miscidx: index inbox min/max article numbers

This will be used to speed up NNTP group listings and IMAP startup
with thousands of inboxes.

5 files changed, 38 insertions, 16 deletions

diff --git a/lib/PublicInbox/Inbox.pm b/lib/PublicInbox/Inbox.pm
index 0ad68810..3f70e69d 100644
--- a/lib/PublicInbox/Inbox.pm
+++ b/lib/PublicInbox/Inbox.pm
@@ -409,6 +409,16 @@ sub uidvalidity { $_[0]->{uidvalidity} //= eval { $_[0]->mm->created_at } }
 
 sub eidx_key { $_[0]->{newsgroup} // $_[0]->{inboxdir} }
 
+# only used by NNTP, so we need ->mm anyways
+sub art_min { $_[0]->{-art_min} //= eval { $_[0]->mm(1)->min } }
+
+# used by IMAP, too, which tries to avoid ->mm (but ->{mm} is likely
+# faster since it's smaller iff available)
+sub art_max {
+        $_[0]->{-art_max} //= eval { $_[0]->{mm}->max } //
+                                eval { $_[0]->over(1)->max };
+}
+
 sub mailboxid { # rfc 8474, 8620, 8621
         my ($self, $imap_slice) = @_;
         my $pfx = defined($imap_slice) ? $self->{newsgroup} : $self->{name};
diff --git a/lib/PublicInbox/MiscIdx.pm b/lib/PublicInbox/MiscIdx.pm
index 5faf5c66..76b33b16 100644
--- a/lib/PublicInbox/MiscIdx.pm
+++ b/lib/PublicInbox/MiscIdx.pm
@@ -108,12 +108,16 @@ EOF
         $doc->add_boolean_term('Q'.$eidx_key); # uniQue id
         $doc->add_boolean_term('T'.'inbox'); # Type
 
+        # force reread from disk, {description} could be loaded from {misc}
+        delete @$ibx{qw(-art_min -art_max description)};
         if (defined($ibx->{newsgroup}) && $ibx->nntp_usable) {
                 $doc->add_boolean_term('T'.'newsgroup'); # additional Type
+                my $n = $ibx->art_min;
+                add_val($doc, $PublicInbox::MiscSearch::ART_MIN, $n) if $n;
+                $n = $ibx->art_max;
+                add_val($doc, $PublicInbox::MiscSearch::ART_MAX, $n) if $n;
         }
 
-        # force reread from disk, {description} could be loaded from {misc}
-        delete $ibx->{description};
         my $desc = $ibx->description;
 
         # description = S/Subject (or title)
diff --git a/lib/PublicInbox/MiscSearch.pm b/lib/PublicInbox/MiscSearch.pm
index c6d2a062..5fb47d03 100644
--- a/lib/PublicInbox/MiscSearch.pm
+++ b/lib/PublicInbox/MiscSearch.pm
@@ -1,4 +1,4 @@
-# Copyright (C) 2020-2021 all contributors <meta@public-inbox.org>
+# Copyright (C) all contributors <meta@public-inbox.org>
 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
 
 # read-only counterpart to MiscIdx
@@ -11,6 +11,8 @@ my $json;
 # Xapian value columns:
 our $MODIFIED = 0;
 our $UIDVALIDITY = 1; # (created time)
+our $ART_MIN = 2; # NNTP article number
+our $ART_MAX = 3; # NNTP article number
 
 # avoid conflicting with message Search::prob_prefix for UI/UX reasons
 my %PROB_PREFIX = (
@@ -87,14 +89,13 @@ sub ibx_data_once {
         my $term = 'Q'.$ibx->eidx_key; # may be {inboxdir}, so private
         my $head = $xdb->postlist_begin($term);
         my $tail = $xdb->postlist_end($term);
-        if ($head != $tail) {
-                my $doc = $xdb->get_document($head->get_docid);
-                $ibx->{uidvalidity} //= int_val($doc, $UIDVALIDITY);
-                $ibx->{-modified} = int_val($doc, $MODIFIED);
-                $doc->get_data;
-        } else {
-                undef;
-        }
+        return if $head == $tail;
+        my $doc = $xdb->get_document($head->get_docid);
+        $ibx->{uidvalidity} //= int_val($doc, $UIDVALIDITY);
+        $ibx->{-modified} = int_val($doc, $MODIFIED);
+        $ibx->{-art_min} = int_val($doc, $ART_MIN);
+        $ibx->{-art_max} = int_val($doc, $ART_MAX);
+        $doc->get_data;
 }
 
 sub doc2ibx_cache_ent { # @_ == ($self, $doc) OR ($doc)
@@ -109,6 +110,8 @@ sub doc2ibx_cache_ent { # @_ == ($self, $doc) OR ($doc)
         {
                 uidvalidity => int_val($doc, $UIDVALIDITY),
                 -modified => int_val($doc, $MODIFIED),
+                -art_min => int_val($doc, $ART_MIN), # may be undef
+                -art_max => int_val($doc, $ART_MAX), # may be undef
                 # extract description from manifest.js.gz epoch description
                 description => $d
         };
diff --git a/lib/PublicInbox/Msgmap.pm b/lib/PublicInbox/Msgmap.pm
index 1041cd17..cb4bb295 100644
--- a/lib/PublicInbox/Msgmap.pm
+++ b/lib/PublicInbox/Msgmap.pm
@@ -144,13 +144,17 @@ sub max {
         $sth->fetchrow_array // 0;
 }
 
-sub minmax {
-        # breaking MIN and MAX into separate queries speeds up from 250ms
-        # to around 700us with 2.7million messages.
+sub min {
         my $sth = $_[0]->{dbh}->prepare_cached('SELECT MIN(num) FROM msgmap',
                                                 undef, 1);
         $sth->execute;
-        ($sth->fetchrow_array // 0, max($_[0]));
+        $sth->fetchrow_array // 0;
+}
+
+sub minmax {
+        # breaking MIN and MAX into separate queries speeds up from 250ms
+        # to around 700us with 2.7million messages.
+        (min($_[0]), max($_[0]));
 }
 
 sub mid_delete {
diff --git a/lib/PublicInbox/Search.pm b/lib/PublicInbox/Search.pm
index b6141f68..2feb3e13 100644
--- a/lib/PublicInbox/Search.pm
+++ b/lib/PublicInbox/Search.pm
@@ -543,9 +543,10 @@ sub help {
         \@ret;
 }
 
+# always returns a scalar value
 sub int_val ($$) {
         my ($doc, $col) = @_;
-        my $val = $doc->get_value($col) or return; # undefined is '' in Xapian
+        my $val = $doc->get_value($col) or return undef; # undef is '' in Xapian
         sortable_unserialise($val) + 0; # PV => IV conversion
 }