Merge pull request #74 from SoapBox/2.0.0-beta.4

2.0.0 beta.4

Author: Nick Frasser

run-tests.sh

@@ -1,4 +1,4 @@
-if [[ `echo $TRAVIS_BRANCH` = "master" ]]; then
+if [[ `echo $TRAVIS_BRANCH` = "master" && `echo $SAUCE_USERNAME` != "" ]]; then
 	# Run basic and SauceLabs tests
 	echo "Running complete test suite..."
 	npm test || exit 1

src/linkify-element.js

@@ -8,7 +8,7 @@ const HTML_NODE = 1, TXT_NODE = 3;
 
 /**
 	Given a parent element and child node that the parent contains, replaces
-	that child with the given aary of new children
+	that child with the given array of new children
 */
 function replaceChildWithChildren(parent, oldChild, newChildren) {
 	let lastNewChild = newChildren[newChildren.length - 1];
@@ -22,9 +22,10 @@ function replaceChildWithChildren(parent, oldChild, newChildren) {
 /**
 	Given an array of MultiTokens, return an array of Nodes that are either
 	(a) Plain Text nodes (node type 3)
-	(b) Anchor tag nodes (usually, unless tag name is overriden in the options)
+	(b) Anchor tag nodes (usually, unless tag name is overridden in the options)
 
-	Takes the same options as linkifyElement and an optional doc element (this should be passed in by linkifyElement)
+	Takes the same options as linkifyElement and an optional doc element
+	(this should be passed in by linkifyElement)
 */
 function tokensToNodes(tokens, opts, doc) {
 	let result = [];
@@ -96,8 +97,7 @@ function linkifyElementHelper(element, opts, doc) {
 		return element;
 	}
 
-	let
-	childElement = element.firstChild;
+	let childElement = element.firstChild;
 
 	while (childElement) {
 
@@ -145,8 +145,7 @@ function linkifyElement(element, opts, doc=null) {
 	return linkifyElementHelper(element, opts, doc);
 }
 
-// Maintain reference to the recursive helper to save some option-normalization
-// cycles
+// Maintain reference to the recursive helper to cache option-normalization
 linkifyElement.helper = linkifyElementHelper;
 linkifyElement.normalize = options.normalize;
 

src/linkify/core/parser.js

@@ -23,6 +23,7 @@ TT_DOMAIN		= TEXT_TOKENS.DOMAIN,
 TT_AT			= TEXT_TOKENS.AT,
 TT_COLON		= TEXT_TOKENS.COLON,
 TT_DOT			= TEXT_TOKENS.DOT,
+TT_PUNCTUATION	= TEXT_TOKENS.PUNCTUATION,
 TT_LOCALHOST	= TEXT_TOKENS.LOCALHOST,
 TT_NL			= TEXT_TOKENS.NL,
 TT_NUM			= TEXT_TOKENS.NUM,
@@ -86,9 +87,10 @@ S_PROTOCOL_SLASH.on(TT_SLASH, S_PROTOCOL_SLASH_SLASH);
 S_START.on(TT_TLD, S_DOMAIN);
 S_START.on(TT_DOMAIN, S_DOMAIN);
 S_START.on(TT_LOCALHOST, S_TLD);
-S_START.on(TT_NUM, S_LOCALPART);
+S_START.on(TT_NUM, S_DOMAIN);
 S_PROTOCOL_SLASH_SLASH.on(TT_TLD, S_PSS_DOMAIN);
 S_PROTOCOL_SLASH_SLASH.on(TT_DOMAIN, S_PSS_DOMAIN);
+S_PROTOCOL_SLASH_SLASH.on(TT_NUM, S_PSS_DOMAIN);
 S_PROTOCOL_SLASH_SLASH.on(TT_LOCALHOST, S_PSS_TLD);
 
 // Account for dots and hyphens
@@ -102,12 +104,15 @@ S_EMAIL_DOMAIN.on(TT_DOT, S_EMAIL_DOMAIN_DOT);
 // After the first domain and a dot, we can find either a URL or another domain
 S_DOMAIN_DOT.on(TT_TLD, S_TLD);
 S_DOMAIN_DOT.on(TT_DOMAIN, S_DOMAIN);
+S_DOMAIN_DOT.on(TT_NUM, S_DOMAIN);
 S_DOMAIN_DOT.on(TT_LOCALHOST, S_DOMAIN);
 S_PSS_DOMAIN_DOT.on(TT_TLD, S_PSS_TLD);
 S_PSS_DOMAIN_DOT.on(TT_DOMAIN, S_PSS_DOMAIN);
+S_PSS_DOMAIN_DOT.on(TT_NUM, S_PSS_DOMAIN);
 S_PSS_DOMAIN_DOT.on(TT_LOCALHOST, S_PSS_DOMAIN);
 S_EMAIL_DOMAIN_DOT.on(TT_TLD, S_EMAIL);
 S_EMAIL_DOMAIN_DOT.on(TT_DOMAIN, S_EMAIL_DOMAIN);
+S_EMAIL_DOMAIN_DOT.on(TT_NUM, S_EMAIL_DOMAIN);
 S_EMAIL_DOMAIN_DOT.on(TT_LOCALHOST, S_EMAIL_DOMAIN);
 
 // S_TLD accepts! But the URL could be longer, try to find a match greedily
@@ -133,14 +138,14 @@ S_EMAIL_COLON.on(TT_NUM, S_EMAIL_PORT);
 let qsAccepting = [
 	TT_DOMAIN,
 	TT_AT,
-
 	TT_LOCALHOST,
 	TT_NUM,
 	TT_PLUS,
 	TT_POUND,
 	TT_PROTOCOL,
 	TT_SLASH,
-	TT_TLD
+	TT_TLD,
+	TT_SYM
 ];
 
 // Types of tokens that can follow a URL and be part of the query string
@@ -150,7 +155,7 @@ let qsNonAccepting = [
 	TT_COLON,
 	TT_DOT,
 	TT_QUERY,
-	TT_SYM
+	TT_PUNCTUATION
 ];
 
 // Account for the query string
@@ -167,7 +172,6 @@ S_URL_SYMS.on(qsNonAccepting, S_URL_SYMS);
 // Tokens allowed in the localpart of the email
 let localpartAccepting = [
 	TT_DOMAIN,
-	TT_COLON,
 	TT_NUM,
 	TT_PLUS,
 	TT_POUND,
@@ -183,12 +187,6 @@ S_DOMAIN.on(TT_AT, S_LOCALPART_AT);
 S_DOMAIN_DOT.on(localpartAccepting, S_LOCALPART);
 S_TLD.on(localpartAccepting, S_LOCALPART);
 S_TLD.on(TT_AT, S_LOCALPART_AT);
-S_TLD_COLON.on(localpartAccepting, S_LOCALPART);
-S_TLD_COLON.on(TT_DOT, S_LOCALPART);
-S_TLD_COLON.on(TT_AT, S_LOCALPART_AT);
-S_TLD_PORT.on(localpartAccepting, S_LOCALPART);
-S_TLD_PORT.on(TT_DOT, S_LOCALPART_DOT);
-S_TLD_PORT.on(TT_AT, S_LOCALPART_AT);
 
 // Okay we're on a localpart. Now what?
 // TODO: IP addresses and what if the email starts with numbers?

src/linkify/core/scanner.js

@@ -44,6 +44,7 @@ S_START.on('#', makeState(TOKENS.POUND));
 S_START.on('?', makeState(TOKENS.QUERY));
 S_START.on('/', makeState(TOKENS.SLASH));
 S_START.on(COLON, makeState(TOKENS.COLON));
+S_START.on(/[,;!]/, makeState(TOKENS.PUNCTUATION));
 
 // Whitespace jumps
 // Tokens of only non-newline whitespace are arbitrarily long

src/linkify/core/tokens.js

@@ -63,6 +63,15 @@ class DOT extends TextToken {
 	constructor() { super('.'); }
 }
 
+/**
+	A character class that can surround the URL, but which the URL cannot begin
+	or end with. Does not include certain English punctuation like parentheses.
+
+	@class PUNCTUATION
+	@extends TextToken
+*/
+class PUNCTUATION extends TextToken {}
+
 /**
 	The word localhost (by itself)
 	@class LOCALHOST
@@ -158,6 +167,7 @@ let text = {
 	AT,
 	COLON,
 	DOT,
+	PUNCTUATION,
 	LOCALHOST,
 	NL: TNL,
 	NUM,

test/spec/linkify/core/parser-test.js

@@ -15,7 +15,6 @@ EMAIL	= MULTI_TOKENS.EMAIL;
 	[2] - The values of the tokens the text should result in
 */
 var tests = [
-	// BEGIN: Original linkify tests
 	[
 		'google.com',
 		[URL],
@@ -33,12 +32,12 @@ var tests = [
 		['there are two tests, ', 'brennan.com', ' and ', 'nick.ca', ' -- do they work?']
 	], [
 		'there are two tests!brennan.com. and nick.ca? -- do they work?',
-		[TEXT, URL, TEXT],
-		['there are two tests!brennan.com. and ', 'nick.ca', '? -- do they work?']
+		[TEXT, URL, TEXT, URL, TEXT],
+		['there are two tests!', 'brennan.com', '. and ', 'nick.ca', '? -- do they work?']
 	], [
 		'This [i.imgur.com/ckSj2Ba.jpg)] should also work',
 		[TEXT, URL, TEXT],
-		['This [', 'i.imgur.com/ckSj2Ba.jpg', ')] should also work']
+		['This [', 'i.imgur.com/ckSj2Ba.jpg)]', ' should also work']
 	], [
 		'A link is http://nick.is.awesome/?q=nick+amazing&nick=yo%29%30hellp another is http://nick.con/?q=look',
 		[TEXT, URL, TEXT],
@@ -63,13 +62,10 @@ var tests = [
 		'This port is too short someport.com: this port is too long http://googgle.com:789023/myQuery this port is just right https://github.com:8080/SoapBox/jQuery-linkify/',
 		[TEXT, URL, TEXT, URL, TEXT, URL],
 		['This port is too short ', 'someport.com', ': this port is too long ', 'http://googgle.com:789023/myQuery', ' this port is just right ', 'https://github.com:8080/SoapBox/jQuery-linkify/']
-	],
-	// END: Original linkifiy tests
-	// BEGIN: New linkify tests
-	[
-		'The best URL http://google.com/?love=true and t.co',
+	], [
+		'The best URL http://google.com/?love=true, and t.co',
 		[TEXT, URL, TEXT, URL],
-		['The best URL ', 'http://google.com/?love=true', ' and ', 't.co']
+		['The best URL ', 'http://google.com/?love=true', ', and ', 't.co']
 	], [
 		'Please email me at [email protected]',
 		[TEXT, EMAIL],
@@ -83,15 +79,42 @@ var tests = [
 		[URL, TEXT],
 		['http://500-px.com', ' is a real domain?']
 	], [
-		'IP loops like email? [email protected] works!!',
+		'IP loops like email? [email protected]! works!!',
 		[TEXT, EMAIL, TEXT],
-		['IP loops like email? ', '[email protected]', ' works!!']
+		['IP loops like email? ', '[email protected]', '! works!!']
 	], [
-		'Url like bro-215.co with a hyphen?',
+		'Url like bro-215.co; with a hyphen?',
 		[TEXT, URL, TEXT],
-		['Url like ', 'bro-215.co', ' with a hyphen?']
+		['Url like ', 'bro-215.co', '; with a hyphen?']
+	], [
+		'This URL http://23456789098.sydney is a number',
+		[TEXT, URL, TEXT],
+		['This URL ', 'http://23456789098.sydney', ' is a number']
+	], [
+		'This URL http://23456789098.sydney is a number',
+		[TEXT, URL, TEXT],
+		['This URL ', 'http://23456789098.sydney', ' is a number']
+	], [
+		'A URL with only numbers is 123.456.ca another is //7.8.com/?wat=1 is valid',
+		[TEXT, URL, TEXT, URL, TEXT],
+		['A URL with only numbers is ', '123.456.ca', ' another is ', '//7.8.com/?wat=1', ' is valid']
+	], [
+		'URL Numbers 6.wat.78.where.eu and u.0.e.9.kp',
+		[TEXT, URL, TEXT, URL],
+		['URL Numbers ', '6.wat.78.where.eu', ' and ', 'u.0.e.9.kp']
+	], [
+		'Emails like nick:[email protected] do not have colons in them',
+		[TEXT, EMAIL, TEXT],
+		['Emails like nick:', '[email protected]', ' do not have colons in them']
+	], [
+		'Emails cannot have two dots, e.g.: [email protected]',
+		[TEXT, EMAIL],
+		['Emails cannot have two dots, e.g.: nick..', '[email protected]']
+	], [
+		'The `mailto:` part should not be included in mailto:[email protected]',
+		[TEXT, EMAIL],
+		['The `mailto:` part should not be included in mailto:', '[email protected]']
 	]
-	// END: New linkify tests
 ];
 
 describe('linkify/core/parser#run()', function () {

test/spec/linkify/core/scanner-test.js

@@ -7,6 +7,7 @@ DOMAIN		= TEXT_TOKENS.DOMAIN,
 AT			= TEXT_TOKENS.AT,
 COLON		= TEXT_TOKENS.COLON,
 DOT			= TEXT_TOKENS.DOT,
+PUNCTUATION	= TEXT_TOKENS.PUNCTUATION,
 LOCALHOST	= TEXT_TOKENS.LOCALHOST,
 NL			= TEXT_TOKENS.NL,
 NUM			= TEXT_TOKENS.NUM,
@@ -35,6 +36,7 @@ var tests = [
 	['/', [SLASH], ['/']],
 	['&', [SYM], ['&']],
 	['&?<>(', [SYM, QUERY, SYM, SYM, SYM], ['&', '?', '<', '>', '(']],
+	['!,;', [PUNCTUATION, PUNCTUATION, PUNCTUATION], ['!', ',', ';']],
 	['hello', [DOMAIN], ['hello']],
 	['Hello123', [DOMAIN], ['Hello123']],
 	['hello123world', [DOMAIN], ['hello123world']],