Search Gitlab project for notebook files

gitlab_search.py

from requests import get

project_id = '16454639'

def gen_items(project_id, path=''):
  """ List all files in this folder """
  url = f'https://gitlab.com/api/v4/projects/{project_id}/repository/tree'
  params = {'page': page, 'per_page': 100}
  if path:
    params['path'] = path
  while True:
    items = get(url, params).json()
    if not items: break
    yield from items
    params['page'] += 1
    
def search_ipynb(project_id):
  """ Search only 3 levels to avoid thousands files """
  dirs = []
  # search root dir
  for item in gen_items(project_id):
    if item['name'].endswith('.ipynb'):
      yield item['path']
    if item['type']=='tree':
      dirs.append(item['path'])
  # search 1st subdir
  dirs2 = []
  for d in dirs:
    for item in gen_items(project_id, d):
      if item['name'].endswith('.ipynb'):
        yield item['path']
      if item['type']=='tree':
        dirs2.append(item['path'])
  # search 2nd subdir
  for d in dirs2:
    for item in gen_items(project_id, d):
      if item['name'].endswith('.ipynb'):
        yield item['path']