Add scripts to import videos

Gemfile

@@ -100,11 +100,8 @@ gem "meilisearch-rails"
 gem "ahoy_matey", "~> 4.2"
 gem "vite_rails"
 gem "meta-tags", "~> 2.18"
-
 gem "groupdate", "~> 6.2"
-
 gem "appsignal", "~> 3.4"
-
 gem "chartkick", "~> 5.0"
 
 gem "rails_autolink", "~> 1.1"
@@ -126,3 +123,4 @@ gem "ruby-openai"
 gem "json-repair", "~> 0.2.0"
 
 gem "redcarpet", "~> 3.6"
+gem "country_select", "~> 8.0"

Gemfile.lock

@@ -115,10 +115,11 @@ GEM
     chartkick (5.0.7)
     concurrent-ruby (1.3.3)
     connection_pool (2.4.1)
-    countries (6.0.1)
+    countries (5.7.2)
       unaccent (~> 0.3)
-    crack (1.0.0)
-      bigdecimal
+    country_select (8.0.3)
+      countries (~> 5.0)
+    crack (0.4.5)
       rexml
     crass (1.0.6)
     date (3.3.4)
@@ -177,15 +178,9 @@ GEM
       raabro (~> 1.4)
     globalid (1.2.1)
       activesupport (>= 6.1)
-    google-protobuf (4.27.2-aarch64-linux)
-      bigdecimal
-      rake (>= 13)
     google-protobuf (4.27.2-arm64-darwin)
       bigdecimal
       rake (>= 13)
-    google-protobuf (4.27.2-x86_64-darwin)
-      bigdecimal
-      rake (>= 13)
     google-protobuf (4.27.2-x86_64-linux)
       bigdecimal
       rake (>= 13)
@@ -215,15 +210,9 @@ GEM
     json-repair (0.2.0)
     language_server-protocol (3.17.0.3)
     lint_roller (1.1.0)
-    litestream (0.10.4)
-      logfmt (>= 0.0.10)
-      sqlite3
     litestream (0.10.4-arm64-darwin)
       logfmt (>= 0.0.10)
       sqlite3
-    litestream (0.10.4-x86_64-darwin)
-      logfmt (>= 0.0.10)
-      sqlite3
     litestream (0.10.4-x86_64-linux)
       logfmt (>= 0.0.10)
       sqlite3
@@ -269,12 +258,8 @@ GEM
     net-smtp (0.5.0)
       net-protocol
     nio4r (2.7.3)
-    nokogiri (1.16.6-aarch64-linux)
-      racc (~> 1.4)
     nokogiri (1.16.6-arm64-darwin)
       racc (~> 1.4)
-    nokogiri (1.16.6-x86_64-darwin)
-      racc (~> 1.4)
     nokogiri (1.16.6-x86_64-linux)
       racc (~> 1.4)
     pagy (6.5.0)
@@ -406,9 +391,7 @@ GEM
       fugit (~> 1.11.0)
       railties (>= 7.1)
     sorbet-runtime (0.5.11475)
-    sqlite3 (2.0.2-aarch64-linux-gnu)
     sqlite3 (2.0.2-arm64-darwin)
-    sqlite3 (2.0.2-x86_64-darwin)
     sqlite3 (2.0.2-x86_64-linux-gnu)
     standard (1.39.1)
       language_server-protocol (~> 3.17.0.2)
@@ -437,8 +420,8 @@ GEM
     tzinfo (2.0.6)
       concurrent-ruby (~> 1.0)
     unaccent (0.4.0)
-    unicode-display_width (2.5.0)
-    uri (0.13.0)
+    unicode-display_width (2.4.2)
+    uri (0.12.2)
     useragent (0.16.10)
     vcr (6.2.0)
     view_component (3.12.1)
@@ -471,11 +454,7 @@ GEM
     zeitwerk (2.6.16)
 
 PLATFORMS
-  aarch64-linux
-  arm64-darwin-22
   arm64-darwin-23
-  x86_64-darwin-19
-  x86_64-darwin-23
   x86_64-linux
 
 DEPENDENCIES
@@ -491,6 +470,7 @@ DEPENDENCIES
   capybara
   chartkick (~> 5.0)
   countries
+  country_select (~> 8.0)
   debug
   dockerfile-rails (>= 1.2)
   dotenv-rails

app/clients/youtube/playlists.rb

@@ -1,5 +1,6 @@
 module Youtube
   class Playlists < Youtube::Client
+    DEFAULT_METADATA_PARSER = "Youtube::VideoMetadata"
     def all(channel_id:, title_matcher: nil)
       items = all_items("/playlists", query: {channelId: channel_id, part: "snippet,contentDetails"}).map do |metadata|
         OpenStruct.new({
@@ -10,6 +11,7 @@ def all(channel_id:, title_matcher: nil)
           channel_id: metadata.snippet.channelId,
           year: metadata.snippet.title.match(/\d{4}/).to_s.presence || DateTime.parse(metadata.snippet.publishedAt).year,
           videos_count: metadata.contentDetails.itemCount,
+          metadata_parser: DEFAULT_METADATA_PARSER,
           slug: metadata.snippet.title.parameterize
         })
       end

app/models/event.rb

@@ -26,4 +26,6 @@ class Event < ApplicationRecord
 
   # validations
   validates :name, presence: true
+  VALID_COUNTRY_CODES = ISO3166::Country.codes
+  validates :country_code, inclusion: {in: VALID_COUNTRY_CODES}, allow_nil: true
 end

app/models/youtube/video_metadata.rb

@@ -0,0 +1,92 @@
+# require "active_support/core_ext/hash/keys"
+
+# This class is used to extract the metadata from a youtube video
+# it will try to:
+# - extract the speakers from the title
+# - remove the event_name from the title to make less redondant
+# - remove leading separators from the title
+module Youtube
+  class VideoMetadata
+    SPEAKERS_SECTION_SEPARATOR = " by "
+    SEPARATOR_IN_BETWEEN_SPEAKERS = / & |, | and /
+
+    def initialize(metadata:, event_name:, options: {})
+      @metadata = metadata
+      @event_name = event_name
+    end
+
+    def cleaned
+      OpenStruct.new(
+        {
+          title: title,
+          raw_title: raw_title,
+          speakers: speakers,
+          event_name: @event_name,
+          published_at: @metadata.published_at,
+          description: description,
+          video_id: @metadata.video_id
+        }
+      )
+    end
+
+    def keynote?
+      title_without_event_name.match(/keynote/i)
+    end
+
+    def lighting?
+      title_without_event_name.match(/lightning talks/i)
+    end
+
+    private
+
+    def extract_info_from_title
+      title_parts = title_without_event_name.split(SPEAKERS_SECTION_SEPARATOR)
+      speakers = title_parts.last.split(SEPARATOR_IN_BETWEEN_SPEAKERS).map(&:strip)
+      title = title_parts[0..-2].join(SPEAKERS_SECTION_SEPARATOR).gsub(/^\s*-/, "").strip
+
+      {
+        title: keynote? ? remove_leading_and_trailing_separators_from(title_without_event_name) : remove_leading_and_trailing_separators_from(title),
+        speakers: speakers
+      }
+    end
+
+    def speakers
+      return [] if lighting?
+
+      title_parts = title_without_event_name.split(SPEAKERS_SECTION_SEPARATOR)
+      title_parts.last.split(SEPARATOR_IN_BETWEEN_SPEAKERS).map(&:strip)
+    end
+
+    def raw_title
+      @metadata.title
+    end
+
+    def title_without_event_name
+      # RubyConf AU 2013: From Stubbies to Longnecks by Geoffrey Giesemann
+      # will return "From Stubbies to Longnecks by Geoffrey Giesemann"
+      remove_leading_and_trailing_separators_from(raw_title.gsub(@event_name, "").gsub(/\s+/, " "))
+    end
+
+    ## remove : - and other separators from the title
+    def remove_leading_and_trailing_separators_from(title)
+      title.gsub(/^[-:]?/, "").strip.then do |title|
+        title.gsub(/[-:,]$/, "").strip
+      end
+    end
+
+    def title
+      if keynote? || lighting?
+        # when it is a keynote or lighting, usually we want to keep the full title without the event name
+        remove_leading_and_trailing_separators_from(title_without_event_name)
+      else
+        title_parts = title_without_event_name.split(SPEAKERS_SECTION_SEPARATOR)
+        title = title_parts[0..-2].join(SPEAKERS_SECTION_SEPARATOR).gsub(/^\s*-/, "").strip
+        remove_leading_and_trailing_separators_from(title)
+      end
+    end
+
+    def description
+      @metadata.description
+    end
+  end
+end

app/models/youtube/video_metadata_rails_world.rb

@@ -0,0 +1,84 @@
+# require "active_support/core_ext/hash/keys"
+
+# This class is used to extract the metadata from a youtube video
+# it will try to:
+# - extract the speakers from the title
+# - remove the event_name from the title to make less redondant
+# - remove leading separators from the title
+module Youtube
+  class VideoMetadataRailsWorld
+    SPEAKERS_SECTION_SEPARATOR = " - "
+    SEPARATOR_IN_BETWEEN_SPEAKERS = / & |, | and /
+
+    def initialize(metadata:, event_name:, options: {})
+      @metadata = metadata
+      @event_name = event_name
+    end
+
+    def cleaned
+      OpenStruct.new(
+        {
+          title: title,
+          raw_title: raw_title,
+          speakers: speakers,
+          event_name: @event_name,
+          published_at: @metadata.published_at,
+          description: description,
+          video_id: @metadata.video_id
+        }
+      )
+    end
+
+    def keynote?
+      title_without_event_name.match(/keynote/i)
+    end
+
+    private
+
+    def extract_info_from_title
+      title_parts = title_without_event_name.split(SPEAKERS_SECTION_SEPARATOR)
+      speakers = title_parts.last.split(SEPARATOR_IN_BETWEEN_SPEAKERS).map(&:strip)
+      title = title_parts[0..-2].join(SPEAKERS_SECTION_SEPARATOR).gsub(/^\s*-/, "").strip
+
+      {
+        title: keynote? ? remove_leading_and_trailing_separators_from(title_without_event_name) : remove_leading_and_trailing_separators_from(title),
+        speakers: speakers
+      }
+    end
+
+    def speakers
+      raw_title_parts.first.split(SEPARATOR_IN_BETWEEN_SPEAKERS).map(&:strip)
+    end
+
+    def raw_title
+      @metadata.title
+    end
+
+    def title_without_event_name
+      # RubyConf AU 2013: From Stubbies to Longnecks by Geoffrey Giesemann
+      # will return "From Stubbies to Longnecks by Geoffrey Giesemann"
+      remove_leading_and_trailing_separators_from(raw_title.gsub(@event_name, "").gsub(/\s+/, " "))
+    end
+
+    ## remove : - and other separators from the title
+    def remove_leading_and_trailing_separators_from(title)
+      return title if title.blank?
+
+      title.gsub(/^[-:]?/, "").strip.then do |title|
+        title.gsub(/[-:,]$/, "").strip
+      end
+    end
+
+    def title
+      remove_leading_and_trailing_separators_from(raw_title_parts[1])
+    end
+
+    def description
+      @metadata.description
+    end
+
+    def raw_title_parts
+      title_without_event_name.split(SPEAKERS_SECTION_SEPARATOR)
+    end
+  end
+end

config/credentials.yml.enc

@@ -0,0 +1 @@
+uAKB/HPZ9hm7lyAIpWgUe+24TG/N8bi2smtvEa7aD0nZNIU1sb0ctVEYNcSQbKJjHWSED68aF9/O7iYBt1QYU93in3J14rDvj0bHJ8GCWNEQFI6QekRPcOVrX4qehGn4Y29ikhsrc1/th6V0EHrkP4q3d+AtK0zazPLpLQgMFSMX5dYjck2vT8HfK3nOvX77+RUPXhyyO9roIBDBpSf8kEayzUi0q/h/k8cGsqbM3aL2InfXNM8gaQgJ7atEAzJsKoyt5yGoFxjWTDoOFEqnIIhXu1uRmW+S5aKUC/h0Kil/nEgsvufZGuPiDQgHvXZmO/3W3MmaCLaO2LHcp4qMySaH5SDBOXdzAmGuEl0SMhdyvtuZBCw60WmWRhh9GnhwBwJs/FdlaGnAGLZyB+lQhYYNTOww--9Tq1/zY6N+kujjV9--iOafU6Imt5fluW+yk+MPWQ==