pumi 0.19.0 → 0.20.0

data/data/provinces.yml

@@ -10,6 +10,15 @@ provinces:
       en: Province
     links:
       wikipedia: https://en.wikipedia.org/wiki/Banteay_Meanchey_province
+    geodata:
+      lat: '13.7989147'
+      long: '102.8862666'
+      bounding_box:
+      - '13.3579351'
+      - '14.2502283'
+      - '102.3338282'
+      - '103.4427283'
+    iso3166_2: KH-1
   '02':
     name:
       km: បាត់ដំបង
@@ -20,6 +29,15 @@ provinces:
       en: Province
     links:
       wikipedia: https://en.wikipedia.org/wiki/Battambang_province
+    geodata:
+      lat: '12.9256791'
+      long: '103.23171364274523'
+      bounding_box:
+      - '12.3711193'
+      - '13.4793632'
+      - '102.3450807'
+      - '103.9011713'
+    iso3166_2: KH-2
   '03':
     name:
       km: កំពង់ចាម
@@ -30,6 +48,15 @@ provinces:
       en: Province
     links:
       wikipedia: https://en.wikipedia.org/wiki/Kampong_Cham_province
+    geodata:
+      lat: '12.1174857'
+      long: '105.20800220742942'
+      bounding_box:
+      - '11.7342165'
+      - '12.5014623'
+      - '104.8198072'
+      - '105.7759428'
+    iso3166_2: KH-3
   '04':
     name:
       km: កំពង់ឆ្នាំង
@@ -40,6 +67,15 @@ provinces:
       en: Province
     links:
       wikipedia: https://en.wikipedia.org/wiki/Kampong_Chhnang_province
+    geodata:
+      lat: '12.1382025'
+      long: '104.3354098'
+      bounding_box:
+      - '11.7344035'
+      - '12.6074435'
+      - '104.1710946'
+      - '104.9019904'
+    iso3166_2: KH-4
   '05':
     name:
       km: កំពង់ស្ពឺ
@@ -50,6 +86,15 @@ provinces:
       en: Province
     links:
       wikipedia: https://en.wikipedia.org/wiki/Kampong_Speu_province
+    geodata:
+      lat: '11.57753675'
+      long: '104.27140233242656'
+      bounding_box:
+      - '11.0779221'
+      - '12.0768453'
+      - '103.7729787'
+      - '104.7925761'
+    iso3166_2: KH-5
   '06':
     name:
       km: កំពង់ធំ
@@ -60,6 +105,15 @@ provinces:
       en: Province
     links:
       wikipedia: https://en.wikipedia.org/wiki/Kampong_Thom_province
+    geodata:
+      lat: '12.6687923'
+      long: '104.8879197'
+      bounding_box:
+      - '12.1886895'
+      - '13.4495556'
+      - '104.1734824'
+      - '105.7348383'
+    iso3166_2: KH-6
   '07':
     name:
       km: កំពត
@@ -70,6 +124,15 @@ provinces:
       en: Province
     links:
       wikipedia: https://en.wikipedia.org/wiki/Kampot_province
+    geodata:
+      lat: '10.6314754'
+      long: '104.132637'
+      bounding_box:
+      - '10.3536198'
+      - '11.1722266'
+      - '103.8390342'
+      - '104.727076'
+    iso3166_2: KH-7
   '08':
     name:
       km: កណ្តាល
@@ -80,6 +143,15 @@ provinces:
       en: Province
     links:
       wikipedia: https://en.wikipedia.org/wiki/Kandal_province
+    geodata:
+      lat: '11.396262499999999'
+      long: '105.02078685074581'
+      bounding_box:
+      - '10.8964776'
+      - '11.8943134'
+      - '104.6562154'
+      - '105.2971605'
+    iso3166_2: KH-8
   '09':
     name:
       km: កោះកុង
@@ -90,6 +162,15 @@ provinces:
       en: Province
     links:
       wikipedia: https://en.wikipedia.org/wiki/Koh_Kong_province
+    geodata:
+      lat: '11.6813417'
+      long: '103.0258043'
+      bounding_box:
+      - '10.6542293'
+      - '12.116263'
+      - '102.5849839'
+      - '104.0645553'
+    iso3166_2: KH-9
   '10':
     name:
       km: ក្រចេះ
@@ -100,6 +181,15 @@ provinces:
       en: Province
     links:
       wikipedia: https://en.wikipedia.org/wiki/Krati%C3%A9_province
+    geodata:
+      lat: '12.678301699999999'
+      long: '106.05817329824376'
+      bounding_box:
+      - '11.9494318'
+      - '13.4073182'
+      - '105.5871802'
+      - '106.8019914'
+    iso3166_2: KH-10
   '11':
     name:
       km: មណ្ឌលគិរី
@@ -110,6 +200,15 @@ provinces:
       en: Province
     links:
       wikipedia: https://en.wikipedia.org/wiki/Mondulkiri_province
+    geodata:
+      lat: '12.7396454'
+      long: '107.00882593208544'
+      bounding_box:
+      - '12.0614615'
+      - '13.4175148'
+      - '106.3385777'
+      - '107.6033126'
+    iso3166_2: KH-11
   '12':
     name:
       km: ភ្នំពេញ
@@ -120,6 +219,15 @@ provinces:
       en: Capital
     links:
       wikipedia: https://en.wikipedia.org/wiki/Phnom_Penh
+    geodata:
+      lat: '11.5730391'
+      long: '104.857807'
+      bounding_box:
+      - '11.4200852'
+      - '11.7349524'
+      - '104.7204046'
+      - '105.0440261'
+    iso3166_2: KH-12
   '13':
     name:
       km: ព្រះវិហារ
@@ -130,6 +238,15 @@ provinces:
       en: Province
     links:
       wikipedia: https://en.wikipedia.org/wiki/Preah_Vihear_province
+    geodata:
+      lat: '13.746298150000001'
+      long: '104.97403002709669'
+      bounding_box:
+      - '13.0553454'
+      - '14.4374188'
+      - '104.3579998'
+      - '105.8992487'
+    iso3166_2: KH-13
   '14':
     name:
       km: ព្រៃវែង
@@ -140,6 +257,15 @@ provinces:
       en: Province
     links:
       wikipedia: https://en.wikipedia.org/wiki/Prey_Veng_province
+    geodata:
+      lat: '11.365959700000001'
+      long: '105.45572772976189'
+      bounding_box:
+      - '10.8598854'
+      - '11.8712687'
+      - '105.1039948'
+      - '105.8091078'
+    iso3166_2: KH-14
   '15':
     name:
       km: ពោធិ៍សាត់
@@ -150,6 +276,15 @@ provinces:
       en: Province
     links:
       wikipedia: https://en.wikipedia.org/wiki/Pursat_province
+    geodata:
+      lat: '12.4662592'
+      long: '103.7426829518918'
+      bounding_box:
+      - '11.8432598'
+      - '13.0891438'
+      - '102.70351'
+      - '104.3951534'
+    iso3166_2: KH-15
   '16':
     name:
       km: រតនគិរី
@@ -160,6 +295,15 @@ provinces:
       en: Province
     links:
       wikipedia: https://en.wikipedia.org/wiki/Ratanakiri_province
+    geodata:
+      lat: '13.9311155'
+      long: '107.03914990288844'
+      bounding_box:
+      - '13.1721622'
+      - '14.6904224'
+      - '106.5301893'
+      - '107.6276788'
+    iso3166_2: KH-16
   '17':
     name:
       km: សៀមរាប
@@ -170,6 +314,15 @@ provinces:
       en: Province
     links:
       wikipedia: https://en.wikipedia.org/wiki/Siem_Reap_province
+    geodata:
+      lat: '13.314962'
+      long: '103.805433'
+      bounding_box:
+      - '12.7720495'
+      - '13.9718578'
+      - '103.3892306'
+      - '104.6832474'
+    iso3166_2: KH-17
   '18':
     name:
       km: ព្រះសីហនុ
@@ -180,6 +333,15 @@ provinces:
       en: Province
     links:
       wikipedia: https://en.wikipedia.org/wiki/Sihanoukville_province
+    geodata:
+      lat: '10.50614075'
+      long: '103.26205644863938'
+      bounding_box:
+      - '9.4110961'
+      - '11.436044'
+      - '102.6295604'
+      - '104.0772547'
+    iso3166_2: KH-18
   '19':
     name:
       km: ស្ទឹងត្រែង
@@ -190,6 +352,15 @@ provinces:
       en: Province
     links:
       wikipedia: https://en.wikipedia.org/wiki/Stung_Treng_province
+    geodata:
+      lat: '13.852468'
+      long: '106.0889848986395'
+      bounding_box:
+      - '13.1137001'
+      - '14.5919671'
+      - '105.5194671'
+      - '106.7891813'
+    iso3166_2: KH-19
   '20':
     name:
       km: ស្វាយរៀង
@@ -200,6 +371,15 @@ provinces:
       en: Province
     links:
       wikipedia: https://en.wikipedia.org/wiki/Svay_Rieng_province
+    geodata:
+      lat: '11.1942275'
+      long: '105.7973275083619'
+      bounding_box:
+      - '10.7889074'
+      - '11.5996145'
+      - '105.5796471'
+      - '106.206119'
+    iso3166_2: KH-20
   '21':
     name:
       km: តាកែវ
@@ -210,6 +390,15 @@ provinces:
       en: Province
     links:
       wikipedia: https://en.wikipedia.org/wiki/Tak%C3%A9o_province
+    geodata:
+      lat: '10.93539205'
+      long: '104.78988300047047'
+      bounding_box:
+      - '10.5187794'
+      - '11.3530403'
+      - '104.4101661'
+      - '105.0983022'
+    iso3166_2: KH-21
   '22':
     name:
       km: ឧត្តរមានជ័យ
@@ -220,6 +409,15 @@ provinces:
       en: Province
     links:
       wikipedia: https://en.wikipedia.org/wiki/Oddar_Meanchey_province
+    geodata:
+      lat: '14.1585631'
+      long: '103.78279595655918'
+      bounding_box:
+      - '13.8769057'
+      - '14.4403696'
+      - '103.0318699'
+      - '104.5085773'
+    iso3166_2: KH-22
   '23':
     name:
       km: កែប
@@ -230,6 +428,15 @@ provinces:
       en: Province
     links:
       wikipedia: https://en.wikipedia.org/wiki/Kep_province
+    geodata:
+      lat: '10.402153949999999'
+      long: '104.26506181616313'
+      bounding_box:
+      - '10.1548084'
+      - '10.5908859'
+      - '104.0710119'
+      - '104.4259394'
+    iso3166_2: KH-23
   '24':
     name:
       km: ប៉ៃលិន
@@ -240,6 +447,15 @@ provinces:
       en: Province
     links:
       wikipedia: https://en.wikipedia.org/wiki/Pailin_province
+    geodata:
+      lat: '12.89053285'
+      long: '102.62946042446026'
+      bounding_box:
+      - '12.663855'
+      - '13.1166697'
+      - '102.485444'
+      - '102.7533088'
+    iso3166_2: KH-24
   '25':
     name:
       km: ត្បូងឃ្មុំ
@@ -250,3 +466,12 @@ provinces:
       en: Province
     links:
       wikipedia: https://en.wikipedia.org/wiki/Tboung_Khmum_province
+    geodata:
+      lat: '11.950072599999999'
+      long: '105.824571891073'
+      bounding_box:
+      - '11.5995257'
+      - '12.3005412'
+      - '105.408445'
+      - '106.4674381'
+    iso3166_2: KH-25

data/lib/pumi/bot/wikipedia/article.rb

@@ -0,0 +1,13 @@
+module Pumi
+  module Bot
+    module Wikipedia
+      class Article
+        attr_reader :client
+
+        def initialize(client: Pumi::Wikipedia::Client.new)
+          @client = client
+        end
+      end
+    end
+  end
+end

data/lib/pumi/bot/wikipedia/communes_in_cambodia_article.rb

@@ -0,0 +1,157 @@
+require "ostruct"
+
+module Pumi
+  module Bot
+    module Wikipedia
+      class CommunesInCambodiaArticle < Article
+        class Location < SimpleDelegator
+          def communes_summary
+            summary = communes_by_type.map do |administrative_unit, communes|
+              next if Array(communes).empty?
+
+              text = "#{communes.size} #{administrative_unit.name_en}"
+              text << "s" if communes.size > 1
+              text << " (#{administrative_unit.name_km} #{administrative_unit.name_latin})"
+            end
+
+            summary << "#{format_number(villages_count)} Villages (ភូមិ Phum)"
+
+            if summary.size <= 2
+              summary.join(" and ")
+            else
+              [summary[0..-2].join(", "), summary[-1]].join(" and ")
+            end
+          end
+
+          private
+
+          def communes_by_type
+            communes.each_with_object({}) do |commune, result|
+              administrative_unit = commune.administrative_unit
+              result[administrative_unit] ||= []
+              result[administrative_unit] << commune
+            end
+          end
+
+          def format_number(number)
+            number.to_s.gsub(/(\d)(?=(\d\d\d)+(?!\d))/, "\\1,")
+          end
+        end
+
+        class Province < Location
+          def districts
+            Pumi::District.where(province_id: id).map { |district| District.new(district) }
+          end
+
+          def communes
+            Pumi::Commune.where(province_id: id)
+          end
+
+          def villages_count
+            Pumi::Village.where(province_id: id).count
+          end
+        end
+
+        class District < Location
+          def communes
+            Pumi::Commune.where(district_id: id)
+          end
+
+          def villages_count
+            Pumi::Village.where(district_id: id).count
+          end
+        end
+
+        PAGE_TITLE = "List_of_communes_in_Cambodia".freeze
+        INTRO_TEXT = "<div id=\"intro\">The '''communes of Cambodia''' ({{lang|km|ឃុំ}} ''[[khum]]''/{{lang|km|សង្កាត់}} ''[[sangkat]]'') are the third-level administrative divisions in  Cambodia. They are the subdivisions of the [[List of districts in Cambodia|districts and municipalities of Cambodia]]. Communes can consist of as few as %<min_villages>s<ref>{{cite web|url=http://db.ncdd.gov.kh/gazetteer/view/commune.castle?cm=%<smallest_commune_id>s |title=%<smallest_commune_name>s |publisher=National Committee for Sub-National Democratic Development }}</ref> or as many as %<max_villages>s<ref>{{cite web|url=http://db.ncdd.gov.kh/gazetteer/view/commune.castle?cm=%<largest_commune_id>s |title=%<largest_commune_name>s |publisher=National Committee for Sub-National Democratic Development }}</ref> villages (''[[phum]]''), depending on the population.\nThere are a total of %<communes_count>s communes and %<villages_count>s villages in Cambodia.</div>".freeze
+
+        TEMPLATE = File.read("#{__dir__}/templates/commune_list.wikitext.erb")
+
+        def publish
+          replace_intro
+          replace_number_of_communes
+          replace_communes_list
+          client.update_page(title: PAGE_TITLE, source:, comment: "Update page")
+        end
+
+        private
+
+        def source_html
+          @source_html ||= Nokogiri::HTML(source)
+        end
+
+        def source
+          @source ||= page.fetch(:source).dup
+        end
+
+        def page
+          @page ||= client.get_page(title: PAGE_TITLE)
+        end
+
+        def intro_section
+          result = source_html.at_css('[id="intro"]')
+          raise "Could not find section with <div id=\"intro\">" if result.nil?
+
+          result
+        end
+
+        def communes_list_section
+          result = source_html.at_css('[id="communes-list"]')
+          raise "Could not find section with <div id=\"communes-list\">" if result.nil?
+
+          result
+        end
+
+        def replace_communes_list
+          provinces = Pumi::Province.all.map { |province| Province.new(province) }
+          data = OpenStruct.new(provinces:)
+          communes_list = ERB.new(TEMPLATE).result(data.instance_eval { binding })
+          source.sub!(communes_list_section.to_html, communes_list)
+        end
+
+        def replace_number_of_communes
+          source.sub!(
+            /current_number\s*=\s*(?:^|\s)(\d*\.?\d+|\d{1,3}(?:,\d{3})*(?:\.\d+)?)(?!\S)/,
+            "current_number = #{format_number(Pumi::Commune.all.size)}"
+          )
+        end
+
+        def replace_intro
+          villages = Pumi::Village.all.each_with_object({}) do |village, result|
+            result[village.commune] ||= []
+
+            result[village.commune] << village
+          end
+
+          village_sizes = villages.each_with_object({}) do |(commune, villages), result|
+            result[villages.size] = commune
+          end
+
+          min_villages = village_sizes.keys.min
+          max_villages = village_sizes.keys.max
+
+          smallest_commune = village_sizes[min_villages]
+          largest_commune = village_sizes[max_villages]
+
+          intro = format(
+            INTRO_TEXT,
+            min_villages:,
+            max_villages:,
+            smallest_commune_id: smallest_commune.id,
+            smallest_commune_name: smallest_commune.name_en,
+            largest_commune_id: largest_commune.id,
+            largest_commune_name: largest_commune.name_en,
+            communes_count: format_number(Pumi::Commune.all.count),
+            villages_count: format_number(Pumi::Village.all.count)
+          )
+
+          source.sub!(intro_section.to_html, intro)
+        end
+
+        def format_number(number)
+          number.to_s.gsub(/(\d)(?=(\d\d\d)+(?!\d))/, "\\1,")
+        end
+      end
+    end
+  end
+end

data/lib/pumi/bot/wikipedia/districts_in_cambodia_article.rb

@@ -0,0 +1,122 @@
+require "ostruct"
+
+module Pumi
+  module Bot
+    module Wikipedia
+      class DistrictsInCambodiaArticle < Article
+        PAGE_TITLE = "List_of_districts,_municipalities_and_sections_in_Cambodia".freeze
+
+        INTRO_TEXT = "This is a list of [[Cambodia]]'s %<districts_count>s districts ({{lang|km|ស្រុក}} ''srok''), %<municipalities_count>s district-level municipalities ({{lang|km|ក្រុង}} ''krong'') and %<sections_count>s sections ({{lang|km|ខណ្ឌ}} ''khan'') organized by each [[Provinces of Cambodia|province]] and an [[Provinces of Cambodia|autonomous municipality]] ([[Phnom Penh]]).".freeze
+        DISTRICTS_TEMPLATE = File.read("#{__dir__}/templates/district_list.wikitext.erb")
+
+        Misspelling = Struct.new(:incorrect_text, :correct_text, keyword_init: true)
+
+        MISSPELLINGS = [
+          Misspelling.new(incorrect_text: "Kratié", correct_text: "Kratie"),
+          Misspelling.new(incorrect_text: "Mondulkiri", correct_text: "Mondul Kiri"),
+          Misspelling.new(
+            incorrect_text: "Phnom Penh (autonomous municipality)",
+            correct_text: "Phnom Penh"
+          ),
+          Misspelling.new(incorrect_text: "Ratanakiri", correct_text: "Ratanak Kiri"),
+          Misspelling.new(incorrect_text: "Siem Reap", correct_text: "Siemreap"),
+          Misspelling.new(incorrect_text: "Takéo", correct_text: "Takeo")
+        ].freeze
+
+        def publish
+          page = client.get_page(title: PAGE_TITLE)
+          source = page.fetch(:source).dup
+
+          replace_intro(source:, replacement: generate_intro)
+          replace_number_of_districts(source:, replacement: District.all.size)
+
+          Pumi::Province.all.each do |province|
+            section = find_section_by_title(source:, title: province.name_en)
+            new_section = generate_districts_section(province:)
+            source.sub!(section[1], new_section)
+          end
+
+          client.update_page(title: PAGE_TITLE, source:, comment: "Update page")
+        end
+
+        private
+
+        def generate_districts_section(province:)
+          districts = Pumi::District.where(province_id: province.id)
+
+          data = OpenStruct.new(
+            province:,
+            districts:,
+            districts_summary: generate_districts_summary(districts:)
+          )
+          result = ERB.new(DISTRICTS_TEMPLATE).result(data.instance_eval { binding })
+          "\n\n#{result}\n"
+        end
+
+        def generate_districts_summary(districts:)
+          summary = districts_by_type(collection: districts).map do |au, d|
+            next if Array(d).empty?
+
+            text = "#{d.size} #{au.name_en}"
+            if d.size > 1
+              au.name_en.end_with?("y") ? text.chomp!("y") << "ies" : text << "s"
+            end
+            text << " (#{au.name_km} #{au.name_latin})"
+          end
+
+          if summary.size <= 2
+            summary.join(" and ")
+          else
+            [summary[0..-2].join(", "), summary[-1]].join(" and ")
+          end
+        end
+
+        def find_section_by_title(source:, title:)
+          misspelling = MISSPELLINGS.find { |m| m.correct_text == title }
+          search_term = misspelling&.incorrect_text || title
+
+          section = source.match(/==\s*\[\[[[:word:]\s]+\|#{Regexp.escape(search_term)}\]\].+?==(.+?(?===))/m)
+
+          raise "Missing section for #{search_term}" if section.nil?
+
+          section
+        end
+
+        def generate_intro
+          all_districts = districts_by_type
+
+          format(
+            INTRO_TEXT,
+            districts_count: fetch_districts_by_type("District", collection: all_districts).size,
+            municipalities_count: fetch_districts_by_type(
+              "Municipality",
+              collection: all_districts
+            ).size,
+            sections_count: fetch_districts_by_type("Section", collection: all_districts).size
+          )
+        end
+
+        def replace_number_of_districts(source:, replacement:)
+          source.sub!(/current_number\s*=\s*\d+/, "current_number = #{replacement}")
+        end
+
+        def replace_intro(source:, replacement:)
+          source.sub!(/This is a list of.+/, replacement)
+        end
+
+        def districts_by_type(collection: Pumi::District.all)
+          collection.each_with_object({}) do |district, result|
+            administrative_unit = district.administrative_unit
+            result[administrative_unit] ||= []
+            result[administrative_unit] << district
+          end
+        end
+
+        def fetch_districts_by_type(type, collection:)
+          administrative_unit = collection.keys.find { |au| au.name_en == type }
+          Array(collection[administrative_unit])
+        end
+      end
+    end
+  end
+end