arboretum 0.0.3 → 0.0.4

data/lib/arboretum/xml.rb

@@ -0,0 +1,169 @@
+module Arboretum
+  module XML
+    module IO
+      require 'ox'
+      require_relative 'doctree'
+      
+      class XMLParseException < StandardError
+        def initialize(msg="An error occurred while parsing XML")
+          super(msg)
+        end
+      end
+
+      class ArboretumIO
+        def self.read(file_path, style: :clean, type: :auto)
+          sax_parser = ArboretumSax.new(style)
+          type = file_path[(file_path.rindex(/\..+/))+1..-1].to_sym if type == :auto
+          File.open(file_path, 'r') do |f|
+            case type
+            when :xml, :xhtml
+              Ox.sax_parse(sax_parser, f, :skip => :skip_off)
+            when :html
+              Ox.sax_parse(sax_parser, f, :skip => :skip_off, :smart => true)
+            else
+              puts "Warning: Invalid file type `#{type}` given for read. Using `:xml` instead..."
+              Ox.sax_parse(sax_parser, f, :skip => :skip_off)
+            end
+          end
+          sax_parser.tree
+        end
+
+        def self.write(tree, file_path, style: :pretty, type: :auto)
+          type = file_path[(file_path.rindex(/\..+/))+1..-1].to_sym if type == :auto
+          if ![:xml, :xhtml, :html].include?(type)
+            puts "Warning: Invalid file type `#{type}` given for write. Using `:xml` instead..."
+            type = :xml
+          end
+          File.open(file_path, 'w') do |f|
+            f.write(tree.dump_markup(style, type))
+          end
+        end
+      end
+
+      class ArboretumSax < Ox::Sax
+        include Arboretum::DocTree::Elements
+
+        attr_accessor :tree
+
+        def initialize(style)
+          # Must initalize line to have access
+          @line = nil # Integer
+
+          # Root document element
+          root = DocRootElement.new # DocRootElement
+
+          # Intialize a tree for the parser to build on
+          @tree = Tree.new(root)    # Tree
+
+          # Contextual information for parsing process
+          @open_elements = [root]    # Array of Elements
+
+          # Read style
+          @style = style
+        end
+
+        def print_read_failure(method_name, args, reason="None given...")
+          puts "Reading of line #{@line} failed..."
+          puts "Reason: #{reason}"
+          puts "Method: #{method_name.to_s}"
+          puts "Args: #{args}"
+          puts "--------"
+        end
+
+        def start_element(name)
+          # Extract namespace if one exists
+          element_ns = nil
+          element_tag = name
+          tag_str = name.to_s
+          if tag_str.include?(':')
+            tag_split = tag_str.split(':')
+            raise XMLParseException if tag_split.length != 2
+            # Final information
+            element_ns = tag_split[0].to_sym
+            element_tag = tag_split[1].to_sym
+          end
+
+          # Add TaggedElement to tree
+          opened_element = TaggedElement.new(element_ns, element_tag)
+          @open_elements.last.append_child(opened_element)
+
+          # Open the element if paired
+          @open_elements.push(opened_element)
+        end
+        def attr(name, str)
+          # Get the most recently opened element
+          most_recent_open = @open_elements.last
+          # Append found values to attribute
+          most_recent_open.add_attr_value(name, str)
+        end
+        def comment(str)
+          # Add CommentElement to tree
+          comment_element = CommentElement.new(str)
+          @open_elements.last.append_child(comment_element)
+          # Do not open the element (no children)
+        end
+        def text(str)
+          if @style == :clean
+            # Compress, but preserve, whitespace
+            str.gsub!(/\s+/, ' ')
+
+            # Add TextElement to tree
+            text_element = TextElement.new(str)
+            @open_elements.last.append_child(text_element)
+            # Do not open the element (no children)
+
+          elsif @style == :preserve
+            # Add TextElement to tree
+            text_element = TextElement.new(str)
+            @open_elements.last.append_child(text_element)
+            # Do not open the element (no children)
+          end
+        end
+        def end_element(name)
+          # Close the most recent element/Get the closed element
+          closed_element = @open_elements.pop
+
+          # Partially ensure the correct element has been closed
+          if not "#{closed_element.namespaced_tag}".eql?(name.to_s)
+            args = method(__method__).parameters.map {|_,arg_name| binding.local_variable_get(arg_name)}
+            reason = "Non-matching element open/close tags: #{closed_element.tag} and #{name}"
+            self.print_read_failure(__method__, args, reason)
+          end
+        end
+
+        # Unimplemented calls w/debug output
+        def instruct(target)
+          args = method(__method__).parameters.map {|_,arg_name| binding.local_variable_get(arg_name)}
+          reason = "Unimplemented method"
+          self.print_read_failure(__method__, args, reason)
+        end
+        def end_instruct(target)
+          args = method(__method__).parameters.map {|_,arg_name| binding.local_variable_get(arg_name)}
+          reason = "Unimplemented method"
+          self.print_read_failure(__method__, args, reason)
+        end
+        def doctype(str)
+          args = method(__method__).parameters.map {|_,arg_name| binding.local_variable_get(arg_name)}
+          reason = "Unimplemented method"
+          self.print_read_failure(__method__, args, reason)
+        end
+        def cdata(str)
+          args = method(__method__).parameters.map {|_,arg_name| binding.local_variable_get(arg_name)}
+          reason = "Unimplemented method"
+          self.print_read_failure(__method__, args, reason)
+        end
+        def error(msg, line, column)
+          args = method(__method__).parameters.map {|_,arg_name| binding.local_variable_get(arg_name)}
+          reason = "A read error occured"
+          self.print_read_failure(__method__, args, reason)
+        end
+        def abort(name)
+          args = method(__method__).parameters.map {|_,arg_name| binding.local_variable_get(arg_name)}
+          reason = "Unimplemented method"
+          self.print_read_failure(__method__, args, reason)
+        end
+      end
+    end # IO
+
+  end # XML
+end # Arboretum

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: arboretum
 version: !ruby/object:Gem::Version
-  version: 0.0.3
+  version: 0.0.4
 platform: ruby
 authors:
 - tomjw64
@@ -37,6 +37,9 @@ extensions: []
 extra_rdoc_files: []
 files:
 - lib/arboretum.rb
+- lib/arboretum/doctree.rb
+- lib/arboretum/scandent.rb
+- lib/arboretum/xml.rb
 homepage: https://github.com/tomjw64/arboretum
 licenses:
 - MIT