Browse Source

Fix quadratic time on backticks

This commit adds a cache `StateInline->backticks` which remembers
positions for every possible backtick closer (`, ``, ```, etc.).

Algorithm is similar to one described here:
685b714453

close https://github.com/markdown-it/markdown-it/issues/733
pull/745/head
Alex Kocharin 4 years ago
parent
commit
1e8aff0084
  1. 1
      CHANGELOG.md
  2. 55
      lib/rules_inline/backticks.js
  3. 3
      lib/rules_inline/state_inline.js

1
CHANGELOG.md

@ -10,6 +10,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
### Fixed
- `[](<foo<bar>)` is no longer a valid link.
- Fix performance issues when parsing links, #732.
- Fix performance issues when parsing backticks, #733.
## [12.0.2] - 2020-10-23

55
lib/rules_inline/backticks.js

@ -2,8 +2,18 @@
'use strict';
function addCodeToken(state, marker, pos, matchStart) {
var token = state.push('code_inline', 'code', 0);
token.markup = marker;
token.content = state.src.slice(pos, matchStart)
.replace(/\n/g, ' ')
.replace(/^ (.+) $/, '$1');
}
module.exports = function backtick(state, silent) {
var start, max, marker, matchStart, matchEnd, token,
var start, max, marker, matchStart, matchEnd, startLength, endLength,
pos = state.pos,
ch = state.src.charCodeAt(pos);
@ -13,31 +23,54 @@ module.exports = function backtick(state, silent) {
pos++;
max = state.posMax;
// scan marker length
while (pos < max && state.src.charCodeAt(pos) === 0x60/* ` */) { pos++; }
marker = state.src.slice(start, pos);
startLength = marker.length;
// Look for required marker length in the cache first
if (state.backticks[startLength] && state.backticks[startLength] > start) {
if (state.backticks[startLength] === Infinity) {
if (!silent) state.pending += marker;
state.pos += startLength;
} else {
if (!silent) addCodeToken(state, marker, pos, state.backticks[startLength]);
state.pos = matchEnd;
}
return true;
}
matchStart = matchEnd = pos;
// Nothing found in the cache, scan until the end of the line (or until marker is found)
while ((matchStart = state.src.indexOf('`', matchEnd)) !== -1) {
matchEnd = matchStart + 1;
// scan marker length
while (matchEnd < max && state.src.charCodeAt(matchEnd) === 0x60/* ` */) { matchEnd++; }
if (matchEnd - matchStart === marker.length) {
if (!silent) {
token = state.push('code_inline', 'code', 0);
token.markup = marker;
token.content = state.src.slice(pos, matchStart)
.replace(/\n/g, ' ')
.replace(/^ (.+) $/, '$1');
}
endLength = matchEnd - matchStart;
if (endLength === marker.length) {
// Found matching closer length.
if (!silent) addCodeToken(state, marker, pos, matchStart);
state.pos = matchEnd;
return true;
}
// Some different length found, put it in cache just in case
if (!state.backticks[endLength] || state.backticks[endLength] <= start) {
state.backticks[endLength] = matchStart;
}
// Scanned through the end, didn't find anything. Mark "no matches" for this length;
if (matchEnd >= max) {
state.backticks[startLength] = Infinity;
}
}
if (!silent) { state.pending += marker; }
state.pos += marker.length;
if (!silent) state.pending += marker;
state.pos += startLength;
return true;
};

3
lib/rules_inline/state_inline.js

@ -31,6 +31,9 @@ function StateInline(src, md, env, outTokens) {
// Stack of delimiter lists for upper level tags
this._prev_delimiters = [];
// backtick length => last seen position
this.backticks = {};
}

Loading…
Cancel
Save