plagiarism2 0.0.6 → 0.0.7

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: a5ae907124afcc6b114cb2fd1a575b7d0a4c5c85
-  data.tar.gz: b6c4a942896b1933ce60a7e3216b8c839bf422a5
+  metadata.gz: 8f4e43615c24c041d05466d83091b775a26bcde1
+  data.tar.gz: 2fc69b5c28c9a5da3726765071fc21323bc80ac3
 SHA512:
-  metadata.gz: bb93fde868664ee036d0fdaa13ee7de53235a372ec41daf5d72109f490b6985094e44afbd8cd189b77693f1a9283d2216010b7c040a5792505874b58e321ffdd
-  data.tar.gz: 93f3cf1b8fed1d7071a80ce4c92383beec0a1b49de3d27f6956c645cc269b8140e4b16d86bfd63384def1ab4ffa8feb9a70b8bf24dfe610f61f120e7b7ea65ef
+  metadata.gz: 5f97108d3ff54b682628ae1328f67726871761cd2121aace55debb7784cdbd5fe21b6d4c7e1661562ed30853bf0afba3397540d3c3e8d62bdd89f2f5dd315841
+  data.tar.gz: e4e2014486db66aecb23159296c1e03578542e2b1ce3b13a8bbbcf883967640335588e476882d2f1fcf4a67fcbd2649eb73e31b94d7233c90039633cde942904

data/README.md

@@ -44,20 +44,27 @@ config.google_key = xxx
 config.google_cx = xx
 ```
 
-After that you can check the unique of content
+###### Plagiarism.unique?
 
 ```ruby
 text = 'Latte user story paradigm affordances experiential innovate venture capital physical computing. Ship it agile actionable insight iterate thought leader pitch deck experiential iterate. Venture capital food-truck quantitative vs. qualitative SpaceTeam convergence agile.'
-Plagiarism.unique? text
+Plagiarism.unique? text # => true
 ```
 
 `Plagiarism.unique?` is true when all strategies is true
 
+###### Plagiarism.match
+
+```ruby
+Plagiarism.match 'ringmd' # => 'www.ring.md'
+```
+
 #### Bash
 
 ```bash
-plagiarism init # generate config
-plagiarism unique -c="ringmd" # check unique
+plagiarism init # generate config, default ~/.plagiarism.yml
+plagiarism unique -c="ringmd" # => false
+plagiarism match -c="ringmd" # => 'www.ring.md'
 ```
 
 For more details, you can use help

data/lib/plagiarism/strategries/bing.rb

@@ -9,9 +9,9 @@ module Plagiarism
           Typhoeus.get(URL, params: params.merge('$format' => :json, 'Query' => "'#{content}'"), userpwd: ":#{Config.bing_key}")
         end
 
-        def iterate(response)
+        def iterate(response, action = :all?)
           results = JSON.parse(response)['d']['results'] rescue []
-          results.all? do |r|
+          results.send(action) do |r|
             uri = URI.parse URI::encode(r['Url'])
             yield uri
           end

data/lib/plagiarism/strategries/duck.rb

@@ -9,10 +9,10 @@ module Plagiarism
           Typhoeus.get(URL, params: params.merge(q: content))
         end
 
-        def iterate(response)
+        def iterate(response, action = :all?)
           doc = Nokogiri::HTML response
-          doc.css('.results_links_deep:not(.result--no-result)').all? do |row|
-            href = row.at_css('.result__a').attributes['href'].value rescue ''
+          doc.css('.results_links_deep:not(.result--no-result) .result__a').send(action) do |row|
+            href = row.attributes['href'].value rescue ''
             uri = URI.parse URI::encode(href)
             yield uri
           end

data/lib/plagiarism/strategries/engine.rb

@@ -10,7 +10,7 @@ module Plagiarism
           raise
         end
 
-        def iterate(r)
+        def iterate(r, a = :all?)
           raise
         end
 
@@ -51,7 +51,7 @@ module Plagiarism
       end
 
       def retrieve_link(response)
-        self.class.iterate(response) { |uri| uri.host !~ self.class.whitelists_regex and return uri.to_s }
+        self.class.iterate(response, :find) { |uri| uri.host !~ self.class.whitelists_regex and return uri.to_s }
       end
 
     end

data/lib/plagiarism/strategries/free_google.rb

@@ -10,9 +10,9 @@ module Plagiarism
           Typhoeus.get URL, params: params.merge(v: VERSION, q: content, rsz: :large)
         end
 
-        def iterate(response)
+        def iterate(response, action = :all?)
           results = JSON.parse(response)['responseData']['results'] rescue []
-          results.all? do |r|
+          results.send(action) do |r|
             uri = URI.parse URI::encode(r['unescapedUrl'])
             yield uri
           end

data/lib/plagiarism/strategries/google.rb

@@ -16,9 +16,9 @@ module Plagiarism
           )
         end
 
-        def iterate(response)
+        def iterate(response, action = :all?)
           results = JSON.parse(response)['items'] || []
-          results.all? do |r|
+          results.send(action) do |r|
             uri = URI.parse URI::encode(r['link'])
             yield uri
           end

data/lib/plagiarism/strategries/yahoo.rb

@@ -6,13 +6,13 @@ module Plagiarism
       class << self
 
         def fetch(content, params)
-          Typhoeus.get(URL, params: params.merge(p: content))
+          Typhoeus.get(URL, params: params.merge(p: content), headers: { 'User-Agent' => 'Mozilla/5.0 (Windows NT 6.1) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/41.0.2228.0 Safari/537.36' })
         end
 
-        def iterate(response)
+        def iterate(response, action = :all?)
           doc = Nokogiri::HTML response
-          doc.css('.searchCenterMiddle li').all? do |row|
-            href = row.at_css('.compTitle a').attributes['href'].value rescue ''
+          doc.css('.searchCenterMiddle li .compTitle a').send(action) do |row|
+            href = row.attributes['href'].value
             uri = URI.parse URI::encode(href)
             yield uri
           end

data/lib/plagiarism/version.rb

@@ -1,3 +1,3 @@
 module Plagiarism
-  VERSION = "0.0.6"
+  VERSION = "0.0.7"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: plagiarism2
 version: !ruby/object:Gem::Version
-  version: 0.0.6
+  version: 0.0.7
 platform: ruby
 authors:
 - MQuy