searxng/searx/engines/google.py
Markus Heiser 2499899554 [mod] Google: reversed engineered & upgrade to data_type: traits_v1
Partial reverse engineering of the Google engines including a improved language
and region handling based on the engine.traits_v1 data.

When ever possible the implementations of the Google engines try to make use of
the async REST APIs.  The get_lang_info() has been generalized to a
get_google_info() function / especially the region handling has been improved by
adding the cr parameter.

searx/data/engine_traits.json
  Add data type "traits_v1" generated by the fetch_traits() functions from:

  - Google (WEB),
  - Google images,
  - Google news,
  - Google scholar and
  - Google videos

  and remove data from obsolete data type "supported_languages".

  A traits.custom type that maps region codes to *supported_domains* is fetched
  from https://www.google.com/supported_domains

searx/autocomplete.py:
  Reversed engineered autocomplete from Google WEB.  Supports Google's languages and
  subdomains.  The old API suggestqueries.google.com/complete has been replaced
  by the async REST API: https://{subdomain}/complete/search?{args}

searx/engines/google.py
  Reverse engineering and extensive testing ..
  - fetch_traits():  Fetch languages & regions from Google properties.
  - always use the async REST API (formally known as 'use_mobile_ui')
  - use *supported_domains* from traits
  - improved the result list by fetching './/div[@data-content-feature]'
    and parsing the type of the various *content features* --> thumbnails are
    added

searx/engines/google_images.py
  Reverse engineering and extensive testing ..
  - fetch_traits():  Fetch languages & regions from Google properties.
  - use *supported_domains* from traits
  - if exists, freshness_date is added to the result
  - issue 1864: result list has been improved a lot (due to the new cr parameter)

searx/engines/google_news.py
  Reverse engineering and extensive testing ..
  - fetch_traits():  Fetch languages & regions from Google properties.
    *supported_domains* is not needed but a ceid list has been added.
  - different region handling compared to Google WEB
  - fixed for various languages & regions (due to the new ceid parameter) /
    avoid CONSENT page
  - Google News do no longer support time range
  - result list has been fixed: XPath of pub_date and pub_origin

searx/engines/google_videos.py
  - fetch_traits():  Fetch languages & regions from Google properties.
  - use *supported_domains* from traits
  - add paging support
  - implement a async request ('asearch': 'arc' & 'async':
    'use_ac:true,_fmt:html')
  - simplified code (thanks to '_fmt:html' request)
  - issue 1359: fixed xpath of video length data

searx/engines/google_scholar.py
  - fetch_traits():  Fetch languages & regions from Google properties.
  - use *supported_domains* from traits
  - request(): include patents & citations
  - response(): fixed CAPTCHA detection (Scholar has its own CATCHA manager)
  - hardening XPath to iterate over results
  - fixed XPath of pub_type (has been change from gs_ct1 to gs_cgt2 class)
  - issue 1769 fixed: new request implementation is no longer incompatible

Signed-off-by: Markus Heiser <markus.heiser@darmarit.de>
2023-03-24 10:37:42 +01:00

511 lines
18 KiB
Python

# SPDX-License-Identifier: AGPL-3.0-or-later
# lint: pylint
"""This is the implementation of the Google WEB engine. Some of this
implementations (manly the :py:obj:`get_google_info`) are shared by other
engines:
- :ref:`google images engine`
- :ref:`google news engine`
- :ref:`google videos engine`
- :ref:`google scholar engine`
- :ref:`google autocomplete`
"""
from typing import TYPE_CHECKING
import re
from urllib.parse import urlencode
from lxml import html
import babel
import babel.core
import babel.languages
from searx.utils import extract_text, eval_xpath, eval_xpath_list, eval_xpath_getindex
from searx.locales import language_tag, region_tag, get_offical_locales
from searx import network
from searx.exceptions import SearxEngineCaptchaException
from searx.enginelib.traits import EngineTraits
if TYPE_CHECKING:
import logging
logger: logging.Logger
traits: EngineTraits
# about
about = {
"website": 'https://www.google.com',
"wikidata_id": 'Q9366',
"official_api_documentation": 'https://developers.google.com/custom-search/',
"use_official_api": False,
"require_api_key": False,
"results": 'HTML',
}
# engine dependent config
categories = ['general', 'web']
paging = True
time_range_support = True
safesearch = True
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[@data-sokoban-container]'
title_xpath = './/a/h3[1]'
href_xpath = './/a[h3]/@href'
content_xpath = './/div[@data-content-feature]'
# google *sections* are no usual *results*, we ignore them
g_section_with_header = './g-section-with-header'
# 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, "EIaa9b")]//a'
# UI_ASYNC = 'use_ac:true,_fmt:html' # returns a HTTP 500 when user search for
# # celebrities like '!google natasha allegri'
# # or '!google chris evans'
UI_ASYNC = 'use_ac:true,_fmt:prog'
"""Format of the response from UI's async request."""
def get_google_info(params, eng_traits):
"""Composing various (language) properties for the google engines (:ref:`google
API`).
This function is called by the various google engines (:ref:`google web
engine`, :ref:`google images engine`, :ref:`google news engine` and
:ref:`google videos engine`).
:param dict param: Request parameters of the engine. At least
a ``searxng_locale`` key should be in the dictionary.
:param eng_traits: Engine's traits fetched from google preferences
(:py:obj:`searx.enginelib.traits.EngineTraits`)
:rtype: dict
:returns:
Py-Dictionary with the key/value pairs:
language:
The language code that is used by google (e.g. ``lang_en`` or
``lang_zh-TW``)
country:
The country code that is used by google (e.g. ``US`` or ``TW``)
locale:
A instance of :py:obj:`babel.core.Locale` build from the
``searxng_locale`` value.
subdomain:
Google subdomain :py:obj:`google_domains` that fits to the country
code.
params:
Py-Dictionary with additional request arguments (can be passed to
:py:func:`urllib.parse.urlencode`).
- ``hl`` parameter: specifies the interface language of user interface.
- ``lr`` parameter: restricts search results to documents written in
a particular language.
- ``cr`` parameter: restricts search results to documents
originating in a particular country.
- ``ie`` parameter: sets the character encoding scheme that should
be used to interpret the query string ('utf8').
- ``oe`` parameter: sets the character encoding scheme that should
be used to decode the XML result ('utf8').
headers:
Py-Dictionary with additional HTTP headers (can be passed to
request's headers)
- ``Accept: '*/*``
"""
ret_val = {
'language': None,
'country': None,
'subdomain': None,
'params': {},
'headers': {},
'cookies': {},
'locale': None,
}
sxng_locale = params.get('searxng_locale', 'all')
try:
locale = babel.Locale.parse(sxng_locale, sep='-')
except babel.core.UnknownLocaleError:
locale = None
eng_lang = eng_traits.get_language(sxng_locale, 'lang_en')
lang_code = eng_lang.split('_')[-1] # lang_zh-TW --> zh-TW / lang_en --> en
country = eng_traits.get_region(sxng_locale, eng_traits.all_locale)
# Test zh_hans & zh_hant --> in the topmost links in the result list of list
# TW and HK you should a find wiktionary.org zh_hant link. In the result
# list of zh-CN should not be no hant link instead you should find
# zh.m.wikipedia.org/zh somewhere in the top.
# '!go 日 :zh-TW' --> https://zh.m.wiktionary.org/zh-hant/%E6%97%A5
# '!go 日 :zh-CN' --> https://zh.m.wikipedia.org/zh/%E6%97%A5
ret_val['language'] = eng_lang
ret_val['country'] = country
ret_val['locale'] = locale
ret_val['subdomain'] = eng_traits.custom['supported_domains'].get(country.upper(), 'www.google.com')
# hl parameter:
# The hl parameter specifies the interface language (host language) of
# your user interface. To improve the performance and the quality of your
# search results, you are strongly encouraged to set this parameter
# explicitly.
# https://developers.google.com/custom-search/docs/xml_results#hlsp
# The Interface Language:
# https://developers.google.com/custom-search/docs/xml_results_appendices#interfaceLanguages
ret_val['params']['hl'] = lang_code
# lr parameter:
# The lr (language restrict) parameter restricts search results to
# documents written in a particular language.
# https://developers.google.com/custom-search/docs/xml_results#lrsp
# Language Collection Values:
# https://developers.google.com/custom-search/docs/xml_results_appendices#languageCollections
#
# To select 'all' languages an empty 'lr' value is used.
#
# Different to other google services, Google Schloar supports to select more
# than one language. The languages are seperated by a pipe '|' (logical OR).
# By example: &lr=lang_zh-TW%7Clang_de selects articles written in
# traditional chinese OR german language.
ret_val['params']['lr'] = eng_lang
if sxng_locale == 'all':
ret_val['params']['lr'] = ''
# cr parameter:
# The cr parameter restricts search results to documents originating in a
# particular country.
# https://developers.google.com/custom-search/docs/xml_results#crsp
ret_val['params']['cr'] = 'country' + country
if sxng_locale == 'all':
ret_val['params']['cr'] = ''
# gl parameter: (mandatory by Geeogle News)
# The gl parameter value is a two-letter country code. For WebSearch
# results, the gl parameter boosts search results whose country of origin
# matches the parameter value. See the Country Codes section for a list of
# valid values.
# Specifying a gl parameter value in WebSearch requests should improve the
# relevance of results. This is particularly true for international
# customers and, even more specifically, for customers in English-speaking
# countries other than the United States.
# https://developers.google.com/custom-search/docs/xml_results#glsp
ret_val['params']['gl'] = country
# ie parameter:
# The ie parameter sets the character encoding scheme that should be used
# to interpret the query string. The default ie value is latin1.
# https://developers.google.com/custom-search/docs/xml_results#iesp
ret_val['params']['ie'] = 'utf8'
# oe parameter:
# The oe parameter sets the character encoding scheme that should be used
# to decode the XML result. The default oe value is latin1.
# https://developers.google.com/custom-search/docs/xml_results#oesp
ret_val['params']['oe'] = 'utf8'
# num parameter:
# The num parameter identifies the number of search results to return.
# The default num value is 10, and the maximum value is 20. If you request
# more than 20 results, only 20 results will be returned.
# https://developers.google.com/custom-search/docs/xml_results#numsp
# HINT: seems to have no effect (tested in google WEB & Images)
# ret_val['params']['num'] = 20
# HTTP headers
ret_val['headers']['Accept'] = '*/*'
# Cookies
# - https://github.com/searxng/searxng/pull/1679#issuecomment-1235432746
# - https://github.com/searxng/searxng/issues/1555
ret_val['cookies']['CONSENT'] = "YES+"
return ret_val
def detect_google_sorry(resp):
if resp.url.host == 'sorry.google.com' or resp.url.path.startswith('/sorry'):
raise SearxEngineCaptchaException()
def request(query, params):
"""Google search request"""
# pylint: disable=line-too-long
offset = (params['pageno'] - 1) * 10
google_info = get_google_info(params, traits)
# https://www.google.de/search?q=corona&hl=de&lr=lang_de&start=0&tbs=qdr%3Ad&safe=medium
query_url = (
'https://'
+ google_info['subdomain']
+ '/search'
+ "?"
+ urlencode(
{
'q': query,
**google_info['params'],
'filter': '0',
'start': offset,
# 'vet': '12ahUKEwik3ZbIzfn7AhXMX_EDHbUDBh0QxK8CegQIARAC..i',
# 'ved': '2ahUKEwik3ZbIzfn7AhXMX_EDHbUDBh0Q_skCegQIARAG',
# 'cs' : 1,
# 'sa': 'N',
# 'yv': 3,
# 'prmd': 'vin',
# 'ei': 'GASaY6TxOcy_xc8PtYeY6AE',
# 'sa': 'N',
# 'sstk': 'AcOHfVkD7sWCSAheZi-0tx_09XDO55gTWY0JNq3_V26cNN-c8lfD45aZYPI8s_Bqp8s57AHz5pxchDtAGCA_cikAWSjy9kw3kgg'
# formally known as use_mobile_ui
'asearch': 'arc',
'async': UI_ASYNC,
}
)
)
if params['time_range'] in time_range_dict:
query_url += '&' + urlencode({'tbs': 'qdr:' + time_range_dict[params['time_range']]})
if params['safesearch']:
query_url += '&' + urlencode({'safe': filter_mapping[params['safesearch']]})
params['url'] = query_url
params['cookies'] = google_info['cookies']
params['headers'].update(google_info['headers'])
return params
# (function(){var s='data:image/jpeg;base64,/9j/4AAQ ...
# ... DX/Ff5XSpSgdU32xSlKDJ//9k\x3d';var ii=['dimg_21'];_setImagesSrc(ii,s);})();
RE_DATA_IMAGE = re.compile(r"'(data:image[^']*)'[^']*ii=\['([^']*)'\];_setImagesSrc")
def _parse_data_images(dom):
data_image_map = {}
for _script in eval_xpath_list(dom, "//script[@nonce]"):
script = _script.text
if not script:
continue
script = RE_DATA_IMAGE.search(script)
if not script:
continue
data_image_map[script.group(2)] = script.group(1).replace(r'\x3d', '=')
logger.debug('data:image objects --> %s', list(data_image_map.keys()))
return data_image_map
def response(resp):
"""Get response from google's search request"""
# pylint: disable=too-many-branches, too-many-statements
detect_google_sorry(resp)
results = []
# convert the text to dom
dom = html.fromstring(resp.text)
data_image_map = {}
if '_fmt:html' in UI_ASYNC:
# in this format images are embedded by a bse64 encoded 'data:image'
data_image_map = _parse_data_images(dom)
# results --> answer
answer_list = eval_xpath(dom, '//div[contains(@class, "LGOjhe")]')
if answer_list:
answer_list = [_.xpath("normalize-space()") for _ in answer_list]
results.append({'answer': ' '.join(answer_list)})
else:
logger.debug("did not find 'answer'")
# parse results
for result in eval_xpath_list(dom, results_xpath): # pylint: disable=too-many-nested-blocks
# google *sections*
if extract_text(eval_xpath(result, g_section_with_header)):
logger.debug("ignoring <g-section-with-header>")
continue
try:
title_tag = eval_xpath_getindex(result, title_xpath, 0, default=None)
if title_tag is None:
# this not one of the common google results *section*
logger.debug('ignoring item from the result_xpath list: missing title')
continue
title = extract_text(title_tag)
url = eval_xpath_getindex(result, href_xpath, 0, None)
if url is None:
continue
content = []
img_list = []
for content_feature in eval_xpath(result, content_xpath):
val = content_feature.attrib['data-content-feature']
if val in ['1', '2']:
txt = extract_text(content_feature, allow_none=True)
if txt:
content.append(txt)
elif '0' in val:
img = content_feature.xpath('.//img/@src')
if img:
img = img[0]
if img.startswith('data:image'):
img_id = content_feature.xpath('.//img/@id')
if img_id:
img = data_image_map.get(img_id[0])
img_list.append(img)
if not content:
logger.debug('ignoring item from the result_xpath list: missing content of title "%s"', title)
continue
content = ' / '.join(content)
img_src = img_list[0] if img_list else None
results.append({'url': url, 'title': title, 'content': content, 'img_src': img_src})
except Exception as e: # pylint: disable=broad-except
logger.error(e, exc_info=True)
continue
# parse suggestion
for suggestion in eval_xpath_list(dom, suggestion_xpath):
# append suggestion
results.append({'suggestion': extract_text(suggestion)})
# return results
return results
# get supported languages from their site
skip_countries = [
# official language of google-country not in google-languages
'AL', # Albanien (sq)
'AZ', # Aserbaidschan (az)
'BD', # Bangladesch (bn)
'BN', # Brunei Darussalam (ms)
'BT', # Bhutan (dz)
'ET', # Äthiopien (am)
'GE', # Georgien (ka, os)
'GL', # Grönland (kl)
'KH', # Kambodscha (km)
'LA', # Laos (lo)
'LK', # Sri Lanka (si, ta)
'ME', # Montenegro (sr)
'MK', # Nordmazedonien (mk, sq)
'MM', # Myanmar (my)
'MN', # Mongolei (mn)
'MV', # Malediven (dv) // dv_MV is unknown by babel
'MY', # Malaysia (ms)
'NP', # Nepal (ne)
'TJ', # Tadschikistan (tg)
'TM', # Turkmenistan (tk)
'UZ', # Usbekistan (uz)
]
def fetch_traits(engine_traits: EngineTraits, add_domains: bool = True):
"""Fetch languages from Google."""
# pylint: disable=import-outside-toplevel, too-many-branches
engine_traits.custom['supported_domains'] = {}
resp = network.get('https://www.google.com/preferences')
if not resp.ok:
raise RuntimeError("Response from Google's preferences is not OK.")
dom = html.fromstring(resp.text)
# supported language codes
lang_map = {'no': 'nb'}
for x in eval_xpath_list(dom, '//*[@id="langSec"]//input[@name="lr"]'):
eng_lang = x.get("value").split('_')[-1]
try:
locale = babel.Locale.parse(lang_map.get(eng_lang, eng_lang), sep='-')
except babel.UnknownLocaleError:
print("ERROR: %s -> %s is unknown by babel" % (x.get("data-name"), eng_lang))
continue
sxng_lang = language_tag(locale)
conflict = engine_traits.languages.get(sxng_lang)
if conflict:
if conflict != eng_lang:
print("CONFLICT: babel %s --> %s, %s" % (sxng_lang, conflict, eng_lang))
continue
engine_traits.languages[sxng_lang] = 'lang_' + eng_lang
# alias languages
engine_traits.languages['zh'] = 'lang_zh-CN'
# supported region codes
for x in eval_xpath_list(dom, '//*[@name="region"]/..//input[@name="region"]'):
eng_country = x.get("value")
if eng_country in skip_countries:
continue
if eng_country == 'ZZ':
engine_traits.all_locale = 'ZZ'
continue
sxng_locales = get_offical_locales(eng_country, engine_traits.languages.keys(), regional=True)
if not sxng_locales:
print("ERROR: can't map from google country %s (%s) to a babel region." % (x.get('data-name'), eng_country))
continue
for sxng_locale in sxng_locales:
engine_traits.regions[region_tag(sxng_locale)] = eng_country
# alias regions
engine_traits.regions['zh-CN'] = 'HK'
# supported domains
if add_domains:
resp = network.get('https://www.google.com/supported_domains')
if not resp.ok:
raise RuntimeError("Response from https://www.google.com/supported_domains is not OK.")
for domain in resp.text.split():
domain = domain.strip()
if not domain or domain in [
'.google.com',
]:
continue
region = domain.split('.')[-1].upper()
engine_traits.custom['supported_domains'][region] = 'www' + domain
if region == 'HK':
# There is no google.cn, we use .com.hk for zh-CN
engine_traits.custom['supported_domains']['CN'] = 'www' + domain