anystyle-parser 0.5.1 → 0.5.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: f5e9afa705ff5d976d3b2ea79ad875779c102d62
-  data.tar.gz: e20422c80e4be4c2effe81adba4f11675b70ebe0
+  metadata.gz: c5471f4465579c6aec53ffee37f8e337075934fd
+  data.tar.gz: e32a68bbbe093abd3c44df90df5524325afc9f7e
 SHA512:
-  metadata.gz: c0af42d7c7d3a4f9216d20d3b71db2d481c454e105d03994e4e8d3ecc038fb9c533e81b650ab35e94f1415bd031d1f67c4b2634a3307658f2c001349b7742cd9
-  data.tar.gz: 7270eebab80be4539e0655eaefd0cf390dd7ff58c74c67ffb7ca2634ed626057cbb5fdc415e47e6f0c53038d9607f7aa6080cedfea5869be4dead98021c05e26
+  metadata.gz: cf520db7b3e31bfe1bd02b03b3d8a951a6288e6d10055ceeccf0b515d3f45e1a83abce911cb698839169c75cf9cac8d2e4169ea54d5db900b25b134b7caffd39
+  data.tar.gz: ebc4173949674553f396dbba6631fae165493f47f748288edad5ccf4fdee33b4e0a2fe7b6a54c5fc0c95e79f8bbbf991a03f62dff202fb6ec949a1098ec1ab24

data/HISTORY.md

@@ -1,6 +1,7 @@
-0.5.0 / 2014-03-13
+0.5.2 / 2014-03-13
 ==================
 * Add XML output
+* Improve ULR normalizer
 
 0.4.4 & 0.4.5 / 2014-03-10
 ==========================

data/lib/anystyle/parser/normalizer.rb

@@ -312,8 +312,7 @@ module Anystyle
         url, *dangling = hash[:url]
         unmatched(:url, hash, dangling) unless dangling.empty?
 
-        url.gsub!(/^\s+|[,\s]+$/, '')
-        hash[:isbn] = isbn
+        hash[:url] = url[/([a-z]+:\/\/)?\w+\.\w+[\w\.\/%-]+/i] || url
         hash
       end
 

data/lib/anystyle/parser/version.rb

@@ -1,5 +1,5 @@
 module Anystyle
   module Parser
-    VERSION = '0.5.1'.freeze
+    VERSION = '0.5.2'.freeze
   end
 end

data/spec/anystyle/parser/normalizer_spec.rb

@@ -58,6 +58,18 @@ module Anystyle
         end
       end
 
+      describe 'URL extraction' do
+        it 'recognizes full URLs' do
+          n.normalize_url(:url => 'Available at: https://www.example.org/x.pdf').should == { :url => 'https://www.example.org/x.pdf' }
+          n.normalize_url(:url => 'Available at: https://www.example.org/x.pdf [Retrieved today]').should == { :url => 'https://www.example.org/x.pdf' }
+        end
+
+        it 'tries to detect URLs without protocol' do
+          n.normalize_url(:url => 'Available at: www.example.org/x.pdf').should == { :url => 'www.example.org/x.pdf' }
+          n.normalize_url(:url => 'Available at: example.org/x.pdf [Retrieved today]').should == { :url => 'example.org/x.pdf' }
+        end
+      end
+
       describe 'date extraction' do
         it 'extracts month and year from a string like "(July 2009)"' do
           h = Normalizer.instance.normalize_date(:date => '(July 2009)')

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: anystyle-parser
 version: !ruby/object:Gem::Version
-  version: 0.5.1
+  version: 0.5.2
 platform: ruby
 authors:
 - Sylvester Keil