datahen 0.16.2 → 0.17.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 8945724e5d11f40eba22a9ffca7ca7d024b8565dae4ba1d3da9e486eac262575
-  data.tar.gz: 867319a0c6358c593951e6241b24d9c3fd9dcb759ce1287eeb640ade0c23e69a
+  metadata.gz: 82f77763f1977307312821247750463f03c0a7dcb52d78d1ae8ca344265b6e56
+  data.tar.gz: 5f04de4cc0df0b7edcddc2514ef90bb1627e141d8a7b3fcc3ccbc2272b5a629a
 SHA512:
-  metadata.gz: bad8dea41951df061c84934fa63a0594bc8caf22b8665c3d6746bb29ca2821103ecd3814a9a9eba020009e8165390900b1c45a1fd0a7175fb7ea52f7a077fdab
-  data.tar.gz: 74342e01eaa21a590ef998219282fb40342cde4bf8db58617d24583bfa00b967df3f8c38e41c9b92868e90113b7cc5f6346fb2062b446edd3bb4612e053c5da7
+  metadata.gz: cabad1dc36b89878f2a361ab4d4f0c62af932ab3d69b0ca6f929c24809be74d3e82b00fa577215f9019cd243fac286f684cacdcedf8f4cb221400b7c7014c5e3
+  data.tar.gz: b86c0f71a4c555b2e8dce19509d93b888c8dacd72ea0e57db623b3ad1ddb9f3a0ac5bafaac4fdcaf28673c3a77e2c2da4a4aaf0d00fa04465dda641ada86eb52

data/lib/datahen/cli/scraper.rb

@@ -33,6 +33,7 @@ module Datahen
       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'
       option :max_page_size, type: :numeric, desc: 'Set a value to set max page size when fetching a page. Set a value grather than 0 to set it as limit, 0 means any size. Default: 0'
+      option :enable_global_cache, type: :boolean, desc: 'Set true to enable page cache. Default: false'
       def create(scraper_name, git_repository)
         # puts "options #{options}"
         client = Client::Scraper.new(options)
@@ -59,6 +60,7 @@ module Datahen
       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'
       option :max_page_size, type: :numeric, desc: 'Set a value to set max page size when fetching a page. Set a value grather than 0 to set it as limit, 0 means any size. Default: 0'
+      option :enable_global_cache, type: :boolean, desc: 'Set true to enable page cache. Default: false'
       def update(scraper_name)
         client = Client::Scraper.new(options)
         puts "#{client.update(scraper_name, options)}"

data/lib/datahen/cli/scraper_job.rb

@@ -105,6 +105,7 @@ module Datahen
       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'
       option :max_page_size, type: :numeric, desc: 'Set a value to set max page size when fetching a page. Set a value grather than 0 to set it as limit, 0 means any size. Default: 0'
+      option :enable_global_cache, type: :boolean, desc: 'Set true to enable page cache. Default: false'
       def update(scraper_name)
         if options[:job]
           client = Client::Job.new(options)

data/lib/datahen/client/job.rb

@@ -22,6 +22,7 @@ module Datahen
         body[:proxy_type] = opts[:proxy_type] if opts[:proxy_type]
         body[:profile] = opts[:profile] if opts[:profile]
         body[:max_page_size] = opts[:max_page_size] if opts[:max_page_size]
+        body[:enable_global_cache] = opts[:enable_global_cache] if opts.has_key?("enable_global_cache") || opts.has_key?(:enable_global_cache)
         params = @options.merge({body: body.to_json})
 
         self.class.put("/jobs/#{job_id}", params)

data/lib/datahen/client/job_page.rb

@@ -16,6 +16,7 @@ module Datahen
         body[:priority] = opts[:priority] if opts[:priority]
         body[:vars] = opts[:vars] if opts[:vars]
         body[:max_size] = opts[:max_size] if opts[:max_size]
+        body[:enable_global_cache] = opts[:enable_global_cache] if opts.has_key?("enable_global_cache") || opts.has_key?(:enable_global_cache)
 
         params = @options.merge({body: body.to_json})
 
@@ -38,6 +39,7 @@ module Datahen
         body[:no_redirect] = opts[:no_redirect] if opts[:no_redirect]
         body[:cookie] = opts[:cookie] if opts[:cookie]
         body[:max_size] = opts[:max_size] if opts[:max_size]
+        body[:enable_global_cache] = opts[:enable_global_cache] if opts.has_key?("enable_global_cache") || opts.has_key?(:enable_global_cache)
 
         params = @options.merge({body: body.to_json})
 

data/lib/datahen/client/scraper.rb

@@ -29,6 +29,7 @@ module Datahen
         body[:multiple_jobs] = opts[:multiple_jobs] if opts[:multiple_jobs]
         body[:max_job_count] = opts[:max_job_count] if opts[:max_job_count]
         body[:max_page_size] = opts[:max_page_size] if opts[:max_page_size]
+        body[:enable_global_cache] = opts[:enable_global_cache] if opts.has_key?("enable_global_cache") || opts.has_key?(:enable_global_cache)
         params = @options.merge({body: body.to_json})
         self.class.post("/scrapers", params)
       end
@@ -51,6 +52,7 @@ module Datahen
         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)
         body[:max_page_size] = opts[:max_page_size] if opts.has_key?("max_page_size") || opts.has_key?(:max_page_size)
+        body[:enable_global_cache] = opts[:enable_global_cache] if opts.has_key?("enable_global_cache") || opts.has_key?(:enable_global_cache)
         params = @options.merge({body: body.to_json})
 
         self.class.put("/scrapers/#{scraper_name}", params)

data/lib/datahen/client/scraper_job.rb

@@ -12,6 +12,7 @@ module Datahen
         body[:browser_worker_count] = opts[:browsers] if opts[:browsers]
         body[:proxy_type] = opts[:proxy_type] if opts[:proxy_type]
         body[:max_page_size] = opts[:max_page_size] if opts[:max_page_size]
+        body[:enable_global_cache] = opts[:enable_global_cache] if opts.has_key?("enable_global_cache") || opts.has_key?(:enable_global_cache)
         if opts[:vars]
           if opts[:vars].is_a?(Array)
             body[:vars] = opts[:vars]
@@ -39,6 +40,7 @@ module Datahen
         body[:proxy_type] = opts[:proxy_type] if opts[:proxy_type]
         body[:profile] = opts[:profile] if opts[:profile]
         body[:max_page_size] = opts[:max_page_size] if opts[:max_page_size]
+        body[:enable_global_cache] = opts[:enable_global_cache] if opts.has_key?("enable_global_cache") || opts.has_key?(:enable_global_cache)
         params = @options.merge({body: body.to_json})
 
         self.class.put("/scrapers/#{scraper_name}/current_job", params)

data/lib/datahen/client/scraper_job_page.rb

@@ -16,6 +16,7 @@ module Datahen
         body[:priority] = opts[:priority] if opts[:priority]
         body[:vars] = opts[:vars] if opts[:vars]
         body[:max_size] = opts[:max_size] if opts[:max_size]
+        body[:enable_global_cache] = opts[:enable_global_cache] if opts.has_key?("enable_global_cache") || opts.has_key?(:enable_global_cache)
 
         params = @options.merge({body: body.to_json})
 
@@ -61,6 +62,7 @@ module Datahen
         body[:no_redirect] = opts[:no_redirect] if opts[:no_redirect]
         body[:cookie] = opts[:cookie] if opts[:cookie]
         body[:max_size] = opts[:max_size] if opts[:max_size]
+        body[:enable_global_cache] = opts[:enable_global_cache] if opts.has_key?("enable_global_cache") || opts.has_key?(:enable_global_cache)
 
         params = @options.merge({body: body.to_json})
 

data/lib/datahen/scraper/ruby_parser_executor.rb

@@ -8,7 +8,11 @@ module Datahen
       attr_accessor :refetch_self
       # Reparse self page flag.
       # @return [Boollean]
+      # @note It is stronger than #limbo_self flag.
       attr_accessor :reparse_self
+      # Limbo self page flag.
+      # @return [Boollean]
+      attr_accessor :limbo_self
 
       def initialize(options={})
         @filename = options.fetch(:filename) { raise "Filename is required"}
@@ -31,7 +35,8 @@ module Datahen
           :find_output,
           :find_outputs,
           :refetch,
-          :reparse
+          :reparse,
+          :limbo
         ].freeze
       end
 
@@ -104,7 +109,7 @@ module Datahen
 
       def refetch_page gid
         if save
-          Client::ScraperJobPage.new({gid: gid}).refetch_by_job(self.job_id)
+          Client::JobPage.new({gid: gid}).refetch(self.job_id)
           puts "Refetch page #{gid}"
         else
           puts "Would have refetch page #{gid}"
@@ -122,7 +127,7 @@ module Datahen
 
       def reparse_page gid
         if save
-          Client::ScraperJobPage.new({gid: gid}).reparse_by_job(self.job_id)
+          Client::JobPage.new({gid: gid}).reparse(self.job_id)
           puts "Reparse page #{gid}"
         else
           puts "Would have reparse page #{gid}"
@@ -138,6 +143,24 @@ module Datahen
         reparse_page page_gid
       end
 
+      def limbo_page gid
+        if save
+          Client::JobPage.new({gid: gid}).limbo(self.job_id)
+          puts "Limbo page #{gid}"
+        else
+          puts "Would have limbo page #{gid}"
+        end
+      end
+
+      def limbo page_gid
+        raise ArgumentError.new("page_gid needs to be a String.") unless page_gid.is_a?(String)
+        if page_gid == gid
+          self.limbo_self = true
+          raise Error::SafeTerminateError
+        end
+        limbo_page page_gid
+      end
+
       def eval_parser_script(save=false)
         update_parsing_starting_status
 
@@ -148,6 +171,7 @@ module Datahen
           page = init_page_vars(page)
           self.refetch_self = false
           self.reparse_self = false
+          self.limbo_self = false
 
           begin
             context = isolated_binding({
@@ -178,6 +202,8 @@ module Datahen
             refetch_page gid
           elsif reparse_self
             reparse_page gid
+          elsif limbo_self
+            limbo_page gid
           else
             update_parsing_done_status
           end

data/lib/datahen/version.rb

@@ -1,3 +1,3 @@
 module Datahen
-  VERSION = "0.16.2"
+  VERSION = "0.17.0"
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: datahen
 version: !ruby/object:Gem::Version
-  version: 0.16.2
+  version: 0.17.0
 platform: ruby
 authors:
 - Parama Danoesubroto
 autorequire: 
 bindir: exe
 cert_chain: []
-date: 2021-08-11 00:00:00.000000000 Z
+date: 2021-08-20 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: thor