TypeError: 'NoneType' object is not iterable still present in latest release

[timminator]

Hi there,

first of all thank you for your work.
I'm having one issue though. When running paddleocr from the command line on an image where there is no text present it errors out. This is an issue that was already more than once reported, for example in #11203. This bug is really annoying because it prevents running paddleocr on a whole folder via command line.
SHWL mentioned that this bug was fixed in the main branch in june, but in the latest release from october this bug is still present. Can you give my any update when this will be fixed with a new release?

I know this can be prevented when running paddleocr via a python script and using a try-catch block but this is not what i would like to do.

[GreatV]

The issue you are encountering, where a TypeError: 'NoneType' object is not iterable occurs when using PaddleOCR on images without text, has been a known problem and has been discussed in multiple previous issues, such as #11203 and #14130. Here's a detailed explanation and an update:

Cause of the Issue

The root cause of the issue is that the result object returned by the OCR process is None when no text is detected in the image. The current implementation does not handle this None case properly, causing the script to fail when it attempts to iterate over a NoneType object.

The problematic code snippet (as highlighted in the issue report) is:

if result is not None:
    lines = []
    for res in result:
        for line in res:
            logger.info(line)
            lines.append(pprint.pformat(line) + "\n")

If result is None, the for res in result loop will raise a TypeError.

Status of the Fix

A fix for this issue was committed to the main branch in June 2024, as noted by SHWL. However, it appears that this fix has not yet been included in the latest official release (October 2024).

The delay in releasing this fix could be due to various reasons:

1. The fix might be part of a larger set of changes that have not yet been finalized for a stable release.
2. The maintainers might be waiting for more testing before merging the fix into a stable release.

Workarounds

While waiting for the official release that includes the fix, you can consider the following workarounds:

1. Modify the Installed Code

If you are comfortable editing the installed library, you can manually apply the fix to handle the None case. Update the relevant section of the code in your local paddleocr.py file to:

if result:
    lines = []
    for res in result:
        for line in res:
            logger.info(line)
            lines.append(pprint.pformat(line) + "\n")
else:
    logger.info("No text detected in the image.")

This ensures that the script will not attempt to iterate over None.

2. Use the Main Branch

If you need the fix immediately, you can install PaddleOCR directly from the main branch on GitHub, which should include the changes:

pip install --upgrade git+https://github.com/PaddlePaddle/PaddleOCR.git

This will replace your current installation with the latest development version.

3. Use a Python Script with Error Handling

As you mentioned, running PaddleOCR via a Python script with a try-except block is another option. Here's an example:

from paddleocr import PaddleOCR

ocr = PaddleOCR()
result = ocr.ocr('path_to_image.jpg')

if result:
    for res in result:
        print(res)
else:
    print("No text detected in the image.")

4. Wait for the Next Release

If you prefer not to modify the code or use the main branch, you can wait for the next release. It is recommended to monitor the releases page for updates.

Next Steps and Recommendations

1. Monitor the Issue Tracker: Follow the ongoing discussion in #14130 to stay updated on the progress of the fix being included in the next stable release.
2. Request an Update: If this issue is critical for your workflow, consider commenting on the issue to request an update from the maintainers regarding the expected timeline for the release.
3. Use Workarounds Temporarily: Implement one of the suggested workarounds to mitigate the issue until the official fix is released.

If you encounter further issues or have additional concerns, feel free to ask!

---

Response generated by feifei-bot | chatgpt-4o-latest

[SWHL]

Could you provide the reproduce code? I will trace it.

[timminator]

The auto generated answer was actually helpful to some extent. The actual fix seems to be the PR #14132, merged on October 31st 2024.
I'm not sure what you meant that got fixed in june but this PR seems to be the actual solution. I updated my paddleocr.py file manually and that solved it, but it would still be nice if this was working out of the box.
This PR was also after the latest release, that explains why its still present in the latest version.
Do you know when you plan to release the next update with this fix included?

[SWHL]

OK , and I am not sure when the next release will be. Welcome to continue to pay attention.

If the problem in the disscussion has been solved, please close it and mark your answer as the best answer. Thanks.