Output schema

The primary artifact is a GeoJSON FeatureCollection of LineString features in EPSG:4326 (coordinates are [longitude, latitude]). A machine-readable JSON Schema is committed at docs/schema/coverage.schema.json.

Feature properties

Property	Type	Always?	Meaning
source	"google" | "mapillary" | "kartaview"	✅	provenance — which source produced this line
length_m	number	✅	great-circle length in metres
n_points	integer ≥ 2	✅	vertex count
id	string	—	segment id (e.g. google-seg-42, or the Mapillary sequence id)
capture_date	string	—	YYYY-MM (Google, --precision) or ISO date
pano_ids	string[]	—	the official pano ids composing a Google line (provenance)
is_pano	boolean	—	Mapillary 360 flag
sequence_id	string	—	Mapillary/KartaView sequence id

Every geometry is a LineString with ≥ 2 points; degenerate features are dropped on export.

Example

{
  "type": "FeatureCollection",
  "features": [
    {
      "type": "Feature",
      "geometry": {
        "type": "LineString",
        "coordinates": [[26.050403, 44.434487], [26.050512, 44.434971]]
      },
      "properties": {
        "source": "google",
        "length_m": 390.83,
        "n_points": 39,
        "id": "google-seg-42",
        "capture_date": "2024-07"
      }
    }
  ]
}

Validate downstream

import json, jsonschema  # pip install jsonschema
schema = json.load(open("docs/schema/coverage.schema.json"))
data = json.load(open("bucharest.geojson"))
jsonschema.validate(data, schema)   # raises if malformed

# The HARD RULE is also checkable on the output: no Google feature should carry a photosphere id.
from tracelines.models import is_official_panoid
for f in data["features"]:
    if f["properties"]["source"] == "google":
        for pid in f["properties"].get("pano_ids", []):
            assert is_official_panoid(pid), f"non-official id in output: {pid}"

The source property is your provenance record — keep it, because it determines what you may do with each line (see Data provenance).