vore 0.2.5-x86_64-windows → 0.2.8-x86_64-windows

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: b244d29c525e65f76e7cffc59c453f26883d3c7c5d7a7ed2248ecaf319f22515
-  data.tar.gz: 9bd4a0446694812585e68cf8c0b750f808979f0732a136d6d15273a84e7afd4b
+  metadata.gz: 85a50437d0557e28ff3eeb155a8e527163aeb5f90fd98f0e5cb8e09e3d81bb6e
+  data.tar.gz: 61d1255a042db43b6e50bc749fccceb452080afd07e1eac04461ef4894fb8027
 SHA512:
-  metadata.gz: 2783cd4297442c2f4aee698adf301b130873d8e38d01bbdea0f32ec59fc70cdd332cd08f4b7bf3fbfd5e796cb003bc332a9e55d199af36bbacc40cac56e746ca
-  data.tar.gz: caea7c9488653fa4ad09eafc82dc6b1584379016a6e45256ba4bf7f83f8ddfe9ba8218144591ad62593f38b8dd6abc32e6d599932ec0a53c41f09efb3c594a7f
+  metadata.gz: d5c8a45adf9b4402d3600ae26728afb89b5e4d24953ad905462ad6b7c2d682a2806ed452673c486000621e5bf0373de6dd85a28ba592cb2e5eb1e86d83eb973f
+  data.tar.gz: 0077fcc1c2173c46be332da7727f6560a1cf8b2351491bc147b0d1111cefc82dc8bf8081e0f2370e2f7432d1905b2419e8bc168294dfef8a4387d609d512dae8

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: x86_64-windows
 authors:
 - Garen J. Torikian