fix Unicode planes (#52)

.changeset/modern-cooks-laugh.md

@@ -0,0 +1,5 @@
+---
+"unicode-segmenter": patch
+---
+
+Fix `isSMP`, and add more plane utils (`isSIP`, `isTIP`, `isSSP`)

src/grapheme.js

@@ -14,7 +14,7 @@
 // @ts-check
 
 import { bsearchRange } from './core.js';
-import { isSMP } from './utils.js';
+import { isBMP } from './utils.js';
 import {
   searchGraphemeCategory,
   GraphemeCategory,
@@ -96,7 +96,7 @@ export function* graphemeSegments(input) {
 
   while (true) {
     segment += input[cursor++];
-    if (isSMP(cp)) {
+    if (!isBMP(cp)) {
       segment += input[cursor++];
     }
 

src/utils.js

@@ -89,5 +89,35 @@ export function isBMP(c) {
  * @return {boolean}
  */
 export function isSMP(c) {
-  return 0xffff < c;
+  return 0x10000 <= c && c <= 0x1ffff;
+}
+
+/**
+ * Check if given code point is within the SIP(Supplementary Ideographic Plane)
+ *
+ * @param {number} c Unicode code point
+ * @return {boolean}
+ */
+export function isSIP(c) {
+  return 0x20000 <= c && c <= 0x2ffff;
+}
+
+/**
+ * Check if given code point is within the TIP(Tertiary Ideographic Plane)
+ *
+ * @param {number} c Unicode code point
+ * @return {boolean}
+ */
+export function isTIP(c) {
+  return 0x30000 <= c && c <= 0x3ffff;
+}
+
+/**
+ * Check if given code point is within the SSP(Supplementary Special-purpose Plane)
+ *
+ * @param {number} c Unicode code point
+ * @return {boolean}
+ */
+export function isSSP(c) {
+  return 0xe0000 <= c && c <= 0xeffff;
 }

test/utils.js

@@ -9,6 +9,9 @@ import {
   takeCodePoint,
   isBMP,
   isSMP,
+  isSIP,
+  isTIP,
+  isSSP,
 } from 'unicode-segmenter/utils';
 
 fc.configureGlobal({
@@ -83,20 +86,50 @@ test('takeChar', async t => {
 
 test('isBMP', () => {
   fc.assert(
+    fc.property(fc.integer({ min: 0, max: 0xffff }),
     // @ts-ignore
-    fc.property(fc.fullUnicode(), (data) => {
-      // @ts-ignore
-      assert.equal(isBMP(data.codePointAt(0)) ? 1 : 2, data.length);
+    (data) => {
+      assert.ok(isBMP(data));
     }),
   );
 });
 
 test('isSMP', () => {
   fc.assert(
+    fc.property(fc.integer({ min: 0x10000, max: 0x1ffff }),
     // @ts-ignore
-    fc.property(fc.fullUnicode(), (data) => {
-      // @ts-ignore
-      assert.equal(isSMP(data.codePointAt(0)) ? 2 : 1, data.length);
+    (data) => {
+      assert.ok(isSMP(data));
+    }),
+  );
+});
+
+test('isSIP', () => {
+  fc.assert(
+    fc.property(fc.integer({ min: 0x20000, max: 0x2ffff }),
+    // @ts-ignore
+    (data) => {
+      assert.ok(isSIP(data));
+    }),
+  );
+});
+
+test('isTIP', () => {
+  fc.assert(
+    fc.property(fc.integer({ min: 0x30000, max: 0x3ffff }),
+    // @ts-ignore
+    (data) => {
+      assert.ok(isTIP(data));
+    }),
+  );
+});
+
+test('isSSP', () => {
+  fc.assert(
+    fc.property(fc.integer({ min: 0xe0000, max: 0xeffff }),
+    // @ts-ignore
+    (data) => {
+      assert.ok(isSSP(data));
     }),
   );
 });