コード例 #1
0
def build_page(content, url):
    page = BeautifulSoup(content, 'lxml')
    subtitle = page.find('p', class_="novel_subtitle").get_text()
    content = page.find('div', id="novel_honbun", class_="novel_view")
    correct_point_ruby_as_bold(content)
    if fullruby:
        content = yomituki.ruby_div(content)
    else:
        content = content.prettify()
    append = page.find('div', id="novel_a", class_="novel_view")
    if append is not None:
        correct_point_ruby_as_bold(append)
        if fullruby:
            append = yomituki.ruby_div(append)
        else:
            append = append.prettify()
        append = '<hr>' + append
    else:
        append = ''
    html = '<html>\n<head>\n' + '<title>' + subtitle + '</title>\n</head>\n<body>\n<div>\n<h3>' + subtitle + '</h3>\n' + content + append + '</div>\n</body>\n</html>'
    name = url.split('/')[-2]
    built_page = epub.EpubHtml(title=subtitle,
                               file_name=name + '.xhtml',
                               content=html,
                               lang='ja_jp')
    return name, built_page
コード例 #2
0
def build_page(content, url):
    page = BeautifulSoup(content, 'lxml')
    subtitle = page.find('p', class_="widget-episodeTitle js-vertical-composition-item").get_text()
    content = page.find('div', class_="widget-episodeBody js-episode-body")
    if fullruby:
        content = yomituki.ruby_div(content)
    else:
        content = content.prettify()
    html = '<html>\n<head>\n' + '<title>' + subtitle + '</title>\n</head>\n<body>\n<div>\n<h3>' + subtitle + '</h3>\n' + content + '</div>\n</body>\n</html>'
    name = url.split('/')[-1]
    built_page = epub.EpubHtml(title=subtitle, file_name=name + '.xhtml', content=html, lang='ja_jp')
    return name, built_page
コード例 #3
0
ファイル: alphapolis.py プロジェクト: exc4l/WebNovelCrawler
def build_page(content, url):
    page = BeautifulSoup(content, 'lxml')
    subtitle = page.find('h2', class_="episode-title").get_text().replace(
        '\n', '').replace('\t', '')
    content = page.find('div', id="novelBoby", class_="text")
    if fullruby:
        content = yomituki.ruby_div(content)
    else:
        content = content.prettify()
    html = '<html>\n<head>\n' + '<title>' + subtitle + '</title>\n</head>\n<body>\n<div>\n<h3>' + subtitle + '</h3>\n' + content + '</div>\n</body>\n</html>'
    name = url.split('/')[-1]
    built_page = epub.EpubHtml(title=subtitle,
                               file_name=name + '.xhtml',
                               content=html,
                               lang='ja_jp')
    return name, built_page