Fix ArgumentError: invalid byte sequence in UTF-8 for non-UTF-8 files

CHANGELOG.md

@@ -1,6 +1,7 @@
 ### 0.5.1 (Next)
 
 * [#45](https://github.com/dblock/fui/pull/45): Migrated from Travis CI to GitHub Actions with danger-pr-comment workflow - [@dblock](https://github.com/dblock).
+* [#37](https://github.com/dblock/fui/issues/37): Fixed `ArgumentError: invalid byte sequence in UTF-8` when processing files with non-UTF-8 encoding - [@dblock](https://github.com/dblock).
 * Your contribution here.
 
 ### 0.5.0 (2018/12/19)

lib/fui/finder.rb

@@ -84,26 +84,23 @@ def global_imported(file_contents, header)
     end
 
     def process_code(references, path)
-      File.open(path) do |file|
-        yield path if block_given?
-        headers.each do |header|
-          filename_without_extension = File.basename(path, File.extname(path))
-          file_contents = File.read(file)
-          global_import_exists = global_imported(file_contents, header)
-          local_import_exists = local_imported(file_contents, header)
-          references[header] << path if filename_without_extension != header.filename_without_extension && (local_import_exists || global_import_exists)
-        end
+      yield path if block_given?
+      file_contents = File.read(path, encoding: 'binary').encode('UTF-8', invalid: :replace, undef: :replace)
+      headers.each do |header|
+        filename_without_extension = File.basename(path, File.extname(path))
+        global_import_exists = global_imported(file_contents, header)
+        local_import_exists = local_imported(file_contents, header)
+        references[header] << path if filename_without_extension != header.filename_without_extension && (local_import_exists || global_import_exists)
       end
     end
 
     def process_xml(references, path)
-      File.open(path) do |file|
-        yield path if block_given?
-        headers.each do |header|
-          filename_without_extension = File.basename(path, File.extname(path))
-          check_xibs = !options['ignore-xib-files']
-          references[header] << path if (check_xibs || filename_without_extension != header.filename_without_extension) && File.read(file).include?("customClass=\"#{header.filename_without_extension}\"")
-        end
+      yield path if block_given?
+      file_contents = File.read(path, encoding: 'binary').encode('UTF-8', invalid: :replace, undef: :replace)
+      headers.each do |header|
+        filename_without_extension = File.basename(path, File.extname(path))
+        check_xibs = !options['ignore-xib-files']
+        references[header] << path if (check_xibs || filename_without_extension != header.filename_without_extension) && file_contents.include?("customClass=\"#{header.filename_without_extension}\"")
       end
     end
   end

spec/fixtures/non_utf8/main.m

@@ -0,0 +1,3 @@
+#import "used_class.h"
+// ÄãºÃ
+- void main() {}

spec/fixtures/non_utf8/used_class.h

@@ -0,0 +1,8 @@
+//
+//  UsedClass.h
+//
+
+#import <UIKit/UIKit.h>
+
+@interface UsedClass
+@end

spec/fui/finder_spec.rb

@@ -179,4 +179,19 @@
       end
     end
   end
+  context 'files with non-UTF-8 encoding' do
+    before :each do
+      @fixtures_dir = File.expand_path(File.join(__FILE__, '../../fixtures/non_utf8'))
+    end
+    describe '#references' do
+      it 'handles non-UTF-8 encoded files without raising an error' do
+        finder = Fui::Finder.new(@fixtures_dir)
+        expect { finder.references }.not_to raise_error
+      end
+      it 'finds references in files with non-UTF-8 encoding' do
+        finder = Fui::Finder.new(@fixtures_dir)
+        expect(Hash[finder.references.map { |k, v| [k.filename, v.count] }]).to eq('used_class.h' => 1)
+      end
+    end
+  end
 end