From 410c2f903ddf601d75cb3d917e60374b1ebb8667 Mon Sep 17 00:00:00 2001 From: Markus Heiser Date: Tue, 7 Jul 2020 21:50:59 +0200 Subject: [PATCH] [fix] revise google engine this commit is picked from #1985 --- searx/engines/google.py | 518 ++++++++++++++++------------------------ 1 file changed, 207 insertions(+), 311 deletions(-) diff --git a/searx/engines/google.py b/searx/engines/google.py index eed3a044e..24d46b97c 100644 --- a/searx/engines/google.py +++ b/searx/engines/google.py @@ -1,210 +1,207 @@ -# Google (Web) -# -# @website https://www.google.com -# @provide-api yes (https://developers.google.com/custom-search/) -# -# @using-api no -# @results HTML -# @stable no (HTML can change) -# @parse url, title, content, suggestion +# SPDX-License-Identifier: AGPL-3.0-or-later +"""Google (Web) -import re +:website: https://www.google.com +:provide-api: yes (https://developers.google.com/custom-search/) +:using-api: not the offical, since it needs registration to another service +:results: HTML +:stable: no +:parse: url, title, content, number_of_results, answer, suggestion, correction + +For detailed description of the *REST-full* API see: `Query Parameter +Definitions`_. + +.. _Query Parameter Definitions: + https://developers.google.com/custom-search/docs/xml_results#WebSearch_Query_Parameter_Definitions + +""" + +# pylint: disable=invalid-name, missing-function-docstring + +from lxml import html from flask_babel import gettext -from lxml import html, etree -from searx.engines.xpath import extract_text, extract_url +from searx.engines.xpath import extract_text from searx import logger -from searx.url_utils import urlencode, urlparse, parse_qsl +from searx.url_utils import urlencode, urlparse from searx.utils import match_language, eval_xpath logger = logger.getChild('google engine') - # engine dependent config categories = ['general'] paging = True language_support = True -use_locale_domain = True time_range_support = True - -# based on https://en.wikipedia.org/wiki/List_of_Google_domains and tests -default_hostname = 'www.google.com' - -country_to_hostname = { - 'BG': 'www.google.bg', # Bulgaria - 'CZ': 'www.google.cz', # Czech Republic - 'DE': 'www.google.de', # Germany - 'DK': 'www.google.dk', # Denmark - 'AT': 'www.google.at', # Austria - 'CH': 'www.google.ch', # Switzerland - 'GR': 'www.google.gr', # Greece - 'AU': 'www.google.com.au', # Australia - 'CA': 'www.google.ca', # Canada - 'GB': 'www.google.co.uk', # United Kingdom - 'ID': 'www.google.co.id', # Indonesia - 'IE': 'www.google.ie', # Ireland - 'IN': 'www.google.co.in', # India - 'MY': 'www.google.com.my', # Malaysia - 'NZ': 'www.google.co.nz', # New Zealand - 'PH': 'www.google.com.ph', # Philippines - 'SG': 'www.google.com.sg', # Singapore - # 'US': 'www.google.us', # United States, redirect to .com - 'ZA': 'www.google.co.za', # South Africa - 'AR': 'www.google.com.ar', # Argentina - 'CL': 'www.google.cl', # Chile - 'ES': 'www.google.es', # Spain - 'MX': 'www.google.com.mx', # Mexico - 'EE': 'www.google.ee', # Estonia - 'FI': 'www.google.fi', # Finland - 'BE': 'www.google.be', # Belgium - 'FR': 'www.google.fr', # France - 'IL': 'www.google.co.il', # Israel - 'HR': 'www.google.hr', # Croatia - 'HU': 'www.google.hu', # Hungary - 'IT': 'www.google.it', # Italy - 'JP': 'www.google.co.jp', # Japan - 'KR': 'www.google.co.kr', # South Korea - 'LT': 'www.google.lt', # Lithuania - 'LV': 'www.google.lv', # Latvia - 'NO': 'www.google.no', # Norway - 'NL': 'www.google.nl', # Netherlands - 'PL': 'www.google.pl', # Poland - 'BR': 'www.google.com.br', # Brazil - 'PT': 'www.google.pt', # Portugal - 'RO': 'www.google.ro', # Romania - 'RU': 'www.google.ru', # Russia - 'SK': 'www.google.sk', # Slovakia - 'SI': 'www.google.si', # Slovenia - 'SE': 'www.google.se', # Sweden - 'TH': 'www.google.co.th', # Thailand - 'TR': 'www.google.com.tr', # Turkey - 'UA': 'www.google.com.ua', # Ukraine - # 'CN': 'www.google.cn', # China, only from China ? - 'HK': 'www.google.com.hk', # Hong Kong - 'TW': 'www.google.com.tw' # Taiwan -} - -# osm -url_map = 'https://www.openstreetmap.org/'\ - + '?lat={latitude}&lon={longitude}&zoom={zoom}&layers=M' - -# search-url -search_path = '/search' -search_url = ('https://{hostname}' + - search_path + - '?{query}&start={offset}&gws_rd=cr&gbv=1&lr={lang}&hl={lang_short}&ei=x') - -time_range_search = "&tbs=qdr:{range}" -time_range_dict = {'day': 'd', - 'week': 'w', - 'month': 'm', - 'year': 'y'} - -# other URLs -map_hostname_start = 'maps.google.' -maps_path = '/maps' -redirect_path = '/url' -images_path = '/images' +safesearch = True supported_languages_url = 'https://www.google.com/preferences?#languages' +# based on https://en.wikipedia.org/wiki/List_of_Google_domains and tests +google_domains = { + 'BG': 'google.bg', # Bulgaria + 'CZ': 'google.cz', # Czech Republic + 'DE': 'google.de', # Germany + 'DK': 'google.dk', # Denmark + 'AT': 'google.at', # Austria + 'CH': 'google.ch', # Switzerland + 'GR': 'google.gr', # Greece + 'AU': 'google.com.au', # Australia + 'CA': 'google.ca', # Canada + 'GB': 'google.co.uk', # United Kingdom + 'ID': 'google.co.id', # Indonesia + 'IE': 'google.ie', # Ireland + 'IN': 'google.co.in', # India + 'MY': 'google.com.my', # Malaysia + 'NZ': 'google.co.nz', # New Zealand + 'PH': 'google.com.ph', # Philippines + 'SG': 'google.com.sg', # Singapore + # 'US': 'google.us', # United States, redirect to .com + 'ZA': 'google.co.za', # South Africa + 'AR': 'google.com.ar', # Argentina + 'CL': 'google.cl', # Chile + 'ES': 'google.es', # Spain + 'MX': 'google.com.mx', # Mexico + 'EE': 'google.ee', # Estonia + 'FI': 'google.fi', # Finland + 'BE': 'google.be', # Belgium + 'FR': 'google.fr', # France + 'IL': 'google.co.il', # Israel + 'HR': 'google.hr', # Croatia + 'HU': 'google.hu', # Hungary + 'IT': 'google.it', # Italy + 'JP': 'google.co.jp', # Japan + 'KR': 'google.co.kr', # South Korea + 'LT': 'google.lt', # Lithuania + 'LV': 'google.lv', # Latvia + 'NO': 'google.no', # Norway + 'NL': 'google.nl', # Netherlands + 'PL': 'google.pl', # Poland + 'BR': 'google.com.br', # Brazil + 'PT': 'google.pt', # Portugal + 'RO': 'google.ro', # Romania + 'RU': 'google.ru', # Russia + 'SK': 'google.sk', # Slovakia + 'SI': 'google.si', # Slovenia + 'SE': 'google.se', # Sweden + 'TH': 'google.co.th', # Thailand + 'TR': 'google.com.tr', # Turkey + 'UA': 'google.com.ua', # Ukraine + # 'CN': 'google.cn', # China, only from China ? + 'HK': 'google.com.hk', # Hong Kong + 'TW': 'google.com.tw' # Taiwan +} + +time_range_dict = { + 'day': 'd', + 'week': 'w', + 'month': 'm', + 'year': 'y' +} + +# Filter results. 0: None, 1: Moderate, 2: Strict +filter_mapping = { + 0 : 'off', + 1 : 'medium', + 2 : 'high' +} + # specific xpath variables -results_xpath = '//div[contains(@class, "ZINbbc")]' -url_xpath = './/div[@class="kCrYT"][1]/a/@href' -title_xpath = './/div[@class="kCrYT"][1]/a/div[1]' -content_xpath = './/div[@class="kCrYT"][2]//div[contains(@class, "BNeawe")]//div[contains(@class, "BNeawe")]' -suggestion_xpath = '//div[contains(@class, "ZINbbc")][last()]//div[@class="rVLSBd"]/a//div[contains(@class, "BNeawe")]' -spelling_suggestion_xpath = '//div[@id="scc"]//a' +# ------------------------ -# map : detail location -map_address_xpath = './/div[@class="s"]//table//td[2]/span/text()' -map_phone_xpath = './/div[@class="s"]//table//td[2]/span/span' -map_website_url_xpath = 'h3[2]/a/@href' -map_website_title_xpath = 'h3[2]' +# google results are grouped into
+results_xpath = '//div[@class="g"]' -# map : near the location -map_near = 'table[@class="ts"]//tr' -map_near_title = './/h4' -map_near_url = './/h4/a/@href' -map_near_phone = './/span[@class="nobr"]' +# google *sections* are no usual *results*, we ignore them +g_section_with_header='./g-section-with-header' -# images -images_xpath = './/div/a' -image_url_xpath = './@href' -image_img_src_xpath = './img/@src' +# the title is a h3 tag relative to the result group +title_xpath = './/h3[1]' -# property names -# FIXME : no translation -property_address = "Address" -property_phone = "Phone number" +# in the result group there is
it's first child is a (on some results, the is the first "descendant", not ""child") +href_xpath = './/div[@class="r"]//a/@href' +# in the result group there is
containing he *content* +content_xpath = './/div[@class="s"]' -# remove google-specific tracking-url -def parse_url(url_string, google_hostname): - # sanity check - if url_string is None: - return url_string +# Suggestions are links placed in a *card-section*, we extract only the text +# from the links not the links itself. +suggestion_xpath = '//div[contains(@class, "card-section")]//a' - # normal case - parsed_url = urlparse(url_string) - if (parsed_url.netloc in [google_hostname, ''] - and parsed_url.path == redirect_path): - query = dict(parse_qsl(parsed_url.query)) - return query['q'] - else: - return url_string +# Since google does *auto-correction* on the first query these are not really +# *spelling suggestions*, we use them anyway. +spelling_suggestion_xpath = '//div[@class="med"]/p/a' - -# returns extract_text on the first result selected by the xpath or None def extract_text_from_dom(result, xpath): + """returns extract_text on the first result selected by the xpath or None""" r = eval_xpath(result, xpath) if len(r) > 0: return extract_text(r[0]) return None - -# do search-request -def request(query, params): - offset = (params['pageno'] - 1) * 10 - - if params['language'] == 'all' or params['language'] == 'en-US': - language = 'en-GB' - else: - language = match_language(params['language'], supported_languages, language_aliases) +def get_lang_country(params, lang_list, custom_aliases): + """Returns a tuple with *langauage* on its first and *country* on its second + position.""" + language = params['language'] + if language == 'all': + language = 'en-US' language_array = language.split('-') - if params['language'].find('-') > 0: - country = params['language'].split('-')[1] - elif len(language_array) == 2: + + if len(language_array) == 2: country = language_array[1] else: - country = 'US' + country = language_array[0].upper() - url_lang = 'lang_' + language + language = match_language(language, lang_list, custom_aliases) + lang_country = '%s-%s' % (language, country) + if lang_country == 'en-EN': + lang_country = 'en' - if use_locale_domain: - google_hostname = country_to_hostname.get(country.upper(), default_hostname) - else: - google_hostname = default_hostname + return language, country, lang_country + +def request(query, params): + """Google search request""" + + offset = (params['pageno'] - 1) * 10 + language, country, lang_country = get_lang_country( + # pylint: disable=undefined-variable + params, supported_languages, language_aliases + ) + subdomain = 'www.' + google_domains.get(country.upper(), 'google.com') + + # https://www.google.de/search?q=corona&hl=de-DE&lr=lang_de&start=0&tbs=qdr%3Ad&safe=medium + query_url = 'https://'+ subdomain + '/search' + "?" + urlencode({ + 'q': query, + 'hl': lang_country, + 'lr': "lang_" + language, + 'ie': "utf8", + 'oe': "utf8", + 'start': offset, + }) - # original format: ID=3e2b6616cee08557:TM=5556667580:C=r:IP=4.1.12.5-:S=23ASdf0soFgF2d34dfgf-_22JJOmHdfgg - params['cookies']['GOOGLE_ABUSE_EXEMPTION'] = 'x' - params['url'] = search_url.format(offset=offset, - query=urlencode({'q': query}), - hostname=google_hostname, - lang=url_lang, - lang_short=language) if params['time_range'] in time_range_dict: - params['url'] += time_range_search.format(range=time_range_dict[params['time_range']]) + query_url += '&' + urlencode({'tbs': 'qdr:' + time_range_dict[params['time_range']]}) + if params['safesearch']: + query_url += '&' + urlencode({'safe': filter_mapping[params['safesearch']]}) - params['headers']['Accept-Language'] = language + ',' + language + '-' + country - params['headers']['Accept'] = 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8' + params['url'] = query_url + logger.debug("query_url --> %s", query_url) - params['google_hostname'] = google_hostname + # en-US,en;q=0.8,en;q=0.5 + params['headers']['Accept-Language'] = ( + lang_country + ',' + language + ';q=0.8,' + language + ';q=0.5' + ) + logger.debug("HTTP header Accept-Language --> %s", + params['headers']['Accept-Language']) + params['headers']['Accept'] = ( + 'text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,*/*;q=0.8' + ) + #params['google_subdomain'] = subdomain return params - -# get response from search-request def response(resp): + """Get response from google's search request""" results = [] # detect google sorry @@ -215,68 +212,53 @@ def response(resp): if resp_url.path.startswith('/sorry'): raise RuntimeWarning(gettext('CAPTCHA required')) - # which hostname ? - google_hostname = resp.search_params.get('google_hostname') - google_url = "https://" + google_hostname + # which subdomain ? + # subdomain = resp.search_params.get('google_subdomain') # convert the text to dom dom = html.fromstring(resp.text) - instant_answer = eval_xpath(dom, '//div[@id="_vBb"]//text()') - if instant_answer: - results.append({'answer': u' '.join(instant_answer)}) + # results --> answer + answer = eval_xpath(dom, '//div[contains(@class, "LGOjhe")]//text()') + if answer: + results.append({'answer': ' '.join(answer)}) + else: + logger.debug("did not found 'answer'") + + # results --> number_of_results try: - results_num = int(eval_xpath(dom, '//div[@id="resultStats"]//text()')[0] - .split()[1].replace(',', '')) - results.append({'number_of_results': results_num}) - except: - pass + _txt = eval_xpath(dom, '//div[@id="result-stats"]//text()')[0] + _digit = ''.join([n for n in _txt if n.isdigit()]) + number_of_results = int(_digit) + results.append({'number_of_results': number_of_results}) + + except Exception as e: # pylint: disable=broad-except + logger.debug("did not 'number_of_results'") + logger.error(e, exc_info=True) # parse results for result in eval_xpath(dom, results_xpath): + + # google *sections* + if extract_text(eval_xpath(result, g_section_with_header)): + logger.debug("ingoring ") + continue + try: title = extract_text(eval_xpath(result, title_xpath)[0]) - url = parse_url(extract_url(eval_xpath(result, url_xpath), google_url), google_hostname) - parsed_url = urlparse(url, google_hostname) - - # map result - if parsed_url.netloc == google_hostname: - # TODO fix inside links - continue - # if parsed_url.path.startswith(maps_path) or parsed_url.netloc.startswith(map_hostname_start): - # print "yooooo"*30 - # x = eval_xpath(result, map_near) - # if len(x) > 0: - # # map : near the location - # results = results + parse_map_near(parsed_url, x, google_hostname) - # else: - # # map : detail about a location - # results = results + parse_map_detail(parsed_url, result, google_hostname) - # # google news - # elif parsed_url.path == search_path: - # # skipping news results - # pass - - # # images result - # elif parsed_url.path == images_path: - # # only thumbnail image provided, - # # so skipping image results - # # results = results + parse_images(result, google_hostname) - # pass - - else: - # normal result - content = extract_text_from_dom(result, content_xpath) - if content is None: - continue - - # append result - results.append({'url': url, - 'title': title, - 'content': content - }) - except: - logger.debug('result parse error in:\n%s', etree.tostring(result, pretty_print=True)) + url = eval_xpath(result, href_xpath)[0] + content = extract_text_from_dom(result, content_xpath) + results.append({ + 'url': url, + 'title': title, + 'content': content + }) + except Exception as e: # pylint: disable=broad-except + logger.error(e, exc_info=True) + #from lxml import etree + #logger.debug(etree.tostring(result, pretty_print=True)) + #import pdb + #pdb.set_trace() continue # parse suggestion @@ -290,102 +272,16 @@ def response(resp): # return results return results - -def parse_images(result, google_hostname): - results = [] - for image in eval_xpath(result, images_xpath): - url = parse_url(extract_text(eval_xpath(image, image_url_xpath)[0]), google_hostname) - img_src = extract_text(eval_xpath(image, image_img_src_xpath)[0]) - - # append result - results.append({'url': url, - 'title': '', - 'content': '', - 'img_src': img_src, - 'template': 'images.html' - }) - - return results - - -def parse_map_near(parsed_url, x, google_hostname): - results = [] - - for result in x: - title = extract_text_from_dom(result, map_near_title) - url = parse_url(extract_text_from_dom(result, map_near_url), google_hostname) - attributes = [] - phone = extract_text_from_dom(result, map_near_phone) - add_attributes(attributes, property_phone, phone, 'tel:' + phone) - results.append({'title': title, - 'url': url, - 'content': attributes_to_html(attributes) - }) - - return results - - -def parse_map_detail(parsed_url, result, google_hostname): - results = [] - - # try to parse the geoloc - m = re.search(r'@([0-9\.]+),([0-9\.]+),([0-9]+)', parsed_url.path) - if m is None: - m = re.search(r'll\=([0-9\.]+),([0-9\.]+)\&z\=([0-9]+)', parsed_url.query) - - if m is not None: - # geoloc found (ignored) - lon = float(m.group(2)) # noqa - lat = float(m.group(1)) # noqa - zoom = int(m.group(3)) # noqa - - # attributes - attributes = [] - address = extract_text_from_dom(result, map_address_xpath) - phone = extract_text_from_dom(result, map_phone_xpath) - add_attributes(attributes, property_address, address, 'geo:' + str(lat) + ',' + str(lon)) - add_attributes(attributes, property_phone, phone, 'tel:' + phone) - - # title / content / url - website_title = extract_text_from_dom(result, map_website_title_xpath) - content = extract_text_from_dom(result, content_xpath) - website_url = parse_url(extract_text_from_dom(result, map_website_url_xpath), google_hostname) - - # add a result if there is a website - if website_url is not None: - results.append({'title': website_title, - 'content': (content + '
' if content is not None else '') - + attributes_to_html(attributes), - 'url': website_url - }) - - return results - - -def add_attributes(attributes, name, value, url): - if value is not None and len(value) > 0: - attributes.append({'label': name, 'value': value, 'url': url}) - - -def attributes_to_html(attributes): - retval = '' - for a in attributes: - value = a.get('value') - if 'url' in a: - value = '' + value + '' - retval = retval + '' - retval = retval + '
' + a.get('label') + '' + value + '
' - return retval - - # get supported languages from their site def _fetch_supported_languages(resp): - supported_languages = {} + ret_val = {} dom = html.fromstring(resp.text) - options = eval_xpath(dom, '//*[@id="langSec"]//input[@name="lr"]') - for option in options: - code = eval_xpath(option, './@value')[0].split('_')[-1] - name = eval_xpath(option, './@data-name')[0].title() - supported_languages[code] = {"name": name} - return supported_languages + radio_buttons = eval_xpath(dom, '//*[@id="langSec"]//input[@name="lang"]') + + for x in radio_buttons: + name = x.get("data-name") + code = x.get("value") + ret_val[code] = {"name": name} + + return ret_val