From da2d18495845c9a6bde6ff206e02f9889cab621c Mon Sep 17 00:00:00 2001
From: Ezra Lim <ezralim1992@gmail.com>
Date: Tue, 25 Mar 2025 14:33:13 +0800
Subject: [PATCH] fix: ensure git diff output is properly encoded as UTF-8

burn: Remove unused fixture
---
 lib/committer/git_helper.rb             |  2 +-
 spec/committer/commit_generator_spec.rb |  2 +-
 spec/committer/git_helper_spec.rb       | 23 +++++++++++++++++++++++
 spec/fixtures/invalid-utf8-diff.txt     |  7 +++++++
 spec/fixtures/sample_diff.txt           | 11 -----------
 spec/fixtures/utf-8-diff.txt            |  7 +++++++
 6 files changed, 39 insertions(+), 13 deletions(-)
 create mode 100644 spec/fixtures/invalid-utf8-diff.txt
 delete mode 100644 spec/fixtures/sample_diff.txt
 create mode 100644 spec/fixtures/utf-8-diff.txt

diff --git a/lib/committer/git_helper.rb b/lib/committer/git_helper.rb
index 98ae36b..04cd7b5 100644
--- a/lib/committer/git_helper.rb
+++ b/lib/committer/git_helper.rb
@@ -22,7 +22,7 @@ def staged_diff
         stdout, stderr, status = Open3.capture3('git diff --staged')
         raise Committer::Error, "Failed to get git diff: #{stderr}" unless status.success?
 
-        stdout
+        stdout.dup.force_encoding('UTF-8').scrub
       end
     end
   end
diff --git a/spec/committer/commit_generator_spec.rb b/spec/committer/commit_generator_spec.rb
index ad02827..6a79d25 100644
--- a/spec/committer/commit_generator_spec.rb
+++ b/spec/committer/commit_generator_spec.rb
@@ -4,7 +4,7 @@
 require 'committer/commit_generator'
 
 RSpec.describe Committer::CommitGenerator do
-  let(:diff) { load_fixture('sample_diff.txt') }
+  let(:diff) { load_fixture('utf-8-diff.txt') }
   let(:commit_context) { nil }
   let(:summary_response) { JSON.parse(load_fixture('claude_response_summary.json')) }
   let(:body_response) { JSON.parse(load_fixture('claude_response_with_body.json')) }
diff --git a/spec/committer/git_helper_spec.rb b/spec/committer/git_helper_spec.rb
index d3ffaab..64ec4a3 100644
--- a/spec/committer/git_helper_spec.rb
+++ b/spec/committer/git_helper_spec.rb
@@ -25,6 +25,29 @@
         Committer::GitHelper.staged_diff
       end.to raise_error(Committer::Error, 'Failed to get git diff: error message')
     end
+
+    it 'properly handles UTF-8 characters in diff output' do
+      utf8_diff = load_fixture('utf-8-diff.txt')
+      allow(Open3).to receive(:capture3).with('git diff --staged').and_return([utf8_diff, '',
+                                                                               double(success?: true)])
+      result = Committer::GitHelper.staged_diff
+      expect(result.encoding.name).to eq('UTF-8')
+      expect(result.valid_encoding?).to be true
+      expect(result).to eq(utf8_diff)
+    end
+
+    it 'handles invalid UTF-8 characters in diff output' do
+      invalid_utf8_diff = load_fixture('invalid-utf8-diff.txt')
+      # Make sure the test fixture has invalid UTF-8 characters
+      invalid_utf8_data = invalid_utf8_diff.dup.force_encoding('UTF-8')
+      expect(invalid_utf8_data.valid_encoding?).to be false
+
+      allow(Open3).to receive(:capture3).with('git diff --staged').and_return([invalid_utf8_diff, '',
+                                                                               double(success?: true)])
+      result = Committer::GitHelper.staged_diff
+      expect(result.encoding.name).to eq('UTF-8')
+      expect(result.valid_encoding?).to be true
+    end
   end
 
   describe '.commit' do
diff --git a/spec/fixtures/invalid-utf8-diff.txt b/spec/fixtures/invalid-utf8-diff.txt
new file mode 100644
index 0000000..0e6a299
--- /dev/null
+++ b/spec/fixtures/invalid-utf8-diff.txt
@@ -0,0 +1,7 @@
+diff --git a/test.txt b/test.txt
+new file mode 100644
+index 0000000..ea49386
+--- /dev/null
++++ b/test.txt
+@@ -0,0 +1 @@
++This is a string with invalid UTF-8 bytes: �( �( �(�(
\ No newline at end of file
diff --git a/spec/fixtures/sample_diff.txt b/spec/fixtures/sample_diff.txt
deleted file mode 100644
index 42b6c72..0000000
--- a/spec/fixtures/sample_diff.txt
+++ /dev/null
@@ -1,11 +0,0 @@
-diff --git a/lib/committer/version.rb b/lib/committer/version.rb
-index 12345678..87654321 100644
---- a/lib/committer/version.rb
-+++ b/lib/committer/version.rb
-@@ -1,5 +1,5 @@
- # frozen_string_literal: true
-
- module Committer
--  VERSION = '0.1.0'
-+  VERSION = '0.1.1'
- end
\ No newline at end of file
diff --git a/spec/fixtures/utf-8-diff.txt b/spec/fixtures/utf-8-diff.txt
new file mode 100644
index 0000000..6162d40
--- /dev/null
+++ b/spec/fixtures/utf-8-diff.txt
@@ -0,0 +1,7 @@
+diff --git a/test.txt b/test.txt
+new file mode 100644
+index 0000000..ea49386
+--- /dev/null
++++ b/test.txt
+@@ -0,0 +1 @@
++Привіт, світ