1
0
mirror of https://github.com/gorhill/uBlock.git synced 2024-09-29 22:27:12 +02:00
uBlock/src/js/utils.js

742 lines
24 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
*/
'use strict';
2014-06-24 00:42:43 +02:00
/******************************************************************************/
// A standalone URL tokenizer will allow us to use URL tokens in more than
// just static filtering engine. This opens the door to optimize other
// filtering engine parts aside static filtering. This also allows:
// - Tokenize only on demand.
// - To potentially avoid tokenizing when same URL is fed to tokenizer.
// - Benchmarking shows this to be a common occurrence.
2017-05-19 14:45:19 +02:00
//
// https://github.com/gorhill/uBlock/issues/2630
2017-05-20 02:22:26 +02:00
// Slice input URL into a list of safe-integer token values, instead of a list
2017-05-19 14:45:19 +02:00
// of substrings. The assumption is that with dealing only with numeric
// values, less underlying memory allocations, and also as a consequence
// less work for the garbage collector down the road.
// Another assumption is that using a numeric-based key value for Map() is
// more efficient than string-based key value (but that is something I would
// have to benchmark).
2017-05-20 02:22:26 +02:00
// Benchmark for string-based tokens vs. safe-integer token values:
2017-05-19 14:45:19 +02:00
// https://gorhill.github.io/obj-vs-set-vs-map/tokenize-to-str-vs-to-int.html
Add HNTrie-based filter classes to store origin-only filters Related issue: - https://github.com/uBlockOrigin/uBlock-issues/issues/528#issuecomment-484408622 Following STrie-related work in above issue, I noticed that a large number of filters in EasyList were filters which only had to match against the document origin. For instance, among just the top 10 most populous buckets, there were four such buckets with over hundreds of entries each: - bits: 72, token: "http", 146 entries - bits: 72, token: "https", 139 entries - bits: 88, token: "http", 122 entries - bits: 88, token: "https", 118 entries These filters in these buckets have to be matched against all the network requests. In order to leverage HNTrie for these filters[1], they are now handled in a special way so as to ensure they all end up in a single HNTrie (per bucket), which means that instead of scanning hundreds of entries per URL, there is now a single scan per bucket per URL for these apply-everywhere filters. Now, any filter which fulfill ALL the following condition will be processed in a special manner internally: - Is of the form `|https://` or `|http://` or `*`; and - Does have a `domain=` option; and - Does not have a negated domain in its `domain=` option; and - Does not have `csp=` option; and - Does not have a `redirect=` option If a filter does not fulfill ALL the conditions above, no change in behavior. A filter which matches ALL of the above will be processed in a special manner: - The `domain=` option will be decomposed so as to create as many distinct filter as there is distinct value in the `domain=` option - This also apply to the `badfilter` version of the filter, which means it now become possible to `badfilter` only one of the distinct filter without having to `badfilter` all of them. - The logger will always report these special filters with only a single hostname in the `domain=` option. *** [1] HNTrie is currently WASM-ed on Firefox.
2019-04-19 22:33:46 +02:00
µBlock.urlTokenizer = new (class {
constructor() {
this._chars = '0123456789%abcdefghijklmnopqrstuvwxyz';
this._validTokenChars = new Uint8Array(128);
for ( let i = 0, n = this._chars.length; i < n; i++ ) {
this._validTokenChars[this._chars.charCodeAt(i)] = i + 1;
}
// Four upper bits of token hash are reserved for built-in predefined
// token hashes, which should never end up being used when tokenizing
// any arbitrary string.
this.dotTokenHash = 0x10000000;
this.anyTokenHash = 0x20000000;
this.anyHTTPSTokenHash = 0x30000000;
this.anyHTTPTokenHash = 0x40000000;
this.noTokenHash = 0x50000000;
this.emptyTokenHash = 0xF0000000;
Add HNTrie-based filter classes to store origin-only filters Related issue: - https://github.com/uBlockOrigin/uBlock-issues/issues/528#issuecomment-484408622 Following STrie-related work in above issue, I noticed that a large number of filters in EasyList were filters which only had to match against the document origin. For instance, among just the top 10 most populous buckets, there were four such buckets with over hundreds of entries each: - bits: 72, token: "http", 146 entries - bits: 72, token: "https", 139 entries - bits: 88, token: "http", 122 entries - bits: 88, token: "https", 118 entries These filters in these buckets have to be matched against all the network requests. In order to leverage HNTrie for these filters[1], they are now handled in a special way so as to ensure they all end up in a single HNTrie (per bucket), which means that instead of scanning hundreds of entries per URL, there is now a single scan per bucket per URL for these apply-everywhere filters. Now, any filter which fulfill ALL the following condition will be processed in a special manner internally: - Is of the form `|https://` or `|http://` or `*`; and - Does have a `domain=` option; and - Does not have a negated domain in its `domain=` option; and - Does not have `csp=` option; and - Does not have a `redirect=` option If a filter does not fulfill ALL the conditions above, no change in behavior. A filter which matches ALL of the above will be processed in a special manner: - The `domain=` option will be decomposed so as to create as many distinct filter as there is distinct value in the `domain=` option - This also apply to the `badfilter` version of the filter, which means it now become possible to `badfilter` only one of the distinct filter without having to `badfilter` all of them. - The logger will always report these special filters with only a single hostname in the `domain=` option. *** [1] HNTrie is currently WASM-ed on Firefox.
2019-04-19 22:33:46 +02:00
this._urlIn = '';
this._urlOut = '';
this._tokenized = false;
Expand bidi-trie usage in static network filtering engine Related issues: - https://github.com/uBlockOrigin/uBlock-issues/issues/761 - https://github.com/uBlockOrigin/uBlock-issues/issues/528 The previous bidi-trie code could only hold filters which are plain pattern, i.e. no wildcard characters, and which had no origin option (`domain=`), right and/or left anchor, and no `csp=` option. Example of filters that could be moved into a bidi-trie data structure: &ad_box_ /w/d/capu.php?z=$script,third-party ||liveonlinetv247.com/images/muvixx-150x50-watch-now-in-hd-play-btn.gif Examples of filters that could NOT be moved to a bidi-trie: -adap.$domain=~l-adap.org /tsc.php?*&ses= ||ibsrv.net/*forumsponsor$domain=[...] @@||imgspice.com/jquery.cookie.js|$script ||view.atdmt.com^*/iview/$third-party ||postimg.cc/image/$csp=[...] Ideally the filters above should be able to be moved to a bidi-trie since they are basically plain patterns, or at least partially moved to a bidi-trie when there is only a single wildcard (i.e. made of two plain patterns). Also, there were two distinct bidi-tries in which plain-pattern filters can be moved to: one for patterns without hostname anchoring and another one for patterns with hostname-anchoring. This was required because the hostname-anchored patterns have an extra condition which is outside the bidi-trie knowledge. This commit expands the number of filters which can be stored in the bidi-trie, and also remove the need to use two distinct bidi-tries. - Added ability to associate a pattern with an integer in the bidi-trie [1]. - The bidi-trie match code passes this externally provided integer when calling an externally provided method used for testing extra conditions that may be present for a plain pattern found to be matching in the bidi-trie. - Decomposed existing filters into smaller logical units: - FilterPlainLeftAnchored => FilterPatternPlain + FilterAnchorLeft - FilterPlainRightAnchored => FilterPatternPlain + FilterAnchorRight - FilterExactMatch => FilterPatternPlain + FilterAnchorLeft + FilterAnchorRight - FilterPlainHnAnchored => FilterPatternPlain + FilterAnchorHn - FilterWildcard1 => FilterPatternPlain + [ FilterPatternLeft or FilterPatternRight ] - FilterWildcard1HnAnchored => FilterPatternPlain + [ FilterPatternLeft or FilterPatternRight ] + FilterAnchorHn - FilterGenericHnAnchored => FilterPatternGeneric + FilterAnchorHn - FilterGenericHnAndRightAnchored => FilterPatternGeneric + FilterAnchorRight + FilterAnchorHn - FilterOriginMixedSet => FilterOriginMissSet + FilterOriginHitSet - Instances of FilterOrigin[...], FilterDataHolder can also be added to a composite filter to represent `domain=` and `csp=` options. - Added a new filter class, FilterComposite, for filters which are a combination of two or more logical units. A FilterComposite instance is a match when *all* filters composing it are a match. Since filters are now encoded into combination of smaller units, it becomes possible to extract the FilterPatternPlain component and store it in the bidi-trie, and use the integer as a handle for the remaining extra conditions, if any. Since a single pattern in the bidi-trie may be a component for different filters, the associated integer points to a sequence of extra conditions, and a match occurs as soon as one of the extra conditions (which may itself be a sequence of conditions) is fulfilled. Decomposing filters which are currently single instance into sequences of smaller logical filters means increasing the storage and CPU overhead when evaluating such filters. The CPU overhead is compensated by the fact that more filters can now moved into the bidi-trie, where the first match is efficiently evaluated. The extra conditions have to be evaluated if and only if there is a match in the bidi-trie. The storage overhead is compensated by the bidi-trie's intrinsic nature of merging similar patterns. Furthermore, the storage overhead is reduced by no longer using JavaScript array to store collection of filters (which is what FilterComposite is): the same technique used in [2] is imported to store sequences of filters. A sequence of filters is a sequence of integer pairs where the first integer is an index to an actual filter instance stored in a global array of filters (`filterUnits`), while the second integer is an index to the next pair in the sequence -- which means all sequences of filters are encoded in one single array of integers (`filterSequences` => Uint32Array). As a result, a sequence of filters can be represented by one single integer -- an index to the first pair -- regardless of the number of filters in the sequence. This representation is further leveraged to replace the use of JavaScript array in FilterBucket [3], which used a JavaScript array to store collection of filters. Doing so means there is no more need for FilterPair [4], which purpose was to be a lightweight representation when there was only two filters in a collection. As a result of the above changes, the map of `token` (integer) => filter instance (object) used to associate tokens to filters or collections of filters is replaced with a more efficient map of `token` (integer) to filter unit index (integer) to lookup a filter object from the global `filterUnits` array. Another consequence of using one single global array to store all filter instances means we can reuse existing instances when a logical filter instance is parameter-less, which is the case for FilterAnchorLeft, FilterAnchorRight, FilterAnchorHn, the index to these single instances is reused where needed. `urlTokenizer` now stores the character codes of the scanned URL into a bidi-trie buffer, for reuse when string matching methods are called. New method: `tokenHistogram()`, used to generate histograms of occurrences of token extracted from URLs in built-in benchmark. The top results of the "miss" histogram are used as "bad tokens", i.e. tokens to avoid if possible when compiling filter lists. All plain pattern strings are now stored in the bidi-trie memory buffer, regardless of whether they will be used in the trie proper or not. Three methods have been added to the bidi-trie to test stored string against the URL which is also stored in then bidi-trie. FilterParser is now instanciated on demand and released when no longer used. *** [1] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/strie.js#L120 [2] https://github.com/gorhill/uBlock/commit/e94024d350b066e4e04a772b0a3dbc69daab3fb7 [3] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/static-net-filtering.js#L1630 [4] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/static-net-filtering.js#L1566
2019-10-21 14:15:58 +02:00
this._tokens = new Uint32Array(1024);
this.knownTokens = new Uint8Array(65536);
this.resetKnownTokens();
Expand bidi-trie usage in static network filtering engine Related issues: - https://github.com/uBlockOrigin/uBlock-issues/issues/761 - https://github.com/uBlockOrigin/uBlock-issues/issues/528 The previous bidi-trie code could only hold filters which are plain pattern, i.e. no wildcard characters, and which had no origin option (`domain=`), right and/or left anchor, and no `csp=` option. Example of filters that could be moved into a bidi-trie data structure: &ad_box_ /w/d/capu.php?z=$script,third-party ||liveonlinetv247.com/images/muvixx-150x50-watch-now-in-hd-play-btn.gif Examples of filters that could NOT be moved to a bidi-trie: -adap.$domain=~l-adap.org /tsc.php?*&ses= ||ibsrv.net/*forumsponsor$domain=[...] @@||imgspice.com/jquery.cookie.js|$script ||view.atdmt.com^*/iview/$third-party ||postimg.cc/image/$csp=[...] Ideally the filters above should be able to be moved to a bidi-trie since they are basically plain patterns, or at least partially moved to a bidi-trie when there is only a single wildcard (i.e. made of two plain patterns). Also, there were two distinct bidi-tries in which plain-pattern filters can be moved to: one for patterns without hostname anchoring and another one for patterns with hostname-anchoring. This was required because the hostname-anchored patterns have an extra condition which is outside the bidi-trie knowledge. This commit expands the number of filters which can be stored in the bidi-trie, and also remove the need to use two distinct bidi-tries. - Added ability to associate a pattern with an integer in the bidi-trie [1]. - The bidi-trie match code passes this externally provided integer when calling an externally provided method used for testing extra conditions that may be present for a plain pattern found to be matching in the bidi-trie. - Decomposed existing filters into smaller logical units: - FilterPlainLeftAnchored => FilterPatternPlain + FilterAnchorLeft - FilterPlainRightAnchored => FilterPatternPlain + FilterAnchorRight - FilterExactMatch => FilterPatternPlain + FilterAnchorLeft + FilterAnchorRight - FilterPlainHnAnchored => FilterPatternPlain + FilterAnchorHn - FilterWildcard1 => FilterPatternPlain + [ FilterPatternLeft or FilterPatternRight ] - FilterWildcard1HnAnchored => FilterPatternPlain + [ FilterPatternLeft or FilterPatternRight ] + FilterAnchorHn - FilterGenericHnAnchored => FilterPatternGeneric + FilterAnchorHn - FilterGenericHnAndRightAnchored => FilterPatternGeneric + FilterAnchorRight + FilterAnchorHn - FilterOriginMixedSet => FilterOriginMissSet + FilterOriginHitSet - Instances of FilterOrigin[...], FilterDataHolder can also be added to a composite filter to represent `domain=` and `csp=` options. - Added a new filter class, FilterComposite, for filters which are a combination of two or more logical units. A FilterComposite instance is a match when *all* filters composing it are a match. Since filters are now encoded into combination of smaller units, it becomes possible to extract the FilterPatternPlain component and store it in the bidi-trie, and use the integer as a handle for the remaining extra conditions, if any. Since a single pattern in the bidi-trie may be a component for different filters, the associated integer points to a sequence of extra conditions, and a match occurs as soon as one of the extra conditions (which may itself be a sequence of conditions) is fulfilled. Decomposing filters which are currently single instance into sequences of smaller logical filters means increasing the storage and CPU overhead when evaluating such filters. The CPU overhead is compensated by the fact that more filters can now moved into the bidi-trie, where the first match is efficiently evaluated. The extra conditions have to be evaluated if and only if there is a match in the bidi-trie. The storage overhead is compensated by the bidi-trie's intrinsic nature of merging similar patterns. Furthermore, the storage overhead is reduced by no longer using JavaScript array to store collection of filters (which is what FilterComposite is): the same technique used in [2] is imported to store sequences of filters. A sequence of filters is a sequence of integer pairs where the first integer is an index to an actual filter instance stored in a global array of filters (`filterUnits`), while the second integer is an index to the next pair in the sequence -- which means all sequences of filters are encoded in one single array of integers (`filterSequences` => Uint32Array). As a result, a sequence of filters can be represented by one single integer -- an index to the first pair -- regardless of the number of filters in the sequence. This representation is further leveraged to replace the use of JavaScript array in FilterBucket [3], which used a JavaScript array to store collection of filters. Doing so means there is no more need for FilterPair [4], which purpose was to be a lightweight representation when there was only two filters in a collection. As a result of the above changes, the map of `token` (integer) => filter instance (object) used to associate tokens to filters or collections of filters is replaced with a more efficient map of `token` (integer) to filter unit index (integer) to lookup a filter object from the global `filterUnits` array. Another consequence of using one single global array to store all filter instances means we can reuse existing instances when a logical filter instance is parameter-less, which is the case for FilterAnchorLeft, FilterAnchorRight, FilterAnchorHn, the index to these single instances is reused where needed. `urlTokenizer` now stores the character codes of the scanned URL into a bidi-trie buffer, for reuse when string matching methods are called. New method: `tokenHistogram()`, used to generate histograms of occurrences of token extracted from URLs in built-in benchmark. The top results of the "miss" histogram are used as "bad tokens", i.e. tokens to avoid if possible when compiling filter lists. All plain pattern strings are now stored in the bidi-trie memory buffer, regardless of whether they will be used in the trie proper or not. Three methods have been added to the bidi-trie to test stored string against the URL which is also stored in then bidi-trie. FilterParser is now instanciated on demand and released when no longer used. *** [1] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/strie.js#L120 [2] https://github.com/gorhill/uBlock/commit/e94024d350b066e4e04a772b0a3dbc69daab3fb7 [3] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/static-net-filtering.js#L1630 [4] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/static-net-filtering.js#L1566
2019-10-21 14:15:58 +02:00
this.MAX_TOKEN_LENGTH = 7;
this.charCodes = new Uint8Array(2048);
this.charCodeCount = 0;
Add HNTrie-based filter classes to store origin-only filters Related issue: - https://github.com/uBlockOrigin/uBlock-issues/issues/528#issuecomment-484408622 Following STrie-related work in above issue, I noticed that a large number of filters in EasyList were filters which only had to match against the document origin. For instance, among just the top 10 most populous buckets, there were four such buckets with over hundreds of entries each: - bits: 72, token: "http", 146 entries - bits: 72, token: "https", 139 entries - bits: 88, token: "http", 122 entries - bits: 88, token: "https", 118 entries These filters in these buckets have to be matched against all the network requests. In order to leverage HNTrie for these filters[1], they are now handled in a special way so as to ensure they all end up in a single HNTrie (per bucket), which means that instead of scanning hundreds of entries per URL, there is now a single scan per bucket per URL for these apply-everywhere filters. Now, any filter which fulfill ALL the following condition will be processed in a special manner internally: - Is of the form `|https://` or `|http://` or `*`; and - Does have a `domain=` option; and - Does not have a negated domain in its `domain=` option; and - Does not have `csp=` option; and - Does not have a `redirect=` option If a filter does not fulfill ALL the conditions above, no change in behavior. A filter which matches ALL of the above will be processed in a special manner: - The `domain=` option will be decomposed so as to create as many distinct filter as there is distinct value in the `domain=` option - This also apply to the `badfilter` version of the filter, which means it now become possible to `badfilter` only one of the distinct filter without having to `badfilter` all of them. - The logger will always report these special filters with only a single hostname in the `domain=` option. *** [1] HNTrie is currently WASM-ed on Firefox.
2019-04-19 22:33:46 +02:00
}
setURL(url) {
if ( url !== this._urlIn ) {
this._urlIn = url;
this._urlOut = url.toLowerCase();
this._tokenized = false;
}
return this._urlOut;
Add HNTrie-based filter classes to store origin-only filters Related issue: - https://github.com/uBlockOrigin/uBlock-issues/issues/528#issuecomment-484408622 Following STrie-related work in above issue, I noticed that a large number of filters in EasyList were filters which only had to match against the document origin. For instance, among just the top 10 most populous buckets, there were four such buckets with over hundreds of entries each: - bits: 72, token: "http", 146 entries - bits: 72, token: "https", 139 entries - bits: 88, token: "http", 122 entries - bits: 88, token: "https", 118 entries These filters in these buckets have to be matched against all the network requests. In order to leverage HNTrie for these filters[1], they are now handled in a special way so as to ensure they all end up in a single HNTrie (per bucket), which means that instead of scanning hundreds of entries per URL, there is now a single scan per bucket per URL for these apply-everywhere filters. Now, any filter which fulfill ALL the following condition will be processed in a special manner internally: - Is of the form `|https://` or `|http://` or `*`; and - Does have a `domain=` option; and - Does not have a negated domain in its `domain=` option; and - Does not have `csp=` option; and - Does not have a `redirect=` option If a filter does not fulfill ALL the conditions above, no change in behavior. A filter which matches ALL of the above will be processed in a special manner: - The `domain=` option will be decomposed so as to create as many distinct filter as there is distinct value in the `domain=` option - This also apply to the `badfilter` version of the filter, which means it now become possible to `badfilter` only one of the distinct filter without having to `badfilter` all of them. - The logger will always report these special filters with only a single hostname in the `domain=` option. *** [1] HNTrie is currently WASM-ed on Firefox.
2019-04-19 22:33:46 +02:00
}
resetKnownTokens() {
this.knownTokens.fill(0);
this.addKnownToken(this.dotTokenHash);
this.addKnownToken(this.anyTokenHash);
this.addKnownToken(this.anyHTTPSTokenHash);
this.addKnownToken(this.anyHTTPTokenHash);
this.addKnownToken(this.noTokenHash);
}
addKnownToken(th) {
this.knownTokens[th & 0xFFFF ^ th >>> 16] = 1;
}
// Tokenize on demand.
Expand bidi-trie usage in static network filtering engine Related issues: - https://github.com/uBlockOrigin/uBlock-issues/issues/761 - https://github.com/uBlockOrigin/uBlock-issues/issues/528 The previous bidi-trie code could only hold filters which are plain pattern, i.e. no wildcard characters, and which had no origin option (`domain=`), right and/or left anchor, and no `csp=` option. Example of filters that could be moved into a bidi-trie data structure: &ad_box_ /w/d/capu.php?z=$script,third-party ||liveonlinetv247.com/images/muvixx-150x50-watch-now-in-hd-play-btn.gif Examples of filters that could NOT be moved to a bidi-trie: -adap.$domain=~l-adap.org /tsc.php?*&ses= ||ibsrv.net/*forumsponsor$domain=[...] @@||imgspice.com/jquery.cookie.js|$script ||view.atdmt.com^*/iview/$third-party ||postimg.cc/image/$csp=[...] Ideally the filters above should be able to be moved to a bidi-trie since they are basically plain patterns, or at least partially moved to a bidi-trie when there is only a single wildcard (i.e. made of two plain patterns). Also, there were two distinct bidi-tries in which plain-pattern filters can be moved to: one for patterns without hostname anchoring and another one for patterns with hostname-anchoring. This was required because the hostname-anchored patterns have an extra condition which is outside the bidi-trie knowledge. This commit expands the number of filters which can be stored in the bidi-trie, and also remove the need to use two distinct bidi-tries. - Added ability to associate a pattern with an integer in the bidi-trie [1]. - The bidi-trie match code passes this externally provided integer when calling an externally provided method used for testing extra conditions that may be present for a plain pattern found to be matching in the bidi-trie. - Decomposed existing filters into smaller logical units: - FilterPlainLeftAnchored => FilterPatternPlain + FilterAnchorLeft - FilterPlainRightAnchored => FilterPatternPlain + FilterAnchorRight - FilterExactMatch => FilterPatternPlain + FilterAnchorLeft + FilterAnchorRight - FilterPlainHnAnchored => FilterPatternPlain + FilterAnchorHn - FilterWildcard1 => FilterPatternPlain + [ FilterPatternLeft or FilterPatternRight ] - FilterWildcard1HnAnchored => FilterPatternPlain + [ FilterPatternLeft or FilterPatternRight ] + FilterAnchorHn - FilterGenericHnAnchored => FilterPatternGeneric + FilterAnchorHn - FilterGenericHnAndRightAnchored => FilterPatternGeneric + FilterAnchorRight + FilterAnchorHn - FilterOriginMixedSet => FilterOriginMissSet + FilterOriginHitSet - Instances of FilterOrigin[...], FilterDataHolder can also be added to a composite filter to represent `domain=` and `csp=` options. - Added a new filter class, FilterComposite, for filters which are a combination of two or more logical units. A FilterComposite instance is a match when *all* filters composing it are a match. Since filters are now encoded into combination of smaller units, it becomes possible to extract the FilterPatternPlain component and store it in the bidi-trie, and use the integer as a handle for the remaining extra conditions, if any. Since a single pattern in the bidi-trie may be a component for different filters, the associated integer points to a sequence of extra conditions, and a match occurs as soon as one of the extra conditions (which may itself be a sequence of conditions) is fulfilled. Decomposing filters which are currently single instance into sequences of smaller logical filters means increasing the storage and CPU overhead when evaluating such filters. The CPU overhead is compensated by the fact that more filters can now moved into the bidi-trie, where the first match is efficiently evaluated. The extra conditions have to be evaluated if and only if there is a match in the bidi-trie. The storage overhead is compensated by the bidi-trie's intrinsic nature of merging similar patterns. Furthermore, the storage overhead is reduced by no longer using JavaScript array to store collection of filters (which is what FilterComposite is): the same technique used in [2] is imported to store sequences of filters. A sequence of filters is a sequence of integer pairs where the first integer is an index to an actual filter instance stored in a global array of filters (`filterUnits`), while the second integer is an index to the next pair in the sequence -- which means all sequences of filters are encoded in one single array of integers (`filterSequences` => Uint32Array). As a result, a sequence of filters can be represented by one single integer -- an index to the first pair -- regardless of the number of filters in the sequence. This representation is further leveraged to replace the use of JavaScript array in FilterBucket [3], which used a JavaScript array to store collection of filters. Doing so means there is no more need for FilterPair [4], which purpose was to be a lightweight representation when there was only two filters in a collection. As a result of the above changes, the map of `token` (integer) => filter instance (object) used to associate tokens to filters or collections of filters is replaced with a more efficient map of `token` (integer) to filter unit index (integer) to lookup a filter object from the global `filterUnits` array. Another consequence of using one single global array to store all filter instances means we can reuse existing instances when a logical filter instance is parameter-less, which is the case for FilterAnchorLeft, FilterAnchorRight, FilterAnchorHn, the index to these single instances is reused where needed. `urlTokenizer` now stores the character codes of the scanned URL into a bidi-trie buffer, for reuse when string matching methods are called. New method: `tokenHistogram()`, used to generate histograms of occurrences of token extracted from URLs in built-in benchmark. The top results of the "miss" histogram are used as "bad tokens", i.e. tokens to avoid if possible when compiling filter lists. All plain pattern strings are now stored in the bidi-trie memory buffer, regardless of whether they will be used in the trie proper or not. Three methods have been added to the bidi-trie to test stored string against the URL which is also stored in then bidi-trie. FilterParser is now instanciated on demand and released when no longer used. *** [1] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/strie.js#L120 [2] https://github.com/gorhill/uBlock/commit/e94024d350b066e4e04a772b0a3dbc69daab3fb7 [3] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/static-net-filtering.js#L1630 [4] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/static-net-filtering.js#L1566
2019-10-21 14:15:58 +02:00
getTokens(encodeInto) {
Add HNTrie-based filter classes to store origin-only filters Related issue: - https://github.com/uBlockOrigin/uBlock-issues/issues/528#issuecomment-484408622 Following STrie-related work in above issue, I noticed that a large number of filters in EasyList were filters which only had to match against the document origin. For instance, among just the top 10 most populous buckets, there were four such buckets with over hundreds of entries each: - bits: 72, token: "http", 146 entries - bits: 72, token: "https", 139 entries - bits: 88, token: "http", 122 entries - bits: 88, token: "https", 118 entries These filters in these buckets have to be matched against all the network requests. In order to leverage HNTrie for these filters[1], they are now handled in a special way so as to ensure they all end up in a single HNTrie (per bucket), which means that instead of scanning hundreds of entries per URL, there is now a single scan per bucket per URL for these apply-everywhere filters. Now, any filter which fulfill ALL the following condition will be processed in a special manner internally: - Is of the form `|https://` or `|http://` or `*`; and - Does have a `domain=` option; and - Does not have a negated domain in its `domain=` option; and - Does not have `csp=` option; and - Does not have a `redirect=` option If a filter does not fulfill ALL the conditions above, no change in behavior. A filter which matches ALL of the above will be processed in a special manner: - The `domain=` option will be decomposed so as to create as many distinct filter as there is distinct value in the `domain=` option - This also apply to the `badfilter` version of the filter, which means it now become possible to `badfilter` only one of the distinct filter without having to `badfilter` all of them. - The logger will always report these special filters with only a single hostname in the `domain=` option. *** [1] HNTrie is currently WASM-ed on Firefox.
2019-04-19 22:33:46 +02:00
if ( this._tokenized ) { return this._tokens; }
Expand bidi-trie usage in static network filtering engine Related issues: - https://github.com/uBlockOrigin/uBlock-issues/issues/761 - https://github.com/uBlockOrigin/uBlock-issues/issues/528 The previous bidi-trie code could only hold filters which are plain pattern, i.e. no wildcard characters, and which had no origin option (`domain=`), right and/or left anchor, and no `csp=` option. Example of filters that could be moved into a bidi-trie data structure: &ad_box_ /w/d/capu.php?z=$script,third-party ||liveonlinetv247.com/images/muvixx-150x50-watch-now-in-hd-play-btn.gif Examples of filters that could NOT be moved to a bidi-trie: -adap.$domain=~l-adap.org /tsc.php?*&ses= ||ibsrv.net/*forumsponsor$domain=[...] @@||imgspice.com/jquery.cookie.js|$script ||view.atdmt.com^*/iview/$third-party ||postimg.cc/image/$csp=[...] Ideally the filters above should be able to be moved to a bidi-trie since they are basically plain patterns, or at least partially moved to a bidi-trie when there is only a single wildcard (i.e. made of two plain patterns). Also, there were two distinct bidi-tries in which plain-pattern filters can be moved to: one for patterns without hostname anchoring and another one for patterns with hostname-anchoring. This was required because the hostname-anchored patterns have an extra condition which is outside the bidi-trie knowledge. This commit expands the number of filters which can be stored in the bidi-trie, and also remove the need to use two distinct bidi-tries. - Added ability to associate a pattern with an integer in the bidi-trie [1]. - The bidi-trie match code passes this externally provided integer when calling an externally provided method used for testing extra conditions that may be present for a plain pattern found to be matching in the bidi-trie. - Decomposed existing filters into smaller logical units: - FilterPlainLeftAnchored => FilterPatternPlain + FilterAnchorLeft - FilterPlainRightAnchored => FilterPatternPlain + FilterAnchorRight - FilterExactMatch => FilterPatternPlain + FilterAnchorLeft + FilterAnchorRight - FilterPlainHnAnchored => FilterPatternPlain + FilterAnchorHn - FilterWildcard1 => FilterPatternPlain + [ FilterPatternLeft or FilterPatternRight ] - FilterWildcard1HnAnchored => FilterPatternPlain + [ FilterPatternLeft or FilterPatternRight ] + FilterAnchorHn - FilterGenericHnAnchored => FilterPatternGeneric + FilterAnchorHn - FilterGenericHnAndRightAnchored => FilterPatternGeneric + FilterAnchorRight + FilterAnchorHn - FilterOriginMixedSet => FilterOriginMissSet + FilterOriginHitSet - Instances of FilterOrigin[...], FilterDataHolder can also be added to a composite filter to represent `domain=` and `csp=` options. - Added a new filter class, FilterComposite, for filters which are a combination of two or more logical units. A FilterComposite instance is a match when *all* filters composing it are a match. Since filters are now encoded into combination of smaller units, it becomes possible to extract the FilterPatternPlain component and store it in the bidi-trie, and use the integer as a handle for the remaining extra conditions, if any. Since a single pattern in the bidi-trie may be a component for different filters, the associated integer points to a sequence of extra conditions, and a match occurs as soon as one of the extra conditions (which may itself be a sequence of conditions) is fulfilled. Decomposing filters which are currently single instance into sequences of smaller logical filters means increasing the storage and CPU overhead when evaluating such filters. The CPU overhead is compensated by the fact that more filters can now moved into the bidi-trie, where the first match is efficiently evaluated. The extra conditions have to be evaluated if and only if there is a match in the bidi-trie. The storage overhead is compensated by the bidi-trie's intrinsic nature of merging similar patterns. Furthermore, the storage overhead is reduced by no longer using JavaScript array to store collection of filters (which is what FilterComposite is): the same technique used in [2] is imported to store sequences of filters. A sequence of filters is a sequence of integer pairs where the first integer is an index to an actual filter instance stored in a global array of filters (`filterUnits`), while the second integer is an index to the next pair in the sequence -- which means all sequences of filters are encoded in one single array of integers (`filterSequences` => Uint32Array). As a result, a sequence of filters can be represented by one single integer -- an index to the first pair -- regardless of the number of filters in the sequence. This representation is further leveraged to replace the use of JavaScript array in FilterBucket [3], which used a JavaScript array to store collection of filters. Doing so means there is no more need for FilterPair [4], which purpose was to be a lightweight representation when there was only two filters in a collection. As a result of the above changes, the map of `token` (integer) => filter instance (object) used to associate tokens to filters or collections of filters is replaced with a more efficient map of `token` (integer) to filter unit index (integer) to lookup a filter object from the global `filterUnits` array. Another consequence of using one single global array to store all filter instances means we can reuse existing instances when a logical filter instance is parameter-less, which is the case for FilterAnchorLeft, FilterAnchorRight, FilterAnchorHn, the index to these single instances is reused where needed. `urlTokenizer` now stores the character codes of the scanned URL into a bidi-trie buffer, for reuse when string matching methods are called. New method: `tokenHistogram()`, used to generate histograms of occurrences of token extracted from URLs in built-in benchmark. The top results of the "miss" histogram are used as "bad tokens", i.e. tokens to avoid if possible when compiling filter lists. All plain pattern strings are now stored in the bidi-trie memory buffer, regardless of whether they will be used in the trie proper or not. Three methods have been added to the bidi-trie to test stored string against the URL which is also stored in then bidi-trie. FilterParser is now instanciated on demand and released when no longer used. *** [1] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/strie.js#L120 [2] https://github.com/gorhill/uBlock/commit/e94024d350b066e4e04a772b0a3dbc69daab3fb7 [3] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/static-net-filtering.js#L1630 [4] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/static-net-filtering.js#L1566
2019-10-21 14:15:58 +02:00
let i = this._tokenize(encodeInto);
this._tokens[i+0] = this.anyTokenHash;
this._tokens[i+1] = 0;
i += 2;
Add HNTrie-based filter classes to store origin-only filters Related issue: - https://github.com/uBlockOrigin/uBlock-issues/issues/528#issuecomment-484408622 Following STrie-related work in above issue, I noticed that a large number of filters in EasyList were filters which only had to match against the document origin. For instance, among just the top 10 most populous buckets, there were four such buckets with over hundreds of entries each: - bits: 72, token: "http", 146 entries - bits: 72, token: "https", 139 entries - bits: 88, token: "http", 122 entries - bits: 88, token: "https", 118 entries These filters in these buckets have to be matched against all the network requests. In order to leverage HNTrie for these filters[1], they are now handled in a special way so as to ensure they all end up in a single HNTrie (per bucket), which means that instead of scanning hundreds of entries per URL, there is now a single scan per bucket per URL for these apply-everywhere filters. Now, any filter which fulfill ALL the following condition will be processed in a special manner internally: - Is of the form `|https://` or `|http://` or `*`; and - Does have a `domain=` option; and - Does not have a negated domain in its `domain=` option; and - Does not have `csp=` option; and - Does not have a `redirect=` option If a filter does not fulfill ALL the conditions above, no change in behavior. A filter which matches ALL of the above will be processed in a special manner: - The `domain=` option will be decomposed so as to create as many distinct filter as there is distinct value in the `domain=` option - This also apply to the `badfilter` version of the filter, which means it now become possible to `badfilter` only one of the distinct filter without having to `badfilter` all of them. - The logger will always report these special filters with only a single hostname in the `domain=` option. *** [1] HNTrie is currently WASM-ed on Firefox.
2019-04-19 22:33:46 +02:00
if ( this._urlOut.startsWith('https://') ) {
Expand bidi-trie usage in static network filtering engine Related issues: - https://github.com/uBlockOrigin/uBlock-issues/issues/761 - https://github.com/uBlockOrigin/uBlock-issues/issues/528 The previous bidi-trie code could only hold filters which are plain pattern, i.e. no wildcard characters, and which had no origin option (`domain=`), right and/or left anchor, and no `csp=` option. Example of filters that could be moved into a bidi-trie data structure: &ad_box_ /w/d/capu.php?z=$script,third-party ||liveonlinetv247.com/images/muvixx-150x50-watch-now-in-hd-play-btn.gif Examples of filters that could NOT be moved to a bidi-trie: -adap.$domain=~l-adap.org /tsc.php?*&ses= ||ibsrv.net/*forumsponsor$domain=[...] @@||imgspice.com/jquery.cookie.js|$script ||view.atdmt.com^*/iview/$third-party ||postimg.cc/image/$csp=[...] Ideally the filters above should be able to be moved to a bidi-trie since they are basically plain patterns, or at least partially moved to a bidi-trie when there is only a single wildcard (i.e. made of two plain patterns). Also, there were two distinct bidi-tries in which plain-pattern filters can be moved to: one for patterns without hostname anchoring and another one for patterns with hostname-anchoring. This was required because the hostname-anchored patterns have an extra condition which is outside the bidi-trie knowledge. This commit expands the number of filters which can be stored in the bidi-trie, and also remove the need to use two distinct bidi-tries. - Added ability to associate a pattern with an integer in the bidi-trie [1]. - The bidi-trie match code passes this externally provided integer when calling an externally provided method used for testing extra conditions that may be present for a plain pattern found to be matching in the bidi-trie. - Decomposed existing filters into smaller logical units: - FilterPlainLeftAnchored => FilterPatternPlain + FilterAnchorLeft - FilterPlainRightAnchored => FilterPatternPlain + FilterAnchorRight - FilterExactMatch => FilterPatternPlain + FilterAnchorLeft + FilterAnchorRight - FilterPlainHnAnchored => FilterPatternPlain + FilterAnchorHn - FilterWildcard1 => FilterPatternPlain + [ FilterPatternLeft or FilterPatternRight ] - FilterWildcard1HnAnchored => FilterPatternPlain + [ FilterPatternLeft or FilterPatternRight ] + FilterAnchorHn - FilterGenericHnAnchored => FilterPatternGeneric + FilterAnchorHn - FilterGenericHnAndRightAnchored => FilterPatternGeneric + FilterAnchorRight + FilterAnchorHn - FilterOriginMixedSet => FilterOriginMissSet + FilterOriginHitSet - Instances of FilterOrigin[...], FilterDataHolder can also be added to a composite filter to represent `domain=` and `csp=` options. - Added a new filter class, FilterComposite, for filters which are a combination of two or more logical units. A FilterComposite instance is a match when *all* filters composing it are a match. Since filters are now encoded into combination of smaller units, it becomes possible to extract the FilterPatternPlain component and store it in the bidi-trie, and use the integer as a handle for the remaining extra conditions, if any. Since a single pattern in the bidi-trie may be a component for different filters, the associated integer points to a sequence of extra conditions, and a match occurs as soon as one of the extra conditions (which may itself be a sequence of conditions) is fulfilled. Decomposing filters which are currently single instance into sequences of smaller logical filters means increasing the storage and CPU overhead when evaluating such filters. The CPU overhead is compensated by the fact that more filters can now moved into the bidi-trie, where the first match is efficiently evaluated. The extra conditions have to be evaluated if and only if there is a match in the bidi-trie. The storage overhead is compensated by the bidi-trie's intrinsic nature of merging similar patterns. Furthermore, the storage overhead is reduced by no longer using JavaScript array to store collection of filters (which is what FilterComposite is): the same technique used in [2] is imported to store sequences of filters. A sequence of filters is a sequence of integer pairs where the first integer is an index to an actual filter instance stored in a global array of filters (`filterUnits`), while the second integer is an index to the next pair in the sequence -- which means all sequences of filters are encoded in one single array of integers (`filterSequences` => Uint32Array). As a result, a sequence of filters can be represented by one single integer -- an index to the first pair -- regardless of the number of filters in the sequence. This representation is further leveraged to replace the use of JavaScript array in FilterBucket [3], which used a JavaScript array to store collection of filters. Doing so means there is no more need for FilterPair [4], which purpose was to be a lightweight representation when there was only two filters in a collection. As a result of the above changes, the map of `token` (integer) => filter instance (object) used to associate tokens to filters or collections of filters is replaced with a more efficient map of `token` (integer) to filter unit index (integer) to lookup a filter object from the global `filterUnits` array. Another consequence of using one single global array to store all filter instances means we can reuse existing instances when a logical filter instance is parameter-less, which is the case for FilterAnchorLeft, FilterAnchorRight, FilterAnchorHn, the index to these single instances is reused where needed. `urlTokenizer` now stores the character codes of the scanned URL into a bidi-trie buffer, for reuse when string matching methods are called. New method: `tokenHistogram()`, used to generate histograms of occurrences of token extracted from URLs in built-in benchmark. The top results of the "miss" histogram are used as "bad tokens", i.e. tokens to avoid if possible when compiling filter lists. All plain pattern strings are now stored in the bidi-trie memory buffer, regardless of whether they will be used in the trie proper or not. Three methods have been added to the bidi-trie to test stored string against the URL which is also stored in then bidi-trie. FilterParser is now instanciated on demand and released when no longer used. *** [1] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/strie.js#L120 [2] https://github.com/gorhill/uBlock/commit/e94024d350b066e4e04a772b0a3dbc69daab3fb7 [3] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/static-net-filtering.js#L1630 [4] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/static-net-filtering.js#L1566
2019-10-21 14:15:58 +02:00
this._tokens[i+0] = this.anyHTTPSTokenHash;
this._tokens[i+1] = 0;
i += 2;
Add HNTrie-based filter classes to store origin-only filters Related issue: - https://github.com/uBlockOrigin/uBlock-issues/issues/528#issuecomment-484408622 Following STrie-related work in above issue, I noticed that a large number of filters in EasyList were filters which only had to match against the document origin. For instance, among just the top 10 most populous buckets, there were four such buckets with over hundreds of entries each: - bits: 72, token: "http", 146 entries - bits: 72, token: "https", 139 entries - bits: 88, token: "http", 122 entries - bits: 88, token: "https", 118 entries These filters in these buckets have to be matched against all the network requests. In order to leverage HNTrie for these filters[1], they are now handled in a special way so as to ensure they all end up in a single HNTrie (per bucket), which means that instead of scanning hundreds of entries per URL, there is now a single scan per bucket per URL for these apply-everywhere filters. Now, any filter which fulfill ALL the following condition will be processed in a special manner internally: - Is of the form `|https://` or `|http://` or `*`; and - Does have a `domain=` option; and - Does not have a negated domain in its `domain=` option; and - Does not have `csp=` option; and - Does not have a `redirect=` option If a filter does not fulfill ALL the conditions above, no change in behavior. A filter which matches ALL of the above will be processed in a special manner: - The `domain=` option will be decomposed so as to create as many distinct filter as there is distinct value in the `domain=` option - This also apply to the `badfilter` version of the filter, which means it now become possible to `badfilter` only one of the distinct filter without having to `badfilter` all of them. - The logger will always report these special filters with only a single hostname in the `domain=` option. *** [1] HNTrie is currently WASM-ed on Firefox.
2019-04-19 22:33:46 +02:00
} else if ( this._urlOut.startsWith('http://') ) {
Expand bidi-trie usage in static network filtering engine Related issues: - https://github.com/uBlockOrigin/uBlock-issues/issues/761 - https://github.com/uBlockOrigin/uBlock-issues/issues/528 The previous bidi-trie code could only hold filters which are plain pattern, i.e. no wildcard characters, and which had no origin option (`domain=`), right and/or left anchor, and no `csp=` option. Example of filters that could be moved into a bidi-trie data structure: &ad_box_ /w/d/capu.php?z=$script,third-party ||liveonlinetv247.com/images/muvixx-150x50-watch-now-in-hd-play-btn.gif Examples of filters that could NOT be moved to a bidi-trie: -adap.$domain=~l-adap.org /tsc.php?*&ses= ||ibsrv.net/*forumsponsor$domain=[...] @@||imgspice.com/jquery.cookie.js|$script ||view.atdmt.com^*/iview/$third-party ||postimg.cc/image/$csp=[...] Ideally the filters above should be able to be moved to a bidi-trie since they are basically plain patterns, or at least partially moved to a bidi-trie when there is only a single wildcard (i.e. made of two plain patterns). Also, there were two distinct bidi-tries in which plain-pattern filters can be moved to: one for patterns without hostname anchoring and another one for patterns with hostname-anchoring. This was required because the hostname-anchored patterns have an extra condition which is outside the bidi-trie knowledge. This commit expands the number of filters which can be stored in the bidi-trie, and also remove the need to use two distinct bidi-tries. - Added ability to associate a pattern with an integer in the bidi-trie [1]. - The bidi-trie match code passes this externally provided integer when calling an externally provided method used for testing extra conditions that may be present for a plain pattern found to be matching in the bidi-trie. - Decomposed existing filters into smaller logical units: - FilterPlainLeftAnchored => FilterPatternPlain + FilterAnchorLeft - FilterPlainRightAnchored => FilterPatternPlain + FilterAnchorRight - FilterExactMatch => FilterPatternPlain + FilterAnchorLeft + FilterAnchorRight - FilterPlainHnAnchored => FilterPatternPlain + FilterAnchorHn - FilterWildcard1 => FilterPatternPlain + [ FilterPatternLeft or FilterPatternRight ] - FilterWildcard1HnAnchored => FilterPatternPlain + [ FilterPatternLeft or FilterPatternRight ] + FilterAnchorHn - FilterGenericHnAnchored => FilterPatternGeneric + FilterAnchorHn - FilterGenericHnAndRightAnchored => FilterPatternGeneric + FilterAnchorRight + FilterAnchorHn - FilterOriginMixedSet => FilterOriginMissSet + FilterOriginHitSet - Instances of FilterOrigin[...], FilterDataHolder can also be added to a composite filter to represent `domain=` and `csp=` options. - Added a new filter class, FilterComposite, for filters which are a combination of two or more logical units. A FilterComposite instance is a match when *all* filters composing it are a match. Since filters are now encoded into combination of smaller units, it becomes possible to extract the FilterPatternPlain component and store it in the bidi-trie, and use the integer as a handle for the remaining extra conditions, if any. Since a single pattern in the bidi-trie may be a component for different filters, the associated integer points to a sequence of extra conditions, and a match occurs as soon as one of the extra conditions (which may itself be a sequence of conditions) is fulfilled. Decomposing filters which are currently single instance into sequences of smaller logical filters means increasing the storage and CPU overhead when evaluating such filters. The CPU overhead is compensated by the fact that more filters can now moved into the bidi-trie, where the first match is efficiently evaluated. The extra conditions have to be evaluated if and only if there is a match in the bidi-trie. The storage overhead is compensated by the bidi-trie's intrinsic nature of merging similar patterns. Furthermore, the storage overhead is reduced by no longer using JavaScript array to store collection of filters (which is what FilterComposite is): the same technique used in [2] is imported to store sequences of filters. A sequence of filters is a sequence of integer pairs where the first integer is an index to an actual filter instance stored in a global array of filters (`filterUnits`), while the second integer is an index to the next pair in the sequence -- which means all sequences of filters are encoded in one single array of integers (`filterSequences` => Uint32Array). As a result, a sequence of filters can be represented by one single integer -- an index to the first pair -- regardless of the number of filters in the sequence. This representation is further leveraged to replace the use of JavaScript array in FilterBucket [3], which used a JavaScript array to store collection of filters. Doing so means there is no more need for FilterPair [4], which purpose was to be a lightweight representation when there was only two filters in a collection. As a result of the above changes, the map of `token` (integer) => filter instance (object) used to associate tokens to filters or collections of filters is replaced with a more efficient map of `token` (integer) to filter unit index (integer) to lookup a filter object from the global `filterUnits` array. Another consequence of using one single global array to store all filter instances means we can reuse existing instances when a logical filter instance is parameter-less, which is the case for FilterAnchorLeft, FilterAnchorRight, FilterAnchorHn, the index to these single instances is reused where needed. `urlTokenizer` now stores the character codes of the scanned URL into a bidi-trie buffer, for reuse when string matching methods are called. New method: `tokenHistogram()`, used to generate histograms of occurrences of token extracted from URLs in built-in benchmark. The top results of the "miss" histogram are used as "bad tokens", i.e. tokens to avoid if possible when compiling filter lists. All plain pattern strings are now stored in the bidi-trie memory buffer, regardless of whether they will be used in the trie proper or not. Three methods have been added to the bidi-trie to test stored string against the URL which is also stored in then bidi-trie. FilterParser is now instanciated on demand and released when no longer used. *** [1] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/strie.js#L120 [2] https://github.com/gorhill/uBlock/commit/e94024d350b066e4e04a772b0a3dbc69daab3fb7 [3] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/static-net-filtering.js#L1630 [4] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/static-net-filtering.js#L1566
2019-10-21 14:15:58 +02:00
this._tokens[i+0] = this.anyHTTPTokenHash;
this._tokens[i+1] = 0;
i += 2;
}
Expand bidi-trie usage in static network filtering engine Related issues: - https://github.com/uBlockOrigin/uBlock-issues/issues/761 - https://github.com/uBlockOrigin/uBlock-issues/issues/528 The previous bidi-trie code could only hold filters which are plain pattern, i.e. no wildcard characters, and which had no origin option (`domain=`), right and/or left anchor, and no `csp=` option. Example of filters that could be moved into a bidi-trie data structure: &ad_box_ /w/d/capu.php?z=$script,third-party ||liveonlinetv247.com/images/muvixx-150x50-watch-now-in-hd-play-btn.gif Examples of filters that could NOT be moved to a bidi-trie: -adap.$domain=~l-adap.org /tsc.php?*&ses= ||ibsrv.net/*forumsponsor$domain=[...] @@||imgspice.com/jquery.cookie.js|$script ||view.atdmt.com^*/iview/$third-party ||postimg.cc/image/$csp=[...] Ideally the filters above should be able to be moved to a bidi-trie since they are basically plain patterns, or at least partially moved to a bidi-trie when there is only a single wildcard (i.e. made of two plain patterns). Also, there were two distinct bidi-tries in which plain-pattern filters can be moved to: one for patterns without hostname anchoring and another one for patterns with hostname-anchoring. This was required because the hostname-anchored patterns have an extra condition which is outside the bidi-trie knowledge. This commit expands the number of filters which can be stored in the bidi-trie, and also remove the need to use two distinct bidi-tries. - Added ability to associate a pattern with an integer in the bidi-trie [1]. - The bidi-trie match code passes this externally provided integer when calling an externally provided method used for testing extra conditions that may be present for a plain pattern found to be matching in the bidi-trie. - Decomposed existing filters into smaller logical units: - FilterPlainLeftAnchored => FilterPatternPlain + FilterAnchorLeft - FilterPlainRightAnchored => FilterPatternPlain + FilterAnchorRight - FilterExactMatch => FilterPatternPlain + FilterAnchorLeft + FilterAnchorRight - FilterPlainHnAnchored => FilterPatternPlain + FilterAnchorHn - FilterWildcard1 => FilterPatternPlain + [ FilterPatternLeft or FilterPatternRight ] - FilterWildcard1HnAnchored => FilterPatternPlain + [ FilterPatternLeft or FilterPatternRight ] + FilterAnchorHn - FilterGenericHnAnchored => FilterPatternGeneric + FilterAnchorHn - FilterGenericHnAndRightAnchored => FilterPatternGeneric + FilterAnchorRight + FilterAnchorHn - FilterOriginMixedSet => FilterOriginMissSet + FilterOriginHitSet - Instances of FilterOrigin[...], FilterDataHolder can also be added to a composite filter to represent `domain=` and `csp=` options. - Added a new filter class, FilterComposite, for filters which are a combination of two or more logical units. A FilterComposite instance is a match when *all* filters composing it are a match. Since filters are now encoded into combination of smaller units, it becomes possible to extract the FilterPatternPlain component and store it in the bidi-trie, and use the integer as a handle for the remaining extra conditions, if any. Since a single pattern in the bidi-trie may be a component for different filters, the associated integer points to a sequence of extra conditions, and a match occurs as soon as one of the extra conditions (which may itself be a sequence of conditions) is fulfilled. Decomposing filters which are currently single instance into sequences of smaller logical filters means increasing the storage and CPU overhead when evaluating such filters. The CPU overhead is compensated by the fact that more filters can now moved into the bidi-trie, where the first match is efficiently evaluated. The extra conditions have to be evaluated if and only if there is a match in the bidi-trie. The storage overhead is compensated by the bidi-trie's intrinsic nature of merging similar patterns. Furthermore, the storage overhead is reduced by no longer using JavaScript array to store collection of filters (which is what FilterComposite is): the same technique used in [2] is imported to store sequences of filters. A sequence of filters is a sequence of integer pairs where the first integer is an index to an actual filter instance stored in a global array of filters (`filterUnits`), while the second integer is an index to the next pair in the sequence -- which means all sequences of filters are encoded in one single array of integers (`filterSequences` => Uint32Array). As a result, a sequence of filters can be represented by one single integer -- an index to the first pair -- regardless of the number of filters in the sequence. This representation is further leveraged to replace the use of JavaScript array in FilterBucket [3], which used a JavaScript array to store collection of filters. Doing so means there is no more need for FilterPair [4], which purpose was to be a lightweight representation when there was only two filters in a collection. As a result of the above changes, the map of `token` (integer) => filter instance (object) used to associate tokens to filters or collections of filters is replaced with a more efficient map of `token` (integer) to filter unit index (integer) to lookup a filter object from the global `filterUnits` array. Another consequence of using one single global array to store all filter instances means we can reuse existing instances when a logical filter instance is parameter-less, which is the case for FilterAnchorLeft, FilterAnchorRight, FilterAnchorHn, the index to these single instances is reused where needed. `urlTokenizer` now stores the character codes of the scanned URL into a bidi-trie buffer, for reuse when string matching methods are called. New method: `tokenHistogram()`, used to generate histograms of occurrences of token extracted from URLs in built-in benchmark. The top results of the "miss" histogram are used as "bad tokens", i.e. tokens to avoid if possible when compiling filter lists. All plain pattern strings are now stored in the bidi-trie memory buffer, regardless of whether they will be used in the trie proper or not. Three methods have been added to the bidi-trie to test stored string against the URL which is also stored in then bidi-trie. FilterParser is now instanciated on demand and released when no longer used. *** [1] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/strie.js#L120 [2] https://github.com/gorhill/uBlock/commit/e94024d350b066e4e04a772b0a3dbc69daab3fb7 [3] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/static-net-filtering.js#L1630 [4] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/static-net-filtering.js#L1566
2019-10-21 14:15:58 +02:00
this._tokens[i+0] = this.noTokenHash;
this._tokens[i+1] = 0;
this._tokens[i+2] = 0;
Add HNTrie-based filter classes to store origin-only filters Related issue: - https://github.com/uBlockOrigin/uBlock-issues/issues/528#issuecomment-484408622 Following STrie-related work in above issue, I noticed that a large number of filters in EasyList were filters which only had to match against the document origin. For instance, among just the top 10 most populous buckets, there were four such buckets with over hundreds of entries each: - bits: 72, token: "http", 146 entries - bits: 72, token: "https", 139 entries - bits: 88, token: "http", 122 entries - bits: 88, token: "https", 118 entries These filters in these buckets have to be matched against all the network requests. In order to leverage HNTrie for these filters[1], they are now handled in a special way so as to ensure they all end up in a single HNTrie (per bucket), which means that instead of scanning hundreds of entries per URL, there is now a single scan per bucket per URL for these apply-everywhere filters. Now, any filter which fulfill ALL the following condition will be processed in a special manner internally: - Is of the form `|https://` or `|http://` or `*`; and - Does have a `domain=` option; and - Does not have a negated domain in its `domain=` option; and - Does not have `csp=` option; and - Does not have a `redirect=` option If a filter does not fulfill ALL the conditions above, no change in behavior. A filter which matches ALL of the above will be processed in a special manner: - The `domain=` option will be decomposed so as to create as many distinct filter as there is distinct value in the `domain=` option - This also apply to the `badfilter` version of the filter, which means it now become possible to `badfilter` only one of the distinct filter without having to `badfilter` all of them. - The logger will always report these special filters with only a single hostname in the `domain=` option. *** [1] HNTrie is currently WASM-ed on Firefox.
2019-04-19 22:33:46 +02:00
this._tokenized = true;
return this._tokens;
Add HNTrie-based filter classes to store origin-only filters Related issue: - https://github.com/uBlockOrigin/uBlock-issues/issues/528#issuecomment-484408622 Following STrie-related work in above issue, I noticed that a large number of filters in EasyList were filters which only had to match against the document origin. For instance, among just the top 10 most populous buckets, there were four such buckets with over hundreds of entries each: - bits: 72, token: "http", 146 entries - bits: 72, token: "https", 139 entries - bits: 88, token: "http", 122 entries - bits: 88, token: "https", 118 entries These filters in these buckets have to be matched against all the network requests. In order to leverage HNTrie for these filters[1], they are now handled in a special way so as to ensure they all end up in a single HNTrie (per bucket), which means that instead of scanning hundreds of entries per URL, there is now a single scan per bucket per URL for these apply-everywhere filters. Now, any filter which fulfill ALL the following condition will be processed in a special manner internally: - Is of the form `|https://` or `|http://` or `*`; and - Does have a `domain=` option; and - Does not have a negated domain in its `domain=` option; and - Does not have `csp=` option; and - Does not have a `redirect=` option If a filter does not fulfill ALL the conditions above, no change in behavior. A filter which matches ALL of the above will be processed in a special manner: - The `domain=` option will be decomposed so as to create as many distinct filter as there is distinct value in the `domain=` option - This also apply to the `badfilter` version of the filter, which means it now become possible to `badfilter` only one of the distinct filter without having to `badfilter` all of them. - The logger will always report these special filters with only a single hostname in the `domain=` option. *** [1] HNTrie is currently WASM-ed on Firefox.
2019-04-19 22:33:46 +02:00
}
Add HNTrie-based filter classes to store origin-only filters Related issue: - https://github.com/uBlockOrigin/uBlock-issues/issues/528#issuecomment-484408622 Following STrie-related work in above issue, I noticed that a large number of filters in EasyList were filters which only had to match against the document origin. For instance, among just the top 10 most populous buckets, there were four such buckets with over hundreds of entries each: - bits: 72, token: "http", 146 entries - bits: 72, token: "https", 139 entries - bits: 88, token: "http", 122 entries - bits: 88, token: "https", 118 entries These filters in these buckets have to be matched against all the network requests. In order to leverage HNTrie for these filters[1], they are now handled in a special way so as to ensure they all end up in a single HNTrie (per bucket), which means that instead of scanning hundreds of entries per URL, there is now a single scan per bucket per URL for these apply-everywhere filters. Now, any filter which fulfill ALL the following condition will be processed in a special manner internally: - Is of the form `|https://` or `|http://` or `*`; and - Does have a `domain=` option; and - Does not have a negated domain in its `domain=` option; and - Does not have `csp=` option; and - Does not have a `redirect=` option If a filter does not fulfill ALL the conditions above, no change in behavior. A filter which matches ALL of the above will be processed in a special manner: - The `domain=` option will be decomposed so as to create as many distinct filter as there is distinct value in the `domain=` option - This also apply to the `badfilter` version of the filter, which means it now become possible to `badfilter` only one of the distinct filter without having to `badfilter` all of them. - The logger will always report these special filters with only a single hostname in the `domain=` option. *** [1] HNTrie is currently WASM-ed on Firefox.
2019-04-19 22:33:46 +02:00
tokenHashFromString(s) {
const l = s.length;
if ( l === 0 ) { return this.emptyTokenHash; }
const vtc = this._validTokenChars;
Add HNTrie-based filter classes to store origin-only filters Related issue: - https://github.com/uBlockOrigin/uBlock-issues/issues/528#issuecomment-484408622 Following STrie-related work in above issue, I noticed that a large number of filters in EasyList were filters which only had to match against the document origin. For instance, among just the top 10 most populous buckets, there were four such buckets with over hundreds of entries each: - bits: 72, token: "http", 146 entries - bits: 72, token: "https", 139 entries - bits: 88, token: "http", 122 entries - bits: 88, token: "https", 118 entries These filters in these buckets have to be matched against all the network requests. In order to leverage HNTrie for these filters[1], they are now handled in a special way so as to ensure they all end up in a single HNTrie (per bucket), which means that instead of scanning hundreds of entries per URL, there is now a single scan per bucket per URL for these apply-everywhere filters. Now, any filter which fulfill ALL the following condition will be processed in a special manner internally: - Is of the form `|https://` or `|http://` or `*`; and - Does have a `domain=` option; and - Does not have a negated domain in its `domain=` option; and - Does not have `csp=` option; and - Does not have a `redirect=` option If a filter does not fulfill ALL the conditions above, no change in behavior. A filter which matches ALL of the above will be processed in a special manner: - The `domain=` option will be decomposed so as to create as many distinct filter as there is distinct value in the `domain=` option - This also apply to the `badfilter` version of the filter, which means it now become possible to `badfilter` only one of the distinct filter without having to `badfilter` all of them. - The logger will always report these special filters with only a single hostname in the `domain=` option. *** [1] HNTrie is currently WASM-ed on Firefox.
2019-04-19 22:33:46 +02:00
let th = vtc[s.charCodeAt(0)];
for ( let i = 1; i !== 7 && i !== l; i++ ) {
th = th << 4 ^ vtc[s.charCodeAt(i)];
2017-05-19 14:45:19 +02:00
}
return th;
Add HNTrie-based filter classes to store origin-only filters Related issue: - https://github.com/uBlockOrigin/uBlock-issues/issues/528#issuecomment-484408622 Following STrie-related work in above issue, I noticed that a large number of filters in EasyList were filters which only had to match against the document origin. For instance, among just the top 10 most populous buckets, there were four such buckets with over hundreds of entries each: - bits: 72, token: "http", 146 entries - bits: 72, token: "https", 139 entries - bits: 88, token: "http", 122 entries - bits: 88, token: "https", 118 entries These filters in these buckets have to be matched against all the network requests. In order to leverage HNTrie for these filters[1], they are now handled in a special way so as to ensure they all end up in a single HNTrie (per bucket), which means that instead of scanning hundreds of entries per URL, there is now a single scan per bucket per URL for these apply-everywhere filters. Now, any filter which fulfill ALL the following condition will be processed in a special manner internally: - Is of the form `|https://` or `|http://` or `*`; and - Does have a `domain=` option; and - Does not have a negated domain in its `domain=` option; and - Does not have `csp=` option; and - Does not have a `redirect=` option If a filter does not fulfill ALL the conditions above, no change in behavior. A filter which matches ALL of the above will be processed in a special manner: - The `domain=` option will be decomposed so as to create as many distinct filter as there is distinct value in the `domain=` option - This also apply to the `badfilter` version of the filter, which means it now become possible to `badfilter` only one of the distinct filter without having to `badfilter` all of them. - The logger will always report these special filters with only a single hostname in the `domain=` option. *** [1] HNTrie is currently WASM-ed on Firefox.
2019-04-19 22:33:46 +02:00
}
Add HNTrie-based filter classes to store origin-only filters Related issue: - https://github.com/uBlockOrigin/uBlock-issues/issues/528#issuecomment-484408622 Following STrie-related work in above issue, I noticed that a large number of filters in EasyList were filters which only had to match against the document origin. For instance, among just the top 10 most populous buckets, there were four such buckets with over hundreds of entries each: - bits: 72, token: "http", 146 entries - bits: 72, token: "https", 139 entries - bits: 88, token: "http", 122 entries - bits: 88, token: "https", 118 entries These filters in these buckets have to be matched against all the network requests. In order to leverage HNTrie for these filters[1], they are now handled in a special way so as to ensure they all end up in a single HNTrie (per bucket), which means that instead of scanning hundreds of entries per URL, there is now a single scan per bucket per URL for these apply-everywhere filters. Now, any filter which fulfill ALL the following condition will be processed in a special manner internally: - Is of the form `|https://` or `|http://` or `*`; and - Does have a `domain=` option; and - Does not have a negated domain in its `domain=` option; and - Does not have `csp=` option; and - Does not have a `redirect=` option If a filter does not fulfill ALL the conditions above, no change in behavior. A filter which matches ALL of the above will be processed in a special manner: - The `domain=` option will be decomposed so as to create as many distinct filter as there is distinct value in the `domain=` option - This also apply to the `badfilter` version of the filter, which means it now become possible to `badfilter` only one of the distinct filter without having to `badfilter` all of them. - The logger will always report these special filters with only a single hostname in the `domain=` option. *** [1] HNTrie is currently WASM-ed on Firefox.
2019-04-19 22:33:46 +02:00
stringFromTokenHash(th) {
if ( th === 0 ) { return ''; }
return th.toString(16);
Add HNTrie-based filter classes to store origin-only filters Related issue: - https://github.com/uBlockOrigin/uBlock-issues/issues/528#issuecomment-484408622 Following STrie-related work in above issue, I noticed that a large number of filters in EasyList were filters which only had to match against the document origin. For instance, among just the top 10 most populous buckets, there were four such buckets with over hundreds of entries each: - bits: 72, token: "http", 146 entries - bits: 72, token: "https", 139 entries - bits: 88, token: "http", 122 entries - bits: 88, token: "https", 118 entries These filters in these buckets have to be matched against all the network requests. In order to leverage HNTrie for these filters[1], they are now handled in a special way so as to ensure they all end up in a single HNTrie (per bucket), which means that instead of scanning hundreds of entries per URL, there is now a single scan per bucket per URL for these apply-everywhere filters. Now, any filter which fulfill ALL the following condition will be processed in a special manner internally: - Is of the form `|https://` or `|http://` or `*`; and - Does have a `domain=` option; and - Does not have a negated domain in its `domain=` option; and - Does not have `csp=` option; and - Does not have a `redirect=` option If a filter does not fulfill ALL the conditions above, no change in behavior. A filter which matches ALL of the above will be processed in a special manner: - The `domain=` option will be decomposed so as to create as many distinct filter as there is distinct value in the `domain=` option - This also apply to the `badfilter` version of the filter, which means it now become possible to `badfilter` only one of the distinct filter without having to `badfilter` all of them. - The logger will always report these special filters with only a single hostname in the `domain=` option. *** [1] HNTrie is currently WASM-ed on Firefox.
2019-04-19 22:33:46 +02:00
}
toSelfie() {
return µBlock.base64.encode(
this.knownTokens.buffer,
this.knownTokens.byteLength
);
}
fromSelfie(selfie) {
return µBlock.base64.decode(selfie, this.knownTokens.buffer);
}
// https://github.com/chrisaljoudi/uBlock/issues/1118
// We limit to a maximum number of tokens.
Expand bidi-trie usage in static network filtering engine Related issues: - https://github.com/uBlockOrigin/uBlock-issues/issues/761 - https://github.com/uBlockOrigin/uBlock-issues/issues/528 The previous bidi-trie code could only hold filters which are plain pattern, i.e. no wildcard characters, and which had no origin option (`domain=`), right and/or left anchor, and no `csp=` option. Example of filters that could be moved into a bidi-trie data structure: &ad_box_ /w/d/capu.php?z=$script,third-party ||liveonlinetv247.com/images/muvixx-150x50-watch-now-in-hd-play-btn.gif Examples of filters that could NOT be moved to a bidi-trie: -adap.$domain=~l-adap.org /tsc.php?*&ses= ||ibsrv.net/*forumsponsor$domain=[...] @@||imgspice.com/jquery.cookie.js|$script ||view.atdmt.com^*/iview/$third-party ||postimg.cc/image/$csp=[...] Ideally the filters above should be able to be moved to a bidi-trie since they are basically plain patterns, or at least partially moved to a bidi-trie when there is only a single wildcard (i.e. made of two plain patterns). Also, there were two distinct bidi-tries in which plain-pattern filters can be moved to: one for patterns without hostname anchoring and another one for patterns with hostname-anchoring. This was required because the hostname-anchored patterns have an extra condition which is outside the bidi-trie knowledge. This commit expands the number of filters which can be stored in the bidi-trie, and also remove the need to use two distinct bidi-tries. - Added ability to associate a pattern with an integer in the bidi-trie [1]. - The bidi-trie match code passes this externally provided integer when calling an externally provided method used for testing extra conditions that may be present for a plain pattern found to be matching in the bidi-trie. - Decomposed existing filters into smaller logical units: - FilterPlainLeftAnchored => FilterPatternPlain + FilterAnchorLeft - FilterPlainRightAnchored => FilterPatternPlain + FilterAnchorRight - FilterExactMatch => FilterPatternPlain + FilterAnchorLeft + FilterAnchorRight - FilterPlainHnAnchored => FilterPatternPlain + FilterAnchorHn - FilterWildcard1 => FilterPatternPlain + [ FilterPatternLeft or FilterPatternRight ] - FilterWildcard1HnAnchored => FilterPatternPlain + [ FilterPatternLeft or FilterPatternRight ] + FilterAnchorHn - FilterGenericHnAnchored => FilterPatternGeneric + FilterAnchorHn - FilterGenericHnAndRightAnchored => FilterPatternGeneric + FilterAnchorRight + FilterAnchorHn - FilterOriginMixedSet => FilterOriginMissSet + FilterOriginHitSet - Instances of FilterOrigin[...], FilterDataHolder can also be added to a composite filter to represent `domain=` and `csp=` options. - Added a new filter class, FilterComposite, for filters which are a combination of two or more logical units. A FilterComposite instance is a match when *all* filters composing it are a match. Since filters are now encoded into combination of smaller units, it becomes possible to extract the FilterPatternPlain component and store it in the bidi-trie, and use the integer as a handle for the remaining extra conditions, if any. Since a single pattern in the bidi-trie may be a component for different filters, the associated integer points to a sequence of extra conditions, and a match occurs as soon as one of the extra conditions (which may itself be a sequence of conditions) is fulfilled. Decomposing filters which are currently single instance into sequences of smaller logical filters means increasing the storage and CPU overhead when evaluating such filters. The CPU overhead is compensated by the fact that more filters can now moved into the bidi-trie, where the first match is efficiently evaluated. The extra conditions have to be evaluated if and only if there is a match in the bidi-trie. The storage overhead is compensated by the bidi-trie's intrinsic nature of merging similar patterns. Furthermore, the storage overhead is reduced by no longer using JavaScript array to store collection of filters (which is what FilterComposite is): the same technique used in [2] is imported to store sequences of filters. A sequence of filters is a sequence of integer pairs where the first integer is an index to an actual filter instance stored in a global array of filters (`filterUnits`), while the second integer is an index to the next pair in the sequence -- which means all sequences of filters are encoded in one single array of integers (`filterSequences` => Uint32Array). As a result, a sequence of filters can be represented by one single integer -- an index to the first pair -- regardless of the number of filters in the sequence. This representation is further leveraged to replace the use of JavaScript array in FilterBucket [3], which used a JavaScript array to store collection of filters. Doing so means there is no more need for FilterPair [4], which purpose was to be a lightweight representation when there was only two filters in a collection. As a result of the above changes, the map of `token` (integer) => filter instance (object) used to associate tokens to filters or collections of filters is replaced with a more efficient map of `token` (integer) to filter unit index (integer) to lookup a filter object from the global `filterUnits` array. Another consequence of using one single global array to store all filter instances means we can reuse existing instances when a logical filter instance is parameter-less, which is the case for FilterAnchorLeft, FilterAnchorRight, FilterAnchorHn, the index to these single instances is reused where needed. `urlTokenizer` now stores the character codes of the scanned URL into a bidi-trie buffer, for reuse when string matching methods are called. New method: `tokenHistogram()`, used to generate histograms of occurrences of token extracted from URLs in built-in benchmark. The top results of the "miss" histogram are used as "bad tokens", i.e. tokens to avoid if possible when compiling filter lists. All plain pattern strings are now stored in the bidi-trie memory buffer, regardless of whether they will be used in the trie proper or not. Three methods have been added to the bidi-trie to test stored string against the URL which is also stored in then bidi-trie. FilterParser is now instanciated on demand and released when no longer used. *** [1] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/strie.js#L120 [2] https://github.com/gorhill/uBlock/commit/e94024d350b066e4e04a772b0a3dbc69daab3fb7 [3] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/static-net-filtering.js#L1630 [4] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/static-net-filtering.js#L1566
2019-10-21 14:15:58 +02:00
_tokenize(encodeInto) {
Add HNTrie-based filter classes to store origin-only filters Related issue: - https://github.com/uBlockOrigin/uBlock-issues/issues/528#issuecomment-484408622 Following STrie-related work in above issue, I noticed that a large number of filters in EasyList were filters which only had to match against the document origin. For instance, among just the top 10 most populous buckets, there were four such buckets with over hundreds of entries each: - bits: 72, token: "http", 146 entries - bits: 72, token: "https", 139 entries - bits: 88, token: "http", 122 entries - bits: 88, token: "https", 118 entries These filters in these buckets have to be matched against all the network requests. In order to leverage HNTrie for these filters[1], they are now handled in a special way so as to ensure they all end up in a single HNTrie (per bucket), which means that instead of scanning hundreds of entries per URL, there is now a single scan per bucket per URL for these apply-everywhere filters. Now, any filter which fulfill ALL the following condition will be processed in a special manner internally: - Is of the form `|https://` or `|http://` or `*`; and - Does have a `domain=` option; and - Does not have a negated domain in its `domain=` option; and - Does not have `csp=` option; and - Does not have a `redirect=` option If a filter does not fulfill ALL the conditions above, no change in behavior. A filter which matches ALL of the above will be processed in a special manner: - The `domain=` option will be decomposed so as to create as many distinct filter as there is distinct value in the `domain=` option - This also apply to the `badfilter` version of the filter, which means it now become possible to `badfilter` only one of the distinct filter without having to `badfilter` all of them. - The logger will always report these special filters with only a single hostname in the `domain=` option. *** [1] HNTrie is currently WASM-ed on Firefox.
2019-04-19 22:33:46 +02:00
const tokens = this._tokens;
let url = this._urlOut;
let l = url.length;
if ( l === 0 ) { return 0; }
2017-05-19 14:45:19 +02:00
if ( l > 2048 ) {
url = url.slice(0, 2048);
l = 2048;
}
Expand bidi-trie usage in static network filtering engine Related issues: - https://github.com/uBlockOrigin/uBlock-issues/issues/761 - https://github.com/uBlockOrigin/uBlock-issues/issues/528 The previous bidi-trie code could only hold filters which are plain pattern, i.e. no wildcard characters, and which had no origin option (`domain=`), right and/or left anchor, and no `csp=` option. Example of filters that could be moved into a bidi-trie data structure: &ad_box_ /w/d/capu.php?z=$script,third-party ||liveonlinetv247.com/images/muvixx-150x50-watch-now-in-hd-play-btn.gif Examples of filters that could NOT be moved to a bidi-trie: -adap.$domain=~l-adap.org /tsc.php?*&ses= ||ibsrv.net/*forumsponsor$domain=[...] @@||imgspice.com/jquery.cookie.js|$script ||view.atdmt.com^*/iview/$third-party ||postimg.cc/image/$csp=[...] Ideally the filters above should be able to be moved to a bidi-trie since they are basically plain patterns, or at least partially moved to a bidi-trie when there is only a single wildcard (i.e. made of two plain patterns). Also, there were two distinct bidi-tries in which plain-pattern filters can be moved to: one for patterns without hostname anchoring and another one for patterns with hostname-anchoring. This was required because the hostname-anchored patterns have an extra condition which is outside the bidi-trie knowledge. This commit expands the number of filters which can be stored in the bidi-trie, and also remove the need to use two distinct bidi-tries. - Added ability to associate a pattern with an integer in the bidi-trie [1]. - The bidi-trie match code passes this externally provided integer when calling an externally provided method used for testing extra conditions that may be present for a plain pattern found to be matching in the bidi-trie. - Decomposed existing filters into smaller logical units: - FilterPlainLeftAnchored => FilterPatternPlain + FilterAnchorLeft - FilterPlainRightAnchored => FilterPatternPlain + FilterAnchorRight - FilterExactMatch => FilterPatternPlain + FilterAnchorLeft + FilterAnchorRight - FilterPlainHnAnchored => FilterPatternPlain + FilterAnchorHn - FilterWildcard1 => FilterPatternPlain + [ FilterPatternLeft or FilterPatternRight ] - FilterWildcard1HnAnchored => FilterPatternPlain + [ FilterPatternLeft or FilterPatternRight ] + FilterAnchorHn - FilterGenericHnAnchored => FilterPatternGeneric + FilterAnchorHn - FilterGenericHnAndRightAnchored => FilterPatternGeneric + FilterAnchorRight + FilterAnchorHn - FilterOriginMixedSet => FilterOriginMissSet + FilterOriginHitSet - Instances of FilterOrigin[...], FilterDataHolder can also be added to a composite filter to represent `domain=` and `csp=` options. - Added a new filter class, FilterComposite, for filters which are a combination of two or more logical units. A FilterComposite instance is a match when *all* filters composing it are a match. Since filters are now encoded into combination of smaller units, it becomes possible to extract the FilterPatternPlain component and store it in the bidi-trie, and use the integer as a handle for the remaining extra conditions, if any. Since a single pattern in the bidi-trie may be a component for different filters, the associated integer points to a sequence of extra conditions, and a match occurs as soon as one of the extra conditions (which may itself be a sequence of conditions) is fulfilled. Decomposing filters which are currently single instance into sequences of smaller logical filters means increasing the storage and CPU overhead when evaluating such filters. The CPU overhead is compensated by the fact that more filters can now moved into the bidi-trie, where the first match is efficiently evaluated. The extra conditions have to be evaluated if and only if there is a match in the bidi-trie. The storage overhead is compensated by the bidi-trie's intrinsic nature of merging similar patterns. Furthermore, the storage overhead is reduced by no longer using JavaScript array to store collection of filters (which is what FilterComposite is): the same technique used in [2] is imported to store sequences of filters. A sequence of filters is a sequence of integer pairs where the first integer is an index to an actual filter instance stored in a global array of filters (`filterUnits`), while the second integer is an index to the next pair in the sequence -- which means all sequences of filters are encoded in one single array of integers (`filterSequences` => Uint32Array). As a result, a sequence of filters can be represented by one single integer -- an index to the first pair -- regardless of the number of filters in the sequence. This representation is further leveraged to replace the use of JavaScript array in FilterBucket [3], which used a JavaScript array to store collection of filters. Doing so means there is no more need for FilterPair [4], which purpose was to be a lightweight representation when there was only two filters in a collection. As a result of the above changes, the map of `token` (integer) => filter instance (object) used to associate tokens to filters or collections of filters is replaced with a more efficient map of `token` (integer) to filter unit index (integer) to lookup a filter object from the global `filterUnits` array. Another consequence of using one single global array to store all filter instances means we can reuse existing instances when a logical filter instance is parameter-less, which is the case for FilterAnchorLeft, FilterAnchorRight, FilterAnchorHn, the index to these single instances is reused where needed. `urlTokenizer` now stores the character codes of the scanned URL into a bidi-trie buffer, for reuse when string matching methods are called. New method: `tokenHistogram()`, used to generate histograms of occurrences of token extracted from URLs in built-in benchmark. The top results of the "miss" histogram are used as "bad tokens", i.e. tokens to avoid if possible when compiling filter lists. All plain pattern strings are now stored in the bidi-trie memory buffer, regardless of whether they will be used in the trie proper or not. Three methods have been added to the bidi-trie to test stored string against the URL which is also stored in then bidi-trie. FilterParser is now instanciated on demand and released when no longer used. *** [1] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/strie.js#L120 [2] https://github.com/gorhill/uBlock/commit/e94024d350b066e4e04a772b0a3dbc69daab3fb7 [3] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/static-net-filtering.js#L1630 [4] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/static-net-filtering.js#L1566
2019-10-21 14:15:58 +02:00
encodeInto.haystackSize = l;
const knownTokens = this.knownTokens;
Add HNTrie-based filter classes to store origin-only filters Related issue: - https://github.com/uBlockOrigin/uBlock-issues/issues/528#issuecomment-484408622 Following STrie-related work in above issue, I noticed that a large number of filters in EasyList were filters which only had to match against the document origin. For instance, among just the top 10 most populous buckets, there were four such buckets with over hundreds of entries each: - bits: 72, token: "http", 146 entries - bits: 72, token: "https", 139 entries - bits: 88, token: "http", 122 entries - bits: 88, token: "https", 118 entries These filters in these buckets have to be matched against all the network requests. In order to leverage HNTrie for these filters[1], they are now handled in a special way so as to ensure they all end up in a single HNTrie (per bucket), which means that instead of scanning hundreds of entries per URL, there is now a single scan per bucket per URL for these apply-everywhere filters. Now, any filter which fulfill ALL the following condition will be processed in a special manner internally: - Is of the form `|https://` or `|http://` or `*`; and - Does have a `domain=` option; and - Does not have a negated domain in its `domain=` option; and - Does not have `csp=` option; and - Does not have a `redirect=` option If a filter does not fulfill ALL the conditions above, no change in behavior. A filter which matches ALL of the above will be processed in a special manner: - The `domain=` option will be decomposed so as to create as many distinct filter as there is distinct value in the `domain=` option - This also apply to the `badfilter` version of the filter, which means it now become possible to `badfilter` only one of the distinct filter without having to `badfilter` all of them. - The logger will always report these special filters with only a single hostname in the `domain=` option. *** [1] HNTrie is currently WASM-ed on Firefox.
2019-04-19 22:33:46 +02:00
const vtc = this._validTokenChars;
Expand bidi-trie usage in static network filtering engine Related issues: - https://github.com/uBlockOrigin/uBlock-issues/issues/761 - https://github.com/uBlockOrigin/uBlock-issues/issues/528 The previous bidi-trie code could only hold filters which are plain pattern, i.e. no wildcard characters, and which had no origin option (`domain=`), right and/or left anchor, and no `csp=` option. Example of filters that could be moved into a bidi-trie data structure: &ad_box_ /w/d/capu.php?z=$script,third-party ||liveonlinetv247.com/images/muvixx-150x50-watch-now-in-hd-play-btn.gif Examples of filters that could NOT be moved to a bidi-trie: -adap.$domain=~l-adap.org /tsc.php?*&ses= ||ibsrv.net/*forumsponsor$domain=[...] @@||imgspice.com/jquery.cookie.js|$script ||view.atdmt.com^*/iview/$third-party ||postimg.cc/image/$csp=[...] Ideally the filters above should be able to be moved to a bidi-trie since they are basically plain patterns, or at least partially moved to a bidi-trie when there is only a single wildcard (i.e. made of two plain patterns). Also, there were two distinct bidi-tries in which plain-pattern filters can be moved to: one for patterns without hostname anchoring and another one for patterns with hostname-anchoring. This was required because the hostname-anchored patterns have an extra condition which is outside the bidi-trie knowledge. This commit expands the number of filters which can be stored in the bidi-trie, and also remove the need to use two distinct bidi-tries. - Added ability to associate a pattern with an integer in the bidi-trie [1]. - The bidi-trie match code passes this externally provided integer when calling an externally provided method used for testing extra conditions that may be present for a plain pattern found to be matching in the bidi-trie. - Decomposed existing filters into smaller logical units: - FilterPlainLeftAnchored => FilterPatternPlain + FilterAnchorLeft - FilterPlainRightAnchored => FilterPatternPlain + FilterAnchorRight - FilterExactMatch => FilterPatternPlain + FilterAnchorLeft + FilterAnchorRight - FilterPlainHnAnchored => FilterPatternPlain + FilterAnchorHn - FilterWildcard1 => FilterPatternPlain + [ FilterPatternLeft or FilterPatternRight ] - FilterWildcard1HnAnchored => FilterPatternPlain + [ FilterPatternLeft or FilterPatternRight ] + FilterAnchorHn - FilterGenericHnAnchored => FilterPatternGeneric + FilterAnchorHn - FilterGenericHnAndRightAnchored => FilterPatternGeneric + FilterAnchorRight + FilterAnchorHn - FilterOriginMixedSet => FilterOriginMissSet + FilterOriginHitSet - Instances of FilterOrigin[...], FilterDataHolder can also be added to a composite filter to represent `domain=` and `csp=` options. - Added a new filter class, FilterComposite, for filters which are a combination of two or more logical units. A FilterComposite instance is a match when *all* filters composing it are a match. Since filters are now encoded into combination of smaller units, it becomes possible to extract the FilterPatternPlain component and store it in the bidi-trie, and use the integer as a handle for the remaining extra conditions, if any. Since a single pattern in the bidi-trie may be a component for different filters, the associated integer points to a sequence of extra conditions, and a match occurs as soon as one of the extra conditions (which may itself be a sequence of conditions) is fulfilled. Decomposing filters which are currently single instance into sequences of smaller logical filters means increasing the storage and CPU overhead when evaluating such filters. The CPU overhead is compensated by the fact that more filters can now moved into the bidi-trie, where the first match is efficiently evaluated. The extra conditions have to be evaluated if and only if there is a match in the bidi-trie. The storage overhead is compensated by the bidi-trie's intrinsic nature of merging similar patterns. Furthermore, the storage overhead is reduced by no longer using JavaScript array to store collection of filters (which is what FilterComposite is): the same technique used in [2] is imported to store sequences of filters. A sequence of filters is a sequence of integer pairs where the first integer is an index to an actual filter instance stored in a global array of filters (`filterUnits`), while the second integer is an index to the next pair in the sequence -- which means all sequences of filters are encoded in one single array of integers (`filterSequences` => Uint32Array). As a result, a sequence of filters can be represented by one single integer -- an index to the first pair -- regardless of the number of filters in the sequence. This representation is further leveraged to replace the use of JavaScript array in FilterBucket [3], which used a JavaScript array to store collection of filters. Doing so means there is no more need for FilterPair [4], which purpose was to be a lightweight representation when there was only two filters in a collection. As a result of the above changes, the map of `token` (integer) => filter instance (object) used to associate tokens to filters or collections of filters is replaced with a more efficient map of `token` (integer) to filter unit index (integer) to lookup a filter object from the global `filterUnits` array. Another consequence of using one single global array to store all filter instances means we can reuse existing instances when a logical filter instance is parameter-less, which is the case for FilterAnchorLeft, FilterAnchorRight, FilterAnchorHn, the index to these single instances is reused where needed. `urlTokenizer` now stores the character codes of the scanned URL into a bidi-trie buffer, for reuse when string matching methods are called. New method: `tokenHistogram()`, used to generate histograms of occurrences of token extracted from URLs in built-in benchmark. The top results of the "miss" histogram are used as "bad tokens", i.e. tokens to avoid if possible when compiling filter lists. All plain pattern strings are now stored in the bidi-trie memory buffer, regardless of whether they will be used in the trie proper or not. Three methods have been added to the bidi-trie to test stored string against the URL which is also stored in then bidi-trie. FilterParser is now instanciated on demand and released when no longer used. *** [1] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/strie.js#L120 [2] https://github.com/gorhill/uBlock/commit/e94024d350b066e4e04a772b0a3dbc69daab3fb7 [3] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/static-net-filtering.js#L1630 [4] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/static-net-filtering.js#L1566
2019-10-21 14:15:58 +02:00
const charCodes = encodeInto.haystack;
let i = 0, j = 0, c, v, n, ti, th;
2017-05-19 14:45:19 +02:00
for (;;) {
for (;;) {
Add HNTrie-based filter classes to store origin-only filters Related issue: - https://github.com/uBlockOrigin/uBlock-issues/issues/528#issuecomment-484408622 Following STrie-related work in above issue, I noticed that a large number of filters in EasyList were filters which only had to match against the document origin. For instance, among just the top 10 most populous buckets, there were four such buckets with over hundreds of entries each: - bits: 72, token: "http", 146 entries - bits: 72, token: "https", 139 entries - bits: 88, token: "http", 122 entries - bits: 88, token: "https", 118 entries These filters in these buckets have to be matched against all the network requests. In order to leverage HNTrie for these filters[1], they are now handled in a special way so as to ensure they all end up in a single HNTrie (per bucket), which means that instead of scanning hundreds of entries per URL, there is now a single scan per bucket per URL for these apply-everywhere filters. Now, any filter which fulfill ALL the following condition will be processed in a special manner internally: - Is of the form `|https://` or `|http://` or `*`; and - Does have a `domain=` option; and - Does not have a negated domain in its `domain=` option; and - Does not have `csp=` option; and - Does not have a `redirect=` option If a filter does not fulfill ALL the conditions above, no change in behavior. A filter which matches ALL of the above will be processed in a special manner: - The `domain=` option will be decomposed so as to create as many distinct filter as there is distinct value in the `domain=` option - This also apply to the `badfilter` version of the filter, which means it now become possible to `badfilter` only one of the distinct filter without having to `badfilter` all of them. - The logger will always report these special filters with only a single hostname in the `domain=` option. *** [1] HNTrie is currently WASM-ed on Firefox.
2019-04-19 22:33:46 +02:00
if ( i === l ) { return j; }
Expand bidi-trie usage in static network filtering engine Related issues: - https://github.com/uBlockOrigin/uBlock-issues/issues/761 - https://github.com/uBlockOrigin/uBlock-issues/issues/528 The previous bidi-trie code could only hold filters which are plain pattern, i.e. no wildcard characters, and which had no origin option (`domain=`), right and/or left anchor, and no `csp=` option. Example of filters that could be moved into a bidi-trie data structure: &ad_box_ /w/d/capu.php?z=$script,third-party ||liveonlinetv247.com/images/muvixx-150x50-watch-now-in-hd-play-btn.gif Examples of filters that could NOT be moved to a bidi-trie: -adap.$domain=~l-adap.org /tsc.php?*&ses= ||ibsrv.net/*forumsponsor$domain=[...] @@||imgspice.com/jquery.cookie.js|$script ||view.atdmt.com^*/iview/$third-party ||postimg.cc/image/$csp=[...] Ideally the filters above should be able to be moved to a bidi-trie since they are basically plain patterns, or at least partially moved to a bidi-trie when there is only a single wildcard (i.e. made of two plain patterns). Also, there were two distinct bidi-tries in which plain-pattern filters can be moved to: one for patterns without hostname anchoring and another one for patterns with hostname-anchoring. This was required because the hostname-anchored patterns have an extra condition which is outside the bidi-trie knowledge. This commit expands the number of filters which can be stored in the bidi-trie, and also remove the need to use two distinct bidi-tries. - Added ability to associate a pattern with an integer in the bidi-trie [1]. - The bidi-trie match code passes this externally provided integer when calling an externally provided method used for testing extra conditions that may be present for a plain pattern found to be matching in the bidi-trie. - Decomposed existing filters into smaller logical units: - FilterPlainLeftAnchored => FilterPatternPlain + FilterAnchorLeft - FilterPlainRightAnchored => FilterPatternPlain + FilterAnchorRight - FilterExactMatch => FilterPatternPlain + FilterAnchorLeft + FilterAnchorRight - FilterPlainHnAnchored => FilterPatternPlain + FilterAnchorHn - FilterWildcard1 => FilterPatternPlain + [ FilterPatternLeft or FilterPatternRight ] - FilterWildcard1HnAnchored => FilterPatternPlain + [ FilterPatternLeft or FilterPatternRight ] + FilterAnchorHn - FilterGenericHnAnchored => FilterPatternGeneric + FilterAnchorHn - FilterGenericHnAndRightAnchored => FilterPatternGeneric + FilterAnchorRight + FilterAnchorHn - FilterOriginMixedSet => FilterOriginMissSet + FilterOriginHitSet - Instances of FilterOrigin[...], FilterDataHolder can also be added to a composite filter to represent `domain=` and `csp=` options. - Added a new filter class, FilterComposite, for filters which are a combination of two or more logical units. A FilterComposite instance is a match when *all* filters composing it are a match. Since filters are now encoded into combination of smaller units, it becomes possible to extract the FilterPatternPlain component and store it in the bidi-trie, and use the integer as a handle for the remaining extra conditions, if any. Since a single pattern in the bidi-trie may be a component for different filters, the associated integer points to a sequence of extra conditions, and a match occurs as soon as one of the extra conditions (which may itself be a sequence of conditions) is fulfilled. Decomposing filters which are currently single instance into sequences of smaller logical filters means increasing the storage and CPU overhead when evaluating such filters. The CPU overhead is compensated by the fact that more filters can now moved into the bidi-trie, where the first match is efficiently evaluated. The extra conditions have to be evaluated if and only if there is a match in the bidi-trie. The storage overhead is compensated by the bidi-trie's intrinsic nature of merging similar patterns. Furthermore, the storage overhead is reduced by no longer using JavaScript array to store collection of filters (which is what FilterComposite is): the same technique used in [2] is imported to store sequences of filters. A sequence of filters is a sequence of integer pairs where the first integer is an index to an actual filter instance stored in a global array of filters (`filterUnits`), while the second integer is an index to the next pair in the sequence -- which means all sequences of filters are encoded in one single array of integers (`filterSequences` => Uint32Array). As a result, a sequence of filters can be represented by one single integer -- an index to the first pair -- regardless of the number of filters in the sequence. This representation is further leveraged to replace the use of JavaScript array in FilterBucket [3], which used a JavaScript array to store collection of filters. Doing so means there is no more need for FilterPair [4], which purpose was to be a lightweight representation when there was only two filters in a collection. As a result of the above changes, the map of `token` (integer) => filter instance (object) used to associate tokens to filters or collections of filters is replaced with a more efficient map of `token` (integer) to filter unit index (integer) to lookup a filter object from the global `filterUnits` array. Another consequence of using one single global array to store all filter instances means we can reuse existing instances when a logical filter instance is parameter-less, which is the case for FilterAnchorLeft, FilterAnchorRight, FilterAnchorHn, the index to these single instances is reused where needed. `urlTokenizer` now stores the character codes of the scanned URL into a bidi-trie buffer, for reuse when string matching methods are called. New method: `tokenHistogram()`, used to generate histograms of occurrences of token extracted from URLs in built-in benchmark. The top results of the "miss" histogram are used as "bad tokens", i.e. tokens to avoid if possible when compiling filter lists. All plain pattern strings are now stored in the bidi-trie memory buffer, regardless of whether they will be used in the trie proper or not. Three methods have been added to the bidi-trie to test stored string against the URL which is also stored in then bidi-trie. FilterParser is now instanciated on demand and released when no longer used. *** [1] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/strie.js#L120 [2] https://github.com/gorhill/uBlock/commit/e94024d350b066e4e04a772b0a3dbc69daab3fb7 [3] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/static-net-filtering.js#L1630 [4] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/static-net-filtering.js#L1566
2019-10-21 14:15:58 +02:00
c = url.charCodeAt(i);
charCodes[i] = c;
v = vtc[c];
i += 1;
2017-05-19 14:45:19 +02:00
if ( v !== 0 ) { break; }
}
th = v; ti = i - 1; n = 1;
for (;;) {
if ( i === l ) { break; }
Expand bidi-trie usage in static network filtering engine Related issues: - https://github.com/uBlockOrigin/uBlock-issues/issues/761 - https://github.com/uBlockOrigin/uBlock-issues/issues/528 The previous bidi-trie code could only hold filters which are plain pattern, i.e. no wildcard characters, and which had no origin option (`domain=`), right and/or left anchor, and no `csp=` option. Example of filters that could be moved into a bidi-trie data structure: &ad_box_ /w/d/capu.php?z=$script,third-party ||liveonlinetv247.com/images/muvixx-150x50-watch-now-in-hd-play-btn.gif Examples of filters that could NOT be moved to a bidi-trie: -adap.$domain=~l-adap.org /tsc.php?*&ses= ||ibsrv.net/*forumsponsor$domain=[...] @@||imgspice.com/jquery.cookie.js|$script ||view.atdmt.com^*/iview/$third-party ||postimg.cc/image/$csp=[...] Ideally the filters above should be able to be moved to a bidi-trie since they are basically plain patterns, or at least partially moved to a bidi-trie when there is only a single wildcard (i.e. made of two plain patterns). Also, there were two distinct bidi-tries in which plain-pattern filters can be moved to: one for patterns without hostname anchoring and another one for patterns with hostname-anchoring. This was required because the hostname-anchored patterns have an extra condition which is outside the bidi-trie knowledge. This commit expands the number of filters which can be stored in the bidi-trie, and also remove the need to use two distinct bidi-tries. - Added ability to associate a pattern with an integer in the bidi-trie [1]. - The bidi-trie match code passes this externally provided integer when calling an externally provided method used for testing extra conditions that may be present for a plain pattern found to be matching in the bidi-trie. - Decomposed existing filters into smaller logical units: - FilterPlainLeftAnchored => FilterPatternPlain + FilterAnchorLeft - FilterPlainRightAnchored => FilterPatternPlain + FilterAnchorRight - FilterExactMatch => FilterPatternPlain + FilterAnchorLeft + FilterAnchorRight - FilterPlainHnAnchored => FilterPatternPlain + FilterAnchorHn - FilterWildcard1 => FilterPatternPlain + [ FilterPatternLeft or FilterPatternRight ] - FilterWildcard1HnAnchored => FilterPatternPlain + [ FilterPatternLeft or FilterPatternRight ] + FilterAnchorHn - FilterGenericHnAnchored => FilterPatternGeneric + FilterAnchorHn - FilterGenericHnAndRightAnchored => FilterPatternGeneric + FilterAnchorRight + FilterAnchorHn - FilterOriginMixedSet => FilterOriginMissSet + FilterOriginHitSet - Instances of FilterOrigin[...], FilterDataHolder can also be added to a composite filter to represent `domain=` and `csp=` options. - Added a new filter class, FilterComposite, for filters which are a combination of two or more logical units. A FilterComposite instance is a match when *all* filters composing it are a match. Since filters are now encoded into combination of smaller units, it becomes possible to extract the FilterPatternPlain component and store it in the bidi-trie, and use the integer as a handle for the remaining extra conditions, if any. Since a single pattern in the bidi-trie may be a component for different filters, the associated integer points to a sequence of extra conditions, and a match occurs as soon as one of the extra conditions (which may itself be a sequence of conditions) is fulfilled. Decomposing filters which are currently single instance into sequences of smaller logical filters means increasing the storage and CPU overhead when evaluating such filters. The CPU overhead is compensated by the fact that more filters can now moved into the bidi-trie, where the first match is efficiently evaluated. The extra conditions have to be evaluated if and only if there is a match in the bidi-trie. The storage overhead is compensated by the bidi-trie's intrinsic nature of merging similar patterns. Furthermore, the storage overhead is reduced by no longer using JavaScript array to store collection of filters (which is what FilterComposite is): the same technique used in [2] is imported to store sequences of filters. A sequence of filters is a sequence of integer pairs where the first integer is an index to an actual filter instance stored in a global array of filters (`filterUnits`), while the second integer is an index to the next pair in the sequence -- which means all sequences of filters are encoded in one single array of integers (`filterSequences` => Uint32Array). As a result, a sequence of filters can be represented by one single integer -- an index to the first pair -- regardless of the number of filters in the sequence. This representation is further leveraged to replace the use of JavaScript array in FilterBucket [3], which used a JavaScript array to store collection of filters. Doing so means there is no more need for FilterPair [4], which purpose was to be a lightweight representation when there was only two filters in a collection. As a result of the above changes, the map of `token` (integer) => filter instance (object) used to associate tokens to filters or collections of filters is replaced with a more efficient map of `token` (integer) to filter unit index (integer) to lookup a filter object from the global `filterUnits` array. Another consequence of using one single global array to store all filter instances means we can reuse existing instances when a logical filter instance is parameter-less, which is the case for FilterAnchorLeft, FilterAnchorRight, FilterAnchorHn, the index to these single instances is reused where needed. `urlTokenizer` now stores the character codes of the scanned URL into a bidi-trie buffer, for reuse when string matching methods are called. New method: `tokenHistogram()`, used to generate histograms of occurrences of token extracted from URLs in built-in benchmark. The top results of the "miss" histogram are used as "bad tokens", i.e. tokens to avoid if possible when compiling filter lists. All plain pattern strings are now stored in the bidi-trie memory buffer, regardless of whether they will be used in the trie proper or not. Three methods have been added to the bidi-trie to test stored string against the URL which is also stored in then bidi-trie. FilterParser is now instanciated on demand and released when no longer used. *** [1] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/strie.js#L120 [2] https://github.com/gorhill/uBlock/commit/e94024d350b066e4e04a772b0a3dbc69daab3fb7 [3] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/static-net-filtering.js#L1630 [4] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/static-net-filtering.js#L1566
2019-10-21 14:15:58 +02:00
c = url.charCodeAt(i);
charCodes[i] = c;
v = vtc[c];
i += 1;
2017-05-19 14:45:19 +02:00
if ( v === 0 ) { break; }
if ( n === 7 ) { continue; }
th = th << 4 ^ v;
n += 1;
2017-05-19 14:45:19 +02:00
}
if ( knownTokens[th & 0xFFFF ^ th >>> 16] !== 0 ) {
tokens[j+0] = th;
tokens[j+1] = ti;
j += 2;
}
2017-05-19 14:45:19 +02:00
}
Add HNTrie-based filter classes to store origin-only filters Related issue: - https://github.com/uBlockOrigin/uBlock-issues/issues/528#issuecomment-484408622 Following STrie-related work in above issue, I noticed that a large number of filters in EasyList were filters which only had to match against the document origin. For instance, among just the top 10 most populous buckets, there were four such buckets with over hundreds of entries each: - bits: 72, token: "http", 146 entries - bits: 72, token: "https", 139 entries - bits: 88, token: "http", 122 entries - bits: 88, token: "https", 118 entries These filters in these buckets have to be matched against all the network requests. In order to leverage HNTrie for these filters[1], they are now handled in a special way so as to ensure they all end up in a single HNTrie (per bucket), which means that instead of scanning hundreds of entries per URL, there is now a single scan per bucket per URL for these apply-everywhere filters. Now, any filter which fulfill ALL the following condition will be processed in a special manner internally: - Is of the form `|https://` or `|http://` or `*`; and - Does have a `domain=` option; and - Does not have a negated domain in its `domain=` option; and - Does not have `csp=` option; and - Does not have a `redirect=` option If a filter does not fulfill ALL the conditions above, no change in behavior. A filter which matches ALL of the above will be processed in a special manner: - The `domain=` option will be decomposed so as to create as many distinct filter as there is distinct value in the `domain=` option - This also apply to the `badfilter` version of the filter, which means it now become possible to `badfilter` only one of the distinct filter without having to `badfilter` all of them. - The logger will always report these special filters with only a single hostname in the `domain=` option. *** [1] HNTrie is currently WASM-ed on Firefox.
2019-04-19 22:33:46 +02:00
}
})();
2014-06-24 00:42:43 +02:00
/******************************************************************************/
µBlock.formatCount = function(count) {
if ( typeof count !== 'number' ) {
return '';
}
2019-09-11 14:08:30 +02:00
let s = count.toFixed(0);
if ( count >= 1000 ) {
if ( count < 10000 ) {
2014-12-24 14:11:22 +01:00
s = '>' + s.slice(0,1) + 'k';
} else if ( count < 100000 ) {
2014-12-24 14:11:22 +01:00
s = s.slice(0,2) + 'k';
} else if ( count < 1000000 ) {
2014-12-24 14:11:22 +01:00
s = s.slice(0,3) + 'k';
} else if ( count < 10000000 ) {
s = s.slice(0,1) + 'M';
} else {
s = s.slice(0,-6) + 'M';
}
}
return s;
2014-06-24 00:42:43 +02:00
};
2014-08-20 15:24:16 +02:00
// https://www.youtube.com/watch?v=DyvzfyqYm_s
/******************************************************************************/
2016-08-13 22:42:58 +02:00
2016-10-13 19:25:57 +02:00
µBlock.dateNowToSensibleString = function() {
2019-09-11 14:08:30 +02:00
const now = new Date(Date.now() - (new Date()).getTimezoneOffset() * 60000);
2016-10-13 19:25:57 +02:00
return now.toISOString().replace(/\.\d+Z$/, '')
.replace(/:/g, '.')
.replace('T', '_');
};
/******************************************************************************/
2019-09-11 14:08:30 +02:00
µBlock.LineIterator = class {
constructor(text, offset) {
this.text = text;
this.textLen = this.text.length;
this.offset = offset || 0;
}
2019-09-11 14:08:30 +02:00
next(offset) {
if ( offset !== undefined ) {
this.offset += offset;
}
let lineEnd = this.text.indexOf('\n', this.offset);
2016-08-13 22:42:58 +02:00
if ( lineEnd === -1 ) {
2019-09-11 14:08:30 +02:00
lineEnd = this.text.indexOf('\r', this.offset);
if ( lineEnd === -1 ) {
lineEnd = this.textLen;
}
2016-08-13 22:42:58 +02:00
}
2019-09-11 14:08:30 +02:00
const line = this.text.slice(this.offset, lineEnd);
this.offset = lineEnd + 1;
return line;
}
charCodeAt(offset) {
return this.text.charCodeAt(this.offset + offset);
}
eot() {
return this.offset >= this.textLen;
2016-08-13 22:42:58 +02:00
}
};
/******************************************************************************/
// The field iterator is less CPU-intensive than when using native
// String.split().
2019-09-11 14:08:30 +02:00
µBlock.FieldIterator = class {
constructor(sep) {
this.text = '';
this.sep = sep;
this.sepLen = sep.length;
this.offset = 0;
}
first(text) {
this.text = text;
this.offset = 0;
return this.next();
}
next() {
let end = this.text.indexOf(this.sep, this.offset);
if ( end === -1 ) {
end = this.text.length;
}
const field = this.text.slice(this.offset, end);
this.offset = end + this.sepLen;
return field;
}
remainder() {
return this.text.slice(this.offset);
}
};
/******************************************************************************/
µBlock.CompiledLineIO = {
serialize: JSON.stringify,
unserialize: JSON.parse,
blockStartPrefix: '#block-start-', // ensure no special regex characters
blockEndPrefix: '#block-end-', // ensure no special regex characters
2019-09-11 14:08:30 +02:00
Writer: class {
constructor() {
this.io = µBlock.CompiledLineIO;
this.blockId = undefined;
this.block = undefined;
this.stringifier = this.io.serialize;
this.blocks = new Map();
this.properties = new Map();
}
push(args) {
Expand bidi-trie usage in static network filtering engine Related issues: - https://github.com/uBlockOrigin/uBlock-issues/issues/761 - https://github.com/uBlockOrigin/uBlock-issues/issues/528 The previous bidi-trie code could only hold filters which are plain pattern, i.e. no wildcard characters, and which had no origin option (`domain=`), right and/or left anchor, and no `csp=` option. Example of filters that could be moved into a bidi-trie data structure: &ad_box_ /w/d/capu.php?z=$script,third-party ||liveonlinetv247.com/images/muvixx-150x50-watch-now-in-hd-play-btn.gif Examples of filters that could NOT be moved to a bidi-trie: -adap.$domain=~l-adap.org /tsc.php?*&ses= ||ibsrv.net/*forumsponsor$domain=[...] @@||imgspice.com/jquery.cookie.js|$script ||view.atdmt.com^*/iview/$third-party ||postimg.cc/image/$csp=[...] Ideally the filters above should be able to be moved to a bidi-trie since they are basically plain patterns, or at least partially moved to a bidi-trie when there is only a single wildcard (i.e. made of two plain patterns). Also, there were two distinct bidi-tries in which plain-pattern filters can be moved to: one for patterns without hostname anchoring and another one for patterns with hostname-anchoring. This was required because the hostname-anchored patterns have an extra condition which is outside the bidi-trie knowledge. This commit expands the number of filters which can be stored in the bidi-trie, and also remove the need to use two distinct bidi-tries. - Added ability to associate a pattern with an integer in the bidi-trie [1]. - The bidi-trie match code passes this externally provided integer when calling an externally provided method used for testing extra conditions that may be present for a plain pattern found to be matching in the bidi-trie. - Decomposed existing filters into smaller logical units: - FilterPlainLeftAnchored => FilterPatternPlain + FilterAnchorLeft - FilterPlainRightAnchored => FilterPatternPlain + FilterAnchorRight - FilterExactMatch => FilterPatternPlain + FilterAnchorLeft + FilterAnchorRight - FilterPlainHnAnchored => FilterPatternPlain + FilterAnchorHn - FilterWildcard1 => FilterPatternPlain + [ FilterPatternLeft or FilterPatternRight ] - FilterWildcard1HnAnchored => FilterPatternPlain + [ FilterPatternLeft or FilterPatternRight ] + FilterAnchorHn - FilterGenericHnAnchored => FilterPatternGeneric + FilterAnchorHn - FilterGenericHnAndRightAnchored => FilterPatternGeneric + FilterAnchorRight + FilterAnchorHn - FilterOriginMixedSet => FilterOriginMissSet + FilterOriginHitSet - Instances of FilterOrigin[...], FilterDataHolder can also be added to a composite filter to represent `domain=` and `csp=` options. - Added a new filter class, FilterComposite, for filters which are a combination of two or more logical units. A FilterComposite instance is a match when *all* filters composing it are a match. Since filters are now encoded into combination of smaller units, it becomes possible to extract the FilterPatternPlain component and store it in the bidi-trie, and use the integer as a handle for the remaining extra conditions, if any. Since a single pattern in the bidi-trie may be a component for different filters, the associated integer points to a sequence of extra conditions, and a match occurs as soon as one of the extra conditions (which may itself be a sequence of conditions) is fulfilled. Decomposing filters which are currently single instance into sequences of smaller logical filters means increasing the storage and CPU overhead when evaluating such filters. The CPU overhead is compensated by the fact that more filters can now moved into the bidi-trie, where the first match is efficiently evaluated. The extra conditions have to be evaluated if and only if there is a match in the bidi-trie. The storage overhead is compensated by the bidi-trie's intrinsic nature of merging similar patterns. Furthermore, the storage overhead is reduced by no longer using JavaScript array to store collection of filters (which is what FilterComposite is): the same technique used in [2] is imported to store sequences of filters. A sequence of filters is a sequence of integer pairs where the first integer is an index to an actual filter instance stored in a global array of filters (`filterUnits`), while the second integer is an index to the next pair in the sequence -- which means all sequences of filters are encoded in one single array of integers (`filterSequences` => Uint32Array). As a result, a sequence of filters can be represented by one single integer -- an index to the first pair -- regardless of the number of filters in the sequence. This representation is further leveraged to replace the use of JavaScript array in FilterBucket [3], which used a JavaScript array to store collection of filters. Doing so means there is no more need for FilterPair [4], which purpose was to be a lightweight representation when there was only two filters in a collection. As a result of the above changes, the map of `token` (integer) => filter instance (object) used to associate tokens to filters or collections of filters is replaced with a more efficient map of `token` (integer) to filter unit index (integer) to lookup a filter object from the global `filterUnits` array. Another consequence of using one single global array to store all filter instances means we can reuse existing instances when a logical filter instance is parameter-less, which is the case for FilterAnchorLeft, FilterAnchorRight, FilterAnchorHn, the index to these single instances is reused where needed. `urlTokenizer` now stores the character codes of the scanned URL into a bidi-trie buffer, for reuse when string matching methods are called. New method: `tokenHistogram()`, used to generate histograms of occurrences of token extracted from URLs in built-in benchmark. The top results of the "miss" histogram are used as "bad tokens", i.e. tokens to avoid if possible when compiling filter lists. All plain pattern strings are now stored in the bidi-trie memory buffer, regardless of whether they will be used in the trie proper or not. Three methods have been added to the bidi-trie to test stored string against the URL which is also stored in then bidi-trie. FilterParser is now instanciated on demand and released when no longer used. *** [1] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/strie.js#L120 [2] https://github.com/gorhill/uBlock/commit/e94024d350b066e4e04a772b0a3dbc69daab3fb7 [3] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/static-net-filtering.js#L1630 [4] https://github.com/gorhill/uBlock/blob/135a45a878f5b93bc538f822981e3a42b1e9073f/src/js/static-net-filtering.js#L1566
2019-10-21 14:15:58 +02:00
this.block.push(this.stringifier(args));
}
last() {
if ( Array.isArray(this.block) && this.block.length !== 0 ) {
return this.block[this.block.length - 1];
}
2019-09-11 14:08:30 +02:00
}
select(blockId) {
if ( blockId === this.blockId ) { return; }
this.blockId = blockId;
this.block = this.blocks.get(blockId);
if ( this.block === undefined ) {
this.blocks.set(blockId, (this.block = []));
}
}
2019-09-11 14:08:30 +02:00
toString() {
let result = [];
for ( let [ id, lines ] of this.blocks ) {
if ( lines.length === 0 ) { continue; }
result.push(
this.io.blockStartPrefix + id,
lines.join('\n'),
this.io.blockEndPrefix + id
);
}
return result.join('\n');
}
},
2019-09-11 14:08:30 +02:00
Reader: class {
constructor(raw, blockId) {
this.io = µBlock.CompiledLineIO;
this.block = '';
this.len = 0;
this.offset = 0;
this.line = '';
2019-09-11 14:08:30 +02:00
this.parser = this.io.unserialize;
this.blocks = new Map();
this.properties = new Map();
let reBlockStart = new RegExp(
'^' + this.io.blockStartPrefix + '(\\d+)\\n',
'gm'
);
let match = reBlockStart.exec(raw);
while ( match !== null ) {
let beg = match.index + match[0].length;
let end = raw.indexOf(this.io.blockEndPrefix + match[1], beg);
this.blocks.set(parseInt(match[1], 10), raw.slice(beg, end));
reBlockStart.lastIndex = end;
match = reBlockStart.exec(raw);
}
if ( blockId !== undefined ) {
this.select(blockId);
}
}
2019-09-11 14:08:30 +02:00
next() {
if ( this.offset === this.len ) {
this.line = '';
return false;
}
let pos = this.block.indexOf('\n', this.offset);
if ( pos !== -1 ) {
this.line = this.block.slice(this.offset, pos);
this.offset = pos + 1;
} else {
this.line = this.block.slice(this.offset);
this.offset = this.len;
}
return true;
}
select(blockId) {
this.block = this.blocks.get(blockId) || '';
this.len = this.block.length;
this.offset = 0;
return this;
}
fingerprint() {
return this.line;
}
args() {
return this.parser(this.line);
}
}
};
/******************************************************************************/
2016-09-16 23:41:17 +02:00
µBlock.openNewTab = function(details) {
if ( details.url.startsWith('logger-ui.html') ) {
if ( details.shiftKey ) {
this.changeUserSettings(
'alwaysDetachLogger',
!this.userSettings.alwaysDetachLogger
);
2016-09-16 23:41:17 +02:00
}
if ( this.userSettings.alwaysDetachLogger ) {
details.popup = this.hiddenSettings.loggerPopupType;
const url = new URL(vAPI.getURL(details.url));
url.searchParams.set('popup', '1');
details.url = url.href;
let popupLoggerBox;
try {
popupLoggerBox = JSON.parse(
vAPI.localStorage.getItem('popupLoggerBox')
);
} catch(ex) {
}
if ( popupLoggerBox !== undefined ) {
details.box = popupLoggerBox;
}
}
2016-09-16 23:41:17 +02:00
}
vAPI.tabs.open(details);
};
/******************************************************************************/
2017-01-27 19:44:52 +01:00
2019-09-11 14:08:30 +02:00
µBlock.MRUCache = class {
constructor(size) {
this.size = size;
this.array = [];
this.map = new Map();
this.resetTime = Date.now();
}
add(key, value) {
const found = this.map.has(key);
2017-10-21 19:43:46 +02:00
this.map.set(key, value);
if ( !found ) {
if ( this.array.length === this.size ) {
this.map.delete(this.array.pop());
}
this.array.unshift(key);
}
2019-09-11 14:08:30 +02:00
}
remove(key) {
2017-10-21 19:43:46 +02:00
if ( this.map.has(key) ) {
this.array.splice(this.array.indexOf(key), 1);
}
2019-09-11 14:08:30 +02:00
}
lookup(key) {
const value = this.map.get(key);
2017-10-21 19:43:46 +02:00
if ( value !== undefined && this.array[0] !== key ) {
2019-09-11 14:08:30 +02:00
let i = this.array.indexOf(key);
do {
this.array[i] = this.array[i-1];
} while ( --i );
this.array[0] = key;
2017-10-21 19:43:46 +02:00
}
return value;
2019-09-11 14:08:30 +02:00
}
reset() {
2017-10-21 19:43:46 +02:00
this.array = [];
this.map.clear();
this.resetTime = Date.now();
2017-10-21 19:43:46 +02:00
}
};
/******************************************************************************/
2017-11-09 18:53:05 +01:00
// https://developer.mozilla.org/en-US/docs/Web/JavaScript/Guide/Regular_Expressions
µBlock.escapeRegex = function(s) {
return s.replace(/[.*+?^${}()|[\]\\]/g, '\\$&');
};
/******************************************************************************/
2019-09-11 14:08:30 +02:00
µBlock.decomposeHostname = (( ) => {
// For performance purpose, as simple tests as possible
2019-09-11 14:08:30 +02:00
const reHostnameVeryCoarse = /[g-z_-]/;
const reIPv4VeryCoarse = /\.\d+$/;
2019-09-11 14:08:30 +02:00
const toBroaderHostname = function(hostname) {
const pos = hostname.indexOf('.');
if ( pos !== -1 ) {
return hostname.slice(pos + 1);
}
return hostname !== '*' && hostname !== '' ? '*' : '';
};
2019-09-11 14:08:30 +02:00
const toBroaderIPv4Address = function(ipaddress) {
2018-08-09 17:31:25 +02:00
if ( ipaddress === '*' || ipaddress === '' ) { return ''; }
2019-09-11 14:08:30 +02:00
const pos = ipaddress.lastIndexOf('.');
2018-08-09 17:31:25 +02:00
if ( pos === -1 ) { return '*'; }
return ipaddress.slice(0, pos);
};
2019-09-11 14:08:30 +02:00
const toBroaderIPv6Address = function(ipaddress) {
return ipaddress !== '*' && ipaddress !== '' ? '*' : '';
};
return function decomposeHostname(hostname, decomposed) {
if ( decomposed.length === 0 || decomposed[0] !== hostname ) {
2018-08-09 17:31:25 +02:00
let broaden;
if ( reHostnameVeryCoarse.test(hostname) === false ) {
if ( reIPv4VeryCoarse.test(hostname) ) {
2018-08-09 17:31:25 +02:00
broaden = toBroaderIPv4Address;
} else if ( hostname.startsWith('[') ) {
broaden = toBroaderIPv6Address;
}
}
if ( broaden === undefined ) {
broaden = toBroaderHostname;
}
decomposed[0] = hostname;
let i = 1;
for (;;) {
hostname = broaden(hostname);
if ( hostname === '' ) { break; }
decomposed[i++] = hostname;
}
decomposed.length = i;
}
return decomposed;
};
})();
/******************************************************************************/
// TODO: evaluate using TextEncoder/TextDecoder
µBlock.orphanizeString = function(s) {
return JSON.parse(JSON.stringify(s));
};
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
/******************************************************************************/
// Custom base64 encoder/decoder
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
//
// TODO:
// Could expand the LZ4 codec API to be able to return UTF8-safe string
// representation of a compressed buffer, and thus the code below could be
// moved LZ4 codec-side.
// https://github.com/uBlockOrigin/uBlock-issues/issues/461
// Provide a fallback encoding for Chromium 59 and less by issuing a plain
// JSON string. The fallback can be removed once min supported version is
// above 59.
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
µBlock.base64 = new (class {
constructor() {
this.valToDigit = new Uint8Array(64);
this.digitToVal = new Uint8Array(128);
const chars = "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz@%";
for ( let i = 0, n = chars.length; i < n; i++ ) {
const c = chars.charCodeAt(i);
this.valToDigit[i] = c;
this.digitToVal[c] = i;
}
this.magic = 'Base64_1';
}
encode(arrbuf, arrlen) {
const inputLength = (arrlen + 3) >>> 2;
const inbuf = new Uint32Array(arrbuf, 0, inputLength);
const outputLength = this.magic.length + 7 + inputLength * 7;
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 outbuf = new Uint8Array(outputLength);
let j = 0;
for ( let i = 0; i < this.magic.length; i++ ) {
outbuf[j++] = this.magic.charCodeAt(i);
}
let v = inputLength;
do {
outbuf[j++] = this.valToDigit[v & 0b111111];
v >>>= 6;
} while ( v !== 0 );
outbuf[j++] = 0x20 /* ' ' */;
for ( let i = 0; i < inputLength; i++ ) {
v = inbuf[i];
do {
outbuf[j++] = this.valToDigit[v & 0b111111];
v >>>= 6;
} while ( v !== 0 );
outbuf[j++] = 0x20 /* ' ' */;
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 ( typeof TextDecoder === 'undefined' ) {
return JSON.stringify(
Array.from(new Uint32Array(outbuf.buffer, 0, j >>> 2))
);
}
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 textDecoder = new TextDecoder();
return textDecoder.decode(new Uint8Array(outbuf.buffer, 0, j));
}
decode(instr, arrbuf) {
if ( instr.charCodeAt(0) === 0x5B /* '[' */ ) {
const inbuf = JSON.parse(instr);
if ( arrbuf instanceof ArrayBuffer === false ) {
return new Uint32Array(inbuf);
}
const outbuf = new Uint32Array(arrbuf);
outbuf.set(inbuf);
return outbuf;
}
if ( instr.startsWith(this.magic) === false ) {
throw new Error('Invalid µBlock.base64 encoding');
}
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 inputLength = instr.length;
const outbuf = arrbuf instanceof ArrayBuffer === false
? new Uint32Array(this.decodeSize(instr) >> 2)
: new Uint32Array(arrbuf);
let i = instr.indexOf(' ', this.magic.length) + 1;
if ( i === -1 ) {
throw new Error('Invalid µBlock.base64 encoding');
}
let j = 0;
for (;;) {
if ( i === inputLength ) { break; }
let v = 0, l = 0;
for (;;) {
const c = instr.charCodeAt(i++);
if ( c === 0x20 /* ' ' */ ) { break; }
v += this.digitToVal[c] << l;
l += 6;
}
outbuf[j++] = v;
}
return outbuf;
}
decodeSize(instr) {
if ( instr.startsWith(this.magic) === false ) { return 0; }
let v = 0, l = 0, i = this.magic.length;
for (;;) {
const c = instr.charCodeAt(i++);
if ( c === 0x20 /* ' ' */ ) { break; }
v += this.digitToVal[c] << l;
l += 6;
}
return v << 2;
}
})();
/******************************************************************************/
// The requests.json.gz file can be downloaded from:
// https://cdn.cliqz.com/adblocking/requests_top500.json.gz
//
// Which is linked from:
// https://whotracks.me/blog/adblockers_performance_study.html
//
// Copy the file into ./tmp/requests.json.gz
//
// If the file is present when you build uBO using `make-[target].sh` from
// the shell, the resulting package will have `./assets/requests.json`, which
// will be looked-up by the method below to launch a benchmark session.
//
// From uBO's dev console, launch the benchmark:
// µBlock.staticNetFilteringEngine.benchmark();
//
// The advanced setting `consoleLogLevel` must be set to `info` to see the
// results in uBO's dev console, see:
// https://github.com/gorhill/uBlock/wiki/Advanced-settings#consoleloglevel
//
// The usual browser dev tools can be used to obtain useful profiling
// data, i.e. start the profiler, call the benchmark method from the
// console, then stop the profiler when it completes.
//
// Keep in mind that the measurements at the blog post above where obtained
// with ONLY EasyList. The CPU reportedly used was:
// https://www.cpubenchmark.net/cpu.php?cpu=Intel+Core+i7-6600U+%40+2.60GHz&id=2608
//
// Rename ./tmp/requests.json.gz to something else if you no longer want
// ./assets/requests.json in the build.
2019-09-11 14:08:30 +02:00
µBlock.loadBenchmarkDataset = (( ) => {
let datasetPromise;
let ttlTimer;
return function() {
if ( ttlTimer !== undefined ) {
clearTimeout(ttlTimer);
ttlTimer = undefined;
}
vAPI.setTimeout(( ) => {
ttlTimer = undefined;
datasetPromise = undefined;
}, 60000);
if ( datasetPromise !== undefined ) {
return datasetPromise;
}
console.info(`Loading benchmark dataset...`);
const url = vAPI.getURL('/assets/requests.json');
datasetPromise = µBlock.assets.fetchText(url).then(details => {
console.info(`Parsing benchmark dataset...`);
const requests = [];
const lineIter = new µBlock.LineIterator(details.content);
while ( lineIter.eot() === false ) {
let request;
try {
request = JSON.parse(lineIter.next());
} catch(ex) {
}
if ( request instanceof Object === false ) { continue; }
if ( !request.frameUrl || !request.url ) { continue; }
requests.push(request);
}
return requests;
}).catch(details => {
console.info(`Not found: ${details.url}`);
datasetPromise = undefined;
});
return datasetPromise;
};
})();
/******************************************************************************/
µBlock.fireDOMEvent = function(name) {
if (
window instanceof Object &&
window.dispatchEvent instanceof Function &&
window.CustomEvent instanceof Function
) {
window.dispatchEvent(new CustomEvent(name));
}
};
/******************************************************************************/
µBlock.getMessageSenderDetails = function(sender) {
const r = {};
if ( sender instanceof Object ) {
r.url = sender.url;
r.frameId = sender.frameId;
const tab = sender.tab;
if ( tab instanceof Object ) {
r.tabId = tab.id;
}
}
return r;
};