vore 0.2.5-arm64-darwin → 0.2.8-arm64-darwin

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: bc0dadc7e35ecbe250d467f65c9c24443ed535523d4874476b27628ac680fabf
-  data.tar.gz: 83a30b479e3d23585e8f001dc9879402b97f4a921d22ef0644925a484822168a
+  metadata.gz: cea3bfc2d1ad6b14200a35b52eaa739bb14b2f6d5ace761be3d77a53366f19b2
+  data.tar.gz: '03861e5e9ccdbca488f9397bc59f480f48e6d62e03b4cf40ee652e3cc57745da'
 SHA512:
-  metadata.gz: aa3f33ea776f0b4775f650dacaaa350d9a9c0f4fa05cdf2380c2626f46c30d0cb6440972a6eab472e65ec86de1dc771bd9792d183ef5830a5fddc02716805c73
-  data.tar.gz: 5e3b7fb092abb1d341ac4514ea171affb6df9eace0d6a0183ef477ec3ae3f1c97a33517c05f09720bed51c0219bd2eb819f6653c613e683d45c37e3fec677d25
+  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}")
 
@@ -51,6 +47,7 @@ module Vore
         results[:pages_visited] += 1
 
         html_file = File.read(path).force_encoding("UTF-8")
+        rewritten_html_file = ""
 
         if html_file.empty?
           results[:pages_unprocessed] += 1
@@ -58,7 +55,13 @@ module Vore
           next
         end
 
-        rewritten_html_file = @selma.rewrite(html_file)
+        begin
+          rewritten_html_file = @selma.rewrite(html_file)
+        rescue StandardError => e
+          Vore.logger.warn("Error rewriting #{path}: #{e}")
+          results[:pages_unprocessed] += 1
+          next
+        end
 
         # drops the first 3 parts of the path, which are "tmp", "vore", and the site name
         url_path = path.split(FILE_SEPERATOR)[3..].join("/")
@@ -83,6 +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 = "0.2.5"
+  VERSION = "0.2.8"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: vore
 version: !ruby/object:Gem::Version
-  version: 0.2.5
+  version: 0.2.8
 platform: arm64-darwin
 authors:
 - Garen J. Torikian