linguist-js/dist/input/fromRawContent.js

Analyse the programming languages used in a folder or from raw content, using the same rules that GitHub Linguist does.

import { getFileExtension } from '../program/fs/normalisedPath.js';
export default function fromRawContent(contents, vendorPaths = []) {
    return Object.entries(contents).map(([path, content]) => {
        const extension = getFileExtension(path);
        const metadata = {};
        if (vendorPaths.some((pathPtn) => RegExp(pathPtn, 'i').test(path))) {
            metadata.vendored = true;
        }
        return {
            path,
            content,
            firstLine: content.split(/\r?\n/)[0] ?? '',
            size: Buffer.byteLength(content, 'utf-8'),
            extension,
            isBinary: false,
            metadata: Object.keys(metadata).length ? metadata : undefined,
        };
    });
}