Is it possible to add a command line option to output only the first N words?
It is useful for example for speed benchmarking.
Thanks 😀

Given this input:

PassWord123!@#пароль

... it would be great to be able to optionally generate at least one of the following sets of output, treating the "boundary" between changes in character class as the split point, optionally case-sensitive, case-insensitive, (or both):

P,ass,W,ord,123,!@#,пароль
Pass,Word,123,!@#,пароль

... resulting in the following superset of output strings:

!@#
123
P
Pass
W
Word
ass
ord
пароль

In the simplest case, all non-ASCII could be handled as an entirely separate class, and in a later revision, checked for case and class - similar to Perl's \p character classes:

\p{Number}
\p{Letter}
\p{Upper}
\p{Lower}
\p{Mark}
\p[Punctuation}
\p{Symbol}

I've been referring to this concept of splitting on character-class boundary as "tokenization", and I've been doing it with the Perl \p classes, but I suspect that there's a lot of room for more efficiency by doing it within chasm (and being able to combine it with other chasm options).

Would be good to clarify which cmdline parameters are mutually exclusive, which perform primary actions, which augment the functionality of a primary action, etc. Grouping them into primary and then modifier subgroups would be good, and then sorting each of those sections alphabetically (in compressed manpage style).

Would also be good if -t mentioned that the resulting filename has '.tokens' appended. I assumed that I'd be supplying the filename entirely, so I got a 'blah.tokens.tokens' output filename. :) I see the basefilename in the usage output, but explicitly mentioning "this flag will append '.tokens' to the basefilename" or something like that would be good.

A "chasm: a CHaracter Aware Splitting Method for producing password candidates" line at the top of the usage output is customary.

-s and -t appear to be mutually exclusive - requesting -s with -t has no effect on the output order. Naively, I would expect -s to modify -t, so that I could get a frequency-ordered list. (I'd also like that frequency order to include the frequency count itself, if I had my druthers.)

The overall burden necessary for new users to understand the output could use a little reduction. For example, the usage output should describe the scheme used to name the output files.

It also seems counterintuitive for the default behavior to be to create a bunch of output files.

More-specific usage feedback would be useful. For example, if the basefilename is omitted, prior to printing usage, an "error: basefilename not specified" would help new users.