www-lib-markdown/js/markdown.js

// Blocks
// - Paragraph
// - Header 1-6   # ## ### #### ##### ######  or   ===  ---
// - Blockquote (nestable)   >
// - Unordered list (nestable)   *_
// - Ordered list (nestable)    1._
// - Code block   ```\ncode\n```  or  4 spaces/tab indent
// - Horizontal rule   ---    - - -    * * *  etc
// - Table   -|-
// - Definition list     term\n: definition\n: alternate definition
// - Footnote (bottom)   citation[^1]
// - Abbreviation (definition)   *[ABC]: Abbrev Blah Cat
// Inline
// - Link   [text](https://url)
// - Emphasis   *emphasized*
// - Strong   **bold**
// - Inline code   `code`
// - Strikethrough   ~strike~
// - Image   ![alt text](https://image){.cssclass}
// - Footnote (inline)   [^1]: footnote text
// - Abbreviation (inline)

class _MDBlock {
	toHTML(config) {
		throw new Error(self.constructor.name + ".toHTML not implemented");
	}

	/**
	 * @param {_MDBlock[]} blocks
	 * @returns {String}
	 */
	static toHTML(blocks, config) {
		return blocks.map((block) => block.toHTML(config)).join("\n");
	}
}

class _MDMultiBlock extends _MDBlock {
	/** @var {_MDBlock[]} */
	#blocks;
	/**
	 * @param {_MDBlock[]} blocks
	 */
	constructor(blocks) {
		super();
		this.#blocks = blocks;
	}
	toHTML(config) {
		return _MDBlock.toHTML(this.#blocks, config);
	}
}

class _MDParagraph extends _MDBlock {
	/** @var {_MDBlock} */
	content;

	/**
	 * @param {_MDBlock} content
	 */
	constructor(content) {
		super();
		this.content = content;
	}

	toHTML(config) {
		let contentHTML = this.content.toHTML(config);
		return `<p>${contentHTML}</p>\n`;
	}
}

class _MDHeader extends _MDBlock {
	/** @var {number} */
	level;
	/** @var {_MDBlock} */
	content;

	/**
	 * @param {number} level
	 * @param {_MDBlock} content
	 */
	constructor(level, content) {
		super();
		this.level = level;
		this.content = content;
	}

	toHTML(config) {
		let contentHTML = this.content.toHTML(config);
		return `<h${this.level}>${contentHTML}</h${this.level}>\n`;
	}
}

class _MDBlockquote extends _MDBlock {
	/** @var {_MDBlock[]} */
	content;
	/**
	 * @param {_MDBlock[]} content
	 */
	constructor(content) {
		super();
		this.content = content;
	}
	toHTML(config) {
		let contentHTML = _MDBlock.toHTML(this.content, config);
		return `<blockquote>\n${contentHTML}\n</blockquote>`;
	}
}

class _MDUnorderedList extends _MDBlock {
	/** @var {_MDListItem[]} */
	items;
	/**
	 * @param {_MDListItem[]} items
	 */
	constructor(items) {
		super();
		this.items = items;
	}
	toHTML(config) {
		let contentHTML = _MDBlock.toHTML(this.items);
		return `<ul>\n${contentHTML}\n</ul>`;
	}
}

class _MDOrderedList extends _MDBlock {
	/** @var {_MDListItem[]} */
	items;
	/**
	 * @param {_MDListItem[]} items
	 */
	constructor(items) {
		super();
		this.items = items;
	}
	toHTML(config) {
		let contentHTML = _MDBlock.toHTML(this.items);
		return `<ol>\n${contentHTML}\n</ol>`;
	}
}

class _MDListItem extends _MDBlock {
	/** @var {_MDBlock} */
	content;

	/**
	 * @param {_MDBlock} content
	 */
	constructor(content) {
		super();
		this.content = content;
	}

	toHTML(config) {
		let contentHTML = this.content.toHTML(config);
		return `<li>${contentHTML}</li>`;
	}
}

class _MDCodeBlock extends _MDBlock {
	/** @var {String} */
	#code;
	/**
	 * @param {String} code
	 */
	constructor(code) {
		super();
		this.#code = code;
	}
	toHTML(config) {
		return `<pre><code>${this.#code}</code></pre>`;
	}
}

class _MDHorizontalRule extends _MDBlock {
	toHTML(config) {
		return "<hr>\n";
	}
}

class _MDTableHeaderCell extends _MDBlock {
	/** @var {_MDBlock} */
	#content;
	/**
	 * @param {_MDBlock} content
	 */
	constructor(content) {
		super();
		this.#content = content;
	}
	toHTML(config) {
		let contentHTML = this.#content.toHTML(config);
		return `<th>${contentHTML}</th>`;
	}
}

class _MDTableCell extends _MDBlock {
	/** @var {_MDBlock} */
	#content;
	/**
	 * @param {_MDBlock} content
	 */
	constructor(content) {
		super();
		this.#content = content;
	}
	toHTML(config) {
		let contentHTML = this.#content.toHTML(config);
		return `<th>${contentHTML}</th>`;
	}
}

class _MDTableRow extends _MDBlock {
	/** @var {_MDTableCell[]|_MDTableHeaderCell[]} */
	#cells;
	/**
	 * @param {_MDTableCell[]|_MDTableHeaderCell[]} cells
	 */
	constructor(cells) {
		super();
		this.#cells = cells;
	}
	toHTML(config) {
		cellsHTML = _MDBlock.toHTML(this.#cells, config);
		return `<tr>\n${cellsHTML}\n</tr>`;
	}
}

class _MDTable extends _MDBlock {
	/** @var {_MDTableRow} */
	#headerRow;
	/** @var {_MDTableRow[]} */
	#bodyRows;
	toHTML(config) {
		let headerRowHTML = this.#headerRow.toHTML(config);
		let bodyRowsHTML = _MDBlock.toHTML(this.#bodyRows);
		return `<table>\n<thead>\n${headerRowHTML}\n</thead>\n<tbody>\n${bodyRowsHTML}\n</tbody>\n</table>`;
	}
}

class _MDDefinitionList extends _MDBlock {
	/** @var {_MDBlock[]} */
	#content;
	/**
	 * @param {_MDBlock[]} content
	 */
	constructor(content) {
		super();
		this.#content = content;
	}
	toHTML(config) {
		let contentHTML = _MDBlock.toHTML(this.#content);
		return `<dl>\n${contentHTML}\n</dl>`;
	}
}

class _MDDefinitionTerm extends _MDBlock {
	/** @var {_MDBlock} */
	#content;
	/**
	 * @param {_MDBlock} content
	 */
	constructor(content) {
		super();
		this.#content = content;
	}
	toHTML(config) {
		let contentHTML = this.#content.toHTML(config);
		return `<dt>${contentHTML}</dt>`;
	}
}

class _MDDefinitionDefinition extends _MDBlock {
	/** @var {_MDBlock} */
	#content;
	/**
	 * @param {_MDBlock} content
	 */
	constructor(content) {
		super();
		this.#content = content;
	}
	toHTML(config) {
		let contentHTML = this.#content.toHTML(config);
		return `<dd>${contentHTML}</dd>`;
	}
}

class _MDFootnoteReference extends _MDBlock {
	/** @var {String} */
	#id;
	/**
	 * @param {String} id
	 */
	constructor(id) {
		super();
		this.#id = id;
	}
	toHTML(config) {
		return `<sup><a href="#footnote${this.#id}">${this.#id}</a></sup>`;
	}
}

class _MDFootnoteContent extends _MDBlock {
	/** @var {String} */
	#id;
	/** @var {_MDBlock} */
	#content;
	/**
	 * @param {String} id
	 * @param {_MDBlock} content
	 */
	constructor(id, content) {
		super();
		this.#id = id;
		this.#content = content;
	}
	toHTML(config) {
		// TODO: Forward and back links
		// TODO: Deferring footnotes to end of document
		//<ol>
		//<li id="fn:1" role="doc-endnote">
		//<p>Footnote&nbsp;<a href="#fnref:1" class="footnote-backref" role="doc-backlink">↩︎</a></p>
		//</li>
		//</ol>
		return '';
	}
}

class _MDAbbreviationOccurrence extends _MDBlock {
	/** @var {String} */
	#label;
	/** @var {String} */
	#definition;

	/**
	 * @param {String} label
	 * @param {String} definition
	 */
	constructor(label, definition) {
		super();
		this.#label = label;
		this.#definition = definition;
	}

	toHTML(config) {
		return `<abbr title="${this.#definition.replace('"', '&quot;')}">${this.#label}</abbr>`;
	}
}

class _MDInline extends _MDBlock {
	/** @var {String} */
	#raw;
	/**
	 * @param {String} raw
	 */
	constructor(raw) {
		super();
		this.#raw = raw;
	}

	toHTML(config) {
		return this.#raw;
	}
}

// Blocks that immediately start a new block
// - Headers
// - Blockquote
// - Code block   ```\ncode\n```
// Blocks that need blank line first
// - HR   ---   - - -   ***   * * * * * *
// - Lists
// - Table
// - Code block   [4+spaces]code
// - Definition list   term\n: definition\n: alternate def
// Unknown blocks
// - Footnotes   some text[^1]   [^1]: first footnote content
// - Abbreviations   *[HTML]: Hyper Text
// Inline styles
// - Links
// - Italic
// - Bold
// - `code`
// - Strikethrough
// - Images   ![alt text](url){.cssclass}
// - Literals   \*
class _MDState {
	/** @var {String[]} */
	lines = [];

	/** @var {object} */
	abbreviations = {};

	/** @var {object} */
	footnotes = {};

	/** @var {number} */
	p = 0;

	copy() {
		let cp = new _MDState();
		cp.abbreviations = this.abbreviations;
		cp.footnotes = this.footnotes;
		cp.p = this.p;
		return cp;
	}

	/** @param {_MDState} other */
	apply(other) {
		this.abbreviations = other.abbreviations;
		this.footnotes = other.footnotes;
		this.p = other.p;
	}

	hasLines(minCount) {
		return this.p + minCount <= this.lines.length;
	}
}

class MDConfig {

}

class Markdown {
	/**
	 * @param {String} line
	 */
	static #stripIndent(line) {
		return line.replace(/^(?: {1,4}|\t)/, '');
	}

	/**
	 * @param {_MDState} state
	 * @returns {_MDBlock[]}
	 */
	static #readBlocks(state) {
		var blocks = [];
		while (state.hasLines(1)) {
			let block = this.#readNextBlock(state);
			if (block) {
				blocks.push(block);
			} else {
				break;
			}
		}
		return blocks;
	}

	/**
	 * @param {_MDState} state
	 * @returns {_MDBlock}
	 */
	static #readNextBlock(state) {
		while (state.hasLines(1) && state.lines[state.p].trim().length == 0) {
			console.info("Skipping blank line " + state.p);
			state.p++;
		}
		var block;
		block = this.#readUnderlineHeader(state); if (block) return block;
		block = this.#readHashHeader(state); if (block) return block;
		block = this.#readBlockQuote(state); if (block) return block;
		block = this.#readUnorderedList(state); if (block) return block;
		block = this.#readOrderedList(state); if (block) return block;
		block = this.#readFencedCodeBlock(state); if (block) return block;
		block = this.#readIndentedCodeBlock(state); if (block) return block;
		block = this.#readHorizontalRule(state); if (block) return block;
		block = this.#readTable(state); if (block) return block;
		block = this.#readDefinitionList(state); if (block) return block;
		block = this.#readFootnoteDef(state); if (block) return block;
		block = this.#readAbbreviationDef(state); if (block) return block;
		block = this.#readParagraph(state); if (block) return block;
		return null;
	}

	/**
	 * @param {_MDState} state
	 * @returns {_MDBlock|null}
	 */
	static #readInline(state, line) {
		return new _MDInline(line);
	}

	/**
	 * Reads the contents of something like a list item
	 * @param {_MDState} state
	 * @param {number} firstLineStartPos
	 * @param {RegExp} stopRegex
	 * @returns {_MDBlock}
	 */
	static #readInteriorContent(state, firstLineStartPos, stopRegex) {
		var p = state.p;
		var seenBlankLine = false;
		var needsBlocks = false;
		var lines = [];
		while (p < state.lines.length) {
			let line = state.lines[p++];
			if (p == state.p + 1) {
				line = line.substring(firstLineStartPos);
			}
			let isBlank = line.trim().length == 0;
			let isIndented = /^\s+/.exec(line) !== null;
			if (isBlank) {
				seenBlankLine = true;
				lines.push(line.trim());
			} else if (stopRegex && stopRegex.exec(line)) {
				p--;
				break;
			} else if (isIndented) {
				if (seenBlankLine) {
					needsBlocks = true;
				}
				lines.push(this.#stripIndent(line));
			} else {
				if (seenBlankLine) {
					p--;
					break;
				}
				lines.push(this.#stripIndent(line));
			}
		}
		while (lines.length > 0 && lines[lines.length - 1].trim().length == 0) {
			lines.pop();
		}
		if (needsBlocks) {
			let substate = new _MDState();
			substate.lines = lines;
			substate.abbreviations = state.abbreviations;
			substate.footnotes = state.footnotes;
			let blocks = this.#readBlocks(substate);
			state.p = p;
			return new _MDMultiBlock(blocks);
		} else {
			state.p = p;
			return this.#readInline(state, lines.join("\n"));
		}
	}

	/**
	 * @param {_MDState} state
	 * @returns {_MDBlock|null}
	 */
	static #readUnderlineHeader(state) {
		var p = state.p;
		if (!state.hasLines(2)) return null;
		let contentLine = state.lines[p++].trim();
		let underLine = state.lines[p++].trim();
		if (contentLine == '') return null;
		if (/^=+$/.exec(underLine)) {
			state.p = p;
			return new _MDHeader(1, this.#readInline(state, contentLine));
		}
		if (/^\-+$/.exec(underLine)) {
			state.p = p;
			return new _MDHeader(2, this.#readInline(state, contentLine));
		}
		return null;
	}

	/**
	 * @param {_MDState} state
	 * @returns {_MDBlock|null}
	 */
	static #readHashHeader(state) {
		var p = state.p;
		var groups = /^(#{1,6})\s*([^#].*)\s*$/.exec(state.lines[p++]);
		if (groups === null) return null;
		state.p = p;
		return new _MDHeader(groups[1].length, this.#readInline(state, groups[2]));
	}

	/**
	 * @param {_MDState} state
	 * @returns {_MDBlock|null}
	 */
	static #readBlockQuote(state) {
		var blockquoteLines = [];
		var p = state.p;
		while (p < state.lines.length) {
			let line = state.lines[p++];
			if (line.startsWith(">")) {
				blockquoteLines.push(line);
			} else {
				break;
			}
		}
		if (blockquoteLines.length > 0) {
			let contentLines = blockquoteLines.map(function(line) {
				return line.substring(1).replace(/^ {0,3}\t?/, '');
			});
			let substate = new _MDState();
			substate.lines = contentLines;
			substate.abbreviations = state.abbreviations;
			substate.footnotes = state.footnotes;
			let quotedBlocks = this.#readBlocks(substate);
			state.p = p;
			return new _MDBlockquote(quotedBlocks);
		}
		return null;
	}

	/**
	 * @param {_MDState} state
	 * @returns {_MDListItem|null}
	 */
	static #readUnorderedListItem(state) {
		var p = state.p;
		let line = state.lines[p];
		let groups = /^([\*\+\-]\s+)(.*)$/.exec(line);
		if (groups === null) return null;
		return new _MDListItem(this.#readInteriorContent(state, groups[1].length, /^[\*\+\-]\s+/));
	}

	/**
	 * @param {_MDState} state
	 * @returns {_MDBlock|null}
	 */
	static #readUnorderedList(state) {
		var p = state.p;
		var items = [];
		var item = null;
		do {
			item = this.#readUnorderedListItem(state);
			if (item) items.push(item);
		} while (item);
		if (items.length == 0) return null;
		return new _MDUnorderedList(items);
	}

	/**
	 * @param {_MDState} state
	 * @returns {_MDBlock|null}
	 */
	static #readOrderedList(state) {
		return null;
	}

	/**
	 * @param {_MDState} state
	 * @returns {_MDBlock|null}
	 */
	static #readFencedCodeBlock(state) {
		return null;
	}

	/**
	 * @param {_MDState} state
	 * @returns {_MDBlock|null}
	 */
	static #readIndentedCodeBlock(state) {
		return null;
	}

	/**
	 * @param {_MDState} state
	 * @returns {_MDBlock|null}
	 */
	static #readHorizontalRule(state) {
		var p = state.p;
		let line = state.lines[p++];
		if (/^\s*(?:\-(?:\s*\-){2,}|\*(?:\s*\*){2,})\s*$/.exec(line)) {
			state.p = p;
			return new _MDHorizontalRule();
		}
		return null;
	}

	/**
	 * @param {_MDState} state
	 * @returns {_MDBlock|null}
	 */
	static #readTable(state) {
		return null;
	}

	/**
	 * @param {_MDState} state
	 * @returns {_MDBlock|null}
	 */
	static #readDefinitionList(state) {
		return null;
	}

	/**
	 * @param {_MDState} state
	 * @returns {_MDBlock|null}
	 */
	static #readFootnoteDef(state) {
		return null;
	}

	/**
	 * @param {_MDState} state
	 * @returns {_MDBlock|null}
	 */
	static #readAbbreviationDef(state) {
		return null;
	}

	/**
	 * @param {_MDState} state
	 * @returns {_MDBlock|null}
	 */
	static #readParagraph(state) {
		if (!state.hasLines(1)) return null;
		var paragraphLines = [];
		var p = state.p;
		while (p < state.lines.length) {
			let line = state.lines[p++];
			if (line.trim().length == 0) {
				break;
			}
			paragraphLines.push(line);
		}
		if (paragraphLines.length > 0) {
			state.p = p;
			let content = paragraphLines.join("\n");
			return new _MDParagraph(this.#readInline(state, content));
		}
		return null;
	}

	/**
	 * @param  {String} markdown
	 * @returns {String} HTML
	 */
	static toHTML(markdown, config=new MDConfig()) {
		var state = new _MDState();
		let lines = markdown.replace("\r", "").split("\n");
		state.lines = lines;
		let blocks = this.#readBlocks(state);
		let html = _MDBlock.toHTML(blocks);
		return html;
	}
}