class Puppet::Provider

@!attribute [r] doc

The (full) documentation for this provider class. The documentation for the provider class itself
should be set with the DSL method {desc=}. Setting the documentation with with {doc=} has the same effect
as setting it with {desc=} (only the class documentation part is set). In essence this means that
there is no getter for the class documentation part (since the getter returns the full
documentation when there are additional contributors).

@return [String] Returns the full documentation for the provider.

@see Puppet::Utils::Docs @comment This is puzzling … a write only doc attribute??? The generated setter never seems to be

used, instead the instance variable @doc is set in the `desc` method. This seems wrong. It is instead
documented as a read only attribute (to get the full documentation). Also see doc below for
desc.

@!attribute [w] desc

Sets the documentation of this provider class. (The full documentation is read via the
{doc} attribute).

@dsl type

@todo Original = _“LAK 2007-05-09: Keep the model stuff around for backward compatibility”_

Is this really needed? The comment about backwards compatibility was made in 2007.

@return [???] A model kept for backwards compatibility. @api private @deprecated This attribute is available for backwards compatibility reasons.

@return [String] The name of the provider

@todo What is this type? A reference to a Puppet::Type ? @return [Puppet::Type] the resource type (that this provider is … WHAT?)

@todo Original = _“The source parameter exists so that providers using the same

source can specify this, so reading doesn't attempt to read the
same package multiple times."_ This seems to be a package type specific attribute. Is this really
used?

@return [???] The source is WHAT?

@todo original = _“LAK 2007-05-09: Keep the model stuff around for backward compatibility”_, why is it

both here (instance) and at class level? Is this a different model?

@return [???] model is WHAT?

@return [???] This resource is what? Is an instance of a provider attached to one particular Puppet::Resource?

Returns the absolute path to the executable for the command referenced by the given name. @raise [Puppet::DevError] if the name does not reference an existing command. @return [String] the absolute path to the found executable for the command @see which

# File lib/puppet/provider.rb, line 154
def self.command(name)
  name = name.intern

  if defined?(@commands) and command = @commands[name]
    # nothing
  elsif superclass.respond_to? :command and command = superclass.command(name)
    # nothing
  else
    raise Puppet::DevError, "No command #{name} defined for provider #{self.name}"
  end

  which(command)
end

Confines this provider to be suitable only on hosts where the given commands are present. Also see {Puppet::Provider::Confiner#confine} for other types of confinement of a provider by use of other types of predicates.

@note It is preferred if the commands are not entered with absolute paths as this allows puppet

to search for them using the PATH variable.

@param command_specs [Hash{String => String}] Map of name to command that the provider will

be executing on the system. Each command is specified with a name and the path of the executable.

@return [void] @see ::optional_commands

# File lib/puppet/provider.rb, line 180
def self.commands(command_specs)
  command_specs.each do |name, path|
    has_command(name, path)
  end
end

@return [Boolean] Return whether the given feature has been declared or not.

# File lib/puppet/provider.rb, line 277
def self.declared_feature?(name)
  defined?(@declared_features) and @declared_features.include?(name)
end

@return [Boolean] Returns whether this implementation satisfies all of the default requirements or not.

Returns false If defaults are empty.

@see ::defaultfor

# File lib/puppet/provider.rb, line 285
def self.default?
  return false if @defaults.empty?
  if @defaults.find do |fact, values|
      values = [values] unless values.is_a? Array
      if fval = Facter.value(fact).to_s and fval != ""
        fval = fval.to_s.downcase.intern
      else
        return false
      end

      # If any of the values match, we're a default.
      if values.find do |value| fval == value.to_s.downcase.intern end
        false
      else
        true
      end
    end
    return false
  else
    return true
  end
end

Sets a facts filter that determine which of several suitable providers should be picked by default. This selection only kicks in if there is more than one suitable provider. To filter on multiple facts the given hash may contain more than one fact name/value entry. The filter picks the provider if all the fact/value entries match the current set of facts. (In case there are still more than one provider after this filtering, the first found is picked). @param hash [Hash<{String => Object}>] hash of fact name to fact value. @return [void]

# File lib/puppet/provider.rb, line 316
def self.defaultfor(hash)
  hash.each do |d,v|
    @defaults[d] = v
  end
end

(see Puppet::Util::Execution.execfail)

# File lib/puppet/provider.rb, line 146
def self.execfail(*args)
  Puppet::Util::Execution.execfail(*args)
end

(see Puppet::Util::Execution.execpipe)

# File lib/puppet/provider.rb, line 134
def self.execpipe(*args, &block)
  Puppet::Util::Execution.execpipe(*args, &block)
end

(see Puppet::Util::Execution.execute)

# File lib/puppet/provider.rb, line 122
def self.execute(*args)
  Puppet::Util::Execution.execute(*args)
end

Creates a convenience method for invocation of a command.

This generates a Provider method that allows easy execution of the command. The generated method may take arguments that will be passed through to the executable as the command line arguments when it is invoked.

@example Use it like this:

has_command(:echo, "/bin/echo")
def some_method
  echo("arg 1", "arg 2")
end

@comment the . . . below is intentional to avoid the three dots to become an illegible ellipsis char. @example . . . or like this

has_command(:echo, "/bin/echo") do
  is_optional
  environment :HOME => "/var/tmp", :PWD => "/tmp"
end

@param name [Symbol] The name of the command (will become the name of the generated method that executes the command) @param path [String] The path to the executable for the command @yield [ ] A block that configures the command (see {Puppet::Provider::Command}) @comment a yield [ ] produces {|| …} in the signature, do not remove the space. @note the name ´has_command´ looks odd in an API context, but makes more sense when seen in the internal

DSL context where a Provider is declaratively defined.

# File lib/puppet/provider.rb, line 228
def self.has_command(name, path, &block)
  name = name.intern
  command = CommandDefiner.define(name, path, self, &block)

  @commands[name] = command.executable

  # Now define the class and instance methods.
  create_class_and_instance_method(name) do |*args|
    return command.execute(*args)
  end
end

Initializes defaults and commands (i.e. clears them). @return [void]

# File lib/puppet/provider.rb, line 342
def self.initvars
  @defaults = {}
  @commands = {}
end

Returns a list of system resources (entities) this provider may/can manage. This is a query mechanism that lists entities that the provider may manage on a given system. It is is directly used in query services, but is also the foundation for other services; prefetching, and purging.

As an example, a package provider lists all installed packages. (In contrast, the File provider does not list all files on the file-system as that would make execution incredibly slow). An implementation of this method should be made if it is possible to quickly (with a single system call) provide all instances.

An implementation of this method should only cache the values of properties if they are discovered as part of the process for finding existing resources. Resource properties that require additional commands (than those used to determine existence/identity) should be implemented in their respective getter method. (This is important from a performance perspective; it may be expensive to compute, as well as wasteful as all discovered resources may perhaps not be managed).

An implementation may return an empty list (naturally with the effect that it is not possible to query for manageable entities).

By implementing this method, it is possible to use the `resources´ resource type to specify purging of all non managed entities.

@note The returned instances are instance of some subclass of Provider, not resources. @return [Array<Puppet::Provider>] a list of providers referencing the system entities @abstract this method must be implemented by a subclass and this super method should never be called as it raises an exception. @raise [Puppet::DevError] Error indicating that the method should have been implemented by subclass. @see prefetch

# File lib/puppet/provider.rb, line 374
def self.instances
  raise Puppet::DevError, "Provider #{self.name} has not defined the 'instances' class method"
end

Creates the methods for a given command. @api private @deprecated Use {commands}, {optional_commands}, or {has_command} instead. This was not meant to be part of a public API

# File lib/puppet/provider.rb, line 381
def self.make_command_methods(name)
  Puppet.deprecation_warning "Provider.make_command_methods is deprecated; use Provider.commands or Provider.optional_commands instead for creating command methods"

  # Now define a method for that command
  unless singleton_class.method_defined?(name)
    meta_def(name) do |*args|
      # This might throw an ExecutionFailure, but the system above
      # will catch it, if so.
      command = Puppet::Provider::Command.new(name, command(name), Puppet::Util, Puppet::Util::Execution)
      return command.execute(*args)
    end

    # And then define an instance method that just calls the class method.
    # We need both, so both instances and classes can easily run the commands.
    unless method_defined?(name)
      define_method(name) do |*args|
        self.class.send(name, *args)
      end
    end
  end
end

Creates getter- and setter- methods for each property supported by the resource type. Call this method to generate simple accessors for all properties supported by the resource type. These simple accessors lookup and sets values in the property hash. The generated methods may be overridden by more advanced implementations if something else than a straight forward getter/setter pair of methods is required. (i.e. define such overriding methods after this method has been called)

An implementor of a provider that makes use of `prefetch` and `flush` can use this method since it uses the internal `@property_hash` variable to store values. An implementation would then update the system state on a call to `flush` based on the current values in the `@property_hash`.

@return [void]

# File lib/puppet/provider.rb, line 416
def self.mk_resource_methods
  [resource_type.validproperties, resource_type.parameters].flatten.each do |attr|
    attr = attr.intern
    next if attr == :name
    define_method(attr) do
      @property_hash[attr] || :absent
    end

    define_method(attr.to_s + "=") do |val|
      @property_hash[attr] = val
    end
  end
end

Creates a new provider that is optionally initialized from a resource or a hash of properties. If no argument is specified, a new non specific provider is initialized. If a resource is given it is remembered for further operations. If a hash is used it becomes the internal `@property_hash` structure of the provider - this hash holds the current state property values of system entities as they are being discovered by querying or other operations (typically getters).

@todo The use of a hash as a parameter needs a better exaplanation; why is this done? What is the intent? @param resource [Puppet::Resource, Hash] optional resource or hash

# File lib/puppet/provider.rb, line 539
def initialize(resource = nil)
  if resource.is_a?(Hash)
    # We don't use a duplicate here, because some providers (ParsedFile, at least)
    # use the hash here for later events.
    @property_hash = resource
  elsif resource
    @resource = resource
    # LAK 2007-05-09: Keep the model stuff around for backward compatibility
    @model = resource
    @property_hash = {}
  else
    @property_hash = {}
  end
end

Defines optional commands. Since Puppet 2.7.8 this is typically not needed as evaluation of provider suitability is lazy (when a resource is evaluated) and the absence of commands that will be present after other resources have been applied no longer needs to be specified as optional. @param [Hash{String => String}] command_specs Named commands that the provider will

be executing on the system. Each command is specified with a name and the path of the executable.

(@see has_command) @see commands

# File lib/puppet/provider.rb, line 195
def self.optional_commands(hash)
  hash.each do |name, target|
    has_command(name, target) do
      is_optional
    end
  end
end

@return [String] Returns the data source, which is the provider name if no other source has been set. @todo Unclear what “the source” is used for?

# File lib/puppet/provider.rb, line 451
def self.source
  @source ||= self.name
end

@return [Integer] Returns a numeric specificity for this provider based on how many requirements it has

and number of _ancestors_. The higher the number the more specific the provider.

The number of requirements is based on the number of defaults set up with {Provider.defaultfor}.

The ancestors is the Ruby Module::ancestors method and the number of classes returned is used to boost the score. The intent is that if two providers are equal, but one is more “derived” than the other (i.e. includes more classes), it should win because it is more specific). @note Because of how this value is

calculated there could be surprising side effects if a provider included an excessive amount of classes.

# File lib/puppet/provider.rb, line 332
def self.specificity
  # This strange piece of logic attempts to figure out how many parent providers there
  # are to increase the score. What is will actually do is count all classes that Ruby Module::ancestors
  # returns (which can be other classes than those the parent chain) - in a way, an odd measure of the
  # complexity of a provider).
  (@defaults.length * 100) + ancestors.select { |a| a.is_a? Class }.length
end

Returns true if the given attribute/parameter is supported by the provider. The check is made that the parameter is a valid parameter for the resource type, and then if all its required features (if any) are supported by the provider.

@param param [Class, Puppet::Parameter] the parameter class, or a parameter instance @return [Boolean] Returns whether this provider supports the given parameter or not. @raise [Puppet::DevError] if the given parameter is not valid for the resource type

# File lib/puppet/provider.rb, line 463
def self.supports_parameter?(param)
  if param.is_a?(Class)
    klass = param
  else
    unless klass = resource_type.attrclass(param)
      raise Puppet::DevError, "'#{param}' is not a valid parameter for #{resource_type.name}"
    end
  end
  return true unless features = klass.required_features

  !!satisfies?(*features)
end

Compares this provider against another provider. Comparison is only possible with another provider (no other class). The ordering is based on the class name of the two providers.

@return [-1,0,+1, nil] A comparison result -1, 0, +1 if this is before other, equal or after other. Returns

nil oif not comparable to other.

@see Comparable

# File lib/puppet/provider.rb, line 593
def <=>(other)
  # We can only have ordering against other providers.
  return nil unless other.is_a? Puppet::Provider
  # Otherwise, order by the providers class name.
  return self.class.name <=> other.class.name
end

Clears this provider instance to allow GC to clean up.

# File lib/puppet/provider.rb, line 512
def clear
  @resource = nil
  @model = nil
end

(see command)

# File lib/puppet/provider.rb, line 518
def command(name)
  self.class.command(name)
end

Convenience methods - see class method with the same name. @see execfail @return (see execfail)

# File lib/puppet/provider.rb, line 141
def execfail(*args)
  Puppet::Util::Execution.execfail(*args)
end

Convenience methods - see class method with the same name. @see execpipe @return (see execpipe)

# File lib/puppet/provider.rb, line 129
def execpipe(*args, &block)
  Puppet::Util::Execution.execpipe(*args, &block)
end

Convenience methods - see class method with the same name. @see execute @return (see execute)

# File lib/puppet/provider.rb, line 117
def execute(*args)
  Puppet::Util::Execution.execute(*args)
end

Returns the value of a parameter value, or `:absent` if it is not defined. @param param [Puppet::Parameter] the parameter to obtain the value of @return [Object] the value of the parameter or `:absent` if not defined.

# File lib/puppet/provider.rb, line 526
def get(param)
  @property_hash[param.intern] || :absent
end

Returns the name of the resource this provider is operating on. @return [String] the name of the resource instance (e.g. the file path of a File). @raise [Puppet::DevError] if no resource is set, or no name defined.

# File lib/puppet/provider.rb, line 558
def name
  if n = @property_hash[:name]
    return n
  elsif self.resource
    resource.name
  else
    raise Puppet::DevError, "No resource and no name in property hash in #{self.class.name} instance"
  end
end

Sets the given parameters values as the current values for those parameters. Other parameters are unchanged. @param [Array<Puppet::Parameter] the parameters with values that should be set @return [void]

# File lib/puppet/provider.rb, line 573
def set(params)
  params.each do |param, value|
    @property_hash[param.intern] = value
  end
end

@return [String] Returns a human readable string with information about the resource and the provider.

# File lib/puppet/provider.rb, line 580
def to_s
  "#{@resource}(provider=#{self.class.name})"
end