Config.cpp

#include "Config.hpp"
#include "util.hpp"

Config::Config(const Config & other)
{
  this->lines = other.lines;
  this->predicted = other.predicted;
  this->lastPoppedStack = other.lastPoppedStack;
  this->lastAttached = other.lastAttached;
  this->currentWordId = other.currentWordId;
  this->appliableSplitTransitions = other.appliableSplitTransitions;
  this->appliableTransitions = other.appliableTransitions;

  this->strategy.reset(new Strategy(*other.strategy));

  this->rawInput = other.rawInput;
  this->wordIndex = other.wordIndex;
  this->characterIndex = other.characterIndex;
  this->state = other.state;
  this->history = other.history;
  this->stack = other.stack;
  this->extraColumns = this->extraColumns;
}

std::size_t Config::getIndexOfLine(int lineIndex) const
{
  return lineIndex * getNbColumns() * (nbHypothesesMax+1);
}

std::size_t Config::getIndexOfCol(int colIndex) const
{
  return colIndex * (nbHypothesesMax+1);
}

void Config::addLines(unsigned int nbLines)
{
  lines.resize(lines.size() + nbLines*getNbColumns()*(nbHypothesesMax+1));
}

void Config::addComment()
{
  lines.resize(lines.size() + getNbColumns()*(nbHypothesesMax+1));
  get(0, getNbLines()-1, 0) = "#";
  getLastNotEmptyHyp(0, getNbLines()-1) = "#";
}

void Config::resizeLines(unsigned int nbLines)
{
  lines.resize(nbLines*getNbColumns()*(nbHypothesesMax+1));
}

bool Config::has(int colIndex, int lineIndex, int hypothesisIndex) const
{
  return colIndex >= 0 && colIndex < (int)getNbColumns() && lineIndex >= (int)getFirstLineIndex() && lineIndex < (int)getFirstLineIndex() + (int)getNbLines() && hypothesisIndex >= 0 && hypothesisIndex < nbHypothesesMax+1;
}

bool Config::has(const std::string & colName, int lineIndex, int hypothesisIndex) const
{
  return hasColIndex(colName) && has(getColIndex(colName), lineIndex, hypothesisIndex);
}

Config::String & Config::get(const std::string & colName, int lineIndex, int hypothesisIndex)
{
  return get(getColIndex(colName), lineIndex, hypothesisIndex);
}

const Config::String & Config::getConst(const std::string & colName, int lineIndex, int hypothesisIndex) const
{
  return getConst(getColIndex(colName), lineIndex, hypothesisIndex);
}

Config::String & Config::get(int colIndex, int lineIndex, int hypothesisIndex)
{
  return *getIterator(colIndex, lineIndex, hypothesisIndex);
}

const Config::String & Config::getConst(int colIndex, int lineIndex, int hypothesisIndex) const
{
  return *getConstIterator(colIndex, lineIndex, hypothesisIndex);
}

std::size_t Config::getNbLines() const
{
  return lines.size() / getIndexOfCol(getNbColumns());
}

void Config::print(FILE * dest) const
{
  std::vector<std::string> currentSequence;
  std::vector<std::string> currentSequenceComments;

  auto flushCurrentSequence = [&dest, &currentSequence, &currentSequenceComments]()
  {
    if (currentSequence.empty() && currentSequenceComments.empty())
      return;

    for (auto & comment : currentSequenceComments)
      fmt::print(dest, "{}", comment);

    for (auto & line : currentSequence)
      fmt::print(dest, "{}", line);

    fmt::print(dest, "\n");

    currentSequence.clear();
    currentSequenceComments.clear();
  };

  for (unsigned int line = 0; line < getNbLines(); line++)
  {
    if (isComment(getFirstLineIndex()+line))
    {
      currentSequenceComments.emplace_back(fmt::format("{}\n", getConst(0, getFirstLineIndex()+line, 0)));
      continue;
    }
    for (unsigned int i = 0; i < getNbColumns()-1; i++)
    {
      if (isExtraColumn(getColName(i)) and getColName(i) != EOSColName)
      {
        if (i == getNbColumns()-2)
          currentSequence.back().back() = '\n';
        continue;
      }
      auto & colContent = getAsFeature(i, getFirstLineIndex()+line);
      std::string valueToPrint = colContent;
      try
      {
        if (getColName(i) == headColName)
          if (valueToPrint != "0")
            valueToPrint = getAsFeature(idColName, std::stoi(valueToPrint));
      } catch(std::exception &) {}
      if (valueToPrint.empty())
        valueToPrint = "_";

      currentSequence.emplace_back(fmt::format("{}{}", valueToPrint, i < getNbColumns()-2 ? "\t" : "\n"));
    }
    auto & eosColContent = getAsFeature(EOSColName, getFirstLineIndex()+line);
    if (eosColContent == EOSSymbol1)
      flushCurrentSequence();
  }

  flushCurrentSequence();
}

void Config::printForDebug(FILE * dest) const
{
  static constexpr int windowSize = 10;
  static constexpr int lettersWindowSize = 40;
  static constexpr int maxWordLength = 7;

  int firstLineToPrint = wordIndex;
  int lastLineToPrint = wordIndex;
  while (wordIndex-firstLineToPrint < windowSize and has(0, firstLineToPrint-1, 0))
    --firstLineToPrint;
  while (lastLineToPrint - wordIndex < windowSize and has(0, lastLineToPrint+1, 0))
    ++lastLineToPrint;

  std::vector<std::vector<std::string>> toPrint;

  toPrint.emplace_back();
  toPrint.back().emplace_back("");
  for (unsigned int i = 0; i < getNbColumns(); i++)
  {
    if (isExtraColumn(getColName(i)) and getColName(i) != EOSColName)
      continue;
    toPrint.back().emplace_back(getColName(i));
  }

  for (int line = firstLineToPrint; line <= lastLineToPrint; line++)
  {
    if (isComment(line))
      continue;
    toPrint.emplace_back();
    toPrint.back().emplace_back(line == (int)wordIndex ? "=>" : "");
    for (unsigned int i = 0; i < getNbColumns(); i++)
    {
      if (isExtraColumn(getColName(i)) and getColName(i) != EOSColName)
        continue;
      std::string colContent = has(i,line,0) ? getAsFeature(i, line).get() : "?";
      std::string toPrintCol = colContent;
      try
      {
        if (getColName(i) == headColName && toPrintCol != "_" && !toPrintCol.empty())
          if (toPrintCol != "0" && toPrintCol != "?")
            toPrintCol = has(0,std::stoi(toPrintCol),0) ? getAsFeature(idColName, std::stoi(toPrintCol)).get() : "?";
      } catch(std::exception & e) {util::myThrow(fmt::format("toPrintCol='{}' {}", toPrintCol, e.what()));}
      toPrint.back().emplace_back(util::shrink(toPrintCol, maxWordLength));
    }
  }

  std::vector<std::size_t> colLength(toPrint[0].size(), 0);
  for (auto & line : toPrint)
    for (unsigned int col = 0; col < line.size()-1; col++)
      colLength[col] = std::max((int)colLength[col], util::printedLength(line[col]));

  int lengthSum = 2*getNbColumns();
  for (auto & val : colLength)
    lengthSum += val;
  std::string longLine = fmt::format("{:-<{}}", "", lengthSum);

  std::string historyStr = "";
  for (auto & h : history)
  {
    historyStr += h;
    historyStr += ",";
  }
  if (!historyStr.empty())
    historyStr.pop_back();
  std::string stackStr = "";
  for (auto & s : stack)
  {
    if (hasColIndex(idColName))
    {
      if (has(idColName, s, 0))
        stackStr += getAsFeature(idColName, s);
      else
        stackStr += "?";
    }
    else
      stackStr += std::to_string(s);
    stackStr += ",";
  }
  if (!stackStr.empty())
    stackStr.pop_back();
  fmt::print(dest, "{}\n", longLine);
  for (std::size_t index = characterIndex; index < util::getSize(rawInput) and index - characterIndex < lettersWindowSize; index++)
    fmt::print(dest, "{}", getLetter(index));
  if (!util::isEmpty(rawInput))
    fmt::print(dest, "\n{}\n", longLine);
  fmt::print(dest, "State={}\nwordIndex={} characterIndex={}\nhistory=({})\nstack=({})\n", state, wordIndex, characterIndex, historyStr, stackStr);
  fmt::print(dest, "{}\n", longLine);

  for (unsigned int line = 0; line < toPrint.size(); line++)
  {
    if (line == 1)
      fmt::print(dest, "{}\n", longLine);
    for (unsigned int col = 0; col < toPrint[line].size()-1; col++)
      fmt::print(dest, "{}{:>{}}{}", toPrint[line][col], "", colLength[col]-util::printedLength(toPrint[line][col]), col == toPrint[line].size()-2 ? "\n" : "  ");
    if (toPrint[line].back() == EOSSymbol1)
      fmt::print(dest, "\n");
  }

  fmt::print(dest, "{}\n", longLine);
}

Config::String & Config::getLastNotEmpty(int colIndex, int lineIndex)
{
  if (!has(colIndex, lineIndex, 0))
    util::myThrow(fmt::format("asked for line {} but last line = {}", lineIndex, getNbLines()+getFirstLineIndex()-1));
  int baseIndex = getIndexOfLine(lineIndex-getFirstLineIndex()) + getIndexOfCol(colIndex);

  for (int i = nbHypothesesMax; i > 0; --i)
    if (!util::isEmpty(lines[baseIndex+i]))
      return lines[baseIndex+i];

  return lines[baseIndex];
}

Config::String & Config::getLastNotEmptyHyp(int colIndex, int lineIndex)
{
  if (!has(colIndex, lineIndex, 0))
    util::myThrow(fmt::format("asked for line {} but nbLines = {}", lineIndex, getNbLines()));
  int baseIndex = getIndexOfLine(lineIndex-getFirstLineIndex()) + getIndexOfCol(colIndex);

  for (int i = nbHypothesesMax; i > 0; --i)
    if (!util::isEmpty(lines[baseIndex+i]))
      return lines[baseIndex+i];

  return lines[baseIndex+1];
}

Config::String & Config::getFirstEmpty(int colIndex, int lineIndex)
{
  if (!has(colIndex, lineIndex, 0))
    util::myThrow(fmt::format("asked for line {} but nbLines = {}", lineIndex, getNbLines()));
  int baseIndex = getIndexOfLine(lineIndex-getFirstLineIndex()) + getIndexOfCol(colIndex);

  for (int i = 1; i < nbHypothesesMax; ++i)
    if (util::isEmpty(lines[baseIndex+i]))
      return lines[baseIndex+i];

  return lines[baseIndex+nbHypothesesMax]; 
}

Config::String & Config::getFirstEmpty(const std::string & colName, int lineIndex)
{
  return getFirstEmpty(getColIndex(colName), lineIndex);
}

const Config::String & Config::getLastNotEmptyConst(int colIndex, int lineIndex) const
{
  if (!has(colIndex, lineIndex, 0))
    util::myThrow(fmt::format("asked for line {} but nbLines = {}", lineIndex, getNbLines()));
  int baseIndex = getIndexOfLine(lineIndex-getFirstLineIndex()) + getIndexOfCol(colIndex);

  for (int i = nbHypothesesMax; i > 0; --i)
    if (!util::isEmpty(lines[baseIndex+i]))
      return lines[baseIndex+i];

  return lines[baseIndex];
}

const Config::String & Config::getLastNotEmptyHypConst(int colIndex, int lineIndex) const
{
  if (!has(colIndex, lineIndex, 0))
    util::myThrow(fmt::format("asked for line {} but nbLines = {}", lineIndex, getNbLines()));
  int baseIndex = getIndexOfLine(lineIndex-getFirstLineIndex()) + getIndexOfCol(colIndex);

  for (int i = nbHypothesesMax; i > 0; --i)
    if (!util::isEmpty(lines[baseIndex+i]))
      return lines[baseIndex+i];

  return lines[baseIndex+1];
}

const Config::String & Config::getAsFeature(int colIndex, int lineIndex) const
{
  if (isPredicted(getColName(colIndex)))
    return getLastNotEmptyHypConst(colIndex, lineIndex);

  return getLastNotEmptyConst(colIndex, lineIndex);
}

const Config::String & Config::getAsFeature(const std::string & colName, int lineIndex) const
{
  return getAsFeature(getColIndex(colName), lineIndex);
}

Config::String & Config::getLastNotEmpty(const std::string & colName, int lineIndex)
{
  return getLastNotEmpty(getColIndex(colName), lineIndex);
}

Config::String & Config::getLastNotEmptyHyp(const std::string & colName, int lineIndex)
{
  return getLastNotEmptyHyp(getColIndex(colName), lineIndex);
}

const Config::String & Config::getLastNotEmptyConst(const std::string & colName, int lineIndex) const
{
  return getLastNotEmptyConst(getColIndex(colName), lineIndex);
}

const Config::String & Config::getLastNotEmptyHypConst(const std::string & colName, int lineIndex) const
{
  return getLastNotEmptyHypConst(getColIndex(colName), lineIndex);
}

Config::ValueIterator Config::getIterator(int colIndex, int lineIndex, int hypothesisIndex)
{
  return lines.begin() + getIndexOfLine(lineIndex-getFirstLineIndex()) + getIndexOfCol(colIndex) + hypothesisIndex;
}

Config::ConstValueIterator Config::getConstIterator(int colIndex, int lineIndex, int hypothesisIndex) const
{
  return lines.begin() + getIndexOfLine(lineIndex-getFirstLineIndex()) + getIndexOfCol(colIndex) + hypothesisIndex;
}

void Config::addToHistory(const std::string & transition)
{
  history.push_back(String(transition));
}

void Config::addToStack(std::size_t index)
{
  stack.push_back(index);
}

void Config::popStack()
{
  lastPoppedStack = getStack(0);
  stack.pop_back();
}

void Config::swapStack(int relIndex1, int relIndex2)
{
  int tmp = getStack(relIndex1);
  getStackRef(relIndex1) = getStack(relIndex2);
  getStackRef(relIndex2) = tmp;
}

bool Config::hasCharacter(int letterIndex) const
{
  return letterIndex >= 0 and letterIndex < (int)util::getSize(rawInput);
}

util::utf8char Config::getLetter(int letterIndex) const
{
  return rawInput[letterIndex];
}

bool Config::isComment(std::size_t lineIndex) const
{
  auto iter = getConstIterator(0, lineIndex, 0);
  return !iter->get().empty() and iter->get()[0] == '#';
}

bool Config::isCommentPredicted(std::size_t lineIndex) const
{
  auto & col0Pred = getAsFeature(0, lineIndex);
  auto & col0Gold = getConst(0, lineIndex, 0);
  return (!util::isEmpty(col0Pred) and col0Pred.get()[0] == '#') or (!util::isEmpty(col0Gold) and col0Gold.get()[0] == '#');
}

bool Config::isMultiword(std::size_t lineIndex) const
{
  return hasColIndex(idColName) && getConst(idColName, lineIndex, 0).get().find('-') != std::string::npos;
}

bool Config::isMultiwordPredicted(std::size_t lineIndex) const
{
  return hasColIndex(idColName) && getAsFeature(idColName, lineIndex).get().find('-') != std::string::npos;
}

int Config::getMultiwordSize(std::size_t lineIndex) const
{
  auto splited = util::split(getConst(idColName, lineIndex, 0).get(), '-');
  return std::stoi(std::string(splited[1])) - std::stoi(std::string(splited[0]));
}

int Config::getMultiwordSizePredicted(std::size_t lineIndex) const
{
  auto splited = util::split(getAsFeature(idColName, lineIndex).get(), '-');
  return std::stoi(std::string(splited[1])) - std::stoi(std::string(splited[0]));
}

bool Config::isEmptyNode(std::size_t lineIndex) const
{
  return hasColIndex(idColName) && getConst(idColName, lineIndex, 0).get().find('.') != std::string::npos;
}

bool Config::isEmptyNodePredicted(std::size_t lineIndex) const
{
  return hasColIndex(idColName) && getAsFeature(idColName, lineIndex).get().find('.') != std::string::npos;
}

bool Config::isToken(std::size_t lineIndex) const
{
  return !isComment(lineIndex) && !isMultiword(lineIndex) && !isEmptyNode(lineIndex);
}

bool Config::isTokenPredicted(std::size_t lineIndex) const
{
  return !isCommentPredicted(lineIndex) && !isMultiwordPredicted(lineIndex) && !isEmptyNodePredicted(lineIndex);
}

bool Config::moveWordIndex(int relativeMovement)
{
  int nbMovements = 0;
  int oldVal = wordIndex;
  while (nbMovements != relativeMovement)
  {
    do
    {
      relativeMovement > 0 ? wordIndex++ : wordIndex--;
      if (!has(0,wordIndex,0))
      {
        wordIndex = oldVal;
        return false;
      }
    }
    while (isComment(wordIndex));
    nbMovements += relativeMovement > 0 ? 1 : -1;
  }

  return true;
}

void Config::moveWordIndexRelaxed(int relativeMovement)
{
  int nbMovements = 0;
  int increment = relativeMovement > 0 ? 1 : -1;
  while (nbMovements != relativeMovement)
  {
    do
    {
      if (!has(0,wordIndex+increment,0))
        break;
      wordIndex += increment;
    }
    while (isComment(wordIndex));
    nbMovements += relativeMovement > 0 ? 1 : -1;
  }

  if (!isComment(wordIndex))
    return;

  moveWordIndex(-increment);
}

bool Config::canMoveWordIndex(int relativeMovement) const
{
  int nbMovements = 0;
  int oldVal = wordIndex;
  while (nbMovements != relativeMovement)
  {
    do
    {
      relativeMovement > 0 ? oldVal++ : oldVal--;
      if (!has(0,oldVal,0))
        return false;
    }
    while (isComment(oldVal));
    nbMovements += relativeMovement > 0 ? 1 : -1;
  }

  return true;
}

bool Config::moveCharacterIndex(int relativeMovement)
{
  int oldVal = characterIndex;
  characterIndex = std::max(0, (int)std::min(characterIndex+relativeMovement, util::getSize(rawInput)));
  return (int)characterIndex == oldVal + relativeMovement;
}

bool Config::canMoveCharacterIndex(int relativeMovement) const
{
  int target = std::max(0, (int)std::min(characterIndex+relativeMovement, util::getSize(rawInput)));
  return target == (int)characterIndex + relativeMovement;
}

bool Config::rawInputOnlySeparatorsLeft() const
{
  for (unsigned int i = characterIndex; i < util::getSize(rawInput); i++)
    if (!util::isSeparator(rawInput[i]))
      return false;

  return true;
}

std::size_t Config::getWordIndex() const
{
  return wordIndex;
}

std::size_t Config::getCharacterIndex() const
{
  return characterIndex;
}

const Config::String & Config::getHistory(int relativeIndex) const
{
  return history[history.size()-1-relativeIndex];
}

std::size_t Config::getStack(int relativeIndex) const
{
  if (relativeIndex == -1)
    return getLastPoppedStack();
  return stack[stack.size()-1-relativeIndex];
}

std::size_t & Config::getStackRef(int relativeIndex)
{
  return stack[stack.size()-1-relativeIndex];
}

bool Config::hasHistory(int relativeIndex) const
{
  return relativeIndex >= 0 && relativeIndex < (int)history.size();
}

bool Config::hasStack(int relativeIndex) const
{
  if (relativeIndex == -1)
    return has(0,getLastPoppedStack(),0);
  return relativeIndex >= 0 && relativeIndex < (int)stack.size();
}

Config::String Config::getState() const
{
  return state;
}

void Config::setState(const std::string state)
{
  this->state = state;
}

bool Config::stateIsDone() const
{
  if (!util::isEmpty(rawInput))
    return rawInputOnlySeparatorsLeft() and !has(0, wordIndex+1, 0) and !hasStack(0);

  return !has(0, wordIndex+1, 0) and !hasStack(0);
}

void Config::addPredicted(const std::set<std::string> & predicted)
{
  for (auto & col : predicted)
  {
    if (!hasColIndex(col))
      util::myThrow(fmt::format("unknown column '{}'", col));
    this->predicted.insert(col);
  }

  for (auto & col : extraColumns)
    if (col != EOSColName)
      this->predicted.insert(col);
}

bool Config::isPredicted(const std::string & colName) const
{
  return predicted.count(colName);
}

int Config::getLastPoppedStack() const
{
  return lastPoppedStack;
}

int Config::getCurrentWordId() const
{
  return currentWordId;
}

void Config::setCurrentWordId(int currentWordId)
{
  this->currentWordId = currentWordId;
}

void Config::addMissingColumns()
{
  int firstIndex = 0;
  for (unsigned int index = 0; index < getNbLines(); index++)
  {
    if (!isTokenPredicted(index))
      continue;

    if (util::isEmpty(getAsFeature(idColName, index)))
    {
      int last = 0;
      if (index > 0 and isTokenPredicted(index-1))
        last = std::stoi(getAsFeature(idColName, index-1));
      getLastNotEmptyHyp(idColName, index) = std::to_string(last+1);
    }

    int curId = std::stoi(getAsFeature(idColName, index));
    if (curId == 1)
      firstIndex = index;

    if (hasColIndex(headColName))
      if (util::isEmpty(getAsFeature(headColName, index)))
        getLastNotEmptyHyp(headColName, index) = (curId == 1) ? "0" : std::to_string(firstIndex);
  }
}

long Config::getRelativeWordIndex(int relativeIndex) const
{
  if (relativeIndex < 0)
  {
    for (int index = getWordIndex()-1, counter = 0; has(0,index,0); --index)
        if (!isCommentPredicted(index))
        {
          --counter;
          if (counter == relativeIndex)
            return index;
        }
  }
  else
  {
    for (int index = getWordIndex(), counter = 0; has(0,index,0); ++index)
        if (!isCommentPredicted(index))
        {
          if (counter == relativeIndex)
            return index;
          ++counter;
        }
  }

  return -1;
}

long Config::getRelativeWordIndex(Object object, int relativeIndex) const
{
  if (object == Object::Buffer)
    return getRelativeWordIndex(relativeIndex);

  return getStack(relativeIndex);
}

bool Config::hasRelativeWordIndex(Object object, int relativeIndex) const
{
  if (object == Object::Buffer)
    return has(0,getRelativeWordIndex(relativeIndex),0);

  return hasStack(relativeIndex);
}

void Config::setAppliableSplitTransitions(const std::vector<Transition *> & appliableSplitTransitions)
{
  this->appliableSplitTransitions = appliableSplitTransitions;
}

void Config::setAppliableTransitions(const std::vector<int> & appliableTransitions)
{
  this->appliableTransitions = appliableTransitions;
}

const std::vector<Transition *> & Config::getAppliableSplitTransitions() const
{
  return appliableSplitTransitions;
}

const std::vector<int> & Config::getAppliableTransitions() const
{
  return appliableTransitions;
}

Config::Object Config::str2object(const std::string & s)
{
  if (s == "b")
    return Object::Buffer;
  if (s == "s")
    return Object::Stack;

  util::myThrow(fmt::format("Invalid object '{}'", s));
  return Object::Buffer;
}

bool Config::isExtraColumn(const std::string & colName) const
{
  for (auto & extraCol : extraColumns)
    if (extraCol == colName)
      return true;
  return false;
}

int Config::getLastAttached() const
{
  return lastAttached;
}

void Config::setLastAttached(int lastAttached)
{
  this->lastAttached = lastAttached;
}

std::size_t Config::getStackSize() const
{
  return stack.size();
}

void Config::setStrategy(const std::vector<std::string> & strategyDefinition)
{
  strategy.reset(new Strategy(strategyDefinition));
}

Strategy & Config::getStrategy()
{
  if (strategy.get() == nullptr)
    util::myThrow("strategy was not set");

  return *strategy.get();
}