get_textpage_ocr() No OCR support: TESSDATA_PREFIX not set

[yonglee7015]

get_textpage_ocr() not work
Google Colab. I follow instruction as below .

Enabling Integrated OCR Support
If you do not intend to use this feature, skip this step. Otherwise, it is required for both installation paths: from wheels and from sources.

PyMuPDF will already contain all the logic to support OCR functions. But it additionally does need Tesseract’s language support data, so installation of Tesseract-OCR is still required.

The language support folder location must currently [1] be communicated via storing it in the environment variable "TESSDATA_PREFIX".

So for a working OCR functionality, make sure to complete this checklist:

Install Tesseract.

Locate Tesseract’s language support folder. Typically you will find it here:
Windows: C:Program FilesTesseract-OCRtessdata

Unix systems: /usr/share/tesseract-ocr/4.00/tessdata

Set the environment variable TESSDATA_PREFIX
Windows: set TESSDATA_PREFIX=C:Program FilesTesseract-OCRtessdata

Unix systems: export TESSDATA_PREFIX=/usr/share/tesseract-ocr/4.00/tessdata

!apt install tesseract-ocr
set environment TESSDATA_PREFIX='/usr/share/tesseract-ocr/4.00/tessdata'

RuntimeError: No OCR support: TESSDATA_PREFIX not set

[JorjMcKie]

Shouldn't you use the command !export TESSDATA_PREFIX=/usr/share/tesseract-ocr/4.00/tessdata in this environment?

[yonglee7015]

Then what should I do in Google Colab?
I use !export TESSDATA_PREFIX=/usr/share/tesseract-ocr/4.00/tessdata
But it still show this runtime error

[JorjMcKie]

Hm, I looked around in the Colab documentation. It seems that the Python environment is already running when you enter Colab. And that environment variable has to be set before Python is running.
So an idea could be to start a subprocess to execute the export command and then also modify os.environ and set the variable ... IAW something like this:

import subprocess,os
subprocess.call(["export", "TESSDATA_PREFIX=/usr/share/tesseract-ocr/4.00/tessdata"])
os.environ["TESSDATA_PREFIX"] = "/usr/share/tesseract-ocr/4.00/tessdata"

Maybe, considering the !export may be doing exactly the same, calling subprocess is not even necessary ...

[JorjMcKie]

Ah, and in PyMuPDF there is a variable being set in __init__.py that checks the os.environ variable, so the above must happen before import fitz.

[JorjMcKie]

Thinking more about it, you should be fine simply setting fitz.TESSDATA_PREFIX = "/usr/share/tesseract-ocr/4.00/tessdata"

[yonglee7015]

Hm, I looked around in the Colab documentation. It seems that the Python environment is already running when you enter Colab. And that environment variable has to be set before Python is running. So an idea could be to start a subprocess to execute the export command and then also modify os.environ and set the variable ... IAW something like this:

import subprocess,os
subprocess.call(["export", "TESSDATA_PREFIX=/usr/share/tesseract-ocr/4.00/tessdata"])
os.environ["TESSDATA_PREFIX"] = "/usr/share/tesseract-ocr/4.00/tessdata"

Maybe, considering the !export may be doing exactly the same, calling subprocess is not even necessary ...

[Errno 2] No such file or directory: 'export'
subprocess not work

[yonglee7015]

Thinking more about it, you should be fine simply setting fitz.TESSDATA_PREFIX = "/usr/share/tesseract-ocr/4.00/tessdata"

still not work.

RuntimeError: No OCR support: TESSDATA_PREFIX not set

[yonglee7015]

Ah, and in PyMuPDF there is a variable being set in __init__.py that checks the os.environ variable, so the above must happen before import fitz.

yes ,must happen before import fitz.

import os
os.environ["TESSDATA_PREFIX"] = "/usr/share/tesseract-ocr/4.00/tessdata"

import fitz

Then it works. thank you. Runtime error disappeared.

but the OCR result is so terrible.
set full = True, it seems not so bad