Optimization for FlatNetworkFilterList.

Author: boocmp

src/filters/fb_network.rs

@@ -21,6 +21,7 @@ pub struct FlatNetworkFiltersListBuilder<'a> {
     filters: Vec<WIPOffset<fb::NetworkFilter<'a>>>,
 
     unique_domains: Vec<Hash>,
+    unique_domains_map: HashMap<Hash, u16>,
 }
 
 impl<'a> FlatNetworkFiltersListBuilder<'a> {
@@ -29,34 +30,31 @@ impl<'a> FlatNetworkFiltersListBuilder<'a> {
             builder: flatbuffers::FlatBufferBuilder::new(),
             filters: vec![],
             unique_domains: vec![],
+            unique_domains_map: HashMap::new(),
         }
     }
 
-    fn get_or_insert(arr: &mut Vec<Hash>, h: Hash) -> u16 {
-        if let Some(index) = arr.iter().position(|&x| x == h) {
-            u16::try_from(index).expect("< u16 max")
-        } else {
-            arr.push(h);
-            u16::try_from(arr.len() - 1).expect("< u16 max")
+    fn get_or_insert(&mut self, h: &Hash) -> u16 {
+        if let Some(&index) = self.unique_domains_map.get(h) {
+            return index;
         }
+        let index = self.unique_domains.len() as u16;
+        self.unique_domains.push(*h);
+        self.unique_domains_map.insert(*h, index);
+        return index;
     }
 
     pub fn add(&mut self, network_filter: &NetworkFilter) -> u32 {
         let opt_domains = network_filter.opt_domains.as_ref().map(|v| {
-            let mut o: Vec<u16> = v
-                .into_iter()
-                .map(|x| Self::get_or_insert(&mut self.unique_domains, *x))
-                .collect();
+            let mut o: Vec<u16> = v.iter().map(|x| self.get_or_insert(x)).collect();
+
             o.sort_unstable();
             o.dedup();
             self.builder.create_vector(&o)
         });
 
         let opt_not_domains = network_filter.opt_not_domains.as_ref().map(|v| {
-            let mut o: Vec<u16> = v
-                .into_iter()
-                .map(|x| Self::get_or_insert(&mut self.unique_domains, *x))
-                .collect();
+            let mut o: Vec<u16> = v.iter().map(|x| self.get_or_insert(x)).collect();
             o.sort_unstable();
             o.dedup();
             self.builder.create_vector(&o)
@@ -65,23 +63,23 @@ impl<'a> FlatNetworkFiltersListBuilder<'a> {
         let modifier_option = network_filter
             .modifier_option
             .as_ref()
-            .map(|s| self.builder.create_shared_string(&s));
+            .map(|s| self.builder.create_string(&s));
 
         let hostname = network_filter
             .hostname
             .as_ref()
-            .map(|s| self.builder.create_shared_string(&s));
+            .map(|s| self.builder.create_string(&s));
 
         let tag = network_filter
             .tag
             .as_ref()
-            .map(|s| self.builder.create_shared_string(&s));
+            .map(|s| self.builder.create_string(&s));
 
         let patterns = if network_filter.filter.iter().len() > 0 {
             let offsets: Vec<WIPOffset<&str>> = network_filter
                 .filter
                 .iter()
-                .map(|s| self.builder.create_shared_string(s))
+                .map(|s| self.builder.create_string(s))
                 .collect();
             Some(self.builder.create_vector(&offsets))
         } else {

src/network_filter_list.rs

@@ -267,15 +267,59 @@ pub struct FlatNetworkFilterList {
 
 impl NetworkFilterListTrait for FlatNetworkFilterList {
     fn new(filters: Vec<NetworkFilter>, optimize: bool) -> Self {
-        let mut temp_list = NetworkFilterList::new(filters, optimize);
+        // Compute tokens for all filters
+        let filter_tokens: Vec<_> = filters
+            .into_iter()
+            .map(|filter| {
+                let tokens = filter.get_tokens();
+                (filter, tokens)
+            })
+            .collect();
+        // compute the tokens' frequency histogram
+        let (total_number_of_tokens, tokens_histogram) = token_histogram(&filter_tokens);
 
         let mut flat_builder = FlatNetworkFiltersListBuilder::new();
         let mut filter_map = HashMap::<Hash, Vec<u32>>::new();
+        let mut optimizable = HashMap::<Hash, Vec<NetworkFilter>>::new();
+        {
+            for (network_filter, multi_tokens) in filter_tokens {
+                let index = if !optimizer::is_filter_optimizable_by_patterns(&network_filter) {
+                    Some(flat_builder.add(&network_filter))
+                } else {
+                    None
+                };
+
+                for tokens in multi_tokens {
+                    let mut best_token: Hash = 0;
+                    let mut min_count = total_number_of_tokens + 1;
+                    for token in tokens {
+                        match tokens_histogram.get(&token) {
+                            None => {
+                                min_count = 0;
+                                best_token = token
+                            }
+                            Some(&count) if count < min_count => {
+                                min_count = count;
+                                best_token = token
+                            }
+                            _ => {}
+                        }
+                    }
+                    if let Some(index) = index {
+                        insert_dup(&mut filter_map, best_token, index);
+                    } else {
+                        insert_dup(&mut optimizable, best_token, network_filter.clone());
+                    }
+                } // tokens
+            }
+        }
+
+        for (token, v) in optimizable {
+            let optimized = optimizer::optimize_by_groupping_patterns(v);
 
-        for (key, vec) in temp_list.filter_map.drain() {
-            for filter in vec.into_iter() {
-                let index = flat_builder.add(&(*filter));
-                insert_dup(&mut filter_map, key, index);
+            for filter in optimized {
+                let index = flat_builder.add(&filter);
+                insert_dup(&mut filter_map, token, index);
             }
         }
 

src/optimizer.rs

@@ -10,6 +10,29 @@ trait Optimization {
     fn select(&self, filter: &NetworkFilter) -> bool;
 }
 
+pub fn is_filter_optimizable_by_patterns(filter: &NetworkFilter) -> bool {
+    filter.opt_domains.is_none()
+        && filter.opt_not_domains.is_none()
+        && !filter.is_hostname_anchor()
+        && !filter.is_redirect()
+        && !filter.is_csp()
+}
+
+pub fn optimize_by_groupping_patterns(filters: Vec<NetworkFilter>) -> Vec<NetworkFilter> {
+    let mut optimized: Vec<NetworkFilter> = Vec::new();
+
+    let simple_pattern_group = SimplePatternGroup {};
+    let (mut fused, mut unfused) = apply_optimisation(&simple_pattern_group, filters);
+    optimized.append(&mut fused);
+
+    // Append whatever is still left unfused
+    optimized.append(&mut unfused);
+
+    // Re-sort the list, now that the order has been perturbed
+    optimized.sort_by_key(|f| f.id);
+    optimized
+}
+
 /// Fuse `NetworkFilter`s together by applying optimizations sequentially.
 pub fn optimize(filters: Vec<NetworkFilter>) -> Vec<NetworkFilter> {
     let mut optimized: Vec<NetworkFilter> = Vec::new();
@@ -129,11 +152,7 @@ impl Optimization for SimplePatternGroup {
         format!("{:b}:{:?}", filter.mask, filter.is_complete_regex())
     }
     fn select(&self, filter: &NetworkFilter) -> bool {
-        filter.opt_domains.is_none()
-            && filter.opt_not_domains.is_none()
-            && !filter.is_hostname_anchor()
-            && !filter.is_redirect()
-            && !filter.is_csp()
+        is_filter_optimizable_by_patterns(filter)
     }
 }