Ponysearch/searx/engines/derpibooru.py

63 lines
1.8 KiB
Python
Raw Normal View History

2022-07-14 12:29:41 +02:00
# SPDX-License-Identifier: AGPL-3.0-or-later
# lint: pylint
"""
Derpibooru (Images)
"""
2022-07-15 09:22:21 +02:00
from urllib.parse import urlencode, urlparse, urlunparse, parse_qsl
2022-07-14 12:29:41 +02:00
from json import loads
from searx import logger
2022-07-14 12:29:41 +02:00
# about
about = {
"website": 'https://derpibooru.org/',
"wikidata_id": 'Q28233552',
"official_api_documentation": 'https://derpibooru.org/pages/api/',
2022-07-15 08:08:47 +02:00
"use_official_api": True,
2022-07-14 12:29:41 +02:00
"require_api_key": False,
"results": 'JSON',
}
base_url = 'https://derpibooru.org/'
search_url = base_url + 'api/v1/json/search/images?'
categories = ['images']
page_size = 20
paging = True
filter_id = 214606
sort_field = "wilson_score"
sort_direction = "desc"
2022-07-14 12:29:41 +02:00
2022-07-15 12:19:16 +02:00
def clean_url(url):
parsed = urlparse(url)
query = [(k, v) for (k, v) in parse_qsl(parsed.query) if k not in ['ixid', 's']]
return urlunparse((parsed.scheme, parsed.netloc, parsed.path, parsed.params, urlencode(query), parsed.fragment))
2022-07-14 12:29:41 +02:00
def request(query, params):
2022-07-14 13:53:55 +02:00
params['url'] = search_url + urlencode(
{'q': query, 'filter_id': filter_id, 'page': params['pageno'], 'per_page': page_size, 'sf': sort_field, 'sd': sort_direction}
2022-07-14 14:14:58 +02:00
)
2022-07-14 12:29:41 +02:00
logger.debug("query_url --> %s", params['url'])
return params
def response(resp):
results = []
json_data = loads(resp.text)
2022-07-15 08:49:04 +02:00
if 'images' in json_data:
2022-07-14 12:29:41 +02:00
for result in json_data['images']:
results.append(
{
'template': 'images.html',
'url': base_url + 'images/' + str(result.get('id')),
2022-07-15 12:19:16 +02:00
'thumbnail_src': clean_url(result['representations']['thumb']),
'img_src': clean_url(result['representations']['full']),
2022-07-14 12:29:41 +02:00
'title': result.get('name') or 'unknown',
'content': result.get('description') or '',
}
)
2022-07-15 08:39:42 +02:00
2022-07-14 12:29:41 +02:00
return results