michael/minify-html
michael/minify-html
Archived
1
0
Fork 0
Code Issues Pull requests Projects Releases Wiki Activity

Use regex crate for substring searches

Browse source
This commit is contained in:
Wilson Lin 2020-07-03 20:37:52 +10:00
commit a519394c3a
10 changed files with 40 additions and 84 deletions
Download patch file Download diff file Expand all files Collapse all files

1
gen/_common.ts
View file

@ -13,7 +13,6 @@ try {
writeFileSync(join(RUST_OUT_DIR, 'mod.rs'), `
pub mod attrs;
pub mod entities;
pub mod patterns;
`);
export const DATA_DIR = join(__dirname, 'data');

6
gen/data/patterns.json
View file

@ -1,6 +0,0 @@
{
"COMMENT_END": "-->",
"STYLE_END": "</style",
"SCRIPT_END": "</script",
"INSTRUCTION_END": "?>"
}

31
gen/patterns.ts
View file

@ -1,31 +0,0 @@
import {readFileSync, writeFileSync} from 'fs';
import {DATA_DIR, RUST_OUT_DIR} from './_common';
import {join} from 'path';
import {EOL} from 'os';
const patterns: {[name: string]: string} = JSON.parse(readFileSync(join(DATA_DIR, 'patterns.json'), 'utf8'));
const chr = (str: string, char: number) => str.charCodeAt(char);
const buildPattern = (seq: string): string => {
const dfa = Array.from({length: 256}, () => Array(seq.length).fill(0));
dfa[chr(seq, 0)][0] = 1;
let x = 0;
let j = 1;
while (j < seq.length) {
for (let c = 0; c < 256; c++) {
dfa[c][j] = dfa[c][x];
}
dfa[chr(seq, j)][j] = j + 1;
x = dfa[chr(seq, j)][x];
j += 1;
}
return `crate::pattern::SinglePattern::prebuilt(&[${dfa.flat().join(', ')}], ${seq.length})`;
};
const output = Object.entries(patterns)
.map(([name, pattern]) => `pub static ${name}: &crate::pattern::SinglePattern = &${buildPattern(pattern)};`);
writeFileSync(join(RUST_OUT_DIR, 'patterns.rs'), output.join(EOL));