ooxml_parser 0.4.0 → 0.7.0

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/comments.rb

@@ -0,0 +1,36 @@
+# frozen_string_literal: true
+
+require_relative 'comments/comment'
+module OoxmlParser
+  # Class for parsing `comments.xml` file
+  class Comments < OOXMLDocumentObject
+    # @return [Array<Comment>] list of comments
+    attr_reader :comments_array
+
+    def initialize(params = {})
+      @comments_array = []
+      @parent = params[:parent]
+      @file = params.fetch(:file, OOXMLDocumentObject.path_to_folder + 'word/comments.xml')
+    end
+
+    # @return [Comment] accessor
+    def [](key)
+      @comments_array[key]
+    end
+
+    # Parse CommentsExtended object
+    # @return [Comments] result of parsing
+    def parse
+      return nil unless File.file?(@file)
+
+      doc = parse_xml(@file)
+      doc.xpath('w:comments/*').each do |node_child|
+        case node_child.name
+        when 'comment'
+          @comments_array << Comment.new(parent: self).parse(node_child)
+        end
+      end
+      self
+    end
+  end
+end

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/comments/comment.rb

@@ -0,0 +1,37 @@
+# frozen_string_literal: true
+
+module OoxmlParser
+  # Comment Data
+  class Comment < OOXMLDocumentObject
+    # @return [Integer] id of comment
+    attr_reader :id
+    # @return [Array<DocxParagraph>] array of paragraphs
+    attr_reader :paragraphs
+
+    def initialize(id = nil, paragraphs = [], parent: nil)
+      @id = id
+      @paragraphs = paragraphs
+      @parent = parent
+    end
+
+    # Parse Comment object
+    # @param node [Nokogiri::XML:Element] node to parse
+    # @return [Comment] result of parsing
+    def parse(node)
+      node.attributes.each do |key, value|
+        case key
+        when 'id'
+          @id = value.value.to_i
+        end
+      end
+
+      node.xpath('*').each do |node_child|
+        case node_child.name
+        when 'p'
+          @paragraphs << DocxParagraph.new.parse(node_child, 0, parent: self)
+        end
+      end
+      self
+    end
+  end
+end

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/comments_extended.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require_relative 'comments_extended/comment_extended'
 module OoxmlParser
   # Class for parsing `commentsExtended.xml` file
@@ -17,7 +19,8 @@ module OoxmlParser
     def parse
       file_to_parse = OOXMLDocumentObject.path_to_folder + 'word/commentsExtended.xml'
       return nil unless File.exist?(file_to_parse)
-      doc = Nokogiri::XML(File.open(file_to_parse))
+
+      doc = parse_xml(file_to_parse)
       doc.xpath('w15:commentsEx/*').each do |node_child|
         case node_child.name
         when 'commentEx'

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/comments_extended/comment_extended.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module OoxmlParser
   # Class for parsing `w15:commentEx` tag
   class CommentExtended < OOXMLDocumentObject

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/document_background.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module OoxmlParser
   # Class for describing Document Background `w:background`
   class DocumentBackground < OOXMLDocumentObject

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/document_properties.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module OoxmlParser
   # Document Properties
   class DocumentProperties < OOXMLDocumentObject
@@ -11,7 +13,7 @@ module OoxmlParser
         warn "There is no 'docProps/app.xml' in docx. It may be some problem with it"
         return self
       end
-      node = Nokogiri::XML(File.open(properties_file))
+      node = parse_xml(properties_file)
       node.xpath('*').each do |node_child|
         case node_child.name
         when 'Properties'

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/document_settings.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module OoxmlParser
   # Class for parsing `settings.xml` file
   class DocumentSettings < OOXMLDocumentObject
@@ -7,7 +9,10 @@ module OoxmlParser
     # Parse Settings object
     # @return [DocumentSettings] result of parsing
     def parse
-      doc = Nokogiri::XML(File.open(OOXMLDocumentObject.path_to_folder + 'word/settings.xml'))
+      settings_path = OOXMLDocumentObject.path_to_folder + 'word/settings.xml'
+      return nil unless File.exist?(settings_path)
+
+      doc = parse_xml(settings_path)
       doc.xpath('w:settings/*').each do |node_child|
         case node_child.name
         when 'defaultTabStop'

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/document_structure_helpers.rb

@@ -1,9 +1,10 @@
+# frozen_string_literal: true
+
 module OoxmlParser
   # method to help to work with DocumentStructure
   module DocumentStructureHelpers
     # @return [True, false] if structure contain any user data
     def with_data?
-      return true if @elements.length > 1
       @elements.each do |current_element|
         return true if current_element.with_data?
       end

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/document_style.rb

@@ -1,9 +1,14 @@
+# frozen_string_literal: true
+
 require_relative 'document_style/document_style_helper'
 module OoxmlParser
   # Class for describing styles containing in +styles.xml+
   class DocumentStyle < OOXMLDocumentObject
     include TableStylePropertiesHelper
     include DocumentStyleHelper
+
+    # @return [True, False] is style default
+    attr_reader :default
     # @return [Symbol] Type of style (+:paragraph+ or +:table+)
     attr_accessor :type
     # @return [FixNum] number of style
@@ -30,6 +35,7 @@ module OoxmlParser
     # Used to determine if current style is visible in style list in editors
     # According to http://www.wordarticles.com/Articles/WordStyles/LatentStyles.php
     attr_accessor :q_format
+
     alias visible? q_format
 
     def initialize(parent: nil)
@@ -38,10 +44,12 @@ module OoxmlParser
       @parent = parent
     end
 
+    # @return [String] result of convert of object to string
     def to_s
       "Table style properties list: #{@table_style_properties_list.join(',')}"
     end
 
+    # @return [String] inspect of object for debug means
     def inspect
       to_s
     end
@@ -55,6 +63,8 @@ module OoxmlParser
           @type = value.value.to_sym
         when 'styleId'
           @style_id = value.value
+        when 'default'
+          @default = attribute_enabled?(value.value)
         end
       end
       node.xpath('*').each do |subnode|
@@ -66,7 +76,7 @@ module OoxmlParser
         when 'next'
           @next_style = subnode.attribute('val').value
         when 'rPr'
-          @run_properties = DocxParagraphRun.new.parse_properties(subnode)
+          @run_properties = DocxParagraphRun.new(parent: self).parse_properties(subnode)
         when 'pPr'
           @paragraph_properties = ParagraphProperties.new(parent: self).parse(subnode)
         when 'tblPr'
@@ -84,16 +94,5 @@ module OoxmlParser
       fill_empty_table_styles
       self
     end
-
-    # Parse all document style list
-    # @return [Array, DocumentStyle]
-    def self.parse_list(parent)
-      styles_array = []
-      doc = Nokogiri::XML(File.open(OOXMLDocumentObject.path_to_folder + 'word/styles.xml'))
-      doc.search('//w:style').each do |style|
-        styles_array << DocumentStyle.new(parent: parent).parse(style)
-      end
-      styles_array
-    end
   end
 end

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/document_style/document_style_helper.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module OoxmlParser
   # Helper methods for working with Style List
   module DocumentStyleHelper

data/lib/ooxml_parser/docx_parser/docx_data/document_structure/docx_paragraph.rb

@@ -1,7 +1,13 @@
+# frozen_string_literal: true
+
 # noinspection RubyTooManyInstanceVariablesInspection
-require_relative 'docx_paragraph/bookmark'
+require_relative 'docx_paragraph/bookmark_start'
+require_relative 'docx_paragraph/bookmark_end'
+require_relative 'docx_paragraph/comment_range_end'
+require_relative 'docx_paragraph/comment_range_start'
 require_relative 'docx_paragraph/docx_paragraph_helper'
 require_relative 'docx_paragraph/docx_paragraph_run'
+require_relative 'docx_paragraph/field_simple'
 require_relative 'docx_paragraph/indents'
 require_relative 'docx_paragraph/inserted'
 require_relative 'docx_paragraph/structured_document_tag'
@@ -9,24 +15,27 @@ require_relative 'docx_paragraph/frame_properties'
 require_relative 'docx_paragraph/docx_formula'
 require_relative 'docx_paragraph/style_parametres'
 module OoxmlParser
+  # Class for data of DocxParagraph
   class DocxParagraph < OOXMLDocumentObject
     include DocxParagraphHelper
     attr_accessor :number, :bookmark_start, :bookmark_end, :align, :spacing, :background_color, :ind, :numbering,
-                  :character_style_array, :horizontal_line, :page_break, :kinoku, :borders, :keep_lines,
+                  :character_style_array, :page_break, :borders, :keep_lines,
                   :contextual_spacing, :sector_properties, :page_numbering, :section_break, :style, :keep_next,
-                  :orphan_control, :frame_properties
+                  :orphan_control
     # @return [Hyperlink] hyperlink in paragraph
     attr_reader :hyperlink
     # @return [ParagraphProperties] Properties of current paragraph
     attr_accessor :paragraph_properties
     # @return [Inserted] data inserted by review
     attr_accessor :inserted
+    # @return [FieldSimple] field simple
+    attr_reader :field_simple
     # @return [Integer] id of paragraph (for comment)
     attr_accessor :paragraph_id
+    # @return [MathParagraph] math paragraph
+    attr_reader :math_paragraph
     # @return [Integer] id of text (for comment)
     attr_accessor :text_id
-    # @return [StructuredDocumentTag] structured document tag data
-    attr_accessor :sdt
 
     def initialize(parent: nil)
       @number = 0
@@ -35,9 +44,7 @@ module OoxmlParser
       @align = 'left'
       @spacing = Spacing.new
       @ind = Indents.new
-      @kinoku = false
       @character_style_array = []
-      @horizontal_line = false
       @page_break = false
       @borders = Borders.new
       @keep_lines = false
@@ -48,6 +55,11 @@ module OoxmlParser
       @parent = parent
     end
 
+    alias elements character_style_array
+
+    # Constructor for copy of object
+    # @param source [DocxParagraph] original object
+    # @return [void]
     def initialize_copy(source)
       super
       @bookmark_start = source.bookmark_start.clone
@@ -56,18 +68,17 @@ module OoxmlParser
       @spacing = source.spacing.clone
     end
 
+    # @return [Array<OOXMLDocumentObject>] array of child objects that contains data
     def nonempty_runs
       @character_style_array.select do |cur_run|
-        if cur_run.is_a?(DocxParagraphRun)
-          (!cur_run.text.empty? ||
-              !cur_run.alternate_content.nil? ||
-              !cur_run.drawing.nil? ||
-              !cur_run.object.nil? ||
-              !cur_run.shape.nil? ||
-              !cur_run.footnote.nil? ||
-              !cur_run.endnote.nil?
-          )
-        elsif cur_run.is_a?(DocxFormula)
+        if cur_run.is_a?(DocxParagraphRun) || cur_run.is_a?(ParagraphRun)
+          !cur_run.empty?
+        elsif cur_run.is_a?(DocxFormula) ||
+              cur_run.is_a?(StructuredDocumentTag) ||
+              cur_run.is_a?(BookmarkStart) ||
+              cur_run.is_a?(BookmarkEnd) ||
+              cur_run.is_a?(CommentRangeStart) ||
+              cur_run.is_a?(CommentRangeEnd)
           true
         end
       end
@@ -77,16 +88,12 @@ module OoxmlParser
 
     # @return [True, false] if structure contain any user data
     def with_data?
-      !nonempty_runs.empty?
-    end
-
-    def remove_empty_runs
-      nonempty = nonempty_runs
-      @character_style_array.each do |cur_run|
-        @character_style_array.delete(cur_run) unless nonempty.include?(cur_run)
-      end
+      !nonempty_runs.empty? || paragraph_properties.section_properties
     end
 
+    # Compare this object to other
+    # @param other [Object] any other object
+    # @return [True, False] result of comparision
     def ==(other)
       ignored_attributes = %i[@number @parent]
       all_instance_variables = instance_variables
@@ -97,11 +104,18 @@ module OoxmlParser
       true
     end
 
+    # Parse object
+    # @param node [Nokogiri::XML:Node] node with DocxParagraph
+    # @param par_number [Integer] number of paragraph
+    # @param default_character [DocxParagraphRun] style for paragraph
+    # @param parent [OOXMLDocumentObject] parent of run
+    # @return [DocxParagraph] result of parse
     def parse(node, par_number = 0, default_character = DocxParagraphRun.new, parent: nil)
-      @parent = parent
+      @parent ||= parent
       default_character_style = default_character.dup
       character_styles_array = []
       custom_character_style = default_character_style.dup
+      custom_character_style.parent = self
       char_number = 0
       comments = []
       node.attributes.each do |key, value|
@@ -115,30 +129,20 @@ module OoxmlParser
       node.xpath('*').each do |node_child|
         case node_child.name
         when 'bookmarkStart'
-          @bookmark_start << Bookmark.new(parent: self).parse(node_child)
+          character_styles_array << BookmarkStart.new(parent: self).parse(node_child)
         when 'bookmarkEnd'
-          @bookmark_end << Bookmark.new(parent: self).parse(node_child)
+          character_styles_array << BookmarkEnd.new(parent: self).parse(node_child)
         when 'pPr'
           parse_paragraph_style(node_child, custom_character_style)
-          node.xpath('w:pict').each do |pict|
-            pict.xpath('v:rect').each do
-              @horizontal_line = true
-            end
-          end
           @paragraph_properties = ParagraphProperties.new(parent: self).parse(node_child)
         when 'commentRangeStart'
-          comments << node_child.attribute('id').value
+          character_styles_array << CommentRangeStart.new(parent: self).parse(node_child)
+        when 'commentRangeEnd'
+          character_styles_array << CommentRangeEnd.new(parent: self).parse(node_child)
         when 'fldSimple'
-          instruction = node_child.attribute('instr').to_s
-          @page_numbering = true if instruction.include?('PAGE')
-          node_child.xpath('w:r').each do |r_tag|
-            character_style = default_character_style.dup
-            character_style.parse(r_tag, char_number, parent: parent)
-            character_style.page_number = @page_numbering
-            character_style.instruction = instruction
-            character_styles_array << character_style.dup
-            char_number += 1
-          end
+          @field_simple = FieldSimple.new(parent: self).parse(node_child)
+          @page_numbering = true if field_simple.page_numbering?
+          character_styles_array += field_simple.runs
         when 'r'
           character_style = custom_character_style.dup
           node_child.xpath('w:instrText').each do |insrt_text|
@@ -147,47 +151,33 @@ module OoxmlParser
           character_style.parse(node_child, char_number, parent: self)
           character_style.comments = comments.dup
           character_styles_array << character_style.dup
-          character_styles_array.last.shape = character_style.shape unless character_style.shape.nil?
+          character_styles_array.last.shape = character_style.shape if character_style.shape
           char_number += 1
         when 'hyperlink'
           @hyperlink = Hyperlink.new(parent: self).parse(node_child)
           character_style = default_character_style.dup
-          if !node_child.attribute('id').nil?
+          character_style.parent = self
+          if node_child.attribute('id')
             character_style.link = Hyperlink.new(parent: character_style).parse(node_child)
-          else
-            character_style.link = node_child.attribute('anchor').value unless node_child.attribute('anchor').nil?
+          elsif node_child.attribute('anchor')
+            character_style.link = node_child.attribute('anchor').value
           end
           node_child.xpath('w:r').each do |r_tag|
-            character_style.parse(r_tag, char_number, parent: parent)
+            character_style.parse(r_tag, char_number, parent: self)
             character_styles_array << character_style.dup
             char_number += 1
           end
           node_child.xpath('w:fldSimple').each do |simple_field|
-            instruction = simple_field.attribute('instr').to_s
-            @page_numbering = true if instruction.include?('PAGE')
-            simple_field.xpath('w:r').each do |r_tag|
-              character_style.parse(r_tag, char_number, parent: self)
-              character_style.page_number = @page_numbering
-              character_style.instruction = instruction
-              character_styles_array << character_style.dup
-              char_number += 1
-            end
+            hyperlink_field_simple = FieldSimple.new(parent: self).parse(simple_field)
+            character_styles_array += hyperlink_field_simple.runs
           end
         when 'oMathPara'
-          node_child.xpath('m:oMath').each do |o_math|
-            character_styles_array << DocxFormula.new(parent: self).parse(o_math)
-          end
-        when 'commentRangeEnd'
-          comments.each_with_index do |comment, index|
-            if comment == node_child.attribute('id').value
-              comments.delete_at(index)
-              break
-            end
-          end
+          @math_paragraph = MathParagraph.new(parent: self).parse(node_child)
+          character_styles_array << math_paragraph.math
         when 'ins'
           @inserted = Inserted.new(parent: self).parse(node_child)
         when 'sdt'
-          @sdt = StructuredDocumentTag.new(parent: self).parse(node_child)
+          character_styles_array << StructuredDocumentTag.new(parent: self).parse(node_child)
         end
       end
       @number = par_number
@@ -197,7 +187,11 @@ module OoxmlParser
       self
     end
 
-    def parse_paragraph_style(node, default_char_style = DocxParagraphRun.new)
+    # Parse style
+    # @param node [Nokogiri::XML:Node] node with style
+    # @param default_char_style [DocxParagraphRun] style for paragraph
+    # @return [DocxParagraph] result of parse
+    def parse_paragraph_style(node, default_char_style = DocxParagraphRun.new(parent: self))
       node.xpath('*').each do |node_child|
         case node_child.name
         when 'pageBreakBefore'
@@ -221,13 +215,9 @@ module OoxmlParser
           @background_color = Color.new(parent: self).parse_hex_string(background_color_string)
           @background_color.set_style(node_child.attribute('val').value.to_sym) unless node_child.attribute('val').nil?
         when 'pStyle'
-          DocxParagraph.parse_paragraph_style_xml(node_child.attribute('val').value, self, default_char_style)
+          parse_paragraph_style_xml(node_child.attribute('val').value, default_char_style)
         when 'ind'
           @ind = DocumentStructure.default_paragraph_style.ind.dup.parse(node_child)
-        when 'kinoku'
-          @kinoku = true
-        when 'framePr'
-          @frame_properties = FrameProperties.new(parent: self).parse(node_child)
         when 'numPr'
           @numbering = NumberingProperties.new(parent: self).parse(node_child)
         when 'jc'
@@ -245,8 +235,6 @@ module OoxmlParser
           @section_break = case @sector_properties.type
                            when 'oddPage'
                              'Odd page'
-                           when 'evenPage'
-                             'Even page'
                            when 'continuous'
                              'Current Page'
                            else
@@ -258,19 +246,42 @@ module OoxmlParser
       self
     end
 
-    def self.parse_paragraph_style_xml(id, paragraph_style, character_style)
-      doc = Nokogiri::XML(File.open(OOXMLDocumentObject.path_to_folder + 'word/styles.xml'))
+    # Parse style xml
+    # @param id [String] id of style to parse
+    # @param character_style [DocxParagraphRun] style to parse
+    # @return [void]
+    def parse_paragraph_style_xml(id, character_style)
+      doc = parse_xml(OOXMLDocumentObject.path_to_folder + 'word/styles.xml')
       doc.search('//w:style').each do |style|
         next unless style.attribute('styleId').value == id
+
         style.xpath('w:pPr').each do |p_pr|
-          paragraph_style.parse_paragraph_style(p_pr, character_style)
-          paragraph_style.style = StyleParametres.new(parent: paragraph_style).parse(style)
+          parse_paragraph_style(p_pr, character_style)
+          @style = StyleParametres.new(parent: self).parse(style)
         end
         style.xpath('w:rPr').each do |r_pr|
-          character_style.parse_properties(r_pr, DocumentStructure.default_run_style)
+          character_style.parse_properties(r_pr)
         end
         break
       end
     end
+
+    extend Gem::Deprecate
+    deprecate :page_numbering, 'field_simple.page_numbering?', 2020, 1
+
+    # @return [OoxmlParser::StructuredDocumentTag] Return first sdt element
+    def sdt
+      @character_style_array.each do |cur_element|
+        return cur_element if cur_element.is_a?(StructuredDocumentTag)
+      end
+      nil
+    end
+    deprecate :sdt, 'nonempty_runs[i]', 2020, 1
+
+    # @return [OoxmlParser::FrameProperties] Return frame properties
+    def frame_properties
+      paragraph_properties.frame_properties
+    end
+    deprecate :frame_properties, 'paragraph_properties.frame_properties', 2020, 1
   end
 end