feat!(@formatjs/cli-lib): repurpose en-XB for bidi pseudo locale (#3978)

BREAKING CHANGE: `en-XA` is a pseudo locale for accented and lengthned English with markers, and `en-XB` is now a bidi pseudo locale.

Author: pyrocat101

packages/cli-lib/src/pseudo_locale.ts

@@ -6,8 +6,26 @@ import {
   isPluralElement,
   isSelectElement,
   isTagElement,
+  LiteralElement,
 } from '@formatjs/icu-messageformat-parser'
 
+function forEachLiteralElement(
+  ast: MessageFormatElement[],
+  fn: (el: LiteralElement) => void
+): void {
+  ast.forEach(el => {
+    if (isLiteralElement(el)) {
+      fn(el)
+    } else if (isPluralElement(el) || isSelectElement(el)) {
+      for (const opt of Object.values(el.options)) {
+        forEachLiteralElement(opt.value, fn)
+      }
+    } else if (isTagElement(el)) {
+      forEachLiteralElement(el.children, fn)
+    }
+  })
+}
+
 export function generateXXLS(
   msg: string | MessageFormatElement[]
 ): MessageFormatElement[] {
@@ -24,16 +42,8 @@ export function generateXXAC(
   msg: string | MessageFormatElement[]
 ): MessageFormatElement[] {
   const ast = typeof msg === 'string' ? parse(msg) : msg
-  ast.forEach(el => {
-    if (isLiteralElement(el)) {
-      el.value = el.value.toUpperCase()
-    } else if (isPluralElement(el) || isSelectElement(el)) {
-      for (const opt of Object.values(el.options)) {
-        generateXXAC(opt.value)
-      }
-    } else if (isTagElement(el)) {
-      generateXXAC(el.children)
-    }
+  forEachLiteralElement(ast, el => {
+    el.value = el.value.toUpperCase()
   })
   return ast
 }
@@ -50,64 +60,104 @@ export function generateXXHA(
   return [{type: TYPE.literal, value: '[javascript]'}, ...ast]
 }
 
-const ASCII = 'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ'
-const ACCENTED_ASCII = 'âḃćḋèḟĝḫíĵǩĺṁńŏṗɋŕśṭůṿẘẋẏẓḀḂḈḊḔḞḠḢḬĴḴĻḾŊÕṔɊŔṠṮŨṼẄẌŸƵ'
+type _TupleOf<T, N extends number, R extends unknown[]> = R['length'] extends N
+  ? R
+  : _TupleOf<T, N, [T, ...R]>
+
+type Tuple<T, N extends number> = N extends N
+  ? number extends N
+    ? T[]
+    : _TupleOf<T, N, []>
+  : never
+
+type PseudoLocaleTransformMap = {
+  caps: Tuple<number, 26>
+  small: Tuple<number, 26>
+}
+
+const ACCENTED_MAP: PseudoLocaleTransformMap = {
+  // ȦƁƇḒḖƑƓĦĪĴĶĿḾȠǾƤɊŘŞŦŬṼẆẊẎẐ
+  // prettier-ignore
+  "caps": [550, 385, 391, 7698, 7702, 401, 403, 294, 298, 308, 310, 319, 7742, 544, 510, 420, 586, 344, 350, 358, 364, 7804, 7814, 7818, 7822, 7824],
+  // ȧƀƈḓḗƒɠħīĵķŀḿƞǿƥɋřşŧŭṽẇẋẏẑ
+  // prettier-ignore
+  "small": [551, 384, 392, 7699, 7703, 402, 608, 295, 299, 309, 311, 320, 7743, 414, 511, 421, 587, 345, 351, 359, 365, 7805, 7815, 7819, 7823, 7825],
+}
+
+const FLIPPED_MAP: PseudoLocaleTransformMap = {
+  // ∀ԐↃᗡƎℲ⅁HIſӼ⅂WNOԀÒᴚS⊥∩ɅMX⅄Z
+  // prettier-ignore
+  "caps": [8704, 1296, 8579, 5601, 398, 8498, 8513, 72, 73, 383, 1276, 8514, 87, 78, 79, 1280, 210, 7450, 83, 8869, 8745, 581, 77, 88, 8516, 90],
+  // ɐqɔpǝɟƃɥıɾʞʅɯuodbɹsʇnʌʍxʎz
+  // prettier-ignore
+  "small": [592, 113, 596, 112, 477, 607, 387, 613, 305, 638, 670, 645, 623, 117, 111, 100, 98, 633, 115, 647, 110, 652, 653, 120, 654, 122],
+}
 
+/**
+ * Based on: https://hg.mozilla.org/mozilla-central/file/a1f74e8c8fb72390d22054d6b00c28b1a32f6c43/intl/l10n/L10nRegistry.jsm#l425
+ */
+function transformString(
+  map: PseudoLocaleTransformMap,
+  elongate = false,
+  msg: string
+) {
+  return msg.replace(/[a-z]/gi, ch => {
+    const cc = ch.charCodeAt(0)
+    if (cc >= 97 && cc <= 122) {
+      const newChar = String.fromCodePoint(map.small[cc - 97])
+      // duplicate "a", "e", "o" and "u" to emulate ~30% longer text
+      if (elongate && (cc === 97 || cc === 101 || cc === 111 || cc === 117)) {
+        return newChar + newChar
+      }
+      return newChar
+    }
+    if (cc >= 65 && cc <= 90) {
+      return String.fromCodePoint(map.caps[cc - 65])
+    }
+    return ch
+  })
+}
+
+/**
+ * accented - Ȧȧƈƈḗḗƞŧḗḗḓ Ḗḗƞɠŀīīşħ
+ * --------------------------------
+ *
+ * This locale replaces all Latin characters with their accented equivalents, and duplicates some
+ * vowels to create roughly 30% longer strings. Strings are wrapped in markers (square brackets),
+ * which help with detecting truncation.
+ */
 export function generateENXA(
   msg: string | MessageFormatElement[]
 ): MessageFormatElement[] {
   const ast = typeof msg === 'string' ? parse(msg) : msg
-  ast.forEach(el => {
-    if (isLiteralElement(el)) {
-      el.value = el.value
-        .split('')
-        .map(c => {
-          const i = ASCII.indexOf(c)
-          if (i < 0) {
-            return c
-          }
-          return ACCENTED_ASCII[i]
-        })
-        .join('')
-    } else if (isPluralElement(el) || isSelectElement(el)) {
-      for (const opt of Object.values(el.options)) {
-        generateENXA(opt.value)
-      }
-    } else if (isTagElement(el)) {
-      generateENXA(el.children)
-    }
+  forEachLiteralElement(ast, el => {
+    el.value = transformString(ACCENTED_MAP, true, el.value)
   })
-  return ast
+  return [
+    {type: TYPE.literal, value: '['},
+    ...ast,
+    {type: TYPE.literal, value: ']'},
+  ]
 }
 
+/**
+ * bidi - ɥsıʅƃuƎ ıpıԐ
+ * -------------------
+ *
+ * This strategy replaces all Latin characters with their 180 degree rotated versions and enforces
+ * right to left text flow using Unicode UAX#9 Explicit Directional Embeddings. In this mode, the UI
+ * directionality will also be set to right-to-left.
+ */
 export function generateENXB(
   msg: string | MessageFormatElement[]
 ): MessageFormatElement[] {
   const ast = typeof msg === 'string' ? parse(msg) : msg
-  ast.forEach(el => {
-    if (isLiteralElement(el)) {
-      const pseudoString = el.value
-        .split('')
-        .map((c, index) => {
-          const i = ASCII.indexOf(c)
-          const canPad = (index + 1) % 3 === 0
-
-          if (i < 0) {
-            return c
-          }
-
-          return canPad ? ACCENTED_ASCII[i].repeat(3) : ACCENTED_ASCII[i]
-        })
-        .join('')
-
-      el.value = `[!! ${pseudoString} !!]`
-    } else if (isPluralElement(el) || isSelectElement(el)) {
-      for (const opt of Object.values(el.options)) {
-        generateENXB(opt.value)
-      }
-    } else if (isTagElement(el)) {
-      generateENXB(el.children)
-    }
+  forEachLiteralElement(ast, el => {
+    el.value = transformString(FLIPPED_MAP, false, el.value)
   })
-  return ast
+  return [
+    {type: TYPE.literal, value: '\u202e'},
+    ...ast,
+    {type: TYPE.literal, value: '\u202c'},
+  ]
 }