// lab2.cc
// Sequence matching driver for CS 241 Lab 2, Spring 2003
//
// SYNTAX: lab2 <corpus file> <pattern file> [ <mask file> ]
//
// This sequence matching program reads two sequences, a CORPUS and a PATTERN,
// from files on disk and finds all strings of a given length M (given
// interactively by the user) that occur in both corpus and pattern.  
// Matching substrings are printed in the form
//      <index 1> <index 2> <substring>
// where the two indices are the offsets of the match within the corpus and
// pattern, respectively, and <substring> is the actual matching string.
//
// As an optional third argument, the program can take a file containing a
// MASK sequence.  Substrings of the mask sequence are considered
// "uninteresting" and so must not be reported by the matching code.
// To implement this requirement, we delete any occurrences of substrings
// in the mask sequence from our pattern table before performing the search.
//

#include <iostream>

using namespace std;

#include "SeqReader.h"
#include "StringDictionary.h"

//
// local prototypes
//
static StringDictionary *createTable(const char *patternSeq, int patternLength,
				int matchLength);
static void maskTable(StringDictionary *table, const char *maskSeq, int maskLength,
		      int matchLength);
static void findMatches(StringDictionary *table, const char *corpusSeq, 
			int corpusLength, int matchLength);

int main(int argc, char *argv[]){
  const char *corpusSeq  = NULL;
  const char *patternSeq = NULL;
  const char *maskSeq    = NULL;
  int corpusLength, patternLength, maskLength;
  
  if (argc < 3) {
    cout << "Syntax: Lab2 <corpus file> <pattern file> [<mask file>]\n";
    exit(1);
  }
  else {
    corpusSeq  = readSeq(argv[1], &corpusLength);
    patternSeq = readSeq(argv[2], &patternLength);

    cout << "CORPUS: "  << corpusLength << " bases" << endl;;
    cout << "PATTERN: " << patternLength << " bases" << endl;
      
    if (argc > 3){
      maskSeq = readSeq(argv[3], &maskLength);
      cout << "MASK: "  << maskLength << " bases" << endl;
    }
  }
  
  // Interactively get the desired match length
  //
  int matchLength;
  cout << "Match length? ";
  cin >> matchLength;
  
  StringDictionary *table = createTable(patternSeq, patternLength, matchLength);

  cout << "\nAfter creating the table, it holds "; 
  cout << table->size() << " sequences of length " << matchLength << endl;
  
  if (maskSeq != NULL)
    maskTable(table, maskSeq, maskLength, matchLength);
  
  cout << "After removing the mask sequences, the table holds ";
  cout << table->size() << " sequences of length " << matchLength << endl << endl;
  
  findMatches(table, corpusSeq, corpusLength, matchLength);
  return 0;
}


// Create a new StringDictionary containing all substrings of the pattern
// sequence.
// 
static StringDictionary *createTable(const char *patternSeq, int patternLength,
				int matchLength)
{
  StringDictionary *table = new StringDictionary(patternLength, matchLength);
  Record *rec;

  for (int j = 0; j < patternLength - matchLength + 1; j++){
    const char *key = patternSeq + j;

    if (table->contains(key)){
      rec = table->get(key);
      rec->positions.add(j);
    }
    else {
      rec = new Record();
      rec->positions.add(j);
      table->put(key,rec);
    }
  }
  return table;
}


// Remove all substrings in the mask sequence from a StringDictionary.
// 
static void maskTable(StringDictionary *table, const char *maskSeq, int maskLength,
		      int matchLength){
  for (int j = 0; j < maskLength - matchLength + 1; j++){
    const char *key = maskSeq + j;
    Record* rec = table->remove(key);
    if (rec != NULL)
      delete rec;           // don't need this struct anymore
  }
}


// Find and print all matches between the corpus sequence and any
// string in a StringDictionary.
//
static void findMatches(StringDictionary *table, const char *corpusSeq, 
			int corpusLength, int matchLength){
  int numMatches = 0;
  for (int j = 0; j < corpusLength - matchLength + 1; j++){
    const char *key = corpusSeq + j;
    if (table->contains(key)){
      const Record *rec = table->get(key);
      for (unsigned int k = 0; k < rec->positions.length(); k++){
	cout << j << ' ' << rec->positions[k] << ' ';
	for (int p = 0; p < matchLength; p++)
	  cout << key[p];
	numMatches++;
	cout << endl;
      }
    }
  }
  cout << "\nThere were " << numMatches << " matches found." << endl;
}