vore 0.2.8-arm64-darwin → 0.3.0-arm64-darwin

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: cea3bfc2d1ad6b14200a35b52eaa739bb14b2f6d5ace761be3d77a53366f19b2
-  data.tar.gz: '03861e5e9ccdbca488f9397bc59f480f48e6d62e03b4cf40ee652e3cc57745da'
+  metadata.gz: 54a3f9525133d20d29eadd67c83ef0c996c3e85ae6843b3ea239e3cff0d9bea3
+  data.tar.gz: 065af90bf1234459fe430a2c49882d988e4b1367786dc506a7cf571804812339
 SHA512:
-  metadata.gz: b7c448fce4a1c6b26b6bff99fbe91053588f56b86b077567913902b4c3405d805d56e41ae430d302fbe2c8d834b50514a945f7361b8996c992d33355cdd7269a
-  data.tar.gz: aa7a8facb68f637b2642b4ee1fbb7437b4cf78ede8f31c8505e9d8141f45bdb4ef689f80ca498af4eb681cd6d5c6c3cbbdba8059f8890ad9fcb37eac5a480095
+  metadata.gz: 540ceeac482a7b9274161a5b64c7d5d0fd570dd3756934ae8a6dcaa760869b81c3f56d89b7c959f7a18c66fffe94c9b215efff8e648d5580aaeaa823e10405e3
+  data.tar.gz: f1d56bc655ac4e42e720837d00df756d497f55ceca844edc4fd7ceb887736458afb02532b90a1fe0f95a7e8b1fe3e551edd3ea196fa8d995c3bd3a5f42db7f32

data/lib/vore/configuration.rb

@@ -5,5 +5,9 @@ module Vole
     DEFAULT_SANITIZATION_CONFIG = Selma::Sanitizer::Config::RELAXED.dup.merge({
       allow_doctype: false,
     })
+
+    DEFAULT_OPTIONS = {
+      delay: 3500,
+    }
   end
 end

data/lib/vore/crawler.rb

@@ -12,7 +12,7 @@ module Vore
 
     # Creates a crawler
     # denylist: Sets a denylist filter, allows a regexp, string or array of either to be matched.
-    def initialize(denylist: /a^/, sanitization_config: Vole::Configuration::DEFAULT_SANITIZATION_CONFIG)
+    def initialize(denylist: /a^/, sanitization_config: Vole::Configuration::DEFAULT_SANITIZATION_CONFIG, options: Vole::Configuration::DEFAULT_OPTIONS)
       @denylist_regexp = Regexp.union(denylist)
 
       @content_extractor = Vole::Handlers::ContentExtractor.new
@@ -20,6 +20,7 @@ module Vore
       ext = PLATFORM.include?("windows") ? ".exe" : ""
       @executable = File.expand_path([__FILE__, "..", "..", "..", "exe", "vore-spider#{ext}"].join(FILE_SEPERATOR))
       @parent_output_dir = "tmp/vore"
+      @options = options
 
       return if File.exist?(@executable)
 
@@ -31,7 +32,7 @@ module Vore
       @output_dir = "#{@parent_output_dir}/#{website.gsub(/[^a-zA-Z0-9]/, "_").squeeze("_")}"
       Vore.logger.info("Vore started crawling #{website}, outputting to #{output_dir}")
 
-      output = run_command(website, @output_dir)
+      output = run_command(website, delay: @options[:delay])
 
       Vore.logger.info("Vore finished crawling #{website}: #{output}")
 
@@ -86,14 +87,14 @@ module Vore
     #   crawl_site(site)
     # end
 
-    def run_command(website, output_dir)
+    def run_command(website, delay: 3500)
       %x(#{@executable} \
         --user-agent #{user_agent} \
-        --delay 3500 \
+        --delay #{delay} \
         --url #{website} \
         download \
         -t \
-        #{output_dir})
+        #{@output_dir})
     end
 
     def user_agent

data/lib/vore/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Vore
-  VERSION = "0.2.8"
+  VERSION = "0.3.0"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: vore
 version: !ruby/object:Gem::Version
-  version: 0.2.8
+  version: 0.3.0
 platform: arm64-darwin
 authors:
 - Garen J. Torikian