From d2091e8f71ce2dda90c86e3b3a4edce3ec723981 Mon Sep 17 00:00:00 2001
From: Valentin Lorentz <progval+git@progval.net>
Date: Thu, 11 Nov 2021 19:07:31 +0100
Subject: [PATCH] Geography: Always prefer IANA timezones over UTC offsets

Even if the UTC offsets are given on a more specific entity,
they are usually worthless to us (eg. DST).
---
 plugins/Geography/test.py     |  6 ++++++
 plugins/Geography/wikidata.py | 25 ++++++++++++++++---------
 2 files changed, 22 insertions(+), 9 deletions(-)

diff --git a/plugins/Geography/test.py b/plugins/Geography/test.py
index 7accfa44b..7cbb7a77b 100644
--- a/plugins/Geography/test.py
+++ b/plugins/Geography/test.py
@@ -180,6 +180,12 @@ class GeographyWikidataTestCase(SupyTestCase):
             utils.time.iana_timezone("America/New_York"),
         )
 
+        self.assertEqual(
+            # Paris, France
+            wikidata.timezone_from_uri("http://www.wikidata.org/entity/Q90"),
+            utils.time.iana_timezone("Europe/Paris"),
+        )
+
     @skipIf(not network, "Network test")
     def testParentQualifiedIgnorePreferred(self):
         # The queried object does not have a TZ property,
diff --git a/plugins/Geography/wikidata.py b/plugins/Geography/wikidata.py
index df7ff7ced..712fe01a1 100644
--- a/plugins/Geography/wikidata.py
+++ b/plugins/Geography/wikidata.py
@@ -79,6 +79,9 @@ SELECT ?item ?itemLabel ?rank ?endtime ?appliestopart ?utcoffset ?tzid (MIN(?are
   # Get the area, will be used to order by specificity
   OPTIONAL { ?statementsubject wdt:P2046 ?area. }
 
+  # Require that ?timezone be an instance of...
+  ?timezone (wdt:P31/wdt:P279*) <$tztype>.
+
   {
     # Get either an IANA timezone ID...
     ?timezone wdt:P6687 ?tzid.
@@ -130,15 +133,19 @@ def _query_sparql(query):
 def timezone_from_uri(location_uri):
     """Returns a :class:datetime.tzinfo object, given a Wikidata Q-ID.
     eg. ``"Q60"`` for New York City."""
-    data = _query_sparql(TIMEZONE_QUERY.substitute(subject=location_uri))
-    results = data["results"]["bindings"]
-    for result in results:
-        if "tzid" in result:
-            return utils.time.iana_timezone(result["tzid"]["value"])
-        else:
-            assert "utcoffset" in result
-            utc_offset = float(result["utcoffset"]["value"])
-            return datetime.timezone(datetime.timedelta(hours=utc_offset))
+    for tztype in [
+        "http://www.wikidata.org/entity/Q17272692", # IANA timezones first
+        "http://www.wikidata.org/entity/Q12143", # any timezone as a fallback
+    ]:
+        data = _query_sparql(TIMEZONE_QUERY.substitute(subject=location_uri, tztype=tztype))
+        results = data["results"]["bindings"]
+        for result in results:
+            if "tzid" in result:
+                return utils.time.iana_timezone(result["tzid"]["value"])
+            else:
+                assert "utcoffset" in result
+                utc_offset = float(result["utcoffset"]["value"])
+                return datetime.timezone(datetime.timedelta(hours=utc_offset))
 
     return None