virastar 0.0.4 → 0.0.5

data/TODO

@@ -1,10 +1 @@
-- destories /n if the line ends with ""
-- do not destroy urls dots and colons
-/https?://([-\w\.]+)+(:\d+)?(/([\w/_\.]*(\?\S+)?)?)?/
-
-- translate to js
-
-
-DONE:
-- spacing after , : ; causing a lot of problem in this case (,) => (, )
-- do not put space after : in the context of numbers like hour 19:45 => ۱۹:۴۵
+- translate to js

data/lib/virastar/version.rb

@@ -1,3 +1,3 @@
 module Virastar
-  VERSION = "0.0.4"
+  VERSION = "0.0.5"
 end

data/lib/virastar.rb

@@ -29,6 +29,16 @@ module Virastar
 
     def cleanup
       text = @text
+
+      # removing URLS bringing them back at the end of process
+      urls = []
+      i = 0
+      text.gsub!(/https?:\/\/([-\w\.]+)+(:\d+)?(\/([\w\/_\.]*(\?\S+)?)?)?/) do |s|
+        urls[i] = s.dup
+        i += 1
+        "__urls__#{i}__"
+      end
+
       # replace double dash to ndash and triple dash to mdash
       if @fix_dashes
         text.gsub!(/-{3}/,'—')
@@ -89,33 +99,45 @@ module Virastar
       end
       # ----------------------------------------------------------------
 
+      # should fix outside and inside spacing for () [] {}  “” «»
+      if @fix_spacing_for_braces_and_quotes
+        text.gsub!(/[   ‌]*(\()\s*([^)]+?)\s*?(\))[   ‌]*/,' \1\2\3 ')
+        text.gsub!(/[   ‌]*(\[)\s*([^)]+?)\s*?(\])[   ‌]*/,' \1\2\3 ')
+        text.gsub!(/[   ‌]*(\{)\s*([^)]+?)\s*?(\})[   ‌]*/,' \1\2\3 ')
+        text.gsub!(/[   ‌]*(“)\s*([^)]+?)\s*?(”)[   ‌]*/,' \1\2\3 ')
+        text.gsub!(/[   ‌]*(«)\s*([^)]+?)\s*?(»)[   ‌]*/,' \1\2\3 ')
+      end
+
       # : ; , . ! ? and their persian equivalents should have one space after and no space before
       if @fix_spacing_for_braces_and_quotes
-        text.gsub!(/[ ‌	]*([:;,؛،.؟!]{1})[ ‌	]*/, '\1 ')
+        text.gsub!(/[ ‌  ]*([:;,؛،.؟!]{1})[ ‌  ]*/, '\1 ')
         # do not put space after colon that separates time parts
         text.gsub!(/([۰-۹]+):\s+([۰-۹]+)/, '\1:\2')
       end
-      
-      
 
-      # should fix spacing for () [] {}  “” «»
+      # should fix inside spacing for () [] {}  “” «»
       if @fix_spacing_for_braces_and_quotes
-        text.gsub!(/\s*(\()\s*([^)]+?)\s*?(\))\s*/,' \1\2\3 ')
-        text.gsub!(/\s*(\[)\s*([^)]+?)\s*?(\])\s*/,' \1\2\3 ')
-        text.gsub!(/\s*(\{)\s*([^)]+?)\s*?(\})\s*/,' \1\2\3 ')
-        text.gsub!(/\s*(“)\s*([^)]+?)\s*?(”)\s*/,' \1\2\3 ')
-        text.gsub!(/\s*(«)\s*([^)]+?)\s*?(»)\s*/,' \1\2\3 ')
+        text.gsub!(/(\()\s*([^)]+?)\s*?(\))/,'\1\2\3')
+        text.gsub!(/(\[)\s*([^)]+?)\s*?(\])/,'\1\2\3')
+        text.gsub!(/(\{)\s*([^)]+?)\s*?(\})/,'\1\2\3')
+        text.gsub!(/(“)\s*([^)]+?)\s*?(”)/,'\1\2\3')
+        text.gsub!(/(«)\s*([^)]+?)\s*?(»)/,'\1\2\3')
       end
 
       # should replace more than one space with just a single one
       if @cleanup_spacing
         text.gsub!(/[ ]+/,' ')
-        text.gsub!(/([\n]+)[ 	‌]*/,'\1')
+        text.gsub!(/([\n]+)[   ‌]*/,'\1')
       end
 
       # remove spaces, tabs, and new lines from the beginning and enf of file
       text.strip! if @cleanup_begin_and_end
 
+      # bringing back urls
+      text.gsub!(/__urls__\d+__/) do |s|
+        urls[s.split("__").last.to_i - 1]
+      end
+
       text
     end
 

data/spec/virastar_spec.rb

@@ -193,16 +193,24 @@ describe Virastar do
     test.persian_cleanup.should  == result
   end
 
-  it "should not destroy URLs"
-   # do
-   #    test = "http://virastar.heroku.com"
-   #    result = "http://virastar.heroku.com"
-   #    test.persian_cleanup.should  == result
-   #end
+  it "should not destroy URLs" do
+      test = "http://virastar.heroku.com"
+      result = "http://virastar.heroku.com"
+      test2 = "http://virastar.heroku.com\nhttp://balatarin.com"
+      result2 = "http://virastar.heroku.com\nhttp://balatarin.com"
+      test.persian_cleanup.should  == result
+      test2.persian_cleanup.should  == result2
+   end
 
   it "should not replace line breaks when the line ends with quotes" do
-    test = 'استفاده از "گيومه های فارسي"\nساده است'
-    result = 'استفاده از «گیومه‌های فارسی» \nساده است'
+    test = "salam \"khoobi\" \n chetori"
+    result = "salam «khoobi» \nchetori"
+    test.persian_cleanup.should  == result
+  end
+  
+  it "should not put space after quotes, {}, () or [] if there's ,.; just after that" do
+    test = "«This», {this}, (this), [this] or {this}. sometimes (this)."
+    result = "«This»، {this}، (this)، [this] or {this}. sometimes (this)."
     test.persian_cleanup.should  == result
   end
 

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: virastar
 version: !ruby/object:Gem::Version 
-  hash: 23
+  hash: 21
   prerelease: 
   segments: 
   - 0
   - 0
-  - 4
-  version: 0.0.4
+  - 5
+  version: 0.0.5
 platform: ruby
 authors: 
 - Allen A. Bargi
@@ -15,7 +15,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2011-01-22 00:00:00 +01:00
+date: 2011-01-23 00:00:00 +01:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency