searxng/searx/engines/duckduckgo_definitions.py

# SPDX-License-Identifier: AGPL-3.0-or-later
# lint: pylint
"""
DuckDuckGo Instant Answer API
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

The `DDG-API <https://duckduckgo.com/api>`__ is no longer documented but from
reverse engineering we can see that some services (e.g. instant answers) still
in use from the DDG search engine.

As far we can say the *instant answers* API does not support languages, or at
least we could not find out how language support should work.  It seems that
most of the features are based on English terms.

"""

from typing import TYPE_CHECKING

from urllib.parse import urlencode, urlparse, urljoin
from lxml import html

from searx.data import WIKIDATA_UNITS
from searx.utils import extract_text, html_to_text, get_string_replaces_function
from searx.external_urls import get_external_url, get_earth_coordinates_url, area_to_osm_zoom

if TYPE_CHECKING:
    import logging

    logger: logging.Logger

# about
about = {
    "website": 'https://duckduckgo.com/',
    "wikidata_id": 'Q12805',
    "official_api_documentation": 'https://duckduckgo.com/api',
    "use_official_api": True,
    "require_api_key": False,
    "results": 'JSON',
}

send_accept_language_header = True

URL = 'https://api.duckduckgo.com/' + '?{query}&format=json&pretty=0&no_redirect=1&d=1'

WIKIDATA_PREFIX = ['http://www.wikidata.org/entity/', 'https://www.wikidata.org/entity/']

replace_http_by_https = get_string_replaces_function({'http:': 'https:'})


def is_broken_text(text):
    """duckduckgo may return something like ``<a href="xxxx">http://somewhere Related website<a/>``

    The href URL is broken, the "Related website" may contains some HTML.

    The best solution seems to ignore these results.
    """
    return text.startswith('http') and ' ' in text


def result_to_text(text, htmlResult):
    # TODO : remove result ending with "Meaning" or "Category"  # pylint: disable=fixme
    result = None
    dom = html.fromstring(htmlResult)
    a = dom.xpath('//a')
    if len(a) >= 1:
        result = extract_text(a[0])
    else:
        result = text
    if not is_broken_text(result):
        return result
    return None


def request(query, params):
    params['url'] = URL.format(query=urlencode({'q': query}))
    return params


def response(resp):
    # pylint: disable=too-many-locals, too-many-branches, too-many-statements
    results = []

    search_res = resp.json()

    # search_res.get('Entity') possible values (not exhaustive) :
    # * continent / country / department / location / waterfall
    # * actor / musician / artist
    # * book / performing art / film / television  / media franchise / concert tour / playwright
    # * prepared food
    # * website / software / os / programming language / file format / software engineer
    # * company

    content = ''
    heading = search_res.get('Heading', '')
    attributes = []
    urls = []
    infobox_id = None
    relatedTopics = []

    # add answer if there is one
    answer = search_res.get('Answer', '')
    if answer:
        logger.debug('AnswerType="%s" Answer="%s"', search_res.get('AnswerType'), answer)
        if search_res.get('AnswerType') not in ['calc', 'ip']:
            results.append({'answer': html_to_text(answer)})

    # add infobox
    if 'Definition' in search_res:
        content = content + search_res.get('Definition', '')

    if 'Abstract' in search_res:
        content = content + search_res.get('Abstract', '')

    # image
    image = search_res.get('Image')
    image = None if image == '' else image
    if image is not None and urlparse(image).netloc == '':
        image = urljoin('https://duckduckgo.com', image)

    # urls
    # Official website, Wikipedia page
    for ddg_result in search_res.get('Results', []):
        firstURL = ddg_result.get('FirstURL')
        text = ddg_result.get('Text')
        if firstURL is not None and text is not None:
            urls.append({'title': text, 'url': firstURL})
            results.append({'title': heading, 'url': firstURL})

    # related topics
    for ddg_result in search_res.get('RelatedTopics', []):
        if 'FirstURL' in ddg_result:
            firstURL = ddg_result.get('FirstURL')
            text = ddg_result.get('Text')
            if not is_broken_text(text):
                suggestion = result_to_text(text, ddg_result.get('Result'))
                if suggestion != heading and suggestion is not None:
                    results.append({'suggestion': suggestion})
        elif 'Topics' in ddg_result:
            suggestions = []
            relatedTopics.append({'name': ddg_result.get('Name', ''), 'suggestions': suggestions})
            for topic_result in ddg_result.get('Topics', []):
                suggestion = result_to_text(topic_result.get('Text'), topic_result.get('Result'))
                if suggestion != heading and suggestion is not None:
                    suggestions.append(suggestion)

    # abstract
    abstractURL = search_res.get('AbstractURL', '')
    if abstractURL != '':
        # add as result ? problem always in english
        infobox_id = abstractURL
        urls.append({'title': search_res.get('AbstractSource'), 'url': abstractURL, 'official': True})
        results.append({'url': abstractURL, 'title': heading})

    # definition
    definitionURL = search_res.get('DefinitionURL', '')
    if definitionURL != '':
        # add as result ? as answer ? problem always in english
        infobox_id = definitionURL
        urls.append({'title': search_res.get('DefinitionSource'), 'url': definitionURL})

    # to merge with wikidata's infobox
    if infobox_id:
        infobox_id = replace_http_by_https(infobox_id)

    # attributes
    # some will be converted to urls
    if 'Infobox' in search_res:
        infobox = search_res.get('Infobox')
        if 'content' in infobox:
            osm_zoom = 17
            coordinates = None
            for info in infobox.get('content'):
                data_type = info.get('data_type')
                data_label = info.get('label')
                data_value = info.get('value')

                # Workaround: ddg may return a double quote
                if data_value == '""':
                    continue

                # Is it an external URL ?
                # * imdb_id / facebook_profile / youtube_channel / youtube_video / twitter_profile
                # * instagram_profile / rotten_tomatoes / spotify_artist_id / itunes_artist_id / soundcloud_id
                # * netflix_id
                external_url = get_external_url(data_type, data_value)
                if external_url is not None:
                    urls.append({'title': data_label, 'url': external_url})
                elif data_type in ['instance', 'wiki_maps_trigger', 'google_play_artist_id']:
                    # ignore instance: Wikidata value from "Instance Of" (Qxxxx)
                    # ignore wiki_maps_trigger: reference to a javascript
                    # ignore google_play_artist_id: service shutdown
                    pass
                elif data_type == 'string' and data_label == 'Website':
                    # There is already an URL for the website
                    pass
                elif data_type == 'area':
                    attributes.append({'label': data_label, 'value': area_to_str(data_value), 'entity': 'P2046'})
                    osm_zoom = area_to_osm_zoom(data_value.get('amount'))
                elif data_type == 'coordinates':
                    if data_value.get('globe') == 'http://www.wikidata.org/entity/Q2':
                        # coordinate on Earth
                        # get the zoom information from the area
                        coordinates = info
                    else:
                        # coordinate NOT on Earth
                        attributes.append({'label': data_label, 'value': data_value, 'entity': 'P625'})
                elif data_type == 'string':
                    attributes.append({'label': data_label, 'value': data_value})

            if coordinates:
                data_label = coordinates.get('label')
                data_value = coordinates.get('value')
                latitude = data_value.get('latitude')
                longitude = data_value.get('longitude')
                url = get_earth_coordinates_url(latitude, longitude, osm_zoom)
                urls.append({'title': 'OpenStreetMap', 'url': url, 'entity': 'P625'})

    if len(heading) > 0:
        # TODO get infobox.meta.value where .label='article_title'    # pylint: disable=fixme
        if image is None and len(attributes) == 0 and len(urls) == 1 and len(relatedTopics) == 0 and len(content) == 0:
            results.append({'url': urls[0]['url'], 'title': heading, 'content': content})
        else:
            results.append(
                {
                    'infobox': heading,
                    'id': infobox_id,
                    'content': content,
                    'img_src': image,
                    'attributes': attributes,
                    'urls': urls,
                    'relatedTopics': relatedTopics,
                }
            )

    return results


def unit_to_str(unit):
    for prefix in WIKIDATA_PREFIX:
        if unit.startswith(prefix):
            wikidata_entity = unit[len(prefix) :]
            return WIKIDATA_UNITS.get(wikidata_entity, unit)
    return unit


def area_to_str(area):
    """parse ``{'unit': 'https://www.wikidata.org/entity/Q712226', 'amount': '+20.99'}``"""
    unit = unit_to_str(area.get('unit'))
    if unit is not None:
        try:
            amount = float(area.get('amount'))
            return '{} {}'.format(amount, unit)
        except ValueError:
            pass
    return '{} {}'.format(area.get('amount', ''), area.get('unit', ''))