A library for defining parsers (character- or binary-oriented) inline. ... mlp Meredith L. Patterson mlp@upstandinghackers.com lead LGPL 0.1 2013-11-15 alpha First cut of extension. Hammer C library
void free() Deallocate a parse result Free the memory allocated to a parse result when the result is no longer needed. void report() Output benchmarking results in a human-readable format object HParseResult parse(string str) Parse an input Top-level function to call a parser that has been built over some piece of input. --> int compile(string backend, array params) Generate parse tables for the given parsing backend Some of the parsing backends must transform a parser into a set of parse tables before they can be used. See the documentation for the parser backend in question for information about the [params] parameter, or pass in NULL for the defaults. Returns -1 if the grammar cannot be compiled with the specified options; 0 otherwise. object HBenchmarkResults benchmark(array testcases) Benchmark this parser using all the applicable parsing backends against a set of test cases you provide object HParser hammer_token(string token) Parse a token (primitive) Given a string, returns a parser that parses that string value. Result token type: string object HParser hammer_ch(string ch) Parse a character (primitive) Given a single character, returns a parser that parses that character. Result token type: string object HParser hammer_ch_range(string lower, string upper) Parse a character within the given range (primitive) Given two single-character bounds, lower and upper, returns a parser that parses a single character within the range [lower, upper] (inclusive). Result token type: string object HParser hammer_int_range(object HParser p, int lower, int upper) Parse an integer within the given range (primitive) Given an integer parser, p, and two integer bounds, lower and upper, returns a parser that parses an integral value within the range [lower, upper] (inclusive). Result token type: int object HParser hammer_bits(int len, bool sign) Parse a specified number of bits (primitive) Returns a parser that parses the specified number of bits. sign == true if signed, false if unsigned. Result token type: int object HParser hammer_int64() Parse a 64-bit signed integer (primitive) Returns a parser that parses a signed 8-byte integer value. Result token type: int object HParser hammer_int32() Parse a 32-bit signed integer (primitive) Returns a parser that parses a signed 4-byte integer value. Result token type: int object HParser hammer_int16() Parse a 16-bit signed integer (primitive) Returns a parser that parses a signed 2-byte integer value. Result token type: int object HParser hammer_int8() Parse an 8-bit signed integer (primitive) Returns a parser that parses a signed 1-byte integer value. Result token type: int object HParser hammer_uint64() Parse a 64-bit unsigned integer (primitive) Returns a parser that parses an unsigned 8-byte integer value. Result token type: int object HParser hammer_uint32() Parse a 32-bit unsigned integer (primitive) Returns a parser that parses an unsigned 4-byte integer value. Result token type: int object HParser hammer_uint16() Parse a 16-bit unsigned integer (primitive) Returns a parser that parses an unsigned 2-byte integer value. Result token type: int object HParser hammer_uint8() Parse an 8-bit unsigned integer (primitive) Returns a parser that parses an unsigned 1-byte integer value. Result token type: int object HParser hammer_whitespace(object HParser p) Parse something preceded by whitespace (higher-order) Given a parser, p, returns a parser that skips any whitespace and then applies p. Result token type: p's result type object HParser hammer_left(object HParser p, object HParser q) Take the leftmost result of two parsers (higher-order) Given two parsers, p and q, returns a parser that parses them in sequence but only returns p's result. Result token type: p's result type object HParser hammer_right(object HParser p, object HParser q) Take the rightmost result of two parsers (higher-order) Given two parsers, p and q, returns a parser that parses them in sequence but only returns q's result. Result token type: q's result type object HParser hammer_middle(object HParser p, object HParser x, object HParser q) Take the middle result of three parsers (higher-order) Given three parsers, p, x, and q, returns a parser that parses them in sequence but only returns x's result. Result token type: x's result type object HParser hammer_action(object HParser p, callback f) Attach a semantic action to a parser (higher-order) Given another parser, p, and a function, f, returns a parser that applies p, then applies f to everything in the AST of p's result. Result token type: any object HParser hammer_in(string charset) Parse a character that appears in charset (primitive) Parse a single character in the given charset. Result token type: string object HParser hammer_not_in(string charset) Parse a character that does not appear in charset (primitive) Parse a single character *not* in the given charset. Result token type: string object HParser hammer_end_p() Recognize the end-of-input (primitive) A no-argument parser that succeeds if there is no more input to parse. Result token type: none. object HParser hammer_nothing_p() This parser always fails (primitive) hammer_nothing_p is primarily useful when stubbing out a larger parser. If you define a rule as hammer_nothing_p(), the top-level parser will run, but if a rule containing hammer_nothing_p is invoked, the parse will fail. Result token type: there is no result token for this parser. object HParser hammer_sequence(array p_array) Apply a list of parsers sequentially (higher-order) Given an array of parsers, p_array, apply each parser in order. The parse succeeds only if all parsers succeed. Result token type: array object HParser hammer_choice(array p_array) Apply any one of a list of parsers (higher-order) Given an array of parsers, p_array, apply each parser in order. The first parser to succeed produces the result; if no parsers succeed, the parse fails. Result token type: the type of the first successful parser's result object HParser hammer_butnot(object HParser p1, object HParser p2) Parse p1 but not p2 (higher-order) Given two parsers, p1 and p2, this parser succeeds in the following cases: * if p1 succeeds and p2 fails (starting from the same place in the input stream) * if both succeed but p1's result is as long as or longer than p2's Result token type: p1's result type. object HParser hammer_difference(object HParser p1, object HParser p2) Parse p1 but not p2 (higher-order) Given two parsers, p1 and p2, this parser succeeds in the following cases: * if p1 succeeds and p2 fails (starting from the same place in the input stream) * if both succeed but p2's result is *strictly* shorter than p1's Result token type: p1's result type. object HParser hammer_xor(object HParser p1, object HParser p2) Parse p1 or p2 but not both (higher-order) Given two parsers, p1 and p2, this parser suceeds if *either* p1 or p2 succeed, but not if they both do. Result token type: the result type of whichever parser succeeded object HParser hammer_many(object HParser p) Parse zero or more instances of p (higher-order) Given a parser, p, this parser succeeds for zero or more repetitions of p. Result token type: array object HParser hammer_many1(object HParser p) Parse one or more instances of p (higher-order) Given a parser, p, this parser succeeds for one or more repetitions of p. Result token type: array object HParser hammer_repeat_n(object HParser p, int n) Parse exactly n instances of p (higher-order) Given a parser, p, this parser succeeds for exactly N repetitions of p. Result token type: array object HParser hammer_optional(object HParser p) Mark a parser as optional, like ? in regular expressions (higher-order) Given a parser, p, this parser succeeds with the value p parsed or with an empty result. Result token type: If p succeeded, the type of its result; if not, nothing. object HParser hammer_ignore(object HParser p) Apply p but leave its result out of the final AST (higher-order) Given a parser, p, this parser succeeds if p succeeds, but doesn't include p's result in the final result. Result token type: none. object HParser hammer_sepBy(object HParser p, object HParser sep) Parse a (possibly empty) sequence of values separated by a separator (higher-order) Given a parser, p, and a parser for a separator, sep, this parser matches a (possibly empty) list of things that p can parse, separated by sep. For example, if p is hammer_many1(hammer_ch_range('0', '9')) and sep is hammer_ch(','), hammer_sepBy(p, sep) will match a comma-separated list of integers. Result token type: array object HParser hammer_sepBy1(object HParser p, object HParser sep) Parse a sequence of values separated by a separator (higher-order) Given a parser, p, and a parser for a separator, sep, this parser matches a list of things that p can parse, separated by sep. Unlike hammer_sepBy, this ensures that the result has at least one element. For example, if p is hammer_many(hammer_ch_range('0', '9')), hammer_sepBy1(p, sep) will match a comma-separated list of integers. Result token type: array object HParser hammer_epsilon_p() Parse the empty string (primitive) This parser always returns a zero-length match, i.e., the empty string. Result token type: none object HParser hammer_length_value(object HParser length, object HParser value) Parse (length) repetitions of (value) (higher-order) This parser applies its first argument to read an unsigned integer value, then applies its second argument that many times. length should produce an integer value; this is checked at runtime. Specifically, length's token type must be int. Result token type: array object HParser hammer_attr_bool(object HParser p, callback pred) Ensure that some property holds for the AST (higher-order) This parser attaches a predicate function, which returns true or false, to a parser. The function is evaluated over the parser's result. The parse only succeeds if pred returns true. hammer_attr_bool will check whether p's result exists and whether p's result AST exists; you do not need to check for this in your predicate function. Result token type: p's result type if pred succeeds, otherwise the parse fails. object HParser hammer_and(object HParser p) Verify that p succeeds, but don't actually apply it (higher-order) This parser asserts that a conditional syntax is satisfied, but doesn't consume that conditional syntax. This is useful for lookahead. As an example: Suppose you already have a parser, hex_p, that parses numbers in hexadecimal format (including the leading "0x"). Then hammer_sequence(hammer_and(hammer_token("0x")), hex_p) checks to see whether there is a leading "0x", does not consume the "0x", then applies hex_p to parse the hexadecimal number. This parser succeeds if p succeeds, and fails if p fails. Result token type: none object HParser hammer_not(object HParser p) Verify that p does not succeed, and don't consume any input (higher-order) This parser asserts that a conditional syntax is *not* satisfied, and doesn't consume the additional syntax. As a somewhat contrived example: Since hammer_choice applies its arguments in order, the following parser: hammer_sequence(hammer_ch('a'), hammer_choice(hammer_ch('+'), hammer_token('++')), hammer_ch('b')) will not parse "a++b", because once hammer_choice has succeeded, it will not backtrack and try other alternatives if a later parser in the sequence fails. Instead, you can force the use of the second alternative by turning the hammer_ch('+') alternative into a sequence with hammer_not: hammer_sequence(hammer_ch('a'), hammer_choice(hammer_sequence(hammer_ch('+'), hammer_not('+')), hammer_token('++')), hammer_ch('b')) If the input string is "a+b", the first alternative is applied; if the input string is "a++b", the second alternative is applied. Result token type: none object HParser hammer_indirect() Forward-declare a parser to be used recursively (higher-order) Create a parser that calls out to another, as yet unknown, parser. Note that the inner parser must be bound later, using hammer_bind_indirect(). This can be used to create recursive parsers. Result token type: the type of whatever parser is bound to it with hammer_bind_indirect(). void hammer_bind_indirect(object HParser p) Set the inner parser of an indirect parser. See hammer_indirect() for details.