vore 0.2.6-arm64-linux → 0.3.0-arm64-linux

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 72fc70a334c49c0cd3d49fa18a0754e593024555276c128503f6d2897c3bb25a
-  data.tar.gz: 47d4b811f806876617702d1d599848eaa3948f4ebae715a686d05919a37a9f1c
+  metadata.gz: 7c3033ea973b09cb40e841e5be9403126d3aa9acc1c82bcd6ef718086b03b397
+  data.tar.gz: e3fb4483dfa4b3cc0406c5e99ed623cade66f05b50204121a5325307b74a8abc
 SHA512:
-  metadata.gz: 5b6d5a89270fdf1b4630cc09ea1e7e0e9622b3af61a5965ba48718a2b6b2706d5984f59ffa510a8b192910604e7a1ff79f3b373466f3181d3f69ebd714bad3e3
-  data.tar.gz: 9906bcb9a83a4ea7c12de3be523fa3b534a6e0faeb2bf4dcc586a3e097c4efc5fd511d091641a902b4b050be6cff65e25809fb453298b587074aeda8ae3f932d
+  metadata.gz: 2cde527795fe34275c0a8ad93218d38674b01197eae21337d898809eb7da05d38fbdeaa405082cfd4d4e1ba2184703190f337544aa979f640830a0172e5ee732
+  data.tar.gz: 5a867187798343e9ba6736937ef54f9feee35537421652969d4a3558c538d183ca422a25fc1b9d4f909e7e99e9a113262f257915c0f9f000e51afe553149177d

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: arm64-linux
 authors:
 - Garen J. Torikian