remote_table 1.1.6 → 1.1.7

data/Rakefile

@@ -21,3 +21,5 @@ begin
 rescue LoadError
    puts "Rdoc is not available"
 end
+
+task :default => :test

data/lib/remote_table/format.rb

@@ -1,3 +1,7 @@
+if ::RUBY_VERSION >= '1.9'
+  require 'ensure/encoding'
+end
+
 class RemoteTable  
   class Format
     autoload :Excel, 'remote_table/format/excel'
@@ -16,6 +20,14 @@ class RemoteTable
       @t = t
     end
     
+    def utf8(str)
+      if ::RUBY_VERSION >= '1.9'
+        str.ensure_encoding 'UTF-8', :external_encoding => t.properties.encoding, :invalid_characters => :transcode
+      else
+        str
+      end
+    end
+    
     include ::Enumerable
     def each
       raise "must be defined by format"

data/lib/remote_table/format/delimited.rb

@@ -1,9 +1,10 @@
 if RUBY_VERSION >= '1.9'
   require 'csv'
-  ::FasterCSV = ::CSV
+  ::RemoteTable::CSV = ::CSV
 else
   begin
     require 'fastercsv'
+    ::RemoteTable::CSV = ::FasterCSV
   rescue ::LoadError
     $stderr.puts "[remote_table gem] You probably need to manually install the fastercsv gem and/or require it in your Gemfile."
     raise $!
@@ -15,25 +16,24 @@ class RemoteTable
     class Delimited < Format
       include Textual
       def each(&blk)
-        convert_file_to_utf8!
         remove_useless_characters!
         skip_rows!
-        ::FasterCSV.foreach(t.local_file.path, fastercsv_options) do |row|
+        CSV.foreach(t.local_file.path, fastercsv_options) do |row|
           ordered_hash = ::ActiveSupport::OrderedHash.new
           filled_values = 0
           case row
-          when ::FasterCSV::Row
+          when CSV::Row
             row.each do |header, value|
               next if header.blank?
               value = '' if value.nil?
-              ordered_hash[header] = value
+              ordered_hash[header] = utf8 value
               filled_values += 1 if value.present?
             end
           when ::Array
             index = 0
             row.each do |value|
               value = '' if value.nil?
-              ordered_hash[index] = value
+              ordered_hash[index] = utf8 value
               filled_values += 1 if value.present?
               index += 1
             end

data/lib/remote_table/format/fixed_width.rb

@@ -4,13 +4,15 @@ class RemoteTable
     class FixedWidth < Format
       include Textual
       def each(&blk)
-        convert_file_to_utf8!
         remove_useless_characters!
         crop_rows!
         skip_rows!
         cut_columns!
         parser.parse[:rows].each do |hash|
           hash.reject! { |k, v| k.blank? }
+          hash.each do |k, v|
+            hash[k] = utf8 v
+          end
           yield hash if t.properties.keep_blank_rows or hash.any? { |k, v| v.present? }
         end
       ensure

data/lib/remote_table/format/html.rb

@@ -5,11 +5,10 @@ class RemoteTable
     class HTML < Format
       include Textual
       def each(&blk)
-        convert_file_to_utf8!
         remove_useless_characters!
         html_headers = (t.properties.headers.is_a?(::Array)) ? t.properties.headers : nil
         ::Nokogiri::HTML(unescaped_html_without_soft_hyphens, nil, 'UTF-8').xpath(t.properties.row_xpath).each do |row|
-          values = row.xpath(t.properties.column_xpath).map { |td| td.content.gsub(/\s+/, ' ').strip }
+          values = row.xpath(t.properties.column_xpath).map { |td| utf8 td.content.gsub(/\s+/, ' ').strip }
           if html_headers.nil?
             html_headers = values
             next

data/lib/remote_table/format/mixins/textual.rb

@@ -2,11 +2,7 @@ require 'fileutils'
 require 'escape'
 class RemoteTable
   class Format
-    module Textual
-      def convert_file_to_utf8!
-        ::RemoteTable.executor.bang t.local_file.path, "iconv -c -f #{::Escape.shell_single_word t.properties.encoding} -t UTF-8"
-      end
-      
+    module Textual      
       USELESS_CHARACTERS = [
         '\xef\xbb\xbf',   # UTF-8 byte order mark
         '\xc2\xad'        # soft hyphen, often inserted by MS Office (html: ­)

data/lib/remote_table/properties.rb

@@ -48,11 +48,11 @@ class RemoteTable
       t.options['skip'].to_i
     end
     
-    # The encoding
+    # Likely external encoding
     #
     # Default: "UTF-8"
     def encoding
-      t.options['encoding'] || 'UTF-8'
+      @encoding ||= ::Array.wrap(t.options['encoding'] || [ 'UTF-8', 'US-ASCII', 'ASCII-8BIT' ])
     end
     
     # The delimiter

data/lib/remote_table/version.rb

@@ -1,3 +1,3 @@
 class RemoteTable
-  VERSION = "1.1.6"
+  VERSION = "1.1.7"
 end

data/remote_table.gemspec

@@ -30,6 +30,7 @@ Gem::Specification.new do |s|
   s.add_dependency 'google-spreadsheet-ruby' #roo
   s.add_dependency 'escape', '>=0.0.4'
   s.add_dependency 'posix-spawn'
+  s.add_dependency 'ensure-encoding'
   unless RUBY_VERSION >= '1.9'
     s.add_dependency 'fastercsv', '>=1.5.0'
   end

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: remote_table
 version: !ruby/object:Gem::Version 
-  hash: 31
+  hash: 29
   prerelease: 
   segments: 
   - 1
   - 1
-  - 6
-  version: 1.1.6
+  - 7
+  version: 1.1.7
 platform: ruby
 authors: 
 - Seamus Abshere
@@ -16,7 +16,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2011-03-31 00:00:00 -05:00
+date: 2011-04-20 00:00:00 -05:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency 
@@ -183,9 +183,23 @@ dependencies:
   type: :runtime
   version_requirements: *id011
 - !ruby/object:Gem::Dependency 
-  name: fastercsv
+  name: ensure-encoding
   prerelease: false
   requirement: &id012 !ruby/object:Gem::Requirement 
+    none: false
+    requirements: 
+    - - ">="
+      - !ruby/object:Gem::Version 
+        hash: 3
+        segments: 
+        - 0
+        version: "0"
+  type: :runtime
+  version_requirements: *id012
+- !ruby/object:Gem::Dependency 
+  name: fastercsv
+  prerelease: false
+  requirement: &id013 !ruby/object:Gem::Requirement 
     none: false
     requirements: 
     - - ">="
@@ -197,11 +211,11 @@ dependencies:
         - 0
         version: 1.5.0
   type: :runtime
-  version_requirements: *id012
+  version_requirements: *id013
 - !ruby/object:Gem::Dependency 
   name: errata
   prerelease: false
-  requirement: &id013 !ruby/object:Gem::Requirement 
+  requirement: &id014 !ruby/object:Gem::Requirement 
     none: false
     requirements: 
     - - ">="
@@ -213,11 +227,11 @@ dependencies:
         - 0
         version: 0.2.0
   type: :development
-  version_requirements: *id013
+  version_requirements: *id014
 - !ruby/object:Gem::Dependency 
   name: test-unit
   prerelease: false
-  requirement: &id014 !ruby/object:Gem::Requirement 
+  requirement: &id015 !ruby/object:Gem::Requirement 
     none: false
     requirements: 
     - - ">="
@@ -227,11 +241,11 @@ dependencies:
         - 0
         version: "0"
   type: :development
-  version_requirements: *id014
+  version_requirements: *id015
 - !ruby/object:Gem::Dependency 
   name: shoulda
   prerelease: false
-  requirement: &id015 !ruby/object:Gem::Requirement 
+  requirement: &id016 !ruby/object:Gem::Requirement 
     none: false
     requirements: 
     - - ">="
@@ -241,11 +255,11 @@ dependencies:
         - 0
         version: "0"
   type: :development
-  version_requirements: *id015
+  version_requirements: *id016
 - !ruby/object:Gem::Dependency 
   name: ruby-debug
   prerelease: false
-  requirement: &id016 !ruby/object:Gem::Requirement 
+  requirement: &id017 !ruby/object:Gem::Requirement 
     none: false
     requirements: 
     - - ">="
@@ -255,7 +269,7 @@ dependencies:
         - 0
         version: "0"
   type: :development
-  version_requirements: *id016
+  version_requirements: *id017
 description: Gives you a standard way to parse various formats and treat them as an array of hashes.
 email: 
 - seamus@abshere.net
@@ -268,7 +282,6 @@ extra_rdoc_files: []
 files: 
 - .document
 - .gitignore
-- CHANGELOG
 - Gemfile
 - LICENSE
 - README.rdoc

data/CHANGELOG

@@ -1,12 +0,0 @@
-1.1.0
-* Fixed difference in row hashes between Ruby 1.8 and 1.9
-* Properly required fastercsv depending on Ruby version
-1.0.0
-* Using bundler and gemspec instead of jeweler
-* Refactored to follow more Ruby conventions
-* Suggesting new syntax that looks more like an Enumerable... t[5] instead of t.rows[5]
-* Switching to string option keys (but old syntax is supported)
-[...no changelog for 0.1.6--1.0.0...sorry]
-0.1.6
-* For CSVs, force convert headers using String#toutf8. :encoding => 'N'|'U' didn't work.
-* Fix handling of long urls when passing off to Tempfile.