From 748b7afc934164fa5b0e2e3aaf9ead198a708901 Mon Sep 17 00:00:00 2001
From: Kate F <kate@elide.org>
Date: Thu, 30 May 2024 15:27:51 +0100
Subject: [PATCH] Add fsm_new_statealloc(), to pre-allocate to a known number
 of states.

---
 include/fsm/fsm.h      |  9 +++++++++
 src/libfsm/charset.c   |  3 +--
 src/libfsm/clone.c     |  2 +-
 src/libfsm/fsm.c       | 14 ++++++++++++--
 src/libfsm/reverse.c   |  2 +-
 src/libre/re_strings.c |  1 +
 6 files changed, 25 insertions(+), 6 deletions(-)

diff --git a/include/fsm/fsm.h b/include/fsm/fsm.h
index 1904814bd..2fea117cd 100644
--- a/include/fsm/fsm.h
+++ b/include/fsm/fsm.h
@@ -60,6 +60,15 @@ struct fsm_end_ids {
 struct fsm *
 fsm_new(const struct fsm_options *opt);
 
+/*
+ * As fsm_new(), but with an explicit number of pre-allocated states.
+ * This is intended to save on internal reallocations for situations
+ * where it's known in advance exactly how many states an FSM will have.
+ * You can still add more states per usual.
+ */
+struct fsm *
+fsm_new_statealloc(const struct fsm_options *opt, size_t statealloc);
+
 /*
  * Free a structure created by fsm_new(), and all of its contents.
  * No other pointers returned by this API are to be freed individually.
diff --git a/src/libfsm/charset.c b/src/libfsm/charset.c
index dd8a679ed..662c0ac1f 100644
--- a/src/libfsm/charset.c
+++ b/src/libfsm/charset.c
@@ -36,8 +36,7 @@ fsm_intersect_charset(struct fsm *a, size_t n, const char *charset)
 	{
 		fsm_state_t state;
 
-		// TODO: pass .statealloc explicitly, we know it's 1. the default is overkill
-		b = fsm_new(a->opt);
+		b = fsm_new_statealloc(a->opt, 1);
 		if (b == NULL) {
 			return NULL;
 		}
diff --git a/src/libfsm/clone.c b/src/libfsm/clone.c
index bec16bb0f..78511b12c 100644
--- a/src/libfsm/clone.c
+++ b/src/libfsm/clone.c
@@ -35,7 +35,7 @@ fsm_clone(const struct fsm *fsm)
 	assert(fsm != NULL);
 	assert(fsm->opt != NULL);
 
-	new = fsm_new(fsm->opt);
+	new = fsm_new_statealloc(fsm->opt, fsm->statecount);
 	if (new == NULL) {
 		return NULL;
 	}
diff --git a/src/libfsm/fsm.c b/src/libfsm/fsm.c
index d027825ff..0e46962d1 100644
--- a/src/libfsm/fsm.c
+++ b/src/libfsm/fsm.c
@@ -44,11 +44,15 @@ free_contents(struct fsm *fsm)
 }
 
 struct fsm *
-fsm_new(const struct fsm_options *opt)
+fsm_new_statealloc(const struct fsm_options *opt, size_t statealloc)
 {
 	static const struct fsm_options defaults;
 	struct fsm *new, f;
 
+	if (statealloc == 0) {
+		return fsm_new(opt);
+	}
+
 	if (opt == NULL) {
 		opt = &defaults;
 	}
@@ -60,7 +64,7 @@ fsm_new(const struct fsm_options *opt)
 		return NULL;
 	}
 
-	new->statealloc = FSM_DEFAULT_STATEALLOC;
+	new->statealloc = statealloc;
 	new->statecount = 0;
 	new->endcount   = 0;
 	new->capture_info = NULL;
@@ -92,6 +96,12 @@ fsm_new(const struct fsm_options *opt)
 	return new;
 }
 
+struct fsm *
+fsm_new(const struct fsm_options *opt)
+{
+	return fsm_new_statealloc(opt, FSM_DEFAULT_STATEALLOC);
+}
+
 void
 fsm_free(struct fsm *fsm)
 {
diff --git a/src/libfsm/reverse.c b/src/libfsm/reverse.c
index e896a1bff..b802f364d 100644
--- a/src/libfsm/reverse.c
+++ b/src/libfsm/reverse.c
@@ -54,7 +54,7 @@ fsm_reverse(struct fsm *fsm)
 	if (fsm->endcount == 0 || !fsm_getstart(fsm, &prevstart)) {
 		struct fsm *new;
 
-		new = fsm_new(fsm->opt);
+		new = fsm_new_statealloc(fsm->opt, 1);
 		if (new == NULL) {
 			return 0;
 		}
diff --git a/src/libre/re_strings.c b/src/libre/re_strings.c
index b2f04fec1..ff8b2b4ab 100644
--- a/src/libre/re_strings.c
+++ b/src/libre/re_strings.c
@@ -96,6 +96,7 @@ re_strings_build(struct re_strings *g,
 		}
 	}
 
+	/* TODO: count trie nodes and fsm_new_statealloc() */
 	fsm = fsm_new(opt);
 	if (fsm == NULL) {
 		goto error;