[mod] wikipedia & wikidata: upgrade to data_type: traits_v1

BTW this fix an issue in wikipedia: SearXNG's locales zh-TW and zh-HK are now
using language `zh-classical` from wikipedia (and not `zh`).

Signed-off-by: Markus Heiser <markus.heiser@darmarit.de>

docs/src/searx.engines.wikipedia.rst

@@ -0,0 +1,27 @@
+.. _wikimedia engines:
+
+=========
+Wikimedia
+=========
+
+.. contents:: Contents
+   :depth: 2
+   :local:
+   :backlinks: entry
+
+
+.. _wikipedia engine:
+
+Wikipedia
+=========
+
+.. automodule:: searx.engines.wikipedia
+  :members:
+
+.. _wikidata engine:
+
+Wikidata
+=========
+
+.. automodule:: searx.engines.wikidata
+  :members:

searx/autocomplete.py

@@ -143,14 +143,31 @@ def qwant(query, sxng_locale):
     return results
 
 
-def wikipedia(query, lang):
-    # wikipedia autocompleter
-    url = 'https://' + lang + '.wikipedia.org/w/api.php?action=opensearch&{0}&limit=10&namespace=0&format=json'
+def wikipedia(query, sxng_locale):
+    """Autocomplete from Wikipedia. Supports Wikipedia's languages (aka netloc)."""
+    results = []
+    eng_traits = engines['wikipedia'].traits
+    wiki_lang = eng_traits.get_language(sxng_locale, 'en')
+    wiki_netloc = eng_traits.custom['wiki_netloc'].get(wiki_lang, 'en.wikipedia.org')
+
+    url = 'https://{wiki_netloc}/w/api.php?{args}'
+    args = urlencode(
+        {
+            'action': 'opensearch',
+            'format': 'json',
+            'formatversion': '2',
+            'search': query,
+            'namespace': '0',
+            'limit': '10',
+        }
+    )
+    resp = get(url.format(args=args, wiki_netloc=wiki_netloc))
+    if resp.ok:
+        data = resp.json()
+        if len(data) > 1:
+            results = data[1]
 
-    resp = loads(get(url.format(urlencode(dict(search=query)))).text)
-    if len(resp) > 1:
-        return resp[1]
-    return []
+    return results
 
 
 def yandex(query, _lang):

searx/engines/wikidata.py

@@ -1,9 +1,12 @@
 # SPDX-License-Identifier: AGPL-3.0-or-later
 # lint: pylint
-"""Wikidata
+"""This module implements the Wikidata engine.  Some implementations are shared
+from :ref:`wikipedia engine`.
+
 """
 # pylint: disable=missing-class-docstring
 
+from typing import TYPE_CHECKING
 from hashlib import md5
 from urllib.parse import urlencode, unquote
 from json import loads
@@ -13,13 +16,17 @@ from babel.dates import format_datetime, format_date, format_time, get_datetime_
 
 from searx.data import WIKIDATA_UNITS
 from searx.network import post, get
-from searx.utils import match_language, searx_useragent, get_string_replaces_function
+from searx.utils import searx_useragent, get_string_replaces_function
 from searx.external_urls import get_external_url, get_earth_coordinates_url, area_to_osm_zoom
-from searx.engines.wikipedia import (  # pylint: disable=unused-import
-    fetch_traits,
-    _fetch_supported_languages,
-    supported_languages_url,
-)
+from searx.engines.wikipedia import fetch_traits as _fetch_traits
+from searx.enginelib.traits import EngineTraits
+
+if TYPE_CHECKING:
+    import logging
+
+    logger: logging.Logger
+
+traits: EngineTraits
 
 # about
 about = {
@@ -155,33 +162,35 @@ def send_wikidata_query(query, method='GET'):
 
 
 def request(query, params):
-    language = params['language'].split('-')[0]
-    if language == 'all':
-        language = 'en'
-    else:
-        language = match_language(params['language'], supported_languages, language_aliases).split('-')[0]
+
+    # wikidata does not support zh-classical (zh_Hans) / zh-TW, zh-HK and zh-CN
+    # mapped to zh
+    sxng_lang = params['searxng_locale'].split('-')[0]
+    language = traits.get_language(sxng_lang, 'en')
 
     query, attributes = get_query(query, language)
+    logger.debug("request --> language %s // len(attributes): %s", language, len(attributes))
 
     params['method'] = 'POST'
     params['url'] = SPARQL_ENDPOINT_URL
     params['data'] = {'query': query}
     params['headers'] = get_headers()
-
     params['language'] = language
     params['attributes'] = attributes
+
     return params
 
 
 def response(resp):
+
     results = []
     jsonresponse = loads(resp.content.decode())
 
-    language = resp.search_params['language'].lower()
+    language = resp.search_params['language']
     attributes = resp.search_params['attributes']
+    logger.debug("request --> language %s // len(attributes): %s", language, len(attributes))
 
     seen_entities = set()
-
     for result in jsonresponse.get('results', {}).get('bindings', []):
         attribute_result = {key: value['value'] for key, value in result.items()}
         entity_url = attribute_result['item']
@@ -757,3 +766,15 @@ def init(engine_settings=None):  # pylint: disable=unused-argument
         lang = result['name']['xml:lang']
         entity_id = result['item']['value'].replace('http://www.wikidata.org/entity/', '')
         WIKIDATA_PROPERTIES[(entity_id, lang)] = name.capitalize()
+
+
+def fetch_traits(engine_traits: EngineTraits):
+    """Use languages evaluated from :py:obj:`wikipedia.fetch_traits
+    <searx.engines.wikipedia.fetch_traits>` except zh-classical (zh_Hans) what
+    is not supported by wikidata."""
+
+    _fetch_traits(engine_traits)
+    # wikidata does not support zh-classical (zh_Hans)
+    engine_traits.languages.pop('zh_Hans')
+    # wikidata does not have net-locations for the languages
+    engine_traits.custom['wiki_netloc'] = {}

searx/engines/wikipedia.py

@@ -1,16 +1,26 @@
 # SPDX-License-Identifier: AGPL-3.0-or-later
-"""
- Wikipedia (Web)
+# lint: pylint
+"""This module implements the Wikipedia engine.  Some of this implementations
+are shared by other engines:
+
+- :ref:`wikidata engine`
+
+The list of supported languages is fetched from the article linked by
+:py:obj:`wikipedia_article_depth`.  Unlike traditional search engines, wikipedia
+does not support one Wikipedia for all the languages, but there is one Wikipedia
+for every language (:py:obj:`fetch_traits`).
 """
 
-from urllib.parse import quote
-from json import loads
+import urllib.parse
+import babel
+
 from lxml import html
-from searx.utils import match_language, searx_useragent
+
 from searx import network
+from searx.locales import language_tag
 from searx.enginelib.traits import EngineTraits
 
-engine_traits: EngineTraits
+traits: EngineTraits
 
 # about
 about = {
@@ -22,32 +32,40 @@ about = {
     "results": 'JSON',
 }
 
-
 send_accept_language_header = True
 
-# search-url
-search_url = 'https://{language}.wikipedia.org/api/rest_v1/page/summary/{title}'
-supported_languages_url = 'https://meta.wikimedia.org/wiki/List_of_Wikipedias'
-language_variants = {"zh": ("zh-cn", "zh-hk", "zh-mo", "zh-my", "zh-sg", "zh-tw")}
+wikipedia_article_depth = 'https://meta.wikimedia.org/wiki/Wikipedia_article_depth'
+"""The *editing depth* of Wikipedia is one of several possible rough indicators
+of the encyclopedia's collaborative quality, showing how frequently its articles
+are updated.  The measurement of depth was introduced after some limitations of
+the classic measurement of article count were realized.
+"""
+
+# example: https://zh-classical.wikipedia.org/api/rest_v1/page/summary/日
+rest_v1_summary_url = 'https://{wiki_netloc}/api/rest_v1/page/summary/{title}'
+"""`wikipedia rest_v1 summary API`_: The summary response includes an extract of
+the first paragraph of the page in plain text and HTML as well as the type of
+page. This is useful for page previews (fka. Hovercards, aka. Popups) on the web
+and link previews in the apps.
 
+.. _wikipedia rest_v1 summary API: https://en.wikipedia.org/api/rest_v1/#/Page%20content/get_page_summary__title_
 
-# set language in base_url
-def url_lang(lang):
-    lang_pre = lang.split('-')[0]
-    if lang_pre == 'all' or lang_pre not in supported_languages and lang_pre not in language_aliases:
-        return 'en'
-    return match_language(lang, supported_languages, language_aliases).split('-')[0]
+"""
 
 
-# do search-request
 def request(query, params):
+    """Assemble a request (`wikipedia rest_v1 summary API`_)."""
     if query.islower():
         query = query.title()
 
-    language = url_lang(params['language'])
-    params['url'] = search_url.format(title=quote(query), language=language)
+    engine_language = traits.get_language(params['searxng_locale'], 'en')
+    wiki_netloc = traits.custom['wiki_netloc'].get(engine_language, 'https://en.wikipedia.org/wiki/')
+    title = urllib.parse.quote(query)
+
+    # '!wikipedia 日 :zh-TW' --> https://zh-classical.wikipedia.org/
+    # '!wikipedia 日 :zh' --> https://zh.wikipedia.org/
+    params['url'] = rest_v1_summary_url.format(wiki_netloc=wiki_netloc, title=title)
 
-    params['headers']['User-Agent'] = searx_useragent()
     params['raise_for_httperror'] = False
     params['soft_max_redirects'] = 2
 
@@ -56,13 +74,14 @@ def request(query, params):
 
 # get response from search-request
 def response(resp):
+
+    results = []
     if resp.status_code == 404:
         return []
-
     if resp.status_code == 400:
         try:
-            api_result = loads(resp.text)
-        except:
+            api_result = resp.json()
+        except Exception:  # pylint: disable=broad-except
             pass
         else:
             if (
@@ -73,52 +92,25 @@ def response(resp):
 
     network.raise_for_httperror(resp)
 
-    results = []
-    api_result = loads(resp.text)
-
-    # skip disambiguation pages
-    if api_result.get('type') != 'standard':
-        return []
-
+    api_result = resp.json()
     title = api_result['title']
     wikipedia_link = api_result['content_urls']['desktop']['page']
-
-    results.append({'url': wikipedia_link, 'title': title})
-
-    results.append(
-        {
-            'infobox': title,
-            'id': wikipedia_link,
-            'content': api_result.get('extract', ''),
-            'img_src': api_result.get('thumbnail', {}).get('source'),
-            'urls': [{'title': 'Wikipedia', 'url': wikipedia_link}],
-        }
-    )
+    results.append({'url': wikipedia_link, 'title': title, 'content': api_result.get('description', '')})
+
+    if api_result.get('type') == 'standard':
+        results.append(
+            {
+                'infobox': title,
+                'id': wikipedia_link,
+                'content': api_result.get('extract', ''),
+                'img_src': api_result.get('thumbnail', {}).get('source'),
+                'urls': [{'title': 'Wikipedia', 'url': wikipedia_link}],
+            }
+        )
 
     return results
 
 
-# get supported languages from their site
-def _fetch_supported_languages(resp):
-    supported_languages = {}
-    dom = html.fromstring(resp.text)
-    tables = dom.xpath('//table[contains(@class,"sortable")]')
-    for table in tables:
-        # exclude header row
-        trs = table.xpath('.//tr')[1:]
-        for tr in trs:
-            td = tr.xpath('./td')
-            code = td[3].xpath('./a')[0].text
-            name = td[1].xpath('./a')[0].text
-            english_name = td[1].xpath('./a')[0].text
-            articles = int(td[4].xpath('./a')[0].text.replace(',', ''))
-            # exclude languages with too few articles
-            if articles >= 100:
-                supported_languages[code] = {"name": name, "english_name": english_name}
-
-    return supported_languages
-
-
 # Nonstandard language codes
 #
 # These Wikipedias use language codes that do not conform to the ISO 639
@@ -135,104 +127,57 @@ lang_map = {
     'nrm': 'nrf',
     'roa-rup': 'rup',
     'nds-nl': 'nds',
-    #'roa-tara: – invented code used for the Tarantino Wikipedia (again, roa is the standard code for the large family of Romance languages that the Tarantino dialect falls within)
     #'simple: – invented code used for the Simple English Wikipedia (not the official IETF code en-simple)
-    'zh-classical': 'zh_Hant',
     'zh-min-nan': 'nan',
     'zh-yue': 'yue',
     'an': 'arg',
+    'zh-classical': 'zh-Hant',  # babel maps classical to zh-Hans (for whatever reason)
 }
 
 unknown_langs = [
-    'ab',  # Abkhazian
-    'alt',  # Southern Altai
     'an',  # Aragonese
-    'ang',  # Anglo-Saxon
-    'arc',  # Aramaic
-    'ary',  # Moroccan Arabic
-    'av',  # Avar
     'ba',  # Bashkir
-    'be-tarask',
     'bar',  # Bavarian
     'bcl',  # Central Bicolano
-    'bh',  # Bhojpuri
-    'bi',  # Bislama
-    'bjn',  # Banjar
-    'blk',  # Pa'O
-    'bpy',  # Bishnupriya Manipuri
-    'bxr',  # Buryat
-    'cbk-zam',  # Zamboanga Chavacano
-    'co',  # Corsican
-    'cu',  # Old Church Slavonic
-    'dty',  # Doteli
-    'dv',  # Divehi
-    'ext',  # Extremaduran
-    'fj',  # Fijian
-    'frp',  # Franco-Provençal
-    'gan',  # Gan
-    'gom',  # Goan Konkani
+    'be-tarask',  # Belarusian variant / Belarusian is already covered by 'be'
+    'bpy',  # Bishnupriya Manipuri is unknown by babel
     'hif',  # Fiji Hindi
     'ilo',  # Ilokano
-    'inh',  # Ingush
-    'jbo',  # Lojban
-    'kaa',  # Karakalpak
-    'kbd',  # Kabardian Circassian
-    'kg',  # Kongo
-    'koi',  # Komi-Permyak
-    'krc',  # Karachay-Balkar
-    'kv',  # Komi
-    'lad',  # Ladino
-    'lbe',  # Lak
-    'lez',  # Lezgian
     'li',  # Limburgish
-    'ltg',  # Latgalian
-    'mdf',  # Moksha
-    'mnw',  # Mon
-    'mwl',  # Mirandese
-    'myv',  # Erzya
-    'na',  # Nauruan
-    'nah',  # Nahuatl
-    'nov',  # Novial
-    'nrm',  # Norman
-    'pag',  # Pangasinan
-    'pam',  # Kapampangan
-    'pap',  # Papiamentu
-    'pdc',  # Pennsylvania German
-    'pfl',  # Palatinate German
-    'roa-rup',  # Aromanian
-    'sco',  # Scots
-    'sco',  # Scots (https://sco.wikipedia.org) is not known by babel, Scottish Gaelic (https://gd.wikipedia.org) is known by babel
+    'sco',  # Scots (sco) is not known by babel, Scottish Gaelic (gd) is known by babel
     'sh',  # Serbo-Croatian
     'simple',  # simple english is not know as a natural language different to english (babel)
-    'sm',  # Samoan
-    'srn',  # Sranan
-    'stq',  # Saterland Frisian
-    'szy',  # Sakizaya
-    'tcy',  # Tulu
-    'tet',  # Tetum
-    'tpi',  # Tok Pisin
-    'trv',  # Seediq
-    'ty',  # Tahitian
-    'tyv',  # Tuvan
-    'udm',  # Udmurt
-    'vep',  # Vepsian
-    'vls',  # West Flemish
     'vo',  # Volapük
     'wa',  # Walloon
-    'xal',  # Kalmyk
 ]
 
 
 def fetch_traits(engine_traits: EngineTraits):
-    """Fetch languages from Wikipedia"""
-    # pylint: disable=import-outside-toplevel
+    """Fetch languages from Wikipedia.
+
+    The location of the Wikipedia address of a language is mapped in a
+    :py:obj:`custom field <searx.enginelib.traits.EngineTraits.custom>`
+    (``wiki_netloc``).  Here is a reduced example:
+
+    .. code:: python
 
-    engine_traits.data_type = 'supported_languages'  # deprecated
+       traits.custom['wiki_netloc'] = {
+           "en": "en.wikipedia.org",
+           ..
+           "gsw": "als.wikipedia.org",
+           ..
+           "zh": "zh.wikipedia.org",
+           "zh-classical": "zh-classical.wikipedia.org"
+       }
 
-    import babel
-    from searx.locales import language_tag
+    """
 
-    resp = network.get('https://meta.wikimedia.org/wiki/List_of_Wikipedias')
+    engine_traits.custom['wiki_netloc'] = {}
+
+    # insert alias to map from a region like zh-CN to a language zh_Hans
+    engine_traits.languages['zh_Hans'] = 'zh'
+
+    resp = network.get(wikipedia_article_depth)
     if not resp.ok:
         print("ERROR: response from Wikipedia is not OK.")
 
@@ -242,34 +187,31 @@ def fetch_traits(engine_traits: EngineTraits):
         cols = row.xpath('./td')
         if not cols:
             continue
-
         cols = [c.text_content().strip() for c in cols]
-        articles = int(cols[4].replace(',', '').replace('-', '0'))
-        users = int(cols[8].replace(',', '').replace('-', '0'))
-        depth = cols[11].strip('-')
 
-        if articles < 1000:
+        depth = float(cols[3].replace('-', '0').replace(',', ''))
+        articles = int(cols[4].replace(',', '').replace(',', ''))
+
+        if articles < 10000:
             # exclude languages with too few articles
             continue
 
-        # depth: rough indicator of a Wikipedia’s quality, showing how
-        #        frequently its articles are updated.
-        if depth == '':
-            if users < 1000:
-                # depth is not calculated --> at least 1000 user should registered
-                continue
-        elif int(depth) < 20:
+        if int(depth) < 20:
+            # Rough indicator of a Wikipedia’s quality, showing how frequently
+            # its articles are updated.
             continue
 
-        eng_tag = cols[3]
+        eng_tag = cols[2]
+        wiki_url = row.xpath('./td[3]/a/@href')[0]
+        wiki_url = urllib.parse.urlparse(wiki_url)
 
         if eng_tag in unknown_langs:
             continue
 
         try:
-            sxng_tag = language_tag(babel.Locale.parse(lang_map.get(eng_tag, eng_tag)))
+            sxng_tag = language_tag(babel.Locale.parse(lang_map.get(eng_tag, eng_tag), sep='-'))
         except babel.UnknownLocaleError:
-            print("ERROR: %s -> %s is unknown by babel" % (cols[1], eng_tag))
+            print("ERROR: %s [%s] is unknown by babel" % (cols[0], eng_tag))
             continue
 
         conflict = engine_traits.languages.get(sxng_tag)
@@ -277,6 +219,6 @@ def fetch_traits(engine_traits: EngineTraits):
             if conflict != eng_tag:
                 print("CONFLICT: babel %s --> %s, %s" % (sxng_tag, conflict, eng_tag))
             continue
-        engine_traits.languages[sxng_tag] = eng_tag
 
-    engine_traits.languages['zh_Hans'] = 'zh'
+        engine_traits.languages[sxng_tag] = eng_tag
+        engine_traits.custom['wiki_netloc'][eng_tag] = wiki_url.netloc