Stride/app/records.py

"""Estimated distance best efforts from full-resolution recorded samples."""

from bisect import bisect_left, bisect_right

import click

from .analytics import numeric

EFFORTS_VERSION = 1
DISTANCES = (
    ("1 km", 1000),
    ("2 km", 2000),
    ("5 km", 5000),
    ("10 km", 10000),
    ("Half marathon", 21097.5),
    ("Marathon", 42195),
)


def best_efforts(activity):
    """Search piecewise-linear distance/time curves, counting internal stops.

    A minimum starts or ends at a recorded distance breakpoint. Evaluate both
    sets, interpolating the opposite boundary. At stationary boundaries use
    latest departure and earliest arrival, excluding stops outside the effort.
    Missing samples, distance resets and non-increasing times break the curve.
    Timestamp gaps remain elapsed time, including pauses in watch recording.
    """
    if activity.sport != "running":
        return []
    segments, segment = [], []
    for sample in activity.samples:
        distance, time = sample.get("distance"), sample.get("time")
        if not numeric(distance) or not numeric(time) or distance < 0:
            if segment:
                segments.append(segment)
            segment = []
            continue
        if segment and (distance < segment[-1][0] or time <= segment[-1][1]):
            segments.append(segment)
            segment = []
        segment.append((distance, time))
    if segment:
        segments.append(segment)
    best = {}
    for segment in segments:
        distances = [point[0] for point in segment]
        times = [point[1] for point in segment]

        def at(distance, departure=False, distances=distances, times=times):
            index = (
                bisect_right(distances, distance) - 1
                if departure
                else bisect_left(distances, distance)
            )
            if distances[index] == distance:
                return times[index]
            if departure:
                index += 1
            fraction = (distance - distances[index - 1]) / (distances[index] - distances[index - 1])
            return times[index - 1] + fraction * (times[index] - times[index - 1])

        for label, length in DISTANCES:
            if distances[-1] - distances[0] < length:
                continue
            starts = {d for d in distances if d + length <= distances[-1]}
            starts.update(d - length for d in distances if d - length >= distances[0])
            for start in sorted(starts):
                seconds = at(start + length) - at(start, departure=True)
                if seconds > 0 and (label not in best or seconds < best[label]["seconds"]):
                    best[label] = {
                        "label": label,
                        "distance": length,
                        "seconds": seconds,
                        "start_m": start,
                        "end_m": start + length,
                    }
    return [best[label] for label, _ in DISTANCES if label in best]


def persist_efforts(activity):
    """Store derived data in the same transaction as the imported run."""
    activity.best_efforts = best_efforts(activity)
    activity.efforts_version = EFFORTS_VERSION
    activity.has_route = any(sample.get("point") for sample in activity.samples)


def record_book(activities):
    """Compare persisted efforts in chronology; never read samples here.

    First efforts establish a record. Strict improvements earn historical PR
    badges, which remain on the run after a later activity beats that record.
    """
    records, runs = {}, {}
    for activity in sorted(activities, key=lambda row: (row.started_at, row.id)):
        efforts = (
            [dict(effort) for effort in activity.best_efforts]
            if activity.efforts_version == EFFORTS_VERSION and activity.sport == "running"
            else []
        )
        for effort in efforts:
            previous = records.get(effort["label"])
            effort["pr"] = previous is None or effort["seconds"] < previous["seconds"] - 1e-6
            if effort["pr"]:
                records[effort["label"]] = dict(effort, activity=activity)
        runs[activity.id] = efforts
    return {"records": [records[label] for label, _ in DISTANCES if label in records], "runs": runs}


def register_cli(app):
    @app.cli.command("backfill-efforts")
    @click.option("--batch-size", default=25, type=click.IntRange(1, 1000), show_default=True)
    @click.option("--owner-id", type=click.IntRange(1), help="Limit to one athlete.")
    @click.option("--force", is_flag=True, help="Recalculate even current-version efforts.")
    def backfill(batch_size, owner_id, force):
        """Backfill persisted best efforts in bounded, resumable transactions."""
        from sqlalchemy.orm import undefer

        from .extensions import db
        from .models import Activity

        last_id, count = 0, 0
        # A fixed upper bound prevents concurrent imports extending this run forever.
        maximum = db.session.scalar(db.select(db.func.max(Activity.id))) or 0
        while last_id < maximum:
            query = db.select(Activity).where(Activity.id > last_id, Activity.id <= maximum)
            if owner_id:
                query = query.where(Activity.owner_id == owner_id)
            if not force:
                query = query.where(Activity.efforts_version != EFFORTS_VERSION)
            rows = db.session.scalars(
                query.options(undefer(Activity.samples))
                .order_by(Activity.id)
                .limit(batch_size)
                .with_for_update()
            ).all()
            if not rows:
                break
            try:
                for row in rows:
                    persist_efforts(row)
                last_id = rows[-1].id
                db.session.commit()
            except Exception as error:
                db.session.rollback()
                raise click.ClickException(
                    "Backfill failed; completed batches are saved. Rerun to resume."
                ) from error
            count += len(rows)
            db.session.expunge_all()
            click.echo(f"Backfilled {count} activities (through ID {last_id}).")
        click.echo(f"Complete: {count} activities backfilled; effort version {EFFORTS_VERSION}.")