Add bonuspunkt's parser

Fixes #15.
Fixes #199.
Fixes #583.
Fixes #654.
Fixes #928.
Fixes #1001.

client/js/libs/handlebars/ircmessageparser/anyIntersection.js

@@ -0,0 +1,10 @@
+"use strict";
+
+function anyIntersection(a, b) {
+	return a.start <= b.start && b.start < a.end ||
+		a.start < b.end && b.end <= a.end ||
+		b.start <= a.start && a.start < b.end ||
+		b.start < a.end && a.end <= b.end;
+}
+
+module.exports = anyIntersection;

client/js/libs/handlebars/ircmessageparser/fill.js

@@ -0,0 +1,29 @@
+"use strict";
+
+function fill(existingEntries, text) {
+	let position = 0;
+	const result = [];
+
+	for (let i = 0; i < existingEntries.length; i++) {
+		const textSegment = existingEntries[i];
+
+		if (textSegment.start > position) {
+			result.push({
+				start: position,
+				end: textSegment.start
+			});
+		}
+		position = textSegment.end;
+	}
+
+	if (position < text.length) {
+		result.push({
+			start: position,
+			end: text.length
+		});
+	}
+
+	return result;
+}
+
+module.exports = fill;

client/js/libs/handlebars/ircmessageparser/findChannels.js

@@ -0,0 +1,32 @@
+"use strict";
+
+const escapeRegExp = require("lodash/escapeRegExp");
+
+// NOTE: channel prefixes should be RPL_ISUPPORT.CHANTYPES
+// NOTE: userModes should be RPL_ISUPPORT.PREFIX
+function findChannels(text, channelPrefixes, userModes) {
+	const userModePattern = userModes.map(escapeRegExp).join("");
+	const channelPrefixPattern = channelPrefixes.map(escapeRegExp).join("");
+
+	const channelPattern = `(?:^|\\s)[${ userModePattern }]*([${ channelPrefixPattern }][^ \u0007]+)`;
+	const channelRegExp = new RegExp(channelPattern, "g");
+
+	const result = [];
+	let match;
+
+	do {
+		match = channelRegExp.exec(text);
+
+		if (match) {
+			result.push({
+				start: match.index + match[0].length - match[1].length,
+				end: match.index + match[0].length,
+				channel: match[1]
+			});
+		}
+	} while (match);
+
+	return result;
+}
+
+module.exports = findChannels;

client/js/libs/handlebars/ircmessageparser/merge.js

@@ -0,0 +1,47 @@
+"use strict";
+
+const anyIntersection = require("./anyIntersection");
+const fill = require("./fill");
+
+let Object_assign = Object.assign;
+
+if (typeof Object_assign !== "function") {
+	Object_assign = function(target) {
+		Array.prototype.slice.call(arguments, 1).forEach(function(obj) {
+			Object.keys(obj).forEach(function(key) {
+				target[key] = obj[key];
+			});
+		});
+		return target;
+	};
+}
+
+function assign(textPart, fragment) {
+	const fragStart = fragment.start;
+	const start = Math.max(fragment.start, textPart.start);
+	const end = Math.min(fragment.end, textPart.end);
+
+	return Object_assign({}, fragment, {
+		start: start,
+		end: end,
+		text: fragment.text.slice(start - fragStart, end - fragStart)
+	});
+}
+
+function merge(textParts, styleFragments) {
+	const cleanText = styleFragments.map(fragment => fragment.text).join("");
+
+	const allParts = textParts
+		.concat(fill(textParts, cleanText))
+		.sort((a, b) => a.start - b.start);
+
+	return allParts.map(textPart => {
+		textPart.fragments = styleFragments
+			.filter(fragment => anyIntersection(textPart, fragment))
+			.map(fragment => assign(textPart, fragment));
+
+		return textPart;
+	});
+}
+
+module.exports = merge;

client/js/libs/handlebars/ircmessageparser/parseStyle.js

@@ -0,0 +1,131 @@
+"use strict";
+
+const BOLD = "\x02";
+const COLOR = "\x03";
+const RESET = "\x0f";
+const REVERSE = "\x16";
+const ITALIC = "\x1d";
+const UNDERLINE = "\x1f";
+
+const colorRx = /^(\d{1,2})(?:,(\d{1,2}))?/;
+const controlCodesRx = /[\u0000-\u001F]/g;
+
+function parseStyle(text) {
+	const result = [];
+	let start = 0;
+	let position = 0;
+
+	let colorCodes, bold, textColor, bgColor, reverse, italic, underline;
+
+	const resetStyle = () => {
+		bold = false;
+		textColor = undefined;
+		bgColor = undefined;
+		reverse = false;
+		italic = false;
+		underline = false;
+	};
+	resetStyle();
+
+	const emitFragment = () => {
+		const textPart = text.slice(start, position);
+		start = position + 1;
+
+		const processedText = textPart.replace(controlCodesRx, "");
+
+		if (!processedText.length) {
+			return;
+		}
+
+		result.push({
+			bold,
+			textColor,
+			bgColor,
+			reverse,
+			italic,
+			underline,
+			text: processedText
+		});
+	};
+
+	while (position < text.length) {
+		switch (text[position]) {
+
+		case RESET:
+			emitFragment();
+			resetStyle();
+			break;
+
+		case BOLD:
+			emitFragment();
+			bold = !bold;
+			break;
+
+		case COLOR:
+			emitFragment();
+
+			colorCodes = text.slice(position + 1).match(colorRx);
+
+			if (colorCodes) {
+				textColor = Number(colorCodes[1]);
+				bgColor = Number(colorCodes[2]);
+				if (Number.isNaN(bgColor)) {
+					bgColor = undefined;
+				}
+				position += colorCodes[0].length;
+			} else {
+				textColor = undefined;
+				bgColor = undefined;
+			}
+			start = position + 1;
+			break;
+
+		case REVERSE:
+			emitFragment();
+			reverse = !reverse;
+			break;
+
+		case ITALIC:
+			emitFragment();
+			italic = !italic;
+			break;
+
+		case UNDERLINE:
+			emitFragment();
+			underline = !underline;
+			break;
+		}
+		position += 1;
+	}
+
+	emitFragment();
+
+	return result;
+}
+
+const properties = ["bold", "textColor", "bgColor", "italic", "underline", "reverse"];
+
+function prepare(text) {
+	return parseStyle(text)
+		.filter(fragment => fragment.text.length)
+		.reduce((prev, curr, i) => {
+			if (i === 0) {
+				return prev.concat([curr]);
+			}
+
+			const lastEntry = prev[prev.length - 1];
+			if (properties.some(key => curr[key] !== lastEntry[key])) {
+				return prev.concat([curr]);
+			}
+
+			lastEntry.text += curr.text;
+			return prev;
+		}, [])
+		.map((fragment, i, array) => {
+			fragment.start = i === 0 ? 0 : array[i - 1].end;
+			fragment.end = fragment.start + fragment.text.length;
+			return fragment;
+		});
+}
+
+module.exports = prepare;

client/js/libs/handlebars/parse.js

@@ -2,125 +2,113 @@
 
 const Handlebars = require("handlebars/runtime");
 const URI = require("urijs");
+const parseStyle = require("./ircmessageparser/parseStyle");
+const findChannels = require("./ircmessageparser/findChannels");
+const merge = require("./ircmessageparser/merge");
 
-module.exports = function(text) {
-	text = Handlebars.Utils.escapeExpression(text);
-	text = colors(text);
-	text = channels(text);
-	text = uri(text);
-	return text;
-};
+const commonSchemes = [
+	"http", "https",
+	"ftp", "sftp",
+	"smb", "file",
+	"irc", "ircs",
+	"svn", "git",
+	"steam", "mumble", "ts3server",
+	"svn+ssh", "ssh",
+];
 
-function uri(text) {
-	return URI.withinString(text, function(url) {
-		if (url.indexOf("javascript:") === 0) {
-			return url;
-		}
-		var split = url.split("<");
-		url = "<a href='" + split[0].replace(/^www/, "http://www") + "' target='_blank' rel='noopener'>" + split[0] + "</a>";
-		if (split.length > 1) {
-			url += "<" + split.slice(1).join("<");
-		}
-		return url;
-	});
-}
+function findLinks(text) {
+	let result = [];
+	let lastPosition = 0;
 
-/**
- * Channels names are strings of length up to fifty (50) characters.
- * The only restriction on a channel name is that it SHALL NOT contain
- * any spaces (' '), a control G (^G or ASCII 7), a comma (',').
- * Channel prefix '&' is handled as '&amp;' because this parser is executed
- * after entities in the message have been escaped. This prevents a couple of bugs.
- */
-function channels(text) {
-	return text.replace(
-		/(^|\s|\x07|,)((?:#|&amp;)[^\x07\s,]{1,49})/g,
-		'$1<span class="inline-channel" role="button" tabindex="0" data-chan="$2">$2</span>'
-	);
-}
-
-/**
- * MIRC compliant colour and style parser
- * Unfortuanately this is a non trivial operation
- * See this branch for source and tests
- * https://github.com/megawac/irc-style-parser/tree/shout
- */
-var styleCheck_Re = /[\x00-\x1F]/,
-	back_re = /^([0-9]{1,2})(,([0-9]{1,2}))?/,
-	colourKey = "\x03",
-	// breaks all open styles ^O (\x0F)
-	styleBreak = "\x0F";
-
-function styleTemplate(settings) {
-	return "<span class='" + settings.style + "'>" + settings.text + "</span>";
-}
-
-var styles = [
-	["normal", "\x00", ""], ["underline", "\x1F"],
-	["bold", "\x02"], ["italic", "\x1D"]
-].map(function(style) {
-	var escaped = encodeURI(style[1]).replace("%", "\\x");
-	return {
-		name: style[0],
-		style: style[2] ? style[2] : "irc-" + style[0],
-		key: style[1],
-		keyregex: new RegExp(escaped + "(.*?)(" + escaped + "|$)")
-	};
-});
-
-function colors(line) {
-	// http://www.mirc.com/colors.html
-	// http://www.aviran.org/stripremove-irc-client-control-characters/
-	// https://github.com/perl6/mu/blob/master/examples/rules/Grammar-IRC.pm
-	// regexs are cruel to parse this thing
-
-	// already done?
-	if (!styleCheck_Re.test(line)) {
-		return line;
-	}
-
-	// split up by the irc style break character ^O
-	if (line.indexOf(styleBreak) >= 0) {
-		return line.split(styleBreak).map(colors).join("");
-	}
-
-	var result = line;
-	var parseArr = result.split(colourKey);
-	var text, match, colour, background = "";
-	for (var i = 0; i < parseArr.length; i++) {
-		text = parseArr[i];
-		match = text.match(back_re);
-		if (!match) {
-			// ^C (no colour) ending. Escape current colour and carry on
-			background = "";
-			continue;
-		}
-		colour = "irc-fg" + +match[1];
-		// set the background colour
-		if (match[3]) {
-			background = " irc-bg" + +match[3];
-		}
-		// update the parsed text result
-		result = result.replace(colourKey + text, styleTemplate({
-			style: colour + background,
-			text: text.slice(match[0].length)
-		}));
-	}
-
-	// Matching styles (italics/bold/underline)
-	// if only colours were this easy...
-	styles.forEach(function(style) {
-		if (result.indexOf(style.key) < 0) {
+	URI.withinString(text, function(url, start, end) {
+		// v-- fix: url was modified and does not match input string -> cant be mapped
+		if (text.indexOf(url, lastPosition) < 0) {
 			return;
 		}
+		// ^-- /fix: url was modified and does not match input string -> cant be mapped
 
-		result = result.replace(style.keyregex, function(matchedTrash, matchedText) {
-			return styleTemplate({
-				style: style.style,
-				text: matchedText
-			});
+		// v-- fix: use prefered scheme
+		const parsed = URI(url);
+		const parsedScheme = parsed.scheme().toLowerCase();
+		const matchedScheme = commonSchemes.find(scheme => parsedScheme.endsWith(scheme));
+
+		if (matchedScheme) {
+			const prefix = parsedScheme.length - matchedScheme.length;
+			start += prefix;
+			url = url.slice(prefix);
+		}
+		// ^-- /fix: use prefered scheme
+
+		// URL matched, but does not start with a protocol, add it
+		if (!parsedScheme.length) {
+			url = "http://" + url;
+		}
+
+		result.push({
+			start: start,
+			end: end,
+			link: url
 		});
 	});
 
 	return result;
 }
+
+function createFragment(fragment) {
+	let className = "";
+	if (fragment.bold) {
+		className += " irc-bold";
+	}
+	if (fragment.textColor !== undefined) {
+		className += " irc-fg" + fragment.textColor;
+	}
+	if (fragment.bgColor !== undefined) {
+		className += " irc-bg" + fragment.bgColor;
+	}
+	if (fragment.italic) {
+		className += " irc-italic";
+	}
+	if (fragment.underline) {
+		className += " irc-underline";
+	}
+	const escapedText = Handlebars.Utils.escapeExpression(fragment.text);
+	if (className) {
+		return "<span class='" + className.trim() + "'>" + escapedText + "</span>";
+	}
+	return escapedText;
+}
+
+module.exports = function parse(text) {
+	const styleFragments = parseStyle(text);
+	const cleanText = styleFragments.map(fragment => fragment.text).join("");
+
+	const channelPrefixes = ["#", "&"]; // RPL_ISUPPORT.CHANTYPES
+	const userModes = ["!", "@", "%", "+"]; // RPL_ISUPPORT.PREFIX
+	const channelParts = findChannels(cleanText, channelPrefixes, userModes);
+
+	const linkParts = findLinks(cleanText);
+
+	const parts = channelParts
+		.concat(linkParts)
+		.sort((a, b) => a.start - b.start);
+
+	return merge(parts, styleFragments).map(textPart => {
+		const fragments = textPart.fragments.map(createFragment).join("");
+
+		if (textPart.link) {
+			const escapedLink = Handlebars.Utils.escapeExpression(textPart.link);
+			return (
+				"<a href='" + escapedLink + "' target='_blank' rel='noopener'>" +
+					fragments +
+				"</a>");
+		} else if (textPart.channel) {
+			const escapedChannel = Handlebars.Utils.escapeExpression(textPart.channel);
+			return (
+				"<span class='inline-channel' role='button' tabindex='0' data-chan='" + escapedChannel + "'>" +
+					fragments +
+				"</span>");
+		}
+
+		return fragments;
+	}).join("");
+};