Script: Split "Subspecies of" history items

[jonkerz]

• Unsplit: results

• Re-run Script: Replace hardcoded names with missing2 tags #1113

[jonkerz]

• Run script - https://antcat.org/activities/92936

Batch 1

BATCH = 1
EDIT_SUMMARY = "Split 'Subspecies of' history items [batch #{BATCH}], see %github1120"
ANTCATBOT = User.find_by!(name: 'AntCatBot')

# For activities and PaperTrail.
RequestStore.store[:current_request_uuid] = SecureRandom.uuid
PaperTrail.request.whodunnit = ANTCATBOT.id

def split_subspecies_of_item! old_item, expected_tax:, dry_run: false
  old_taxt = old_item.taxt.dup
  split = old_taxt.split('; of ')

  num_tax_tags = old_taxt.scan('{tax').size

  if expected_tax == :balanced
    raise "tax and 'in' unbalanced" unless num_tax_tags == split.size
  else
   raise "should be #{expected_tax} tax tag" unless old_taxt.scan('{tax').size == expected_tax
   raise "should be #{expected_tax} '; of'" unless split.size == expected_tax
  end

  taxt_to_update = split.first
  taxt_to_update << '.' unless taxt_to_update.ends_with?('.')

  tail_taxts = split[1..]
  taxts_to_create = tail_taxts.map do |taxt_to_create_part|
    string = "Subspecies of " << taxt_to_create_part
    string << '.' unless string.ends_with?('.')
    string
  end

  puts "ID: #{old_item.id}"
  puts "num_tax_tags: #{num_tax_tags}"
  puts "split.size: #{split.size}"
  puts "original: #{old_taxt}".yellow
  puts "update:   #{taxt_to_update}".green
  taxts_to_create.each do |taxt_to_create|
    puts "create:   #{taxt_to_create}".blue
  end
  puts

  return if dry_run

  old_item.update!(taxt: taxt_to_update)
  new_items = taxts_to_create.reverse.map do |taxt_to_create|
    new_item = TaxonHistoryItem.create!(taxon: old_item.taxon, taxt: taxt_to_create, position: old_item.position + 1)
    create_activity new_item, :create, "from ##{old_item.id}"
  end

  create_activity old_item, :update, "to ##{new_items.map(&:id)}"
end

def create_activity history_item, action, item_edit_summary
  history_item.create_activity(
    action,
    ANTCATBOT,
    edit_summary: "Split 'Subspecies of' history items (split #{item_edit_summary}) [batch #{BATCH}], see %github1120"
  )
end

def regex_split! mysql_regex
  items = TaxonHistoryItem.where("taxt REGEXP ?", mysql_regex); nil
  item_ids = items.pluck(:id); nil # Hack for `dry_run`.

  puts "items: #{items.count}".yellow
  puts "item_ids: #{item_ids.size}".yellow
  puts "total history items: #{TaxonHistoryItem.count}".yellow

  TaxonHistoryItem.where(id: item_ids).find_each do |history_item|
    split_subspecies_of_item! history_item, expected_tax: :balanced, dry_run: false
  end; nil

  puts "items: #{items.count}".yellow
  puts "item_ids: #{item_ids.size}".yellow
  puts "total history items: #{TaxonHistoryItem.count}".yellow
end

tax = "{tax [0-9]+}"
tax_ish = "({tax(ac)? [0-9]+}|{missing[0-9]? .+}|[A-z][a-z]+ [a-z]+( [a-z]+)?)"
one_or_many_tax_ish = "(#{tax_ish}|(#{tax_ish}, )+#{tax_ish}?)"

ref = '({ref [0-9]+}: [0-9a-z]+)'
one_or_many_refs = "(#{ref}|(#{ref}; )+#{ref}?)"

ref_w_extras = '({ref [0-9]+}: [0-9a-z]+( \((in key|in text|footnote)\))?)'
one_or_many_refs_w_extras = "(#{ref_w_extras}|(#{ref_w_extras}; )+#{ref_w_extras}?)"

# ---

regex_split! "^Subspecies of #{tax}: #{one_or_many_refs}; of #{tax}: #{one_or_many_refs}$"
regex_split! "^Subspecies of #{tax}: #{one_or_many_refs}; of #{tax}: #{one_or_many_refs}.$"

regex_split! "^Subspecies of #{tax}: #{one_or_many_refs};( of #{tax}: #{one_or_many_refs};)+ of #{tax}: #{one_or_many_refs}$"
regex_split! "^Subspecies of #{tax}: #{one_or_many_refs};( of #{tax}: #{one_or_many_refs};)+ of #{tax}: #{one_or_many_refs}.$"

# ---

regex_split! "^Subspecies of #{tax}: #{one_or_many_refs_w_extras}; of #{tax}: #{one_or_many_refs_w_extras}$"
regex_split! "^Subspecies of #{tax}: #{one_or_many_refs_w_extras}; of #{tax}: #{one_or_many_refs_w_extras}.$"

regex_split! "^Subspecies of #{tax}: #{one_or_many_refs_w_extras};( of #{tax}: #{one_or_many_refs_w_extras};)+ of #{tax}: #{one_or_many_refs_w_extras}$"
regex_split! "^Subspecies of #{tax}: #{one_or_many_refs_w_extras};( of #{tax}: #{one_or_many_refs_w_extras};)+ of #{tax}: #{one_or_many_refs_w_extras}.$"

# ---

Activity.execute_script_activity User.find_by!(name: 'Fredrik Palmkron'), EDIT_SUMMARY

[jonkerz]

• Run script - https://antcat.org/activities/92964

Batch 2

BATCH = 2
EDIT_SUMMARY = "Split 'Subspecies of' history items [batch #{BATCH}], see %github1120"
ANTCATBOT = User.find_by!(name: 'AntCatBot')

# For activities and PaperTrail.
RequestStore.store[:current_request_uuid] = SecureRandom.uuid
PaperTrail.request.whodunnit = ANTCATBOT.id

def split_subspecies_of_item! old_item, expected_tax: :unknown, dry_run: false
  old_taxt = old_item.taxt.dup
  split = old_taxt.split('; of ')

  taxt_to_update = split.first
  taxt_to_update << '.' unless taxt_to_update.ends_with?('.')

  tail_taxts = split[1..]
  taxts_to_create = tail_taxts.map do |taxt_to_create_part|
    string = "Subspecies of " << taxt_to_create_part
    string << '.' unless string.ends_with?('.')
    string
  end

  puts "ID: #{old_item.id}"
  puts "split.size: #{split.size}"
  puts "original: #{old_taxt}".yellow
  puts "update:   #{taxt_to_update}".green
  taxts_to_create.each do |taxt_to_create|
    puts "create:   #{taxt_to_create}".blue
  end
  puts

  return if dry_run

  old_item.update!(taxt: taxt_to_update)
  new_items = taxts_to_create.reverse.map do |taxt_to_create|
    new_item = TaxonHistoryItem.create!(taxon: old_item.taxon, taxt: taxt_to_create, position: old_item.position + 1)
    create_activity new_item, :create, "from ##{old_item.id}"
  end

  create_activity old_item, :update, "to ##{new_items.map(&:id)}"
end

def create_activity history_item, action, item_edit_summary
  history_item.create_activity(
    action,
    ANTCATBOT,
    edit_summary: "Split 'Subspecies of' history items (split #{item_edit_summary}) [batch #{BATCH}], see %github1120"
  )
end

def regex_split! mysql_regex
  items = TaxonHistoryItem.where("taxt REGEXP ?", mysql_regex); nil
  item_ids = items.pluck(:id); nil # Hack for `dry_run`.

  puts "items: #{items.count}".yellow
  puts "item_ids: #{item_ids.size}".yellow
  puts "total history items: #{TaxonHistoryItem.count}".yellow

  TaxonHistoryItem.where(id: item_ids).find_each do |history_item|
    split_subspecies_of_item! history_item, expected_tax: :balanced, dry_run: false
  end; nil

  puts "items: #{items.count}".yellow
  puts "item_ids: #{item_ids.size}".yellow
  puts "total history items: #{TaxonHistoryItem.count}".yellow
end

tax = "{tax [0-9]+}"
tax_ish = "({tax(ac)? [0-9]+}|{missing[0-9]? .+}|[A-z][a-z]+ [a-z]+( [a-z]+)?)"
one_or_many_tax_ish = "(#{tax_ish}|(#{tax_ish}, )+#{tax_ish}?)"

ref = '({ref [0-9]+}: [0-9a-z]+)'
one_or_many_refs = "(#{ref}|(#{ref}; )+#{ref}?)"

ref_w_extras = '({ref [0-9]+}: [0-9a-z]+( \((in key|in text|footnote)\))?)'
one_or_many_refs_w_extras = "(#{ref_w_extras}|(#{ref_w_extras}; )+#{ref_w_extras}?)"

# ---

regex_split! "^Subspecies of #{tax_ish}: #{one_or_many_refs}; of #{tax_ish}: #{one_or_many_refs}$"
regex_split! "^Subspecies of #{tax_ish}: #{one_or_many_refs}; of #{tax_ish}: #{one_or_many_refs}.$"

regex_split! "^Subspecies of #{tax_ish}: #{one_or_many_refs};( of #{tax_ish}: #{one_or_many_refs};)+ of #{tax_ish}: #{one_or_many_refs}$"
regex_split! "^Subspecies of #{tax_ish}: #{one_or_many_refs};( of #{tax_ish}: #{one_or_many_refs};)+ of #{tax_ish}: #{one_or_many_refs}.$"

# ---

regex_split! "^Subspecies of #{tax_ish}: #{one_or_many_refs_w_extras}; of #{tax_ish}: #{one_or_many_refs_w_extras}$"
regex_split! "^Subspecies of #{tax_ish}: #{one_or_many_refs_w_extras}; of #{tax_ish}: #{one_or_many_refs_w_extras}.$"

regex_split! "^Subspecies of #{tax_ish}: #{one_or_many_refs_w_extras};( of #{tax_ish}: #{one_or_many_refs_w_extras};)+ of #{tax_ish}: #{one_or_many_refs_w_extras}$"
regex_split! "^Subspecies of #{tax_ish}: #{one_or_many_refs_w_extras};( of #{tax_ish}: #{one_or_many_refs_w_extras};)+ of #{tax_ish}: #{one_or_many_refs_w_extras}.$"

# ---

Activity.execute_script_activity User.find_by!(name: 'Fredrik Palmkron'), EDIT_SUMMARY