pre-0.12.x => master

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "adblock"
-version = "0.11.1"
+version = "0.12.0"
 authors = ["Anton Lazarev <alazarev@brave.com>", "Andrius Aucinas"]
 edition = "2021"
 
@@ -39,6 +39,7 @@ rustc-hash = { version = "1.1.0", default-features = false }
 memchr = "2.4"
 base64 = "0.22"
 rmp-serde = "0.15"
+arrayvec = "0.7"
 cssparser = { version = "0.34", optional = true }
 selectors = { version = "0.26", optional = true }
 precomputed-hash = "0.1"

js/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "adblock-rs"
-version = "0.11.1"
+version = "0.12.0"
 authors = ["Anton Lazarev <alazarev@brave.com>", "Andrius Aucinas"]
 edition = "2021"
 license = "MPL-2.0"

package.json

@@ -1,6 +1,6 @@
 {
   "name": "adblock-rs",
-  "version": "0.11.1",
+  "version": "0.12.0",
   "description": "Very fast, Rust-based, native implementation of ad-blocker engine for Node",
   "keywords": [
     "adblock",

src/blocker.rs

@@ -79,6 +79,11 @@ pub struct Blocker {
     pub(crate) filter_data_context: FilterDataContextRef,
 }
 
+#[cfg(feature = "single-thread")]
+pub(crate) type RegexManagerRef<'a> = std::cell::RefMut<'a, RegexManager>;
+#[cfg(not(feature = "single-thread"))]
+pub(crate) type RegexManagerRef<'a> = std::sync::MutexGuard<'a, RegexManager>;
+
 impl Blocker {
     /// Decide if a network request (usually from WebRequest API) should be
     /// blocked, redirected or allowed.
@@ -130,31 +135,36 @@ impl Blocker {
         self.get_list(NetworkFilterListId::TaggedFiltersAll)
     }
 
-    #[cfg(feature = "single-thread")]
-    fn borrow_regex_manager(&self) -> std::cell::RefMut<'_, RegexManager> {
+    /// Borrow mutable reference to the regex manager for the ['Blocker`].
+    /// Only one caller can borrow the regex manager at a time.
+    pub(crate) fn borrow_regex_manager(&self) -> RegexManagerRef<'_> {
+        #[cfg(feature = "single-thread")]
         #[allow(unused_mut)]
         let mut manager = self.regex_manager.borrow_mut();
+        #[cfg(not(feature = "single-thread"))]
+        let mut manager = self.regex_manager.lock().unwrap();
 
         #[cfg(not(target_arch = "wasm32"))]
         manager.update_time();
 
         manager
     }
 
-    #[cfg(not(feature = "single-thread"))]
-    fn borrow_regex_manager(&self) -> std::sync::MutexGuard<'_, RegexManager> {
-        let mut manager = self.regex_manager.lock().unwrap();
-        manager.update_time();
-        manager
-    }
-
     pub fn check_generic_hide(&self, hostname_request: &Request) -> bool {
         let mut regex_manager = self.borrow_regex_manager();
         self.generic_hide()
             .check(hostname_request, &HashSet::new(), &mut regex_manager)
             .is_some()
     }
 
+    #[cfg(test)]
+    pub(crate) fn check_exceptions(&self, request: &Request) -> bool {
+        let mut regex_manager = self.borrow_regex_manager();
+        self.exceptions()
+            .check(request, &HashSet::new(), &mut regex_manager)
+            .is_some()
+    }
+
     pub fn check_parameterised(
         &self,
         request: &Request,

src/engine.rs

@@ -139,6 +139,11 @@ impl Engine {
         self.blocker.check(request, &self.resources)
     }
 
+    #[cfg(test)]
+    pub(crate) fn check_network_request_exceptions(&self, request: &Request) -> bool {
+        self.blocker.check_exceptions(request)
+    }
+
     pub fn check_network_request_subset(
         &self,
         request: &Request,
@@ -266,6 +271,11 @@ impl Engine {
         self.blocker.set_regex_discard_policy(new_discard_policy);
     }
 
+    #[cfg(test)]
+    pub fn borrow_regex_manager(&self) -> crate::blocker::RegexManagerRef<'_> {
+        self.blocker.borrow_regex_manager()
+    }
+
     #[cfg(feature = "debug-info")]
     pub fn discard_regex(&mut self, regex_id: u64) {
         self.blocker.discard_regex(regex_id);

src/filters/fb_network.rs

@@ -171,9 +171,4 @@ impl NetworkMatchable for FlatNetworkFilter<'_> {
             regex_manager,
         )
     }
-
-    #[cfg(test)]
-    fn matches_test(&self, request: &Request) -> bool {
-        self.matches(request, &mut RegexManager::default())
-    }
 }

src/filters/fb_network_builder.rs

@@ -7,6 +7,7 @@ use flatbuffers::WIPOffset;
 use crate::filters::fb_builder::EngineFlatBuilder;
 use crate::filters::network::{FilterTokens, NetworkFilter};
 use crate::filters::token_selector::TokenSelector;
+use crate::utils::TokensBuffer;
 
 use crate::filters::network::NetworkFilterMaskHelper;
 use crate::flatbuffers::containers::flat_multimap::FlatMultiMapBuilder;
@@ -134,6 +135,7 @@ impl<'a> FlatSerialize<'a, EngineFlatBuilder<'a>> for NetworkFilterListBuilder {
         let mut optimizable = HashMap::<ShortHash, Vec<NetworkFilter>>::new();
 
         let mut token_frequencies = TokenSelector::new(rule_list.filters.len());
+        let mut tokens_buffer = TokensBuffer::default();
 
         {
             for network_filter in rule_list.filters {
@@ -157,7 +159,7 @@ impl<'a> FlatSerialize<'a, EngineFlatBuilder<'a>> for NetworkFilterListBuilder {
                     }
                 };
 
-                let multi_tokens = network_filter.get_tokens_optimized();
+                let multi_tokens = network_filter.get_tokens(&mut tokens_buffer);
                 match multi_tokens {
                     FilterTokens::Empty => {
                         // No tokens, add to fallback bucket (token 0)
@@ -171,7 +173,7 @@ impl<'a> FlatSerialize<'a, EngineFlatBuilder<'a>> for NetworkFilterListBuilder {
                         }
                     }
                     FilterTokens::Other(tokens) => {
-                        let best_token = token_frequencies.select_least_used_token(&tokens);
+                        let best_token = token_frequencies.select_least_used_token(tokens);
                         token_frequencies.record_usage(best_token);
                         store_filter(best_token);
                     }

src/filters/network.rs

@@ -15,9 +15,7 @@ use crate::filters::abstract_network::{
 use crate::lists::ParseOptions;
 use crate::regex_manager::RegexManager;
 use crate::request;
-use crate::utils::{self, Hash};
-
-pub(crate) const TOKENS_BUFFER_SIZE: usize = 200;
+use crate::utils::{self, Hash, TokensBuffer};
 
 /// For now, only support `$removeparam` with simple alphanumeric/dash/underscore patterns.
 static VALID_PARAM: Lazy<Regex> = Lazy::new(|| Regex::new(r"^[a-zA-Z0-9_\-]+$").unwrap());
@@ -312,10 +310,10 @@ pub enum FilterPart {
 }
 
 #[derive(Debug, PartialEq)]
-pub enum FilterTokens {
+pub(crate) enum FilterTokens<'a> {
     Empty,
-    OptDomains(Vec<Hash>),
-    Other(Vec<Hash>),
+    OptDomains(&'a [Hash]),
+    Other(&'a [Hash]),
 }
 
 pub struct FilterPartIterator<'a> {
@@ -883,19 +881,11 @@ impl NetworkFilter {
         )
     }
 
-    #[deprecated(since = "0.11.1", note = "use get_tokens_optimized instead")]
-    pub fn get_tokens(&self) -> Vec<Vec<Hash>> {
-        match self.get_tokens_optimized() {
-            FilterTokens::OptDomains(domains) => {
-                domains.into_iter().map(|domain| vec![domain]).collect()
-            }
-            FilterTokens::Other(tokens) => vec![tokens],
-            FilterTokens::Empty => vec![],
-        }
-    }
-
-    pub fn get_tokens_optimized(&self) -> FilterTokens {
-        let mut tokens: Vec<Hash> = Vec::with_capacity(TOKENS_BUFFER_SIZE);
+    pub(crate) fn get_tokens<'a>(
+        &'a self,
+        tokens_buffer: &'a mut TokensBuffer,
+    ) -> FilterTokens<'a> {
+        tokens_buffer.clear();
 
         // If there is only one domain and no domain negation, we also use this
         // domain as a token.
@@ -905,7 +895,7 @@ impl NetworkFilter {
         {
             if let Some(domains) = self.opt_domains.as_ref() {
                 if let Some(domain) = domains.first() {
-                    tokens.push(*domain)
+                    tokens_buffer.push(*domain);
                 }
             }
         }
@@ -918,7 +908,7 @@ impl NetworkFilter {
                         (self.is_plain() || self.is_regex()) && !self.is_right_anchor();
                     let skip_first_token = self.is_right_anchor();
 
-                    utils::tokenize_filter_to(f, skip_first_token, skip_last_token, &mut tokens);
+                    utils::tokenize_filter_to(f, skip_first_token, skip_last_token, tokens_buffer);
                 }
             }
             FilterPart::AnyOf(_) => (), // across AnyOf set of filters no single token is guaranteed to match to a request
@@ -928,45 +918,55 @@ impl NetworkFilter {
         // Append tokens from hostname, if any
         if !self.mask.contains(NetworkFilterMask::IS_HOSTNAME_REGEX) {
             if let Some(hostname) = self.hostname.as_ref() {
-                utils::tokenize_to(hostname, &mut tokens);
+                utils::tokenize_to(hostname, tokens_buffer);
             }
         } else if let Some(hostname) = self.hostname.as_ref() {
             // Find last dot to tokenize the prefix
             let last_dot_pos = hostname.rfind('.');
             if let Some(last_dot_pos) = last_dot_pos {
-                utils::tokenize_to(&hostname[..last_dot_pos], &mut tokens);
+                utils::tokenize_to(&hostname[..last_dot_pos], tokens_buffer);
             }
         }
 
-        if tokens.is_empty() && self.mask.contains(NetworkFilterMask::IS_REMOVEPARAM) {
+        if tokens_buffer.is_empty() && self.mask.contains(NetworkFilterMask::IS_REMOVEPARAM) {
             if let Some(removeparam) = &self.modifier_option {
                 if VALID_PARAM.is_match(removeparam) {
-                    utils::tokenize_to(&removeparam.to_ascii_lowercase(), &mut tokens);
+                    utils::tokenize_to(&removeparam.to_ascii_lowercase(), tokens_buffer);
                 }
             }
         }
 
         // If we got no tokens for the filter/hostname part, then we will dispatch
         // this filter in multiple buckets based on the domains option.
-        if tokens.is_empty() && self.opt_domains.is_some() && self.opt_not_domains.is_none() {
+        if tokens_buffer.is_empty() && self.opt_domains.is_some() && self.opt_not_domains.is_none()
+        {
             if let Some(opt_domains) = self.opt_domains.as_ref() {
                 if !opt_domains.is_empty() {
-                    return FilterTokens::OptDomains(opt_domains.clone());
+                    return FilterTokens::OptDomains(opt_domains);
                 }
             }
             FilterTokens::Empty
         } else {
             // Add optional token for protocol
             if self.for_http() && !self.for_https() {
-                tokens.push(utils::fast_hash("http"));
+                tokens_buffer.push(utils::fast_hash("http"));
             } else if self.for_https() && !self.for_http() {
-                tokens.push(utils::fast_hash("https"));
+                tokens_buffer.push(utils::fast_hash("https"));
             }
 
-            // Remake a vector to drop extra capacity.
-            let mut t = Vec::with_capacity(tokens.len());
-            t.extend(tokens);
-            FilterTokens::Other(t)
+            FilterTokens::Other(tokens_buffer.as_slice())
+        }
+    }
+
+    #[cfg(test)]
+    pub(crate) fn matches_test(&self, request: &request::Request) -> bool {
+        let filter_set = crate::FilterSet::new_with_rules(vec![self.clone()], vec![], true);
+        let engine = crate::Engine::from_filter_set(filter_set, true);
+
+        if self.is_exception() {
+            engine.check_network_request_exceptions(request)
+        } else {
+            engine.check_network_request(request).matched
         }
     }
 }
@@ -986,35 +986,8 @@ impl fmt::Display for NetworkFilter {
     }
 }
 
-pub trait NetworkMatchable {
+pub(crate) trait NetworkMatchable {
     fn matches(&self, request: &request::Request, regex_manager: &mut RegexManager) -> bool;
-
-    #[cfg(test)]
-    fn matches_test(&self, request: &request::Request) -> bool;
-}
-
-impl NetworkMatchable for NetworkFilter {
-    fn matches(&self, request: &request::Request, regex_manager: &mut RegexManager) -> bool {
-        use crate::filters::network_matchers::{
-            check_excluded_domains, check_included_domains, check_options, check_pattern,
-        };
-        check_options(self.mask, request)
-            && check_included_domains(self.opt_domains.as_deref(), request)
-            && check_excluded_domains(self.opt_not_domains.as_deref(), request)
-            && check_pattern(
-                self.mask,
-                self.filter.iter(),
-                self.hostname.as_deref(),
-                (self as *const NetworkFilter) as u64,
-                request,
-                regex_manager,
-            )
-    }
-
-    #[cfg(test)]
-    fn matches_test(&self, request: &request::Request) -> bool {
-        self.matches(request, &mut RegexManager::default())
-    }
 }
 
 // ---------------------------------------------------------------------------