vore 0.2.6-arm64-darwin → 0.2.8-arm64-darwin

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 402d817b4979ae3cb7caee99653fc2ddfccf726b53f49c76dc078d54a9868af3
-  data.tar.gz: 1768208e84f98fbbcb5e8bb683528f6f476c928d79c1e006cb1b6a8bf521cae7
+  metadata.gz: cea3bfc2d1ad6b14200a35b52eaa739bb14b2f6d5ace761be3d77a53366f19b2
+  data.tar.gz: '03861e5e9ccdbca488f9397bc59f480f48e6d62e03b4cf40ee652e3cc57745da'
 SHA512:
-  metadata.gz: '0091d5f55c923e91abfa57b01e7857072196c85e1053568b5799c715c6d188a27075a82c9393e9fdadade34ff02832c98780dcf9e1f70970f45db29a77fbc733'
-  data.tar.gz: 5915151dbdd4d055f5cf6305dada2a0a82d817a8563ddffab52582f6c580903a2f29269b81c3591c407775ab522d66e55c5a901d3afb4640aa0e959c56f88e2c
+  metadata.gz: b7c448fce4a1c6b26b6bff99fbe91053588f56b86b077567913902b4c3405d805d56e41ae430d302fbe2c8d834b50514a945f7361b8996c992d33355cdd7269a
+  data.tar.gz: aa7a8facb68f637b2642b4ee1fbb7437b4cf78ede8f31c8505e9d8141f45bdb4ef689f80ca498af4eb681cd6d5c6c3cbbdba8059f8890ad9fcb37eac5a480095

data/lib/vore/crawler.rb

@@ -8,6 +8,8 @@ module Vore
     PLATFORM = [:cpu, :os].map { |m| Gem::Platform.local.send(m) }.join("-")
     FILE_SEPERATOR = PLATFORM.include?("windows") ? File::ALT_SEPARATOR : File::SEPARATOR
 
+    attr_reader :output_dir
+
     # 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)
@@ -26,16 +28,10 @@ module Vore
     end
 
     def scrape_each_page(website, &block)
-      output_dir = "#{@parent_output_dir}/#{website.gsub(/[^a-zA-Z0-9]/, "_").squeeze("_")}"
+      @output_dir = "#{@parent_output_dir}/#{website.gsub(/[^a-zA-Z0-9]/, "_").squeeze("_")}"
       Vore.logger.info("Vore started crawling #{website}, outputting to #{output_dir}")
 
-      output = %x(#{@executable} \
-        --user-agent #{user_agent} \
-        --delay 3500 \
-        --url #{website} \
-        download \
-        -t \
-        #{output_dir})
+      output = run_command(website, @output_dir)
 
       Vore.logger.info("Vore finished crawling #{website}: #{output}")
 
@@ -54,7 +50,6 @@ module Vore
         rewritten_html_file = ""
 
         if html_file.empty?
-          Vore.logger.warn("HTML file empty: #{path}")
           results[:pages_unprocessed] += 1
           results[:unprocessed_pages] << path
           next
@@ -91,6 +86,16 @@ module Vore
     #   crawl_site(site)
     # end
 
+    def run_command(website, output_dir)
+      %x(#{@executable} \
+        --user-agent #{user_agent} \
+        --delay 3500 \
+        --url #{website} \
+        download \
+        -t \
+        #{output_dir})
+    end
+
     def user_agent
       "'Mozilla/5.0 (compatible; Vore/#{Vore::VERSION}; +https://github.com/gjtorikian/vore)'"
     end

data/lib/vore/handlers/content_extractor.rb

@@ -20,7 +20,6 @@ module Vole
 
       def handle_element(element)
         if element.tag_name == "pre" ||
-            element.tag_name == "code" ||
             element.tag_name == "form" ||
             element.tag_name == "style" ||
             element.tag_name == "noscript" ||

data/lib/vore/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Vore
-  VERSION = "0.2.6"
+  VERSION = "0.2.8"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: vore
 version: !ruby/object:Gem::Version
-  version: 0.2.6
+  version: 0.2.8
 platform: arm64-darwin
 authors:
 - Garen J. Torikian