tests/fixtures/normalization_benchmark_v1.json

{
  "comment": "Golden expectations for propose_merge_clusters; extend after reviewing export JSONL. Random block was generated by scripts/generate_normalization_benchmark_fixture.py.",
  "random_seed": 42,
  "random_case_only_count": 20,
  "merge_cluster_cases": [
    {
      "rows": [
        [
          "snug_a",
          "Snuggles",
          "fetlife_fetish",
          10.0
        ],
        [
          "snug_b",
          "snuggles",
          "fetlife_fetish",
          200.0
        ]
      ],
      "expect_canonical_id": "snug_b",
      "expect_duplicate_ids": [
        "snug_a"
      ]
    },
    {
      "rows": [
        [
          "dup_fmf",
          "threesome fmf",
          "fetlife_fetish",
          100.0
        ],
        [
          "canon_fmf",
          "FMF threesomes",
          "fetlife_fetish",
          500.0
        ],
        [
          "mfm_other",
          "MFM threesomes",
          "fetlife_fetish",
          400.0
        ]
      ],
      "expect_canonical_id": "canon_fmf",
      "expect_duplicate_ids": [
        "dup_fmf"
      ],
      "expect_excluded_ids": [
        "mfm_other"
      ]
    },
    {
      "rows": [
        [
          "rnd_00_lo",
          "Glasshouse",
          "sbench_3257",
          46.0
        ],
        [
          "rnd_00_hi",
          "glasshouse",
          "sbench_3257",
          199.0
        ]
      ],
      "expect_canonical_id": "rnd_00_hi",
      "expect_duplicate_ids": [
        "rnd_00_lo"
      ]
    },
    {
      "rows": [
        [
          "rnd_01_lo",
          "Nightfall",
          "sbench_6873",
          15.0
        ],
        [
          "rnd_01_hi",
          "nightfall",
          "sbench_6873",
          174.0
        ]
      ],
      "expect_canonical_id": "rnd_01_hi",
      "expect_duplicate_ids": [
        "rnd_01_lo"
      ]
    },
    {
      "rows": [
        [
          "rnd_02_lo",
          "Starlight",
          "sbench_9654",
          18.0
        ],
        [
          "rnd_02_hi",
          "starlight",
          "sbench_9654",
          61.0
        ]
      ],
      "expect_canonical_id": "rnd_02_hi",
      "expect_duplicate_ids": [
        "rnd_02_lo"
      ]
    },
    {
      "rows": [
        [
          "rnd_03_lo",
          "Doorframe",
          "sbench_2615",
          45.0
        ],
        [
          "rnd_03_hi",
          "doorframe",
          "sbench_2615",
          168.0
        ]
      ],
      "expect_canonical_id": "rnd_03_hi",
      "expect_duplicate_ids": [
        "rnd_03_lo"
      ]
    },
    {
      "rows": [
        [
          "rnd_04_lo",
          "Candlewax",
          "sbench_5574",
          18.0
        ],
        [
          "rnd_04_hi",
          "candlewax",
          "sbench_5574",
          99.0
        ]
      ],
      "expect_canonical_id": "rnd_04_hi",
      "expect_duplicate_ids": [
        "rnd_04_lo"
      ]
    },
    {
      "rows": [
        [
          "rnd_05_lo",
          "Raspberry",
          "sbench_3527",
          49.0
        ],
        [
          "rnd_05_hi",
          "raspberry",
          "sbench_3527",
          146.0
        ]
      ],
      "expect_canonical_id": "rnd_05_hi",
      "expect_duplicate_ids": [
        "rnd_05_lo"
      ]
    },
    {
      "rows": [
        [
          "rnd_06_lo",
          "Riverbank",
          "sbench_1674",
          6.0
        ],
        [
          "rnd_06_hi",
          "riverbank",
          "sbench_1674",
          157.0
        ]
      ],
      "expect_canonical_id": "rnd_06_hi",
      "expect_duplicate_ids": [
        "rnd_06_lo"
      ]
    },
    {
      "rows": [
        [
          "rnd_07_lo",
          "Dragonfly",
          "sbench_1584",
          23.0
        ],
        [
          "rnd_07_hi",
          "dragonfly",
          "sbench_1584",
          148.0
        ]
      ],
      "expect_canonical_id": "rnd_07_hi",
      "expect_duplicate_ids": [
        "rnd_07_lo"
      ]
    },
    {
      "rows": [
        [
          "rnd_08_lo",
          "Headlights",
          "sbench_9891",
          17.0
        ],
        [
          "rnd_08_hi",
          "headlights",
          "sbench_9891",
          71.0
        ]
      ],
      "expect_canonical_id": "rnd_08_hi",
      "expect_duplicate_ids": [
        "rnd_08_lo"
      ]
    },
    {
      "rows": [
        [
          "rnd_09_lo",
          "Butterfly",
          "sbench_7527",
          35.0
        ],
        [
          "rnd_09_hi",
          "butterfly",
          "sbench_7527",
          91.0
        ]
      ],
      "expect_canonical_id": "rnd_09_hi",
      "expect_duplicate_ids": [
        "rnd_09_lo"
      ]
    },
    {
      "rows": [
        [
          "rnd_10_lo",
          "Lighthouse",
          "sbench_6201",
          6.0
        ],
        [
          "rnd_10_hi",
          "lighthouse",
          "sbench_6201",
          135.0
        ]
      ],
      "expect_canonical_id": "rnd_10_hi",
      "expect_duplicate_ids": [
        "rnd_10_lo"
      ]
    },
    {
      "rows": [
        [
          "rnd_11_lo",
          "Scholarship",
          "sbench_5925",
          37.0
        ],
        [
          "rnd_11_hi",
          "scholarship",
          "sbench_5925",
          109.0
        ]
      ],
      "expect_canonical_id": "rnd_11_hi",
      "expect_duplicate_ids": [
        "rnd_11_lo"
      ]
    },
    {
      "rows": [
        [
          "rnd_12_lo",
          "Fireworks",
          "sbench_1139",
          3.0
        ],
        [
          "rnd_12_hi",
          "fireworks",
          "sbench_1139",
          118.0
        ]
      ],
      "expect_canonical_id": "rnd_12_hi",
      "expect_duplicate_ids": [
        "rnd_12_lo"
      ]
    },
    {
      "rows": [
        [
          "rnd_13_lo",
          "Minefields",
          "sbench_4741",
          6.0
        ],
        [
          "rnd_13_hi",
          "minefields",
          "sbench_4741",
          119.0
        ]
      ],
      "expect_canonical_id": "rnd_13_hi",
      "expect_duplicate_ids": [
        "rnd_13_lo"
      ]
    },
    {
      "rows": [
        [
          "rnd_14_lo",
          "Haystacks",
          "sbench_1654",
          25.0
        ],
        [
          "rnd_14_hi",
          "haystacks",
          "sbench_1654",
          131.0
        ]
      ],
      "expect_canonical_id": "rnd_14_hi",
      "expect_duplicate_ids": [
        "rnd_14_lo"
      ]
    },
    {
      "rows": [
        [
          "rnd_15_lo",
          "Strawberry",
          "sbench_7428",
          41.0
        ],
        [
          "rnd_15_hi",
          "strawberry",
          "sbench_7428",
          153.0
        ]
      ],
      "expect_canonical_id": "rnd_15_hi",
      "expect_duplicate_ids": [
        "rnd_15_lo"
      ]
    },
    {
      "rows": [
        [
          "rnd_16_lo",
          "Roadhouses",
          "sbench_2664",
          24.0
        ],
        [
          "rnd_16_hi",
          "roadhouses",
          "sbench_2664",
          150.0
        ]
      ],
      "expect_canonical_id": "rnd_16_hi",
      "expect_duplicate_ids": [
        "rnd_16_lo"
      ]
    },
    {
      "rows": [
        [
          "rnd_17_lo",
          "Volleyball",
          "sbench_3432",
          43.0
        ],
        [
          "rnd_17_hi",
          "volleyball",
          "sbench_3432",
          128.0
        ]
      ],
      "expect_canonical_id": "rnd_17_hi",
      "expect_duplicate_ids": [
        "rnd_17_lo"
      ]
    },
    {
      "rows": [
        [
          "rnd_18_lo",
          "Flashlight",
          "sbench_1169",
          39.0
        ],
        [
          "rnd_18_hi",
          "flashlight",
          "sbench_1169",
          103.0
        ]
      ],
      "expect_canonical_id": "rnd_18_hi",
      "expect_duplicate_ids": [
        "rnd_18_lo"
      ]
    },
    {
      "rows": [
        [
          "rnd_19_lo",
          "Battlefield",
          "sbench_8751",
          47.0
        ],
        [
          "rnd_19_hi",
          "battlefield",
          "sbench_8751",
          122.0
        ]
      ],
      "expect_canonical_id": "rnd_19_hi",
      "expect_duplicate_ids": [
        "rnd_19_lo"
      ]
    }
  ]
}