Merge pull request #329 from peakji/fast-seek

250x faster iterator seek
Level · Dec 19, 2016 · 959b78f · 959b78f
2 parents 2936971 + 9d44401
commit 959b78f
Show file tree

Hide file tree

Showing 3 changed files with 43 additions and 11 deletions.
diff --git a/src/iterator.cc b/src/iterator.cc
@@ -58,6 +58,7 @@ Iterator::Iterator (
   count      = 0;
   target     = NULL;
   seeking    = false;
+  landed     = false;
   nexting    = false;
   ended      = false;
   endWorker  = NULL;
@@ -215,6 +216,11 @@ bool Iterator::IteratorNext (std::vector<std::pair<std::string, std::string> >&
       result.push_back(std::make_pair(key, value));
       size = size + key.size() + value.size();
 
+      if (!landed) {
+        landed = true;
+        return true;
+      }
+
       if (size > highWaterMark)
         return true;
 
@@ -273,6 +279,7 @@ NAN_METHOD(Iterator::Seek) {
 
   dbIterator->Seek(*iterator->target);
   iterator->seeking = true;
+  iterator->landed = false;
 
   if (iterator->OutOfRange(iterator->target)) {
     if (iterator->reverse) {

diff --git a/src/iterator.h b/src/iterator.h
@@ -64,6 +64,7 @@ class Iterator : public Nan::ObjectWrap {
   leveldb::Slice* target;
   std::string* end;
   bool seeking;
+  bool landed;
   bool reverse;
   bool keys;
   bool values;

diff --git a/test/iterator-test.js b/test/iterator-test.js
@@ -98,17 +98,41 @@ make('reverse seek from invalid range', function (db, t, done) {
   })
 })
 
-make('iterator seek resets state', function (db, t, done) {
-  var ite = db.iterator()
-  ite.next(function (err, key, value) {
-    t.error(err, 'no error from next()')
-    t.equal(key.toString(), 'one', 'key matches')
-    t.ok(ite.cache, 'has cached items')
-    t.equal(ite.finished, true, 'finished')
-    ite.seek('two')
-    t.notOk(ite.cache, 'cache is removed')
-    t.equal(ite.finished, false, 'resets finished state')
-    ite.end(done)
+make('iterator optimized for seek', function (db, t, done) {
+  var batch = db.batch()
+  batch.put('a', 1)
+  batch.put('b', 1)
+  batch.put('c', 1)
+  batch.put('d', 1)
+  batch.put('e', 1)
+  batch.put('f', 1)
+  batch.put('g', 1)
+  batch.write(function (err) {
+    var ite = db.iterator()
+    t.error(err, 'no error from batch')
+    ite.next(function (err, key, value) {
+      t.error(err, 'no error from next()')
+      t.equal(key.toString(), 'a', 'key matches')
+      t.equal(ite.cache.length, 0, "no cache")
+      ite.next(function (err, key, value) {
+        t.error(err, 'no error from next()')
+        t.equal(key.toString(), 'b', 'key matches')
+        t.ok(ite.cache.length > 0, "has cached items")
+        ite.seek('d')
+        t.notOk(ite.cache, 'cache is removed')
+        ite.next(function (err, key, value) {
+          t.error(err, 'no error from next()')
+          t.equal(key.toString(), 'd', 'key matches')
+          t.equal(ite.cache.length, 0, "no cache")
+          ite.next(function (err, key, value) {
+            t.error(err, 'no error from next()')
+            t.equal(key.toString(), 'e', 'key matches')
+            t.ok(ite.cache.length > 0, "has cached items")
+            done();
+          })
+        })
+      })
+    })
   })
 })