mirror of
https://github.com/zulip/zulip.git
synced 2025-11-03 13:33:24 +00:00
df001db1a9
Black 23 enforces some slightly more specific rules about empty line counts and redundant parenthesis removal, but the result is still compatible with Black 22. (This does not actually upgrade our Python environment to Black 23 yet.) Signed-off-by: Anders Kaseorg <anders@zulip.com>
82 lines
3.5 KiB
Python
82 lines
3.5 KiB
Python
from typing import Dict, Iterable, List
|
|
|
|
import ahocorasick
|
|
from django.db import transaction
|
|
|
|
from zerver.lib.cache import (
|
|
cache_with_key,
|
|
realm_alert_words_automaton_cache_key,
|
|
realm_alert_words_cache_key,
|
|
)
|
|
from zerver.models import AlertWord, Realm, UserProfile, flush_realm_alert_words
|
|
|
|
|
|
@cache_with_key(realm_alert_words_cache_key, timeout=3600 * 24)
|
|
def alert_words_in_realm(realm: Realm) -> Dict[int, List[str]]:
|
|
user_ids_and_words = AlertWord.objects.filter(realm=realm, user_profile__is_active=True).values(
|
|
"user_profile_id", "word"
|
|
)
|
|
user_ids_with_words: Dict[int, List[str]] = {}
|
|
for id_and_word in user_ids_and_words:
|
|
user_ids_with_words.setdefault(id_and_word["user_profile_id"], [])
|
|
user_ids_with_words[id_and_word["user_profile_id"]].append(id_and_word["word"])
|
|
return user_ids_with_words
|
|
|
|
|
|
@cache_with_key(realm_alert_words_automaton_cache_key, timeout=3600 * 24)
|
|
def get_alert_word_automaton(realm: Realm) -> ahocorasick.Automaton:
|
|
user_id_with_words = alert_words_in_realm(realm)
|
|
alert_word_automaton = ahocorasick.Automaton()
|
|
for user_id, alert_words in user_id_with_words.items():
|
|
for alert_word in alert_words:
|
|
alert_word_lower = alert_word.lower()
|
|
if alert_word_automaton.exists(alert_word_lower):
|
|
(key, user_ids_for_alert_word) = alert_word_automaton.get(alert_word_lower)
|
|
user_ids_for_alert_word.add(user_id)
|
|
else:
|
|
alert_word_automaton.add_word(alert_word_lower, (alert_word_lower, {user_id}))
|
|
alert_word_automaton.make_automaton()
|
|
# If the kind is not AHOCORASICK after calling make_automaton, it means there is no key present
|
|
# and hence we cannot call items on the automaton yet. To avoid it we return None for such cases
|
|
# where there is no alert-words in the realm.
|
|
# https://pyahocorasick.readthedocs.io/en/latest/#make-automaton
|
|
if alert_word_automaton.kind != ahocorasick.AHOCORASICK:
|
|
return None
|
|
return alert_word_automaton
|
|
|
|
|
|
def user_alert_words(user_profile: UserProfile) -> List[str]:
|
|
return list(AlertWord.objects.filter(user_profile=user_profile).values_list("word", flat=True))
|
|
|
|
|
|
@transaction.atomic
|
|
def add_user_alert_words(user_profile: UserProfile, new_words: Iterable[str]) -> List[str]:
|
|
existing_words_lower = {word.lower() for word in user_alert_words(user_profile)}
|
|
|
|
# Keeping the case, use a dictionary to get the set of
|
|
# case-insensitive distinct, new alert words
|
|
word_dict: Dict[str, str] = {}
|
|
for word in new_words:
|
|
if word.lower() in existing_words_lower:
|
|
continue
|
|
word_dict[word.lower()] = word
|
|
|
|
AlertWord.objects.bulk_create(
|
|
AlertWord(user_profile=user_profile, word=word, realm=user_profile.realm)
|
|
for word in word_dict.values()
|
|
)
|
|
# Django bulk_create operations don't flush caches, so we need to do this ourselves.
|
|
flush_realm_alert_words(user_profile.realm)
|
|
|
|
return user_alert_words(user_profile)
|
|
|
|
|
|
@transaction.atomic
|
|
def remove_user_alert_words(user_profile: UserProfile, delete_words: Iterable[str]) -> List[str]:
|
|
# TODO: Ideally, this would be a bulk query, but Django doesn't have a `__iexact`.
|
|
# We can clean this up if/when PostgreSQL has more native support for case-insensitive fields.
|
|
# If we turn this into a bulk operation, we will need to call flush_realm_alert_words() here.
|
|
for delete_word in delete_words:
|
|
AlertWord.objects.filter(user_profile=user_profile, word__iexact=delete_word).delete()
|
|
return user_alert_words(user_profile)
|