Python module "_extra" could not be loaded

[workflowsguy]

Description of the bug

After updating pymupdf from 1.25.5 to 1.26, importing the module now fails with the following exception

Traceback (most recent call last):
  File "/Users/guy/bin/detect sender.py", line 84, in <module>
    import fitz as PyMuPDF
  File "/Users/guy/Library/Python/3.12/lib/python/site-packages/fitz/__init__.py", line 2, in <module>
    from pymupdf import *
  File "/Users/guy/Library/Python/3.12/lib/python/site-packages/pymupdf/__init__.py", line 30, in <module>
    from . import extra
  File "/Users/guy/Library/Python/3.12/lib/python/site-packages/pymupdf/extra.py", line 10, in <module>
    from . import _extra
ImportError: dlopen(/Users/guy/Library/Python/3.12/lib/python/site-packages/pymupdf/_extra.so, 2): Symbol not found: __ZNKSt3__14__fs10filesystem28recursive_directory_iterator13__dereferenceEv
  Referenced from: /Users/guy/Library/Python/3.12/lib/python/site-packages/pymupdf/libmupdf.dylib
  Expected in: /usr/lib/libc++.1.dylib
 in /Users/guy/Library/Python/3.12/lib/python/site-packages/pymupdf/libmupdf.dylib

Reverting pymupdf back to 1.25.5, my script works again.

How to reproduce the bug

import fitz as PyMuPDF

PyMuPDF version

1.26.0

Operating system

macOS

Python version

3.12.10

[julian-smith-artifex-com]

Thanks for the report. It looks like your system has an old version of the C++ libraries.

PyMuPDF-1.26.0 uses MuPDF-1.26.1 which in turn uses a version of Tesseract that uses C++'s relatively new std::filesystem::recursive_directory_iterator.

Could you post the full output of these commands?:

uname -a
clang --version

[workflowsguy]

uname -a:

Darwin Guy-iMac.local 17.7.0 Darwin Kernel Version 17.7.0: Fri Oct 30 13:34:27 PDT 2020; root:xnu-4570.71.82.8~1/RELEASE_X86_64 x86_64 i386 Darwin

clang --version:

Apple LLVM version 10.0.0 (clang-1000.11.45.5) Target: x86_64-apple-darwin17.7.0 Thread model: posix InstalledDir: /Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin

[julian-smith-artifex-com]

Thanks for posting that information.

I'm afraid your system is too old to use the latest PyMuPDF. We build wheels with Github's macos-13, which has Clang/LLVM 14.0.0, which i suspect is necessary for building and running Tesseract.

You might be able to get things to work by building PyMuPDF yourself with PYMUPDF_SETUP_MUPDF_TESSERACT=0, but otherwise the only solution is to upgrade your system.