datahen 0.14.3 → 0.14.11

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: b00adfb4f357beeae276a130cc5e0ee1d34ddd8bdef4a0374f4b55f89f894460
-  data.tar.gz: 6363d591d93d99addcaaea6d964b2fb07e4d8222f873c3a8f6496f48c97b1483
+  metadata.gz: 678f01c798cd52a29da298f48e65b1139f13c80a1214d2c1328f4a4c49abbc63
+  data.tar.gz: 0f787dc429274bb1cd0521237883615e7f8c83316d36fcb319e542cd0fe22d7c
 SHA512:
-  metadata.gz: 75e9f1b1e5ba61563c3ff9d12071cb76d77382353003776d88817940623e37aeb300df2114260f86a5a18a59e1dc9f74a33c00029e183c0681c616dce109d6c6
-  data.tar.gz: 8e2a07b11d20fe88c93aa0af0abf10ceb6593d5d2db63d9b5c3744c6f8b59754274a3949b6331d7d71a3018f7cf65e1a7d31e2ab6f4dbbcde494b33357dab40a
+  metadata.gz: dcaf8fbd6dcfb04f74b8b65ffdcbc127c3e8f179bf0b6da3f2e55acbd2f8d0e425f09a79287bc3ff1791b3fd20e1503c31af3a3128af6eaaedf43e149f1bc995
+  data.tar.gz: 577dda3f28fe7303c08aa290f2fa82150e6f1e679184fac739f839f9de6200be11df7111eddd79f7db7fcfa3aa03f925933db85bed382a1c5c0211534b384e67

data/lib/datahen/cli.rb

@@ -16,6 +16,7 @@ require 'datahen/cli/parser'
 require 'datahen/cli/seeder'
 require 'datahen/cli/finisher'
 require 'datahen/cli/env_var'
+require 'datahen/cli/account'
 
 
 
@@ -41,5 +42,8 @@ module Datahen
 
     desc "var SUBCOMMAND ...ARGS", "for environment variable related activities"
     subcommand "var", EnvVar
+
+    desc "account SUBCOMMAND ...ARGS", "for account related activities"
+    subcommand "account", Account
   end
 end

data/lib/datahen/cli/account.rb

@@ -0,0 +1,17 @@
+module Datahen
+  class CLI < Thor
+    class Account < Thor
+
+      desc "profile", "displays the account applied profile"
+      long_desc <<-LONGDESC
+        Displays the account applied profile
+      LONGDESC
+      def profile()
+        client = Client::Account.new(options)
+        puts "#{client.profile()}"
+      end
+
+    end
+  end
+
+end

data/lib/datahen/cli/env_var.rb

@@ -2,7 +2,7 @@ module Datahen
   class CLI < Thor
     class EnvVar < Thor
       desc "list", "List environment variables on the account"
-      
+
       long_desc <<-LONGDESC
         List all environment variables on the account.
       LONGDESC
@@ -19,7 +19,7 @@ module Datahen
           <name>: Var name can only consist of alphabets, numbers, underscores. Name must be unique to your account, otherwise it will be overwritten.\x5
           <value>: Value of variable.\x5
           LONGDESC
-      option :secret, type: :boolean, desc: 'Set true to make it decrypt the value. Default: false' 
+      option :secret, type: :boolean, desc: 'Set true to make it decrypt the value. Default: false'
       def set(name, value)
         # puts "options #{options}"
         client = Client::EnvVar.new(options)
@@ -38,10 +38,6 @@ module Datahen
         puts "#{client.unset(name)}"
       end
 
-
-      
-
-
     end
   end
 

data/lib/datahen/cli/job.rb

@@ -10,6 +10,7 @@ module Datahen
       desc "list", "gets a list of jobs"
       option :page, :aliases => :p, type: :numeric, desc: 'Get the next set of records by page.'
       option :per_page, :aliases => :P, type: :numeric, desc: 'Number of records per page. Max 500 per page.'
+      option :status, type: :string, desc: 'Returns jobs with a certain status'
       long_desc <<-LONGDESC
         List scrape jobs.
       LONGDESC

data/lib/datahen/cli/scraper.rb

@@ -1,8 +1,8 @@
 module Datahen
   class CLI < Thor
     class Scraper < Thor
-      desc "list", "List scrapers"
 
+      desc "list", "List scrapers"
       long_desc <<-LONGDESC
         List all scrapers.
       LONGDESC
@@ -29,6 +29,7 @@ module Datahen
       option :cancel_current_job, type: :boolean, desc: 'Set true to cancel currently active job if scheduler starts. Default: false'
       option :schedule, type: :string, desc: 'Set the schedule of the scraper to run. Must be in CRON format.'
       option :timezone, type: :string, desc: "Set the scheduler's timezone. Must be in IANA Timezone format. Defaults to \"America/Toronto\""
+      option :profile, type: :string, desc: 'Set the profiles (comma separated) to apply to the job. Default: default'
       def create(scraper_name, git_repository)
         # puts "options #{options}"
         client = Client::Scraper.new(options)
@@ -51,6 +52,7 @@ module Datahen
       option :cancel_current_job, type: :boolean, desc: 'Set true to cancel currently active job if scheduler starts. Default: false'
       option :schedule, type: :string, desc: 'Set the schedule of the scraper to run. Must be in CRON format.'
       option :timezone, type: :string, desc: "Set the scheduler's timezone. Must be in IANA Timezone format. Defaults to \"America/Toronto\""
+      option :profile, type: :string, desc: 'Set the profiles (comma separated) to apply to the job. Default: default'
       def update(scraper_name)
         client = Client::Scraper.new(options)
         puts "#{client.update(scraper_name, options)}"
@@ -164,15 +166,25 @@ module Datahen
       option :"max-timestamp", type: :string, desc: 'Ending timestamp point in time to query historic stats (inclusive)'
       option :"limit", type: :numeric, desc: 'Limit stats retrieved'
       option :"order", type: :numeric, desc: 'Order stats by timestamp [DESC]'
+      option :live, type: :boolean, desc: 'Get data from the live stats history, not cached stats history.'
+      option :filter, type: :string, desc: 'Filter results on `day` or `hour`, if not specified will return all records.'
       def history(scraper_name)
         client = Client::JobStat.new(options)
         if options[:job]
-          puts "#{client.job_stats_history(options[:job])}"
+          puts "#{client.job_stats_history(options[:job], options)}"
         else
-          puts "#{client.scraper_job_stats_history(scraper_name)}"
+          puts "#{client.scraper_job_stats_history(scraper_name, options)}"
         end
       end
 
+      desc "profile <scraper_name>", "displays the scraper applied profile"
+      long_desc <<-LONGDESC
+        Displays the account applied profile
+      LONGDESC
+      def profile(scraper_name)
+        client = Client::Scraper.new(options)
+        puts "#{client.profile(scraper_name)}"
+      end
 
       desc "job SUBCOMMAND ...ARGS", "manage scrapers jobs"
       subcommand "job", ScraperJob

data/lib/datahen/cli/scraper_job.rb

@@ -26,6 +26,7 @@ module Datahen
       LONGDESC
       option :page, :aliases => :p, type: :numeric, desc: 'Get the next set of records by page.'
       option :per_page, :aliases => :P, type: :numeric, desc: 'Number of records per page. Max 500 per page.'
+      option :status, type: :string, desc: 'Returns jobs with a certain status'
       def list(scraper_name)
         client = Client::ScraperJob.new(options)
         puts "#{client.all(scraper_name)}"
@@ -64,16 +65,17 @@ module Datahen
 
       desc "pause <scraper_name>", "pauses a scraper's current job"
       long_desc <<-LONGDESC
-        pauses a scraper's current job
+        Pauses a scraper's current job
       LONGDESC
       option :job, :aliases => :j, type: :numeric, desc: 'Set a specific job ID'
+      option :force, :aliases => :f, type: :boolean, desc: 'Force a job to be paused from a done or cancelled status'
       def pause(scraper_name)
         if options[:job]
           client = Client::Job.new(options)
-          puts "#{client.pause(options[:job])}"
+          puts "#{client.pause(options[:job], options)}"
         else
           client = Client::ScraperJob.new(options)
-          puts "#{client.pause(scraper_name)}"
+          puts "#{client.pause(scraper_name, options)}"
         end
       end
 
@@ -85,6 +87,7 @@ module Datahen
       option :workers, :aliases => :w, type: :numeric, desc: 'Set how many standard workers to use. Scraper job must be restarted(paused then resumed, or cancelled then resumed) for it to take effect. Default: 1. '
       option :browsers, type: :numeric, desc: 'Set how many browser workers to use. Scraper job must be restarted(paused then resumed, or cancelled then resumed) for it to take effect. Default: 0. '
       option :proxy_type, desc: 'Set the Proxy type. Default: standard'
+      option :profile, type: :string, desc: 'Set the profiles (comma separated) to apply to the job. Default: default'
       option :job, :aliases => :j, type: :numeric, desc: 'Set a specific job ID'
       def update(scraper_name)
         if options[:job]
@@ -96,6 +99,21 @@ module Datahen
         end
       end
 
+      desc "profile <scraper_name>", "displays a scraper's current job applied profile"
+      long_desc <<-LONGDESC
+        Displays a scraper's current job applied profile
+      LONGDESC
+      option :job, :aliases => :j, type: :numeric, desc: 'Set a specific job ID'
+      def profile(scraper_name)
+        if options[:job]
+          client = Client::Job.new(options)
+          puts "#{client.profile(options[:job])}"
+        else
+          client = Client::ScraperJob.new(options)
+          puts "#{client.profile(scraper_name)}"
+        end
+      end
+
       desc "var SUBCOMMAND ...ARGS", "for managing scraper's job variables"
       subcommand "var", ScraperJobVar
 

data/lib/datahen/client.rb

@@ -1,4 +1,5 @@
 require "datahen/client/base"
+require "datahen/client/account"
 require "datahen/client/auth_token"
 require "datahen/client/deploy_key"
 require 'datahen/client/export'
@@ -24,7 +25,6 @@ require "datahen/client/job_var"
 require "datahen/client/scraper_job_var"
 require "datahen/client/job_finisher"
 
-
 module Datahen
   module Client
   end

data/lib/datahen/client/account.rb

@@ -0,0 +1,13 @@
+module Datahen
+  module Client
+    class Account < Datahen::Client::Base
+
+      def profile(opts={})
+        params = @options.merge(opts)
+
+        self.class.get("/profile", params)
+      end
+
+    end
+  end
+end

data/lib/datahen/client/base.rb

@@ -57,6 +57,8 @@ module Datahen
         query[:"max-timestamp"] = opts[:"max-timestamp"] if opts[:"max-timestamp"]
         query[:limit] = opts[:limit] if opts[:limit]
         query[:order] = opts[:order] if opts[:order]
+        query[:filter] = opts[:filter] if opts[:filter]
+        query[:force] = opts[:force] if opts[:force]
 
         if opts[:query]
           if opts[:query].is_a?(Hash)

data/lib/datahen/client/job.rb

@@ -20,6 +20,7 @@ module Datahen
         body[:standard_worker_count] = opts[:workers] if opts[:workers]
         body[:browser_worker_count] = opts[:browsers] if opts[:browsers]
         body[:proxy_type] = opts[:proxy_type] if opts[:proxy_type]
+        body[:profile] = opts[:profile] if opts[:profile]
         params = @options.merge({body: body.to_json})
 
         self.class.put("/jobs/#{job_id}", params)
@@ -64,6 +65,12 @@ module Datahen
         self.class.put("/jobs/#{job_id}/finisher_update", params)
       end
 
+      def profile(job_id, opts={})
+        params = @options.merge(opts)
+
+        self.class.get("/jobs/#{job_id}/profile", params)
+      end
+
     end
 
   end

data/lib/datahen/client/job_stat.rb

@@ -18,12 +18,20 @@ module Datahen
         end
       end
 
-      def job_stats_history(job_id)
-        self.class.get("/jobs/#{job_id}/stats/history", @options)
+      def job_stats_history(job_id, opts={})
+        if opts[:live]
+          self.class.get("/jobs/#{job_id}/stats/history", @options)
+        else
+          self.class.get("/cached/jobs/#{job_id}/stats/history", @options)
+        end
       end
 
-      def scraper_job_stats_history(scraper_name)
-        self.class.get("/scrapers/#{scraper_name}/current_job/stats/history", @options)
+      def scraper_job_stats_history(scraper_name, opts={})
+        if opts[:live]
+          self.class.get("/scrapers/#{scraper_name}/current_job/stats/history", @options)
+        else
+          self.class.get("/cached/scrapers/#{scraper_name}/current_job/stats/history", @options)
+        end
       end
 
     end

data/lib/datahen/client/scraper.rb

@@ -25,6 +25,7 @@ module Datahen
         body[:cancel_current_job] = opts[:cancel_current_job] if opts[:cancel_current_job]
         body[:schedule] = opts[:schedule] if opts[:schedule]
         body[:timezone] = opts[:timezone] if opts[:timezone]
+        body[:profile] = opts[:profile] if opts[:profile]
         params = @options.merge({body: body.to_json})
         self.class.post("/scrapers", params)
       end
@@ -43,6 +44,7 @@ module Datahen
         body[:cancel_current_job] = opts[:cancel_current_job] if opts.has_key?("cancel_current_job") || opts.has_key?(:cancel_current_job)
         body[:schedule] = opts[:schedule] if opts[:schedule]
         body[:timezone] = opts[:timezone] if opts[:timezone]
+        body[:profile] = opts[:profile] if opts[:profile]
         params = @options.merge({body: body.to_json})
 
         self.class.put("/scrapers/#{scraper_name}", params)
@@ -52,6 +54,12 @@ module Datahen
         params = @options.merge(opts)
         self.class.delete("/scrapers/#{scraper_name}", params)
       end
+
+      def profile(scraper_name, opts={})
+        params = @options.merge(opts)
+
+        self.class.get("/scrapers/#{scraper_name}/profile", params)
+      end
     end
   end
 end

data/lib/datahen/client/scraper_job.rb

@@ -29,6 +29,7 @@ module Datahen
         body[:standard_worker_count] = opts[:workers] if opts[:workers]
         body[:browser_worker_count] = opts[:browsers] if opts[:browsers]
         body[:proxy_type] = opts[:proxy_type] if opts[:proxy_type]
+        body[:profile] = opts[:profile] if opts[:profile]
         params = @options.merge({body: body.to_json})
 
         self.class.put("/scrapers/#{scraper_name}/current_job", params)
@@ -48,6 +49,12 @@ module Datahen
         opts[:status] = 'paused'
         update(scraper_name, opts)
       end
+
+      def profile(scraper_name, opts={})
+        params = @options.merge(opts)
+
+        self.class.get("/scrapers/#{scraper_name}/current_job/profile", params)
+      end
     end
   end
 end

data/lib/datahen/version.rb

@@ -1,3 +1,3 @@
 module Datahen
-  VERSION = "0.14.3"
+  VERSION = "0.14.11"
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: datahen
 version: !ruby/object:Gem::Version
-  version: 0.14.3
+  version: 0.14.11
 platform: ruby
 authors:
 - Parama Danoesubroto
 autorequire: 
 bindir: exe
 cert_chain: []
-date: 2020-08-02 00:00:00.000000000 Z
+date: 2020-08-24 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: thor
@@ -189,6 +189,7 @@ files:
 - exe/hen
 - lib/datahen.rb
 - lib/datahen/cli.rb
+- lib/datahen/cli/account.rb
 - lib/datahen/cli/env_var.rb
 - lib/datahen/cli/finisher.rb
 - lib/datahen/cli/global_page.rb
@@ -206,6 +207,7 @@ files:
 - lib/datahen/cli/scraper_var.rb
 - lib/datahen/cli/seeder.rb
 - lib/datahen/client.rb
+- lib/datahen/client/account.rb
 - lib/datahen/client/auth_token.rb
 - lib/datahen/client/backblaze_content.rb
 - lib/datahen/client/base.rb
@@ -264,7 +266,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.1.2
+rubygems_version: 3.0.3
 signing_key: 
 specification_version: 4
 summary: DataHen toolbelt for developers