path: root/grammer.cpp

#include "grammer.h"

#include "debug.h"

#include <algorithm>
#include <limits>

using namespace Gram;

void Compiler::clear()
{
 symbol_variants.clear();
 nodes.clear();
}

std::string Compiler::GetTypeOfNode(index_t node_id) const
{
 if (node_id >= nodes.size())
  throw std::runtime_error("GetTypeOfNode(): node_id="s + std::to_string(node_id) + ", nodes.size()="s + std::to_string(nodes.size()));
 return nodes[node_id].type;
}

bool Gram::ChildIdIsEmpty(int32_t child_id)
{
 return child_id == 0;
}

bool Gram::ChildIdIsToken(int32_t child_id)
{
 return child_id < 0;
}

bool Gram::ChildIdIsNode(int32_t child_id)
{
 return child_id > 0;
}

index_t Gram::TokenIdFromChildId(int32_t child_id)
{
 return index_t(-child_id) - 1;
}

int32_t Gram::ChildIdFromTokenId(index_t token_id)
{
 return -1 - int32_t(token_id);
}

void Compiler::DumpTree()
{
 Debug("= Dump =======================================");
 Debug("nodes.size()="s + std::to_string(nodes.size()));
 if (nodes.size() > 0) {
  if (0) {
   Debug("--- Nodes ------------------------------------");
   for (const auto& node : nodes) {
    std::string line{"("s + std::to_string(node.node_id) + "):"s};
    for (const auto& child : node.child_ids) {
     line += " "s;
     if (ChildIdIsToken(child))
      line += "t"s + std::to_string(TokenIdFromChildId(child));
     else
      line += std::to_string(child);
    }
    Debug(line);
   }
  }

  Debug("--- Tree -------------------------------------");
  std::deque<std::pair<int32_t, size_t>> todo{std::pair<int32_t, size_t>{static_cast<int32_t>(0), 0}}; // id, indent
  while (!todo.empty()) {
   auto [current_index, indent] {todo.front()};
   todo.pop_front();

   std::string line(indent, ' ');
   if (ChildIdIsToken(current_index)) {
    index_t token_id {TokenIdFromChildId(current_index)};
    line += "Token("s + std::to_string(token_id) + "): "s + tokens[token_id].type + "("s + tokens[token_id].value + ")"s;
   } else {
    auto& node {nodes[current_index]};
    line += "Node("s + std::to_string(current_index) + "): "s + node.type + "/" + std::to_string(node.variant);
   
    auto child_ids{node.child_ids};
    for (int i = 0; i < child_ids.size(); i++) {
     todo.insert(todo.begin() + i, std::pair<int32_t, size_t>{child_ids[i], indent + 1});
    }
   }
   Debug(line);
  }
 }
 Debug("==============================================");
}

index_t Compiler::AddNode(const std::string& type, index_t variant, NodePosition pos)
{
 auto& list { bnf[type][variant]};
 index_t node_id{nodes.size()};
 if (nodes.size() > 0)
  nodes[pos.node_id].child_ids[pos.child_pos] = node_id;
 nodes.emplace_back(TreeNode{pos, node_id, type, variant, std::vector<int32_t>(size_t(list.size()), 0)});

 return node_id;
}

Compiler::AddNodeGuard::AddNodeGuard(Compiler& compiler, index_t variant): m_compiler(compiler)
{
 m_compiler.symbol_variants.push_back(variant);
}

Compiler::AddNodeGuard::~AddNodeGuard()
{
 m_compiler.symbol_variants.pop_back();
}

void Compiler::IncNodePosition(NodePosition& pos)
{
 if (nodes.size() == 0) // special case: empty tree 
  return;
 if (pos.node_id >= nodes.size())
  throw std::runtime_error("ICE: NodePosition with node_id "s + std::to_string(pos.node_id) + " doesn't exist."s);
 if (pos.child_pos >= nodes[pos.node_id].child_ids.size())
  throw std::runtime_error("ICE: NodePosition with child_pos "s + std::to_string(pos.child_pos) + " in node_id "s + std::to_string(pos.node_id) + " doesn't exist."s);

 int32_t child_id{nodes[pos.node_id].child_ids[pos.child_pos]};

 if (ChildIdIsEmpty(child_id))
  throw std::runtime_error("ICE: NodePosition is empty");

 // Actually, advance
 if (ChildIdIsToken(child_id)) {
  pos.child_pos++;
 } else {
  pos.node_id = child_id;
  pos.child_pos = 0;
 }
 
 // Go to parent if child ids completely traversed
 while (pos.node_id > 0 && pos.child_pos >= nodes[pos.node_id].child_ids.size()) {
  pos = nodes[pos.node_id].pos;
  pos.child_pos++;
 }

 // Advancing at root node for last child is allowed: Finished
 if (pos.child_pos >= nodes[pos.node_id].child_ids.size())
  return;

 if (ChildIdIsNode(nodes[pos.node_id].child_ids[pos.child_pos]))
  throw std::runtime_error("ICE: No node expected at "s + std::to_string(pos.node_id) + "/"s + std::to_string(pos.child_pos));
}

size_t Compiler::minimumSymbolsNeeded(std::string symbol)
{
 if (isTerminal(bnf, symbol)) {
  return 1;
 } else {
  auto it_min{m_min.find(symbol)};
  if (it_min != m_min.end())
   return it_min->second;
  m_min[symbol] = std::numeric_limits<size_t>::max();

  auto it{bnf.find(symbol)};
  if (it != bnf.end()) {
   size_t minimum{std::numeric_limits<size_t>::max()};
   
   for (const auto& list: it->second) {
    minimum = std::min(minimum, minimumSymbolsNeeded(list));
   }

   m_min[symbol] = minimum;

   return minimum;
  } else
   throw std::runtime_error("ICE: Symbol "s + symbol + " expected in BNF"s);
 }
}

size_t Compiler::minimumSymbolsNeeded(std::vector<std::string> symbol_list)
{
 size_t result{0};

 for (const auto& symbol: symbol_list) {
  size_t min { minimumSymbolsNeeded(symbol) };
  if (min == std::numeric_limits<size_t>::max())
   return min;
  result += min;
 }

 return result;
}

/// begin, end: indexes in tokens list
bool Compiler::match(std::vector<std::string> symbol_list, size_t begin, size_t end)
{
 // match terminal symbols at start
 while (begin < end && symbol_list.size() > 0 && symbol_list.front() == tokens[begin].type) {
  begin++;
  symbol_list.erase(symbol_list.begin());
 }

 // match terminal symbols at end
 while (begin < end && symbol_list.size() > 0 && symbol_list.back() == tokens[end - 1].type) {
  end--;
  symbol_list.erase(symbol_list.end() - 1);
 }

 // matching of empty lists
 if (symbol_list.size() == 0) {
  if (begin == end) { // only match real empty list
   // this is the point of the final match
   constructTree();
   return true;
  }
  return false;
 }

 // now, symbol_list has size > 0 and contains non-terminal symbols at start and end
 
 // resolve first symbol
 auto it{bnf.find(symbol_list.front())};
 if (it != bnf.end()) {
  for (size_t i = 0; i < it->second.size(); i++) { // iterate over alternatives
   AddNodeGuard guard(*this, i);
   std::vector<std::string> list {it->second[i]};
   list.insert(list.end(), symbol_list.begin() + 1, symbol_list.end());
   if (minimumSymbolsNeeded(list) > end - begin) // stop recursion
    continue;

   // TODO: recurse last?
   if (match(list, begin, end))
    return true;
  }
 } else
  return false; // terminal symbol not found in bnf, non-matching
 
 return false; // no match found
}

bool Compiler::match(std::string symbol, size_t begin, size_t end)
{
 std::vector<std::string> symbol_list{symbol};
 return match(symbol_list, begin, end);
}

void Compiler::constructTree()
{
 symbol_variants_it = symbol_variants.begin();
 
 m_symbol_list = {m_top};
 m_symbol_list_pos = 0;

 NodePosition tree_pos;

 while (m_symbol_list_pos < m_symbol_list.size()) {
  std::string symbol{m_symbol_list[m_symbol_list_pos]};

  if (isTerminal(bnf, symbol)) {
   // Advance terminal symbol
   nodes[tree_pos.node_id].child_ids[tree_pos.child_pos] = ChildIdFromTokenId(m_symbol_list_pos);
   IncNodePosition(tree_pos);
   m_symbol_list_pos++;
  } else {
   // Replace non-terminal symbol
   m_symbol_list.erase(m_symbol_list.begin() + m_symbol_list_pos);
   std::vector<std::string> list {bnf[symbol][*symbol_variants_it]};
   m_symbol_list.insert(m_symbol_list.begin() + m_symbol_list_pos, list.begin(), list.end());

   index_t node_id {AddNode(symbol, *symbol_variants_it, tree_pos)};
   
   if (node_id > 0) {
    nodes[tree_pos.node_id].child_ids[tree_pos.child_pos] = node_id;
    IncNodePosition(tree_pos);
   }
   
   symbol_variants_it++;
  }
 }

}

Compiler::Compiler(BNF& bnf, const std::string& top): bnf(bnf), m_top(top), ReverseBNF{Reverse(bnf)}, reversedFirst{reverseFirst(bnf)}
{
 //std::cout << "DEBUG: " << m_top << std::endl;
 
 //
 // prepare helper cache (TODO: remove this ugly workaround for remaining bad marker elements)
 //
 minimumSymbolsNeeded("translation-unit");
 // remove bad marker elements
 auto it{m_min.begin()};
 while (it != m_min.end()) {
  if (it->second == std::numeric_limits<size_t>::max()) {
   it = m_min.erase(it);
  } else {
   ++it;
  }
 }
 minimumSymbolsNeeded("translation-unit");
}

std::vector<TreeNode> Compiler::compile(std::vector<Token> p_tokens)
{
 clear();
 tokens = p_tokens;

 if (tokens.size() == 0)
  throw std::runtime_error("No tokens");

 //
 // top-down algorithm:
 //
 // 1. Match linear tokens list to bnf, building up list of used variants (symbol_variants)
 // 2. Construct Node Tree from symbol_variants
 //
 if (!match(m_top, 0, tokens.size()))
  throw std::runtime_error("Compile error");

 DumpTree();

 return nodes;
}