code_public/lidar_rendu
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Garder un doublon de voisine si sa taille diffère de la copie d'input/

Browse Source 
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
Antoine Jacquin
2026-09-27 16:31:58 +02:00
parent 32d694bbd6
commit c9b7504346
2 changed files with 25 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files

17
lidar_pipeline/pipeline.py
View File

@ -533,7 +533,9 @@ class LidarArchaeoPipeline:
voisines) ; l'exemplaire d'input/ fait foi (_neighbor_laz_files voisines) ; l'exemplaire d'input/ fait foi (_neighbor_laz_files
cherche dans input/ avant edge_neighbors/), le doublon ne sert donc cherche dans input/ avant edge_neighbors/), le doublon ne sert donc
à rien et peut peser plusieurs centaines de Mo. Les fichiers ".part" à rien et peut peser plusieurs centaines de Mo. Les fichiers ".part"
(téléchargement en cours) ne sont jamais touchés. (téléchargement en cours) ne sont jamais touchés, et un doublon de
taille différente est conservé (l'exemplaire d'input/ pourrait être
tronqué : on ne supprime jamais ce qui serait la seule copie saine).
""" """
from .dtm import EDGE_NEIGHBORS_DIRNAME from .dtm import EDGE_NEIGHBORS_DIRNAME
if not edge_dir.is_dir(): if not edge_dir.is_dir():
@ -541,13 +543,20 @@ class LidarArchaeoPipeline:
removed = 0 removed = 0
freed = 0 freed = 0
for path in edge_dir.iterdir(): for path in edge_dir.iterdir():
if not path.is_file() or path.suffix == ".part" or path.name.endswith(".part"): if not path.is_file() or path.suffix == ".part":
continue continue
twin = self.input_dir / path.name twin = self.input_dir / path.name
if twin.exists(): if not twin.exists():
continue
try: try:
freed += path.stat().st_size size = path.stat().st_size
if twin.stat().st_size != size:
logger.warning(f"Raccord des bords : {path.name} présent dans input/ "
f"et {EDGE_NEIGHBORS_DIRNAME}/ avec des tailles "
f"différentes — doublon conservé")
continue
path.unlink() path.unlink()
freed += size
removed += 1 removed += 1
except OSError: except OSError:
pass pass

8
lidar_pipeline/tests/test_pipeline.py
View File

@ -121,9 +121,14 @@ class TestCleanupEdgeNeighborDuplicates:
dup_name = "LHD_FXX_1000_6779_PTS_LAMB93_IGN69.copc.laz" dup_name = "LHD_FXX_1000_6779_PTS_LAMB93_IGN69.copc.laz"
unique_name = "LHD_FXX_1001_6779_PTS_LAMB93_IGN69.copc.laz" unique_name = "LHD_FXX_1001_6779_PTS_LAMB93_IGN69.copc.laz"
part_name = "LHD_FXX_1002_6779_PTS_LAMB93_IGN69.copc.laz.part" part_name = "LHD_FXX_1002_6779_PTS_LAMB93_IGN69.copc.laz.part"
mismatch_name = "LHD_FXX_1003_6779_PTS_LAMB93_IGN69.copc.laz"
(input_dir / dup_name).write_bytes(b"authoritative") (input_dir / dup_name).write_bytes(b"authoritative")
(edge_dir / dup_name).write_bytes(b"duplicata") (edge_dir / dup_name).write_bytes(b"authoritative")
# input/ tronqué (taille différente) : la voisine peut être la seule
# copie saine, elle doit rester.
(input_dir / mismatch_name).write_bytes(b"")
(edge_dir / mismatch_name).write_bytes(b"complete")
(edge_dir / unique_name).write_bytes(b"voisine-unique") (edge_dir / unique_name).write_bytes(b"voisine-unique")
(edge_dir / part_name).write_bytes(b"en-cours") (edge_dir / part_name).write_bytes(b"en-cours")
@ -135,6 +140,7 @@ class TestCleanupEdgeNeighborDuplicates:
assert (input_dir / dup_name).read_bytes() == b"authoritative" assert (input_dir / dup_name).read_bytes() == b"authoritative"
assert (edge_dir / unique_name).exists() assert (edge_dir / unique_name).exists()
assert (edge_dir / part_name).exists() assert (edge_dir / part_name).exists()
assert (edge_dir / mismatch_name).read_bytes() == b"complete"
def test_noop_when_edge_dir_missing(self, tmp_path): def test_noop_when_edge_dir_missing(self, tmp_path):
from lidar_pipeline.pipeline import LidarArchaeoPipeline from lidar_pipeline.pipeline import LidarArchaeoPipeline