From f48503cd31826f8cfb0f0e2b6964879dbe318385 Mon Sep 17 00:00:00 2001
From: Sebastian McKenzie <sebmck@gmail.com>
Date: Sun, 19 Oct 2014 18:01:34 +1100
Subject: [PATCH] add unicode flag support to regex

---
 acorn.js              | 38 ++++++++++++++++++++++++++----
 test/tests-harmony.js | 29 +++++++++++++++++++++++
 test/tests.js         | 54 +++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 117 insertions(+), 4 deletions(-)

diff --git a/acorn.js b/acorn.js
index a36b5dac1a..efd9e42369 100644
--- a/acorn.js
+++ b/acorn.js
@@ -934,14 +934,36 @@
     // Need to use `readWord1` because '\uXXXX' sequences are allowed
     // here (don't ask).
     var mods = readWord1();
-    if (mods && !/^[gmsiy]*$/.test(mods)) raise(start, "Invalid regular expression flag");
+    var tmp = content;
+    if (mods) {
+      var validFlags = /^[gmsiy]*$/;
+      if (options.ecmaVersion >= 6) validFlags = /^[gmsiyu]*$/;
+      if (!validFlags.test(mods)) raise(start, "Invalid regular expression flag");
+      if (mods.indexOf('u') >= 0) {
+        // Replace each astral symbol and every Unicode code point
+        // escape sequence that represents such a symbol with a single
+        // ASCII symbol to avoid throwing on regular expressions that
+        // are only valid in combination with the `/u` flag.
+        tmp = tmp
+          .replace(/\\u\{([0-9a-fA-F]{5,6})\}/g, 'x')
+          .replace(/[\uD800-\uDBFF][\uDC00-\uDFFF]/g, 'x');
+      }
+    }
+    // Detect invalid regular expressions.
     try {
-      var value = new RegExp(content, mods);
+      new RegExp(tmp);
     } catch (e) {
       if (e instanceof SyntaxError) raise(start, "Error parsing regular expression: " + e.message);
       raise(e);
     }
-    return finishToken(_regexp, value);
+    // Get a regular expression object for this pattern-flag pair, or `null` in
+    // case the current environment doesn't support the flags it uses.
+    try {
+      var value = new RegExp(content, mods);
+    } catch (err) {
+      value = null;
+    }
+    return finishToken(_regexp, {pattern: content, flags: mods, value: value});
   }
 
   // Read an integer in the given radix. Return null if zero digits
@@ -2005,7 +2027,15 @@
       }
       return id;
 
-    case _num: case _string: case _regexp:
+    case _regexp:
+      var node = startNode();
+      node.regex = {pattern: tokVal.pattern, flags: tokVal.flags};
+      node.value = tokVal.value;
+      node.raw = input.slice(tokStart, tokEnd);
+      next();
+      return finishNode(node, "Literal");
+
+    case _num: case _string:
       var node = startNode();
       node.value = tokVal;
       node.raw = input.slice(tokStart, tokEnd);
diff --git a/test/tests-harmony.js b/test/tests-harmony.js
index 66a802be25..b35c91920f 100644
--- a/test/tests-harmony.js
+++ b/test/tests-harmony.js
@@ -14792,6 +14792,35 @@ test("func(...a, b)", {
   locations: true
 });
 
+test("/[a-z]/u", {
+  type: "Program",
+  body: [
+    {
+      type: "ExpressionStatement",
+      expression: {
+        type: "Literal",
+        regex: {
+          pattern: "[a-z]",
+          flags: "u"
+        },
+        loc: {
+          start: {
+            line: 1,
+            column: 0
+          },
+          end: {
+            line: 1,
+            column: 8
+          }
+        }
+      }
+    }
+  ]
+}, {
+  locations: true,
+  ecmaVersion: 6
+});
+
 // Harmony Invalid syntax
 
 testFail("0o", "Expected number in radix 8 (1:2)", {ecmaVersion: 6});
diff --git a/test/tests.js b/test/tests.js
index 4fedc72ca1..287d98f84d 100644
--- a/test/tests.js
+++ b/test/tests.js
@@ -135,6 +135,60 @@ test("\n    42\n\n", {
   }
 });
 
+test("/foobar/", {
+  type: "Program",
+  body: [
+    {
+      type: "ExpressionStatement",
+      expression: {
+        type: "Literal",
+        value: /foobar/,
+        regex: {
+          pattern: "foobar",
+          flags: ""
+        },
+        loc: {
+          start: {
+            line: 1,
+            column: 0
+          },
+          end: {
+            line: 1,
+            column: 8
+          }
+        }
+      }
+    }
+  ]
+});
+
+test("/[a-z]/g", {
+  type: "Program",
+  body: [
+    {
+      type: "ExpressionStatement",
+      expression: {
+        type: "Literal",
+        value: /[a-z]/,
+        regex: {
+          pattern: "[a-z]",
+          flags: "g"
+        },
+        loc: {
+          start: {
+            line: 1,
+            column: 0
+          },
+          end: {
+            line: 1,
+            column: 8
+          }
+        }
+      }
+    }
+  ]
+});
+
 test("(1 + 2 ) * 3", {
   type: "Program",
   body: [