1 /* Header file for the buffer manipulation primitives.
2 Copyright (C) 1985, 1986, 1992, 1993, 1994, 1995
3 Free Software Foundation, Inc.
4 Copyright (C) 1995 Sun Microsystems, Inc.
6 This file is part of XEmacs.
8 XEmacs is free software; you can redistribute it and/or modify it
9 under the terms of the GNU General Public License as published by the
10 Free Software Foundation; either version 2, or (at your option) any
13 XEmacs is distributed in the hope that it will be useful, but WITHOUT
14 ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
15 FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
18 You should have received a copy of the GNU General Public License
19 along with XEmacs; see the file COPYING. If not, write to
20 the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
21 Boston, MA 02111-1307, USA. */
23 /* Synched up with: FSF 19.30. */
28 JWZ: separated out bufslots.h, early in Lemacs.
29 Ben Wing: almost completely rewritten for Mule, 19.12.
32 #ifndef _XEMACS_BUFFER_H_
33 #define _XEMACS_BUFFER_H_
35 #include "character.h"
41 /************************************************************************/
43 /* definition of Lisp buffer object */
45 /************************************************************************/
47 /* Note: we keep both Bytind and Bufpos versions of some of the
48 important buffer positions because they are accessed so much.
49 If we didn't do this, we would constantly be invalidating the
50 bufpos<->bytind cache under Mule.
52 Note that under non-Mule, both versions will always be the
53 same so we don't really need to keep track of them. But it
54 simplifies the logic to go ahead and do so all the time and
55 the memory loss is insignificant. */
57 /* Formerly, it didn't much matter what went inside the struct buffer_text
58 and what went outside it. Now it does, with the advent of "indirect
59 buffers" that share text with another buffer. An indirect buffer
60 shares the same *text* as another buffer, but has its own buffer-local
61 variables, its own accessible region, and its own markers and extents.
62 (Due to the nature of markers, it doesn't actually matter much whether
63 we stick them inside or out of the struct buffer_text -- the user won't
64 notice any difference -- but we go ahead and put them outside for
65 consistency and overall saneness of algorithm.)
67 FSFmacs gets away with not maintaining any "children" pointers from
68 a buffer to the indirect buffers that refer to it by putting the
69 markers inside of the struct buffer_text, using markers to keep track
70 of BEGV and ZV in indirect buffers, and relying on the fact that
71 all intervals (text properties and overlays) use markers for their
72 start and end points. We don't do this for extents (markers are
73 inefficient anyway and take up space), so we have to maintain
74 children pointers. This is not terribly hard, though, and the
75 code to maintain this is just like the code already present in
76 extent-parent and extent-children.
81 Bufbyte *beg; /* Actual address of buffer contents. */
82 Bytind gpt; /* Index of gap in buffer. */
83 Bytind z; /* Index of end of buffer. */
84 Bufpos bufz; /* Equivalent as a Bufpos. */
85 int gap_size; /* Size of buffer's gap */
86 int end_gap_size; /* Size of buffer's end gap */
87 long modiff; /* This counts buffer-modification events
88 for this buffer. It is incremented for
89 each such event, and never otherwise
91 long save_modiff; /* Previous value of modiff, as of last
92 time buffer visited or saved a file. */
95 /* We keep track of a "known" region for very fast access.
96 This information is text-only so it goes here. */
97 Bufpos mule_bufmin, mule_bufmax;
98 Bytind mule_bytmin, mule_bytmax;
102 int mule_shifter, mule_three_p;
105 /* And we also cache 16 positions for fairly fast access near those
107 Bufpos mule_bufpos_cache[16];
108 Bytind mule_bytind_cache[16];
111 /* Similar to the above, we keep track of positions for which line
112 number has last been calculated. See line-number.c. */
113 Lisp_Object line_number_cache;
115 /* Change data that goes with the text. */
116 struct buffer_text_change_data *changes;
122 struct lcrecord_header header;
124 /* This structure holds the coordinates of the buffer contents
125 in ordinary buffers. In indirect buffers, this is not used. */
126 struct buffer_text own_text;
128 /* This points to the `struct buffer_text' that is used for this buffer.
129 In an ordinary buffer, this is the own_text field above.
130 In an indirect buffer, this is the own_text field of another buffer. */
131 struct buffer_text *text;
133 Bytind pt; /* Position of point in buffer. */
134 Bufpos bufpt; /* Equivalent as a Bufpos. */
135 Bytind begv; /* Index of beginning of accessible range. */
136 Bufpos bufbegv; /* Equivalent as a Bufpos. */
137 Bytind zv; /* Index of end of accessible range. */
138 Bufpos bufzv; /* Equivalent as a Bufpos. */
140 int face_change; /* This is set when a change in how the text should
141 be displayed (e.g., font, color) is made. */
143 /* change data indicating what portion of the text has changed
144 since the last time this was reset. Used by redisplay.
145 Logically we should keep this with the text structure, but
146 redisplay resets it for each buffer individually and we don't
147 want interference between an indirect buffer and its base
149 struct each_buffer_change_data *changes;
151 #ifdef REGION_CACHE_NEEDS_WORK
152 /* If the long line scan cache is enabled (i.e. the buffer-local
153 variable cache-long-line-scans is non-nil), newline_cache
154 points to the newline cache, and width_run_cache points to the
157 The newline cache records which stretches of the buffer are
158 known *not* to contain newlines, so that they can be skipped
159 quickly when we search for newlines.
161 The width run cache records which stretches of the buffer are
162 known to contain characters whose widths are all the same. If
163 the width run cache maps a character to a value > 0, that value
164 is the character's width; if it maps a character to zero, we
165 don't know what its width is. This allows compute_motion to
166 process such regions very quickly, using algebra instead of
167 inspecting each character. See also width_table, below. */
168 struct region_cache *newline_cache;
169 struct region_cache *width_run_cache;
170 #endif /* REGION_CACHE_NEEDS_WORK */
172 /* The markers that refer to this buffer. This is actually a single
173 marker -- successive elements in its marker `chain' are the other
174 markers referring to this buffer */
175 struct Lisp_Marker *markers;
177 /* The buffer's extent info. This is its own type, an extent-info
178 object (done this way for ease in marking / finalizing). */
179 Lisp_Object extent_info;
181 /* ----------------------------------------------------------------- */
182 /* All the stuff above this line is the responsibility of insdel.c,
183 with some help from marker.c and extents.c.
184 All the stuff below this line is the responsibility of buffer.c. */
186 /* In an indirect buffer, this points to the base buffer.
187 In an ordinary buffer, it is 0.
188 We DO mark through this slot. */
189 struct buffer *base_buffer;
191 /* List of indirect buffers whose base is this buffer.
192 If we are an indirect buffer, this will be nil.
193 Do NOT mark through this. */
194 Lisp_Object indirect_children;
196 /* Flags saying which DEFVAR_PER_BUFFER variables
197 are local to this buffer. */
200 /* Set to the modtime of the visited file when read or written.
201 -1 means visited file was nonexistent.
202 0 means visited file modtime unknown; in no case complain
203 about any mismatch on next save attempt. */
206 /* the value of text->modiff at the last auto-save. */
207 int auto_save_modified;
209 /* The time at which we detected a failure to auto-save,
210 Or -1 if we didn't have a failure. */
211 int auto_save_failure_time;
213 /* Position in buffer at which display started
214 the last time this buffer was displayed. */
215 int last_window_start;
217 /* Everything from here down must be a Lisp_Object */
219 #define MARKED_SLOT(x) Lisp_Object x
220 #include "bufslots.h"
224 DECLARE_LRECORD (buffer, struct buffer);
225 #define XBUFFER(x) XRECORD (x, buffer, struct buffer)
226 #define XSETBUFFER(x, p) XSETRECORD (x, p, buffer)
227 #define BUFFERP(x) RECORDP (x, buffer)
228 #define GC_BUFFERP(x) GC_RECORDP (x, buffer)
229 #define CHECK_BUFFER(x) CHECK_RECORD (x, buffer)
230 #define CONCHECK_BUFFER(x) CONCHECK_RECORD (x, buffer)
232 #define BUFFER_LIVE_P(b) (!NILP ((b)->name))
234 #define CHECK_LIVE_BUFFER(x) do { \
236 if (!BUFFER_LIVE_P (XBUFFER (x))) \
237 dead_wrong_type_argument (Qbuffer_live_p, (x)); \
240 #define CONCHECK_LIVE_BUFFER(x) do { \
241 CONCHECK_BUFFER (x); \
242 if (!BUFFER_LIVE_P (XBUFFER (x))) \
243 x = wrong_type_argument (Qbuffer_live_p, (x)); \
247 #define BUFFER_BASE_BUFFER(b) ((b)->base_buffer ? (b)->base_buffer : (b))
249 /* Map over buffers sharing the same text as MPS_BUF. MPS_BUFVAR is a
250 variable that gets the buffer values (beginning with the base
251 buffer, then the children), and MPS_BUFCONS should be a temporary
252 Lisp_Object variable. */
253 #define MAP_INDIRECT_BUFFERS(mps_buf, mps_bufvar, mps_bufcons) \
254 for (mps_bufcons = Qunbound, \
255 mps_bufvar = BUFFER_BASE_BUFFER (mps_buf); \
256 UNBOUNDP (mps_bufcons) ? \
257 (mps_bufcons = mps_bufvar->indirect_children, \
259 : (!NILP (mps_bufcons) \
260 && (mps_bufvar = XBUFFER (XCAR (mps_bufcons)), 1) \
261 && (mps_bufcons = XCDR (mps_bufcons), 1)); \
266 /************************************************************************/
268 /* working with raw internal-format data */
270 /************************************************************************/
272 /* NOTE: In all the following macros, we follow these rules concerning
273 multiple evaluation of the arguments:
275 1) Anything that's an lvalue can be evaluated more than once.
276 2) Anything that's a Lisp Object can be evaluated more than once.
277 This should probably be changed, but this follows the way
278 that all the macros in lisp.h do things.
279 3) 'struct buffer *' arguments can be evaluated more than once.
280 4) Nothing else can be evaluated more than once. Use inline
281 functions, if necessary, to prevent multiple evaluation.
282 5) An exception to (4) is that there are some macros below that
283 may evaluate their arguments more than once. They are all
284 denoted with the word "unsafe" in their name and are generally
285 meant to be called only by other macros that have already
286 stored the calling values in temporary variables.
289 Use the following functions/macros on contiguous strings of data.
290 If the text you're operating on is known to come from a buffer, use
291 the buffer-level functions below -- they know about the gap and may
295 (A) For working with charptr's (pointers to internally-formatted text):
296 -----------------------------------------------------------------------
298 VALID_CHARPTR_P (ptr):
299 Given a charptr, does it point to the beginning of a character?
301 ASSERT_VALID_CHARPTR (ptr):
302 If error-checking is enabled, assert that the given charptr
303 points to the beginning of a character. Otherwise, do nothing.
306 Given a charptr (assumed to point at the beginning of a character),
307 modify that pointer so it points to the beginning of the next
311 Given a charptr (assumed to point at the beginning of a
312 character or at the very end of the text), modify that pointer
313 so it points to the beginning of the previous character.
315 VALIDATE_CHARPTR_BACKWARD (ptr):
316 Make sure that PTR is pointing to the beginning of a character.
317 If not, back up until this is the case. Note that there are not
318 too many places where it is legitimate to do this sort of thing.
319 It's an error if you're passed an "invalid" char * pointer.
320 NOTE: PTR *must* be pointing to a valid part of the string (i.e.
321 not the very end, unless the string is zero-terminated or
322 something) in order for this function to not cause crashes.
324 VALIDATE_CHARPTR_FORWARD (ptr):
325 Make sure that PTR is pointing to the beginning of a character.
326 If not, move forward until this is the case. Note that there
327 are not too many places where it is legitimate to do this sort
328 of thing. It's an error if you're passed an "invalid" char *
332 (B) For working with the length (in bytes and characters) of a
333 section of internally-formatted text:
334 --------------------------------------------------------------
336 bytecount_to_charcount (ptr, nbi):
337 Given a pointer to a text string and a length in bytes,
338 return the equivalent length in characters.
340 charcount_to_bytecount (ptr, nch):
341 Given a pointer to a text string and a length in characters,
342 return the equivalent length in bytes.
344 charptr_n_addr (ptr, n):
345 Return a pointer to the beginning of the character offset N
346 (in characters) from PTR.
349 (C) For retrieving or changing the character pointed to by a charptr:
350 ---------------------------------------------------------------------
352 charptr_emchar (ptr):
353 Retrieve the character pointed to by PTR as an Emchar.
355 charptr_emchar_n (ptr, n):
356 Retrieve the character at offset N (in characters) from PTR,
359 set_charptr_emchar (ptr, ch):
360 Store the character CH (an Emchar) as internally-formatted
361 text starting at PTR. Return the number of bytes stored.
363 charptr_copy_char (ptr, ptr2):
364 Retrieve the character pointed to by PTR and store it as
365 internally-formatted text in PTR2.
368 (D) For working with Emchars:
369 -----------------------------
371 [Note that there are other functions/macros for working with Emchars
372 in mule-charset.h, for retrieving the charset of an Emchar
373 and such. These are only valid when MULE is defined.]
376 Return whether the given Emchar is valid.
379 Return whether the given Lisp_Object is a character.
381 CHECK_CHAR_COERCE_INT (ch):
382 Signal an error if CH is not a valid character or integer Lisp_Object.
383 If CH is an integer Lisp_Object, convert it to a character Lisp_Object,
384 but merely by repackaging, without performing tests for char validity.
387 Maximum number of buffer bytes per Emacs character.
392 /* ---------------------------------------------------------------------- */
393 /* (A) For working with charptr's (pointers to internally-formatted text) */
394 /* ---------------------------------------------------------------------- */
397 # define VALID_CHARPTR_P(ptr) BUFBYTE_FIRST_BYTE_P (* (unsigned char *) ptr)
399 # define VALID_CHARPTR_P(ptr) 1
402 #ifdef ERROR_CHECK_BUFPOS
403 # define ASSERT_VALID_CHARPTR(ptr) assert (VALID_CHARPTR_P (ptr))
405 # define ASSERT_VALID_CHARPTR(ptr)
408 /* Note that INC_CHARPTR() and DEC_CHARPTR() have to be written in
409 completely separate ways. INC_CHARPTR() cannot use the DEC_CHARPTR()
410 trick of looking for a valid first byte because it might run off
411 the end of the string. DEC_CHARPTR() can't use the INC_CHARPTR()
412 method because it doesn't have easy access to the first byte of
413 the character it's moving over. */
415 #define REAL_INC_CHARPTR(ptr) \
416 ((void) ((ptr) += REP_BYTES_BY_FIRST_BYTE (* (unsigned char *) (ptr))))
418 #define REAL_INC_CHARBYTIND(ptr,pos) \
419 (pos += REP_BYTES_BY_FIRST_BYTE (* (unsigned char *) (ptr)))
421 #define REAL_DEC_CHARPTR(ptr) do { \
423 } while (!VALID_CHARPTR_P (ptr))
425 #ifdef ERROR_CHECK_BUFPOS
426 #define INC_CHARPTR(ptr) do { \
427 ASSERT_VALID_CHARPTR (ptr); \
428 REAL_INC_CHARPTR (ptr); \
431 #define INC_CHARBYTIND(ptr,pos) do { \
432 ASSERT_VALID_CHARPTR (ptr); \
433 REAL_INC_CHARBYTIND (ptr,pos); \
436 #define DEC_CHARPTR(ptr) do { \
437 CONST Bufbyte *dc_ptr1 = (ptr); \
438 CONST Bufbyte *dc_ptr2 = dc_ptr1; \
439 REAL_DEC_CHARPTR (dc_ptr2); \
440 assert (dc_ptr1 - dc_ptr2 == \
441 REP_BYTES_BY_FIRST_BYTE (*dc_ptr2)); \
445 #else /* ! ERROR_CHECK_BUFPOS */
446 #define INC_CHARBYTIND(ptr,pos) REAL_INC_CHARBYTIND (ptr,pos)
447 #define INC_CHARPTR(ptr) REAL_INC_CHARPTR (ptr)
448 #define DEC_CHARPTR(ptr) REAL_DEC_CHARPTR (ptr)
449 #endif /* ! ERROR_CHECK_BUFPOS */
453 #define VALIDATE_CHARPTR_BACKWARD(ptr) do { \
454 while (!VALID_CHARPTR_P (ptr)) ptr--; \
457 /* This needs to be trickier to avoid the possibility of running off
458 the end of the string. */
460 #define VALIDATE_CHARPTR_FORWARD(ptr) do { \
461 Bufbyte *vcf_ptr = (ptr); \
462 VALIDATE_CHARPTR_BACKWARD (vcf_ptr); \
463 if (vcf_ptr != (ptr)) \
471 #define VALIDATE_CHARPTR_BACKWARD(ptr)
472 #define VALIDATE_CHARPTR_FORWARD(ptr)
473 #endif /* not MULE */
475 /* -------------------------------------------------------------- */
476 /* (B) For working with the length (in bytes and characters) of a */
477 /* section of internally-formatted text */
478 /* -------------------------------------------------------------- */
480 INLINE CONST Bufbyte *charptr_n_addr (CONST Bufbyte *ptr, Charcount offset);
481 INLINE CONST Bufbyte *
482 charptr_n_addr (CONST Bufbyte *ptr, Charcount offset)
484 return ptr + charcount_to_bytecount (ptr, offset);
487 /* -------------------------------------------------------------------- */
488 /* (C) For retrieving or changing the character pointed to by a charptr */
489 /* -------------------------------------------------------------------- */
491 #define simple_charptr_emchar(ptr) ((Emchar) (ptr)[0])
492 #define simple_set_charptr_emchar(ptr, x) ((ptr)[0] = (Bufbyte) (x), 1)
493 #define simple_charptr_copy_char(ptr, ptr2) ((ptr2)[0] = *(ptr), 1)
497 Emchar non_ascii_charptr_emchar (CONST Bufbyte *ptr);
498 Bytecount non_ascii_set_charptr_emchar (Bufbyte *ptr, Emchar c);
499 Bytecount non_ascii_charptr_copy_char (CONST Bufbyte *ptr, Bufbyte *ptr2);
501 INLINE Emchar charptr_emchar (CONST Bufbyte *ptr);
503 charptr_emchar (CONST Bufbyte *ptr)
505 return BYTE_ASCII_P (*ptr) ?
506 simple_charptr_emchar (ptr) :
507 non_ascii_charptr_emchar (ptr);
510 INLINE Bytecount set_charptr_emchar (Bufbyte *ptr, Emchar x);
512 set_charptr_emchar (Bufbyte *ptr, Emchar x)
514 return !CHAR_MULTIBYTE_P (x) ?
515 simple_set_charptr_emchar (ptr, x) :
516 non_ascii_set_charptr_emchar (ptr, x);
519 INLINE Bytecount charptr_copy_char (CONST Bufbyte *ptr, Bufbyte *ptr2);
521 charptr_copy_char (CONST Bufbyte *ptr, Bufbyte *ptr2)
523 return BYTE_ASCII_P (*ptr) ?
524 simple_charptr_copy_char (ptr, ptr2) :
525 non_ascii_charptr_copy_char (ptr, ptr2);
530 # define charptr_emchar(ptr) simple_charptr_emchar (ptr)
531 # define set_charptr_emchar(ptr, x) simple_set_charptr_emchar (ptr, x)
532 # define charptr_copy_char(ptr, ptr2) simple_charptr_copy_char (ptr, ptr2)
534 #endif /* not MULE */
536 #define charptr_emchar_n(ptr, offset) \
537 charptr_emchar (charptr_n_addr (ptr, offset))
540 /* ---------------------------- */
541 /* (D) For working with Emchars */
542 /* ---------------------------- */
545 # define MAX_EMCHAR_LEN 6
548 # define MAX_EMCHAR_LEN 4
550 # define MAX_EMCHAR_LEN 1
555 /*----------------------------------------------------------------------*/
556 /* Accessor macros for important positions in a buffer */
557 /*----------------------------------------------------------------------*/
559 /* We put them here because some stuff below wants them before the
560 place where we would normally put them. */
562 /* None of these are lvalues. Use the settor macros below to change
565 /* Beginning of buffer. */
566 #define BI_BUF_BEG(buf) ((Bytind) 1)
567 #define BUF_BEG(buf) ((Bufpos) 1)
569 /* Beginning of accessible range of buffer. */
570 #define BI_BUF_BEGV(buf) ((buf)->begv + 0)
571 #define BUF_BEGV(buf) ((buf)->bufbegv + 0)
573 /* End of accessible range of buffer. */
574 #define BI_BUF_ZV(buf) ((buf)->zv + 0)
575 #define BUF_ZV(buf) ((buf)->bufzv + 0)
578 #define BI_BUF_Z(buf) ((buf)->text->z + 0)
579 #define BUF_Z(buf) ((buf)->text->bufz + 0)
582 #define BI_BUF_PT(buf) ((buf)->pt + 0)
583 #define BUF_PT(buf) ((buf)->bufpt + 0)
585 /*----------------------------------------------------------------------*/
586 /* Converting between positions and addresses */
587 /*----------------------------------------------------------------------*/
589 /* Convert the address of a byte in the buffer into a position. */
590 INLINE Bytind BI_BUF_PTR_BYTE_POS (struct buffer *buf, Bufbyte *ptr);
592 BI_BUF_PTR_BYTE_POS (struct buffer *buf, Bufbyte *ptr)
594 return ((ptr) - (buf)->text->beg + 1
595 - ((ptr - (buf)->text->beg + 1) > (buf)->text->gpt
596 ? (buf)->text->gap_size : 0));
599 #define BUF_PTR_BYTE_POS(buf, ptr) \
600 bytind_to_bufpos (buf, BI_BUF_PTR_BYTE_POS (buf, ptr))
602 /* Address of byte at position POS in buffer. */
603 INLINE Bufbyte * BI_BUF_BYTE_ADDRESS (struct buffer *buf, Bytind pos);
605 BI_BUF_BYTE_ADDRESS (struct buffer *buf, Bytind pos)
607 return ((buf)->text->beg +
608 ((pos >= (buf)->text->gpt ? (pos + (buf)->text->gap_size) : pos)
612 #define BUF_BYTE_ADDRESS(buf, pos) \
613 BI_BUF_BYTE_ADDRESS (buf, bufpos_to_bytind (buf, pos))
615 /* Address of byte before position POS in buffer. */
616 INLINE Bufbyte * BI_BUF_BYTE_ADDRESS_BEFORE (struct buffer *buf, Bytind pos);
618 BI_BUF_BYTE_ADDRESS_BEFORE (struct buffer *buf, Bytind pos)
620 return ((buf)->text->beg +
621 ((pos > (buf)->text->gpt ? (pos + (buf)->text->gap_size) : pos)
625 #define BUF_BYTE_ADDRESS_BEFORE(buf, pos) \
626 BI_BUF_BYTE_ADDRESS_BEFORE (buf, bufpos_to_bytind (buf, pos))
628 /*----------------------------------------------------------------------*/
629 /* Converting between byte indices and memory indices */
630 /*----------------------------------------------------------------------*/
632 INLINE int valid_memind_p (struct buffer *buf, Memind x);
634 valid_memind_p (struct buffer *buf, Memind x)
636 return ((x >= 1 && x <= (Memind) (buf)->text->gpt) ||
637 (x > (Memind) ((buf)->text->gpt + (buf)->text->gap_size) &&
638 x <= (Memind) ((buf)->text->z + (buf)->text->gap_size)));
641 INLINE Memind bytind_to_memind (struct buffer *buf, Bytind x);
643 bytind_to_memind (struct buffer *buf, Bytind x)
645 return (Memind) ((x > (buf)->text->gpt) ? (x + (buf)->text->gap_size) : x);
649 INLINE Bytind memind_to_bytind (struct buffer *buf, Memind x);
651 memind_to_bytind (struct buffer *buf, Memind x)
653 #ifdef ERROR_CHECK_BUFPOS
654 assert (valid_memind_p (buf, x));
656 return (Bytind) ((x > (Memind) (buf)->text->gpt) ?
657 x - (buf)->text->gap_size :
661 #define memind_to_bufpos(buf, x) \
662 bytind_to_bufpos (buf, memind_to_bytind (buf, x))
663 #define bufpos_to_memind(buf, x) \
664 bytind_to_memind (buf, bufpos_to_bytind (buf, x))
666 /* These macros generalize many standard buffer-position functions to
667 either a buffer or a string. */
669 /* Converting between Meminds and Bytinds, for a buffer-or-string.
670 For strings, this is a no-op. For buffers, this resolves
671 to the standard memind<->bytind converters. */
673 #define buffer_or_string_bytind_to_memind(obj, ind) \
674 (BUFFERP (obj) ? bytind_to_memind (XBUFFER (obj), ind) : (Memind) ind)
676 #define buffer_or_string_memind_to_bytind(obj, ind) \
677 (BUFFERP (obj) ? memind_to_bytind (XBUFFER (obj), ind) : (Bytind) ind)
679 /* Converting between Bufpos's and Bytinds, for a buffer-or-string.
680 For strings, this maps to the bytecount<->charcount converters. */
682 #define buffer_or_string_bufpos_to_bytind(obj, pos) \
683 (BUFFERP (obj) ? bufpos_to_bytind (XBUFFER (obj), pos) : \
684 (Bytind) charcount_to_bytecount (XSTRING_DATA (obj), pos))
686 #define buffer_or_string_bytind_to_bufpos(obj, ind) \
687 (BUFFERP (obj) ? bytind_to_bufpos (XBUFFER (obj), ind) : \
688 (Bufpos) bytecount_to_charcount (XSTRING_DATA (obj), ind))
690 /* Similar for Bufpos's and Meminds. */
692 #define buffer_or_string_bufpos_to_memind(obj, pos) \
693 (BUFFERP (obj) ? bufpos_to_memind (XBUFFER (obj), pos) : \
694 (Memind) charcount_to_bytecount (XSTRING_DATA (obj), pos))
696 #define buffer_or_string_memind_to_bufpos(obj, ind) \
697 (BUFFERP (obj) ? memind_to_bufpos (XBUFFER (obj), ind) : \
698 (Bufpos) bytecount_to_charcount (XSTRING_DATA (obj), ind))
700 /************************************************************************/
702 /* working with buffer-level data */
704 /************************************************************************/
708 (A) Working with byte indices:
709 ------------------------------
711 VALID_BYTIND_P(buf, bi):
712 Given a byte index, does it point to the beginning of a character?
714 ASSERT_VALID_BYTIND_UNSAFE(buf, bi):
715 If error-checking is enabled, assert that the given byte index
716 is within range and points to the beginning of a character
717 or to the end of the buffer. Otherwise, do nothing.
719 ASSERT_VALID_BYTIND_BACKWARD_UNSAFE(buf, bi):
720 If error-checking is enabled, assert that the given byte index
721 is within range and satisfies ASSERT_VALID_BYTIND() and also
722 does not refer to the beginning of the buffer. (i.e. movement
723 backwards is OK.) Otherwise, do nothing.
725 ASSERT_VALID_BYTIND_FORWARD_UNSAFE(buf, bi):
726 If error-checking is enabled, assert that the given byte index
727 is within range and satisfies ASSERT_VALID_BYTIND() and also
728 does not refer to the end of the buffer. (i.e. movement
729 forwards is OK.) Otherwise, do nothing.
731 VALIDATE_BYTIND_BACKWARD(buf, bi):
732 Make sure that the given byte index is pointing to the beginning
733 of a character. If not, back up until this is the case. Note
734 that there are not too many places where it is legitimate to do
735 this sort of thing. It's an error if you're passed an "invalid"
738 VALIDATE_BYTIND_FORWARD(buf, bi):
739 Make sure that the given byte index is pointing to the beginning
740 of a character. If not, move forward until this is the case.
741 Note that there are not too many places where it is legitimate
742 to do this sort of thing. It's an error if you're passed an
743 "invalid" byte index.
746 Given a byte index (assumed to point at the beginning of a
747 character), modify that value so it points to the beginning
748 of the next character.
751 Given a byte index (assumed to point at the beginning of a
752 character), modify that value so it points to the beginning
753 of the previous character. Unlike for DEC_CHARPTR(), we can
754 do all the assert()s because there are sentinels at the
755 beginning of the gap and the end of the buffer.
758 A constant representing an invalid Bytind. Valid Bytinds
759 can never have this value.
762 (B) Converting between Bufpos's and Bytinds:
763 --------------------------------------------
765 bufpos_to_bytind(buf, bu):
766 Given a Bufpos, return the equivalent Bytind.
768 bytind_to_bufpos(buf, bi):
769 Given a Bytind, return the equivalent Bufpos.
771 make_bufpos(buf, bi):
772 Given a Bytind, return the equivalent Bufpos as a Lisp Object.
776 /*----------------------------------------------------------------------*/
777 /* working with byte indices */
778 /*----------------------------------------------------------------------*/
781 # define VALID_BYTIND_P(buf, x) \
782 BUFBYTE_FIRST_BYTE_P (*BI_BUF_BYTE_ADDRESS (buf, x))
784 # define VALID_BYTIND_P(buf, x) 1
787 #ifdef ERROR_CHECK_BUFPOS
789 # define ASSERT_VALID_BYTIND_UNSAFE(buf, x) do { \
790 assert (BUFFER_LIVE_P (buf)); \
791 assert ((x) >= BI_BUF_BEG (buf) && x <= BI_BUF_Z (buf)); \
792 assert (VALID_BYTIND_P (buf, x)); \
794 # define ASSERT_VALID_BYTIND_BACKWARD_UNSAFE(buf, x) do { \
795 assert (BUFFER_LIVE_P (buf)); \
796 assert ((x) > BI_BUF_BEG (buf) && x <= BI_BUF_Z (buf)); \
797 assert (VALID_BYTIND_P (buf, x)); \
799 # define ASSERT_VALID_BYTIND_FORWARD_UNSAFE(buf, x) do { \
800 assert (BUFFER_LIVE_P (buf)); \
801 assert ((x) >= BI_BUF_BEG (buf) && x < BI_BUF_Z (buf)); \
802 assert (VALID_BYTIND_P (buf, x)); \
805 #else /* not ERROR_CHECK_BUFPOS */
806 # define ASSERT_VALID_BYTIND_UNSAFE(buf, x)
807 # define ASSERT_VALID_BYTIND_BACKWARD_UNSAFE(buf, x)
808 # define ASSERT_VALID_BYTIND_FORWARD_UNSAFE(buf, x)
810 #endif /* not ERROR_CHECK_BUFPOS */
812 /* Note that, although the Mule version will work fine for non-Mule
813 as well (it should reduce down to nothing), we provide a separate
814 version to avoid compilation warnings and possible non-optimal
815 results with stupid compilers. */
818 # define VALIDATE_BYTIND_BACKWARD(buf, x) do { \
819 Bufbyte *VBB_ptr = BI_BUF_BYTE_ADDRESS (buf, x); \
820 while (!BUFBYTE_FIRST_BYTE_P (*VBB_ptr)) \
824 # define VALIDATE_BYTIND_BACKWARD(buf, x)
827 /* Note that, although the Mule version will work fine for non-Mule
828 as well (it should reduce down to nothing), we provide a separate
829 version to avoid compilation warnings and possible non-optimal
830 results with stupid compilers. */
833 # define VALIDATE_BYTIND_FORWARD(buf, x) do { \
834 Bufbyte *VBF_ptr = BI_BUF_BYTE_ADDRESS (buf, x); \
835 while (!BUFBYTE_FIRST_BYTE_P (*VBF_ptr)) \
839 # define VALIDATE_BYTIND_FORWARD(buf, x)
842 /* Note that in the simplest case (no MULE, no ERROR_CHECK_BUFPOS),
843 this crap reduces down to simply (x)++. */
845 #define INC_BYTIND(buf, x) do \
847 ASSERT_VALID_BYTIND_FORWARD_UNSAFE (buf, x); \
848 /* Note that we do the increment first to \
849 make sure that the pointer in \
850 VALIDATE_BYTIND_FORWARD() ends up on \
851 the correct side of the gap */ \
853 VALIDATE_BYTIND_FORWARD (buf, x); \
856 /* Note that in the simplest case (no MULE, no ERROR_CHECK_BUFPOS),
857 this crap reduces down to simply (x)--. */
859 #define DEC_BYTIND(buf, x) do \
861 ASSERT_VALID_BYTIND_BACKWARD_UNSAFE (buf, x); \
862 /* Note that we do the decrement first to \
863 make sure that the pointer in \
864 VALIDATE_BYTIND_BACKWARD() ends up on \
865 the correct side of the gap */ \
867 VALIDATE_BYTIND_BACKWARD (buf, x); \
870 INLINE Bytind prev_bytind (struct buffer *buf, Bytind x);
872 prev_bytind (struct buffer *buf, Bytind x)
878 INLINE Bytind next_bytind (struct buffer *buf, Bytind x);
880 next_bytind (struct buffer *buf, Bytind x)
886 #define BYTIND_INVALID ((Bytind) -1)
888 /*----------------------------------------------------------------------*/
889 /* Converting between buffer positions and byte indices */
890 /*----------------------------------------------------------------------*/
894 Bytind bufpos_to_bytind_func (struct buffer *buf, Bufpos x);
895 Bufpos bytind_to_bufpos_func (struct buffer *buf, Bytind x);
897 /* The basic algorithm we use is to keep track of a known region of
898 characters in each buffer, all of which are of the same width. We
899 keep track of the boundaries of the region in both Bufpos and
900 Bytind coordinates and also keep track of the char width, which
901 is 1 - 4 bytes. If the position we're translating is not in
902 the known region, then we invoke a function to update the known
903 region to surround the position in question. This assumes
904 locality of reference, which is usually the case.
906 Note that the function to update the known region can be simple
907 or complicated depending on how much information we cache.
908 For the moment, we don't cache any information, and just move
909 linearly forward or back from the known region, with a few
910 shortcuts to catch all-ASCII buffers. (Note that this will
911 thrash with bad locality of reference.) A smarter method would
912 be to keep some sort of pseudo-extent layer over the buffer;
913 maybe keep track of the bufpos/bytind correspondence at the
914 beginning of each line, which would allow us to do a binary
915 search over the pseudo-extents to narrow things down to the
916 correct line, at which point you could use a linear movement
917 method. This would also mesh well with efficiently
918 implementing a line-numbering scheme.
920 Note also that we have to multiply or divide by the char width
921 in order to convert the positions. We do some tricks to avoid
922 ever actually having to do a multiply or divide, because that
923 is typically an expensive operation (esp. divide). Multiplying
924 or dividing by 1, 2, or 4 can be implemented simply as a
925 shift left or shift right, and we keep track of a shifter value
926 (0, 1, or 2) indicating how much to shift. Multiplying by 3
927 can be implemented by doubling and then adding the original
928 value. Dividing by 3, alas, cannot be implemented in any
929 simple shift/subtract method, as far as I know; so we just
930 do a table lookup. For simplicity, we use a table of size
931 128K, which indexes the "divide-by-3" values for the first
932 64K non-negative numbers. (Note that we can increase the
933 size up to 384K, i.e. indexing the first 192K non-negative
934 numbers, while still using shorts in the array.) This also
935 means that the size of the known region can be at most
936 64K for width-three characters.
940 extern short three_to_one_table[];
943 INLINE int real_bufpos_to_bytind (struct buffer *buf, Bufpos x);
945 real_bufpos_to_bytind (struct buffer *buf, Bufpos x)
947 if (x >= buf->text->mule_bufmin && x <= buf->text->mule_bufmax)
948 return (buf->text->mule_bytmin +
950 (x - buf->text->mule_bufmin) * buf->text->mule_size
952 ((x - buf->text->mule_bufmin) << buf->text->mule_shifter) +
953 (buf->text->mule_three_p ? (x - buf->text->mule_bufmin) : 0)
957 return bufpos_to_bytind_func (buf, x);
960 INLINE int real_bytind_to_bufpos (struct buffer *buf, Bytind x);
962 real_bytind_to_bufpos (struct buffer *buf, Bytind x)
964 if (x >= buf->text->mule_bytmin && x <= buf->text->mule_bytmax)
965 return (buf->text->mule_bufmin +
967 (buf->text->mule_size == 0 ? 0 :
968 (x - buf->text->mule_bytmin) / buf->text->mule_size)
970 ((buf->text->mule_three_p
971 ? three_to_one_table[x - buf->text->mule_bytmin]
972 : (x - buf->text->mule_bytmin) >> buf->text->mule_shifter))
976 return bytind_to_bufpos_func (buf, x);
981 # define real_bufpos_to_bytind(buf, x) ((Bytind) x)
982 # define real_bytind_to_bufpos(buf, x) ((Bufpos) x)
984 #endif /* not MULE */
986 #ifdef ERROR_CHECK_BUFPOS
988 Bytind bufpos_to_bytind (struct buffer *buf, Bufpos x);
989 Bufpos bytind_to_bufpos (struct buffer *buf, Bytind x);
991 #else /* not ERROR_CHECK_BUFPOS */
993 #define bufpos_to_bytind real_bufpos_to_bytind
994 #define bytind_to_bufpos real_bytind_to_bufpos
996 #endif /* not ERROR_CHECK_BUFPOS */
998 #define make_bufpos(buf, ind) make_int (bytind_to_bufpos (buf, ind))
1000 /*----------------------------------------------------------------------*/
1001 /* Converting between buffer bytes and Emacs characters */
1002 /*----------------------------------------------------------------------*/
1004 /* The character at position POS in buffer. */
1005 #define BI_BUF_FETCH_CHAR(buf, pos) \
1006 charptr_emchar (BI_BUF_BYTE_ADDRESS (buf, pos))
1007 #define BUF_FETCH_CHAR(buf, pos) \
1008 BI_BUF_FETCH_CHAR (buf, bufpos_to_bytind (buf, pos))
1010 /* The character at position POS in buffer, as a string. This is
1011 equivalent to set_charptr_emchar (str, BUF_FETCH_CHAR (buf, pos))
1012 but is faster for Mule. */
1014 # define BI_BUF_CHARPTR_COPY_CHAR(buf, pos, str) \
1015 charptr_copy_char (BI_BUF_BYTE_ADDRESS (buf, pos), str)
1016 #define BUF_CHARPTR_COPY_CHAR(buf, pos, str) \
1017 BI_BUF_CHARPTR_COPY_CHAR (buf, bufpos_to_bytind (buf, pos), str)
1022 /************************************************************************/
1024 /* working with externally-formatted data */
1026 /************************************************************************/
1028 /* Sometimes strings need to be converted into one or another
1029 external format, for passing to a library function. (Note
1030 that we encapsulate and automatically convert the arguments
1031 of some functions, but not others.) At times this conversion
1032 also has to go the other way -- i.e. when we get external-
1033 format strings back from a library function.
1038 /* WARNING: These use a static buffer. This can lead to disaster if
1039 these functions are not used *very* carefully. Under normal
1040 circumstances, do not call these functions; call the front ends
1043 Extbyte *convert_to_external_format (CONST Bufbyte *ptr,
1046 enum external_data_format fmt);
1047 Bufbyte *convert_from_external_format (CONST Extbyte *ptr,
1050 enum external_data_format fmt);
1054 #define convert_to_external_format(ptr, len, len_out, fmt) \
1055 (*(len_out) = (int) (len), (Extbyte *) (ptr))
1056 #define convert_from_external_format(ptr, len, len_out, fmt) \
1057 (*(len_out) = (Bytecount) (len), (Bufbyte *) (ptr))
1061 /* In all of the following macros we use the following general principles:
1063 -- Functions that work with charptr's accept two sorts of charptr's:
1065 a) Pointers to memory with a length specified. The pointer will be
1066 fundamentally of type `unsigned char *' (although labelled
1067 as `Bufbyte *' for internal-format data and `Extbyte *' for
1068 external-format data) and the length will be fundamentally of
1069 type `int' (although labelled as `Bytecount' for internal-format
1070 data and `Extcount' for external-format data). The length is
1071 always a count in bytes.
1072 b) Zero-terminated pointers; no length specified. The pointer
1073 is of type `char *', whether the data pointed to is internal-format
1074 or external-format. These sorts of pointers are available for
1075 convenience in working with C library functions and literal
1076 strings. In general you should use these sorts of pointers only
1077 to interface to library routines and not for general manipulation,
1078 as you are liable to lose embedded nulls and such. This could
1079 be a big problem for routines that want Unicode-formatted data,
1080 which is likely to have lots of embedded nulls in it.
1081 (In the real world, though, external Unicode data will be UTF-8,
1082 which will not have embedded nulls and is ASCII-compatible - martin)
1084 -- Functions that work with Lisp strings accept strings as Lisp Objects
1085 (as opposed to the `struct Lisp_String *' for some of the other
1086 string accessors). This is for convenience in working with the
1087 functions, as otherwise you will almost always have to call
1088 XSTRING() on the object.
1090 -- Functions that work with charptr's are not guaranteed to copy
1091 their data into alloca()ed space. Functions that work with
1092 Lisp strings are, however. The reason is that Lisp strings can
1093 be relocated any time a GC happens, and it could happen at some
1094 rather unexpected times. The internal-external conversion is
1095 rarely done in time-critical functions, and so the slight
1096 extra time required for alloca() and copy is well-worth the
1097 safety of knowing your string data won't be relocated out from
1102 /* Maybe convert charptr's data into ext-format and store the result in
1105 You may wonder why this is written in this fashion and not as a
1106 function call. With a little trickery it could certainly be
1107 written this way, but it won't work because of those DAMN GCC WANKERS
1108 who couldn't be bothered to handle alloca() properly on the x86
1109 architecture. (If you put a call to alloca() in the argument to
1110 a function call, the stack space gets allocated right in the
1111 middle of the arguments to the function call and you are unbelievably
1116 #define GET_CHARPTR_EXT_DATA_ALLOCA(ptr, len, fmt, ptr_out, len_out) do \
1118 Bytecount gceda_len_in = (Bytecount) (len); \
1119 Extcount gceda_len_out; \
1120 CONST Bufbyte *gceda_ptr_in = (ptr); \
1121 Extbyte *gceda_ptr_out = \
1122 convert_to_external_format (gceda_ptr_in, gceda_len_in, \
1123 &gceda_len_out, fmt); \
1124 /* If the new string is identical to the old (will be the case most \
1125 of the time), just return the same string back. This saves \
1126 on alloca()ing, which can be useful on C alloca() machines and \
1127 on stack-space-challenged environments. */ \
1129 if (gceda_len_in == gceda_len_out && \
1130 !memcmp (gceda_ptr_in, gceda_ptr_out, gceda_len_out)) \
1132 (ptr_out) = (Extbyte *) gceda_ptr_in; \
1136 (ptr_out) = (Extbyte *) alloca (1 + gceda_len_out); \
1137 memcpy ((void *) ptr_out, gceda_ptr_out, 1 + gceda_len_out); \
1139 (len_out) = gceda_len_out; \
1144 #define GET_CHARPTR_EXT_DATA_ALLOCA(ptr, len, fmt, ptr_out, len_out) do \
1146 (ptr_out) = (Extbyte *) (ptr); \
1147 (len_out) = (Extcount) (len); \
1152 #define GET_C_CHARPTR_EXT_DATA_ALLOCA(ptr, fmt, ptr_out) do \
1154 Extcount gcceda_ignored_len; \
1155 CONST Bufbyte *gcceda_ptr_in = (CONST Bufbyte *) (ptr); \
1156 Extbyte *gcceda_ptr_out; \
1158 GET_CHARPTR_EXT_DATA_ALLOCA (gcceda_ptr_in, \
1159 strlen ((char *) gcceda_ptr_in), \
1162 gcceda_ignored_len); \
1163 (ptr_out) = (char *) gcceda_ptr_out; \
1166 #define GET_C_CHARPTR_EXT_BINARY_DATA_ALLOCA(ptr, ptr_out) \
1167 GET_C_CHARPTR_EXT_DATA_ALLOCA (ptr, FORMAT_BINARY, ptr_out)
1168 #define GET_CHARPTR_EXT_BINARY_DATA_ALLOCA(ptr, len, ptr_out, len_out) \
1169 GET_CHARPTR_EXT_DATA_ALLOCA (ptr, len, FORMAT_BINARY, ptr_out, len_out)
1171 #define GET_C_CHARPTR_EXT_FILENAME_DATA_ALLOCA(ptr, ptr_out) \
1172 GET_C_CHARPTR_EXT_DATA_ALLOCA (ptr, FORMAT_FILENAME, ptr_out)
1173 #define GET_CHARPTR_EXT_FILENAME_DATA_ALLOCA(ptr, len, ptr_out, len_out) \
1174 GET_CHARPTR_EXT_DATA_ALLOCA (ptr, len, FORMAT_FILENAME, ptr_out, len_out)
1176 #define GET_C_CHARPTR_EXT_CTEXT_DATA_ALLOCA(ptr, ptr_out) \
1177 GET_C_CHARPTR_EXT_DATA_ALLOCA (ptr, FORMAT_CTEXT, ptr_out)
1178 #define GET_CHARPTR_EXT_CTEXT_DATA_ALLOCA(ptr, len, ptr_out, len_out) \
1179 GET_CHARPTR_EXT_DATA_ALLOCA (ptr, len, FORMAT_CTEXT, ptr_out, len_out)
1181 /* Maybe convert external charptr's data into internal format and store
1182 the result in alloca()'ed space.
1184 You may wonder why this is written in this fashion and not as a
1185 function call. With a little trickery it could certainly be
1186 written this way, but it won't work because of those DAMN GCC WANKERS
1187 who couldn't be bothered to handle alloca() properly on the x86
1188 architecture. (If you put a call to alloca() in the argument to
1189 a function call, the stack space gets allocated right in the
1190 middle of the arguments to the function call and you are unbelievably
1195 #define GET_CHARPTR_INT_DATA_ALLOCA(ptr, len, fmt, ptr_out, len_out) do \
1197 Extcount gcida_len_in = (Extcount) (len); \
1198 Bytecount gcida_len_out; \
1199 CONST Extbyte *gcida_ptr_in = (ptr); \
1200 Bufbyte *gcida_ptr_out = \
1201 convert_from_external_format (gcida_ptr_in, gcida_len_in, \
1202 &gcida_len_out, fmt); \
1203 /* If the new string is identical to the old (will be the case most \
1204 of the time), just return the same string back. This saves \
1205 on alloca()ing, which can be useful on C alloca() machines and \
1206 on stack-space-challenged environments. */ \
1208 if (gcida_len_in == gcida_len_out && \
1209 !memcmp (gcida_ptr_in, gcida_ptr_out, gcida_len_out)) \
1211 (ptr_out) = (Bufbyte *) gcida_ptr_in; \
1215 (ptr_out) = (Extbyte *) alloca (1 + gcida_len_out); \
1216 memcpy ((void *) ptr_out, gcida_ptr_out, 1 + gcida_len_out); \
1218 (len_out) = gcida_len_out; \
1223 #define GET_CHARPTR_INT_DATA_ALLOCA(ptr, len, fmt, ptr_out, len_out) do \
1225 (ptr_out) = (Bufbyte *) (ptr); \
1226 (len_out) = (Bytecount) (len); \
1231 #define GET_C_CHARPTR_INT_DATA_ALLOCA(ptr, fmt, ptr_out) do \
1233 Bytecount gccida_ignored_len; \
1234 CONST Extbyte *gccida_ptr_in = (CONST Extbyte *) (ptr); \
1235 Bufbyte *gccida_ptr_out; \
1237 GET_CHARPTR_INT_DATA_ALLOCA (gccida_ptr_in, \
1238 strlen ((char *) gccida_ptr_in), \
1241 gccida_ignored_len); \
1242 (ptr_out) = gccida_ptr_out; \
1245 #define GET_C_CHARPTR_INT_BINARY_DATA_ALLOCA(ptr, ptr_out) \
1246 GET_C_CHARPTR_INT_DATA_ALLOCA (ptr, FORMAT_BINARY, ptr_out)
1247 #define GET_CHARPTR_INT_BINARY_DATA_ALLOCA(ptr, len, ptr_out, len_out) \
1248 GET_CHARPTR_INT_DATA_ALLOCA (ptr, len, FORMAT_BINARY, ptr_out, len_out)
1250 #define GET_C_CHARPTR_INT_FILENAME_DATA_ALLOCA(ptr, ptr_out) \
1251 GET_C_CHARPTR_INT_DATA_ALLOCA (ptr, FORMAT_FILENAME, ptr_out)
1252 #define GET_CHARPTR_INT_FILENAME_DATA_ALLOCA(ptr, len, ptr_out, len_out) \
1253 GET_CHARPTR_INT_DATA_ALLOCA (ptr, len, FORMAT_FILENAME, ptr_out, len_out)
1255 #define GET_C_CHARPTR_INT_CTEXT_DATA_ALLOCA(ptr, ptr_out) \
1256 GET_C_CHARPTR_INT_DATA_ALLOCA (ptr, FORMAT_CTEXT, ptr_out)
1257 #define GET_CHARPTR_INT_CTEXT_DATA_ALLOCA(ptr, len, ptr_out, len_out) \
1258 GET_CHARPTR_INT_DATA_ALLOCA (ptr, len, FORMAT_CTEXT, ptr_out, len_out)
1261 /* Maybe convert Lisp string's data into ext-format and store the result in
1264 You may wonder why this is written in this fashion and not as a
1265 function call. With a little trickery it could certainly be
1266 written this way, but it won't work because of those DAMN GCC WANKERS
1267 who couldn't be bothered to handle alloca() properly on the x86
1268 architecture. (If you put a call to alloca() in the argument to
1269 a function call, the stack space gets allocated right in the
1270 middle of the arguments to the function call and you are unbelievably
1273 #define GET_STRING_EXT_DATA_ALLOCA(s, fmt, ptr_out, len_out) do \
1275 Extcount gseda_len_out; \
1276 struct Lisp_String *gseda_s = XSTRING (s); \
1277 Extbyte * gseda_ptr_out = \
1278 convert_to_external_format (string_data (gseda_s), \
1279 string_length (gseda_s), \
1280 &gseda_len_out, fmt); \
1281 (ptr_out) = (Extbyte *) alloca (1 + gseda_len_out); \
1282 memcpy ((void *) ptr_out, gseda_ptr_out, 1 + gseda_len_out); \
1283 (len_out) = gseda_len_out; \
1287 #define GET_C_STRING_EXT_DATA_ALLOCA(s, fmt, ptr_out) do \
1289 Extcount gcseda_ignored_len; \
1290 Extbyte *gcseda_ptr_out; \
1292 GET_STRING_EXT_DATA_ALLOCA (s, fmt, gcseda_ptr_out, \
1293 gcseda_ignored_len); \
1294 (ptr_out) = (char *) gcseda_ptr_out; \
1297 #define GET_STRING_BINARY_DATA_ALLOCA(s, ptr_out, len_out) \
1298 GET_STRING_EXT_DATA_ALLOCA (s, FORMAT_BINARY, ptr_out, len_out)
1299 #define GET_C_STRING_BINARY_DATA_ALLOCA(s, ptr_out) \
1300 GET_C_STRING_EXT_DATA_ALLOCA (s, FORMAT_BINARY, ptr_out)
1302 #define GET_STRING_FILENAME_DATA_ALLOCA(s, ptr_out, len_out) \
1303 GET_STRING_EXT_DATA_ALLOCA (s, FORMAT_FILENAME, ptr_out, len_out)
1304 #define GET_C_STRING_FILENAME_DATA_ALLOCA(s, ptr_out) \
1305 GET_C_STRING_EXT_DATA_ALLOCA (s, FORMAT_FILENAME, ptr_out)
1307 #define GET_STRING_OS_DATA_ALLOCA(s, ptr_out, len_out) \
1308 GET_STRING_EXT_DATA_ALLOCA (s, FORMAT_OS, ptr_out, len_out)
1309 #define GET_C_STRING_OS_DATA_ALLOCA(s, ptr_out) \
1310 GET_C_STRING_EXT_DATA_ALLOCA (s, FORMAT_OS, ptr_out)
1312 #define GET_STRING_CTEXT_DATA_ALLOCA(s, ptr_out, len_out) \
1313 GET_STRING_EXT_DATA_ALLOCA (s, FORMAT_CTEXT, ptr_out, len_out)
1314 #define GET_C_STRING_CTEXT_DATA_ALLOCA(s, ptr_out) \
1315 GET_C_STRING_EXT_DATA_ALLOCA (s, FORMAT_CTEXT, ptr_out)
1319 /************************************************************************/
1321 /* fake charset functions */
1323 /************************************************************************/
1325 /* used when MULE is not defined, so that Charset-type stuff can still
1330 typedef int Charset_ID;
1332 #define Vcharset_ascii Qnil
1334 #define CHAR_CHARSET(ch) Vcharset_ascii
1335 #define CHAR_LEADING_BYTE(ch) LEADING_BYTE_ASCII
1336 #define LEADING_BYTE_ASCII 0x80
1337 #define NUM_LEADING_BYTES 1
1338 #define MIN_LEADING_BYTE 0x80
1339 #define CHARSETP(cs) 1
1340 #define CHARSET_BY_LEADING_BYTE(lb) Vcharset_ascii
1341 #define XCHARSET_LEADING_BYTE(cs) LEADING_BYTE_ASCII
1342 #define XCHARSET_GRAPHIC(cs) -1
1343 #define XCHARSET_COLUMNS(cs) 1
1344 #define XCHARSET_DIMENSION(cs) 1
1345 #define REP_BYTES_BY_FIRST_BYTE(fb) 1
1346 #define BREAKUP_CHAR(ch, charset, byte1, byte2) do { \
1347 (charset) = Vcharset_ascii; \
1351 #define BYTE_ASCII_P(byte) 1
1355 /************************************************************************/
1357 /* higher-level buffer-position functions */
1359 /************************************************************************/
1361 /*----------------------------------------------------------------------*/
1362 /* Settor macros for important positions in a buffer */
1363 /*----------------------------------------------------------------------*/
1365 /* Set beginning of accessible range of buffer. */
1366 #define SET_BOTH_BUF_BEGV(buf, val, bival) \
1369 (buf)->begv = (bival); \
1370 (buf)->bufbegv = (val); \
1373 /* Set end of accessible range of buffer. */
1374 #define SET_BOTH_BUF_ZV(buf, val, bival) \
1377 (buf)->zv = (bival); \
1378 (buf)->bufzv = (val); \
1382 /* Since BEGV and ZV are almost never set, it's reasonable to enforce
1383 the restriction that the Bufpos and Bytind values must both be
1384 specified. However, point is set in lots and lots of places. So
1385 we provide the ability to specify both (for efficiency) or just
1387 #define BOTH_BUF_SET_PT(buf, val, bival) set_buffer_point (buf, val, bival)
1388 #define BI_BUF_SET_PT(buf, bival) \
1389 BOTH_BUF_SET_PT (buf, bytind_to_bufpos (buf, bival), bival)
1390 #define BUF_SET_PT(buf, value) \
1391 BOTH_BUF_SET_PT (buf, value, bufpos_to_bytind (buf, value))
1395 /* These macros exist in FSFmacs because SET_PT() in FSFmacs incorrectly
1396 does too much stuff, such as moving out of invisible extents. */
1397 #define TEMP_SET_PT(position) (temp_set_point ((position), current_buffer))
1398 #define SET_BUF_PT(buf, value) ((buf)->pt = (value))
1399 #endif /* FSFmacs */
1401 /*----------------------------------------------------------------------*/
1402 /* Miscellaneous buffer values */
1403 /*----------------------------------------------------------------------*/
1405 /* Number of characters in buffer */
1406 #define BUF_SIZE(buf) (BUF_Z (buf) - BUF_BEG (buf))
1408 /* Is this buffer narrowed? */
1409 #define BUF_NARROWED(buf) \
1410 ((BI_BUF_BEGV (buf) != BI_BUF_BEG (buf)) || \
1411 (BI_BUF_ZV (buf) != BI_BUF_Z (buf)))
1413 /* Modification count. */
1414 #define BUF_MODIFF(buf) ((buf)->text->modiff)
1416 /* Saved modification count. */
1417 #define BUF_SAVE_MODIFF(buf) ((buf)->text->save_modiff)
1420 #define BUF_FACECHANGE(buf) ((buf)->face_change)
1422 #define POINT_MARKER_P(marker) \
1423 (XMARKER (marker)->buffer != 0 && \
1424 EQ ((marker), XMARKER (marker)->buffer->point_marker))
1426 #define BUF_MARKERS(buf) ((buf)->markers)
1430 The new definitions of CEILING_OF() and FLOOR_OF() differ semantically
1431 from the old ones (in FSF Emacs and XEmacs 19.11 and before).
1432 Conversion is as follows:
1434 OLD_BI_CEILING_OF(n) = NEW_BI_CEILING_OF(n) - 1
1435 OLD_BI_FLOOR_OF(n) = NEW_BI_FLOOR_OF(n + 1)
1437 The definitions were changed because the new definitions are more
1438 consistent with the way everything else works in Emacs.
1441 /* Properties of CEILING_OF and FLOOR_OF (also apply to BI_ variants):
1443 1) FLOOR_OF (CEILING_OF (n)) = n
1444 CEILING_OF (FLOOR_OF (n)) = n
1446 2) CEILING_OF (n) = n if and only if n = ZV
1447 FLOOR_OF (n) = n if and only if n = BEGV
1449 3) CEILING_OF (CEILING_OF (n)) = ZV
1450 FLOOR_OF (FLOOR_OF (n)) = BEGV
1452 4) The bytes in the regions
1454 [BYTE_ADDRESS (n), BYTE_ADDRESS_BEFORE (CEILING_OF (n))]
1458 [BYTE_ADDRESS (FLOOR_OF (n)), BYTE_ADDRESS_BEFORE (n)]
1464 /* Return the maximum index in the buffer it is safe to scan forwards
1465 past N to. This is used to prevent buffer scans from running into
1466 the gap (e.g. search.c). All characters between N and CEILING_OF(N)
1467 are located contiguous in memory. Note that the character *at*
1468 CEILING_OF(N) is not contiguous in memory. */
1469 #define BI_BUF_CEILING_OF(b, n) \
1470 ((n) < (b)->text->gpt && (b)->text->gpt < BI_BUF_ZV (b) ? \
1471 (b)->text->gpt : BI_BUF_ZV (b))
1472 #define BUF_CEILING_OF(b, n) \
1473 bytind_to_bufpos (b, BI_BUF_CEILING_OF (b, bufpos_to_bytind (b, n)))
1475 /* Return the minimum index in the buffer it is safe to scan backwards
1476 past N to. All characters between FLOOR_OF(N) and N are located
1477 contiguous in memory. Note that the character *at* N may not be
1478 contiguous in memory. */
1479 #define BI_BUF_FLOOR_OF(b, n) \
1480 (BI_BUF_BEGV (b) < (b)->text->gpt && (b)->text->gpt < (n) ? \
1481 (b)->text->gpt : BI_BUF_BEGV (b))
1482 #define BUF_FLOOR_OF(b, n) \
1483 bytind_to_bufpos (b, BI_BUF_FLOOR_OF (b, bufpos_to_bytind (b, n)))
1485 #define BI_BUF_CEILING_OF_IGNORE_ACCESSIBLE(b, n) \
1486 ((n) < (b)->text->gpt && (b)->text->gpt < BI_BUF_Z (b) ? \
1487 (b)->text->gpt : BI_BUF_Z (b))
1488 #define BUF_CEILING_OF_IGNORE_ACCESSIBLE(b, n) \
1490 (b, BI_BUF_CEILING_OF_IGNORE_ACCESSIBLE (b, bufpos_to_bytind (b, n)))
1492 #define BI_BUF_FLOOR_OF_IGNORE_ACCESSIBLE(b, n) \
1493 (BI_BUF_BEG (b) < (b)->text->gpt && (b)->text->gpt < (n) ? \
1494 (b)->text->gpt : BI_BUF_BEG (b))
1495 #define BUF_FLOOR_OF_IGNORE_ACCESSIBLE(b, n) \
1497 (b, BI_BUF_FLOOR_OF_IGNORE_ACCESSIBLE (b, bufpos_to_bytind (b, n)))
1500 extern struct buffer *current_buffer;
1502 /* This is the initial (startup) directory, as used for the *scratch* buffer.
1503 We're making this a global to make others aware of the startup directory.
1504 `initial_directory' is stored in external format.
1506 extern char initial_directory[];
1507 extern void init_initial_directory (void); /* initialize initial_directory */
1509 EXFUN (Fbuffer_disable_undo, 1);
1510 EXFUN (Fbuffer_modified_p, 1);
1511 EXFUN (Fbuffer_name, 1);
1512 EXFUN (Fcurrent_buffer, 0);
1513 EXFUN (Ferase_buffer, 1);
1514 EXFUN (Fget_buffer, 1);
1515 EXFUN (Fget_buffer_create, 1);
1516 EXFUN (Fget_file_buffer, 1);
1517 EXFUN (Fkill_buffer, 1);
1518 EXFUN (Fother_buffer, 3);
1519 EXFUN (Frecord_buffer, 1);
1520 EXFUN (Fset_buffer, 1);
1521 EXFUN (Fset_buffer_modified_p, 2);
1523 extern Lisp_Object QSscratch, Qafter_change_function, Qafter_change_functions;
1524 extern Lisp_Object Qbefore_change_function, Qbefore_change_functions;
1525 extern Lisp_Object Qbuffer_or_string_p, Qdefault_directory, Qfirst_change_hook;
1526 extern Lisp_Object Qpermanent_local, Vafter_change_function;
1527 extern Lisp_Object Vafter_change_functions, Vbefore_change_function;
1528 extern Lisp_Object Vbefore_change_functions, Vbuffer_alist, Vbuffer_defaults;
1529 extern Lisp_Object Vinhibit_read_only, Vtransient_mark_mode;
1531 /* This structure marks which slots in a buffer have corresponding
1532 default values in Vbuffer_defaults.
1533 Each such slot has a nonzero value in this structure.
1534 The value has only one nonzero bit.
1536 When a buffer has its own local value for a slot,
1537 the bit for that slot (found in the same slot in this structure)
1538 is turned on in the buffer's local_var_flags slot.
1540 If a slot in this structure is zero, then even though there may
1541 be a DEFVAR_BUFFER_LOCAL for the slot, there is no default value for it;
1542 and the corresponding slot in Vbuffer_defaults is not used. */
1544 extern struct buffer buffer_local_flags;
1547 /* Allocation of buffer data. */
1551 char *r_alloc (unsigned char **, unsigned long);
1552 char *r_re_alloc (unsigned char **, unsigned long);
1553 void r_alloc_free (unsigned char **);
1555 #define BUFFER_ALLOC(data, size) \
1556 ((Bufbyte *) r_alloc ((unsigned char **) &data, (size) * sizeof(Bufbyte)))
1557 #define BUFFER_REALLOC(data, size) \
1558 ((Bufbyte *) r_re_alloc ((unsigned char **) &data, (size) * sizeof(Bufbyte)))
1559 #define BUFFER_FREE(data) r_alloc_free ((unsigned char **) &(data))
1560 #define R_ALLOC_DECLARE(var,data) r_alloc_declare (&(var), data)
1562 #else /* !REL_ALLOC */
1564 #define BUFFER_ALLOC(data,size)\
1565 (data = xnew_array (Bufbyte, size))
1566 #define BUFFER_REALLOC(data,size)\
1567 ((Bufbyte *) xrealloc (data, (size) * sizeof(Bufbyte)))
1568 /* Avoid excess parentheses, or syntax errors may rear their heads. */
1569 #define BUFFER_FREE(data) xfree (data)
1570 #define R_ALLOC_DECLARE(var,data)
1572 #endif /* !REL_ALLOC */
1574 extern Lisp_Object Vbuffer_alist;
1575 void set_buffer_internal (struct buffer *b);
1576 struct buffer *decode_buffer (Lisp_Object buffer, int allow_string);
1578 /* from editfns.c */
1579 void widen_buffer (struct buffer *b, int no_clip);
1580 int beginning_of_line_p (struct buffer *b, Bufpos pt);
1583 void set_buffer_point (struct buffer *buf, Bufpos pos, Bytind bipos);
1584 void find_charsets_in_bufbyte_string (Charset_ID *charsets,
1587 void find_charsets_in_emchar_string (Charset_ID *charsets,
1590 int bufbyte_string_displayed_columns (CONST Bufbyte *str, Bytecount len);
1591 int emchar_string_displayed_columns (CONST Emchar *str, Charcount len);
1592 void convert_bufbyte_string_into_emchar_dynarr (CONST Bufbyte *str,
1594 Emchar_dynarr *dyn);
1595 Charcount convert_bufbyte_string_into_emchar_string (CONST Bufbyte *str,
1598 void convert_emchar_string_into_bufbyte_dynarr (Emchar *arr, int nels,
1599 Bufbyte_dynarr *dyn);
1600 Bufbyte *convert_emchar_string_into_malloced_string (Emchar *arr, int nels,
1601 Bytecount *len_out);
1603 void init_buffer_markers (struct buffer *b);
1604 void uninit_buffer_markers (struct buffer *b);
1606 /* flags for get_buffer_pos_char(), get_buffer_range_char(), etc. */
1607 /* At most one of GB_COERCE_RANGE and GB_NO_ERROR_IF_BAD should be
1608 specified. At most one of GB_NEGATIVE_FROM_END and GB_NO_ERROR_IF_BAD
1609 should be specified. */
1611 #define GB_ALLOW_PAST_ACCESSIBLE (1 << 0)
1612 #define GB_ALLOW_NIL (1 << 1)
1613 #define GB_CHECK_ORDER (1 << 2)
1614 #define GB_COERCE_RANGE (1 << 3)
1615 #define GB_NO_ERROR_IF_BAD (1 << 4)
1616 #define GB_NEGATIVE_FROM_END (1 << 5)
1617 #define GB_HISTORICAL_STRING_BEHAVIOR (GB_NEGATIVE_FROM_END | GB_ALLOW_NIL)
1619 Bufpos get_buffer_pos_char (struct buffer *b, Lisp_Object pos,
1620 unsigned int flags);
1621 Bytind get_buffer_pos_byte (struct buffer *b, Lisp_Object pos,
1622 unsigned int flags);
1623 void get_buffer_range_char (struct buffer *b, Lisp_Object from, Lisp_Object to,
1624 Bufpos *from_out, Bufpos *to_out,
1625 unsigned int flags);
1626 void get_buffer_range_byte (struct buffer *b, Lisp_Object from, Lisp_Object to,
1627 Bytind *from_out, Bytind *to_out,
1628 unsigned int flags);
1629 Charcount get_string_pos_char (Lisp_Object string, Lisp_Object pos,
1630 unsigned int flags);
1631 Bytecount get_string_pos_byte (Lisp_Object string, Lisp_Object pos,
1632 unsigned int flags);
1633 void get_string_range_char (Lisp_Object string, Lisp_Object from,
1634 Lisp_Object to, Charcount *from_out,
1635 Charcount *to_out, unsigned int flags);
1636 void get_string_range_byte (Lisp_Object string, Lisp_Object from,
1637 Lisp_Object to, Bytecount *from_out,
1638 Bytecount *to_out, unsigned int flags);
1639 Bufpos get_buffer_or_string_pos_char (Lisp_Object object, Lisp_Object pos,
1640 unsigned int flags);
1641 Bytind get_buffer_or_string_pos_byte (Lisp_Object object, Lisp_Object pos,
1642 unsigned int flags);
1643 void get_buffer_or_string_range_char (Lisp_Object object, Lisp_Object from,
1644 Lisp_Object to, Bufpos *from_out,
1645 Bufpos *to_out, unsigned int flags);
1646 void get_buffer_or_string_range_byte (Lisp_Object object, Lisp_Object from,
1647 Lisp_Object to, Bytind *from_out,
1648 Bytind *to_out, unsigned int flags);
1649 Bufpos buffer_or_string_accessible_begin_char (Lisp_Object object);
1650 Bufpos buffer_or_string_accessible_end_char (Lisp_Object object);
1651 Bytind buffer_or_string_accessible_begin_byte (Lisp_Object object);
1652 Bytind buffer_or_string_accessible_end_byte (Lisp_Object object);
1653 Bufpos buffer_or_string_absolute_begin_char (Lisp_Object object);
1654 Bufpos buffer_or_string_absolute_end_char (Lisp_Object object);
1655 Bytind buffer_or_string_absolute_begin_byte (Lisp_Object object);
1656 Bytind buffer_or_string_absolute_end_byte (Lisp_Object object);
1657 void record_buffer (Lisp_Object buf);
1658 Lisp_Object get_buffer (Lisp_Object name,
1659 int error_if_deleted_or_does_not_exist);
1660 int map_over_sharing_buffers (struct buffer *buf,
1661 int (*mapfun) (struct buffer *buf,
1666 /************************************************************************/
1667 /* Case conversion */
1668 /************************************************************************/
1670 /* A "trt" table is a mapping from characters to other characters,
1671 typically used to convert between uppercase and lowercase. For
1672 compatibility reasons, trt tables are currently in the form of
1673 a Lisp string of 256 characters, specifying the conversion for each
1674 of the first 256 Emacs characters (i.e. the 256 Latin-1 characters).
1675 This should be generalized at some point to support conversions for
1676 all of the allowable Mule characters.
1679 /* The _1 macros are named as such because they assume that you have
1680 already guaranteed that the character values are all in the range
1681 0 - 255. Bad lossage will happen otherwise. */
1683 # define MAKE_TRT_TABLE() Fmake_string (make_int (256), make_char (0))
1684 # define TRT_TABLE_AS_STRING(table) XSTRING_DATA (table)
1685 # define TRT_TABLE_CHAR_1(table, ch) \
1686 string_char (XSTRING (table), (Charcount) ch)
1687 # define SET_TRT_TABLE_CHAR_1(table, ch1, ch2) \
1688 set_string_char (XSTRING (table), (Charcount) ch1, ch2)
1691 # define MAKE_MIRROR_TRT_TABLE() make_opaque (256, 0)
1692 # define MIRROR_TRT_TABLE_AS_STRING(table) ((Bufbyte *) XOPAQUE_DATA (table))
1693 # define MIRROR_TRT_TABLE_CHAR_1(table, ch) \
1694 ((Emchar) (MIRROR_TRT_TABLE_AS_STRING (table)[ch]))
1695 # define SET_MIRROR_TRT_TABLE_CHAR_1(table, ch1, ch2) \
1696 (MIRROR_TRT_TABLE_AS_STRING (table)[ch1] = (Bufbyte) (ch2))
1699 # define IN_TRT_TABLE_DOMAIN(c) (((EMACS_UINT) (c)) <= 255)
1702 #define MIRROR_DOWNCASE_TABLE_AS_STRING(buf) \
1703 MIRROR_TRT_TABLE_AS_STRING (buf->mirror_downcase_table)
1704 #define MIRROR_UPCASE_TABLE_AS_STRING(buf) \
1705 MIRROR_TRT_TABLE_AS_STRING (buf->mirror_upcase_table)
1706 #define MIRROR_CANON_TABLE_AS_STRING(buf) \
1707 MIRROR_TRT_TABLE_AS_STRING (buf->mirror_case_canon_table)
1708 #define MIRROR_EQV_TABLE_AS_STRING(buf) \
1709 MIRROR_TRT_TABLE_AS_STRING (buf->mirror_case_eqv_table)
1711 #define MIRROR_DOWNCASE_TABLE_AS_STRING(buf) \
1712 TRT_TABLE_AS_STRING (buf->downcase_table)
1713 #define MIRROR_UPCASE_TABLE_AS_STRING(buf) \
1714 TRT_TABLE_AS_STRING (buf->upcase_table)
1715 #define MIRROR_CANON_TABLE_AS_STRING(buf) \
1716 TRT_TABLE_AS_STRING (buf->case_canon_table)
1717 #define MIRROR_EQV_TABLE_AS_STRING(buf) \
1718 TRT_TABLE_AS_STRING (buf->case_eqv_table)
1721 INLINE Emchar TRT_TABLE_OF (Lisp_Object trt, Emchar c);
1723 TRT_TABLE_OF (Lisp_Object trt, Emchar c)
1725 return IN_TRT_TABLE_DOMAIN (c) ? TRT_TABLE_CHAR_1 (trt, c) : c;
1728 /* Macros used below. */
1729 #define DOWNCASE_TABLE_OF(buf, c) TRT_TABLE_OF (buf->downcase_table, c)
1730 #define UPCASE_TABLE_OF(buf, c) TRT_TABLE_OF (buf->upcase_table, c)
1732 /* 1 if CH is upper case. */
1734 INLINE int UPPERCASEP (struct buffer *buf, Emchar ch);
1736 UPPERCASEP (struct buffer *buf, Emchar ch)
1738 return DOWNCASE_TABLE_OF (buf, ch) != ch;
1741 /* 1 if CH is lower case. */
1743 INLINE int LOWERCASEP (struct buffer *buf, Emchar ch);
1745 LOWERCASEP (struct buffer *buf, Emchar ch)
1747 return (UPCASE_TABLE_OF (buf, ch) != ch &&
1748 DOWNCASE_TABLE_OF (buf, ch) == ch);
1751 /* 1 if CH is neither upper nor lower case. */
1753 INLINE int NOCASEP (struct buffer *buf, Emchar ch);
1755 NOCASEP (struct buffer *buf, Emchar ch)
1757 return UPCASE_TABLE_OF (buf, ch) == ch;
1760 /* Upcase a character, or make no change if that cannot be done. */
1762 INLINE Emchar UPCASE (struct buffer *buf, Emchar ch);
1764 UPCASE (struct buffer *buf, Emchar ch)
1766 return (DOWNCASE_TABLE_OF (buf, ch) == ch) ? UPCASE_TABLE_OF (buf, ch) : ch;
1769 /* Upcase a character known to be not upper case. Unused. */
1771 #define UPCASE1(buf, ch) UPCASE_TABLE_OF (buf, ch)
1773 /* Downcase a character, or make no change if that cannot be done. */
1775 #define DOWNCASE(buf, ch) DOWNCASE_TABLE_OF (buf, ch)
1777 #endif /* _XEMACS_BUFFER_H_ */