[js] Turn some uses of the graphemeBreaker modules into use of our homebrew regexp

Author: pmurias

src/vm/js/nqp-runtime/.bignum.js.swo

@@ -43,6 +43,8 @@ const stripMarks = require('./strip-marks.js');
 
 const codecs = require('./codecs.js');
 
+const graphemeRegexp = require('./graphemes').regexp;
+
 const foldCase = require('fold-case');
 
 const xregexp = require('xregexp');
@@ -1722,31 +1724,45 @@ op.flip = function(str) {
 };
 
 op.charsnfg = function(str) {
-  return graphemeBreaker.countBreaks(str);
+  let count = 0;
+  graphemeRegexp.lastIndex = 0;
+  while (graphemeRegexp.test(str)) {
+    count++;
+  }
+  return count;
 };
 
 op.substr3nfg = function(str, start, length) {
   let startGraphemes = start;
-  let startChars = 0;
+
+  graphemeRegexp.lastIndex = 0;
   while (startGraphemes--) {
-    startChars = graphemeBreaker.nextBreak(str, startChars);
+    if (!graphemeRegexp.test(str)) return '';
   }
 
-  let lengthGraphemes = length;
+  let startChars = graphemeRegexp.lastIndex;
+
   let substringEnd = startChars;
+  let lengthGraphemes = length;
   while (lengthGraphemes--) {
-     substringEnd = graphemeBreaker.nextBreak(str, substringEnd);
+    if (graphemeRegexp.test(str)) {
+      substringEnd = graphemeRegexp.lastIndex;
+    } else {
+      break;
+    }
   }
 
   return str.substring(startChars, substringEnd);
 };
 
 op.substr2nfg = function(str, start) {
   let startGraphemes = start;
-  let startChars = 0;
+
+  graphemeRegexp.lastIndex = 0;
   while (startGraphemes--) {
-    startChars = graphemeBreaker.nextBreak(str, startChars);
+    if (!graphemeRegexp.test(str)) return '';
   }
+  let startChars = graphemeRegexp.lastIndex;
 
   return str.substr(startChars);
 };

src/vm/js/nqp-runtime/.core.js.swn

@@ -0,0 +1,24 @@
+'use strict';
+const data = require('nqp-unicode-data');
+
+function build(source) {
+  return source.replace(/ /g, '').replace(/<([^>]+)>/g, function(match, part) {
+    const content = parts[part] ? build(parts[part]) : data.regex(part);
+    if (content === undefined) throw "Can't do anything with: " + part;
+    return '(?:' + content + ')';
+  });
+}
+
+const parts = {};
+
+parts.CRLF = '<CR><LF>';
+parts.RISequence = '<Regional_Indicator><Regional_Indicator>';
+parts.HangulSyllable = '<L>*<V>+<T>*|<L>*<LV><V>*<T>*|<L>*<LVT><T>*|<L>+|<T>+';
+parts.Grapheme_Extend = '<Extend> | (?:<ZWJ><Glue_After_Zwj>) | (?:<ZWJ><E_Base_GAZ><Extend>*<E_Modifier>?) | <ZWJ>'
+parts.Modified = '<Prepend>*(?:<RISequence>|<HangulSyllable>|(?:(?:<E_Base>|<E_Base_GAZ>)<Extend>*<E_Modifier>)|(?:(?!<Control>).))(<Grapheme_Extend>|<SpacingMark>)*';
+
+parts.Degenerate = '<ZWJ>(?:<Glue_After_Zwj>|<E_Base_GAZ><Extend>*<E_Modifier>?)';
+
+const graphemePattern = build('<CRLF>|<Degenerate>|<Modified>|[^]');
+
+exports.regexp = new RegExp(graphemePattern, 'yu');