slacken 0.1.0 → 0.1.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: ffc05a0da1b3b3fc3534666b1de2a36697448557
-  data.tar.gz: 7d0fda248b1f4a64500e5e6e92b3941adc8320ab
+  metadata.gz: 5fc0f32867586c7b5d1bcc0f8d6f50a7397c8fc7
+  data.tar.gz: 437c297ac0aec496d3002bc88770fce4919a5ec7
 SHA512:
-  metadata.gz: 8ee43ac2dbef90cf74aced8f4e940dde59d80bbbd693fdcc7d822624ea21f51b8a7a13e19345dc9d06cf3363c6c8e1c168e9811f2808484e7f7cc7dfcf0c9067
-  data.tar.gz: 19f4898e6ee95503572ea4bb70005f54d84779f647c93c1c982250685bed83b7f95acf7a4506347f8bc535776e4fd18864edb871bff1f71b071673babde0105d
+  metadata.gz: 4e4283f2ff319d82403f1465b2e8fe74da8a7f2e6d61e3489bfd44224038e7e719e8cba5165a88b8fd3902ed4113810b0baf51e392450fec8d7fca1554251a47
+  data.tar.gz: 6ee2318990b5e0a67579a8114f5e9f1a5b8215b21b5f6110c717c3b154db4b5ae08a962628b057a97124eaa043ce9d5f0ce6f2cda80fd715b8c7b9c2311fecc0

data/.travis.yml

@@ -0,0 +1,3 @@
+language: ruby
+rvm: 2.2
+script: bundle exec rake spec

data/README.md

@@ -1,15 +1,17 @@
-# slacken
+# slacken [![Build Status](https://travis-ci.org/increments/slacken.svg)](https://travis-ci.org/increments/slacken)
 
 ## Description
 This gem translates a html source into a markup text for Slack.
 
+Official description of Slack message formatting is [here](https://api.slack.com/docs/formatting).
+
 ## Examples
 
-Sample input source and output texts are in `sample/`.
+Sample input source and output texts are in `examples/`.
 
 ```
 > require 'slacken'
-> puts Slacken.translate(File.read('sample/source.html'))
+> puts Slacken.translate(File.read('examples/source.html'))
 # *Slacken*
 #
 # This gem translates a html source into *a markup text for Slack*.

data/lib/slacken/document_component.rb

@@ -1,41 +1,36 @@
 require 'forwardable'
+require 'slacken/filters'
 
-require 'slacken/document_component/elim_blanks'
-require 'slacken/document_component/elim_invalid_links'
-require 'slacken/document_component/elim_line_breaks'
-require 'slacken/document_component/extract_img_alt'
-require 'slacken/document_component/group_inlines'
-require 'slacken/document_component/group_indent'
-require 'slacken/document_component/sanitize_link_containers'
-require 'slacken/document_component/sanitize_special_tag_containers'
-require 'slacken/document_component/stringfy_checkbox'
-require 'slacken/document_component/stringfy_emoji'
-
-# Public: An intermediate object that is used when a HTML source is translated into a MarkupElement
-#         representing structure of a markup text.
-#         A DocumentComponent has tree structure and has child nodes as `children`.
 module Slacken
+  # Public: An intermediate object that is used when a HTML source is translated into a MarkupElement
+  #         representing structure of a markup text.
+  #         A DocumentComponent has tree structure and has child nodes as `children`.
   class DocumentComponent
-    include ElimBlanks
-    include ElimInvalidLinks
-    include ElimLineBreaks
-    include ExtractImgAlt
-    include GroupInlines
-    include GroupIndent
-    include SanitizeLinkContainers
-    include SanitizeSpecialTagContainers
-    include StringfyCheckbox
-    include StringfyEmoji
+    NormalizeFilters = [
+      Filters::StringfyEmoji,
+      Filters::StringfyCheckbox,
+      Filters::ExtractImgAlt,
+      Filters::ElimInvalidLinks,
+      Filters::SanitizeHeadline,
+      Filters::SanitizeLink,
+      Filters::SanitizeList,
+      Filters::SanitizeTable,
+      Filters::GroupInlines,
+      Filters::GroupIndent,
+      Filters::ElimBlanks,
+      Filters::ElimLineBreaks,
+    ]
 
     extend Forwardable
     def_delegators :@type, :block?, :inline?
 
-    attr_reader :type, :attrs, :children
+    attr_reader :type, :attrs, :children, :marks
 
     def initialize(type, children = [], attrs = {})
       @type = NodeType.create(type)
       @attrs = attrs
       @children = children
+      @marks = {}
     end
 
     def derive(new_children, updates = {})
@@ -49,19 +44,11 @@ module Slacken
     end
 
     def normalize
-      stringfy_emoji
-        .stringfy_checkbox
-        .extract_img_alt
-        .elim_invalid_links
-        .sanitize_link_containers
-        .sanitize_special_tag_containers
-        .group_inlines
-        .group_indent
-        .elim_blanks
-        .elim_line_breaks
+      NormalizeFilters.reduce(self) do |component, filter_klass|
+        filter_klass.new.call(component)
+      end
     end
 
-    # Private: Convert this element to a MarkupElement.
     def produce_element
       if type.member_of?(:table)
         TableElement.new(children.map(&:produce_element))

data/lib/slacken/dom_container.rb

@@ -1,7 +1,7 @@
 require 'nokogiri'
 
-# Public: a DOM tree container parsed by Nokogiri.
 module Slacken
+  # Public: a DOM tree container parsed by Nokogiri.
   class DomContainer
     attr_reader :root
 

data/lib/slacken/filter.rb

@@ -0,0 +1,27 @@
+module Slacken
+  # Public: Base class of filters for DocumentComponent.
+  class Filter
+    attr_reader :options
+    def initialize(options = {})
+      @options = options
+    end
+
+    # Public: Create a new refined DocumentComponent from the given component.
+    #
+    # component - A DocumentComponent object.
+    #
+    # Returns a new refined DocumentComponent object.
+    def call(component)
+      fail NotImplementedError
+    end
+
+    # Public: Check if the given component passes postcondition of the filter.
+    #
+    # component - A DocumentComponent object to check.
+    #
+    # Returns true if the component passes the postcondition.
+    def valid?(component)
+      true
+    end
+  end
+end

data/lib/slacken/filters.rb

@@ -0,0 +1,19 @@
+module Slacken
+  module Filters
+  end
+end
+
+require 'slacken/filter'
+
+require 'slacken/filters/elim_blanks'
+require 'slacken/filters/elim_invalid_links'
+require 'slacken/filters/elim_line_breaks'
+require 'slacken/filters/extract_img_alt'
+require 'slacken/filters/group_inlines'
+require 'slacken/filters/group_indent'
+require 'slacken/filters/sanitize_headline'
+require 'slacken/filters/sanitize_link'
+require 'slacken/filters/sanitize_list'
+require 'slacken/filters/sanitize_table'
+require 'slacken/filters/stringfy_checkbox'
+require 'slacken/filters/stringfy_emoji'

data/lib/slacken/filters/elim_blanks.rb

@@ -0,0 +1,41 @@
+module Slacken::Filters
+  # Public: Reject blank elements.
+  class ElimBlanks < Slacken::Filter
+    def call(component)
+      if component.type.member_of?(:pre)
+        component
+      else
+        component.derive(
+          component.children.reject(&method(:blank?)).map(&method(:call))
+        )
+      end
+    end
+
+    def valid?(component)
+      if component.type.member_of?(:pre)
+        true
+      else
+        !blank?(component) && component.children.all?(&method(:valid?))
+      end
+    end
+
+    private
+
+    def blank?(component)
+      # Reduce complexity of calculation by marking.
+      # (`blank?` traces the given tree to its leaf nodes.)
+      return component.marks[:blank] if component.marks.has_key?(:blank)
+
+      component.marks[:blank] =
+        case component.type.name
+        when :pre, :ul, :li, :br, :hr, :img, :checkbox
+          false
+        when :text, :emoji
+          content = component.attrs[:content]
+          content.nil? || !content.match(/\A\s*\Z/).nil?
+        else
+          component.children.empty? || component.children.all?(&method(:blank?))
+        end
+    end
+  end
+end

data/lib/slacken/filters/elim_invalid_links.rb

@@ -0,0 +1,33 @@
+module Slacken::Filters
+  # Public: Eliminate internal links and blank links.
+  class ElimInvalidLinks < Slacken::Filter
+    def call(component)
+      if invalid_link?(component)
+        component.derive(
+          component.children.map(&method(:call)),
+          type: :span
+        )
+      else
+        component.derive(
+          component.children.map(&method(:call)),
+        )
+      end
+    end
+
+    def valid?(component)
+      return false if invalid_link?(component)
+      component.children.all?(&method(:valid?))
+    end
+
+    private
+
+    def invalid_link?(component)
+      if component.type.member_of?(:a)
+        link = component.attrs[:href]
+        link.nil? || !link.match(%r{\Ahttps?://})
+      else
+        false
+      end
+    end
+  end
+end

data/lib/slacken/filters/elim_line_breaks.rb

@@ -0,0 +1,36 @@
+module Slacken::Filters
+  # Public: Remove line breaks from texts.
+  class ElimLineBreaks < Slacken::Filter
+    def call(component)
+      case component.type.name
+      when :pre
+        component
+      when :text
+        new_content = component.attrs[:content].gsub(/[\r\n]/, '')
+        component.derive(
+          component.children,
+          attrs: component.attrs.merge(content: new_content)
+        )
+      else
+        component.derive(
+          component.children.map(&method(:call))
+        )
+      end
+    end
+
+    def valid?(component)
+      has_no_line_breaks?(component)
+    end
+
+    def has_no_line_breaks?(component)
+      case component.type.name
+      when :pre
+        true
+      when :text
+        !component.attrs[:content].match(/[\r\n]/)
+      else
+        component.children.all?(&method(:has_no_line_breaks?))
+      end
+    end
+  end
+end

data/lib/slacken/filters/extract_img_alt.rb

@@ -0,0 +1,18 @@
+module Slacken::Filters
+  # Public: Convert alt attribute of img node to child text node.
+  class ExtractImgAlt < Slacken::Filter
+    def call(component)
+      if component.type.member_of?(:img)
+        component.derive([
+          component.class.new(
+            :text, [], content: component.attrs[:alt] || component.attrs[:src]
+          )
+        ])
+      else
+        component.derive(
+          component.children.map(&method(:call))
+        )
+      end
+    end
+  end
+end

data/lib/slacken/filters/group_indent.rb

@@ -0,0 +1,28 @@
+module Slacken::Filters
+  # Public: Wrap child content nodes of each li or dd node with an indent node.
+  class GroupIndent < Slacken::Filter
+    def call(component)
+      if component.type.member_of?(%i(li dd))
+        head, *tails = component.children.map(&method(:call))
+        component.derive([head, component.class.new(:indent, tails)])
+      else
+        component.derive(
+          component.children.map(&method(:call))
+        )
+      end
+    end
+
+    def valid?(component)
+      if component.type.member_of?(%i(li dd))
+        head, tail = component.children
+        if tail
+          tail.type.member_of?(:indent) && valid?(tail)
+        else
+          true
+        end
+      else
+        component.children.all?(&method(:valid?))
+      end
+    end
+  end
+end

data/lib/slacken/filters/group_inlines.rb

@@ -0,0 +1,31 @@
+module Slacken::Filters
+  # Public: Group inline elements and wrap them in a wrapper node.
+  #         Wrapper nodes are introduced to group inline nodes in a paragraph.
+  class GroupInlines < Slacken::Filter
+    def call(component)
+      if component.block?
+        component.derive(group_component(component))
+      else
+        component.derive(component.children.map(&method(:call)))
+      end
+    end
+
+    def valid?(component)
+      if component.type.member_of?(:wrapper)
+        true
+      elsif component.inline?
+        false
+      else
+        component.children.all?(&method(:valid?))
+      end
+    end
+
+    private
+
+    def group_component(component)
+      component.children.map(&method(:call)).chunk(&:inline?).map do |is_inline, group|
+        is_inline ? component.class.new(:wrapper, group) : group
+      end.flatten
+    end
+  end
+end

data/lib/slacken/filters/sanitize_headline.rb

@@ -0,0 +1,46 @@
+module Slacken::Filters
+  # Public: Sanitize not allowed tags in headline.
+  class SanitizeHeadline < Slacken::Filter
+    def call(component)
+      # NOTE: each special tag (list, headline, and table) is not allowed to occur
+      #       in another type special tags.
+      if component.type.name =~ /h\d/
+        component.derive(component.children.map(&method(:sanitize_headline)))
+      else
+        component.derive(component.children.map(&method(:call)))
+      end
+    end
+
+    def valid?(component)
+      if component.type.name =~ /h\d/
+        component.children.all?(&method(:headline_sanitized?))
+      else
+        component.children.all?(&method(:valid?))
+      end
+    end
+
+    private
+
+    def sanitize_headline(component)
+      if component.type.allowed_in_headline?
+        component.derive(
+          component.children.map(&method(:sanitize_headline))
+        )
+      else
+        # No block tags are allowed.
+        component.derive(
+          component.children.map(&method(:sanitize_headline)),
+          type: :span
+        )
+      end
+    end
+
+    def headline_sanitized?(component)
+      if component.type.allowed_in_headline?
+        component.children.all?(&method(:headline_sanitized?))
+      else
+        false
+      end
+    end
+  end
+end

data/lib/slacken/filters/sanitize_link.rb

@@ -0,0 +1,48 @@
+module Slacken::Filters
+  # Public: Sanitize not allowed tags in links.
+  class SanitizeLink < Slacken::Filter
+    def call(component)
+      if component.type.member_of?(%i(img a iframe))
+        component.derive(
+          component.children.map(&method(:sanitize))
+        )
+      else
+        component.derive(
+          component.children.map(&method(:call))
+        )
+      end
+    end
+
+    def valid?(component)
+      if component.type.member_of?(%i(img a iframe))
+        component.children.all?(&method(:link_containers_sanitized?))
+      else
+        component.children.all?(&method(:valid?))
+      end
+    end
+
+    private
+
+    def sanitize(component)
+      if component.type.allowed_in_link?
+        component.derive(
+          component.children.map(&method(:sanitize))
+        )
+      else
+        # No block tags are allowed.
+        component.derive(
+          component.children.map(&method(:sanitize)),
+          type: :span
+        )
+      end
+    end
+
+    def link_containers_sanitized?(component)
+      if component.type.allowed_in_link?
+        component.children.all?(&method(:link_containers_sanitized?))
+      else
+        false
+      end
+    end
+  end
+end