amenezes / aiopytesseract

A Python asyncio wrapper for Tesseract-OCR.

Hi, I'm Alexandre Menezes

Currently I am a programmer at Supremo Tribunal Federal.

Status

• 👷 Working on STF

Interests

• 🐍 Programming: 🐍 Python/Cython / 💻 C / 🦀 Rust / ⚡ Zig
• 🤖 Machine Learning
• 🍥 Debian OS

aiopytesseract's Issues

got nothing string output

Platform

Linux

Steps to Reproduce

1 install
2 add to code
3 run
4 get nothing

Expected Result

get some text output

Actual Result

when using tesseract

Char Whitelist support

Need/Ideas Statement

It would be nice to have access to tessedit_char_whitelist so that I can whitelist only specific characters.

Configurable TESSERACT_CMD

Need/Ideas Statement

Hello,

It would be really nice if TESSERACT_CMD could be not constant but configurable to be able to use it in Windows too. Thanks.

Exception from one parent

Need/Ideas Statement

Now check exception very hard. Need always take all exception.
May create from one, or two. But it is better from one.

Example:

class TesseractRuntimeError(RuntimeError):
    pass


class TesseractError(Exception):
    """Base exception for tesseract"""

    def __init__(self, message=""):
        self.message = message
    
    def __str__(self):
        return self.message


class PSMInvalidException(TesseractError):    
    def __init__(self, message="PSM Invalid"):
        super().__init__(message)


class OEMInvalidException(TesseractError):
    def __init__(self, message="OEM Invalid"):
        super().__init__(message)


class NoSuchFileException(TesseractError):
    def __init__(self, message="No such file"):
        super().__init__(message)


class LanguageInvalidException(TesseractError):
    def __init__(self, message="Language invalid"):
        super().__init__(message)

It would be great if and 'TesseractRuntimeError' was also inherited from 'TesseractError'. On you decision...

Default dpi

Need/Ideas Statement

Next )
Exists yet problem.

Tesseract by default not need dpi, then it will be tried to select appropriate dpi himself.

# man tesseract
--dpi N
           Specify the resolution N in DPI for the input image(s). A typical value for N is 300. Without this option, the resolution is read from the metadata
           included in the image. If an image does not include that information, Tesseract tries to guess it.

This is not possible in your implementation. The value for 'dpi' always has something value.
May be change it? If need I may create PR )

Absence of parameters

Platform

Linux

Steps to Reproduce

import aiopytesseract

params = await aiopytesseract.tesseract_parameters()
[p for p in params if p.name == 'tessedit_char_blacklist']

Expected Result

[Parameter(name='tessedit_char_blacklist', description='Blacklist of chars not to recogniz', value='-')]

Actual Result

[]

image_to_data psm and output_type

Need/Ideas Statement

Hi,
Thanks for sharing this code!

There is any option to add this two parameters to image_to_data function?

Best regards!