[PATCH 0/5] cgit + WwwHighlight.pm integration

[PATCH 0/5] cgit + WwwHighlight.pm integration

[Eric Wong]

cgit-wwwhighlight-filter.lua can benefit cgit installations by
allowing them to avoid spawning a new process for about-filter
and source-filter in cgitrc.

Instead of spawning highlight(1) or similar every time a cgit
filter is invoked, we can have a long-lived PSGI server running
WwwHighlight.pm(*) and rely on lua-http to access the server
over loopback.

WwwHighlight.pm is also updated to support the same
linkification logic we use for emails.


(*) see examples/highlight.psgi

Eric Wong (5):
  wwwhighlight: do linkification, too
  examples/cgit-commit-filter: remove unused variable
  wwwhighlight: deal with no-op highlight
  examples: cgit filter for use with WwwHighlight
  examples/cgit-commit-filter.lua: some doc updates

 MANIFEST                              |  1 +
 examples/cgit-commit-filter.lua       | 10 ++++-
 examples/cgit-wwwhighlight-filter.lua | 57 +++++++++++++++++++++++++++
 lib/PublicInbox/WwwHighlight.pm       | 11 +++++-
 4 files changed, 77 insertions(+), 2 deletions(-)
 create mode 100644 examples/cgit-wwwhighlight-filter.lua

[PATCH 1/5] wwwhighlight: do linkification, too

[Eric Wong]

Otherwise, there's no reason to use this API over highlight(1).
Maybe this can be an option in the future; but I'm struggling to
find a reason to not do it by default.
---
 lib/PublicInbox/WwwHighlight.pm | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/lib/PublicInbox/WwwHighlight.pm b/lib/PublicInbox/WwwHighlight.pm
index 0191640..d8101b7 100644
--- a/lib/PublicInbox/WwwHighlight.pm
+++ b/lib/PublicInbox/WwwHighlight.pm
@@ -24,6 +24,7 @@ use warnings;
 use bytes (); # only for bytes::length
 use HTTP::Status qw(status_message);
 use parent qw(PublicInbox::HlMod);
+use PublicInbox::Linkify qw();
 
 # TODO: support highlight(1) for distros which don't package the
 # SWIG extension.  Also, there may be admins who don't want to
@@ -64,7 +65,10 @@ sub call {
 	return r(405) if $req_method ne 'PUT';
 
 	my $bref = read_in_full($env) or return r(500);
+	my $l = PublicInbox::Linkify->new;
+	$l->linkify_1($$bref);
 	$bref = $self->do_hl($bref, $env->{PATH_INFO});
+	$l->linkify_2($$bref);
 
 	my $h = [ 'Content-Type', 'text/html; charset=UTF-8' ];
 	push @$h, 'Content-Length', bytes::length($$bref);
-- 
EW

[PATCH 2/5] examples/cgit-commit-filter: remove unused variable

[Eric Wong]

---
 examples/cgit-commit-filter.lua | 1 -
 1 file changed, 1 deletion(-)

diff --git a/examples/cgit-commit-filter.lua b/examples/cgit-commit-filter.lua
index 7799bef..a8f9e1d 100644
--- a/examples/cgit-commit-filter.lua
+++ b/examples/cgit-commit-filter.lua
@@ -17,7 +17,6 @@ urls['public-inbox.git'] = 'https://public-inbox.org/meta/'
 function filter_open(...)
 	lineno = 0
 	buffer = ""
-	subject = ""
 end
 
 function filter_close()
-- 
EW

[PATCH 3/5] wwwhighlight: deal with no-op highlight

[Eric Wong]

Not all inputs are highlight-able, so reuse the original
input and just linkify it if it can't be highlighted.
---
 lib/PublicInbox/WwwHighlight.pm | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/lib/PublicInbox/WwwHighlight.pm b/lib/PublicInbox/WwwHighlight.pm
index d8101b7..bc349f8 100644
--- a/lib/PublicInbox/WwwHighlight.pm
+++ b/lib/PublicInbox/WwwHighlight.pm
@@ -25,6 +25,7 @@ use bytes (); # only for bytes::length
 use HTTP::Status qw(status_message);
 use parent qw(PublicInbox::HlMod);
 use PublicInbox::Linkify qw();
+use PublicInbox::Hval qw(ascii_html);
 
 # TODO: support highlight(1) for distros which don't package the
 # SWIG extension.  Also, there may be admins who don't want to
@@ -67,7 +68,11 @@ sub call {
 	my $bref = read_in_full($env) or return r(500);
 	my $l = PublicInbox::Linkify->new;
 	$l->linkify_1($$bref);
-	$bref = $self->do_hl($bref, $env->{PATH_INFO});
+	if (my $res = $self->do_hl($bref, $env->{PATH_INFO})) {
+		$bref = $res;
+	} else {
+		$$bref = ascii_html($$bref);
+	}
 	$l->linkify_2($$bref);
 
 	my $h = [ 'Content-Type', 'text/html; charset=UTF-8' ];
-- 
EW

[PATCH 4/5] examples: cgit filter for use with WwwHighlight

[Eric Wong]

I'm using this as the cgit about-filter and source-filter
in https://80x24.org/public-inbox.git
---
 MANIFEST                              |  1 +
 examples/cgit-wwwhighlight-filter.lua | 57 +++++++++++++++++++++++++++
 2 files changed, 58 insertions(+)
 create mode 100644 examples/cgit-wwwhighlight-filter.lua

diff --git a/MANIFEST b/MANIFEST
index 9858b9e..ed8ff49 100644
--- a/MANIFEST
+++ b/MANIFEST
@@ -39,6 +39,7 @@ examples/apache2_perl.conf
 examples/apache2_perl_old.conf
 examples/cgi-webrick.rb
 examples/cgit-commit-filter.lua
+examples/cgit-wwwhighlight-filter.lua
 examples/cgit.psgi
 examples/highlight.psgi
 examples/logrotate.conf
diff --git a/examples/cgit-wwwhighlight-filter.lua b/examples/cgit-wwwhighlight-filter.lua
new file mode 100644
index 0000000..dafb7a8
--- /dev/null
+++ b/examples/cgit-wwwhighlight-filter.lua
@@ -0,0 +1,57 @@
+-- Copyright (C) 2019 all contributors <meta@public-inbox.org>
+-- License: GPL-2.0+ <https://www.gnu.org/licenses/gpl-2.0.txt>
+--
+-- This filter accesses the PublicInbox::WwwHighlight PSGI endpoint
+-- (see examples/highlight.psgi)
+--
+-- Dependencies: lua-http
+--
+-- disclaimer: written by someone who does not know Lua.
+--
+-- This requires cgit linked with Lua
+-- Usage (in your cgitrc(5) config file):
+--
+--   source-filter=lua:/path/to/this/script.lua
+--   about-filter=lua:/path/to/this/script.lua
+--
+local wwwhighlight_url = 'http://127.0.0.1:9090/'
+local req_timeout = 10
+
+-- this is nil when used as about-filter
+local fn
+
+function filter_open(...)
+	req_body = ""
+	fn = select(1, ...)
+	if fn then
+		local http_util = require 'http.util'
+		wwwhighlight_url = wwwhighlight_url .. http_util.encodeURI(fn)
+	end
+end
+
+-- buffer the entire source in memory
+function filter_write(str)
+	req_body = req_body .. str
+end
+
+function filter_close()
+	local request = require 'http.request'
+	local req = request.new_from_uri(wwwhighlight_url)
+	req.headers:upsert(':method', 'PUT')
+	req:set_body(req_body)
+	local headers, stream = req:go(req_timeout)
+	local body, err = stream:get_body_as_string()
+	if fn then
+		html("<pre>")
+	end
+	if not body and err then
+		io.stderr:write(tostring(err), "\n")
+		html_txt(req_body)
+	else
+		html(body)
+	end
+	if fn then
+		html("</pre>")
+	end
+	return 0
+end
-- 
EW

[PATCH 5/5] examples/cgit-commit-filter.lua: some doc updates

[Eric Wong]

It's been a while since I wrote this, and it needs to be kept
up-to-date with some advances in our Perl code.
---
 examples/cgit-commit-filter.lua | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/examples/cgit-commit-filter.lua b/examples/cgit-commit-filter.lua
index a8f9e1d..1677253 100644
--- a/examples/cgit-commit-filter.lua
+++ b/examples/cgit-commit-filter.lua
@@ -13,6 +13,9 @@
 local urls = {}
 urls['public-inbox.git'] = 'https://public-inbox.org/meta/'
 -- additional URLs here...
+-- TODO we should be able to auto-generate this based on "coderepo"
+-- directives in the public-inbox config file; but keep in mind
+-- the mapping is M:N between inboxes and coderepos
 
 function filter_open(...)
 	lineno = 0
@@ -20,6 +23,9 @@ function filter_open(...)
 end
 
 function filter_close()
+	-- cgit opens and closes this filter for the commit subject
+	-- and body separately, and we only generate the link based
+	-- on the commit subject:
 	if lineno == 1 and string.find(buffer, "\n") == nil then
 		u = urls[os.getenv('CGIT_REPO_URL')]
 		if u == nil then
@@ -32,6 +38,9 @@ function filter_close()
 			html('</tt></a>')
 		end
 	else
+		-- pass the body-through as-is
+		-- TODO: optionally use WwwHighlight for linkification like
+		-- cgit-wwwhighlight-filter.lua
 		html(buffer)
 	end
 	return 0
-- 
EW

[PATCH v2] examples: cgit filter for use with WwwHighlight

[Eric Wong]

I'm using this as the cgit about-filter and source-filter
in https://80x24.org/public-inbox.git
---
 v2:
 * proper error handling and behavior w.r.t. large files
 * do not send Expect: header (and waiting for 100-continue)
 * correctly distinguish between about-filter and source-filter
   (v1 was creating nested <pre> when used as source-filter)

 interdiff:
 $ U=https://public-inbox.org/meta
 $ interdiff <(curl -s $U/dafb7a8/s/?b=x.lua) <(curl -s $U/a267d1c/s/?b=x.lua)

 MANIFEST                              |   1 +
 examples/cgit-wwwhighlight-filter.lua | 105 ++++++++++++++++++++++++++
 2 files changed, 106 insertions(+)
 create mode 100644 examples/cgit-wwwhighlight-filter.lua

diff --git a/MANIFEST b/MANIFEST
index 9858b9e..ed8ff49 100644
--- a/MANIFEST
+++ b/MANIFEST
@@ -39,6 +39,7 @@ examples/apache2_perl.conf
 examples/apache2_perl_old.conf
 examples/cgi-webrick.rb
 examples/cgit-commit-filter.lua
+examples/cgit-wwwhighlight-filter.lua
 examples/cgit.psgi
 examples/highlight.psgi
 examples/logrotate.conf
diff --git a/examples/cgit-wwwhighlight-filter.lua b/examples/cgit-wwwhighlight-filter.lua
new file mode 100644
index 0000000..a267d1c
--- /dev/null
+++ b/examples/cgit-wwwhighlight-filter.lua
@@ -0,0 +1,105 @@
+-- Copyright (C) 2019 all contributors <meta@public-inbox.org>
+-- License: GPL-2.0+ <https://www.gnu.org/licenses/gpl-2.0.txt>
+--
+-- This filter accesses the PublicInbox::WwwHighlight PSGI endpoint
+-- (see examples/highlight.psgi)
+--
+-- Dependencies: lua-http
+--
+-- disclaimer: written by someone who does not know Lua.
+--
+-- This requires cgit linked with Lua
+-- Usage (in your cgitrc(5) config file):
+--
+--   source-filter=lua:/path/to/this/script.lua
+--   about-filter=lua:/path/to/this/script.lua
+--
+local wwwhighlight_url = 'http://127.0.0.1:9090/'
+local req_timeout = 10
+local too_big = false
+
+-- match $PublicInbox::HTTP::MAX_REQUEST_BUFFER
+local max_len = 10 * 1024 * 1024
+
+-- about-filter needs surrounding <pre> tags if all we do is
+-- highlight and linkify
+local pre = true
+
+function filter_open(...)
+	req_body = ""
+
+	-- detect when we're used in an about-filter
+	local repo_url = os.getenv('CGIT_REPO_URL')
+	if repo_url then
+		local path_info = os.getenv('PATH_INFO')
+		rurl = path_info:match("^/(.+)/about/?$")
+		pre = rurl == repo_url
+	end
+
+	-- hand filename off for language detection
+	local fn = select(1, ...)
+	if fn then
+		local http_util = require 'http.util'
+		wwwhighlight_url = wwwhighlight_url .. http_util.encodeURI(fn)
+	end
+end
+
+-- try to buffer the entire source in memory
+function filter_write(str)
+	if too_big then
+		html(str)
+	elseif (req_body:len() + str:len()) > max_len then
+		too_big = true
+		req_body = ""
+		html(req_body)
+		html(str)
+	else
+		req_body = req_body .. str
+	end
+end
+
+function fail(err)
+	io.stderr:write(tostring(err), "\n")
+	if pre then
+		html("<pre>")
+	end
+	html_txt(req_body)
+	if pre then
+		html("</pre>")
+	end
+	return 1
+end
+
+function filter_close()
+	if too_big then
+		return 0
+	end
+	local request = require 'http.request'
+	local req = request.new_from_uri(wwwhighlight_url)
+	req.headers:upsert(':method', 'PUT')
+	req:set_body(req_body)
+
+	-- don't wait for 100-Continue message from the PSGI app
+	req.headers:delete('expect')
+
+	local headers, stream = req:go(req_timeout)
+	if headers == nil then
+		return fail(stream)
+	end
+	local status = headers:get(':status')
+	if status ~= '200' then
+		return fail('status ' .. status)
+	end
+	local body, err = stream:get_body_as_string()
+	if not body and err then
+		return fail(err)
+	end
+	if pre then
+		html("<pre>")
+	end
+	html(body)
+	if pre then
+		html("</pre>")
+	end
+	return 0
+end
-- 
EW

Re: [PATCH v2] examples: cgit filter for use with WwwHighlight

[Eric Wong]

>  interdiff:
>  $ U=https://public-inbox.org/meta
>  $ interdiff <(curl -s $U/dafb7a8/s/?b=x.lua) <(curl -s $U/a267d1c/s/?b=x.lua)

Oops, I got that wrong :x  It should be:

	diff -u <(curl -s $U/dafb7a8/s/x.lua) <(curl -s $U/a267d1c/s/x.lua)