date_extractor 0.1.0 → 0.1.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 6ccaf1114d7c608b8f2cccb38d902db374839807
-  data.tar.gz: f101665c1df1a41c6c4b7eeda73de833e5ff6ebf
+  metadata.gz: 2b729b3c4a7d4f759cb83eb4f3caa8b28290556e
+  data.tar.gz: 0a3921dc386b53084a8839ad6132dabb5eeaacf4
 SHA512:
-  metadata.gz: 46b468a50ed84fbf06888fff0447acd7ad602aaf2919e89e9d686cd098240ed28b4be8d740553370a28d2d70a9299479e1d5912f1aba4246b4a3327eb2f1b8a1
-  data.tar.gz: 58f2827dc897ee64acde531cd6dbd77a5fbb144abcf7a6c24a5e19aa188d57cad7d7557fffc92e60f2058fcf7500141a997bb55616249bfaeda54b7e4f109868
+  metadata.gz: 436e6fbffd2ffbec1c5693fc2761efcf004e0bb089c9e92e5f48f3bddb0b154029f0f9d9e2e0a86f23ba9c29971a8b1916ad6b1dd58a5b726c3247da270b12b0
+  data.tar.gz: e6b70609de5f70b98570cc12d00688e86c408957ef51de3ae33915a5a1b0eb8c2bc7357e6c46de36fb5dc5313f433b5029efaf016dc60e8f9465b689a7584904

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    date_extractor (0.1.0)
+    date_extractor (0.1.1)
 
 GEM
   remote: https://rubygems.org/

data/date_extractor.gemspec

@@ -11,7 +11,7 @@ Gem::Specification.new do |spec|
 
   spec.summary       = %q{Extract dates from a text.}
   spec.description   = %q{Extract dates from a text.}
-  spec.homepage      = "https://github.com/south37"
+  spec.homepage      = "https://github.com/south37/date_extractor"
   spec.license       = "MIT"
 
   spec.files         = `git ls-files -z`.split("\x0").reject do |f|

data/lib/date_extractor/version.rb

@@ -1,3 +1,3 @@
 module DateExtractor
-  VERSION = "0.1.0"
+  VERSION = "0.1.1"
 end

data/lib/date_extractor.rb

@@ -111,195 +111,199 @@ module DateExtractor
 
   RE = /(?:#{DAY_RE})|(?:#{ONLY_DAY_RE})/x
 
-  def self.get_match_and_positions(body, re)
-    body.to_enum(:scan, re).map { [Regexp.last_match, Regexp.last_match.begin(0), Regexp.last_match.end(0)] }
-  end
+  class << self
+    # @param [String] body
+    # @param [Integer | NilClass] fallback_month
+    # @param [Integer | NilClass] fallback_year
+    # @param [Boolean] debug
+    # @return [[String], [[Date, DateTime | NilClass, DateTime | NilClass]] matched strings and dates
+    def extract(body, fallback_month: nil, fallback_year: nil, debug: false)
+      today = Date.today
+      fallback_month ||= Date.today.month
+      fallback_year  ||= Date.today.year
+
+      day_matches = get_match_and_positions(body, RE)  # [[MatchData, start, end], [...], ...]
+
+      day_with_hours = days_from_matches(day_matches.map(&:first), fallback_month, fallback_year, debug: debug)  # [[MatchData, Date, DateTime, DateTime], [MatchData, Date, DateTime, nil]...]
+      day_with_hours_size = day_matches.size
+
+      timeslots_container = Array.new(day_with_hours_size) { Array.new }  # contains timeslots in each day
+
+      timeslot_matches = get_match_and_positions(body, TIMESLOT_RE)  # [[MatchData, start, end], [...], ...]
+      timeslot_matches.each do |(timeslot_match, start_pos, end_pos)|
+        i = 0  # index of left_day
+
+        while i < day_with_hours_size
+          left_day = day_with_hours[i]
+          if left_day[1].nil?  # If failed to `Date.new(~)`, nil is set to left_day[1] which is `Date`
+            i += 1
+            next end
+
+          right_day = day_with_hours[i+1]
+          if !right_day.nil? && right_day[1].nil?  # When failed to `Date.new(~)`
+            right_day = day_with_hours[i+2]
+          end
 
-  def self.get_hour_from_timeslot_match(match)
-    begin
-      start_hour = to_downer_letter(match[:start_hour])
-    rescue
-      start_hour = nil
-    end
+          if right_day.nil?  # left_day is on the last
+            # Check if timeslot is on the right of left_day
+            if left_day[0].end(0) <= start_pos
+              timeslots_container[i].push timeslot_match
+            end
+          else
+            # Check if timeslot is between left_day and right_day
+            if left_day[0].end(0) <= start_pos && (end_pos - 1) < right_day[0].begin(0)
+              timeslots_container[i].push timeslot_match
+            end
+          end
 
-    begin
-      start_min = to_downer_letter(match[:start_min])
-    rescue
-      if match.names.include?('start_half_hour_unit') && match[:start_half_hour_unit] == '半'
-        start_min = 30
-      else
-        start_min = nil
+          i += 1
+        end
       end
-    end
 
-    begin
-      end_hour = to_downer_letter(match[:end_hour])
-    rescue
-      end_hour = nil
-    end
+      days_from_timeslots = days_from_timeslot_matches(timeslots_container, day_with_hours)  # days contains day whidh has same index with timeslots_container
+
+      result_datetimes = days_from_timeslots.map { |(match, day, start_t, end_t)| [day, start_t, end_t] }
+      result_strs      = days_from_timeslots.map { |(match, _, _, _)| match&.[](0) }
 
-    begin
-      end_min = to_downer_letter(match[:end_min])
-    rescue
-      if match.names.include?('end_half_hour_unit') && match[:end_half_hour_unit] == '半'
-        end_min = 30
+      if !debug  # Reject nil dates
+        exists           = result_datetimes.map { |arr| !arr[0].nil? }
+        result_strs      = result_strs.select.with_index { |str, i| exists[i] }
+        result_datetimes = result_datetimes.select.with_index { |arr, i| exists[i] }
+        [result_strs, result_datetimes]
       else
-        end_min = nil
+        [result_strs, result_datetimes]
       end
     end
 
-    [start_hour, start_min, end_hour, end_min]
-  end
+  private
+
+    def get_match_and_positions(body, re)
+      body.to_enum(:scan, re).map { [Regexp.last_match, Regexp.last_match.begin(0), Regexp.last_match.end(0)] }
+    end
 
-  def self.create_datetime_if_exists(year, month, day, hour, min)
-    if !hour.nil?
+    def get_hour_from_timeslot_match(match)
       begin
-        result = DateTime.new(year, month, day, hour.to_i, min.to_i)
+        start_hour = to_downer_letter(match[:start_hour])
       rescue
-        result = nil
+        start_hour = nil
       end
-    else
-      result = nil
-    end
-    result
-  end
 
-  # @return [[MatchData, Date | NilClass, DateTime | NilClass, DateTime | NilClass]]
-  # If month is not specified, fallback_month is used as month. This value is
-  # updated by discovering other month specification. Same for fallback_year.
-  def self.days_from_matches(matches, fallback_month, fallback_year, debug: false)
-    matches.map do |match|
       begin
-        year = to_downer_letter(match[:year])
-        fallback_year = year
+        start_min = to_downer_letter(match[:start_min])
       rescue
-        year = fallback_year
+        if match.names.include?('start_half_hour_unit') && match[:start_half_hour_unit] == '半'
+          start_min = 30
+        else
+          start_min = nil
+        end
       end
 
-      # When ONLY_DAY_RE is used, month is nil
       begin
-        month = to_downer_letter(match[:month]).to_i
-        fallback_month = month
+        end_hour = to_downer_letter(match[:end_hour])
       rescue
-        month = fallback_month
+        end_hour = nil
       end
 
-      day = to_downer_letter(match[:day]).to_i
-
-      start_hour, start_min, end_hour, end_min = get_hour_from_timeslot_match(match)
-
       begin
-        date = Date.new(year, month, day)
+        end_min = to_downer_letter(match[:end_min])
       rescue
-        date = nil
+        if match.names.include?('end_half_hour_unit') && match[:end_half_hour_unit] == '半'
+          end_min = 30
+        else
+          end_min = nil
+        end
       end
 
-      start_t = self.create_datetime_if_exists(year, month, day, start_hour, start_min)
-      end_t   = self.create_datetime_if_exists(year, month, day, end_hour, end_min)
+      [start_hour, start_min, end_hour, end_min]
+    end
 
-      if !date.nil?
-        [match, date, start_t, end_t]
+    def create_datetime_if_exists(year, month, day, hour, min)
+      if !hour.nil?
+        begin
+          result = DateTime.new(year, month, day, hour.to_i, min.to_i)
+        rescue
+          result = nil
+        end
       else
-        [match, nil, nil, nil]
+        result = nil
       end
+      result
     end
-  end
 
-  # days contains day whidh has same index with timeslots_container
-  def self.days_from_timeslot_matches(timeslots_container, day_with_hours)
-    result = []
-
-    day_with_hours.each_with_index do |day_with_hour, i|
-      result.push(day_with_hour)
-      _, day, _, _ = day_with_hour  #
-      next if day.nil?
+    # @return [[MatchData, Date | NilClass, DateTime | NilClass, DateTime | NilClass]]
+    # If month is not specified, fallback_month is used as month. This value is
+    # updated by discovering other month specification. Same for fallback_year.
+    def days_from_matches(matches, fallback_month, fallback_year, debug: false)
+      matches.map do |match|
+        begin
+          year = to_downer_letter(match[:year])
+          fallback_year = year
+        rescue
+          year = fallback_year
+        end
 
-      timeslot_matches = timeslots_container[i]
-      next if (timeslot_matches.size == 0)
+        # When ONLY_DAY_RE is used, month is nil
+        begin
+          month = to_downer_letter(match[:month]).to_i
+          fallback_month = month
+        rescue
+          month = fallback_month
+        end
 
-      timeslot_matches.each do |timeslot_match|
-        start_hour, start_min, end_hour, end_min = self.get_hour_from_timeslot_match(timeslot_match)
+        day = to_downer_letter(match[:day]).to_i
 
-        start_t = self.create_datetime_if_exists(day.year, day.month, day.day, start_hour, start_min)
-        end_t   = self.create_datetime_if_exists(day.year, day.month, day.day, end_hour, end_min)
+        start_hour, start_min, end_hour, end_min = get_hour_from_timeslot_match(match)
 
-        result.push([timeslot_match, day, start_t, end_t])
-      end
-    end
+        begin
+          date = Date.new(year, month, day)
+        rescue
+          date = nil
+        end
 
-    result
-  end
+        start_t = create_datetime_if_exists(year, month, day, start_hour, start_min)
+        end_t   = create_datetime_if_exists(year, month, day, end_hour, end_min)
 
-  def self.to_downer_letter(upper_or_downer_letter)
-    upper_or_downer_letter.split('').map do |c|
-      if /[０-９]/.match(c)
-        (c.ord - "０".ord).to_s
-      else
-        c
+        if !date.nil?
+          [match, date, start_t, end_t]
+        else
+          [match, nil, nil, nil]
+        end
       end
-    end.join
-  end
-
-  # @param [String] body
-  # @param [Integer | NilClass] fallback_month
-  # @param [Integer | NilClass] fallback_year
-  # @param [Boolean] debug
-  # @return [[String], [[Date, DateTime | NilClass, DateTime | NilClass]] matched strings and dates
-  def self.extract(body, fallback_month: nil, fallback_year: nil, debug: false)
-    today = Date.today
-    fallback_month ||= Date.today.month
-    fallback_year  ||= Date.today.year
-
-    day_matches = get_match_and_positions(body, RE)  # [[MatchData, start, end], [...], ...]
+    end
 
-    day_with_hours = days_from_matches(day_matches.map(&:first), fallback_month, fallback_year, debug: debug)  # [[MatchData, Date, DateTime, DateTime], [MatchData, Date, DateTime, nil]...]
-    day_with_hours_size = day_matches.size
+    # days contains day whidh has same index with timeslots_container
+    def days_from_timeslot_matches(timeslots_container, day_with_hours)
+      result = []
 
-    timeslots_container = Array.new(day_with_hours_size) { Array.new }  # contains timeslots in each day
+      day_with_hours.each_with_index do |day_with_hour, i|
+        result.push(day_with_hour)
+        _, day, _, _ = day_with_hour  #
+        next if day.nil?
 
-    timeslot_matches = get_match_and_positions(body, TIMESLOT_RE)  # [[MatchData, start, end], [...], ...]
-    timeslot_matches.each do |(timeslot_match, start_pos, end_pos)|
-      i = 0  # index of left_day
+        timeslot_matches = timeslots_container[i]
+        next if (timeslot_matches.size == 0)
 
-      while i < day_with_hours_size
-        left_day = day_with_hours[i]
-        if left_day[1].nil?  # If failed to `Date.new(~)`, nil is set to left_day[1] which is `Date`
-          i += 1
-          next end
+        timeslot_matches.each do |timeslot_match|
+          start_hour, start_min, end_hour, end_min = get_hour_from_timeslot_match(timeslot_match)
 
-        right_day = day_with_hours[i+1]
-        if !right_day.nil? && right_day[1].nil?  # When failed to `Date.new(~)`
-          right_day = day_with_hours[i+2]
-        end
+          start_t = create_datetime_if_exists(day.year, day.month, day.day, start_hour, start_min)
+          end_t   = create_datetime_if_exists(day.year, day.month, day.day, end_hour, end_min)
 
-        if right_day.nil?  # left_day is on the last
-          # Check if timeslot is on the right of left_day
-          if left_day[0].end(0) <= start_pos
-            timeslots_container[i].push timeslot_match
-          end
-        else
-          # Check if timeslot is between left_day and right_day
-          if left_day[0].end(0) <= start_pos && (end_pos - 1) < right_day[0].begin(0)
-            timeslots_container[i].push timeslot_match
-          end
+          result.push([timeslot_match, day, start_t, end_t])
         end
-
-        i += 1
       end
-    end
-
-    days_from_timeslots = days_from_timeslot_matches(timeslots_container, day_with_hours)  # days contains day whidh has same index with timeslots_container
 
-    result_datetimes = days_from_timeslots.map { |(match, day, start_t, end_t)| [day, start_t, end_t] }
-    result_strs      = days_from_timeslots.map { |(match, _, _, _)| match&.[](0) }
+      result
+    end
 
-    if !debug  # Reject nil dates
-      exists           = result_datetimes.map { |arr| !arr[0].nil? }
-      result_strs      = result_strs.select.with_index { |str, i| exists[i] }
-      result_datetimes = result_datetimes.select.with_index { |arr, i| exists[i] }
-      [result_strs, result_datetimes]
-    else
-      [result_strs, result_datetimes]
+    def to_downer_letter(upper_or_downer_letter)
+      upper_or_downer_letter.split('').map do |c|
+        if /[０-９]/.match(c)
+          (c.ord - "０".ord).to_s
+        else
+          c
+        end
+      end.join
     end
   end
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: date_extractor
 version: !ruby/object:Gem::Version
-  version: 0.1.0
+  version: 0.1.1
 platform: ruby
 authors:
 - Nao Minami
@@ -85,7 +85,7 @@ files:
 - date_extractor.gemspec
 - lib/date_extractor.rb
 - lib/date_extractor/version.rb
-homepage: https://github.com/south37
+homepage: https://github.com/south37/date_extractor
 licenses:
 - MIT
 metadata: {}