Mailing List Archive

Author: jm
Date: Sat Dec 16 03:54:04 2006
New Revision: 487789

URL: http://svn.apache.org/viewvc?view=rev&rev=487789
Log:
hack hack hack; update P595Body plugin to use the new blead feature of REGMARK, thanks to demerphq. still needs speeding up unfortunately

Modified:
spamassassin/trunk/lib/Mail/SpamAssassin/Plugin/P595Body.pm

Modified: spamassassin/trunk/lib/Mail/SpamAssassin/Plugin/P595Body.pm
URL: http://svn.apache.org/viewvc/spamassassin/trunk/lib/Mail/SpamAssassin/Plugin/P595Body.pm?view=diff&rev=487789&r1=487788&r2=487789
==============================================================================
--- spamassassin/trunk/lib/Mail/SpamAssassin/Plugin/P595Body.pm (original)
+++ spamassassin/trunk/lib/Mail/SpamAssassin/Plugin/P595Body.pm Sat Dec 16 03:54:04 2006
@@ -37,6 +37,10 @@

$self->{one_line_body} = Mail::SpamAssassin::Plugin::OneLineBodyRuleType->new();

+ if ($] < 5.009005) {
+ die "this plugin requires perl 5.9.5 or later";
+ }
+
return $self;
}

@@ -47,23 +51,8 @@
my $conf = $params->{conf};

my $main = $self->{main};
- $main->{base_extract} = 1;
- $main->{bases_must_be_casei} = 1;
- $main->{bases_can_use_alternations} = 0; # /(foo|bar|baz)/
- $main->{bases_can_use_quantifiers} = 0; # /foo.*bar/ or /foo*bar/ or /foooo?bar/
- $main->{bases_can_use_char_classes} = 0; # /fo[opqr]bar/
- $main->{bases_split_out_alternations} = 1; # /(foo|bar|baz)/ => ["foo", "bar", "baz"]
-
- my $ext_start = time;
- my $basextor = Mail::SpamAssassin::Plugin::BodyRuleBaseExtractor->new
- ($self->{main});
- $basextor->extract_bases($conf);
- my $ext_dur = time - $ext_start;
- warn "base extraction took $ext_dur seconds\n";

$conf->{skip_body_rules} ||= { };
- $conf->{need_one_line_sub} ||= { };
-
$self->setup_test_set ($conf, $conf->{body_tests}, 'body');
}

@@ -71,7 +60,8 @@
my ($self, $conf, $test_set, $ruletype) = @_;
foreach my $pri (keys %{$test_set}) {
my $nicepri = $pri; $nicepri =~ s/-/neg/g;
- $self->setup_test_set_pri($conf, $test_set->{$pri}, $ruletype.'_'.$nicepri, $pri);
+ $self->setup_test_set_pri($conf, $test_set->{$pri},
+ $ruletype.'_'.$nicepri, $pri);
}
}

@@ -79,49 +69,36 @@
my ($self, $conf, $rules, $ruletype, $pri) = @_;

my $alternates = [];
- my $trie_rules = {};
+ while (my ($rule, $pat) = each %{$conf->{body_tests}->{$pri}}) {
+ $pat = Mail::SpamAssassin::Util::regexp_remove_delimiters($pat);

- # while (my ($rule, $pat) = each %{$pms->{conf}->{body_tests}->{$priority}}) {
- # push @{$alternates}, $pat;
- # }
+ # ignore rules marked for ReplaceTags work!
+ next if ($conf->{rules_to_replace}->{$rule});

- foreach my $base (keys %{$conf->{base_string}->{$ruletype}})
- {
- push @{$alternates}, $base;
- my @rules = split(' ', $conf->{base_string}->{$ruletype}->{$base});
- $trie_rules->{$base} = \@rules;
-
- foreach my $rule (@rules) {
- # ignore rules marked for ReplaceTags work!
- # TODO: we should be able to order the 'finish_parsing_end'
- # plugin calls to do this.
- next if ($conf->{rules_to_replace}->{$rule});
-
- # TODO: need a cleaner way to do this. I expect when rule types
- # are implementable in plugins, I can do it that way
- $conf->{skip_body_rules}->{$rule} = 1;
-
- # ensure that the one-liner version of the function call is
- # created, though
- $conf->{generate_body_one_line_sub}->{$rule} = 1;
- }
+ # use the REGMARK feature:
+ # see http://taint.org/2006/11/16/154546a.html#comment-1011
+ #
+ push @{$alternates}, "$pat(*:$rule)";
+
+ # TODO: need a cleaner way to do this. I expect when rule types
+ # are implementable in plugins, I can do it that way
+ $conf->{skip_body_rules}->{$rule} = 1;
}

my $sub = '
sub {
+ our $REGMARK;
our @matched = ();
$_[0] =~ m#('.join('|', @{$alternates}).')(?{
- push @matched, $1;
- })(*FAIL)#i;
+ push @matched, $REGMARK;
+ })(*FAIL)#;
return @matched;
}
';
# warn "JMD $sub";

$conf->{$ruletype}->{trie_re_sub} = eval $sub;
- if ($@) { warn "trie sub compilation failed: $@"; }
-
- $conf->{$ruletype}->{trie_rules} = $trie_rules;
+ if ($@) { warn "REGMARK sub compilation failed: $@"; }
}

###########################################################################
@@ -151,8 +128,7 @@
my $conf = $scanner->{conf};

my $trie_re_sub = $conf->{$ruletype}->{trie_re_sub};
- my $trie_rules = $conf->{$ruletype}->{trie_rules};
- if (!$trie_re_sub || !$trie_rules)
+ if (!$trie_re_sub)
{
dbg("zoom: run_body_fast_scan for $ruletype skipped, no rules");
return;
@@ -167,13 +143,11 @@
no strict "refs";
foreach my $line (@{$params->{lines}})
{
- my $sub = $trie_re_sub;
- my @caught = $sub->($line);
+ my @caught = $trie_re_sub->($line);
next unless (scalar @caught > 0);

my %alreadydone = ();
- foreach my $caught (@caught) {
- foreach my $rulename (@{$trie_rules->{$caught}})
+ foreach my $rulename (@caught) {
{
# only try each rule once per line
next if exists $alreadydone{$rulename};
@@ -182,18 +156,7 @@
# ignore 0-scored rules, of course
next unless $scoresptr->{$rulename};

- # dbg("zoom: base found for $rulename: $line");
-
- # TODO: ick, this shouldn't have to use this package name. A good
- # solution would be for compiled rule methods to live in their own
- # namespace anyway; Mail::SpamAssassin::CompiledRules or something
- my $fn = 'Mail::SpamAssassin::Plugin::Check::'.
- $rulename.'_one_line_body_test';
-
- # run the real regexp -- on this line alone.
- if (!&{$fn} ($scanner, $line) && $do_dbg) {
- $self->{rule2xs_misses}->{$rulename}++;
- }
+ $scanner->got_hit($rulename, "BODY: ", ruletype => "p595_body");
}
}
}