Regexp compiler used from literals

db4
Daniel Ehrenberg 2009-02-26 18:06:57 -06:00
parent 85432bd267
commit af2d380a7f
8 changed files with 130 additions and 120 deletions

View File

@ -1,34 +1,43 @@
! Copyright (C) 2009 Daniel Ehrenberg.
! See http://factorcode.org/license.txt for BSD license.
USING: regexp regexp.private regexp.classes kernel sequences regexp.negation
USING: regexp.classes kernel sequences regexp.negation
quotations regexp.minimize assocs fry math locals combinators
accessors words compiler.units ;
accessors words compiler.units kernel.private strings
sequences.private arrays regexp.matchers call ;
IN: regexp.compiler
: literals>cases ( literal-transitions -- case-body )
[ 1quotation ] assoc-map ;
: non-literals>dispatch ( non-literal-transitions -- quot )
[ [ '[ dup _ class-member? ] ] [ 1quotation ] bi* ] assoc-map
[ 3drop f ] suffix '[ _ cond ] ;
[ [ '[ dup _ class-member? ] ] [ '[ drop _ execute ] ] bi* ] assoc-map
[ 3drop ] suffix '[ _ cond ] ;
: expand-one-or ( or-class transition -- alist )
[ seq>> ] dip '[ _ 2array ] map ;
: expand-or ( alist -- new-alist )
[
first2 over or-class?
[ expand-one-or ] [ 2array 1array ] if
] map concat ;
: split-literals ( transitions -- case default )
! Convert disjunction of literals to literals. Also maybe small ranges.
>alist [ first integer? ] partition
>alist expand-or [ first integer? ] partition
[ literals>cases ] [ non-literals>dispatch ] bi* ;
USING: kernel.private strings sequences.private ;
:: step ( index str case-body final? -- match? )
:: step ( last-match index str case-body final? -- last-index/f )
final? index last-match ?
index str bounds-check? [
index 1+ str
index str nth-unsafe
case-body case
] [ final? ] if ; inline
] when ; inline
: transitions>quot ( transitions final-state? -- quot )
[ split-literals suffix ] dip
'[ { array-capacity string } declare _ _ step ] ;
'[ _ _ step ] ;
! '[ { array-capacity string } declare _ _ step ] ;
: word>quot ( word dfa -- quot )
[ transitions>> at ]
@ -39,7 +48,8 @@ USING: kernel.private strings sequences.private ;
'[
[
dup _ word>quot
(( index string -- ? )) define-declared
(( last-match index string -- ? ))
define-declared
] each
] with-compilation-unit ;
@ -59,7 +69,13 @@ USING: kernel.private strings sequences.private ;
states>words [ states>code ] keep start-state>> ;
: run-regexp ( string word -- ? )
[ 0 ] 2dip execute ; inline
[ f 0 ] 2dip execute ; inline
: regexp>quotation ( regexp -- quot )
compile-regexp dfa>> dfa>word '[ _ run-regexp ] ;
: dfa>quotation ( dfa -- quot )
dfa>word '[ _ run-regexp ] ;
TUPLE: quot-matcher quot ;
C: <quot-matcher> quot-matcher
M: quot-matcher match-index
quot>> call( string -- i/f ) ;

View File

@ -0,0 +1,61 @@
! Copyright (C) 2008, 2009 Daniel Ehrenberg, Doug Coleman.
! See http://factorcode.org/license.txt for BSD license.
USING: kernel sequences math splitting make fry ;
IN: regexp.matchers
! For now, a matcher is just something with a method to do the
! equivalent of match.
! matcher protocol:
GENERIC: match-index ( string matcher -- index/f )
: match ( string matcher -- slice/f )
dupd match-index [ head-slice ] [ drop f ] if* ;
: matches? ( string matcher -- ? )
dupd match-index
[ swap length = ] [ drop f ] if* ;
: match-head ( string matcher -- end/f ) match [ length ] [ f ] if* ;
: match-at ( string m matcher -- n/f finished? )
[
2dup swap length > [ 2drop f f ] [ tail-slice t ] if
] dip swap [ match-head f ] [ 2drop f t ] if ;
: match-range ( string m matcher -- a/f b/f )
3dup match-at over [
drop nip rot drop dupd +
] [
[ 3drop drop f f ] [ drop [ 1+ ] dip match-range ] if
] if ;
: first-match ( string matcher -- slice/f )
dupd 0 swap match-range rot over [ <slice> ] [ 3drop f ] if ;
: re-cut ( string matcher -- end/f start )
dupd first-match
[ split1-slice swap ] [ "" like f swap ] if* ;
<PRIVATE
: (re-split) ( string matcher -- )
over [ [ re-cut , ] keep (re-split) ] [ 2drop ] if ;
PRIVATE>
: re-split ( string matcher -- seq )
[ (re-split) ] { } make ;
: re-replace ( string matcher replacement -- result )
[ re-split ] dip join ;
: next-match ( string matcher -- end/f match/f )
dupd first-match dup
[ [ split1-slice nip ] keep ] [ 2drop f f ] if ;
: all-matches ( string matcher -- seq )
[ dup ] swap '[ _ next-match ] [ ] produce nip harvest ;
: count-matches ( string matcher -- n )
all-matches length ;

View File

@ -1,6 +1,7 @@
! Copyright (C) 2009 Daniel Ehrenberg
! See http://factorcode.org/license.txt for BSD license.
USING: tools.test regexp.minimize assocs regexp accessors regexp.transition-tables ;
USING: tools.test regexp.minimize assocs regexp regexp.syntax
accessors regexp.transition-tables ;
IN: regexp.minimize.tests
[ t ] [ 1 2 H{ { { 1 2 } t } } same-partition? ] unit-test

View File

@ -1,6 +1,6 @@
! Copyright (C) 2008, 2009 Doug Coleman, Daniel Ehrenberg.
! See http://factorcode.org/license.txt for BSD license.
USING: kernel strings help.markup help.syntax ;
USING: kernel strings help.markup help.syntax regexp.matchers ;
IN: regexp
ABOUT: "regexp"

View File

@ -1,5 +1,5 @@
USING: regexp tools.test kernel sequences regexp.parser
regexp.traversal eval strings multiline accessors ;
regexp.traversal eval strings multiline accessors regexp.matchers ;
IN: regexp-tests
\ <regexp> must-infer
@ -241,8 +241,8 @@ IN: regexp-tests
[ t ] [ "abc" <reversed> R/ abc/r matches? ] unit-test
[ t ] [ "abc" <reversed> R/ a[bB][cC]/r matches? ] unit-test
[ t ] [ "xabc" R/ abc/ match-from-end >boolean ] unit-test
[ t ] [ "xabc" R/ a[bB][cC]/ match-from-end >boolean ] unit-test
[ t ] [ "xabc" R/ abc/ <reverse-matcher> match >boolean ] unit-test
[ t ] [ "xabc" R/ a[bB][cC]/ <reverse-matcher> match >boolean ] unit-test
[ t ] [ "s@f" "[a-z.-]@[a-z]" <regexp> matches? ] unit-test
[ f ] [ "a" "[a-z.-]@[a-z]" <regexp> matches? ] unit-test

View File

@ -5,26 +5,29 @@ assocs prettyprint.backend prettyprint.custom make lexer
namespaces parser arrays fry locals regexp.minimize
regexp.parser regexp.nfa regexp.dfa regexp.traversal
regexp.transition-tables splitting sorting regexp.ast
regexp.negation ;
regexp.negation regexp.matchers regexp.compiler ;
IN: regexp
TUPLE: regexp
{ raw read-only }
{ parse-tree read-only }
{ options read-only }
dfa reverse-dfa ;
dfa reverse-dfa dfa-quot ;
: make-regexp ( string ast -- regexp )
f f <options> f f regexp boa ; foldable
f f <options> f f f regexp boa ; foldable
! Foldable because, when the dfa slot is set,
! it'll be set to the same thing regardless of who sets it
: <optioned-regexp> ( string options -- regexp )
[ dup parse-regexp ] [ string>options ] bi*
f f regexp boa ;
f f f regexp boa ;
: <regexp> ( string -- regexp ) "" <optioned-regexp> ;
TUPLE: reverse-matcher regexp ;
C: <reverse-matcher> reverse-matcher
<PRIVATE
: get-ast ( regexp -- ast )
@ -33,76 +36,24 @@ TUPLE: regexp
: compile-regexp ( regexp -- regexp )
dup '[ [ _ get-ast ast>dfa ] unless* ] change-dfa ;
: compile-dfa-quot ( regexp -- regexp )
dup '[ [ _ compile-regexp dfa>> dfa>quotation ] unless* ] change-dfa-quot ;
: <reversed-option> ( ast -- reversed )
"r" string>options <with-options> ;
: compile-reverse ( regexp -- regexp )
dup '[ [ _ get-ast <reversed-option> ast>dfa ] unless* ] change-reverse-dfa ;
: (match) ( string regexp -- dfa-traverser )
compile-regexp dfa>> <dfa-traverser> do-match ;
M: regexp match-index ( string regexp -- index/f )
dup dfa-quot>>
[ <quot-matcher> ]
[ compile-regexp dfa>> <dfa-matcher> ] ?if
match-index ;
: (match-reversed) ( string regexp -- dfa-traverser )
[ <reversed> ] [ compile-reverse reverse-dfa>> ] bi*
<dfa-traverser> do-match ;
PRIVATE>
: match ( string regexp -- slice/f )
(match) return-match ;
: match-from-end ( string regexp -- slice/f )
(match-reversed) return-match ;
: matches? ( string regexp -- ? )
dupd match
[ [ length ] bi@ = ] [ drop f ] if* ;
: match-head ( string regexp -- end/f ) match [ length ] [ f ] if* ;
: match-at ( string m regexp -- n/f finished? )
[
2dup swap length > [ 2drop f f ] [ tail-slice t ] if
] dip swap [ match-head f ] [ 2drop f t ] if ;
: match-range ( string m regexp -- a/f b/f )
3dup match-at over [
drop nip rot drop dupd +
] [
[ 3drop drop f f ] [ drop [ 1+ ] dip match-range ] if
] if ;
: first-match ( string regexp -- slice/f )
dupd 0 swap match-range rot over [ <slice> ] [ 3drop f ] if ;
: re-cut ( string regexp -- end/f start )
dupd first-match
[ split1-slice swap ] [ "" like f swap ] if* ;
<PRIVATE
: (re-split) ( string regexp -- )
over [ [ re-cut , ] keep (re-split) ] [ 2drop ] if ;
PRIVATE>
: re-split ( string regexp -- seq )
[ (re-split) ] { } make ;
: re-replace ( string regexp replacement -- result )
[ re-split ] dip join ;
: next-match ( string regexp -- end/f match/f )
dupd first-match dup
[ [ split1-slice nip ] keep ] [ 2drop f f ] if ;
: all-matches ( string regexp -- seq )
[ dup ] swap '[ _ next-match ] [ ] produce nip harvest ;
: count-matches ( string regexp -- n )
all-matches length ;
<PRIVATE
M: reverse-matcher match-index ( string regexp -- index/f )
[ <reversed> ] [ regexp>> compile-reverse reverse-dfa>> ] bi*
<dfa-traverser> do-match match-index>> ;
: find-regexp-syntax ( string -- prefix suffix )
{
@ -131,7 +82,7 @@ PRIVATE>
: parsing-regexp ( accum end -- accum )
lexer get [ take-until ] [ parse-noblank-token ] bi
<optioned-regexp> compile-regexp parsed ;
<optioned-regexp> compile-dfa-quot parsed ;
PRIVATE>

View File

@ -1,7 +1,7 @@
! Copyright (C) 2008 Doug Coleman.
! See http://factorcode.org/license.txt for BSD license.
USING: accessors assocs combinators kernel math
quotations sequences regexp.classes fry arrays
quotations sequences regexp.classes fry arrays regexp.matchers
combinators.short-circuit prettyprint regexp.nfa ;
IN: regexp.traversal
@ -9,16 +9,14 @@ TUPLE: dfa-traverser
dfa-table
current-state
text
start-index current-index
matches ;
current-index
match-index ;
: <dfa-traverser> ( text dfa -- match )
dfa-traverser new
swap [ start-state>> >>current-state ] [ >>dfa-table ] bi
swap >>text
0 >>start-index
0 >>current-index
V{ } clone >>matches ;
0 >>current-index ;
: final-state? ( dfa-traverser -- ? )
[ current-state>> ]
@ -33,25 +31,11 @@ TUPLE: dfa-traverser
[ end-of-text? ]
} 1|| ;
: save-final-state ( dfa-straverser -- )
[ current-index>> ] [ matches>> ] bi push ;
: save-final-state ( dfa-traverser -- dfa-traverser )
dup current-index>> >>match-index ;
: match-done? ( dfa-traverser -- ? )
dup final-state? [
dup save-final-state
] when text-finished? ;
: text-character ( dfa-traverser n -- ch )
[ text>> ] swap '[ current-index>> _ + ] bi nth ;
: previous-text-character ( dfa-traverser -- ch )
-1 text-character ;
: current-text-character ( dfa-traverser -- ch )
0 text-character ;
: next-text-character ( dfa-traverser -- ch )
1 text-character ;
dup final-state? [ save-final-state ] when text-finished? ;
: increment-state ( dfa-traverser state -- dfa-traverser )
>>current-state
@ -79,10 +63,7 @@ TUPLE: dfa-traverser
[ increment-state do-match ] when*
] unless ;
: return-match ( dfa-traverser -- slice/f )
dup matches>>
[ drop f ]
[
[ [ start-index>> ] [ text>> ] bi ]
[ peek ] bi* swap <slice>
] if-empty ;
TUPLE: dfa-matcher dfa ;
C: <dfa-matcher> dfa-matcher
M: dfa-matcher match-index
dfa>> <dfa-traverser> do-match match-index>> ;

View File

@ -1,7 +1,7 @@
! Copyright (C) 2008 Slava Pestov.
! See http://factorcode.org/license.txt for BSD license.
USING: accessors regexp prettyprint io io.encodings.ascii
io.files kernel sequences assocs namespaces ;
USING: accessors regexp.matchers prettyprint io io.encodings.ascii
io.files kernel sequences assocs namespaces regexp ;
IN: benchmark.regex-dna
! Based on http://shootout.alioth.debian.org/gp4/benchmark.php?test=regexdna&lang=ruby&id=1