boilerpipe-ruby 0.4.4 → 0.5.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 835e5122c287082ef39874ee0a3365f08910a7a3f70b76c45dc264e8e1301edc
-  data.tar.gz: edff0860c01277adfc703e453e22b1d24c2db1817cfd4ad6800dfc1a6c0f339f
+  metadata.gz: 7fec2bd11d29c4b5d14f70e10fcac76beb95c61e25fbe5cac15b82e8c64fbf69
+  data.tar.gz: 766ea373235462c3678cc2487c647d6211fd2fc066626d5c10ab7e4d31f303ad
 SHA512:
-  metadata.gz: 7cbeb85f9dd74c930e9999e850d2abca6fd3e9470971461a978295de640195192f633a71a19a20b96824fb5becec2e7841e7f3e257b83b90550c935b6d16ff7b
-  data.tar.gz: 393803c95416f77a467724cd0b451961dcbb9c6e544c6d22a4f71853a041e5f2c32513ffd3f43e9733de43624fce31dad18b9cb7d0c37bd3cd722a6fdf223481
+  metadata.gz: be90614a1c2efa29356e9b3b255a5e5d4374474fd6b711d4ed9ab575c4ab8466a1d6903c23de46276133d1621727dea8525422e49608185c1e6294af4f6e0f54
+  data.tar.gz: 5b368e59ced5b794b8e2033b632de67a09bf43e3c45070e96a66ce695bddb5966130e37d9862179a9c989d5166645e182bd2973c179cb8f369c7c4942e238f30

data/CHANGELOG.md

@@ -1,3 +1,6 @@
+# 0.5.0 / 2021-02-15
+* internal refactoring for clarity
+
 # 0.4.4 / 2021-02-13
 * Do a better job of stripping out script tags
 

data/README.md

@@ -1,6 +1,6 @@
 # Boilerpipe
 
-[![CircleCI](https://circleci.com/gh/gregors/boilerpipe-ruby/tree/master.svg?style=shield)](https://circleci.com/gh/gregors/boilerpipe-ruby/tree/master)
+[![CircleCI](https://circleci.com/gh/gregors/boilerpipe-ruby/tree/main.svg?style=shield)](https://circleci.com/gh/gregors/boilerpipe-ruby/tree/main)
 [![Gem Version](https://badge.fury.io/rb/boilerpipe-ruby.svg)](https://badge.fury.io/rb/boilerpipe-ruby)
 
 A pure ruby implemenation of the boilerpipe algorithm.

data/lib/boilerpipe/filters/expand_title_to_content_filter.rb

@@ -8,33 +8,23 @@ module Boilerpipe::Filters
     def self.process(doc)
       tbs = doc.text_blocks
 
-      #     slower and more ruby-like
-      #     comeback and let's do some benchmarking
-      #     titles = tbs.select{ |tb| tb.has_label?(:TITLE) }
-      #     title = tbs.index(titles.last)
-      #     content_start = tbs.find_index(&:is_content?)
+      title = tbs.select{ |tb| tb.has_label?(:TITLE) }.last
+      title_idx = tbs.index(title)
 
-      i = 0
-      title = nil
-      content_start = nil
+      content_start = tbs.find_index(&:is_content?)
 
-      tbs.each do |tb|
-        title = i if content_start.nil? && tb.has_label?(:TITLE)
-        content_start = i if content_start.nil? && tb.is_content?
-        i += 1
-      end
+      return doc if no_title_with_subsequent_content?(content_start, title_idx)
 
-      return doc if no_title_with_subsequent_content?(content_start, title)
-
-      tbs.slice(title...content_start).each do |tb|
-        tb.content = true if tb.has_label?(:MIGHT_BE_CONTENT)
-      end
+      tbs.slice(title_idx...content_start)
+        .select{ |tb| tb.has_label?(:MIGHT_BE_CONTENT) }
+        .each{ |tb| tb.content = true }
 
       doc
     end
 
-    def self.no_title_with_subsequent_content?(content_start, title)
-      title.nil? || content_start.nil? || content_start <= title
+    def self.no_title_with_subsequent_content?(content_start, title_idx)
+      # title has to start before content
+      title_idx.nil? || content_start.nil? || title_idx >= content_start
     end
   end
 end

data/lib/boilerpipe/sax/html_content_handler.rb

@@ -17,7 +17,7 @@ module Boilerpipe::SAX
       @flush = false
       @block_tag_level = -1
 
-      @in_body = 0
+      @in_body_tag = 0
       @in_anchor_tag = 0
       @in_ignorable_element = 0
       @in_anchor_text = false
@@ -92,9 +92,15 @@ module Boilerpipe::SAX
       @label_stacks.pop
     end
 
+    def not_in_body_tag?
+      @in_body_tag == 0
+    end
+
     def flush_block
       @flush = false
-      if @in_body == 0
+
+      # set title
+      if not_in_body_tag?
         @title = @token_buffer.strip if :TITLE == @last_start_tag
         clear_buffers
         return
@@ -205,12 +211,12 @@ module Boilerpipe::SAX
       @in_ignorable_element -= 1
     end
 
-    def increase_in_body!
-      @in_body += 1
+    def enter_body_tag!
+      @in_body_tag += 1
     end
 
-    def decrease_in_body!
-      @in_body -= 1
+    def exit_body_tag!
+      @in_body_tag -= 1
     end
 
     def in_ignorable_element?

data/lib/boilerpipe/sax/tag_actions/body.rb

@@ -4,13 +4,13 @@ module Boilerpipe::SAX::TagActions
   class Body
     def start(handler, name, attrs)
       handler.flush_block
-      handler.increase_in_body!
+      handler.enter_body_tag!
       false
     end
 
     def end_tag(handler, name)
       handler.flush_block
-      handler.decrease_in_body!
+      handler.exit_body_tag!
       false
     end
 

data/lib/boilerpipe/version.rb

@@ -1,3 +1,3 @@
 module Boilerpipe
-  VERSION = '0.4.4'
+  VERSION = '0.5.0'
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: boilerpipe-ruby
 version: !ruby/object:Gem::Version
-  version: 0.4.4
+  version: 0.5.0
 platform: ruby
 authors:
 - Gregory Ostermayr
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2021-02-13 00:00:00.000000000 Z
+date: 2021-02-15 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler