vore 0.2.1-x86_64-windows → 0.2.2-x86_64-windows

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 3c6f05422a96c467db934bcc38dfe52e031a8a4515780c55ebb1f78993825e1b
-  data.tar.gz: d26f5e75aa905b9ab9e61dd93dedff583739ef1e509a0e8ad999b8f69872d494
+  metadata.gz: 1f43312e386621e74f76efadca6c5e74f19f62fc5d259f75963d2d39f642f504
+  data.tar.gz: 6e189804c155718fe5b739282f2130f344da01325acab55fa1900019d97b935b
 SHA512:
-  metadata.gz: e9cf15323bc49343f834eca443d9231e709e1e9690f9fa0fc89820508e8c356556049b1ff6292fe6b4578fd8e4e91f9e38d05f158dd9d52f8196159906019bed
-  data.tar.gz: 02ac5c1fed91bc8b4136fbe119eb4c195a6facac16b431224785fc6ad812ffc8180b40dbbfee11cdba4213e99316d9d7de7dffff3dba95daccde734348a0c8e6
+  metadata.gz: 9c6fb7dd1cd7da0730ca5caac2ac0d4beea60c88374b721c7181fab02f56a34dae880b3c4ee4c11da287eca26ae766030ff471f8ffaf29c9211bc5b55e32c155
+  data.tar.gz: 76ebda58d854ff7f286f0307a32136dbc6b25e9eabc095ed993ac233428365438d741458aa0adf1786841bc391240691bed46576a441c5f030291fb723628669

data/lib/vore/crawler.rb

@@ -31,6 +31,7 @@ module Vore
 
       output = %x(#{@executable} \
         --user-agent #{user_agent} \
+        --delay 3000
         --url #{website} \
         download \
         -t \
@@ -38,12 +39,25 @@ module Vore
 
       Vore.logger.info("Vore finished crawling #{website}: #{output}")
 
+      results = {
+        pages_visited: 0,
+        pages_unprocessed: 0,
+        unprocessed_pages: [],
+      }
+
       Dir.glob(File.join(output_dir, "**", "*")).each do |path|
         next unless File.file?(path)
 
         html_file = File.read(path).force_encoding("UTF-8")
         rewritten_html_file = @selma.rewrite(html_file)
 
+        results[:pages_visited] += 1
+        if rewritten_html_file.empty?
+          results[:pages_unprocessed] += 1
+          results[:unprocessed_pages] << path
+          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("/")
 
@@ -58,6 +72,8 @@ module Vore
       ensure
         File.delete(path) if File.file?(path)
       end
+
+      results
     end
 
     # def crawl(site, block)

data/lib/vore/handlers/content_extractor.rb

@@ -19,7 +19,13 @@ module Vole
       end
 
       def handle_element(element)
-        if element.tag_name == "pre" || element.tag_name == "code" || element.tag_name == "script" || element.tag_name == "form"
+        if element.tag_name == "pre" ||
+            element.tag_name == "code" ||
+            element.tag_name == "form" ||
+            element.tag_name == "style" ||
+            element.tag_name == "noscript" ||
+            element.tag_name == "script" ||
+            element.tag_name == "svg"
           element.remove
         elsif element.tag_name == "title"
           @within_title = true

data/lib/vore/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Vore
-  VERSION = "0.2.1"
+  VERSION = "0.2.2"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: vore
 version: !ruby/object:Gem::Version
-  version: 0.2.1
+  version: 0.2.2
 platform: x86_64-windows
 authors:
 - Garen J. Torikian