@gmod/bam/esm/htsget.js

Parser for BAM and BAM index (bai) files

import { unzip } from '@gmod/bgzf-filehandle';
import BamFile, { BAM_MAGIC } from "./bamFile.js";
import Chunk from "./chunk.js";
import { parseHeaderText } from "./sam.js";
import { appendInRange, concatUint8Array } from "./util.js";
import { VirtualOffset } from "./virtualOffset.js";
async function fetchOk(url, opts) {
    const res = await fetch(url, opts);
    if (!res.ok) {
        throw new Error(`HTTP ${res.status} fetching ${url}: ${await res.text()}`);
    }
    return res;
}
async function fetchChunk({ url, headers }, opts) {
    // pass base64 data URLs straight to fetch; otherwise apply headers (minus
    // referer, which isn't a permitted client-set header).
    // https://stackoverflow.com/a/54123275/2129219
    const { referer: _referer, ...rest } = headers ?? {};
    const res = url.startsWith('data:')
        ? await fetchOk(url)
        : await fetchOk(url, { ...opts, headers: rest });
    return new Uint8Array(await res.arrayBuffer());
}
async function fetchAndConcat(arr, opts) {
    // Pipeline unzip after each fetch so decompression overlaps later fetches.
    return concatUint8Array(await Promise.all(arr.map(async (c) => unzip(await fetchChunk(c, opts)))));
}
export default class HtsgetFile extends BamFile {
    baseUrl;
    trackId;
    constructor(args) {
        super({ htsget: true, recordClass: args.recordClass });
        this.baseUrl = args.baseUrl;
        this.trackId = args.trackId;
    }
    async getRecordsForRange(chr, min, max, opts) {
        const base = `${this.baseUrl}/${this.trackId}`;
        const url = `${base}?referenceName=${chr}&start=${min}&end=${max}&format=BAM`;
        const chrId = this.chrToIndex?.[chr];
        if (chrId === undefined) {
            return [];
        }
        const result = await fetchOk(url, opts);
        const data = await result.json();
        const uncba = await fetchAndConcat(data.htsget.urls.slice(1), {
            signal: opts?.signal,
        });
        const zero = new VirtualOffset(0, 0);
        const allRecords = await this.readBamFeatures(uncba, [], [], new Chunk(zero, zero, 0));
        return appendInRange(allRecords, chrId, min, max);
    }
    async getHeaderPre(opts = {}) {
        const url = `${this.baseUrl}/${this.trackId}?referenceName=na&class=header`;
        const result = await fetchOk(url, opts);
        const data = await result.json();
        const uncba = await fetchAndConcat(data.htsget.urls, {
            signal: opts.signal,
        });
        const dataView = new DataView(uncba.buffer);
        if (dataView.getInt32(0, true) !== BAM_MAGIC) {
            throw new Error('Not a BAM file');
        }
        const headLen = dataView.getInt32(4, true);
        const decoder = new TextDecoder();
        const headerText = decoder.decode(uncba.subarray(8, 8 + headLen));
        const samHeader = parseHeaderText(headerText);
        // use the @SQ lines in the header to figure out the
        // mapping between ref ref ID numbers and names
        const idToName = [];
        const nameToId = {};
        const sqLines = samHeader.filter(l => l.tag === 'SQ');
        for (const [refId, sqLine] of sqLines.entries()) {
            let refName = '';
            let length = 0;
            for (const item of sqLine.data) {
                if (item.tag === 'SN') {
                    refName = item.value;
                }
                else if (item.tag === 'LN') {
                    length = +item.value;
                }
            }
            nameToId[refName] = refId;
            idToName[refId] = { refName, length };
        }
        this.chrToIndex = nameToId;
        this.indexToChr = idToName;
        return samHeader;
    }
}
//# sourceMappingURL=htsget.js.map