tiny.profiling.ingest.selector

1pub const selector = @import("selector.zig");

1const std = @import("std");2const ingest = @import("root.zig");34pub const upstream_repository = "https://github.com/simdjson/simdjson";5pub const upstream_revision = "8e6bac94877f2d3d026000d36ce81e0aaf38d26f";67const slot_count = 64;8const empty = std.math.maxInt(u8);910const Entry = struct {11    name: []const u8,12    field: ingest.event.Field,13};1415const entries = [_]Entry{16    .{ .name = "schema", .field = .schema },17    .{ .name = "event", .field = .event },18    .{ .name = "kind", .field = .kind },19    .{ .name = "summary", .field = .summary },20    .{ .name = "sample_ns", .field = .sample_ns },21    .{ .name = "mean_ns", .field = .mean_ns },22    .{ .name = "median_ns", .field = .median_ns },23    .{ .name = "p95_ns", .field = .p95_ns },24    .{ .name = "p99_ns", .field = .p99_ns },25    .{ .name = "alloc_bytes", .field = .alloc_bytes },26    .{ .name = "alloc_count", .field = .alloc_count },27    .{ .name = "value", .field = .value },28    .{ .name = "mean", .field = .mean },29    .{ .name = "median", .field = .median },30    .{ .name = "allocated_bytes", .field = .allocated_bytes },31    .{ .name = "retained_bytes", .field = .retained_bytes },32    .{ .name = "high_water_live_bytes", .field = .high_water_live_bytes },33    .{ .name = "len", .field = .len },34    .{ .name = "old_len", .field = .old_len },35    .{ .name = "new_len", .field = .new_len },36};3738const slots = buildSlots();3940comptime {41    if (!std.math.isPowerOfTwo(slot_count)) @compileError("selector slot count must be a power of two");42    if (entries.len >= empty) @compileError("selector entry index exceeds its representation");43}4445pub fn field(hash: u64, key: []const u8) ingest.event.Field {46    const slot = @as(usize, @intCast(hash & (slot_count - 1)));47    const index = slots[slot];48    if (index == empty) return .other;49    const entry = entries[index];50    return if (std.mem.eql(u8, key, entry.name)) entry.field else .other;51}5253fn buildSlots() [slot_count]u8 {54    var result: [slot_count]u8 = @splat(empty);55    for (entries, 0..) |entry, index| {56        const hash = std.hash.Wyhash.hash(ingest.duplicate.hash_seed, entry.name);57        const slot = @as(usize, @intCast(hash & (slot_count - 1)));58        if (result[slot] != empty) @compileError("selector hash seed is not collision free");59        result[slot] = @intCast(index);60    }61    return result;62}6364test "on demand field selector has one exact candidate per known key" {65    try std.testing.expectEqual(@as(usize, 20), entries.len);66    for (entries, 0..) |entry, index| {67        try std.testing.expectEqual(index, @backingInt(entry.field));68        const hash = std.hash.Wyhash.hash(ingest.duplicate.hash_seed, entry.name);69        try std.testing.expectEqual(entry.field, field(hash, entry.name));70    }71    var occupied_miss = false;72    for ([_][]const u8{73        "schemas",74        "events",75        "allocated_count",76        "high_water_live_byte",77        "new_lens",78        "unknown",79        "kindred",80        "sample_ns_extra",81    }) |unknown| {82        const hash = std.hash.Wyhash.hash(ingest.duplicate.hash_seed, unknown);83        const slot = @as(usize, @intCast(hash & (slot_count - 1)));84        occupied_miss = occupied_miss or slots[slot] != empty;85        try std.testing.expectEqual(ingest.event.Field.other, field(hash, unknown));86    }87    try std.testing.expect(occupied_miss);88}