officex2str 0.0.3 → 0.0.4

data/README.md

@@ -1,6 +1,6 @@
 # Officex2str
 
-Convert office 2010 files to string.
+Convert Microsoft office 2007/2010 files(docx/xlsx/pptx) to string.
 
 ## Installation
 
@@ -18,5 +18,5 @@ Or install it yourself as:
 
 ## Usage
 
-   Officex2str.convert("file_path")
+    Officex2str.convert("file_path")
 

data/fixtures/sample.txt

@@ -0,0 +1 @@
+sample

data/lib/officex2str/version.rb

@@ -1,3 +1,3 @@
 module Officex2str
-  VERSION = "0.0.3"
+  VERSION = "0.0.4"
 end

data/lib/officex2str.rb

@@ -1,37 +1,60 @@
 require 'nokogiri'
 require 'zipruby'
 require 'mime/types'
-#require "officex2str/version"
 
-module Officex2str
+class Officex2str
+  DOCX_CONTENT_TYPE = "application/vnd.openxmlformats-officedocument.wordprocessingml.document"
+  XLSX_CONTENT_TYPE = "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet"
+  PPTX_CONTENT_TYPE = "application/vnd.openxmlformats-officedocument.presentationml.presentation"
+  VALID_CONTENT_TYPE = [DOCX_CONTENT_TYPE, XLSX_CONTENT_TYPE, PPTX_CONTENT_TYPE].freeze
+
+  attr_accessor :path, :content_type
+
   def self.convert(file_path)
-    archives   = Zip::Archive.open(file_path) { |archive| archive.map(&:name) }
-    pages      = self.pickup_pages(file_path, archives)
-    xmls       = self.extract_xmls(file_path, pages)
-    self.xml_to_str(xmls)
+    self.new(file_path).convert
+  end
+
+  def initialize(file_path)
+    @path = file_path
+    @content_type = MIME::Types.type_for(path).first.content_type
+  end
+
+  def convert
+    if valid_file?
+      archives   = Zip::Archive.open(path) { |archive| archive.map(&:name) }
+      pages      = pickup_pages(archives)
+      xmls       = extract_xmls(pages)
+      xml_to_str(xmls)
+    else
+      raise InvaildFileTypeError, "Not recognized file type"
+    end
   end
 
 private
-  def self.pickup_pages file_path, archives
-    case content_type = MIME::Types.type_for(file_path).first.content_type
-    when "application/vnd.openxmlformats-officedocument.wordprocessingml.document"
+  def valid_file?
+    !!VALID_CONTENT_TYPE.include?(content_type)
+  end
+
+  def pickup_pages archives
+    case content_type
+    when DOCX_CONTENT_TYPE
       archives.select{|a| /^word\/document/ =~ a}
-    when "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet"
+    when XLSX_CONTENT_TYPE
       archives.select{|a| /^xl\/worksheets\/sheet/ =~ a or /^xl\/sharedStrings/ =~ a or /^xl\/comments/ =~ a }
-    when "application/vnd.openxmlformats-officedocument.presentationml.presentation"
+    when PPTX_CONTENT_TYPE
       archives.select{|a| /^ppt\/slides\/slide/ =~ a}
     else
-      nil
+      raise InvalidContentTypeError, "Not recognized content type"
     end
   end
 
-  def self.extract_xmls file_path, pages
+  def extract_xmls pages
     xml_text = []
-    Zip::Archive.open(file_path) { |archive| pages.each{ |page| archive.fopen(page) do |f| xml_text << f.read end; } }
+    Zip::Archive.open(path) { |archive| pages.each{ |page| archive.fopen(page) do |f| xml_text << f.read end; } }
     xml_text
   end
 
-  def self.xml_to_str xml_text
+  def xml_to_str xml_text
     text = ""
     xml_text.each{|xml_t| text << Nokogiri.XML(xml_t.toutf8, nil, 'utf8').to_str } unless xml_text.empty?
     text

data/spec/officex2str_spec.rb

@@ -2,44 +2,56 @@
 require 'spec_helper'
 
 describe Officex2str do
+  context "#valid_file?" do
+    subject do
+      Officex2str.new(@file_path).send(:valid_file?)
+    end
+    context "extname is docx" do
+      before { @file_path = "fixtures/sample.docx" }
+      it { subject.should be_true }
+    end
+    context "extname is xlsx" do
+      before { @file_path = "fixtures/sample.xlsx" }
+      it { subject.should be_true }
+    end
+    context "extname is pptx" do
+      before { @file_path = "fixtures/sample.pptx" }
+      it { subject.should be_true }
+    end
+    context "extname is txt" do
+      before { @file_path = "fixtures/sample.txt" }
+      it { subject.should be_false }
+    end
+  end
+
   context "#pickup_pages" do
     subject do
       archives = Zip::Archive.open(@file_path) { |archive| archive.map(&:name) }
-      Officex2str.send(:pickup_pages, @file_path, archives).sort
+      Officex2str.new(@file_path).send(:pickup_pages, archives).sort
     end
     context "extname is docx" do
-      before do
-        @file_path = "fixtures/sample.docx"
-      end
+      before { @file_path = "fixtures/sample.docx" }
       it { subject.should == ["word/document.xml"] }
     end
 
     context "extname is xlsx" do
-      before do
-        @file_path = "fixtures/sample.xlsx"
-      end
+      before { @file_path = "fixtures/sample.xlsx" }
       it { subject.should == ["xl/comments1.xml", "xl/sharedStrings.xml", "xl/worksheets/sheet1.xml", "xl/worksheets/sheet2.xml"] }
     end
 
     context "extname is pptx" do
-      before do
-        @file_path = "fixtures/sample.pptx"
-      end
+      before { @file_path = "fixtures/sample.pptx" }
       it { subject.should == ["ppt/slides/slide1.xml", "ppt/slides/slide2.xml"] }
     end
+
   end
 
   context "#convert" do
     subject do
-      archives = Zip::Archive.open(@file_path) { |archive| archive.map(&:name) }
-      pages = Officex2str.send(:pickup_pages, @file_path, archives)
-      xmls = Officex2str.send(:extract_xmls, @file_path, pages)
       Officex2str.convert(@file_path)
     end
     context "extname is xlsx" do
-      before do
-        @file_path = "fixtures/sample.xlsx"
-      end
+      before { @file_path = "fixtures/sample.xlsx" }
       it do
         subject.should include("複数シート対応")
         subject.should include("ソニックガーデン")
@@ -52,9 +64,7 @@ describe Officex2str do
     end
 
     context "extname is docx" do
-      before do
-        @file_path = "fixtures/sample.docx"
-      end
+      before { @file_path = "fixtures/sample.docx" }
       it do
         subject.should include("複数ページ対応")
         subject.should include("ソニックガーデン")
@@ -65,9 +75,7 @@ describe Officex2str do
     end
 
     context "extname is pptx" do
-      before do
-        @file_path = "fixtures/sample.pptx"
-      end
+      before { @file_path = "fixtures/sample.pptx" }
       it do
         subject.should include("Aタイトル")
         subject.should include("Aサブタイトル")
@@ -78,6 +86,6 @@ describe Officex2str do
         subject.should_not include("sheet")
       end
     end
-
   end
+
 end

metadata

@@ -5,8 +5,8 @@ version: !ruby/object:Gem::Version
   segments: 
   - 0
   - 0
-  - 3
-  version: 0.0.3
+  - 4
+  version: 0.0.4
 platform: ruby
 authors: 
 - interu
@@ -14,7 +14,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2012-07-18 00:00:00 +09:00
+date: 2012-07-19 00:00:00 +09:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency 
@@ -94,6 +94,7 @@ files:
 - Rakefile
 - fixtures/sample.docx
 - fixtures/sample.pptx
+- fixtures/sample.txt
 - fixtures/sample.xlsx
 - lib/officex2str.rb
 - lib/officex2str/version.rb