randoffee

Library to handle REG random coffees.

---

Workflow for generating random coffees

TODO: describe the algorithm. Until I bother writing this up, you can find it in the comments scattered around run.py.

1. Install the library.

   git clone git@github.com:alan-turing-institute/randoffee.git
   cd randoffee
   python -m pip install -e .
   

2. cd to the Coffee folder in the REG SharePoint.

3. Run

   python -m randoffee.run
   

4. The script will print the groupings, some statistics about them, and ask you whether these are your final groupings.

   Before answering yes or no, you should double-check the groups to make sure that they are sensible. (In practice, we have never interfered with any of the groups.)

   If you are satisfied with these groupings, then you can enter y to save the permutation to disk. It will be saved in previous/YYYY-MM-DD.json. This file will be used for future coffee rounds (the script will make sure to generate groups that are sufficiently different from it).

   If you enter n, the permutation will still be saved as previous/.latest.json. You can just mv this to the desired date if you realise that you do want those groups.

5. Change the random topics in the email to anything you like.

6. Send the email to the list of email addresses the script gives you.

---

Usage as library

More functionality to follow.

Example usage (to investigate how similar previous coffees have been) follows. The COFFEE_DIR variable must point to the Coffee folder in the REG SharePoint folder (in my case, I've symlinked it to ~/coffee).

from pathlib import Path

from randoffee import Permutation


## Read in all previous permutations

ALL_PERMS = []
COFFEE_DIR = Path.home() / 'coffee'

for file in (COFFEE_DIR / 'previous').iterdir():
    if file.is_file() and file.suffix == '.json':
        ALL_PERMS.append(Permutation.from_json_file(file))

ALL_PERMS = sorted(ALL_PERMS, key=lambda x: x.date)


## Calculate similarity between all consecutive permutations

for (perm1, perm2) in zip(ALL_PERMS, ALL_PERMS[1:]):
    print(f'-------- {perm1.date} x {perm2.date} --------')
    print(perm1.similarity_to(perm2))
    print('\n\n')