Parsit

Description

This library provides a very simple and lightweight parser (recursive descendant ll(1)) to combine and express a grammar.

The library uses Logos as a lexical analyzer and tokenizer.

The premise

This library major incentives were:

• lightweight : very small and does not require a deep dive
• transparency : literally 3 structs with a handful of methods
• speed : good speed (with a gratitude to Logos)

The steps to implement

Create a set of tokens using Logos

Add logos to dependency

    logos = "*"

use logos::Logos;

#[derive(Logos, Debug, PartialEq)]
enum Token {
    // Tokens can be literal strings, of any length.
    #[token("fast")]
    Fast,

    #[token(".")]
    Period,

    // Or regular expressions.
    #[regex("[a-zA-Z]+")]
    Text,

    // Logos requires one token variant to handle errors,
    // it can be named anything you wish.
    #[error]
    // We can also use this variant to define whitespace,
    // or any other matches we wish to skip.
    #[regex(r"[ \t\n\f]+", logos::skip)]
    Error,
}

Create a parser that will be able to parse the given set of tokens

The library provides Parsit<'a,T> instance that encompasses a set of tokens and auxiliary methods

struct Parser<'a> {
    inner: Parsit<'a, Token<'a>>,
}

Implement a parsing functions using Parsit instance and auxiliary methods from the Step

The helpers:

• the macros token! that alleviates comparing and matching single tokens
• methods then, then_zip and others from Step
• methods one_or_more, zero_or_more from Parsit

Transform the result into Result<Structure, ParserError<'a>>

      fn text(&self, pos: usize) -> Result<Vec<Sentence<'a>>, ParseError<'a>> {
    self.inner.zero_or_more(pos, |p| self.sentence(p)).into()
}

Complete example

  use crate::parser::Parsit;
use crate::token;
use crate::step::Step;
use crate::parser::EmptyToken;
use crate::error::ParseError;
use logos::Logos;


#[derive(Logos, Debug, Copy, Clone, PartialEq)]
pub enum Token<'a> {
    #[regex(r"[a-zA-Z-]+")]
    Word(&'a str),

    #[token(",")]
    Comma,
    #[token(".")]
    Dot,

    #[token("!")]
    Bang,
    #[token("?")]
    Question,

    #[regex(r"[ \t\r\n\u000C\f]+", logos::skip)]
    Whitespace,
    #[error]
    Error,
}

#[derive(Debug, Copy, Clone, PartialEq)]
enum Item<'a> {
    Word(&'a str),
    Comma,
}

#[derive(Debug, Clone, PartialEq)]
enum Sentence<'a> {
    Sentence(Vec<Item<'a>>),
    Question(Vec<Item<'a>>),
    Exclamation(Vec<Item<'a>>),
}

struct Parser<'a> {
    inner: Parsit<'a, Token<'a>>,
}

impl<'a> Parser<'a> {
    fn new(text: &'a str) -> Parser<'a> {
        let delegate: Parsit<Token> = Parsit::new(text).unwrap();
        Parser { inner: delegate }
    }

    fn sentence(&self, pos: usize) -> Step<'a, Sentence<'a>> {
        let items = |p| self.inner.one_or_more(p, |p| self.word(p));

        let sentence = |p| items(p)
            .then_zip(|p| token!(self.inner.token(p) => Token::Dot))
            .take_left()
            .map(Sentence::Sentence);

        let exclamation = |p| items(p)
            .then_zip(|p| token!(self.inner.token(p) => Token::Bang))
            .take_left()
            .map(Sentence::Exclamation);
        let question = |p| items(p)
            .then_zip(|p| token!(self.inner.token(p) => Token::Question))
            .take_left()
            .map(Sentence::Question);

        sentence(pos)
            .or_from(pos)
            .or(exclamation)
            .or(question)
            .into()
    }
    fn word(&self, pos: usize) -> Step<'a, Item<'a>> {
        token!(self.inner.token(pos) =>
                     Token::Word(v) => Item::Word(v),
                     Token::Comma => Item::Comma
            )
    }
    fn text(&self, pos: usize) -> Result<Vec<Sentence<'a>>, ParseError<'a>> {
        self.inner.zero_or_more(pos, |p| self.sentence(p)).into()
    }
}


#[test]
fn test() {
    let parser = Parser::new(r#"
            I have a strange addiction,
            It often sets off sparks!
            I really cannot seem to stop,
            Using exclamation marks!
            Anyone heard of the interrobang?
            The poem is for kids.
        "#);

    let result = parser.text(0).unwrap();
    println!("{:?}", result);
}

The base auxiliary methods

On parser

• token - gives a possibility to pull out a curren token
• one_or_more - gives a one or more semantic
• zero_or_more - gives a zero or more semantic
• validate_eof - ensure the parser reaches end of the input

Macros

• token! - parses the current token. In general, it is used the following token!(p.token(pos) => T::Bang => "!")
• wrap! - implements a simple pattern in grammar like left value right, for instance [1,2,3] or (a,b)
  • can handle the default value like wrap!(0 => left; value or default; right)
  • can handle the option value like wrap!(0 => left; value ?; right)
• seq! - implements a simple pattern of sequence like el sep el ... , for instance 1,2,3
  • can have a , at the end signaling the separator can be at the ned of the seq like 1,2,3 (,)?

On step

To alternate

• or - gives an alternative in a horizon of one token
• or_from - gives a backtracking option

To combine

• then - gives a basic combination with a next rule omitting the current one
• then_zip - combines a current result and a next one into a pair
• then_skip - parses the next one but drops the result keeping only current one
• then_or_none -combines a next one in an option with a current one or return a none otherwise

To collect

• take_left - drops a right value from a pair
• take_right - drops a left value from a pair
• merge - merge a value into a list
• to_map - transforms a list of pairs into a map

To transform

• or_val - replaces a value with a default value if it is not presented
• or_none - replaces a value with a none if it is not presented

To work with value

• ok - transforms a value into an option
• error - transforms an error into an option
• map - transforms a value
• combine - combines a value with another value from a given step
• validate - validates a given value and transforms into an error if a validation failed

To print

• print - print a step
• print_with - print a step with a given prefix
• print_as - print a step with a transformation of value
• print_with_as - print a step with a transformation of value with a given prefix
• parsit.env - Prints a position and env from the source text(with a radius of 3 tokens )

Testing

Lexer

To test a lexer there are methods from crate::parsit::test::lexer_test::* for service

 use logos::Logos;
use crate::parsit::test::lexer_test::*;

#[derive(Logos, Debug, PartialEq)]
pub enum T<'a> {
    #[regex(r"[a-zA-Z-]+")]
    Word(&'a str),

    #[token(",")]
    Comma,
    #[token(".")]
    Dot,

    #[token("!")]
    Bang,
    #[token("?")]
    Question,

    #[regex(r"[ \t\r\n]+", logos::skip)]
    Whitespace,
    #[error]
    Error,
}

#[test]
fn test() {
    expect::<T>("abc, bcs!", vec![T::Word("abc"), T::Comma, T::Word("bcs"), T::Bang]);
    expect_succeed::<T>("abc, bcs!");
    expect_failed::<T>("abc, bcs >> !");
    expect_failed_with::<T, _>("abc, bcs > !", |e| e.is_bad_token_on(">"));
}

Parser

To test a parser there are methods from crate::parsit::test::parser_test::* for service

• expect : expect to parse a given value
• expect_or_env : expect to parse a given value otherwise it will print an env (parsit.env)
• expect_pos : expect to parse and get a cursor on a given pos
• expect_pos_or_env : expect to parse and get a cursor on a given pos otherwise it will print an env (parsit.env)
• fail : should fail parsing
• fail_on : should fail parsing on a given position

use logos::Logos;
use crate::parsit::test::parser_test::fail;
use crate::parsit::test::parser_test::parsit;
use crate::parsit::token;
use crate::parsit::parser::Parsit;
use crate::parsit::step::Step;

#[derive(Logos, Debug, PartialEq)]
pub enum T<'a> {
    #[regex(r"[a-zA-Z-]+")]
    Word(&'a str),

    #[token(",")]
    Comma,
    #[token(".")]
    Dot,

    #[token("!")]
    Bang,
    #[token("?")]
    Question,

    #[regex(r"[ \t\r\n]+", logos::skip)]
    Whitespace,
    #[error]
    Error,
}

#[test]
fn test_expect() {
    let p = parsit("abc!");
    let bang = |pos: usize| token!(p.token(pos) => T::Bang => "!");
    let word = |pos: usize| token!(p.token(pos) => T::Word(v) => *v);
    let step =
        word(0)
            .then_or_val_zip(bang, "")
            .map(|(a, b)| format!("{}{}", a, b));

    expect(step, "abc!".to_string());
}
#[test]
fn test_expect_or_env() {
    let p = parsit("abc!");
    let bang = |pos: usize| token!(p.token(pos) => T::Bang => "!");
    let word = |pos: usize| token!(p.token(pos) => T::Word(v) => *v);
    let step =
        word(0)
            .then_or_val_zip(bang, "")
            .map(|(a, b)| format!("{}{}", a, b));

    expect_or_env(p,step, "abc!".to_string());
}

#[test]
fn test_pos() {
    let p = parsit("abc!");
    let bang = |pos: usize| token!(p.token(pos) => T::Bang => "!");
    let word = |pos: usize| token!(p.token(pos) => T::Word(v) => v);
    let step = word(0).then_or_val_zip(bang, "");

    expect_pos(step, 2); // the next position to parse
}
#[test]
fn test_pos_or_env() {
    let p = parsit("abc!");
    let bang = |pos: usize| token!(p.token(pos) => T::Bang => "!");
    let word = |pos: usize| token!(p.token(pos) => T::Word(v) => v);
    let step = word(0).then_or_val_zip(bang, "");

    expect_pos_or_env(p,step, 2); // the next position to parse
}

#[test]
fn test_fail() {
    let p = parsit("abc?!");
    let bang = |pos: usize| token!(p.token(pos) => T::Bang => "!");
    let word = |pos: usize| token!(p.token(pos) => T::Word(v) => v);
    let step = word(0).then_zip(bang);

    fail(step);
}

#[test]
fn test_fail_on() {
    let p = parsit("abc?!");
    let bang = |pos: usize| token!(p.token(pos) => T::Bang => "!");
    let word = |pos: usize| token!(p.token(pos) => T::Word(v) => v);
    let step = word(0).then_zip(bang);

    fail_on(step, 1);
}