@jbrowse/core 5.0.0-beta.1 → 5.0.0-beta.11

package/esm/util/sessionSharing.js

@@ -38,32 +38,39 @@ export function b64PadSuffix(b64) {
     }
     return b64 + '='.repeat(num);
 }
+async function transformBytes(bytes, transform) {
+    const stream = new Response(bytes).body.pipeThrough(transform);
+    return new Uint8Array(await new Response(stream).arrayBuffer());
+}
 /**
  * Decode and inflate a url-safe base64 to a string
  * See {@link https://en.wikipedia.org/wiki/Base64#URL_applications}
+ *
+ * A truncated input rejects rather than inflating to a prefix: a link cut short
+ * in transit has to say so, not fail later as a JSON syntax error.
  */
 export async function fromUrlSafeB64(b64) {
-    const originalB64 = b64PadSuffix(b64.replaceAll('-', '+').replaceAll('_', '/'));
-    const { inflate } = await import('pako-esm2');
-    const bytes = Uint8Array.from(atob(originalB64), c => c.charCodeAt(0));
-    const inflated = inflate(bytes, undefined);
-    return new TextDecoder('utf8').decode(inflated);
+    try {
+        const bytes = Uint8Array.from(atob(b64PadSuffix(b64.replaceAll('-', '+').replaceAll('_', '/'))), c => c.charCodeAt(0));
+        return new TextDecoder().decode(await transformBytes(bytes, new DecompressionStream('deflate')));
+    }
+    catch (e) {
+        throw new Error('This session is incomplete or damaged, which is how a link cut short in transit arrives. A short link survives being sent where a long one may not.', { cause: e });
+    }
 }
 /**
  * Compress and encode a string as url-safe base64
  * See {@link https://en.wikipedia.org/wiki/Base64#URL_applications}
  */
 export async function toUrlSafeB64(str) {
-    const bytes = new TextEncoder().encode(str);
-    const { deflate } = await import('pako-esm2');
-    const deflated = deflate(bytes, undefined);
+    const deflated = await transformBytes(new TextEncoder().encode(str), new CompressionStream('deflate'));
     const encoded = btoa(bytesToBinaryString(deflated)).replace(/=+$/, '');
     return encoded.replaceAll('+', '-').replaceAll('/', '_');
 }
 /**
- * The `?session=` value prefixes, named here because this module is what
- * produces them. The decoder half (jbrowse-web's SessionLoader) has its own
- * list covering the two prefixes nothing here writes (`spec-`, `local-`).
+ * The `?session=` value prefixes for a session carried by or behind the link.
+ * The decoder half (jbrowse-web's SessionLoader) has its own list covering
+ * `spec-` and `local-` too.
  */
 export const SHARE_PREFIX = 'share-';
 export const ENCODED_PREFIX = 'encoded-';
@@ -102,23 +109,16 @@ async function shareSessionToDynamo(session, shareURL, referer) {
     return { sessionId, password };
 }
 // Encodes a session snapshot into the `?session=` value jbrowse-web decodes:
-// `share-<id>` (uploaded + encrypted), `encoded-<b64>` (compressed inline), or
-// `json-<json>` (plaintext inline). Single source of these prefixes, shared by
-// jbrowse-web's ShareDialog and jbrowse-desktop's ExportToWebDialog so the two
-// producers can't drift from the decoder.
+// `share-<id>` (uploaded + encrypted) or `encoded-<b64>` (compressed inline).
+// Single source of these prefixes, shared by jbrowse-web's ShareDialog and
+// jbrowse-desktop's ExportToWebDialog so the two producers can't drift from the
+// decoder. The decoder also reads `json-{"session":…}`, which nothing here
+// writes: it is the form a person writes by hand.
 export async function encodeSessionParam(mode, session, options) {
     if (mode === 'short') {
         const { sessionId, password } = await shareSessionToDynamo(session, options.shareURL, options.referer);
         return { sessionParam: `${SHARE_PREFIX}${sessionId}`, password };
     }
-    else if (mode === 'json') {
-        // compact in the link (this mode's URL is the longest of the three
-        // already), indented in the dialog's readable-JSON panel
-        return {
-            sessionParam: `${JSON_PREFIX}${JSON.stringify({ session })}`,
-            plaintext: JSON.stringify({ session }, null, 2),
-        };
-    }
     else {
         const encoded = await toUrlSafeB64(JSON.stringify(session));
         return { sessionParam: `${ENCODED_PREFIX}${encoded}` };

package/esm/util/shapeNameList.d.ts

@@ -0,0 +1,2 @@
+/** The shape names in range order: what an unlisted `range` hands out. */
+export declare const SHAPE_NAMES: ('circle' | 'triangle-down' | 'diamond')[];

package/esm/util/shapeNameList.js

@@ -0,0 +1,6 @@
+/** The shape names in range order: what an unlisted `range` hands out. */
+export const SHAPE_NAMES = [
+    'circle',
+    'triangle-down',
+    'diamond',
+];

package/esm/util/shapeNames.d.ts

@@ -0,0 +1,7 @@
+import type { ShapeName } from './markEncodingTypes.ts';
+/**
+ * #api
+ * The `point` mark's painter code for each shape an encoding can name.
+ */
+export declare const SHAPE_CODES: Record<ShapeName, number>;
+export { SHAPE_NAMES } from './shapeNameList.ts';

package/esm/util/shapeNames.js

@@ -0,0 +1,11 @@
+import { GLYPH_DIAMOND, GLYPH_DISC, GLYPH_TRIANGLE, } from '@jbrowse/render-core/shaders/pointMarkConsts';
+/**
+ * #api
+ * The `point` mark's painter code for each shape an encoding can name.
+ */
+export const SHAPE_CODES = {
+    circle: GLYPH_DISC,
+    'triangle-down': GLYPH_TRIANGLE,
+    diamond: GLYPH_DIAMOND,
+};
+export { SHAPE_NAMES } from './shapeNameList.js';

package/esm/util/simpleFeature.d.ts

@@ -90,7 +90,6 @@ export interface SimpleFeatureSerializedNoId {
     name?: string;
     id?: string | number;
     uniqueId?: string;
-    __jbrowsefmt?: Record<string, unknown>;
     mate?: {
         refName: string;
         start: number;

package/esm/util/springAnimate.js

@@ -61,7 +61,7 @@ export function springAnimate({ from, to, write, read, onFinish = () => { }, pre
             animationFrameId = requestAnimationFrame(() => {
                 update({
                     lastPosition: position,
-                    lastTime: time,
+                    lastTime: lastTime + numSteps,
                     lastVelocity: velocity,
                 });
             });

package/esm/util/starFusion.d.ts

@@ -0,0 +1,26 @@
+export type FusionBreakpoint = {
+    refName: string;
+    start: number;
+    end: number;
+    strand?: number;
+    /** which way the sequence this end keeps runs from it: 1 right, -1 left, 0 unknown */
+    mateDirection: number;
+};
+/**
+ * The column names of a STAR-Fusion `fusion_predictions.tsv` header line. The
+ * leading `#` is not universal across STAR-Fusion versions and wrappers, so it
+ * is stripped when present. Throws a named error for a header without the two
+ * breakpoint columns, which is the wrong file or the wrong File Type guess.
+ */
+export declare function starFusionColumns(header: string | undefined): string[];
+/**
+ * A STAR-Fusion `refName:pos:strand` breakpoint as an interbase feature, read
+ * from the RIGHT because a refName may contain colons (GRCh38's
+ * `HLA-A*01:01:01:01`). The position is 1-based in the file. The strand is the
+ * gene's transcription strand, and the fusion transcript keeps the donor's
+ * sequence 5' of the junction and the acceptor's 3' of it, so `mateDirection`
+ * is the strand negated for a donor and taken as read for an acceptor: the
+ * convention `parseSvAlt` states for a breakend and the paired-arc display
+ * draws its ticks from.
+ */
+export declare function parseStarFusionBreakpoint(str: string, isDonor: boolean): FusionBreakpoint;

package/esm/util/starFusion.js

@@ -0,0 +1,37 @@
+/**
+ * The column names of a STAR-Fusion `fusion_predictions.tsv` header line. The
+ * leading `#` is not universal across STAR-Fusion versions and wrappers, so it
+ * is stripped when present. Throws a named error for a header without the two
+ * breakpoint columns, which is the wrong file or the wrong File Type guess.
+ */
+export function starFusionColumns(header) {
+    const columns = (header?.startsWith('#') ? header.slice(1) : header)?.split('\t') ?? [];
+    const missing = ['LeftBreakpoint', 'RightBreakpoint'].filter(c => !columns.includes(c));
+    if (missing.length) {
+        throw new Error(`Not a STAR-Fusion file: no ${missing.join(' or ')} column. Found: ${columns.slice(0, 6).join(', ')}`);
+    }
+    return columns;
+}
+/**
+ * A STAR-Fusion `refName:pos:strand` breakpoint as an interbase feature, read
+ * from the RIGHT because a refName may contain colons (GRCh38's
+ * `HLA-A*01:01:01:01`). The position is 1-based in the file. The strand is the
+ * gene's transcription strand, and the fusion transcript keeps the donor's
+ * sequence 5' of the junction and the acceptor's 3' of it, so `mateDirection`
+ * is the strand negated for a donor and taken as read for an acceptor: the
+ * convention `parseSvAlt` states for a breakend and the paired-arc display
+ * draws its ticks from.
+ */
+export function parseStarFusionBreakpoint(str, isDonor) {
+    const parts = str.split(':');
+    const strandStr = parts.length >= 3 ? parts.pop() : undefined;
+    const pos = parts.length >= 2 ? +parts.pop() : Number.NaN;
+    const strand = strandStr === '+' ? 1 : strandStr === '-' ? -1 : undefined;
+    return {
+        refName: parts.join(':'),
+        start: pos - 1,
+        end: pos,
+        strand,
+        mateDirection: strand === undefined ? 0 : isDonor ? -strand : strand,
+    };
+}

package/esm/util/stepTables.d.ts

@@ -0,0 +1,96 @@
+import { DerivedTable } from './featureTable.ts';
+import type { Column, FeatureTable, MadeRows } from './featureTable.ts';
+import type { AggregateOp, AggregateStep, BinStep } from './markEncodingTypes.ts';
+import type { Feature, SimpleFeatureSerialized } from './simpleFeature.ts';
+export declare const DEFAULT_BIN_AS: [string, string];
+/** What a `bin` over `fields` writes on each piece: the bases of its interval inside its bin. */
+export declare const BIN_OVERLAP_FIELD = "overlap";
+/**
+ * A table's rows split into sections, each a contiguous range: `bounds[s]` is
+ * where section `s` starts and `bounds[s + 1]` where it ends. A step keeps
+ * its rows in section order, and one that groups rows groups within a
+ * section, so a faceted request runs each step once over every section.
+ */
+export type Bounds = Uint32Array;
+export interface Staged {
+    table: FeatureTable;
+    bounds: Bounds;
+}
+export declare function boundsThrough(bounds: Bounds, parentRow: Uint32Array): Bounds;
+/**
+ * The entries a `flatten` fanned out, one row each over the container row it
+ * came from. A container kept with nothing to fan out is a row of its own
+ * that reads everything off itself.
+ */
+export declare abstract class FannedTable extends DerivedTable {
+    get madeFrom(): boolean;
+    /** A container kept with nothing to fan out: the row is the container. */
+    abstract isKept(i: number): boolean;
+    abstract get hasKept(): boolean;
+    row(i: number): Feature;
+}
+export declare function stepReader(table: FeatureTable, ref: string, step: string): (i: number) => unknown;
+export declare function stepNumberReader(table: FeatureTable, ref: string, step: string): (i: number) => number;
+export declare function groupKey(value: unknown): unknown;
+export declare function binSize({ step }: BinStep): number;
+/**
+ * Each row's interval and the bins it overlaps, `first` to `last`, with each
+ * section's lowest and highest bin: `first` NaN where the ends are not
+ * numbers or run backwards, so the row is no piece. A zero-length interval
+ * is its position's bin alone.
+ */
+export declare function intervalsOf(table: FeatureTable, [startField, endField]: [string, string], size: number, bounds: Bounds): {
+    start: Float64Array<ArrayBuffer>;
+    end: Float64Array<ArrayBuffer>;
+    first: Float64Array<ArrayBuffer>;
+    last: Float64Array<ArrayBuffer>;
+    pieces: number;
+    lowBin: Float64Array<ArrayBuffer>;
+    highBin: Float64Array<ArrayBuffer>;
+};
+/**
+ * Rows a step made from nothing: an `aggregate`'s groups and a `coverage`'s
+ * runs, carrying only the fields they wrote and an id from their span, with
+ * no feature to hang from.
+ */
+export declare class MadeTable implements MadeRows {
+    readonly length: number;
+    private readonly columns;
+    private readonly tags;
+    constructor(length: number, columns: ReadonlyMap<string, Column>, tags: (i: number) => string);
+    column(field: string): Column;
+    id(i: number): string;
+    parentOfRow(): undefined;
+    row(i: number): Feature;
+    json(i: number): SimpleFeatureSerialized;
+}
+export declare function valueColumn(values: readonly unknown[]): Column;
+export declare function madeGroups({ groups, keys, ops, step, refNames, start, end, sectionOf, sections, }: {
+    groups: number;
+    keys: readonly [string, unknown[]][];
+    ops: readonly Column[];
+    step: AggregateStep;
+    refNames: unknown[];
+    start: Float64Array;
+    end: Float64Array;
+    sectionOf: ArrayLike<number>;
+    sections: number;
+}): Staged;
+export declare function isWeighted({ op, weight }: AggregateOp): boolean;
+/**
+ * One op's running sums per group, which the aggregate and the binned
+ * aggregate both add rows to in row order, so the two answer the same
+ * doubles. An unweighted row adds a weight of 1, which leaves every sum what
+ * it was before weights.
+ */
+export declare class OpSums {
+    readonly op: AggregateOp['op'];
+    private readonly readsValue;
+    readonly sum: Float64Array;
+    readonly weight: Float64Array;
+    readonly min: Float64Array;
+    readonly max: Float64Array;
+    constructor({ op, field }: AggregateOp, groups: number);
+    add(g: number, v: number, w: number): void;
+    column(groups: number): Column;
+}

package/esm/util/stepTables.js

@@ -0,0 +1,277 @@
+// The pieces of the transform steps that the fused kernels share with the
+// steps themselves. Imported by relative path only, so none of it is a
+// package subpath or part of the plugin ABI.
+import { aggregateFieldName } from './aggregateFieldName.js';
+import { DerivedTable, NO_COLUMN, TableRow, numberReaderOf, readerOf, valueAt, } from './featureTable.js';
+import { fieldReader, isPlainFieldRef } from './fieldReader.js';
+import { isJexl } from './jexlStrings.js';
+import { numericValue } from './numericValue.js';
+export const DEFAULT_BIN_AS = ['start', 'end'];
+/** What a `bin` over `fields` writes on each piece: the bases of its interval inside its bin. */
+export const BIN_OVERLAP_FIELD = 'overlap';
+// Section starts through a step whose rows came from its input's in order:
+// each section starts at the first row whose parent row is in it.
+export function boundsThrough(bounds, parentRow) {
+    const out = new Uint32Array(bounds.length);
+    let lo = 0;
+    for (let s = 0; s < bounds.length; s++) {
+        const edge = bounds[s];
+        let hi = parentRow.length;
+        while (lo < hi) {
+            const mid = (lo + hi) >>> 1;
+            if (parentRow[mid] < edge) {
+                lo = mid + 1;
+            }
+            else {
+                hi = mid;
+            }
+        }
+        out[s] = lo;
+    }
+    return out;
+}
+/**
+ * The entries a `flatten` fanned out, one row each over the container row it
+ * came from. A container kept with nothing to fan out is a row of its own
+ * that reads everything off itself.
+ */
+export class FannedTable extends DerivedTable {
+    get madeFrom() {
+        return true;
+    }
+    row(i) {
+        return this.isKept(i) ? this.parent.row(this.parentOf(i)) : super.row(i);
+    }
+}
+// A step reads a name or a dotted path. A plain name reads the table's column;
+// a path walks each row, chosen once per step. A computed field is a `formula`
+// step's to make.
+export function stepReader(table, ref, step) {
+    if (isJexl(ref)) {
+        throw new Error(`${step} field is a name or a dotted path, and a formula step in front computes one (${ref})`);
+    }
+    if (isPlainFieldRef(ref)) {
+        return readerOf(table.column(ref));
+    }
+    const read = fieldReader(ref, undefined);
+    return (i) => read(table.row(i));
+}
+export function stepNumberReader(table, ref, step) {
+    if (isPlainFieldRef(ref) && !isJexl(ref)) {
+        return numberReaderOf(table.column(ref));
+    }
+    const read = stepReader(table, ref, step);
+    return (i) => numericValue(read(i));
+}
+// A list holding one value is that value, as a VCF's ALT and INFO fields
+// arrive, and a longer one is its text: a Map keys a list by identity, which
+// made every row a group of its own. A missing value is one group, whether
+// the field is absent or holds a VCF's `.`.
+export function groupKey(value) {
+    return Array.isArray(value)
+        ? value.length === 1
+            ? (value[0] ?? undefined)
+            : value.join(',')
+        : (value ?? undefined);
+}
+export function binSize({ step }) {
+    if (!(step > 0)) {
+        throw new Error(`a bin step needs a positive size (${step})`);
+    }
+    return step;
+}
+/**
+ * Each row's interval and the bins it overlaps, `first` to `last`, with each
+ * section's lowest and highest bin: `first` NaN where the ends are not
+ * numbers or run backwards, so the row is no piece. A zero-length interval
+ * is its position's bin alone.
+ */
+export function intervalsOf(table, [startField, endField], size, bounds) {
+    const n = table.length;
+    const readStart = stepNumberReader(table, startField, 'a bin');
+    const readEnd = stepNumberReader(table, endField, 'a bin');
+    const start = new Float64Array(n);
+    const end = new Float64Array(n);
+    const firstBin = new Float64Array(n);
+    const lastBin = new Float64Array(n);
+    const sections = bounds.length - 1;
+    const lowBin = new Float64Array(sections).fill(Infinity);
+    const highBin = new Float64Array(sections).fill(-Infinity);
+    let pieces = 0;
+    for (let s = 0; s < sections; s++) {
+        let low = Infinity;
+        let high = -Infinity;
+        for (let i = bounds[s]; i < bounds[s + 1]; i++) {
+            const rowS = readStart(i);
+            const rowE = readEnd(i);
+            start[i] = rowS;
+            end[i] = rowE;
+            if (!(Number.isFinite(rowS) && Number.isFinite(rowE) && rowE >= rowS)) {
+                firstBin[i] = Number.NaN;
+                continue;
+            }
+            const first = Math.floor(rowS / size);
+            // The last bin starting before the end, found by the product the
+            // pieces are cut at.
+            let last = first;
+            while ((last + 1) * size < rowE) {
+                last++;
+            }
+            firstBin[i] = first;
+            lastBin[i] = last;
+            pieces += last - first + 1;
+            if (first < low) {
+                low = first;
+            }
+            if (last > high) {
+                high = last;
+            }
+        }
+        lowBin[s] = low;
+        highBin[s] = high;
+    }
+    return {
+        start,
+        end,
+        first: firstBin,
+        last: lastBin,
+        pieces,
+        lowBin,
+        highBin,
+    };
+}
+/**
+ * Rows a step made from nothing: an `aggregate`'s groups and a `coverage`'s
+ * runs, carrying only the fields they wrote and an id from their span, with
+ * no feature to hang from.
+ */
+export class MadeTable {
+    length;
+    columns;
+    tags;
+    constructor(length, columns, tags) {
+        this.length = length;
+        this.columns = columns;
+        this.tags = tags;
+    }
+    column(field) {
+        return field === 'uniqueId'
+            ? { kind: 'value', read: i => this.id(i) }
+            : (this.columns.get(field) ?? NO_COLUMN);
+    }
+    id(i) {
+        const at = (field) => String(valueAt(this.column(field), i));
+        return `${at('refName')}:${at('start')}-${at('end')}${this.tags(i)}`;
+    }
+    parentOfRow() {
+        return undefined;
+    }
+    row(i) {
+        return new TableRow(this, i);
+    }
+    json(i) {
+        const out = {};
+        for (const [field, column] of this.columns) {
+            out[field] = valueAt(column, i);
+        }
+        out.uniqueId = this.id(i);
+        return out;
+    }
+}
+export function valueColumn(values) {
+    return { kind: 'value', read: i => values[i] };
+}
+// An aggregate's groups as rows, section by section: the groupby fields, the
+// ops, and the reference name and extent each group spans.
+export function madeGroups({ groups, keys, ops, step, refNames, start, end, sectionOf, sections, }) {
+    const columns = new Map();
+    for (const [field, values] of keys) {
+        columns.set(field, valueColumn(values));
+    }
+    for (const [k, agg] of step.ops.entries()) {
+        columns.set(aggregateFieldName(agg), ops[k]);
+    }
+    columns.set('refName', valueColumn(refNames));
+    columns.set('start', { kind: 'number', values: start, at: undefined });
+    columns.set('end', { kind: 'number', values: end, at: undefined });
+    const serial = new Uint32Array(groups);
+    const out = new Uint32Array(sections);
+    let s = 0;
+    for (let g = 0; g < groups; g++) {
+        while (s < sectionOf[g]) {
+            s++;
+            out[s] = g;
+        }
+        serial[g] = g - out[s];
+    }
+    for (s++; s < sections; s++) {
+        out[s] = groups;
+    }
+    return {
+        table: new MadeTable(groups, columns, g => `#${serial[g]}`),
+        bounds: out,
+    };
+}
+// Whether an op reads its `weight`: a weight moves no minimum or maximum.
+export function isWeighted({ op, weight }) {
+    return weight !== undefined && op !== 'min' && op !== 'max';
+}
+/**
+ * One op's running sums per group, which the aggregate and the binned
+ * aggregate both add rows to in row order, so the two answer the same
+ * doubles. An unweighted row adds a weight of 1, which leaves every sum what
+ * it was before weights.
+ */
+export class OpSums {
+    op;
+    readsValue;
+    sum;
+    weight;
+    min;
+    max;
+    constructor({ op, field }, groups) {
+        if (op !== 'count' && field === undefined) {
+            throw new Error(`an aggregate ${op} needs a field`);
+        }
+        this.op = op;
+        this.readsValue = op !== 'count';
+        this.sum = new Float64Array(groups);
+        this.weight = new Float64Array(groups);
+        this.min = new Float64Array(groups).fill(Infinity);
+        this.max = new Float64Array(groups).fill(-Infinity);
+    }
+    add(g, v, w) {
+        if ((this.readsValue && !Number.isFinite(v)) || !Number.isFinite(w)) {
+            return;
+        }
+        this.weight[g] += w;
+        if (this.readsValue) {
+            this.sum[g] += v * w;
+            if (v < this.min[g]) {
+                this.min[g] = v;
+            }
+            if (v > this.max[g]) {
+                this.max[g] = v;
+            }
+        }
+    }
+    column(groups) {
+        const { op } = this;
+        if (op === 'count' || op === 'sum') {
+            const values = (op === 'count' ? this.weight : this.sum).slice(0, groups);
+            return { kind: 'number', values, at: undefined };
+        }
+        const values = new Float64Array(groups);
+        for (let g = 0; g < groups; g++) {
+            values[g] =
+                this.weight[g] === 0
+                    ? Number.NaN
+                    : op === 'mean'
+                        ? this.sum[g] / this.weight[g]
+                        : op === 'min'
+                            ? this.min[g]
+                            : this.max[g];
+        }
+        return { kind: 'number', values, at: undefined, nanIsAbsent: true };
+    }
+}