Source code for oursin.volumes

"""Volumetric datasets (x*y*z matrix)"""

from . import client
from . import utils
import numpy as np
import zlib
import json
import csv
import base64

from vbl_aquarium.models.urchin import VolumeMetaModel, VolumeDataChunk

counter = 0
volumes = []

CHUNK_LIMIT = 1000000

click_list = []
verbose = False

def _volume_click(data):
	"""Internal callback function
	"""
	click_pos = json.loads(data)
	if verbose:
		print(click_pos)
	
	click_list.append(click_pos)

[docs] def clear_clicks(): """Clear the volumes click list """ global click_list click_list = []
[docs] def save_clicks(fpath): """Save the current click list to a CSV file. Use `urchin.volumes.clear_clicks()` before starting your click sequence. Parameters ---------- fpath : string Relative filepath """ global click_list # Extract headers and data headers = ['ap', 'ml', 'dv'] data = [[entry[header] for header in headers] for entry in click_list] with open(fpath, mode='w', newline='') as file: writer = csv.writer(file) # Write headers writer.writerow(headers) # Write data writer.writerows(data)
[docs] def clear(): """Clear all custom meshes """ global volumes for volume in volumes: volume.delete() volumes = []
[docs] class Volume: """Volumetric dataset represented in a compressed format by using a colormap to translate uint8 x/y/z data into full RGB color. Volumes should be created in (AP, ML, DV) """ def __init__(self, volume_data, colormap = None): """_summary_ Parameters ---------- volume_data : _type_ _description_ colormap : _type_, optional _description_, by default None """ global counter, volumes self.id = f'volume{counter}' counter += 1 volume_data[np.isnan(volume_data)] = 255 flattened_data = volume_data.flatten().astype(np.uint8).tobytes() compressed_data = zlib.compress(flattened_data) compressed_data = base64.b64encode(compressed_data).decode('utf-8') if colormap is None: colormap = ['#000000'] * 255 self.data = VolumeMetaModel( name = f'volume{counter}', n_bytes = len(compressed_data), colormap = [utils.formatted_color(color) for color in colormap], visible = True ) self.update() # send data packets # split data into chunks n_chunks = int(np.ceil(self.data.n_bytes / CHUNK_LIMIT)) print(f'Data fits in {n_chunks} chunks of 1MB or less') offset = 0 for chunk in range(n_chunks): # get the data chunk_size = np.min((self.data.n_bytes - offset, CHUNK_LIMIT)) chunk_data = VolumeDataChunk( name = self.data.name, # bytes = base64.b64encode(compressed_data[offset : offset + chunk_size]).decode('utf-8') bytes = compressed_data[offset : offset + chunk_size] ) client.sio.emit('SetVolumeData', chunk_data.to_json_string()) offset += chunk_size volumes.append(self)
[docs] def update(self): client.sio.emit('UpdateVolume', self.data.to_json_string())
[docs] def delete(self): client.sio.emit('DeleteVolume', self.id)
[docs] def compress_volume(volume_data, n_colors=254): """Compress a volume of float data into a uint8 volume by quantiles. NaN values are mapped to 255 (transparent) for Urchin. This is required for use with the urchin.volume.Volume object type. Parameters ---------- volume_data : float volume 3D matrix of float data n_colors : int (optional) Default to 254, number of un-reserved colors. 255 must always be reserved for NaN / transparency Returns ------- (uint8 volume, float[] map) """ valid_values = volume_data[~np.isnan(volume_data)] quantiles = np.quantile(valid_values.flatten(), np.linspace(0,1,n_colors)) out = np.digitize(volume_data, quantiles, right=True).astype(np.uint8) out[np.isnan(volume_data)] = 255 return out.astype(np.uint8), quantiles
[docs] def colormap(colormap_name='greens', reserved_colors=[], datapoints=None): """Build a colormap This function has two parts: 1. It builds a standard colormap in indexes 0->n_colors 2. It leaves "reserved" colors at the end, by default this is just 255 which becomes transparent in Urchin. But you can add a list of additional colors which will be added to the end of the colormap. The order will match the list you pass in, so e.g. 3. If you pass in datapoints, it will generate a non-uniform colormap going from the min to maximum value. indexes: [0->252, 253->254, 255] colors: [greens, your reserved colors, transparent] Colormap options ---------- reds: 0->255 R channel greens: 0->255 G channel blues: 0->255 B channel Parameters ---------- colormap_name : str, optional _description_, by default 'greens' reserved_colors : _type_, optional _description_, by default None Returns ------- list of string List of colormap hex colors in Urchin-compatible format """ colormap = [] n_unreserved = 254 - len(reserved_colors) datapoints = (datapoints - np.min(datapoints)) / (np.max(datapoints) - np.min(datapoints)) datapoints = datapoints / np.max(datapoints) for i in range(0, n_unreserved): if datapoints is not None: v = int(np.round(datapoints[i] * 255)) else: v = int(np.round(i / n_unreserved * 255)) if colormap_name == 'reds': colormap.append(utils.rgba_to_hex((v,0,0,255))) elif colormap_name == 'greens': colormap.append(utils.rgba_to_hex((0,v,0,255))) elif colormap_name == 'blues': colormap.append(utils.rgba_to_hex((0,0,v,255))) else: raise Exception(f'{colormap_name} is not a valid colormap option') colormap.extend(reserved_colors) return colormap