Audit: corrections de bugs identifiés

- rendering.py: colorbar cassée quand NaN mask actif — créer un
  ScalarMappable avec le cmap sauvegardé au lieu de rely sur
  l'image RGBA qui n'a plus de cmap
- rendering.py: nettoyage du PNG temporaire avec try/finally et
  missing_ok=True pour éviter les fichiers orphelins
- gpu.py: to_gpu() convertit en float32 au lieu de float64 pour
  réduire la consommation mémoire GPU
- dtm.py: utiliser _file_basename() de pipeline.py au lieu de
  dupliquer la logique d'extraction du basename
- pipeline.py: docstring corrigé (18 visualisations, pas 19)
- cli.py: --file supporte aussi les noms sans .copc
  (recherche .copc.laz et .copc.las en plus de .laz et .las)

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

lidar_pipeline/cli.py

@@ -171,9 +171,13 @@ Exemples:
             from pathlib import Path
             input_dir = Path(args.input)
             # Each pattern is the full filename without extension (e.g. LHD_FXX_1000_6882_PTS_LAMB93_IGN69.copc)
+            # Also supports bare name without .copc (e.g. LHD_FXX_1000_6882_PTS_LAMB93_IGN69)
             selected_files = []
             for pattern in args.file:
-                matches = list(input_dir.glob(f"{pattern}.laz")) + list(input_dir.glob(f"{pattern}.las"))
+                matches = (list(input_dir.glob(f"{pattern}.laz"))
+                          + list(input_dir.glob(f"{pattern}.las"))
+                          + list(input_dir.glob(f"{pattern}.copc.laz"))
+                          + list(input_dir.glob(f"{pattern}.copc.las")))
                 # Remove duplicates
                 matches = list(dict.fromkeys(matches))
                 if not matches:

lidar_pipeline/dtm.py

@@ -218,12 +218,9 @@ def classify_ground(laz_file, temp_dir, method='auto', force=False):
     else:
         logger.info(f"  Classification sol: {method.upper()} (forcé)")
 
-    # Strip all known LiDAR extensions (.copc.laz, .laz, .las)
+    # Use shared basename extraction function
-    laz_base = laz_file.name
+    from .pipeline import _file_basename
-    for ext in ['.copc.laz', '.copc.las', '.laz', '.las']:
+    laz_base = _file_basename(laz_file)
-        if laz_base.lower().endswith(ext):
-            laz_base = laz_base[:-len(ext)]
-            break
 
     output_las = temp_dir / f"{laz_base}_ground_{method}.las"
 

lidar_pipeline/gpu.py

@@ -57,16 +57,17 @@ def log_gpu_status():
 
 
 def to_gpu(arr):
-    """Send array to GPU if available, otherwise return as float64 numpy.
+    """Send array to GPU if available, otherwise return as float32 numpy.
 
-    Falls back to CPU if GPU is unavailable (e.g. in forked subprocess).
+    Uses float32 to reduce GPU memory usage. Falls back to CPU if GPU
+    is unavailable (e.g. in forked subprocess).
     """
     if _gpu_available():
         try:
-            return _cp.asarray(arr.astype(np.float64))
+            return _cp.asarray(arr.astype(np.float32))
         except Exception:
             pass  # Fall back to CPU
-    return arr.astype(np.float64)
+    return arr.astype(np.float32)
 
 
 def to_cpu(arr):

lidar_pipeline/pipeline.py

@@ -3,7 +3,7 @@
 LidarArchaeoPipeline coordinates the full processing chain:
 1. Ground classification (PDAL/SMRF)
 2. DTM generation
-3. Visualization generation (19 products)
+3. Visualization generation (18 products)
 4. Rendering (WebP + PDF report)
 """
 

lidar_pipeline/rendering.py

@@ -345,13 +345,19 @@ def tif_to_png(tif_file, vis_dir, resolution):
             has_nan_mask = nan_mask is not None and not is_rgb_result
             if has_nan_mask:
                 # data is normalized 0-1 from _apply_colormap; apply cmap to get RGBA
-                cmap_obj = plt.get_cmap(cmap) if isinstance(cmap, str) else cmap
+                # Save the colormap for colorbar before converting to RGBA
-                rgba = cmap_obj(data)  # (H, W, 4) float RGBA
+                saved_cmap = plt.get_cmap(cmap) if isinstance(cmap, str) else cmap
+                saved_vmin = float(np.nanmin(data)) if not nan_mask.all() else 0
+                saved_vmax = float(np.nanmax(data)) if not nan_mask.all() else 1
+                rgba = saved_cmap(data)  # (H, W, 4) float RGBA
                 rgba[nan_mask, 3] = 0.0  # transparent where no data
                 data = rgba
                 is_rgba = True
             else:
                 is_rgba = False
+                saved_cmap = None
+                saved_vmin = None
+                saved_vmax = None
 
             # Create figure — adapt width to data resolution for sharp rendering
             # At high res (5000+px wide), we need a larger figure to avoid downsampling artifacts
@@ -375,7 +381,14 @@ def tif_to_png(tif_file, vis_dir, resolution):
             ax.set_title(f"{title}\n{description}", fontsize=15, fontweight='bold', pad=10)
 
             if not is_rgb:
-                cbar = plt.colorbar(im, ax=ax, pad=0.02, shrink=0.85, aspect=30)
+                if is_rgba and saved_cmap is not None:
+                    # Create a ScalarMappable for the colorbar from the saved colormap
+                    sm = plt.cm.ScalarMappable(cmap=saved_cmap,
+                                               norm=plt.Normalize(vmin=saved_vmin, vmax=saved_vmax))
+                    sm.set_array([])
+                    cbar = plt.colorbar(sm, ax=ax, pad=0.02, shrink=0.85, aspect=30)
+                else:
+                    cbar = plt.colorbar(im, ax=ax, pad=0.02, shrink=0.85, aspect=30)
                 cbar.ax.tick_params(labelsize=9, width=1.5)
                 cbar.outline.set_linewidth(1.5)
                 cbar.set_label(legend_label, fontsize=10, fontweight='bold')
@@ -486,16 +499,18 @@ def tif_to_png(tif_file, vis_dir, resolution):
             # Save as PNG then convert to WebP — use higher DPI for large data
             save_dpi = 200 if width > 3000 else 150
             png_temp = vis_dir / f"{tif_file.stem}_temp.png"
-            plt.savefig(png_temp, dpi=save_dpi, bbox_inches='tight', pad_inches=0.15,
+            try:
-                       facecolor='white', format='png')
+                plt.savefig(png_temp, dpi=save_dpi, bbox_inches='tight', pad_inches=0.15,
-            plt.close()
+                           facecolor='white', format='png')
+            finally:
+                plt.close()
 
             img = PILImage.open(str(png_temp))
             img.save(str(webp_file), format='WEBP', lossless=True)
-            png_temp.unlink()
+            png_temp.unlink(missing_ok=True)
 
             # Delete source TIFF
-            tif_file.unlink()
+            tif_file.unlink(missing_ok=True)
 
             return webp_file