tests/testthat/test-benchmark.R

In bigANNOY: Approximate k-Nearest Neighbour Search for 'bigmemory' Matrices with Annoy

skip_if_not_installed("bigmemory")
skip_if_not_installed("RcppAnnoy")

test_that("benchmark comparison reports bigANNOY and RcppAnnoy rows with byte metrics", {
    td <- tempfile("bigannoy-benchmark-compare-")
    dir.create(td, recursive = TRUE)
    on.exit(unlink(td, recursive = TRUE, force = TRUE), add = TRUE)

    result <- benchmark_annoy_vs_rcppannoy(
        n_ref = 40L,
        n_query = 8L,
        n_dim = 4L,
        k = 3L,
        n_trees = 10L,
        search_k = 50L,
        exact = FALSE,
        path_dir = td,
        keep_files = TRUE,
        load_mode = "eager"
    )

    summary <- result$summary

    expect_setequal(summary$implementation, c("bigANNOY", "RcppAnnoy"))
    expect_true(all(summary$ref_bytes > 0))
    expect_true(all(summary$total_data_bytes >= summary$ref_bytes))
    expect_true(all(summary$index_bytes > 0))
    expect_true(all(summary$build_elapsed >= 0))
    expect_true(all(summary$search_elapsed >= 0))

    big_row <- summary[summary$implementation == "bigANNOY", , drop = FALSE]
    rcpp_row <- summary[summary$implementation == "RcppAnnoy", , drop = FALSE]

    expect_equal(big_row$metadata_bytes, benchmark_file_bytes(result$artifacts$bigannoy$metadata_path))
    expect_true(big_row$metadata_bytes > 0)
    expect_identical(rcpp_row$metadata_bytes, 0)
    expect_true(result$validation$valid)
})

test_that("benchmark volume suite expands across requested size grid", {
    td <- tempfile("bigannoy-benchmark-volume-")
    dir.create(td, recursive = TRUE)

    result <- benchmark_annoy_volume_suite(
        n_ref = c(25L, 50L),
        n_query = c(5L, 10L),
        n_dim = c(3L, 6L),
        k = 2L,
        n_trees = 8L,
        search_k = 20L,
        exact = FALSE,
        path_dir = td
    )

    summary <- result$summary

    expect_equal(nrow(summary), 2L * 2L * 2L * 2L)
    expect_setequal(summary$implementation, c("bigANNOY", "RcppAnnoy"))
    expect_true(all(summary$total_data_bytes > 0))
    expect_true(all(summary$artifact_bytes >= summary$index_bytes))
})