ReadingMachine.cpp

#include "ReadingMachine.hpp"
#include "util.hpp"

ReadingMachine::ReadingMachine(std::filesystem::path path, bool train) : path(path), train(train)
{
  readFromFile(path);
}

void ReadingMachine::readFromFile(std::filesystem::path path)
{
  std::FILE * file = std::fopen(path.c_str(), "r");
  if (not file)
    util::myThrow(fmt::format("can't open file '{}'", path.string()));

  char buffer[1024];
  std::string fileContent;
  std::vector<std::string> lines;
  while (!std::feof(file))
  {
    if (buffer != std::fgets(buffer, 1024, file))
      break;
    // If line is blank or commented (# or //), ignore it
    if (util::doIfNameMatch(std::regex("((\\s|\\t)*)(((#|//).*)|)(\n|)"), buffer, [](auto){}))
      continue;

    if (buffer[std::strlen(buffer)-1] == '\n')
      buffer[std::strlen(buffer)-1] = '\0';

    lines.emplace_back(buffer);
  }
  std::fclose(file);

  try
  {
    unsigned int curLine = 0;
    if (!util::doIfNameMatch(std::regex("Name : (.+)"), lines[curLine++], [this](auto sm){name = sm[1];}))
      util::myThrow("No name specified");

    while (util::doIfNameMatch(std::regex("Classifier : (.+)"), lines[curLine], [this,path,&lines,&curLine](auto sm)
      {
        curLine++;
        classifierDefinitions.emplace_back();
        classifierNames.emplace_back(sm.str(1));
        if (lines[curLine] != "{")
          util::myThrow(fmt::format("Expected '{}', got '{}' instead", "{", lines[curLine]));

        for (curLine++; curLine < lines.size(); curLine++)
        {
          if (lines[curLine] == "}")
          {
            curLine++;
            break;
          }
          classifierDefinitions.back().emplace_back(lines[curLine]);
        }
        classifiers.emplace_back(new Classifier(sm.str(1), path.parent_path(), classifierDefinitions.back(), train));
        for (auto state : classifiers.back()->getStates())
          state2classifier[state] = classifiers.size()-1;
      }));
    if (classifiers.empty())
      util::myThrow("No Classifier specified");

    util::doIfNameMatch(std::regex("Splitwords : (.+)"), lines[curLine], [this,path,&curLine](auto sm)
    {
      this->splitWordTransitionSet.reset(new TransitionSet(path.parent_path() / sm.str(1)));
      curLine++;
    });

    if (!util::doIfNameMatch(std::regex("Predictions : (.+)"), lines[curLine++], [this](auto sm)
    {