vore 0.2.1-x86_64-darwin → 0.2.3-x86_64-darwin

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 31497cb9cec566256acca79e43bfb5b6fe8791bea446bba60ca895f2be3ab91b
-  data.tar.gz: 2252352ba0f823215117c143d0608431bfd6e0037f9b122f6711c191dc7f99c2
+  metadata.gz: 29d98a8f9581f5effc5c5ec76b1eaa4878a81876350126848fda82e73eba2a12
+  data.tar.gz: e29a8806131f0ce5ef176e048b3e75d1dee90a5bfab0f244a6e1f401e7a749d3
 SHA512:
-  metadata.gz: a0f51ccfc30cb95ff87b6275c033b343ca7c07994f9937fad18a670d8973dd2c0d2db8c40f5849d2e32a372c1e755e97b3489b46532bd9c0317e7bfdc88786e0
-  data.tar.gz: 9517a87119025f30bdb105fa89440990ec260b32db10e497a6775271fec547358a4552edf6033508d27225a12427bd61272b13e0585824802c4490d7187173de
+  metadata.gz: f82525deb3f2555eeca00d562b46ee8816f7dbd85dce3b705e30b1daa4d40569605159300fe6c001e67779c1ebcb837065dd04e71a55f26a1e9fa1abb4d19f8a
+  data.tar.gz: 7d2d633b9977dfd716547b330ecb84a77e989a57aeae029a6270cc7263269038439d5f6663314a53a1d5be05bb050b8ec26ffc486f351377f272caff061366ff

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 = "0.2.3"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: vore
 version: !ruby/object:Gem::Version
-  version: 0.2.1
+  version: 0.2.3
 platform: x86_64-darwin
 authors:
 - Garen J. Torikian