From e46378286942fff2baff7a17ab2f254bcbe5da74 Mon Sep 17 00:00:00 2001
From: Thomas Sileo <t@a4.io>
Date: Tue, 16 Jul 2019 21:33:32 +0200
Subject: [PATCH] Add webmentions utils

---
 utils/webmentions.py | 54 ++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 54 insertions(+)
 create mode 100644 utils/webmentions.py

diff --git a/utils/webmentions.py b/utils/webmentions.py
new file mode 100644
index 0000000..307ccc1
--- /dev/null
+++ b/utils/webmentions.py
@@ -0,0 +1,54 @@
+import logging
+from typing import Optional
+from urllib.parse import urlparse
+
+import requests
+from bs4 import BeautifulSoup
+from little_boxes.urlutils import is_url_valid
+
+logger = logging.getLogger(__name__)
+
+
+def _make_abs(url: Optional[str], parent: str) -> Optional[str]:
+    if url is None:
+        return None
+
+    if url.startswith("http"):
+        return url
+
+    return (
+        urlparse(parent)._replace(path=url, params="", query="", fragment="").geturl()
+    )
+
+
+def _discover_webmention_endoint(url: str) -> Optional[str]:
+    try:
+        resp = requests.get(url, timeout=3)
+    except Exception:
+        return None
+
+    for k, v in resp.links.items():
+        if "webmention" in k:
+            return _make_abs(resp.links[k].get("url"), url)
+
+    soup = BeautifulSoup(resp.text, "html5lib")
+    wlinks = soup.find_all(["link", "a"], attrs={"rel": "webmention"})
+    for wlink in wlinks:
+        if "href" in wlink.attrs:
+            return _make_abs(wlink.attrs["href"], url)
+
+    return None
+
+
+def discover_webmention_endoint(url: str) -> Optional[str]:
+    """Discover the Webmention endpoint of a given URL, if any.
+
+    Passes all the tests at https://webmention.rocks!
+
+    """
+    wurl = _discover_webmention_endoint(url)
+    if wurl is None:
+        return None
+    if not is_url_valid(wurl):
+        return None
+    return wurl