// Oldham, Jeffrey D.
// 2000 Jan 25
// CS1321

// Modified by:  Massingill, Berna L.
// 2000 Jan 27

// CS1321 Homework 3: Search Engine Code

#include <iostream.h>
#include <stdlib.h>		// has EXIT_SUCCESS
#include "types.h"
#include <fstream.h>
#include <strstream.h>


// We require a very strict (too strict) format for the database file:
// one line of keywords separated by _one_ space end ending with one
//   space followed by a newline character
// one line per document beginning with the document's name and
//  followed by as many doubles as there are keywords.  These doubles
//  are separated by spaces, and the last double is followed by a space.


// ==== FUNCTION DECLARATIONS ====

// Read the document names and vectors stored in the specified file.
// precondition: "databaseFileName" is the name of a file containing
//		   documents and vector contents.
// postconditions: "vdv" is filled with the documents and vectors
//		     specified in "databaseFileName".
//		   "km" is filled with the keywords
//		     specified in "databaseFileName".
//		   returns true only if succeeds (otherwise exits or
//                   returns false).
bool obtainDatabase(const char databaseFileName[],
		    vector<DocVec> & vdv,
		    KeywordMapping & km);

// Read the keyword mapping stored in the specified file.
// precondition: "databaseFileStream" has been successfully opened.
// postconditions: "km" is filled with the keywords.
//		   returns true if succeeds, false otherwise.
bool obtainKM(ifstream & databaseFileStream, KeywordMapping & km);

// Read the document names and vectors stored in the specified file.
// preconditions: "databaseFileStream" has been successfully opened.
//		  sz is the number of vector entries.
// postconditions: "vdv" is filled with the documents and vectors.
//		   returns true if succeeds, false otherwise.
bool obtainVDV(ifstream & databaseFileStream,
	       const std::vector<double>::size_type & sz,
	       vector<DocVec> & vdv);

// Read one document name and its vector stored on one line of the specified file.
// preconditions: "databaseFileStream" has been successfully opened.
//		  sz is the number of vector entries.
// postconditions: "dv" is filled with a document and a vector.
//		   returns true if succeeds, false otherwise.
bool obtainDV(ifstream & databaseFileStream,
	      const std::vector<double>::size_type & sz,
	      DocVec & dv);

// Ask the user for a list of keywords and the number of desired documents.
// preconditions: "km" has been filled using a call to obtainDatabase().
// postconditions: "searchVector" is filled with values corresponding to
//		    what the user wants to search for.
//		   "nuDesiredDocs" is the number of desired documents
//		     the user wants.  For example, she may want the
//		     top 10 documents.
//		   returns true if succeeds, false otherwise.
bool obtainSearchWords(const KeywordMapping & km,
		       vector<double> & searchVector, int& nuDesiredDocs);

// Prompt for and obtain a list of words.
// preconditions:  "prompt" contains text for the prompt.
// postconditions:  user is prompted for a list of words, all on one line.
//                  "words" contains words user enters in response.
//		   returns true if succeeds, false otherwise.
bool obtainWords(const string & prompt, vector<string> & words);

// Prompt for and obtain an integer.
// preconditions:  "prompt" contains text for the prompt.
// postconditions:  user is prompted for an integer.
//                 "i" contains integer user enters in response.
//		   returns true if succeeds, false otherwise.
bool obtainInteger(const string & prompt, int & i);

// Convert keywords to search vector.
// preconditions: "km" has been filled using a call to obtainDatabase().
//                 "words" contains words to convert.
// postconditions:  "searchVector" contains result of converting "words"
//		      and has unit length.
// Words not appearing in km are omitted from searchVector.
void convertWordsToSV(const KeywordMapping & km,
		      const vector<string> & words,
		      vector<double> & searchVector);

// 2000Feb04 JDO adds declaration:
// Convert to a unit vector.
// preconditions: "v" has nonzero length.
//		  type T can be divided by a double and multiplied with itself.
// postcondition: "v" has been normalized to have unit length.
// template <class T>
void unitVector(vector<double> & v);


// Feel free to ADD function declarations here. ============================


// ==== MAIN PROGRAM ====

// Command-line argument should be:
// 1. the filename of the database of documents and vectors

int main(int argc, char * argv[])
{
  KeywordMapping km;		// mapping from keywords to vector components
				// no need to look inside this code
  vector<DocVec> vdv;		// collection of documents and their vectors
  int nuDesiredDocs;		// Each search should produce this
				// many documents, e.g., the top 10 documents.

  if (argc != 2) {
    cerr << argv[0] << ": name-of-database-file\n";
    return EXIT_FAILURE;
  }

  // ADD code here ==========================================================

    return EXIT_SUCCESS;
}


// ==== FUNCTION DEFINITIONS ====

bool obtainDatabase(const char databaseFileName[],
		    vector<DocVec> & vdv,
		    KeywordMapping & km)
{
  ifstream dbfile;
  dbfile.open(databaseFileName);
  if (dbfile.fail())
    {
      cerr << "File " << databaseFileName << " not found.\n";
      exit(EXIT_FAILURE);
    }

  bool returnValue = obtainKM(dbfile, km) && obtainVDV(dbfile, km.size(), vdv);
  dbfile.close();
  return returnValue;
}

// -----------------

bool obtainKM(ifstream & databaseFileStream, KeywordMapping & km)
{
  string s;
  for (std::vector<double>::size_type index = 0;
       databaseFileStream.peek() != '\n' && getline(databaseFileStream,s,' ');
       ++index)
    km[s] = index;
  return databaseFileStream;
}

// -----------------

bool obtainVDV(ifstream & databaseFileStream,
	       const std::vector<double>::size_type & sz,
	       vector<DocVec> & vdv)
{
  DocVec dv;
  while (obtainDV(databaseFileStream, sz, dv))
    vdv.push_back(dv);
  return databaseFileStream.eof();
}

// -----------------

bool obtainDV(ifstream & databaseFileStream,
	      const std::vector<double>::size_type & sz,
	      DocVec & dv) {
  dv.second.resize(sz);
  if (databaseFileStream >> dv.first)
    for (std::vector<double>::size_type index = 0;
	 index < sz && databaseFileStream >> dv.second[index];
	 ++index)
      ;
  return databaseFileStream;
}

// -----------------

bool obtainSearchWords(const KeywordMapping & km,
		       vector<double> & searchVector, int& nuDesiredDocs)
{
  vector<string> keyWords;
  bool returnValue =
    obtainWords("keywords to search for", keyWords) &&
    obtainInteger("number of results you want to see", nuDesiredDocs);
  convertWordsToSV(km, keyWords, searchVector);
  return returnValue;
}

// ----------------

bool obtainWords(const string & prompt, vector<string> & words)
{
  string line, word;
  cout << "Please enter " << prompt << ", all on one line:\n";
  if (!getline(cin, line)) {
    cerr << "failed to enter the search words\n";
    return false;
  }
  istrstream linestream(line.c_str());
  while (linestream >> word)
    words.push_back(word);
  return true;
}

// ----------------

bool obtainInteger(const string & prompt, int & i)
{
  cout << "Please enter " << prompt << ":  ";
  return (cin >> i);
}

// ----------------

void convertWordsToSV(const KeywordMapping & km,
		      const vector<string> & words,
		      vector<double> & searchVector)
{
  KeywordMapping::const_iterator kmpos;
  searchVector.resize(km.size());
  for (vector<string>::const_iterator pos = words.begin();
       pos != words.end();
       ++pos)
    if ((kmpos = km.find(*pos)) != km.end()) // clearest code I ever
      ++searchVector[(*kmpos).second];	      // wrote, JDO :)
  unitVector(searchVector);
  return;
}

// ----------------

void unitVector(vector<double> & v) {
  // ADD code here
  return;
}

// ADD code here ============================================================