The initialization of the DB schema ("base schema") has so far been done on
demand, which causes race conditions with competing threads and processes.
The DDL statements for creating the "base schema" are now executed as part of
the initialization of the app.
Further improvements were made to harden the database applications:
- Wikidata & Radio-Browser engine perform their initialization only once (so far
the initialization was carried out in each thread/process).
- If multiple processes try to set DB's WAL mode when opening the DB at the same
time, this usually leads to another race condition, which is now also caught.
Related:
- https://github.com/searxng/searxng/issues/6181#issuecomment-4586705Closes: #6181
Signed-off-by: Markus Heiser <markus.heiser@darmarit.de>
The type checker in my IDE shut down after over 500 errors / after this
patch there are still 125 criticisms, however its an improvement and a better
starting point.
Signed-off-by: Markus Heiser <markus.heiser@darmarit.de>
The big queries for initializing and updating the currencies take longer than
the default of the wikidata engine, which is only 3sec.
Signed-off-by: Markus Heiser <markus.heiser@darmarit.de>
Wikipedia's zh-classical is not zh_Hant (see doc-string of engines.wikipedia).
Fixed the example in the doc-string of locales.get_engine_locale() to 'zh_TW'.
Signed-off-by: Markus Heiser <markus.heiser@darmarit.de>
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>
Implements a fetch_traits function for the Wikipedia engines.
.. note::
Does not include migration of the request methode from 'supported_languages'
to 'traits' (EngineTraits) object!
Signed-off-by: Markus Heiser <markus.heiser@darmarit.de>
Wikidata info box images are now loaded from uploads.wikimedia.org instead of commons.wikimedia.org to prevent redirects
Co-authored-by: Markus Heiser <markus.heiser@darmarit.de>
Fix pylint issues from commit (3d96a983)
[format.python] initial formatting of the python code
Signed-off-by: Markus Heiser <markus.heiser@darmarit.de>
Disable the python code formatting from python-black, where the readability of
code suffers by formatting.
Signed-off-by: Markus Heiser <markus.heiser@darmarit.de>
settings.yml:
* outgoing.networks:
* can contains network definition
* propertiers: enable_http, verify, http2, max_connections, max_keepalive_connections,
keepalive_expiry, local_addresses, support_ipv4, support_ipv6, proxies, max_redirects, retries
* retries: 0 by default, number of times searx retries to send the HTTP request (using different IP & proxy each time)
* local_addresses can be "192.168.0.1/24" (it supports IPv6)
* support_ipv4 & support_ipv6: both True by default
see https://github.com/searx/searx/pull/1034
* each engine can define a "network" section:
* either a full network description
* either reference an existing network
* all HTTP requests of engine use the same HTTP configuration (it was not the case before, see proxy configuration in master)
check HTTP response:
* detect some comme CAPTCHA challenge (no solving). In this case the engine is suspended for long a time.
* otherwise raise HTTPError as before
the check is done in poolrequests.py (was before in search.py).
update qwant, wikipedia, wikidata to use raise_for_httperror instead of raise_for_status
use
from searx.engines.duckduckgo import _fetch_supported_languages, supported_languages_url # NOQA
so it is possible to easily remove all unused import using autoflake:
autoflake --in-place --recursive --remove-all-unused-imports searx tests
* Search URL is https://www.wikidata.org/w/index.php?{query}&ns0=1 (with ns0=1 at the end to avoid an HTTP redirection)
* url_detail: remove the disabletidy=1 deprecated parameter
* Add eval_xpath function: compile once for all xpath.
* Add get_id_cache: retrieve all HTML with an id, avoid the slow to procress dynamic xpath '//div[@id="{propertyid}"]'.replace('{propertyid}')
* Create an etree.HTMLParser() instead of using the global one (see #1575)
Markus Heiser