Clean up parser combinators

release
Slava Pestov 2007-11-25 17:07:32 -05:00
parent 7925cf2266
commit 12599e03c4
1 changed files with 153 additions and 151 deletions

View File

@ -1,22 +1,23 @@
! Copyright (C) 2004 Chris Double.
! See http://factorcode.org/license.txt for BSD license.
USING: lazy-lists promises kernel sequences strings math io
arrays namespaces splitting ;
USING: lazy-lists promises kernel sequences strings math
arrays splitting ;
IN: parser-combinators
! Parser combinator protocol
GENERIC: (parse) ( input parser -- list )
GENERIC: parse ( input parser -- list )
M: promise (parse) ( input parser -- list )
force (parse) ;
: parse ( input parser -- promise )
(parse) ;
M: promise parse ( input parser -- list )
force parse ;
TUPLE: parse-result parsed unparsed ;
: parse-1 ( input parser -- result )
parse car parse-result-parsed ;
parse dup nil? [
"Parse error" throw
] [
car parse-result-parsed
] if ;
C: <parse-result> parse-result
@ -24,7 +25,7 @@ TUPLE: token-parser string ;
C: token token-parser ( string -- parser )
M: token-parser (parse) ( input parser -- list )
M: token-parser parse ( input parser -- list )
token-parser-string swap over ?head-slice [
<parse-result> 1list
] [
@ -35,7 +36,7 @@ TUPLE: satisfy-parser quot ;
C: satisfy satisfy-parser ( quot -- parser )
M: satisfy-parser (parse) ( input parser -- list )
M: satisfy-parser parse ( input parser -- list )
#! A parser that succeeds if the predicate,
#! when passed the first character in the input, returns
#! true.
@ -56,7 +57,7 @@ TUPLE: epsilon-parser ;
C: epsilon epsilon-parser ( -- parser )
M: epsilon-parser (parse) ( input parser -- list )
M: epsilon-parser parse ( input parser -- list )
#! A parser that parses the empty string. It
#! does not consume any input and always returns
#! an empty list as the parse tree with the
@ -67,7 +68,7 @@ TUPLE: succeed-parser result ;
C: succeed succeed-parser ( result -- parser )
M: succeed-parser (parse) ( input parser -- list )
M: succeed-parser parse ( input parser -- list )
#! A parser that always returns 'result' as a
#! successful parse with no input consumed.
succeed-parser-result swap <parse-result> 1list ;
@ -76,7 +77,7 @@ TUPLE: fail-parser ;
C: fail fail-parser ( -- parser )
M: fail-parser (parse) ( input parser -- list )
M: fail-parser parse ( input parser -- list )
#! A parser that always fails and returns
#! an empty list of successes.
2drop nil ;
@ -88,7 +89,7 @@ TUPLE: and-parser parsers ;
>r and-parser-parsers r> add
] [
2array
] if \ and-parser construct-boa ;
] if and-parser construct-boa ;
: and-parser-parse ( list p1 -- list )
swap [
@ -100,18 +101,19 @@ TUPLE: and-parser parsers ;
] lmap-with
] lmap-with lconcat ;
M: and-parser (parse) ( input parser -- list )
M: and-parser parse ( input parser -- list )
#! Parse 'input' by sequentially combining the
#! two parsers. First parser1 is applied to the
#! input then parser2 is applied to the rest of
#! the input strings from the first parser.
and-parser-parsers unclip swapd parse [ [ and-parser-parse ] reduce ] 2curry promise ;
and-parser-parsers unclip swapd parse
[ [ and-parser-parse ] reduce ] 2curry promise ;
TUPLE: or-parser p1 p2 ;
C: <|> or-parser ( parser1 parser2 -- parser )
M: or-parser (parse) ( input parser1 -- list )
M: or-parser parse ( input parser1 -- list )
#! Return the combined list resulting from the parses
#! of parser1 and parser2 being applied to the same
#! input. This implements the choice parsing operator.
@ -130,7 +132,7 @@ TUPLE: sp-parser p1 ;
#! calling the original parser.
C: sp sp-parser ( p1 -- parser )
M: sp-parser (parse) ( input parser -- list )
M: sp-parser parse ( input parser -- list )
#! Skip all leading whitespace from the input then call
#! the parser on the remaining input.
>r left-trim-slice r> sp-parser-p1 parse ;
@ -139,7 +141,7 @@ TUPLE: just-parser p1 ;
C: just just-parser ( p1 -- parser )
M: just-parser (parse) ( input parser -- result )
M: just-parser parse ( input parser -- result )
#! Calls the given parser on the input removes
#! from the results anything where the remaining
#! input to be parsed is not empty. So ensures a
@ -150,7 +152,7 @@ TUPLE: apply-parser p1 quot ;
C: <@ apply-parser ( parser quot -- parser )
M: apply-parser (parse) ( input parser -- result )
M: apply-parser parse ( input parser -- result )
#! Calls the parser on the input. For each successfull
#! parse the quot is call with the parse result on the stack.
#! The result of that quotation then becomes the new parse result.
@ -166,14 +168,13 @@ TUPLE: some-parser p1 ;
C: some some-parser ( p1 -- parser )
M: some-parser (parse) ( input parser -- result )
M: some-parser parse ( input parser -- result )
#! Calls the parser on the input, guarantees
#! the parse is complete (the remaining input is empty),
#! picks the first solution and only returns the parse
#! tree since the remaining input is empty.
some-parser-p1 just parse-1 ;
: <& ( parser1 parser2 -- parser )
#! Same as <&> except discard the results of the second parser.
<&> [ first ] <@ ;
@ -184,15 +185,15 @@ M: some-parser (parse) ( input parser -- result )
: <:&> ( parser1 parser2 -- result )
#! Same as <&> except flatten the result.
<&> [ dup second swap first [ % , ] { } make ] <@ ;
<&> [ first2 add ] <@ ;
: <&:> ( parser1 parser2 -- result )
#! Same as <&> except flatten the result.
<&> [ dup second swap first [ , % ] { } make ] <@ ;
<&> [ first2 swap add* ] <@ ;
: <:&:> ( parser1 parser2 -- result )
#! Same as <&> except flatten the result.
<&> [ dup second swap first [ % % ] { } make ] <@ ;
<&> [ first2 append ] <@ ;
LAZY: <*> ( parser -- parser )
dup <*> <&:> { } succeed <|> ;
@ -208,10 +209,11 @@ LAZY: <?> ( parser -- parser )
[ 1array ] <@ f succeed <|> ;
TUPLE: only-first-parser p1 ;
LAZY: only-first ( parser -- parser )
\ only-first-parser construct-boa ;
M: only-first-parser (parse) ( input parser -- list )
LAZY: only-first ( parser -- parser )
only-first-parser construct-boa ;
M: only-first-parser parse ( input parser -- list )
#! Transform a parser into a parser that only yields
#! the first possibility.
only-first-parser-p1 parse 1 swap ltake ;