valeri/src/reader.cpp

#include "reader.hpp"

#include "common.hpp"
#include "error.hpp"

static bool is_digit(char32_t c) { return c >= '0' && c <= '9'; }
static bool is_alpha(char32_t c) {
  return (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z');
}

static bool is_hex_digit(char32_t c) {
  return is_digit(c) || (c >= 'a' && c <= 'f') || (c >= 'A' && c <= 'F');
}

static bool is_symbol_char(char32_t c) {
  static char valid_char[] = {'!', '$', '%', '&', '*', '+', '-', '.', '/',
                              ':', '<', '=', '>', '?', '@', '^', '_', '~'};

  if (is_digit(c)) return true;
  if (is_alpha(c)) return true;

  for (size_t i = 0; i < sizeof(valid_char); ++i) {
    if (char32_t(valid_char[i]) == c) return true;
  }
  return false;
}

static bool is_whitespace(char32_t c) {
  return c == ' ' || c == '\t' || c == '\v' || c == '\f';
}

static bool is_brace(char32_t c) {
  return c == '[' || c == ']' || c == '(' || c == ')' || c == '{' || c == '}';
}

static bool is_newline(char32_t c) { return c == '\r' || c == '\n'; }

static bool is_separator(char32_t c) {
  return is_whitespace(c) || is_brace(c) || is_newline(c) || c == ';';
}

#define EOS 0

Result<Value> Reader::read_one() {
  forward_whitespace();

  auto saved_position = position_;
  if (is_numeric_start()) {
    auto res = read_number();
    if (res.has_value()) return res;
  } else if (match("true") || match("false")) {
    auto res = read_bool();
    if (res.has_value()) return res;
  } else if (is_string_start()) {
    return read_string();
  } else if (match('(')) {
    return read_list();
  } else if (match('[')) {
    // TODO: implement array
    // return read_array();
    return ErrorCode::ReadError;
  } else if (match('{')) {
    return read_dict();
    return ErrorCode::ReadError;
  }

  position_ = saved_position;
  if (is_symbol_start()) {
    return read_symbol();
  }

  return ErrorCode::ReadError;
}

Result<Value> Reader::read_multiple() {
  Value res = TRY(Nil::create(_arena));

  while (1) {
    forward_whitespace();
    if (is_eof()) {
      return reverse(_arena, res);
    }

    auto val = TRY(read_one());
    res = Value(TRY(Pair::create(_arena, val, res)));
  }

  return ErrorCode::ReadError;
}

Result<Value> Reader::read_list() {
  if (!match('(')) return ErrorCode::ReadError;

  forward();

  Value res = TRY(Nil::create(_arena));

  while (1) {
    forward_whitespace();

    if (is_eof()) {
      return ErrorCode::ReadError;
    }

    if (match(')')) {
      forward();
      return reverse(_arena, res);
    }

    auto val = TRY(read_one());

    res = Value(TRY(Pair::create(_arena, val, res)));
  }

  return ErrorCode::ReadError;
}

Result<Value> Reader::read_dict() {
  if (!match('{')) return ErrorCode::ReadError;

  forward();

  auto res = TRY(Dict::create(_arena));

  while (1) {
    forward_whitespace();

    if (is_eof()) {
      return ErrorCode::ReadError;
    }

    if (match('}')) {
      forward();
      return Value(std::move(res));
    }

    auto val1 = TRY(read_one());
    auto val2 = TRY(read_one());

    res = TRY(res.insert(_arena, val1, val2));
  }

  return ErrorCode::ReadError;
}
Result<Value> Reader::read_bool() {
  if (match("true")) {
    forward(4);
    if (!is_separator(get()) && !is_eof()) return ErrorCode::ReadError;
    return Value(TRY(Bool::create(_arena, true)));
  }
  if (match("false")) {
    forward(5);
    if (!is_separator(get()) && !is_eof()) return ErrorCode::ReadError;
    return Value(TRY(Bool::create(_arena, false)));
  }

  return ErrorCode::ReadError;
}

Result<Value> Reader::read_string() {
  if (!match('"')) return ErrorCode::ReadError;
  size_t start = position_.offset + 1;

  forward();

  while (!match('"') && !match('\r') && !match('\n') && !is_eof()) {
    if (match('\\')) {
      forward();
    }

    forward();
  }

  if (!match('"')) return ErrorCode::UnterminatedStringLiteral;

  forward();

  String result = TRY(String::create(_arena, ""));

  for (size_t i = 0; i < position_.offset - start - 1; i++) {
    char32_t cur = TRY(_str[start + i]);
    if (cur != '\\') {
      // TODO: optimize this
      result = TRY(result.concat(_arena, &cur, 1));
      continue;
    }
    ++i;
    if (i >= position_.offset - start) continue;
    char32_t next = TRY(_str[start + i]);

    if (next == '\0') {
      result = TRY(result.concat(_arena, "\0"));
      continue;
    }
    switch (next) {
      case 'b':
        result = TRY(result.concat(_arena, "\b"));
        continue;
      case 'f':
        result = TRY(result.concat(_arena, "\f"));
        continue;
      case 'n':
        result = TRY(result.concat(_arena, "\n"));
        continue;
      case 'r':
        result = TRY(result.concat(_arena, "\r"));
        continue;
      case 't':
        result = TRY(result.concat(_arena, "\t"));
        continue;
      case 'v':
        result = TRY(result.concat(_arena, "\v"));
        continue;
    }

    result = TRY(result.concat(_arena, "\\"));
    result = TRY(result.concat(_arena, &next, 1));
  }

  if (!is_separator(get()) && !is_eof()) return ErrorCode::ReadError;

  return Value(std::move(result));
}

Result<Value> Reader::read_number() {
  if (!is_numeric_start()) return ErrorCode::ReadError;
  size_t start = position_.offset;

  bool is_float = false;

  if (match('+') || match('-')) forward();

  if (match('0')) {
    forward();
    if (match('.')) {
      is_float = true;
      forward();
      while (is_digit(get())) forward();

      if (match('e') || match('E')) {
        if (!forward_exponent()) return ErrorCode::InvalidNumericLiteral;
      }
    } else if (match('e') || match('E')) {
      is_float = true;
      if (!forward_exponent()) return ErrorCode::InvalidNumericLiteral;
    } else if (match('x') || match('X')) {
      if (!forward_hex_number()) return ErrorCode::InvalidNumericLiteral;
    } else if (is_digit(get())) {
      do {
        forward();
      } while (is_digit(get()));
    }
  } else {
    while (is_digit(get())) forward();
    if (match('.')) {
      is_float = true;
      forward();
      while (is_digit(get())) forward();
    }
    if (match('e') || match('E')) {
      is_float = true;
      if (!forward_exponent()) return ErrorCode::InvalidNumericLiteral;
    }
  }

  Value res;

  if (position_.offset - start >= 32) return ErrorCode::InvalidNumericLiteral;

  char buf[32];
  for (size_t i = 0; i < position_.offset - start; ++i) {
    buf[i] = char(TRY(_str[start + i]));
  }
  buf[position_.offset - start] = '\0';

  if (is_float) {
    res = Value(TRY(Float::create(_arena, strtod(buf, 0))));
    if (match("f64")) {
      forward(3);
    }
  } else {
    res = Value(TRY(Int64::create(_arena, strtoll(buf, 0, 10))));
  }

  if (!is_separator(get()) && !is_eof()) return ErrorCode::ReadError;

  return res;
}

Result<Value> Reader::read_symbol() {
  if (!is_symbol_char(get())) return ErrorCode::ReadError;

  size_t start = position_.offset;

  while (is_symbol_char(get())) forward();

  size_t end = position_.offset;

  String str = TRY(_str.sub(_arena, start, end));

  if (!is_separator(get()) && !is_eof()) return ErrorCode::ReadError;

  return Value(TRY(Symbol::create(_arena, str)));
}

char32_t Reader::get(size_t offset) {
  size_t pos = position_.offset + offset;
  auto res = _str[pos];
  if (res.has_value()) return *res;
  return 0;
}

bool Reader::is_eol() { return match('\n') || match('\r'); }

bool Reader::is_eof() { return position_.offset == _str.size(); }

bool Reader::is_whitespace() {
  return match(' ') || match('\t') || match('\v') || match('\f');

  return false;
}

bool Reader::is_comment_start() { return match(';'); }

bool Reader::is_string_start() { return match('"'); }

bool Reader::is_symbol_start() { return is_symbol_char(get()); }

bool Reader::is_numeric_start() {
  char32_t c = get();
  if (is_digit(c)) return true;
  if (c == '.' && is_digit(get(1))) return true;
  if ((c == '+' || c == '-') && is_digit(get(1))) return true;
  if ((c == '+' || c == '-') && get(1) == '.' && is_digit(get(2))) return true;
  return false;
}

bool Reader::match(const char* str) {
  size_t slen = strlen(str);
  for (size_t i = 0; i < slen; i++) {
    if (get(i) != char32_t(str[i])) return false;
  }
  return true;
}

bool Reader::match(char c) { return get() == char32_t(c); }

void Reader::forward(size_t n) {
  for (size_t i = 0; i < n; i++) {
    forward();
  }
}

void Reader::forward() {
  if (is_eof()) {
    return;
  }

  if (is_eol()) {
    if (match("\r\n")) {
      position_.offset++;
    }
    position_.line++;
    position_.column = 1;
  } else {
    position_.column++;
  }

  position_.offset++;
}

void Reader::forward_whitespace() {
  while (true) {
    if (is_eol()) {
      forward();
    } else if (is_whitespace()) {
      forward();
    } else if (is_comment_start()) {
      forward();
      do {
        forward();
      } while (!is_eof() && !is_eol());
    } else {
      break;
    }
  }
}

bool Reader::forward_decimal_number() {
  if (!is_digit(get())) return false;

  while (is_digit(get())) forward();
  return true;
}

bool Reader::forward_hex_number() {
  if (!is_hex_digit(get())) return false;

  while (is_hex_digit(get())) forward();
  return true;
}

bool Reader::forward_exponent() {
  if (!match('e') && !match('E')) return false;
  forward();

  if (match('-') || match('+')) forward();

  return forward_decimal_number();
}

Result<Value> read_one(Arena& arena, Value& value) {
  if (!value.is<String>()) return ErrorCode::TypeMismatch;
  auto r = Reader(arena, *value.to<String>());
  return r.read_one();
}
Result<Value> read_one(Arena& arena, String& value) {
  auto r = Reader(arena, value);
  return r.read_one();
}
Result<Value> read_one(Arena& arena, const char* value) {
  auto s = TRY(String::create(arena, value));
  auto r = Reader(arena, s);
  return r.read_one();
}