imw 0.2.7 → 0.2.8

data/lib/imw/tools/summarizer.rb

@@ -11,24 +11,30 @@ module IMW
     # files.  Few large files will not cause a problem.
     class Summarizer
 
+      # Options for this Summarizer.
+      attr_accessor :options
+
       # The inputs given to this Summarizer.
       attr_reader :inputs
 
-      # The resources to this Summarizer, calculated recursively from
-      # its +inputs+.
+      # The resources analyzed, calculated recursively from the
+      # +inputs+.
       attr_reader :resources
 
       include IMW::Tools::ExtensionAnalyzer      
 
       # Initialize a new Summarizer with the given +inputs+.
       #
+      # A Hash of options can be given as the last parameter.
+      #
       # @param [Array<String, IMW::Resource>] inputs
       # @return [IMW::Tools::Summarizer]
       def initialize *inputs
-        self.inputs = inputs.flatten
+        self.options = (inputs.last.is_a?(Hash) && inputs.pop) || {}
+        self.inputs  = inputs.flatten
       end
 
-      # Return the total size.
+      # Return the total size of all resources.
       #
       # @return [Integer]
       def total_size
@@ -43,14 +49,20 @@ module IMW
       # @return [Array<Hash>]
       def summary
         @summary ||= inputs.map do |input|
-          if input.respond_to?(:summary)
-            input.summary rescue {}
-          else
-            {}
-          end
+          #input.guess_schema! if input.schema.nil? && input.respond_to?(:guess_schema!)
+          input.respond_to?(:summary) ? input.summary : {}
         end
       end
 
+      # The metadata employed by this Summarizer.
+      #
+      # It can be set by setting <tt>options[:metadata]</tt>.
+      #
+      # @return [IMW::Metadata, nil]
+      def metadata
+        @metadata ||= options[:metadata] && IMW::Metadata.load(options[:metadata])
+      end
+
       protected
       # Set new inputs for this summarizer.
       #
@@ -61,7 +73,6 @@ module IMW
       def inputs= new_inputs
         @inputs = new_inputs.map do |path_or_resource|
           input = IMW.open(path_or_resource)
-          input.should_exist!("Cannot summarize.")
         end
         @resources = inputs.map do |input|
           input.is_local? && input.is_directory? ? input.all_resources : input

data/lib/imw/utils.rb

@@ -1,8 +1,18 @@
 require 'rubygems'
 require 'fileutils'
+require 'active_support'
 require 'imw/utils/error'
 require 'imw/utils/log'
 require 'imw/utils/paths'
 require 'imw/utils/misc'
 require 'imw/utils/extensions'
 
+module IMW
+
+  # Utility modules.
+  module Utils
+    autoload :DynamicallyExtendable, 'imw/utils/dynamically_extendable'
+    autoload :HasURI,                'imw/utils/has_uri'
+  end
+end
+

data/lib/imw/utils/dynamically_extendable.rb

@@ -0,0 +1,137 @@
+module IMW
+  module Utils
+
+    # Provides an including class with a class-level array of
+    # "handlers" that it can use to dynamically extend its instances
+    # with specific modules only if certain conditions are met.
+    #
+    # This allows different instances of a class to implement very
+    # different behavior at runtime.
+    #
+    # An example use case might be a Database class which dynamically
+    # extends its instances with an adaptor module appropriate to the
+    # particular database the object refers to.
+    module DynamicallyExtendable
+
+      def self.included obj
+        obj.extend(ClassMethods)
+      end
+      
+      # Return the modules this object has been extended by.
+      #
+      # @return [Array]
+      def modules
+        @modules ||= []
+      end
+
+      # Works just like Object#extend except it keeps track of the
+      # modules it has extended.
+      #
+      # @see IMW::Utils::DynamicallyExtendable#modules
+      def extend mod
+        modules << mod
+        super mod
+      end
+
+      # Iterate through this object's class's handlers and extend this
+      # object with the module referred to by any matching handlers.
+      def extend_appropriately! options={}
+        self.class.extend_instance! self, options
+      end
+
+      # A collection of methods which provide a class including
+      # IMW::Utils::DynamicallyExtendable with a class-level Array of
+      # handlers that can be applied to instances by calling the
+      # instance's +extend_appropriately!+ method.
+      module ClassMethods
+
+        # The handlers an including class has defined.
+        #
+        # @return [Array<Array>]
+        def handlers
+          @handlers ||= []
+        end
+        
+        # Register a new handler for an including class.
+        #
+        # +handler+ must be one of
+        #
+        # 1. Regexp 
+        # 2. Proc 
+        # 3. +true+
+        #
+        # In case (1), if the regular expression matches the
+        # instance's +to_s+ method then the module (+mod+) will be
+        # used..
+        #
+        # In case (2), if the Proc returns a value other than +false+
+        # or +nil+ after being passed an instance then the module will
+        # be used.
+        #
+        # In case (3), the module will be used.
+        #
+        # @param [String, Module] mod
+        # @param [Regexp, Proc, true] handler
+        def register_handler mod, handler
+          raise IMW::ArgumentError.new("Module must be either a Module or String")       unless mod.is_a?(Module)     || mod.is_a?(String)
+          raise IMW::ArgumentError.new("Handler must be either a Regexp, Proc, or true") unless handler.is_a?(Regexp) || handler.is_a?(Proc) || handler == true
+          handlers << [mod, handler]
+        end
+
+        # Register a collection of handlers.
+        #
+        # @see IMW::Utils::DynamicallyExtendable::ClassMethods#register_handler
+        def register_handlers *pairs
+          pairs.each { |pair| register_handler *pair }
+        end
+        
+        # Iterate through this class's handlers and extend the given
+        # object with modules whose handler conditions match the
+        # instance.
+        #
+        # Passing in <tt>:use_modules</tt> or <tt>:skip_modules</tt>
+        # allows overriding the default behavior of handlers.
+        #
+        # @param [Object] instance
+        # @param [Hash] options
+        # @option options [Array<String,Module>] use_modules a list of modules used regardless of handlers
+        # @option options [Array<String,Module>] skip_modules a list of modules not to be used regardless of handlers
+        # @return [Object] the newly extended object
+        def extend_instance! instance, options={}
+          return if options[:no_modules]
+          options.reverse_merge!(:use_modules => [], :skip_modules => [])
+          handlers.each do |mod_name, handler|
+            case handler
+            when Regexp    then extend_instance_with_mod_or_string!(instance, mod_name, options[:skip_modules]) if handler =~ instance.to_s
+            when Proc      then extend_instance_with_mod_or_string!(instance, mod_name, options[:skip_modules]) if handler.call(instance)
+            when TrueClass then extend_instance_with_mod_or_string!(instance, mod_name, options[:skip_modules])
+            else           raise IMW::TypeError("A handler must be Regexp, Proc, or true")
+            end
+          end
+          options[:use_modules].each { |mod_name| extend_instance_with_mod_or_string!(instance, mod_name, options[:skip_modules]) }
+          instance
+        end
+        
+        # Extend +instance+ with +mod_or_string+.  Will work hard to
+        # try and interpret +mod_or_string+ as a module if it's a
+        # string.
+        #
+        # @param [Object] instance
+        #
+        # @param [Module, String] mod_or_string the module or string
+        # representing a module to extend the instance with
+        #
+        # @param [Array<Module,String>] skip_modules modules to exclude
+        def extend_instance_with_mod_or_string! instance, mod_or_string, skip_modules
+          return if skip_modules.include?(mod_or_string)
+          if mod_or_string.is_a?(Module)
+            instance.extend(mod_or_string)
+          else
+            m = IMW.class_eval(mod_or_string)
+            instance.extend(m) unless skip_modules.include?(m)
+          end
+        end    
+      end
+    end
+  end
+end

data/lib/imw/utils/error.rb

@@ -25,6 +25,9 @@ module IMW
   # Error communicating with a remote entity.
   ArgumentError = Class.new(Error)
 
+  # Error in defining or matching a schema.
+  SchemaError = Class.new(Error)
+
   # An error meant to be used when a system call goes awry.  It will
   # report exit status and the process id of the offending call.
   class SystemCallError < IMW::Error

data/lib/imw/utils/extensions.rb

@@ -4,10 +4,6 @@ require 'imw/utils/extensions/hash'
 require 'imw/utils/extensions/struct'
 require 'imw/utils/extensions/symbol'
 
-require 'active_support/core_ext/object/blank'
-require 'active_support/core_ext/object/misc'
-
-
 module IMW
   # A replacement for the standard system call which raises an
   # IMW::SystemCallError if the command fails which prints better

data/lib/imw/utils/extensions/array.rb

@@ -1,7 +1,6 @@
-require 'active_support/core_ext/array/extract_options'
-class Array #:nodoc:
-  include ActiveSupport::CoreExtensions::Array::ExtractOptions
-end
+# class Array #:nodoc:
+#   include ActiveSupport::CoreExtensions::Array::ExtractOptions
+# end
 
 class Array
 
@@ -107,9 +106,9 @@ class Array
     terminals
   end
 
-  # Dump the data in this array to the resource at the given +uri+.
-  def dump uri
-    IMW.open!(uri).dump(self)
+  # Emit the data in this array to the resource at the given +uri+.
+  def emit uri
+    IMW.open!(uri).emit(self)
   end
 
 end

data/lib/imw/utils/extensions/hash.rb

@@ -1,5 +1,3 @@
-require 'active_support/core_ext/hash/reverse_merge'
-
 class Hash
   # Return the elements of this hash in a pretty-printed string,
   # inserting +final_string+ between the last two items.
@@ -201,8 +199,8 @@ class Hash
     terminals
   end
 
-  # Dump the data from this Hash into the given +uri+.
-  def dump uri
-    IMW.open!(uri).dump(self)
+  # Emit the data from this Hash into the given +uri+.
+  def emit uri
+    IMW.open!(uri).emit(self)
   end
 end

data/lib/imw/utils/extensions/string.rb

@@ -31,9 +31,9 @@ class String
     self.downcase.underscore.to_sym
   end
 
-  # Dump this string into the given +uri+.
-  def dump uri
-    IMW.open!(uri).dump(self)
+  # Emit this string into the given +uri+.
+  def emit uri
+    IMW.open!(uri).emit(self)
   end
 
 end

data/lib/imw/utils/has_uri.rb

@@ -0,0 +1,114 @@
+require 'addressable/uri'
+
+module IMW
+  module Utils
+
+    # Endows an including class with a wrapper for Addressable::URI
+    module HasURI
+
+      # The URI of this object.
+      attr_reader :uri
+
+      # Set the URI of this resource by parsing the given +uri+ (if
+      # necessary).
+      #
+      # @param [String, Addressable::URI] uri the uri to parse
+      def uri= uri
+        if uri.is_a?(Addressable::URI)
+          @uri = uri
+        else
+          begin
+            @uri = Addressable::URI.parse(uri.to_s)
+          rescue URI::InvalidURIError
+            @uri = Addressable::URI.parse(URI.encode(uri.to_s))
+            @encoded_uri = true
+          end
+        end
+      end
+
+      # The scheme of this resource.  Will be +nil+ for local resources.
+      #
+      # @return [String]
+      def scheme
+        @scheme ||= uri.scheme
+      end
+
+      # The directory name of this resource's path.
+      #
+      # @return [String]
+      def dirname
+        @dirname  ||= File.dirname(path)
+      end
+
+      # The basename of this resource's path.
+      #
+      # @return [String]
+      def basename
+        @basename ||= File.basename(path)
+      end
+
+      # Returns the extension (INCLUDING the '.') of this resource's
+      # path.  Redefine this in an including class for which this is
+      # weird ('.tar.gz' I'm talking to you...)
+      #
+      # @return [String]
+      def extname
+        @extname ||= File.extname(path)
+      end
+
+      # Returns the extension (WITHOUT the '.') of this resource's path.
+      #
+      # @return [String]
+      def extension
+        @extension ||= extname[1..-1] || ''
+      end
+
+      # Returns the basename of the file with its extension removed
+      #
+      #   IMW.open('/path/to/some_file.tar.gz').name # => some_file
+      #
+      # @return [String]
+      def name
+        @name ||= extname ? basename[0,basename.length - extname.length] : basename
+      end
+
+      # Returns the user associated with the host of this URI.
+      #
+      # @return [String]
+      def user
+        @user ||= uri.user
+      end
+
+      # Return the fragment part of this resource's URI.
+      #
+      # Will likely be +nil+ for local resources.
+      #
+      # @return [String]
+      def fragment
+        @fragment ||= uri.fragment
+      end
+
+      # Return the URI of this resource with any query strings and
+      # fragments removed.
+      #
+      # @return [URI::Generic]
+      def stripped_uri
+        uri_args = returning({}) do |args|
+          %w[scheme userinfo host port path].each do |method|
+            args[method.to_sym] = respond_to?(method) ? send(method) : uri.send(method)
+          end
+        end
+        uri.class.new(uri_args)
+      end
+
+      def to_s
+        uri.to_s
+      end
+    end
+  end
+end
+  
+
+  
+  
+

data/spec/data/{sample.csv → formats/delimited/sample.csv}

@@ -3,7 +3,7 @@ ID,Name,Genus,Species
 002,Panamanian Night Monkey,Aotus,zonalis
 003,Hernández-Camacho's Night Monkey,Aotus,jorgehernandezi
 004,Gray-handed Night Monkey,Aotus,griseimembra
-005,Hershkovitz's Night Monkey,Aotus,hershkovitzi
+005,Hershkovitz's Night Monkey,Aotus,hershkovitzi          
 006,Brumback's Night Monkey,Aotus,brumbacki
 007,Three-striped Night Monkey,Aotus,trivirgatus
 008,Spix's Night Monkey,Aotus,vociferans

data/spec/data/formats/delimited/with_schema/ace-hardware-locations.tsv

@@ -0,0 +1,11 @@
+Store Number	Store Name	Address	City	State	Zip	Phone Number	Latitude	Longitude
+10344	Ace Gambles of Hotchkiss	121 E. Bridge Street	Hotchkiss	CO	81419	(970) 872-3535	38.7988	-107.7205
+10347	Al's Ace Hardware	4010 E 29th St	Tucson	AZ	85711	(520) 747-8144	32.1996	-110.9059
+10349	Al's Ace Hardware	4722 E Broadway Blvd	Tucson	AZ	85711	(520) 318-0414	32.2216	-110.8916
+10351	Al's Ace Hardware	2884 N Campbell Ave	Tucson	AZ	85719	(520) 325-2432	32.259	-110.9428
+10353	Lewistown Ace Hardware	815 NE Main St	Lewistown	MT	59457	(406) 538-4000	47.0702	-109.409
+10356	Byram Revell Ace Hardware	5726 Terry Rd	Jackson	MS	39272	(601) 371-8429	32.1919	-90.2542
+10357	Mr C's Ace Hardware	1201 Precinct Line Rd	Hurst	TX	76053	(817) 282-3428	32.831	-97.186
+10358	Gus Bock Ace Hardware	3455 Ridge Rd	Lansing	IL	60438	(708) 474-5940	41.5643	-87.5321
+10359	Elberton Ace Hardware	886 Elbert St	Elberton	GA	30635	(706) 283-3054	34.1008	-82.8495
+10361	Phil's Hardware	13 SW H St	Madras	OR	97741	(541) 475-9392	44.627737	-121.13004