1 files changed, 104 insertions, 22 deletions

diff --git a/km.h b/km.h
index 0c49832..642560b 100644
--- a/km.h
+++ b/km.h
@@ -2,6 +2,8 @@
 #define KM_H
 
 #include <stddef.h> // size_t
+#include <stdint.h> // uint8_t
+#include <stdio.h> // FILE
 
 // forward typedef for callbacks
 typedef struct km_rand_src_t_ km_rand_src_t;
@@ -33,9 +35,19 @@ typedef struct {
          num_ints;
 } km_shape_t;
 
+typedef enum {
+  KM_SET_STATE_INIT,
+  KM_SET_STATE_NORMALIZED,
+  KM_SET_STATE_FINI,
+  KM_SET_STATE_LAST,
+} km_set_state_t;
+
 // data set
 typedef struct {
-  float *floats;
+  km_set_state_t state;
+
+  float *floats,
+        *bounds;
   int *ints;
 
   km_shape_t shape;
@@ -50,14 +62,28 @@ _Bool km_set_init(km_set_t * const, const km_shape_t *, const size_t);
 void km_set_fini(km_set_t * const);
 
 // append rows to data set, growing set if necessary
-_Bool km_set_push_rows(km_set_t *, const size_t, const float *, const int *);
+_Bool km_set_push(km_set_t *, const size_t, const float *, const int *);
 
-// get row from data set
+// normalize data set
+_Bool km_set_normalize(km_set_t * const);
+
+// get pointer to data set row
 float *km_set_get_row(const km_set_t *, const size_t);
 
-// init a cluster set of size N from a data set by picking random
-// cluster centers
-_Bool km_clusters_rand_init(
+typedef struct {
+  float d2;
+  size_t cluster;
+} km_row_map_t;
+
+typedef struct {
+  const float *means;
+  const float *variances;
+  const size_t num_clusters;
+} km_solve_stats_t;
+
+// init a set with num_clusters clusters of shape num_floats by picking
+// random cluster centers
+_Bool km_set_init_rand_clusters(
   km_set_t *,
   const size_t,
   const size_t,
@@ -65,40 +91,96 @@ _Bool km_clusters_rand_init(
 );
 
 typedef struct {
-  void (*on_step)(const km_set_t *, void *);
-  void (*on_means)(const km_set_t *, const float *, const size_t, void *);
-} km_clusters_solve_cbs_t;
+  void (*on_step)(const km_set_t *, const km_row_map_t *, void *);
+  void (*on_stats)(const km_set_t *, const km_solve_stats_t *, void *);
+} km_solve_cbs_t;
 
 // use k-means to iteratively update the cluster centroids until there
 // are no more changes to the centroids
 _Bool
-km_clusters_solve(
+km_solve(
   km_set_t * const,
   const km_set_t * const,
-  const km_clusters_solve_cbs_t * const,
+  const km_solve_cbs_t * const,
   void * const
 );
 
 typedef struct {
   const km_set_t * const cluster_set;
-  const size_t num_clusters;
-  const float mean_distance;
-  const size_t num_empty_clusters;
-} km_search_row_t;
 
-typedef void (*km_search_row_cb_t)(
-  const km_search_row_t * const,
+  const float mean_distance,
+              mean_variance,
+              mean_cluster_size;
+
+  const size_t num_clusters,
+               num_empty_clusters;
+} km_find_data_t;
+
+typedef _Bool (*km_find_init_cb_t)(
+  km_set_t * const cluster_set,
+  const size_t num_floats,
+  const size_t num_clusters,
+  void *cb_data
+);
+
+typedef _Bool (*km_find_fini_cb_t)(
+  km_set_t * const cluster_set,
+  void *cb_data
+);
+
+typedef void (*km_find_data_cb_t)(
+  const km_find_data_t * const,
   void *
 );
 
+typedef struct {
+  size_t max_clusters;
+  size_t num_tests;
+  km_rand_src_t *rand_src;
+
+  km_find_init_cb_t on_init;
+  km_find_fini_cb_t on_fini;
+  km_find_data_cb_t on_data;
+} km_find_cbs_t;
+
 // repeatedly test different cluster sizes and report results
 _Bool
-km_search(
+km_find(
   const km_set_t * const set,
-  const size_t max_clusters,
-  const size_t num_tests,
-  const km_search_row_cb_t on_row,
-  void *cb_data
+  const km_find_cbs_t * const cbs,
+  void * const cb_data
+);
+
+// render set to rgb buffer
+void
+km_set_draw(
+  const km_set_t * const set,
+  uint8_t * const rgb,
+  const size_t width,
+  const size_t height,
+  const uint32_t color
+);
+
+// load callbacks
+typedef struct {
+  _Bool (*on_shape)(const km_shape_t * const, void *);
+  _Bool (*on_row)(const float *, const int *, void *);
+  void (*on_error)(const char * const, void *);
+} km_load_cbs_t;
+
+// load set from file
+_Bool
+km_load(
+  FILE * const fh,
+  const km_load_cbs_t * const cbs,
+  void * const cb_data
+);
+
+_Bool
+km_load_path(
+  const char * const path,
+  const km_load_cbs_t * const cbs,
+  void * const cb_data
 );
 
 #endif /* KM_H */