I wrote a parser for Redis protocol so you don't have to

Summary

I wrote a parser for Redis protocol so you don't have to There are a lot of "Write your own Redis" guides online. The general idea is neat: you get to write a simple KV storage and a network interface for it in a guided fashion. You will work with network and file I/O, do some bytes encoding to support Redis wire protocol, write a simple data structure to hold your data. It's a nice way to get acquainted with a new language. In this spirit, I decided to write my own Redis. To be honest, I just wanted to implement a Log-Structured Merge Tree, but it occurred to me that it would be nicer with an interactive interface, so I could use it from other software. My plan was simple: Redis is an established product, it's protocol (also called RESP - Redis Serialization Protocol) is well-documented and has to be relatively simple. Considering amount of "Redis-like" software out there, I expected to find a dozen different parsers to choose from. However, I soon found out that this is not the case. At least for Go, there are a couple projects on Github, but they all share some of these properties: Integrated into existing databases, so you can't just use them as a library Archived 7 years ago Have little to none testing Don't implement half the specification In the end, I shelved the whole LSM Tree idea and decided to write a RESP parser/encoder. Should be easy, after all. The good RESP basis RESP is a text-based protocol. It's very easy to understand what is happening under the hood, and interacting with your server doesn't require any tools more complicated then netcat. It also doesn't have any complicated semantics. All types start with a single byte identifying a type, and end with a CRLF terminator (\r\n). For instance, encoding of string MEOW looks like this: +MEOW\r\n. + means that all text until the next \r\n is a string. Simple enough. There also similar representations for errors (-ERROR\r\n) and integers (+42\r\n). Some types have information about their length encode...