vore 0.2.1-arm64-darwin → 0.2.3-arm64-darwin

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 119b1cdc06e20e708ceda86505a02361c19bf0de881300222e164bac1bbe6cc6
-  data.tar.gz: 75febe4d2f4cd048722d94a2f265b6239db67f08282a26de70a5c85bc4735933
+  metadata.gz: 1fb9a7a5a0d624b1c5a6829a03adbb7fa3deef8415ce92a28441c73c17261eb3
+  data.tar.gz: b3345789f05d1eda10743a535f086c1ab8d7ffbd3014affe7170ea13e78e71fb
 SHA512:
-  metadata.gz: 364c154a766444e457769925a2c0a4fe890c53fbd9a7726d9896afac6389807806a6c36a1a3e3d27ae5e7c8de837c0081dab4b8397e9f23a8a7655607f5c62d7
-  data.tar.gz: c32f3ba74037034ff6f957cd397a61ad6c8328a2f282196b45e69584c5693b7647e4dd1cad105aa7b20475106ebe9933f80b81585061f42338b073005b7fe217
+  metadata.gz: 9dae767278a15953dcbec4c350d98ca0acdfe0a04940b11d7088acfadb10354227db753783b7132a597dd817acbea234f54c82b7e10af1fb31afdc1d293f6211
+  data.tar.gz: de6c5b45e9e1bc9ceb0520ca1e3ffcc86e4a8a9493fc7b078f8854b59c2a6b518024ee2802c98efff0d8b73281a67d92e4b157891797d64d9683a51e575aeaa2

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: arm64-darwin
 authors:
 - Garen J. Torikian