Count sum of number of pages in a folder. Can be called with `-r` to check subdirectories also.

pdfpagecount.py

"""count pdf pages"""
from __future__ import print_function
import sys
from getopt import GetoptError, getopt
from PyPDF2 import PdfFileReader
try:
    from os import scandir
except ImportError:
    from scandir import scandir


def get_page_count(paths=None, recursive=False):
    """Returns sum of pages from all found pdf files within directory

Keyword arguments:
paths -- list of path strings (default current path)
recursive -- search subfolders recursively (default False)
    """
    paths = paths if paths else ["."]
    pdffiles = list()
    count = 0

    while paths:
        curr = paths.pop()
        for direntry in scandir(curr):
            if recursive and direntry.is_dir():
                paths.append(direntry.path)
            elif direntry.is_file and direntry.path[-3:] == "pdf":
                pdffiles.append(direntry.path)

    for filename in pdffiles:
        count += get_page_number(filename)

    return count


def get_page_number(filename):
    """Returns number of pages in filename"""
    with open(filename, mode='rb'):
        return PdfFileReader(filename, strict=False).numPages


def main(argv):
    """usage pdfpagecount.py [-r] <path>"""
    recursive = False
    paths = ["."]
    try:
        opts, args = getopt(argv, "hr", ["recursive"])
    except GetoptError:
        print('pdfpagecount.py [-r] <path>')
        sys.exit(2)
    for opt in opts:
        opt = opt.first
        if opt == '-h':
            print('pdfpagecount.py [-r] <path>')
            sys.exit()
        elif opt in ("-r", "--recursive"):
            recursive = True
    if args:
        paths = args
    print(get_page_count(paths, recursive))

if __name__ == "__main__":
    main(sys.argv[1:])

pdfpagecount_test.py

"""pdfpagecount test"""
import unittest
import os
from urllib import urlretrieve
import pdfpagecount


class Test(unittest.TestCase):
    """Test Case"""
    def setUp(self):
        """change to correct testfiles folder

This testCase needs a folder structure like this:

+---pdfpagecount.py
+---pdfpagecount_test.py
|
+---testfiles
|   \---get_page_count
|       +---empty
|       \---full
|           +---file.pdf

file.pdf can have an arbitrary name but must contain only one page
"""
        try:
            os.chdir("./testfiles/get_page_count")
        except os.error:
            os.makedirs("./testfiles/get_page_count")
            os.makedirs("./testfiles/get_page_count/full")
            os.makedirs("./testfiles/get_page_count/empty")
            urlretrieve("https://stlab.adobe.com/wiki/images/d/d3/Test.pdf", 
                        "./testfiles/get_page_count/full/file.pdf")
            os.chdir("./testfiles/get_page_count")

    def test_get_page_count_empty(self):
        """Empty folder should contain no pages"""
        os.chdir("./empty")
        self.assertEqual(0, pdfpagecount.get_page_count())

    def test_get_page_count_full(self):
        """Folder contains pdf file with one page"""
        os.chdir("./full")
        self.assertEqual(1, pdfpagecount.get_page_count())

    def test_get_page_count_recursive(self):
        """Recursive search"""
        self.assertEqual(1, pdfpagecount.get_page_count(recursive=True))

if __name__ == '__main__':
    unittest.main()