Allow to customize how regexes are provided

marmeladema · marmeladema · commit af100dc28a2a · 2025-09-02T09:01:12.000+02:00
diff --git a/engine/src/ast/field_expr.rs b/engine/src/ast/field_expr.rs
@@ -11,7 +11,7 @@ use crate::{
     filter::CompiledExpr,
     lex::{Lex, LexErrorKind, LexResult, LexWith, expect, skip_space, span},
     range_set::RangeSet,
-    rhs_types::{Bytes, ExplicitIpRange, ListName, Regex, Wildcard},
+    rhs_types::{Bytes, ExplicitIpRange, ListName, RegexExpr, Wildcard},
     scheme::{Field, Identifier, List},
     searcher::{EmptySearcher, TwoWaySearcher},
     strict_partial_ord::StrictPartialOrd,
@@ -152,7 +152,7 @@ pub enum ComparisonOpExpr {
 
     /// "matches / ~" comparison
     #[serde(serialize_with = "serialize_matches")]
-    Matches(Regex),
+    Matches(RegexExpr),
 
     /// "wildcard" comparison
     #[serde(serialize_with = "serialize_wildcard")]
@@ -205,7 +205,7 @@ fn serialize_contains<S: Serializer>(rhs: &Bytes, ser: S) -> Result<S::Ok, S::Er
     serialize_op_rhs("Contains", rhs, ser)
 }
 
-fn serialize_matches<S: Serializer>(rhs: &Regex, ser: S) -> Result<S::Ok, S::Error> {
+fn serialize_matches<S: Serializer>(rhs: &RegexExpr, ser: S) -> Result<S::Ok, S::Error> {
     serialize_op_rhs("Matches", rhs, ser)
 }
 
@@ -376,7 +376,7 @@ impl ComparisonExpr {
                         (ComparisonOpExpr::Contains(bytes), input)
                     }
                     BytesOp::Matches => {
-                        let (regex, input) = Regex::lex_with(input, parser)?;
+                        let (regex, input) = RegexExpr::lex_with(input, parser)?;
                         (ComparisonOpExpr::Matches(regex), input)
                     }
                     BytesOp::Wildcard => {
@@ -688,7 +688,9 @@ impl Expr for ComparisonExpr {
 
                 search!(TwoWaySearcher::new(bytes))
             }
-            ComparisonOpExpr::Matches(regex) => lhs.compile_with(compiler, false, regex),
+            ComparisonOpExpr::Matches(regex) => {
+                lhs.compile_with(compiler, false, regex.into_regex())
+            }
             ComparisonOpExpr::Wildcard(wildcard) => lhs.compile_with(compiler, false, wildcard),
             ComparisonOpExpr::StrictWildcard(wildcard) => {
                 lhs.compile_with(compiler, false, wildcard)
@@ -2806,7 +2808,12 @@ mod tests {
 
         // Matches operator
         let parser = FilterParser::new(&SCHEME);
-        let r = Regex::new("a.b", RegexFormat::Literal, parser.settings()).unwrap();
+        let r = RegexExpr::new(
+            "a.b",
+            RegexFormat::Literal,
+            &parser.settings().regex_provider,
+        )
+        .unwrap();
         let expr = assert_ok!(
             parser.lex_as("http.host matches r###\"a.b\"###"),
             ComparisonExpr {
diff --git a/engine/src/ast/parse.rs b/engine/src/ast/parse.rs
@@ -1,11 +1,14 @@
 use super::{FilterAst, FilterValueAst};
 use crate::{
+    RegexProvider,
     lex::{LexErrorKind, LexResult, LexWith, complete},
+    rhs_types::DefaultRegexProvider,
     scheme::Scheme,
 };
 use std::cmp::{max, min};
 use std::error::Error;
 use std::fmt::{self, Debug, Display, Formatter};
+use std::sync::{Arc, LazyLock};
 
 /// An opaque filter parsing error associated with the original input.
 ///
@@ -96,28 +99,37 @@ impl Display for ParseError<'_> {
     }
 }
 
+static DEFAULT_REGEX_PROVIDER: LazyLock<Arc<dyn RegexProvider>> =
+    LazyLock::new(|| Arc::new(DefaultRegexProvider::default()));
+
 /// Parser settings.
-#[derive(Clone, Debug, PartialEq, Eq)]
+#[derive(Clone, Debug)]
 pub struct ParserSettings {
-    /// Approximate size of the cache used by the DFA of a regex.
-    /// Default: 10MB
-    pub regex_dfa_size_limit: usize,
-    /// Approximate size limit of the compiled regular expression.
-    /// Default: 2MB
-    pub regex_compiled_size_limit: usize,
+    /// Regex provider.
+    pub regex_provider: Arc<dyn RegexProvider>,
     /// Maximum number of star metacharacters allowed in a wildcard.
     /// Default: unlimited
     pub wildcard_star_limit: usize,
 }
 
+impl PartialEq for ParserSettings {
+    fn eq(&self, other: &Self) -> bool {
+        let Self {
+            regex_provider,
+            wildcard_star_limit,
+        } = self;
+        Arc::ptr_eq(regex_provider, &other.regex_provider)
+            && *wildcard_star_limit == other.wildcard_star_limit
+    }
+}
+
+impl Eq for ParserSettings {}
+
 impl Default for ParserSettings {
     #[inline]
     fn default() -> Self {
         Self {
-            // Default value extracted from the regex crate.
-            regex_compiled_size_limit: 10 * (1 << 20),
-            // Default value extracted from the regex crate.
-            regex_dfa_size_limit: 2 * (1 << 20),
+            regex_provider: DEFAULT_REGEX_PROVIDER.clone(),
             wildcard_star_limit: usize::MAX,
         }
     }
@@ -176,30 +188,6 @@ impl<'s> FilterParser<'s> {
         &self.settings
     }
 
-    /// Set the approximate size limit of the compiled regular expression.
-    #[inline]
-    pub fn regex_set_compiled_size_limit(&mut self, regex_compiled_size_limit: usize) {
-        self.settings.regex_compiled_size_limit = regex_compiled_size_limit;
-    }
-
-    /// Get the approximate size limit of the compiled regular expression.
-    #[inline]
-    pub fn regex_get_compiled_size_limit(&self) -> usize {
-        self.settings.regex_compiled_size_limit
-    }
-
-    /// Set the approximate size of the cache used by the DFA of a regex.
-    #[inline]
-    pub fn regex_set_dfa_size_limit(&mut self, regex_dfa_size_limit: usize) {
-        self.settings.regex_dfa_size_limit = regex_dfa_size_limit;
-    }
-
-    /// Get the approximate size of the cache used by the DFA of a regex.
-    #[inline]
-    pub fn regex_get_dfa_size_limit(&self) -> usize {
-        self.settings.regex_dfa_size_limit
-    }
-
     /// Set the maximum number of star metacharacters allowed in a wildcard.
     #[inline]
     pub fn wildcard_set_star_limit(&mut self, wildcard_star_limit: usize) {
diff --git a/engine/src/lib.rs b/engine/src/lib.rs
@@ -112,8 +112,8 @@ pub use self::{
         panic_catcher_get_backtrace, panic_catcher_set_fallback_mode, panic_catcher_set_hook,
     },
     rhs_types::{
-        Bytes, BytesFormat, ExplicitIpRange, IntRange, IpCidr, IpRange, Regex, RegexError,
-        RegexFormat,
+        Bytes, BytesFormat, DefaultRegexProvider, ExplicitIpRange, IntRange, IpCidr, IpRange,
+        Regex, RegexError, RegexExpr, RegexFormat, RegexProvider,
     },
     scheme::{
         Field, FieldIndex, FieldRedefinitionError, FieldRef, Function, FunctionRedefinitionError,
@@ -125,3 +125,6 @@ pub use self::{
         TypeMismatchError,
     },
 };
+
+#[cfg(feature = "regex")]
+pub use self::rhs_types::RegexSettings;
diff --git a/engine/src/rhs_types/mod.rs b/engine/src/rhs_types/mod.rs
@@ -16,6 +16,11 @@ pub use self::{
     ip::{ExplicitIpRange, IpCidr, IpRange},
     list::ListName,
     map::UninhabitedMap,
-    regex::{Error as RegexError, Regex, RegexFormat},
+    regex::{
+        DefaultRegexProvider, Error as RegexError, Regex, RegexExpr, RegexFormat, RegexProvider,
+    },
     wildcard::{Wildcard, WildcardError},
 };
+
+#[cfg(feature = "regex")]
+pub use self::regex::RegexSettings;
diff --git a/engine/src/rhs_types/regex/imp_real.rs b/engine/src/rhs_types/regex/imp_real.rs
@@ -1,19 +1,51 @@
-use regex_automata::MatchKind;
-
-use super::Error;
-use crate::{ParserSettings, RegexFormat};
-use std::ops::Deref;
+use super::{Error, Regex};
+use crate::RegexProvider;
 use std::sync::Arc;
 
-/// Wrapper around [`regex_automata::meta::Regex`]
-#[derive(Clone)]
-pub struct Regex {
-    pattern: Arc<str>,
-    regex: regex_automata::meta::Regex,
-    format: RegexFormat,
+pub(crate) type MetaRegex = regex_automata::meta::Regex;
+
+impl Regex for MetaRegex {
+    #[inline]
+    fn is_match(&self, input: &[u8]) -> bool {
+        MetaRegex::is_match(self, input)
+    }
+}
+
+/// Regex settings.
+#[derive(Clone, Debug, PartialEq, Eq)]
+pub struct RegexSettings {
+    /// Approximate size of the cache used by the DFA of a regex.
+    /// Default: 10MB
+    pub dfa_size_limit: usize,
+    /// Approximate size limit of the compiled regular expression.
+    /// Default: 2MB
+    pub compiled_size_limit: usize,
+}
+
+impl Default for RegexSettings {
+    #[inline]
+    fn default() -> Self {
+        Self {
+            // Default value extracted from the regex crate.
+            compiled_size_limit: 10 * (1 << 20),
+            // Default value extracted from the regex crate.
+            dfa_size_limit: 2 * (1 << 20),
+        }
+    }
+}
+
+/// Default regex provider.
+#[derive(Debug, Default)]
+pub struct DefaultRegexProvider {
+    settings: RegexSettings,
 }
 
-impl Regex {
+impl DefaultRegexProvider {
+    /// Creates a new default regex provider.
+    pub const fn new(settings: RegexSettings) -> Self {
+        Self { settings }
+    }
+
     /// Retrieves the syntax configuration that will be used to build the regex.
     #[inline]
     fn syntax_config() -> regex_automata::util::syntax::Config {
@@ -24,32 +56,25 @@ impl Regex {
 
     /// Retrieves the meta configuration that will be used to build the regex.
     #[inline]
-    fn meta_config(settings: &ParserSettings) -> regex_automata::meta::Config {
+    fn meta_config(settings: &RegexSettings) -> regex_automata::meta::Config {
         regex_automata::meta::Config::new()
-            .match_kind(MatchKind::LeftmostFirst)
+            .match_kind(regex_automata::MatchKind::LeftmostFirst)
             .utf8_empty(false)
             .dfa(false)
-            .nfa_size_limit(Some(settings.regex_compiled_size_limit))
-            .onepass_size_limit(Some(settings.regex_compiled_size_limit))
-            .dfa_size_limit(Some(settings.regex_compiled_size_limit))
-            .hybrid_cache_capacity(settings.regex_dfa_size_limit)
+            .nfa_size_limit(Some(settings.compiled_size_limit))
+            .onepass_size_limit(Some(settings.compiled_size_limit))
+            .dfa_size_limit(Some(settings.compiled_size_limit))
+            .hybrid_cache_capacity(settings.dfa_size_limit)
     }
+}
 
-    /// Compiles a regular expression.
-    pub fn new(
-        pattern: &str,
-        format: RegexFormat,
-        settings: &ParserSettings,
-    ) -> Result<Self, Error> {
+impl RegexProvider for DefaultRegexProvider {
+    fn lookup_regex(&self, pattern: &str) -> Result<Arc<dyn Regex>, Error> {
         ::regex_automata::meta::Builder::new()
-            .configure(Self::meta_config(settings))
+            .configure(Self::meta_config(&self.settings))
             .syntax(Self::syntax_config())
             .build(pattern)
-            .map(|regex| Regex {
-                pattern: Arc::from(pattern),
-                regex,
-                format,
-            })
+            .map(|re| Arc::new(re) as Arc<dyn Regex>)
             .map_err(|err| {
                 if let Some(limit) = err.size_limit() {
                     Error::CompiledTooBig(limit)
@@ -60,45 +85,21 @@ impl Regex {
                 }
             })
     }
-
-    /// Returns the pattern of this regex.
-    #[inline]
-    pub fn as_str(&self) -> &str {
-        &self.pattern
-    }
-
-    /// Returns the format used by the pattern.
-    #[inline]
-    pub fn format(&self) -> RegexFormat {
-        self.format
-    }
-}
-
-impl From<Regex> for regex_automata::meta::Regex {
-    #[inline]
-    fn from(regex: Regex) -> Self {
-        regex.regex
-    }
-}
-
-impl Deref for Regex {
-    type Target = regex_automata::meta::Regex;
-
-    #[inline]
-    fn deref(&self) -> &Self::Target {
-        &self.regex
-    }
 }
 
 #[test]
 fn test_compiled_size_limit() {
+    use super::{DefaultRegexProvider, RegexSettings};
+    use crate::{RegexExpr, RegexFormat};
+
     const COMPILED_SIZE_LIMIT: usize = 1024 * 1024;
-    let settings = ParserSettings {
-        regex_compiled_size_limit: COMPILED_SIZE_LIMIT,
+    let settings = RegexSettings {
+        compiled_size_limit: COMPILED_SIZE_LIMIT,
         ..Default::default()
     };
+    let regex_provider = DefaultRegexProvider::new(settings);
     assert_eq!(
-        Regex::new(".{4079,65535}", RegexFormat::Literal, &settings),
+        RegexExpr::new(".{4079,65535}", RegexFormat::Literal, &regex_provider),
         Err(Error::CompiledTooBig(COMPILED_SIZE_LIMIT))
     );
 }
diff --git a/engine/src/rhs_types/regex/imp_stub.rs b/engine/src/rhs_types/regex/imp_stub.rs
@@ -1,40 +1,21 @@
-use thiserror::Error;
-
 use crate::{FilterParser, RegexFormat};
+use thiserror::Error;
 
-/// Dummy regex error.
-#[derive(Debug, PartialEq, Error)]
-pub enum Error {}
-
-/// Dummy regex wrapper that can only store a pattern
-/// but not actually be used for matching.
-#[derive(Clone)]
-pub struct Regex {
-    pattern: String,
-    format: RegexFormat,
-}
-
-impl Regex {
-    /// Creates a new dummy regex.
-    pub fn new(pattern: &str, format: RegexFormat, _: &FilterParser<'_>) -> Result<Self, Error> {
-        Ok(Self {
-            pattern: pattern.to_string(),
-            format,
-        })
-    }
+pub(crate) struct StubRegex {}
 
+impl Regex for StubRegex {
     /// Not implemented and will panic if called.
-    pub fn is_match(&self, _text: &[u8]) -> bool {
+    fn is_match(&self, _text: &[u8]) -> bool {
         unimplemented!("Engine was built without regex support")
     }
+}
 
-    /// Returns the original string of this dummy regex wrapper.
-    pub fn as_str(&self) -> &str {
-        self.pattern.as_str()
-    }
+/// Default regex provider.
+#[derive(Debug, Default)]
+pub struct DefaultRegexProvider;
 
-    /// Returns the format behind the regex
-    pub fn format(&self) -> RegexFormat {
-        self.format
+impl RegexProvider for DefaultRegexProvider {
+    fn lookup(&self, pattern: &str) -> Result<Arc<dyn AsRegex>, Error> {
+        Ok(Arc::new(StubRegex {}))
     }
 }
diff --git a/engine/src/rhs_types/regex/mod.rs b/engine/src/rhs_types/regex/mod.rs
