pyvcad-attribute-resolver 3.0.0

Solver-based attribute resolver for OpenVCAD.

pyvcad_attribute_resolver

Solver-based attribute resolution for OpenVCAD models. Automatically converts design-intent attributes (e.g. density, shore hardness) into machine-level attributes (e.g. nozzle temperature, flow-rate multiplier) using registered conversion data.

See the main OpenVCAD package for more info. For a deep-dive into the architecture, see design_doc.md.

Quick Start

import pyvcad as pv
import pyvcad_attribute_resolver as resolver

# Load built-in foaming filament data
resolver.register_foaming_conversions()

# Build a design with a physical attribute
sphere = pv.Sphere(pv.Vec3(0, 0, 0), 10.0)
sphere.set_attribute(pv.DefaultAttributes.TEMPERATURE,
                     pv.FloatAttribute("x + 210"))

# The resolver figures out: temperature -> flow_rate
result = resolver.adapt(sphere, ["flow_rate"], tags=["foaming_tpu"])

API Reference

Function	Description
register_conversion(source, target, converter_factory, ...)	Register a new conversion edge
unregister_conversion(name)	Remove a conversion by name
list_conversions()	List all registered ConversionEntry objects
clear_conversions()	Remove all registered conversions
adapt(design, targets, tags=..., ...)	Wrap a design node with automatic conversions
list_modules()	Discover installed domain modules
register_foaming_conversions()	Register all 5 foaming filament conversions
register_tpu_conversions()	Register VarioShore TPU conversions only
register_pla_conversions()	Register LW-PLA conversions only
register_j750_shore_hardness_conversions()	Register J750 shore-hardness to volume-fractions conversions
register_j750_modulus_toughness_conversions()	Register the J750 modulus+toughness inverse-design conversion
register_generic_color_conversions(alpha=1.0)	Register machine-agnostic COLOR_RGB <-> COLOR_RGBA conversions
generate_j750_modulus_toughness_lookup_table(...)	Regenerate the bundled J750 inverse lookup table
generate_j750_modulus_toughness_reachable_region_plot(...)	Regenerate the J750 reachable-region validation plot

---

Adding a New Conversion Module

This section walks through adding your own material data to the resolver.

1. Create the Module Folder

Create a new sub-package under modules/:

resolver/src/pyvcad_attribute_resolver/modules/
└── your_material/
    ├── __init__.py
    └── data.py          # or .csv, .json — whatever suits your data

2. Add Your Data

Put your empirical measurements in data.py. Each dataset is a list of (input_value, output_value) tuples:

# data.py

#: Density (g/cm³) -> temperature (°C)
DENSITY_TO_TEMPERATURE = [
    (0.50,  260),
    (0.65,  250),
    (0.80,  240),
    (1.00,  230),
]

#: Temperature (°C) -> flow rate multiplier
TEMPERATURE_TO_FLOW_RATE = [
    (230, 1.20),
    (240, 1.00),
    (250, 0.85),
    (260, 0.75),
]

Tip: If your data lives in CSV files, you can load them in data.py using importlib.resources or pathlib.Path(__file__).parent to locate the files relative to the module.

3. Write Registration Functions

In __init__.py, create factory functions and register them:

# __init__.py
import pyvcad as pv
from pyvcad_attribute_resolver.registry import register_conversion
from .data import DENSITY_TO_TEMPERATURE, TEMPERATURE_TO_FLOW_RATE


def _lut_factory(source_attr, target_attr, data):
    """Return a callable that builds a fresh LookupTableConverter."""
    entries = [pv.LookupTableEntry(kv[0], kv[0], kv[1]) for kv in data]

    def _factory():
        return pv.LookupTableConverter(
            [source_attr], [target_attr],
            entries, pv.InterpolationMode.LINEAR,
        )
    return _factory


def register_your_material_conversions():
    """Register all conversions for your material."""
    register_conversion(
        source="density",
        target="temperature",
        converter_factory=_lut_factory(
            pv.DefaultAttributes.DENSITY,
            pv.DefaultAttributes.TEMPERATURE,
            DENSITY_TO_TEMPERATURE,
        ),
        name="your_material_density_to_temperature",
        priority=0,
        tags=["your_material"],
    )
    register_conversion(
        source="temperature",
        target="flow_rate",
        converter_factory=_lut_factory(
            pv.DefaultAttributes.TEMPERATURE,
            pv.DefaultAttributes.FLOW_RATE,
            TEMPERATURE_TO_FLOW_RATE,
        ),
        name="your_material_temperature_to_flow_rate",
        priority=0,
        tags=["your_material"],
    )

Key points:

• name must be globally unique across all modules.
• tags should identify your material so users can filter with tags=["your_material"]. Use distinct tags to avoid ambiguity with other modules that define the same conversion edges.
• required_inputs should list every attribute the converter actually consumes. Leave it unset for ordinary one-input conversions; set it explicitly for cases like modulus + toughness -> volume_fractions.
• converter_factory must be a zero-arg callable that returns a new converter instance each time.

J750 Modulus/Toughness Module

The built-in j750_modulus_toughness module inverse-designs volume_fractions from two design-intent attributes:

• modulus in MPa
• toughness in MJ/m^3

It uses the legacy C++ model to generate a 128 x 128 lookup table over:

• log10(modulus [MPa]) in [-1.0, 3.5]
• toughness [MJ/m^3] in [0.0, 6.5]

The shipped runtime table assumes the J750 materials are:

• Agilus30Clr
• VeroBlack
• M.Cleanser

and uses a default maximum cleanser volume fraction of 0.30.

To register the built-in table:

import pyvcad_attribute_resolver as resolver

resolver.register_j750_modulus_toughness_conversions()

To regenerate the lookup table with a different liquid cap or resolution:

resolver.generate_j750_modulus_toughness_lookup_table(
    output_path="j750_custom_lut.json",
    table_size=192,
    max_liquid_volume=0.25,
    fail_threshold=0.05,
)

Then register that table explicitly:

resolver.register_j750_modulus_toughness_conversions(
    lookup_table_path="j750_custom_lut.json",
)

The generator stores a separate validity mask alongside the fraction grid. Invalid cells represent unreachable modulus/toughness targets; runtime lookup raises immediately instead of clamping to a nearby printable mixture.

4. Wire It Into the Package

Add a convenience re-export in the top-level __init__.py:

# resolver/src/pyvcad_attribute_resolver/__init__.py
from .modules.your_material import register_your_material_conversions

5. Use It

import pyvcad as pv
import pyvcad_attribute_resolver as resolver

resolver.register_your_material_conversions()

design = pv.Sphere(pv.Vec3(0, 0, 0), 10.0)
design.set_attribute(pv.DefaultAttributes.DENSITY,
                     pv.FloatAttribute(0.7))

# Automatically resolves: density -> temperature -> flow_rate
result = resolver.adapt(design, ["temperature", "flow_rate"],
                        tags=["your_material"])

6. Add Tests

Create a test file in resolver/tests/ (see test_foaming.py for an example):

# resolver/tests/test_your_material.py
import pytest
import pyvcad as pv
import pyvcad_attribute_resolver as resolver

@pytest.fixture(autouse=True)
def clean():
    resolver.clear_conversions()
    yield
    resolver.clear_conversions()

def test_registration():
    resolver.register_your_material_conversions()
    names = {e.name for e in resolver.list_conversions()}
    assert "your_material_density_to_temperature" in names
    assert "your_material_temperature_to_flow_rate" in names

def test_adapt_chain():
    resolver.register_your_material_conversions()
    sphere = pv.Sphere(pv.Vec3(0, 0, 0), 1.0)
    sphere.set_attribute("density", pv.FloatAttribute(0.7))
    result = resolver.adapt(sphere,
                            ["temperature", "flow_rate"],
                            tags=["your_material"])
    assert "temperature" in result.attribute_list()
    assert "flow_rate" in result.attribute_list()

Run with:

cd resolver && python -m pytest tests/ -v

---

Module Directory Structure

resolver/src/pyvcad_attribute_resolver/
├── __init__.py              # core public API
├── adapt.py                 # adapt() entry point
├── conversion_entry.py      # ConversionEntry dataclass
├── conversion_graph.py      # directed graph
├── registry.py              # global register/unregister/list/clear
├── solver.py                # BFS pathfinder
└── modules/
    ├── __init__.py           # list_modules() discovery
    ├── foaming/              # built-in foaming filament data
    │   ├── __init__.py
    │   └── data.py
    └── your_model(s)/        # your new module
        ├── __init__.py
        └── data.py