MichelleDB

A very simple dynamically mapped document database with lucene-like filters and indexes in Rust.

Majorly influenced by this blog post

Run

Run with Rust 1.62+

cargo run 

To reindex documents, run with re_index feature

cargo run --features re_index

Usage

With the server running on port 9001 in a terminal, in another terminal:

Add document

curl -X POST -H 'Content-Type: application/json' -d '{"name": "Ogechi", "age": "45"}' http://localhost:9001/docs

{"body":{"id":"b6a921c3-eb7d-4946-ba05-c0edad0eff4c"},"status":"ok"}

GET Document:

From the example above:

 curl http://localhost:9001/docs/b6a921c3-eb7d-4946-ba05-c0edad0eff4c | jq   
 
 {
  "body": {
    "body": {
      "age": "45",
      "name": "Ogechi"
    },
    "id": "b6a921c3-eb7d-4946-ba05-c0edad0eff4c"
  },
  "status": "ok"
}

GET Document with filters

curl http://localhost:9001/docs?q=name:Ogechi | jq 
{
  "body": {
    "count": 1,
    "documents": [
      {
        "body": {
          "age": "45",
          "name": "Ogechi"
        },
        "id": "b6a921c3-eb7d-4946-ba05-c0edad0eff4c"
      }
    ]
  },
  "status": "ok"
}

By default, only indexes are used and if you want range queries e.g < > without a direct (equal) query to skip the index and result in a scan, you can easily set the optional skipIndex query to true which will scan the table applying the range filter

curl -G --data-urlencode 'q=age:<50' --data-urlencode 'skipIndex=true' http://localhost:9001/docs | jq
{
  "body": {
    "count": 1,
    "documents": [
      {
        "body": {
          "age": "45",
          "name": "Ogechi"
        },
        "id": "b6a921c3-eb7d-4946-ba05-c0edad0eff4c"
      }
    ]
  },
  "status": "ok"
}

Load test

To load in mock data, simply run:

./script.sh movies.json

and Ctrl+C when you're tired of waiting :)

Sample implementation with mutexes