datacatalog-importer 0.1.7 → 0.1.8

data/VERSION

@@ -1 +1 @@
-0.1.7
+0.1.8

data/datacatalog-importer.gemspec

@@ -5,11 +5,11 @@
 
 Gem::Specification.new do |s|
   s.name = %q{datacatalog-importer}
-  s.version = "0.1.7"
+  s.version = "0.1.8"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
   s.authors = ["David James"]
-  s.date = %q{2010-03-12}
+  s.date = %q{2010-03-16}
   s.description = %q{This framework makes it easier to write importers for the National Data Catalog.}
   s.email = %q{djames@sunlightfoundation.com}
   s.extra_rdoc_files = [

data/lib/pusher.rb

@@ -7,6 +7,10 @@ module DataCatalog
       include DataCatalog::Importer::Shared
       
       REQUIRED = %w(api_key base_uri cache_folder)
+      
+      # These keys should not be passed along directly; they need to be
+      # examined so that real ID's can be passed along instead.
+      LOOKUP_KEYS = [:organization, :downloads]
 
       def initialize(options)
         REQUIRED.each do |r|
@@ -92,23 +96,24 @@ module DataCatalog
       end
       
       def find_organization_by(field, name)
-        docs = DataCatalog::Organization.all(field => name)
+        verify_one_result(field, name, "Organization") do
+          DataCatalog::Organization.all(field => name)
+        end
+      end
+      
+      def verify_one_result(field, name, model)
+        docs = yield
         n = docs.length
         case n
-        when 0
-          nil
-        when 1
-          docs[0]
-        else
-          multiple_matches("Organization", { field => name }, n)
+        when 0 then nil
+        when 1 then docs[0]
+        else multiple_matches(model, { field => name }, n)
         end
       end
       
       # Important: do not modify data
       def create_or_update_source(data)
-        data = data.reject do |key, value|
-          [:organization, :downloads].include?(key)
-        end
+        data = data.reject { |k, v| LOOKUP_KEYS.include?(k) }
         docs = DataCatalog::Source.all(:url => data[:url])
         n = docs.length
         case n
@@ -157,21 +162,29 @@ module DataCatalog
       #
       # Note: modifies data (that is why I use the !)
       def link_to_existing_organization!(data, organization_id_key)
-        organization_hash = data.delete(:organization)
-        raise "Could not find :organization key" unless organization_hash
-        name = organization_hash[:name]
-        url  = organization_hash[:url]
+        hash = data.delete(:organization)
+        raise "Could not find :organization key" unless hash
 
-        organization, error_message = if url
-          [find_organization_by(:url, url), "url : #{url}"]
-        elsif name
-          [find_organization_by(:name, name), "name : #{name}"]
+        keys = [:url, :home_url, :name]
+        unless hash.any? { |key, value| keys.include?(key) }
+          raise "Need #{keys.join(' or ')} to lookup an organization"
+        end
+
+        attempts = []
+        organization = nil
+        keys.each do |key|
+          value = hash[key]
+          organization = if value
+            attempts << "#{key}:#{value}"
+            find_organization_by(key, value)
+          end
+          break if organization
         end
 
         if organization
           data[organization_id_key] = organization.id
         else
-          puts "  - Could not find organization with #{error_message}"
+          puts "- Could not find organization with #{attempts.join(' or ')}"
         end
         true # return value not important
       end
@@ -179,8 +192,8 @@ module DataCatalog
       # ---
       
       def create_failed(model, text, data, error)
-        puts "  - Failed. #{error}"
-        puts "    Uploading Report to API."
+        puts "- Failed. #{error}"
+        puts "  Uploading Report to API."
         DataCatalog::Report.create({
           :status => "new",
           :text   => "Cannot create #{model} : #{text}",
@@ -193,8 +206,8 @@ module DataCatalog
       end
       
       def update_failed(model, data, id, error)
-        puts "  - Failed. #{error}"
-        puts "    Uploading Report to API."
+        puts "- Failed. #{error}"
+        puts "  Uploading Report to API."
         DataCatalog::Report.create({
           :status => "new",
           :text   => "Cannot update #{model} of id : #{id}",
@@ -207,13 +220,13 @@ module DataCatalog
       end  
       
       def multiple_matches(model, data, n)
-        puts "  - Failed. #{n} matches for #{model}."
-        puts "    Uploading Report to API."
+        puts "- Failed. #{n} matches for #{model}."
+        puts "  Uploading Report to API."
         DataCatalog::Report.create({
           :status => "new",
           :text   => "Multiple matches for url : #{data[:url]}",
           :object => {
-            :error  => "Cannot automatically update #{model}since there " +
+            :error  => "Cannot automatically update #{model} since there " +
               "are #{n} matches for url : #{data[:url]}",
             :params => data,
           },

metadata

@@ -5,8 +5,8 @@ version: !ruby/object:Gem::Version
   segments: 
   - 0
   - 1
-  - 7
-  version: 0.1.7
+  - 8
+  version: 0.1.8
 platform: ruby
 authors: 
 - David James
@@ -14,7 +14,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2010-03-12 00:00:00 -05:00
+date: 2010-03-16 00:00:00 -04:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency