dbf 4.3.2 → 5.0.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: c754ead4e347035e5997c61369dc602784ed7b1c6abda474d4f5678980d6280a
-  data.tar.gz: 5b8494f61b66720690b77cb11fc4bc2bfa778dd51e269b5091c9a77cb3c74c4e
+  metadata.gz: 4db2b075394f8f3cb60fc36e71a71eb9f8367e7b34b8863481fe6c2ab0a2aeac
+  data.tar.gz: f3e7ba5bd793d1732d2ae20b23b3a56285fa81228d06db9d38b055049591b76e
 SHA512:
-  metadata.gz: bffdced2430256755264936f67bbabc996add3b10b89f39e0c03b53203b859eabf760d5d6c0660679e741daee619f4dc69c4707c7cca6fea4c86719709b99461
-  data.tar.gz: 447cd7a12f995dcd2fcedf4e0c9c009b9d5b990e2ce2bd0eac65515a98247bb112fd230b5fd4023bf1478ee407dedcac42e50102ed1cbd73ade5608056aa0ea9
+  metadata.gz: b7c5847999bf422df92d90bbdea11bf0d6baa09ace5649cd65814b22cb5192a5ec085e8b25b9a2c56cb889f0f23d0fd0e3e20561c737ae05857afb63db69aea4
+  data.tar.gz: e1ed01f59b6f13e5bd3a91b327cdcd2060f85f2a608dfaf3e4dae1723f9a97b0ef2221465b0ae14aed1345e99b0c6c8edf9de00259f8a741914d889212ca5dc6

data/CHANGELOG.md

@@ -1,5 +1,15 @@
 # Changelog
 
+## 5.0.1
+
+- Raise ArgumentError data is not a string or StringIO object
+
+## 5.0.0
+
+- Refactor the Column class to support non-ASCII header names
+- Output encoding is now set to UTF-8 if there is no embedded encoding and one is not
+  specified during DBF::Table initialization.
+
 ## 4.3.2
 
 - Fixes to maintain support for Ruby 3.0.x until it's EOL

data/bin/dbf

@@ -41,6 +41,7 @@ else
     puts
     puts "Database: #{filename}"
     puts "Type: (#{table.version}) #{table.version_description}"
+    puts "Encoding: #{table.header_encoding}" if table.header_encoding
     puts "Memo File: #{table.has_memo_file? ? 'true' : 'false'}"
     puts "Records: #{table.record_count}"
 

data/lib/dbf/column.rb

@@ -8,7 +8,7 @@ module DBF
     class NameError < StandardError
     end
 
-    attr_reader :table, :name, :type, :length, :decimal
+    attr_reader :table, :name, :type, :length, :decimal, :encoding
 
     def_delegator :type_cast_class, :type_cast
 
@@ -38,13 +38,14 @@ module DBF
     # @param length [Integer]
     # @param decimal [Integer]
     def initialize(table, name, type, length, decimal)
+      @encoding = table.encoding
+
       @table = table
       @name = clean(name)
       @type = type
       @length = length
       @decimal = decimal
       @version = table.version
-      @encoding = table.encoding
 
       validate_length
       validate_name
@@ -72,37 +73,18 @@ module DBF
     # @return [String]
     def underscored_name
       @underscored_name ||= name.gsub(/([a-z\d])([A-Z])/, '\1_\2').tr('-', '_').downcase
-      
     end
 
     private
 
     def clean(value) # :nodoc:
-      value.strip.partition("\x00").first.gsub(/[^\x20-\x7E]/, '')
-    end
-
-    def encode(value, strip_output: false) # :nodoc:
-      return value unless value.respond_to?(:encoding)
-
-      output = @encoding ? encode_string(value) : value
-      strip_output ? output.strip : output
-    end
-
-    def encoding_args # :nodoc:
-      @encoding_args ||= [
-        Encoding.default_external,
-        {undef: :replace, invalid: :replace}
-      ]
-    end
-
-    def encode_string(string) # :nodoc:
-      string.force_encoding(@encoding).encode(*encoding_args)
+      table.encode_string(value.strip.partition("\x00").first)
     end
 
     def type_cast_class # :nodoc:
       @type_cast_class ||= begin
         klass = @length == 0 ? ColumnType::Nil : TYPE_CAST_CLASS[type.to_sym]
-        klass.new(@decimal, @encoding)
+        klass.new(self)
       end
     end
 

data/lib/dbf/column_type.rb

@@ -5,9 +5,9 @@ module DBF
 
       # @param decimal [Integer]
       # @param encoding [String, Encoding]
-      def initialize(decimal, encoding)
-        @decimal = decimal
-        @encoding = encoding
+      def initialize(column)
+        @decimal = column.decimal
+        @encoding = column.encoding
       end
     end
 

data/lib/dbf/table.rb

@@ -76,7 +76,7 @@ module DBF
     # @param encoding [optional String, Encoding] encoding Name of the encoding or an Encoding object
     def initialize(data, memo = nil, encoding = nil)
       @data = open_data(data)
-      @encoding = encoding || header.encoding
+      @encoding = encoding || header.encoding || Encoding.default_external
       @memo = open_memo(data, memo)
       yield self if block_given?
     end
@@ -211,6 +211,21 @@ module DBF
       VERSIONS[version]
     end
 
+    # Encode string
+    #
+    # @param [String] string
+    # @return [String]
+    def encode_string(string) # :nodoc:
+      string.force_encoding(@encoding).encode(Encoding.default_external, undef: :replace, invalid: :replace)
+    end
+
+    # Encoding specified in the file header
+    #
+    # @return [Encoding]
+    def header_encoding
+      header.encoding
+    end
+
     private
 
     def build_columns # :nodoc:
@@ -293,7 +308,14 @@ module DBF
     end
 
     def open_data(data) # :nodoc:
-      data.is_a?(StringIO) ? data : File.open(data, 'rb')
+      case data
+      when StringIO
+        data
+      when String
+        File.open(data, 'rb')
+      else
+        raise ArgumentError, 'data must be a file path or StringIO object'
+      end
     rescue Errno::ENOENT
       raise DBF::FileNotFoundError, "file not found: #{data}"
     end

data/lib/dbf/version.rb

@@ -1,3 +1,3 @@
 module DBF
-  VERSION = '4.3.2'.freeze
+  VERSION = '5.0.1'.freeze
 end

data/spec/dbf/column_spec.rb

@@ -37,7 +37,7 @@ RSpec.describe DBF::Column do
 
     describe 'with empty column name' do
       it 'raises DBF::Column::NameError' do
-        expect { DBF::Column.new table, "\xFF\xFC", 'N', 1, 0 }.to raise_error(DBF::Column::NameError)
+        expect { DBF::Column.new table, '', 'N', 1, 0 }.to raise_error(DBF::Column::NameError)
       end
     end
   end
@@ -283,16 +283,4 @@ RSpec.describe DBF::Column do
       end
     end
   end
-
-  describe '#name' do
-    it 'contains only ASCII characters' do
-      column = DBF::Column.new table, "--\x1F-\x68\x65\x6C\x6C\x6F \x00world-\x80--", 'N', 1, 0
-      expect(column.name).to eq '---hello '
-    end
-
-    it 'is truncated at the null character' do
-      column = DBF::Column.new table, "--\x1F-\x68\x65\x6C\x6C\x6F \x00world-\x80\x80--", 'N', 1, 0
-      expect(column.name).to eq '---hello '
-    end
-  end
 end

data/spec/dbf/encoding_spec.rb

@@ -0,0 +1,47 @@
+# encoding: utf-8
+
+require 'spec_helper'
+
+RSpec.describe 'default encoding' do
+  let(:dbf_path) { fixture('dbase_03_cyrillic.dbf') }
+  let(:table) { DBF::Table.new dbf_path }
+
+  it 'defaults to UTF-8 encoding' do
+    expect(table.encoding).to eq Encoding::UTF_8
+  end
+
+  it 'uses the table encoding for column encoding' do
+    column = table.columns.first
+    expect(column.encoding).to eq table.encoding
+  end
+
+  it 'encodes column names' do
+    expect(table.column_names).to eq ['ШАР', 'ПЛОЩА']
+  end
+
+  it 'encodes record values' do
+    expect(table.record(0).attributes['ШАР']).to eq 'Номер'
+  end
+end
+
+RSpec.describe 'embedded encoding' do
+  let(:dbf_path) { fixture('cp1251.dbf') }
+  let(:table) { DBF::Table.new dbf_path }
+
+  it 'defaults to UTF-8 encoding' do
+    expect(table.encoding).to eq 'cp1251'
+  end
+
+  it 'uses the table encoding for column encoding' do
+    column = table.columns.first
+    expect(column.encoding).to eq table.encoding
+  end
+
+  it 'encodes column names' do
+    expect(table.column_names).to eq ['RN', 'NAME']
+  end
+
+  it 'encodes record values' do
+    expect(table.record(0).attributes['NAME']).to eq 'амбулаторно-поликлиническое'
+  end
+end

data/spec/dbf/table_spec.rb

@@ -29,6 +29,12 @@ RSpec.describe DBF::Table do
       end
     end
 
+    describe 'when data is nil' do
+      it 'raises ArgumentError' do
+        expect { DBF::Table.new nil }.to raise_error(ArgumentError, 'data must be a file path or StringIO object')
+      end
+    end
+
     describe 'when given paths to existing dbf and memo files' do
       it 'does not raise an error' do
         expect { DBF::Table.new dbf_path, memo_path }.to_not raise_error

data/spec/fixtures/cp1251_summary.txt

@@ -1,6 +1,7 @@
 
 Database: cp1251.dbf
 Type: (30) Visual FoxPro
+Encoding: cp1251
 Memo File: false
 Records: 4
 

data/spec/fixtures/dbase_03_cyrillic_summary.txt

@@ -0,0 +1,11 @@
+
+Database: dbase_03_cyrillic.dbf
+Type: (03) dBase III without memo file
+Memo File: false
+Records: 2
+
+Fields:
+Name             Type       Length     Decimal
+------------------------------------------------------------------------------
+ШАР              C          25         0         
+ПЛОЩА            N          15         2         

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: dbf
 version: !ruby/object:Gem::Version
-  version: 4.3.2
+  version: 5.0.1
 platform: ruby
 authors:
 - Keith Morrison
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2024-01-23 00:00:00.000000000 Z
+date: 2024-04-25 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: csv
@@ -56,6 +56,7 @@ files:
 - lib/dbf/version.rb
 - spec/dbf/column_spec.rb
 - spec/dbf/database/foxpro_spec.rb
+- spec/dbf/encoding_spec.rb
 - spec/dbf/file_formats_spec.rb
 - spec/dbf/record_spec.rb
 - spec/dbf/table_spec.rb
@@ -64,6 +65,8 @@ files:
 - spec/fixtures/dbase_02.dbf
 - spec/fixtures/dbase_02_summary.txt
 - spec/fixtures/dbase_03.dbf
+- spec/fixtures/dbase_03_cyrillic.dbf
+- spec/fixtures/dbase_03_cyrillic_summary.txt
 - spec/fixtures/dbase_03_summary.txt
 - spec/fixtures/dbase_30.dbf
 - spec/fixtures/dbase_30.fpt
@@ -125,7 +128,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: 1.3.0
 requirements: []
-rubygems_version: 3.4.12
+rubygems_version: 3.5.8
 signing_key:
 specification_version: 4
 summary: Read xBase files