143, in export_pdf(args.imgdir, 300) File "/opt/homebrew/bin/hocr-pdf", line 51, in export_pdf load_invisible_font() File "/opt/homebrew/bin/hocr-pdf", line 134, in load_invisible_font uncompressed = bytearray(zlib.decompress(base64.decodestring(font))) AttributeError: module 'base64' has no attribute 'decodestring'

Name: hocr-tools Version: 1.1.1 Name: lxml Version: 4.6.4 Name: Pillow Version: 8.4.0 Name: reportlab Version: 3.6.2

kba commented 2 years ago

base64.decodestring has been removed in Python 3.9, it has been deprecated since 3.1 - can you run with a less recent Python version?

Related: https://github.com/ocropus/hocr-tools/issues/158

ghost commented 2 years ago

Thank you! I installed python 3.8.12 with pyenv and now it works perfectly... and so fast!

FriedrichFroebel commented 1 year ago

While using an older Python version might have been the solution, this does not seem to hold true any more, with recent Ubuntu versions shipping Python 3.10 by default and Python 3.11 being available. Adding a corresponding check should be easy enough while retaining compatibility with Python 2 (although this is broken for now: https://github.com/ocropus/hocr-tools/issues/158#issuecomment-1353653532).

stweil commented 1 year ago

There is no need to support any Python version before 3.7 (which is the oldest version which is still officially supported). And you are right, the code should be fixed to work with recent Python versions, too.

FriedrichFroebel commented 1 year ago

This apparently already has been fixed in 2019, but not yet released: https://github.com/ocropus/hocr-tools/commit/d756f75ce8cf1224a78c9eab5db4952be17b9d70