petitparser

PetitParser for Dart

PetitParser makes writing parsers in Dart fast, enjoyable, and fully type-safe.

Instead of writing fragile regular expressions or configuring complex external code generators, PetitParser lets you build parsers directly in plain Dart code. You assemble small, specialized parsers like LEGO bricks to model everything from simple configuration values to full programming languages.

✨ Why PetitParser?

• 🧩 Composable: Combine simple parsers into rich grammars using familiar operators and methods.
• 🛡️ Type-Safe: Take advantage of modern Dart records, generics, and pattern matching without casting dynamic types.
• ⚡ Fast & Efficient: Optimized character predicates and dedicated fast-path parsing keep execution blazing fast.
• 🔍 Debuggable: Step through parsing in your IDE, inspect tokens, and visualize execution trees with built-in tools.
• 🛠️ Extensible: Transform raw text directly into strongly typed models, syntax trees, or domain objects on the fly.
• 🌐 Open Source & Battle-Tested: Highly tested, well-documented, and trusted across production Dart and Flutter applications.

🚀 Quick Start

Add PetitParser to your Dart or Flutter project:

dart pub add petitparser

Import the package, assemble your parser, and parse key-value pairs:

import 'package:petitparser/petitparser.dart';

void main() {
  // 1. Match a key of one or more letters:
  final key = letter().plus().flatten();

  // 2. Match a numeric value and convert it directly into an integer:
  final value = digit().plus().flatten().map(int.parse);

  // 3. Combine them: match key, discard '=', and match value:
  final entry = key.trim().skip(after: char('=')).then(value.trim());

  // 4. Transform the parsed pair into a typed Dart record:
  final parser = entry.map2((k, v) => (key: k, value: v));

  // 5. Parse the input:
  final result = parser.parse('port = 8080');

  print(result.value); // (key: port, value: 8080)
}

Notice what happened here:

• letter().plus() matched consecutive letters, and .flatten() turned them into a String.
• digit().plus() matched consecutive digits, and .map(int.parse) transformed them immediately into an int.
• .skip(after: char('=')) consumed the separator without cluttering the result.
• .then(...) combined both parts into a strongly typed record (String, int).
• .map2(...) mapped both positional values into a readable named record.

🧩 Core Building Blocks

Every grammar in PetitParser is constructed from three fundamental concepts: matching text, combining parsers, and transforming results.

🔡 Matching Text

Primitive parsers recognize basic character patterns:

• char('a') matches an exact single character:

  char('a').parse('a').value; // 'a'
  

• string('dart') matches an entire string:

  string('dart').parse('dart').value; // 'dart'
  

• digit() matches any digit from 0 to 9:

  digit().parse('7').value; // '7'
  

• letter() matches any uppercase or lowercase letter:

  letter().parse('x').value; // 'x'
  

• word() matches a letter, digit, or underscore:

  word().parse('_').value; // '_'
  

• whitespace() matches spaces, tabs, and newlines:

  whitespace().parse(' ').value; // ' '
  

• pattern('0-9a-fA-F') matches any character within a range:

  pattern('0-9a-fA-F').parse('f').value; // 'f'
  

• any() matches any character:

  any().parse('!').value; // '!'
  

Most character parsers accept optional parameters such as ignoreCase: true for case-insensitive matching and unicode: true for full Unicode code-point support.

🔗 Combining Parsers

Snap individual parsers together to recognize structured input:

• Sequences: Run parsers in order. letter().then(digit()) returns a typed Dart record with each result:

  letter().then(digit()).parse('a1').value; // ('a', '1')
  

• Choices: Try options in order until one succeeds. [letter(), digit()].toChoiceParser() matches either a letter or a digit:

  [letter(), digit()].toChoiceParser().parse('a').value; // 'a'
  

• Repetition: Repeat patterns to collect multiple matches into a List:

  • letter().star() matches zero or more times:

    letter().star().parse('abc').value; // ['a', 'b', 'c']
    

  • digit().plus() matches one or more times:

    digit().plus().parse('123').value; // ['1', '2', '3']
    

  • char('-').optional() matches zero or one time, returning null when absent:

    char('-').optional().parse('-').value; // '-'
    char('-').optional().parse('+').value; // null
    

  • letter().times(3) matches an exact number of occurrences:

    letter().times(3).parse('abc').value; // ['a', 'b', 'c']
    

  • digit().plusSeparated(char(',')) matches items separated by a delimiter:

    digit().plusSeparated(char(',')).parse('1,2,3').value.elements; // ['1', '2', '3']
    

🎨 Transforming Results

Rather than traversing an untyped syntax tree after the fact, PetitParser lets you transform data while parsing:

• parser.flatten() merges matched characters into a single String:

  digit().plus().flatten().parse('123').value; // '123'
  

• parser.trim() discards leading and trailing whitespace around a match:

  string('true').trim().parse('  true  ').value; // 'true'
  

• parser.map(callback) transforms the parsed value with a custom function:

  digit().plus().flatten().map(int.parse).parse('42').value; // 42
  

• parser.map2(callback) unpacks a two-element record and passes each value as an argument:

  letter().then(digit()).map2((l, d) => '$l:$d').parse('x9').value; // 'x:9'
  

• parser.map3(callback) similarly unpacks a three-element record:

  (letter(), char(':'), digit())
      .toSequenceParser()
      .map3((l, sep, d) => '$l$sep$d')
      .parse('a:1')
      .value; // 'a:1'
  

🎯 Handling Results

Calling parse on a string returns a Result. You can inspect the outcome using Dart pattern matching:

final parser = digit().plus().flatten().map(int.parse);

void handleResult(Result<int> result) {
  switch (result) {
    case Success(value: final value):
      print('Parsed $value');
    case Failure(message: final message, position: final position):
      print('Error at $position: $message');
  }
}

handleResult(parser.parse('123')); // Parsed 123
handleResult(parser.parse('abc')); // Error at 0: digit expected

If you only need to know whether an input string conforms to your grammar, use accept:

if (parser.accept('123')) {
  print('Valid input');
}

To extract all matching occurrences across a larger body of text, use allMatches:

final words = letter().plus().flatten();
final matches = words.allMatches('two words 123');

print(matches); // ('two', 'words')

🛠️ Practical Examples

📋 Delimited Lists

This example parses a comma-separated list of integers like '1, 2, 3' into a Dart List<int>:

final number = digit().plus().flatten().map(int.parse);
final numbers = number
    .plusSeparated(char(',').trim())
    .map((list) => list.elements);

final result = numbers.parse('1, 2, 3');
print(result.value); // [1, 2, 3]

list.elements gives you a typed List<int> containing only the parsed numbers, neatly dropping the comma characters.

🧮 Expression Builder

This example parses and evaluates mathematical expressions containing numbers, parentheses, negation, exponentiation, multiplication, division, addition, and subtraction:

import 'dart:math' as math;
import 'package:petitparser/petitparser.dart';

final builder = ExpressionBuilder<num>();

// Define primitives like floating-point or integer numbers:
builder.primitive(
  digit()
      .plus()
      .then(char('.').then(digit().plus()).optional())
      .flatten()
      .trim()
      .map(num.parse),
);

// Add parentheses with highest priority:
builder.group().wrapper(
  char('(').trim(),
  char(')').trim(),
  (left, value, right) => value,
);

// Add prefix operators:
builder.group().prefix(char('-').trim(), (op, value) => -value);

// Add right-associative power operator:
builder.group().right(
  char('^').trim(),
  (left, op, right) => math.pow(left, right),
);

// Add left-associative multiplication and division:
builder.group()
  ..left(char('*').trim(), (left, op, right) => left * right)
  ..left(char('/').trim(), (left, op, right) => left / right);

// Add left-associative addition and subtraction:
builder.group()
  ..left(char('+').trim(), (left, op, right) => left + right)
  ..left(char('-').trim(), (left, op, right) => left - right);

final parser = builder.build().end();

print(parser.parse('1 + 2 * 3').value);   // 7
print(parser.parse('(1 + 2) * 3').value); // 9
print(parser.parse('2 ^ 2 ^ 3').value);   // 256
print(parser.parse('-8 + 2').value);      // -6

🌲 Recursive Structures

This example parses nested JSON-style arrays of numbers like '[1, [2, 3], 4]' into a hierarchy of Dart lists:

final value = undefined<Object>();

// An array contains zero or more values separated by commas:
final array = char('[')
    .trim()
    .then(value.starSeparated(char(',').trim()))
    .then(char(']').trim())
    .map3((open, elements, close) => elements.elements);

final number = digit().plus().flatten().map(int.parse);

// Wire the recursive definition into the placeholder:
value.set([number, array].toChoiceParser());

final parser = value.end();

print(parser.parse('[1, [2, 3], 4]').value); // [1, [2, 3], 4]

🏛️ Large Grammars

When a grammar grows to dozens or hundreds of productions, organizing rules as separate variables can become difficult to maintain. Subclassing GrammarDefinition lets you organize rules as methods on a class, resolving complex mutual recursions cleanly.

This example parses and evaluates arithmetic expressions using a modular, strongly typed grammar definition:

class ExpressionGrammarDefinition extends GrammarDefinition<num> {
  @override
  Parser<num> start() => ref0(term).end();

  Parser<num> term() => [ref0(add), ref0(prod)].toChoiceParser();

  Parser<num> add() => ref0(prod)
      .then(char('+').trim())
      .then(ref0(term))
      .map3((left, _, right) => left + right);

  Parser<num> prod() => [ref0(mul), ref0(prim)].toChoiceParser();

  Parser<num> mul() => ref0(prim)
      .then(char('*').trim())
      .then(ref0(prod))
      .map3((left, _, right) => left * right);

  Parser<num> prim() => [ref0(parens), ref0(number)].toChoiceParser();

  Parser<num> parens() => char('(')
      .trim()
      .then(ref0(term))
      .then(char(')').trim())
      .map3((_, value, _) => value);

  Parser<num> number() => digit().plus().flatten().trim().map(num.parse);
}

References between rules use ref0(rule) to preserve strong types and resolve mutual dependencies dynamically when building the parser:

final definition = ExpressionGrammarDefinition();
final parser = definition.build();

print(parser.parse('1 + 2 * 3').value); // 7
print(parser.parse('(1 + 2) * 3').value); // 9

You can also build a parser starting from any specific production rule in the grammar using buildFrom:

final numberParser = definition.buildFrom(ref0(definition.number));
print(numberParser.parse('42').value); // 42

Organizing grammars this way makes them easy to navigate, refactor, and test production by production in your IDE.

🔬 Advanced Capabilities

📍 Source Positions and Tokens

If you are writing a compiler, linter, or syntax highlighter, you often need to know exactly where in the source file an element appeared. Wrap any parser with .token() to capture its start and stop offsets, line number, and column number:

final id = letter().plus().flatten().token();
final token = id.parse('hello').value;

print(token.value);  // 'hello'
print(token.start);  // 0
print(token.stop);   // 5
print(token.line);   // 1
print(token.column); // 1

🔭 Lookaheads and Semantic Guards

You can inspect the upcoming input stream without consuming characters:

• parser.and() performs positive lookahead, succeeding only if parser matches next without advancing the input position.
• parser.not() performs negative lookahead, succeeding only if parser fails to match next.
• parser.where(predicate) rejects a match dynamically if a condition is not met.

For example, match a keyword like let only when it is not immediately followed by more word characters:

final keyword = string('let').skip(after: word().not());

print(keyword.parse('let').value); // 'let'
print(keyword.accept('letter'));   // false

⏳ Lazy Repetition

Standard repetition combinators like star() are possessive and consume as much input as possible. When matching delimited content like multi-line comments or quoted strings, use starLazy or plusLazy to stop as soon as a boundary parser succeeds:

final comment = string('/*')
    .then(any().starLazy(string('*/')).flatten())
    .then(string('*/'))
    .map3((start, body, end) => body);

print(comment.parse('/* note */').value); // ' note '

🩺 Grammar Linter

Grammars can contain subtle issues such as unreachable branches, infinite loops, or accidental left-recursions. PetitParser provides a built-in linter in package:petitparser/reflection.dart that you can run in your tests:

import 'package:petitparser/reflection.dart';
import 'package:test/test.dart';

test('grammar verification', () {
  final parser = letter().plus();
  expect(linter(parser), isEmpty);
});

⏱️ Debugging, Tracing, and Profiling

When a parser behaves unexpectedly on a particular input, wrap it with trace to see each step it takes:

final parser = letter().then(digit());
trace(parser).parse('a1');

This prints a clear activation tree showing which parsers were called and how they succeeded or failed:

SequenceParser2<String, String>
  SingleCharacterParser[letter expected]
  Success<String>[1:2]: a
  SingleCharacterParser[digit expected]
  Success<String>[1:3]: 1
Success<(String, String)>[1:3]: (a, 1)

For performance tuning, import package:petitparser/debug.dart and use profile(parser) to measure invocation counts and execution time per parser, or progress(parser) to inspect backtracking behavior.

📚 Resources

• Website: Interactive web demos and ports to other languages.
• GitHub Repository: Source code, discussions, and issue tracker.
• API Documentation: Detailed class and method reference.
• Examples Repository: Full implementations for JSON, Lisp, Math, Smalltalk, and more.

📜 History

Grammars for programming languages are traditionally specified statically. They are hard to compose and reuse due to ambiguities that inevitably arise. PetitParser combines ideas from scannerless parsing, parser combinators, parsing expression grammars (PEG), and packrat parsers to model grammars and parsers as objects that can be reconfigured dynamically.

PetitParser was originally created in Smalltalk. Later on, as a way to learn and explore these platforms, I reimplemented PetitParser in Java and Dart. All implementations share a consistent architecture while adopting the idioms and modern best practices of each target language.

📄 License

The MIT License, see LICENSE.