Paste: parser for markup language

USING: accessors arrays assocs combinators fry io
io.encodings.utf8 io.files io.streams.string kernel lists
locals macros math math.order namespaces sequences splitting
strings unicode.categories urls.encoding vectors
xml.data xml.syntax xml.writer xmode.code2html
 ;
IN: t2t

! Predicates

: strip ( line -- line' )
    [ "\t " member? ] trim ;

: heading? ( string -- ? )
    [ f ]
    [
        [ first "=+" member? ]
        [ last  "=+" member? ] bi and 
    ] if-empty ;

: list? ( string -- ? )
    [ f ]
    [
        [ first "-+" member? ]
        [ heading? not ] bi and
    ] if-empty ;

: comment? ( line -- ? )
    first CHAR: % = ;

: table-row? ( line -- ? )
    strip
    [ f ]
    [
        [ first CHAR: | = ]
        [ last  CHAR: | = ] bi and 
    ] if-empty ;
    
! Parsing helpers

SYMBOL: lines

: eof? ( -- ? )
    lines get empty? ;

: skip-line ( -- )
    lines get
    [ "skip-line found EOF" throw ]
    [ 1 tail-slice lines set ] if-empty ;

: peek-line ( -- line )
    lines get
    [ "peek-line found EOF" throw ]
    [ first ] if-empty ;

! Parsing words

SYMBOL: ast

: add-node ( x -- )
    ast get swap suffix ast set ;

: parse-heading ( -- )
    "heading" add-node skip-line ;

: parse-list ( -- )
    "list" add-node skip-line 
    peek-line list? [ skip-line ] when ; ! example of peeking the next item

: parse-text ( -- )
    "text" add-node skip-line ;

: parse-lines ( -- )
    [ eof? not ]
    [
        {
            { [ peek-line empty? ]   [ "empty" add-node skip-line ] }
            { [ peek-line comment? ] [ skip-line ] }
            { [ peek-line heading? ] [ parse-heading ] }
            { [ peek-line list?    ] [ parse-list ] }
            [ parse-text ]
        } cond
    ] while ;

: parse ( filename -- ast )
    { } ast set
    utf8 file-lines lines set
    parse-lines
    ast get ;