redis2.0

A custom Redis server built from scratch in Node.js — implements the RESP (REdis Serialization Protocol) parser and a small set of commands over raw TCP, without using any Redis libraries.

Features

• Hand-written RESP2 parser (decoder + encoder), handling streamed/fragmented TCP data and inline (telnet-style) commands.
• Command dispatch for PING, SET, GET, DEL, EXISTS, FLUSHDB, LPUSH, RPUSH, RPOP, LRANGE, HSET, HGET, HGETALL.
• Shared in-memory key-value store, with values as strings, arrays (lists), or plain objects (hashes) — WRONGTYPE errors when a command is used against the wrong kind of value for a key.
• Error logging to a file (logs/error.log) in addition to the console.

Folder structure

index.js              Entry point — starts the TCP server on port 8000
server/
  server.js            net.createServer setup, per-connection RESP decoding,
                        dispatches parsed commands to the command handlers
commands/
  index.js             Dispatch table: command name -> handler
  ping.js
  set.js
  get.js
  del.js
  exists.js
  flushdb.js
  lpush.js
  rpush.js
  rpop.js
  lrange.js
  hset.js
  hget.js
  hgetall.js
store/
  store.js             Shared in-memory Map used by all commands
redis-parser/
  decoder.js            RESPDecoder — parses raw buffers into commands
  encoder.js            encode() and friends — turns JS values into RESP replies
  index.js              Public exports for the parser module
utils/
  logger.js             logError() — writes timestamped errors to logs/error.log
  typeCheck.js         isString/isList/isHash — value-type guards for WRONGTYPE checks
logs/
  error.log             Generated at runtime

Running the server

node index.js

Starts listening on 127.0.0.1:8000 and logs Custom Redis Server running on port 8000.

Testing it

The easiest way is redis-cli, since it speaks proper RESP regardless of what you type:

redis-cli -p 8000
127.0.0.1:8000> ping
PONG
127.0.0.1:8000> set name munishwar
OK
127.0.0.1:8000> get name
"munishwar"

Raw inline commands also work via a client that sends CRLF line endings, e.g. ncat -C 127.0.0.1 8000 (plain nc without -C won't work — it only sends \n, not \r\n).

Supported commands

Command	Usage	Reply
PING	PING [msg]	PONG, or echoes msg if given
SET	SET key value	OK
GET	GET key	the stored value, or nil if not found
DEL	DEL key [key ...]	integer count of keys actually deleted
EXISTS	EXISTS key [key ...]	integer count of given keys that exist
FLUSHDB	FLUSHDB	OK — removes all keys from the store
LPUSH	LPUSH key value [value ...]	integer length of the list after pushing
RPUSH	RPUSH key value [value ...]	integer length of the list after pushing
RPOP	RPOP key	the popped value, or nil if the list/key is empty
LRANGE	LRANGE key start stop	array of elements in that range (supports negative indices)
HSET	HSET key field value [field value ...]	integer count of fields newly added
HGET	HGET key field	the field's value, or nil if missing
HGETALL	HGETALL key	flattened array [field, value, field, value, ...]

LPUSH/RPUSH/LRANGE error with WRONGTYPE if key already holds a string; HSET/HGET/HGETALL do the same for lists. Popping the last element of a list removes the key entirely, same as real Redis.

Any other command falls back to replying OK.

Known limitations

• Data is in-memory only — nothing is persisted to disk, so a restart wipes the store.
• No expiry (EX/PX).
• Lists are missing LPOP/LLEN; hashes are missing HDEL/HEXISTS/HLEN. No sets or sorted sets yet.
• No authentication.
• This is a learning project, not intended for production use.