5 Commits

Author SHA1 Message Date
Markus Heiser
8ef5fbca4e [fix] cache.ExpireCache: definition of a context name for the key
The definition of a context name belongs in the abstract base class (was
previously only in the concrete implementation for the SQLite adapter).

Suggested-by: @dalf https://github.com/searxng/searxng/pull/4650#discussion_r2069873853
Signed-off-by: Markus Heiser <markus.heiser@darmarit.de>
2025-05-03 08:39:12 +02:00
Markus Heiser
7351c38e6c [fix] (armv7) cache.ExpireCache: remove option ENCRYPT_VALUE
Prophylactic encryption of the value currently makes no sense; on the contrary,
since the ``cryptography`` package is not available on armv7, it would cause
further problems.

Suggested-by: @dalf https://github.com/searxng/searxng/pull/4650#issuecomment-2830786661
Signed-off-by: Markus Heiser <markus.heiser@darmarit.de>
2025-05-03 08:39:12 +02:00
Markus Heiser
bdfe1c2a15 [mod] engines: migration of the individual cache solutions to EngineCache
The EngineCache class replaces all previously individual solutions for caches in
the context of the engines.

- demo_offline.py
- duckduckgo.py
- radio_browser.py
- soundcloud.py
- startpage.py
- wolframalpha_api.py
- wolframalpha_noapi.py

Search term to test most of the modified engines::

    !ddg !rb !sc !sp !wa test

    !ddg !rb !sc !sp !wa foo

For introspection of the DB, jump into developer environment and run command to
show cache state::

    $ ./manage pyenv.cmd bash --norc --noprofile
    (py3) python -m searx.enginelib cache state

    cache tables and key/values
    ===========================
    [demo_offline        ] 2025-04-22 11:32:50 count        --> (int) 4
    [startpage           ] 2025-04-22 12:32:30 SC_CODE      --> (str) fSOBnhEMlDfE20
    [duckduckgo          ] 2025-04-22 12:32:31 4dff493e.... --> (str) 4-128634958369380006627592672385352473325
    [duckduckgo          ] 2025-04-22 12:40:06 3e2583e2.... --> (str) 4-263126175288871260472289814259666848451
    [radio_browser       ] 2025-04-23 11:33:08 servers      --> (list) ['https://de2.api.radio-browser.info',  ...]
    [soundcloud          ] 2025-04-29 11:40:06 guest_client_id --> (str) EjkRJG0BLNEZquRiPZYdNtJdyGtTuHdp
    [wolframalpha        ] 2025-04-22 12:40:06 code         --> (str) 5aa79f86205ad26188e0e26e28fb7ae7
    number of tables: 6
    number of key/value pairs: 7

In the "cache tables and key/values" section, the table name (engine name) is at
first position on the second there is the calculated expire date and on the
third and fourth position the key/value is shown.

About duckduckgo: The *vqd coode* of ddg depends on the query term and therefore
the key is a hash value of the query term (to not to store the raw query term).

In the "properties of ENGINES_CACHE" section all properties of the SQLiteAppl /
ExpireCache and their last modification date are shown::

    properties of ENGINES_CACHE
    ===========================
    [last modified: 2025-04-22 11:32:27] DB_SCHEMA           : 1
    [last modified: 2025-04-22 11:32:27] LAST_MAINTENANCE    :
    [last modified: 2025-04-22 11:32:27] crypt_hash          : ca612e3566fdfd7cf7efe2b1c9349f461158d07cb78a3750e5c5be686aa8ebdc
    [last modified: 2025-04-22 11:32:30] CACHE-TABLE--demo_offline: demo_offline
    [last modified: 2025-04-22 11:32:30] CACHE-TABLE--startpage: startpage
    [last modified: 2025-04-22 11:32:31] CACHE-TABLE--duckduckgo: duckduckgo
    [last modified: 2025-04-22 11:33:08] CACHE-TABLE--radio_browser: radio_browser
    [last modified: 2025-04-22 11:40:06] CACHE-TABLE--soundcloud: soundcloud
    [last modified: 2025-04-22 11:40:06] CACHE-TABLE--wolframalpha: wolframalpha

These properties provide information about the state of the ExpireCache and
control the behavior.  For example, the maintenance intervals are controlled by
the last modification date of the LAST_MAINTENANCE property and the hash value
of the password can be used to detect whether the password has been changed (in
this case the DB entries can no longer be decrypted and the entire cache must be
discarded).

Signed-off-by: Markus Heiser <markus.heiser@darmarit.de>
2025-05-03 08:39:12 +02:00
Markus Heiser
4cbfba9d7b [mod] ExpireCache - sqlite based key/value cache with expire time 2025-05-03 08:39:12 +02:00
Markus Heiser
4a594f1b53 [fix] ResourceWarning: unclosed database in sqlite3
Reported:

- https://github.com/inetol-infrastructure/searxng-container/issues/5

Related:

- https://github.com/searxng/searxng/issues/4405#issuecomment-2692352352

Signed-off-by: Markus Heiser <markus.heiser@darmarit.de>
2025-05-03 08:39:12 +02:00
17 changed files with 962 additions and 187 deletions

View File

@@ -29,6 +29,8 @@
directly using ``python searx/webapp.py``. Doesn't apply to a SearXNG directly using ``python searx/webapp.py``. Doesn't apply to a SearXNG
services running behind a proxy and using socket communications. services running behind a proxy and using socket communications.
.. _server.secret_key:
``secret_key`` : ``$SEARXNG_SECRET`` ``secret_key`` : ``$SEARXNG_SECRET``
Used for cryptography purpose. Used for cryptography purpose.

View File

@@ -4,19 +4,13 @@
Engine Library Engine Library
============== ==============
.. contents::
:depth: 2
:local:
:backlinks: entry
.. automodule:: searx.enginelib .. automodule:: searx.enginelib
:members: :members:
.. _searx.enginelib.traits: .. _searx.enginelib.traits:
Engine traits Engine traits
============= =============
.. automodule:: searx.enginelib.traits .. automodule:: searx.enginelib.traits
:members: :members:

View File

@@ -29,7 +29,7 @@ Programming Interface
parameter. This function can be omitted, if there is no need to setup anything parameter. This function can be omitted, if there is no need to setup anything
in advance. in advance.
:py:func:`search(query, params) <searx.engines.demo_offline.searc>` :py:func:`search(query, params) <searx.engines.demo_offline.search>`
Each offline engine has a function named ``search``. This function is Each offline engine has a function named ``search``. This function is
responsible to perform a search and return the results in a presentable responsible to perform a search and return the results in a presentable
format. (Where *presentable* means presentable by the selected result format. (Where *presentable* means presentable by the selected result

8
docs/src/searx.cache.rst Normal file
View File

@@ -0,0 +1,8 @@
.. _searx.cache:
======
Caches
======
.. automodule:: searx.cache
:members:

405
searx/cache.py Normal file
View File

@@ -0,0 +1,405 @@
"""Implementation of caching solutions.
- :py:obj:`searx.cache.ExpireCache` and its :py:obj:`searx.cache.ExpireCacheCfg`
----
"""
from __future__ import annotations
__all__ = ["ExpireCacheCfg", "ExpireCacheStats", "ExpireCache", "ExpireCacheSQLite"]
import abc
import dataclasses
import datetime
import hashlib
import hmac
import os
import pickle
import sqlite3
import string
import tempfile
import time
import typing
import msgspec
from searx import sqlitedb
from searx import logger
from searx import get_setting
log = logger.getChild("cache")
class ExpireCacheCfg(msgspec.Struct): # pylint: disable=too-few-public-methods
"""Configuration of a :py:obj:`ExpireCache` cache."""
name: str
"""Name of the cache."""
db_url: str = ""
"""URL of the SQLite DB, the path to the database file. If unset a default
DB will be created in `/tmp/sxng_cache_{self.name}.db`"""
MAX_VALUE_LEN: int = 1024 * 10
"""Max lenght of a *serialized* value."""
MAXHOLD_TIME: int = 60 * 60 * 24 * 7 # 7 days
"""Hold time (default in sec.), after which a value is removed from the cache."""
MAINTENANCE_PERIOD: int = 60 * 60 # 2h
"""Maintenance period in seconds / when :py:obj:`MAINTENANCE_MODE` is set to
``auto``."""
MAINTENANCE_MODE: typing.Literal["auto", "off"] = "auto"
"""Type of maintenance mode
``auto``:
Maintenance is carried out automatically as part of the maintenance
intervals (:py:obj:`MAINTENANCE_PERIOD`); no external process is required.
``off``:
Maintenance is switched off and must be carried out by an external process
if required.
"""
password: bytes = get_setting("server.secret_key").encode() # type: ignore
"""Password used by :py:obj:`ExpireCache.secret_hash`.
The default password is taken from :ref:`secret_key <server.secret_key>`.
When the password is changed, the hashed keys in the cache can no longer be
used, which is why all values in the cache are deleted when the password is
changed.
"""
def __post_init__(self):
# if db_url is unset, use a default DB in /tmp/sxng_cache_{name}.db
if not self.db_url:
self.db_url = tempfile.gettempdir() + os.sep + f"sxng_cache_{ExpireCache.normalize_name(self.name)}.db"
@dataclasses.dataclass
class ExpireCacheStats:
"""Dataclass wich provides information on the status of the cache."""
cached_items: dict[str, list[tuple[str, typing.Any, int]]]
"""Values in the cache mapped by context name.
.. code: python
{
"context name": [
("foo key": "foo value", <expire>),
("bar key": "bar value", <expire>),
# ...
],
# ...
}
"""
def report(self):
c_ctx = 0
c_kv = 0
lines = []
for ctx_name, kv_list in self.cached_items.items():
c_ctx += 1
if not kv_list:
lines.append(f"[{ctx_name:20s}] empty")
continue
for key, value, expire in kv_list:
valid_until = datetime.datetime.fromtimestamp(expire).strftime("%Y-%m-%d %H:%M:%S")
c_kv += 1
lines.append(f"[{ctx_name:20s}] {valid_until} {key:12}" f" --> ({type(value).__name__}) {value} ")
lines.append(f"Number of contexts: {c_ctx}")
lines.append(f"number of key/value pairs: {c_kv}")
return "\n".join(lines)
class ExpireCache(abc.ABC):
"""Abstract base class for the implementation of a key/value cache
with expire date."""
cfg: ExpireCacheCfg
hash_token = "hash_token"
@abc.abstractmethod
def set(self, key: str, value: typing.Any, expire: int | None, ctx: str | None = None) -> bool:
"""Set *key* to *value*. To set a timeout on key use argument
``expire`` (in sec.). If expire is unset the default is taken from
:py:obj:`ExpireCacheCfg.MAXHOLD_TIME`. After the timeout has expired,
the key will automatically be deleted.
The ``ctx`` argument specifies the context of the ``key``. A key is
only unique in its context.
The concrete implementations of this abstraction determine how the
context is mapped in the connected database. In SQL databases, for
example, the context is a DB table or in a Key/Value DB it could be
a prefix for the key.
If the context is not specified (the default is ``None``) then a
default context should be used, e.g. a default table for SQL databases
or a default prefix in a Key/Value DB.
"""
@abc.abstractmethod
def get(self, key: str, default=None, ctx: str | None = None) -> typing.Any:
"""Return *value* of *key*. If key is unset, ``None`` is returned."""
@abc.abstractmethod
def maintenance(self, force: bool = False, truncate: bool = False) -> bool:
"""Performs maintenance on the cache.
``force``:
Maintenance should be carried out even if the maintenance interval has
not yet been reached.
``truncate``:
Truncate the entire cache, which is necessary, for example, if the
password has changed.
"""
@abc.abstractmethod
def state(self) -> ExpireCacheStats:
"""Returns a :py:obj:`ExpireCacheStats`, which provides information
about the status of the cache."""
@staticmethod
def build_cache(cfg: ExpireCacheCfg) -> ExpireCache:
"""Factory to build a caching instance.
.. note::
Currently, only the SQLite adapter is available, but other database
types could be implemented in the future, e.g. a Valkey (Redis)
adapter.
"""
return ExpireCacheSQLite(cfg)
@staticmethod
def normalize_name(name: str) -> str:
"""Returns a normalized name that can be used as a file name or as a SQL
table name (is used, for example, to normalize the context name)."""
_valid = "-_." + string.ascii_letters + string.digits
return "".join([c for c in name if c in _valid])
def serialize(self, value: typing.Any) -> bytes:
dump: bytes = pickle.dumps(value)
return dump
def deserialize(self, value: bytes) -> typing.Any:
obj = pickle.loads(value)
return obj
def secret_hash(self, name: str | bytes) -> str:
"""Creates a hash of the argument ``name``. The hash value is formed
from the ``name`` combined with the :py:obj:`password
<ExpireCacheCfg.password>`. Can be used, for example, to make the
``key`` stored in the DB unreadable for third parties."""
if isinstance(name, str):
name = bytes(name, encoding='utf-8')
m = hmac.new(name + self.cfg.password, digestmod='sha256')
return m.hexdigest()
class ExpireCacheSQLite(sqlitedb.SQLiteAppl, ExpireCache):
"""Cache that manages key/value pairs in a SQLite DB. The DB model in the
SQLite DB is implemented in abstract class :py:obj:`SQLiteAppl
<searx.sqlitedb.SQLiteAppl>`.
The following configurations are required / supported:
- :py:obj:`ExpireCacheCfg.db_url`
- :py:obj:`ExpireCacheCfg.MAXHOLD_TIME`
- :py:obj:`ExpireCacheCfg.MAINTENANCE_PERIOD`
- :py:obj:`ExpireCacheCfg.MAINTENANCE_MODE`
"""
DB_SCHEMA = 1
# The key/value tables will be created on demand by self.create_table
DDL_CREATE_TABLES = {}
CACHE_TABLE_PREFIX = "CACHE-TABLE-"
def __init__(self, cfg: ExpireCacheCfg):
"""An instance of the SQLite expire cache is build up from a
:py:obj:`config <ExpireCacheCfg>`."""
self.cfg = cfg
if cfg.db_url == ":memory:":
log.critical("don't use SQLite DB in :memory: in production!!")
super().__init__(cfg.db_url)
def init(self, conn: sqlite3.Connection) -> bool:
ret_val = super().init(conn)
if not ret_val:
return False
new = hashlib.sha256(self.cfg.password).hexdigest()
old = self.properties(self.hash_token)
if old != new:
if old is not None:
log.warning("[%s] hash token changed: truncate all cache tables", self.cfg.name)
self.maintenance(force=True, truncate=True)
self.properties.set(self.hash_token, new)
return True
def maintenance(self, force: bool = False, truncate: bool = False) -> bool:
if not force and int(time.time()) < self.next_maintenance_time:
# log.debug("no maintenance required yet, next maintenance interval is in the future")
return False
# Prevent parallel DB maintenance cycles from other DB connections
# (e.g. in multi thread or process environments).
self.properties.set("LAST_MAINTENANCE", "") # hint: this (also) sets the m_time of the property!
if truncate:
self.truncate_tables(self.table_names)
return True
# drop items by expire time stamp ..
expire = int(time.time())
with self.connect() as conn:
for table in self.table_names:
res = conn.execute(f"DELETE FROM {table} WHERE expire < ?", (expire,))
log.debug("deleted %s keys from table %s (expire date reached)", res.rowcount, table)
# Vacuuming the WALs
# https://www.theunterminatedstring.com/sqlite-vacuuming/
conn.execute("PRAGMA wal_checkpoint(TRUNCATE)")
conn.close()
return True
def create_table(self, table: str) -> bool:
"""Create DB ``table`` if it has not yet been created, no recreates are
initiated if the table already exists.
"""
if table in self.table_names:
# log.debug("key/value table %s exists in DB (no need to recreate)", table)
return False
log.info("key/value table '%s' NOT exists in DB -> create DB table ..", table)
sql_table = "\n".join(
[
f"CREATE TABLE IF NOT EXISTS {table} (",
" key TEXT,",
" value BLOB,",
f" expire INTEGER DEFAULT (strftime('%s', 'now') + {self.cfg.MAXHOLD_TIME}),",
"PRIMARY KEY (key))",
]
)
sql_index = f"CREATE INDEX IF NOT EXISTS index_expire_{table} ON {table}(expire);"
with self.connect() as conn:
conn.execute(sql_table)
conn.execute(sql_index)
conn.close()
self.properties.set(f"{self.CACHE_TABLE_PREFIX}-{table}", table)
return True
@property
def table_names(self) -> list[str]:
"""List of key/value tables already created in the DB."""
sql = f"SELECT value FROM properties WHERE name LIKE '{self.CACHE_TABLE_PREFIX}%%'"
rows = self.DB.execute(sql).fetchall() or []
return [r[0] for r in rows]
def truncate_tables(self, table_names: list[str]):
log.debug("truncate table: %s", ",".join(table_names))
with self.connect() as conn:
for table in table_names:
conn.execute(f"DELETE FROM {table}")
conn.close()
return True
@property
def next_maintenance_time(self) -> int:
"""Returns (unix epoch) time of the next maintenance."""
return self.cfg.MAINTENANCE_PERIOD + self.properties.m_time("LAST_MAINTENANCE", int(time.time()))
# implement ABC methods of ExpireCache
def set(self, key: str, value: typing.Any, expire: int | None, ctx: str | None = None) -> bool:
"""Set key/value in DB table given by argument ``ctx``. If expire is
unset the default is taken from :py:obj:`ExpireCacheCfg.MAXHOLD_TIME`.
If ``ctx`` argument is ``None`` (the default), a table name is
generated from the :py:obj:`ExpireCacheCfg.name`. If DB table does not
exists, it will be created (on demand) by :py:obj:`self.create_table
<ExpireCacheSQLite.create_table>`.
"""
table = ctx
self.maintenance()
value = self.serialize(value=value)
if len(value) > self.cfg.MAX_VALUE_LEN:
log.warning("ExpireCache.set(): %s.key='%s' - value too big to cache (len: %s) ", table, value, len(value))
return False
if not expire:
expire = self.cfg.MAXHOLD_TIME
expire = int(time.time()) + expire
table_name = table
if not table_name:
table_name = self.normalize_name(self.cfg.name)
self.create_table(table_name)
sql = (
f"INSERT INTO {table_name} (key, value, expire) VALUES (?, ?, ?)"
f" ON CONFLICT DO "
f"UPDATE SET value=?, expire=?"
)
if table:
with self.DB:
self.DB.execute(sql, (key, value, expire, value, expire))
else:
with self.connect() as conn:
conn.execute(sql, (key, value, expire, value, expire))
conn.close()
return True
def get(self, key: str, default=None, ctx: str | None = None) -> typing.Any:
"""Get value of ``key`` from table given by argument ``ctx``. If
``ctx`` argument is ``None`` (the default), a table name is generated
from the :py:obj:`ExpireCacheCfg.name`. If ``key`` not exists (in
table), the ``default`` value is returned.
"""
table = ctx
self.maintenance()
if not table:
table = self.normalize_name(self.cfg.name)
if table not in self.table_names:
return default
sql = f"SELECT value FROM {table} WHERE key = ?"
row = self.DB.execute(sql, (key,)).fetchone()
if row is None:
return default
return self.deserialize(row[0])
def state(self) -> ExpireCacheStats:
cached_items = {}
for table in self.table_names:
cached_items[table] = []
for row in self.DB.execute(f"SELECT key, value, expire FROM {table}"):
cached_items[table].append((row[0], self.deserialize(row[1]), row[2]))
return ExpireCacheStats(cached_items=cached_items)

View File

@@ -1,6 +1,16 @@
# SPDX-License-Identifier: AGPL-3.0-or-later # SPDX-License-Identifier: AGPL-3.0-or-later
"""Implementations of the framework for the SearXNG engines. """Implementations of the framework for the SearXNG engines.
- :py:obj:`searx.enginelib.EngineCache`
- :py:obj:`searx.enginelib.Engine`
- :py:obj:`searx.enginelib.traits`
There is a command line for developer purposes and for deeper analysis. Here is
an example in which the command line is called in the development environment::
$ ./manage pyenv.cmd bash --norc --noprofile
(py3) python -m searx.enginelib --help
.. hint:: .. hint::
The long term goal is to modularize all implementations of the engine The long term goal is to modularize all implementations of the engine
@@ -9,16 +19,158 @@
- move implementations of the :ref:`searx.engines loader` to a new module in - move implementations of the :ref:`searx.engines loader` to a new module in
the :py:obj:`searx.enginelib` namespace. the :py:obj:`searx.enginelib` namespace.
-----
""" """
from __future__ import annotations from __future__ import annotations
from typing import List, Callable, TYPE_CHECKING
__all__ = ["EngineCache", "Engine", "ENGINES_CACHE"]
from typing import List, Callable, TYPE_CHECKING, Any
import string
import typer
from ..cache import ExpireCache, ExpireCacheCfg
if TYPE_CHECKING: if TYPE_CHECKING:
from searx.enginelib import traits from searx.enginelib import traits
ENGINES_CACHE = ExpireCache.build_cache(
ExpireCacheCfg(
name="ENGINES_CACHE",
MAXHOLD_TIME=60 * 60 * 24 * 7, # 7 days
MAINTENANCE_PERIOD=60 * 60, # 2h
)
)
"""Global :py:obj:`searx.cache.ExpireCacheSQLite` instance where the cached
values from all engines are stored. The `MAXHOLD_TIME` is 7 days and the
`MAINTENANCE_PERIOD` is set to two hours."""
app = typer.Typer()
@app.command()
def state():
"""Show state for the caches of the engines."""
title = "cache tables and key/values"
print(title)
print("=" * len(title))
print(ENGINES_CACHE.state().report())
print()
title = f"properties of {ENGINES_CACHE.cfg.name}"
print(title)
print("=" * len(title))
print(str(ENGINES_CACHE.properties)) # type: ignore
@app.command()
def maintenance(force: bool = True):
"""Carry out maintenance on cache of the engines."""
ENGINES_CACHE.maintenance(force=force)
class EngineCache:
"""Persistent (SQLite) key/value cache that deletes its values again after
``expire`` seconds (default/max: :py:obj:`MAXHOLD_TIME
<searx.cache.ExpireCacheCfg.MAXHOLD_TIME>`). This class is a wrapper around
:py:obj:`ENGINES_CACHE` (:py:obj:`ExpireCacheSQLite
<searx.cache.ExpireCacheSQLite>`).
In the :origin:`searx/engines/demo_offline.py` engine you can find an
exemplary implementation of such a cache other exaples are implemeted
in:
- :origin:`searx/engines/radio_browser.py`
- :origin:`searx/engines/soundcloud.py`
- :origin:`searx/engines/startpage.py`
.. code: python
from searx.enginelib import EngineCache
CACHE: EngineCache
def init(engine_settings):
global CACHE
CACHE = EngineCache(engine_settings["name"])
def request(query, params):
token = CACHE.get(key="token")
if token is None:
token = get_token()
# cache token of this engine for 1h
CACHE.set(key="token", value=token, expire=3600)
...
For introspection of the DB, jump into developer environment and run command to
show cache state::
$ ./manage pyenv.cmd bash --norc --noprofile
(py3) python -m searx.enginelib cache state
cache tables and key/values
===========================
[demo_offline ] 2025-04-22 11:32:50 count --> (int) 4
[startpage ] 2025-04-22 12:32:30 SC_CODE --> (str) fSOBnhEMlDfE20
[duckduckgo ] 2025-04-22 12:32:31 4dff493e.... --> (str) 4-128634958369380006627592672385352473325
[duckduckgo ] 2025-04-22 12:40:06 3e2583e2.... --> (str) 4-263126175288871260472289814259666848451
[radio_browser ] 2025-04-23 11:33:08 servers --> (list) ['https://de2.api.radio-browser.info', ...]
[soundcloud ] 2025-04-29 11:40:06 guest_client_id --> (str) EjkRJG0BLNEZquRiPZYdNtJdyGtTuHdp
[wolframalpha ] 2025-04-22 12:40:06 code --> (str) 5aa79f86205ad26188e0e26e28fb7ae7
number of tables: 6
number of key/value pairs: 7
In the "cache tables and key/values" section, the table name (engine name) is at
first position on the second there is the calculated expire date and on the
third and fourth position the key/value is shown.
About duckduckgo: The *vqd coode* of ddg depends on the query term and therefore
the key is a hash value of the query term (to not to store the raw query term).
In the "properties of ENGINES_CACHE" section all properties of the SQLiteAppl /
ExpireCache and their last modification date are shown::
properties of ENGINES_CACHE
===========================
[last modified: 2025-04-22 11:32:27] DB_SCHEMA : 1
[last modified: 2025-04-22 11:32:27] LAST_MAINTENANCE :
[last modified: 2025-04-22 11:32:27] crypt_hash : ca612e3566fdfd7cf7efe...
[last modified: 2025-04-22 11:32:30] CACHE-TABLE--demo_offline: demo_offline
[last modified: 2025-04-22 11:32:30] CACHE-TABLE--startpage: startpage
[last modified: 2025-04-22 11:32:31] CACHE-TABLE--duckduckgo: duckduckgo
[last modified: 2025-04-22 11:33:08] CACHE-TABLE--radio_browser: radio_browser
[last modified: 2025-04-22 11:40:06] CACHE-TABLE--soundcloud: soundcloud
[last modified: 2025-04-22 11:40:06] CACHE-TABLE--wolframalpha: wolframalpha
These properties provide information about the state of the ExpireCache and
control the behavior. For example, the maintenance intervals are controlled by
the last modification date of the LAST_MAINTENANCE property and the hash value
of the password can be used to detect whether the password has been changed (in
this case the DB entries can no longer be decrypted and the entire cache must be
discarded).
"""
def __init__(self, engine_name: str, expire: int | None = None):
self.expire = expire or ENGINES_CACHE.cfg.MAXHOLD_TIME
_valid = "-_." + string.ascii_letters + string.digits
self.table_name = "".join([c if c in _valid else "_" for c in engine_name])
def set(self, key: str, value: Any, expire: int | None = None) -> bool:
return ENGINES_CACHE.set(
key=key,
value=value,
expire=expire or self.expire,
ctx=self.table_name,
)
def get(self, key: str, default=None) -> Any:
return ENGINES_CACHE.get(key, default=default, ctx=self.table_name)
def secret_hash(self, name: str | bytes) -> str:
return ENGINES_CACHE.secret_hash(name=name)
class Engine: # pylint: disable=too-few-public-methods class Engine: # pylint: disable=too-few-public-methods
"""Class of engine instances build from YAML settings. """Class of engine instances build from YAML settings.

View File

@@ -0,0 +1,21 @@
"""Implementation of a command line for development purposes. To start a
command, switch to the environment and run library module as a script::
$ ./manage pyenv.cmd bash --norc --noprofile
(py3) python -m searx.enginelib --help
The following commands can be used for maintenance and introspection
(development) of the engine cache::
(py3) python -m searx.enginelib cache state
(py3) python -m searx.enginelib cache maintenance
"""
import typer
from .. import enginelib
app = typer.Typer()
app.add_typer(enginelib.app, name="cache", help="Commands related to the cache of the engines.")
app()

View File

@@ -15,6 +15,7 @@ close to the implementation, its just a simple example. To get in use of this
import json import json
from searx.result_types import EngineResults from searx.result_types import EngineResults
from searx.enginelib import EngineCache
engine_type = 'offline' engine_type = 'offline'
categories = ['general'] categories = ['general']
@@ -32,14 +33,18 @@ about = {
# if there is a need for globals, use a leading underline # if there is a need for globals, use a leading underline
_my_offline_engine: str = "" _my_offline_engine: str = ""
CACHE: EngineCache
"""Persistent (SQLite) key/value cache that deletes its values after ``expire``
seconds."""
def init(engine_settings=None):
def init(engine_settings):
"""Initialization of the (offline) engine. The origin of this demo engine is a """Initialization of the (offline) engine. The origin of this demo engine is a
simple json string which is loaded in this example while the engine is simple json string which is loaded in this example while the engine is
initialized. initialized."""
global _my_offline_engine, CACHE # pylint: disable=global-statement
""" CACHE = EngineCache(engine_settings["name"]) # type:ignore
global _my_offline_engine # pylint: disable=global-statement
_my_offline_engine = ( _my_offline_engine = (
'[ {"value": "%s"}' '[ {"value": "%s"}'
@@ -57,8 +62,8 @@ def search(query, request_params) -> EngineResults:
results. results.
""" """
res = EngineResults() res = EngineResults()
count = CACHE.get("count", 0)
count = 0
for row in json.loads(_my_offline_engine): for row in json.loads(_my_offline_engine):
count += 1 count += 1
kvmap = { kvmap = {
@@ -75,4 +80,7 @@ def search(query, request_params) -> EngineResults:
) )
) )
res.add(res.types.LegacyResult(number_of_results=count)) res.add(res.types.LegacyResult(number_of_results=count))
# cache counter value for 20sec
CACHE.set("count", count, expire=20)
return res return res

View File

@@ -6,16 +6,17 @@ DuckDuckGo WEB
from __future__ import annotations from __future__ import annotations
from typing import TYPE_CHECKING
import re
from urllib.parse import quote_plus
import json import json
import re
import typing
from urllib.parse import quote_plus
import babel import babel
import lxml.html import lxml.html
from searx import ( from searx import (
locales, locales,
redislib,
external_bang, external_bang,
) )
from searx.utils import ( from searx.utils import (
@@ -25,12 +26,12 @@ from searx.utils import (
extract_text, extract_text,
) )
from searx.network import get # see https://github.com/searxng/searxng/issues/762 from searx.network import get # see https://github.com/searxng/searxng/issues/762
from searx import redisdb
from searx.enginelib.traits import EngineTraits from searx.enginelib.traits import EngineTraits
from searx.enginelib import EngineCache
from searx.exceptions import SearxEngineCaptchaException from searx.exceptions import SearxEngineCaptchaException
from searx.result_types import EngineResults from searx.result_types import EngineResults
if TYPE_CHECKING: if typing.TYPE_CHECKING:
import logging import logging
logger: logging.Logger logger: logging.Logger
@@ -61,28 +62,18 @@ url = "https://html.duckduckgo.com/html"
time_range_dict = {'day': 'd', 'week': 'w', 'month': 'm', 'year': 'y'} time_range_dict = {'day': 'd', 'week': 'w', 'month': 'm', 'year': 'y'}
form_data = {'v': 'l', 'api': 'd.js', 'o': 'json'} form_data = {'v': 'l', 'api': 'd.js', 'o': 'json'}
__CACHE = []
CACHE: EngineCache
"""Persistent (SQLite) key/value cache that deletes its values after ``expire``
seconds."""
def _cache_key(query: str, region: str): def init(_): # pylint: disable=unused-argument
return 'SearXNG_ddg_web_vqd' + redislib.secret_hash(f"{query}//{region}") global CACHE # pylint: disable=global-statement
CACHE = EngineCache("duckduckgo") # type:ignore
def cache_vqd(query: str, region: str, value: str): def get_vqd(query: str, region: str, force_request: bool = False) -> str:
"""Caches a ``vqd`` value from a query."""
c = redisdb.client()
if c:
logger.debug("VALKEY cache vqd value: %s (%s)", value, region)
c.set(_cache_key(query, region), value, ex=600)
else:
logger.debug("MEM cache vqd value: %s (%s)", value, region)
if len(__CACHE) > 100: # cache vqd from last 100 queries
__CACHE.pop(0)
__CACHE.append((_cache_key(query, region), value))
def get_vqd(query: str, region: str, force_request: bool = False):
"""Returns the ``vqd`` that fits to the *query*. """Returns the ``vqd`` that fits to the *query*.
:param query: The query term :param query: The query term
@@ -114,31 +105,34 @@ def get_vqd(query: str, region: str, force_request: bool = False):
seems the block list is a sliding window: to get my IP rid from the bot list seems the block list is a sliding window: to get my IP rid from the bot list
I had to cool down my IP for 1h (send no requests from that IP to DDG). I had to cool down my IP for 1h (send no requests from that IP to DDG).
""" """
key = _cache_key(query, region) key = CACHE.secret_hash(f"{query}//{region}")
value = CACHE.get(key=key)
if value is not None and not force_request:
logger.debug("vqd: re-use cached value: %s", value)
return value
c = redisdb.client() logger.debug("vqd: request value from from duckduckgo.com")
if c: resp = get(f'https://duckduckgo.com/?q={quote_plus(query)}')
value = c.get(key) if resp.status_code == 200: # type: ignore
if value or value == b'': value = extr(resp.text, 'vqd="', '"') # type: ignore
value = value.decode('utf-8') # type: ignore if value:
logger.debug("re-use CACHED vqd value: %s", value) logger.debug("vqd value from duckduckgo.com request: '%s'", value)
return value else:
logger.error("vqd: can't parse value from ddg response (return empty string)")
return ""
else:
logger.error("vqd: got HTTP %s from duckduckgo.com", resp.status_code)
for k, value in __CACHE: if value:
if k == key: CACHE.set(key=key, value=value)
logger.debug("MEM re-use CACHED vqd value: %s", value) else:
return value logger.error("vqd value from duckduckgo.com ", resp.status_code)
return value
if force_request:
resp = get(f'https://duckduckgo.com/?q={quote_plus(query)}')
if resp.status_code == 200: # type: ignore
value = extr(resp.text, 'vqd="', '"') # type: ignore
if value:
logger.debug("vqd value from DDG request: %s", value)
cache_vqd(query, region, value)
return value
return None def set_vqd(query: str, region: str, value: str):
key = CACHE.secret_hash(f"{query}//{region}")
CACHE.set(key=key, value=value, expire=3600)
def get_ddg_lang(eng_traits: EngineTraits, sxng_locale, default='en_US'): def get_ddg_lang(eng_traits: EngineTraits, sxng_locale, default='en_US'):
@@ -373,8 +367,11 @@ def response(resp) -> EngineResults:
# some locales (at least China) does not have a "next page" button # some locales (at least China) does not have a "next page" button
form = form[0] form = form[0]
form_vqd = eval_xpath(form, '//input[@name="vqd"]/@value')[0] form_vqd = eval_xpath(form, '//input[@name="vqd"]/@value')[0]
set_vqd(
cache_vqd(resp.search_params['data']['q'], resp.search_params['data']['kl'], form_vqd) query=resp.search_params['data']['q'],
region=resp.search_params['data']['kl'],
value=str(form_vqd),
)
# just select "web-result" and ignore results of class "result--ad result--ad--small" # just select "web-result" and ignore results of class "result--ad result--ad--small"
for div_result in eval_xpath(doc, '//div[@id="links"]/div[contains(@class, "web-result")]'): for div_result in eval_xpath(doc, '//div[@id="links"]/div[contains(@class, "web-result")]'):
@@ -401,7 +398,7 @@ def response(resp) -> EngineResults:
results.add( results.add(
results.types.Answer( results.types.Answer(
answer=zero_click, answer=zero_click,
url=eval_xpath_getindex(doc, '//div[@id="zero_click_abstract"]/a/@href', 0), url=eval_xpath_getindex(doc, '//div[@id="zero_click_abstract"]/a/@href', 0), # type: ignore
) )
) )

View File

@@ -5,7 +5,9 @@
https://de1.api.radio-browser.info/#Advanced_station_search https://de1.api.radio-browser.info/#Advanced_station_search
""" """
from __future__ import annotations
import typing
import random import random
import socket import socket
from urllib.parse import urlencode from urllib.parse import urlencode
@@ -13,9 +15,15 @@ import babel
from flask_babel import gettext from flask_babel import gettext
from searx.network import get from searx.network import get
from searx.enginelib import EngineCache
from searx.enginelib.traits import EngineTraits from searx.enginelib.traits import EngineTraits
from searx.locales import language_tag from searx.locales import language_tag
if typing.TYPE_CHECKING:
import logging
logger = logging.getLogger()
traits: EngineTraits traits: EngineTraits
about = { about = {
@@ -52,11 +60,24 @@ none filters are applied. Valid filters are:
""" """
servers = [] CACHE: EngineCache
"""Persistent (SQLite) key/value cache that deletes its values after ``expire``
seconds."""
def init(_): def init(_):
# see https://api.radio-browser.info global CACHE # pylint: disable=global-statement
CACHE = EngineCache("radio_browser")
server_list()
def server_list() -> list[str]:
servers = CACHE.get("servers", [])
if servers:
return servers
# hint: can take up to 40sec!
ips = socket.getaddrinfo("all.api.radio-browser.info", 80, 0, 0, socket.IPPROTO_TCP) ips = socket.getaddrinfo("all.api.radio-browser.info", 80, 0, 0, socket.IPPROTO_TCP)
for ip_tuple in ips: for ip_tuple in ips:
_ip: str = ip_tuple[4][0] # type: ignore _ip: str = ip_tuple[4][0] # type: ignore
@@ -65,8 +86,22 @@ def init(_):
if srv not in servers: if srv not in servers:
servers.append(srv) servers.append(srv)
# update server list once in 24h
CACHE.set(key="servers", value=servers, expire=60 * 60 * 24)
return servers
def request(query, params): def request(query, params):
servers = server_list()
if not servers:
logger.error("Fetched server list is empty!")
params["url"] = None
return
server = random.choice(servers)
args = { args = {
'name': query, 'name': query,
'order': 'votes', 'order': 'votes',
@@ -87,8 +122,7 @@ def request(query, params):
if countrycode in traits.custom['countrycodes']: # type: ignore if countrycode in traits.custom['countrycodes']: # type: ignore
args['countrycode'] = countrycode args['countrycode'] = countrycode
params['url'] = f"{random.choice(servers)}/json/stations/search?{urlencode(args)}" params['url'] = f"{server}/json/stations/search?{urlencode(args)}"
return params
def response(resp): def response(resp):
@@ -154,8 +188,9 @@ def fetch_traits(engine_traits: EngineTraits):
babel_reg_list = get_global("territory_languages").keys() babel_reg_list = get_global("territory_languages").keys()
language_list = get(f'{servers[0]}/json/languages').json() # type: ignore server = server_list()[0]
country_list = get(f'{servers[0]}/json/countries').json() # type: ignore language_list = get(f'{server}/json/languages').json() # type: ignore
country_list = get(f'{server}/json/countries').json() # type: ignore
for lang in language_list: for lang in language_list:

View File

@@ -1,14 +1,23 @@
# SPDX-License-Identifier: AGPL-3.0-or-later # SPDX-License-Identifier: AGPL-3.0-or-later
"""SoundCloud is a German audio streaming service.""" """SoundCloud is a German audio streaming service."""
from __future__ import annotations
import re import re
from urllib.parse import quote_plus, urlencode import typing
import datetime import datetime
from urllib.parse import quote_plus, urlencode
from dateutil import parser from dateutil import parser
from lxml import html from lxml import html
from searx.network import get as http_get from searx.network import get as http_get
from searx.enginelib import EngineCache
if typing.TYPE_CHECKING:
import logging
logger: logging.Logger
about = { about = {
"website": "https://soundcloud.com", "website": "https://soundcloud.com",
@@ -28,7 +37,6 @@ HTML frontend of the common WEB site.
""" """
cid_re = re.compile(r'client_id:"([^"]*)"', re.I | re.U) cid_re = re.compile(r'client_id:"([^"]*)"', re.I | re.U)
guest_client_id = ""
results_per_page = 10 results_per_page = 10
soundcloud_facet = "model" soundcloud_facet = "model"
@@ -48,6 +56,10 @@ app_locale_map = {
"sv": "sv", "sv": "sv",
} }
CACHE: EngineCache
"""Persistent (SQLite) key/value cache that deletes its values after ``expire``
seconds."""
def request(query, params): def request(query, params):
@@ -55,6 +67,12 @@ def request(query, params):
# - user_id=451561-497874-703312-310156 # - user_id=451561-497874-703312-310156
# - app_version=1740727428 # - app_version=1740727428
guest_client_id = CACHE.get("guest_client_id")
if guest_client_id is None:
guest_client_id = get_client_id()
if guest_client_id:
CACHE.set(key="guest_client_id", value=guest_client_id)
args = { args = {
"q": query, "q": query,
"offset": (params['pageno'] - 1) * results_per_page, "offset": (params['pageno'] - 1) * results_per_page,
@@ -104,12 +122,12 @@ def response(resp):
return results return results
def init(engine_settings=None): # pylint: disable=unused-argument def init(engine_settings): # pylint: disable=unused-argument
global guest_client_id # pylint: disable=global-statement global CACHE # pylint: disable=global-statement
guest_client_id = get_client_id() CACHE = EngineCache(engine_settings["name"]) # type:ignore
def get_client_id() -> str: def get_client_id() -> str | None:
client_id = "" client_id = ""
url = "https://soundcloud.com" url = "https://soundcloud.com"
@@ -143,4 +161,4 @@ def get_client_id() -> str:
logger.info("using client_id '%s' for soundclud queries", client_id) logger.info("using client_id '%s' for soundclud queries", client_id)
else: else:
logger.warning("missing valid client_id for soundclud queries") logger.warning("missing valid client_id for soundclud queries")
return client_id return client_id or None

View File

@@ -84,7 +84,6 @@ from typing import TYPE_CHECKING, Any
from collections import OrderedDict from collections import OrderedDict
import re import re
from unicodedata import normalize, combining from unicodedata import normalize, combining
from time import time
from datetime import datetime, timedelta from datetime import datetime, timedelta
from json import loads from json import loads
@@ -97,6 +96,7 @@ from searx.network import get # see https://github.com/searxng/searxng/issues/7
from searx.exceptions import SearxEngineCaptchaException from searx.exceptions import SearxEngineCaptchaException
from searx.locales import region_tag from searx.locales import region_tag
from searx.enginelib.traits import EngineTraits from searx.enginelib.traits import EngineTraits
from searx.enginelib import EngineCache
if TYPE_CHECKING: if TYPE_CHECKING:
import logging import logging
@@ -159,10 +159,21 @@ search_form_xpath = '//form[@id="search"]'
</form> </form>
""" """
# timestamp of the last fetch of 'sc' code
sc_code_ts = 0 CACHE: EngineCache
sc_code = '' """Persistent (SQLite) key/value cache that deletes its values after ``expire``
sc_code_cache_sec = 30 seconds."""
def init(_):
global CACHE # pylint: disable=global-statement
# hint: all three startpage engines (WEB, Images & News) can/should use the
# same sc_code ..
CACHE = EngineCache("startpage") # type:ignore
sc_code_cache_sec = 3600
"""Time in seconds the sc-code is cached in memory :py:obj:`get_sc_code`.""" """Time in seconds the sc-code is cached in memory :py:obj:`get_sc_code`."""
@@ -176,14 +187,10 @@ def get_sc_code(searxng_locale, params):
Startpage's search form generates a new sc-code on each request. This Startpage's search form generates a new sc-code on each request. This
function scrap a new sc-code from Startpage's home page every function scrap a new sc-code from Startpage's home page every
:py:obj:`sc_code_cache_sec` seconds. :py:obj:`sc_code_cache_sec` seconds."""
""" sc_code = CACHE.get("SC_CODE", "")
if sc_code:
global sc_code_ts, sc_code # pylint: disable=global-statement
if sc_code and (time() < (sc_code_ts + sc_code_cache_sec)):
logger.debug("get_sc_code: reuse '%s'", sc_code)
return sc_code return sc_code
headers = {**params['headers']} headers = {**params['headers']}
@@ -233,8 +240,9 @@ def get_sc_code(searxng_locale, params):
message="get_sc_code: [PR-695] query new sc time-stamp failed! (%s)" % resp.url, # type: ignore message="get_sc_code: [PR-695] query new sc time-stamp failed! (%s)" % resp.url, # type: ignore
) from exc ) from exc
sc_code_ts = time() sc_code = str(sc_code)
logger.debug("get_sc_code: new value is: %s", sc_code) logger.debug("get_sc_code: new value is: %s", sc_code)
CACHE.set(key="SC_CODE", value=sc_code, expire=sc_code_cache_sec)
return sc_code return sc_code

View File

@@ -5,7 +5,7 @@
from urllib.parse import urlencode from urllib.parse import urlencode
from lxml import etree import lxml.etree
# about # about
about = { about = {
@@ -72,7 +72,7 @@ def replace_pua_chars(text):
def response(resp): def response(resp):
results = [] results = []
search_results = etree.XML(resp.content) search_results = lxml.etree.XML(resp.content)
# return empty array if there are no results # return empty array if there are no results
if search_results.xpath(failure_xpath): if search_results.xpath(failure_xpath):

View File

@@ -3,11 +3,13 @@
Wolfram|Alpha (Science) Wolfram|Alpha (Science)
""" """
from __future__ import annotations
from json import loads from json import loads
from time import time
from urllib.parse import urlencode from urllib.parse import urlencode
from searx.network import get as http_get from searx.network import get as http_get
from searx.enginelib import EngineCache
# about # about
about = { about = {
@@ -40,41 +42,39 @@ search_url = (
referer_url = url + 'input/?{query}' referer_url = url + 'input/?{query}'
token = {'value': '', 'last_updated': None}
# pods to display as image in infobox # pods to display as image in infobox
# this pods do return a plaintext, but they look better and are more useful as images # this pods do return a plaintext, but they look better and are more useful as images
image_pods = {'VisualRepresentation', 'Illustration', 'Symbol'} image_pods = {'VisualRepresentation', 'Illustration', 'Symbol'}
# seems, wolframalpha resets its token in every hour CACHE: EngineCache
def obtain_token(): """Persistent (SQLite) key/value cache that deletes its values after ``expire``
update_time = time() - (time() % 3600) seconds."""
try:
token_response = http_get('https://www.wolframalpha.com/input/api/v1/code?ts=9999999999999999999', timeout=2.0)
token['value'] = loads(token_response.text)['code'] def init(engine_settings):
token['last_updated'] = update_time global CACHE # pylint: disable=global-statement
except: # pylint: disable=bare-except CACHE = EngineCache(engine_settings["name"]) # type:ignore
pass
def obtain_token() -> str:
token = CACHE.get(key="token")
if token is None:
resp = http_get('https://www.wolframalpha.com/input/api/v1/code?ts=9999999999999999999', timeout=2.0)
token = resp.json()["code"]
# seems, wolframalpha resets its token in every hour
CACHE.set(key="code", value=token, expire=3600)
return token return token
def init(engine_settings=None): # pylint: disable=unused-argument
obtain_token()
# do search-request
def request(query, params): def request(query, params):
# obtain token if last update was more than an hour token = obtain_token()
if time() - (token['last_updated'] or 0) > 3600: params['url'] = search_url.format(query=urlencode({'input': query}), token=token)
obtain_token()
params['url'] = search_url.format(query=urlencode({'input': query}), token=token['value'])
params['headers']['Referer'] = referer_url.format(query=urlencode({'i': query})) params['headers']['Referer'] = referer_url.format(query=urlencode({'i': query}))
return params return params
# get response from search-request
def response(resp): def response(resp):
results = [] results = []

View File

@@ -1,5 +1,12 @@
# SPDX-License-Identifier: AGPL-3.0-or-later # SPDX-License-Identifier: AGPL-3.0-or-later
"""Implementations for providing the favicons in SearXNG""" """Implementations for providing the favicons in SearXNG.
There is a command line for developer purposes and for deeper analysis. Here is
an example in which the command line is called in the development environment::
$ ./manage pyenv.cmd bash --norc --noprofile
(py3) python -m searx.favicons --help
"""
from __future__ import annotations from __future__ import annotations

View File

@@ -236,6 +236,12 @@ class FaviconCacheSQLite(sqlitedb.SQLiteAppl, FaviconCache):
model in the SQLite DB is implemented using the abstract class model in the SQLite DB is implemented using the abstract class
:py:obj:`sqlitedb.SQLiteAppl`. :py:obj:`sqlitedb.SQLiteAppl`.
For introspection of the DB, jump into developer environment and run command
to show cache state::
$ ./manage pyenv.cmd bash --norc --noprofile
(py3) python -m searx.favicons cache state
The following configurations are required / supported: The following configurations are required / supported:
- :py:obj:`FaviconCacheConfig.db_url` - :py:obj:`FaviconCacheConfig.db_url`
@@ -357,6 +363,10 @@ CREATE TABLE IF NOT EXISTS blob_map (
if sha256 != FALLBACK_ICON: if sha256 != FALLBACK_ICON:
conn.execute(self.SQL_INSERT_BLOBS, (sha256, bytes_c, mime, data)) conn.execute(self.SQL_INSERT_BLOBS, (sha256, bytes_c, mime, data))
conn.execute(self.SQL_INSERT_BLOB_MAP, (sha256, resolver, authority)) conn.execute(self.SQL_INSERT_BLOB_MAP, (sha256, resolver, authority))
# hint: the with context of the connection object closes the transaction
# but not the DB connection. The connection has to be closed by the
# caller of self.connect()!
conn.close()
return True return True
@@ -376,7 +386,8 @@ CREATE TABLE IF NOT EXISTS blob_map (
return return
self.properties.set("LAST_MAINTENANCE", "") # hint: this (also) sets the m_time of the property! self.properties.set("LAST_MAINTENANCE", "") # hint: this (also) sets the m_time of the property!
# do maintenance tasks # Do maintenance tasks. This can be take a little more time, to avoid
# DB locks, etablish a new DB connecton.
with self.connect() as conn: with self.connect() as conn:
@@ -407,6 +418,12 @@ CREATE TABLE IF NOT EXISTS blob_map (
conn.execute("DELETE FROM blob_map WHERE sha256 IN ('%s')" % "','".join(sha_list)) conn.execute("DELETE FROM blob_map WHERE sha256 IN ('%s')" % "','".join(sha_list))
logger.debug("dropped %s blobs with total size of %s bytes", len(sha_list), c) logger.debug("dropped %s blobs with total size of %s bytes", len(sha_list), c)
# Vacuuming the WALs
# https://www.theunterminatedstring.com/sqlite-vacuuming/
conn.execute("PRAGMA wal_checkpoint(TRUNCATE)")
conn.close()
def _query_val(self, sql, default=None): def _query_val(self, sql, default=None):
val = self.DB.execute(sql).fetchone() val = self.DB.execute(sql).fetchone()
if val is not None: if val is not None:

View File

@@ -7,20 +7,83 @@
:py:obj:`SQLiteProperties`: :py:obj:`SQLiteProperties`:
Class to manage properties stored in a database. Class to manage properties stored in a database.
---- Examplarical implementations based on :py:obj:`SQLiteAppl`:
:py:obj:`searx.cache.ExpireCacheSQLite` :
Cache that manages key/value pairs in a SQLite DB, in which the key/value
pairs are deleted after an "expire" time. This type of cache is used, for
example, for the engines, see :py:obj:`searx.enginelib.EngineCache`.
:py:obj:`searx.favicons.cache.FaviconCacheSQLite` :
Favicon cache that manages the favicon BLOBs in a SQLite DB.
----
""" """
from __future__ import annotations from __future__ import annotations
import sys import abc
import datetime
import re import re
import sqlite3 import sqlite3
import sys
import threading import threading
import abc import uuid
from searx import logger from searx import logger
logger = logger.getChild('sqlitedb') logger = logger.getChild("sqlitedb")
THREAD_LOCAL = threading.local()
class DBSession:
"""A *thead-local* DB session"""
@classmethod
def get_connect(cls, app: SQLiteAppl) -> sqlite3.Connection:
"""Returns a thread local DB connection. The connection is only
established once per thread.
"""
if getattr(THREAD_LOCAL, "DBSession_map", None) is None:
THREAD_LOCAL.DBSession_map = {}
session = THREAD_LOCAL.DBSession_map.get(app.db_url)
if session is None:
session = cls(app)
return session.conn
def __init__(self, app: SQLiteAppl):
self.uuid = uuid.uuid4()
self.app = app
self._conn = None
# self.__del__ will be called, when thread ends
if getattr(THREAD_LOCAL, "DBSession_map", None) is None:
THREAD_LOCAL.DBSession_map = {}
THREAD_LOCAL.DBSession_map[self.app.db_url] = self
@property
def conn(self) -> sqlite3.Connection:
msg = f"[{threading.current_thread().ident}] DBSession: " f"{self.app.__class__.__name__}({self.app.db_url})"
if self._conn is None:
self._conn = self.app.connect()
logger.debug("%s --> created new connection", msg)
# else:
# logger.debug("%s --> already connected", msg)
return self._conn
def __del__(self):
try:
if self._conn is not None:
# HINT: Don't use Python's logging facility in a destructor, it
# will produce error reports when python aborts the process or
# thread, because at this point objects that the logging module
# needs, do not exist anymore.
# msg = f"DBSession: close [{self.uuid}] {self.app.__class__.__name__}({self.app.db_url})"
# logger.debug(msg)
self._conn.close()
except Exception: # pylint: disable=broad-exception-caught
pass
class SQLiteAppl(abc.ABC): class SQLiteAppl(abc.ABC):
@@ -51,13 +114,18 @@ class SQLiteAppl(abc.ABC):
""" """
SQLITE_JOURNAL_MODE = "WAL" SQLITE_JOURNAL_MODE = "WAL"
"""``SQLiteAppl`` applications are optimzed for WAL_ mode, its not recommend
to change the journal mode (see :py:obj:`SQLiteAppl.tear_down`).
.. _WAL: https://sqlite.org/wal.html
"""
SQLITE_CONNECT_ARGS = { SQLITE_CONNECT_ARGS = {
# "timeout": 5.0, # "timeout": 5.0,
# "detect_types": 0, # "detect_types": 0,
"check_same_thread": bool(SQLITE_THREADING_MODE != "serialized"), "check_same_thread": bool(SQLITE_THREADING_MODE != "serialized"),
"cached_statements": 0, # https://github.com/python/cpython/issues/118172 "cached_statements": 0, # https://github.com/python/cpython/issues/118172
# "uri": False, # "uri": False,
"autocommit": False, "isolation_level": None,
} # fmt:skip } # fmt:skip
"""Connection arguments (:py:obj:`sqlite3.connect`) """Connection arguments (:py:obj:`sqlite3.connect`)
@@ -66,10 +134,6 @@ class SQLiteAppl(abc.ABC):
``serialized``. The check is more of a hindrance in this case because it ``serialized``. The check is more of a hindrance in this case because it
would prevent a DB connector from being used in multiple threads. would prevent a DB connector from being used in multiple threads.
``autocommit``:
Is disabled by default. Note: autocommit option has been added in Python
3.12.
``cached_statements``: ``cached_statements``:
Is set to ``0`` by default. Note: Python 3.12+ fetch result are not Is set to ``0`` by default. Note: Python 3.12+ fetch result are not
consistent in multi-threading application and causing an API misuse error. consistent in multi-threading application and causing an API misuse error.
@@ -89,9 +153,17 @@ class SQLiteAppl(abc.ABC):
self.db_url = db_url self.db_url = db_url
self.properties = SQLiteProperties(db_url) self.properties = SQLiteProperties(db_url)
self.thread_local = threading.local()
self._init_done = False self._init_done = False
self._compatibility() self._compatibility()
# atexit.register(self.tear_down)
# def tear_down(self):
# """:ref:`Vacuuming the WALs` upon normal interpreter termination
# (:py:obj:`atexit.register`).
# .. _SQLite: Vacuuming the WALs: https://www.theunterminatedstring.com/sqlite-vacuuming/
# """
# self.DB.execute("PRAGMA wal_checkpoint(TRUNCATE)")
def _compatibility(self): def _compatibility(self):
@@ -113,19 +185,31 @@ class SQLiteAppl(abc.ABC):
"SQLite runtime library version %s is not supported (require >= 3.35)", sqlite3.sqlite_version "SQLite runtime library version %s is not supported (require >= 3.35)", sqlite3.sqlite_version
) )
def _connect(self) -> sqlite3.Connection:
conn = sqlite3.Connection(self.db_url, **self.SQLITE_CONNECT_ARGS) # type: ignore
conn.execute(f"PRAGMA journal_mode={self.SQLITE_JOURNAL_MODE}")
self.register_functions(conn)
return conn
def connect(self) -> sqlite3.Connection: def connect(self) -> sqlite3.Connection:
"""Creates a new DB connection (:py:obj:`SQLITE_CONNECT_ARGS`). If not """Creates a new DB connection (:py:obj:`SQLITE_CONNECT_ARGS`). If not
already done, the DB schema is set up already done, the DB schema is set up. The caller must take care of
closing the resource. Alternatively, :py:obj:`SQLiteAppl.DB` can also
be used (the resource behind `self.DB` is automatically closed when the
process or thread is terminated).
""" """
if sys.version_info < (3, 12): if sys.version_info < (3, 12):
# Prior Python 3.12 there is no "autocommit" option # Prior Python 3.12 there is no "autocommit" option
self.SQLITE_CONNECT_ARGS.pop("autocommit", None) self.SQLITE_CONNECT_ARGS.pop("autocommit", None)
self.init() msg = (
logger.debug("%s: connect to DB: %s // %s", self.__class__.__name__, self.db_url, self.SQLITE_CONNECT_ARGS) f"[{threading.current_thread().ident}] {self.__class__.__name__}({self.db_url})"
conn = sqlite3.Connection(self.db_url, **self.SQLITE_CONNECT_ARGS) # type: ignore f" {self.SQLITE_CONNECT_ARGS} // {self.SQLITE_JOURNAL_MODE}"
conn.execute(f"PRAGMA journal_mode={self.SQLITE_JOURNAL_MODE}") )
self.register_functions(conn) logger.debug(msg)
with self._connect() as conn:
self.init(conn)
return conn return conn
def register_functions(self, conn): def register_functions(self, conn):
@@ -150,7 +234,7 @@ class SQLiteAppl(abc.ABC):
.. _re.search: https://docs.python.org/3/library/re.html#re.search .. _re.search: https://docs.python.org/3/library/re.html#re.search
""" """
conn.create_function('regexp', 2, lambda x, y: 1 if re.search(x, y) else 0, deterministic=True) conn.create_function("regexp", 2, lambda x, y: 1 if re.search(x, y) else 0, deterministic=True)
@property @property
def DB(self) -> sqlite3.Connection: def DB(self) -> sqlite3.Connection:
@@ -168,57 +252,66 @@ class SQLiteAppl(abc.ABC):
https://docs.python.org/3/library/sqlite3.html#sqlite3-controlling-transactions https://docs.python.org/3/library/sqlite3.html#sqlite3-controlling-transactions
""" """
if getattr(self.thread_local, 'DB', None) is None: conn = None
self.thread_local.DB = self.connect()
# Theoretically it is possible to reuse the DB cursor across threads as if self.SQLITE_THREADING_MODE == "serialized":
# of Python 3.12, in practice the threading of the cursor seems to me to # Theoretically it is possible to reuse the DB cursor across threads
# be so faulty that I prefer to establish one connection per thread # as of Python 3.12, in practice the threading of the cursor seems
# to me a little faulty that I prefer to establish one connection
# per thread.
#
# may we can activate this code one day ..
# if self._DB is None:
# self._DB = self.connect()
# conn = self._DB
conn = DBSession.get_connect(self)
else:
conn = DBSession.get_connect(self)
self.thread_local.DB.commit() # Since more than one instance of SQLiteAppl share the same DB
return self.thread_local.DB # connection, we need to make sure that each SQLiteAppl instance has run
# its init method at least once.
self.init(conn)
# In "serialized" mode, SQLite can be safely used by multiple threads return conn
# with no restriction.
#
# if self.SQLITE_THREADING_MODE != "serialized":
# if getattr(self.thread_local, 'DB', None) is None:
# self.thread_local.DB = self.connect()
# return self.thread_local.DB
#
# if self._DB is None:
# self._DB = self.connect() # pylint: disable=attribute-defined-outside-init
# return self._DB
def init(self): def init(self, conn: sqlite3.Connection) -> bool:
"""Initializes the DB schema and properties, is only executed once even """Initializes the DB schema and properties, is only executed once even
if called several times.""" if called several times.
If the initialization has not yet taken place, it is carried out and a
`True` is returned to the caller at the end. If the initialization has
already been carried out in the past, `False` is returned.
"""
if self._init_done: if self._init_done:
return return False
self._init_done = True self._init_done = True
logger.debug("init DB: %s", self.db_url) logger.debug("init DB: %s", self.db_url)
self.properties.init() self.properties.init(conn)
ver = self.properties("DB_SCHEMA") ver = self.properties("DB_SCHEMA")
if ver is None: if ver is None:
with self.properties.DB: with conn:
self.create_schema(self.properties.DB) self.create_schema(conn)
else: else:
ver = int(ver) ver = int(ver)
if ver != self.DB_SCHEMA: if ver != self.DB_SCHEMA:
raise sqlite3.DatabaseError("Expected DB schema v%s, DB schema is v%s" % (self.DB_SCHEMA, ver)) raise sqlite3.DatabaseError("Expected DB schema v%s, DB schema is v%s" % (self.DB_SCHEMA, ver))
logger.debug("DB_SCHEMA = %s", ver) logger.debug("DB_SCHEMA = %s", ver)
def create_schema(self, conn): return True
def create_schema(self, conn: sqlite3.Connection):
logger.debug("create schema ..") logger.debug("create schema ..")
self.properties.set("DB_SCHEMA", self.DB_SCHEMA)
self.properties.set("LAST_MAINTENANCE", "")
with conn: with conn:
for table_name, sql in self.DDL_CREATE_TABLES.items(): for table_name, sql in self.DDL_CREATE_TABLES.items():
conn.execute(sql) conn.execute(sql)
self.properties.set(f"Table {table_name} created", table_name) self.properties.set(f"Table {table_name} created", table_name)
self.properties.set("DB_SCHEMA", self.DB_SCHEMA)
self.properties.set("LAST_MAINTENANCE", "")
class SQLiteProperties(SQLiteAppl): class SQLiteProperties(SQLiteAppl):
@@ -253,33 +346,32 @@ CREATE TABLE IF NOT EXISTS properties (
" ON CONFLICT(name) DO UPDATE" " ON CONFLICT(name) DO UPDATE"
" SET value=excluded.value, m_time=strftime('%s', 'now')" " SET value=excluded.value, m_time=strftime('%s', 'now')"
) )
SQL_DELETE = "DELETE FROM properties WHERE name = ?"
SQL_TABLE_EXISTS = ( SQL_TABLE_EXISTS = (
"SELECT name FROM sqlite_master" "SELECT name FROM sqlite_master"
" WHERE type='table' AND name='properties'" " WHERE type='table' AND name='properties'"
) # fmt:skip ) # fmt:skip
SQLITE_CONNECT_ARGS = dict(SQLiteAppl.SQLITE_CONNECT_ARGS) SQLITE_CONNECT_ARGS = dict(SQLiteAppl.SQLITE_CONNECT_ARGS)
SQLITE_CONNECT_ARGS["autocommit"] = True # This option has no effect before Python 3.12
def __init__(self, db_url: str): # pylint: disable=super-init-not-called def __init__(self, db_url: str): # pylint: disable=super-init-not-called
self.db_url = db_url self.db_url = db_url
self.thread_local = threading.local()
self._init_done = False self._init_done = False
self._compatibility() self._compatibility()
def init(self): def init(self, conn: sqlite3.Connection) -> bool:
"""Initializes DB schema of the properties in the DB.""" """Initializes DB schema of the properties in the DB."""
if self._init_done: if self._init_done:
return return False
self._init_done = True self._init_done = True
logger.debug("init properties of DB: %s", self.db_url) logger.debug("init properties of DB: %s", self.db_url)
with self.DB as conn: res = conn.execute(self.SQL_TABLE_EXISTS)
res = conn.execute(self.SQL_TABLE_EXISTS) if res.fetchone() is None: # DB schema needs to be be created
if res.fetchone() is None: # DB schema needs to be be created self.create_schema(conn)
self.create_schema(conn) return True
def __call__(self, name, default=None): def __call__(self, name: str, default=None):
"""Returns the value of the property ``name`` or ``default`` if property """Returns the value of the property ``name`` or ``default`` if property
not exists in DB.""" not exists in DB."""
@@ -288,36 +380,47 @@ CREATE TABLE IF NOT EXISTS properties (
return default return default
return res[0] return res[0]
def set(self, name, value): def set(self, name: str, value: str | int):
"""Set ``value`` of property ``name`` in DB. If property already """Set ``value`` of property ``name`` in DB. If property already
exists, update the ``m_time`` (and the value).""" exists, update the ``m_time`` (and the value)."""
self.DB.execute(self.SQL_SET, (name, value)) with self.DB:
self.DB.execute(self.SQL_SET, (name, value))
if sys.version_info <= (3, 12): def delete(self, name: str) -> int:
# Prior Python 3.12 there is no "autocommit" option / lets commit """Delete of property ``name`` from DB."""
# explicitely. with self.DB:
self.DB.commit() cur = self.DB.execute(self.SQL_DELETE, (name,))
return cur.rowcount
def row(self, name, default=None): def row(self, name: str, default=None):
"""Returns the DB row of property ``name`` or ``default`` if property """Returns the DB row of property ``name`` or ``default`` if property
not exists in DB.""" not exists in DB."""
cur = self.DB.cursor() res = self.DB.execute("SELECT * FROM properties WHERE name = ?", (name,))
cur.execute("SELECT * FROM properties WHERE name = ?", (name,)) row = res.fetchone()
res = cur.fetchone() if row is None:
if res is None:
return default return default
col_names = [column[0] for column in cur.description]
return dict(zip(col_names, res))
def m_time(self, name, default: int = 0) -> int: col_names = [column[0] for column in row.description]
return dict(zip(col_names, row))
def m_time(self, name: str, default: int = 0) -> int:
"""Last modification time of this property.""" """Last modification time of this property."""
res = self.DB.execute(self.SQL_M_TIME, (name,)).fetchone() res = self.DB.execute(self.SQL_M_TIME, (name,))
if res is None: row = res.fetchone()
if row is None:
return default return default
return int(res[0]) return int(row[0])
def create_schema(self, conn): def create_schema(self, conn):
with conn: with conn:
conn.execute(self.DDL_PROPERTIES) conn.execute(self.DDL_PROPERTIES)
def __str__(self) -> str:
lines = []
for row in self.DB.execute("SELECT name, value, m_time FROM properties"):
name, value, m_time = row
m_time = datetime.datetime.fromtimestamp(m_time).strftime("%Y-%m-%d %H:%M:%S")
lines.append(f"[last modified: {m_time}] {name:20s}: {value}")
return "\n".join(lines)