NimYAML - YAML implementation for Nim

NimYAML is a YAML implementation in Nim. It aims to implement the complete YAML 1.2 specification and has a streaming interface that makes it possible to parse YAML input sequentially, without loading all data into memory at once. It is able to automatically serialize Nim object to YAML and deserialize them again.

Attention: NimYAML is work in progress. There is no release yet, and some features are highly experimental.

Quickstart

Using the sequential parser

import yaml

let input = """
an integer: 42
a boolean: yes
a list:
 - 3.14159
 - !!str 23
 - null
"""

var
  parser = newParser(coreTagLibrary())
  events = parser.parse(newStringStream(input))

for event in events():
  echo $event

Output:

yamlStartDocument()
yamlStartMap(tag=?)
yamlScalar(tag=?, typeHint=yTypeUnknown, content="an integer")
yamlScalar(tag=?, typeHint=yTypeInteger, content="42")
yamlScalar(tag=?, typeHint=yTypeUnknown, content="a boolean")
yamlScalar(tag=?, typeHint=yTypeBoolTrue, content="yes")
yamlScalar(tag=?, typeHint=yTypeUnknown, content="a list")
yamlStartSequence(tag=?)
yamlScalar(tag=?, typeHint=yTypeFloat, content="3.14159")
yamlScalar(tag=!!str, typeHint=yTypeInteger, content="23")
yamlScalar(tag=?, typeHint=yTypeNull, content="null")
yamlEndSequence()
yamlEndMap()
yamlEndDocument()

Dumping YAML

import yaml, streams

proc example(): YamlStream =
  result = iterator(): YamlStreamEvent =
    yield startDocEvent()
    yield startMapEvent(mayHaveKeyObjects = false)
    yield scalarEvent("an integer")
    yield scalarEvent("42", tag = yTagInteger)
    yield scalarEvent("a list")
    yield startSeqEvent(tag = yTagSequence)
    yield scalarEvent("item", tag = yTagString)
    yield scalarEvent("no", tag = yTagBoolean, typeHint = yTypeBoolFalse)
    yield scalarEvent("")
    yield endSeqEvent()
    yield scalarEvent("a float")
    yield scalarEvent("3.14159", tag = yTagFloat)
    yield endMapEvent()
    yield endDocEvent()

present(example(), newFileStream(stdout), coreTagLibrary(), yDumpBlockOnly)
echo "\n\n"
present(example(), newFileStream(stdout), coreTagLibrary(), yDumpCanonical)
echo "\n\n"
present(example(), newFileStream(stdout), coreTagLibrary(), yDumpJson)

Output:

%YAML 1.2
---
an integer: !!int 42
a list: !!seq
  - !!str item
  - !!bool no
  - ""
a float: !!float 3.14159


%YAML 1.2
---
{
  ? "an integer"
  : !!int "42",
  ? "a list"
  : !!seq [
    !!str "item",
    !!bool "no",
    ""
  ],
  ? "a float"
  : !!float "3.14159"
}


{
  "an integer": 42,
  "a list": [
    "item",
    false,
    ""
  ],
  "a float": 3.14159
}

Using Nim Type Serialization

Attention: This feature is highly experimental!

import yaml.serialization
import tables

make_serializable:
    type
        Person = object
            firstname, surname: string
            age: int
            additionalAttributes: Table[string, string]

# loading

let input = """
-
  firstname: Peter
  surname: Pan
  age: 12
  additionalAttributes:
    canFly: yes
    location: Neverland
-
  firstname: Karl
  surname: Koch
  age: 23
  additionalAttributes:
    location: Celle
    occupation: Hacker
"""

var persons: seq[Person]
load(newStringStream(input), persons)
assert persons[0].surname == "Pan"
assert persons[1].additionalAttributes["location"] == "Celle"

# dumping

dump(persons, newFileStream(stdout), ypsCanonical)

Output:

%YAML 1.2
--- !nim:seq(nim:Person)
[
  !nim:Person {
    ? !!str "firstname"
    : !!str "Peter",
    ? !!str "surname"
    : !!str "Pan",
    ? !!str "age"
    : !!int "12",
    ? !!str "additionalAttributes"
    : {
      ? !!str "canFly"
      : !!str "yes",
      ? !!str "location"
      : !!str "Neverland"
    }
  },
  !nim:Person {
    ? !!str "firstname"
    : !!str "Karl",
    ? !!str "surname"
    : !!str "Koch",
    ? !!str "age"
    : !!int "23",
    ? !!str "additionalAttributes"
    : {
      ? !!str "occupation"
      : !!str "Hacker",
      ? !!str "location"
      : !!str "Celle"
    }
  }
]

TODO list

• Documentation:
  • Document yaml.serialization
  • Setup github pages site with proper API documentation
• Lexer:
  • Add type hints for more scalar types
• Parser:
  • Properly handle leading spaces in block scalars
• Serialization:
  • Support for more standard library types
  • Support for ref and ptr types
  • Support for anchors and aliases (requires ref and ptr types)
  • Support polymorphism (requires ref and ptr types)
  • Support variant objects
  • Support transient fields (i.e. fields that will not be (de-)serialized on objects and tuples)
  • Make it possible for user to define a tag URI for custom types
  • Use concept type class Serializable or something
  • Check for and avoid name clashes when generating local tags for custom object types.
  • Possibly use genSym for predefined and generated yamlTag procs because they are an implementation detail and should not be visible to the caller. same goes for lazyLoadTag and safeLoadUri.
• General:
  • Proper error handling (do not use ValueError for everything)
  • Proper error handling, seriously
  • Document exceptions with raises pragmas in code

License

MIT