fix: Fixed handling surrogare characters in insert, replace, delete mode in Vim

Author: andrewnester

Diff for: src/keyboard/vim.js

@@ -923,6 +923,20 @@ domLib.importCssString(`.normal-mode .ace_cursor{
     return range.head;
   }
 
+  function updateSelectionForSurrogateCharacters(cm, curStart, curEnd) {
+    // start and character position when no selection 
+    // is the same in visual mode, and differs in 1 character in normal mode
+    if (curStart.line === curEnd.line && curStart.ch >= curEnd.ch - 1) {
+      var text = cm.getLine(curStart.line);
+      var charCode = text.charCodeAt(curStart.ch);
+      if (0xD800 <= charCode && charCode <= 0xD8FF) {
+        curEnd.ch += 1;
+      }
+    }
+
+    return {start: curStart, end: curEnd};
+  }
+
   var defaultKeymap = [
     // Key to key mapping. This goes first to make it possible to override
     // existing mappings.
@@ -2540,9 +2554,10 @@ domLib.importCssString(`.normal-mode .ace_cursor{
             mode = vim.visualBlock ? 'block' :
                    linewise ? 'line' :
                    'char';
+            var newPositions = updateSelectionForSurrogateCharacters(cm, curStart, curEnd);
             cmSel = makeCmSelection(cm, {
-              anchor: curStart,
-              head: curEnd
+              anchor: newPositions.start,
+              head: newPositions.end
             }, mode);
             if (linewise) {
               var ranges = cmSel.ranges;
@@ -2574,9 +2589,10 @@ domLib.importCssString(`.normal-mode .ace_cursor{
             }
             mode = 'char';
             var exclusive = !motionArgs.inclusive || linewise;
+            var newPositions = updateSelectionForSurrogateCharacters(cm, curStart, curEnd);
             cmSel = makeCmSelection(cm, {
-              anchor: curStart,
-              head: curEnd
+              anchor: newPositions.start,
+              head: newPositions.end
             }, mode, exclusive);
           }
           cm.setSelections(cmSel.ranges, cmSel.primary);
@@ -3449,9 +3465,10 @@ domLib.importCssString(`.normal-mode .ace_cursor{
           vim.visualBlock = !!actionArgs.blockwise;
           head = clipCursorToContent(
               cm, new Pos(anchor.line, anchor.ch + repeat - 1));
+          var newPosition = updateSelectionForSurrogateCharacters(cm, anchor, head)
           vim.sel = {
-            anchor: anchor,
-            head: head
+            anchor: newPosition.start,
+            head: newPosition.end
           };
           CodeMirror.signal(cm, "vim-mode-change", {mode: "visual", subMode: vim.visualLine ? "linewise" : vim.visualBlock ? "blockwise" : ""});
           updateCmSelection(cm);
@@ -3749,18 +3766,25 @@ domLib.importCssString(`.normal-mode .ace_cursor{
           }
           curEnd = new Pos(curStart.line, replaceTo);
         }
+
+        var newPositions = updateSelectionForSurrogateCharacters(cm, curStart, curEnd);
+        curStart = newPositions.start;
+        curEnd = newPositions.end;
         if (replaceWith=='\n') {
           if (!vim.visualMode) cm.replaceRange('', curStart, curEnd);
           // special case, where vim help says to replace by just one line-break
           (CodeMirror.commands.newlineAndIndentContinueComment || CodeMirror.commands.newlineAndIndent)(cm);
         } else {
           var replaceWithStr = cm.getRange(curStart, curEnd);
+          // replace all surrogate characters with selected character
+          replaceWithStr = replaceWithStr.replace(/[\uD800-\uDBFF][\uDC00-\uDFFF]/g, replaceWith);
           //replace all characters in range by selected, but keep linebreaks
           replaceWithStr = replaceWithStr.replace(/[^\n]/g, replaceWith);
           if (vim.visualBlock) {
             // Tabs are split in visua block before replacing
             var spaces = new Array(cm.getOption("tabSize")+1).join(' ');
             replaceWithStr = cm.getSelection();
+            replaceWithStr = replaceWithStr.replace(/[\uD800-\uDBFF][\uDC00-\uDFFF]/g, replaceWith);
             replaceWithStr = replaceWithStr.replace(/\t/g, spaces).replace(/[^\n]/g, replaceWith).split('\n');
             cm.replaceSelections(replaceWithStr);
           } else {

Diff for: src/keyboard/vim_test.js

@@ -1863,6 +1863,13 @@ testVim('i', function(cm, vim, helpers) {
   helpers.assertCursorAt(0, 1);
   eq('vim-insert', cm.getOption('keyMap'));
 });
+testVim('i with surrogate characters', function(cm, vim, helpers) {
+  cm.setCursor(0, 0);
+  helpers.doKeys('i');
+  helpers.doKeys('test');
+  helpers.doKeys('<Esc>');
+  eq('test😀', cm.getValue());
+}, { value: '😀' });
 testVim('i_repeat', function(cm, vim, helpers) {
   helpers.doKeys('3', 'i');
   helpers.doKeys('test')
@@ -2111,6 +2118,11 @@ testVim('r', function(cm, vim, helpers) {
   helpers.doKeys('r', '<CR>');
   eq('\nx', cm.getValue());
 }, { value: 'wordet\nanother' });
+testVim('r with surrogate characters', function(cm, vim, helpers) {
+  cm.setCursor(0, 0);
+  helpers.doKeys('r', 'u');
+  eq('u', cm.getValue());
+}, { value: '😀' });
 testVim('r_visual_block', function(cm, vim, helpers) {
   cm.ace.setOptions({tabSize: 4, useSoftTabs: false}); // ace_patch TODO
   cm.setCursor(2, 3);
@@ -2125,6 +2137,16 @@ testVim('r_visual_block', function(cm, vim, helpers) {
   helpers.doKeys('<C-v>', 'h', 'h', 'r', 'r');
   eq('1  l\n5  l\nalllefg\nrrrrrrrr', cm.getValue());
 }, {value: '1234\n5678\nabcdefg', indentWithTabs: true});
+testVim('r_visual with surrogate characters', function(cm, vim, helpers) {
+  cm.setCursor(0, 0);
+  helpers.doKeys('v', 'r', 'u');
+  eq('u', cm.getValue());
+}, { value: '😀' });
+testVim('r_visual_block with surrogate characters', function(cm, vim, helpers) {
+  cm.setCursor(0, 0);
+  helpers.doKeys('<C-v>', 'r', 'u');
+  eq('u', cm.getValue());
+}, { value: '😀' });
 testVim('R', function(cm, vim, helpers) {
   cm.setCursor(0, 1);
   helpers.doKeys('R');
@@ -2673,13 +2695,27 @@ testVim('s_normal', function(cm, vim, helpers) {
   helpers.doKeys('<Esc>');
   eq('ac', cm.getValue());
 }, { value: 'abc'});
+testVim('s_normal surrogate character', function(cm, vim, helpers) {
+  cm.setCursor(0, 0);
+  helpers.doKeys('s');
+  helpers.doKeys('test');
+  helpers.doKeys('<Esc>');
+  eq('test', cm.getValue());
+}, { value: '😀' });
 testVim('s_visual', function(cm, vim, helpers) {
   cm.setCursor(0, 1);
   helpers.doKeys('v', 's');
   helpers.doKeys('<Esc>');
   helpers.assertCursorAt(0, 0);
   eq('ac', cm.getValue());
 }, { value: 'abc'});
+testVim('d with surrogate character', function(cm, vim, helpers) {
+  cm.setCursor(0, 0);
+  helpers.doKeys('v');
+  helpers.doKeys('d');
+  helpers.doKeys('<Esc>');
+  eq('', cm.getValue());
+}, { value: '😀' });
 testVim('o_visual', function(cm, vim, helpers) {
   cm.setCursor(0,0);
   helpers.doKeys('v','l','l','l','o');