chimps 0.1.2 → 0.1.3

data/VERSION

@@ -1 +1 @@
-0.1.2
+0.1.3

data/lib/chimps/commands/base.rb

@@ -39,7 +39,7 @@ module Chimps
       @argv = argv
       run_options_definers
       parse_command_line!
-      Chimps::Config.load
+      resolve_options!
     end
 
     # The name of this command, including the
@@ -68,6 +68,14 @@ module Chimps
         raise CLIError.new("#{e.message}.  Try `chimps help #{name}'")
       end
     end
+
+    # Ensure that certain options (verbosity, log file) that can be
+    # passed on the command-line override those stored in a
+    # configuration file (if present).
+    def resolve_options!
+      Chimps::Config.load       # load defaults from config file
+      Chimps::CONFIG.merge!(Chimps::COMMAND_LINE_OPTIONS) # overwrites from command line if necessary
+    end
       
     # Run all methods beginning with +define+ and ending with +option+
     # or +options+.
@@ -83,16 +91,21 @@ module Chimps
     # such options at the moment are <tt>-v</tt> (or
     # <tt>--[no-]verbose</tt>) for verbosity, and <tt>-i</tt> (or
     # <tt>--identity-file</tt>) for setting the identify file to use.
+    # <tt>--log-file</tt>) for setting the log file to use.
     def define_common_options
       separator self.class::HELP
       separator "\nOptions include:"
       
-      on("-v", "--[no-]verbose", "Be verbose, or not.") do |v|
-        Chimps::CONFIG[:verbose] = v
+      on("-v", "--[no-]verbose", "Be verbose, or not.") do |verbose|
+        Chimps::COMMAND_LINE_OPTIONS[:verbose] = verbose
       end
       
-      on("-i", "--identity-file PATH", "Use the given YAML identify file to authenticate with Infochimps instead of the default (~/.chimps) ") do |i|
-        Chimps::CONFIG[:identity_file] = File.expand_path(i)
+      on("-i", "--identity-file PATH", "Use the given YAML identify file to authenticate with Infochimps instead of the default (~/.chimps) ") do |path|
+        Chimps::COMMAND_LINE_OPTIONS[:identity_file] = File.expand_path(path)
+      end
+
+      on("-l", "--log-file PATH", "Use the given path to log Chimps output (`-' is interpreted as $stdout).") do |path|
+        Chimps::COMMAND_LINE_OPTIONS[:log_file] = path # don't expand_path as it might be a `-'
       end
     end
 

data/lib/chimps/commands/help.rb

@@ -57,9 +57,9 @@ for any of the commands above.
 
 = Setup
 
-Once you have obtained an API key and secret from Infochimps, place them
-in a file Chimps::CONFIG[:identity_file] in your home directory with the
-following format
+Once you have obtained an API key and secret from Infochimps, place
+them in a file Chimps::CONFIG[:identity_file] in your home directory
+with the following format
 
   ---
   # API credentials for use on the main Infochimps site

data/lib/chimps/config.rb

@@ -1,5 +1,12 @@
 module Chimps
 
+  # Options that can be overriden by the command-line.
+  COMMAND_LINE_OPTIONS = {
+    :identity_file    => File.expand_path(ENV["CHIMPS_RC"] || "~/.chimps"),
+    # log_file -- will be specified on command line
+    # verbose  -- will be specified on command line
+  }
+
   # Default configuration for Chimps.  User-specific configuration
   # lives in a YAML file <tt>~/.chimps</tt>.
   CONFIG = {
@@ -9,14 +16,12 @@ module Chimps
     :site => {
       :host => ENV["CHIMPS_HOST"]       || 'http://infochimps.org'
     },
-    :identity_file    => File.expand_path(ENV["CHIMPS_RC"] || "~/.chimps"),
-    :verbose          => nil,
     :timestamp_format => "%Y-%m-%d_%H-%M-%S"
   }
 
   # Is Chimps in verbose mode?
   #
-  # @return [true, false]
+  # @return [true, false, nil]
   def self.verbose?
     CONFIG[:verbose]
   end
@@ -42,9 +47,9 @@ module Chimps
     # file.
     def self.load
       # FIXME this is a terrible hack...and it only goes to 2 deep!
-      if File.exist?(CONFIG[:identity_file])
+      if File.exist?(COMMAND_LINE_OPTIONS[:identity_file])
         require 'yaml'
-        YAML.load_file(CONFIG[:identity_file]).each_pair do |key, value|
+        YAML.load_file(COMMAND_LINE_OPTIONS[:identity_file]).each_pair do |key, value|
           if value.is_a?(Hash) && CONFIG.include?(key)
             CONFIG[key].merge!(value)
           else

data/lib/chimps/request.rb

@@ -103,7 +103,7 @@ module Chimps
     # @return [Chimps::Response]
     def get options={}
       handle_exceptions do
-        puts "GET #{url}" if Chimps.verbose?
+        Chimps.log.info("GET #{url}")
         Response.new(super(DEFAULT_HEADERS.merge(options)))
       end
     end
@@ -117,7 +117,7 @@ module Chimps
     # @return [Chimps::Response]
     def post options={}
       handle_exceptions do
-        puts "POST #{url}" if Chimps.verbose?
+        Chimps.log.info("POST #{url}")
         Response.new(super(data_text, DEFAULT_HEADERS.merge(options)))
       end
     end
@@ -131,7 +131,7 @@ module Chimps
     # @return [Chimps::Response]
     def put options={}
       handle_exceptions do
-        puts "PUT #{url}" if Chimps.verbose?
+        Chimps.log.info("PUT #{url}")
         Response.new(super(data_text, DEFAULT_HEADERS.merge(options)))
       end
     end
@@ -146,7 +146,7 @@ module Chimps
     # @return [Chimps::Response]
     def delete options={}
       handle_exceptions do
-        puts "DELETE #{url}" if Chimps.verbose?
+        Chimps.log.info("DELETE #{url}")
         Response.new(super(DEFAULT_HEADERS.merge(options)))
       end
     end
@@ -175,20 +175,10 @@ module Chimps
       query_params[:api_key]      = Chimps::CONFIG[:site][:key]
     end
 
-    # Return the sorted keys of the query params.
-    #
-    # @return [Array]
-    def alphabetical_params
-      query_params.keys.map(&:to_s).sort
-    end
-
     # Return an unsigned query string for this request.
     #
-    # Query parameters will be used in alphabetical order.
-    #
     # @return [String]
     def unsigned_query_string
-      # alphabetical_params.map { |key| "#{CGI::escape(key.to_s)}=#{CGI::escape(query_params[key.to_sym].to_s)}" }.join("&") # doesn't flatten nested hashes properly
       RestClient::Payload.generate(query_params)
     end
 
@@ -200,8 +190,7 @@ module Chimps
     #
     # @return [String]
     def unsigned_query_string_stripped
-      require 'cgi'      
-      @query_params_text ||= alphabetical_params.map { |key| CGI::escape(key.to_s) + CGI::escape(query_params[key.to_sym].to_s) }.join('')
+      @query_params_text ||= obj_to_stripped_string(query_params)
     end
 
     # Return the data of this request as a string.
@@ -238,6 +227,18 @@ module Chimps
       "#{unsigned_query_string}&signature=#{signature}"
     end
 
+    # Turn +obj+ into a string, sorting on internal keys.
+    #
+    # @param [Hash, Array, String] obj
+    # @return [String]
+    def obj_to_stripped_string obj
+      case obj
+      when Hash   then obj.keys.map(&:to_s).sort.map { |key| [key.to_s.downcase, obj_to_stripped_string(obj[key.to_sym])].join('') }.join('')
+      when Array  then obj.map { |e| obj_to_stripped_string(e) }.join('')
+      else             obj.to_s
+      end
+    end
+    
   end
 
   # A class to encapsulate requests made against the Infochimps paid

data/lib/chimps/utils/log.rb

@@ -0,0 +1,50 @@
+module Chimps
+
+  # The Chimps logger.  Set via Chimps::CONFIG[:log_file] and defaults
+  # to $stdout.
+  #
+  # @return [Logger]
+  def self.log
+    @log ||= Log.new_logger
+  end
+
+  # Set the Chimps logger.
+  #
+  # @param [Logger] new_log
+  def self.log= new_log
+    @log = new_log
+  end
+
+  # Module for initializing the Chimps logger from configuration
+  # settings.
+  module Log
+
+    # Initialize a new Logger instance with the log level set by
+    # Chimps.verbose?
+    #
+    # @return [Logger]
+    def self.new_logger
+      require 'logger'
+      returning(Logger.new(log_file)) do |log|
+        log.progname = "Chimps"
+        log.level    = Chimps.verbose? ? Logger::INFO : Logger::WARN
+      end
+    end
+
+    protected
+    # Return either the path to the log file in
+    # Chimps::CONFIG[:log_file] or $stdout if the path is blank or
+    # equal to `-'.
+    #
+    # @return [String, $stdout] the path to the log or $stdout
+    def self.log_file
+      if Chimps::CONFIG[:log_file]
+        Chimps::CONFIG[:log_file].strip == '-' ? $stdout : Chimps::CONFIG[:log_file]
+      else
+        $stdout
+      end
+    end
+  end
+end
+
+    

data/lib/chimps/utils.rb

@@ -1,6 +1,7 @@
 require 'chimps/config'
 require 'chimps/utils/extensions'
 require 'chimps/utils/error'
+require 'chimps/utils/log'
 
 module Chimps
   module Utils
@@ -8,4 +9,5 @@ module Chimps
     autoload :UsesModel,    'chimps/utils/uses_model'
     autoload :UsesYamlData, 'chimps/utils/uses_yaml_data'
   end
+
 end

data/lib/chimps/workflows/uploader.rb

@@ -47,8 +47,7 @@ module Chimps
       # @option options [String, IMW::Resource] archive the path to the archive to create (defaults to IMW::Workflows::Downloader#default_archive_path)
       # @option options [String] fmt the data format to annotate the upload with
       def initialize options={}
-        require 'imw'
-        IMW.verbose      = Chimps.verbose?
+        require_imw
         @dataset         = options[:dataset] or raise PackagingError.new("Must provide the ID or handle of a dataset to upload data to.")
         self.local_paths = options[:local_paths]   # must come before self.archive=
         self.archive     = options[:archive]
@@ -105,24 +104,43 @@ module Chimps
         @archive = potential_package
       end
 
+      # Return the summarizer responsible for summarizing data on this
+      # upload.
+      #
+      # @return [IMW::Tools::Summarizer]
+      def summarizer
+        @summarizer ||= IMW::Tools::Summarizer.new(local_paths)
+      end
+
       # Set the data format to annotate the upload with.
       #
       # If not provided, Chimps will use the Infinite Monkeywrench
       # (IMW) to try and guess the data format.  See
       # IMW::Tools::Summarizer for more information.
       def fmt= new_fmt=nil
-        @fmt ||= new_fmt || IMW::Tools::Summarizer.new(local_paths).most_common_data_format
+        @fmt ||= new_fmt || summarizer.most_common_data_format
       end
 
       # The default path to the archive that will be built.
       #
-      # Defaults to a ZIP file in the current directory named after
-      # the +dataset+'s ID or handle and the current time.
+      # Defaults to a file in the current directory named after the
+      # +dataset+'s ID or handle and the current time.  The package
+      # format (<tt>.zip</tt> or <tt>.tar.bz2</tt>) is determined by
+      # size, see
+      # Chimps::Workflows::Uploader#default_archive_extension.
       #
       # @return [String]
       def default_archive_path
         # in current working directory...
-        "chimps_#{dataset}-#{Time.now.strftime(Chimps::CONFIG[:timestamp_format])}.zip"
+        "chimps_#{dataset}-#{Time.now.strftime(Chimps::CONFIG[:timestamp_format])}.#{default_archive_extension}"
+      end
+
+      # Use <tt>zip</tt> if the data is less than 500 MB in size and
+      # <tt>tar.bz2</tt> otherwise.
+      #
+      # @return ['tar.bz2', 'zip']
+      def default_archive_extension
+        summarizer.total_size >= 524288000 ? 'tar.bz2' : 'zip'
       end
 
       # The URL to the <tt>README-infochimps</tt> file on Infochimps'
@@ -222,7 +240,7 @@ module Chimps
       #
       # @return [Hash]
       def package_data
-        { :package => {:path => token['key'], :fmt => token['fmt'], :pkg_size => archive.size, :pkg_fmt => archive.extension} }
+        { :package => {:path => token['key'], :fmt => token['fmt'], :pkg_size => archive.size, :pkg_fmt => archive.extension, :summary => summarizer.summary, :token_timestamp => token['timestamp'] } }
       end
 
       # Make a final POST request to Infochimps, creating the final
@@ -232,6 +250,17 @@ module Chimps
         package_creation_response.print
         raise UploadError.new("Unable to notify Infochimps of newly uploaded data.") if package_creation_response.error?
       end
+
+      protected
+      # Require IMW and match the IMW logger to the Chimps logger.
+      def require_imw
+        begin
+          require 'imw'
+        rescue LoadError
+          raise Chimps::Error.new("The Infinite Monkeywrench (IMW) gem is required to upload.")
+        end
+        IMW.verbose = Chimps.verbose?
+      end
       
     end
   end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification 
 name: chimps
 version: !ruby/object:Gem::Version 
-  version: 0.1.2
+  version: 0.1.3
 platform: ruby
 authors: 
 - Dhruv Bansal
@@ -9,7 +9,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2010-05-25 00:00:00 -05:00
+date: 2010-06-06 00:00:00 -05:00
 default_executable: chimps
 dependencies: 
 - !ruby/object:Gem::Dependency 
@@ -105,6 +105,7 @@ files:
 - lib/chimps/utils.rb
 - lib/chimps/utils/error.rb
 - lib/chimps/utils/extensions.rb
+- lib/chimps/utils/log.rb
 - lib/chimps/utils/uses_curl.rb
 - lib/chimps/utils/uses_model.rb
 - lib/chimps/utils/uses_yaml_data.rb