infoboxer 0.1.2.1 → 0.2.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 5a8db6382e2ffef87c2685aa1a6ef9ad37f3b57b
-  data.tar.gz: c4a622a22d275e1098f6957fad0ec0de13a88001
+  metadata.gz: d3081274989109208504796d1357e7ab78dd8981
+  data.tar.gz: 255f2ffa01c283fd11cbe1a1b308223d276c3b22
 SHA512:
-  metadata.gz: 15945396bc46ea107235d2e1f2c9b496a81b6cc28fe6d6ecfc2c9424ded26cb8cf4a3ec06cc8c63a52b4e0839ee5e7deac941f59784dcb2c1f3769dea468d3d0
-  data.tar.gz: 0dbce96e10c402c1676a4f5b694862e9fec9cff7d25606b5217ce3e31e64776bfa36ac31ee0c8e7964b32010ad93e9ad4453e605cbd4abcd9eff02350b58f35e
+  metadata.gz: 47ff1c7ac1f6e34ba4e5491cd7f5a6e180f18c02c4bf6061d08c6589ca3b66cd8ac1c600cc6e03dda244c3dd37a1986356e47d86f79602416e0eba021182fe00
+  data.tar.gz: c40d2bb3f4b2d336830d56e8b8cc2b126807022f409fe41fd63f0d229f139030b4ef9c18116f802016c3e33af6f4aba1f1766c03619d3e185cdce2b949d63bf6

data/CHANGELOG.md

@@ -1,5 +1,17 @@
 # Infoboxer's change log
 
+## 0.2.0 (2015-12-21)
+
+* MediaWiki backend changed to (our own handcrafted)
+  [mediawiktory](https://github.com/molybdenum-99/mediawiktory);
+* Added page lists fetching like `MediaWiki#category(categoryname)`,
+  `MediaWiki#search(search_phrase)`;
+* `MediaWiki#get` now can fetch any number of pages at once (it was only
+  50 in previous versions);
+* `bin/infoboxer` console added for quick experimenting;
+* `Template#to_h` added for quick information extraction;
+* many small bugfixes and echancements. 
+
 ## 0.1.2.1 (2015-12-04)
 
 * Small bug with newlines in templates fixed.
@@ -22,6 +34,6 @@ Basically, preparing for wider release!
 
 ## 0.1.0 (2015-08-07)
 
-Initial (ok, I know it's typically called 0.1.1, but here's work of
+Initial (ok, I know it's typically called 0.0.1, but here's work of
 three monthes, numerous documentations and examples and so on... so, let
 it be 0.1.0).

data/README.md

@@ -4,6 +4,7 @@
 [![Build Status](https://travis-ci.org/molybdenum-99/infoboxer.svg?branch=master)](https://travis-ci.org/molybdenum-99/infoboxer)
 [![Coverage Status](https://coveralls.io/repos/molybdenum-99/infoboxer/badge.svg?branch=master&service=github)](https://coveralls.io/github/molybdenum-99/infoboxer?branch=master)
 [![Code Climate](https://codeclimate.com/github/molybdenum-99/infoboxer/badges/gpa.svg)](https://codeclimate.com/github/molybdenum-99/infoboxer)
+[![Molybdenum-99 Gitter](https://badges.gitter.im/molybdenum-99.png)](https://gitter.im/molybdenum-99)
 
 **Infoboxer** is pure-Ruby Wikipedia (and generic MediaWiki) client and
 parser, targeting information extraction (hence the name).
@@ -97,6 +98,25 @@ See [Navigation shortcuts](https://github.com/molybdenum-99/infoboxer/wiki/Navig
 
 To put it all in one piece, also take a look at [Data extraction tips and tricks](https://github.com/molybdenum-99/infoboxer/wiki/Tips-and-tricks).
 
+### infoboxer executable
+
+Just try `infoboxer` command.
+
+Without any options, it starts IRB session with infoboxer required and
+included into main namespace.
+
+With `-w` option, it provides a shortcut to MediaWiki instance you want.
+Like this:
+
+```
+$ infoboxer -w https://en.wikipedia.org/w/api.php
+> get('Argentina')
+ => #<Page(title: "Argentina", url: "https://en.wikipedia.org/wiki/Argentina"): ....
+```
+
+You can also use shortcuts like `infoboxer -w wikipedia` for common
+wikies (and, just for fun, `infoboxer -wikipedia` also).
+
 ## Advanced topics
 
 * [Reasons](https://github.com/molybdenum-99/infoboxer/wiki/Reasons) for
@@ -114,9 +134,10 @@ To put it all in one piece, also take a look at [Data extraction tips and tricks
 
 ## Compatibility
 
-As of now, Infoboxer reported to be compatible with any MRI Ruby since 1.9.3.
-In Travis-CI tests, JRuby is failing due to bug in old Java 7/Java 8 SSL
-certificate support ([see here](https://github.com/jruby/jruby/issues/2599)),
+As of now, Infoboxer reported to be compatible with any MRI Ruby since 2.0.0
+(1.9.3 previously, dropped since Infoboxer 0.2.0). In Travis-CI tests,
+JRuby is failing due to bug in old Java 7/Java 8 SSL certificate support
+([see here](https://github.com/jruby/jruby/issues/2599)),
 and Rubinius failing 3 specs of 500 by mystery, which is uninvestigated yet.
 
 Therefore, those Ruby versions are excluded from Travis config, though,
@@ -129,10 +150,10 @@ they may still work for you.
   * **NB**: ↑ this is "current version" link, but RubyDoc.info unfortunately
     sometimes fails to update it to really _current_; in case you feel
     something seriously underdocumented, please-please look at
-    [0.1.2 docs](http://www.rubydoc.info/gems/infoboxer/0.1.2).
+    [0.2.0 docs](http://www.rubydoc.info/gems/infoboxer/0.2.0).
 * [Contributing](https://github.com/molybdenum-99/infoboxer/wiki/Contributing)
 * [Roadmap](https://github.com/molybdenum-99/infoboxer/wiki/Roadmap)
 
 ## License
 
-MIT.
+[MIT](https://github.com/molybdenum-99/infoboxer/blob/master/LICENSE.txt).

data/bin/infoboxer

@@ -0,0 +1,45 @@
+#!/usr/bin/env ruby
+require 'rubygems'
+require 'bundler/setup'
+require 'infoboxer'
+
+include Infoboxer
+
+require 'optparse'
+
+wiki_url = nil
+
+OptionParser.new do |opts|
+  opts.banner = "Usage: bin/infoboxer [-w wiki_api_url]"
+
+  opts.on("-w", "--wiki WIKI_API_URL",
+              "Make wiki by WIKI_API_URL a default wiki, and use it with just get('Pagename')") do |w|
+    wiki_url = w
+  end
+end.parse!
+
+if wiki_url
+  if wiki_url =~ /^[a-z]+$/
+    wiki_url = case
+    when domain = Infoboxer::WIKIMEDIA_PROJECTS[wiki_url.to_sym]
+      "https://en.#{domain}/w/api.php"
+    when domain = Infoboxer::WIKIMEDIA_PROJECTS[('w' + wiki_url).to_sym]
+      "https://en.#{domain}/w/api.php"
+    else
+      fail("Unidentified wiki: #{wiki_url}")
+    end
+  end
+  
+  DEFAULT_WIKI = Infoboxer.wiki(wiki_url)
+  puts "Default Wiki selected: #{wiki_url}.\nNow you can use `get('Pagename')`, `category('Categoryname')` and so on.\n\n"
+  [:raw, :get, :category, :search, :prefixsearch].each do |m|
+    define_method(m){|*arg|
+      DEFAULT_WIKI.send(m, *arg)
+    }
+  end
+end
+
+require 'irb'
+ARGV.shift until ARGV.empty?
+IRB.start
+

data/infoboxer.gemspec

@@ -29,7 +29,7 @@ Gem::Specification.new do |s|
 
   s.add_dependency 'htmlentities'
   s.add_dependency 'procme'
-  s.add_dependency 'rest-client'
+  s.add_dependency 'mediawiktory', '>= 0.0.2'
   s.add_dependency 'addressable'
   s.add_dependency 'terminal-table'
   s.add_dependency 'backports'

data/lib/infoboxer/definitions/en.wikipedia.org.rb

@@ -24,7 +24,6 @@ module Infoboxer
         '!((' => '[[',
         '!-' => '|-',
         '!:' => ':',
-        '&' => '&',
         "'" => " '",
         "''" => '″',
         "'s" => "'‍s",

data/lib/infoboxer/media_wiki/page.rb

@@ -7,15 +7,19 @@ module Infoboxer
     # Alongside with document tree structure, knows document's title as
     # represented by MediaWiki and human (non-API) URL.
     class Page < Tree::Document
-      def initialize(client, children, raw)
-        @client = client
-        super(children, raw)
+      def initialize(client, children, source)
+        @client, @source = client, source
+        super(children, title: source.title, url: source.fullurl)
       end
 
       # Instance of {MediaWiki} which this page was received from
       # @return {MediaWiki}
       attr_reader :client
 
+      # Instance of MediaWiktory::Page class with source data
+      # @return {MediaWiktory::Page}
+      attr_reader :source
+
       # @!attribute [r] title
       #   Page title.
       #   @return [String]
@@ -24,11 +28,15 @@ module Infoboxer
       #   Page friendly URL.
       #   @return [String]
 
-      def_readers :title, :url, :traits
+      def_readers :title, :url
+
+      def traits
+        client.traits
+      end
 
       private
 
-      PARAMS_TO_INSPECT = [:url, :title, :domain]
+      PARAMS_TO_INSPECT = [:url, :title] #, :domain]
 
       def show_params
         super(params.select{|k, v| PARAMS_TO_INSPECT.include?(k)})

data/lib/infoboxer/media_wiki/traits.rb

@@ -68,14 +68,14 @@ module Infoboxer
 
       def initialize(options = {})
         @options = options
-        @file_prefix = [DEFAULTS[:file_prefix], options.delete(:file_prefix)].
+        @file_namespace = [DEFAULTS[:file_namespace], namespace_aliases(options, 'File')].
           flatten.compact.uniq
-        @category_prefix = [DEFAULTS[:category_prefix], options.delete(:category_prefix)].
+        @category_namespace = [DEFAULTS[:category_namespace], namespace_aliases(options, 'Category')].
           flatten.compact.uniq
       end
 
       # @private
-      attr_reader :file_prefix, :category_prefix
+      attr_reader :file_namespace, :category_namespace
 
       # @private
       def templates
@@ -84,9 +84,15 @@ module Infoboxer
 
       private
 
+      def namespace_aliases(options, canonical)
+        namespace = (options[:namespaces] || []).detect{|v| v.canonical == canonical}
+        return nil unless namespace
+        [namespace['*'], *namespace.aliases]
+      end
+
       DEFAULTS = {
-        file_prefix: 'File',
-        category_prefix: 'Category'
+        file_namespace: 'File',
+        category_namespace: 'Category'
       }
 
     end

data/lib/infoboxer/media_wiki.rb

@@ -1,6 +1,7 @@
 # encoding: utf-8
-require 'rest-client'
-require 'json'
+#require 'rest-client'
+#require 'json'
+require 'mediawiktory'
 require 'addressable/uri'
 
 require_relative 'media_wiki/traits'
@@ -36,7 +37,7 @@ module Infoboxer
       attr_accessor :user_agent
     end
 
-    attr_reader :api_base_url
+    attr_reader :api_base_url, :traits
 
     # Creating new MediaWiki client. {Infoboxer.wiki} provides shortcut
     # for it, as well as shortcuts for some well-known wikis, like
@@ -49,7 +50,8 @@ module Infoboxer
     #   * `:user_agent` (also aliased as `:ua`) -- custom User-Agent header.
     def initialize(api_base_url, options = {})
       @api_base_url = Addressable::URI.parse(api_base_url)
-      @resource = RestClient::Resource.new(api_base_url, headers: headers(options))
+      @client = MediaWiktory::Client.new(api_base_url, user_agent: user_agent(options))
+      @traits = Traits.get(@api_base_url.host, namespaces: extract_namespaces)
     end
 
     # Receive "raw" data from Wikipedia (without parsing or wrapping in
@@ -57,18 +59,22 @@ module Infoboxer
     #
     # @return [Array<Hash>]
     def raw(*titles)
-      postprocess @resource.get(
-        params: DEFAULT_PARAMS.merge(titles: titles.join('|'))
-      )
+      titles.each_slice(50).map{|part|
+        @client.query.
+          titles(*part).
+          prop(revisions: {prop: :content}, info: {prop: :url}).
+          redirects(true). # FIXME: should be done transparently by MediaWiktory?
+          perform.pages
+      }.inject(:concat) # somehow flatten(1) fails!
     end
 
-    # Receive list of parsed wikipedia pages for list of titles provided.
+    # Receive list of parsed MediaWiki pages for list of titles provided.
     # All pages are received with single query to MediaWiki API.
     #
-    # **NB**: currently, if you are requesting more than 50 titles at
-    # once (MediaWiki limitation for single request), Infoboxer will
-    # **not** try to get other pages with subsequent queries. This will
-    # be fixed in future.
+    # **NB**: if you are requesting more than 50 titles at once
+    # (MediaWiki limitation for single request), Infoboxer will do as
+    # many queries as necessary to extract them all (it will be like
+    # `(titles.count / 50.0).ceil` requests)
     #
     # @return [Tree::Nodes<Page>] array of parsed pages. Notes:
     #   * if you call `get` with only one title, one page will be
@@ -87,76 +93,118 @@ module Infoboxer
     #     NotFound.
     #
     def get(*titles)
-      pages = raw(*titles).reject{|raw| raw[:content].nil?}.
+      pages = raw(*titles).
+        tap{|pages| pages.detect(&:invalid?).tap{|i| i && fail(i.raw.invalidreason)}}.
+        select(&:exists?).
         map{|raw|
-          traits = Traits.get(@api_base_url.host, extract_traits(raw))
-          
           Page.new(self,
-            Parser.paragraphs(raw[:content], traits),
-            raw.merge(traits: traits))
+            Parser.paragraphs(raw.content, traits),
+            raw)
         }
       titles.count == 1 ? pages.first : Tree::Nodes[*pages]
     end
 
-    private
+    # Receive list of parsed MediaWiki pages from specified category.
+    #
+    # **NB**: currently, this API **always** fetches all pages from
+    # category, there is no option to "take first 20 pages". Pages are
+    # fetched in 50-page batches, then parsed. So, for large category
+    # it can really take a while to fetch all pages.
+    #
+    # @param title Category title. You can use namespaceless title (like
+    #     `"Countries in South America"`), title with namespace (like 
+    #     `"Category:Countries in South America"`) or title with local
+    #     namespace (like `"Catégorie:Argentine"` for French Wikipedia)
+    #
+    # @return [Tree::Nodes<Page>] array of parsed pages.
+    #
+    def category(title)
+      title = normalize_category_title(title)
+      
+      list(categorymembers: {title: title, limit: 50})
+    end
 
-    # @private
-    PROP = [
-      'revisions',    # to extract content of the page
-      'info',         # to extract page canonical url
-      'categories',   # to extract default category prefix
-      'images'        # to extract default media prefix
-    ].join('|')
-
-    # @private
-    DEFAULT_PARAMS = {
-      action:    :query,
-      format:    :json,
-      redirects: true,
-
-      prop:      PROP,
-      rvprop:    :content,
-      inprop:    :url,
-    }
-
-    def headers(options)
-      {'User-Agent' => options[:user_agent] || options[:ua] || self.class.user_agent || UA}
+    # Receive list of parsed MediaWiki pages for provided search query.
+    # See [MediaWiki API docs](https://www.mediawiki.org/w/api.php?action=help&modules=query%2Bsearch)
+    # for details.
+    #
+    # **NB**: currently, this API **always** fetches all pages from
+    # category, there is no option to "take first 20 pages". Pages are
+    # fetched in 50-page batches, then parsed. So, for large category
+    # it can really take a while to fetch all pages.
+    #
+    # @param query Search query. For old installations, look at
+    #     https://www.mediawiki.org/wiki/Help:Searching
+    #     for search syntax. For new ones (including Wikipedia), see at
+    #     https://www.mediawiki.org/wiki/Help:CirrusSearch.
+    #
+    # @return [Tree::Nodes<Page>] array of parsed pages.
+    #
+    def search(query)
+      list(search: {search: query, limit: 50})
+    end
+
+    # Receive list of parsed MediaWiki pages with titles startin from prefix.
+    # See [MediaWiki API docs](https://www.mediawiki.org/w/api.php?action=help&modules=query%2Bprefixsearch)
+    # for details.
+    #
+    # **NB**: currently, this API **always** fetches all pages from
+    # category, there is no option to "take first 20 pages". Pages are
+    # fetched in 50-page batches, then parsed. So, for large category
+    # it can really take a while to fetch all pages.
+    #
+    # @param prefix page title prefix.
+    #
+    # @return [Tree::Nodes<Page>] array of parsed pages.
+    #
+    def prefixsearch(prefix)
+      list(prefixsearch: {search: prefix, limit: 100})
     end
 
-    def extract_traits(raw)
-      raw.select{|k, v| [:file_prefix, :category_prefix].include?(k)}
+    def inspect
+      "#<#{self.class}(#{@api_base_url.host})>"
     end
 
-    def guess_traits(pages)
-      categories = pages.map{|p| p['categories']}.compact.flatten
-      images = pages.map{|p| p['images']}.compact.flatten
-      {
-        file_prefix: images.map{|i| i['title'].scan(/^([^:]+):/)}.flatten.uniq,
-        category_prefix: categories.map{|i| i['title'].scan(/^([^:]+):/)}.flatten.uniq,
-      }
+    private
+
+    def list(query)
+      response = @client.query.
+        generator(query).
+        prop(revisions: {prop: :content}, info: {prop: :url}).
+        redirects(true). # FIXME: should be done transparently by MediaWiktory?
+        perform
+
+      response.continue! while response.continue?
+
+      pages = response.pages.select(&:exists?).
+        map{|raw|
+          Page.new(self,
+            Parser.paragraphs(raw.content, traits),
+            raw)
+        }
+
+      Tree::Nodes[*pages]
     end
 
-    def postprocess(response)
-      pages = JSON.parse(response)['query']['pages']
-      traits = guess_traits(pages.values)
+    def normalize_category_title(title)
+      # FIXME: shouldn't it go to MediaWiktory?..
+      namespace, titl = title.include?(':') ? title.split(':', 2) : [nil, title]
+      namespace, titl = nil, title unless traits.category_namespace.include?(namespace)
       
-      pages.map{|id, data|
-        if id.to_i < 0
-          {
-            title: data['title'],
-            content: nil,
-            not_found: true
-          }
-        else
-          {
-            title: data['title'],
-            content: data['revisions'].first['*'],
-            url: data['fullurl'],
-          }.merge(traits)
-        end
+      namespace ||= traits.category_namespace.first
+      [namespace, titl].join(':')
+    end
+
+    def user_agent(options)
+      options[:user_agent] || options[:ua] || self.class.user_agent || UA
+    end
+
+    def extract_namespaces
+      siteinfo = @client.query.meta(siteinfo: {prop: [:namespaces, :namespacealiases]}).perform
+      siteinfo.raw.query.namespaces.map{|_, namespace|
+        aliases = siteinfo.raw.query.namespacealiases.select{|a| a.id == namespace.id}.map{|a| a['*']}
+        namespace.merge(aliases: aliases)
       }
-    rescue JSON::ParserError
-      fail RuntimeError, "Not a JSON response, seems there's not a MediaWiki API: #{@api_base_url}"
     end
   end
 end

data/lib/infoboxer/navigation/shortcuts.rb

@@ -118,7 +118,7 @@ module Infoboxer
         #
         # @return {Tree::Nodes}
         def categories
-          lookup(Tree::Wikilink, namespace: /^#{ensure_traits.category_prefix.join('|')}$/)
+          lookup(Tree::Wikilink, namespace: /^#{ensure_traits.category_namespace.join('|')}$/)
         end
 
         # As users accustomed to have only one infobox on a page

data/lib/infoboxer/parser/context.rb

@@ -1,4 +1,6 @@
 # encoding: utf-8
+require 'strscan'
+
 module Infoboxer
   class Parser
     class Context
@@ -86,11 +88,23 @@ module Infoboxer
         res
       end
 
+      def push_eol_sign(re)
+        @inline_eol_sign = re
+      end
+
+      def pop_eol_sign
+        @inline_eol_sign = nil
+      end
+
+      attr_reader :inline_eol_sign
+
       def inline_eol?(exclude = nil)
         # not using StringScanner#check, as it will change #matched value
         eol? ||
-          (current =~ %r[^(</ref>|}})] &&
-            (!exclude || $1 !~ exclude)) # FIXME: ugly, but no idea of prettier solution
+          (
+            (current =~ %r[^(</ref>|}})] || @inline_eol_sign && current =~ @inline_eol_sign) && 
+            (!exclude || $1 !~ exclude)
+          ) # FIXME: ugly, but no idea of prettier solution
       end
 
       def scan_continued_until(re, leave_pattern = false)

data/lib/infoboxer/parser/image.rb

@@ -5,7 +5,7 @@ module Infoboxer
       include Tree
       
       def image
-        @context.skip(re.file_prefix) or
+        @context.skip(re.file_namespace) or
           @context.fail!("Something went wrong: it's not image?")
 
         path = @context.scan_until(/\||\]\]/)

data/lib/infoboxer/parser/inline.rb

@@ -32,7 +32,12 @@ module Infoboxer
       def short_inline(until_pattern = nil)
         nodes = Nodes[]
         guarded_loop do
-          chunk = @context.scan_until(re.short_inline_until_cache[until_pattern])
+          # FIXME: quick and UGLY IS HELL JUST TRYING TO MAKE THE SHIT WORK
+          if @context.inline_eol_sign
+            chunk = @context.scan_until(re.short_inline_until_cache_brackets[until_pattern])
+          else
+            chunk = @context.scan_until(re.short_inline_until_cache[until_pattern])
+          end
           nodes << chunk
 
           break if @context.matched_inline?(until_pattern)
@@ -82,7 +87,7 @@ module Infoboxer
           when "''"
             Italic.new(short_inline(/''/))
           when '[['
-            if @context.check(re.file_prefix)
+            if @context.check(re.file_namespace)
               image
             else
               wikilink
@@ -118,7 +123,11 @@ module Infoboxer
         # [http://www.example.org link name]
         def external_link(protocol)
           link = @context.scan_continued_until(/\s+|\]/)
-          caption = inline(/\]/) if @context.matched =~ /\s+/
+          if @context.matched =~ /\s+/
+            @context.push_eol_sign(/^\]/)
+            caption = short_inline(/\]/) 
+            @context.pop_eol_sign
+          end
           ExternalLink.new(protocol + link, caption)
         end
 

data/lib/infoboxer/parser/template.rb

@@ -4,8 +4,8 @@ module Infoboxer
     module Template
       include Tree
       
-      # NB: here we are not distingish templates like {{Infobox|variable}}
-      # and "magic words" like {{formatnum:123}}
+      # NB: here we are not distingish templates like `{{Infobox|variable}}`
+      # and "magic words" like `{{formatnum:123}}`
       # Just calling all of them "templates". This behaviour will change
       # in future, I presume
       # More about magic words: https://www.mediawiki.org/wiki/Help:Magic_words
@@ -29,6 +29,7 @@ module Infoboxer
             @context.skip(/\s*=\s*/)
           else
             name = num
+            num += 1
           end
 
           value = long_inline(/\||}}/)
@@ -38,8 +39,6 @@ module Infoboxer
 
           break if @context.eat_matched?('}}')
           @context.eof? and @context.fail!("Unexpected break of template variables: #{res}")
-
-          num += 1
         end
         res
       end

data/lib/infoboxer/parser/util.rb

@@ -16,20 +16,31 @@ module Infoboxer
 
       INLINE_EOL = %r[(?=   # if we have ahead... (not scanned, just checked
         </ref>        |     # <ref> closed
-        }}                  # or template closed
+        }}            
+      )]x
+
+      INLINE_EOL_BR = %r[(?=   # if we have ahead... (not scanned, just checked
+        </ref>        |     # <ref> closed
+        }}            |     # or template closed
+        (?<!\])\](?!\])     # or ext.link closed,
+                            # the madness with look-ahead/behind means "match single bracket but not double"
       )]x
 
 
       def make_regexps
         {
-          file_prefix: /(#{@context.traits.file_prefix.join('|')}):/,
+          file_namespace: /(#{@context.traits.file_namespace.join('|')}):/,
           formatting: FORMATTING,
           inline_until_cache: Hash.new{|h, r|
             h[r] = Regexp.union(*[r, FORMATTING, /$/].compact.uniq)
           },
           short_inline_until_cache: Hash.new{|h, r|
             h[r] = Regexp.union(*[r, INLINE_EOL, FORMATTING, /$/].compact.uniq)
+          },
+          short_inline_until_cache_brackets: Hash.new{|h, r|
+            h[r] = Regexp.union(*[r, INLINE_EOL_BR, FORMATTING, /$/].compact.uniq)
           }
+          
         }
       end
 
@@ -46,7 +57,7 @@ module Infoboxer
             scan.skip(/=\s*/)
             q = scan.scan(/['"]/)
             if q
-              value = scan.scan_until(/#{q}/).sub(q, '')
+              value = scan.scan_until(/#{q}|$/).sub(q, '')
             else
               value = scan.scan_until(/\s|$/)
             end

data/lib/infoboxer/tree/image.rb

@@ -43,7 +43,7 @@ module Infoboxer
         super(level) +
           if caption && !caption.empty?
             indent(level+1) + "caption:\n" +
-              caption.map(&call(to_tree: level+2)).join
+              caption.children.map(&call(to_tree: level+2)).join
           else
             ''
           end