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 02/43] wwwstream: oneshot: perform gzip without middleware
Date: Sun,  5 Jul 2020 23:27:18 +0000	[thread overview]
Message-ID: <20200705232759.3161-3-e@yhbt.net> (raw)
In-Reply-To: <20200705232759.3161-1-e@yhbt.net>

Plack::Middleware::Deflater forces us to use a memory-intensive
closure.  Instead, work towards building compressed strings in
memory to reduce the overhead of buffering large HTML output.
---
 lib/PublicInbox/GzipFilter.pm | 13 +++++++++++++
 lib/PublicInbox/WwwStream.pm  | 27 ++++++++++++++++++++++-----
 2 files changed, 35 insertions(+), 5 deletions(-)

diff --git a/lib/PublicInbox/GzipFilter.pm b/lib/PublicInbox/GzipFilter.pm
index a7355a8df..115660cb1 100644
--- a/lib/PublicInbox/GzipFilter.pm
+++ b/lib/PublicInbox/GzipFilter.pm
@@ -4,7 +4,9 @@
 # Qspawn filter
 package PublicInbox::GzipFilter;
 use strict;
+use parent qw(Exporter);
 use Compress::Raw::Zlib qw(Z_FINISH Z_OK);
+our @EXPORT_OK = qw(gzip_maybe);
 my %OPT = (-WindowBits => 15 + 16, -AppendOutput => 1);
 
 sub new { bless {}, shift }
@@ -16,6 +18,17 @@ sub attach {
 	$self
 }
 
+sub gzip_maybe ($) {
+	my ($env) = @_;
+	return if (($env->{HTTP_ACCEPT_ENCODING}) // '') !~ /\bgzip\b/;
+
+	# in case Plack::Middleware::Deflater is loaded:
+	$env->{'plack.skip-deflater'} = 1;
+
+	my ($gz, $err) = Compress::Raw::Zlib::Deflate->new(%OPT);
+	$err == Z_OK ? $gz : undef;
+}
+
 # for GetlineBody (via Qspawn) when NOT using $env->{'pi-httpd.async'}
 sub translate ($$) {
 	my $self = $_[0];
diff --git a/lib/PublicInbox/WwwStream.pm b/lib/PublicInbox/WwwStream.pm
index 915a71ba0..79ed6871e 100644
--- a/lib/PublicInbox/WwwStream.pm
+++ b/lib/PublicInbox/WwwStream.pm
@@ -13,6 +13,8 @@ use base qw(Exporter);
 our @EXPORT_OK = qw(html_oneshot);
 use bytes (); # length
 use PublicInbox::Hval qw(ascii_html prurl);
+use Compress::Raw::Zlib qw(Z_FINISH Z_OK);
+use PublicInbox::GzipFilter qw(gzip_maybe);
 our $TOR_URL = 'https://www.torproject.org/';
 our $CODE_URL = 'https://public-inbox.org/public-inbox.git';
 
@@ -178,13 +180,28 @@ sub html_oneshot ($$;$) {
 		ctx => $ctx,
 		base_url => base_url($ctx),
 	}, __PACKAGE__;
-	my @x = (_html_top($self), $sref ? $$sref : (), _html_end($self));
+	my @x;
+	my @h = ('Content-Type' => 'text/html; charset=UTF-8');
+	if (my $gz = gzip_maybe($ctx->{env})) {
+		my $err = $gz->deflate(_html_top($self), $x[0]);
+		die "gzip->deflate: $err" if $err != Z_OK;
+		if ($sref) {
+			$err = $gz->deflate($sref, $x[0]);
+			die "gzip->deflate: $err" if $err != Z_OK;
+		}
+		$err = $gz->deflate(_html_end($self), $x[0]);
+		die "gzip->deflate: $err" if $err != Z_OK;
+		$err = $gz->flush($x[0], Z_FINISH);
+		die "gzip->flush: $err" if $err != Z_OK;
+		push @h, qw(Vary Accept-Encoding Content-Encoding gzip);
+	} else {
+		@x = (_html_top($self), $sref ? $$sref : (), _html_end($self));
+	}
+
 	my $len = 0;
 	$len += bytes::length($_) for @x;
-	[ $code, [
-		'Content-Type' => 'text/html; charset=UTF-8',
-		'Content-Length' => $len
-	], \@x ];
+	push @h, 'Content-Length', $len;
+	[ $code, \@h, \@x ]
 }
 
 1;

  parent reply	other threads:[~2020-07-05 23:28 UTC|newest]

Thread overview: 44+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-07-05 23:27 [PATCH 00/43] www: async git cat-file w/ -httpd Eric Wong
2020-07-05 23:27 ` [PATCH 01/43] gzipfilter: minor cleanups Eric Wong
2020-07-05 23:27 ` Eric Wong [this message]
2020-07-05 23:27 ` [PATCH 03/43] www*stream: gzip ->getline responses Eric Wong
2020-07-05 23:27 ` [PATCH 04/43] wwwtext: gzip text/plain responses, as well Eric Wong
2020-07-05 23:27 ` [PATCH 05/43] wwwtext: switch to html_oneshot Eric Wong
2020-07-05 23:27 ` [PATCH 06/43] www: need: use WwwStream::html_oneshot Eric Wong
2020-07-05 23:27 ` [PATCH 07/43] wwwlisting: use GzipFilter for HTML Eric Wong
2020-07-05 23:27 ` [PATCH 08/43] gzipfilter: replace Compress::Raw::Deflate usages Eric Wong
2020-07-05 23:27 ` [PATCH 09/43] {gzip,noop}filter: ->zmore returns undef, always Eric Wong
2020-07-05 23:27 ` [PATCH 10/43] mbox: remove html_oneshot import Eric Wong
2020-07-05 23:27 ` [PATCH 11/43] wwwstatic: support gzipped directory listings Eric Wong
2020-07-05 23:27 ` [PATCH 12/43] qspawn: learn to gzip streaming responses Eric Wong
2020-07-05 23:27 ` [PATCH 13/43] stop auto-loading Plack::Middleware::Deflater Eric Wong
2020-07-05 23:27 ` [PATCH 14/43] mboxgz: do asynchronous git blob retrievals Eric Wong
2020-07-05 23:27 ` [PATCH 15/43] mboxgz: reduce hash depth Eric Wong
2020-07-05 23:27 ` [PATCH 16/43] mbox: async blob fetch for "single message" raw mboxrd Eric Wong
2020-07-05 23:27 ` [PATCH 17/43] wwwatomstream: simplify feed_update callers Eric Wong
2020-07-05 23:27 ` [PATCH 18/43] wwwatomstream: use PublicInbox::Inbox->modified for feed_updated Eric Wong
2020-07-05 23:27 ` [PATCH 19/43] wwwatomstream: reuse $ctx as $self Eric Wong
2020-07-05 23:27 ` [PATCH 20/43] xt/httpd-async-stream: allow more options Eric Wong
2020-07-05 23:27 ` [PATCH 21/43] wwwatomstream: support async blob fetch Eric Wong
2020-07-05 23:27 ` [PATCH 22/43] wwwstream: reduce object graph depth Eric Wong
2020-07-05 23:27 ` [PATCH 23/43] wwwstream: reduce blob fetch paths for ->getline Eric Wong
2020-07-05 23:27 ` [PATCH 24/43] www: start making gzipfilter the parent response class Eric Wong
2020-07-05 23:27 ` [PATCH 25/43] remove unused/redundant zlib-related imports Eric Wong
2020-07-05 23:27 ` [PATCH 26/43] wwwstream: use parent.pm and no warnings Eric Wong
2020-07-05 23:27 ` [PATCH 27/43] wwwstream: subclass off GzipFilter Eric Wong
2020-07-05 23:27 ` [PATCH 28/43] view: make /$INBOX/$MSGID/ permalink async Eric Wong
2020-07-05 23:27 ` [PATCH 29/43] view: /$INBOX/$MSGID/t/ reads blobs asynchronously Eric Wong
2020-07-05 23:27 ` [PATCH 30/43] view: update /$INBOX/$MSGID/T/ to be async Eric Wong
2020-07-05 23:27 ` [PATCH 31/43] feed: generate_i: eliminate pointless loop Eric Wong
2020-07-05 23:27 ` [PATCH 32/43] feed: /$INBOX/new.html fetches blobs asynchronously Eric Wong
2020-07-05 23:27 ` [PATCH 33/43] ssearchview: /$INBOX/?q=$QUERY&x=t uses async blobs Eric Wong
2020-07-05 23:27 ` [PATCH 34/43] view: eml_entry: reduce parameters Eric Wong
2020-07-05 23:27 ` [PATCH 35/43] view: /$INBOX/$MSGID/t/: avoid extra hash lookup in eml case Eric Wong
2020-07-05 23:27 ` [PATCH 36/43] wwwstream: eliminate ::response, use html_oneshot Eric Wong
2020-07-05 23:27 ` [PATCH 37/43] www: update internal docs Eric Wong
2020-07-05 23:27 ` [PATCH 38/43] view: simplify eml_entry callers further Eric Wong
2020-07-05 23:27 ` [PATCH 39/43] wwwtext: simplify gzf_maybe use Eric Wong
2020-07-05 23:27 ` [PATCH 40/43] wwwattach: support async blob retrievals Eric Wong
2020-07-05 23:27 ` [PATCH 41/43] gzipfilter: drop HTTP connection on bugs or data corruption Eric Wong
2020-07-05 23:27 ` [PATCH 42/43] daemon: warn on missing blobs Eric Wong
2020-07-05 23:27 ` [PATCH 43/43] gzipfilter: check http->{forward} for client disconnects 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=20200705232759.3161-3-e@yhbt.net \
    --to=e@yhbt.net \
    --cc=meta@public-inbox.org \
    /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
Be sure your reply has a Subject: header at the top and a blank line before the message body.
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).