Working with Meshes¶

This page is task-oriented. For the complete object model and filtering semantics, see Modern Data Model.

Read a path¶

import gmshparser

mesh = gmshparser.read("path/to/mesh.msh")

read() detects the supported MSH version automatically and returns an immutable gmshparser.api.Mesh.

Read a text stream¶

from io import StringIO

mesh = gmshparser.read(StringIO(msh_text), name="generated.msh")

The optional name appears in parser errors and mesh.name.

Summarize a mesh¶

print(mesh.name)
print(mesh.version)
print(mesh.dimension)
print(mesh.bounds)
print(len(mesh.nodes), len(mesh.elements), len(mesh.entities))
print(mesh.element_types)

mesh.bounds is ((xmin, ymin, zmin), (xmax, ymax, zmax)). Empty meshes have no geometric dimension or bounds.

Look up nodes by tag¶

node = mesh.nodes[42]
print(node.coordinates)
print(node.entity_key)
print(node.physical_tags)

Integer indexing always means the original Gmsh tag. It is not a positional array index.

Use get() when absence is expected:

node = mesh.nodes.get(42)
if node is not None:
    print(node.x, node.y, node.z)

Filter nodes¶

surface_nodes = mesh.nodes.where(dimension=2)
entity_nodes = mesh.nodes.by_entity(dimension=2, tag=7)
parametric_nodes = mesh.nodes.where(parametric=True)
physical_nodes = mesh.nodes.where(physical_tag=10)

Filtered collections retain normal tag lookup and parser order.

Inspect elements¶

from gmshparser import ElementType

for element in mesh.elements:
    print(element.tag, element.element_type, element.node_tags)
    for node in element:
        print(node.coordinates)

triangles = mesh.elements.by_type(ElementType.TRIANGLE)

element.nodes contains resolved Node objects. element.node_tags and element.connectivity retain the original Gmsh connectivity tags.

surface_quads = mesh.elements.where(
    element_type=ElementType.QUADRANGLE,
    dimension=2,
)
entity_elements = mesh.elements.by_entity(dimension=2, tag=7)
physical_elements = mesh.elements.where(physical_tag=10)

Work with entities¶

surface = mesh.entity(dimension=2, tag=7)

print(surface.nodes)
print(surface.elements)
print(surface.element_types)
print(surface.physical_tags)

Dimension views are available directly:

print(mesh.points)
print(mesh.curves)
print(mesh.surfaces)
print(mesh.volumes)

Filter the complete entity collection when several conditions matter:

triangle_surfaces = mesh.entities.where(
    dimension=2,
    element_type=ElementType.TRIANGLE,
    has_elements=True,
)

Export nodes to CSV¶

import csv

with open("nodes.csv", "w", newline="") as stream:
    writer = csv.writer(stream)
    writer.writerow(["tag", "x", "y", "z"])
    writer.writerows((node.tag, *node.coordinates) for node in mesh.nodes)

Build an element-to-coordinate table¶

rows = [
    (
        element.tag,
        element.element_type,
        tuple(node.coordinates for node in element),
    )
    for element in mesh.elements
]

For numerical arrays and zero-based connectivity, use NumPy Interoperability instead.

Physical and periodic metadata¶

Named boundaries, materials, and periodic node relations are covered in Physical Groups and Periodicity.

Error handling¶

try:
    mesh = gmshparser.read("mesh.msh")
except FileNotFoundError:
    print("Mesh file not found")
except gmshparser.ParseError as error:
    print(error)

See Error Handling for the complete hierarchy and source context.

Existing compatibility applications¶

legacy_mesh = gmshparser.parse("mesh.msh")

See Migrating from the Compatibility API when moving existing get_* code to the modern model.