migration to remove duplicated github handles

adrienpoly · Feb 22, 2025 · bfa3644 · bfa3644
1 parent 486fdea
commit bfa3644
Show file tree

Hide file tree

Showing 4 changed files with 33 additions and 8 deletions.
diff --git a/app/models/speaker.rb b/app/models/speaker.rb
@@ -7,7 +7,7 @@
 #  bio             :text             default(""), not null
 #  bsky            :string           default(""), not null
 #  bsky_metadata   :json             not null
-#  github          :string           default(""), not null
+#  github          :string           default(""), not null, indexed
 #  github_metadata :json             not null
 #  linkedin        :string           default(""), not null
 #  mastodon        :string           default(""), not null
@@ -26,6 +26,7 @@
 # Indexes
 #
 #  index_speakers_on_canonical_id  (canonical_id)
+#  index_speakers_on_github        (github) UNIQUE WHERE github IS NOT NULL AND github != ''
 #  index_speakers_on_name          (name)
 #  index_speakers_on_slug          (slug) UNIQUE
 #
@@ -66,7 +67,7 @@ class Speaker < ApplicationRecord
 
   # validations
   validates :canonical, exclusion: {in: ->(speaker) { [speaker] }, message: "can't be itself"}
-  validates :github, uniqueness: {case_sensitive: false}
+  validates :github, uniqueness: {case_sensitive: false}, allow_blank: true
 
   # scope
   scope :with_talks, -> { where.not(talks_count: 0) }

diff --git a/db/migrate/20250128070756_add_unique_index_on_github_in_speaker.rb b/db/migrate/20250128070756_add_unique_index_on_github_in_speaker.rb
@@ -1,8 +1,30 @@
 class AddUniqueIndexOnGitHubInSpeaker < ActiveRecord::Migration[8.0]
-  def change
-    # remove all github handles for speakers having a canonical speaker
-    Speaker.where.not(canonical_id: nil).update_all(github: "")
-    puts Speaker.group(:github).having("count(*) > 1").pluck(:github)
-    add_index :speakers, :github, unique: true
+  def up
+    # remove duplicate with no talks
+    Speaker.group(:github).having("count(*) > 1").pluck(:github).each do |github|
+      Speaker.where(github: github, talks_count: 0).update_all(github: "")
+    end
+
+    # some speaker with a canonical speaker still have talks attached to them let's reasign them
+    Speaker.not_canonical.where.not(talks_count: 0).each do |speaker|
+      speaker.assign_canonical_speaker!(canonical_speaker: speaker.canonical)
+    end
+
+    # fix one by one
+    Speaker.find_by(name: "Andrew").assign_canonical_speaker!(canonical_speaker: Speaker.find_by(name: "Andrew Nesbitt"))
+    Speaker.find_by(name: "HASUMI Hitoshi").assign_canonical_speaker!(canonical_speaker: Speaker.find_by(name: "Hitoshi Hasumi"))
+    Speaker.find_by(name: "hogelog").assign_canonical_speaker!(canonical_speaker: Speaker.find_by(name: "Sunao Hogelog Komuro"))
+    Speaker.find_by(name: "Jônatas Paganini").assign_canonical_speaker!(canonical_speaker: Speaker.find_by(name: "Jônatas Davi Paganini"))
+    Speaker.find_by(name: "Sutou Kouhei").assign_canonical_speaker!(canonical_speaker: Speaker.find_by(name: "Kouhei Sutou"))
+    Speaker.find_by(slug: "maciek-rzasa").assign_canonical_speaker!(canonical_speaker: Speaker.find_by(slug: "maciej-rzasa"))
+    Speaker.find_by(slug: "mario-alberto-chavez").assign_canonical_speaker!(canonical_speaker: Speaker.find_by(slug: "mario-chavez"))
+    Speaker.find_by(slug: "enrique-morellon").assign_canonical_speaker!(canonical_speaker: Speaker.find_by(slug: "enrique-mogollan"))
+    Speaker.find_by(slug: "masafumi-okura").assign_canonical_speaker!(canonical_speaker: Speaker.find_by(slug: "okura-masafumi"))
+    Speaker.find_by(slug: "oliver-lacan").assign_canonical_speaker!(canonical_speaker: Speaker.find_by(slug: "olivier-lacan"))
+    add_index :speakers, :github, unique: true, where: "github IS NOT NULL AND github != ''"
+  end
+
+  def down
+    remove_index :speakers, :github
   end
 end
diff --git a/db/schema.rb b/db/schema.rb
diff --git a/test/fixtures/speakers.yml b/test/fixtures/speakers.yml
@@ -7,7 +7,7 @@
 #  bio             :text             default(""), not null
 #  bsky            :string           default(""), not null
 #  bsky_metadata   :json             not null
-#  github          :string           default(""), not null
+#  github          :string           default(""), not null, indexed
 #  github_metadata :json             not null
 #  linkedin        :string           default(""), not null
 #  mastodon        :string           default(""), not null
@@ -26,6 +26,7 @@
 # Indexes
 #
 #  index_speakers_on_canonical_id  (canonical_id)
+#  index_speakers_on_github        (github) UNIQUE WHERE github IS NOT NULL AND github != ''
 #  index_speakers_on_name          (name)
 #  index_speakers_on_slug          (slug) UNIQUE
 #