Merge pull request #120 from raumonmar1/Task-064

SpamDetector added

src/main/java/acme/components/SpamDetector.java

@@ -0,0 +1,67 @@
+package acme.components;
+
+import java.util.Arrays;
+import java.util.HashMap;
+import java.util.List;
+import java.util.Map;
+
+import acme.entities.systemConfiguration.SystemConfiguration;
+
+public class SpamDetector {
+
+	public static Boolean isSpam(final String text, final SystemConfiguration systemConfiguration) {
+		boolean result = false;
+
+		final List<String> wordsChecking = SpamDetector.getWords(text);		
+		final Map<String,Double> enTuples = SpamDetector.getSpamWords(systemConfiguration.getSpamTuplesEn());		
+		final Map<String,Double> esTuples = SpamDetector.getSpamWords(systemConfiguration.getSpamTuplesEs());
+		final Map<String,Double> spamTerms = new HashMap<>();
+		spamTerms.putAll(enTuples);
+		spamTerms.putAll(esTuples);
+
+		final Double spamThreshold = systemConfiguration.getSpamThreshold();
+
+		final Double spamRatio = SpamDetector.spam(wordsChecking, spamTerms);
+
+		if(spamRatio >= spamThreshold) {
+			result = true;
+		}
+
+		return result;
+	}
+
+	private static List<String> getWords(final String originalText){
+		return Arrays.asList(originalText.replaceAll("[.,:;/*=|()¡!¿?{}`´<>]"," ").replace("\""," ").replace("\\"," ")
+			.trim().split("\\s+"));		
+	}
+
+	private static Map<String,Double> getSpamWords(final String spamTuples){
+		final Map<String,Double> spamWords = new HashMap<String,Double>();
+
+		for(final String keyValue : spamTuples.split(",")) {
+			final String[] pair = keyValue.replace("("," ").replace(")"," ").replace("'", "").trim().split(":");
+			spamWords.put(pair[0], Double.valueOf(pair[1]));
+			}
+
+		return spamWords;
+	}
+
+	private static Double spam(final List<String> words, final Map<String,Double> spamTerms) {
+		Double spamWeight = 0.;
+		Integer palabrasDobles = 0;
+		String palabraAnterior= "";
+
+		for(final String word: words) {
+			if(spamTerms.keySet().contains(word.toLowerCase())) {
+				spamWeight += spamTerms.get(word.toLowerCase());
+			}
+			if(spamTerms.keySet().contains(palabraAnterior.toLowerCase() + " " + word.toLowerCase())) {
+				spamWeight += spamTerms.get(palabraAnterior.toLowerCase() + " " + word.toLowerCase());
+				palabrasDobles += 1;
+			}
+			palabraAnterior = word;
+		}
+
+		return (spamWeight/(words.size() - 1 * palabrasDobles)) * 10;
+	}	
+}

src/main/java/acme/entities/systemConfiguration/SystemConfiguration.java

@@ -26,7 +26,10 @@ public class SystemConfiguration extends AbstractEntity{
 	public String systemCurrency;
 
 	@NotBlank
-	public String spamTuples;
+	public String spamTuplesEn;
+
+	@NotBlank
+	public String spamTuplesEs;
 
 	@NotNull
 	@Range(min = 0, max = 1)

src/main/webapp/WEB-INF/resources/initial-data/system-configuration.csv

@@ -1,2 +1,2 @@
-key,accepted-currencies,system-currency,spam-tuples,spam-threshold,money-exchange-name,money-exchange-link
-system-configuration-01,"EUR,USD,GBP",EUR,"(sex,0.10),(sexo,0.10),(viagra,0.10),(cialis,0.10),(hard core,0.10),(muy duro,0.10),(sexy,0.05),(nigeria,0.05),(you've won,0.05),(has ganado,0.05),(one million,0.05),(un millon,0.05)",0.10,Exchange rate API,https://exchangerate.host
+key,accepted-currencies,system-currency,spam-tuples-en,spam-tuples-es,spam-threshold,money-exchange-name,money-exchange-link
+system-configuration-01,"EUR,USD,GBP",EUR,"(sex,0.10),(viagra,0.10),(cialis,0.10),(hard core,0.10),(sexy,0.05),(nigeria,0.05),(you've won,0.05),(one million,0.05)","(sexo,0.10),(viagra,0.10),(cialis,0.10),(muy duro,0.10),(sexy,0.05),(nigeria,0.05),(has ganado,0.05),(un millon,0.05)",0.10,Exchange rate API,https://exchangerate.host