Sharkey/src/mfm/parser.ts

import * as P from 'parsimmon';
import parseAcct from '../misc/acct/parse';
import { toUnicode } from 'punycode';
import { takeWhile, cumulativeSum } from '../prelude/array';
import { Tree } from '../prelude/tree';
import * as T from '../prelude/tree';

const emojiRegex = /((?:\ud83d[\udc68\udc69])(?:\ud83c[\udffb-\udfff])?\u200d(?:\u2695\ufe0f|\u2696\ufe0f|\u2708\ufe0f|\ud83c[\udf3e\udf73\udf93\udfa4\udfa8\udfeb\udfed]|\ud83d[\udcbb\udcbc\udd27\udd2c\ude80\ude92]|\ud83e[\uddb0-\uddb3])|(?:\ud83c[\udfcb\udfcc]|\ud83d[\udd74\udd75]|\u26f9)((?:\ud83c[\udffb-\udfff]|\ufe0f)\u200d[\u2640\u2642]\ufe0f)|(?:\ud83c[\udfc3\udfc4\udfca]|\ud83d[\udc6e\udc71\udc73\udc77\udc81\udc82\udc86\udc87\ude45-\ude47\ude4b\ude4d\ude4e\udea3\udeb4-\udeb6]|\ud83e[\udd26\udd35\udd37-\udd39\udd3d\udd3e\uddb8\uddb9\uddd6-\udddd])(?:\ud83c[\udffb-\udfff])?\u200d[\u2640\u2642]\ufe0f|(?:\ud83d\udc68\u200d\u2764\ufe0f\u200d\ud83d\udc8b\u200d\ud83d\udc68|\ud83d\udc68\u200d\ud83d\udc68\u200d\ud83d\udc66\u200d\ud83d\udc66|\ud83d\udc68\u200d\ud83d\udc68\u200d\ud83d\udc67\u200d\ud83d[\udc66\udc67]|\ud83d\udc68\u200d\ud83d\udc69\u200d\ud83d\udc66\u200d\ud83d\udc66|\ud83d\udc68\u200d\ud83d\udc69\u200d\ud83d\udc67\u200d\ud83d[\udc66\udc67]|\ud83d\udc69\u200d\u2764\ufe0f\u200d\ud83d\udc8b\u200d\ud83d[\udc68\udc69]|\ud83d\udc69\u200d\ud83d\udc69\u200d\ud83d\udc66\u200d\ud83d\udc66|\ud83d\udc69\u200d\ud83d\udc69\u200d\ud83d\udc67\u200d\ud83d[\udc66\udc67]|\ud83d\udc68\u200d\u2764\ufe0f\u200d\ud83d\udc68|\ud83d\udc68\u200d\ud83d\udc66\u200d\ud83d\udc66|\ud83d\udc68\u200d\ud83d\udc67\u200d\ud83d[\udc66\udc67]|\ud83d\udc68\u200d\ud83d\udc68\u200d\ud83d[\udc66\udc67]|\ud83d\udc68\u200d\ud83d\udc69\u200d\ud83d[\udc66\udc67]|\ud83d\udc69\u200d\u2764\ufe0f\u200d\ud83d[\udc68\udc69]|\ud83d\udc69\u200d\ud83d\udc66\u200d\ud83d\udc66|\ud83d\udc69\u200d\ud83d\udc67\u200d\ud83d[\udc66\udc67]|\ud83d\udc69\u200d\ud83d\udc69\u200d\ud83d[\udc66\udc67]|\ud83c\udff3\ufe0f\u200d\ud83c\udf08|\ud83c\udff4\u200d\u2620\ufe0f|\ud83d\udc41\u200d\ud83d\udde8|\ud83d\udc68\u200d\ud83d[\udc66\udc67]|\ud83d\udc69\u200d\ud83d[\udc66\udc67]|\ud83d\udc6f\u200d\u2640\ufe0f|\ud83d\udc6f\u200d\u2642\ufe0f|\ud83e\udd3c\u200d\u2640\ufe0f|\ud83e\udd3c\u200d\u2642\ufe0f|\ud83e\uddde\u200d\u2640\ufe0f|\ud83e\uddde\u200d\u2642\ufe0f|\ud83e\udddf\u200d\u2640\ufe0f|\ud83e\udddf\u200d\u2642\ufe0f)|[\u0023\u002a\u0030-\u0039]\ufe0f?\u20e3|(?:[\u00a9\u00ae\u2122\u265f]\ufe0f)|(?:\ud83c[\udc04\udd70\udd71\udd7e\udd7f\ude02\ude1a\ude2f\ude37\udf21\udf24-\udf2c\udf36\udf7d\udf96\udf97\udf99-\udf9b\udf9e\udf9f\udfcd\udfce\udfd4-\udfdf\udff3\udff5\udff7]|\ud83d[\udc3f\udc41\udcfd\udd49\udd4a\udd6f\udd70\udd73\udd76-\udd79\udd87\udd8a-\udd8d\udda5\udda8\uddb1\uddb2\uddbc\uddc2-\uddc4\uddd1-\uddd3\udddc-\uddde\udde1\udde3\udde8\uddef\uddf3\uddfa\udecb\udecd-\udecf\udee0-\udee5\udee9\udef0\udef3]|[\u203c\u2049\u2139\u2194-\u2199\u21a9\u21aa\u231a\u231b\u2328\u23cf\u23ed-\u23ef\u23f1\u23f2\u23f8-\u23fa\u24c2\u25aa\u25ab\u25b6\u25c0\u25fb-\u25fe\u2600-\u2604\u260e\u2611\u2614\u2615\u2618\u2620\u2622\u2623\u2626\u262a\u262e\u262f\u2638-\u263a\u2640\u2642\u2648-\u2653\u2660\u2663\u2665\u2666\u2668\u267b\u267f\u2692-\u2697\u2699\u269b\u269c\u26a0\u26a1\u26aa\u26ab\u26b0\u26b1\u26bd\u26be\u26c4\u26c5\u26c8\u26cf\u26d1\u26d3\u26d4\u26e9\u26ea\u26f0-\u26f5\u26f8\u26fa\u26fd\u2702\u2708\u2709\u270f\u2712\u2714\u2716\u271d\u2721\u2733\u2734\u2744\u2747\u2757\u2763\u2764\u27a1\u2934\u2935\u2b05-\u2b07\u2b1b\u2b1c\u2b50\u2b55\u3030\u303d\u3297\u3299])(?:\ufe0f|(?!\ufe0e))|(?:(?:\ud83c[\udfcb\udfcc]|\ud83d[\udd74\udd75\udd90]|[\u261d\u26f7\u26f9\u270c\u270d])(?:\ufe0f|(?!\ufe0e))|(?:\ud83c[\udf85\udfc2-\udfc4\udfc7\udfca]|\ud83d[\udc42\udc43\udc46-\udc50\udc66-\udc69\udc6e\udc70-\udc78\udc7c\udc81-\udc83\udc85-\udc87\udcaa\udd7a\udd95\udd96\ude45-\ude47\ude4b-\ude4f\udea3\udeb4-\udeb6\udec0\udecc]|\ud83e[\udd18-\udd1c\udd1e\udd1f\udd26\udd30-\udd39\udd3d\udd3e\uddb5\uddb6\uddb8\uddb9\uddd1-\udddd]|[\u270a\u270b]))(?:\ud83c[\udffb-\udfff])?|(?:\ud83c\udff4\udb40\udc67\udb40\udc62\udb40\udc65\udb40\udc6e\udb40\udc67\udb40\udc7f|\ud83c\udff4\udb40\udc67\udb40\udc62\udb40\udc73\udb40\udc63\udb40\udc74\udb40\udc7f|\ud83c\udff4\udb40\udc67\udb40\udc62\udb40\udc77\udb40\udc6c\udb40\udc73\udb40\udc7f|\ud83c\udde6\ud83c[\udde8-\uddec\uddee\uddf1\uddf2\uddf4\uddf6-\uddfa\uddfc\uddfd\uddff

type Node<T, P> = { type: T, props: P };

export type MentionNode = Node<'mention', {
	canonical: string,
	username: string,
	host: string,
	acct: string
}>;

export type HashtagNode = Node<'hashtag', {
	hashtag: string
}>;

export type EmojiNode = Node<'emoji', {
	name: string
}>;

export type MfmNode =
	MentionNode |
	HashtagNode |
	EmojiNode |
	Node<string, any>;

export type MfmTree = Tree<MfmNode>;

export type MfmForest = MfmTree[];

export function createLeaf(type: string, props: any): MfmTree {
	return T.createLeaf({ type, props });
}

export function createTree(type: string, children: MfmForest, props: any): MfmTree {
	return T.createTree({ type, props }, children);
}

export function removeOrphanedBrackets(s: string): string {
	const openBrackets = ['(', '「'];
	const closeBrackets = [')', '」'];
	const xs = cumulativeSum(s.split('').map(c => {
		if (openBrackets.includes(c)) return 1;
		if (closeBrackets.includes(c)) return -1;
		return 0;
	}));
	const firstOrphanedCloseBracket = xs.findIndex(x => x < 0);
	if (firstOrphanedCloseBracket !== -1) return s.substr(0, firstOrphanedCloseBracket);
	const lastMatched = xs.lastIndexOf(0);
	return s.substr(0, lastMatched + 1);
}

const newline = P((input, i) => {
	if (i == 0 || input[i] == '\n' || input[i - 1] == '\n') {
		return P.makeSuccess(i, null);
	} else {
		return P.makeFailure(i, 'not newline');
	}
});

export const plainParser = P.createLanguage({
	root: r => P.alt(
		r.emoji,
		r.text
	).atLeast(1),

	text: () => P.any.map(x => createLeaf('text', { text: x })),

	//#region Emoji
	emoji: r =>
		P.alt(
			P.regexp(/:([a-z0-9_+-]+):/i, 1)
			.map(x => createLeaf('emoji', {
				name: x
			})),
			P.regexp(emojiRegex)
			.map(x => createLeaf('emoji', {
				emoji: x
			})),
		),
	//#endregion
});

const mfm = P.createLanguage({
	root: r => P.alt(
		r.big,
		r.small,
		r.bold,
		r.strike,
		r.italic,
		r.motion,
		r.url,
		r.link,
		r.mention,
		r.hashtag,
		r.emoji,
		r.blockCode,
		r.inlineCode,
		r.quote,
		r.math,
		r.search,
		r.title,
		r.center,
		r.text
	).atLeast(1),

	text: () => P.any.map(x => createLeaf('text', { text: x })),

	//#region Big
	big: r =>
		P.regexp(/^\*\*\*([\s\S]+?)\*\*\*/, 1)
		.map(x => createTree('big', P.alt(
			r.strike,
			r.italic,
			r.mention,
			r.hashtag,
			r.emoji,
			r.math,
			r.text
		).atLeast(1).tryParse(x), {})),
	//#endregion

	//#region Small
	small: r =>
		P.regexp(/<small>([\s\S]+?)<\/small>/, 1)
		.map(x => createTree('small', P.alt(
			r.strike,
			r.italic,
			r.mention,
			r.hashtag,
			r.emoji,
			r.math,
			r.text
		).atLeast(1).tryParse(x), {})),
	//#endregion

	//#region Block code
	blockCode: r =>
		newline.then(
			P((input, i) => {
				const text = input.substr(i);
				const match = text.match(/^```(.+?)?\n([\s\S]+?)\n```(\n|$)/i);
				if (!match) return P.makeFailure(i, 'not a blockCode');
				return P.makeSuccess(i + match[0].length, createLeaf('blockCode', { code: match[2], lang: match[1] ? match[1].trim() : null }));
			})
		),
	//#endregion

	//#region Bold
	bold: r =>
		P.alt(P.regexp(/\*\*([\s\S]+?)\*\*/, 1), P.regexp(/__([a-zA-Z0-9\s]+?)__/, 1))
		.map(x => createTree('bold', P.alt(
			r.strike,
			r.italic,
			r.mention,
			r.hashtag,
			r.url,
			r.link,
			r.emoji,
			r.text
		).atLeast(1).tryParse(x), {})),
	//#endregion

	//#region Center
	center: r =>
		P.regexp(/<center>([\s\S]+?)<\/center>/, 1)
		.map(x => createTree('center', P.alt(
			r.big,
			r.small,
			r.bold,
			r.strike,
			r.italic,
			r.motion,
			r.mention,
			r.hashtag,
			r.emoji,
			r.math,
			r.url,
			r.link,
			r.text
		).atLeast(1).tryParse(x), {})),
	//#endregion

	//#region Emoji
	emoji: r =>
		P.alt(
			P.regexp(/:([a-z0-9_+-]+):/i, 1)
			.map(x => createLeaf('emoji', {
				name: x
			})),
			P.regexp(emojiRegex)
			.map(x => createLeaf('emoji', {
				emoji: x
			})),
		),
	//#endregion

	//#region Hashtag
	hashtag: r =>
		P((input, i) => {
			const text = input.substr(i);
			const match = text.match(/^#([^\s\.,!\?'"#:]+)/i);
			if (!match) return P.makeFailure(i, 'not a hashtag');
			let hashtag = match[1];
			hashtag = removeOrphanedBrackets(hashtag);
			if (hashtag.match(/^[0-9]+$/)) return P.makeFailure(i, 'not a hashtag');
			if (input[i - 1] != null && input[i - 1].match(/[a-z0-9]/i)) return P.makeFailure(i, 'not a hashtag');
			if (hashtag.length > 50) return P.makeFailure(i, 'not a hashtag');
			return P.makeSuccess(i + ('#' + hashtag).length, createLeaf('hashtag', { hashtag: hashtag }));
		}),
	//#endregion

	//#region Inline code
	inlineCode: r =>
		P.regexp(/`([^´\n]+?)`/, 1)
		.map(x => createLeaf('inlineCode', { code: x })),
	//#endregion

	//#region Italic
	italic: r =>
		P.alt(P.regexp(/<i>([\s\S]+?)<\/i>/, 1), P.regexp(/(\*|_)([a-zA-Z0-9]+?[\s\S]*?)\1/, 2))
		.map(x => createTree('italic', P.alt(
			r.bold,
			r.strike,
			r.mention,
			r.hashtag,
			r.url,
			r.link,
			r.emoji,
			r.text
		).atLeast(1).tryParse(x), {})),
	//#endregion

	//#region Link
	link: r =>
		P.seqObj(
			['silent', P.string('?').fallback(null).map(x => x != null)] as any,
			P.string('['),
			['text', P.regexp(/[^\n\[\]]+/)] as any,
			P.string(']'),
			P.string('('),
			['url', r.url] as any,
			P.string(')'),
		)
		.map((x: any) => {
			return createTree('link', P.alt(
				r.big,
				r.small,
				r.bold,
				r.strike,
				r.italic,
				r.motion,
				r.emoji,
				r.text
			).atLeast(1).tryParse(x.text), {
				silent: x.silent,
				url: x.url.node.props.url
			});
		}),
	//#endregion

	//#region Math
	math: r =>
		P.regexp(/\\\((.+?)\\\)/, 1)
		.map(x => createLeaf('math', { formula: x })),
	//#endregion

	//#region Mention
	mention: r =>
		P((input, i) => {
			const text = input.substr(i);
			const match = text.match(/^@\w([\w-]*\w)?(?:@[\w\.\-]+\w)?/);
			if (!match) return P.makeFailure(i, 'not a mention');
			if (input[i - 1] != null && input[i - 1].match(/[a-z0-9]/i)) return P.makeFailure(i, 'not a mention');
			return P.makeSuccess(i + match[0].length, match[0]);
		})
		.map(x => {
			const { username, host } = parseAcct(x.substr(1));
			const canonical = host != null ? `@${username}@${toUnicode(host)}` : x;
			return createLeaf('mention', {
				canonical, username, host, acct: x
			});
		}),
	//#endregion

	//#region Motion
	motion: r =>
		P.alt(P.regexp(/\(\(\(([\s\S]+?)\)\)\)/, 1), P.regexp(/<motion>(.+?)<\/motion>/, 1))
		.map(x => createTree('motion', P.alt(
			r.bold,
			r.small,
			r.strike,
			r.italic,
			r.mention,
			r.hashtag,
			r.emoji,
			r.url,
			r.link,
			r.math,
			r.text
		).atLeast(1).tryParse(x), {})),
	//#endregion

	//#region Quote
	quote: r =>
		newline.then(P((input, i) => {
			const text = input.substr(i);
			if (!text.match(/^>[\s\S]+?/)) return P.makeFailure(i, 'not a quote');
			const quote = takeWhile(line => line.startsWith('>'), text.split('\n'));
			const qInner = quote.join('\n').replace(/^>/gm, '').replace(/^ /gm, '');
			if (qInner == '') return P.makeFailure(i, 'not a quote');
			const contents = r.root.tryParse(qInner);
			return P.makeSuccess(i + quote.join('\n').length + 1, createTree('quote', contents, {}));
		})),
	//#endregion

	//#region Search
	search: r =>
		newline.then(P((input, i) => {
			const text = input.substr(i);
			const match = text.match(/^(.+?)( |　)(検索|\[検索\]|Search|\[Search\])(\n|$)/i);
			if (!match) return P.makeFailure(i, 'not a search');
			return P.makeSuccess(i + match[0].length, createLeaf('search', { query: match[1], content: match[0].trim() }));
		})),
	//#endregion

	//#region Strike
	strike: r =>
		P.regexp(/~~(.+?)~~/, 1)
		.map(x => createTree('strike', P.alt(
			r.bold,
			r.italic,
			r.mention,
			r.hashtag,
			r.url,
			r.link,
			r.emoji,
			r.text
		).atLeast(1).tryParse(x), {})),
	//#endregion

	//#region Title
	title: r =>
		newline.then(P((input, i) => {
			const text = input.substr(i);
			const match = text.match(/^((【|\[)(.+?)(】|]))(\n|$)/);
			if (!match) return P.makeFailure(i, 'not a title');
			const q = match[1].trim().substring(1, match[1].length - 1);
			const contents = P.alt(
				r.big,
				r.small,
				r.bold,
				r.strike,
				r.italic,
				r.motion,
				r.url,
				r.link,
				r.mention,
				r.hashtag,
				r.emoji,
				r.inlineCode,
				r.text
			).atLeast(1).tryParse(q);
			return P.makeSuccess(i + match[0].length, createTree('title', contents, {}));
		})),
	//#endregion

	//#region URL
	url: r =>
		P((input, i) => {
			const text = input.substr(i);
			const match = text.match(/^https?:\/\/[\w\/:%#@\$&\?!\(\)\[\]~\.,=\+\-]+/);
			if (!match) return P.makeFailure(i, 'not a url');
			let url = match[0];
			url = removeOrphanedBrackets(url);
			if (url.endsWith('.')) url = url.substr(0, url.lastIndexOf('.'));
			if (url.endsWith(',')) url = url.substr(0, url.lastIndexOf(','));
			return P.makeSuccess(i + url.length, url);
		})
		.map(x => createLeaf('url', { url: x })),
	//#endregion
});

export default mfm;