parse-xml/index.js

"use strict";

// FIXME: Document that this *may* not strip whitespace around text elements, and that it includes potentially-whitespace-only text elements between tags!

const Promise = require("bluebird");
const saxes = require("saxes");
const consumable = require("@joepie91/consumable");
const pEvent = require("p-event");
const debug = require("debug")("promistream:parse-xml");

const pipe = require("@promistream/pipe");
const map = require("@promistream/map");
const buffer = require("@promistream/buffer");
const lastWill = require("@promistream/last-will");

const { validateOptions, validateValue } = require("@validatem/core");
const required = require("@validatem/required");
const defaultTo = require("@validatem/default-to");
const isString = require("@validatem/is-string");
const allowExtraProperties = require("@validatem/allow-extra-properties");
const arrayOf = require("@validatem/array-of");

module.exports = function parseXML(_options) {
	let options = validateOptions(arguments, [
		defaultTo({}),
		allowExtraProperties({
			events: [ required, arrayOf(isString) ]
		})
	]);

	debug("Creating Saxes parser");
	let parser = new saxes.SaxesParser(options);
	let queue = consumable([]);
	let endEventPromise = pEvent(parser, "end");

	for (let eventName of options.events) {
		parser.on(eventName, (value) => {
			debug(`${eventName} event emitted by parser`);
			queue.peek().push({ type: eventName, value: value });
		});
	}

	return pipe([
		map((value) => {
			validateValue(value, [ isString ]); // FIXME: Turn this into a clearer error

			debug("Writing chunk to parser");
			parser.write(value);

			debug(`Returning ${queue.peek().length} parser items`);
			return queue.replace([]);
		}),
		lastWill({ onAllEnds: () => {
			return Promise.try(() => {
				parser.close();

				return endEventPromise;
			}).then(() => {
				debug("Parser finished processing");
				return queue.consume();
			});
		}}),
		buffer()
	]);
};