fix malformed urls

jg/plucker/meetups_pehapkari/spider.py

@@ -43,7 +43,7 @@ def parse_event(
 
         self.logger.info(f"Event: {event.summary} {event.begin}")
         try:
-            url = html.fromstring(event.description).xpath("//a/@href")[-1]
+            url = fix_url(html.fromstring(event.description).xpath("//a/@href")[-1])
         except IndexError:
             url = self.default_event_url
         return Meetup(
@@ -58,3 +58,11 @@ def parse_event(
             series_org="komunita kolem PHP",
             series_url="https://www.pehapkari.cz/",
         )
+
+
+def fix_url(url: str) -> str:
+    if url.startswith("http"):
+        return url
+    if url.startswith("www."):
+        return f"https://{url}"
+    raise ValueError(f"Invalid URL: {url}")

tests/test_meetups_pehapkari_spider.py

@@ -0,0 +1,20 @@
+import pytest
+
+from jg.plucker.meetups_pehapkari.spider import fix_url
+
+
+@pytest.mark.parametrize(
+    "url, expected",
+    [
+        (
+            "www.facebook.com/events/1118251933124168",
+            "https://www.facebook.com/events/1118251933124168",
+        ),
+        (
+            "https://www.meetup.com/pra%C5%BEske-srazy-p%C5%99atel-php-pehapkari-cz/events/305454246/",
+            "https://www.meetup.com/pra%C5%BEske-srazy-p%C5%99atel-php-pehapkari-cz/events/305454246/",
+        ),
+    ],
+)
+def test_fix_url(url: str, expected: str):
+    assert fix_url(url) == expected