html-renderer 0.0.3 → 0.0.4

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 3bbe90456ebf54ed08a1d78e5056fa82a4ccc68e6487cd9fce72f807669464e8
-  data.tar.gz: 2106840d7e1b57aedaced18c6cf122d928f6ce6def84cf191c8c4d84373750da
+  metadata.gz: ef89dd4c00345deca8c54a7ac62823dd8a506a0f941186cdd6d82ebd51ae7d47
+  data.tar.gz: 4a2fd1e8b670ff87fd63b7e94a7e50af7bd66e73f94bb48b197b9f8374b8d196
 SHA512:
-  metadata.gz: ca6fda95a4b8d7ae62310d2600486420f2d239e443474068e00953e86d3c61e20680651c2b59d688aa4e9e73884bd98db131bbf053597d993d508e37dba9fe00
-  data.tar.gz: 7bea3114914af7a5122adad1e74b0beb44a7d2cc5658f88b056aae1b1ac1b37180999b774151e7def45e4f6d6e2b276c05259009721d250a96aa4fcf21351f53
+  metadata.gz: b0dd0122febdc3018b1d2d1f38cd6fc19c23e2495989b6b8ccca1f5e7cf096d159a38aa85fdb600f48767d75827ce6f025282a9b0a206e74aa93c60e5c02bedc
+  data.tar.gz: 8164fb8d54a3f2895bb288d32eb9e03716ff8bd3a40b68382385c499cae363c1d0ff685f77abeac96a83c0b313edfd9c9c3b013173d878ac46b155b28b84466c

data/VERSION

@@ -1 +1 @@
-0.0.3
+0.0.4

data/bin/html2ansi

@@ -69,6 +69,9 @@ private
     s&.downcase&.scan(/\w+/)&.join
   end
 
+  def subscript(s)
+    "[#{s}]"
+  end
 
 public
 
@@ -139,7 +142,7 @@ public
       when 1 then title.bold.yellow
       when 2 then title.bold.cyan
       when 3 then title.bold.blue
-      else title.purple
+      else title.magenta
     end
 
     "#{bar}\n  #{title}\n#{bar}\n\n"
@@ -219,7 +222,7 @@ def render(stream, paged=false)
     lesspipe { |less| less.puts output }
   else
     puts output
-  end 
+  end
 end
 
 #######################################################################################

data/examples/ansi_renderer.rb

@@ -102,7 +102,7 @@ public
       when 1 then title.bold.yellow
       when 2 then title.bold.cyan
       when 3 then title.bold.blue
-      else title.purple
+      else title.magenta
     end
 
     "#{bar}\n  #{title}\n#{bar}\n\n"

data/lib/html-renderer/base.rb

@@ -96,6 +96,8 @@ private
             emphasis(render_children(node, state))
           when "sup"
             superscript(render_children(node, state))
+          when "sub"
+            subscript(render_children(node, state))
           when "u"
             underline(render_children(node, state))
           when "br"
@@ -118,7 +120,7 @@ private
             list_item(render_children(node, state), state.list_order)
 
           when "code"
-            block_code(render_children(node, state))
+            block_code(render_children(node, state), nil)
           when "blockquote"
             block_quote(render_children(node, state))
 
@@ -149,9 +151,9 @@ private
             table(header, rows)
 
           when "html", "body", "nav", "span", "form", "label", "input", "button", "section", "fieldset",
-               "menu", "article", "header", "time", "aside", "footer", "nobr", "wbr", 
+               "menu", "article", "header", "time", "aside", "footer", "nobr", "wbr",
                "table", "tr", "td", "th", "tt", "thead", "tbody", "noscript", "select",
-               "address", "center"
+               "address", "center", "small"
             render_children(node, state)
 
           when "head", "script", "link", "style"

data/lib/html-renderer/html_parser.rb

@@ -0,0 +1,105 @@
+require 'strscan'
+
+class Stack < Array
+  alias_method :top, :last
+  alias_method :peek, :last
+end
+
+class String
+  def recursive_inspect(depth)
+    ("  "*depth)+inspect
+  end
+end
+
+class HTMLParser
+
+  OPEN_TAG_RE  = %r{<([^>]+)>}
+  CLOSE_TAG_RE = %r{</([^>]+)>}
+  TEXT_RE      = %r{[^<]+}
+  ATTR_RE      = %r{(\w+)=(?:"([^"]+)"|'([^']+)'|(\w+))}
+
+  class Tag
+
+    attr_accessor :name, :attrs, :children
+
+    def self.from_str(s)
+      name, rest = s.split(/\s+/, 2)
+      
+      if rest
+        attrs = rest.scan(HTMLParser::ATTR_RE).flatten.compact.each_slice(2).to_h
+      else
+        attrs = {}
+      end
+      new(name, attrs)
+    end
+
+    def initialize(name, attrs={}, children=[])
+      @name     = name
+      @attrs    = attrs
+      @children = children
+    end
+
+    def recursive_inspect(depth=0)
+      curdent = "  "*depth
+      indent = "  "*(depth+1)
+      "#{curdent}<#{name} #{attrs}>\n#{indent}#{children.map{|c| c.recursive_inspect(depth+1)}}\n#{curdent}</#{name}>"
+    end
+
+  end
+
+  def initialize(html)
+    @s = StringScanner.new(html)
+    # @s = html
+  end
+
+  def each_tag
+      until @s.eos?
+      if @s.scan(CLOSE_TAG_RE)
+        yield [:close_tag, @s.captures.first]
+      elsif @s.scan(OPEN_TAG_RE)
+        tag = Tag.from_str(@s.captures.first)
+        yield [:open_tag, tag]
+      elsif @s.scan(TEXT_RE)
+        yield [:text, @s.matched]
+      end
+    end
+  end
+
+  def as_tree
+    tree.map { |e| e.recursive_inspect }
+  end
+
+  def tree
+    stack = Stack.new
+    stack.push Tag.new("root")
+
+    each_tag do |type, elem|
+      case type
+      when :text
+        text = elem.strip
+        stack.top.children << text unless text.empty?
+      when :open_tag
+        stack.top.children << elem
+        stack.push elem
+      when :close_tag
+        stack.pop
+      else
+        raise "wat"
+      end
+    end
+
+    stack
+  end
+
+end
+
+unless file = ARGV.first
+  file = "test.html"
+end
+
+html = File.read(file)
+
+r = HTMLParser.new(html)
+r.each_tag{|t| p t}
+
+# puts r.as_tree

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: html-renderer
 version: !ruby/object:Gem::Version
-  version: 0.0.3
+  version: 0.0.4
 platform: ruby
 authors:
 - epitron
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2018-06-11 00:00:00.000000000 Z
+date: 2018-11-05 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: oga
@@ -61,6 +61,7 @@ files:
 - examples/test.html
 - lib/html-renderer.rb
 - lib/html-renderer/base.rb
+- lib/html-renderer/html_parser.rb
 homepage: http://github.com/epitron/html-renderer/
 licenses:
 - WTFPL