curation 1.0 → 1.5

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 0f56e1c572c081875ce6e71608496d64a3d34ef84837c60e15c06e358ab0af44
-  data.tar.gz: e7b3899aee83e21bad48c59b68d15d792cebdf9c84b95c24012d4329a3de08cf
+  metadata.gz: 15ed9203a4164d0cfdfb7e973523008ddfe518ce21016731ae8a1dcddc74b0ab
+  data.tar.gz: e2269fbae2d6a6355f90709409de0fa1dd3e555dfd533b10d00d1073af0ac9ec
 SHA512:
-  metadata.gz: a71487cae498b7f04e028f7e6f71793e08047bddefc28e8e3700167c5bf1734cccdd5972a997cd30b07d9c69a2ebb86c4e744e9c32f63a3f0b7c75ba23bf92ef
-  data.tar.gz: b33659760a649ff74aff8fcb821b28d073d4324ce671f1cb737107d4515043de5916d7315a385dfa0cb2517de177aa39f745f3aaad95149339cb4cd116b8cb52
+  metadata.gz: 72df82702be64fcee4e1e7725ecb3c1f831c036040c2ef445388feb493a43cd78b979ed63e99dd0168399c2bd014a0a00e2f7211f561b0ba88f13d0098daaa68
+  data.tar.gz: dd439269683719541f49204484e7d7fd42cbc0f0320e488d2f12a5a36bc472cd343343c4697d0a84a3f8746a308dad03b5e84e33b03adedd3984de73082e5bbb

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    curation (0.1.0)
+    curation (1.5)
       metainspector
       nokogiri
 

data/lib/curation.rb

@@ -1,4 +1,6 @@
 require "curation/version"
+require "metainspector"
+require "open-uri"
 
 module Curation
   class Error < StandardError; end
@@ -17,22 +19,18 @@ 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
-      if json_ld.any?
-        json_ld.each do |ld|
-          return ld['headline'] if ld.has_key? 'headline'
-        end
-      end
-      metainspector.best_title
+      @title = find_title
     end
 
     def image
       @image = find_image
-      @image = @image.gsub('http://', 'https://')
+      @image = @image.to_s.gsub('http://', 'https://')
       @image
     end
 
@@ -44,8 +42,7 @@ module Curation
           return ld['articleBody'] if ld.has_key? 'articleBody'
         end
       end
-      text = ''
-      h = html.dup
+      h = nokogiri.dup
       BLACKLIST.each do |tag|
         h.css(tag).remove
       end
@@ -58,54 +55,88 @@ module Curation
 
     protected
 
+    def find_title
+      if json_ld.any?
+        json_ld.each do |ld|
+          return ld['headline'] if ld.has_key? 'headline'
+        end
+      end
+      begin
+        [
+          metainspector.best_title,
+          metainspector.title,
+          nokogiri.css('[itemprop="headline"]')&.first&.inner_text,
+          nokogiri.css('title')&.first&.inner_text
+        ].each do |possibility|
+          return possibility unless possibility.blank?
+        end
+      rescue
+        puts 'Curation::Page find_title error'
+      end
+      return ''
+    end
+
     def find_image
       if json_ld.any?
         json_ld.each do |ld|
           if ld.has_key? 'image'
             image_data = ld['image']
             return image_data if image_data.is_a? String
-            return image_data.first if image_data.is_a? Array
+            if image_data.is_a? Array
+              first = image_data.first
+              return first if first.is_a? String
+              return first['url'] if first.is_a? Hash
+            end
             return image_data['url'] if image_data.is_a? Hash
           end
         end
       end
-      metainspector.images.best
-    end
-
-    def html
-      @html ||= Nokogiri::HTML data
-    rescue
-      puts "Nokogiri error"
+      begin
+        [
+          metainspector.images.best,
+          nokogiri.css('[property="og:image"]').first&.attributes['content'].value
+        ].each do |possibility|
+            return possibility unless possibility.blank?
+        end
+      rescue
+        puts 'Curation::Page find_image error'
+      end
+      return ''
     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)
             @json_ld << hash
           end
         rescue
-          puts "JSON LD error"
+          puts 'Curation::Page json_ld error'
         end
       end
       @json_ld
     end
 
-    def data
-      require 'open-uri'
-      URI.open url
+    def html
+      @html ||= URI.open url
     rescue
       puts "Impossible to open #{url}"
     end
 
+    def nokogiri
+      @nokogiri ||= Nokogiri::HTML html
+    rescue
+      puts 'Curation::Page nokogiri error'
+    end
+
     def metainspector
-      @metainspector ||= MetaInspector.new url
+      @metainspector ||= MetaInspector.new url, document: html
     rescue
-      puts "MetaInspector error"
+      puts 'Curation::Page metainspector error'
     end
   end
 end

data/lib/curation/version.rb

@@ -1,3 +1,3 @@
 module Curation
-  VERSION = "1.0"
+  VERSION = "1.5"
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: curation
 version: !ruby/object:Gem::Version
-  version: '1.0'
+  version: '1.5'
 platform: ruby
 authors:
 - Arnaud Levy
 autorequire: 
 bindir: exe
 cert_chain: []
-date: 2020-07-20 00:00:00.000000000 Z
+date: 2020-07-22 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: metainspector