diff --git a/src/backends/glr.c b/src/backends/glr.c
index 9828c2b30fc2a53eab1d2de5fd77b6f0893b6065..5619b26a4558b7ffbfe5b0ac33689e4625848cc2 100644
--- a/src/backends/glr.c
+++ b/src/backends/glr.c
@@ -174,9 +174,9 @@ static bool glr_step(HParseResult **result, HSlist *engines,
HSlistNode *x;
for(x=engines->head; x; x=x->next) {
HLREngine *eng = x->elem;
- if(eng->state == engine->state) {
- x->elem = lrengine_merge(eng, engine);
- break;
+ if(eng->state == engine->state && eng->input.index == engine->input.index) {
+ x->elem = lrengine_merge(eng, engine);
+ break;
}
}
if(!x) // no merge happened
diff --git a/src/t_regression.c b/src/t_regression.c
index 2c28b99efe6a36e69e5831044dadcbcc381f4d18..3bf763600962bdcdad6208863ba27f92966195f3 100644
--- a/src/t_regression.c
+++ b/src/t_regression.c
@@ -1,5 +1,6 @@
#include <glib.h>
#include <stdint.h>
+#include <stdlib.h>
#include "glue.h"
#include "hammer.h"
#include "test_suite.h"
@@ -472,6 +473,141 @@ static void test_issue83() {
g_check_cmp_int(r, ==, 0);
}
+/*
+ * This is Meg's cut-down bug 60 test case
+ */
+
+static void test_bug60() {
+ /* There is probably an even smaller example that shows the issue */
+
+ HParser *zed = NULL;
+ HParser *alpha = NULL;
+ HParser *vchar = NULL;
+ HParser *why = NULL;
+ HParser *plural_zed = NULL;
+ HParser *plural_zed_zed = NULL;
+ HParser *a_to_zed = NULL;
+ HParser *alphas = NULL;
+ HParser *rule = NULL;
+ HParser *rulelist = NULL;
+ HParser *p = NULL;
+ HParseResult *r = NULL;
+ int n;
+
+ zed = h_ch('z');
+
+ vchar = h_ch_range(0x79, 0x7a); /* allows y and z */
+
+ alpha = h_ch('a');
+
+ why = h_ch('y');
+
+ plural_zed = h_sequence(
+ why,
+ h_many(h_choice(alpha, vchar, NULL)),
+ NULL);
+ plural_zed_zed = h_choice(plural_zed, zed, NULL);
+ alphas = h_choice(alpha, h_sequence(plural_zed_zed, alpha, NULL), NULL);
+
+ a_to_zed = h_sequence(
+ zed,
+ h_many(h_sequence(h_many1(alphas), zed, NULL)),
+ NULL);
+ rule = h_sequence(a_to_zed, plural_zed_zed, NULL);
+ rulelist = h_many1(h_choice(
+ rule,
+ h_sequence(h_many(alphas), plural_zed_zed, NULL),
+ NULL));
+
+ p = rulelist;
+
+ g_check_parse_ok(p, PB_GLR, "ayzza", 5);
+ g_check_parse_match(p, PB_GLR, "ayzza", 5, "(((u0x61) (u0x79 (u0x7a u0x7a u0x61))))");
+
+}
+
+/*
+ * This is the original bug60 test case; cut down from an ABNF parser
+ */
+
+#define BUG60_ABNF_SCAN_UP_TO 64
+
+static void test_bug60_abnf() {
+ HParser *newline = NULL;
+ HParser *alpha = NULL;
+ HParser *sp = NULL;
+ HParser *vchar = NULL;
+ HParser *equal = NULL;
+ HParser *semicolon = NULL;
+ HParser *comment = NULL;
+ HParser *c_nl = NULL;
+ HParser *c_sp = NULL;
+ HParser *defined_as = NULL;
+ HParser *alphas = NULL;
+ HParser *rule = NULL;
+ HParser *rulelist = NULL;
+ HParser *p = NULL;
+ int i, j, r, s_size;
+ char *s = NULL;
+ const char *test_string_template = "x = y z%s;%s\n\n";
+ char buf_1[BUG60_ABNF_SCAN_UP_TO+1];
+ char buf_2[2*BUG60_ABNF_SCAN_UP_TO+1];
+
+ newline = h_ch('\n');
+ alpha = h_choice(h_ch_range('A', 'Z'), h_ch_range('a', 'z'), NULL);
+ sp = h_ch(' ');
+ vchar = h_ch_range(0x21, 0x7e);
+ equal = h_ch('=');
+ semicolon = h_ch(';');
+ comment = h_sequence(
+ semicolon,
+ h_many(h_choice(sp, vchar, NULL)),
+ newline,
+ NULL);
+ c_nl = h_choice(comment, newline, NULL);
+ c_sp = h_choice(sp, h_sequence(c_nl, sp, NULL), NULL);
+ defined_as = h_sequence(h_many(c_sp), equal, h_many(c_sp), NULL);
+ alphas = h_sequence(
+ alpha,
+ h_many(h_sequence(h_many1(c_sp), alpha, NULL)),
+ h_many(c_sp),
+ NULL);
+ rule = h_sequence(alpha, defined_as, alphas, c_nl, NULL);
+ rulelist = h_many1(h_choice(
+ rule,
+ h_sequence(h_many(c_sp), c_nl, NULL),
+ NULL));
+
+ p = rulelist;
+ g_check_compilable(p, PB_GLR, 1);
+
+ /* Have a buffer for the string */
+ s_size = strlen(test_string_template) + 3*BUG60_ABNF_SCAN_UP_TO + 1;
+ s = malloc(s_size);
+ g_check_cmp_ptr(s, !=, NULL);
+
+ /*
+ * Try to parse all the different strings according to the template up to
+ * the scan limit.
+ *
+ * Correct behavior: it parses for all values of i, j.
+ * Bugged behavior: when i % 3 != 0, parse failures begin to occur at
+ * j == (i / 3) + (i % 3).
+ */
+ for (i = 0; i < BUG60_ABNF_SCAN_UP_TO; ++i) {
+ memset(buf_1, ' ', i);
+ buf_1[i] = 0;
+ for (j = 0; j < 2*BUG60_ABNF_SCAN_UP_TO; ++j) {
+ memset(buf_2, 'x', j);
+ buf_2[j] = 0;
+ snprintf(s, s_size, test_string_template, buf_1, buf_2);
+ g_check_parse_ok_no_compile(p, s, strlen(s));
+ }
+ }
+
+ free(s);
+}
+
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);
@@ -488,4 +624,6 @@ void register_regression_tests(void) {
g_test_add_func("/core/regression/issue87", test_issue87);
g_test_add_func("/core/regression/issue92", test_issue92);
g_test_add_func("/core/regression/issue83", test_issue83);
+ g_test_add_func("/core/regression/bug60", test_bug60);
+ g_test_add_func("/core/regression/bug60_abnf", test_bug60_abnf);
}
diff --git a/src/test_suite.h b/src/test_suite.h
index 6f95e077650e86d4c5ef1e0f01f2ee2b20ab31d2..775c8818a1f79cb2fb50275a55c0e8e1f8a235fe 100644
--- a/src/test_suite.h
+++ b/src/test_suite.h
@@ -109,24 +109,36 @@
} while(0)
#define g_check_compilable(lang, backend, params) do { \
- if (!h_compile(lang, backend, params)) { \
- g_test_message("Language is not %s(%s)", #backend, params); \
+ int r = h_compile((HParser *)(lang), (HParserBackend)(backend), (void *)(params)); \
+ if (r != 0) { \
+ g_test_message("Language is not %s(%s)", #backend, #params); \
g_test_fail(); \
} \
} while(0)
-
+#define print_arena_stats(arena) do { \
+ if (g_test_verbose()) { \
+ HArenaStats stats; \
+ h_allocator_stats(arena, &stats); \
+ g_test_message("Parse used %zd bytes, wasted %zd bytes. " \
+ "Inefficiency: %5f%%", \
+ stats.used, stats.wasted, \
+ stats.wasted * 100. / (stats.used+stats.wasted)); \
+ } \
+ } while(0)
+
#define g_check_parse_failed__m(mm__, parser, backend, input, inp_len) do { \
int skip = h_compile__m(mm__, (HParser *)(parser), (HParserBackend)backend, NULL); \
if(skip != 0) { \
- g_test_message("Compile failed"); \
+ g_test_message("Compile failed on line %d", __LINE__); \
g_test_fail(); \
break; \
} \
- HParseResult *result = h_parse__m(mm__, parser, (const uint8_t*)input, inp_len); \
- if (NULL != result) { \
- h_parse_result_free(result); \
- g_test_message("Check failed: shouldn't have succeeded, but did"); \
+ HParseResult *res = h_parse__m(mm__, parser, (const uint8_t*)input, inp_len); \
+ if (NULL != res) { \
+ print_arena_stats(res->arena); \
+ h_parse_result_free(res); \
+ g_test_message("Check failed: parse shouldn't have succeeded, but did on line %d", __LINE__); \
g_test_fail(); \
} \
} while(0)
@@ -134,25 +146,27 @@
#define g_check_parse_failed(p, be, input, len) \
g_check_parse_failed__m(&system_allocator, p, be, input, len)
-#define print_arena_stats(arena) do { \
- if (g_test_verbose()) { \
- HArenaStats stats; \
- h_allocator_stats(arena, &stats); \
- g_test_message("Parse used %zd bytes, wasted %zd bytes. " \
- "Inefficiency: %5f%%", \
- stats.used, stats.wasted, \
- stats.wasted * 100. / (stats.used+stats.wasted)); \
+#define g_check_parse_failed_no_compile__m(mm__, parser, input, inp_len) do { \
+ HParseResult *res = h_parse__m(mm__, parser, (const uint8_t*)input, inp_len); \
+ if (NULL != res) { \
+ print_arena_stats(res->arena); \
+ h_parse_result_free(res); \
+ g_test_message("Check failed: shouldn't have succeeded, but did on ;ine %d", __LINE__); \
+ g_test_fail(); \
} \
} while(0)
-#define g_check_parse_ok(parser, backend, input, inp_len) do { \
- int skip = h_compile((HParser *)(parser), (HParserBackend) backend, NULL); \
+#define g_check_parse_failed_no_compile(p, input, len) \
+ g_check_parse_failed__m(&system_allocator, p, input, len)
+
+#define g_check_parse_ok__m(mm__, parser, backend, input, inp_len) do { \
+ int skip = h_compile__m(mm__, (HParser *)(parser), (HParserBackend) backend, NULL); \
if(skip) { \
g_test_message("Compile failed"); \
g_test_fail(); \
break; \
} \
- HParseResult *res = h_parse(parser, (const uint8_t*)input, inp_len); \
+ HParseResult *res = h_parse__m(mm__, parser, (const uint8_t*)input, inp_len); \
if (!res) { \
g_test_message("Parse failed on line %d", __LINE__); \
g_test_fail(); \
@@ -162,14 +176,31 @@
} \
} while(0)
-#define g_check_parse_match(parser, backend, input, inp_len, result) do { \
- int skip = h_compile((HParser *)(parser), (HParserBackend) backend, NULL); \
+#define g_check_parse_ok(p, be, input, len) \
+ g_check_parse_ok__m(&system_allocator, p, be, input, len)
+
+#define g_check_parse_ok_no_compile__m(mm__, parser, input, inp_len) do { \
+ HParseResult *res = h_parse__m(mm__, parser, (const uint8_t*)input, inp_len); \
+ if (!res) { \
+ g_test_message("Parse failed on line %d", __LINE__); \
+ g_test_fail(); \
+ } else { \
+ print_arena_stats(res->arena); \
+ h_parse_result_free(res); \
+ } \
+ } while(0)
+
+#define g_check_parse_ok_no_compile(p, input, len) \
+ g_check_parse_ok_no_compile__m(&system_allocator, p, input, len)
+
+#define g_check_parse_match__m(mm__, parser, backend, input, inp_len, result) do { \
+ int skip = h_compile__m(mm__, (HParser *)(parser), (HParserBackend) backend, NULL); \
if(skip) { \
g_test_message("Compile failed"); \
g_test_fail(); \
break; \
} \
- HParseResult *res = h_parse(parser, (const uint8_t*)input, inp_len); \
+ HParseResult *res = h_parse__m(mm__, (HParser *)(parser), (const uint8_t*)input, inp_len); \
if (!res) { \
g_test_message("Parse failed on line %d", __LINE__); \
g_test_fail(); \
@@ -182,6 +213,26 @@
} \
} while(0)
+#define g_check_parse_match(parser, backend, input, inp_len, result) \
+ g_check_parse_match__m(&system_allocator, parser, backend, input, inp_len, result)
+
+#define g_check_parse_match_no_compile__m(mm__, parser, input, inp_len, result) do { \
+ HParseResult *res = h_parse__m(mm__, (HParser *)(parser), (const uint8_t*)input, inp_len); \
+ if (!res) { \
+ g_test_message("Parse failed on line %d", __LINE__); \
+ g_test_fail(); \
+ } else { \
+ char* cres = h_write_result_unamb(res->ast); \
+ g_check_string(cres, ==, result); \
+ (&system_allocator)->free(&system_allocator, cres); \
+ print_arena_stats(res->arena); \
+ h_parse_result_free(res); \
+ } \
+ } while(0)
+
+#define g_check_parse_match_no_compile(parser, input, inp_len, result) \
+ g_check_parse_match_no_compile__m(&system_allocator, parser, input, inp_len, result)
+
#define g_check_parse_chunks_failed__m(mm__, parser, backend, chunk1, c1_len, chunk2, c2_len) do { \
int skip = h_compile__m(mm__, (HParser *)(parser), (HParserBackend)backend, NULL); \
if(skip) { \
@@ -193,7 +244,7 @@
} while(0)
#define g_check_parse_chunks_failed___m(mm__, parser, chunk1, c1_len, chunk2, c2_len) do { \
- HSuspendedParser *s = h_parse_start__m(mm__, parser); \
+ HSuspendedParser *s = h_parse_start__m(mm__, (HParser *)(parser)); \
if(!s) { \
g_test_message("Chunk-wise parsing not available"); \
g_test_fail(); \
@@ -226,7 +277,7 @@
} while(0)
#define g_check_parse_chunks_ok_(parser, chunk1, c1_len, chunk2, c2_len) do { \
- HSuspendedParser *s = h_parse_start(parser); \
+ HSuspendedParser *s = h_parse_start((HParser *)(parser)); \
if(!s) { \
g_test_message("Chunk-wise parsing not available"); \
g_test_fail(); \
@@ -255,7 +306,7 @@
} while(0)
#define g_check_parse_chunks_match_(parser, chunk1, c1_len, chunk2, c2_len, result) do { \
- HSuspendedParser *s = h_parse_start(parser); \
+ HSuspendedParser *s = h_parse_start((HParser *)(parser)); \
if(!s) { \
g_test_message("Chunk-wise parsing not available"); \
g_test_fail(); \