ageng/ClearUrls
1
0
Fork 0
mirror of https://gitlab.com/KevinRoebert/ClearUrls synced 2025-12-16 06:05:37 +07:00
Code Issues Projects Releases Wiki Activity

Pre-compiled Regular Expressions

Browse Source 
Previously, regular expressions were compiled inside loops for every URL processed.

Now RegExp objects are created once when rules are loaded and stored in the provider instance.
This commit is contained in:
Kevin R
2025-12-07 21:36:15 +01:00
parent 6d904144ec
commit 7c8d6b0eee
1 changed files with 54 additions and 72 deletions
Download Patch File Download Diff File Expand all files Collapse all files

126
clearurls.js
View File

@@ -89,14 +89,14 @@ function removeFieldsFormURL(provider, pureUrl, quiet = false, request = null) {
/* /*
* Apply raw rules to the URL. * Apply raw rules to the URL.
*/ */
rawRules.forEach(function (rawRule) { rawRules.forEach(function ({ rule: rawRuleStr, regex: rawRuleRegex }) {
let beforeReplace = url; let beforeReplace = url;
url = url.replace(new RegExp(rawRule, "gi"), ""); url = url.replace(rawRuleRegex, "");
if (beforeReplace !== url) { if (beforeReplace !== url) {
//Log the action //Log the action
if (storage.loggingStatus && !quiet) { if (storage.loggingStatus && !quiet) {
pushToLog(beforeReplace, url, rawRule); pushToLog(beforeReplace, url, rawRuleStr);
} }
increaseBadged(quiet, request); increaseBadged(quiet, request);
@@ -113,13 +113,16 @@ function removeFieldsFormURL(provider, pureUrl, quiet = false, request = null) {
* Only test for matches, if there are fields or fragments that can be cleaned. * Only test for matches, if there are fields or fragments that can be cleaned.
*/ */
if (fields.toString() !== "" || fragments.toString() !== "") { if (fields.toString() !== "" || fragments.toString() !== "") {
rules.forEach(rule => { rules.forEach(({ rule, regex }) => {
const beforeFields = fields.toString(); const beforeFields = fields.toString();
const beforeFragments = fragments.toString(); const beforeFragments = fragments.toString();
let localChange = false; let localChange = false;
// Use the pre-compiled regex.
// Note: reusing a regex without 'g' flag is safe for .test()
for (const field of fields.keys()) { for (const field of fields.keys()) {
if (new RegExp("^"+rule+"$", "gi").test(field)) { if (regex.test(field)) {
fields.delete(field); fields.delete(field);
changes = true; changes = true;
localChange = true; localChange = true;
@@ -127,7 +130,7 @@ function removeFieldsFormURL(provider, pureUrl, quiet = false, request = null) {
} }
for (const fragment of fragments.keys()) { for (const fragment of fragments.keys()) {
if (new RegExp("^"+rule+"$", "gi").test(fragment)) { if (regex.test(fragment)) {
fragments.delete(fragment); fragments.delete(fragment);
changes = true; changes = true;
localChange = true; localChange = true;
@@ -360,7 +363,8 @@ function start() {
let methods = []; let methods = [];
if (_completeProvider) { if (_completeProvider) {
enabled_rules[".*"] = true; // enabled_rules[".*"] = true; // Original
enabled_rules[".*"] = new RegExp("^.*$", "i");
} }
/** /**
@@ -402,29 +406,26 @@ function start() {
* @return {boolean} ProviderURL as RegExp * @return {boolean} ProviderURL as RegExp
*/ */
this.matchURL = function (url) { this.matchURL = function (url) {
// Use matchException internal logic
return urlPattern.test(url) && !(this.matchException(url)); return urlPattern.test(url) && !(this.matchException(url));
}; };
/** /**
* Apply a rule to a given tuple of rule array. * Helper to update rule maps with compiled regexes.
* @param enabledRuleArray array for enabled rules
* @param disabledRulesArray array for disabled rules
* @param {String} rule RegExp as string
* @param {boolean} isActive Is this rule active?
*/ */
this.applyRule = (enabledRuleArray, disabledRulesArray, rule, isActive = true) => { const updateRule = (enabledMap, disabledMap, rule, isActive, compileFn) => {
if (isActive) { if (isActive) {
enabledRuleArray[rule] = true; if (!enabledMap[rule]) {
try {
if (disabledRulesArray[rule] !== undefined) { enabledMap[rule] = compileFn(rule);
delete disabledRulesArray[rule]; } catch (e) {
console.error("Invalid regex", rule, e);
}
} }
if (disabledMap[rule]) delete disabledMap[rule];
} else { } else {
disabledRulesArray[rule] = true; disabledMap[rule] = true;
if (enabledMap[rule]) delete enabledMap[rule];
if (enabledRuleArray[rule] !== undefined) {
delete enabledRuleArray[rule];
}
} }
}; };
@@ -436,20 +437,22 @@ function start() {
* @param {boolean} isActive Is this rule active? * @param {boolean} isActive Is this rule active?
*/ */
this.addRule = function (rule, isActive = true) { this.addRule = function (rule, isActive = true) {
this.applyRule(enabled_rules, disabled_rules, rule, isActive); updateRule(enabled_rules, disabled_rules, rule, isActive, r => new RegExp("^" + r + "$", "i"));
}; };
/** /**
* Return all active rules as an array. * Return all active rules as an array of {rule, regex}.
* *
* @return Array RegExp strings * @return Array Objects
*/ */
this.getRules = function () { this.getRules = function () {
if (!storage.referralMarketing) { let source = enabled_rules;
return Object.keys(Object.assign(enabled_rules, enabled_referralMarketing)); if (storage.referralMarketing) {
// Determine if we need to merge referral marketing rules
// We use a new object to avoid mutating enabled_rules via Object.assign if that was happening
source = Object.assign({}, enabled_rules, enabled_referralMarketing);
} }
return Object.entries(source).map(([rule, regex]) => ({ rule, regex }));
return Object.keys(enabled_rules);
}; };
/** /**
@@ -460,16 +463,17 @@ function start() {
* @param {boolean} isActive Is this rule active? * @param {boolean} isActive Is this rule active?
*/ */
this.addRawRule = function (rule, isActive = true) { this.addRawRule = function (rule, isActive = true) {
this.applyRule(enabled_rawRules, disabled_rawRules, rule, isActive); updateRule(enabled_rawRules, disabled_rawRules, rule, isActive, r => new RegExp(r, "gi"));
}; };
/** /**
* Return all active raw rules as an array. * Return all active raw rules as an array.
* *
* @return Array RegExp strings * @return Array Objects {rule, regex}
*/ */
this.getRawRules = function () { this.getRawRules = function () {
return Object.keys(enabled_rawRules); // return Object.keys(enabled_rawRules);
return Object.entries(enabled_rawRules).map(([rule, regex]) => ({ rule, regex }));
}; };
/** /**
@@ -480,7 +484,7 @@ function start() {
* @param {boolean} isActive Is this rule active? * @param {boolean} isActive Is this rule active?
*/ */
this.addReferralMarketing = function (rule, isActive = true) { this.addReferralMarketing = function (rule, isActive = true) {
this.applyRule(enabled_referralMarketing, disabled_referralMarketing, rule, isActive); updateRule(enabled_referralMarketing, disabled_referralMarketing, rule, isActive, r => new RegExp("^" + r + "$", "i"));
}; };
/** /**
@@ -491,19 +495,7 @@ function start() {
* @param {Boolean} isActive Is this exception active? * @param {Boolean} isActive Is this exception active?
*/ */
this.addException = function (exception, isActive = true) { this.addException = function (exception, isActive = true) {
if (isActive) { updateRule(enabled_exceptions, disabled_exceptions, exception, isActive, r => new RegExp(r, "i"));
enabled_exceptions[exception] = true;
if (disabled_exceptions[exception] !== undefined) {
delete disabled_exceptions[exception];
}
} else {
disabled_exceptions[exception] = true;
if (enabled_exceptions[exception] !== undefined) {
delete enabled_exceptions[exception];
}
}
}; };
/** /**
@@ -541,11 +533,10 @@ function start() {
//Add the site blocked alert to every exception //Add the site blocked alert to every exception
if (url === siteBlockedAlert) return true; if (url === siteBlockedAlert) return true;
for (const exception in enabled_exceptions) { for (const [exception, regex] of Object.entries(enabled_exceptions)) {
if (result) break; if (result) break;
// let exception_regex = new RegExp(exception, "i"); // Old
let exception_regex = new RegExp(exception, "i"); result = regex.test(url);
result = exception_regex.test(url);
} }
return result; return result;
@@ -559,19 +550,7 @@ function start() {
* @param {Boolean} isActive Is this redirection active? * @param {Boolean} isActive Is this redirection active?
*/ */
this.addRedirection = function (redirection, isActive = true) { this.addRedirection = function (redirection, isActive = true) {
if (isActive) { updateRule(enabled_redirections, disabled_redirections, redirection, isActive, r => new RegExp(r, "i"));
enabled_redirections[redirection] = true;
if (disabled_redirections[redirection] !== undefined) {
delete disabled_redirections[redirection];
}
} else {
disabled_redirections[redirection] = true;
if (enabled_redirections[redirection] !== undefined) {
delete enabled_redirections[redirection];
}
}
}; };
/** /**
@@ -582,11 +561,14 @@ function start() {
this.getRedirection = function (url) { this.getRedirection = function (url) {
let re = null; let re = null;
for (const redirection in enabled_redirections) { for (const [redirection, regex] of Object.entries(enabled_redirections)) {
let result = (url.match(new RegExp(redirection, "i"))); // let result = (url.match(new RegExp(redirection, "i"))); // Old
let result = url.match(regex);
if (result && result.length > 0 && redirection) { if (result && result.length > 0 && redirection) {
re = (new RegExp(redirection, "i")).exec(url)[1]; // re = (new RegExp(redirection, "i")).exec(url)[1];
// Reuse match result if possible, but exec is same as match for non-g.
re = result[1]; // Capture group 1
break; break;
} }
@@ -623,7 +605,7 @@ function start() {
pushToLog(request.url, request.url, translate('log_ping_blocked')); pushToLog(request.url, request.url, translate('log_ping_blocked'));
increaseBadged(false, request); increaseBadged(false, request);
increaseTotalCounter(1); increaseTotalCounter(1);
return {cancel: true}; return { cancel: true };
} }
/* /*
@@ -642,8 +624,8 @@ function start() {
if (result.redirect) { if (result.redirect) {
if (providers[i].shouldForceRedirect() && if (providers[i].shouldForceRedirect() &&
request.type === 'main_frame') { request.type === 'main_frame') {
browser.tabs.update(request.tabId, {url: result.url}).catch(handleError); browser.tabs.update(request.tabId, { url: result.url }).catch(handleError);
return {cancel: true}; return { cancel: true };
} }
return { return {
@@ -658,9 +640,9 @@ function start() {
if (result.cancel) { if (result.cancel) {
if (request.type === 'main_frame') { if (request.type === 'main_frame') {
const blockingPage = browser.runtime.getURL("html/siteBlockedAlert.html?source=" + encodeURIComponent(request.url)); const blockingPage = browser.runtime.getURL("html/siteBlockedAlert.html?source=" + encodeURIComponent(request.url));
browser.tabs.update(request.tabId, {url: blockingPage}).catch(handleError); browser.tabs.update(request.tabId, { url: blockingPage }).catch(handleError);
return {cancel: true}; return { cancel: true };
} else { } else {
return { return {
redirectUrl: siteBlockedAlert redirectUrl: siteBlockedAlert
@@ -723,7 +705,7 @@ function start() {
*/ */
browser.webRequest.onBeforeRequest.addListener( browser.webRequest.onBeforeRequest.addListener(
promise, promise,
{urls: ["<all_urls>"], types: getData("types").concat(getData("pingRequestTypes"))}, { urls: ["<all_urls>"], types: getData("types").concat(getData("pingRequestTypes")) },
["blocking"] ["blocking"]
); );
} }