From c42186671c9808a7192504f5556c48718607f7d3 Mon Sep 17 00:00:00 2001
From: Peter Wu <peter@lekensteyn.nl>
Date: Wed, 1 Apr 2015 09:19:46 +0200
Subject: Count unigrams

---
 spellchecker/src/CorpusReader.java | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/spellchecker/src/CorpusReader.java b/spellchecker/src/CorpusReader.java
index 536a41b..f815cfd 100644
--- a/spellchecker/src/CorpusReader.java
+++ b/spellchecker/src/CorpusReader.java
@@ -15,6 +15,7 @@ public class CorpusReader {
 
     private HashMap<String, Integer> ngrams;
     private Set<String> vocabulary;
+    private int unigramCount = 0;
 
     public CorpusReader() throws IOException {
         readNGrams();
@@ -56,6 +57,10 @@ public class CorpusReader {
             try {
                 count = Integer.parseInt(s1);
                 ngrams.put(s2, count);
+                // unigram
+                if (s2.indexOf(' ') == -1) {
+                    unigramCount += count;
+                }
             } catch (NumberFormatException nfe) {
                 throw new NumberFormatException("NumberformatError: " + s1);
             }
@@ -115,4 +120,8 @@ public class CorpusReader {
 
         return smoothedCount;
     }
+
+    public int getUnigramCount() {
+        return unigramCount;
+    }
 }
-- 
cgit v1.2.1