From effd741ab029932a7a1061f36e3c02c1082a361b Mon Sep 17 00:00:00 2001
From: Michael Hashizume <michael.hashizume@puppet.com>
Date: Wed, 1 Jun 2022 14:14:37 -0700
Subject: [PATCH] (FACT-3113) Encode strings as UTF-8 when reading files

Currently, when Facter reads a file with non-UTF-8 characters, it
will crash. This commit explicitly uses the IO class's `:encoding`
option when reading files.
---
 lib/facter/util/file_helper.rb       |  4 ++--
 spec/facter/util/file_helper_spec.rb | 12 ++++++++----
 2 files changed, 10 insertions(+), 6 deletions(-)

diff --git a/lib/facter/util/file_helper.rb b/lib/facter/util/file_helper.rb
index 20f08ef561..6401a45446 100644
--- a/lib/facter/util/file_helper.rb
+++ b/lib/facter/util/file_helper.rb
@@ -9,14 +9,14 @@ class << self
         DEBUG_MESSAGE = 'File at: %s is not accessible.'
 
         def safe_read(path, default_return = '')
-          return File.read(path) if File.readable?(path)
+          return File.read(path, encoding: Encoding::UTF_8) if File.readable?(path)
 
           log_failed_to_read(path)
           default_return
         end
 
         def safe_readlines(path, default_return = [])
-          return File.readlines(path) if File.readable?(path)
+          return File.readlines(path, encoding: Encoding::UTF_8) if File.readable?(path)
 
           log_failed_to_read(path)
           default_return
diff --git a/spec/facter/util/file_helper_spec.rb b/spec/facter/util/file_helper_spec.rb
index fb1e4d8487..325abf98e5 100644
--- a/spec/facter/util/file_helper_spec.rb
+++ b/spec/facter/util/file_helper_spec.rb
@@ -32,7 +32,7 @@
 
   describe '#safe_read' do
     before do
-      allow(File).to receive(:read).with(path).and_return(content)
+      allow(File).to receive(:read).with(path, anything).and_return(content)
     end
 
     context 'when successfully read the file content' do
@@ -42,6 +42,10 @@
         expect(file_helper.safe_read(path)).to eq(content)
       end
 
+      it 'returns the file content as UTF-8' do
+        expect(file_helper.safe_read(path).encoding.name).to eq('UTF-8')
+      end
+
       it 'File.readable? is called with the correct path' do
         file_helper.safe_read(path)
 
@@ -51,7 +55,7 @@
       it 'File.read is called with the correct path' do
         file_helper.safe_read(path)
 
-        expect(File).to have_received(:read).with(path)
+        expect(File).to have_received(:read).with(path, anything)
       end
 
       it "doesn't log anything" do
@@ -129,7 +133,7 @@
 
   describe '#safe_read_lines' do
     before do
-      allow(File).to receive(:readlines).with(path).and_return(array_content)
+      allow(File).to receive(:readlines).with(path, anything).and_return(array_content)
     end
 
     context 'when successfully read the file lines' do
@@ -148,7 +152,7 @@
       it 'File.readlines is called with the correct path' do
         file_helper.safe_readlines(path)
 
-        expect(File).to have_received(:readlines).with(path)
+        expect(File).to have_received(:readlines).with(path, anything)
       end
 
       it "doesn't log anything" do