more tests

Author: ChALkeR

tests/whatwg.test.js

@@ -1,20 +1,88 @@
 import '@exodus/bytes/encoding.js'
 import { percentEncodeAfterEncoding } from '@exodus/bytes/whatwg.js'
-import { test } from 'node:test'
+import { describe, test } from 'node:test'
+import { labels } from './encoding/fixtures/encodings.cjs'
 
-// https://url.spec.whatwg.org/#example-percent-encode-operations
-test('percent-encode after encoding', (t) => {
-  const userinfo = ' "#/:;<=>?@[\\]^`{|}' // https://url.spec.whatwg.org/#userinfo-percent-encode-set
+const userinfo = ' "#/:;<=>?@[\\]^`{|}' // https://url.spec.whatwg.org/#userinfo-percent-encode-set
+const jsuri = ' "%<>[\\]^`{|}' // https://tc39.es/ecma262/#sec-encodeuri-uri
+const jsuricomponent = ' "#$%&+,/:;<=>?@[\\]^`{|}' // https://tc39.es/ecma262/#sec-encodeuricomponent-uricomponent
+
+const sets = ['', userinfo, jsuri, jsuricomponent]
+const invalid = ['replacement', 'utf-16le', 'utf-16be'] // https://encoding.spec.whatwg.org/#get-an-encoder
+
+const slowEngine =
+  process.env.EXODUS_TEST_PLATFORM === 'quickjs' ||
+  process.env.EXODUS_TEST_PLATFORM === 'xs' ||
+  process.env.EXODUS_TEST_PLATFORM === 'engine262'
+
+describe('percent-encode after encoding', () => {
   const f = percentEncodeAfterEncoding
-  t.assert.strictEqual(f('Shift_JIS', ' ', userinfo), '%20')
-  t.assert.strictEqual(f('Shift_JIS', '≡', userinfo), '%81%DF')
-  t.assert.strictEqual(f('Shift_JIS', '‽', userinfo), '%26%238253%3B')
-  // t.assert.strictEqual(f('ISO-2022-JP', '¥', userinfo), "%1B(J\\%1B(B") // https://github.com/whatwg/url/issues/895
-  t.assert.strictEqual(
-    f('Shift_JIS', '1+1 ≡ 2%20‽', userinfo, true),
-    '1+1+%81%DF+2%20%26%238253%3B'
-  )
-  t.assert.strictEqual(f('UTF-8', '≡', userinfo), '%E2%89%A1')
-  t.assert.strictEqual(f('UTF-8', '‽', userinfo), '%E2%80%BD')
-  t.assert.strictEqual(f('UTF-8', 'Say what‽', userinfo), 'Say%20what%E2%80%BD')
+
+  // https://url.spec.whatwg.org/#example-percent-encode-operations
+  test('examples from spec', (t) => {
+    t.assert.strictEqual(f('Shift_JIS', ' ', userinfo), '%20')
+    t.assert.strictEqual(f('Shift_JIS', '≡', userinfo), '%81%DF')
+    t.assert.strictEqual(f('Shift_JIS', '‽', userinfo), '%26%238253%3B')
+    // t.assert.strictEqual(f('ISO-2022-JP', '¥', userinfo), "%1B(J\\%1B(B") // https://github.com/whatwg/url/issues/895
+    t.assert.strictEqual(
+      f('Shift_JIS', '1+1 ≡ 2%20‽', userinfo, true),
+      '1+1+%81%DF+2%20%26%238253%3B'
+    )
+    t.assert.strictEqual(f('UTF-8', '≡', userinfo), '%E2%89%A1')
+    t.assert.strictEqual(f('UTF-8', '‽', userinfo), '%E2%80%BD')
+    t.assert.strictEqual(f('UTF-8', 'Say what‽', userinfo), 'Say%20what%E2%80%BD')
+  })
+
+  // https://encoding.spec.whatwg.org/#get-an-encoder
+  describe('throws on unkown, utf-16 and replacement', () => {
+    for (const encoding of [...invalid, 'what', 'UTF-16', 'unicode']) {
+      test(encoding, (t) => {
+        for (const set of sets) {
+          t.assert.throws(() => f(encoding, '', set), /encoding/)
+          t.assert.throws(() => f(encoding, ' ', set), /encoding/)
+          t.assert.throws(() => f(encoding, ' ', set, true), /encoding/)
+          t.assert.throws(() => f(encoding, '\uFFFD', set, true), /encoding/)
+        }
+      })
+    }
+  })
+
+  describe('all valid encodings are recognized', () => {
+    for (const encoding of labels) {
+      if (encoding.includes(encoding)) continue
+      test(encoding, (t) => {
+        for (const set of sets) {
+          t.assert.strictEqual(f(encoding, '', set), '')
+          // Even non-ASCII encodings passthrough on a lone space
+          t.assert.strictEqual(f(encoding, ' ', set), set.includes(' ') ? '%20' : ' ')
+          t.assert.strictEqual(f(encoding, ' ', set, true), '+')
+        }
+      })
+    }
+  })
+
+  describe('throws on non-scalarvalue', () => {
+    for (const encoding of labels) {
+      test(encoding, (t) => {
+        for (let cp = 0xd8_00; cp < 0xe0_00; cp++) {
+          const s = String.fromCodePoint(cp)
+          t.assert.throws(() => f(encoding, s, userinfo))
+          t.assert.throws(() => f(encoding, s, jsuri))
+        }
+      })
+    }
+  })
+
+  test('encodeURI / encodeURIComponent', (t) => {
+    // https://tc39.es/ecma262/#sec-encodeuri-uri step 2, coherence check
+    t.assert.deepStrictEqual([...(jsuri + ';/?:@&=+$,#')].sort(), [...jsuricomponent].sort())
+
+    const MAX = slowEngine ? 0x1_ff_ff : 0x10_ff_ff // Max Unicode codepoint
+    for (let cp = 0; cp <= MAX; cp++) {
+      if (cp >= 0xd8_00 && cp < 0xe0_00) continue
+      const s = String.fromCodePoint(cp)
+      t.assert.strictEqual(f('utf8', s, jsuricomponent), encodeURIComponent(s))
+      t.assert.strictEqual(f('utf8', s, jsuri), encodeURI(s))
+    }
+  })
 })