Add fsm_vacuum, which reduces the state array when over-allocated.

include/fsm/fsm.h

@@ -493,5 +493,14 @@ int fsm_fgetc(void *opaque); /* expects opaque to be FILE *  */
 int
 fsm_shuffle(struct fsm *fsm, unsigned seed);
 
+/* Attempt to reclaim memory if an FSM has significantly reduced its
+ * state count. Returns 1 if */
+enum fsm_vacuum_res {
+	FSM_VACUUM_NO_CHANGE,
+	FSM_VACUUM_REDUCED_MEMORY,
+	FSM_VACUUM_ERROC_REALLOC = -1,
+}
+fsm_vacuum(struct fsm *fsm);
+
 #endif
 

src/libfsm/Makefile

@@ -22,6 +22,7 @@ SRC += src/libfsm/trim.c
 SRC += src/libfsm/example.c
 SRC += src/libfsm/getc.c
 SRC += src/libfsm/shuffle.c
+SRC += src/libfsm/vacuum.c
 SRC += src/libfsm/vm.c
 
 # graph things

src/libfsm/fsm.c

@@ -57,7 +57,7 @@ fsm_new(const struct fsm_options *opt)
 		return NULL;
 	}
 
-	new->statealloc = 128; /* guess */
+	new->statealloc = FSM_DEFAULT_STATEALLOC;
 	new->statecount = 0;
 	new->endcount   = 0;
 	new->capture_info = NULL;

src/libfsm/internal.h

@@ -44,6 +44,9 @@ struct state_array;
 
 #define FSM_CAPTURE_MAX INT_MAX
 
+/* guess for default state allocation */
+#define FSM_DEFAULT_STATEALLOC 128
+
 struct fsm_edge {
 	fsm_state_t state; /* destination */
 	unsigned char symbol;

src/libfsm/libfsm.syms

@@ -68,6 +68,7 @@ fsm_addstate
 fsm_addstate_bulk
 fsm_removestate
 fsm_shuffle
+fsm_vacuum
 
 fsm_addedge_any
 fsm_addedge_epsilon

src/libfsm/vacuum.c

@@ -0,0 +1,40 @@
+/*
+ * Copyright 2024 Scott Vokes
+ *
+ * See LICENCE for the full copyright terms.
+ */
+
+#include <stdlib.h>
+#include <fsm/fsm.h>
+#include <fsm/alloc.h>
+
+#include <assert.h>
+
+#include "internal.h"
+
+enum fsm_vacuum_res
+fsm_vacuum(struct fsm *fsm)
+{
+	assert(fsm != NULL);
+	assert(fsm->statecount <= fsm->statealloc);
+	size_t nceil = fsm->statealloc;
+
+	while ((fsm->statecount < nceil/2) && nceil > 1) {
+		nceil /= 2;
+	}
+
+	if (nceil == fsm->statealloc) {
+		return FSM_VACUUM_NO_CHANGE;
+	}
+
+	struct fsm_state *nstates = f_realloc(fsm->opt->alloc,
+	    fsm->states, nceil * sizeof(nstates[0]));
+	if (nstates == NULL) {
+		return FSM_VACUUM_ERROC_REALLOC;
+	}
+
+	fsm->states = nstates;
+	fsm->statealloc = nceil;
+
+	return FSM_VACUUM_REDUCED_MEMORY;
+}