Added UiString.charStartPos

Author: viktor-podzigun

src/UiString.d.ts

@@ -1,6 +1,13 @@
 export interface UiString {
   strWidth(): number;
   toString(): string;
+  charStartPos(from: number): UiCharStartPos;
   slice(from: number, until: number): string;
   ensureWidth(width: number, padCh: string): string;
 }
+
+export interface UiCharStartPos {
+  readonly lcw: number;
+  readonly pos: number;
+  readonly rcw: number;
+}

src/UiString.mjs

@@ -1,3 +1,6 @@
+/**
+ * @typedef {import('./UiString').UiCharStartPos} UiCharStartPos
+ */
 import Blessed from "@farjs/blessed";
 
 const { unicode } = Blessed;
@@ -32,10 +35,10 @@ function UiString(str) {
       cw = unicode.charWidth(str, i);
 
       if (sw + cw <= width) {
-        if (
-          unicode.isSurrogate(str, i) ||
-          (i + 1 < str.length && unicode.isCombining(str, i + 1))
-        ) {
+        if (unicode.isSurrogate(str, i)) {
+          i += 1;
+        }
+        while (i + 1 < str.length && unicode.charWidth(str, i + 1) === 0) {
           i += 1;
         }
         i += 1;
@@ -86,6 +89,22 @@ function UiString(str) {
 
     toString: () => str,
 
+    charStartPos: (from) => {
+      if (strWidth() === 0) {
+        return { lcw: 0, pos: 0, rcw: 0 };
+      }
+
+      const pos = Math.min(Math.max(from, 0), strWidth());
+      const { i, sw, cw } = skipWidth(0, pos === 0 ? 1 : pos);
+      if (pos === 0) {
+        return { lcw: 0, pos, rcw: cw };
+      }
+      const start = sw + cw;
+      const next = start > pos ? i + 1 : i;
+      const rcw = start === strWidth() ? 0 : skipWidth(next, 1).cw;
+      return { lcw: cw, pos: start, rcw };
+    },
+
     slice: (from, until) => {
       const start = from > 0 ? skipWidth(0, from).i : 0;
       const { i: end } = skipWidth(start, until - from);

test/UiString.test.mjs

@@ -1,3 +1,6 @@
+/**
+ * @typedef {import('../src/UiString').UiCharStartPos} UiCharStartPos
+ */
 import Blessed from "@farjs/blessed";
 import assert from "node:assert/strict";
 import UiString from "../src/UiString.mjs";
@@ -32,6 +35,60 @@ describe("UiString.test.mjs", () => {
     assert.deepEqual(UiString(str).toString(), str);
   });
 
+  it("should return left/right char widths and start pos when charStartPos", () => {
+    /**
+     * @param {string} str
+     * @param {number} pos
+     * @param {UiCharStartPos} expected
+     */
+    function check(str, pos, expected) {
+      assert.deepEqual(UiString(str).charStartPos(pos), expected);
+    }
+
+    //when & then
+    check("", 0, { lcw: 0, pos: 0, rcw: 0 });
+    check("abc", -1, { lcw: 0, pos: 0, rcw: 1 });
+    check("abc", 0, { lcw: 0, pos: 0, rcw: 1 });
+    check("abc", 1, { lcw: 1, pos: 1, rcw: 1 });
+    check("abc", 2, { lcw: 1, pos: 2, rcw: 1 });
+    check("abc", 3, { lcw: 1, pos: 3, rcw: 0 });
+    check("abc", 4, { lcw: 1, pos: 3, rcw: 0 });
+    check("й", 0, { lcw: 0, pos: 0, rcw: 1 });
+    check("й", 1, { lcw: 1, pos: 1, rcw: 0 });
+    check("й", 2, { lcw: 1, pos: 1, rcw: 0 });
+    check("aй", 0, { lcw: 0, pos: 0, rcw: 1 });
+    check("aй", 1, { lcw: 1, pos: 1, rcw: 1 });
+    check("aй", 2, { lcw: 1, pos: 2, rcw: 0 });
+    check("йa", 0, { lcw: 0, pos: 0, rcw: 1 });
+    check("йa", 1, { lcw: 1, pos: 1, rcw: 1 });
+    check("йa", 2, { lcw: 1, pos: 2, rcw: 0 });
+    check("\uD83C\uDF31", 0, { lcw: 0, pos: 0, rcw: 2 });
+    check("\uD83C\uDF31", 1, { lcw: 2, pos: 2, rcw: 0 });
+    check("\uD83C\uDF31", 2, { lcw: 2, pos: 2, rcw: 0 });
+    check("a\uD83C\uDF31", 0, { lcw: 0, pos: 0, rcw: 1 });
+    check("a\uD83C\uDF31", 1, { lcw: 1, pos: 1, rcw: 2 });
+    check("a\uD83C\uDF31", 2, { lcw: 2, pos: 3, rcw: 0 });
+    check("a\uD83C\uDF31", 3, { lcw: 2, pos: 3, rcw: 0 });
+    check("\uff01", 0, { lcw: 0, pos: 0, rcw: 2 });
+    check("\uff01", 1, { lcw: 2, pos: 2, rcw: 0 });
+    check("\uff01", 2, { lcw: 2, pos: 2, rcw: 0 });
+    check("a\uff01b", 0, { lcw: 0, pos: 0, rcw: 1 });
+    check("a\uff01b", 1, { lcw: 1, pos: 1, rcw: 2 });
+    check("a\uff01b", 2, { lcw: 2, pos: 3, rcw: 1 });
+    check("a\uff01b", 3, { lcw: 2, pos: 3, rcw: 1 });
+    check("a\uff01b", 4, { lcw: 1, pos: 4, rcw: 0 });
+    check("\u200D", 0, { lcw: 0, pos: 0, rcw: 0 });
+    check("\u200Dй", 0, { lcw: 0, pos: 0, rcw: 1 });
+    check("\u200Dй", 1, { lcw: 1, pos: 1, rcw: 0 });
+    check("\u200Dй", 2, { lcw: 1, pos: 1, rcw: 0 });
+    check("\u200Dй\u200Dй", 0, { lcw: 0, pos: 0, rcw: 1 });
+    check("\u200Dй\u200Dй", 1, { lcw: 1, pos: 1, rcw: 1 });
+    check("\u200Dй\u200Dй", 2, { lcw: 1, pos: 2, rcw: 0 });
+    check("double 🉐", 4, { lcw: 1, pos: 4, rcw: 1 });
+    check("double 🉐", 7, { lcw: 1, pos: 7, rcw: 2 });
+    check("double 🉐", 8, { lcw: 2, pos: 9, rcw: 0 });
+  });
+
   it("should return part of str when slice", () => {
     //given
     const str = "abcd";
@@ -54,6 +111,8 @@ describe("UiString.test.mjs", () => {
 
   it("should handle combining chars when slice", () => {
     //given
+    assert.deepEqual(unicode.isCombining("\u200D", 0), true);
+    assert.deepEqual(unicode.charWidth("\u200D", 0), 0);
     assert.deepEqual(unicode.isCombining("й", 0), false);
     assert.deepEqual(unicode.isCombining("й", 1), true);
     assert.deepEqual(unicode.strWidth("й"), 1);
@@ -63,14 +122,37 @@ describe("UiString.test.mjs", () => {
     assert.deepEqual(UiString("Валютный").slice(6, 7), "ы");
     assert.deepEqual(UiString("Валютный").slice(7, 8), "й");
     assert.deepEqual(UiString("й").slice(0, 1), "й");
+    assert.deepEqual(UiString("й\u200D").slice(0, 1), "й\u200D");
+    assert.deepEqual(UiString("й\u200D\u200Db").slice(0, 1), "й\u200D\u200D");
+    assert.deepEqual(UiString("й\u200D\u200Db").slice(1, 2), "b");
     assert.deepEqual(UiString("1й").slice(0, 1), "1");
     assert.deepEqual(UiString("1й").slice(0, 2), "1й");
     assert.deepEqual(UiString("й2").slice(0, 2), "й2");
     assert.deepEqual(UiString("й2").slice(0, 1), "й");
     assert.deepEqual(UiString("й2").slice(1, 2), "2");
   });
 
-  it("should handle surrogate chars when slice", () => {
+  it("should handle high/low surrogate chars when slice", () => {
+    //given
+    assert.deepEqual(unicode.isSurrogate("\uD83C", 0), false);
+    assert.deepEqual(unicode.isSurrogate("\uD83Ca", 0), false);
+    assert.deepEqual(unicode.isSurrogate("\uD83C\uDF31", 0), true);
+    assert.deepEqual(unicode.isSurrogate("\uD83C\uDF31", 1), false);
+    assert.deepEqual(unicode.charWidth("\uD800", 0), 0);
+    assert.deepEqual(unicode.charWidth("\uD83C", 0), 0);
+    assert.deepEqual(unicode.charWidth("\uDBFF", 0), 0);
+    assert.deepEqual(unicode.charWidth("\uDC00", 0), 0);
+    assert.deepEqual(unicode.charWidth("\uDFFF", 0), 0);
+    assert.deepEqual(unicode.isCombining("\uD83C", 0), false);
+
+    //when & then
+    assert.deepEqual(UiString("\uD83C").slice(0, 1), "\uD83C");
+    assert.deepEqual(UiString("a\uD83C").slice(0, 1), "a\uD83C");
+    assert.deepEqual(UiString("a\uD83C\uD800b").slice(0, 1), "a\uD83C\uD800");
+    assert.deepEqual(UiString("a\uD83C\uD800b").slice(1, 2), "b");
+  });
+
+  it("should handle surrogate pairs when slice", () => {
     //given
     assert.deepEqual(unicode.isSurrogate("\uD83C\uDF31", 0), true);
     assert.deepEqual(unicode.isSurrogate("\uD83C\uDF31", 1), false);