web_stat 0.2.6 → 0.2.7

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 1f508dae86c4163037aa8ca2dd299be2f98ad01a2b4e61cff8c20b7562cb91e4
-  data.tar.gz: 2a4e70f10396243d5917decf75714302ded5959cebb517fca6d27bc3c45db5eb
+  metadata.gz: 32af8e84a325614bf546a76fdf603e592abbf4201a559857bc0964fe49226cc1
+  data.tar.gz: 8b63075dbd78d2e4b7368cb6a0a4aa2f8ae655cf7f85b526bd413a5fa210fbc0
 SHA512:
-  metadata.gz: c6e4b4066d42003bb4e07ac9bfd4184e73dfea3a3d4e3d7a5b239c227f009a631e01cfb7b5d17c0240453d5f943b9e1f49c846d6d2167dd9ad2122372e709fe3
-  data.tar.gz: 597cef76252fdcc77beefcc5d24e68625c93fa5293bed0d7a40f26847b0db6a0b981c83a8e6f3beafee93778c1ff2bf16899a29c6df1866bf6b929b5d29c2975
+  metadata.gz: 45a0fe16c58679cbea01c56710324a461da3bd5c48caa1ca44cf981b2f5443b6ed06ce1db84577edac968e7651cd47ec998ef331dbf47c7306b243a1190fa76d
+  data.tar.gz: c391ce77bf97673c708821fa9067ba64e5b89fbf67d3490376341931f3b9ee6ecebf4a51e0abe69dcc827f7f35d7d6986af353ea2ab4f1d0945f4de2a7d34b2f

data/Gemfile.lock

@@ -7,7 +7,7 @@ GIT
 PATH
   remote: .
   specs:
-    web_stat (0.2.6)
+    web_stat (0.2.7)
       bundler (>= 2.0.2)
       cld (>= 0.8.0)
       mechanize (>= 2.7)

data/lib/web_stat/fetch.rb

@@ -53,9 +53,13 @@ module WebStat
     def save_local_path(url)
       return nil if url.nil?
       tmp_file = "/tmp/#{Digest::SHA1.hexdigest(url)}"
-      URI.open(original_url(url)) do |remote_file|
-        File.open(tmp_file, "w+b") do |_file|
-          _file.puts(remote_file.read)
+      agent = Mechanize.new { |_agent| _agent.user_agent = WebStat::Configure.get["user_agent"] }
+      image = agent.get(url)
+      File.open(tmp_file, "w+b") do |_file|
+        if image.class == Mechanize::File
+          _file.puts(image.body)
+        else
+          _file.puts(image.body_io.read)
         end
       end
       tmp_file
@@ -68,18 +72,30 @@ module WebStat
       # Enable to read Robots.txt
       agent.robots = true
       document = agent.get(url, [], nil, { 'Accept-Language' => 'ja'})
-      document.body.encode('UTF-8', document.encoding)
+      if document.class == Mechanize::File
+        document.body
+      else
+        document.body.encode('UTF-8', document.encoding)
+      end
     end
     
     # Get the informations of @url
-    def stat
+    # @param [Hash] Specify a dictionary for each language code. example ) {"ja": /***/**.dic, "other": /***/***.dic}
+    def stat(userdics: nil)
       clean_content = content.scrub('').gsub(/[\s　]/, "")
-      tag = WebStat::Tag.new(content, userdic: WebStat::Configure.get["userdic"])
+      language_code = CLD.detect_language(clean_content)[:code]
+      if userdics && userdics.has_key?(language_code)
+        tag = WebStat::Tag.new("#{title} #{content}", userdic: userdics[language_code])
+      elsif userdics && userdics.has_key?("other")
+        tag = WebStat::Tag.new("#{title} #{content}", userdic: userdics["other"])
+      else
+        tag = WebStat::Tag.new("#{title} #{content}", userdic: WebStat::Configure.get["userdic"])
+      end
       {
         title: title,
         site_name: site_name,
         content: clean_content,
-        language_code: CLD.detect_language(clean_content)[:code],
+        language_code: language_code,
         url: @url,
         eyecatch_image_path: save_local_path(eyecatch_image_path),
         tags: tag.nouns

data/lib/web_stat/version.rb

@@ -1,3 +1,3 @@
 module WebStat
-  VERSION = "0.2.6"
+  VERSION = "0.2.7"
 end

data/lib/web_stat.rb

@@ -21,24 +21,26 @@ require "web_stat/fetch/fetch_as_web"
 module WebStat
   class << self
     # Get web page's stat by url
-    def stat_by_web(url)
+    # @param [Hash] Specify a dictionary for each language code. example ) {"ja": /***/**.dic, "other": /***/***.dic}
+    def stat_by_web(url, userdics: nil)
       web_stat = WebStat::FetchAsWeb.new(url)
-      web_stat.stat
+      web_stat.stat(userdics: userdics)
     end
 
     # Get web page's stat by url
     # @param String url
-    def stat_by_url(url)
-      stat_by_web(url)
+    def stat_by_url(url, userdics: nil)
+      stat_by_web(url, userdics: userdics)
     end
     
     # Get web page's stat by html
     # @param String html
     # @param [String] url
-    def stat_by_html(html, url=nil)
+    # @param [Hash] Specify a dictionary for each language code. example ) {"ja": /***/**.dic, "other": /***/***.dic}
+    def stat_by_html(html, url=nil, userdics: nil)
       web_stat = WebStat::FetchAsHtml.new(html)
       web_stat.url = url unless url.nil?
-      web_stat.stat
+      web_stat.stat(userdics: userdics)
     end
   end
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: web_stat
 version: !ruby/object:Gem::Version
-  version: 0.2.6
+  version: 0.2.7
 platform: ruby
 authors:
 - yusuke abe