uBlock/src/js/storage.js

1411 lines
48 KiB
JavaScript
Raw Normal View History

2014-06-24 00:42:43 +02:00
/*******************************************************************************
uBlock Origin - a browser extension to block requests.
Copyright (C) 2014-present Raymond Hill
2014-06-24 00:42:43 +02:00
This program is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see {http://www.gnu.org/licenses/}.
Home: https://github.com/gorhill/uBlock
*/
/* global punycode, publicSuffixList */
'use strict';
2014-06-24 00:42:43 +02:00
/******************************************************************************/
2015-03-07 05:36:09 +01:00
µBlock.getBytesInUse = function(callback) {
if ( typeof callback !== 'function' ) {
callback = this.noopFunc;
}
let bytesInUse;
let countdown = 0;
Refactor selfie generation into a more flexible persistence mechanism The motivation is to address the higher peak memory usage at launch time with 3rd-gen HNTrie when a selfie was present. The selfie generation prior to this change was to collect all filtering data into a single data structure, and then to serialize that whole structure at once into storage (using JSON.stringify). However, HNTrie serialization requires that a large UintArray32 be converted into a plain JS array, which itslef would be indirectly converted into a JSON string. This was the main reason why peak memory usage would be higher at launch from selfie, since the JSON string would need to be wholly unserialized into JS objects, which themselves would need to be converted into more specialized data structures (like that Uint32Array one). The solution to lower peak memory usage at launch is to refactor selfie generation to allow a more piecemeal approach: each filtering component is given the ability to serialize itself rather than to be forced to be embedded in the master selfie. With this approach, the HNTrie buffer can now serialize to its own storage by converting the buffer data directly into a string which can be directly sent to storage. This avoiding expensive intermediate steps such as converting into a JS array and then to a JSON string. As part of the refactoring, there was also opportunistic code upgrade to ES6 and Promise (eventually all of uBO's code will be proper ES6). Additionally, the polyfill to bring getBytesInUse() to Firefox has been revisited to replace the rather expensive previous implementation with an implementation with virtually no overhead.
2019-02-14 19:33:55 +01:00
const process = count => {
if ( typeof count === 'number' ) {
if ( bytesInUse === undefined ) {
bytesInUse = 0;
}
bytesInUse += count;
}
countdown -= 1;
if ( countdown > 0 ) { return; }
2014-06-24 00:42:43 +02:00
µBlock.storageUsed = bytesInUse;
2015-03-07 05:36:09 +01:00
callback(bytesInUse);
2014-06-24 00:42:43 +02:00
};
// Not all platforms implement this method.
if ( vAPI.storage.getBytesInUse instanceof Function ) {
countdown += 1;
vAPI.storage.getBytesInUse(null, process);
}
if (
navigator.storage instanceof Object &&
navigator.storage.estimate instanceof Function
) {
countdown += 1;
navigator.storage.estimate().then(estimate => {
process(estimate.usage);
Refactor selfie generation into a more flexible persistence mechanism The motivation is to address the higher peak memory usage at launch time with 3rd-gen HNTrie when a selfie was present. The selfie generation prior to this change was to collect all filtering data into a single data structure, and then to serialize that whole structure at once into storage (using JSON.stringify). However, HNTrie serialization requires that a large UintArray32 be converted into a plain JS array, which itslef would be indirectly converted into a JSON string. This was the main reason why peak memory usage would be higher at launch from selfie, since the JSON string would need to be wholly unserialized into JS objects, which themselves would need to be converted into more specialized data structures (like that Uint32Array one). The solution to lower peak memory usage at launch is to refactor selfie generation to allow a more piecemeal approach: each filtering component is given the ability to serialize itself rather than to be forced to be embedded in the master selfie. With this approach, the HNTrie buffer can now serialize to its own storage by converting the buffer data directly into a string which can be directly sent to storage. This avoiding expensive intermediate steps such as converting into a JS array and then to a JSON string. As part of the refactoring, there was also opportunistic code upgrade to ES6 and Promise (eventually all of uBO's code will be proper ES6). Additionally, the polyfill to bring getBytesInUse() to Firefox has been revisited to replace the rather expensive previous implementation with an implementation with virtually no overhead.
2019-02-14 19:33:55 +01:00
});
}
if ( countdown === 0 ) {
callback();
}
2014-06-24 00:42:43 +02:00
};
/******************************************************************************/
2015-11-29 23:06:58 +01:00
µBlock.saveLocalSettings = (function() {
let saveAfter = 4 * 60 * 1000;
2014-06-24 00:42:43 +02:00
let onTimeout = ( ) => {
let µb = µBlock;
2016-10-08 16:15:31 +02:00
if ( µb.localSettingsLastModified > µb.localSettingsLastSaved ) {
µb.saveLocalSettings();
2015-11-29 23:06:58 +01:00
}
vAPI.setTimeout(onTimeout, saveAfter);
};
2014-06-24 00:42:43 +02:00
2015-11-29 23:06:58 +01:00
vAPI.setTimeout(onTimeout, saveAfter);
2018-07-26 00:24:14 +02:00
return function(callback) {
this.localSettingsLastSaved = Date.now();
vAPI.storage.set(this.localSettings, callback);
};
2015-11-29 23:06:58 +01:00
})();
2014-06-24 00:42:43 +02:00
/******************************************************************************/
µBlock.saveUserSettings = function() {
vAPI.storage.set(this.userSettings);
2014-06-24 00:42:43 +02:00
};
/******************************************************************************/
2018-02-21 19:29:36 +01:00
µBlock.loadHiddenSettings = function() {
return new Promise(resolve => {
// >>>> start of executor
vAPI.storage.get('hiddenSettings', bin => {
if ( bin instanceof Object === false ) {
return resolve();
}
Refactor selfie generation into a more flexible persistence mechanism The motivation is to address the higher peak memory usage at launch time with 3rd-gen HNTrie when a selfie was present. The selfie generation prior to this change was to collect all filtering data into a single data structure, and then to serialize that whole structure at once into storage (using JSON.stringify). However, HNTrie serialization requires that a large UintArray32 be converted into a plain JS array, which itslef would be indirectly converted into a JSON string. This was the main reason why peak memory usage would be higher at launch from selfie, since the JSON string would need to be wholly unserialized into JS objects, which themselves would need to be converted into more specialized data structures (like that Uint32Array one). The solution to lower peak memory usage at launch is to refactor selfie generation to allow a more piecemeal approach: each filtering component is given the ability to serialize itself rather than to be forced to be embedded in the master selfie. With this approach, the HNTrie buffer can now serialize to its own storage by converting the buffer data directly into a string which can be directly sent to storage. This avoiding expensive intermediate steps such as converting into a JS array and then to a JSON string. As part of the refactoring, there was also opportunistic code upgrade to ES6 and Promise (eventually all of uBO's code will be proper ES6). Additionally, the polyfill to bring getBytesInUse() to Firefox has been revisited to replace the rather expensive previous implementation with an implementation with virtually no overhead.
2019-02-14 19:33:55 +01:00
const hs = bin.hiddenSettings;
2018-02-21 19:29:36 +01:00
if ( hs instanceof Object ) {
Refactor selfie generation into a more flexible persistence mechanism The motivation is to address the higher peak memory usage at launch time with 3rd-gen HNTrie when a selfie was present. The selfie generation prior to this change was to collect all filtering data into a single data structure, and then to serialize that whole structure at once into storage (using JSON.stringify). However, HNTrie serialization requires that a large UintArray32 be converted into a plain JS array, which itslef would be indirectly converted into a JSON string. This was the main reason why peak memory usage would be higher at launch from selfie, since the JSON string would need to be wholly unserialized into JS objects, which themselves would need to be converted into more specialized data structures (like that Uint32Array one). The solution to lower peak memory usage at launch is to refactor selfie generation to allow a more piecemeal approach: each filtering component is given the ability to serialize itself rather than to be forced to be embedded in the master selfie. With this approach, the HNTrie buffer can now serialize to its own storage by converting the buffer data directly into a string which can be directly sent to storage. This avoiding expensive intermediate steps such as converting into a JS array and then to a JSON string. As part of the refactoring, there was also opportunistic code upgrade to ES6 and Promise (eventually all of uBO's code will be proper ES6). Additionally, the polyfill to bring getBytesInUse() to Firefox has been revisited to replace the rather expensive previous implementation with an implementation with virtually no overhead.
2019-02-14 19:33:55 +01:00
const hsDefault = this.hiddenSettingsDefault;
for ( const key in hsDefault ) {
2018-02-21 19:29:36 +01:00
if (
hsDefault.hasOwnProperty(key) &&
hs.hasOwnProperty(key) &&
typeof hs[key] === typeof hsDefault[key]
) {
this.hiddenSettings[key] = hs[key];
2018-02-21 19:29:36 +01:00
}
}
if ( typeof this.hiddenSettings.suspendTabsUntilReady === 'boolean' ) {
this.hiddenSettings.suspendTabsUntilReady =
this.hiddenSettings.suspendTabsUntilReady
? 'yes'
: 'unset';
}
2018-02-21 19:29:36 +01:00
}
if ( vAPI.localStorage.getItem('immediateHiddenSettings') === null ) {
this.saveImmediateHiddenSettings();
2018-02-21 19:29:36 +01:00
}
Refactor selfie generation into a more flexible persistence mechanism The motivation is to address the higher peak memory usage at launch time with 3rd-gen HNTrie when a selfie was present. The selfie generation prior to this change was to collect all filtering data into a single data structure, and then to serialize that whole structure at once into storage (using JSON.stringify). However, HNTrie serialization requires that a large UintArray32 be converted into a plain JS array, which itslef would be indirectly converted into a JSON string. This was the main reason why peak memory usage would be higher at launch from selfie, since the JSON string would need to be wholly unserialized into JS objects, which themselves would need to be converted into more specialized data structures (like that Uint32Array one). The solution to lower peak memory usage at launch is to refactor selfie generation to allow a more piecemeal approach: each filtering component is given the ability to serialize itself rather than to be forced to be embedded in the master selfie. With this approach, the HNTrie buffer can now serialize to its own storage by converting the buffer data directly into a string which can be directly sent to storage. This avoiding expensive intermediate steps such as converting into a JS array and then to a JSON string. As part of the refactoring, there was also opportunistic code upgrade to ES6 and Promise (eventually all of uBO's code will be proper ES6). Additionally, the polyfill to bring getBytesInUse() to Firefox has been revisited to replace the rather expensive previous implementation with an implementation with virtually no overhead.
2019-02-14 19:33:55 +01:00
self.log.verbosity = this.hiddenSettings.consoleLogLevel;
resolve();
});
// <<<< end of executor
});
2018-02-21 19:29:36 +01:00
};
// Note: Save only the settings which values differ from the default ones.
// This way the new default values in the future will properly apply for those
// which were not modified by the user.
2018-02-21 19:29:36 +01:00
µBlock.saveHiddenSettings = function(callback) {
Refactor selfie generation into a more flexible persistence mechanism The motivation is to address the higher peak memory usage at launch time with 3rd-gen HNTrie when a selfie was present. The selfie generation prior to this change was to collect all filtering data into a single data structure, and then to serialize that whole structure at once into storage (using JSON.stringify). However, HNTrie serialization requires that a large UintArray32 be converted into a plain JS array, which itslef would be indirectly converted into a JSON string. This was the main reason why peak memory usage would be higher at launch from selfie, since the JSON string would need to be wholly unserialized into JS objects, which themselves would need to be converted into more specialized data structures (like that Uint32Array one). The solution to lower peak memory usage at launch is to refactor selfie generation to allow a more piecemeal approach: each filtering component is given the ability to serialize itself rather than to be forced to be embedded in the master selfie. With this approach, the HNTrie buffer can now serialize to its own storage by converting the buffer data directly into a string which can be directly sent to storage. This avoiding expensive intermediate steps such as converting into a JS array and then to a JSON string. As part of the refactoring, there was also opportunistic code upgrade to ES6 and Promise (eventually all of uBO's code will be proper ES6). Additionally, the polyfill to bring getBytesInUse() to Firefox has been revisited to replace the rather expensive previous implementation with an implementation with virtually no overhead.
2019-02-14 19:33:55 +01:00
const bin = { hiddenSettings: {} };
for ( const prop in this.hiddenSettings ) {
if (
this.hiddenSettings.hasOwnProperty(prop) &&
this.hiddenSettings[prop] !== this.hiddenSettingsDefault[prop]
) {
bin.hiddenSettings[prop] = this.hiddenSettings[prop];
}
}
vAPI.storage.set(bin, callback);
2018-02-21 19:29:36 +01:00
this.saveImmediateHiddenSettings();
Refactor selfie generation into a more flexible persistence mechanism The motivation is to address the higher peak memory usage at launch time with 3rd-gen HNTrie when a selfie was present. The selfie generation prior to this change was to collect all filtering data into a single data structure, and then to serialize that whole structure at once into storage (using JSON.stringify). However, HNTrie serialization requires that a large UintArray32 be converted into a plain JS array, which itslef would be indirectly converted into a JSON string. This was the main reason why peak memory usage would be higher at launch from selfie, since the JSON string would need to be wholly unserialized into JS objects, which themselves would need to be converted into more specialized data structures (like that Uint32Array one). The solution to lower peak memory usage at launch is to refactor selfie generation to allow a more piecemeal approach: each filtering component is given the ability to serialize itself rather than to be forced to be embedded in the master selfie. With this approach, the HNTrie buffer can now serialize to its own storage by converting the buffer data directly into a string which can be directly sent to storage. This avoiding expensive intermediate steps such as converting into a JS array and then to a JSON string. As part of the refactoring, there was also opportunistic code upgrade to ES6 and Promise (eventually all of uBO's code will be proper ES6). Additionally, the polyfill to bring getBytesInUse() to Firefox has been revisited to replace the rather expensive previous implementation with an implementation with virtually no overhead.
2019-02-14 19:33:55 +01:00
self.log.verbosity = this.hiddenSettings.consoleLogLevel;
2018-02-21 19:29:36 +01:00
};
/******************************************************************************/
µBlock.hiddenSettingsFromString = function(raw) {
var out = Object.assign({}, this.hiddenSettingsDefault),
lineIter = new this.LineIterator(raw),
line, matches, name, value;
while ( lineIter.eot() === false ) {
line = lineIter.next();
matches = /^\s*(\S+)\s+(.+)$/.exec(line);
if ( matches === null || matches.length !== 3 ) { continue; }
name = matches[1];
if ( out.hasOwnProperty(name) === false ) { continue; }
value = matches[2];
switch ( typeof out[name] ) {
case 'boolean':
if ( value === 'true' ) {
out[name] = true;
} else if ( value === 'false' ) {
out[name] = false;
}
break;
case 'string':
out[name] = value;
break;
case 'number':
out[name] = parseInt(value, 10);
if ( isNaN(out[name]) ) {
out[name] = this.hiddenSettingsDefault[name];
}
break;
default:
break;
}
}
2018-02-21 19:29:36 +01:00
return out;
};
µBlock.stringFromHiddenSettings = function() {
var out = [],
2018-02-21 19:29:36 +01:00
keys = Object.keys(this.hiddenSettings).sort();
for ( var key of keys ) {
out.push(key + ' ' + this.hiddenSettings[key]);
}
return out.join('\n');
};
/******************************************************************************/
2018-02-21 19:29:36 +01:00
// These settings must be available immediately on startup, without delay
// through the vAPI.localStorage. Add/remove settings as needed.
µBlock.saveImmediateHiddenSettings = function() {
vAPI.localStorage.setItem(
'immediateHiddenSettings',
JSON.stringify({
consoleLogLevel: this.hiddenSettings.consoleLogLevel,
disableWebAssembly: this.hiddenSettings.disableWebAssembly,
2018-02-21 19:29:36 +01:00
suspendTabsUntilReady: this.hiddenSettings.suspendTabsUntilReady,
})
);
};
/******************************************************************************/
µBlock.savePermanentFirewallRules = function() {
2018-09-03 20:06:49 +02:00
vAPI.storage.set({
dynamicFilteringString: this.permanentFirewall.toString()
});
2015-03-27 18:00:55 +01:00
};
/******************************************************************************/
2015-05-21 20:15:17 +02:00
µBlock.savePermanentURLFilteringRules = function() {
2018-09-03 20:06:49 +02:00
vAPI.storage.set({
urlFilteringString: this.permanentURLFiltering.toString()
});
2015-05-21 20:15:17 +02:00
};
/******************************************************************************/
2015-03-27 18:00:55 +01:00
µBlock.saveHostnameSwitches = function() {
2018-09-03 20:06:49 +02:00
vAPI.storage.set({
hostnameSwitchesString: this.permanentSwitches.toString()
});
};
/******************************************************************************/
µBlock.saveWhitelist = function() {
2018-09-03 20:06:49 +02:00
vAPI.storage.set({
netWhitelist: this.stringFromWhitelist(this.netWhitelist)
});
this.netWhitelistModifyTime = Date.now();
};
/*******************************************************************************
TODO(seamless migration):
The code related to 'remoteBlacklist' can be removed when I am confident
all users have moved to a version of uBO which no longer depends on
the property 'remoteBlacklists, i.e. v1.11 and beyond.
**/
µBlock.loadSelectedFilterLists = function() {
return new Promise(resolve => {
// >>>> start of executor
vAPI.storage.get('selectedFilterLists', bin => {
2017-12-14 22:42:54 +01:00
// Select default filter lists if first-time launch.
if (
bin instanceof Object === false ||
Array.isArray(bin.selectedFilterLists) === false
) {
this.assets.metadata(availableLists => {
this.saveSelectedFilterLists(
this.autoSelectRegionalFilterLists(availableLists)
2018-01-01 13:52:03 +01:00
);
resolve();
2017-01-26 16:17:38 +01:00
});
return;
}
this.selectedFilterLists = bin.selectedFilterLists;
resolve();
});
// <<<< end of executor
});
};
µBlock.saveSelectedFilterLists = function(newKeys, append, callback) {
if ( typeof append === 'function' ) {
callback = append;
append = false;
}
2017-01-26 16:17:38 +01:00
var oldKeys = this.selectedFilterLists.slice();
if ( append ) {
newKeys = newKeys.concat(oldKeys);
}
var newSet = new Set(newKeys);
// Purge unused filter lists from cache.
for ( var i = 0, n = oldKeys.length; i < n; i++ ) {
if ( newSet.has(oldKeys[i]) === false ) {
this.removeFilterList(oldKeys[i]);
2017-01-22 22:05:16 +01:00
}
2017-01-26 16:17:38 +01:00
}
newKeys = Array.from(newSet);
2017-01-26 16:17:38 +01:00
var bin = {
2017-12-14 22:42:54 +01:00
selectedFilterLists: newKeys
2017-01-26 16:17:38 +01:00
};
this.selectedFilterLists = newKeys;
vAPI.storage.set(bin, callback);
};
/******************************************************************************/
2017-01-22 22:05:16 +01:00
µBlock.applyFilterListSelection = function(details, callback) {
var µb = this,
selectedListKeySet = new Set(this.selectedFilterLists),
externalLists = this.userSettings.externalLists,
i, n, assetKey;
// Filter lists to select
if ( Array.isArray(details.toSelect) ) {
if ( details.merge ) {
for ( i = 0, n = details.toSelect.length; i < n; i++ ) {
selectedListKeySet.add(details.toSelect[i]);
}
} else {
selectedListKeySet = new Set(details.toSelect);
}
}
// Imported filter lists to remove
if ( Array.isArray(details.toRemove) ) {
var removeURLFromHaystack = function(haystack, needle) {
return haystack.replace(
new RegExp(
'(^|\\n)' +
2017-11-10 12:56:38 +01:00
µb.escapeRegex(needle) +
2017-01-22 22:05:16 +01:00
'(\\n|$)', 'g'),
'\n'
).trim();
};
for ( i = 0, n = details.toRemove.length; i < n; i++ ) {
assetKey = details.toRemove[i];
selectedListKeySet.delete(assetKey);
externalLists = removeURLFromHaystack(externalLists, assetKey);
this.removeFilterList(assetKey);
}
}
// Filter lists to import
if ( typeof details.toImport === 'string' ) {
// https://github.com/gorhill/uBlock/issues/1181
// Try mapping the URL of an imported filter list to the assetKey of an
// existing stock list.
var assetKeyFromURL = function(url) {
var needle = url.replace(/^https?:/, '');
var assets = µb.availableFilterLists, asset;
for ( var assetKey in assets ) {
asset = assets[assetKey];
if ( asset.content !== 'filters' ) { continue; }
if ( typeof asset.contentURL === 'string' ) {
if ( asset.contentURL.endsWith(needle) ) { return assetKey; }
continue;
}
if ( Array.isArray(asset.contentURL) === false ) { continue; }
for ( i = 0, n = asset.contentURL.length; i < n; i++ ) {
if ( asset.contentURL[i].endsWith(needle) ) {
return assetKey;
}
}
}
return url;
};
var importedSet = new Set(this.listKeysFromCustomFilterLists(externalLists)),
toImportSet = new Set(this.listKeysFromCustomFilterLists(details.toImport));
for ( var urlKey of toImportSet ) {
if ( importedSet.has(urlKey) ) { continue; }
assetKey = assetKeyFromURL(urlKey);
if ( assetKey === urlKey ) {
importedSet.add(urlKey);
2017-01-22 22:05:16 +01:00
}
selectedListKeySet.add(assetKey);
}
externalLists = Array.from(importedSet).sort().join('\n');
2017-01-22 22:05:16 +01:00
}
var result = Array.from(selectedListKeySet);
2017-01-22 22:05:16 +01:00
if ( externalLists !== this.userSettings.externalLists ) {
this.userSettings.externalLists = externalLists;
vAPI.storage.set({ externalLists: externalLists });
}
this.saveSelectedFilterLists(result);
if ( typeof callback === 'function' ) {
callback(result);
}
};
/******************************************************************************/
µBlock.listKeysFromCustomFilterLists = function(raw) {
var out = new Set(),
reIgnore = /^[!#]/,
reValid = /^[a-z-]+:\/\/\S+/,
lineIter = new this.LineIterator(raw),
location;
while ( lineIter.eot() === false ) {
location = lineIter.next().trim();
if ( reIgnore.test(location) || !reValid.test(location) ) {
continue;
}
out.add(location);
}
return Array.from(out);
2017-01-22 22:05:16 +01:00
};
/******************************************************************************/
2014-07-13 02:32:44 +02:00
µBlock.saveUserFilters = function(content, callback) {
2015-12-07 14:59:22 +01:00
// https://github.com/gorhill/uBlock/issues/1022
// Be sure to end with an empty line.
content = content.trim();
if ( content !== '' ) { content += '\n'; }
2015-08-11 21:29:14 +02:00
this.assets.put(this.userFiltersPath, content, callback);
this.removeCompiledFilterList(this.userFiltersPath);
2014-07-13 02:32:44 +02:00
};
µBlock.loadUserFilters = function(callback) {
return this.assets.get(this.userFiltersPath, callback);
};
/******************************************************************************/
µBlock.appendUserFilters = function(filters, options) {
filters = filters.trim();
if ( filters.length === 0 ) { return; }
2015-02-24 00:31:29 +01:00
// https://github.com/uBlockOrigin/uBlock-issues/issues/372
// Auto comment using user-defined template.
let comment = '';
if (
options instanceof Object &&
options.autoComment === true &&
this.hiddenSettings.autoCommentFilterTemplate.indexOf('{{') !== -1
) {
const d = new Date();
comment =
'! ' +
this.hiddenSettings.autoCommentFilterTemplate
.replace('{{date}}', d.toLocaleDateString())
.replace('{{time}}', d.toLocaleTimeString())
.replace('{{origin}}', options.origin);
}
const onSaved = ( ) => {
const compiledFilters = this.compileFilters(
filters,
{ assetKey: this.userFiltersPath }
);
const snfe = this.staticNetFilteringEngine;
const cfe = this.cosmeticFilteringEngine;
const acceptedCount = snfe.acceptedCount + cfe.acceptedCount;
const discardedCount = snfe.discardedCount + cfe.discardedCount;
this.applyCompiledFilters(compiledFilters, true);
const entry = this.availableFilterLists[this.userFiltersPath];
const deltaEntryCount =
snfe.acceptedCount +
cfe.acceptedCount - acceptedCount;
const deltaEntryUsedCount =
deltaEntryCount -
(snfe.discardedCount + cfe.discardedCount - discardedCount);
2015-04-27 00:31:51 +02:00
entry.entryCount += deltaEntryCount;
entry.entryUsedCount += deltaEntryUsedCount;
vAPI.storage.set({ 'availableFilterLists': this.availableFilterLists });
this.staticNetFilteringEngine.freeze();
this.redirectEngine.freeze();
this.staticExtFilteringEngine.freeze();
this.selfieManager.destroy();
2015-02-24 00:31:29 +01:00
};
const onLoaded = details => {
if ( details.error ) { return; }
// The comment, if any, will be applied if and only if it is different
// from the last comment found in the user filter list.
if ( comment !== '' ) {
const pos = details.content.lastIndexOf(comment);
if (
pos === -1 ||
details.content.indexOf('\n!', pos + 1) !== -1
) {
filters = '\n' + comment + '\n' + filters;
}
}
2015-04-07 03:26:05 +02:00
// https://github.com/chrisaljoudi/uBlock/issues/976
// If we reached this point, the filter quite probably needs to be
// added for sure: do not try to be too smart, trying to avoid
// duplicates at this point may lead to more issues.
this.saveUserFilters(details.content.trim() + '\n' + filters, onSaved);
2014-07-13 02:32:44 +02:00
};
2014-09-08 23:46:58 +02:00
2015-02-24 00:31:29 +01:00
this.loadUserFilters(onLoaded);
2014-07-13 02:32:44 +02:00
};
/******************************************************************************/
µBlock.autoSelectRegionalFilterLists = function(lists) {
2017-01-27 19:44:52 +01:00
var selectedListKeys = [ this.userFiltersPath ],
list;
for ( var key in lists ) {
if ( lists.hasOwnProperty(key) === false ) { continue; }
list = lists[key];
if ( list.off !== true ) {
selectedListKeys.push(key);
continue;
2015-02-25 22:51:04 +01:00
}
2017-11-09 18:53:05 +01:00
if ( this.listMatchesEnvironment(list) ) {
selectedListKeys.push(key);
list.off = false;
}
}
return selectedListKeys;
};
2015-02-25 22:51:04 +01:00
/******************************************************************************/
µBlock.getAvailableLists = function(callback) {
var µb = this,
oldAvailableLists = {},
newAvailableLists = {};
// User filter list.
newAvailableLists[this.userFiltersPath] = {
group: 'user',
title: vAPI.i18n('1pPageName')
};
// Custom filter lists.
var importedListKeys = this.listKeysFromCustomFilterLists(µb.userSettings.externalLists),
i = importedListKeys.length, listKey, entry;
while ( i-- ) {
listKey = importedListKeys[i];
entry = {
content: 'filters',
contentURL: listKey,
external: true,
group: 'custom',
submitter: 'user',
title: ''
};
newAvailableLists[listKey] = entry;
this.assets.registerAssetSource(listKey, entry);
}
// Convert a no longer existing stock list into an imported list.
var customListFromStockList = function(assetKey) {
var oldEntry = oldAvailableLists[assetKey];
if ( oldEntry === undefined || oldEntry.off === true ) { return; }
var listURL = oldEntry.contentURL;
if ( Array.isArray(listURL) ) {
listURL = listURL[0];
}
var newEntry = {
content: 'filters',
contentURL: listURL,
external: true,
group: 'custom',
submitter: 'user',
title: oldEntry.title || ''
};
newAvailableLists[listURL] = newEntry;
µb.assets.registerAssetSource(listURL, newEntry);
importedListKeys.push(listURL);
µb.userSettings.externalLists += '\n' + listURL;
µb.userSettings.externalLists = µb.userSettings.externalLists.trim();
vAPI.storage.set({ externalLists: µb.userSettings.externalLists });
µb.saveSelectedFilterLists([ listURL ], true);
};
// Final steps:
// - reuse existing list metadata if any;
// - unregister unreferenced imported filter lists if any.
var finalize = function() {
var assetKey, newEntry, oldEntry;
// Reuse existing metadata.
for ( assetKey in oldAvailableLists ) {
oldEntry = oldAvailableLists[assetKey];
newEntry = newAvailableLists[assetKey];
// List no longer exists. If a stock list, try to convert to
// imported list if it was selected.
if ( newEntry === undefined ) {
µb.removeFilterList(assetKey);
if ( assetKey.indexOf('://') === -1 ) {
customListFromStockList(assetKey);
}
continue;
}
if ( oldEntry.entryCount !== undefined ) {
newEntry.entryCount = oldEntry.entryCount;
}
if ( oldEntry.entryUsedCount !== undefined ) {
newEntry.entryUsedCount = oldEntry.entryUsedCount;
2014-07-25 22:12:20 +02:00
}
2015-03-11 04:46:18 +01:00
// This may happen if the list name was pulled from the list
// content.
2015-04-07 03:26:05 +02:00
// https://github.com/chrisaljoudi/uBlock/issues/982
2015-03-11 04:46:18 +01:00
// There is no guarantee the title was successfully extracted from
// the list content.
if (
newEntry.title === '' &&
typeof oldEntry.title === 'string' &&
oldEntry.title !== ''
2015-03-11 04:46:18 +01:00
) {
newEntry.title = oldEntry.title;
}
2014-06-24 00:42:43 +02:00
}
2016-01-03 19:58:25 +01:00
// Remove unreferenced imported filter lists.
var dict = new Set(importedListKeys);
for ( assetKey in newAvailableLists ) {
newEntry = newAvailableLists[assetKey];
if ( newEntry.submitter !== 'user' ) { continue; }
if ( dict.has(assetKey) ) { continue; }
delete newAvailableLists[assetKey];
µb.assets.unregisterAssetSource(assetKey);
µb.removeFilterList(assetKey);
2016-01-03 19:58:25 +01:00
}
2014-06-24 00:42:43 +02:00
};
2017-01-26 16:17:38 +01:00
// Built-in filter lists loaded.
var onBuiltinListsLoaded = function(entries) {
for ( var assetKey in entries ) {
if ( entries.hasOwnProperty(assetKey) === false ) { continue; }
entry = entries[assetKey];
if ( entry.content !== 'filters' ) { continue; }
newAvailableLists[assetKey] = Object.assign({}, entry);
2014-07-25 22:12:20 +02:00
}
// Load set of currently selected filter lists.
2017-01-26 16:17:38 +01:00
var listKeySet = new Set(µb.selectedFilterLists);
for ( listKey in newAvailableLists ) {
if ( newAvailableLists.hasOwnProperty(listKey) ) {
newAvailableLists[listKey].off = !listKeySet.has(listKey);
}
}
finalize();
callback(newAvailableLists);
};
2016-01-03 19:58:25 +01:00
// Available lists previously computed.
var onOldAvailableListsLoaded = function(bin) {
oldAvailableLists = bin && bin.availableFilterLists || {};
µb.assets.metadata(onBuiltinListsLoaded);
};
2016-01-03 19:58:25 +01:00
// Load previously saved available lists -- these contains data
// computed at run-time, we will reuse this data if possible.
vAPI.storage.get('availableFilterLists', onOldAvailableListsLoaded);
2015-02-24 00:31:29 +01:00
};
/******************************************************************************/
2016-01-03 19:58:25 +01:00
// This is used to be re-entrancy resistant.
µBlock.loadingFilterLists = false;
2014-09-08 23:46:58 +02:00
µBlock.loadFilterLists = function(callback) {
2016-01-03 19:58:25 +01:00
// Callers are expected to check this first.
Squashed commit of the following: commit 7c6cacc59b27660fabacb55d668ef099b222a9e6 Author: Raymond Hill <rhill@raymondhill.net> Date: Sat Nov 3 08:52:51 2018 -0300 code review: finalize support for wasm-based hntrie commit 8596ed80e3bdac2c36e3c860b51e7189f6bc8487 Merge: cbe1f2e 000eb82 Author: Raymond Hill <rhill@raymondhill.net> Date: Sat Nov 3 08:41:40 2018 -0300 Merge branch 'master' of github.com:gorhill/uBlock into trie-wasm commit cbe1f2e2f38484d42af3204ec7f1b5decd30f99e Merge: 270fc7f dbb7e80 Author: Raymond Hill <rhill@raymondhill.net> Date: Fri Nov 2 17:43:20 2018 -0300 Merge branch 'master' of github.com:gorhill/uBlock into trie-wasm commit 270fc7f9b3b73d79e6355522c1a42ce782fe7e5c Merge: d2a89cf d693d4f Author: Raymond Hill <rhill@raymondhill.net> Date: Fri Nov 2 16:21:08 2018 -0300 Merge branch 'master' of github.com:gorhill/uBlock into trie-wasm commit d2a89cf28f0816ffd4617c2c7b4ccfcdcc30e1b4 Merge: d7afc78 649f82f Author: Raymond Hill <rhill@raymondhill.net> Date: Fri Nov 2 14:54:58 2018 -0300 Merge branch 'master' of github.com:gorhill/uBlock into trie-wasm commit d7afc78b5f5675d7d34c5a1d0ec3099a77caef49 Author: Raymond Hill <rhill@raymondhill.net> Date: Fri Nov 2 13:56:11 2018 -0300 finalize wasm-based hntrie implementation commit e7b9e043cf36ad055791713e34eb0322dec84627 Author: Raymond Hill <rhill@raymondhill.net> Date: Fri Nov 2 08:14:02 2018 -0300 add first-pass implementation of wasm version of hntrie commit 1015cb34624f3ef73ace58b58fe4e03dfc59897f Author: Raymond Hill <rhill@raymondhill.net> Date: Wed Oct 31 17:16:47 2018 -0300 back up draft work toward experimenting with wasm hntries
2018-11-03 12:58:46 +01:00
if ( this.loadingFilterLists ) { return; }
2016-01-03 19:58:25 +01:00
this.loadingFilterLists = true;
2015-02-24 00:31:29 +01:00
const µb = µBlock;
const loadedListKeys = [];
let filterlistsCount = 0;
2014-07-25 22:12:20 +02:00
2014-09-08 23:46:58 +02:00
if ( typeof callback !== 'function' ) {
callback = this.noopFunc;
}
const onDone = function() {
µb.staticNetFilteringEngine.freeze();
µb.staticExtFilteringEngine.freeze();
µb.redirectEngine.freeze();
vAPI.storage.set({ 'availableFilterLists': µb.availableFilterLists });
2015-02-24 19:48:03 +01:00
vAPI.messaging.broadcast({
what: 'staticFilteringDataChanged',
parseCosmeticFilters: µb.userSettings.parseAllABPHideFilters,
ignoreGenericCosmeticFilters: µb.userSettings.ignoreGenericCosmeticFilters,
listKeys: loadedListKeys
});
2016-09-24 20:36:08 +02:00
2014-09-08 23:46:58 +02:00
callback();
2015-02-24 00:31:29 +01:00
2017-05-06 19:19:05 +02:00
µb.selfieManager.destroy();
µb.lz4Codec.relinquish();
2016-01-03 19:58:25 +01:00
µb.loadingFilterLists = false;
2014-07-25 22:12:20 +02:00
};
const applyCompiledFilters = function(assetKey, compiled) {
const snfe = µb.staticNetFilteringEngine;
const sxfe = µb.staticExtFilteringEngine;
let acceptedCount = snfe.acceptedCount + sxfe.acceptedCount,
discardedCount = snfe.discardedCount + sxfe.discardedCount;
µb.applyCompiledFilters(compiled, assetKey === µb.userFiltersPath);
if ( µb.availableFilterLists.hasOwnProperty(assetKey) ) {
const entry = µb.availableFilterLists[assetKey];
entry.entryCount = snfe.acceptedCount + sxfe.acceptedCount -
acceptedCount;
entry.entryUsedCount = entry.entryCount -
(snfe.discardedCount + sxfe.discardedCount - discardedCount);
2014-06-24 00:42:43 +02:00
}
loadedListKeys.push(assetKey);
2014-06-24 00:42:43 +02:00
};
const onCompiledListLoaded = function(details) {
applyCompiledFilters(details.assetKey, details.content);
2015-02-24 00:31:29 +01:00
filterlistsCount -= 1;
if ( filterlistsCount === 0 ) {
onDone();
}
};
const onFilterListsReady = function(lists) {
µb.availableFilterLists = lists;
2015-02-24 00:31:29 +01:00
µb.redirectEngine.reset();
µb.staticExtFilteringEngine.reset();
2015-02-24 00:31:29 +01:00
µb.staticNetFilteringEngine.reset();
2015-11-29 23:06:58 +01:00
µb.selfieManager.destroy();
µb.staticFilteringReverseLookup.resetLists();
2014-06-24 00:42:43 +02:00
2015-02-24 00:31:29 +01:00
// We need to build a complete list of assets to pull first: this is
// because it *may* happens that some load operations are synchronous:
// This happens for assets which do not exist, ot assets with no
// content.
const toLoad = [];
for ( const assetKey in lists ) {
if ( lists.hasOwnProperty(assetKey) === false ) { continue; }
if ( lists[assetKey].off ) { continue; }
toLoad.push(assetKey);
2015-02-06 07:20:04 +01:00
}
2015-02-24 00:31:29 +01:00
filterlistsCount = toLoad.length;
if ( filterlistsCount === 0 ) {
2016-01-03 19:58:25 +01:00
return onDone();
2015-02-24 00:31:29 +01:00
}
let i = toLoad.length;
2015-02-24 00:31:29 +01:00
while ( i-- ) {
µb.getCompiledFilterList(toLoad[i], onCompiledListLoaded);
2014-06-24 00:42:43 +02:00
}
};
2015-02-24 00:31:29 +01:00
this.getAvailableLists(onFilterListsReady);
this.loadRedirectResources();
2014-06-24 00:42:43 +02:00
};
/******************************************************************************/
µBlock.getCompiledFilterList = function(assetKey, callback) {
var µb = this,
compiledPath = 'compiled/' + assetKey,
rawContent;
2014-06-24 00:42:43 +02:00
var onCompiledListLoaded2 = function(details) {
if ( details.content === '' ) {
details.content = µb.compileFilters(
rawContent,
{ assetKey: assetKey }
);
µb.assets.put(compiledPath, details.content);
}
rawContent = undefined;
details.assetKey = assetKey;
callback(details);
};
var onRawListLoaded = function(details) {
2015-06-08 14:09:08 +02:00
if ( details.content === '' ) {
details.assetKey = assetKey;
2015-06-08 14:09:08 +02:00
callback(details);
return;
}
µb.extractFilterListMetadata(assetKey, details.content);
// Fectching the raw content may cause the compiled content to be
// generated somewhere else in uBO, hence we try one last time to
// fetch the compiled content in case it has become available.
rawContent = details.content;
µb.assets.get(compiledPath, onCompiledListLoaded2);
2015-02-24 00:31:29 +01:00
};
2014-09-25 21:44:18 +02:00
var onCompiledListLoaded1 = function(details) {
2015-02-24 00:31:29 +01:00
if ( details.content === '' ) {
µb.assets.get(assetKey, onRawListLoaded);
2015-02-24 00:31:29 +01:00
return;
}
details.assetKey = assetKey;
2015-02-24 00:31:29 +01:00
callback(details);
};
2014-09-25 21:44:18 +02:00
this.assets.get(compiledPath, onCompiledListLoaded1);
2015-02-24 00:31:29 +01:00
};
2015-02-24 00:31:29 +01:00
/******************************************************************************/
2018-02-23 12:42:17 +01:00
// https://github.com/gorhill/uBlock/issues/3406
// Lower minimum update period to 1 day.
µBlock.extractFilterListMetadata = function(assetKey, raw) {
let listEntry = this.availableFilterLists[assetKey];
if ( listEntry === undefined ) { return; }
// Metadata expected to be found at the top of content.
let head = raw.slice(0, 1024);
// https://github.com/gorhill/uBlock/issues/313
// Always try to fetch the name if this is an external filter list.
if ( listEntry.title === '' || listEntry.group === 'custom' ) {
let matches = head.match(/(?:^|\n)(?:!|# )[\t ]*Title[\t ]*:([^\n]+)/i);
if ( matches !== null ) {
// https://bugs.chromium.org/p/v8/issues/detail?id=2869
// orphanizeString is to work around String.slice()
// potentially causing the whole raw filter list to be held in
// memory just because we cut out the title as a substring.
listEntry.title = this.orphanizeString(matches[1].trim());
}
}
// Extract update frequency information
let matches = head.match(/(?:^|\n)(?:!|# )[\t ]*Expires[\t ]*:[\t ]*(\d+)[\t ]*(h)?/i);
if ( matches !== null ) {
let v = Math.max(parseInt(matches[1], 10), 1);
2018-05-16 20:55:12 +02:00
if ( matches[2] !== undefined ) {
v = Math.ceil(v / 24);
}
if ( v !== listEntry.updateAfter ) {
this.assets.registerAssetSource(assetKey, { updateAfter: v });
}
}
2014-09-25 21:44:18 +02:00
};
/******************************************************************************/
µBlock.removeCompiledFilterList = function(assetKey) {
this.assets.remove('compiled/' + assetKey);
};
µBlock.removeFilterList = function(assetKey) {
this.removeCompiledFilterList(assetKey);
this.assets.remove(assetKey);
};
/******************************************************************************/
2014-09-25 21:44:18 +02:00
µBlock.compileFilters = function(rawText, details) {
let writer = new this.CompiledLineIO.Writer();
2014-06-24 00:42:43 +02:00
// Populate the writer with information potentially useful to the
// client compilers.
if ( details ) {
if ( details.assetKey ) {
writer.properties.set('assetKey', details.assetKey);
}
}
2014-06-24 00:42:43 +02:00
// Useful references:
// https://adblockplus.org/en/filter-cheatsheet
// https://adblockplus.org/en/filters
const staticNetFilteringEngine = this.staticNetFilteringEngine;
const staticExtFilteringEngine = this.staticExtFilteringEngine;
const reIsWhitespaceChar = /\s/;
const reMaybeLocalIp = /^[\d:f]/;
const reIsLocalhostRedirect = /\s+(?:0\.0\.0\.0|broadcasthost|localhost|local|ip6-\w+)\b/;
const reLocalIp = /^(?:0\.0\.0\.0|127\.0\.0\.1|::1|fe80::1%lo0)/;
const lineIter = new this.LineIterator(this.processDirectives(rawText));
while ( lineIter.eot() === false ) {
2014-06-24 00:42:43 +02:00
// rhill 2014-04-18: The trim is important here, as without it there
// could be a lingering `\r` which would cause problems in the
// following parsing code.
let line = lineIter.next().trim();
if ( line.length === 0 ) { continue; }
2015-01-23 17:32:49 +01:00
2014-06-24 00:42:43 +02:00
// Strip comments
const c = line.charAt(0);
if ( c === '!' || c === '[' ) { continue; }
2014-06-24 00:42:43 +02:00
2014-09-25 21:44:18 +02:00
// Parse or skip cosmetic filters
2015-01-23 17:32:49 +01:00
// All cosmetic filters are caught here
if ( staticExtFilteringEngine.compile(line, writer) ) { continue; }
2014-06-24 00:42:43 +02:00
2015-01-23 17:32:49 +01:00
// Whatever else is next can be assumed to not be a cosmetic filter
// Most comments start in first column
if ( c === '#' ) { continue; }
2014-06-24 00:42:43 +02:00
2015-01-23 17:32:49 +01:00
// Catch comments somewhere on the line
// Remove:
// ... #blah blah blah
// ... # blah blah blah
// Don't remove:
// ...#blah blah blah
// because some ABP filters uses the `#` character (URL fragment)
const pos = line.indexOf('#');
2015-01-23 17:32:49 +01:00
if ( pos !== -1 && reIsWhitespaceChar.test(line.charAt(pos - 1)) ) {
line = line.slice(0, pos).trim();
}
2014-06-24 00:42:43 +02:00
// https://github.com/gorhill/httpswitchboard/issues/15
// Ensure localhost et al. don't end up in the ubiquitous blacklist.
2015-01-23 17:32:49 +01:00
// With hosts files, we need to remove local IP redirection
if ( reMaybeLocalIp.test(c) ) {
// Ignore hosts file redirect configuration
// 127.0.0.1 localhost
// 255.255.255.255 broadcasthost
if ( reIsLocalhostRedirect.test(line) ) { continue; }
2015-01-23 17:32:49 +01:00
line = line.replace(reLocalIp, '').trim();
}
2015-03-07 19:20:18 +01:00
if ( line.length === 0 ) { continue; }
2014-06-24 00:42:43 +02:00
staticNetFilteringEngine.compile(line, writer);
2015-02-24 00:31:29 +01:00
}
return writer.toString();
2015-02-24 00:31:29 +01:00
};
2014-06-24 00:42:43 +02:00
2015-02-24 00:31:29 +01:00
/******************************************************************************/
2014-06-24 00:42:43 +02:00
2016-02-17 15:28:20 +01:00
// https://github.com/gorhill/uBlock/issues/1395
2016-02-17 16:04:55 +01:00
// Added `firstparty` argument: to avoid discarding cosmetic filters when
2016-02-17 15:28:20 +01:00
// applying 1st-party filters.
µBlock.applyCompiledFilters = function(rawText, firstparty) {
if ( rawText === '' ) { return; }
let reader = new this.CompiledLineIO.Reader(rawText);
this.staticNetFilteringEngine.fromCompiledContent(reader);
this.staticExtFilteringEngine.fromCompiledContent(reader, {
skipGenericCosmetic: this.userSettings.ignoreGenericCosmeticFilters,
skipCosmetic: !firstparty && !this.userSettings.parseAllABPHideFilters
});
2014-06-24 00:42:43 +02:00
};
/******************************************************************************/
// https://github.com/AdguardTeam/AdguardBrowserExtension/issues/917
µBlock.processDirectives = function(content) {
const reIf = /^!#(if|endif)\b([^\n]*)/gm;
const parts = [];
let beg = 0, depth = 0, discard = false;
while ( beg < content.length ) {
const match = reIf.exec(content);
if ( match === null ) { break; }
if ( match[1] === 'if' ) {
let expr = match[2].trim();
const target = expr.startsWith('!');
if ( target ) { expr = expr.slice(1); }
const token = this.processDirectives.tokens.get(expr);
if (
depth === 0 &&
discard === false &&
token !== undefined &&
vAPI.webextFlavor.soup.has(token) === target
) {
parts.push(content.slice(beg, match.index));
discard = true;
}
depth += 1;
continue;
}
depth -= 1;
if ( depth < 0 ) { break; }
if ( depth === 0 && discard ) {
beg = match.index + match[0].length + 1;
discard = false;
}
}
if ( depth === 0 && parts.length !== 0 ) {
parts.push(content.slice(beg));
content = parts.join('\n');
}
return content.trim();
};
µBlock.processDirectives.tokens = new Map([
[ 'ext_ublock', 'ublock' ],
[ 'env_chromium', 'chromium' ],
[ 'env_edge', 'edge' ],
[ 'env_firefox', 'firefox' ],
[ 'env_mobile', 'mobile' ],
[ 'env_safari', 'safari' ],
[ 'cap_html_filtering', 'html_filtering' ],
[ 'cap_user_stylesheet', 'user_stylesheet' ]
]);
/******************************************************************************/
2018-02-15 23:25:38 +01:00
µBlock.loadRedirectResources = function(updatedContent) {
Refactor selfie generation into a more flexible persistence mechanism The motivation is to address the higher peak memory usage at launch time with 3rd-gen HNTrie when a selfie was present. The selfie generation prior to this change was to collect all filtering data into a single data structure, and then to serialize that whole structure at once into storage (using JSON.stringify). However, HNTrie serialization requires that a large UintArray32 be converted into a plain JS array, which itslef would be indirectly converted into a JSON string. This was the main reason why peak memory usage would be higher at launch from selfie, since the JSON string would need to be wholly unserialized into JS objects, which themselves would need to be converted into more specialized data structures (like that Uint32Array one). The solution to lower peak memory usage at launch is to refactor selfie generation to allow a more piecemeal approach: each filtering component is given the ability to serialize itself rather than to be forced to be embedded in the master selfie. With this approach, the HNTrie buffer can now serialize to its own storage by converting the buffer data directly into a string which can be directly sent to storage. This avoiding expensive intermediate steps such as converting into a JS array and then to a JSON string. As part of the refactoring, there was also opportunistic code upgrade to ES6 and Promise (eventually all of uBO's code will be proper ES6). Additionally, the polyfill to bring getBytesInUse() to Firefox has been revisited to replace the rather expensive previous implementation with an implementation with virtually no overhead.
2019-02-14 19:33:55 +01:00
let content = '';
Refactor selfie generation into a more flexible persistence mechanism The motivation is to address the higher peak memory usage at launch time with 3rd-gen HNTrie when a selfie was present. The selfie generation prior to this change was to collect all filtering data into a single data structure, and then to serialize that whole structure at once into storage (using JSON.stringify). However, HNTrie serialization requires that a large UintArray32 be converted into a plain JS array, which itslef would be indirectly converted into a JSON string. This was the main reason why peak memory usage would be higher at launch from selfie, since the JSON string would need to be wholly unserialized into JS objects, which themselves would need to be converted into more specialized data structures (like that Uint32Array one). The solution to lower peak memory usage at launch is to refactor selfie generation to allow a more piecemeal approach: each filtering component is given the ability to serialize itself rather than to be forced to be embedded in the master selfie. With this approach, the HNTrie buffer can now serialize to its own storage by converting the buffer data directly into a string which can be directly sent to storage. This avoiding expensive intermediate steps such as converting into a JS array and then to a JSON string. As part of the refactoring, there was also opportunistic code upgrade to ES6 and Promise (eventually all of uBO's code will be proper ES6). Additionally, the polyfill to bring getBytesInUse() to Firefox has been revisited to replace the rather expensive previous implementation with an implementation with virtually no overhead.
2019-02-14 19:33:55 +01:00
const onDone = ( ) => {
this.redirectEngine.resourcesFromString(content);
2017-03-05 18:54:47 +01:00
};
Refactor selfie generation into a more flexible persistence mechanism The motivation is to address the higher peak memory usage at launch time with 3rd-gen HNTrie when a selfie was present. The selfie generation prior to this change was to collect all filtering data into a single data structure, and then to serialize that whole structure at once into storage (using JSON.stringify). However, HNTrie serialization requires that a large UintArray32 be converted into a plain JS array, which itslef would be indirectly converted into a JSON string. This was the main reason why peak memory usage would be higher at launch from selfie, since the JSON string would need to be wholly unserialized into JS objects, which themselves would need to be converted into more specialized data structures (like that Uint32Array one). The solution to lower peak memory usage at launch is to refactor selfie generation to allow a more piecemeal approach: each filtering component is given the ability to serialize itself rather than to be forced to be embedded in the master selfie. With this approach, the HNTrie buffer can now serialize to its own storage by converting the buffer data directly into a string which can be directly sent to storage. This avoiding expensive intermediate steps such as converting into a JS array and then to a JSON string. As part of the refactoring, there was also opportunistic code upgrade to ES6 and Promise (eventually all of uBO's code will be proper ES6). Additionally, the polyfill to bring getBytesInUse() to Firefox has been revisited to replace the rather expensive previous implementation with an implementation with virtually no overhead.
2019-02-14 19:33:55 +01:00
const onUserResourcesLoaded = details => {
2017-03-05 18:54:47 +01:00
if ( details.content !== '' ) {
content += '\n\n' + details.content;
}
onDone();
};
Refactor selfie generation into a more flexible persistence mechanism The motivation is to address the higher peak memory usage at launch time with 3rd-gen HNTrie when a selfie was present. The selfie generation prior to this change was to collect all filtering data into a single data structure, and then to serialize that whole structure at once into storage (using JSON.stringify). However, HNTrie serialization requires that a large UintArray32 be converted into a plain JS array, which itslef would be indirectly converted into a JSON string. This was the main reason why peak memory usage would be higher at launch from selfie, since the JSON string would need to be wholly unserialized into JS objects, which themselves would need to be converted into more specialized data structures (like that Uint32Array one). The solution to lower peak memory usage at launch is to refactor selfie generation to allow a more piecemeal approach: each filtering component is given the ability to serialize itself rather than to be forced to be embedded in the master selfie. With this approach, the HNTrie buffer can now serialize to its own storage by converting the buffer data directly into a string which can be directly sent to storage. This avoiding expensive intermediate steps such as converting into a JS array and then to a JSON string. As part of the refactoring, there was also opportunistic code upgrade to ES6 and Promise (eventually all of uBO's code will be proper ES6). Additionally, the polyfill to bring getBytesInUse() to Firefox has been revisited to replace the rather expensive previous implementation with an implementation with virtually no overhead.
2019-02-14 19:33:55 +01:00
const onResourcesLoaded = details => {
if ( details.content !== '' ) {
2017-03-05 18:54:47 +01:00
content = details.content;
}
Refactor selfie generation into a more flexible persistence mechanism The motivation is to address the higher peak memory usage at launch time with 3rd-gen HNTrie when a selfie was present. The selfie generation prior to this change was to collect all filtering data into a single data structure, and then to serialize that whole structure at once into storage (using JSON.stringify). However, HNTrie serialization requires that a large UintArray32 be converted into a plain JS array, which itslef would be indirectly converted into a JSON string. This was the main reason why peak memory usage would be higher at launch from selfie, since the JSON string would need to be wholly unserialized into JS objects, which themselves would need to be converted into more specialized data structures (like that Uint32Array one). The solution to lower peak memory usage at launch is to refactor selfie generation to allow a more piecemeal approach: each filtering component is given the ability to serialize itself rather than to be forced to be embedded in the master selfie. With this approach, the HNTrie buffer can now serialize to its own storage by converting the buffer data directly into a string which can be directly sent to storage. This avoiding expensive intermediate steps such as converting into a JS array and then to a JSON string. As part of the refactoring, there was also opportunistic code upgrade to ES6 and Promise (eventually all of uBO's code will be proper ES6). Additionally, the polyfill to bring getBytesInUse() to Firefox has been revisited to replace the rather expensive previous implementation with an implementation with virtually no overhead.
2019-02-14 19:33:55 +01:00
if ( this.hiddenSettings.userResourcesLocation === 'unset' ) {
2017-03-05 18:54:47 +01:00
return onDone();
}
Refactor selfie generation into a more flexible persistence mechanism The motivation is to address the higher peak memory usage at launch time with 3rd-gen HNTrie when a selfie was present. The selfie generation prior to this change was to collect all filtering data into a single data structure, and then to serialize that whole structure at once into storage (using JSON.stringify). However, HNTrie serialization requires that a large UintArray32 be converted into a plain JS array, which itslef would be indirectly converted into a JSON string. This was the main reason why peak memory usage would be higher at launch from selfie, since the JSON string would need to be wholly unserialized into JS objects, which themselves would need to be converted into more specialized data structures (like that Uint32Array one). The solution to lower peak memory usage at launch is to refactor selfie generation to allow a more piecemeal approach: each filtering component is given the ability to serialize itself rather than to be forced to be embedded in the master selfie. With this approach, the HNTrie buffer can now serialize to its own storage by converting the buffer data directly into a string which can be directly sent to storage. This avoiding expensive intermediate steps such as converting into a JS array and then to a JSON string. As part of the refactoring, there was also opportunistic code upgrade to ES6 and Promise (eventually all of uBO's code will be proper ES6). Additionally, the polyfill to bring getBytesInUse() to Firefox has been revisited to replace the rather expensive previous implementation with an implementation with virtually no overhead.
2019-02-14 19:33:55 +01:00
this.assets.fetchText(
this.hiddenSettings.userResourcesLocation,
onUserResourcesLoaded
);
};
2018-02-15 23:25:38 +01:00
if ( typeof updatedContent === 'string' && updatedContent.length !== 0 ) {
return onResourcesLoaded({ content: updatedContent });
}
Refactor selfie generation into a more flexible persistence mechanism The motivation is to address the higher peak memory usage at launch time with 3rd-gen HNTrie when a selfie was present. The selfie generation prior to this change was to collect all filtering data into a single data structure, and then to serialize that whole structure at once into storage (using JSON.stringify). However, HNTrie serialization requires that a large UintArray32 be converted into a plain JS array, which itslef would be indirectly converted into a JSON string. This was the main reason why peak memory usage would be higher at launch from selfie, since the JSON string would need to be wholly unserialized into JS objects, which themselves would need to be converted into more specialized data structures (like that Uint32Array one). The solution to lower peak memory usage at launch is to refactor selfie generation to allow a more piecemeal approach: each filtering component is given the ability to serialize itself rather than to be forced to be embedded in the master selfie. With this approach, the HNTrie buffer can now serialize to its own storage by converting the buffer data directly into a string which can be directly sent to storage. This avoiding expensive intermediate steps such as converting into a JS array and then to a JSON string. As part of the refactoring, there was also opportunistic code upgrade to ES6 and Promise (eventually all of uBO's code will be proper ES6). Additionally, the polyfill to bring getBytesInUse() to Firefox has been revisited to replace the rather expensive previous implementation with an implementation with virtually no overhead.
2019-02-14 19:33:55 +01:00
this.redirectEngine.resourcesFromSelfie().then(success => {
2018-02-15 23:25:38 +01:00
if ( success !== true ) {
Refactor selfie generation into a more flexible persistence mechanism The motivation is to address the higher peak memory usage at launch time with 3rd-gen HNTrie when a selfie was present. The selfie generation prior to this change was to collect all filtering data into a single data structure, and then to serialize that whole structure at once into storage (using JSON.stringify). However, HNTrie serialization requires that a large UintArray32 be converted into a plain JS array, which itslef would be indirectly converted into a JSON string. This was the main reason why peak memory usage would be higher at launch from selfie, since the JSON string would need to be wholly unserialized into JS objects, which themselves would need to be converted into more specialized data structures (like that Uint32Array one). The solution to lower peak memory usage at launch is to refactor selfie generation to allow a more piecemeal approach: each filtering component is given the ability to serialize itself rather than to be forced to be embedded in the master selfie. With this approach, the HNTrie buffer can now serialize to its own storage by converting the buffer data directly into a string which can be directly sent to storage. This avoiding expensive intermediate steps such as converting into a JS array and then to a JSON string. As part of the refactoring, there was also opportunistic code upgrade to ES6 and Promise (eventually all of uBO's code will be proper ES6). Additionally, the polyfill to bring getBytesInUse() to Firefox has been revisited to replace the rather expensive previous implementation with an implementation with virtually no overhead.
2019-02-14 19:33:55 +01:00
this.assets.get('ublock-resources', onResourcesLoaded);
2018-02-15 23:25:38 +01:00
}
Refactor selfie generation into a more flexible persistence mechanism The motivation is to address the higher peak memory usage at launch time with 3rd-gen HNTrie when a selfie was present. The selfie generation prior to this change was to collect all filtering data into a single data structure, and then to serialize that whole structure at once into storage (using JSON.stringify). However, HNTrie serialization requires that a large UintArray32 be converted into a plain JS array, which itslef would be indirectly converted into a JSON string. This was the main reason why peak memory usage would be higher at launch from selfie, since the JSON string would need to be wholly unserialized into JS objects, which themselves would need to be converted into more specialized data structures (like that Uint32Array one). The solution to lower peak memory usage at launch is to refactor selfie generation to allow a more piecemeal approach: each filtering component is given the ability to serialize itself rather than to be forced to be embedded in the master selfie. With this approach, the HNTrie buffer can now serialize to its own storage by converting the buffer data directly into a string which can be directly sent to storage. This avoiding expensive intermediate steps such as converting into a JS array and then to a JSON string. As part of the refactoring, there was also opportunistic code upgrade to ES6 and Promise (eventually all of uBO's code will be proper ES6). Additionally, the polyfill to bring getBytesInUse() to Firefox has been revisited to replace the rather expensive previous implementation with an implementation with virtually no overhead.
2019-02-14 19:33:55 +01:00
});
};
/******************************************************************************/
Squashed commit of the following: commit 7c6cacc59b27660fabacb55d668ef099b222a9e6 Author: Raymond Hill <rhill@raymondhill.net> Date: Sat Nov 3 08:52:51 2018 -0300 code review: finalize support for wasm-based hntrie commit 8596ed80e3bdac2c36e3c860b51e7189f6bc8487 Merge: cbe1f2e 000eb82 Author: Raymond Hill <rhill@raymondhill.net> Date: Sat Nov 3 08:41:40 2018 -0300 Merge branch 'master' of github.com:gorhill/uBlock into trie-wasm commit cbe1f2e2f38484d42af3204ec7f1b5decd30f99e Merge: 270fc7f dbb7e80 Author: Raymond Hill <rhill@raymondhill.net> Date: Fri Nov 2 17:43:20 2018 -0300 Merge branch 'master' of github.com:gorhill/uBlock into trie-wasm commit 270fc7f9b3b73d79e6355522c1a42ce782fe7e5c Merge: d2a89cf d693d4f Author: Raymond Hill <rhill@raymondhill.net> Date: Fri Nov 2 16:21:08 2018 -0300 Merge branch 'master' of github.com:gorhill/uBlock into trie-wasm commit d2a89cf28f0816ffd4617c2c7b4ccfcdcc30e1b4 Merge: d7afc78 649f82f Author: Raymond Hill <rhill@raymondhill.net> Date: Fri Nov 2 14:54:58 2018 -0300 Merge branch 'master' of github.com:gorhill/uBlock into trie-wasm commit d7afc78b5f5675d7d34c5a1d0ec3099a77caef49 Author: Raymond Hill <rhill@raymondhill.net> Date: Fri Nov 2 13:56:11 2018 -0300 finalize wasm-based hntrie implementation commit e7b9e043cf36ad055791713e34eb0322dec84627 Author: Raymond Hill <rhill@raymondhill.net> Date: Fri Nov 2 08:14:02 2018 -0300 add first-pass implementation of wasm version of hntrie commit 1015cb34624f3ef73ace58b58fe4e03dfc59897f Author: Raymond Hill <rhill@raymondhill.net> Date: Wed Oct 31 17:16:47 2018 -0300 back up draft work toward experimenting with wasm hntries
2018-11-03 12:58:46 +01:00
µBlock.loadPublicSuffixList = function() {
if ( this.hiddenSettings.disableWebAssembly === false ) {
publicSuffixList.enableWASM();
}
Refactor selfie generation into a more flexible persistence mechanism The motivation is to address the higher peak memory usage at launch time with 3rd-gen HNTrie when a selfie was present. The selfie generation prior to this change was to collect all filtering data into a single data structure, and then to serialize that whole structure at once into storage (using JSON.stringify). However, HNTrie serialization requires that a large UintArray32 be converted into a plain JS array, which itslef would be indirectly converted into a JSON string. This was the main reason why peak memory usage would be higher at launch from selfie, since the JSON string would need to be wholly unserialized into JS objects, which themselves would need to be converted into more specialized data structures (like that Uint32Array one). The solution to lower peak memory usage at launch is to refactor selfie generation to allow a more piecemeal approach: each filtering component is given the ability to serialize itself rather than to be forced to be embedded in the master selfie. With this approach, the HNTrie buffer can now serialize to its own storage by converting the buffer data directly into a string which can be directly sent to storage. This avoiding expensive intermediate steps such as converting into a JS array and then to a JSON string. As part of the refactoring, there was also opportunistic code upgrade to ES6 and Promise (eventually all of uBO's code will be proper ES6). Additionally, the polyfill to bring getBytesInUse() to Firefox has been revisited to replace the rather expensive previous implementation with an implementation with virtually no overhead.
2019-02-14 19:33:55 +01:00
return this.assets.get(
'compiled/' + this.pslAssetKey
).then(details =>
publicSuffixList.fromSelfie(details.content, µBlock.base128)
).catch(reason => {
console.info(reason);
return false;
}).then(success => {
if ( success ) { return; }
Refactor selfie generation into a more flexible persistence mechanism The motivation is to address the higher peak memory usage at launch time with 3rd-gen HNTrie when a selfie was present. The selfie generation prior to this change was to collect all filtering data into a single data structure, and then to serialize that whole structure at once into storage (using JSON.stringify). However, HNTrie serialization requires that a large UintArray32 be converted into a plain JS array, which itslef would be indirectly converted into a JSON string. This was the main reason why peak memory usage would be higher at launch from selfie, since the JSON string would need to be wholly unserialized into JS objects, which themselves would need to be converted into more specialized data structures (like that Uint32Array one). The solution to lower peak memory usage at launch is to refactor selfie generation to allow a more piecemeal approach: each filtering component is given the ability to serialize itself rather than to be forced to be embedded in the master selfie. With this approach, the HNTrie buffer can now serialize to its own storage by converting the buffer data directly into a string which can be directly sent to storage. This avoiding expensive intermediate steps such as converting into a JS array and then to a JSON string. As part of the refactoring, there was also opportunistic code upgrade to ES6 and Promise (eventually all of uBO's code will be proper ES6). Additionally, the polyfill to bring getBytesInUse() to Firefox has been revisited to replace the rather expensive previous implementation with an implementation with virtually no overhead.
2019-02-14 19:33:55 +01:00
return this.assets.get(this.pslAssetKey, details => {
Squashed commit of the following: commit 7c6cacc59b27660fabacb55d668ef099b222a9e6 Author: Raymond Hill <rhill@raymondhill.net> Date: Sat Nov 3 08:52:51 2018 -0300 code review: finalize support for wasm-based hntrie commit 8596ed80e3bdac2c36e3c860b51e7189f6bc8487 Merge: cbe1f2e 000eb82 Author: Raymond Hill <rhill@raymondhill.net> Date: Sat Nov 3 08:41:40 2018 -0300 Merge branch 'master' of github.com:gorhill/uBlock into trie-wasm commit cbe1f2e2f38484d42af3204ec7f1b5decd30f99e Merge: 270fc7f dbb7e80 Author: Raymond Hill <rhill@raymondhill.net> Date: Fri Nov 2 17:43:20 2018 -0300 Merge branch 'master' of github.com:gorhill/uBlock into trie-wasm commit 270fc7f9b3b73d79e6355522c1a42ce782fe7e5c Merge: d2a89cf d693d4f Author: Raymond Hill <rhill@raymondhill.net> Date: Fri Nov 2 16:21:08 2018 -0300 Merge branch 'master' of github.com:gorhill/uBlock into trie-wasm commit d2a89cf28f0816ffd4617c2c7b4ccfcdcc30e1b4 Merge: d7afc78 649f82f Author: Raymond Hill <rhill@raymondhill.net> Date: Fri Nov 2 14:54:58 2018 -0300 Merge branch 'master' of github.com:gorhill/uBlock into trie-wasm commit d7afc78b5f5675d7d34c5a1d0ec3099a77caef49 Author: Raymond Hill <rhill@raymondhill.net> Date: Fri Nov 2 13:56:11 2018 -0300 finalize wasm-based hntrie implementation commit e7b9e043cf36ad055791713e34eb0322dec84627 Author: Raymond Hill <rhill@raymondhill.net> Date: Fri Nov 2 08:14:02 2018 -0300 add first-pass implementation of wasm version of hntrie commit 1015cb34624f3ef73ace58b58fe4e03dfc59897f Author: Raymond Hill <rhill@raymondhill.net> Date: Wed Oct 31 17:16:47 2018 -0300 back up draft work toward experimenting with wasm hntries
2018-11-03 12:58:46 +01:00
if ( details.content !== '' ) {
this.compilePublicSuffixList(details.content);
}
});
});
};
µBlock.compilePublicSuffixList = function(content) {
publicSuffixList.parse(content, punycode.toASCII);
this.assets.put(
'compiled/' + this.pslAssetKey,
Refactor selfie generation into a more flexible persistence mechanism The motivation is to address the higher peak memory usage at launch time with 3rd-gen HNTrie when a selfie was present. The selfie generation prior to this change was to collect all filtering data into a single data structure, and then to serialize that whole structure at once into storage (using JSON.stringify). However, HNTrie serialization requires that a large UintArray32 be converted into a plain JS array, which itslef would be indirectly converted into a JSON string. This was the main reason why peak memory usage would be higher at launch from selfie, since the JSON string would need to be wholly unserialized into JS objects, which themselves would need to be converted into more specialized data structures (like that Uint32Array one). The solution to lower peak memory usage at launch is to refactor selfie generation to allow a more piecemeal approach: each filtering component is given the ability to serialize itself rather than to be forced to be embedded in the master selfie. With this approach, the HNTrie buffer can now serialize to its own storage by converting the buffer data directly into a string which can be directly sent to storage. This avoiding expensive intermediate steps such as converting into a JS array and then to a JSON string. As part of the refactoring, there was also opportunistic code upgrade to ES6 and Promise (eventually all of uBO's code will be proper ES6). Additionally, the polyfill to bring getBytesInUse() to Firefox has been revisited to replace the rather expensive previous implementation with an implementation with virtually no overhead.
2019-02-14 19:33:55 +01:00
publicSuffixList.toSelfie(µBlock.base128)
);
2014-06-24 00:42:43 +02:00
};
/******************************************************************************/
2014-09-08 23:46:58 +02:00
// This is to be sure the selfie is generated in a sane manner: the selfie will
// be generated if the user doesn't change his filter lists selection for
2014-09-08 23:46:58 +02:00
// some set time.
2014-09-25 21:44:18 +02:00
2015-11-29 23:06:58 +01:00
µBlock.selfieManager = (function() {
Refactor selfie generation into a more flexible persistence mechanism The motivation is to address the higher peak memory usage at launch time with 3rd-gen HNTrie when a selfie was present. The selfie generation prior to this change was to collect all filtering data into a single data structure, and then to serialize that whole structure at once into storage (using JSON.stringify). However, HNTrie serialization requires that a large UintArray32 be converted into a plain JS array, which itslef would be indirectly converted into a JSON string. This was the main reason why peak memory usage would be higher at launch from selfie, since the JSON string would need to be wholly unserialized into JS objects, which themselves would need to be converted into more specialized data structures (like that Uint32Array one). The solution to lower peak memory usage at launch is to refactor selfie generation to allow a more piecemeal approach: each filtering component is given the ability to serialize itself rather than to be forced to be embedded in the master selfie. With this approach, the HNTrie buffer can now serialize to its own storage by converting the buffer data directly into a string which can be directly sent to storage. This avoiding expensive intermediate steps such as converting into a JS array and then to a JSON string. As part of the refactoring, there was also opportunistic code upgrade to ES6 and Promise (eventually all of uBO's code will be proper ES6). Additionally, the polyfill to bring getBytesInUse() to Firefox has been revisited to replace the rather expensive previous implementation with an implementation with virtually no overhead.
2019-02-14 19:33:55 +01:00
const µb = µBlock;
let timer;
2014-09-08 23:46:58 +02:00
// As of 2018-05-31:
Refactor selfie generation into a more flexible persistence mechanism The motivation is to address the higher peak memory usage at launch time with 3rd-gen HNTrie when a selfie was present. The selfie generation prior to this change was to collect all filtering data into a single data structure, and then to serialize that whole structure at once into storage (using JSON.stringify). However, HNTrie serialization requires that a large UintArray32 be converted into a plain JS array, which itslef would be indirectly converted into a JSON string. This was the main reason why peak memory usage would be higher at launch from selfie, since the JSON string would need to be wholly unserialized into JS objects, which themselves would need to be converted into more specialized data structures (like that Uint32Array one). The solution to lower peak memory usage at launch is to refactor selfie generation to allow a more piecemeal approach: each filtering component is given the ability to serialize itself rather than to be forced to be embedded in the master selfie. With this approach, the HNTrie buffer can now serialize to its own storage by converting the buffer data directly into a string which can be directly sent to storage. This avoiding expensive intermediate steps such as converting into a JS array and then to a JSON string. As part of the refactoring, there was also opportunistic code upgrade to ES6 and Promise (eventually all of uBO's code will be proper ES6). Additionally, the polyfill to bring getBytesInUse() to Firefox has been revisited to replace the rather expensive previous implementation with an implementation with virtually no overhead.
2019-02-14 19:33:55 +01:00
// JSON.stringify-ing ourselves results in a better baseline
// memory usage at selfie-load time. For some reasons.
const create = function() {
Promise.all([
µb.assets.put(
'selfie/main',
JSON.stringify({
magic: µb.systemSettings.selfieMagic,
availableFilterLists: µb.availableFilterLists,
})
),
µb.redirectEngine.toSelfie('selfie/redirectEngine'),
µb.staticExtFilteringEngine.toSelfie('selfie/staticExtFilteringEngine'),
µb.staticNetFilteringEngine.toSelfie('selfie/staticNetFilteringEngine'),
]).then(( ) => {
µb.lz4Codec.relinquish();
2018-08-06 18:34:41 +02:00
});
};
2015-11-29 23:06:58 +01:00
Refactor selfie generation into a more flexible persistence mechanism The motivation is to address the higher peak memory usage at launch time with 3rd-gen HNTrie when a selfie was present. The selfie generation prior to this change was to collect all filtering data into a single data structure, and then to serialize that whole structure at once into storage (using JSON.stringify). However, HNTrie serialization requires that a large UintArray32 be converted into a plain JS array, which itslef would be indirectly converted into a JSON string. This was the main reason why peak memory usage would be higher at launch from selfie, since the JSON string would need to be wholly unserialized into JS objects, which themselves would need to be converted into more specialized data structures (like that Uint32Array one). The solution to lower peak memory usage at launch is to refactor selfie generation to allow a more piecemeal approach: each filtering component is given the ability to serialize itself rather than to be forced to be embedded in the master selfie. With this approach, the HNTrie buffer can now serialize to its own storage by converting the buffer data directly into a string which can be directly sent to storage. This avoiding expensive intermediate steps such as converting into a JS array and then to a JSON string. As part of the refactoring, there was also opportunistic code upgrade to ES6 and Promise (eventually all of uBO's code will be proper ES6). Additionally, the polyfill to bring getBytesInUse() to Firefox has been revisited to replace the rather expensive previous implementation with an implementation with virtually no overhead.
2019-02-14 19:33:55 +01:00
const load = function() {
return Promise.all([
µb.assets.get('selfie/main').then(details => {
if (
details instanceof Object === false ||
typeof details.content !== 'string' ||
details.content === ''
) {
return false;
}
let selfie;
try {
selfie = JSON.parse(details.content);
} catch(ex) {
}
if (
selfie instanceof Object === false ||
selfie.magic !== µb.systemSettings.selfieMagic
) {
return false;
}
µb.availableFilterLists = selfie.availableFilterLists;
return true;
}),
µb.redirectEngine.fromSelfie('selfie/redirectEngine'),
µb.staticExtFilteringEngine.fromSelfie('selfie/staticExtFilteringEngine'),
µb.staticNetFilteringEngine.fromSelfie('selfie/staticNetFilteringEngine'),
]).then(results =>
results.reduce((acc, v) => acc && v, true)
).catch(reason => {
log.info(reason);
return false;
});
};
Refactor selfie generation into a more flexible persistence mechanism The motivation is to address the higher peak memory usage at launch time with 3rd-gen HNTrie when a selfie was present. The selfie generation prior to this change was to collect all filtering data into a single data structure, and then to serialize that whole structure at once into storage (using JSON.stringify). However, HNTrie serialization requires that a large UintArray32 be converted into a plain JS array, which itslef would be indirectly converted into a JSON string. This was the main reason why peak memory usage would be higher at launch from selfie, since the JSON string would need to be wholly unserialized into JS objects, which themselves would need to be converted into more specialized data structures (like that Uint32Array one). The solution to lower peak memory usage at launch is to refactor selfie generation to allow a more piecemeal approach: each filtering component is given the ability to serialize itself rather than to be forced to be embedded in the master selfie. With this approach, the HNTrie buffer can now serialize to its own storage by converting the buffer data directly into a string which can be directly sent to storage. This avoiding expensive intermediate steps such as converting into a JS array and then to a JSON string. As part of the refactoring, there was also opportunistic code upgrade to ES6 and Promise (eventually all of uBO's code will be proper ES6). Additionally, the polyfill to bring getBytesInUse() to Firefox has been revisited to replace the rather expensive previous implementation with an implementation with virtually no overhead.
2019-02-14 19:33:55 +01:00
const destroy = function() {
if ( timer !== undefined ) {
2015-11-29 23:06:58 +01:00
clearTimeout(timer);
Refactor selfie generation into a more flexible persistence mechanism The motivation is to address the higher peak memory usage at launch time with 3rd-gen HNTrie when a selfie was present. The selfie generation prior to this change was to collect all filtering data into a single data structure, and then to serialize that whole structure at once into storage (using JSON.stringify). However, HNTrie serialization requires that a large UintArray32 be converted into a plain JS array, which itslef would be indirectly converted into a JSON string. This was the main reason why peak memory usage would be higher at launch from selfie, since the JSON string would need to be wholly unserialized into JS objects, which themselves would need to be converted into more specialized data structures (like that Uint32Array one). The solution to lower peak memory usage at launch is to refactor selfie generation to allow a more piecemeal approach: each filtering component is given the ability to serialize itself rather than to be forced to be embedded in the master selfie. With this approach, the HNTrie buffer can now serialize to its own storage by converting the buffer data directly into a string which can be directly sent to storage. This avoiding expensive intermediate steps such as converting into a JS array and then to a JSON string. As part of the refactoring, there was also opportunistic code upgrade to ES6 and Promise (eventually all of uBO's code will be proper ES6). Additionally, the polyfill to bring getBytesInUse() to Firefox has been revisited to replace the rather expensive previous implementation with an implementation with virtually no overhead.
2019-02-14 19:33:55 +01:00
timer = undefined;
2015-11-29 23:06:58 +01:00
}
Refactor selfie generation into a more flexible persistence mechanism The motivation is to address the higher peak memory usage at launch time with 3rd-gen HNTrie when a selfie was present. The selfie generation prior to this change was to collect all filtering data into a single data structure, and then to serialize that whole structure at once into storage (using JSON.stringify). However, HNTrie serialization requires that a large UintArray32 be converted into a plain JS array, which itslef would be indirectly converted into a JSON string. This was the main reason why peak memory usage would be higher at launch from selfie, since the JSON string would need to be wholly unserialized into JS objects, which themselves would need to be converted into more specialized data structures (like that Uint32Array one). The solution to lower peak memory usage at launch is to refactor selfie generation to allow a more piecemeal approach: each filtering component is given the ability to serialize itself rather than to be forced to be embedded in the master selfie. With this approach, the HNTrie buffer can now serialize to its own storage by converting the buffer data directly into a string which can be directly sent to storage. This avoiding expensive intermediate steps such as converting into a JS array and then to a JSON string. As part of the refactoring, there was also opportunistic code upgrade to ES6 and Promise (eventually all of uBO's code will be proper ES6). Additionally, the polyfill to bring getBytesInUse() to Firefox has been revisited to replace the rather expensive previous implementation with an implementation with virtually no overhead.
2019-02-14 19:33:55 +01:00
µb.cacheStorage.remove('selfie'); // TODO: obsolete, remove eventually.
µb.assets.remove(/^selfie\//);
timer = vAPI.setTimeout(( ) => {
timer = undefined;
create();
}, µb.hiddenSettings.selfieAfter * 60000);
};
2015-11-29 23:06:58 +01:00
return {
load: load,
2015-11-29 23:06:58 +01:00
destroy: destroy
};
})();
2014-09-08 23:46:58 +02:00
/******************************************************************************/
2015-07-27 16:10:34 +02:00
// https://github.com/gorhill/uBlock/issues/531
// Overwrite user settings with admin settings if present.
//
// Admin settings match layout of a uBlock backup. Not all data is
// necessarily present, i.e. administrators may removed entries which
// values are left to the user's choice.
µBlock.restoreAdminSettings = function() {
return new Promise(resolve => {
// >>>> start of executor
if ( vAPI.adminStorage instanceof Object === false ) {
return resolve();
}
vAPI.adminStorage.getItem('adminSettings', json => {
let data;
2015-10-21 17:53:03 +02:00
if ( typeof json === 'string' && json !== '' ) {
try {
data = JSON.parse(json);
} catch (ex) {
console.error(ex);
}
2015-07-27 16:10:34 +02:00
}
if ( data instanceof Object === false ) {
return resolve();
2015-10-21 17:53:03 +02:00
}
2015-07-27 16:10:34 +02:00
const bin = {};
let binNotEmpty = false;
2015-07-27 16:10:34 +02:00
// Allows an admin to set their own 'assets.json' file, with their own
// set of stock assets.
if ( typeof data.assetsBootstrapLocation === 'string' ) {
bin.assetsBootstrapLocation = data.assetsBootstrapLocation;
binNotEmpty = true;
}
2015-10-21 17:53:03 +02:00
if ( typeof data.userSettings === 'object' ) {
for ( const name in this.userSettings ) {
if ( this.userSettings.hasOwnProperty(name) === false ) {
2015-10-21 17:53:03 +02:00
continue;
}
if ( data.userSettings.hasOwnProperty(name) === false ) {
continue;
}
bin[name] = data.userSettings[name];
binNotEmpty = true;
2015-07-27 16:10:34 +02:00
}
2015-10-21 17:53:03 +02:00
}
// 'selectedFilterLists' is an array of filter list tokens. Each token
// is a reference to an asset in 'assets.json'.
if ( Array.isArray(data.selectedFilterLists) ) {
bin.selectedFilterLists = data.selectedFilterLists;
binNotEmpty = true;
2015-07-27 16:10:34 +02:00
}
2015-10-21 17:53:03 +02:00
if ( typeof data.netWhitelist === 'string' ) {
bin.netWhitelist = data.netWhitelist;
binNotEmpty = true;
}
2015-07-27 16:10:34 +02:00
2015-10-21 17:53:03 +02:00
if ( typeof data.dynamicFilteringString === 'string' ) {
bin.dynamicFilteringString = data.dynamicFilteringString;
binNotEmpty = true;
}
2015-07-27 16:10:34 +02:00
2015-10-21 17:53:03 +02:00
if ( typeof data.urlFilteringString === 'string' ) {
bin.urlFilteringString = data.urlFilteringString;
binNotEmpty = true;
}
2015-07-27 16:10:34 +02:00
2015-10-21 17:53:03 +02:00
if ( typeof data.hostnameSwitchesString === 'string' ) {
bin.hostnameSwitchesString = data.hostnameSwitchesString;
binNotEmpty = true;
}
2015-07-27 16:10:34 +02:00
2015-10-21 17:53:03 +02:00
if ( binNotEmpty ) {
vAPI.storage.set(bin);
}
2015-07-27 16:10:34 +02:00
2015-10-21 17:53:03 +02:00
if ( typeof data.userFilters === 'string' ) {
this.assets.put(this.userFiltersPath, data.userFilters);
2015-10-21 17:53:03 +02:00
}
2015-07-27 16:10:34 +02:00
resolve();
});
2015-10-21 17:53:03 +02:00
// <<<< end of executor
});
2015-07-27 16:10:34 +02:00
};
/******************************************************************************/
2017-11-09 18:53:05 +01:00
// https://github.com/gorhill/uBlock/issues/2344
// Support mutliple locales per filter list.
// https://github.com/gorhill/uBlock/issues/3210
// Support ability to auto-enable a filter list based on user agent.
µBlock.listMatchesEnvironment = function(details) {
var re;
// Matches language?
if ( typeof details.lang === 'string' ) {
re = this.listMatchesEnvironment.reLang;
if ( re === undefined ) {
2017-11-09 18:53:05 +01:00
re = new RegExp('\\b' + self.navigator.language.slice(0, 2) + '\\b');
this.listMatchesEnvironment.reLang = re;
}
if ( re.test(details.lang) ) { return true; }
}
// Matches user agent?
if ( typeof details.ua === 'string' ) {
re = new RegExp('\\b' + this.escapeRegex(details.ua) + '\\b', 'i');
if ( re.test(self.navigator.userAgent) ) { return true; }
}
return false;
};
/******************************************************************************/
µBlock.scheduleAssetUpdater = (function() {
var timer, next = 0;
return function(updateDelay) {
if ( timer ) {
clearTimeout(timer);
timer = undefined;
2014-09-08 23:46:58 +02:00
}
if ( updateDelay === 0 ) {
next = 0;
2015-02-24 00:31:29 +01:00
return;
}
var now = Date.now();
// Use the new schedule if and only if it is earlier than the previous
// one.
if ( next !== 0 ) {
updateDelay = Math.min(updateDelay, Math.max(next - now, 0));
2015-02-24 00:31:29 +01:00
}
next = now + updateDelay;
timer = vAPI.setTimeout(function() {
timer = undefined;
next = 0;
var µb = µBlock;
µb.assets.updateStart({
delay: µb.hiddenSettings.autoUpdateAssetFetchPeriod * 1000 || 120000
});
}, updateDelay);
2015-02-24 00:31:29 +01:00
};
})();
/******************************************************************************/
µBlock.assetObserver = function(topic, details) {
// Do not update filter list if not in use.
if ( topic === 'before-asset-updated' ) {
2017-05-08 20:00:41 +02:00
if ( details.type === 'filters' ) {
if (
this.availableFilterLists.hasOwnProperty(details.assetKey) === false ||
this.selectedFilterLists.indexOf(details.assetKey) === -1
) {
return;
2017-05-08 20:00:41 +02:00
}
}
// https://github.com/gorhill/uBlock/issues/2594
if ( details.assetKey === 'ublock-resources' ) {
if (
this.hiddenSettings.ignoreRedirectFilters === true &&
this.hiddenSettings.ignoreScriptInjectFilters === true
) {
return;
2017-05-08 20:00:41 +02:00
}
}
return true;
}
// Compile the list while we have the raw version in memory
if ( topic === 'after-asset-updated' ) {
Refactor selfie generation into a more flexible persistence mechanism The motivation is to address the higher peak memory usage at launch time with 3rd-gen HNTrie when a selfie was present. The selfie generation prior to this change was to collect all filtering data into a single data structure, and then to serialize that whole structure at once into storage (using JSON.stringify). However, HNTrie serialization requires that a large UintArray32 be converted into a plain JS array, which itslef would be indirectly converted into a JSON string. This was the main reason why peak memory usage would be higher at launch from selfie, since the JSON string would need to be wholly unserialized into JS objects, which themselves would need to be converted into more specialized data structures (like that Uint32Array one). The solution to lower peak memory usage at launch is to refactor selfie generation to allow a more piecemeal approach: each filtering component is given the ability to serialize itself rather than to be forced to be embedded in the master selfie. With this approach, the HNTrie buffer can now serialize to its own storage by converting the buffer data directly into a string which can be directly sent to storage. This avoiding expensive intermediate steps such as converting into a JS array and then to a JSON string. As part of the refactoring, there was also opportunistic code upgrade to ES6 and Promise (eventually all of uBO's code will be proper ES6). Additionally, the polyfill to bring getBytesInUse() to Firefox has been revisited to replace the rather expensive previous implementation with an implementation with virtually no overhead.
2019-02-14 19:33:55 +01:00
// Skip selfie-related content.
if ( details.assetKey.startsWith('selfie/') ) { return; }
var cached = typeof details.content === 'string' && details.content !== '';
if ( this.availableFilterLists.hasOwnProperty(details.assetKey) ) {
if ( cached ) {
2017-01-22 22:05:16 +01:00
if ( this.selectedFilterLists.indexOf(details.assetKey) !== -1 ) {
this.extractFilterListMetadata(
details.assetKey,
details.content
);
this.assets.put(
'compiled/' + details.assetKey,
this.compileFilters(
details.content,
{ assetKey: details.assetKey }
)
);
}
} else {
this.removeCompiledFilterList(details.assetKey);
2015-09-13 16:26:36 +02:00
}
} else if ( details.assetKey === this.pslAssetKey ) {
if ( cached ) {
this.compilePublicSuffixList(details.content);
}
} else if ( details.assetKey === 'ublock-resources' ) {
2018-02-15 23:25:38 +01:00
this.redirectEngine.invalidateResourcesSelfie();
if ( cached ) {
2018-02-15 23:25:38 +01:00
this.loadRedirectResources(details.content);
}
}
vAPI.messaging.broadcast({
what: 'assetUpdated',
key: details.assetKey,
cached: cached
});
2017-05-06 19:19:05 +02:00
// https://github.com/gorhill/uBlock/issues/2585
Refactor selfie generation into a more flexible persistence mechanism The motivation is to address the higher peak memory usage at launch time with 3rd-gen HNTrie when a selfie was present. The selfie generation prior to this change was to collect all filtering data into a single data structure, and then to serialize that whole structure at once into storage (using JSON.stringify). However, HNTrie serialization requires that a large UintArray32 be converted into a plain JS array, which itslef would be indirectly converted into a JSON string. This was the main reason why peak memory usage would be higher at launch from selfie, since the JSON string would need to be wholly unserialized into JS objects, which themselves would need to be converted into more specialized data structures (like that Uint32Array one). The solution to lower peak memory usage at launch is to refactor selfie generation to allow a more piecemeal approach: each filtering component is given the ability to serialize itself rather than to be forced to be embedded in the master selfie. With this approach, the HNTrie buffer can now serialize to its own storage by converting the buffer data directly into a string which can be directly sent to storage. This avoiding expensive intermediate steps such as converting into a JS array and then to a JSON string. As part of the refactoring, there was also opportunistic code upgrade to ES6 and Promise (eventually all of uBO's code will be proper ES6). Additionally, the polyfill to bring getBytesInUse() to Firefox has been revisited to replace the rather expensive previous implementation with an implementation with virtually no overhead.
2019-02-14 19:33:55 +01:00
// Whenever an asset is overwritten, the current selfie is quite
// likely no longer valid.
2017-05-06 19:19:05 +02:00
this.selfieManager.destroy();
return;
}
2017-01-22 22:05:16 +01:00
// Update failed.
if ( topic === 'asset-update-failed' ) {
vAPI.messaging.broadcast({
what: 'assetUpdated',
key: details.assetKey,
failed: true
});
return;
}
// Reload all filter lists if needed.
if ( topic === 'after-assets-updated' ) {
if ( details.assetKeys.length !== 0 ) {
this.loadFilterLists();
}
if ( this.userSettings.autoUpdate ) {
this.scheduleAssetUpdater(this.hiddenSettings.autoUpdatePeriod * 3600000 || 25200000);
} else {
this.scheduleAssetUpdater(0);
}
vAPI.messaging.broadcast({
what: 'assetsUpdated',
assetKeys: details.assetKeys
});
return;
}
// New asset source became available, if it's a filter list, should we
// auto-select it?
if ( topic === 'builtin-asset-source-added' ) {
if ( details.entry.content === 'filters' ) {
if (
details.entry.off !== true ||
2017-11-09 18:53:05 +01:00
this.listMatchesEnvironment(details.entry)
) {
this.saveSelectedFilterLists([ details.assetKey ], true);
}
}
return;
}
};