0.2.2 • Published 1 year ago

peg-express v0.2.2

Weekly downloads
-
License
MIT
Repository
github
Last release
1 year ago

peg-express

ok-da Maintainability Test Coverage DeepScan grade Codacy Badge

Installation

To install peg-express, run the following command in the project directory:

$ npm install peg-express

Getting Started

To use peg-express, you need to write a grammar definition and optional semantic actions. Without semantics actions, generated parsers create a parse tree of the input. The user can easily write their semantic actions by extending the generated parser class.

Typical steps to use peg-express are:

  1. Write a grammar definition in PEG.
  2. Generate a parser from the grammar definition.
  3. Write semantic actions by inheriting the generated parser.
  4. Use the parser.

Step1: Write a grammar definition in PEG

peg-express uses PEG for grammar definition. PEG is a simple and powerful grammar definition language.

You can write a grammar definition for a simple calculator as follows in calculator.peg:

Expression <- Term (('+' / '-') Term)*
Term       <- Factor (('*' / '/') Factor)*
Factor     <- Number / '(' Expression ')'
Number     <- r'\d+'

The start symbol of the grammar is Expression, which matches an arithmetic expression like (1+2)*3. Note that you can use regular expressions in the grammar definition. r'\d+' is a regular expression that matches a sequence of one or more digits. r is a prefix of a regular expression literal. You can write arbitrary JavaScript regular expressions between the prefix r' and '.

Step2: Generate a parser from the grammar definition

To generate a parser from the grammar definition, run the following command:

peg-express <grammar>

<grammar> is a path to a grammar definition file. The above command generates a file named 'parser.ts'.

For example, if you save the grammar definition in Step 2 as calculator.peg, you can generate a parser as follows:

peg-express calculator.peg

This generates a parser file parser.ts in the current directory.

Step3: Write semantic actions by inheriting the generated parser

This step is optional. If you do not write semantic actions, the generated parser creates a parse tree of the input. You can write semantic actions by extending the generated parser class.

For example, if you want to evaluate the input expression, you can write semantic actions as follows:

import { Expression, Factor, Parser, Term, NodeTerminal } from './parser';

export class Calculator extends Parser {
  override Expression([term, rest]: Expression) {
    return rest.reduce(
      (acc, [[add], term]): number =>
        add ? acc + term.value : acc - term.value,
      term.value
    );
  }

  override Term([factor, rest]: Term) {
    return rest.reduce(
      (acc, [[mul], factor]): number =>
        mul ? acc * factor.value : acc / factor.value,
      factor.value
    );
  }

  override Factor([number, parenthesizedExpression]: Factor) {
    if (number) {
      return number.value;
    }
    const [, expr] = parenthesizedExpression;
    return expr.value;
  }

  override Number(node: NodeTerminal) {
    return parseInt(node.text, 10);
  }
}

When you write the above code, the IDE (e.g., VSCode) can provide code completion and type-checking for semantic actions as follows:

The parser class implements default semantic actions that you can override.

Step4: Use the parser

You can use the generated parser by instantiating the parser class and calling parse() method. If you extend the generated parser class, you need to instantiate the extended class.

For example, if you save the semantic actions above as calculator.ts, you can use the parser as follows:

import { Calculator } from './calculator';

const parser = new Calculator();
const result = parser.parse('(1+2)*3', 'Expression');
if (result instanceof Error) {
  console.error(result.message);
} else {
  console.log(result.value);
}

The output of the above program is 9.

PEG Syntax

A rule in PEG is defined as follows:

name <- parsing_expression

name is a rule name called a nonterminal symbol. parsing_expression is a sequence of one or more parsing expressions.

peg-express supports the following parsing expressions:

ExpressionExampleDescription
wildcard.Matches any character.
string'abc'Matches the character string abc.
regular expressionr'abc'Is equivalent to /abc/ in TypeScript
nonterminal symbolnameMatches the rule name.
ordered choicee1 / e2Matches e1 or e1.
sequencee1 e2Matches e1 and e2.
zero or moree*Matches zero or more e.
one or moree+Matches one or more e.
optionale?Matches e or nothing.
grouping(e)Matches e.
and predicate&eMatches if e matches.
not predicate!eMatches if e does not match.

Generating a Parser

Generated parsers consist of the following two files:

  • parser.ts: The generated parser class.
  • SemanticValue.ts: The types of semantic values for each node in the parse tree.

Parser Class parser.ts

peg-express generates a parser class from a grammar definition. The generated parser implements default semantic actions that you can override. A semantic action is a method that is called when a node in the parse tree is matched. The name of the semantic action is the same as the name of the node in the parse tree. In other words, the name of the semantic action is the same as the name of the rule in the grammar definition. The return value of the semantic action is the semantic value of the node in the parse tree. The parameter of the semantic action is a collection of placeholders, each of which holds the semantic value of a child node in the parse tree. Its type depends on the right-hand side of the rule in the grammar definition. For example, if the right-hand side of the rule is A B where A and B are nonterminal symbols, the parameter of the semantic action is [item0, item1]: [Value<SemanticValue.A>, Value<SemanticValue.B>]. The type of each parsing expression is defined as follows:

ExpressionExampleType
wildcard.NodeTerminal
string'abc'NodeTerminal
regular expressionr'abc'NodeTerminal
nonterminal symbolnameValue<SemanticType.name>
ordered choicee1 / e2[e1:Type(e1), e2: null] \| [e1:null, e2: Type(e2)]
sequencee1 e2[e1:Type(e1), e2: Type(e2)]
zero or moree*Type(e)[]
one or moree+Type(e)[]
optionale?Type(e)[]
grouping(e)Type(e)
and predicate&enull
not predicate!enull

In the above table, Type(e) is the type of the semantic value of the parsing expression e.

Type definitions for semantic values SemanticValue.ts

peg-express generates a type definition file for semantic values. By default, the type of the semantic value of each node in the parse tree is any. You can override the default type by specifying the type of the semantic value in the grammar definition. This enables the IDE to provide better type checking when you write semantic actions.

For example, if you want to specify the type of the semantic value of the Expression node as number, you can modify the type definition for the Expression node as follows:

type Expression = number;

When you accidentally write a semantic action that returns a value of a different type, the IDE will notify you of the error.

Currently, SemanticValue.ts is overwritten every time you generate a parser. In the future, peg-express will support the update of the type definition file.

License

MIT License

Feedback

If you have any questions or comments, please contact me at peg-express@okuda.xyz