parser.rb in fbo-0.1.0

- old
+ new
@@ -1,66 +1,63 @@
-require "strscan"
+require 'treetop'
+require 'fbo/node_extensions'
 
+base_path = ::File.expand_path(::File.dirname(__FILE__))
+Treetop.load(::File.join(base_path, './dump.treetop'))
+
 module FBO
   class Parser
-    TAG_PATTERN = /<\w+>/m
-    PATTERN = /<([A-Z]+)>.*<\/\1>/m
-
-    def initialize(file, tag_pattern: TAG_PATTERN)
+    def initialize(file)
       @file = file
-      @tag_pattern = tag_pattern
     end
 
-    def notices
-      if @notices.nil?
-        parse_notices(split_file_contents)
+    def parse(data = nil)
+      data ||= @file.contents
+
+      puts "Class = #{ data.class.name }"
+      if data.respond_to? :each
+        @tree = parse_collection(data)
+      else
+        @tree = parse_string(data)
       end
-      @notices
+      @tree
     end
 
     private
 
-    def split_file_contents
-      contents = @file.read
-      contents = cleanup_contents(contents)
-      scanner = StringScanner.new(contents)
-      text_notices = []
+    def parser
+      @parser ||= FBO::DumpParser.new
+    end
 
-      while !scanner.eos?
-        initial_tag = scanner.check_until(@tag_pattern)
-        if initial_tag
-          initial_tag.strip!
-        else
-          break
-        end
+    def parse_string(data)
+      tree = parser.parse(data)
+      if tree.nil?
+        line = parser.failure_line
+        column = parser.failure_column
+        reason = parser.failure_reason
+        raise Exception, "Parse error at line #{ line }, column #{ column }: #{ reason }"
+      end
+      clean_tree(tree)
+    end
 
-        text_notice = scanner.scan_until(/#{ closing_tag_for(initial_tag) }\s*/)
-        if text_notice
-          text_notices << text_notice.strip
+    def parse_collection(data)
+      super_tree = nil
+      data.each do |string|
+        tree = parse_string(string)
+        if super_tree
+          super_tree = FBO::Dump::DumpNode.new(super_tree.input, super_tree.interval,
+                                               super_tree.elements + tree.elements)
         else
-          break
+          super_tree = tree
         end
       end
-      text_notices
+      super_tree
     end
-
-    def parse_notices(text_notices)
-      @notices = []
-      text_notices.each do |text|
-        handler = FBO::Parser::HandlerSelector.select(text)
-        @notices << handler.parse(text)
-      end
-    end
-
-    def cleanup_contents(contents)
-      contents
-        .encode('UTF-16le', :invalid => :replace, :replace => '')
-        .encode('UTF-8')
-        .gsub(/\r\n/, "\n")
-        .gsub(/^M/, "")
-    end
-
-    def closing_tag_for(tag)
-      tag.sub(/</, "</")
+    
+    def clean_tree(node)
+      return if node.elements.nil?
+      node.elements.delete_if { |node| node.class.name == "Treetop::Runtime::SyntaxNode" }
+      node.elements.each { |node| clean_tree(node) }
+      node
     end
   end
 end