iter01-validator.py

import os
import re
from bs4 import BeautifulSoup, NavigableString


GIT_DIR = '/home/jakub/github/pb138/'

def validate(url):
    page = open(url)

    soup = BeautifulSoup(page.read(), 'html.parser')

    forbidden_tags = soup.find_all(['table', 'td', 'tr', 'strong', 'small', 'h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'h7', 'svg', 'br', 'style'])

    if len(forbidden_tags) > 0:
        print('[ FAIL ] submission contains forbidden tags', forbidden_tags)

    if len(soup.find_all('body')) != 1:
        print('[ FAIL] I expected to find exactly one body tag')
        return

    forbidden_attrs = ['style', 'width', 'height']
    for tag in soup.body.children:
        if isinstance(tag, NavigableString):
            continue
        for attr in forbidden_attrs:
            if tag.has_attr(attr):
                print(f'[ FAIL ] {attr} attrs are forbidden: {tag}')

    for img in soup.find_all('img'):
        if not img.has_attr('alt'):
            print('[ FAIL ]Imgs must have alt text', img)
        elif len(img['alt']) <= 3:
            print('[ WARN ] ALt tag should be something meaningful', img)

    for row in (soup.find_all('row') + soup.find_all('div', {'class', re.compile(r'.*=row( )?.*')})):
        if row is not None and len(row.find_all('col')) + len(row.find_all('div', {'class', re.compile(r'.*=col( )?.*')})) < 1:
            print('[ WARN ]There are rows without cols!', row)

    for inp in soup.find_all('input'):
        if len(inp.find_parents('form')) != 1:
            print('Every input should be wrapped in one form', inp)



for repo in os.listdir(GIT_DIR):
    flag = False
    for root, dirs, files in os.walk(os.path.join(GIT_DIR, repo)):
        for file in files:
            if file.endswith('.html') and flag:
                print('Duplicate html file', os.path.join(root, file))

            if file.endswith('.html') and not flag:
                flag = True
                print('~~~~~~')
                print(repo)
                print('~~~~~~')
                validate(os.path.join(root, file))

    if not flag:
        print('~~~~~\nHtml file for ' + repo + ' not found!\n~~~~~~')