t-ocr

T-OCR Library

This is a Python library that contains methods for easy and fast OCR parsing with various tools. Used to read scanned PDF documents and images with text.

|

Key Features

• File parsing is done in the AWS cloud, eliminating the need for additional dependencies on your machine.

• Parsing of each page is done in parallel in AWS, resulting in a significantly faster parsing process.

• Access to the T-OCR AWS part is required to use the library.

• Combines the best OCR tools and provides a consistent interface, making it easy to use and switch between tools.

  .. code-block:: python

    # Parsing with Tesseract (Free OCR)
    free_ocr = FreeOCR()
    pages = free_ocr.read_pdf("1.pdf")
    for page in pages:
        print(page.full_text)
  

  .. code-block:: python

    # Parsing with Textract
    textract = Textract()
    pages = textract.read_pdf("1.pdf")
    for page in pages:
        print(page.full_text)
  

• The package contains a large number of powerful methods that will help you save time and read parsed data faster and better.

  .. code-block:: python

    # Get dates etc. from text
    date_matches = get_valid_dates("Text - February 17, 2009, text...")  # 02/17/2009
  
    # Compare the real text with what the OCR read
    does_strings_match_by_fuzzy("Apple", "Appla", percentage=0.8)  # True
    does_text_contains_string_by_fuzzy("Apple", "Appla watch", percentage=0.8) # True
  
    text = "Long article about Appla watch"
    find_string_in_text_by_fuzzy("Apple", text, percentage=0.8)  # 19
  

• Caching OCR results locally in t_ocr_cache folder is supported. Use this during development to avoid blocking parsing time.

  .. code-block:: python

    pdf_file = "file.pdf"
    # First run
    free_ocr.read_pdf("1.pdf", cache_data=True) # Takes 15 seconds
    # Every next run
    free_ocr.read_pdf("1.pdf", cache_data=True) # Takes 0.01 seconds
  
    # If the state of the file or arguments changes, it will save the cache for this call too
    free_ocr.read_pdf("1.pdf", dpi=500, cache_data=True) # Takes 15 seconds
    free_ocr.read_pdf("1.pdf", dpi=500, cache_data=True) # Takes 0.01 seconds
  

|

Supported OCR Tools

1. Tesseract (Free OCR)

- 🆓 Free
- ✅ Read text
- ❌ Read tables, vertical field values
- ⬛ Simple form design preferred

2. AWS Textract

- 💸 Paid
- ✅ Read text
- ✅ Read tables, vertical field values
- 🌃 Flexible for different file designs