Fixing find URL

git-svn-id: https://svn.d4science.research-infrastructures.eu/gcube/trunk/social-networking/social-util-library@176813 82a268e6-3cf1-43bd-a215-b396298e98cf

src/main/java/org/gcube/social_networking/socialutillibrary/Utils.java

@@ -39,21 +39,25 @@ public class Utils {
 	/**
 	 * Hashtag regex enhanced for ticket #4937
 	 */
+	@Deprecated
 	private static final String HASHTAG_REGEX = "^#\\w+([-_.]?\\w+)*|\\s#\\w+([-_.]?\\w+)*|(?<=[\\[({])#\\w+([-_.]?\\w+)";
 
 	/**
 	 * IPv4 regex
 	 */
+	@Deprecated
 	public static final String IPV4_REGEX = "\\b(?:(?:25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?)\\.){3}(?:25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?)\\b";
 
 	/**
 	 * IPV6 regex needs to be put in square brackets, see below
 	 */
+	@Deprecated
 	public static final String IPV6_REGEX = "\\b(?:[0-9a-fA-F]{1,4}:){7}[0-9a-fA-F]{1,4}\\b";
 
 	/**
 	 * Pattern for URLS
 	 */
+	@Deprecated
 	private static final Pattern URL_PATTERN = Pattern.compile(
 			"\\b(((ht|f)tp(s?)\\:\\/\\/|~\\/|\\/)|www.)" + 
 					"(\\w+:\\w+@)?((([-\\w]+\\.)+(com|org|net|gov" + 
@@ -174,6 +178,7 @@ public class Utils {
 	 * @param feedText
 	 * @return the text with the clickable url in it
 	 */
+	@Deprecated
 	public static String extractURL(String feedText) {
 		// separate input by spaces ( URLs have no spaces )
 		feedText = feedText.replaceAll("(\r\n|\n)"," <br/> ");
@@ -271,6 +276,7 @@ public class Utils {
 	 * @param item a text token
 	 * @return the actual http link
 	 */
+	@Deprecated
 	public static String getHttpToken(String originalItem) {
 		// apply pattern
 		String item = null;
@@ -374,6 +380,7 @@ public class Utils {
 	 * @param text
 	 * @return a String without html tags
 	 */
+	@Deprecated
 	public static String removeHTMLFromText(String text){
 
 		if(text == null)

src/main/java/org/gcube/socialnetworking/socialtoken/SocialMessageParser.java

@@ -23,7 +23,10 @@ public class SocialMessageParser {
 	private final String originalMessage;
 	
 	private StringWriter stringWriter;
+	
 	private List<ReplaceableToken> tokens;
+	private List<TagToken> tagTokens;
+	private List<URLToken> urlTokens;
 	
 	private List<String> hashtags;
 	
@@ -34,12 +37,18 @@ public class SocialMessageParser {
 	public List<ReplaceableToken> getTokens() {
 		if(tokens==null){
 			tokens = new ArrayList<>();
+			tagTokens = new ArrayList<>();
+			urlTokens = new ArrayList<>();
+			hashtags = new ArrayList<>();
+			
 			GCubeStringTokenizer socialStringTokenizer = new GCubeStringTokenizer(originalMessage);
 			for(Token token : socialStringTokenizer.getTokens()) {
 				String tokenString = token.getToken();
 				if(tokenString.startsWith("#")) {
 					TagToken tagToken = new TagToken(token);
 					tokens.add(tagToken);
+					tagTokens.add(tagToken);
+					hashtags.add(tagToken.getToken());
 					continue;
 				}
 				
@@ -47,6 +56,7 @@ public class SocialMessageParser {
 				if(url!=null) {
 					URLToken urlToken = new URLToken(token);
 					tokens.add(urlToken);
+					urlTokens.add(urlToken);
 					continue;
 				}
 				
@@ -60,13 +70,9 @@ public class SocialMessageParser {
 	public String getParsedMessage() {
 		if(stringWriter==null) {
 			stringWriter = new StringWriter();
-			hashtags = new ArrayList<>();
 			for(ReplaceableToken token : getTokens()) {
 				stringWriter.append(token.getTokenReplacement());
 				stringWriter.append(token.getDelimiter());
-				if(token instanceof TagToken) {
-					hashtags.add(token.getToken());
-				}
 			}
 		}
 		return stringWriter.toString();
@@ -108,9 +114,23 @@ public class SocialMessageParser {
 	}
 	
 	public List<String> getHashtags() {
-		if(stringWriter==null) {
-			getParsedMessage();
+		if(tokens==null){
+			getTokens();
 		}
 		return hashtags;
 	}
+	
+	public List<TagToken> getTagTokens() {
+		if(tokens==null){
+			getTokens();
+		}
+		return tagTokens;
+	}
+	
+	public List<URLToken> getURLTokens() {
+		if(tokens==null){
+			getTokens();
+		}
+		return urlTokens;
+	}
 }

src/main/java/org/gcube/socialnetworking/socialtoken/URLToken.java

@@ -9,6 +9,8 @@ import org.gcube.socialnetworking.tokenization.Token;
 
 public class URLToken extends ReplaceableToken {
 	
+	protected SanitizedURL sanitizedURL;
+	
 	public URLToken(String token, String delimiter, int start, int end) {
 		super(token, delimiter, start, end);
 	}
@@ -33,8 +35,7 @@ public class URLToken extends ReplaceableToken {
 			try {
 				Map<String,String> anchorAttibutes = new HashMap<>(1);
 				anchorAttibutes.put("target", "_blank");
-				SanitizedURL sanitizedURL = new SanitizedURL(token);
-				String url = sanitizedURL.getURL().toString();
+				String url = getExtractedURL().toString();
 				tokenReplacement = sanitizedURL.getPrefix() + ReplaceableToken.createLink(url, url, anchorAttibutes) + sanitizedURL.getPostfix();
 			}catch(MalformedURLException e) {
 				tokenReplacement = token;
@@ -43,5 +44,12 @@ public class URLToken extends ReplaceableToken {
 		}
 		return tokenReplacement;
 	}
+
+	public URL getExtractedURL() throws MalformedURLException {
+		if(sanitizedURL==null) {
+			sanitizedURL = new SanitizedURL(token);
+		}
+		return sanitizedURL.getURL();
+	}
 	
 }

src/test/java/org/gcube/social_networking/socialutillibrary/TestUnit.java

@@ -2,6 +2,7 @@ package org.gcube.social_networking.socialutillibrary;
 
 import java.util.List;
 
+import org.gcube.socialnetworking.socialtoken.SocialMessageParser;
 import org.gcube.socialnetworking.socialtoken.URLToken;
 import org.junit.Test;
 import org.slf4j.Logger;
@@ -38,5 +39,22 @@ public class TestUnit {
 			logger.debug(urlToken.getTokenReplacement());
 		}
 	}
+	
+	
+	protected String findFirstLink(String message) {
+		try {
+			SocialMessageParser messageParser = new SocialMessageParser(message);
+			List<URLToken> urlTokens = messageParser.getURLTokens();
+			return urlTokens.get(0).getExtractedURL().toString();
+		}catch (Exception e) {
+			return null;
+		}
+	}
+	
+	@Test
+	public void testTest() {
+		 String text = "Hello (https://doodle.com/poll/not-existing-poll)";
+		 logger.debug(findFirstLink(text));
+	}
 
 }