423A35C7/pytorch3d
1
0
Fork 0
mirror of https://github.com/facebookresearch/pytorch3d.git synced 2026-10-20 13:12:25 +08:00
Code Issues Packages Projects Releases Wiki Activity

Accept tab-separated lines in load_obj (#2061)

Browse Source 
Summary:
`load_obj` recognizes vertex, texture, normal and face lines with `line.startswith("v ")`, `"vt "`, `"vn "` and `"f "`. The OBJ format separates fields with any whitespace, and some exporters write a tab after the keyword. Those lines are skipped without any error, so the mesh silently loses vertices and faces:

```python
from io import StringIO
from pytorch3d.io import load_obj

verts, faces, _ = load_obj(StringIO("v\t0 0 0\nv 1 0 0\nv 0 1 0\nf\t1 2 3\n"))
verts.shape, faces.verts_idx.shape
# before: (2, 3), (0, 3)   after: (3, 3), (1, 3)
```

This dispatches on the first token of the already split line (`tokens[:1] == ["v"]`), the same way the `usemtl` branch already does.

Pull Request resolved: https://github.com/facebookresearch/pytorch3d/pull/2061

Test Plan: added `test_load_obj_tab_separated` to `tests/test_io_obj.py`, using tabs in `v`, `vt`, `vn` and `f` lines. It fails on main and passes with the change. I ran `tests/test_io_obj.py` on Windows without the C++ extension: the same 19 tests fail before and after (they reopen a `NamedTemporaryFile` or need `_C`), and the rest pass. flake8 is clean on the changed lines.

Reviewed By: MichaelRamamonjisoa

Differential Revision: D124656223

Pulled By: bottler

fbshipit-source-id: ab0ef3dc0f3c998c44874e6a416e6a9649f29378
This commit is contained in:
Raashish Aggarwal
2026-10-19 14:36:14 -07:00
committed by meta-codesync[bot]
parent 88e182f989
commit 8b3ba6ede4
2 changed files with 27 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
pytorch3d/io/obj_io.py
View File

@@ -472,26 +472,26 @@ def _parse_obj(f, data_dir: str):
materials_idx = len(material_names) - 1
else:
materials_idx = material_names.index(material_name)
elif line.startswith("v "): # Line is a vertex.
elif tokens[:1] == ["v"]: # Line is a vertex.
vert = [float(x) for x in tokens[1:4]]
if len(vert) != 3:
msg = "Vertex %s does not have 3 values. Line: %s"
raise ValueError(msg % (str(vert), str(line)))
verts.append(vert)
elif line.startswith("vt "): # Line is a texture.
elif tokens[:1] == ["vt"]: # Line is a texture.
tx = [float(x) for x in tokens[1:3]]
if len(tx) != 2:
raise ValueError(
"Texture %s does not have 2 values. Line: %s" % (str(tx), str(line))
)
verts_uvs.append(tx)
elif line.startswith("vn "): # Line is a normal.
elif tokens[:1] == ["vn"]: # Line is a normal.
norm = [float(x) for x in tokens[1:4]]
if len(norm) != 3:
msg = "Normal %s does not have 3 values. Line: %s"
raise ValueError(msg % (str(norm), str(line)))
normals.append(norm)
elif line.startswith("f "): # Line is a face.
elif tokens[:1] == ["f"]: # Line is a face.
# Update face properties info.
_parse_face(
line,

23
tests/test_io_obj.py
View File

@@ -319,6 +319,29 @@ class TestMeshObjIO(TestCaseMixin, unittest.TestCase):
self.assertTrue(materials is None)
self.assertTrue(tex_maps is None)
def test_load_obj_tab_separated(self):
# Tabs are whitespace too, including between the keyword and the values.
obj_file = "\n".join(
[
"v\t0.1\t0.2\t0.3",
"v 0.2 0.3 0.4",
"v\t0.3 0.4 0.5",
"vt\t0.5\t0.5",
"vn\t0.0\t0.0\t1.0",
"f\t1/1/1\t2/1/1\t3/1/1",
]
)
verts, faces, aux = load_obj(StringIO(obj_file))
expected_verts = torch.tensor(
[[0.1, 0.2, 0.3], [0.2, 0.3, 0.4], [0.3, 0.4, 0.5]], dtype=torch.float32
)
self.assertClose(verts, expected_verts)
self.assertClose(faces.verts_idx, torch.tensor([[0, 1, 2]]))
self.assertClose(faces.textures_idx, torch.tensor([[0, 0, 0]]))
self.assertClose(faces.normals_idx, torch.tensor([[0, 0, 0]]))
self.assertClose(aux.verts_uvs, torch.tensor([[0.5, 0.5]]))
self.assertClose(aux.normals, torch.tensor([[0.0, 0.0, 1.0]]))
def test_load_obj_error_textures(self):
obj_file = "\n".join(["vt 0.1"])
with NamedTemporaryFile(mode="w", suffix=".obj") as f: