Merge pull request #66 from visoft/master

Added opening streams and outputting to a stream

Author: satoryu

README.md

@@ -51,6 +51,17 @@ doc.bookmarks.each_pair do |bookmark_name, bookmark_object|
 end
 ```
 
+Don't have a local file but a buffer? Docx handles those to:
+
+```ruby
+require 'docx'
+
+# Create a Docx::Document object from a remote file
+doc = Docx::Document.open(buffer)
+
+# Everything about reading is the same as shown above
+```
+
 ### Rendering html
 ``` ruby
 require 'docx'

lib/docx/document.rb

@@ -20,9 +20,16 @@ module Docx
   class Document
     attr_reader :xml, :doc, :zip, :styles
 
-    def initialize(path, &block)
+    def initialize(path_or_io, options = {})
       @replace = {}
-      @zip = Zip::File.open(path)
+
+      # if path-or_io is string && does not contain a null byte
+      if (path_or_io.instance_of?(String) && !/\u0000/.match?(path_or_io))
+        @zip = Zip::File.open(path_or_io)
+      else
+        @zip = Zip::File.open_buffer(path_or_io)
+      end
+
       @document_xml = @zip.read('word/document.xml')
       @doc = Nokogiri::XML(@document_xml)
       load_styles
@@ -32,33 +39,31 @@ def initialize(path, &block)
       end
     end
 
-
     # This stores the current global document properties, for now
     def document_properties
       {
         font_size: font_size
       }
     end
 
-
     # With no associated block, Docx::Document.open is a synonym for Docx::Document.new. If the optional code block is given, it will be passed the opened +docx+ file as an argument and the Docx::Document oject will automatically be closed when the block terminates. The values of the block will be returned from Docx::Document.open.
     # call-seq:
     #   open(filepath) => file
     #   open(filepath) {|file| block } => obj
     def self.open(path, &block)
-      self.new(path, &block)
+      new(path, &block)
     end
 
     def paragraphs
       @doc.xpath('//w:document//w:body/w:p').map { |p_node| parse_paragraph_from p_node }
     end
 
     def bookmarks
-      bkmrks_hsh = Hash.new
+      bkmrks_hsh = {}
       bkmrks_ary = @doc.xpath('//w:bookmarkStart').map { |b_node| parse_bookmark_from b_node }
       # auto-generated by office 2010
-      bkmrks_ary.reject! {|b| b.name == "_GoBack" }
-      bkmrks_ary.each {|b| bkmrks_hsh[b.name] = b }
+      bkmrks_ary.reject! { |b| b.name == '_GoBack' }
+      bkmrks_ary.each { |b| bkmrks_hsh[b.name] = b }
       bkmrks_hsh
     end
 
@@ -104,6 +109,7 @@ def save(path)
       Zip::OutputStream.open(path) do |out|
         zip.each do |entry|
           next unless entry.file?
+
           out.put_next_entry(entry.name)
 
           if @replace[entry.name]
@@ -116,7 +122,28 @@ def save(path)
       zip.close
     end
 
-    alias_method :text, :to_s
+    # Output entire document as a StringIO object
+    def stream
+      update
+      stream = Zip::OutputStream.write_buffer do |out|
+        zip.each do |entry|
+          next unless entry.file?
+
+          out.put_next_entry(entry.name)
+
+          if @replace[entry.name]
+            out.write(@replace[entry.name])
+          else
+            out.write(zip.read(entry.name))
+          end
+        end
+      end
+
+      stream.rewind
+      stream
+    end
+
+    alias text to_s
 
     def replace_entry(entry_path, file_contents)
       @replace[entry_path] = file_contents
@@ -138,7 +165,7 @@ def load_styles
     # end of methods that make edits?
     #++
     def update
-      replace_entry "word/document.xml", doc.serialize(:save_with => 0)
+      replace_entry 'word/document.xml', doc.serialize(save_with: 0)
     end
 
     # generate Elements::Containers::Paragraph from paragraph XML node

spec/docx/document_spec.rb

@@ -1,18 +1,13 @@
-# coding: utf-8
 require 'docx'
 require 'tempfile'
 
 describe Docx::Document do
   before(:all) do
-    @fixtures_path = "spec/fixtures"
+    @fixtures_path = 'spec/fixtures'
     @formatting_line_count = 12 # number of lines the formatting.docx file has
   end
 
-  describe 'reading' do
-    before do
-      @doc = Docx::Document.open(@fixtures_path + '/basic.docx')
-    end
-
+  shared_examples_for 'reading' do
     it 'should read the document' do
       expect(@doc.paragraphs.size).to eq(2)
       expect(@doc.paragraphs.first.text).to eq('hello')
@@ -53,12 +48,57 @@
     end
   end
 
+  shared_examples_for 'saving to file' do
+    it 'should save to a normal file path' do
+      @new_doc_path = @fixtures_path + '/new_save.docx'
+      @doc.save(@new_doc_path)
+      @new_doc = Docx::Document.open(@new_doc_path)
+      expect(@new_doc.paragraphs.size).to eq(@doc.paragraphs.size)
+    end
+
+    it 'should save to a tempfile' do
+      temp_file = Tempfile.new(['docx_gem', '.docx'])
+      @new_doc_path = temp_file.path
+      @doc.save(@new_doc_path)
+      @new_doc = Docx::Document.open(@new_doc_path)
+      expect(@new_doc.paragraphs.size).to eq(@doc.paragraphs.size)
+
+      temp_file.close
+      temp_file.unlink
+      # ensure temp file has been removed
+      expect(File.exist?(@new_doc_path)).to eq(false)
+    end
+
+    after do
+      File.delete(@new_doc_path) if File.exist?(@new_doc_path)
+    end
+  end
+
+  describe 'reading' do
+    context 'using normal file' do
+      before do
+        @doc = Docx::Document.open(@fixtures_path + '/basic.docx')
+      end
+
+      it_behaves_like 'reading'
+    end
+
+    context 'using stream' do
+      before do
+        stream = File.binread(@fixtures_path + '/basic.docx')
+        @doc = Docx::Document.open(stream)
+      end
+
+      it_behaves_like 'reading'
+    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
+    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)
@@ -71,7 +111,7 @@
       end
     end
 
-    it "should have tables with columns and cells" do
+    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)
@@ -82,23 +122,23 @@
       end
     end
 
-    it "should have proper count" do
+    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
-      expect(@doc.tables[0].rows[0].cells[0].text).to eq "ENGLISH"
-      expect(@doc.tables[0].rows[0].cells[1].text).to eq "FRANÇAIS"
-      expect(@doc.tables[1].rows[0].cells[0].text).to eq "Second table"
-      expect(@doc.tables[1].rows[0].cells[1].text).to eq "Second tableau"
-      expect(@doc.tables[0].columns[0].cells[5].text).to eq "aphids"
-      expect(@doc.tables[0].columns[1].cells[5].text).to eq "puceron"
+    it 'should have tables with proper text' do
+      expect(@doc.tables[0].rows[0].cells[0].text).to eq 'ENGLISH'
+      expect(@doc.tables[0].rows[0].cells[1].text).to eq 'FRANÇAIS'
+      expect(@doc.tables[1].rows[0].cells[0].text).to eq 'Second table'
+      expect(@doc.tables[1].rows[0].cells[1].text).to eq 'Second tableau'
+      expect(@doc.tables[0].columns[0].cells[5].text).to eq 'aphids'
+      expect(@doc.tables[0].columns[1].cells[5].text).to eq 'puceron'
     end
 
-    it "should read embedded links" do
+    it 'should read embedded links' do
       expect(@doc.tables[0].columns[1].cells[1].text).to match(/^Directive/)
     end
 
@@ -109,7 +149,7 @@
     end
   end
 
-  describe 'editing'  do
+  describe 'editing' do
     before do
       @doc = Docx::Document.open(@fixtures_path + '/editing.docx')
     end
@@ -173,8 +213,7 @@
     end
 
     it 'should allow content deletion' do
-      expect{@doc.paragraphs.first.remove!}.to change{@doc.paragraphs.size}.by(-1)
-
+      expect { @doc.paragraphs.first.remove! }.to change { @doc.paragraphs.size }.by(-1)
     end
   end
 
@@ -223,7 +262,6 @@
     end
 
     it 'should contain a paragraph with multiple text runs' do
-
     end
 
     it 'should detect normal formatting' do
@@ -316,34 +354,21 @@
   end
 
   describe 'saving' do
-    before do
-      @doc = Docx::Document.open(@fixtures_path + '/saving.docx')
-    end
-
-    it 'should save to a normal file path' do
-      @new_doc_path = @fixtures_path + '/new_save.docx'
-      @doc.save(@new_doc_path)
-      @new_doc = Docx::Document.open(@new_doc_path)
-      expect(@new_doc.paragraphs.size).to eq(@doc.paragraphs.size)
-    end
-
-    it 'should save to a tempfile' do
-      temp_file = Tempfile.new(['docx_gem', '.docx'])
-      @new_doc_path = temp_file.path
-      @doc.save(@new_doc_path)
-      @new_doc = Docx::Document.open(@new_doc_path)
-      expect(@new_doc.paragraphs.size).to eq(@doc.paragraphs.size)
+    context 'from a normal file' do
+      before do
+        @doc = Docx::Document.open(@fixtures_path + '/saving.docx')
+      end
 
-      temp_file.close
-      temp_file.unlink
-      # ensure temp file has been removed
-      expect(File.exists?(@new_doc_path)).to eq(false)
+      it_behaves_like 'saving to file'
     end
 
-    after do
-      if File.exists?(@new_doc_path)
-        File.delete(@new_doc_path)
+    context 'from a stream' do
+      before do
+        stream = File.binread(@fixtures_path + '/saving.docx')
+        @doc = Docx::Document.open(stream)
       end
+
+      it_behaves_like 'saving to file'
     end
 
     context 'wps modified docx file' do
@@ -357,6 +382,24 @@
     end
   end
 
+  describe 'streaming' do
+    it 'should return a StringIO to send over HTTP' do
+      doc = Docx::Document.open(@fixtures_path + '/basic.docx')
+      expect(doc.stream).to be_a(StringIO)
+    end
+
+    context 'should return a valid docx stream' do
+      before do
+        doc = Docx::Document.open(@fixtures_path + '/basic.docx')
+        result = doc.stream
+
+        @doc = Docx::Document.open(result)
+      end
+
+      it_behaves_like 'reading'
+    end
+  end
+
   describe 'outputting html' do
     before do
       @doc = Docx::Document.open(@fixtures_path + '/formatting.docx')
@@ -400,7 +443,7 @@
       expect(@doc.paragraphs[8].to_html.scan(regex).flatten.first.split(';').include?('text-align:right')).to eq(true)
     end
 
-    it "should set font size on styled paragraphs" do
+    it 'should set font size on styled paragraphs' do
       regex = /(\<p{1})[^\>]+style\=\"([^\"]+).+(<\/p>)/
       scan = @doc.paragraphs[9].to_html.scan(regex).flatten
       expect(scan.first).to eq '<p'
@@ -441,32 +484,28 @@
     it 'should output styled html' do
       expect(@formatted_line.to_html.scan('<span style="text-decoration:underline;"><strong><em>all</em></strong></span>').size).to eq 1
     end
-
   end
 
   describe 'replacing contents' do
     let(:replacement_file_path) { @fixtures_path + '/replacement.png' }
-    let(:temp_file_path){ Tempfile.new(['docx_gem', '.docx']).path }
-    let(:entry_path){ 'word/media/image1.png' }
-    let(:doc){ Docx::Document.open(@fixtures_path + '/replacement.docx') }
+    let(:temp_file_path) { Tempfile.new(['docx_gem', '.docx']).path }
+    let(:entry_path) { 'word/media/image1.png' }
+    let(:doc) { Docx::Document.open(@fixtures_path + '/replacement.docx') }
 
     it 'should replace existing file within the document' do
-      File.open replacement_file_path, "rb" do |io|
+      File.open replacement_file_path, 'rb' do |io|
         doc.replace_entry entry_path, io.read
       end
 
       doc.save(temp_file_path)
 
-      File.open replacement_file_path, "rb" do |io|
-        expect(Zip::File.open(temp_file_path).read entry_path).to eq io.read
+      File.open replacement_file_path, 'rb' do |io|
+        expect(Zip::File.open(temp_file_path).read(entry_path)).to eq io.read
       end
     end
 
     after do
-      if File.exists?(temp_file_path)
-        File.delete(temp_file_path)
-      end
+      File.delete(temp_file_path) if File.exist?(temp_file_path)
     end
   end
 end
-