From 9c4e9d381462c16e93ef74fd5133512f720a77e5 Mon Sep 17 00:00:00 2001 From: Bnyro Date: Thu, 3 Aug 2023 18:33:10 +0200 Subject: [PATCH] [feat] implementation of Wikimedia commons for images --- searx/engines/wikicommons.py | 71 ++++++++++++++++++++++++++++++++++++ searx/settings.yml | 6 +++ 2 files changed, 77 insertions(+) create mode 100644 searx/engines/wikicommons.py diff --git a/searx/engines/wikicommons.py b/searx/engines/wikicommons.py new file mode 100644 index 000000000..47b2a5bea --- /dev/null +++ b/searx/engines/wikicommons.py @@ -0,0 +1,71 @@ +# SPDX-License-Identifier: AGPL-3.0-or-later +# lint: pylint +"""Wikimedia Commons (images) + +""" + +from urllib.parse import urlencode + +# about +about = { + "website": 'https://commons.wikimedia.org/', + "wikidata_id": 'Q565', + "official_api_documentation": 'https://commons.wikimedia.org/w/api.php', + "use_official_api": True, + "require_api_key": False, + "results": 'JSON', +} + +base_url = "https://commons.wikimedia.org" +search_prefix = ( + '?action=query' + '&format=json' + '&generator=search' + '&gsrnamespace=6' + '&gsrprop=snippet' + '&prop=info|imageinfo' + '&iiprop=url|size|mime' + '&iiurlheight=180' # needed for the thumb url +) +paging = True +number_of_results = 10 + + +def request(query, params): + language = 'en' + if params['language'] != 'all': + language = params['language'].split('-')[0] + + args = { + 'uselang': language, + 'gsrlimit': number_of_results, + 'gsroffset': number_of_results * (params["pageno"] - 1), + 'gsrsearch': "filetype:bitmap|drawing " + query, + } + + params["url"] = f"{base_url}/w/api.php{search_prefix}&{urlencode(args)}" + return params + + +def response(resp): + results = [] + json = resp.json() + + if not json.get("query", {}).get("pages"): + return results + + for item in json["query"]["pages"].values(): + imageinfo = item["imageinfo"][0] + title = item["title"].replace("File:", "").rsplit('.', 1)[0] + result = { + 'url': imageinfo["descriptionurl"], + 'title': title, + 'content': item["snippet"], + 'img_src': imageinfo["url"], + 'img_format': f'{imageinfo["width"]} x {imageinfo["height"]}', + 'thumbnail_src': imageinfo["thumburl"], + 'template': 'images.html', + } + results.append(result) + + return results diff --git a/searx/settings.yml b/searx/settings.yml index 9f53c4d0f..85434370d 100644 --- a/searx/settings.yml +++ b/searx/settings.yml @@ -1640,6 +1640,12 @@ engines: website: https://www.wikivoyage.org/ wikidata_id: Q373 + - name: wikicommons.images + engine: wikicommons + shortcut: wc + categories: images + number_of_results: 10 + - name: wolframalpha shortcut: wa # You can use the engine using the official stable API, but you need an API