Added url extraction from activitypub profiles

retribute_api/search/activitypub.py

+import lxml.html
+from rest_framework import serializers
+
+
+def get_urls(content):
+    links = []
+    dom = lxml.html.fromstring(content)
+    for link in dom.xpath("//a/@href"):
+        links.append(link)
+    return links
+
+
+def extract_urls_from_attachments(attachments):
+    data = []
+    for attachment in attachments:
+        if attachment["type"] != "PropertyValue":
+            continue
+
+        urls = get_urls(attachment["value"])
+        if not urls:
+            continue
+        row = {"summary": attachment["name"], "url": urls[0]}
+        data.append(row)
+
+    return data
+
+
+class TagSerializer(serializers.Serializer):
+    name = serializers.CharField()
+    type = serializers.CharField()
+
+
+class AttachmentSerializer(serializers.Serializer):
+    type = serializers.CharField()
+    name = serializers.CharField()
+    value = serializers.CharField()
+
+
+class ActorSerializer(serializers.Serializer):
+    attachment = serializers.ListField(child=AttachmentSerializer(), min_length=1)
+    tag = serializers.ListField(child=TagSerializer(), min_length=0)

setup.cfg

@@ -31,6 +31,8 @@ install_requires =
     django-redis
     djangorestframework
     psycopg2-binary
+    lxml
+
 
 
 [options.entry_points]

tests/search/test_activitypub.py

+from retribute_api.search import activitypub
+
+
+def test_profile_serializer():
+    payload = {
+        "tag": [{"type": "Hashtag", "name": "#nobot"}],
+        "attachment": [
+            {
+                "type": "PropertyValue",
+                "name": "patreon",
+                "value": '<a href="https://patreon.com/username" rel="me nofollow noopener">Test</a>',
+            }
+        ],
+    }
+    serializer = activitypub.ActorSerializer(data=payload)
+    assert serializer.is_valid(raise_exception=True) is True
+
+    assert serializer.validated_data == payload
+
+
+def test_extract_urls_from_attachments():
+    attachments = [
+        {
+            "type": "PropertyValue",
+            "name": "Support me on Patreon",
+            "value": '<a href="https://patreon.com/username" rel="me nofollow noopener" target="_blank">This is my Patreon</a>',
+        },
+        {
+            "type": "PropertyValue",
+            "name": "Support me on Ko-Fi",
+            "value": '<a href="https://ko-fi.com/username" rel="me nofollow noopener" target="_blank">This is my Ko-Fi</a>',
+        },
+        {"type": "PropertyValue", "name": "Irrelevant text", "value": "No link"},
+    ]
+
+    expected = [
+        {"summary": "Support me on Patreon", "url": "https://patreon.com/username"},
+        {"summary": "Support me on Ko-Fi", "url": "https://ko-fi.com/username"},
+    ]
+
+    assert activitypub.extract_urls_from_attachments(attachments) == expected