Merge pull request #10378 from SeekingMeaning/version-regexes

version: add new `Version::Parser` class
2025-07-14 16:09:03 +08:00 · 2021-01-25 09:24:42 -08:00 · 2021-01-25 09:24:42 -08:00 · 0a0f435a88
commit 0a0f435a88
parent 1f008d52df a06bd4e45d
3 changed files with 188 additions and 60 deletions
--- a/Library/Homebrew/test/version/parser_spec.rb
+++ b/Library/Homebrew/test/version/parser_spec.rb
@ -0,0 +1,81 @@
+# typed: false
+# frozen_string_literal: true
+
+require "version/parser"
+
+describe Version::Parser do
+  specify "::new" do
+    expect { described_class.new }
+      .to raise_error("Version::Parser is declared as abstract; it cannot be instantiated")
+  end
+
+  describe Version::RegexParser do
+    specify "::new" do
+      # TODO: see https://github.com/sorbet/sorbet/issues/2374
+      # expect { described_class.new(/[._-](\d+(?:\.\d+)+)/) }
+      #   .to raise_error("Version::RegexParser is declared as abstract; it cannot be instantiated")
+      expect { described_class.new(/[._-](\d+(?:\.\d+)+)/) }.not_to raise_error
+    end
+
+    specify "::process_spec" do
+      expect { described_class.process_spec(Pathname(TEST_TMPDIR)) }
+        .to raise_error("The method `process_spec` on #<Class:Version::RegexParser> is declared as `abstract`. " \
+                        "It does not have an implementation.")
+    end
+  end
+
+  describe Version::UrlParser do
+    specify "::new" do
+      expect { described_class.new(/[._-](\d+(?:\.\d+)+)/) }.not_to raise_error
+    end
+
+    specify "::process_spec" do
+      expect(described_class.process_spec(Pathname("#{TEST_TMPDIR}/testdir-0.1.test")))
+        .to eq("#{TEST_TMPDIR}/testdir-0.1.test")
+
+      expect(described_class.process_spec(Pathname("https://sourceforge.net/foo_bar-1.21.tar.gz/download")))
+        .to eq("https://sourceforge.net/foo_bar-1.21.tar.gz/download")
+
+      expect(described_class.process_spec(Pathname("https://sf.net/foo_bar-1.21.tar.gz/download")))
+        .to eq("https://sf.net/foo_bar-1.21.tar.gz/download")
+
+      expect(described_class.process_spec(Pathname("https://brew.sh/testball-0.1")))
+        .to eq("https://brew.sh/testball-0.1")
+
+      expect(described_class.process_spec(Pathname("https://brew.sh/testball-0.1.tgz")))
+        .to eq("https://brew.sh/testball-0.1.tgz")
+    end
+  end
+
+  describe Version::StemParser do
+    before { Pathname("#{TEST_TMPDIR}/testdir-0.1.test").mkpath }
+
+    after { Pathname("#{TEST_TMPDIR}/testdir-0.1.test").unlink }
+
+    specify "::new" do
+      expect { described_class.new(/[._-](\d+(?:\.\d+)+)/) }.not_to raise_error
+    end
+
+    describe "::process_spec" do
+      it "works with directories" do
+        expect(described_class.process_spec(Pathname("#{TEST_TMPDIR}/testdir-0.1.test"))).to eq("testdir-0.1.test")
+      end
+
+      it "works with SourceForge URLs with /download suffix" do
+        expect(described_class.process_spec(Pathname("https://sourceforge.net/foo_bar-1.21.tar.gz/download")))
+          .to eq("foo_bar-1.21")
+
+        expect(described_class.process_spec(Pathname("https://sf.net/foo_bar-1.21.tar.gz/download")))
+          .to eq("foo_bar-1.21")
+      end
+
+      it "works with URLs without file extension" do
+        expect(described_class.process_spec(Pathname("https://brew.sh/testball-0.1"))).to eq("testball-0.1")
+      end
+
+      it "works with URLs with file extension" do
+        expect(described_class.process_spec(Pathname("https://brew.sh/testball-0.1.tgz"))).to eq("testball-0.1")
+      end
+    end
+  end
+end
--- a/Library/Homebrew/version.rb
+++ b/Library/Homebrew/version.rb
@ -3,6 +3,7 @@

 require "pkg_version"
 require "version/null"
+require "version/parser"

 # A formula's version.
 #
@ -369,105 +370,86 @@ class Version

    spec = Pathname.new(spec) unless spec.is_a? Pathname

-    spec_s = spec.to_s
-
-    stem = if spec.directory?
-      spec.basename.to_s
-    elsif spec_s.match?(%r{((?:sourceforge\.net|sf\.net)/.*)/download$})
-      spec.dirname.stem
-    elsif spec_s.match?(/\.[^a-zA-Z]+$/) # rubocop:disable Lint/DuplicateBranch
-      spec.basename.to_s
-    else
-      spec.stem
+    VERSION_PARSERS.each do |parser|
+      version = parser.parse(spec)
+      return version if version.present?
    end

+    nil
+  end
+  private_class_method :_parse
+
+  VERSION_PARSERS = [
    # date-based versioning
    # e.g. ltopers-v2017-04-14.tar.gz
-    m = /-v?(\d{4}-\d{2}-\d{2})/.match(stem)
-    return m.captures.first unless m.nil?
+    StemParser.new(/-v?(\d{4}-\d{2}-\d{2})/),

    # GitHub tarballs
    # e.g. https://github.com/foo/bar/tarball/v1.2.3
    # e.g. https://github.com/sam-github/libnet/tarball/libnet-1.1.4
    # e.g. https://github.com/isaacs/npm/tarball/v0.2.5-1
    # e.g. https://github.com/petdance/ack/tarball/1.93_02
-    m = %r{github\.com/.+/(?:zip|tar)ball/(?:v|\w+-)?((?:\d+[-._])+\d*)$}.match(spec_s)
-    return m.captures.first unless m.nil?
+    UrlParser.new(%r{github\.com/.+/(?:zip|tar)ball/(?:v|\w+-)?((?:\d+[-._])+\d*)$}),

    # e.g. https://github.com/erlang/otp/tarball/OTP_R15B01 (erlang style)
-    m = /[-_]([Rr]\d+[AaBb]\d*(?:-\d+)?)/.match(spec_s)
-    return m.captures.first unless m.nil?
+    UrlParser.new(/[-_]([Rr]\d+[AaBb]\d*(?:-\d+)?)/),

    # e.g. boost_1_39_0
-    m = /((?:\d+_)+\d+)$/.match(stem)
-    return T.must(m.captures.first).tr("_", ".") unless m.nil?
+    StemParser.new(/((?:\d+_)+\d+)$/) { |s| s.tr("_", ".") },

    # e.g. foobar-4.5.1-1
    # e.g. unrtf_0.20.4-1
    # e.g. ruby-1.9.1-p243
-    m = /[-_]((?:\d+\.)*\d+\.\d+-(?:p|rc|RC)?\d+)(?:[-._](?i:bin|dist|stable|src|sources?|final|full))?$/.match(stem)
-    return m.captures.first unless m.nil?
+    StemParser.new(/[-_]((?:\d+\.)*\d+\.\d+-(?:p|rc|RC)?\d+)(?:[-._](?i:bin|dist|stable|src|sources?|final|full))?$/),

    # URL with no extension
    # e.g. https://waf.io/waf-1.8.12
    # e.g. https://codeload.github.com/gsamokovarov/jump/tar.gz/v0.7.1
-    m = /[-v]((?:\d+\.)*\d+)$/.match(spec_s)
-    return m.captures.first unless m.nil?
+    UrlParser.new(/[-v]((?:\d+\.)*\d+)$/),

    # e.g. lame-398-1
-    m = /-(\d+-\d+)/.match(stem)
-    return m.captures.first unless m.nil?
+    StemParser.new(/-(\d+-\d+)/),

    # e.g. foobar-4.5.1
-    m = /-((?:\d+\.)*\d+)$/.match(stem)
-    return m.captures.first unless m.nil?
+    StemParser.new(/-((?:\d+\.)*\d+)$/),

    # e.g. foobar-4.5.1.post1
-    m = /-((?:\d+\.)*\d+(.post\d+)?)$/.match(stem)
-    return m.captures.first unless m.nil?
+    StemParser.new(/-((?:\d+\.)*\d+(.post\d+)?)$/),

    # e.g. foobar-4.5.1b
-    m = /-((?:\d+\.)*\d+(?:[abc]|rc|RC)\d*)$/.match(stem)
-    return m.captures.first unless m.nil?
+    StemParser.new(/-((?:\d+\.)*\d+(?:[abc]|rc|RC)\d*)$/),

    # e.g. foobar-4.5.0-alpha5, foobar-4.5.0-beta1, or foobar-4.50-beta
-    m = /-((?:\d+\.)*\d+-(?:alpha|beta|rc)\d*)$/.match(stem)
-    return m.captures.first unless m.nil?
+    StemParser.new(/-((?:\d+\.)*\d+-(?:alpha|beta|rc)\d*)$/),

    # e.g. https://ftpmirror.gnu.org/libidn/libidn-1.29-win64.zip
    # e.g. https://ftpmirror.gnu.org/libmicrohttpd/libmicrohttpd-0.9.17-w32.zip
-    m = /-(\d+\.\d+(?:\.\d+)?)-w(?:in)?(?:32|64)$/.match(stem)
-    return m.captures.first unless m.nil?
+    StemParser.new(/-(\d+\.\d+(?:\.\d+)?)-w(?:in)?(?:32|64)$/),

    # Opam packages
    # e.g. https://opam.ocaml.org/archives/sha.1.9+opam.tar.gz
    # e.g. https://opam.ocaml.org/archives/lablgtk.2.18.3+opam.tar.gz
    # e.g. https://opam.ocaml.org/archives/easy-format.1.0.2+opam.tar.gz
-    m = /\.(\d+\.\d+(?:\.\d+)?)\+opam$/.match(stem)
-    return m.captures.first unless m.nil?
+    StemParser.new(/\.(\d+\.\d+(?:\.\d+)?)\+opam$/),

    # e.g. https://ftpmirror.gnu.org/mtools/mtools-4.0.18-1.i686.rpm
    # e.g. https://ftpmirror.gnu.org/autogen/autogen-5.5.7-5.i386.rpm
    # e.g. https://ftpmirror.gnu.org/libtasn1/libtasn1-2.8-x86.zip
    # e.g. https://ftpmirror.gnu.org/libtasn1/libtasn1-2.8-x64.zip
    # e.g. https://ftpmirror.gnu.org/mtools/mtools_4.0.18_i386.deb
-    m = /[-_](\d+\.\d+(?:\.\d+)?(?:-\d+)?)[-_.](?:i[36]86|x86|x64(?:[-_](?:32|64))?)$/.match(stem)
-    return m.captures.first unless m.nil?
+    StemParser.new(/[-_](\d+\.\d+(?:\.\d+)?(?:-\d+)?)[-_.](?:i[36]86|x86|x64(?:[-_](?:32|64))?)$/),

    # e.g. https://registry.npmjs.org/@angular/cli/-/cli-1.3.0-beta.1.tgz
    # e.g. https://github.com/dlang/dmd/archive/v2.074.0-beta1.tar.gz
    # e.g. https://github.com/dlang/dmd/archive/v2.074.0-rc1.tar.gz
    # e.g. https://github.com/premake/premake-core/releases/download/v5.0.0-alpha10/premake-5.0.0-alpha10-src.zip
-    m = /[-.vV]?((?:\d+\.)+\d+[-_.]?(?i:alpha|beta|pre|rc)\.?\d{,2})/.match(stem)
-    return m.captures.first unless m.nil?
+    StemParser.new(/[-.vV]?((?:\d+\.)+\d+[-_.]?(?i:alpha|beta|pre|rc)\.?\d{,2})/),

    # e.g. foobar4.5.1
-    m = /((?:\d+\.)*\d+)$/.match(stem)
-    return m.captures.first unless m.nil?
+    StemParser.new(/((?:\d+\.)*\d+)$/),

    # e.g. foobar-4.5.0-bin
-    m = /[-vV]((?:\d+\.)+\d+[abc]?)[-._](?i:bin|dist|stable|src|sources?|final|full)$/.match(stem)
-    return m.captures.first unless m.nil?
+    StemParser.new(/[-vV]((?:\d+\.)+\d+[abc]?)[-._](?i:bin|dist|stable|src|sources?|final|full)$/),

    # dash version style
    # e.g. http://www.antlr.org/download/antlr-3.4-complete.jar
@ -475,20 +457,16 @@ class Version
    # e.g. https://search.maven.org/remotecontent?filepath=com/facebook/presto/presto-cli/0.181/presto-cli-0.181-executable.jar
    # e.g. https://search.maven.org/remotecontent?filepath=org/fusesource/fuse-extra/fusemq-apollo-mqtt/1.3/fusemq-apollo-mqtt-1.3-uber.jar
    # e.g. https://search.maven.org/remotecontent?filepath=org/apache/orc/orc-tools/1.2.3/orc-tools-1.2.3-uber.jar
-    m = /-((?:\d+\.)+\d+)-/.match(stem)
-    return m.captures.first unless m.nil?
+    StemParser.new(/-((?:\d+\.)+\d+)-/),

    # e.g. dash_0.5.5.1.orig.tar.gz (Debian style)
-    m = /_((?:\d+\.)+\d+[abc]?)[.]orig$/.match(stem)
-    return m.captures.first unless m.nil?
+    StemParser.new(/_((?:\d+\.)+\d+[abc]?)[.]orig$/),

    # e.g. https://www.openssl.org/source/openssl-0.9.8s.tar.gz
-    m = /-v?(\d[^-]+)/.match(stem)
-    return m.captures.first unless m.nil?
+    StemParser.new(/-v?(\d[^-]+)/),

    # e.g. astyle_1.23_macosx.tar.gz
-    m = /_v?(\d[^_]+)/.match(stem)
-    return m.captures.first unless m.nil?
+    StemParser.new(/_v?(\d[^_]+)/),

    # e.g. http://mirrors.jenkins-ci.org/war/1.486/jenkins.war
    # e.g. https://github.com/foo/bar/releases/download/0.10.11/bar.phar
@ -497,18 +475,15 @@ class Version
    # e.g. https://wwwlehre.dhbw-stuttgart.de/~sschulz/WORK/E_DOWNLOAD/V_1.9/E.tgz
    # e.g. https://github.com/JustArchi/ArchiSteamFarm/releases/download/2.3.2.0/ASF.zip
    # e.g. https://people.gnome.org/~newren/eg/download/1.7.5.2/eg
-    m = %r{/([rvV]_?)?(\d+\.\d+(\.\d+){,2})}.match(spec_s)
-    return m.captures.second unless m.nil?
+    UrlParser.new(%r{/(?:[rvV]_?)?(\d+\.\d+(?:\.\d+){,2})}),

    # e.g. https://www.ijg.org/files/jpegsrc.v8d.tar.gz
-    m = /\.v(\d+[a-z]?)/.match(stem)
-    return m.captures.first unless m.nil?
+    StemParser.new(/\.v(\d+[a-z]?)/),

    # e.g. https://secure.php.net/get/php-7.1.10.tar.bz2/from/this/mirror
-    m = /[-.vV]?((?:\d+\.)+\d+(?:[-_.]?(?i:alpha|beta|pre|rc)\.?\d{,2})?)/.match(spec_s)
-    return m.captures.first unless m.nil?
-  end
-  private_class_method :_parse
+    UrlParser.new(/[-.vV]?((?:\d+\.)+\d+(?:[-_.]?(?i:alpha|beta|pre|rc)\.?\d{,2})?)/),
+  ].freeze
+  private_constant :VERSION_PARSERS

  sig { params(val: T.any(PkgVersion, String, Version), detected_from_url: T::Boolean).void }
  def initialize(val, detected_from_url: false)
--- a/Library/Homebrew/version/parser.rb
+++ b/Library/Homebrew/version/parser.rb
@ -0,0 +1,72 @@
+# typed: true
+# frozen_string_literal: true
+
+class Version
+  # @api private
+  class Parser
+    extend T::Sig
+    extend T::Helpers
+    abstract!
+
+    sig { abstract.params(spec: Pathname).returns(T.nilable(String)) }
+    def parse(spec); end
+  end
+
+  # @api private
+  class RegexParser < Parser
+    extend T::Sig
+    extend T::Helpers
+    abstract!
+
+    sig { params(regex: Regexp, block: T.nilable(T.proc.params(arg0: String).returns(String))).void }
+    def initialize(regex, &block)
+      super()
+      @regex = regex
+      @block = block
+    end
+
+    sig { override.params(spec: Pathname).returns(T.nilable(String)) }
+    def parse(spec)
+      match = @regex.match(self.class.process_spec(spec))
+      return if match.blank?
+
+      version = match.captures.first
+      return if version.blank?
+      return @block.call(version) if @block.present?
+
+      version
+    end
+
+    sig { abstract.params(spec: Pathname).returns(String) }
+    def self.process_spec(spec); end
+  end
+
+  # @api private
+  class UrlParser < RegexParser
+    extend T::Sig
+
+    sig { override.params(spec: Pathname).returns(String) }
+    def self.process_spec(spec)
+      spec.to_s
+    end
+  end
+
+  # @api private
+  class StemParser < RegexParser
+    extend T::Sig
+
+    SOURCEFORGE_DOWNLOAD_REGEX = %r{(?:sourceforge\.net|sf\.net)/.*/download$}.freeze
+    NO_FILE_EXTENSION_REGEX = /\.[^a-zA-Z]+$/.freeze
+
+    sig { override.params(spec: Pathname).returns(String) }
+    def self.process_spec(spec)
+      return spec.basename.to_s if spec.directory?
+
+      spec_s = spec.to_s
+      return spec.dirname.stem if spec_s.match?(SOURCEFORGE_DOWNLOAD_REGEX)
+      return spec.basename.to_s if spec_s.match?(NO_FILE_EXTENSION_REGEX)
+
+      spec.stem
+    end
+  end
+end