discovery.py

import requests
from bs4 import BeautifulSoup
from bs4 import Comment
from urllib.parse import urlparse

from indieweb_utils import canonicalize_url

# from ..utils.urls import canonicalize_url

class TrackbackError(Exception):
    """Base class for trackback errors."""
    pass

class ConnectionError(TrackbackError):
    """Raised when a connection error occurs."""
    pass

class InvalidStatusCodeError(TrackbackError):
    """Raised when the server returns an invalid status code."""
    pass

def discover_trackback_url(url: str):
    """Discover the trackback URL from a URL.
    
    :param url: The URL to discover the trackback URL from.
    :returns: The trackback URL.
    """

    get_trackback_url_request = requests.get(url)

    if get_trackback_url_request.status_code != 200:
        raise InvalidStatusCodeError("The server returned a status code of {}.".format(get_trackback_url_request.status_code))
    
    soup = BeautifulSoup(get_trackback_url_request.text, "html.parser")

    # get all comments
    comments = soup.find_all(string=lambda text:isinstance(text, Comment))

    for c in comments:
        soup = BeautifulSoup(c, "html.parser")

        # print(soup)
        
        rdf = soup.find("rdf:description")

        if not rdf:
            continue

        trackback_url = rdf.get("trackback:ping")

        if not trackback_url:
            raise TrackbackError("No trackback URL found in RDF.")
        
        domain = urlparse(url).netloc

        return canonicalize_url(trackback_url, domain)
    
    return ""


def send_trackback(url, title: str = None, excerpt: str = None, blog_name: str = None):
    """Send a trackback to a URL.

    :param url: The URL to send the trackback to.
    :param title: The title of the post.
    :param excerpt: The excerpt of the post.
    :param blog_name: The name of the blog.
    :returns: The status code and message from the server.

    """

    try:
        send_trackback_request = requests.get(url)
    except requests.exceptions.ConnectionError:
        raise ConnectionError("Could not connect to the server.")
    
    if send_trackback_request.status_code != 200:
        raise InvalidStatusCodeError("The server returned a status code of {}.".format(send_trackback_request.status_code))
    
    soup = BeautifulSoup(send_trackback_request.text, "html.parser")["response"]

    # validate response

    if soup.find("error") and soup.find("error").text != "0":
        raise TrackbackError("The server returned an error: {}".format(soup.find("message").text))
    
new_trackback_url = discover_trackback_url("https://arxiv.org/abs/1706.03762")

print(new_trackback_url)