//An instance of this class maintains information on every word that occurs
//in a text file: the word, its frequency, the line-numbers where it occurs
//
import java.util.*;
import java.io.*;
public class Concordance
{
   //Instance Variables
   private String fileName;
   private ArrayList<WordRecord> table;
   
   //Constructor
   public Concordance(String fileName)
   {
      //Initialize Instance Variables
      this.fileName = fileName;
      this.table = new ArrayList<WordRecord>();
      
      //Construct this Concordance by scanning words from the text file
      try
      {
         //Open the text file, and create a scanner over the file
         Scanner source = new Scanner( new FileReader(fileName) );

         //Scan words from the file to create the concordance
         this.scanFile( source );
      }
      catch (FileNotFoundException fnfe)
      {
         throw new RuntimeException( fnfe.getMessage() );
      } 
   }
   
   //Constructor Helper
   //Update this concordance table with each word from the text file
   private void scanFile(Scanner source)
   {
      int lineNumber = 0;
      
      //Scan all lines from the text file
      while ( source.hasNext() )
      {
         //Input and count the next line from the file
         //Create a Scanner to process the input line
         lineNumber++;
         String input = source.nextLine().trim();
         Scanner stream = new Scanner( input );
         
         //Scan all "words" from the current input line
         while ( stream.hasNext() )
         {
            //Extract, standardize the next "word"
            String word = prune(stream.next()).toLowerCase() ;
            //Update the concordance with the "word"
            if (word.length() > 0)
               this.update( word, lineNumber );
         }
      }
   }
   
   //Accessor
   // @return: the name of the text file of this Concordance
   public String getFileName()
   {
      return this.fileName;
   }
   
   //Accessor
   // @return: the number of entries/words recorded in this Concordance
   public int getTableSize()
   {
      return this.table.size();
   }
   
   //Accessor
   // @return: a list of all (unique) words recorded in this Concordance
   public String[] getWords()
   {
      String[] words = new String[this.table.size()];
      for (int i = 0; i < this.table.size(); i++)
         words[i] = this.table.get(i).getWord();
      return words;
   }
   
   //File Query Method
   // @return: the total number of words processed in constructing this Concordance
   public int numberOfWordsInFile()
   {
      int count = 0;
      for (WordRecord record : this.table)
         count += record.getFrequency();
      return count;
   }
   
   //File Query Method
   // @return: the total number of text lines scanned in constructing this Concordance
   public int numberOfLinesInFile()
   {
      int maxLineNumber = 0;
      for (WordRecord record : this.table)
      {
         int hiIndex = record.getLineNumbers().length - 1;
         if (record.getLineNumbers()[hiIndex] > maxLineNumber)
            maxLineNumber = record.getLineNumbers()[hiIndex];
      }
      return maxLineNumber;
   }
   
   //Word Query Method
   // @param word: a word to be searched in this Concordance
   // @return: true iff parameter word is recorded in an entry of this Concordance
   public boolean contains(String word)
   {
      return this.lookUp( word ) != null;
   }
        
   //Word Query Method
   // @param word: a word to be searched in this Concordance
   // @return: the number of times (parameter) word occurs in the text file of this Concordance
   public int frequency(String word)
   {
      WordRecord record = this.lookUp( word );
      if (record != null)
         return record.getFrequency();
         
      return 0;
   }
   
   //Word Query Method
   // @param word: a word to be searched in this Concordance
   // @return: a list of the line-numbers of the text file of this Concordance where (parameter) word occurs
   public int[] lineNumbers(String word)
   {
      WordRecord record = this.lookUp( word );
      if (record != null)
         return record.getLineNumbers();
         
      return null;
   }
   
   //Override
   public String toString()
   {
      String image = "CONCORDANCE File: " + this.fileName;
      for (WordRecord entry : this.table)
         image += "\n" + entry;
      return image.trim();
   }
   
   //Eliminate leading and trailing white-space and punction symbols
   // @param word: a word to be "pruned" of white-space and punctuation
   // @return: the "pruned" word, meeting the following post-conditions
   //          2) a substring of the original word
   //          1) no LEADING or TRAILING white-space
   //          2) no LEADING or TRAILING punctuation symbols
   private static String prune(String word)
   {
      word = word.trim();
      if ( word.length() == 0 )
         return word;
         
      if ( isPunctuationSymbol(word.charAt(0)) )
         return prune( word.substring(1) );
      
      int endIndex = word.length() - 1 ;  
      if ( isPunctuationSymbol(word.charAt(endIndex)) )
         return prune( word.substring(0, endIndex) );

      return word;
   }
   
   private static boolean isPunctuationSymbol(char ch)
   {
      return ".,;:!?()'\"".indexOf(ch) != -1;
   }
   
   //Helper Method - Update a concordance with a word from the text file 
   //If the word is already in the concordance it's word-record is updated
   // with the line-number, otherwise a new word-record is created for the
   // new word and added to the concordance
   private void update(String word, int lineNumber)
   {
      WordRecord record = new WordRecord(word, lineNumber);
      
      //Locate the word in the Concordance table
      int index = this.table.size() - 1;
      while (index >= 0 && record.compareTo(this.table.get(index)) < 0)
         index--;
         
      //Update an existing WrodRecord, or insert a new WordRecord     
      if (index < 0 || record.compareTo(this.table.get(index)) > 0)
         this.table.add(index + 1, record);        //Insert a new WordRecord
      else
         this.table.get(index).update(lineNumber); //Update an existing WordRecord
   }
   
   //Helper Method - Find a given word in this Concordance
   // @param: a word to be searched in this Concordance
   // @return: a reference to the WordRecord for the given word in this Concordance 
   //          null if the parameter word is not recorded in this Concordance
   private WordRecord lookUp(String word)
   {
      WordRecord target = new WordRecord(prune(word).toLowerCase(), 1);
      
      int loIndex = 0;
      int hiIndex = this.table.size() - 1;
      while (loIndex <= hiIndex)
      {
         int middle = (loIndex + hiIndex)/2;
         int comparison = this.table.get(middle).compareTo(target);
         
         if ( comparison == 0 )
            return this.table.get(middle);
            
         if ( comparison < 0 )
            loIndex = middle + 1;
         else
            hiIndex = middle - 1;
      }
      return null;
   }
}