Switch to regex-automata

Author: marmeladema

Cargo.lock

@@ -26,7 +26,7 @@ libc = "0.2.42"
 memmem = "0.1.1"
 num_enum = "0.7"
 rand = "0.8"
-regex = { version = "1.3.6" }
+regex-automata = { version = "0.4.9" }
 serde = { version = "1.0.113", features = [ "derive" ] }
 serde_json = "1.0.56"
 sliceslice = "0.4.3"

Cargo.toml

@@ -26,7 +26,7 @@ cidr.workspace = true
 fnv.workspace = true
 memmem.workspace = true
 rand.workspace = true
-regex = { workspace = true, optional = true }
+regex-automata = { workspace = true, optional = true }
 serde.workspace = true
 serde_json.workspace = true
 sliceslice.workspace = true
@@ -38,4 +38,5 @@ criterion.workspace = true
 indoc.workspace = true
 
 [features]
+regex = ["dep:regex-automata"]
 default = [ "regex" ]

engine/Cargo.toml

@@ -1,51 +1,92 @@
-use crate::{ParserSettings, RegexFormat};
+use regex_automata::MatchKind;
 
-pub use regex::Error;
+use super::Error;
+use crate::{ParserSettings, RegexFormat};
+use std::ops::Deref;
+use std::sync::Arc;
 
-/// Wrapper around [`regex::bytes::Regex`]
+/// Wrapper around [`regex_automata::meta::Regex`]
 #[derive(Clone)]
 pub struct Regex {
-    compiled_regex: regex::bytes::Regex,
+    pattern: Arc<str>,
+    regex: regex_automata::meta::Regex,
     format: RegexFormat,
 }
 
 impl Regex {
+    /// Retrieves the syntax configuration that will be used to build the regex.
+    #[inline]
+    pub fn syntax_config() -> regex_automata::util::syntax::Config {
+        regex_automata::util::syntax::Config::new()
+            .unicode(false)
+            .utf8(false)
+    }
+
+    /// Retrieves the meta configuration that will be used to build the regex.
+    #[inline]
+    pub fn meta_config(settings: &ParserSettings) -> regex_automata::meta::Config {
+        regex_automata::meta::Config::new()
+            .match_kind(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)
+    }
+
     /// Compiles a regular expression.
     pub fn new(
         pattern: &str,
         format: RegexFormat,
         settings: &ParserSettings,
     ) -> Result<Self, Error> {
-        ::regex::bytes::RegexBuilder::new(pattern)
-            .unicode(false)
-            .size_limit(settings.regex_compiled_size_limit)
-            .dfa_size_limit(settings.regex_dfa_size_limit)
-            .build()
-            .map(|r| Regex {
-                compiled_regex: r,
+        ::regex_automata::meta::Builder::new()
+            .configure(Self::meta_config(settings))
+            .syntax(Self::syntax_config())
+            .build(pattern)
+            .map(|regex| Regex {
+                pattern: Arc::from(pattern),
+                regex,
                 format,
             })
+            .map_err(|err| {
+                if let Some(limit) = err.size_limit() {
+                    Error::CompiledTooBig(limit)
+                } else if let Some(syntax) = err.syntax_error() {
+                    Error::Syntax(syntax.to_string())
+                } else {
+                    unreachable!()
+                }
+            })
     }
 
-    /// Returns true if and only if the regex matches the string given.
-    pub fn is_match(&self, text: &[u8]) -> bool {
-        self.compiled_regex.is_match(text)
-    }
-
-    /// Returns the original string of this regex.
+    /// Returns the pattern of this regex.
+    #[inline]
     pub fn as_str(&self) -> &str {
-        self.compiled_regex.as_str()
+        &self.pattern
     }
 
-    /// Returns the format behind the regex
+    /// Returns the format used by the pattern.
+    #[inline]
     pub fn format(&self) -> RegexFormat {
         self.format
     }
 }
 
-impl From<Regex> for regex::bytes::Regex {
+impl From<Regex> for regex_automata::meta::Regex {
+    #[inline]
     fn from(regex: Regex) -> Self {
-        regex.compiled_regex
+        regex.regex
+    }
+}
+
+impl Deref for Regex {
+    type Target = regex_automata::meta::Regex;
+
+    #[inline]
+    fn deref(&self) -> &Self::Target {
+        &self.regex
     }
 }
 

engine/src/rhs_types/regex/imp_real.rs

@@ -3,10 +3,9 @@ use crate::rhs_types::bytes::lex_raw_string_as_str;
 use crate::FilterParser;
 use cfg_if::cfg_if;
 use serde::{Serialize, Serializer};
-use std::{
-    fmt::{self, Debug, Formatter},
-    hash::{Hash, Hasher},
-};
+use std::fmt::{self, Debug, Display, Formatter};
+use std::hash::{Hash, Hasher};
+use thiserror::Error;
 
 cfg_if! {
     if #[cfg(feature = "regex")] {
@@ -19,7 +18,7 @@ cfg_if! {
 }
 
 /// RegexFormat describes the format behind the regex
-#[derive(PartialEq, Eq, Copy, Clone)]
+#[derive(PartialEq, Eq, Copy, Clone, Debug)]
 pub enum RegexFormat {
     /// Literal string was used to define the expression
     Literal,
@@ -41,9 +40,20 @@ impl Hash for Regex {
     }
 }
 
+impl Display for Regex {
+    /// Shows the original regular expression.
+    fn fmt(&self, f: &mut Formatter<'_>) -> fmt::Result {
+        write!(f, "{}", self.as_str())
+    }
+}
+
 impl Debug for Regex {
+    /// Shows the original regular expression.
     fn fmt(&self, f: &mut Formatter<'_>) -> fmt::Result {
-        f.write_str(self.as_str())
+        f.debug_struct("Regex")
+            .field("pattern", &self.as_str())
+            .field("format", &self.format())
+            .finish()
     }
 }
 
@@ -120,6 +130,28 @@ impl Serialize for Regex {
     }
 }
 
+/// An error that occurred during parsing or compiling a regular expression.
+#[non_exhaustive]
+#[derive(Clone, Debug, Error, PartialEq)]
+pub enum Error {
+    /// A syntax error.
+    Syntax(String),
+    /// The compiled regex exceeded the configured
+    /// regex compiled size limit.
+    CompiledTooBig(usize),
+}
+
+impl Display for Error {
+    fn fmt(&self, f: &mut core::fmt::Formatter<'_>) -> core::fmt::Result {
+        match *self {
+            Error::Syntax(ref err) => Display::fmt(err, f),
+            Error::CompiledTooBig(limit) => {
+                write!(f, "Compiled regex exceeds size limit of {} bytes.", limit)
+            }
+        }
+    }
+}
+
 #[cfg(test)]
 mod test {
     use super::*;

engine/src/rhs_types/regex/mod.rs

@@ -25,7 +25,7 @@ wirefilter.workspace = true
 
 [dev-dependencies]
 indoc.workspace = true
-regex.workspace = true
+regex-automata.workspace = true
 
 [build-dependencies]
 cbindgen = "0.27"

ffi/Cargo.toml

@@ -1039,7 +1039,7 @@ pub extern "C" fn wirefilter_get_version() -> StaticRustAllocatedString {
 #[allow(clippy::bool_assert_comparison)]
 mod ffi_test {
     use super::*;
-    use regex::Regex;
+    use regex_automata::meta::Regex;
     use std::ffi::CStr;
 
     impl RustAllocatedString {