/* Frobby: Software for monomial ideal computations.
   Copyright (C) 2007 Bjarke Hammersholt Roune (www.broune.com)

   This program is free software; you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by
   the Free Software Foundation; either version 2 of the License, or
   (at your option) any later version.

   This program is distributed in the hope that it will be useful,
   but WITHOUT ANY WARRANTY; without even the implied warranty of
   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
   GNU General Public License for more details.

   You should have received a copy of the GNU General Public License
   along with this program.  If not, see http://www.gnu.org/licenses/.
*/
#include "stdinc.h"
#include "randomDataGenerators.h"

#include "BigIdeal.h"
#include "Ideal.h"
#include "Term.h"
#include "error.h"
#include "FrobbyStringStream.h"

#include <limits>
#include <ctime>
#include <unistd.h>

void generateLinkedListIdeal(BigIdeal& ideal, size_t variableCount) {
  VarNames names(variableCount);
  ideal.clearAndSetNames(variableCount);
  ideal.reserve(variableCount);
  for (size_t var = 1; var < variableCount; ++var) {
    ideal.newLastTerm();
    ideal.getLastTermExponentRef(var) = 1;
    ideal.getLastTermExponentRef(var - 1) = 1;
  }
}

void generateChessIdeal(BigIdeal& bigIdeal,
                        size_t rowCount,
                        size_t columnCount,
                        int deltaRow[],
                        int deltaColumn[],
                        size_t deltaCount) {
  if (mpz_class(rowCount) * mpz_class(columnCount) >
      numeric_limits<size_t>::max())
    reportError("Number of positions on requested chess board too large.");

  // Generate names
  VarNames names;
  for (size_t row = 0; row < rowCount; ++row) {
    for (size_t column = 0; column < columnCount; ++column) {
      FrobbyStringStream name;
      name << 'r' << (row + 1) << 'c' << (column + 1);
      names.addVar(name);
    }
  }
  bigIdeal.clearAndSetNames(names);
  Ideal ideal(bigIdeal.getVarCount());

  // Generate ideal
  for (size_t row = 0; row < rowCount; ++row) {
    for (size_t column = 0; column < columnCount; ++column) {
      for (size_t delta = 0; delta < deltaCount; ++delta) {
        // Check that the target position is within the board.

        if (deltaRow[delta] == numeric_limits<int>::min() ||
            (deltaRow[delta] < 0 &&
             row < (size_t)-deltaRow[delta]) ||
            (deltaRow[delta] > 0 &&
             rowCount - row <= (size_t)deltaRow[delta]))
          continue;

        if (deltaColumn[delta] == numeric_limits<int>::min() ||
            (deltaColumn[delta] < 0 &&
             column < (size_t)-deltaColumn[delta]) ||
            (deltaColumn[delta] > 0 &&
             columnCount - column <= (size_t)deltaColumn[delta]))
          continue;

        Term chessMove(ideal.getVarCount());
        chessMove[row * columnCount + column] = 1;

        size_t targetRow = row + deltaRow[delta];
        size_t targetColumn = column + deltaColumn[delta];
        ASSERT(targetRow < rowCount);
        ASSERT(targetColumn < columnCount);

        chessMove[targetRow * columnCount + targetColumn] = 1;
        ideal.insert(chessMove);
      }
    }
  }

  ideal.sortReverseLex();
  bigIdeal.insert(ideal);
}

void generateKingChessIdeal(BigIdeal& ideal, size_t rowsAndColumns) {
  int deltaRow[]    = {-1, 0, 1, 1}; // the other moves follow by symmetry
  int deltaColumn[] = { 1, 1, 1, 0};
  ASSERT(sizeof(deltaRow) == sizeof(deltaColumn));

  size_t deltaCount = sizeof(deltaRow) / sizeof(int);

  generateChessIdeal(ideal, rowsAndColumns, rowsAndColumns,
                     deltaRow, deltaColumn, deltaCount);
}

void generateKnightChessIdeal(BigIdeal& ideal, size_t rowsAndColumns) {
  int deltaRow[]    = {-1,  1, 2,  2}; // the other moves follow by symmetry
  int deltaColumn[] = { 2,  2, 1, -1};
  ASSERT(sizeof(deltaRow) == sizeof(deltaColumn));

  size_t deltaCount = sizeof(deltaRow) / sizeof(int);

  generateChessIdeal(ideal, rowsAndColumns, rowsAndColumns,
                     deltaRow, deltaColumn, deltaCount);
}

namespace {
  /** Consider the entries of pattern as the bits in a binary
   number. This method adds 1 and returns true if the resulting
   pattern is all zeroes. */
  bool nextBitPattern(vector<char>& pattern) {
    typedef vector<char>::iterator iterator;
    for (iterator it = pattern.begin(); it != pattern.end(); ++it) {
      if (*it)
        *it = 0;
      else {
        *it = 1;
        ASSERT(pattern != vector<char>(pattern.size()));
        return true;
      }
    }

    ASSERT(pattern == vector<char>(pattern.size()));
    return false;
  }
}

void generateTreeIdeal(BigIdeal& ideal, size_t varCount) {
  ideal.clearAndSetNames(VarNames(varCount));

  // Declare outside of loop to avoid repeated initialization.
  mpz_class exponent;

  // Using vector<char> to avoid vector<bool> which has special
  // properties. Going through all "bit" patterns by simulating adding
  // one at each step. pattern starts at all zero, which represents
  // the identity, so we take the next bit pattern even in the first
  // iteration to go past that.
  vector<char> pattern(varCount);
  while (nextBitPattern(pattern)) {
    size_t setSize = 0;
    typedef vector<char>::iterator iterator;
    for (iterator it = pattern.begin(); it != pattern.end(); ++it)
      setSize += (size_t)*it;

    exponent = varCount - setSize + 1;
    ideal.newLastTerm();
    for (size_t var = 0; var < varCount; ++var)
      if (pattern[var])
        ideal.getLastTermExponentRef(var) = exponent;
  }
}

void generateRookChessIdeal(BigIdeal& bigIdeal, size_t n, size_t k) {
  if (n == 0 || k == 0)
    reportError("One side of rook ideal has zero vertices.");
  if (n > 1000 || k > 1000)
    reportError("Number of variables in rook ideal too large.");
  if (n > k)
	std::swap(n, k);

  size_t varCount = n * k;
  Ideal ideal(varCount);
  Term term(varCount);

  vector<char> taken(k);
  vector<size_t> choice(n);
  size_t level = 0;
  while (true) {
	if (choice[level] == k) {
	  if (level == 0)
		break;
	  --level;
	  ASSERT(static_cast<bool>(taken[choice[level]]) == true);
      ASSERT(term[level * k + choice[level]] == 1);
	  taken[choice[level]] = false;
      term[level * k + choice[level]] = 0;
	  ++choice[level];
	  continue;
	}
    if (taken[choice[level]]) {
	  ++choice[level];
	  continue;
	}
	taken[choice[level]] = true;
    ASSERT(term[level * k + choice[level]] == 0);
	term[level * k + choice[level]] = 1;

    if (level < n - 1) {
	  ++level;
	  choice[level] = 0;
	} else {
	  ideal.insert(term);
	  ASSERT(static_cast<bool>(taken[choice[level]]) == true);
      ASSERT(term[level * k + choice[level]] == 1);
	  taken[choice[level]] = false;
      term[level * k + choice[level]] = 0;
	  ++choice[level];
	}
  }

  VarNames names(varCount);
  bigIdeal.clearAndSetNames(names);
  bigIdeal.insert(ideal);
}

void generateMatchingIdeal(BigIdeal& bigIdeal, size_t n) {
  if (n == 0)
    reportError("Too few variables in matching ideal.");
  if (n > 1000 || n > 1000)
    reportError("Number of variables in matching ideal too large.");

  class State {
  public:
	State(size_t nodeCount):
	  _notTaken(-1), _nodes(nodeCount), _isAnchor(nodeCount) {
	  std::fill(_nodes.begin(), _nodes.end(), _notTaken);
	  const size_t varCount = nodeCount * (nodeCount - 1) / 2; // n choose 2
	  _term.reset(varCount);
	}

	void takeEdge(size_t anchor, size_t other) {
	  ASSERT(anchor < _nodes.size());
	  ASSERT(other < _nodes.size());
	  ASSERT(!isTaken(anchor));
	  ASSERT(!isTaken(other));
	  _nodes[anchor] = other;
	  _nodes[other] = anchor;
      _isAnchor[anchor] = true;

	  const size_t var = edgeToVar(anchor, other);
	  ASSERT(_term[var] == 0);
	  _term[var] = 1;
	}

	void takeNode(size_t node) {
	  ASSERT(node < getNodeCount());
	  ASSERT(!isTaken(node));
	  ASSERT(!isAnchor(node));
	  _nodes[node] = node;
	}

	void dropNode(size_t node) {
	  ASSERT(node < getNodeCount());
	  ASSERT(isTaken(node));
	  ASSERT(!isAnchor(node));
	  ASSERT(_nodes[node] == node);
	  _nodes[node] = _notTaken;
	}

	void dropEdge(size_t anchor) {
	  ASSERT(anchor < _nodes.size());
	  ASSERT(isTaken(anchor));
	  ASSERT(isAnchor(anchor));
	  _isAnchor[anchor] = false;
	  const size_t other = _nodes[anchor];
	  _nodes[other] = _notTaken;
	  _nodes[anchor] = _notTaken;

	  const size_t var = edgeToVar(anchor, other);
	  ASSERT(_term[var] == 1);
	  _term[var] = 0;
	}

	size_t getNeighbor(size_t node) const {
	  ASSERT(isTaken(node));
	  return _nodes[node];
	}

    bool isAnchor(size_t node) const {
	  ASSERT(node < _nodes.size());
	  return _isAnchor[node];
	}

	bool isTaken(size_t node) const {
	  ASSERT(node < _nodes.size());
	  return _nodes[node] != _notTaken;
	}

	const Term& getTerm() const {return _term;}
	size_t getNodeCount() const {return _nodes.size();}

	// Returns static_cast<size_t>(-1) if there are no anchors to the
	// left (negative direction).
	size_t getAnchorLeft(size_t node) const {
	  ASSERT(node <= getNodeCount());
	  for (--node; node != static_cast<size_t>(-1); --node)
		if (isAnchor(node))
		  break;
	  return node;
	}

	// returns getNodeCount() if all are taken to right (positive
	// direction).
	size_t getNotTakenRight(size_t node) const {
	  ASSERT(node < getNodeCount());
	  for (++node; node < getNodeCount(); ++node)
		if (!isTaken(node))
		  break;
	  return node;
	}

  private:
	size_t edgeToVar(size_t a, size_t b) const {
	  ASSERT(a != b);
	  ASSERT(a < _nodes.size());
	  ASSERT(b < _nodes.size());
	  if (a < b)
		std::swap(a, b);
	  const size_t var = (a * (a - 1)) / 2 + b;
	  ASSERT(var < _term.getVarCount());
	  return var;
	}

	const size_t _notTaken; // cannot be static when local class
	std::vector<size_t> _nodes;
    std::vector<size_t> _isAnchor;
	Term _term;
  };

  State state(n);
  Ideal ideal(state.getTerm().getVarCount());
  size_t node = 0;

  // one node cannot be used in maximum matching if odd number of nodes.
  size_t notUsed = state.getNodeCount();
  if (state.getNodeCount() % 2 == 1) {
	notUsed = 0;
	state.takeNode(notUsed);
	++node;
  }
  while (true) {
	if (node == static_cast<size_t>(-1)) {
	  if (notUsed < state.getNodeCount()) {
		state.dropNode(notUsed);
		++notUsed;
	  }
	  if (notUsed == state.getNodeCount())
		break;
	  state.takeNode(notUsed);
	  node = 0; // start over with next node unused
	}
	ASSERT(node <= state.getNodeCount());
	if (node == state.getNodeCount()) {
	  ideal.insert(state.getTerm());
	  node = state.getAnchorLeft(node);
	} else if (!state.isTaken(node)) {
	  const size_t neighbor = state.getNotTakenRight(node);
	  if (neighbor == state.getNodeCount()) {
		node = state.getAnchorLeft(node);
	  }	else {
		state.takeEdge(node, neighbor);
		node = state.getNotTakenRight(neighbor);
	  }
	} else {
	  ASSERT(state.isTaken(node));
	  ASSERT(state.isAnchor(node));
	  const size_t neighbor = state.getNeighbor(node);
	  const size_t nextNeighbor = state.getNotTakenRight(neighbor);
	  state.dropEdge(node);
	  if (nextNeighbor == state.getNodeCount()) {
		node = state.getAnchorLeft(node);
	  } else {
		state.takeEdge(node, nextNeighbor);
		node = state.getNotTakenRight(node);
	  }
	}
  }

  VarNames names(state.getTerm().getVarCount());
  bigIdeal.clearAndSetNames(names);
  bigIdeal.insert(ideal);
}

bool generateRandomEdgeIdeal
(BigIdeal& bigIdeal, size_t variableCount, size_t generatorCount) {
  Ideal ideal(variableCount);
  Term term(variableCount);

  size_t generatorsToGo = generatorCount;
  size_t triesLeft = (size_t)4 * 1000 * 1000;
  while (generatorsToGo > 0 && triesLeft > 0) {
    --triesLeft;

    size_t a = rand() % variableCount;
    size_t b = rand() % variableCount;
    if (a == b)
      continue;

    term[a] = 1;
    term[b] = 1;

    if (ideal.isIncomparable(term)) {
      ideal.insert(term);
      --generatorsToGo;
    }

    term[a] = 0;
    term[b] = 0;

    --triesLeft;
  }

  VarNames names(variableCount);
  bigIdeal.clearAndSetNames(names);
  bigIdeal.insert(ideal);

  return generatorsToGo == 0;
}


bool generateRandomIdeal(BigIdeal& bigIdeal,
                         size_t exponentRange,
                         size_t variableCount,
                         size_t generatorCount) {
  Ideal ideal(variableCount);
  Term term(variableCount);

  size_t generatorsToGo = generatorCount;
  size_t triesLeft = (size_t)4 * 1000 * 1000;
  while (generatorsToGo > 0 && triesLeft > 0) {
    --triesLeft;

    for (size_t var = 0; var < variableCount; ++var) {
      term[var] = rand();
      if (exponentRange != numeric_limits<size_t>::max())
        term[var] %= exponentRange + 1;
    }

    if (ideal.isIncomparable(term)) {
      ideal.insert(term);
      --generatorsToGo;
    }
  }

  VarNames names(variableCount);
  bigIdeal.clearAndSetNames(names);
  bigIdeal.insert(ideal);

  return generatorsToGo == 0;
}

void generateRandomFrobeniusInstance(vector<mpz_class>& instance,
                                     size_t entryCount,
                                     const mpz_class& maxEntry) {
  ASSERT(entryCount >= 1);
  ASSERT(maxEntry >= 1);

  gmp_randclass random(gmp_randinit_default);

  /// @todo: preserve state across calls.
  random.seed((unsigned long)time(0) +
#ifdef __GNUC__ // Only GCC defines this macro.
              (unsigned long)getpid() +
#endif
              (unsigned long)clock());

  instance.resize(entryCount);

  // Populate instance with random numbers in range [1,maxEntry].
  for (size_t i = 0; i < entryCount; ++i)
    instance[i] = random.get_z_range(maxEntry) + 1;

  // Calculate greatest common divisor of instance.
  mpz_class gcd = instance[0];
  for (size_t i = 1; i < entryCount; ++i)
    mpz_gcd(gcd.get_mpz_t(), gcd.get_mpz_t(), instance[i].get_mpz_t());

  // Ensure that instance are relatively prime.
  instance.front() /= gcd;

  sort(instance.begin(), instance.end());
}