Extractor for Pathfinder Humble Bundle.

extract.py

#!/usr/bin/env python3

import os
import re
import zipfile

fileMatch = re.compile(
    r"(Pathfinder|Starfinder)(RoleplayingGame|CampaignSetting|Module|AdventurePath|Chronicles|Companion|PlayerCompanion|Playtest|CoreRulebook|SocietyScenario)(.*)(PDF|Bundle)(-SingleFile)?.zip")

mainPDFName = re.compile(r"(PZO(PSS)?[0-9\-]+B?E).pdf")


def get_main_pdf_name(names):
    for name in names:
        m = re.search(mainPDFName, name)

        if m != None:
            return m.group(1)

    return None

def get_additional_names(names, main_name):
    ret = []

    for name in names:
        if name.startswith(main_name + "-"):
            ret += [name]

    return ret

def extract_file(zip, filename, base_filename):
    input = zip.open(filename, "r")

    if "-" in filename:
        filename = filename[(filename.find("-")):]
    else:
        filename = ".pdf"
    
    print("[DEBUG  ] Extracting %s to %s%s" % (filename, base_filename, filename, ))

    output = open("%s%s" %
                (base_filename, filename, ), "wb")

    output.write(input.read())

def extract_files_in_path(path):
    for file in os.listdir(path):
        filename = os.path.join(path, file)

        if not filename.endswith(".zip"):
            continue

        print("[INFO   ] Extracting", filename)

        m = re.search(fileMatch, filename)

        if m == None:
            print("[WARNING] This doesn't look like a Pazio filename.")
            continue

        game_name, file_type, file_name, download_type, _ = m.groups()

        print("[DEBUG  ]", game_name, file_type, file_name, download_type)

        if game_name == "Starfinder":
            file_type = "Starfinder"
            file_name = "CoreRulebook"
        
        if not os.path.exists(file_type):
            os.mkdir(file_type)

        zip = zipfile.ZipFile(filename)

        names = zip.namelist()

        print("[DEBUG  ]", "Zip Names: ", names)

        main_pdf_name = get_main_pdf_name(names)

        if main_pdf_name == None:
            print("[ERROR  ] Could not determine main file")
            continue

        base_filename = "%s/%s" % (file_type, file_name, )

        extract_file(zip, main_pdf_name + ".pdf", base_filename)

        for add_file in get_additional_names(names, main_pdf_name):
            extract_file(zip, add_file, base_filename)

downloads_dir = os.path.expanduser("~\\Downloads")

extract_files_in_path(".")
extract_files_in_path(downloads_dir)