[][src]Crate lexpr

This crate provides facilities for parsing, printing and manipulating S-expression data. S-expressions are the format used to represent code and data in the Lisp language family.

((name . "John Doe")
 (age . 43)
 (address
  (street "10 Downing Street")
  (city "London"))
 (phones "+44 1234567" "+44 2345678"))

lexpr also supports more complex types; including keywords and configurable tokens for true, false and nil, by default using Scheme syntax:

(define-class rectangle ()
 (width
   #:init-value #nil ;; Nil value
   #:settable #t     ;; true
   #:guard (> width 10)
 )
 (height
   #:init-value 10
   #:writable #f ;; false
  ))

Note that keywords, and the corresponding #: notation, is not part of standard Scheme, but is supported by lexpr's default parser settings.

There are three common ways that you might find yourself needing to work with S-expression data in Rust:

Only the first two items of this list are handled by lexpr; for conversion from and to statically typed Rust data structures see the serde-lexpr crate.

Operating on dynamically typed S-expression data

Any valid S-expression can be manipulated using the Value data structure.

Constructing S-expression values

 use lexpr::{Value, parse::Error};

// Some s-expressions a &str.
let data = r#"((name . "John Doe")
               (age . 43)
               (phones "+44 1234567" "+44 2345678"))"#;

// Parse the string of data into lexpr::Value.
let v = lexpr::from_str(data)?;

// Access parts of the data by indexing with square brackets.
println!("Please call {} at the number {}", v["name"], v["phones"][1]);

Ok(())

What are S-expressions?

S-expressions, as mentioned above, are the notation used by various dialects of Lisp to represent data (and code). As a data format, it is roughly comparable to JSON (JavaScript Object Notation), but syntactically more lightweight. Also, JSON is designed for consumption and generation by machines, which is reflected by the fact that it does not specify a syntax for comments. S-expressions on the other hand, are intended to be written and read by humans as well as machines. In this respect, they are more like YAML, but have a simpler and less syntactically rigid structure. For example, indentation does not convey any information to the parser, but is used only to allow for easier digestion by humans.

Different Lisp dialects have notational differences for some data types, and some may lack specific data types completely. This section tries to give an overview over the different types of values representable by the Value data type and how it relates to different Lisp dialects. All examples are given in the syntax used in Guile Scheme implementation.

The parser and serializer implementation in lexpr can be tailored to parse and generate S-expression data in various "dialects" in use by different Lisp variants; the aim is to cover large parts of R6RS and R7RS Scheme with some Guile and Racket extensions, as well as Emacs Lisp.

In the following, the S-expression values that are modeled by lexpr are introduced, In general, S-expression values can be split into the two categories primitive types and compound types.

Primitive types

Primitive, or non-compound types are those that can not recursively contain arbitrary other values. Numbers, strings and booleans fall into this category.

Symbols and keywords

Lisp has a data type not commonly found in other languages, namely "symbols". A symbol is conceptually similar to identifiers in other languages, but allow for a much richer set of characters than typically allowed for identifiers in other languages. Also, identifiers in other languages can usually not be used in data; Lisps expose them as a primitive data type, a result of the homoiconicity of the Lisp language family.

this-is-a-symbol ; A single symbol, dashes are allowed
another.symbol   ; Periods are allowed as well
foo$bar!<_>?     ; As are quite a few other characters

Another data type, present in some Lisp dialects, such as Emacs Lisp, Common Lisp, and several Scheme implementations, are keywords. These are also supported by lexpr. Keywords are very similiar to symbols, but are typically prefixed by : or #: and are used for different purposes in the language.

#:foo ; A keyword named "foo", written in Guile/Racket notation
:bar  ; A keyword named "bar", written in Emacs Lisp or Common Lisp notation

Booleans

While Scheme has a primitive boolean data type, more traditional Lisps such as Emacs Lisp and Common Lisp do not; they instead use the symbols t and nil to represent boolean values. Using parser options, lexpr allows to parse these symbols as booleans, which may be desirable in some circumstances, as booleans are simpler to handle than symbols.

#t ; The literal representing true
#f ; The literal representing false

The empty list and "nil"

In traditional Lisps, the end of list is represented as by a special atom written as nil. In Scheme, the empty list is an atom written as (), and there nil is just a regular symbol. Both nil and the empty list are present and distinguishable in lexpr.

Numbers

Numbers are represented by the Number abstract data type. It can handle signed and unsigned integers, each up to 64 bit size, as well as floating point numbers. The Scheme syntax for hexadecimal, octal, and binary literals is supported.

1 -4 3.14  ; A postive, negative, and a floating point number
#xDEADBEEF ; An integer written using decimal notation
#o0677     ; Octal
#b10110    ; Binary

Scheme has an elaborate numerical type hierarchy (called "numeric tower"), which supports fractionals, numbers of arbitrary size, and complex numbers. These more advanced number types are not yet supported by lexpr.

Characters

Characters are unicode codepoints, represented by Rust's char data type embedded in the Value::Char variant.

Strings

"Hello World!"

Lists

Lists are a sequence of values, of either atoms or lists. In fact, Lisp does not have a "real" list data type, but instead lists are represented by chains of so-called "cons cells", which are used to form a singly-linked list, terminated by the empty list (or nil in tradional Lisps). It is also possible for the terminator to not be the empty list, but instead be af an arbitrary other data type. In this case, the list is refered to as an "improper" or "dotted" list. Here are some examples:

("Hello" "World")   ; A regular list
;; A list having with another, single-element, list as
;; its second item
("Hello" ("World"))
(1 . 2) ; A cons cell, represented as an improper list by `lexpr`
(1 2 . 3) ; A dotted (improper) list

Lists are not only used to represent sequences of values, but also associative arrays, also known as maps. A map is represented as a list containing cons cells, where the first field of each cons cell, called car, for obscure historical reasons, is the key, and the second field (cdr) of the cons cell is the associated value.

;; An association list with the symbols `a` and `b` as keys
((a . 42) (b . 43))

Vectors

In contrast to lists, which are represented as singly-linked chains of "cons cells", vectors allow O(1) indexing, and thus are quite similar to Rusts Vec datatype.

#(1 2 "three") ; A vector in Scheme notation

Byte vectors

Byte vectors are similar to regular vectors, but are uniform: each element only holds a single byte, i.e. an exact integer in the range of 0 to 255, inclusive.

#u8(41 42 43) ; A byte vector

Modules

cons

List "cons cell" data type and accompanying iterator types.

datum

S-expression values including source location.

number

Dynamically typed number type.

parse

S-expression parser and options.

print

Converting S-expression values into text.

value

The Value enum, a dynamically typed way of representing any valid S-expression value.

Macros

sexp

Construct a Value using syntax similar to regular S-expressions.

Structs

Cons

A Lisp "cons cell".

Datum

Combines an S-expression value with location information.

Number

Represents an S-expression number, whether integer or floating point.

Parser

Parser for the S-expression text representation.

Printer

A printer for S-expression values.

Enums

Value

Represents an S-expression value.

Traits

Index

A type that can be used to index into a lexpr::Value.

Functions

from_reader

Parse a value from an IO stream of S-expressions, using the default parser options.

from_reader_custom

Parse a value from an IO stream containing a single S-expression.

from_slice

Parse a value from bytes representing a single S-expressions, using the default parser options.

from_slice_custom

Parse a value from bytes representing a single S-expression.

from_str

Parse a value from a string slice representing a single S-expressions, using the default parser options.

from_str_custom

Parse a value from a string slice representing a single S-expression.

to_string

Serialize the given value an S-expression string, using the default printer options.

to_string_custom

Serialize the given value an S-expression string.

to_vec

Serialize the given value as byte vector containing S-expression text, using the default printer options.

to_vec_custom

Serialize the given value as byte vector containing S-expression text.

to_writer

Serialize the given value value as S-expression text into the IO stream, using the default printer options.

to_writer_custom

Serialize the given value value as S-expression text into the IO stream.