code.delx.au - gnu-emacs/blob - src/region-cache.h

   1 /* Header file: Caching facts about regions of the buffer, for optimization.
   2    Copyright (C) 1985-1986, 1993, 1995, 2001-2011  Free Software Foundation, Inc.
   3
   4 This file is part of GNU Emacs.
   5
   6 GNU Emacs is free software: you can redistribute it and/or modify
   7 it under the terms of the GNU General Public License as published by
   8 the Free Software Foundation, either version 3 of the License, or
   9 (at your option) any later version.
  10
  11 GNU Emacs is distributed in the hope that it will be useful,
  12 but WITHOUT ANY WARRANTY; without even the implied warranty of
  13 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  14 GNU General Public License for more details.
  15
  16 You should have received a copy of the GNU General Public License
  17 along with GNU Emacs.  If not, see <http://www.gnu.org/licenses/>.  */
  18
  19
  20 /* This code was written by Jim Blandy <jimb@cs.oberlin.edu> to help
  21    GNU Emacs better support the gene editor written for the University
  22    of Illinois at Urbana-Champagne's Ribosome Database Project (RDP).
  23
  24    Emacs implements line operations (finding the beginning/end of the
  25    line, vertical motion, all the redisplay stuff) by searching for
  26    newlines in the buffer.  Usually, this is a good design; it's very
  27    clean to just represent the buffer as an unstructured string of
  28    characters, and the lines in most files are very short (less than
  29    eighty characters), meaning that scanning usually costs about the
  30    same as the overhead of maintaining some more complicated data
  31    structure.
  32
  33    However, some applications, like gene editing, make use of very
  34    long lines --- on the order of tens of kilobytes.  In such cases,
  35    it may well be worthwhile to try to avoid scanning, because the
  36    scans have become two orders of magnitude more expensive.  It would
  37    be nice if this speedup could preserve the simplicity of the
  38    existing data structure, and disturb as little of the existing code
  39    as possible.
  40
  41    So here's the tack.  We add some caching to the scan_buffer
  42    function, so that when it searches for a newline, it notes that the
  43    region between the start and end of the search contained no
  44    newlines; then, the next time around, it consults this cache to see
  45    if there are regions of text it can skip over completely.  The
  46    buffer modification primitives invalidate this cache.
  47
  48    (Note: Since the redisplay code needs similar information on
  49    modified regions of the buffer, we can use the code that helps out
  50    redisplay as a guide to where we need to add our own code to
  51    invalidate our cache.  prepare_to_modify_buffer seems to be the
  52    central spot.)
  53
  54    Note that the cache code itself never mentions newlines
  55    specifically, so if you wanted to cache other properties of regions
  56    of the buffer, you could use this code pretty much unchanged.  So
  57    this cache really holds "known/unknown" information --- "I know
  58    this region has property P" vs. "I don't know if this region has
  59    property P or not."  */
  60
  61
  62 /* Allocate, initialize and return a new, empty region cache.  */
  63 struct region_cache *new_region_cache (void);
  64
  65 /* Free a region cache.  */
  66 void free_region_cache (struct region_cache *);
  67
  68 /* Assert that the region of BUF between START and END (absolute
  69    buffer positions) is "known," for the purposes of CACHE (e.g. "has
  70    no newlines", in the case of the line cache).  */
  71 extern void know_region_cache (struct buffer *BUF,
  72                                struct region_cache *CACHE,
  73                                EMACS_INT START, EMACS_INT END);
  74
  75 /* Indicate that a section of BUF has changed, to invalidate CACHE.
  76    HEAD is the number of chars unchanged at the beginning of the buffer.
  77    TAIL is the number of chars unchanged at the end of the buffer.
  78       NOTE: this is *not* the same as the ending position of modified
  79       region.
  80    (This way of specifying regions makes more sense than absolute
  81    buffer positions in the presence of insertions and deletions; the
  82    args to pass are the same before and after such an operation.)  */
  83 extern void invalidate_region_cache (struct buffer *BUF,
  84                                      struct region_cache *CACHE,
  85                                      EMACS_INT HEAD, EMACS_INT TAIL);
  86
  87 /* The scanning functions.
  88
  89    Basically, if you're scanning forward/backward from position POS,
  90    and region_cache_forward/backward returns true, you can skip all
  91    the text between POS and *NEXT.  And if the function returns false,
  92    you should examine all the text from POS to *NEXT, and call
  93    know_region_cache depending on what you find there; this way, you
  94    might be able to avoid scanning it again.  */
  95
  96 /* Return true if the text immediately after POS in BUF is known, for
  97    the purposes of CACHE.  If NEXT is non-zero, set *NEXT to the nearest
  98    position after POS where the knownness changes.  */
  99 extern int region_cache_forward (struct buffer *BUF,
 100                                  struct region_cache *CACHE,
 101                                  EMACS_INT POS,
 102                                  EMACS_INT *NEXT);
 103
 104 /* Return true if the text immediately before POS in BUF is known, for
 105    the purposes of CACHE.  If NEXT is non-zero, set *NEXT to the nearest
 106    position before POS where the knownness changes.  */
 107 extern int region_cache_backward (struct buffer *BUF,
 108                                   struct region_cache *CACHE,
 109                                   EMACS_INT POS,
 110                                   EMACS_INT *NEXT);
 111