mirror of
https://github.com/kuhyx/diet-guard.git
synced 2026-11-11 16:33:56 +01:00
A Duration is an absolute span, so stepping a local DateTime by
Duration(days: 1) moves 23 or 25 hours across a DST boundary and lands on
the wrong wall-clock day. Under Europe/Warsaw, spring-forward 2026-03-29:
lastCompleteDay(2026-03-30) -> 2026-03-28 23:00, the day before
yesterday, not yesterday
weeksAgo: 1 from 2026-04-01 -> 2026-03-24 23:00, a Tuesday, so the
wrong ISO week entirely
a 03-27..03-31 period -> emits four day keys, silently dropping
2026-03-31
The Python half computes all three correctly, because timedelta on a date
has no clock to drift against. So for about two weeks a year the phone and
the PC would report different averages for the same log -- exactly the kind
of device-local divergence the shared-source-of-truth rule exists to stop.
Passing out-of-range components to DateTime's constructor normalizes on the
calendar (day 0 is the previous month's last day), which is what timedelta
already does. The same fix goes to day_status_service's streak walker: it
had the identical pattern and predates this feature, so a streak that
spanned the clock change could break when it should have held.
None of the existing tests could see any of this -- every fixture date sits
in a DST-free span -- so the regression tests use dates that straddle
2026-03-29 and assert on the calendar day rather than the DateTime, which
keeps them meaningful whatever zone the runner is in. Verified they fail
against the old arithmetic before fixing it.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DjJqpcEsaV5WRhPXyxCThd
186 lines
6.8 KiB
Dart
186 lines
6.8 KiB
Dart
/// Weekly and monthly average intake, and how that average met the budget.
|
|
///
|
|
/// PERIOD AVERAGE SPEC (keep in sync with `diet_guard/_averages.py`):
|
|
///
|
|
/// * the average is over **logged days only** -- `sum(kcal) / loggedDays`. A
|
|
/// day you forgot to log is not a zero-calorie day, and counting it as one
|
|
/// would make every gap in the log read as "well under budget", which is the
|
|
/// exact opposite of the truth. `loggedDays`/`elapsedDays` travel with the
|
|
/// average so a caller can show how much of the period it covers.
|
|
/// * the yardstick is the **mean of the per-day budgets over those same logged
|
|
/// days**, resolved through a [BudgetSchedule] -- never today's budget.
|
|
/// Comparing a past week against a budget lowered yesterday would
|
|
/// retroactively reclassify it, which is what `budget_schedule.dart` exists
|
|
/// to prevent.
|
|
/// * bands reuse the calendar's own over-budget boundary
|
|
/// ([kOverBudgetYellowCeiling]), so a period and its days can never disagree
|
|
/// about what "over" means:
|
|
///
|
|
/// - `under`: `avgKcal <= avgBudget`
|
|
/// - `slightlyOver`: `avgBudget < avgKcal <= avgBudget * ceiling`
|
|
/// - `veryOver`: `avgKcal > avgBudget * ceiling`
|
|
///
|
|
/// * **today is excluded.** A period ending at "now" mixes complete days with
|
|
/// one that is three hours old, and a half-logged today drags the mean far
|
|
/// enough to flip the band. Every period ends at [lastCompleteDay] --
|
|
/// yesterday -- so "this week" means "this week so far, in finished days". A
|
|
/// period with no finished days yet (Monday, or the 1st) reports
|
|
/// `elapsedDays == 0` and a null average rather than a flattering fake one.
|
|
/// * weeks are **ISO weeks, Monday through Sunday**; months are calendar
|
|
/// months.
|
|
///
|
|
/// Every function here is a pure function of its explicit `log`/`schedule`/
|
|
/// `today` arguments and never reaches into on-disk state.
|
|
library;
|
|
|
|
import 'package:diet_guard_app/models/period_average.dart';
|
|
import 'package:diet_guard_app/services/budget_schedule.dart';
|
|
import 'package:diet_guard_app/services/day_status_service.dart';
|
|
import 'package:diet_guard_app/services/log_storage_service.dart';
|
|
|
|
/// Classifies [avgKcal] against [avgBudget].
|
|
AverageBand bandFor(double avgKcal, double avgBudget) {
|
|
if (avgKcal <= avgBudget) return AverageBand.under;
|
|
if (avgKcal <= avgBudget * kOverBudgetYellowCeiling) {
|
|
return AverageBand.slightlyOver;
|
|
}
|
|
return AverageBand.veryOver;
|
|
}
|
|
|
|
/// Returns the last day whose log is finished: the day before [today].
|
|
DateTime lastCompleteDay(DateTime today) => _addDays(today, -1);
|
|
|
|
/// Returns the Monday and Sunday of the ISO week containing [day].
|
|
(DateTime, DateTime) weekBounds(DateTime day) {
|
|
final monday = _addDays(day, DateTime.monday - day.weekday);
|
|
return (monday, _addDays(monday, 6));
|
|
}
|
|
|
|
/// Returns the first and last dates of [day]'s calendar month.
|
|
(DateTime, DateTime) monthBounds(DateTime day) => (
|
|
DateTime(day.year, day.month),
|
|
// Day 0 of the *next* month is the last day of this one, resolved by
|
|
// DateTime's constructor rather than by subtracting 24 hours.
|
|
DateTime(day.year, day.month + 1, 0),
|
|
);
|
|
|
|
/// Returns the first of the month [months] before [day]'s month.
|
|
DateTime shiftMonths(DateTime day, int months) =>
|
|
DateTime(day.year, day.month - months);
|
|
|
|
/// Averages [log]'s intake over `[start, end]` and classifies it.
|
|
///
|
|
/// An [end] before [start] is an empty period, not an error.
|
|
PeriodAverage periodAverage(
|
|
DayLog log, {
|
|
required BudgetSchedule schedule,
|
|
required DateTime start,
|
|
required DateTime end,
|
|
}) {
|
|
final from = _dateOnly(start);
|
|
final to = _dateOnly(end);
|
|
var totalKcal = 0.0;
|
|
var totalBudget = 0;
|
|
var loggedDays = 0;
|
|
var elapsedDays = 0;
|
|
for (var day = from; !day.isAfter(to); day = _addDays(day, 1)) {
|
|
elapsedDays++;
|
|
final key = dateKey(day);
|
|
final entries = (log[key] ?? const []).where((e) => !e.deleted);
|
|
if (entries.isEmpty) continue;
|
|
loggedDays++;
|
|
totalKcal += sumKcal(entries);
|
|
totalBudget += schedule.forDay(key);
|
|
}
|
|
if (loggedDays == 0) {
|
|
return PeriodAverage(
|
|
start: dateKey(from),
|
|
end: dateKey(to),
|
|
loggedDays: 0,
|
|
elapsedDays: elapsedDays,
|
|
avgKcal: null,
|
|
avgBudget: null,
|
|
band: null,
|
|
);
|
|
}
|
|
final avgKcal = totalKcal / loggedDays;
|
|
final avgBudget = totalBudget / loggedDays;
|
|
return PeriodAverage(
|
|
start: dateKey(from),
|
|
end: dateKey(to),
|
|
loggedDays: loggedDays,
|
|
elapsedDays: elapsedDays,
|
|
avgKcal: avgKcal,
|
|
avgBudget: avgBudget,
|
|
band: bandFor(avgKcal, avgBudget),
|
|
);
|
|
}
|
|
|
|
/// Returns the average for an ISO week, [weeksAgo] weeks back.
|
|
///
|
|
/// [weeksAgo] 0 is the current week through yesterday, 1 the previous complete
|
|
/// week, and so on.
|
|
PeriodAverage weeklyAverage(
|
|
DayLog log, {
|
|
required BudgetSchedule schedule,
|
|
int weeksAgo = 0,
|
|
DateTime? today,
|
|
}) {
|
|
final ref = _dateOnly(today ?? DateTime.now());
|
|
final anchor = _addDays(ref, -DateTime.daysPerWeek * weeksAgo);
|
|
return _capped(log, schedule, weekBounds(anchor), ref);
|
|
}
|
|
|
|
/// Returns the average for a calendar month, [monthsAgo] months back.
|
|
PeriodAverage monthlyAverage(
|
|
DayLog log, {
|
|
required BudgetSchedule schedule,
|
|
int monthsAgo = 0,
|
|
DateTime? today,
|
|
}) {
|
|
final ref = _dateOnly(today ?? DateTime.now());
|
|
return _capped(log, schedule, monthBounds(shiftMonths(ref, monthsAgo)), ref);
|
|
}
|
|
|
|
/// Averages [bounds], truncated so it never includes [ref] itself.
|
|
///
|
|
/// The single place the "today is excluded" rule is applied, so a caller
|
|
/// cannot construct a period that half-counts an unfinished day.
|
|
PeriodAverage _capped(
|
|
DayLog log,
|
|
BudgetSchedule schedule,
|
|
(DateTime, DateTime) bounds,
|
|
DateTime ref,
|
|
) {
|
|
final (start, end) = bounds;
|
|
final cap = lastCompleteDay(ref);
|
|
return periodAverage(
|
|
log,
|
|
schedule: schedule,
|
|
start: start,
|
|
end: end.isBefore(cap) ? end : cap,
|
|
);
|
|
}
|
|
|
|
DateTime _dateOnly(DateTime d) => DateTime(d.year, d.month, d.day);
|
|
|
|
/// Returns midnight [days] calendar days from [d], safely across DST.
|
|
///
|
|
/// Deliberately NOT `d.add(Duration(days: days))`: a [Duration] is an absolute
|
|
/// span, so on a local [DateTime] it moves by 23 or 25 hours across a DST
|
|
/// boundary and lands on the wrong wall-clock day. In Europe/Warsaw that made
|
|
/// `lastCompleteDay(2026-03-30)` return the day *before* yesterday, put the
|
|
/// `weeksAgo` anchor in the wrong ISO week, and silently drop the last day of
|
|
/// any period spanning spring-forward -- so the phone and the PC disagreed
|
|
/// about the same log for two weeks a year. Passing out-of-range components
|
|
/// to the constructor normalizes on the *calendar* instead, which is what
|
|
/// Python's `timedelta` on a `date` already does.
|
|
DateTime _addDays(DateTime d, int days) =>
|
|
DateTime(d.year, d.month, d.day + days);
|
|
|
|
/// Formats [d] as a `YYYY-MM-DD` log key.
|
|
String dateKey(DateTime d) =>
|
|
'${d.year.toString().padLeft(4, '0')}-'
|
|
'${d.month.toString().padLeft(2, '0')}-'
|
|
'${d.day.toString().padLeft(2, '0')}';
|