readline: improve Unicode handling

Prevents moving left or right from placing the cursor in between code
units comprising a code point.

PR-URL: #25723
Fixes: #25693
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Ruben Bridgewater <ruben@bridgewater.de>

lib/readline.js

@@ -579,27 +579,48 @@ Interface.prototype._wordLeft = function() {
 Interface.prototype._wordRight = function() {
  if (this.cursor < this.line.length) {
  var trailing = this.line.slice(this.cursor);
- var match = trailing.match(/^(?:\s+|\W+|\w+)\s*/);
+ var match = trailing.match(/^(?:\s+|[^\w\s]+|\w+)\s*/);
  this._moveCursor(match[0].length);
  }
 };
 
+function charLengthLeft(str, i) {
+ if (i <= 0)
+ return 0;
+ if (i > 1 && str.codePointAt(i - 2) >= 2 ** 16 ||
+ str.codePointAt(i - 1) >= 2 ** 16) {
+ return 2;
+ }
+ return 1;
+}
+
+function charLengthAt(str, i) {
+ if (str.length <= i)
+ return 0;
+ return str.codePointAt(i) >= 2 ** 16 ? 2 : 1;
+}
 
 Interface.prototype._deleteLeft = function() {
  if (this.cursor > 0 && this.line.length > 0) {
- this.line = this.line.slice(0, this.cursor - 1) +
+ // The number of UTF-16 units comprising the character to the left
+ const charSize = charLengthLeft(this.line, this.cursor);
+ this.line = this.line.slice(0, this.cursor - charSize) +
  this.line.slice(this.cursor, this.line.length);
 
- this.cursor--;
+ this.cursor -= charSize;
  this._refreshLine();
  }
 };
 
 
 Interface.prototype._deleteRight = function() {
- this.line = this.line.slice(0, this.cursor) +
- this.line.slice(this.cursor + 1, this.line.length);
- this._refreshLine();
+ if (this.cursor < this.line.length) {
+ // The number of UTF-16 units comprising the character to the left
+ const charSize = charLengthAt(this.line, this.cursor);
+ this.line = this.line.slice(0, this.cursor) +
+ this.line.slice(this.cursor + charSize, this.line.length);
+ this._refreshLine();
+ }
 };
 
 
@@ -833,11 +854,11 @@ Interface.prototype._ttyWrite = function(s, key) {
  break;
 
  case 'b': // back one character
- this._moveCursor(-1);
+ this._moveCursor(-charLengthLeft(this.line, this.cursor));
  break;
 
  case 'f': // forward one character
- this._moveCursor(+1);
+ this._moveCursor(+charLengthAt(this.line, this.cursor));
  break;
 
  case 'l': // clear the whole screen
@@ -951,11 +972,12 @@ Interface.prototype._ttyWrite = function(s, key) {
  break;
 
  case 'left':
- this._moveCursor(-1);
+ // obtain the code point to the left
+ this._moveCursor(-charLengthLeft(this.line, this.cursor));
  break;
 
  case 'right':
- this._moveCursor(+1);
+ this._moveCursor(+charLengthAt(this.line, this.cursor));
  break;
 
  case 'home':

test/parallel/test-readline-interface.js

@@ -650,6 +650,115 @@ function isWarned(emitter) {
  rli.close();
  }
 
+ // Back and Forward one astral character
+ {
+ const fi = new FakeInput();
+ const rli = new readline.Interface({
+ input: fi,
+ output: fi,
+ prompt: '',
+ terminal: terminal
+ });
+ fi.emit('data', '💻');
+
+ // move left one character/code point
+ fi.emit('keypress', '.', { name: 'left' });
+ let cursorPos = rli._getCursorPos();
+ assert.strictEqual(cursorPos.rows, 0);
+ assert.strictEqual(cursorPos.cols, 0);
+
+ // move right one character/code point
+ fi.emit('keypress', '.', { name: 'right' });
+ cursorPos = rli._getCursorPos();
+ assert.strictEqual(cursorPos.rows, 0);
+ if (common.hasIntl) {
+ assert.strictEqual(cursorPos.cols, 2);
+ } else {
+ assert.strictEqual(cursorPos.cols, 1);
+ }
+
+ rli.on('line', common.mustCall((line) => {
+ assert.strictEqual(line, '💻');
+ }));
+ fi.emit('data', '\n');
+ rli.close();
+ }
+
+ // Two astral characters left
+ {
+ const fi = new FakeInput();
+ const rli = new readline.Interface({
+ input: fi,
+ output: fi,
+ prompt: '',
+ terminal: terminal
+ });
+ fi.emit('data', '💻');
+
+ // move left one character/code point
+ fi.emit('keypress', '.', { name: 'left' });
+ let cursorPos = rli._getCursorPos();
+ assert.strictEqual(cursorPos.rows, 0);
+ assert.strictEqual(cursorPos.cols, 0);
+
+ fi.emit('data', '🐕');
+ cursorPos = rli._getCursorPos();
+ assert.strictEqual(cursorPos.rows, 0);
+
+ if (common.hasIntl) {
+ assert.strictEqual(cursorPos.cols, 2);
+ } else {
+ assert.strictEqual(cursorPos.cols, 1);
+ // Fix cursor position without internationalization
+ fi.emit('keypress', '.', { name: 'left' });
+ }
+
+ rli.on('line', common.mustCall((line) => {
+ assert.strictEqual(line, '🐕💻');
+ }));
+ fi.emit('data', '\n');
+ rli.close();
+ }
+
+ // Two astral characters right
+ {
+ const fi = new FakeInput();
+ const rli = new readline.Interface({
+ input: fi,
+ output: fi,
+ prompt: '',
+ terminal: terminal
+ });
+ fi.emit('data', '💻');
+
+ // move left one character/code point
+ fi.emit('keypress', '.', { name: 'right' });
+ let cursorPos = rli._getCursorPos();
+ assert.strictEqual(cursorPos.rows, 0);
+ if (common.hasIntl) {
+ assert.strictEqual(cursorPos.cols, 2);
+ } else {
+ assert.strictEqual(cursorPos.cols, 1);
+ // Fix cursor position without internationalization
+ fi.emit('keypress', '.', { name: 'right' });
+ }
+
+ fi.emit('data', '🐕');
+ cursorPos = rli._getCursorPos();
+ assert.strictEqual(cursorPos.rows, 0);
+ if (common.hasIntl) {
+ assert.strictEqual(cursorPos.cols, 4);
+ } else {
+ assert.strictEqual(cursorPos.cols, 2);
+ }
+
+ rli.on('line', common.mustCall((line) => {
+ assert.strictEqual(line, '💻🐕');
+ }));
+ fi.emit('data', '\n');
+ rli.close();
+ }
+
  {
  // `wordLeft` and `wordRight`
  const fi = new FakeInput();
@@ -791,6 +900,35 @@ function isWarned(emitter) {
  rli.close();
  }
 
+ // deleteLeft astral character
+ {
+ const fi = new FakeInput();
+ const rli = new readline.Interface({
+ input: fi,
+ output: fi,
+ prompt: '',
+ terminal: terminal
+ });
+ fi.emit('data', '💻');
+ let cursorPos = rli._getCursorPos();
+ assert.strictEqual(cursorPos.rows, 0);
+ if (common.hasIntl) {
+ assert.strictEqual(cursorPos.cols, 2);
+ } else {
+ assert.strictEqual(cursorPos.cols, 1);
+ }
+ // Delete left character
+ fi.emit('keypress', '.', { ctrl: true, name: 'h' });
+ cursorPos = rli._getCursorPos();
+ assert.strictEqual(cursorPos.rows, 0);
+ assert.strictEqual(cursorPos.cols, 0);
+ rli.on('line', common.mustCall((line) => {
+ assert.strictEqual(line, '');
+ }));
+ fi.emit('data', '\n');
+ rli.close();
+ }
+
  // deleteRight
  {
  const fi = new FakeInput();
@@ -820,6 +958,34 @@ function isWarned(emitter) {
  rli.close();
  }
 
+ // deleteRight astral character
+ {
+ const fi = new FakeInput();
+ const rli = new readline.Interface({
+ input: fi,
+ output: fi,
+ prompt: '',
+ terminal: terminal
+ });
+ fi.emit('data', '💻');
+
+ // Go to the start of the line
+ fi.emit('keypress', '.', { ctrl: true, name: 'a' });
+ let cursorPos = rli._getCursorPos();
+ assert.strictEqual(cursorPos.rows, 0);
+ assert.strictEqual(cursorPos.cols, 0);
+
+ // Delete right character
+ fi.emit('keypress', '.', { ctrl: true, name: 'd' });
+ cursorPos = rli._getCursorPos();
+ assert.strictEqual(cursorPos.rows, 0);
+ assert.strictEqual(cursorPos.cols, 0);
+ rli.on('line', common.mustCall((line) => {
+ assert.strictEqual(line, '');
+ }));
+ fi.emit('data', '\n');
+ rli.close();
+ }
 
  // deleteLineLeft
  {