xtotxt 0.4 → 0.5

data/lib/xtotxt.rb

@@ -1,6 +1,11 @@
 require 'yaml'
 
+class XtotxtError < StandardError; end
+
 class Xtotxt
+  VERSION = 0.5
+  SUPPORTED_EXTENSIONS = %w{pdf doc docx odt rtf html}
+
   @@config_file_name = "xtotxt.yml"
   @@dirs_to_check    = %w{. ~ /etc}
   @@ext              = nil
@@ -15,6 +20,7 @@ class Xtotxt
         end
       end
     end
+    @@ext = @ext_default
   end
 
   def convert(input_file_name)
@@ -22,7 +28,7 @@ class Xtotxt
 
     ext = path_list.pop
 
-    raise("not a supported document extension: #{ext}") unless %w{pdf doc docx odt rtf html}.member?(ext)
+    raise XtotxtError.new("not a supported document extension: #{ext}") unless SUPPORTED_EXTENSIONS.member?(ext)
 
     output_file = (path_list << "txt").join(".")
 
@@ -40,7 +46,7 @@ class Xtotxt
     when "html":
         "#{@ext[:html]} -o #{output_file} #{input_file_name}"
     else
-        raise "have no way to convert #{ext} yet"
+        raise XtotxtError.new("have no way to convert #{ext} yet")
     end
 
     #puts "executing: #{command_line}"
@@ -49,7 +55,7 @@ class Xtotxt
     text = if $? == 0
       File.read(output_file)
     else
-      raise "Failed to convert #{input_file_name}. Exit status: #{$?.exitstatus}.  Output: #{command_output}"
+      raise XtotxtError.new("Failed to convert #{input_file_name}. Exit status: #{$?.exitstatus}.  Output: #{command_output}")
     end
 
     case ext

data/spec/xtotxt.yml

@@ -1,6 +1,6 @@
 :pdf:  "/opt/local/bin/xpdf-pdftotext"
 :doc:  "/opt/local/bin/antiword"
 :docx: "/usr/local/bin/docx2txt.pl"
-:odt:  "/usr/local/bin/odt2txt"
+:odt:  "/opt/local/bin/odt2txt"
 :rtf:  "/opt/local/bin/unrtf -P /opt/local/lib/unrtf"
 :html: "/opt/local/bin/html2text"

data/spec/xtotxt_spec.rb

@@ -4,6 +4,7 @@ require 'xtotxt'
 describe Xtotxt do
   before(:all) do
     @input_prefix = "#{Pathname.new(__FILE__).dirname}/fixtures/test"
+    @text = "three pigheaded piglets had a plan"
   end
 
   before do
@@ -36,37 +37,37 @@ describe Xtotxt do
   it "converts a pdf document correctly" do
     text = @x.convert("#{@input_prefix}.pdf")
 
-    text.should == "three pigheaded piglets had a plan\n\n\f"
+    text.strip.should == @text
   end
 
   it "converts a doc document correctly" do
     text = @x.convert("#{@input_prefix}.doc")
 
-    text.should == "\nthree pigheaded piglets had a plan\n\n"
+    text.strip.should == @text
   end
 
   it "converts a docx document correctly" do
     text = @x.convert("#{@input_prefix}.docx")
 
-    text.should == "three pigheaded piglets had a plan\n\n"
+    text.strip.should == @text
   end
 
   it "converts an odt document correctly" do
     text = @x.convert("#{@input_prefix}.odt")
 
-    text.should == "\nthree pigheaded piglets had a plan\n\n"
+    text.strip.should == @text
   end
 
   it "converts an rtf document correctly" do
     text = @x.convert("#{@input_prefix}.rtf")
 
-    text.should == "three pigheaded piglets had a plan\n"
+    text.strip.should == @text
   end
 
   it "converts an html document correctly" do
     text = @x.convert("#{@input_prefix}.html")
 
-    text.should == "three pigheaded piglets had a plan\n"
+    text.strip.should == @text
   end
 
 

data/xtotxt.gemspec

@@ -1,6 +1,6 @@
 # -*- encoding: utf-8 -*-
 $:.push File.expand_path("../lib", __FILE__)
-require "xtotxt/version"
+require "lib/xtotxt"
 
 Gem::Specification.new do |s|
   s.name        = "xtotxt"

metadata

@@ -1,12 +1,12 @@
 --- !ruby/object:Gem::Specification 
 name: xtotxt
 version: !ruby/object:Gem::Version 
-  hash: 3
+  hash: 1
   prerelease: 
   segments: 
   - 0
-  - 4
-  version: "0.4"
+  - 5
+  version: "0.5"
 platform: ruby
 authors: 
 - Alexy Khrabrov
@@ -14,7 +14,8 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2011-09-23 00:00:00 Z
+date: 2011-09-30 00:00:00 -07:00
+default_executable: 
 dependencies: []
 
 description: A simple wrapper calling, for each supported input format, a given command-line tool
@@ -41,12 +42,10 @@ files:
 - spec/xtotxt.yml
 - spec/xtotxt_spec.rb
 - xtotxt.gemspec
-- xtotxt/version.rb
+has_rdoc: true
 homepage: http://www.topprospect.com
 licenses: []
 
-metadata: {}
-
 post_install_message: 
 rdoc_options: []
 
@@ -73,9 +72,9 @@ required_rubygems_version: !ruby/object:Gem::Requirement
 requirements: []
 
 rubyforge_project: xtotxt
-rubygems_version: 1.8.10
+rubygems_version: 1.6.2
 signing_key: 
-specification_version: 4
+specification_version: 3
 summary: Convert pdf, doc and docx to plain text
 test_files: 
 - spec/fixtures/test.doc

data/xtotxt/version.rb

@@ -1,3 +0,0 @@
-module Xtotxt
-  VERSION = 0.4
-end