support for reading tables

Author: yannp

README.md

@@ -1,6 +1,6 @@
 # docx
 
-a ruby library/gem for interacting with `.docx` files. currently capabilities include reading paragraphs/bookmarks, inserting text at bookmarks, and saving the document.
+a ruby library/gem for interacting with `.docx` files. currently capabilities include reading paragraphs/bookmarks, inserting text at bookmarks, reading tables/rows/columns/cells and saving the document.
 
 ## usage
 
@@ -29,6 +29,36 @@ doc.bookmarks.each_pair do |bookmark_name, bookmark_object|
 end
 ```
 
+### reading tables
+
+``` ruby
+require 'docx'
+
+# Create a Docx::Document object for our existing docx file
+doc = Docx::Document.open('tables.docx')
+
+first_table = doc.tables[0]
+puts first_table.row_count
+puts first_table.column_count
+puts first_table.rows[0].cells[0].text
+puts first_table.columns[0].cells[0].text
+
+# Iterate through tables
+doc.tables.each do |table|
+  table.rows.each do |row| # Row-based iteration
+    row.cells.each do |cell|
+      puts cell.text
+    end
+  end
+  
+  table.columns.each do |column| # Column-based iteration
+    column.cells.each do |cell|
+      puts cell.text
+    end
+  end
+end
+```
+
 ### writing
 
 ``` ruby

docx.gemspec

@@ -6,9 +6,9 @@ Gem::Specification.new do |s|
   s.version     = Docx::VERSION
   s.summary     = 'a ruby library/gem for interacting with .docx files'
   s.description = s.summary
-  s.authors     = ['Christopher Hunt', 'Marcus Ortiz']
+  s.authors     = ['Christopher Hunt', 'Marcus Ortiz', 'Yann Plancqueel']
   s.email       = ['chrahunt@gmail.com']
-  s.homepage    = 'https://github.com/chrahunt/docx'
+  s.homepage    = 'https://github.com/yannp/docx'
   s.files       = Dir["README.md", "LICENSE.md", "lib/**/*.rb"]
 
   s.add_dependency 'nokogiri', '~> 1.5'

lib/docx/containers.rb

@@ -1,3 +1,4 @@
 require 'docx/containers/container'
 require 'docx/containers/text_run'
-require 'docx/containers/paragraph'
+require 'docx/containers/paragraph'
+require 'docx/containers/table'

lib/docx/containers/table.rb

@@ -0,0 +1,51 @@
+require 'docx/containers/table_row'
+require 'docx/containers/table_column'
+require 'docx/containers/container'
+
+module Docx
+  module Elements
+    module Containers
+      class Table
+        include Container
+        include Elements::Element
+
+        def self.tag
+          'tbl'
+        end
+
+        def initialize(node)
+          @node = node
+          @properties_tag = 'tblGrid'
+        end
+
+        # Array of row
+        def rows
+          @node.xpath('w:tr').map {|r_node| Containers::TableRow.new(r_node) }
+        end
+
+        def row_count
+          @node.xpath('w:tr').count
+        end
+
+        # Array of column
+        def columns
+          columns_containers = []
+          (0..(column_count-1)).each do |i|
+            columns_containers[i] = Containers::TableColumn.new @node.xpath("w:tr//w:tc[#{i+1}]")
+          end
+          columns_containers
+        end
+
+        def column_count
+          @node.xpath('w:tblGrid/w:gridCol').count
+        end
+
+        # Iterate over each row within a table
+        def each_rows
+          rows.each { |r| yield(r) }
+        end
+        
+      end
+    end
+  end
+end

lib/docx/containers/table_cell.rb

@@ -0,0 +1,39 @@
+require 'docx/containers/text_run'
+require 'docx/containers/container'
+
+module Docx
+  module Elements
+    module Containers
+      class TableCell
+        include Container
+        include Elements::Element
+
+        def self.tag
+          'tc'
+        end
+
+        def initialize(node)
+          @node = node
+          @properties_tag = 'tcPr'
+        end
+
+        # Return text of paragraph's cell
+        def to_s
+          paragraphs.map(&:text).join('')
+        end
+
+        # Array of paragraphs contained within cell
+        def paragraphs
+          @node.xpath('w:p').map {|p_node| Containers::Paragraph.new(p_node) }
+        end
+
+        # Iterate over each text run within a paragraph's cell
+        def each_paragraph
+          paragraphs.each { |tr| yield(tr) }
+        end
+        
+        alias_method :text, :to_s
+      end
+    end
+  end
+end

lib/docx/containers/table_column.rb

@@ -0,0 +1,29 @@
+require 'docx/containers/table_cell'
+require 'docx/containers/container'
+
+module Docx
+  module Elements
+    module Containers
+      class TableColumn
+        include Container
+        include Elements::Element
+
+        def self.tag
+          'w:gridCol'
+        end
+
+        def initialize(cell_nodes)
+          @node = ''
+          @properties_tag = ''
+          @cells = cell_nodes.map { |c_node| Containers::TableCell.new(c_node) }
+        end
+
+        # Array of cells contained within row
+        def cells
+          @cells
+        end
+        
+      end
+    end
+  end
+end

lib/docx/containers/table_row.rb

@@ -0,0 +1,28 @@
+require 'docx/containers/table_cell'
+require 'docx/containers/container'
+
+module Docx
+  module Elements
+    module Containers
+      class TableRow
+        include Container
+        include Elements::Element
+
+        def self.tag
+          'tr'
+        end
+
+        def initialize(node)
+          @node = node
+          @properties_tag = ''
+        end
+
+        # Array of cells contained within row
+        def cells
+          @node.xpath('w:tc').map {|c_node| Containers::TableCell.new(c_node) }
+        end
+        
+      end
+    end
+  end
+end

lib/docx/document.rb

@@ -16,7 +16,7 @@ module Docx
   #     puts d.text
   #   end
   class Document
-    delegate :paragraphs, :bookmarks, :to => :@parser
+    delegate :paragraphs, :bookmarks, :tables, :to => :@parser
     delegate :doc, :xml, :zip, :to => :@parser
     def initialize(path, &block)
       @replace = {}

lib/docx/parser.rb

@@ -30,6 +30,10 @@ def bookmarks
       bkmrks_ary.each {|b| bkmrks_hsh[b.name] = b }
       bkmrks_hsh
     end
+
+    def tables
+      @doc.xpath('//w:document//w:body//w:tbl').map { |t_node| parse_table_from t_node }
+    end
 
     private
 
@@ -42,5 +46,9 @@ def parse_paragraph_from(p_node)
     def parse_bookmark_from(b_node)
       Elements::Bookmark.new(b_node)
     end
+
+    def parse_table_from(t_node)
+      Elements::Containers::Table.new(t_node)
+    end
   end
 end

spec/docx/document_spec.rb

@@ -39,6 +39,52 @@
     end
   end
 
+  describe 'read tables' do
+    before do
+      @doc = Docx::Document.open(@fixtures_path + '/tables.docx')
+    end
+
+    it "should have tables with rows and cells" do
+      expect(@doc.tables.count).to eq 2
+      @doc.tables.each do |table|
+        expect(table).to be_an_instance_of(Docx::Elements::Containers::Table)
+        table.rows.each do |row|
+          expect(row).to be_an_instance_of(Docx::Elements::Containers::TableRow)
+          row.cells.each do |cell|
+            expect(cell).to be_an_instance_of(Docx::Elements::Containers::TableCell)
+          end
+        end
+      end
+    end
+
+    it "should have tables with columns and cells" do
+      @doc.tables.each do |table|
+        table.columns.each do |column|
+          expect(column).to be_an_instance_of(Docx::Elements::Containers::TableColumn)
+          column.cells.each do |cell|
+            expect(cell).to be_an_instance_of(Docx::Elements::Containers::TableCell)
+          end
+        end
+      end
+    end
+
+    it "should have proper count" do
+      expect(@doc.tables[0].row_count).to eq 171
+      expect(@doc.tables[1].row_count).to eq 2
+      expect(@doc.tables[0].column_count).to eq 2
+      expect(@doc.tables[1].column_count).to eq 2
+    end
+
+    it "should have tables with proper text" do
+      @doc.tables[0].rows[0].cells[0].text.should eq "ENGLISH"
+      @doc.tables[0].rows[0].cells[1].text.should eq "FRANÇAIS"
+      @doc.tables[1].rows[0].cells[0].text.should eq "Second table"
+      @doc.tables[1].rows[0].cells[1].text.should eq "Second tableau"
+      @doc.tables[0].columns[0].cells[5].text.should eq "aphids"
+      @doc.tables[0].columns[1].cells[5].text.should eq "puceron"
+    end
+  end
+
   describe 'editing'  do
     before do
       @doc = Docx::Document.open(@fixtures_path + '/editing.docx')