Upload digital files programmatically using Alma APIs

files.csv

upload.py

import os, requests, boto3, uuid, csv, sys
from requests.models import HTTPError
from multiprocessing import Pool

S3 = boto3.resource('s3')
INST_CODE = 'TR_INTEGRATION_INST'
LIBRARY_CODE = 'MAIN'
THREADS = 3
URL = 'https://api-na.hosted.exlibrisgroup.com'
HEADERS = {
    'authorization': f"apikey {os.environ['ALMA_APIKEY']}",
    'accept': 'application/json',
    'content-type': 'application/json',
  }

def fix_url(url):
  return url if url.startswith('https://') else URL + url

def get(url):
  response = requests.get(fix_url(url), headers=HEADERS)
  handle_http_error(response)
  return response.json()

def post(url, body):
  response = requests.post(fix_url(url), json=body, headers=HEADERS)
  handle_http_error(response)
  return response.json()

def add_rep(mms_id):
  rep = {
    "is_remote": False,
    "library": { "value": LIBRARY_CODE },
    "usage_type": { "value": "PRESERVATION_MASTER" }
  }
  return post(f"/almaws/v1/bibs/{mms_id}/representations", rep)

def add_file(mms_id, rep_id, path):
  file = { "path": path }
  return post(f"/almaws/v1/bibs/{mms_id}/representations/{rep_id}/files", file)

def upload_file(path):
  bucket = S3.Bucket('na-st01.ext.exlibrisgroup.com')
  kwargs = {'Body': open(path, 'rb'), 'Key': f"{INST_CODE}/upload/migration/{uuid.uuid4()}/{os.path.basename(path)}"}
  bucket.put_object(**kwargs)
  return kwargs['Key']

def handle_http_error(resp):
  try:
    resp.raise_for_status()
  except HTTPError as err:
    msg=''
    try:
      msg = err.response.json()['errorList']['error'][0]['errorMessage']
    except ValueError:
      msg = RuntimeError(err.response.text)
    if (msg): raise RuntimeError(msg) from None
    else: raise

def get_next_line():
  with open(sys.argv[1], newline='') as csvfile: 
    filereader = csv.reader(csvfile)
    for row in filereader:
      yield row

def process_line(l):
  print('Processing line:', l)
  rep = add_rep(l[0])
  key = upload_file(l[1])
  print('Uploaded file', key)
  file = add_file(l[0], rep["id"], key)
  print('Added file to rep', file['pid'])

def main():
  f = get_next_line()
  t = Pool(processes=THREADS)
  for i in f:
    t.apply_async(process_line, (i,), error_callback=handle_error)
  t.close()
  t.join()

def handle_error(e):
  print('An error occurred:', e)

if __name__ == '__main__':
  if len(sys.argv) <= 1:
    print("Missing CSV file")
    exit()

  main()

Be sure your API key is in the environment (ALMA_APIKEY). Uses the default AWS credential profile (or specify a profile with the AWS_PROFILE environment variable).

E.g.:

$ AWS_PROFILE=TR_INTEGRATION_INST python upload.py files.csv

See this blog post for more details.