cec73b5 · searx · git.iko.soy

git.iko.soy/searx

Merge pull request #2350 from dalf/mod-engines-report-captcha

[mod] stackoverflow & yandex: detect CAPTCHA response

commitcec73b5dcf6e94918798c6e1e6457dd021a34a39 authorAlexandre Flament <alex@al-f.net> dateThu 3 Dec 2020, 13:50 (6 years ago) parent7905d41 fa909c7

2 files changed+12−2

compare with master

unifiedsplit

searx/engines/stackoverflow.py+6−1 searx/engines/yandex.py+6−1

searx/engines/stackoverflow.py+6−1

@@ -10,9 +10,10 @@

1010 @parse url, title, content

1111"""

1212

13−from urllib.parse import urlencode, urljoin

13+from urllib.parse import urlencode, urljoin, urlparse

1414from lxml import html

1515from searx.utils import extract_text

16+from searx.exceptions import SearxEngineCaptchaException

1617

1718# engine dependent config

1819categories = ['it']

@@ -37,6 +38,10 @@ def request(query, params):

3738

3839# get response from search-request

3940def response(resp):

41+ resp_url = urlparse(resp.url)

42+ if resp_url.path.startswith('/nocaptcha'):

43+ raise SearxEngineCaptchaException()

44+

4045 results = []

4146

4247 dom = html.fromstring(resp.text)

searx/engines/yandex.py+6−1

@@ -9,9 +9,10 @@

99 @parse url, title, content

1010"""

1111

12−from urllib.parse import urlencode

12+from urllib.parse import urlencode, urlparse

1313from lxml import html

1414from searx import logger

15+from searx.exceptions import SearxEngineCaptchaException

1516

1617logger = logger.getChild('yandex engine')

1718

@@ -47,6 +48,10 @@ def request(query, params):

4748

4849# get response from search-request

4950def response(resp):

51+ resp_url = urlparse(resp.url)

52+ if resp_url.path.startswith('/showcaptcha'):

53+ raise SearxEngineCaptchaException()

54+

5055 dom = html.fromstring(resp.text)

5156 results = []

5257