vore 0.2.6-arm64-linux → 0.2.8-arm64-linux

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 72fc70a334c49c0cd3d49fa18a0754e593024555276c128503f6d2897c3bb25a
-  data.tar.gz: 47d4b811f806876617702d1d599848eaa3948f4ebae715a686d05919a37a9f1c
+  metadata.gz: 0f77381da3968a295e7b13edb86877a2d94646c9f9730bcd2e891c1bd885956a
+  data.tar.gz: b392e9cf87a1bf80ec9b03603c6984b55c76a6da569eba836af48c090f865d5b
 SHA512:
-  metadata.gz: 5b6d5a89270fdf1b4630cc09ea1e7e0e9622b3af61a5965ba48718a2b6b2706d5984f59ffa510a8b192910604e7a1ff79f3b373466f3181d3f69ebd714bad3e3
-  data.tar.gz: 9906bcb9a83a4ea7c12de3be523fa3b534a6e0faeb2bf4dcc586a3e097c4efc5fd511d091641a902b4b050be6cff65e25809fb453298b587074aeda8ae3f932d
+  metadata.gz: abb134c68955daa0ef225183f2bf1563a449395cc93a4ff6bf7f8161d1fcde9e315327d3fa50bfe2081f6ab301cc869c0fd231774c5794ad34043c6093e797c1
+  data.tar.gz: 858cdee603239b6e5c7cdbbf3538d5566ca726d08e986376d527d549d8f4bf4de8bccb325758b5449baae37575b7e954bd78c02ab816875589c5273c66b11ea1

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-linux
 authors:
 - Garen J. Torikian