best_type 0.0.4 → 0.0.5

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 065e858091f8c79dc0a6b0ae0ef5a4afeeaa300007a866e0b2072fd1cffee991
-  data.tar.gz: 348dd2f451a751baba7af76ad3f40c9dd394f1921fee3eb2196a65693cf10073
+  metadata.gz: 9e8b2cfe247c6e8a1a94dada5b9ffc4538b2154d5afcc002a0718007c7be5cd9
+  data.tar.gz: 313248015c4f2f2e364ed4da2197173038c183b3083ae8e26a0e82cd4066de1f
 SHA512:
-  metadata.gz: a39c2b4c490befb5f4f2b554e39c2f26af4773dbfda28c9ccdc878ad9731c8a8cb5eb37cae301a13427de2d2235a7d608493b61c791ed57d29bf1136e605bb37
-  data.tar.gz: f5989894ec08dbe385b866990fe4477da5028efbbc5533391681b1b165b05daa39087e11098344425ca04d260cb034c7fc3912e43447437c1d35226c6408447b
+  metadata.gz: aa0414efd2ee4510098cf769dd18b54759edeaa9d833a9f9ce2fed3b32a2921325a16d15cba5f4609ffb5b3334c39b9f7d56c94c427c42933c55f7be9db40d0f
+  data.tar.gz: ef907355f8e95552c84d55661b81f4ddb3d6bd4ec9b4901d01f7a98cc3eea27f59ae807c5d933fa3c2281d79b253d81d388e69af3295fb2eead756347c083eee

data/README.md

@@ -28,6 +28,12 @@ BestType.dc_type.for_file_name('/path/to/some/file.jpg') # 'StillImage'
 
 # detect dc type for mime types
 BestType.dc_type.for_mime_type('image/jpeg') # 'StillImage'
+
+# detect dc type for file names (including full file path)
+BestType.pcdm_type.for_file_name('/path/to/some/file.jpg') # 'Image'
+
+# detect dc type for mime types
+BestType.pcdm_type.for_mime_type('image/jpeg') # 'Image'
 ```
 
 ### Add Custom Overrides
@@ -36,12 +42,14 @@ BestType.configure({
   extension_to_mime_type_overrides:
     'custom': 'custom/type'
   mime_type_to_dc_type_overrides:
-    'custom/type': 'Custom'
+    'custom/type': 'CustomDC'
+  mime_type_to_pcdm_type_overrides:
+    'custom/type': 'CustomPCDM'
 })
 
 BestType.mime_type.for_file_name('myfile.custom') # 'custom/type'
-BestType.dc_type.for_file_name('myfile.custom') # 'Custom'
-BestType.dc_type.for_mime_type('custom/type') # 'Custom'
+BestType.dc_type.for_file_name('myfile.custom') # 'CustomDC'
+BestType.pcdm_type.for_mime_type('custom/type') # 'CustomPCDM'
 
 ```
 
@@ -63,6 +71,8 @@ BestType.configure({
     'custom': 'custom/type'
   mime_type_to_dc_type_overrides:
     'custom/type': 'Custom'
+  mime_type_to_pcdm_type_overrides:
+    'custom/type': 'Custom'
 })
 ```
 
@@ -81,18 +91,24 @@ development:
     'good': 'good/type'
   mime_type_to_dc_type_overrides:
     'good/type': 'Good'
+  mime_type_to_pcdm_type_overrides:
+    'good/type': 'Goodly'
     
 test:
   extension_to_mime_type_overrides:
     'better': 'better/type'
   mime_type_to_dc_type_overrides:
     'better/type': 'Better'
+  mime_type_to_pcdm_type_overrides:
+    'best/type': 'Betterly'
 
 production:
   extension_to_mime_type_overrides:
     'best': 'best/type'
   mime_type_to_dc_type_overrides:
     'best/type': 'Best'
+  mime_type_to_pcdm_type_overrides:
+    'best/type': 'Bestly'
 ```
 
 ### Running Tests (for developers):

data/config/internal_config_options.yml

@@ -6,3 +6,5 @@ mime_type_to_dc_type_overrides:
   'application/mxf': 'MovingImage'
   'application/vnd.openxmlformats-officedocument.wordprocessingml.document': 'Text'
   'application/vnd.openxmlformats-officedocument.spreadsheetml.sheet': 'Dataset'
+mime_type_to_pcdm_type_overrides:
+  'test/type': 'Test'

data/lib/best_type.rb

@@ -2,6 +2,7 @@ require 'best_type/version'
 require 'best_type/config'
 require 'best_type/mime_type_lookup'
 require 'best_type/dc_type_lookup'
+require 'best_type/pcdm_type_lookup'
 require 'yaml'
 
 module BestType
@@ -15,6 +16,10 @@ module BestType
     @dc_type ||= BestType::DcTypeLookup.new(mime_type)
   end
 
+  def self.pcdm_type
+    @pcdm_type ||= BestType::PcdmTypeLookup.new(mime_type)
+  end
+
   def self.config(reload = false, user_config_options = {})
     if @config.nil? || reload
       @semaphore.synchronize do

data/lib/best_type/config.rb

@@ -1,7 +1,7 @@
 module BestType
   class Config
 
-    attr_reader :extension_to_mime_type_overrides, :mime_type_to_dc_type_overrides
+    attr_reader :extension_to_mime_type_overrides, :mime_type_to_dc_type_overrides, :mime_type_to_pcdm_type_overrides
 
     def initialize(user_config_options = {})
       # Get defaults from internal_custom_mapping.yml in gem
@@ -9,12 +9,14 @@ module BestType
       internal_config_file_path = File.join(gem_dir, 'config/internal_config_options.yml')
       internal_config_options = YAML.load_file(internal_config_file_path)
 
-      @extension_to_mime_type_overrides = internal_config_options['extension_to_mime_type_overrides']
-      @mime_type_to_dc_type_overrides = internal_config_options['mime_type_to_dc_type_overrides']
+      @extension_to_mime_type_overrides = internal_config_options['extension_to_mime_type_overrides'] || {}
+      @mime_type_to_dc_type_overrides = internal_config_options['mime_type_to_dc_type_overrides'] || {}
+      @mime_type_to_pcdm_type_overrides = internal_config_options['mime_type_to_pcdm_type_overrides'] || {}
 
       stringify_user_config_options_keys!(user_config_options)
       add_extension_to_mime_type_overrides(user_config_options['extension_to_mime_type_overrides']) if user_config_options.key?('extension_to_mime_type_overrides')
       add_mime_type_to_dc_type_overrides(user_config_options['mime_type_to_dc_type_overrides']) if user_config_options.key?('mime_type_to_dc_type_overrides')
+      add_mime_type_to_pcdm_type_overrides(user_config_options['mime_type_to_pcdm_type_overrides']) if user_config_options.key?('mime_type_to_pcdm_type_overrides')
     end
 
     private
@@ -27,6 +29,10 @@ module BestType
       @mime_type_to_dc_type_overrides.merge!(overrides)
     end
 
+    def add_mime_type_to_pcdm_type_overrides(overrides)
+      @mime_type_to_pcdm_type_overrides.merge!(overrides)
+    end
+
     def stringify_user_config_options_keys!(user_config_options)
       user_config_options_keys = user_config_options.keys
       user_config_options_keys.each do |key|

data/lib/best_type/dc_type_lookup.rb

@@ -3,13 +3,39 @@ module BestType
 
     attr_reader :config
 
-    FALLBACK_DC_TYPE = 'Software'.freeze
+    COLLECTION = 'Collection'.freeze
+    DATASET = 'Dataset'.freeze
+    EVENT = 'Event'.freeze
+    INTERACTIVE_RESOURCE = 'InteractiveResource'.freeze
+    MOVING_IMAGE = 'MovingImage'.freeze
+    PHYSICAL_OBJECT = 'PhysicalObject'.freeze
+    SERVICE = 'Service'.freeze
+    SOFTWARE = 'Software'.freeze
+    SOUND = 'Sound'.freeze
+    STILL_IMAGE = 'StillImage'.freeze
+    TEXT = 'Text'.freeze
+
+    # these include values that will not be derived from MIME/content types
+    VALID_TYPES = [
+      COLLECTION, EVENT, INTERACTIVE_RESOURCE, MOVING_IMAGE, PHYSICAL_OBJECT,
+      SERVICE, SOFTWARE, SOUND, STILL_IMAGE, TEXT
+    ].freeze
+
+    FALLBACK_DC_TYPE = SOFTWARE
 
     def initialize(mime_type_lookup_instance)
       @mime_type_lookup = mime_type_lookup_instance
       @config = @mime_type_lookup.config
     end
 
+    def fallback_type
+      FALLBACK_DC_TYPE
+    end
+
+    def valid_type?(value)
+      VALID_TYPES.include? value
+    end
+
     def for_file_name(file_name_or_path)
       for_mime_type(@mime_type_lookup.for_file_name(file_name_or_path))
     end
@@ -20,13 +46,13 @@ module BestType
       return dc_type unless dc_type.nil?
 
       mimes_to_dc = {
-        /^image/ => 'StillImage',
-        /^video/ => 'MovingImage',
-        /^audio/ => 'Sound',
-        /^text/ => 'Text',
-        /^application\/(pdf|msword)/ => 'Text',
-        /excel|spreadsheet|xls|application\/sql/ => 'Dataset',
-        /^application/ => 'Software'
+        /^image/ => STILL_IMAGE,
+        /^video/ => MOVING_IMAGE,
+        /^audio/ => SOUND,
+        /^text/ => TEXT,
+        /^application\/(pdf|msword)/ => TEXT,
+        /excel|spreadsheet|xls|application\/sql/ => DATASET,
+        /^application/ => SOFTWARE
       }
 
       dc_type = mimes_to_dc.find { |pattern, _type_val| mime_type =~ pattern }

data/lib/best_type/pcdm_type_lookup.rb

@@ -0,0 +1,113 @@
+module BestType
+  class PcdmTypeLookup
+
+    attr_reader :config
+
+    # https://github.com/duraspace/pcdm/blob/master/pcdm-ext/file-format-types.rdf
+    ARCHIVE = "Archive".freeze
+    AUDIO = "Audio".freeze
+    DATABASE = "Database".freeze
+    DATASET = "Dataset".freeze
+    EMAIL = "Email".freeze
+    FONT = "Font".freeze
+    HTML = "HTML".freeze
+    IMAGE = "Image".freeze
+    PAGE_DESCRIPTION = "PageDescription".freeze
+    PRESENTATION = "Presentation".freeze
+    SOFTWARE = "Software".freeze
+    SOURCE_CODE = "SourceCode".freeze
+    SPREADSHEET = "Spreadsheet".freeze
+    STRUCTURED_TEXT = "StructuredText".freeze
+    TEXT = "Text".freeze
+    UNKNOWN = "Unknown".freeze
+    UNSTRUCTURED_TEXT = "UnstructuredText".freeze
+    VIDEO = "Video".freeze
+    WEBSITE = "Website".freeze
+
+    # these include values that will not be derived from MIME/content types
+    VALID_TYPES = [
+      ARCHIVE, AUDIO, DATABASE, DATASET, EMAIL, FONT, HTML, IMAGE, PAGE_DESCRIPTION, PRESENTATION, SOFTWARE,
+      SOURCE_CODE, SPREADSHEET, STRUCTURED_TEXT, TEXT, UNKNOWN, UNSTRUCTURED_TEXT, VIDEO, WEBSITE
+    ].freeze
+
+    FALLBACK_TYPE = UNKNOWN
+
+    def initialize(mime_type_lookup_instance)
+      @mime_type_lookup = mime_type_lookup_instance
+      @config = @mime_type_lookup.config
+    end
+
+    def fallback_type
+      FALLBACK_TYPE
+    end
+
+    def valid_type?(value)
+      VALID_TYPES.include? value
+    end
+
+    def for_file_name(file_name_or_path)
+      for_mime_type(@mime_type_lookup.for_file_name(file_name_or_path))
+    end
+
+    def for_mime_type(mime_type)
+      # Check config overrides first
+      file_type = @config.mime_type_to_pcdm_type_overrides.fetch(mime_type, nil)
+      return file_type unless file_type.nil?
+
+      mimes_to_type = {
+        /^image/i => IMAGE,
+        /^video/i => VIDEO,
+        /^audio/i => AUDIO,
+        /^text/i => {
+          /\/css/i => SOURCE_CODE,
+          /\/html/i => HTML,
+          /.+/ => TEXT
+        },
+        /excel|spreadsheet|xls/i => SPREADSHEET,
+        /application\/sql/i => DATABASE,
+        /csv/i => DATASET,
+        /octet.stream/i => UNKNOWN,
+        /^application/i => {
+          /\/access/i => DATABASE,
+          /\/css/i => SOURCE_CODE,
+          /\/html/i => HTML,
+          /\/mbox/i => EMAIL,
+          /\/mp4/i => VIDEO,
+          /\/mp4a/i => AUDIO,
+          /\/msaccess/i => DATABASE,
+          /\/mxf/i => VIDEO,
+          /\/(pdf|msword)/i => PAGE_DESCRIPTION,
+          /\/postscript/i => PAGE_DESCRIPTION,
+          /\/powerpoint/i => PRESENTATION,
+          /\/rtf/i => PAGE_DESCRIPTION,
+          /\/sql/i => DATABASE,
+          /\/swf/ => VIDEO,
+          /\/vnd.ms-asf/i => VIDEO,
+          /\/vnd.ms-word/i => PAGE_DESCRIPTION,
+          /\/vnd.ms-wpl/i => PAGE_DESCRIPTION,
+          /\/vnd.oasis.opendocument.text/i => PAGE_DESCRIPTION,
+          /\/vnd.openxmlformats-officedocument.presentation/i => PRESENTATION,
+          /\/vnd.openxmlformats-officedocument.wordprocessingml/i => PAGE_DESCRIPTION,
+          /\/vnd.ms-powerpoint/i => PRESENTATION,
+          /\/vnd.sun.xml.calc/i => SPREADSHEET,
+          /\/vnd.sun.xml.impress/i => PRESENTATION,
+          /\/vnd.sun.xml.writer/i => PAGE_DESCRIPTION,
+          /\/xml/i => STRUCTURED_TEXT,
+          /\/x.mspublisher/i => PAGE_DESCRIPTION,
+          /\/x.shockwave-flash/ => VIDEO,
+          /\/x.spss/i => DATASET,
+          /\/zip/i => ARCHIVE,
+          /.+/ => UNKNOWN
+        }
+      }
+
+      file_type = mimes_to_type.detect { |pattern, _type_val| mime_type =~ pattern }
+      return fallback_type unless file_type
+      if file_type&.last.is_a? Hash
+        file_type = file_type.last.detect { |pattern, _type_val| mime_type =~ pattern }
+      end
+      file_type.nil? ? fallback_type : file_type.last
+    end
+
+  end
+end

data/lib/best_type/version.rb

@@ -1,6 +1,6 @@
 module BestType
 
-  VERSION = '0.0.4'.freeze
+  VERSION = '0.0.5'.freeze
 
   def self.version
     VERSION

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: best_type
 version: !ruby/object:Gem::Version
-  version: 0.0.4
+  version: 0.0.5
 platform: ruby
 authors:
 - Eric O'Hanlon
@@ -106,6 +106,7 @@ files:
 - lib/best_type/config.rb
 - lib/best_type/dc_type_lookup.rb
 - lib/best_type/mime_type_lookup.rb
+- lib/best_type/pcdm_type_lookup.rb
 - lib/best_type/version.rb
 - lib/tasks/best_type.rake
 - lib/tasks/best_type/ci.rake
@@ -128,8 +129,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubyforge_project: 
-rubygems_version: 2.7.7
+rubygems_version: 3.0.6
 signing_key: 
 specification_version: 4
 summary: A library for selecting the best mime type or dc type for a file.