GithubFollow/Surge_by_SukkaW
1
0
Fork 0
mirror of https://github.com/SukkaW/Surge.git synced 2025-12-13 01:30:37 +08:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

Simplify Trie

Browse Source
This commit is contained in:
SukkaW 2024-10-15 16:16:56 +08:00
parent e6f7a98ee9
commit b742701f59
3 changed files with 231 additions and 159 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
Build/lib/rules/base.ts
View File

@ -12,7 +12,7 @@ import { readFileByLine } from '../fetch-text-by-line';
import { asyncWriteToStream } from '../async-write-to-stream';
export abstract class RuleOutput<TPreprocessed = unknown> {
protected domainTrie = createTrie<string>(null, true);
protected domainTrie = createTrie(null, true);
protected domainKeywords = new Set<string>();
protected domainWildcard = new Set<string>();
protected userAgent = new Set<string>();
@ -97,7 +97,7 @@ export abstract class RuleOutput<TPreprocessed = unknown> {
}
addDomain(domain: string) {
this.domainTrie.add(domain, domain);
this.domainTrie.add(domain);
return this;
}
@ -109,7 +109,8 @@ export abstract class RuleOutput<TPreprocessed = unknown> {
}
addDomainSuffix(domain: string) {
return this.addDomain(domain[0] === '.' ? domain : '.' + domain);
this.domainTrie.add(domain, true);
return this;
}
bulkAddDomainSuffix(domains: string[]) {

131
Build/lib/trie.test.ts
View File

@ -1,37 +1,37 @@
import { createTrie, hostnameToTokens } from './trie';
import { createTrie } from './trie';
import { describe, it } from 'mocha';
import { expect } from 'chai';
describe('hostname to tokens', () => {
it('should split hostname into tokens.', () => {
expect(hostnameToTokens('.blog.skk.moe')).to.deep.equal([
'.',
'blog',
'.',
'skk',
'.',
'moe'
]);
// describe('hostname to tokens', () => {
// it('should split hostname into tokens.', () => {
// expect(hostnameToTokens('.blog.skk.moe')).to.deep.equal([
// '.',
// 'blog',
// '.',
// 'skk',
// '.',
// 'moe'
// ]);
expect(hostnameToTokens('blog.skk.moe')).to.deep.equal([
'blog',
'.',
'skk',
'.',
'moe'
]);
// expect(hostnameToTokens('blog.skk.moe')).to.deep.equal([
// 'blog',
// '.',
// 'skk',
// '.',
// 'moe'
// ]);
expect(hostnameToTokens('skk.moe')).to.deep.equal([
'skk',
'.',
'moe'
]);
// expect(hostnameToTokens('skk.moe')).to.deep.equal([
// 'skk',
// '.',
// 'moe'
// ]);
expect(hostnameToTokens('moe')).to.deep.equal([
'moe'
]);
});
});
// expect(hostnameToTokens('moe')).to.deep.equal([
// 'moe'
// ]);
// });
// });
describe('Trie', () => {
it('should be possible to add domains to a Trie.', () => {
@ -43,12 +43,12 @@ describe('Trie', () => {
expect(trie.size).to.equal(3);
expect(trie.has('a.skk.moe')).to.equal(true);
expect(trie.has('skk.moe')).to.equal(true);
expect(trie.has('anotherskk.moe')).to.equal(true);
expect(trie.has('example.com')).to.equal(false);
expect(trie.has('skk.mo')).to.equal(false);
expect(trie.has('another.skk.moe')).to.equal(false);
expect(trie.has('a.skk.moe'), 'a.skk.moe').to.equal(true);
expect(trie.has('skk.moe'), 'skk.moe').to.equal(true);
expect(trie.has('anotherskk.moe'), 'anotherskk.moe').to.equal(true);
expect(trie.has('example.com'), 'example.com').to.equal(false);
expect(trie.has('skk.mo'), 'skk.mo').to.equal(false);
expect(trie.has('another.skk.moe'), 'another.skk.moe').to.equal(false);
});
it('adding the same item several times should not increase size.', () => {
@ -78,6 +78,7 @@ describe('Trie', () => {
const trie = createTrie(null, false);
trie.add('skk.moe');
trie.add('blog.skk.moe');
trie.add('example.com');
trie.add('moe.sb');
@ -89,12 +90,12 @@ describe('Trie', () => {
expect(trie.has('skk.moe')).to.equal(false);
expect(trie.has('moe.sb')).to.equal(true);
expect(trie.size).to.equal(2);
expect(trie.size).to.equal(3);
expect(trie.delete('example.com')).to.equal(true);
expect(trie.size).to.equal(1);
expect(trie.size).to.equal(2);
expect(trie.delete('moe.sb')).to.equal(true);
expect(trie.size).to.equal(0);
expect(trie.size).to.equal(1);
});
it('should be possible to check the existence of a sequence in the Trie.', () => {
@ -116,15 +117,15 @@ describe('Trie', () => {
trie.add('cdn.example.com');
trie.add('example.org');
expect(trie.find('example.com')).to.deep.equal(['example.com', 'cdn.example.com', 'blog.example.com']);
expect(trie.find('com')).to.deep.equal(['example.com', 'cdn.example.com', 'blog.example.com']);
expect(trie.find('.example.com')).to.deep.equal(['cdn.example.com', 'blog.example.com']);
expect(trie.find('org')).to.deep.equal(['example.org']);
expect(trie.find('example.net')).to.deep.equal([]);
expect(trie.find('')).to.deep.equal(['example.org', 'example.com', 'cdn.example.com', 'blog.example.com']);
expect(trie.find('example.com'), 'example.com').to.deep.equal(['example.com', 'cdn.example.com', 'blog.example.com']);
expect(trie.find('com'), 'com').to.deep.equal(['example.com', 'cdn.example.com', 'blog.example.com']);
expect(trie.find('.example.com'), '.example.com').to.deep.equal(['cdn.example.com', 'blog.example.com']);
expect(trie.find('org'), 'prg').to.deep.equal(['example.org']);
expect(trie.find('example.net'), 'example.net').to.deep.equal([]);
expect(trie.find(''), '').to.deep.equal(['example.org', 'example.com', 'cdn.example.com', 'blog.example.com']);
});
it('should be possible to retrieve items matching the given prefix even with a smol trie.', () => {
it('should be possible to retrieve items matching the given prefix even with a smol trie', () => {
const trie = createTrie(null, true);
trie.add('.example.com');
@ -206,7 +207,7 @@ describe('smol tree', () => {
]);
});
it('should create simple tree - 2', () => {
it('should create simple tree - 3', () => {
const trie = createTrie([
'.blog.sub.example.com', 'cdn.sub.example.com', '.sub.example.com'
], true);
@ -258,10 +259,16 @@ describe('smol tree', () => {
'skk.moe',
'anotherskk.moe',
'blog.anotherskk.moe',
'blog.skk.moe'
'blog.skk.moe',
'.cdn.local',
'blog.img.skk.local',
'img.skk.local'
], true);
expect(trie.dump()).to.deep.equal([
expect(trie.dump(), '1').to.deep.equal([
'img.skk.local',
'blog.img.skk.local',
'.cdn.local',
'anotherskk.moe',
'blog.anotherskk.moe',
'skk.moe',
@ -270,20 +277,44 @@ describe('smol tree', () => {
trie.whitelist('.skk.moe');
expect(trie.dump()).to.deep.equal([
expect(trie.dump(), '2').to.deep.equal([
'img.skk.local',
'blog.img.skk.local',
'.cdn.local',
'anotherskk.moe',
'blog.anotherskk.moe'
]);
trie.whitelist('anotherskk.moe');
expect(trie.dump()).to.deep.equal([
expect(trie.dump(), '3').to.deep.equal([
'img.skk.local',
'blog.img.skk.local',
'.cdn.local',
'blog.anotherskk.moe'
]);
trie.add('anotherskk.moe');
trie.whitelist('.anotherskk.moe');
expect(trie.dump()).to.deep.equal([]);
expect(trie.dump(), '4').to.deep.equal([
'img.skk.local',
'blog.img.skk.local',
'.cdn.local'
]);
trie.whitelist('img.skk.local');
expect(trie.dump(), '5').to.deep.equal([
'blog.img.skk.local',
'.cdn.local'
]);
trie.whitelist('cdn.local');
expect(trie.dump(), '6').to.deep.equal([
'blog.img.skk.local'
]);
trie.whitelist('.skk.local');
expect(trie.dump(), '7').to.deep.equal([]);
});
it('should whitelist trie correctly', () => {

244
Build/lib/trie.ts
View File

@ -7,7 +7,8 @@ import util from 'node:util';
import { noop } from 'foxact/noop';
type TrieNode<Meta = any> = [
boolean, /** sentinel */
boolean, /** end */
boolean, /** includeAllSubdoain (.example.org, ||example.com) */
TrieNode | null, /** parent */
Map<string, TrieNode>, /** children */
Meta /** meta */
@ -19,59 +20,56 @@ function deepTrieNodeToJSON(node: TrieNode,
if (node[0]) {
obj['[start]'] = node[0];
}
if (node[3] != null) {
obj['[subdomain]'] = node[1];
if (node[4] != null) {
if (unpackMeta) {
obj['[meta]'] = unpackMeta(node[3]);
} else {
obj['[meta]'] = node[3];
}
}
node[2].forEach((value, key) => {
node[3].forEach((value, key) => {
obj[key] = deepTrieNodeToJSON(value, unpackMeta);
});
return obj;
}
const createNode = <Meta = any>(parent: TrieNode | null = null): TrieNode => [false, parent, new Map<string, TrieNode>(), null] as TrieNode<Meta>;
const createNode = <Meta = any>(allSubdomain = false, parent: TrieNode | null = null): TrieNode => [false, allSubdomain, parent, new Map<string, TrieNode>(), null] as TrieNode<Meta>;
export function hostnameToTokens(hostname: string): string[] {
const tokens = hostname.split('.');
const results: string[] = [];
let token = '';
for (let i = 0, l = tokens.length; i < l; i++) {
if (i > 0) {
results.push('.');
}
for (let i = 0, l = tokens.length; i < l; i++) {
token = tokens[i];
if (token.length > 0) {
results.push(token);
}
}
return results;
}
function walkHostnameTokens(hostname: string, onToken: (token: string) => boolean | null): boolean | null {
const tokens = hostname.split('.');
let token = '';
const l = tokens.length - 1;
// we are at the first of hostname, no splitor there
let token = '';
for (let i = l; i >= 0; i--) {
if (
i < l // when i === l, we are at the first of hostname, no splitor there
// when onToken returns true, we should skip the rest of the loop
&& onToken('.')
) {
token = tokens[i];
if (token.length > 0) {
const t = onToken(token);
if (t === null) {
return null;
}
// if the callback returns true, we should skip the rest
if (t) {
return true;
}
token = tokens[i];
if (
token.length > 0
// when onToken returns true, we should skip the rest of the loop
&& onToken(token)
) {
return true;
}
}
@ -104,7 +102,7 @@ abstract class Triebase<Meta = any> {
}
}
public abstract add(suffix: string, meta?: Meta): void;
public abstract add(suffix: string, includeAllSubdoain?: boolean, meta?: Meta): void;
protected walkIntoLeafWithTokens(
tokens: string[],
@ -124,8 +122,8 @@ abstract class Triebase<Meta = any> {
parent = node;
if (node[2].has(token)) {
node = node[2].get(token)!;
if (node[3].has(token)) {
node = node[3].get(token)!;
} else {
return null;
}
@ -144,14 +142,14 @@ abstract class Triebase<Meta = any> {
let parent: TrieNode = node;
const onToken = (token: string) => {
if (token === '') {
return true;
}
// if (token === '') {
// return true;
// }
parent = node;
if (node[2].has(token)) {
node = node[2].get(token)!;
if (node[3].has(token)) {
node = node[3].get(token)!;
} else {
return null;
}
@ -168,10 +166,18 @@ abstract class Triebase<Meta = any> {
return { node, parent };
};
public contains(suffix: string): boolean { return this.walkIntoLeafWithSuffix(suffix) !== null; };
public contains(suffix: string, includeAllSubdoain = suffix[0] === '.'): boolean {
if (suffix[0] === '.') {
suffix = suffix.slice(1);
}
const res = this.walkIntoLeafWithSuffix(suffix);
if (!res) return false;
if (includeAllSubdoain) return res.node[1];
return true;
};
private walk(
onMatches: (suffix: string[], meta: Meta) => void,
onMatches: (suffix: string[], subdomain: boolean, meta: Meta) => void,
initialNode = this.$root,
initialSuffix: string[] = []
) {
@ -185,7 +191,7 @@ abstract class Triebase<Meta = any> {
node = nodeStack.pop()!;
const suffix = suffixStack.pop()!;
node[2].forEach((childNode, k) => {
node[3].forEach((childNode, k) => {
// Pushing the child node to the stack for next iteration of DFS
nodeStack.push(childNode);
@ -194,7 +200,7 @@ abstract class Triebase<Meta = any> {
// If the node is a sentinel, we push the suffix to the results
if (node[0]) {
onMatches(suffix, node[3]);
onMatches(suffix, node[1], node[4]);
}
} while (nodeStack.length);
};
@ -208,7 +214,7 @@ abstract class Triebase<Meta = any> {
// Even if the node size is 1, but the single child is ".", we should retain the branch
// Since the "." could be special if it is the leaf-est node
const onlyChild = node[2].size < 2 && !node[2].has('.');
const onlyChild = node[3].size === 0 && !node[2];
if (toPrune != null) { // the top-est branch that could potentially being pruned
if (!onlyChild) {
@ -236,26 +242,29 @@ abstract class Triebase<Meta = any> {
*/
public find(
inputSuffix: string,
/** @default true */ includeEqualWithSuffix = true
subdomainOnly = inputSuffix[0] === '.'
// /** @default true */ includeEqualWithSuffix = true
): string[] {
// if (smolTree) {
// throw new Error('A Trie with smolTree enabled cannot perform find!');
// }
if (inputSuffix[0] === '.') {
inputSuffix = inputSuffix.slice(1);
}
const inputTokens = hostnameToTokens(inputSuffix);
const res = this.walkIntoLeafWithTokens(inputTokens);
if (res === null) return [];
const matches: string[][] = [];
const results: string[] = [];
const onMatches = includeEqualWithSuffix
// fast path (default option)
? (suffix: string[]) => matches.push(suffix)
// slow path
: (suffix: string[]) => {
if (!deepEqualArray(suffix, inputTokens)) {
matches.push(suffix);
const onMatches = subdomainOnly
? (suffix: string[], subdomain: boolean) => { // fast path (default option)
const d = fastStringArrayJoin(suffix, '.');
if (!subdomain && d === inputSuffix) return;
results.push(subdomain ? '.' + d : d);
}
: (suffix: string[], subdomain: boolean) => { // fast path (default option)
const d = fastStringArrayJoin(suffix, '.');
results.push(subdomain ? '.' + d : d);
};
this.walk(
@ -264,7 +273,7 @@ abstract class Triebase<Meta = any> {
inputTokens
);
return matches.map((m) => fastStringArrayJoin(m, ''));
return results;
};
/**
@ -280,7 +289,7 @@ abstract class Triebase<Meta = any> {
const { node, toPrune, tokenToPrune } = res;
if (tokenToPrune && toPrune) {
toPrune[2].delete(tokenToPrune);
toPrune[3].delete(tokenToPrune);
} else {
node[0] = false;
}
@ -288,18 +297,23 @@ abstract class Triebase<Meta = any> {
return true;
};
// eslint-disable-next-line @typescript-eslint/unbound-method -- alias class methods
// eslint-disable-next-line @typescript-eslint/unbound-method -- safe
public delete = this.remove;
/**
* Method used to assert whether the given prefix exists in the Trie.
*/
public has(suffix: string): boolean {
public has(suffix: string, includeAllSubdoain = suffix[0] === '.'): boolean {
if (suffix[0] === '.') {
suffix = suffix.slice(1);
}
const res = this.walkIntoLeafWithSuffix(suffix);
return res
? res.node[0]
: false;
if (res === null) return false;
if (!res.node[0]) return false;
if (includeAllSubdoain) return res.node[1];
return true;
};
public dump(onSuffix: (suffix: string) => void): void;
@ -308,8 +322,14 @@ abstract class Triebase<Meta = any> {
const results: string[] = [];
const handleSuffix = onSuffix
? (suffix: string[]) => onSuffix(fastStringArrayJoin(suffix, ''))
: (suffix: string[]) => results.push(fastStringArrayJoin(suffix, ''));
? (suffix: string[], subdomain: boolean) => {
const d = fastStringArrayJoin(suffix, '.');
onSuffix(subdomain ? '.' + d : d);
}
: (suffix: string[], subdomain: boolean) => {
const d = fastStringArrayJoin(suffix, '.');
results.push(subdomain ? '.' + d : d);
};
this.walk(handleSuffix);
@ -322,8 +342,8 @@ abstract class Triebase<Meta = any> {
const results: Meta[] = [];
const handleMeta = onMeta
? (_suffix: string[], meta: Meta) => onMeta(meta)
: (_suffix: string[], meta: Meta) => results.push(meta);
? (_suffix: string[], _subdomain: boolean, meta: Meta) => onMeta(meta)
: (_suffix: string[], _subdomain: boolean, meta: Meta) => results.push(meta);
this.walk(handleMeta);
@ -331,13 +351,19 @@ abstract class Triebase<Meta = any> {
};
public dumpWithMeta(onSuffix: (suffix: string, meta: Meta | undefined) => void): void;
public dumpWithMeta(): string[];
public dumpWithMeta(onSuffix?: (suffix: string, meta: Meta | undefined) => void): string[] | void {
const results: string[] = [];
public dumpWithMeta(): Array<[string, Meta | undefined]>;
public dumpWithMeta(onSuffix?: (suffix: string, meta: Meta | undefined) => void): Array<[string, Meta | undefined]> | void {
const results: Array<[string, Meta | undefined]> = [];
const handleSuffix = onSuffix
? (suffix: string[], meta: Meta | undefined) => onSuffix(fastStringArrayJoin(suffix, ''), meta)
: (suffix: string[]) => results.push(fastStringArrayJoin(suffix, ''));
? (suffix: string[], subdomain: boolean, meta: Meta | undefined) => {
const d = fastStringArrayJoin(suffix, '.');
return onSuffix(subdomain ? '.' + d : d, meta);
}
: (suffix: string[], subdomain: boolean, meta: Meta | undefined) => {
const d = fastStringArrayJoin(suffix, '.');
results.push([subdomain ? '.' + d : d, meta]);
};
this.walk(handleSuffix);
@ -359,21 +385,25 @@ abstract class Triebase<Meta = any> {
export class HostnameSmolTrie<Meta = any> extends Triebase<Meta> {
public smolTree = true;
add(suffix: string, meta?: Meta): void {
add(suffix: string, includeAllSubdoain = suffix[0] === '.', meta?: Meta): void {
let node: TrieNode<Meta> = this.$root;
let curNodeChildren: Map<string, TrieNode<Meta>> = node[2];
let curNodeChildren: Map<string, TrieNode<Meta>> = node[3];
if (suffix[0] === '.') {
suffix = suffix.slice(1);
}
const onToken = (token: string) => {
curNodeChildren = node[2];
curNodeChildren = node[3];
if (curNodeChildren.has(token)) {
node = curNodeChildren.get(token)!;
// During the adding of `[start]blog|.skk.moe` and find out that there is a `[start].skk.moe` in the trie, skip adding the rest of the node
if (node[0] && token === '.') {
if (node[1]) {
return true;
}
} else {
const newNode = createNode(node);
const newNode = createNode(false, node);
curNodeChildren.set(token, newNode);
node = newNode;
}
@ -387,55 +417,57 @@ export class HostnameSmolTrie<Meta = any> extends Triebase<Meta> {
}
// If we are in smolTree mode, we need to do something at the end of the loop
if (suffix[0] === '.') {
// Trying to add `[start].sub.example.com` where there is already a `[start]blog.sub.example.com` in the trie
if (includeAllSubdoain) {
// Trying to add `[.]sub.example.com` where there is already a `blog.sub.example.com` in the trie
// Make sure parent `[start]sub.example.com` (without dot) is removed (SETINEL to false)
(/** parent */ node[1]!)[0] = false;
// (/** parent */ node[2]!)[0] = false;
// Removing the rest of the parent's child nodes
node[2].clear();
node[3].clear();
// The SENTINEL of this node will be set to true at the end of the function, so we don't need to set it here
// we can use else-if here, because the children is now empty, we don't need to check the leading "."
} else if (node[2].get('.')?.[0] === true) {
} else if (node[1]) {
// Trying to add `example.com` when there is already a `.example.com` in the trie
// No need to increment size and set SENTINEL to true (skip this "new" item)
return;
}
node[0] = true;
node[3] = meta!;
node[1] = includeAllSubdoain;
node[4] = meta!;
}
public whitelist(suffix: string, includeAllSubdoain = suffix[0] === '.') {
if (suffix[0] === '.') {
suffix = suffix.slice(1);
}
public whitelist(suffix: string) {
const tokens = hostnameToTokens(suffix);
const res = this.getSingleChildLeaf(tokens);
if (res === null) return;
const { node, toPrune, tokenToPrune, parent } = res;
const { node, toPrune, tokenToPrune } = res;
// Trying to whitelist `[start].sub.example.com` where there is already a `[start]blog.sub.example.com` in the trie
if (tokens[0] === '.') {
// Trying to whitelist `[start].sub.example.com` where there might already be a `[start]blog.sub.example.com` in the trie
if (includeAllSubdoain) {
// If there is a `[start]sub.example.com` here, remove it
parent[0] = false;
node[0] = false;
node[1] = false;
// Removing all the child nodes by empty the children
// This removes the only child ".", which removes "blog.sub.example.com"
parent[2].clear();
node[3].clear();
} else {
// Trying to whitelist `example.com` when there is already a `.example.com` in the trie
const dotNode = node[2].get('.');
if (dotNode) {
dotNode[0] = false;
}
node[1] = false;
}
// return early if not found
if (!node[0]) return;
if (tokenToPrune && toPrune) {
toPrune[2].delete(tokenToPrune);
toPrune[3].delete(tokenToPrune);
} else {
node[0] = false;
}
@ -447,31 +479,39 @@ export class HostnameTrie<Meta = any> extends Triebase<Meta> {
return this.$size;
}
add(suffix: string, meta?: Meta): void {
add(suffix: string, includeAllSubdoain = suffix[0] === '.', meta?: Meta): void {
let node: TrieNode<Meta> = this.$root;
const onToken = (token: string) => {
if (node[2].has(token)) {
node = node[2].get(token)!;
if (node[3].has(token)) {
node = node[3].get(token)!;
} else {
const newNode = createNode(node);
node[2].set(token, newNode);
const newNode = createNode(false, node);
node[3].set(token, newNode);
node = newNode;
}
return false;
};
if (suffix[0] === '.') {
suffix = suffix.slice(1);
}
// When walkHostnameTokens returns true, we should skip the rest
if (walkHostnameTokens(suffix, onToken)) {
return;
}
if (!node[0]) {
// if same entry has been added before, skip
if (node[0]) {
return;
}
this.$size++;
node[0] = true;
node[3] = meta!;
}
node[1] = includeAllSubdoain;
node[4] = meta!;
}
}
@ -486,11 +526,11 @@ export function createTrie<_Meta = any>(from?: string[] | Set<string> | null, sm
export type Trie = ReturnType<typeof createTrie>;
function deepEqualArray(a: string[], b: string[]) {
let len = a.length;
if (len !== b.length) return false;
while (len--) {
if (a[len] !== b[len]) return false;
}
return true;
};
// function deepEqualArray(a: string[], b: string[]) {
// let len = a.length;
// if (len !== b.length) return false;
// while (len--) {
// if (a[len] !== b[len]) return false;
// }
// return true;
// };