FIX: Return proper results when searching for a topic in Japanese

Currently, when the default locale is Japanese, the search for a topic
using its URL, path or ID doesn’t work as expected. It will either
return wrong results or no result at all.

The problem lies with how we process the provided terms in Japanese
mode. For example, if `http://localhost/t/-/55` is provided, currently
this will result in `http localhost t 5 5` to be searched for.

This patch addresses the issue by checking whether the provided term
needs segmenting. If the provided term is a number, or a path or a full
URL, then it doesn’t need segmenting. When that happens we skip the
processing we normally apply for Japanese, making the search return the
expected results.

lib/search.rb

@@ -117,7 +117,7 @@ class Search
     data.force_encoding("UTF-8")
     data = clean_term(data)
 
-    if purpose != :topic
+    if purpose != :topic && need_segmenting?(data)
       if segment_chinese?
         require "cppjieba_rb" unless defined?(CppjiebaRb)
 
@@ -226,6 +226,13 @@ class Search
       end
   end
 
+  def self.need_segmenting?(data)
+    return false if data.match?(/\A\d+\z/)
+    !URI.parse(data).path.start_with?("/")
+  rescue URI::InvalidURIError
+    true
+  end
+
   attr_accessor :term
   attr_reader :clean_term, :guardian
 

spec/lib/search_spec.rb

@@ -9,6 +9,36 @@ RSpec.describe Search do
     Jobs.run_immediately!
   end
 
+  describe ".need_segmenting?" do
+    subject(:search) { described_class }
+
+    context "when data only contains digits" do
+      let(:data) { "510" }
+
+      it { is_expected.not_to be_need_segmenting(data) }
+    end
+
+    context "when data does not only contain digits" do
+      context "when data is a full URL" do
+        let(:data) { "http://localhost/t/-/510" }
+
+        it { is_expected.not_to be_need_segmenting(data) }
+      end
+
+      context "when data is a path" do
+        let(:data) { "/t/-/510" }
+
+        it { is_expected.not_to be_need_segmenting(data) }
+      end
+
+      context "when data is something else" do
+        let(:data) { "text" }
+
+        it { is_expected.to be_need_segmenting(data) }
+      end
+    end
+  end
+
   describe "#ts_config" do
     it "maps locales to correct Postgres dictionaries" do
       expect(Search.ts_config).to eq("english")
@@ -1597,10 +1627,38 @@ RSpec.describe Search do
     let!(:post_2) { Fabricate(:post, topic: topic_2) }
 
     describe ".prepare_data" do
-      it "removes punctuations" do
-        SiteSetting.search_tokenize_japanese = true
+      subject(:prepared_data) { Search.prepare_data(data) }
 
-        expect(Search.prepare_data(post.raw)).to eq("This is some japanese text 日本 が 大好き です")
+      let(:data) { post.raw }
+
+      before { SiteSetting.search_tokenize_japanese = true }
+
+      it "removes punctuations" do
+        expect(prepared_data).to eq("This is some japanese text 日本 が 大好き です")
+      end
+
+      context "when providing only an URL" do
+        let(:data) { "http://localhost/t/-/51" }
+
+        it "does not change it" do
+          expect(prepared_data).to eq(data)
+        end
+      end
+
+      context "when providing only a path" do
+        let(:data) { "/t/-/51" }
+
+        it "does not change it" do
+          expect(prepared_data).to eq(data)
+        end
+      end
+
+      context "when providing only an ID" do
+        let(:data) { "51" }
+
+        it "does not change it" do
+          expect(prepared_data).to eq(data)
+        end
       end
     end
 
@@ -1616,23 +1674,24 @@ RSpec.describe Search do
         SiteSetting.refresh!
       end
 
-      it "finds posts containing Japanese text if tokenization is forced" do
-        SiteSetting.search_tokenize_japanese = true
+      context "when tokenization is forced" do
+        before { SiteSetting.search_tokenize_japanese = true }
 
+        it "finds posts containing Japanese text" do
           expect(Search.execute("日本").posts.map(&:id)).to eq([post_2.id, post.id])
           expect(Search.execute("日").posts.map(&:id)).to eq([post_2.id, post.id])
         end
+      end
 
-      it "find posts containing search term when site's locale is set to Japanese" do
-        SiteSetting.default_locale = "ja"
+      context "when default locale is set to Japanese" do
+        before { SiteSetting.default_locale = "ja" }
 
+        it "find posts containing search term" do
           expect(Search.execute("日本").posts.map(&:id)).to eq([post_2.id, post.id])
           expect(Search.execute("日").posts.map(&:id)).to eq([post_2.id, post.id])
         end
 
         it "does not include superfluous spaces in blurbs" do
-        SiteSetting.default_locale = "ja"
-
           post.update!(
             raw: "場サアマネ織企ういかせ竹域ヱイマ穂基ホ神3予読ずねいぱ松査ス禁多サウ提懸イふ引小43改こょドめ。深とつぐ主思料農ぞかル者杯検める活分えほづぼ白犠",
           )
@@ -1642,6 +1701,42 @@ RSpec.describe Search do
           expect(results.posts.length).to eq(1)
           expect(results.blurb(results.posts.first)).to include("ういかせ竹域")
         end
+
+        context "when searching for a topic in particular" do
+          subject(:results) do
+            described_class.execute(
+              term,
+              guardian: Discourse.system_user.guardian,
+              type_filter: "topic",
+              search_for_id: true,
+            )
+          end
+
+          context "when searching by topic ID" do
+            let(:term) { topic.id }
+
+            it "finds the proper post" do
+              expect(results.posts.first).to have_attributes(topic: topic, post_number: 1)
+            end
+          end
+
+          context "when searching by topic URL" do
+            let(:term) { "http://#{Discourse.current_hostname}/t/-/#{topic.id}" }
+
+            it "finds the proper post" do
+              expect(results.posts.first).to have_attributes(topic: topic, post_number: 1)
+            end
+          end
+
+          context "when searching by topic path" do
+            let(:term) { "/t/-/#{topic.id}" }
+
+            it "finds the proper post" do
+              expect(results.posts.first).to have_attributes(topic: topic, post_number: 1)
+            end
+          end
+        end
+      end
     end
   end