Blazingly fast code search 🏎️

About

Sourcebot is a fast code indexing and search tool for your codebases. It is built ontop of the zoekt indexer, originally authored by Han-Wen Nienhuys and now maintained by Sourcegraph.

Getting Started

Using Docker

0. Install Docker

1. Create a config.json file and list the repositories you want to index. The JSON schema index.json defines the structure of the config file and the available options. For example, if we want to index Sourcebot on its own code, we could use the following config found in sample-config.json:

   {
       "$schema": "https://raw.githubusercontent.com/TaqlaAI/sourcebot/main/schemas/index.json",
       "Configs": [
           {
               "Type": "github",
               "GitHubOrg": "TaqlaAI",
               "Name": "^sourcebot$"
           }
       ]
   }
   

Sourcebot also supports indexing GitLab & BitBucket. Checkout the index.json for a full list of available options.

2. Create a Personal Access Token (PAT) to authenticate with a code host(s):

   GitHub

   Generate a GitHub Personal Access Token (PAT) [here](https://github.com/settings/tokens/new). If you're only indexing public repositories select the `public_repo` scope; otherwise, select the `repo` scope.

   You'll need to pass this PAT each time you run Sourcebot, so we recommend adding it as an environment variable. In this guide, we'll add the Github PAT as an environment variable called GITHUB_TOKEN:

   export GITHUB_TOKEN=<your-token-here>
   

   If you'd like to persist this environment variable across shell sessions, please add this line to your shell config file (ex. ~/.bashrc, ~/.bash_profile, etc)

   GitLab

   TODO

   BitBucket

   TODO

3. Launch the latest image from the ghcr registry:

   GitHub

   Run the sourcebot docker image, passing in the Github PAT you generated in the previous step as an environment variable called GITHUB_TOKEN:

   docker run -p 3000:3000 --rm --name sourcebot -v $(pwd):/data -e GITHUB_TOKEN=$GITHUB_TOKEN ghcr.io/taqlaai/sourcebot:main
   

   GitLab

   docker run -p 3000:3000 --rm --name sourcebot -v $(pwd):/data -e GITLAB_TOKEN=<token> ghcr.io/taqlaai/sourcebot:main
   

   BitBucket

   TODO

   Two things should happen: (1) a .sourcebot directory will be created containing the mirror repositories and indexes, and (2) you will see output similar to:

   INFO spawned: 'node-server' with pid 10
   INFO spawned: 'zoekt-indexserver' with pid 11
   INFO spawned: 'zoekt-webserver' with pid 12
   run [zoekt-mirror-github -dest /data/.sourcebot/repos -delete -org <org>]
   ...
   INFO success: node-server entered RUNNING state, process has stayed up for > than 1 seconds (startsecs)
   INFO success: zoekt-indexserver entered RUNNING state, process has stayed up for > than 1 seconds (startsecs)
   INFO success: zoekt-webserver entered RUNNING state, process has stayed up for > than 1 seconds (startsecs)
   

   zoekt will now index your repositories (at HEAD). By default, it will re-index existing repositories every hour, and discover new repositories every 24 hours.

4. Go to http://localhost:3000 - once an index has been created, you can start searching.

Building Sourcebot

1. Install go and NodeJS. Note that a NodeJS version of at least 21.1.0 is required.

2. Install ctags (required by zoekt-indexserver): Mac: brew install universal-ctags Ubuntu: apt-get install universal-ctags

3. Clone the repository with submodules:

   git clone --recurse-submodules https://github.com/TaqlaAI/sourcebot.git
   

4. Run make to build zoekt and install dependencies:

   cd sourcebot
   make
   

The zoekt binaries and web dependencies are placed into bin and node_modules respectively.

3. Create a config.json file and list the repositories you want to index. The JSON schema defined in index.json defines the structure of the config file and the available options. For example, if we want to index Sourcebot on its own code, we could use the following config found in sample-config.json:

   {
       "$schema": "https://raw.githubusercontent.com/TaqlaAI/sourcebot/main/schemas/index.json",
       "Configs": [
           {
               "Type": "github",
               "GitHubOrg": "TaqlaAI",
               "Name": "^sourcebot$"
           }
       ]
   }
   

4. Create a Personal Access Token (PAT) to authenticate with a code host:

   GitHub

   Generate a GitHub Personal Access Token (PAT) [here](https://github.com/settings/tokens/new). If you are indexing public repositories only, you can select the `public_repo` scope, otherwise you will need the `repo` scope.

   Create a text file named .github-token in your home directory and paste the token in it. The file should look like:

   ghp_...
   

   zoekt will read this file to authenticate with GitHub.

   GitLab

   TODO

   BitBucket

   TODO

5. Start Sourcebot with the command:

   yarn dev
   

   A .sourcebot directory will be created and zoekt will begin to index the repositories found given config.json.

6. Go to http://localhost:3000 - once an index has been created, you can start searching.

Telemetry

By default, Sourcebot collects anonymized usage data through PostHog to help us improve the performance and reliability of our tool. We do not collect or transmit any information related to your codebase. All events are sanitized to ensure that no sensitive or identifying details leave your machine. The data we collect includes general usage statistics and metadata such as query performance (e.g., search duration, error rates) to monitor the application's health and functionality. This information helps us better understand how Sourcebot is used and where improvements can be made :)

If you'd like to disable all telemetry, you can do so by setting the environment variable SOURCEBOT_TELEMETRY_DISABLED to 1 in the docker run command:

docker run -e SOURCEBOT_TELEMETRY_DISABLED=1 /* additional args */ ghcr.io/taqlaai/sourcebot:main

Or if you are building locally, add the following to your .env file:

SOURCEBOT_TELEMETRY_DISABLED=1
NEXT_PUBLIC_SOURCEBOT_TELEMETRY_DISABLED=1