chupa-text 1.0.0 → 1.0.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: b70738567326cdcde32a77574ef71c9981bb0ca2
-  data.tar.gz: a7d5c2ffa050864b2d7448ded88dbda543695909
+  metadata.gz: c04261f7c3b4c26a0d67f18f473c8adf1d435080
+  data.tar.gz: 4e1a8da97abfa817226f919fcb672fb55d570d4d
 SHA512:
-  metadata.gz: f64768ada9e980ee3ce8e052258ecce419c4474bb25a2b8ebee170c9e27510536412731c3587ca1eb4f5a027f3262104e636556074fcc95cf0f784550dca3c0e
-  data.tar.gz: e3d862b6b5aa0ff30a49276327e1282fd2891a236b0822560f6c8f8aefbbfdf256ef27ad9ef7ba786e9c0120965f65c2f9d5a7a469cf98a17c7b5122bf8e1f50
+  metadata.gz: 00ada22d44d1f41ca4e2f08f02c43b41b193453ee33477dd02e9f5a9fbc0d87f5de46c53bc36a77b1494c2f8729a01e72419f9c9c9738d8819c04dde06c50d44
+  data.tar.gz: 7d404268095b308d4b7158e70458ac31d2dade5a0a131431559e84b34b563fb94454d5a3e900684dfe84da2642b382a14c55a479192992b141136a5f8659faf9

data/data/mime-types.conf

@@ -1,19 +1,21 @@
 # -*- ruby -*-
 
-mime_type["txt"] = "text/plain"
+mime_types["txt"] = "text/plain"
 
-mime_type["gz"]  = "application/x-gzip"
-mime_type["tgz"] = "application/x-gtar-compressed"
+mime_types["gz"]  = "application/x-gzip"
+mime_types["tgz"] = "application/x-gtar-compressed"
 
-mime_type["tar"] = "application/x-tar"
+mime_types["tar"] = "application/x-tar"
 
-mime_type["htm"]   = "text/html"
-mime_type["html"]  = "text/html"
-mime_type["xhtml"] = "application/xhtml+xml"
+mime_types["htm"]   = "text/html"
+mime_types["html"]  = "text/html"
+mime_types["xhtml"] = "application/xhtml+xml"
 
-mime_type["xml"] = "text/xml"
+mime_types["xml"] = "text/xml"
 
-mime_type["css"] = "text/css"
+mime_types["css"] = "text/css"
 
-mime_type["csv"] = "text/csv"
-mime_type["tsv"] = "text/tab-separated-values"
+mime_types["csv"] = "text/csv"
+mime_types["tsv"] = "text/tab-separated-values"
+
+mime_types["pdf"] = "application/pdf"

data/doc/text/command-line.md

@@ -123,14 +123,14 @@ command. You can use glob pattern for decomposer name such as
 The default is `["*"]`. It means that all installed decomposers are
 used.
 
-`mime_type["<extension>"] = "<MIME type>"`
+`mime_types["<extension>"] = "<MIME type>"`
 
 It specifies a map to a MIME type from path extension.
 
 Here is an example that maps `"html"` to `"text/html"`:
 
 ```
-mime_type["html"] = "text/html"
+mime_types["html"] = "text/html"
 ```
 
 Th default configuration file registers popular MIME types.

data/doc/text/news.md

@@ -1,5 +1,17 @@
 # News
 
+## 1.0.1: 2014-01-05
+
+  * chupa-text: Supported loading decomposers installed by RubyGems.
+  * chupa-text: Added `--disable-gems` option that disable loading
+    decomposers installed by RubyGems.
+  * chupa-text: Added `-I` option to use decomposers that are not
+    installed by RubyGems.
+  * Added {ChupaText::Data#text_plain?}.
+  * configuration: Changed `mime_types` from `mime_type` because they
+    processes about a set of MIME types.
+  * configuration: Added PDF to the default MIME type mappings.
+
 ## 1.0.0: 2014-01-05
 
 The first release!!!

data/lib/chupa-text/command/chupa-text.rb

@@ -29,12 +29,13 @@ module ChupaText
       def initialize
         @input = nil
         @configuration = Configuration.default
+        @enable_gems = true
       end
 
       def run(*arguments)
         return false unless parse_arguments(arguments)
 
-        Decomposers.load
+        load_decomposers
         extractor = create_extractor
         data = create_data
         formatter = create_formatter
@@ -77,9 +78,22 @@ module ChupaText
                   "Read configuration from FILE.") do |path|
           load_configuration(path)
         end
+        parser.on("--disable-gems",
+                  "Disable decomposers installed by RubyGems.") do
+          @enable_gems = false
+        end
+        parser.on("-I=PATH",
+                  "Append PATH to decomposer load path.") do |path|
+          $LOAD_PATH << path
+        end
         parser
       end
 
+      def load_decomposers
+        Decomposers.enable_all_gems if @enable_gems
+        Decomposers.load
+      end
+
       def create_extractor
         extractor = Extractor.new
         extractor.apply_configuration(@configuration)

data/lib/chupa-text/configuration-loader.rb

@@ -19,11 +19,11 @@ require "pathname"
 module ChupaText
   class ConfigurationLoader
     attr_reader :decomposer
-    attr_reader :mime_type
+    attr_reader :mime_types
     def initialize(configuration)
       @configuration = configuration
       @decomposer = DecomposerLoader.new(@configuration.decomposer)
-      @mime_type = MIMETypeLoader.new(@configuration.mime_type_registry)
+      @mime_types = MIMETypesLoader.new(@configuration.mime_type_registry)
       @load_paths = []
       data_dir = File.join(File.dirname(__FILE__), "..", "..", "data")
       @load_paths << File.expand_path(data_dir)
@@ -82,7 +82,7 @@ module ChupaText
       end
     end
 
-    class MIMETypeLoader
+    class MIMETypesLoader
       def initialize(registry)
         @registry = registry
       end

data/lib/chupa-text/data.rb

@@ -117,6 +117,12 @@ module ChupaText
       (mime_type || "").start_with?("text/")
     end
 
+    # @return [Bool] true if MIME type is "text/plain", false
+    #   otherwise.
+    def text_plain?
+      mime_type == "text/plain"
+    end
+
     private
     def guess_mime_type
       guess_mime_type_from_uri or

data/lib/chupa-text/decomposers.rb

@@ -17,6 +17,20 @@
 module ChupaText
   module Decomposers
     class << self
+      def enable_all_gems
+        decomposer_specs = Gem::Specification.find_all do |spec|
+          spec.name.start_with?("chupa-text-decomposer-")
+        end
+        grouped_decomposer_specs = decomposer_specs.group_by(&:name)
+        latest_decomposer_specs = []
+        grouped_decomposer_specs.each do |name, specs|
+          latest_decomposer_specs << specs.sort_by(&:version).last
+        end
+        latest_decomposer_specs.each do |spec|
+          gem(spec.name, spec.version)
+        end
+      end
+
       def load
         paths = []
         $LOAD_PATH.each do |load_path|

data/lib/chupa-text/extractor.rb

@@ -59,6 +59,10 @@ module ChupaText
       targets = [ensure_data(input)]
       until targets.empty?
         target = targets.pop
+        if target.text_plain?
+          yield(target)
+          next
+        end
         decomposer = find_decomposer(target)
         if decomposer.nil?
           yield(target) if target.text?
@@ -72,13 +76,10 @@ module ChupaText
 
     private
     def ensure_data(input)
-      case input
-      when String, Pathname, URI::Generic
-        data = Data.new
-        data.uri = input.to_s
-        data
-      else
+      if input.is_a?(Data)
         input
+      else
+        InputData.new(input)
       end
     end
 

data/lib/chupa-text/version.rb

@@ -15,5 +15,5 @@
 # Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
 
 module ChupaText
-  VERSION = "1.0.0"
+  VERSION = "1.0.1"
 end

data/test/command/test-chupa-text.rb

@@ -44,7 +44,7 @@ class TestCommandChupaText < Test::Unit::TestCase
 
   def run_command(*arguments)
     succeeded = wrap_io do
-      ChupaText::Command::ChupaText.run(*arguments)
+      ChupaText::Command::ChupaText.run("--disable-gems", *arguments)
     end
     [succeeded, JSON.parse(@stdout.string)]
   end

data/test/test-data.rb

@@ -56,6 +56,40 @@ class TestData < Test::Unit::TestCase
         end
       end
     end
+
+    sub_test_case("text?") do
+      def test_text_plain
+        @data.mime_type = "text/plain"
+        assert_true(@data.text?)
+      end
+
+      def test_text_html
+        @data.mime_type = "text/html"
+        assert_true(@data.text?)
+      end
+
+      def test_application_xhtml_xml
+        @data.mime_type = "application/xhtml+xml"
+        assert_false(@data.text?)
+      end
+    end
+
+    sub_test_case("text_plain?") do
+      def test_text_plain
+        @data.mime_type = "text/plain"
+        assert_true(@data.text_plain?)
+      end
+
+      def test_text_html
+        @data.mime_type = "text/html"
+        assert_false(@data.text_plain?)
+      end
+
+      def test_application_xhtml_xml
+        @data.mime_type = "application/xhtml+xml"
+        assert_false(@data.text_plain?)
+      end
+    end
   end
 
   sub_test_case("extension") do

data/test/test-extractor.rb

@@ -97,12 +97,12 @@ class TestExtractor < Test::Unit::TestCase
     sub_test_case("multi decomposed") do
       class CopyDecomposer < ChupaText::Decomposer
         def target?(data)
-          data["copied"].nil?
+          data.mime_type == "text/x-plain"
         end
 
         def decompose(data)
           copied_data = data.dup
-          copied_data["copied"] = true
+          copied_data.mime_type = "text/plain"
           yield(copied_data.dup)
           yield(copied_data.dup)
         end
@@ -116,7 +116,7 @@ class TestExtractor < Test::Unit::TestCase
 
       def test_decompose
         data = ChupaText::Data.new
-        data.mime_type = "text/plain"
+        data.mime_type = "text/x-plain"
         data.body = "Hello"
         assert_equal(["Hello", "Hello"], extract(data))
       end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: chupa-text
 version: !ruby/object:Gem::Version
-  version: 1.0.0
+  version: 1.0.1
 platform: ruby
 authors:
 - Kouhei Sutou
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2014-01-04 00:00:00.000000000 Z
+date: 2014-01-05 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler