Merge pull request #127 from liuxingbaoyu/modifiers

feat: support modifiers proposal
jviereck · Jun 26, 2022 · 524fbd5 · 524fbd5
2 parents e66d3d0 + 9c14b21
commit 524fbd5
Show file tree

Hide file tree

Showing 6 changed files with 658 additions and 11 deletions.
diff --git a/parser.d.ts b/parser.d.ts
@@ -5,6 +5,7 @@ export type Features = {
   namedGroups?: boolean;
   unicodePropertyEscape?: boolean;
   unicodeSet?: boolean;
+  modifiers?: boolean;
 };
 
 export type AstNodeType =
@@ -98,15 +99,35 @@ export type CharacterClass<F extends Features = {}> = Base<"characterClass"> & {
   kind: "union" | _If<F["unicodeSet"], "intersection" | "subtraction", never>;
 };
 
-export type NonCapturingGroup<F extends Features = {}> = Base<"group"> & {
-  behavior:
-    | "ignore"
-    | "lookahead"
-    | "lookbehind"
-    | "negativeLookahead"
-    | "negativeLookbehind";
-  body: RootNode<F>[];
-};
+export type ModifierFlags = {
+  enabling: string,
+  disabling: string
+}
+
+export type NonCapturingGroup<F extends Features = {}> = Base<"group"> &
+  (
+    | {
+        behavior:
+          | "lookahead"
+          | "lookbehind"
+          | "negativeLookahead"
+          | "negativeLookbehind";
+        body: RootNode<F>[];
+      }
+    | ({
+        behavior: "ignore";
+        body: RootNode<F>[];
+      } & _If<
+        F["modifiers"],
+        {
+          modifierFlags?: ModifierFlags;
+        },
+        {
+          modifierFlags: undefined;
+        }
+      >)
+  );
+
 
 export type CapturingGroup<F extends Features = {}> = Base<"group"> & {
   behavior: "normal";

diff --git a/parser.js b/parser.js
@@ -207,6 +207,16 @@
 // ClassHalfOfDouble ::
 //      one of & - ! # % , : ; < = > @ _ ` ~
 //
+// --------------------------------------------------------------
+// NOTE: The following productions refer to the
+//       "Regular Expression Pattern Modifiers for ECMAScript" proposal.
+//       https://github.com/tc39/proposal-regexp-modifiers
+// --------------------------------------------------------------
+//
+// Atom ::
+//      ( ? RegularExpressionFlags : Disjunction )
+//      ( ? RegularExpressionFlags - RegularExpressionFlags : Disjunction )
+//
 
 "use strict";
 (function() {
@@ -718,7 +728,8 @@
       //      \ AtomEscape
       //      CharacterClass
       //      ( GroupSpecifier Disjunction )
-      //      ( ? : Disjunction )
+      //      ( ? RegularExpressionFlags : Disjunction )
+      //      ( ? RegularExpressionFlags - RegularExpressionFlags : Disjunction )
       // ExtendedAtom ::
       //      ExtendedPatternCharacter
       // ExtendedPatternCharacter ::
@@ -766,13 +777,60 @@
         group.name = name;
         return group;
       }
+      else if (features.modifiers && str.indexOf("(?") == pos && str[pos+2] != ":") {
+        return parseModifiersGroup();
+      }
       else {
         //      ( Disjunction )
         //      ( ? : Disjunction )
         return parseGroup('(?:', 'ignore', '(', 'normal');
       }
     }
 
+    function parseModifiersGroup() {
+      function hasDupChar(str) {
+        var i = 0;
+        while (i < str.length) {
+          if (str.indexOf(str[i], i + 1) != -1) {
+            return true;
+          }
+          i++;
+        }
+        return false;
+      }
+
+      var from = pos;
+      incr(2);
+
+      var enablingFlags = matchReg(/^[sim]+/);
+      var disablingFlags;
+      if(match("-")){
+        disablingFlags = matchReg(/^[sim]+/);
+        if (!disablingFlags) {
+          bail('Invalid flags for modifiers group');
+        }
+      } else if(!enablingFlags){
+        bail('Invalid flags for modifiers group');
+      }
+
+      enablingFlags = enablingFlags ? enablingFlags[0] : "";
+      disablingFlags = disablingFlags ? disablingFlags[0] : "";
+
+      var flags = enablingFlags + disablingFlags;
+      if(flags.length > 3 || hasDupChar(flags)) {
+        bail('flags cannot be duplicated for modifiers group');
+      }
+
+      var modifiersGroup = finishGroup("ignore", from);
+
+      modifiersGroup.modifierFlags = {
+          enabling: enablingFlags,
+          disabling: disablingFlags
+        };
+
+      return modifiersGroup;
+    }
+
     function parseUnicodeSurrogatePairEscape(firstEscape) {
       if (isUnicodeMode) {
         var first, second;

diff --git a/test/index.js b/test/index.js
@@ -62,6 +62,15 @@ runTests('./test-data-unicode-set.json', 'v', {
   unicodeSet: true,
   unicodePropertyEscape: true
 });
+runTests('./test-data-modifiers-group.json', '', {
+  modifiers: true,
+  namedGroups: true,
+});
+runTests('./test-data-named-groups.json', '', {
+  modifiers: true,
+  namedGroups: true,
+});
+
 
 (function testUVError() {
   var message = 'It should throw an error when using both the "u" and "v" flags.';