adSegmentIdentity.test.ts

/**
 * Unit Tests for Episode Identity Resolution
 *
 * Tests the episode key generation logic:
 * - Primary: RSS <guid>
 * - Fallback: hash(feedUrl + enclosureUrl + pubDate)
 */

import {
  generateEpisodeKey,
  isHashBasedKey,
  batchGenerateEpisodeKeys,
} from '~/services/adSegments/adSegmentIdentity'
import { EpisodeIdentityInput } from '~/services/adSegments/adSegmentIdentity'

describe('Generate Episode Key - GUID-based', () => {
  test('uses GUID when available', () => {
    const input: EpisodeIdentityInput = {
      guid: 'https://example.com/podcast/episode-123',
      feedUrl: 'https://example.com/podcast/feed.xml',
      enclosureUrl: 'https://example.com/podcast/episode-123.mp3',
    }

    const key = generateEpisodeKey(input)

    expect(key).toBe('https://example.com/podcast/episode-123')
    expect(isHashBasedKey(key)).toBe(false)
  })

  test('trims whitespace from GUID', () => {
    const input: EpisodeIdentityInput = {
      guid: '  https://example.com/podcast/episode-123  ',
      feedUrl: 'https://example.com/podcast/feed.xml',
      enclosureUrl: 'https://example.com/podcast/episode-123.mp3',
    }

    const key = generateEpisodeKey(input)

    expect(key).toBe('https://example.com/podcast/episode-123')
  })

  test('strips CDATA wrapper from GUID', () => {
    const input: EpisodeIdentityInput = {
      guid: '<![CDATA[ b47746b8-daa3-11f0-b08a-233a79f7667c ]]>',
      feedUrl: 'https://example.com/podcast/feed.xml',
      enclosureUrl: 'https://example.com/podcast/episode.mp3',
    }

    const key = generateEpisodeKey(input)

    expect(key).toBe('b47746b8-daa3-11f0-b08a-233a79f7667c')
    expect(isHashBasedKey(key)).toBe(false)
  })

  test('handles CDATA with extra whitespace', () => {
    const input: EpisodeIdentityInput = {
      guid: '<![CDATA[   ec8b4655-cb61-4d69-87ee-dbdba77c0a62   ]]>',
      feedUrl: 'https://example.com/podcast/feed.xml',
      enclosureUrl: 'https://example.com/podcast/episode.mp3',
    }

    const key = generateEpisodeKey(input)

    expect(key).toBe('ec8b4655-cb61-4d69-87ee-dbdba77c0a62')
  })

  test('handles real-world test episode GUIDs', () => {
    // Test 1: Plain UUID (What Bitcoin Did)
    const test1: EpisodeIdentityInput = {
      guid: 'ec8b4655-cb61-4d69-87ee-dbdba77c0a62',
      feedUrl: 'https://feeds.fountain.fm/UZSKQcrOnhqYS1JopxGg',
      enclosureUrl: 'https://feeds.fountain.fm/UZSKQcrOnhqYS1JopxGg/items/RpJavE7EEgZNW5hpgUDJ/files/AUDIO---DEFAULT---dcfdc6d6-489e-4ac7-b9a4-5629cc632395.mp3',
    }

    const key1 = generateEpisodeKey(test1)
    expect(key1).toBe('ec8b4655-cb61-4d69-87ee-dbdba77c0a62')

    // Test 2: Plain UUID (TFTC)
    const test2: EpisodeIdentityInput = {
      guid: '2d4217bb-799b-4192-a559-348e82e2ea0e',
      feedUrl: 'https://feeds.fountain.fm/ZwwaDULvAj0yZvJ5kdB9/',
      enclosureUrl: 'https://feeds.fountain.fm/ZwwaDULvAj0yZvJ5kdB9/items/oQBhreJIa5P98Zh0AUmh/files/AUDIO---DEFAULT---34eb3a45-d135-453e-a790-58a744da35ba.mp3',
    }

    const key2 = generateEpisodeKey(test2)
    expect(key2).toBe('2d4217bb-799b-4192-a559-348e82e2ea0e')

    // Test 3: CDATA wrapped UUID (TIP)
    const test3: EpisodeIdentityInput = {
      guid: '<![CDATA[ b47746b8-daa3-11f0-b08a-233a79f7667c ]]>',
      feedUrl: 'https://feeds.megaphone.fm/PPLLC8974708240',
      enclosureUrl: 'https://www.podtrac.com/pts/redirect.mp3/pdst.fm/e/traffic.megaphone.fm/PPLLC1896786790.mp3?updated=1765909572',
    }

    const key3 = generateEpisodeKey(test3)
    expect(key3).toBe('b47746b8-daa3-11f0-b08a-233a79f7667c')
  })

  test('handles GUID with special characters', () => {
    const input: EpisodeIdentityInput = {
      guid: 'urn:uuid:12345678-1234-1234-1234-123456789abc',
      feedUrl: 'https://example.com/podcast/feed.xml',
      enclosureUrl: 'https://example.com/podcast/episode.mp3',
    }

    const key = generateEpisodeKey(input)

    expect(key).toBe('urn:uuid:12345678-1234-1234-1234-123456789abc')
  })
})

describe('Generate Episode Key - Hash-based Fallback', () => {
  test('generates hash when GUID is missing', () => {
    const input: EpisodeIdentityInput = {
      guid: null,
      feedUrl: 'https://example.com/podcast/feed.xml',
      enclosureUrl: 'https://example.com/podcast/episode.mp3',
    }

    const key = generateEpisodeKey(input)

    expect(key).toMatch(/^hash:[0-9a-f]{16}$/)
    expect(isHashBasedKey(key)).toBe(true)
  })

  test('generates hash when GUID is empty string', () => {
    const input: EpisodeIdentityInput = {
      guid: '',
      feedUrl: 'https://example.com/podcast/feed.xml',
      enclosureUrl: 'https://example.com/podcast/episode.mp3',
    }

    const key = generateEpisodeKey(input)

    expect(key).toMatch(/^hash:[0-9a-f]{16}$/)
    expect(isHashBasedKey(key)).toBe(true)
  })

  test('generates hash when GUID is only whitespace', () => {
    const input: EpisodeIdentityInput = {
      guid: '   ',
      feedUrl: 'https://example.com/podcast/feed.xml',
      enclosureUrl: 'https://example.com/podcast/episode.mp3',
    }

    const key = generateEpisodeKey(input)

    expect(key).toMatch(/^hash:[0-9a-f]{16}$/)
    expect(isHashBasedKey(key)).toBe(true)
  })

  test('hash is deterministic (same input = same hash)', () => {
    const input1: EpisodeIdentityInput = {
      guid: null,
      feedUrl: 'https://example.com/podcast/feed.xml',
      enclosureUrl: 'https://example.com/podcast/episode.mp3',
    }

    const input2: EpisodeIdentityInput = {
      guid: null,
      feedUrl: 'https://example.com/podcast/feed.xml',
      enclosureUrl: 'https://example.com/podcast/episode.mp3',
    }

    const key1 = generateEpisodeKey(input1)
    const key2 = generateEpisodeKey(input2)

    expect(key1).toBe(key2)
  })

  test('hash is unique (different input = different hash)', () => {
    const input1: EpisodeIdentityInput = {
      guid: null,
      feedUrl: 'https://example.com/podcast/feed.xml',
      enclosureUrl: 'https://example.com/podcast/episode1.mp3',
    }

    const input2: EpisodeIdentityInput = {
      guid: null,
      feedUrl: 'https://example.com/podcast/feed.xml',
      enclosureUrl: 'https://example.com/podcast/episode2.mp3',
    }

    const key1 = generateEpisodeKey(input1)
    const key2 = generateEpisodeKey(input2)

    expect(key1).not.toBe(key2)
  })

  test('includes pubDate in hash when available', () => {
    const inputWithPubDate: EpisodeIdentityInput = {
      guid: null,
      feedUrl: 'https://example.com/podcast/feed.xml',
      enclosureUrl: 'https://example.com/podcast/episode.mp3',
      pubDate: '2025-01-01T00:00:00Z',
    }

    const inputWithoutPubDate: EpisodeIdentityInput = {
      guid: null,
      feedUrl: 'https://example.com/podcast/feed.xml',
      enclosureUrl: 'https://example.com/podcast/episode.mp3',
    }

    const keyWith = generateEpisodeKey(inputWithPubDate)
    const keyWithout = generateEpisodeKey(inputWithoutPubDate)

    // Different pubDate should result in different hash
    expect(keyWith).not.toBe(keyWithout)
  })

  test('normalizes URLs (case-insensitive)', () => {
    const input1: EpisodeIdentityInput = {
      guid: null,
      feedUrl: 'HTTPS://EXAMPLE.COM/PODCAST/FEED.XML',
      enclosureUrl: 'HTTPS://EXAMPLE.COM/PODCAST/EPISODE.MP3',
    }

    const input2: EpisodeIdentityInput = {
      guid: null,
      feedUrl: 'https://example.com/podcast/feed.xml',
      enclosureUrl: 'https://example.com/podcast/episode.mp3',
    }

    const key1 = generateEpisodeKey(input1)
    const key2 = generateEpisodeKey(input2)

    // Should generate same hash despite case differences
    expect(key1).toBe(key2)
  })

  test('normalizes trailing slashes', () => {
    const input1: EpisodeIdentityInput = {
      guid: null,
      feedUrl: 'https://example.com/podcast/feed.xml/',
      enclosureUrl: 'https://example.com/podcast/episode.mp3/',
    }

    const input2: EpisodeIdentityInput = {
      guid: null,
      feedUrl: 'https://example.com/podcast/feed.xml',
      enclosureUrl: 'https://example.com/podcast/episode.mp3',
    }

    const key1 = generateEpisodeKey(input1)
    const key2 = generateEpisodeKey(input2)

    // Should generate same hash despite trailing slashes
    expect(key1).toBe(key2)
  })
})

describe('Generate Episode Key - Error Cases', () => {
  test('throws error when GUID is missing and feedUrl is missing', () => {
    const input: EpisodeIdentityInput = {
      guid: null,
      feedUrl: null,
      enclosureUrl: 'https://example.com/podcast/episode.mp3',
    }

    expect(() => generateEpisodeKey(input)).toThrow()
  })

  test('throws error when GUID is missing and enclosureUrl is missing', () => {
    const input: EpisodeIdentityInput = {
      guid: null,
      feedUrl: 'https://example.com/podcast/feed.xml',
      enclosureUrl: null,
    }

    expect(() => generateEpisodeKey(input)).toThrow()
  })

  test('throws error when GUID is missing and both feedUrl and enclosureUrl are missing', () => {
    const input: EpisodeIdentityInput = {
      guid: null,
      feedUrl: null,
      enclosureUrl: null,
    }

    expect(() => generateEpisodeKey(input)).toThrow()
  })
})

describe('Batch Generate Episode Keys', () => {
  test('generates keys for multiple episodes', () => {
    const inputs: EpisodeIdentityInput[] = [
      {
        guid: 'https://example.com/episode/1',
        feedUrl: 'https://example.com/feed.xml',
        enclosureUrl: 'https://example.com/episode1.mp3',
      },
      {
        guid: null,
        feedUrl: 'https://example.com/feed.xml',
        enclosureUrl: 'https://example.com/episode2.mp3',
      },
      {
        guid: 'urn:uuid:12345',
        feedUrl: 'https://example.com/feed.xml',
        enclosureUrl: 'https://example.com/episode3.mp3',
      },
    ]

    const keys = batchGenerateEpisodeKeys(inputs)

    expect(keys).toHaveLength(3)
    expect(keys[0]).toBe('https://example.com/episode/1')
    expect(keys[1]).toMatch(/^hash:[0-9a-f]{16}$/)
    expect(keys[2]).toBe('urn:uuid:12345')
  })

  test('handles failed key generation gracefully', () => {
    const inputs: EpisodeIdentityInput[] = [
      {
        guid: 'valid-guid',
        feedUrl: 'https://example.com/feed.xml',
        enclosureUrl: 'https://example.com/episode.mp3',
      },
      {
        guid: null,
        feedUrl: null,
        enclosureUrl: null, // This will fail
      },
    ]

    const keys = batchGenerateEpisodeKeys(inputs)

    expect(keys).toHaveLength(2)
    expect(keys[0]).toBe('valid-guid')
    expect(keys[1]).toBe('') // Failed keys return empty string
  })
})

describe('Is Hash-Based Key', () => {
  test('identifies hash-based keys', () => {
    expect(isHashBasedKey('hash:abc123def456')).toBe(true)
    expect(isHashBasedKey('hash:0000000000000000')).toBe(true)
  })

  test('identifies GUID-based keys', () => {
    expect(isHashBasedKey('https://example.com/episode/1')).toBe(false)
    expect(isHashBasedKey('urn:uuid:12345')).toBe(false)
    expect(isHashBasedKey('episode-guid-123')).toBe(false)
  })
})

describe('Alternative Field Names', () => {
  test('accepts mediaUrl as alternative to enclosureUrl', () => {
    const input: EpisodeIdentityInput = {
      guid: null,
      feedUrl: 'https://example.com/feed.xml',
      mediaUrl: 'https://example.com/episode.mp3', // Using mediaUrl instead of enclosureUrl
    }

    const key = generateEpisodeKey(input)

    expect(key).toMatch(/^hash:[0-9a-f]{16}$/)
  })

  test('prefers enclosureUrl over mediaUrl when both are present', () => {
    const input: EpisodeIdentityInput = {
      guid: null,
      feedUrl: 'https://example.com/feed.xml',
      enclosureUrl: 'https://example.com/episode1.mp3',
      mediaUrl: 'https://example.com/episode2.mp3',
    }

    const key1 = generateEpisodeKey(input)

    const inputOnlyEnclosure: EpisodeIdentityInput = {
      guid: null,
      feedUrl: 'https://example.com/feed.xml',
      enclosureUrl: 'https://example.com/episode1.mp3',
    }

    const key2 = generateEpisodeKey(inputOnlyEnclosure)

    // Should use enclosureUrl, so keys should match
    expect(key1).toBe(key2)
  })
})