ohloh_scm 2.5.1 → 3.0.1

data/ohloh_scm.gemspec

@@ -1,22 +1,25 @@
-$:.push File.expand_path("../lib", __FILE__)
+# frozen_string_literal: true
+
+$LOAD_PATH << File.expand_path('lib', __dir__)
 require 'ohloh_scm/version'
 
 Gem::Specification.new do |gem|
   gem.name          = 'ohloh_scm'
   gem.version       = OhlohScm::Version::STRING
-  gem.authors       = ["BlackDuck Software"]
-  gem.email         = ["info@openhub.net"]
-  gem.summary       = %[Source Control Management]
-  gem.description   = %[The Ohloh source control management library for interacting with Git, SVN, CVS, Hg and Bzr repositories.]
-  gem.homepage      = %[https://github.com/blackducksw/ohloh_scm/]
-  gem.license       = %[GPL-2.0]
-
-  gem.files         = `git ls-files -z`.force_encoding('utf-8').split("\x0")
-  gem.executables   = gem.files.grep(%r{^bin/}).map { |f| File.basename(f) }
-  gem.test_files    = gem.files.grep(%r{^(test|spec|features)/})
-  gem.require_paths = %w(lib)
+  gem.authors       = ['OpenHub Team at Synopsys']
+  gem.email         = ['info@openhub.net']
+  gem.summary       = 'Source Control Management'
+  gem.description   = 'The OpenHub source control management library for \
+                       interacting with Git, SVN, CVS, Hg and Bzr repositories.'
+  gem.homepage      = 'https://github.com/blackducksoftware/ohloh_scm/'
+  gem.license       = 'GPL-2.0'
 
-  gem.add_runtime_dependency 'posix-spawn', '~> 0.3'
-  gem.add_runtime_dependency 'nokogiri', '~> 1.8', '>= 1.8.1'
-  gem.add_runtime_dependency 'test-unit', '~> 3.2', '>= 3.2.7'
+  gem.files         = `git ls-files -z`.split("\x0")
+  gem.test_files    = gem.files.grep(/^spec\//)
+  gem.require_paths = %w[lib]
+  gem.post_install_message = "Ohloh SCM is depending on Git #{OhlohScm::Version::GIT}, "\
+                             "SVN #{OhlohScm::Version::SVN}, CVSNT #{OhlohScm::Version::CVSNT}, "\
+                             "Mercurial #{OhlohScm::Version::HG} and Bazaar "\
+                             "#{OhlohScm::Version::BZR}. If the installed version is different, "\
+                             'Ohloh SCM may not operate as expected.'
 end

data/spec/.rubocop.yml

@@ -0,0 +1,22 @@
+inherit_from: ../.rubocop.yml
+
+Style/FrozenStringLiteralComment:
+  Enabled: false
+
+Metrics/LineLength:
+  Max: 120
+
+Metrics/AbcSize:
+  Enabled: false
+
+ClassLength:
+  Enabled: false
+
+AbcSize:
+  Enabled: false
+
+Metrics/MethodLength:
+  Enabled: false
+
+Metrics/BlockLength:
+  Enabled: false

data/spec/benchmarks/hg_bzr_bash_vs_py_api.rb

@@ -0,0 +1,39 @@
+# NOTE: Setup before running benchmark. Replace hg with bzr in the following commands for bzr benchmarks.
+# cd spec/scm_fixtures
+# tar xf hg_large.tgz && cd hg_large
+# bash create-large-file.sh 14 # Create a file as per given factor # 14=19MB # 15=38MB # 16=76MB # 17=151MB
+# hg add && hg commit -m 'temp'
+# cd ../../../
+# ruby -I lib spec/benchmarks/hg_bzr_bash_vs_py_api.rb hg
+# hg rollback  # bzr uncommit # revert last commit to try different file sizes.
+
+require_relative '../../lib/ohloh_scm'
+require 'benchmark'
+
+OhlohScm::System.logger.level = :error
+
+scm = ARGV[0] || :hg
+repo_path = File.expand_path("../scm_fixtures/#{scm}_large", __dir__)
+
+puts 'Benchmarks for `cat_file`'
+
+activity = OhlohScm::Factory.get_core(scm_type: scm, url: repo_path).activity
+commit = OhlohScm::Commit.new(token: '1')
+diff = OhlohScm::Diff.new(path: 'large.php')
+
+puts `du -sh #{repo_path}/large.php`
+
+Benchmark.bmbm 20 do |reporter|
+  reporter.report("#{scm}[bash api]  ") do
+    if scm.to_s == 'hg'
+      `cd #{activity.url} && hg cat -r #{commit.token} #{diff.path}`
+    else
+      token = commit.token.to_i + 1
+      `cd #{activity.url} && bzr cat --name-from-revision -r #{token} '#{diff.path}'`
+    end
+  end
+
+  reporter.report("#{scm}[python api]") do
+    activity.cat_file(commit, diff)
+  end
+end

data/spec/benchmarks/process_spawn_benchmark.rb

@@ -0,0 +1,133 @@
+# Usage: ruby process_spawn_benchmark.rb [-n <count>] [-m <mem-size>]
+# Run posix-spawn (Ruby extension) benchmarks and report to standard output.
+#
+# Options:
+#   -n, --count=NUM         total number of processes to spawn.
+#   -m, --mem-size=MB       RES size to bloat to before performing benchmarks.
+#
+# Benchmarks run with -n 500 -m 100 by default.
+require 'optparse'
+require 'posix-spawn'
+require 'benchmark'
+require 'digest/sha1'
+require 'open3'
+
+class ProcessSpawnBenchmark
+  include Benchmark
+
+  def initialize(allocate, iterations)
+    bloat_main_process_memory(allocate)
+    repo_path = setup_test_repository
+    @cmd = "cd #{repo_path} && git log"
+    @iterations = iterations
+    benchmark_all
+  end
+
+  private
+
+  def bloat_main_process_memory(allocate)
+    _ = 'x' * allocate
+    memory_used = `ps aux | grep #{File.basename(__FILE__)} | grep -v grep | awk '{print $6/1000}'`.strip.to_i
+    puts "Parent process: #{memory_used}MB RESidual memory"
+  end
+
+  def setup_test_repository
+    repo_name = 'git'
+    repo_path = File.expand_path("../scm_fixtures/#{repo_name}.tgz", __dir__)
+    dest_path = "/tmp/#{repo_name}"
+    system("rm -rf #{dest_path} && tar xvf #{repo_path} -C /tmp/ > /dev/null")
+    dest_path
+  end
+
+  def benchmark_all
+    puts "Benchmarking: #{@iterations} child processes spawned for each"
+    bmbm 40 do |reporter|
+      @reporter = reporter
+      benchmark_open3
+      benchmark_posix_spawn_child
+      benchmark_posix_spawn
+      benchmark_popen
+      benchmark_system
+      benchmark_spawn if Process.respond_to?(:spawn)
+      benchmark_fork_exec
+    end
+  end
+
+  def benchmark_open3
+    @reporter.report('Open3.capture3 => ') do
+      @iterations.times do
+        stdout, = Open3.capture3(@cmd)
+        verify_output(stdout, __method__)
+      end
+    end
+  end
+
+  def benchmark_posix_spawn_child
+    @reporter.report('POSIX::Spawn::Child => ') do
+      @iterations.times do
+        child = POSIX::Spawn::Child.new(@cmd)
+        verify_output(child.out, __method__)
+      end
+    end
+  end
+
+  def benchmark_posix_spawn
+    @reporter.report('pspawn (posix_spawn) => ') do
+      @iterations.times do
+        pid = POSIX::Spawn.pspawn("#{@cmd} > /dev/null")
+        Process.wait(pid)
+      end
+    end
+  end
+
+  def benchmark_popen
+    @reporter.report('IO.popen:') do
+      @iterations.times do
+        IO.popen(@cmd).each {}
+      end
+    end
+  end
+
+  def benchmark_system
+    @reporter.report('``:') do
+      @iterations.times do
+        stdout = `#{@cmd}`
+        verify_output(stdout, __method__)
+      end
+    end
+  end
+
+  def benchmark_spawn
+    @reporter.report('spawn (native):') do
+      @iterations.times do
+        pid = Process.spawn("#{@cmd} > /dev/null")
+        Process.wait(pid)
+      end
+    end
+  end
+
+  def benchmark_fork_exec
+    @reporter.report('fspawn (fork/exec):') do
+      @iterations.times do
+        pid = POSIX::Spawn.fspawn("#{@cmd} > /dev/null")
+        Process.wait(pid)
+      end
+    end
+  end
+
+  def verify_output(stdout, method_name)
+    return if Digest::SHA1.hexdigest(stdout) == 'df31df68785baa8725e40e2a2583bb8f7e9dd3c5'
+
+    raise "Git log output did not match for #{method_name.slice(/benchmark_(.+)$/, 1)}"
+  end
+end
+
+allocate   = 100 * (1024**2)
+iterations = 500
+ARGV.options do |o|
+  o.on('-n', '--count=num')   { |val| iterations = val.to_i }
+  o.on('-m', '--mem-size=MB') { |val| allocate   = val.to_i * (1024**2) }
+  o.parse!
+end
+
+ProcessSpawnBenchmark.new(allocate, iterations)

data/spec/helpers/assert_scm_attr_helper.rb

@@ -0,0 +1,31 @@
+# frozen_string_literal: true
+
+module AssertScmAttrHelper
+  def get_core(scm_type, opts = {})
+    OhlohScm::Factory.get_core({ scm_type: scm_type, url: 'foobar' }.merge(opts))
+  end
+
+  def assert_url_error(scm_type, *urls)
+    urls.each do |url|
+      core = get_core(scm_type, url: url)
+      core.validation.send(:url_errors).wont_be :empty?
+    end
+  end
+
+  def assert_url_valid(scm_type, url)
+    core = get_core(scm_type, url: url)
+    core.validation.send(:url_errors).must_be_nil
+  end
+
+  def assert_branch_name_error(scm_type, *branches)
+    branches.each do |branch_name|
+      core = get_core(scm_type, url: ':pserver:cvs:cvs@cvs.test.org:/test', branch_name: branch_name)
+      core.validation.send(:branch_name_errors).wont_be :empty?
+    end
+  end
+
+  def assert_branch_name_valid(scm_type, branch_name)
+    core = get_core(scm_type, url: ':pserver:cvs:cvs@cvs.test.org:/test', branch_name: branch_name)
+    core.validation.send(:branch_name_errors).must_be_nil
+  end
+end

data/spec/helpers/commit_tokens_helper.rb

@@ -0,0 +1,26 @@
+class CommitTokensHelper
+  def initialize(core, commit_labels, trunk_only: false)
+    @core = core
+    @trunk_only = trunk_only
+    @commit_labels = commit_labels
+  end
+
+  def between(from, to)
+    to_labels(@core.activity.commit_tokens(after: from_label(from), up_to: from_label(to),
+                                           trunk_only: @trunk_only))
+  end
+
+  private
+
+  def to_label(sha1)
+    @commit_labels.invert[sha1.to_s]
+  end
+
+  def to_labels(sha1s)
+    sha1s.map { |sha1| to_label(sha1) }
+  end
+
+  def from_label(label)
+    @commit_labels[label]
+  end
+end

data/spec/helpers/generic_helper.rb

@@ -0,0 +1,5 @@
+module GenericHelper
+  def tmpdir(prefix = 'oh_scm_repo_')
+    Dir.mktmpdir(prefix) { |path| yield path }
+  end
+end

data/spec/helpers/repository_helper.rb

@@ -0,0 +1,36 @@
+# frozen_string_literal: true
+
+module RepositoryHelper
+  %w[git svn git_svn cvs hg bzr].each do |scm_type|
+    define_method("with_#{scm_type}_repository") do |name, branch_name = nil, &block|
+      with_repository(scm_type, name, branch_name) { |core| block.call(core) }
+    end
+  end
+
+  private
+
+  def with_repository(scm_type, name, branch_name = nil)
+    source_path = get_fixture_folder_path(name)
+    Dir.mktmpdir('oh_scm_fixture_') do |dir_path|
+      setup_repository_archive(source_path, dir_path)
+      path_prefix = scm_type == 'svn' ? 'file://' : ''
+      yield OhlohScm::Factory.get_core(scm_type: scm_type, url: "#{path_prefix}#{File.join(dir_path, name)}",
+                                       branch_name: branch_name)
+    end
+  end
+
+  def setup_repository_archive(source_path, dir_path)
+    if Dir.exist?(source_path)
+      `cp -R #{source_path} #{dir_path}`
+    elsif File.exist?("#{source_path}.tgz")
+      `tar xzf #{source_path}.tgz --directory #{dir_path}`
+    else
+      raise "Repository archive #{source_path} not found."
+    end
+  end
+
+  def get_fixture_folder_path(name)
+    fixture_dir = File.expand_path('../scm_fixtures', __dir__)
+    File.join(fixture_dir, name)
+  end
+end

data/spec/helpers/system_helper.rb

@@ -0,0 +1,25 @@
+# We intend to keep System functions protected.
+class SystemPrivateAccessor
+  extend OhlohScm::System
+
+  class << self
+    def run_private(cmd)
+      run(cmd)
+    end
+
+    def run_with_error_private(cmd)
+      run_with_err(cmd)
+    end
+  end
+end
+
+module SystemHelper
+  # Cannot use the name `run` since it conflicts with Minitest#run.
+  def run_p(cmd)
+    SystemPrivateAccessor.run_private(cmd)
+  end
+
+  def run_with_error_p(cmd)
+    SystemPrivateAccessor.run_with_error_private(cmd)
+  end
+end

data/spec/ohloh_scm/bzr/activity_spec.rb

@@ -0,0 +1,460 @@
+# frozen_string_literal: true
+
+require 'spec_helper'
+
+describe 'Bzr::Activity' do
+  it 'must export contents of a repository' do
+    with_bzr_repository('bzr') do |bzr|
+      tmpdir do |dir|
+        bzr.activity.export(dir)
+        entries = ['.', '..', 'Cédric.txt', 'file1.txt', 'file3.txt', 'file4.txt', 'file5.txt']
+        Dir.entries(dir).sort.must_equal entries
+      end
+    end
+  end
+
+  it 'must return head and parents correctly' do
+    with_bzr_repository('bzr') do |bzr|
+      activity = bzr.activity
+
+      activity.head_token.must_equal 'test@example.com-20111222183733-y91if5npo3pe8ifs'
+      activity.head.token.must_equal 'test@example.com-20111222183733-y91if5npo3pe8ifs'
+      assert activity.head.diffs.any? # diffs should be populated
+
+      activity.parents(activity.head).first.token.must_equal 'obnox@samba.org-20090204004942-73rnw0izen42f154'
+      assert activity.parents(activity.head).first.diffs.any?
+    end
+  end
+
+  it 'must return file contents' do
+    with_bzr_repository('bzr') do |bzr|
+      expected = <<-EXPECTED.gsub(/ {8}/, '')
+        first file
+        second line
+      EXPECTED
+
+      commit = OhlohScm::Commit.new(token: 6)
+      diff = OhlohScm::Diff.new(path: 'file1.txt')
+      bzr.activity.cat_file(commit, diff).must_equal expected
+
+      # file2.txt has been removed in commit #5
+      diff2 = OhlohScm::Diff.new(path: 'file2.txt')
+      bzr.activity.cat_file(bzr.activity.commits.last, diff2).must_be_nil
+    end
+  end
+
+  it 'cat_file must work with non-ascii name' do
+    with_bzr_repository('bzr') do |bzr|
+      expected = <<-EXPECTED.gsub(/ {8}/, '')
+        first file
+        second line
+      EXPECTED
+
+      commit = OhlohScm::Commit.new(token: 7)
+      diff = OhlohScm::Diff.new(path: 'Cédric.txt')
+      bzr.activity.cat_file(commit, diff).must_equal expected
+    end
+  end
+
+  it 'must get file contents by parent rev' do
+    with_bzr_repository('bzr') do |bzr|
+      expected = <<-EXPECTED.gsub(/ {8}/, '')
+        first file
+        second line
+      EXPECTED
+
+      commit = OhlohScm::Commit.new(token: 6)
+      diff = OhlohScm::Diff.new(path: 'file1.txt')
+      bzr.activity.cat_file_parent(commit, diff).must_equal expected
+
+      # file2.txt has been removed in commit #5
+      expected = <<-EXPECTED.gsub(/ {8}/, '')
+        another file
+      EXPECTED
+
+      commit2 = OhlohScm::Commit.new(token: 5)
+      diff2 = OhlohScm::Diff.new(path: 'file2.txt')
+      bzr.activity.cat_file_parent(commit2, diff2).must_equal expected
+    end
+  end
+
+  describe 'commits' do
+    it 'must test_commit_count' do
+      with_bzr_repository('bzr') do |bzr|
+        bzr.activity.commit_count.must_equal 7
+        bzr.activity.commit_count(after: revision_ids.first).must_equal 6
+        bzr.activity.commit_count(after: revision_ids[5]).must_equal 1
+        bzr.activity.commit_count(after: revision_ids.last).must_equal 0
+      end
+    end
+
+    it 'must test_commit_count_with_branches' do
+      with_bzr_repository('bzr_with_branch') do |bzr|
+        # Only 3 commits are on main line... make sure we catch the branch commit as well
+        bzr.activity.commit_count.must_equal 4
+      end
+    end
+
+    it 'must test_commit_count_after_merge' do
+      with_bzr_repository('bzr_with_branch') do |bzr|
+        last_commit = bzr.activity.commits.last
+        assert_equal 0, bzr.activity.commit_count(after: last_commit.token)
+      end
+    end
+
+    it 'must test_commit_count_trunk_only' do
+      with_bzr_repository('bzr_with_branch') do |bzr|
+        # Only 3 commits are on main line
+        bzr.activity.commit_count(trunk_only: true).must_equal 3
+      end
+    end
+
+    it 'must test_commit_tokens_after' do
+      with_bzr_repository('bzr') do |bzr|
+        bzr.activity.commit_tokens.must_equal revision_ids
+        bzr.activity.commit_tokens(after: revision_ids.first).must_equal revision_ids[1..6]
+        bzr.activity.commit_tokens(after: revision_ids[5]).must_equal revision_ids[6..6]
+        bzr.activity.commit_tokens(after: revision_ids.last).must_equal []
+      end
+    end
+
+    it 'must test_commit_tokens_after_merge' do
+      with_bzr_repository('bzr_with_branch') do |bzr|
+        last_commit = bzr.activity.commits.last
+        assert_equal [], bzr.activity.commit_tokens(after: last_commit.token)
+      end
+    end
+
+    it 'must test_commit_tokens_after_nested_merge' do
+      with_bzr_repository('bzr_with_nested_branches') do |bzr|
+        last_commit = bzr.activity.commits.last
+        assert_equal [], bzr.activity.commit_tokens(after: last_commit.token)
+      end
+    end
+
+    it 'must test_commit_tokens_trunk_only_false' do
+      # Funny business with commit ordering has been fixed by BzrXmlParser.
+      # Now we always see branch commits before merge commit.
+      with_bzr_repository('bzr_with_branch') do |bzr|
+        expected = ['test@example.com-20090206214301-s93cethy9atcqu9h',
+                    'test@example.com-20090206214451-lzjngefdyw3vmgms',
+                    'test@example.com-20090206214350-rqhdpz92l11eoq2t', # branch commit
+                    'test@example.com-20090206214515-21lkfj3dbocao5pr']  # merge commit
+        bzr.activity.commit_tokens.must_equal expected
+      end
+    end
+
+    it 'must test_commit_tokens_trunk_only_true' do
+      with_bzr_repository('bzr_with_branch') do |bzr|
+        expected = ['test@example.com-20090206214301-s93cethy9atcqu9h',
+                    'test@example.com-20090206214451-lzjngefdyw3vmgms',
+                    'test@example.com-20090206214515-21lkfj3dbocao5pr']  # merge commit
+        bzr.activity.commit_tokens(trunk_only: true).must_equal expected
+      end
+    end
+
+    it 'must test_nested_branches_commit_tokens_trunk_only_false' do
+      with_bzr_repository('bzr_with_nested_branches') do |bzr|
+        expected = ['obnox@samba.org-20090204002342-5r0q4gejk69rk6uv',
+                    'obnox@samba.org-20090204002422-5ylnq8l4713eqfy0',
+                    'obnox@samba.org-20090204002453-u70a3ehf3ae9kay1',
+                    'obnox@samba.org-20090204002518-yb0x153oa6mhoodu',
+                    'obnox@samba.org-20090204002540-gmana8tk5f9gboq9',
+                    'obnox@samba.org-20090204004942-73rnw0izen42f154',
+                    'test@example.com-20110803170302-fz4mbr89n8f5agha',
+                    'test@example.com-20110803170341-v1icvy05b430t68l',
+                    'test@example.com-20110803170504-z7xz5uxj02e5x3z6',
+                    'test@example.com-20110803170522-asv6i9z6m22jc8zz',
+                    'test@example.com-20110803170648-o0xcbni7lwp97azj',
+                    'test@example.com-20110803170818-v44umypquqg8migo']
+        bzr.activity.commit_tokens.must_equal expected
+      end
+    end
+
+    it 'must test_nested_branches_commit_tokens_trunk_only_true' do
+      with_bzr_repository('bzr_with_nested_branches') do |bzr|
+        expected = ['obnox@samba.org-20090204002342-5r0q4gejk69rk6uv',
+                    'obnox@samba.org-20090204002422-5ylnq8l4713eqfy0',
+                    'obnox@samba.org-20090204002453-u70a3ehf3ae9kay1',
+                    'obnox@samba.org-20090204002518-yb0x153oa6mhoodu',
+                    'obnox@samba.org-20090204002540-gmana8tk5f9gboq9',
+                    'obnox@samba.org-20090204004942-73rnw0izen42f154',
+                    'test@example.com-20110803170818-v44umypquqg8migo']
+        bzr.activity.commit_tokens(trunk_only: true).must_equal expected
+      end
+    end
+
+    it 'must test_commits_trunk_only_false' do
+      with_bzr_repository('bzr_with_branch') do |bzr|
+        expected = ['test@example.com-20090206214301-s93cethy9atcqu9h',
+                    'test@example.com-20090206214451-lzjngefdyw3vmgms',
+                    'test@example.com-20090206214350-rqhdpz92l11eoq2t', # branch commit
+                    'test@example.com-20090206214515-21lkfj3dbocao5pr']  # merge commit
+
+        bzr.activity.commits.map(&:token).must_equal expected
+      end
+    end
+
+    it 'must test_commits_trunk_only_true' do
+      with_bzr_repository('bzr_with_branch') do |bzr|
+        expected = ['test@example.com-20090206214301-s93cethy9atcqu9h',
+                    'test@example.com-20090206214451-lzjngefdyw3vmgms',
+                    'test@example.com-20090206214515-21lkfj3dbocao5pr']  # merge commit
+        bzr.activity.commits(trunk_only: true).map(&:token).must_equal expected
+      end
+    end
+
+    it 'must test_commits_after_merge' do
+      with_bzr_repository('bzr_with_branch') do |bzr|
+        last_commit = bzr.activity.commits.last
+        bzr.activity.commits(after: last_commit.token).must_be :empty?
+      end
+    end
+
+    it 'must test_commits_after_nested_merge' do
+      with_bzr_repository('bzr_with_nested_branches') do |bzr|
+        last_commit = bzr.activity.commits.last
+        bzr.activity.commits(after: last_commit.token).must_be :empty?
+      end
+    end
+
+    it 'must test_nested_branches_commits_trunk_only_false' do
+      with_bzr_repository('bzr_with_nested_branches') do |bzr|
+        expected = ['obnox@samba.org-20090204002342-5r0q4gejk69rk6uv',
+                    'obnox@samba.org-20090204002422-5ylnq8l4713eqfy0',
+                    'obnox@samba.org-20090204002453-u70a3ehf3ae9kay1',
+                    'obnox@samba.org-20090204002518-yb0x153oa6mhoodu',
+                    'obnox@samba.org-20090204002540-gmana8tk5f9gboq9',
+                    'obnox@samba.org-20090204004942-73rnw0izen42f154',
+                    'test@example.com-20110803170302-fz4mbr89n8f5agha',
+                    'test@example.com-20110803170341-v1icvy05b430t68l',
+                    'test@example.com-20110803170504-z7xz5uxj02e5x3z6',
+                    'test@example.com-20110803170522-asv6i9z6m22jc8zz',
+                    'test@example.com-20110803170648-o0xcbni7lwp97azj',
+                    'test@example.com-20110803170818-v44umypquqg8migo']
+        bzr.activity.commits.map(&:token).must_equal expected
+      end
+    end
+
+    it 'must test_nested_branches_commits_trunk_only_true' do
+      with_bzr_repository('bzr_with_nested_branches') do |bzr|
+        expected = ['obnox@samba.org-20090204002342-5r0q4gejk69rk6uv',
+                    'obnox@samba.org-20090204002422-5ylnq8l4713eqfy0',
+                    'obnox@samba.org-20090204002453-u70a3ehf3ae9kay1',
+                    'obnox@samba.org-20090204002518-yb0x153oa6mhoodu',
+                    'obnox@samba.org-20090204002540-gmana8tk5f9gboq9',
+                    'obnox@samba.org-20090204004942-73rnw0izen42f154',
+                    'test@example.com-20110803170818-v44umypquqg8migo']
+        bzr.activity.commits(trunk_only: true).map(&:token).must_equal expected
+      end
+    end
+
+    it 'must test_commits' do
+      with_bzr_repository('bzr') do |bzr|
+        bzr.activity.commits.collect(&:token).must_equal revision_ids
+        bzr.activity.commits(after: revision_ids[5]).collect(&:token).must_equal revision_ids[6..6]
+        bzr.activity.commits(after: revision_ids.last).collect(&:token).must_equal []
+
+        # Check that the diffs are not populated
+        bzr.activity.commits.first.diffs.must_equal []
+      end
+    end
+
+    it 'must test_each_commit' do
+      with_bzr_repository('bzr') do |bzr|
+        commits = []
+        bzr.activity.each_commit do |c|
+          assert c.committer_name
+          assert c.committer_date.is_a?(Time)
+          refute c.message.empty?
+          assert c.diffs.any?
+          # Check that the diffs are populated
+          c.diffs.each do |d|
+            assert d.action =~ /^[MAD]$/
+            refute d.path.empty?
+          end
+          commits << c
+        end
+
+        # Make sure we cleaned up after ourselves
+        assert !FileTest.exist?(bzr.activity.log_filename)
+
+        # Verify that we got the commits in forward chronological order
+        commits.collect(&:token).must_equal revision_ids
+      end
+    end
+
+    it 'must test_each_commit_trunk_only_false' do
+      with_bzr_repository('bzr_with_branch') do |bzr|
+        commits = []
+        bzr.activity.each_commit { |c| commits << c }
+        expected = ['test@example.com-20090206214301-s93cethy9atcqu9h',
+                    'test@example.com-20090206214451-lzjngefdyw3vmgms',
+                    'test@example.com-20090206214350-rqhdpz92l11eoq2t', # branch commit
+                    'test@example.com-20090206214515-21lkfj3dbocao5pr'] # merge commit]
+        commits.map(&:token).must_equal expected
+      end
+    end
+
+    it 'must test_each_commit_trunk_only_true' do
+      with_bzr_repository('bzr_with_branch') do |bzr|
+        commits = []
+        bzr.activity.each_commit(trunk_only: true) { |c| commits << c }
+        expected = [
+          'test@example.com-20090206214301-s93cethy9atcqu9h',
+          'test@example.com-20090206214451-lzjngefdyw3vmgms',
+          'test@example.com-20090206214515-21lkfj3dbocao5pr' # merge commit
+          # 'test@example.com-20090206214350-rqhdpz92l11eoq2t' # branch commit -- after merge!
+        ]
+        commits.map(&:token).must_equal expected
+      end
+    end
+
+    it 'must test_each_commit_after_merge' do
+      with_bzr_repository('bzr_with_branch') do |bzr|
+        last_commit = bzr.activity.commits.last
+
+        commits = []
+        bzr.activity.each_commit(after: last_commit.token) { |c| commits << c }
+        commits.must_equal []
+      end
+    end
+
+    it 'must test_each_commit_after_nested_merge_at_tip' do
+      with_bzr_repository('bzr_with_nested_branches') do |bzr|
+        last_commit = bzr.activity.commits.last
+
+        commits = []
+        bzr.activity.each_commit(after: last_commit.token) { |c| commits << c }
+        commits.must_equal []
+      end
+    end
+
+    it 'must test_each_commit_after_nested_merge_not_at_tip' do
+      with_bzr_repository('bzr_with_nested_branches') do |bzr|
+        last_commit = bzr.activity.commits.last
+        next_to_last_commit = bzr.activity.commits[-2]
+
+        yielded_commits = []
+        bzr.activity.each_commit(after: next_to_last_commit.token) { |c| yielded_commits << c }
+        yielded_commits.map(&:token).must_equal [last_commit.token]
+      end
+    end
+
+    it 'must test_nested_branches_each_commit_trunk_only_false' do
+      with_bzr_repository('bzr_with_nested_branches') do |bzr|
+        commits = []
+        bzr.activity.each_commit { |c| commits << c }
+        expected = ['obnox@samba.org-20090204002342-5r0q4gejk69rk6uv',
+                    'obnox@samba.org-20090204002422-5ylnq8l4713eqfy0',
+                    'obnox@samba.org-20090204002453-u70a3ehf3ae9kay1',
+                    'obnox@samba.org-20090204002518-yb0x153oa6mhoodu',
+                    'obnox@samba.org-20090204002540-gmana8tk5f9gboq9',
+                    'obnox@samba.org-20090204004942-73rnw0izen42f154',
+                    'test@example.com-20110803170302-fz4mbr89n8f5agha',
+                    'test@example.com-20110803170341-v1icvy05b430t68l',
+                    'test@example.com-20110803170504-z7xz5uxj02e5x3z6',
+                    'test@example.com-20110803170522-asv6i9z6m22jc8zz',
+                    'test@example.com-20110803170648-o0xcbni7lwp97azj',
+                    'test@example.com-20110803170818-v44umypquqg8migo']
+        commits.map(&:token).must_equal expected
+      end
+    end
+
+    it 'must test_nested_branches_each_commit_trunk_only_true' do
+      with_bzr_repository('bzr_with_nested_branches') do |bzr|
+        commits = []
+        bzr.activity.each_commit(trunk_only: true) { |c| commits << c }
+        expected = ['obnox@samba.org-20090204002342-5r0q4gejk69rk6uv',
+                    'obnox@samba.org-20090204002422-5ylnq8l4713eqfy0',
+                    'obnox@samba.org-20090204002453-u70a3ehf3ae9kay1',
+                    'obnox@samba.org-20090204002518-yb0x153oa6mhoodu',
+                    'obnox@samba.org-20090204002540-gmana8tk5f9gboq9',
+                    'obnox@samba.org-20090204004942-73rnw0izen42f154',
+                    'test@example.com-20110803170818-v44umypquqg8migo']
+        commits.map(&:token).must_equal expected
+      end
+    end
+
+    # This bzr repository contains the following tree structure
+    #    /foo/
+    #    /foo/helloworld.c
+    #    /bar/
+    # Ohloh doesn't care about directories, so only /foo/helloworld.c should be reported.
+    it 'must test_each_commit_excludes_directories' do
+      with_bzr_repository('bzr_with_subdirectories') do |bzr|
+        commits = []
+        bzr.activity.each_commit do |c|
+          commits << c
+        end
+        commits.size.must_equal 1
+        commits.first.diffs.size.must_equal 1
+        commits.first.diffs.first.path.must_equal 'foo/helloworld.c'
+      end
+    end
+
+    # Verfies OTWO-344
+    it 'must test_commit_tokens_with_colon_character' do
+      with_bzr_repository('bzr_colon') do |bzr|
+        bzr.activity.commit_tokens.must_equal ['svn-v4:364a429a-ab12-11de-804f-e3d9c25ff3d2::0']
+      end
+    end
+
+    it 'must test_committer_and_author_name' do
+      with_bzr_repository('bzr_with_authors') do |bzr|
+        commits = []
+        bzr.activity.each_commit do |c|
+          commits << c
+        end
+        commits.size.must_equal 3
+
+        commits[0].message.must_equal 'Initial.'
+        commits[0].committer_name.must_equal 'Abhay Mujumdar'
+        commits[0].author_name.must_be_nil
+        commits[0].author_email.must_be_nil
+
+        commits[1].message.must_equal 'Updated.'
+        commits[1].committer_name.must_equal 'Abhay Mujumdar'
+        commits[1].author_name.must_equal 'John Doe'
+        commits[1].author_email.must_equal 'johndoe@example.com'
+
+        # When there are multiple authors, first one is captured.
+        commits[2].message.must_equal 'Updated by two authors.'
+        commits[2].committer_name.must_equal 'test'
+        commits[2].author_name.must_equal 'John Doe'
+        commits[2].author_email.must_equal 'johndoe@example.com'
+      end
+    end
+
+    # Bzr converts invalid utf-8 characters into valid format before commit.
+    # So no utf-8 encoding issues are seen in ruby when dealing with Bzr.
+    it 'must test_commits_encoding' do
+      with_bzr_repository('bzr_with_invalid_encoding') do |bzr|
+        assert bzr.activity.commits
+      end
+    end
+
+    protected
+
+    def revision_ids
+      ['obnox@samba.org-20090204002342-5r0q4gejk69rk6uv', # 1
+       'obnox@samba.org-20090204002422-5ylnq8l4713eqfy0', # 2
+       'obnox@samba.org-20090204002453-u70a3ehf3ae9kay1', # 3
+       'obnox@samba.org-20090204002518-yb0x153oa6mhoodu', # 4
+       'obnox@samba.org-20090204002540-gmana8tk5f9gboq9', # 5
+       'obnox@samba.org-20090204004942-73rnw0izen42f154', # 6
+       'test@example.com-20111222183733-y91if5npo3pe8ifs'] # 7
+    end
+  end
+
+  it 'must get the tags correctly' do
+    with_bzr_repository('bzr') do |bzr|
+      time1 = Time.parse('2009-02-04 00:25:40 +0000')
+      time2 = Time.parse('2011-12-22 18:37:33 +0000')
+      time3 = Time.parse('2009-02-04 00:24:22 +0000')
+
+      bzr.activity.tags.must_equal [['v1.0.0', '5', time1],
+                                    ['v2.0.0', '7', time2], ['v 3.0.0', '2', time3]]
+    end
+  end
+end