curation 1.1 → 1.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 14637da5bf7b047f9c34b40ff4bf108110dce7408d3c12af9e76222141bdcf65
-  data.tar.gz: 144487e448354476958895db783fdc681d0116ec3eca5e7ede1cd3cf1bd1d13a
+  metadata.gz: 21a638c75deffb70db8694ecb99d94b1b711fba922ecbd0826d3a69597031df6
+  data.tar.gz: 621519940fe6bc44302201f64211eba932d52ed366ebe0c6f11c02688b5ae0ba
 SHA512:
-  metadata.gz: 303babe96daee5e792672e1ece05b49d77c115e393af00ded6902b10b9d4adb51210923c7d6c0e68a359246078618a8d09606ab57dd57fa0882db7e3a474e3e4
-  data.tar.gz: 451a1ad8866adc91850ff9b150847564bb9de43b15432de2dbf06fb2936a3788ed2c0bf4d9e3c9c462e13bdb5c6236bf5dc0c97a6697a8c3ea9c307bf2f35410
+  metadata.gz: bf5f63d05793d71ac052b94dcaf5f4e34943355799e042361ae704ca9423b501ab3e157ae8b6509da642e6fa680768205030a24ddc1eb4e24c31863284289575
+  data.tar.gz: d09140b59ba119e85e9c955a40586bd1ce6410fe392a59b5e4a92e2a1ae80b4608fd458c0b84b24be8ba31267bcee9550432264e6ca4068be47b90db46228ed1

data/lib/curation.rb

@@ -19,8 +19,9 @@ module Curation
       '[style*="display: none;"]', '[style*="display: none"]', '[aria-hidden="true"]'
     ]
 
-    def initialize(url)
+    def initialize(url, html = nil)
       @url = url
+      @html = html
     end
 
     def title
@@ -29,7 +30,8 @@ module Curation
           return ld['headline'] if ld.has_key? 'headline'
         end
       end
-      metainspector.best_title
+      metainspector.best_title unless metainspector.best_title.blank?
+      metainspector.title
     end
 
     def image
@@ -46,7 +48,7 @@ module Curation
           return ld['articleBody'] if ld.has_key? 'articleBody'
         end
       end
-      h = html.dup
+      h = nokogiri.dup
       BLACKLIST.each do |tag|
         h.css(tag).remove
       end
@@ -73,17 +75,11 @@ module Curation
       metainspector.images.best
     end
 
-    def html
-      @html ||= Nokogiri::HTML data
-    rescue
-      puts "Nokogiri error"
-    end
-
     def json_ld
       unless @json_ld
         @json_ld = []
         begin
-          options = html.css('[type="application/ld+json"]')
+          options = nokogiri.css('[type="application/ld+json"]')
           options.each do |option|
             string = option.inner_text
             hash = JSON.parse(string)
@@ -96,14 +92,20 @@ module Curation
       @json_ld
     end
 
-    def data
-      URI.open url
+    def html
+      @html ||= URI.open url
     rescue
       puts "Impossible to open #{url}"
     end
 
+    def nokogiri
+      @nokogiri ||= Nokogiri::HTML html
+    rescue
+      puts "Nokogiri error"
+    end
+
     def metainspector
-      @metainspector ||= MetaInspector.new url
+      @metainspector ||= MetaInspector.new url, document: html
     rescue
       puts "MetaInspector error"
     end

data/lib/curation/version.rb

@@ -1,3 +1,3 @@
 module Curation
-  VERSION = "1.1"
+  VERSION = "1.2"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: curation
 version: !ruby/object:Gem::Version
-  version: '1.1'
+  version: '1.2'
 platform: ruby
 authors:
 - Arnaud Levy