From cd58b96fad79e15d23a0bbb63977ae13b8ffffef Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Javi=20Mart=C3=ADn?= <javim@elretirao.net>
Date: Mon, 20 Dec 2021 16:22:14 +0100
Subject: [PATCH] Support geozone segments with non-Latin characters
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

The `parameterize` method uses the `I18n.transliterate` method, whose
documentation says:

```
I18n.transliterate("Ærøskøbing")
=> "AEroskobing"

I18n.transliterate("日本語")
=> "???"
```

That means we can't use it for dictionaries where characters don't have
a transliteration to the latin alphabet.

So we're changing the code in order to only transliterate characters
with a transliteration to the latin alphabet.

Note the first example ("Česká republika") already worked with the
previous code; the test has been added to make sure accented characters
are handled properly.
---
 lib/user_segments.rb           |  8 +++++++-
 spec/lib/user_segments_spec.rb | 10 ++++++++++
 2 files changed, 17 insertions(+), 1 deletion(-)

diff --git a/lib/user_segments.rb b/lib/user_segments.rb
index 5ee5bfe49..7e4e8a154 100644
--- a/lib/user_segments.rb
+++ b/lib/user_segments.rb
@@ -84,6 +84,12 @@ class UserSegments
     end
 
     def self.geozones
-      Geozone.order(:name).map { |geozone| [geozone.name.parameterize.underscore, geozone] }.to_h
+      Geozone.order(:name).map do |geozone|
+        [geozone.name.gsub(/./) { |char| character_approximation(char) }.underscore.tr(" ", "_"), geozone]
+      end.to_h
+    end
+
+    def self.character_approximation(char)
+      I18n::Backend::Transliterator::HashTransliterator::DEFAULT_APPROXIMATIONS[char] || char
     end
 end
diff --git a/spec/lib/user_segments_spec.rb b/spec/lib/user_segments_spec.rb
index 4aa454cbe..79b6d176f 100644
--- a/spec/lib/user_segments_spec.rb
+++ b/spec/lib/user_segments_spec.rb
@@ -31,6 +31,16 @@ describe UserSegments do
         expect(UserSegments.segment_name("lowlands_and_highlands")).to eq "Lowlands and Highlands"
       end
 
+      it "supports international alphabets" do
+        create(:geozone, name: "Česká republika")
+        create(:geozone, name: "България")
+        create(:geozone, name: "日本")
+
+        expect(UserSegments.segment_name("ceska_republika")).to eq "Česká republika"
+        expect(UserSegments.segment_name("България")).to eq "България"
+        expect(UserSegments.segment_name("日本")).to eq "日本"
+      end
+
       it "returns regular segments when the geozone doesn't exist" do
         expect(UserSegments.segment_name("all_users")).to eq "All users"
       end