Merge pull request #33 from fastly/sv/cdata-codegen-updates-part-3

CDATA codegen updates, part 3

Author: silentbicycle

fuzz/target.c

@@ -508,8 +508,6 @@ fuzz_eager_output(const uint8_t *data, size_t size)
 		}
 	}
 
-	enum re_is_anchored_res anchorage[MAX_PATTERNS] = {0};
-
 	/* for each pattern, attempt to compile to a DFA */
 	for (size_t p_i = 0; p_i < env.pattern_count; p_i++) {
 		const char *p = env.patterns[p_i];
@@ -528,14 +526,9 @@ fuzz_eager_output(const uint8_t *data, size_t size)
 			continue; /* invalid regex */
 		}
 
-		const fsm_output_id_t endid = (fsm_output_id_t)p_i;
-		ret = fsm_seteageroutputonends(fsm, endid);
-		assert(ret == 1);
-
 		if (verbose) {
 			fprintf(stderr, "==== pattern %zd, pre det\n", p_i);
 			fsm_dump(stderr, fsm);
-			fsm_eager_output_dump(stderr, fsm);
 			fprintf(stderr, "====\n");
 
 			fsm_state_t c = fsm_countstates(fsm);
@@ -544,12 +537,6 @@ fuzz_eager_output(const uint8_t *data, size_t size)
 			}
 		}
 
-		ret = fsm_determinise(fsm);
-		assert(ret == 1);
-
-		ret = fsm_minimise(fsm);
-		assert(ret == 1);
-
 		fsm_state_t start;
 		if (!fsm_getstart(fsm, &start)) {
 			fsm_free(fsm);
@@ -578,7 +565,7 @@ fuzz_eager_output(const uint8_t *data, size_t size)
 	/* copy and combine fsms into one DFA */
 	{
 		size_t used = 0;
-		struct fsm_union_entry entries[MAX_PATTERNS] = {0};
+		struct fsm *nfas[MAX_PATTERNS] = {0};
 
 		for (size_t i = 0; i < env.fsm_count; i++) {
 			/* there can be gaps, fsms[] lines up with patterns[] */
@@ -604,9 +591,7 @@ fuzz_eager_output(const uint8_t *data, size_t size)
 				}
 			}
 
-			entries[used].fsm = cp;
-			entries[used].anchored_start = anchorage[i] & RE_IS_ANCHORED_START;
-			entries[used].anchored_end = anchorage[i] & RE_IS_ANCHORED_END;
+			nfas[used] = cp;
 			used++;
 		}
 
@@ -615,7 +600,7 @@ fuzz_eager_output(const uint8_t *data, size_t size)
 		}
 
 		/* consumes entries[] */
-		struct fsm *fsm = fsm_union_repeated_pattern_group(used, entries, NULL);
+		struct fsm *fsm = fsm_union_repeated_pattern_group(used, nfas, NULL, 0);
 		assert(fsm != NULL);
 
 		if (verbose) {

include/fsm/bool.h

@@ -52,15 +52,22 @@ struct fsm *
 fsm_union_array(size_t fsm_count,
     struct fsm **fsms, struct fsm_combined_base_pair *bases);
 
-struct fsm_union_entry {
-	struct fsm *fsm;
-	bool anchored_start;
-	bool anchored_end;
-};
-
+/* Combine an array of NFAs into a single NFA that attempts to match them
+ * all in one pass, with an extra loop so that more than one pattern with
+ * eager outputs can match. Ownership of the NFAs is transferred, they will
+ * be combined (or freed, if they don't have a start state).
+ *
+ * This MUST be called with NFAs constructed via re_comp, Calling it with
+ * manually constructed NFAs or DFAs is unsupported.
+ *
+ * This will set end IDs and/or output IDs representing matching each
+ * of the original NFAs on the combined result, where nfas[i] will
+ * get ID of (id_base + i).
+ *
+ * Returns NULL on error. */
 struct fsm *
-fsm_union_repeated_pattern_group(size_t entry_count,
-    struct fsm_union_entry *entries, struct fsm_combined_base_pair *bases);
+fsm_union_repeated_pattern_group(size_t nfa_count,
+    struct fsm **nfas, struct fsm_combined_base_pair *bases, size_t id_base);
 
 struct fsm *
 fsm_intersect(struct fsm *a, struct fsm *b);

src/libfsm/closure.c

@@ -128,7 +128,7 @@ epsilon_closure_single(const struct fsm *fsm, struct state_set **closures, fsm_s
 }
 
 struct state_set **
-epsilon_closure(struct fsm *fsm)
+epsilon_closure(const struct fsm *fsm)
 {
 	struct state_set **closures;
 	fsm_state_t s;
@@ -190,7 +190,7 @@ epsilon_closure(struct fsm *fsm)
 }
 
 void
-closure_free(struct fsm *fsm, struct state_set **closures, size_t n)
+closure_free(const struct fsm *fsm, struct state_set **closures, size_t n)
 {
 	fsm_state_t s;
 

src/libfsm/determinise.c

@@ -240,6 +240,7 @@ fsm_determinise_with_config(struct fsm *nfa,
 			assert(dfa->states[m->dfastate].edges == NULL);
 
 			dfa->states[m->dfastate].edges = m->edges;
+			m->edges = NULL; /* transfer ownership */
 
 			/*
 			 * The current DFA state is an end state if any of its associated NFA
@@ -616,6 +617,8 @@ map_free(struct map *map)
 		if (b == NULL) {
 			continue;
 		}
+		/* free any edge sets that didn't get transferred */
+		edge_set_free(map->alloc, b->edges);
 		f_free(map->alloc, b);
 	}
 

src/libfsm/eager_output.c

@@ -408,7 +408,10 @@ fsm_eager_output_compact(struct fsm *fsm, fsm_state_t *mapping, size_t mapping_c
 
 		assert(ob->state < mapping_count);
 		const fsm_state_t nstate = mapping[ob->state];
-		if (nstate == FSM_STATE_REMAP_NO_STATE) { continue; }
+		if (nstate == FSM_STATE_REMAP_NO_STATE) {
+			f_free(fsm->alloc, ob->entry);
+			continue;
+		}
 
 		const uint64_t hash = hash_id(nstate);
 

src/libfsm/internal.h

@@ -94,10 +94,10 @@ state_hasnondeterminism(const struct fsm *fsm, fsm_state_t state, struct bm *bm)
  * for states, with wrapper to populate malloced array of user-facing structs.
  */
 struct state_set **
-epsilon_closure(struct fsm *fsm);
+epsilon_closure(const struct fsm *fsm);
 
 void
-closure_free(struct fsm *fsm, struct state_set **closures, size_t n);
+closure_free(const struct fsm *fsm, struct state_set **closures, size_t n);
 
 /*
  * Internal free function that invokes free(3) by default, or a user-provided