secs-gem/src/secs2/sml.cpp

#include "secsgem/secs2/sml.hpp"

#include <cctype>
#include <cstdlib>
#include <string>
#include <vector>

namespace secsgem::secs2 {

namespace {

struct Cursor {
  const std::string& s;
  std::size_t i = 0;

  bool done() const { return i >= s.size(); }
  char peek() const { return s[i]; }
  char get() { return s[i++]; }

  void skip_ws() {
    while (i < s.size() && std::isspace(static_cast<unsigned char>(s[i]))) ++i;
  }

  void expect(char c) {
    skip_ws();
    if (i >= s.size() || s[i] != c)
      throw SmlError(std::string("expected `") + c + "` at " + std::to_string(i));
    ++i;
  }

  bool consume(char c) {
    skip_ws();
    if (i < s.size() && s[i] == c) { ++i; return true; }
    return false;
  }

  std::string read_word() {
    skip_ws();
    const std::size_t start = i;
    while (i < s.size() &&
           !std::isspace(static_cast<unsigned char>(s[i])) &&
           s[i] != '<' && s[i] != '>' && s[i] != '[' && s[i] != ']' &&
           s[i] != '"') {
      ++i;
    }
    if (i == start) throw SmlError("expected word at " + std::to_string(i));
    return s.substr(start, i - start);
  }

  std::string read_quoted() {
    skip_ws();
    if (i >= s.size() || s[i] != '"') throw SmlError("expected `\"`");
    ++i;
    const std::size_t start = i;
    while (i < s.size() && s[i] != '"') {
      // No escape sequences in the SML emitter; mirror that here.
      ++i;
    }
    if (i >= s.size()) throw SmlError("unterminated string");
    std::string out = s.substr(start, i - start);
    ++i;  // consume closing quote
    return out;
  }
};

Format parse_tag(const std::string& tag) {
  if (tag == "L")       return Format::List;
  if (tag == "A")       return Format::ASCII;
  if (tag == "J")       return Format::JIS8;
  if (tag == "C")       return Format::C2;
  if (tag == "B")       return Format::Binary;
  if (tag == "BOOLEAN") return Format::Boolean;
  if (tag == "U1")      return Format::U1;
  if (tag == "U2")      return Format::U2;
  if (tag == "U4")      return Format::U4;
  if (tag == "U8")      return Format::U8;
  if (tag == "I1")      return Format::I1;
  if (tag == "I2")      return Format::I2;
  if (tag == "I4")      return Format::I4;
  if (tag == "I8")      return Format::I8;
  if (tag == "F4")      return Format::F4;
  if (tag == "F8")      return Format::F8;
  throw SmlError("unknown SML tag `" + tag + "`");
}

// Parse one byte literal — accepts 0xHH and decimal forms.
uint8_t parse_byte(const std::string& w) {
  return static_cast<uint8_t>(std::strtoul(w.c_str(), nullptr, 0));
}

Item parse_item(Cursor& c);

Item parse_scalar_body(Cursor& c, Format fmt) {
  switch (fmt) {
    case Format::ASCII:
    case Format::JIS8: {
      auto s = c.read_quoted();
      return fmt == Format::ASCII ? Item::ascii(s) : Item::jis8(s);
    }
    case Format::C2: {
      // C2 is a vector of 16-bit code points; SML round-trip for C2 is
      // not yet specified by the emitter — accept whitespace-separated
      // integers for now.
      std::vector<uint16_t> v;
      while (true) {
        c.skip_ws();
        if (c.done() || c.peek() == '>') break;
        v.push_back(static_cast<uint16_t>(std::strtoul(c.read_word().c_str(), nullptr, 0)));
      }
      return Item::c2(v);
    }
    case Format::Binary: {
      std::vector<uint8_t> bytes;
      while (true) {
        c.skip_ws();
        if (c.done() || c.peek() == '>') break;
        bytes.push_back(parse_byte(c.read_word()));
      }
      return Item::binary(bytes);
    }
    case Format::Boolean: {
      std::vector<uint8_t> bytes;
      while (true) {
        c.skip_ws();
        if (c.done() || c.peek() == '>') break;
        auto w = c.read_word();
        bool b = (w == "T" || w == "TRUE" || w == "true" || w == "1");
        bytes.push_back(b ? 1 : 0);
      }
      return Item::boolean(bytes);
    }
#define SML_NUM(fmtT, ctype, factory, parser)                                 \
    case Format::fmtT: {                                                      \
      std::vector<ctype> v;                                                   \
      while (true) {                                                          \
        c.skip_ws();                                                          \
        if (c.done() || c.peek() == '>') break;                               \
        auto w = c.read_word();                                               \
        v.push_back(static_cast<ctype>(parser(w.c_str(), nullptr, 0)));       \
      }                                                                       \
      return factory(v);                                                      \
    }
    SML_NUM(U1, uint8_t,  Item::u1, std::strtoul)
    SML_NUM(U2, uint16_t, Item::u2, std::strtoul)
    SML_NUM(U4, uint32_t, Item::u4, std::strtoul)
    SML_NUM(U8, uint64_t, Item::u8, std::strtoull)
    SML_NUM(I1, int8_t,   Item::i1, std::strtol)
    SML_NUM(I2, int16_t,  Item::i2, std::strtol)
    SML_NUM(I4, int32_t,  Item::i4, std::strtol)
    SML_NUM(I8, int64_t,  Item::i8, std::strtoll)
#undef SML_NUM
    case Format::F4: {
      std::vector<float> v;
      while (true) {
        c.skip_ws();
        if (c.done() || c.peek() == '>') break;
        v.push_back(std::strtof(c.read_word().c_str(), nullptr));
      }
      return Item::f4(v);
    }
    case Format::F8: {
      std::vector<double> v;
      while (true) {
        c.skip_ws();
        if (c.done() || c.peek() == '>') break;
        v.push_back(std::strtod(c.read_word().c_str(), nullptr));
      }
      return Item::f8(v);
    }
    case Format::List:
      // Unreachable; handled in parse_item.
      throw SmlError("internal: List shape in scalar dispatch");
  }
  throw SmlError("unhandled format in scalar dispatch");
}

Item parse_item(Cursor& c) {
  c.expect('<');
  auto tag = c.read_word();
  Format fmt = parse_tag(tag);

  // Optional `[n]` count — accepted, ignored.
  if (c.consume('[')) {
    (void)c.read_word();
    c.expect(']');
  }

  if (fmt == Format::List) {
    Item::List children;
    while (true) {
      c.skip_ws();
      if (c.done()) throw SmlError("unterminated list");
      if (c.peek() == '>') { c.get(); break; }
      children.push_back(parse_item(c));
    }
    return Item::list(std::move(children));
  }

  Item result = parse_scalar_body(c, fmt);
  c.expect('>');
  return result;
}

}  // namespace

Item parse_sml(const std::string& text) {
  Cursor c{text};
  c.skip_ws();
  return parse_item(c);
}

std::optional<Item> try_parse_sml(const std::string& text) {
  try {
    return parse_sml(text);
  } catch (const SmlError&) {
    return std::nullopt;
  }
}

}  // namespace secsgem::secs2