factor/extra/gap-buffer/gap-buffer.factor

295 lines
8.5 KiB
Factor

! Copyright (C) 2007 Alex Chapman All Rights Reserved.
! See http://factorcode.org/license.txt for BSD license.
!
! gap buffer -- largely influenced by Strandh and Villeneuve's Flexichain
! for a good introduction see:
! http://p-cos.net/lisp-ecoop/submissions/StrandhVilleneuveMoore.pdf
USING: kernel arrays sequences sequences.private circular math
math.order math.functions generic ;
IN: gap-buffer
! gap-start -- the first element of the gap
! gap-end -- the first element after the gap
! expand-factor -- should be > 1
! min-size -- < 5 is not sensible
TUPLE: gb
gap-start
gap-end
expand-factor
min-size ;
GENERIC: gb-seq ( gb -- seq )
GENERIC: set-gb-seq ( seq gb -- )
M: gb gb-seq ( gb -- seq ) delegate ;
M: gb set-gb-seq ( seq gb -- ) set-delegate ;
: required-space ( n gb -- n )
tuck gb-expand-factor * ceiling >fixnum swap gb-min-size max ;
: <gb> ( seq -- gb )
gb new
5 over set-gb-min-size
1.5 over set-gb-expand-factor
[ >r length r> set-gb-gap-start ] 2keep
[ swap length over required-space swap set-gb-gap-end ] 2keep
[
over length over required-space rot { } like resize-array <circular> swap set-gb-seq
] keep ;
M: gb like ( seq gb -- seq ) drop <gb> ;
: gap-length ( gb -- n ) [ gb-gap-end ] keep gb-gap-start - ;
: buffer-length ( gb -- n ) gb-seq length ;
M: gb length ( gb -- n ) [ buffer-length ] keep gap-length - ;
: valid-position? ( pos gb -- ? )
#! one element past the end of the buffer is a valid position when we're inserting
length -1 swap between? ;
: valid-index? ( i gb -- ? )
buffer-length -1 swap between? ;
TUPLE: position-out-of-bounds position gap-buffer ;
C: <position-out-of-bounds> position-out-of-bounds
: position>index ( pos gb -- i )
2dup valid-position? [
2dup gb-gap-start >= [
gap-length +
] [ drop ] if
] [
<position-out-of-bounds> throw
] if ;
TUPLE: index-out-of-bounds index gap-buffer ;
C: <index-out-of-bounds> index-out-of-bounds
: index>position ( i gb -- pos )
2dup valid-index? [
2dup gb-gap-end >= [
gap-length -
] [ drop ] if
] [
<index-out-of-bounds> throw
] if ;
M: gb virtual@ ( n gb -- n seq ) [ position>index ] keep gb-seq ;
M: gb nth ( n gb -- elt ) bounds-check virtual@ nth-unsafe ;
M: gb nth-unsafe ( n gb -- elt ) virtual@ nth-unsafe ;
M: gb set-nth ( elt n seq -- ) bounds-check virtual@ set-nth-unsafe ;
M: gb set-nth-unsafe ( elt n seq -- ) virtual@ set-nth-unsafe ;
M: gb virtual-seq gb-seq ;
INSTANCE: gb virtual-sequence
! ------------- moving the gap -------------------------------
: (copy-element) ( to start seq -- ) tuck nth -rot set-nth ;
: copy-element ( dst start seq -- ) >r [ + ] keep r> (copy-element) ;
: copy-elements-back ( dst start seq n -- )
dup 0 > [
>r [ copy-element ] 3keep >r 1+ r> r> 1- copy-elements-back
] [ 3drop drop ] if ;
: copy-elements-forward ( dst start seq n -- )
dup 0 > [
>r [ copy-element ] 3keep >r 1- r> r> 1- copy-elements-forward
] [ 3drop drop ] if ;
: copy-elements ( dst start end seq -- )
pick pick > [
>r dupd - r> swap copy-elements-forward
] [
>r over - r> swap copy-elements-back
] if ;
! the gap can be moved either forward or back. Moving the gap 'inside' means
! moving elements across the gap. Moving the gap 'around' means changing the
! start of the circular buffer to avoid moving as many elements.
! We decide which method (inside or around) to pick based on the number of
! elements that will need to be moved. We always try to move as few elements as
! possible.
: move-gap? ( i gb -- i gb ? ) 2dup gb-gap-end = not ;
: move-gap-forward? ( i gb -- i gb ? ) 2dup gb-gap-start >= ;
: move-gap-back-inside? ( i gb -- i gb ? )
#! is it cheaper to move the gap inside than around?
2dup [ gb-gap-start swap 2 * - ] keep [ buffer-length ] keep gb-gap-end - <= ;
: move-gap-forward-inside? ( i gb -- i gb ? )
#! is it cheaper to move the gap inside than around?
2dup [ gb-gap-end >r 2 * r> - ] keep [ gb-gap-start ] keep buffer-length + <= ;
: move-gap-forward-inside ( i gb -- )
[ dup gap-length neg swap gb-gap-end rot ] keep gb-seq copy-elements ;
: move-gap-back-inside ( i gb -- )
[ dup gap-length swap gb-gap-start 1- rot 1- ] keep gb-seq copy-elements ;
: move-gap-forward-around ( i gb -- )
0 over move-gap-back-inside [
dup buffer-length [
swap gap-length - neg swap
] keep
] keep [
gb-seq copy-elements
] keep dup gap-length swap gb-seq change-circular-start ;
: move-gap-back-around ( i gb -- )
dup buffer-length over move-gap-forward-inside [
length swap -1
] keep [
gb-seq copy-elements
] keep dup length swap gb-seq change-circular-start ;
: move-gap-forward ( i gb -- )
move-gap-forward-inside? [
move-gap-forward-inside
] [
move-gap-forward-around
] if ;
: move-gap-back ( i gb -- )
move-gap-back-inside? [
move-gap-back-inside
] [
move-gap-back-around
] if ;
: (move-gap) ( i gb -- )
move-gap? [
move-gap-forward? [
move-gap-forward
] [
move-gap-back
] if
] [ 2drop ] if ;
: fix-gap ( n gb -- )
2dup [ gap-length + ] keep set-gb-gap-end set-gb-gap-start ;
! moving the gap to position 5 means that the element in position 5 will be immediately after the gap
GENERIC: move-gap ( n gb -- )
M: gb move-gap ( n gb -- ) 2dup [ position>index ] keep (move-gap) fix-gap ;
! ------------ resizing -------------------------------------
: enough-room? ( n gb -- ? )
#! is there enough room to add 'n' elements to gb?
tuck length + swap buffer-length <= ;
: set-new-gap-end ( array gb -- )
[ buffer-length swap length swap - ] keep
[ gb-gap-end + ] keep set-gb-gap-end ;
: after-gap ( gb -- gb )
dup gb-seq swap gb-gap-end tail ;
: before-gap ( gb -- gb )
dup gb-gap-start head ;
: copy-after-gap ( array gb -- )
#! copy everything after the gap in 'gb' into the end of 'array',
#! and change 'gb's gap-end to reflect the gap-end in 'array'
dup after-gap >r 2dup set-new-gap-end gb-gap-end swap r> -rot copy ;
: copy-before-gap ( array gb -- )
#! copy everything before the gap in 'gb' into the start of 'array'
before-gap 0 rot copy ; ! gap start doesn't change
: resize-buffer ( gb new-size -- )
f <array> swap 2dup copy-before-gap 2dup copy-after-gap
>r <circular> r> set-gb-seq ;
: decrease-buffer-size ( gb -- )
#! the gap is too big, so resize to something sensible
dup length over required-space resize-buffer ;
: increase-buffer-size ( n gb -- )
#! increase the buffer to fit at least 'n' more elements
tuck length + over required-space resize-buffer ;
: gb-too-big? ( gb -- ? )
dup buffer-length over gb-min-size > [
dup length over buffer-length rot gb-expand-factor sq / <
] [ drop f ] if ;
: ?decrease ( gb -- )
dup gb-too-big? [
decrease-buffer-size
] [ drop ] if ;
: ensure-room ( n gb -- )
#! ensure that ther will be enough room for 'n' more elements
2dup enough-room? [ 2drop ] [
increase-buffer-size
] if ;
! ------- editing operations ---------------
GENERIC# insert* 2 ( seq position gb -- )
: prepare-insert ( seq position gb -- seq gb )
tuck move-gap over length over ensure-room ;
: insert-elements ( seq gb -- )
dup gb-gap-start swap gb-seq copy ;
: increment-gap-start ( gb n -- )
over gb-gap-start + swap set-gb-gap-start ;
! generic dispatch identifies numbers as sequences before numbers...
! M: number insert* ( elem position gb -- ) >r >r 1array r> r> insert* ;
: number-insert ( num position gb -- ) >r >r 1array r> r> insert* ;
M: sequence insert* ( seq position gb -- )
pick number? [
number-insert
] [
prepare-insert [ insert-elements ] 2keep swap length increment-gap-start
] if ;
: (delete*) ( gb -- )
dup gb-gap-end 1+ over set-gb-gap-end ?decrease ;
GENERIC: delete* ( pos gb -- )
M: gb delete* ( position gb -- )
tuck move-gap (delete*) ;
! -------- stack/queue operations -----------
: push-start ( obj gb -- ) 0 swap insert* ;
: push-end ( obj gb -- ) [ length ] keep insert* ;
: pop-elem ( position gb -- elem ) [ nth ] 2keep delete* ;
: pop-start ( gb -- elem ) 0 swap pop-elem ;
: pop-end ( gb -- elem ) [ length 1- ] keep pop-elem ;
: rotate ( n gb -- )
dup length 1 > [
swap dup 0 > [
[ dup [ pop-end ] keep push-start ]
] [
neg [ dup [ pop-start ] keep push-end ]
] if times drop
] [ 2drop ] if ;