path: root/src/nvim/charset.c

/// @file charset.c
///
/// Code related to character sets.

#include <assert.h>
#include <errno.h>
#include <inttypes.h>
#include <limits.h>
#include <stdlib.h>
#include <string.h>
#include <sys/types.h>

#include "auto/config.h"
#include "klib/kvec.h"
#include "nvim/ascii_defs.h"
#include "nvim/buffer_defs.h"
#include "nvim/charset.h"
#include "nvim/cursor.h"
#include "nvim/eval/typval_defs.h"
#include "nvim/garray.h"
#include "nvim/globals.h"
#include "nvim/keycodes.h"
#include "nvim/macros_defs.h"
#include "nvim/mbyte.h"
#include "nvim/memory.h"
#include "nvim/option.h"
#include "nvim/path.h"
#include "nvim/pos_defs.h"
#include "nvim/strings.h"
#include "nvim/vim_defs.h"

#ifdef INCLUDE_GENERATED_DECLARATIONS
# include "charset.c.generated.h"
#endif

static bool chartab_initialized = false;

// b_chartab[] is an array with 256 bits, each bit representing one of the
// characters 0-255.
#define SET_CHARTAB(buf, c) \
  (buf)->b_chartab[(unsigned)(c) >> 6] |= (1ull << ((c) & 0x3f))
#define RESET_CHARTAB(buf, c) \
  (buf)->b_chartab[(unsigned)(c) >> 6] &= ~(1ull << ((c) & 0x3f))
#define GET_CHARTAB_TAB(chartab, c) \
  ((chartab)[(unsigned)(c) >> 6] & (1ull << ((c) & 0x3f)))

// Table used below, see init_chartab() for an explanation
static uint8_t g_chartab[256];

// Flags for g_chartab[].
#define CT_CELL_MASK  0x07  ///< mask: nr of display cells (1, 2 or 4)
#define CT_PRINT_CHAR 0x10  ///< flag: set for printable chars
#define CT_ID_CHAR    0x20  ///< flag: set for ID chars
#define CT_FNAME_CHAR 0x40  ///< flag: set for file name chars

/// Fill g_chartab[].  Also fills curbuf->b_chartab[] with flags for keyword
/// characters for current buffer.
///
/// Depends on the option settings 'iskeyword', 'isident', 'isfname',
/// 'isprint' and 'encoding'.
///
/// The index in g_chartab[] is the character when first byte is up to 0x80,
/// if the first byte is 0x80 and above it depends on further bytes.
///
/// The contents of g_chartab[]:
/// - The lower two bits, masked by CT_CELL_MASK, give the number of display
///   cells the character occupies (1 or 2).  Not valid for UTF-8 above 0x80.
/// - CT_PRINT_CHAR bit is set when the character is printable (no need to
///   translate the character before displaying it).  Note that only DBCS
///   characters can have 2 display cells and still be printable.
/// - CT_FNAME_CHAR bit is set when the character can be in a file name.
/// - CT_ID_CHAR bit is set when the character can be in an identifier.
///
/// @return FAIL if 'iskeyword', 'isident', 'isfname' or 'isprint' option has
/// an error, OK otherwise.
int init_chartab(void)
{
  return buf_init_chartab(curbuf, true);
}

/// Helper for init_chartab
///
/// @param global false: only set buf->b_chartab[]
///
/// @return FAIL if 'iskeyword', 'isident', 'isfname' or 'isprint' option has
/// an error, OK otherwise.
int buf_init_chartab(buf_T *buf, int global)
{
  int c;

  if (global) {
    // Set the default size for printable characters:
    // From <Space> to '~' is 1 (printable), others are 2 (not printable).
    // This also inits all 'isident' and 'isfname' flags to false.
    c = 0;

    while (c < ' ') {
      g_chartab[c++] = (dy_flags & DY_UHEX) ? 4 : 2;
    }

    while (c <= '~') {
      g_chartab[c++] = 1 + CT_PRINT_CHAR;
    }

    while (c < 256) {
      if (c >= 0xa0) {
        // UTF-8: bytes 0xa0 - 0xff are printable (latin1)
        // Also assume that every multi-byte char is a filename character.
        g_chartab[c++] = (CT_PRINT_CHAR | CT_FNAME_CHAR) + 1;
      } else {
        // the rest is unprintable by default
        g_chartab[c++] = (dy_flags & DY_UHEX) ? 4 : 2;
      }
    }
  }