dbf 2.0.7 → 2.0.8

data/Rakefile

@@ -1,4 +1,3 @@
-require 'rubygems'
 require 'bundler/setup';
 Bundler.setup(:default, :development)
 
@@ -17,9 +16,3 @@ desc "Open an irb session preloaded with this library"
 task :console do
   sh "irb -rubygems -I lib -r dbf.rb"
 end
-
-# require 'metric_fu'
-# MetricFu::Configuration.run do |config|
-#   config.rcov[:test_files] = ['spec/**/*_spec.rb']
-#   config.rcov[:rcov_opts] << "-Ispec"
-# end

data/dbf.gemspec

@@ -14,22 +14,11 @@ Gem::Specification.new do |s|
 
   s.executables = ['dbf']
   s.rdoc_options = ['--charset=UTF-8']
-  s.extra_rdoc_files = ['README.md', 'CHANGELOG.md', 'MIT-LICENSE']
+  s.extra_rdoc_files = ['README.md', 'CHANGELOG.md', 'LICENSE']
   s.files = Dir['[A-Z]*', '{bin,docs,lib,spec}/**/*', 'dbf.gemspec']
   s.test_files = Dir.glob('spec/**/*_spec.rb')
   s.require_paths = ['lib']
 
   s.required_rubygems_version = '>= 1.3.0'
   s.add_dependency 'fastercsv', '~> 1.5.4'
-
-  s.add_development_dependency 'rspec'
-  s.add_development_dependency 'rake', '>= 0.9.2'
-
-  # if RUBY_VERSION.to_f >= 1.9
-  #   s.add_development_dependency 'ruby-debug19'
-  # elsif RUBY_VERSION != '1.8.6'
-  #   s.add_development_dependency 'ruby-debug'
-  # end
-  # s.add_development_dependency 'metric_fu'
 end
-

data/lib/dbf.rb

@@ -5,6 +5,7 @@ if CSV.const_defined? :Reader
   require 'fastercsv'
 end
 
+require 'dbf/schema'
 require 'dbf/record'
 require 'dbf/column/base'
 require 'dbf/column/dbase'

data/lib/dbf/column/base.rb

@@ -7,7 +7,7 @@ module DBF
     class NameError < StandardError; end
 
     class Base
-      attr_reader :name, :type, :length, :decimal
+      attr_reader :table, :name, :type, :length, :decimal
 
       # Initialize a new DBF::Column
       #
@@ -15,11 +15,22 @@ module DBF
       # @param [String] type
       # @param [Fixnum] length
       # @param [Fixnum] decimal
-      def initialize(name, type, length, decimal, version, encoding=nil)
-        @name, @type, @length, @decimal, @version, @encoding = clean(name), type, length, decimal, version, encoding
+      def initialize(table, name, type, length, decimal)
+        @table = table
+        @name = clean(name)
+        @type = type
+        @length = length
+        @decimal = decimal
+        @version = table.version
+        @encoding = table.encoding
+
+        unless length > 0
+          raise LengthError, "field length must be greater than 0"
+        end
 
-        raise LengthError, "field length must be greater than 0" unless length > 0
-        raise NameError, "column name cannot be empty" if @name.length == 0
+        if @name.empty?
+          raise NameError, "column name cannot be empty"
+        end
       end
 
       # Cast value to native type
@@ -31,7 +42,7 @@ module DBF
           when 'N' then unpack_number(value)
           when 'I' then unpack_unsigned_long(value)
           when 'F' then value.to_f
-          when 'Y' then unpack_unsigned_long(value) / 10000.0
+          when 'Y' then (unpack_unsigned_long(value) / 10000.0).to_f
           when 'D' then decode_date(value)
           when 'T' then decode_datetime(value)
           when 'L' then boolean(value)
@@ -40,6 +51,9 @@ module DBF
         end
       end
 
+      # Returns true if the column is a memo
+      #
+      # @return [Boolean]
       def memo?
         @memo ||= type == 'M'
       end
@@ -51,16 +65,20 @@ module DBF
         "\"#{underscored_name}\", #{schema_data_type}\n"
       end
 
-      def self.underscore_name(string)
-        string.gsub(/::/, '/').
-          gsub(/([A-Z]+)([A-Z][a-z])/,'\1_\2').
-          gsub(/([a-z\d])([A-Z])/,'\1_\2').
-          tr('-', '_').
-          downcase
-      end
-
+      # Underscored name
+      #
+      # This is the column name converted to underscore format.
+      # For example, MyColumn will be returned as my_column.
+      #
+      # @return [String]
       def underscored_name
-        @underscored_name ||= self.class.underscore_name(name)
+        @underscored_name ||= begin
+          name.gsub(/::/, '/').
+            gsub(/([A-Z]+)([A-Z][a-z])/,'\1_\2').
+            gsub(/([a-z\d])([A-Z])/,'\1_\2').
+            tr('-', '_').
+            downcase
+        end
       end
 
       private
@@ -73,9 +91,11 @@ module DBF
       end
 
       def decode_datetime(value) #nodoc
-        days, milliseconds = value.unpack('l2')
-        seconds = (milliseconds / 1000).to_i
-        DateTime.jd(days, (seconds/3600).to_i, (seconds/60).to_i % 60, seconds % 60) rescue nil
+        days, msecs = value.unpack('l2')
+        secs = (msecs / 1000).to_i
+        DateTime.jd(days, (secs/3600).to_i, (secs/60).to_i % 60, secs % 60)
+      rescue
+        nil
       end
 
       def decode_memo(value) #nodoc
@@ -95,10 +115,10 @@ module DBF
       end
 
       def encode_string(value) #nodoc
-        if @encoding
-          if String.new.respond_to?(:encoding)
-            value.force_encoding(@encoding).encode(Encoding.default_external, :undef => :replace, :invalid => :replace)
-          else
+        if @encoding && table.supports_encoding?
+          if table.supports_string_encoding?
+            value.force_encoding(@encoding).encode(*encoding_args)
+          elsif table.supports_iconv?
             Iconv.conv('UTF-8', @encoding, value)
           end
         else
@@ -106,6 +126,10 @@ module DBF
         end
       end
 
+      def encoding_args #nodoc
+        [Encoding.default_external, {:undef => :replace, :invalid => :replace}]
+      end
+
       def schema_data_type #nodoc
         case type
         when "N", "F"
@@ -134,9 +158,8 @@ module DBF
       end
 
       def clean(value) #nodoc
-        first_null = value.index("\x00")
-        value = value[0, first_null] if first_null
-        value.gsub(/[^\x20-\x7E]/, "")
+        truncated_value = value.strip.partition("\x00").first
+        truncated_value.gsub(/[^\x20-\x7E]/, '')
       end
 
     end

data/lib/dbf/column/foxpro.rb

@@ -1,9 +1,9 @@
 module DBF
   module Column
     class Foxpro < Base
-      def unpack_binary(value) #nodoc
-        value.unpack('d')[0]
-      end
+      # def unpack_binary(value) #nodoc
+      #   value.unpack('d')[0]
+      # end
     end
   end
-end
+end

data/lib/dbf/header.rb

@@ -10,8 +10,12 @@ module DBF
 
     def initialize(data, set_encoding)
       @data = data
-      @version, @record_count, @header_length, @record_length, @encoding_key = data.unpack("H2 x3 V v2 x17H2")
+      @version, @record_count, @header_length, @record_length, @encoding_key = unpack_header
       @encoding = DBF::ENCODINGS[@encoding_key] if set_encoding
     end
+
+    def unpack_header
+      data.unpack("H2 x3 V v2 x17H2")
+    end
   end
 end

data/lib/dbf/memo/base.rb

@@ -5,7 +5,7 @@ module DBF
       BLOCK_SIZE = 512
 
       def self.open(filename, version)
-        self.new File.open(filename, 'rb'), version
+        new(File.open(filename, 'rb'), version)
       end
 
       def initialize(data, version)
@@ -45,4 +45,4 @@ module DBF
       end
     end
   end
-end
+end

data/lib/dbf/record.rb

@@ -36,40 +36,49 @@ module DBF
     end
 
     # Reads attributes by column name
+    #
+    # @param [String, Symbol] key
     def [](key)
       key = key.to_s
       if attributes.has_key?(key)
         attributes[key]
-      elsif index = column_names.index(key)
+      elsif index = underscored_column_names.index(key)
         attributes[@columns[index].name]
       end
     end
 
+    # Record attributes
+    #
     # @return [Hash]
     def attributes
       @attributes ||= Hash[@columns.map {|column| [column.name, init_attribute(column)]}]
     end
 
+    # Overrides standard Object.respond_to? to return true if a
+    # matching column name is found.
+    #
+    # @param [String, Symbol] method
+    # @return [Boolean]
     def respond_to?(method, *args)
-      if column_names.include?(method.to_s)
+      if underscored_column_names.include?(method.to_s)
         true
       else
         super
       end
     end
 
-    def method_missing(method, *args)
-      if index = column_names.index(method.to_s)
+    private
+
+    def method_missing(method, *args) #nodoc
+      if index = underscored_column_names.index(method.to_s)
         attributes[@columns[index].name]
       else
         super
       end
     end
 
-    private
-
-    def column_names
-      @column_names ||= @columns.map {|column| column.underscored_name}
+    def underscored_column_names # nodoc
+      @underscored_column_names ||= @columns.map {|column| column.underscored_name}
     end
 
     def init_attribute(column) #nodoc

data/lib/dbf/schema.rb

@@ -0,0 +1,34 @@
+module DBF
+  module Schema
+    # Generate an ActiveRecord::Schema
+    #
+    # xBase data types are converted to generic types as follows:
+    # - Number columns with no decimals are converted to :integer
+    # - Number columns with decimals are converted to :float
+    # - Date columns are converted to :datetime
+    # - Logical columns are converted to :boolean
+    # - Memo columns are converted to :text
+    # - Character columns are converted to :string and the :limit option is set
+    #   to the length of the character column
+    #
+    # Example:
+    #   create_table "mydata" do |t|
+    #     t.column :name, :string, :limit => 30
+    #     t.column :last_update, :datetime
+    #     t.column :is_active, :boolean
+    #     t.column :age, :integer
+    #     t.column :notes, :text
+    #   end
+    #
+    # @return [String]
+    def schema
+      s = "ActiveRecord::Schema.define do\n"
+      s << "  create_table \"#{File.basename(@data.path, ".*")}\" do |t|\n"
+      columns.each do |column|
+        s << "    t.column #{column.schema_definition}"
+      end
+      s << "  end\nend"
+      s
+    end
+  end
+end

data/lib/dbf/table.rb

@@ -1,9 +1,12 @@
 module DBF
+  class FileNotFoundError < StandardError
+  end
 
   # DBF::Table is the primary interface to a single DBF file and provides
   # methods for enumerating and searching the records.
   class Table
     include Enumerable
+    include Schema
 
     DBF_HEADER_SIZE = 32
 
@@ -59,11 +62,15 @@ module DBF
     # @param [optional String, StringIO] memo Path to the memo file or a StringIO object
     # @param [optional String, Encoding] encoding Name of the encoding or an Encoding object
     def initialize(data, memo = nil, encoding = nil)
-      @data = open_data(data)
-      @data.rewind
-      @header = Header.new(@data.read(DBF_HEADER_SIZE), supports_encoding? || supports_iconv?)
-      @encoding = encoding || header.encoding
-      @memo = open_memo(data, memo)
+      begin
+        @data = open_data(data)
+        @data.rewind
+        @header = Header.new(@data.read(DBF_HEADER_SIZE), supports_encoding?)
+        @encoding = encoding || header.encoding
+        @memo = open_memo(data, memo)
+      rescue Errno::ENOENT => error
+        raise DBF::FileNotFoundError.new("file not found: #{data}")
+      end
     end
 
     # @return [TrueClass, FalseClass]
@@ -108,7 +115,7 @@ module DBF
     # @param [Fixnum] index
     # @return [DBF::Record, NilClass]
     def record(index)
-      seek(index * header.record_length)
+      seek_to_record(index)
       if !deleted_record?
         DBF::Record.new(@data.read(header.record_length), columns, version, @memo)
       end
@@ -137,37 +144,6 @@ module DBF
       VERSIONS[version]
     end
 
-    # Generate an ActiveRecord::Schema
-    #
-    # xBase data types are converted to generic types as follows:
-    # - Number columns with no decimals are converted to :integer
-    # - Number columns with decimals are converted to :float
-    # - Date columns are converted to :datetime
-    # - Logical columns are converted to :boolean
-    # - Memo columns are converted to :text
-    # - Character columns are converted to :string and the :limit option is set
-    #   to the length of the character column
-    #
-    # Example:
-    #   create_table "mydata" do |t|
-    #     t.column :name, :string, :limit => 30
-    #     t.column :last_update, :datetime
-    #     t.column :is_active, :boolean
-    #     t.column :age, :integer
-    #     t.column :notes, :text
-    #   end
-    #
-    # @return [String]
-    def schema
-      s = "ActiveRecord::Schema.define do\n"
-      s << "  create_table \"#{File.basename(@data.path, ".*")}\" do |t|\n"
-      columns.each do |column|
-        s << "    t.column #{column.schema_definition}"
-      end
-      s << "  end\nend"
-      s
-    end
-
     # Dumps all records to a CSV file.  If no filename is given then CSV is
     # output to STDOUT.
     #
@@ -215,38 +191,58 @@ module DBF
       end
     end
 
-    # Retrieves column information from the database
+    # All columns
+    #
+    # @return [Array]
     def columns
-      @columns ||= begin
-        @data.seek(DBF_HEADER_SIZE)
-        columns = []
-        while !["\0", "\r"].include?(first_byte = @data.read(1))
-          column_data = first_byte + @data.read(31)
-          name, type, length, decimal = column_data.unpack('a10 x a x4 C2')
-          if length > 0
-            columns << column_class.new(name.strip, type, length, decimal, version, encoding)
-          end
-        end
-        columns
-      end
+      @columns ||= build_columns
     end
 
+    # Column names
+    #
+    # @return [String]
+    def column_names
+      columns.map { |column| column.name }
+    end
+
+    # Is string encoding supported?
+    # String encoding is always supported in Ruby 1.9+.
+    # Ruby 1.8.x requires that Ruby be compiled with iconv support.
     def supports_encoding?
-      String.new.respond_to?(:encoding)
+      supports_string_encoding? || supports_iconv?
+    end
+
+    # Does String support encoding?  Should be true in Ruby 1.9+
+    def supports_string_encoding?
+      ''.respond_to?(:encoding)
     end
 
-    def supports_iconv?
+    def supports_iconv? #nodoc
       require 'iconv'
       true
     rescue
       false
     end
 
-    def foxpro?
-      FOXPRO_VERSIONS.keys.include? version
+    private
+
+    def build_columns #nodoc
+      columns = []
+      @data.seek(DBF_HEADER_SIZE)
+      while !["\0", "\r"].include?(first_byte = @data.read(1))
+        column_data = first_byte + @data.read(DBF_HEADER_SIZE - 1)
+        name, type, length, decimal = column_data.unpack('a10 x a x4 C2')
+        if length > 0
+          columns << column_class.new(self, name, type, length, decimal)
+        end
+      end
+      columns
     end
 
-    private
+
+    def foxpro? #nodoc
+      FOXPRO_VERSIONS.keys.include? version
+    end
 
     def column_class #nodoc
       @column_class ||= foxpro? ? Column::Foxpro : Column::Dbase
@@ -312,6 +308,10 @@ module DBF
       @data.seek header.header_length + offset
     end
 
+    def seek_to_record(index) #nodoc
+      seek(index * header.record_length)
+    end
+
     def csv_class #nodoc
       @csv_class ||= CSV.const_defined?(:Reader) ? FCSV : CSV
     end