vore 0.2.6-x86_64-linux → 0.3.0-x86_64-linux

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 53cb93bee6f3736369589c1eaaf7b739ca6f898e814d33d4ff747f4daa2077fc
-  data.tar.gz: 100cf434502ce236ce6f31aa24c110ed43274043f1ea7a678f4cad9905079e7a
+  metadata.gz: 5fa0065651385809a53579488f9985cc3332197d4fa9818508859bb24274a16b
+  data.tar.gz: a0754878a08d651215dd5df40220e3002255ef65fedcb365e0605ff228db27fd
 SHA512:
-  metadata.gz: b66ad8d522524716bc1f5a904b428934ef5006916cad623893301569afcfd7b255c021d1bc3f1205cd757365a165b6b88dfe278e9979aba4d900fc913f074d73
-  data.tar.gz: 9e176c95395e40503fc8a5af9f014a5befc37a63d04b71a0d64549be100700b92abbb439f57be4ab4699a3543ad2ca32203095b83e402774b35e70ae58d6cde2
+  metadata.gz: 48298f9bf6de3e76b443a4ea46fa0adc687e4a39b90fc94ae6a97e0ce0153c6e983e292af0c8d6a53dcb44e94b8ecba94f8adc75fe5ce8cd7857e947ab113795
+  data.tar.gz: 404712a662f24fff36346998e791cfac7c9ac44661005209f0a2a65b38146f0233dfcdc9cfc115c5772340d310d085dbadbca8766201fad810843359ccff1c98

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

@@ -8,9 +8,11 @@ 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)
+    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
@@ -18,6 +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)
 
@@ -26,16 +29,10 @@ 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, delay: @options[:delay])
 
       Vore.logger.info("Vore finished crawling #{website}: #{output}")
 
@@ -54,7 +51,6 @@ 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 +87,16 @@ module Vore
     #   crawl_site(site)
     # end
 
+    def run_command(website, delay: 3500)
+      %x(#{@executable} \
+        --user-agent #{user_agent} \
+        --delay #{delay} \
+        --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 = "0.3.0"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: vore
 version: !ruby/object:Gem::Version
-  version: 0.2.6
+  version: 0.3.0
 platform: x86_64-linux
 authors:
 - Garen J. Torikian