Merge pull request #276 from covid19-miyazaki/ticket-275-fix-scraipin…

…g-news

closed #275 fix scraiping news

.github/workflows/news_scraping.yml

@@ -21,10 +21,10 @@ jobs:
     - uses: actions/checkout@v2
     - uses: actions/setup-ruby@v1
       with:
-        ruby-version: '2.5'
+        ruby-version: '2.5.8'
     - name: Scraping News
       env:
-        URL: "https://www.pref.miyazaki.lg.jp/kenko/hoken/kansensho/covid19/hassei.html"
+        URL: "https://www.pref.miyazaki.lg.jp/covid-19/index.html"
         SELENIUM_HOST: localhost
         TZ: Asia/Tokyo
       run: |

.ruby-version

@@ -0,0 +1 @@
+2.5.8

Gemfile.lock

@@ -1,13 +1,11 @@
 GEM
   remote: https://rubygems.org/
   specs:
-    childprocess (2.0.0)
-      rake (< 13.0)
-    rake (12.3.3)
-    rubyzip (1.3.0)
-    selenium-webdriver (3.142.4)
-      childprocess (>= 0.5, < 3.0)
-      rubyzip (~> 1.2, >= 1.2.2)
+    childprocess (3.0.0)
+    rubyzip (2.3.0)
+    selenium-webdriver (3.142.7)
+      childprocess (>= 0.5, < 4.0)
+      rubyzip (>= 1.2.2)
 
 PLATFORMS
   ruby

components/WhatsNew.vue

@@ -16,12 +16,6 @@
           target="_blank"
           rel="noopener"
         >
-          <time
-            class="WhatsNew-list-item-anchor-time px-2"
-            :datetime="formattedDate(item.date)"
-          >
-            {{ item.date }}
-          </time>
           <span class="WhatsNew-list-item-anchor-link">
             <!-- @todo t-i18n化が必要 -->
             {{ item.text }}

data/data.json

@@ -151,7 +151,7 @@
     },
     "main_summary": {
         "attr": "検査実施人数",
-        "value": 1609,
+        "value": 4431,
         "children": [
             {
                 "attr": "陽性患者数",
@@ -269,7 +269,8 @@
     },
     "inspections_summary": {
         "date": "2020/03/27 00:00",
-        "data": {},
+        "data": {
+        },
         "labels": []
     }
-}
+}

data/news.json

@@ -1,14 +1,44 @@
 {
     "newsItems": [
         {
-            "date": "2020年5月29日",
-            "url": "https://www.pref.miyazaki.lg.jp/sogoseisaku/kenko/hoken/20200529_message.html",
-            "text": "新型コロナウイルス感染症に関する知事メッセージ（令和2年5月29日）"
+            "url": "https://www.pref.miyazaki.lg.jp/kansensho-taisaku/covid-19/hassei_list.html",
+            "text": "新型コロナウイルス感染症患者（142～157例目)の発生について（令和2年7月31日）"
         },
         {
-            "date": "2020年5月29日",
-            "url": "https://www.pref.miyazaki.lg.jp/kansensho-taisaku/kenko/hoken/kinkyujitaisengen_covid19.html",
-            "text": "「新しい生活様式」の実践へ"
+            "url": "https://www.pref.miyazaki.lg.jp/fukushihoken/covid-19/jigyosha/20200730202201.html",
+            "text": "「休業要請等」を県下全域に拡大します（8月1日～）"
+        },
+        {
+            "url": "https://www.pref.miyazaki.lg.jp/jinji/covid-19/chiji/20200801_message.html",
+            "text": "【知事メッセージ】知事部局職員の新型コロナウイルス感染について（令和2年8月1日）"
+        },
+        {
+            "url": "https://www.pref.miyazaki.lg.jp/kohosenryaku/covid-19/chiji/20200730.html",
+            "text": "【知事記者会見】「休業要請等」の県下全域への拡大について（令和2年7月30日）"
+        },
+        {
+            "url": "https://www.pref.miyazaki.lg.jp/kansensho-taisaku/covid-19/yobo/hassei.html",
+            "text": "県内の感染状況（警戒レベル）及び注意すべき県外の地域"
+        },
+        {
+            "url": "https://www.pref.miyazaki.lg.jp/kansensho-taisaku/covid-19/yobo/20200726104446.html",
+            "text": "高鍋町内の接待を伴う飲食店の従業員やご利用された方の相談先"
+        },
+        {
+            "url": "https://www.pref.miyazaki.lg.jp/ky-somu/covid-19/torikumi/20200725163141.html",
+            "text": "県立学校における新型コロナウイルス感染症対策の対応について（令和2年7月30日時点）"
+        },
+        {
+            "url": "https://www.pref.miyazaki.lg.jp/choju/covid-19/jigyosha/20200713095950.html",
+            "text": "新型コロナウイルス感染症対応従事者等慰労金（医療・介護・障害福祉分）について"
+        },
+        {
+            "url": "https://www.pref.miyazaki.lg.jp/kansensho-taisaku/covid-19/yobo/20200623app.html",
+            "text": "新型コロナウイルス接触確認アプリ（COCOA）を利用しましょう"
+        },
+        {
+            "url": "https://www.pref.miyazaki.lg.jp/sogoseisaku/kense/sogoseisaku/20200511164006.html",
+            "text": "新型コロナ宮崎復興応援寄附金"
         }
     ]
 }

scrapingSource/scraping.rb

@@ -4,26 +4,41 @@
 
 # スクレイピング
 driver.navigate.to(ENV['URL'])
-if (driver.find_elements(:class => "list_table").size == 0)
+if (driver.find_elements(:class => "info_list").size == 0)
+  puts "no info_list"
   exit
 end
-list_table = driver.find_element(:class => "list_table")
-dates = list_table.find_elements(:class => "date")
-urls = list_table.find_elements(:tag_name => "a")
-texts = list_table.find_elements(:tag_name => "a")
-count = dates.length - 1
+infoList = driver.find_element(:class => "info_list")
+scrapedNews = infoList.find_elements(:tag_name => "a")
+count = scrapedNews.length - 1
 newsItems = []
 for i in 0..count do
-  newsItem = { "date" => dates[i].text, "url" => urls[i].attribute("href"), "text" => texts[i].text }
+  newsItem = { "url" => scrapedNews[i].attribute("href"), "text" => scrapedNews[i].text }
   newsItems.push(newsItem)
 end
 news = { "newsItems" => newsItems }
 puts news
 
 # JSON出力
-news_json = JSON.pretty_generate(news, {:indent => "    "})
+newsJson = JSON.pretty_generate(news, {:indent => "    "})
 File.open("data/news.json", mode = "w") { |f|
-  f.write(news_json)
+  f.write(newsJson)
+}
+
+pcrTable = driver.find_elements(:class => "datatable").last
+rows = pcrTable.find_elements(:tag_name => "tr").last
+total = rows.find_element(:tag_name => "td").find_element(:tag_name => "p")
+
+dataHash = {}
+File.open("data/data.json") do |file|
+  dataHash = JSON.load(file)
+end
+
+dataHash["main_summary"]["value"] = total.text.delete(",").to_i
+
+dataJson = JSON.pretty_generate(dataHash, {:indent => "    "})
+File.open("data/data.json", mode = "w") { |f|
+  f.write(dataJson)
 }
 
 exit