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

…atch--remove-struct-bm' into sv/tmp-integration-branch-to-vendor-for-da2lx-subside-go-or-no-go

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/endids
 SUBDIR += tests/epsilons

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 <stdint.h>
+
 struct fsm;
 struct fsm_state;
 
@@ -128,5 +130,30 @@ 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.
+ *
+ * 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 bitmap[4], size_t *char_count);
+
 #endif
 

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/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/edge.c
 SRC += src/libfsm/empty.c
 SRC += src/libfsm/end.c