Fix Git::Lib#commit_data for GPG-signed commits

lib/git/lib.rb

@@ -221,54 +221,57 @@ def object_size(sha)
     def commit_data(sha)
       sha = sha.to_s
       cdata = command_lines('cat-file', 'commit', sha)
-      process_commit_data(cdata, sha, 0)
+      process_commit_data(cdata, sha)
     end
 
-    def process_commit_data(data, sha = nil, indent = 4)
+    def process_commit_data(data, sha)
       hsh = {
-        'sha'     => sha,
-        'message' => '',
-        'parent'  => []
+        'sha'    => sha,
+        'parent' => []
       }
 
-      loop do
-        key, *value = data.shift.split
-
-        break if key.nil?
-
+      each_cat_file_header(data) do |key, value|
         if key == 'parent'
-          hsh['parent'] << value.join(' ')
+          hsh['parent'] << value
         else
-          hsh[key] = value.join(' ')
+          hsh[key] = value
         end
       end
 
-      hsh['message'] = data.collect {|line| line[indent..-1]}.join("\n") + "\n"
+      hsh['message'] = data.join("\n") + "\n"
 
       return hsh
     end
 
+    CAT_FILE_HEADER_LINE = /\A(?<key>\w+) (?<value>.*)\z/
+
+    def each_cat_file_header(data)
+      while (match = CAT_FILE_HEADER_LINE.match(data.shift))
+        key = match[:key]
+        value_lines = [match[:value]]
+
+        while data.first.start_with?(' ')
+          value_lines << data.shift.lstrip
+        end
+
+        yield key, value_lines.join("\n")
+      end
+    end
+
     def tag_data(name)
       sha = sha.to_s
       tdata = command_lines('cat-file', 'tag', name)
-      process_tag_data(tdata, name, 0)
+      process_tag_data(tdata, name)
     end
 
-    def process_tag_data(data, name, indent=4)
-      hsh = {
-        'name'    => name,
-        'message' => ''
-      }
-
-      loop do
-        key, *value = data.shift.split
-
-        break if key.nil?
+    def process_tag_data(data, name)
+      hsh = { 'name' => name }
 
-        hsh[key] = value.join(' ')
+      each_cat_file_header(data) do |key, value|
+        hsh[key] = value
       end
 
-      hsh['message'] = data.collect {|line| line[indent..-1]}.join("\n") + "\n"
+      hsh['message'] = data.join("\n") + "\n"
 
       return hsh
     end

tests/files/signed_commits/dot_git/HEAD

@@ -0,0 +1 @@
+ref: refs/heads/main

tests/files/signed_commits/dot_git/config

@@ -0,0 +1,7 @@
+[core]
+	repositoryformatversion = 0
+	filemode = true
+	bare = false
+	logallrefupdates = true
+	ignorecase = true
+	precomposeunicode = true

tests/files/signed_commits/dot_git/logs/HEAD

@@ -0,0 +1 @@
+0000000000000000000000000000000000000000 a043c677c93d9f2b285771a29742cc73715e41ea Simon Coffey <simon.coffey@futurelearn.com> 1673868871 +0000	commit (initial): Signed commit

tests/files/signed_commits/dot_git/logs/refs/heads/main

@@ -0,0 +1 @@
+0000000000000000000000000000000000000000 a043c677c93d9f2b285771a29742cc73715e41ea Simon Coffey <simon.coffey@futurelearn.com> 1673868871 +0000	commit (initial): Signed commit

tests/files/signed_commits/dot_git/refs/heads/main

@@ -0,0 +1 @@
+a043c677c93d9f2b285771a29742cc73715e41ea

tests/files/signed_commits/notes.txt

@@ -0,0 +1 @@
+it is very important that changes to this file are verified

tests/units/test_signed_commits.rb

@@ -0,0 +1,58 @@
+#!/usr/bin/env ruby
+
+require File.dirname(__FILE__) + '/../test_helper'
+require "fileutils"
+
+class TestSignedCommits < Test::Unit::TestCase
+  def git_working_dir
+    cwd = FileUtils.pwd
+    if File.directory?(File.join(cwd, 'files'))
+      test_dir = File.join(cwd, 'files')
+    elsif File.directory?(File.join(cwd, '..', 'files'))
+      test_dir = File.join(cwd, '..', 'files')
+    elsif File.directory?(File.join(cwd, 'tests', 'files'))
+      test_dir = File.join(cwd, 'tests', 'files')
+    end
+
+    create_temp_repo(File.expand_path(File.join(test_dir, 'signed_commits')))
+  end
+
+  def create_temp_repo(clone_path)
+    filename = 'git_test' + Time.now.to_i.to_s + rand(300).to_s.rjust(3, '0')
+    @tmp_path = File.join("/tmp/", filename)
+    FileUtils.mkdir_p(@tmp_path)
+    FileUtils.cp_r(clone_path, @tmp_path)
+    tmp_path = File.join(@tmp_path, File.basename(clone_path))
+    Dir.chdir(tmp_path) do
+      FileUtils.mv('dot_git', '.git')
+    end
+    tmp_path
+  end
+
+  def setup
+    @lib = Git.open(git_working_dir).lib
+  end
+
+  def test_commit_data
+    data = @lib.commit_data('a043c677c93d9f2b')
+
+    assert_equal('Simon Coffey <simon.coffey@futurelearn.com> 1673868871 +0000', data['author'])
+    assert_equal('92fd5b7c1aeb6a4e2602799f01608b3deebbad2d', data['tree'])
+    assert_equal(<<~EOS.chomp, data['gpgsig'])
+      -----BEGIN PGP SIGNATURE-----
+
+      iHUEABYKAB0WIQRmiEtd91BkbBpcgV2yCJ+VnJz/iQUCY8U2cgAKCRCyCJ+VnJz/
+      ibjyAP48dGdoFgWL2BjV3CnmebdVjEjTCQtF2QGUybJsyJhhcwEAwbzAAGt3YHfS
+      uuLNH9ki9Sqd+/CH+L8Q2dPM5F4l3gg=
+      =3ATn
+      -----END PGP SIGNATURE-----
+    EOS
+    assert_equal(<<~EOS, data['message'])
+      Signed commit
+
+      This will allow me to test commit data extraction for signed commits.
+      I'm making the message multiline to make sure that message extraction is
+      preserved.
+    EOS
+  end
+end