Merge pull request #46 from KentoMoriwaki/emoji2

Simpler handling surrogate pairs

diff.js

@@ -28,6 +28,9 @@ import { equalityStrict } from './function.js'
  * @template T
  */
 
+const highSurrogateRegex = /[\uD800-\uDBFF]/
+const lowSurrogateRegex = /[\uDC00-\uDFFF]/
+
 /**
  * Create a diff between two strings. This diff implementation is highly
  * efficient, but not very sophisticated.
@@ -44,9 +47,13 @@ export const simpleDiffString = (a, b) => {
   while (left < a.length && left < b.length && a[left] === b[left]) {
     left++
   }
+  // If the last same character is a high surrogate, we need to rollback to the previous character
+  if (highSurrogateRegex.test(a[left - 1])) left--
   while (right + left < a.length && right + left < b.length && a[a.length - right - 1] === b[b.length - right - 1]) {
     right++
   }
+  // If the last same character is a low surrogate, we need to rollback to the previous character
+  if (lowSurrogateRegex.test(a[a.length - right])) right--
   return {
     index: left,
     remove: a.length - left - right,
@@ -111,6 +118,8 @@ export const simpleDiffStringWithCursor = (a, b, cursor) => {
   ) {
     left++
   }
+  // If the last same character is a high surrogate, we need to rollback to the previous character
+  if (highSurrogateRegex.test(a[left - 1])) left--
   // Iterate right to the left until we find a changed character
   while (
     right + left < a.length &&
@@ -119,6 +128,8 @@ export const simpleDiffStringWithCursor = (a, b, cursor) => {
   ) {
     right++
   }
+  // If the last same character is a low surrogate, we need to rollback to the previous character
+  if (lowSurrogateRegex.test(a[a.length - right])) right--
   // Try to iterate left further to the right without caring about the current cursor position
   while (
     right + left < a.length &&
@@ -127,6 +138,7 @@ export const simpleDiffStringWithCursor = (a, b, cursor) => {
   ) {
     left++
   }
+  if (highSurrogateRegex.test(a[left - 1])) left--
   return {
     index: left,
     remove: a.length - left - right,

diff.test.js

@@ -2,7 +2,6 @@ import { simpleDiffString, simpleDiffArray, simpleDiffStringWithCursor } from '.
 import * as prng from './prng.js'
 import * as f from './function.js'
 import * as t from './testing.js'
-import * as object from './object.js'
 import * as str from './string.js'
 
 /**
@@ -14,8 +13,12 @@ function runDiffTest (a, b, expected) {
   const result = simpleDiffString(a, b)
   t.compare(result, expected)
   t.compare(result, simpleDiffStringWithCursor(a, b, a.length)) // check that the withCursor approach returns the same result
-  const arrResult = simpleDiffArray(a.split(''), b.split(''))
-  t.compare(arrResult, object.assign({}, result, { insert: result.insert.split('') }))
+  const recomposed = str.splice(a, result.index, result.remove, result.insert)
+  t.compareStrings(recomposed, b)
+  const arrResult = simpleDiffArray(Array.from(a), Array.from(b))
+  const arrRecomposed = Array.from(a)
+  arrRecomposed.splice(arrResult.index, arrResult.remove, ...arrResult.insert)
+  t.compareStrings(arrRecomposed.join(''), b)
 }
 
 /**
@@ -30,6 +33,11 @@ export const testDiffing = tc => {
   runDiffTest('abc', 'xyz', { index: 0, remove: 3, insert: 'xyz' })
   runDiffTest('axz', 'au', { index: 1, remove: 2, insert: 'u' })
   runDiffTest('ax', 'axy', { index: 2, remove: 0, insert: 'y' })
+  // These strings share high-surrogate characters
+  runDiffTest('\u{d83d}\u{dc77}'/* '👷' */, '\u{d83d}\u{dea7}\u{d83d}\u{dc77}'/* '🚧👷' */, { index: 0, remove: 0, insert: '🚧' })
+  runDiffTest('\u{d83d}\u{dea7}\u{d83d}\u{dc77}'/* '🚧👷' */, '\u{d83d}\u{dc77}'/* '👷' */, { index: 0, remove: 2, insert: '' })
+  // These strings share low-surrogate characters
+  runDiffTest('\u{d83d}\u{dfe6}\u{d83d}\u{dfe6}'/* '🟦🟦' */, '\u{d83c}\u{dfe6}\u{d83d}\u{dfe6}'/* '🏦🟦' */, { index: 0, remove: 2, insert: '🏦' })
 }
 
 /**
@@ -74,6 +82,20 @@ export const testSimpleDiffWithCursor = tc => {
     const recomposed = str.splice(initial, change.index, change.remove, change.insert)
     t.compareStrings(expected, recomposed)
   }
+  {
+    const initial = '🚧🚧🚧'
+    const change = simpleDiffStringWithCursor(initial, '🚧🚧', 2) // Should delete after the midst of 🚧
+    t.compare(change, { insert: '', remove: 2, index: 2 })
+    const recomposed = str.splice(initial, change.index, change.remove, change.insert)
+    t.compareStrings('🚧🚧', recomposed)
+  }
+  {
+    const initial = '🚧👷🚧👷'
+    const change = simpleDiffStringWithCursor(initial, '🚧🚧', 2) // Should delete after the first 🚧 and insert 🚧
+    t.compare(change, { insert: '🚧', remove: 6, index: 2 })
+    const recomposed = str.splice(initial, change.index, change.remove, change.insert)
+    t.compareStrings('🚧🚧', recomposed)
+  }
 }
 
 /**