[PATCH] experimenting with new parsing rules

[Matthias Andree]

David Relson <relson at osagesoftware.com> writes:

> If you're interested in experimenting with Michael's parsing rules, I
> have attached patch that can enable/disable them using environment
> variable BOGOTEST.  Sample uses would be:
>
> BOGOTEST=0 bogolexer -p < message > 0.out
> BOGOTEST=1 bogolexer -p < message > 1.out
>
> where BOGOTEST=0 gives the current results and BOGOTEST=1 uses Michael's
> rules.  I often run with redirected stdout (as shown above) so I can run
> "gtkdiff 0.out 1.out" to see what's different.

Looks broken, from bogofilter -vvv:

"List-Help"                         34  0.176101  0.024000  0.120026 +
"List-Post"                         34  0.176101  0.024000  0.120026 +
"List-Subscribe"                    34  0.176101  0.024000  0.120026 +
"List-Unsubscribe"                  34  0.176101  0.024000  0.120026 +

These should have been tagged.

"rtrn:ESMTP"                         0  0.000000  0.000000  0.415000 -
"rtrn:MET"                           0  0.000000  0.000000  0.415000 -
"rtrn:Mon"                           0  0.000000  0.000000  0.415000 -
"rtrn:NETSYS.COM"                    0  0.000000  0.000000  0.415000 -
"rtrn:POP3"                          0  0.000000  0.000000  0.415000 -

This tag is wrong, these are from Received: headers, not from a reverse
path header such as Return-Path:.

Sorry, I don't have the time to investigate this further now.

> Actually, BOGOTEST is bit coded with 1 enabling the modified rules and 2
> enabling the lexer's debug mode (so one can see which rules are being
> applied).

Not quite. Here's a fix, replace this function in lexer_v3.l (mind the
"& 2"):

void lexer_set_debug(int v)
{
#ifndef FLEX_DEBUG
    (void) v;
#else
    yy_flex_debug = v & 2;
#endif
}

-- 
Matthias Andree

Encrypt your mail: my GnuPG key ID is 0x052E7D95