wovnrb 0.2.23 → 0.2.24

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: ea8678343beb5b165e98756cdae5701487ab3e4f
-  data.tar.gz: bd2976576eb5e64c6cbc7f822ccab117f427765b
+  metadata.gz: 0a7e70b6cdf98b2eefab0f135987d9930f915955
+  data.tar.gz: abbfc8770a7600d87c1ff514ba9fbfc6c4ce5d78
 SHA512:
-  metadata.gz: 372502decc7bb4e4e903f297b34b2d7161fa793f8c6bc075a3176d95d86a4642160f3afb63744ab8c7ebfe0d1380ba7b70d2bd56a799d1a06811e5d826f239a8
-  data.tar.gz: 283eacde612587dedc974226aa35c8320330ecd3a806d5d6718f0e005672c511bd1f3eaa4a7b12819c42a69aa8b69c9605556f377123108cc753107a6ecd10db
+  metadata.gz: e130de085fa3f0873242214df53b7e5798afda355edc300527467b05be107d5605642c264f6807017f2d37056ba2ee78c429ab5049e16f5b9e8bf890ea0c1baf
+  data.tar.gz: a4d0b9e0d39c5c19f8eb83c8c467d99bc6d1c502655c7cacd172049c9196324c626e20b3f1891443cc4f6cec5a5c3cb78227c3276ef17c4db24a9418843d8d2b

data/lib/wovnrb/html_replacers/link_replacer.rb

@@ -1,5 +1,12 @@
 module Wovnrb
   class LinkReplacer < ReplacerBase
+    module FileExtension
+      IMG_FILES = '(?!jp$)jpe?g?|bmp|gif|png|btif|tiff?|psd|djvu?|xif|wbmp|webp|p(n|b|g|p)m|rgb|tga|x(b|p)m|xwd|pic|ico|fh(c|4|5|7)?|xif|f(bs|px|st)'
+      AUDIO_FILES = 'mp(3|2)|m(p?2|3|p?4|pg)a|midi?|kar|rmi|web(m|a)|aif(f?|c)|w(ma|av|ax)|m(ka|3u)|sil|s3m|og(a|g)|uvv?a'
+      VIDEO_FILES = 'm(x|4)u|fl(i|v)|3g(p|2)|jp(gv|g?m)|mp(4v?|g4|(?!$)e?g?)|m(1|2)v|ogv|m(ov|ng)|qt|uvv?(h|m|p|s|v)|dvb|mk(v|3d|s)|f4v|as(x|f)|w(m(v|x)|vx)|xvid'
+      DOC_FILES = 'zip|tar|ez|aw|atom(cat|svc)?|(cc)?xa?ml|cdmi(a|c|d|o|q)?|epub|g(ml|px|xf)|jar|js|ser|class|json(ml)?|do(c|t)m?|xps|pp(a|tx?|s)m?|potm?|sldm|mp(p|t)|bin|dms|lrf|mar|so|dist|distz|m?pkg|bpk|dump|rtf|tfi|pdf|pgp|apk|o(t|d)(b|c|ft?|g|h|i|p|s|t)'
+    end
+
     def initialize(pattern, headers)
       @pattern = pattern
       @headers = headers
@@ -12,9 +19,18 @@ module Wovnrb
         href = node.get_attribute('href')
         next if href =~ /^\s*\{\{.+\}\}\s*$/
         next if href =~ /javascript:void\(/
+        next if is_file?(href)
         new_href = lang.add_lang_code(href, @pattern, @headers)
         node.set_attribute('href', new_href)
       end
     end
+
+    def is_file?(href)
+      img_files = /^(https?:\/\/)?.*(\.(#{FileExtension::IMG_FILES}))((\?|#).*)?$/i
+      audio_files = /^(https?:\/\/)?.*(\.(#{FileExtension::AUDIO_FILES}))((\?|#).*)?$/i
+      video_files = /^(https?:\/\/)?.*(\.(#{FileExtension::VIDEO_FILES}))((\?|#).*)?$/i
+      doc_files = /^(https?:\/\/)?.*(\.(#{FileExtension::DOC_FILES}))((\?|#).*)?$/i
+      href =~ img_files || href =~ audio_files || href =~ video_files || href =~ doc_files
+    end
   end
 end

data/lib/wovnrb/version.rb

@@ -1,3 +1,3 @@
 module Wovnrb
-  VERSION = "0.2.23"
+  VERSION = "0.2.24"
 end

data/test/lib/html_replacers/link_replacer_test.rb

@@ -51,6 +51,111 @@ module Wovnrb
       assert_equal('javascript:void(0);', link)
     end
 
+    def test_replace_link_path
+      replacer = LinkReplacer.new('path', get_header)
+      dom = Wovnrb.get_dom('<a href="/index.html">link text</a>')
+      replacer.replace(dom, Lang.new('en'))
+
+      link = dom.xpath('//a')[0].get_attribute('href')
+      assert_equal('/en/index.html', link)
+    end
+
+    def test_replace_img_link_path
+      replacer = LinkReplacer.new('path', get_header)
+      dom = Wovnrb.get_dom('<a href="http://favy.tips/index.jpg">link text</a>')
+      replacer.replace(dom, Lang.new('ja'))
+
+      link = dom.xpath('//a')[0].get_attribute('href')
+      assert_equal('http://favy.tips/index.jpg', link)
+    end
+
+    def test_replace_img_link_path_with_query_or_hash
+      replacer = LinkReplacer.new('path', get_header)
+      dom = Wovnrb.get_dom('<a href="http://favy.tips/index.jpg?test=1">link text</a>')
+      replacer.replace(dom, Lang.new('ja'))
+
+      link = dom.xpath('//a')[0].get_attribute('href')
+      assert_equal('http://favy.tips/index.jpg?test=1', link)
+
+      dom = Wovnrb.get_dom('<a href="http://favy.tips/index.jpg#test=1">link text</a>')
+      replacer.replace(dom, Lang.new('ja'))
+
+      link = dom.xpath('//a')[0].get_attribute('href')
+      assert_equal('http://favy.tips/index.jpg#test=1', link)
+    end
+
+    def test_replace_audio_link_path
+      replacer = LinkReplacer.new('path', get_header)
+      dom = Wovnrb.get_dom('<a href="/index.mp3">link text</a>')
+      replacer.replace(dom, Lang.new('ja'))
+
+      link = dom.xpath('//a')[0].get_attribute('href')
+      assert_equal('/index.mp3', link)
+    end
+
+    def test_replace_audio_link_path_with_query_or_hash
+      replacer = LinkReplacer.new('path', get_header)
+      dom = Wovnrb.get_dom('<a href="/index.mp3?test=1">link text</a>')
+      replacer.replace(dom, Lang.new('ja'))
+
+      link = dom.xpath('//a')[0].get_attribute('href')
+      assert_equal('/index.mp3?test=1', link)
+
+      dom = Wovnrb.get_dom('<a href="/index.mp3#test=1">link text</a>')
+      replacer.replace(dom, Lang.new('ja'))
+
+      link = dom.xpath('//a')[0].get_attribute('href')
+      assert_equal('/index.mp3#test=1', link)
+    end
+
+    def test_replace_video_link_path
+      replacer = LinkReplacer.new('path', get_header)
+      dom = Wovnrb.get_dom('<a href="/index.mpeg">link text</a>')
+      replacer.replace(dom, Lang.new('ja'))
+
+      link = dom.xpath('//a')[0].get_attribute('href')
+      assert_equal('/index.mpeg', link)
+    end
+
+    def test_replace_video_link_path_with_query_or_hash
+      replacer = LinkReplacer.new('path', get_header)
+      dom = Wovnrb.get_dom('<a href="/index.mp4?test=1">link text</a>')
+      replacer.replace(dom, Lang.new('ja'))
+
+      link = dom.xpath('//a')[0].get_attribute('href')
+      assert_equal('/index.mp4?test=1', link)
+
+      dom = Wovnrb.get_dom('<a href="/index.mp4#test=1">link text</a>')
+      replacer.replace(dom, Lang.new('ja'))
+
+      link = dom.xpath('//a')[0].get_attribute('href')
+      assert_equal('/index.mp4#test=1', link)
+    end
+
+    def test_replace_doc_link_path
+      replacer = LinkReplacer.new('path', get_header)
+      dom = Wovnrb.get_dom('<a href="/index.pptx">link text</a>')
+      replacer.replace(dom, Lang.new('ja'))
+
+      link = dom.xpath('//a')[0].get_attribute('href')
+      assert_equal('/index.pptx', link)
+    end
+
+    def test_replace_doc_link_path_with_query_or_hash
+      replacer = LinkReplacer.new('path', get_header)
+      dom = Wovnrb.get_dom('<a href="/index.pptx?test=1">link text</a>')
+      replacer.replace(dom, Lang.new('ja'))
+
+      link = dom.xpath('//a')[0].get_attribute('href')
+      assert_equal('/index.pptx?test=1', link)
+
+      dom = Wovnrb.get_dom('<a href="/index.pptx#test=1">link text</a>')
+      replacer.replace(dom, Lang.new('ja'))
+
+      link = dom.xpath('//a')[0].get_attribute('href')
+      assert_equal('/index.pptx#test=1', link)
+    end
+
     def test_replace_javascript_link_subdomain
       replacer = LinkReplacer.new('subdomain', get_header)
       dom = Wovnrb.get_dom('<a href="javascript:void(0);">link text</a>')
@@ -82,4 +187,3 @@ module Wovnrb
     end
   end
 end
-

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: wovnrb
 version: !ruby/object:Gem::Version
-  version: 0.2.23
+  version: 0.2.24
 platform: ruby
 authors:
 - Jeff Sandford
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2017-08-29 00:00:00.000000000 Z
+date: 2017-09-21 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: nokogumbo