ooxml_parser 0.28.0 → 0.29.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: ddd7d84f5f19a2b12fad76d455ebc614f9eb1b7a5c6f96f3585e649dc2569c4a
-  data.tar.gz: ca87a76367cb58ed7d02c7be6bcdb2ffe1e6635cccccff5a5ca33f44e44ed60b
+  metadata.gz: ae537423c384b461a5c2eb77bfe60b692ab334e1bd6ae22cc928e28230302997
+  data.tar.gz: c8696409b7862346b631367d7fa014ceea48e18deb3a6b9832fbb4dcc57dab13
 SHA512:
-  metadata.gz: 8e85e11b6a5faf18784cc9f3e9b216a1aad962aaf2be54f6dc35d1e3c8fbb73e9b4ed86270e796f01260fcdf2404729c2a36b5b9d133020916f142e12f32b903
-  data.tar.gz: 5e3c2129a515ce309a33e5f8c75847cde732a6ce4108f36bb8c3222d3b02f48b42bef3faa9eecaafc063f691e58d39e020f1eda3e44767b9452813ab8d3ca087
+  metadata.gz: 9b3655be06b3aa59bdb700ba359f0dffd6bbe54840978665870673edb2e7249c179807b102cc7b49a503c638b0508a3cb46f0183afa4be7a8b8baff93b3dbab6
+  data.tar.gz: 36b999f281320ca0792e6504060eb65ba0f97005ed82b131dea276f6b136e6a0cca0281a3ab4a0c67ce213fe31e26263a1603bc7cf4ac10a847e264e64ccd512

data/lib/ooxml_parser/common_parser/common_data/alternate_content/drawing/graphic/picture/docx_blip/file_reference.rb

@@ -36,7 +36,7 @@ module OoxmlParser
       full_path_to_file = OOXMLDocumentObject.path_to_folder + OOXMLDocumentObject.root_subfolder + @path.gsub('..', '')
       if File.exist?(full_path_to_file)
         @content = if File.extname(@path) == '.xlsx'
-                     OoxmlParser::Parser.parse(full_path_to_file)
+                     parse_ole_xlsx(full_path_to_file)
                    else
                      File.binread(full_path_to_file)
                    end
@@ -45,5 +45,21 @@ module OoxmlParser
       end
       self
     end
+
+    private
+
+    # Parse ole xlsx file
+    # @param [String] full_path to file
+    # @return [XLSXWorkbook]
+    def parse_ole_xlsx(full_path)
+      # TODO: Fix this ugly hack with global vars
+      #   by replacing all global variables
+      stack = OOXMLDocumentObject.xmls_stack
+      dir = OOXMLDocumentObject.path_to_folder
+      result = OoxmlParser::Parser.parse(full_path)
+      OOXMLDocumentObject.xmls_stack = stack
+      OOXMLDocumentObject.path_to_folder = dir
+      result
+    end
   end
 end

data/lib/ooxml_parser/common_parser/parser.rb

@@ -29,7 +29,7 @@ module OoxmlParser
         format = Parser.recognize_folder_format
         case format
         when :docx
-          DocumentStructure.parse
+          DocumentStructure.new.parse
         when :xlsx
           XLSXWorkbook.new.parse
         when :pptx

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

@@ -21,8 +21,12 @@ module OoxmlParser
     attr_accessor :next_style
     # @return [DocxParagraphRun] run properties
     attr_accessor :run_properties
+    # @return [Nokogiri::XML:Node] run properties node
+    attr_accessor :run_properties_node
     # @return [DocxParagraph] run properties
     attr_accessor :paragraph_properties
+    # @return [Nokogiri::XML:Node] paragraph properties node
+    attr_accessor :paragraph_properties_node
     # @return [TableProperties] properties of table
     attr_accessor :table_properties
     # @return [Array, TableStyleProperties] list of table style properties
@@ -76,9 +80,11 @@ module OoxmlParser
         when 'next'
           @next_style = subnode.attribute('val').value
         when 'rPr'
-          @run_properties = DocxParagraphRun.new(parent: self).parse_properties(subnode)
+          @run_properties_node = subnode
+          @run_properties = DocxParagraphRun.new(parent: self).parse_properties(@run_properties_node)
         when 'pPr'
-          @paragraph_properties = ParagraphProperties.new(parent: self).parse(subnode)
+          @paragraph_properties_node = subnode
+          @paragraph_properties = ParagraphProperties.new(parent: self).parse(@paragraph_properties_node)
         when 'tblPr'
           @table_properties = TableProperties.new(parent: self).parse(subnode)
         when 'trPr'

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

@@ -12,7 +12,6 @@ require_relative 'docx_paragraph/inserted'
 require_relative 'docx_paragraph/structured_document_tag'
 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
@@ -202,7 +201,8 @@ module OoxmlParser
         when 'contextualSpacing'
           @contextual_spacing = true
         when 'pStyle'
-          parse_paragraph_style_xml(node_child.attribute('val').value, default_char_style)
+          @paragraph_style_ref = ParagraphStyleRef.new(parent: self).parse(node_child)
+          fill_style_data(default_char_style)
         when 'ind'
           @ind = DocumentStructure.default_paragraph_style.ind.dup.parse(node_child)
         when 'numPr'
@@ -229,24 +229,13 @@ module OoxmlParser
       self
     end
 
-    # Parse style xml
-    # @param id [String] id of style to parse
+    # Fill data from styles
     # @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|
-          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)
-        end
-        break
-      end
+    def fill_style_data(character_style)
+      @style = root_object.document_style_by_id(@paragraph_style_ref.value)
+      parse_paragraph_style(@style.paragraph_properties_node, character_style) if @style.paragraph_properties_node
+      character_style.parse_properties(@style.run_properties_node) if @style.run_properties_node
     end
 
     extend Gem::Deprecate

data/lib/ooxml_parser/docx_parser/document_structure.rb

@@ -156,61 +156,60 @@ module OoxmlParser
 
     # Parse docx file
     # @return [DocumentStructure] parsed structure
-    def self.parse
-      doc_structure = DocumentStructure.new
-      doc_structure.content_types = ContentTypes.new(parent: doc_structure).parse
+    def parse
+      @content_types = ContentTypes.new(parent: self).parse
       OOXMLDocumentObject.root_subfolder = 'word/'
       OOXMLDocumentObject.xmls_stack = []
       @comments = []
       DocumentStructure.default_paragraph_style = DocxParagraph.new
-      DocumentStructure.default_run_style = DocxParagraphRun.new(parent: doc_structure)
-      doc_structure.theme = PresentationTheme.parse('word/theme/theme1.xml')
-      doc_structure.relationships = Relationships.new(parent: self).parse_file("#{OOXMLDocumentObject.path_to_folder}word/_rels/document.xml.rels")
-      doc_structure.parse_styles
+      DocumentStructure.default_run_style = DocxParagraphRun.new(parent: self)
+      @theme = PresentationTheme.parse('word/theme/theme1.xml')
+      @relationships = Relationships.new(parent: self).parse_file("#{OOXMLDocumentObject.path_to_folder}word/_rels/document.xml.rels")
+      parse_styles
       number = 0
       OOXMLDocumentObject.add_to_xmls_stack('word/document.xml')
-      doc = doc_structure.parse_xml(OOXMLDocumentObject.current_xml)
+      doc = parse_xml(OOXMLDocumentObject.current_xml)
       doc.search('//w:document').each do |document|
         document.xpath('w:background').each do |background|
-          doc_structure.background = DocumentBackground.new(parent: doc_structure).parse(background)
+          @background = DocumentBackground.new(parent: self).parse(background)
         end
         document.xpath('w:body').each do |body|
           body.xpath('*').each do |element|
             case element.name
             when 'p'
               child = element.child
-              unless child.nil? && doc_structure.elements.last.instance_of?(Table)
-                paragraph_style = DocumentStructure.default_paragraph_style.dup.parse(element, number, DocumentStructure.default_run_style, parent: doc_structure)
+              unless child.nil? && @elements.last.instance_of?(Table)
+                paragraph_style = DocumentStructure.default_paragraph_style.dup.parse(element, number, DocumentStructure.default_run_style, parent: self)
                 number += 1
-                doc_structure.elements << paragraph_style.dup
+                @elements << paragraph_style.dup
               end
             when 'tbl'
-              table = Table.new(parent: doc_structure).parse(element,
-                                                             number,
-                                                             TableProperties.new)
+              table = Table.new(parent: self).parse(element,
+                                                    number,
+                                                    TableProperties.new)
               number += 1
-              doc_structure.elements << table
+              @elements << table
             when 'sdt'
-              doc_structure.elements << StructuredDocumentTag.new(parent: doc_structure).parse(element)
+              @elements << StructuredDocumentTag.new(parent: self).parse(element)
             end
           end
           body.xpath('w:sectPr').each do |sect_pr|
-            doc_structure.page_properties = PageProperties.new(parent: doc_structure).parse(sect_pr,
-                                                                                            DocumentStructure.default_paragraph_style,
-                                                                                            DocumentStructure.default_run_style)
-            doc_structure.notes = doc_structure.page_properties.notes # keep copy of notes to compatibility with previous docx models
+            @page_properties = PageProperties.new(parent: self).parse(sect_pr,
+                                                                      DocumentStructure.default_paragraph_style,
+                                                                      DocumentStructure.default_run_style)
+            @notes = page_properties.notes # keep copy of notes to compatibility with previous docx models
           end
         end
       end
       OOXMLDocumentObject.xmls_stack.pop
-      doc_structure.document_properties = DocumentProperties.new(parent: doc_structure).parse
-      doc_structure.comments = Comments.new(parent: doc_structure).parse
-      doc_structure.comments_extended = CommentsExtended.new(parent: doc_structure).parse
-      doc_structure.comments_document = Comments.new(parent: doc_structure,
-                                                     file: "#{OOXMLDocumentObject.path_to_folder}word/#{doc_structure.relationships.target_by_type('commentsDocument').first}")
-                                                .parse
-      doc_structure.settings = DocumentSettings.new(parent: doc_structure).parse
-      doc_structure
+      @document_properties = DocumentProperties.new(parent: self).parse
+      @comments = Comments.new(parent: self).parse
+      @comments_extended = CommentsExtended.new(parent: self).parse
+      @comments_document = Comments.new(parent: self,
+                                        file: "#{OOXMLDocumentObject.path_to_folder}word/#{relationships.target_by_type('commentsDocument').first}")
+                                   .parse
+      @settings = DocumentSettings.new(parent: self).parse
+      self
     end
 
     # Parse default style

data/lib/ooxml_parser/docx_parser.rb

@@ -11,7 +11,7 @@ module OoxmlParser
     # @return [DocumentStructure] result of parse
     def self.parse_docx(path_to_file)
       Parser.parse_format(path_to_file) do
-        DocumentStructure.parse
+        DocumentStructure.new.parse
       end
     end
   end

data/lib/ooxml_parser/version.rb

@@ -4,6 +4,6 @@ module OoxmlParser
   # This module holds the RuboCop version information.
   module Version
     # [String] Version of Gem
-    STRING = '0.28.0'
+    STRING = '0.29.0'
   end
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: ooxml_parser
 version: !ruby/object:Gem::Version
-  version: 0.28.0
+  version: 0.29.0
 platform: ruby
 authors:
 - ONLYOFFICE
@@ -423,7 +423,6 @@ files:
 - lib/ooxml_parser/docx_parser/document_structure/docx_paragraph/sdt/sdt_content.rb
 - lib/ooxml_parser/docx_parser/document_structure/docx_paragraph/sdt/sdt_properties.rb
 - lib/ooxml_parser/docx_parser/document_structure/docx_paragraph/structured_document_tag.rb
-- lib/ooxml_parser/docx_parser/document_structure/docx_paragraph/style_parametres.rb
 - lib/ooxml_parser/docx_parser/document_structure/header_footer.rb
 - lib/ooxml_parser/docx_parser/document_structure/numbering.rb
 - lib/ooxml_parser/docx_parser/document_structure/numbering/abstract_numbering.rb

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

@@ -1,30 +0,0 @@
-# frozen_string_literal: true
-
-module OoxmlParser
-  # Style Parameter Data
-  class StyleParametres < OOXMLDocumentObject
-    attr_accessor :q_format, :hidden, :name
-
-    def initialize(params = {})
-      @name = params[:name]
-      @q_format = params.fetch(:q_format, false)
-      @hidden = params.fetch(:hidden, false)
-      super(parent: params[:parent])
-    end
-
-    # Parse StyleParametres data
-    # @param [Nokogiri::XML:Element] node with StyleParametres data
-    # @return [StyleParametres] value of Columns data
-    def parse(node)
-      node.xpath('*').each do |node_child|
-        case node_child.name
-        when 'name'
-          @name = node_child.attribute('val').value
-        when 'qFormat'
-          @q_format = option_enabled?(node_child)
-        end
-      end
-      self
-    end
-  end
-end