Naive port of mod_pagespeed for nginx

luislavena · June 6, 2011, 8:25pm

Hi all,

This is just to announce my naive yet functional port of Google’s
mod_pagespeed to nginx.
I’m not sure I’ll have much time to improve on this, but this might
interest
you.

Right now it compiles with the latest stable version of mod_pagespeed
and
does a very simple parsing using the AttrValuesSaverFilter that I found
in
the unittests. (it just saves all the html tag attributes found while
parsing in a string).

The code and instructions to compile it can be found here :

Happy hacking,
Matthieu.

Matthieu_T · June 7, 2011, 6:10am

Matthieu this is great! Nice work! Please keep us posted on this.

-Josh

On Mon, Jun 6, 2011 at 2:23 PM, Matthieu T.

Matthieu_T · June 7, 2011, 6:11am

Out of curiosity, have you defined a mechanism to configure nginx to use
any
of the predefined filters? It looks like you were aware of
RewriteOptions,
which is the API-based equivalent to pagespeed.conf, but you are not
using
it yet in

github.com

mtourne/ngx_instaweb/blob/master/ngx_http_instaweb_module.cpp

/**
 * Copyright (c) 2010, 2011 CloudFlare, Inc. (http://www.cloudflare.com)
 * Copyright (c) 2010, 2011 Matthieu Tourne <[email protected]>
 *
 * @author Matthieu Tourne <[email protected]>
 */

#include "ngx_http_instaweb_module.h"
#include "net/instaweb/htmlparse/public/html_parse.h"
#include "net/instaweb/htmlparse/public/empty_html_filter.h"
#include "net/instaweb/rewriter/public/rewrite_driver.h"
#include "net/instaweb/rewriter/public/rewrite_options.h"

namespace net_instaweb {

class AttrValuesSaverFilter : public EmptyHtmlFilter {
 public:
  AttrValuesSaverFilter() { }

  virtual void StartElement(HtmlElement* element) {

This file has been truncated. show original

.

The next step in your nginx integration looks to be resource-handling
and
fetching. Specifically, you need to define a way to fetch external HTTP
resources referenced on a page, and a way to handle fetches for
rewritten
resources. We can help you if you have any questions feel free to reach
out.

Please also check this out:

We’ve been working along the same lines you have: starting with a
mod_pagespeed build, generate a linkable library that has the kernel of
functionality needed to integrate mod_pagespeed’s parser & algorithms
into a
new platform.

-Josh

Matthieu_T · June 7, 2011, 10:26am

On Mon, Jun 6, 2011 at 6:37 PM, Matthieu T.
[email protected]wrote:

I think I started looking at creating a RewriteOptions object, but got side
tracked and this project has been idle for a little while.
I just updated it this weekend for the latest mod_pagespeed, and put it out
there.

Cool. As you get time to push this forward, please don’t hesitate to
ask
questions.

I like how the instaweb API is designed, and how you provide "backend

functionalities" through adapter objects (if I remember correctly).
With nginx extensive subrequests mechanism, fetching external resources
shouldn’t be too difficult.

Thanks! It was a goal up front to make ports like this possible.

This Makefile could definitely help. I was also wondering if there was a

python lib that could make sense of Google’s internal build files ?

I’m not 100% sure – we are not experts in the .gyp file system
ourselves
although there is doc here:
gyp - Generate Your Projects - Monorail. What problem
are
you trying to solve?

Matthieu_T · June 7, 2011, 9:08am

I think I started looking at creating a RewriteOptions object, but got
side
tracked and this project has been idle for a little while.
I just updated it this weekend for the latest mod_pagespeed, and put it
out
there.

I like how the instaweb API is designed, and how you provide “backend
functionalities” through adapter objects (if I remember correctly).
With nginx extensive subrequests mechanism, fetching external resources
shouldn’t be too difficult.

This Makefile could definitely help. I was also wondering if there was a
python lib that could make sense of Google’s internal build files ?

Thank you!

Matthieu.