extractor.rb in solrizer-1.0.1

- old
+ new

@@ -1,7 +1,5 @@
-require 'solr'
-require 'rexml/document'
 require "nokogiri"
 require 'yaml'
 
 module Solrizer
   
@@ -10,39 +8,53 @@
 #
 # Each of the Solrizer implementations provides its own Extractor module that extends the behaviors of Solrizer::Extractor
 # with methods specific to that implementation (ie. extract_tag, extract_rels_ext, xml_to_solr, html_to_solr)
 #
 class Extractor
-
-  # Populates a solr doc with values from a hash.  
-  # Accepts two forms of hashes:
-  # => {'technology'=>["t1", "t2"], 'company'=>"c1", "person"=>["p1", "p2"]}
-  # or
-  # => {:facets => {'technology'=>["t1", "t2"], 'company'=>"c1", "person"=>["p1", "p2"]} }
-  #
-  # Note that values for individual fields can be a single string or an array of strings.
-  def extract_hash( input_hash, solr_doc=Solr::Document.new )    
-    facets = input_hash.has_key?(:facets) ? input_hash[:facets] : input_hash
-    facets.each_pair do |facet_name, value|
-      case value.class.to_s
-      when "String"
-        solr_doc << Solr::Field.new( :"#{facet_name}_facet" => "#{value}" )
-      when "Array"
-        value.each { |v| solr_doc << Solr::Field.new( :"#{facet_name}_facet" => "#{v}" ) } 
-      end
+  
+  # Insert +field_value+ for +field_name+ into +solr_doc+
+  # Ensures that field values are always appended to arrays within the values hash. 
+  # Also ensures that values are run through format_node_value
+  # @param [Hash] solr_doc
+  # @param [String] field_name
+  # @param [String] field_value
+  def self.insert_solr_field_value(solr_doc, field_name, field_value)
+    formatted_value = self.format_node_value(field_value)
+    if solr_doc.has_key?(field_name)
+      solr_doc[field_name] << formatted_value
+    else
+      solr_doc.merge!( {field_name => [formatted_value]} ) 
     end
-    
-    if input_hash.has_key?(:symbols) 
-      input_hash[:symbols].each do |symbol_name, value|
-        case value.class.to_s
-        when "String"
-          solr_doc << Solr::Field.new( :"#{symbol_name}_s" => "#{value}" )
-	      when "Array"
-          value.each { |v| solr_doc << Solr::Field.new( :"#{symbol_name}_s" => "#{v}" ) } 
-        end
-      end
-    end
     return solr_doc
+  end
+  
+  # Strips the majority of whitespace from the values array and then joins them with a single blank delimitter
+  # @param [Array] values Array of strings representing the values returned 
+  def self.format_node_value values
+    values.map{|val| val.gsub(/\s+/,' ').strip}.join(" ")
+  end
+  
+  # Instance Methods
+  
+  # Alias for Solrizer::Extractor#insert_solr_field_value
+  def insert_solr_field_value(solr_doc, field_name, field_value)
+    Solrizer::Extractor.insert_solr_field_value(solr_doc, field_name, field_value)
+  end
+  
+  # Alias for Solrizer::Extractor#format_node_value
+  def format_node_value values
+    Solrizer::Extractor.format_node_value(values)
+  end
+  
+  # Deprecated.
+  # merges input_hash into solr_hash
+  # @param [Hash] input_hash the input hash of values
+  # @param [Hash] solr_hash the solr values hash to add the values into
+  # @return [Hash] the populated Solr values hash
+  # 
+  def extract_hash( input_hash, solr_hash=Hash.new )   
+    warn "[DEPRECATION] `extract_hash` is deprecated.  Just pass values directly into your solr values hash" 
+    return solr_hash.merge!(input_hash)
   end
   
 end
 end