OCR not extracting all text

[robkooper]

When processing the following file, only the following text is extracted:

ocr_text: File File Uploaded Uploaded Uploaded kB f9d7460ae8 copy Jul 13 2021 Benjamin Galewsky PROCESSED

[ddey2]

Hi @robkooper, I am looking into the issue now. I have found out some reasons why it's not able to properly extract the info.

1. Tesseract is able to get the texts from the image. For this one, it considers the text in 2 columns and retrieves it column by column. Hence it's a bit jumbled up. I am investigating ways to fix that.

2. clean_word cleans up each word from extraction. There is a condition on line no 46

if cw.isalnum() and len(cw) >= 2:

This above condition gets rid of words like t33_9.png, 967.3, 16:55:57 , image/png etc. Is there any reason for this check? What should be the expected behavior here?

[robkooper]

Lets add 2 parameters:

1. sets the psm option, 3 == multi-column, 6 == single column (default)
2. instead of splitting text in words, keep lines, but parameter to find minimum line length (default 10 characters)