import pip

def install(package):
    if hasattr(pip, 'main'):
        pip.main(['install', package])
    else:
        pip._internal.main(['install', package])

install('git+https://github.com/huggingface/transformers.git')
install('datasets sentencepiece')
install('PyPDF2')
install('pdfminer.six')
install('pdfplumber')
install('poppler-utils')
install('tesseract-ocr')
install('libtesseract-dev')