Example semi-automated bot merge

kite_runner_merge.py

# Steps to merge a number of Kite Runner editions and works:

from catharbot import catharbot
bot = catharbot.CatharBot()

def merge_unique_lists(lists, hash_fn=None):
    """
    Combine unique lists into a new unique list. Preserves ordering.
    """
    result = []
    seen = set()
    for lst in lists:
        for el in lst:
            hsh = hash_fn(el) if hash_fn else el
            if hsh not in seen:
                result.append(el)
                seen.add(hsh)
    return result

orphan_olids = ['OL24057181M', 'OL24237802M']
master_olid = 'OL5782001W' # Chosen as master because: Most Editions, On the most Lists, on Staff Picks List
dupe_olids = [ 'OL5781992W',
               'OL17174243W',
               'OL16068777W',
               'OL17348501W',
               'OL16049475W'
             ]

# Move orphans to master work
orphans = [ bot.get_move_edition(e, master_olid) for e in orphan_olids ]
bot.save_many(orphans, "Associate with work")

master = bot.load_doc(master_olid)

edition_with_desc = bot.load_doc(orphan_olids[1]) # OL24237802M has a good description, use that on the master work

master['description'] = edition_with_desc['description']
master['title'] = "The Kite Runner"   # Title case the master work

# Update lists on master from all duplicates...
for w in dupe_olids:
  dupe = bot.load_doc(w)
  for prop in ['subjects', 'covers', 'subject_places']:
      master[prop] = merge_unique_lists([master.get(prop, []), dupe.get(prop, [])])
      
bot.save_one(master, "update master before merge")
bot.merge_works(dupe_olids, master_olid) 

merge.py

# These are the commands I used to merge 
# https://openlibrary.org/works/OL7920112W?v=5 (dupe)
#   and
# https://openlibrary.org/works/OL483502W/House_of_Sand_and_Fog?v=7  (master, because it was on the most lists. Lists don't like redirects)

# Uses my bot code, https://github.com/hornc/catharbot
# which requires the openlibrary-client, https://github.com/internetarchive/openlibrary-client/
#

from catharbot import catharbot
bot = catharbot.CatharBot()

def merge_subjects(a, b):
    return list(set(a['subjects']) | set(b['subjects']))

def merge_covers(a, b):
    return list(set(a['covers']) | set(b['covers']))

master_olid = "OL483502W"
dupe_olid   = "OL7920112W"

master = bot.load_doc(master_olid)
dupe   = bot.load_doc(dupe_olid)

master['title'] = dupe['title']  # The dupe title happened to be in proper title case
master['description'] = dupe['description'] # dupe had a description, master didn't

master['subjects'] = merge_subjects(master, dupe) # dupe only had one subject that the master already had, so no-op
master['covers'] = merge_covers(master, dupe)     # dupe had one cover not on the master

bot.save_one(master, "update master before merge")

bot.merge_works([dupe_olid], master_olid)   # This moves the dupe editions and makes the dupe work a redirect