Merge branch 'sv/determine-which-characters-must-appear-in-input-to-m…

…atch' into sv/tmp-integration-branch-to-vendor-for-da2lx-subside-development

Makefile

@@ -116,6 +116,7 @@ SUBDIR += tests/intersect
 SUBDIR += tests/eclosure
 SUBDIR += tests/equals
 SUBDIR += tests/subtract
+SUBDIR += tests/detect_required
 SUBDIR += tests/determinise
 SUBDIR += tests/eager_output
 SUBDIR += tests/endids

include/adt/bitmap.h

@@ -8,6 +8,9 @@
 #define ADT_BITMAP_H
 
 #include <stdint.h>
+#include <limits.h>
+
+#include <stdio.h>
 #include "print/esc.h"
 
 struct fsm_state;
@@ -23,6 +26,9 @@ bm_get(const struct bm *bm, size_t i);
 void
 bm_set(struct bm *bm, size_t i);
 
+void
+bm_unset(struct bm *bm, size_t i);
+
 /* Get a writeable pointer to the Nth word of the char set bitmap,
  * or NULL if out of bounds. */
 uint64_t *
@@ -51,5 +57,17 @@ bm_snprint(const struct bm *bm, const struct fsm_options *opt,
 	int boxed,
 	escputc *escputc);
 
+void
+bm_copy(struct bm *dst, const struct bm *src);
+
+void
+bm_intersect(struct bm *dst, const struct bm *src);
+
+void
+bm_union(struct bm *dst, const struct bm *src);
+
+int
+bm_any(const struct bm *bm);
+
 #endif
 

include/fsm/walk.h

@@ -7,6 +7,8 @@
 #ifndef FSM_WALK_H
 #define FSM_WALK_H
 
+#include <adt/bitmap.h>
+
 struct fsm;
 struct fsm_state;
 
@@ -128,5 +130,34 @@ fsm_generate_matches_cb fsm_generate_cb_printf;
  * to escape all characters or just nonprintable ones. */
 fsm_generate_matches_cb fsm_generate_cb_printf_escaped;
 
+/* Walk a DFA and detect which characters MUST appear in the input for a
+ * match to be possible. For example, if input for the DFA corresponding
+ * to /^(abc|dbe)$/ does not contain 'b' at all, there's no way it can
+ * ever match, so executing the regex is unnecessary. This does not detect
+ * which characters must appear before/after others or how many times, just
+ * which must be present.
+ *
+ * The input must be a DFA. When run with EXPENSIVE_CHECKS this will
+ * check and return ERROR_MISUSE if it is not, otherwise this is an
+ * unchecked error.
+ *
+ * The character map will be cleared before populating. If *count is
+ * non-NULL it will be updated with how many required characters were
+ * found.
+ *
+ * There is an optional step_limit -- if this is reached, then it will
+ * return FSM_DETECT_REQUIRED_CHARACTERS_STEP_LIMIT_REACHED and a
+ * cleared bitmap, because any partial information could still have been
+ * contradicted later. If the step_limit is 0 it will be ignored. */
+enum fsm_detect_required_characters_res {
+	FSM_DETECT_REQUIRED_CHARACTERS_WRITTEN,
+	FSM_DETECT_REQUIRED_CHARACTERS_STEP_LIMIT_REACHED,
+	FSM_DETECT_REQUIRED_CHARACTERS_ERROR_MISUSE = -1,
+	FSM_DETECT_REQUIRED_CHARACTERS_ERROR_ALLOC = -2,
+};
+enum fsm_detect_required_characters_res
+fsm_detect_required_characters(const struct fsm *dfa, size_t step_limit,
+    uint64_t charmap[4], size_t *count);
+
 #endif
 

man/fsm.1/fsm.1.xml

@@ -33,6 +33,7 @@
 	<!ENTITY G.opt "<option>-G</option>&nbsp;&length.arg;">
 	<!ENTITY k.opt "<option>-k</option>&nbsp;&io.arg;">
 	<!ENTITY i.opt "<option>-i</option>&nbsp;&iterations.arg;">
+	<!ENTITY G.opt "<option>-S</option>&nbsp;&limit.arg;">
 	<!ENTITY U.opt "<option>-U</option>&nbsp;&charset.arg;">
 	<!ENTITY X.opt "<option>-X</option>">
 
@@ -325,6 +326,14 @@
 				</listitem>
 			</varlistentry>
 
+			<varlistentry>
+				<term>&S.opt;</term>
+
+				<listitem>
+					<para>Set a step limit for long-running operations.</para>
+				</listitem>
+			</varlistentry>
+
 			<varlistentry>
 				<term>&t.opt;</term>
 
@@ -487,6 +496,13 @@
 									of each state in the &fsm;.
 									Printed to &stdout.lit;; exit status is always true.</td>
 							</tr>
+							<tr>
+								<td><code>requiredchars</code></td>
+								<td rowspan="1" role="na">&ndash;</td>
+								<td rowspan="1">Determine characters that must appear in any
+									inputs that could match the &fsm;. Exit status is true
+									unless the step limit was reached.</td>
+							</tr>
 						</tbody>
 					</table>
 				</listitem>

src/adt/bitmap.c

@@ -10,6 +10,7 @@
 #include <limits.h>
 #include <ctype.h>
 #include <stdint.h>
+#include <string.h>
 
 #include <adt/bitmap.h>
 #include <adt/u64bitset.h>
@@ -34,6 +35,15 @@ bm_set(struct bm *bm, size_t i)
 	u64bitset_set(bm->map, i);
 }
 
+void
+bm_unset(struct bm *bm, size_t i)
+{
+	assert(bm != NULL);
+	assert(i <= UCHAR_MAX);
+
+	u64bitset_clear(bm->map, i);
+}
+
 uint64_t *
 bm_nth_word(struct bm *bm, size_t n)
 {
@@ -325,3 +335,34 @@ bm_snprint(const struct bm *bm, const struct fsm_options *opt,
 
 	return -1;
 }
+
+void
+bm_copy(struct bm *dst, const struct bm *src)
+{
+	memcpy(dst, src, sizeof(*src));
+}
+
+void
+bm_intersect(struct bm *dst, const struct bm *src)
+{
+	for (size_t i = 0; i < sizeof(src->map)/sizeof(src->map[0]); i++) {
+		dst->map[i] &= src->map[i];
+	}
+}
+
+void
+bm_union(struct bm *dst, const struct bm *src)
+{
+	for (size_t i = 0; i < sizeof(src->map)/sizeof(src->map[0]); i++) {
+		dst->map[i] |= src->map[i];
+	}
+}
+
+int
+bm_any(const struct bm *bm)
+{
+	for (size_t i = 0; i < sizeof(bm->map)/sizeof(bm->map[0]); i++) {
+		if (bm->map[i]) { return 1; }
+	}
+	return 0;
+}

src/fsm/main.c

@@ -25,6 +25,7 @@
 #include <fsm/parser.h>
 
 #include <adt/stateset.h> /* XXX */
+#include <adt/u64bitset.h>
 
 #include "libfsm/internal.h" /* XXX */
 
@@ -101,6 +102,16 @@ query_epsilonclosure(const struct fsm *fsm, fsm_state_t state)
 	abort();
 }
 
+static int
+query_required_chars(const struct fsm *fsm, fsm_state_t state)
+{
+	(void) fsm;
+	(void) state;
+
+	/* never called */
+	abort();
+}
+
 static void
 usage(void)
 {
@@ -227,7 +238,9 @@ static int
 		{ "hasambiguity",      fsm_has, fsm_hasnondeterminism },
 		{ "hasnondeterminism", fsm_has, fsm_hasnondeterminism },
 		{ "hasepsilons",       fsm_has, fsm_hasepsilons       },
-		{ "epsilons",          fsm_has, fsm_hasepsilons       }
+		{ "epsilons",          fsm_has, fsm_hasepsilons       },
+		{ "requiredchars",     NULL,    query_required_chars  },
+		{ "chars",             NULL,    query_required_chars  },
 	};
 
 	assert(name != NULL);
@@ -378,6 +391,7 @@ main(int argc, char *argv[])
 	int xfiles;
 	int r;
 	size_t generate_bounds = 0;
+	size_t step_limit = 0;
 
 	int (*query)(const struct fsm *, fsm_state_t);
 	int (*walk )(const struct fsm *,
@@ -404,7 +418,7 @@ main(int argc, char *argv[])
 	{
 		int c;
 
-		while (c = getopt(argc, argv, "h" "aCcgwXe:k:i:" "xpq:l:dG:mrt:EU:W:"), c != -1) {
+		while (c = getopt(argc, argv, "h" "aCcgwXe:k:i:" "xpq:l:dG:mrt:ES:U:W:"), c != -1) {
 			switch (c) {
 			case 'a': opt.anonymous_states  = 1;          break;
 			case 'c': opt.consolidate_edges = 1;          break;
@@ -451,6 +465,10 @@ main(int argc, char *argv[])
 				}
 				break;
 
+			case 'S':
+				step_limit = strtoul(optarg, NULL, 10);
+				break; /* can be 0 */
+
 			case 'h':
 				usage();
 				exit(EXIT_SUCCESS);
@@ -669,6 +687,34 @@ main(int argc, char *argv[])
 			closure_free(fsm, closures, fsm->statecount);
 
 			return 0;
+		} else if (query == query_required_chars) {
+			assert(walk == NULL);
+			uint64_t charmap[4];
+			size_t count;
+			enum fsm_detect_required_characters_res res;
+			res = fsm_detect_required_characters(fsm, step_limit, charmap, &count);
+			if (res == FSM_DETECT_REQUIRED_CHARACTERS_STEP_LIMIT_REACHED) {
+				fprintf(stderr, "fsm_detect_required_characters: step limit reached (%zd)\n", step_limit);
+				exit(EXIT_FAILURE);
+			} else {
+				assert(res == FSM_DETECT_REQUIRED_CHARACTERS_WRITTEN);
+				char buf[257] = {0};
+				size_t used = 0;
+				for (size_t i = 0; i < 256; i++) {
+					if (u64bitset_get(charmap, i)) {
+						buf[used++] = (char)i;
+					}
+				}
+				printf("%zd ", count);
+				for (size_t i = 0; i < used; i++) {
+					c_escputc_str(stdout, &opt, buf[i]);
+				}
+				printf("\n");
+
+				fsm_free(fsm);
+				fsm_to_cleanup = NULL;
+				return EXIT_SUCCESS;
+			}
 		} else {
 			assert(walk != NULL);
 			r |= !walk(fsm, query);

src/libfsm/Makefile

@@ -7,6 +7,7 @@ SRC += src/libfsm/complete.c
 SRC += src/libfsm/consolidate.c
 SRC += src/libfsm/clone.c
 SRC += src/libfsm/closure.c
+SRC += src/libfsm/detect_required.c
 SRC += src/libfsm/eager_output.c
 SRC += src/libfsm/edge.c
 SRC += src/libfsm/empty.c