From 53667583c094252818f203b9233a7726d0a4e8a0 Mon Sep 17 00:00:00 2001
From: "Pablo M. Bermudo Garay" <pablombg@gmail.com>
Date: Wed, 9 Feb 2022 19:58:30 +0100
Subject: Add proactive de-duplication to the memory backend

---
 objectstore/memory_backend.go | 42 +++++++++++++++++++++++++++++++++++++-----
 1 file changed, 37 insertions(+), 5 deletions(-)

(limited to 'objectstore/memory_backend.go')

diff --git a/objectstore/memory_backend.go b/objectstore/memory_backend.go
index 7dcfe4a..36ee7b6 100644
--- a/objectstore/memory_backend.go
+++ b/objectstore/memory_backend.go
@@ -5,12 +5,22 @@
 package objectstore
 
 import (
+	"crypto/sha256"
+	"encoding/hex"
 	"fmt"
 	"sync"
 )
 
+type memObject struct {
+	payload string
+	hash    string
+}
+
 type memBucket struct {
-	objects map[string]string
+	objects map[string]memObject
+	// Map used to check if the hash already exists, also
+	// storing the objectId to identify de duplicate
+	hashes map[string]string
 }
 
 type MemObjectStore struct {
@@ -24,18 +34,37 @@ func NewMemBackend() *MemObjectStore {
 	return os
 }
 
-func (os *MemObjectStore) CreateObject(bucketId string, objectId string, object string) {
+func (os *MemObjectStore) CreateObject(bucketId string, objectId string, payload string) error {
 	os.Lock()
 	defer os.Unlock()
 
+	// Create bucket if it doesn't exist
 	bucket, ok := os.buckets[bucketId]
 	if !ok {
 		bucket = memBucket{}
-		bucket.objects = make(map[string]string)
+		bucket.objects = make(map[string]memObject)
+		bucket.hashes = make(map[string]string)
 		os.buckets[bucketId] = bucket
 	}
 
+	// Hash the object content
+	hasher := sha256.New()
+	hasher.Write([]byte(payload))
+	hash := hex.EncodeToString(hasher.Sum(nil))
+
+	// Check for duplicates
+	if objectId, dup := bucket.hashes[hash]; dup {
+		return NewDuplicateError(objectId)
+	}
+
+	// Store the object
+	object := memObject{
+		payload: payload,
+		hash:    hash,
+	}
 	bucket.objects[objectId] = object
+	bucket.hashes[hash] = objectId
+	return nil
 }
 
 func (os *MemObjectStore) GetObject(bucketId string, objectId string) (string, error) {
@@ -52,7 +81,7 @@ func (os *MemObjectStore) GetObject(bucketId string, objectId string) (string, e
 		return "", fmt.Errorf("Object not found")
 	}
 
-	return object, nil
+	return object.payload, nil
 }
 
 func (os *MemObjectStore) DeleteObject(bucketId string, objectId string) error {
@@ -64,12 +93,15 @@ func (os *MemObjectStore) DeleteObject(bucketId string, objectId string) error {
 		return fmt.Errorf("Bucket not found")
 	}
 
-	_, ok = bucket.objects[objectId]
+	object, ok := bucket.objects[objectId]
 	if !ok {
 		return fmt.Errorf("Object not found")
 	}
 
+	// Delete both, object and hash
 	delete(bucket.objects, objectId)
+	delete(bucket.hashes, object.hash)
+
 	// Delete the bucket if it's empty
 	if len(bucket.objects) == 0 {
 		delete(os.buckets, bucketId)
-- 
cgit v1.2.3-70-g09d2