web_stat 0.2.3 → 0.2.4

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 90030fe0082c3bd528c2a2dcec6d6539df7a42ca024c9710e110d42af56d0d49
-  data.tar.gz: 7ad6b5178e4892229623a5f5d89368ec4dd1e899ef501422bc3b3ff38c207154
+  metadata.gz: 7d09f9fede0272da84e8174294f32872a95a39059e9af66002bcb54761110cf9
+  data.tar.gz: 2156038d5b220e2d9b5175a03bb8ff9030c779408a5b94982e449a37fca60c12
 SHA512:
-  metadata.gz: 5d2c7ac403071588b6f1e5e1afb62f7032c55736a23057d7bb4e2147d15935224f51b2ad74dc987c8f5e84166a2baae6565f207c32f3b3cdb88147183b29fcb0
-  data.tar.gz: d4774d7f3bb27c06c893d90fb1be43a52e3f0d9ed528854e3dadcdf49e122263abcfbca4a872ee30375c3a931de7b29678aa59b9a1656c9ccdd5ea9a803f2b09
+  metadata.gz: 31e2770f42e1763e625001042212ef50afaa5ead1473763fe51b9edc0ecceee4df365457531ac43cc81cbd60b77a20e8109882feba35e425d79799c9cc505375
+  data.tar.gz: 5960afd25f49e1fcea5b66c8fe45aadbb8740e3b77a97fba3f32a3e855e42224310224f86936d32df6da05210184dedeb49989c562c87ad1fbc728d7a183a5e7

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    web_stat (0.2.3)
+    web_stat (0.2.4)
       bundler (>= 2.0.2)
       cld (>= 0.8.0)
       final_redirect_url (>= 0.1.0)

data/bin/fetch_as_html

@@ -0,0 +1,6 @@
+#!/usr/bin/env ruby
+APP_PATH = File.expand_path('../', __dir__)
+require "bundler/setup"
+require "#{APP_PATH}/lib/web_stat.rb"
+p "[ERROR] Not enough argments. example) fetch_as_html https://newsdict.jp" if ARGV[0].nil?
+p WebStat.stat_by_url(ARGV[0])

data/lib/web_stat/fetch.rb

@@ -72,13 +72,13 @@ module WebStat
     
     # Get the informations of @url
     def stat
-      content.gsub!(/(\n|\t|\s|　)/, "")
+      clean_content = content.scrub('').gsub(/[\s　]/, "")
       tag = WebStat::Tag.new(content, userdic: WebStat::Configure.get["userdic"])
       {
         title: title,
         site_name: site_name,
-        content: content,
-        language_code: CLD.detect_language(content)[:code],
+        content: clean_content,
+        language_code: CLD.detect_language(clean_content)[: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 = "0.2.3"
+  VERSION = "0.2.4"
 end

data/spec/web_stat/fetch_spec.rb

@@ -97,8 +97,8 @@ RSpec.describe WebStat::Fetch do
   
   it "valid url" do
     web_stat_fetch_web_class = WebStat::FetchAsWeb.new("https://newsdict.blog/content/images/size/w100/2019/03/facebook-3.jpg")
-    #expect(web_stat_fetch_web_class.url_valid?("http://status.aws.amazon.com/#cloudfront_12345")).to be true
-    #expect(web_stat_fetch_web_class.url_valid?("https://findy-code.io?h=NWsZey5UgJ51u&t=omikuji-22")).to be true
+    expect(web_stat_fetch_web_class.url_valid?("http://status.aws.amazon.com/#cloudfront_12345")).to be true
+    expect(web_stat_fetch_web_class.url_valid?("https://findy-code.io?h=NWsZey5UgJ51u&t=omikuji-22")).to be true
     expect(web_stat_fetch_web_class.url_valid?("https://www.meetup.com/pro/docker")).to be true
     expect(web_stat_fetch_web_class.url_valid?("https://gxyt4.app.goo.gl/Mn64U")).to be true
     expect(web_stat_fetch_web_class.url_valid?("https://status.cloud.google.com/incident/cloud-functions/19010")).to be true

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: web_stat
 version: !ruby/object:Gem::Version
-  version: 0.2.3
+  version: 0.2.4
 platform: ruby
 authors:
 - yusuke abe
@@ -181,7 +181,8 @@ dependencies:
 description: Fetch the web pages and stat.
 email:
 - yube@newsdict.jp
-executables: []
+executables:
+- fetch_as_html
 extensions: []
 extra_rdoc_files: []
 files:
@@ -195,6 +196,7 @@ files:
 - LICENSE.txt
 - README.md
 - Rakefile
+- bin/fetch_as_html
 - lib/web_stat.rb
 - lib/web_stat/categorize.rb
 - lib/web_stat/config/web_stat.yml