ecosystems-bibliothecary 14.2.0 → 15.0.0

data/lib/bibliothecary/parsers/nuget.rb

@@ -7,13 +7,12 @@ module Bibliothecary
   module Parsers
     class Nuget
       include Bibliothecary::Analyser
-      extend Bibliothecary::MultiParsers::JSONRuntime
 
       def self.mapping
         {
           match_filename("Project.json") => {
             kind: "manifest",
-            parser: :parse_json_runtime_manifest,
+            parser: :parse_project_json,
           },
           match_filename("Project.lock.json") => {
             kind: "lockfile",
@@ -46,9 +45,19 @@ module Bibliothecary
         }
       end
 
-      add_multi_parser(Bibliothecary::MultiParsers::CycloneDX)
-      add_multi_parser(Bibliothecary::MultiParsers::DependenciesCSV)
-      add_multi_parser(Bibliothecary::MultiParsers::Spdx)
+      def self.parse_project_json(file_contents, options: {})
+        manifest = JSON.parse(file_contents)
+        dependencies = manifest.fetch("dependencies", {}).map do |name, requirement|
+          Dependency.new(
+            name: name,
+            requirement: requirement,
+            type: "runtime",
+            source: options.fetch(:filename, nil),
+            platform: platform_name
+          )
+        end
+        ParserResult.new(dependencies: dependencies)
+      end
 
       def self.parse_project_lock_json(file_contents, options: {})
         manifest = JSON.parse file_contents
@@ -68,32 +77,24 @@ module Bibliothecary
       def self.parse_packages_lock_json(file_contents, options: {})
         manifest = JSON.parse file_contents
 
-        frameworks = {}
-        manifest.fetch("dependencies", []).each do |framework, deps|
-          frameworks[framework] = deps
-            .reject { |_name, details| details["type"] == "Project" } # Projects do not have versions
+        # Merge dependencies from all target frameworks, deduping by name.
+        # Different frameworks may have different versions of the same package,
+        # but we can only report one version per package.
+        dependencies = manifest.fetch("dependencies", {}).flat_map do |_framework, deps|
+          deps
+            .reject { |_name, details| details["type"] == "Project" }
             .map do |name, details|
               Dependency.new(
                 name: name,
-                # 'resolved' has been set in all examples so far
-                # so fallback to requested is pure paranoia
                 requirement: details.fetch("resolved", details.fetch("requested", "*")),
                 type: "runtime",
                 source: options.fetch(:filename, nil),
                 platform: platform_name
               )
             end
-        end
+        end.uniq(&:name)
 
-        unless frameworks.empty?
-          # we should really return multiple manifests, but bibliothecary doesn't
-          # do that yet so at least pick deterministically.
-
-          # Note, frameworks can be empty, so remove empty ones and then return the last sorted item if any
-          frameworks.delete_if { |_k, v| v.empty? }
-          return ParserResult.new(dependencies: frameworks[frameworks.keys.max]) unless frameworks.empty?
-        end
-        ParserResult.new(dependencies: [])
+        ParserResult.new(dependencies: dependencies)
       end
 
       def self.parse_packages_config(file_contents, options: {})
@@ -201,7 +202,7 @@ module Bibliothecary
 
       def self.parse_paket_lock(file_contents, options: {})
         lines = file_contents.split("\n")
-        package_version_re = /\s+(?<name>\S+)\s\((?<version>\d+\.\d+[.\d+[.\d+]*]*)\)/
+        package_version_re = /\s+(?<name>\S+)\s\((?<version>\d+(?:\.\d+)+)\)/
         packages = lines.select { |line| package_version_re.match(line) }.map { |line| package_version_re.match(line) }.map do |match|
           Dependency.new(
             name: match[:name].strip,

data/lib/bibliothecary/parsers/ollama.rb

@@ -15,8 +15,6 @@ module Bibliothecary
         }
       end
 
-      add_multi_parser(Bibliothecary::MultiParsers::CycloneDX)
-      add_multi_parser(Bibliothecary::MultiParsers::DependenciesCSV)
 
       def self.parse_modelfile(file_contents, options: {})
         source = options.fetch(:filename, 'Modelfile')

data/lib/bibliothecary/parsers/packagist.rb

@@ -20,9 +20,6 @@ module Bibliothecary
         }
       end
 
-      add_multi_parser(Bibliothecary::MultiParsers::CycloneDX)
-      add_multi_parser(Bibliothecary::MultiParsers::DependenciesCSV)
-      add_multi_parser(Bibliothecary::MultiParsers::Spdx)
 
       def self.parse_lockfile(file_contents, options: {})
         manifest = JSON.parse file_contents

data/lib/bibliothecary/parsers/pub.rb

@@ -20,8 +20,6 @@ module Bibliothecary
         }
       end
 
-      add_multi_parser(Bibliothecary::MultiParsers::CycloneDX)
-      add_multi_parser(Bibliothecary::MultiParsers::DependenciesCSV)
 
       def self.parse_yaml_manifest(file_contents, options: {})
         manifest = YAML.load file_contents

data/lib/bibliothecary/parsers/pypi.rb

@@ -88,34 +88,40 @@ module Bibliothecary
         }
       end
 
-      add_multi_parser(Bibliothecary::MultiParsers::CycloneDX)
-      add_multi_parser(Bibliothecary::MultiParsers::DependenciesCSV)
-      add_multi_parser(Bibliothecary::MultiParsers::Spdx)
 
       def self.parser_pylock(file_contents, options: {})
-        lockfile = Tomlrb.parse(file_contents)
-        dependencies = lockfile["packages"].map do |d|
-          is_local = true if d.key?("archive") || d.key?("directory")
-          Dependency.new(
+        dependencies = []
+        # Split into [[packages]] blocks and extract fields from each
+        file_contents.split(/\[\[packages\]\]/).drop(1).each do |block|
+          name = block[/^name\s*=\s*"([^"]+)"/m, 1]
+          version = block[/^version\s*=\s*"([^"]+)"/m, 1]
+          # Local packages have [packages.archive] or [packages.directory] sections
+          is_local = block.include?("[packages.archive]") || block.include?("[packages.directory]")
+
+          dependencies << Dependency.new(
             platform: platform_name,
-            name: d["name"],
+            name: name,
             type: "runtime",
             source: options.fetch(:filename, nil),
-            requirement: d["version"] || "*",
-            local: is_local
+            requirement: version || "*",
+            local: is_local || nil
           )
         end
         ParserResult.new(dependencies: dependencies)
       end
 
       def self.parse_uv_lock(file_contents, options: {})
-        manifest = Tomlrb.parse(file_contents)
         source = options.fetch(:filename, nil)
-        dependencies = manifest.fetch("package", []).map do |package|
-          Dependency.new(
+        dependencies = []
+        # Split into [[package]] blocks and extract fields from each
+        file_contents.split(/\[\[package\]\]/).drop(1).each do |block|
+          name = block[/name\s*=\s*"([^"]+)"/, 1]
+          version = block[/version\s*=\s*"([^"]+)"/, 1]
+
+          dependencies << Dependency.new(
             platform: platform_name,
-            name: package["name"],
-            requirement: map_requirements(package),
+            name: name,
+            requirement: version,
             type: "runtime", # All dependencies are considered runtime
             source: source
           )
@@ -230,32 +236,43 @@ module Bibliothecary
       end
 
       def self.parse_poetry_lock(file_contents, options: {})
-        manifest = Tomlrb.parse(file_contents)
         deps = []
-        manifest["package"].each do |package|
-          # next if group == "_meta"
+        # Split into [[package]] blocks and extract fields from each
+        file_contents.split(/\[\[package\]\]/).drop(1).each do |block|
+          name = block[/^name\s*=\s*"([^"]+)"/m, 1]
+          version = block[/^version\s*=\s*"([^"]+)"/m, 1]
 
           # Poetry <1.2.0 used singular "category" for kind
           # Poetry >=1.2.0 uses plural "groups" field for kind(s)
-          groups = package.values_at("category", "groups").flatten.compact
-            .map do |g|
-              if g == "dev"
-                "develop"
-              else
-                (g == "main" ? "runtime" : g)
-              end
+          # Use ^ anchor to avoid matching commented lines
+          category = block[/^category\s*=\s*"([^"]+)"/m, 1]
+          groups_match = block[/^groups\s*=\s*\[([^\]]+)\]/m, 1]
+          groups = if groups_match
+                     groups_match.scan(/"([^"]+)"/).flatten
+                   elsif category
+                     [category]
+                   else
+                     []
+                   end
+
+          groups = groups.map do |g|
+            if g == "dev"
+              "develop"
+            else
+              (g == "main" ? "runtime" : g)
             end
+          end
 
           groups = ["runtime"] if groups.empty?
 
           groups.each do |group|
-            # Poetry lockfiles should already contain normalizated names, but we'll
+            # Poetry lockfiles should already contain normalized names, but we'll
             # apply it here as well just to be consistent with pyproject.toml parsing.
-            normalized_name = normalize_name(package["name"])
+            normalized_name = normalize_name(name)
             deps << Dependency.new(
               name: normalized_name,
-              original_name: normalized_name == package["name"] ? nil : package["name"],
-              requirement: map_requirements(package),
+              original_name: normalized_name == name ? nil : name,
+              requirement: version,
               type: group,
               source: options.fetch(:filename, nil),
               platform: platform_name
@@ -314,7 +331,8 @@ module Bibliothecary
       # Invalid lines in requirements.txt are skipped.
       def self.parse_requirements_txt(file_contents, options: {})
         deps = []
-        type = case options[:filename]
+        source = options.fetch(:filename, nil)
+        type = case source
                when /dev/ || /docs/ || /tools/
                  "development"
                when /test/
@@ -323,21 +341,22 @@ module Bibliothecary
                  "runtime"
                end
 
-        file_contents.split("\n").each do |line|
-          if line["://"]
+        file_contents.each_line do |line|
+          line = line.chomp
+          if line.include?("://")
             begin
-              result = parse_requirements_txt_url(line, type, options.fetch(:filename, nil))
+              result = parse_requirements_txt_url(line, type, source)
             rescue URI::Error, NoEggSpecified
               next
             end
 
             deps << result
-          elsif (match = line.delete(" ").match(REQUIREMENTS_REGEXP))
+          elsif (match = line.tr(" ", "").match(REQUIREMENTS_REGEXP))
             deps << Dependency.new(
               name: match[1],
               requirement: match[-1],
               type: type,
-              source: options.fetch(:filename, nil),
+              source: source,
               platform: platform_name
             )
           end

data/lib/bibliothecary/parsers/rubygems.rb

@@ -1,18 +1,24 @@
 # frozen_string_literal: true
 
 require "bundler"
-require "gemnasium/parser"
 
 module Bibliothecary
   module Parsers
     class Rubygems
       include Bibliothecary::Analyser
-      extend Bibliothecary::MultiParsers::BundlerLikeManifest
 
       NAME_VERSION = '(?! )(.*?)(?: \(([^-]*)(?:-(.*))?\))?'
       NAME_VERSION_4 = /^ {4}#{NAME_VERSION}$/
       BUNDLED_WITH = /BUNDLED WITH/
 
+      # Gemfile patterns
+      GEM_REGEXP = /^\s*gem\s+['"]([^'"]+)['"]\s*(?:,\s*['"]([^'"]+)['"])?/
+      GROUP_START = /^\s*group\s+(.+?)\s+do/
+      BLOCK_END = /^\s*end\s*$/
+
+      # Gemspec pattern - captures type in first group
+      GEMSPEC_DEPENDENCY = /\.add_(development_|runtime_)?dependency\s*\(?\s*['"]([^'"]+)['"]\s*(?:,\s*['"]([^'"]+)['"])?(?:\s*,\s*['"]([^'"]+)['"])?\s*\)?/
+
       def self.mapping
         {
           match_filenames("Gemfile", "gems.rb") => {
@@ -33,55 +39,114 @@ module Bibliothecary
         }
       end
 
-      add_multi_parser(Bibliothecary::MultiParsers::CycloneDX)
-      add_multi_parser(Bibliothecary::MultiParsers::DependenciesCSV)
-      add_multi_parser(Bibliothecary::MultiParsers::Spdx)
 
       def self.parse_gemfile_lock(file_contents, options: {})
-        lockfile = Bundler::LockfileParser.new(file_contents)
         source = options.fetch(:filename, nil)
+        dependencies = []
 
-        dependencies = lockfile.specs.map do |spec|
-          Dependency.new(
-            platform: platform_name,
-            name: spec.name,
-            requirement: spec.version.to_s,
-            type: "runtime",
-            source: source
-          )
-        end
+        file_contents.each_line do |line|
+          line = line.chomp.gsub(/\r$/, "")
+          next unless (match = line.match(NAME_VERSION_4))
+
+          name, version, _platform = match.captures
+          next if name.nil? || name.empty?
 
-        bundler_version = lockfile.bundler_version
-        if bundler_version
           dependencies << Dependency.new(
             platform: platform_name,
-            name: "bundler",
-            requirement: bundler_version.to_s,
+            name: name,
+            requirement: version,
             type: "runtime",
             source: source
           )
         end
 
+        if (bundler_dep = parse_bundler(file_contents, source))
+          dependencies << bundler_dep
+        end
+
         ParserResult.new(dependencies: dependencies)
       end
 
       def self.parse_gemfile(file_contents, options: {})
-        manifest = Gemnasium::Parser.send(:gemfile, file_contents)
-        dependencies = parse_ruby_manifest(manifest, platform_name, options.fetch(:filename, nil))
-        ParserResult.new(dependencies: dependencies)
+        source = options.fetch(:filename, nil)
+        deps = []
+        current_type = "runtime"
+        block_depth = 0
+
+        file_contents.each_line do |line|
+          # Track group blocks
+          if (group_match = line.match(GROUP_START))
+            block_depth += 1
+            groups = group_match[1]
+            current_type = groups.include?(":development") ? "development" : "runtime"
+            next
+          end
+
+          if line.match?(BLOCK_END) && block_depth > 0
+            block_depth -= 1
+            current_type = "runtime" if block_depth == 0
+            next
+          end
+
+          # Match gem declarations
+          if (match = line.match(GEM_REGEXP))
+            name = match[1]
+            version = match[2]
+            requirement = version ? "= #{version}" : ">= 0"
+
+            deps << Dependency.new(
+              platform: platform_name,
+              name: name,
+              requirement: requirement,
+              type: current_type,
+              source: source
+            )
+          end
+        end
+
+        ParserResult.new(dependencies: deps)
       end
 
       def self.parse_gemspec(file_contents, options: {})
-        manifest = Gemnasium::Parser.send(:gemspec, file_contents)
-        dependencies = parse_ruby_manifest(manifest, platform_name, options.fetch(:filename, nil))
-        ParserResult.new(dependencies: dependencies)
+        source = options.fetch(:filename, nil)
+        deps = []
+
+        file_contents.each_line do |line|
+          match = line.match(GEMSPEC_DEPENDENCY)
+          next unless match
+
+          type_prefix, name, ver1, ver2 = match.captures
+          type = type_prefix == "development_" ? "development" : "runtime"
+          requirement = build_requirement(ver1, ver2)
+
+          deps << Dependency.new(
+            platform: platform_name,
+            name: name,
+            requirement: requirement,
+            type: type,
+            source: source
+          )
+        end
+
+        ParserResult.new(dependencies: deps)
+      end
+
+      def self.build_requirement(ver1, ver2)
+        if ver1 && ver2
+          "#{ver1}, #{ver2}"
+        elsif ver1
+          ver1
+        else
+          ">= 0"
+        end
       end
 
       def self.parse_bundler(file_contents, source = nil)
         bundled_with_index = file_contents.lines(chomp: true).find_index { |line| line.match(BUNDLED_WITH) }
-        version = file_contents.lines(chomp: true).fetch(bundled_with_index + 1)&.strip
+        return nil unless bundled_with_index
 
-        return nil unless version
+        version = file_contents.lines(chomp: true).fetch(bundled_with_index + 1, nil)&.strip
+        return nil unless version && !version.empty?
 
         Dependency.new(
           name: "bundler",

data/lib/bibliothecary/parsers/shard.rb

@@ -20,7 +20,6 @@ module Bibliothecary
         }
       end
 
-      add_multi_parser(Bibliothecary::MultiParsers::DependenciesCSV)
 
       def self.parse_yaml_lockfile(file_contents, options: {})
         manifest = YAML.load file_contents

data/lib/bibliothecary/parsers/swift_pm.rb

@@ -1,75 +1,123 @@
+require "json"
+
 module Bibliothecary
   module Parsers
     class SwiftPM
       include Bibliothecary::Analyser
 
+      # Matches .Package(url: "...", majorVersion: X, minor: Y)
+      # Also matches .package(url: "...", from: "X.Y.Z") (Swift 4+)
+      PACKAGE_REGEXP_LEGACY = /\.Package\s*\(\s*url:\s*"([^"]+)"[^)]*majorVersion:\s*(\d+)[^)]*minor:\s*(\d+)/
+      PACKAGE_REGEXP_FROM = /\.package\s*\(\s*(?:name:\s*"[^"]+",\s*)?url:\s*"([^"]+)"[^)]*from:\s*"([^"]+)"/i
+      PACKAGE_REGEXP_EXACT = /\.package\s*\(\s*(?:name:\s*"[^"]+",\s*)?url:\s*"([^"]+)"[^)]*(?:\.exact|exact)\s*\(\s*"([^"]+)"\s*\)/i
+      PACKAGE_REGEXP_RANGE = /\.package\s*\(\s*(?:name:\s*"[^"]+",\s*)?url:\s*"([^"]+)"[^)]*"([^"]+)"\s*(?:\.\.|\.\.\.)\s*"([^"]+)"/i
+
       def self.mapping
         {
           match_filename("Package.swift", case_insensitive: true) => {
-            kind: 'manifest',
+            kind: "manifest",
             parser: :parse_package_swift,
-            related_to: ['lockfile']
+            related_to: ["lockfile"],
           },
           match_filename("Package.resolved", case_insensitive: true) => {
-            kind: 'lockfile',
+            kind: "lockfile",
             parser: :parse_package_resolved,
-            related_to: ['manifest']
-          }
+            related_to: ["manifest"],
+          },
         }
       end
 
-      add_multi_parser(Bibliothecary::MultiParsers::CycloneDX)
-      add_multi_parser(Bibliothecary::MultiParsers::DependenciesCSV)
-      add_multi_parser(Bibliothecary::MultiParsers::Spdx)
 
       def self.parse_package_swift(file_contents, options: {})
-        source = options.fetch(:filename, 'Package.swift')
-        response = Typhoeus.post("#{Bibliothecary.configuration.swift_parser_host}/to-json", body: file_contents, timeout: 60)
-        raise Bibliothecary::RemoteParsingError.new("Http Error #{response.response_code} when contacting: #{Bibliothecary.configuration.swift_parser_host}/to-json", response.response_code) unless response.success?
-        json = JSON.parse(response.body)
-        deps = json["dependencies"].map do |dependency|
-          name = dependency["url"].gsub(/^https?:\/\//, "").gsub(/\.git$/,"")
-          version = "#{dependency['version']['lowerBound']} - #{dependency['version']['upperBound']}"
-          Bibliothecary::Dependency.new(
+        source = options.fetch(:filename, "Package.swift")
+        deps = []
+
+        # Remove comments (but not :// in URLs)
+        content = file_contents.gsub(%r{(?<!:)//.*$}, "")
+
+        # Legacy format: .Package(url: "...", majorVersion: X, minor: Y)
+        content.scan(PACKAGE_REGEXP_LEGACY) do |url, major, minor|
+          name = url.gsub(%r{^https?://}, "").gsub(/\.git$/, "")
+          # Match the remote parser format: lowerBound - upperBound
+          lower = "#{major}.#{minor}.0"
+          upper = "#{major}.#{minor}.9223372036854775807"
+          deps << Dependency.new(
+            platform: platform_name,
+            name: name,
+            requirement: "#{lower} - #{upper}",
+            type: "runtime",
+            source: source
+          )
+        end
+
+        # Swift 4+ format: .package(url: "...", from: "X.Y.Z")
+        content.scan(PACKAGE_REGEXP_FROM) do |url, version|
+          name = url.gsub(%r{^https?://}, "").gsub(/\.git$/, "")
+          deps << Dependency.new(
             platform: platform_name,
             name: name,
-            requirement: version,
+            requirement: ">= #{version}",
             type: "runtime",
             source: source
           )
         end
-        Bibliothecary::ParserResult.new(dependencies: deps)
+
+        # Swift 4+ exact version: .package(url: "...", .exact("X.Y.Z"))
+        content.scan(PACKAGE_REGEXP_EXACT) do |url, version|
+          name = url.gsub(%r{^https?://}, "").gsub(/\.git$/, "")
+          deps << Dependency.new(
+            platform: platform_name,
+            name: name,
+            requirement: "= #{version}",
+            type: "runtime",
+            source: source
+          )
+        end
+
+        # Swift 4+ range: .package(url: "...", "1.0.0"..<"2.0.0")
+        content.scan(PACKAGE_REGEXP_RANGE) do |url, lower, upper|
+          name = url.gsub(%r{^https?://}, "").gsub(/\.git$/, "")
+          deps << Dependency.new(
+            platform: platform_name,
+            name: name,
+            requirement: "#{lower} - #{upper}",
+            type: "runtime",
+            source: source
+          )
+        end
+
+        ParserResult.new(dependencies: deps)
       end
 
       def self.parse_package_resolved(file_contents, options: {})
-        source = options.fetch(:filename, 'Package.resolved')
+        source = options.fetch(:filename, "Package.resolved")
         json = JSON.parse(file_contents)
         deps = if json["version"] == 1
           json["object"]["pins"].map do |dependency|
-            name = dependency['repositoryURL'].gsub(/^https?:\/\//, '').gsub(/\.git$/,'')
-            version = dependency['state']['version']
-            Bibliothecary::Dependency.new(
+            name = dependency["repositoryURL"].gsub(%r{^https?://}, "").gsub(/\.git$/, "")
+            version = dependency["state"]["version"]
+            Dependency.new(
               platform: platform_name,
               name: name,
               requirement: version,
-              type: 'runtime',
+              type: "runtime",
               source: source
             )
           end
-        else # version 2
+        else # version 2+
           json["pins"].map do |dependency|
-            name = dependency['location'].gsub(/^https?:\/\//, '').gsub(/\.git$/,'')
-            version = dependency['state']['version']
-            Bibliothecary::Dependency.new(
+            name = dependency["location"].gsub(%r{^https?://}, "").gsub(/\.git$/, "")
+            version = dependency["state"]["version"]
+            Dependency.new(
               platform: platform_name,
               name: name,
               requirement: version,
-              type: 'runtime',
+              type: "runtime",
               source: source
             )
           end
         end
-        Bibliothecary::ParserResult.new(dependencies: deps)
+        ParserResult.new(dependencies: deps)
       end
     end
   end