|
Lines 405-410
Link Here
|
| 405 |
# now do some token abstraction; in other words, make them act like |
405 |
# now do some token abstraction; in other words, make them act like |
| 406 |
# patterns instead of text copies. |
406 |
# patterns instead of text copies. |
| 407 |
|
407 |
|
|
|
408 |
# Chris Thielen Anti Obfu Proof of Concept patch |
| 409 |
# 2004-01-08 |
| 410 |
if (!$in_headers) |
| 411 |
{ |
| 412 |
my $tokcpy = $token; |
| 413 |
# nuke all non-alphas. |
| 414 |
$tokcpy =~ s/^sk://; # optimize this; we check if it happened later |
| 415 |
$tokcpy =~ tr/A-Za-z//cd; |
| 416 |
# case insensitive. (good? bad?) |
| 417 |
$tokcpy =~ tr/A-Z/a-z/; |
| 418 |
# sort the chars in string |
| 419 |
$tokcpy = join('',sort(split(//,$tokcpy))); |
| 420 |
if ($tokcpy) |
| 421 |
{ |
| 422 |
$tokcpy = "sk:".$tokcpy if $token =~ m/^sk:/; |
| 423 |
# add token |
| 424 |
push (@{$self->{tokens}}, $tokprefix.$tokcpy); $wc++; |
| 425 |
# remove any repeated chars |
| 426 |
my $tokcpycpy = $tokcpy; |
| 427 |
$tokcpy =~ tr/A-Za-z/A-Za-z/s; |
| 428 |
dbg("BAYES TRANSLATE: $token: $tokcpycpy, $tokcpy"); |
| 429 |
if ($tokcpy) |
| 430 |
{ |
| 431 |
# add token |
| 432 |
push (@{$self->{tokens}}, $tokprefix.$tokcpy); $wc++; |
| 433 |
} |
| 434 |
} |
| 435 |
} |
| 436 |
|
| 408 |
# replace digits with 'N'... |
437 |
# replace digits with 'N'... |
| 409 |
if ($token =~ /\d/ && (!$in_headers || !NO_NUMERIC_IN_HEADERS)) |
438 |
if ($token =~ /\d/ && (!$in_headers || !NO_NUMERIC_IN_HEADERS)) |
| 410 |
{ |
439 |
{ |
