ruby_docx_beta 0.1.0

data/lib/ruby_docx/elements.rb

@@ -0,0 +1,54 @@
+require "ruby_docx/elements/element"
+require "ruby_docx/elements/equation"
+require "ruby_docx/elements/image"
+require "ruby_docx/elements/drawing"
+require "ruby_docx/elements/ole_object"
+require "ruby_docx/elements/style"
+require "ruby_docx/elements/table_style"
+require "ruby_docx/elements/table_row"
+require "ruby_docx/elements/table_column"
+require "ruby_docx/elements/table_cell"
+require "ruby_docx/elements/table"
+
+require "ruby_docx/elements/text"
+require "ruby_docx/elements/text_run"
+require "ruby_docx/elements/link"
+require "ruby_docx/elements/bookmark"
+require "ruby_docx/elements/paragraph"
+require "ruby_docx/elements/break_line"
+
+module RubyDocx::Elements
+
+  class Parser
+    ELEMENTS_ALIAS = {
+      "pPr" => RubyDocx::Elements::Style,
+      "rPr" => RubyDocx::Elements::Style,
+      "tblPr" => RubyDocx::Elements::TableStyle,
+      "tcPr" => RubyDocx::Elements::TableStyle,
+      "pict" => RubyDocx::Elements::Image,
+      "drawing" => RubyDocx::Elements::Drawing,
+      "p" => RubyDocx::Elements::Paragraph,
+      "r" => RubyDocx::Elements::TextRun,
+      "t" => RubyDocx::Elements::Text,
+      "hyperlink" => RubyDocx::Elements::Link,
+      "bookmarkStart" => RubyDocx::Elements::Bookmark,
+      "tbl" => RubyDocx::Elements::Table,
+      "tr" => RubyDocx::Elements::TableRow,
+      "tc" => RubyDocx::Elements::TableCell,
+      "tblGrid" => RubyDocx::Elements::TableColumn,
+      "object" => RubyDocx::Elements::OleObject,
+      # "bookmarkEnd" => RubyDocx::Elements::Link,
+      "br" => RubyDocx::Elements::BreakLine,
+      "oMathPara" => RubyDocx::Elements::Equation,
+    }
+
+    def self.parse_by_name(doc, c_node)
+      if ELEMENTS_ALIAS.keys.index(c_node.name.to_s)
+        ELEMENTS_ALIAS[c_node.name.to_s].new(doc, c_node)
+      end
+
+    end
+
+  end
+
+end

data/lib/ruby_docx/elements/bookmark.rb

@@ -0,0 +1,42 @@
+module RubyDocx::Elements
+  class Bookmark < Element
+    attr_reader :anchor
+
+    def element
+      @element ||= @node.children.first
+    end
+
+    def anchor
+      return @anchor if @anchor
+      return "" if @node.attributes["name"].to_s.size <= 0
+
+      name = @node.attributes["name"].value.to_s
+
+      if name == "_GoBack"
+        nil
+      else
+        @anchor = name
+      end
+    end
+
+    def anchor=(an)
+      @anchor = an
+    end
+
+    def replace(an)
+      self.anchor = an
+    end
+
+    def to_html
+      if self.anchor
+        "<span name=\"#{self.anchor}\"></span>"
+      else
+        ""
+      end
+    end
+
+    def to_s
+      ""
+    end
+  end
+end

data/lib/ruby_docx/elements/break_line.rb

@@ -0,0 +1,13 @@
+module RubyDocx::Elements
+  class BreakLine < Element
+
+    def to_html
+      "<br>"
+    end
+
+    def to_s
+      "\n"
+    end
+
+  end
+end

data/lib/ruby_docx/elements/drawing.rb

@@ -0,0 +1,76 @@
+require 'mimemagic'
+require "base64"
+
+module RubyDocx::Elements
+  class Drawing < Element
+    attr_accessor :path, :zip, :link
+    attr_reader :width, :height
+
+    def data
+      if self.path
+        @zip.read("word/#{@path}")
+      else
+
+      end
+
+    end
+
+    def relation_id
+      element = @node.xpath(".//pic:blipFill/a:blip", "a" => "http://schemas.openxmlformats.org/drawingml/2006/main", "pic" => "http://schemas.openxmlformats.org/drawingml/2006/picture").first
+      if element && element.attributes.keys.index("embed")
+        element.attributes["embed"].value
+      else
+        nil
+      end
+    end
+
+    def replace(lnk)
+      @link = lnk
+    end
+
+    def style
+      element = @node.xpath(".//v:shape").first
+
+      if element && element.attributes.keys.index("style")
+        element.attributes["style"].value
+      else
+        nil
+      end
+    end
+
+    def title
+      element = @node.xpath(".//v:imagedata").first
+      if element && element.attributes.keys.index("title")
+        element.attributes["title"].value
+      else
+        nil
+      end
+    end
+
+    def mime
+      MimeMagic.by_magic(self.data).type
+    end
+
+    def base64_data
+      "data:#{self.mime};base64,#{Base64.strict_encode64(self.data)}"
+    end
+
+    def to_html
+      if @link.to_s.size > 0
+        "<img src='#{@link}' alt=\"#{self.title}\" style=\"#{self.style}\" />"
+      else
+        "<img src='#{self.base64_data}' alt=\"#{self.title}\" style=\"#{self.style}\" />"
+      end
+
+    end
+
+    def save(path)
+      if @zip && @path
+        file = File.new(path, "wb")
+        file.write(self.data)
+        file.close
+      end
+    end
+
+  end
+end

data/lib/ruby_docx/elements/element.rb

@@ -0,0 +1,53 @@
+
+module RubyDocx::Elements
+  class Element
+
+    attr_reader :node, :doc, :style, :grid
+
+
+    def initialize(doc, node)
+      @node = node
+      @doc = doc
+
+      if self.node.name.to_s == "pict"
+        @doc.setup_image(self)
+      elsif self.node.name.to_s == "drawing"
+        @doc.setup_drawing(self)
+      elsif self.node.name.to_s == "object"
+        @doc.setup_object(self)
+      end
+
+      self.parse_elements
+    end
+
+    def to_s
+      @node.to_s
+    end
+
+    def to_xml
+      @node.to_s
+    end
+
+    def elements
+      @elements ||= @node.children.map do |c_node|
+        v = RubyDocx::Elements::Parser.parse_by_name(@doc, c_node)
+
+        if ["pPr", "rPr", "tblPr", "tcPr"].index(c_node.name.to_s)
+          @style = v
+          nil
+        elsif ["tblGrid"].index(c_node.name.to_s)
+          @grid = v
+          nil
+        else
+          v
+        end
+      end.compact
+    end
+
+    alias :parse_elements :elements
+
+    def inspect
+      "#<#{self.class} @node=#{@node.name}>"
+    end
+  end
+end

data/lib/ruby_docx/elements/equation.rb

@@ -0,0 +1,51 @@
+require 'texmath'
+require 'calculus'
+
+module RubyDocx::Elements
+  class Equation < Element
+
+    def to_omml
+      ele = @node.xpath(".//m:oMath").first
+      ele.to_s
+    end
+
+    def to_latex
+       TeXMath.convert("#{self.to_omml}", :from => :omml, :to => :tex)
+    end
+
+    def to_mathml
+       TeXMath.convert("#{self.to_omml}", :from => :omml, :to => :mathml)
+    end
+
+
+    def replace_with_mathml(txt)
+    end
+
+    def replace_with_latex(txt)
+    end
+
+    def to_png
+      @path ||= Calculus::Expression.new("#{self.to_latex}", :parse => false).to_png
+      File.read(@path)
+    end
+
+    def base64_data
+      "data:image/png;base64,#{Base64.strict_encode64(self.to_png)}"
+    end
+
+    def save(path)
+      file = File.new(path, "wb")
+      file.write(self.to_png)
+      file.close
+    end
+
+    def to_html
+      "<img src='#{self.base64_data}' data-latex=\"#{self.to_latex}\" style='height: 13px;' />"
+    end
+
+    def to_s
+      self.to_latex.to_s
+    end
+
+  end
+end

data/lib/ruby_docx/elements/image.rb

@@ -0,0 +1,76 @@
+require 'mimemagic'
+require "base64"
+
+module RubyDocx::Elements
+  class Image < Element
+    attr_accessor :path, :zip, :link
+    attr_reader :width, :height
+
+    def data
+      if self.path
+        @zip.read("word/#{@path}")
+      else
+
+      end
+
+    end
+
+    def relation_id
+      element = @node.xpath(".//v:imagedata").first
+      if element && element.attributes.keys.index("id")
+        element.attributes["id"].value
+      else
+        nil
+      end
+    end
+
+    def replace(lnk)
+      @link = lnk
+    end
+
+    def style
+      element = @node.xpath(".//v:shape").first
+
+      if element && element.attributes.keys.index("style")
+        element.attributes["style"].value
+      else
+        nil
+      end
+    end
+
+    def title
+      element = @node.xpath(".//v:imagedata").first
+      if element && element.attributes.keys.index("title")
+        element.attributes["title"].value
+      else
+        nil
+      end
+    end
+
+    def mime
+      MimeMagic.by_magic(self.data).type
+    end
+
+    def base64_data
+      "data:#{self.mime};base64,#{Base64.strict_encode64(self.data)}"
+    end
+
+    def to_html
+      if @link.to_s.size > 0
+        "<img src='#{@link}' alt=\"#{self.title}\" style=\"#{self.style}\" />"
+      else
+        "<img src='#{self.base64_data}' alt=\"#{self.title}\" style=\"#{self.style}\" />"
+      end
+
+    end
+
+    def save(path)
+      if @zip && @path
+        file = File.new(path, "wb")
+        file.write(self.data)
+        file.close
+      end
+    end
+
+  end
+end

data/lib/ruby_docx/elements/link.rb

@@ -0,0 +1,51 @@
+module RubyDocx::Elements
+  class Link < Element
+    attr_reader :href, :content, :html_content
+
+    def element
+      @element ||= @node.children.first
+    end
+
+    def href
+      @href ||= @node.attributes["anchor"].value.to_s
+    end
+
+    def href=(an)
+      @href = an
+    end
+
+    def content
+      @content ||= self.elements.map(&:to_s).join("")
+    end
+
+    def content=(txt)
+      @content = txt
+    end
+
+    def html_content
+      @html_content ||= self.elements.map(&:to_html).join("")
+    end
+
+    def html_content=(html)
+      @html_content = html
+    end
+
+    def replace(url)
+      self.href = url
+    end
+
+    def to_html
+      if !self.href
+        "#{self.html_content}"
+      elsif self.href.to_s =~ /^http/
+        "<a href=\"#{self.href}\">#{self.html_content}</a>"
+      else
+        "<a href=\"##{self.href}\">#{self.html_content}</a>"
+      end
+    end
+
+    def to_s
+      self.content.to_s
+    end
+  end
+end

data/lib/ruby_docx/elements/ole_object.rb

@@ -0,0 +1,89 @@
+require 'texmath'
+require 'calculus'
+require 'mathtype'
+require 'mathtype_to_mathml'
+
+module Mathtype
+  class Converter
+    def initialize(equation)
+      ole = Ole::Storage.open(equation)
+      eq = ole.file.read("Equation Native")[28..-1]
+
+      data = Mathtype::Equation.read(eq).snapshot
+      @builder = Nokogiri::XML::Builder.new do |xml|
+        @xml = xml
+        xml.root do
+          process(object: data)
+        end
+      end
+    end
+
+  end
+end
+
+module RubyDocx::Elements
+  class OleObject < Element
+    attr_accessor :path, :zip, :link
+    attr_accessor :image_path
+
+    def style
+      element = @node.xpath(".//v:shape").first
+      if element && element.attributes.keys.index("style")
+        element.attributes["style"].value
+      else
+        nil
+      end
+    end
+
+    def relation_id
+      element = @node.xpath(".//o:OLEObject").first
+      if element && element.attributes.keys.index("id")
+        element.attributes["id"].value
+      else
+        nil
+      end
+    end
+
+    def to_mathml
+      io = StringIO.new(@zip.read("word/" + @path))
+      # xml = Mathtype::Converter.new(io)
+      # # puts xml.to_xml
+      # xml.to_xml
+
+      mathml = MathTypeToMathML::Converter.new(io).convert
+    end
+
+    def to_latex
+      TeXMath.convert("#{self.to_mathml}", :from => :mathml, :to => :tex)
+    end
+
+    def to_png
+      @image_path ||= Calculus::Expression.new("#{self.to_latex}", :parse => false).to_png
+      File.read(@image_path)
+    end
+
+    def base64_data
+      "data:image/png;base64,#{Base64.strict_encode64(self.to_png)}"
+    end
+
+    def save(path)
+      file = File.new(path, "wb")
+      file.write(self.to_png)
+      file.close
+    end
+
+    def to_html
+      if self.style
+        "<img src='#{self.base64_data}' data-latex=\"#{self.to_latex}\" style='#{self.style}' />"
+      else
+        "<img src='#{self.base64_data}' data-latex=\"#{self.to_latex}\" style='height: 13px;' />"
+      end
+
+    end
+
+    def to_s
+      self.to_latex.to_s
+    end
+
+  end
+end