Overlapping ConcurrentHashMap puts using putIfAbsent

Vote count:

0

There seems to be an issue with inserting into the hashtable. I create about 8 threads, and in each thread I do the following code. Each thread receives a char[] array. The job of each thread is to tokenize this array (look for spaces). Once a token is found, I need to add it to the hashtable if it doesn't exist. If it does exist, then I need to add 1 to the current value of that token (the key).

Questions you might ask:

Why not convert from char[] to String?

I tried this, and since strings are immutable, I eventually ran out of memory (I am processing a 10g file), or I spend too long garbage collecting. With Character[], I am able to reuse the same variable and not take up extra space in memory.

What is the issue?

When I am done processing the entire file, I run the code:

for (Entry<Character [], Integer> e : wordCountMap.entrySet()) {
    System.out.println(Arrays.toString(e.getKey()) + " = " + e.getValue());
}

in my main function. What I get as a result is less than 100 key/value pairs. I know that there should be around 20,000. There somehow seems to be some overlap.

    Character [] charArray = new Character[8];
    for (i = 0; i < newbyte.length; i++) { //newbyte is a char[] from main
        if (newbyte[i] != ' ') {
            charArray[counter] = newbyte[i];
            counter++;
        }
        else { 
            check = wordCountMap.putIfAbsent(charArray, 1);
            if (check != null) { 
                wordCountMap.put(charArray, wordCountMap.get(charArray) + 1);
            }
            for (j = 0; j < counter; j++) {
                charArray[j] = null;
            }//Null out the array

ConcurrentMap<Character [], Integer> wordCountMap //this is the definition in main

asked 1 min ago

DanGordon

1