remote_table 0.2.8 → 0.2.9

data/VERSION

@@ -1 +1 @@
-0.2.8
+0.2.9

data/lib/remote_table.rb

@@ -1,6 +1,11 @@
 require 'digest/md5'
 require 'iconv'
 require 'active_support'
+begin; require 'active_support/core_ext/object/blank'; rescue MissingSourceFile; end
+begin; require 'active_support/core_ext/string/inflections'; rescue MissingSourceFile; end
+begin; require 'active_support/core_ext/array/wrap'; rescue MissingSourceFile; end
+begin; require 'active_support/core_ext/hash/except'; rescue MissingSourceFile; end
+begin; require 'active_support/core_ext/string/starts_ends_with'; rescue MissingSourceFile; end
 require 'tempfile'
 require 'fastercsv'
 require 'slither'

data/lib/remote_table/request.rb

@@ -1,51 +1,46 @@
 class RemoteTable
   class Request
-    attr_accessor :url, :post_data, :username, :password
+    attr_accessor :parsed_url, :post_data, :username, :password
     attr_accessor :form_data
     
     # TODO: support post_data
     # TODO: support HTTP basic auth
     def initialize(bus)
-      @url = bus[:url] or raise "need url"
+      @parsed_url = URI.parse(bus[:url]) or raise "need url"
+      if @parsed_url.host == 'spreadsheets.google.com' and (bus[:format].blank? or bus[:format].to_s == 'csv')
+        @parsed_url.query = 'output=csv&' + @parsed_url.query.sub(/\&*output=.*(\&|\z)/, '')
+      end
       @form_data = bus[:form_data]
     end
     
     def download
-      path = ::File.join(staging_dir_path, 'REMOTE_TABLE_PACKAGE')
-      cmd = %{
-        curl \
-        --silent \
-        --header "Expect: " \
-        --location \
-        #{"--data \"#{form_data}\"" if form_data.present?} \
-        "#{url_with_google_docs_handling}" \
-        --output "#{path}"
-      }
-      `#{cmd}`
-      path
+      path = ::File.join staging_dir_path, 'REMOTE_TABLE_PACKAGE'
+      if parsed_url.scheme == 'file'
+        parsed_url.path
+      else
+        cmd = %{
+          curl \
+          --silent \
+          --header "Expect: " \
+          --location \
+          #{"--data \"#{form_data}\"" if form_data.present?} \
+          "#{parsed_url}" \
+          --output "#{path}"
+        }
+        `#{cmd}`
+        path
+      end
     end
     
     private
     
-    def staging_dir_path
-      path = tempfile_path_from_url
-      FileUtils.rm_f(path)
-      FileUtils.mkdir(path)
-      at_exit { FileUtils.rm_rf(path) }
-      path
-    end
-    
-    def tempfile_path_from_url
-      Tempfile.open(url.gsub(/[^a-z0-9]+/i, '_')[0,100]).path
-    end
-    
-    def url_with_google_docs_handling
-      url = self.url
-      if url.include?('spreadsheets.google.com')
-        url = url.gsub(/\&output=.*(\&|\z)/, '')
-        url << "&output=csv"
-      end
-      url
+    def staging_dir_path    
+      return @_staging_dir_path if @_staging_dir_path
+      @_staging_dir_path = Tempfile.open(parsed_url.to_s.gsub(/[^a-z0-9]+/i, '_')[0,100]).path
+      FileUtils.rm_f @_staging_dir_path
+      FileUtils.mkdir @_staging_dir_path
+      at_exit { FileUtils.rm_rf @_staging_dir_path }
+      @_staging_dir_path
     end
   end
 end

data/remote_table.gemspec

@@ -5,11 +5,11 @@
 
 Gem::Specification.new do |s|
   s.name = %q{remote_table}
-  s.version = "0.2.8"
+  s.version = "0.2.9"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
   s.authors = ["Seamus Abshere", "Andy Rossmeissl"]
-  s.date = %q{2010-03-30}
+  s.date = %q{2010-04-15}
   s.description = %q{Remotely open and parse Excel XLS, ODS, CSV and fixed-width tables.}
   s.email = %q{seamus@abshere.net}
   s.extra_rdoc_files = [

data/test/remote_table_test.rb

@@ -11,16 +11,18 @@ class FuelOilParser
   end
   def apply(row)
     virtual_rows = []
-    row.keys.grep(/(.*) Residual Fuel Oil/) do |location_column_name|
+    row.keys.grep(/(.+) Residual Fuel Oil/) do |location_column_name|
+      first_part = $1
       next if (cost = row[location_column_name]).blank? or (date = row['Date']).blank?
-      if $1.starts_with?('U.S.')
+      if first_part.starts_with?('U.S.')
         locatable = "united_states (Country)"
-      elsif $1.include?('PADD')
-        /\(PADD (.*)\)/.match($1)
-        next if $1 == '1' # skip PADD 1 because we always prefer subdistricts
-        locatable = "#{$1} (PetroleumAdministrationForDefenseDistrict)"
+      elsif first_part.include?('PADD')
+        /\(PADD (.*)\)/.match(first_part)
+        padd_part = $1
+        next if padd_part == '1' # skip PADD 1 because we always prefer subdistricts
+        locatable = "#{padd_part} (PetroleumAdministrationForDefenseDistrict)"
       else
-        locatable = "#{$1} (State)"
+        locatable = "#{first_part} (State)"
       end
       date = Time.parse(date)
       virtual_rows << {
@@ -160,7 +162,7 @@ class RemoteTableTest < Test::Unit::TestCase
       assert_equal row.except('row_hash'), @test2_rows[index]
     end
   end
-
+  
   should "read fixed width correctly, keeping blank rows" do
     t = RemoteTable.new(:url => 'http://cloud.github.com/downloads/seamusabshere/remote_table/test2.fixed_width.txt',
                         :format => :fixed_width,
@@ -257,6 +259,7 @@ class RemoteTableTest < Test::Unit::TestCase
   
     t = RemoteTable.new(:url => 'http://tonto.eia.doe.gov/dnav/pet/xls/PET_PRI_RESID_A_EPPR_PTA_CPGAL_M.xls',
                         :transform => { :class => FuelOilParser })
+  
     assert t.rows.include?(ma_1990_01)
     assert t.rows.include?(ga_1990_01)
   end

metadata

@@ -5,8 +5,8 @@ version: !ruby/object:Gem::Version
   segments: 
   - 0
   - 2
-  - 8
-  version: 0.2.8
+  - 9
+  version: 0.2.9
 platform: ruby
 authors: 
 - Seamus Abshere
@@ -15,7 +15,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2010-03-30 00:00:00 -04:00
+date: 2010-04-15 00:00:00 -04:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency