* [PATCH 08/13] www: remove smsg_mime API and adjust callers
2020-06-01 10:06 7% [PATCH 00/13] smsg: remove tricky {mime} field Eric Wong
@ 2020-06-01 10:06 4% ` Eric Wong
0 siblings, 0 replies; 2+ results
From: Eric Wong @ 2020-06-01 10:06 UTC (permalink / raw)
To: meta
To further simplify callers and avoid embarrasing memory
explosions[1], we can finally eliminate this method in
favor of smsg_eml.
[1] commit 7d02b9e64455831d3bda20cd2e64e0c15dc07df5
("view: stop storing all MIME objects on large threads")
fixed a huge memory blowup.
---
Documentation/mknews.perl | 7 ++--
lib/PublicInbox/Feed.pm | 6 ++--
lib/PublicInbox/Inbox.pm | 12 ++-----
lib/PublicInbox/SearchView.pm | 4 +--
lib/PublicInbox/Smsg.pm | 7 ++--
lib/PublicInbox/View.pm | 63 +++++++++++++++++------------------
t/v2mda.t | 4 +--
7 files changed, 50 insertions(+), 53 deletions(-)
diff --git a/Documentation/mknews.perl b/Documentation/mknews.perl
index 3bdebfce7a5..965c30c1dcb 100755
--- a/Documentation/mknews.perl
+++ b/Documentation/mknews.perl
@@ -102,9 +102,10 @@ sub mime2txt {
}
sub mime2html {
- my ($out, $mime, $ctx) = @_;
- my $smsg = bless { mime => $mime }, 'PublicInbox::Smsg';
- print $out PublicInbox::View::index_entry($smsg, $ctx, 1) or die;
+ my ($out, $eml, $ctx) = @_;
+ my $smsg = bless {}, 'PublicInbox::Smsg';
+ $smsg->populate($eml);
+ print $out PublicInbox::View::eml_entry($ctx, $smsg, $eml, 1) or die;
}
sub html_start {
diff --git a/lib/PublicInbox/Feed.pm b/lib/PublicInbox/Feed.pm
index b770a35077c..4c1056b4665 100644
--- a/lib/PublicInbox/Feed.pm
+++ b/lib/PublicInbox/Feed.pm
@@ -53,9 +53,9 @@ sub new_html_i {
my ($nr, $ctx) = @_;
my $msgs = $ctx->{msgs};
while (my $smsg = shift @$msgs) {
- my $m = $ctx->{-inbox}->smsg_mime($smsg) or next;
- my $more = scalar @$msgs;
- return PublicInbox::View::index_entry($m, $ctx, $more);
+ my $eml = $ctx->{-inbox}->smsg_eml($smsg) or next;
+ return PublicInbox::View::eml_entry($ctx, $smsg, $eml,
+ scalar @$msgs);
}
PublicInbox::View::pagination_footer($ctx, './new.html');
}
diff --git a/lib/PublicInbox/Inbox.pm b/lib/PublicInbox/Inbox.pm
index 38abdfe5847..af034358b15 100644
--- a/lib/PublicInbox/Inbox.pm
+++ b/lib/PublicInbox/Inbox.pm
@@ -326,18 +326,12 @@ sub msg_by_smsg ($$) {
git($self)->cat_file($blob);
}
-sub smsg_mime {
- my ($self, $smsg) = @_;
- if (my $s = msg_by_smsg($self, $smsg)) {
- $smsg->{mime} = PublicInbox::Eml->new($s);
- return $smsg;
- }
-}
-
sub smsg_eml {
my ($self, $smsg) = @_;
my $bref = msg_by_smsg($self, $smsg) or return;
- PublicInbox::Eml->new($bref);
+ my $eml = PublicInbox::Eml->new($bref);
+ $smsg->populate($eml) unless exists($smsg->{num}); # v1 w/o SQLite
+ $eml;
}
sub mid2num($$) {
diff --git a/lib/PublicInbox/SearchView.pm b/lib/PublicInbox/SearchView.pm
index 249cf53926d..d53a533e53c 100644
--- a/lib/PublicInbox/SearchView.pm
+++ b/lib/PublicInbox/SearchView.pm
@@ -300,8 +300,8 @@ sub mset_thread_i {
my ($nr, $ctx) = @_;
my $msgs = $ctx->{msgs} or return;
while (my $smsg = pop @$msgs) {
- $ctx->{-inbox}->smsg_mime($smsg) or next;
- return PublicInbox::View::index_entry($smsg, $ctx,
+ my $eml = $ctx->{-inbox}->smsg_eml($smsg) or next;
+ return PublicInbox::View::eml_entry($ctx, $smsg, $eml,
scalar @$msgs);
}
my ($skel) = delete @$ctx{qw(skel msgs)};
diff --git a/lib/PublicInbox/Smsg.pm b/lib/PublicInbox/Smsg.pm
index 446bca81b48..9688c5592a2 100644
--- a/lib/PublicInbox/Smsg.pm
+++ b/lib/PublicInbox/Smsg.pm
@@ -12,7 +12,7 @@ use strict;
use warnings;
use base qw(Exporter);
our @EXPORT_OK = qw(subject_normalized);
-use PublicInbox::MID qw/mid_mime/;
+use PublicInbox::MID qw(mid_mime mids);
use PublicInbox::Address;
use PublicInbox::MsgTime qw(msg_timestamp msg_datestamp);
use Time::Local qw(timegm);
@@ -105,7 +105,7 @@ sub __hdr ($$) {
};
}
-# for Import and v1 WWW code paths
+# for Import and v1 non-SQLite WWW code paths
sub populate {
my ($self, $hdr, $v2w) = @_;
for my $f (qw(From To Cc Subject)) {
@@ -133,6 +133,9 @@ sub populate {
$self->{-ts} = [ my @ts = msg_timestamp($hdr, $v2w->{cotime}) ];
$self->{ds} //= $ds[0]; # no zone
$self->{ts} //= $ts[0];
+
+ # for v1 users w/o SQLite
+ $self->{mid} //= eval { mids($hdr)->[0] } // '';
}
sub subject ($) { __hdr($_[0], 'Subject') }
diff --git a/lib/PublicInbox/View.pm b/lib/PublicInbox/View.pm
index a05ac4142f2..0bc2b06e4dc 100644
--- a/lib/PublicInbox/View.pm
+++ b/lib/PublicInbox/View.pm
@@ -74,11 +74,10 @@ sub msg_page_more { # cold
my $ibx = $ctx->{-inbox};
my $next = $ibx->over->next_by_mid($ctx->{mid}, \$id, \$prev);
$ctx->{more} = [ $id, $prev, $next ] if $next;
- $smsg = $ibx->smsg_mime($smsg) or return '';
+ my $eml = $ibx->smsg_eml($smsg) or return '';
$ctx->{mhref} = '../' . mid_href($smsg->{mid}) . '/';
- my $mime = delete $smsg->{mime};
- $ctx->{obuf} = _msg_page_prepare_obuf($mime->header_obj, $ctx, $nr);
- multipart_text_as_html($mime, $ctx);
+ $ctx->{obuf} = _msg_page_prepare_obuf($eml->header_obj, $ctx, $nr);
+ multipart_text_as_html($eml, $ctx);
${delete $ctx->{obuf}} .= '</pre><hr>';
}
@@ -181,14 +180,14 @@ sub nr_to_s ($$$) {
# human-friendly format
sub fmt_ts ($) { strftime('%Y-%m-%d %k:%M', gmtime($_[0])) }
+# Displays the text of of the message for /$INBOX/$MSGID/[Tt]/ endpoint
# this is already inside a <pre>
-sub index_entry {
- my ($smsg, $ctx, $more) = @_;
- my $subj = $smsg->subject;
- my $mid_raw = $smsg->mid;
+sub eml_entry {
+ my ($ctx, $smsg, $eml, $more) = @_;
+ my $subj = delete $smsg->{subject};
+ my $mid_raw = $smsg->{mid};
my $id = id_compress($mid_raw, 1);
my $id_m = 'm'.$id;
-
my $root_anchor = $ctx->{root_anchor} || '';
my $irt;
my $obfs_ibx = $ctx->{-obfs_ibx};
@@ -201,12 +200,12 @@ sub index_entry {
$rv .= $subj . "\n";
$rv .= _th_index_lite($mid_raw, \$irt, $id, $ctx);
my @tocc;
- my $ds = $smsg->ds; # for v1 non-Xapian/SQLite users
- # deleting {mime} is critical to memory use,
- # the rest of the fields saves about 400K as we iterate across 1K msgs
- my ($mime) = delete @$smsg{qw(mime ds ts blob subject)};
+ my $ds = delete $smsg->{ds}; # for v1 non-Xapian/SQLite users
+
+ # Deleting these fields saves about 400K as we iterate across 1K msgs
+ delete @$smsg{qw(ts blob)};
- my $hdr = $mime->header_obj;
+ my $hdr = $eml->header_obj;
my $from = _hdr_names_html($hdr, 'From');
obfuscate_addrs($obfs_ibx, $from) if $obfs_ibx;
$rv .= "From: $from @ ".fmt_ts($ds)." UTC";
@@ -244,7 +243,7 @@ sub index_entry {
# scan through all parts, looking for displayable text
$ctx->{mhref} = $mhref;
$ctx->{obuf} = \$rv;
- $mime->each_part(\&add_text_body, $ctx, 1);
+ $eml->each_part(\&add_text_body, $ctx, 1);
delete $ctx->{obuf};
# add the footer
@@ -372,10 +371,10 @@ sub pre_thread { # walk_thread callback
skel_dump($ctx, $level, $node);
}
-sub thread_index_entry {
- my ($ctx, $level, $smsg) = @_;
+sub thread_eml_entry {
+ my ($ctx, $level, $smsg, $eml) = @_;
my ($beg, $end) = thread_adj_level($ctx, $level);
- $beg . '<pre>' . index_entry($smsg, $ctx, 0) . '</pre>' . $end;
+ $beg . '<pre>' . eml_entry($ctx, $smsg, $eml, 0) . '</pre>' . $end;
}
sub stream_thread_i { # PublicInbox::WwwStream::getline callback
@@ -387,8 +386,8 @@ sub stream_thread_i { # PublicInbox::WwwStream::getline callback
my $node = shift @$q or next;
my $cl = $level + 1;
unshift @$q, map { ($cl, $_) } @{$node->{children}};
- if ($ctx->{-inbox}->smsg_mime($node)) {
- return thread_index_entry($ctx, $level, $node);
+ if (my $eml = $ctx->{-inbox}->smsg_eml($node)) {
+ return thread_eml_entry($ctx, $level, $node, $eml);
} else {
return ghost_index_entry($ctx, $level, $node);
}
@@ -400,19 +399,19 @@ sub stream_thread ($$) {
my ($rootset, $ctx) = @_;
my $ibx = $ctx->{-inbox};
my @q = map { (0, $_) } @$rootset;
- my ($smsg, $level);
+ my ($smsg, $eml, $level);
while (@q) {
$level = shift @q;
- my $node = shift @q or next;
+ $smsg = shift @q or next;
my $cl = $level + 1;
- unshift @q, map { ($cl, $_) } @{$node->{children}};
- $smsg = $ibx->smsg_mime($node) and last;
+ unshift @q, map { ($cl, $_) } @{$smsg->{children}};
+ $eml = $ibx->smsg_eml($smsg) and last;
}
- return missing_thread($ctx) unless $smsg;
+ return missing_thread($ctx) unless $eml;
$ctx->{-obfs_ibx} = $ibx->{obfuscate} ? $ibx : undef;
$ctx->{-title_html} = ascii_html($smsg->{subject});
- $ctx->{-html_tip} = thread_index_entry($ctx, $level, $smsg);
+ $ctx->{-html_tip} = thread_eml_entry($ctx, $level, $smsg, $eml);
$ctx->{-queue} = \@q;
PublicInbox::WwwStream->response($ctx, 200, \&stream_thread_i);
}
@@ -452,13 +451,13 @@ sub thread_html {
return stream_thread($rootset, $ctx) unless $ctx->{flat};
# flat display: lazy load the full message from smsg
- my $smsg;
- while (my $m = shift @$msgs) {
- $smsg = $ibx->smsg_mime($m) and last;
+ my ($smsg, $eml);
+ while ($smsg = shift @$msgs) {
+ $eml = $ibx->smsg_eml($smsg) and last;
}
return missing_thread($ctx) unless $smsg;
$ctx->{-title_html} = ascii_html($smsg->{subject});
- $ctx->{-html_tip} = '<pre>'.index_entry($smsg, $ctx, scalar @$msgs);
+ $ctx->{-html_tip} = '<pre>'.eml_entry($ctx, $smsg, $eml, scalar @$msgs);
$ctx->{msgs} = $msgs;
PublicInbox::WwwStream->response($ctx, 200, \&thread_html_i);
}
@@ -467,8 +466,8 @@ sub thread_html_i { # PublicInbox::WwwStream::getline callback
my ($nr, $ctx) = @_;
my $msgs = $ctx->{msgs} or return;
while (my $smsg = shift @$msgs) {
- $ctx->{-inbox}->smsg_mime($smsg) or next;
- return index_entry($smsg, $ctx, scalar @$msgs);
+ my $eml = $ctx->{-inbox}->smsg_eml($smsg) or next;
+ return eml_entry($ctx, $smsg, $eml, scalar @$msgs);
}
my ($skel) = delete @$ctx{qw(skel msgs)};
$$skel;
diff --git a/t/v2mda.t b/t/v2mda.t
index 36f43ff096c..7666eb2dacd 100644
--- a/t/v2mda.t
+++ b/t/v2mda.t
@@ -52,8 +52,8 @@ if ($V == 1) {
}
my $msgs = $ibx->search->query('');
is(scalar(@$msgs), 1, 'only got one message');
-my $saved = $ibx->smsg_mime($msgs->[0]);
-is($saved->{mime}->as_string, $mime->as_string, 'injected message');
+my $eml = $ibx->smsg_eml($msgs->[0]);
+is($eml->as_string, $mime->as_string, 'injected message');
{
my @new = glob("$faildir/new/*");
^ permalink raw reply related [relevance 4%]
* [PATCH 00/13] smsg: remove tricky {mime} field
@ 2020-06-01 10:06 7% Eric Wong
2020-06-01 10:06 4% ` [PATCH 08/13] www: remove smsg_mime API and adjust callers Eric Wong
0 siblings, 1 reply; 2+ results
From: Eric Wong @ 2020-06-01 10:06 UTC (permalink / raw)
To: meta
Storing a large PublicInbox::Eml (or in the past, Email::MIME)
object inside a small PublicInbox::Smsg object has historically
been bloat-prone[1] since there may be many small smsgs in
memory at once
Hundreds or thousands of $smsg objects can linger in memory due
to search results and message threading operations. So keep
$eml and $smsg objects independent of each other, for now.
Instead, we'll introduce a $smsg->populate($eml) API to handle
filling in the keys for the importer, indexer, and
non-SQLite-using WWW users.
Furthermore, $smsg->$field dispatch has always been measurably
faster than $smsg->{$field} access in NNTP. Since $smsg->$field
became read-only with the removal of $smsg->{mime}, we can
abandon the $smsg->$field invocations and favor of direct hash
access.
[1] the prime example being what commit 7d02b9e64455831d fixed
("view: stop storing all MIME objects on large threads")
Eric Wong (13):
inbox: introduce smsg_eml method
wwwatomstream: convert callers to use smsg_eml
v2writable: fix non-sensical interpolation in BUG message
import: modernize to use Perl 5.10 features
smsg: introduce ->populate method
smsg: get rid of ->wrap initializer, too
inbox: msg_by_*: remove $(size)ref args
www: remove smsg_mime API and adjust callers
nntp: smsg_range_i: favor ->{$field} lookups when possible
smsg: get rid of remaining {mime} users
smsg: remove ->bytes and ->lines methods
smsg: remove remaining accessor methods
wwwatomstream: drop smsg->{mid} fallback for non-SQLite
Documentation/mknews.perl | 7 +-
lib/PublicInbox/ExtMsg.pm | 2 +-
lib/PublicInbox/Feed.pm | 8 +-
lib/PublicInbox/Import.pm | 69 ++++++++---------
lib/PublicInbox/Inbox.pm | 32 ++++----
lib/PublicInbox/Mbox.pm | 2 +-
lib/PublicInbox/NNTP.pm | 14 +++-
lib/PublicInbox/OverIdx.pm | 3 +-
lib/PublicInbox/SearchIdx.pm | 33 ++++-----
lib/PublicInbox/SearchView.pm | 6 +-
lib/PublicInbox/Smsg.pm | 123 +++++++++++--------------------
lib/PublicInbox/SolverGit.pm | 4 +-
lib/PublicInbox/V2Writable.pm | 11 +--
lib/PublicInbox/View.pm | 63 ++++++++--------
lib/PublicInbox/WwwAtomStream.pm | 8 +-
t/altid.t | 3 +-
t/altid_v2.t | 3 +-
t/import.t | 3 +-
t/search.t | 46 +++++++-----
t/v2mda.t | 4 +-
t/v2writable.t | 5 +-
21 files changed, 207 insertions(+), 242 deletions(-)
^ permalink raw reply [relevance 7%]
Results 1-2 of 2 | reverse | options above
-- pct% links below jump to the message on this page, permalinks otherwise --
2020-06-01 10:06 7% [PATCH 00/13] smsg: remove tricky {mime} field Eric Wong
2020-06-01 10:06 4% ` [PATCH 08/13] www: remove smsg_mime API and adjust callers Eric Wong
Code repositories for project(s) associated with this public 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).