htmlclipping 0.1.7 → 0.1.8

data/lib/htmlclipping.rb

@@ -33,7 +33,7 @@
 require 'iconv'
 
 class HtmlClipping
-	Version = '0.1.7'
+	Version = '0.1.8'
 	
 	# html:: The HTML of the referring web page.
 	# referred_uri:: The URI that is being referred to.
@@ -70,8 +70,10 @@ class HtmlClipping
 		end
 		excerpt = Detokenizer.new( to_detokenize, @referred_uri ).execute
 		excerpt = Clipper.new( excerpt, @excerpt_limit ).execute
-		excerpt.gsub( /[\200-\377]/ ) { |c| "&#%04d;" % c[0] }
-		convert( excerpt )
+		if excerpt
+			excerpt.gsub( /[\200-\377]/ ) { |c| "&#%04d;" % c[0] }
+			convert( excerpt )
+		end
 	end
 
 	class ArrayOfWordsAndBrTags < Array # :nodoc:

data/lib/htmlclipping.rb~

@@ -33,7 +33,7 @@
 require 'iconv'
 
 class HtmlClipping
-	Version = '0.1.6'
+	Version = '0.1.7'
 	
 	# html:: The HTML of the referring web page.
 	# referred_uri:: The URI that is being referred to.
@@ -57,17 +57,19 @@ class HtmlClipping
 	
 	# Returns the clipping as a string suitable for use as XML text.
 	def to_s
-		if @contents =~ %r{<body[^>]*>(.*)</body>}mi
-			to_detokenize = $1
-		else
-			@contents =~ %r{<body[^>]*>(.*)}mi
-			to_detokenize = $1
+		regexes = [
+			%r{<body[^>]*>(.*)</body>}mi, %r{<body[^>]*>(.*)}mi,
+			%r{</head>(.*)</(body|html)>}mi
+		]
+		to_detokenize = nil
+		until to_detokenize or regexes.empty?
+			regex = regexes.shift
+			if @contents =~ regex
+				to_detokenize = $1
+			end
 		end
-p to_detokenize
 		excerpt = Detokenizer.new( to_detokenize, @referred_uri ).execute
-p 'DETOKENIZED ' + excerpt
 		excerpt = Clipper.new( excerpt, @excerpt_limit ).execute
-p excerpt
 		excerpt.gsub( /[\200-\377]/ ) { |c| "&#%04d;" % c[0] }
 		convert( excerpt )
 	end

metadata

@@ -3,8 +3,8 @@ rubygems_version: 0.8.11
 specification_version: 1
 name: htmlclipping
 version: !ruby/object:Gem::Version 
-  version: 0.1.7
-date: 2006-08-26 00:00:00 -04:00
+  version: 0.1.8
+date: 2006-12-12 00:00:00 -05:00
 summary: HtmlClipping generates excerpts from an HTML page that has a link pointing to a particular URI.
 require_paths: 
 - lib