The geeky way to play a game

From: "flushy flushy.net via chugalug" 
------------------------------------------------------

(If meuon see this, maybe he can reset my bounce flag - as I can only 
see messages on the website chatter link).

Talking with meuon this past weekend, he mentioned some cool things 
folks were doing in the most geekiest way possible. So I mentioned a 
geeky way I solve cryptic crossword-like clues for a puzzle in a game I 
play online.

https://github.com/gonoph/misc-scripts

keywords.py - is a python3 script that I use to generate all possible 
combos from a list of letters. I then use that list to search my logs or 
query a database of unique items/names for the clue solution. Since some 
clues are unique names which are fantasy based names, they are not 
always real names, nor spelled using common language patterns.

It uses a recursive function to do the bulk of the work.

Also, the letters used can be included multiple times, so there can be 
duplicate keywords generated - which the script uses a "set" to ensure 
uniqueness.

One of the last clues I had to solve involved 2 words, joined together 
to form a sequence of 12 letters.

This posed a couple of problems.

Problem 1: since the number of combinations is at worst 12 factorial, 
that's 479 million keywords. I didn't have the memory to generate that 
type of word list

Problem 2: I also didn't have the time as my original script was single 
threaded.

I solved Problem-2 by altering my script to be multi-threaded using 
Python's ProcessPoolExecutor class.

I then solved Problem-1 by running a memory optimized instance in AWS 
for the time it took to generate the keywords I needed. I ran a single 
instance of r8g.8xlarge (32x vcores arm64 with 256 GiB of RAM) @ 
$1.88512 / hour.

With my 2 problems solved, I generated the keywords in 10 minutes (for a 
cost of ~$0.31), then transferred the (compressed) file to my other 
system, and powered off and deleted the instance.

End result: 59 million keywords, 743 MiB in size (uncompressed).

Then started the hard part of scanning those keywords in logs and in the 
database. This process I used mysql and liberal use of cut, sort -u, 
split, and grep - with some shell to glue it together.

I used a multi-step process:

1. trim the keywords from 12 to 6 letters, removing duplicates (trims it 
to ~150k words, and ~1 MiB).
2. perform partial match of the logs/database for those 150k words
3. record what entry and partial match
4. use the partial matches to filter the 59 million keywords to a 
filtered list
5. use the filtered list to re-match the previously matched entries
6. anything that has survived this point is potentially the solution

Anyways, I'd love to hear how other folks are using their geeky skill to 
perform mundane tasks for fun!

--b



===============================================================
From: Wil Wade via chugalug 
------------------------------------------------------
I use a little bookmarklet to clip every coupon at publix. So I put in my
phone number at checkout and get random coupons!

https://gist.github.com/wilwade/aafaf972e6e442c6224b8dcc676f1e4b




===============================================================
From: "flushy flushy.net via chugalug" 
------------------------------------------------------

Oh that's a great idea!

--b





===============================================================
From: Eric Beavers via chugalug 
------------------------------------------------------
Thanks for sharing. This is delightful!