From a8f954805ecb9f27f0e813ecec86d46a2c3c0251 Mon Sep 17 00:00:00 2001
From: "s0wlz (Matthias Puchstein)" <matthias@puchstein.lu>
Date: Thu, 22 Jan 2026 00:59:37 +0100
Subject: [PATCH] Extend lpolpg split workflow

Changelog:

- add --split-lpolpg-delete-source to remove source files after split

- auto-split after downloads when lpolpg is enabled in config
---
 geodata_pipeline/lpolpg_split.py |  7 +++++++
 geodata_to_unity.py              | 35 ++++++++++++++++++++++++++++++++
 2 files changed, 42 insertions(+)

diff --git a/geodata_pipeline/lpolpg_split.py b/geodata_pipeline/lpolpg_split.py
index a25c33e..81044c3 100644
--- a/geodata_pipeline/lpolpg_split.py
+++ b/geodata_pipeline/lpolpg_split.py
@@ -43,6 +43,7 @@ def split_lpolpg(
     formats: Iterable[str] = ("laz", "xyz"),
     ground_classes: Iterable[int] = (2,),
     overwrite: bool = False,
+    delete_source: bool = False,
 ) -> int:
     try:
         import laspy
@@ -139,5 +140,11 @@ def split_lpolpg(
             f"[lpolpg] {name}: ground={stats.ground_points}, "
             f"objects={stats.object_points}, total={stats.total}"
         )
+        if delete_source:
+            try:
+                os.remove(in_path)
+                print(f"[lpolpg] removed source {name}")
+            except OSError as exc:
+                print(f"[lpolpg] warn: could not remove {name}: {exc}")
 
     return 0
diff --git a/geodata_to_unity.py b/geodata_to_unity.py
index 9956b72..68fffba 100644
--- a/geodata_to_unity.py
+++ b/geodata_to_unity.py
@@ -7,6 +7,11 @@ import os
 import sys
 from typing import Iterable
 
+try:
+    import tomllib
+except ImportError:  # pragma: no cover - tomllib is required
+    tomllib = None
+
 from geodata_download import run_download
 
 from geodata_pipeline.config import Config, DEFAULT_CONFIG_PATH, ensure_default_config
@@ -113,6 +118,11 @@ def parse_args(argv: Iterable[str] | None = None) -> argparse.Namespace:
         action="store_true",
         help="Overwrite existing LPG/LPO outputs when splitting lpolpg.",
     )
+    parser.add_argument(
+        "--split-lpolpg-delete-source",
+        action="store_true",
+        help="Delete lpolpg source files after a successful split.",
+    )
     return parser.parse_args(argv)
 
 
@@ -129,6 +139,27 @@ def load_config(args: argparse.Namespace) -> Config:
     return cfg.with_overrides(raw_dgm1_path=args.raw_dgm1_path, raw_dop20_path=args.raw_dop20_path)
 
 
+def _download_requests_lpolpg(download_config: str, requested: list[str] | None) -> bool:
+    lpolpg_keys = {"lpolpg", "lpg", "lpo"}
+    if requested:
+        return any(name in lpolpg_keys for name in requested)
+    if tomllib is None:
+        return False
+    try:
+        with open(download_config, "rb") as fh:
+            cfg = tomllib.load(fh)
+    except OSError:
+        return False
+    datasets = cfg.get("datasets", {})
+    if not isinstance(datasets, dict):
+        return False
+    for key in lpolpg_keys:
+        dataset_cfg = datasets.get(key)
+        if isinstance(dataset_cfg, dict) and dataset_cfg.get("enabled", True):
+            return True
+    return False
+
+
 def main(argv: Iterable[str] | None = None) -> int:
     args = parse_args(argv)
     cfg = load_config(args)
@@ -167,6 +198,9 @@ def main(argv: Iterable[str] | None = None) -> int:
         )
         if download_exit != 0:
             return download_exit
+        if not args.split_lpolpg and _download_requests_lpolpg(args.download_config, datasets):
+            print("[download] lpolpg detected; splitting into lpg/lpo.")
+            args.split_lpolpg = True
     if args.split_lpolpg:
         formats = [fmt.strip() for fmt in args.split_lpolpg_formats.split(",") if fmt.strip()]
         ground = [int(val) for val in args.split_lpolpg_ground_classes.split(",") if val.strip()]
@@ -175,6 +209,7 @@ def main(argv: Iterable[str] | None = None) -> int:
             formats=formats,
             ground_classes=ground,
             overwrite=args.split_lpolpg_overwrite,
+            delete_source=args.split_lpolpg_delete_source,
         )
         if split_exit != 0:
             return split_exit