readline,repl: add substring history search

doc/api/repl.md

@@ -22,10 +22,11 @@ be connected to any Node.js [stream][].
 
 Instances of [`repl.REPLServer`][] support automatic completion of inputs,
 completion preview, simplistic Emacs-style line editing, multi-line inputs,
-[ZSH][] like reverse-i-search, ANSI-styled output, saving and restoring current
-REPL session state, error recovery, and customizable evaluation functions.
-Terminals that do not support ANSI-styles and Emacs-style line editing
-automatically fall back to a limited feature set.
+[ZSH][]-like reverse-i-search, [ZSH][]-like substring-based history search,
+ANSI-styled output, saving and restoring current REPL session state, error
+recovery, and customizable evaluation functions. Terminals that do not support
+ANSI styles and Emacs-style line editing automatically fall back to a limited
+feature set.
 
 ### Commands and Special Keys
 

lib/internal/readline/utils.js

@@ -1,9 +1,8 @@
 'use strict';
 
 const {
-  Boolean,
-  NumberIsInteger,
   RegExp,
+  Symbol,
 } = primordials;
 
 // Regex used for ansi escape code splitting
@@ -17,9 +16,9 @@ const ansi = new RegExp(ansiPattern, 'g');
 
 const kUTF16SurrogateThreshold = 0x10000; // 2 ** 16
 const kEscape = '\x1b';
+const kSubstringSearch = Symbol('kSubstringSearch');
 
 let getStringWidth;
-let isFullWidthCodePoint;
 
 function CSI(strings, ...args) {
   let ret = `${kEscape}[`;
@@ -37,65 +36,65 @@ CSI.kClearToLineEnd = CSI`0K`;
 CSI.kClearLine = CSI`2K`;
 CSI.kClearScreenDown = CSI`0J`;
 
+// TODO(BridgeAR): Treat combined characters as single character, i.e,
+// 'a\u0301' and '\u0301a' (both have the same visual output).
+// Check Canonical_Combining_Class in
+// http://userguide.icu-project.org/strings/properties
+function charLengthLeft(str, i) {
+  if (i <= 0)
+    return 0;
+  if ((i > 1 && str.codePointAt(i - 2) >= kUTF16SurrogateThreshold) ||
+      str.codePointAt(i - 1) >= kUTF16SurrogateThreshold) {
+    return 2;
+  }
+  return 1;
+}
+
+function charLengthAt(str, i) {
+  if (str.length <= i) {
+    // Pretend to move to the right. This is necessary to autocomplete while
+    // moving to the right.
+    return 1;
+  }
+  return str.codePointAt(i) >= kUTF16SurrogateThreshold ? 2 : 1;
+}
+
 if (internalBinding('config').hasIntl) {
   const icu = internalBinding('icu');
-  getStringWidth = function getStringWidth(str, options) {
-    options = options || {};
-    if (NumberIsInteger(str)) {
-      // Provide information about the character with code point 'str'.
-      return icu.getStringWidth(
-        str,
-        Boolean(options.ambiguousAsFullWidth),
-        false
-      );
-    }
-    str = stripVTControlCharacters(String(str));
+  // icu.getStringWidth(string, ambiguousAsFullWidth, expandEmojiSequence)
+  // Defaults: ambiguousAsFullWidth = false; expandEmojiSequence = true;
+  // TODO(BridgeAR): Expose the options to the user. That is probably the
+  // best thing possible at the moment, since it's difficult to know what
+  // the receiving end supports.
+  getStringWidth = function getStringWidth(str) {
     let width = 0;
+    str = stripVTControlCharacters(str);
     for (let i = 0; i < str.length; i++) {
       // Try to avoid calling into C++ by first handling the ASCII portion of
       // the string. If it is fully ASCII, we skip the C++ part.
       const code = str.charCodeAt(i);
-      if (code < 127) {
-        width += code >= 32;
-        continue;
+      if (code >= 127) {
+        width += icu.getStringWidth(str.slice(i));
+        break;
       }
-      width += icu.getStringWidth(
-        str.slice(i),
-        Boolean(options.ambiguousAsFullWidth),
-        Boolean(options.expandEmojiSequence)
-      );
-      break;
+      width += code >= 32 ? 1 : 0;
     }
     return width;
   };
-  isFullWidthCodePoint =
-    function isFullWidthCodePoint(code, options) {
-      if (typeof code !== 'number')
-        return false;
-      return icu.getStringWidth(code, options) === 2;
-    };
 } else {
   /**
    * Returns the number of columns required to display the given string.
    */
   getStringWidth = function getStringWidth(str) {
-    if (NumberIsInteger(str))
-      return isFullWidthCodePoint(str) ? 2 : 1;
-
     let width = 0;
 
-    str = stripVTControlCharacters(String(str));
-
-    for (let i = 0; i < str.length; i++) {
-      const code = str.codePointAt(i);
-
-      if (code >= kUTF16SurrogateThreshold) { // Surrogates.
-        i++;
-      }
+    str = stripVTControlCharacters(str);
 
+    for (const char of str) {
+      const code = char.codePointAt(0);
       if (isFullWidthCodePoint(code)) {
         width += 2;
-      } else {
+      } else if (!isZeroWidthCodePoint(code)) {
         width++;
       }
     }
@@ -107,10 +106,10 @@ if (internalBinding('config').hasIntl) {
    * Returns true if the character represented by a given
    * Unicode code point is full-width. Otherwise returns false.
    */
-  isFullWidthCodePoint = function isFullWidthCodePoint(code) {
-    // Code points are derived from:
+  const isFullWidthCodePoint = (code) => {
+    // Code points are partially derived from:
     // http://www.unicode.org/Public/UNIDATA/EastAsianWidth.txt
-    return NumberIsInteger(code) && code >= 0x1100 && (
+    return code >= 0x1100 && (
       code <= 0x115f ||  // Hangul Jamo
       code === 0x2329 || // LEFT-POINTING ANGLE BRACKET
       code === 0x232a || // RIGHT-POINTING ANGLE BRACKET
@@ -137,10 +136,23 @@ if (internalBinding('config').hasIntl) {
       (code >= 0x1b000 && code <= 0x1b001) ||
       // Enclosed Ideographic Supplement
       (code >= 0x1f200 && code <= 0x1f251) ||
+      // Miscellaneous Symbols and Pictographs 0x1f300 - 0x1f5ff
+      // Emoticons 0x1f600 - 0x1f64f
+      (code >= 0x1f300 && code <= 0x1f64f) ||
       // CJK Unified Ideographs Extension B .. Tertiary Ideographic Plane
       (code >= 0x20000 && code <= 0x3fffd)
     );
   };
+
+  const isZeroWidthCodePoint = (code) => {
+    return code <= 0x1F || // C0 control codes
+      (code > 0x7F && code <= 0x9F) || // C1 control codes
+      (code >= 0x0300 && code <= 0x036F) || // Combining Diacritical Marks
+      (code >= 0x200B && code <= 0x200F) || // Modifying Invisible Characters
+      (code >= 0xFE00 && code <= 0xFE0F) || // Variation Selectors
+      (code >= 0xFE20 && code <= 0xFE2F) || // Combining Half Marks
+      (code >= 0xE0100 && code <= 0xE01EF); // Variation Selectors
+  };
 }
 
 /**
@@ -466,11 +478,12 @@ function commonPrefix(strings) {
 }
 
 module.exports = {
+  charLengthAt,
+  charLengthLeft,
   commonPrefix,
   emitKeys,
   getStringWidth,
-  isFullWidthCodePoint,
-  kUTF16SurrogateThreshold,
+  kSubstringSearch,
   stripVTControlCharacters,
   CSI
 };