feat(db): idempotent startup normalizer rewriting List values to native JSON

This commit is contained in:
Joseph Doherty
2026-06-16 17:50:19 -04:00
parent e3d804a1a6
commit f4b101b532
3 changed files with 360 additions and 0 deletions
@@ -0,0 +1,123 @@
using Microsoft.EntityFrameworkCore;
using Microsoft.Extensions.Logging;
using ZB.MOM.WW.ScadaBridge.Commons.Types;
using ZB.MOM.WW.ScadaBridge.Commons.Types.Enums;
namespace ZB.MOM.WW.ScadaBridge.ConfigurationDatabase;
/// <summary>
/// Idempotent central startup normalizer that rewrites already-persisted List attribute
/// values from the old array-of-strings JSON form (<c>["10","20"]</c>) to the new
/// native-typed form (<c>[10,20]</c>).
///
/// <para>
/// This is a safety net: at the time of writing there is no deployed List attribute data,
/// so in practice it finds nothing to rewrite. It runs on every central startup after
/// migrations apply and MUST never abort startup for data reasons — every row's work is
/// wrapped in a per-row try/catch that logs and skips on malformed data. A second run
/// finds nothing to change (native → native re-encode is byte-identical).
/// </para>
/// </summary>
public static class ListValueNormalizer
{
/// <summary>
/// Rewrites old-form List attribute values to the native-typed JSON form across
/// <see cref="ScadaBridgeDbContext.TemplateAttributes"/> and
/// <see cref="ScadaBridgeDbContext.InstanceAttributeOverrides"/>. Idempotent and
/// best-effort: malformed rows are logged and skipped, never rethrown.
/// </summary>
/// <param name="db">The configuration database context.</param>
/// <param name="logger">Optional logger for diagnostics.</param>
/// <param name="ct">Cancellation token.</param>
/// <returns>A task that represents the asynchronous operation.</returns>
public static async Task NormalizeAsync(
ScadaBridgeDbContext db,
ILogger? logger = null,
CancellationToken ct = default)
{
var rewritten = 0;
// TemplateAttributes: List rows carry the element type on the row itself.
var templateRows = await db.TemplateAttributes
.Where(a => a.DataType == DataType.List)
.ToListAsync(ct);
foreach (var a in templateRows)
{
try
{
var native = AttributeValueCodec.Encode(
AttributeValueCodec.Decode(a.Value, DataType.List, a.ElementDataType));
if (native != a.Value)
{
a.Value = native;
rewritten++;
}
}
catch (FormatException ex)
{
logger?.LogWarning(ex,
"List value normalizer: skipping unparseable list value for TemplateAttribute {Id}.",
a.Id);
}
}
// InstanceAttributeOverrides: only rows that carry an element type are List rows.
// Rows with a null ElementDataType are scalar/legacy rows (no deployed List data
// exists, so none in practice) and are skipped.
var overrideRows = await db.InstanceAttributeOverrides
.Where(o => o.ElementDataType != null)
.ToListAsync(ct);
foreach (var o in overrideRows)
{
if (o.ElementDataType is null)
{
logger?.LogDebug(
"List value normalizer: skipping InstanceAttributeOverride {Id} with no element type.",
o.Id);
continue;
}
try
{
var native = AttributeValueCodec.Encode(
AttributeValueCodec.Decode(o.OverrideValue, DataType.List, o.ElementDataType));
if (native != o.OverrideValue)
{
o.OverrideValue = native;
rewritten++;
}
}
catch (FormatException ex)
{
logger?.LogWarning(ex,
"List value normalizer: skipping unparseable list value for InstanceAttributeOverride {Id}.",
o.Id);
}
}
try
{
await db.SaveChangesAsync(ct);
}
catch (Exception ex)
{
// A catastrophic DB failure on SaveChanges may propagate, but log it first so
// startup diagnostics are not silent. Per-row data problems are already handled
// above and never reach here.
logger?.LogError(ex, "List value normalizer: SaveChanges failed.");
throw;
}
if (rewritten > 0)
{
logger?.LogInformation(
"List value normalizer: rewrote {n} attribute value(s) to native JSON.", rewritten);
}
else
{
logger?.LogDebug("List value normalizer: no attribute values required rewriting.");
}
}
}
@@ -47,6 +47,12 @@ public static class MigrationHelper
"Apply migrations using 'dotnet ef database update' or the generated SQL scripts before starting in production mode.");
}
}
// Safety-net normalizer: rewrite any already-persisted List attribute values from the
// old array-of-strings JSON form to the new native-typed form. Idempotent and never
// aborts startup for data reasons (per-row skip + log). Safe even if both central
// nodes run it concurrently on startup.
await ListValueNormalizer.NormalizeAsync(dbContext, logger, cancellationToken);
}
private static async Task WaitForDatabaseReadyAsync(
@@ -0,0 +1,231 @@
using Microsoft.EntityFrameworkCore;
using ZB.MOM.WW.ScadaBridge.Commons.Entities.Instances;
using ZB.MOM.WW.ScadaBridge.Commons.Entities.Sites;
using ZB.MOM.WW.ScadaBridge.Commons.Entities.Templates;
using ZB.MOM.WW.ScadaBridge.Commons.Types.Enums;
using ZB.MOM.WW.ScadaBridge.ConfigurationDatabase;
namespace ZB.MOM.WW.ScadaBridge.ConfigurationDatabase.Tests;
/// <summary>
/// Tests for <see cref="ListValueNormalizer"/> — the idempotent startup normalizer that
/// rewrites already-persisted List attribute values from the old array-of-strings JSON
/// form (<c>["10","20"]</c>) to the new native-typed form (<c>[10,20]</c>).
/// </summary>
public class ListValueNormalizerTests : IDisposable
{
private readonly ScadaBridgeDbContext _context;
public ListValueNormalizerTests()
{
_context = SqliteTestHelper.CreateInMemoryContext();
}
public void Dispose()
{
_context.Database.CloseConnection();
_context.Dispose();
}
// Seeds a Template parent (satisfies the TemplateAttribute -> Template FK) and returns its Id.
private async Task<int> SeedTemplateAsync()
{
var template = new Template("T1");
_context.Templates.Add(template);
await _context.SaveChangesAsync();
return template.Id;
}
// Seeds Site + Template + Instance (satisfies the InstanceAttributeOverride -> Instance FK)
// and returns the Instance Id.
private async Task<int> SeedInstanceAsync()
{
var site = new Site("Site1", "S-001");
var template = new Template("T1");
_context.Sites.Add(site);
_context.Templates.Add(template);
await _context.SaveChangesAsync();
var instance = new Instance("Inst1") { SiteId = site.Id, TemplateId = template.Id };
_context.Instances.Add(instance);
await _context.SaveChangesAsync();
return instance.Id;
}
[Fact]
public async Task TemplateAttribute_OldStringForm_RewrittenToNative()
{
var templateId = await SeedTemplateAsync();
_context.TemplateAttributes.Add(new TemplateAttribute("intList")
{
TemplateId = templateId,
DataType = DataType.List,
ElementDataType = DataType.Int32,
Value = "[\"10\",\"20\"]",
});
await _context.SaveChangesAsync();
await ListValueNormalizer.NormalizeAsync(_context);
var reloaded = await _context.TemplateAttributes.AsNoTracking().SingleAsync();
Assert.Equal("[10,20]", reloaded.Value);
}
[Fact]
public async Task TemplateAttribute_AlreadyNative_IsNotRewritten()
{
var templateId = await SeedTemplateAsync();
_context.TemplateAttributes.Add(new TemplateAttribute("intList")
{
TemplateId = templateId,
DataType = DataType.List,
ElementDataType = DataType.Int32,
Value = "[10,20]",
});
await _context.SaveChangesAsync();
await ListValueNormalizer.NormalizeAsync(_context);
// No tracked entity should be marked Modified — idempotent no-op.
var tracked = _context.ChangeTracker.Entries<TemplateAttribute>()
.Where(e => e.State == EntityState.Modified)
.ToList();
Assert.Empty(tracked);
var reloaded = await _context.TemplateAttributes.AsNoTracking().SingleAsync();
Assert.Equal("[10,20]", reloaded.Value);
}
[Fact]
public async Task TemplateAttribute_StringList_IsUnchanged()
{
var templateId = await SeedTemplateAsync();
_context.TemplateAttributes.Add(new TemplateAttribute("stringList")
{
TemplateId = templateId,
DataType = DataType.List,
ElementDataType = DataType.String,
Value = "[\"a\",\"b\"]",
});
await _context.SaveChangesAsync();
await ListValueNormalizer.NormalizeAsync(_context);
var reloaded = await _context.TemplateAttributes.AsNoTracking().SingleAsync();
Assert.Equal("[\"a\",\"b\"]", reloaded.Value);
}
[Fact]
public async Task TemplateAttribute_Malformed_IsSkipped_AndSiblingStillNormalized()
{
var templateId = await SeedTemplateAsync();
_context.TemplateAttributes.Add(new TemplateAttribute("badList")
{
TemplateId = templateId,
DataType = DataType.List,
ElementDataType = DataType.Int32,
Value = "[\"a\"", // malformed JSON
});
_context.TemplateAttributes.Add(new TemplateAttribute("goodList")
{
TemplateId = templateId,
DataType = DataType.List,
ElementDataType = DataType.Int32,
Value = "[\"5\"]", // old form, valid
});
await _context.SaveChangesAsync();
// Must NOT throw despite the malformed row.
await ListValueNormalizer.NormalizeAsync(_context);
var bad = await _context.TemplateAttributes.AsNoTracking()
.SingleAsync(a => a.Name == "badList");
var good = await _context.TemplateAttributes.AsNoTracking()
.SingleAsync(a => a.Name == "goodList");
Assert.Equal("[\"a\"", bad.Value); // skipped, untouched
Assert.Equal("[5]", good.Value); // normalized
}
[Fact]
public async Task TemplateAttribute_NonListRow_IsUnchanged()
{
var templateId = await SeedTemplateAsync();
_context.TemplateAttributes.Add(new TemplateAttribute("scalar")
{
TemplateId = templateId,
DataType = DataType.Int32,
ElementDataType = null,
Value = "42",
});
await _context.SaveChangesAsync();
await ListValueNormalizer.NormalizeAsync(_context);
var reloaded = await _context.TemplateAttributes.AsNoTracking().SingleAsync();
Assert.Equal("42", reloaded.Value);
}
[Fact]
public async Task InstanceAttributeOverride_OldStringForm_RewrittenToNative()
{
var instanceId = await SeedInstanceAsync();
_context.InstanceAttributeOverrides.Add(new InstanceAttributeOverride("intList")
{
InstanceId = instanceId,
ElementDataType = DataType.Int32,
OverrideValue = "[\"5\"]",
});
await _context.SaveChangesAsync();
await ListValueNormalizer.NormalizeAsync(_context);
var reloaded = await _context.InstanceAttributeOverrides.AsNoTracking().SingleAsync();
Assert.Equal("[5]", reloaded.OverrideValue);
}
[Fact]
public async Task InstanceAttributeOverride_NullElementType_IsUntouched()
{
var instanceId = await SeedInstanceAsync();
_context.InstanceAttributeOverrides.Add(new InstanceAttributeOverride("scalar")
{
InstanceId = instanceId,
ElementDataType = null,
OverrideValue = "42",
});
await _context.SaveChangesAsync();
await ListValueNormalizer.NormalizeAsync(_context);
var reloaded = await _context.InstanceAttributeOverrides.AsNoTracking().SingleAsync();
Assert.Equal("42", reloaded.OverrideValue);
}
[Fact]
public async Task NormalizeAsync_IsIdempotent_SecondRunChangesNothing()
{
var templateId = await SeedTemplateAsync();
_context.TemplateAttributes.Add(new TemplateAttribute("intList")
{
TemplateId = templateId,
DataType = DataType.List,
ElementDataType = DataType.Int32,
Value = "[\"10\",\"20\"]",
});
await _context.SaveChangesAsync();
await ListValueNormalizer.NormalizeAsync(_context);
var afterFirst = await _context.TemplateAttributes.AsNoTracking().SingleAsync();
Assert.Equal("[10,20]", afterFirst.Value);
await ListValueNormalizer.NormalizeAsync(_context);
var afterSecond = await _context.TemplateAttributes.AsNoTracking().SingleAsync();
Assert.Equal("[10,20]", afterSecond.Value);
var tracked = _context.ChangeTracker.Entries<TemplateAttribute>()
.Where(e => e.State == EntityState.Modified)
.ToList();
Assert.Empty(tracked);
}
}