From f7dd8a6e8ab694800526c02fe91b75df92162ab6 Mon Sep 17 00:00:00 2001
From: Christopher Jeffrey <chjjeffrey@gmail.com>
Date: Sat, 21 May 2016 16:37:43 -0700
Subject: [PATCH] hash bloom

---
 lib/bcoin/bloom.js | 105 +++++++++++++++++++++++++++++++++++++++++++++
 test/bloom-test.js |  22 ++++++++++
 2 files changed, 127 insertions(+)

diff --git a/lib/bcoin/bloom.js b/lib/bcoin/bloom.js
index 01e22a5c..24dbaa0c 100644
--- a/lib/bcoin/bloom.js
+++ b/lib/bcoin/bloom.js
@@ -372,6 +372,110 @@ RollingFilter.prototype.added = function added(val, enc) {
   return false;
 };
 
+/**
+ * An object which uses a hash table initially, but
+ * switches to a bloom filter once a limit is reached.
+ * @exports HashFilter
+ * @constructor
+ * @param {Number} items - Expected number of items.
+ * @param {Number} rate - False positive rate (0.0-1.0).
+ * @param {Number} [limit=5000] - Threshold at which
+ * to switch to a bloom filter.
+ */
+
+function HashFilter(items, rate, limit) {
+  if (!(this instanceof HashFilter))
+    return new HashFilter(items, rate, limit);
+
+  this.items = items;
+  this.rate = rate;
+  this.limit = limit || 5000;
+
+  this.filter = null;
+  this.table = {};
+  this.count = 0;
+}
+
+/**
+ * Reset the filter.
+ */
+
+HashFilter.prototype.reset = function reset() {
+  if (this.filter)
+    return this.filter.reset();
+
+  this.table = {};
+  this.count = 0;
+};
+
+/**
+ * Add data to the filter.
+ * @param {Buffer|String}
+ * @param {String?} enc - Can be any of the Buffer object's encodings.
+ */
+
+HashFilter.prototype.add = function add(val, enc) {
+  var i, keys;
+
+  if (this.filter)
+    return this.filter.add(val, enc);
+
+  if (Buffer.isBuffer(val))
+    val = val.toString('hex');
+
+  if (this.table[val])
+    return false;
+
+  this.table[val] = true;
+  this.count++;
+
+  if (this.count > this.limit) {
+    this.filter = new RollingFilter(this.items, this.rate);
+
+    keys = Object.keys(this.table);
+
+    for (i = 0; i < keys.length; i++)
+      this.filter.add(keys[i], 'hex');
+
+    this.table = {};
+    this.count = 0;
+  }
+
+  return true;
+};
+
+/**
+ * Test whether data is present in the filter.
+ * @param {Buffer|String} val
+ * @param {String?} enc - Can be any of the Buffer object's encodings.
+ * @returns {Boolean}
+ */
+
+HashFilter.prototype.test = function test(val, enc) {
+  if (this.filter)
+    return this.filter.test(val, enc);
+
+  if (Buffer.isBuffer(val))
+    val = val.toString('hex');
+
+  return this.table[val] === true;
+};
+
+/**
+ * Test whether data is present in the
+ * filter and potentially add data.
+ * @param {Buffer|String} val
+ * @param {String?} enc - Can be any of the Buffer object's encodings.
+ * @returns {Boolean} Whether data was added.
+ */
+
+HashFilter.prototype.added = function added(val, enc) {
+  if (this.filter)
+    return this.filter.added(val, enc);
+
+  return this.add(val, enc);
+};
+
 /*
  * Murmur3
  */
@@ -491,5 +595,6 @@ function write(data, value, off) {
 exports = Bloom;
 exports.murmur3 = murmur3;
 exports.rolling = RollingFilter;
+exports.hash = HashFilter;
 
 module.exports = Bloom;
diff --git a/test/bloom-test.js b/test/bloom-test.js
index 4691cc55..aba76820 100644
--- a/test/bloom-test.js
+++ b/test/bloom-test.js
@@ -60,6 +60,28 @@ describe('Bloom', function() {
     }
   });
 
+  it('should test hash filter', function() {
+    var filter = new bcoin.bloom.hash(210000, 0.00001, 700);
+    filter.tweak = 0xdeadbeef;
+    // ~1m operations
+    for (var i = 0; i < 1000; i++) {
+      var n = i.toString(16);
+      if (n.length % 2 !== 0)
+        n = '0' + n;
+      var str = 'deadbeef' + n;
+      filter.add(str, 'hex');
+      var j = i;
+      do {
+        var n = j.toString(16);
+        if (n.length % 2 !== 0)
+          n = '0' + n;
+        var str = 'deadbeef' + n;
+        assert(filter.test(str, 'hex') === true);
+        assert(filter.test('00' + str, 'hex') === false);
+      } while (j--);
+    }
+  });
+
   it('should handle rolling generations', function() {
     var filter = new bcoin.bloom.rolling(50, 0.00001);
     filter.tweak = 0xdeadbeee;