From aa999f87f16456145c29ac7c3ea21e13a9bdf11b Mon Sep 17 00:00:00 2001
From: Brian Hulette <brian.hulette@ccri.com>
Date: Wed, 10 Jan 2018 14:38:49 -0500
Subject: [PATCH] Add DictionaryVector optimization for equals predicate

---
 js/perf/table_config.js       |  2 +-
 js/src/dataframe/predicate.ts | 26 +++++++++++++++++++++++++-
 2 files changed, 26 insertions(+), 2 deletions(-)

diff --git a/js/perf/table_config.js b/js/perf/table_config.js
index 06c9198353b80..7bface6d2cdde 100644
--- a/js/perf/table_config.js
+++ b/js/perf/table_config.js
@@ -25,7 +25,7 @@ const filenames = glob.sync(path.resolve(__dirname, `../test/data/tables/`, `*.a
 tests = [
     {col: 0, test: 'gteq', value: 0        },
     {col: 1, test: 'gteq', value: 0        },
-    //{col: 2, test:   'eq', value: 'Seattle'},
+    {col: 2, test:   'eq', value: 'Seattle'},
 ]
 
 for (const filename of filenames) {
diff --git a/js/src/dataframe/predicate.ts b/js/src/dataframe/predicate.ts
index 4438c0adbaa98..263b8646d71fc 100644
--- a/js/src/dataframe/predicate.ts
+++ b/js/src/dataframe/predicate.ts
@@ -1,4 +1,5 @@
 import { Vector } from "../vector/vector";
+import { DictionaryVector } from "../vector/dictionary";
 
 export type ValueFunc<T> = (idx: number, cols: Vector[]) => T|null;
 export type PredicateFunc = (idx: number, cols: Vector[]) => boolean;
@@ -118,7 +119,30 @@ class Equals extends ComparisonPredicate {
 
     protected _bindColLit(cols: Vector<any>[], col: Col     , lit: Literal  ): PredicateFunc {
         const col_func = col.bind(cols);
-        return (idx: number, cols: Vector[]) => col_func(idx, cols) == lit.v;
+        if (col.vector instanceof DictionaryVector) {
+            // Assume that there is only one key with the value `lit.v`
+            let key = -1
+            for (; ++key < col.vector.data.length;) {
+                if (col.vector.data.get(key) === lit.v) {
+                    break;
+                }
+            }
+
+            if (key == col.vector.data.length) {
+                // the value doesn't exist in the dictionary - always return
+                // false
+                // TODO: special-case of PredicateFunc that encapsulates this
+                // "always false" behavior. That way filtering operations don't
+                // have to bother checking
+                return () => false;
+            } else {
+                return (idx: number) => {
+                    return (col.vector as DictionaryVector<any>).getKey(idx) === key;
+                }
+            }
+        } else {
+            return (idx: number, cols: Vector[]) => col_func(idx, cols) == lit.v;
+        }
     }
 }