INNER CODE UNIT · Python

get_links_and_destinations_from_html

zcash/zips · links_and_dests.py:54

def get_links_and_destinations_from_html(f):
    links = set()
    internal = set()
    dests = set()

    soup = BeautifulSoup(f.read(), "html5lib")

    # First try to find this: <script type="application/json" data-target="react-app.embeddedData">
    # If it exists, its content is some JSON that we need to parse to get the real content.
    for script in soup.find_all('script'):
        if script.get('data-target') == "react-app.embeddedData":
            content = json.loads(script.string).get('payload', {}).get('blob', {}).get('richText')
            if content is not None:
                (links, dests) = get_links_and_destinations_from_html(StringIO(content))
                break

    for link in soup.find_all('a'):
        if link.has_attr('href'):

View source record →