Fix empty search match before surrogate pair causing infinite loop (#5389)

* Fix empty search match causing infinite loop

* Improve regex u flag feature detection

* Use RegExp.unicode instead of $supportsUnicodeFlag

* Fix mistake in last commit

* Fix failing tests

* Add new unicode mode tests

* Add missing semi colons

Author: FIameCaster

src/ext/searchbox.js

@@ -167,6 +167,7 @@ class SearchBox {
     updateCounter() {
         var editor = this.editor;
         var regex = editor.$search.$options.re;
+        var supportsUnicodeFlag = regex.unicode;
         var all = 0;
         var before = 0;
         if (regex) {
@@ -188,7 +189,7 @@ class SearchBox {
                 if (all > MAX_COUNT)
                     break;
                 if (!m[0]) {
-                    regex.lastIndex = last += 1;
+                    regex.lastIndex = last += lang.skipEmptyMatch(value, last, supportsUnicodeFlag);
                     if (last >= value.length)
                         break;
                 }

src/lib/lang.js

@@ -181,11 +181,6 @@ exports.supportsLookbehind = function () {
     return true;
 };
 
-exports.supportsUnicodeFlag = function () {
-    try {
-        new RegExp('^.$', 'u');
-    } catch (error) {
-        return false;
-    }
-    return true;
+exports.skipEmptyMatch = function(line, last, supportsUnicodeFlag) {
+    return supportsUnicodeFlag && line.codePointAt(last) > 0xffff ? 2 : 1;
 };

src/search.js

@@ -218,28 +218,22 @@ class Search {
 
         if (!options.needle)
             return options.re = false;
-        
-        if (options.$supportsUnicodeFlag === undefined) {
-            options.$supportsUnicodeFlag = lang.supportsUnicodeFlag();
-        }
+
+        if (!options.regExp)
+            needle = lang.escapeRegExp(needle);
+
+        var modifier = options.caseSensitive ? "gm" : "gmi";
 
         try {
             new RegExp(needle, "u");
+            options.$supportsUnicodeFlag = true;
+            modifier += "u";
         } catch (e) {
             options.$supportsUnicodeFlag = false; //left for backward compatibility with previous versions for cases like /ab\{2}/gu
         }
-        
-        if (!options.regExp)
-            needle = lang.escapeRegExp(needle);
 
         if (options.wholeWord)
             needle = addWordBoundary(needle, options);
-
-        var modifier = options.caseSensitive ? "gm" : "gmi";
-
-        if (options.$supportsUnicodeFlag) {
-            modifier += "u";
-        }
 
         options.$isMultiLine = !$disableFakeMultiline && /[\n\r]/.test(needle);
         if (options.$isMultiLine)
@@ -270,6 +264,7 @@ class Search {
             return false;
         var backwards = options.backwards == true;
         var skipCurrent = options.skipCurrent != false;
+        var supportsUnicodeFlag = re.unicode;
 
         var range = options.range;
         var start = options.start;
@@ -343,7 +338,7 @@ class Search {
                     last = m.index;
                     if (!length) {
                         if (last >= line.length) break;
-                        re.lastIndex = last += 1;
+                        re.lastIndex = last += lang.skipEmptyMatch(line, last, supportsUnicodeFlag);
                     }
                     if (m.index + length > endIndex)
                         break;
@@ -369,7 +364,7 @@ class Search {
                     if (callback(row, last, row,last + length))
                         return true;
                     if (!length) {
-                        re.lastIndex = last += 1;
+                        re.lastIndex = last += lang.skipEmptyMatch(line, last, supportsUnicodeFlag);
                         if (last >= line.length) return false;
                     }
                 }

src/search_test.js

@@ -9,6 +9,7 @@ var MockRenderer = require("./test/mockrenderer").MockRenderer;
 var Editor = require("./editor").Editor;
 var Search = require("./search").Search;
 var assert = require("./test/assertions");
+var Range = require("./range").Range;
 
 module.exports = {
     "test: configure the search object" : function() {
@@ -172,6 +173,38 @@ module.exports = {
         assert.position(range.end, 1, 6);
     },
 
+    "test: return to unicode mode when possible": function() {
+        var session = new EditSession(["𝓕oo"]);
+
+        var search = new Search().set({
+            needle: "}",
+            regExp: true
+        });
+
+        search.find(session);
+        search.set({
+            needle: "."
+        });
+
+        var range = search.find(session);
+        assert.position(range.start, 0, 0);
+        assert.position(range.end, 0, 2);
+    },
+
+    "test: empty match before surrogate pair": function() {
+        var session = new EditSession(["𝓕oo"]);
+
+        var search = new Search().set({
+            needle: "()",
+            regExp: true,
+            start: new Range(0, 0, 0, 0)
+        });
+
+        var range = search.find(session);
+        assert.position(range.start, 0, 2);
+        assert.position(range.end, 0, 2);
+    },
+
     "test: find backwards": function() {
         var session = new EditSession(["juhu juhu juhu juhu"]);
         session.getSelection().moveCursorTo(0, 10);