datahen 0.14.10 → 0.14.17

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 68b2f97e2b71ab8e79edacc9fa0b719c1d4b93bac684631ab084e46f44407795
-  data.tar.gz: 2e4a453502ac9a72c7cbea81b3f36100373adf9894e5c2ca4250cb1fa709c548
+  metadata.gz: a3cf1bf50610ab9ba523ca8f9ae1fdb307831cb56018036076f31353b357edfa
+  data.tar.gz: '0199ea98a2f171675168699adc523932418b9822ff156c71786458c6362b6cdb'
 SHA512:
-  metadata.gz: e73e87d091b590fd1efc4c31d5589b930e0fae458a8b8e5439c47dfae008c1db9186672f49ddcf41e7a0c32e31398b1703b45f1cde81a3a4e397c8f16008a4ea
-  data.tar.gz: 52fd87fe09082f7ebd3a6e96ce0fc699b8a474ceb867a2187f21fb54431ac91f4cffdaf3a0ede51feb2a18fa9085a13a00048ad83a362d63399f2c4b87739c78
+  metadata.gz: 8d66226573dbd9bd3ef795ce021eb9ee202b21cce1d0b79211093bb9abda6f1982a721bf8ae1fbc9bf84aae6e50d1878bf9d389ee67d685e4d1c440fd88cceeb
+  data.tar.gz: 6ea1a7748cf77ae1cb8f750da71df86120f68e671c84e535e8536ee3324ed71026034a042662d081108ff7b72ccb3f0ef3ab3f049bdb45c92fc0e86ba2e10d46

data/lib/datahen/cli/global_page.rb

@@ -15,7 +15,10 @@ module Datahen
 
         if result['available'] == true
           puts "Preview content url: \"#{result['preview_url']}\""
-          `open "#{result['preview_url']}"`
+          begin
+            `open "#{result['preview_url']}"`
+          rescue
+          end
         else
           puts "Content does not exist"
         end

data/lib/datahen/cli/job.rb

@@ -6,7 +6,6 @@ module Datahen
         "#{basename} #{@package_name} #{command.usage}"
       end
 
-
       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.'
@@ -35,8 +34,7 @@ module Datahen
         client = Client::JobStat.new(options)
         puts "#{client.job_current_stats(job_id, options)}"
       end
-
-
+      
     end
   end
 

data/lib/datahen/cli/scraper.rb

@@ -30,6 +30,8 @@ module Datahen
       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'
+      option :multiple_jobs, type: :boolean, desc: 'Set true to enable multiple jobs. Default: false'
+      option :max_job_count, type: :numeric, desc: 'Set a value to set max number of jobs available. Set -1 for unlimited. Default: 3'
       def create(scraper_name, git_repository)
         # puts "options #{options}"
         client = Client::Scraper.new(options)
@@ -53,6 +55,8 @@ module Datahen
       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'
+      option :multiple_jobs, type: :boolean, desc: 'Set true to enable multiple jobs. Default: false'
+      option :max_job_count, type: :numeric, desc: 'Set a value to set max number of jobs available. Set -1 for unlimited. Default: 3'
       def update(scraper_name)
         client = Client::Scraper.new(options)
         puts "#{client.update(scraper_name, options)}"

data/lib/datahen/cli/scraper_export.rb

@@ -36,7 +36,10 @@ module Datahen
 
         if result['signed_url']
           puts "Download url: \"#{result['signed_url']}\""
-          `open "#{result['signed_url']}"`
+          begin
+            `open "#{result['signed_url']}"`
+          rescue
+          end
         else
           puts "Exported file does not exist"
         end

data/lib/datahen/cli/scraper_job.rb

@@ -48,6 +48,21 @@ module Datahen
         end
       end
 
+      desc "delete <scraper_name>", "delete a scraper's current job"
+      long_desc <<-LONGDESC
+        Delete a scraper's current job
+      LONGDESC
+      option :job, :aliases => :j, type: :numeric, desc: 'Set a specific job ID'
+      def delete(scraper_name)
+        if options[:job]
+          client = Client::Job.new(options)
+          puts "#{client.delete(options[:job])}"
+        else
+          client = Client::ScraperJob.new(options)
+          puts "#{client.delete(scraper_name)}"
+        end
+      end
+
       desc "resume <scraper_name>", "resumes a scraper's current job"
       long_desc <<-LONGDESC
         Resumes a scraper's current job
@@ -68,13 +83,14 @@ module Datahen
         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
 

data/lib/datahen/cli/scraper_page.rb

@@ -111,6 +111,7 @@ module Datahen
           puts "Must specify either a --gid, --fetch-fail, --parse-fail or --status"
           return
         end
+
         if options[:job]
           client = Client::JobPage.new(options)
           puts "#{client.refetch(options[:job])}"
@@ -129,26 +130,39 @@ module Datahen
       option :status, type: :string, desc: 'Reparse only pages with a specific status.'
       option :job, :aliases => :j, type: :numeric, desc: 'Set a specific job ID'
       def reparse(scraper_name)
-        begin
-          options[:vars] = JSON.parse(options[:vars]) if options[:vars]
+        if !options.key?(:gid) && !options.key?(:parse_fail) && !options.key?(:status)
+          puts "Must specify either a --gid, --parse-fail or --status"
+          return
+        end
 
-          if !options.key?(:gid) && !options.key?(:parse_fail) && !options.key?(:status)
-            puts "Must specify either a --gid, --parse-fail or --status"
-            return
-          end
+        if options[:job]
+          client = Client::JobPage.new(options)
+          puts "#{client.reparse(options[:job])}"
+        else
+          client = Client::ScraperJobPage.new(options)
+          puts "#{client.reparse(scraper_name)}"
+        end
+      end
 
-          if options[:job]
-            client = Client::JobPage.new(options)
-            puts "#{client.reparse(options[:job])}"
-          else
-            client = Client::ScraperJobPage.new(options)
-            puts "#{client.reparse(scraper_name)}"
-          end
+      desc "limbo <scraper_name>", "Move pages on a scraper's current job to limbo"
+      long_desc <<-LONGDESC
+        Move pages in a scraper's current job to limbo. You need to specify either a --gid or --status.\x5
+      LONGDESC
+      option :gid, :aliases => :g, type: :string, desc: 'Move a specific GID to limbo'
+      option :status, type: :string, desc: 'Move pages with a specific status to limbo.'
+      option :job, :aliases => :j, type: :numeric, desc: 'Set a specific job ID'
+      def limbo(scraper_name)
+        if !options.key?(:gid) && !options.key?(:status)
+          puts "Must specify either a --gid or --status"
+          return
+        end
 
-        rescue JSON::ParserError
-          if options[:vars]
-            puts "Error: #{options[:vars]} on vars is not a valid JSON"
-          end
+        if options[:job]
+          client = Client::JobPage.new(options)
+          puts "#{client.limbo(options[:job])}"
+        else
+          client = Client::ScraperJobPage.new(options)
+          puts "#{client.limbo(scraper_name)}"
         end
       end
 
@@ -224,7 +238,10 @@ module Datahen
 
         if result['available'] == true
           puts "Preview content url: \"#{result['preview_url']}\""
-          `open "#{result['preview_url']}"`
+          begin
+            `open "#{result['preview_url']}"`
+          rescue
+          end
         else
           puts "Content does not exist"
         end
@@ -244,7 +261,10 @@ module Datahen
 
         if result['available'] == true
           puts "Preview failed content url: \"#{result['preview_url']}\""
-          `open "#{result['preview_url']}"`
+          begin
+            `open "#{result['preview_url']}"`
+          rescue
+          end
         else
           puts "Failed Content does not exist"
         end

data/lib/datahen/client/auth_token.rb

@@ -16,7 +16,7 @@ module Datahen
             role: role,
             description: description}
 
-        params = @options.merge({body: body.to_json})
+        params = @options.merge({body: body.to_json}).merge(opts)
         self.class.post("/auth_tokens", params)
       end
 

data/lib/datahen/client/base.rb

@@ -58,6 +58,7 @@ module Datahen
         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

@@ -71,6 +71,11 @@ module Datahen
         self.class.get("/jobs/#{job_id}/profile", params)
       end
 
+      def delete(job_id, opts={})
+        params = @options.merge(opts)
+        self.class.delete("/jobs/#{job_id}", params)
+      end
+
     end
 
   end

data/lib/datahen/client/job_page.rb

@@ -72,6 +72,11 @@ module Datahen
         params = @options.merge(opts)
         self.class.put("/jobs/#{job_id}/pages/refetch", params)
       end
+
+      def limbo(job_id, opts={})
+        params = @options.merge(opts)
+        self.class.put("/jobs/#{job_id}/pages/limbo", params)
+      end
     end
   end
 end

data/lib/datahen/client/scraper.rb

@@ -26,6 +26,8 @@ module Datahen
         body[:schedule] = opts[:schedule] if opts[:schedule]
         body[:timezone] = opts[:timezone] if opts[:timezone]
         body[:profile] = opts[:profile] if opts[:profile]
+        body[:multiple_jobs] = opts[:multiple_jobs] if opts[:multiple_jobs]
+        body[:max_job_count] = opts[:max_job_count] if opts[:max_job_count]
         params = @options.merge({body: body.to_json})
         self.class.post("/scrapers", params)
       end
@@ -45,6 +47,8 @@ module Datahen
         body[:schedule] = opts[:schedule] if opts[:schedule]
         body[:timezone] = opts[:timezone] if opts[:timezone]
         body[:profile] = opts[:profile] if opts[:profile]
+        body[:multiple_jobs] = opts[:multiple_jobs] if opts.has_key?("multiple_jobs") || opts.has_key?(:multiple_jobs)
+        body[:max_job_count] = opts[:max_job_count] if opts.has_key?("max_job_count") || opts.has_key?(:max_job_count)
         params = @options.merge({body: body.to_json})
 
         self.class.put("/scrapers/#{scraper_name}", params)

data/lib/datahen/client/scraper_job.rb

@@ -55,6 +55,11 @@ module Datahen
 
         self.class.get("/scrapers/#{scraper_name}/current_job/profile", params)
       end
+
+      def delete(scraper_name, opts={})
+        params = @options.merge(opts)
+        self.class.delete("/scrapers/#{scraper_name}/current_job", params)
+      end
     end
   end
 end

data/lib/datahen/client/scraper_job_page.rb

@@ -26,7 +26,7 @@ module Datahen
         self.class.put("/scrapers/#{scraper_name}/current_job/pages/refetch", params)
       end
 
-      # Deprecated, please use Datahen::Client::JobVar#refetch instead.
+      # Deprecated, please use Datahen::Client::JobPage#refetch instead.
       #
       # @note This method will be removed at some point in the future.
       def refetch_by_job(job_id, opts={})
@@ -39,6 +39,11 @@ module Datahen
         self.class.put("/scrapers/#{scraper_name}/current_job/pages/reparse", params)
       end
 
+      def limbo(scraper_name, opts={})
+        params = @options.merge(opts)
+        self.class.put("/scrapers/#{scraper_name}/current_job/pages/limbo", params)
+      end
+
       def enqueue(scraper_name, method, url, opts={})
         body = {}
         body[:method] =  method != "" ? method : "GET"

data/lib/datahen/version.rb

@@ -1,3 +1,3 @@
 module Datahen
-  VERSION = "0.14.10"
+  VERSION = "0.14.17"
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: datahen
 version: !ruby/object:Gem::Version
-  version: 0.14.10
+  version: 0.14.17
 platform: ruby
 authors:
 - Parama Danoesubroto
 autorequire: 
 bindir: exe
 cert_chain: []
-date: 2020-08-10 00:00:00.000000000 Z
+date: 2020-10-20 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: thor
@@ -266,7 +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