factor/core/parser/parser.factor

! Copyright (C) 2005, 2008 Slava Pestov.
! See http://factorcode.org/license.txt for BSD license.
USING: arrays definitions generic assocs kernel math
namespaces prettyprint sequences strings vectors words
quotations inspector io.styles io combinators sorting
splitting math.parser effects continuations debugger 
io.files io.streams.string io.streams.lines vocabs
source-files classes hashtables compiler.errors compiler.units ;
IN: parser

TUPLE: lexer text line line-text line-length column ;

: next-line ( lexer -- )
    0 over set-lexer-column
    dup lexer-line over lexer-text ?nth over set-lexer-line-text
    dup lexer-line-text length over set-lexer-line-length
    dup lexer-line 1+ swap set-lexer-line ;

: <lexer> ( text -- lexer )
    0 { set-lexer-text set-lexer-line } lexer construct
    dup lexer-text empty? [ dup next-line ] unless ;

: location ( -- loc )
    file get lexer get lexer-line 2dup and
    [ >r source-file-path r> 2array ] [ 2drop f ] if ;

: save-location ( definition -- )
    location remember-definition ;

: save-class-location ( class -- )
    location remember-class ;

SYMBOL: parser-notes

t parser-notes set-global

: parser-notes? ( -- ? )
    parser-notes get "quiet" get not and ;

: file. ( file -- )
    [
        source-file-path <pathname> pprint
    ] [
        "<interactive>" write
    ] if* ":" write ;

: note. ( str -- )
    parser-notes? [
        file get file.
        lexer get [
            lexer-line number>string print
        ] [
            nl
        ] if*
        "Note: " write dup print
    ] when drop ;

: skip ( i seq ? -- n )
    over >r
    [ swap CHAR: \s eq? xor ] curry find* drop
    [ r> drop ] [ r> length ] if* ;

: change-column ( lexer quot -- )
    swap
    [ dup lexer-column swap lexer-line-text rot call ] keep
    set-lexer-column ; inline

GENERIC: skip-blank ( lexer -- )

M: lexer skip-blank ( lexer -- )
    [ t skip ] change-column ;

GENERIC: skip-word ( lexer -- )

M: lexer skip-word ( lexer -- )
    [
        2dup nth CHAR: " eq? [ drop 1+ ] [ f skip ] if
    ] change-column ;

: still-parsing? ( lexer -- ? )
    dup lexer-line swap lexer-text length <= ;

: still-parsing-line? ( lexer -- ? )
    dup lexer-column swap lexer-line-length < ;

: (parse-token) ( lexer -- str )
    [ lexer-column ] keep
    [ skip-word ] keep
    [ lexer-column ] keep
    lexer-line-text subseq ;

:  parse-token ( lexer -- str/f )
    dup still-parsing? [
        dup skip-blank
        dup still-parsing-line?
        [ (parse-token) ] [ dup next-line parse-token ] if
    ] [ drop f ] if ;

: scan ( -- str/f ) lexer get parse-token ;

TUPLE: bad-escape ;

: bad-escape ( -- * )
    \ bad-escape construct-empty throw ;

M: bad-escape summary drop "Bad escape code" ;

: escape ( escape -- ch )
    H{
        { CHAR: e  CHAR: \e }
        { CHAR: n  CHAR: \n }
        { CHAR: r  CHAR: \r }
        { CHAR: t  CHAR: \t }
        { CHAR: s  CHAR: \s }
        { CHAR: \s CHAR: \s }
        { CHAR: 0  CHAR: \0 }
        { CHAR: \\ CHAR: \\ }
        { CHAR: \" CHAR: \" }
    } at [ bad-escape ] unless* ;

: next-escape ( m str -- n ch )
    2dup nth CHAR: u =
    [ >r 1+ dup 6 + tuck r> subseq hex> ]
    [ over 1+ -rot nth escape ] if ;

: next-char ( m str -- n ch )
    2dup nth CHAR: \\ =
    [ >r 1+ r> next-escape ] [ over 1+ -rot nth ] if ;

: (parse-string) ( m str -- n )
    2dup nth CHAR: " =
    [ drop 1+ ] [ [ next-char , ] keep (parse-string) ] if ;

: parse-string ( -- str )
    lexer get [
        [ (parse-string) ] "" make swap
    ] change-column ;

TUPLE: parse-error file line col text ;

: <parse-error> ( msg -- error )
    file get
    lexer get
    { lexer-line lexer-column lexer-line-text } get-slots
    parse-error construct-boa
    [ set-delegate ] keep ;

: parse-dump ( error -- )
    dup parse-error-file file.
    dup parse-error-line number>string print
    dup parse-error-text dup string? [ print ] [ drop ] if
    parse-error-col 0 or CHAR: \s <string> write
    "^" print ;

M: parse-error error.
    dup parse-dump  delegate error. ;

SYMBOL: use
SYMBOL: in

: word/vocab% ( word -- )
    "(" % dup word-vocabulary % " " % word-name % ")" % ;

: shadow-warning ( new old -- )
    2dup eq? [
        2drop
    ] [
        [ word/vocab% " shadowed by " % word/vocab% ] "" make
        note.
    ] if ;

: shadow-warnings ( vocab vocabs -- )
    [
        swapd assoc-stack dup
        [ shadow-warning ] [ 2drop ] if
    ] curry assoc-each ;

: (use+) ( vocab -- )
    vocab-words use get 2dup shadow-warnings push ;

: use+ ( vocab -- )
    load-vocab (use+) ;

: add-use ( seq -- ) [ use+ ] each ;

: set-use ( seq -- )
    [ vocab-words ] map [ ] subset >vector use set ;

: check-vocab-string ( name -- name )
    dup string?
    [ "Vocabulary name must be a string" throw ] unless ;

: set-in ( name -- )
    check-vocab-string dup in set create-vocab (use+) ;

: create-in ( string -- word )
    in get create dup set-word dup save-location ;

TUPLE: unexpected want got ;

: unexpected ( want got -- * )
    \ unexpected construct-boa throw ;

PREDICATE: unexpected unexpected-eof
    unexpected-got not ;

: unexpected-eof ( word -- * ) f unexpected ;

: (parse-tokens) ( accum end -- accum )
    scan 2dup = [
        2drop
    ] [
        [ pick push (parse-tokens) ] [ unexpected-eof ] if*
    ] if ;

: parse-tokens ( end -- seq )
    100 <vector> swap (parse-tokens) >array ;

: CREATE ( -- word ) scan create-in ;

: CREATE-CLASS ( -- word )
    scan in get create
    dup save-class-location
    dup predicate-word dup set-word save-location ;

: word-restarts ( possibilities -- restarts )
    natural-sort [
        [ "Use the word " swap summary append ] keep
    ] { } map>assoc ;

TUPLE: no-word name ;

M: no-word summary
    drop "Word not found in current vocabulary search path" ;

: no-word ( name -- newword )
    dup \ no-word construct-boa
    swap words-named [ forward-reference? not ] subset
    word-restarts throw-restarts
    dup word-vocabulary (use+) ;

: check-forward ( str word -- word/f )
    dup forward-reference? [
        drop
        use get
        [ at ] with map [ ] subset
        [ forward-reference? not ] find nip
    ] [
        nip
    ] if ;

: search ( str -- word/f )
    dup use get assoc-stack check-forward ;

: scan-word ( -- word/number/f )
    scan dup [
        dup search [ ] [
            dup string>number [ ] [ no-word ] ?if
        ] ?if
    ] when ;

TUPLE: staging-violation word ;

: staging-violation ( word -- * )
    \ staging-violation construct-boa throw ;

M: staging-violation summary
    drop
    "A parsing word cannot be used in the same file it is defined in." ;

: execute-parsing ( word -- )
    new-definitions get [
        dupd first key? [ staging-violation ] when
    ] when*
    execute ;

: parse-step ( accum end -- accum ? )
    scan-word {
        { [ 2dup eq? ] [ 2drop f ] }
        { [ dup not ] [ drop unexpected-eof t ] }
        { [ dup delimiter? ] [ unexpected t ] }
        { [ dup parsing? ] [ nip execute-parsing t ] }
        { [ t ] [ pick push drop t ] }
    } cond ;

: (parse-until) ( accum end -- accum )
    dup >r parse-step [ r> (parse-until) ] [ r> drop ] if ;

: parse-until ( end -- vec )
    100 <vector> swap (parse-until) ;

: parsed ( accum obj -- accum ) over push ;

: with-parser ( lexer quot -- newquot )
    swap lexer set
    [ call >quotation ] [ <parse-error> rethrow ] recover ;

: (parse-lines) ( lexer -- quot )
    [ f parse-until ] with-parser ;

SYMBOL: lexer-factory

[ <lexer> ] lexer-factory set-global

: parse-lines ( lines -- quot )
    lexer-factory get call (parse-lines) ;

! Parsing word utilities
: parse-effect ( -- effect )
    ")" parse-tokens "(" over member? [
        "Stack effect declaration must not contain (" throw
    ] [
        { "--" } split1 dup [
            <effect>
        ] [
            "Stack effect declaration must contain --" throw
        ] if
    ] if ;

TUPLE: bad-number ;

: bad-number ( -- * ) \ bad-number construct-boa throw ;

: parse-base ( parsed base -- parsed )
    scan swap base> [ bad-number ] unless* parsed ;

: parse-literal ( accum end quot -- accum )
    >r parse-until r> call parsed ; inline

: parse-definition ( -- quot )
    \ ; parse-until >quotation ;

GENERIC: expected>string ( obj -- str )

M: f expected>string drop "end of input" ;
M: word expected>string word-name ;
M: string expected>string ;

M: unexpected error.
    "Expected " write
    dup unexpected-want expected>string write
    " but got " write
    unexpected-got expected>string print ;

M: bad-number summary
    drop "Bad number literal" ;

SYMBOL: bootstrap-syntax

: with-file-vocabs ( quot -- )
    [
        "scratchpad" in set
        { "syntax" "scratchpad" } set-use
        bootstrap-syntax get [ use get push ] when*
        call
    ] with-scope ; inline

SYMBOL: interactive-vocabs

{
    "arrays"
    "assocs"
    "combinators"
    "compiler.errors"
    "continuations"
    "debugger"
    "definitions"
    "editors"
    "generic"
    "help"
    "inspector"
    "io"
    "io.files"
    "kernel"
    "listener"
    "math"
    "memory"
    "namespaces"
    "prettyprint"
    "sequences"
    "slicing"
    "sorting"
    "strings"
    "syntax"
    "tools.annotations"
    "tools.crossref"
    "tools.memory"
    "tools.profiler"
    "tools.test"
    "tools.time"
    "vocabs"
    "vocabs.loader"
    "words"
    "scratchpad"
} interactive-vocabs set-global

: with-interactive-vocabs ( quot -- )
    [
        "scratchpad" in set
        interactive-vocabs get set-use
        call
    ] with-scope ; inline

: parse-fresh ( lines -- quot )
    [ parse-lines ] with-file-vocabs ;

: parsing-file ( file -- )
    "quiet" get [
        drop
    ] [
        "Loading " write <pathname> . flush
    ] if ;

: smudged-usage-warning ( usages removed -- )
    parser-notes? [
        "Warning: the following definitions were removed from sources," print
        "but are still referenced from other definitions:" print
        nl
        dup stack.
        nl
        "The following definitions need to be updated:" print
        nl
        over stack.
    ] when 2drop ;

: filter-moved ( assoc -- newassoc )
    [
        drop where dup [ first ] when
        file get source-file-path =
    ] assoc-subset ;

: removed-definitions ( -- definitions )
    new-definitions old-definitions
    [ get first2 union ] 2apply diff ;

: smudged-usage ( -- usages referenced removed )
    removed-definitions filter-moved keys [
        outside-usages
        [ empty? swap pathname? or not ] assoc-subset
        dup values concat prune swap keys
    ] keep ;

: forget-smudged ( -- )
    smudged-usage forget-all
    over empty? [ 2dup smudged-usage-warning ] unless 2drop ;

: finish-parsing ( lines quot -- )
    file get
    [ record-form ] keep
    [ record-modified ] keep
    [ record-definitions ] keep
    record-checksum ;

: parse-stream ( stream name -- quot )
    [
        [
            lines dup parse-fresh
            tuck finish-parsing
            forget-smudged
        ] with-source-file
    ] with-compilation-unit ;

: parse-file-restarts ( file -- restarts )
    "Load " swap " again" 3append t 2array 1array ;

: parse-file ( file -- quot )
    [
        [
            [ parsing-file ] keep
            [ ?resource-path <file-reader> ] keep
            parse-stream
        ] with-compiler-errors
    ] [
        over parse-file-restarts rethrow-restarts
        drop parse-file
    ] recover ;

: run-file ( file -- )
    [ [ parse-file call ] keep ] assert-depth drop ;

: ?run-file ( path -- )
    dup ?resource-path exists? [ run-file ] [ drop ] if ;

: bootstrap-file ( path -- )
    [ parse-file % ] [ run-file ] if-bootstrapping ;

: eval ( str -- )
    [ string-lines parse-fresh ] with-compilation-unit call ;

: eval>string ( str -- output )
    [
        parser-notes off
        [ [ eval ] keep ] try drop
    ] string-out ;