import os

from urllib.parse import urljoin

import pywikibot

from pywikibot.bot import QuitKeyboardInterrupt
from pywikibot import pagegenerators
from pywikibot.comms.http import fetch
from pywikibot.specialbots import UploadRobot
from bs4 import BeautifulSoup

first_run = False
pages_checked = 0
oni2_images = 0
file_formats = ('.jpg', '.jpeg', '.png', '.gif', '.svg')

# Scrapes the HTML at the given URL for image tags
def get_image_links(url, shown):
    links = []
    global oni2_images
    global pages_checked

    response = fetch(url)
    if response.status_code != 200:
        pywikibot.output('Skipping url: {}'.format(url))
        return links

    soup = BeautifulSoup(response.text, 'html.parser')
    pages_checked = pages_checked + 1
    if not shown:
        tagname = 'a'
    elif shown == 'just':
        tagname = 'img'
    else:
        tagname = ['a', 'img']
    #pywikibot.output('Looking at tags.')
    for tag in soup.findAll(tagname):
        link = tag.get('src', tag.get('href', None))
        if not link:
            #pywikibot.output('It is not a link.')
            continue
        #pywikibot.output('Got link {0}.'.format(link))
        _, ext = os.path.splitext(link)
        if ext.lower() in file_formats:
            pywikibot.output('Found image link {0}.'.format(ext))
            if "oni2.net" in link:
                pywikibot.stdout('Found an oni2.net image: {0}'.format(link))
                oni2_images = oni2_images + 1
    return links


def main(*args):
    cat = ''
    url = ''
    image_url = False
    shown = False
    desc = []

    local_args = pywikibot.handle_args(args)
    genFactory = pagegenerators.GeneratorFactory()

    for arg in local_args:
        if arg.startswith('-cat:'):
            cat = arg[5:]
        elif arg == '-shown':
            shown = True
        elif arg == '-justshown':
            shown = 'just'
        elif url == '':
            url = arg
        else:
            desc += [arg]
    desc = ' '.join(desc)

    site = pywikibot.Site()
    cat_obj = pywikibot.Category(site, cat)
    generator = pagegenerators.CategorizedPageGenerator(cat_obj, recurse=True)
    for page in pagegenerators.PreloadingGenerator(generator, 100):
        pywikibot.stdout('Checking page {0}'.format(page.title()))
        page_url = page.full_url().replace("%2F", "/")
        get_image_links(page_url, shown)

    global pages_checked
    global oni2_images
    pywikibot.stdout('Checked {0} page(s) and found {1} image(s) from oni2.net.'.format(pages_checked, oni2_images))

if __name__ == '__main__':
    main()