training/make_dataset.py

"""Build a synthetic instruction dataset: thread (+image descriptions) -> ActionPlan JSON.

Reuses the agent's SYSTEM prompt and ActionPlan schema so training and serving
stay in sync. This is a template with a few hand-written seeds plus simple
augmentation — expand toward ~500-2000 examples covering relative dates, ranges,
conflicts, no-event chitchat (empty), multiple events, and image-derived events.

Output: training/data/dataset.jsonl  (one {messages:[...]} chat record per line)
Generating training data may use any offline tooling — the "no cloud API" rule
applies to the running app's inference, not dataset prep.

Vision note: we keep the vision encoder frozen at train time, so image examples
are represented by a short text *description* of the image (what the model would
"see"); this teaches the decision/format behavior without multimodal training.
"""
from __future__ import annotations

import json
import sys
from pathlib import Path

sys.path.insert(0, str(Path(__file__).resolve().parent.parent))
from server.agent import SYSTEM  # noqa: E402  reuse the serving prompt

# (current_datetime, thread, action_plan_dict)
SEEDS = [
    (
        "2026-06-08T09:00:00",
        "Alex: lunch tomorrow at 1pm?\nMe: works, the usual place",
        {
            "reasoning": "Alex proposed lunch tomorrow at 1pm and it was accepted.",
            "events": [
                {
                    "title": "Lunch with Alex",
                    "start": "2026-06-09T13:00:00",
                    "end": "2026-06-09T14:00:00",
                    "location": "the usual place",
                    "attendees": ["Alex"],
                    "reminder_minutes": 30,
                    "notes": None,
                }
            ],
            "conflicts": [],
            "proposed_times": [],
            "reply_draft": "Perfect, see you at 1 at the usual place!",
            "needs_clarification": None,
        },
    ),
    (
        "2026-06-08T09:00:00",
        "Sam: dentist moved to next Thursday 3:30\nMe: noted",
        {
            "reasoning": "Sam rescheduled the dentist to next Thursday at 3:30pm.",
            "events": [
                {
                    "title": "Dentist",
                    "start": "2026-06-11T15:30:00",
                    "end": "2026-06-11T16:30:00",
                    "location": None,
                    "attendees": [],
                    "reminder_minutes": 60,
                    "notes": None,
                }
            ],
            "conflicts": [],
            "proposed_times": [],
            "reply_draft": "Got it, I'll be there Thursday at 3:30.",
            "needs_clarification": None,
        },
    ),
    (
        "2026-06-08T09:00:00",
        "Jo: haha that meme killed me\nMe: lol",
        {
            "reasoning": "Casual chat, no plan to schedule.",
            "events": [],
            "conflicts": [],
            "proposed_times": [],
            "reply_draft": "😂",
            "needs_clarification": None,
        },
    ),
    (
        "2026-06-08T09:00:00",
        "Pat: you free sometime this week to grab coffee?",
        {
            "reasoning": "A meetup is proposed but no concrete day/time was given.",
            "events": [],
            "conflicts": [],
            "proposed_times": [],
            "reply_draft": "Yeah! What day works for you?",
            "needs_clarification": "Which day/time this week works for coffee?",
        },
    ),
    (
        # image-derived example (vision encoder frozen -> describe the image in text)
        "2026-06-08T09:00:00",
        "Riley: [image: screenshot of a concert ticket — 'The Lumens, Fri Jun 12, "
        "doors 7:30pm, The Fillmore']\nMe: can't wait!",
        {
            "reasoning": "The screenshot is a concert ticket with a concrete date, time, and venue.",
            "events": [
                {
                    "title": "The Lumens concert",
                    "start": "2026-06-12T19:30:00",
                    "end": "2026-06-12T22:30:00",
                    "location": "The Fillmore",
                    "attendees": ["Riley"],
                    "reminder_minutes": 120,
                    "notes": "From ticket screenshot",
                }
            ],
            "conflicts": [],
            "proposed_times": [],
            "reply_draft": "Added it — doors at 7:30, see you Friday!",
            "needs_clarification": None,
        },
    ),
    # --- multiple events in one thread (mirrors the live "Try a sample") ---
    (
        "2026-06-08T08:00:00",
        "Room parent: picture day is this Thursday at 9am, green shirts!\n"
        "Coach Dana: heads up, soccer moves to Tuesday 5pm this week\nMe: thanks, adding both",
        {
            "reasoning": "Two concrete events: picture day Thursday 9am and the rescheduled soccer Tuesday 5pm.",
            "events": [
                {
                    "title": "Picture day (green shirt)",
                    "start": "2026-06-11T09:00:00",
                    "end": "2026-06-11T09:30:00",
                    "location": None,
                    "attendees": [],
                    "reminder_minutes": 720,
                    "notes": "Wear the green class shirt",
                },
                {
                    "title": "Soccer practice",
                    "start": "2026-06-09T17:00:00",
                    "end": "2026-06-09T18:00:00",
                    "location": None,
                    "attendees": ["Coach Dana"],
                    "reminder_minutes": 60,
                    "notes": None,
                },
            ],
            "conflicts": [],
            "proposed_times": [],
            "reply_draft": "Got both down — picture day Thursday 9am and soccer Tuesday 5pm!",
            "needs_clarification": None,
        },
    ),
    # --- relative date: "next Friday" ---
    (
        "2026-06-08T09:00:00",
        "Teacher: our class field trip permission slips are due next Friday\nMe: ok!",
        {
            "reasoning": "A due date stated as 'next Friday' relative to Monday Jun 8 resolves to Jun 19.",
            "events": [
                {
                    "title": "Permission slips due",
                    "start": "2026-06-19T09:00:00",
                    "end": None,
                    "location": None,
                    "attendees": [],
                    "reminder_minutes": 1440,
                    "notes": "Class field trip",
                }
            ],
            "conflicts": [],
            "proposed_times": [],
            "reply_draft": "Thanks — I'll have them signed before Friday.",
            "needs_clarification": None,
        },
    ),
    # --- multi-hour range: field trip 9-3 ---
    (
        "2026-06-08T09:00:00",
        "Room parent: field trip to the science museum is Wed 9am–3pm, bring a bagged lunch\nMe: noted",
        {
            "reasoning": "A single all-day-ish event with an explicit start and end on Wednesday.",
            "events": [
                {
                    "title": "Science museum field trip",
                    "start": "2026-06-10T09:00:00",
                    "end": "2026-06-10T15:00:00",
                    "location": "Science museum",
                    "attendees": [],
                    "reminder_minutes": 720,
                    "notes": "Bring a bagged lunch",
                }
            ],
            "conflicts": [],
            "proposed_times": [],
            "reply_draft": "Got it — bagged lunch, 9 to 3 Wednesday.",
            "needs_clarification": None,
        },
    ),
    # --- multi-day range: book fair (3 days) ---
    (
        "2026-06-08T09:00:00",
        "PTA: the book fair runs Mon–Wed next week in the library, open during pickup\nMe: 👍",
        {
            "reasoning": "A multi-day event spanning Monday to Wednesday of next week.",
            "events": [
                {
                    "title": "Book fair",
                    "start": "2026-06-15T08:00:00",
                    "end": "2026-06-17T16:00:00",
                    "location": "School library",
                    "attendees": [],
                    "reminder_minutes": 1440,
                    "notes": "Open during pickup",
                }
            ],
            "conflicts": [],
            "proposed_times": [],
            "reply_draft": "Thanks — we'll stop by after pickup.",
            "needs_clarification": None,
        },
    ),
    # --- no-event chitchat (empty) ---
    (
        "2026-06-08T09:00:00",
        "Jess: did you see the class photo album?? so cute\nMe: omg yes 😍",
        {
            "reasoning": "Friendly chatter, nothing to schedule.",
            "events": [],
            "conflicts": [],
            "proposed_times": [],
            "reply_draft": "They grow up so fast 🥹",
            "needs_clarification": None,
        },
    ),
    (
        "2026-06-08T09:00:00",
        "Coach Dana: great hustle from the kids today, proud of them\nMe: such a fun season",
        {
            "reasoning": "Encouragement, no plan or date mentioned.",
            "events": [],
            "conflicts": [],
            "proposed_times": [],
            "reply_draft": "Thanks Coach — they love it!",
            "needs_clarification": None,
        },
    ),
    # --- conflict against existing calendar + proposed alternatives ---
    (
        "2026-06-08T09:00:00",
        "Existing calendar:\n- Dentist (Mia): 2026-06-10T14:00..2026-06-10T15:00\n\n"
        "Room parent: makeup class photo is Wednesday at 2pm\nMe: hmm Mia has the dentist then",
        {
            "reasoning": "The makeup photo at Wed 2pm overlaps Mia's existing dentist appointment.",
            "events": [
                {
                    "title": "Makeup class photo",
                    "start": "2026-06-10T14:00:00",
                    "end": "2026-06-10T14:30:00",
                    "location": None,
                    "attendees": [],
                    "reminder_minutes": 120,
                    "notes": None,
                }
            ],
            "conflicts": [
                {
                    "event_index": 0,
                    "clashes_with": "Dentist (Mia) 2:00-3:00pm",
                    "severity": "overlap",
                }
            ],
            "proposed_times": ["2026-06-10T15:00:00", "2026-06-11T14:00:00"],
            "reply_draft": "Mia has the dentist at 2 — could she do the makeup photo at 3, or Thursday?",
            "needs_clarification": None,
        },
    ),
    (
        "2026-06-08T09:00:00",
        "Existing calendar:\n- Work standup: 2026-06-09T09:00..2026-06-09T09:30\n\n"
        "Teacher: parent volunteers needed in class Tue 9–10am\nMe: I'll try",
        {
            "reasoning": "Volunteering Tue 9-10am butts against the 9:00-9:30 standup (overlap).",
            "events": [
                {
                    "title": "Classroom volunteering",
                    "start": "2026-06-09T09:00:00",
                    "end": "2026-06-09T10:00:00",
                    "location": "Classroom",
                    "attendees": [],
                    "reminder_minutes": 60,
                    "notes": None,
                }
            ],
            "conflicts": [
                {
                    "event_index": 0,
                    "clashes_with": "Work standup 9:00-9:30am",
                    "severity": "overlap",
                }
            ],
            "proposed_times": ["2026-06-09T10:00:00", "2026-06-11T09:00:00"],
            "reply_draft": "I have a 9am call — could I come in at 10 instead?",
            "needs_clarification": None,
        },
    ),
    # --- image-described event: PTA flyer ---
    (
        "2026-06-08T09:00:00",
        "Room parent: [image: flyer — 'Spring Family Picnic, Sat Jun 13, 11am, Lincoln Park "
        "Shelter B, bring a dish to share']\nMe: we'll be there",
        {
            "reasoning": "The flyer gives a concrete date, time, and location for the family picnic.",
            "events": [
                {
                    "title": "Spring Family Picnic",
                    "start": "2026-06-13T11:00:00",
                    "end": "2026-06-13T14:00:00",
                    "location": "Lincoln Park, Shelter B",
                    "attendees": [],
                    "reminder_minutes": 120,
                    "notes": "Bring a dish to share (from flyer)",
                }
            ],
            "conflicts": [],
            "proposed_times": [],
            "reply_draft": "Added it — we'll bring a salad!",
            "needs_clarification": None,
        },
    ),
    # --- image-described event: permission slip photo ---
    (
        "2026-06-08T09:00:00",
        "Mom: [image: photo of a permission slip — 'Zoo trip Thursday June 11, depart 8:30am, "
        "return ~2:30pm, $12 due Mon']\nMe: signing tonight",
        {
            "reasoning": "Permission slip photo encodes the trip date/time and a separate payment deadline.",
            "events": [
                {
                    "title": "Zoo trip",
                    "start": "2026-06-11T08:30:00",
                    "end": "2026-06-11T14:30:00",
                    "location": "Zoo",
                    "attendees": [],
                    "reminder_minutes": 720,
                    "notes": "$12 due Monday; from permission slip",
                }
            ],
            "conflicts": [],
            "proposed_times": [],
            "reply_draft": "Signed and the $12 is ready for Monday!",
            "needs_clarification": None,
        },
    ),
    # --- ambiguity: needs clarification, no event yet ---
    (
        "2026-06-08T09:00:00",
        "Teacher: can a couple of parents help with the class party sometime next week?\nMe: maybe!",
        {
            "reasoning": "A request without a fixed day or time — ask before scheduling anything.",
            "events": [],
            "conflicts": [],
            "proposed_times": [],
            "reply_draft": "Happy to help — what day and time is the party?",
            "needs_clarification": "Which day/time next week is the class party?",
        },
    ),
    # --- explicit date "the 14th" ---
    (
        "2026-06-08T09:00:00",
        "Coach Dana: end-of-season pizza party is the 14th at 6pm at Tony's\nMe: yum, count us in",
        {
            "reasoning": "'The 14th' resolves to Jun 14; concrete time and venue given.",
            "events": [
                {
                    "title": "End-of-season pizza party",
                    "start": "2026-06-14T18:00:00",
                    "end": "2026-06-14T20:00:00",
                    "location": "Tony's",
                    "attendees": ["Coach Dana"],
                    "reminder_minutes": 120,
                    "notes": None,
                }
            ],
            "conflicts": [],
            "proposed_times": [],
            "reply_draft": "Count us in — see you at Tony's at 6!",
            "needs_clarification": None,
        },
    ),
    # --- "in two weeks" ---
    (
        "2026-06-08T09:00:00",
        "Teacher: the spring recital is in two weeks, Friday at 6pm in the gym\nMe: 📅",
        {
            "reasoning": "Friday two weeks out is Jun 19? The teacher pins it to Friday 6pm — Jun 19.",
            "events": [
                {
                    "title": "Spring recital",
                    "start": "2026-06-19T18:00:00",
                    "end": "2026-06-19T19:30:00",
                    "location": "School gym",
                    "attendees": [],
                    "reminder_minutes": 1440,
                    "notes": None,
                }
            ],
            "conflicts": [],
            "proposed_times": [],
            "reply_draft": "Can't wait — we'll be there Friday at 6!",
            "needs_clarification": None,
        },
    ),
    # --- early dismissal ---
    (
        "2026-06-08T09:00:00",
        "School office: reminder, early dismissal this Wednesday at 12:30pm\nMe: thanks for the heads up",
        {
            "reasoning": "A one-off schedule change parents need to plan pickup around.",
            "events": [
                {
                    "title": "Early dismissal",
                    "start": "2026-06-10T12:30:00",
                    "end": None,
                    "location": "School",
                    "attendees": [],
                    "reminder_minutes": 60,
                    "notes": "Pickup at 12:30 instead of usual",
                }
            ],
            "conflicts": [],
            "proposed_times": [],
            "reply_draft": "Noted — I'll be there at 12:30 Wednesday.",
            "needs_clarification": None,
        },
    ),
    # --- parent-teacher conference (specific slot) ---
    (
        "2026-06-08T09:00:00",
        "Teacher: your parent–teacher conference slot is Thursday 4:15pm, ~15 min\nMe: works for me",
        {
            "reasoning": "A confirmed 15-minute conference slot on Thursday.",
            "events": [
                {
                    "title": "Parent–teacher conference",
                    "start": "2026-06-11T16:15:00",
                    "end": "2026-06-11T16:30:00",
                    "location": "Classroom",
                    "attendees": ["Teacher"],
                    "reminder_minutes": 120,
                    "notes": None,
                }
            ],
            "conflicts": [],
            "proposed_times": [],
            "reply_draft": "4:15 Thursday works — see you then!",
            "needs_clarification": None,
        },
    ),
    # --- birthday party RSVP with location ---
    (
        "2026-06-08T09:00:00",
        "Other parent: Leo's bday party Sat 2–4pm at Jump Zone, can Mia come?\nMe: she'd love to!",
        {
            "reasoning": "A party invite with a clear window and venue, accepted.",
            "events": [
                {
                    "title": "Leo's birthday party",
                    "start": "2026-06-13T14:00:00",
                    "end": "2026-06-13T16:00:00",
                    "location": "Jump Zone",
                    "attendees": ["Leo"],
                    "reminder_minutes": 180,
                    "notes": None,
                }
            ],
            "conflicts": [],
            "proposed_times": [],
            "reply_draft": "She'd love to — see you Saturday at Jump Zone!",
            "needs_clarification": None,
        },
    ),
    # --- carpool reminder ---
    (
        "2026-06-08T09:00:00",
        "Neighbor: I've got carpool drop-off Friday at 7:45am this week\nMe: perfect, thank you!",
        {
            "reasoning": "A specific carpool commitment on Friday morning.",
            "events": [
                {
                    "title": "Carpool drop-off (neighbor driving)",
                    "start": "2026-06-12T07:45:00",
                    "end": None,
                    "location": None,
                    "attendees": [],
                    "reminder_minutes": 30,
                    "notes": None,
                }
            ],
            "conflicts": [],
            "proposed_times": [],
            "reply_draft": "Thank you! I've got next Friday.",
            "needs_clarification": None,
        },
    ),
    # --- reschedule (supersede an earlier time) ---
    (
        "2026-06-08T09:00:00",
        "Coach Dana: practice this Tue is now 6pm, not 5\nMe: ok updating",
        {
            "reasoning": "A reschedule: practice Tuesday moves to 6pm.",
            "events": [
                {
                    "title": "Soccer practice",
                    "start": "2026-06-09T18:00:00",
                    "end": "2026-06-09T19:00:00",
                    "location": None,
                    "attendees": ["Coach Dana"],
                    "reminder_minutes": 60,
                    "notes": "Moved from 5pm to 6pm",
                }
            ],
            "conflicts": [],
            "proposed_times": [],
            "reply_draft": "Updated — 6pm Tuesday it is.",
            "needs_clarification": None,
        },
    ),
    # --- bake sale drop-off deadline ---
    (
        "2026-06-08T09:00:00",
        "PTA: bake sale items drop off Thursday by 8am at the front office\nMe: I'll bring cookies",
        {
            "reasoning": "A drop-off deadline Thursday morning.",
            "events": [
                {
                    "title": "Bake sale drop-off",
                    "start": "2026-06-11T08:00:00",
                    "end": None,
                    "location": "Front office",
                    "attendees": [],
                    "reminder_minutes": 720,
                    "notes": "Bringing cookies",
                }
            ],
            "conflicts": [],
            "proposed_times": [],
            "reply_draft": "Cookies will be there by 8 Thursday!",
            "needs_clarification": None,
        },
    ),
    # --- vague meetup, needs clarification ---
    (
        "2026-06-08T09:00:00",
        "Other parent: we should set up a playdate soon!\nMe: yes definitely",
        {
            "reasoning": "Mutual interest but no concrete day, time, or place.",
            "events": [],
            "conflicts": [],
            "proposed_times": [],
            "reply_draft": "Let's do it — what day works for you two?",
            "needs_clarification": "Which day/time and where for the playdate?",
        },
    ),
]


def to_record(now: str, thread: str, plan: dict) -> dict:
    user = f"Current datetime: {now}\nExisting calendar: (none provided)\n\nConversation:\n{thread}\n\nReturn the ActionPlan JSON now."
    return {
        "messages": [
            {"role": "system", "content": SYSTEM},
            {"role": "user", "content": user},
            {"role": "assistant", "content": json.dumps(plan, ensure_ascii=False)},
        ]
    }


def main():
    out_dir = Path(__file__).with_name("data")
    out_dir.mkdir(exist_ok=True)
    out = out_dir / "dataset.jsonl"
    with out.open("w", encoding="utf-8") as f:
        for now, thread, plan in SEEDS:
            f.write(json.dumps(to_record(now, thread, plan), ensure_ascii=False) + "\n")
    print(f"wrote {out} ({len(SEEDS)} seed records) — expand this before training")


if __name__ == "__main__":
    main()