htmlclipping 0.1.2 → 0.1.3

data/lib/htmlclipping.rb

@@ -33,7 +33,7 @@
 require 'iconv'
 
 class HtmlClipping
-	Version = '0.1.2'
+	Version = '0.1.3'
 	
 	# html:: The HTML of the referring web page.
 	# referred_uri:: The URI that is being referred to.
@@ -41,7 +41,11 @@ class HtmlClipping
 	def initialize( html, referred_uri, excerpt_limit )
 		@contents = html
 		if @contents =~ %r{<meta.*text/html; charset=(.*?)('|")}i
-			@converter = Iconv.new( 'UTF-8', $1 )
+			begin
+				@converter = Iconv.new( 'UTF-8', $1 )
+			rescue Errno::EINVAL
+				# skip it
+			end
 		end
 		@referred_uri = referred_uri
 		@excerpt_limit = excerpt_limit

data/lib/htmlclipping.rb~

@@ -33,7 +33,7 @@
 require 'iconv'
 
 class HtmlClipping
-	Version = '0.1.1'
+	Version = '0.1.2'
 	
 	# html:: The HTML of the referring web page.
 	# referred_uri:: The URI that is being referred to.
@@ -132,13 +132,14 @@ class HtmlClipping
 				elsif @text_after.nil?
 					@text_after = part
 					choices << get_elts.join( ' ' )
-					@text_before, @strong_text, @text_after = nil, nil, nil
+					@text_before, @strong_text, @text_after = @text_after, nil, nil
 				end
 			end
-			unless @text_before.nil?
+			unless @strong_text.nil?
 				@text_after = ''
 				choices << get_elts.join( ' ' )
 			end
+			choices.delete_if { |choice| choice =~ %r{<strong>\s*</strong>} }
 			choices.sort_by { |choice| choice.split( %r{<br />} ).size }.first
 		end
 
@@ -175,11 +176,11 @@ class HtmlClipping
 			@detokenized = HtmlWithFixedAttributes.new( @contents )
 			@detokenized.gsub!( /<!--.*?-->/m, '' )
 			@detokenized.gsub!(
-				%r{</?(h\d|p|blockquote|table|tr|br|div|form|ul|li|center|ol|dl|dd|dt|fieldset|option|select|object|o:p).*?>}i,
+				%r{</?(h\d|p|blockquote|table|tr|th|br|div|form|ul|li|center|ol|dl|dd|dt|fieldset|option|select|object|o:p).*?>}i,
 				' <br /> '
 			)
 			@detokenized.gsub!(
-				%r{</?(acronym|abbr|strong|td|tt|small|em|img|font|span|input|hr|noscript|legend|address).*?>}im, ''
+				%r{</?(acronym|nobr|abbr|strong|td|tt|small|em|img|font|span|input|hr|noscript|legend|address).*?>}im, ''
 			)
 			@detokenized.gsub!( %r{</?(b|i)(\s+.*?)?>}i, '' )
 			substitute_links

metadata

@@ -3,8 +3,8 @@ rubygems_version: 0.8.6
 specification_version: 1
 name: htmlclipping
 version: !ruby/object:Gem::Version 
-  version: 0.1.2
-date: 2005-10-01
+  version: 0.1.3
+date: 2005-10-27
 summary: HtmlClipping generates excerpts from an HTML page that has a link pointing to a particular URI.
 require_paths: 
   - lib