Mercurial > hg > xemacs-beta
annotate src/line-number.c @ 5580:a0e81357194e
Move macros with shadows in bytecomp.el to the end of the files, cl-macs
lisp/ChangeLog addition:
2011-10-08 Aidan Kehoe <kehoea@parhasard.net>
* cl-macs.el:
* cl-macs.el (load-time-value):
* cl-macs.el (flet):
* cl-macs.el (labels):
* cl-macs.el (the):
* cl-macs.el (declare):
Move all these macros to the end of the file, since they're in
byte-compile-initial-macro-environment, and we don't want their
definitions to override that for the rest of the file during
byte-compilation. Happens not to matter right now, but avoids
surprises for anyone using the macros elsewhere in cl-macs down
the line.
author | Aidan Kehoe <kehoea@parhasard.net> |
---|---|
date | Sat, 08 Oct 2011 12:26:09 +0100 |
parents | 308d34e9f07d |
children | 56144c8593a8 |
rev | line source |
---|---|
428 | 1 /* Line number cache. |
2 Copyright (C) 1997 Free Software Foundation, Inc. | |
3 | |
4 This file is part of XEmacs. | |
5 | |
5402
308d34e9f07d
Changed bulk of GPLv2 or later files identified by script
Mats Lidell <matsl@xemacs.org>
parents:
867
diff
changeset
|
6 XEmacs is free software: you can redistribute it and/or modify it |
428 | 7 under the terms of the GNU General Public License as published by the |
5402
308d34e9f07d
Changed bulk of GPLv2 or later files identified by script
Mats Lidell <matsl@xemacs.org>
parents:
867
diff
changeset
|
8 Free Software Foundation, either version 3 of the License, or (at your |
308d34e9f07d
Changed bulk of GPLv2 or later files identified by script
Mats Lidell <matsl@xemacs.org>
parents:
867
diff
changeset
|
9 option) any later version. |
428 | 10 |
11 XEmacs is distributed in the hope that it will be useful, but WITHOUT | |
12 ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or | |
13 FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License | |
14 for more details. | |
15 | |
16 You should have received a copy of the GNU General Public License | |
5402
308d34e9f07d
Changed bulk of GPLv2 or later files identified by script
Mats Lidell <matsl@xemacs.org>
parents:
867
diff
changeset
|
17 along with XEmacs. If not, see <http://www.gnu.org/licenses/>. */ |
428 | 18 |
19 /* Synched up with: Not in FSF. */ | |
20 | |
21 /* To calculate the line numbers, redisplay must count the newlines | |
22 from a known position. This used to be BUF_BEGV, but this made the | |
23 line numbering extremely slow for large buffers, because Emacs had | |
24 to rescan the whole buffer at each redisplay. | |
25 | |
26 To make line numbering efficient, we maintain a buffer-local cache | |
27 of recently used positions and their line numbers. The cache is | |
28 implemented as a small ring of cache positions. A cache position | |
29 is either nil or a cons of a buffer position (marker) and the | |
30 corresponding line number. | |
31 | |
32 When calculating the line numbers, this cache is consulted if it | |
33 would otherwise take too much time to count the newlines in the | |
34 buffer (see the comment to buffer_line_number().) | |
35 | |
36 Insertion and deletions that contain/delete newlines invalidate the | |
37 cached positions after the insertion point. This guarantees | |
38 relatively fast line numbers caching (even in buffers where point | |
39 moves a lot), and low memory usage. All of this is done only in | |
40 the buffers where the cache is actually initialized -- i.e. where | |
41 line-numbering is on, and you move the point farther than | |
42 LINE_NUMBER_FAR from the beginning of buffer. In this sense, the | |
43 cache is lazy -- if you don't use it, you don't pay for it. | |
44 | |
45 NOTE: line-number cache should not be confused with line-start | |
46 cache. Line-start cache (a part of redisplay) works with the | |
47 display lines, whereas this works with the buffer lines (literally | |
48 counting the newlines). */ | |
49 | |
50 #include <config.h> | |
51 #include "lisp.h" | |
52 #include "buffer.h" | |
53 | |
54 #include "line-number.h" | |
55 | |
56 /* #### The following three values could stand more exploration for | |
57 best performance. */ | |
58 | |
59 /* Size of the ring. The current code expects this to be a small | |
60 number. If you make it larger, you should probably optimize the | |
61 code below to keep it sorted. */ | |
62 #define LINE_NUMBER_RING_SIZE 8 | |
63 | |
64 /* How much traversal has to be exceeded for two points to be | |
65 considered "far" from each other. When two points are far, cache | |
66 will be used. */ | |
67 #define LINE_NUMBER_FAR 16384 | |
68 | |
69 /* How large a string has to be to give up searching it for newlines, | |
70 before change. */ | |
71 #define LINE_NUMBER_LARGE_STRING 256 | |
72 | |
73 /* To be used only when you *know* the cache has been allocated! */ | |
74 #define LINE_NUMBER_RING(b) (XCAR ((b)->text->line_number_cache)) | |
75 #define LINE_NUMBER_BEGV(b) (XCDR ((b)->text->line_number_cache)) | |
76 | |
77 | |
78 /* Initialize the cache. Cache is (in pseudo-BNF): | |
79 | |
80 CACHE = nil | INITIALIZED-CACHE | |
81 INITIALIZED-CACHE = cons (RING, BEGV-LINE) | |
82 RING = vector (*RING-ELEMENT) | |
83 RING-ELEMENT = nil | RING-PAIR | |
84 RING-PAIR = cons (marker, integer) | |
85 BEGV-LINE = integer | |
86 | |
87 Line number cache should never, ever, be visible to Lisp (because | |
88 destructively modifying its elements can cause crashes.) Debug it | |
89 using debug_print (current_buffer->text->last_number_cache). */ | |
90 static void | |
91 allocate_line_number_cache (struct buffer *b) | |
92 { | |
93 b->text->line_number_cache = Fcons (make_vector (LINE_NUMBER_RING_SIZE, Qnil), | |
94 Qzero); | |
95 narrow_line_number_cache (b); | |
96 } | |
97 | |
98 /* Flag LINE_NUMBER_BEGV (b) as dirty. Do it only if the line number | |
442 | 99 cache is already initialized. */ |
428 | 100 void |
101 narrow_line_number_cache (struct buffer *b) | |
102 { | |
103 if (NILP (b->text->line_number_cache)) | |
104 return; | |
105 | |
106 if (BUF_BEG (b) == BUF_BEGV (b)) | |
107 /* The is the case Fwiden and save_restriction_restore. Since we | |
108 know the correct value, we can update it now. */ | |
109 LINE_NUMBER_BEGV (b) = Qzero; | |
110 else | |
111 /* Calculating the line number of BUF_BEGV here is a bad idea, | |
112 because there is absolutely no reason to do it before the next | |
113 redisplay. We simply mark it as dirty instead. */ | |
114 LINE_NUMBER_BEGV (b) = make_int (-1); | |
115 } | |
116 | |
117 /* Invalidate the line number cache positions that lie after POS. */ | |
118 static void | |
665 | 119 invalidate_line_number_cache (struct buffer *b, Charbpos pos) |
428 | 120 { |
121 EMACS_INT i, j; | |
122 Lisp_Object *ring = XVECTOR_DATA (LINE_NUMBER_RING (b)); | |
123 | |
124 for (i = 0; i < LINE_NUMBER_RING_SIZE; i++) | |
125 { | |
126 if (!CONSP (ring[i])) | |
127 break; | |
128 /* As the marker stays behind the insertions, this check might | |
129 as well be `>'. However, Finsert_before_markers can advance | |
130 the marker anyway, which bites in shell buffers. | |
131 | |
132 #### This forces recreation of the cached marker (and | |
133 recalculation of newlines) every time a newline is inserted | |
134 at point, which is way losing. Isn't there a way to make a | |
135 marker impervious to Finsert_before_markers()?? Maybe I | |
136 should convert the code to use extents. */ | |
137 if (marker_position (XCAR (ring[i])) >= pos) | |
138 { | |
139 /* Get the marker out of the way. */ | |
140 Fset_marker (XCAR (ring[i]), Qnil, Qnil); | |
141 /* ...and shift the ring elements, up to the first nil. */ | |
142 for (j = i; !NILP (ring[j]) && j < LINE_NUMBER_RING_SIZE - 1; j++) | |
143 ring[j] = ring[j + 1]; | |
144 ring[j] = Qnil; | |
145 /* Must recheck position i. */ | |
146 i--; | |
147 } | |
148 } | |
149 } | |
150 | |
151 /* Invalidate the cache positions after POS, if the string to be | |
152 inserted contains a newline. If the string is too large (larger | |
153 than LINE_NUMBER_LARGE_STRING), invalidate the cache positions | |
154 after POS without prior search. | |
155 | |
156 This will do nothing if the cache is uninitialized. */ | |
157 void | |
665 | 158 insert_invalidate_line_number_cache (struct buffer *b, Charbpos pos, |
867 | 159 const Ibyte *nonreloc, Bytecount length) |
428 | 160 { |
161 if (NILP (b->text->line_number_cache)) | |
162 return; | |
163 | |
164 if (length > LINE_NUMBER_LARGE_STRING | |
165 || | |
166 /* We could also count how many newlines there are in the string | |
167 and update the cache accordingly, but it would be too much | |
168 work for too little gain. */ | |
647 | 169 memchr ((void *)nonreloc, '\n', length)) |
428 | 170 invalidate_line_number_cache (b, pos); |
171 } | |
172 | |
173 /* Invalidate the cache positions after FROM, if the region to be | |
174 deleted contains a newline. If the region-to-be-deleted is larger | |
175 than LINE_NUMBER_LARGE_STRING, invalidate the cache positions after | |
176 FROM without unconditionally. | |
177 | |
178 This will do nothing if the cache is uninitialized. */ | |
179 void | |
665 | 180 delete_invalidate_line_number_cache (struct buffer *b, Charbpos from, Charbpos to) |
428 | 181 { |
182 if (NILP (b->text->line_number_cache)) | |
183 return; | |
184 | |
185 if ((to - from) > LINE_NUMBER_LARGE_STRING) | |
186 invalidate_line_number_cache (b, from); | |
187 else | |
188 { | |
189 EMACS_INT shortage; | |
190 scan_buffer (b, '\n', from, to, 1, &shortage, 0); | |
191 if (!shortage) | |
192 invalidate_line_number_cache (b, from); | |
193 } | |
194 } | |
195 | |
196 /* Get the nearest known position we know the line number of | |
197 (i.e. BUF_BEGV, and cached positions). The return position will be | |
198 either closer than BEG, or BEG. The line of this known position | |
199 will be stored in LINE. | |
200 | |
201 *LINE should be initialized to the line number of BEG (normally, | |
202 BEG will be BUF_BEGV, and *LINE will be XINT (LINE_NUMBER_BEGV). | |
203 This will initialize the cache, if necessary. */ | |
204 static void | |
665 | 205 get_nearest_line_number (struct buffer *b, Charbpos *beg, Charbpos pos, |
428 | 206 EMACS_INT *line) |
207 { | |
208 EMACS_INT i; | |
209 Lisp_Object *ring = XVECTOR_DATA (LINE_NUMBER_RING (b)); | |
210 Charcount length = pos - *beg; | |
211 | |
212 if (length < 0) | |
213 length = -length; | |
214 | |
215 /* Find the ring entry closest to POS, if it is closer than BEG. */ | |
216 for (i = 0; i < LINE_NUMBER_RING_SIZE && CONSP (ring[i]); i++) | |
217 { | |
665 | 218 Charbpos newpos = marker_position (XCAR (ring[i])); |
428 | 219 Charcount howfar = newpos - pos; |
220 if (howfar < 0) | |
221 howfar = -howfar; | |
222 if (howfar < length) | |
223 { | |
224 length = howfar; | |
225 *beg = newpos; | |
226 *line = XINT (XCDR (ring[i])); | |
227 } | |
228 } | |
229 } | |
230 | |
231 /* Add a (POS . LINE) pair to the ring, and rotate it. */ | |
232 static void | |
665 | 233 add_position_to_cache (struct buffer *b, Charbpos pos, EMACS_INT line) |
428 | 234 { |
235 Lisp_Object *ring = XVECTOR_DATA (LINE_NUMBER_RING (b)); | |
236 int i = LINE_NUMBER_RING_SIZE - 1; | |
237 | |
238 /* Set the last marker in the ring to point nowhere. */ | |
239 if (CONSP (ring[i])) | |
240 Fset_marker (XCAR (ring[i]), Qnil, Qnil); | |
241 | |
242 /* Rotate the ring... */ | |
243 for (; i > 0; i--) | |
244 ring[i] = ring[i - 1]; | |
245 | |
246 /* ...and update it. */ | |
247 ring[0] = Fcons (Fset_marker (Fmake_marker (), make_int (pos), | |
771 | 248 wrap_buffer (b)), |
428 | 249 make_int (line)); |
250 } | |
251 | |
252 /* Calculate the line number in buffer B at position POS. If CACHEP | |
253 is non-zero, initialize and facilitate the line-number cache. The | |
254 line number of the first line is 0. If narrowing is in effect, | |
255 count the lines are counted from the beginning of the visible | |
256 portion of the buffer. | |
257 | |
258 The cache works as follows: To calculate the line number, we need | |
259 two positions: position of point (POS) and the position from which | |
260 to count newlines (BEG). We start by setting BEG to BUF_BEGV. If | |
261 this would require too much searching (i.e. pos - BUF_BEGV > | |
262 LINE_NUMBER_FAR), try to find a closer position in the ring. If it | |
263 is found, use that position for BEG, and increment the line number | |
264 appropriately. | |
265 | |
266 If the calculation (with or without the cache lookup) required more | |
267 than LINE_NUMBER_FAR characters of traversal, update the cache. */ | |
268 EMACS_INT | |
665 | 269 buffer_line_number (struct buffer *b, Charbpos pos, int cachep) |
428 | 270 { |
665 | 271 Charbpos beg = BUF_BEGV (b); |
428 | 272 EMACS_INT cached_lines = 0; |
273 EMACS_INT shortage, line; | |
274 | |
275 if ((pos > beg ? pos - beg : beg - pos) <= LINE_NUMBER_FAR) | |
276 cachep = 0; | |
277 | |
278 if (cachep) | |
279 { | |
280 if (NILP (b->text->line_number_cache)) | |
281 allocate_line_number_cache (b); | |
282 /* If we don't know the line number of BUF_BEGV, calculate it now. */ | |
283 if (XINT (LINE_NUMBER_BEGV (b)) == -1) | |
284 { | |
285 LINE_NUMBER_BEGV (b) = Qzero; | |
286 /* #### This has a side-effect of changing the cache. */ | |
287 LINE_NUMBER_BEGV (b) = | |
288 make_int (buffer_line_number (b, BUF_BEGV (b), 1)); | |
289 } | |
290 cached_lines = XINT (LINE_NUMBER_BEGV (b)); | |
291 get_nearest_line_number (b, &beg, pos, &cached_lines); | |
292 } | |
293 | |
294 scan_buffer (b, '\n', beg, pos, pos > beg ? EMACS_INT_MAX : -EMACS_INT_MAX, | |
295 &shortage, 0); | |
296 | |
297 line = EMACS_INT_MAX - shortage; | |
298 if (beg > pos) | |
299 line = -line; | |
300 line += cached_lines; | |
301 | |
302 if (cachep) | |
303 { | |
304 /* If too far, update the cache. */ | |
305 if ((pos > beg ? pos - beg : beg - pos) > LINE_NUMBER_FAR) | |
306 add_position_to_cache (b, pos, line); | |
307 /* Account for narrowing. If cache is not used, this is | |
308 unnecessary, because we counted from BUF_BEGV anyway. */ | |
309 line -= XINT (LINE_NUMBER_BEGV (b)); | |
310 } | |
311 | |
312 return line; | |
313 } |