* Add simple implementation of classifiers abstraction and winnow classifier

* Force statfile to work with float values

3 files changed, 168 insertions, 0 deletions

diff --git a/src/classifiers/classifiers.c b/src/classifiers/classifiers.c
new file mode 100644
index 000000000..5dab03122
--- /dev/null
+++ b/src/classifiers/classifiers.c
@@ -0,0 +1,28 @@
+/*
+ * Common classifier functions
+ */
+
+#include <sys/types.h>
+#include "classifiers.h"
+
+struct classifier classifiers[] = {
+	{"winnow", winnow_classify, winnow_learn },
+};
+
+struct classifier*
+get_classifier (char *name)
+{
+	int i;
+
+	for (i = 0; i < sizeof (classifiers) / sizeof (classifiers[0]); i ++) {
+		if (strcmp (classifiers[i].name, name) == 0) {
+			return &classifiers[i];
+		}
+	}
+
+	return NULL;
+}
+
+/*
+ * vi:ts=4
+ */
diff --git a/src/classifiers/classifiers.h b/src/classifiers/classifiers.h
new file mode 100644
index 000000000..eb5e5de4e
--- /dev/null
+++ b/src/classifiers/classifiers.h
@@ -0,0 +1,33 @@
+#ifndef CLASSIFIERS_H
+#define CLASSIFIERS_H
+
+#include <sys/types.h>
+#include "../config.h"
+
+#ifdef HAVE_STDINT_H
+#include <stdint.h>
+#endif
+#include "../mem_pool.h"
+#include "../statfile.h"
+#include "../tokenizers/tokenizers.h"
+
+/* Common classifier structure */
+struct classifier {
+	char *name;
+	double (*classify_func)(statfile_pool_t *pool, char *statfile, GTree *input);
+	void (*learn_func)(statfile_pool_t *pool, char *statfile, GTree *input, int in_class);
+};
+
+/* Get classifier structure by name or return NULL if this name is not found */
+struct classifier* get_classifier (char *name);
+/* Winnow algorithm */
+double winnow_classify (statfile_pool_t *pool, char *statfile, GTree *input);
+void winnow_learn (statfile_pool_t *pool, char *statfile, GTree *input, int in_class);
+
+/* Array of all defined classifiers */
+extern struct classifier classifiers[];
+
+#endif
+/*
+ * vi:ts=4
+ */
diff --git a/src/classifiers/winnow.c b/src/classifiers/winnow.c
new file mode 100644
index 000000000..552f054b2
--- /dev/null
+++ b/src/classifiers/winnow.c
@@ -0,0 +1,107 @@
+/*
+ * Winnow classifier
+ */
+
+#include <sys/types.h>
+#include "classifiers.h"
+
+#define WINNOW_PROMOTION 1.23
+#define WINNOW_DEMOTION 0.83
+
+struct winnow_callback_data {
+	statfile_pool_t *pool;
+	char *filename;
+	double sum;
+	int count;
+	int in_class;
+	time_t now;
+};
+
+static gboolean
+classify_callback (gpointer key, gpointer value, gpointer data) 
+{
+	token_node_t *node = key;
+	struct winnow_callback_data *cd = data;
+	float v;
+	
+	/* Consider that not found blocks have value 1 */
+	if ((v = statfile_pool_get_block (cd->pool, cd->filename, node->h1, node->h2, cd->now)) < 0.00001) {
+		cd->sum += 1;
+	}
+	else {
+		cd->sum += v;
+	}
+
+	cd->count ++;
+
+	return FALSE;
+}
+
+static gboolean
+learn_callback (gpointer key, gpointer value, gpointer data) 
+{
+	token_node_t *node = key;
+	struct winnow_callback_data *cd = data;
+	float v, c;
+
+	c = (cd->in_class) ? WINNOW_PROMOTION : WINNOW_DEMOTION;
+
+	/* Consider that not found blocks have value 1 */
+	if ((v = statfile_pool_get_block (cd->pool, cd->filename, node->h1, node->h2, cd->now)) < 0.00001) {
+		statfile_pool_set_block (cd->pool, cd->filename, node->h1, node->h2, cd->now, c);
+	}
+	else {
+		statfile_pool_set_block (cd->pool, cd->filename, node->h1, node->h2, cd->now, v * c);
+	}
+
+	cd->count ++;
+	
+	return FALSE;
+}
+
+
+double 
+winnow_classify (statfile_pool_t *pool, char *statfile, GTree *input)
+{
+	struct winnow_callback_data data;
+
+	data.pool = pool;
+	data.filename = statfile;
+	data.sum = 0;
+	data.count = 0;
+	data.now = time (NULL);
+
+	if (!statfile_pool_is_open (pool, statfile)) {
+		if (statfile_pool_open (pool, statfile) == -1) {
+			return 0;
+		}
+	}
+
+	g_tree_foreach (input, classify_callback, &data);
+	
+	return data.sum / data.count;
+}
+
+void
+winnow_learn (statfile_pool_t *pool, char *statfile, GTree *input, int in_class)
+{
+	struct winnow_callback_data data;
+
+	data.pool = pool;
+	data.filename = statfile;
+	data.sum = 0;
+	data.count = 0;
+	data.in_class = in_class;
+	data.now = time (NULL);
+
+	if (!statfile_pool_is_open (pool, statfile)) {
+		if (statfile_pool_open (pool, statfile) == -1) {
+			return;
+		}
+	}
+
+	statfile_pool_lock_file (pool, statfile);
+	g_tree_foreach (input, learn_callback, &data);
+	statfile_pool_unlock_file (pool, statfile);
+	
+}