From 6c865cb0b71abd768882fe7275bd0a4a7eb517b3 Mon Sep 17 00:00:00 2001
From: Marwin Hochfelsner <50826859+hlfan@users.noreply.github.com>
Date: Mon, 27 Oct 2025 21:16:24 +0100
Subject: [PATCH] Increase tag2link item leniency

---
 lib/tag2link.rb           | 39 +++++++++++++++++++++++++++++++--------
 test/lib/tag2link_test.rb | 35 +++++++++++++++++++++++++++++++++++
 2 files changed, 66 insertions(+), 8 deletions(-)

diff --git a/lib/tag2link.rb b/lib/tag2link.rb
index b4ab9c3b7..58bdb16dc 100644
--- a/lib/tag2link.rb
+++ b/lib/tag2link.rb
@@ -19,13 +19,36 @@ module Tag2link
     data
       # exclude deprecated and third-party URLs
       .reject { |item| item["rank"] == "deprecated" || item["source"] == "wikidata:P3303" }
-      .group_by { |item| item["key"] }
-      .transform_keys { |key| key.sub(/^Key:/, "") }
-      # move preferred to the start of the array
-      .transform_values { |items| items.sort_by { |item| item["rank"] == "preferred" ? 0 : 1 }.uniq { |item| item["url"] } }
-      # exclude any that are ambiguous, i.e. the best and second-best have the same rank
-      .reject { |_key, value| value[1] && value[0]["rank"] == value[1]["rank"] }
-      # keep only the best match
-      .transform_values { |items| items[0]["url"] }
+      .group_by { |item| item["key"].sub(/^Key:/, "") }
+      .transform_values { |items| choose_best_item(items) }
+      .compact
+      .transform_values { |items| items["url"] }
   end
+
+  def self.choose_best_item(items)
+    return nil if items.blank?
+
+    return items.first if items.size == 1
+
+    # move preferred to the start of the array
+    ranked = items.sort_by { |item| item["rank"] == "preferred" ? 0 : 1 }.uniq { |item| item["url"] }
+    top_rank = ranked.first["rank"]
+    top_items = ranked.select { |i| i["rank"] == top_rank }
+
+    # if only one top-ranked item, prefer that
+    return top_items.first if top_items.size == 1
+
+    grouped = top_items.group_by { |i| i["source"] }
+    return nil if grouped.size > 2
+
+    # if both sources have exactly one preferred, prefer osmwiki
+    return grouped["osmwiki:P8"]&.first || grouped.values.flatten.first if grouped.all? { |_s, vals| vals.size == 1 }
+
+    # if one source has multiple preferreds and the other has one, prefer the single one
+    return grouped.min_by { |_s, vals| vals.size }.last.first if grouped.any? { |_s, vals| vals.size == 1 }
+
+    # exclude any that are ambiguous
+    nil
+  end
+  private_class_method :choose_best_item
 end
diff --git a/test/lib/tag2link_test.rb b/test/lib/tag2link_test.rb
index f7dbd6abd..cc207cfa8 100644
--- a/test/lib/tag2link_test.rb
+++ b/test/lib/tag2link_test.rb
@@ -36,6 +36,14 @@ class Tag2linkTest < ActiveSupport::TestCase
     dict = Tag2link.build_dict(data)
     assert_equal "http://example.com/$1", dict["example"]
 
+    data = [
+      { "key" => "Key:example", "url" => "http://example2.com/$1", "rank" => "preferred", "source" => "osmwiki:P8" },
+      { "key" => "Key:example", "url" => "http://example.com/$1", "rank" => "preferred", "source" => "wikidata:P1630" },
+      { "key" => "Key:example", "url" => "http://example3.com/$1", "rank" => "preferred", "source" => "osmwiki:P8" }
+    ]
+    dict = Tag2link.build_dict(data)
+    assert_equal "http://example.com/$1", dict["example"]
+
     data = [
       { "key" => "Key:example", "url" => "http://example2.com/$1", "rank" => "preferred", "source" => "osmwiki:P8" }
     ]
@@ -67,4 +75,31 @@ class Tag2linkTest < ActiveSupport::TestCase
     dict = Tag2link.build_dict(data)
     assert_not_includes dict, "example"
   end
+
+  def test_build_dict_chooses_osmwiki_when_both_have_single_preferred
+    data = [
+      { "key" => "Key:example", "url" => "http://example1.com/$1", "rank" => "preferred", "source" => "osmwiki:P8" },
+      { "key" => "Key:example", "url" => "http://example2.com/$1", "rank" => "preferred", "source" => "wikidata:P1630" }
+    ]
+    dict = Tag2link.build_dict(data)
+    assert_equal "http://example1.com/$1", dict["example"]
+
+    data = [
+      { "key" => "Key:example", "url" => "http://example1.com/$1", "rank" => "normal", "source" => "osmwiki:P8" },
+      { "key" => "Key:example", "url" => "http://example2.com/$1", "rank" => "normal", "source" => "wikidata:P1630" }
+    ]
+    dict = Tag2link.build_dict(data)
+    assert_equal "http://example1.com/$1", dict["example"]
+  end
+
+  def test_build_dict_multiple_sources_more_than_two
+    data = [
+      { "key" => "Key:example", "url" => "http://example1.com/$1", "rank" => "normal", "source" => "osmwiki:P8" },
+      { "key" => "Key:example", "url" => "http://example2.com/$1", "rank" => "normal", "source" => "wikidata:P1630" },
+      { "key" => "Key:example", "url" => "http://example3.com/$1", "rank" => "normal", "source" => "other:source" }
+    ]
+    dict = Tag2link.build_dict(data)
+    # Should not happen with current tag2link schema, but ensure we handle it gracefully
+    assert_not_includes dict, "example"
+  end
 end
-- 
2.39.5