pars

README

pars - Simple parser combinators for Go

For documentation please read the godoc.

Documentation

Overview

Package pars is a simple framework for parser combinators. It is designed to be easy to use, yet powerful enough to allow solving real world problems. Parsers can be arranged into flexible stacks of more elemental parsers. Parser results can be transformed via transformers to easily convert their results into a more fitting format or to implement additional conditions that must be fulfilled for a successful parse. Complex parsers can be debugged by wrapping them with a logger. pars parsers can read from a string or from an io.Reader, so streaming parsing is possible if you need it.

Index

• func ParseFromReader(ior io.Reader, p Parser) (interface{}, error)
• func ParseString(s string, p Parser) (interface{}, error)
• type Clause
  • func (c Clause) Parsers() []Parser
  • func (c Clause) TransformError(err error) error
  • func (c Clause) TransformResult(val []interface{}) interface{}
• type DescribeClause
  • func (d DescribeClause) TransformError(err error) error
• type DispatchClause
• type Logger
• type Parser
  • func AnyByte() Parser
  • func AnyRune() Parser
  • func BigInt() Parser
  • func Byte(b byte) Parser
  • func Char(r rune) Parser
  • func CharPred(pred func(rune) bool) Parser
  • func DelimitedString(beginDelimiter, endDelimiter string) Parser
  • func DiscardLeft(left, right Parser) Parser
  • func DiscardRight(left, right Parser) Parser
  • func Dispatch(clauses ...DispatchClause) Parser
  • func Error(err error) Parser
  • func ErrorTransformer(parser Parser, transformer func(error) (interface{}, error)) Parser
  • func Except(parser, except Parser) Parser
  • func Float() Parser
  • func Int() Parser
  • func JoinString(parser Parser) Parser
  • func Many(parser Parser) Parser
  • func NewAnyByte() Parser
  • func NewAnyRune() Parser
  • func NewBigInt() Parser
  • func NewByte(b byte) Parser
  • func NewChar(r rune) Parser
  • func NewCharPred(pred func(rune) bool) Parser
  • func NewDelimitedString(beginDelimiter, endDelimiter string) Parser
  • func NewDiscardLeft(left, right Parser) Parser
  • func NewDiscardRight(left, right Parser) Parser
  • func NewError(err error) Parser
  • func NewExcept(parser, except Parser) Parser
  • func NewFloat() Parser
  • func NewInt() Parser
  • func NewLogger(parser Parser, logger *log.Logger) Parser
  • func NewMany(parser Parser) Parser
  • func NewOptional(parser Parser) Parser
  • func NewOr(parsers ...Parser) Parser
  • func NewRecursive(factory func() Parser) Parser
  • func NewRunesToString(parser Parser) Parser
  • func NewRunesUntil(endCondition Parser) Parser
  • func NewSep(item, separator Parser) Parser
  • func NewSeq(parsers ...Parser) Parser
  • func NewSome(parser Parser) Parser
  • func NewSplicingSeq(parsers ...Parser) Parser
  • func NewStdLogger(parser Parser, prefix string) Parser
  • func NewString(expected string) Parser
  • func NewStringCI(expected string) Parser
  • func NewSwallowLeadingWhitespace(parser Parser) Parser
  • func NewSwallowTrailingWhitespace(parser Parser) Parser
  • func NewSwallowWhitespace(parser Parser) Parser
  • func NewTransformer(parser Parser, transformer func(interface{}) (interface{}, error)) Parser
  • func Optional(parser Parser) Parser
  • func Or(parsers ...Parser) Parser
  • func Recursive(factory func() Parser) Parser
  • func RunesToString(parser Parser) Parser
  • func RunesUntil(endCondition Parser) Parser
  • func Sep(item, separator Parser) Parser
  • func Seq(parsers ...Parser) Parser
  • func Some(parser Parser) Parser
  • func SplicingSeq(parsers ...Parser) Parser
  • func String(expected string) Parser
  • func StringCI(expected string) Parser
  • func SwallowLeadingWhitespace(parser Parser) Parser
  • func SwallowTrailingWhitespace(parser Parser) Parser
  • func SwallowWhitespace(parser Parser) Parser
  • func Transformer(parser Parser, transformer func(interface{}) (interface{}, error)) Parser
  • func WithLogging(parser Parser, logger Logger) Parser
  • func WithStdLogging(parser Parser, prefix string) Parser
• type Reader
  • func NewReader(r io.Reader) *Reader
  • func (br *Reader) Read(p []byte) (n int, err error)
  • func (br *Reader) Unread(p []byte)
• type Scanner
  • func NewScanner(r *Reader, p Parser) Scanner
  • func (s Scanner) Err() error
  • func (s Scanner) Result() interface{}
  • func (s Scanner) ResultString() string
  • func (s *Scanner) Scan() bool
• type StringJoiningClause
  • func (s StringJoiningClause) TransformResult(vals []interface{}) interface{}

Examples

• DiscardLeft
• Or
• Scanner
• Seq
• Transformer

Functions

func ParseFromReader

func ParseFromReader(ior io.Reader, p Parser) (interface{}, error)

ParseFromReader parses from an io.Reader.

func ParseString

func ParseString(s string, p Parser) (interface{}, error)

ParseString is a helper function to directly use a parser on a string.

Types

type Clause

type Clause []Parser

Clause is the most simple DispatchClause. It is just a slice of parsers without any transformations.

func (Clause) Parsers

func (c Clause) Parsers() []Parser

Parsers returns the parser slice for this clause.

func (Clause) TransformError

func (c Clause) TransformError(err error) error

TransformError returns the given error unchanged.

func (Clause) TransformResult

func (c Clause) TransformResult(val []interface{}) interface{}

TransformResult returns the only value if the slice of values has only one element. Otherwise it returns the slice of values unchanged.

type DescribeClause

type DescribeClause struct {
	DispatchClause
	Description string
}

DescribeClause extends the error message of a clause so that a custom description is part of the message.

func (DescribeClause) TransformError

func (d DescribeClause) TransformError(err error) error

TransformError extends the error message of a clause so that a custom description is part of the message.

type DispatchClause

type DispatchClause interface {
	//Parsers returns the parsers of the clause.
	Parsers() []Parser
	//TransformResult allows the DispatchClause to combine the results of its parsers to a single result.
	TransformResult([]interface{}) interface{}
	//TransformError allows the DispatchClause to replace or extend the error returned on a failed match.
	TransformError(error) error
}

DispatchClause is the interface of a clause used by Dispatch.

type Logger

type Logger interface {
	Println(...interface{})
}

Logger is anything that lines can be printed to.

type Parser

type Parser interface {
	//Parse is used for the actual parsing. It reads from the reader and returns the result or an error value.
	//Each parser must remember enough from the call to this method to undo the reading in case of a parsing error that occurs later.
	//When Parse returns with an error, Parse must make sure that all read bytes are unread so that another parser could try to parse them.
	Parse(*Reader) (interface{}, error)
	//Unread puts read bytes back to the reader so that they can be read again by other parsers.
	Unread(*Reader)
	//Clone creates a parser that works the same as the receiver. This allows to create a single parser as a blueprint for other parsers.
	//Internal state from reading operations should not be cloned.
	Clone() Parser
}

Parser contains the methods that each parser in this framework has to provide.

var EOF Parser = eof{}

EOF is a parser that never yields a value but that succeeds if and only if the source reached EOF

func AnyByte

func AnyByte() Parser

AnyByte returns a parser that reads exactly one byte from the source.

func AnyRune

func AnyRune() Parser

AnyRune returns a parser that parses a single valid rune. If no such rune can be read, ErrRuneExpected is returned.

func BigInt

func BigInt() Parser

BigInt returns a parser that parses an integer. The parsed integer is returned as a math/big.Int.

func Byte

func Byte(b byte) Parser

Byte returns a parser used to read a single known byte. A different byte is treated as a parsing error.

func Char

func Char(r rune) Parser

Char returns a parser used to read a single known rune. A different rune is treated as a parsing error.

func CharPred

func CharPred(pred func(rune) bool) Parser

CharPred returns a parser that parses a single rune as long as it fulfills the given predicate.

func DelimitedString

func DelimitedString(beginDelimiter, endDelimiter string) Parser

DelimitedString returns a parser that parses a string between two given delimiter strings and returns the value between.

func DiscardLeft

func DiscardLeft(left, right Parser) Parser

DiscardLeft returns a parser that calls two other parsers but only returns the result of the second parser. Both parsers must succeed.

Example

data := "$123"

dollarParser := DiscardLeft(Char('$'), Int())

result, err := ParseString(data, dollarParser)
if err != nil {
	fmt.Println("Error while parsing:", err)
	return
}

fmt.Printf("%v: %T\n", result, result)

Output:

123: int

func DiscardRight

func DiscardRight(left, right Parser) Parser

DiscardRight returns a parser that calls two other parsers but only returns the result of the first parser. Both parsers must succeed.

func Dispatch

func Dispatch(clauses ...DispatchClause) Parser

Dispatch returns a parser that is like a combination of Seq and Or with limited backtracking.

A Dispatch contains multiple clauses consisting of parsers. Dispatch parses by trying the clauses one by one. The first matching clause is used, later clauses are not tried. Each clause can contain multiple parsers. Clauses are special because they limit the backtracking: If the first parser of a clause matches, that clause is selected even if a later parser of that clause fails. If no clause matches, the error from the last clause is returned.

The motivation for limited backtracking is in better error reporting. When an Or parser fails, all you know is that not a single parser succeeded. When a Dispatch parser fails after a clause was selected, you know which subclause was supposed to be parsed and can return a fitting error message.

func Error

func Error(err error) Parser

Error returns a parser that always fails with the given error

func ErrorTransformer

func ErrorTransformer(parser Parser, transformer func(error) (interface{}, error)) Parser

ErrorTransformer wraps a parser so that an error result is transformed according to the given function. If the wrapped parser was successful, the result is not changed.

func Except

func Except(parser, except Parser) Parser

Except returns a parser that wraps another parser so that it fails if a third, excepted parser would succeed.

func Float

func Float() Parser

Float returns a parser that parses a floating point number. The supported format is an optional minus sign followed by digits optionally followed by a decimal point and more digits.

func Int

func Int() Parser

Int returns a parser that parses an integer. The parsed integer is converted via strconv.Atoi.

func JoinString

func JoinString(parser Parser) Parser

JoinString wraps a parser that returns a slice of runes or strings so that it returns a single string instead. Runes and strings can be mixed in the same slice. The slice also can contain other slices of runes and strings, recursively. The returned parser WILL PANIC if the wrapped parser returns something that is not a slice of runes or strings!

func Many

func Many(parser Parser) Parser

Many returns a parser that matches a given parser one or more times. Not matching at all is an error.

func NewAnyByte

func NewAnyByte() Parser

NewAnyByte returns a parser that reads exactly one byte from the source.

Deprecated: Use AnyByte instead.

func NewAnyRune

func NewAnyRune() Parser

NewAnyRune returns a parser that parses a single valid rune. If no such rune can be read, ErrRuneExpected is returned.

Deprecated: Use AnyRune instead.

func NewBigInt

func NewBigInt() Parser

NewBigInt returns a parser that parses an integer. The parsed integer is returned as a math/big.Int.

Deprecated: Use BigInt instead.

func NewByte

func NewByte(b byte) Parser

NewByte returns a parser used to read a single known byte. A different byte is treated as a parsing error.

Deprecated: Use Byte instead.

func NewChar

func NewChar(r rune) Parser

NewChar returns a parser used to read a single known rune. A different rune is treated as a parsing error.

Deprecated: Use Char instead.

func NewCharPred

func NewCharPred(pred func(rune) bool) Parser

NewCharPred returns a parser that parses a single rune as long as it fulfills the given predicate.

Deprecated: Use CharPred instead.

func NewDelimitedString

func NewDelimitedString(beginDelimiter, endDelimiter string) Parser

NewDelimitedString returns a parser that parses a string between two given delimiter strings and returns the value between.

Deprecated: Use DelimitedString instead.

func NewDiscardLeft

func NewDiscardLeft(left, right Parser) Parser

NewDiscardLeft returns a parser that calls two other parsers but only returns the result of the second parser. Both parsers must succeed.

Deprecated: Use DiscardLeft instead.

func NewDiscardRight

func NewDiscardRight(left, right Parser) Parser

NewDiscardRight returns a parser that calls two other parsers but only returns the result of the first parser. Both parsers must succeed.

Deprecated: Use DiscardRight instead.

func NewError

func NewError(err error) Parser

NewError returns a parser that always fails with the given error

Deprecated: Use Error instead.

func NewExcept

func NewExcept(parser, except Parser) Parser

NewExcept returns a parser that wraps another parser so that it fails if a third, excepted parser would succeed.

Deprecated: Use Except instead.

func NewFloat

func NewFloat() Parser

NewFloat returns a parser that parses a floating point number. The supported format is an optional minus sign followed by digits optionally followed by a decimal point and more digits.

Deprecated: Use Float instead.

func NewInt

func NewInt() Parser

NewInt returns a parser that parses an integer. The parsed integer is converted via strconv.Atoi.

Deprecated: Use Int instead.

func NewLogger

func NewLogger(parser Parser, logger *log.Logger) Parser

NewLogger wraps a parser so that calls to it are logged to a given logger.

Deprecated: Use WithLogging instead.

func NewMany

func NewMany(parser Parser) Parser

NewMany returns a parser that matches a given parser one or more times. Not matching at all is an error.

Deprecated: Use Many instead.

func NewOptional

func NewOptional(parser Parser) Parser

NewOptional returns a parser that reads exactly one result according to a given other parser. If it fails, the error is discarded and nil is returned.

Deprecated: Use Optional instead.

func NewOr

func NewOr(parsers ...Parser) Parser

NewOr returns a parser that matches the first of a given set of parsers. A later parser will not be tried if an earlier match was found. The returned parser uses the error message of the last parser verbatim.

Deprecated: Use Or instead.

func NewRecursive

func NewRecursive(factory func() Parser) Parser

NewRecursive allows to recursively define a parser in terms of itself.

Deprecated: Use Recursive instead.

func NewRunesToString

func NewRunesToString(parser Parser) Parser

NewRunesToString wraps a parser that returns a slice of runes so that it returns a string instead. The returned parser WILL PANIC if the wrapped parser returns something that is not a slice of runes!

Deprecated: Use JoinString instead.

func NewRunesUntil

func NewRunesUntil(endCondition Parser) Parser

NewRunesUntil returns a parser that parses runes as long as the given endCondition parser does not match.

Deprecated: Use RunesUntil instead.

func NewSep

func NewSep(item, separator Parser) Parser

NewSep returns a parser that parses a sequence of items according to a first parser that are separated by matches of a second parser.

Deprecated: Use Sep instead.

func NewSeq

func NewSeq(parsers ...Parser) Parser

NewSeq returns a parser that matches all of its given parsers in order or none of them.

Deprecated: Use Seq instead.

func NewSome

func NewSome(parser Parser) Parser

NewSome returns a parser that matches a given parser zero or more times. Not matching at all is not an error.

Deprecated: Use Some instead.

func NewSplicingSeq

func NewSplicingSeq(parsers ...Parser) Parser

NewSplicingSeq returns a parser that works like a Seq but joins slices returned by its subparsers into a single slice.

Deprecated: Use SplicingSeq instead.

func NewStdLogger

func NewStdLogger(parser Parser, prefix string) Parser

NewStdLogger wraps a parser so that calls to it are logged to a logger logging to StdErr with a given prefix.

Deprecated: Use WithStdLogging instead.

func NewString

func NewString(expected string) Parser

NewString returns a parser for a single known string. Different strings are treated as a parsing error.

Deprecated: Use String instead.

func NewStringCI

func NewStringCI(expected string) Parser

NewStringCI returns a case-insensitive parser for a single known string. Different strings are treated as a parsing error.

Deprecated: Use StringCI instead.

func NewSwallowLeadingWhitespace

func NewSwallowLeadingWhitespace(parser Parser) Parser

NewSwallowLeadingWhitespace wraps a parser so that it removes leading whitespace.

Deprecated: Use SwallowLeadingWhitespace instead.

func NewSwallowTrailingWhitespace

func NewSwallowTrailingWhitespace(parser Parser) Parser

NewSwallowTrailingWhitespace wraps a parser so that it removes trailing whitespace.

Deprecated: Use SwallowTrailingWhitespace instead.

func NewSwallowWhitespace

func NewSwallowWhitespace(parser Parser) Parser

NewSwallowWhitespace wraps a parser so that it removes leading and trailing whitespace.

Deprecated: Use SwallowWhitespace instead.

func NewTransformer

func NewTransformer(parser Parser, transformer func(interface{}) (interface{}, error)) Parser

NewTransformer wraps a parser so that the result is transformed according to the given function. If the transformer returns an error, the parsing is handled as failed.

Deprecated: Use Transformer instead.

func Optional

func Optional(parser Parser) Parser

Optional returns a parser that reads exactly one result according to a given other parser. If it fails, the error is discarded and nil is returned.

func Or

func Or(parsers ...Parser) Parser

Or returns a parser that matches the first of a given set of parsers. A later parser will not be tried if an earlier match was found. The returned parser uses the error message of the last parser verbatim.

Example

data := "124"

parser := Or(String("123"), String("124"))

result, err := ParseString(data, parser)
if err != nil {
	fmt.Println("Error while parsing:", err)
	return
}

fmt.Printf("%v: %T\n", result, result)

Output:

124: string

func Recursive

func Recursive(factory func() Parser) Parser

Recursive allows to recursively define a parser in terms of itself.

func RunesToString

func RunesToString(parser Parser) Parser

RunesToString wraps a parser that returns a slice of runes or strings so that it returns a single string instead. Runes and strings can be mixed in the same slice. The returned parser WILL PANIC if the wrapped parser returns something that is not a slice of runes or strings!

Deprecated: Use JoinString instead.

func RunesUntil

func RunesUntil(endCondition Parser) Parser

RunesUntil returns a parser that parses runes as long as the given endCondition parser does not match.

func Sep

func Sep(item, separator Parser) Parser

Sep returns a parser that parses a sequence of items according to a first parser that are separated by matches of a second parser.

func Seq

func Seq(parsers ...Parser) Parser

Seq returns a parser that matches all of its given parsers in order or none of them.

Example

data := "$123"

dollarParser := Seq(Char('$'), Int())

result, err := ParseString(data, dollarParser)
if err != nil {
	fmt.Println("Error while parsing:", err)
	return
}

values := result.([]interface{})
fmt.Printf("%c: %T\n", values[0], values[0])
fmt.Printf("%v: %T\n", values[1], values[1])

Output:

$: int32
123: int

func Some

func Some(parser Parser) Parser

Some returns a parser that matches a given parser zero or more times. Not matching at all is not an error.

func SplicingSeq

func SplicingSeq(parsers ...Parser) Parser

SplicingSeq returns a parser that works like a Seq but joins slices returned by its subparsers into a single slice.

func String

func String(expected string) Parser

String returns a parser for a single known string. Different strings are treated as a parsing error.

func StringCI

func StringCI(expected string) Parser

StringCI returns a case-insensitive parser for a single known string. Different strings are treated as a parsing error.

func SwallowLeadingWhitespace

func SwallowLeadingWhitespace(parser Parser) Parser

SwallowLeadingWhitespace wraps a parser so that it removes leading whitespace.

func SwallowTrailingWhitespace

func SwallowTrailingWhitespace(parser Parser) Parser

SwallowTrailingWhitespace wraps a parser so that it removes trailing whitespace.

func SwallowWhitespace

func SwallowWhitespace(parser Parser) Parser

SwallowWhitespace wraps a parser so that it removes leading and trailing whitespace.

func Transformer

func Transformer(parser Parser, transformer func(interface{}) (interface{}, error)) Parser

Transformer wraps a parser so that the result is transformed according to the given function. If the transformer returns an error, the parsing is handled as failed.

Example

package main

import (
	"bitbucket.org/ragnara/pars"
	"fmt"
)

// Celsius contains a temperature in degree celsius.
type Celsius int

func (c Celsius) String() string {
	return fmt.Sprintf("%v°C", int(c))
}

// TemperatureParser is a parser for temperature strings returning Celsius instances.
type TemperatureParser struct {
	pars.Parser
}

// NewTemperatureParser creates a new TemperatureParser instance.
func NewTemperatureParser() TemperatureParser {
	//Define the format
	simpleParser := pars.Seq(pars.Int(), pars.Or(pars.String("°C"), pars.String("°F")))
	//Add an conversion
	transformedParser := pars.Transformer(simpleParser, transformParsedTemperatureToCelsius)

	return TemperatureParser{Parser: transformedParser}
}

// Parse returns the Celsius instance for a temperature string containing an integer followed by either "°C" or "°F". Fahrenheit strings are converted to celsius.
// For other strings, an error is returned.
func (t TemperatureParser) Parse(s string) (Celsius, error) {
	val, err := pars.ParseString(s, t.Parser)
	if err != nil {
		return Celsius(0), err
	}
	return val.(Celsius), nil
}

// MustParse parses exactly like Parse but panics if an invalid string was found. It should not be used on user input!
func (t TemperatureParser) MustParse(s string) Celsius {
	val, err := t.Parse(s)
	if err != nil {
		panic(err)
	}
	return val
}

func transformParsedTemperatureToCelsius(parserResult interface{}) (interface{}, error) {
	values := parserResult.([]interface{})
	degrees := values[0].(int)
	unit := values[1].(string)

	switch unit {
	case "°C":
		return Celsius(degrees), nil
	case "°F":
		return Celsius((degrees - 32) * 5 / 9), nil
	default:
		panic("Impossible unit: " + unit)
	}
}

func main() {
	sample1 := "32°C"
	sample2 := "104°F"
	sample3 := "128K"

	fmt.Println("Sample1:", NewTemperatureParser().MustParse(sample1))
	fmt.Println("Sample2:", NewTemperatureParser().MustParse(sample2))

	val, err := NewTemperatureParser().Parse(sample3)
	fmt.Println("Sample3:", val)
	fmt.Println("Sample3 error:", err.Error())

}

Output:

Sample1: 32°C
Sample2: 40°C
Sample3: 0°C
Sample3 error: Could not find expected sequence item 1: Could not parse expected string "°F": EOF

func WithLogging

func WithLogging(parser Parser, logger Logger) Parser

WithLogging wraps a parser so that calls to it are logged to a given logger.

func WithStdLogging

func WithStdLogging(parser Parser, prefix string) Parser

WithStdLogging wraps a parser so that calls to it are logged to a logger logging to StdErr with a given prefix.

type Reader

type Reader struct {
	// contains filtered or unexported fields
}

Reader is an io.Reader that can Unread as many bytes as necessary.

func NewReader

func NewReader(r io.Reader) *Reader

NewReader creates a new Reader from an io.Reader.

func (*Reader) Read

func (br *Reader) Read(p []byte) (n int, err error)

Read reads a slice of bytes.

func (*Reader) Unread

func (br *Reader) Unread(p []byte)

Unread unreads a slice of bytes so that they will be read again by Read.

type Scanner

type Scanner struct {
	// contains filtered or unexported fields
}

Scanner provides a convenient interface to use a single parser multiple times on the same reader. Successive calls to Scan will parse the input and allow the results to be accessed one at a time. Scanner stops at the first error.

Example

data := "this is a text of words"
reader := NewReader(strings.NewReader(data))

wordParser := SwallowTrailingWhitespace(JoinString(RunesUntil(CharPred(unicode.IsSpace))))

scanner := NewScanner(reader, wordParser)

for scanner.Scan() {
	fmt.Println(scanner.ResultString())
}
fmt.Println(scanner.Err())

Output:

this
is
a
text
of
words
<nil>

func NewScanner

func NewScanner(r *Reader, p Parser) Scanner

NewScanner returns a new scanner using a given Reader and Parser.

func (Scanner) Err

func (s Scanner) Err() error

Err returns the last encountered error that is not io.EOF. It returns nil otherwise.

func (Scanner) Result

func (s Scanner) Result() interface{}

Result returns the most recently parsed value from a call to Scan.

func (Scanner) ResultString

func (s Scanner) ResultString() string

ResultString returns the most recently parsed value from a call to Scan, cast to a String. This will panic if the last result is not a string!

func (*Scanner) Scan

func (s *Scanner) Scan() bool

Scan invokes the parser on the reader and makes the results available via Result and Err. Scan returns true if the parsing succeeded and returns false otherwise.

type StringJoiningClause

type StringJoiningClause struct {
	DispatchClause
}

StringJoiningClause extends a clause that consists of parsers that return runes or strings so that it returnes a single string instead. Slices are handled recursively. StringJoiningClause WILL PANIC if any of the parsers return something other than a rune or a string or a slice of these types.

func (StringJoiningClause) TransformResult

func (s StringJoiningClause) TransformResult(vals []interface{}) interface{}

TransformResult joins runes and strings together like JoinString.