class Webgen::PathHandler

Namespace for all path handlers.

About ↑

A path handler is a webgen extension that uses source Path objects to create Node objects and that provides methods for rendering these nodes. The nodes are stored in a hierarchy, the root of which is a Tree object. Path handlers can do simple things, like copying a path from the source to the destination, or a complex things, like generating a whole set of nodes from one input path (e.g. generating a whole image gallery)!

The paths that are handled by a path handler are generally specified via path patterns. The create_nodes method of a path handler is called for each source path that should be handled. And when it is time to write out a node, the content method on the path handler associated with the node is called to retrieve the rendered content of the node.

Tree creation ↑

The method populate_tree is used for creating the initial node tree, the internal representation of all paths. It is only the initial tree because it is possible that additional, secondary nodes are created during the rendering phase by using the create_secondary_nodes method.

Tree creation works like this:

  1. All path handlers on the invocation list are used in turn. The order is important; it allows avoiding unnecessary write phases and it makes sure that, for example, directory nodes are created before their file nodes.

  2. When a path handler is used for creating nodes, all source paths (retrieved by using Webgen::Source#paths method) that match one of the associated patterns and/or all path with the 'handler' meta information set to the path handler are used.

  3. The meta information of a used source path is then updated with the meta information applied by methods registered for the :apply_meta_info_to_path blackboard message.

    After that the source path is given to the parse_meta_info! method of the path handler so that meta information of the path can be updated with meta information stored in the content of the path itself.

    Then the meta information 'versions' is used to determine if multiple version of the path should be used for creating nodes and each path version is then given to the create_nodes method of the path handler so that it can create one or more nodes.

  4. Nodes returned by creates_nodes of a path handler are assumed to have the Webgen::Node#node_info keys :path and :path_handler and the meta info key 'modified_at' correctly set (this is automatically done if the Webgen::PathHandler::Base#create_node method is used).

Path Patterns and Invocation order ↑

Path patterns define which paths are handled by a specific path handler. These patterns are specified when a path handler is registered using register method. The patterns need to have a format that Dir.glob can handle. Note that a user can always associate any path with a path handler through a meta information path and the 'handler' meta information key.

In addition to specifying the patterns a path handler uses, one can also specify the place in the invocation list which the path handler should use. The invocation list is used from the front to the back when the Tree is created.

Implementing a path handler ↑

A path handler must take the website as the only parameter on initialization and needs to define the following methods:

parse_meta_info!(path)

Update path.meta_info with meta information found in the content of the path. The return values of this method are given to the create_nodes method as additional parameters!

This allows one to use a single pass for reading the meta information and the normal content of the path.

create_nodes(path, …)

Create one or more nodes from the path and return them. If parse_meta_info! returns one or more values, these values are provided as additional parameters to this method.

It is a good idead to use the helper method Webgen::PathHandler::Base#create_node for actually creating a node.

content(node)

Return the content of the given node. This method is only called for nodes that have been created by the path handler.

Also note that a path handler does not need to reside under the Webgen::PathHandler namespace but all built-in ones do so that auto-loading of the path handlers works.

The Webgen::PathHandler::Base module provides default implementations of the needed methods (except for create_nodes) and should be used by all path handlers! If a path handler processes paths in Webgen Page Format, it should probably also use Webgen::PathHandler::PageUtils.

Information that is used by a path handler only for processing purposes should be stored in the node_info hash of a node as the meta_info hash is reserved for user provided node meta information.

Following is a simple path handler class example which copies paths from the source to the destination and modifies the extension in the process:

class SimpleCopy

  include Webgen::PathHandler::Base

  def create_nodes(path)
    path.ext += '.copied'
    create_node(path)
  end

  def content(node)
    node.node_info[:path]
  end

end

website.ext.path_handler.register(SimpleCopy, patterns: ['**  .jpg', '   *.png'])

Attributes

current_dest_node[R]

The destination node if one is currently written (only during the invocation of write_tree) or nil otherwise.

Public Class Methods

new(website) click to toggle source

Create a new path handler object for the given website.

Calls superclass method Webgen::ExtensionManager.new
# File lib/webgen/path_handler.rb, line 133
def initialize(website)
  super()
  @website = website
  @current_dest_node = nil
  @invocation_order = []
  @instances = {}
  @secondary_nodes = {}

  @website.blackboard.add_listener(:website_generated, 'path_handler') do
    @website.cache[:path_handler_secondary_nodes] = @secondary_nodes
  end

  used_secondary_paths = {}
  written_nodes = Set.new
  @website.blackboard.add_listener(:before_secondary_nodes_created, 'path_handler') do |path, source_alcn|
    (used_secondary_paths[source_alcn] ||= Set.new) << path if source_alcn
  end
  @website.blackboard.add_listener(:before_all_nodes_written, 'path_handler') do |node|
    used_secondary_paths = {}
    written_nodes = Set.new
  end
  @website.blackboard.add_listener(:after_node_written, 'path_handler') do |node|
    written_nodes << node.alcn
  end
  @website.blackboard.add_listener(:after_all_nodes_written, 'path_handler') do
    @secondary_nodes.delete_if do |path, data|
      if written_nodes.include?(data[1]) && (!used_secondary_paths[data[1]] ||
                                             !used_secondary_paths[data[1]].include?(path))
        data[2].each {|alcn| @website.tree.delete_node(@website.tree[alcn])}
        true
      end
    end
  end
end

Public Instance Methods

create_secondary_nodes(path, content = '', source_alcn = nil) click to toggle source

Create nodes for the given path (a Path object which must not be a source path).

The content of the path also needs to be specified. Note that if an IO block is associated with the path, it is discarded!

If the parameter handler is present, nodes from the given path are only created with the specified handler.

If the secondary nodes are created during the rendering phase (and not during node creation, ie. in a create_nodes method of a path handler), the source_alcn has to be set to the node alcn from which these nodes are created!

# File lib/webgen/path_handler.rb, line 341
def create_secondary_nodes(path, content = '', source_alcn = nil)
  if (sn = @secondary_nodes[path]) && sn[1] != source_alcn
    raise Webgen::NodeCreationError.new("Duplicate secondary path name <#{path}>", 'path_handler', path)
  end
  @website.blackboard.dispatch_msg(:before_secondary_nodes_created, path, source_alcn)

  path['modified_at'] ||= @website.tree[source_alcn]['modified_at'] if source_alcn
  path.set_io { StringIO.new(content) }

  nodes = if path['handler']
            @website.blackboard.dispatch_msg(:apply_meta_info_to_path, path)
            create_nodes_with_path_handler(path, path['handler'])
          else
            create_nodes([path])
          end
  @website.blackboard.dispatch_msg(:after_secondary_nodes_created, path, nodes)

  if source_alcn
    path.set_io(&nil)
    _, _, stored_alcns = @secondary_nodes.delete(path)
    cur_alcns = nodes.map {|n| n.alcn}
    (stored_alcns - cur_alcns).each {|n| @website.tree.delete_node(@website.tree[n])} if stored_alcns
    @secondary_nodes[path.dup] = [content, source_alcn, cur_alcns]
  end

  nodes
end
instance(handler) click to toggle source

Return the instance of the path handler class with the given name.

# File lib/webgen/path_handler.rb, line 212
def instance(handler)
  @instances[handler.intern] ||= extension(handler).new(@website)
end
populate_tree() click to toggle source

Populate the website tree with nodes.

Can only be called once because the tree can only be populated once!

# File lib/webgen/path_handler.rb, line 226
def populate_tree
  raise Webgen::NodeCreationError.new("Can't populate tree twice", 'path_handler') if @website.tree.root

  time = Benchmark.measure do
    meta_info, rest = @website.ext.source.paths.partition {|path| path.path =~ /[\/.]metainfo$/}

    used_paths = []

    @website.blackboard.add_listener(:before_node_created, 'path_handler (temp_populate_tree)') do |path|
      used_paths << path
    end
    create_nodes(meta_info, [:meta_info])
    create_nodes(rest)
    @website.blackboard.remove_listener(:before_node_created, 'path_handler (temp_populate_tree)')

    unused_paths = rest - used_paths
    @website.logger.vinfo do
      "The following source paths have not been used: #{unused_paths.join(', ')}"
    end if unused_paths.length > 0

    (@website.cache[:path_handler_secondary_nodes] || {}).each do |path, (content, source_alcn, _)|
      next if !@website.tree[source_alcn]
      create_secondary_nodes(path, content, source_alcn)
    end
  end
  @website.logger.vinfo do
    "Populating node tree took " << ('%2.2f' % time.real) << ' seconds'
  end

  @website.blackboard.dispatch_msg(:after_tree_populated)
end
register(klass, options={}, &block) click to toggle source

Register a path handler.

The parameter klass has to contain the name of the path handler class or the class object itself. If the class is located under this namespace, only the class name without the hierarchy part is needed, otherwise the full class name including parent module/class names is needed.

Options: ↑

:name

The name for the path handler. If not set, it defaults to the snake-case version of the class name (without the hierarchy part). It should only contain letters.

:patterns

A list of path patterns for which the path handler should be used. If not specified, defaults to an empty list.

:insert_at

Specifies the position in the invocation list. If not specified or if :end is specified, the handler is added to the end of the list. If :front is specified, it is added to the beginning of the list. Otherwise the value is expected to be a position number and the path handler is added at the specified position in the list.

Examples: ↑

path_handler.register('Template')     # registers Webgen::PathHandler::Template

path_handler.register('::Template')   # registers Template !!!

path_handler.register('MyModule::Doit', name: 'template', patterns: ['**/*.template'])
# File lib/webgen/path_handler.rb, line 197
def register(klass, options={}, &block)
  name = do_register(klass, options, false, &block)
  ext_data(name).patterns = options[:patterns] || []
  pos = if options[:insert_at].nil? || options[:insert_at] == :end
          -1
        elsif options[:insert_at] == :front
          0
        else
          options[:insert_at].to_i
        end
  @invocation_order.delete(name)
  @invocation_order.insert(pos, name)
end
write_tree() click to toggle source

Write all changed nodes of the website tree to their respective destination using the Destination object at website.ext.destination.

Returns the number of passes needed for correctly writing out all paths.

# File lib/webgen/path_handler.rb, line 262
def write_tree
  passes = 0
  content = nil

  begin
    at_least_one_node_written = false
    @website.cache.reset_volatile_cache
    @website.blackboard.dispatch_msg(:before_all_nodes_written)
    @website.tree.node_access[:alcn].sort_by {|a, n| [n['write_order'].to_s, a]}.each do |name, node|
      begin
        next if node == @website.tree.dummy_root ||
          (node['passive'] && !node['no_output'] && !@website.ext.item_tracker.node_referenced?(node)) ||
          ((@website.config['website.dry_run'] || node['no_output'] || @website.ext.destination.exists?(node.dest_path)) &&
           !@website.ext.item_tracker.node_changed?(node))

        @website.blackboard.dispatch_msg(:before_node_written, node)
        if !node['no_output']
          content = write_node(node)
          at_least_one_node_written = true
        end
        @website.blackboard.dispatch_msg(:after_node_written, node, content)
      rescue Webgen::Error => e
        e.path = node.alcn if e.path.to_s.empty?
        e.location = "path_handler.#{name_of_instance(node.node_info[:path_handler])}" unless e.location
        raise
      rescue Exception => e
        raise Webgen::RenderError.new(e, "path_handler.#{name_of_instance(node.node_info[:path_handler])}", node)
      end
    end
    @website.blackboard.dispatch_msg(:after_all_nodes_written)
    passes += 1 if at_least_one_node_written
  end while at_least_one_node_written

  @website.blackboard.dispatch_msg(:website_generated)
  passes
end

Private Instance Methods

create_nodes(paths, handlers = @invocation_order) click to toggle source

Use the registered path handlers to create nodes which are all returned.

# File lib/webgen/path_handler.rb, line 318
def create_nodes(paths, handlers = @invocation_order)
  nodes = []
  paths.each {|path| @website.blackboard.dispatch_msg(:apply_meta_info_to_path, path)}
  handlers.each do |name|
    paths_for_handler(name.to_s, paths).each do |path|
      nodes += create_nodes_with_path_handler(path, name)
    end
  end
  nodes
end
create_nodes_with_path_handler(path, handler) { |path| ... } click to toggle source

Prepare everything to create nodes from the path using the given handler. After the nodes are created, it is checked if they have all needed properties.

Returns an array with all created nodes.

# File lib/webgen/path_handler.rb, line 388
def create_nodes_with_path_handler(path, handler) #:yields: path
  *data = instance(handler).parse_meta_info!(path)

  (path.meta_info.delete('versions') || {'default' => {}}).map do |name, mi|
    vpath = path.dup
    (mi ||= {})['version'] ||= name
    vpath.meta_info.merge!(mi)
    @website.logger.debug do
      "Creating node version '#{vpath['version']}' from path <#{vpath}> with #{handler} handler"
    end
    @website.blackboard.dispatch_msg(:before_node_created, vpath)
    instance(handler).create_nodes(vpath, *data)
  end.flatten.compact.each do |node|
    @website.blackboard.dispatch_msg(:after_node_created, node)
  end
rescue Webgen::Error => e
  e.path = path.to_s if e.path.to_s.empty?
  e.location = "path_handler.#{handler}" unless e.location
  raise
rescue Exception => e
  raise Webgen::NodeCreationError.new(e, "path_handler.#{handler}", path)
end
name_of_instance(instance) click to toggle source

Return the name of the give path handler instance.

# File lib/webgen/path_handler.rb, line 217
def name_of_instance(instance)
  @instances.rassoc(instance)[0] || 'unknown'
end
paths_for_handler(name, paths) click to toggle source

Return the paths which are handled by the path handler name (where name is a String).

# File lib/webgen/path_handler.rb, line 370
def paths_for_handler(name, paths)
  patterns = ext_data(name).patterns || []

  options = (@website.config['path_handler.patterns.case_sensitive'] ? 0 : File::FNM_CASEFOLD) |
    (@website.config['path_handler.patterns.match_leading_dot'] ? File::FNM_DOTMATCH : 0) |
    File::FNM_PATHNAME

  paths.select do |path|
    path.meta_info['handler'] == name ||
      patterns.any? {|pat| Webgen::Path.matches_pattern?(path, pat, options)}
  end.sort {|a,b| a.path <=> b.path}
end
write_node(node) click to toggle source

Write the given node to the destination.

# File lib/webgen/path_handler.rb, line 300
def write_node(node)
  @current_dest_node = node
  @website.logger.info do
    "[#{(@website.ext.destination.exists?(node.dest_path) ? 'update' : 'create')}] <#{node.dest_path}>"
  end
  content = nil
  time = Benchmark.measure { content = node.content }
  @website.ext.destination.write(node.dest_path, content)
  @website.logger.vinfo do
    "[timing] <#{node.dest_path}> rendered in " << ('%2.2f' % time.real) << ' seconds'
  end
  content
ensure
  @current_dest_node = nil
end