infoboxer 0.3.0 → 0.4.0

data/lib/infoboxer/media_wiki/page.rb

@@ -1,4 +1,4 @@
-# encoding: utf-8
+# frozen_string_literal: true
 
 module Infoboxer
   class MediaWiki
@@ -35,6 +35,15 @@ module Infoboxer
         client.traits
       end
 
+      # FIXME: take from siteinfo!
+      def namespace
+        Traits::STANDARD_NAMESPACES[source.fetch('ns') + 2] # Media = -2, Specia = -1, Main = 0
+      end
+
+      def category?
+        namespace == 'Category'
+      end
+
       private
 
       PARAMS_TO_INSPECT = %i[url title].freeze

data/lib/infoboxer/media_wiki/traits.rb

@@ -1,4 +1,4 @@
-# encoding: utf-8
+# frozen_string_literal: true
 
 module Infoboxer
   class MediaWiki
@@ -34,9 +34,8 @@ module Infoboxer
         end
 
         # @private
-        def get(domain, options = {})
-          cls = Traits.domains[domain]
-          cls ? cls.new(options) : Traits.new(options)
+        def get(domain, site_info = {})
+          (Traits.domains[domain] || Traits).new(site_info)
         end
 
         # @private
@@ -60,7 +59,7 @@ module Infoboxer
         # [English Wikipedia traits](https://github.com/molybdenum-99/infoboxer/blob/master/lib/infoboxer/definitions/en.wikipedia.org.rb)
         # for example implementation.
         def for(domain, &block)
-          Traits.domains[domain].tap { |c| c && c.instance_eval(&block) } ||
+          Traits.domains[domain]&.instance_eval(&block) ||
             Class.new(self, &block).domain(domain)
         end
 
@@ -68,18 +67,27 @@ module Infoboxer
         alias_method :default, :new
       end
 
-      def initialize(options = {})
-        @options = options
-        @file_namespace =
-          [DEFAULTS[:file_namespace], namespace_aliases(options, 'File')]
-          .flatten.compact.uniq
-        @category_namespace =
-          [DEFAULTS[:category_namespace], namespace_aliases(options, 'Category')]
-          .flatten.compact.uniq
+      def initialize(site_info = {})
+        @site_info = site_info
+      end
+
+      def namespace?(prefix)
+        known_namespaces.include?(prefix)
+      end
+
+      def interwiki?(prefix)
+        known_interwikis.key?(prefix)
       end
 
       # @private
-      attr_reader :file_namespace, :category_namespace
+      def file_namespace
+        @file_namespace ||= ns_aliases('File')
+      end
+
+      # @private
+      def category_namespace
+        @category_namespace ||= ns_aliases('Category')
+      end
 
       # @private
       def templates
@@ -88,16 +96,55 @@ module Infoboxer
 
       private
 
-      def namespace_aliases(options, canonical)
-        namespace = (options[:namespaces] || []).detect { |v| v['canonical'] == canonical }
-        return nil unless namespace
-        [namespace['*'], *namespace['aliases']]
+      def known_namespaces
+        @known_namespaces ||=
+          if @site_info.empty?
+            STANDARD_NAMESPACES
+          else
+            (@site_info['namespaces'].values + @site_info['namespacealiases']).map { |n| n['*'] }
+          end
+      end
+
+      def known_interwikis
+        @known_interwikis ||=
+          if @site_info.empty?
+            {}
+          else
+            @site_info['interwikimap'].map { |iw| [iw['prefix'], iw] }.to_h
+          end
+      end
+
+      def ns_aliases(base)
+        return [base] if @site_info.empty?
+
+        main = @site_info['namespaces'].values.detect { |n| n['canonical'] == base }
+        [base, main['*']] +
+          @site_info['namespacealiases']
+          .select { |a| a['id'] == main['id'] }.flat_map { |n| n['*'] }
+          .compact.uniq
       end
 
-      DEFAULTS = {
-        file_namespace: 'File',
-        category_namespace: 'Category'
-      }.freeze
+      # See https://www.mediawiki.org/wiki/Help:Namespaces#Standard_namespaces
+      STANDARD_NAMESPACES = [
+        'Media',            # Direct linking to media files.
+        'Special',          # Special (non-editable) pages.
+        '',                 # (Main)
+        'Talk',             # Article discussion.
+        'User',             #
+        'User talk',        #
+        'Project',          # Meta-discussions related to the operation and development of the wiki.
+        'Project talk',     #
+        'File',             # Metadata for images, videos, sound files and other media.
+        'File talk',        #
+        'MediaWiki',        # System messages and other important content.
+        'MediaWiki talk',   #
+        'Template',         # Templates: blocks of text or wikicode that are intended to be transcluded.
+        'Template talk',    #
+        'Help',             # Help files, instructions and "how-to" guides.
+        'Help talk',        #
+        'Category',         # Categories: dynamic lists of other pages.
+        'Category talk',    #
+      ].freeze
     end
   end
 end

data/lib/infoboxer/navigation.rb

@@ -1,4 +1,4 @@
-# encoding: utf-8
+# frozen_string_literal: true
 
 module Infoboxer
   # Navigation is one of the things Infoboxer is proud about. It tries
@@ -96,5 +96,11 @@ module Infoboxer
     class Tree::Document
       include Navigation::Sections::Container
     end
+
+    module Helpers
+      def W(*arg, &block) # rubocop:disable Naming/MethodName
+        Lookup::Selector.new(*arg, &block)
+      end
+    end
   end
 end

data/lib/infoboxer/navigation/lookup.rb

@@ -1,4 +1,4 @@
-# encoding: utf-8
+# frozen_string_literal: true
 
 require_relative 'selector'
 
@@ -98,9 +98,13 @@ module Infoboxer
         #   Selects matching nodes from current node's siblings, which
         #   are above current node in parents children list.
 
+        # @!method lookup_prev_sibling(*selectors, &block)
+        #   Selects first matching nodes from current node's siblings, which
+        #   are above current node in parents children list.
+
         # Underscored version of {#matches?}
         def _matches?(selector)
-          selector.matches?(self)
+          selector === self
         end
 
         # Underscored version of {#lookup}
@@ -136,6 +140,11 @@ module Infoboxer
           prev_siblings._find(selector)
         end
 
+        # Underscored version of {#lookup_prev_sibling}
+        def _lookup_prev_sibling(selector)
+          prev_siblings.reverse.detect { |n| selector === n }
+        end
+
         # Underscored version of {#lookup_next_siblings}
         def _lookup_next_siblings(selector)
           next_siblings._find(selector)
@@ -146,14 +155,14 @@ module Infoboxer
           lookup lookup_children lookup_parents
           lookup_siblings
           lookup_next_siblings lookup_prev_siblings
+          lookup_prev_sibling
         ]
           .map { |sym| [sym, :"_#{sym}"] }
           .each do |sym, underscored|
-
-          define_method(sym) do |*args, &block|
-            send(underscored, Selector.new(*args, &block))
+            define_method(sym) do |*args, &block|
+              send(underscored, Selector.new(*args, &block))
+            end
           end
-        end
 
         # Checks if node has any parent matching selectors.
         def parent?(*selectors, &block)
@@ -209,7 +218,6 @@ module Infoboxer
           lookup_siblings
           lookup_next_siblings lookup_prev_siblings
         ].map { |sym| [sym, :"_#{sym}"] }.each do |sym, underscored|
-
           define_method(sym) do |*args, &block|
             send(underscored, Selector.new(*args, &block))
           end

data/lib/infoboxer/navigation/sections.rb

@@ -1,4 +1,4 @@
-# encoding: utf-8
+# frozen_string_literal: true
 
 module Infoboxer
   module Navigation
@@ -77,10 +77,23 @@ module Infoboxer
           when 1
             @sections.select { |s| names.first === s.heading.text_ }
           else
-            @sections.select { |s| names.first === s.heading.text_ }.sections(*names[1..-1])
+            @sections.select { |s| names.first === s.heading.text_ }.sections(*names[1..])
           end
         end
 
+        def subsections(*names)
+          sections = names.map { |name|
+            heading = lookup_children(:Heading, text_: name).first
+            next unless heading
+
+            body = heading.next_siblings
+                          .take_while { |n| !n.is_a?(Tree::Heading) || n.level > heading.level }
+
+            Section.new(heading, body)
+          }.compact
+          Tree::Nodes.new(sections)
+        end
+
         def lookup_children(*arg)
           if arg.include?(:Section)
             sections.find(*(arg - [:Section]))
@@ -94,6 +107,7 @@ module Infoboxer
         def make_sections
           res = Tree::Nodes[]
           return res if headings.empty?
+
           level = headings.first.level
 
           children
@@ -123,21 +137,25 @@ module Infoboxer
         #
         # @return {Tree::Nodes<Section>}
         def in_sections
-          main_node = parent.is_a?(Tree::Document) ? self : lookup_parents[-2]
+          return parent.in_sections unless parent.is_a?(Tree::Document)
+          return @in_sections if @in_sections
 
           heading =
-            if main_node.is_a?(Tree::Heading)
-              main_node.lookup_prev_siblings(Tree::Heading, level: main_node.level - 1).last
+            if is_a?(Tree::Heading)
+              lookup_prev_sibling(Tree::Heading, level: level - 1)
             else
-              main_node.lookup_prev_siblings(Tree::Heading).last
+              lookup_prev_sibling(Tree::Heading)
             end
-          return Tree::Nodes[] unless heading
+          unless heading
+            @in_sections = Tree::Nodes[]
+            return @in_sections
+          end
 
           body = heading.next_siblings
-                        .take_while { |n| !n.is_a?(Tree::Heading) || n.level < heading.level }
+                        .take_while { |n| !n.is_a?(Tree::Heading) || n.level > heading.level }
 
           section = Section.new(heading, body)
-          Tree::Nodes[section, *heading.in_sections]
+          @in_sections = Tree::Nodes[section, *heading.in_sections]
         end
       end
 

data/lib/infoboxer/navigation/selector.rb

@@ -1,4 +1,4 @@
-# encoding: utf-8
+# frozen_string_literal: true
 
 module Infoboxer
   module Navigation
@@ -10,7 +10,7 @@ module Infoboxer
         def initialize(*arg, &block)
           @arg = [arg, block].flatten.compact.map(&method(:sym_to_class))
           @arg.each do |a|
-            a.reject! { |_k, v| v.nil? } if a.is_a?(Hash)
+            a.compact! if a.is_a?(Hash)
           end
         end
 
@@ -24,8 +24,8 @@ module Infoboxer
           "#<Selector(#{@arg.map(&:to_s).join(', ')})>"
         end
 
-        def matches?(node)
-          @arg.all? { |a| arg_matches?(a, node) }
+        def ===(other)
+          @arg.all? { |a| arg_matches?(a, other) }
         end
 
         private
@@ -44,8 +44,8 @@ module Infoboxer
             check.call(node)
           when Hash
             check.all? { |attr, value|
-              node.respond_to?(attr) && value === node.send(attr) ||
-                node.params.key?(attr) && value === node.params[attr]
+              node.respond_to?(attr) && value_matches?(value, node.send(attr)) ||
+                node.params.key?(attr) && value_matches?(value, node.params[attr])
             }
           when Symbol
             node.respond_to?(check) && node.send(check)
@@ -53,6 +53,14 @@ module Infoboxer
             check === node
           end
         end
+
+        def value_matches?(matcher, value)
+          if matcher.is_a?(String) && value.is_a?(String)
+            matcher.casecmp(value).zero?
+          else
+            matcher === value
+          end
+        end
       end
     end
   end

data/lib/infoboxer/navigation/shortcuts.rb

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

data/lib/infoboxer/navigation/wikipath.rb

@@ -1,4 +1,4 @@
-# encoding: utf-8
+# frozen_string_literal: true
 
 require_relative '../wiki_path'
 

data/lib/infoboxer/parser.rb

@@ -1,4 +1,4 @@
-# encoding: utf-8
+# frozen_string_literal: true
 
 require 'ostruct'
 require 'logger'
@@ -54,7 +54,7 @@ module Infoboxer
     def initialize(context)
       @context = context
       @re = OpenStruct.new(make_regexps)
-      @logger = Logger.new(STDOUT).tap { |l| l.level = Logger::FATAL }
+      @logger = Logger.new($stdout).tap { |l| l.level = Logger::FATAL }
     end
 
     require_relative 'parser/inline'

data/lib/infoboxer/parser/context.rb

@@ -1,16 +1,16 @@
-# encoding: utf-8
+# frozen_string_literal: true
 
 require 'strscan'
 
 module Infoboxer
   class Parser
-    class Context
+    class Context # rubocop:disable Metrics/ClassLength
       attr_reader :lineno
       attr_reader :traits
 
       def initialize(text, traits = nil)
         @lines = text
-                 .gsub(/<!--.+?-->/m, '') # FIXME: will also kill comments inside <nowiki> tag
+                 .gsub(/<!--.*?-->/m, '') # FIXME: will also kill comments inside <nowiki> tag
                  .split(/[\r\n]/)
         @lineno = -1
         @traits = traits || MediaWiki::Traits.default
@@ -21,22 +21,23 @@ module Infoboxer
       attr_reader :next_lines
 
       def colno
-        @scanner && @scanner.pos || 0
+        @scanner&.pos || 0
       end
 
       def matched
-        @matched ||= @scanner && @scanner.matched
+        @matched ||= @scanner&.matched
       end
 
       # check which works only once
       def eat_matched?(str)
         return false unless matched == str
+
         @matched = 'DUMMY'
         true
       end
 
       def rest
-        @rest ||= @scanner && @scanner.rest
+        @rest ||= @scanner&.rest
       end
 
       alias_method :current, :rest
@@ -109,7 +110,7 @@ module Infoboxer
       end
 
       def scan_continued_until(re, leave_pattern = false)
-        res = ''
+        res = +''
 
         loop do
           chunk = _scan_until(re)
@@ -130,7 +131,13 @@ module Infoboxer
 
       # state inspection
       def matched_inline?(re)
-        re.nil? ? (matched.empty? && eol?) : matched =~ re
+        if re.nil?
+          matched.empty? && eol?
+        elsif re.inspect.start_with?('/^') # was it REALLY at the beginning of the line?..
+          @scanner.pos == matched.length && matched =~ re
+        else
+          matched =~ re
+        end
       end
 
       def matched?(re)
@@ -146,6 +153,13 @@ module Infoboxer
         fail(ParsingError, "#{text} at line #{@lineno}:\n\t#{current}")
       end
 
+      def unscan_matched!
+        return unless @matched
+
+        @scanner.pos -= @matched.size
+        @rest = nil
+      end
+
       private
 
       # we do hard use of #matched and #rest, its wiser to memoize them
@@ -163,7 +177,7 @@ module Infoboxer
       def shift(amount)
         @lineno += amount
         current = @lines[lineno]
-        @next_lines = @lines[(lineno + 1)..-1]
+        @next_lines = @lines[(lineno + 1)..]
         if current
           @scanner.string = current
           @rest = current