How to mask an email address using m-code?

For example my email could be masked as follows:

[char([21 8 4 13 8 0 21 0]+97) '@gmail.com']

But this is trivial and very redundant. I am curious about your concepts on this. Please paste your mask on your (or mine or other real-like one) email address - Lets find out how it can be done in the most clever way.

[Paulo Silva]

just leave it in plain text and generate a p-code file

[Vieniava]

Paulo, good point; but I thought about anti-spam masking rather than about 'anti-reverse-engineering' method.

[Matt Tearle]

I don't understand what you're after. What's wrong with your solution for anti-spam purposes? Can you explain your criteria more fully? Thx

[Vieniava]

Matt, my solution is to long. I would like to have it as short as possible and also maybe there is the way without direct "char numbering" approach.

[Matt Tearle]

OK, sorry, I'm being dense, I guess. You can easily use double('[e-mail]') to get a numeric representation, then char to flip back, so... I'm missing something. Are you looking for a simple encoding (cipher)? Because most anti-spam obscuring just adds stuff into a string (like 'f o o(at)barNOSPAM(dot)com'). But that's just string manipulation. For a cipher approach, you're assuming someone can decipher it (eg char <-> double), and the result isn't human-readable -- is that what you're after?

Answer 1

[Matt Fig]

I don't quite know what you are after either, even with your response to Matt. If that is too long, I don't know what is shorter except just writing out the email address. How about these:

fliplr('moc.liam@liameym')
char(max(['m)m-i)@0a&l+c/m';'3y/a&l-m/i0.%o-']))
strrep('[e-mail]','t','m')
strrep('[e-mail]',116,109)
char('nznbjmAnbjm/dpn' - 1)

[Andrew Newell]

This could be the start of the first Cryptic Matlab Contest!

[Vieniava]

'fliplr' solution is really nice

[Matt Fig]

I prefer the last one myself.

[Vieniava]

the last one is quasi-Caesar-cipher

Answer 2

[Jan]

I'm using a trivial masking by using an uppercase string instead of a - character, e.g. "[e-mail]". Although I never got any spam to this address, I decided to insert the current year: "[e-mail]", such that I can change the address each year.

But this is already too complicated for some human readers: I got not a single spam, but about 10 messages (through the "Contact the author" form) from CSSM users per year complaining about a not working address...

Therefore I think, that "[char([21 8 4 13 8 0 21 0]+97) '@gmail.com']" will increase the mail noise instead of reducing it. I assume "[e-mail]" would be more efficient.

Answer 3

[Sean de Wolski]

I think Brett Shoelson won this one:

http://www.mathworks.com/matlabcentral/fileexchange/8987-makesig

[Vieniava]

'cumsum' is tricky

Answer 4

[Walter Roberson]

Sounds like you are looking for Kolmogorov Complexity -- the shortest algorithm for producing a given output.

Answer 5

[Walter Roberson]

char(23+mod(1.332.^'ihRcilh? yoly^UP`0th^',96))

char(23+mod(1.145.^'UL:QL?U?Ppz?Lst Gz',96))

char(22+mod(1.3107.^'8s~+sn,ScPP>%1o8Y%,aq',96))

Encoder:

S='[e-mail]';
L=(10001:13354)/10000;for d=0:45;for P=L;[tf,idx]=ismember(S,char(d+mod(P.^(0:127),96)));if sum(tf)==length(S)&all(idx>31 & idx<128);fprintf('char(%d+mod(%0.4f.^''%s'',96))\n',d,P,char(idx-1));return;end;end;end

Note: using 1.0001:.0001:1.3354 does not work because of colon round-off error. The upper bound 1.3354 is the largest 4-decimal-place number such that P^127 <= 2^53

Jan's address with the year was the only one of the group that required the 4 digit base.