Script to manage assets with Blender.
Blender Asset Tracer, a.k.a. BAT🦇, is the replacement of BAM and blender-file
Development is driven by choices explained in T54125.
python3.9 -m venv .venv
. ./.venv/bin/activate
pip install -U pip
pip install poetry black
poetry install
mypy --install-types
BAT Pack supports uploading to S3-compatible storage. This requires a credentials file in
~/.aws/credentials
. Replace the all-capital words to suit your situation.
[ENDPOINT]
aws_access_key_id = YOUR_ACCESS_KEY_ID
aws_secret_access_key = YOUR_SECRET
You can then send a BAT Pack to the storage using a target s3:/ENDPOINT/bucketname/path-in-bucket
,
for example:
bat pack my_blendfile.blend s3:/storage.service.cloud/jobs/awesome_work
This will upload the blend file and its dependencies to awesome_work/my_blendfile.blend
in
the jobs
bucket.
There are two object types used to represent file paths. Those are strictly separated.
bpathlib.BlendPath
represents a path as stored in a blend file. It consists of bytes, and is blendfile-relative when it starts with//
. It can represent any path from any OS Blender supports, and as such should be used carefully.pathlib.Path
represents an actual path, possibly on the local filesystem of the computer running BAT. Any filesystem operation (such as checking whether it exists) must be done using apathlib.Path
.
When it is necessary to interpret a bpathlib.BlendPath
as a real path instead of a sequence of
bytes, BAT first attempts to decode it as UTF-8. If that fails, the local filesystem encoding is
used. The latter is also no guarantee of correctness, though.
The code statically type-checked with mypy.
Mypy likes to see the return type of __init__
methods explicitly declared as None
. Until issue
#604 is resolved, we just do this in our code too.
BAT can be used as a Python library to inspect the contents of blend files, without having to open Blender itself. Here is an example showing how to determine the render engine used:
#!/usr/bin/env python3.7
import json
import sys
from pathlib import Path
from blender_asset_tracer import blendfile
from blender_asset_tracer.blendfile import iterators
if len(sys.argv) != 2:
print(f'Usage: {sys.argv[0]} somefile.blend', file=sys.stderr)
sys.exit(1)
bf_path = Path(sys.argv[1])
bf = blendfile.open_cached(bf_path)
# Get the first window manager (there is probably exactly one).
window_managers = bf.find_blocks_from_code(b'WM')
assert window_managers, 'The Blend file has no window manager'
window_manager = window_managers[0]
# Get the scene from the first window.
windows = window_manager.get_pointer((b'windows', b'first'))
for window in iterators.listbase(windows):
scene = window.get_pointer(b'scene')
break
# BAT can only return simple values, so it can't return the embedded
# struct 'r'. 'r.engine' is a simple string, though.
engine = scene[b'r', b'engine'].decode('utf8')
xsch = scene[b'r', b'xsch']
ysch = scene[b'r', b'ysch']
size = scene[b'r', b'size'] / 100.0
render_info = {
'engine': engine,
'frame_pixels': {
'x': int(xsch * size),
'y': int(ysch * size),
},
}
json.dump(render_info, sys.stdout, indent=4, sort_keys=True)
print()
To understand the naming of the properties, look at Blender's DNA_xxxx.h
files with struct
definitions. It is those names that are accessed via blender_asset_tracer.blendfile
. The
mapping to the names accessible in Blender's Python interface can be found in the rna_yyyy.c
files.
This section documents some guidelines for the code in BAT.
All imports should be done at the top level of the file, and not inside functions. The goal is to ensure that, once imported, a (sub)module of BAT can be used without having to import more parts of BAT.
This requirement helps to keep Blender add-ons separated, as an add-on can
import the modules of BAT it needs, then remove them from sys.modules
and
sys.path
so that other add-ons don't see them. This should reduce problems
with various add-ons shipping different versions of BAT.
For uploading packages to PyPi, an API key is required; username+password will not work.
First, generate an API token at https://pypi.org/manage/account/token/. Then, use this token when publishing instead of your username and password.
As username, use __token__
.
As password, use the token itself, including the pypi-
prefix.
See https://pypi.org/help/#apitoken for help using API tokens to publish. This
is what I have in ~/.pypirc
:
[distutils]
index-servers =
bat
# Use `twine upload -r bat` to upload with this token.
[bat]
repository = https://upload.pypi.org/legacy/
username = __token__
password = pypi-abc-123-blablabla
. ./.venv/bin/activate
pip install twine
poetry build
twine check dist/blender_asset_tracer-1.15.tar.gz dist/blender_asset_tracer-1.15-*.whl
twine upload -r bat dist/blender_asset_tracer-1.15.tar.gz dist/blender_asset_tracer-1.15-*.whl