infoboxer 0.3.0 → 0.4.0

data/lib/infoboxer/tree/linkable.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module Infoboxer
   module Tree
     # Module included into everything, that can be treated as
@@ -15,7 +17,7 @@ module Infoboxer
       # * {Tree::Nodes#follow} for extracting multiple links at once;
       # * {MediaWiki#get} for basic information on page extraction.
       def follow
-        client.get(link)
+        client.get(link, interwiki: interwiki)
       end
 
       # Human-readable page URL
@@ -28,6 +30,9 @@ module Infoboxer
 
       protected
 
+      # redefined in {Wikilink}
+      def interwiki; end
+
       def page
         lookup_parents(MediaWiki::Page).first or fail('Not in a page from real source')
       end

data/lib/infoboxer/tree/list.rb

@@ -1,4 +1,4 @@
-# encoding: utf-8
+# frozen_string_literal: true
 
 module Infoboxer
   module Tree
@@ -15,9 +15,8 @@ module Infoboxer
       # Internal, used by {Parser}
       def merge!(other)
         ochildren = other.children.dup
-        if children.last && children.last.can_merge?(ochildren.first)
-          children.last.merge!(ochildren.shift)
-        end
+        children.last.merge!(ochildren.shift) \
+          if children.last&.can_merge?(ochildren.first)
         push_children(*ochildren)
       end
 
@@ -81,7 +80,7 @@ module Infoboxer
     # Represents ordered list (list with numbers).
     class OrderedList < List
       def make_marker(item)
-        list_text_indent + "#{(item.index + 1)}. "
+        list_text_indent + "#{item.index + 1}. "
       end
     end
 

data/lib/infoboxer/tree/math.rb

@@ -1,12 +1,11 @@
+# frozen_string_literal: true
+
 module Infoboxer
   module Tree
     # Represents node of math formulae marked with TeX
     #
     # See also: https://en.wikipedia.org/wiki/Help:Displaying_a_formula
     class Math < Text
-      def text
-        "<math>#{super}</math>"
-      end
     end
   end
 end

data/lib/infoboxer/tree/node.rb

@@ -1,4 +1,4 @@
-# encoding: utf-8
+# frozen_string_literal: true
 
 require 'htmlentities'
 
@@ -11,7 +11,7 @@ module Infoboxer
     # you will receive it from tree and use for navigations.
     #
     class Node
-      def initialize(params = {})
+      def initialize(**params)
         @params = params
       end
 
@@ -154,7 +154,7 @@ module Infoboxer
       end
 
       def show_params(prms = nil)
-        (prms || params).map { |k, v| "#{k}: #{v.inspect}" }.join(', ')
+        (prms || params).compact.map { |k, v| "#{k}: #{v.inspect}" }.join(', ')
       end
 
       def indent(level)
@@ -162,7 +162,7 @@ module Infoboxer
       end
 
       def _eq(_other)
-        fail(NotImplementedError, "#_eq should be defined in subclasses (called for #{self.class})")
+        false
       end
 
       def decode(str)

data/lib/infoboxer/tree/nodes.rb

@@ -1,4 +1,4 @@
-# encoding: utf-8
+# frozen_string_literal: true
 
 module Infoboxer
   module Tree
@@ -38,15 +38,32 @@ module Infoboxer
       # @!method compact
       #    Just like Array#compact, but returns Nodes
 
+      # @!method grep(pattern)
+      #    Just like Array#grep, but returns Nodes
+
+      # @!method grep_v(pattern)
+      #    Just like Array#grep_v, but returns Nodes
+
       # @!method -(other)
       #    Just like Array#-, but returns Nodes
 
-      %i[select reject sort_by flatten compact -].each do |sym|
+      # @!method +(other)
+      #    Just like Array#+, but returns Nodes
+
+      # NB: Since Ruby 3.0, we need to redefine all Enumerable methods (otherwise they return Array).
+      # TODO: Check those lacking overrides!
+
+      %i[
+        select reject sort_by flatten compact grep grep_v - +
+        take_while drop_while
+      ].each do |sym|
         define_method(sym) do |*args, &block|
           Nodes[*super(*args, &block)]
         end
       end
 
+      alias_method :filter, :select
+
       # Just like Array#first, but returns Nodes, if provided with `n` of elements.
       def first(n = nil)
         if n.nil?
@@ -75,6 +92,21 @@ module Infoboxer
         end
       end
 
+      # Just like Array#flat_map, but returns Nodes, **if** all map results are Node
+      def flat_map
+        res = super
+        if res.all? { |n| n.is_a?(Node) || n.is_a?(Nodes) }
+          Nodes[*res]
+        else
+          res
+        end
+      end
+
+      # Just like Array#group, but returns hash with `{<grouping variable> => Nodes}`
+      def group_by
+        super.transform_values { |group| Nodes[*group] }
+      end
+
       # @!method prev_siblings
       #   Previous siblings (flat list) of all nodes inside.
 
@@ -129,6 +161,12 @@ module Infoboxer
         map(&:text).join
       end
 
+      alias_method :to_s, :text
+
+      def unwrap
+        map { |n| n.respond_to?(:unwrap) ? n.unwrap : n }
+      end
+
       # Fetches pages by ALL wikilinks inside in ONE query to MediaWiki
       # API.
       #
@@ -139,23 +177,27 @@ module Infoboxer
       # @return [Nodes<MediaWiki::Page>] It is still `Nodes`, so you
       #   still can process them uniformely.
       def follow
-        links = select { |n| n.respond_to?(:link) }.map(&:link)
+        links = grep(Linkable)
         return Nodes[] if links.empty?
+
         page = first.lookup_parents(MediaWiki::Page).first or
           fail('Not in a page from real source')
         page.client or fail('MediaWiki client not set')
-        page.client.get(*links)
+        pages = links.group_by(&:interwiki)
+                     .flat_map { |iw, ls| page.client.get(*ls.map(&:link), interwiki: iw) }
+        pages.count == 1 ? pages.first : Nodes[*pages]
       end
 
       # @private
       # Internal, used by {Parser}
-      def <<(node)
+      def <<(node) # rubocop:disable Metrics/PerceivedComplexity
         if node.is_a?(Array)
           node.each { |n| self << n }
-        elsif last && last.can_merge?(node)
+        elsif last&.can_merge?(node)
           last.merge!(node)
         else
           return if !node || node.empty?
+
           node = Text.new(node) if node.is_a?(String)
           super
         end
@@ -173,7 +215,9 @@ module Infoboxer
       # @private
       # Internal, used by {Parser}
       def flow_templates
-        make_nodes(map { |n| n.is_a?(Paragraph) ? n.to_templates? : n })
+        # TODO: will it be better?..
+        # make_nodes(map { |n| n.is_a?(Paragraph) ? n.to_templates? : n })
+        self
       end
 
       private

data/lib/infoboxer/tree/paragraphs.rb

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

data/lib/infoboxer/tree/ref.rb

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

data/lib/infoboxer/tree/table.rb

@@ -1,4 +1,4 @@
-# encoding: utf-8
+# frozen_string_literal: true
 
 require 'terminal-table'
 
@@ -26,13 +26,13 @@ module Infoboxer
       #
       # FIXME: it can easily be several table heading rows
       def heading_row
-        rows.first if rows.first && rows.first.children.all? { |c| c.is_a?(TableHeading) }
+        rows.first if rows.first&.children&.all? { |c| c.is_a?(TableHeading) }
       end
 
       # For now, returns all table rows except {#heading_row}
       def body_rows
-        if rows.first && rows.first.children.all? { |c| c.is_a?(TableHeading) }
-          rows[1..-1]
+        if rows.first&.children&.all? { |c| c.is_a?(TableHeading) }
+          rows[1..]
         else
           rows
         end

data/lib/infoboxer/tree/template.rb

@@ -1,4 +1,4 @@
-# encoding: utf-8
+# frozen_string_literal: true
 
 require_relative 'linkable'
 
@@ -22,6 +22,10 @@ module Infoboxer
         false
       end
 
+      def named?
+        name !~ /^\d+$/
+      end
+
       protected
 
       def descr
@@ -110,12 +114,17 @@ module Infoboxer
       alias_method :variables, :children
 
       def initialize(name, variables = Nodes[])
-        super(variables, extract_params(variables))
+        super(variables, **extract_params(variables))
         @name = name
       end
 
       def text
-        ''
+        res = unnamed_variables.map(&:text).join('|')
+        res.empty? ? '' : "{#{name}:#{res}}"
+      end
+
+      def unwrap
+        unnamed_variables.flat_map(&:children).unwrap
       end
 
       # See {Node#to_tree}
@@ -139,7 +148,11 @@ module Infoboxer
       #
       # @return [Nodes<Var>]
       def unnamed_variables
-        variables.find(name: /^\d+$/)
+        variables.reject(&:named?)
+      end
+
+      def named_variables
+        variables.select(&:named?)
       end
 
       # Fetches template variable(s) by name(s) or patterns.
@@ -242,7 +255,7 @@ module Infoboxer
       def extract_params(vars)
         vars
           .select { |v| v.children.count == 1 && v.children.first.is_a?(Text) }
-          .map { |v| [v.name, v.children.first.raw_text] }.to_h
+          .map { |v| [v.name.to_sym, v.children.first.raw_text] }.to_h
       end
 
       def inspect_variables(depth)

data/lib/infoboxer/tree/text.rb

@@ -1,4 +1,4 @@
-# encoding: utf-8
+# frozen_string_literal: true
 
 module Infoboxer
   module Tree
@@ -15,9 +15,9 @@ module Infoboxer
       # Text fragment without decodint of HTML entities.
       attr_accessor :raw_text
 
-      def initialize(text, params = {})
-        super(params)
-        @raw_text = text
+      def initialize(text, **params)
+        super(**params)
+        @raw_text = +text
       end
 
       # See {Node#text}
@@ -39,13 +39,13 @@ module Infoboxer
       # @private
       # Internal, used by {Parser}
       def merge!(other)
-        if other.is_a?(String)
-          @raw_text << other
-        elsif other.is_a?(Text)
-          @raw_text << other.raw_text
-        else
-          fail("Not mergeable into text: #{other.inspect}")
-        end
+        @raw_text <<
+          case other
+          when String then other
+          when Text then other.raw_text
+          else
+            fail("Not mergeable into text: #{other.inspect}")
+          end
       end
 
       # @private

data/lib/infoboxer/tree/wikilink.rb

@@ -1,4 +1,4 @@
-# encoding: utf-8
+# frozen_string_literal: true
 
 require_relative 'linkable'
 
@@ -12,14 +12,23 @@ module Infoboxer
     # Note, that Wikilink is {Linkable}, so you can {Linkable#follow #follow}
     # it to obtain linked pages.
     class Wikilink < Link
-      def initialize(*)
-        super
-        parse_link!
+      def initialize(link, label = nil, namespace: nil, interwiki: nil)
+        super(link, label, namespace: namespace, interwiki: interwiki)
+        @namespace = namespace || ''
+        @interwiki = interwiki
+        parse_name!
       end
 
       # "Clean" wikilink name, for ex., `Cities` for `[Category:Cities]`
       attr_reader :name
 
+      # Interwiki identifier. For example, `[[wikt:Argentina]]`
+      # will have `"Argentina"` as its {#name} and `"wikt"` (wiktionary) as an
+      # interwiki. TODO: how to use it.
+      #
+      # See [Wikipedia docs](https://en.wikipedia.org/wiki/Help:Interwiki_linking) for details.
+      attr_reader :interwiki
+
       # Wikilink namespace, `Category` for `[Category:Cities]`, empty
       # string (not `nil`!) for just `[Cities]`
       attr_reader :namespace
@@ -46,10 +55,8 @@ module Infoboxer
 
       private
 
-      def parse_link!
-        @name, @namespace = link.split(':', 2).reverse
-        @namespace ||= ''
-
+      def parse_name!
+        @name = namespace.empty? ? link : link.sub(/^#{namespace}:/, '')
         @name, @anchor = @name.split('#', 2)
         @anchor ||= ''
 
@@ -68,6 +75,7 @@ module Infoboxer
 
         return unless children.count == 1 &&
                       children.first.is_a?(Text) && children.first.raw_text.empty?
+
         children.first.raw_text = @topic
       end
     end

data/lib/infoboxer/version.rb

@@ -1,8 +1,9 @@
-# encoding: utf-8
+# frozen_string_literal: true
 
 module Infoboxer
   MAJOR = 0
-  MINOR = 3
+  MINOR = 4
   PATCH = 0
-  VERSION = [MAJOR, MINOR, PATCH].join('.')
+  PRE = nil
+  VERSION = [MAJOR, MINOR, PATCH, PRE].compact.join('.')
 end

data/lib/infoboxer/wiki_path.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module Infoboxer
   # @private
   class WikiPath
@@ -36,7 +38,7 @@ module Infoboxer
           attrs[attr.to_sym] = process_value(value)
         end
         res = op == '//' ? {op: :lookup} : {}
-        res[:type] = type.gsub(/(?:^|_)([a-z])/, &:upcase).tr('_', '').to_sym unless type.empty?
+        res[:type] = process_type(type) unless type.empty?
         res.merge(attrs) # TODO: raise if empty selector
       end
 
@@ -51,6 +53,15 @@ module Infoboxer
         end
       end
 
+      def process_type(type)
+        type.gsub(/(?:^|_)([a-z])/, &:upcase).tr('_', '').to_sym
+            .tap { |t| valid_type?(t) or fail(ParseError, "Unrecognized node type: #{type}") }
+      end
+
+      def valid_type?(t)
+        t == :Section || Infoboxer::Tree.const_defined?(t)
+      end
+
       def unexpected(scanner, expected)
         place = scanner.eos? ? 'end of pattern' : scanner.rest.inspect
         fail ParseError, "Unexpected #{place}, expecting #{expected}"