user/dev discussion of public-inbox itself
 help / color / mirror / code / Atom feed
From: Eric Wong <e@yhbt.net>
To: meta@public-inbox.org
Subject: [PATCH 75/82] imap: STATUS/EXAMINE: rely on SQLite overview
Date: Wed, 10 Jun 2020 07:05:12 +0000	[thread overview]
Message-ID: <20200610070519.18252-76-e@yhbt.net> (raw)
In-Reply-To: <20200610070519.18252-1-e@yhbt.net>

We can get exact values for EXISTS, UIDNEXT using SQLite
rather than calculating off $ibx->mm->max ourselves.

Furthermore, $ibx->mm is less useful than $ibx->over for IMAP
(and for our read-only daemons in general) so do not depend on
$ibx->mm outside of startup/reload to save FDs and reduce kernel
page cache footprint.
---
 lib/PublicInbox/DummyInbox.pm | 11 ++++++-----
 lib/PublicInbox/IMAP.pm       | 30 ++++++++++++++++++------------
 lib/PublicInbox/IMAPD.pm      |  2 +-
 lib/PublicInbox/Over.pm       | 30 ++++++++++++++++++++++++++++++
 t/over.t                      |  3 +++
 5 files changed, 58 insertions(+), 18 deletions(-)

diff --git a/lib/PublicInbox/DummyInbox.pm b/lib/PublicInbox/DummyInbox.pm
index b6c48db1a0e..c3f4f5c6674 100644
--- a/lib/PublicInbox/DummyInbox.pm
+++ b/lib/PublicInbox/DummyInbox.pm
@@ -9,13 +9,14 @@ use strict;
 
 sub created_at { 0 } # Msgmap::created_at
 sub mm { shift }
-sub max { undef } # Msgmap::max
-sub msg_range { [] } # Msgmap::msg_range
+sub uid_range { [] } # Over::uid_range
+sub subscribe_unlock { undef };
 
 no warnings 'once';
-*uid_range = *query_xover = \&msg_range;
+*max = \&created_at;
+*query_xover = \&uid_range;
 *over = \&mm;
-*subscribe_unlock = *unsubscribe_unlock =
-	*get_art = *description = *base_url = \&max;
+*unsubscribe_unlock =
+	*get_art = *description = *base_url = \&subscribe_unlock;
 
 1;
diff --git a/lib/PublicInbox/IMAP.pm b/lib/PublicInbox/IMAP.pm
index d0683530988..41c8066446c 100644
--- a/lib/PublicInbox/IMAP.pm
+++ b/lib/PublicInbox/IMAP.pm
@@ -183,7 +183,7 @@ sub cmd_noop ($$) { "$_[1] OK Noop done\r\n" }
 # called by PublicInbox::InboxIdle
 sub on_inbox_unlock {
 	my ($self, $ibx) = @_;
-	my $new = $ibx->mm->max;
+	my $new = $ibx->over->max;
 	my $uid_base = $self->{uid_base} // 0;
 	my $uid_end = $uid_base + UID_BLOCK;
 	defined(my $old = $self->{-idle_max}) or die 'BUG: -idle_max unset';
@@ -223,7 +223,7 @@ sub cmd_idle ($$) {
 	# IDLE seems allowed by dovecot w/o a mailbox selected *shrug*
 	my $ibx = $self->{ibx} or return "$tag BAD no mailbox selected\r\n";
 	$self->{-idle_tag} = $tag;
-	my $max = $ibx->mm->max // 0;
+	my $max = $ibx->over->max;
 	my $uid_end = $self->{uid_base} + UID_BLOCK;
 	my $sock = $self->{sock} or return;
 	my $fd = fileno($sock);
@@ -283,14 +283,20 @@ sub inbox_lookup ($$) {
 		my $mb_top = $1;
 		$uid_base = $2 * UID_BLOCK;
 		$ibx = $self->{imapd}->{mailboxes}->{lc $mailbox} or return;
-		$exists = $ibx->mm->max // 0;
-		ensure_ranges_exist($self->{imapd}, $ibx, $exists);
-		my $uid_end = $uid_base + UID_BLOCK;
-		$exists = $uid_end if $exists > $uid_end;
-		$uidnext = $exists + 1;
-		$exists -= $uid_base;
+		my $max;
+		($exists, $uidnext, $max) = $ibx->over->imap_status($uid_base,
+							$uid_base + UID_BLOCK);
+		ensure_ranges_exist($self->{imapd}, $ibx, $max);
 	} else { # check for dummy inboxes
-		$ibx = $self->{imapd}->{mailboxes}->{lc $mailbox} or return;
+		$mailbox = lc $mailbox;
+		$ibx = $self->{imapd}->{mailboxes}->{$mailbox} or return;
+
+		# if "INBOX.foo.bar" is selected and "INBOX.foo.bar.0",
+		# check for new UID ranges (e.g. "INBOX.foo.bar.1")
+		if (my $z = $self->{imapd}->{mailboxes}->{"$mailbox.0"}) {
+			ensure_ranges_exist($self->{imapd}, $z, $z->over->max);
+		}
+
 		$uid_base = $exists = 0;
 		$uidnext = 1;
 	}
@@ -590,7 +596,7 @@ sub range_step ($$) {
 		($beg, $end) = ($1 + 0, $2 + 0);
 	} elsif ($range =~ /\A([0-9]+):\*\z/) {
 		$beg = $1 + 0;
-		$end = $self->{ibx}->mm->max // 0;
+		$end = $self->{ibx}->over->max;
 		my $uid_end = $self->{uid_base} + UID_BLOCK;
 		$end = $uid_end if $end > $uid_end;
 		$beg = $end if $beg > $end;
@@ -1093,14 +1099,14 @@ sub cmd_uid_search ($$$;) {
 	if (!scalar(keys %$q)) {
 		$self->msg_more('* SEARCH');
 		my $beg = 1;
-		my $end = $ibx->mm->max // 0;
+		my $end = $ibx->over->max;
 		uid_clamp($self, \$beg, \$end);
 		long_response($self, \&uid_search_uid_range,
 				$tag, \$beg, $end, $sql);
 	} elsif (my $uid = $q->{uid}) {
 		if ($uid =~ /\A([0-9]+):([0-9]+|\*)\z/s) {
 			my ($beg, $end) = ($1, $2);
-			$end = $ibx->mm->max if $end eq '*';
+			$end = $ibx->over->max if $end eq '*';
 			uid_clamp($self, \$beg, \$end);
 			$self->msg_more('* SEARCH');
 			long_response($self, \&uid_search_uid_range,
diff --git a/lib/PublicInbox/IMAPD.pm b/lib/PublicInbox/IMAPD.pm
index 186ec7b0062..e4dc90e8d0e 100644
--- a/lib/PublicInbox/IMAPD.pm
+++ b/lib/PublicInbox/IMAPD.pm
@@ -44,7 +44,7 @@ sub imapd_refresh_ibx { # pi_config->each_inbox cb
 	# this case is a 32-bit representation of the creation
 	# date/time of the mailbox"
 	defined($ibx->{uidvalidity} = $mm->created_at) or return;
-	PublicInbox::IMAP::ensure_ranges_exist($imapd, $ibx, $mm->max // 1);
+	PublicInbox::IMAP::ensure_ranges_exist($imapd, $ibx, $mm->max // 0);
 
 	# preload to avoid fragmentation:
 	$ibx->description;
diff --git a/lib/PublicInbox/Over.pm b/lib/PublicInbox/Over.pm
index 1faeff418f4..e0f20ea6d3d 100644
--- a/lib/PublicInbox/Over.pm
+++ b/lib/PublicInbox/Over.pm
@@ -229,4 +229,34 @@ sub uid_range {
 	$dbh->selectcol_arrayref($q, undef, $beg, $end);
 }
 
+sub max {
+	my ($self) = @_;
+	my $sth = $self->connect->prepare_cached(<<'', undef, 1);
+SELECT MAX(num) FROM over WHERE num > 0
+
+	$sth->execute;
+	$sth->fetchrow_array // 0;
+}
+
+sub imap_status {
+	my ($self, $uid_base, $uid_end) = @_;
+	my $dbh = $self->connect;
+	my $sth = $dbh->prepare_cached(<<'', undef, 1);
+SELECT COUNT(num) FROM over WHERE num > ? AND num <= ?
+
+	$sth->execute($uid_base, $uid_end);
+	my $exists = $sth->fetchrow_array;
+
+	$sth = $dbh->prepare_cached(<<'', undef, 1);
+SELECT MAX(num) + 1 FROM over WHERE num <= ?
+
+	$sth->execute($uid_end);
+	my $uidnext = $sth->fetchrow_array;
+
+	$sth = $dbh->prepare_cached(<<'', undef, 1);
+SELECT MAX(num) FROM over WHERE num > 0
+
+	($exists, $uidnext, $sth->fetchrow_array // 0);
+}
+
 1;
diff --git a/t/over.t b/t/over.t
index 5586aa310e6..734fdaa3604 100644
--- a/t/over.t
+++ b/t/over.t
@@ -10,6 +10,8 @@ use_ok 'PublicInbox::OverIdx';
 my ($tmpdir, $for_destroy) = tmpdir();
 my $over = PublicInbox::OverIdx->new("$tmpdir/over.sqlite3");
 $over->connect;
+is($over->max, 0, 'max is zero on new DB (scalar context)');
+is_deeply([$over->max], [0], 'max is zero on new DB (list context)');
 my $x = $over->next_tid;
 is(int($x), $x, 'integer tid');
 my $y = $over->next_tid;
@@ -58,6 +60,7 @@ foreach my $mid (qw(a b)) {
 	my $msgs = [ map { $_->{num} } @{$over->get_thread($mid)} ];
 	is_deeply([98, 99], $msgs, "linked messages by Message-ID: <$mid>");
 }
+isnt($over->max, 0, 'max is non-zero');
 
 $over->rollback_lazy;
 

  parent reply	other threads:[~2020-06-10  7:08 UTC|newest]

Thread overview: 84+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-06-10  7:03 [PATCH 00/82] public-inbox-imapd: read-only IMAP server Eric Wong
2020-06-10  7:03 ` [PATCH 01/82] doc: add some IMAP standards Eric Wong
2020-06-10  7:03 ` [PATCH 02/82] nntpd: restrict allowed newsgroup names Eric Wong
2020-06-10  7:04 ` [PATCH 03/82] preliminary imap server implementation Eric Wong
2020-06-10  7:04 ` [PATCH 04/82] inboxidle: new class to detect inbox changes Eric Wong
2020-06-10  7:04 ` [PATCH 05/82] imap: support IDLE Eric Wong
2020-06-10  7:04 ` [PATCH 06/82] msgmap: split ->max into its own method Eric Wong
2020-06-10  7:04 ` [PATCH 07/82] imap: delay InboxIdle start, support refresh Eric Wong
2020-06-10  7:04 ` [PATCH 08/82] imap: implement STATUS command Eric Wong
2020-06-10  7:04 ` [PATCH 09/82] imap: use Text::ParseWords::parse_line to handle quoted words Eric Wong
2020-06-10  7:04 ` [PATCH 10/82] imap: support LIST command Eric Wong
2020-06-10  7:04 ` [PATCH 11/82] t/imapd: support FakeInotify and KQNotify Eric Wong
2020-06-10  7:04 ` [PATCH 12/82] imap: support fetch for BODYSTRUCTURE and BODY Eric Wong
2020-06-10  7:04 ` [PATCH 13/82] eml: each_part: single part $idx is 1 Eric Wong
2020-06-10  7:04 ` [PATCH 14/82] imap: allow fetch of partial of BODY[...] and headers Eric Wong
2020-06-10  7:04 ` [PATCH 15/82] imap: always include `resp-text' in responses Eric Wong
2020-06-10  7:04 ` [PATCH 16/82] imap: split out unit tests and benchmarks Eric Wong
2020-06-10  7:04 ` [PATCH 17/82] imap: fix multi-message partial header fetches Eric Wong
2020-06-10  7:04 ` [PATCH 18/82] imap: simplify partial fetch structure Eric Wong
2020-06-10  7:04 ` [PATCH 19/82] imap: support sequence number FETCH Eric Wong
2020-06-10  7:04 ` [PATCH 20/82] imap: do not include ".PEEK" in responses Eric Wong
2020-06-10  7:04 ` [PATCH 21/82] imap: support the CLOSE command Eric Wong
2020-06-10  7:04 ` [PATCH 22/82] imap: speed up HEADER.FIELDS[.NOT] range fetches Eric Wong
2020-06-10  7:04 ` [PATCH 23/82] git: async: flatten the inflight array Eric Wong
2020-06-10  7:04 ` [PATCH 24/82] git: do our own read buffering for cat-file Eric Wong
2020-06-10  7:04 ` [PATCH 25/82] imap: use git-cat-file asynchronously Eric Wong
2020-06-10  7:04 ` [PATCH 26/82] git: idle rbuf for async Eric Wong
2020-06-10  7:04 ` [PATCH 27/82] imap: support LSUB command Eric Wong
2020-06-10  7:04 ` [PATCH 28/82] imap: FETCH: support comma-delimited ranges Eric Wong
2020-06-10  7:04 ` [PATCH 29/82] add imapd compression test Eric Wong
2020-06-10  7:04 ` [PATCH 30/82] testcommon: tcp_(server|connect): BAIL_OUT on failure Eric Wong
2020-06-10  7:04 ` [PATCH 31/82] *deflate: drop invalid comment about rbuf Eric Wong
2020-06-10  7:04 ` [PATCH 32/82] imap: fix pipelining with async git Eric Wong
2020-06-10  7:04 ` [PATCH 33/82] git: cat_async: provide requested OID + "missing" on missing blobs Eric Wong
2020-06-10  7:04 ` [PATCH 34/82] git: move async_cat reference to PublicInbox::Git Eric Wong
2020-06-10  7:04 ` [PATCH 35/82] git: async: automatic retry on alternates change Eric Wong
2020-06-10  7:04 ` [PATCH 36/82] imapclient: wrapper for Mail::IMAPClient Eric Wong
2020-06-10  7:04 ` [PATCH 37/82] xt: add imapd-validate and imapd-mbsync-oimap Eric Wong
2020-06-10  7:04 ` [PATCH 38/82] imap: support out-of-bounds ranges Eric Wong
2020-06-10  7:04 ` [PATCH 39/82] xt/perf-imap-list: time refresh_inboxlist Eric Wong
2020-06-10  7:04 ` [PATCH 40/82] imap: case-insensitive mailbox name comparisons Eric Wong
2020-06-10  7:04 ` [PATCH 41/82] imap: break giant inboxes into sub-inboxes of 50K messages Eric Wong
2020-06-10  7:04 ` [PATCH 42/82] imap: start doing iterative config reloading Eric Wong
2020-06-10  7:04 ` [PATCH 43/82] imap: require ".$UID_MIN-$UID_END" suffix Eric Wong
2020-06-10  7:04 ` [PATCH 44/82] imapd: ensure LIST is sorted alphabetically, for now Eric Wong
2020-06-10  7:04 ` [PATCH 45/82] imap: omit $UID_END from mailbox name, use index Eric Wong
2020-06-10  7:04 ` [PATCH 46/82] t/config.t: always compare against git bool behavior Eric Wong
2020-06-10  7:04 ` [PATCH 47/82] xt/*: show some tunable parameters Eric Wong
2020-06-10  7:04 ` [PATCH 48/82] imap: STATUS and LIST are case-insensitive, too Eric Wong
2020-06-10  7:04 ` [PATCH 49/82] imap: EXAMINE/STATUS: return correct counts Eric Wong
2020-06-10  7:04 ` [PATCH 50/82] imap: avoid uninitialized warnings on incomplete commands Eric Wong
2020-06-10  7:04 ` [PATCH 51/82] imap: start parsing out queries for SQLite and Xapian Eric Wong
2020-06-10  7:04 ` [PATCH 52/82] imap: SEARCH: clamp results to the 50K UID range Eric Wong
2020-06-10  7:04 ` [PATCH 53/82] imap: allow UID range search on timestamps Eric Wong
2020-06-10  7:04 ` [PATCH 54/82] over: get_art: use dbh->prepare_cached Eric Wong
2020-06-10  7:04 ` [PATCH 55/82] search: index byte size of a message for IMAP search Eric Wong
2020-06-10  7:04 ` [PATCH 56/82] search: index UID for IMAP search, too Eric Wong
2020-06-10  7:04 ` [PATCH 57/82] imap: remove dummies from sequence number FETCH Eric Wong
2020-06-10  7:04 ` [PATCH 58/82] imap: compile UID FETCH to opcodes Eric Wong
2020-06-10  7:04 ` [PATCH 59/82] imap: UID FETCH: optimize for smsg-only case Eric Wong
2020-06-10  7:04 ` [PATCH 60/82] imap: UID FETCH: optimize (UID FLAGS) harder Eric Wong
2020-06-10  7:04 ` [PATCH 61/82] imap: IDLE: avoid extraneous wakeups, keep-alive Eric Wong
2020-06-10  7:04 ` [PATCH 62/82] imap: 30 minute auto-logout timer Eric Wong
2020-06-10  7:05 ` [PATCH 63/82] imap: split ->logged_in attribute into a separate class Eric Wong
2020-06-10  7:05 ` [PATCH 64/82] searchidx: v1 (re)-index uses git asynchronously Eric Wong
2020-06-10  7:05 ` [PATCH 65/82] index: account for CRLF conversion when storing bytes Eric Wong
2020-06-10  7:05 ` [PATCH 66/82] imap: rely on smsg->{bytes} for RFC822.SIZE Eric Wong
2020-06-10  7:05 ` [PATCH 67/82] imap: UID FETCH requires at least one data item Eric Wong
2020-06-10  7:05 ` [PATCH 68/82] imap: LIST shows "INBOX" in all caps Eric Wong
2020-06-10  7:05 ` [PATCH 69/82] imap: support 8000 octet lines Eric Wong
2020-06-10  7:05 ` [PATCH 70/82] imap: reinstate some message sequence number support Eric Wong
2020-06-10  7:05 ` [PATCH 71/82] imap: cleanup ->{uid_base} usage Eric Wong
2020-06-10  7:05 ` [PATCH 72/82] imap: FETCH: more granular CRLF conversion Eric Wong
2020-06-10  7:05 ` [PATCH 73/82] imap: further speed up HEADER.FIELDS FETCH requests Eric Wong
2020-06-10  7:05 ` [PATCH 74/82] imap: FETCH: try to make fake MSNs sequentially Eric Wong
2020-06-10  7:05 ` Eric Wong [this message]
2020-06-10  7:05 ` [PATCH 76/82] imap: UID SEARCH: support multiple ranges Eric Wong
2020-06-10  7:05 ` [PATCH 77/82] imap: wire up Xapian, MSN SEARCH and multi sequence-sets Eric Wong
2020-06-10  7:05 ` [PATCH 78/82] imap: misc cleanups and notes Eric Wong
2020-06-10  7:05 ` [PATCH 79/82] imapd: don't bother sorting LIST output Eric Wong
2020-06-10  7:05 ` [PATCH 80/82] imap: remove non-UID SEARCH for now Eric Wong
2020-06-10  7:05 ` [PATCH 81/82] over: uid_range: remove LIMIT Eric Wong
2020-06-10  7:05 ` [PATCH 82/82] imap: FETCH: proper MSN => UID mapping for requests Eric Wong
2020-06-12 23:49 ` [PATCH 83/82] imap: introduce memory-efficient uo2m mapping Eric Wong

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

  List information: https://public-inbox.org/README

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20200610070519.18252-76-e@yhbt.net \
    --to=e@yhbt.net \
    --cc=meta@public-inbox.org \
    --subject='Re: [PATCH 75/82] imap: STATUS/EXAMINE: rely on SQLite overview' \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

Code repositories for project(s) associated with this inbox:

	https://80x24.org/public-inbox.git

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).