package pratter
Install
Dune Dependency
Authors
Maintainers
Sources
sha256=7dd9a7b970d3f660a957a54ae257ac2228f8203a133f8ceb7a73ce61f0663833
sha512=040a36c6d61761701d7f93d620f5a466caa0d578f6dfb0a93028fd8f693b4abef8b22c9b7695971e21ce347865dfc84f26cc032f1c4bfc070ee6e87cd9a72318
Description
Pratter is a library that provides a parser that transforms streams of terms to applied terms. Terms may contain infix or prefix operators and native applications. The parser is based on the Pratt parsing algorithm and extended to handle term application and non associative operators.
Published: 10 May 2022
README
Pratter: A parser for terms with operators and applications
Pratter is a library that provides a parser that transforms streams of terms to applied terms. Terms may contain infix or prefix operators and applications. The parser is an extension of the Pratt parsing algorithm.
Examples can be seen in the tests inside the t/
folder.
Example: simple terms
The idea is that terms are made of symbols and applications and some symbol can be declared to be infix or prefix operators.
Start by defining the terms,
type term = Appl of term * term  Symb of string
and the data structure that maps symbols identifiers to operators properties,
type table = {
unary : (string * Pratter.priority) list
; binary : (string * (Pratter.priority * Pratter.associativity)) list
}
let empty : table = { unary = [] ; binary = [] }
Next, define a module to pack these two types, and two functions:
get
that is able to retrieve the properties of a symbol if it is declared as an operator,make_appl
that creates an application out of two terms.
module Terms : Pratter.SUPPORT with type term = term and type table = table =
struct
type nonrec term = term
type nonrec table = table
let get { unary; binary } (t: term) =
match t with
 Symb id > (
try Some (Pratter.Una, List.assoc id unary)
with Not_found > (
try
let bp, assoc = List.assoc id binary in
Some (Bin assoc, bp)
with Not_found > None ) )
 _ > None
let make_appl t u = Appl (t, u)
end
module Parser = Pratter.Make (Terms)
Then that's it, we can parse streams of terms with operators. For instance, assume that we want to parse x + y * z
where
+
and*
are infix operators, with*
having a higher binding power than+
;
is a unary operator having a higher binding power than+
and*
.
Create a table holding these operators:
let tbl =
{ unary = [ "", 1.0 ]
; binary = [ ("+", (0.5, Pratter.Left)) ; ("*", (0.6, Pratter.Left)) ] }
Priority (also called binding power) can be any float, and associativity may be Pratter.Left
, Pratter.Right
or Pratter.Neither
.
Finally parse the input using Parser.expression
:
let input = [ Symb "x"; Symb "+"; Symb "y"; Symb "*"; Symb ""; Symb "z"]
Parser.expression tbl (Stream.of_list input)
we obtain the term x + (y * (z))
representend by
Appl (Appl (Symb "+", Symb "x"),
Appl (Appl (Symb "*", Symb "y"), Appl (Symb "", Symb "z")))
Dependencies (2)
 camlpstreams

dune
>= "2.7"
Dev Dependencies (5)

odoc
withdoc

qcheckalcotest
withtest

qcheck
withtest

alcotest
withtest

ocaml
>= "4.02"  withtest & >= "4.03"
Used by (1)

lambdapi
>= "2.0.0" & < "2.2.1"
Conflicts
None