Golinks is a keyword-to-URL mapping service. It allows the user to enter go/<keyword> in their browser, and is redirected to the corresponding URL.

This services can be used for personal purposes (e.g. go/me opens your website) or for internal link shortening in organizations (e.g. go/kpi opens the company's KPIs dashboard).

This repository consists of a few services:

Server-side services:

• redirector - a simple HTTP server that redirects keyword requests to the corresponding URL.
• crud - a gRPC service that provides CRUD operations for the keyword-to-URL mappings.
• crud_http - an HTTP service that provides CRUD operations for the keyword-to-URL mappings.
• web - a web app that allows easier management of the mappings and the server. (WIP)
• browser/chrome - a Chrome extension that allows the user to configure the Golinks server URL.

The mapping can be stored in various ways including local files or in database. This behavior is configurable via a configuration file.

You can run both services with all default configurations by:

go run .

Then, go to your browser and try open localhost:8080/gh. It should redirect you to https://github.com.

If you would like to just type in go/gh (instead of localhost:8080/go/gh) and make it work, you need to add 127.0.0.1 go to your host file, which usually is /etc/hosts on Linux/MacOS and C:\Windows\System32\drivers\etc\hosts on Windows.

If you are using Chrome, you can also install the extension that is included in browser/chrome/src directory to your Chrome browser. Basically, go to chrome://extensions, enable "Developer mode" (using top-right corner switch), click "Load unpacked" and select the browser/chrome/src directory.

Then, in extension configuration page, you can specify the server URL where you are hosting the redirector service, which by default is http://localhost:8080.

You can specify a configuration file by running:

go run . -config <path_to_config_file>

See files/config.yaml for schema.

Mappers are key-value stores that map keywords to URLs.

You can specify the mapper in the configuration file. The redirector services supports 4 types of mappers:

readonly mappers does not support put or delete operations. singleton mappers can only exist once in the system. You can specify one single such mapper in the configuration file.

You can only designate one mapper as persistor in the configuration file, and that mapper has to be a non-readonly mapper.

If the persistor field is not specified, then the entire system would be readonly.

If there are multiple mappers configured, CRUD operations would be resolved by the following rules:

• GET: any query would run through the list of mappers and the first successful match would be returned.
• PUT:
  • insert: the mapper designated as persistor would be used.
  • update: the key-value pair would be updated in all mappers that support the GET operation.
• DELETE: the key-value pair would be deleted from the mapper that returns a match by the GET operation rule.

See code comments in pkg/sanitizer for details.

The CRUD operations are exposed as a gRPC service. You can use the grpcurl tool to interact with the service.

grpcurl -plaintext -d '{"path": "gh"}' localhost:8081 pb.Golinks/GetUrl

The CRUD operations are also exposed as a HTTP service. You can use the curl tool to interact with the service.

curl -v http://localhost:8082/go
curl -v http://localhost:8082/go/gh
curl -X PUT -v localhost:8082/go \
     -H "Content-Type: application/json" \
     -d '{"path":"prom","url":"https://prometheus.io"}'
curl -X DELETE -v http://localhost:8082/go/prom

This is intended to be interacted with by a CLI tool.

WIP. Should provide:

• basic CRUD operations for user to manage the URL mappings.
• listing all the existing keywords, the list should be searchable and sortable (esp by use count).
• route go/d/* to the web interface. (need to change chrome extension code)
• fuzzy search for keywords (need to implement in backend too)

• redirector,crud,crud_http: See Makefile for commands to run tests, build and clean the server project.
• web: Go to web. npm i to install dependencies. npm run dev to start the development server.
• browser/chrome: Go to browser/chrome. npm i to install dependencies. npm test to run the tests. See chrome doc to load the extension in browser.

• Web UI/CLI for easier management of the mappings.
• Fuzzy search for keywords
• Deployment scheme
  • containerize and use Kubernetes, Terraform for deployment. Will be more necessary if we want to scale/use stuff like envoy (for grpc-web proxying for example) or connecting to logging/monitoring services.
• Authentication/Authorization
  • integrate with org-specific auth services for access control (who gets to modify/view the mappings etc).
• Monitoring and logging
  • the logs are now going into stdout only.