Adding Pauper in the stop-words list

All supported format names in stopwords have been grouped into a constant for easy update and re-use.
umekonn · May 20, 2023 · f03a804 · f03a804
1 parent 5ceb541
commit f03a804
Showing 1 changed file with 5 additions and 5 deletions.
diff --git a/forge-lda/src/main/java/forge/lda/LDAModelGenetrator.java b/forge-lda/src/main/java/forge/lda/LDAModelGenetrator.java
@@ -38,6 +38,7 @@
  */
 public final class LDAModelGenetrator {
 
+    public static final String SUPPORTED_LDA_FORMATS = "Historic|Modern|Pioneer|Standard|Legacy|Vintage|Pauper";
     public static Map<String, Map<String,List<List<Pair<String, Double>>>>> ldaPools = new HashMap<>();
     public static Map<String, List<Archetype>> ldaArchetypes = new HashMap<>();
 
@@ -167,7 +168,6 @@ public static Map<String,List<List<Pair<String, Double>>>> loadFormat(GameFormat
 
     public static List<Archetype> initializeFormat(GameFormat format) throws Exception{
         Dataset dataset = new Dataset(format);
-
         //estimate number of topics to attempt to find using power law
         final int numTopics = Float.valueOf(347f*dataset.getNumDocs()/(2892f + dataset.getNumDocs())).intValue();
         System.out.println("Num Topics = " + numTopics);
@@ -223,8 +223,8 @@ public static List<Archetype> initializeFormat(GameFormat format) throws Excepti
             }
             LinkedHashMap<String, Integer> wordCounts = new LinkedHashMap<>();
             for( Deck deck: decks){
-                String name = deck.getName().replaceAll(".* Version - ","").replaceAll(" \\((Historic|Modern|Pioneer|Standard|Legacy|Vintage), #[0-9]+\\)","");
-                name = name.replaceAll("\\(Historic|Modern|Pioneer|Standard|Legacy|Vintage|Fuck|Shit|Cunt\\)","");
+                String name = deck.getName().replaceAll(".* Version - ","").replaceAll(" \\((" + SUPPORTED_LDA_FORMATS + "), #[0-9]+\\)","");
+                name = name.replaceAll("\\(" + SUPPORTED_LDA_FORMATS + "|Fuck|Shit|Cunt|Ass|Arse|Dick|Pussy\\)","");
                 String[] tokens = name.split(" ");
                 for(String rawtoken: tokens){
                     String token = rawtoken.toLowerCase();
@@ -256,7 +256,7 @@ public static List<Archetype> initializeFormat(GameFormat format) throws Excepti
             System.out.println("============ " + deckName);
             System.out.println(decks.toString());
 
-            unfilteredTopics.add(new Archetype(topRankVocabs,deckName,decks.size()));
+            unfilteredTopics.add(new Archetype(topRankVocabs, deckName, decks.size()));
         }
         Comparator<Archetype> archetypeComparator = new Comparator<Archetype>() {
             @Override
@@ -376,7 +376,7 @@ public static void updateLegendMatrix(Deck deck, PaperCard legend, Map<String, I
                     int old = matrix[legendIntegerMap.get(legend.getName())][cardIntegerMap.get(pairCard.getName())];
                     matrix[legendIntegerMap.get(legend.getName())][cardIntegerMap.get(pairCard.getName())] = old + 1;
                 }catch (NullPointerException ne){
-                    //Todo: Not sure what was failing here
+                    //TODO: Not sure what was failing here
                     ne.printStackTrace();
                 }
             }