infoboxer 0.3.0 → 0.4.0

data/lib/infoboxer/parser/html.rb

@@ -1,4 +1,4 @@
-# encoding: utf-8
+# frozen_string_literal: true
 
 module Infoboxer
   class Parser

data/lib/infoboxer/parser/image.rb

@@ -1,4 +1,4 @@
-# encoding: utf-8
+# frozen_string_literal: true
 
 module Infoboxer
   class Parser
@@ -11,7 +11,7 @@ module Infoboxer
 
         path = @context.scan_until(/\||\]\]/)
         attrs = @context.matched == '|' ? image_attrs : {}
-        Tree::Image.new(path, attrs)
+        Tree::Image.new(path, **attrs)
       end
 
       def image_attrs

data/lib/infoboxer/parser/inline.rb

@@ -1,4 +1,4 @@
-# encoding: utf-8
+# frozen_string_literal: true
 
 module Infoboxer
   class Parser
@@ -18,7 +18,8 @@ module Infoboxer
 
           if @context.eof?
             break unless until_pattern
-            @context.fail!("#{until_pattern} not found, starting from #{start}")
+
+            @context.fail!("#{until_pattern.source} not found, starting from #{start}")
           end
 
           if @context.eol?
@@ -35,7 +36,7 @@ module Infoboxer
         guarded_loop do
           # FIXME: quick and UGLY IS HELL JUST TRYING TO MAKE THE SHIT WORK
           chunk =
-            if @context.inline_eol_sign == /^\]/
+            if @context.inline_eol_sign == /^\]/ # rubocop:disable Style/CaseLikeIf
               @context.scan_until(re.short_inline_until_cache_brackets[until_pattern])
             elsif @context.inline_eol_sign == /^\]\]/
               @context.scan_until(re.short_inline_until_cache_brackets2[until_pattern])
@@ -66,7 +67,8 @@ module Infoboxer
 
           if @context.eof?
             break unless until_pattern
-            @context.fail!("#{until_pattern} not found")
+
+            @context.fail!("#{until_pattern.source} not found")
           end
 
           if @context.eol?
@@ -83,7 +85,7 @@ module Infoboxer
 
       private
 
-      def inline_formatting(match)
+      def inline_formatting(match) # rubocop:disable Metrics/MethodLength, Metrics/CyclomaticComplexity, Metrics/AbcSize
         case match
         when "'''''"
           BoldItalic.new(short_inline(/'''''/))
@@ -109,6 +111,8 @@ module Infoboxer
           reference(Regexp.last_match(1))
         when /<math>/
           math
+        when /<gallery([^>]*)>/
+          gallery(Regexp.last_match(1))
         when '<'
           html || Text.new(match) # it was not HTML, just accidental <
         else
@@ -126,8 +130,18 @@ module Infoboxer
           caption = inline(/\]\]/)
           @context.pop_eol_sign
         end
+        name, namespace = link.split(':', 2).reverse
+        lnk, params =
+          if @context.traits.namespace?(namespace)
+            [link, {namespace: namespace}]
+          elsif @context.traits.interwiki?(namespace)
+            [name, {interwiki: namespace}]
+          else
+            [link, {}]
+          end
 
-        Wikilink.new(link, caption)
+        puts @context.rest if lnk.nil?
+        Wikilink.new(lnk, caption, **params)
       end
 
       # http://en.wikipedia.org/wiki/Help:Link#External_links
@@ -145,7 +159,7 @@ module Infoboxer
 
       def reference(param_str, closed = false)
         children = closed ? Nodes[] : long_inline(%r{</ref>})
-        Ref.new(children, parse_params(param_str))
+        Ref.new(children, **parse_params(param_str))
       end
 
       def math
@@ -159,6 +173,35 @@ module Infoboxer
           Text.new(@context.scan_continued_until(%r{</nowiki>}))
         end
       end
+
+      def gallery(tag_rest)
+        params = parse_params(tag_rest)
+        images = []
+        guarded_loop do
+          @context.next! if @context.eol?
+          path = @context.scan_until(%r{</gallery>|\||$})
+          attrs = @context.matched == '|' ? gallery_image_attrs : {}
+          unless path.empty?
+            # FIXME: what if path NOT matches the namespace?
+            images << Tree::Image.new(path.sub(/^#{re.file_namespace.source}/i, ''), **attrs)
+          end
+          break if @context.matched == '</gallery>'
+        end
+        Gallery.new(images, **params)
+      end
+
+      def gallery_image_attrs
+        nodes = []
+
+        guarded_loop do
+          nodes << short_inline(%r{\||</gallery>})
+          break if @context.eol? || @context.matched?(%r{</gallery>})
+        end
+
+        nodes.map(&method(:image_attr))
+             .inject(&:merge)
+             .reject { |_k, v| v.nil? || v.empty? }
+      end
     end
 
     require_relative 'image'

data/lib/infoboxer/parser/paragraphs.rb

@@ -1,4 +1,4 @@
-# encoding: utf-8
+# frozen_string_literal: true
 
 module Infoboxer
   class Parser
@@ -14,7 +14,7 @@ module Infoboxer
 
           @context.next!
         end
-        nodes.flow_templates
+        nodes
       end
 
       private
@@ -25,7 +25,7 @@ module Infoboxer
           heading(Regexp.last_match[:text], Regexp.last_match[:level])
         when /^\s*{\|/
           table
-        when /^[\*\#:;]./
+        when /^[*\#:;]./
           list(until_pattern)
         when /^-{4,}/
           HR.new

data/lib/infoboxer/parser/table.rb

@@ -1,4 +1,4 @@
-# encoding: utf-8
+# frozen_string_literal: true
 
 module Infoboxer
   class Parser
@@ -14,12 +14,13 @@ module Infoboxer
 
         prms = table_params
         log "Table params found #{prms}"
-        table = Tree::Table.new(Nodes[], prms)
+        table = Tree::Table.new(Nodes[], **prms)
 
         @context.next!
 
-        loop do
+        guarded_loop do
           table_next_line(table) or break
+          log 'Next table row'
           @context.next!
         end
 
@@ -52,6 +53,9 @@ module Infoboxer
           table_template(table)
         when nil
           return false
+        when /^(?<level>={2,})\s*(?<text>.+?)\s*\k<level>$/ # heading implicitly closes the table
+          @context.prev!
+          return false
         else
           return table_cell_cont(table)
         end
@@ -61,16 +65,25 @@ module Infoboxer
 
       def table_row(table, param_str)
         log 'Table row found'
-        table.push_children(TableRow.new(Nodes[], parse_params(param_str)))
+        table.push_children(TableRow.new(Nodes[], **parse_params(param_str)))
       end
 
       def table_caption(table)
         log 'Table caption found'
         @context.skip(/^\s*\|\+\s*/)
 
+        params = if @context.check(/[^|{\[]+\|([^|]|$)/)
+                   parse_params(@context.scan_until(/\|/))
+                 else
+                   {}
+                 end
+
         children = inline(/^\s*([|!]|{\|)/)
-        @context.prev! if @context.eol? # compensate next! which will be done in table()
-        table.push_children(TableCaption.new(children.strip))
+        if @context.matched
+          @context.unscan_matched!
+          @context.prev! # compensate next! which will be done in table()
+        end
+        table.push_children(TableCaption.new(children.strip, **params))
       end
 
       def table_cells(table, cell_class = TableCell)
@@ -80,13 +93,13 @@ module Infoboxer
 
         @context.skip(/\s*[!|]\s*/)
         guarded_loop do
-          params = if @context.check(/[^|{|\[]+\|([^\|]|$)/)
+          params = if @context.check(/[^|{\[]+\|([^|]|$)/)
                      parse_params(@context.scan_until(/\|/))
                    else
                      {}
                    end
           content = short_inline(/(\|\||!!)/)
-          row.push_children(cell_class.new(content, params))
+          row.push_children(cell_class.new(content, **params))
           break if @context.eol?
         end
       end
@@ -94,15 +107,17 @@ module Infoboxer
       def table_template(table)
         contents = paragraph(/^\s*([|!]|{\|)/).to_templates?
 
-        if (row = table.children.last).is_a?(TableRow)
-          if (cell = row.children.last).is_a?(BaseCell)
-            cell.push_children(*contents)
-          else
-            row.push_children(*contents)
-          end
-        else
-          table.push_children(*contents)
-        end
+        # Note: in fact, without full template parsing, we CAN'T know what level to insert it:
+        # Template can be something like <tr><td>Foo</td></tr>
+        # But for consistency, we insert all templates inside the <td>, forcing this <td>
+        # to exist.
+
+        table.push_children(TableRow.new) unless table.children.last.is_a?(TableRow)
+        row = table.children.last
+        row.push_children(TableCell.new) unless row.children.last.is_a?(BaseCell)
+        cell = row.children.last
+
+        cell.push_children(*contents)
       end
 
       # Good news, everyone! Table can be IMPLICITLY closed when it's
@@ -122,6 +137,7 @@ module Infoboxer
         unless container
           # return "table not continued" unless row is empty
           return true if @context.current.empty?
+
           @context.prev!
           return false
         end

data/lib/infoboxer/parser/template.rb

@@ -1,4 +1,4 @@
-# encoding: utf-8
+# frozen_string_literal: true
 
 module Infoboxer
   class Parser
@@ -29,8 +29,8 @@ module Infoboxer
 
         guarded_loop do
           @context.next! while @context.eol?
-          if @context.check(/\s*([^ =}|<]+)\s*=\s*/)
-            name = @context.scan(/\s*([^ =]+)/).strip
+          if @context.check(/\s*([^=}|<]+)\s*=\s*/)
+            name = @context.scan(/\s*([^=]+)/).strip
             @context.skip(/\s*=\s*/)
           else
             name = num
@@ -46,13 +46,14 @@ module Infoboxer
           log 'Variable value found'
 
           break if @context.eat_matched?('}}')
+
           @context.eof? and @context.fail!("Unexpected break of template variables: #{res}")
         end
         res
       end
 
       def sanitize_value(nodes)
-        nodes.pop if nodes.last.is_a?(Pre) && nodes.last.text =~ /^\s*$/ # FIXME: dirty!
+        nodes.pop if (nodes.last.is_a?(Pre) || nodes.last.is_a?(Text)) && nodes.last.text =~ /^\s*$/ # FIXME: dirty!
         nodes
       end
     end

data/lib/infoboxer/parser/util.rb

@@ -1,4 +1,4 @@
-# encoding: utf-8
+# frozen_string_literal: true
 
 module Infoboxer
   class Parser
@@ -12,6 +12,7 @@ module Infoboxer
         \[[a-z]+://   |     # external link
         <nowiki[^>]*> |     # nowiki
         <ref[^>]*>    |     # reference
+        <gallery[^>]*>|     # gallery
         <math>        |     # math
         <                   # HTML tag
       ))x

data/lib/infoboxer/templates.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module Infoboxer
   # This module covers advanced MediaWiki templates usage.
   #

data/lib/infoboxer/templates/base.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module Infoboxer
   module Templates
     class Base < Tree::Template

data/lib/infoboxer/templates/set.rb

@@ -1,4 +1,4 @@
-# encoding: utf-8
+# frozen_string_literal: true
 
 module Infoboxer
   module Templates

data/lib/infoboxer/tree.rb

@@ -1,4 +1,4 @@
-# encoding: utf-8
+# frozen_string_literal: true
 
 module Infoboxer
   # Infoboxer provides you with tree structure of the Wikipedia page,
@@ -63,7 +63,7 @@ module Infoboxer
     require_relative 'tree/nodes'
 
     %w[text compound inline
-       image html paragraphs list template table ref math
+       image gallery html paragraphs list template table ref math
        document].each do |type|
       require_relative "tree/#{type}"
     end

data/lib/infoboxer/tree/compound.rb

@@ -1,11 +1,11 @@
-# encoding: utf-8
+# frozen_string_literal: true
 
 module Infoboxer
   module Tree
     # Base class for all nodes with children.
     class Compound < Node
-      def initialize(children = Nodes.new, params = {})
-        super(params)
+      def initialize(children = Nodes.new, **params)
+        super(**params)
         @children = Nodes[*children]
         @children.each { |c| c.parent = self }
       end

data/lib/infoboxer/tree/document.rb

@@ -1,4 +1,4 @@
-# encoding: utf-8
+# frozen_string_literal: true
 
 module Infoboxer
   module Tree

data/lib/infoboxer/tree/gallery.rb

@@ -0,0 +1,12 @@
+# frozen_string_literal: true
+
+module Infoboxer
+  module Tree
+    # Represents gallery of images (contents of `<gallery>` special tag).
+    #
+    # See [Wikipedia Tutorial](https://en.wikipedia.org/wiki/Help:Gallery_tag)
+    # for explanation of attributes.
+    class Gallery < Compound
+    end
+  end
+end

data/lib/infoboxer/tree/html.rb

@@ -1,4 +1,4 @@
-# encoding: utf-8
+# frozen_string_literal: true
 
 module Infoboxer
   module Tree
@@ -13,7 +13,7 @@ module Infoboxer
     # Represents HTML tag, surrounding some contents.
     class HTMLTag < Compound
       def initialize(tag, attrs, children = Nodes.new)
-        super(children, attrs)
+        super(children, **attrs)
         @tag = tag
       end
 
@@ -45,7 +45,7 @@ module Infoboxer
     #
     class HTMLOpeningTag < Node
       def initialize(tag, attrs)
-        super(attrs)
+        super(**attrs)
         @tag = tag
       end
 

data/lib/infoboxer/tree/image.rb

@@ -1,4 +1,4 @@
-# encoding: utf-8
+# frozen_string_literal: true
 
 module Infoboxer
   module Tree
@@ -7,9 +7,9 @@ module Infoboxer
     # See [Wikipedia Tutorial](https://en.wikipedia.org/wiki/Wikipedia:Extended_image_syntax)
     # for explanation of attributes.
     class Image < Node
-      def initialize(path, params = {})
-        @caption = params.delete(:caption)
-        super({path: path}.merge(params))
+      def initialize(path, caption: nil, **params)
+        @caption = caption
+        super(path: path, **params)
       end
 
       # Image caption. Can have (sometimes many) other nodes inside.

data/lib/infoboxer/tree/inline.rb

@@ -1,4 +1,4 @@
-# encoding: utf-8
+# frozen_string_literal: true
 
 module Infoboxer
   module Tree
@@ -17,8 +17,8 @@ module Infoboxer
 
     # Base class for internal/external links,
     class Link < Compound
-      def initialize(link, label = nil)
-        super(label || Nodes.new([Text.new(link)]), link: link)
+      def initialize(link, label = nil, **attr)
+        super(label || Nodes.new([Text.new(link)]), link: link, **attr)
       end
 
       # @!attribute [r] link