|
Some checks are pending
dependency freshness / dependencies on newest stable (push) Waiting to run
file length / file-length (push) Waiting to run
focus-project pre-commit / pre-commit (push) Waiting to run
focus-project tests / test (push) Waiting to run
md-naming / markdown naming convention (push) Waiting to run
palette drift / palette (push) Waiting to run
repo-contract / repo contract (push) Waiting to run
shared gate tests / gate unit tests (push) Waiting to run
A score row completes mock only after a contiguous >= 90 min focus session (600 s gap rule) that ended by the score row. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01WzxdWZUTC9Zzdzsx7VW7dm |
||
|---|---|---|
| .. | ||
| focus_project | ||
| systemd | ||
| .pre-commit-config.yaml | ||
| install_harvest.sh | ||
| install.sh | ||
| pyproject.toml | ||
| README.md | ||
| requirements.txt | ||
focus_project
Time-boxed focus projects ("until Thursday I prepare for X"). The active project switches chosen gates off and defines a daily-minutes gate.
import focus_project as fp
fp.stands_down("leetcode", today) # gate in leetcode|book|reading|tutor; never raises
fp.active(today) # Project | None
fp.status(today) # dict for i3bar / GUIs
fp.record_activity("timer", "start")
Data: $FOCUS_PROJECT_DATA or ~/.local/share/focus_project/ (projects.json,
activity.jsonl). minutes.py is owned by the minutes/harvester agent.
Time rules: minutes_for clips every interval at "now" (an event stamped ahead of
the clock earns only once the clock reaches it); record_activity rejects a ts
more than 120 s ahead (ValueError). Timer segments (start -> checkin ... -> stop)
lapse after a 600 s gap; a tutor/heartbeat is a beat too (0.8.0: a graded check-in card
extends the open segment, never opens one). end_project(id, day) rejects a past day
(ValueError); today or later is fine. The harvester pays every project day of the
last 31 days, so late-synced events for older days are credited (idempotent).
Overtime carries over: own minutes past a day's target bank (max 60) and pay on the
next project day; the bank never crosses the project's end. minutes_for(day) is the
capped pay figure, status(day)["minutes"] the real count (121/120) with carried_in,
carry_out, own_minutes, counted_minutes. Carry ledger rows: <project>-<YYYYMMDD>-carry
(detail.carry), never over daily_minutes together with own rows.
Edits only ratchet (0.7.0): update_project/apply_update(id, today=...) never lowers
today's obligation. A lower daily_minutes/unlock_minutes, or a gate added to replaces,
on a started project holds from TOMORROW (a history row ends today with the old values,
replaces_on(day)/daily_on/unlock_on read it); a mixed edit defers as a whole. Raises,
gates removed and edits to a project that has not started hold from today.
apply_update(...).effective_from says which. end cannot be before today, so an edit never
uncovers today. A NEW project starting today may replace gates and set any targets: it is a
new obligation, not a reduction (but see the floor).
Floor (0.7.1, coordinator decision for kuhy): daily_minutes >= 30 x replaced gates (min 30; book = reading),
enforced on create and update (daily_floor(replaces)), so a day cannot be made cheap.
Open the work: python3 -m focus_project.open [--no-site] (tutor focus screen, work_url).
Install: ./install.sh (stdlib only, editable user install for /usr/bin/python3).
Interview prep (kind="interview", 0.9): create_project(..., kind="interview", test_at=, invite_url=, trips=, weekend_answers=);
end must be the test day. Budget: 30 min weekdays / 120 Sat-Sun (daily_on), exempt from daily_floor, unlock clamped.
Milestones are generated, never stored: milestones(project, now) -> rules, trip_ask:<sat>, phone:<trip start>, mock, slot
with states pending/done/missed/late_done/unknown. owed_lock(now) = earliest overdue uncompleted one (never on test day; wire it
after focus-guard's sick/free-day stand-down). status(day) gains milestones and missed.
Evidence is append-only (evidence/<id>.jsonl + copied files): record_evidence(id, milestone, kind, value, file=, now=)
(file, tap, score, slot, slot_confirm, tutor's phone_check). add_trip / answer_weekend only get stricter (ValueError).
run_checks(id) runs the phone source (reads the latest phone_check row) and latches changes; register_source swaps it.