Merge pull request #948 from julia-thorn/use-pip-defaults-homepage

Pip: Fallback Homepage to Project-URLs

Author: MikeMcQuaid

lib/licensed/sources/pip.rb

@@ -20,7 +20,7 @@ def enumerate_dependencies
             metadata: {
               "type"        => self.class.type,
               "summary"     => package["Summary"],
-              "homepage"    => package["Home-page"]
+              "homepage"    => homepage(package)
             }
           )
         end
@@ -83,14 +83,30 @@ def package_names
       end
 
       # Returns a hash filled with package info parsed from the email-header formatted output
-      # returned by `pip show`
+      # returned by `pip show --verbose`, including continuation lines for multi-line fields.
       def parse_package_info(package_info)
+        current_key = nil
+
         package_info.lines.each_with_object(Hash.new(0)) do |pkg, a|
-          next if pkg.start_with?(/^\s/)
+          if pkg.match?(/^\s/)
+            if current_key
+              current_value = a[current_key]
+              continuation = pkg.strip
+              a[current_key] =
+                if current_value.to_s.empty?
+                  continuation
+                else
+                  "#{current_value}\n#{continuation}"
+                end
+            end
+            next
+          end
 
           k, v = pkg.split(":", 2)
           next if k.nil? || k.empty?
-          a[k.strip] = v&.strip
+
+          current_key = k.strip
+          a[current_key] = v&.strip
         end
       end
 
@@ -101,7 +117,39 @@ def pip_list_command
 
       # Returns the output from `pip show <package> <package> ...`
       def pip_show_command(package)
-        Licensed::Shell.execute(*pip_command, "--disable-pip-version-check", "show", package)
+        Licensed::Shell.execute(*pip_command, "--disable-pip-version-check", "show", "--verbose", package)
+      end
+
+      # Returns the package homepage from pip package metadata
+      def homepage(package)
+        home_page = package["Home-page"]
+        return home_page unless home_page.to_s.empty?
+
+        homepage_from_project_urls(package["Project-URL"] || package["Project-URLs"]) || home_page
+      end
+
+      # Returns best-effort homepage URL extracted from Project-URL(s) metadata
+      # With priority given to Home > Repository > Source, otherwise the first URL
+      def homepage_from_project_urls(project_urls)
+        return if project_urls.to_s.empty?
+
+        entries = project_urls
+          .to_s
+          .split("\n")
+          .map(&:strip)
+          .reject(&:empty?)
+
+        candidates = entries.filter_map do |entry|
+          label, url = entry.split(",", 2).map { |value| value&.strip }
+          next unless url&.match?(%r{^https?://})
+
+          [label.to_s, url]
+        end
+
+        preferred = candidates.find { |label, _| label.match?(/home/i) } ||
+          candidates.find { |label, _| label.match?(/repo/i) } ||
+          candidates.find { |label, _| label.match?(/source/i) }
+        preferred&.last || candidates.first&.last
       end
 
       def virtual_env_dir

test/sources/pip_test.rb

@@ -4,6 +4,103 @@
 require "tmpdir"
 
 describe Licensed::Sources::Pip do
+  class TestablePipSource < Licensed::Sources::Pip
+    public :parse_package_info, :homepage
+  end
+
+  let(:config) { Licensed::AppConfiguration.new({ "source_path" => Dir.pwd }) }
+  let(:source) { TestablePipSource.new(config) }
+
+  it "parses pip show continuation lines" do
+    parsed = source.parse_package_info(
+      <<~INFO
+        Name: azure-core
+        Project-URLs:
+          Repository, https://github.com/Azure/azure-sdk-for-python/tree/main/sdk/core/azure-core
+      INFO
+    )
+
+    assert_equal(
+      "Repository, https://github.com/Azure/azure-sdk-for-python/tree/main/sdk/core/azure-core",
+      parsed["Project-URLs"]
+    )
+  end
+
+  it "falls back to Project-URLs when Home-page is empty" do
+    homepage = source.homepage(
+      {
+        "Home-page" => "",
+        "Project-URLs" => "Repository, https://github.com/Azure/azure-sdk-for-python/tree/main/sdk/core/azure-core"
+      }
+    )
+
+    assert_equal "https://github.com/Azure/azure-sdk-for-python/tree/main/sdk/core/azure-core", homepage
+  end
+
+  it "prefers Home URL from Project-URLs even when listed last" do
+    homepage = source.homepage(
+      {
+        "Home-page" => "",
+        "Project-URLs" => [
+          "Documentation, https://learn.microsoft.com/azure/",
+          "Repository, https://github.com/Azure/azure-sdk-for-python/tree/main/sdk/core/azure-core",
+          "Home, https://azure.microsoft.com/en-us/products/"
+        ].join("\n")
+      }
+    )
+
+    assert_equal "https://azure.microsoft.com/en-us/products/", homepage
+  end
+
+  it "prefers Repository URL when Home is not present" do
+    homepage = source.homepage(
+      {
+        "Home-page" => "",
+        "Project-URLs" => [
+          "Documentation, https://learn.microsoft.com/azure/",
+          "Repository, https://github.com/Azure/azure-sdk-for-python/tree/main/sdk/core/azure-core",
+          "Source, https://example.com/source"
+        ].join("\n")
+      }
+    )
+
+    assert_equal "https://github.com/Azure/azure-sdk-for-python/tree/main/sdk/core/azure-core", homepage
+  end
+
+  it "filters malformed and non-http Project-URLs entries" do
+    homepage = source.homepage(
+      {
+        "Home-page" => "",
+        "Project-URLs" => [
+          "NoCommaEntry",
+          "Documentation, ftp://learn.microsoft.com/azure/",
+          "Documentation, https://learn.microsoft.com/azure/",
+          "Source, not-a-url"
+        ].join("\n")
+      }
+    )
+
+    assert_equal "https://learn.microsoft.com/azure/", homepage
+  end
+
+  it "prefers Home-page when present" do
+    homepage = source.homepage(
+      {
+        "Home-page" => "https://example.com/home",
+        "Project-URLs" => "Repository, https://github.com/example/repo"
+      }
+    )
+
+    assert_equal "https://example.com/home", homepage
+  end
+
+  it "returns empty Home-page when Project-URLs is a non-string value" do
+    package = Hash.new(0)
+    package["Home-page"] = ""
+
+    assert_equal "", source.homepage(package)
+  end
+
   it "finds lowercase dist-info directories for mixed-case package names" do
     Dir.mktmpdir do |dir|
       dist_info = File.join(dir, "pyjwt-2.12.0.dist-info")