diff --git a/src/backends/lalr.c b/src/backends/lalr.c
index 14f64cd1a23cf2276a1377e0d1b78c3a24125ed8..272f00d6c598e59bdf253363c8f4440e619e47e3 100644
--- a/src/backends/lalr.c
+++ b/src/backends/lalr.c
@@ -52,7 +52,7 @@ static void transform_productions(const HLRTable *table, HLREnhGrammar *eg,
if (xAy->type != HCF_CHOICE) {
return;
}
- // XXX CHARSET?
+ // NB: nothing to do on quasi-terminal CHARSET which carries no list of rhs's
HArena *arena = eg->arena;
@@ -286,14 +286,28 @@ int h_lalr_compile(HAllocator* mm__, HParser* parser, const void* params)
HHashSet *lhss = h_hashtable_get(eg->corr, item->lhs);
assert(lhss != NULL);
H_FOREACH_KEY(lhss, HCFChoice *lhs)
- assert(lhs->type == HCF_CHOICE); // XXX could be CHARSET?
-
- for(HCFSequence **p=lhs->seq; *p; p++) {
- HCFChoice **rhs = (*p)->items;
- if(!match_production(eg, rhs, item->rhs, state)) {
- continue;
- }
-
+ assert(lhs->type == HCF_CHOICE || lhs->type == HCF_CHARSET);
+
+ bool match = false;
+ if(lhs->type == HCF_CHOICE) {
+ for(HCFSequence **p=lhs->seq; *p; p++) {
+ HCFChoice **rhs = (*p)->items;
+ if(match_production(eg, rhs, item->rhs, state)) {
+ match = true;
+ break;
+ }
+ }
+ } else { // HCF_CHARSET
+ assert(item->rhs[0] != NULL);
+ assert(item->rhs[1] == NULL);
+ assert(item->rhs[0]->type == HCF_CHAR);
+ HLRTransition *t = h_hashtable_get(eg->smap, lhs);
+ assert(t != NULL);
+ match = (t->to == state
+ && charset_isset(lhs->charset, item->rhs[0]->chr));
+ }
+
+ if(match) {
// the left-hand symbol's follow set is this production's
// contribution to the lookahead
const HStringMap *fs = h_follow(1, eg->grammar, lhs);
@@ -304,7 +318,8 @@ int h_lalr_compile(HAllocator* mm__, HParser* parser, const void* params)
// for each lookahead symbol, put action into table cell
if(terminals_put(table->tmap[state], fs, action) < 0)
inadeq = true;
- } H_END_FOREACH // enhanced production
+ }
+ H_END_FOREACH // enhanced production
H_END_FOREACH // reducible item
if(inadeq) {
diff --git a/src/cfgrammar.c b/src/cfgrammar.c
index a8761b8d537ec236f7a4876e1ad86a30742df988..6cd87ee574d2819ec7a54f6b0d140add1c52209e 100644
--- a/src/cfgrammar.c
+++ b/src/cfgrammar.c
@@ -672,7 +672,7 @@ static void stringset_extend(HCFGrammar *g, HStringMap *ret,
}
-void h_pprint_char(FILE *f, char c)
+void h_pprint_char(FILE *f, uint8_t c)
{
switch(c) {
case '"': fputs("\\\"", f); break;
@@ -685,12 +685,12 @@ void h_pprint_char(FILE *f, char c)
if (isprint((int)c)) {
fputc(c, f);
} else {
- fprintf(f, "\\x%.2X", c);
+ fprintf(f, "\\x%.2X", (unsigned int)c);
}
}
}
-static void pprint_charset_char(FILE *f, char c)
+static void pprint_charset_char(FILE *f, uint8_t c)
{
switch(c) {
case '"': fputc(c, f); break;
@@ -896,8 +896,8 @@ pprint_stringmap_elems(FILE *file, bool first, char *prefix, size_t n, char sep,
if (map->epsilon_branch) {
if (!first) {
fputc(sep, file);
- first=false;
}
+ first=false;
if (n==0) {
fputs("\"\"", file);
} else {
@@ -915,8 +915,8 @@ pprint_stringmap_elems(FILE *file, bool first, char *prefix, size_t n, char sep,
if (map->end_branch) {
if (!first) {
fputs(",\"", file);
- first=false;
}
+ first=false;
if (n>0) {
fputs("\"\"", file);
}
diff --git a/src/cfgrammar.h b/src/cfgrammar.h
index 9cefc62e83f07048dc2a24f0cda1bde28ca72066..0d31ef57116fadad6700241cf805bf325f550c69 100644
--- a/src/cfgrammar.h
+++ b/src/cfgrammar.h
@@ -102,4 +102,4 @@ void h_pprint_stringset(FILE *file, const HStringMap *set, int indent);
void h_pprint_stringmap(FILE *file, char sep,
void (*valprint)(FILE *f, void *env, void *val), void *env,
const HStringMap *map);
-void h_pprint_char(FILE *file, char c);
+void h_pprint_char(FILE *file, uint8_t c);
diff --git a/src/parsers/many.c b/src/parsers/many.c
index 51d733fcf87e3191e6f413a9513ac7900d29d8f2..cae2b0eade03450cae13f48e8f53c37db4237721 100644
--- a/src/parsers/many.c
+++ b/src/parsers/many.c
@@ -59,6 +59,32 @@ static bool many_isValidCF(void *env) {
repeat->sep->vtable->isValidCF(repeat->sep->env)));
}
+// turn (_ x (_ y (_ z ()))) into (x y z) where '_' are optional
+static HParsedToken *reshape_many(const HParseResult *p, void *user)
+{
+ HCountedArray *seq = h_carray_new(p->arena);
+
+ const HParsedToken *tok = p->ast;
+ while(tok) {
+ assert(tok->token_type == TT_SEQUENCE);
+ if(tok->seq->used > 0) {
+ size_t n = tok->seq->used;
+ assert(n <= 3);
+ h_carray_append(seq, tok->seq->elements[n-2]);
+ tok = tok->seq->elements[n-1];
+ } else {
+ tok = NULL;
+ }
+ }
+
+ HParsedToken *res = a_new_(p->arena, HParsedToken, 1);
+ res->token_type = TT_SEQUENCE;
+ res->seq = seq;
+ res->index = p->ast->index;
+ res->bit_offset = p->ast->bit_offset;
+ return res;
+}
+
static void desugar_many(HAllocator *mm__, HCFStack *stk__, void *env) {
// TODO: refactor this.
HRepeat *repeat = (HRepeat*)env;
@@ -93,7 +119,7 @@ static void desugar_many(HAllocator *mm__, HCFStack *stk__, void *env) {
HCFS_BEGIN_CHOICE() { // Mar
HCFS_BEGIN_SEQ() {
if (repeat->sep != NULL) {
- HCFS_DESUGAR(h_ignore__m(mm__, repeat->sep));
+ HCFS_DESUGAR(repeat->sep);
}
//stk__->last_completed->reshape = h_act_ignore; // BUG: This modifies a memoized entry.
HCFS_DESUGAR(repeat->p);
@@ -108,7 +134,7 @@ static void desugar_many(HAllocator *mm__, HCFStack *stk__, void *env) {
//HCFS_DESUGAR(h_ignore__m(mm__, h_epsilon_p()));
} HCFS_END_SEQ();
}
- HCFS_THIS_CHOICE->reshape = h_act_flatten;
+ HCFS_THIS_CHOICE->reshape = reshape_many;
} HCFS_END_CHOICE();
}
diff --git a/src/t_regression.c b/src/t_regression.c
index 1eeaf8c2bed340fd103c696b1e386d529647d1ed..4097fb6cf1a64392438e9dddb4ac17427e5e0065 100644
--- a/src/t_regression.c
+++ b/src/t_regression.c
@@ -118,9 +118,33 @@ static void test_llk_zero_end(void) {
g_check_parse_failed(aze, be, "a", 1);
}
+static void test_lalr_charset_lhs(void) {
+ HParserBackend be = PB_LALR;
+
+ HParser *p = h_choice(h_ch('A'), h_uint8(), NULL);
+
+ // the above would fail to compile because of an unhandled case in trying
+ // to resolve a conflict where an item's left-hand-side was an HCF_CHARSET.
+
+ g_check_parse_match(p, be, "A",1, "u0x41");
+ g_check_parse_match(p, be, "B",1, "u0x42");
+}
+
+static void test_cfg_many_seq(void) {
+ HParser *p = h_many(h_sequence(h_ch('A'), h_ch('B'), NULL));
+
+ g_check_parse_match(p, PB_LLk, "ABAB",4, "((u0x41 u0x42) (u0x41 u0x42))");
+ g_check_parse_match(p, PB_LALR, "ABAB",4, "((u0x41 u0x42) (u0x41 u0x42))");
+ g_check_parse_match(p, PB_GLR, "ABAB",4, "((u0x41 u0x42) (u0x41 u0x42))");
+ // these would instead parse as (u0x41 u0x42 u0x41 u0x42) due to a faulty
+ // reshape on h_many.
+}
+
void register_regression_tests(void) {
g_test_add_func("/core/regression/bug118", test_bug118);
g_test_add_func("/core/regression/seq_index_path", test_seq_index_path);
g_test_add_func("/core/regression/read_bits_48", test_read_bits_48);
g_test_add_func("/core/regression/llk_zero_end", test_llk_zero_end);
+ g_test_add_func("/core/regression/lalr_charset_lhs", test_lalr_charset_lhs);
+ g_test_add_func("/core/regression/cfg_many_seq", test_cfg_many_seq);
}