mirror of
https://github.com/kuhyx/diet-guard.git
synced 2026-11-11 14:07:40 +01:00
Krzysztof kuhy Rudnicki
c7371d7e0d
The button ran a full sync tick inline on the Tk thread, so the fullscreen lock was dead for its whole duration -- ~18-27s measured against the live remote, and unbounded in the worst case. Two changes, and the second is what makes the first cheap: * the pull moves to a daemon worker thread, with the result marshalled back through a queue polled on the Tk thread. The worker touches no widget, by construction: Tcl is not thread-safe and that includes `after`. Polling also makes the completion path testable with no thread at all. * the button now runs the *narrow* pull. Its question is identical to the gate's -- "did a peer log this slot?" -- and `_reconcile_after_fetch` reads only `due_slots()`, so the budget and food banks it used to fetch were never consulted here. The full tick still runs after the window closes. Measured 84ms against the live remote. Narrowing also shrinks the write race from tens of seconds to imperceptible: the pull persists its merge, so a meal logged mid-fetch could be overwritten by a snapshot taken before it. A second click while a fetch is in flight is now ignored, which is the reentrancy guard and the race fix in one. That guard started out as a disabled button, which is how it surfaced a latent bug worth recording: the fetch button widget is never stored on `GateWidgets`, so `self._widgets.fetch_button` did not exist and the disable was a silent no-op -- hidden because the suppression tuple included `AttributeError`. Coverage would have reported 100% on a guard that did nothing. Guarding on `_fetch_result` needs no widget at all. Also adds `INTERACTIVE_TIMEOUT_SECONDS = 2.0` for the paths a user waits on. `SYNC_TIMEOUT_SECONDS` only ever reached GitHub: crdt_sync v0.6.0's `firebase_client_for` takes no timeout, so Firebase used its 15s default and one hung session refresh stalled the lock window -- the journal has five such refresh failures in five days. Verified to fail closed to the local decision. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01LJ5YnA3CERitQ57HjMu2Qz
124 lines
7.3 KiB
Python
124 lines
7.3 KiB
Python
"""Constants for the diet_guard calorie tracker and gate."""
|
|
|
|
from __future__ import annotations
|
|
|
|
from pathlib import Path
|
|
|
|
# --- Daily target -----------------------------------------------------------
|
|
# There is deliberately NO budget number here. It is computed once from
|
|
# biometrics at ``init`` time, written to BUDGET_FILE below, and freely
|
|
# editable and synced afterward from either device -- see _budget.py and
|
|
# _sync.py. Read via diet_guard._budget.daily_budget() for over/under
|
|
# decisions; freely shown in local CLI/GUI/app output.
|
|
#
|
|
# Fraction of the budget at which status flips from "on track" to "approaching
|
|
# limit". Also mirrored above 100% as the day-status yellow/red boundary for
|
|
# the calendar -- see _daystatus.py.
|
|
BUDGET_WARN_FRACTION: float = 0.80
|
|
|
|
# --- Storage ----------------------------------------------------------------
|
|
# The food log is personal and high-churn, so it lives in the XDG data dir and
|
|
# is deliberately NOT committed to git (unlike wake_state.json).
|
|
DATA_DIR: Path = Path.home() / ".local" / "share" / "diet_guard"
|
|
FOOD_LOG_FILE: Path = DATA_DIR / "food_log.json"
|
|
|
|
# Revision cache for cross-device sync. Beside the log it describes and
|
|
# cleared with it: skipping an unchanged peer is only sound because that
|
|
# peer's records are already merged into the local log, so state that outlived
|
|
# its log would skip peers whose data had been lost.
|
|
SYNC_STATE_FILE: Path = DATA_DIR / "sync_state.json"
|
|
# The user's personal "food bank": every food they have logged before, with its
|
|
# full macros, keyed by name. This is the ONLY corpus the gate's autocomplete
|
|
# searches -- Open Food Facts is used to *fill* a new food's macros, never to
|
|
# search. Local-only, git-ignored.
|
|
FOOD_BANK_FILE: Path = DATA_DIR / "food_bank.json"
|
|
# Hand-curated bank entries (see _foodbank_manual.py): foods added without
|
|
# ever logging them, so they are NOT derivable from the food log and must
|
|
# sync in their own right. Kept separate from FOOD_BANK_FILE because that
|
|
# one is rewritten wholesale on every log write.
|
|
MANUAL_BANK_FILE: Path = DATA_DIR / "food_bank_manual.json"
|
|
# The budget: a plain JSON dotfile alongside the log, freely editable on this
|
|
# device or the phone app and synced between them (see _sync.py).
|
|
# Git-ignored, never committed. "Hidden" here means never-online (it lives
|
|
# outside the repo) -- the number itself is shown freely in local CLI/GUI/app
|
|
# output.
|
|
BUDGET_FILE: Path = DATA_DIR / ".budget"
|
|
# The effective-from history of that budget (see _budget_history.py), kept in
|
|
# its own file so ``.budget``'s schema stays at v2 and every existing reader
|
|
# of it is untouched. Classifying a *past* day reads this; everything about
|
|
# *today* still reads BUDGET_FILE. Git-ignored, never committed.
|
|
BUDGET_HISTORY_FILE: Path = DATA_DIR / ".budget_history"
|
|
# The effective-from history of the meal schedule (see _meal_schedule_store.py)
|
|
# -- when the user eats their first and last meal, and how many meals fall in
|
|
# between. Forward-only for the same reason as the budget history: switching
|
|
# from four meals to five must not retroactively mark every past day as having
|
|
# missed a checkpoint. Git-ignored, never committed.
|
|
MEAL_SCHEDULE_FILE: Path = DATA_DIR / ".meal_schedule"
|
|
|
|
# --- Estimator (Open Food Facts) -------------------------------------------
|
|
# The default backend is Open Food Facts' "Search-a-licious" full-text search:
|
|
# free, no key, strongest for branded/packaged foods (including fast food).
|
|
# (The older cgi/search.pl endpoint is heavily rate-limited and returns an HTML
|
|
# "temporarily unavailable" page to API clients, and /api/v2/search ignores the
|
|
# query term, so neither is usable here.) Swappable for a local/remote LLM
|
|
# backend later without touching the log or CLI layers.
|
|
OFF_SEARCH_URL: str = "https://search.openfoodfacts.org/search"
|
|
OFF_TIMEOUT_SECONDS: float = 8.0
|
|
OFF_PAGE_SIZE: int = 5
|
|
# Open Food Facts asks API clients to identify themselves with a descriptive
|
|
# User-Agent string so abusive clients can be told apart from polite ones.
|
|
OFF_USER_AGENT: str = "diet_guard/1.0 (personal diet tracker)"
|
|
# Portion assumed when neither --grams nor an OFF serving size is available.
|
|
DEFAULT_PORTION_GRAMS: float = 100.0
|
|
|
|
# --- Gate (log-to-unlock) ---------------------------------------------------
|
|
# The gate is driven by FIXED MEAL SLOTS, not by a gap timer. Starting at the
|
|
# day-start hour, a slot opens every interval; once a slot's hour has passed,
|
|
# that slot must carry a logged meal or the screen locks until it does. This
|
|
# makes tracking fully automatic (you are prompted on a schedule rather than
|
|
# trusted to log voluntarily) and nudges regular eating. Coming home late
|
|
# naturally produces several unlogged elapsed slots at once -> one lock that
|
|
# backfills the whole day, which is the "requirement to access the PC" behavior.
|
|
GATE_DAY_START_HOUR: int = 8 # first slot (08:00); also the "beginning of day"
|
|
GATE_SLOT_INTERVAL_HOURS: int = 4 # slots at 08:00, 12:00, 16:00, 20:00
|
|
# Past this hour the gate never fires, so an unlogged late slot lapses quietly
|
|
# instead of locking you out overnight. (A new day resets all slots at 00:00.)
|
|
GATE_EATING_END_HOUR: int = 22 # exclusive (22:00)
|
|
# flock single-instance guard: stops a timer from stacking lock windows.
|
|
GATE_LOCK_FILE: Path = DATA_DIR / ".gate.lock"
|
|
|
|
# --- Sync (cross-device log merge) ------------------------------------------
|
|
# GitHub is used purely as dumb file storage via the REST Contents API (not a
|
|
# git clone) -- mirrors ~/todo's sync transport. Each device pushes its own
|
|
# full current log as one file under devices/<id>/food_log.json; merging
|
|
# happens client-side (see _sync_merge.py), never via git.
|
|
SYNC_REPO_OWNER: str = "kuhyx"
|
|
SYNC_REPO_NAME: str = "syncs"
|
|
# The id this machine pushed under before per-install uuids. The sync reader
|
|
# still treats it as this device's own, so the log pushed under it is not
|
|
# pulled back and re-merged as a peer's; drop to None once devices/pc/ has
|
|
# been reclaimed. See SYNC_DEVICE_ID_FILE.
|
|
SYNC_LEGACY_DEVICE_ID: str | None = "pc"
|
|
# This device's sync id, minted once and persisted. A fixed "pc" collides the
|
|
# moment a second machine takes the same role, and after a reinstall the new
|
|
# install would inherit the old one's CRDT identity; a uuid makes both
|
|
# impossible by construction. Lives beside the other state, not in ~/.config,
|
|
# because losing it means being seen as a brand-new device.
|
|
SYNC_DEVICE_ID_FILE: Path = DATA_DIR / ".device_id"
|
|
# A fine-grained GitHub PAT, scoped to just SYNC_REPO_NAME's contents. The
|
|
# user creates this once via github.com (see CLAUDE.md) and saves it here,
|
|
# mode 600. Never committed -- this path is outside the repo entirely.
|
|
SYNC_TOKEN_FILE: Path = Path.home() / ".config" / "diet_guard" / "sync_token"
|
|
SYNC_TIMEOUT_SECONDS: float = 10.0
|
|
|
|
#: Per-request budget for a pull the user is actively waiting on (the gate's
|
|
#: pre-lock refresh and the lock screen's "Fetch from sync" button).
|
|
#:
|
|
#: Deliberately far below :data:`SYNC_TIMEOUT_SECONDS`: those paths run 1-2
|
|
#: requests and must resolve inside a second, whereas the background tick can
|
|
#: afford to be patient. The default is worse than it looks -- ``crdt_sync``
|
|
#: v0.6.0's ``firebase_client_for`` takes no timeout at all, so Firebase falls
|
|
#: back to *15s per request* and one hung session refresh stalls the lock
|
|
#: window. The journal records five such refresh failures in five days.
|
|
INTERACTIVE_TIMEOUT_SECONDS: float = 2.0
|