HTML to SXML

html2sxml.js

#!/usr/bin/env node

/* USAGE:
 * html2sxml /path/to/index.html > output.sxml
 * html2sxml < input.html > output.sxml
 */

'use strict';

const path = require("path");
const { PassThrough } = require("stream");

const htmlparser = require("htmlparser2");

const html2sxml = input => {

	const output = new PassThrough();
	output.write(";generated by html2sxml\n")
	output.write(";https://github.com/aditi/util\n");

	let indentLevel = 0;

	const parser = new htmlparser.Parser({

		onopentag (name, attr) {

			output.write("\n" + "\t".repeat(indentLevel++) + `(${name}`);
			let attrKeys = Object.keys(attr);
			if (attrKeys.length) {

				let selector = "";
				if (attr.id) selector += "#" + attr.id.split(" ").join("#");
				if (attr.class) selector += "." + attr.class.split(" ").join(".");
				if (selector) output.write(selector);

				attrKeys = attrKeys.filter(name => ![ "class", "id" ].includes(name));
				if (attrKeys.length) {

					output.write(" (@ ");
					attrKeys.forEach(name => output.write(`(${name} "${attr[name]}") `));
					output.write(")");

				}
			}

		},

		oncomment (comments) {

			(comments = comments.trim())
				&& comments.split("\n")
					.forEach(comment =>
						output.write(
							"\n; " + "\t".repeat(indentLevel) + comment.trim()
							// Strip whitespace and prepend ; to every line of comment
						)
					);

				},

		ontext (text) {

			text.trim() && output.write(
				` "${text.replace(/"/g, String.raw`\"`)}"`
			);

		},

		onclosetag () {

			output.write(")");
			indentLevel--;

		},

		onend () {

			output.write("\n");

		}

	}, { decodeEntities: true });

	input.pipe(parser);
	return output;

};

if (require.main === module) {

	if (process.argv[2]) {

		if (process.argv[2].match(/.*htm(|l)$/)) {

			// mode = 'file'

			let pathToFile = process.argv[2];
			if (!path.isAbsolute(pathToFile)) {
				pathToFile = path.resolve(`${process.cwd()}/${process.argv[2]}`);
			}

			html2sxml(require("fs").createReadStream(pathToFile, "utf8"))
				.pipe(process.stdout);

		} else {

			console.error("Path not a HTML file");
			process.exit(1);

		}

	} else {

		// mode = 'stdin'
		html2sxml(process.stdin).pipe(process.stdout);

	}

}

module.exports = html2sxml;

Note: The SXML output produced by this module is not fully compliant with SXML spec[1][2]. It is in fact a variant of SXML used as a DSL in the aditi blog engine. I might name this variant ASX.