Refactor name mapping in tools/version.rb (#13637)

* Add get_clean_name function to PackageUtils

* Use PackageUtils.get_clean_name in tools/version.rb and refactor name mapping

lib/const.rb

@@ -4,7 +4,7 @@ require 'etc'
 require 'open3'
 
 OLD_CREW_VERSION = defined?(CREW_VERSION) ? CREW_VERSION : '1.0'
-CREW_VERSION = '1.68.2' unless defined?(CREW_VERSION) && CREW_VERSION == OLD_CREW_VERSION
+CREW_VERSION = '1.68.3' unless defined?(CREW_VERSION) && CREW_VERSION == OLD_CREW_VERSION
 
 # Kernel architecture.
 KERN_ARCH = Etc.uname[:machine]
@@ -397,9 +397,6 @@ CREW_ANITYA_PACKAGE_NAME_MAPPINGS = Set[
   { pkg_name: 'docbook_xml51', anitya_pkg: 'docbook-xml', comments: '' },
   { pkg_name: 'doxygen', anitya_pkg: 'doxygen', comments: '' },
   { pkg_name: 'filecmd', anitya_pkg: 'file', comments: '' },
-  { pkg_name: 'gcc_build', anitya_pkg: 'gcc', comments: '' },
-  { pkg_name: 'gcc_dev', anitya_pkg: 'gcc', comments: '' },
-  { pkg_name: 'gcc_lib', anitya_pkg: 'gcc', comments: '' },
   { pkg_name: 'gcr_3', anitya_pkg: 'gcr~3', comments: '' },
   { pkg_name: 'gcr_4', anitya_pkg: 'gcr', comments: '' },
   { pkg_name: 'gemacs', anitya_pkg: 'emacs', comments: '' },
@@ -482,8 +479,7 @@ CREW_ANITYA_PACKAGE_NAME_MAPPINGS = Set[
   { pkg_name: 'yad', anitya_pkg: 'yad', comments: 'Prefer to GitHub' },
   { pkg_name: 'zig13', anitya_pkg: 'zig', comments: '' },
   { pkg_name: 'zimg', anitya_pkg: 'zimg', comments: 'Prefer to GitHub' },
-  { pkg_name: 'zoneinfo', anitya_pkg: 'tzdata', comments: '' },
-  { pkg_name: 'zstd_static', anitya_pkg: 'zstd', comments: '' }
+  { pkg_name: 'zoneinfo', anitya_pkg: 'tzdata', comments: '' }
 ].to_h { |h| [h[:pkg_name], h[:anitya_pkg]] }
 
 CREW_LICENSE = <<~LICENSESTRING

lib/package_utils.rb

@@ -80,7 +80,7 @@ class PackageUtils
     pkg_version.gsub!(/-dfsg.*/, '')
     # Delete -gcc14, futureproofed until gcc 100
     pkg_version.gsub!(/-gcc\d{2}/, '')
-    # Trim kde- suffixes in qt5 packages so nothing else gets confused.
+    # Trim kde- prefixes in qt5 packages so nothing else gets confused.
     pkg_version.delete_prefix!('kde-')
     # Delete -py3.12, futureproofed until Python 4
     pkg_version.gsub!(/-py3\.\d{2}/, '')
@@ -100,6 +100,23 @@ class PackageUtils
     return pkg_version
   end
 
+  # Remove our language-specific prefixes and any build splitting suffixes.
+  # This is mostly for use when querying Anitya in tools/version.rb, and is not suitable for Repology.
+  def self.get_clean_name(pkg_name)
+    # Delete language-specific prefixes.
+    pkg_name.delete_prefix!('perl_')
+    pkg_name.delete_prefix!('py3_')
+    pkg_name.delete_prefix!('ruby_')
+    # Delete suffixes for split packages.
+    pkg_name.delete_suffix!('_build')
+    pkg_name.delete_suffix!('_dev')
+    pkg_name.delete_suffix!('_lib')
+    # Delete the _static suffix for statically built packages.
+    pkg_name.delete_suffix!('_static')
+
+    return pkg_name
+  end
+
   def self.get_gitlab_pkginfo(pkg_name, pkg_version, pkg_arch, build = nil, verbose = nil)
     # This is largely rehashing self.get_binary_url(pkg) using the
     # curl and jq binaries for debugging purposes.

tests/lib/package_utils.rb

@@ -230,4 +230,26 @@ class PackageUtilsTest < Minitest::Test
     assert_equal('579-4', PackageUtils.get_clean_version('579-4-1628457'))
     assert_equal('2.1.5-20220429', PackageUtils.get_clean_version('2.1.5-20220429'))
   end
+
+  def test_get_clean_perl_name
+    assert_equal('wakeonlan', PackageUtils.get_clean_name('perl_wakeonlan'))
+  end
+
+  def test_get_clean_python_name
+    assert_equal('maturin', PackageUtils.get_clean_name('py3_maturin'))
+  end
+
+  def test_get_clean_ruby_name
+    assert_equal('typeprof', PackageUtils.get_clean_name('ruby_typeprof'))
+  end
+
+  def test_get_clean_split_name
+    assert_equal('gcc', PackageUtils.get_clean_name('gcc_build'))
+    assert_equal('gcc', PackageUtils.get_clean_name('gcc_dev'))
+    assert_equal('gcc', PackageUtils.get_clean_name('gcc_lib'))
+  end
+
+  def test_get_clean_static_name
+    assert_equal('zstd', PackageUtils.get_clean_name('zstd_static'))
+  end
 end

tests/tools/version.rb

@@ -26,6 +26,31 @@ class VersionMonitorTest < Minitest::Test
     assert_nil(get_anitya_id('99_still_notreal', 'ftp://com.crystal', 'Package'))
   end
 
+  def test_get_prefixed_anitya_id
+    assert_equal(2636, get_anitya_id('perl_app_cpanminus', 'https://metacpan.org/pod/App::cpanminus', 'PERL'))
+    # We use the Anitya homepage for maturin here to simplify testing.
+    assert_equal(42653, get_anitya_id('py3_maturin', 'https://www.maturin.rs', 'Python'))
+    # Every ruby_ package has the RUBY buildsystem, so we lie here to test the behavior.
+    # Technically that means we don't need the behavior right now, but things may change.
+    assert_equal(139836, get_anitya_id('ruby_typeprof', 'https://github.com/ruby/typeprof', 'Package'))
+  end
+
+  def test_get_split_anitya_id
+    assert_equal(6502, get_anitya_id('gcc_build', 'https://www.gnu.org/software/gcc/', 'Package'))
+    assert_equal(6502, get_anitya_id('gcc_dev', 'https://www.gnu.org/software/gcc/', 'Package'))
+    assert_equal(6502, get_anitya_id('gcc_lib', 'https://www.gnu.org/software/gcc/', 'Package'))
+  end
+
+  def test_get_static_anitya_id
+    assert_equal(12083, get_anitya_id('zstd_static', 'https://facebook.github.io/zstd/', 'Package'))
+  end
+
+  def test_get_mapped_anitya_id
+    assert_equal(1783, get_anitya_id('libxml2_autotools', 'http://xmlsoft.org/', 'Autotools'))
+    assert_equal(6502, get_anitya_id('libssp', 'https://gcc.gnu.org/', 'Autotools'))
+    assert_equal(4194, get_anitya_id('rest', 'https://wiki.gnome.org/Projects/Librest', 'Meson'))
+  end
+
   def test_get_multi_ecosystem_anitya_id
     # This should eliminate all the other candidates due to their ecosystems, so the homepage isn't even checked.
     assert_equal(217, get_anitya_id('bridge_utils', 'https://wiki.linuxfoundation.org/networking/bridge', 'Package'))

tools/version.rb

@@ -1,5 +1,5 @@
 #!/usr/bin/env ruby
-# version.rb version 3.16 (for Chromebrew)
+# version.rb version 3.17 (for Chromebrew)
 
 OPTIONS = %w[-h --help -j --json -u --update-package-files -v --verbose -vv]
 
@@ -51,15 +51,11 @@ versions_updated = {}
 versions = []
 
 def get_version(name, homepage, source, version)
-  anitya_name_mapping_idx = CREW_ANITYA_PACKAGE_NAME_MAPPINGS.keys.find_index { |i| i == name }
-  anitya_name = name.gsub(/\Apy\d_|\Aperl_|\Aruby_/, '')
-  anitya_name = CREW_ANITYA_PACKAGE_NAME_MAPPINGS.values[anitya_name_mapping_idx] unless anitya_name_mapping_idx.nil?
-  anitya_id = get_anitya_id(anitya_name, homepage, @pkg.superclass.to_s)
-  anitya_name = @new_anitya_name unless @new_anitya_name.nil?
+  anitya_id = get_anitya_id(name, homepage, @pkg.superclass.to_s)
   # If anitya_id cannot be determined, a Range can be returned, and
   # .nonzero? does not work with Ranges.
   anitya_id = nil if anitya_id.is_a? Range
-  puts "anitya_name: #{anitya_name} anitya_id: #{anitya_id}" if VERBOSE
+  puts "anitya_id: #{anitya_id}" if VERBOSE
   if anitya_id&.nonzero?
     # Get the latest stable version of the package from anitya.
     json = JSON.parse(Net::HTTP.get(URI("https://release-monitoring.org/api/v2/versions/?project_id=#{anitya_id}")))
@@ -197,6 +193,14 @@ def get_anitya_id(name, homepage, buildsystem)
   # Ignore python pip and ruby gem packages with the Pip and RUBY buildsystems.
   return if %w[Pip RUBY].include?(buildsystem)
 
+  # Change the name into something Anitya will prefer.
+  original_name = name.dup
+  # Remove any language-specific prefixes and build splitting suffixes.
+  name = PackageUtils.get_clean_name(name)
+  # If this package has a hardcoded mapping, use it.
+  name = CREW_ANITYA_PACKAGE_NAME_MAPPINGS[name] if CREW_ANITYA_PACKAGE_NAME_MAPPINGS.include?(name)
+  puts "anitya_name: #{name} #{"(instead of #{original_name})" if name != original_name}" if VERBOSE
+
   # Find out how many packages Anitya has with the provided name.
   puts "url is https://release-monitoring.org/api/v2/projects/?name=#{CGI.escape(name)}" if VERY_VERBOSE
   json = JSON.parse(Net::HTTP.get(URI("https://release-monitoring.org/api/v2/projects/?name=#{CGI.escape(name)}")))