From 93e88121de827eb29b28d0d138b6b80690585021 Mon Sep 17 00:00:00 2001
From: Luca Frosini <luca.frosini@isti.cnr.it>
Date: Thu, 24 Jan 2019 10:38:34 +0000
Subject: [PATCH] Better implementation of URL identification

git-svn-id: https://svn.d4science.research-infrastructures.eu/gcube/trunk/social-networking/social-util-library@176778 82a268e6-3cf1-43bd-a215-b396298e98cf
---
 .../socialnetworking/token/SanitizedURL.java  | 47 +++++++++++++++++++
 .../socialnetworking/token/URLToken.java      | 37 +++++----------
 2 files changed, 58 insertions(+), 26 deletions(-)
 create mode 100644 src/main/java/org/gcube/socialnetworking/token/SanitizedURL.java

diff --git a/src/main/java/org/gcube/socialnetworking/token/SanitizedURL.java b/src/main/java/org/gcube/socialnetworking/token/SanitizedURL.java
new file mode 100644
index 0000000..96fef58
--- /dev/null
+++ b/src/main/java/org/gcube/socialnetworking/token/SanitizedURL.java
@@ -0,0 +1,47 @@
+package org.gcube.socialnetworking.token;
+
+import java.net.MalformedURLException;
+import java.net.URL;
+
+public class SanitizedURL {
+	
+	private static String FINAL_CHARACTERS_TO_REMOVE_FROM_LINK = "[\\.\\,\\;\\)\\:]";
+	
+	protected String prefix;
+	protected String postfix;
+	protected final URL url;
+	
+	public SanitizedURL(String urlString) throws MalformedURLException {
+		prefix = "";
+		if(urlString.startsWith("(")) {
+			prefix = urlString.substring(0, 1);
+			urlString = urlString.substring(1);
+		}
+		
+		if(urlString.startsWith("www.")) {
+			urlString = "http://" + urlString;
+		}
+		
+		postfix = urlString.substring(urlString.length()-1);
+		if(postfix.matches(FINAL_CHARACTERS_TO_REMOVE_FROM_LINK)) {
+			urlString = urlString.substring(0, urlString.length()-1);
+		}else {
+			postfix = "";
+		}
+		
+		url = new URL(urlString);
+	}
+
+	public String getPrefix() {
+		return prefix;
+	}
+
+	public String getPostfix() {
+		return postfix;
+	}
+
+	public URL getURL() {
+		return url;
+	}
+	
+}
\ No newline at end of file
diff --git a/src/main/java/org/gcube/socialnetworking/token/URLToken.java b/src/main/java/org/gcube/socialnetworking/token/URLToken.java
index e33dcaf..856785f 100644
--- a/src/main/java/org/gcube/socialnetworking/token/URLToken.java
+++ b/src/main/java/org/gcube/socialnetworking/token/URLToken.java
@@ -7,8 +7,6 @@ import java.util.Map;
 
 public class URLToken extends ReplaceableToken {
 	
-	private static String FINAL_CHARACTERS_TO_REMOVE_FROM_LINK = "[\\.\\,\\;\\)\\:]";
-	
 	public URLToken(String token, String delimiter, int start, int end) {
 		super(token, delimiter, start, end);
 	}
@@ -17,15 +15,10 @@ public class URLToken extends ReplaceableToken {
 		super(token);
 	}
 	
-	public static URL isURL(String string) {
+	public static URL isURL(String url) {
 		try {
-			if(string.startsWith("(")) {
-				string = string.substring(1);
-			}
-			if(string.startsWith("www.")) {
-				string = "http://" + string;
-			}
-			return new URL(string);
+			SanitizedURL sanitizedURL = new SanitizedURL(url);
+			return sanitizedURL.getURL();
 		} catch(MalformedURLException e) {
 			// not an URL
 			return null;
@@ -35,23 +28,15 @@ public class URLToken extends ReplaceableToken {
 	
 	public String getTokenReplacement() {
 		if(!replaced) {
-			// Testing if it is an URL preceded by (
-			String prefix = ""; 
-			if(tokenReplacement.startsWith("(")) {
-				tokenReplacement = token.substring(1);
-				prefix = "(";
+			try {
+				Map<String,String> anchorAttibutes = new HashMap<>(1);
+				anchorAttibutes.put("target", "_blank");
+				SanitizedURL sanitizedURL = new SanitizedURL(token);
+				String url = sanitizedURL.getURL().toString();
+				tokenReplacement = sanitizedURL.getPrefix() + ReplaceableToken.createLink(url, url, anchorAttibutes) + sanitizedURL.getPostfix();
+			}catch(MalformedURLException e) {
+				tokenReplacement = token;
 			}
-			
-			Map<String,String> anchorAttibutes = new HashMap<>(1);
-			anchorAttibutes.put("target", "_blank");
-			String url = tokenReplacement;
-			String postfix = tokenReplacement.substring(tokenReplacement.length()-1);
-			if(postfix.matches(FINAL_CHARACTERS_TO_REMOVE_FROM_LINK)) {
-				url = tokenReplacement.substring(0, token.length()-1);
-			}else {
-				postfix = "";
-			}
-			tokenReplacement = prefix + ReplaceableToken.createLink(url, url, anchorAttibutes) + postfix;
 			replaced = true;
 		}
 		return tokenReplacement;