ScadaBridge/tests/ZB.MOM.WW.ScadaBridge.SiteEventLogging.Tests/SiteEventsSchemaCalibrationTests.cs

using Microsoft.Data.Sqlite;

namespace ZB.MOM.WW.ScadaBridge.SiteEventLogging.Tests;

/// <summary>
/// Pins the <c>site_events</c> schema to the shape the storage-cap size model was calibrated
/// against (register row 93, <c>docs/plans/2026-08-17-site-event-log-cap-semantics-decision.md</c> §4).
/// </summary>
/// <remarks>
/// <see cref="EventLogPurgeService.EstimateSiteEventsSize"/> hardcodes this schema twice over:
/// the sampling SQL sums <c>octet_length</c> over the exact column list, and
/// <c>IndexedColumnCount = 4</c> plus the fill-factor/overhead constants are empirical for
/// exactly these four secondary indexes. A schema migration that adds a column or an index
/// invalidates both SILENTLY — the new column's bytes are never sampled, a fifth index is
/// never charged — so this test turns that drift into a loud failure. The fix on failure is
/// NEVER to just update the assertions: re-run the memo §4 calibration, update the
/// <see cref="EventLogPurgeService"/> constants and sampling SQL together, and only then
/// re-pin the schema here.
/// </remarks>
public class SiteEventsSchemaCalibrationTests
{
    private const string RecalibrationNotice =
        "site_events schema drifted from the size-model calibration. Re-run the calibration in " +
        "docs/plans/2026-08-17-site-event-log-cap-semantics-decision.md §4 and update " +
        "EventLogPurgeService's constants + sampling SQL together with this pin (register row 93).";

    private static SqliteConnection OpenWithSchema()
    {
        var connection = new SqliteConnection("Data Source=:memory:");
        connection.Open();
        SiteEventLogSchema.Apply(connection);
        return connection;
    }

    private static List<string> ReadStrings(SqliteConnection connection, string sql)
    {
        using var cmd = connection.CreateCommand();
        cmd.CommandText = sql;
        var values = new List<string>();
        using var reader = cmd.ExecuteReader();
        while (reader.Read())
            values.Add(reader.GetString(0));
        return values;
    }

    /// <summary>Verifies that the modelled columns exactly match the sampled octet-length sum.</summary>
    [Fact]
    public void Columns_MatchTheSampledOctetLengthSum_Exactly()
    {
        // The sampler's payload_bytes expression sums octet_length over exactly these eight
        // columns, in this order. A ninth column would contribute real disk bytes that the
        // model never counts, under-sizing every estimate.
        using var connection = OpenWithSchema();

        var columns = ReadStrings(connection, "SELECT name FROM pragma_table_info('site_events')");

        Assert.True(
            columns.SequenceEqual(
                ["id", "timestamp", "event_type", "severity", "instance_id", "source", "message", "details"]),
            $"{RecalibrationNotice} Actual columns: {string.Join(", ", columns)}");
    }

    /// <summary>Verifies that the table has exactly the four indexes the model charges for, plus the primary-key autoindex.</summary>
    [Fact]
    public void Indexes_AreExactlyTheFourTheModelCharges_PlusThePkAutoindex()
    {
        // IndexedColumnCount = 4 charges four secondary index entries per row; the TEXT
        // primary key's autoindex is absorbed into the empirically-measured fill factors.
        // A fifth secondary index would be silently under-counted.
        using var connection = OpenWithSchema();

        var indexes = ReadStrings(
            connection, "SELECT name FROM pragma_index_list('site_events') ORDER BY name");

        Assert.True(
            indexes.SequenceEqual(
                [
                    "idx_events_instance",
                    "idx_events_severity_timestamp",
                    "idx_events_timestamp",
                    "idx_events_type_timestamp",
                    "sqlite_autoindex_site_events_1",
                ]),
            $"{RecalibrationNotice} Actual indexes: {string.Join(", ", indexes)}");
    }

    /// <summary>Verifies that each index covers exactly the columns the sampler sums for it.</summary>
    /// <param name="indexName">The index name.</param>
    /// <param name="expectedColumns">The columns the index is expected to cover, comma-separated, in key order.</param>
    [Theory]
    [InlineData("idx_events_timestamp", "timestamp")]
    [InlineData("idx_events_instance", "instance_id")]
    [InlineData("idx_events_type_timestamp", "event_type,timestamp")]
    [InlineData("idx_events_severity_timestamp", "severity,timestamp")]
    public void EachIndex_CoversExactlyTheColumnsTheSamplerSums(string indexName, string expectedColumns)
    {
        // The sampler's indexed_bytes expression sums 3 * octet_length(timestamp) +
        // octet_length(event_type) + octet_length(severity) + octet_length(instance_id):
        // timestamp once for its own index and once for each of the two composites (#123).
        // Widening or re-pointing an index changes the real entry width without changing
        // what is sampled.
        using var connection = OpenWithSchema();

        var columns = ReadStrings(
            connection, $"SELECT name FROM pragma_index_info('{indexName}') ORDER BY seqno");

        Assert.True(
            columns.SequenceEqual(expectedColumns.Split(',')),
            $"{RecalibrationNotice} Index {indexName} covers: {string.Join(", ", columns)}");
    }
}