Overview

Psych is a YAML parser and emitter. Psych leverages libyaml [Home page: pyyaml.org/wiki/LibYAML] or [git repo: github.com/yaml/libyaml] for its YAML parsing and emitting capabilities. In addition to wrapping libyaml, Psych also knows how to serialize and de-serialize most Ruby objects to and from the YAML format.

I NEED TO PARSE OR EMIT YAML RIGHT NOW!

# Parse some YAML
Psych.load("--- foo") # => "foo"

# Emit some YAML
Psych.dump("foo")     # => "--- foo\n...\n"
{ :a => 'b'}.to_yaml  # => "---\n:a: b\n"

Got more time on your hands? Keep on reading!

YAML Parsing

Psych provides a range of interfaces for parsing a YAML document ranging from low level to high level, depending on your parsing needs. At the lowest level, is an event based parser. Mid level is access to the raw YAML AST, and at the highest level is the ability to unmarshal YAML to Ruby objects.

YAML Emitting

Psych provides a range of interfaces ranging from low to high level for producing YAML documents. Very similar to the YAML parsing interfaces, Psych provides at the lowest level, an event based system, mid-level is building a YAML AST, and the highest level is converting a Ruby object straight to a YAML document.

High-level API

Parsing

The high level YAML parser provided by Psych simply takes YAML as input and returns a Ruby data structure. For information on using the high level parser see Psych.load

Reading from a string

Psych.safe_load("--- a")             # => 'a'
Psych.safe_load("---\n - a\n - b")   # => ['a', 'b']
# From a trusted string:
Psych.load("--- !ruby/range\nbegin: 0\nend: 42\nexcl: false\n") # => 0..42

Reading from a file

Psych.safe_load_file("data.yml", permitted_classes: [Date])
Psych.load_file("trusted_database.yml")

Exception handling

begin
  # The second argument changes only the exception contents
  Psych.parse("--- `", "file.txt")
rescue Psych::SyntaxError => ex
  ex.file    # => 'file.txt'
  ex.message # => "(file.txt): found character that cannot start any token"
end

Emitting

The high level emitter has the easiest interface. Psych simply takes a Ruby data structure and converts it to a YAML document. See Psych.dump for more information on dumping a Ruby data structure.

Writing to a string

# Dump an array, get back a YAML string
Psych.dump(['a', 'b'])  # => "---\n- a\n- b\n"

# Dump an array to an IO object
Psych.dump(['a', 'b'], StringIO.new)  # => #<StringIO:0x000001009d0890>

# Dump an array with indentation set
Psych.dump(['a', ['b']], :indentation => 3) # => "---\n- a\n-  - b\n"

# Dump an array to an IO with indentation set
Psych.dump(['a', ['b']], StringIO.new, :indentation => 3)

Writing to a file

Currently there is no direct API for dumping Ruby structure to file:

File.open('database.yml', 'w') do |file|
  file.write(Psych.dump(['a', 'b']))
end

Mid-level API

Parsing

Psych provides access to an AST produced from parsing a YAML document. This tree is built using the Psych::Parser and Psych::TreeBuilder. The AST can be examined and manipulated freely. Please see Psych::parse_stream, Psych::Nodes, and Psych::Nodes::Node for more information on dealing with YAML syntax trees.

Reading from a string

# Returns Psych::Nodes::Stream
Psych.parse_stream("---\n - a\n - b")

# Returns Psych::Nodes::Document
Psych.parse("---\n - a\n - b")

Reading from a file

# Returns Psych::Nodes::Stream
Psych.parse_stream(File.read('database.yml'))

# Returns Psych::Nodes::Document
Psych.parse_file('database.yml')

Exception handling

begin
  # The second argument changes only the exception contents
  Psych.parse("--- `", "file.txt")
rescue Psych::SyntaxError => ex
  ex.file    # => 'file.txt'
  ex.message # => "(file.txt): found character that cannot start any token"
end

Emitting

At the mid level is building an AST. This AST is exactly the same as the AST used when parsing a YAML document. Users can build an AST by hand and the AST knows how to emit itself as a YAML document. See Psych::Nodes, Psych::Nodes::Node, and Psych::TreeBuilder for more information on building a YAML AST.

Writing to a string

# We need Psych::Nodes::Stream (not Psych::Nodes::Document)
stream = Psych.parse_stream("---\n - a\n - b")

stream.to_yaml # => "---\n- a\n- b\n"

Writing to a file

# We need Psych::Nodes::Stream (not Psych::Nodes::Document)
stream = Psych.parse_stream(File.read('database.yml'))

File.open('database.yml', 'w') do |file|
  file.write(stream.to_yaml)
end

Low-level API

Parsing

The lowest level parser should be used when the YAML input is already known, and the developer does not want to pay the price of building an AST or automatic detection and conversion to Ruby objects. See Psych::Parser for more information on using the event based parser.

Reading to Psych::Nodes::Stream structure

parser = Psych::Parser.new(TreeBuilder.new) # => #<Psych::Parser>
parser = Psych.parser                       # it's an alias for the above

parser.parse("---\n - a\n - b")             # => #<Psych::Parser>
parser.handler                              # => #<Psych::TreeBuilder>
parser.handler.root                         # => #<Psych::Nodes::Stream>

Receiving an events stream

recorder = Psych::Handlers::Recorder.new
parser = Psych::Parser.new(recorder)

parser.parse("---\n - a\n - b")
recorder.events # => [list of [event, args] lists]
                # event is one of: Psych::Handler::EVENTS
                # args are the arguments passed to the event

Emitting

The lowest level emitter is an event based system. Events are sent to a Psych::Emitter object. That object knows how to convert the events to a YAML document. This interface should be used when document format is known in advance or speed is a concern. See Psych::Emitter for more information.

Writing to a Ruby structure

Psych.parser.parse("--- a")       # => #<Psych::Parser>

parser.handler.first              # => #<Psych::Nodes::Stream>
parser.handler.first.to_ruby      # => ["a"]

parser.handler.root.first         # => #<Psych::Nodes::Document>
parser.handler.root.first.to_ruby # => "a"

# You can instantiate an Emitter manually
Psych::Visitors::ToRuby.new.accept(parser.handler.root.first)
# => "a"

If an object defines encode_with, then an instance of Psych::Coder will be passed to the method when the object is being serialized. The Coder automatically assumes a Psych::Nodes::Mapping is being emitted. Other objects like Sequence and Scalar may be emitted if seq= or scalar= are called, respectively.

Psych::Handler is an abstract base class that defines the events used when dealing with Psych::Parser. Clients who want to use Psych::Parser should implement a class that inherits from Psych::Handler and define events that they can handle.

Psych::Handler defines all events that Psych::Parser can possibly send to event handlers.

See Psych::Parser for more details

YAML event parser class. This class parses a YAML document and calls events on the handler that is passed to the constructor. The events can be used for things such as constructing a YAML AST or deserializing YAML documents. It can even be fed back to Psych::Emitter to emit the same document that was parsed.

See Psych::Handler for documentation on the events that Psych::Parser emits.

Here is an example that prints out ever scalar found in a YAML document:

# Handler for detecting scalar values
class ScalarHandler < Psych::Handler
  def scalar value, anchor, tag, plain, quoted, style
    puts value
  end
end

parser = Psych::Parser.new(ScalarHandler.new)
parser.parse(yaml_document)

Here is an example that feeds the parser back in to Psych::Emitter. The YAML document is read from STDIN and written back out to STDERR:

parser = Psych::Parser.new(Psych::Emitter.new($stderr))
parser.parse($stdin)

Psych uses Psych::Parser in combination with Psych::TreeBuilder to construct an AST of the parsed YAML document.

Scan scalars for built in types

Psych::Stream is a streaming YAML emitter. It will not buffer your YAML, but send it straight to an IO.

Here is an example use:

stream = Psych::Stream.new($stdout)
stream.start
stream.push({:foo => 'bar'})
stream.finish

YAML will be immediately emitted to $stdout with no buffering.

Psych::Stream#start will take a block and ensure that Psych::Stream#finish is called, so you can do this form:

stream = Psych::Stream.new($stdout)
stream.start do |em|
  em.push(:foo => 'bar')
end

This class works in conjunction with Psych::Parser to build an in-memory parse tree that represents a YAML document.

Example

parser = Psych::Parser.new Psych::TreeBuilder.new
parser.parse('--- foo')
tree = parser.handler.root

See Psych::Handler for documentation on the event methods used in this class.

Overview

When using Psych.load to deserialize a YAML document, the document is translated to an intermediary AST. That intermediary AST is then translated in to a Ruby object graph.

In the opposite direction, when using Psych.dump, the Ruby object graph is translated to an intermediary AST which is then converted to a YAML document.

Psych::Nodes contains all of the classes that make up the nodes of a YAML AST. You can manually build an AST and use one of the visitors (see Psych::Visitors) to convert that AST to either a YAML document or to a Ruby object graph.

Here is an example of building an AST that represents a list with one scalar:

# Create our nodes
stream = Psych::Nodes::Stream.new
doc    = Psych::Nodes::Document.new
seq    = Psych::Nodes::Sequence.new
scalar = Psych::Nodes::Scalar.new('foo')

# Build up our tree
stream.children << doc
doc.children    << seq
seq.children    << scalar

The stream is the root of the tree. We can then convert the tree to YAML:

stream.to_yaml => "---\n- foo\n"

Or convert it to Ruby:

stream.to_ruby => [["foo"]]

YAML AST Requirements

A valid YAML AST must have one Psych::Nodes::Stream at the root. A Psych::Nodes::Stream node must have 1 or more Psych::Nodes::Document nodes as children.

Psych::Nodes::Document nodes must have one and only one child. That child may be one of:

• Psych::Nodes::Sequence

• Psych::Nodes::Mapping

• Psych::Nodes::Scalar

Psych::Nodes::Sequence and Psych::Nodes::Mapping nodes may have many children, but Psych::Nodes::Mapping nodes should have an even number of children.

All of these are valid children for Psych::Nodes::Sequence and Psych::Nodes::Mapping nodes:

• Psych::Nodes::Sequence

• Psych::Nodes::Mapping

• Psych::Nodes::Scalar

• Psych::Nodes::Alias

Psych::Nodes::Scalar and Psych::Nodes::Alias are both terminal nodes and should not have any children.

Configuration options for dumping YAML.

This handler will capture an event and record the event. Recorder events are available vial Psych::Handlers::Recorder#events.

For example:

recorder = Psych::Handlers::Recorder.new
parser = Psych::Parser.new recorder
parser.parse '--- foo'

recorder.events # => [list of events]

# Replay the events

emitter = Psych::Emitter.new $stdout
recorder.events.each do |m, args|
  emitter.send m, *args
end