fig bugs in csv parsing and don't throw an exception on blank token file

lib/reckon/app.rb

@@ -80,7 +80,7 @@ def learn_from_ledger(ledger)
 
     # Add tokens from account_tokens_file to accounts
     def extract_account_tokens(subtree, account = nil)
-      if subtree.nil?
+      if subtree.nil? || !subtree
         puts "Warning: empty #{account} tree"
         {}
       elsif subtree.is_a?(Array)
@@ -168,15 +168,18 @@ def each_row_backwards
                   :money => @csv_parser.money_for(index),
                   :description => @csv_parser.description_for(index) }
       end
-      rows.sort_by { |n| [n[:date], -n[:money], n[:description]] }.each { |row| yield row }
+      rows.sort_by { |n|
+        [n[:date], -n[:money], n[:description]]
+      }.each { |row| yield row }
     end
 
     def print_transaction(rows, fh = $stdout)
       str = "\n"
       header = %w[Date Amount Description Note]
       maxes = header.map(&:length)
-
-      rows = rows.map { |r| [r[:pretty_date], r[:pretty_money], r[:description], r[:note]] }
+      rows = rows.map { |r|
+        [r[:pretty_date], r[:pretty_money], r[:description], r[:note]]
+      }
 
       rows.each do |r|
         r.length.times { |i| l = r[i] ? r[i].length : 0; maxes[i] = l if maxes[i] < l }

lib/reckon/cosine_similarity.rb

@@ -12,6 +12,7 @@
 # These weights and measures are used to suggest which account a transaction should be
 # assigned to.
 module Reckon
+  # Calculates cosine similarity for tf/idf
   class CosineSimilarity
     DocumentInfo = Struct.new(:tokens, :accounts)
 

lib/reckon/csv_parser.rb

@@ -64,13 +64,13 @@ def row(index)
     private
 
     def filter_csv
-      if options[:ignore_columns]
-        new_columns = []
-        columns.each_with_index do |column, index|
-          new_columns << column unless options[:ignore_columns].include?(index + 1)
-        end
-        @columns = new_columns
+      return unless options[:ignore_columns]
+
+      new_columns = []
+      columns.each_with_index do |column, index|
+        new_columns << (options[:ignore_columns].include?(index + 1) ? [''] * column.length : column)
       end
+      @columns = new_columns
     end
 
     def evaluate_columns(cols)
@@ -222,7 +222,8 @@ def parse(data, filename = nil)
       # convert to a stringio object to handle multi-line fields
       parser_opts = {
         col_sep: separator,
-        skip_blanks: true
+        skip_blanks: true,
+        row_sep: :auto
       }
       begin
         rows = CSV.parse(StringIO.new(data), **parser_opts)
@@ -235,7 +236,7 @@ def parse(data, filename = nil)
           index = data.index("\n", index) + 1 # skip over newline character
           count += 1
         end
-        rows = CSV.parse(StringIO.new(data[index..-1]), **parser_opts)
+        rows = CSV.parse(StringIO.new(data[index..]), **parser_opts)
         rows[0..-footer_lines_to_skip]
       end
     end

lib/reckon/ledger_parser.rb

@@ -175,7 +175,7 @@ def to_csv(ledger)
     end
 
     def format_row(row, line1, line2)
-      note = row[:note] ? "\t; row[:note]" : ""
+      note = row[:note] ? "\t; #{row[:note]}" : ""
       out = "#{row[:pretty_date]}\t#{row[:description]}#{note}\n"
       out += "\t#{line1.first}\t\t\t#{line1.last}\n"
       out += "\t#{line2.first}\t\t\t#{line2.last}\n\n"