Merge pull request #9074 from vladimyr/livecheck

livecheck: refactor url preprocessing

Library/Homebrew/download_strategy.rb

@@ -1225,7 +1225,8 @@ class DownloadStrategyDetector
     when %r{^https?://github\.com/[^/]+/[^/]+\.git$}
       GitHubGitDownloadStrategy
     when %r{^https?://.+\.git$},
-         %r{^git://}
+         %r{^git://},
+         %r{^https?://git\.sr\.ht/[^/]+/[^/]+$}
       GitDownloadStrategy
     when %r{^https?://www\.apache\.org/dyn/closer\.cgi},
          %r{^https?://www\.apache\.org/dyn/closer\.lua}

Library/Homebrew/livecheck/livecheck.rb

@@ -3,6 +3,7 @@
 
 require "livecheck/strategy"
 require "ruby-progressbar"
+require "uri"
 
 module Homebrew
   # The {Livecheck} module consists of methods used by the `brew livecheck`
@@ -13,18 +14,15 @@ module Homebrew
   module Livecheck
     module_function
 
-    GITHUB_SPECIAL_CASES = %w[
+    GITEA_INSTANCES = %w[
-      api.github.com
+      codeberg.org
-      /latest
+      gitea.com
-      mednafen
+      opendev.org
-      camlp5
+      tildegit.org
-      kotlin
+    ].freeze
-      osrm-backend
+
-      prometheus
+    GOGS_INSTANCES = %w[
-      pyenv-virtualenv
+      lolg.it
-      sysdig
-      shairport-sync
-      yuicompressor
     ].freeze
 
     UNSTABLE_VERSION_KEYWORDS = %w[
@@ -316,26 +314,35 @@ module Homebrew
     # Preprocesses and returns the URL used by livecheck.
     # @return [String]
     def preprocess_url(url)
-      # Check for GitHub repos on github.com, not AWS
+      begin
-      url = url.sub("github.s3.amazonaws.com", "github.com") if url.include?("github")
+        uri = URI.parse url
+      rescue URI::InvalidURIError
+        return url
+      end
 
-      # Use repo from GitHub or GitLab inferred from download URL
+      host = uri.host == "github.s3.amazonaws.com" ? "github.com" : uri.host
-      if url.include?("github.com") && GITHUB_SPECIAL_CASES.none? { |sc| url.include? sc }
+      path = uri.path.delete_prefix("/").delete_suffix(".git")
-        if url.include? "archive"
+      scheme = uri.scheme
-          url = url.sub(%r{/archive/.*}, ".git") if url.include? "github"
-        elsif url.include? "releases"
-          url = url.sub(%r{/releases/.*}, ".git")
-        elsif url.include? "downloads"
-          url = "#{Pathname.new(url.sub(%r{/downloads(.*)}, "\\1")).dirname}.git"
-        elsif !url.end_with?(".git")
-          # Truncate the URL at the user/repo part, if possible
-          %r{(?<github_repo_url>(?:[a-z]+://)?github.com/[^/]+/[^/#]+)} =~ url
-          url = github_repo_url if github_repo_url.present?
 
-          url.delete_suffix!("/") if url.end_with?("/")
+      if host.end_with?("github.com")
-          url += ".git"
+        return url if path.match? %r{/releases/latest/?$}
-        end
+
-      elsif url.include?("/-/archive/")
+        owner, repo = path.delete_prefix("downloads/").split("/")
+        url = "#{scheme}://#{host}/#{owner}/#{repo}.git"
+      elsif host.end_with?(*GITEA_INSTANCES)
+        return url if path.match? %r{/releases/latest/?$}
+
+        owner, repo = path.split("/")
+        url = "#{scheme}://#{host}/#{owner}/#{repo}.git"
+      elsif host.end_with?(*GOGS_INSTANCES)
+        owner, repo = path.split("/")
+        url = "#{scheme}://#{host}/#{owner}/#{repo}.git"
+      # sourcehut
+      elsif host.end_with?("git.sr.ht")
+        owner, repo = path.split("/")
+        url = "#{scheme}://#{host}/#{owner}/#{repo}"
+      # GitLab (gitlab.com or self-hosted)
+      elsif path.include?("/-/archive/")
         url = url.sub(%r{/-/archive/.*$}i, ".git")
       end
 

Library/Homebrew/test/livecheck/livecheck_spec.rb

@@ -156,6 +156,12 @@ describe Homebrew::Livecheck do
   describe "::preprocess_url" do
     let(:github_git_url_with_extension) { "https://github.com/Homebrew/brew.git" }
 
+    it "returns the unmodified URL for an unparseable URL" do
+      # Modeled after the `head` URL in the `ncp` formula
+      expect(livecheck.preprocess_url(":something:cvs:@cvs.brew.sh:/cvs"))
+        .to eq(":something:cvs:@cvs.brew.sh:/cvs")
+    end
+
     it "returns the unmodified URL for a GitHub URL ending in .git" do
       expect(livecheck.preprocess_url(github_git_url_with_extension))
         .to eq(github_git_url_with_extension)
@@ -200,5 +206,35 @@ describe Homebrew::Livecheck do
       expect(livecheck.preprocess_url("https://brew.sh/Homebrew/brew/-/archive/1.0.0/brew-1.0.0.tar.gz"))
         .to eq("https://brew.sh/Homebrew/brew.git")
     end
+
+    it "returns the Git repository URL for a Codeberg archive URL" do
+      expect(livecheck.preprocess_url("https://codeberg.org/Homebrew/brew/archive/brew-1.0.0.tar.gz"))
+        .to eq("https://codeberg.org/Homebrew/brew.git")
+    end
+
+    it "returns the Git repository URL for a Gitea archive URL" do
+      expect(livecheck.preprocess_url("https://gitea.com/Homebrew/brew/archive/brew-1.0.0.tar.gz"))
+        .to eq("https://gitea.com/Homebrew/brew.git")
+    end
+
+    it "returns the Git repository URL for an Opendev archive URL" do
+      expect(livecheck.preprocess_url("https://opendev.org/Homebrew/brew/archive/brew-1.0.0.tar.gz"))
+        .to eq("https://opendev.org/Homebrew/brew.git")
+    end
+
+    it "returns the Git repository URL for a tildegit archive URL" do
+      expect(livecheck.preprocess_url("https://tildegit.org/Homebrew/brew/archive/brew-1.0.0.tar.gz"))
+        .to eq("https://tildegit.org/Homebrew/brew.git")
+    end
+
+    it "returns the Git repository URL for a LOL Git archive URL" do
+      expect(livecheck.preprocess_url("https://lolg.it/Homebrew/brew/archive/brew-1.0.0.tar.gz"))
+        .to eq("https://lolg.it/Homebrew/brew.git")
+    end
+
+    it "returns the Git repository URL for a sourcehut archive URL" do
+      expect(livecheck.preprocess_url("https://git.sr.ht/~Homebrew/brew/archive/1.0.0.tar.gz"))
+        .to eq("https://git.sr.ht/~Homebrew/brew")
+    end
   end
 end