Press n or j to go to the next uncovered block, b, p or k for the previous block.
| 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 | 1x 1x 1x 1x 1x 15x 47x 15x 47x 47x 44x 34x 31x 19x 15x 15x 15x 15x 15x 15x 15x 34x 10x 34x 1x 7x 7x 7x 7x 7x 1x 6x 6x 6x 6x 6x 6x 6x 6x 6x 6x 6x 6x 6x 1x 1x 1x 1x 6x 6x 6x 6x 4x 4x 3x 7x 7x 7x 7x 7x 7x 1x 1x 6x 6x 6x 6x 6x 30x 4x 4x 4x 4x 6x 6x 6x 6x 6x 6x 6x | /**
* Lazy tensor accessor returned by getTensor().
*
* Wraps a TensorDescriptor and accumulates slice/scale parameters, then
* issues a single POST /api/slice request on .compute().
*
* Axis semantics are inferred from dim_labels (t/time, z/depth/plane,
* c/channel, y/height/row, x/width/col) with a positional heuristic fallback
* for unknown labels.
*/
import type { TensorDescriptor, TypedNdArray, SliceRequest } from "./types.js";
import type { TensorHttpClient } from "./client.js";
// ---------------------------------------------------------------------------
// Axis helpers
// ---------------------------------------------------------------------------
const SPATIAL_Y = new Set(["y", "height", "row", "rows"]);
const SPATIAL_X = new Set(["x", "width", "col", "cols", "column", "columns"]);
const SPATIAL_Z = new Set(["z", "depth", "plane", "planes", "slice"]);
const TEMPORAL = new Set(["t", "time", "frame", "frames"]);
const CHANNEL = new Set(["c", "channel", "channels", "band", "bands"]);
export interface AxisMap {
t: number | null;
z: number | null;
c: number | null;
y: number | null;
x: number | null;
}
/** Derive axis→dimension-index mapping from dim_labels. */
export function buildAxisMap(dimLabels: string[]): AxisMap {
const map: AxisMap = { t: null, z: null, c: null, y: null, x: null };
const labels = dimLabels.map((l) => l.toLowerCase().trim());
for (let i = 0; i < labels.length; i++) {
const l = labels[i] as string;
if (TEMPORAL.has(l) && map.t === null) map.t = i;
else if (SPATIAL_Z.has(l) && map.z === null) map.z = i;
else if (CHANNEL.has(l) && map.c === null) map.c = i;
else if (SPATIAL_Y.has(l) && map.y === null) map.y = i;
else if (SPATIAL_X.has(l) && map.x === null) map.x = i;
}
// Positional heuristic fallback for unlabelled / unknown axes:
// Last two unassigned dimensions → Y, X; one before that → Z
const ndim = labels.length;
if (ndim >= 1 && map.x === null) map.x = ndim - 1;
if (ndim >= 2 && map.y === null) map.y = ndim - 2;
if (ndim >= 3 && map.z === null) map.z = ndim - 3;
if (ndim >= 4 && map.c === null) map.c = ndim - 4;
if (ndim >= 5 && map.t === null) map.t = ndim - 5;
return map;
}
/** Return true when any axis was inferred by heuristic (not explicit label). */
export function isAxisMapAmbiguous(dimLabels: string[]): boolean {
const labels = dimLabels.map((l) => l.toLowerCase().trim());
const allKnown = [...TEMPORAL, ...SPATIAL_Z, ...CHANNEL, ...SPATIAL_Y, ...SPATIAL_X];
return labels.some((l) => !allKnown.includes(l));
}
// ---------------------------------------------------------------------------
// Scale selector
// ---------------------------------------------------------------------------
/** Per-dimension integer downsampling factors aligned to powers of two. */
export interface ScaleVector {
factors: number[];
/** True if any factor was snapped to a different power-of-two. */
snapped: boolean;
}
const HYSTERESIS = 0.2; // 20% band before switching scale level
/**
* Compute power-of-two scale hint for a 2D viewport.
*
* @param tensorShape Full shape of the tensor.
* @param axisMap Axis index mapping from buildAxisMap().
* @param viewportW Viewport width in physical pixels (already DPR-scaled).
* @param viewportH Viewport height in physical pixels (already DPR-scaled).
* @param pixelBudget Maximum output megapixels (default 1.0).
* @param prevFactors Previously used scale factors (for hysteresis).
*/
export function computeScaleHint(
tensorShape: number[],
axisMap: AxisMap,
viewportW: number,
viewportH: number,
pixelBudget = 1_000_000,
prevFactors?: number[],
): ScaleVector {
const ndim = tensorShape.length;
const factors = new Array<number>(ndim).fill(1);
const yIdx = axisMap.y;
const xIdx = axisMap.x;
if (yIdx === null || xIdx === null) {
return { factors, snapped: false };
}
const dataH = tensorShape[yIdx] as number;
const dataW = tensorShape[xIdx] as number;
Iif (dataH <= 0 || dataW <= 0) {
return { factors, snapped: false };
}
// How many data pixels fit into the viewport?
const rawScaleY = dataH / viewportH;
const rawScaleX = dataW / viewportW;
// Use the larger factor (more conservative / lower resolution)
const rawScale = Math.max(rawScaleY, rawScaleX);
// Budget limit: don't request more pixels than the budget
const budgetPixels = pixelBudget;
const budgetFactor = Math.sqrt((dataH * dataW) / budgetPixels);
const targetScale = Math.max(rawScale, budgetFactor, 1);
// Snap to nearest power of two
const log2 = Math.log2(targetScale);
const snappedLog2 = Math.round(log2);
let snappedFactor = Math.max(1, Math.pow(2, snappedLog2));
// Hysteresis: don't switch scale unless deviation exceeds 20%
if (prevFactors && prevFactors[yIdx] !== undefined) {
const prev = prevFactors[yIdx];
const ratio = snappedFactor / prev;
Eif (ratio > 1 - HYSTERESIS && ratio < 1 + HYSTERESIS) {
snappedFactor = prev; // stay on current level
}
}
const snapped = Math.round(snappedFactor) !== Math.round(targetScale);
factors[yIdx] = snappedFactor;
factors[xIdx] = snappedFactor;
// All other axes (T, Z, C) stay at 1
return { factors: factors.map(Math.round), snapped };
}
// ---------------------------------------------------------------------------
// TensorArray
// ---------------------------------------------------------------------------
export interface SliceOptions {
/** Fixed index (or start of range) per axis label: t, z, c, y, x. */
t?: number | [number, number];
z?: number | [number, number];
c?: number | [number, number];
y?: number | [number, number];
x?: number | [number, number];
scaleHint?: number[];
reductionMethod?: string;
pixelBudget?: number;
}
/** Expand a scalar or [start, stop] into a [start, stop] pair. */
function toRange(
val: number | [number, number] | undefined,
fullSize: number,
): [number, number] {
Iif (val === undefined) return [0, fullSize];
if (typeof val === "number") return [val, val + 1];
return val;
}
/**
* Lazy accessor for a single tensor within a data source.
*
* Call .compute(options) to fetch data from the server.
*/
export class TensorArray {
readonly descriptor: TensorDescriptor;
readonly sourceId: string;
readonly axisMap: AxisMap;
readonly axisMapAmbiguous: boolean;
private readonly _client: TensorHttpClient;
constructor(
client: TensorHttpClient,
sourceId: string,
descriptor: TensorDescriptor,
) {
this._client = client;
this.sourceId = sourceId;
this.descriptor = descriptor;
this.axisMap = buildAxisMap(descriptor.dim_labels);
this.axisMapAmbiguous = isAxisMapAmbiguous(descriptor.dim_labels);
}
get ndim(): number {
return this.descriptor.shape.length;
}
get shape(): number[] {
return this.descriptor.shape;
}
get dtype(): string {
return this.descriptor.dtype;
}
/**
* Fetch a sub-region of the tensor.
*
* @param options Per-axis slice ranges + scale/reduction settings.
* @returns TypedNdArray with raw bytes, shape, dtype, and dim labels.
*/
async compute(options: SliceOptions = {}): Promise<TypedNdArray> {
const ndim = this.ndim;
const shape = this.descriptor.shape;
const sliceStart: number[] = new Array(ndim).fill(0);
const sliceStop: number[] = [...shape];
const setAxis = (idx: number | null, val: number | [number, number] | undefined) => {
if (idx === null || val === undefined) return;
const fullSize = shape[idx] as number;
const [s, e] = toRange(val, fullSize);
sliceStart[idx] = Math.max(0, s);
sliceStop[idx] = Math.min(fullSize, e);
};
setAxis(this.axisMap.t, options.t);
setAxis(this.axisMap.z, options.z);
setAxis(this.axisMap.c, options.c);
setAxis(this.axisMap.y, options.y);
setAxis(this.axisMap.x, options.x);
const req: SliceRequest = {
source_id: this.sourceId,
tensor_id: this.descriptor.array_id,
slice_start: sliceStart,
slice_stop: sliceStop,
scale_hint: options.scaleHint,
reduction_method: options.reductionMethod,
pixel_budget: options.pixelBudget,
};
return this._client.slice(req);
}
}
|