Build regex using parser settings instead of full filter parser

cloudflare · Feb 24, 2025 · 10b13f1 · 10b13f1
1 parent ca9a4a2
commit 10b13f1
Show file tree

Hide file tree

Showing 3 changed files with 24 additions and 22 deletions.
diff --git a/engine/src/ast/field_expr.rs b/engine/src/ast/field_expr.rs
@@ -2646,9 +2646,9 @@ mod tests {
 
         // Matches operator
         let parser = FilterParser::new(&SCHEME);
-        let r = Regex::new("a.b", RegexFormat::Literal, &parser).unwrap();
+        let r = Regex::new("a.b", RegexFormat::Literal, parser.settings()).unwrap();
         let expr = assert_ok!(
-            FilterParser::new(&SCHEME).lex_as("http.host matches r###\"a.b\"###"),
+            parser.lex_as("http.host matches r###\"a.b\"###"),
             ComparisonExpr {
                 lhs: IndexExpr {
                     identifier: IdentifierExpr::Field(field("http.host")),

diff --git a/engine/src/rhs_types/regex/imp_real.rs b/engine/src/rhs_types/regex/imp_real.rs
@@ -1,4 +1,4 @@
-use crate::{FilterParser, RegexFormat};
+use crate::{ParserSettings, RegexFormat};
 
 pub use regex::Error;
 
@@ -14,12 +14,12 @@ impl Regex {
     pub fn new(
         pattern: &str,
         format: RegexFormat,
-        parser: &FilterParser<'_>,
+        settings: &ParserSettings,
     ) -> Result<Self, Error> {
         ::regex::bytes::RegexBuilder::new(pattern)
             .unicode(false)
-            .size_limit(parser.settings.regex_compiled_size_limit)
-            .dfa_size_limit(parser.settings.regex_dfa_size_limit)
+            .size_limit(settings.regex_compiled_size_limit)
+            .dfa_size_limit(settings.regex_dfa_size_limit)
             .build()
             .map(|r| Regex {
                 compiled_regex: r,
@@ -51,15 +51,13 @@ impl From<Regex> for regex::bytes::Regex {
 
 #[test]
 fn test_compiled_size_limit() {
-    use crate::Scheme;
-
-    let scheme = Scheme::default();
-
     const COMPILED_SIZE_LIMIT: usize = 1024 * 1024;
-    let mut parser = FilterParser::new(&scheme);
-    parser.regex_set_compiled_size_limit(COMPILED_SIZE_LIMIT);
+    let settings = ParserSettings {
+        regex_compiled_size_limit: COMPILED_SIZE_LIMIT,
+        ..Default::default()
+    };
     assert_eq!(
-        Regex::new(".{4079,65535}", RegexFormat::Literal, &parser),
+        Regex::new(".{4079,65535}", RegexFormat::Literal, &settings),
         Err(Error::CompiledTooBig(COMPILED_SIZE_LIMIT))
     );
 }
diff --git a/engine/src/rhs_types/regex/mod.rs b/engine/src/rhs_types/regex/mod.rs
@@ -52,7 +52,7 @@ fn lex_regex_from_raw_string<'i>(
     parser: &FilterParser<'_>,
 ) -> LexResult<'i, Regex> {
     let ((lexed, hashes), input) = lex_raw_string_as_str(input)?;
-    match Regex::new(lexed, RegexFormat::Raw(hashes), parser) {
+    match Regex::new(lexed, RegexFormat::Raw(hashes), parser.settings()) {
         Ok(regex) => Ok((regex, input)),
         Err(err) => Err((LexErrorKind::ParseRegex(err), input)),
     }
@@ -94,7 +94,7 @@ fn lex_regex_from_literal<'i>(input: &'i str, parser: &FilterParser<'_>) -> LexR
             };
         }
     };
-    match Regex::new(&regex_buf, RegexFormat::Literal, parser) {
+    match Regex::new(&regex_buf, RegexFormat::Literal, parser.settings()) {
         Ok(regex) => Ok((regex, input)),
         Err(err) => Err((LexErrorKind::ParseRegex(err), regex_str)),
     }
@@ -123,17 +123,19 @@ impl Serialize for Regex {
 #[cfg(test)]
 mod test {
     use super::*;
-    use crate::Scheme;
+    use crate::{ParserSettings, Scheme};
 
     #[test]
     fn test() {
         let scheme = Scheme::new();
+        let parser = FilterParser::new(&scheme);
+
         let expr = assert_ok!(
-            Regex::lex_with(r#""[a-z"\]]+\d{1,10}\"";"#, &FilterParser::new(&scheme)),
+            Regex::lex_with(r#""[a-z"\]]+\d{1,10}\"";"#, &parser),
             Regex::new(
                 r#"[a-z"\]]+\d{1,10}""#,
                 RegexFormat::Literal,
-                &FilterParser::new(&scheme)
+                &ParserSettings::default(),
             )
             .unwrap(),
             ";"
@@ -142,7 +144,7 @@ mod test {
         assert_json!(expr, r#"[a-z"\]]+\d{1,10}""#);
 
         assert_err!(
-            Regex::lex_with(r#""abcd\"#, &FilterParser::new(&scheme)),
+            Regex::lex_with(r#""abcd\"#, &parser),
             LexErrorKind::MissingEndingQuote,
             "abcd\\"
         );
@@ -151,6 +153,8 @@ mod test {
     #[test]
     fn test_raw_string() {
         let scheme = Scheme::new();
+        let parser = FilterParser::new(&scheme);
+
         let expr = assert_ok!(
             Regex::lex_with(
                 r###"r#"[a-z"\]]+\d{1,10}""#;"###,
@@ -159,7 +163,7 @@ mod test {
             Regex::new(
                 r#"[a-z"\]]+\d{1,10}""#,
                 RegexFormat::Raw(1),
-                &FilterParser::new(&scheme)
+                parser.settings(),
             )
             .unwrap(),
             ";"
@@ -170,12 +174,12 @@ mod test {
         let expr = assert_ok!(
             Regex::lex_with(
                 r##"r#"(?u)\*\a\f\t\n\r\v\x7F\x{10FFFF}\u007F\u{7F}\U0000007F\U{7F}"#"##,
-                &FilterParser::new(&scheme)
+                &parser,
             ),
             Regex::new(
                 r#"(?u)\*\a\f\t\n\r\v\x7F\x{10FFFF}\u007F\u{7F}\U0000007F\U{7F}"#,
                 RegexFormat::Raw(1),
-                &FilterParser::new(&scheme)
+                parser.settings(),
             )
             .unwrap(),
             ""