site_maps 0.0.1.beta3 → 0.1.1

data/lib/site_maps/primitive/string.rb

@@ -0,0 +1,106 @@
+# frozen_string_literal: true
+
+begin
+  require "dry/inflector"
+rescue LoadError
+  # noop
+end
+
+begin
+  require "active_support/inflector"
+rescue LoadError
+  # noop
+end
+
+module SiteMaps::Primitive
+  class String < ::String
+    def self.inflector
+      return @inflector if defined?(@inflector)
+
+      @inflector = if defined?(::ActiveSupport::Inflector)
+        ::ActiveSupport::Inflector
+      elsif defined?(::Dry::Inflector)
+        ::Dry::Inflector.new
+      end
+    end
+
+    def classify
+      new_str = inflector&.classify(self) || split("/").collect do |c|
+        c.split("_").collect(&:capitalize).join
+      end.join("::")
+
+      self.class.new(new_str)
+    end
+
+    def constantize
+      inflector&.constantize(self) || Object.const_get(self)
+    end
+
+    def underscore
+      new_str = sub(/^::/, "")
+        .gsub("::", "/")
+        .gsub(/([A-Z]+)([A-Z][a-z])/, '\1_\2')
+        .gsub(/([a-z\d])([A-Z])/, '\1_\2')
+        .tr("-", "_")
+        .tr(".", "_")
+        .gsub(/\s/, "_")
+        .gsub(/__+/, "_")
+        .downcase
+
+      self.class.new(new_str)
+    end
+
+    def pluralize
+      new_str = inflector&.pluralize(self) || begin
+        # dummy pluralize
+        if /y$/.match?(self)
+          sub(/y$/, "ies")
+        elsif /s$/.match?(self)
+          self
+        else
+          self + "s"
+        end
+      end
+
+      new_str.is_a?(self.class) ? new_str : self.class.new(new_str)
+    end
+
+    def singularize
+      new_str = inflector&.singularize(self) || begin
+        # dummy singularize
+        if /ies$/.match?(self)
+          sub(/ies$/, "y")
+        elsif /s$/.match?(self)
+          sub(/s$/, "")
+        else
+          self
+        end
+      end
+
+      new_str.is_a?(self.class) ? new_str : self.class.new(new_str)
+    end
+
+    def camelize(uppercase_first_letter = true)
+      new_str = inflector&.camelize(self, uppercase_first_letter) || begin
+        # dummy camelize
+        str = to_s
+        str = str.sub(/^[a-z\d]*/) { $&.capitalize }
+        str = str.tr("-", "_")
+        str = str.gsub(/(?:_|(\/))([a-z\d]*)/i) { "#{$1}#{$2.capitalize}" }
+        str = str.gsub("/", "::")
+        unless uppercase_first_letter
+          str = str.sub(/^[A-Z]*/) { $&.downcase }
+        end
+        str
+      end
+
+      self.class.new(new_str)
+    end
+
+    protected
+
+    def inflector
+      self.class.inflector
+    end
+  end
+end

data/lib/site_maps/robots_txt.rb

@@ -0,0 +1,21 @@
+# frozen_string_literal: true
+
+module SiteMaps
+  module RobotsTxt
+    class << self
+      def sitemap_directive(url = nil)
+        url ||= SiteMaps.current_adapter&.config&.url
+        raise ArgumentError, "No sitemap URL provided and no adapter configured" unless url
+
+        "Sitemap: #{url}"
+      end
+
+      def render(sitemap_url: nil, extra_directives: [])
+        lines = ["User-agent: *", "Allow: /"]
+        extra_directives.each { |d| lines << d }
+        lines << sitemap_directive(sitemap_url)
+        lines.join("\n") + "\n"
+      end
+    end
+  end
+end

data/lib/site_maps/runner/event_listener.rb

@@ -2,12 +2,12 @@
 
 module SiteMaps
   module Runner::EventListener
-    extend Primitives::Output
+    extend Primitive::Output
 
     module_function
 
     def [](event_name)
-      method_name = Primitives::String.new(event_name).underscore.to_sym
+      method_name = Primitive::String.new(event_name).underscore.to_sym
       return unless respond_to?(:"on_#{method_name}")
 
       method(:"on_#{method_name}")

data/lib/site_maps/runner.rb

@@ -4,8 +4,9 @@ module SiteMaps
   class Runner
     attr_reader :adapter
 
-    def initialize(adapter = SiteMaps.current_adapter, max_threads: 4)
+    def initialize(adapter = SiteMaps.current_adapter, max_threads: 4, ping: nil)
       @adapter = adapter.tap(&:reset!)
+      @ping = ping
       @pool = Concurrent::FixedThreadPool.new(max_threads)
       @execution = Concurrent::Hash.new
       @failed = Concurrent::AtomicBoolean.new(false)
@@ -62,8 +63,9 @@ module SiteMaps
                 builder = SiteMaps::SitemapBuilder.new(
                   adapter: adapter,
                   location: process.location(**kwargs),
-                  notification_payload: { process: process }
+                  notification_payload: {process: process}
                 )
+                adapter.process_mixins.each { |mixin| builder.extend(mixin) }
                 process.call(builder, **kwargs)
                 builder.finalize!
               end
@@ -88,9 +90,21 @@ module SiteMaps
       adapter.repo.remaining_index_links.each do |item|
         adapter.sitemap_index.add(item)
       end
+      adapter.external_sitemaps.each do |item|
+        adapter.sitemap_index.add(item)
+      end
       unless adapter.sitemap_index.empty?
         raw_data = adapter.sitemap_index.to_xml
-        adapter.write(adapter.config.url, raw_data, last_modified: Time.now)
+        adapter.write(adapter.config.url, raw_data, last_modified: adapter.sitemap_index.last_modified)
+      end
+      should_ping = @ping.nil? ? adapter.config.respond_to?(:ping_search_engines?) && adapter.config.ping_search_engines? : @ping
+      ping_search_engines if should_ping
+    end
+
+    def ping_search_engines
+      engines = adapter.config.respond_to?(:ping_engines) ? adapter.config.ping_engines : nil
+      SiteMaps::Notification.instrument("sitemaps.ping") do |payload|
+        payload[:results] = SiteMaps::Ping.ping(adapter.config.url, engines: engines)
       end
     end
 

data/lib/site_maps/sitemap_builder.rb

@@ -6,7 +6,7 @@ module SiteMaps
 
     def initialize(adapter:, location: nil, notification_payload: {})
       @adapter = adapter
-      @url_set = SiteMaps::Builder::URLSet.new
+      @url_set = build_url_set
       @location = location
       @mutex = Mutex.new
       @notification_payload = notification_payload
@@ -15,10 +15,13 @@ module SiteMaps
     def add(path, params: nil, **options)
       @mutex.synchronize do
         link = build_link(path, params)
-        url_set.add(link, **options)
+        filtered_options = adapter.apply_url_filters(link, options)
+        return if filtered_options.nil?
+
+        url_set.add(link, **filtered_options)
       rescue SiteMaps::FullSitemapError
         finalize_and_start_next_urlset!
-        url_set.add(link, **options)
+        url_set.add(link, **filtered_options)
       end
     end
 
@@ -66,7 +69,16 @@ module SiteMaps
         adapter.write(sitemap_url, raw_data, last_modified: url_set.last_modified)
         add_sitemap_index(sitemap_url, lastmod: url_set.last_modified)
       end
-      @url_set = SiteMaps::Builder::URLSet.new
+      @url_set = build_url_set
+    end
+
+    def build_url_set
+      options = {}
+      options[:max_links] = config.max_links if config.respond_to?(:max_links)
+      options[:emit_priority] = config.emit_priority if config.respond_to?(:emit_priority)
+      options[:emit_changefreq] = config.emit_changefreq if config.respond_to?(:emit_changefreq)
+      options[:xsl_url] = config.xsl_stylesheet_url if config.respond_to?(:xsl_stylesheet_url)
+      SiteMaps::Builder::URLSet.new(**options)
     end
 
     def build_link(path, params)

data/lib/site_maps/sitemap_reader.rb

@@ -43,6 +43,9 @@ module SiteMaps
       raise FileNotFoundError.new("The file #{@location} does not exist")
     rescue OpenURI::HTTPError
       raise FileNotFoundError.new("The file #{@location} could not be opened")
+    rescue SocketError, Errno::ECONNREFUSED, Errno::ETIMEDOUT,
+      Net::OpenTimeout, Net::ReadTimeout => e
+      raise FileNotFoundError.new("The file #{@location} could not be reached: #{e.message}")
     end
 
     def compressed?

data/lib/site_maps/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module SiteMaps
-  VERSION = "0.0.1.beta3"
+  VERSION = "0.1.1"
 end

data/lib/site_maps.rb

@@ -3,6 +3,7 @@
 require_relative "site_maps/version"
 
 require "builder"
+require "logger"
 require "concurrent-ruby"
 require "date"
 require "fileutils"
@@ -23,6 +24,7 @@ loader.inflector.inflect "dsl" => "DSL"
 loader.inflector.inflect "url_set" => "URLSet"
 loader.inflector.inflect "url" => "URL"
 loader.inflector.inflect "xml" => "XML"
+loader.inflector.inflect "xsl_stylesheet" => "XSLStylesheet"
 loader.log! if ENV["DEBUG_ZEITWERK"]
 loader.setup
 
@@ -33,7 +35,7 @@ module SiteMaps
     news: 1_000
   }
   MAX_FILESIZE = 50_000_000 # bytes
-  DEFAULT_LOGGER = Logger.new($stdout)
+  DEFAULT_LOGGER = ::Logger.new($stdout)
 
   Error = Class.new(StandardError)
   AdapterNotFound = Class.new(Error)
@@ -42,6 +44,10 @@ module SiteMaps
   FullSitemapError = Class.new(Error)
   ConfigurationError = Class.new(Error)
 
+  SCOPE_KEY = :__site_maps_scope__
+
+  @mutex = Mutex.new
+
   class << self
     attr_reader :current_adapter
     attr_writer :logger
@@ -54,18 +60,51 @@ module SiteMaps
       adapter_class = if adapter.is_a?(Class) # && adapter < Adapters::Adapter
         adapter
       else
-        const_name = Primitives::String.new(adapter.to_s).classify
+        const_name = Primitive::String.new(adapter.to_s).classify
         begin
           Adapters.const_get(const_name)
         rescue NameError
           raise AdapterNotFound, "Adapter #{adapter.inspect} not found"
         end
       end
-      @current_adapter = adapter_class.new(**options, &block)
+      instance = adapter_class.new(**options, &block)
+      if (scope = Thread.current[SCOPE_KEY])
+        scope[:adapter] = instance
+      else
+        @current_adapter = instance
+      end
+      instance
+    end
+
+    # Register a context-aware sitemap definition. The block is stored and
+    # called when {.generate} is invoked with a `context:` hash. The hash
+    # keys are passed as keyword arguments to the block.
+    #
+    # Example:
+    #   # config/sitemap.rb
+    #   SiteMaps.define do |site:|
+    #     use(:file_system) do
+    #       config.url = "https://#{site.domain}/sitemap.xml"
+    #       process { |s| site.pages.each { |p| s.add(p.path) } }
+    #     end
+    #   end
+    #
+    #   # Usage:
+    #   SiteMaps.generate(config_file: "config/sitemap.rb", context: {site: site})
+    #     .enqueue_all
+    #     .run
+    #
+    # @param block [Proc] Receives keyword arguments from the `context:` hash
+    def define(&block)
+      if (scope = Thread.current[SCOPE_KEY])
+        scope[:definition] = block
+      else
+        @definition = block
+      end
     end
 
     def config
-      @config ||= Configuration.new
+      @mutex.synchronize { @config ||= Configuration.new }
       yield(@config) if block_given?
       @config
     end
@@ -88,21 +127,53 @@ module SiteMaps
     #     .enqueue_remaining # Enqueue all other non-enqueued processes
     #     .run
     #
+    # For multi-tenant / context-aware configurations, the config file can
+    # use {.define} and pass runtime context as keyword arguments via the
+    # `context:` kwarg:
+    #
+    # Example:
+    #   SiteMaps.generate(config_file: "config/sitemap.rb", context: {site: site})
+    #     .enqueue_all
+    #     .run
+    #
     # @param config_file [String] The path to a configuration file
+    # @param context [Hash] Keyword arguments passed to the block registered
+    #   via {.define}. Must be a Hash (or nil for no context).
     # @param options [Hash] Options to pass to the runner
     # @return [Runner] An instance of the runner
-    def generate(config_file: nil, **options)
+    def generate(config_file: nil, context: nil, **options)
+      adapter = nil
       if config_file
-        @current_adapter = nil
-        load(config_file)
+        previous_scope = Thread.current[SCOPE_KEY]
+        scope = {adapter: nil, definition: nil}
+        Thread.current[SCOPE_KEY] = scope
+        begin
+          load(config_file)
+          if scope[:definition]
+            kwargs = context || {}
+            raise ArgumentError, "context: must be a Hash, got #{context.class}" unless kwargs.is_a?(Hash)
+
+            instance_exec(**kwargs, &scope[:definition])
+          end
+          adapter = scope[:adapter]
+        ensure
+          Thread.current[SCOPE_KEY] = previous_scope
+        end
+        # Preserve backward-compat: expose the generated adapter through
+        # the `current_adapter` singleton for single-tenant callers. In
+        # multi-tenant concurrent use, last-writer-wins — each Runner still
+        # gets its own isolated adapter from the thread-local scope above.
+        @current_adapter = adapter if adapter
+      else
+        adapter = current_adapter
       end
-      raise AdapterNotSetError, "No adapter set. Use SiteMaps.use to set an adapter" unless current_adapter
+      raise AdapterNotSetError, "No adapter set. Use SiteMaps.use to set an adapter" unless adapter
 
-      Runner.new(current_adapter, **options)
+      Runner.new(adapter, **options)
     end
 
     def logger
-      @logger ||= DEFAULT_LOGGER
+      @mutex.synchronize { @logger ||= DEFAULT_LOGGER }
     end
   end
 end

data/site_maps.gemspec

@@ -17,7 +17,7 @@ Gem::Specification.new do |spec|
 
   spec.homepage = "https://github.com/marcosgz/site_maps"
   spec.license = "MIT"
-  spec.required_ruby_version = Gem::Requirement.new(">= 2.7.0")
+  spec.required_ruby_version = Gem::Requirement.new(">= 3.2.0")
 
   raise "RubyGems 2.0 or newer is required to protect against public gem pushes." unless spec.respond_to?(:metadata)
 

metadata

@@ -1,14 +1,13 @@
 --- !ruby/object:Gem::Specification
 name: site_maps
 version: !ruby/object:Gem::Version
-  version: 0.0.1.beta3
+  version: 0.1.1
 platform: ruby
 authors:
 - Marcos G. Zimmermann
-autorequire:
 bindir: exec
 cert_chain: []
-date: 2024-11-23 00:00:00.000000000 Z
+date: 1980-01-02 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rack
@@ -96,7 +95,9 @@ files:
 - ".rspec"
 - ".rubocop.yml"
 - ".tool-versions"
+- AGENTS.md
 - CHANGELOG.md
+- CLAUDE.md
 - Gemfile
 - Gemfile.lock
 - LICENSE.txt
@@ -104,6 +105,16 @@ files:
 - Rakefile
 - bin/console
 - bin/setup
+- docs/README.md
+- docs/adapters.md
+- docs/api.md
+- docs/cli.md
+- docs/events.md
+- docs/extensions.md
+- docs/getting-started.md
+- docs/middleware.md
+- docs/processes.md
+- docs/rails.md
 - exec/site_maps
 - lib/site-maps.rb
 - lib/site_maps.rb
@@ -124,18 +135,22 @@ files:
 - lib/site_maps/builder/sitemap_index/item.rb
 - lib/site_maps/builder/url.rb
 - lib/site_maps/builder/url_set.rb
+- lib/site_maps/builder/xsl_stylesheet.rb
 - lib/site_maps/cli.rb
 - lib/site_maps/configuration.rb
 - lib/site_maps/incremental_location.rb
+- lib/site_maps/middleware.rb
 - lib/site_maps/notification.rb
 - lib/site_maps/notification/bus.rb
 - lib/site_maps/notification/event.rb
 - lib/site_maps/notification/publisher.rb
-- lib/site_maps/primitives/array.rb
-- lib/site_maps/primitives/output.rb
-- lib/site_maps/primitives/string.rb
+- lib/site_maps/ping.rb
+- lib/site_maps/primitive/array.rb
+- lib/site_maps/primitive/output.rb
+- lib/site_maps/primitive/string.rb
 - lib/site_maps/process.rb
 - lib/site_maps/railtie.rb
+- lib/site_maps/robots_txt.rb
 - lib/site_maps/runner.rb
 - lib/site_maps/runner/event_listener.rb
 - lib/site_maps/sitemap_builder.rb
@@ -150,7 +165,6 @@ metadata:
   bug_tracker_uri: https://github.com/marcosgz/site_maps/issues
   documentation_uri: https://github.com/marcosgz/site_maps
   source_code_uri: https://github.com/marcosgz/site_maps
-post_install_message:
 rdoc_options: []
 require_paths:
 - lib
@@ -158,15 +172,14 @@ required_ruby_version: !ruby/object:Gem::Requirement
   requirements:
   - - ">="
     - !ruby/object:Gem::Version
-      version: 2.7.0
+      version: 3.2.0
 required_rubygems_version: !ruby/object:Gem::Requirement
   requirements:
   - - ">="
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.5.21
-signing_key:
+rubygems_version: 3.6.9
 specification_version: 4
 summary: Concurrent and Incremental sitemap.xml builder for ruby applications
 test_files: []

data/lib/site_maps/primitives/string.rb

@@ -1,43 +0,0 @@
-# frozen_string_literal: true
-
-begin
-  require "dry/inflector"
-rescue LoadError
-  # noop
-end
-
-begin
-  require "active_support/inflector"
-rescue LoadError
-  # noop
-end
-
-module SiteMaps::Primitives
-  class String < ::String
-    def classify
-      new_str = if defined?(Dry::Inflector)
-        Dry::Inflector.new.classify(self)
-      elsif defined?(ActiveSupport::Inflector)
-        ActiveSupport::Inflector.classify(self)
-      else
-        split("_").map(&:capitalize).join
-      end
-
-      self.class.new(new_str)
-    end
-
-    def underscore
-      new_str = sub(/^::/, "")
-        .gsub("::", "/")
-        .gsub(/([A-Z]+)([A-Z][a-z])/, '\1_\2')
-        .gsub(/([a-z\d])([A-Z])/, '\1_\2')
-        .tr("-", "_")
-        .tr(".", "_")
-        .gsub(/\s/, "_")
-        .gsub(/__+/, "_")
-        .downcase
-
-      self.class.new(new_str)
-    end
-  end
-end