gbyukg / gbyukg.github.com Goto Github PK

from bs4 import BeautifulSoup
import itertools

wrapFile = "./.tmp/cheatsheets/bash/index.html"

def wrapH2(wrapEls):
    for wrapEls in h2s:
        els = [i for i in itertools.takewhile(lambda x: x.name not in [wrapEls.name, 'script'], wrapEls.next_siblings)]

        h2_section = soup.new_tag('div', **{'class':'h2-section'})
        h3_section_list = soup.new_tag('div', **{'class':'h3-section-list'})

        wrapEls.wrap(h2_section)

        for tag in els:
            h3_section_list.append(tag)

        h2_section.append(h3_section_list)

        # for tag in els:
        #     h2_section.append(tag)

def wrapH3(wrapEls):
    for wrapEls in h3s:
        els = [i for i in itertools.takewhile(lambda x: x.name not in [wrapEls.name, 'script'], wrapEls.next_siblings)]

        h3_section = soup.new_tag('div', **{'class':'h3-section'})
        body_section = soup.new_tag('div', **{'class':'cheatsheet-body'})

        wrapEls.wrap(h3_section)

        for tag in els:
            body_section.append(tag)

        h3_section.append(body_section)

with open(wrapFile, 'r', encoding='utf-8') as source:
    soup = BeautifulSoup(source)

    main = soup.find('article', class_="article") # None
    h2s = main.find_all('h2', recursive=False)
    h3s = main.find_all('h3', recursive=False)

    wrapH2(h2s)
    wrapH3(h3s)

with open(wrapFile, 'w', encoding='utf-8') as desc:
    desc.write(soup.prettify(formatter="None")) # 禁止转义 ⁢ 字符

print(soup.prettify())