embulk 0.7.2 → 0.7.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 69167ac2d89cb01cce7fe04d3380290604141720
-  data.tar.gz: cfcc577be3131bd9792d420425fd10dedf2df974
+  metadata.gz: e54fd2e5f50348148e43d19d4bd832c1fc23b2bc
+  data.tar.gz: e9ee753c60761e9eb9304e2cc107808409e2e94a
 SHA512:
-  metadata.gz: ebe4adc87fd92f411e19db78e3ea58569c7902bf8e7f9dee97b797d5d7b3f502c056804d8e4d8a9989b6e61dc8e36476588fbf279d8232d48a4ea7842fa3411b
-  data.tar.gz: b2721835cd90a56cd9c97644a1af1429225542a8c0f1845014ce833e4f548dd48fed4511329f3f196da00d85783f8bb8bab73cf1e1f5effb6ff680fe5a13cacf
+  metadata.gz: 6ff353d37efad5f385422cc78af6fa6ea7530a0734ab33fbde2883cbde9363a9b45aa00c801087aa9631d805c3e62e5525363c7c8b094207377005fc00660aca
+  data.tar.gz: 43a9a324e3682d2f116843a7ace68c6b5fab138a9d06d9bbdf7e8c831f74c32069538de4f141826b58ed99a5cc8c26e5b09f818aac984b8c641ae27c1301ba6b

data/build.gradle

@@ -16,7 +16,7 @@ def release_projects = [project(":embulk-core"), project(":embulk-standards")]
 
 allprojects {
     group = 'org.embulk'
-    version = '0.7.2'
+    version = '0.7.3'
 
     ext {
         jrubyVersion = '9.0.0.0'

data/embulk-docs/src/release.rst

@@ -4,6 +4,7 @@ Release Notes
 .. toctree::
     :maxdepth: 1
 
+    release/release-0.7.3
     release/release-0.7.2
     release/release-0.7.1
     release/release-0.7.0

data/embulk-docs/src/release/release-0.7.3.rst

@@ -0,0 +1,21 @@
+Release 0.7.3
+==================================
+
+General Changes
+------------------
+
+* ``embulk selfupdate`` command gets optoinal ``<version>`` argument to upgrade or downgrade to specific version (@cosmo0920++).
+* ``embulk migrate`` command assumes that source code file encoding is UTF-8. This fixes encoding exception on Windows environment.
+
+
+Java Plugin API
+------------------
+
+* Fixed ``Embulk::Exec.preview?``.
+* Fixed ``raise ConfigException``.
+* added missing ``PluginLoadError`` class.
+
+
+Release Date
+------------------
+2015-08-23

data/lib/embulk/buffer.rb

@@ -1,22 +1,20 @@
 
 module Embulk
   class Buffer < String
-    if Embulk.java?
-      def self.from_java(java_buffer)
-        byte_list = org.jruby.util.ByteList.new(java_buffer.array(), java_buffer.offset(), java_buffer.limit(), false)
-        buffer = new
-        buffer.replace(org.jruby.RubyString.new(JRuby.runtime, self, byte_list).dup)  # TODO simplify
-        buffer
-      end
+    def self.from_java(java_buffer)
+      byte_list = org.jruby.util.ByteList.new(java_buffer.array(), java_buffer.offset(), java_buffer.limit(), false)
+      buffer = new
+      buffer.replace(org.jruby.RubyString.new(JRuby.runtime, self, byte_list).dup)  # TODO simplify
+      buffer
+    end
 
-      def self.from_ruby_string(string)
-        b = Buffer.new(string)
-        b.force_encoding('ASCII-8BIT')
-      end
+    def self.from_ruby_string(string)
+      b = Buffer.new(string)
+      b.force_encoding('ASCII-8BIT')
+    end
 
-      def to_java
-        Java::Buffer.wrap(to_java_bytes)
-      end
+    def to_java
+      Java::Buffer.wrap(to_java_bytes)
     end
   end
 end

data/lib/embulk/column.rb

@@ -20,49 +20,45 @@ module Embulk
       end
     end
 
-    if Embulk.java?
-      def self.from_java(java_column)
-        type = Type.from_java(java_column.getType)
-        if type == :timestamp
-          format = java_column.getType.getFormat
-        else
-          format = nil
-        end
-
-        Column.new(java_column.getIndex, java_column.getName, type, format)
+    def self.from_java(java_column)
+      type = Type.from_java(java_column.getType)
+      if type == :timestamp
+        format = java_column.getType.getFormat
+      else
+        format = nil
       end
 
-      def to_java
-        if type == :timestamp && format
-          Java::Column.new(index, name, Type.new_java_type(type).withFormat(format))
-        else
-          Java::Column.new(index, name, Type.new_java_type(type))
-        end
+      Column.new(java_column.getIndex, java_column.getName, type, format)
+    end
+
+    def to_java
+      if type == :timestamp && format
+        Java::Column.new(index, name, Type.new_java_type(type).withFormat(format))
+      else
+        Java::Column.new(index, name, Type.new_java_type(type))
       end
     end
   end
 
   module Type
-    if Embulk.java?
-      def self.from_java(java_type)
-        java_type.getName.to_sym
-      end
+    def self.from_java(java_type)
+      java_type.getName.to_sym
+    end
 
-      def self.new_java_type(ruby_type)
-        case ruby_type
-        when :boolean
-          Java::Types::BOOLEAN
-        when :long
-          Java::Types::LONG
-        when :double
-          Java::Types::DOUBLE
-        when :string
-          Java::Types::STRING
-        when :timestamp
-          Java::Types::TIMESTAMP
-        else
-          raise ArgumentError, "Unknown type #{ruby_type.inspect}: supported types are :boolean, :long, :double, :string and :timestamp"
-        end
+    def self.new_java_type(ruby_type)
+      case ruby_type
+      when :boolean
+        Java::Types::BOOLEAN
+      when :long
+        Java::Types::LONG
+      when :double
+        Java::Types::DOUBLE
+      when :string
+        Java::Types::STRING
+      when :timestamp
+        Java::Types::TIMESTAMP
+      else
+        raise ArgumentError, "Unknown type #{ruby_type.inspect}: supported types are :boolean, :long, :double, :string and :timestamp"
       end
     end
   end

data/lib/embulk/command/embulk_migrate_plugin.rb

@@ -96,7 +96,7 @@ module Embulk
 
     def match(glob, pattern)
       ms = Dir[File.join(@path, glob)].map do |file|
-        File.read(file).match(pattern)
+        read(file).match(pattern)
       end.compact
       return nil if ms.empty?
       ms
@@ -104,7 +104,7 @@ module Embulk
 
     def replace(glob, pattern, text=nil)
       ms = Dir[File.join(@path, glob)].map do |file|
-        data = File.read(file)
+        data = read(file)
         first = nil
         pos = 0
         while pos < data.length
@@ -126,7 +126,7 @@ module Embulk
 
     def insert_line(glob, pattern, text=nil)
       ms = Dir[File.join(@path, glob)].map do |file|
-        data = File.read(file)
+        data = read(file)
         if m = data.match(pattern)
           ln = m.pre_match.split("\n").count
           replace = text || yield(m)
@@ -142,7 +142,7 @@ module Embulk
     end
 
     def data(file)
-      File.read(File.join(@path, file))
+      read(File.join(@path, file))
     end
 
     def write(file, data)
@@ -152,7 +152,7 @@ module Embulk
     private
 
     def modify(path, data)
-      orig = File.read(path) rescue nil
+      orig = read(path) rescue nil
       if orig != data
         File.write(path, data)
         unless @modified_files.has_key?(path)
@@ -166,5 +166,11 @@ module Embulk
       end
       nil
     end
+
+    def read(path)
+      # external_encoding: assumes source code is written in UTF-8.
+      # internal_encoding: process files using UTF-8 so that modified data (data written to the file) becomes UTF-8.
+      File.read(path, external_encoding: 'UTF-8', internal_encoding: 'UTF-8')
+    end
   end
 end

data/lib/embulk/command/embulk_run.rb

@@ -175,7 +175,7 @@ examples:
       op.on('-f', "Skip corruption check", TrueClass) do |b|
         options[:force] = true
       end
-      args = 0..0
+      args = 0..1
 
     when :example
       args = 0..1
@@ -253,6 +253,7 @@ examples:
 
     when :selfupdate
       require 'embulk/command/embulk_selfupdate'
+      options[:version] = ARGV[0]
       Embulk.selfupdate(options)
 
     else
@@ -424,16 +425,16 @@ examples:
     STDERR.puts "Embulk v#{Embulk::VERSION}"
     STDERR.puts "usage: <command> [--options]"
     STDERR.puts "commands:"
-    STDERR.puts "   bundle    [directory]                              # create or update plugin environment."
-    STDERR.puts "   run       <config.yml>                             # run a bulk load transaction."
-    STDERR.puts "   preview   <config.yml>                             # dry-run the bulk load without output and show preview."
-    STDERR.puts "   guess     <partial-config.yml> -o <output.yml>     # guess missing parameters to create a complete configuration file."
-    STDERR.puts "   gem       <install | list | help>                  # install a plugin or show installed plugins."
+    STDERR.puts "   bundle     [directory]                             # create or update plugin environment."
+    STDERR.puts "   run        <config.yml>                            # run a bulk load transaction."
+    STDERR.puts "   preview    <config.yml>                            # dry-run the bulk load without output and show preview."
+    STDERR.puts "   guess      <partial-config.yml> -o <output.yml>    # guess missing parameters to create a complete configuration file."
+    STDERR.puts "   gem        <install | list | help>                 # install a plugin or show installed plugins."
     STDERR.puts "                                                      # plugin path is #{ENV['GEM_HOME']}"
-    STDERR.puts "   new       <category> <name>                        # generates new plugin template"
-    STDERR.puts "   migrate   <path>                                   # modify plugin code to use the latest Embulk plugin API"
-    STDERR.puts "   example   [path]                                   # creates an example config file and csv file to try embulk."
-    STDERR.puts "   selfupdate                                         # upgrades embulk to the latest released version."
+    STDERR.puts "   new        <category> <name>                       # generates new plugin template"
+    STDERR.puts "   migrate    <path>                                  # modify plugin code to use the latest Embulk plugin API"
+    STDERR.puts "   example    [path]                                  # creates an example config file and csv file to try embulk."
+    STDERR.puts "   selfupdate [version]                               # upgrades embulk to the latest released version or to the specified version."
     STDERR.puts ""
     if message
       system_exit "error: #{message}"

data/lib/embulk/command/embulk_selfupdate.rb

@@ -14,15 +14,27 @@ module Embulk
       raise SystemExit.new(1)
     end
 
-    puts "Checking the latest version..."
-    latest_version = check_latest_version
+    if version = options[:version]
+      puts "Checking version #{version}..."
+      target_version = check_target_version(version)
 
-    if Gem::Version.new(latest_version) <= Gem::Version.new(Embulk::VERSION)
-      puts "Already up-to-date. #{latest_version} is the latest version."
-      return
-    end
+      unless target_version
+        puts "Specified version does not exist: #{version}"
+        raise SystemExit.new(1)
+      end
+      puts "Found version #{target_version}."
+
+    else
+      puts "Checking the latest version..."
+      target_version = check_latest_version
 
-    puts "Found new version #{latest_version}."
+      if Gem::Version.new(target_version) <= Gem::Version.new(Embulk::VERSION)
+        puts "Already up-to-date. #{target_version} is the latest version."
+        return
+      end
+
+      puts "Found new version #{target_version}."
+    end
 
     unless File.writable?(jar_path)
       STDERR.puts ""
@@ -32,7 +44,7 @@ module Embulk
       raise SystemExit.new(1)
     end
 
-    url = "https://dl.bintray.com/embulk/maven/embulk-#{latest_version}.jar"
+    url = "https://dl.bintray.com/embulk/maven/embulk-#{target_version}.jar"
     puts "Downloading #{url} ..."
 
     require 'open-uri'
@@ -50,8 +62,8 @@ module Embulk
           data = File.read("jar:#{java.io.File.new(tmp.path).toURI.toURL}!/embulk/version.rb")
           m = Module.new
           m.module_eval(data)
-          unless m::Embulk::VERSION == latest_version
-            raise "Embulk::VERSION does not match with #{latest_version}"
+          unless m::Embulk::VERSION == target_version
+            raise "Embulk::VERSION does not match with #{target_version}"
           end
         rescue => e
           STDERR.puts "Corruption checking failed (#{e})."
@@ -64,21 +76,44 @@ module Embulk
       File.rename(tmp.path, jar_path)
     end
 
-    puts "Updated to #{latest_version}."
+    puts "Updated to #{target_version}."
   end
 
+  private
+
   def self.check_latest_version
-    require 'net/https'
-    bintray = Net::HTTP.new('bintray.com', 443)
-    bintray.use_ssl = true
-    bintray.verify_mode = OpenSSL::SSL::VERIFY_NONE
-    bintray.start do
-      response = bintray.get('/embulk/maven/embulk/_latestVersion')
-      raise "Expected response code 302 Found but got #{response.code}" if response.code != "302"
+    start_bintray_http do |http|
+      response = http.get('/embulk/maven/embulk/_latestVersion')
+      if response.code != "302"
+        raise "Expected response code 302 Found but got #{response.code}"
+      end
       location = response["Location"].to_s
       m = /(\d+\.\d+[^\/]+)/.match(location)
-      raise "Cound not find version number in Location header '#{location}'" unless m
+      unless m
+        raise "Cound not find version number in Location header '#{location}'"
+      end
       return m[1]
     end
   end
+
+  def self.check_target_version(version=nil)
+    start_bintray_http do |http|
+      response = http.get("/embulk/maven/embulk/#{version}")
+      if response.code == "404"
+        return nil
+      elsif response.code != "200"
+        raise "Unexpected response code: #{response.code}"
+      else
+        return version
+      end
+    end
+  end
+
+  def self.start_bintray_http(&block)
+    require 'net/https'
+    bintray = Net::HTTP.new('bintray.com', 443)
+    bintray.use_ssl = true
+    bintray.verify_mode = OpenSSL::SSL::VERIFY_NONE
+    bintray.start(&block)
+  end
 end

data/lib/embulk/data_source.rb

@@ -145,31 +145,31 @@ module Embulk
             begin
               Integer(v)
             rescue => e
-              raise ConfigError, e
+              raise ConfigError.new e
             end
           when :float
             begin
               Float(v)
             rescue => e
-              raise ConfigError, e
+              raise ConfigError.new e
             end
           when :string
             begin
               String(v).dup
             rescue => e
-              raise ConfigError, e
+              raise ConfigError.new e
             end
           when :bool
             begin
               !!v  # TODO validation
             rescue => e
-              raise ConfigError, e
+              raise ConfigError.new e
             end
           when :hash
-            raise ConfigError, "Invalid value for :hash" unless v.is_a?(Hash)
+            raise ConfigError.new "Invalid value for :hash" unless v.is_a?(Hash)
             DataSource.new.merge!(v)
           when :array
-            raise ConfigError, "Invalid value for :array" unless v.is_a?(Array)
+            raise ConfigError.new "Invalid value for :array" unless v.is_a?(Array)
             v.dup
           else
             unless type.respond_to?(:load)
@@ -178,7 +178,7 @@ module Embulk
             begin
               type.load(v)
             rescue => e
-              raise ConfigError, e
+              raise ConfigError.new e
             end
           end
 
@@ -186,34 +186,32 @@ module Embulk
         value = options[:default]
 
       else
-        raise ConfigError, "Required field #{key.to_s.dump} is not set"
+        raise ConfigError.new "Required field #{key.to_s.dump} is not set"
       end
 
       return value
     end
 
-    if Embulk.java?
-      def self.from_java(java_data_source_impl)
-        json = java_data_source_impl.toString
-        new.merge!(JSON.parse(json))
-      end
+    def self.from_java(java_data_source_impl)
+      json = java_data_source_impl.toString
+      new.merge!(JSON.parse(json))
+    end
 
-      def self.from_ruby_hash(hash)
-        new.merge!(hash)
-      end
+    def self.from_ruby_hash(hash)
+      new.merge!(hash)
+    end
 
-      def to_java
-        json = to_json
-        Java::Injected::ModelManager.readObject(Java::DataSourceImpl.java_class, json.to_java)
-      end
+    def to_java
+      json = to_json
+      Java::Injected::ModelManager.readObject(Java::DataSourceImpl.java_class, json.to_java)
+    end
 
-      def load_config(task_type)
-        Java::Injected::ModelManager.readObjectWithConfigSerDe(task_type.java_class, to_json.to_java)
-      end
+    def load_config(task_type)
+      Java::Injected::ModelManager.readObjectWithConfigSerDe(task_type.java_class, to_json.to_java)
+    end
 
-      def load_task(task_type)
-        Java::Injected::ModelManager.readObject(task_type.java_class, to_json.to_java)
-      end
+    def load_task(task_type)
+      Java::Injected::ModelManager.readObject(task_type.java_class, to_json.to_java)
     end
   end