feat: move only_literals optimization to regex compilation

vthib · Jul 16, 2023 · 6b9e899 · 6b9e899
1 parent e16f8fa
commit 6b9e899
Show file tree

Hide file tree

Showing 3 changed files with 18 additions and 19 deletions.
diff --git a/boreal/src/compiler/variable.rs b/boreal/src/compiler/variable.rs
@@ -89,20 +89,7 @@ pub(crate) fn compile_variable(
             regex::compile_regex(&ast.into(), dot_all, &modifiers)
         }
         VariableDeclarationValue::HexString(hex_string) => {
-            // Nocase, fullword and wide is not compatible with hex strings
-            modifiers.nocase = false;
-            modifiers.fullword = false;
-            modifiers.wide = false;
-
-            let hir = hex_string.into();
-            match only_literals::hir_to_only_literals(&hir) {
-                Some(literals) => Ok(CompiledVariable {
-                    literals,
-                    matcher_kind: matcher::MatcherKind::Literals,
-                    non_wide_regex: None,
-                }),
-                None => regex::compile_regex(&hir, true, &modifiers),
-            }
+            regex::compile_regex(&hex_string.into(), true, &modifiers)
         }
     };
 

diff --git a/boreal/src/compiler/variable/only_literals.rs b/boreal/src/compiler/variable/only_literals.rs
@@ -1,9 +1,9 @@
 use crate::regex::{visit, Hir, VisitAction, Visitor};
 
 /// Can the hex string be expressed using only literals.
-pub(super) fn hir_to_only_literals(hir: &Hir) -> Option<Vec<Vec<u8>>> {
+pub(super) fn hir_to_only_literals(hir: &Hir, dot_all: bool) -> Option<Vec<Vec<u8>>> {
     // TODO: move this count into the HirStatistics visitor
-    match visit(hir, CountLiterals::new(true)) {
+    match visit(hir, CountLiterals::new(dot_all)) {
         Some(count) if count < 100 => visit(hir, Literals::new()),
         Some(_) | None => None,
     }
@@ -297,7 +297,7 @@ mod tests {
             let regex = parse_regex_string(regex);
             let hir = regex.ast.into();
 
-            let res = hir_to_only_literals(&hir);
+            let res = hir_to_only_literals(&hir, true);
             match &res {
                 Some(v) => assert_eq!(v, expected.unwrap()),
                 None => assert!(expected.is_none()),
@@ -342,7 +342,7 @@ mod tests {
             let hex_string = parse_hex_string(hex_string);
             let hir = hex_string.into();
 
-            let res = hir_to_only_literals(&hir);
+            let res = hir_to_only_literals(&hir, true);
             match &res {
                 Some(v) => assert_eq!(v, expected.unwrap()),
                 None => assert!(expected.is_none()),

diff --git a/boreal/src/compiler/variable/regex.rs b/boreal/src/compiler/variable/regex.rs
@@ -5,7 +5,7 @@ use crate::regex::{regex_hir_to_string, visit, Hir, Regex, VisitAction, Visitor}
 
 use super::literals::LiteralsDetails;
 use super::matcher::MatcherKind;
-use super::{CompiledVariable, VariableCompilationError};
+use super::{only_literals, CompiledVariable, VariableCompilationError};
 
 /// Build a matcher for the given regex and string modifiers.
 ///
@@ -18,6 +18,18 @@ pub(super) fn compile_regex(
     dot_all: bool,
     modifiers: &VariableModifiers,
 ) -> Result<CompiledVariable, VariableCompilationError> {
+    // Try to convert into only literals if possible
+    // TODO: handle more modifiers
+    if !modifiers.nocase && !modifiers.wide {
+        if let Some(literals) = only_literals::hir_to_only_literals(hir, dot_all) {
+            return Ok(CompiledVariable {
+                literals,
+                matcher_kind: MatcherKind::Literals,
+                non_wide_regex: None,
+            });
+        }
+    }
+
     let LiteralsDetails {
         mut literals,
         pre_hir,