sdf-parser/lib/iterator.js

SDF parser

import { parseString } from 'dynamic-typing';
import { MolfileStream } from "./MolfileStream.js";
/**
 * Asynchronously iterate over molecules from a text-decoded SDF stream.
 * @param readStream - A `ReadableStream<string>` supplying SDF text content.
 * @param options - Iterator options.
 * @yields {IteratorMolecule} Individual molecule objects.
 * @example
 * ```ts
 * import { openAsBlob } from 'node:fs';
 * import { iterator } from 'sdf-parser';
 *
 * const blob = await openAsBlob('compounds.sdf');
 * const textDecoder = new TextDecoderStream();
 * for await (const molecule of iterator(blob.stream().pipeThrough(textDecoder))) {
 *   console.log(molecule.molfile);
 * }
 * ```
 */
export async function* iterator(readStream, options = {}) {
    const { eol = '\n', mixedEOL, dynamicTyping = true } = options;
    const moleculeStream = readStream.pipeThrough(new MolfileStream({ mixedEOL }));
    const effectiveEol = mixedEOL !== false ? '\n' : eol;
    for await (const entry of moleculeStream) {
        const molecule = parseMolecule(entry, { eol: effectiveEol, dynamicTyping });
        if (!options.filter || options.filter(molecule)) {
            yield molecule;
        }
    }
}
function parseMolecule(sdfPart, options) {
    const { eol, dynamicTyping } = options;
    const parts = sdfPart.split(`${eol}>`);
    const molecule = {
        molfile: parts.length > 0 && parts[0].length > 5 ? parts[0] + eol : '',
    };
    for (let j = 1; j < parts.length; j++) {
        const lines = parts[j].split(eol);
        const from = lines[0].indexOf('<');
        const to = lines[0].indexOf('>');
        const label = lines[0].slice(from + 1, to);
        for (let k = 1; k < lines.length - 1; k++) {
            if (molecule[label]) {
                molecule[label] = `${molecule[label]}${eol}${lines[k]}`;
            }
            else {
                molecule[label] = lines[k];
            }
        }
        if (dynamicTyping) {
            molecule[label] = parseString(molecule[label]);
        }
    }
    return molecule;
}
//# sourceMappingURL=iterator.js.map