package compbio.data.sequence;

import java.io.BufferedReader;
import java.io.InputStream;
import java.io.InputStreamReader;
import java.io.IOException;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.List;
import java.util.Scanner;
import java.util.TreeSet;
import java.util.regex.Pattern;

import org.apache.log4j.Logger;

// Utility class for reading alifold output

public class RNAStructReader {

	private static Logger log = Logger.getLogger(RNAStructReader.class);
	
	// Whitespace patterns
	static String s = "[+\\s=]+";
	static String bracket = "\\(|\\)|\\{|\\}|\\[|\\]";
	static String notData = "[\\s=+]+";

	// RNAOut data type patterns 
	static String seqP = "[_\\-a-zA-Z]{2,}"; // Has to match --mis output aswell (not just ACGU_)
	static String structP = "[\\.)({}\\[\\],]{2,}";
	static String floatP = "-?\\d+\\.\\d*(e[+\\-]\\d+)?";
	static String energyP = "-?[0-9]*\\.?[0-9]{2}";
	static String freqP = "^-?\\d\\.\\d{6,}(e[+\\-]\\d+)?$";
	
	// alifold out line patterns
	static String ps = "\\s*";
	static String alignmentP = "^"+seqP+ps+"$";
	static String mfeStructP = "^"+structP+s+"\\("+ps+floatP+s+floatP+s+floatP+ps+"\\)"+ps+"$";
	static String justStructP = "^"+structP+ps+"$";
	static String stochBTStructP = "^"+structP+s+floatP+s+floatP+ps+"$";
	static String PStructP = "^"+structP+s+"\\["+ps+floatP+ps+"\\]"+ps+"$";
	static String centStructP = "^"+structP+s+floatP+ps+"\\{"+ps+floatP+s+floatP+ps+"\\}"+ps+"$";
	static String MEAStructP = "^"+structP+s+"\\{"+ps+floatP+s+"MEA="+floatP+ps+"\\}"+ps+"$";
	static String freeEnergyP = "^"+ps+"free energy of ensemble"+ps+"="+ps+floatP+ps+"kcal/mol"+ps+"$";
	static String ensembleFreqP = "^"+ps+"frequency of mfe structure in ensemble "+floatP+ps+"$";
	
	
	public static RNAStructScoreManager readRNAStructStream(InputStream stdout)
			throws IOException {
		
		String error = "Error in parsing alifold stdout file: ";
		// The Lists required to construct a ScoreManager Using the new constructor
		List<String> structs = new ArrayList<String>();
		List<TreeSet<Score>> data = new ArrayList<TreeSet<Score>>();

		// Allocate necessry data structures for creating Score objects
		ArrayList<Float> scores = new ArrayList<Float>();

		BufferedReader reader = new BufferedReader(new InputStreamReader(stdout));
		// The first 2 lines of the alifold stdout file are always the same format
		String fline = reader.readLine();
		assert (Pattern.matches(AlifoldLine.alignment.regex, fline)) :
			error + "Sequence Alignment Expected";
		structs.add(fline.trim());
		data.add(newEmptyScore(AlifoldResult.consensusAlignment));
		
		fline = reader.readLine();
		assert (Pattern.matches(AlifoldLine.mfeStruct.regex, fline)) :
			error + "Consensus Structure and Energy Expected";
		Scanner sc = new Scanner(fline);
		structs.add(sc.next());
		for (int i = 0; i < 3; i++) {
			scores.add(Float.parseFloat(sc.findInLine(floatP)));
		}
		data.add(newSetScore(AlifoldResult.mfeStructure, scores));
		
		// Now the alifold stdout file formats diverge based on arguments
		fline = reader.readLine();
		String sline;
		Scanner nsc = null;
		while ( fline != null) {
			scores.clear();
			AlifoldLine ftype = identifyLine(fline);
			sline = reader.readLine(); // Look ahead
			sc = new Scanner(fline);
			if (sline != null) nsc = new Scanner(sline);

			if (ftype.equals(AlifoldLine.PStruct)) {
				// The -p or --MEA option is specified
				// The next line should always be frequency of mfe structure
				assert ( sline != null && Pattern.matches(AlifoldLine.ensembleFreq.regex, sline)) :
					error + "Expected frequency of mfe structure";
				structs.add(sc.next());
				scores.add(Float.parseFloat(sc.findInLine(floatP)));
				scores.add(Float.parseFloat(nsc.findInLine(floatP)));
				data.add(newSetScore(AlifoldResult.contactProbabilityStructure, scores));
				// Jump line
				sline = reader.readLine();
			}
			else if (ftype.equals(AlifoldLine.centStruct)) {
				structs.add(sc.next());
				for (int i = 0; i < 3; i++) {
					scores.add(Float.parseFloat(sc.findInLine(floatP)));
				}
				data.add(newSetScore(AlifoldResult.centroidStructure, scores));
			}
			else if (ftype.equals(AlifoldLine.MEAStruct)) {
				structs.add(sc.next());
				for (int i = 0; i < 2; i++) {
					scores.add(Float.parseFloat(sc.findInLine(floatP)));
				}
				data.add(newSetScore(AlifoldResult.MEAStucture, scores));
			}
			else if (ftype.equals(AlifoldLine.justStruct)) {
				structs.add(sc.next());
				data.add(newEmptyScore(AlifoldResult.stochBTStructure));
			}
			else if (ftype.equals(AlifoldLine.stochBTStruct)) {
				structs.add(sc.next());
				scores.add(sc.nextFloat());
				scores.add(sc.nextFloat());
				data.add(newSetScore(AlifoldResult.stochBTStructure, scores));
			}
			else if (ftype.equals(AlifoldLine.freeEnergy)) {
				assert (sline != null 
						&& Pattern.matches(AlifoldLine.ensembleFreq.regex, sline)) :
						error + "Found 'freeEnergy' line on its own";
				structs.add("Free energy of ensemble (kcal/mol) followed by "
						+ "frequency of mfe structure in ensemble");
				scores.add(Float.parseFloat(sc.findInLine(floatP)));
				scores.add(Float.parseFloat(nsc.findInLine(floatP)));
				data.add(newSetScore(AlifoldResult.ensembleValues, scores));
				// jump line
				sline = reader.readLine();
			}
			

			assert(!ftype.equals(AlifoldLine.ensembleFreq)) :
				error + "Wasn't expecting 'frequency of mfe structure'!";
			assert(!ftype.equals(AlifoldLine.mfeStruct)) :
				error + "'Standard output' line at a place other than line 2!";
			assert(!ftype.equals(AlifoldLine.alignment)) :
				error + "Wasn't expecting an alignment sequence!";
			assert(!ftype.equals(AlifoldLine.OTHER)) :
				error + "Wasn't expecting this whatever it is: " + fline;
			if (Pattern.matches("^\\s*$", fline)) {
				log.warn("While parsing alifold stdout: A line is either empty or"
						+ " contains only whitespace");
			}
			
			fline = sline;
		}
				
		sc.close();
		if (nsc != null) nsc.close();
		
		return new RNAStructScoreManager(structs, data);
	}
	
	// Just for the purpose of creating new TreeSet<Score> objects of length one
	// for adding to a 'data' list to make a ScoreManager
	private static TreeSet<Score> newSetScore(Enum<?> res, List<Float> scores) {
		// first convert List<Float> to float[]
		float[] scoresf = new float[scores.size()];
		Float f;
		for (int i = 0; i < scoresf.length; i++) {
			f = scores.get(i);
			scoresf[i] = ( f != null ? f : Float.NaN);
		}
		return new TreeSet<Score>(Arrays.asList(new Score(res, scoresf)));
	}

	// A method just for the purpose of neatly creating Almost Empty score objects
	// that can't be null
	public static TreeSet<Score> newEmptyScore(Enum<?> res) {
		return new TreeSet<Score>(Arrays.asList(new Score(res, new float[0])));
	}

	public static RNAStructScoreManager readRNAStructStream(InputStream stdout, 
			InputStream alifold) throws IOException {
		
		// The Lists required to construct a ScoreManager Using the new constructor
		List<String> structs;
		List<TreeSet<Score>> data; 
		
		// Get a ScoreManager that takes the std output but ignores alifold.out (-p)
		RNAStructScoreManager stdSM = readRNAStructStream(stdout);
		
		// Unpack this into the structs and data lists
		structs = stdSM.getStructs();
		data = stdSM.getData();
		
		// Now parse alifold.out
		Scanner sc = new Scanner(alifold);
		sc.useDelimiter("[\\s%]+");
		
		// jump two lines to the data 
		sc.nextLine(); sc.nextLine();
		
		// Read the first, second and fourth columns. Ignoring everything else.
		// Allocate necessry data structures for creating Score objects
		ArrayList<Float> scores = new ArrayList<Float>();
		List<Range> rangeHolder = new ArrayList<Range>();
		String s = "null";
		while (true) {
			s = sc.next();
			if (java.util.regex.Pattern.matches("^[\\.)(]{2,}$", s)) break;
			if (!sc.hasNextLine()) break;
			int t = sc.nextInt();
			rangeHolder.add(new Range(Integer.parseInt(s), t));
			sc.next();
			scores.add(sc.nextFloat());
			sc.nextLine();
		}
		sc.close();
		
		// Update the first ScoreHolder TreeSet<Score> element
		assert (rangeHolder.size() == scores.size());
		TreeSet<Score> sHolder = new TreeSet<Score>();
		for (int i = 0; i < rangeHolder.size(); i++) {
			ArrayList<Float> singleS = new ArrayList<Float>(Arrays.asList(scores.get(i)));
			TreeSet<Range> singleR = new TreeSet<Range>(Arrays.asList(rangeHolder.get(i)));
			sHolder.add(new Score(AlifoldResult.contactProbabilities, singleS, singleR));
		}
		
		data.set(0, sHolder);
		
		return new RNAStructScoreManager(structs, data);
	}

	private static RNAOut identify(String token) {
		if (Pattern.matches(seqP, token)) {
			return RNAOut.SEQ;
		} else if (Pattern.matches(structP, token)) {
			return RNAOut.STRUCT;
		} else if (Pattern.matches(energyP, token)) {
			return RNAOut.ENERGY;
		} else if (Pattern.matches(freqP, token)) {
			return RNAOut.FREQ;
		}
		
		return RNAOut.OTHER;
	}
	
	private static AlifoldLine identifyLine(String line) {
		
		for (AlifoldLine il : AlifoldLine.values()) {
			if (Pattern.matches(il.regex, line)) return il;
		}
		return AlifoldLine.OTHER;
	}
	
	static enum AlifoldLine {
		mfeStruct (mfeStructP),
		justStruct (justStructP),
		stochBTStruct (stochBTStructP),
		PStruct (PStructP),
		centStruct (centStructP),
		MEAStruct (MEAStructP),
		freeEnergy (freeEnergyP),
		ensembleFreq (ensembleFreqP),
		alignment (alignmentP), 
		OTHER (".*");
		
		String regex;
		AlifoldLine(String regex) { this.regex = regex; }

	}
	
	//The types of data in an RNAalifold stdout file
	static enum RNAOut {
		SEQ, STRUCT, ENERGY, FREQ, OTHER
	}

	//Something to put in the Score objects of the alifold result which gives information
	//about what kind of sequence it is holding in its String Id.

	public static enum AlifoldResult {
		mfeStructure, contactProbabilityStructure, MEAStucture, centroidStructure, stochBTStructure, consensusAlignment, ensembleValues, contactProbabilities
	}
	

	// Print the full regex Strings for testing 
	public static void main(String[] args) {
		for (AlifoldLine l : AlifoldLine.values()) {
			System.out.println(l.toString() + ": " + l.regex.replace("^","").replace("$",""));
		}
	}
	

	
}