neo_reader/neo_node.py

"""NEO nodes implementation in ngclib repository"""
from pathlib import Path
import numpy as np
from codecs import encode
from overrides import overrides
import torch as tr

from .multitask_dataset import NpzRepresentation

def _cmap_hex_to_rgb(hex_list):
    res = []
    for hex_data in hex_list:
        r = int(hex_data[1: 3], 16)
        g = int(hex_data[3: 5], 16)
        b = int(hex_data[5: 7], 16)
        res.append([r, g, b])
    return np.array(res)

def _act_to_cmap(act_file: Path) -> np.ndarray:
    """converts the .act file to a matplotlib cmap representation"""
    with open(act_file, "rb") as act:
        raw_data = act.read()                           # Read binary data
    hex_data = encode(raw_data, "hex")                  # Convert it to hexadecimal values
    total_colors_count = int(hex_data[-7:-4], 16)       # Get last 3 digits to get number of colors total
    total_colors_count = 256

    # Decode colors from hex to string and split it by 6 (because colors are #1c1c1c)
    colors = [hex_data[i: i + 6].decode() for i in range(0, total_colors_count * 6, 6)]

    # Add # to each item and filter empty items if there is a corrupted total_colors_count bit
    hex_colors = [f"#{i}" for i in colors if len(i)]
    rgb_colors = _cmap_hex_to_rgb(hex_colors)
    return rgb_colors

class NEONode(NpzRepresentation):
    """NEO nodes implementation"""
    def __init__(self, node_type: str, name: str):
        self.node_type = node_type
        self.name = name
        act_path = Path(__file__).absolute().parent / "cmaps" / f"{self.node_type}.act"
        assert act_path.exists(), f"Node type '{node_type}' not valid. No act file found: '{act_path}'"
        self.cmap = _act_to_cmap(act_path)

    @overrides
    def load_from_disk(self, path: Path) -> tr.Tensor:
        data = np.load(path, allow_pickle=False)
        y = data if isinstance(data, np.ndarray) else data["arr_0"] # in case on npz, we need this as well
        y = y[0] if y.shape[0] == 1 else y # pylint: disable=unsubscriptable-object
        y = np.expand_dims(y, axis=-1) if len(y.shape) == 2 else y
        y[y == 0] = float("nan")
        return tr.from_numpy(y).float()

    @overrides
    def save_to_disk(self, data: tr.Tensor, path: Path):
        return super().save_to_disk(data.clip(0, 1), path)

    def plot_fn(self, x: tr.Tensor) -> np.ndarray:
        y = np.clip(x.cpu().detach().numpy(), 0, 1)
        y = y * 255
        y[np.isnan(y)] = 255
        y = y.astype(np.uint).squeeze()
        y_rgb = self.cmap[y].astype(np.uint8)
        return y_rgb