From 4ce39ebf32d79982a31458b7b70d9fecde40cd35 Mon Sep 17 00:00:00 2001
From: magic-akari <akari.ccino@gmail.com>
Date: Wed, 6 Mar 2024 12:40:25 +0800
Subject: [PATCH] feat(es/parser): Support Regular Expression `v` flag (#8690)

**Related issue:**

- Closes #8462
---
 crates/swc_ecma_compat_common/src/regexp.rs | 3 +++
 crates/swc_ecma_compat_es2015/src/lib.rs    | 1 +
 crates/swc_ecma_compat_es2018/src/lib.rs    | 1 +
 crates/swc_ecma_compat_es2022/src/lib.rs    | 1 +
 crates/swc_ecma_parser/src/parser/expr.rs   | 3 ++-
 crates/swc_ecma_preset_env/src/lib.rs       | 2 ++
 6 files changed, 10 insertions(+), 1 deletion(-)

diff --git a/crates/swc_ecma_compat_common/src/regexp.rs b/crates/swc_ecma_compat_common/src/regexp.rs
index fc68ee1631ab..7c24107baf05 100644
--- a/crates/swc_ecma_compat_common/src/regexp.rs
+++ b/crates/swc_ecma_compat_common/src/regexp.rs
@@ -23,6 +23,8 @@ pub struct Config {
     pub unicode_property_regex: bool,
     /// [RegExp.prototype.unicode](https://tc39.es/ecma262/multipage/text-processing.html#sec-get-regexp.prototype.unicode)
     pub unicode_regex: bool,
+    // [RegExp.prototype.unicodeSets](https://github.com/tc39/proposal-regexp-v-flag)
+    pub unicode_sets_regex: bool,
 }
 
 struct RegExp {
@@ -39,6 +41,7 @@ impl VisitMut for RegExp {
             if (self.config.dot_all_regex && regex.flags.contains('s'))
                 || (self.config.sticky_regex && regex.flags.contains('y'))
                 || (self.config.unicode_regex && regex.flags.contains('u'))
+                || (self.config.unicode_sets_regex && regex.flags.contains('v'))
                 || (self.config.has_indices && regex.flags.contains('d'))
                 || (self.config.named_capturing_groups_regex && regex.exp.contains("(?<"))
                 || (self.config.lookbehind_assertion && regex.exp.contains("(?<=")
diff --git a/crates/swc_ecma_compat_es2015/src/lib.rs b/crates/swc_ecma_compat_es2015/src/lib.rs
index 95f883c2fd5d..e536e0704b4a 100644
--- a/crates/swc_ecma_compat_es2015/src/lib.rs
+++ b/crates/swc_ecma_compat_es2015/src/lib.rs
@@ -68,6 +68,7 @@ where
             sticky_regex: true,
             unicode_property_regex: false,
             unicode_regex: true,
+            unicode_sets_regex: false,
         }),
         block_scoped_functions(),
         template_literal(c.template_literal),
diff --git a/crates/swc_ecma_compat_es2018/src/lib.rs b/crates/swc_ecma_compat_es2018/src/lib.rs
index 9ec1d7c69485..041032b2102c 100644
--- a/crates/swc_ecma_compat_es2018/src/lib.rs
+++ b/crates/swc_ecma_compat_es2018/src/lib.rs
@@ -22,6 +22,7 @@ pub fn es2018(c: Config) -> impl Fold {
             sticky_regex: false,
             unicode_property_regex: true,
             unicode_regex: false,
+            unicode_sets_regex: false,
         }),
         object_rest_spread(c.object_rest_spread)
     )
diff --git a/crates/swc_ecma_compat_es2022/src/lib.rs b/crates/swc_ecma_compat_es2022/src/lib.rs
index 2611710a02e7..bfe0bf226af8 100644
--- a/crates/swc_ecma_compat_es2022/src/lib.rs
+++ b/crates/swc_ecma_compat_es2022/src/lib.rs
@@ -24,6 +24,7 @@ pub fn es2022<C: Comments>(cm: Option<C>, config: Config, unresolved_mark: Mark)
             sticky_regex: false,
             unicode_property_regex: true,
             unicode_regex: false,
+            unicode_sets_regex: false,
         }),
         static_blocks(config.class_properties.static_blocks_mark),
         class_properties(cm, config.class_properties, unresolved_mark),
diff --git a/crates/swc_ecma_parser/src/parser/expr.rs b/crates/swc_ecma_parser/src/parser/expr.rs
index 6ec36478ad47..cbbffbef53a0 100644
--- a/crates/swc_ecma_parser/src/parser/expr.rs
+++ b/crates/swc_ecma_parser/src/parser/expr.rs
@@ -340,7 +340,8 @@ impl<I: Tokens> Parser<I> {
                                     AHashMap::<char, usize>::default(),
                                     |mut map, flag| {
                                         let key = match flag {
-                                            'g' | 'i' | 'm' | 's' | 'u' | 'y' | 'd' => flag,
+                                            // https://tc39.es/ecma262/#sec-isvalidregularexpressionliteral
+                                            'd' | 'g' | 'i' | 'm' | 's' | 'u' | 'v' | 'y' => flag,
                                             _ => '\u{0000}', // special marker for unknown flags
                                         };
                                         map.entry(key).and_modify(|count| *count += 1).or_insert(1);
diff --git a/crates/swc_ecma_preset_env/src/lib.rs b/crates/swc_ecma_preset_env/src/lib.rs
index e86def9631b6..cb19217b090d 100644
--- a/crates/swc_ecma_preset_env/src/lib.rs
+++ b/crates/swc_ecma_preset_env/src/lib.rs
@@ -99,6 +99,7 @@ where
         let enable_sticky_regex = should_enable!(StickyRegex, false);
         let enable_unicode_property_regex = should_enable!(UnicodePropertyRegex, false);
         let enable_unicode_regex = should_enable!(UnicodeRegex, false);
+        let enable_unicode_sets_regex = should_enable!(UnicodeSetsRegex, false);
 
         let enable = enable_dot_all_regex
             || enable_named_capturing_groups_regex
@@ -119,6 +120,7 @@ where
                     sticky_regex: enable_sticky_regex,
                     unicode_property_regex: enable_unicode_property_regex,
                     unicode_regex: enable_unicode_regex,
+                    unicode_sets_regex: enable_unicode_sets_regex,
                 }),
                 enable
             )