gmshparser - parse Gmsh .msh file format

Package author: Jukka Aho (@ahojukka5)

Gmshparser is a small Python package which aims to do only one thing: parse Gmsh mesh file format. Package does not have any external dependencies to other packages and it aims to be a simple stand-alone solution for a common problem: how to import mesh to your favourite research FEM code?

• Project source in GitHub: https://github.com/ahojukka5/gmshparser
• Project documentation in ReadTheDocs: https://gmshparser.readthedocs.io
• Project releases in PyPi: https://pypi.org/project/gmshparser

Installing package

To install the most recent package from Python Package Index (PyPi), use git:

pip install gmshparser

To install the development version, you can install the package directly from the GitHub:

pip install git+git://github.com/ahojukka5/gmshparser.git

Using application programming interface

To read mesh into Mesh object, use command parse. It takes a filename and parses the file with the set of parsers, defined in DEFAULT_PARSERS (see developing package section for more info..!)

import gmshparser
mesh = gmshparser.parse("data/testmesh.msh")
print(mesh)

Mesh name: data/testmesh.msh
Mesh version: 4.1
Number of nodes: 6
Minimum node tag: 1
Maximum node tag: 6
Number of node entities: 1
Number of elements: 2
Minimum element tag: 1
Maximum element tag: 2
Number of element entities: 1

After reading the model, you can querying your data form mesh object. For example, to extract all nodes from the model:

for entity in mesh.get_node_entities():
    for node in entity.get_nodes():
        nid = node.get_tag()
        ncoords = node.get_coordinates()
        print("Node id = %s, node coordinates = %s" % (nid, ncoords))

Node id = 1, node coordinates = (0.0, 0.0, 0.0)
Node id = 2, node coordinates = (1.0, 0.0, 0.0)
Node id = 3, node coordinates = (1.0, 1.0, 0.0)
Node id = 4, node coordinates = (0.0, 1.0, 0.0)
Node id = 5, node coordinates = (2.0, 0.0, 0.0)
Node id = 6, node coordinates = (2.0, 1.0, 0.0)

Extract all elements from the model:

for entity in mesh.get_element_entities():
    eltype = entity.get_element_type()
    print("Element type: %s" % eltype)
    for element in entity.get_elements():
        elid = element.get_tag()
        elcon = element.get_connectivity()
        print("Element id = %s, connectivity = %s" % (elid, elcon))

Element type: 3
Element id = 1, connectivity = [1, 2, 3, 4]
Element id = 2, connectivity = [2, 5, 6, 3]

If you are writing your FEM stuff with Python, then you have access to the all relevant properties of the model using mesh object. For further information on how to access nodes, elements, physical groups, and other things what Gmsh provides, take a look of documentation.

Using command line interface

gmshparser can also be useful even if you don't make FEM code in Python. The above loops used to extract nodes and elements are actually so common, that you can use them from the command line. This way you can print nodes and elements in a simpler format with command-line tools, making it easier to read an element mesh with C ++ or Fortran, for example. To extract nodes:

jukka@jukka-XPS-13-9380:~$ gmshparser data/testmesh.msh nodes

6
1 0.000000 0.000000 0.000000
2 1.000000 0.000000 0.000000
3 1.000000 1.000000 0.000000
4 0.000000 1.000000 0.000000
5 2.000000 0.000000 0.000000
6 2.000000 1.000000 0.000000

To extract elements, use choice elements. The first line is having the total number of elements, and the rest of the lines are in format element_id element_type element_connectivity. The length of the line naturally depends on how many nodes the element is having.

jukka@jukka-XPS-13-9380:~$ gmshparser data/testmesh.msh elements

2
1 3 1 2 3 4
2 3 2 5 6 3

Visualizing meshes using gmshparser and matplotlib

The intention of the package is not to visualize meshes. But as it is a quite common task to visualize 2-dimensional triangluar meshes in acedemic papers, lecture notes, and things like that, it can be done easily using gmshparser and matplotlib. There's a helper function gmshparser.helpers.get_triangles, which returns a tuple (X, Y, T) which can then be passed to matplotlib to get a mesh plot:

import gmshparser
mesh = gmshparser.parse("data/example_mesh.msh")
X, Y, T = gmshparser.helpers.get_triangles(mesh)

import matplotlib.pylab as plt
plt.figure()
plt.triplot(X, Y, T, color='black')
plt.axis('equal')
plt.axis('off')
plt.tight_layout()
plt.savefig('docs/example_mesh.svg')

Developing package

gmshparser is written such a way, that it's easy to define your own parsers which are responsible for parsing some section, starting with $SectionName and ending with $EndSectionName. For example, a parser which is responsible to parse MeshFormat setion is MainFormatParser and it is defined with the following code:

class MeshFormatParser(AbstractParser):

    @staticmethod
    def get_section_name():
        return "$MeshFormat"

    @staticmethod
    def parse(mesh: Mesh, io: TextIO) -> None:
        s = io.readline().strip().split(" ")
        mesh.set_version(float(s[0]))
        mesh.set_ascii(int(s[1]) == 0)
        mesh.set_precision(int(s[2]))

All the active parsers used in parsing are then appended to the list of parsers in MainParser, from where they are called when an appropriate get_section_name() is found from file considered to be parsed. The MainParser itself is then called in parse to get things done:

def parse(filename: str) -> Mesh:
    """Parse Gmsh .msh file and return `Mesh` object."""
    mesh = Mesh()
    mesh.set_name(filename)
    parser = MainParser()
    with open(filename, "r") as io:
        parser.parse(mesh, io)
    return mesh

If you want to learn how to write your own parser, you can e.g. take of look of NodesParser which is responsible for parsing nodes and ElementsParser which is responsible for parsing elements, to get an idea how things are implemented.

Contributing to project

Like in all other open source projects, contributions are always welcome to this project too! If you have some great ideas how to make this package better, feature requests etc., you can open an issue on gmshparser's issue tracker or contact me ([email protected]) directly.