Convert pdf file to jpeg.

pdf_to_jpg.py

"""
Convert PDF images to jpg files

Requirements:
    * Install imagemagick http://www.imagemagick.org/download/binaries/
    * Install ghostscript https://www.ghostscript.com/download.html
    * pip install wand
"""
import os
from wand.image import Image


__all__ = ['make_new_filename', 'input_files', 'convert']


def make_new_filename(filename, ext):
    """Make a new filename.

    Args:
        filename (str): Original filename
        ext (str): New filename extension.
    """
    base_name = os.path.splitext(filename)[0]
    new_filename = '{}{}'.format(base_name, ext)
    i = 0
    while os.path.exists(new_filename):
        new_filename = '{} [{}]{}'.format(base_name, i, ext)  # 'filename [1].jpeg'
        i += 1
    return new_filename


def input_files(*files, ext='.jpeg', modify_ext=True):
    """Ask the user's input for files and a new extension.

    Args:
        files (list)[None]: List of string filenames.
        ext (str)['.jpeg']: New extension for the files.
        modify_ext (bool)[True]: If True the user can change the extension through the input.

    """
    if modify_ext:
        print('Enter a filenames (blank line to stop, "ext = .png" for new extension):')
    else:
        print('Enter a filenames (blank line to stop:')

    files = list(files)

    while True:
        value = input('')
        if value.strip() == '':
            break
        elif modify_ext and value.lower().replace(' ', '').startswith('ext='):
            # Only get the extension with no quotes
            ext = value.split('=', 1)[-1].strip(' "\'')
        else:
            # Remove quotes from filename
            files.append(value.strip('"\''))

    return files, ext


def convert(*files, ext='.jpeg', new_filename_creator=None):
    """Convert the images in the given pdf files to a new image file with the given extension.

    Args:
        files (tuple/str): List of filenames to convert.
        ext (str)[.jpeg]: Extension to convert the images to.
        new_filename_creator (function/Callable)[make_new_filename]: Function that takes in an
            old filename and extension and returns a new filename to save to.

    Returns:
        new_images (list/str): List of new image filenames that were saved.
    """
    if new_filename_creator is None:
        new_filename_creator = make_new_filename

    new_images = []
    add = new_images.append

    for file in files:
        with(Image(filename=file, resolution=120)) as source:
            for image in source.sequence:
                new_filename = make_new_filename(file, ext)
                Image(image).save(filename=new_filename)
                add(new_filename)

    return new_images


if __name__ == '__main__':
    import argparse

    P = argparse.ArgumentParser('Extract PDF images')
    P.add_argument('files', nargs='*', type=str, help='Files to convert. Leave empty to enter manually.')
    P.add_argument('-e', '--ext', type=str, default='.jpeg', help='Extension to convert to (".jpeg")')

    ARGS = P.parse_args()
    FILES = list(ARGS.files)
    EXT = ARGS.ext

    if len(FILES) == 0:
        FILES, EXT = input_files(ext=EXT, modify_ext=True)

    # Run the command
    convert(*FILES, ext=EXT)