"""
CAD shape to tetrahedral FEA mesh
=================================

Drive a CAD shape through gmsh to obtain a tetrahedral mesh suitable
for finite-element analysis, then bring the volumetric result back
into PyVista as an :class:`pyvista.UnstructuredGrid`.

This imports a real STEP part, the NIST AM Bench 2022 LPBF bridge
specimen (public domain, https://doi.org/10.18434/mds2-2607),
directly into gmsh via ``gmsh.model.occ.importShapes``, the canonical
way to move from CAD to FEA, and meshes the volume tetrahedrally.
"""

# %%

import gmsh
import pandas as pd
import pyvista as pv

import pyvista_cad
from pyvista_cad.examples import downloads

# %%
# Import the real STEP into gmsh's OCC kernel and mesh the volume.

gmsh.initialize()
try:
    gmsh.model.add('part_fea')
    gmsh.model.occ.importShapes(downloads.step_part_path())
    gmsh.model.occ.synchronize()

    gmsh.option.setNumber('Mesh.MeshSizeMax', 2.0)
    gmsh.option.setNumber('Mesh.MeshSizeMin', 0.5)
    gmsh.model.mesh.generate(3)
    grid = pyvista_cad.from_gmsh('part_fea')
finally:
    gmsh.finalize()

# gmsh returns a mixed grid (vertex, line, triangle, tet). For a
# volumetric FEA mesh we only keep VTK_TETRA (cell type 10).
VTK_TETRA = 10
grid = grid.extract_cells(grid.celltypes == VTK_TETRA)
grid

# %%
# Quality check: a healthy tet mesh has no zero-volume elements.

vols = grid.compute_cell_sizes(length=False, area=False, volume=True)['Volume']

pd.DataFrame(
    {
        'metric': ['min', 'mean', 'max'],
        'tet_volume': [vols.min(), vols.mean(), vols.max()],
    }
)

# %%
# Slice through the part to expose the tet interior.

bnds = grid.bounds
clip = grid.clip(normal='x', crinkle=True)
clip['Z'] = clip.points[:, 2]

pl = pv.Plotter()
pl.add_mesh(grid, color='lightgray', opacity=0.15)
pl.add_mesh(clip, scalars='Z', cmap='inferno', show_edges=True, edge_color='black')
pl.show()