gouv-calendar-compilator 0.0.2

data/README.md

@@ -0,0 +1,57 @@
+# Gouvernement Calendar Data Compilator
+
+This library fetches data from the governement sources and APIs for national days off and school holidays, and compile it into a JSON file.
+
+# Library usage
+
+Add the gem to your Gemfile or \*.gemspec file:
+
+    gem 'gouv-calendar-compilator'
+
+Then use it:
+
+    require 'gouv-calendar-compilator'
+
+Wherever you wanna use the data, create a new GouvCalendarCompilator::DataCompilator object and call the `compile` method  like so:
+```
+data_compilator = GouvCalendarCompilator::DataCompilator.new
+data_compilator.compile
+```
+
+The compile method will return a ruby Array of Hashes where each Hash represents a period of time containing the data formatted like so:
+```
+[
+    {
+        start: "2021-10-10",
+        end: "2021-10-11",
+        coefficient: 0
+    },
+    ...
+]
+```
+
+**start** is a string containing the beginning date of the period of time (should always be the same than the previous period of time **end** field)
+**end** is a string containing the date of the day after the end of the period of time 
+**coefficient** is an integer (possible values 0, 1, 2) representing the type of period it is, correspnding to:
+
+0 = Working Day
+1 = School Holidays
+2 = Holiday
+
+# Contributing
+
+Thanks to think of contributing.
+
+We are open to contributions. Please feal free to create an issue covering what you wanted to do and
+to create a PR directed to the develop branch for it to be accepted.
+
+We do not have clear development roadmap, because we will contribute to this project on the need for
+us. This won't prevent to have time answering to your needs, and welcome test cases or other bug reports.
+
+Please:
+
+- respect each others (take example to existing [code of conducts like the one of Atom for example](https://github.com/atom/atom/blob/master/CODE_OF_CONDUCT.md)).
+- do not use issues or PR comments for anything else than the purpose of this project.
+
+We will use censorship when a comment is deemed (by us - project managers) as inappropriate (Trolling,
+insults or any other non-productive discussions which does not deserve to be stored).

data/lib/gouv-calendar-compilator.rb

@@ -0,0 +1,8 @@
+#--
+# Copyright (C) 2021 - Octopus Lab SAS
+# All rights are described by the GPLv3 license present in the file /LICENSE available in the
+# original repository of this file or [here](https://www.gnu.org/licenses/gpl-3.0.fr.html).
+#++
+# frozen_string_literal: true
+
+require 'gouv_calendar_compilator'

data/lib/gouv_calendar_compilator/data_compilator.rb

@@ -0,0 +1,74 @@
+#--
+# Copyright (C) 2021 - Octopus Lab SAS
+# All rights are described by the GPLv3 license present in the file /LICENSE available in the
+# original repository of this file or [here](https://www.gnu.org/licenses/gpl-3.0.fr.html).
+#++
+# frozen_string_literal: true
+
+require 'open-uri'
+require 'json'
+
+module GouvCalendarCompilator
+  # This is the Data Compilator Object for the Gouv Data Compilator gem
+  # It manages the compilation of different data sources and compile them into
+  # a new, properly formatted one
+  class DataCompilator
+    # Initializer for Data Compiler class.
+    def initialize
+      @sh_data_compiler = ::GouvCalendarCompilator::France::SchoolHolidaysDataCompilator.new
+      @ndo_data_compiler = ::GouvCalendarCompilator::France::NationalDaysOffDataCompilator.new
+    end
+
+    # This is the method you want to call to compile the sources into a proper dataset.
+    # It returns a ruby Array of Hashes
+    def compile
+      sh_calendar_data = @sh_data_compiler.compile
+      ndo_and_sh_data = self.class.sort_calendar_data(@ndo_data_compiler.compile(sh_calendar_data))
+      self.class.fill_operating_days(ndo_and_sh_data)
+    end
+
+    # Sorts calendar data by date
+    def self.sort_calendar_data(calendar)
+      calendar.each_key do |zone_name|
+        calendar[zone_name].sort_by! { |rec| ::Date.parse(rec[:start]) }
+      end
+      calendar
+    end
+
+    # Fill in the gaps in school holidays + national days off calendar with operatiog periods
+    def self.fill_operating_days(calendar)
+      calendar = sort_calendar_data(calendar)
+      calendar.each_key do |zone|
+        new_entries = []
+        calendar[zone].each_with_index do |vacation_period, index|
+          current_end_date = ::Date.parse(vacation_period[:end])
+          if calendar[zone][index + 1].nil?
+            if (current_end_date...::GouvCalendarCompilator::DATETIME_END).count > 1
+              new_entries.push(
+                {
+                  start: current_end_date.to_date.to_s,
+                  end: (::GouvCalendarCompilator::DATETIME_END + 1).to_date.to_s,
+                  coefficient: ::GouvCalendarCompilator::OPERATING_DAY
+                }
+              )
+            end
+          else
+            next_start_date = ::Date.parse(calendar[zone][index + 1][:start])
+
+            if (current_end_date...next_start_date).count >= 1
+              new_entries.push(
+                {
+                  start: current_end_date.to_date.to_s,
+                  end: next_start_date.to_date.to_s,
+                  coefficient: ::GouvCalendarCompilator::OPERATING_DAY
+                }
+              )
+            end
+          end
+        end
+        calendar[zone] += new_entries
+      end
+      calendar
+    end
+  end
+end

data/lib/gouv_calendar_compilator/data_compilators/france/national_days_off_data_compilator.rb

@@ -0,0 +1,111 @@
+#--
+# Copyright (C) 2021 - Octopus Lab SAS
+# All rights are described by the GPLv3 license present in the file /LICENSE available in the
+# original repository of this file or [here](https://www.gnu.org/licenses/gpl-3.0.fr.html).
+#++
+# frozen_string_literal: true
+
+module GouvCalendarCompilator
+  module France
+    # Data compiler class for national days off
+    class NationalDaysOffDataCompilator
+      # Initializer for France National Days off Compiler class.
+      def initialize
+        @data_fetcher = ::GouvCalendarCompilator::DataFetcher.new
+      end
+
+      # This is the method you want to call to compile the sources into a proper dataset.
+      # It returns a ruby Array of Hashes
+      #
+      # @param calendar Array<Hash> : previously formatted has of School holidays
+      def compile(calendar)
+        @data_fetcher.fetch_national_days_off_data.each do |zone_specific_ndo_data|
+          calendar = ndo_parse_data_region(
+            zone_specific_ndo_data[:data],
+            zone_specific_ndo_data[:zones],
+            ::GouvCalendarCompilator::DataCompilator.sort_calendar_data(calendar)
+          )
+        end
+
+        calendar
+      end
+
+      private
+
+      # NATIONAL DAYS OFF DEDICATED FUNCTIONS
+      def ndo_parse_data_region(ndo_dataset, targeted_zones, existing_calendar)
+        ndo_dataset.each_key do |day_off|
+          existing_calendar = ndo_insert_day_off_date_in_calendar(
+            ::Date.parse(day_off),
+            targeted_zones,
+            existing_calendar
+          )
+        end
+        existing_calendar
+      end
+
+      # Splits the periods of time when a national day off falls into a school holiday period
+      def ndo_split_holidays_period_insert_day_off(holidays_period, day_off_date)
+        [
+          {
+            start: holidays_period[:start],
+            end: day_off_date.to_s,
+            coefficient: ::GouvCalendarCompilator::SCHOOL_HOLIDAYS
+          },
+          {
+            start: day_off_date.to_s,
+            end: (day_off_date + 1).to_s,
+            coefficient: ::GouvCalendarCompilator::NATIONAL_DAY_OFF
+          },
+          {
+            start: (day_off_date + 1).to_s,
+            end: (::Date.parse(holidays_period[:end]) + 1).to_s,
+            coefficient: ::GouvCalendarCompilator::SCHOOL_HOLIDAYS
+          }
+        ]
+      end
+
+      # checks if the given date already exists in the generated calendar for the given zones
+      def ndo_insert_day_off_date_in_calendar(day_off_date, zones, calendar)
+        zones.each do |zone_name|
+          included_in_vacation = false
+          calendar[zone_name].each_with_index do |vacation_period, index|
+            start_date = ::Date.parse(vacation_period[:start])
+            end_date = ::Date.parse(vacation_period[:end])
+            date_range = (start_date...end_date)
+            unless date_range.include?(day_off_date) &&
+                   vacation_period[:coefficient] != ::GouvCalendarCompilator::NATIONAL_DAY_OFF
+
+              next
+            end
+
+            included_in_vacation = true
+            splitted_vacation = ndo_split_holidays_period_insert_day_off(calendar[zone_name][index], day_off_date)
+
+            # replacing current holiday with the first part of the splitted one
+            # except if day off is first day of holiday (then deleting the first part of the splitted vacation)
+            if calendar[zone_name][index][:start] == day_off_date.to_s
+              calendar[zone_name].delete(calendar[zone_name][index])
+            else
+              calendar[zone_name][index] = splitted_vacation[0]
+            end
+            # inserting national day off
+            calendar[zone_name] << splitted_vacation[1]
+            # inserting other part of the splitted vacation
+            calendar[zone_name] << splitted_vacation[2]
+          end
+
+          # if day off not in the middle of a vacation period, simply insert it inside the dataset
+          next if included_in_vacation
+
+          calendar[zone_name] << {
+            start: day_off_date.to_s,
+            end: (day_off_date + 1).to_s,
+            coefficient: ::GouvCalendarCompilator::NATIONAL_DAY_OFF
+          }
+        end
+        calendar
+      end
+    end
+  end
+end

data/lib/gouv_calendar_compilator/data_compilators/france/school_holidays_data_compilator.rb

@@ -0,0 +1,143 @@
+#--
+# Copyright (C) 2021 - Octopus Lab SAS
+# All rights are described by the GPLv3 license present in the file /LICENSE available in the
+# original repository of this file or [here](https://www.gnu.org/licenses/gpl-3.0.fr.html).
+#++
+# frozen_string_literal: true
+
+module GouvCalendarCompilator
+  module France
+    # Manages the compilation of different school holidays data sources and compile them into
+    # a new, properly formatted one
+    class SchoolHolidaysDataCompilator
+      # Initializer for School Holidays data compiler
+      def initialize
+        @data_fetcher = ::GouvCalendarCompilator::DataFetcher.new
+      end
+
+      # This is the method you want to call to compile the school holidays into a proper dataset.
+      # It returns a ruby Array of Hashes
+      def compile
+        school_holidays_data_full = @data_fetcher.fetch_school_holidays_data
+        school_holidays_data_trimmed = sh_trim_not_necessary_data(school_holidays_data_full)
+        zones = sh_define_possible_zones(school_holidays_data_trimmed)
+        sh_format_zones_data(school_holidays_data_trimmed, zones, school_holidays_data_full)
+      end
+
+      private
+
+      # Calculates an approximate end date based on same vacations, for same region of the world, from an other year
+      def calculate_approximate_end_date(trimmed_dataset, raw_dataset, vacation_index)
+        vacation_type = trimmed_dataset[vacation_index]['fields']['description']
+        vacation_location = trimmed_dataset[vacation_index]['fields']['location']
+
+        filtered_data =
+          raw_dataset.select do |record|
+            record['fields']['description'] == vacation_type && record['fields']['location'] == vacation_location
+          end
+        date_range = nil
+
+        filtered_data.each do |vacation|
+          next unless vacation['fields']['end_date']
+
+          start_date = ::Date.parse(vacation['fields']['start_date'])
+          end_date = ::Date.parse(vacation['fields']['end_date'])
+          date_range = (start_date...end_date)
+          break
+        end
+
+        approximate_end_date = ::Date.parse(trimmed_dataset[vacation_index]['fields']['start_date']) + date_range.count
+        approximate_end_date.to_date.to_s
+      end
+
+      # Calculates an approximate start date based on same vacations, for same region of the world, from an other year
+      def calculate_approximate_start_date(trimmed_dataset, raw_dataset, vacation_index)
+        vacation_type = trimmed_dataset[vacation_index]['fields']['description']
+        vacation_location = trimmed_dataset[vacation_index]['fields']['location']
+
+        filtered_data =
+          raw_dataset.select do |record|
+            record['fields']['description'] == vacation_type && record['fields']['location'] == vacation_location
+          end
+        date_range = nil
+
+        filtered_data.each do |vacation|
+          next unless vacation['fields']['start_date'] && vacation['fields']['description'] == vacation_type
+
+          start_date = ::Date.parse(vacation['fields']['start_date'])
+          end_date = ::Date.parse(vacation['fields']['end_date'])
+          date_range = (start_date...end_date)
+          break
+        end
+
+        approximate_start_date = ::Date.parse(trimmed_dataset[vacation_index]['fields']['end_date']) - date_range.count
+        approximate_start_date.to_date.to_s
+      end
+
+      # Trims dataset to desired time era
+      def sh_trim_not_necessary_data(raw_sh_dataset)
+        date_range = ::GouvCalendarCompilator::DATETIME_START...::GouvCalendarCompilator::DATETIME_END
+        filtered_data =
+          raw_sh_dataset.select do |record|
+            date_range.include?(::Date.parse(record['fields']['start_date'])) &&
+              (record['fields']['end_date'] ? date_range.include?(::Date.parse(record['fields']['end_date'])) : true)
+          end
+        filtered_data.uniq do |record|
+          [
+            record['fields']['start_date'],
+            record['fields']['description'],
+            record['fields']['zones']
+          ].join(':')
+        end
+      end
+
+      # Returns an array of strings corresponding to each of the existing zones in the school holidays dataset
+      def sh_define_possible_zones(school_holidays)
+        zones = []
+        school_holidays.each do |record|
+          zones << record['fields']['zones']
+        end
+        zones.uniq
+      end
+
+      # Returns the formatted school holidays data for all zones
+      def sh_format_zones_data(school_holidays, zones, sh_data_full)
+        formatted_zones_data = {}
+        zones.each do |zone_name|
+          zone_specific_holidays_data = school_holidays.select { |record| record['fields']['zones'] == zone_name }
+          zone_specific_holidays_data_full = sh_data_full.select { |record| record['fields']['zones'] == zone_name }
+          records_contianer = []
+          zone_specific_holidays_data.each_with_index do |holiday_record, index|
+            holiday_start_date = (if holiday_record['fields']['start_date'].nil?
+                                    calculate_approximate_start_date(
+                                      zone_specific_holidays_data,
+                                      zone_specific_holidays_data_full,
+                                      index
+                                    )
+                                  else
+                                    ::Date.parse(holiday_record['fields']['start_date']).to_s
+                                  end)
+            holiday_end_date = (if holiday_record['fields']['end_date'].nil?
+                                  calculate_approximate_end_date(
+                                    zone_specific_holidays_data,
+                                    zone_specific_holidays_data_full,
+                                    index
+                                  )
+                                else
+                                  (::Date.parse(holiday_record['fields']['end_date']) + 1).to_s
+                                end)
+            records_contianer.push(
+              {
+                start: holiday_start_date,
+                end: holiday_end_date,
+                coefficient: ::GouvCalendarCompilator::SCHOOL_HOLIDAYS
+              }
+            )
+          end
+          formatted_zones_data[zone_name] = records_contianer
+        end
+        formatted_zones_data
+      end
+    end
+  end
+end

data/lib/gouv_calendar_compilator/data_fetcher.rb

@@ -0,0 +1,34 @@
+#--
+# Copyright (C) 2021 - Octopus Lab SAS
+# All rights are described by the GPLv3 license present in the file /LICENSE available in the
+# original repository of this file or [here](https://www.gnu.org/licenses/gpl-3.0.fr.html).
+#++
+# frozen_string_literal: true
+
+require 'open-uri'
+require 'json'
+
+module GouvCalendarCompilator
+  # This is the Data Fetcher Object for the Gouv Data Compilator gem
+  # It manages the fetching of the needed data through the different sources
+  class DataFetcher
+    # Method used to fetch national days off data from governement open data sources
+    def fetch_national_days_off_data
+      ndo_data_full = []
+      ::GouvCalendarCompilator::DAYS_OFF_ZONES_MATCHING.each_key do |zone_name|
+        zone_data = {}
+        (::GouvCalendarCompilator::YEAR_START..::GouvCalendarCompilator::YEAR_END).each do |year|
+          url = ::GouvCalendarCompilator::DAYS_OFF_API_BASE_URL + "#{zone_name}/#{year}.json"
+          zone_data.merge!(::JSON.parse(::URI.parse(url).read))
+        end
+        ndo_data_full.push({ data: zone_data, zones: ::GouvCalendarCompilator::DAYS_OFF_ZONES_MATCHING[zone_name] })
+      end
+      ndo_data_full
+    end
+
+    # Method used to fetch school holidays data from governement open data sources
+    def fetch_school_holidays_data
+      ::JSON.parse(::URI.parse(::GouvCalendarCompilator::SCHOOL_HOLIDAYS_DATA_URL).read)
+    end
+  end
+end

data/lib/gouv_calendar_compilator/zone_department_matching/france.rb

@@ -0,0 +1,123 @@
+# frozen_string_literal: true
+
+#--
+# Copyright (C) 2021 - Octopus Lab SAS
+# All rights are described by the GPLv3 license present in the file /LICENSE available in the
+# original repository of this file or [here](https://www.gnu.org/licenses/gpl-3.0.fr.html).
+#++
+module GouvCalendarCompilator
+  module ZoneDepartmentMatching
+    FRANCE = {
+      '01': 'Zone A',
+      '02': 'Zone B',
+      '03': 'Zone A',
+      '04': 'Zone B',
+      '05': 'Zone B',
+      '06': 'Zone B',
+      '07': 'Zone A',
+      '08': 'Zone B',
+      '09': 'Zone C',
+      '10': 'Zone B',
+      '11': 'Zone C',
+      '12': 'Zone C',
+      '13': 'Zone B',
+      '14': 'Zone B',
+      '15': 'Zone A',
+      '16': 'Zone A',
+      '17': 'Zone A',
+      '18': 'Zone B',
+      '19': 'Zone A',
+      '21': 'Zone A',
+      '22': 'Zone B',
+      '23': 'Zone A',
+      '24': 'Zone A',
+      '25': 'Zone A',
+      '26': 'Zone A',
+      '27': 'Zone B',
+      '28': 'Zone B',
+      '29': 'Zone B',
+      '2a': 'Corse',
+      '2b': 'Corse',
+      '30': 'Zone C',
+      '31': 'Zone C',
+      '32': 'Zone C',
+      '33': 'Zone A',
+      '34': 'Zone C',
+      '35': 'Zone B',
+      '36': 'Zone B',
+      '37': 'Zone B',
+      '38': 'Zone A',
+      '39': 'Zone A',
+      '40': 'Zone A',
+      '41': 'Zone B',
+      '42': 'Zone A',
+      '43': 'Zone A',
+      '44': 'Zone B',
+      '45': 'Zone B',
+      '46': 'Zone C',
+      '47': 'Zone A',
+      '48': 'Zone C',
+      '49': 'Zone B',
+      '50': 'Zone B',
+      '51': 'Zone B',
+      '52': 'Zone B',
+      '53': 'Zone B',
+      '54': 'Zone B',
+      '55': 'Zone B',
+      '56': 'Zone B',
+      '57': 'Zone B',
+      '58': 'Zone A',
+      '59': 'Zone B',
+      '60': 'Zone B',
+      '61': 'Zone B',
+      '62': 'Zone B',
+      '63': 'Zone A',
+      '64': 'Zone A',
+      '65': 'Zone C',
+      '66': 'Zone C',
+      '67': 'Zone B',
+      '68': 'Zone B',
+      '69': 'Zone A',
+      '70': 'Zone A',
+      '71': 'Zone A',
+      '72': 'Zone B',
+      '73': 'Zone A',
+      '74': 'Zone A',
+      '75': 'Zone C',
+      '76': 'Zone B',
+      '77': 'Zone C',
+      '78': 'Zone C',
+      '79': 'Zone A',
+      '80': 'Zone B',
+      '81': 'Zone C',
+      '82': 'Zone C',
+      '83': 'Zone B',
+      '84': 'Zone B',
+      '85': 'Zone B',
+      '86': 'Zone A',
+      '87': 'Zone A',
+      '88': 'Zone B',
+      '89': 'Zone A',
+      '90': 'Zone A',
+      '91': 'Zone C',
+      '92': 'Zone C',
+      '93': 'Zone C',
+      '94': 'Zone C',
+      '95': 'Zone C',
+      '971': 'Guadeloupe',
+      '972': 'Martinique',
+      '973': 'Guyane',
+      '974': 'Réunion',
+      '975': 'Saint Pierre et Miquelon',
+      '976': 'Mayotte',
+      '977': 'Saint-Barthélemy',
+      '978': 'Saint-Martin',
+      '986': 'Wallis et Futuna',
+      '987': 'Polynésie',
+      '988': 'Nouvelle Calédonie',
+      '984': 'Terres australes et antarctiques françaises',
+      '989': 'Clipperton'
+    }.freeze
+    public_constant :FRANCE
+  end
+end

data/lib/gouv_calendar_compilator.rb

@@ -0,0 +1,62 @@
+#--
+# Copyright (C) 2021 - Octopus Lab SAS
+# All rights are described by the GPLv3 license present in the file /LICENSE available in the
+# original repository of this file or [here](https://www.gnu.org/licenses/gpl-3.0.fr.html).
+#++
+# frozen_string_literal: true
+
+# The GouvCalendarCompilator Module
+# Compiles School Holidays Data and National Days Off data from Government sources into a new dataset
+# and returns it as an Array of Hashes
+module GouvCalendarCompilator
+  YEAR_START = 2020
+  public_constant :YEAR_START
+  YEAR_END = (::Date.today.year + 1)
+  public_constant :YEAR_END
+  SCHOOL_HOLIDAYS_DATA_URL = 'https://www.data.gouv.fr/fr/datasets/r/000ae493-9fa8-4088-9f53-76d375204036'
+  public_constant :SCHOOL_HOLIDAYS_DATA_URL
+  DAYS_OFF_API_BASE_URL = 'https://calendrier.api.gouv.fr/jours-feries/'
+  public_constant :DAYS_OFF_API_BASE_URL
+
+  DAYS_OFF_ZONES_MATCHING = {
+    guadeloupe: ['Guadeloupe'],
+    guyane: ['Guyane'],
+    'la-reunion': ['Réunion'],
+    martinique: ['Martinique'],
+    mayotte: ['Mayotte'],
+    metropole: ['Zone A', 'Zone B', 'Zone C', 'Corse'],
+    'nouvelle-caledonie': ['Nouvelle Calédonie'],
+    'polynesie-francaise': ['Polynésie'],
+    'saint-pierre-et-miquelon': ['Saint Pierre et Miquelon'],
+    'wallis-et-futuna': ['Wallis et Futuna']
+  }.freeze
+
+  public_constant :DAYS_OFF_ZONES_MATCHING
+
+  OPERATING_DAY = 0
+  public_constant :OPERATING_DAY
+  SCHOOL_HOLIDAYS = 1
+  public_constant :SCHOOL_HOLIDAYS
+  NATIONAL_DAY_OFF = 2
+  public_constant :NATIONAL_DAY_OFF
+  WEEK_END = 3
+  public_constant :WEEK_END
+
+  DATETIME_START = ::Date.parse("#{::GouvCalendarCompilator::YEAR_START}-01-01T00:00:00").freeze
+  public_constant :DATETIME_START
+  DATETIME_END = ::Date.parse("#{::GouvCalendarCompilator::YEAR_END}-12-31T23:59:59").freeze
+  public_constant :DATETIME_END
+  ONE_SECOND = (1.0 / (60 * 24 * 24))
+  public_constant :ONE_SECOND
+  # Base error class for all of the errors raised by this library.
+  class Error < StandardError; end
+
+  # Error related to bad argument given to a method.
+  class ArgumentError < Error; end
+end
+
+require 'gouv_calendar_compilator/data_compilator'
+require 'gouv_calendar_compilator/data_fetcher'
+require 'gouv_calendar_compilator/data_compilators/france/school_holidays_data_compilator'
+require 'gouv_calendar_compilator/data_compilators/france/national_days_off_data_compilator'
+require 'gouv_calendar_compilator/zone_department_matching/france'

data/lib/version.rb

@@ -0,0 +1,11 @@
+#--
+# Copyright (C) 2021 - Octopus Lab SAS
+# All rights are described by the GPLv3 license present in the file /LICENSE available in the
+# original repository of this file or [here](https://www.gnu.org/licenses/gpl-3.0.fr.html).
+#++
+# frozen_string_literal: true
+
+module GouvCalendarCompilator
+  VERSION = '0.0.2'
+  public_constant :VERSION
+end