diff --git a/HACKING b/HACKING
index d923217e8007694f45de5f99802321ac6f9b37c0..970a2491e9bcc03b39cc7e3fd18b74c2da136d0a 100644
--- a/HACKING
+++ b/HACKING
@@ -52,3 +52,11 @@ documented), assume that
In other words, don't let the (memory manager) streams cross.
+Language-independent test suite
+===============================
+
+There is a language-independent representation of the Hammer test
+suite in `lib/test-suite`. This is intended to be used with the
+tsparser.pl prolog library, along with a language-specific frontend.
+
+No language-specific frontends have been written yet.
diff --git a/lib/test-suite b/lib/test-suite
new file mode 100644
index 0000000000000000000000000000000000000000..207ece4054fabb0c31041e8c394cbe28b60c33ef
--- /dev/null
+++ b/lib/test-suite
@@ -0,0 +1,312 @@
+# This file is to be processed with testgen.pl to produce test suites
+# for bindings. The only escape code that is valid is \x
+
+# The only type that may change between bindings is "char", which gets
+# converted into the most convenient representation of a character in
+# a given language. In C, this is a TT_UINT; in P*, it's a
+# single-character string. In this file, a character is wrapped with
+# single quotes.
+
+# Strings may be given in C syntax, with no escapes other than \xHH or
+# in hex syntax, like <31.32.33> for "123".
+
+# indirect/bind_indirect can be used via subparsers. A subparser is
+# introduced with a "subparser $<name> = <definition()>;" command.
+# First, all subparsers are declared with h_indirect(). Then, they
+# are all bound. Inside a definition, all subparsers can be
+# referenced by $<name>.
+
+token {
+ parser token("95\xa2");
+ test "95\xa2" --> "95\xa2";
+ test "95\xa2" --> fail;
+}
+
+ch {
+ parser ch(0xA2);
+ test "\xa2" --> '\xa2';
+ test "\xa3" --> fail;
+}
+
+ch_range {
+ parser ch_range(0x61,0x63);
+ test "b" --> 'b';
+ test "d" --> fail;
+}
+
+int64 {
+ parser int64();
+ test <ff.ff.ff.fe.00.00.00.00> --> s-0x200000000;
+ test <ff.ff.ff.fe.00.00.00> --> fail;
+}
+
+int32 {
+ parser int32();
+ test <ff.fe.00.00> --> s-0x20000;
+ test <ff.fe.00> --> fail;
+ test <00.02.00.00> --> s0x20000;
+ test <00.02.00> --> fail;
+}
+
+int16 {
+ parser int16();
+ test <fe.00> --> s-0x200;
+ test <fe> --> fail;
+ test <02.00> --> s0x200;
+ test <02> --> fail;
+}
+
+int8 {
+ parser int8();
+ test <88> --> s-0x78;
+ test <> --> fail;
+}
+
+uint64 {
+ parser uint64();
+ test <00.00.00.02.00.00.00.00> --> u0x200000000;
+ test <00.00.00.02.00.00.00> --> fail;
+}
+
+uint32 {
+ parser uint32();
+ test <00.02.00.00> --> u0x20000;
+ test <00.02.00> --> fail;
+}
+
+uint16 {
+ parser uint16();
+ test <02.00> --> u0x200;
+ test <02> --> fail;
+}
+
+uint8 {
+ parser uint8();
+ test <78> --> u0x78;
+ test <> --> fail;
+}
+
+int_range {
+ parser int_range(uint8(), 0x3, 0x10);
+ test <05> --> u0x05;
+ test <0b> --> fail;
+}
+
+whitespace {
+ parser whitespace(ch(0x61));
+ test "a" --> 'a';
+ test " a" --> 'a';
+ test " a" --> 'a';
+ test "\x09a" --> 'a';
+ test "_a" --> fail;
+
+ parser whitespace(end_p());
+ test "" --> none;
+ test " " --> none;
+ test " x" --> fail;
+}
+
+left {
+ parser left(ch(0x61), ch(0x20));
+ test "a " --> 'a';
+ test "a" --> fail;
+ test " " --> fail;
+ test "ba" --> fail;
+}
+
+middle {
+ parser middle(ch(' '), ch('a'), ch(' '));
+ test " a " --> 'a';
+ test "a" --> fail;
+ test " a" --> fail;
+ test "a " --> fail;
+ test " b " --> fail;
+ test "ba " --> fail;
+ test " ab" --> fail;
+}
+
+# TODO: action
+
+in {
+ parser in("abc");
+ test "b" --> 'b';
+ test "d" --> fail;
+}
+
+not_in {
+ parser not_in("abc");
+ test "d" --> 'd';
+ test "a" --> fail;
+}
+
+end_p {
+ parser sequence(ch('a'), end_p());
+ test "a" --> ['a'];
+ test "aa" --> fail;
+}
+
+nothing_p {
+ parser nothing_p();
+ test "a" --> fail;
+}
+
+sequence {
+ parser sequence(ch('a'), ch('b'));
+ test "ab" --> ['a', 'b'];
+ test "a" --> fail;
+ test "b" --> fail;
+
+ parser sequence(ch('a'), whitespace(ch('b')));
+ test "ab" --> ['a','b'];
+ test "a b" --> ['a','b'];
+ test "a b" --> ['a','b'];
+}
+
+choice {
+ parser choice(ch('a'), ch('b'));
+ test "a" --> 'a';
+ test "b" --> 'b';
+ test "ab" --> 'a';
+ test "c" --> fail;
+}
+
+butnot {
+ parser butnot(ch('a'), token("ab"));
+ test "a" --> 'a';
+ test "ab" --> fail;
+ test "aa" --> 'a';
+
+ parser butnot(ch_range('0','9'), ch('6'));
+ test "5" --> '5';
+ test "6" --> fail;
+}
+
+difference {
+ # Not sure this actually tests anything
+ parser difference(token("ab"), ch('a'));
+ test "ab" --> "ab";
+ test "a" --> fail;
+}
+
+xor {
+ parser xor(ch_range('0','6'), ch_range('5','9'));
+ test "0" --> '0';
+ test "9" --> '9';
+ test "5" --> fail;
+ test "a" --> fail;
+}
+
+many {
+ # TODO: Some of the parsers in the original were commented out.
+ parser many(choice(ch('a'), ch('b')));
+ test "" --> [];
+ test "a" --> ['a'];
+ test "b" --> ['b'];
+ test "aabbaba" --> ['a','a','b','b','a','b','a'];
+}
+
+many1 {
+ # TODO: Some of the checks in the original were commented out.
+ parser many1(choice(ch('a'),ch('b')));
+ test "" --> fail;
+ test "a" --> ['a'];
+ test "b" --> ['b'];
+ test "aabbaba" --> ['a','a','b','b','a','b','a'];
+ test "daabbabadef" --> fail;
+}
+
+repeat-n {
+ parser repeat_n(choice(ch('a'),ch('b')),0x2);
+ test "adef" --> fail;
+ test "abdef" --> ['a','b'];
+ test "dabdef" --> fail;
+}
+
+optional {
+ parser sequence(ch('a'), optional(choice(ch('b'),ch('c'))), ch('d'));
+ test "abd" --> ['a','b','d'];
+ test "acd" --> ['a','c','d'];
+ test "ad" --> ['a',none,'d'];
+ test "aed" --> fail;
+ test "ab" --> fail;
+ test "ac" --> fail;
+}
+
+ignore {
+ parser sequence(ch('a'), ignore(ch('b')), ch('c'));
+ test "abc" --> ['a','c'];
+ test "ac" --> fail;
+}
+
+sepBy {
+ parser sepBy(choice(ch('1'), ch('2'), ch('3')), ch(','));
+ test "1,2,3" --> ['1','2','3'];
+ test "1,3,2" --> ['1','3','2'];
+ test "1,3" --> ['1','3'];
+ test "3" --> ['3'];
+ test "" --> [];
+}
+
+sepBy1 {
+ parser sepBy1(choice(ch('1'), ch('2'), ch('3')), ch(','));
+ test "1,2,3" --> ['1','2','3'];
+ test "1,3,2" --> ['1','3','2'];
+ test "1,3" --> ['1','3'];
+ test "3" --> ['3'];
+ test "" --> fail;
+}
+
+# TODO: attr_bool;
+and {
+ parser sequence(and(ch('0')), ch('0'));
+ test "0" --> ['0'];
+ test "1" --> fail;
+ parser sequence(and(ch('0')), ch('1'));
+ test "0" --> fail;
+ test "1" --> fail;
+ parser sequence(ch('1'), and(ch('2')));
+ test "12" --> ['1'];
+ test "13" --> fail;
+}
+
+not {
+ parser sequence(ch('a'), choice(token('+'), token("++")), ch('b'));
+ test "a+b" --> ['a',"+",'b'];
+ test "a++b" --> fail;
+
+ parser sequence(ch('a'), choice(sequence(token('+'), not(ch('+'))),
+ token("++")),
+ ch('b'));
+ test "a+b" --> ['a', ["+"], 'b'];
+ test "a++b" --> ['a', "++", 'b'];
+}
+
+leftrec {
+ subparser $lr = choice(sequence($lr, ch('a')), epsilon_p());
+ parser $lr;
+ test "a" --> ['a'];
+ test "aa" --> [['a'],'a'];
+ test "aaa" --> [[['a'],'a'],'a'];
+}
+
+rightrec {
+ subparser $rr = choice(sequence(ch('a'), $rr), epsilon_p());
+ parser $rr;
+ test "a" --> ['a'];
+ test "aa" --> ['a',['a']];
+ test "aaa" --> ['a',['a',['a']]];
+}
+
+ambiguous {
+ subparser $d = ch('d');
+ subparser $p = ch('+');
+ subparser $e = choice(sequence($e, $p, $e), $d);
+ # TODO: implement action/h_act_flatten
+ parser $e;
+
+ test "d" --> 'd';
+ test "d+d" --> ['d','+','d'];
+ test "d+d+d" --> [['d','+','d'],'+','d'];
+}
+
+
diff --git a/lib/testgen.pl b/lib/testgen.pl
new file mode 100644
index 0000000000000000000000000000000000000000..c8ff7e0888f08ab0be5af405e70b934ab1b825f9
--- /dev/null
+++ b/lib/testgen.pl
@@ -0,0 +1,197 @@
+% -*- prolog -*-
+
+% test suite grammar
+:- module(tsparser,
+ [read_tc/1]).
+:- expects_dialect(swi).
+:- use_module(library(pure_input)).
+
+
+nl --> "\r", !.
+nl --> "\n", !.
+ws --> " ", !.
+ws --> "\t", !.
+ws --> nl, !.
+nonws --> ws, !, {fail}.
+nonws --> [_]. % any character other than whitespace
+tilEOL --> nl, !.
+tilEOL --> [_], tilEOL.
+comment --> "#", tilEOL.
+layout1 --> ws, !; comment.
+layout --> layout1, !, layout.
+layout --> [].
+eof([],[]).
+
+idChar(Chr) -->
+ [Chr],
+ {memberchk(Chr, "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ")}.
+idTailChar(Chr) --> idChar(Chr), !.
+idTailChar(Chr) -->
+ [Chr],
+ {memberchk(Chr, "1234567890-_")}.
+
+id(Name) --> % toplevel lexical token
+ idChar(Chr),
+ idTail(Rest),
+ {atom_codes(Name, [Chr|Rest])}.
+idTail([Chr|Rest]) -->
+ idTailChar(Chr), !,
+ idTail(Rest).
+idTail([]) --> [].
+
+string(Str) --> "\"", qstring(Str), "\"", !.
+string(Str) --> "<", hexstring(Str), ">".
+
+qchar(C) --> "\\x", !, hexbyte(C).
+qchar(C) --> [C], {C >= 0x20, C < 0x7F}.
+qstring([], [34|R], [34|R]) :- !.
+qstring([C|Rest]) -->
+ qchar(C), qstring(Rest).
+qstring([]).
+hexstring([C|Rest]) -->
+ hexbyte(C), !,
+ hexstring_post(Rest).
+hexstring([]) -->
+ [].
+hexstring_post([C|Rest]) -->
+ ".", !,
+ hexbyte(C),
+ hexstring_post(Rest).
+hexstring_post([]) --> [].
+
+hexchr(C) --> [C0], {memberchk(C0, "0123456789"), C is C0 - 0x30}, !.
+hexchr(C) --> [C0], {memberchk(C0, "abcdef"), C is C0 - 0x61 + 10}, !.
+hexchr(C) --> [C0], {memberchk(C0, "ABCDEF"), C is C0 - 0x41 + 10}, !.
+hexbyte(C) -->
+ hexchr(C0),
+ hexchr(C1),
+ {C is C0 * 16 + C1}.
+hexnum(A) -->
+ hexchr(C0),
+ hexnum(A, C0).
+hexnum(A, Acc) -->
+ hexchr(C0), !,
+ {AccP is Acc * 16 + C0},
+ hexnum(A, AccP).
+hexnum(Acc, Acc) --> [].
+
+
+parse_result_seq([A|Ar]) -->
+ parse_result(A), layout,
+ parse_result_seq_tail(Ar).
+parse_result_seq([]) -->
+ layout.
+parse_result_seq_tail([]) --> [].
+parse_result_seq_tail([A|Ar]) -->
+ ",", layout,
+ parse_result(A), layout,
+ parse_result_seq_tail(Ar).
+
+parse_result(seq(A)) -->
+ "[", layout,
+ parse_result_seq(A), % eats trailing ws
+ "]", !.
+parse_result(char(A)) -->
+ "'", qchar(A), "'", !.
+parse_result(uint(A)) -->
+ "u0x", hexnum(A), !.
+parse_result(sint(A)) -->
+ "s0x", hexnum(A), !.
+parse_result(sint(A)) -->
+ "s-0x", hexnum(A0), {A is -A0}, !.
+parse_result(string(A)) -->
+ string(A), !.
+parse_result(none) --> "none".
+
+parser_arg(A) --> parser(A), !.
+parser_arg(string(A)) --> string(A), !.
+parser_arg(char(A)) --> "'", qchar(A), "'", !.
+parser_arg(num(A)) --> "0x", hexnum(A).
+
+parser_arglist_tail([A0|As]) -->
+ % eats leading whitespace
+ ",", !, layout,
+ parser_arg(A0), layout,
+ parser_arglist_tail(As).
+parser_arglist_tail([]) --> [].
+
+parser_arglist([]) --> [].
+parser_arglist([A0|As]) -->
+ parser_arg(A0),
+ layout,
+ parser_arglist_tail(As). % eats trailing whitespace
+
+
+parser(ref(Name)) -->
+ "$", !, id(Name).
+parser(parser(Name, Args)) -->
+ id(Name), layout,
+ "(", layout,
+ parser_arglist(Args), % eats trailing whitespace
+ ")".
+
+testcase_elem(subparser(Name,Value)) -->
+ "subparser", layout1, layout, !,
+ "$", id(Name), layout,
+ "=", layout,
+ parser(Value), layout,
+ ";".
+testcase_elem(parser(Value)) -->
+ "parser", layout1, layout, !,
+ parser(Value), layout,
+ ";".
+testcase_elem(testFail(Input)) -->
+ "test", layout1, layout,
+ string(Input), layout,
+ "-->", layout,
+ "fail", layout,
+ ";", !.
+testcase_elem(test(Input, Expected)) -->
+ "test", layout1, layout,
+ string(Input), layout,
+ "-->", layout,
+ parse_result(Expected), layout,
+ ";", !.
+
+testcase_body([Elem|Rest]) -->
+ testcase_elem(Elem), layout, !,
+ testcase_body(Rest).
+testcase_body([]) --> [].
+
+testcase(testcase(Name,Content)) -->
+ id(Name),
+ layout,
+ "{",
+ layout,
+ testcase_body(Content),
+ layout,
+ "}".
+
+testcases([]) -->
+ layout, eof, !.
+testcases([TC|Rest]) -->
+ layout,
+ testcase(TC),
+ testcases(Rest).
+testsuite(A) -->
+ testcases(A).
+
+
+%slurp(File, Lst) :-
+% get_code(File, C),
+% (C == -1 ->
+% Lst = [];
+% slurp(File, Rest),
+% Lst = [C|Rest]), !.
+
+read_tc(A) :-
+ phrase_from_file(testsuite(A), 'test-suite').
+
+%read_tc(A) :-
+% read_tc('test-suite', A).
+%read_tc(Name, A) :-
+% open(Name, read, File),
+% slurp(File, Contents),
+% close(File), !,
+% phrase(testsuite(A), Contents).
+