web_stat 0.3.4 → 0.3.9

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 746fe3666bc5a47c315e0499fd23dc84319142a8d850b872e5118553b1e711b6
-  data.tar.gz: d2335c2d0324fc81d5b891e7cf383646f8573ef0147ac4b34672add61ecdb6c6
+  metadata.gz: ff258b9a50e224dfb465ad698871b2dc95dc2aaf028f5e966b0b920ca127f5f6
+  data.tar.gz: 0fe37dff66de85ac77137f95e94ef2e3a2e42ae1931b3b138017481c3cda9734
 SHA512:
-  metadata.gz: d23e332cb9114aec4c7302aafb1ed4ca6fc081b3b71a21586779ec5d67de74a196508c38dd70bb8c1e5af3828ced4bdeda7f7c86ea84336fbd6fc476ba87de31
-  data.tar.gz: 9c4e7955432b59b8f66edf71d814412e6eb7720c81f1c647e9f9420f7592febbcfe349c7ceb7a2b6d371c1b7d1de7d98148f7c015d5cc1f4ef52cf08a84afcbe
+  metadata.gz: c2158946ad78a192e902e6b74b9fd72ecfc001afe2c3b531c2e8cea9d7b37239d280174066329e3f908d843c0173ddc03181bf8af529870326f29bd19bf5ee06
+  data.tar.gz: eea91e8addd528e53989c94385c9d339895d138fea4fa911383b6c8157ade28a9c1c75566f2f40ca40d0f5c3b73267188b3695a573bc155a14150b22e3bebb47

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    web_stat (0.3.4)
+    web_stat (0.3.7)
       bundler (>= 2.0.2)
       cld (>= 0.8.0)
       mechanize (>= 2.7)
@@ -83,7 +83,7 @@ GEM
       nokogiri (>= 1.6.0)
     rubyzip (2.3.0)
     safe_yaml (1.0.5)
-    sanitize (5.2.0)
+    sanitize (5.2.1)
       crass (~> 1.0.2)
       nokogiri (>= 1.8.0)
       nokogumbo (~> 2.0)

data/lib/helpers/web_drive_helper.rb

@@ -10,9 +10,7 @@ module WebStat
         options = Selenium::WebDriver::Chrome::Options.new(args: [
           'headless',
           'no-sandbox',
-          'disable-gpu',
-          'start-maximized',
-          'window-size=1920,1080'
+          'disable-gpu'
           ])
         driver = Selenium::WebDriver.for(:chrome, options: options)
         driver.manage.timeouts.implicit_wait = 10

data/lib/web_stat/fetch.rb

@@ -47,6 +47,9 @@ module WebStat
           break
         end
       end
+      if path.nil? || path.empty? || @nokogiri.at('body').xpath('//img').first
+        path = @nokogiri.at('body').xpath('//img').first.attr('src')
+      end
       if ! path.nil? && path.match(/^\//)
         "#{URI.parse(@url).scheme}://#{URI.parse(@url).host}#{path}"
       else
@@ -57,7 +60,7 @@ module WebStat
     # Get local path to save url
     # @param [String] url
     def save_local_path(url)
-      return nil if url.nil? || url.empty?
+      return nil if url.nil? || ! url.match(%{^http})
       tmp_file = "/tmp/#{Digest::SHA1.hexdigest(url)}"
       agent = Mechanize.new { |_agent| _agent.user_agent = WebStat::Configure.get["user_agent"] }
       image = agent.get(url)
@@ -75,17 +78,25 @@ module WebStat
     # @param [String] url
     # @param [String] body
     def get_url(url)
-      agent = Mechanize.new { |_agent| _agent.user_agent = WebStat::Configure.get["user_agent"] }
+      mech = Mechanize.new { |_mech| _mech.user_agent = WebStat::Configure.get["user_agent"] }
       # Enable to read Robots.txt
-      agent.robots = true
+      mech.robots = true
       begin
-        document = agent.get(url, [], nil, { 'Accept-Language' => 'ja'})
-        if document.class == Mechanize::File
-          body = document.body
+        if mech.agent.robots_disallowed?(url)
+          raise Mechanize::RobotsDisallowedError.new(url)
+        end
+        if WebStat::Configure.get["use_chromedirver"]
+          document = WebStat::WebDriverHelper.get_last_url(url)
+          @status = 200
         else
-          body = document.body.encode('UTF-8', document.encoding)
+          document = mech.get(url, [], nil, { 'Accept-Language' => 'ja'})
+          if document.class == Mechanize::File
+            body = document.body
+          else
+            body = document.body.encode('UTF-8', document.encoding)
+          end
+          @status = document.code
         end
-        @status = document.code
       rescue Mechanize::ResponseCodeError => e
         body = e.page.body
         @status = e.page.code
@@ -96,7 +107,7 @@ module WebStat
     # Get the informations of @url
     # @param [Hash] Specify a dictionary for each language code. example ) {"ja": /***/**.dic, "other": /***/***.dic}
     def stat(userdics: nil)
-      clean_content = content.scrub('').gsub(/[\n\t\r　]/, "").gsub(/\s{2,}/, "\s")
+      clean_content = content.scrub('').gsub(/[\n\t\r　]/, "").gsub(/\s{2,}/, "\s").gsub(URI.regexp, "")
       language_code = CLD.detect_language(clean_content)[:code]
       if userdics && userdics.has_key?(language_code) && File.exists?(userdics[language_code])
         tag = WebStat::Tag.new("#{title} #{content}", userdic: userdics[language_code])

data/lib/web_stat/version.rb

@@ -1,3 +1,3 @@
 module WebStat
-  VERSION = "0.3.4"
+  VERSION = "0.3.9"
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: web_stat
 version: !ruby/object:Gem::Version
-  version: 0.3.4
+  version: 0.3.9
 platform: ruby
 authors:
 - yusuke abe
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2020-06-13 00:00:00.000000000 Z
+date: 2020-06-18 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler