package anotacja;

public class BCBagOfWords {
	
	public InvertedIndex[] index;
	
	public BCBagOfWords(InvertedIndex[] index) {
		this.index = index;
	}
	
	public float[] measure(QueryData query) {
		
		//extract the most important information for BoW
		int[] data = query.groups;
		
		//tablica pomiarem podobienstwa dla calej bazy
		float[] similarity = new float[InvertedIndex.content.size()];
		
		//wyznacz L1 norme dla zapytania
		float queryL1 = 0;
		for (int i = 0; i < data.length; i++) {
			int clust = data[i];
			queryL1 += InvertedIndex.idf[clust];
		}
					
		//System.out.print("INDEX: ");
		int clust = -1; int cQuery = 0;
		for (int i = 0; i < data.length; i++) {
			cQuery++;
			if ((i == data.length - 1) || (data[i] != data[i + 1])) {
				clust = data[i];
				if (this.index[clust] == null) {
					this.index[clust] = new InvertedIndex(clust);
					if (!this.index[clust].readIndex()) {
						//nastapil blad odczytu, przerwij
						return null;
					}
					if (this.index[clust].images == null) {
						//indeks nie zostal wczytany, przerwij
						return null;
					}
				}
				float vQuery = (float)Math.sqrt(cQuery * InvertedIndex.idf[clust] / (data.length - 1));
				
				int cDatab = 0;
				int[] clInd = this.index[clust].images;
				for (int j = 0; j < clInd.length - 1; j++)
				{
					cDatab++;
					if (clInd[j] != clInd[j + 1])
					{
						//if (clInd[j] == 43) { System.out.print(clust + "/" + cDatab + " "); }
						float vDatab = (float)Math.sqrt(cDatab * InvertedIndex.idf[clust] / InvertedIndex.len[clInd[j]]);
						similarity[clInd[j]] += vQuery * vDatab;
						cDatab = 0;
					}
				}
				cQuery = 0;
			}
		}
		//System.out.println();
		
		//System.out.print("SIM: ");
		//for (int i = 0; i < content.size(); i++) { System.out.format("%7.5f ", similarity[i]); }
		//System.out.println();
		
		for (int i = 0; i < InvertedIndex.content.size(); i++)
		{
			float norm = (float)Math.sqrt((queryL1 / (data.length - 1)) * (InvertedIndex.nL1[i] / InvertedIndex.len[i]));
			similarity[i] = similarity[i] / norm;
			//za duzo punktow kluczowych, utnij podobienstwo
			if (InvertedIndex.len[i] > 7000) similarity[i] = 0.001f;
		}
		
		return similarity;
		
		//System.out.print("S-N: ");
		//for (int i = 0; i < content.size(); i++) { System.out.format("%7.5f ", similarity[i]); }
		//System.out.println();
		
		//List<Ordered<String>> result = new ArrayList<Ordered<String>>();
		//for (int i = 0; i < content.size(); i++) result.add(new Ordered<String>(content.get(i), similarity[i]));
		//Collections.sort(result);
	}
}