chimps-cli 0.0.1

data/lib/chimps-cli/commands/list.rb

@@ -0,0 +1,48 @@
+module Chimps
+  module Commands
+
+    # A command to issue a GET request against an index of resources
+    # at Infochimps.
+    class List < Chimps::Command
+
+      include Chimps::Utils::HttpFormat
+      include Chimps::Utils::ActsOnResource
+
+      def self.default_response_fmt
+        'tsv'
+      end
+
+      def self.allowed_response_fmts
+        super().concat(['tsv'])
+      end
+
+      def self.allowed_resources
+        %w[collection dataset source license]
+      end
+      
+      USAGE = "usage: chimps [OPTIONS] list [RESOURCE]"
+      HELP   = <<EOF
+
+List #{default_resources_type} on Infochimps.
+
+#{resources_listing}
+
+Examples:
+
+  $ chimps list                 # list all datasets
+  $ chimps list --my            # list your datasets
+  $ chimps list licenses        # list licenses
+  $ chimps list --my sources    # list your sources
+EOF
+
+      # Issue the GET request.
+      def execute!
+        query_params = {}
+        query_params[:skip_header] = true if config[:skip_column_names]
+        Request.new(resources_path, :query_params => query_params, :sign => config[:my]).get(headers).print(config)
+      end
+
+    end
+  end
+end
+

data/lib/chimps-cli/commands/me.rb

@@ -0,0 +1,30 @@
+module Chimps
+  module Commands
+
+    class Me < Chimps::Command
+
+      include Chimps::Utils::HttpFormat
+
+      def self.default_response_fmt
+        'yaml'
+      end
+
+      USAGE = "usage: chimps me [OPTIONS]"
+      HELP   = <<EOF
+
+Show a summary of your account.
+
+Examples:
+
+  $ chimps me
+  $ chimps me --response_format=json
+EOF
+
+      def execute!
+        Chimps::Request.new("/me" + ".#{response_fmt}", :sign => true).get.print
+      end
+
+    end
+  end
+end
+

data/lib/chimps-cli/commands/post.rb

@@ -0,0 +1,33 @@
+module Chimps
+  module Commands
+
+    class Post < Chimps::Command
+
+      include Chimps::Utils::HttpFormat
+      include Chimps::Utils::ExplicitPath
+      include Chimps::Utils::UsesParamValueData
+
+      USAGE = "usage: chimps post [OPTIONS] PATH [PARAM=VALUE] ..."
+      HELP   = <<EOF
+
+Send a GET request to the given PATH at Infochimps.
+
+Examples:
+
+  # create a dataset
+  $ chimps post /datasets title="List of Chimps" description="Awesome..."
+
+  # create a source
+  $ chimps post /sources --data=/path/to/some/metadata.json
+EOF
+      
+      def execute!
+        response = Chimps::Request.new(path, :query_params => query_params, :data => (data || true), :sign => config[:sign]).post(headers)
+        response.print_headers if config[:headers]
+        response.print
+      end
+
+    end
+  end
+end
+

data/lib/chimps-cli/commands/put.rb

@@ -0,0 +1,33 @@
+module Chimps
+  module Commands
+
+    class Put < Chimps::Command
+
+      include Chimps::Utils::HttpFormat
+      include Chimps::Utils::ExplicitPath
+      include Chimps::Utils::UsesParamValueData
+
+      USAGE = "usage: chimps put [OPTIONS] PATH [PARAM=VALUE] ..."
+      HELP   = <<EOF
+
+Send a GET request to the given PATH at Infochimps.
+
+Examples:
+
+  # update a dataset
+  $ chimps put /datasets/my-dataset title="A new title"
+
+  # update a source
+  $ chimps put /sources/my-source --data=/path/to/new/metadata.yaml
+EOF
+      
+      def execute!
+        response = Chimps::Request.new(path, :query_params => query_params, :data => (data || true), :sign => config[:sign]).put(headers)
+        response.print_headers if config[:headers]
+        response.print
+      end
+
+    end
+  end
+end
+

data/lib/chimps-cli/commands/query.rb

@@ -0,0 +1,58 @@
+module Chimps
+  module Commands
+
+    # A command to issue a GET request against the Infochimps paid
+    # query API.
+    class Query < Chimps::Command
+
+      include Chimps::Utils::UsesParamValueData      
+
+      USAGE = "usage: chimps query [OPTIONS] ENDPOINT [PROP=VALUE] ..."
+      HELP   = <<EOF
+
+Send a query to the given ENDPOINT on the Infochimps Query API with the
+given parameters.
+
+#{how_to_input_data}
+You can learn more about the Infochimps query API, discover datasets
+to query, and look up the available parameters at
+
+  http://api.infochimps.com
+
+Examples:
+
+  $ chimps query soc/net/tw/trstrank screen_name=infochimps
+  $ chimps query soc/net/tw/trstrank?screen_name=infochimps
+  $ chimps query web/an/de/demographics ip=146.6.180.1
+  $ chimps query soc/net/tw/trstrank --data=trstrank_queries.yml
+EOF
+
+      def endpoint
+        raise CLIError.new(self.class::USAGE) unless config.argv.size > 1 && (! config.argv[1].empty?)
+        config.argv[1]
+      end
+
+      def requests
+        # ensure_data_is_present! unless endpoint.include?("?")
+        if data.is_a?(Hash)
+          [QueryRequest.new(endpoint, :query_params => data, :sign => true)]
+        else
+          data.map { |params| QueryRequest.new(endpoint, :query_params => params, :sign => true) }
+        end
+      end
+      
+      def execute!
+        requests.each do |request|
+          response = request.get
+          if response.error?
+            response.print :to => $stderr
+          else
+            puts(config[:pretty] ? JSON.pretty_generate(JSON.parse(response.body)) : response.body)
+          end
+        end
+      end
+      
+    end
+  end
+end
+

data/lib/chimps-cli/commands/search.rb

@@ -0,0 +1,54 @@
+module Chimps
+  module Commands
+
+    class Search < Chimps::Command
+
+      include Chimps::Utils::HttpFormat
+      include Chimps::Utils::ActsOnResource
+
+      def self.allowed_resources
+        %w[dataset]
+      end
+
+      def self.allowed_response_fmts
+        super().concat(['tsv'])
+      end
+
+      def self.default_response_fmt
+        'tsv'
+      end
+      
+      USAGE = "usage: chimps search [OPTIONS] QUERY"
+      HELP   = <<EOF
+
+Search for #{default_resources_type} on Infochimps for the given
+QUERY.
+
+Examples:
+
+  $ chimps search music                    # search all datasets for 'music'
+  $ chimps search --my 'finance AND nyse'  # search your datasets for 'finance' and 'nyse'
+EOF
+      
+      def query
+        raise CLIError.new(self.class::USAGE) if config.rest.size < 2
+        config.rest[1..-1].join(' ')
+      end
+      
+      def query_params
+        {:query => query}.tap do |qp|
+          qp[:skip_header] = true if config[:skip_column_names]
+        end
+      end
+
+      def path
+        config[:my] ? "/my/search.#{response_fmt}" : "/search.#{response_fmt}"
+      end
+
+      def execute!
+        Request.new(path, :query_params => query_params, :sign => config[:my]).get(headers).print(config)
+      end
+    end
+  end
+end
+

data/lib/chimps-cli/commands/show.rb

@@ -0,0 +1,40 @@
+module Chimps
+  module Commands
+
+    class Show < Chimps::Command
+
+      include Chimps::Utils::ActsOnResource
+      include Chimps::Utils::HttpFormat
+
+      def self.default_response_fmt
+        'yaml'
+      end
+
+      def self.allowed_resources
+        %w[dataset collection source license user]
+      end
+      
+      USAGE = "usage: chimps show [OPTIONS] [RESOURCE] SLUG_OR_ID"
+      HELP   = <<EOF
+
+Return a description of the #{default_resource_type} with the given
+ID or or slug.
+
+#{resources_listing}
+
+Examples:
+
+  $ chimps show musicbrainz           # show musicbrainz dataset
+  $ chimps show 11598                 # show musicbrainz dataset
+  $ chimps show source musicbrainzorg # show musicbrainz.org source page
+  $ chimps show user Infochimps       # show Infochimps' user page
+EOF
+
+      def execute!
+        puts Chimps::Request.new(resource_path + ".#{response_fmt}").get.body
+      end
+
+    end
+  end
+end
+

data/lib/chimps-cli/commands/test.rb

@@ -0,0 +1,37 @@
+module Chimps
+  module Commands
+
+    class Test < Chimps::Command
+
+      USAGE = "usage: chimps [OPTIONS] test"
+      HELP = <<EOF
+
+Print diagnostic information on the API credentials being used by chimps
+and send a test request to Infochimps to make sure the API credentials
+work.
+EOF
+
+      def execute!
+        dataset_response = Chimps::Request.new("/me", :sign => true).get
+        query_response   = Chimps::QueryRequest.new("soc/net/tw/trstrank", :query => {:screen_name => :infochimps}).get
+
+        if dataset_response.error?
+          puts "Could not authenticate with Infochimps Dataset API."
+        else
+          
+          dataset_response.parse!
+          username = dataset_response['user']['username']
+          puts "Authenticated as user '#{username}' for Infochimps Dataset API at http://www.infochimps.com"
+        end
+
+        if query_response.error?
+          puts "Could not authenticate with Infochimps Query API."
+        else
+          puts "Authenticated for Infochimps Query API at http://api.infochimps.com"
+        end
+      end
+
+    end
+  end
+end
+

data/lib/chimps-cli/commands/update.rb

@@ -0,0 +1,38 @@
+module Chimps
+  module Commands
+
+    # A command to issue a PUT request to update a resource at
+    # Infochimps.
+    class Update < Chimps::Command
+
+      include Chimps::Utils::ActsOnResource
+      include Chimps::Utils::UsesParamValueData
+
+      def self.allowed_models
+        %w[dataset source license]
+      end
+      
+      USAGE = "usage: chimps update [OPTIONS] [RESOURCE] ID_OR_SLUG [PROP=VALUE] ..."
+      HELP   = <<EOF
+
+Updates a #{default_resource_type} identified by the given ID or slug using the properties
+ and values supplied.
+
+#{how_to_input_data}
+#{resources_listing}
+
+Examples:
+
+  $ chimps update my-awesome-dataset title='Yet More Awesome Dataset' description="It is even cooler"
+  $ chimps update source my-source -d my_source.yml
+EOF
+      
+      def execute!
+        ensure_data_is_present!
+        Request.new(resource_path, :body => {resource_type => data } , :sign => true).put.print
+      end
+      
+    end
+  end
+end
+

data/lib/chimps-cli/commands/upload.rb

@@ -0,0 +1,86 @@
+module Chimps
+  module Commands
+
+    class Upload < Chimps::Command
+
+      USAGE  = "usage: chimps upload [OPTIONS] ID_OR_SLUG [PATH|LINK] ..."
+      HELP   = <<EOF
+
+Create, modify, inspect, or submit an upload for the dataset with the
+given ID or slug on Infochimps.
+
+Check the status of the current upload for 'my-dataset'
+
+  $ chimps upload my-dataset
+
+If this returns a "404" then it just means that the upload does not
+exist yet.  You will also see whether the upload has started or not.
+
+Assuming the upload hasn't started, you can add files from your local
+machine or links that you want Infochimps to scrape:
+
+  $ chimps upload my-dataset /path/to/some/data.tsv
+  $ chimps upload my-dataset http://www.cooldata.com/some/great/data.html
+  $ chimps upload my-dataset s3://mybucket/some/open/data.xls
+
+You can also pass more than one local file or link in each call to
+`chimps upload' if you wish.
+
+When you're ready to have Infochimps start processing the upload run
+
+  $ chimps upload --start my-dataset
+
+You can watch your upload's status by running
+
+  $ chimps upload my-dataset
+
+at any time.
+EOF
+      
+      def upload
+        return @upload if @upload
+        raise CLIError.new(self.class::USAGE) if config.argv.size < 2 or config.argv[1].blank?
+        @upload = Chimps::Upload.new(config.argv[1])
+      end
+
+      def objs_to_upload
+        config.argv[2..-1] if config.argv.size > 2
+      end
+
+      def execute!
+        case
+        when config[:create]
+          upload.create         # don't print b/c output is messy
+        when config[:start]
+          upload.start.print
+        when config[:restart]
+          upload.restart.print
+        when config[:destroy]
+          upload.destroy.print
+        when objs_to_upload
+          links = []
+          upload.create if upload.show.error?
+          objs_to_upload.each do |obj|
+            if File.exist?(obj)
+              tok = upload.upload_token
+              upload.upload_file(obj, tok)
+            else
+              validate_link(obj)
+              links << obj
+            end
+            upload.create_links(*links) unless links.empty?
+          end
+        else
+          upload.show.print
+        end
+      end
+
+      def validate_link link
+        scheme = link.split("://").first
+        raise UploadError.new("Infochimps only accepts links with a URI scheme (prefix) of http, https, ftp, or s3 -- you gave #{scheme}") unless ['http', 'https', 's3','ftp'].include?(scheme)
+      end
+      
+    end
+  end
+end
+