Add comparision function for shingles.

2 files changed, 25 insertions, 0 deletions

diff --git a/src/libutil/shingles.c b/src/libutil/shingles.c
index 2307a318c..653fa5356 100644
--- a/src/libutil/shingles.c
+++ b/src/libutil/shingles.c
@@ -135,3 +135,18 @@ rspamd_shingles_default_filter (guint64 *input, gsize count,
 
 	return minimal;
 }
+
+
+gdouble rspamd_shingles_compare (const struct rspamd_shingle *a,
+		const struct rspamd_shingle *b)
+{
+	gint i, common = 0;
+
+	for (i = 0; i < RSPAMD_SHINGLE_SIZE; i ++) {
+		if (a->hashes[i] == b->hashes[i]) {
+			common ++;
+		}
+	}
+
+	return (gdouble)common / 84.0;
+}
diff --git a/src/libutil/shingles.h b/src/libutil/shingles.h
index 571199992..39e9cf38d 100644
--- a/src/libutil/shingles.h
+++ b/src/libutil/shingles.h
@@ -58,6 +58,16 @@ struct rspamd_shingle* rspamd_shingles_generate (GArray *input,
 		gpointer filterd);
 
 /**
+ * Compares two shingles and return result as a floating point value - 1.0
+ * for completely similar shingles and 0.0 for completely different ones
+ * @param a
+ * @param b
+ * @return
+ */
+gdouble rspamd_shingles_compare (const struct rspamd_shingle *a,
+		const struct rspamd_shingle *b);
+
+/**
  * Default filtering function
  */
 guint64 rspamd_shingles_default_filter (guint64 *input, gsize count,