Merge pull request katef#479 from katef/kate/endleaf-rejiggle

Endleaf rejiggle, .fsm syntax for endids, refactor endid get/set api, various related bugfixes

Makefile

@@ -117,6 +117,7 @@ SUBDIR += tests/subtract
 SUBDIR += tests/determinise
 SUBDIR += tests/endids
 SUBDIR += tests/epsilons
+SUBDIR += tests/fsm
 SUBDIR += tests/glob
 SUBDIR += tests/like
 SUBDIR += tests/literal

include/fsm/fsm.h

@@ -37,12 +37,6 @@ typedef unsigned int fsm_end_id_t;
 
 #define FSM_END_ID_MAX UINT_MAX
 
-/* struct used to return a collection of end IDs. */
-struct fsm_end_ids {
-	unsigned count;
-	fsm_end_id_t ids[1];
-};
-
 /*
  * Create a new FSM. This is to be freed with fsm_free(). A structure allocated
  * from fsm_new() is expected to be passed as the "fsm" argument to the
@@ -222,29 +216,31 @@ fsm_setendid(struct fsm *fsm, fsm_end_id_t id);
  * Returns 1 on success, 0 on error.
  * */
 int
-fsm_setendidstate(struct fsm *fsm, fsm_state_t end_state, fsm_end_id_t id);
+fsm_endid_set(struct fsm *fsm, fsm_state_t end_state, fsm_end_id_t id);
 
 /* Get the end IDs associated with an end state, if any.
- * If id_buf has enough cells to store all the end IDs (according
- * to id_buf_count) then they are written into id_buf[] and
- * *ids_written is set to the number of IDs. The end IDs in the
- * buffer may appear in any order, but should not have duplicates.
+ * id_buf is expected to have enough cells (according to id_buf_count)
+ * to store all the end IDs. You can find this with fsm_endid_count().
+ *
+ * The end IDs in the buffer may appear in any order,
+ * but will not have duplicates.
+ *
+ * A state with no end IDs set is considered equivalent to a state
+ * that has the empty set, this API does not distinguish these cases.
+ * This is not an error.
+ *
+ * It is an error to attempt to get end IDs associated with a state
+ * that is not marked as an end state.
  *
  * Returns 0 if there is not enough space in id_buf for the
  * end IDs, or 1 if zero or more end IDs were returned. */
-enum fsm_getendids_res {
-	FSM_GETENDIDS_NOT_FOUND,
-	FSM_GETENDIDS_FOUND,
-	FSM_GETENDIDS_ERROR_INSUFFICIENT_SPACE = -1
-};
-enum fsm_getendids_res
-fsm_getendids(const struct fsm *fsm, fsm_state_t end_state,
-    size_t id_buf_count, fsm_end_id_t *id_buf,
-    size_t *ids_written);
+int
+fsm_endid_get(const struct fsm *fsm, fsm_state_t end_state,
+    size_t id_buf_count, fsm_end_id_t *id_buf);
 
 /* Get the number of end IDs associated with an end state. */
 size_t
-fsm_getendidcount(const struct fsm *fsm, fsm_state_t end_state);
+fsm_endid_count(const struct fsm *fsm, fsm_state_t end_state);
 
 /* Callback function to remap the end ids of a state.  This function can
  * remap to fewer end ids, but cannot add additional end ids, and cannot

include/fsm/options.h

@@ -65,12 +65,15 @@ struct fsm_options {
 	const char *cp;
 
 	/* TODO: explain. for C code fragment output */
-	int (*leaf)(FILE *, const struct fsm_end_ids *ids,
+	int (*leaf)(FILE *, const fsm_end_id_t *ids, size_t count,
 	    const void *leaf_opaque);
 	void *leaf_opaque;
 
 	/* TODO: explain. for C code fragment output */
-	int (*endleaf)(FILE *, const struct fsm_end_ids *ids,
+	/* Placement in the output stream depends on the format.
+	 * This replaces an entire "return xyz;" statement for C-like formats,
+	 * but appends extra information for others. */
+	int (*endleaf)(FILE *, const fsm_end_id_t *ids, size_t count,
 	    const void *endleaf_opaque);
 	void *endleaf_opaque;
 

src/fsm/main.c

@@ -477,17 +477,14 @@ main(int argc, char *argv[])
 		struct fsm *q;
 
 		if ((op & OP_ARITY) == 1) {
-			if (argc > 1) {
-				usage();
-				exit(EXIT_FAILURE);
-			}
+			/* argc < 1 is okay */
 
 			q = fsm_parse((argc == 0) ? stdin : xopen(argv[0]), &opt);
 			if (q == NULL) {
 				exit(EXIT_FAILURE);
 			}
 		} else {
-			if (argc != 2) {
+			if (argc < 2) {
 				usage();
 				exit(EXIT_FAILURE);
 			}
@@ -610,6 +607,17 @@ main(int argc, char *argv[])
 		printf("=> total %g ms (avg %g ms)\n", elapsed, elapsed / iterations);
 	}
 
+	/* we're done consuming filenames, remaining argv is text to match */
+	if ((op & OP_ARITY) == 1) {
+		if (argc > 0) {
+			argc -= 1;
+			argv += 1;
+		}
+	} else {
+		argc -= 2;
+		argv += 2;
+	}
+
 	/* henceforth, r is $?-convention (0 for success) */
 
 	if (fsm == NULL) {
@@ -661,14 +669,15 @@ main(int argc, char *argv[])
 		}
 	}
 
-	/* TODO: optional -- to delimit texts as opposed to .fsm filenames */
-	if (op == OP_IDENTITY && argc > 0) {
+	/* match text */
+	if (argc > 0) {
 		int i;
 
 		/* TODO: option to print input texts which match. like grep(1) does.
 		 * This is not the same as printing patterns which match (by associating
 		 * a pattern to the end state), like lx(1) does */
 
+		/* TODO: optional -- to delimit texts as opposed to .fsm filenames */
 		for (i = 0; i < argc; i++) {
 			fsm_state_t state;
 			int e;
@@ -694,7 +703,7 @@ main(int argc, char *argv[])
 				continue;
 			}
 
-			/* TODO: option to print state number? */
+			/* TODO: option to print matching end-ids */
 		}
 	}
 

src/libfsm/clone.c

@@ -127,15 +127,13 @@ static int
 copy_end_ids_cb(fsm_state_t state, const fsm_end_id_t id, void *opaque)
 {
 	struct copy_end_ids_env *env = opaque;
-	enum fsm_endid_set_res sres;
 	assert(env->tag == 'c');
 
 #if LOG_CLONE_ENDIDS
 	fprintf(stderr, "clone[%d] <- %d\n", state, id);
 #endif
 
-	sres = fsm_endid_set(env->dst, state, id);
-	if (sres == FSM_ENDID_SET_ERROR_ALLOC_FAIL) {
+	if (!fsm_endid_set(env->dst, state, id)) {
 		env->ok = 0;
 		return 0;
 	}

src/libfsm/consolidate.c

@@ -231,19 +231,14 @@ static int
 consolidate_end_ids_cb(fsm_state_t state, const fsm_end_id_t *ids, size_t num_ids, void *opaque)
 {
 	struct consolidate_end_ids_env *env = opaque;
-	enum fsm_endid_set_res sres;
 	fsm_state_t s;
 	assert(env->tag == 'C');
 
 	assert(state < env->mapping_count);
 	s = env->mapping[state];
 
-	sres = fsm_endid_set_bulk(env->dst, s, num_ids, ids, FSM_ENDID_BULK_APPEND);
-	if (sres == FSM_ENDID_SET_ERROR_ALLOC_FAIL) {
-		return 0;
-	}
-
-	return 1;
+	return fsm_endid_set_bulk(env->dst, s,
+		num_ids, ids, FSM_ENDID_BULK_APPEND);
 }
 
 static int