patches to include/exclude based on From field in header
[email protected] Wed, 15 Jan 2003 17:37:28 -0500 (EST)
| Newsgroups | gmane.network.ubh |
|---|---|
| Message-ID | <[email protected]> |
This requires you add a column authors to your table headers which can be done with the following commands: mysql> use ubhcache mysql> alter table headers add authors char(255) not null after subject; mysql> quit It addes two new switches and two new ubhrc opts -K or OPT_K include these authors -N or OPT_N don't include these authors Like -I/-X you can only have one of these not both. You can have one of -I/-X and one of -K/-N. ------------------------ Yahoo! Groups Sponsor ---------------------~--> Flexible Keyboard is the ideal accessory for PDA users that are on the move. http://us.click.yahoo.com/dCBVZC/WnCFAA/xGHJAA/IHFolB/TM ---------------------------------------------------------------------~-> To unsubscribe from this group, send an email to: [email protected] Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
ubh.newer.patch
(text/plain, 8.1 KB)
--- ubh.old 2003-01-15 15:19:01.000000000 -0500
+++ ubh 2003-01-15 17:22:14.000000000 -0500
@@ -138,8 +138,10 @@
-c <file> Use file as config file instead of default (ubhrc).
-G <group> Process this one group.
-a Process all articles (disregards whether article is marked read).
- -I <regexp> Inclusion search filter (double quote on command line)
- -X <regexp> Exclusion search filter (double quote on command line)
+ -I <regexp> Inclusion search filter for Subject (double quote on command line)
+ -X <regexp> Exclusion search filter for Subject (double quote on command line)
+ -K <regexp> Inclusion search filter for From (double quote on command line)
+ -N <regexp> Exclusion search filter for From (double quote on command line)
-C Clean filenames (change all non alphanumeric to '_')
-L Long filenames (use subject as the filename)
-O [yes|no|skip] Overwrite strategy
@@ -186,6 +188,21 @@
}
}
+ if ($opt{'K'}) {
+ eval { "" =~ /$opt{'K'}/ };
+ if ($@) {
+ $log->fatal("invalid -K pattern: ", $@);
+ $error = 1;
+ }
+ }
+
+ if ($opt{'N'}) {
+ eval { "" =~ /$opt{'N'}/ };
+ if ($@) {
+ $log->fatal("invalid -N pattern: ", $@);
+ $error = 1;
+ }
+
if ($opt{'w'}) {
print_warranty;
exit(0);
@@ -288,6 +305,12 @@
elsif ($keyword =~ /OPT_X/) {
$opt{'X'} = $value;
}
+ elsif ($keyword =~ /OPT_K/) {
+ $opt{'K'} = $value;
+ }
+ elsif ($keyword =~ /OPT_N/) {
+ $opt{'N'} = $value;
+ }
elsif ($keyword =~ /OPT_C/) {
$opt{'C'} = 1;
}
@@ -555,7 +578,7 @@
exit(-1);
}
else {
- print DUMPARTICLE @{$body};
+ print DUMPARTICLE @{$body} or die "cannot print: $!";
close DUMPARTICLE;
}
}
@@ -790,7 +813,7 @@
}
binmode(FH);
seek(FH, $offset, 0);
- print FH $decoded;
+ print FH $decoded or die "cannot print: $!";
close(FH);
$log->debug("Finished decoding file.");
}
@@ -815,7 +838,7 @@
}
else {
binmode(FH);
- print FH $decoded;
+ print FH $decoded or die "cannot print: $!";
close(FH);
$log->info("Finished decoding.");
}
@@ -891,7 +914,7 @@
else {
binmode YENCTEMP;
foreach (@{$article}) {
- print YENCTEMP;
+ print YENCTEMP or die "cannot print: $!";
}
close YENCTEMP;
@@ -1242,7 +1265,7 @@
my $mimefilename = $platform->get_tempfilename;
open(SINGFILE, "> ".$mimefilename)
or die "ERROR: open(".$mimefilename.") failed.\n";
- print SINGFILE @{$article};
+ print SINGFILE @{$article} or die "cannot print: $!";
close SINGFILE;
if ($opt{'D'}) {
@@ -1569,7 +1592,7 @@
# Part was successfully downloaded.
if ($opt{'A'}) {
- print PARTSFILE @{$p};
+ print PARTSFILE @{$p} or die "cannot print: $!";
}
else {
push(@parts, @{$p});
@@ -1722,7 +1745,7 @@
$log->info("Begin processing single part articles.");
my $stmt =
- "SELECT h.msgid, h.subject, h.category ".
+ "SELECT h.msgid, h.subject, h.authors, h.category ".
"FROM headers h, xref_headers x ".
"WHERE h.header_id = x.header_id ".
"AND x.newsgroup_id = ".$group->{newsgroup_id}." ".
@@ -1752,7 +1775,7 @@
$sth->execute;
while (my @cols = $sth->fetchrow_array) {
- my ($msgid, $subject, $category) = @cols;
+ my ($msgid, $subject, $authors, $category) = @cols;
my $title;
if (!$opt{'g'}) {
@@ -1790,6 +1813,20 @@
next;
}
+ # Inclusion/exclusion filtering.
+ # Skip authors that don't match the inclusion filter.
+ # Skip authors that do match the exclusion filter.
+ if ((defined $opt{'K'} && $authors !~ /$opt{'K'}/o) ||
+ (defined $opt{'N'} && $authors =~ /$opt{'N'}/o)) {
+
+ # It was either not included or was excluded.
+ if ($opt{'z'}) {
+ $dbh->do(
+ "UPDATE headers SET status = 'READ' ".
+ "WHERE msgid = ".$dbh->quote($msgid));
+ }
+ next;
+ }
# Skip it if there is a batch file and it isn't mentioned.
next if (defined $batchfile && !$batchfile->contains($title));
@@ -1837,7 +1874,7 @@
$log->info("Begin processing multipart articles.");
my $stmt =
- "SELECT a.assemble_id, h.msgid, h.subject, h.category ".
+ "SELECT a.assemble_id, h.msgid, h.subject, h.authors, h.category ".
"FROM assemble a, headers h, xref_headers x ".
"WHERE a.assemble_id = x.assemble_id ".
"AND h.header_id = x.header_id ".
@@ -1863,7 +1900,7 @@
$sth->execute;
while (my @cols = $sth->fetchrow_array) {
- my ($assemble_id, $msgid, $subject, $category) = @cols;
+ my ($assemble_id, $msgid, $subject, $authors, $category) = @cols;
# Try to pick a title out of the subject using filename extension filter.
my $title;
@@ -1898,6 +1935,21 @@
next;
}
+ # Inclusion/exclusion filtering.
+ # Skip authors that don't match the inclusion filter.
+ # Skip authors that do match the exclusion filter.
+ if ((defined $opt{'K'} && $authors !~ /$opt{'K'}/o) ||
+ (defined $opt{'N'} && $authors =~ /$opt{'N'}/o)) {
+
+ # It was either not included or was excluded.
+ if ($opt{'z'}) {
+ $dbh->do(
+ "UPDATE headers SET status = 'READ' ".
+ "WHERE msgid = ".$dbh->quote($msgid));
+ }
+ next;
+ }
+
# Skip it if there is a batch file and it isn't mentioned.
next if (defined $batchfile && !$batchfile->contains($title));
@@ -1969,7 +2021,7 @@
=head1 SYNOPSIS
-ubh [-c file] [(-S|-M)] [-A] [-b file] [-C] [-d] [-D] [-g] [-a] [(-I|-X) pattern] [-L] [-O (yes|no|skip)] [-u] [-w] [-y] [-z]
+ubh [-c file] [(-S|-M)] [-A] [-b file] [-C] [-d] [-D] [-g] [-a] [(-I|-X) pattern] [(-K|-N) pattern] [-L] [-O (yes|no|skip)] [-u] [-w] [-y] [-z]
=head1 DESCRIPTION
@@ -2024,6 +2076,7 @@
C<ubhcatalog>), one per line, to be downloaded.
Mutually exclusive with B<-I> and B<-X>.
+Mutually exclusive with B<-K> and B<-N>.
=item -d
@@ -2075,12 +2128,22 @@
=item -I I<regexp>
-Inclusion search filter (double quote on command line).
+Inclusion search filter for Subject field(double quote on command line).
+I<regexp> is any valid Perl regular expression.
+
+=item -K I<regexp>
+
+Inclusion search filter for From field(double quote on command line).
I<regexp> is any valid Perl regular expression.
=item -X I<regexp>
-Exclusion search filter (double quote on command line).
+Exclusion search filter for Subject field(double quote on command line).
+I<regexp> is any valid Perl regular expression.
+
+=item -N I<regexp>
+
+Exclusion search filter for From field(double quote on command line).
I<regexp> is any valid Perl regular expression.
=item -C
ubhdb.newer.patch
(text/plain, 1.4 KB)
--- ubhdb.old 2003-01-15 15:02:32.000000000 -0500
+++ ubhdb 2003-01-15 17:07:45.000000000 -0500
@@ -58,7 +58,7 @@
my $log = get_logger("ubhdb::add_articles");
my %headers;
- foreach my $h (qw(Subject Message-ID Date)) {
+ foreach my $h (qw(Subject From Message-ID Date)) {
$headers{$h} = $nntp->xhdr($h, "$first-$last");
}
@@ -70,7 +70,8 @@
str2time($headers{'Date'}->{$i}),
"GMT"));
my $number = $i;
- my $subject = $headers{'Subject'}->{$i};
+ my $subject = $headers{'Subject'}->{$i};
+ my $authors = $headers{'From'}->{$i};
next if (!$subject || $subject =~ /^\s+$/);
# Categorize this header based on its subject.
@@ -150,13 +151,14 @@
# until it has been assembled.
my $stmt =
"INSERT IGNORE INTO headers ".
- "(msgid,date,subject,category,status".
+ "(msgid,date,subject,authors,category,status".
(defined $part && defined $total ? ",part,totalparts" : '').
") ".
"VALUES (".
$msgid.",".
$date.",".
$dbh->quote($subject).",".
+ $dbh->quote($authors).",".
$dbh->quote($category).",".
$dbh->quote('IMPORTING').
(defined $part && defined $total ? ",".$part.",".$total : '').