diff --git a/src/bindings/lua/hammer.lua b/src/bindings/lua/hammer.lua
new file mode 100644
index 0000000000000000000000000000000000000000..fe0c47fae66933df813221ee867521315e2b9277
--- /dev/null
+++ b/src/bindings/lua/hammer.lua
@@ -0,0 +1,308 @@
+local ffi = require("ffi")
+ffi.cdef[[
+static const BYTE_BIG_ENDIAN = 0x1
+static const BIT_BIG_ENDIAN = 0x2
+static const BYTE_LITTLE_ENDIAN = 0x0
+static const BIT_LITTLE_ENDIAN = 0x0
+
+typedef enum HParserBackend_ {
+ PB_MIN = 0,
+ PB_PACKRAT = PB_MIN, // PB_MIN is always the default.
+ PB_REGULAR,
+ PB_LLk,
+ PB_LALR,
+ PB_GLR,
+ PB_MAX = PB_GLR
+} HParserBackend;
+
+typedef enum HTokenType_ {
+ TT_NONE = 1,
+ TT_BYTES = 2,
+ TT_SINT = 4,
+ TT_UINT = 8,
+ TT_SEQUENCE = 16,
+ TT_RESERVED_1, // reserved for backend-specific internal use
+ TT_ERR = 32,
+ TT_USER = 64,
+ TT_MAX
+} HTokenType;
+
+typedef struct HBytes_ {
+ const uint8_t *token;
+ size_t len;
+} HBytes;
+
+typedef struct HArena_ HArena ; // hidden implementation
+
+typedef struct HCountedArray_ {
+ size_t capacity;
+ size_t used;
+ HArena * arena;
+ struct HParsedToken_ **elements;
+} HCountedArray;
+
+typedef struct HParsedToken_ {
+ HTokenType token_type;
+ union {
+ HBytes bytes;
+ int64_t sint;
+ uint64_t uint;
+ double dbl;
+ float flt;
+ HCountedArray *seq; // a sequence of HParsedToken's
+ void *user;
+ };
+ size_t index;
+ size_t bit_length;
+ char bit_offset;
+} HParsedToken;
+
+typedef struct HParseResult_ {
+ const HParsedToken *ast;
+ int64_t bit_length;
+ HArena * arena;
+} HParseResult;
+
+typedef struct HParserVtable_ HParserVtable;
+typedef struct HCFChoice_ HCFChoice;
+
+typedef struct HParser_ {
+ const HParserVtable *vtable;
+ HParserBackend backend;
+ void* backend_data;
+ void *env;
+ HCFChoice *desugared;
+} HParser;
+
+typedef HParsedToken* (*HAction)(const HParseResult *p, void* user_data);
+typedef bool (*HPredicate)(HParseResult *p, void* user_data);
+typedef HParser* (*HContinuation)(HAllocator *mm__, const HParsedToken *x, void *env);
+
+HParseResult* h_parse(const HParser* parser, const uint8_t* input, size_t length);
+HParser* h_token(const uint8_t *str, const size_t len);
+HParser* h_ch(const uint8_t c);
+HParser* h_ch_range(const uint8_t lower, const uint8_t upper);
+HParser* h_int_range(const HParser *p, const int64_t lower, const int64_t upper);
+HParser* h_bits(size_t len, bool sign);
+HParser* h_int64();
+HParser* h_int32();
+HParser* h_int16();
+HParser* h_int8();
+HParser* h_uint64();
+HParser* h_uint32();
+HParser* h_uint16();
+HParser* h_uint8();
+HParser* h_whitespace(const HParser* p);
+HParser* h_left(const HParser* p, const HParser* q);
+HParser* h_right(const HParser* p, const HParser* q);
+HParser* h_middle(const HParser* p, const HParser* x, const HParser* q);
+HParser* h_action(const HParser* p, const HAction a, void* user_data);
+HParser* h_in(const uint8_t *charset, size_t length);
+HParser* h_not_in(const uint8_t *charset, size_t length);
+HParser* h_end_p();
+HParser* h_nothing_p();
+HParser* h_sequence__a(void *args[]);
+HParser* h_choice__a(void *args[]);
+HParser* h_permutation__a(void *args[]);
+HParser* h_butnot(const HParser* p1, const HParser* p2);
+HParser* h_difference(const HParser* p1, const HParser* p2);
+HParser* h_xor(const HParser* p1, const HParser* p2);
+HParser* h_many(const HParser* p);
+HParser* h_many1(const HParser* p);
+HParser* h_repeat_n(const HParser* p, const size_t n);
+HParser* h_optional(const HParser* p);
+HParser* h_ignore(const HParser* p);
+HParser* h_sepBy(const HParser* p);
+HParser* h_sepBy1(const HParser* p);
+HParser* h_epsilon_p();
+HParser* h_length_value(const HParser* length, const HParser* value);
+HParser* h_attr_bool(const HParser* p, HPredicate pred, void* user_data);
+HParser* h_and(const HParser* p);
+HParser* h_not(const HParser* p);
+HParser* h_indirect(const HParser* p);
+void h_bind_indirect(HParser* indirect, const HParser* inner);
+HParser* h_with_endianness(char endianness, const HParser* p);
+HParser* h_put_value(const HParser* p, const char* name);
+HParser* h_get_value(const char* name);
+HParser* h_bind(const HParser *p, HContinuation k, void *env);
+
+int h_compile(HParser* parser, HParserBackend backend, const void* params);
+]]
+local h = ffi.load("hammer")
+
+local function helper(a, n, b, ...)
+ if n == 0 then return a
+ else return b, helper(a, n-1, ...) end
+end
+local function append(a, ...)
+ return helper(a, select('#', ...), ...)
+end
+
+local parser
+local mt = {
+ __index = {
+ parse = function(p, str) return h.h_parse(p, str, #str) end,
+ },
+}
+parser = ffi.metatype("HParser", mt)
+
+local function token(str)
+ return h.h_token(str, #str)
+end
+local function ch(c)
+ if type(c) == "number" then
+ return h.h_ch(c)
+ else
+ return h.h_ch(c:byte)
+ end
+end
+local function ch_range(lower, upper)
+ if type(lower) == "number" and type(upper) == "number" then
+ return h.h_ch_range(lower, upper)
+ -- FIXME this is really not thorough type checking
+ else
+ return h.h_ch_range(lower:byte, upper:byte)
+ end
+end
+local function int_range(parser, lower, upper)
+ return h.h_int_range(parser, lower, upper)
+end
+local function bits(len, sign)
+ return h.h_bits(len, sign)
+end
+local function int64()
+ return h.h_int64()
+end
+local function int32()
+ return h.h_int32()
+end
+local function int16()
+ return h.h_int16()
+end
+local function int8()
+ return h.h_int8()
+end
+local function uint64()
+ return h.h_uint64()
+end
+local function uint32()
+ return h.h_uint32()
+end
+local function uint16()
+ return h.h_uint16()
+end
+local function uint8()
+ return h.h_uint8()
+end
+local function whitespace(parser)
+ return h.h_whitespace(parser)
+end
+local function left(parser1, parser2)
+ return h.h_left(parser1, parser2)
+end
+local function right(parser1, parser2)
+ return h.h_right(parser1, parser2)
+end
+local function middle(parser1, parser2, parser3)
+ return h.h_middle(parser1, parser2, parser3)
+end
+-- There could also be an overload of this that doesn't
+-- bother with the env pointer, and passes it as NIL by
+-- default, but I'm not going to deal with overloads now.
+local function action(parser, action, user_data)
+ local cb = ffi.cast("HAction", action)
+ return h.h_action(parser, cb, user_data)
+end
+local function in(charset)
+ return h.h_in(charset, #charset)
+end
+local function not_in(charset)
+ return h.h_not_in(charset, #charset)
+end
+local function end_p()
+ return h.h_end_p()
+end
+local function nothing_p()
+ return h.h_nothing_p()
+end
+local function sequence(...)
+ local parsers = append(nil, ...)
+ return h.h_sequence__a(parsers)
+end
+local function choice(...)
+ local parsers = append(nil, ...)
+ return h.h_choice__a(parsers)
+end
+local function permutation(...)
+ local parsers = append(nil, ...)
+ return h.h_permutation__a(parsers)
+end
+local function butnot(parser1, parser2)
+ return h.h_butnot(parser1, parser2)
+end
+local function difference(parser1, parser2)
+ return h.h_difference(parser1, parser2)
+end
+local function xor(parser1, parser2)
+ return h.h_xor(parser1, parser2)
+end
+local function many(parser)
+ return h.h_many(parser)
+end
+local function many1(parser)
+ return h.h_many1(parser)
+end
+local function repeat_n(parser, n)
+ return h.h_repeat_n(parser, n)
+end
+local function optional(parser)
+ return h.h_optional(parser)
+end
+local function ignore(parser)
+ return h.h_ignore(parser)
+end
+local function sepBy(parser)
+ return h.h_sepBy(parser)
+end
+local function sepBy1(parser)
+ return h.h_sepBy1(parser)
+end
+local function epsilon_p()
+ return h.h_epsilon_p()
+end
+local function length_value(length, value)
+ return h.h_length_value(length, value)
+end
+local function attr_bool(parser, predicate, user_data)
+ local cb = ffi.cast("HPredicate", predicate)
+ return h.h_attr_bool(parser, cb, user_data)
+end
+local function and(parser)
+ return h.h_and(parser)
+end
+local function not(parser)
+ return h.h_not(parser)
+end
+local function indirect(parser)
+ return h.h_indirect(parser)
+end
+local function bind_indirect(indirect, inner)
+ return h.h_bind_indirect(indirect, inner)
+end
+local function with_endianness(endianness, parser)
+ return h.h_with_endianness(endianness, parser)
+end
+local function put_value(parser, name)
+ return h.h_put_value(parser, name)
+end
+local function get_value(name)
+ return h.h_get_value(parser, name)
+end
+local function bind(parser, continuation, env)
+ local cb = ffi.cast("HContinuation", continuation)
+ return h.h_bind(parser, cb, env)
+end
+
+local function compile(parser, backend, params)
+ return h.h_compile(parser, backend, params)
+end
diff --git a/src/bindings/lua/test.lua b/src/bindings/lua/test.lua
new file mode 100644
index 0000000000000000000000000000000000000000..a5ee79773ecb590bc351ce15b817d8810db6ce5d
--- /dev/null
+++ b/src/bindings/lua/test.lua
@@ -0,0 +1,840 @@
+describe("Combinator tests", function()
+ local hammer
+
+ setup(function()
+ hammer = require("hammer")
+ end)
+
+ teardown(function()
+ hammer = nil
+ end)
+
+ describe("Token tests", function()
+ local parser = hammer.token("95" .. string.char(0xa2))
+ it("parses a token", function()
+ local ret = parser:parse("95" .. string.char(0xa2))
+ assert.are.same(ret.ast.bytes, "95" .. string.char(0xa2))
+ end)
+ it("does not parse an incomplete token", function()
+ local ret = parser:parse("95")
+ assert.is_falsy(ret)
+ end)
+ end)
+
+ describe("Char tests", function()
+ local parser = hammer.ch(0xa2)
+ it("parses a matching char", function()
+ local ret = parser:parse(string.char(0xa2))
+ assert.are.same(ret.ast.uint, string.char(0xa2))
+ end)
+ it("rejects a non-matching char", function()
+ local ret = parser:parse(string.char(0xa3))
+ assert.is_falsy(ret)
+ end)
+ end)
+
+ describe("Char range tests", function()
+ local parser = hammer.ch_range("a", "c")
+ it("parses a char in the range", function()
+ local ret = parser:parse("b")
+ assert.are.same(ret.ast.uint, "b")
+ end)
+ it("rejects a char outside the range", function()
+ local ret = parser:parse("d")
+ assert.is_falsy(ret)
+ )
+ end)
+
+ describe("Signed 64-bit int tests", function()
+ local parser = hammer.int64()
+ it("parses a valid 64-bit int", function()
+ local ret = parser:parse(string.char(0xff, 0xff, 0xff, 0xfe, 0x00, 0x00, 0x00, 0x00))
+ assert.are.same(ret.ast.sint, -0x200000000)
+ end)
+ it("does not parse an invalid 64-bit int", function()
+ local ret = parser:parse(string.char(0xff, 0xff, 0xff, 0xfe, 0x00, 0x00, 0x00))
+ assert.is_falsy(ret)
+ end)
+ end)
+
+ describe("Signed 32-bit int tests", function()
+ local parser = hammer.int32()
+ it("parses a valid 32-bit int", function()
+ local ret = parser:parse(string.char(\xff, 0xfe, 0x00, 0x00))
+ assert.are.same(ret.ast.sint, -0x20000)
+ end)
+ it("does not parse an invalid 32-bit int", function()
+ local ret = parser:parse(string.char(0xff, 0xfe, 0x00))
+ assert.is_falsy(ret)
+ end)
+ end)
+
+ describe("Signed 16-bit int tests", function()
+ local parser = hammer.int16()
+ it("parses a valid 16-bit int", function()
+ local ret = parser:parse(string.char(0xfe, 0x00))
+ assert.are.same(ret.ast.sint, -0x200)
+ end)
+ it("does not parse an invalid 16-bit int", function()
+ local ret = parser:parse((string.char(0xfe))
+ assert.is_falsy(ret)
+ end)
+ end)
+
+ describe("Signed 8-bit int tests", function()
+ local parser = hammer.int8()
+ it("parses a valid 8-bit int", function()
+ local ret = parser:parse(string.char(0x88))
+ assert.are.same(ret.ast.sint, -0x78)
+ end)
+ it("does not parse an invalid 8-bit int", function()
+ local ret = parser:parse("")
+ assert.is_falsy(ret)
+ end)
+ end)
+
+ describe("Unsigned 64-bit int tests", function()
+ local parser = hammer.uint64()
+ it("parses a valid 64-bit unsigned int", function()
+ local ret = parser:parse(string.char(0x00, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00, 0x00))
+ assert.are.same(ret.ast.uint, 0x200000000)
+ end)
+ it("does not parse an invalid 64-bit unsigned int", function()
+ local ret = parser:parse(string.char(0x00, 0x00, 0x00, 0x02, 0x00, 0x00, 0x00))
+ assert.is_falsy(ret)
+ end)
+ end)
+
+ describe("Unsigned 32-bit int tests", function()
+ local parser = hammer.uint32()
+ it("parses a valid 32-bit unsigned int", function()
+ local ret = parser:parse(string.char(0x00, 0x02, 0x00, 0x00))
+ assert.are.same(ret.ast.uint, -0x20000)
+ end)
+ it("does not parse an invalid 32-bit unsigned int", function()
+ local ret = parser:parse(string.char(0x00, 0x02, 0x00))
+ assert.is_falsy(ret)
+ end)
+ end)
+
+ describe("Unsigned 16-bit int tests", function()
+ local parser = hammer.uint16()
+ it("parses a valid 16-bit unsigned int", function()
+ local ret = parser:parse(string.char(0x02, 0x00))
+ assert.are.same(ret.ast.uint, 0x200)
+ end)
+ it("does not parse an invalid 16-bit unsigned int", function()
+ local ret = parser:parse(string.char(0x02))
+ assert.is_falsy(ret)
+ end)
+ end)
+
+ describe("Unsigned 8-bit int tests", function()
+ local parser = hammer.uint8()
+ it("parses a valid 8-bit unsigned int", function()
+ local ret = parser:parse(string.char(0x78))
+ assert.are.same(ret.ast.uint, 0x78)
+ end)
+ it("does not parse an invalid 8=bit unsigned int", function()
+ local ret = parser:parse("")
+ assert.is_falsy(ret)
+ end)
+ end)
+
+ describe("Integer range tests", function()
+ local parser = hammer.int_range(hammer.uint8(), 3, 10)
+ it("parses a value in the range", function()
+ local ret = parser:parse(string.char(0x05))
+ assert.are.same(ret.ast.uint, 5)
+ end)
+ it("does not parse a value outside the range", function()
+ local ret = parser:parse(string.char(0xb))
+ assert.is_falsy(ret)
+ end)
+ end)
+
+ describe("Whitespace tests", function()
+ local parser = hammer.whitespace(hammer.ch("a"))
+ local parser2 = hammer.whitespace(hammer.end_p())
+ it("parses a string with no whitespace", function()
+ local ret = parser:parse("a")
+ assert.are.same(ret.ast.uint, "a")
+ end)
+ it("parses a string with a leading space", function()
+ local ret = parser:parse(" a")
+ assert.are.same(ret.ast.uint, "a")
+ end)
+ it("parses a string with leading spaces", function()
+ local ret = parser:parse(" a")
+ assert.are.same(ret.ast.uint, "a")
+ end)
+ it("parses a string with a leading tab", function()
+ local ret = parser:parse("\ta")
+ assert.are.same(ret.ast.uint, "a")
+ end)
+ it("does not parse a string with a leading underscore", function()
+ local ret = parser:parse("_a")
+ assert.is_falsy(ret)
+ end)
+ it("parses an empty string", function()
+ local ret = parser2:parse("")
+ assert.are.same(ret.ast, nil)
+ end)
+ it("parses a whitespace-only string", function()
+ local ret = parser2:parse(" ")
+ assert.are.same(ret.ast, nil)
+ end)
+ it("does not parse a string with leading whitespace and a trailing character", function()
+ local ret = parser2:parse(" x")
+ assert.is_falsy(ret)
+ end)
+ end)
+
+ describe("Leftmost-parser tests", function()
+ local parser = hammer.left(hammer.ch("a"), hammer.ch(" "))
+ it("parses the leftmost character", function()
+ local ret = parser:parse("a ")
+ assert.are.same(ret.ast.uint, "a")
+ end)
+ it("does not parse a string that is too short", function()
+ local ret = parser:parse("a")
+ assert.is_falsy(ret)
+ end)
+ it("does not parse a string that starts with the wrong character", function()
+ local ret = parser:parse(" ")
+ assert.is_falsy(ret)
+ end)
+ it("does not parse a string with the wrong character in the second place", function()
+ local ret = parser:parse("ab")
+ assert.is_falsy(ret)
+ end)
+ end)
+
+ describe("Rightmost-parser tests", function()
+ local parser = hammer.right(hammer.ch(" "), hammer.ch("a"))
+ it("parses the rightmost character", function()
+ local ret = parser:parse(" a")
+ assert.are.same(ret.ast.uint, "a")
+ end)
+ it("does not parse a string that starts with the wrong character", function()
+ local ret = parser:parse("a")
+ assert.is_falsy(ret)
+ end)
+ it("does not parse a string that is too short", function()
+ local ret = parser:parse(" ")
+ assert.is_falsy(ret)
+ end)
+ it("does not parse a string with the characters in the wrong order", function()
+ local ret = parser:parse("ba")
+ assert.is_falsy(ret)
+ end)
+ end)
+
+ describe("Middle-parser tests", function()
+ local parser = hammer.middle(hammer.ch(" "), hammer.ch("a"), hammer.ch(" ")
+ it("parses the middle character", function()
+ local ret = parser:parse(" a ")
+ assert.are.same(ret.ast.uint, "a")
+ end)
+ it("does not parse a string that is too short", function()
+ local ret = parser:parse("a")
+ assert.is_falsy(ret)
+ ret = parser:parse(" ")
+ assert.is_falsy(ret)
+ ret = parser:parse(" a")
+ assert.is_falsy(ret)
+ ret = parser:parse("a ")
+ assert.is_falsy(ret)
+ end)
+ it("does not parse a string with the wrong character in the middle", function()
+ ret = parser:parse(" b ")
+ assert.is_falsy(ret)
+ end)
+ it("does not parse a string that starts with the wrong character", function()
+ ret = parser:parse("ba ")
+ assert.is_falsy(ret)
+ end)
+ it("does not parse a string that ends with the wrong character", function()
+ ret = parser:parse(" ab")
+ assert.is_falsy(ret)
+ end)
+ end)
+
+ describe("Semantic action tests", function()
+ local function upcase(result, user_data)
+ local chars = result.ast.seq.elements
+ local ret = ""
+ for i, v in ipairs(chars)
+ do ret = ret .. string.char(v.uint):upper()
+ end
+ return ret
+ end
+ local parser = hammer.action(hammer.sequence(hammer.choice(hammer.ch("a"), hammer.ch("A")), hammer.choice(hammer.ch("b"), hammer.ch("B"))), upcase, nil)
+ it("converts a lowercase 'ab' to uppercase", function()
+ local ret = parser:parse("ab")
+ assert.are.same(ret.ast.seq, {"A", "B"})
+ end)
+ it("accepts an uppercase 'AB' unchanged", function()
+ local ret = parser:parse("AB")
+ assert.are.same(ret.ast.seq, {"A", "B"})
+ end)
+ it("rejects strings that don't match the underlying parser", function()
+ local ret = parser:parse()
+ assert.is_falsy(ret)
+ end)
+ end)
+
+ describe("Character set membership tests", function()
+ local parser = hammer.in({"a", "b", "c"})
+ it("parses a character that is in the included set", function()
+ local ret = parser:parse("b")
+ assert.are.same(ret.ast.uint, "b")
+ end)
+ it("does not parse a character that is not in the included set", function()
+ local ret = parser:parse("d")
+ assert.is_falsy(ret)
+ end)
+ end)
+
+ describe("Character set non-membership tests", function()
+ local parser = hammer.not_in({"a", "b", "c"})
+ it("parses a character that is not in the excluded set", function()
+ local ret = parser:parse("d")
+ assert.are.same(ret.ast.uint, "d")
+ end)
+ it("does not parse a character that is in the excluded set", function()
+ local ret = parser:parse("a")
+ assert.is_falsy(ret)
+ end)
+ end)
+
+ describe("End-of-input tests", function()
+ local parser = hammer.sequence(hammer.ch("a"), hammer.end_p())
+ it("parses a string that ends where it is expected to", function()
+ local ret = parser:parse("a")
+ assert.are.same(ret.ast.seq, {"a"})
+ end)
+ it("does not parse a string that is too long", function()
+ local ret = parser:parse("aa")
+ assert.is_falsy(ret)
+ end)
+ end)
+
+ describe("Bottom parser tests", function()
+ local parser = hammer.nothing_p()
+ it("always fails", function()
+ local ret = parser:parse("a")
+ assert.is_falsy(ret)
+ end)
+ end)
+
+ describe("Parser sequence tests", function()
+ local parser = hammer.sequence(hammer.ch("a"), hammer.ch("b"))
+ local parser2 = hammer.sequence(hammer.ch("a"), hammer.whitespace(hammer.ch("b")))
+ it("parses a string matching the sequence", function()
+ local ret = parser:parse("ab")
+ assert.are.same(ret.ast.seq, {"a", "b"})
+ end)
+ it("does not parse a string that is too short", function()
+ local ret = parser:parse("a")
+ assert.is_falsy(ret)
+ end)
+ it("does not parse a string with the sequence out of order", function()
+ local ret = parser:parse("ba")
+ assert.is_falsy(ret)
+ end)
+ it("parses a whitespace-optional string with no whitespace", function()
+ local ret = parser2:parse("ab")
+ assert.are.same(ret.ast.seq, {"a", "b"})
+ end)
+ it("parses a whitespace-optional string containing whitespace", function()
+ local ret = parser:parse("a b")
+ assert.are.same(ret.ast.seq, {"a", "b"})
+ ret = parser:parse("a b")
+ assert.are.same(ret.ast.seq, {"a", "b"})
+ end)
+ end)
+
+ describe("Choice-of-parsers tests", function()
+ local parser = hammer.choice(hammer.ch("a"), hammer.ch("b"))
+ it("parses a character in the choice set", function()
+ local ret = parser:parse("a")
+ assert.are.same(ret.ast.uint, "a")
+ ret = parser:parse("b")
+ assert.are.same(ret.ast.uint, "b")
+ end)
+ it("does not parse a character not in the choice set", function()
+ local ret = parser:parse("c")
+ assert.is_falsy(ret)
+ end)
+ end)
+
+ describe("X-but-not-Y tests", function()
+ local parser = hammer.butnot(hammer.ch("a"), hammer.token("ab"))
+ local parser2 = hammer.butnot(hammer.ch_range("0", "9"), hammer.ch("6"))
+ it("succeeds when 'a' matches but 'ab' doesn't", function()
+ local ret = parser:parse("a")
+ assert.are.same(ret.ast.uint, "a")
+ ret = parser:parse("aa")
+ assert.are.same(ret.ast.uint, "a")
+ end)
+ it("fails when p2's result is longer than p1's", function()
+ local ret = parser:parse("ab")
+ assert.is_falsy(ret)
+ end)
+ it("fails when p2's result is the same length as p1's", function()
+ local ret = parser2:parse("6")
+ assert.is_falsy(ret)
+ end)
+ end)
+
+ describe("Difference-of-parsers tests", function()
+ local parser = hammer.difference(hammer.token("ab"), hammer.ch("a"))
+ it("succeeds when 'ab' matches and its result is longer than the result for 'a'", function()
+ local ret = parser:parse("ab")
+ assert.are.same(ret.ast., )
+ end)
+ it("fails if 'ab' doesn't match", function()
+ local ret = parser:parse("a")
+ assert.is_falsy(ret)
+ end)
+ end)
+
+ describe("XOR-of-parsers tests", function()
+ local parser = hammer.xor(hammer.ch_range("0", "6"), hammer.ch_range("5", "9"))
+ it("parses a value only in the first range", function()
+ local ret = parser:parse("0")
+ assert.are.same(ret.ast.uint, "0")
+ end)
+ it("parses a value only in the second range", function()
+ local ret = parser:parse("9")
+ assert.are.same(ret.ast.uint, "9")
+ end)
+ it("does not parse a value inside both ranges", function()
+ local ret = parser:parse("5")
+ assert.is_falsy(ret)
+ end)
+ it("does not parse a value outside the range", function()
+ local ret = parser:parse("a")
+ assert.is_falsy(ret)
+ end)
+ end)
+
+ describe("Kleene * tests", function()
+ local parser = hammer.many(hammer.choice(hammer.ch("a"), hammer.ch("b")))
+ it("parses an empty string", function()
+ local ret = parser:parse("")
+ assert.are.same(ret.ast.seq, {})
+ end)
+ it("parses a single repetition of the pattern", function()
+ local ret = parser:parse("a")
+ assert.are.same(ret.ast.seq, {"a"})
+ ret = parser:parse("b")
+ assert.are.same(ret.ast.seq, {"b"})
+ end)
+ it("parses multiple repetitions of the pattern", function()
+ local ret = parser:parse("aabbaba")
+ assert.are.same(ret.ast.seq, {"a", "a", "b", "b", "a", "b", "a"})
+ end)
+ end)
+
+ describe("Kleene + tests", function()
+ local parser = hammer.many1(hammer.choice(hammer.ch("a"), hammer.ch("b")))
+ it("does not parse an empty string", function()
+ local ret = parser:parse("")
+ assert.is_falsy(ret)
+ end)
+ it("parses a single repetition of the pattern", function()
+ local ret = parser:parse("a")
+ assert.are.same(ret.ast.seq, {"a"})
+ ret = parser:parse("b")
+ assert.are.same(ret.ast.seq, {"b"})
+ end)
+ it("parses multiple repetitions of the pattern", function()
+ local ret = parser:parse("aabbaba")
+ assert.are.same(ret.ast.seq, {"a", "a", "b", "b", "a", "b", "a"})
+ end)
+ it("does not parse a string that does not start with one of the patterns to repeat", function()
+ local ret = parser:parse("daabbabadef")
+ assert.is_falsy(ret)
+ end)
+ end)
+
+ describe("Fixed-number-of-repetitions tests", function()
+ local parser = hammer.repeat_n(hammer.choice(hammer.ch("a"), hammer.ch("b")), 2)
+ it("does not parse a string without enough repetitions", function()
+ local ret = parser:parse("adef")
+ assert.is_falsy(ret)
+ end)
+ it("parses a string containing the correct number of repetitions", function()
+ local ret = parser:parse("abdef")
+ assert.are.same(ret.ast.seq, {"a", "b"})
+ end)
+ it("does not parse a string that does not start with a character in the repetition set", function()
+ local ret = parser:parse("dabdef")
+ assert.is_falsy(ret)
+ end)
+ end)
+
+ describe("Kleene ? tests", function()
+ local parser = hammer.sequence(hammer.ch("a"), hammer.optional(hammer.choice(hammer.ch("b"), hammer.ch("c"))), hammer.ch("d"))
+ it("parses a string containing either optional character", function()
+ local ret = parser:parse("abd")
+ assert.are.same(ret.ast.seq, {"a", "b", "d"})
+ ret = parser:parse("acd")
+ assert.are.same(ret.ast.seq, {"a", "c", "d"})
+ end)
+ it("parses a string missing one of the optional characters", function()
+ local ret = parser:parse("ad")
+ assert.are.same(ret.ast.seq, {"a", {}, "d"})
+ end
+ it("does not parse a string containing a character not among the optional ones", function()
+ local ret = parser:parse("aed")
+ assert.is_falsy(ret)
+ end)
+ end)
+
+ describe("'ignore' decorator tests", function()
+ local parser = hammer.sequence(hammer.ch("a"), hammer.ignore(hammer.ch("b")), hammer.ch("c"))
+ it("parses a string containing the pattern to ignore, and leaves that pattern out of the result", function()
+ local ret = parser:parse("abc")
+ assert.are.same(ret.ast.seq, {"a", "c"})
+ end)
+ it("does not parse a string not containing the pattern to ignore", function()
+ local ret = parser:parse("ac")
+ assert.is_falsy(ret)
+ end)
+ end)
+
+ describe("Possibly-empty separated lists", function()
+ local parser = hammer.sepBy(hammer.choice(hammer.ch("1"), hammer.ch("2"), hammer.ch("3")), hammer.ch(","))
+ it("parses an ordered list", function()
+ local ret = parser:parse("1,2,3")
+ assert.are.same(ret.ast.seq, {"1", "2", "3"})
+ end)
+ it("parses an unordered list", function()
+ local ret = parser:parse("1,3,2")
+ assert.are.same(ret.ast.seq, {"1", "3", "2"})
+ end)
+ it("parses a list not containing all options", function()
+ local ret = parser:parse("1,3")
+ assert.are.same(ret.ast.seq, {"1", "3"})
+ end)
+ it("parses a unary list", function()
+ local ret = parser:parse("3")
+ assert.are.same(ret.ast.seq, {"3"})
+ end)
+ it("parses an empty list", function()
+ local ret = parser:parse("")
+ assert.are.same(ret.ast.seq, {})
+ end)
+ end)
+
+ describe("Non-empty separated lists", function()
+ local parser = hammer.sepBy1(hammer.choice(hammer.ch("1"), hammer.ch("2"), hammer.ch("3")), hammer.ch(","))
+ it("parses an ordered list", function()
+ local ret = parser:parse("1,2,3")
+ assert.are.same(ret.ast.seq, {"1", "2", "3"})
+ end)
+ it("parses an unordered list", function()
+ local ret = parser:parse("1,3,2")
+ assert.are.same(ret.ast.seq, {"1", "3", "2"})
+ end)
+ it("parses a list not containing all options", function()
+ local ret = parser:parse("1,3")
+ assert.are.same(ret.ast.seq, {"1", "3"})
+ end)
+ it("parses a unary list", function()
+ local ret = parser:parse("3")
+ assert.are.same(ret.ast.seq, {"3"})
+ end)
+ it("does not parse an empty list", function()
+ local ret = parser:parse("")
+ assert.is_falsy(ret)
+ end)
+ end)
+
+ describe("Empty string tests", function()
+ local parser = hammer.sequence(hammer.ch("a"), hammer.epsilon_p(), hammer.ch("b"))
+ local parser2 = hammer.sequence(hammer.epsilon_p(), hammer.ch("a"))
+ local parser3 = hammer.sequence(hammer.ch("a"), hammer.epsilon_p())
+ it("parses an empty string between two characters", function()
+ local ret = parser:parse("ab")
+ assert.are.same(ret.ast.seq, {"a", "b"})
+ end)
+ it("parses an empty string before a character", function()
+ local ret = parser2:parse("a")
+ assert.are.same(ret.ast.seq, {"a"})
+ end)
+ it("parses a ", function()
+ local ret = parser3:parse("a")
+ assert.are.same(ret.ast.seq, {"a"})
+ end)
+ end)
+
+ describe("Attribute validation tests", function()
+ local function equals(result, user_data)
+ return result.ast.seq.elements[0].uint == result.ast.seq.elements[1].uint
+ end
+ local parser = hammer.attr_bool(hammer.many1(hammer.choice(hammer.ch("a"), hammer.ch("b"))), equals)
+ it("parses successfully when both characters are the same (i.e., the validation function succeeds)", function()
+ local ret = parser:parse("aa")
+ assert.are.same(ret.ast.seq, {"a", "a"})
+ ret = parser:parse("bb")
+ assert.are.same(ret.ast.seq, {"b", "b"})
+ end)
+ it("does not parse successfully when the characters are different (i.e., the validation function fails)", function()
+ local ret = parser:parse("ab")
+ assert.is_falsy(ret)
+ end)
+ end)
+
+ describe("Matching lookahead tests", function()
+ local parser = hammer.sequence(hammer.and(hammer.ch("0")), hammer.ch("0"))
+ local parser2 = hammer.sequence(hammer.and(hammer.ch("0")), hammer.ch("1"))
+ local parser3 = hammer.sequence(hammer.ch("1"), hammer.and(hammer.ch("2")))
+ it("parses successfully when the lookahead matches the next character to parse", function()
+ local ret = parser:parse("0")
+ assert.are.same(ret.ast.seq, {"0"})
+ end)
+ it("does not parse successfully when the lookahead does not match the next character to parse", function()
+ local ret = parser2:parse("0")
+ assert.is_falsy(ret)
+ end)
+ it("parses successfully when the lookahead is there", function()
+ local ret = parser3:parse("12")
+ assert.are.same(ret.ast.seq, {"1"})
+ end)
+ end)
+
+ describe("Non-matching lookahead tests", function()
+ local parser = hammer.sequence(hammer.ch("a"), hammer.choice(hammer.ch("+"), hammer.token("++")), hammer.ch("b"))
+ local parser2 = hammer.sequence(hammer.ch("a"), hammer.choice(hammer.sequence(hammer.ch("+"), hammer.not(hammer.ch("+"))), hammer.token("++")), hammer.ch("b"))
+ it("parses a single plus correctly in the 'choice' example", function()
+ local ret = parser:parse("a+b")
+ assert.are.same(ret.ast.seq, {"a", "+", "b"})
+ end)
+ it("does not parse a double plus correctly in the 'choice' example", function()
+ local ret = parser:parse("a++b")
+ assert.is_falsy(ret)
+ end)
+ it("parses a single plus correctly in the 'not' example", function()
+ local ret = parser2:parse("a+b")
+ assert.are.same(ret.ast.seq, {"a", {"+"}, "b"})
+ end)
+ it("parses a double plus correctly in the 'not' example", function()
+ local ret = parser2:parse("a++b")
+ assert.are.same(ret.ast.seq, {"a", "++", "b"})
+ end)
+ end)
+
+ describe("Left recursion tests", function()
+ local parser = hammer.indirect()
+ hammer.bind_indirect(parser, hammer.choice(hammer.sequence(parser, hammer.ch("a")), hammer.ch("a")))
+ it("parses the base case", function()
+ local ret = parser:parse("a")
+ assert.are.same(ret.ast.seq, {"a"})
+ end)
+ it("parses one level of recursion", function()
+ local ret = parser:parse("aa")
+ assert.are.same(ret.ast.seq, {"a", "a"})
+ end)
+ it("parses two levels of recursion", function()
+ local ret = parser:parse("aaa")
+ assert.are.same(ret.ast.seq, {{"a", "a"}, "a"})
+ end)
+ end)
+
+ describe("Right recursion tests", function()
+ local parser = hammer.indirect()
+ hammer.bind_indirect(parser, hammer.choice(hammer.sequence(hammer.ch("a"), parser), hammer.epsilon_p()))
+ it("parses the base case", function()
+ local ret = parser:parse("a")
+ assert.are.same(ret.ast.seq, {"a"})
+ end)
+ it("parses one level of recursion", function()
+ local ret = parser:parse("aa")
+ assert.are.same(ret.ast.seq, {"a", {"a"}})
+ end)
+ it("parses two levels of recursion", function()
+ local ret = parser:parse("aaa")
+ assert.are.same(ret.ast.seq, {"a", {"a", {"a"}}})
+ end)
+ end)
+
+ describe("Endianness tests", function()
+ local bit = require("bit")
+ local u32 = hammer.uint32()
+ local u5 = hammer.bits(5, false)
+ local bb = bit.bor(BYTE_BIG_ENDIAN, BIT_BIG_ENDIAN)
+ local bl = bit.bor(BYTE_BIG_ENDIAN, BIT_LITTLE_ENDIAN)
+ local lb = bit.bor(BYTE_LITTLE_ENDIAN, BIT_BIG_ENDIAN)
+ local ll = bit.bor(BYTE_LITTLE_ENDIAN, BIT_LITTLE_ENDIAN)
+ local parser1 = hammer.with_endianness(bb, u32)
+ local parser2 = hammer.with_endianness(bb, u5)
+ local parser3 = hammer.with_endianness(ll, u32)
+ local parser4 = hammer.with_endianness(ll, u5)
+ local parser5 = hammer.with_endianness(bl, u32)
+ local parser6 = hammer.with_endianness(bl, u5)
+ local parser7 = hammer.with_endianness(lb, u32)
+ local parser8 = hammer.with_endianness(lb, u5)
+ it("parses big-endian cases", function()
+ local ret = parser1:parse("abcd")
+ assert.are.same(ret.ast.uint, 0x61626364)
+ ret = parser2:parse("abcd")
+ assert.are.same(ret.ast.uint, 0xc)
+ end)
+ it("parses little-endian cases", function()
+ local ret = parser3:parse("abcd")
+ assert.are.same(ret.ast.uint, 0x61626364)
+ ret = parser4:parse("abcd")
+ assert.are.same(ret.ast.uint, 0xc)
+ end)
+ it("parses mixed-endian cases", function()
+ local ret = parser5:parse("abcd")
+ assert.are.same(ret.ast.uint, 0x61626364)
+ ret = parser6:parse("abcd")
+ assert.are.same(ret.ast.uint, 0x1)
+ ret = parser7:parse("abcd")
+ assert.are.same(ret.ast.uint, 0x64636261)
+ ret = parser8:parse("abcd")
+ assert.are.same(ret.ast.uint, 0xc)
+ end)
+ end)
+
+ describe("Symbol table tests", function()
+ local parser = hammer.sequence(hammer.put_value(hammer.uint8(),"size"), hammer.token("foo"), hammer.length_value(hammer.get_value("size"), hammer.uint8()))
+ it("parses a string that has enough bytes for the specified length", function()
+ local ret = parser:parse(string.char(0x06) .. "fooabcdef")
+ assert.are.same(ret.ast.seq.elements[1].bytes, "foo")
+ assert.are.same(ret.ast.seq.elements[2].seq, {0x61, 0x62, 0x63, 0x64, 0x65, 0x66})
+ end)
+ it("does not parse a string that does not have enough bytes for the specified length", function()
+ local ret = parser:parse()
+ assert.is_falsy(ret)
+ end)
+ end)
+
+ describe("Permutation tests", function()
+ local parser = hammer.permutation(hammer.ch("a"), hammer.ch("b"), hammer.ch("c"))
+ it("parses a permutation of 'abc'", function()
+ local ret = parser:parse("abc")
+ assert.are.same(ret.ast.seq, {"a", "b", "c"})
+ ret = parser:parse("acb")
+ assert.are.same(ret.ast.seq, {"a", "c", "b"})
+ ret = parser:parse("bac")
+ assert.are.same(ret.ast.seq, {"b", "a", "c"})
+ ret = parser:parse("bca")
+ assert.are.same(ret.ast.seq, {"b", "c", "a"})
+ ret = parser:parse("cab")
+ assert.are.same(ret.ast.seq, {"c", "a", "b"})
+ ret = parser:parse("cba")
+ assert.are.same(ret.ast.seq, {"c", "b", "a"})
+ end)
+ it("does not parse a string that is not a permutation of 'abc'", function()
+ local ret = parser:parse("a")
+ assert.is_falsy(ret)
+ ret = parser:parse("ab")
+ assert.is_falsy(ret)
+ ret = parser:parse("abb")
+ assert.is_falsy(ret)
+ end)
+ parser = hammer.permutation(hammer.ch("a"), hammer.ch("b"), hammer.optional(hammer.ch("c")))
+ it("parses a string that is a permutation of 'ab[c]'", function()
+ local ret = parser:parse("abc")
+ assert.are.same(ret.ast.seq, {"a", "b", "c"})
+ ret = parser:parse("acb")
+ assert.are.same(ret.ast.seq, {"a", "c", "b"})
+ ret = parser:parse("bac")
+ assert.are.same(ret.ast.seq, {"b", "a", "c"})
+ ret = parser:parse("bca")
+ assert.are.same(ret.ast.seq, {"b", "c", "a"})
+ ret = parser:parse("cab")
+ assert.are.same(ret.ast.seq, {"c", "a", "b"})
+ ret = parser:parse("cba")
+ assert.are.same(ret.ast.seq, {"c", "b", "a"})
+ ret = parser:parse("ab")
+ assert.are.same(ret.ast.seq, {"a", "b"})
+ ret = parser:parse("ba")
+ assert.are.same(ret.ast.seq, {"b", "a"})
+ end)
+ it("does not parse a string that is not a permutation of 'ab[c]'", function()
+ local ret = parser:parse("a")
+ assert.is_falsy(ret)
+ ret = parser:parse("b")
+ assert.is_falsy(ret)
+ ret = parser:parse("c")
+ assert.is_falsy(ret)
+ ret = parser:parse("ca")
+ assert.is_falsy(ret)
+ ret = parser:parse("cb")
+ assert.is_falsy(ret)
+ ret = parser:parse("cc")
+ assert.is_falsy(ret)
+ ret = parser:parse("ccab")
+ assert.is_falsy(ret)
+ ret = parser:parse("ccc")
+ assert.is_falsy(ret)
+ end)
+ parser = hammer.permutation(hammer.optional(hammer.ch("c")), hammer.ch("a"), hammer.ch("b"))
+ it("parses a string that is a permutation of '[c]ab'", function()
+ local ret = parser:parse("abc")
+ assert.are.same(ret.ast.seq, {"a", "b", "c"})
+ ret = parser:parse("acb")
+ assert.are.same(ret.ast.seq, {"a", "c", "b"})
+ ret = parser:parse("bac")
+ assert.are.same(ret.ast.seq, {"b", "a", "c"})
+ ret = parser:parse("bca")
+ assert.are.same(ret.ast.seq, {"b", "c", "a"})
+ ret = parser:parse("cab")
+ assert.are.same(ret.ast.seq, {"c", "a", "b"})
+ ret = parser:parse("cba")
+ assert.are.same(ret.ast.seq, {"c", "b", "a"})
+ ret = parser:parse("ab")
+ assert.are.same(ret.ast.seq, {"a", "b"})
+ ret = parser:parse("ba")
+ assert.are.same(ret.ast.seq, {"b", "a"})
+ end)
+ it("does not parse a string that is not a permutation of '[c]ab'", function()
+ local ret = parser:parse("a")
+ assert.is_falsy(ret)
+ ret = parser:parse("b")
+ assert.is_falsy(ret)
+ ret = parser:parse("c")
+ assert.is_falsy(ret)
+ ret = parser:parse("ca")
+ assert.is_falsy(ret)
+ ret = parser:parse("cb")
+ assert.is_falsy(ret)
+ ret = parser:parse("cc")
+ assert.is_falsy(ret)
+ ret = parser:parse("ccab")
+ assert.is_falsy(ret)
+ ret = parser:parse("ccc")
+ assert.is_falsy(ret)
+ end)
+ end)
+
+ -- describe("Monadic binding tests", function()
+ -- local function continuation(allocator, result, env)
+ -- local val = 0
+ -- for k, v in result.seq.elements
+ -- do val = val*10 + v->uint - 48
+ -- end
+ -- if val > 26 then
+ -- return nil
+ -- else
+ -- return hammer.ch
+ -- end
+ -- end
+ -- local parser = hammer.bind(hammer.many1(hammer.ch_range("0", "9")), continuation, "a")
+ -- it("parses a ", function()
+ -- local ret = parser:parse()
+ -- assert.are.same(ret.ast., )
+ -- end)
+ -- it("does not parse a ", function()
+ -- local ret = parser:parse()
+ -- assert.is_falsy(ret)
+ -- end)
+ -- end)
+end)
\ No newline at end of file