Service: SearXNG - EyevinnOSC/community GitHub Wiki

SearXNG

SearXNG (serviceId searxng-searxng) is a free, open source metasearch engine that aggregates results from more than 70 search engines without tracking users or storing search history. In Eyevinn Open Source Cloud (OSC) it is a long-running service instance with up to 10 GiB of storage per instance. License: AGPL-3.0.

What it can do

  • Provide a shared internal search tool for a team, or a privacy-respecting alternative to commercial search engines.
  • Query many engines across categories (Web, News, Images, Videos, Map and others).
  • Expose a JSON search API for search-powered applications.
  • Let users choose engines, language, region, theme, SafeSearch and the autocomplete engine on the Preferences page (stored in a browser cookie).

Creating an instance

SearXNG has no dependencies on other OSC services. Create it with create-service-instance, serviceId: "searxng-searxng":

{ "name": "mysearch", "AutoComplete": "google" }

Use wait-for-service-ready or get-instance-status, then describe-service-instance for the instance URL. With the OSC CLI:

osc create searxng-searxng mysearch -o AutoComplete="google"

Options

Option Required Type Meaning
name Yes string Name of the instance. Alphanumeric, at most 20 characters.
AutoComplete No string Autocomplete engine to use for search suggestions, for example google, duckduckgo or wikipedia.

Using the instance

Open the instance URL; the search interface is available with no login. Use the JSON API:

curl "https://<your-instance-url>/search?q=open+source+cloud&format=json"

The response contains query, number_of_results and a results array whose items have title, url, content, engine and score.

Parameter Description Example
q Search query (required) q=typescript+tutorial
format html (default) or json format=json
categories Comma-separated categories categories=general,news
engines Comma-separated engines to use engines=google,bing
language Language code language=en
pageno Page number (default: 1) pageno=2
const response = await fetch(
  `https://<your-instance-url>/search?q=${encodeURIComponent(query)}&format=json`
);
const data = await response.json();
const results = data.results.map(r => ({ title: r.title, url: r.url, snippet: r.content }));

Limitations

  • Storage is limited to 10 GiB per instance.

Resources