Add suitable links to tags overview page where possible

For some tags such as wikipedia, website, url, wikidata, phone, etc.
this will add one or more links to relevant web sites to the tags
overview tab.

README.md

@@ -28,12 +28,12 @@ It uses:
 * Ruby (must be at least 2.4)
 * [Sinatra web framework](http://www.sinatrarb.com/) and other ruby libraries
 * curl binary
-* sqlite3 binary (version 3.9 or above with FTS5 support)
+* sqlite3 binary (version 3.9 or above with FTS5 and regexp support)
 * Optional: Parallel bzip (pzbip2)
 
 Install the Debian/Ubuntu packages:
 ```sh
-$ sudo apt-get install curl sqlite3
+$ sudo apt-get install curl sqlite3 sqlite3-pcre
 $ sudo apt-get install ruby-passenger libapache2-mod-passenger
 ```
 

taginfo-config-example.json

@@ -48,7 +48,9 @@
         // web server config.
         "download_dir": "../../download",
         // Path to binaries (like 'taginfo-stats' and 'taginfo-similarity').
-        "bin_dir": "../../bin"
+        "bin_dir": "../../bin",
+        // Path to sqlite regexp extension.
+        "sqlite3_pcre_extension": "/usr/lib/sqlite3/pcre.so"
     },
     // "XAPI" and "JOSM" buttons on key/tag pages
     "xapi": {

web/i18n/en.yml

@@ -49,6 +49,7 @@ taginfo:
     key_combinations: Combinations
     comparison: Key/Tag Comparison
     overview: Overview
+    links: Links
     data_from: Data from
     data_from_description: Last update of taginfo database
     instance:

web/lib/sql.rb

@@ -28,6 +28,10 @@ def initialize
             @db = SQLite3::Database.new(filename, { :readonly => true })
             @db.results_as_hash = true
 
+            pcre_extension = TaginfoConfig.get('paths.sqlite3_pcre_extension')
+            if pcre_extension
+                @db.load_extension(pcre_extension)
+            end
             @db.execute('PRAGMA journal_mode = OFF')
             @db.execute('SELECT * FROM languages') do |row|
                 Language.new(row)

web/lib/taglinks.rb

@@ -0,0 +1,80 @@
+
+class TagLink
+
+    attr_reader :title, :url
+
+    def initialize(title, url)
+        @title = title
+        @url = url
+    end
+
+    def html
+        external_link('', @title, @url, true)
+    end
+
+end
+
+class TagMatch
+
+    attr_reader :regex
+
+    def initialize(regex, func)
+        @regex = regex
+        @func = func
+    end
+
+    def match(value)
+        value =~ @regex
+    end
+
+    def call(value)
+        @func.call(value)
+    end
+
+    def links(value)
+        if match(value)
+            return call(value).map{ |link| link.html }
+        end
+        []
+    end
+
+end
+
+def tag_match(regex, func)
+    TagMatch.new(regex, func)
+end
+
+# https://wikistats.wmcloud.org/wikimedias_csv.php
+# grep ,wikipedia, wikimedias.csv | cut -d, -f2 | sort
+TAGLINKS = {
+    'addr:country': tag_match(%r{^[A-Z][A-Z]$}, lambda { |value|
+        return [ TagLink.new('ISO: ' + value, 'https://www.iso.org/obp/ui/#iso:code:3166:' + value),
+                 TagLink.new('Wikipedia: ' + value, 'https://en.wikipedia.org/wiki/ISO_3166-1_alpha-2#' + value) ]
+    }),
+    phone:     tag_match(%r{^\+[0-9. -]+$},  lambda { |value| return [ TagLink.new('Phone number: ' + value, 'tel:' + value.gsub(/[ .-]+/, '-')) ] }),
+    'ref:bag': tag_match(%r{^[0-9]+$},       lambda { |value|
+        id = value.rjust(16, '0')
+        return [ TagLink.new('Basisregistratie Adressen en Gebouwen (BAG): ' + id, 'https://bagviewer.kadaster.nl/lvbag/bag-viewer/index.html#?searchQuery=' + id) ]
+    }),
+    species:   tag_match(%r{^[a-zA-Z -]+$},  lambda { |value| return [ TagLink.new('Wikispecies: ' + value, 'https://species.wikimedia.org/wiki/' + value) ] }),
+    url:       tag_match(%r{^https?://},     lambda { |value| return [ TagLink.new('Website', value) ] }),
+    website:   tag_match(%r{^https?://},     lambda { |value| return [ TagLink.new('Website', value) ] }),
+    wikidata:  tag_match(%r{^Q[0-9]{1,10}$}, lambda { |value| return [ TagLink.new('Wikidata: ' + value, 'https://www.wikidata.org/wiki/' + value) ] }),
+    wikipedia: tag_match(%r{^[a-z-]{2,8}:},  lambda { |value|
+        c = value.split(':')
+        if $WIKIPEDIA_SITES.include?(c[0])
+            return [ TagLink.new('Wikipedia', "https://#{c[0]}.wikipedia.org/wiki/#{c[1]}") ]
+        end
+        return []
+    })
+}
+
+def get_links(key, value)
+    tm = TAGLINKS[key.to_sym]
+    if tm
+        return tm.links(value)
+    end
+
+    return []
+end
+

web/lib/ui/taginfo.rb

@@ -149,6 +149,37 @@ def get_commit
         erb :'taginfo/project_error_log'
     end
 
+    get '/taginfo/taglinks' do
+        @title = 'Taglinks'
+        @section = 'taginfo'
+        @section_title = t.taginfo.meta
+
+        @data = {}
+        TAGLINKS.each do |key, match|
+            row = @db.select("SELECT count_all, values_all FROM db.keys").
+                condition("key = ?", key.to_s).
+                get_first_row
+
+            @data[key] = row
+
+            matching = 0
+
+            pcre_extension = TaginfoConfig.get('paths.sqlite3_pcre_extension')
+            if pcre_extension
+                matching = @db.count('db.tags').
+                    condition("key = ?", key.to_s).
+                    condition("value REGEXP ?", match.regex.to_s).
+                    get_first_i
+            end
+
+            @data[key]['values_match'] = matching;
+
+            @data[key]['links'] = match.call('VALUE');
+        end
+
+        erb :'taginfo/taglinks'
+    end
+
     get '/taginfo/wiki-problems' do
         @title = "Wiki problems"
         @section = 'taginfo'

web/lib/ui/tags.rb

@@ -53,6 +53,8 @@ class Taginfo < Sinatra::Base
         @img_width  = TaginfoConfig.get('geodistribution.width')  * TaginfoConfig.get('geodistribution.scale_image')
         @img_height = TaginfoConfig.get('geodistribution.height') * TaginfoConfig.get('geodistribution.scale_image')
 
+        @links = get_links(@key, @value)
+
         javascript_for(:flexigrid)
         javascript "#{ r18n.locale.code }/tag"
         erb :tag

web/taginfo.rb

@@ -46,6 +46,7 @@
 require 'rack/contrib'
 
 require 'lib/utils.rb'
+require 'lib/taglinks.rb'
 require 'lib/config.rb'
 require 'lib/javascript.rb'
 require 'lib/language.rb'
@@ -117,6 +118,7 @@ def next_update
         expires 0, :no_cache
 
         @db = SQL::Database.new.attach_sources
+        $WIKIPEDIA_SITES = @db.execute('SELECT prefix FROM wikipedia_sites').map{ |row| row['prefix'] }
 
         @data_until = DATA_UNTIL.sub(/:..$/, '')
         @data_until_m = DATA_UNTIL.sub(' ', 'T') + 'Z'

web/views/tag.erb

@@ -52,8 +52,14 @@
         <h2><%= h(t.taginfo.overview) %></h2>
         <table id="grid-overview">
         </table>
+<% if @has_rtype_link or @links.size > 0 %>
+        <h2><%= h(t.taginfo.links) %></h2>
+<% end %>
 <% if @has_rtype_link %>
         <p><%= h(t.pages.tag.overview.see_also) %>: <span id="relationlink"></span></p>
+<% end %>
+<% @links.each do |link| %>
+        <p><%= link %></p>
 <% end %>
     </div>
     <div id="combinations">

web/views/taginfo/index.erb

@@ -13,6 +13,7 @@
         <li><a href="/taginfo/config">Configuration</a></li>
         <li><a href="/taginfo/stats">Statistics</a></li>
         <li><a href="/taginfo/projects">Projects</a></li>
+        <li><a href="/taginfo/taglinks">Tag links</a></li>
         <li><a href="/taginfo/wiki-problems">Problems encountered when parsing wiki</a></li>
     </ul>
     <h2>Software Version</h2>

web/views/taginfo/taglinks.erb

@@ -0,0 +1,22 @@
+<div class="pre" lang="en" dir="ltr">
+    <h1>Tag links</h1>
+    <p>The following keys support tag links, i.e. autogenerated links from tag
+    values to. The regular expression must match for this to work.</p>
+</div>
+<div class="box">
+    <table class="list">
+        <thead><th>Key</th><th>Count</th><th>Unique values</th><th>Matching values</th><th>Regex</th><th>Links (simplified display)</th></thead>
+        <tbody>
+<% TAGLINKS.each_with_index do |(key, match), n| c = (n%2!=0) ? ' even' : '' %>
+        <tr>
+            <td class="<%= c %>"><a href="/keys/<%= h(key) %>"><%= h(key) %></a></td>
+            <td class="tr<%= c %>"><%= @data[key]['count_all'].to_s_with_ts %></td>
+            <td class="tr<%= c %>"><%= @data[key]['values_all'].to_s_with_ts %></td>
+            <td class="tr<%= c %>"><%= @data[key]['values_match'].to_s_with_ts %> (<%= (@data[key]['values_match'] * 100 / @data[key]['values_all']).to_i %>%)</td>
+            <td class="<%= c %>"><tt><%= h(match.regex.inspect[1..-2]) %></tt></td>
+            <td class="<%= c %>"><ul><%= @data[key]['links'].map{ |tl| '<li>' + h(tl.title) + ' &rarr; ' + h(tl.url) + '</li>' }.join %></ul></td>
+        </tr>
+<% end %>
+        </tbody>
+    </table>
+</div>