[ACCEPTED]-Is it irrational to sanitize random character strings for curse words?-filter

Don't generate random strings with vowels 2 and then you don't have to worry about curse 1 words.

Yes, on the grounds that anyone who would 4 be offended by something they saw in a randomly 3 generated string can think of more things 2 they find offensive than you can sanitize.

Don't 1 optimize for the insane.

Microsoft omits the following from their 20 product keys:

0 1 2 5 A E I O U L N S Z

I omit those from [0-9A-Z], and 19 once the key is generated, I match against 18 a list I found of two-letter combinations most common 17 in English, and regenerate the key if there 16 is a match. For speed, I edit the list of 15 letter pairs by first culling from that 14 list the pairs that are already prevented 13 due to their inclusion of a character in 12 the stripped list ('HE' can't exist if the 11 key is generated from a character set that 10 does not include 'E'), then convert some 9 from 'E' to '3', as in 'H3' instead of 'HE', etc. I 8 have also added a few of my own, like 'KK' and 7 'CK' for edge cases. One could also omit 6 '3' for speed as necessary, although the 5 more characters you omit the fewer unique 4 keys can be generated.

Probably not a perfect 3 solution, but it's fast enough for my needs 2 and prevents almost all English words from 1 being generated, offensive or not.

Simplest solution is to generate from a 8 'sanitized' alphabet; use a set of characters 7 that cannot possibly form words. One suggestion 6 in one of the answers is hexadecimal which 5 is an excellent choice, or otherwise drop 4 some critical letters from the alphabet.

Note 3 that just dropping vowels is not going to 2 do the job... it is all too easy to infer 1 them from the remaining consonants.

That makes sense to me. I mean, it would 6 be a pretty bad PR disaster if someone posts 5 a picture of your product, with this stamped 4 on the back of the CD case:

12345-67890-F**KU-ABCDE-FGHIJ

It 3 sounds funny but you never know what kind 2 of sense of humor the person will have who 1 happens to pick up that package.

See those items tagged with clbuttic

0

I think it's better to plainly avoid vowels. A 3 product key like JKL-YOUAREMYFRIEND-0001-KK 2 may not be offensive but it doesn't sound 1 like serious business either.

Limit your randomly generated "words" to 8 hex characters and I don't believe you'll 7 have any English-language curses. This also 6 pushes you down a path of not spending too 5 much time on your random word generator.

Of 4 course, there may be some language where 3 you can curse with hex digits, but then 2 you're not likely to know/filter those curses 1 anyway.

If you are just worried about product keys, I 8 would stick to hexadecimal digits, maybe 7 even a guid would work for you. Probably 6 no chance of a "naughty" word being generated 5 with these constraints. You could also just 4 stick to numbers as well. If you must have 3 random strings with all letters of the alphabet, it 2 is probably better safe than sorry so I 1 would do the filtering.

We are using a random string generator for 8 a security key that will be sent to customers, and 7 did not omit vowels from the allowable characters, etc. No 6 word of a lie, one of the strings it generated 5 was "7D9WAF*CKS" ...! Luckily, this was 4 discovered during development, and we are 3 now going to restrict the allowable characters. Whew!! That 2 was a close one - happy to have found this 1 posting !

A very fun tale of woe to read with a similar situation 1 to you.

It's certain conceivable, but I wouldn't 2 devote much time to it, especially if you've 1 got letters and numbers.

I'm using randomly generated, phonetic-sounding 5 passwords for one webapp I wrote. I did 4 end up hard-coding a list of "dirty" words 3 that aren't acceptable, but the list that 2 matched my pattern ended up being pretty 1 short.

No. You have no chance to collect all curse 2 words in all world languages. Those words 1 usually don't appear in dictionaries.