Read and write Valve's KeyValues3 format
Project description
keyvalues3
KeyValues3 is a Valve developed data format. It is similar in structure to JSON, but supports binary encodings, versioning, and data annotations. The text syntax also has some minor ergonomic improvements (support for single- and multi-line comments, trailing commas, and multi-line strings.)
Usage
import keyvalues3 as kv3
bt_config = kv3.read("tests/documents/bt_config.kv3")
>>> bt_config.keys()
dict_keys(['default', 'low', 'fair', 'normal', 'tough', 'hard', 'very_hard', 'expert', 'elite'])
>>> bt_config["elite"]["reaction_time"]
0.12
# The root value is most of the time a dict
>>> type(bt_config.value)
<class 'dict'>
>>> bt_config.original_encoding
Encoding(name='text', version=UUID('e21c7f3c-8a33-41c5-9977-a76d3a32aa0d'))
>>> bt_config.format
Format(name='generic', version=UUID('7412167c-06e9-4698-aff2-e63eb59037e7'))
# To write it back
>>> kv3.write(bt_config, "tests/documents/bt_config.kv3", use_original_encoding=True)
# Write to a stream
>>> import sys
>>> kv3.write({"key": [1,2,3]}, sys.stdout)
<!-- kv3 encoding:text:version{e21c7f3c-8a33-41c5-9977-a76d3a32aa0d} format:generic:version{7412167c-06e9-4698-aff2-e63eb59037e7} -->
{
key = [1, 2, 3]
}
Install 
pip install keyvalues3
Supported encodings
Encoding 👩💻 | Read 📖 | Write ✍️ |
---|---|---|
Text UTF-8 | Yes ✔️ | Yes ✔️ |
Text UTF-8 Headerless | Yes ✔️ | Yes ✔️ |
Binary Uncompressed | No ⛔ | Yes ✔️ |
Binary LZ4 | No ⛔ | Yes ✔️ |
Binary (Other newer) | No ⛔ | No ⛔ |
Using dataclasses
Dataclasses (marked with @dataclass
) are also supported for serialization (uses asdict()
)
@dataclass
class RenderMeshFile(_Node): # _Node defines _class, name, notes, and children (+ helper methods)
filename: resourcepath = ""
import_translation: list[float] = field(default_factory=lambda: [0, 0, 0])
import_rotation: list[float] = field(default_factory=lambda: [0, 0, 0])
import_scale: float = 1.0
align_origin_x_type: str = "None"
align_origin_y_type: str = "None"
align_origin_z_type: str = "None"
parent_bone: namelink = ""
import_filter: dict = field(
default_factory=lambda:dict(
exclude_by_default = False,
exception_list = [ ]
)
)
You can find more ModelDoc dataclasses here.
Credits
Valve Corporation® for making KeyValues3.
ValveResourceFormat for reverse engineering the binary formats.