// file      : libbutl/char-scanner.mxx -*- C++ -*-
// copyright : Copyright (c) 2014-2019 Code Synthesis Ltd
// license   : MIT; see accompanying LICENSE file

#ifndef __cpp_modules_ts
#pragma once
#endif

#include <cassert>

#ifndef __cpp_lib_modules_ts
#include <string>  // char_traits
#include <cstdint> // uint64_t
#include <istream>
#endif

// Other includes.

#ifdef __cpp_modules_ts
export module butl.char_scanner;
#ifdef __cpp_lib_modules_ts
import std.core;
import std.io;
#endif
import butl.fdstream;
#else
#include <libbutl/fdstream.mxx>
#endif

#include <libbutl/export.hxx>

LIBBUTL_MODEXPORT namespace butl
{
  // Low-level character stream scanner. Normally used as a base for
  // higher-level lexers.
  //
  class LIBBUTL_SYMEXPORT char_scanner
  {
  public:
    // If the crlf argument is true, then recognize Windows newlines (0x0D
    // 0x0A) and convert them to just '\n' (0x0A). Note that a standalone
    // 0x0D is treated "as if" it was followed by 0x0A and multiple 0x0D
    // are treated as one.
    //
    // Note also that if the stream happens to be ifdstream, then it includes
    // a number of optimizations that assume nobody else is messing with the
    // stream.
    //
    // The line and position arguments can be used to override the start line
    // and position in the stream (useful when re-scanning data saved with the
    // save_* facility).
    //
    char_scanner (std::istream& is,
                  bool crlf = true,
                  std::uint64_t line = 1,
                  std::uint64_t position = 0);

    char_scanner (const char_scanner&) = delete;
    char_scanner& operator= (const char_scanner&) = delete;

    // Scanner interface.
    //
  public:

    // Extended character. It includes line/column/position information and is
    // capable of representing EOF.
    //
    // Note that implicit conversion of EOF to char_type results in NUL
    // character (which means in most cases it is safe to compare xchar to
    // char without checking for EOF).
    //
    class xchar
    {
    public:
      using traits_type = std::char_traits<char>;
      using int_type = traits_type::int_type;
      using char_type = traits_type::char_type;

      int_type value;
      std::uint64_t line;
      std::uint64_t column;

      // Logical character position (see ifdstream for details on the logical
      // part) if the scanned stream is ifdstream and always zero otherwise.
      //
      std::uint64_t position;

      operator char_type () const
      {
        return value != traits_type::eof ()
          ? static_cast<char_type> (value)
          : char_type (0);
      }

      xchar (int_type v,
             std::uint64_t l = 0,
             std::uint64_t c = 0,
             std::uint64_t p = 0)
          : value (v), line (l), column (c), position (p) {}
    };

    xchar
    get ();

    void
    get (const xchar& peeked); // Get previously peeked character (faster).

    void
    unget (const xchar&);

    // Note that if there is an "ungot" character, peek() will return
    // that.
    //
    xchar
    peek ();

    // Tests. In the future we can add tests line alpha(), alnum(),
    // etc.
    //
    static bool
    eos (const xchar& c) {return c.value == xchar::traits_type::eof ();}

    // Line, column and position of the next character to be extracted from
    // the stream by peek() or get().
    //
    std::uint64_t line;
    std::uint64_t column;
    std::uint64_t position;

    // Ability to save raw data as it is being scanned. Note that the
    // character is only saved when it is got, not peeked.
    //
  public:
    void
    save_start (std::string& b)
    {
      assert (save_ == nullptr);
      save_ = &b;
    }

    void
    save_stop ()
    {
      assert (save_ != nullptr);
      save_ = nullptr;
    }

    struct save_guard
    {
      explicit
      save_guard (char_scanner& s, std::string& b): s_ (&s) {s.save_start (b);}

      void
      stop () {if (s_ != nullptr) {s_->save_stop (); s_ = nullptr;}}

      ~save_guard () {stop ();}

    private:
      char_scanner* s_;
    };

  protected:
    using int_type = xchar::int_type;
    using char_type = xchar::char_type;

    int_type
    peek_ ();

    void
    get_ ();

    std::uint64_t
    pos_ () const;

  protected:
    std::istream& is_;

    // Note that if you are reading from the buffer directly, then it is
    // also your responsibility to save the data.
    //
    fdbuf* buf_; // NULL if not ifdstream.
    const char_type* gptr_;
    const char_type* egptr_;

    std::string* save_ = nullptr;

    bool crlf_;
    bool eos_ = false;

    bool unget_ = false;
    bool unpeek_ = false;

    xchar ungetc_ = '\0';
    xchar unpeekc_ = '\0';
  };
}

#include <libbutl/char-scanner.ixx>