perf: use slacc on check-word-mute (#10721)

* perf: use slacc on check-word-mute when all of specified words are single word

* perf: use slacc as possible

* build: avoid tarball

* chore: update slacc

* build: update package name
This commit is contained in:
Acid Chicken (硫酸鶏) 2023-05-05 19:49:34 +09:00 committed by GitHub
parent 14e364a74a
commit 4a72941eda
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
4 changed files with 236 additions and 6 deletions

View file

@ -34,7 +34,18 @@
"@swc/core-win32-ia32-msvc": "1.3.56",
"@swc/core-win32-x64-msvc": "1.3.56",
"@tensorflow/tfjs": "4.4.0",
"@tensorflow/tfjs-node": "4.4.0"
"@tensorflow/tfjs-node": "4.4.0",
"slacc-android-arm-eabi": "0.0.7",
"slacc-android-arm64": "0.0.7",
"slacc-darwin-arm64": "0.0.7",
"slacc-darwin-universal": "0.0.7",
"slacc-darwin-x64": "0.0.7",
"slacc-linux-arm-gnueabihf": "0.0.7",
"slacc-linux-arm64-gnu": "0.0.7",
"slacc-linux-arm64-musl": "0.0.7",
"slacc-linux-x64-gnu": "0.0.7",
"slacc-win32-arm64-msvc": "0.0.7",
"slacc-win32-x64-msvc": "0.0.7"
},
"dependencies": {
"@aws-sdk/client-s3": "3.321.1",
@ -128,6 +139,7 @@
"semver": "7.5.0",
"sharp": "0.32.1",
"sharp-read-bmp": "github:misskey-dev/sharp-read-bmp",
"slacc": "0.0.7",
"strict-event-emitter-types": "2.0.0",
"stringz": "2.1.0",
"summaly": "github:misskey-dev/summaly",

View file

@ -1,3 +1,4 @@
import { AhoCorasick } from 'slacc';
import RE2 from 're2';
import type { Note } from '@/models/entities/Note.js';
import type { User } from '@/models/entities/User.js';
@ -12,6 +13,8 @@ type UserLike = {
id: User['id'];
};
const acCache = new Map<string, AhoCorasick>();
export async function checkWordMute(note: NoteLike, me: UserLike | null | undefined, mutedWords: Array<string | string[]>): Promise<boolean> {
// 自分自身
if (me && (note.userId === me.id)) return false;
@ -21,7 +24,22 @@ export async function checkWordMute(note: NoteLike, me: UserLike | null | undefi
if (text === '') return false;
const matched = mutedWords.some(filter => {
const acable = mutedWords.filter(filter => Array.isArray(filter) && filter.length === 1).map(filter => filter[0]).sort();
const unacable = mutedWords.filter(filter => !Array.isArray(filter) || filter.length !== 1);
const acCacheKey = acable.join('\n');
const ac = acCache.get(acCacheKey) ?? AhoCorasick.withPatterns(acable);
acCache.delete(acCacheKey);
for (const obsoleteKeys of acCache.keys()) {
if (acCache.size > 1000) {
acCache.delete(obsoleteKeys);
}
}
acCache.set(acCacheKey, ac);
if (ac.isMatch(text)) {
return true;
}
const matched = unacable.some(filter => {
if (Array.isArray(filter)) {
return filter.every(keyword => text.includes(keyword));
} else {

View file

@ -0,0 +1,49 @@
import { checkWordMute } from '@/misc/check-word-mute.js';
describe(checkWordMute, () => {
describe('Slacc boost mode', () => {
it('should return false if mutedWords is empty', async () => {
expect(await checkWordMute({ userId: '1', text: 'foo' }, null, [])).toBe(false);
});
it('should return true if mutedWords is not empty and text contains muted word', async () => {
expect(await checkWordMute({ userId: '1', text: 'foo' }, null, [['foo']])).toBe(true);
});
it('should return false if mutedWords is not empty and text does not contain muted word', async () => {
expect(await checkWordMute({ userId: '1', text: 'foo' }, null, [['bar']])).toBe(false);
});
it('should return false when the note is written by me even if mutedWords is not empty and text contains muted word', async () => {
expect(await checkWordMute({ userId: '1', text: 'foo' }, { id: '1' }, [['foo']])).toBe(false);
});
it('should return true if mutedWords is not empty and text contains muted word in CW', async () => {
expect(await checkWordMute({ userId: '1', text: 'foo', cw: 'bar' }, null, [['bar']])).toBe(true);
});
it('should return true if mutedWords is not empty and text contains muted word in both CW and text', async () => {
expect(await checkWordMute({ userId: '1', text: 'foo', cw: 'bar' }, null, [['foo'], ['bar']])).toBe(true);
});
it('should return true if mutedWords is not empty and text does not contain muted word in both CW and text', async () => {
expect(await checkWordMute({ userId: '1', text: 'foo', cw: 'bar' }, null, [['foo'], ['baz']])).toBe(true);
});
});
describe('normal mode', () => {
it('should return false if text does not contain muted words', async () => {
expect(await checkWordMute({ userId: '1', text: 'foo' }, null, [['foo', 'bar']])).toBe(false);
});
it('should return true if text contains muted words', async () => {
expect(await checkWordMute({ userId: '1', text: 'foobar' }, null, [['foo', 'bar']])).toBe(true);
});
it('should return false when the note is written by me even if text contains muted words', async () => {
expect(await checkWordMute({ userId: '1', text: 'foo bar' }, { id: '1' }, [['foo', 'bar']])).toBe(false);
});
});
describe('RegExp mode', () => {
it('should return false if text does not contain muted words', async () => {
expect(await checkWordMute({ userId: '1', text: 'foo' }, null, ['/bar/'])).toBe(false);
});
it('should return true if text contains muted words', async () => {
expect(await checkWordMute({ userId: '1', text: 'foobar' }, null, ['/bar/'])).toBe(true);
});
it('should return false when the note is written by me even if text contains muted words', async () => {
expect(await checkWordMute({ userId: '1', text: 'foo bar' }, { id: '1' }, ['/bar/'])).toBe(false);
});
});
});

View file

@ -337,6 +337,9 @@ importers:
sharp-read-bmp:
specifier: github:misskey-dev/sharp-read-bmp
version: github.com/misskey-dev/sharp-read-bmp/02d9dc189fa7df0c4bea09330be26741772dac01
slacc:
specifier: 0.0.7
version: 0.0.7
strict-event-emitter-types:
specifier: 2.0.0
version: 2.0.0
@ -345,7 +348,7 @@ importers:
version: 2.1.0
summaly:
specifier: github:misskey-dev/summaly
version: github.com/misskey-dev/summaly/c7d71a9ec2467268b3911dc2ac805c2b8a898d3e
version: github.com/misskey-dev/summaly/2d63e2a0066f89871e777cc81d43c1ade8c97517
systeminformation:
specifier: 5.17.12
version: 5.17.12
@ -434,6 +437,39 @@ importers:
'@tensorflow/tfjs-node':
specifier: 4.4.0
version: 4.4.0(seedrandom@3.0.5)
slacc-android-arm-eabi:
specifier: 0.0.7
version: 0.0.7
slacc-android-arm64:
specifier: 0.0.7
version: 0.0.7
slacc-darwin-arm64:
specifier: 0.0.7
version: 0.0.7
slacc-darwin-universal:
specifier: 0.0.7
version: 0.0.7
slacc-darwin-x64:
specifier: 0.0.7
version: 0.0.7
slacc-linux-arm-gnueabihf:
specifier: 0.0.7
version: 0.0.7
slacc-linux-arm64-gnu:
specifier: 0.0.7
version: 0.0.7
slacc-linux-arm64-musl:
specifier: 0.0.7
version: 0.0.7
slacc-linux-x64-gnu:
specifier: 0.0.7
version: 0.0.7
slacc-win32-arm64-msvc:
specifier: 0.0.7
version: 0.0.7
slacc-win32-x64-msvc:
specifier: 0.0.7
version: 0.0.7
devDependencies:
'@jest/globals':
specifier: 29.5.0
@ -948,7 +984,7 @@ importers:
version: github.com/misskey-dev/storybook-addon-misskey-theme/cf583db098365b2ccc81a82f63ca9c93bc32b640(@storybook/blocks@7.0.7)(@storybook/components@7.0.7)(@storybook/core-events@7.0.7)(@storybook/manager-api@7.0.7)(@storybook/preview-api@7.0.7)(@storybook/theming@7.0.7)(@storybook/types@7.0.7)(react-dom@18.2.0)(react@18.2.0)
summaly:
specifier: github:misskey-dev/summaly
version: github.com/misskey-dev/summaly/c7d71a9ec2467268b3911dc2ac805c2b8a898d3e
version: github.com/misskey-dev/summaly/2d63e2a0066f89871e777cc81d43c1ade8c97517
vite-plugin-turbosnap:
specifier: 1.0.2
version: 1.0.2
@ -17783,6 +17819,121 @@ packages:
resolution: {integrity: sha512-bLGGlR1QxBcynn2d5YmDX4MGjlZvy2MRBDRNHLJ8VI6l6+9FUiyTFNJ0IveOSP0bcXgVDPRcfGqA0pjaqUpfVg==}
dev: true
/slacc-android-arm-eabi@0.0.7:
resolution: {integrity: sha512-6TikZlR1jsQscxwphhrf0U4xbsRy6zKJ0zmEULopTzbohgo5OLdZ7L3tQazkYlaaFe3YjGnVLW3FfGhhrajVog==}
engines: {node: '>= 10'}
cpu: [arm]
os: [android]
requiresBuild: true
dev: false
optional: true
/slacc-android-arm64@0.0.7:
resolution: {integrity: sha512-aol/9Rg0Hfqu81hpK+HXcx9sGYu4qqYU+djBCgLtb8I6ZMdWUdE0dp8ACBoTOmYn34hYGcUu4FlJUZ8r7Utucg==}
engines: {node: '>= 10'}
cpu: [arm64]
os: [android]
requiresBuild: true
dev: false
optional: true
/slacc-darwin-arm64@0.0.7:
resolution: {integrity: sha512-PkV7rO/c9AImNYDacP+kxtOjVuxjy06IIOAxbWerIWvoeqsCNRtiF/dh+OqIACRFBuHIDe0oAyUCEMGUTnzjyQ==}
engines: {node: '>= 10'}
cpu: [arm64]
os: [darwin]
requiresBuild: true
dev: false
optional: true
/slacc-darwin-universal@0.0.7:
resolution: {integrity: sha512-Y9zXpL40m4Yq3dE5vdnAgfmn0Fxc0Bf0ixC9TSl96gKeIZEd6drkjfpHFdsIDNImzOksIAUo0HHiDdbEfE7zdQ==}
engines: {node: '>= 10'}
os: [darwin]
requiresBuild: true
dev: false
optional: true
/slacc-darwin-x64@0.0.7:
resolution: {integrity: sha512-yKaGjX2YJl1QHe4NgqQVsY83jees3hjFxEUPoKpuZEQzWbMNn0XSyceFRGXIk1oDqiKU40UcsdcCedjYjSEd0Q==}
engines: {node: '>= 10'}
cpu: [x64]
os: [darwin]
requiresBuild: true
dev: false
optional: true
/slacc-linux-arm-gnueabihf@0.0.7:
resolution: {integrity: sha512-pdWMdQeX6uA9JfSoWo9EHH0yRiwXKMbaKoS9gflDSyt/hjeR3Qx/KK7Wihd7HeXx7njlNdpr9ycTRmm5NgapQQ==}
engines: {node: '>= 10'}
cpu: [arm]
os: [linux]
requiresBuild: true
dev: false
optional: true
/slacc-linux-arm64-gnu@0.0.7:
resolution: {integrity: sha512-hz9TK/w6fxeNZXyFzuLq5cJD/XRyJbo6BaIdW+VrKKnb9nkLnWlqDQtdtJk7Fw7zHjdY3Uqufjwm0iT6qBVpUQ==}
engines: {node: '>= 10'}
cpu: [arm64]
os: [linux]
requiresBuild: true
dev: false
optional: true
/slacc-linux-arm64-musl@0.0.7:
resolution: {integrity: sha512-wCDAYL7e+lh3XL7g87Ui/Bb2Ap9GcBqeJuj2yHIx6MYC8ontwFSXhqRTmd2zmPLmZA5Nc11aKGN11YNu0Pnwlw==}
engines: {node: '>= 10'}
cpu: [arm64]
os: [linux]
requiresBuild: true
dev: false
optional: true
/slacc-linux-x64-gnu@0.0.7:
resolution: {integrity: sha512-E5+2cveizpfHXCk/Hu5VfslWFeDVw47nywODiJ8CsofT2l5ITfYPMFEBXm9ORY25mGBTgsO6lJYiF9Hz4FlS9Q==}
engines: {node: '>= 10'}
cpu: [x64]
os: [linux]
requiresBuild: true
dev: false
optional: true
/slacc-win32-arm64-msvc@0.0.7:
resolution: {integrity: sha512-3a+qnkZbP+Pr5RZuzd0Vi1uCal137QiJajRAWT4r7qwu+Zidd50x2oikQ4rAegqZVTm8qTwVmWA+WmH8WHI7iw==}
engines: {node: '>= 10'}
cpu: [arm64]
os: [win32]
requiresBuild: true
dev: false
optional: true
/slacc-win32-x64-msvc@0.0.7:
resolution: {integrity: sha512-ydFdZ7wEXQPsw2Tg+yG9uJdCGTehyPtrWBVUMa7fojr3j1gbtThXS2l9Ad/6fYYi2VwdaYPLWbwV3GYElPGL8g==}
engines: {node: '>= 10'}
cpu: [x64]
os: [win32]
requiresBuild: true
dev: false
optional: true
/slacc@0.0.7:
resolution: {integrity: sha512-rwi2F3oJaGPST9JdCoUd5fnSZaoZFgTL00GFKhKufT48uwtUEAHlOL0t8gEVmon71X+53f9nEdsGWhwtOutJTQ==}
engines: {node: '>= 10'}
optionalDependencies:
slacc-android-arm-eabi: 0.0.7
slacc-android-arm64: 0.0.7
slacc-darwin-arm64: 0.0.7
slacc-darwin-universal: 0.0.7
slacc-darwin-x64: 0.0.7
slacc-linux-arm-gnueabihf: 0.0.7
slacc-linux-arm64-gnu: 0.0.7
slacc-linux-arm64-musl: 0.0.7
slacc-linux-x64-gnu: 0.0.7
slacc-win32-arm64-msvc: 0.0.7
slacc-win32-x64-msvc: 0.0.7
dev: false
/slash@3.0.0:
resolution: {integrity: sha512-g9Q1haeby36OSStwb4ntCGGGaKsaVSjQ68fBxoQcutl5fS1vuY18H3wSt3jFyFtrkx+Kz0V1G85A4MyAdDMi2Q==}
engines: {node: '>=8'}
@ -20248,8 +20399,8 @@ packages:
react-dom: 18.2.0(react@18.2.0)
dev: true
github.com/misskey-dev/summaly/c7d71a9ec2467268b3911dc2ac805c2b8a898d3e:
resolution: {tarball: https://codeload.github.com/misskey-dev/summaly/tar.gz/c7d71a9ec2467268b3911dc2ac805c2b8a898d3e}
github.com/misskey-dev/summaly/2d63e2a0066f89871e777cc81d43c1ade8c97517:
resolution: {tarball: https://codeload.github.com/misskey-dev/summaly/tar.gz/2d63e2a0066f89871e777cc81d43c1ade8c97517}
name: summaly
version: 4.0.2
dependencies: