Stripped "\" from Windows Tessdata Path

[tallsmallone]

When trying to run pytesseract from a windows 10 environment, I found that the config was stripping the \ in my tessdata path.

I found that the solution to fix this was in pytesseract.py: line 251:

if config:
cmd_args += shlex.split(config, posix=False)

To confirm it was the issue, here was the config before this section of code:
--tessdata-dir E:\Projects\fhm-test-automation\tessdata --oem 3 --psm 3

And after the shlex.split(), the cmd_args was now:
['tesseract', 'C:\\Users\\nickr\\AppData\\Local\\Temp\\tess_ai78chdv.PNG', 'C:\\Users\\nickr\\AppData\\Local\\Temp\\tess_ai78chdv', '-l', 'eng', '--tessdata-dir', 'E:Projectsfhm-test-automationtessdata', '--oem', '3', '--psm', '3']

Please note that I truncated off the extra, unneeded information.

As you can see, the --tessdata-dir was altered, thus giving the dreaded TESSDATA_PREFIX error.

I would make a PR myself, but I do not know enough about the system to know the consequences on my actions.

[bozhodimitrov]

Hi @tallsmallone
Thank you for reporting the issue.
This seems like a platform specific bug.
For reference, this is how numpy does the parsing: https://github.com/numpy/numpy/blob/main/numpy/distutils/_shell_utils.py
Maybe we can use some built-in method for solving this.