Add property tests for post link parsing

Cover parsePostLinks invariants over broad seeded random inputs: round trip reconstruction, link/text segment shape, trailing-punctuation handling, input ordering, email exclusion, and determinism. By refactorer.
2026-09-21 16:52:01 -07:00 · 2026-09-15 19:20:48 -07:00
parent 6c7952f469
commit e17c515065
1 changed files with 173 additions and 0 deletions
@@ -0,0 +1,173 @@
+/*
+  Property tests for the post link parser.
+
+  The unit tests probe parsePostLinks at a few fixed fixtures. These
+  properties pin down the parser's invariants over broad random inputs:
+
+    - Round trip: concatenating the visible text of every segment reconstructs
+      the original input exactly, including trailing punctuation that was
+      stripped from a link.
+    - Segment shape: every segment is text or link; link segments carry a
+      non-empty, whitespace-free href. An explicit http(s) URL keeps its
+      scheme; a bare domain is linked with an https scheme while its visible
+      text stays as written.
+    - Trailing sentence punctuation never becomes part of a link.
+    - Links appear in the same order as in the input.
+    - An email address is never linked.
+    - Parsing is deterministic.
+*/
+
+'use strict'
+
+const test = require('node:test')
+const assert = require('node:assert/strict')
+const { seededRandom, forAll, intGen } = require('./harness')
+const { parsePostLinks } = require('../../src/services/post-links')
+
+const rng = seededRandom(20260916)
+
+const HTTP_URLS = [
+  'https://memo.fullstackcash.net',
+  'http://example.com/path',
+  'https://example.com/other/path?q=1',
+  'https://youtu.be/dQw4w9WgXcQ',
+  'https://example.com/a/b/c',
+  'http://sub.domain.example.co'
+]
+
+const BARE_DOMAINS = [
+  'memo.fullstackcash.net',
+  'memo.fullstackcash.net/feed',
+  'www.example.com',
+  'example.com',
+  'sub.domain.example.co/path'
+]
+
+const EMAILS = ['chris@example.com', 'first.last@example.com', 'a@b.co']
+
+const WORDS = [
+  'hello', 'world', 'check', 'this', 'out', 'memo', 'post', 'a', 'the',
+  'visit', 'go', 'to', 'now', 'read', 'see'
+]
+
+const PUNCTUATION = [' ', '  ', '.', ',', '!', '?', ':', ';', ')', ']']
+
+const TRAILING_PUNCTUATION_RE = /[.,;:!?)\]]$/
+
+const TOKENS = [...WORDS, ...PUNCTUATION, ...HTTP_URLS, ...BARE_DOMAINS, ...EMAILS]
+
+function randomText () {
+  const n = intGen(rng, 0, 14)()
+  let text = ''
+  for (let i = 0; i < n; i++) {
+    text += TOKENS[Math.floor(rng() * TOKENS.length)]
+  }
+  return text
+}
+
+function reconstruct (segments) {
+  return segments.map((segment) => segment.text).join('')
+}
+
+function isHttpUrl (value) {
+  return /^https?:\/\//.test(value)
+}
+
+test('parsePostLinks round-trips: segments reconstruct the original text', async () => {
+  await forAll(
+    () => randomText(),
+    async (text) => reconstruct(parsePostLinks(text)) === text,
+    { label: 'post-links round trip', samples: 3000 }
+  )
+})
+
+test('parsePostLinks yields only valid text and link segments', async () => {
+  await forAll(
+    () => randomText(),
+    async (text) => {
+      for (const segment of parsePostLinks(text)) {
+        if (segment.type === 'text') {
+          if (typeof segment.text !== 'string') return false
+          continue
+        }
+        if (segment.type !== 'link') return false
+        if (!segment.href || typeof segment.href !== 'string') return false
+        if (!segment.text || typeof segment.text !== 'string') return false
+        if (/\s/.test(segment.text)) return false
+        if (!isHttpUrl(segment.href)) return false
+        // Explicit URLs preserve their scheme; bare domains get https.
+        const expectedHref = isHttpUrl(segment.text) ? segment.text : `https://${segment.text}`
+        if (segment.href !== expectedHref) return false
+      }
+      return true
+    },
+    { label: 'post-links segment shape', samples: 3000 }
+  )
+})
+
+test('parsePostLinks never leaves trailing punctuation in a link', async () => {
+  await forAll(
+    () => randomText(),
+    async (text) => {
+      for (const segment of parsePostLinks(text)) {
+        if (segment.type === 'link' && TRAILING_PUNCTUATION_RE.test(segment.text)) {
+          return false
+        }
+      }
+      return true
+    },
+    { label: 'post-links trailing punctuation', samples: 3000 }
+  )
+})
+
+test('parsePostLinks yields links in input order', async () => {
+  await forAll(
+    () => randomText(),
+    async (text) => {
+      let cursor = 0
+      for (const segment of parsePostLinks(text)) {
+        if (segment.type !== 'link') continue
+        const at = text.indexOf(segment.text, cursor)
+        if (at < 0) return false
+        cursor = at + segment.text.length
+      }
+      return true
+    },
+    { label: 'post-links ordering', samples: 3000 }
+  )
+})
+
+test('parsePostLinks never links an email address', async () => {
+  await forAll(
+    () => {
+      const local = ['chris', 'first.last', 'a_b', 'x'][Math.floor(rng() * 4)]
+      const domain = ['example.com', 'b.co', 'mail.example.org'][Math.floor(rng() * 3)]
+      return `${local}@${domain}`
+    },
+    async (email) => {
+      const segments = parsePostLinks(email)
+      return segments.length === 1 &&
+        segments[0].type === 'text' &&
+        segments[0].text === email
+    },
+    { label: 'post-links email stays text', samples: 500 }
+  )
+})
+
+test('parsePostLinks is deterministic', async () => {
+  await forAll(
+    () => randomText(),
+    async (text) => {
+      const first = parsePostLinks(text)
+      const second = parsePostLinks(text)
+      return JSON.stringify(first) === JSON.stringify(second)
+    },
+    { label: 'post-links determinism', samples: 2000 }
+  )
+})
+
+test('parsePostLinks stringifies nullish and non-string input', () => {
+  assert.deepEqual(parsePostLinks(null), [{ type: 'text', text: '' }])
+  assert.deepEqual(parsePostLinks(undefined), [{ type: 'text', text: '' }])
+  assert.deepEqual(parsePostLinks(12345), [{ type: 'text', text: '12345' }])
+})