pdf-language-detector 0.0.11

A python script to iterate over a list of PDF in a directory and try to guess their language with Tesseract OCR.

PLD (PDF Language Detector)

PLD is a Python program that analyzes PDF files, extracts images, processes them using Optical Character Recognition (OCR), and detects the dominant language of the text. It provides language detection information in JSON format and calculates the average confidence coefficient for each language.

Requirements

• Python 3.8 or above
• Tesseract OCR
• pdftoppm

Installation

Install Tesseract OCR and pdftoppm using your package manager. For example, on Ubuntu:

sudo apt install tesseract-ocr tesseract-ocr-all poppler-utils

From PyPi

Install with pip:

python3 -m pip install --user pdf-language-detector

Then run directly from your terminal:

pld --help

From the sources

Clone the PLD repository:

git clone [e-mail]:github.com/icij/pld.git

Install the required Python packages with poetry:

poetry install

Then run inside a virtual env managed by poetry:

poetry run pld --help

From Docker

Install with Docker:

docker pull icij/pld

Then run inside a container:

docker run -it icij/pld pld --help

Usage

Detect

This command process PDF files and detect the dominant language.

pld detect --help

    --language A list of ISO3 language codes to detect.
    --input-dir: Path to the input directory containing PDF files. Default is the current directory.
    --output-dir (optional): Path to the output directory. Default is 'out' directory in the current directory.
    --max-pages (optional): Maximum number of pages to process per PDF file. Default is 5.
    --resume (optional): Skip PDF files already analyzed.
    --skip-images (optional): Skip the extraction of PDF files a images.
    --skip-ocr (optional): Skip the OCR of images from PDF files.
    --parallel (optional): Number of threads to run in parallel.
    --relative-to (optional): Path to the directory relative to which build the output dir path.

Report

This command print a report from the previously detected language (using the same output dir).

pld report --help

    --output-dir: Path to the output directory. Default is 'out' directory in the current directory.

Test

You can run the test suite (propulsed by pytest) with this command:

make test

Examples

Process PDF files in the current directory, detect English and Spanish languages, and save the results in the 'results' directory:

pld --language eng --language spa --input-dir documents --output-dir results

Process PDF files in the 'documents' directory, detect French and Greek languages, and limit the processing to 3 pages per file:

pld --language fra --language ell --input-dir documents --max-pages 3

License

This project is licensed under the MIT License.