Add support for classification of non-fiction books.

[quanlib.git] / classify / classset.rb

diff --git a/classify/classset.rb b/classify/classset.rb
new file mode 100644 (file)
index 0000000..710db7a
--- /dev/null
+++ b/classify/classset.rb
@@ -0,0 +1,120 @@
+require 'csv'
+
+require 'bookclass'
+
+class ClassSet
+  @@class_csv_file = 'class.csv'
+
+  def initialize
+    @entries = {}
+    load!(@@class_csv_file)
+  end
+
+  def add!(info)
+    key = construct_key(info.grouping, info.title)
+    @entries[key] = info
+  end
+
+  def construct_key(author_grouping, title)
+    author_grouping.to_s + '|' + title.to_s
+  end
+
+  def get(author_grouping, title)
+    key = construct_key(author_grouping, title)
+    if @entries.has_key?(key)
+      return @entries[key]
+    else
+      return nil
+    end
+  end
+
+  def has_key?(author_grouping, title)
+    @entries.has_key?(construct_key(author_grouping, title))
+  end
+
+  def ensure_contains!(info)
+    if ! has_key?(info.grouping, info.title)
+      add!(info)
+    end
+  end
+
+  def inspect 
+    data = []
+
+    if nil != @entries 
+      data.push('entries=' + @entries.inspect + '') 
+    end
+
+    return '(ClassSet:' + data.join(',') + ')'
+  end
+
+  def load!(file_name)
+    first = true
+    @entries = {}
+
+    if ! File.exist?(file_name)
+      puts 'WARNING:  file "' + file_name + '" not found.'
+      return 
+    end
+
+    File.open(file_name, 'r:UTF-8') do |fd|
+      csv = CSV.new(fd)
+      csv.to_a.each do |row|
+        if first
+          first = false
+        elsif row.length >= 6
+          ddc = row[0]
+          lcc = row[1]
+          grouping = row[2]
+          author = row[3]
+          title = row[4]
+          fast = []
+          if nil != row[5]
+            fast = row[5].split(';')
+          end
+  
+          bookclass = BookClass.new(grouping, title)
+          bookclass.ddc = ddc
+          bookclass.lcc = lcc
+          bookclass.author = author
+  
+          fast.each do |id|
+            bookclass.add_fast(id)
+          end
+  
+          key = construct_key(grouping, title)
+          @entries[key] = bookclass
+        end
+      end
+    end
+  end
+
+  def save(file_name)
+    CSV.open(file_name, 'w:UTF-8') do |csv|
+      csv << ['Dewey', 'LCC', 'Grouping', 'Author', 'Title', 'FAST']
+
+      @entries.keys.sort.each do |key|
+        info = @entries[key]
+
+        ddc = info.ddc
+        lcc = info.lcc
+        grouping = info.grouping
+        author = info.author
+        title = info.title
+        fast_list = info.fast
+        fast_ids = []
+        fast_list.each do |tuple|
+          fast_ids.push(tuple[0])
+        end
+        fast = fast_ids.join(';')
+        
+        csv << [ ddc, lcc, grouping, author, title, fast ]
+      end
+    end
+  end
+
+  def save_state
+    save(@@class_csv_file)
+  end
+end
+