markdown2html/src/footnotes.ts

import { marked } from 'marked';
import { MarkdownOptions } from './render';

const footnotes = Symbol('footnotes');

// We're going to hang some extra data off of the lexer so we can reference it
// later when generating links back to references
declare module 'marked' {
	export interface Lexer {
		[footnotes]: Record<string, number>;
	}
}

export interface FootnoteLinkToken extends marked.Tokens.Generic {
	id: string;
	inst: number;
}

export function footnote_ref_ext(renderer: marked.Renderer, opts: MarkdownOptions) : marked.TokenizerExtension & marked.RendererExtension {
	return {
		name: 'footnote_ref',
		level: 'inline',
		start: (src) => src.match(/\[\^/)?.index,
		tokenizer(src, tokens) {
			const rule = /^\[\^([a-zA-Z0-9-\._, §]+)]/;
			const match = rule.exec(src);

			if (match) {
				const id = match[1];

				return {
					type: 'footnote_ref',
					raw: match[0],
					id: id,
					inst: next_cite_inst(this.lexer, id)
				};
			}
		},
		renderer(token: FootnoteLinkToken) {
			return `<sup id="cite:ref-${token.id}-${token.inst}"><a href="#cite:note-${token.id}">[${token.id}]</a></sup>`;
		}
	};
}

export interface FootnoteListToken extends marked.Tokens.Generic {
	text: string;
	items: FootnoteToken[];
}

export interface FootnoteToken extends marked.Tokens.Generic {
	id: string;
	text: string;
	inst_count() : number;
}

export function footnote_list_ext(renderer: marked.Renderer, opts: MarkdownOptions) : marked.TokenizerExtension & marked.RendererExtension {
	return {
		name: 'footnote_list',
		level: 'block',
		start: (src) => src.match(/^\[/)?.index,
		tokenizer(src, tokens) {
			const token: FootnoteListToken = {
				type: 'footnote_list',
				raw: '',
				text: '',
				items: [ ]
			};

			let remaining = src;
			const prefix_rule = /^\[\^([a-zA-Z0-9\., _§-]+)]:/;
			const whitespace_rule = /^\s*(?:\n|$)/;

			if (! prefix_rule.test(src)) {
				return null;
			}

			const items: { prefix: string, content: string }[] = [ ];
			let current: { prefix: string, content: string };

			function take(str: string) {
				token.raw += str;
				remaining = remaining.slice(str.length);
				return str;
			}

			line_loop:
			while (true) {
				const prefix_match = prefix_rule.exec(remaining);

				if (prefix_match) {
					take(prefix_match[0]);

					items.push(
						current = {
							prefix: prefix_match[1],
							content: '',
						}
					);

					continue line_loop;
				}

				const whitespace_match = whitespace_rule.exec(remaining);

				if (whitespace_match) {
					if (! take(whitespace_match[0])) {
						break line_loop;
					}

					current = null;
					continue line_loop;
				}

				if (current) {
					const next_newline = remaining.indexOf('\n');

					if (next_newline < 0) {
						current.content += take(remaining);
						break;
					}

					current.content += take(remaining.slice(0, next_newline + 1));
					continue line_loop;
				}

				break;
			}

			token.text = token.raw.trim();
			token.items = items.map(({ prefix, content }) => {
				const token = {
					type: 'footnote',
					raw: `[^${prefix}]:${content}`,
					id: prefix,
					text: content,
					tokens: [ ],
					inst_count: () => get_cite_inst_count(this.lexer, prefix)
				};

				this.lexer.inline(token.text, token.tokens); 
				return token;
			});

			return token;
		},
		renderer(token: FootnoteListToken) {
			const items = token.items.map((item) => (`
				<li role="doc-footnote" id="cite:note-${item.id}">
					<span class="cite-label">[${item.id}]</span>
					${footnote_link_backs(item.id, item.inst_count())}
					${this.parser.parseInline(item.tokens, renderer)}
				</li>
			`));

			return `<ul role="doc-endnotes">${items.join('')}</ul>`;
		}
	};
}

function get_cite_inst_count(lexer: marked.Lexer, id: string) {
	if (! lexer[footnotes]) {
		lexer[footnotes] = { };
	}
	
	return lexer[footnotes][id] || 0;
}

function next_cite_inst(lexer: marked.Lexer, id: string) {
	if (! lexer[footnotes]) {
		lexer[footnotes] = { };
	}

	if (! lexer[footnotes][id]) {
		lexer[footnotes][id] = 0;
	}

	return ++lexer[footnotes][id];
}

const letters = 'abcdefghijklmnopqrstuvwxyz';

function footnote_link_backs(id: string, count: number) {
	if (! count) {
		return '';
	}

	if (count === 1) {
		return `<sup><a href="#cite:ref-${id}-1" title="Back to reference">^</a></sup>`;
	}

	// note: We're using letters for link backs; If we run out, only
	// show the first 26 references
	// todo: A more ideal solution would be to start using double-letters,
	// like "aa", "ab", "ac", etc. after running out of single letter options
	count = Math.min(count, 26);

	const links: string[] = [ ];

	for (let i = 0; i < count; i++) {
		const letter = letters[i];
		links[i] = `<a href="#cite:ref-${id}-${i + 1}" title="Back to reference ${letter}">${letter}</a>`;
	}

	return `<sup>^ ${links.join(' ')}</sup>`;
}