haystack-brave-search

A Haystack component for web search using the Brave Search API.

Installation

pip install haystack-brave-search

Requirements

• Python 3.12+
• A Brave Search API key

Usage

Standalone

from haystack_brave_search import BraveWebSearch
import os

# Set the API key using an environment variable
brave_api_key = os.environ["BRAVE_API_KEY"]

# Parameters set at initialisation are used as defaults for every call
searcher = BraveWebSearch(api_key=brave_api_key, top_k=5, country="BE", language="en")
result = searcher.run(query="Haystack AI framework")

# Any parameter can be overridden at run time
result = searcher.run(query="Haystack AI framework", top_k=3, country="US", language="en")

for doc in result["documents"]:
    print(doc.meta["title"])
    print(doc.meta["url"])
    print(doc.content)

In a Haystack Pipeline

from haystack import Pipeline
from haystack.components.builders import PromptBuilder
from haystack_brave_search import BraveWebSearch

prompt_template = """
Given the following search results:
{% for doc in documents %}
- {{ doc.meta.title }}: {{ doc.content }}
{% endfor %}

Answer the question: {{ query }}
"""

pipe = Pipeline()
pipe.add_component("search", BraveWebSearch(api_key="your-api-key", top_k=5))
pipe.add_component("prompt", PromptBuilder(template=prompt_template))

pipe.connect("search.documents", "prompt.documents")

result = pipe.run({
    "search": {"query": "What is RAG?"},
    "prompt": {"query": "What is RAG?"}
})

# Override search parameters at run time without changing the component
result = pipe.run({
    "search": {"query": "What is RAG?", "top_k": 3, "country": "BE", "language": "en"},
    "prompt": {"query": "What is RAG?"}
})

print(result["prompt"]["prompt"])

With AsyncPipeline

from haystack import AsyncPipeline
from haystack_brave_search import BraveWebSearch

pipe = AsyncPipeline()
pipe.add_component("search", BraveWebSearch(api_key="your-api-key"))

Using a Proxy

Pass a proxy_url to the constructor to route requests through a proxy. When set, api_key is used as the Bearer token and the default X-Subscription-Token header is removed.

searcher = BraveWebSearch(api_key="your-token", proxy_url="https://your-proxy-url.com")

Parameters

Initialisation (__init__)

Run time (run)

Run-time values take precedence over the defaults set at initialisation. Omitting a run-time parameter falls back to the initialisation value.

Output

The run() method returns a dict with a documents key containing a list of Haystack Document objects:

• document.content — snippet/description of the result

• document.meta["title"] — page title

• document.meta["url"] — page URL

Environment Variable

You can avoid hardcoding the API key by using an environment variable:

import os
from haystack_brave_search import BraveWebSearch

searcher = BraveWebSearch(api_key=os.environ["BRAVE_API_KEY"])

License

MIT