backend: initial commit

This commit is contained in:
Davide Scaini
2026-03-28 13:57:12 +01:00
commit 38c5423aeb
36 changed files with 2463 additions and 0 deletions
+198
View File
@@ -0,0 +1,198 @@
"""Write a processed activity to BAS JSON files."""
import json
import re
import unicodedata
from pathlib import Path
from bincio.extract.metrics import ComputedMetrics
from bincio.extract.models import LapData, ParsedActivity
from bincio.extract.simplify import build_geojson
from bincio.extract.timeseries import build_timeseries
def make_activity_id(activity: ParsedActivity) -> str:
"""Generate a BAS activity ID from started_at + optional title slug."""
ts = activity.started_at
# Compact ISO format: 2024-06-01T073012+0200
tz_str = ts.strftime("%z") # e.g. "+0200" or ""
ts_part = ts.strftime("%Y-%m-%dT%H%M%S") + (tz_str or "Z")
if activity.title:
slug = _slugify(activity.title)
return f"{ts_part}-{slug}" if slug else ts_part
return ts_part
def write_activity(
activity: ParsedActivity,
metrics: ComputedMetrics,
output_dir: Path,
privacy: str = "public",
duplicate_of: str | None = None,
rdp_epsilon: float = 0.0001,
) -> str:
"""Write {id}.json and (if GPS available) {id}.geojson. Returns the ID."""
activity_id = make_activity_id(activity)
acts_dir = output_dir / "activities"
acts_dir.mkdir(parents=True, exist_ok=True)
source = _infer_source(activity)
has_gps = metrics.bbox is not None and privacy not in ("no_gps", "private")
# ── detail JSON ──────────────────────────────────────────────────────────
detail: dict = {
"bas_version": "1.0",
"id": activity_id,
"title": activity.title or _auto_title(activity),
"description": activity.description,
"sport": activity.sport,
"sub_sport": activity.sub_sport,
"started_at": activity.started_at.isoformat(),
"distance_m": metrics.distance_m,
"duration_s": metrics.duration_s,
"moving_time_s": metrics.moving_time_s,
"elevation_gain_m": metrics.elevation_gain_m,
"elevation_loss_m": metrics.elevation_loss_m,
"avg_speed_kmh": metrics.avg_speed_kmh,
"max_speed_kmh": metrics.max_speed_kmh,
"avg_hr_bpm": metrics.avg_hr_bpm,
"max_hr_bpm": metrics.max_hr_bpm,
"avg_cadence_rpm": metrics.avg_cadence_rpm,
"avg_power_w": metrics.avg_power_w,
"max_power_w": metrics.max_power_w,
"gear": activity.gear,
"device": activity.device,
"bbox": list(metrics.bbox) if metrics.bbox else None,
"start_latlng": list(metrics.start_latlng) if metrics.start_latlng else None,
"end_latlng": list(metrics.end_latlng) if metrics.end_latlng else None,
"laps": [_serialise_lap(lap) for lap in activity.laps],
"timeseries": build_timeseries(activity.points, activity.started_at, privacy),
"source": source,
"source_file": activity.source_file,
"source_hash": activity.source_hash,
"strava_id": activity.strava_id,
"duplicate_of": duplicate_of,
"privacy": privacy,
"custom": {},
}
json_path = acts_dir / f"{activity_id}.json"
json_path.write_text(json.dumps(detail, indent=2, ensure_ascii=False))
# ── GeoJSON track ────────────────────────────────────────────────────────
if has_gps:
geojson = build_geojson(activity.points, activity_id, epsilon=rdp_epsilon)
geojson_path = acts_dir / f"{activity_id}.geojson"
geojson_path.write_text(json.dumps(geojson, indent=2, ensure_ascii=False))
return activity_id
def build_summary(
activity: ParsedActivity,
metrics: ComputedMetrics,
activity_id: str,
privacy: str = "public",
) -> dict:
"""Build the Activity Summary object for index.json."""
has_gps = metrics.bbox is not None and privacy not in ("no_gps", "private")
return {
"id": activity_id,
"title": activity.title or _auto_title(activity),
"sport": activity.sport,
"sub_sport": activity.sub_sport,
"started_at": activity.started_at.isoformat(),
"distance_m": metrics.distance_m,
"duration_s": metrics.duration_s,
"moving_time_s": metrics.moving_time_s,
"elevation_gain_m": metrics.elevation_gain_m,
"avg_speed_kmh": metrics.avg_speed_kmh,
"max_speed_kmh": metrics.max_speed_kmh,
"avg_hr_bpm": metrics.avg_hr_bpm,
"max_hr_bpm": metrics.max_hr_bpm,
"avg_cadence_rpm": metrics.avg_cadence_rpm,
"avg_power_w": metrics.avg_power_w,
"source": _infer_source(activity),
"privacy": privacy,
"detail_url": f"activities/{activity_id}.json",
"track_url": f"activities/{activity_id}.geojson" if has_gps else None,
}
def write_index(summaries: list[dict], output_dir: Path, owner: dict) -> None:
"""Write index.json (sorted newest first)."""
sorted_summaries = sorted(
summaries,
key=lambda s: s["started_at"],
reverse=True,
)
index = {
"bas_version": "1.0",
"owner": owner,
"generated_at": _now_iso(),
"shards": [],
"activities": sorted_summaries,
}
(output_dir / "index.json").write_text(
json.dumps(index, indent=2, ensure_ascii=False)
)
# ── helpers ──────────────────────────────────────────────────────────────────
def _now_iso() -> str:
from datetime import datetime, timezone
return datetime.now(timezone.utc).isoformat()
def _auto_title(activity: ParsedActivity) -> str:
ts = activity.started_at
hour = ts.hour
if 5 <= hour < 12:
time_of_day = "Morning"
elif 12 <= hour < 17:
time_of_day = "Afternoon"
elif 17 <= hour < 21:
time_of_day = "Evening"
else:
time_of_day = "Night"
sport = activity.sport.capitalize()
return f"{time_of_day} {sport}"
def _infer_source(activity: ParsedActivity) -> str | None:
if activity.strava_id:
return "strava_export"
name = activity.source_file.lower()
# Karoo uses UUID-style names
if "activity" in name and len(name.split(".")) >= 3:
return "karoo"
if name.endswith(".fit") or name.endswith(".fit.gz"):
return "fit_file"
if name.endswith(".gpx") or name.endswith(".gpx.gz"):
return "gpx_file"
if name.endswith(".tcx") or name.endswith(".tcx.gz"):
return "tcx_file"
return None
def _slugify(text: str) -> str:
text = unicodedata.normalize("NFKD", text)
text = text.encode("ascii", "ignore").decode("ascii")
text = text.lower()
text = re.sub(r"[^a-z0-9]+", "-", text)
return text.strip("-")[:60]
def _serialise_lap(lap: LapData) -> dict:
return {
"index": lap.index,
"started_at": lap.started_at.isoformat(),
"duration_s": lap.duration_s,
"distance_m": lap.distance_m,
"elevation_gain_m": lap.elevation_gain_m,
"avg_speed_kmh": lap.avg_speed_kmh,
"avg_hr_bpm": lap.avg_hr_bpm,
"avg_power_w": lap.avg_power_w,
}