htmlclipping 0.1.1 → 0.1.2

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.
@@ -123,11 +123,24 @@ class HtmlClipping
 		end
 
 		def execute
-			@excerpt =~ %r{(.*)(<strong>.*?</strong>\S*)(.*)}m
-			@text_before = $1
-			@strong_text = $2
-			@text_after = $3
-			get_elts.join( ' ' )
+			choices = []
+			@excerpt.split( %r{(<strong>.*?</strong>\S*)} ).each do |part|
+				if @text_before.nil?
+					@text_before = part
+				elsif @strong_text.nil?
+					@strong_text = part
+				elsif @text_after.nil?
+					@text_after = part
+					choices << get_elts.join( ' ' )
+					@text_before, @strong_text, @text_after = @text_after, nil, nil
+				end
+			end
+			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
 
 		def get_elts
@@ -163,11 +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
@@ -218,7 +231,7 @@ class HtmlClipping
 			}
 			fixed
 		end
-		
+
 		def fix_bracket_plus( part )
 			inside_attribute = false
 			between_tags = true
@@ -232,7 +245,9 @@ class HtmlClipping
 						last_attribute_quote = delimiter
 					end
 				end
-				between_tags = !between_tags if %w( < > ).include? delimiter
+				if delimiter == '>' or ( delimiter == '<' and $' !~ /^!/ )
+					between_tags = !between_tags
+				end
 				fixed << pre_delimiter
 				fixed << delimiter
 			}

data/lib/htmlclipping.rb~

@@ -33,7 +33,7 @@
 require 'iconv'
 
 class HtmlClipping
-	Version = '0.1.0'
+	Version = '0.1.1'
 	
 	# html:: The HTML of the referring web page.
 	# referred_uri:: The URI that is being referred to.
@@ -123,11 +123,23 @@ class HtmlClipping
 		end
 
 		def execute
-			@excerpt =~ %r{(.*)(<strong>.*?</strong>\S*)(.*)}m
-			@text_before = $1
-			@strong_text = $2
-			@text_after = $3
-			get_elts.join( ' ' )
+			choices = []
+			@excerpt.split( %r{(<strong>.*?</strong>\S*)} ).each do |part|
+				if @text_before.nil?
+					@text_before = part
+				elsif @strong_text.nil?
+					@strong_text = part
+				elsif @text_after.nil?
+					@text_after = part
+					choices << get_elts.join( ' ' )
+					@text_before, @strong_text, @text_after = nil, nil, nil
+				end
+			end
+			unless @text_before.nil?
+				@text_after = ''
+				choices << get_elts.join( ' ' )
+			end
+			choices.sort_by { |choice| choice.split( %r{<br />} ).size }.first
 		end
 
 		def get_elts
@@ -173,6 +185,7 @@ class HtmlClipping
 			substitute_links
 			compact_brs
 			@detokenized.gsub!( /\s+/, ' ' )
+			@detokenized.gsub!( /&([^\s;]*\s)/, '&amp;\1' )
 			@detokenized
 		end
 		
@@ -217,7 +230,7 @@ class HtmlClipping
 			}
 			fixed
 		end
-		
+
 		def fix_bracket_plus( part )
 			inside_attribute = false
 			between_tags = true
@@ -231,7 +244,9 @@ class HtmlClipping
 						last_attribute_quote = delimiter
 					end
 				end
-				between_tags = !between_tags if %w( < > ).include? delimiter
+				if delimiter == '>' or ( delimiter == '<' and $' !~ /^!/ )
+					between_tags = !between_tags
+				end
 				fixed << pre_delimiter
 				fixed << delimiter
 			}

metadata

@@ -3,8 +3,8 @@ rubygems_version: 0.8.6
 specification_version: 1
 name: htmlclipping
 version: !ruby/object:Gem::Version 
-  version: 0.1.1
-date: 2005-06-26
+  version: 0.1.2
+date: 2005-10-01
 summary: HtmlClipping generates excerpts from an HTML page that has a link pointing to a particular URI.
 require_paths: 
   - lib