chronicle-etl 0.1.2 → 0.2.2

data/lib/chronicle/etl/cli/jobs.rb

@@ -0,0 +1,110 @@
+require 'pp'
+
+module Chronicle
+  module ETL
+    module CLI
+      # CLI commands for working with ETL jobs
+      class Jobs < SubcommandBase
+        default_task "start"
+        namespace :jobs
+
+        class_option :extractor, aliases: '-e', desc: 'Extractor class (available: stdin, csv, file)', default: 'stdin', banner: 'extractor-name'
+        class_option :'extractor-opts', desc: 'Extractor options', type: :hash, default: {}
+        class_option :transformer, aliases: '-t', desc: 'Transformer class (available: null)', default: 'null', banner: 'transformer-name'
+        class_option :'transformer-opts', desc: 'Transformer options', type: :hash, default: {}
+        class_option :loader, aliases: '-l', desc: 'Loader class (available: stdout, csv, table)', default: 'stdout', banner: 'loader-name'
+        class_option :'loader-opts', desc: 'Loader options', type: :hash, default: {}
+        class_option :job, aliases: '-j', desc: 'Job configuration name (or filename)'
+
+        map run: :start  # Thor doesn't like `run` as a command name
+        desc "run", "Start a job"
+        long_desc <<-LONG_DESC
+          This will run an ETL job. Each job needs three parts:
+
+            1. #{'Extractor'.underline}: pulls data from an external source. By default, this is stdout. Other common options including pulling data from an API or reading JSON from a file.
+
+            2. #{'Transformer'.underline}: transforms data into a new format. If none is specified, we use the `null` transformer which does nothing to the data.
+
+            3. #{'Loader'.underline}: takes that transformed data and loads it externally. This can be an API, flat files, (or by default), stdout.
+
+            If you do not want to use the command line flags, you can also configure a job with a .yml config file. You can either specify the path to this file or use the filename and place the file in ~/.config/chronicle/etl/jobs/NAME.yml and call it with `--job NAME`
+LONG_DESC
+        # Run an ETL job
+        def start
+          runner_options = build_runner_options(options)
+          runner = Chronicle::ETL::Runner.new(runner_options)
+          runner.run!
+        end
+
+        desc "create", "Create a job"
+        # Create an ETL job
+        def create
+          runner_options = build_runner_options(options)
+          path = File.join('chronicle', 'etl', 'jobs', options[:job])
+          Chronicle::ETL::Config.write(path, runner_options)
+        end
+
+        desc "show", "Show details about a job"
+        # Show an ETL job
+        def show
+          runner_options = build_runner_options(options)
+          pp runner_options
+        end
+
+        desc "list", "List all available jobs"
+        # List available ETL jobs
+        def list
+          jobs = Chronicle::ETL::Config.jobs
+
+          job_details = jobs.map do |job|
+            r = Chronicle::ETL::Config.load("chronicle/etl/jobs/#{job}.yml")
+
+            extractor = r[:extractor][:name] if r[:extractor]
+            transformer = r[:transformer][:name] if r[:transformer]
+            loader = r[:loader][:name] if r[:loader]
+
+            [job, extractor, transformer, loader]
+          end
+
+          headers = ['name', 'extractor', 'transformer', 'loader'].map{|h| h.upcase.bold }
+
+          table = TTY::Table.new(headers, job_details)
+          puts table.render(indent: 0, padding: [0, 2])
+        end
+
+        private
+
+        # Create runner options by reading config file and then overwriting with flag options
+        def build_runner_options options
+          flag_options = process_flag_options(options)
+          job_options = load_job(options[:job])
+          flag_options.merge(job_options)
+        end
+
+        def load_job job
+          yml_config = Chronicle::ETL::Config.load("chronicle/etl/jobs/#{job}.yml")
+          # FIXME: use better trick to depely symbolize keys
+          JSON.parse(yml_config.to_json, symbolize_names: true)
+        end
+
+        # Takes flag options and turns them into a runner config
+        def process_flag_options options
+          {
+            extractor: {
+              name: options[:extractor],
+              options: options[:'extractor-opts']
+            },
+            transformer: {
+              name: options[:transformer],
+              options: options[:'transformer-opts']
+            },
+            loader: {
+              name: options[:loader],
+              options: options[:'loader-opts']
+            }
+          }
+        end
+      end
+    end
+  end
+end

data/lib/chronicle/etl/cli/main.rb

@@ -0,0 +1,83 @@
+require 'thor'
+require 'chronicle/etl'
+require 'colorize'
+
+require 'chronicle/etl/cli/subcommand_base'
+require 'chronicle/etl/cli/connectors'
+require 'chronicle/etl/cli/jobs'
+
+module Chronicle
+  module ETL
+    module CLI
+      # Main entrypoint for CLI app
+      class Main < Thor
+        class_option "verbose", type: :boolean, default: false
+        default_task "jobs"
+
+        desc 'connectors:COMMAND', 'Connectors available for ETL jobs', hide: true
+        subcommand 'connectors', Connectors
+
+        desc 'jobs:COMMAND', 'Configure and run jobs', hide: true
+        subcommand 'jobs', Jobs
+
+        # Entrypoint for the CLI
+        def self.start(given_args = ARGV, config = {})
+          if given_args.none?
+            abort "No command entered or job specified. To see commands, run `chronicle-etl help`".red
+          end
+
+          # take a subcommand:command and splits them so Thor knows how to hand off to the subcommand class
+          if given_args.any? && given_args[0].include?(':')
+            commands = given_args.shift.split(':')
+            given_args = given_args.unshift(commands).flatten
+          end
+
+          super(given_args, config)
+        end
+
+        # Displays help options for chronicle-etl
+        def help(meth = nil, subcommand = false)
+          if meth && !respond_to?(meth)
+            klass, task = Thor::Util.find_class_and_task_by_namespace("#{meth}:#{meth}")
+            klass.start(['-h', task].compact, shell: shell)
+          else
+            shell.say "ABOUT".bold
+            shell.say "  #{'chronicle-etl'.italic} is a utility tool for #{'extracting'.underline}, #{'transforming'.underline}, and #{'loading'.underline} personal data."
+            shell.say
+            shell.say "USAGE".bold
+            shell.say "  $ chronicle-etl COMMAND"
+            shell.say
+            shell.say "EXAMPLES".bold
+            shell.say "  Show available connectors:".italic.light_black
+            shell.say "  $ chronicle-etl connectors:list"
+            shell.say
+            shell.say "  Run a simple job:".italic.light_black
+            shell.say "  $ chronicle-etl jobs:start --extractor stdin --transformer null --loader stdout"
+            shell.say
+            shell.say "  Show full job options:".italic.light_black
+            shell.say "  $ chronicle-etl jobs help start"
+
+            list = []
+
+            Thor::Util.thor_classes_in(Chronicle::ETL::CLI).each do |thor_class|
+              list += thor_class.printable_tasks(false)
+            end
+            list.sort! { |a, b| a[0] <=> b[0] }
+            list.unshift ["help", "# This help menu"]
+
+            shell.say
+            shell.say 'ALL COMMANDS'.bold
+            shell.print_table(list, indent: 2, truncate: true)
+            shell.say
+            shell.say "VERSION".bold
+            shell.say "  #{Chronicle::ETL::VERSION}"
+            shell.say
+            shell.say "FULL DOCUMENTATION".bold
+            shell.say "  https://github.com/chronicle-app/chronicle-etl".blue
+            shell.say
+          end
+        end
+      end
+    end
+  end
+end

data/lib/chronicle/etl/cli/subcommand_base.rb

@@ -0,0 +1,37 @@
+module Chronicle
+  module ETL
+    module CLI
+      # Base class for CLI subcommands. Overrides Thor methods so we can use command:subcommand syntax
+      class SubcommandBase < Thor
+        # Print usage instructions for a subcommand
+        def self.help(shell, subcommand = false)
+          list = printable_commands(true, subcommand)
+          Thor::Util.thor_classes_in(self).each do |klass|
+            list += klass.printable_commands(false)
+          end
+          list.sort! { |a, b| a[0] <=> b[0] }
+
+          shell.say "COMMANDS".bold
+          shell.print_table(list, indent: 2, truncate: true)
+          shell.say
+          class_options_help(shell)
+        end
+
+        # Show docs with command:subcommand pattern.
+        # For `help` command, don't use colon
+        def self.banner(command, namespace = nil, subcommand = false)
+          if command.name == 'help'
+            "#{subcommand_prefix} #{command.usage}"
+          else
+            "#{subcommand_prefix}:#{command.usage}"
+          end
+        end
+
+        # Use subcommand classname to derive display name for subcommand
+        def self.subcommand_prefix
+          self.name.gsub(%r{.*::}, '').gsub(%r{^[A-Z]}) { |match| match[0].downcase }.gsub(%r{[A-Z]}) { |match| "-#{match[0].downcase}" }
+        end
+      end
+    end
+  end
+end

data/lib/chronicle/etl/config.rb

@@ -0,0 +1,32 @@
+require 'runcom'
+
+module Chronicle
+  module ETL
+    # Utility methods to read, write, and access config files
+    module Config
+      # Loads a yml config file
+      def self.load(path)
+        config = Runcom::Config.new(path)
+        # FIXME: hack to deeply symbolize keys
+        JSON.parse(config.to_h.to_json, symbolize_names: true)
+      end
+
+      # Writes a hash as a yml config file
+      def self.write(path, data)
+        config = Runcom::Config.new(path)
+        filename = config.all[0].to_s + '.yml'
+        File.open(filename, 'w') do |f|
+          f << data.to_yaml
+        end
+      end
+
+      # Returns all jobs available in ~/.config/chronicle/etl/jobs/*.yml
+      def self.jobs
+        job_directory = Runcom::Config.new('chronicle/etl/jobs').current
+        Dir.glob(File.join(job_directory, "*.yml")).map do |filename|
+          File.basename(filename, ".*")
+        end
+      end
+    end
+  end
+end

data/lib/chronicle/etl/exceptions.rb

@@ -0,0 +1,17 @@
+module Chronicle
+  module ETL
+    class Error < StandardError; end;
+
+    class ConnectorNotAvailableError < Error
+      def initialize(message, provider: nil, name: nil)
+        super(message)
+        @provider = provider
+        @name = name
+      end
+      attr_reader :name, :provider
+    end
+
+    class ProviderNotAvailableError < ConnectorNotAvailableError; end
+    class ProviderConnectorNotAvailableError < ConnectorNotAvailableError; end
+  end
+end

data/lib/chronicle/etl/extractors/{csv.rb → csv_extractor.rb}

@@ -1,5 +1,5 @@
 require 'csv'
-class Chronicle::Etl::Extractors::Csv < Chronicle::Etl::Extractors::Extractor
+class Chronicle::ETL::CsvExtractor < Chronicle::ETL::Extractor
   DEFAULT_OPTIONS = {
     headers: true,
     filename: $stdin
@@ -18,7 +18,7 @@ class Chronicle::Etl::Extractors::Csv < Chronicle::Etl::Extractors::Extractor
   end
 
   def results_count
-    CSV.read(@options[:filename], { headers: @options[:headers] }).count if read_from_file?
+    CSV.read(@options[:filename], headers: @options[:headers]).count if read_from_file?
   end
 
   private
@@ -33,7 +33,7 @@ class Chronicle::Etl::Extractors::Csv < Chronicle::Etl::Extractors::Extractor
     }
 
     stream = read_from_file? ? File.open(@options[:filename]) : @options[:filename]
-    CSV.new(stream, csv_options)
+    CSV.new(stream, **csv_options)
   end
 
   def read_from_file?

data/lib/chronicle/etl/extractors/extractor.rb

@@ -1,20 +1,31 @@
+require 'chronicle/etl'
+
 module Chronicle
-  module Etl
-    module Extractors
-      class Extractor
-        def initialize(options = {})
-          @options = options.transform_keys!(&:to_sym) 
-        end
+  module ETL
+    # Abstract class representing an Extractor for an ETL job
+    class Extractor
+      extend Chronicle::ETL::Catalog
 
-        def extract
-          raise NotImplementedError
-        end
+      # Construct a new instance of this extractor. Options are passed in from a Runner
+      # == Paramters:
+      # options::
+      #   Options for configuring this Extractor
+      def initialize(options = {})
+        @options = options.transform_keys!(&:to_sym)
+      end
 
-        def results_count; end
+      # Entrypoint for this Extractor. Called by a Runner. Expects a series of records to be yielded
+      def extract
+        raise NotImplementedError
       end
+
+      # An optional method to calculate how many records there are to extract. Used primarily for
+      # building the progress bar
+      def results_count; end
     end
   end
 end
 
-require_relative 'stdin'
-require_relative 'csv'
+require_relative 'csv_extractor'
+require_relative 'file_extractor'
+require_relative 'stdin_extractor'

data/lib/chronicle/etl/extractors/file_extractor.rb

@@ -0,0 +1,52 @@
+require 'pathname'
+
+module Chronicle
+  module ETL
+    class FileExtractor < Chronicle::ETL::Extractor
+      def extract
+        if file?
+          extract_file do |data, metadata|
+            yield(data, metadata)
+          end
+        elsif directory?
+          extract_from_directory do |data, metadata|
+            yield(data, metadata)
+          end
+        end
+      end
+
+      def results_count
+        if file?
+          return 1
+        else
+          search_pattern = File.join(@options[:filename], '**/*.eml')
+          Dir.glob(search_pattern).count
+        end
+      end
+
+      private
+
+      def extract_from_directory
+        search_pattern = File.join(@options[:filename], '**/*.eml')
+        filenames = Dir.glob(search_pattern)
+        filenames.each do |filename|
+          file = File.open(filename)
+          yield(file.read, {filename: file})
+        end
+      end
+
+      def extract_file
+        file = File.open(@options[:filename])
+        yield(file.read, {filename: @options[:filename]})
+      end
+
+      def directory?
+        Pathname.new(@options[:filename]).directory?
+      end
+
+      def file?
+        Pathname.new(@options[:filename]).file?
+      end
+    end
+  end
+end

data/lib/chronicle/etl/extractors/stdin_extractor.rb

@@ -0,0 +1,11 @@
+module Chronicle
+  module ETL
+    class StdinExtractor < Chronicle::ETL::Extractor
+      def extract
+        $stdin.read.each_line do |line|
+          yield line
+        end
+      end
+    end
+  end
+end

data/lib/chronicle/etl/loaders/csv_loader.rb

@@ -0,0 +1,29 @@
+require 'csv'
+
+module Chronicle
+  module ETL
+    class CsvLoader < Chronicle::ETL::Loader
+      def initialize(options={})
+        super(options)
+        @rows = []
+      end
+
+      def load(result)
+        if (result.is_a? Hash)
+          @rows << result.values
+        else
+          @rows << result
+        end
+      end
+
+      def finish
+        z = $stdout
+        CSV(z) do |csv|
+          @rows.each do |row|
+            csv << row
+          end
+        end
+      end
+    end
+  end
+end