Data.SCargot.Repr

Documentation

This module contains several different representations for s-expressions. The s-cargot library underlying uses the SExpr type as its representation type, which is a binary tree representation with an arbitrary type for its leaves.

This type is not always convenient to manipulate in Haskell code, this module defines two alternate representations which turn a sequence of nested right-branching cons pairs into Haskell lists: that is to say, they transform between

SCons a (SCons b (SCons c SNil))  <=>  RSList [a, b, c]

These two types differ in how they handle non-well-formed lists, i.e. lists that end with an atom. The RichSExpr format handles this with a special constructor for lists that end in an atom:

SCons a (SCons b (SAtom c))  <=>  RSDotted [a, b] c

On the other hand, the WellFormedSExpr type elects not to handle this case. This is unusual for Lisp source code, but is a reasonable choice for configuration or data storage formats that use s-expressions, where non-well-formed lists would be an unnecessary complication.

To make working with these types less verbose, there are other modules that export pattern aliases and helper functions: these can be found at Data.SCargot.Repr.Basic, Data.SCargot.Repr.Rich, and Data.SCargot.Repr.WellFormed.

data SExpr atom

All S-Expressions can be understood as a sequence of cons cells (represented here by SCons), the empty list nil (represented by SNil) or an atom.

Constructors

SCons (SExpr atom) (SExpr atom)

SAtom atom

SNil

Instances

Functor SExpr

Foldable SExpr

Traversable SExpr

IsList (SExpr atom)

Eq atom => Eq (SExpr atom)

Data atom => Data (SExpr atom)

Read atom => Read (SExpr atom)

Show atom => Show (SExpr atom)

IsString atom => IsString (SExpr atom)

type Item (SExpr atom) = SExpr atom

Rich SExpr representation

data RichSExpr atom

Sometimes the cons-based interface is too low level, and we'd rather have the lists themselves exposed. In this case, we have RSList to represent a well-formed cons list, and RSDotted to represent an improper list of the form (a b c . d). This representation is based on the structure of the parsed S-Expression, and not on how it was originally represented: thus, (a . (b)) is going to be represented as RSList[RSAtom a, RSAtom b] despite having been originally represented as a dotted list.

Constructors

RSList [RichSExpr atom]

RSDotted [RichSExpr atom] atom

RSAtom atom

Instances

Functor RichSExpr

Foldable RichSExpr

Traversable RichSExpr

IsList (RichSExpr atom)

Eq atom => Eq (RichSExpr atom)

Data atom => Data (RichSExpr atom)

Read atom => Read (RichSExpr atom)

Show atom => Show (RichSExpr atom)

IsString atom => IsString (RichSExpr atom)

type Item (RichSExpr atom) = RichSExpr atom

toRich :: SExpr atom -> RichSExpr atom

It should always be true that

fromRich (toRich x) == x

and that

toRich (fromRich x) == x

fromRich :: RichSExpr atom -> SExpr atom

This follows the same laws as toRich.

Well-Formed SExpr representation

data WellFormedSExpr atom

A well-formed s-expression is one which does not contain any dotted lists. This means that not every value of SExpr a can be converted to a WellFormedSExpr a, although the opposite is fine.

Constructors

WFSList [WellFormedSExpr atom]

WFSAtom atom

Instances

Functor WellFormedSExpr

Foldable WellFormedSExpr

Traversable WellFormedSExpr

IsList (WellFormedSExpr atom)

Eq atom => Eq (WellFormedSExpr atom)

Data atom => Data (WellFormedSExpr atom)

Read atom => Read (WellFormedSExpr atom)

Show atom => Show (WellFormedSExpr atom)

IsString atom => IsString (WellFormedSExpr atom)

type Item (WellFormedSExpr atom) = WellFormedSExpr atom

toWellFormed :: SExpr atom -> Either String (WellFormedSExpr atom)

This will be Nothing if the argument contains an improper list. It should hold that

toWellFormed (fromWellFormed x) == Right x

and also (more tediously) that

case toWellFormed x of
  Left _  -> True
  Right y -> x == fromWellFormed y

fromWellFormed :: WellFormedSExpr atom -> SExpr atom

Convert a WellFormedSExpr back into a SExpr.