Properly unescape diff paths

lib/git/diff.rb

@@ -129,8 +129,8 @@ def process_full_diff
         final = {}
         current_file = nil
         @full_diff.split("\n").each do |line|
-          if m = /^diff --git a\/(.*?) b\/(.*?)/.match(line)
-            current_file = m[1]
+          if m = %r{\Adiff --git ("?)a/(.+?)\1 ("?)b/(.+?)\3\z}.match(line)
+            current_file = Git::Lib.unescape_path(m[2])
             final[current_file] = defaults.merge({:patch => line, :path => current_file})
           else
             if m = /^index ([0-9a-f]{4,40})\.\.([0-9a-f]{4,40})( ......)*/.match(line)

lib/git/lib.rb

@@ -1085,7 +1085,8 @@ def command(cmd, *opts, &block)
       global_opts = []
       global_opts << "--git-dir=#{@git_dir}" if !@git_dir.nil?
       global_opts << "--work-tree=#{@git_work_dir}" if !@git_work_dir.nil?
-      global_opts << ["-c", "color.ui=false"]
+      global_opts << %w[-c core.quotePath=true]
+      global_opts << %w[-c color.ui=false]
 
       opts = [opts].flatten.map {|s| escape(s) }.join(' ')
 
@@ -1226,5 +1227,38 @@ def windows_platform?
       RUBY_PLATFORM =~ win_platform_regex || RUBY_DESCRIPTION =~ win_platform_regex
     end
 
+    UNESCAPES = {
+      'a' => 0x07,
+      'b' => 0x08,
+      't' => 0x09,
+      'n' => 0x0a,
+      'v' => 0x0b,
+      'f' => 0x0c,
+      'r' => 0x0d,
+      'e' => 0x1b,
+      "\\" => 0x5c,
+      "\"" => 0x22,
+      "'" => 0x27
+    }
+
+    def self.unescape_path(path)
+      index = 0
+      bytes = []
+      while index < path.length
+        if path[index] == '\\'
+          if '0' <= path[index + 1] && path[index + 1] <= '7'
+            bytes << path[index + 1..index + 4].to_i(8)
+            index += 4
+          elsif UNESCAPES.include?(path[index + 1])
+            bytes << UNESCAPES[path[index + 1]]
+            index += 2
+          end
+        else
+          bytes << path[index].ord
+          index += 1
+        end
+      end
+      bytes.pack('c*').force_encoding(Encoding::UTF_8)
+    end
   end
 end

tests/units/test_archive.rb

@@ -45,7 +45,7 @@ def test_archive
 
     f = @git.object('v2.6').archive(tempfile, :format => 'tar', :prefix => 'test/', :path => 'ex_dir/')
     assert(File.exist?(f))
- 
+
     lines = Minitar::Input.open(f).each.to_a.map(&:full_name)
     assert_match(%r{test/}, lines[1])
     assert_match(%r{test/ex_dir/ex\.txt}, lines[3])

tests/units/test_diff_with_escaped_path.rb

@@ -0,0 +1,22 @@
+#!/usr/bin/env ruby
+# encoding: utf-8
+
+require File.dirname(__FILE__) + '/../test_helper'
+
+# Test diff when the file path has to be quoted according to core.quotePath
+# See https://git-scm.com/docs/git-config#Documentation/git-config.txt-corequotePath
+#
+class TestDiffWithEscapedPath < Test::Unit::TestCase
+  def test_diff_with_non_ascii_filename
+    in_temp_dir do |path|
+      create_file('my_other_file_☠', "First Line\n")
+      `git init`
+      `git add .`
+      `git config --local core.safecrlf false` if Gem.win_platform?
+      `git commit -m "First Commit"`
+      update_file('my_other_file_☠', "Second Line\n")
+      diff_paths = Git.open('.').diff.map(&:path)
+      assert_equal(["my_other_file_☠"], diff_paths)
+    end
+  end
+end

tests/units/test_logger.rb

@@ -7,32 +7,49 @@ class TestLogger < Test::Unit::TestCase
   def setup
     set_file_paths
   end
-
+
+  def missing_log_entry
+    'Did not find expected log entry.'
+  end
+
+  def unexpected_log_entry
+    'Unexpected log entry found'
+  end
+
   def test_logger
     log = Tempfile.new('logfile')
     log.close
-    
+
     logger = Logger.new(log.path)
     logger.level = Logger::DEBUG
-    
+
     @git = Git.open(@wdir, :log => logger)
     @git.branches.size
-    
+
     logc = File.read(log.path)
-    assert(/INFO -- : git ['"]--git-dir=[^'"]+['"] ['"]--work-tree=[^'"]+['"] ['"]-c['"] ['"]color.ui=false['"] branch ['"]-a['"]/.match(logc))
-    assert(/DEBUG -- :   cherry\n  diff_over_patches\n\* git_grep/m.match(logc))
 
+    expected_log_entry = /INFO -- : git (?<global_options>.*?) branch ['"]-a['"]/
+    assert_match(expected_log_entry, logc, missing_log_entry)
+
+    expected_log_entry = /DEBUG -- :   cherry/
+    assert_match(expected_log_entry, logc, missing_log_entry)
+  end
+
+  def test_logging_at_info_level_should_not_show_debug_messages
     log = Tempfile.new('logfile')
     log.close
     logger = Logger.new(log.path)
     logger.level = Logger::INFO
-    
+
     @git = Git.open(@wdir, :log => logger)
     @git.branches.size
-    
+
     logc = File.read(log.path)
-    assert(/INFO -- : git ['"]--git-dir=[^'"]+['"] ['"]--work-tree=[^'"]+['"] ['"]-c['"] ['"]color.ui=false['"] branch ['"]-a['"]/.match(logc))
-    assert(!/DEBUG -- :   cherry\n  diff_over_patches\n\* git_grep/m.match(logc))
+
+    expected_log_entry = /INFO -- : git (?<global_options>.*?) branch ['"]-a['"]/
+    assert_match(expected_log_entry, logc, missing_log_entry)
+
+    expected_log_entry = /DEBUG -- :   cherry/
+    assert_not_match(expected_log_entry, logc, unexpected_log_entry)
   end
-
 end

tests/units/test_unescape_path.rb

@@ -0,0 +1,35 @@
+#!/usr/bin/env ruby
+# encoding: utf-8
+
+require File.dirname(__FILE__) + '/../test_helper'
+
+# Test diff when the file path has to be quoted according to core.quotePath
+# See https://git-scm.com/docs/git-config#Documentation/git-config.txt-corequotePath
+# See https://www.jvt.me/posts/2020/06/23/byte-array-to-string-ruby/
+#
+class TestUnescapePath < Test::Unit::TestCase
+  def test_simple_path
+    path = 'my_other_file'
+    expected_unescaped_path = 'my_other_file'
+    assert_equal(expected_unescaped_path, Git::Lib.unescape_path(path))
+  end
+
+  def test_unicode_path
+    path = 'my_other_file_\\342\\230\\240'
+    expected_unescaped_path = 'my_other_file_☠'
+    assert_equal(expected_unescaped_path, Git::Lib.unescape_path(path))
+  end
+
+  def test_single_char_escapes
+    Git::Lib::UNESCAPES.each_pair do |escape_char, expected_char|
+      path = "\\#{escape_char}"
+      assert_equal(expected_char.chr, Git::Lib.unescape_path(path))
+    end
+  end
+
+  def test_compound_escape
+    path = "my_other_file_\\\"\\342\\230\\240\\n\\\""
+    expected_unescaped_path = 'my_other_file_"☠' + "\n" + '"'
+    assert_equal(expected_unescaped_path, Git::Lib.unescape_path(path))
+  end
+end