add km_init(), refactor existing init

6 files changed, 112 insertions, 74 deletions

diff --git a/Makefile b/Makefile
index f3a71d5..cfb7ae2 100644
--- a/Makefile
+++ b/Makefile
@@ -1,7 +1,7 @@
 APP=km-test
 CFLAGS=-W -Wall -Wextra -Werror -pedantic -std=c11 -O2
 OBJS=km-set.o km-draw.o km-load.o km-find.o km-rand.o km-solve.o \
-     km-init-type.o km-init-rand.o km-init-forgy.o main.o
+     km-init-rand.o km-init-forgy.o km-init.o main.o
 LIBS=-lm
 
 .PHONY=all clean
diff --git a/km-init-rand.c b/km-init-rand.c
index f9d6eb6..c6000bd 100644
--- a/km-init-rand.c
+++ b/km-init-rand.c
@@ -9,9 +9,12 @@ bool
 km_init_rand(
   km_set_t * const cs,
   const size_t num_clusters,
-  const size_t num_floats,
+  const km_set_t * const set,
   km_rand_t * const rs
 ) {
+  // get number of floats from data set
+  const size_t num_floats = set->shape.num_floats;
+
   // init cluster shape
   const km_shape_t shape = {
     .num_floats = num_floats,
diff --git a/km-init-type.c b/km-init-type.c
deleted file mode 100644
index 27fc1ee..0000000
--- a/km-init-type.c
+++ /dev/null
@@ -1,32 +0,0 @@
-#include <stddef.h> // size_t
-#include <string.h> // strcmp()
-#include "km.h"
-
-static const struct {
-  const km_init_type_t type;
-  const char * const name;
-} TYPES[] = {{
-  .name = "rand",
-  .type = KM_INIT_TYPE_RAND,
-}, {
-  .name = "forgy",
-  .type = KM_INIT_TYPE_FORGY,
-}};
-
-#define NUM_TYPES (sizeof(TYPES) / sizeof(TYPES[0]))
-
-km_init_type_t
-km_get_init_type(
-  const char * const s
-) {
-  // find init method
-  for (size_t i = 0; i < NUM_TYPES; i++) {
-    if (!strcmp(s, TYPES[i].name)) {
-      // return type
-      return TYPES[i].type;
-    }
-  }
-
-  // return failure
-  return KM_INIT_TYPE_LAST;
-}
diff --git a/km-init.c b/km-init.c
new file mode 100644
index 0000000..f04fca0
--- /dev/null
+++ b/km-init.c
@@ -0,0 +1,80 @@
+#include <stdbool.h> // bool
+#include <stddef.h> // size_t
+#include <string.h> // strcmp()
+#include "util.h"
+#include "km.h"
+
+typedef bool (*km_init_fn_t)(
+  km_set_t *,
+  const size_t,
+  const km_set_t *,
+  km_rand_t *
+);
+
+bool km_init_rand(km_set_t *, const size_t, const km_set_t *, km_rand_t *);
+bool km_init_forgy(km_set_t *, const size_t, const km_set_t *, km_rand_t *);
+
+static const struct {
+  const km_init_type_t type;
+  const char * const name;
+  const km_init_fn_t init;
+} TYPES[] = {{
+  .name = "rand",
+  .type = KM_INIT_TYPE_RAND,
+  .init = km_init_rand,
+}, {
+  .name = "forgy",
+  .type = KM_INIT_TYPE_FORGY,
+  .init = km_init_forgy,
+}};
+
+#define NUM_TYPES (sizeof(TYPES) / sizeof(TYPES[0]))
+
+km_init_type_t
+km_init_get_type(
+  const char * const s
+) {
+  // find init method
+  for (size_t i = 0; i < NUM_TYPES; i++) {
+    if (!strcmp(s, TYPES[i].name)) {
+      // return type
+      return TYPES[i].type;
+    }
+  }
+
+  // return failure
+  return KM_INIT_TYPE_LAST;
+}
+
+static km_init_fn_t
+get_init_func(
+  const km_init_type_t type
+) {
+  // find init method
+  for (size_t i = 0; i < NUM_TYPES; i++) {
+    if (type == TYPES[i].type) {
+      // return init func
+      return TYPES[i].init;
+    }
+  }
+
+  // return failure
+  return NULL;
+}
+
+bool
+km_init(
+  km_set_t * const cs,
+  const km_init_type_t init_type,
+  const size_t num_clusters,
+  const km_set_t * const set,
+  km_rand_t * const rs
+) {
+  km_init_fn_t init_fn = get_init_func(init_type);
+  if (!init_fn) {
+    die("unknown cluster init method");
+  }
+
+  // call init, return result
+  return init_fn(cs, num_clusters, set, rs);
+}
diff --git a/km.h b/km.h
index 6f09009..866e941 100644
--- a/km.h
+++ b/km.h
@@ -5,14 +5,6 @@
 #include <stdint.h> // uint8_t
 #include <stdio.h> // FILE
 
-typedef enum {
-  KM_INIT_TYPE_RAND,
-  KM_INIT_TYPE_FORGY,
-  KM_INIT_TYPE_LAST,
-} km_init_type_t;
-
-km_init_type_t km_get_init_type(const char * const);
-
 // forward typedef for callbacks
 typedef struct km_rand_t_ km_rand_t;
 
@@ -85,22 +77,6 @@ _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 cluster set by picking random centroids
-_Bool km_init_rand(
-  km_set_t *,
-  const size_t,
-  const size_t,
-  km_rand_t *
-);
-
-// init cluster set by picking random rows from set
-_Bool km_init_forgy(
-  km_set_t * const,
-  const size_t,
-  const km_set_t * const,
-  km_rand_t *
-);
-
 typedef struct {
   float d2;
   size_t cluster;
@@ -197,7 +173,7 @@ typedef struct {
   void (*on_error)(const char * const, void *);
 } km_load_cbs_t;
 
-// load set from file
+// load set from file handle
 _Bool
 km_load(
   FILE * const fh,
@@ -205,11 +181,32 @@ km_load(
   void * const cb_data
 );
 
-_Bool
-km_load_path(
+// load set from path
+_Bool km_load_path(
   const char * const path,
   const km_load_cbs_t * const cbs,
   void * const cb_data
 );
 
+typedef enum {
+  // init cluster set by picking random centroids
+  KM_INIT_TYPE_RAND,
+
+  // init cluster set by picking random rows from set
+  KM_INIT_TYPE_FORGY,
+
+  KM_INIT_TYPE_LAST,
+} km_init_type_t;
+
+km_init_type_t km_init_get_type(const char * const);
+
+// init cluster set using given method
+_Bool km_init(
+  km_set_t * const,
+  const km_init_type_t,
+  const size_t,
+  const km_set_t * const,
+  km_rand_t * const
+);
+
 #endif /* KM_H */
diff --git a/main.c b/main.c
index f071804..6acaf60 100644
--- a/main.c
+++ b/main.c
@@ -141,17 +141,7 @@ find_on_init(
   void *cb_data
 ) {
   ctx_t * const ctx = cb_data;
-  km_rand_t * const rs = &(ctx->rs);
-
-  switch(ctx->init_type) {
-  case KM_INIT_TYPE_RAND:
-    return km_init_rand(cs, num_clusters, set->shape.num_floats, rs);
-  case KM_INIT_TYPE_FORGY:
-    return km_init_forgy(cs, num_clusters, set, rs);
-  default:
-    die("unknown cluster init method");
-    return false;
-  }
+  return km_init(cs, ctx->init_type, num_clusters, set, &(ctx->rs));
 }
 
 static bool
@@ -338,7 +328,7 @@ int main(int argc, char *argv[]) {
   ctx_t ctx;
   memset(&ctx, 0, sizeof(ctx_t));
   km_rand_init_system(&(ctx.rs));
-  ctx.init_type = km_get_init_type(init_type_name);
+  ctx.init_type = km_init_get_type(init_type_name);
 
   // init data set
   km_set_t set;
@@ -346,7 +336,7 @@ int main(int argc, char *argv[]) {
     die("km_load_path(\"%s\") failed", data_path);
   }
 
-  // init data set
+  // normalize data set
   if (!km_set_normalize(&set)) {
     die("km_set_normalize() failed");
   }