Fix bugs and improve pipeline flexibility

- Fix gpu_cleanup import missing in visualizations.py (NameError in workers)
- Fix t_pdf referenced before assignment when PDF is skipped
- Skip classification+DTM when DTM exists regardless of --force
- --force now only regenerates WebP/PDF, not classification/DTM
- --force-classification forces reclassification when needed
- Add laspy repair fallback for corrupt LAZ files (EVLR errors)
- Keep DTM TIF by default for reuse (--no-keep-tif to delete)
- Increase space between image and bottom cartouche (0.12→0.19)

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

CLAUDE.md

@@ -17,7 +17,8 @@ All commands run inside Docker. Use `./run.sh` as the primary interface.
 ./run.sh --test                                 # Run unit tests
 ./run.sh -g --file LHD_FXX_1000_6882_PTS_LAMB93_IGN69.copc  # Single file
 ./run.sh --ground-classification pmf            # Force PMF ground classification
-./run.sh -g --keep-tif                          # Keep intermediate TIFF files
+./run.sh -g --keep-tif                          # Keep intermediate TIFF files (default: kept)
+./run.sh -g --no-keep-tif                       # Delete intermediate TIFF files
 ./run.sh                                        # Print help (no args)
 ```
 
@@ -82,6 +83,6 @@ Uses `ProcessPoolExecutor` with `'spawn'` start method (required for CUDA). Each
 - **Language**: UI messages and comments in French. Code identifiers in English.
 - **Logging**: Use `logger = logging.getLogger("lidar")`. Prefix per-file logs via `_file_filter.basename`.
 - **GPU pattern**: `arr_gpu = to_gpu(arr)` → compute → `result = to_cpu(arr_gpu)` → `gpu_cleanup()` between visualizations.
-- **Output format**: Visualizations saved as WebP. TIFF intermediates deleted after conversion unless `--keep-tif`. No COGs or viewer — only WebP + PDF report remain.
+- **Output format**: Visualizations saved as WebP. DTM TIFF kept by default for reuse (use `--no-keep-tif` to delete). `--force` regenerates WebPs without re-classifying if DTM exists. No COGs or viewer — only WebP + PDF report remain.
 - **Compression**: TIF intermediates use `deflate` compression (faster than LZW for float32 data).
 - **Tests**: Run only inside Docker via `./run.sh --test`. Synthetic DEM fixture in `tests/conftest.py`.

lidar_pipeline/cli.py

@@ -118,9 +118,9 @@ Exemples:
         help="Reclassifier le sol même si le fichier .las existe déjà"
     )
     parser.add_argument(
-        "--keep-tif",
+        "--no-keep-tif",
         action="store_true",
-        help="Conserver les fichiers TIFF intermédiaires (sinon supprimés après conversion WebP)"
+        help="Supprimer les fichiers TIFF intermédiaires après conversion WebP (par défaut: conservés)"
     )
     parser.add_argument(
         "--ground-classification",
@@ -176,7 +176,7 @@ Exemples:
             force=args.force,
             ground_method=args.ground_classification,
             force_classify=args.force_classification,
-            keep_tif=args.keep_tif,
+            keep_tif=not args.no_keep_tif,
         )
 
         # If --file is specified, process only matching files

lidar_pipeline/dtm.py

@@ -246,10 +246,57 @@ def classify_ground(laz_file, temp_dir, method='auto', force=False):
         logger.info(f"  ✓ Classification sol {method.upper()} terminée")
         return output_las
     except subprocess.CalledProcessError as e:
-        logger.error(f"  ✗ Erreur classification PDAL ({method.upper()}): {e.stderr.decode()}")
+        error_msg = e.stderr.decode() if e.stderr else str(e)
+        logger.warning(f"  ✗ Erreur classification PDAL ({method.upper()}): {error_msg}")
+
+        # Try repairing file with laspy if PDAL fails on EVLR/VLR
+        if 'VLR' in error_msg or 'Invalid' in error_msg:
+            logger.info(f"  → Tentative de réparation du fichier avec laspy...")
+            repaired_las = temp_dir / f"{laz_base}_repaired.las"
+            if _repair_laz_with_laspy(laz_file, repaired_las):
+                # Retry PDAL pipeline with repaired file
+                pipeline_json = _create_ground_pipeline(repaired_las, output_las, method)
+                with open(pipeline_file, 'w') as f:
+                    f.write(pipeline_json)
+                try:
+                    subprocess.run(
+                        ["pdal", "pipeline", str(pipeline_file)],
+                        capture_output=True, check=True
+                    )
+                    logger.info(f"  ✓ Classification sol {method.upper()} terminée (fichier réparé)")
+                    return output_las
+                except subprocess.CalledProcessError as e2:
+                    error_msg2 = e2.stderr.decode() if e2.stderr else str(e2)
+                    logger.error(f"  ✗ Échec classification même après réparation: {error_msg2}")
+            else:
+                logger.error(f"  ✗ Impossible de réparer le fichier")
         return None
 
 
+def _repair_laz_with_laspy(input_laz, output_las):
+    """Try to repair a corrupt LAZ file by re-reading with laspy and saving as LAS.
+
+    Works around PDAL errors like 'Invalid Extended VLR size' by stripping
+    problematic VLR/EVLR metadata during re-save.
+
+    Args:
+        input_laz: Path to corrupt LAZ/LAS file.
+        output_las: Path for repaired LAS output.
+
+    Returns:
+        True if repair succeeded, False otherwise.
+    """
+    import laspy
+    try:
+        las = laspy.read(str(input_laz))
+        las.write(str(output_las))
+        logger.info(f"  ✓ Fichier réparé via laspy ({len(las.points):,} points)")
+        return True
+    except Exception as e:
+        logger.warning(f"  ✗ Réparation laspy échouée: {e}")
+        return False
+
+
 def create_dtm_fast(las_file, basename, dtm_dir, resolution, force=False):
     """Create DTM using fast binning method with gap filling.
 

lidar_pipeline/pipeline.py

@@ -275,9 +275,11 @@ class LidarArchaeoPipeline:
         logger.info(f"FICHIER : {basename}")
         logger.info("=" * 60)
 
-        # Skip ground classification + DTM if DTM already exists (unless --force)
+        # Skip ground classification + DTM if DTM already exists
+        # --force only affects visualizations/PDF, not classification/DTM
+        # Use --force-classification to force reclassification
         dtm_path = self.dtm_dir / f"{basename}_dtm.tif"
-        if dtm_path.exists() and not self.force:
+        if dtm_path.exists():
             logger.info("[1/5] Classification du sol — sautée (DTM existant)")
             logger.info("[2/5] Génération DTM — sautée (DTM existant)")
             dtm_file = dtm_path
@@ -297,7 +299,7 @@ class LidarArchaeoPipeline:
             # Step 2: Generate DTM
             logger.info("[2/5] Génération DTM...")
             t2 = time.time()
-            dtm_file = create_dtm_fast(las_file, basename, self.dtm_dir, self.resolution, force=self.force)
+            dtm_file = create_dtm_fast(las_file, basename, self.dtm_dir, self.resolution)
             t_dtm = time.time() - t2
             if not dtm_file:
                 logger.error(f"  ✗ Échec DTM ({t_dtm:.1f}s)")
@@ -309,6 +311,7 @@ class LidarArchaeoPipeline:
         self.generate_all_visualizations(dtm_file, basename)
 
         # Step 4: PDF report
+        t_pdf = 0
         file_vis_dir = self.vis_dir / basename
         pdf_file = self.pdf_dir / f"{basename}_rapport.pdf"
         if pdf_file.exists() and not self.force:
@@ -320,10 +323,8 @@ class LidarArchaeoPipeline:
             t_pdf = time.time() - t4
             logger.info(f"  ✓ Rapport PDF terminé ({t_pdf:.1f}s)")
 
-        # Step 5: Clean up DTM TIF unless --keep-tif
+        # Step 5: Keep DTM TIF for reuse (regenerating WebPs skips classification)
-        if not self.keep_tif:
+        # Use --no-keep-tif to delete DTM files after processing
-            for dtm_file in sorted(self.dtm_dir.glob(f"{basename}_dtm.tif")):
-                dtm_file.unlink(missing_ok=True)
 
         t_total = time.time() - t_start
         logger.info(f"✓ {basename} terminé en {t_total:.1f}s")

lidar_pipeline/rendering.py

@@ -387,9 +387,9 @@ def tif_to_png(tif_file, vis_dir, resolution, keep_tif=False, source_info=None):
             # Fixed data area position — identical for ALL visualization types
             # This ensures overlay/superposition works across all WebP images
             data_left = 0.08
-            data_bottom = 0.12
+            data_bottom = 0.19
             data_width_frac = 0.74
-            data_height_frac = 0.78
+            data_height_frac = 0.71
 
             ax = fig.add_axes([data_left, data_bottom, data_width_frac, data_height_frac])
             if is_rgba or is_rgb:

lidar_pipeline/visualizations.py

@@ -18,7 +18,7 @@ import rasterio
 from scipy.ndimage import generic_filter
 from scipy.stats import binned_statistic_2d
 
-from .gpu import HAS_GPU, to_gpu, to_cpu, xp_gaussian_filter, xp_uniform_filter, xp_minimum_filter
+from .gpu import HAS_GPU, to_gpu, to_cpu, xp_gaussian_filter, xp_uniform_filter, xp_minimum_filter, gpu_cleanup
 
 logger = logging.getLogger("lidar")
 

run.sh

@@ -9,7 +9,7 @@
 #   -v              Mode verbeux (timestamps + niveaux)
 #   --debug         Mode debug (détails internes fichier:ligne)
 #   -f / --force    Régénérer tous les fichiers même si existants
-#   --keep-tif       Conserver les fichiers TIFF intermédiaires
+#   --no-keep-tif    Supprimer les TIFF intermédiaires (par défaut: conservés)
 #   --force-classification
 #                   Reclassifier le sol même si le fichier .las existe déjà
 #   --ground-classification {auto,smrf,pmf,csf}
@@ -40,7 +40,7 @@ if [ $# -eq 0 ]; then
     echo "  -f / --force    Régénérer tous les fichiers même si les WebP existent"
     echo "  --force-classification"
     echo "                  Reclassifier le sol même si le fichier .las existe"
-    echo "  --keep-tif      Conserver les fichiers TIFF intermédiaires"
+    echo "  --no-keep-tif   Supprimer les TIFF intermédiaires (défaut: conservés)"
     echo "  --ground-classification {auto,smrf,pmf,csf}"
     echo "                  Méthode de classification du sol (défaut: auto)"
     echo "  --file NOM...   Traiter un ou plusieurs fichiers LAZ (nom complet sans .laz/.las)"
@@ -52,7 +52,7 @@ if [ $# -eq 0 ]; then
     echo "  $0 -g -w 4                               # GPU + 4 workers"
     echo "  $0 -g -v                                 # GPU + verbeux"
     echo "  $0 -g -r 0.2                             # Haute résolution"
-    echo "  $0 -g --force                            # Tout régénérer (WebP + classification)"
+    echo "  $0 -g --force                            # Régénérer les WebP"
     echo "  $0 -g --force-classification             # Reclassifier le sol seulement"
     echo "  $0 -g --ground-classification pmf        # Forcer PMF"
     echo "  $0 -g --file LHD_...IGN69.copc           # Un fichier"
@@ -67,7 +67,7 @@ FORCE_FLAG=""
 FILE_ARGS=""
 GROUND_METHOD=""
 FORCE_CLASSIFY_FLAG=""
-KEEP_TIF_FLAG=""
+NO_KEEP_TIF_FLAG=""
 
 # Parse arguments manually (more robust than getopts for mixed short/long options)
 while [ $# -gt 0 ]; do
@@ -80,7 +80,7 @@ while [ $# -gt 0 ]; do
         --debug) VERBOSE_FLAG="--debug"; shift ;;
         --force) FORCE_FLAG="--force"; shift ;;
         --force-classification) FORCE_CLASSIFY_FLAG="--force-classification"; shift ;;
-        --keep-tif) KEEP_TIF_FLAG="--keep-tif"; shift ;;
+        --no-keep-tif) NO_KEEP_TIF_FLAG="--no-keep-tif"; shift ;;
         --ground-classification) GROUND_METHOD="$2"; shift 2 ;;
         --ground-classification=*) GROUND_METHOD="${1#--ground-classification=}"; shift ;;
         --file) shift; while [ $# -gt 0 ] && [[ ! "$1" =~ ^- ]]; do FILE_ARGS="$FILE_ARGS $1"; shift; done ;;
@@ -99,7 +99,7 @@ while [ $# -gt 0 ]; do
             echo "  -f / --force    Régénérer tous les fichiers même si les WebP existent"
             echo "  --force-classification"
             echo "                  Reclassifier le sol même si le fichier .las existe"
-            echo "  --keep-tif      Conserver les fichiers TIFF intermédiaires"
+            echo "  --no-keep-tif   Supprimer les TIFF intermédiaires (défaut: conservés)"
             echo "  --ground-classification {auto,smrf,pmf,csf}"
             echo "                  Méthode de classification du sol (défaut: auto)"
             echo "  --file NOM...   Traiter un ou plusieurs fichiers LAZ (nom complet sans .laz/.las)"
@@ -111,7 +111,7 @@ while [ $# -gt 0 ]; do
             echo "  $0 -g -w 4                               # GPU + 4 workers"
             echo "  $0 -g -v                                 # GPU + verbeux"
             echo "  $0 -g -r 0.2                             # Haute résolution"
-            echo "  $0 -g --force                            # Tout régénérer (WebP + classification)"
+            echo "  $0 -g --force                            # Régénérer les WebP"
             echo "  $0 -g --force-classification             # Reclassifier le sol seulement"
             echo "  $0 -g --ground-classification pmf        # Forcer PMF"
             echo "  $0 -g --file LHD_...IGN69.copc           # Un fichier"
@@ -159,14 +159,14 @@ echo "  GPU        : $([ -n "$GPU_FLAG" ] && echo 'OUI' || echo 'non')"
 echo "  Verbeux    : $([ -n "$VERBOSE_FLAG" ] && echo 'OUI' || echo 'non')"
 echo "  Force      : $([ -n "$FORCE_FLAG" ] && echo 'OUI' || echo 'non')"
 echo "  Force classif.: $([ -n "$FORCE_CLASSIFY_FLAG" ] && echo 'OUI' || echo 'non')"
-echo "  Keep TIFF    : $([ -n "$KEEP_TIF_FLAG" ] && echo 'OUI' || echo 'non')"
+echo "  Keep TIFF    : $([ -n "$NO_KEEP_TIF_FLAG" ] && echo 'non' || echo 'OUI')"
 echo "  Classification sol : $([ -n "$GROUND_METHOD" ] && echo "$GROUND_METHOD" || echo 'auto')"
 if [ -n "$FILE_ARGS" ]; then
     echo "  Fichiers   :${FILE_ARGS}"
 fi
 echo "============================================"
 
-CMD_ARGS="-o /data/output -r $RESOLUTION -w $WORKERS $VERBOSE_FLAG $FORCE_FLAG $FORCE_CLASSIFY_FLAG $KEEP_TIF_FLAG"
+CMD_ARGS="-o /data/output -r $RESOLUTION -w $WORKERS $VERBOSE_FLAG $FORCE_FLAG $FORCE_CLASSIFY_FLAG $NO_KEEP_TIF_FLAG"
 if [ -n "$GROUND_METHOD" ]; then
     CMD_ARGS="$CMD_ARGS --ground-classification $GROUND_METHOD"
 fi