-# Copyright (C) 2015-2020 all contributors <meta@public-inbox.org>
+# Copyright (C) 2015-2021 all contributors <meta@public-inbox.org>
# License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
-# based on notmuch, but with no concept of folders, files or flags
+# based on notmuch, but with no concept of folders, files
#
# Indexes mail with Xapian and our (SQLite-based) ::Msgmap for use
# with the web and NNTP interfaces. This index maintains thread
use v5.10.1;
use parent qw(PublicInbox::Search PublicInbox::Lock Exporter);
use PublicInbox::Eml;
+use PublicInbox::Search qw(xap_terms);
use PublicInbox::InboxWritable;
use PublicInbox::MID qw(mids_for_index mids);
use PublicInbox::MsgIter;
use PublicInbox::Spawn qw(spawn nodatacow_dir);
use PublicInbox::Git qw(git_unquote);
use PublicInbox::MsgTime qw(msg_timestamp msg_datestamp);
-our @EXPORT_OK = qw(crlf_adjust log2stack is_ancestor check_size prepare_stack
+use PublicInbox::Address;
+use Config;
+our @EXPORT_OK = qw(log2stack is_ancestor check_size prepare_stack
index_text term_generator add_val is_bad_blob);
my $X = \%PublicInbox::Search::X;
our ($DB_CREATE_OR_OPEN, $DB_OPEN);
our $DB_NO_SYNC = 0;
-our $BATCH_BYTES = $ENV{XAPIAN_FLUSH_THRESHOLD} ? 0x7fffffff : 1_000_000;
+our $BATCH_BYTES = $ENV{XAPIAN_FLUSH_THRESHOLD} ? 0x7fffffff :
+ # assume a typical 64-bit system has 8x more RAM than a
+ # typical 32-bit system:
+ (($Config{ptrsize} >= 8 ? 8192 : 1024) * 1024);
+
use constant DEBUG => !!$ENV{DEBUG};
my $xapianlevels = qr/\A(?:full|medium)\z/;
my $hex = '[a-f0-9]';
my $OID = $hex .'{40,}';
+my @VMD_MAP = (kw => 'K', L => 'L');
our $INDEXLEVELS = qr/\A(?:full|medium|basic)\z/;
sub new {
}
}
$ibx = PublicInbox::InboxWritable->new($ibx);
- my $self = bless {
- ibx => $ibx,
- xpfx => $inboxdir, # for xpfx_init
- -altid => $altid,
- ibx_ver => $version,
- indexlevel => $indexlevel,
- }, $class;
- $self->xpfx_init;
+ my $self = PublicInbox::Search->new($ibx);
+ bless $self, $class;
+ $self->{ibx} = $ibx;
+ $self->{-altid} = $altid;
+ $self->{indexlevel} = $indexlevel;
$self->{-set_indexlevel_once} = 1 if $indexlevel eq 'medium';
if ($ibx->{-skip_docdata}) {
$self->{-set_skip_docdata_once} = 1;
sub idx_release {
my ($self, $wake) = @_;
if (need_xapian($self)) {
- my $xdb = delete $self->{xdb} or croak 'not acquired';
+ my $xdb = delete $self->{xdb} or croak '{xdb} not acquired';
$xdb->close;
}
$self->lock_release($wake) if $self->{creat};
sub load_xapian_writable () {
return 1 if $X->{WritableDatabase};
- PublicInbox::Search::load_xapian() or return;
+ PublicInbox::Search::load_xapian() or die "failed to load Xapian: $@\n";
my $xap = $PublicInbox::Search::Xap;
for (qw(Document TermGenerator WritableDatabase)) {
$X->{$_} = $xap.'::'.$_;
$DB_CREATE_OR_OPEN = eval($xap.'::DB_CREATE_OR_OPEN()');
$DB_OPEN = eval($xap.'::DB_OPEN()');
my $ver = (eval($xap.'::major_version()') << 16) |
- (eval($xap.'::minor_version()') << 8);
+ (eval($xap.'::minor_version()') << 8) |
+ eval($xap.'::revision()');
$DB_NO_SYNC = 0x4 if $ver >= 0x10400;
+ # Xapian v1.2.21..v1.2.24 were missing close-on-exec on OFD locks
+ $X->{CLOEXEC_UNSET} = 1 if $ver >= 0x010215 && $ver <= 0x010218;
1;
}
}
}
+sub index_phrase ($$$$) {
+ my ($self, $text, $wdf_inc, $prefix) = @_;
+
+ my $tg = term_generator($self);
+ $tg->index_text($text, $wdf_inc, $prefix);
+ $tg->increase_termpos;
+}
+
sub index_text ($$$$) {
my ($self, $text, $wdf_inc, $prefix) = @_;
- my $tg = term_generator($self); # man Search::Xapian::TermGenerator
if ($self->{indexlevel} eq 'full') {
- $tg->index_text($text, $wdf_inc, $prefix);
- $tg->increase_termpos;
+ index_phrase($self, $text, $wdf_inc, $prefix);
} else {
+ my $tg = term_generator($self);
$tg->index_text_without_positions($text, $wdf_inc, $prefix);
}
}
sub index_headers ($$) {
my ($self, $smsg) = @_;
- my @x = (from => 'A', # Author
- subject => 'S', to => 'XTO', cc => 'XCC');
+ my @x = (from => 'A', to => 'XTO', cc => 'XCC'); # A: Author
+ while (my ($field, $pfx) = splice(@x, 0, 2)) {
+ my $val = $smsg->{$field};
+ next if $val eq '';
+ # include "(comments)" after the address, too, so not using
+ # PublicInbox::Address::names or pairs
+ index_text($self, $val, 1, $pfx);
+
+ # we need positional info for email addresses since they
+ # can be considered phrases
+ if ($self->{indexlevel} eq 'medium') {
+ for my $addr (PublicInbox::Address::emails($val)) {
+ index_phrase($self, $addr, 1, $pfx);
+ }
+ }
+ }
+ @x = (subject => 'S');
while (my ($field, $pfx) = splice(@x, 0, 2)) {
my $val = $smsg->{$field};
index_text($self, $val, 1, $pfx) if $val ne '';
index_text($self, join("\n", @$xnq), 1, 'XNQ');
@$xnq = ();
}
- index_text($self, $text, 1, $pfx);
+ if ($pfx eq 'XDFN') {
+ index_phrase($self, $text, 1, $pfx);
+ } else {
+ index_text($self, $text, 1, $pfx);
+ }
}
sub index_old_diff_fn {
my $ct = $part->content_type || 'text/plain';
my $fn = $part->filename;
if (defined $fn && $fn ne '') {
- index_text($self, $fn, 1, 'XFN');
+ index_phrase($self, $fn, 1, 'XFN');
}
if ($part->{is_submsg}) {
my $mids = mids_for_index($part);
$l =~ /<([^>]+)>/ or next;
my $lid = lc $1;
$doc->add_boolean_term('G' . $lid);
- index_text($self, $lid, 1, 'XL'); # probabilistic
+ index_phrase($self, $lid, 1, 'XL'); # probabilistic
}
}
sub index_ids ($$$$) {
my ($self, $doc, $hdr, $mids) = @_;
for my $mid (@$mids) {
- index_text($self, $mid, 1, 'XM');
+ index_phrase($self, $mid, 1, 'XM');
# because too many Message-IDs are prefixed with
# "Pine.LNX."...
if ($mid =~ /\w{12,}/) {
my @long = ($mid =~ /(\w{3,}+)/g);
- index_text($self, join(' ', @long), 1, 'XM');
+ index_phrase($self, join(' ', @long), 1, 'XM');
}
}
$doc->add_boolean_term('Q' . $_) for @$mids;
index_headers($self, $smsg);
if (defined(my $eidx_key = $smsg->{eidx_key})) {
- $doc->add_boolean_term('O'.$eidx_key);
+ $doc->add_boolean_term('O'.$eidx_key) if $eidx_key ne '.';
}
msg_iter($eml, \&index_xapian, [ $self, $doc ]);
index_ids($self, $doc, $eml, $mids);
if (!$self->{-skip_docdata}) {
# WWW doesn't need {to} or {cc}, only NNTP
$smsg->{to} = $smsg->{cc} = '';
- PublicInbox::OverIdx::parse_references($smsg, $eml, $mids);
+ $smsg->parse_references($eml, $mids);
my $data = $smsg->to_doc_data;
$doc->set_data($data);
}
sub add_xapian ($$$$) {
my ($self, $eml, $smsg, $mids) = @_;
+ begin_txn_lazy($self);
+ my $merge_vmd = delete $smsg->{-merge_vmd};
my $doc = eml2doc($self, $eml, $smsg, $mids);
+ if (my $old = $merge_vmd ? _get_doc($self, $smsg->{num}) : undef) {
+ my @x = @VMD_MAP;
+ while (my ($field, $pfx) = splice(@x, 0, 2)) {
+ for my $term (xap_terms($pfx, $old)) {
+ $doc->add_boolean_term($pfx.$term);
+ }
+ }
+ }
$self->{xdb}->replace_document($smsg->{num}, $doc);
}
sub _msgmap_init ($) {
my ($self) = @_;
- die "BUG: _msgmap_init is only for v1\n" if $self->{ibx_ver} != 1;
+ die "BUG: _msgmap_init is only for v1\n" if $self->{ibx}->version != 1;
$self->{mm} //= eval {
require PublicInbox::Msgmap;
- my $rw = $self->{ibx}->{-no_fsync} ? 2 : 1;
- PublicInbox::Msgmap->new($self->{ibx}->{inboxdir}, $rw);
+ PublicInbox::Msgmap->new_file($self->{ibx}, 1);
};
}
sub add_message {
# mime = PublicInbox::Eml or Email::MIME object
my ($self, $mime, $smsg, $sync) = @_;
+ begin_txn_lazy($self);
my $mids = mids_for_index($mime);
$smsg //= bless { blob => '' }, 'PublicInbox::Smsg'; # test-only compat
$smsg->{mid} //= $mids->[0]; # v1 compatibility
begin_txn_lazy($self);
my $doc = _get_doc($self, $docid) or return;
term_generator($self)->set_document($doc);
- $doc->add_boolean_term('O'.$eidx_key);
+
+ # '.' is special for lei_store
+ $doc->add_boolean_term('O'.$eidx_key) if $eidx_key ne '.';
+
index_list_id($self, $doc, $eml);
$self->{xdb}->replace_document($docid, $doc);
}
$self->{xdb}->replace_document($docid, $doc);
}
-sub smsg_from_doc ($) {
- my ($doc) = @_;
- my $data = $doc->get_data or return;
- my $smsg = bless {}, 'PublicInbox::Smsg';
- $smsg->{ts} = int_val($doc, PublicInbox::Search::TS());
- my $dt = int_val($doc, PublicInbox::Search::DT());
- my ($yyyy, $mon, $dd, $hh, $mm, $ss) = unpack('A4A2A2A2A2A2', $dt);
- $smsg->{ds} = timegm($ss, $mm, $hh, $dd, $mon - 1, $yyyy);
- $smsg->load_from_data($data);
- $smsg;
+sub set_vmd {
+ my ($self, $docid, $vmd) = @_;
+ begin_txn_lazy($self);
+ my $doc = _get_doc($self, $docid) or return;
+ my ($end, @rm, @add);
+ my @x = @VMD_MAP;
+ while (my ($field, $pfx) = splice(@x, 0, 2)) {
+ my $set = $vmd->{$field} // next;
+ my %keep = map { $_ => 1 } @$set;
+ my %add = %keep;
+ $end //= $doc->termlist_end;
+ for (my $cur = $doc->termlist_begin; $cur != $end; $cur++) {
+ $cur->skip_to($pfx);
+ last if $cur == $end;
+ my $v = $cur->get_termname;
+ $v =~ s/\A$pfx//s or next;
+ $keep{$v} ? delete($add{$v}) : push(@rm, $pfx.$v);
+ }
+ push(@add, map { $pfx.$_ } keys %add);
+ }
+ return unless scalar(@rm) || scalar(@add);
+ $doc->remove_term($_) for @rm;
+ $doc->add_boolean_term($_) for @add;
+ $self->{xdb}->replace_document($docid, $doc);
+}
+
+sub apply_vmd_mod ($$) {
+ my ($doc, $vmd_mod) = @_;
+ my $updated = 0;
+ my @x = @VMD_MAP;
+ while (my ($field, $pfx) = splice(@x, 0, 2)) {
+ # field: "L" or "kw"
+ for my $val (@{$vmd_mod->{"-$field"} // []}) {
+ eval {
+ $doc->remove_term($pfx . $val);
+ ++$updated;
+ };
+ }
+ for my $val (@{$vmd_mod->{"+$field"} // []}) {
+ $doc->add_boolean_term($pfx . $val);
+ ++$updated;
+ }
+ }
+ $updated;
+}
+
+sub add_vmd {
+ my ($self, $docid, $vmd) = @_;
+ begin_txn_lazy($self);
+ my $doc = _get_doc($self, $docid) or return;
+ my @x = @VMD_MAP;
+ my $updated = 0;
+ while (my ($field, $pfx) = splice(@x, 0, 2)) {
+ my $add = $vmd->{$field} // next;
+ $doc->add_boolean_term($pfx . $_) for @$add;
+ $updated += scalar(@$add);
+ }
+ $updated += apply_vmd_mod($doc, $vmd);
+ $self->{xdb}->replace_document($docid, $doc) if $updated;
+}
+
+sub remove_vmd {
+ my ($self, $docid, $vmd) = @_;
+ begin_txn_lazy($self);
+ my $doc = _get_doc($self, $docid) or return;
+ my $replace;
+ my @x = @VMD_MAP;
+ while (my ($field, $pfx) = splice(@x, 0, 2)) {
+ my $rm = $vmd->{$field} // next;
+ for (@$rm) {
+ eval {
+ $doc->remove_term($pfx . $_);
+ $replace = 1;
+ };
+ }
+ }
+ $self->{xdb}->replace_document($docid, $doc) if $replace;
+}
+
+sub update_vmd {
+ my ($self, $docid, $vmd_mod) = @_;
+ begin_txn_lazy($self);
+ my $doc = _get_doc($self, $docid) or return;
+ my $updated = apply_vmd_mod($doc, $vmd_mod);
+ $self->{xdb}->replace_document($docid, $doc) if $updated;
+ $updated;
}
sub xdb_remove {
my ($self, @docids) = @_;
- my $xdb = $self->{xdb} or return;
+ begin_txn_lazy($self);
+ my $xdb = $self->{xdb} // die 'BUG: missing {xdb}';
for my $docid (@docids) {
eval { $xdb->delete_document($docid) };
warn "E: #$docid not in in Xapian? $@\n" if $@;
}
}
-sub remove_by_docid {
- my ($self, $num) = @_;
- die "BUG: remove_by_docid is v2-only\n" if $self->{oidx};
- $self->begin_txn_lazy;
- xdb_remove($self, $num) if need_xapian($self);
+sub xdb_remove_quiet {
+ my ($self, $docid) = @_;
+ begin_txn_lazy($self);
+ my $xdb = $self->{xdb} // die 'BUG: missing {xdb}';
+ eval { $xdb->delete_document($docid) };
+ ++$self->{-quiet_rm} unless $@;
}
+sub nr_quiet_rm { delete($_[0]->{-quiet_rm}) // 0 }
+
sub index_git_blob_id {
my ($doc, $pfx, $objid) = @_;
}
}
-# returns the number of bytes to add if given a non-CRLF arg
-sub crlf_adjust ($) {
- if (index($_[0], "\r\n") < 0) {
- # common case is LF-only, every \n needs an \r;
- # so favor a cheap tr// over an expensive m//g
- $_[0] =~ tr/\n/\n/;
- } else { # count number of '\n' w/o '\r', expensive:
- scalar(my @n = ($_[0] =~ m/(?<!\r)\n/g));
- }
-}
-
sub is_bad_blob ($$$$) {
my ($oid, $type, $size, $expect_oid) = @_;
if ($type ne 'blob') {
my ($nr, $max) = @$sync{qw(nr max)};
++$$nr;
$$max -= $size;
- $size += crlf_adjust($$bref);
- my $smsg = bless { bytes => $size, blob => $oid }, 'PublicInbox::Smsg';
+ my $smsg = bless { blob => $oid }, 'PublicInbox::Smsg';
+ $smsg->set_bytes($$bref, $size);
my $self = $sync->{sidx};
local $self->{current_info} = "$self->{current_info}: $oid";
my $eml = PublicInbox::Eml->new($bref);
$smsg->{num} = index_mm($self, $eml, $oid, $sync) or
die "E: could not generate NNTP article number for $oid";
add_message($self, $eml, $smsg, $sync);
+ ++$self->{nidx};
my $cur_cmt = $sync->{cur_cmt} // die 'BUG: {cur_cmt} missing';
${$sync->{latest_cmt}} = $cur_cmt;
}
if (defined(my $cur_cmt = $sync->{cur_cmt})) {
${$sync->{latest_cmt}} = $cur_cmt;
}
+ ++$self->{nidx};
}
sub with_umask {
my ($self, $opt) = @_;
delete $self->{lock_path} if $opt->{-skip_lock};
$self->with_umask(\&_index_sync, $self, $opt);
- if ($opt->{reindex} && !$opt->{quit}) {
+ if ($opt->{reindex} && !$opt->{quit} &&
+ !grep(defined, @$opt{qw(since until)})) {
my %again = %$opt;
delete @again{qw(rethread reindex)};
index_sync($self, \%again);
# $newest may be undef
my $newest = $stk ? $stk->{latest_cmt} : ${$sync->{latest_cmt}};
if (defined($newest)) {
- my $cur = $self->{mm}->last_commit || '';
- if (need_update($self, $cur, $newest)) {
+ my $cur = $self->{mm}->last_commit;
+ if (need_update($self, $sync, $cur, $newest)) {
$self->{mm}->last_commit($newest);
}
}
${$sync->{max}} = $self->{batch_bytes};
$self->{mm}->{dbh}->commit;
- my $xdb = need_xapian($self) ? $self->{xdb} : undef;
+ my $xdb = $self->{xdb};
if ($newest && $xdb) {
my $cur = $xdb->get_metadata('last_commit');
- if (need_update($self, $cur, $newest)) {
+ if (need_update($self, $sync, $cur, $newest)) {
$xdb->set_metadata('last_commit', $newest);
}
}
# Count the new files so they can be added newest to oldest
# and still have numbers increasing from oldest to newest
- my $fh = $git->popen(qw(log --raw -r --pretty=tformat:%at-%ct-%H
- --no-notes --no-color --no-renames --no-abbrev),
- $range);
+ my @cmd = qw(log --raw -r --pretty=tformat:%at-%ct-%H
+ --no-notes --no-color --no-renames --no-abbrev);
+ for my $k (qw(since until)) {
+ my $v = $sync->{-opt}->{$k} // next;
+ next if !$sync->{-opt}->{reindex};
+ push @cmd, "--$k=$v";
+ }
+ my $fh = $git->popen(@cmd, $range);
my ($at, $ct, $stk, $cmt);
while (<$fh>) {
return if $sync->{quit};
$? == 0;
}
-sub need_update ($$$) {
- my ($self, $cur, $new) = @_;
+sub need_update ($$$$) {
+ my ($self, $sync, $cur, $new) = @_;
my $git = $self->{ibx}->git;
- return 1 if $cur && !is_ancestor($git, $cur, $new);
+ $cur //= ''; # XS Search::Xapian ->get_metadata doesn't give undef
+
+ # don't rewind if --{since,until,before,after} are in use
+ return if $cur ne '' &&
+ grep(defined, @{$sync->{-opt}}{qw(since until)}) &&
+ is_ancestor($git, $new, $cur);
+
+ return 1 if $cur ne '' && !is_ancestor($git, $cur, $new);
my $range = $cur eq '' ? $new : "$cur..$new";
chomp(my $n = $git->qx(qw(rev-list --count), $range));
($n eq '' || $n > 0);
$self->with_umask(\&_commit_txn, $self);
}
-sub worker_done {
- my ($self) = @_;
- if (need_xapian($self)) {
- die "$$ $0 xdb not released\n" if $self->{xdb};
- }
- die "$$ $0 still in transaction\n" if $self->{txn};
-}
-
sub eidx_shard_new {
my ($class, $eidx, $shard) = @_;
my $self = bless {
$self;
}
-# ensure there's no stale Xapian docs by treating $over as canonical
-sub over_check {
- my ($self, $over) = @_;
- begin_txn_lazy($self);
- my $sth = $over->dbh->prepare(<<'');
-SELECT COUNT(*) FROM over WHERE num = ?
-
- my $xdb = $self->{xdb};
- my $cur = $xdb->postlist_begin('');
- my $end = $xdb->postlist_end('');
- my $xdir = $self->xdir;
- for (; $cur != $end; $cur++) {
- my $docid = $cur->get_docid;
- $sth->execute($docid);
- my $x = $sth->fetchrow_array;
- next if $x > 0;
- warn "I: removing $xdir #$docid, not in `over'\n";
- $xdb->delete_document($docid);
- }
-}
-
1;