mirror of
https://github.com/kuhyx/diet-guard.git
synced 2026-11-11 13:07:40 +01:00
Without this the schedule would be device-local, which is the failure this
whole feature has to avoid: the phone sets five meals, the PC still derives
four, and the PC's gate nags for a 12:00 checkpoint the phone never offered
and can never satisfy. So sync lands before either edit surface, not after.
The history rides the existing `budget` CRDT record as one
`sched:<YYYY-MM-DD>` field per entry, exactly like the budget's own `hist:`
fields. merge_record is per-field LWW over the *union* of field names and both
devices push the merged record, so a device that predates the feature relays
the fields untouched -- no coordinated release. There is a test for precisely
that: merging our push with a budget-only peer keeps both schedule entries.
Verified the two languages agree on the wire, not just in their own suites --
Python and Dart both emit
{"sched:1970-01-01":{"f":8,"l":20,"n":4},"sched:2026-08-16":{"f":8,"l":20,"n":5}}
The Hlc is derived from the *parsed* timestamp on both sides, which matters
because the epoch fallback formats differently in each language
(1970-01-01T00:00:00+00:00 vs ...000Z) and a string comparison would diverge.
Two guards worth naming:
- A device with no history contributes no `sched:` fields at all, so an unset
default can never outrank a peer's real schedule; and an empty merge result
is never written back, so a pre-feature peer cannot erase a real history.
- Dart's `_syncBudget` synthesises a local record when the schedule has been
edited but the budget never has -- `budgetToLog(null, ...)` returns an empty
Log and would otherwise drop the schedule fields too. Its budget value
carries an epoch stamp, so it loses every LWW race against a real edit.
sync_merge/_schedule.py and sync_merge_schedule.dart are separate modules
rather than additions to the budget adapters, which were at 186 and 170 lines
against the 250 cap. Each declares its own record-id constant so the Dart
pair does not import each other.
772 Python tests at 100% branch coverage, 619 Dart.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0159b4E5hh512jYUeaJVNxfM
177 lines
7.0 KiB
Dart
177 lines
7.0 KiB
Dart
/// Budget and budget-history <-> crdt_sync.Record adapters.
|
|
///
|
|
/// Split out of `sync_merge.dart` for file size; that file re-exports these
|
|
/// so existing importers keep working.
|
|
///
|
|
/// A budget record is edited repeatedly rather than immutable-after-creation
|
|
/// like a food-log entry, so its [Hlc] derives from a `t` edit timestamp
|
|
/// rather than a birth time that never changes. History rides along as
|
|
/// `hist:<YYYY-MM-DD>` fields on the same record, so a device predating the
|
|
/// feature relays them untouched. Mirrors `diet_guard/_sync_merge.py`.
|
|
library;
|
|
|
|
import 'dart:convert';
|
|
|
|
import 'package:crdt_sync/crdt_sync.dart';
|
|
import 'package:diet_guard_app/services/budget_schedule.dart';
|
|
import 'package:diet_guard_app/services/meal_schedule_history.dart';
|
|
import 'package:diet_guard_app/services/sync_device_id.dart';
|
|
import 'package:diet_guard_app/services/sync_merge_schedule.dart';
|
|
|
|
/// Stable id: exactly one budget record per device-pushed `budget.json`.
|
|
const budgetRecordId = 'budget';
|
|
|
|
/// Derives a deterministic [Hlc] for a raw budget record from its `t` field.
|
|
///
|
|
/// Mirrors [entryHlc]'s determinism -- the same unedited record always
|
|
/// yields the same Hlc, so re-syncing an unchanged budget is a no-op -- but
|
|
/// reads `t` (bumped on every explicit edit) rather than a fixed birth
|
|
/// time, since a budget can be edited repeatedly and the *edit* time is
|
|
/// what last-writer-wins must compare.
|
|
Hlc budgetHlc(Map<String, dynamic> record) {
|
|
final wallTimeMs =
|
|
DateTime.tryParse(
|
|
record['t'] as String? ?? '',
|
|
)?.toUtc().millisecondsSinceEpoch ??
|
|
0;
|
|
return Hlc.newTick(currentSyncDeviceId, wallTimeMsOverride: wallTimeMs);
|
|
}
|
|
|
|
/// Field-name prefix for the effective-from budget history, one field per
|
|
/// date. A separate *field* rather than a separate document, because
|
|
/// `mergeRecord` unions field names -- see [budgetToLog].
|
|
const budgetHistoryFieldPrefix = 'hist:';
|
|
|
|
/// Derives a deterministic [Hlc] for one history entry from its edit time.
|
|
///
|
|
/// Same trick as [budgetHlc]: identical inputs always yield the same clock,
|
|
/// so re-syncing unchanged history is a no-op. Two devices that seed the
|
|
/// history independently derive the same wall time and the same value and
|
|
/// differ only in node id, so whichever side wins the field-level LWW, the
|
|
/// value is identical and the merge converges in one round.
|
|
Hlc historyHlc(BudgetEntry entry) {
|
|
final wallTimeMs =
|
|
DateTime.tryParse(entry.editedAt)?.toUtc().millisecondsSinceEpoch ?? 0;
|
|
return Hlc.newTick(currentSyncDeviceId, wallTimeMsOverride: wallTimeMs);
|
|
}
|
|
|
|
/// Converts a raw local budget record plus its history into a [Log].
|
|
///
|
|
/// Returns an empty [Log] when [record] is null (this device has never
|
|
/// explicitly set a budget), so it contributes nothing to the merge rather
|
|
/// than clobbering another device's real value with the unset default.
|
|
///
|
|
/// The history rides along as one `hist:<YYYY-MM-DD>` field per entry on the
|
|
/// *same* record. crdt_sync's `mergeRecord` is per-field LWW over the union
|
|
/// of field names, so a device that predates the history neither clobbers
|
|
/// those fields nor blocks them: it merges them in from the remote and
|
|
/// pushes them straight back out (both sides push the *merged* log). That is
|
|
/// what makes this safe to roll out without a coordinated release.
|
|
///
|
|
/// `w` (body weight) is stripped from `value` for the same reason the
|
|
/// history is not stored there: inside the shared map it was collateral
|
|
/// damage of whole-map LWW, since this device rebuilds that map without it.
|
|
/// Python carries it as its own `weight` field instead, and this device
|
|
/// relays that field untouched through the merged log even though it has no
|
|
/// weight of its own to contribute -- so both devices still see one value.
|
|
Log budgetToLog(
|
|
Map<String, dynamic>? record, [
|
|
List<BudgetEntry> entries = const [],
|
|
List<ScheduleEntry> scheduleEntries = const [],
|
|
]) {
|
|
if (record == null) return {};
|
|
final hlc = budgetHlc(record);
|
|
final value = Map<String, dynamic>.from(record)
|
|
..remove('t')
|
|
..remove('w');
|
|
final fields = <String, (dynamic, Hlc)>{'value': (value, hlc)};
|
|
for (final entry in entries) {
|
|
fields['$budgetHistoryFieldPrefix${entry.effectiveFrom}'] = (
|
|
entry.kcal,
|
|
historyHlc(entry),
|
|
);
|
|
}
|
|
// The meal-schedule history rides the same record as its own `sched:`
|
|
// fields; see `sync_merge_schedule.dart`.
|
|
fields.addAll(scheduleFields(scheduleEntries));
|
|
return {budgetRecordId: Record(id: budgetRecordId, fields: fields)};
|
|
}
|
|
|
|
/// Extracts the effective-from history from a merged budget [Log].
|
|
///
|
|
/// Each entry's `editedAt` is reconstructed from its field Hlc rather than
|
|
/// carried separately, so the stored timestamp and the clock the merge
|
|
/// compared can never drift apart.
|
|
List<BudgetEntry> logToHistory(Log log) {
|
|
final record = log[budgetRecordId];
|
|
if (record == null) return const [];
|
|
final entries = <BudgetEntry>[];
|
|
for (final field in record.fields.entries) {
|
|
if (!field.key.startsWith(budgetHistoryFieldPrefix)) continue;
|
|
final kcal = field.value.$1;
|
|
if (kcal is! int) continue;
|
|
entries.add(
|
|
BudgetEntry(
|
|
effectiveFrom: field.key.substring(budgetHistoryFieldPrefix.length),
|
|
kcal: kcal,
|
|
editedAt: DateTime.fromMillisecondsSinceEpoch(
|
|
field.value.$2.wallTimeMs,
|
|
isUtc: true,
|
|
).toLocal().toIso8601String(),
|
|
),
|
|
);
|
|
}
|
|
entries.sort((a, b) => a.effectiveFrom.compareTo(b.effectiveFrom));
|
|
return entries;
|
|
}
|
|
|
|
/// Converts a merged budget [Log] back into a raw budget record.
|
|
///
|
|
/// Returns null when the log has no budget record at all (neither device
|
|
/// has ever set one yet) -- callers treat that as "nothing to apply
|
|
/// locally", not an error.
|
|
Map<String, dynamic>? logToBudget(Log log) {
|
|
final record = log[budgetRecordId];
|
|
if (record == null) return null;
|
|
final field = record.fields['value'];
|
|
final value = field?.$1;
|
|
final result = value is Map
|
|
? Map<String, dynamic>.from(value.cast<String, dynamic>())
|
|
: <String, dynamic>{};
|
|
final hlc = field?.$2;
|
|
if (hlc != null) {
|
|
result['t'] = DateTime.fromMillisecondsSinceEpoch(
|
|
hlc.wallTimeMs,
|
|
isUtc: true,
|
|
).toLocal().toIso8601String();
|
|
}
|
|
return result;
|
|
}
|
|
|
|
/// Parses one device's pushed `budget.json` text into a crdt_sync [Log].
|
|
///
|
|
/// Throwing [FormatException] or [TypeError] is treated as unparsable by
|
|
/// the caller (`syncLog`'s `decode` callback), matching [parseRemoteLog]'s
|
|
/// tolerance for a corrupt/truncated push. There is no legacy plain-format
|
|
/// fallback here -- `budget.json` is a brand-new sync payload.
|
|
Log parseRemoteBudget(String text) {
|
|
final raw = jsonDecode(text);
|
|
if (raw is! Map) {
|
|
throw const FormatException(
|
|
'top-level budget payload is not a JSON object',
|
|
);
|
|
}
|
|
final rawMap = raw.cast<String, dynamic>();
|
|
return rawMap.map(
|
|
(id, data) => MapEntry(id, Record.fromJson(data as Map<String, dynamic>)),
|
|
);
|
|
}
|
|
|
|
/// Serializes a merged budget [Log] for push.
|
|
String encodeBudgetForPush(Log log) {
|
|
final encoded = <String, dynamic>{
|
|
for (final entry in log.entries) entry.key: entry.value.toJson(),
|
|
};
|
|
return jsonEncode(encoded);
|
|
}
|