Fresh start

5 files changed, 168 insertions, 0 deletions

diff --git a/jni/ruby/lib/rexml/dtd/attlistdecl.rb b/jni/ruby/lib/rexml/dtd/attlistdecl.rb
new file mode 100644
index 0000000..25955ee
--- /dev/null
+++ b/jni/ruby/lib/rexml/dtd/attlistdecl.rb
@@ -0,0 +1,10 @@
+require "rexml/child"
+module REXML
+  module DTD
+    class AttlistDecl < Child
+      START = "<!ATTLIST"
+      START_RE = /^\s*#{START}/um
+      PATTERN_RE = /\s*(#{START}.*?>)/um
+    end
+  end
+end
diff --git a/jni/ruby/lib/rexml/dtd/dtd.rb b/jni/ruby/lib/rexml/dtd/dtd.rb
new file mode 100644
index 0000000..62317ba
--- /dev/null
+++ b/jni/ruby/lib/rexml/dtd/dtd.rb
@@ -0,0 +1,46 @@
+require "rexml/dtd/elementdecl"
+require "rexml/dtd/entitydecl"
+require "rexml/comment"
+require "rexml/dtd/notationdecl"
+require "rexml/dtd/attlistdecl"
+require "rexml/parent"
+
+module REXML
+  module DTD
+    class Parser
+      def Parser.parse( input )
+        case input
+        when String
+          parse_helper input
+        when File
+          parse_helper input.read
+        end
+      end
+
+      # Takes a String and parses it out
+      def Parser.parse_helper( input )
+        contents = Parent.new
+        while input.size > 0
+          case input
+          when ElementDecl.PATTERN_RE
+            match = $&
+            contents << ElementDecl.new( match )
+          when AttlistDecl.PATTERN_RE
+            matchdata = $~
+            contents << AttlistDecl.new( matchdata )
+          when EntityDecl.PATTERN_RE
+            matchdata = $~
+            contents << EntityDecl.new( matchdata )
+          when Comment.PATTERN_RE
+            matchdata = $~
+            contents << Comment.new( matchdata )
+          when NotationDecl.PATTERN_RE
+            matchdata = $~
+            contents << NotationDecl.new( matchdata )
+          end
+        end
+        contents
+      end
+    end
+  end
+end
diff --git a/jni/ruby/lib/rexml/dtd/elementdecl.rb b/jni/ruby/lib/rexml/dtd/elementdecl.rb
new file mode 100644
index 0000000..f90b27d
--- /dev/null
+++ b/jni/ruby/lib/rexml/dtd/elementdecl.rb
@@ -0,0 +1,17 @@
+require "rexml/child"
+module REXML
+  module DTD
+    class ElementDecl < Child
+      START = "<!ELEMENT"
+      START_RE = /^\s*#{START}/um
+      # PATTERN_RE = /^\s*(#{START}.*?)>/um
+      PATTERN_RE = /^\s*#{START}\s+((?:[:\w][-\.\w]*:)?[-!\*\.\w]*)(.*?)>/
+      #\s*((((["']).*?\5)|[^\/'">]*)*?)(\/)?>/um, true)
+
+      def initialize match
+        @name = match[1]
+        @rest = match[2]
+      end
+    end
+  end
+end
diff --git a/jni/ruby/lib/rexml/dtd/entitydecl.rb b/jni/ruby/lib/rexml/dtd/entitydecl.rb
new file mode 100644
index 0000000..a9286b2
--- /dev/null
+++ b/jni/ruby/lib/rexml/dtd/entitydecl.rb
@@ -0,0 +1,56 @@
+require "rexml/child"
+module REXML
+  module DTD
+    class EntityDecl < Child
+      START = "<!ENTITY"
+      START_RE = /^\s*#{START}/um
+      PUBLIC = /^\s*#{START}\s+(?:%\s+)?(\w+)\s+PUBLIC\s+((["']).*?\3)\s+((["']).*?\5)\s*>/um
+      SYSTEM = /^\s*#{START}\s+(?:%\s+)?(\w+)\s+SYSTEM\s+((["']).*?\3)(?:\s+NDATA\s+\w+)?\s*>/um
+      PLAIN = /^\s*#{START}\s+(\w+)\s+((["']).*?\3)\s*>/um
+      PERCENT = /^\s*#{START}\s+%\s+(\w+)\s+((["']).*?\3)\s*>/um
+      # <!ENTITY name SYSTEM "...">
+      # <!ENTITY name "...">
+      def initialize src
+        super()
+        md = nil
+        if src.match( PUBLIC )
+          md = src.match( PUBLIC, true )
+          @middle = "PUBLIC"
+          @content = "#{md[2]} #{md[4]}"
+        elsif src.match( SYSTEM )
+          md = src.match( SYSTEM, true )
+          @middle = "SYSTEM"
+          @content = md[2]
+        elsif src.match( PLAIN )
+          md = src.match( PLAIN, true )
+          @middle = ""
+          @content = md[2]
+        elsif src.match( PERCENT )
+          md = src.match( PERCENT, true )
+          @middle = ""
+          @content = md[2]
+        end
+        raise ParseException.new("failed Entity match", src) if md.nil?
+        @name = md[1]
+      end
+
+      def to_s
+        rv = "<!ENTITY #@name "
+        rv << "#@middle " if @middle.size > 0
+        rv << @content
+        rv
+      end
+
+      def write( output, indent )
+        indent( output, indent )
+        output << to_s
+      end
+
+      def EntityDecl.parse_source source, listener
+        md = source.match( PATTERN_RE, true )
+        thing = md[0].squeeze(" \t\n\r")
+        listener.send inspect.downcase, thing
+      end
+    end
+  end
+end
diff --git a/jni/ruby/lib/rexml/dtd/notationdecl.rb b/jni/ruby/lib/rexml/dtd/notationdecl.rb
new file mode 100644
index 0000000..17d1b9e
--- /dev/null
+++ b/jni/ruby/lib/rexml/dtd/notationdecl.rb
@@ -0,0 +1,39 @@
+require "rexml/child"
+module REXML
+  module DTD
+    class NotationDecl < Child
+      START = "<!NOTATION"
+      START_RE = /^\s*#{START}/um
+      PUBLIC = /^\s*#{START}\s+(\w[\w-]*)\s+(PUBLIC)\s+((["']).*?\4)\s*>/um
+      SYSTEM = /^\s*#{START}\s+(\w[\w-]*)\s+(SYSTEM)\s+((["']).*?\4)\s*>/um
+      def initialize src
+        super()
+        if src.match( PUBLIC )
+          md = src.match( PUBLIC, true )
+        elsif src.match( SYSTEM )
+          md = src.match( SYSTEM, true )
+        else
+          raise ParseException.new( "error parsing notation: no matching pattern", src )
+        end
+        @name = md[1]
+        @middle = md[2]
+        @rest = md[3]
+      end
+
+      def to_s
+        "<!NOTATION #@name #@middle #@rest>"
+      end
+
+      def write( output, indent )
+        indent( output, indent )
+        output << to_s
+      end
+
+      def NotationDecl.parse_source source, listener
+        md = source.match( PATTERN_RE, true )
+        thing = md[0].squeeze(" \t\n\r")
+        listener.send inspect.downcase, thing
+      end
+    end
+  end
+end