deba 0.10.0 → 0.11.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: d717c720283ce4e2ec8ff902b269c8fc79e15d92
-  data.tar.gz: e589c45ed86c7b6929ffdee81f8228a4197f963a
+  metadata.gz: 64da76c5a1c04969f66f580d20e350c7bca13715
+  data.tar.gz: fa665cf07572193d1a3f6ffcfa976885bf50a9ad
 SHA512:
-  metadata.gz: ff5850362d73ad3fcb1e3ed1708ed2fb5ae35ab6d364f6f9efec83d41a02bff57fc3f52968c253941d07d16ff9e16364247a1d226b6925a2f9ae45d14c226a3c
-  data.tar.gz: e21024b05d064f279ee581f6841bc8e0d63dc4e7733b44151a2e9a0ef9d3aebf68a8103860503ed3c174f92af30c9851bf89f29c4e030ab936495f78bde4d71c
+  metadata.gz: ef43a8c834a219a8b4fa5ad78cc4a59f10b09ec637ca46574a2ef61d22cc879ac9da32fea5fbf904d396b6b95c9c64301c730a818914149d9e9db840776dc215
+  data.tar.gz: 0f06e42c7267719f135a4bbcd0e6d24565979b19c3ade21bf49e622fccb7bec3c7cfe332651dde6fcbb2158064e08f8a560b047a1f9c029a49398711d9678cb1

data/lib/deba/definition_description.rb

@@ -1,13 +1,10 @@
 class Deba::DefinitionDescription
-  attr_reader :segments
-
-  def initialize(segments, line_prefix, last)
+  def initialize(segments, last)
     @segments = segments
-    @line_prefix = line_prefix
     @last = last
   end
 
-  def to_s
-    "#{Deba::Stringifier.new(@segments, @line_prefix).stringify}\n#{"\n" if @last}"
+  def to_a
+    @segments + ["\n#{"\n" if @last}"]
   end
 end

data/lib/deba/definition_term.rb

@@ -1,12 +1,9 @@
 class Deba::DefinitionTerm
-  attr_reader :segments
-
-  def initialize(segments, line_prefix)
+  def initialize(segments)
     @segments = segments
-    @line_prefix = line_prefix
   end
 
-  def to_s
-    "#{Deba::Stringifier.new(@segments, @line_prefix).stringify}:\n"
+  def to_a
+    @segments + [":\n"]
   end
 end

data/lib/deba/document.rb

@@ -1,15 +1,39 @@
 class Deba::Document
-  attr_reader :blocks
+  attr_reader :content
 
-  def initialize
-    @blocks = []
+  def initialize(extractor)
+    @extractor = extractor
+    @content = ""
+
+    start
+  end
+
+  def <<(segment)
+    @segments << segment
+  end
+
+  def break(*args)
+    finish
+    start(*args)
+  end
+
+  def finish
+    return unless present?
+
+    @args.unshift(@segments)
+    block = @block_type.new(*@args).to_a
+    block.unshift("> ") if @extractor.in_blockquote?
+
+    @content << Deba::Stringifier.new(block).stringify
   end
 
-  def <<(block)
-    @blocks << block
+  def start(*args)
+    @segments = []
+    @block_type = args.shift
+    @args = args
   end
 
-  def to_s
-    @blocks.map { |block| block.to_s }.join.chomp("\n")
+  def present?
+    @segments.any? { |segment| segment.is_a?(Deba::Span) && Deba::Utils.present?(segment.to_s) }
   end
 end

data/lib/deba/extractor.rb

@@ -12,12 +12,12 @@ class Deba::Extractor
 
   def extract
     @just_appended_br = false
-    @document = Deba::Document.new
-    @text_run = Deba::TextRunner.new(@document)
+    @in_blockquote = false
+    @document = Deba::Document.new(self)
 
     process(@node)
 
-    @document
+    @document.content.chomp("\n")
   end
 
   def process(node)
@@ -34,7 +34,7 @@ class Deba::Extractor
       if @just_appended_br
         @just_appended_br = false
 
-        @text_run.break(Deba::Paragraph, line_prefix(node))
+        @document.break(Deba::Paragraph)
 
         return
       else
@@ -43,11 +43,11 @@ class Deba::Extractor
     elsif @just_appended_br
       @just_appended_br = false
 
-      @text_run << Deba::Break.new
+      @document << "\n"
     end
 
     if node.text?
-      @text_run << node.inner_text if Deba::Utils.present?(node.inner_text)
+      @document << Deba::Span.new(node.inner_text) if Deba::Utils.present?(node.inner_text)
 
       return
     end
@@ -55,55 +55,68 @@ class Deba::Extractor
     if ENHANCERS.keys.flatten.include?(node_name)
       ENHANCERS.each_pair do |tags, nsf_rep|
         if tags.include?(node_name)
-          @text_run << nsf_rep
+          @document << nsf_rep
           node.children.each { |n| process(n) }
-          @text_run << nsf_rep
+          @document << nsf_rep
         end
       end
 
       return
     end
 
+    if node_name == 'blockquote'
+      @in_blockquote = true
+
+      @document.break(Deba::Paragraph)
+      node.children.each { |n| process(n) }
+      @document.break(Deba::Paragraph)
+
+      @in_blockquote = false
+
+      return
+    end
+
     if node_name == 'li'
       last_item = node.xpath('count(following-sibling::li)').to_i == 0
       index = node.xpath('boolean(ancestor::ol)') ? (node.xpath('count(preceding-sibling::li)').to_i + 1) : nil
-      @text_run.break(Deba::ListItem, line_prefix(node), last_item, index)
+      
+      @document.break(Deba::ListItem, last_item, index)
       node.children.each { |n| process(n) }
-      @text_run.break(Deba::Paragraph, line_prefix(node))
+      @document.break(Deba::Paragraph)
 
       return
     end
 
     if node_name == 'dt'
-      @text_run.break(Deba::DefinitionTerm, line_prefix(node))
+      @document.break(Deba::DefinitionTerm)
       node.children.each { |n| process(n) }
-      @text_run.break(Deba::Paragraph, line_prefix(node))
+      @document.break(Deba::Paragraph)
 
       return
     end
 
     if node_name == 'dd'
       last_item = node.xpath('count(following-sibling::dd)').to_i == 0
-      @text_run.break(Deba::DefinitionDescription, line_prefix(node), last_item)
+      @document.break(Deba::DefinitionDescription, last_item)
       node.children.each { |n| process(n) }
-      @text_run.break(Deba::Paragraph, line_prefix(node))
+      @document.break(Deba::Paragraph)
 
       return
     end
 
     #These tags terminate the current paragraph, if present, and start a new paragraph
     if BLOCK_INITIATING_TAGS.include?(node_name)
-      @text_run.break(Deba::Paragraph, line_prefix(node))
+      @document.break(Deba::Paragraph)
       node.children.each { |n| process(n) }
-      @text_run.break(Deba::Paragraph, line_prefix(node))
+      @document.break(Deba::Paragraph)
 
       return
     end
 
     if HEADING_TAGS.include?(node_name)
-      @text_run.break(Deba::Heading, node_name[1..-1].to_i)
+      @document.break(Deba::Heading, node_name[1..-1].to_i)
       node.children.each { |n| process(n) }
-      @text_run.break(Deba::Paragraph, line_prefix(node))
+      @document.break(Deba::Paragraph)
 
       return
     end
@@ -112,11 +125,7 @@ class Deba::Extractor
     node.children.each { |n| process(n) }
   end
 
-  def line_prefix(node)
-    if node.xpath('boolean(ancestor::blockquote)')
-      Deba::Blockquote.new
-    else
-      nil
-    end
+  def in_blockquote?
+    @in_blockquote
   end
 end

data/lib/deba/heading.rb

@@ -1,12 +1,10 @@
 class Deba::Heading
-  attr_reader :segments, :level
-
   def initialize(segments, level)
     @segments = segments
     @level = level
   end
 
-  def to_s
-    "#{"#" * @level} #{Deba::Stringifier.new(@segments).stringify}\n\n"
+  def to_a
+    ["#" * @level] + @segments + ["\n\n"]
   end
 end

data/lib/deba/list_item.rb

@@ -1,15 +1,12 @@
 class Deba::ListItem
-  attr_reader :segments
-
-  def initialize(segments, line_prefix, last, index)
+  def initialize(segments, last, index)
     @segments = segments
-    @line_prefix = line_prefix
     @last = last
     @index = index
   end
 
-  def to_s
-    "#{Deba::Stringifier.new([prefix] + @segments, @line_prefix).stringify}\n#{"\n" if @last}"
+  def to_a
+    [prefix] + @segments + ["\n#{"\n" if @last}"]
   end
 
   def prefix

data/lib/deba/paragraph.rb

@@ -1,12 +1,13 @@
 class Deba::Paragraph
-  attr_reader :segments
-
-  def initialize(segments, line_prefix)
+  def initialize(segments)
     @segments = segments
-    @line_prefix = line_prefix
   end
 
-  def to_s
-    "#{Deba::Stringifier.new(@segments, @line_prefix).stringify}\n\n"
+  def always?
+    false
+  end
+
+  def to_a
+    @segments + ["\n\n"]
   end
 end

data/lib/deba/span.rb

@@ -0,0 +1,9 @@
+class Deba::Span
+  def initialize(text)
+    @text = text
+  end
+
+  def to_s
+    @text
+  end
+end

data/lib/deba/stringifier.rb

@@ -1,28 +1,17 @@
 class Deba::Stringifier
-  def initialize(segments, line_prefix = nil)
+  def initialize(segments)
     @segments = segments
-    @line_prefix = line_prefix
   end
 
   def stringify
-    prefix(chunkify)
-  end
-
-  def chunkify
     chunks = @segments.chunk { |segment| segment.class }
 
     chunks.map do |type, chunk_segments|
-      if type == String
-        Deba::Utils.normalise(chunk_segments.join)
-      elsif type == Deba::Break
-        chunk_segments.map { |s| s.to_s }.join
+      if type == Deba::Span
+        Deba::Utils.normalise(chunk_segments.map { |s| s.to_s }.join)
+      else
+        chunk_segments.join
       end
     end.join
   end
-
-  def prefix(text)
-    return text if @line_prefix.nil?
-
-    text.gsub(/^/, @line_prefix.to_s)
-  end
 end

data/lib/deba.rb

@@ -1,28 +1,22 @@
 require "nokogiri"
 
 module Deba
-  VERSION = "0.10.0"
+  VERSION = "0.11.0"
 end
 
 require "deba/utils"
 require "deba/stringifier"
-require "deba/document"
-require "deba/break"
+require "deba/span"
 require "deba/heading"
 require "deba/list_item"
-require "deba/blockquote"
 require "deba/definition_term"
 require "deba/definition_description"
 require "deba/paragraph"
-require "deba/text_runner"
+require "deba/document"
 require "deba/extractor"
 
 module Deba
   def self.extract(html, options = {})
-    document(html, options).to_s
-  end
-
-  def self.document(html, options = {})
     doc = html.is_a?(Nokogiri::XML::Node) ? html : Nokogiri.HTML(html)
     Deba::Extractor.new(doc, options).extract
   end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: deba
 version: !ruby/object:Gem::Version
-  version: 0.10.0
+  version: 0.11.0
 platform: ruby
 authors:
 - Brenton "B-Train" Fletcher
 autorequire: 
 bindir: exe
 cert_chain: []
-date: 2017-02-02 00:00:00.000000000 Z
+date: 2017-02-04 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler
@@ -71,8 +71,6 @@ files:
 - deba.gemspec
 - exe/deba
 - lib/deba.rb
-- lib/deba/blockquote.rb
-- lib/deba/break.rb
 - lib/deba/definition_description.rb
 - lib/deba/definition_term.rb
 - lib/deba/document.rb
@@ -80,8 +78,8 @@ files:
 - lib/deba/heading.rb
 - lib/deba/list_item.rb
 - lib/deba/paragraph.rb
+- lib/deba/span.rb
 - lib/deba/stringifier.rb
-- lib/deba/text_runner.rb
 - lib/deba/utils.rb
 homepage: http://example.com
 licenses:

data/lib/deba/blockquote.rb

@@ -1,5 +0,0 @@
-class Deba::Blockquote
-  def to_s
-    "> "
-  end
-end

data/lib/deba/break.rb

@@ -1,5 +0,0 @@
-class Deba::Break
-  def to_s
-    "\n"
-  end
-end

data/lib/deba/text_runner.rb

@@ -1,33 +0,0 @@
-class Deba::TextRunner
-  def initialize(document)
-    @document = document
-
-    start
-  end
-
-  def <<(segment)
-    @segments << segment
-  end
-
-  def break(*args)
-    finish
-    start(*args)
-  end
-
-  def finish
-    return unless present?
-
-    @args.unshift(@segments)
-    @document << @block_type.new(*@args)
-  end
-
-  def start(*args)
-    @segments = []
-    @block_type = args.shift
-    @args = args
-  end
-
-  def present?
-    @segments.any? { |segment| segment.is_a?(String) && Deba::Utils.present?(segment) }
-  end
-end