DTDs are a separate type now; all variables in xml.state

basis/xml/autoencoding/autoencoding.factor

@@ -10,8 +10,8 @@ IN: xml.autoencoding
 
 : start-utf16le ( -- tag )
     utf16le decode-input-if
-    CHAR: ? expect
-    0 expect check instruct ;
+    "?\0" expect
+    check instruct ;
 
 : 10xxxxxx? ( ch -- ? )
     -6 shift 3 bitand 2 = ;
@@ -36,10 +36,10 @@ IN: xml.autoencoding
 
 : skip-utf8-bom ( -- tag )
     "\u0000bb\u0000bf" expect utf8 decode-input
-    CHAR: < expect check make-tag ;
+    "<" expect check make-tag ;
 
 : decode-expecting ( encoding string -- tag )
-    [ decode-input-if next ] [ expect-string ] bi* check make-tag ;
+    [ decode-input-if next ] [ expect ] bi* check make-tag ;
 
 : start-utf16be ( -- tag )
     utf16be "<" decode-expecting ;

basis/xml/data/data-docs.factor

@@ -150,3 +150,52 @@ HELP: assure-name
 HELP: <simple-name>
 { $values { "string" string } { "name" name } }
 { $description "Converts a string into an XML name with an empty prefix and URL." } ;
+
+HELP: element-decl
+{ $class-description "Describes the class of element declarations, like <!ELEMENT  greeting (#PCDATA)>." } ;
+
+HELP: <element-decl>
+{ $values { "name" name } { "content-spec" string } { "element-decl" entity-decl } }
+{ $description "Creates an element declaration object, of the class " { $link element-decl } } ;
+
+HELP: attlist-decl
+{ $class-description "Describes the class of element declarations, like <!ATTLIST pre xml:space (preserve) #FIXED 'preserve'>." } ;
+
+HELP: <attlist-decl>
+{ $values { "name" name } { "att-defs" string } { "attlist-decl" attlist-decl } }
+{ $description "Creates an element declaration object, of the class " { $link attlist-decl } } ;
+
+HELP: entity-decl
+{ $class-description "Describes the class of element declarations, like <!ENTITY foo 'bar'>." } ;
+
+HELP: <entity-decl>
+{ $values { "name" name } { "def" string } { "pe?" "t or f" } { "entity-decl" entity-decl } }
+{ $description "Creates an entity declaration object, of the class " { $link entity-decl } ". The pe? slot should be t if the object is a DTD-internal entity, like <!ENTITY % foo 'bar'> and f if the object is like <!ENTITY foo 'bar'>, that is, it can be used outside of the DTD." } ;
+
+HELP: system-id
+{ $class-description "Describes the class of system identifiers within an XML DTD directive, such as <!DOCTYPE greeting " { $emphasis "SYSTEM 'hello.dtd'" } ">" } ;
+
+HELP: <system-id>
+{ $values { "system-literal" string } { "system-id" system-id } }
+{ $description "Constructs a " { $link system-id } " tuple." } ;
+
+HELP: public-id
+{ $class-description "Describes the class of public identifiers within an XML DTD directive, such as <!DOCTYPE open-hatch " { $emphasis "PUBLIC '-//Textuality//TEXT Standard open-hatch boilerplate//EN' 'http://www.textuality.com/boilerplate/OpenHatch.xml'" } ">" } ;
+
+HELP: <public-id>
+{ $values { "pubid-literal" string } { "system-literal" string } { "public-id" public-id } }
+{ $description "Constructs a " { $link system-id } " tuple." } ;
+
+HELP: notation-decl
+{ $class-description "Describes the class of element declarations, like <!NOTATION jpg SYSTEM './jpgviewer'>." } ;
+
+HELP: <notation-decl>
+{ $values { "name" string } { "id" id } { "notation-decl" notation-decl } }
+{ $description "Creates an notation declaration object, of the class " { $link notation-decl } "." } ;
+
+HELP: doctype-decl
+{ $class-description "Describes the class of doctype declarations." } ;
+
+HELP: <doctype-decl>
+{ $values { "name" name } { "external-id" id } { "internal-subset" sequence } { "doctype-decl" doctype-decl } }
+{ $description "Creates a new doctype declaration object, of the class " { $link doctype-decl } ". Only one of external-id or internal-subset will be non-null." } ;

basis/xml/data/data.factor

@@ -84,11 +84,13 @@ C: <comment> comment
 TUPLE: directive ;
 
 TUPLE: element-decl < directive
-    { name string } { content-spec string } ;
+    { name string }
+    { content-spec string } ;
 C: <element-decl> element-decl
 
 TUPLE: attlist-decl < directive
-    { name string } { att-defs string } ;
+    { name string }
+    { att-defs string } ;
 C: <attlist-decl> attlist-decl
 
 UNION: boolean t POSTPONE: f ;
@@ -107,13 +109,23 @@ C: <public-id> public-id
 
 UNION: id system-id public-id POSTPONE: f ;
 
+TUPLE: dtd
+    { directives sequence }
+    { entities assoc }
+    { parameter-entities assoc } ;
+C: <dtd> dtd
+
+UNION: dtd/f dtd POSTPONE: f ;
+
 TUPLE: doctype-decl < directive
     { name string }
     { external-id id }
-    { internal-subset sequence } ;
+    { internal-subset dtd/f } ;
 C: <doctype-decl> doctype-decl
 
-TUPLE: notation-decl < directive name id ;
+TUPLE: notation-decl < directive
+    { name string }
+    { id string } ;
 C: <notation-decl> notation-decl
 
 TUPLE: instruction { text string } ;

basis/xml/dtd/dtd.factor

@@ -2,12 +2,9 @@
 ! See http://factorcode.org/license.txt for BSD license.
 USING: xml.tokenize xml.data xml.state kernel sequences ascii
 fry xml.errors combinators hashtables namespaces xml.entities
-strings ;
+strings xml.name ;
 IN: xml.dtd
 
-: take-word ( -- string )
-    [ get-char blank? ] take-until ;
-
 : take-decl-contents ( -- first second )
     pass-blank take-word pass-blank ">" take-string ;
 
@@ -20,36 +17,15 @@ IN: xml.dtd
 : take-notation-decl ( -- notation-decl )
     take-decl-contents <notation-decl> ; 
 
-: take-until-one-of ( seps -- str sep )
-    '[ get-char _ member? ] take-until get-char ;
-
-: take-system-id ( -- system-id )
-    parse-quote <system-id> close ;
-
-: take-public-id ( -- public-id )
-    parse-quote parse-quote <public-id> close ;
-
 UNION: dtd-acceptable
     directive comment instruction ;
 
-: (take-external-id) ( token -- external-id )
-    pass-blank {
-        { "SYSTEM" [ take-system-id ] }
-        { "PUBLIC" [ take-public-id ] }
-        [ bad-external-id ]
-    } case ;
-
-: take-external-id ( -- external-id )
-    take-word (take-external-id) ;
-
-: only-blanks ( str -- )
-    [ blank? ] all? [ bad-decl ] unless ;
 : take-entity-def ( var -- entity-name entity-def )
     [
         take-word pass-blank get-char {
             { CHAR: ' [ parse-quote ] }
             { CHAR: " [ parse-quote ] }
-            [ drop take-external-id ]
+            [ drop take-external-id close ]
         } case
    ] dip '[ swap _ [ ?set-at ] change ] 2keep ;
 

basis/xml/elements/elements.factor

@@ -3,11 +3,11 @@
 USING: kernel namespaces xml.tokenize xml.state xml.name
 xml.data accessors arrays make xml.char-classes fry assocs sequences
 math xml.errors sets combinators io.encodings io.encodings.iana
-unicode.case xml.dtd strings ;
+unicode.case xml.dtd strings xml.entities ;
 IN: xml.elements
 
 : parse-attr ( -- )
-    parse-name pass-blank CHAR: = expect pass-blank
+    parse-name pass-blank "=" expect pass-blank
     t parse-quote* 2array , ;
 
 : start-tag ( -- name ? )
@@ -31,14 +31,14 @@ IN: xml.elements
 
 : end-tag ( name attrs-alist -- tag )
     tag-ns pass-blank get-char CHAR: / =
-    [ pop-ns <contained> next CHAR: > expect ]
+    [ pop-ns <contained> next ">" expect ]
     [ depth inc <opener> close ] if ;
 
 : take-comment ( -- comment )
-    "--" expect-string
+    "--" expect
     "--" take-string
     <comment>
-    CHAR: > expect ;
+    ">" expect ;
 
 : assure-no-extra ( seq -- )
     [ first ] map {
@@ -80,7 +80,7 @@ SYMBOL: string-input?
     string-input? get [ drop ] [ decode-input ] if ;
 
 : parse-prolog ( -- prolog )
-    pass-blank middle-tag "?>" expect-string
+    pass-blank middle-tag "?>" expect
     dup assure-no-extra prolog-attrs
     dup encoding>> dup "UTF-16" =
     [ drop ] [ name>encoding [ decode-input-if ] when* ] if
@@ -96,45 +96,45 @@ SYMBOL: string-input?
 
 : take-cdata ( -- string )
     depth get zero? [ bad-cdata ] when
-    "[CDATA[" expect-string "]]>" take-string ;
+    "[CDATA[" expect "]]>" take-string ;
 
 DEFER: make-tag ! Is this unavoidable?
 
 : expand-pe ( -- ) ; ! Make this run the contents of the pe within a DOCTYPE
 
-: (take-internal-subset) ( -- )
+: dtd-loop ( -- )
     pass-blank get-char {
         { CHAR: ] [ next ] }
         { CHAR: % [ expand-pe ] }
         { CHAR: < [
             next make-tag dup dtd-acceptable?
-            [ bad-doctype ] unless , (take-internal-subset)
+            [ bad-doctype ] unless , dtd-loop
         ] }
+        { f [ ] }
         [ 1string bad-doctype ]
     } case ;
 
-: take-internal-subset ( -- seq )
+: take-internal-subset ( -- dtd )
     [
-        H{ } pe-table set
+        H{ } clone pe-table set
         t in-dtd? set
-        (take-internal-subset)
-    ] { } make ;
+        dtd-loop
+        pe-table get
+    ] { } make swap extra-entities get swap <dtd> ;
 
-: nontrivial-doctype ( -- external-id internal-subset )
-    pass-blank get-char CHAR: [ = [
-        next take-internal-subset f swap close
-    ] [
-        " >" take-until-one-of {
-            { CHAR: \s [ (take-external-id) ] }
-            { CHAR: > [ only-blanks f ] }
-        } case f
-    ] if ;
+: take-optional-id ( -- id/f )
+    get-char "SP" member?
+    [ take-external-id ] [ f ] if ;
+
+: take-internal ( -- dtd/f )
+    get-char CHAR: [ =
+    [ next take-internal-subset ] [ f ] if ;
 
 : take-doctype-decl ( -- doctype-decl )
-    pass-blank " >" take-until-one-of {
-        { CHAR: \s [ nontrivial-doctype ] }
-        { CHAR: > [ f f ] }
-    } case <doctype-decl> ;
+    pass-blank take-name
+    pass-blank take-optional-id
+    pass-blank take-internal
+    <doctype-decl> close ;
 
 : take-directive ( -- doctype )
     take-name dup "DOCTYPE" =

basis/xml/entities/entities.factor

@@ -1,7 +1,7 @@
 ! Copyright (C) 2005, 2006 Daniel Ehrenberg
 ! See http://factorcode.org/license.txt for BSD license.
 USING: namespaces make kernel assocs sequences fry values
-io.files io.encodings.binary ;
+io.files io.encodings.binary xml.state ;
 IN: xml.entities
 
 : entities-out
@@ -37,7 +37,5 @@ IN: xml.entities
         { "quot"  CHAR: "  }
     } ;
 
-SYMBOL: extra-entities
-
 : with-entities ( entities quot -- )
     [ swap extra-entities set call ] with-scope ; inline

basis/xml/entities/html/html.factor

@@ -1,13 +1,13 @@
 ! Copyright (C) 2009 Daniel Ehrenberg.
 ! See http://factorcode.org/license.txt for BSD license.
 USING: assocs io.encodings.binary io.files kernel namespaces sequences
-values xml xml.entities ;
+values xml xml.entities accessors xml.state ;
 IN: xml.entities.html
 
 VALUE: html-entities
 
 : read-entities-file ( file -- table )
-    file>dtd nip ;
+    file>dtd entities>> ;
 
 : get-html ( -- table )
     { "lat1" "special" "symbol" } [

basis/xml/name/name.factor

@@ -2,7 +2,7 @@
 ! See http://factorcode.org/license.txt for BSD license.
 USING: kernel namespaces accessors xml.tokenize xml.data assocs
 xml.errors xml.char-classes combinators.short-circuit splitting
-fry xml.state sequences ;
+fry xml.state sequences combinators ascii ;
 IN: xml.name
 
 ! XML namespace processing: ns = namespace
@@ -74,3 +74,21 @@ SYMBOL: ns-stack
 : parse-name-starting ( string -- name )
     take-name append interpret-name ;
 
+: take-system-id ( -- system-id )
+    parse-quote <system-id> ;
+
+: take-public-id ( -- public-id )
+    parse-quote parse-quote <public-id> ;
+
+: (take-external-id) ( token -- external-id )
+    pass-blank {
+        { "SYSTEM" [ take-system-id ] }
+        { "PUBLIC" [ take-public-id ] }
+        [ bad-external-id ]
+    } case ;
+
+: take-word ( -- string )
+    [ get-char blank? ] take-until ;
+
+: take-external-id ( -- external-id )
+    take-word (take-external-id) ;

basis/xml/state/state.factor

@@ -23,3 +23,11 @@ SYMBOL: xml-stack
 SYMBOL: prolog-data
 
 SYMBOL: depth
+
+SYMBOL: interpolating?
+
+SYMBOL: in-dtd?
+
+SYMBOL: pe-table
+
+SYMBOL: extra-entities

basis/xml/tests/test.factor

@@ -49,10 +49,10 @@ SYMBOL: xml-file
 [ "foo" ] [ "<boo><![CDATA[foo]]></boo>" string>xml children>string ] unit-test
 [ "<!-- B+, B, or B--->" string>xml ] must-fail
 [ ] [ "<?xml version='1.0'?><!-- declarations for <head> & <body> --><foo/>" string>xml drop ] unit-test
-[ T{ element-decl f "br" "EMPTY" } ] [ "<!ELEMENT br EMPTY>" string>dtd drop second ] unit-test
-[ T{ element-decl f "p" "(#PCDATA|emph)*" } ] [ "<!ELEMENT p (#PCDATA|emph)*>" string>dtd drop second ] unit-test
-[ T{ element-decl f "%name.para;" "%content.para;" } ] [ "<!ELEMENT %name.para; %content.para;>" string>dtd drop second ] unit-test
-[ T{ element-decl f "container" "ANY" } ] [ "<!ELEMENT container ANY>" string>dtd drop second ] unit-test
+[ T{ element-decl f "br" "EMPTY" } ] [ "<!ELEMENT br EMPTY>" string>dtd directives>> first ] unit-test
+[ T{ element-decl f "p" "(#PCDATA|emph)*" } ] [ "<!ELEMENT p (#PCDATA|emph)*>" string>dtd directives>> first ] unit-test
+[ T{ element-decl f "%name.para;" "%content.para;" } ] [ "<!ELEMENT %name.para; %content.para;>" string>dtd directives>> first ] unit-test
+[ T{ element-decl f "container" "ANY" } ] [ "<!ELEMENT container ANY>" string>dtd directives>> first ] unit-test
 [ T{ doctype-decl f "foo" } ] [ "<!DOCTYPE foo>" string>xml-chunk first ] unit-test
 [ T{ doctype-decl f "foo" } ] [ "<!DOCTYPE foo >" string>xml-chunk first ] unit-test
 [ T{ doctype-decl f "foo" T{ system-id f "blah.dtd" } } ] [ "<!DOCTYPE foo SYSTEM 'blah.dtd'>" string>xml-chunk first ] unit-test

basis/xml/tests/xmode-dtd.factor

@@ -1,8 +1,8 @@
 ! Copyright (C) 2008 Slava Pestov.
 ! See http://factorcode.org/license.txt for BSD license.
-USING: xml io.encodings.utf8 io.files kernel tools.test ;
+USING: xml xml.data kernel tools.test ;
 IN: xml.tests
 
-[ ] [
-    "resource:basis/xmode/xmode.dtd" file>dtd 2drop
+[ t ] [
+    "resource:basis/xmode/xmode.dtd" file>dtd dtd?
 ] unit-test

basis/xml/tokenize/tokenize.factor

@@ -75,12 +75,7 @@ IN: xml.tokenize
     dup length rot length 1- - head
     get-char [ missing-close ] unless next ;
 
-: expect ( ch -- )
-    get-char 2dup = [ 2drop ] [
-        [ 1string ] bi@ expected
-    ] if next ;
-
-: expect-string ( string -- )
+: expect ( string -- )
     dup [ get-char next ] replicate 2dup =
     [ 2drop ] [ expected ] if ;
 
@@ -96,9 +91,6 @@ IN: xml.tokenize
         "x" ?head 16 10 ? base> ,
     ] [ parse-named-entity ] if ;
 
-SYMBOL: pe-table
-SYMBOL: in-dtd?
-
 : parse-pe ( -- )
     next CHAR: ; take-char dup next
     pe-table get at [ % ] [ no-entity ] ?if ;
@@ -131,7 +123,7 @@ SYMBOL: in-dtd?
     ] parse-char ;
 
 : close ( -- )
-    pass-blank CHAR: > expect ;
+    pass-blank ">" expect ;
 
 : normalize-quote ( str -- str )
     [ dup "\t\r\n" member? [ drop CHAR: \s ] when ] map ;

basis/xml/writer/writer.factor

@@ -136,10 +136,10 @@ M: public-id write-xml-chunk
     [ pubid-literal>> write "' '" write ]
     [ system-literal>> write "'" write ] bi ;
 
-: write-internal-subset ( seq -- )
+: write-internal-subset ( dtd -- )
     [
         "[" write indent
-        [ ?indent write-xml-chunk ] each
+        directives>> [ ?indent write-xml-chunk ] each
         unindent ?indent "]" write
     ] when* ;
 

basis/xml/xml.factor

@@ -164,21 +164,15 @@ TUPLE: pull-xml scope ;
 : file>xml ( filename -- xml )
     binary <file-reader> read-xml ;
 
-: (read-dtd) ( -- dtd )
-    ! should filter out blanks, throw error on non-dtd stuff
-    V{ } clone dup [ push ] curry sax-loop ;
-
-: read-dtd ( stream -- dtd entities )
+: read-dtd ( stream -- dtd )
     [
-        t in-dtd? set
         reset-prolog
         H{ } clone extra-entities set
-        (read-dtd)
-        extra-entities get
+        take-internal-subset
     ] with-state ;
 
-: file>dtd ( filename -- dtd entities )
+: file>dtd ( filename -- dtd )
     utf8 <file-reader> read-dtd ;
 
-: string>dtd ( string -- dtd entities )
+: string>dtd ( string -- dtd )
     <string-reader> read-dtd ;