Frisk/RcGcDb
1
0
Fork 0
mirror of https://gitlab.com/chicken-riders/RcGcDb.git synced 2025-02-23 00:54:09 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

Merge branch 'postgres' into 'master'

Browse source 
Change Database provider from SQLite to Postgres DB

See merge request chicken-riders/RcGcDb!18
This commit is contained in:
Frisk 2021-03-20 15:12:20 +00:00
parent 9a5f57cda6 fa541fefb9
commit b952e62730
7 changed files with 253 additions and 195 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
requirements.txt
View file

@ -3,3 +3,5 @@ aiohttp >= 3.6.2
lxml >= 4.2.1 lxml >= 4.2.1
nest-asyncio >= 1.4.0 nest-asyncio >= 1.4.0
irc >= 19.0.1 irc >= 19.0.1
beautifulsoup4>=4.9.3
asyncpg>=0.22.0

6
settings.json.example
View file

@ -4,10 +4,14 @@
}, },
"max_requests_per_minute": 30, "max_requests_per_minute": 30,
"minimal_cooldown_per_wiki_in_sec": 60, "minimal_cooldown_per_wiki_in_sec": 60,
"database_path": "rcgcdb.db",
"monitoring_webhook": "111111111111111111/AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA", "monitoring_webhook": "111111111111111111/AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA",
"support": "https://discord.gg/v77RTk5", "support": "https://discord.gg/v77RTk5",
"irc_overtime": 3600, "irc_overtime": 3600,
"pg_user": "postgres",
"pg_host": "localhost",
"pg_db": "rcgcdb",
"pg_pass": "secret_password",
"pg_port": "5432",
"irc_servers": { "irc_servers": {
"your custom name for the farm": { "your custom name for the farm": {
"domains": ["wikipedia.org", "otherwikipedia.org"], "domains": ["wikipedia.org", "otherwikipedia.org"],

333
src/bot.py
View file

@ -11,7 +11,7 @@ from typing import Generator
from contextlib import asynccontextmanager from contextlib import asynccontextmanager
from src.argparser import command_line_args from src.argparser import command_line_args
from src.config import settings from src.config import settings
from src.database import db_cursor, db_connection from src.database import db
from src.exceptions import * from src.exceptions import *
from src.misc import get_paths, get_domain from src.misc import get_paths, get_domain
from src.msgqueue import messagequeue, send_to_discord from src.msgqueue import messagequeue, send_to_discord
@ -35,13 +35,17 @@ all_wikis: dict = {}
mw_msgs: dict = {} # will have the type of id: tuple mw_msgs: dict = {} # will have the type of id: tuple
main_tasks: dict = {} main_tasks: dict = {}
# First populate the all_wikis list with every wiki # First populate the all_wikis list with every wiki
# Reasons for this: 1. we require amount of wikis to calculate the cooldown between requests # Reasons for this: 1. we require amount of wikis to calculate the cooldown between requests
# 2. Easier to code # 2. Easier to code
for db_wiki in db_cursor.execute('SELECT wiki, rcid FROM rcgcdw GROUP BY wiki ORDER BY ROWID'): async def populate_allwikis():
all_wikis[db_wiki["wiki"]] = Wiki() # populate all_wikis async with db.pool().acquire() as connection:
all_wikis[db_wiki["wiki"]].rc_active = db_wiki["rcid"] async with connection.transaction():
async for db_wiki in connection.cursor('SELECT DISTINCT wiki, rcid FROM rcgcdw'):
all_wikis[db_wiki["wiki"]] = Wiki() # populate all_wikis
all_wikis[db_wiki["wiki"]].rc_active = db_wiki["rcid"]
queue_limit = settings.get("queue_limit", 30) queue_limit = settings.get("queue_limit", 30)
QueuedWiki = namedtuple("QueuedWiki", ['url', 'amount']) QueuedWiki = namedtuple("QueuedWiki", ['url', 'amount'])
@ -92,7 +96,7 @@ class RcQueue:
all_wikis[wiki].rc_active = -1 all_wikis[wiki].rc_active = -1
if not self[group]["query"]: # if there is no wiki left in the queue, get rid of the task if not self[group]["query"]: # if there is no wiki left in the queue, get rid of the task
logger.debug(f"{group} no longer has any wikis queued!") logger.debug(f"{group} no longer has any wikis queued!")
if not self.check_if_domain_in_db(group): if not await self.check_if_domain_in_db(group):
await self.stop_task_group(group) await self.stop_task_group(group)
else: else:
logger.debug(f"But there are still wikis for it in DB!") logger.debug(f"But there are still wikis for it in DB!")
@ -101,12 +105,12 @@ class RcQueue:
self[group]["task"].cancel() self[group]["task"].cancel()
del self.domain_list[group] del self.domain_list[group]
def check_if_domain_in_db(self, domain): async def check_if_domain_in_db(self, domain):
fetch_all = db_cursor.execute( async with db.pool().acquire() as connection:
'SELECT ROWID, webhook, wiki, lang, display, rcid FROM rcgcdw WHERE rcid != -1 GROUP BY wiki ORDER BY ROWID ASC') async with connection.transaction():
for wiki in fetch_all.fetchall(): async for wiki in connection.cursor('SELECT DISTINCT wiki FROM rcgcdw WHERE rcid != -1;'):
if get_domain(wiki["wiki"]) == domain: if get_domain(wiki["wiki"]) == domain:
return True return True
return False return False
@asynccontextmanager @asynccontextmanager
@ -143,50 +147,50 @@ class RcQueue:
async def update_queues(self): async def update_queues(self):
"""Makes a round on rcgcdb DB and looks for updates to the queues in self.domain_list""" """Makes a round on rcgcdb DB and looks for updates to the queues in self.domain_list"""
try: try:
fetch_all = db_cursor.execute(
'SELECT ROWID, webhook, wiki, lang, display, rcid FROM rcgcdw WHERE rcid != -1 OR rcid IS NULL GROUP BY wiki ORDER BY ROWID ASC')
self.to_remove = [x[0] for x in filter(self.filter_rc_active, all_wikis.items())] # first populate this list and remove wikis that are still in the db, clean up the rest self.to_remove = [x[0] for x in filter(self.filter_rc_active, all_wikis.items())] # first populate this list and remove wikis that are still in the db, clean up the rest
full = set() full = set()
for db_wiki in fetch_all.fetchall(): async with db.pool().acquire() as connection:
domain = get_domain(db_wiki["wiki"]) async with connection.transaction():
try: async for db_wiki in connection.cursor('SELECT DISTINCT wiki, row_number() OVER (ORDER BY webhook) AS rowid, webhook, lang, display, rcid FROM rcgcdw WHERE rcid != -1 OR rcid IS NULL ORDER BY webhook'):
if db_wiki["wiki"] not in all_wikis: domain = get_domain(db_wiki["wiki"])
raise AssertionError try:
self.to_remove.remove(db_wiki["wiki"]) if db_wiki["wiki"] not in all_wikis:
except AssertionError: raise AssertionError
all_wikis[db_wiki["wiki"]] = Wiki() self.to_remove.remove(db_wiki["wiki"])
all_wikis[db_wiki["wiki"]].rc_active = db_wiki["rcid"] except AssertionError:
except ValueError: all_wikis[db_wiki["wiki"]] = Wiki()
pass all_wikis[db_wiki["wiki"]].rc_active = db_wiki["rcid"]
if domain in full: except ValueError:
continue
try:
current_domain: dict = self[domain]
if current_domain["irc"]:
logger.debug("DOMAIN LIST FOR IRC: {}".format(current_domain["irc"].updated))
logger.debug("CURRENT DOMAIN INFO: {}".format(domain))
logger.debug("IS WIKI IN A LIST?: {}".format(db_wiki["wiki"] in current_domain["irc"].updated))
logger.debug("LAST CHECK FOR THE WIKI {} IS {}".format(db_wiki["wiki"], all_wikis[db_wiki["wiki"]].last_check))
if db_wiki["wiki"] in current_domain["irc"].updated: # Priority wikis are the ones with IRC, if they get updated forcefully add them to queue
current_domain["irc"].updated.remove(db_wiki["wiki"])
current_domain["query"].append(QueuedWiki(db_wiki["wiki"], 20), forced=True)
logger.debug("Updated in IRC so adding to queue.")
continue
elif all_wikis[db_wiki["wiki"]].last_check+settings["irc_overtime"] < time.time(): # if time went by and wiki should be updated now use default mechanics
logger.debug("Overtime so adding to queue.")
pass pass
else: # Continue without adding if domain in full:
logger.debug("No condition fulfilled so skipping.") continue
try:
current_domain: dict = self[domain]
if current_domain["irc"]:
logger.debug("DOMAIN LIST FOR IRC: {}".format(current_domain["irc"].updated))
logger.debug("CURRENT DOMAIN INFO: {}".format(domain))
logger.debug("IS WIKI IN A LIST?: {}".format(db_wiki["wiki"] in current_domain["irc"].updated))
logger.debug("LAST CHECK FOR THE WIKI {} IS {}".format(db_wiki["wiki"], all_wikis[db_wiki["wiki"]].last_check))
if db_wiki["wiki"] in current_domain["irc"].updated: # Priority wikis are the ones with IRC, if they get updated forcefully add them to queue
current_domain["irc"].updated.remove(db_wiki["wiki"])
current_domain["query"].append(QueuedWiki(db_wiki["wiki"], 20), forced=True)
logger.debug("Updated in IRC so adding to queue.")
continue
elif all_wikis[db_wiki["wiki"]].last_check+settings["irc_overtime"] < time.time(): # if time went by and wiki should be updated now use default mechanics
logger.debug("Overtime so adding to queue.")
pass
else: # Continue without adding
logger.debug("No condition fulfilled so skipping.")
continue
if not db_wiki["rowid"] < current_domain["last_rowid"]:
current_domain["query"].append(QueuedWiki(db_wiki["wiki"], 20))
except KeyError:
await self.start_group(domain, [QueuedWiki(db_wiki["wiki"], 20)])
logger.info("A new domain group ({}) has been added since last time, adding it to the domain_list and starting a task...".format(domain))
except ListFull:
full.add(domain)
current_domain["last_rowid"] = db_wiki["rowid"]
continue continue
if not db_wiki["ROWID"] < current_domain["last_rowid"]:
current_domain["query"].append(QueuedWiki(db_wiki["wiki"], 20))
except KeyError:
await self.start_group(domain, [QueuedWiki(db_wiki["wiki"], 20)])
logger.info("A new domain group ({}) has been added since last time, adding it to the domain_list and starting a task...".format(domain))
except ListFull:
full.add(domain)
current_domain["last_rowid"] = db_wiki["ROWID"]
continue
for wiki in self.to_remove: for wiki in self.to_remove:
await self.remove_wiki_from_group(wiki) await self.remove_wiki_from_group(wiki)
for group, data in self.domain_list.items(): for group, data in self.domain_list.items():
@ -226,24 +230,29 @@ def calculate_delay_for_group(group_length: int) -> float:
return 0.0 return 0.0
def generate_targets(wiki_url: str, additional_requirements: str) -> defaultdict: async def generate_targets(wiki_url: str, additional_requirements: str) -> defaultdict:
"""To minimize the amount of requests, we generate a list of language/display mode combinations to create messages for """To minimize the amount of requests, we generate a list of language/display mode combinations to create messages for
this way we can send the same message to multiple webhooks which have the same wiki and settings without doing another this way we can send the same message to multiple webhooks which have the same wiki and settings without doing another
request to the wiki just to duplicate the message. request to the wiki just to duplicate the message.
""" """
combinations = defaultdict(list) combinations = defaultdict(list)
for webhook in db_cursor.execute('SELECT webhook, lang, display FROM rcgcdw WHERE wiki = ? {}'.format(additional_requirements), (wiki_url,)): async with db.pool().acquire() as connection:
combination = (webhook["lang"], webhook["display"]) async with connection.transaction():
combinations[combination].append(webhook["webhook"]) async for webhook in connection.cursor('SELECT webhook, lang, display FROM rcgcdw WHERE wiki = $1 {}'.format(additional_requirements), wiki_url):
combination = (webhook["lang"], webhook["display"])
combinations[combination].append(webhook["webhook"])
return combinations return combinations
async def generate_domain_groups(): async def generate_domain_groups():
"""Generate a list of wikis per domain (fandom.com, wikipedia.org etc.)""" """Generate a list of wikis per domain (fandom.com, wikipedia.org etc.)
:returns tuple[str, list]"""
domain_wikis = defaultdict(list) domain_wikis = defaultdict(list)
fetch_all = db_cursor.execute('SELECT ROWID, webhook, wiki, lang, display, rcid FROM rcgcdw WHERE rcid != -1 OR rcid IS NULL GROUP BY wiki ORDER BY ROWID ASC') async with db.pool().acquire() as connection:
for db_wiki in fetch_all.fetchall(): async with connection.transaction():
domain_wikis[get_domain(db_wiki["wiki"])].append(QueuedWiki(db_wiki["wiki"], 20)) async for db_wiki in connection.cursor('SELECT DISTINCT wiki, webhook, lang, display, rcid FROM rcgcdw WHERE rcid != -1 OR rcid IS NULL'):
domain_wikis[get_domain(db_wiki["wiki"])].append(QueuedWiki(db_wiki["wiki"], 20))
for group, db_wikis in domain_wikis.items(): for group, db_wikis in domain_wikis.items():
yield group, db_wikis yield group, db_wikis
@ -301,10 +310,10 @@ async def scan_group(group: str):
else: else:
local_wiki.rc_active = 0 local_wiki.rc_active = 0
DBHandler.add(queued_wiki.url, 0) DBHandler.add(queued_wiki.url, 0)
DBHandler.update_db() await DBHandler.update_db()
continue continue
categorize_events = {} categorize_events = {}
targets = generate_targets(queued_wiki.url, "AND (rcid != -1 OR rcid IS NULL)") targets = await generate_targets(queued_wiki.url, "AND (rcid != -1 OR rcid IS NULL)")
paths = get_paths(queued_wiki.url, recent_changes_resp) paths = get_paths(queued_wiki.url, recent_changes_resp)
new_events = 0 new_events = 0
local_wiki.last_check = time.time() # on successful check, save new last check time local_wiki.last_check = time.time() # on successful check, save new last check time
@ -347,7 +356,7 @@ async def scan_group(group: str):
if recent_changes: # we don't have to test for highest_rc being null, because if there are no RC entries recent_changes will be an empty list which will result in false in here and DO NOT save the value if recent_changes: # we don't have to test for highest_rc being null, because if there are no RC entries recent_changes will be an empty list which will result in false in here and DO NOT save the value
local_wiki.rc_active = highest_rc local_wiki.rc_active = highest_rc
DBHandler.add(queued_wiki.url, highest_rc) DBHandler.add(queued_wiki.url, highest_rc)
DBHandler.update_db() await DBHandler.update_db()
except asyncio.CancelledError: except asyncio.CancelledError:
return return
except QueueEmpty: except QueueEmpty:
@ -393,110 +402,109 @@ async def message_sender():
async def discussion_handler(): async def discussion_handler():
try: try:
while True: while True:
fetch_all = db_cursor.execute( async with db.pool().acquire() as connection:
"SELECT wiki, rcid, postid FROM rcgcdw WHERE postid != '-1' OR postid IS NULL GROUP BY wiki") async with connection.transaction():
for db_wiki in fetch_all.fetchall(): async for db_wiki in connection.cursor("SELECT DISTINCT wiki, rcid, postid FROM rcgcdw WHERE postid != '-1' OR postid IS NULL"):
try: try:
local_wiki = all_wikis[db_wiki["wiki"]] # set a reference to a wiki object from memory local_wiki = all_wikis[db_wiki["wiki"]] # set a reference to a wiki object from memory
except KeyError: except KeyError:
local_wiki = all_wikis[db_wiki["wiki"]] = Wiki() local_wiki = all_wikis[db_wiki["wiki"]] = Wiki()
local_wiki.rc_active = db_wiki["rcid"] local_wiki.rc_active = db_wiki["rcid"]
if db_wiki["wiki"] not in rcqueue.irc_mapping["fandom.com"].updated_discussions and local_wiki.last_discussion_check+settings["irc_overtime"] > time.time(): # I swear if another wiki farm ever starts using Fandom discussions I'm gonna use explosion magic if db_wiki["wiki"] not in rcqueue.irc_mapping["fandom.com"].updated_discussions and local_wiki.last_discussion_check+settings["irc_overtime"] > time.time(): # I swear if another wiki farm ever starts using Fandom discussions I'm gonna use explosion magic
continue continue
else:
try:
rcqueue.irc_mapping["fandom.com"].updated_discussions.remove(db_wiki["wiki"])
except KeyError:
pass # to be expected
header = settings["header"]
header["Accept"] = "application/hal+json"
async with aiohttp.ClientSession(headers=header,
timeout=aiohttp.ClientTimeout(6.0)) as session:
try:
feeds_response = await local_wiki.fetch_feeds(db_wiki["wiki"], session)
except (WikiServerError, WikiError):
continue # ignore this wiki if it throws errors
try:
discussion_feed_resp = await feeds_response.json(encoding="UTF-8")
if "error" in discussion_feed_resp:
error = discussion_feed_resp["error"]
if error == "NotFoundException": # Discussions disabled
if db_wiki["rcid"] != -1: # RC feed is disabled
db_cursor.execute("UPDATE rcgcdw SET postid = ? WHERE wiki = ?",
("-1", db_wiki["wiki"],))
else:
await local_wiki.remove(db_wiki["wiki"], 1000)
DBHandler.update_db()
continue
raise WikiError
discussion_feed = discussion_feed_resp["_embedded"]["doc:posts"]
discussion_feed.reverse()
except aiohttp.ContentTypeError:
logger.exception("Wiki seems to be resulting in non-json content.")
continue
except asyncio.TimeoutError:
logger.debug("Timeout on reading JSON of discussion post feeed.")
continue
except:
logger.exception("On loading json of response.")
continue
if db_wiki["postid"] is None: # new wiki, just get the last post to not spam the channel
if len(discussion_feed) > 0:
DBHandler.add(db_wiki["wiki"], discussion_feed[-1]["id"], True)
else:
DBHandler.add(db_wiki["wiki"], "0", True)
DBHandler.update_db()
continue
comment_events = []
targets = generate_targets(db_wiki["wiki"], "AND NOT postid = '-1'")
for post in discussion_feed:
if post["_embedded"]["thread"][0]["containerType"] == "ARTICLE_COMMENT" and post["id"] > db_wiki["postid"]:
comment_events.append(post["forumId"])
comment_pages: dict = {}
if comment_events:
try:
comment_pages = await local_wiki.safe_request(
"{wiki}wikia.php?controller=FeedsAndPosts&method=getArticleNamesAndUsernames&stablePageIds={pages}&format=json".format(
wiki=db_wiki["wiki"], pages=",".join(comment_events)
), RateLimiter(), "articleNames")
except aiohttp.ClientResponseError: # Fandom can be funny sometimes... See #30
comment_pages = None
except:
if command_line_args.debug:
logger.exception("Exception on Feeds article comment request")
shutdown(loop=asyncio.get_event_loop())
else: else:
logger.exception("Exception on Feeds article comment request")
await generic_msg_sender_exception_logger(traceback.format_exc(),
"Exception on Feeds article comment request",
Post=str(post)[0:1000], Wiki=db_wiki["wiki"])
message_list = defaultdict(list)
for post in discussion_feed: # Yeah, second loop since the comments require an extra request
if post["id"] > db_wiki["postid"]:
for target in targets.items():
try: try:
message = await essential_feeds(post, comment_pages, db_wiki, target) rcqueue.irc_mapping["fandom.com"].updated_discussions.remove(db_wiki["wiki"])
if message is not None: except KeyError:
message_list[target[0]].append(message) pass # to be expected
except asyncio.CancelledError: header = settings["header"]
raise header["Accept"] = "application/hal+json"
async with aiohttp.ClientSession(headers=header,
timeout=aiohttp.ClientTimeout(6.0)) as session:
try:
feeds_response = await local_wiki.fetch_feeds(db_wiki["wiki"], session)
except (WikiServerError, WikiError):
continue # ignore this wiki if it throws errors
try:
discussion_feed_resp = await feeds_response.json(encoding="UTF-8")
if "error" in discussion_feed_resp:
error = discussion_feed_resp["error"]
if error == "NotFoundException": # Discussions disabled
if db_wiki["rcid"] != -1: # RC feed is disabled
await connection.execute("UPDATE rcgcdw SET postid = $1 WHERE wiki = $2", "-1", db_wiki["wiki"])
else:
await local_wiki.remove(db_wiki["wiki"], 1000)
await DBHandler.update_db()
continue
raise WikiError
discussion_feed = discussion_feed_resp["_embedded"]["doc:posts"]
discussion_feed.reverse()
except aiohttp.ContentTypeError:
logger.exception("Wiki seems to be resulting in non-json content.")
continue
except asyncio.TimeoutError:
logger.debug("Timeout on reading JSON of discussion post feeed.")
continue
except:
logger.exception("On loading json of response.")
continue
if db_wiki["postid"] is None: # new wiki, just get the last post to not spam the channel
if len(discussion_feed) > 0:
DBHandler.add(db_wiki["wiki"], discussion_feed[-1]["id"], True)
else:
DBHandler.add(db_wiki["wiki"], "0", True)
await DBHandler.update_db()
continue
comment_events = []
targets = await generate_targets(db_wiki["wiki"], "AND NOT postid = '-1'")
for post in discussion_feed:
if post["_embedded"]["thread"][0]["containerType"] == "ARTICLE_COMMENT" and post["id"] > db_wiki["postid"]:
comment_events.append(post["forumId"])
comment_pages: dict = {}
if comment_events:
try:
comment_pages = await local_wiki.safe_request(
"{wiki}wikia.php?controller=FeedsAndPosts&method=getArticleNamesAndUsernames&stablePageIds={pages}&format=json".format(
wiki=db_wiki["wiki"], pages=",".join(comment_events)
), RateLimiter(), "articleNames")
except aiohttp.ClientResponseError: # Fandom can be funny sometimes... See #30
comment_pages = None
except: except:
if command_line_args.debug: if command_line_args.debug:
logger.exception("Exception on Feeds formatter") logger.exception("Exception on Feeds article comment request")
shutdown(loop=asyncio.get_event_loop()) shutdown(loop=asyncio.get_event_loop())
else: else:
logger.exception("Exception on Feeds formatter") logger.exception("Exception on Feeds article comment request")
await generic_msg_sender_exception_logger(traceback.format_exc(), "Exception in feed formatter", Post=str(post)[0:1000], Wiki=db_wiki["wiki"]) await generic_msg_sender_exception_logger(traceback.format_exc(),
# Lets stack the messages "Exception on Feeds article comment request",
for messages in message_list.values(): Post=str(post)[0:1000], Wiki=db_wiki["wiki"])
messages = stack_message_list(messages) message_list = defaultdict(list)
for message in messages: for post in discussion_feed: # Yeah, second loop since the comments require an extra request
await send_to_discord(message) if post["id"] > db_wiki["postid"]:
if discussion_feed: for target in targets.items():
DBHandler.add(db_wiki["wiki"], post["id"], True) try:
await asyncio.sleep(delay=2.0) # hardcoded really doesn't need much more message = await essential_feeds(post, comment_pages, db_wiki, target)
if message is not None:
message_list[target[0]].append(message)
except asyncio.CancelledError:
raise
except:
if command_line_args.debug:
logger.exception("Exception on Feeds formatter")
shutdown(loop=asyncio.get_event_loop())
else:
logger.exception("Exception on Feeds formatter")
await generic_msg_sender_exception_logger(traceback.format_exc(), "Exception in feed formatter", Post=str(post)[0:1000], Wiki=db_wiki["wiki"])
# Lets stack the messages
for messages in message_list.values():
messages = stack_message_list(messages)
for message in messages:
await send_to_discord(message)
if discussion_feed:
DBHandler.add(db_wiki["wiki"], post["id"], True)
await asyncio.sleep(delay=2.0) # hardcoded really doesn't need much more
await asyncio.sleep(delay=1.0) # Avoid lock on no wikis await asyncio.sleep(delay=1.0) # Avoid lock on no wikis
DBHandler.update_db() await DBHandler.update_db()
except asyncio.CancelledError: except asyncio.CancelledError:
pass pass
except: except:
@ -510,8 +518,6 @@ async def discussion_handler():
def shutdown(loop, signal=None): def shutdown(loop, signal=None):
global main_tasks global main_tasks
DBHandler.update_db()
db_connection.close()
loop.remove_signal_handler(signal) loop.remove_signal_handler(signal)
if len(messagequeue) > 0: if len(messagequeue) > 0:
logger.warning("Some messages are still queued!") logger.warning("Some messages are still queued!")
@ -543,6 +549,9 @@ async def main_loop():
global main_tasks global main_tasks
loop = asyncio.get_event_loop() loop = asyncio.get_event_loop()
nest_asyncio.apply(loop) nest_asyncio.apply(loop)
await db.setup_connection()
logger.debug("Connection type: {}".format(db.connection))
await populate_allwikis()
try: try:
signals = (signal.SIGHUP, signal.SIGTERM, signal.SIGINT) signals = (signal.SIGHUP, signal.SIGTERM, signal.SIGINT)
for s in signals: for s in signals:
@ -558,6 +567,8 @@ async def main_loop():
main_tasks["msg_queue_shield"] = asyncio.shield(main_tasks["message_sender"]) main_tasks["msg_queue_shield"] = asyncio.shield(main_tasks["message_sender"])
await asyncio.gather(main_tasks["wiki_scanner"], main_tasks["discussion_handler"], main_tasks["message_sender"]) await asyncio.gather(main_tasks["wiki_scanner"], main_tasks["discussion_handler"], main_tasks["message_sender"])
except KeyboardInterrupt: except KeyboardInterrupt:
await DBHandler.update_db()
await db.shutdown_connection()
shutdown(loop) shutdown(loop)
except asyncio.CancelledError: except asyncio.CancelledError:
return return

45
src/database.py
View file

@ -1,6 +1,43 @@
import sqlite3 import asyncpg
import logging
from typing import Optional
from src.config import settings from src.config import settings
db_connection = sqlite3.connect(settings.get("database_path", 'rcgcdb.db')) logger = logging.getLogger("rcgcdb.database")
db_connection.row_factory = sqlite3.Row # connection: Optional[asyncpg.Connection] = None
db_cursor = db_connection.cursor()
class db_connection:
connection: Optional[asyncpg.Pool] = None
async def setup_connection(self):
# Establish a connection to an existing database named "test"
# as a "postgres" user.
logger.debug("Setting up the Database connection...")
self.connection = await asyncpg.create_pool(user=settings["pg_user"], host=settings.get("pg_host", "localhost"),
database=settings.get("pg_db", "rcgcdb"), password=settings.get("pg_pass"),
port=settings.get("pg_port", 5432))
logger.debug("Database connection established! Connection: {}".format(self.connection))
async def shutdown_connection(self):
logger.debug("Shutting down database connection...")
await self.connection.close()
def pool(self) -> asyncpg.Pool:
return self.connection
# Tried to make it a decorator but tbh won't probably work
# async def in_transaction(self, func):
# async def single_transaction():
# async with self.connection.acquire() as connection:
# async with connection.transaction():
# await func()
# return single_transaction
# async def query(self, string, *arg):
# async with self.connection.acquire() as connection:
# async with connection.transaction():
# return connection.cursor(string, *arg)
db = db_connection()

29
src/discord.py
View file

@ -3,7 +3,7 @@ from collections import defaultdict
from src.misc import logger from src.misc import logger
from src.config import settings from src.config import settings
from src.database import db_cursor from src.database import db
from src.i18n import langs from src.i18n import langs
from src.exceptions import EmbedListFull from src.exceptions import EmbedListFull
from asyncio import TimeoutError from asyncio import TimeoutError
@ -22,17 +22,19 @@ default_header["X-RateLimit-Precision"] = "millisecond"
# User facing webhook functions # User facing webhook functions
async def wiki_removal(wiki_url, status): async def wiki_removal(wiki_url, status):
for observer in db_cursor.execute('SELECT webhook, lang FROM rcgcdw WHERE wiki = ?', (wiki_url,)): async with db.pool().acquire() as connection:
_ = langs[observer["lang"]]["discord"].gettext async with connection.transaction():
reasons = {410: _("wiki deleted"), 404: _("wiki deleted"), 401: _("wiki inaccessible"), async for observer in connection.cursor('SELECT webhook, lang FROM rcgcdw WHERE wiki = $1', wiki_url):
402: _("wiki inaccessible"), 403: _("wiki inaccessible"), 1000: _("discussions disabled")} _ = langs[observer["lang"]]["discord"].gettext
reason = reasons.get(status, _("unknown error")) reasons = {410: _("wiki deleted"), 404: _("wiki deleted"), 401: _("wiki inaccessible"),
await send_to_discord_webhook(DiscordMessage("compact", "webhook/remove", webhook_url=[], content=_("This recent changes webhook has been removed for `{reason}`!").format(reason=reason), wiki=None), webhook_url=observer["webhook"]) 402: _("wiki inaccessible"), 403: _("wiki inaccessible"), 1000: _("discussions disabled")}
header = settings["header"] reason = reasons.get(status, _("unknown error"))
header['Content-Type'] = 'application/json' await send_to_discord_webhook(DiscordMessage("compact", "webhook/remove", webhook_url=[], content=_("This recent changes webhook has been removed for `{reason}`!").format(reason=reason), wiki=None), webhook_url=observer["webhook"])
header['X-Audit-Log-Reason'] = "Wiki becoming unavailable" header = settings["header"]
async with aiohttp.ClientSession(headers=header, timeout=aiohttp.ClientTimeout(5.0)) as session: header['Content-Type'] = 'application/json'
await session.delete("https://discord.com/api/webhooks/"+observer["webhook"]) header['X-Audit-Log-Reason'] = "Wiki becoming unavailable"
async with aiohttp.ClientSession(headers=header, timeout=aiohttp.ClientTimeout(5.0)) as session:
await session.delete("https://discord.com/api/webhooks/"+observer["webhook"])
async def webhook_removal_monitor(webhook_url: str, reason: int): async def webhook_removal_monitor(webhook_url: str, reason: int):
@ -238,7 +240,8 @@ async def handle_discord_http(code: int, formatted_embed: str, result: aiohttp.C
return 1 return 1
elif code == 401 or code == 404: # HTTP UNAUTHORIZED AND NOT FOUND elif code == 401 or code == 404: # HTTP UNAUTHORIZED AND NOT FOUND
logger.error("Webhook URL is invalid or no longer in use, please replace it with proper one.") logger.error("Webhook URL is invalid or no longer in use, please replace it with proper one.")
db_cursor.execute("DELETE FROM rcgcdw WHERE webhook = ?", (webhook_url,)) async with db.pool().acquire() as connection:
await connection.execute("DELETE FROM rcgcdw WHERE webhook = $1", webhook_url)
await webhook_removal_monitor(webhook_url, code) await webhook_removal_monitor(webhook_url, code)
return 1 return 1
elif code == 429: elif code == 429:

21
src/queue_handler.py
View file

@ -1,5 +1,5 @@
import logging import logging
from src.database import db_cursor, db_connection from src.database import db
logger = logging.getLogger("rcgcdb.queue_handler") logger = logging.getLogger("rcgcdb.queue_handler")
@ -14,15 +14,16 @@ class UpdateDB:
def clear_list(self): def clear_list(self):
self.updated.clear() self.updated.clear()
def update_db(self): async def update_db(self):
for update in self.updated: async with db.pool().acquire() as connection:
if update[2] is None: async with connection.transaction():
sql = "UPDATE rcgcdw SET rcid = ? WHERE wiki = ? AND ( rcid != -1 OR rcid IS NULL )" for update in self.updated:
else: if update[2] is None:
sql = "UPDATE rcgcdw SET postid = ? WHERE wiki = ? AND ( postid != '-1' OR postid IS NULL )" sql = "UPDATE rcgcdw SET rcid = $2 WHERE wiki = $1 AND ( rcid != -1 OR rcid IS NULL )"
db_cursor.execute(sql, (update[1], update[0],)) else:
db_connection.commit() sql = "UPDATE rcgcdw SET postid = $2 WHERE wiki = $1 AND ( postid != '-1' OR postid IS NULL )"
self.clear_list() await connection.execute(sql, update[0], update[1])
self.clear_list()
DBHandler = UpdateDB() DBHandler = UpdateDB()

8
src/wiki.py
View file

@ -2,7 +2,7 @@ from dataclasses import dataclass
import re import re
import logging, aiohttp import logging, aiohttp
from src.exceptions import * from src.exceptions import *
from src.database import db_cursor, db_connection from src.database import db
from src.formatters.rc import embed_formatter, compact_formatter from src.formatters.rc import embed_formatter, compact_formatter
from src.formatters.discussions import feeds_embed_formatter, feeds_compact_formatter from src.formatters.discussions import feeds_embed_formatter, feeds_compact_formatter
from src.misc import parse_link from src.misc import parse_link
@ -109,9 +109,9 @@ class Wiki:
logger.info("Removing a wiki {}".format(wiki_url)) logger.info("Removing a wiki {}".format(wiki_url))
await src.discord.wiki_removal(wiki_url, reason) await src.discord.wiki_removal(wiki_url, reason)
await src.discord.wiki_removal_monitor(wiki_url, reason) await src.discord.wiki_removal_monitor(wiki_url, reason)
db_cursor.execute('DELETE FROM rcgcdw WHERE wiki = ?', (wiki_url,)) async with db.pool().acquire() as connection:
logger.warning('{} rows affected by DELETE FROM rcgcdw WHERE wiki = "{}"'.format(db_cursor.rowcount, wiki_url)) result = await connection.execute('DELETE FROM rcgcdw WHERE wiki = $1', wiki_url)
db_connection.commit() logger.warning('{} rows affected by DELETE FROM rcgcdw WHERE wiki = "{}"'.format(result, wiki_url))
async def pull_comment(self, comment_id, WIKI_API_PATH, rate_limiter): async def pull_comment(self, comment_id, WIKI_API_PATH, rate_limiter):
try: try: