Merge pull request #2 from cto-af/break

Fix break logic to be more useful

Author: hildjj

README.md

@@ -19,7 +19,7 @@ const sw = new StringWidth()
 sw.width('foo') // 3
 sw.width('\u{1F4A9}') // 2: Emoji take two cells
 sw.width('#\ufe0f\u20e3') // 2: More complicated emoji
-sw.break('foobar', 3) // ['foo', 'bar']
+sw.break('foobar', 3) // [{string: 'foo', cells: 3}, {string: 'bar', cells: 3}]
 
 const custom = new StringWidth({
   locale: 'ko-KR',

lib/index.js

@@ -142,30 +142,59 @@ export class StringWidth {
   }
 
   /**
-   * Break a string into two parts, such that the first part is always
-   * shorter or equal to width in display cells.
+   * Break a string into multiple parts, such that each part has maximum
+   * cell length of width, unless a particular grapheme cluster is longer
+   * than width.  Therefore, width should probably be 4 or higher to get
+   * the expected results.
    *
    * @param {string} str String to segment
-   * @param {number} width Maximum number of display cells for first chunk
-   * @returns {[string, string]}
+   * @param {number} width Maximum number of display cells for chunks
+   * @returns {{string: string, cells: number}[]}
    */
   break(str, width) {
+    if (width < 1) {
+      throw new RangeError(`Width must be >= 1.  Got ${width}.`)
+    }
+
+    /** @type {{string: string, cells: number}[]} */
+    const ret = []
+    let string = ''
+    let cells = 0
+
     // Fast-path shortcut for all-ASCII
     if (ASCII_REGEX.test(str)) {
-      return [str.slice(0, width), str.slice(width)]
+      for (cells = 0; cells < str.length; cells += width) {
+        string = str.slice(cells, cells + width)
+        ret.push({
+          string,
+          cells: string.length, // Might be less than width for last chunk
+        })
+      }
+      return ret
     }
 
-    let ret = ''
-    let rest = ''
-    let cur = 0 // Cells
-    for (const {segment, index} of this.#graphemes.segment(str)) {
-      cur += this.#segmentWidth(segment)
-      if (cur > width) { // Cells
-        rest = str.slice(index + segment.length - 1) // Chars
-        break
+    for (const segment of this.graphemes(str)) {
+      const w = this.#segmentWidth(segment)
+      if (w > width) {
+        // Skinny width, fat grapheme cluster
+        if (cells > 0) {
+          ret.push({string, cells})
+          string = ''
+          cells = 0
+        }
+        ret.push({string: segment, cells: w})
+      } else if (cells + w > width) {
+        ret.push({string, cells})
+        string = segment
+        cells = w
+      } else {
+        string += segment
+        cells += w
       }
-      ret += segment
     }
-    return [ret, rest]
+    if (cells > 0) {
+      ret.push({string, cells})
+    }
+    return ret
   }
 }

test/index.test.js

@@ -53,13 +53,29 @@ describe('info', () => {
 
 describe('string breaks', () => {
   it('handles ascii only', () => {
-    assert.deepEqual(sw.break('foo', 10), ['foo', ''])
-    assert.deepEqual(sw.break('foobar', 3), ['foo', 'bar'])
-    assert.deepEqual(sw.break('foo', 0), ['', 'foo'])
+    assert.deepEqual(sw.break('foo', 10), [{string: 'foo', cells: 3}])
+    assert.deepEqual(sw.break('foobar', 3), [
+      {string: 'foo', cells: 3},
+      {string: 'bar', cells: 3},
+    ])
+    assert.throws(() => sw.break('foo', 0))
   })
   it('handles non-ascii', () => {
-    assert.deepEqual(sw.break('foo\u0308', 10), ['foo\u0308', ''])
-    assert.deepEqual(sw.break('foo\u0308ba\u0308r', 3), ['foo\u0308', 'ba\u0308r'])
-    assert.deepEqual(sw.break('foo\u0308', 0), ['', 'foo\u0308'])
+    assert.deepEqual(sw.break('foo\u0308', 10), [
+      {string: 'foo\u0308', cells: 3},
+    ])
+    assert.deepEqual(sw.break('foo\u0308ba\u0308r', 3), [
+      {string: 'foo\u0308', cells: 3},
+      {string: 'ba\u0308r', cells: 3},
+    ])
+    assert.deepEqual(sw.break('\u{1F1F9}\u{1F1FC}b', 1), [
+      {string: '\u{1F1F9}\u{1F1FC}', cells: 2},
+      {string: 'b', cells: 1},
+    ])
+    assert.deepEqual(sw.break('a\u{1F1F9}\u{1F1FC}b', 1), [
+      {string: 'a', cells: 1},
+      {string: '\u{1F1F9}\u{1F1FC}', cells: 2},
+      {string: 'b', cells: 1},
+    ])
   })
 })