Parsec is non-transformer variant of more general ParsecT monad transformer.

ParsecT e s m a is a parser with custom data component of error e, stream type s, underlying monad m and return type a.

parse p file input runs parser p over Identity (see runParserT if you're using the ParsecT monad transformer; parse itself is just a synonym for runParser). It returns either a ParseError (Left) or a value of type a (Right). parseErrorPretty can be used to turn ParseError into the string representation of the error message. See Text.Megaparsec.Error if you need to do more advanced error analysis.

main = case (parse numbers "" "11,2,43") of
         Left err -> putStr (parseErrorPretty err)
         Right xs -> print (sum xs)

numbers = integer `sepBy` char ','

parseMaybe p input runs parser p on input and returns result inside Just on success and Nothing on failure. This function also parses eof, so if the parser doesn't consume all of its input, it will fail.

The function is supposed to be useful for lightweight parsing, where error messages (and thus file name) are not important and entire input should be parsed. For example it can be used when parsing of single number according to specification of its format is desired.

The expression parseTest p input applies a parser p against input input and prints the result to stdout. Useful for testing.

runParser p file input runs parser p on the input list of tokens input, obtained from source file. The file is only used in error messages and may be the empty string. Returns either a ParseError (Left) or a value of type a (Right).

parseFromFile p file = runParser p file <$> readFile file

The function is similar to runParser with the difference that it accepts and returns parser state. This allows to specify arbitrary textual position at the beginning of parsing, for example. This is the most general way to run a parser over the Identity monad.

Since: 4.2.0

runParserT p file input runs parser p on the input list of tokens input, obtained from source file. The file is only used in error messages and may be the empty string. Returns a computation in the underlying monad m that returns either a ParseError (Left) or a value of type a (Right).

This function is similar to runParserT, but like runParser' it accepts and returns parser state. This is thus the most general way to run a parser.

Since: 4.2.0

An associative binary operation

Zero or more.

One or more.

One or none.

The parser unexpected item always fails with an error message telling about unexpected item item without consuming any input.

Return both the result of a parse and the list of tokens that were consumed during parsing. This relies on the change of the stateTokensProcessed value to evaluate how many tokens were consumed.

Since: 5.3.0

Specify how to process ParseErrors that happen inside of this wrapper. As a side effect of current implementation changing errorPos with this combinator will also change the final statePos in parser state.

Since: 5.3.0

The most general way to stop parsing and report ParseError.

unexpected is defined in terms of this function:

unexpected item = failure (Set.singleton item) Set.empty Set.empty

Since: 4.2.0

A synonym for label in the form of an operator.

The parser label name p behaves as parser p, but whenever the parser p fails without consuming any input, it replaces names of “expected” tokens with the name name.

hidden p behaves just like parser p, but it doesn't show any “expected” tokens in error message when p fails.

The parser try p behaves like parser p, except that it backtracks parser state when p fails (either consuming input or not).

This combinator is used whenever arbitrary look ahead is needed. Since it pretends that it hasn't consumed any input when p fails, the (<|>) combinator will try its second alternative even when the first parser failed while consuming input.

For example, here is a parser that is supposed to parse word “let” or “lexical”:

>>> parseTest (string "let" <|> string "lexical") "lexical"
1:1:
unexpected "lex"
expecting "let"

What happens here? First parser consumes “le” and fails (because it doesn't see a “t”). The second parser, however, isn't tried, since the first parser has already consumed some input! try fixes this behavior and allows backtracking to work:

>>> parseTest (try (string "let") <|> string "lexical") "lexical"
"lexical"

try also improves error messages in case of overlapping alternatives, because Megaparsec's hint system can be used:

>>> parseTest (try (string "let") <|> string "lexical") "le"
1:1:
unexpected "le"
expecting "let" or "lexical"

Please note that as of Megaparsec 4.4.0, string backtracks automatically (see tokens), so it does not need try. However, the examples above demonstrate the idea behind try so well that it was decided to keep them. You still need to use try when your alternatives are complex, composite parsers.

If p in lookAhead p succeeds (either consuming input or not) the whole parser behaves like p succeeded without consuming anything (parser state is not updated as well). If p fails, lookAhead has no effect, i.e. it will fail consuming input if p fails consuming input. Combine with try if this is undesirable.

notFollowedBy p only succeeds when parser p fails. This parser never consumes any input and never modifies parser state. It can be used to implement the “longest match” rule.

withRecovery r p allows continue parsing even if parser p fails. In this case r is called with actual ParseError as its argument. Typical usage is to return value signifying failure to parse this particular object and to consume some part of input up to start of next object.

Note that if r fails, original error message is reported as if without withRecovery. In no way recovering parser r can influence error messages.

Since: 4.4.0

observing p allows to “observe” failure of p parser, should it happen, without actually ending parsing, but instead getting the ParseError in Left. On success parsed value is returned in Right as usual. Note that this primitive just allows you to observe parse errors as they happen, it does not backtrack or change how the p parser works in any way.

Since: 5.1.0

This parser only succeeds at the end of the input.

The parser token test mrep accepts a token t with result x when the function test t returns Right x. mrep may provide representation of the token to report in error messages when input stream in empty.

This is the most primitive combinator for accepting tokens. For example, the satisfy parser is implemented as:

satisfy f = token testChar Nothing
  where
    testChar x =
      if f x
        then Right x
        else Left (Set.singleton (Tokens (x:|[])), Set.empty, Set.empty)

The parser tokens test parses list of tokens and returns it. Supplied predicate test is used to check equality of given and parsed tokens.

This can be used for example to write string:

string = tokens (==)

Note that beginning from Megaparsec 4.4.0, this is an auto-backtracking primitive, which means that if it fails, it never consumes any input. This is done to make its consumption model match how error messages for this primitive are reported (which becomes an important thing as user gets more control with primitives like withRecovery):

>>> parseTest (string "abc") "abd"
1:1:
unexpected "abd"
expecting "abc"

This means, in particular, that it's no longer necessary to use try with tokens-based parsers, such as string and string'. This feature does not affect performance in any way.

between open close p parses open, followed by p and close. Returns the value returned by p.

braces = between (symbol "{") (symbol "}")

choice ps tries to apply the parsers in the list ps in order, until one of them succeeds. Returns the value of the succeeding parser.

count n p parses n occurrences of p. If n is smaller or equal to zero, the parser equals to return []. Returns a list of n values.

count' m n p parses from m to n occurrences of p. If n is not positive or m > n, the parser equals to return []. Returns a list of parsed values.

Please note that m may be negative, in this case effect is the same as if it were equal to zero.

Combine two alternatives.

Since: 4.4.0

endBy p sep parses zero or more occurrences of p, separated and ended by sep. Returns a list of values returned by p.

cStatements = cStatement `endBy` semicolon

endBy1 p sep parses one or more occurrences of p, separated and ended by sep. Returns a list of values returned by p.

manyTill p end applies parser p zero or more times until parser end succeeds. Returns the list of values returned by p. This parser can be used to scan comments:

simpleComment = string "<!--" >> manyTill anyChar (string "-->")

someTill p end works similarly to manyTill p end, but p should succeed at least once.

option x p tries to apply parser p. If p fails without consuming input, it returns the value x, otherwise the value returned by p.

priority = option 0 (digitToInt <$> digitChar)

sepBy p sep parses zero or more occurrences of p, separated by sep. Returns a list of values returned by p.

commaSep p = p `sepBy` comma

sepBy1 p sep parses one or more occurrences of p, separated by sep. Returns a list of values returned by p.

sepEndBy p sep parses zero or more occurrences of p, separated and optionally ended by sep. Returns a list of values returned by p.

sepEndBy1 p sep parses one or more occurrences of p, separated and optionally ended by sep. Returns a list of values returned by p.

skipMany p applies the parser p zero or more times, skipping its result.

space = skipMany spaceChar

skipSome p applies the parser p one or more times, skipping its result.

Parse a newline character.

Parse a carriage return character followed by a newline character. Return sequence of characters parsed.

Parse a CRLF (see crlf) or LF (see newline) end of line. Return the sequence of characters parsed.

eol = (pure <$> newline) <|> crlf

Parse a tab character.

Skip zero or more white space characters.

See also: skipMany and spaceChar.

Parse a control character (a non-printing character of the Latin-1 subset of Unicode).

Parse a Unicode space character, and the control characters: tab, newline, carriage return, form feed, and vertical tab.

Parse an upper-case or title-case alphabetic Unicode character. Title case is used by a small number of letter ligatures like the single-character form of Lj.

Parse a lower-case alphabetic Unicode character.

Parse an alphabetic Unicode character: lower-case, upper-case, or title-case letter, or a letter of case-less scripts/modifier letter.

Parse an alphabetic or numeric digit Unicode characters.

Note that the numeric digits outside the ASCII range are parsed by this parser but not by digitChar. Such digits may be part of identifiers but are not used by the printer and reader to represent numbers.

Parse a printable Unicode character: letter, number, mark, punctuation, symbol or space.

Parse an ASCII digit, i.e between “0” and “9”.

Parse an octal digit, i.e. between “0” and “7”.

Parse a hexadecimal digit, i.e. between “0” and “9”, or “a” and “f”, or “A” and “F”.

Parse a Unicode mark character (accents and the like), which combines with preceding characters.

Parse a Unicode numeric character, including digits from various scripts, Roman numerals, etc.

Parse a Unicode punctuation character, including various kinds of connectors, brackets and quotes.

Parse a Unicode symbol characters, including mathematical and currency symbols.

Parse a Unicode space and separator characters.

Parse a character from the first 128 characters of the Unicode character set, corresponding to the ASCII character set.

Parse a character from the first 256 characters of the Unicode character set, corresponding to the ISO 8859-1 (Latin-1) character set.

charCategory cat parses character in Unicode General Category cat, see GeneralCategory.

char c parses a single character c.

semicolon = char ';'

The same as char but case-insensitive. This parser returns actually parsed character preserving its case.

>>> parseTest (char' 'e') "E"
'E'
>>> parseTest (char' 'e') "G"
1:1:
unexpected 'G'
expecting 'E' or 'e'

This parser succeeds for any character. Returns the parsed character.

oneOf cs succeeds if the current character is in the supplied collection of characters cs. Returns the parsed character. Note that this parser doesn't automatically generate “expected” component of error message, so usually you should label it manually with label or (<?>).

See also: satisfy.

digit = oneOf ['0'..'9'] <?> "digit"

The same as oneOf, but case-insensitive. Returns the parsed character preserving its case.

vowel = oneOf' "aeiou" <?> "vowel"

As the dual of oneOf, noneOf cs succeeds if the current character not in the supplied list of characters cs. Returns the parsed character.

The same as noneOf, but case-insensitive.

consonant = noneOf' "aeiou" <?> "consonant"

The parser satisfy f succeeds for any character for which the supplied function f returns True. Returns the character that is actually parsed.

digitChar = satisfy isDigit <?> "digit"
oneOf cs  = satisfy (`elem` cs)

string s parses a sequence of characters given by s. Returns the parsed string (i.e. s).

divOrMod = string "div" <|> string "mod"

The same as string, but case-insensitive. On success returns string cased as actually parsed input.

>>> parseTest (string' "foobar") "foObAr"
"foObAr"

Positive integer that is used to represent line number, column number, and similar things like indentation level. Semigroup instance can be used to safely and purely add Poses together.

Since: 5.0.0

Construction of Pos from an instance of Integral. The function throws InvalidPosException when given non-positive argument. Note that the function is polymorphic with respect to MonadThrow m, so you can get result inside of Maybe, for example.

Since: 5.0.0

Extract Word from Pos.

Since: 5.0.0

Dangerous construction of Pos. Use when you know for sure that argument is positive.

Since: 5.0.0

The exception is thrown by mkPos when its argument is not a positive number.

Since: 5.0.0

The data type SourcePos represents source positions. It contains the name of the source file, a line number, and a column number. Source line and column positions change intensively during parsing, so we need to make them strict to avoid memory leaks.

Construct initial position (line 1, column 1) given name of source file.

Pretty-print a SourcePos.

Since: 5.0.0

Data type that is used to represent “unexpected/expected” items in parse error. The data type is parametrized over token type t.

Since: 5.0.0

The type class defines how to represent information about various exceptional situations. Data types that are used as custom data component in ParseError must be instances of this type class.

Since: 5.0.0

“Default error component”. This is our instance of ErrorComponent provided out-of-box.

Since: 5.0.0

The data type ParseError represents parse errors. It provides the stack of source positions, set of expected and unexpected tokens as well as set of custom associated data. The data type is parametrized over token type t and custom data e.

Note that the stack of source positions contains current position as its head, and the rest of positions allows to track full sequence of include files with topmost source file at the end of the list.

Semigroup (or Monoid) instance of the data type allows to merge parse errors from different branches of parsing. When merging two ParseErrors, longest match is preferred; if positions are the same, custom data sets and collections of message items are combined.

Type class ShowToken includes methods that allow to pretty-print single token as well as stream of tokens. This is used for rendering of error messages.

The type class defines how to print custom data component of ParseError.

Since: 5.0.0

Pretty-print ParseError. The rendered String always ends with a newline.

The function is defined as:

parseErrorPretty e =
  sourcePosStackPretty (errorPos e) ++ ":\n" ++ parseErrorTextPretty e

Since: 5.0.0

dbg label p parser works exactly like p, but when it's evaluated it prints information useful for debugging. The label is only used to refer to this parser in the debugging output. This combinator uses the trace function from Debug.Trace under the hood.

Typical usage is to wrap every sub-parser in misbehaving parser with dbg assigning meaningful labels. Then give it a shot and go through the print-out. As of current version, this combinator prints all available information except for hints, which are probably only interesting to the maintainer of Megaparsec itself and may be quite verbose to output in general. Let me know if you would like to be able to see hints as part of debugging output.

The output itself is pretty self-explanatory, although the following abbreviations should be clarified (they are derived from low-level source code):

• COK — “consumed OK”. The parser consumed input and succeeded.
• CERR — “consumed error”. The parser consumed input and failed.
• EOK — “empty OK”. The parser succeeded without consuming input.
• EERR — “empty error”. The parser failed without consuming input.

Finally, it's not possible to lift this function into some monad transformers without introducing surprising behavior (e.g. unexpected state backtracking) or adding otherwise redundant constraints (e.g. Show instance for state), so this helper is only available for ParsecT monad, not MonadParsec in general.

Since: 5.1.0

An instance of Stream s has stream type s. Token type is determined by the stream and can be found via Token type function.

This is the Megaparsec's state, it's parametrized over stream type s.

Return the current input.

setInput input continues parsing with input. The getInput and setInput functions can for example be used to deal with include files.

Return the current source position.

See also: setPosition, pushPosition, popPosition, and SourcePos.

Get position where the next token in the stream begins. If the stream is empty, return Nothing.

Since: 5.3.0

setPosition pos sets the current source position to pos.

See also: getPosition, pushPosition, popPosition, and SourcePos.

Push given position into stack of positions and continue parsing working with this position. Useful for working with include files and the like.

See also: getPosition, setPosition, popPosition, and SourcePos.

Since: 5.0.0

Pop a position from stack of positions unless it only contains one element (in that case stack of positions remains the same). This is how to return to previous source file after pushPosition.

See also: getPosition, setPosition, pushPosition, and SourcePos.

Since: 5.0.0

Get number of tokens processed so far.

Since: 5.2.0

Set number of tokens processed so far.

Since: 5.2.0

Return tab width. Default tab width is equal to defaultTabWidth. You can set different tab width with help of setTabWidth.

Set tab width. If argument of the function is not positive number, defaultTabWidth will be used.

Returns the full parser state as a State record.

setParserState st set the full parser state to st.

updateParserState f applies function f to the parser state.