Files
searxng/searx/engines/exaapi.py

169 lines
4.7 KiB
Python

# SPDX-License-Identifier: AGPL-3.0-or-later
"""Engine to search using the official `Exa Search API`_. Exa is a search engine for AI agents.
.. _Exa Search API: https://exa.ai/docs/reference/search
Configuration
=============
The engine has the following mandatory setting:
- :py:obj:`api_key`
You can obtain an API key from the `API Key section <https://dashboard.exa.ai/api-keys>`_ in the Exa dashboard.
Optional settings are:
- :py:obj:`results_per_page`
- :py:obj:`search_type`
- :py:obj:`content_mode`
- :py:obj:`content_max_characters`
.. code:: yaml
- name: exaapi
engine: exaapi
shortcut: exa
api_key: "..."
results_per_page: 10
search_type: auto
content_mode: highlights
inactive: false
The API supports SafeSearch and region-aware results.
"""
import typing as t
from dateutil import parser
from searx.exceptions import SearxEngineAPIException
from searx.result_types import EngineResults
from searx.utils import html_to_text
if t.TYPE_CHECKING:
from searx.extended_types import SXNG_Response
from searx.search.processors import OnlineParams
SearchType = t.Literal["fast", "auto", "instant", "deep", "deep-lite", "deep-reasoning"]
ContentMode = t.Literal["highlights", "text"]
about = {
"website": "https://exa.ai",
"wikidata_id": None,
"official_api_documentation": "https://exa.ai/docs/reference/search",
"use_official_api": True,
"require_api_key": True,
"results": "JSON",
}
api_key: str = ""
"""API key for Exa Search API (required)."""
categories = ["general", "web"]
safesearch = True
base_url = "https://api.exa.ai/search"
results_per_page: int = 10
"""Maximum number of results per request. Value must be between 1 and 100, default is 10."""
search_type: SearchType = "auto"
"""Search type. Default is auto, see documentation for more information."""
content_mode: ContentMode = "highlights"
"""Content to request from the API: ``highlights`` (excerpts) or ``text`` (page text)."""
content_max_characters: int = 500
"""Maximum characters for the requested content."""
def init(_):
if not api_key:
raise SearxEngineAPIException("No API key provided")
if not 1 <= results_per_page <= 100:
raise ValueError("results_per_page must be between 1 and 100")
if search_type not in t.get_args(SearchType):
raise ValueError(f"Unsupported search type: {search_type}")
if content_mode not in t.get_args(ContentMode):
raise ValueError(f"Unsupported content mode: {content_mode}")
if content_max_characters < 1:
raise ValueError("content_max_characters must be at least 1")
def _contents_payload() -> dict[str, t.Any]:
if content_mode == "text":
return {"text": {"maxCharacters": content_max_characters, "stripLinks": True}}
return {"highlights": {"maxCharacters": content_max_characters}}
def _extract_content(result: dict[str, t.Any]) -> str:
if content_mode == "text":
return html_to_text(result.get("text") or "")
return html_to_text(" ".join(result.get("highlights") or []))
def request(query: str, params: "OnlineParams") -> None:
"""Create the API request."""
body: dict[str, t.Any] = {
"query": query,
"type": search_type,
"numResults": results_per_page,
"contents": _contents_payload(),
}
# Apply SafeSearch if enabled
if params["safesearch"]:
body["moderation"] = True
# Apply region-aware results if specified
locale_parts = params["searxng_locale"].split("-")
region = locale_parts[-1]
if len(locale_parts) > 1:
body["userLocation"] = region.upper()
params["url"] = base_url
params["method"] = "POST"
params["headers"]["x-api-key"] = api_key
params["json"] = body
def _extract_published_date(value: str | None):
"""Extract and parse the published date from the API response.
Args:
value: Raw date string from the API
Returns:
Parsed datetime object or None if parsing fails
"""
if not value:
return None
try:
return parser.parse(value)
except (parser.ParserError, TypeError, OverflowError):
return None
def response(resp: "SXNG_Response") -> EngineResults:
"""Process the API response and return results."""
res = EngineResults()
for result in resp.json().get("results", []):
url = result.get("url")
if not url:
continue
res.add(
res.types.MainResult(
url=url,
title=html_to_text(result.get("title") or url),
content=_extract_content(result),
thumbnail=result.get("image") or "",
publishedDate=_extract_published_date(result.get("publishedDate")),
author=result.get("author") or "",
)
)
return res