Merge pull request #393 from Ambyjkl/regex-optimization

Disable unicode in network filter regex

Author: antonok-edm

src/engine.rs

@@ -768,14 +768,15 @@ mod tests {
                                        "script").unwrap();
             assert!(engine.check_network_request(&request).matched);
         }*/
-        {
+        // fails - unicode not supported in network filter
+        /*{
             let engine = Engine::from_rules_debug([r#"/tesT߶/$domain=example.com"#], Default::default());
             let request = Request::new("https://example.com/tesT߶",
                                        "https://example.com",
                                        "script").unwrap();
             assert!(engine.check_network_request(&request).matched);
-        }
-        // fails - punycoded domain
+        }*/
+        // fails - unicode not supported in network filter
         /*{
             let engine = Engine::from_rules_debug([r#"/tesT߶/$domain=example.com"#], Default::default());
             let request = Request::new("https://example-tesT߶.com/tesT",

src/filters/network.rs

@@ -3,7 +3,10 @@
 
 use memchr::{memchr as find_char, memmem, memrchr as find_char_reverse};
 use once_cell::sync::Lazy;
-use regex::{Regex, RegexSet};
+use regex::{
+    bytes::Regex as BytesRegex, bytes::RegexBuilder as BytesRegexBuilder,
+    bytes::RegexSet as BytesRegexSet, bytes::RegexSetBuilder as BytesRegexSetBuilder, Regex,
+};
 use serde::{Deserialize, Serialize};
 use thiserror::Error;
 
@@ -180,8 +183,8 @@ impl From<&request::RequestType> for NetworkFilterMask {
 
 #[derive(Debug, Clone)]
 pub enum CompiledRegex {
-    Compiled(Regex),
-    CompiledSet(RegexSet),
+    Compiled(BytesRegex),
+    CompiledSet(BytesRegexSet),
     MatchAll,
     RegexParsingError(regex::Error),
 }
@@ -191,11 +194,11 @@ impl CompiledRegex {
         match &self {
             CompiledRegex::MatchAll => true, // simple case for matching everything, e.g. for empty filter
             CompiledRegex::RegexParsingError(_e) => false, // no match if regex didn't even compile
-            CompiledRegex::Compiled(r) => r.is_match(pattern),
+            CompiledRegex::Compiled(r) => r.is_match(pattern.as_bytes()),
             CompiledRegex::CompiledSet(r) => {
                 // let matches: Vec<_> = r.matches(pattern).into_iter().collect();
                 // println!("Matching {} against RegexSet: {:?}", pattern, matches);
-                r.is_match(pattern)
+                r.is_match(pattern.as_bytes())
             }
         }
     }
@@ -1235,15 +1238,15 @@ pub(crate) fn compile_regex(
         CompiledRegex::MatchAll
     } else if escaped_patterns.len() == 1 {
         let pattern = &escaped_patterns[0];
-        match Regex::new(pattern) {
+        match BytesRegexBuilder::new(pattern).unicode(false).build() {
             Ok(compiled) => CompiledRegex::Compiled(compiled),
             Err(e) => {
                 // println!("Regex parsing failed ({:?})", e);
                 CompiledRegex::RegexParsingError(e)
             }
         }
     } else {
-        match RegexSet::new(escaped_patterns) {
+        match BytesRegexSetBuilder::new(escaped_patterns).unicode(false).build() {
             Ok(compiled) => CompiledRegex::CompiledSet(compiled),
             Err(e) => CompiledRegex::RegexParsingError(e),
         }

src/optimizer.rs

@@ -209,7 +209,7 @@ mod optimization_tests_pattern_group {
     use crate::lists;
     use crate::regex_manager::RegexManager;
     use crate::request::Request;
-    use regex::RegexSet;
+    use regex::bytes::RegexSetBuilder as BytesRegexSetBuilder;
 
     fn check_regex_match(regex: &CompiledRegex, pattern: &str, matches: bool) {
         let is_match = regex.is_match(pattern);
@@ -244,13 +244,15 @@ mod optimization_tests_pattern_group {
 
     #[test]
     fn regex_set_works() {
-        let regex_set = RegexSet::new(&[
+        let regex_set = BytesRegexSetBuilder::new(&[
             r"/static/ad\.",
             "/static/ad-",
             "/static/ad/.*",
             "/static/ads/.*",
             "/static/adv/.*",
-        ]);
+        ])
+        .unicode(false)
+        .build();
 
         let fused_regex = CompiledRegex::CompiledSet(regex_set.unwrap());
         assert!(matches!(fused_regex, CompiledRegex::CompiledSet(_)));