GeoData/scripts/verify_pipeline.py

#!/usr/bin/env python3
"""Lightweight checks for the CityGML → CityJSON → GLB pipeline."""
from __future__ import annotations

import argparse
import sys
from pathlib import Path
from typing import Iterable, Sequence

DEFAULT_RAW_DIR = Path("raw/citygml/lod2")
DEFAULT_CITYJSON_DIR = Path("work/cityjson")
DEFAULT_CITYJSON_TRI_DIR = Path("work/cityjson_tri")
DEFAULT_CITYJSON_SPLIT_DIR = Path("work/cityjson_split")
DEFAULT_GLB_DIR = Path("export_unity/buildings_glb")
DEFAULT_GLB_SPLIT_DIR = Path("export_unity/buildings_glb_split")
DEFAULT_TILE_INDEX = Path("export_unity/tile_index.csv")


def parse_args(argv: Iterable[str] | None = None) -> argparse.Namespace:
    parser = argparse.ArgumentParser(
        description="Validate presence of CityGML/CityJSON/GLB artifacts for the CityGML pipeline."
    )
    parser.add_argument(
        "--mode",
        choices=["cityjson", "glb", "both"],
        default="both",
        help="Which parts of the pipeline to check.",
    )
    parser.add_argument(
        "--tiles",
        nargs="*",
        help="Optional list of tile IDs to check (e.g., LoD2_32_328_5511). Defaults to all GML tiles.",
    )
    parser.add_argument(
        "--raw-dir",
        type=Path,
        default=DEFAULT_RAW_DIR,
        help="Directory containing LoD2 CityGML tiles.",
    )
    parser.add_argument(
        "--cityjson-dir",
        type=Path,
        default=DEFAULT_CITYJSON_DIR,
        help="Directory containing CityJSON exports.",
    )
    parser.add_argument(
        "--tri-dir",
        type=Path,
        default=DEFAULT_CITYJSON_TRI_DIR,
        help="Directory containing cleaned + triangulated CityJSON files.",
    )
    parser.add_argument(
        "--split-dir",
        type=Path,
        default=DEFAULT_CITYJSON_SPLIT_DIR,
        help="Directory containing split CityJSON files (roof/wall).",
    )
    parser.add_argument(
        "--glb-dir",
        type=Path,
        default=DEFAULT_GLB_DIR,
        help="Directory containing base GLB exports.",
    )
    parser.add_argument(
        "--glb-split-dir",
        type=Path,
        default=DEFAULT_GLB_SPLIT_DIR,
        help="Directory containing split GLB exports.",
    )
    parser.add_argument(
        "--tile-index",
        type=Path,
        default=DEFAULT_TILE_INDEX,
        help="Path to the tile_index.csv placement manifest.",
    )
    return parser.parse_args(argv)


def discover_tiles(raw_dir: Path, provided: Sequence[str] | None) -> list[str]:
    if provided:
        return sorted(set(provided))

    if not raw_dir.exists():
        return []

    return sorted(path.stem for path in raw_dir.glob("LoD2_*.gml"))


def path_exists(path: Path) -> bool:
    if path.exists():
        return True
    if path.is_dir():
        return True
    # citygml-tools may output a directory named *.city.json containing a *.json
    if path.suffixes == [".city", ".json"]:
        candidate = path / f"{path.stem}.json"
        if candidate.exists():
            return True
    return False


def collect_missing(tile_ids: Sequence[str], directory: Path, pattern: str) -> list[Path]:
    expected = [directory / pattern.format(tile_id=tile_id) for tile_id in tile_ids]
    return [path for path in expected if not path_exists(path)]


def summarize_missing(label: str, missing: list[Path]) -> str:
    sample = ", ".join(str(path) for path in missing[:5])
    return f"{label}: missing {len(missing)} file(s){' (e.g., ' + sample + ')' if sample else ''}"


def verify_cityjson(tile_ids: Sequence[str], args: argparse.Namespace) -> list[str]:
    issues: list[str] = []
    missing_gml = collect_missing(tile_ids, args.raw_dir, "{tile_id}.gml")
    if missing_gml:
        issues.append(summarize_missing("raw CityGML tiles", missing_gml))

    missing_cityjson = collect_missing(tile_ids, args.cityjson_dir, "{tile_id}.city.json")
    if missing_cityjson:
        issues.append(summarize_missing("CityJSON exports", missing_cityjson))

    missing_tri = collect_missing(tile_ids, args.tri_dir, "{tile_id}.tri.city.json")
    if missing_tri:
        issues.append(summarize_missing("clean + triangulated CityJSON exports", missing_tri))

    missing_roof = collect_missing(tile_ids, args.split_dir, "{tile_id}.roof.city.json")
    if missing_roof:
        issues.append(summarize_missing("split roof CityJSON exports", missing_roof))

    missing_wall = collect_missing(tile_ids, args.split_dir, "{tile_id}.wall.city.json")
    if missing_wall:
        issues.append(summarize_missing("split wall CityJSON exports", missing_wall))

    return issues


def verify_glb(tile_ids: Sequence[str], args: argparse.Namespace) -> list[str]:
    issues: list[str] = []
    missing_base = collect_missing(tile_ids, args.glb_dir, "{tile_id}.glb")
    if missing_base:
        issues.append(summarize_missing("base GLB exports", missing_base))

    missing_roof = collect_missing(tile_ids, args.glb_split_dir, "{tile_id}_roof.glb")
    if missing_roof:
        issues.append(summarize_missing("split roof GLB exports", missing_roof))

    missing_wall = collect_missing(tile_ids, args.glb_split_dir, "{tile_id}_wall.glb")
    if missing_wall:
        issues.append(summarize_missing("split wall GLB exports", missing_wall))

    return issues


def main(argv: Iterable[str] | None = None) -> int:
    args = parse_args(argv)
    tile_ids = discover_tiles(args.raw_dir, args.tiles)
    if not tile_ids:
        print("No tile IDs found. Add LoD2 GML files or pass --tiles.", file=sys.stderr)
        return 1

    issues: list[str] = []
    if args.tile_index and not args.tile_index.exists():
        issues.append(f"tile_index.csv missing: {args.tile_index}")

    if args.mode in ("cityjson", "both"):
        issues.extend(verify_cityjson(tile_ids, args))
    if args.mode in ("glb", "both"):
        issues.extend(verify_glb(tile_ids, args))

    if issues:
        print("Pipeline verification failed:")
        for issue in issues:
            print(f"- {issue}")
        return 1

    print(f"Pipeline verification passed for {len(tile_ids)} tile(s) in mode='{args.mode}'.")
    return 0


if __name__ == "__main__":
    sys.exit(main())