readline: add unicode line separator option

Signed-off-by: Scarab Systems <scarab.systems@yahoo.com>

Author: scarab-systems

doc/api/readline.md

@@ -120,8 +120,10 @@ added: v0.1.98
 -->
 
 The `'line'` event is emitted whenever the `input` stream receives an
-end-of-line input (`\n`, `\r`, or `\r\n`). This usually occurs when the user
-presses <kbd>Enter</kbd> or <kbd>Return</kbd>.
+end-of-line input (`\n`, `\r`, or `\r\n`). By default, Unicode line separator
+(`\u2028`) and paragraph separator (`\u2029`) characters are also treated as
+end-of-line input. This usually occurs when the user presses <kbd>Enter</kbd>
+or <kbd>Return</kbd>.
 
 The `'line'` event is also emitted if new data has been read from a stream and
 that stream ends without a final end-of-line marker.
@@ -716,6 +718,8 @@ added: v17.0.0
     `100`. It can be set to `Infinity`, in which case `\r` followed by `\n`
     will always be considered a single newline (which may be reasonable for
     [reading files][] with `\r\n` line delimiter). **Default:** `100`.
+  * `unicodeLineSeparators` {boolean} If `true`, `\u2028` and `\u2029` will be
+    treated as end-of-line input. **Default:** `true`.
   * `escapeCodeTimeout` {number} The duration `readlinePromises` will wait for a
     character (when reading an ambiguous key sequence in milliseconds one that
     can both form a complete key sequence using the input read so far and can
@@ -981,6 +985,8 @@ changes:
     `100`. It can be set to `Infinity`, in which case `\r` followed by `\n`
     will always be considered a single newline (which may be reasonable for
     [reading files][] with `\r\n` line delimiter). **Default:** `100`.
+  * `unicodeLineSeparators` {boolean} If `true`, `\u2028` and `\u2029` will be
+    treated as end-of-line input. **Default:** `true`.
   * `escapeCodeTimeout` {number} The duration `readline` will wait for a
     character (when reading an ambiguous key sequence in milliseconds one that
     can both form a complete key sequence using the input read so far and can

lib/internal/readline/interface.js

@@ -44,6 +44,7 @@ const {
 
 const {
   validateAbortSignal,
+  validateBoolean,
   validateString,
   validateUint32,
 } = require('internal/validators');
@@ -86,6 +87,7 @@ const kMincrlfDelay = 100;
  *  - \u2029 (Unicode 'PARAGRAPH SEPARATOR')
  */
 const lineEnding = /\r?\n|\r(?!\n)|\u2028|\u2029/g;
+const crlfLineEnding = /\r?\n|\r(?!\n)/g;
 
 const kLineObjectStream = Symbol('line object stream');
 const kQuestionCancel = Symbol('kQuestionCancel');
@@ -116,6 +118,7 @@ const kMoveUpOrHistoryPrev = Symbol('_moveUpOrHistoryPrev');
 const kInsertString = Symbol('_insertString');
 const kLine = Symbol('_line');
 const kLine_buffer = Symbol('_line_buffer');
+const kLineEnding = Symbol('_lineEnding');
 const kKillRing = Symbol('_killRing');
 const kKillRingCursor = Symbol('_killRingCursor');
 const kMoveCursor = Symbol('_moveCursor');
@@ -172,6 +175,7 @@ function InterfaceConstructor(input, output, completer, terminal) {
   let crlfDelay;
   let prompt = '> ';
   let signal;
+  let unicodeLineSeparators = true;
 
   if (input?.input) {
     // An options object was given
@@ -208,6 +212,13 @@ function InterfaceConstructor(input, output, completer, terminal) {
     }
 
     crlfDelay = input.crlfDelay;
+    if (input.unicodeLineSeparators !== undefined) {
+      validateBoolean(
+        input.unicodeLineSeparators,
+        'options.unicodeLineSeparators',
+      );
+      unicodeLineSeparators = input.unicodeLineSeparators;
+    }
     input = input.input;
 
     input.size = historySize;
@@ -250,6 +261,7 @@ function InterfaceConstructor(input, output, completer, terminal) {
     MathMax(kMincrlfDelay, crlfDelay) :
     kMincrlfDelay;
   this.completer = completer;
+  this[kLineEnding] = unicodeLineSeparators ? lineEnding : crlfLineEnding;
 
   this.setPrompt(prompt);
 
@@ -623,6 +635,7 @@ class Interface extends InterfaceConstructor {
     }
 
     // Run test() on the new string chunk, not on the entire line buffer.
+    const lineEnding = this[kLineEnding];
     let newPartContainsEnding = RegExpPrototypeExec(lineEnding, string);
     if (newPartContainsEnding !== null) {
       if (this[kLine_buffer]) {
@@ -1530,6 +1543,7 @@ class Interface extends InterfaceConstructor {
         default:
           if (typeof s === 'string' && s) {
             // Erase state of previous searches.
+            const lineEnding = this[kLineEnding];
             lineEnding.lastIndex = 0;
             let nextMatch;
             // Keep track of the end of the last match.

lib/readline.js

@@ -199,6 +199,7 @@ Interface.prototype.question[promisify.custom] = function question(query, option
  *   removeHistoryDuplicates?: boolean;
  *   prompt?: string;
  *   crlfDelay?: number;
+ *   unicodeLineSeparators?: boolean;
  *   escapeCodeTimeout?: number;
  *   tabSize?: number;
  *   signal?: AbortSignal;

test/parallel/test-readline-line-separators.js

@@ -6,13 +6,45 @@ const { Readable } = require('node:stream');
 
 const str = '012\n345\r67\r\n89\u{2028}ABC\u{2029}DEF';
 
-const rli = new readline.Interface({
-  input: Readable.from(str),
-});
+// Unicode line and paragraph separators are line endings by default.
+{
+  const rli = new readline.Interface({
+    input: Readable.from(str),
+  });
 
-const linesRead = [];
-rli.on('line', (line) => linesRead.push(line));
+  const linesRead = [];
+  rli.on('line', (line) => linesRead.push(line));
 
-rli.on('close', common.mustCall(() => {
-  assert.deepStrictEqual(linesRead, ['012', '345', '67', '89', 'ABC', 'DEF']);
-}));
+  rli.on('close', common.mustCall(() => {
+    assert.deepStrictEqual(linesRead, ['012', '345', '67', '89', 'ABC', 'DEF']);
+  }));
+}
+
+// The option allows file formats such as JSONL to keep Unicode separators
+// inside record contents while still splitting on CR, LF, and CRLF.
+{
+  const rli = new readline.Interface({
+    input: Readable.from(str),
+    unicodeLineSeparators: false,
+  });
+
+  const linesRead = [];
+  rli.on('line', (line) => linesRead.push(line));
+
+  rli.on('close', common.mustCall(() => {
+    assert.deepStrictEqual(
+      linesRead,
+      ['012', '345', '67', '89\u2028ABC\u2029DEF'],
+    );
+  }));
+}
+
+assert.throws(
+  () => new readline.Interface({
+    input: Readable.from(str),
+    unicodeLineSeparators: 'false',
+  }),
+  {
+    code: 'ERR_INVALID_ARG_TYPE',
+  },
+);