Merge pull request #127 from SoapBox/protocol-parsing

Anything that begins with a protocol is recognized as a string

Author: Nick Frasser

src/linkify/core/parser.js

@@ -54,9 +54,6 @@ let S_START = makeState();
 
 // Intermediate states for URLs. Note that domains that begin with a protocol
 // are treated slighly differently from those that don't.
-// (PSS == "PROTOCOL SLASH SLASH")
-// S_DOMAIN* states can generally become prefixes for email addresses, while
-// S_PSS_DOMAIN* cannot
 let
 S_PROTOCOL				= makeState(), // e.g., 'http:'
 S_PROTOCOL_SLASH		= makeState(), // e.g., '/', 'http:/''
@@ -66,11 +63,6 @@ S_DOMAIN_DOT			= makeState(), // (A) domain followed by DOT
 S_TLD					= makeState(T_URL), // (A) Simplest possible URL with no query string
 S_TLD_COLON				= makeState(), // (A) URL followed by colon (potential port number here)
 S_TLD_PORT				= makeState(T_URL), // TLD followed by a port number
-S_PSS_DOMAIN			= makeState(), // parsed string starts with protocol and ends with a potential domain name (B)
-S_PSS_DOMAIN_DOT		= makeState(), // (B) domain followed by DOT
-S_PSS_TLD				= makeState(T_URL), // (B) Simplest possible URL with no query string and a protocol
-S_PSS_TLD_COLON			= makeState(), // (A) URL followed by colon (potential port number here)
-S_PSS_TLD_PORT			= makeState(T_URL), // TLD followed by a port number
 S_URL					= makeState(T_URL), // Long URL with optional port and maybe query string
 S_URL_SYMS				= makeState(), // URL followed by some symbols (will not be part of the final URL)
 S_URL_OPENBRACE			= makeState(), // URL followed by {
@@ -104,15 +96,16 @@ 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_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);
+
+// Force URL for anything sane followed by protocol
+S_PROTOCOL_SLASH_SLASH.on(TT_TLD, S_URL);
+S_PROTOCOL_SLASH_SLASH.on(TT_DOMAIN, S_URL);
+S_PROTOCOL_SLASH_SLASH.on(TT_NUM, S_URL);
+S_PROTOCOL_SLASH_SLASH.on(TT_LOCALHOST, S_URL);
 
 // Account for dots and hyphens
 // hyphens are usually parts of domain names
 S_DOMAIN.on(TT_DOT, S_DOMAIN_DOT);
-S_PSS_DOMAIN.on(TT_DOT, S_PSS_DOMAIN_DOT);
 S_EMAIL_DOMAIN.on(TT_DOT, S_EMAIL_DOMAIN_DOT);
 
 // Hyphen can jump back to a domain name
@@ -122,10 +115,6 @@ 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);
@@ -134,7 +123,6 @@ 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
 // The `run` function should be able to "rollback" to the accepting state
 S_TLD.on(TT_DOT, S_DOMAIN_DOT);
-S_PSS_TLD.on(TT_DOT, S_PSS_DOMAIN_DOT);
 S_EMAIL.on(TT_DOT, S_EMAIL_DOMAIN_DOT);
 
 // Become real URLs after `SLASH` or `COLON NUM SLASH`
@@ -143,10 +131,6 @@ S_TLD.on(TT_COLON, S_TLD_COLON);
 S_TLD.on(TT_SLASH, S_URL);
 S_TLD_COLON.on(TT_NUM, S_TLD_PORT);
 S_TLD_PORT.on(TT_SLASH, S_URL);
-S_PSS_TLD.on(TT_COLON, S_PSS_TLD_COLON);
-S_PSS_TLD.on(TT_SLASH, S_URL);
-S_PSS_TLD_COLON.on(TT_NUM, S_PSS_TLD_PORT);
-S_PSS_TLD_PORT.on(TT_SLASH, S_URL);
 S_EMAIL.on(TT_COLON, S_EMAIL_COLON);
 S_EMAIL_COLON.on(TT_NUM, S_EMAIL_PORT);
 

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

@@ -40,8 +40,8 @@ var tests = [
 		['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],
-		['A link is ', 'http://nick.is', '.awesome/?q=nick+amazing&nick=yo%29%30hellp another is http://nick.con/?q=look']
+		[TEXT, URL, TEXT, URL],
+		['A link is ', 'http://nick.is.awesome/?q=nick+amazing&nick=yo%29%30hellp', ' another is ', 'http://nick.con/?q=look']
 	], [
 		'SOme URLS http://google.com https://google1.com google2.com google.com/search?q=potatoes+oven goo.gl/0192n1 google.com?q=asda test bit.ly/0912j www.bob.com indigo.dev.soapbox.co/mobile google.com/?q=.exe flickr.com/linktoimage.jpg',
 		[TEXT, URL, TEXT, URL, TEXT, URL, TEXT, URL, TEXT, URL, TEXT, URL, TEXT, URL, TEXT, URL, TEXT, URL, TEXT, URL, TEXT, URL],
@@ -130,6 +130,14 @@ var tests = [
 		'A really funky one (example.com/?id=asd2{hellow}and%20it%20continues(23&((@)) and it ends',
 		[TEXT, URL, TEXT],
 		['A really funky one (', 'example.com/?id=asd2{hellow}and%20it%20continues(23&((@)', ') and it ends']
+	], [
+		'Force http:/ and http:// are not but http://a and http://b.local?qeasd3qas=23 are all links',
+		[TEXT, URL, TEXT, URL, TEXT],
+		['Force http:/ and http:// are not but ', 'http://a', ' and ', 'http://b.local?qeasd3qas=23', ' are all links']
+	], [
+		'HTTP Auth URLs should work: http://username:[email protected]',
+		[TEXT, URL],
+		['HTTP Auth URLs should work: ', 'http://username:[email protected]']
 	]
 ];