fix(es/minifier): Bailout regex optimization on invalid flags (#7020)

crates/swc_ecma_minifier/src/compress/pure/misc.rs

@@ -1,6 +1,7 @@
 use std::{fmt::Write, iter::once, num::FpCategory};
 
-use swc_atoms::js_word;
+use rustc_hash::FxHashSet;
+use swc_atoms::{js_word, JsWord};
 use swc_common::{iter::IdentifyLast, util::take::Take, Span, DUMMY_SP};
 use swc_ecma_ast::*;
 use swc_ecma_transforms_optimization::debug_assert_valid;
@@ -247,76 +248,78 @@ impl Pure<'_> {
 
     /// `new RegExp("([Sap]+)", "ig")` => `/([Sap]+)/gi`
     fn optimize_regex(&mut self, args: &mut Vec<ExprOrSpread>, span: &mut Span) -> Option<Expr> {
-        if args.is_empty() || args.len() > 2 {
-            return None;
+        fn valid_pattern(pattern: &Expr) -> Option<JsWord> {
+            if let Expr::Lit(Lit::Str(s)) = pattern {
+                if s.value.contains(|c: char| {
+                    // whitelist
+                    !c.is_ascii_alphanumeric()
+                        && !matches!(c, '$' | '[' | ']' | '(' | ')' | '{' | '}' | '-' | '+' | '_')
+                }) {
+                    None
+                } else {
+                    Some(s.value.clone())
+                }
+            } else {
+                None
+            }
         }
+        fn valid_flag(flag: &Expr, es_version: EsVersion) -> Option<JsWord> {
+            if let Expr::Lit(Lit::Str(s)) = flag {
+                let mut set = FxHashSet::default();
+                for c in s.value.chars() {
+                    if !(matches!(c, 'g' | 'i' | 'm')
+                        || (es_version >= EsVersion::Es2015 && matches!(c, 'u' | 'y'))
+                        || (es_version >= EsVersion::Es2018 && matches!(c, 's')))
+                        || (es_version >= EsVersion::Es2022 && matches!(c, 'd'))
+                    {
+                        return None;
+                    }
 
-        // We aborts the method if arguments are not literals.
-        if args.iter().any(|v| {
-            v.spread.is_some()
-                || match &*v.expr {
-                    Expr::Lit(Lit::Str(s)) => {
-                        if s.value.contains(|c: char| {
-                            // whitelist
-                            !c.is_ascii_alphanumeric()
-                                && !matches!(c, '%' | '[' | ']' | '(' | ')' | '{' | '}' | '-' | '+')
-                        }) {
-                            return true;
-                        }
-                        if s.value.contains("\\\0") || s.value.contains('/') {
-                            return true;
-                        }
-
-                        false
+                    if !set.insert(c) {
+                        return None;
                     }
-                    _ => true,
                 }
-        }) {
-            return None;
-        }
 
-        let pattern = args[0].expr.take();
-
-        let pattern = match *pattern {
-            Expr::Lit(Lit::Str(s)) => s.value,
-            _ => {
-                unreachable!()
+                Some(s.value.clone())
+            } else {
+                None
             }
+        }
+
+        let (pattern, flag) = match args.as_slice() {
+            [ExprOrSpread { spread: None, expr }] => (valid_pattern(expr)?, "".into()),
+            [ExprOrSpread {
+                spread: None,
+                expr: pattern,
+            }, ExprOrSpread {
+                spread: None,
+                expr: flag,
+            }] => (
+                valid_pattern(pattern)?,
+                valid_flag(flag, self.options.ecma)?,
+            ),
+            _ => return None,
         };
 
         if pattern.is_empty() {
             // For some expressions `RegExp()` and `RegExp("")`
             // Theoretically we can use `/(?:)/` to achieve shorter code
             // But some browsers released in 2015 don't support them yet.
-            args[0].expr = pattern.into();
             return None;
         }
 
-        let flags = args
-            .get_mut(1)
-            .map(|v| v.expr.take())
-            .map(|v| match *v {
-                Expr::Lit(Lit::Str(s)) => {
-                    assert!(s.value.is_ascii());
-
-                    let s = s.value.to_string();
-                    let mut bytes = s.into_bytes();
-                    bytes.sort_unstable();
-
-                    String::from_utf8(bytes).unwrap().into()
-                }
-                _ => {
-                    unreachable!()
-                }
-            })
-            .unwrap_or_default();
-
         report_change!("Optimized regex");
 
         Some(Expr::Lit(Lit::Regex(Regex {
             span: *span,
             exp: pattern.into(),
-            flags,
+            flags: {
+                let flag = flag.to_string();
+                let mut bytes = flag.into_bytes();
+                bytes.sort_unstable();
+
+                String::from_utf8(bytes).unwrap().into()
+            },
         })))
     }
 

crates/swc_ecma_minifier/tests/fixture/issues/4234/output.js

@@ -1,4 +1,4 @@
 bar(RegExp(""));
 bar(RegExp("", "u"));
 bar(/a/);
-bar(/a/u);
+bar(RegExp("a", "u"));

crates/swc_ecma_minifier/tests/terser/compress/issue_269/regexp/output.js

@@ -2,4 +2,4 @@
 /bar/gi;
 RegExp(foo);
 RegExp("bar", ig);
-/should/afil;
+RegExp("should", "fail");