/* Header file for the buffer manipulation primitives.
   Copyright (C) 1985, 1986, 1992, 1993, 1994, 1995
   Free Software Foundation, Inc.
   Copyright (C) 1995 Sun Microsystems, Inc.
   Copyright (C) 2001, 2002 Ben Wing.

This file is part of XEmacs.

XEmacs is free software; you can redistribute it and/or modify it
under the terms of the GNU General Public License as published by the
Free Software Foundation; either version 2, or (at your option) any
later version.

XEmacs is distributed in the hope that it will be useful, but WITHOUT
ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
for more details.

You should have received a copy of the GNU General Public License
along with XEmacs; see the file COPYING.  If not, write to
the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
Boston, MA 02111-1307, USA.  */

/* Synched up with: FSF 19.30. */

/* Authorship:

   FSF: long ago.
   JWZ: separated out bufslots.h, early in Lemacs.
   Ben Wing: almost completely rewritten for Mule, 19.12.
 */

#ifndef INCLUDED_buffer_h_
#define INCLUDED_buffer_h_

#include "casetab.h"
#include "chartab.h"

/************************************************************************/
/*                                                                      */
/*                    definition of Lisp buffer object                  */
/*                                                                      */
/************************************************************************/

/* Note: we keep both Bytebpos and Charbpos versions of some of the
   important buffer positions because they are accessed so much.
   If we didn't do this, we would constantly be invalidating the
   charbpos<->bytebpos cache under Mule.

   Note that under non-Mule, both versions will always be the
   same so we don't really need to keep track of them.  But it
   simplifies the logic to go ahead and do so all the time and
   the memory loss is insignificant. */

/* Formerly, it didn't much matter what went inside the struct buffer_text
   and what went outside it.  Now it does, with the advent of "indirect
   buffers" that share text with another buffer.  An indirect buffer
   shares the same *text* as another buffer, but has its own buffer-local
   variables, its own accessible region, and its own markers and extents.
   (Due to the nature of markers, it doesn't actually matter much whether
   we stick them inside or out of the struct buffer_text -- the user won't
   notice any difference -- but we go ahead and put them outside for
   consistency and overall saneness of algorithm.)

   FSFmacs gets away with not maintaining any "children" pointers from
   a buffer to the indirect buffers that refer to it by putting the
   markers inside of the struct buffer_text, using markers to keep track
   of BEGV and ZV in indirect buffers, and relying on the fact that
   all intervals (text properties and overlays) use markers for their
   start and end points.  We don't do this for extents (markers are
   inefficient anyway and take up space), so we have to maintain
   children pointers.  This is not terribly hard, though, and the
   code to maintain this is just like the code already present in
   extent-parent and extent-children.
   */

struct buffer_text
{
  Intbyte *beg;		/* Actual address of buffer contents. */
  Bytebpos gpt;		/* Index of gap in buffer. */
  Bytebpos z;		/* Index of end of buffer. */
  Charbpos bufz;		/* Equivalent as a Charbpos. */
  Bytecount gap_size;/* Size of buffer's gap */
  Bytecount end_gap_size;/* Size of buffer's end gap */
  long modiff;		/* This counts buffer-modification events
			   for this buffer.  It is incremented for
			   each such event, and never otherwise
			   changed.  */
  long save_modiff;	/* Previous value of modiff, as of last
			   time buffer visited or saved a file.  */

#ifdef MULE
  /* We keep track of a "known" region for very fast access.  This
     information is text-only so it goes here.  We update this at each
     change to the buffer, so if it's entirely ASCII, these will always
     contain the minimum and maximum positions of the buffer. */
  Charbpos mule_bufmin, mule_bufmax;
  Bytebpos mule_bytmin, mule_bytmax;
  int mule_shifter, mule_three_p;

  /* And we also cache 16 positions for fairly fast access near those
     positions. */
  Charbpos mule_charbpos_cache[16];
  Bytebpos mule_bytebpos_cache[16];

  int entirely_ascii_p;
#endif

  /* Similar to the above, we keep track of positions for which line
     number has last been calculated.  See line-number.c. */
  Lisp_Object line_number_cache;

  /* Change data that goes with the text. */
  struct buffer_text_change_data *changes;
};

struct buffer
{
  struct lcrecord_header header;

  /* This structure holds the coordinates of the buffer contents
     in ordinary buffers.  In indirect buffers, this is not used.  */
  struct buffer_text own_text;

  /* This points to the `struct buffer_text' that is used for this buffer.
     In an ordinary buffer, this is the own_text field above.
     In an indirect buffer, this is the own_text field of another buffer.  */
  struct buffer_text *text;

  Bytebpos pt;		/* Position of point in buffer. */
  Charbpos bufpt;		/* Equivalent as a Charbpos. */
  Bytebpos begv;		/* Index of beginning of accessible range. */
  Charbpos bufbegv;	/* Equivalent as a Charbpos. */
  Bytebpos zv;		/* Index of end of accessible range. */
  Charbpos bufzv;		/* Equivalent as a Charbpos. */

  int face_change;	/* This is set when a change in how the text should
			   be displayed (e.g., font, color) is made. */

  /* Whether buffer specific face is specified. */
  int buffer_local_face_property;

  /* change data indicating what portion of the text has changed
     since the last time this was reset.  Used by redisplay.
     Logically we should keep this with the text structure, but
     redisplay resets it for each buffer individually and we don't
     want interference between an indirect buffer and its base
     buffer. */
  struct each_buffer_change_data *changes;

#ifdef REGION_CACHE_NEEDS_WORK
  /* If the long line scan cache is enabled (i.e. the buffer-local
     variable cache-long-line-scans is non-nil), newline_cache
     points to the newline cache, and width_run_cache points to the
     width run cache.

     The newline cache records which stretches of the buffer are
     known *not* to contain newlines, so that they can be skipped
     quickly when we search for newlines.

     The width run cache records which stretches of the buffer are
     known to contain characters whose widths are all the same.  If
     the width run cache maps a character to a value > 0, that value
     is the character's width; if it maps a character to zero, we
     don't know what its width is.  This allows compute_motion to
     process such regions very quickly, using algebra instead of
     inspecting each character.  See also width_table, below.  */
  struct region_cache *newline_cache;
  struct region_cache *width_run_cache;
#endif /* REGION_CACHE_NEEDS_WORK */

  /* The markers that refer to this buffer.  This is actually a single
     marker -- successive elements in its marker `chain' are the other
     markers referring to this buffer */
  Lisp_Marker *markers;

  /* The buffer's extent info.  This is its own type, an extent-info
     object (done this way for ease in marking / finalizing). */
  Lisp_Object extent_info;

  /* ----------------------------------------------------------------- */
  /* All the stuff above this line is the responsibility of insdel.c,
     with some help from marker.c and extents.c.
     All the stuff below this line is the responsibility of buffer.c. */

  /* In an indirect buffer, this points to the base buffer.
     In an ordinary buffer, it is 0.
     We DO mark through this slot. */
  struct buffer *base_buffer;

  /* List of indirect buffers whose base is this buffer.
     If we are an indirect buffer, this will be nil.
     Do NOT mark through this. */
  Lisp_Object indirect_children;

  /* Flags saying which DEFVAR_PER_BUFFER variables
     are local to this buffer.  */
  int local_var_flags;

  /* Set to the modtime of the visited file when read or written.
     -1 means visited file was nonexistent.
     0  means visited file modtime unknown; in no case complain
     about any mismatch on next save attempt.  */
  int modtime;

  /* the value of text->modiff at the last auto-save.  */
  long auto_save_modified;

  /* The time at which we detected a failure to auto-save,
     Or -1 if we didn't have a failure.  */
  int auto_save_failure_time;

  /* Position in buffer at which display started
     the last time this buffer was displayed.  */
  int last_window_start;

  /* Everything from here down must be a Lisp_Object */

#define MARKED_SLOT(x) Lisp_Object x
#include "bufslots.h"
#undef MARKED_SLOT
};

DECLARE_LRECORD (buffer, struct buffer);
#define XBUFFER(x) XRECORD (x, buffer, struct buffer)
#define wrap_buffer(p) wrap_record (p, buffer)
#define BUFFERP(x) RECORDP (x, buffer)
#define CHECK_BUFFER(x) CHECK_RECORD (x, buffer)
#define CONCHECK_BUFFER(x) CONCHECK_RECORD (x, buffer)

#define BUFFER_LIVE_P(b) (!NILP ((b)->name))

#define CHECK_LIVE_BUFFER(x) do {			\
  CHECK_BUFFER (x);					\
  if (!BUFFER_LIVE_P (XBUFFER (x)))			\
    dead_wrong_type_argument (Qbuffer_live_p, (x));	\
} while (0)

#define CONCHECK_LIVE_BUFFER(x) do {			\
  CONCHECK_BUFFER (x);					\
  if (!BUFFER_LIVE_P (XBUFFER (x)))			\
    x = wrong_type_argument (Qbuffer_live_p, (x));	\
} while (0)


#define BUFFER_BASE_BUFFER(b) ((b)->base_buffer ? (b)->base_buffer : (b))

/* Map over buffers sharing the same text as MPS_BUF.  MPS_BUFVAR is a
   variable that gets the buffer values (beginning with the base
   buffer, then the children), and MPS_BUFCONS should be a temporary
   Lisp_Object variable.  */
#define MAP_INDIRECT_BUFFERS(mps_buf, mps_bufvar, mps_bufcons)	\
for (mps_bufcons = Qunbound,					\
     mps_bufvar = BUFFER_BASE_BUFFER (mps_buf);			\
     UNBOUNDP (mps_bufcons) ?					\
	(mps_bufcons = mps_bufvar->indirect_children,		\
	1)							\
       : (!NILP (mps_bufcons)					\
	  && (mps_bufvar = XBUFFER (XCAR (mps_bufcons)), 1)	\
	  && (mps_bufcons = XCDR (mps_bufcons), 1));		\
     )


/*----------------------------------------------------------------------*/
/*	    Accessor macros for important positions in a buffer		*/
/*----------------------------------------------------------------------*/

/* We put them here because some stuff below wants them before the
   place where we would normally put them. */

/* None of these are lvalues.  Use the settor macros below to change
   the positions. */

/* Beginning of buffer.  */
#define BI_BUF_BEG(buf) ((Bytebpos) 1)
#define BUF_BEG(buf) ((Charbpos) 1)

/* Beginning of accessible range of buffer.  */
#define BI_BUF_BEGV(buf) ((buf)->begv + 0)
#define BUF_BEGV(buf) ((buf)->bufbegv + 0)

/* End of accessible range of buffer.  */
#define BI_BUF_ZV(buf) ((buf)->zv + 0)
#define BUF_ZV(buf) ((buf)->bufzv + 0)

/* End of buffer.  */
#define BI_BUF_Z(buf) ((buf)->text->z + 0)
#define BUF_Z(buf) ((buf)->text->bufz + 0)

/* Point. */
#define BI_BUF_PT(buf) ((buf)->pt + 0)
#define BUF_PT(buf) ((buf)->bufpt + 0)

/*----------------------------------------------------------------------*/
/*		Converting between positions and addresses		*/
/*----------------------------------------------------------------------*/

/* Convert the address of a byte in the buffer into a position.  */
INLINE_HEADER Bytebpos BI_BUF_PTR_BYTE_POS (struct buffer *buf, Intbyte *ptr);
INLINE_HEADER Bytebpos
BI_BUF_PTR_BYTE_POS (struct buffer *buf, Intbyte *ptr)
{
  return (ptr - buf->text->beg + 1
	  - ((ptr - buf->text->beg + 1) > buf->text->gpt
	     ? buf->text->gap_size : 0));
}

#define BUF_PTR_BYTE_POS(buf, ptr) \
  bytebpos_to_charbpos (buf, BI_BUF_PTR_BYTE_POS (buf, ptr))

/* Address of byte at position POS in buffer. */
INLINE_HEADER Intbyte * BI_BUF_BYTE_ADDRESS (struct buffer *buf, Bytebpos pos);
INLINE_HEADER Intbyte *
BI_BUF_BYTE_ADDRESS (struct buffer *buf, Bytebpos pos)
{
  return (buf->text->beg +
	  ((pos >= buf->text->gpt ? (pos + buf->text->gap_size) : pos)
	   - 1));
}

#define BUF_BYTE_ADDRESS(buf, pos) \
  BI_BUF_BYTE_ADDRESS (buf, charbpos_to_bytebpos (buf, pos))

/* Address of byte before position POS in buffer. */
INLINE_HEADER Intbyte * BI_BUF_BYTE_ADDRESS_BEFORE (struct buffer *buf, Bytebpos pos);
INLINE_HEADER Intbyte *
BI_BUF_BYTE_ADDRESS_BEFORE (struct buffer *buf, Bytebpos pos)
{
  return (buf->text->beg +
	  ((pos > buf->text->gpt ? (pos + buf->text->gap_size) : pos)
	   - 2));
}

#define BUF_BYTE_ADDRESS_BEFORE(buf, pos) \
  BI_BUF_BYTE_ADDRESS_BEFORE (buf, charbpos_to_bytebpos (buf, pos))

/*----------------------------------------------------------------------*/
/*	    Converting between byte indices and memory indices		*/
/*----------------------------------------------------------------------*/

INLINE_HEADER int valid_membpos_p (struct buffer *buf, Membpos x);
INLINE_HEADER int
valid_membpos_p (struct buffer *buf, Membpos x)
{
  return ((x >= 1 && x <= (Membpos) buf->text->gpt) ||
	  (x  > (Membpos) (buf->text->gpt + buf->text->gap_size) &&
	   x <= (Membpos) (buf->text->z   + buf->text->gap_size)));
}

INLINE_HEADER Membpos bytebpos_to_membpos (struct buffer *buf, Bytebpos x);
INLINE_HEADER Membpos
bytebpos_to_membpos (struct buffer *buf, Bytebpos x)
{
  return (Membpos) ((x > buf->text->gpt) ? (x + buf->text->gap_size) : x);
}


INLINE_HEADER Bytebpos membpos_to_bytebpos (struct buffer *buf, Membpos x);
INLINE_HEADER Bytebpos
membpos_to_bytebpos (struct buffer *buf, Membpos x)
{
#ifdef ERROR_CHECK_TEXT
  assert (valid_membpos_p (buf, x));
#endif
  return (Bytebpos) ((x > (Membpos) buf->text->gpt) ?
		   x - buf->text->gap_size :
		   x);
}

#define membpos_to_charbpos(buf, x) \
  bytebpos_to_charbpos (buf, membpos_to_bytebpos (buf, x))
#define charbpos_to_membpos(buf, x) \
  bytebpos_to_membpos (buf, charbpos_to_bytebpos (buf, x))

/* These macros generalize many standard buffer-position functions to
   either a buffer or a string. */

/* Converting between Membposs and Bytebposs, for a buffer-or-string.
   For strings, this is a no-op.  For buffers, this resolves
   to the standard membpos<->bytebpos converters. */

#define buffer_or_string_bytebpos_to_membpos(obj, ind) \
  (BUFFERP (obj) ? bytebpos_to_membpos (XBUFFER (obj), ind) : (Membpos) ind)

#define buffer_or_string_membpos_to_bytebpos(obj, ind) \
  (BUFFERP (obj) ? membpos_to_bytebpos (XBUFFER (obj), ind) : (Bytebpos) ind)

/* Converting between Charbpos's and Bytebposs, for a buffer-or-string.
   For strings, this maps to the bytecount<->charcount converters. */

#define buffer_or_string_charbpos_to_bytebpos(obj, pos)		\
  (BUFFERP (obj) ? charbpos_to_bytebpos (XBUFFER (obj), pos) :	\
   (Bytebpos) string_index_char_to_byte (obj, pos))

#define buffer_or_string_bytebpos_to_charbpos(obj, ind)		\
  (BUFFERP (obj) ? bytebpos_to_charbpos (XBUFFER (obj), ind) :	\
   (Charbpos) string_index_byte_to_char (obj, ind))

/* Similar for Charbpos's and Membposs. */

#define buffer_or_string_charbpos_to_membpos(obj, pos)		\
  (BUFFERP (obj) ? charbpos_to_membpos (XBUFFER (obj), pos) :	\
   (Membpos) string_index_char_to_byte (obj, pos))

#define buffer_or_string_membpos_to_charbpos(obj, ind)		\
  (BUFFERP (obj) ? membpos_to_charbpos (XBUFFER (obj), ind) :	\
   (Charbpos) string_index_byte_to_char (obj, ind))

/************************************************************************/
/*                                                                      */
/*                    working with buffer-level data                    */
/*                                                                      */
/************************************************************************/

/*

   (A) Working with byte indices:
   ------------------------------

   VALID_BYTEBPOS_P(buf, bi):
	Given a byte index, does it point to the beginning of a character?

   ASSERT_VALID_BYTEBPOS_UNSAFE(buf, bi):
	If error-checking is enabled, assert that the given byte index
	is within range and points to the beginning of a character
	or to the end of the buffer.  Otherwise, do nothing.

   ASSERT_VALID_BYTEBPOS_BACKWARD_UNSAFE(buf, bi):
	If error-checking is enabled, assert that the given byte index
	is within range and satisfies ASSERT_VALID_BYTEBPOS() and also
        does not refer to the beginning of the buffer. (i.e. movement
	backwards is OK.) Otherwise, do nothing.

   ASSERT_VALID_BYTEBPOS_FORWARD_UNSAFE(buf, bi):
	If error-checking is enabled, assert that the given byte index
	is within range and satisfies ASSERT_VALID_BYTEBPOS() and also
        does not refer to the end of the buffer. (i.e. movement
	forwards is OK.) Otherwise, do nothing.

   VALIDATE_BYTEBPOS_BACKWARD(buf, bi):
	Make sure that the given byte index is pointing to the beginning
	of a character.  If not, back up until this is the case.  Note
	that there are not too many places where it is legitimate to do
	this sort of thing.  It's an error if you're passed an "invalid"
	byte index.

   VALIDATE_BYTEBPOS_FORWARD(buf, bi):
	Make sure that the given byte index is pointing to the beginning
	of a character.  If not, move forward until this is the case.
	Note that there are not too many places where it is legitimate
	to do this sort of thing.  It's an error if you're passed an
	"invalid" byte index.

   INC_BYTEBPOS(buf, bi):
	Given a byte index (assumed to point at the beginning of a
	character), modify that value so it points to the beginning
	of the next character.

   DEC_BYTEBPOS(buf, bi):
	Given a byte index (assumed to point at the beginning of a
	character), modify that value so it points to the beginning
	of the previous character.  Unlike for DEC_CHARPTR(), we can
	do all the assert()s because there are sentinels at the
	beginning of the gap and the end of the buffer.

   BYTEBPOS_INVALID:
	A constant representing an invalid Bytebpos.  Valid Bytebposs
	can never have this value.


   (B) Converting between Charbpos's and Bytebposs:
   --------------------------------------------

    charbpos_to_bytebpos(buf, bu):
	Given a Charbpos, return the equivalent Bytebpos.

    bytebpos_to_charbpos(buf, bi):
	Given a Bytebpos, return the equivalent Charbpos.

    make_charbpos(buf, bi):
	Given a Bytebpos, return the equivalent Charbpos as a Lisp Object.
 */


/*----------------------------------------------------------------------*/
/*			 working with byte indices			*/
/*----------------------------------------------------------------------*/

#ifdef MULE
# define VALID_BYTEBPOS_P(buf, x) \
  INTBYTE_FIRST_BYTE_P (*BI_BUF_BYTE_ADDRESS (buf, x))
#else
# define VALID_BYTEBPOS_P(buf, x) 1
#endif

#ifdef ERROR_CHECK_TEXT

# define ASSERT_VALID_BYTEBPOS_UNSAFE(buf, x) do {		\
  assert (BUFFER_LIVE_P (buf));					\
  assert ((x) >= BI_BUF_BEG (buf) && x <= BI_BUF_Z (buf));	\
  assert (VALID_BYTEBPOS_P (buf, x));				\
} while (0)
# define ASSERT_VALID_BYTEBPOS_BACKWARD_UNSAFE(buf, x) do {	\
  assert (BUFFER_LIVE_P (buf));					\
  assert ((x) > BI_BUF_BEG (buf) && x <= BI_BUF_Z (buf));	\
  assert (VALID_BYTEBPOS_P (buf, x));				\
} while (0)
# define ASSERT_VALID_BYTEBPOS_FORWARD_UNSAFE(buf, x) do {	\
  assert (BUFFER_LIVE_P (buf));					\
  assert ((x) >= BI_BUF_BEG (buf) && x < BI_BUF_Z (buf));	\
  assert (VALID_BYTEBPOS_P (buf, x));				\
} while (0)

#else /* not ERROR_CHECK_TEXT */
# define ASSERT_VALID_BYTEBPOS_UNSAFE(buf, x)
# define ASSERT_VALID_BYTEBPOS_BACKWARD_UNSAFE(buf, x)
# define ASSERT_VALID_BYTEBPOS_FORWARD_UNSAFE(buf, x)

#endif /* not ERROR_CHECK_TEXT */

/* Note that, although the Mule version will work fine for non-Mule
   as well (it should reduce down to nothing), we provide a separate
   version to avoid compilation warnings and possible non-optimal
   results with stupid compilers. */

#ifdef MULE
# define VALIDATE_BYTEBPOS_BACKWARD(buf, x) do {		\
  Intbyte *VBB_ptr = BI_BUF_BYTE_ADDRESS (buf, x);	\
  while (!INTBYTE_FIRST_BYTE_P (*VBB_ptr))		\
    VBB_ptr--, (x)--;					\
} while (0)
#else
# define VALIDATE_BYTEBPOS_BACKWARD(buf, x)
#endif

/* Note that, although the Mule version will work fine for non-Mule
   as well (it should reduce down to nothing), we provide a separate
   version to avoid compilation warnings and possible non-optimal
   results with stupid compilers. */

#ifdef MULE
# define VALIDATE_BYTEBPOS_FORWARD(buf, x) do {		\
  Intbyte *VBF_ptr = BI_BUF_BYTE_ADDRESS (buf, x);	\
  while (!INTBYTE_FIRST_BYTE_P (*VBF_ptr))		\
    VBF_ptr++, (x)++;					\
} while (0)
#else
# define VALIDATE_BYTEBPOS_FORWARD(buf, x)
#endif

/* Note that in the simplest case (no MULE, no ERROR_CHECK_TEXT),
   this crap reduces down to simply (x)++. */

#define INC_BYTEBPOS(buf, x) do				\
{							\
  ASSERT_VALID_BYTEBPOS_FORWARD_UNSAFE (buf, x);		\
  /* Note that we do the increment first to		\
     make sure that the pointer in			\
     VALIDATE_BYTEBPOS_FORWARD() ends up on		\
     the correct side of the gap */			\
  (x)++;						\
  VALIDATE_BYTEBPOS_FORWARD (buf, x);			\
} while (0)

/* Note that in the simplest case (no MULE, no ERROR_CHECK_TEXT),
   this crap reduces down to simply (x)--. */

#define DEC_BYTEBPOS(buf, x) do				\
{							\
  ASSERT_VALID_BYTEBPOS_BACKWARD_UNSAFE (buf, x);	\
  /* Note that we do the decrement first to		\
     make sure that the pointer in			\
     VALIDATE_BYTEBPOS_BACKWARD() ends up on		\
     the correct side of the gap */			\
  (x)--;						\
  VALIDATE_BYTEBPOS_BACKWARD (buf, x);			\
} while (0)

INLINE_HEADER Bytebpos prev_bytebpos (struct buffer *buf, Bytebpos x);
INLINE_HEADER Bytebpos
prev_bytebpos (struct buffer *buf, Bytebpos x)
{
  DEC_BYTEBPOS (buf, x);
  return x;
}

INLINE_HEADER Bytebpos next_bytebpos (struct buffer *buf, Bytebpos x);
INLINE_HEADER Bytebpos
next_bytebpos (struct buffer *buf, Bytebpos x)
{
  INC_BYTEBPOS (buf, x);
  return x;
}

#define BYTEBPOS_INVALID ((Bytebpos) -1)

/*----------------------------------------------------------------------*/
/*	   Converting between buffer positions and byte indices		*/
/*----------------------------------------------------------------------*/

#ifdef MULE

Bytebpos charbpos_to_bytebpos_func (struct buffer *buf, Charbpos x);
Charbpos bytebpos_to_charbpos_func (struct buffer *buf, Bytebpos x);

/* The basic algorithm we use is to keep track of a known region of
   characters in each buffer, all of which are of the same width.  We keep
   track of the boundaries of the region in both Charbpos and Bytebpos
   coordinates and also keep track of the char width, which is 1 - 4 bytes.
   If the position we're translating is not in the known region, then we
   invoke a function to update the known region to surround the position in
   question.  This assumes locality of reference, which is usually the
   case.

   Note that the function to update the known region can be simple or
   complicated depending on how much information we cache.  In addition to
   the known region, we always cache the correct conversions for point,
   BEGV, and ZV, and in addition to this we cache 16 positions where the
   conversion is known.  We only look in the cache or update it when we
   need to move the known region more than a certain amount (currently 50
   chars), and then we throw away a "random" value and replace it with the
   newly calculated value.

   Finally, we maintain an extra flag that tracks whether the buffer is
   entirely ASCII, to speed up the conversions even more.  This flag is
   actually of dubious value because in an entirely-ASCII buffer the known
   region will always span the entire buffer (in fact, we update the flag
   based on this fact), and so all we're saving is a few machine cycles.

   A potentially smarter method than what we do with known regions and
   cached positions would be to keep some sort of pseudo-extent layer over
   the buffer; maybe keep track of the charbpos/bytebpos correspondence at the
   beginning of each line, which would allow us to do a binary search over
   the pseudo-extents to narrow things down to the correct line, at which
   point you could use a linear movement method.  This would also mesh well
   with efficiently implementing a line-numbering scheme.  However, you
   have to weigh the amount of time spent updating the cache vs. the
   savings that result from it.  In reality, we modify the buffer far less
   often than we access it, so a cache of this sort that provides
   guaranteed LOG (N) performance (or perhaps N * LOG (N), if we set a
   maximum on the cache size) would indeed be a win, particularly in very
   large buffers.  If we ever implement this, we should probably set a
   reasonably high minimum below which we use the old method, because the
   time spent updating the fancy cache would likely become dominant when
   making buffer modifications in smaller buffers.

   Note also that we have to multiply or divide by the char width in order
   to convert the positions.  We do some tricks to avoid ever actually
   having to do a multiply or divide, because that is typically an
   expensive operation (esp. divide).  Multiplying or dividing by 1, 2, or
   4 can be implemented simply as a shift left or shift right, and we keep
   track of a shifter value (0, 1, or 2) indicating how much to shift.
   Multiplying by 3 can be implemented by doubling and then adding the
   original value.  Dividing by 3, alas, cannot be implemented in any
   simple shift/subtract method, as far as I know; so we just do a table
   lookup.  For simplicity, we use a table of size 128K, which indexes the
   "divide-by-3" values for the first 64K non-negative numbers. (Note that
   we can increase the size up to 384K, i.e. indexing the first 192K
   non-negative numbers, while still using shorts in the array.) This also
   means that the size of the known region can be at most 64K for
   width-three characters.

   !!#### We should investigate the algorithm in GNU Emacs.  I think it
   does something similar, but it may differ in some details, and it's
   worth seeing if anything can be gleaned.
   */

extern short three_to_one_table[];

INLINE_HEADER Bytebpos real_charbpos_to_bytebpos (struct buffer *buf, Charbpos x);
INLINE_HEADER Bytebpos
real_charbpos_to_bytebpos (struct buffer *buf, Charbpos x)
{
  if (buf->text->entirely_ascii_p)
    return (Bytebpos) x;
  if (x >= buf->text->mule_bufmin && x <= buf->text->mule_bufmax)
    return (buf->text->mule_bytmin +
	    ((x - buf->text->mule_bufmin) << buf->text->mule_shifter) +
	    (buf->text->mule_three_p ? (x - buf->text->mule_bufmin) : 0));
  else
    return charbpos_to_bytebpos_func (buf, x);
}

INLINE_HEADER Charbpos real_bytebpos_to_charbpos (struct buffer *buf, Bytebpos x);
INLINE_HEADER Charbpos
real_bytebpos_to_charbpos (struct buffer *buf, Bytebpos x)
{
  if (buf->text->entirely_ascii_p)
    return (Charbpos) x;
  if (x >= buf->text->mule_bytmin && x <= buf->text->mule_bytmax)
    return (buf->text->mule_bufmin +
	    ((buf->text->mule_three_p
	      ? three_to_one_table[x - buf->text->mule_bytmin]
	      : (x - buf->text->mule_bytmin) >> buf->text->mule_shifter)));
  else
    return bytebpos_to_charbpos_func (buf, x);
}

#else /* not MULE */

# define real_charbpos_to_bytebpos(buf, x)	((Bytebpos) x)
# define real_bytebpos_to_charbpos(buf, x)	((Charbpos) x)

#endif /* not MULE */

#ifdef ERROR_CHECK_TEXT

Bytebpos charbpos_to_bytebpos (struct buffer *buf, Charbpos x);
Charbpos bytebpos_to_charbpos (struct buffer *buf, Bytebpos x);

#else /* not ERROR_CHECK_TEXT */

#define charbpos_to_bytebpos real_charbpos_to_bytebpos
#define bytebpos_to_charbpos real_bytebpos_to_charbpos

#endif /* not ERROR_CHECK_TEXT */

#define make_charbpos(buf, ind) make_int (bytebpos_to_charbpos (buf, ind))

/*----------------------------------------------------------------------*/
/*         Converting between buffer bytes and Emacs characters         */
/*----------------------------------------------------------------------*/

/* The character at position POS in buffer. */
#define BI_BUF_FETCH_CHAR(buf, pos) \
  charptr_emchar (BI_BUF_BYTE_ADDRESS (buf, pos))
#define BUF_FETCH_CHAR(buf, pos) \
  BI_BUF_FETCH_CHAR (buf, charbpos_to_bytebpos (buf, pos))

/* The character at position POS in buffer, as a string.  This is
   equivalent to set_charptr_emchar (str, BUF_FETCH_CHAR (buf, pos))
   but is faster for Mule. */

# define BI_BUF_CHARPTR_COPY_CHAR(buf, pos, str) \
  charptr_copy_char (BI_BUF_BYTE_ADDRESS (buf, pos), str)
#define BUF_CHARPTR_COPY_CHAR(buf, pos, str) \
  BI_BUF_CHARPTR_COPY_CHAR (buf, charbpos_to_bytebpos (buf, pos), str)


/************************************************************************/
/*                                                                      */
/*                  higher-level buffer-position functions              */
/*                                                                      */
/************************************************************************/

/*----------------------------------------------------------------------*/
/*           Settor macros for important positions in a buffer          */
/*----------------------------------------------------------------------*/

/* Set beginning of accessible range of buffer.  */
#define SET_BOTH_BUF_BEGV(buf, val, bival)	\
do						\
{						\
  (buf)->begv = (bival);			\
  (buf)->bufbegv = (val);			\
} while (0)

/* Set end of accessible range of buffer.  */
#define SET_BOTH_BUF_ZV(buf, val, bival)	\
do						\
{						\
  (buf)->zv = (bival);				\
  (buf)->bufzv = (val);				\
} while (0)

/* Set point. */
/* Since BEGV and ZV are almost never set, it's reasonable to enforce
   the restriction that the Charbpos and Bytebpos values must both be
   specified.  However, point is set in lots and lots of places.  So
   we provide the ability to specify both (for efficiency) or just
   one. */
#define BOTH_BUF_SET_PT(buf, val, bival) set_buffer_point (buf, val, bival)
#define BI_BUF_SET_PT(buf, bival) \
  BOTH_BUF_SET_PT (buf, bytebpos_to_charbpos (buf, bival), bival)
#define BUF_SET_PT(buf, value) \
  BOTH_BUF_SET_PT (buf, value, charbpos_to_bytebpos (buf, value))


#if 0 /* FSFmacs */
/* These macros exist in FSFmacs because SET_PT() in FSFmacs incorrectly
   does too much stuff, such as moving out of invisible extents. */
#define TEMP_SET_PT(position) (temp_set_point ((position), current_buffer))
#define SET_BUF_PT(buf, value) ((buf)->pt = (value))
#endif /* FSFmacs */

/*----------------------------------------------------------------------*/
/*                      Miscellaneous buffer values                     */
/*----------------------------------------------------------------------*/

/* Number of characters in buffer */
#define BUF_SIZE(buf) (BUF_Z (buf) - BUF_BEG (buf))

/* Is this buffer narrowed? */
#define BUF_NARROWED(buf) \
   ((BI_BUF_BEGV (buf) != BI_BUF_BEG (buf)) || \
    (BI_BUF_ZV   (buf) != BI_BUF_Z   (buf)))

/* Modification count.  */
#define BUF_MODIFF(buf) ((buf)->text->modiff)

/* Saved modification count.  */
#define BUF_SAVE_MODIFF(buf) ((buf)->text->save_modiff)

/* Face changed.  */
#define BUF_FACECHANGE(buf) ((buf)->face_change)

#define POINT_MARKER_P(marker) \
   (XMARKER (marker)->buffer != 0 && \
    EQ (marker, XMARKER (marker)->buffer->point_marker))

#define BUF_MARKERS(buf) ((buf)->markers)

/* WARNING:

   The new definitions of CEILING_OF() and FLOOR_OF() differ semantically
   from the old ones (in FSF Emacs and XEmacs 19.11 and before).
   Conversion is as follows:

   OLD_BI_CEILING_OF(n) = NEW_BI_CEILING_OF(n) - 1
   OLD_BI_FLOOR_OF(n) = NEW_BI_FLOOR_OF(n + 1)

   The definitions were changed because the new definitions are more
   consistent with the way everything else works in XEmacs.
 */

/* Properties of CEILING_OF and FLOOR_OF (also apply to BI_ variants):

   1) FLOOR_OF (CEILING_OF (n)) = n
      CEILING_OF (FLOOR_OF (n)) = n

   2) CEILING_OF (n) = n if and only if n = ZV
      FLOOR_OF (n) = n if and only if n = BEGV

   3) CEILING_OF (CEILING_OF (n)) = ZV
      FLOOR_OF (FLOOR_OF (n)) = BEGV

   4) The bytes in the regions

      [BYTE_ADDRESS (n), BYTE_ADDRESS_BEFORE (CEILING_OF (n))]

      and

      [BYTE_ADDRESS (FLOOR_OF (n)), BYTE_ADDRESS_BEFORE (n)]

      are contiguous.

  A typical loop using CEILING_OF to process contiguous ranges of text
  between [from, to) looks like this:
  
  {
    Bytebpos pos = from;

    while (pos < to)
      {
	Bytebpos ceil;

	ceil = BI_BUF_CEILING_OF (buf, pos);
	ceil = min (to, ceil);
	process_intbyte_string (BI_BUF_BYTE_ADDRESS (buf, pos), ceil - pos);
	pos = ceil;
      }
  }

  Currently there will be at most two iterations in the loop, but it is
  written in such a way that it will still work if the buffer
  representation is changed to have multiple gaps in it.
  */


/*  Return the maximum index in the buffer it is safe to scan forwards
    past N to.  This is used to prevent buffer scans from running into
    the gap (e.g. search.c).  All characters between N and CEILING_OF(N)
    are located contiguous in memory.  Note that the character *at*
    CEILING_OF(N) is not contiguous in memory. */
#define BI_BUF_CEILING_OF(b, n)						\
  ((n) < (b)->text->gpt && (b)->text->gpt < BI_BUF_ZV (b) ?		\
   (b)->text->gpt : BI_BUF_ZV (b))
#define BUF_CEILING_OF(b, n)						\
  bytebpos_to_charbpos (b, BI_BUF_CEILING_OF (b, charbpos_to_bytebpos (b, n)))

/*  Return the minimum index in the buffer it is safe to scan backwards
    past N to.  All characters between FLOOR_OF(N) and N are located
    contiguous in memory.  Note that the character *at* N may not be
    contiguous in memory. */
#define BI_BUF_FLOOR_OF(b, n)						\
        (BI_BUF_BEGV (b) < (b)->text->gpt && (b)->text->gpt < (n) ?	\
	 (b)->text->gpt : BI_BUF_BEGV (b))
#define BUF_FLOOR_OF(b, n)						\
  bytebpos_to_charbpos (b, BI_BUF_FLOOR_OF (b, charbpos_to_bytebpos (b, n)))

#define BI_BUF_CEILING_OF_IGNORE_ACCESSIBLE(b, n)			\
  ((n) < (b)->text->gpt && (b)->text->gpt < BI_BUF_Z (b) ?		\
   (b)->text->gpt : BI_BUF_Z (b))
#define BUF_CEILING_OF_IGNORE_ACCESSIBLE(b, n) 				\
  bytebpos_to_charbpos							\
   (b, BI_BUF_CEILING_OF_IGNORE_ACCESSIBLE (b, charbpos_to_bytebpos (b, n)))

#define BI_BUF_FLOOR_OF_IGNORE_ACCESSIBLE(b, n)				\
        (BI_BUF_BEG (b) < (b)->text->gpt && (b)->text->gpt < (n) ?	\
	 (b)->text->gpt : BI_BUF_BEG (b))
#define BUF_FLOOR_OF_IGNORE_ACCESSIBLE(b, n) 				\
  bytebpos_to_charbpos							\
   (b, BI_BUF_FLOOR_OF_IGNORE_ACCESSIBLE (b, charbpos_to_bytebpos (b, n)))

/* This structure marks which slots in a buffer have corresponding
   default values in Vbuffer_defaults.
   Each such slot has a nonzero value in this structure.
   The value has only one nonzero bit.

   When a buffer has its own local value for a slot,
   the bit for that slot (found in the same slot in this structure)
   is turned on in the buffer's local_var_flags slot.

   If a slot in this structure is zero, then even though there may
   be a DEFVAR_BUFFER_LOCAL for the slot, there is no default value for it;
   and the corresponding slot in Vbuffer_defaults is not used.  */

extern struct buffer buffer_local_flags;


/* Allocation of buffer data. */

#ifdef REL_ALLOC

char *r_alloc (unsigned char **, size_t);
char *r_re_alloc (unsigned char **, size_t);
void r_alloc_free (unsigned char **);

#define BUFFER_ALLOC(data, size) \
  ((Intbyte *) r_alloc ((unsigned char **) &data, (size) * sizeof(Intbyte)))
#define BUFFER_REALLOC(data, size) \
  ((Intbyte *) r_re_alloc ((unsigned char **) &data, (size) * sizeof(Intbyte)))
#define BUFFER_FREE(data) r_alloc_free ((unsigned char **) &(data))
#define R_ALLOC_DECLARE(var,data) r_alloc_declare (&(var), data)

#else /* !REL_ALLOC */

#define BUFFER_ALLOC(data,size)\
	(data = xnew_array (Intbyte, size))
#define BUFFER_REALLOC(data,size)\
	((Intbyte *) xrealloc (data, (size) * sizeof(Intbyte)))
/* Avoid excess parentheses, or syntax errors may rear their heads. */
#define BUFFER_FREE(data) xfree (data)
#define R_ALLOC_DECLARE(var,data)

#endif /* !REL_ALLOC */


/************************************************************************/
/*                         Case conversion                              */
/************************************************************************/

/* A "trt" table is a mapping from characters to other characters,
   typically used to convert between uppercase and lowercase.  For
   compatibility reasons, trt tables are currently in the form of
   a Lisp string of 256 characters, specifying the conversion for each
   of the first 256 Emacs characters (i.e. the 256 Latin-1 characters).
   This should be generalized at some point to support conversions for
   all of the allowable Mule characters.
   */

/* The _1 macros are named as such because they assume that you have
   already guaranteed that the character values are all in the range
   0 - 255.  Bad lossage will happen otherwise. */

#define MAKE_TRT_TABLE() Fmake_char_table (Qgeneric)
INLINE_HEADER Emchar TRT_TABLE_CHAR_1 (Lisp_Object table, Emchar c);
INLINE_HEADER Emchar
TRT_TABLE_CHAR_1 (Lisp_Object table, Emchar ch)
{
  Lisp_Object TRT_char;
  TRT_char = get_char_table (ch, XCHAR_TABLE (table));
  if (NILP (TRT_char))
    return ch;
  else
    return XCHAR (TRT_char);
}
#define SET_TRT_TABLE_CHAR_1(table, ch1, ch2)	\
  Fput_char_table (make_char (ch1), make_char (ch2), table);

INLINE_HEADER Emchar TRT_TABLE_OF (Lisp_Object trt, Emchar c);
INLINE_HEADER Emchar
TRT_TABLE_OF (Lisp_Object trt, Emchar c)
{
  return TRT_TABLE_CHAR_1 (trt, c);
}

INLINE_HEADER Lisp_Object BUFFER_CASE_TABLE (struct buffer *buf);
INLINE_HEADER Lisp_Object
BUFFER_CASE_TABLE (struct buffer *buf)
{
  return buf ? buf->case_table : Vstandard_case_table;
}

/* Macros used below. */
#define DOWNCASE_TABLE_OF(buf, c)	\
  TRT_TABLE_OF (XCASE_TABLE_DOWNCASE (BUFFER_CASE_TABLE (buf)), c)
#define UPCASE_TABLE_OF(buf, c)		\
  TRT_TABLE_OF (XCASE_TABLE_UPCASE (BUFFER_CASE_TABLE (buf)), c)

/* 1 if CH is upper case.  */

INLINE_HEADER int UPPERCASEP (struct buffer *buf, Emchar ch);
INLINE_HEADER int
UPPERCASEP (struct buffer *buf, Emchar ch)
{
  return DOWNCASE_TABLE_OF (buf, ch) != ch;
}

/* 1 if CH is lower case.  */

INLINE_HEADER int LOWERCASEP (struct buffer *buf, Emchar ch);
INLINE_HEADER int
LOWERCASEP (struct buffer *buf, Emchar ch)
{
  return (UPCASE_TABLE_OF   (buf, ch) != ch &&
	  DOWNCASE_TABLE_OF (buf, ch) == ch);
}

/* 1 if CH is neither upper nor lower case.  */

INLINE_HEADER int NOCASEP (struct buffer *buf, Emchar ch);
INLINE_HEADER int
NOCASEP (struct buffer *buf, Emchar ch)
{
  return UPCASE_TABLE_OF (buf, ch) == ch;
}

/* Upcase a character, or make no change if that cannot be done.  */

INLINE_HEADER Emchar UPCASE (struct buffer *buf, Emchar ch);
INLINE_HEADER Emchar
UPCASE (struct buffer *buf, Emchar ch)
{
  return (DOWNCASE_TABLE_OF (buf, ch) == ch) ? UPCASE_TABLE_OF (buf, ch) : ch;
}

/* Upcase a character known to be not upper case.  Unused. */

#define UPCASE1(buf, ch) UPCASE_TABLE_OF (buf, ch)

/* Downcase a character, or make no change if that cannot be done. */

#define DOWNCASE(buf, ch) DOWNCASE_TABLE_OF (buf, ch)

#endif /* INCLUDED_buffer_h_ */